diff options
author | Thomas Zimmermann <tzimmermann@suse.de> | 2022-11-24 11:28:05 +0300 |
---|---|---|
committer | Thomas Zimmermann <tzimmermann@suse.de> | 2022-11-24 11:28:05 +0300 |
commit | 1e5b3968a57d7894d5f86a2ecb58fa057cb6f7b2 (patch) | |
tree | a1beadf663bf4a318cccbfba8b4679b5fd0991c7 /drivers/gpu | |
parent | 6fb6c979ca628583d4d0c59a0f8ff977e581ecc0 (diff) | |
parent | d47f9580839eb6fe568e38b2084d94887fbf5ce0 (diff) | |
download | linux-1e5b3968a57d7894d5f86a2ecb58fa057cb6f7b2.tar.xz |
Merge drm/drm-next into drm-misc-next
Backmerging to get v6.1-rc6 into drm-misc-next.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Diffstat (limited to 'drivers/gpu')
936 files changed, 25206 insertions, 18091 deletions
diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig index 722ded626399..315cbdf61979 100644 --- a/drivers/gpu/drm/Kconfig +++ b/drivers/gpu/drm/Kconfig @@ -233,64 +233,8 @@ source "drivers/gpu/drm/i2c/Kconfig" source "drivers/gpu/drm/arm/Kconfig" -config DRM_RADEON - tristate "ATI Radeon" - depends on DRM && PCI && MMU - depends on AGP || !AGP - select FW_LOADER - select DRM_DISPLAY_DP_HELPER - select DRM_DISPLAY_HELPER - select DRM_KMS_HELPER - select DRM_TTM - select DRM_TTM_HELPER - select POWER_SUPPLY - select HWMON - select BACKLIGHT_CLASS_DEVICE - select INTERVAL_TREE - # radeon depends on ACPI_VIDEO when ACPI is enabled, for select to work - # ACPI_VIDEO's dependencies must also be selected. - select INPUT if ACPI - select ACPI_VIDEO if ACPI - # On x86 ACPI_VIDEO also needs ACPI_WMI - select X86_PLATFORM_DEVICES if ACPI && X86 - select ACPI_WMI if ACPI && X86 - help - Choose this option if you have an ATI Radeon graphics card. There - are both PCI and AGP versions. You don't need to choose this to - run the Radeon in plain VGA mode. - - If M is selected, the module will be called radeon. - source "drivers/gpu/drm/radeon/Kconfig" -config DRM_AMDGPU - tristate "AMD GPU" - depends on DRM && PCI && MMU - select FW_LOADER - select DRM_DISPLAY_DP_HELPER - select DRM_DISPLAY_HDMI_HELPER - select DRM_DISPLAY_HELPER - select DRM_KMS_HELPER - select DRM_SCHED - select DRM_TTM - select DRM_TTM_HELPER - select POWER_SUPPLY - select HWMON - select BACKLIGHT_CLASS_DEVICE - select INTERVAL_TREE - select DRM_BUDDY - # amdgpu depends on ACPI_VIDEO when ACPI is enabled, for select to work - # ACPI_VIDEO's dependencies must also be selected. - select INPUT if ACPI - select ACPI_VIDEO if ACPI - # On x86 ACPI_VIDEO also needs ACPI_WMI - select X86_PLATFORM_DEVICES if ACPI && X86 - select ACPI_WMI if ACPI && X86 - help - Choose this option if you have a recent AMD Radeon graphics card. - - If M is selected, the module will be called amdgpu. - source "drivers/gpu/drm/amd/amdgpu/Kconfig" source "drivers/gpu/drm/nouveau/Kconfig" diff --git a/drivers/gpu/drm/amd/amdgpu/Kconfig b/drivers/gpu/drm/amd/amdgpu/Kconfig index 7777d55275de..5fcd510f1abb 100644 --- a/drivers/gpu/drm/amd/amdgpu/Kconfig +++ b/drivers/gpu/drm/amd/amdgpu/Kconfig @@ -1,4 +1,33 @@ # SPDX-License-Identifier: MIT + +config DRM_AMDGPU + tristate "AMD GPU" + depends on DRM && PCI && MMU + select FW_LOADER + select DRM_DISPLAY_DP_HELPER + select DRM_DISPLAY_HDMI_HELPER + select DRM_DISPLAY_HELPER + select DRM_KMS_HELPER + select DRM_SCHED + select DRM_TTM + select DRM_TTM_HELPER + select POWER_SUPPLY + select HWMON + select BACKLIGHT_CLASS_DEVICE + select INTERVAL_TREE + select DRM_BUDDY + # amdgpu depends on ACPI_VIDEO when ACPI is enabled, for select to work + # ACPI_VIDEO's dependencies must also be selected. + select INPUT if ACPI + select ACPI_VIDEO if ACPI + # On x86 ACPI_VIDEO also needs ACPI_WMI + select X86_PLATFORM_DEVICES if ACPI && X86 + select ACPI_WMI if ACPI && X86 + help + Choose this option if you have a recent AMD Radeon graphics card. + + If M is selected, the module will be called amdgpu. + config DRM_AMDGPU_SI bool "Enable amdgpu support for SI parts" depends on DRM_AMDGPU diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 6ad39cf71bdd..712075a491f2 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -250,7 +250,7 @@ endif amdgpu-$(CONFIG_COMPAT) += amdgpu_ioc32.o amdgpu-$(CONFIG_VGA_SWITCHEROO) += amdgpu_atpx_handler.o amdgpu-$(CONFIG_ACPI) += amdgpu_acpi.o -amdgpu-$(CONFIG_HMM_MIRROR) += amdgpu_mn.o +amdgpu-$(CONFIG_HMM_MIRROR) += amdgpu_hmm.o include $(FULL_AMD_PATH)/pm/Makefile diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 8639a4f9c6e8..6b74df446694 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -82,7 +82,6 @@ #include "amdgpu_vce.h" #include "amdgpu_vcn.h" #include "amdgpu_jpeg.h" -#include "amdgpu_mn.h" #include "amdgpu_gmc.h" #include "amdgpu_gfx.h" #include "amdgpu_sdma.h" @@ -219,10 +218,12 @@ extern int amdgpu_use_xgmi_p2p; extern int sched_policy; extern bool debug_evictions; extern bool no_system_mem_limit; +extern int halt_if_hws_hang; #else static const int __maybe_unused sched_policy = KFD_SCHED_POLICY_HWS; static const bool __maybe_unused debug_evictions; /* = false */ static const bool __maybe_unused no_system_mem_limit; +static const int __maybe_unused halt_if_hws_hang; #endif #ifdef CONFIG_HSA_AMD_P2P extern bool pcie_p2p; @@ -675,7 +676,7 @@ enum amd_hw_ip_block_type { MAX_HWIP }; -#define HWIP_MAX_INSTANCE 11 +#define HWIP_MAX_INSTANCE 28 #define HW_ID_MAX 300 #define IP_VERSION(mj, mn, rv) (((mj) << 16) | ((mn) << 8) | (rv)) @@ -1063,6 +1064,7 @@ struct amdgpu_device { struct work_struct reset_work; bool job_hang; + bool dc_enabled; }; static inline struct amdgpu_device *drm_to_adev(struct drm_device *ddev) @@ -1120,6 +1122,8 @@ void amdgpu_device_indirect_wreg64(struct amdgpu_device *adev, bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type); bool amdgpu_device_has_dc_support(struct amdgpu_device *adev); +void amdgpu_device_set_sriov_virtual_display(struct amdgpu_device *adev); + int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, struct amdgpu_reset_context *reset_context); @@ -1293,6 +1297,7 @@ void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev, u32 reg, u32 v); struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev, struct dma_fence *gang); +bool amdgpu_device_has_display_hardware(struct amdgpu_device *adev); /* atpx handler */ #if defined(CONFIG_VGA_SWITCHEROO) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c index b14800ac179e..57b5e11446c6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c @@ -847,7 +847,7 @@ int amdgpu_acpi_init(struct amdgpu_device *adev) struct amdgpu_atif *atif = &amdgpu_acpi_priv.atif; if (atif->notifications.brightness_change) { - if (amdgpu_device_has_dc_support(adev)) { + if (adev->dc_enabled) { #if defined(CONFIG_DRM_AMD_DC) struct amdgpu_display_manager *dm = &adev->dm; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index 046d466b4ee4..f99d4873bf22 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -195,7 +195,7 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev) } adev->kfd.init_complete = kgd2kfd_device_init(adev->kfd.dev, - adev_to_drm(adev), &gpu_resources); + &gpu_resources); amdgpu_amdkfd_total_mem_size += adev->gmc.real_vram_size; @@ -706,6 +706,13 @@ err: void amdgpu_amdkfd_set_compute_idle(struct amdgpu_device *adev, bool idle) { + /* Temporary workaround to fix issues observed in some + * compute applications when GFXOFF is enabled on GFX11. + */ + if (IP_VERSION_MAJ(adev->ip_versions[GC_HWIP][0]) == 11) { + pr_debug("GFXOFF is %s\n", idle ? "enabled" : "disabled"); + amdgpu_gfx_off_ctrl(adev, idle); + } amdgpu_dpm_switch_power_profile(adev, PP_SMC_POWER_PROFILE_COMPUTE, !idle); @@ -753,9 +760,7 @@ bool amdgpu_amdkfd_have_atomics_support(struct amdgpu_device *adev) void amdgpu_amdkfd_ras_poison_consumption_handler(struct amdgpu_device *adev, bool reset) { - struct ras_err_data err_data = {0, 0, 0, NULL}; - - amdgpu_umc_poison_handler(adev, &err_data, reset); + amdgpu_umc_poison_handler(adev, reset); } bool amdgpu_amdkfd_ras_query_utcl2_poison_status(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index 647220a8762d..f50e3ba4d7a5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -353,7 +353,6 @@ int kgd2kfd_init(void); void kgd2kfd_exit(void); struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf); bool kgd2kfd_device_init(struct kfd_dev *kfd, - struct drm_device *ddev, const struct kgd2kfd_shared_resources *gpu_resources); void kgd2kfd_device_exit(struct kfd_dev *kfd); void kgd2kfd_suspend(struct kfd_dev *kfd, bool run_pm); @@ -381,7 +380,7 @@ struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf) } static inline -bool kgd2kfd_device_init(struct kfd_dev *kfd, struct drm_device *ddev, +bool kgd2kfd_device_init(struct kfd_dev *kfd, const struct kgd2kfd_shared_resources *gpu_resources) { return false; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c index c8935d718207..4485bb29bec9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c @@ -41,5 +41,6 @@ const struct kfd2kgd_calls aldebaran_kfd2kgd = { .get_atc_vmid_pasid_mapping_info = kgd_gfx_v9_get_atc_vmid_pasid_mapping_info, .set_vm_context_page_table_base = kgd_gfx_v9_set_vm_context_page_table_base, + .get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy, .program_trap_handler_settings = kgd_gfx_v9_program_trap_handler_settings }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c index 81e3b528bbc9..e92b93557c13 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c @@ -787,7 +787,7 @@ void kgd_gfx_v9_get_cu_occupancy(struct amdgpu_device *adev, int pasid, for (se_idx = 0; se_idx < se_cnt; se_idx++) { for (sh_idx = 0; sh_idx < sh_cnt; sh_idx++) { - gfx_v9_0_select_se_sh(adev, se_idx, sh_idx, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, se_idx, sh_idx, 0xffffffff); queue_map = RREG32_SOC15(GC, 0, mmSPI_CSQ_WF_ACTIVE_STATUS); /* @@ -820,7 +820,7 @@ void kgd_gfx_v9_get_cu_occupancy(struct amdgpu_device *adev, int pasid, } } - gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); soc15_grbm_select(adev, 0, 0, 0, 0); unlock_spi_csq_mutexes(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index 978d3970b5cc..3a763916a5a1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -29,6 +29,7 @@ #include "amdgpu_object.h" #include "amdgpu_gem.h" #include "amdgpu_vm.h" +#include "amdgpu_hmm.h" #include "amdgpu_amdkfd.h" #include "amdgpu_dma_buf.h" #include <uapi/linux/kfd_ioctl.h> @@ -171,9 +172,7 @@ int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev, (kfd_mem_limit.ttm_mem_used + ttm_mem_needed > kfd_mem_limit.max_ttm_mem_limit) || (adev && adev->kfd.vram_used + vram_needed > - adev->gmc.real_vram_size - - atomic64_read(&adev->vram_pin_size) - - reserved_for_pt)) { + adev->gmc.real_vram_size - reserved_for_pt)) { ret = -ENOMEM; goto release; } @@ -405,63 +404,15 @@ static int vm_update_pds(struct amdgpu_vm *vm, struct amdgpu_sync *sync) static uint64_t get_pte_flags(struct amdgpu_device *adev, struct kgd_mem *mem) { - struct amdgpu_device *bo_adev = amdgpu_ttm_adev(mem->bo->tbo.bdev); - bool coherent = mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_COHERENT; - bool uncached = mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_UNCACHED; - uint32_t mapping_flags; - uint64_t pte_flags; - bool snoop = false; + uint32_t mapping_flags = AMDGPU_VM_PAGE_READABLE | + AMDGPU_VM_MTYPE_DEFAULT; - mapping_flags = AMDGPU_VM_PAGE_READABLE; if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_WRITABLE) mapping_flags |= AMDGPU_VM_PAGE_WRITEABLE; if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_EXECUTABLE) mapping_flags |= AMDGPU_VM_PAGE_EXECUTABLE; - switch (adev->asic_type) { - case CHIP_ARCTURUS: - case CHIP_ALDEBARAN: - if (mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) { - if (bo_adev == adev) { - if (uncached) - mapping_flags |= AMDGPU_VM_MTYPE_UC; - else if (coherent) - mapping_flags |= AMDGPU_VM_MTYPE_CC; - else - mapping_flags |= AMDGPU_VM_MTYPE_RW; - if (adev->asic_type == CHIP_ALDEBARAN && - adev->gmc.xgmi.connected_to_cpu) - snoop = true; - } else { - if (uncached || coherent) - mapping_flags |= AMDGPU_VM_MTYPE_UC; - else - mapping_flags |= AMDGPU_VM_MTYPE_NC; - if (amdgpu_xgmi_same_hive(adev, bo_adev)) - snoop = true; - } - } else { - if (uncached || coherent) - mapping_flags |= AMDGPU_VM_MTYPE_UC; - else - mapping_flags |= AMDGPU_VM_MTYPE_NC; - snoop = true; - } - break; - default: - if (uncached || coherent) - mapping_flags |= AMDGPU_VM_MTYPE_UC; - else - mapping_flags |= AMDGPU_VM_MTYPE_NC; - - if (!(mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM)) - snoop = true; - } - - pte_flags = amdgpu_gem_va_map_flags(adev, mapping_flags); - pte_flags |= snoop ? AMDGPU_PTE_SNOOPED : 0; - - return pte_flags; + return amdgpu_gem_va_map_flags(adev, mapping_flags); } /** @@ -510,13 +461,13 @@ kfd_mem_dmamap_userptr(struct kgd_mem *mem, struct ttm_tt *ttm = bo->tbo.ttm; int ret; + if (WARN_ON(ttm->num_pages != src_ttm->num_pages)) + return -EINVAL; + ttm->sg = kmalloc(sizeof(*ttm->sg), GFP_KERNEL); if (unlikely(!ttm->sg)) return -ENOMEM; - if (WARN_ON(ttm->num_pages != src_ttm->num_pages)) - return -EINVAL; - /* Same sequence as in amdgpu_ttm_tt_pin_userptr */ ret = sg_alloc_table_from_pages(ttm->sg, src_ttm->pages, ttm->num_pages, 0, @@ -988,6 +939,7 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr, struct amdkfd_process_info *process_info = mem->process_info; struct amdgpu_bo *bo = mem->bo; struct ttm_operation_ctx ctx = { true, false }; + struct hmm_range *range; int ret = 0; mutex_lock(&process_info->lock); @@ -998,7 +950,7 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr, goto out; } - ret = amdgpu_mn_register(bo, user_addr); + ret = amdgpu_hmm_register(bo, user_addr); if (ret) { pr_err("%s: Failed to register MMU notifier: %d\n", __func__, ret); @@ -1017,7 +969,7 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr, return 0; } - ret = amdgpu_ttm_tt_get_user_pages(bo, bo->tbo.ttm->pages); + ret = amdgpu_ttm_tt_get_user_pages(bo, bo->tbo.ttm->pages, &range); if (ret) { pr_err("%s: Failed to get user pages: %d\n", __func__, ret); goto unregister_out; @@ -1035,10 +987,10 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr, amdgpu_bo_unreserve(bo); release_out: - amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm); + amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, range); unregister_out: if (ret) - amdgpu_mn_unregister(bo); + amdgpu_hmm_unregister(bo); out: mutex_unlock(&process_info->lock); return ret; @@ -1673,6 +1625,11 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu( } } + if (flags & KFD_IOC_ALLOC_MEM_FLAGS_COHERENT) + alloc_flags |= AMDGPU_GEM_CREATE_COHERENT; + if (flags & KFD_IOC_ALLOC_MEM_FLAGS_UNCACHED) + alloc_flags |= AMDGPU_GEM_CREATE_UNCACHED; + *mem = kzalloc(sizeof(struct kgd_mem), GFP_KERNEL); if (!*mem) { ret = -ENOMEM; @@ -1817,7 +1774,7 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu( mutex_unlock(&process_info->lock); /* No more MMU notifiers */ - amdgpu_mn_unregister(mem->bo); + amdgpu_hmm_unregister(mem->bo); ret = reserve_bo_and_cond_vms(mem, NULL, BO_VM_ALL, &ctx); if (unlikely(ret)) @@ -1907,16 +1864,6 @@ int amdgpu_amdkfd_gpuvm_map_memory_to_gpu( */ mutex_lock(&mem->process_info->lock); - /* Lock mmap-sem. If we find an invalid userptr BO, we can be - * sure that the MMU notifier is no longer running - * concurrently and the queues are actually stopped - */ - if (amdgpu_ttm_tt_get_usermm(bo->tbo.ttm)) { - mmap_write_lock(current->mm); - is_invalid_userptr = atomic_read(&mem->invalid); - mmap_write_unlock(current->mm); - } - mutex_lock(&mem->lock); domain = mem->domain; @@ -2372,6 +2319,8 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info, /* Go through userptr_inval_list and update any invalid user_pages */ list_for_each_entry(mem, &process_info->userptr_inval_list, validate_list.head) { + struct hmm_range *range; + invalid = atomic_read(&mem->invalid); if (!invalid) /* BO hasn't been invalidated since the last @@ -2382,7 +2331,8 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info, bo = mem->bo; /* Get updated user pages */ - ret = amdgpu_ttm_tt_get_user_pages(bo, bo->tbo.ttm->pages); + ret = amdgpu_ttm_tt_get_user_pages(bo, bo->tbo.ttm->pages, + &range); if (ret) { pr_debug("Failed %d to get user pages\n", ret); @@ -2401,7 +2351,7 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info, * FIXME: Cannot ignore the return code, must hold * notifier_lock */ - amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm); + amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, range); } /* Mark the BO as valid unless it was invalidated diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c index b81b77a9efa6..9b97fa39d47a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c @@ -101,39 +101,97 @@ void amdgpu_atomfirmware_scratch_regs_init(struct amdgpu_device *adev) } } +static int amdgpu_atomfirmware_allocate_fb_v2_1(struct amdgpu_device *adev, + struct vram_usagebyfirmware_v2_1 *fw_usage, int *usage_bytes) +{ + uint32_t start_addr, fw_size, drv_size; + + start_addr = le32_to_cpu(fw_usage->start_address_in_kb); + fw_size = le16_to_cpu(fw_usage->used_by_firmware_in_kb); + drv_size = le16_to_cpu(fw_usage->used_by_driver_in_kb); + + DRM_DEBUG("atom firmware v2_1 requested %08x %dkb fw %dkb drv\n", + start_addr, + fw_size, + drv_size); + + if ((start_addr & ATOM_VRAM_OPERATION_FLAGS_MASK) == + (uint32_t)(ATOM_VRAM_BLOCK_SRIOV_MSG_SHARE_RESERVATION << + ATOM_VRAM_OPERATION_FLAGS_SHIFT)) { + /* Firmware request VRAM reservation for SR-IOV */ + adev->mman.fw_vram_usage_start_offset = (start_addr & + (~ATOM_VRAM_OPERATION_FLAGS_MASK)) << 10; + adev->mman.fw_vram_usage_size = fw_size << 10; + /* Use the default scratch size */ + *usage_bytes = 0; + } else { + *usage_bytes = drv_size << 10; + } + return 0; +} + +static int amdgpu_atomfirmware_allocate_fb_v2_2(struct amdgpu_device *adev, + struct vram_usagebyfirmware_v2_2 *fw_usage, int *usage_bytes) +{ + uint32_t fw_start_addr, fw_size, drv_start_addr, drv_size; + + fw_start_addr = le32_to_cpu(fw_usage->fw_region_start_address_in_kb); + fw_size = le16_to_cpu(fw_usage->used_by_firmware_in_kb); + + drv_start_addr = le32_to_cpu(fw_usage->driver_region0_start_address_in_kb); + drv_size = le32_to_cpu(fw_usage->used_by_driver_region0_in_kb); + + DRM_DEBUG("atom requested fw start at %08x %dkb and drv start at %08x %dkb\n", + fw_start_addr, + fw_size, + drv_start_addr, + drv_size); + + if ((fw_start_addr & (ATOM_VRAM_BLOCK_NEEDS_NO_RESERVATION << 30)) == 0) { + /* Firmware request VRAM reservation for SR-IOV */ + adev->mman.fw_vram_usage_start_offset = (fw_start_addr & + (~ATOM_VRAM_OPERATION_FLAGS_MASK)) << 10; + adev->mman.fw_vram_usage_size = fw_size << 10; + } + + if ((drv_start_addr & (ATOM_VRAM_BLOCK_NEEDS_NO_RESERVATION << 30)) == 0) { + /* driver request VRAM reservation for SR-IOV */ + adev->mman.drv_vram_usage_start_offset = (drv_start_addr & + (~ATOM_VRAM_OPERATION_FLAGS_MASK)) << 10; + adev->mman.drv_vram_usage_size = drv_size << 10; + } + + *usage_bytes = 0; + return 0; +} + int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev) { struct atom_context *ctx = adev->mode_info.atom_context; int index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1, vram_usagebyfirmware); - struct vram_usagebyfirmware_v2_1 *firmware_usage; - uint32_t start_addr, size; + struct vram_usagebyfirmware_v2_1 *fw_usage_v2_1; + struct vram_usagebyfirmware_v2_2 *fw_usage_v2_2; uint16_t data_offset; + uint8_t frev, crev; int usage_bytes = 0; - if (amdgpu_atom_parse_data_header(ctx, index, NULL, NULL, NULL, &data_offset)) { - firmware_usage = (struct vram_usagebyfirmware_v2_1 *)(ctx->bios + data_offset); - DRM_DEBUG("atom firmware requested %08x %dkb fw %dkb drv\n", - le32_to_cpu(firmware_usage->start_address_in_kb), - le16_to_cpu(firmware_usage->used_by_firmware_in_kb), - le16_to_cpu(firmware_usage->used_by_driver_in_kb)); - - start_addr = le32_to_cpu(firmware_usage->start_address_in_kb); - size = le16_to_cpu(firmware_usage->used_by_firmware_in_kb); - - if ((uint32_t)(start_addr & ATOM_VRAM_OPERATION_FLAGS_MASK) == - (uint32_t)(ATOM_VRAM_BLOCK_SRIOV_MSG_SHARE_RESERVATION << - ATOM_VRAM_OPERATION_FLAGS_SHIFT)) { - /* Firmware request VRAM reservation for SR-IOV */ - adev->mman.fw_vram_usage_start_offset = (start_addr & - (~ATOM_VRAM_OPERATION_FLAGS_MASK)) << 10; - adev->mman.fw_vram_usage_size = size << 10; - /* Use the default scratch size */ - usage_bytes = 0; - } else { - usage_bytes = le16_to_cpu(firmware_usage->used_by_driver_in_kb) << 10; + if (amdgpu_atom_parse_data_header(ctx, index, NULL, &frev, &crev, &data_offset)) { + if (frev == 2 && crev == 1) { + fw_usage_v2_1 = + (struct vram_usagebyfirmware_v2_1 *)(ctx->bios + data_offset); + amdgpu_atomfirmware_allocate_fb_v2_1(adev, + fw_usage_v2_1, + &usage_bytes); + } else if (frev >= 2 && crev >= 2) { + fw_usage_v2_2 = + (struct vram_usagebyfirmware_v2_2 *)(ctx->bios + data_offset); + amdgpu_atomfirmware_allocate_fb_v2_2(adev, + fw_usage_v2_2, + &usage_bytes); } } + ctx->scratch_size_bytes = 0; if (usage_bytes == 0) usage_bytes = 20 * 1024; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c index 2168163aad2d..252a876b0725 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c @@ -209,6 +209,7 @@ void amdgpu_bo_list_get_list(struct amdgpu_bo_list *list, list_add_tail(&e->tv.head, &bucket[priority]); e->user_pages = NULL; + e->range = NULL; } /* Connect the sorted buckets in the output list. */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h index 9caea1688fc3..e4d78491bcc7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h @@ -26,6 +26,8 @@ #include <drm/ttm/ttm_execbuf_util.h> #include <drm/amdgpu_drm.h> +struct hmm_range; + struct amdgpu_device; struct amdgpu_bo; struct amdgpu_bo_va; @@ -36,6 +38,7 @@ struct amdgpu_bo_list_entry { struct amdgpu_bo_va *bo_va; uint32_t priority; struct page **user_pages; + struct hmm_range *range; bool user_invalidated; }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 0528c2b1db6e..8516c814bc9b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -109,6 +109,7 @@ static int amdgpu_cs_p1_ib(struct amdgpu_cs_parser *p, return r; ++(num_ibs[r]); + p->gang_leader_idx = r; return 0; } @@ -287,8 +288,10 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p, } } - if (!p->gang_size) - return -EINVAL; + if (!p->gang_size) { + ret = -EINVAL; + goto free_partial_kdata; + } for (i = 0; i < p->gang_size; ++i) { ret = amdgpu_job_alloc(p->adev, vm, p->entities[i], vm, @@ -296,7 +299,7 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p, if (ret) goto free_all_kdata; } - p->gang_leader = p->jobs[p->gang_size - 1]; + p->gang_leader = p->jobs[p->gang_leader_idx]; if (p->ctx->vram_lost_counter != p->gang_leader->vram_lost_counter) { ret = -ECANCELED; @@ -917,7 +920,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, goto out_free_user_pages; } - r = amdgpu_ttm_tt_get_user_pages(bo, e->user_pages); + r = amdgpu_ttm_tt_get_user_pages(bo, e->user_pages, &e->range); if (r) { kvfree(e->user_pages); e->user_pages = NULL; @@ -995,10 +998,12 @@ out_free_user_pages: if (!e->user_pages) continue; - amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm); + amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, e->range); kvfree(e->user_pages); e->user_pages = NULL; + e->range = NULL; } + mutex_unlock(&p->bo_list->bo_list_mutex); return r; } @@ -1206,10 +1211,9 @@ static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p) return r; } - r = amdgpu_ctx_wait_prev_fence(p->ctx, p->entities[p->gang_size - 1]); + r = amdgpu_ctx_wait_prev_fence(p->ctx, p->entities[p->gang_leader_idx]); if (r && r != -ERESTARTSYS) DRM_ERROR("amdgpu_ctx_wait_prev_fence failed.\n"); - return r; } @@ -1243,9 +1247,12 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p, for (i = 0; i < p->gang_size; ++i) drm_sched_job_arm(&p->jobs[i]->base); - for (i = 0; i < (p->gang_size - 1); ++i) { + for (i = 0; i < p->gang_size; ++i) { struct dma_fence *fence; + if (p->jobs[i] == leader) + continue; + fence = &p->jobs[i]->base.s_fence->scheduled; r = drm_sched_job_add_dependency(&leader->base, fence); if (r) @@ -1270,7 +1277,8 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p, amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) { struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo); - r |= !amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm); + r |= !amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, e->range); + e->range = NULL; } if (r) { r = -EAGAIN; @@ -1281,7 +1289,10 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p, list_for_each_entry(e, &p->validated, tv.head) { /* Everybody except for the gang leader uses READ */ - for (i = 0; i < (p->gang_size - 1); ++i) { + for (i = 0; i < p->gang_size; ++i) { + if (p->jobs[i] == leader) + continue; + dma_resv_add_fence(e->tv.bo->base.resv, &p->jobs[i]->base.s_fence->finished, DMA_RESV_USAGE_READ); @@ -1291,7 +1302,7 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p, e->tv.num_shared = 0; } - seq = amdgpu_ctx_add_fence(p->ctx, p->entities[p->gang_size - 1], + seq = amdgpu_ctx_add_fence(p->ctx, p->entities[p->gang_leader_idx], p->fence); amdgpu_cs_post_dependencies(p); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h index 207e801c24ed..113f39510a72 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h @@ -54,6 +54,7 @@ struct amdgpu_cs_parser { /* scheduler job objects */ unsigned int gang_size; + unsigned int gang_leader_idx; struct drm_sched_entity *entities[AMDGPU_CS_GANG_SIZE]; struct amdgpu_job *jobs[AMDGPU_CS_GANG_SIZE]; struct amdgpu_job *gang_leader; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index f6d9d5da53cd..d2139ac12159 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -326,7 +326,10 @@ static int amdgpu_ctx_init(struct amdgpu_ctx_mgr *mgr, int32_t priority, if (r) return r; - ctx->stable_pstate = current_stable_pstate; + if (mgr->adev->pm.stable_pstate_ctx) + ctx->stable_pstate = mgr->adev->pm.stable_pstate_ctx->stable_pstate; + else + ctx->stable_pstate = current_stable_pstate; return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c index de61a85c4b02..0f16d3c09309 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c @@ -1969,7 +1969,7 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev) amdgpu_ta_if_debugfs_init(adev); #if defined(CONFIG_DRM_AMD_DC) - if (amdgpu_device_has_dc_support(adev)) + if (adev->dc_enabled) dtn_debugfs_init(adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index cc65df9f2419..b2b1c66bfe39 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -1569,7 +1569,7 @@ static int amdgpu_device_check_arguments(struct amdgpu_device *adev) * @pdev: pci dev pointer * @state: vga_switcheroo state * - * Callback for the switcheroo driver. Suspends or resumes the + * Callback for the switcheroo driver. Suspends or resumes * the asics before or after it is powered up using ACPI methods. */ static void amdgpu_switcheroo_set_state(struct pci_dev *pdev, @@ -1916,6 +1916,16 @@ static void amdgpu_device_enable_virtual_display(struct amdgpu_device *adev) } } +void amdgpu_device_set_sriov_virtual_display(struct amdgpu_device *adev) +{ + if (amdgpu_sriov_vf(adev) && !adev->enable_virtual_display) { + adev->mode_info.num_crtc = 1; + adev->enable_virtual_display = true; + DRM_INFO("virtual_display:%d, num_crtc:%d\n", + adev->enable_virtual_display, adev->mode_info.num_crtc); + } +} + /** * amdgpu_device_parse_gpu_info_fw - parse gpu info firmware * @@ -2398,7 +2408,7 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev) adev->ip_blocks[i].status.hw = true; /* right after GMC hw init, we create CSA */ - if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) { + if (amdgpu_mcbp) { r = amdgpu_allocate_static_csa(adev, &adev->virt.csa_obj, AMDGPU_GEM_DOMAIN_VRAM, AMDGPU_CSA_SIZE); @@ -3211,6 +3221,15 @@ static int amdgpu_device_ip_resume_phase2(struct amdgpu_device *adev) return r; } adev->ip_blocks[i].status.hw = true; + + if (adev->in_s0ix && adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) { + /* disable gfxoff for IP resume. The gfxoff will be re-enabled in + * amdgpu_device_resume() after IP resume. + */ + amdgpu_gfx_off_ctrl(adev, false); + DRM_DEBUG("will disable gfxoff for re-initializing other blocks\n"); + } + } return 0; @@ -3339,8 +3358,7 @@ bool amdgpu_device_asic_has_dc_support(enum amd_asic_type asic_type) */ bool amdgpu_device_has_dc_support(struct amdgpu_device *adev) { - if (amdgpu_sriov_vf(adev) || - adev->enable_virtual_display || + if (adev->enable_virtual_display || (adev->harvest_ip_mask & AMD_HARVEST_IP_DMU_MASK)) return false; @@ -4052,15 +4070,18 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev) * at suspend time. * */ -static void amdgpu_device_evict_resources(struct amdgpu_device *adev) +static int amdgpu_device_evict_resources(struct amdgpu_device *adev) { + int ret; + /* No need to evict vram on APUs for suspend to ram or s2idle */ if ((adev->in_s3 || adev->in_s0ix) && (adev->flags & AMD_IS_APU)) - return; + return 0; - if (amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM)) + ret = amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM); + if (ret) DRM_WARN("evicting device resources failed\n"); - + return ret; } /* @@ -4110,7 +4131,9 @@ int amdgpu_device_suspend(struct drm_device *dev, bool fbcon) if (!adev->in_s0ix) amdgpu_amdkfd_suspend(adev, adev->in_runpm); - amdgpu_device_evict_resources(adev); + r = amdgpu_device_evict_resources(adev); + if (r) + return r; amdgpu_fence_driver_hw_fini(adev); @@ -4184,8 +4207,17 @@ int amdgpu_device_resume(struct drm_device *dev, bool fbcon) } /* Make sure IB tests flushed */ + if (amdgpu_sriov_vf(adev)) + amdgpu_irq_gpu_reset_resume_helper(adev); flush_delayed_work(&adev->delayed_init_work); + if (adev->in_s0ix) { + /* re-enable gfxoff after IP resume. This re-enables gfxoff after + * it was disabled for IP resume in amdgpu_device_ip_resume_phase2(). + */ + amdgpu_gfx_off_ctrl(adev, true); + DRM_DEBUG("will enable gfxoff for the mission mode\n"); + } if (fbcon) drm_fb_helper_set_suspend_unlocked(adev_to_drm(adev)->fb_helper, false); @@ -4193,25 +4225,27 @@ int amdgpu_device_resume(struct drm_device *dev, bool fbcon) amdgpu_ras_resume(adev); - /* - * Most of the connector probing functions try to acquire runtime pm - * refs to ensure that the GPU is powered on when connector polling is - * performed. Since we're calling this from a runtime PM callback, - * trying to acquire rpm refs will cause us to deadlock. - * - * Since we're guaranteed to be holding the rpm lock, it's safe to - * temporarily disable the rpm helpers so this doesn't deadlock us. - */ + if (adev->mode_info.num_crtc) { + /* + * Most of the connector probing functions try to acquire runtime pm + * refs to ensure that the GPU is powered on when connector polling is + * performed. Since we're calling this from a runtime PM callback, + * trying to acquire rpm refs will cause us to deadlock. + * + * Since we're guaranteed to be holding the rpm lock, it's safe to + * temporarily disable the rpm helpers so this doesn't deadlock us. + */ #ifdef CONFIG_PM - dev->dev->power.disable_depth++; + dev->dev->power.disable_depth++; #endif - if (!amdgpu_device_has_dc_support(adev)) - drm_helper_hpd_irq_event(dev); - else - drm_kms_helper_hotplug_event(dev); + if (!adev->dc_enabled) + drm_helper_hpd_irq_event(dev); + else + drm_kms_helper_hotplug_event(dev); #ifdef CONFIG_PM - dev->dev->power.disable_depth--; + dev->dev->power.disable_depth--; #endif + } adev->in_suspend = false; if (amdgpu_acpi_smart_shift_update(dev, AMDGPU_SS_DEV_D0)) @@ -4560,6 +4594,10 @@ bool amdgpu_device_should_recover_gpu(struct amdgpu_device *adev) if (amdgpu_gpu_recovery == 0) goto disabled; + /* Skip soft reset check in fatal error mode */ + if (!amdgpu_ras_is_poison_mode_supported(adev)) + return true; + if (!amdgpu_device_ip_check_soft_reset(adev)) { dev_info(adev->dev,"Timeout, but no hardware hang detected.\n"); return false; @@ -5055,94 +5093,6 @@ static int amdgpu_device_suspend_display_audio(struct amdgpu_device *adev) return 0; } -static void amdgpu_device_recheck_guilty_jobs( - struct amdgpu_device *adev, struct list_head *device_list_handle, - struct amdgpu_reset_context *reset_context) -{ - int i, r = 0; - - for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { - struct amdgpu_ring *ring = adev->rings[i]; - int ret = 0; - struct drm_sched_job *s_job; - - if (!ring || !ring->sched.thread) - continue; - - s_job = list_first_entry_or_null(&ring->sched.pending_list, - struct drm_sched_job, list); - if (s_job == NULL) - continue; - - /* clear job's guilty and depend the folowing step to decide the real one */ - drm_sched_reset_karma(s_job); - drm_sched_resubmit_jobs_ext(&ring->sched, 1); - - if (!s_job->s_fence->parent) { - DRM_WARN("Failed to get a HW fence for job!"); - continue; - } - - ret = dma_fence_wait_timeout(s_job->s_fence->parent, false, ring->sched.timeout); - if (ret == 0) { /* timeout */ - DRM_ERROR("Found the real bad job! ring:%s, job_id:%llx\n", - ring->sched.name, s_job->id); - - - amdgpu_fence_driver_isr_toggle(adev, true); - - /* Clear this failed job from fence array */ - amdgpu_fence_driver_clear_job_fences(ring); - - amdgpu_fence_driver_isr_toggle(adev, false); - - /* Since the job won't signal and we go for - * another resubmit drop this parent pointer - */ - dma_fence_put(s_job->s_fence->parent); - s_job->s_fence->parent = NULL; - - /* set guilty */ - drm_sched_increase_karma(s_job); - amdgpu_reset_prepare_hwcontext(adev, reset_context); -retry: - /* do hw reset */ - if (amdgpu_sriov_vf(adev)) { - amdgpu_virt_fini_data_exchange(adev); - r = amdgpu_device_reset_sriov(adev, false); - if (r) - adev->asic_reset_res = r; - } else { - clear_bit(AMDGPU_SKIP_HW_RESET, - &reset_context->flags); - r = amdgpu_do_asic_reset(device_list_handle, - reset_context); - if (r && r == -EAGAIN) - goto retry; - } - - /* - * add reset counter so that the following - * resubmitted job could flush vmid - */ - atomic_inc(&adev->gpu_reset_counter); - continue; - } - - /* got the hw fence, signal finished fence */ - atomic_dec(ring->sched.score); - dma_fence_get(&s_job->s_fence->finished); - dma_fence_signal(&s_job->s_fence->finished); - dma_fence_put(&s_job->s_fence->finished); - - /* remove node from list and free the job */ - spin_lock(&ring->sched.job_list_lock); - list_del_init(&s_job->list); - spin_unlock(&ring->sched.job_list_lock); - ring->sched.ops->free_job(s_job); - } -} - static inline void amdgpu_device_stop_pending_resets(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); @@ -5163,7 +5113,6 @@ static inline void amdgpu_device_stop_pending_resets(struct amdgpu_device *adev) } - /** * amdgpu_device_gpu_recover - reset the asic and recover scheduler * @@ -5186,7 +5135,6 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, int i, r = 0; bool need_emergency_restart = false; bool audio_suspended = false; - int tmp_vram_lost_counter; bool gpu_reset_for_dev_remove = false; gpu_reset_for_dev_remove = @@ -5332,7 +5280,6 @@ retry: /* Rest of adevs pre asic reset from XGMI hive. */ amdgpu_device_stop_pending_resets(tmp_adev); } - tmp_vram_lost_counter = atomic_read(&((adev)->vram_lost_counter)); /* Actual ASIC resets if needed.*/ /* Host driver will handle XGMI hive reset for SRIOV */ if (amdgpu_sriov_vf(adev)) { @@ -5357,32 +5304,16 @@ skip_hw_reset: /* Post ASIC reset for all devs .*/ list_for_each_entry(tmp_adev, device_list_handle, reset_list) { - /* - * Sometimes a later bad compute job can block a good gfx job as gfx - * and compute ring share internal GC HW mutually. We add an additional - * guilty jobs recheck step to find the real guilty job, it synchronously - * submits and pends for the first job being signaled. If it gets timeout, - * we identify it as a real guilty job. - */ - if (amdgpu_gpu_recovery == 2 && - !(tmp_vram_lost_counter < atomic_read(&adev->vram_lost_counter))) - amdgpu_device_recheck_guilty_jobs( - tmp_adev, device_list_handle, reset_context); - for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { struct amdgpu_ring *ring = tmp_adev->rings[i]; if (!ring || !ring->sched.thread) continue; - /* No point to resubmit jobs if we didn't HW reset*/ - if (!tmp_adev->asic_reset_res && !job_signaled) - drm_sched_resubmit_jobs(&ring->sched); - - drm_sched_start(&ring->sched, !tmp_adev->asic_reset_res); + drm_sched_start(&ring->sched, true); } - if (adev->enable_mes) + if (adev->enable_mes && adev->ip_versions[GC_HWIP][0] != IP_VERSION(11, 0, 3)) amdgpu_mes_self_test(tmp_adev); if (!drm_drv_uses_atomic_modeset(adev_to_drm(tmp_adev)) && !job_signaled) { @@ -5421,6 +5352,8 @@ skip_sched_resume: amdgpu_device_resume_display_audio(tmp_adev); amdgpu_device_unset_mp1_state(tmp_adev); + + amdgpu_ras_set_error_query_ready(tmp_adev, true); } recover_end: @@ -5832,8 +5765,6 @@ void amdgpu_pci_resume(struct pci_dev *pdev) if (!ring || !ring->sched.thread) continue; - - drm_sched_resubmit_jobs(&ring->sched); drm_sched_start(&ring->sched, true); } @@ -6024,3 +5955,44 @@ struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev, dma_fence_put(old); return NULL; } + +bool amdgpu_device_has_display_hardware(struct amdgpu_device *adev) +{ + switch (adev->asic_type) { +#ifdef CONFIG_DRM_AMDGPU_SI + case CHIP_HAINAN: +#endif + case CHIP_TOPAZ: + /* chips with no display hardware */ + return false; +#ifdef CONFIG_DRM_AMDGPU_SI + case CHIP_TAHITI: + case CHIP_PITCAIRN: + case CHIP_VERDE: + case CHIP_OLAND: +#endif +#ifdef CONFIG_DRM_AMDGPU_CIK + case CHIP_BONAIRE: + case CHIP_HAWAII: + case CHIP_KAVERI: + case CHIP_KABINI: + case CHIP_MULLINS: +#endif + case CHIP_TONGA: + case CHIP_FIJI: + case CHIP_POLARIS10: + case CHIP_POLARIS11: + case CHIP_POLARIS12: + case CHIP_VEGAM: + case CHIP_CARRIZO: + case CHIP_STONEY: + /* chips with display hardware */ + return true; + default: + /* IP discovery */ + if (!adev->ip_versions[DCE_HWIP][0] || + (adev->harvest_ip_mask & AMD_HARVEST_IP_DMU_MASK)) + return false; + return true; + } +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c index 3993e6134914..6b48178455bc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c @@ -305,8 +305,13 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) goto out; } - if (!amdgpu_discovery_verify_binary_signature(adev->mman.discovery_bin)) { - dev_warn(adev->dev, "get invalid ip discovery binary signature from vram\n"); + if (!amdgpu_discovery_verify_binary_signature(adev->mman.discovery_bin) || amdgpu_discovery == 2) { + /* ignore the discovery binary from vram if discovery=2 in kernel module parameter */ + if (amdgpu_discovery == 2) + dev_info(adev->dev,"force read ip discovery binary from file"); + else + dev_warn(adev->dev, "get invalid ip discovery binary signature from vram\n"); + /* retry read ip discovery binary from file */ r = amdgpu_discovery_read_binary_from_file(adev, adev->mman.discovery_bin); if (r) { @@ -1697,9 +1702,15 @@ static int amdgpu_discovery_set_smu_ip_blocks(struct amdgpu_device *adev) return 0; } +static void amdgpu_discovery_set_sriov_display(struct amdgpu_device *adev) +{ + amdgpu_device_set_sriov_virtual_display(adev); + amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block); +} + static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev) { - if (adev->enable_virtual_display || amdgpu_sriov_vf(adev)) { + if (adev->enable_virtual_display) { amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block); return 0; } @@ -1727,7 +1738,10 @@ static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(3, 1, 6): case IP_VERSION(3, 2, 0): case IP_VERSION(3, 2, 1): - amdgpu_device_ip_block_add(adev, &dm_ip_block); + if (amdgpu_sriov_vf(adev)) + amdgpu_discovery_set_sriov_display(adev); + else + amdgpu_device_ip_block_add(adev, &dm_ip_block); break; default: dev_err(adev->dev, @@ -1740,7 +1754,10 @@ static int amdgpu_discovery_set_display_ip_blocks(struct amdgpu_device *adev) case IP_VERSION(12, 0, 0): case IP_VERSION(12, 0, 1): case IP_VERSION(12, 1, 0): - amdgpu_device_ip_block_add(adev, &dm_ip_block); + if (amdgpu_sriov_vf(adev)) + amdgpu_discovery_set_sriov_display(adev); + else + amdgpu_device_ip_block_add(adev, &dm_ip_block); break; default: dev_err(adev->dev, @@ -2161,6 +2178,7 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) break; case IP_VERSION(10, 3, 1): adev->family = AMDGPU_FAMILY_VGH; + adev->apu_flags |= AMD_APU_IS_VANGOGH; break; case IP_VERSION(10, 3, 3): adev->family = AMDGPU_FAMILY_YC; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index 311a8ea6f065..b22471b3bd63 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -44,6 +44,41 @@ #include <drm/drm_fourcc.h> #include <drm/drm_vblank.h> +/** + * amdgpu_display_hotplug_work_func - work handler for display hotplug event + * + * @work: work struct pointer + * + * This is the hotplug event work handler (all ASICs). + * The work gets scheduled from the IRQ handler if there + * was a hotplug interrupt. It walks through the connector table + * and calls hotplug handler for each connector. After this, it sends + * a DRM hotplug event to alert userspace. + * + * This design approach is required in order to defer hotplug event handling + * from the IRQ handler to a work handler because hotplug handler has to use + * mutexes which cannot be locked in an IRQ handler (since &mutex_lock may + * sleep). + */ +void amdgpu_display_hotplug_work_func(struct work_struct *work) +{ + struct amdgpu_device *adev = container_of(work, struct amdgpu_device, + hotplug_work); + struct drm_device *dev = adev_to_drm(adev); + struct drm_mode_config *mode_config = &dev->mode_config; + struct drm_connector *connector; + struct drm_connector_list_iter iter; + + mutex_lock(&mode_config->mutex); + drm_connector_list_iter_begin(dev, &iter); + drm_for_each_connector_iter(connector, &iter) + amdgpu_connector_hotplug(connector); + drm_connector_list_iter_end(&iter); + mutex_unlock(&mode_config->mutex); + /* Just fire off a uevent and let userspace tell us what to do */ + drm_helper_hpd_irq_event(dev); +} + static int amdgpu_display_framebuffer_init(struct drm_device *dev, struct amdgpu_framebuffer *rfb, const struct drm_mode_fb_cmd2 *mode_cmd, @@ -514,7 +549,7 @@ uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev, */ if ((bo_flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC) && amdgpu_bo_support_uswc(bo_flags) && - amdgpu_device_asic_has_dc_support(adev->asic_type) && + adev->dc_enabled && adev->mode_info.gpu_vm_support) domain |= AMDGPU_GEM_DOMAIN_GTT; #endif @@ -1280,7 +1315,7 @@ int amdgpu_display_modeset_create_props(struct amdgpu_device *adev) "dither", amdgpu_dither_enum_list, sz); - if (amdgpu_device_has_dc_support(adev)) { + if (adev->dc_enabled) { adev->mode_info.abm_level_property = drm_property_create_range(adev_to_drm(adev), 0, "abm level", 0, 4); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h index 560352f7c317..9d19940f73c8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h @@ -35,6 +35,7 @@ #define amdgpu_display_add_encoder(adev, e, s, c) (adev)->mode_info.funcs->add_encoder((adev), (e), (s), (c)) #define amdgpu_display_add_connector(adev, ci, sd, ct, ib, coi, h, r) (adev)->mode_info.funcs->add_connector((adev), (ci), (sd), (ct), (ib), (coi), (h), (r)) +void amdgpu_display_hotplug_work_func(struct work_struct *work); void amdgpu_display_update_priority(struct amdgpu_device *adev); uint32_t amdgpu_display_supported_domains(struct amdgpu_device *adev, uint64_t bo_flags); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c index 7bd8e33b14be..271e30e34d93 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c @@ -328,7 +328,9 @@ amdgpu_dma_buf_create_obj(struct drm_device *dev, struct dma_buf *dma_buf) if (dma_buf->ops == &amdgpu_dmabuf_ops) { struct amdgpu_bo *other = gem_to_amdgpu_bo(dma_buf->priv); - flags |= other->flags & AMDGPU_GEM_CREATE_CPU_GTT_USWC; + flags |= other->flags & (AMDGPU_GEM_CREATE_CPU_GTT_USWC | + AMDGPU_GEM_CREATE_COHERENT | + AMDGPU_GEM_CREATE_UNCACHED); } ret = amdgpu_gem_object_create(adev, dma_buf->size, PAGE_SIZE, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index ca96ee2c2c96..5697d456d7d0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -534,7 +534,7 @@ module_param_named(compute_multipipe, amdgpu_compute_multipipe, int, 0444); * DOC: gpu_recovery (int) * Set to enable GPU recovery mechanism (1 = enable, 0 = disable). The default is -1 (auto, disabled except SRIOV). */ -MODULE_PARM_DESC(gpu_recovery, "Enable GPU recovery mechanism, (2 = advanced tdr mode, 1 = enable, 0 = disable, -1 = auto)"); +MODULE_PARM_DESC(gpu_recovery, "Enable GPU recovery mechanism, (1 = enable, 0 = disable, -1 = auto)"); module_param_named(gpu_recovery, amdgpu_gpu_recovery, int, 0444); /** @@ -1925,9 +1925,6 @@ static const struct pci_device_id pciidlist[] = { {0x1002, 0x73AF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID}, {0x1002, 0x73BF, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_SIENNA_CICHLID}, - /* Van Gogh */ - {0x1002, 0x163F, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_VANGOGH|AMD_IS_APU}, - /* Yellow Carp */ {0x1002, 0x164D, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_YELLOW_CARP|AMD_IS_APU}, {0x1002, 0x1681, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_YELLOW_CARP|AMD_IS_APU}, @@ -2202,7 +2199,8 @@ amdgpu_pci_remove(struct pci_dev *pdev) pm_runtime_forbid(dev->dev); } - if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2)) { + if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2) && + !amdgpu_sriov_vf(adev)) { bool need_to_reset_gpu = false; if (adev->gmc.xgmi.num_physical_nodes > 1) { @@ -2471,7 +2469,7 @@ static int amdgpu_runtime_idle_check_display(struct device *dev) if (ret) return ret; - if (amdgpu_device_has_dc_support(adev)) { + if (adev->dc_enabled) { struct drm_crtc *crtc; drm_for_each_crtc(crtc, drm_dev) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.c index 4d9eb0137f8c..7d2a908438e9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_eeprom.c @@ -79,13 +79,15 @@ * That is, for an I2C EEPROM driver everything is controlled by * the "eeprom_addr". * + * See also top of amdgpu_ras_eeprom.c. + * * P.S. If you need to write, lock and read the Identification Page, * (M24M02-DR device only, which we do not use), change the "7" to * "0xF" in the macro below, and let the client set bit 20 to 1 in * "eeprom_addr", and set A10 to 0 to write into it, and A10 and A1 to * 1 to lock it permanently. */ -#define MAKE_I2C_ADDR(_aa) ((0xA << 3) | (((_aa) >> 16) & 7)) +#define MAKE_I2C_ADDR(_aa) ((0xA << 3) | (((_aa) >> 16) & 0xF)) static int __amdgpu_eeprom_xfer(struct i2c_adapter *i2c_adap, u32 eeprom_addr, u8 *eeprom_buf, u16 buf_size, bool read) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c index e325150879df..2c38ac7bc643 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fru_eeprom.c @@ -29,9 +29,10 @@ #include "amdgpu_fru_eeprom.h" #include "amdgpu_eeprom.h" -#define FRU_EEPROM_MADDR 0x60000 +#define FRU_EEPROM_MADDR_6 0x60000 +#define FRU_EEPROM_MADDR_8 0x80000 -static bool is_fru_eeprom_supported(struct amdgpu_device *adev) +static bool is_fru_eeprom_supported(struct amdgpu_device *adev, u32 *fru_addr) { /* Only server cards have the FRU EEPROM * TODO: See if we can figure this out dynamically instead of @@ -45,6 +46,11 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev) if (amdgpu_sriov_vf(adev)) return false; + /* The default I2C EEPROM address of the FRU. + */ + if (fru_addr) + *fru_addr = FRU_EEPROM_MADDR_8; + /* VBIOS is of the format ###-DXXXYYYY-##. For SKU identification, * we can use just the "DXXX" portion. If there were more models, we * could convert the 3 characters to a hex integer and use a switch @@ -57,21 +63,29 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev) if (strnstr(atom_ctx->vbios_version, "D161", sizeof(atom_ctx->vbios_version)) || strnstr(atom_ctx->vbios_version, "D163", - sizeof(atom_ctx->vbios_version))) + sizeof(atom_ctx->vbios_version))) { + *fru_addr = FRU_EEPROM_MADDR_6; return true; - else + } else { return false; + } case CHIP_ALDEBARAN: - /* All Aldebaran SKUs have the FRU */ + /* All Aldebaran SKUs have an FRU */ + if (!strnstr(atom_ctx->vbios_version, "D673", + sizeof(atom_ctx->vbios_version))) + if (fru_addr) + *fru_addr = FRU_EEPROM_MADDR_6; return true; case CHIP_SIENNA_CICHLID: if (strnstr(atom_ctx->vbios_version, "D603", - sizeof(atom_ctx->vbios_version))) { + sizeof(atom_ctx->vbios_version))) { if (strnstr(atom_ctx->vbios_version, "D603GLXE", - sizeof(atom_ctx->vbios_version))) + sizeof(atom_ctx->vbios_version))) { return false; - else + } else { + *fru_addr = FRU_EEPROM_MADDR_6; return true; + } } else { return false; } @@ -80,41 +94,14 @@ static bool is_fru_eeprom_supported(struct amdgpu_device *adev) } } -static int amdgpu_fru_read_eeprom(struct amdgpu_device *adev, uint32_t addrptr, - unsigned char *buf, size_t buf_size) -{ - int ret; - u8 size; - - ret = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addrptr, buf, 1); - if (ret < 1) { - DRM_WARN("FRU: Failed to get size field"); - return ret; - } - - /* The size returned by the i2c requires subtraction of 0xC0 since the - * size apparently always reports as 0xC0+actual size. - */ - size = buf[0] & 0x3F; - size = min_t(size_t, size, buf_size); - - ret = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addrptr + 1, - buf, size); - if (ret < 1) { - DRM_WARN("FRU: Failed to get data field"); - return ret; - } - - return size; -} - int amdgpu_fru_get_product_info(struct amdgpu_device *adev) { - unsigned char buf[AMDGPU_PRODUCT_NAME_LEN]; - u32 addrptr; + unsigned char buf[8], *pia; + u32 addr, fru_addr; int size, len; + u8 csum; - if (!is_fru_eeprom_supported(adev)) + if (!is_fru_eeprom_supported(adev, &fru_addr)) return 0; /* If algo exists, it means that the i2c_adapter's initialized */ @@ -123,88 +110,102 @@ int amdgpu_fru_get_product_info(struct amdgpu_device *adev) return -ENODEV; } - /* There's a lot of repetition here. This is due to the FRU having - * variable-length fields. To get the information, we have to find the - * size of each field, and then keep reading along and reading along - * until we get all of the data that we want. We use addrptr to track - * the address as we go - */ - - /* The first fields are all of size 1-byte, from 0-7 are offsets that - * contain information that isn't useful to us. - * Bytes 8-a are all 1-byte and refer to the size of the entire struct, - * and the language field, so just start from 0xb, manufacturer size - */ - addrptr = FRU_EEPROM_MADDR + 0xb; - size = amdgpu_fru_read_eeprom(adev, addrptr, buf, sizeof(buf)); - if (size < 1) { - DRM_ERROR("Failed to read FRU Manufacturer, ret:%d", size); - return -EINVAL; + /* Read the IPMI Common header */ + len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, fru_addr, buf, + sizeof(buf)); + if (len != 8) { + DRM_ERROR("Couldn't read the IPMI Common Header: %d", len); + return len < 0 ? len : -EIO; } - /* Increment the addrptr by the size of the field, and 1 due to the - * size field being 1 byte. This pattern continues below. - */ - addrptr += size + 1; - size = amdgpu_fru_read_eeprom(adev, addrptr, buf, sizeof(buf)); - if (size < 1) { - DRM_ERROR("Failed to read FRU product name, ret:%d", size); - return -EINVAL; + if (buf[0] != 1) { + DRM_ERROR("Bad IPMI Common Header version: 0x%02x", buf[0]); + return -EIO; } - len = size; - if (len >= AMDGPU_PRODUCT_NAME_LEN) { - DRM_WARN("FRU Product Name is larger than %d characters. This is likely a mistake", - AMDGPU_PRODUCT_NAME_LEN); - len = AMDGPU_PRODUCT_NAME_LEN - 1; - } - memcpy(adev->product_name, buf, len); - adev->product_name[len] = '\0'; - - addrptr += size + 1; - size = amdgpu_fru_read_eeprom(adev, addrptr, buf, sizeof(buf)); - if (size < 1) { - DRM_ERROR("Failed to read FRU product number, ret:%d", size); - return -EINVAL; + for (csum = 0; len > 0; len--) + csum += buf[len - 1]; + if (csum) { + DRM_ERROR("Bad IPMI Common Header checksum: 0x%02x", csum); + return -EIO; } - len = size; - /* Product number should only be 16 characters. Any more, - * and something could be wrong. Cap it at 16 to be safe - */ - if (len >= sizeof(adev->product_number)) { - DRM_WARN("FRU Product Number is larger than 16 characters. This is likely a mistake"); - len = sizeof(adev->product_number) - 1; - } - memcpy(adev->product_number, buf, len); - adev->product_number[len] = '\0'; + /* Get the offset to the Product Info Area (PIA). */ + addr = buf[4] * 8; + if (!addr) + return 0; - addrptr += size + 1; - size = amdgpu_fru_read_eeprom(adev, addrptr, buf, sizeof(buf)); + /* Get the absolute address to the PIA. */ + addr += fru_addr; - if (size < 1) { - DRM_ERROR("Failed to read FRU product version, ret:%d", size); - return -EINVAL; + /* Read the header of the PIA. */ + len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addr, buf, 3); + if (len != 3) { + DRM_ERROR("Couldn't read the Product Info Area header: %d", len); + return len < 0 ? len : -EIO; } - addrptr += size + 1; - size = amdgpu_fru_read_eeprom(adev, addrptr, buf, sizeof(buf)); + if (buf[0] != 1) { + DRM_ERROR("Bad IPMI Product Info Area version: 0x%02x", buf[0]); + return -EIO; + } - if (size < 1) { - DRM_ERROR("Failed to read FRU serial number, ret:%d", size); - return -EINVAL; + size = buf[1] * 8; + pia = kzalloc(size, GFP_KERNEL); + if (!pia) + return -ENOMEM; + + /* Read the whole PIA. */ + len = amdgpu_eeprom_read(adev->pm.fru_eeprom_i2c_bus, addr, pia, size); + if (len != size) { + kfree(pia); + DRM_ERROR("Couldn't read the Product Info Area: %d", len); + return len < 0 ? len : -EIO; } - len = size; - /* Serial number should only be 16 characters. Any more, - * and something could be wrong. Cap it at 16 to be safe - */ - if (len >= sizeof(adev->serial)) { - DRM_WARN("FRU Serial Number is larger than 16 characters. This is likely a mistake"); - len = sizeof(adev->serial) - 1; + for (csum = 0; size > 0; size--) + csum += pia[size - 1]; + if (csum) { + DRM_ERROR("Bad Product Info Area checksum: 0x%02x", csum); + return -EIO; } - memcpy(adev->serial, buf, len); - adev->serial[len] = '\0'; + /* Now extract useful information from the PIA. + * + * Skip the Manufacturer Name at [3] and go directly to + * the Product Name field. + */ + addr = 3 + 1 + (pia[3] & 0x3F); + if (addr + 1 >= len) + goto Out; + memcpy(adev->product_name, pia + addr + 1, + min_t(size_t, + sizeof(adev->product_name), + pia[addr] & 0x3F)); + adev->product_name[sizeof(adev->product_name) - 1] = '\0'; + + /* Go to the Product Part/Model Number field. */ + addr += 1 + (pia[addr] & 0x3F); + if (addr + 1 >= len) + goto Out; + memcpy(adev->product_number, pia + addr + 1, + min_t(size_t, + sizeof(adev->product_number), + pia[addr] & 0x3F)); + adev->product_number[sizeof(adev->product_number) - 1] = '\0'; + + /* Go to the Product Version field. */ + addr += 1 + (pia[addr] & 0x3F); + + /* Go to the Product Serial Number field. */ + addr += 1 + (pia[addr] & 0x3F); + if (addr + 1 >= len) + goto Out; + memcpy(adev->serial, pia + addr + 1, min_t(size_t, + sizeof(adev->serial), + pia[addr] & 0x3F)); + adev->serial[sizeof(adev->serial) - 1] = '\0'; +Out: + kfree(pia); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index 8ef31d687ef3..a0780a4e3e61 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -38,6 +38,7 @@ #include "amdgpu.h" #include "amdgpu_display.h" #include "amdgpu_dma_buf.h" +#include "amdgpu_hmm.h" #include "amdgpu_xgmi.h" static const struct drm_gem_object_funcs amdgpu_gem_object_funcs; @@ -87,7 +88,7 @@ static void amdgpu_gem_object_free(struct drm_gem_object *gobj) struct amdgpu_bo *robj = gem_to_amdgpu_bo(gobj); if (robj) { - amdgpu_mn_unregister(robj); + amdgpu_hmm_unregister(robj); amdgpu_bo_unref(&robj); } } @@ -378,6 +379,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data, struct amdgpu_device *adev = drm_to_adev(dev); struct drm_amdgpu_gem_userptr *args = data; struct drm_gem_object *gobj; + struct hmm_range *range; struct amdgpu_bo *bo; uint32_t handle; int r; @@ -413,14 +415,13 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data, if (r) goto release_object; - if (args->flags & AMDGPU_GEM_USERPTR_REGISTER) { - r = amdgpu_mn_register(bo, args->addr); - if (r) - goto release_object; - } + r = amdgpu_hmm_register(bo, args->addr); + if (r) + goto release_object; if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE) { - r = amdgpu_ttm_tt_get_user_pages(bo, bo->tbo.ttm->pages); + r = amdgpu_ttm_tt_get_user_pages(bo, bo->tbo.ttm->pages, + &range); if (r) goto release_object; @@ -443,7 +444,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data, user_pages_done: if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE) - amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm); + amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, range); release_object: drm_gem_object_put(gobj); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index 9546adc8a76f..23692e5d4d13 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -583,10 +583,14 @@ void amdgpu_gfx_off_ctrl(struct amdgpu_device *adev, bool enable) if (adev->gfx.gfx_off_req_count == 0 && !adev->gfx.gfx_off_state) { /* If going to s2idle, no need to wait */ - if (adev->in_s0ix) - delay = GFX_OFF_NO_DELAY; - schedule_delayed_work(&adev->gfx.gfx_off_delay_work, + if (adev->in_s0ix) { + if (!amdgpu_dpm_set_powergating_by_smu(adev, + AMD_IP_BLOCK_TYPE_GFX, true)) + adev->gfx.gfx_off_state = true; + } else { + schedule_delayed_work(&adev->gfx.gfx_off_delay_work, delay); + } } } else { if (adev->gfx.gfx_off_req_count == 0) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index 34233a74248c..4365ede42855 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -542,6 +542,7 @@ void amdgpu_gmc_tmz_set(struct amdgpu_device *adev) case IP_VERSION(10, 3, 1): /* YELLOW_CARP*/ case IP_VERSION(10, 3, 3): + case IP_VERSION(11, 0, 1): /* Don't enable it by default yet. */ if (amdgpu_tmz < 1) { @@ -656,7 +657,7 @@ void amdgpu_gmc_get_vbios_allocations(struct amdgpu_device *adev) } if (amdgpu_sriov_vf(adev) || - !amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_DCE)) { + !amdgpu_device_has_display_hardware(adev)) { size = 0; } else { size = amdgpu_gmc_get_vbios_fb_size(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c index b86c0b8252a5..a48ea62b12b0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c @@ -49,9 +49,10 @@ #include "amdgpu.h" #include "amdgpu_amdkfd.h" +#include "amdgpu_hmm.h" /** - * amdgpu_mn_invalidate_gfx - callback to notify about mm change + * amdgpu_hmm_invalidate_gfx - callback to notify about mm change * * @mni: the range (mm) is about to update * @range: details on the invalidation @@ -60,9 +61,9 @@ * Block for operations on BOs to finish and mark pages as accessed and * potentially dirty. */ -static bool amdgpu_mn_invalidate_gfx(struct mmu_interval_notifier *mni, - const struct mmu_notifier_range *range, - unsigned long cur_seq) +static bool amdgpu_hmm_invalidate_gfx(struct mmu_interval_notifier *mni, + const struct mmu_notifier_range *range, + unsigned long cur_seq) { struct amdgpu_bo *bo = container_of(mni, struct amdgpu_bo, notifier); struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); @@ -83,12 +84,12 @@ static bool amdgpu_mn_invalidate_gfx(struct mmu_interval_notifier *mni, return true; } -static const struct mmu_interval_notifier_ops amdgpu_mn_gfx_ops = { - .invalidate = amdgpu_mn_invalidate_gfx, +static const struct mmu_interval_notifier_ops amdgpu_hmm_gfx_ops = { + .invalidate = amdgpu_hmm_invalidate_gfx, }; /** - * amdgpu_mn_invalidate_hsa - callback to notify about mm change + * amdgpu_hmm_invalidate_hsa - callback to notify about mm change * * @mni: the range (mm) is about to update * @range: details on the invalidation @@ -97,9 +98,9 @@ static const struct mmu_interval_notifier_ops amdgpu_mn_gfx_ops = { * We temporarily evict the BO attached to this range. This necessitates * evicting all user-mode queues of the process. */ -static bool amdgpu_mn_invalidate_hsa(struct mmu_interval_notifier *mni, - const struct mmu_notifier_range *range, - unsigned long cur_seq) +static bool amdgpu_hmm_invalidate_hsa(struct mmu_interval_notifier *mni, + const struct mmu_notifier_range *range, + unsigned long cur_seq) { struct amdgpu_bo *bo = container_of(mni, struct amdgpu_bo, notifier); struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); @@ -117,12 +118,12 @@ static bool amdgpu_mn_invalidate_hsa(struct mmu_interval_notifier *mni, return true; } -static const struct mmu_interval_notifier_ops amdgpu_mn_hsa_ops = { - .invalidate = amdgpu_mn_invalidate_hsa, +static const struct mmu_interval_notifier_ops amdgpu_hmm_hsa_ops = { + .invalidate = amdgpu_hmm_invalidate_hsa, }; /** - * amdgpu_mn_register - register a BO for notifier updates + * amdgpu_hmm_register - register a BO for notifier updates * * @bo: amdgpu buffer object * @addr: userptr addr we should monitor @@ -130,25 +131,25 @@ static const struct mmu_interval_notifier_ops amdgpu_mn_hsa_ops = { * Registers a mmu_notifier for the given BO at the specified address. * Returns 0 on success, -ERRNO if anything goes wrong. */ -int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr) +int amdgpu_hmm_register(struct amdgpu_bo *bo, unsigned long addr) { if (bo->kfd_bo) return mmu_interval_notifier_insert(&bo->notifier, current->mm, addr, amdgpu_bo_size(bo), - &amdgpu_mn_hsa_ops); + &amdgpu_hmm_hsa_ops); return mmu_interval_notifier_insert(&bo->notifier, current->mm, addr, amdgpu_bo_size(bo), - &amdgpu_mn_gfx_ops); + &amdgpu_hmm_gfx_ops); } /** - * amdgpu_mn_unregister - unregister a BO for notifier updates + * amdgpu_hmm_unregister - unregister a BO for notifier updates * * @bo: amdgpu buffer object * * Remove any registration of mmu notifier updates from the buffer object. */ -void amdgpu_mn_unregister(struct amdgpu_bo *bo) +void amdgpu_hmm_unregister(struct amdgpu_bo *bo) { if (!bo->notifier.mm) return; @@ -157,10 +158,9 @@ void amdgpu_mn_unregister(struct amdgpu_bo *bo) } int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier, - struct mm_struct *mm, struct page **pages, - uint64_t start, uint64_t npages, - struct hmm_range **phmm_range, bool readonly, - bool mmap_locked, void *owner) + uint64_t start, uint64_t npages, bool readonly, + void *owner, struct page **pages, + struct hmm_range **phmm_range) { struct hmm_range *hmm_range; unsigned long timeout; @@ -193,14 +193,7 @@ int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier, retry: hmm_range->notifier_seq = mmu_interval_read_begin(notifier); - - if (likely(!mmap_locked)) - mmap_read_lock(mm); - r = hmm_range_fault(hmm_range); - - if (likely(!mmap_locked)) - mmap_read_unlock(mm); if (unlikely(r)) { /* * FIXME: This timeout should encompass the retry from diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h index 14a3c1864085..13ed94d3b01b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mn.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h @@ -31,23 +31,22 @@ #include <linux/interval_tree.h> int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier, - struct mm_struct *mm, struct page **pages, - uint64_t start, uint64_t npages, - struct hmm_range **phmm_range, bool readonly, - bool mmap_locked, void *owner); + uint64_t start, uint64_t npages, bool readonly, + void *owner, struct page **pages, + struct hmm_range **phmm_range); int amdgpu_hmm_range_get_pages_done(struct hmm_range *hmm_range); #if defined(CONFIG_HMM_MIRROR) -int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr); -void amdgpu_mn_unregister(struct amdgpu_bo *bo); +int amdgpu_hmm_register(struct amdgpu_bo *bo, unsigned long addr); +void amdgpu_hmm_unregister(struct amdgpu_bo *bo); #else -static inline int amdgpu_mn_register(struct amdgpu_bo *bo, unsigned long addr) +static inline int amdgpu_hmm_register(struct amdgpu_bo *bo, unsigned long addr) { DRM_WARN_ONCE("HMM_MIRROR kernel config option is not enabled, " "add CONFIG_ZONE_DEVICE=y in config file to fix this\n"); return -ENODEV; } -static inline void amdgpu_mn_unregister(struct amdgpu_bo *bo) {} +static inline void amdgpu_hmm_unregister(struct amdgpu_bo *bo) {} #endif #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c index 89011bae7588..a6aef488a822 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c @@ -101,41 +101,6 @@ const char *soc15_ih_clientid_name[] = { }; /** - * amdgpu_hotplug_work_func - work handler for display hotplug event - * - * @work: work struct pointer - * - * This is the hotplug event work handler (all ASICs). - * The work gets scheduled from the IRQ handler if there - * was a hotplug interrupt. It walks through the connector table - * and calls hotplug handler for each connector. After this, it sends - * a DRM hotplug event to alert userspace. - * - * This design approach is required in order to defer hotplug event handling - * from the IRQ handler to a work handler because hotplug handler has to use - * mutexes which cannot be locked in an IRQ handler (since &mutex_lock may - * sleep). - */ -static void amdgpu_hotplug_work_func(struct work_struct *work) -{ - struct amdgpu_device *adev = container_of(work, struct amdgpu_device, - hotplug_work); - struct drm_device *dev = adev_to_drm(adev); - struct drm_mode_config *mode_config = &dev->mode_config; - struct drm_connector *connector; - struct drm_connector_list_iter iter; - - mutex_lock(&mode_config->mutex); - drm_connector_list_iter_begin(dev, &iter); - drm_for_each_connector_iter(connector, &iter) - amdgpu_connector_hotplug(connector); - drm_connector_list_iter_end(&iter); - mutex_unlock(&mode_config->mutex); - /* Just fire off a uevent and let userspace tell us what to do */ - drm_helper_hpd_irq_event(dev); -} - -/** * amdgpu_irq_disable_all - disable *all* interrupts * * @adev: amdgpu device pointer @@ -317,21 +282,6 @@ int amdgpu_irq_init(struct amdgpu_device *adev) } } - if (!amdgpu_device_has_dc_support(adev)) { - if (!adev->enable_virtual_display) - /* Disable vblank IRQs aggressively for power-saving */ - /* XXX: can this be enabled for DC? */ - adev_to_drm(adev)->vblank_disable_immediate = true; - - r = drm_vblank_init(adev_to_drm(adev), adev->mode_info.num_crtc); - if (r) - return r; - - /* Pre-DCE11 */ - INIT_WORK(&adev->hotplug_work, - amdgpu_hotplug_work_func); - } - INIT_WORK(&adev->irq.ih1_work, amdgpu_irq_handle_ih1); INIT_WORK(&adev->irq.ih2_work, amdgpu_irq_handle_ih2); INIT_WORK(&adev->irq.ih_soft_work, amdgpu_irq_handle_ih_soft); @@ -345,11 +295,8 @@ int amdgpu_irq_init(struct amdgpu_device *adev) /* PCI devices require shared interrupts. */ r = request_irq(irq, amdgpu_irq_handler, IRQF_SHARED, adev_to_drm(adev)->driver->name, adev_to_drm(adev)); - if (r) { - if (!amdgpu_device_has_dc_support(adev)) - flush_work(&adev->hotplug_work); + if (r) return r; - } adev->irq.installed = true; adev->irq.irq = irq; adev_to_drm(adev)->max_vblank_count = 0x00ffffff; @@ -366,9 +313,6 @@ void amdgpu_irq_fini_hw(struct amdgpu_device *adev) adev->irq.installed = false; if (adev->irq.msi_enabled) pci_free_irq_vectors(adev->pdev); - - if (!amdgpu_device_has_dc_support(adev)) - flush_work(&adev->hotplug_work); } amdgpu_ih_ring_fini(adev, &adev->irq.ih_soft); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c index de182bfcf85f..6f81ed4fb0d9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c @@ -235,3 +235,20 @@ int amdgpu_jpeg_process_poison_irq(struct amdgpu_device *adev, return 0; } + +void jpeg_set_ras_funcs(struct amdgpu_device *adev) +{ + if (!adev->jpeg.ras) + return; + + amdgpu_ras_register_ras_block(adev, &adev->jpeg.ras->ras_block); + + strcpy(adev->jpeg.ras->ras_block.ras_comm.name, "jpeg"); + adev->jpeg.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__JPEG; + adev->jpeg.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__POISON; + adev->jpeg.ras_if = &adev->jpeg.ras->ras_block.ras_comm; + + /* If don't define special ras_late_init function, use default ras_late_init */ + if (!adev->jpeg.ras->ras_block.ras_late_init) + adev->jpeg.ras->ras_block.ras_late_init = amdgpu_ras_block_late_init; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h index 635dca59a70a..e8ca3e32ad52 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.h @@ -72,5 +72,6 @@ int amdgpu_jpeg_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout); int amdgpu_jpeg_process_poison_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry); +void jpeg_set_ras_funcs(struct amdgpu_device *adev); #endif /*__AMDGPU_JPEG_H__*/ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index b495eff635a3..ba348046fcec 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -338,11 +338,17 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info, fw_info->feature = adev->psp.cap_feature_version; break; case AMDGPU_INFO_FW_MES_KIQ: - fw_info->ver = adev->mes.ucode_fw_version[0]; - fw_info->feature = 0; + fw_info->ver = adev->mes.kiq_version & AMDGPU_MES_VERSION_MASK; + fw_info->feature = (adev->mes.kiq_version & AMDGPU_MES_FEAT_VERSION_MASK) + >> AMDGPU_MES_FEAT_VERSION_SHIFT; break; case AMDGPU_INFO_FW_MES: - fw_info->ver = adev->mes.ucode_fw_version[1]; + fw_info->ver = adev->mes.sched_version & AMDGPU_MES_VERSION_MASK; + fw_info->feature = (adev->mes.sched_version & AMDGPU_MES_FEAT_VERSION_MASK) + >> AMDGPU_MES_FEAT_VERSION_SHIFT; + break; + case AMDGPU_INFO_FW_IMU: + fw_info->ver = adev->gfx.imu_fw_version; fw_info->feature = 0; break; default: @@ -791,7 +797,7 @@ int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) dev_info->ids_flags = 0; if (adev->flags & AMD_IS_APU) dev_info->ids_flags |= AMDGPU_IDS_FLAGS_FUSION; - if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) + if (amdgpu_mcbp) dev_info->ids_flags |= AMDGPU_IDS_FLAGS_PREEMPTION; if (amdgpu_is_tmz(adev)) dev_info->ids_flags |= AMDGPU_IDS_FLAGS_TMZ; @@ -1167,7 +1173,7 @@ int amdgpu_driver_open_kms(struct drm_device *dev, struct drm_file *file_priv) goto error_vm; } - if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) { + if (amdgpu_mcbp) { uint64_t csa_addr = amdgpu_csa_vaddr(adev) & AMDGPU_GMC_HOLE_MASK; r = amdgpu_map_static_csa(adev, &fpriv->vm, adev->virt.csa_obj, @@ -1231,7 +1237,7 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev, if (amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_VCE) != NULL) amdgpu_vce_free_handles(adev, file_priv); - if (amdgpu_mcbp || amdgpu_sriov_vf(adev)) { + if (amdgpu_mcbp) { /* TODO: how to handle reserve failure */ BUG_ON(amdgpu_bo_reserve(adev->virt.csa_obj, true)); amdgpu_vm_bo_del(adev, fpriv->csa_va); @@ -1521,6 +1527,15 @@ static int amdgpu_debugfs_firmware_info_show(struct seq_file *m, void *unused) fw_info.feature, fw_info.ver); } + /* IMU */ + query_fw.fw_type = AMDGPU_INFO_FW_IMU; + query_fw.index = 0; + ret = amdgpu_firmware_info(&fw_info, &query_fw, adev); + if (ret) + return ret; + seq_printf(m, "IMU feature version: %u, firmware version: 0x%08x\n", + fw_info.feature, fw_info.ver); + /* PSP SOS */ query_fw.fw_type = AMDGPU_INFO_FW_SOS; ret = amdgpu_firmware_info(&fw_info, &query_fw, adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h index ad980f4b66e1..97c05d08a551 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h @@ -91,14 +91,12 @@ struct amdgpu_mes { struct amdgpu_bo *ucode_fw_obj[AMDGPU_MAX_MES_PIPES]; uint64_t ucode_fw_gpu_addr[AMDGPU_MAX_MES_PIPES]; uint32_t *ucode_fw_ptr[AMDGPU_MAX_MES_PIPES]; - uint32_t ucode_fw_version[AMDGPU_MAX_MES_PIPES]; uint64_t uc_start_addr[AMDGPU_MAX_MES_PIPES]; /* mes ucode data */ struct amdgpu_bo *data_fw_obj[AMDGPU_MAX_MES_PIPES]; uint64_t data_fw_gpu_addr[AMDGPU_MAX_MES_PIPES]; uint32_t *data_fw_ptr[AMDGPU_MAX_MES_PIPES]; - uint32_t data_fw_version[AMDGPU_MAX_MES_PIPES]; uint64_t data_start_addr[AMDGPU_MAX_MES_PIPES]; /* eop gpu obj */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index effa7df3ddbf..98dbf4e5aae9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -52,6 +52,32 @@ static int psp_load_smu_fw(struct psp_context *psp); static int psp_rap_terminate(struct psp_context *psp); static int psp_securedisplay_terminate(struct psp_context *psp); +static int psp_ring_init(struct psp_context *psp, + enum psp_ring_type ring_type) +{ + int ret = 0; + struct psp_ring *ring; + struct amdgpu_device *adev = psp->adev; + + ring = &psp->km_ring; + + ring->ring_type = ring_type; + + /* allocate 4k Page of Local Frame Buffer memory for ring */ + ring->ring_size = 0x1000; + ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE, + AMDGPU_GEM_DOMAIN_VRAM, + &adev->firmware.rbuf, + &ring->ring_mem_mc_addr, + (void **)&ring->ring_mem); + if (ret) { + ring->ring_size = 0; + return ret; + } + + return 0; +} + /* * Due to DF Cstate management centralized to PMFW, the firmware * loading sequence will be updated as below: @@ -988,6 +1014,8 @@ int psp_ta_unload(struct psp_context *psp, struct ta_context *context) ret = psp_cmd_submit_buf(psp, NULL, cmd, psp->fence_buf_mc_addr); + context->resp_status = cmd->resp.status; + release_psp_cmd_buf(psp); return ret; @@ -1069,42 +1097,6 @@ int psp_ta_init_shared_buf(struct psp_context *psp, &mem_ctx->shared_buf); } -static void psp_prep_ta_invoke_indirect_cmd_buf(struct psp_gfx_cmd_resp *cmd, - uint32_t ta_cmd_id, - struct ta_context *context) -{ - cmd->cmd_id = GFX_CMD_ID_INVOKE_CMD; - cmd->cmd.cmd_invoke_cmd.session_id = context->session_id; - cmd->cmd.cmd_invoke_cmd.ta_cmd_id = ta_cmd_id; - - cmd->cmd.cmd_invoke_cmd.buf.num_desc = 1; - cmd->cmd.cmd_invoke_cmd.buf.total_size = context->mem_context.shared_mem_size; - cmd->cmd.cmd_invoke_cmd.buf.buf_desc[0].buf_size = context->mem_context.shared_mem_size; - cmd->cmd.cmd_invoke_cmd.buf.buf_desc[0].buf_phy_addr_lo = - lower_32_bits(context->mem_context.shared_mc_addr); - cmd->cmd.cmd_invoke_cmd.buf.buf_desc[0].buf_phy_addr_hi = - upper_32_bits(context->mem_context.shared_mc_addr); -} - -int psp_ta_invoke_indirect(struct psp_context *psp, - uint32_t ta_cmd_id, - struct ta_context *context) -{ - int ret; - struct psp_gfx_cmd_resp *cmd = acquire_psp_cmd_buf(psp); - - psp_prep_ta_invoke_indirect_cmd_buf(cmd, ta_cmd_id, context); - - ret = psp_cmd_submit_buf(psp, NULL, cmd, - psp->fence_buf_mc_addr); - - context->resp_status = cmd->resp.status; - - release_psp_cmd_buf(psp); - - return ret; -} - static void psp_prep_ta_invoke_cmd_buf(struct psp_gfx_cmd_resp *cmd, uint32_t ta_cmd_id, uint32_t session_id) @@ -1547,7 +1539,7 @@ int psp_ras_terminate(struct psp_context *psp) return ret; } -static int psp_ras_initialize(struct psp_context *psp) +int psp_ras_initialize(struct psp_context *psp) { int ret; uint32_t boot_cfg = 0xFF; @@ -1610,7 +1602,7 @@ static int psp_ras_initialize(struct psp_context *psp) psp->ras_context.context.mem_context.shared_mem_size = PSP_RAS_SHARED_MEM_SIZE; psp->ras_context.context.ta_load_type = GFX_CMD_ID_LOAD_TA; - if (!psp->ras_context.context.initialized) { + if (!psp->ras_context.context.mem_context.shared_buf) { ret = psp_ta_init_shared_buf(psp, &psp->ras_context.context.mem_context); if (ret) return ret; @@ -1631,7 +1623,9 @@ static int psp_ras_initialize(struct psp_context *psp) else { if (ras_cmd->ras_status) dev_warn(psp->adev->dev, "RAS Init Status: 0x%X\n", ras_cmd->ras_status); - amdgpu_ras_fini(psp->adev); + + /* fail to load RAS TA */ + psp->ras_context.context.initialized = false; } return ret; @@ -1938,10 +1932,15 @@ static int psp_securedisplay_initialize(struct psp_context *psp) } else return ret; + mutex_lock(&psp->securedisplay_context.mutex); + psp_prep_securedisplay_cmd_buf(psp, &securedisplay_cmd, TA_SECUREDISPLAY_COMMAND__QUERY_TA); ret = psp_securedisplay_invoke(psp, TA_SECUREDISPLAY_COMMAND__QUERY_TA); + + mutex_unlock(&psp->securedisplay_context.mutex); + if (ret) { psp_securedisplay_terminate(psp); /* free securedisplay shared memory */ @@ -1990,12 +1989,8 @@ int psp_securedisplay_invoke(struct psp_context *psp, uint32_t ta_cmd_id) ta_cmd_id != TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC) return -EINVAL; - mutex_lock(&psp->securedisplay_context.mutex); - ret = psp_ta_invoke(psp, ta_cmd_id, &psp->securedisplay_context.context); - mutex_unlock(&psp->securedisplay_context.mutex); - return ret; } /* SECUREDISPLAY end */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h index 58ce3ebb446c..cf4f60c66122 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h @@ -118,7 +118,6 @@ struct psp_funcs int (*bootloader_load_dbg_drv)(struct psp_context *psp); int (*bootloader_load_ras_drv)(struct psp_context *psp); int (*bootloader_load_sos)(struct psp_context *psp); - int (*ring_init)(struct psp_context *psp, enum psp_ring_type ring_type); int (*ring_create)(struct psp_context *psp, enum psp_ring_type ring_type); int (*ring_stop)(struct psp_context *psp, @@ -136,6 +135,12 @@ struct psp_funcs int (*vbflash_stat)(struct psp_context *psp); }; +struct ta_funcs { + int (*fn_ta_initialize)(struct psp_context *psp); + int (*fn_ta_invoke)(struct psp_context *psp, uint32_t ta_cmd_id); + int (*fn_ta_terminate)(struct psp_context *psp); +}; + #define AMDGPU_XGMI_MAX_CONNECTED_NODES 64 struct psp_xgmi_node_info { uint64_t node_id; @@ -309,6 +314,7 @@ struct psp_context struct psp_gfx_cmd_resp *cmd; const struct psp_funcs *funcs; + const struct ta_funcs *ta_funcs; /* firmware buffer */ struct amdgpu_bo *fw_pri_bo; @@ -389,7 +395,6 @@ struct amdgpu_psp_funcs { }; -#define psp_ring_init(psp, type) (psp)->funcs->ring_init((psp), (type)) #define psp_ring_create(psp, type) (psp)->funcs->ring_create((psp), (type)) #define psp_ring_stop(psp, type) (psp)->funcs->ring_stop((psp), (type)) #define psp_ring_destroy(psp, type) ((psp)->funcs->ring_destroy((psp), (type))) @@ -463,9 +468,6 @@ int psp_ta_load(struct psp_context *psp, struct ta_context *context); int psp_ta_invoke(struct psp_context *psp, uint32_t ta_cmd_id, struct ta_context *context); -int psp_ta_invoke_indirect(struct psp_context *psp, - uint32_t ta_cmd_id, - struct ta_context *context); int psp_xgmi_initialize(struct psp_context *psp, bool set_extended_data, bool load_ta); int psp_xgmi_terminate(struct psp_context *psp); @@ -479,7 +481,7 @@ int psp_xgmi_get_topology_info(struct psp_context *psp, int psp_xgmi_set_topology_info(struct psp_context *psp, int number_devices, struct psp_xgmi_topology_info *topology); - +int psp_ras_initialize(struct psp_context *psp); int psp_ras_invoke(struct psp_context *psp, uint32_t ta_cmd_id); int psp_ras_enable_features(struct psp_context *psp, union ta_ras_cmd_input *info, bool enable); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c index 0988e00612e5..468a67b302d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.c @@ -41,30 +41,46 @@ static uint32_t get_bin_version(const uint8_t *bin) return hdr->ucode_version; } -static void prep_ta_mem_context(struct psp_context *psp, - struct ta_context *context, +static int prep_ta_mem_context(struct ta_mem_context *mem_context, uint8_t *shared_buf, uint32_t shared_buf_len) { - context->mem_context.shared_mem_size = PAGE_ALIGN(shared_buf_len); - psp_ta_init_shared_buf(psp, &context->mem_context); + if (mem_context->shared_mem_size < shared_buf_len) + return -EINVAL; + memset(mem_context->shared_buf, 0, mem_context->shared_mem_size); + memcpy((void *)mem_context->shared_buf, shared_buf, shared_buf_len); - memcpy((void *)context->mem_context.shared_buf, shared_buf, shared_buf_len); + return 0; } static bool is_ta_type_valid(enum ta_type_id ta_type) { - bool ret = false; + switch (ta_type) { + case TA_TYPE_RAS: + return true; + default: + return false; + } +} + +static const struct ta_funcs ras_ta_funcs = { + .fn_ta_initialize = psp_ras_initialize, + .fn_ta_invoke = psp_ras_invoke, + .fn_ta_terminate = psp_ras_terminate +}; +static void set_ta_context_funcs(struct psp_context *psp, + enum ta_type_id ta_type, + struct ta_context **pcontext) +{ switch (ta_type) { case TA_TYPE_RAS: - ret = true; + *pcontext = &psp->ras_context.context; + psp->ta_funcs = &ras_ta_funcs; break; default: break; } - - return ret; } static const struct file_operations ta_load_debugfs_fops = { @@ -85,8 +101,7 @@ static const struct file_operations ta_invoke_debugfs_fops = { .owner = THIS_MODULE }; - -/** +/* * DOC: AMDGPU TA debugfs interfaces * * Three debugfs interfaces can be opened by a program to @@ -111,15 +126,18 @@ static const struct file_operations ta_invoke_debugfs_fops = { * * - For TA invoke debugfs interface: * Transmit buffer: + * - TA type (4bytes) * - TA ID (4bytes) * - TA CMD ID (4bytes) - * - TA shard buf length (4bytes) + * - TA shard buf length + * (4bytes, value not beyond TA shared memory size) * - TA shared buf * Receive buffer: * - TA shared buf * * - For TA unload debugfs interface: * Transmit buffer: + * - TA type (4bytes) * - TA ID (4bytes) */ @@ -131,59 +149,92 @@ static ssize_t ta_if_load_debugfs_write(struct file *fp, const char *buf, size_t uint32_t copy_pos = 0; int ret = 0; - struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(fp)->i_private; - struct psp_context *psp = &adev->psp; - struct ta_context context = {0}; + struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(fp)->i_private; + struct psp_context *psp = &adev->psp; + struct ta_context *context = NULL; if (!buf) return -EINVAL; ret = copy_from_user((void *)&ta_type, &buf[copy_pos], sizeof(uint32_t)); if (ret || (!is_ta_type_valid(ta_type))) - return -EINVAL; + return -EFAULT; copy_pos += sizeof(uint32_t); ret = copy_from_user((void *)&ta_bin_len, &buf[copy_pos], sizeof(uint32_t)); if (ret) - return -EINVAL; + return -EFAULT; copy_pos += sizeof(uint32_t); ta_bin = kzalloc(ta_bin_len, GFP_KERNEL); if (!ta_bin) - ret = -ENOMEM; + return -ENOMEM; if (copy_from_user((void *)ta_bin, &buf[copy_pos], ta_bin_len)) { ret = -EFAULT; goto err_free_bin; } - ret = psp_ras_terminate(psp); - if (ret) { - dev_err(adev->dev, "Failed to unload embedded RAS TA\n"); + /* Set TA context and functions */ + set_ta_context_funcs(psp, ta_type, &context); + + if (!psp->ta_funcs || !psp->ta_funcs->fn_ta_terminate) { + dev_err(adev->dev, "Unsupported function to terminate TA\n"); + ret = -EOPNOTSUPP; goto err_free_bin; } - context.ta_type = ta_type; - context.ta_load_type = GFX_CMD_ID_LOAD_TA; - context.bin_desc.fw_version = get_bin_version(ta_bin); - context.bin_desc.size_bytes = ta_bin_len; - context.bin_desc.start_addr = ta_bin; + /* + * Allocate TA shared buf in case shared buf was freed + * due to loading TA failed before. + */ + if (!context->mem_context.shared_buf) { + ret = psp_ta_init_shared_buf(psp, &context->mem_context); + if (ret) { + ret = -ENOMEM; + goto err_free_bin; + } + } + + ret = psp_fn_ta_terminate(psp); + if (ret || context->resp_status) { + dev_err(adev->dev, + "Failed to unload embedded TA (%d) and status (0x%X)\n", + ret, context->resp_status); + if (!ret) + ret = -EINVAL; + goto err_free_ta_shared_buf; + } + + /* Prepare TA context for TA initialization */ + context->ta_type = ta_type; + context->bin_desc.fw_version = get_bin_version(ta_bin); + context->bin_desc.size_bytes = ta_bin_len; + context->bin_desc.start_addr = ta_bin; - ret = psp_ta_load(psp, &context); + if (!psp->ta_funcs->fn_ta_initialize) { + dev_err(adev->dev, "Unsupported function to initialize TA\n"); + ret = -EOPNOTSUPP; + goto err_free_ta_shared_buf; + } - if (ret || context.resp_status) { - dev_err(adev->dev, "TA load via debugfs failed (%d) status %d\n", - ret, context.resp_status); + ret = psp_fn_ta_initialize(psp); + if (ret || context->resp_status) { + dev_err(adev->dev, "Failed to load TA via debugfs (%d) and status (0x%X)\n", + ret, context->resp_status); if (!ret) ret = -EINVAL; - goto err_free_bin; + goto err_free_ta_shared_buf; } - context.initialized = true; - if (copy_to_user((char *)buf, (void *)&context.session_id, sizeof(uint32_t))) + if (copy_to_user((char *)buf, (void *)&context->session_id, sizeof(uint32_t))) ret = -EFAULT; +err_free_ta_shared_buf: + /* Only free TA shared buf when returns error code */ + if (ret && context->mem_context.shared_buf) + psp_ta_free_shared_buf(&context->mem_context); err_free_bin: kfree(ta_bin); @@ -192,58 +243,85 @@ err_free_bin: static ssize_t ta_if_unload_debugfs_write(struct file *fp, const char *buf, size_t len, loff_t *off) { - uint32_t ta_id = 0; - int ret = 0; + uint32_t ta_type = 0; + uint32_t ta_id = 0; + uint32_t copy_pos = 0; + int ret = 0; - struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(fp)->i_private; - struct psp_context *psp = &adev->psp; - struct ta_context context = {0}; + struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(fp)->i_private; + struct psp_context *psp = &adev->psp; + struct ta_context *context = NULL; if (!buf) return -EINVAL; - ret = copy_from_user((void *)&ta_id, buf, sizeof(uint32_t)); + ret = copy_from_user((void *)&ta_type, &buf[copy_pos], sizeof(uint32_t)); + if (ret || (!is_ta_type_valid(ta_type))) + return -EFAULT; + + copy_pos += sizeof(uint32_t); + + ret = copy_from_user((void *)&ta_id, &buf[copy_pos], sizeof(uint32_t)); if (ret) - return -EINVAL; + return -EFAULT; - context.session_id = ta_id; + set_ta_context_funcs(psp, ta_type, &context); + context->session_id = ta_id; - ret = psp_ta_unload(psp, &context); - if (!ret) - context.initialized = false; + if (!psp->ta_funcs || !psp->ta_funcs->fn_ta_terminate) { + dev_err(adev->dev, "Unsupported function to terminate TA\n"); + return -EOPNOTSUPP; + } + + ret = psp_fn_ta_terminate(psp); + if (ret || context->resp_status) { + dev_err(adev->dev, "Failed to unload TA via debugfs (%d) and status (0x%X)\n", + ret, context->resp_status); + if (!ret) + ret = -EINVAL; + } + + if (context->mem_context.shared_buf) + psp_ta_free_shared_buf(&context->mem_context); return ret; } static ssize_t ta_if_invoke_debugfs_write(struct file *fp, const char *buf, size_t len, loff_t *off) { + uint32_t ta_type = 0; uint32_t ta_id = 0; uint32_t cmd_id = 0; uint32_t shared_buf_len = 0; - uint8_t *shared_buf = NULL; + uint8_t *shared_buf = NULL; uint32_t copy_pos = 0; int ret = 0; - struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(fp)->i_private; - struct psp_context *psp = &adev->psp; - struct ta_context context = {0}; + struct amdgpu_device *adev = (struct amdgpu_device *)file_inode(fp)->i_private; + struct psp_context *psp = &adev->psp; + struct ta_context *context = NULL; if (!buf) return -EINVAL; + ret = copy_from_user((void *)&ta_type, &buf[copy_pos], sizeof(uint32_t)); + if (ret) + return -EFAULT; + copy_pos += sizeof(uint32_t); + ret = copy_from_user((void *)&ta_id, &buf[copy_pos], sizeof(uint32_t)); if (ret) - return -EINVAL; + return -EFAULT; copy_pos += sizeof(uint32_t); ret = copy_from_user((void *)&cmd_id, &buf[copy_pos], sizeof(uint32_t)); if (ret) - return -EINVAL; + return -EFAULT; copy_pos += sizeof(uint32_t); ret = copy_from_user((void *)&shared_buf_len, &buf[copy_pos], sizeof(uint32_t)); if (ret) - return -EINVAL; + return -EFAULT; copy_pos += sizeof(uint32_t); shared_buf = kzalloc(shared_buf_len, GFP_KERNEL); @@ -254,26 +332,39 @@ static ssize_t ta_if_invoke_debugfs_write(struct file *fp, const char *buf, size goto err_free_shared_buf; } - context.session_id = ta_id; + set_ta_context_funcs(psp, ta_type, &context); + + if (!context->initialized) { + dev_err(adev->dev, "TA is not initialized\n"); + ret = -EINVAL; + goto err_free_shared_buf; + } + + if (!psp->ta_funcs || !psp->ta_funcs->fn_ta_invoke) { + dev_err(adev->dev, "Unsupported function to invoke TA\n"); + ret = -EOPNOTSUPP; + goto err_free_shared_buf; + } - prep_ta_mem_context(psp, &context, shared_buf, shared_buf_len); + context->session_id = ta_id; - ret = psp_ta_invoke_indirect(psp, cmd_id, &context); + ret = prep_ta_mem_context(&context->mem_context, shared_buf, shared_buf_len); + if (ret) + goto err_free_shared_buf; - if (ret || context.resp_status) { - dev_err(adev->dev, "TA invoke via debugfs failed (%d) status %d\n", - ret, context.resp_status); - if (!ret) + ret = psp_fn_ta_invoke(psp, cmd_id); + if (ret || context->resp_status) { + dev_err(adev->dev, "Failed to invoke TA via debugfs (%d) and status (0x%X)\n", + ret, context->resp_status); + if (!ret) { ret = -EINVAL; - goto err_free_ta_shared_buf; + goto err_free_shared_buf; + } } - if (copy_to_user((char *)buf, context.mem_context.shared_buf, shared_buf_len)) + if (copy_to_user((char *)buf, context->mem_context.shared_buf, shared_buf_len)) ret = -EFAULT; -err_free_ta_shared_buf: - psp_ta_free_shared_buf(&context.mem_context); - err_free_shared_buf: kfree(shared_buf); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.h index cfc1542f63ef..14cd1c81c3e6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp_ta.h @@ -24,6 +24,11 @@ #ifndef __AMDGPU_PSP_TA_H__ #define __AMDGPU_PSP_TA_H__ +/* Calling set_ta_context_funcs is required before using the following macros */ +#define psp_fn_ta_initialize(psp) ((psp)->ta_funcs->fn_ta_initialize((psp))) +#define psp_fn_ta_invoke(psp, ta_cmd_id) ((psp)->ta_funcs->fn_ta_invoke((psp), (ta_cmd_id))) +#define psp_fn_ta_terminate(psp) ((psp)->ta_funcs->fn_ta_terminate((psp))) + void amdgpu_ta_if_debugfs_init(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index a4b47e1bd111..077404a9c935 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1561,7 +1561,6 @@ static void amdgpu_ras_interrupt_poison_consumption_handler(struct ras_manager * { bool poison_stat = false; struct amdgpu_device *adev = obj->adev; - struct ras_err_data err_data = {0, 0, 0, NULL}; struct amdgpu_ras_block_object *block_obj = amdgpu_ras_get_ras_block(adev, obj->head.block, 0); @@ -1584,7 +1583,7 @@ static void amdgpu_ras_interrupt_poison_consumption_handler(struct ras_manager * } if (!adev->gmc.xgmi.connected_to_cpu) - amdgpu_umc_poison_handler(adev, &err_data, false); + amdgpu_umc_poison_handler(adev, false); if (block_obj->hw_ops->handle_poison_consumption) poison_stat = block_obj->hw_ops->handle_poison_consumption(adev); @@ -1949,7 +1948,12 @@ static void amdgpu_ras_do_recovery(struct work_struct *work) reset_context.method = AMD_RESET_METHOD_NONE; reset_context.reset_req_dev = adev; - clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags); + + /* Perform full reset in fatal error mode */ + if (!amdgpu_ras_is_poison_mode_supported(ras->adev)) + set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags); + else + clear_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags); amdgpu_device_gpu_recover(ras->adev, NULL, &reset_context); } @@ -2344,7 +2348,8 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev) adev->ras_hw_enabled |= ~(1 << AMDGPU_RAS_BLOCK__UMC | 1 << AMDGPU_RAS_BLOCK__DF); - if (adev->ip_versions[VCN_HWIP][0] == IP_VERSION(2, 6, 0)) + if (adev->ip_versions[VCN_HWIP][0] == IP_VERSION(2, 6, 0) || + adev->ip_versions[VCN_HWIP][0] == IP_VERSION(4, 0, 0)) adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN | 1 << AMDGPU_RAS_BLOCK__JPEG); else @@ -2848,7 +2853,6 @@ static int amdgpu_bad_page_notifier(struct notifier_block *nb, struct amdgpu_device *adev = NULL; uint32_t gpu_id = 0; uint32_t umc_inst = 0, ch_inst = 0; - struct ras_err_data err_data = {0, 0, 0, NULL}; /* * If the error was generated in UMC_V2, which belongs to GPU UMCs, @@ -2887,31 +2891,10 @@ static int amdgpu_bad_page_notifier(struct notifier_block *nb, dev_info(adev->dev, "Uncorrectable error detected in UMC inst: %d, chan_idx: %d", umc_inst, ch_inst); - err_data.err_addr = - kcalloc(adev->umc.max_ras_err_cnt_per_query, - sizeof(struct eeprom_table_record), GFP_KERNEL); - if (!err_data.err_addr) { - dev_warn(adev->dev, - "Failed to alloc memory for umc error record in mca notifier!\n"); + if (!amdgpu_umc_page_retirement_mca(adev, m->addr, ch_inst, umc_inst)) + return NOTIFY_OK; + else return NOTIFY_DONE; - } - - /* - * Translate UMC channel address to Physical address - */ - if (adev->umc.ras && - adev->umc.ras->convert_ras_error_address) - adev->umc.ras->convert_ras_error_address(adev, - &err_data, m->addr, ch_inst, umc_inst); - - if (amdgpu_bad_page_threshold != 0) { - amdgpu_ras_add_bad_pages(adev, err_data.err_addr, - err_data.err_addr_cnt); - amdgpu_ras_save_bad_pages(adev); - } - - kfree(err_data.err_addr); - return NOTIFY_OK; } static struct notifier_block amdgpu_bad_page_nb = { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index 84c241b9a2a1..2d9f3f4cd79e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -33,12 +33,29 @@ #include "amdgpu_reset.h" -#define EEPROM_I2C_MADDR_VEGA20 0x0 -#define EEPROM_I2C_MADDR_ARCTURUS 0x40000 -#define EEPROM_I2C_MADDR_ARCTURUS_D342 0x0 -#define EEPROM_I2C_MADDR_SIENNA_CICHLID 0x0 -#define EEPROM_I2C_MADDR_ALDEBARAN 0x0 -#define EEPROM_I2C_MADDR_SMU_13_0_0 (0x54UL << 16) +/* These are memory addresses as would be seen by one or more EEPROM + * chips strung on the I2C bus, usually by manipulating pins 1-3 of a + * set of EEPROM devices. They form a continuous memory space. + * + * The I2C device address includes the device type identifier, 1010b, + * which is a reserved value and indicates that this is an I2C EEPROM + * device. It also includes the top 3 bits of the 19 bit EEPROM memory + * address, namely bits 18, 17, and 16. This makes up the 7 bit + * address sent on the I2C bus with bit 0 being the direction bit, + * which is not represented here, and sent by the hardware directly. + * + * For instance, + * 50h = 1010000b => device type identifier 1010b, bits 18:16 = 000b, address 0. + * 54h = 1010100b => --"--, bits 18:16 = 100b, address 40000h. + * 56h = 1010110b => --"--, bits 18:16 = 110b, address 60000h. + * Depending on the size of the I2C EEPROM device(s), bits 18:16 may + * address memory in a device or a device on the I2C bus, depending on + * the status of pins 1-3. See top of amdgpu_eeprom.c. + * + * The RAS table lives either at address 0 or address 40000h of EEPROM. + */ +#define EEPROM_I2C_MADDR_0 0x0 +#define EEPROM_I2C_MADDR_4 0x40000 /* * The 2 macros bellow represent the actual size in bytes that @@ -90,6 +107,16 @@ static bool __is_ras_eeprom_supported(struct amdgpu_device *adev) { + if (adev->asic_type == CHIP_IP_DISCOVERY) { + switch (adev->ip_versions[MP1_HWIP][0]) { + case IP_VERSION(13, 0, 0): + case IP_VERSION(13, 0, 10): + return true; + default: + return false; + } + } + return adev->asic_type == CHIP_VEGA20 || adev->asic_type == CHIP_ARCTURUS || adev->asic_type == CHIP_SIENNA_CICHLID || @@ -107,16 +134,30 @@ static bool __get_eeprom_i2c_addr_arct(struct amdgpu_device *adev, if (strnstr(atom_ctx->vbios_version, "D342", sizeof(atom_ctx->vbios_version))) - control->i2c_address = EEPROM_I2C_MADDR_ARCTURUS_D342; + control->i2c_address = EEPROM_I2C_MADDR_0; else - control->i2c_address = EEPROM_I2C_MADDR_ARCTURUS; + control->i2c_address = EEPROM_I2C_MADDR_4; return true; } +static bool __get_eeprom_i2c_addr_ip_discovery(struct amdgpu_device *adev, + struct amdgpu_ras_eeprom_control *control) +{ + switch (adev->ip_versions[MP1_HWIP][0]) { + case IP_VERSION(13, 0, 0): + case IP_VERSION(13, 0, 10): + control->i2c_address = EEPROM_I2C_MADDR_4; + return true; + default: + return false; + } +} + static bool __get_eeprom_i2c_addr(struct amdgpu_device *adev, struct amdgpu_ras_eeprom_control *control) { + struct atom_context *atom_ctx = adev->mode_info.atom_context; u8 i2c_addr; if (!control) @@ -139,27 +180,34 @@ static bool __get_eeprom_i2c_addr(struct amdgpu_device *adev, switch (adev->asic_type) { case CHIP_VEGA20: - control->i2c_address = EEPROM_I2C_MADDR_VEGA20; + control->i2c_address = EEPROM_I2C_MADDR_0; break; case CHIP_ARCTURUS: return __get_eeprom_i2c_addr_arct(adev, control); case CHIP_SIENNA_CICHLID: - control->i2c_address = EEPROM_I2C_MADDR_SIENNA_CICHLID; + control->i2c_address = EEPROM_I2C_MADDR_0; break; case CHIP_ALDEBARAN: - control->i2c_address = EEPROM_I2C_MADDR_ALDEBARAN; + if (strnstr(atom_ctx->vbios_version, "D673", + sizeof(atom_ctx->vbios_version))) + control->i2c_address = EEPROM_I2C_MADDR_4; + else + control->i2c_address = EEPROM_I2C_MADDR_0; break; + case CHIP_IP_DISCOVERY: + return __get_eeprom_i2c_addr_ip_discovery(adev, control); + default: return false; } switch (adev->ip_versions[MP1_HWIP][0]) { case IP_VERSION(13, 0, 0): - control->i2c_address = EEPROM_I2C_MADDR_SMU_13_0_0; + control->i2c_address = EEPROM_I2C_MADDR_4; break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c index cc7597a15fe9..2c1d82fc4c34 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c @@ -121,6 +121,7 @@ static ssize_t amdgpu_securedisplay_debugfs_write(struct file *f, const char __u switch (op) { case 1: + mutex_lock(&psp->securedisplay_context.mutex); psp_prep_securedisplay_cmd_buf(psp, &securedisplay_cmd, TA_SECUREDISPLAY_COMMAND__QUERY_TA); ret = psp_securedisplay_invoke(psp, TA_SECUREDISPLAY_COMMAND__QUERY_TA); @@ -131,8 +132,10 @@ static ssize_t amdgpu_securedisplay_debugfs_write(struct file *f, const char __u else psp_securedisplay_parse_resp_status(psp, securedisplay_cmd->status); } + mutex_unlock(&psp->securedisplay_context.mutex); break; case 2: + mutex_lock(&psp->securedisplay_context.mutex); psp_prep_securedisplay_cmd_buf(psp, &securedisplay_cmd, TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC); securedisplay_cmd->securedisplay_in_message.send_roi_crc.phy_id = phy_id; @@ -146,6 +149,7 @@ static ssize_t amdgpu_securedisplay_debugfs_write(struct file *f, const char __u psp_securedisplay_parse_resp_status(psp, securedisplay_cmd->status); } } + mutex_unlock(&psp->securedisplay_context.mutex); break; default: dev_err(adev->dev, "Invalid input: %s\n", str); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index aea8d26b1724..7b5074e776f4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -58,6 +58,7 @@ #include "amdgpu_amdkfd.h" #include "amdgpu_sdma.h" #include "amdgpu_ras.h" +#include "amdgpu_hmm.h" #include "amdgpu_atomfirmware.h" #include "amdgpu_res_cursor.h" #include "bif/bif_4_1_d.h" @@ -634,9 +635,6 @@ struct amdgpu_ttm_tt { struct task_struct *usertask; uint32_t userflags; bool bound; -#if IS_ENABLED(CONFIG_DRM_AMDGPU_USERPTR) - struct hmm_range *range; -#endif }; #define ttm_to_amdgpu_ttm_tt(ptr) container_of(ptr, struct amdgpu_ttm_tt, ttm) @@ -649,7 +647,8 @@ struct amdgpu_ttm_tt { * Calling function must call amdgpu_ttm_tt_userptr_range_done() once and only * once afterwards to stop HMM tracking */ -int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages) +int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages, + struct hmm_range **range) { struct ttm_tt *ttm = bo->tbo.ttm; struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm); @@ -659,16 +658,15 @@ int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages) bool readonly; int r = 0; + /* Make sure get_user_pages_done() can cleanup gracefully */ + *range = NULL; + mm = bo->notifier.mm; if (unlikely(!mm)) { DRM_DEBUG_DRIVER("BO is not registered?\n"); return -EFAULT; } - /* Another get_user_pages is running at the same time?? */ - if (WARN_ON(gtt->range)) - return -EFAULT; - if (!mmget_not_zero(mm)) /* Happens during process shutdown */ return -ESRCH; @@ -685,9 +683,8 @@ int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages) } readonly = amdgpu_ttm_tt_is_readonly(ttm); - r = amdgpu_hmm_range_get_pages(&bo->notifier, mm, pages, start, - ttm->num_pages, >t->range, readonly, - true, NULL); + r = amdgpu_hmm_range_get_pages(&bo->notifier, start, ttm->num_pages, + readonly, NULL, pages, range); out_unlock: mmap_read_unlock(mm); if (r) @@ -704,30 +701,24 @@ out_unlock: * * Returns: true if pages are still valid */ -bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm) +bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm, + struct hmm_range *range) { struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm); - bool r = false; - if (!gtt || !gtt->userptr) + if (!gtt || !gtt->userptr || !range) return false; DRM_DEBUG_DRIVER("user_pages_done 0x%llx pages 0x%x\n", gtt->userptr, ttm->num_pages); - WARN_ONCE(!gtt->range || !gtt->range->hmm_pfns, - "No user pages to check\n"); - - if (gtt->range) { - /* - * FIXME: Must always hold notifier_lock for this, and must - * not ignore the return code. - */ - r = amdgpu_hmm_range_get_pages_done(gtt->range); - gtt->range = NULL; - } + WARN_ONCE(!range->hmm_pfns, "No user pages to check\n"); - return !r; + /* + * FIXME: Must always hold notifier_lock for this, and must + * not ignore the return code. + */ + return !amdgpu_hmm_range_get_pages_done(range); } #endif @@ -804,20 +795,6 @@ static void amdgpu_ttm_tt_unpin_userptr(struct ttm_device *bdev, /* unmap the pages mapped to the device */ dma_unmap_sgtable(adev->dev, ttm->sg, direction, 0); sg_free_table(ttm->sg); - -#if IS_ENABLED(CONFIG_DRM_AMDGPU_USERPTR) - if (gtt->range) { - unsigned long i; - - for (i = 0; i < ttm->num_pages; i++) { - if (ttm->pages[i] != - hmm_pfn_to_page(gtt->range->hmm_pfns[i])) - break; - } - - WARN((i == ttm->num_pages), "Missing get_user_page_done\n"); - } -#endif } static void amdgpu_ttm_gart_bind(struct amdgpu_device *adev, @@ -1168,8 +1145,9 @@ int amdgpu_ttm_tt_get_userptr(const struct ttm_buffer_object *tbo, * @addr: The address in the current tasks VM space to use * @flags: Requirements of userptr object. * - * Called by amdgpu_gem_userptr_ioctl() to bind userptr pages - * to current task + * Called by amdgpu_gem_userptr_ioctl() and kfd_ioctl_alloc_memory_of_gpu() to + * bind userptr pages to current task and by kfd_ioctl_acquire_vm() to + * initialize GPU VM for a KFD process. */ int amdgpu_ttm_tt_set_userptr(struct ttm_buffer_object *bo, uint64_t addr, uint32_t flags) @@ -1553,6 +1531,23 @@ static void amdgpu_ttm_fw_reserve_vram_fini(struct amdgpu_device *adev) NULL, &adev->mman.fw_vram_usage_va); } +/* + * Driver Reservation functions + */ +/** + * amdgpu_ttm_drv_reserve_vram_fini - free drv reserved vram + * + * @adev: amdgpu_device pointer + * + * free drv reserved vram if it has been reserved. + */ +static void amdgpu_ttm_drv_reserve_vram_fini(struct amdgpu_device *adev) +{ + amdgpu_bo_free_kernel(&adev->mman.drv_vram_usage_reserved_bo, + NULL, + NULL); +} + /** * amdgpu_ttm_fw_reserve_vram_init - create bo vram reservation from fw * @@ -1579,6 +1574,31 @@ static int amdgpu_ttm_fw_reserve_vram_init(struct amdgpu_device *adev) &adev->mman.fw_vram_usage_va); } +/** + * amdgpu_ttm_drv_reserve_vram_init - create bo vram reservation from driver + * + * @adev: amdgpu_device pointer + * + * create bo vram reservation from drv. + */ +static int amdgpu_ttm_drv_reserve_vram_init(struct amdgpu_device *adev) +{ + uint64_t vram_size = adev->gmc.visible_vram_size; + + adev->mman.drv_vram_usage_reserved_bo = NULL; + + if (adev->mman.drv_vram_usage_size == 0 || + adev->mman.drv_vram_usage_size > vram_size) + return 0; + + return amdgpu_bo_create_kernel_at(adev, + adev->mman.drv_vram_usage_start_offset, + adev->mman.drv_vram_usage_size, + AMDGPU_GEM_DOMAIN_VRAM, + &adev->mman.drv_vram_usage_reserved_bo, + NULL); +} + /* * Memoy training reservation functions */ @@ -1747,6 +1767,14 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) } /* + *The reserved vram for driver must be pinned to the specified + *place on the VRAM, so reserve it early. + */ + r = amdgpu_ttm_drv_reserve_vram_init(adev); + if (r) + return r; + + /* * only NAVI10 and onwards ASIC support for IP discovery. * If IP discovery enabled, a block of memory should be * reserved for IP discovey. @@ -1871,6 +1899,7 @@ void amdgpu_ttm_fini(struct amdgpu_device *adev) amdgpu_bo_free_kernel(&adev->mman.sdma_access_bo, NULL, &adev->mman.sdma_access_ptr); amdgpu_ttm_fw_reserve_vram_fini(adev); + amdgpu_ttm_drv_reserve_vram_fini(adev); if (drm_dev_enter(adev_to_drm(adev), &idx)) { @@ -2275,9 +2304,9 @@ static ssize_t amdgpu_iomem_read(struct file *f, char __user *buf, if (p->mapping != adev->mman.bdev.dev_mapping) return -EPERM; - ptr = kmap(p); + ptr = kmap_local_page(p); r = copy_to_user(buf, ptr + off, bytes); - kunmap(p); + kunmap_local(ptr); if (r) return -EFAULT; @@ -2326,9 +2355,9 @@ static ssize_t amdgpu_iomem_write(struct file *f, const char __user *buf, if (p->mapping != adev->mman.bdev.dev_mapping) return -EPERM; - ptr = kmap(p); + ptr = kmap_local_page(p); r = copy_from_user(ptr + off, buf, bytes); - kunmap(p); + kunmap_local(ptr); if (r) return -EFAULT; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h index 6a70818039dd..b391c8d076ff 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h @@ -39,6 +39,8 @@ #define AMDGPU_POISON 0xd0bed0be +struct hmm_range; + struct amdgpu_gtt_mgr { struct ttm_resource_manager manager; struct drm_mm mm; @@ -84,6 +86,11 @@ struct amdgpu_mman { struct amdgpu_bo *fw_vram_usage_reserved_bo; void *fw_vram_usage_va; + /* driver VRAM reservation */ + u64 drv_vram_usage_start_offset; + u64 drv_vram_usage_size; + struct amdgpu_bo *drv_vram_usage_reserved_bo; + /* PAGE_SIZE'd BO for process memory r/w over SDMA. */ struct amdgpu_bo *sdma_access_bo; void *sdma_access_ptr; @@ -149,15 +156,19 @@ void amdgpu_ttm_recover_gart(struct ttm_buffer_object *tbo); uint64_t amdgpu_ttm_domain_start(struct amdgpu_device *adev, uint32_t type); #if IS_ENABLED(CONFIG_DRM_AMDGPU_USERPTR) -int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages); -bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm); +int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, struct page **pages, + struct hmm_range **range); +bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm, + struct hmm_range *range); #else static inline int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, - struct page **pages) + struct page **pages, + struct hmm_range **range) { return -EPERM; } -static inline bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm) +static inline bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm, + struct hmm_range *range) { return false; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c index dd0bc649a57d..5cb62e6249c2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c @@ -698,6 +698,7 @@ FW_VERSION_ATTR(rlc_srlg_fw_version, 0444, gfx.rlc_srlg_fw_version); FW_VERSION_ATTR(rlc_srls_fw_version, 0444, gfx.rlc_srls_fw_version); FW_VERSION_ATTR(mec_fw_version, 0444, gfx.mec_fw_version); FW_VERSION_ATTR(mec2_fw_version, 0444, gfx.mec2_fw_version); +FW_VERSION_ATTR(imu_fw_version, 0444, gfx.imu_fw_version); FW_VERSION_ATTR(sos_fw_version, 0444, psp.sos.fw_version); FW_VERSION_ATTR(asd_fw_version, 0444, psp.asd_context.bin_desc.fw_version); FW_VERSION_ATTR(ta_ras_fw_version, 0444, psp.ras_context.context.bin_desc.fw_version); @@ -719,7 +720,8 @@ static struct attribute *fw_attrs[] = { &dev_attr_ta_ras_fw_version.attr, &dev_attr_ta_xgmi_fw_version.attr, &dev_attr_smc_fw_version.attr, &dev_attr_sdma_fw_version.attr, &dev_attr_sdma2_fw_version.attr, &dev_attr_vcn_fw_version.attr, - &dev_attr_dmcu_fw_version.attr, NULL + &dev_attr_dmcu_fw_version.attr, &dev_attr_imu_fw_version.attr, + NULL }; static const struct attribute_group fw_attr_group = { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h index 1c36235b4539..552e06929229 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h @@ -125,6 +125,7 @@ enum psp_fw_type { PSP_FW_TYPE_PSP_INTF_DRV, PSP_FW_TYPE_PSP_DBG_DRV, PSP_FW_TYPE_PSP_RAS_DRV, + PSP_FW_TYPE_MAX_INDEX, }; /* version_major=2, version_minor=0 */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index aad3c8b4c810..f76c19fc0392 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -22,6 +22,59 @@ */ #include "amdgpu.h" +#include "umc_v6_7.h" + +static int amdgpu_umc_convert_error_address(struct amdgpu_device *adev, + struct ras_err_data *err_data, uint64_t err_addr, + uint32_t ch_inst, uint32_t umc_inst) +{ + switch (adev->ip_versions[UMC_HWIP][0]) { + case IP_VERSION(6, 7, 0): + umc_v6_7_convert_error_address(adev, + err_data, err_addr, ch_inst, umc_inst); + break; + default: + dev_warn(adev->dev, + "UMC address to Physical address translation is not supported\n"); + return AMDGPU_RAS_FAIL; + } + + return AMDGPU_RAS_SUCCESS; +} + +int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev, + uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst) +{ + struct ras_err_data err_data = {0, 0, 0, NULL}; + int ret = AMDGPU_RAS_FAIL; + + err_data.err_addr = + kcalloc(adev->umc.max_ras_err_cnt_per_query, + sizeof(struct eeprom_table_record), GFP_KERNEL); + if (!err_data.err_addr) { + dev_warn(adev->dev, + "Failed to alloc memory for umc error record in MCA notifier!\n"); + return AMDGPU_RAS_FAIL; + } + + /* + * Translate UMC channel address to Physical address + */ + ret = amdgpu_umc_convert_error_address(adev, &err_data, err_addr, + ch_inst, umc_inst); + if (ret) + goto out; + + if (amdgpu_bad_page_threshold != 0) { + amdgpu_ras_add_bad_pages(adev, err_data.err_addr, + err_data.err_addr_cnt); + amdgpu_ras_save_bad_pages(adev); + } + +out: + kfree(err_data.err_addr); + return ret; +} static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, void *ras_error_status, @@ -112,23 +165,29 @@ static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev, return AMDGPU_RAS_SUCCESS; } -int amdgpu_umc_poison_handler(struct amdgpu_device *adev, - void *ras_error_status, - bool reset) +int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset) { - int ret; - struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; - struct ras_common_if head = { - .block = AMDGPU_RAS_BLOCK__UMC, - }; - struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head); + int ret = AMDGPU_RAS_SUCCESS; - ret = - amdgpu_umc_do_page_retirement(adev, ras_error_status, NULL, reset); + if (!adev->gmc.xgmi.connected_to_cpu) { + struct ras_err_data err_data = {0, 0, 0, NULL}; + struct ras_common_if head = { + .block = AMDGPU_RAS_BLOCK__UMC, + }; + struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head); - if (ret == AMDGPU_RAS_SUCCESS && obj) { - obj->err_data.ue_count += err_data->ue_count; - obj->err_data.ce_count += err_data->ce_count; + ret = amdgpu_umc_do_page_retirement(adev, &err_data, NULL, reset); + + if (ret == AMDGPU_RAS_SUCCESS && obj) { + obj->err_data.ue_count += err_data.ue_count; + obj->err_data.ce_count += err_data.ce_count; + } + } else if (reset) { + /* MCA poison handler is only responsible for GPU reset, + * let MCA notifier do page retirement. + */ + kgd2kfd_set_sram_ecc_flag(adev->kfd.dev); + amdgpu_ras_reset_gpu(adev); } return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h index e46439274f3a..a6951160f13a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h @@ -51,9 +51,6 @@ struct amdgpu_umc_ras { struct amdgpu_ras_block_object ras_block; void (*err_cnt_init)(struct amdgpu_device *adev); bool (*query_ras_poison_mode)(struct amdgpu_device *adev); - void (*convert_ras_error_address)(struct amdgpu_device *adev, - struct ras_err_data *err_data, uint64_t err_addr, - uint32_t ch_inst, uint32_t umc_inst); void (*ecc_info_query_ras_error_count)(struct amdgpu_device *adev, void *ras_error_status); void (*ecc_info_query_ras_error_address)(struct amdgpu_device *adev, @@ -86,9 +83,7 @@ struct amdgpu_umc { }; int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block); -int amdgpu_umc_poison_handler(struct amdgpu_device *adev, - void *ras_error_status, - bool reset); +int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset); int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry); @@ -101,4 +96,6 @@ void amdgpu_umc_fill_error_record(struct ras_err_data *err_data, int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev, void *ras_error_status, struct amdgpu_iv_entry *entry); +int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev, + uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c index 3449145ab2bc..33f3415096f7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c @@ -1252,3 +1252,20 @@ int amdgpu_vcn_process_poison_irq(struct amdgpu_device *adev, return 0; } + +void amdgpu_vcn_set_ras_funcs(struct amdgpu_device *adev) +{ + if (!adev->vcn.ras) + return; + + amdgpu_ras_register_ras_block(adev, &adev->vcn.ras->ras_block); + + strcpy(adev->vcn.ras->ras_block.ras_comm.name, "vcn"); + adev->vcn.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__VCN; + adev->vcn.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__POISON; + adev->vcn.ras_if = &adev->vcn.ras->ras_block.ras_comm; + + /* If don't define special ras_late_init function, use default ras_late_init */ + if (!adev->vcn.ras->ras_block.ras_late_init) + adev->vcn.ras->ras_block.ras_late_init = amdgpu_ras_block_late_init; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h index 253ea6b159df..dbb8d68a30c6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h @@ -399,5 +399,6 @@ void amdgpu_debugfs_vcn_fwlog_init(struct amdgpu_device *adev, int amdgpu_vcn_process_poison_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry); +void amdgpu_vcn_set_ras_funcs(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index 9c765b04aae3..a226a6c48fb7 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c @@ -64,6 +64,10 @@ void amdgpu_virt_init_setting(struct amdgpu_device *adev) ddev->driver_features &= ~DRIVER_ATOMIC; adev->cg_flags = 0; adev->pg_flags = 0; + + /* enable mcbp for sriov asic_type before soc21 */ + amdgpu_mcbp = (adev->asic_type < CHIP_IP_DISCOVERY) ? 1 : 0; + } void amdgpu_virt_kiq_reg_write_reg_wait(struct amdgpu_device *adev, @@ -547,6 +551,7 @@ static void amdgpu_virt_populate_vf2pf_ucode_info(struct amdgpu_device *adev) POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_RLC_SRLS, adev->gfx.rlc_srls_fw_version); POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_MEC, adev->gfx.mec_fw_version); POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_MEC2, adev->gfx.mec2_fw_version); + POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_IMU, adev->gfx.imu_fw_version); POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_SOS, adev->psp.sos.fw_version); POPULATE_UCODE_INFO(vf2pf_info, AMD_SRIOV_UCODE_ID_ASD, adev->psp.asd_context.bin_desc.fw_version); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h index 49c4347d154c..2b9d806e23af 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h @@ -75,6 +75,8 @@ struct amdgpu_vf_error_buffer { uint64_t data[AMDGPU_VF_ERROR_ENTRY_SIZE]; }; +enum idh_request; + /** * struct amdgpu_virt_ops - amdgpu device virt operations */ @@ -84,7 +86,8 @@ struct amdgpu_virt_ops { int (*req_init_data)(struct amdgpu_device *adev); int (*reset_gpu)(struct amdgpu_device *adev); int (*wait_reset)(struct amdgpu_device *adev); - void (*trans_msg)(struct amdgpu_device *adev, u32 req, u32 data1, u32 data2, u32 data3); + void (*trans_msg)(struct amdgpu_device *adev, enum idh_request req, + u32 data1, u32 data2, u32 data3); }; /* diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c index 09dec2561adf..53ff91fc6cf6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vkms.c @@ -498,6 +498,10 @@ static int amdgpu_vkms_sw_init(void *handle) adev_to_drm(adev)->mode_config.preferred_depth = 24; adev_to_drm(adev)->mode_config.prefer_shadow = 1; + adev_to_drm(adev)->mode_config.fb_modifiers_not_supported = true; + + adev_to_drm(adev)->mode_config.fb_modifiers_not_supported = true; + r = amdgpu_display_modeset_create_props(adev); if (r) return r; @@ -509,6 +513,10 @@ static int amdgpu_vkms_sw_init(void *handle) return r; } + r = drm_vblank_init(adev_to_drm(adev), adev->mode_info.num_crtc); + if (r) + return r; + drm_kms_helper_poll_init(adev_to_drm(adev)); adev->mode_info.mode_config_initialized = true; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 2291aa14d888..003aa9e47085 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -143,32 +143,6 @@ int amdgpu_vm_set_pasid(struct amdgpu_device *adev, struct amdgpu_vm *vm, return 0; } -/* - * vm eviction_lock can be taken in MMU notifiers. Make sure no reclaim-FS - * happens while holding this lock anywhere to prevent deadlocks when - * an MMU notifier runs in reclaim-FS context. - */ -static inline void amdgpu_vm_eviction_lock(struct amdgpu_vm *vm) -{ - mutex_lock(&vm->eviction_lock); - vm->saved_flags = memalloc_noreclaim_save(); -} - -static inline int amdgpu_vm_eviction_trylock(struct amdgpu_vm *vm) -{ - if (mutex_trylock(&vm->eviction_lock)) { - vm->saved_flags = memalloc_noreclaim_save(); - return 1; - } - return 0; -} - -static inline void amdgpu_vm_eviction_unlock(struct amdgpu_vm *vm) -{ - memalloc_noreclaim_restore(vm->saved_flags); - mutex_unlock(&vm->eviction_lock); -} - /** * amdgpu_vm_bo_evicted - vm_bo is evicted * diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h index 83acb7bd80fe..6546e786bf00 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h @@ -492,7 +492,48 @@ void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m); */ static inline uint64_t amdgpu_vm_tlb_seq(struct amdgpu_vm *vm) { + unsigned long flags; + spinlock_t *lock; + + /* + * Workaround to stop racing between the fence signaling and handling + * the cb. The lock is static after initially setting it up, just make + * sure that the dma_fence structure isn't freed up. + */ + rcu_read_lock(); + lock = vm->last_tlb_flush->lock; + rcu_read_unlock(); + + spin_lock_irqsave(lock, flags); + spin_unlock_irqrestore(lock, flags); + return atomic64_read(&vm->tlb_seq); } +/* + * vm eviction_lock can be taken in MMU notifiers. Make sure no reclaim-FS + * happens while holding this lock anywhere to prevent deadlocks when + * an MMU notifier runs in reclaim-FS context. + */ +static inline void amdgpu_vm_eviction_lock(struct amdgpu_vm *vm) +{ + mutex_lock(&vm->eviction_lock); + vm->saved_flags = memalloc_noreclaim_save(); +} + +static inline bool amdgpu_vm_eviction_trylock(struct amdgpu_vm *vm) +{ + if (mutex_trylock(&vm->eviction_lock)) { + vm->saved_flags = memalloc_noreclaim_save(); + return true; + } + return false; +} + +static inline void amdgpu_vm_eviction_unlock(struct amdgpu_vm *vm) +{ + memalloc_noreclaim_restore(vm->saved_flags); + mutex_unlock(&vm->eviction_lock); +} + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c index 358b91243e37..b5f3bba851db 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c @@ -597,7 +597,9 @@ static int amdgpu_vm_pt_alloc(struct amdgpu_device *adev, if (entry->bo) return 0; + amdgpu_vm_eviction_unlock(vm); r = amdgpu_vm_pt_create(adev, vm, cursor->level, immediate, &pt); + amdgpu_vm_eviction_lock(vm); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c index 18c1a173d187..faa12146635c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c @@ -435,7 +435,7 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man, if (place->flags & TTM_PL_FLAG_TOPDOWN) vres->flags |= DRM_BUDDY_TOPDOWN_ALLOCATION; - if (fpfn || lpfn != man->size) + if (fpfn || lpfn != mgr->mm.size) /* Allocate blocks in desired range */ vres->flags |= DRM_BUDDY_RANGE_ALLOCATION; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h index e78e4c27b62a..6c97148ca0ed 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h @@ -70,6 +70,7 @@ enum amd_sriov_ucode_engine_id { AMD_SRIOV_UCODE_ID_RLC_SRLS, AMD_SRIOV_UCODE_ID_MEC, AMD_SRIOV_UCODE_ID_MEC2, + AMD_SRIOV_UCODE_ID_IMU, AMD_SRIOV_UCODE_ID_SOS, AMD_SRIOV_UCODE_ID_ASD, AMD_SRIOV_UCODE_ID_TA_RAS, diff --git a/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c b/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c index 6be9ac2b9c5b..18ae9433e463 100644 --- a/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c +++ b/drivers/gpu/drm/amd/amdgpu/atombios_encoders.c @@ -2081,8 +2081,11 @@ amdgpu_atombios_encoder_get_lcd_info(struct amdgpu_encoder *encoder) } } record += fake_edid_record->ucFakeEDIDLength ? - fake_edid_record->ucFakeEDIDLength + 2 : - sizeof(ATOM_FAKE_EDID_PATCH_RECORD); + struct_size(fake_edid_record, + ucFakeEDIDString, + fake_edid_record->ucFakeEDIDLength) : + /* empty fake edid record must be 3 bytes long */ + sizeof(ATOM_FAKE_EDID_PATCH_RECORD) + 1; break; case LCD_PANEL_RESOLUTION_RECORD_TYPE: panel_res_record = (ATOM_PANEL_RESOLUTION_PATCH_RECORD *)record; diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c index 90f87b2d985b..248f1a4e915f 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v10_0.c @@ -2829,6 +2829,17 @@ static int dce_v10_0_sw_init(void *handle) if (r) return r; + /* Disable vblank IRQs aggressively for power-saving */ + /* XXX: can this be enabled for DC? */ + adev_to_drm(adev)->vblank_disable_immediate = true; + + r = drm_vblank_init(adev_to_drm(adev), adev->mode_info.num_crtc); + if (r) + return r; + + INIT_WORK(&adev->hotplug_work, + amdgpu_display_hotplug_work_func); + drm_kms_helper_poll_init(adev_to_drm(adev)); adev->mode_info.mode_config_initialized = true; @@ -2891,6 +2902,8 @@ static int dce_v10_0_hw_fini(void *handle) dce_v10_0_pageflip_interrupt_fini(adev); + flush_work(&adev->hotplug_work); + return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c index 0352de72c886..cd9c19060d89 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v11_0.c @@ -2948,6 +2948,17 @@ static int dce_v11_0_sw_init(void *handle) if (r) return r; + /* Disable vblank IRQs aggressively for power-saving */ + /* XXX: can this be enabled for DC? */ + adev_to_drm(adev)->vblank_disable_immediate = true; + + r = drm_vblank_init(adev_to_drm(adev), adev->mode_info.num_crtc); + if (r) + return r; + + INIT_WORK(&adev->hotplug_work, + amdgpu_display_hotplug_work_func); + drm_kms_helper_poll_init(adev_to_drm(adev)); adev->mode_info.mode_config_initialized = true; @@ -3021,6 +3032,8 @@ static int dce_v11_0_hw_fini(void *handle) dce_v11_0_pageflip_interrupt_fini(adev); + flush_work(&adev->hotplug_work); + return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c index 07bd16e82046..76323deecc58 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v6_0.c @@ -2706,6 +2706,18 @@ static int dce_v6_0_sw_init(void *handle) if (r) return r; + /* Disable vblank IRQs aggressively for power-saving */ + /* XXX: can this be enabled for DC? */ + adev_to_drm(adev)->vblank_disable_immediate = true; + + r = drm_vblank_init(adev_to_drm(adev), adev->mode_info.num_crtc); + if (r) + return r; + + /* Pre-DCE11 */ + INIT_WORK(&adev->hotplug_work, + amdgpu_display_hotplug_work_func); + drm_kms_helper_poll_init(adev_to_drm(adev)); return r; @@ -2764,6 +2776,8 @@ static int dce_v6_0_hw_fini(void *handle) dce_v6_0_pageflip_interrupt_fini(adev); + flush_work(&adev->hotplug_work); + return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c index d73df100f2b3..01cf3ab111cb 100644 --- a/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/dce_v8_0.c @@ -2730,6 +2730,18 @@ static int dce_v8_0_sw_init(void *handle) if (r) return r; + /* Disable vblank IRQs aggressively for power-saving */ + /* XXX: can this be enabled for DC? */ + adev_to_drm(adev)->vblank_disable_immediate = true; + + r = drm_vblank_init(adev_to_drm(adev), adev->mode_info.num_crtc); + if (r) + return r; + + /* Pre-DCE11 */ + INIT_WORK(&adev->hotplug_work, + amdgpu_display_hotplug_work_func); + drm_kms_helper_poll_init(adev_to_drm(adev)); adev->mode_info.mode_config_initialized = true; @@ -2790,6 +2802,8 @@ static int dce_v8_0_hw_fini(void *handle) dce_v8_0_pageflip_interrupt_fini(adev); + flush_work(&adev->hotplug_work); + return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index af94ac580d3e..49d34c7bbf20 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -4453,8 +4453,6 @@ static void gfx_v10_0_gpu_early_init(struct amdgpu_device *adev) { u32 gb_addr_config; - adev->gfx.funcs = &gfx_v10_0_gfx_funcs; - switch (adev->ip_versions[GC_HWIP][0]) { case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 1): @@ -6911,6 +6909,8 @@ static int gfx_v10_0_kiq_init_queue(struct amdgpu_ring *ring) mutex_unlock(&adev->srbm_mutex); } else { memset((void *)mqd, 0, sizeof(*mqd)); + if (amdgpu_sriov_vf(adev) && adev->in_suspend) + amdgpu_ring_clear_ring(ring); mutex_lock(&adev->srbm_mutex); nv_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0); amdgpu_ring_init_mqd(ring); @@ -7593,6 +7593,8 @@ static int gfx_v10_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + adev->gfx.funcs = &gfx_v10_0_gfx_funcs; + switch (adev->ip_versions[GC_HWIP][0]) { case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 1): @@ -8489,7 +8491,7 @@ static void gfx_v10_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, control |= ib->length_dw | (vmid << 24); - if ((amdgpu_sriov_vf(ring->adev) || amdgpu_mcbp) && (ib->flags & AMDGPU_IB_FLAG_PREEMPT)) { + if (amdgpu_mcbp && (ib->flags & AMDGPU_IB_FLAG_PREEMPT)) { control |= INDIRECT_BUFFER_PRE_ENB(1); if (flags & AMDGPU_IB_PREEMPTED) @@ -8664,7 +8666,7 @@ static void gfx_v10_0_ring_emit_cntxcntl(struct amdgpu_ring *ring, { uint32_t dw2 = 0; - if (amdgpu_mcbp || amdgpu_sriov_vf(ring->adev)) + if (amdgpu_mcbp) gfx_v10_0_ring_emit_ce_meta(ring, (!amdgpu_sriov_vf(ring->adev) && flags & AMDGPU_IB_PREEMPTED) ? true : false); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c index 671ca5a0f208..9d2c6523f546 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c @@ -843,7 +843,6 @@ static const struct amdgpu_gfx_funcs gfx_v11_0_gfx_funcs = { static int gfx_v11_0_gpu_early_init(struct amdgpu_device *adev) { - adev->gfx.funcs = &gfx_v11_0_gfx_funcs; switch (adev->ip_versions[GC_HWIP][0]) { case IP_VERSION(11, 0, 0): @@ -1626,7 +1625,8 @@ static void gfx_v11_0_constants_init(struct amdgpu_device *adev) u32 tmp; int i; - WREG32_FIELD15_PREREG(GC, 0, GRBM_CNTL, READ_TIMEOUT, 0xff); + if (!amdgpu_sriov_vf(adev)) + WREG32_FIELD15_PREREG(GC, 0, GRBM_CNTL, READ_TIMEOUT, 0xff); gfx_v11_0_setup_rb(adev); gfx_v11_0_get_cu_info(adev, &adev->gfx.cu_info); @@ -4004,6 +4004,8 @@ static int gfx_v11_0_kiq_init_queue(struct amdgpu_ring *ring) mutex_unlock(&adev->srbm_mutex); } else { memset((void *)mqd, 0, sizeof(*mqd)); + if (amdgpu_sriov_vf(adev) && adev->in_suspend) + amdgpu_ring_clear_ring(ring); mutex_lock(&adev->srbm_mutex); soc21_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0); amdgpu_ring_init_mqd(ring); @@ -4390,7 +4392,6 @@ static int gfx_v11_0_hw_fini(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; int r; - uint32_t tmp; amdgpu_irq_put(adev, &adev->gfx.priv_reg_irq, 0); amdgpu_irq_put(adev, &adev->gfx.priv_inst_irq, 0); @@ -4409,15 +4410,14 @@ static int gfx_v11_0_hw_fini(void *handle) amdgpu_mes_kiq_hw_fini(adev); } - if (amdgpu_sriov_vf(adev)) { - gfx_v11_0_cp_gfx_enable(adev, false); - /* Program KIQ position of RLC_CP_SCHEDULERS during destroy */ - tmp = RREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS); - tmp &= 0xffffff00; - WREG32_SOC15(GC, 0, regRLC_CP_SCHEDULERS, tmp); - + if (amdgpu_sriov_vf(adev)) + /* Remove the steps disabling CPG and clearing KIQ position, + * so that CP could perform IDLE-SAVE during switch. Those + * steps are necessary to avoid a DMAR error in gfx9 but it is + * not reproduced on gfx11. + */ return 0; - } + gfx_v11_0_cp_enable(adev, false); gfx_v11_0_enable_gui_idle_interrupt(adev, false); @@ -4656,6 +4656,8 @@ static int gfx_v11_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + adev->gfx.funcs = &gfx_v11_0_gfx_funcs; + adev->gfx.num_gfx_rings = GFX11_NUM_GFX_RINGS; adev->gfx.num_compute_rings = min(amdgpu_gfx_get_num_kcq(adev), AMDGPU_MAX_COMPUTE_RINGS); @@ -4673,6 +4675,26 @@ static int gfx_v11_0_early_init(void *handle) return 0; } +static int gfx_v11_0_ras_late_init(void *handle) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct ras_common_if *gfx_common_if; + int ret; + + gfx_common_if = kzalloc(sizeof(struct ras_common_if), GFP_KERNEL); + if (!gfx_common_if) + return -ENOMEM; + + gfx_common_if->block = AMDGPU_RAS_BLOCK__GFX; + + ret = amdgpu_ras_feature_enable(adev, gfx_common_if, true); + if (ret) + dev_warn(adev->dev, "Failed to enable gfx11 ras feature\n"); + + kfree(gfx_common_if); + return 0; +} + static int gfx_v11_0_late_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; @@ -4686,6 +4708,12 @@ static int gfx_v11_0_late_init(void *handle) if (r) return r; + if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(11, 0, 3)) { + r = gfx_v11_0_ras_late_init(handle); + if (r) + return r; + } + return 0; } @@ -5051,6 +5079,7 @@ static int gfx_v11_0_set_powergating_state(void *handle, switch (adev->ip_versions[GC_HWIP][0]) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 2): + case IP_VERSION(11, 0, 3): amdgpu_gfx_off_ctrl(adev, enable); break; case IP_VERSION(11, 0, 1): @@ -5298,7 +5327,7 @@ static void gfx_v11_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, control |= ib->length_dw | (vmid << 24); - if ((amdgpu_sriov_vf(ring->adev) || amdgpu_mcbp) && (ib->flags & AMDGPU_IB_FLAG_PREEMPT)) { + if (amdgpu_mcbp && (ib->flags & AMDGPU_IB_FLAG_PREEMPT)) { control |= INDIRECT_BUFFER_PRE_ENB(1); if (flags & AMDGPU_IB_PREEMPTED) @@ -6059,6 +6088,7 @@ static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_gfx = { .align_mask = 0xff, .nop = PACKET3(PACKET3_NOP, 0x3FFF), .support_64bit_ptrs = true, + .secure_submission_supported = true, .vmhub = AMDGPU_GFXHUB_0, .get_rptr = gfx_v11_0_ring_get_rptr_gfx, .get_wptr = gfx_v11_0_ring_get_wptr_gfx, diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 7f0b18b0d4c4..d47135606e3e 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -4643,6 +4643,8 @@ static int gfx_v8_0_kiq_init_queue(struct amdgpu_ring *ring) memset((void *)mqd, 0, sizeof(struct vi_mqd_allocation)); ((struct vi_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF; ((struct vi_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF; + if (amdgpu_sriov_vf(adev) && adev->in_suspend) + amdgpu_ring_clear_ring(ring); mutex_lock(&adev->srbm_mutex); vi_srbm_select(adev, ring->me, ring->pipe, ring->queue, 0); gfx_v8_0_mqd_init(ring); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 0320be4a5fc6..676832da75eb 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -1564,7 +1564,7 @@ static void gfx_v9_0_init_always_on_cu_mask(struct amdgpu_device *adev) mask = 1; cu_bitmap = 0; counter = 0; - gfx_v9_0_select_se_sh(adev, i, j, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff); for (k = 0; k < adev->gfx.config.max_cu_per_sh; k ++) { if (cu_info->bitmap[i][j] & mask) { @@ -1583,7 +1583,7 @@ static void gfx_v9_0_init_always_on_cu_mask(struct amdgpu_device *adev) cu_info->ao_cu_bitmap[i][j] = cu_bitmap; } } - gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); mutex_unlock(&adev->grbm_idx_mutex); } @@ -1605,7 +1605,7 @@ static void gfx_v9_0_init_lbpw(struct amdgpu_device *adev) mutex_lock(&adev->grbm_idx_mutex); /* set mmRLC_LB_INIT_CU_MASK thru broadcast mode to enable all SE/SH*/ - gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); WREG32_SOC15(GC, 0, mmRLC_LB_INIT_CU_MASK, 0xffffffff); /* set mmRLC_LB_PARAMS = 0x003F_1006 */ @@ -1654,7 +1654,7 @@ static void gfx_v9_4_init_lbpw(struct amdgpu_device *adev) mutex_lock(&adev->grbm_idx_mutex); /* set mmRLC_LB_INIT_CU_MASK thru broadcast mode to enable all SE/SH*/ - gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); WREG32_SOC15(GC, 0, mmRLC_LB_INIT_CU_MASK, 0xffffffff); /* set mmRLC_LB_PARAMS = 0x003F_1006 */ @@ -1919,8 +1919,6 @@ static int gfx_v9_0_gpu_early_init(struct amdgpu_device *adev) u32 gb_addr_config; int err; - adev->gfx.funcs = &gfx_v9_0_gfx_funcs; - switch (adev->ip_versions[GC_HWIP][0]) { case IP_VERSION(9, 0, 1): adev->gfx.config.max_hw_contexts = 8; @@ -2324,13 +2322,13 @@ static void gfx_v9_0_setup_rb(struct amdgpu_device *adev) mutex_lock(&adev->grbm_idx_mutex); for (i = 0; i < adev->gfx.config.max_shader_engines; i++) { for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) { - gfx_v9_0_select_se_sh(adev, i, j, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff); data = gfx_v9_0_get_rb_active_bitmap(adev); active_rbs |= data << ((i * adev->gfx.config.max_sh_per_se + j) * rb_bitmap_width_per_sh); } } - gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); mutex_unlock(&adev->grbm_idx_mutex); adev->gfx.config.backend_enable_mask = active_rbs; @@ -2467,14 +2465,14 @@ static void gfx_v9_0_wait_for_rlc_serdes(struct amdgpu_device *adev) mutex_lock(&adev->grbm_idx_mutex); for (i = 0; i < adev->gfx.config.max_shader_engines; i++) { for (j = 0; j < adev->gfx.config.max_sh_per_se; j++) { - gfx_v9_0_select_se_sh(adev, i, j, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff); for (k = 0; k < adev->usec_timeout; k++) { if (RREG32_SOC15(GC, 0, mmRLC_SERDES_CU_MASTER_BUSY) == 0) break; udelay(1); } if (k == adev->usec_timeout) { - gfx_v9_0_select_se_sh(adev, 0xffffffff, + amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); mutex_unlock(&adev->grbm_idx_mutex); DRM_INFO("Timeout wait for RLC serdes %u,%u\n", @@ -2483,7 +2481,7 @@ static void gfx_v9_0_wait_for_rlc_serdes(struct amdgpu_device *adev) } } } - gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); mutex_unlock(&adev->grbm_idx_mutex); mask = RLC_SERDES_NONCU_MASTER_BUSY__SE_MASTER_BUSY_MASK | @@ -3583,6 +3581,8 @@ static int gfx_v9_0_kiq_init_queue(struct amdgpu_ring *ring) memset((void *)mqd, 0, sizeof(struct v9_mqd_allocation)); ((struct v9_mqd_allocation *)mqd)->dynamic_cu_mask = 0xFFFFFFFF; ((struct v9_mqd_allocation *)mqd)->dynamic_rb_mask = 0xFFFFFFFF; + if (amdgpu_sriov_vf(adev) && adev->in_suspend) + amdgpu_ring_clear_ring(ring); mutex_lock(&adev->srbm_mutex); soc15_grbm_select(adev, ring->me, ring->pipe, ring->queue, 0); gfx_v9_0_mqd_init(ring); @@ -4539,6 +4539,8 @@ static int gfx_v9_0_early_init(void *handle) { struct amdgpu_device *adev = (struct amdgpu_device *)handle; + adev->gfx.funcs = &gfx_v9_0_gfx_funcs; + if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 1) || adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2)) adev->gfx.num_gfx_rings = 0; @@ -6482,7 +6484,7 @@ static void gfx_v9_0_reset_ras_error_count(struct amdgpu_device *adev) for (i = 0; i < ARRAY_SIZE(gfx_v9_0_edc_counter_regs); i++) { for (j = 0; j < gfx_v9_0_edc_counter_regs[i].se_num; j++) { for (k = 0; k < gfx_v9_0_edc_counter_regs[i].instance; k++) { - gfx_v9_0_select_se_sh(adev, j, 0x0, k); + amdgpu_gfx_select_se_sh(adev, j, 0x0, k); RREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_0_edc_counter_regs[i])); } } @@ -6544,7 +6546,7 @@ static void gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev, for (i = 0; i < ARRAY_SIZE(gfx_v9_0_edc_counter_regs); i++) { for (j = 0; j < gfx_v9_0_edc_counter_regs[i].se_num; j++) { for (k = 0; k < gfx_v9_0_edc_counter_regs[i].instance; k++) { - gfx_v9_0_select_se_sh(adev, j, 0, k); + amdgpu_gfx_select_se_sh(adev, j, 0, k); reg_value = RREG32(SOC15_REG_ENTRY_OFFSET(gfx_v9_0_edc_counter_regs[i])); if (reg_value) @@ -6559,7 +6561,7 @@ static void gfx_v9_0_query_ras_error_count(struct amdgpu_device *adev, err_data->ce_count += sec_count; err_data->ue_count += ded_count; - gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); mutex_unlock(&adev->grbm_idx_mutex); gfx_v9_0_query_utc_edc_status(adev, err_data); @@ -6963,7 +6965,7 @@ static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev, mask = 1; ao_bitmap = 0; counter = 0; - gfx_v9_0_select_se_sh(adev, i, j, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, i, j, 0xffffffff); gfx_v9_0_set_user_cu_inactive_bitmap( adev, disable_masks[i * adev->gfx.config.max_sh_per_se + j]); bitmap = gfx_v9_0_get_cu_active_bitmap(adev); @@ -6996,7 +6998,7 @@ static int gfx_v9_0_get_cu_info(struct amdgpu_device *adev, cu_info->ao_cu_bitmap[i % 4][j + i / 4] = ao_bitmap; } } - gfx_v9_0_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); + amdgpu_gfx_select_se_sh(adev, 0xffffffff, 0xffffffff, 0xffffffff); mutex_unlock(&adev->grbm_idx_mutex); cu_info->number = active_cu_number; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c index 8cf53e039c11..3f8676d23a5e 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v2_1.c @@ -397,6 +397,9 @@ static void gfxhub_v2_1_gart_disable(struct amdgpu_device *adev) ENABLE_ADVANCED_DRIVER_MODEL, 0); WREG32_SOC15(GC, 0, mmGCMC_VM_MX_L1_TLB_CNTL, tmp); + if (amdgpu_sriov_vf(adev)) + return; + /* Setup L2 cache */ WREG32_FIELD15(GC, 0, GCVM_L2_CNTL, ENABLE_L2_CACHE, 0); WREG32_SOC15(GC, 0, mmGCVM_L2_CNTL3, 0); diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0_3.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0_3.c index 5d3fffd4929f..080ff11ca305 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v3_0_3.c @@ -154,6 +154,9 @@ static void gfxhub_v3_0_3_init_system_aperture_regs(struct amdgpu_device *adev) { uint64_t value; + if (amdgpu_sriov_vf(adev)) + return; + /* Disable AGP. */ WREG32_SOC15(GC, 0, regGCMC_VM_AGP_BASE, 0); WREG32_SOC15(GC, 0, regGCMC_VM_AGP_TOP, 0); @@ -354,18 +357,6 @@ static void gfxhub_v3_0_3_program_invalidation(struct amdgpu_device *adev) static int gfxhub_v3_0_3_gart_enable(struct amdgpu_device *adev) { - if (amdgpu_sriov_vf(adev)) { - /* - * GCMC_VM_FB_LOCATION_BASE/TOP is NULL for VF, becuase they are - * VF copy registers so vbios post doesn't program them, for - * SRIOV driver need to program them - */ - WREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_BASE, - adev->gmc.vram_start >> 24); - WREG32_SOC15(GC, 0, regGCMC_VM_FB_LOCATION_TOP, - adev->gmc.vram_end >> 24); - } - /* GART Enable. */ gfxhub_v3_0_3_init_gart_aperture_regs(adev); gfxhub_v3_0_3_init_system_aperture_regs(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c index 657e53708248..21e46817d82d 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c @@ -608,6 +608,8 @@ static void gmc_v10_0_get_vm_pte(struct amdgpu_device *adev, struct amdgpu_bo_va_mapping *mapping, uint64_t *flags) { + struct amdgpu_bo *bo = mapping->bo_va->base.bo; + *flags &= ~AMDGPU_PTE_EXECUTABLE; *flags |= mapping->flags & AMDGPU_PTE_EXECUTABLE; @@ -624,6 +626,11 @@ static void gmc_v10_0_get_vm_pte(struct amdgpu_device *adev, *flags |= AMDGPU_PTE_SYSTEM; *flags &= ~AMDGPU_PTE_VALID; } + + if (bo && bo->flags & (AMDGPU_GEM_CREATE_COHERENT | + AMDGPU_GEM_CREATE_UNCACHED)) + *flags = (*flags & ~AMDGPU_PTE_MTYPE_NV10_MASK) | + AMDGPU_PTE_MTYPE_NV10(MTYPE_UC); } static unsigned gmc_v10_0_get_vbios_fb_size(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c index 66dfb574cc7d..96e52ec0fb69 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c @@ -503,6 +503,8 @@ static void gmc_v11_0_get_vm_pte(struct amdgpu_device *adev, struct amdgpu_bo_va_mapping *mapping, uint64_t *flags) { + struct amdgpu_bo *bo = mapping->bo_va->base.bo; + *flags &= ~AMDGPU_PTE_EXECUTABLE; *flags |= mapping->flags & AMDGPU_PTE_EXECUTABLE; @@ -519,6 +521,11 @@ static void gmc_v11_0_get_vm_pte(struct amdgpu_device *adev, *flags |= AMDGPU_PTE_SYSTEM; *flags &= ~AMDGPU_PTE_VALID; } + + if (bo && bo->flags & (AMDGPU_GEM_CREATE_COHERENT | + AMDGPU_GEM_CREATE_UNCACHED)) + *flags = (*flags & ~AMDGPU_PTE_MTYPE_NV10_MASK) | + AMDGPU_PTE_MTYPE_NV10(MTYPE_UC); } static unsigned gmc_v11_0_get_vbios_fb_size(struct amdgpu_device *adev) @@ -551,7 +558,10 @@ static void gmc_v11_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.node_inst_num = adev->gmc.num_umc; adev->umc.max_ras_err_cnt_per_query = UMC_V8_10_TOTAL_CHANNEL_NUM(adev); adev->umc.channel_offs = UMC_V8_10_PER_CHANNEL_OFFSET; - adev->umc.channel_idx_tbl = &umc_v8_10_channel_idx_tbl[0][0][0]; + if (adev->umc.node_inst_num == 4) + adev->umc.channel_idx_tbl = &umc_v8_10_channel_idx_tbl_ext0[0][0][0]; + else + adev->umc.channel_idx_tbl = &umc_v8_10_channel_idx_tbl[0][0][0]; adev->umc.ras = &umc_v8_10_ras; break; case IP_VERSION(8, 11, 0): diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 67ca16a8027c..50386eb2eec8 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -1113,6 +1113,74 @@ static void gmc_v9_0_get_vm_pde(struct amdgpu_device *adev, int level, } } +static void gmc_v9_0_get_coherence_flags(struct amdgpu_device *adev, + struct amdgpu_bo *bo, + struct amdgpu_bo_va_mapping *mapping, + uint64_t *flags) +{ + struct amdgpu_device *bo_adev = amdgpu_ttm_adev(bo->tbo.bdev); + bool is_vram = bo->tbo.resource->mem_type == TTM_PL_VRAM; + bool coherent = bo->flags & AMDGPU_GEM_CREATE_COHERENT; + bool uncached = bo->flags & AMDGPU_GEM_CREATE_UNCACHED; + unsigned int mtype; + bool snoop = false; + + switch (adev->ip_versions[GC_HWIP][0]) { + case IP_VERSION(9, 4, 1): + case IP_VERSION(9, 4, 2): + if (is_vram) { + if (bo_adev == adev) { + if (uncached) + mtype = MTYPE_UC; + else if (coherent) + mtype = MTYPE_CC; + else + mtype = MTYPE_RW; + /* FIXME: is this still needed? Or does + * amdgpu_ttm_tt_pde_flags already handle this? + */ + if (adev->ip_versions[GC_HWIP][0] == + IP_VERSION(9, 4, 2) && + adev->gmc.xgmi.connected_to_cpu) + snoop = true; + } else { + if (uncached || coherent) + mtype = MTYPE_UC; + else + mtype = MTYPE_NC; + if (mapping->bo_va->is_xgmi) + snoop = true; + } + } else { + if (uncached || coherent) + mtype = MTYPE_UC; + else + mtype = MTYPE_NC; + /* FIXME: is this still needed? Or does + * amdgpu_ttm_tt_pde_flags already handle this? + */ + snoop = true; + } + break; + default: + if (uncached || coherent) + mtype = MTYPE_UC; + else + mtype = MTYPE_NC; + + /* FIXME: is this still needed? Or does + * amdgpu_ttm_tt_pde_flags already handle this? + */ + if (!is_vram) + snoop = true; + } + + if (mtype != MTYPE_NC) + *flags = (*flags & ~AMDGPU_PTE_MTYPE_VG10_MASK) | + AMDGPU_PTE_MTYPE_VG10(mtype); + *flags |= snoop ? AMDGPU_PTE_SNOOPED : 0; +} + static void gmc_v9_0_get_vm_pte(struct amdgpu_device *adev, struct amdgpu_bo_va_mapping *mapping, uint64_t *flags) @@ -1128,14 +1196,9 @@ static void gmc_v9_0_get_vm_pte(struct amdgpu_device *adev, *flags &= ~AMDGPU_PTE_VALID; } - if ((adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 1) || - adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2)) && - !(*flags & AMDGPU_PTE_SYSTEM) && - mapping->bo_va->is_xgmi) - *flags |= AMDGPU_PTE_SNOOPED; - - if (adev->ip_versions[GC_HWIP][0] == IP_VERSION(9, 4, 2)) - *flags |= mapping->flags & AMDGPU_PTE_SNOOPED; + if (mapping->bo_va->base.bo) + gmc_v9_0_get_coherence_flags(adev, mapping->bo_va->base.bo, + mapping, flags); } static unsigned gmc_v9_0_get_vbios_fb_size(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c index f87d0f6ffc93..f2b743a93915 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_5.c @@ -807,16 +807,5 @@ static void jpeg_v2_5_set_ras_funcs(struct amdgpu_device *adev) break; } - if (adev->jpeg.ras) { - amdgpu_ras_register_ras_block(adev, &adev->jpeg.ras->ras_block); - - strcpy(adev->jpeg.ras->ras_block.ras_comm.name, "jpeg"); - adev->jpeg.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__JPEG; - adev->jpeg.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__POISON; - adev->jpeg.ras_if = &adev->jpeg.ras->ras_block.ras_comm; - - /* If don't define special ras_late_init function, use default ras_late_init */ - if (!adev->jpeg.ras->ras_block.ras_late_init) - adev->jpeg.ras->ras_block.ras_late_init = amdgpu_ras_block_late_init; - } + jpeg_set_ras_funcs(adev); } diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c index 63b0d0b810ec..3beb731b2ce5 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.c @@ -27,6 +27,7 @@ #include "soc15.h" #include "soc15d.h" #include "jpeg_v2_0.h" +#include "jpeg_v4_0.h" #include "vcn/vcn_4_0_0_offset.h" #include "vcn/vcn_4_0_0_sh_mask.h" @@ -38,6 +39,7 @@ static void jpeg_v4_0_set_dec_ring_funcs(struct amdgpu_device *adev); static void jpeg_v4_0_set_irq_funcs(struct amdgpu_device *adev); static int jpeg_v4_0_set_powergating_state(void *handle, enum amd_powergating_state state); +static void jpeg_v4_0_set_ras_funcs(struct amdgpu_device *adev); /** * jpeg_v4_0_early_init - set function pointers @@ -55,6 +57,7 @@ static int jpeg_v4_0_early_init(void *handle) jpeg_v4_0_set_dec_ring_funcs(adev); jpeg_v4_0_set_irq_funcs(adev); + jpeg_v4_0_set_ras_funcs(adev); return 0; } @@ -78,6 +81,18 @@ static int jpeg_v4_0_sw_init(void *handle) if (r) return r; + /* JPEG DJPEG POISON EVENT */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, + VCN_4_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst->irq); + if (r) + return r; + + /* JPEG EJPEG POISON EVENT */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, + VCN_4_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst->irq); + if (r) + return r; + r = amdgpu_jpeg_sw_init(adev); if (r) return r; @@ -167,6 +182,8 @@ static int jpeg_v4_0_hw_fini(void *handle) RREG32_SOC15(JPEG, 0, regUVD_JRBC_STATUS)) jpeg_v4_0_set_powergating_state(adev, AMD_PG_STATE_GATE); + amdgpu_irq_put(adev, &adev->jpeg.inst->irq, 0); + return 0; } @@ -524,6 +541,10 @@ static int jpeg_v4_0_process_interrupt(struct amdgpu_device *adev, case VCN_4_0__SRCID__JPEG_DECODE: amdgpu_fence_process(&adev->jpeg.inst->ring_dec); break; + case VCN_4_0__SRCID_DJPEG0_POISON: + case VCN_4_0__SRCID_EJPEG0_POISON: + amdgpu_jpeg_process_poison_irq(adev, source, entry); + break; default: DRM_DEV_ERROR(adev->dev, "Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]); @@ -607,3 +628,63 @@ const struct amdgpu_ip_block_version jpeg_v4_0_ip_block = { .rev = 0, .funcs = &jpeg_v4_0_ip_funcs, }; + +static uint32_t jpeg_v4_0_query_poison_by_instance(struct amdgpu_device *adev, + uint32_t instance, uint32_t sub_block) +{ + uint32_t poison_stat = 0, reg_value = 0; + + switch (sub_block) { + case AMDGPU_JPEG_V4_0_JPEG0: + reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG0_STATUS); + poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG0_STATUS, POISONED_PF); + break; + case AMDGPU_JPEG_V4_0_JPEG1: + reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG1_STATUS); + poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG1_STATUS, POISONED_PF); + break; + default: + break; + } + + if (poison_stat) + dev_info(adev->dev, "Poison detected in JPEG%d sub_block%d\n", + instance, sub_block); + + return poison_stat; +} + +static bool jpeg_v4_0_query_ras_poison_status(struct amdgpu_device *adev) +{ + uint32_t inst = 0, sub = 0, poison_stat = 0; + + for (inst = 0; inst < adev->jpeg.num_jpeg_inst; inst++) + for (sub = 0; sub < AMDGPU_JPEG_V4_0_MAX_SUB_BLOCK; sub++) + poison_stat += + jpeg_v4_0_query_poison_by_instance(adev, inst, sub); + + return !!poison_stat; +} + +const struct amdgpu_ras_block_hw_ops jpeg_v4_0_ras_hw_ops = { + .query_poison_status = jpeg_v4_0_query_ras_poison_status, +}; + +static struct amdgpu_jpeg_ras jpeg_v4_0_ras = { + .ras_block = { + .hw_ops = &jpeg_v4_0_ras_hw_ops, + }, +}; + +static void jpeg_v4_0_set_ras_funcs(struct amdgpu_device *adev) +{ + switch (adev->ip_versions[JPEG_HWIP][0]) { + case IP_VERSION(4, 0, 0): + adev->jpeg.ras = &jpeg_v4_0_ras; + break; + default: + break; + } + + jpeg_set_ras_funcs(adev); +} diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.h index f1ed6ccfedca..07d36c2abd6b 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.h +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0.h @@ -24,6 +24,13 @@ #ifndef __JPEG_V4_0_H__ #define __JPEG_V4_0_H__ +enum amdgpu_jpeg_v4_0_sub_block { + AMDGPU_JPEG_V4_0_JPEG0 = 0, + AMDGPU_JPEG_V4_0_JPEG1, + + AMDGPU_JPEG_V4_0_MAX_SUB_BLOCK, +}; + extern const struct amdgpu_ip_block_version jpeg_v4_0_ip_block; #endif /* __JPEG_V4_0_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c index 067d10073a56..614394118a53 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v10_1.c @@ -121,6 +121,10 @@ static int mes_v10_1_submit_pkt_and_poll_completion(struct amdgpu_mes *mes, if (r < 1) { DRM_ERROR("MES failed to response msg=%d\n", x_pkt->header.opcode); + + while (halt_if_hws_hang) + schedule(); + return -ETIMEDOUT; } @@ -415,10 +419,6 @@ static int mes_v10_1_init_microcode(struct amdgpu_device *adev, mes_hdr = (const struct mes_firmware_header_v1_0 *) adev->mes.fw[pipe]->data; - adev->mes.ucode_fw_version[pipe] = - le32_to_cpu(mes_hdr->mes_ucode_version); - adev->mes.ucode_fw_version[pipe] = - le32_to_cpu(mes_hdr->mes_ucode_data_version); adev->mes.uc_start_addr[pipe] = le32_to_cpu(mes_hdr->mes_uc_start_addr_lo) | ((uint64_t)(le32_to_cpu(mes_hdr->mes_uc_start_addr_hi)) << 32); diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c index fef7d020bc5f..8d9c1e841353 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c @@ -98,7 +98,14 @@ static int mes_v11_0_submit_pkt_and_poll_completion(struct amdgpu_mes *mes, struct amdgpu_device *adev = mes->adev; struct amdgpu_ring *ring = &mes->ring; unsigned long flags; + signed long timeout = adev->usec_timeout; + if (amdgpu_emu_mode) { + timeout *= 100; + } else if (amdgpu_sriov_vf(adev)) { + /* Worst case in sriov where all other 15 VF timeout, each VF needs about 600ms */ + timeout = 15 * 600 * 1000; + } BUG_ON(size % 4 != 0); spin_lock_irqsave(&mes->ring_lock, flags); @@ -118,10 +125,14 @@ static int mes_v11_0_submit_pkt_and_poll_completion(struct amdgpu_mes *mes, DRM_DEBUG("MES msg=%d was emitted\n", x_pkt->header.opcode); r = amdgpu_fence_wait_polling(ring, ring->fence_drv.sync_seq, - adev->usec_timeout * (amdgpu_emu_mode ? 100 : 1)); + timeout); if (r < 1) { DRM_ERROR("MES failed to response msg=%d\n", x_pkt->header.opcode); + + while (halt_if_hws_hang) + schedule(); + return -ETIMEDOUT; } @@ -478,10 +489,6 @@ static int mes_v11_0_init_microcode(struct amdgpu_device *adev, mes_hdr = (const struct mes_firmware_header_v1_0 *) adev->mes.fw[pipe]->data; - adev->mes.ucode_fw_version[pipe] = - le32_to_cpu(mes_hdr->mes_ucode_version); - adev->mes.ucode_fw_version[pipe] = - le32_to_cpu(mes_hdr->mes_ucode_data_version); adev->mes.uc_start_addr[pipe] = le32_to_cpu(mes_hdr->mes_uc_start_addr_lo) | ((uint64_t)(le32_to_cpu(mes_hdr->mes_uc_start_addr_hi)) << 32); @@ -1246,7 +1253,9 @@ static int mes_v11_0_kiq_hw_fini(struct amdgpu_device *adev) if (adev->mes.ring.sched.ready) mes_v11_0_kiq_dequeue_sched(adev); - mes_v11_0_enable(adev, false); + if (!amdgpu_sriov_vf(adev)) + mes_v11_0_enable(adev, false); + return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c index 4d304f22889e..998b5d17b271 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v2_0.c @@ -32,8 +32,6 @@ #include "gc/gc_10_1_0_offset.h" #include "soc15_common.h" -#define mmMM_ATC_L2_MISC_CG_Sienna_Cichlid 0x064d -#define mmMM_ATC_L2_MISC_CG_Sienna_Cichlid_BASE_IDX 0 #define mmDAGB0_CNTL_MISC2_Sienna_Cichlid 0x0070 #define mmDAGB0_CNTL_MISC2_Sienna_Cichlid_BASE_IDX 0 @@ -574,7 +572,6 @@ static void mmhub_v2_0_update_medium_grain_clock_gating(struct amdgpu_device *ad case IP_VERSION(2, 1, 0): case IP_VERSION(2, 1, 1): case IP_VERSION(2, 1, 2): - def = data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG_Sienna_Cichlid); def1 = data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2_Sienna_Cichlid); break; default: @@ -608,8 +605,6 @@ static void mmhub_v2_0_update_medium_grain_clock_gating(struct amdgpu_device *ad case IP_VERSION(2, 1, 0): case IP_VERSION(2, 1, 1): case IP_VERSION(2, 1, 2): - if (def != data) - WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG_Sienna_Cichlid, data); if (def1 != data1) WREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2_Sienna_Cichlid, data1); break; @@ -634,8 +629,8 @@ static void mmhub_v2_0_update_medium_grain_light_sleep(struct amdgpu_device *ade case IP_VERSION(2, 1, 0): case IP_VERSION(2, 1, 1): case IP_VERSION(2, 1, 2): - def = data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG_Sienna_Cichlid); - break; + /* There is no ATCL2 in MMHUB for 2.1.x */ + return; default: def = data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG); break; @@ -646,18 +641,8 @@ static void mmhub_v2_0_update_medium_grain_light_sleep(struct amdgpu_device *ade else data &= ~MM_ATC_L2_MISC_CG__MEM_LS_ENABLE_MASK; - if (def != data) { - switch (adev->ip_versions[MMHUB_HWIP][0]) { - case IP_VERSION(2, 1, 0): - case IP_VERSION(2, 1, 1): - case IP_VERSION(2, 1, 2): - WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG_Sienna_Cichlid, data); - break; - default: - WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG, data); - break; - } - } + if (def != data) + WREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG, data); } static int mmhub_v2_0_set_clockgating(struct amdgpu_device *adev, @@ -695,7 +680,10 @@ static void mmhub_v2_0_get_clockgating(struct amdgpu_device *adev, u64 *flags) case IP_VERSION(2, 1, 0): case IP_VERSION(2, 1, 1): case IP_VERSION(2, 1, 2): - data = RREG32_SOC15(MMHUB, 0, mmMM_ATC_L2_MISC_CG_Sienna_Cichlid); + /* There is no ATCL2 in MMHUB for 2.1.x. Keep the status + * based on DAGB + */ + data = MM_ATC_L2_MISC_CG__ENABLE_MASK; data1 = RREG32_SOC15(MMHUB, 0, mmDAGB0_CNTL_MISC2_Sienna_Cichlid); break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c index ed2293686f0d..9de46fa8f46c 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v10_0.c @@ -126,32 +126,6 @@ out: return err; } -static int psp_v10_0_ring_init(struct psp_context *psp, - enum psp_ring_type ring_type) -{ - int ret = 0; - struct psp_ring *ring; - struct amdgpu_device *adev = psp->adev; - - ring = &psp->km_ring; - - ring->ring_type = ring_type; - - /* allocate 4k Page of Local Frame Buffer memory for ring */ - ring->ring_size = 0x1000; - ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->firmware.rbuf, - &ring->ring_mem_mc_addr, - (void **)&ring->ring_mem); - if (ret) { - ring->ring_size = 0; - return ret; - } - - return 0; -} - static int psp_v10_0_ring_create(struct psp_context *psp, enum psp_ring_type ring_type) { @@ -245,7 +219,6 @@ static void psp_v10_0_ring_set_wptr(struct psp_context *psp, uint32_t value) static const struct psp_funcs psp_v10_0_funcs = { .init_microcode = psp_v10_0_init_microcode, - .ring_init = psp_v10_0_ring_init, .ring_create = psp_v10_0_ring_create, .ring_stop = psp_v10_0_ring_stop, .ring_destroy = psp_v10_0_ring_destroy, diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c index 9518b4394a6e..bd3e3e23a939 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c @@ -360,32 +360,6 @@ static int psp_v11_0_bootloader_load_sos(struct psp_context *psp) return ret; } -static int psp_v11_0_ring_init(struct psp_context *psp, - enum psp_ring_type ring_type) -{ - int ret = 0; - struct psp_ring *ring; - struct amdgpu_device *adev = psp->adev; - - ring = &psp->km_ring; - - ring->ring_type = ring_type; - - /* allocate 4k Page of Local Frame Buffer memory for ring */ - ring->ring_size = 0x1000; - ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->firmware.rbuf, - &ring->ring_mem_mc_addr, - (void **)&ring->ring_mem); - if (ret) { - ring->ring_size = 0; - return ret; - } - - return 0; -} - static int psp_v11_0_ring_stop(struct psp_context *psp, enum psp_ring_type ring_type) { @@ -779,7 +753,6 @@ static const struct psp_funcs psp_v11_0_funcs = { .bootloader_load_spl = psp_v11_0_bootloader_load_spl, .bootloader_load_sysdrv = psp_v11_0_bootloader_load_sysdrv, .bootloader_load_sos = psp_v11_0_bootloader_load_sos, - .ring_init = psp_v11_0_ring_init, .ring_create = psp_v11_0_ring_create, .ring_stop = psp_v11_0_ring_stop, .ring_destroy = psp_v11_0_ring_destroy, diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.c index ff13e1beb49b..5697760a819b 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0_8.c @@ -28,32 +28,6 @@ #include "mp/mp_11_0_8_offset.h" -static int psp_v11_0_8_ring_init(struct psp_context *psp, - enum psp_ring_type ring_type) -{ - int ret = 0; - struct psp_ring *ring; - struct amdgpu_device *adev = psp->adev; - - ring = &psp->km_ring; - - ring->ring_type = ring_type; - - /* allocate 4k Page of Local Frame Buffer memory for ring */ - ring->ring_size = 0x1000; - ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->firmware.rbuf, - &ring->ring_mem_mc_addr, - (void **)&ring->ring_mem); - if (ret) { - ring->ring_size = 0; - return ret; - } - - return 0; -} - static int psp_v11_0_8_ring_stop(struct psp_context *psp, enum psp_ring_type ring_type) { @@ -194,7 +168,6 @@ static void psp_v11_0_8_ring_set_wptr(struct psp_context *psp, uint32_t value) } static const struct psp_funcs psp_v11_0_8_funcs = { - .ring_init = psp_v11_0_8_ring_init, .ring_create = psp_v11_0_8_ring_create, .ring_stop = psp_v11_0_8_ring_stop, .ring_destroy = psp_v11_0_8_ring_destroy, diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c index 0b2ac418e4ac..8ed2281b6557 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v12_0.c @@ -236,34 +236,6 @@ static void psp_v12_0_reroute_ih(struct psp_context *psp) 0x80000000, 0x8000FFFF, false); } -static int psp_v12_0_ring_init(struct psp_context *psp, - enum psp_ring_type ring_type) -{ - int ret = 0; - struct psp_ring *ring; - struct amdgpu_device *adev = psp->adev; - - psp_v12_0_reroute_ih(psp); - - ring = &psp->km_ring; - - ring->ring_type = ring_type; - - /* allocate 4k Page of Local Frame Buffer memory for ring */ - ring->ring_size = 0x1000; - ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->firmware.rbuf, - &ring->ring_mem_mc_addr, - (void **)&ring->ring_mem); - if (ret) { - ring->ring_size = 0; - return ret; - } - - return 0; -} - static int psp_v12_0_ring_create(struct psp_context *psp, enum psp_ring_type ring_type) { @@ -272,6 +244,8 @@ static int psp_v12_0_ring_create(struct psp_context *psp, struct psp_ring *ring = &psp->km_ring; struct amdgpu_device *adev = psp->adev; + psp_v12_0_reroute_ih(psp); + if (amdgpu_sriov_vf(psp->adev)) { /* Write low address of the ring to C2PMSG_102 */ psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr); @@ -425,7 +399,6 @@ static const struct psp_funcs psp_v12_0_funcs = { .init_microcode = psp_v12_0_init_microcode, .bootloader_load_sysdrv = psp_v12_0_bootloader_load_sysdrv, .bootloader_load_sos = psp_v12_0_bootloader_load_sos, - .ring_init = psp_v12_0_ring_init, .ring_create = psp_v12_0_ring_create, .ring_stop = psp_v12_0_ring_stop, .ring_destroy = psp_v12_0_ring_destroy, diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c index 21d822b1d589..ee27bfaba6fd 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c @@ -45,6 +45,7 @@ MODULE_FIRMWARE("amdgpu/psp_13_0_0_ta.bin"); MODULE_FIRMWARE("amdgpu/psp_13_0_7_sos.bin"); MODULE_FIRMWARE("amdgpu/psp_13_0_7_ta.bin"); MODULE_FIRMWARE("amdgpu/psp_13_0_10_sos.bin"); +MODULE_FIRMWARE("amdgpu/psp_13_0_10_ta.bin"); /* For large FW files the time to complete can be very long */ #define USBC_PD_POLLING_LIMIT_S 240 @@ -267,32 +268,6 @@ static int psp_v13_0_bootloader_load_sos(struct psp_context *psp) return ret; } -static int psp_v13_0_ring_init(struct psp_context *psp, - enum psp_ring_type ring_type) -{ - int ret = 0; - struct psp_ring *ring; - struct amdgpu_device *adev = psp->adev; - - ring = &psp->km_ring; - - ring->ring_type = ring_type; - - /* allocate 4k Page of Local Frame Buffer memory for ring */ - ring->ring_size = 0x1000; - ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->firmware.rbuf, - &ring->ring_mem_mc_addr, - (void **)&ring->ring_mem); - if (ret) { - ring->ring_size = 0; - return ret; - } - - return 0; -} - static int psp_v13_0_ring_stop(struct psp_context *psp, enum psp_ring_type ring_type) { @@ -728,7 +703,6 @@ static const struct psp_funcs psp_v13_0_funcs = { .bootloader_load_dbg_drv = psp_v13_0_bootloader_load_dbg_drv, .bootloader_load_ras_drv = psp_v13_0_bootloader_load_ras_drv, .bootloader_load_sos = psp_v13_0_bootloader_load_sos, - .ring_init = psp_v13_0_ring_init, .ring_create = psp_v13_0_ring_create, .ring_stop = psp_v13_0_ring_stop, .ring_destroy = psp_v13_0_ring_destroy, diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c index 321089dfa7db..9d4e24e518e8 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0_4.c @@ -199,32 +199,6 @@ static int psp_v13_0_4_bootloader_load_sos(struct psp_context *psp) return ret; } -static int psp_v13_0_4_ring_init(struct psp_context *psp, - enum psp_ring_type ring_type) -{ - int ret = 0; - struct psp_ring *ring; - struct amdgpu_device *adev = psp->adev; - - ring = &psp->km_ring; - - ring->ring_type = ring_type; - - /* allocate 4k Page of Local Frame Buffer memory for ring */ - ring->ring_size = 0x1000; - ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->firmware.rbuf, - &ring->ring_mem_mc_addr, - (void **)&ring->ring_mem); - if (ret) { - ring->ring_size = 0; - return ret; - } - - return 0; -} - static int psp_v13_0_4_ring_stop(struct psp_context *psp, enum psp_ring_type ring_type) { @@ -373,7 +347,6 @@ static const struct psp_funcs psp_v13_0_4_funcs = { .bootloader_load_intf_drv = psp_v13_0_4_bootloader_load_intf_drv, .bootloader_load_dbg_drv = psp_v13_0_4_bootloader_load_dbg_drv, .bootloader_load_sos = psp_v13_0_4_bootloader_load_sos, - .ring_init = psp_v13_0_4_ring_init, .ring_create = psp_v13_0_4_ring_create, .ring_stop = psp_v13_0_4_ring_stop, .ring_destroy = psp_v13_0_4_ring_destroy, diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c index 01f3bcc62a6c..157147c6c94e 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c @@ -160,32 +160,6 @@ static int psp_v3_1_bootloader_load_sos(struct psp_context *psp) return ret; } -static int psp_v3_1_ring_init(struct psp_context *psp, - enum psp_ring_type ring_type) -{ - int ret = 0; - struct psp_ring *ring; - struct amdgpu_device *adev = psp->adev; - - ring = &psp->km_ring; - - ring->ring_type = ring_type; - - /* allocate 4k Page of Local Frame Buffer memory for ring */ - ring->ring_size = 0x1000; - ret = amdgpu_bo_create_kernel(adev, ring->ring_size, PAGE_SIZE, - AMDGPU_GEM_DOMAIN_VRAM, - &adev->firmware.rbuf, - &ring->ring_mem_mc_addr, - (void **)&ring->ring_mem); - if (ret) { - ring->ring_size = 0; - return ret; - } - - return 0; -} - static void psp_v3_1_reroute_ih(struct psp_context *psp) { struct amdgpu_device *adev = psp->adev; @@ -401,7 +375,6 @@ static const struct psp_funcs psp_v3_1_funcs = { .init_microcode = psp_v3_1_init_microcode, .bootloader_load_sysdrv = psp_v3_1_bootloader_load_sysdrv, .bootloader_load_sos = psp_v3_1_bootloader_load_sos, - .ring_init = psp_v3_1_ring_init, .ring_create = psp_v3_1_ring_create, .ring_stop = psp_v3_1_ring_stop, .ring_destroy = psp_v3_1_ring_destroy, diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c index da3beb0bf2fa..049c26a45d85 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c @@ -455,6 +455,9 @@ static void sdma_v6_0_enable(struct amdgpu_device *adev, bool enable) sdma_v6_0_rlc_stop(adev); } + if (amdgpu_sriov_vf(adev)) + return; + for (i = 0; i < adev->sdma.num_instances; i++) { f32_cntl = RREG32_SOC15_IP(GC, sdma_v6_0_get_reg_offset(adev, i, regSDMA0_F32_CNTL)); f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_F32_CNTL, HALT, enable ? 0 : 1); @@ -1523,6 +1526,7 @@ static const struct amdgpu_ring_funcs sdma_v6_0_ring_funcs = { .align_mask = 0xf, .nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP), .support_64bit_ptrs = true, + .secure_submission_supported = true, .vmhub = AMDGPU_GFXHUB_0, .get_rptr = sdma_v6_0_ring_get_rptr, .get_wptr = sdma_v6_0_ring_get_wptr, diff --git a/drivers/gpu/drm/amd/amdgpu/si_dma.c b/drivers/gpu/drm/amd/amdgpu/si_dma.c index 4d5e718540aa..abca8b529721 100644 --- a/drivers/gpu/drm/amd/amdgpu/si_dma.c +++ b/drivers/gpu/drm/amd/amdgpu/si_dma.c @@ -112,14 +112,12 @@ static void si_dma_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq, static void si_dma_stop(struct amdgpu_device *adev) { - struct amdgpu_ring *ring; u32 rb_cntl; unsigned i; amdgpu_sdma_unset_buffer_funcs_helper(adev); for (i = 0; i < adev->sdma.num_instances; i++) { - ring = &adev->sdma.instance[i].ring; /* dma0 */ rb_cntl = RREG32(DMA_RB_CNTL + sdma_offsets[i]); rb_cntl &= ~DMA_RB_ENABLE; diff --git a/drivers/gpu/drm/amd/amdgpu/soc21.c b/drivers/gpu/drm/amd/amdgpu/soc21.c index e08044008186..b258e9aa0558 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc21.c +++ b/drivers/gpu/drm/amd/amdgpu/soc21.c @@ -322,6 +322,7 @@ soc21_asic_reset_method(struct amdgpu_device *adev) switch (adev->ip_versions[MP1_HWIP][0]) { case IP_VERSION(13, 0, 0): case IP_VERSION(13, 0, 7): + case IP_VERSION(13, 0, 10): return AMD_RESET_METHOD_MODE1; case IP_VERSION(13, 0, 4): return AMD_RESET_METHOD_MODE2; @@ -584,10 +585,6 @@ static int soc21_common_early_init(void *handle) AMD_PG_SUPPORT_JPEG | AMD_PG_SUPPORT_ATHUB | AMD_PG_SUPPORT_MMHUB; - if (amdgpu_sriov_vf(adev)) { - adev->cg_flags = 0; - adev->pg_flags = 0; - } adev->external_rev_id = adev->rev_id + 0x1; // TODO: need update break; case IP_VERSION(11, 0, 2): @@ -645,11 +642,6 @@ static int soc21_common_early_init(void *handle) adev->pg_flags = AMD_PG_SUPPORT_VCN | AMD_PG_SUPPORT_VCN_DPG | AMD_PG_SUPPORT_JPEG; - if (amdgpu_sriov_vf(adev)) { - /* hypervisor control CG and PG enablement */ - adev->cg_flags = 0; - adev->pg_flags = 0; - } adev->external_rev_id = adev->rev_id + 0x20; break; default: @@ -657,6 +649,9 @@ static int soc21_common_early_init(void *handle) return -EINVAL; } + if (amdgpu_sriov_vf(adev)) + amdgpu_virt_init_setting(adev); + return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c index 5d5d031c9e7d..72fd963f178b 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.c @@ -187,9 +187,9 @@ static void umc_v6_7_ecc_info_query_ras_error_count(struct amdgpu_device *adev, } } -static void umc_v6_7_convert_error_address(struct amdgpu_device *adev, - struct ras_err_data *err_data, uint64_t err_addr, - uint32_t ch_inst, uint32_t umc_inst) +void umc_v6_7_convert_error_address(struct amdgpu_device *adev, + struct ras_err_data *err_data, uint64_t err_addr, + uint32_t ch_inst, uint32_t umc_inst) { uint32_t channel_index; uint64_t soc_pa, retired_page, column; @@ -553,5 +553,4 @@ struct amdgpu_umc_ras umc_v6_7_ras = { .query_ras_poison_mode = umc_v6_7_query_ras_poison_mode, .ecc_info_query_ras_error_count = umc_v6_7_ecc_info_query_ras_error_count, .ecc_info_query_ras_error_address = umc_v6_7_ecc_info_query_ras_error_address, - .convert_ras_error_address = umc_v6_7_convert_error_address, }; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h index fe41ed2f5945..105245d5b6e5 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v6_7.h @@ -71,5 +71,7 @@ extern const uint32_t umc_v6_7_channel_idx_tbl_second[UMC_V6_7_UMC_INSTANCE_NUM][UMC_V6_7_CHANNEL_INSTANCE_NUM]; extern const uint32_t umc_v6_7_channel_idx_tbl_first[UMC_V6_7_UMC_INSTANCE_NUM][UMC_V6_7_CHANNEL_INSTANCE_NUM]; - +void umc_v6_7_convert_error_address(struct amdgpu_device *adev, + struct ras_err_data *err_data, uint64_t err_addr, + uint32_t ch_inst, uint32_t umc_inst); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c index 91235df54e22..b7da4528cf0a 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.c @@ -46,6 +46,16 @@ const struct channelnum_map_colbit umc_v8_10_channelnum_map_colbit_table[] = { }; const uint32_t + umc_v8_10_channel_idx_tbl_ext0[] + [UMC_V8_10_UMC_INSTANCE_NUM] + [UMC_V8_10_CHANNEL_INSTANCE_NUM] = { + {{1, 5}, {7, 3}}, + {{14, 15}, {13, 12}}, + {{10, 11}, {9, 8}}, + {{6, 2}, {0, 4}} + }; + +const uint32_t umc_v8_10_channel_idx_tbl[] [UMC_V8_10_UMC_INSTANCE_NUM] [UMC_V8_10_CHANNEL_INSTANCE_NUM] = { diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v8_10.h b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.h index 849ede88e111..25eaf4af5fcf 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v8_10.h +++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.h @@ -66,5 +66,9 @@ extern const uint32_t [UMC_V8_10_UMC_INSTANCE_NUM] [UMC_V8_10_CHANNEL_INSTANCE_NUM]; +extern const uint32_t + umc_v8_10_channel_idx_tbl_ext0[] + [UMC_V8_10_UMC_INSTANCE_NUM] + [UMC_V8_10_CHANNEL_INSTANCE_NUM]; #endif diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index 8a7006d62a87..ce8374ee824d 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -2002,16 +2002,5 @@ static void vcn_v2_5_set_ras_funcs(struct amdgpu_device *adev) break; } - if (adev->vcn.ras) { - amdgpu_ras_register_ras_block(adev, &adev->vcn.ras->ras_block); - - strcpy(adev->vcn.ras->ras_block.ras_comm.name, "vcn"); - adev->vcn.ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__VCN; - adev->vcn.ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__POISON; - adev->vcn.ras_if = &adev->vcn.ras->ras_block.ras_comm; - - /* If don't define special ras_late_init function, use default ras_late_init */ - if (!adev->vcn.ras->ras_block.ras_late_init) - adev->vcn.ras->ras_block.ras_late_init = amdgpu_ras_block_late_init; - } + amdgpu_vcn_set_ras_funcs(adev); } diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c index 897a5ce9c9da..403d054cf51b 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c @@ -31,6 +31,7 @@ #include "soc15_hw_ip.h" #include "vcn_v2_0.h" #include "mmsch_v4_0.h" +#include "vcn_v4_0.h" #include "vcn/vcn_4_0_0_offset.h" #include "vcn/vcn_4_0_0_sh_mask.h" @@ -64,6 +65,7 @@ static int vcn_v4_0_set_powergating_state(void *handle, static int vcn_v4_0_pause_dpg_mode(struct amdgpu_device *adev, int inst_idx, struct dpg_pause_state *new_state); static void vcn_v4_0_unified_ring_set_wptr(struct amdgpu_ring *ring); +static void vcn_v4_0_set_ras_funcs(struct amdgpu_device *adev); /** * vcn_v4_0_early_init - set function pointers @@ -84,6 +86,7 @@ static int vcn_v4_0_early_init(void *handle) vcn_v4_0_set_unified_ring_funcs(adev); vcn_v4_0_set_irq_funcs(adev); + vcn_v4_0_set_ras_funcs(adev); return 0; } @@ -132,6 +135,12 @@ static int vcn_v4_0_sw_init(void *handle) if (r) return r; + /* VCN POISON TRAP */ + r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i], + VCN_4_0__SRCID_UVD_POISON, &adev->vcn.inst[i].irq); + if (r) + return r; + ring = &adev->vcn.inst[i].ring_enc[0]; ring->use_doorbell = true; if (amdgpu_sriov_vf(adev)) @@ -296,6 +305,7 @@ static int vcn_v4_0_hw_fini(void *handle) } } + amdgpu_irq_put(adev, &adev->vcn.inst[i].irq, 0); } return 0; @@ -1939,6 +1949,9 @@ static int vcn_v4_0_process_interrupt(struct amdgpu_device *adev, struct amdgpu_ case VCN_4_0__SRCID__UVD_ENC_GENERAL_PURPOSE: amdgpu_fence_process(&adev->vcn.inst[ip_instance].ring_enc[0]); break; + case VCN_4_0__SRCID_UVD_POISON: + amdgpu_vcn_process_poison_irq(adev, source, entry); + break; default: DRM_ERROR("Unhandled interrupt: %d %d\n", entry->src_id, entry->src_data[0]); @@ -2001,3 +2014,60 @@ const struct amdgpu_ip_block_version vcn_v4_0_ip_block = .rev = 0, .funcs = &vcn_v4_0_ip_funcs, }; + +static uint32_t vcn_v4_0_query_poison_by_instance(struct amdgpu_device *adev, + uint32_t instance, uint32_t sub_block) +{ + uint32_t poison_stat = 0, reg_value = 0; + + switch (sub_block) { + case AMDGPU_VCN_V4_0_VCPU_VCODEC: + reg_value = RREG32_SOC15(VCN, instance, regUVD_RAS_VCPU_VCODEC_STATUS); + poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_VCPU_VCODEC_STATUS, POISONED_PF); + break; + default: + break; + } + + if (poison_stat) + dev_info(adev->dev, "Poison detected in VCN%d, sub_block%d\n", + instance, sub_block); + + return poison_stat; +} + +static bool vcn_v4_0_query_ras_poison_status(struct amdgpu_device *adev) +{ + uint32_t inst, sub; + uint32_t poison_stat = 0; + + for (inst = 0; inst < adev->vcn.num_vcn_inst; inst++) + for (sub = 0; sub < AMDGPU_VCN_V4_0_MAX_SUB_BLOCK; sub++) + poison_stat += + vcn_v4_0_query_poison_by_instance(adev, inst, sub); + + return !!poison_stat; +} + +const struct amdgpu_ras_block_hw_ops vcn_v4_0_ras_hw_ops = { + .query_poison_status = vcn_v4_0_query_ras_poison_status, +}; + +static struct amdgpu_vcn_ras vcn_v4_0_ras = { + .ras_block = { + .hw_ops = &vcn_v4_0_ras_hw_ops, + }, +}; + +static void vcn_v4_0_set_ras_funcs(struct amdgpu_device *adev) +{ + switch (adev->ip_versions[VCN_HWIP][0]) { + case IP_VERSION(4, 0, 0): + adev->vcn.ras = &vcn_v4_0_ras; + break; + default: + break; + } + + amdgpu_vcn_set_ras_funcs(adev); +} diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.h b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.h index 7c5c9d91bb52..7d3d11f40f27 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.h +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.h @@ -24,6 +24,12 @@ #ifndef __VCN_V4_0_H__ #define __VCN_V4_0_H__ +enum amdgpu_vcn_v4_0_sub_block { + AMDGPU_VCN_V4_0_VCPU_VCODEC = 0, + + AMDGPU_VCN_V4_0_MAX_SUB_BLOCK, +}; + extern const struct amdgpu_ip_block_version vcn_v4_0_ip_block; #endif /* __VCN_V4_0_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/vega20_ih.c b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c index 59dfca093155..1706081d054d 100644 --- a/drivers/gpu/drm/amd/amdgpu/vega20_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/vega20_ih.c @@ -267,7 +267,7 @@ static void vega20_ih_reroute_ih(struct amdgpu_device *adev) /* vega20 ih reroute will go through psp this * function is used for newer asics starting arcturus */ - if (adev->asic_type >= CHIP_ARCTURUS) { + if (adev->ip_versions[OSSSYS_HWIP][0] >= IP_VERSION(4, 2, 1)) { /* Reroute to IH ring 1 for VMC */ WREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_INDEX, 0x12); tmp = RREG32_SOC15(OSSSYS, 0, mmIH_CLIENT_CFG_DATA); @@ -308,7 +308,7 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev) adev->nbio.funcs->ih_control(adev); - if (adev->asic_type == CHIP_ARCTURUS && + if ((adev->ip_versions[OSSSYS_HWIP][0] == IP_VERSION(4, 2, 1)) && adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT) { ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN); if (adev->irq.ih.use_bus_addr) { @@ -321,7 +321,7 @@ static int vega20_ih_irq_init(struct amdgpu_device *adev) /* psp firmware won't program IH_CHICKEN for aldebaran * driver needs to program it properly according to * MC_SPACE type in IH_RB_CNTL */ - if (adev->asic_type == CHIP_ALDEBARAN) { + if (adev->ip_versions[OSSSYS_HWIP][0] == IP_VERSION(4, 4, 0)) { ih_chicken = RREG32_SOC15(OSSSYS, 0, mmIH_CHICKEN_ALDEBARAN); if (adev->irq.ih.use_bus_addr) { ih_chicken = REG_SET_FIELD(ih_chicken, IH_CHICKEN, diff --git a/drivers/gpu/drm/amd/amdgpu/vi.c b/drivers/gpu/drm/amd/amdgpu/vi.c index f6ffd7c96ff9..12ef782eb478 100644 --- a/drivers/gpu/drm/amd/amdgpu/vi.c +++ b/drivers/gpu/drm/amd/amdgpu/vi.c @@ -2111,6 +2111,8 @@ void vi_set_virt_ops(struct amdgpu_device *adev) int vi_set_ip_blocks(struct amdgpu_device *adev) { + amdgpu_device_set_sriov_virtual_display(adev); + switch (adev->asic_type) { case CHIP_TOPAZ: /* topaz has no DCE, UVD, VCE */ @@ -2130,7 +2132,7 @@ int vi_set_ip_blocks(struct amdgpu_device *adev) amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block); amdgpu_device_ip_block_add(adev, &sdma_v3_0_ip_block); amdgpu_device_ip_block_add(adev, &pp_smu_ip_block); - if (adev->enable_virtual_display || amdgpu_sriov_vf(adev)) + if (adev->enable_virtual_display) amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block); #if defined(CONFIG_DRM_AMD_DC) else if (amdgpu_device_has_dc_support(adev)) @@ -2150,7 +2152,7 @@ int vi_set_ip_blocks(struct amdgpu_device *adev) amdgpu_device_ip_block_add(adev, &gfx_v8_0_ip_block); amdgpu_device_ip_block_add(adev, &sdma_v3_0_ip_block); amdgpu_device_ip_block_add(adev, &pp_smu_ip_block); - if (adev->enable_virtual_display || amdgpu_sriov_vf(adev)) + if (adev->enable_virtual_display) amdgpu_device_ip_block_add(adev, &amdgpu_vkms_ip_block); #if defined(CONFIG_DRM_AMD_DC) else if (amdgpu_device_has_dc_support(adev)) diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h index c7118843db05..0c4c5499bb5c 100644 --- a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h +++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler.h @@ -2495,442 +2495,444 @@ static const uint32_t cwsr_trap_gfx10_hex[] = { 0xbf9f0000, 0x00000000, }; static const uint32_t cwsr_trap_gfx11_hex[] = { - 0xbfa00001, 0xbfa0021e, + 0xbfa00001, 0xbfa00221, 0xb0804006, 0xb8f8f802, 0x9178ff78, 0x00020006, - 0xb8fbf803, 0xbf0d9f6d, - 0xbfa20006, 0x8b6eff78, - 0x00002000, 0xbfa10009, - 0x8b6eff6d, 0x00ff0000, - 0xbfa2001e, 0x8b6eff7b, - 0x00000400, 0xbfa20041, - 0xbf830010, 0xb8fbf803, - 0xbfa0fffa, 0x8b6eff7b, - 0x00000900, 0xbfa20015, - 0x8b6eff7b, 0x000071ff, - 0xbfa10008, 0x8b6fff7b, - 0x00007080, 0xbfa10001, - 0xbeee1287, 0xb8eff801, - 0x846e8c6e, 0x8b6e6f6e, - 0xbfa2000a, 0x8b6eff6d, - 0x00ff0000, 0xbfa20007, - 0xb8eef801, 0x8b6eff6e, - 0x00000800, 0xbfa20003, + 0xb8fbf803, 0xbf0d9e6d, + 0xbfa10001, 0xbfbd0000, + 0xbf0d9f6d, 0xbfa20006, + 0x8b6eff78, 0x00002000, + 0xbfa10009, 0x8b6eff6d, + 0x00ff0000, 0xbfa2001e, 0x8b6eff7b, 0x00000400, - 0xbfa20026, 0xbefa4d82, - 0xbf89fc07, 0x84fa887a, - 0xf4005bbd, 0xf8000010, - 0xbf89fc07, 0x846e976e, - 0x9177ff77, 0x00800000, - 0x8c776e77, 0xf4045bbd, - 0xf8000000, 0xbf89fc07, - 0xf4045ebd, 0xf8000008, - 0xbf89fc07, 0x8bee6e6e, - 0xbfa10001, 0xbe80486e, - 0x8b6eff6d, 0x01ff0000, - 0xbfa20005, 0x8c78ff78, - 0x00002000, 0x80ec886c, - 0x82ed806d, 0xbfa00005, - 0x8b6eff6d, 0x01000000, - 0xbfa20002, 0x806c846c, - 0x826d806d, 0x8b6dff6d, - 0x0000ffff, 0x8bfe7e7e, - 0x8bea6a6a, 0xb978f802, - 0xbe804a6c, 0x8b6dff6d, - 0x0000ffff, 0xbefa0080, - 0xb97a0283, 0xbeee007e, - 0xbeef007f, 0xbefe0180, - 0xbefe4d84, 0xbf89fc07, - 0x8b7aff7f, 0x04000000, - 0x847a857a, 0x8c6d7a6d, - 0xbefa007e, 0x8b7bff7f, - 0x0000ffff, 0xbefe00c1, - 0xbeff00c1, 0xdca6c000, - 0x007a0000, 0x7e000280, - 0xbefe007a, 0xbeff007b, - 0xb8fb02dc, 0x847b997b, - 0xb8fa3b05, 0x807a817a, - 0xbf0d997b, 0xbfa20002, - 0x847a897a, 0xbfa00001, - 0x847a8a7a, 0xb8fb1e06, - 0x847b8a7b, 0x807a7b7a, + 0xbfa20041, 0xbf830010, + 0xb8fbf803, 0xbfa0fffa, + 0x8b6eff7b, 0x00000900, + 0xbfa20015, 0x8b6eff7b, + 0x000071ff, 0xbfa10008, + 0x8b6fff7b, 0x00007080, + 0xbfa10001, 0xbeee1287, + 0xb8eff801, 0x846e8c6e, + 0x8b6e6f6e, 0xbfa2000a, + 0x8b6eff6d, 0x00ff0000, + 0xbfa20007, 0xb8eef801, + 0x8b6eff6e, 0x00000800, + 0xbfa20003, 0x8b6eff7b, + 0x00000400, 0xbfa20026, + 0xbefa4d82, 0xbf89fc07, + 0x84fa887a, 0xf4005bbd, + 0xf8000010, 0xbf89fc07, + 0x846e976e, 0x9177ff77, + 0x00800000, 0x8c776e77, + 0xf4045bbd, 0xf8000000, + 0xbf89fc07, 0xf4045ebd, + 0xf8000008, 0xbf89fc07, + 0x8bee6e6e, 0xbfa10001, + 0xbe80486e, 0x8b6eff6d, + 0x01ff0000, 0xbfa20005, + 0x8c78ff78, 0x00002000, + 0x80ec886c, 0x82ed806d, + 0xbfa00005, 0x8b6eff6d, + 0x01000000, 0xbfa20002, + 0x806c846c, 0x826d806d, + 0x8b6dff6d, 0x0000ffff, + 0x8bfe7e7e, 0x8bea6a6a, + 0xb978f802, 0xbe804a6c, + 0x8b6dff6d, 0x0000ffff, + 0xbefa0080, 0xb97a0283, + 0xbeee007e, 0xbeef007f, + 0xbefe0180, 0xbefe4d84, + 0xbf89fc07, 0x8b7aff7f, + 0x04000000, 0x847a857a, + 0x8c6d7a6d, 0xbefa007e, 0x8b7bff7f, 0x0000ffff, - 0x807aff7a, 0x00000200, - 0x807a7e7a, 0x827b807b, - 0xd7610000, 0x00010870, - 0xd7610000, 0x00010a71, - 0xd7610000, 0x00010c72, - 0xd7610000, 0x00010e73, - 0xd7610000, 0x00011074, - 0xd7610000, 0x00011275, - 0xd7610000, 0x00011476, - 0xd7610000, 0x00011677, - 0xd7610000, 0x00011a79, - 0xd7610000, 0x00011c7e, - 0xd7610000, 0x00011e7f, - 0xbefe00ff, 0x00003fff, - 0xbeff0080, 0xdca6c040, - 0x007a0000, 0xd760007a, - 0x00011d00, 0xd760007b, - 0x00011f00, 0xbefe007a, - 0xbeff007b, 0xbef4007e, - 0x8b75ff7f, 0x0000ffff, - 0x8c75ff75, 0x00040000, - 0xbef60080, 0xbef700ff, - 0x10807fac, 0xbef1007d, - 0xbef00080, 0xb8f302dc, - 0x84739973, 0xbefe00c1, - 0x857d9973, 0x8b7d817d, - 0xbf06817d, 0xbfa20002, - 0xbeff0080, 0xbfa00002, - 0xbeff00c1, 0xbfa00009, + 0xbefe00c1, 0xbeff00c1, + 0xdca6c000, 0x007a0000, + 0x7e000280, 0xbefe007a, + 0xbeff007b, 0xb8fb02dc, + 0x847b997b, 0xb8fa3b05, + 0x807a817a, 0xbf0d997b, + 0xbfa20002, 0x847a897a, + 0xbfa00001, 0x847a8a7a, + 0xb8fb1e06, 0x847b8a7b, + 0x807a7b7a, 0x8b7bff7f, + 0x0000ffff, 0x807aff7a, + 0x00000200, 0x807a7e7a, + 0x827b807b, 0xd7610000, + 0x00010870, 0xd7610000, + 0x00010a71, 0xd7610000, + 0x00010c72, 0xd7610000, + 0x00010e73, 0xd7610000, + 0x00011074, 0xd7610000, + 0x00011275, 0xd7610000, + 0x00011476, 0xd7610000, + 0x00011677, 0xd7610000, + 0x00011a79, 0xd7610000, + 0x00011c7e, 0xd7610000, + 0x00011e7f, 0xbefe00ff, + 0x00003fff, 0xbeff0080, + 0xdca6c040, 0x007a0000, + 0xd760007a, 0x00011d00, + 0xd760007b, 0x00011f00, + 0xbefe007a, 0xbeff007b, + 0xbef4007e, 0x8b75ff7f, + 0x0000ffff, 0x8c75ff75, + 0x00040000, 0xbef60080, + 0xbef700ff, 0x10807fac, + 0xbef1007d, 0xbef00080, + 0xb8f302dc, 0x84739973, + 0xbefe00c1, 0x857d9973, + 0x8b7d817d, 0xbf06817d, + 0xbfa20002, 0xbeff0080, + 0xbfa00002, 0xbeff00c1, + 0xbfa00009, 0xbef600ff, + 0x01000000, 0xe0685080, + 0x701d0100, 0xe0685100, + 0x701d0200, 0xe0685180, + 0x701d0300, 0xbfa00008, 0xbef600ff, 0x01000000, - 0xe0685080, 0x701d0100, - 0xe0685100, 0x701d0200, - 0xe0685180, 0x701d0300, - 0xbfa00008, 0xbef600ff, - 0x01000000, 0xe0685100, - 0x701d0100, 0xe0685200, - 0x701d0200, 0xe0685300, - 0x701d0300, 0xb8f03b05, - 0x80708170, 0xbf0d9973, - 0xbfa20002, 0x84708970, - 0xbfa00001, 0x84708a70, - 0xb8fa1e06, 0x847a8a7a, - 0x80707a70, 0x8070ff70, - 0x00000200, 0xbef600ff, - 0x01000000, 0x7e000280, - 0x7e020280, 0x7e040280, - 0xbefd0080, 0xd7610002, - 0x0000fa71, 0x807d817d, - 0xd7610002, 0x0000fa6c, - 0x807d817d, 0x917aff6d, - 0x80000000, 0xd7610002, - 0x0000fa7a, 0x807d817d, - 0xd7610002, 0x0000fa6e, - 0x807d817d, 0xd7610002, - 0x0000fa6f, 0x807d817d, - 0xd7610002, 0x0000fa78, - 0x807d817d, 0xb8faf803, - 0xd7610002, 0x0000fa7a, - 0x807d817d, 0xd7610002, - 0x0000fa7b, 0x807d817d, - 0xb8f1f801, 0xd7610002, - 0x0000fa71, 0x807d817d, - 0xb8f1f814, 0xd7610002, - 0x0000fa71, 0x807d817d, - 0xb8f1f815, 0xd7610002, - 0x0000fa71, 0x807d817d, - 0xbefe00ff, 0x0000ffff, - 0xbeff0080, 0xe0685000, - 0x701d0200, 0xbefe00c1, + 0xe0685100, 0x701d0100, + 0xe0685200, 0x701d0200, + 0xe0685300, 0x701d0300, 0xb8f03b05, 0x80708170, 0xbf0d9973, 0xbfa20002, 0x84708970, 0xbfa00001, 0x84708a70, 0xb8fa1e06, 0x847a8a7a, 0x80707a70, + 0x8070ff70, 0x00000200, 0xbef600ff, 0x01000000, - 0xbef90080, 0xbefd0080, - 0xbf800000, 0xbe804100, - 0xbe824102, 0xbe844104, - 0xbe864106, 0xbe884108, - 0xbe8a410a, 0xbe8c410c, - 0xbe8e410e, 0xd7610002, - 0x0000f200, 0x80798179, - 0xd7610002, 0x0000f201, + 0x7e000280, 0x7e020280, + 0x7e040280, 0xbefd0080, + 0xd7610002, 0x0000fa71, + 0x807d817d, 0xd7610002, + 0x0000fa6c, 0x807d817d, + 0x917aff6d, 0x80000000, + 0xd7610002, 0x0000fa7a, + 0x807d817d, 0xd7610002, + 0x0000fa6e, 0x807d817d, + 0xd7610002, 0x0000fa6f, + 0x807d817d, 0xd7610002, + 0x0000fa78, 0x807d817d, + 0xb8faf803, 0xd7610002, + 0x0000fa7a, 0x807d817d, + 0xd7610002, 0x0000fa7b, + 0x807d817d, 0xb8f1f801, + 0xd7610002, 0x0000fa71, + 0x807d817d, 0xb8f1f814, + 0xd7610002, 0x0000fa71, + 0x807d817d, 0xb8f1f815, + 0xd7610002, 0x0000fa71, + 0x807d817d, 0xbefe00ff, + 0x0000ffff, 0xbeff0080, + 0xe0685000, 0x701d0200, + 0xbefe00c1, 0xb8f03b05, + 0x80708170, 0xbf0d9973, + 0xbfa20002, 0x84708970, + 0xbfa00001, 0x84708a70, + 0xb8fa1e06, 0x847a8a7a, + 0x80707a70, 0xbef600ff, + 0x01000000, 0xbef90080, + 0xbefd0080, 0xbf800000, + 0xbe804100, 0xbe824102, + 0xbe844104, 0xbe864106, + 0xbe884108, 0xbe8a410a, + 0xbe8c410c, 0xbe8e410e, + 0xd7610002, 0x0000f200, 0x80798179, 0xd7610002, - 0x0000f202, 0x80798179, - 0xd7610002, 0x0000f203, + 0x0000f201, 0x80798179, + 0xd7610002, 0x0000f202, 0x80798179, 0xd7610002, - 0x0000f204, 0x80798179, - 0xd7610002, 0x0000f205, + 0x0000f203, 0x80798179, + 0xd7610002, 0x0000f204, 0x80798179, 0xd7610002, - 0x0000f206, 0x80798179, - 0xd7610002, 0x0000f207, + 0x0000f205, 0x80798179, + 0xd7610002, 0x0000f206, 0x80798179, 0xd7610002, - 0x0000f208, 0x80798179, - 0xd7610002, 0x0000f209, + 0x0000f207, 0x80798179, + 0xd7610002, 0x0000f208, 0x80798179, 0xd7610002, - 0x0000f20a, 0x80798179, - 0xd7610002, 0x0000f20b, + 0x0000f209, 0x80798179, + 0xd7610002, 0x0000f20a, 0x80798179, 0xd7610002, - 0x0000f20c, 0x80798179, - 0xd7610002, 0x0000f20d, + 0x0000f20b, 0x80798179, + 0xd7610002, 0x0000f20c, 0x80798179, 0xd7610002, - 0x0000f20e, 0x80798179, - 0xd7610002, 0x0000f20f, - 0x80798179, 0xbf06a079, - 0xbfa10006, 0xe0685000, - 0x701d0200, 0x8070ff70, - 0x00000080, 0xbef90080, - 0x7e040280, 0x807d907d, - 0xbf0aff7d, 0x00000060, - 0xbfa2ffbc, 0xbe804100, - 0xbe824102, 0xbe844104, - 0xbe864106, 0xbe884108, - 0xbe8a410a, 0xd7610002, - 0x0000f200, 0x80798179, - 0xd7610002, 0x0000f201, + 0x0000f20d, 0x80798179, + 0xd7610002, 0x0000f20e, 0x80798179, 0xd7610002, - 0x0000f202, 0x80798179, - 0xd7610002, 0x0000f203, + 0x0000f20f, 0x80798179, + 0xbf06a079, 0xbfa10006, + 0xe0685000, 0x701d0200, + 0x8070ff70, 0x00000080, + 0xbef90080, 0x7e040280, + 0x807d907d, 0xbf0aff7d, + 0x00000060, 0xbfa2ffbc, + 0xbe804100, 0xbe824102, + 0xbe844104, 0xbe864106, + 0xbe884108, 0xbe8a410a, + 0xd7610002, 0x0000f200, 0x80798179, 0xd7610002, - 0x0000f204, 0x80798179, - 0xd7610002, 0x0000f205, + 0x0000f201, 0x80798179, + 0xd7610002, 0x0000f202, 0x80798179, 0xd7610002, - 0x0000f206, 0x80798179, - 0xd7610002, 0x0000f207, + 0x0000f203, 0x80798179, + 0xd7610002, 0x0000f204, 0x80798179, 0xd7610002, - 0x0000f208, 0x80798179, - 0xd7610002, 0x0000f209, + 0x0000f205, 0x80798179, + 0xd7610002, 0x0000f206, 0x80798179, 0xd7610002, - 0x0000f20a, 0x80798179, - 0xd7610002, 0x0000f20b, - 0x80798179, 0xe0685000, - 0x701d0200, 0xbefe00c1, - 0x857d9973, 0x8b7d817d, - 0xbf06817d, 0xbfa20002, - 0xbeff0080, 0xbfa00001, - 0xbeff00c1, 0xb8fb4306, - 0x8b7bc17b, 0xbfa10044, - 0xbfbd0000, 0x8b7aff6d, - 0x80000000, 0xbfa10040, - 0x847b867b, 0x847b827b, - 0xbef6007b, 0xb8f03b05, - 0x80708170, 0xbf0d9973, - 0xbfa20002, 0x84708970, - 0xbfa00001, 0x84708a70, - 0xb8fa1e06, 0x847a8a7a, - 0x80707a70, 0x8070ff70, - 0x00000200, 0x8070ff70, - 0x00000080, 0xbef600ff, - 0x01000000, 0xd71f0000, - 0x000100c1, 0xd7200000, - 0x000200c1, 0x16000084, - 0x857d9973, 0x8b7d817d, - 0xbf06817d, 0xbefd0080, - 0xbfa20012, 0xbe8300ff, - 0x00000080, 0xbf800000, - 0xbf800000, 0xbf800000, - 0xd8d80000, 0x01000000, - 0xbf890000, 0xe0685000, - 0x701d0100, 0x807d037d, - 0x80700370, 0xd5250000, - 0x0001ff00, 0x00000080, - 0xbf0a7b7d, 0xbfa2fff4, - 0xbfa00011, 0xbe8300ff, - 0x00000100, 0xbf800000, - 0xbf800000, 0xbf800000, - 0xd8d80000, 0x01000000, - 0xbf890000, 0xe0685000, - 0x701d0100, 0x807d037d, - 0x80700370, 0xd5250000, - 0x0001ff00, 0x00000100, - 0xbf0a7b7d, 0xbfa2fff4, + 0x0000f207, 0x80798179, + 0xd7610002, 0x0000f208, + 0x80798179, 0xd7610002, + 0x0000f209, 0x80798179, + 0xd7610002, 0x0000f20a, + 0x80798179, 0xd7610002, + 0x0000f20b, 0x80798179, + 0xe0685000, 0x701d0200, 0xbefe00c1, 0x857d9973, 0x8b7d817d, 0xbf06817d, - 0xbfa20004, 0xbef000ff, - 0x00000200, 0xbeff0080, - 0xbfa00003, 0xbef000ff, - 0x00000400, 0xbeff00c1, - 0xb8fb3b05, 0x807b817b, - 0x847b827b, 0x857d9973, + 0xbfa20002, 0xbeff0080, + 0xbfa00001, 0xbeff00c1, + 0xb8fb4306, 0x8b7bc17b, + 0xbfa10044, 0xbfbd0000, + 0x8b7aff6d, 0x80000000, + 0xbfa10040, 0x847b867b, + 0x847b827b, 0xbef6007b, + 0xb8f03b05, 0x80708170, + 0xbf0d9973, 0xbfa20002, + 0x84708970, 0xbfa00001, + 0x84708a70, 0xb8fa1e06, + 0x847a8a7a, 0x80707a70, + 0x8070ff70, 0x00000200, + 0x8070ff70, 0x00000080, + 0xbef600ff, 0x01000000, + 0xd71f0000, 0x000100c1, + 0xd7200000, 0x000200c1, + 0x16000084, 0x857d9973, 0x8b7d817d, 0xbf06817d, - 0xbfa20017, 0xbef600ff, - 0x01000000, 0xbefd0084, - 0xbf0a7b7d, 0xbfa10037, - 0x7e008700, 0x7e028701, - 0x7e048702, 0x7e068703, - 0xe0685000, 0x701d0000, - 0xe0685080, 0x701d0100, - 0xe0685100, 0x701d0200, - 0xe0685180, 0x701d0300, - 0x807d847d, 0x8070ff70, - 0x00000200, 0xbf0a7b7d, - 0xbfa2ffef, 0xbfa00025, + 0xbefd0080, 0xbfa20012, + 0xbe8300ff, 0x00000080, + 0xbf800000, 0xbf800000, + 0xbf800000, 0xd8d80000, + 0x01000000, 0xbf890000, + 0xe0685000, 0x701d0100, + 0x807d037d, 0x80700370, + 0xd5250000, 0x0001ff00, + 0x00000080, 0xbf0a7b7d, + 0xbfa2fff4, 0xbfa00011, + 0xbe8300ff, 0x00000100, + 0xbf800000, 0xbf800000, + 0xbf800000, 0xd8d80000, + 0x01000000, 0xbf890000, + 0xe0685000, 0x701d0100, + 0x807d037d, 0x80700370, + 0xd5250000, 0x0001ff00, + 0x00000100, 0xbf0a7b7d, + 0xbfa2fff4, 0xbefe00c1, + 0x857d9973, 0x8b7d817d, + 0xbf06817d, 0xbfa20004, + 0xbef000ff, 0x00000200, + 0xbeff0080, 0xbfa00003, + 0xbef000ff, 0x00000400, + 0xbeff00c1, 0xb8fb3b05, + 0x807b817b, 0x847b827b, + 0x857d9973, 0x8b7d817d, + 0xbf06817d, 0xbfa20017, 0xbef600ff, 0x01000000, 0xbefd0084, 0xbf0a7b7d, - 0xbfa10011, 0x7e008700, + 0xbfa10037, 0x7e008700, 0x7e028701, 0x7e048702, 0x7e068703, 0xe0685000, - 0x701d0000, 0xe0685100, - 0x701d0100, 0xe0685200, - 0x701d0200, 0xe0685300, + 0x701d0000, 0xe0685080, + 0x701d0100, 0xe0685100, + 0x701d0200, 0xe0685180, 0x701d0300, 0x807d847d, - 0x8070ff70, 0x00000400, + 0x8070ff70, 0x00000200, 0xbf0a7b7d, 0xbfa2ffef, - 0xb8fb1e06, 0x8b7bc17b, - 0xbfa1000c, 0x847b837b, - 0x807b7d7b, 0xbefe00c1, - 0xbeff0080, 0x7e008700, + 0xbfa00025, 0xbef600ff, + 0x01000000, 0xbefd0084, + 0xbf0a7b7d, 0xbfa10011, + 0x7e008700, 0x7e028701, + 0x7e048702, 0x7e068703, 0xe0685000, 0x701d0000, - 0x807d817d, 0x8070ff70, - 0x00000080, 0xbf0a7b7d, - 0xbfa2fff8, 0xbfa00146, - 0xbef4007e, 0x8b75ff7f, - 0x0000ffff, 0x8c75ff75, - 0x00040000, 0xbef60080, - 0xbef700ff, 0x10807fac, - 0xb8f202dc, 0x84729972, - 0x8b6eff7f, 0x04000000, - 0xbfa1003a, 0xbefe00c1, - 0x857d9972, 0x8b7d817d, - 0xbf06817d, 0xbfa20002, - 0xbeff0080, 0xbfa00001, - 0xbeff00c1, 0xb8ef4306, - 0x8b6fc16f, 0xbfa1002f, - 0x846f866f, 0x846f826f, - 0xbef6006f, 0xb8f83b05, - 0x80788178, 0xbf0d9972, - 0xbfa20002, 0x84788978, - 0xbfa00001, 0x84788a78, - 0xb8ee1e06, 0x846e8a6e, - 0x80786e78, 0x8078ff78, - 0x00000200, 0x8078ff78, - 0x00000080, 0xbef600ff, - 0x01000000, 0x857d9972, - 0x8b7d817d, 0xbf06817d, - 0xbefd0080, 0xbfa2000c, - 0xe0500000, 0x781d0000, - 0xbf8903f7, 0xdac00000, - 0x00000000, 0x807dff7d, - 0x00000080, 0x8078ff78, - 0x00000080, 0xbf0a6f7d, - 0xbfa2fff5, 0xbfa0000b, - 0xe0500000, 0x781d0000, - 0xbf8903f7, 0xdac00000, - 0x00000000, 0x807dff7d, - 0x00000100, 0x8078ff78, - 0x00000100, 0xbf0a6f7d, - 0xbfa2fff5, 0xbef80080, + 0xe0685100, 0x701d0100, + 0xe0685200, 0x701d0200, + 0xe0685300, 0x701d0300, + 0x807d847d, 0x8070ff70, + 0x00000400, 0xbf0a7b7d, + 0xbfa2ffef, 0xb8fb1e06, + 0x8b7bc17b, 0xbfa1000c, + 0x847b837b, 0x807b7d7b, + 0xbefe00c1, 0xbeff0080, + 0x7e008700, 0xe0685000, + 0x701d0000, 0x807d817d, + 0x8070ff70, 0x00000080, + 0xbf0a7b7d, 0xbfa2fff8, + 0xbfa00146, 0xbef4007e, + 0x8b75ff7f, 0x0000ffff, + 0x8c75ff75, 0x00040000, + 0xbef60080, 0xbef700ff, + 0x10807fac, 0xb8f202dc, + 0x84729972, 0x8b6eff7f, + 0x04000000, 0xbfa1003a, 0xbefe00c1, 0x857d9972, 0x8b7d817d, 0xbf06817d, 0xbfa20002, 0xbeff0080, 0xbfa00001, 0xbeff00c1, - 0xb8ef3b05, 0x806f816f, - 0x846f826f, 0x857d9972, - 0x8b7d817d, 0xbf06817d, - 0xbfa20024, 0xbef600ff, - 0x01000000, 0xbeee0078, + 0xb8ef4306, 0x8b6fc16f, + 0xbfa1002f, 0x846f866f, + 0x846f826f, 0xbef6006f, + 0xb8f83b05, 0x80788178, + 0xbf0d9972, 0xbfa20002, + 0x84788978, 0xbfa00001, + 0x84788a78, 0xb8ee1e06, + 0x846e8a6e, 0x80786e78, 0x8078ff78, 0x00000200, - 0xbefd0084, 0xbf0a6f7d, - 0xbfa10050, 0xe0505000, - 0x781d0000, 0xe0505080, - 0x781d0100, 0xe0505100, - 0x781d0200, 0xe0505180, - 0x781d0300, 0xbf8903f7, - 0x7e008500, 0x7e028501, - 0x7e048502, 0x7e068503, - 0x807d847d, 0x8078ff78, - 0x00000200, 0xbf0a6f7d, - 0xbfa2ffee, 0xe0505000, - 0x6e1d0000, 0xe0505080, - 0x6e1d0100, 0xe0505100, - 0x6e1d0200, 0xe0505180, - 0x6e1d0300, 0xbf8903f7, - 0xbfa00034, 0xbef600ff, - 0x01000000, 0xbeee0078, - 0x8078ff78, 0x00000400, - 0xbefd0084, 0xbf0a6f7d, - 0xbfa10012, 0xe0505000, - 0x781d0000, 0xe0505100, - 0x781d0100, 0xe0505200, - 0x781d0200, 0xe0505300, - 0x781d0300, 0xbf8903f7, - 0x7e008500, 0x7e028501, - 0x7e048502, 0x7e068503, - 0x807d847d, 0x8078ff78, - 0x00000400, 0xbf0a6f7d, - 0xbfa2ffee, 0xb8ef1e06, - 0x8b6fc16f, 0xbfa1000e, - 0x846f836f, 0x806f7d6f, - 0xbefe00c1, 0xbeff0080, + 0x8078ff78, 0x00000080, + 0xbef600ff, 0x01000000, + 0x857d9972, 0x8b7d817d, + 0xbf06817d, 0xbefd0080, + 0xbfa2000c, 0xe0500000, + 0x781d0000, 0xbf8903f7, + 0xdac00000, 0x00000000, + 0x807dff7d, 0x00000080, + 0x8078ff78, 0x00000080, + 0xbf0a6f7d, 0xbfa2fff5, + 0xbfa0000b, 0xe0500000, + 0x781d0000, 0xbf8903f7, + 0xdac00000, 0x00000000, + 0x807dff7d, 0x00000100, + 0x8078ff78, 0x00000100, + 0xbf0a6f7d, 0xbfa2fff5, + 0xbef80080, 0xbefe00c1, + 0x857d9972, 0x8b7d817d, + 0xbf06817d, 0xbfa20002, + 0xbeff0080, 0xbfa00001, + 0xbeff00c1, 0xb8ef3b05, + 0x806f816f, 0x846f826f, + 0x857d9972, 0x8b7d817d, + 0xbf06817d, 0xbfa20024, + 0xbef600ff, 0x01000000, + 0xbeee0078, 0x8078ff78, + 0x00000200, 0xbefd0084, + 0xbf0a6f7d, 0xbfa10050, 0xe0505000, 0x781d0000, + 0xe0505080, 0x781d0100, + 0xe0505100, 0x781d0200, + 0xe0505180, 0x781d0300, 0xbf8903f7, 0x7e008500, - 0x807d817d, 0x8078ff78, - 0x00000080, 0xbf0a6f7d, - 0xbfa2fff7, 0xbeff00c1, + 0x7e028501, 0x7e048502, + 0x7e068503, 0x807d847d, + 0x8078ff78, 0x00000200, + 0xbf0a6f7d, 0xbfa2ffee, 0xe0505000, 0x6e1d0000, - 0xe0505100, 0x6e1d0100, - 0xe0505200, 0x6e1d0200, - 0xe0505300, 0x6e1d0300, - 0xbf8903f7, 0xb8f83b05, - 0x80788178, 0xbf0d9972, - 0xbfa20002, 0x84788978, - 0xbfa00001, 0x84788a78, - 0xb8ee1e06, 0x846e8a6e, - 0x80786e78, 0x8078ff78, - 0x00000200, 0x80f8ff78, - 0x00000050, 0xbef600ff, - 0x01000000, 0xbefd00ff, - 0x0000006c, 0x80f89078, - 0xf428403a, 0xf0000000, - 0xbf89fc07, 0x80fd847d, - 0xbf800000, 0xbe804300, - 0xbe824302, 0x80f8a078, - 0xf42c403a, 0xf0000000, - 0xbf89fc07, 0x80fd887d, - 0xbf800000, 0xbe804300, - 0xbe824302, 0xbe844304, - 0xbe864306, 0x80f8c078, - 0xf430403a, 0xf0000000, - 0xbf89fc07, 0x80fd907d, - 0xbf800000, 0xbe804300, - 0xbe824302, 0xbe844304, - 0xbe864306, 0xbe884308, - 0xbe8a430a, 0xbe8c430c, - 0xbe8e430e, 0xbf06807d, - 0xbfa1fff0, 0xb980f801, - 0x00000000, 0xbfbd0000, + 0xe0505080, 0x6e1d0100, + 0xe0505100, 0x6e1d0200, + 0xe0505180, 0x6e1d0300, + 0xbf8903f7, 0xbfa00034, + 0xbef600ff, 0x01000000, + 0xbeee0078, 0x8078ff78, + 0x00000400, 0xbefd0084, + 0xbf0a6f7d, 0xbfa10012, + 0xe0505000, 0x781d0000, + 0xe0505100, 0x781d0100, + 0xe0505200, 0x781d0200, + 0xe0505300, 0x781d0300, + 0xbf8903f7, 0x7e008500, + 0x7e028501, 0x7e048502, + 0x7e068503, 0x807d847d, + 0x8078ff78, 0x00000400, + 0xbf0a6f7d, 0xbfa2ffee, + 0xb8ef1e06, 0x8b6fc16f, + 0xbfa1000e, 0x846f836f, + 0x806f7d6f, 0xbefe00c1, + 0xbeff0080, 0xe0505000, + 0x781d0000, 0xbf8903f7, + 0x7e008500, 0x807d817d, + 0x8078ff78, 0x00000080, + 0xbf0a6f7d, 0xbfa2fff7, + 0xbeff00c1, 0xe0505000, + 0x6e1d0000, 0xe0505100, + 0x6e1d0100, 0xe0505200, + 0x6e1d0200, 0xe0505300, + 0x6e1d0300, 0xbf8903f7, 0xb8f83b05, 0x80788178, 0xbf0d9972, 0xbfa20002, 0x84788978, 0xbfa00001, 0x84788a78, 0xb8ee1e06, 0x846e8a6e, 0x80786e78, 0x8078ff78, 0x00000200, + 0x80f8ff78, 0x00000050, 0xbef600ff, 0x01000000, - 0xf4205bfa, 0xf0000000, - 0x80788478, 0xf4205b3a, + 0xbefd00ff, 0x0000006c, + 0x80f89078, 0xf428403a, + 0xf0000000, 0xbf89fc07, + 0x80fd847d, 0xbf800000, + 0xbe804300, 0xbe824302, + 0x80f8a078, 0xf42c403a, + 0xf0000000, 0xbf89fc07, + 0x80fd887d, 0xbf800000, + 0xbe804300, 0xbe824302, + 0xbe844304, 0xbe864306, + 0x80f8c078, 0xf430403a, + 0xf0000000, 0xbf89fc07, + 0x80fd907d, 0xbf800000, + 0xbe804300, 0xbe824302, + 0xbe844304, 0xbe864306, + 0xbe884308, 0xbe8a430a, + 0xbe8c430c, 0xbe8e430e, + 0xbf06807d, 0xbfa1fff0, + 0xb980f801, 0x00000000, + 0xbfbd0000, 0xb8f83b05, + 0x80788178, 0xbf0d9972, + 0xbfa20002, 0x84788978, + 0xbfa00001, 0x84788a78, + 0xb8ee1e06, 0x846e8a6e, + 0x80786e78, 0x8078ff78, + 0x00000200, 0xbef600ff, + 0x01000000, 0xf4205bfa, 0xf0000000, 0x80788478, - 0xf4205b7a, 0xf0000000, - 0x80788478, 0xf4205c3a, + 0xf4205b3a, 0xf0000000, + 0x80788478, 0xf4205b7a, 0xf0000000, 0x80788478, - 0xf4205c7a, 0xf0000000, - 0x80788478, 0xf4205eba, + 0xf4205c3a, 0xf0000000, + 0x80788478, 0xf4205c7a, 0xf0000000, 0x80788478, - 0xf4205efa, 0xf0000000, - 0x80788478, 0xf4205e7a, + 0xf4205eba, 0xf0000000, + 0x80788478, 0xf4205efa, 0xf0000000, 0x80788478, - 0xf4205cfa, 0xf0000000, - 0x80788478, 0xf4205bba, + 0xf4205e7a, 0xf0000000, + 0x80788478, 0xf4205cfa, 0xf0000000, 0x80788478, - 0xbf89fc07, 0xb96ef814, 0xf4205bba, 0xf0000000, 0x80788478, 0xbf89fc07, - 0xb96ef815, 0xbefd006f, - 0xbefe0070, 0xbeff0071, - 0x8b6f7bff, 0x000003ff, - 0xb96f4803, 0x8b6f7bff, - 0xfffff800, 0x856f8b6f, - 0xb96fa2c3, 0xb973f801, - 0xb8ee3b05, 0x806e816e, - 0xbf0d9972, 0xbfa20002, - 0x846e896e, 0xbfa00001, - 0x846e8a6e, 0xb8ef1e06, - 0x846f8a6f, 0x806e6f6e, - 0x806eff6e, 0x00000200, - 0x806e746e, 0x826f8075, - 0x8b6fff6f, 0x0000ffff, - 0xf4085c37, 0xf8000050, - 0xf4085d37, 0xf8000060, - 0xf4005e77, 0xf8000074, - 0xbf89fc07, 0x8b6dff6d, - 0x0000ffff, 0x8bfe7e7e, - 0x8bea6a6a, 0xb8eef802, - 0xbf0d866e, 0xbfa20002, - 0xb97af802, 0xbe80486c, - 0xb97af802, 0xbe804a6c, - 0xbfb00000, 0xbf9f0000, + 0xb96ef814, 0xf4205bba, + 0xf0000000, 0x80788478, + 0xbf89fc07, 0xb96ef815, + 0xbefd006f, 0xbefe0070, + 0xbeff0071, 0x8b6f7bff, + 0x000003ff, 0xb96f4803, + 0x8b6f7bff, 0xfffff800, + 0x856f8b6f, 0xb96fa2c3, + 0xb973f801, 0xb8ee3b05, + 0x806e816e, 0xbf0d9972, + 0xbfa20002, 0x846e896e, + 0xbfa00001, 0x846e8a6e, + 0xb8ef1e06, 0x846f8a6f, + 0x806e6f6e, 0x806eff6e, + 0x00000200, 0x806e746e, + 0x826f8075, 0x8b6fff6f, + 0x0000ffff, 0xf4085c37, + 0xf8000050, 0xf4085d37, + 0xf8000060, 0xf4005e77, + 0xf8000074, 0xbf89fc07, + 0x8b6dff6d, 0x0000ffff, + 0x8bfe7e7e, 0x8bea6a6a, + 0xb8eef802, 0xbf0d866e, + 0xbfa20002, 0xb97af802, + 0xbe80486c, 0xb97af802, + 0xbe804a6c, 0xbfb00000, 0xbf9f0000, 0xbf9f0000, 0xbf9f0000, 0xbf9f0000, + 0xbf9f0000, 0x00000000, }; diff --git a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm index 0f81670f6f9c..8b92c33c2a7c 100644 --- a/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm +++ b/drivers/gpu/drm/amd/amdkfd/cwsr_trap_handler_gfx10.asm @@ -186,6 +186,12 @@ L_SKIP_RESTORE: s_getreg_b32 s_save_trapsts, hwreg(HW_REG_TRAPSTS) #if SW_SA_TRAP + // If ttmp1[30] is set then issue s_barrier to unblock dependent waves. + s_bitcmp1_b32 s_save_pc_hi, 30 + s_cbranch_scc0 L_TRAP_NO_BARRIER + s_barrier + +L_TRAP_NO_BARRIER: // If ttmp1[31] is set then trap may occur early. // Spin wait until SAVECTX exception is raised. s_bitcmp1_b32 s_save_pc_hi, 31 diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c index 5feaba6a77de..6d291aa6386b 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c @@ -1950,7 +1950,7 @@ static int criu_checkpoint(struct file *filep, { int ret; uint32_t num_devices, num_bos, num_objects; - uint64_t priv_size, priv_offset = 0; + uint64_t priv_size, priv_offset = 0, bo_priv_offset; if (!args->devices || !args->bos || !args->priv_data) return -EINVAL; @@ -1994,38 +1994,34 @@ static int criu_checkpoint(struct file *filep, if (ret) goto exit_unlock; - ret = criu_checkpoint_bos(p, num_bos, (uint8_t __user *)args->bos, - (uint8_t __user *)args->priv_data, &priv_offset); - if (ret) - goto exit_unlock; + /* Leave room for BOs in the private data. They need to be restored + * before events, but we checkpoint them last to simplify the error + * handling. + */ + bo_priv_offset = priv_offset; + priv_offset += num_bos * sizeof(struct kfd_criu_bo_priv_data); if (num_objects) { ret = kfd_criu_checkpoint_queues(p, (uint8_t __user *)args->priv_data, &priv_offset); if (ret) - goto close_bo_fds; + goto exit_unlock; ret = kfd_criu_checkpoint_events(p, (uint8_t __user *)args->priv_data, &priv_offset); if (ret) - goto close_bo_fds; + goto exit_unlock; ret = kfd_criu_checkpoint_svm(p, (uint8_t __user *)args->priv_data, &priv_offset); if (ret) - goto close_bo_fds; + goto exit_unlock; } -close_bo_fds: - if (ret) { - /* If IOCTL returns err, user assumes all FDs opened in criu_dump_bos are closed */ - uint32_t i; - struct kfd_criu_bo_bucket *bo_buckets = (struct kfd_criu_bo_bucket *) args->bos; - - for (i = 0; i < num_bos; i++) { - if (bo_buckets[i].alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) - close_fd(bo_buckets[i].dmabuf_fd); - } - } + /* This must be the last thing in this function that can fail. + * Otherwise we leak dmabuf file descriptors. + */ + ret = criu_checkpoint_bos(p, num_bos, (uint8_t __user *)args->bos, + (uint8_t __user *)args->priv_data, &bo_priv_offset); exit_unlock: mutex_unlock(&p->mutex); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c index cd5f8b219bf9..af01ba061e1b 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c @@ -50,16 +50,6 @@ static inline unsigned int get_and_inc_gpu_processor_id( return current_id; } -/* Static table to describe GPU Cache information */ -struct kfd_gpu_cache_info { - uint32_t cache_size; - uint32_t cache_level; - uint32_t flags; - /* Indicates how many Compute Units share this cache - * within a SA. Value = 1 indicates the cache is not shared - */ - uint32_t num_cu_shared; -}; static struct kfd_gpu_cache_info kaveri_cache_info[] = { { @@ -795,6 +785,150 @@ static struct kfd_gpu_cache_info yellow_carp_cache_info[] = { }, }; +static struct kfd_gpu_cache_info gfx1037_cache_info[] = { + { + /* TCP L1 Cache per CU */ + .cache_size = 16, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 1, + }, + { + /* Scalar L1 Instruction Cache per SQC */ + .cache_size = 32, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_INST_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 2, + }, + { + /* Scalar L1 Data Cache per SQC */ + .cache_size = 16, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 2, + }, + { + /* GL1 Data Cache per SA */ + .cache_size = 128, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 2, + }, + { + /* L2 Data Cache per GPU (Total Tex Cache) */ + .cache_size = 256, + .cache_level = 2, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 2, + }, +}; + +static struct kfd_gpu_cache_info gc_10_3_6_cache_info[] = { + { + /* TCP L1 Cache per CU */ + .cache_size = 16, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 1, + }, + { + /* Scalar L1 Instruction Cache per SQC */ + .cache_size = 32, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_INST_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 2, + }, + { + /* Scalar L1 Data Cache per SQC */ + .cache_size = 16, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 2, + }, + { + /* GL1 Data Cache per SA */ + .cache_size = 128, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 2, + }, + { + /* L2 Data Cache per GPU (Total Tex Cache) */ + .cache_size = 256, + .cache_level = 2, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 2, + }, +}; + +static struct kfd_gpu_cache_info dummy_cache_info[] = { + { + /* TCP L1 Cache per CU */ + .cache_size = 16, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 1, + }, + { + /* Scalar L1 Instruction Cache per SQC */ + .cache_size = 32, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_INST_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 2, + }, + { + /* Scalar L1 Data Cache per SQC */ + .cache_size = 16, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 2, + }, + { + /* GL1 Data Cache per SA */ + .cache_size = 128, + .cache_level = 1, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 6, + }, + { + /* L2 Data Cache per GPU (Total Tex Cache) */ + .cache_size = 2048, + .cache_level = 2, + .flags = (CRAT_CACHE_FLAGS_ENABLED | + CRAT_CACHE_FLAGS_DATA_CACHE | + CRAT_CACHE_FLAGS_SIMD_CACHE), + .num_cu_shared = 6, + }, +}; + static void kfd_populated_cu_info_cpu(struct kfd_topology_device *dev, struct crat_subtype_computeunit *cu) { @@ -975,8 +1109,12 @@ static int kfd_parse_subtype_cache(struct crat_subtype_cache *cache, props->cachelines_per_tag = cache->lines_per_tag; props->cache_assoc = cache->associativity; props->cache_latency = cache->cache_latency; + memcpy(props->sibling_map, cache->sibling_map, - sizeof(props->sibling_map)); + CRAT_SIBLINGMAP_SIZE); + + /* set the sibling_map_size as 32 for CRAT from ACPI */ + props->sibling_map_size = CRAT_SIBLINGMAP_SIZE; if (cache->flags & CRAT_CACHE_FLAGS_DATA_CACHE) props->cache_type |= HSA_CACHE_TYPE_DATA; @@ -987,7 +1125,6 @@ static int kfd_parse_subtype_cache(struct crat_subtype_cache *cache, if (cache->flags & CRAT_CACHE_FLAGS_SIMD_CACHE) props->cache_type |= HSA_CACHE_TYPE_HSACU; - dev->cache_count++; dev->node_props.caches_count++; list_add_tail(&props->list, &dev->cache_props); @@ -1195,125 +1332,6 @@ err: return ret; } -/* Helper function. See kfd_fill_gpu_cache_info for parameter description */ -static int fill_in_l1_pcache(struct crat_subtype_cache *pcache, - struct kfd_gpu_cache_info *pcache_info, - struct kfd_cu_info *cu_info, - int mem_available, - int cu_bitmask, - int cache_type, unsigned int cu_processor_id, - int cu_block) -{ - unsigned int cu_sibling_map_mask; - int first_active_cu; - - /* First check if enough memory is available */ - if (sizeof(struct crat_subtype_cache) > mem_available) - return -ENOMEM; - - cu_sibling_map_mask = cu_bitmask; - cu_sibling_map_mask >>= cu_block; - cu_sibling_map_mask &= - ((1 << pcache_info[cache_type].num_cu_shared) - 1); - first_active_cu = ffs(cu_sibling_map_mask); - - /* CU could be inactive. In case of shared cache find the first active - * CU. and incase of non-shared cache check if the CU is inactive. If - * inactive active skip it - */ - if (first_active_cu) { - memset(pcache, 0, sizeof(struct crat_subtype_cache)); - pcache->type = CRAT_SUBTYPE_CACHE_AFFINITY; - pcache->length = sizeof(struct crat_subtype_cache); - pcache->flags = pcache_info[cache_type].flags; - pcache->processor_id_low = cu_processor_id - + (first_active_cu - 1); - pcache->cache_level = pcache_info[cache_type].cache_level; - pcache->cache_size = pcache_info[cache_type].cache_size; - - /* Sibling map is w.r.t processor_id_low, so shift out - * inactive CU - */ - cu_sibling_map_mask = - cu_sibling_map_mask >> (first_active_cu - 1); - - pcache->sibling_map[0] = (uint8_t)(cu_sibling_map_mask & 0xFF); - pcache->sibling_map[1] = - (uint8_t)((cu_sibling_map_mask >> 8) & 0xFF); - pcache->sibling_map[2] = - (uint8_t)((cu_sibling_map_mask >> 16) & 0xFF); - pcache->sibling_map[3] = - (uint8_t)((cu_sibling_map_mask >> 24) & 0xFF); - return 0; - } - return 1; -} - -/* Helper function. See kfd_fill_gpu_cache_info for parameter description */ -static int fill_in_l2_l3_pcache(struct crat_subtype_cache *pcache, - struct kfd_gpu_cache_info *pcache_info, - struct kfd_cu_info *cu_info, - int mem_available, - int cache_type, unsigned int cu_processor_id) -{ - unsigned int cu_sibling_map_mask; - int first_active_cu; - int i, j, k; - - /* First check if enough memory is available */ - if (sizeof(struct crat_subtype_cache) > mem_available) - return -ENOMEM; - - cu_sibling_map_mask = cu_info->cu_bitmap[0][0]; - cu_sibling_map_mask &= - ((1 << pcache_info[cache_type].num_cu_shared) - 1); - first_active_cu = ffs(cu_sibling_map_mask); - - /* CU could be inactive. In case of shared cache find the first active - * CU. and incase of non-shared cache check if the CU is inactive. If - * inactive active skip it - */ - if (first_active_cu) { - memset(pcache, 0, sizeof(struct crat_subtype_cache)); - pcache->type = CRAT_SUBTYPE_CACHE_AFFINITY; - pcache->length = sizeof(struct crat_subtype_cache); - pcache->flags = pcache_info[cache_type].flags; - pcache->processor_id_low = cu_processor_id - + (first_active_cu - 1); - pcache->cache_level = pcache_info[cache_type].cache_level; - pcache->cache_size = pcache_info[cache_type].cache_size; - - /* Sibling map is w.r.t processor_id_low, so shift out - * inactive CU - */ - cu_sibling_map_mask = - cu_sibling_map_mask >> (first_active_cu - 1); - k = 0; - for (i = 0; i < cu_info->num_shader_engines; i++) { - for (j = 0; j < cu_info->num_shader_arrays_per_engine; - j++) { - pcache->sibling_map[k] = - (uint8_t)(cu_sibling_map_mask & 0xFF); - pcache->sibling_map[k+1] = - (uint8_t)((cu_sibling_map_mask >> 8) & 0xFF); - pcache->sibling_map[k+2] = - (uint8_t)((cu_sibling_map_mask >> 16) & 0xFF); - pcache->sibling_map[k+3] = - (uint8_t)((cu_sibling_map_mask >> 24) & 0xFF); - k += 4; - cu_sibling_map_mask = - cu_info->cu_bitmap[i % 4][j + i / 4]; - cu_sibling_map_mask &= ( - (1 << pcache_info[cache_type].num_cu_shared) - - 1); - } - } - return 0; - } - return 1; -} - -#define KFD_MAX_CACHE_TYPES 6 static int kfd_fill_gpu_cache_info_from_gfx_config(struct kfd_dev *kdev, struct kfd_gpu_cache_info *pcache_info) @@ -1387,222 +1405,133 @@ static int kfd_fill_gpu_cache_info_from_gfx_config(struct kfd_dev *kdev, return i; } -/* kfd_fill_gpu_cache_info - Fill GPU cache info using kfd_gpu_cache_info - * tables - * - * @kdev - [IN] GPU device - * @gpu_processor_id - [IN] GPU processor ID to which these caches - * associate - * @available_size - [IN] Amount of memory available in pcache - * @cu_info - [IN] Compute Unit info obtained from KGD - * @pcache - [OUT] memory into which cache data is to be filled in. - * @size_filled - [OUT] amount of data used up in pcache. - * @num_of_entries - [OUT] number of caches added - */ -static int kfd_fill_gpu_cache_info(struct kfd_dev *kdev, - int gpu_processor_id, - int available_size, - struct kfd_cu_info *cu_info, - struct crat_subtype_cache *pcache, - int *size_filled, - int *num_of_entries) +int kfd_get_gpu_cache_info(struct kfd_dev *kdev, struct kfd_gpu_cache_info **pcache_info) { - struct kfd_gpu_cache_info *pcache_info; - struct kfd_gpu_cache_info cache_info[KFD_MAX_CACHE_TYPES]; int num_of_cache_types = 0; - int i, j, k; - int ct = 0; - int mem_available = available_size; - unsigned int cu_processor_id; - int ret; - unsigned int num_cu_shared; switch (kdev->adev->asic_type) { case CHIP_KAVERI: - pcache_info = kaveri_cache_info; + *pcache_info = kaveri_cache_info; num_of_cache_types = ARRAY_SIZE(kaveri_cache_info); break; case CHIP_HAWAII: - pcache_info = hawaii_cache_info; + *pcache_info = hawaii_cache_info; num_of_cache_types = ARRAY_SIZE(hawaii_cache_info); break; case CHIP_CARRIZO: - pcache_info = carrizo_cache_info; + *pcache_info = carrizo_cache_info; num_of_cache_types = ARRAY_SIZE(carrizo_cache_info); break; case CHIP_TONGA: - pcache_info = tonga_cache_info; + *pcache_info = tonga_cache_info; num_of_cache_types = ARRAY_SIZE(tonga_cache_info); break; case CHIP_FIJI: - pcache_info = fiji_cache_info; + *pcache_info = fiji_cache_info; num_of_cache_types = ARRAY_SIZE(fiji_cache_info); break; case CHIP_POLARIS10: - pcache_info = polaris10_cache_info; + *pcache_info = polaris10_cache_info; num_of_cache_types = ARRAY_SIZE(polaris10_cache_info); break; case CHIP_POLARIS11: - pcache_info = polaris11_cache_info; + *pcache_info = polaris11_cache_info; num_of_cache_types = ARRAY_SIZE(polaris11_cache_info); break; case CHIP_POLARIS12: - pcache_info = polaris12_cache_info; + *pcache_info = polaris12_cache_info; num_of_cache_types = ARRAY_SIZE(polaris12_cache_info); break; case CHIP_VEGAM: - pcache_info = vegam_cache_info; + *pcache_info = vegam_cache_info; num_of_cache_types = ARRAY_SIZE(vegam_cache_info); break; default: switch (KFD_GC_VERSION(kdev)) { case IP_VERSION(9, 0, 1): - pcache_info = vega10_cache_info; + *pcache_info = vega10_cache_info; num_of_cache_types = ARRAY_SIZE(vega10_cache_info); break; case IP_VERSION(9, 2, 1): - pcache_info = vega12_cache_info; + *pcache_info = vega12_cache_info; num_of_cache_types = ARRAY_SIZE(vega12_cache_info); break; case IP_VERSION(9, 4, 0): case IP_VERSION(9, 4, 1): - pcache_info = vega20_cache_info; + *pcache_info = vega20_cache_info; num_of_cache_types = ARRAY_SIZE(vega20_cache_info); break; case IP_VERSION(9, 4, 2): - pcache_info = aldebaran_cache_info; + *pcache_info = aldebaran_cache_info; num_of_cache_types = ARRAY_SIZE(aldebaran_cache_info); break; case IP_VERSION(9, 1, 0): case IP_VERSION(9, 2, 2): - pcache_info = raven_cache_info; + *pcache_info = raven_cache_info; num_of_cache_types = ARRAY_SIZE(raven_cache_info); break; case IP_VERSION(9, 3, 0): - pcache_info = renoir_cache_info; + *pcache_info = renoir_cache_info; num_of_cache_types = ARRAY_SIZE(renoir_cache_info); break; case IP_VERSION(10, 1, 10): case IP_VERSION(10, 1, 2): case IP_VERSION(10, 1, 3): case IP_VERSION(10, 1, 4): - pcache_info = navi10_cache_info; + *pcache_info = navi10_cache_info; num_of_cache_types = ARRAY_SIZE(navi10_cache_info); break; case IP_VERSION(10, 1, 1): - pcache_info = navi14_cache_info; + *pcache_info = navi14_cache_info; num_of_cache_types = ARRAY_SIZE(navi14_cache_info); break; case IP_VERSION(10, 3, 0): - pcache_info = sienna_cichlid_cache_info; + *pcache_info = sienna_cichlid_cache_info; num_of_cache_types = ARRAY_SIZE(sienna_cichlid_cache_info); break; case IP_VERSION(10, 3, 2): - pcache_info = navy_flounder_cache_info; + *pcache_info = navy_flounder_cache_info; num_of_cache_types = ARRAY_SIZE(navy_flounder_cache_info); break; case IP_VERSION(10, 3, 4): - pcache_info = dimgrey_cavefish_cache_info; + *pcache_info = dimgrey_cavefish_cache_info; num_of_cache_types = ARRAY_SIZE(dimgrey_cavefish_cache_info); break; case IP_VERSION(10, 3, 1): - pcache_info = vangogh_cache_info; + *pcache_info = vangogh_cache_info; num_of_cache_types = ARRAY_SIZE(vangogh_cache_info); break; case IP_VERSION(10, 3, 5): - pcache_info = beige_goby_cache_info; + *pcache_info = beige_goby_cache_info; num_of_cache_types = ARRAY_SIZE(beige_goby_cache_info); break; case IP_VERSION(10, 3, 3): - case IP_VERSION(10, 3, 6): /* TODO: Double check these on production silicon */ - case IP_VERSION(10, 3, 7): /* TODO: Double check these on production silicon */ - pcache_info = yellow_carp_cache_info; + *pcache_info = yellow_carp_cache_info; num_of_cache_types = ARRAY_SIZE(yellow_carp_cache_info); break; + case IP_VERSION(10, 3, 6): + *pcache_info = gc_10_3_6_cache_info; + num_of_cache_types = ARRAY_SIZE(gc_10_3_6_cache_info); + break; + case IP_VERSION(10, 3, 7): + *pcache_info = gfx1037_cache_info; + num_of_cache_types = ARRAY_SIZE(gfx1037_cache_info); + break; case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 1): case IP_VERSION(11, 0, 2): case IP_VERSION(11, 0, 3): - pcache_info = cache_info; num_of_cache_types = - kfd_fill_gpu_cache_info_from_gfx_config(kdev, pcache_info); + kfd_fill_gpu_cache_info_from_gfx_config(kdev, *pcache_info); break; default: - return -EINVAL; - } - } - - *size_filled = 0; - *num_of_entries = 0; - - /* For each type of cache listed in the kfd_gpu_cache_info table, - * go through all available Compute Units. - * The [i,j,k] loop will - * if kfd_gpu_cache_info.num_cu_shared = 1 - * will parse through all available CU - * If (kfd_gpu_cache_info.num_cu_shared != 1) - * then it will consider only one CU from - * the shared unit - */ - - for (ct = 0; ct < num_of_cache_types; ct++) { - cu_processor_id = gpu_processor_id; - if (pcache_info[ct].cache_level == 1) { - for (i = 0; i < cu_info->num_shader_engines; i++) { - for (j = 0; j < cu_info->num_shader_arrays_per_engine; j++) { - for (k = 0; k < cu_info->num_cu_per_sh; - k += pcache_info[ct].num_cu_shared) { - ret = fill_in_l1_pcache(pcache, - pcache_info, - cu_info, - mem_available, - cu_info->cu_bitmap[i % 4][j + i / 4], - ct, - cu_processor_id, - k); - - if (ret < 0) + *pcache_info = dummy_cache_info; + num_of_cache_types = ARRAY_SIZE(dummy_cache_info); + pr_warn("dummy cache info is used temporarily and real cache info need update later.\n"); break; - - if (!ret) { - pcache++; - (*num_of_entries)++; - mem_available -= sizeof(*pcache); - (*size_filled) += sizeof(*pcache); - } - - /* Move to next CU block */ - num_cu_shared = ((k + pcache_info[ct].num_cu_shared) <= - cu_info->num_cu_per_sh) ? - pcache_info[ct].num_cu_shared : - (cu_info->num_cu_per_sh - k); - cu_processor_id += num_cu_shared; } - } - } - } else { - ret = fill_in_l2_l3_pcache(pcache, - pcache_info, - cu_info, - mem_available, - ct, - cu_processor_id); - - if (ret < 0) - break; - - if (!ret) { - pcache++; - (*num_of_entries)++; - mem_available -= sizeof(*pcache); - (*size_filled) += sizeof(*pcache); - } - } } - - pr_debug("Added [%d] GPU cache entries\n", *num_of_entries); - - return 0; + return num_of_cache_types; } static bool kfd_ignore_crat(void) @@ -1961,8 +1890,8 @@ static void kfd_find_numa_node_in_srat(struct kfd_dev *kdev) struct acpi_table_header *table_header = NULL; struct acpi_subtable_header *sub_header = NULL; unsigned long table_end, subtable_len; - u32 pci_id = pci_domain_nr(kdev->pdev->bus) << 16 | - pci_dev_id(kdev->pdev); + u32 pci_id = pci_domain_nr(kdev->adev->pdev->bus) << 16 | + pci_dev_id(kdev->adev->pdev); u32 bdf; acpi_status status; struct acpi_srat_cpu_affinity *cpu; @@ -2037,7 +1966,7 @@ static void kfd_find_numa_node_in_srat(struct kfd_dev *kdev) numa_node = 0; if (numa_node != NUMA_NO_NODE) - set_dev_node(&kdev->pdev->dev, numa_node); + set_dev_node(&kdev->adev->pdev->dev, numa_node); } #endif @@ -2098,14 +2027,14 @@ static int kfd_fill_gpu_direct_io_link_to_cpu(int *avail_size, sub_type_hdr->proximity_domain_from = proximity_domain; #ifdef CONFIG_ACPI_NUMA - if (kdev->pdev->dev.numa_node == NUMA_NO_NODE) + if (kdev->adev->pdev->dev.numa_node == NUMA_NO_NODE) kfd_find_numa_node_in_srat(kdev); #endif #ifdef CONFIG_NUMA - if (kdev->pdev->dev.numa_node == NUMA_NO_NODE) + if (kdev->adev->pdev->dev.numa_node == NUMA_NO_NODE) sub_type_hdr->proximity_domain_to = 0; else - sub_type_hdr->proximity_domain_to = kdev->pdev->dev.numa_node; + sub_type_hdr->proximity_domain_to = kdev->adev->pdev->dev.numa_node; #else sub_type_hdr->proximity_domain_to = 0; #endif @@ -2161,8 +2090,6 @@ static int kfd_create_vcrat_image_gpu(void *pcrat_image, struct kfd_cu_info cu_info; int avail_size = *size; uint32_t total_num_of_cu; - int num_of_cache_entries = 0; - int cache_mem_filled = 0; uint32_t nid = 0; int ret = 0; @@ -2263,31 +2190,12 @@ static int kfd_create_vcrat_image_gpu(void *pcrat_image, crat_table->length += sizeof(struct crat_subtype_memory); crat_table->total_entries++; - /* TODO: Fill in cache information. This information is NOT readily - * available in KGD - */ - sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr + - sub_type_hdr->length); - ret = kfd_fill_gpu_cache_info(kdev, cu->processor_id_low, - avail_size, - &cu_info, - (struct crat_subtype_cache *)sub_type_hdr, - &cache_mem_filled, - &num_of_cache_entries); - - if (ret < 0) - return ret; - - crat_table->length += cache_mem_filled; - crat_table->total_entries += num_of_cache_entries; - avail_size -= cache_mem_filled; - /* Fill in Subtype: IO_LINKS * Only direct links are added here which is Link from GPU to * its NUMA node. Indirect links are added by userspace. */ sub_type_hdr = (typeof(sub_type_hdr))((char *)sub_type_hdr + - cache_mem_filled); + sub_type_hdr->length); ret = kfd_fill_gpu_direct_io_link_to_cpu(&avail_size, kdev, (struct crat_subtype_iolink *)sub_type_hdr, proximity_domain); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_crat.h b/drivers/gpu/drm/amd/amdkfd/kfd_crat.h index 482ba84a728d..8d1e8ba58dee 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.h @@ -29,11 +29,10 @@ #pragma pack(1) /* - * 4CC signature values for the CRAT and CDIT ACPI tables + * 4CC signature value for the CRAT ACPI table */ #define CRAT_SIGNATURE "CRAT" -#define CDIT_SIGNATURE "CDIT" /* * Component Resource Association Table (CRAT) @@ -292,31 +291,22 @@ struct crat_subtype_generic { uint32_t flags; }; -/* - * Component Locality Distance Information Table (CDIT) - */ -#define CDIT_OEMID_LENGTH 6 -#define CDIT_OEMTABLEID_LENGTH 8 - -struct cdit_header { - uint32_t signature; - uint32_t length; - uint8_t revision; - uint8_t checksum; - uint8_t oem_id[CDIT_OEMID_LENGTH]; - uint8_t oem_table_id[CDIT_OEMTABLEID_LENGTH]; - uint32_t oem_revision; - uint32_t creator_id; - uint32_t creator_revision; - uint32_t total_entries; - uint16_t num_domains; - uint8_t entry[1]; -}; - #pragma pack() struct kfd_dev; +/* Static table to describe GPU Cache information */ +struct kfd_gpu_cache_info { + uint32_t cache_size; + uint32_t cache_level; + uint32_t flags; + /* Indicates how many Compute Units share this cache + * within a SA. Value = 1 indicates the cache is not shared + */ + uint32_t num_cu_shared; +}; +int kfd_get_gpu_cache_info(struct kfd_dev *kdev, struct kfd_gpu_cache_info **pcache_info); + int kfd_create_crat_image_acpi(void **crat_image, size_t *size); void kfd_destroy_crat_image(void *crat_image); int kfd_parse_crat_table(void *crat_image, struct list_head *device_list, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c index 65a1d4f9004b..c8dd948966c2 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c @@ -227,7 +227,6 @@ struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf) { struct kfd_dev *kfd = NULL; const struct kfd2kgd_calls *f2g = NULL; - struct pci_dev *pdev = adev->pdev; uint32_t gfx_target_version = 0; switch (adev->asic_type) { @@ -429,7 +428,6 @@ struct kfd_dev *kgd2kfd_probe(struct amdgpu_device *adev, bool vf) kfd->adev = adev; kfd_device_info_init(kfd, vf, gfx_target_version); - kfd->pdev = pdev; kfd->init_complete = false; kfd->kfd2kgd = f2g; atomic_set(&kfd->compute_profile, 0); @@ -497,7 +495,10 @@ static int kfd_gws_init(struct kfd_dev *kfd) (KFD_GC_VERSION(kfd) == IP_VERSION(9, 4, 1) && kfd->mec2_fw_version >= 0x30) || (KFD_GC_VERSION(kfd) == IP_VERSION(9, 4, 2) - && kfd->mec2_fw_version >= 0x28)))) + && kfd->mec2_fw_version >= 0x28) || + (KFD_GC_VERSION(kfd) >= IP_VERSION(10, 3, 0) + && KFD_GC_VERSION(kfd) < IP_VERSION(11, 0, 0) + && kfd->mec2_fw_version >= 0x6b)))) ret = amdgpu_amdkfd_alloc_gws(kfd->adev, kfd->adev->gds.gws_size, &kfd->gws); @@ -511,12 +512,10 @@ static void kfd_smi_init(struct kfd_dev *dev) } bool kgd2kfd_device_init(struct kfd_dev *kfd, - struct drm_device *ddev, const struct kgd2kfd_shared_resources *gpu_resources) { unsigned int size, map_process_packet_size; - kfd->ddev = ddev; kfd->mec_fw_version = amdgpu_amdkfd_get_fw_version(kfd->adev, KGD_ENGINE_MEC1); kfd->mec2_fw_version = amdgpu_amdkfd_get_fw_version(kfd->adev, @@ -541,7 +540,7 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd, kfd->mec_fw_version < kfd->device_info.no_atomic_fw_version)) { dev_info(kfd_device, "skipped device %x:%x, PCI rejects atomics %d<%d\n", - kfd->pdev->vendor, kfd->pdev->device, + kfd->adev->pdev->vendor, kfd->adev->pdev->device, kfd->mec_fw_version, kfd->device_info.no_atomic_fw_version); return false; @@ -650,8 +649,8 @@ bool kgd2kfd_device_init(struct kfd_dev *kfd, kfd_smi_init(kfd); kfd->init_complete = true; - dev_info(kfd_device, "added device %x:%x\n", kfd->pdev->vendor, - kfd->pdev->device); + dev_info(kfd_device, "added device %x:%x\n", kfd->adev->pdev->vendor, + kfd->adev->pdev->device); pr_debug("Starting kfd with the following scheduling policy %d\n", kfd->dqm->sched_policy); @@ -676,7 +675,7 @@ alloc_gtt_mem_failure: amdgpu_amdkfd_free_gws(kfd->adev, kfd->gws); dev_err(kfd_device, "device %x:%x NOT added due to errors\n", - kfd->pdev->vendor, kfd->pdev->device); + kfd->adev->pdev->vendor, kfd->adev->pdev->device); out: return kfd->init_complete; } @@ -789,7 +788,7 @@ int kgd2kfd_resume_iommu(struct kfd_dev *kfd) if (err) dev_err(kfd_device, "Failed to resume IOMMU for device %x:%x\n", - kfd->pdev->vendor, kfd->pdev->device); + kfd->adev->pdev->vendor, kfd->adev->pdev->device); return err; } @@ -801,7 +800,7 @@ static int kfd_resume(struct kfd_dev *kfd) if (err) dev_err(kfd_device, "Error starting queue manager for device %x:%x\n", - kfd->pdev->vendor, kfd->pdev->device); + kfd->adev->pdev->vendor, kfd->adev->pdev->device); return err; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_events.c index 83e3ce9f6049..729d26d648af 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c @@ -506,6 +506,7 @@ int kfd_criu_restore_event(struct file *devkfd, ret = create_other_event(p, ev, &ev_priv->event_id); break; } + mutex_unlock(&p->event_mutex); exit: if (ret) @@ -513,8 +514,6 @@ exit: kfree(ev_priv); - mutex_unlock(&p->event_mutex); - return ret; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_iommu.c b/drivers/gpu/drm/amd/amdkfd/kfd_iommu.c index fbd0afe4da42..ec1bf611624e 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_iommu.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_iommu.c @@ -49,7 +49,7 @@ int kfd_iommu_check_device(struct kfd_dev *kfd) return -ENODEV; iommu_info.flags = 0; - err = amd_iommu_device_info(kfd->pdev, &iommu_info); + err = amd_iommu_device_info(kfd->adev->pdev, &iommu_info); if (err) return err; @@ -71,7 +71,7 @@ int kfd_iommu_device_init(struct kfd_dev *kfd) return 0; iommu_info.flags = 0; - err = amd_iommu_device_info(kfd->pdev, &iommu_info); + err = amd_iommu_device_info(kfd->adev->pdev, &iommu_info); if (err < 0) { dev_err(kfd_device, "error getting iommu info. is the iommu enabled?\n"); @@ -121,7 +121,7 @@ int kfd_iommu_bind_process_to_device(struct kfd_process_device *pdd) return -EINVAL; } - err = amd_iommu_bind_pasid(dev->pdev, p->pasid, p->lead_thread); + err = amd_iommu_bind_pasid(dev->adev->pdev, p->pasid, p->lead_thread); if (!err) pdd->bound = PDD_BOUND; @@ -139,7 +139,8 @@ void kfd_iommu_unbind_process(struct kfd_process *p) for (i = 0; i < p->n_pdds; i++) if (p->pdds[i]->bound == PDD_BOUND) - amd_iommu_unbind_pasid(p->pdds[i]->dev->pdev, p->pasid); + amd_iommu_unbind_pasid(p->pdds[i]->dev->adev->pdev, + p->pasid); } /* Callback for process shutdown invoked by the IOMMU driver */ @@ -222,7 +223,7 @@ static int kfd_bind_processes_to_device(struct kfd_dev *kfd) continue; } - err = amd_iommu_bind_pasid(kfd->pdev, p->pasid, + err = amd_iommu_bind_pasid(kfd->adev->pdev, p->pasid, p->lead_thread); if (err < 0) { pr_err("Unexpected pasid 0x%x binding failure\n", @@ -282,9 +283,9 @@ void kfd_iommu_suspend(struct kfd_dev *kfd) kfd_unbind_processes_from_device(kfd); - amd_iommu_set_invalidate_ctx_cb(kfd->pdev, NULL); - amd_iommu_set_invalid_ppr_cb(kfd->pdev, NULL); - amd_iommu_free_device(kfd->pdev); + amd_iommu_set_invalidate_ctx_cb(kfd->adev->pdev, NULL); + amd_iommu_set_invalid_ppr_cb(kfd->adev->pdev, NULL); + amd_iommu_free_device(kfd->adev->pdev); } /** kfd_iommu_resume - Restore IOMMU after resume @@ -302,20 +303,20 @@ int kfd_iommu_resume(struct kfd_dev *kfd) pasid_limit = kfd_get_pasid_limit(); - err = amd_iommu_init_device(kfd->pdev, pasid_limit); + err = amd_iommu_init_device(kfd->adev->pdev, pasid_limit); if (err) return -ENXIO; - amd_iommu_set_invalidate_ctx_cb(kfd->pdev, + amd_iommu_set_invalidate_ctx_cb(kfd->adev->pdev, iommu_pasid_shutdown_callback); - amd_iommu_set_invalid_ppr_cb(kfd->pdev, + amd_iommu_set_invalid_ppr_cb(kfd->adev->pdev, iommu_invalid_ppr_cb); err = kfd_bind_processes_to_device(kfd); if (err) { - amd_iommu_set_invalidate_ctx_cb(kfd->pdev, NULL); - amd_iommu_set_invalid_ppr_cb(kfd->pdev, NULL); - amd_iommu_free_device(kfd->pdev); + amd_iommu_set_invalidate_ctx_cb(kfd->adev->pdev, NULL); + amd_iommu_set_invalid_ppr_cb(kfd->adev->pdev, NULL); + amd_iommu_free_device(kfd->adev->pdev); return err; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c index 79069c485386..10048ce16aea 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c @@ -28,7 +28,6 @@ #include "amdgpu_sync.h" #include "amdgpu_object.h" #include "amdgpu_vm.h" -#include "amdgpu_mn.h" #include "amdgpu_res_cursor.h" #include "kfd_priv.h" #include "kfd_svm.h" @@ -524,8 +523,8 @@ svm_migrate_ram_to_vram(struct svm_range *prange, uint32_t best_loc, for (addr = start; addr < end;) { unsigned long next; - vma = find_vma(mm, addr); - if (!vma || addr < vma->vm_start) + vma = vma_lookup(mm, addr); + if (!vma) break; next = min(vma->vm_end, end); @@ -793,8 +792,8 @@ int svm_migrate_vram_to_ram(struct svm_range *prange, struct mm_struct *mm, for (addr = start; addr < end;) { unsigned long next; - vma = find_vma(mm, addr); - if (!vma || addr < vma->vm_start) { + vma = vma_lookup(mm, addr); + if (!vma) { pr_debug("failed to find vma for prange %p\n", prange); r = -EFAULT; break; @@ -968,12 +967,10 @@ out_unlock_prange: out_unlock_svms: mutex_unlock(&p->svms.lock); out_unref_process: + pr_debug("CPU fault svms 0x%p address 0x%lx done\n", &p->svms, addr); kfd_unref_process(p); out_mmput: mmput(mm); - - pr_debug("CPU fault svms 0x%p address 0x%lx done\n", &p->svms, addr); - return r ? VM_FAULT_SIGBUS : 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_diq.h b/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_diq.h deleted file mode 100644 index f9cd28690151..000000000000 --- a/drivers/gpu/drm/amd/amdkfd/kfd_pm4_headers_diq.h +++ /dev/null @@ -1,291 +0,0 @@ -/* SPDX-License-Identifier: GPL-2.0 OR MIT */ -/* - * Copyright 2014-2022 Advanced Micro Devices, Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef KFD_PM4_HEADERS_DIQ_H_ -#define KFD_PM4_HEADERS_DIQ_H_ - -/*--------------------_INDIRECT_BUFFER-------------------- */ - -#ifndef _PM4__INDIRECT_BUFFER_DEFINED -#define _PM4__INDIRECT_BUFFER_DEFINED -enum _INDIRECT_BUFFER_cache_policy_enum { - cache_policy___indirect_buffer__lru = 0, - cache_policy___indirect_buffer__stream = 1, - cache_policy___indirect_buffer__bypass = 2 -}; - -enum { - IT_INDIRECT_BUFFER_PASID = 0x5C -}; - -struct pm4__indirect_buffer_pasid { - union { - union PM4_MES_TYPE_3_HEADER header; /* header */ - unsigned int ordinal1; - }; - - union { - struct { - unsigned int reserved1:2; - unsigned int ib_base_lo:30; - } bitfields2; - unsigned int ordinal2; - }; - - union { - struct { - unsigned int ib_base_hi:16; - unsigned int reserved2:16; - } bitfields3; - unsigned int ordinal3; - }; - - union { - unsigned int control; - unsigned int ordinal4; - }; - - union { - struct { - unsigned int pasid:10; - unsigned int reserved4:22; - } bitfields5; - unsigned int ordinal5; - }; - -}; - -#endif - -/*--------------------_RELEASE_MEM-------------------- */ - -#ifndef _PM4__RELEASE_MEM_DEFINED -#define _PM4__RELEASE_MEM_DEFINED -enum _RELEASE_MEM_event_index_enum { - event_index___release_mem__end_of_pipe = 5, - event_index___release_mem__shader_done = 6 -}; - -enum _RELEASE_MEM_cache_policy_enum { - cache_policy___release_mem__lru = 0, - cache_policy___release_mem__stream = 1, - cache_policy___release_mem__bypass = 2 -}; - -enum _RELEASE_MEM_dst_sel_enum { - dst_sel___release_mem__memory_controller = 0, - dst_sel___release_mem__tc_l2 = 1, - dst_sel___release_mem__queue_write_pointer_register = 2, - dst_sel___release_mem__queue_write_pointer_poll_mask_bit = 3 -}; - -enum _RELEASE_MEM_int_sel_enum { - int_sel___release_mem__none = 0, - int_sel___release_mem__send_interrupt_only = 1, - int_sel___release_mem__send_interrupt_after_write_confirm = 2, - int_sel___release_mem__send_data_after_write_confirm = 3 -}; - -enum _RELEASE_MEM_data_sel_enum { - data_sel___release_mem__none = 0, - data_sel___release_mem__send_32_bit_low = 1, - data_sel___release_mem__send_64_bit_data = 2, - data_sel___release_mem__send_gpu_clock_counter = 3, - data_sel___release_mem__send_cp_perfcounter_hi_lo = 4, - data_sel___release_mem__store_gds_data_to_memory = 5 -}; - -struct pm4__release_mem { - union { - union PM4_MES_TYPE_3_HEADER header; /*header */ - unsigned int ordinal1; - }; - - union { - struct { - unsigned int event_type:6; - unsigned int reserved1:2; - enum _RELEASE_MEM_event_index_enum event_index:4; - unsigned int tcl1_vol_action_ena:1; - unsigned int tc_vol_action_ena:1; - unsigned int reserved2:1; - unsigned int tc_wb_action_ena:1; - unsigned int tcl1_action_ena:1; - unsigned int tc_action_ena:1; - unsigned int reserved3:6; - unsigned int atc:1; - enum _RELEASE_MEM_cache_policy_enum cache_policy:2; - unsigned int reserved4:5; - } bitfields2; - unsigned int ordinal2; - }; - - union { - struct { - unsigned int reserved5:16; - enum _RELEASE_MEM_dst_sel_enum dst_sel:2; - unsigned int reserved6:6; - enum _RELEASE_MEM_int_sel_enum int_sel:3; - unsigned int reserved7:2; - enum _RELEASE_MEM_data_sel_enum data_sel:3; - } bitfields3; - unsigned int ordinal3; - }; - - union { - struct { - unsigned int reserved8:2; - unsigned int address_lo_32b:30; - } bitfields4; - struct { - unsigned int reserved9:3; - unsigned int address_lo_64b:29; - } bitfields5; - unsigned int ordinal4; - }; - - unsigned int address_hi; - - unsigned int data_lo; - - unsigned int data_hi; - -}; -#endif - - -/*--------------------_SET_CONFIG_REG-------------------- */ - -#ifndef _PM4__SET_CONFIG_REG_DEFINED -#define _PM4__SET_CONFIG_REG_DEFINED - -struct pm4__set_config_reg { - union { - union PM4_MES_TYPE_3_HEADER header; /*header */ - unsigned int ordinal1; - }; - - union { - struct { - unsigned int reg_offset:16; - unsigned int reserved1:7; - unsigned int vmid_shift:5; - unsigned int insert_vmid:1; - unsigned int reserved2:3; - } bitfields2; - unsigned int ordinal2; - }; - - unsigned int reg_data[1]; /*1..N of these fields */ - -}; -#endif - -/*--------------------_WAIT_REG_MEM-------------------- */ - -#ifndef _PM4__WAIT_REG_MEM_DEFINED -#define _PM4__WAIT_REG_MEM_DEFINED -enum _WAIT_REG_MEM_function_enum { - function___wait_reg_mem__always_pass = 0, - function___wait_reg_mem__less_than_ref_value = 1, - function___wait_reg_mem__less_than_equal_to_the_ref_value = 2, - function___wait_reg_mem__equal_to_the_reference_value = 3, - function___wait_reg_mem__not_equal_reference_value = 4, - function___wait_reg_mem__greater_than_or_equal_reference_value = 5, - function___wait_reg_mem__greater_than_reference_value = 6, - function___wait_reg_mem__reserved = 7 -}; - -enum _WAIT_REG_MEM_mem_space_enum { - mem_space___wait_reg_mem__register_space = 0, - mem_space___wait_reg_mem__memory_space = 1 -}; - -enum _WAIT_REG_MEM_operation_enum { - operation___wait_reg_mem__wait_reg_mem = 0, - operation___wait_reg_mem__wr_wait_wr_reg = 1 -}; - -struct pm4__wait_reg_mem { - union { - union PM4_MES_TYPE_3_HEADER header; /*header */ - unsigned int ordinal1; - }; - - union { - struct { - enum _WAIT_REG_MEM_function_enum function:3; - unsigned int reserved1:1; - enum _WAIT_REG_MEM_mem_space_enum mem_space:2; - enum _WAIT_REG_MEM_operation_enum operation:2; - unsigned int reserved2:24; - } bitfields2; - unsigned int ordinal2; - }; - - union { - struct { - unsigned int reserved3:2; - unsigned int memory_poll_addr_lo:30; - } bitfields3; - struct { - unsigned int register_poll_addr:16; - unsigned int reserved4:16; - } bitfields4; - struct { - unsigned int register_write_addr:16; - unsigned int reserved5:16; - } bitfields5; - unsigned int ordinal3; - }; - - union { - struct { - unsigned int poll_address_hi:16; - unsigned int reserved6:16; - } bitfields6; - struct { - unsigned int register_write_addr:16; - unsigned int reserved7:16; - } bitfields7; - unsigned int ordinal4; - }; - - unsigned int reference; - - unsigned int mask; - - union { - struct { - unsigned int poll_interval:16; - unsigned int reserved8:16; - } bitfields8; - unsigned int ordinal7; - }; - -}; -#endif - - -#endif /* KFD_PM4_HEADERS_DIQ_H_ */ diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h index bf610e3b683b..552c3ac85a13 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h @@ -254,8 +254,6 @@ struct kfd_dev { struct amdgpu_device *adev; struct kfd_device_info device_info; - struct pci_dev *pdev; - struct drm_device *ddev; unsigned int id; /* topology stub index */ @@ -1365,7 +1363,7 @@ void kfd_dec_compute_active(struct kfd_dev *dev); static inline int kfd_devcgroup_check_permission(struct kfd_dev *kfd) { #if defined(CONFIG_CGROUP_DEVICE) || defined(CONFIG_CGROUP_BPF) - struct drm_device *ddev = kfd->ddev; + struct drm_device *ddev = adev_to_drm(kfd->adev); return devcgroup_check_permission(DEVCG_DEV_CHAR, DRM_MAJOR, ddev->render->index, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index 951b63677248..a26257171ab7 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -1050,8 +1050,8 @@ static void kfd_process_destroy_pdds(struct kfd_process *p) * for auto suspend */ if (pdd->runtime_inuse) { - pm_runtime_mark_last_busy(pdd->dev->ddev->dev); - pm_runtime_put_autosuspend(pdd->dev->ddev->dev); + pm_runtime_mark_last_busy(adev_to_drm(pdd->dev->adev)->dev); + pm_runtime_put_autosuspend(adev_to_drm(pdd->dev->adev)->dev); pdd->runtime_inuse = false; } @@ -1633,9 +1633,9 @@ struct kfd_process_device *kfd_bind_process_to_device(struct kfd_dev *dev, * pdd is destroyed. */ if (!pdd->runtime_inuse) { - err = pm_runtime_get_sync(dev->ddev->dev); + err = pm_runtime_get_sync(adev_to_drm(dev->adev)->dev); if (err < 0) { - pm_runtime_put_autosuspend(dev->ddev->dev); + pm_runtime_put_autosuspend(adev_to_drm(dev->adev)->dev); return ERR_PTR(err); } } @@ -1655,8 +1655,8 @@ struct kfd_process_device *kfd_bind_process_to_device(struct kfd_dev *dev, out: /* balance runpm reference count and exit with error */ if (!pdd->runtime_inuse) { - pm_runtime_mark_last_busy(dev->ddev->dev); - pm_runtime_put_autosuspend(dev->ddev->dev); + pm_runtime_mark_last_busy(adev_to_drm(dev->adev)->dev); + pm_runtime_put_autosuspend(adev_to_drm(dev->adev)->dev); } return ERR_PTR(err); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index 64fdf63093a0..814f99888ab1 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -26,7 +26,7 @@ #include "amdgpu_sync.h" #include "amdgpu_object.h" #include "amdgpu_vm.h" -#include "amdgpu_mn.h" +#include "amdgpu_hmm.h" #include "amdgpu.h" #include "amdgpu_xgmi.h" #include "kfd_priv.h" @@ -259,7 +259,7 @@ void svm_range_free_dma_mappings(struct svm_range *prange) pr_debug("failed to find device idx %d\n", gpuidx); continue; } - dev = &pdd->dev->pdev->dev; + dev = &pdd->dev->adev->pdev->dev; svm_range_dma_unmap(dev, dma_addr, 0, prange->npages); kvfree(dma_addr); prange->dma_addr[gpuidx] = NULL; @@ -1586,8 +1586,8 @@ static int svm_range_validate_and_map(struct mm_struct *mm, unsigned long npages; bool readonly; - vma = find_vma(mm, addr); - if (!vma || addr < vma->vm_start) { + vma = vma_lookup(mm, addr); + if (!vma) { r = -EFAULT; goto unreserve_out; } @@ -1596,9 +1596,9 @@ static int svm_range_validate_and_map(struct mm_struct *mm, next = min(vma->vm_end, end); npages = (next - addr) >> PAGE_SHIFT; WRITE_ONCE(p->svms.faulting_task, current); - r = amdgpu_hmm_range_get_pages(&prange->notifier, mm, NULL, - addr, npages, &hmm_range, - readonly, true, owner); + r = amdgpu_hmm_range_get_pages(&prange->notifier, addr, npages, + readonly, owner, NULL, + &hmm_range); WRITE_ONCE(p->svms.faulting_task, NULL); if (r) { pr_debug("failed %d to get svm range pages\n", r); @@ -2542,8 +2542,8 @@ svm_range_get_range_boundaries(struct kfd_process *p, int64_t addr, struct interval_tree_node *node; unsigned long start_limit, end_limit; - vma = find_vma(p->mm, addr << PAGE_SHIFT); - if (!vma || (addr << PAGE_SHIFT) < vma->vm_start) { + vma = vma_lookup(p->mm, addr << PAGE_SHIFT); + if (!vma) { pr_debug("VMA does not exist in address [0x%llx]\n", addr); return -EFAULT; } @@ -2871,8 +2871,8 @@ retry_write_locked: /* __do_munmap removed VMA, return success as we are handling stale * retry fault. */ - vma = find_vma(mm, addr << PAGE_SHIFT); - if (!vma || (addr << PAGE_SHIFT) < vma->vm_start) { + vma = vma_lookup(mm, addr << PAGE_SHIFT); + if (!vma) { pr_debug("address 0x%llx VMA is removed\n", addr); r = 0; goto out_unlock_range; @@ -3152,9 +3152,8 @@ svm_range_is_valid(struct kfd_process *p, uint64_t start, uint64_t size) start <<= PAGE_SHIFT; end = start + (size << PAGE_SHIFT); do { - vma = find_vma(p->mm, start); - if (!vma || start < vma->vm_start || - (vma->vm_flags & device_vma)) + vma = vma_lookup(p->mm, start); + if (!vma || (vma->vm_flags & device_vma)) return -EFAULT; start = min(end, vma->vm_end); } while (start < end); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c index 3f0a4a415907..ef9c6fdfb88d 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c @@ -115,7 +115,7 @@ struct kfd_dev *kfd_device_by_pci_dev(const struct pci_dev *pdev) down_read(&topology_lock); list_for_each_entry(top_dev, &topology_device_list, list) - if (top_dev->gpu && top_dev->gpu->pdev == pdev) { + if (top_dev->gpu && top_dev->gpu->adev->pdev == pdev) { device = top_dev->gpu; break; } @@ -364,7 +364,6 @@ static ssize_t kfd_cache_show(struct kobject *kobj, struct attribute *attr, /* Making sure that the buffer is an empty string */ buffer[0] = 0; - cache = container_of(attr, struct kfd_cache_properties, attr); if (cache->gpu && kfd_devcgroup_check_permission(cache->gpu)) return -EPERM; @@ -379,12 +378,13 @@ static ssize_t kfd_cache_show(struct kobject *kobj, struct attribute *attr, sysfs_show_32bit_prop(buffer, offs, "association", cache->cache_assoc); sysfs_show_32bit_prop(buffer, offs, "latency", cache->cache_latency); sysfs_show_32bit_prop(buffer, offs, "type", cache->cache_type); + offs += snprintf(buffer+offs, PAGE_SIZE-offs, "sibling_map "); - for (i = 0; i < CRAT_SIBLINGMAP_SIZE; i++) + for (i = 0; i < cache->sibling_map_size; i++) for (j = 0; j < sizeof(cache->sibling_map[0])*8; j++) /* Check each bit */ offs += snprintf(buffer+offs, PAGE_SIZE-offs, "%d,", - (cache->sibling_map[i] >> j) & 1); + (cache->sibling_map[i] >> j) & 1); /* Replace the last "," with end of line */ buffer[offs-1] = '\n'; @@ -1169,13 +1169,12 @@ static uint32_t kfd_generate_gpu_id(struct kfd_dev *gpu) local_mem_size = gpu->local_mem_info.local_mem_size_private + gpu->local_mem_info.local_mem_size_public; - - buf[0] = gpu->pdev->devfn; - buf[1] = gpu->pdev->subsystem_vendor | - (gpu->pdev->subsystem_device << 16); - buf[2] = pci_domain_nr(gpu->pdev->bus); - buf[3] = gpu->pdev->device; - buf[4] = gpu->pdev->bus->number; + buf[0] = gpu->adev->pdev->devfn; + buf[1] = gpu->adev->pdev->subsystem_vendor | + (gpu->adev->pdev->subsystem_device << 16); + buf[2] = pci_domain_nr(gpu->adev->pdev->bus); + buf[3] = gpu->adev->pdev->device; + buf[4] = gpu->adev->pdev->bus->number; buf[5] = lower_32_bits(local_mem_size); buf[6] = upper_32_bits(local_mem_size); @@ -1198,7 +1197,6 @@ static struct kfd_topology_device *kfd_assign_gpu(struct kfd_dev *gpu) struct kfd_iolink_properties *iolink; struct kfd_iolink_properties *p2plink; - down_write(&topology_lock); list_for_each_entry(dev, &topology_device_list, list) { /* Discrete GPUs need their own topology device list * entries. Don't assign them to CPU/APU nodes. @@ -1222,7 +1220,6 @@ static struct kfd_topology_device *kfd_assign_gpu(struct kfd_dev *gpu) break; } } - up_write(&topology_lock); return out_dev; } @@ -1269,7 +1266,7 @@ static void kfd_set_iolink_no_atomics(struct kfd_topology_device *dev, if (target_gpu_dev) { uint32_t cap; - pcie_capability_read_dword(target_gpu_dev->gpu->pdev, + pcie_capability_read_dword(target_gpu_dev->gpu->adev->pdev, PCI_EXP_DEVCAP2, &cap); if (!(cap & (PCI_EXP_DEVCAP2_ATOMIC_COMP32 | @@ -1593,6 +1590,221 @@ out: return ret; } + +/* Helper function. See kfd_fill_gpu_cache_info for parameter description */ +static int fill_in_l1_pcache(struct kfd_cache_properties **props_ext, + struct kfd_gpu_cache_info *pcache_info, + struct kfd_cu_info *cu_info, + int cu_bitmask, + int cache_type, unsigned int cu_processor_id, + int cu_block) +{ + unsigned int cu_sibling_map_mask; + int first_active_cu; + struct kfd_cache_properties *pcache = NULL; + + cu_sibling_map_mask = cu_bitmask; + cu_sibling_map_mask >>= cu_block; + cu_sibling_map_mask &= ((1 << pcache_info[cache_type].num_cu_shared) - 1); + first_active_cu = ffs(cu_sibling_map_mask); + + /* CU could be inactive. In case of shared cache find the first active + * CU. and incase of non-shared cache check if the CU is inactive. If + * inactive active skip it + */ + if (first_active_cu) { + pcache = kfd_alloc_struct(pcache); + if (!pcache) + return -ENOMEM; + + memset(pcache, 0, sizeof(struct kfd_cache_properties)); + pcache->processor_id_low = cu_processor_id + (first_active_cu - 1); + pcache->cache_level = pcache_info[cache_type].cache_level; + pcache->cache_size = pcache_info[cache_type].cache_size; + + if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_DATA_CACHE) + pcache->cache_type |= HSA_CACHE_TYPE_DATA; + if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_INST_CACHE) + pcache->cache_type |= HSA_CACHE_TYPE_INSTRUCTION; + if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_CPU_CACHE) + pcache->cache_type |= HSA_CACHE_TYPE_CPU; + if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_SIMD_CACHE) + pcache->cache_type |= HSA_CACHE_TYPE_HSACU; + + /* Sibling map is w.r.t processor_id_low, so shift out + * inactive CU + */ + cu_sibling_map_mask = + cu_sibling_map_mask >> (first_active_cu - 1); + + pcache->sibling_map[0] = (uint8_t)(cu_sibling_map_mask & 0xFF); + pcache->sibling_map[1] = + (uint8_t)((cu_sibling_map_mask >> 8) & 0xFF); + pcache->sibling_map[2] = + (uint8_t)((cu_sibling_map_mask >> 16) & 0xFF); + pcache->sibling_map[3] = + (uint8_t)((cu_sibling_map_mask >> 24) & 0xFF); + + pcache->sibling_map_size = 4; + *props_ext = pcache; + + return 0; + } + return 1; +} + +/* Helper function. See kfd_fill_gpu_cache_info for parameter description */ +static int fill_in_l2_l3_pcache(struct kfd_cache_properties **props_ext, + struct kfd_gpu_cache_info *pcache_info, + struct kfd_cu_info *cu_info, + int cache_type, unsigned int cu_processor_id) +{ + unsigned int cu_sibling_map_mask; + int first_active_cu; + int i, j, k; + struct kfd_cache_properties *pcache = NULL; + + cu_sibling_map_mask = cu_info->cu_bitmap[0][0]; + cu_sibling_map_mask &= + ((1 << pcache_info[cache_type].num_cu_shared) - 1); + first_active_cu = ffs(cu_sibling_map_mask); + + /* CU could be inactive. In case of shared cache find the first active + * CU. and incase of non-shared cache check if the CU is inactive. If + * inactive active skip it + */ + if (first_active_cu) { + pcache = kfd_alloc_struct(pcache); + if (!pcache) + return -ENOMEM; + + memset(pcache, 0, sizeof(struct kfd_cache_properties)); + pcache->processor_id_low = cu_processor_id + + (first_active_cu - 1); + pcache->cache_level = pcache_info[cache_type].cache_level; + pcache->cache_size = pcache_info[cache_type].cache_size; + + if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_DATA_CACHE) + pcache->cache_type |= HSA_CACHE_TYPE_DATA; + if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_INST_CACHE) + pcache->cache_type |= HSA_CACHE_TYPE_INSTRUCTION; + if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_CPU_CACHE) + pcache->cache_type |= HSA_CACHE_TYPE_CPU; + if (pcache_info[cache_type].flags & CRAT_CACHE_FLAGS_SIMD_CACHE) + pcache->cache_type |= HSA_CACHE_TYPE_HSACU; + + /* Sibling map is w.r.t processor_id_low, so shift out + * inactive CU + */ + cu_sibling_map_mask = cu_sibling_map_mask >> (first_active_cu - 1); + k = 0; + + for (i = 0; i < cu_info->num_shader_engines; i++) { + for (j = 0; j < cu_info->num_shader_arrays_per_engine; j++) { + pcache->sibling_map[k] = (uint8_t)(cu_sibling_map_mask & 0xFF); + pcache->sibling_map[k+1] = (uint8_t)((cu_sibling_map_mask >> 8) & 0xFF); + pcache->sibling_map[k+2] = (uint8_t)((cu_sibling_map_mask >> 16) & 0xFF); + pcache->sibling_map[k+3] = (uint8_t)((cu_sibling_map_mask >> 24) & 0xFF); + k += 4; + + cu_sibling_map_mask = cu_info->cu_bitmap[i % 4][j + i / 4]; + cu_sibling_map_mask &= ((1 << pcache_info[cache_type].num_cu_shared) - 1); + } + } + pcache->sibling_map_size = k; + *props_ext = pcache; + return 0; + } + return 1; +} + +#define KFD_MAX_CACHE_TYPES 6 + +/* kfd_fill_cache_non_crat_info - Fill GPU cache info using kfd_gpu_cache_info + * tables + */ +static void kfd_fill_cache_non_crat_info(struct kfd_topology_device *dev, struct kfd_dev *kdev) +{ + struct kfd_gpu_cache_info *pcache_info = NULL; + int i, j, k; + int ct = 0; + unsigned int cu_processor_id; + int ret; + unsigned int num_cu_shared; + struct kfd_cu_info cu_info; + struct kfd_cu_info *pcu_info; + int gpu_processor_id; + struct kfd_cache_properties *props_ext; + int num_of_entries = 0; + int num_of_cache_types = 0; + struct kfd_gpu_cache_info cache_info[KFD_MAX_CACHE_TYPES]; + + amdgpu_amdkfd_get_cu_info(kdev->adev, &cu_info); + pcu_info = &cu_info; + + gpu_processor_id = dev->node_props.simd_id_base; + + pcache_info = cache_info; + num_of_cache_types = kfd_get_gpu_cache_info(kdev, &pcache_info); + if (!num_of_cache_types) { + pr_warn("no cache info found\n"); + return; + } + + /* For each type of cache listed in the kfd_gpu_cache_info table, + * go through all available Compute Units. + * The [i,j,k] loop will + * if kfd_gpu_cache_info.num_cu_shared = 1 + * will parse through all available CU + * If (kfd_gpu_cache_info.num_cu_shared != 1) + * then it will consider only one CU from + * the shared unit + */ + for (ct = 0; ct < num_of_cache_types; ct++) { + cu_processor_id = gpu_processor_id; + if (pcache_info[ct].cache_level == 1) { + for (i = 0; i < pcu_info->num_shader_engines; i++) { + for (j = 0; j < pcu_info->num_shader_arrays_per_engine; j++) { + for (k = 0; k < pcu_info->num_cu_per_sh; k += pcache_info[ct].num_cu_shared) { + + ret = fill_in_l1_pcache(&props_ext, pcache_info, pcu_info, + pcu_info->cu_bitmap[i % 4][j + i / 4], ct, + cu_processor_id, k); + + if (ret < 0) + break; + + if (!ret) { + num_of_entries++; + list_add_tail(&props_ext->list, &dev->cache_props); + } + + /* Move to next CU block */ + num_cu_shared = ((k + pcache_info[ct].num_cu_shared) <= + pcu_info->num_cu_per_sh) ? + pcache_info[ct].num_cu_shared : + (pcu_info->num_cu_per_sh - k); + cu_processor_id += num_cu_shared; + } + } + } + } else { + ret = fill_in_l2_l3_pcache(&props_ext, pcache_info, + pcu_info, ct, cu_processor_id); + + if (ret < 0) + break; + + if (!ret) { + num_of_entries++; + list_add_tail(&props_ext->list, &dev->cache_props); + } + } + } + dev->node_props.caches_count += num_of_entries; + pr_debug("Added [%d] GPU cache entries\n", num_of_entries); +} + int kfd_topology_add_device(struct kfd_dev *gpu) { uint32_t gpu_id; @@ -1617,9 +1829,9 @@ int kfd_topology_add_device(struct kfd_dev *gpu) * CRAT to create a new topology device. Once created assign the gpu to * that topology device */ + down_write(&topology_lock); dev = kfd_assign_gpu(gpu); if (!dev) { - down_write(&topology_lock); proximity_domain = ++topology_crat_proximity_domain; res = kfd_create_crat_image_virtual(&crat_image, &image_size, @@ -1631,6 +1843,7 @@ int kfd_topology_add_device(struct kfd_dev *gpu) topology_crat_proximity_domain--; return res; } + res = kfd_parse_crat_table(crat_image, &temp_topology_device_list, proximity_domain); @@ -1644,23 +1857,28 @@ int kfd_topology_add_device(struct kfd_dev *gpu) kfd_topology_update_device_list(&temp_topology_device_list, &topology_device_list); + dev = kfd_assign_gpu(gpu); + if (WARN_ON(!dev)) { + res = -ENODEV; + goto err; + } + + /* Fill the cache affinity information here for the GPUs + * using VCRAT + */ + kfd_fill_cache_non_crat_info(dev, gpu); + /* Update the SYSFS tree, since we added another topology * device */ res = kfd_topology_update_sysfs(); - up_write(&topology_lock); - if (!res) sys_props.generation_count++; else pr_err("Failed to update GPU (ID: 0x%x) to sysfs topology. res=%d\n", gpu_id, res); - dev = kfd_assign_gpu(gpu); - if (WARN_ON(!dev)) { - res = -ENODEV; - goto err; - } } + up_write(&topology_lock); dev->gpu_id = gpu_id; gpu->id = gpu_id; @@ -1688,13 +1906,13 @@ int kfd_topology_add_device(struct kfd_dev *gpu) cu_info.num_shader_arrays_per_engine; dev->node_props.gfx_target_version = gpu->device_info.gfx_target_version; - dev->node_props.vendor_id = gpu->pdev->vendor; - dev->node_props.device_id = gpu->pdev->device; + dev->node_props.vendor_id = gpu->adev->pdev->vendor; + dev->node_props.device_id = gpu->adev->pdev->device; dev->node_props.capability |= ((dev->gpu->adev->rev_id << HSA_CAP_ASIC_REVISION_SHIFT) & HSA_CAP_ASIC_REVISION_MASK); - dev->node_props.location_id = pci_dev_id(gpu->pdev); - dev->node_props.domain = pci_domain_nr(gpu->pdev->bus); + dev->node_props.location_id = pci_dev_id(gpu->adev->pdev); + dev->node_props.domain = pci_domain_nr(gpu->adev->pdev->bus); dev->node_props.max_engine_clk_fcompute = amdgpu_amdkfd_get_max_engine_clock_in_mhz(dev->gpu->adev); dev->node_props.max_engine_clk_ccompute = diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.h b/drivers/gpu/drm/amd/amdkfd/kfd_topology.h index 9f6c949186c1..fca30d00a9bb 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.h @@ -80,6 +80,8 @@ struct kfd_mem_properties { struct attribute attr; }; +#define CACHE_SIBLINGMAP_SIZE 64 + struct kfd_cache_properties { struct list_head list; uint32_t processor_id_low; @@ -90,10 +92,11 @@ struct kfd_cache_properties { uint32_t cache_assoc; uint32_t cache_latency; uint32_t cache_type; - uint8_t sibling_map[CRAT_SIBLINGMAP_SIZE]; + uint8_t sibling_map[CACHE_SIBLINGMAP_SIZE]; struct kfd_dev *gpu; struct kobject *kobj; struct attribute attr; + uint32_t sibling_map_size; }; struct kfd_iolink_properties { @@ -128,7 +131,6 @@ struct kfd_topology_device { uint32_t proximity_domain; struct kfd_node_properties node_props; struct list_head mem_props; - uint32_t cache_count; struct list_head cache_props; struct list_head io_link_props; struct list_head p2p_link_props; diff --git a/drivers/gpu/drm/amd/display/Kconfig b/drivers/gpu/drm/amd/display/Kconfig index 6925e0280dbe..1cdb379a90d7 100644 --- a/drivers/gpu/drm/amd/display/Kconfig +++ b/drivers/gpu/drm/amd/display/Kconfig @@ -6,7 +6,8 @@ config DRM_AMD_DC bool "AMD DC - Enable new display engine" default y select SND_HDA_COMPONENT if SND_HDA_CORE - select DRM_AMD_DC_DCN if (X86 || PPC_LONG_DOUBLE_128) + # !CC_IS_CLANG: https://github.com/ClangBuiltLinux/linux/issues/1752 + select DRM_AMD_DC_DCN if (X86 || PPC_LONG_DOUBLE_128 || (ARM64 && KERNEL_MODE_NEON && !CC_IS_CLANG)) help Choose this option if you want to use the new display engine support for AMDGPU. This adds required support for Vega and diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index d58dd916488a..a8772082883e 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -146,6 +146,14 @@ MODULE_FIRMWARE(FIRMWARE_NAVI12_DMCU); /* Number of bytes in PSP footer for firmware. */ #define PSP_FOOTER_BYTES 0x100 +/* + * DMUB Async to Sync Mechanism Status + */ +#define DMUB_ASYNC_TO_SYNC_ACCESS_FAIL 1 +#define DMUB_ASYNC_TO_SYNC_ACCESS_TIMEOUT 2 +#define DMUB_ASYNC_TO_SYNC_ACCESS_SUCCESS 3 +#define DMUB_ASYNC_TO_SYNC_ACCESS_INVALID 4 + /** * DOC: overview * @@ -1398,7 +1406,6 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) mutex_init(&adev->dm.dc_lock); mutex_init(&adev->dm.audio_lock); - spin_lock_init(&adev->dm.vblank_lock); if(amdgpu_dm_irq_init(adev)) { DRM_ERROR("amdgpu: failed to initialize DM IRQ support.\n"); @@ -1548,6 +1555,9 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) adev->dm.dc->debug.visual_confirm = amdgpu_dc_visual_confirm; + /* TODO: Remove after DP2 receiver gets proper support of Cable ID feature */ + adev->dm.dc->debug.ignore_cable_id = true; + r = dm_dmub_hw_init(adev); if (r) { DRM_ERROR("DMUB interface failed to initialize: status=%d\n", r); @@ -1633,12 +1643,6 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) } } - if (amdgpu_dm_initialize_drm_device(adev)) { - DRM_ERROR( - "amdgpu: failed to initialize sw for display support.\n"); - goto error; - } - /* Enable outbox notification only after IRQ handlers are registered and DMUB is alive. * It is expected that DMUB will resend any pending notifications at this point, for * example HPD from DPIA. @@ -1646,6 +1650,12 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) if (dc_is_dmub_outbox_supported(adev->dm.dc)) dc_enable_dmub_outbox(adev->dm.dc); + if (amdgpu_dm_initialize_drm_device(adev)) { + DRM_ERROR( + "amdgpu: failed to initialize sw for display support.\n"); + goto error; + } + /* create fake encoders for MST */ dm_dp_create_fake_mst_encoders(adev); @@ -4588,6 +4598,7 @@ static int dm_early_init(void *handle) adev_to_drm(adev)->dev, &dev_attr_s3_debug); #endif + adev->dc_enabled = true; return 0; } @@ -5600,16 +5611,14 @@ static void apply_dsc_policy_for_stream(struct amdgpu_dm_connector *aconnector, { struct drm_connector *drm_connector = &aconnector->base; uint32_t link_bandwidth_kbps; - uint32_t max_dsc_target_bpp_limit_override = 0; struct dc *dc = sink->ctx->dc; uint32_t max_supported_bw_in_kbps, timing_bw_in_kbps; uint32_t dsc_max_supported_bw_in_kbps; + uint32_t max_dsc_target_bpp_limit_override = + drm_connector->display_info.max_dsc_bpp; link_bandwidth_kbps = dc_link_bandwidth_kbps(aconnector->dc_link, dc_link_get_link_cap(aconnector->dc_link)); - if (stream->link && stream->link->local_sink) - max_dsc_target_bpp_limit_override = - stream->link->local_sink->edid_caps.panel_patch.max_dsc_target_bpp_limit; /* Set DSC policy according to dsc_clock_en */ dc_dsc_policy_set_enable_dsc_when_not_needed( @@ -5682,7 +5691,7 @@ create_stream_for_sink(struct amdgpu_dm_connector *aconnector, const struct drm_connector_state *con_state = dm_state ? &dm_state->base : NULL; struct dc_stream_state *stream = NULL; - struct drm_display_mode mode = *drm_mode; + struct drm_display_mode mode; struct drm_display_mode saved_mode; struct drm_display_mode *freesync_mode = NULL; bool native_mode_found = false; @@ -5690,12 +5699,14 @@ create_stream_for_sink(struct amdgpu_dm_connector *aconnector, bool scale = dm_state ? (dm_state->scaling != RMX_OFF) : false; int mode_refresh; int preferred_refresh = 0; + enum color_transfer_func tf = TRANSFER_FUNC_UNKNOWN; #if defined(CONFIG_DRM_AMD_DC_DCN) struct dsc_dec_dpcd_caps dsc_caps; #endif struct dc_sink *sink = NULL; + drm_mode_init(&mode, drm_mode); memset(&saved_mode, 0, sizeof(saved_mode)); if (aconnector == NULL) { @@ -5813,7 +5824,9 @@ create_stream_for_sink(struct amdgpu_dm_connector *aconnector, if (stream->link->dpcd_caps.dprx_feature.bits.VSC_SDP_COLORIMETRY_SUPPORTED) stream->use_vsc_sdp_for_colorimetry = true; } - mod_build_vsc_infopacket(stream, &stream->vsc_infopacket, stream->output_color_space); + if (stream->out_transfer_func->tf == TRANSFER_FUNCTION_GAMMA22) + tf = TRANSFER_FUNC_GAMMA_22; + mod_build_vsc_infopacket(stream, &stream->vsc_infopacket, stream->output_color_space, tf); aconnector->psr_skip_count = AMDGPU_DM_PSR_ENTRY_DELAY; } @@ -6143,6 +6156,70 @@ static void handle_edid_mgmt(struct amdgpu_dm_connector *aconnector) create_eml_sink(aconnector); } +static enum dc_status dm_validate_stream_and_context(struct dc *dc, + struct dc_stream_state *stream) +{ + enum dc_status dc_result = DC_ERROR_UNEXPECTED; + struct dc_plane_state *dc_plane_state = NULL; + struct dc_state *dc_state = NULL; + + if (!stream) + goto cleanup; + + dc_plane_state = dc_create_plane_state(dc); + if (!dc_plane_state) + goto cleanup; + + dc_state = dc_create_state(dc); + if (!dc_state) + goto cleanup; + + /* populate stream to plane */ + dc_plane_state->src_rect.height = stream->src.height; + dc_plane_state->src_rect.width = stream->src.width; + dc_plane_state->dst_rect.height = stream->src.height; + dc_plane_state->dst_rect.width = stream->src.width; + dc_plane_state->clip_rect.height = stream->src.height; + dc_plane_state->clip_rect.width = stream->src.width; + dc_plane_state->plane_size.surface_pitch = ((stream->src.width + 255) / 256) * 256; + dc_plane_state->plane_size.surface_size.height = stream->src.height; + dc_plane_state->plane_size.surface_size.width = stream->src.width; + dc_plane_state->plane_size.chroma_size.height = stream->src.height; + dc_plane_state->plane_size.chroma_size.width = stream->src.width; + dc_plane_state->tiling_info.gfx9.swizzle = DC_SW_UNKNOWN; + dc_plane_state->format = SURFACE_PIXEL_FORMAT_GRPH_ARGB8888; + dc_plane_state->tiling_info.gfx9.swizzle = DC_SW_UNKNOWN; + dc_plane_state->rotation = ROTATION_ANGLE_0; + dc_plane_state->is_tiling_rotated = false; + dc_plane_state->tiling_info.gfx8.array_mode = DC_ARRAY_LINEAR_GENERAL; + + dc_result = dc_validate_stream(dc, stream); + if (dc_result == DC_OK) + dc_result = dc_validate_plane(dc, dc_plane_state); + + if (dc_result == DC_OK) + dc_result = dc_add_stream_to_ctx(dc, dc_state, stream); + + if (dc_result == DC_OK && !dc_add_plane_to_context( + dc, + stream, + dc_plane_state, + dc_state)) + dc_result = DC_FAIL_ATTACH_SURFACES; + + if (dc_result == DC_OK) + dc_result = dc_validate_global_state(dc, dc_state, true); + +cleanup: + if (dc_state) + dc_release_state(dc_state); + + if (dc_plane_state) + dc_plane_state_release(dc_plane_state); + + return dc_result; +} + struct dc_stream_state * create_validate_stream_for_sink(struct amdgpu_dm_connector *aconnector, const struct drm_display_mode *drm_mode, @@ -6169,6 +6246,9 @@ create_validate_stream_for_sink(struct amdgpu_dm_connector *aconnector, if (dc_result == DC_OK && stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST) dc_result = dm_dp_mst_is_port_support_mode(aconnector, stream); + if (dc_result == DC_OK) + dc_result = dm_validate_stream_and_context(adev->dm.dc, stream); + if (dc_result != DC_OK) { DRM_DEBUG_KMS("Mode %dx%d (clk %d) failed DC validation with error %d (%s)\n", drm_mode->hdisplay, @@ -6457,7 +6537,7 @@ static int dm_update_mst_vcpi_slots_for_dsc(struct drm_atomic_state *state, struct drm_connector_state *new_con_state; struct amdgpu_dm_connector *aconnector; struct dm_connector_state *dm_conn_state; - int i, j; + int i, j, ret; int vcpi, pbn_div, pbn, slot_num = 0; for_each_new_connector_in_state(state, connector, new_con_state, i) { @@ -6504,8 +6584,11 @@ static int dm_update_mst_vcpi_slots_for_dsc(struct drm_atomic_state *state, dm_conn_state->pbn = pbn; dm_conn_state->vcpi_slots = slot_num; - drm_dp_mst_atomic_enable_dsc(state, aconnector->port, dm_conn_state->pbn, - false); + ret = drm_dp_mst_atomic_enable_dsc(state, aconnector->port, + dm_conn_state->pbn, false); + if (ret < 0) + return ret; + continue; } @@ -7612,9 +7695,10 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state, bundle->surface_updates[planes_count].plane_info = &bundle->plane_infos[planes_count]; - fill_dc_dirty_rects(plane, old_plane_state, new_plane_state, - new_crtc_state, - &bundle->flip_addrs[planes_count]); + if (acrtc_state->stream->link->psr_settings.psr_feature_enabled) + fill_dc_dirty_rects(plane, old_plane_state, + new_plane_state, new_crtc_state, + &bundle->flip_addrs[planes_count]); /* * Only allow immediate flips for fast updates that don't @@ -7830,6 +7914,9 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state, */ if (acrtc_state->stream->link->psr_settings.psr_version >= DC_PSR_VERSION_SU_1 && acrtc_attach->dm_irq_params.allow_psr_entry && +#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY + !amdgpu_dm_crc_window_is_activated(acrtc_state->base.crtc) && +#endif !acrtc_state->stream->link->psr_settings.psr_allow_active) amdgpu_dm_psr_enable(acrtc_state->stream); } else { @@ -8291,8 +8378,8 @@ static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state) #if defined(CONFIG_DRM_AMD_SECURE_DISPLAY) if (amdgpu_dm_crc_window_is_activated(crtc)) { spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags); - acrtc->dm_irq_params.crc_window.update_win = true; - acrtc->dm_irq_params.crc_window.skip_frame_cnt = 2; + acrtc->dm_irq_params.window_param.update_win = true; + acrtc->dm_irq_params.window_param.skip_frame_cnt = 2; spin_lock_irq(&crc_rd_wrk->crc_rd_work_lock); crc_rd_wrk->crtc = crtc; spin_unlock_irq(&crc_rd_wrk->crc_rd_work_lock); @@ -9518,10 +9605,9 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev, #if defined(CONFIG_DRM_AMD_DC_DCN) if (dc_resource_is_dsc_encoding_supported(dc)) { - if (!pre_validate_dsc(state, &dm_state, vars)) { - ret = -EINVAL; + ret = pre_validate_dsc(state, &dm_state, vars); + if (ret != 0) goto fail; - } } #endif @@ -9616,9 +9702,9 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev, } #if defined(CONFIG_DRM_AMD_DC_DCN) - if (!compute_mst_dsc_configs_for_state(state, dm_state->context, vars)) { + ret = compute_mst_dsc_configs_for_state(state, dm_state->context, vars); + if (ret) { DRM_DEBUG_DRIVER("compute_mst_dsc_configs_for_state() failed\n"); - ret = -EINVAL; goto fail; } @@ -10098,6 +10184,8 @@ static int amdgpu_dm_set_dmub_async_sync_status(bool is_cmd_aux, *operation_result = AUX_RET_ERROR_TIMEOUT; } else if (status_type == DMUB_ASYNC_TO_SYNC_ACCESS_FAIL) { *operation_result = AUX_RET_ERROR_ENGINE_ACQUIRE; + } else if (status_type == DMUB_ASYNC_TO_SYNC_ACCESS_INVALID) { + *operation_result = AUX_RET_ERROR_INVALID_REPLY; } else { *operation_result = AUX_RET_ERROR_UNKNOWN; } @@ -10145,6 +10233,16 @@ int amdgpu_dm_process_dmub_aux_transfer_sync(bool is_cmd_aux, struct dc_context payload->reply[0] = adev->dm.dmub_notify->aux_reply.command; if (!payload->write && adev->dm.dmub_notify->aux_reply.length && payload->reply[0] == AUX_TRANSACTION_REPLY_AUX_ACK) { + + if (payload->length != adev->dm.dmub_notify->aux_reply.length) { + DRM_WARN("invalid read from DPIA AUX %x(%d) got length %d!\n", + payload->address, payload->length, + adev->dm.dmub_notify->aux_reply.length); + return amdgpu_dm_set_dmub_async_sync_status(is_cmd_aux, ctx, + DMUB_ASYNC_TO_SYNC_ACCESS_INVALID, + (uint32_t *)operation_result); + } + memcpy(payload->data, adev->dm.dmub_notify->aux_reply.data, adev->dm.dmub_notify->aux_reply.length); } @@ -10165,8 +10263,8 @@ int amdgpu_dm_process_dmub_aux_transfer_sync(bool is_cmd_aux, struct dc_context */ bool check_seamless_boot_capability(struct amdgpu_device *adev) { - switch (adev->asic_type) { - case CHIP_VANGOGH: + switch (adev->ip_versions[DCE_HWIP][0]) { + case IP_VERSION(3, 0, 1): if (!adev->mman.keep_stolen_vga_memory) return true; break; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h index b5ce15c43bcc..83436ef3b26b 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h @@ -51,12 +51,6 @@ #define AMDGPU_DMUB_NOTIFICATION_MAX 5 /* - * DMUB Async to Sync Mechanism Status - */ -#define DMUB_ASYNC_TO_SYNC_ACCESS_FAIL 1 -#define DMUB_ASYNC_TO_SYNC_ACCESS_TIMEOUT 2 -#define DMUB_ASYNC_TO_SYNC_ACCESS_SUCCESS 3 -/* #include "include/amdgpu_dal_power_if.h" #include "amdgpu_dm_irq.h" */ @@ -366,13 +360,6 @@ struct amdgpu_display_manager { struct mutex audio_lock; /** - * @vblank_lock: - * - * Guards access to deferred vblank work state. - */ - spinlock_t vblank_lock; - - /** * @audio_component: * * Used to notify ELD changes to sound driver. diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c index 8a441a22c46e..66df2394d7e4 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.c @@ -89,13 +89,13 @@ static void amdgpu_dm_set_crc_window_default(struct drm_crtc *crtc) struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc); spin_lock_irq(&drm_dev->event_lock); - acrtc->dm_irq_params.crc_window.x_start = 0; - acrtc->dm_irq_params.crc_window.y_start = 0; - acrtc->dm_irq_params.crc_window.x_end = 0; - acrtc->dm_irq_params.crc_window.y_end = 0; - acrtc->dm_irq_params.crc_window.activated = false; - acrtc->dm_irq_params.crc_window.update_win = false; - acrtc->dm_irq_params.crc_window.skip_frame_cnt = 0; + acrtc->dm_irq_params.window_param.x_start = 0; + acrtc->dm_irq_params.window_param.y_start = 0; + acrtc->dm_irq_params.window_param.x_end = 0; + acrtc->dm_irq_params.window_param.y_end = 0; + acrtc->dm_irq_params.window_param.activated = false; + acrtc->dm_irq_params.window_param.update_win = false; + acrtc->dm_irq_params.window_param.skip_frame_cnt = 0; spin_unlock_irq(&drm_dev->event_lock); } @@ -123,6 +123,8 @@ static void amdgpu_dm_crtc_notify_ta_to_read(struct work_struct *work) phy_id = crc_rd_wrk->phy_inst; spin_unlock_irq(&crc_rd_wrk->crc_rd_work_lock); + mutex_lock(&psp->securedisplay_context.mutex); + psp_prep_securedisplay_cmd_buf(psp, &securedisplay_cmd, TA_SECUREDISPLAY_COMMAND__SEND_ROI_CRC); securedisplay_cmd->securedisplay_in_message.send_roi_crc.phy_id = @@ -133,6 +135,24 @@ static void amdgpu_dm_crtc_notify_ta_to_read(struct work_struct *work) psp_securedisplay_parse_resp_status(psp, securedisplay_cmd->status); } } + + mutex_unlock(&psp->securedisplay_context.mutex); +} + +static void +amdgpu_dm_forward_crc_window(struct work_struct *work) +{ + struct crc_fw_work *crc_fw_wrk; + struct amdgpu_display_manager *dm; + + crc_fw_wrk = container_of(work, struct crc_fw_work, forward_roi_work); + dm = crc_fw_wrk->dm; + + mutex_lock(&dm->dc_lock); + dc_stream_forward_crc_window(dm->dc, &crc_fw_wrk->rect, crc_fw_wrk->stream, crc_fw_wrk->is_stop_cmd); + mutex_unlock(&dm->dc_lock); + + kfree(crc_fw_wrk); } bool amdgpu_dm_crc_window_is_activated(struct drm_crtc *crtc) @@ -142,7 +162,7 @@ bool amdgpu_dm_crc_window_is_activated(struct drm_crtc *crtc) bool ret = false; spin_lock_irq(&drm_dev->event_lock); - ret = acrtc->dm_irq_params.crc_window.activated; + ret = acrtc->dm_irq_params.window_param.activated; spin_unlock_irq(&drm_dev->event_lock); return ret; @@ -187,9 +207,11 @@ int amdgpu_dm_crtc_configure_crc_source(struct drm_crtc *crtc, if (adev->dm.crc_rd_wrk) { flush_work(&adev->dm.crc_rd_wrk->notify_ta_work); spin_lock_irq(&adev->dm.crc_rd_wrk->crc_rd_work_lock); + if (adev->dm.crc_rd_wrk->crtc == crtc) { - dc_stream_stop_dmcu_crc_win_update(stream_state->ctx->dc, - dm_crtc_state->stream); + /* stop ROI update on this crtc */ + dc_stream_forward_crc_window(stream_state->ctx->dc, + NULL, stream_state, true); adev->dm.crc_rd_wrk->crtc = NULL; } spin_unlock_irq(&adev->dm.crc_rd_wrk->crc_rd_work_lock); @@ -439,14 +461,9 @@ void amdgpu_dm_crtc_handle_crc_window_irq(struct drm_crtc *crtc) enum amdgpu_dm_pipe_crc_source cur_crc_src; struct amdgpu_crtc *acrtc = NULL; struct amdgpu_device *adev = NULL; - struct crc_rd_work *crc_rd_wrk = NULL; - struct crc_params *crc_window = NULL, tmp_window; + struct crc_rd_work *crc_rd_wrk; + struct crc_fw_work *crc_fw_wrk; unsigned long flags1, flags2; - struct crtc_position position; - uint32_t v_blank; - uint32_t v_back_porch; - uint32_t crc_window_latch_up_line; - struct dc_crtc_timing *timing_out; if (crtc == NULL) return; @@ -458,74 +475,54 @@ void amdgpu_dm_crtc_handle_crc_window_irq(struct drm_crtc *crtc) spin_lock_irqsave(&drm_dev->event_lock, flags1); stream_state = acrtc->dm_irq_params.stream; cur_crc_src = acrtc->dm_irq_params.crc_src; - timing_out = &stream_state->timing; /* Early return if CRC capture is not enabled. */ if (!amdgpu_dm_is_valid_crc_source(cur_crc_src)) goto cleanup; - if (dm_is_crc_source_crtc(cur_crc_src)) { - if (acrtc->dm_irq_params.crc_window.activated) { - if (acrtc->dm_irq_params.crc_window.update_win) { - if (acrtc->dm_irq_params.crc_window.skip_frame_cnt) { - acrtc->dm_irq_params.crc_window.skip_frame_cnt -= 1; - goto cleanup; - } - crc_window = &tmp_window; - - tmp_window.windowa_x_start = - acrtc->dm_irq_params.crc_window.x_start; - tmp_window.windowa_y_start = - acrtc->dm_irq_params.crc_window.y_start; - tmp_window.windowa_x_end = - acrtc->dm_irq_params.crc_window.x_end; - tmp_window.windowa_y_end = - acrtc->dm_irq_params.crc_window.y_end; - tmp_window.windowb_x_start = - acrtc->dm_irq_params.crc_window.x_start; - tmp_window.windowb_y_start = - acrtc->dm_irq_params.crc_window.y_start; - tmp_window.windowb_x_end = - acrtc->dm_irq_params.crc_window.x_end; - tmp_window.windowb_y_end = - acrtc->dm_irq_params.crc_window.y_end; - - dc_stream_forward_dmcu_crc_window(stream_state->ctx->dc, - stream_state, crc_window); - - acrtc->dm_irq_params.crc_window.update_win = false; - - dc_stream_get_crtc_position(stream_state->ctx->dc, &stream_state, 1, - &position.vertical_count, - &position.nominal_vcount); - - v_blank = timing_out->v_total - timing_out->v_border_top - - timing_out->v_addressable - timing_out->v_border_bottom; - - v_back_porch = v_blank - timing_out->v_front_porch - - timing_out->v_sync_width; - - crc_window_latch_up_line = v_back_porch + timing_out->v_sync_width; - - /* take 3 lines margin*/ - if ((position.vertical_count + 3) >= crc_window_latch_up_line) - acrtc->dm_irq_params.crc_window.skip_frame_cnt = 1; - else - acrtc->dm_irq_params.crc_window.skip_frame_cnt = 0; - } else { - if (acrtc->dm_irq_params.crc_window.skip_frame_cnt == 0) { - if (adev->dm.crc_rd_wrk) { - crc_rd_wrk = adev->dm.crc_rd_wrk; - spin_lock_irqsave(&crc_rd_wrk->crc_rd_work_lock, flags2); - crc_rd_wrk->phy_inst = - stream_state->link->link_enc_hw_inst; - spin_unlock_irqrestore(&crc_rd_wrk->crc_rd_work_lock, flags2); - schedule_work(&crc_rd_wrk->notify_ta_work); - } - } else { - acrtc->dm_irq_params.crc_window.skip_frame_cnt -= 1; - } - } + if (!dm_is_crc_source_crtc(cur_crc_src)) + goto cleanup; + + if (!acrtc->dm_irq_params.window_param.activated) + goto cleanup; + + if (acrtc->dm_irq_params.window_param.update_win) { + if (acrtc->dm_irq_params.window_param.skip_frame_cnt) { + acrtc->dm_irq_params.window_param.skip_frame_cnt -= 1; + goto cleanup; + } + + /* prepare work for dmub to update ROI */ + crc_fw_wrk = kzalloc(sizeof(*crc_fw_wrk), GFP_ATOMIC); + if (!crc_fw_wrk) + goto cleanup; + + INIT_WORK(&crc_fw_wrk->forward_roi_work, amdgpu_dm_forward_crc_window); + crc_fw_wrk->dm = &adev->dm; + crc_fw_wrk->stream = stream_state; + crc_fw_wrk->rect.x = acrtc->dm_irq_params.window_param.x_start; + crc_fw_wrk->rect.y = acrtc->dm_irq_params.window_param.y_start; + crc_fw_wrk->rect.width = acrtc->dm_irq_params.window_param.x_end - + acrtc->dm_irq_params.window_param.x_start; + crc_fw_wrk->rect.height = acrtc->dm_irq_params.window_param.y_end - + acrtc->dm_irq_params.window_param.y_start; + schedule_work(&crc_fw_wrk->forward_roi_work); + + acrtc->dm_irq_params.window_param.update_win = false; + acrtc->dm_irq_params.window_param.skip_frame_cnt = 1; + + } else { + if (acrtc->dm_irq_params.window_param.skip_frame_cnt) { + acrtc->dm_irq_params.window_param.skip_frame_cnt -= 1; + goto cleanup; + } + + if (adev->dm.crc_rd_wrk) { + crc_rd_wrk = adev->dm.crc_rd_wrk; + spin_lock_irqsave(&crc_rd_wrk->crc_rd_work_lock, flags2); + crc_rd_wrk->phy_inst = stream_state->link->link_enc_hw_inst; + spin_unlock_irqrestore(&crc_rd_wrk->crc_rd_work_lock, flags2); + schedule_work(&crc_rd_wrk->notify_ta_work); } } diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.h index f07850db60a6..71bce608d751 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crc.h @@ -40,7 +40,7 @@ enum amdgpu_dm_pipe_crc_source { }; #ifdef CONFIG_DRM_AMD_SECURE_DISPLAY -struct crc_window_parm { +struct crc_window_param { uint16_t x_start; uint16_t y_start; uint16_t x_end; @@ -53,6 +53,7 @@ struct crc_window_parm { int skip_frame_cnt; }; +/* read_work for driver to call PSP to read */ struct crc_rd_work { struct work_struct notify_ta_work; /* To protect crc_rd_work carried fields*/ @@ -60,6 +61,15 @@ struct crc_rd_work { struct drm_crtc *crtc; uint8_t phy_inst; }; + +/* forward_work for driver to forward ROI to dmu */ +struct crc_fw_work { + struct work_struct forward_roi_work; + struct amdgpu_display_manager *dm; + struct dc_stream_state *stream; + struct rect rect; + bool is_stop_cmd; +}; #endif static inline bool amdgpu_dm_is_valid_crc_source(enum amdgpu_dm_pipe_crc_source source) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c index 594fe8a4d02b..22125daf9dcf 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c @@ -127,6 +127,9 @@ static void vblank_control_worker(struct work_struct *work) amdgpu_dm_psr_disable(vblank_work->stream); } else if (vblank_work->stream->link->psr_settings.psr_feature_enabled && !vblank_work->stream->link->psr_settings.psr_allow_active && +#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY + !amdgpu_dm_crc_window_is_activated(&vblank_work->acrtc->base) && +#endif vblank_work->acrtc->dm_irq_params.allow_psr_entry) { amdgpu_dm_psr_enable(vblank_work->stream); } @@ -412,7 +415,7 @@ int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm, { struct amdgpu_crtc *acrtc = NULL; struct drm_plane *cursor_plane; - + bool is_dcn; int res = -ENOMEM; cursor_plane = kzalloc(sizeof(*cursor_plane), GFP_KERNEL); @@ -450,8 +453,14 @@ int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm, acrtc->otg_inst = -1; dm->adev->mode_info.crtcs[crtc_index] = acrtc; - drm_crtc_enable_color_mgmt(&acrtc->base, MAX_COLOR_LUT_ENTRIES, + + /* Don't enable DRM CRTC degamma property for DCE since it doesn't + * support programmable degamma anywhere. + */ + is_dcn = dm->adev->dm.dc->caps.color.dpp.dcn_arch; + drm_crtc_enable_color_mgmt(&acrtc->base, is_dcn ? MAX_COLOR_LUT_ENTRIES : 0, true, MAX_COLOR_LUT_ENTRIES); + drm_mode_crtc_set_gamma_size(&acrtc->base, MAX_COLOR_LEGACY_LUT_ENTRIES); return 0; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c index ee242d9d8b06..2c43cdd2e707 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c @@ -38,6 +38,10 @@ #include "link_hwss.h" #include "dc/dc_dmub_srv.h" +#ifdef CONFIG_DRM_AMD_SECURE_DISPLAY +#include "amdgpu_dm_psr.h" +#endif + struct dmub_debugfs_trace_header { uint32_t entry_count; uint32_t reserved[3]; @@ -299,6 +303,8 @@ static ssize_t dp_link_settings_write(struct file *f, const char __user *buf, case LINK_RATE_HIGH2: case LINK_RATE_HIGH3: case LINK_RATE_UHBR10: + case LINK_RATE_UHBR13_5: + case LINK_RATE_UHBR20: break; default: valid_input = false; @@ -3079,8 +3085,8 @@ static int crc_win_x_start_set(void *data, u64 val) struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc); spin_lock_irq(&drm_dev->event_lock); - acrtc->dm_irq_params.crc_window.x_start = (uint16_t) val; - acrtc->dm_irq_params.crc_window.update_win = false; + acrtc->dm_irq_params.window_param.x_start = (uint16_t) val; + acrtc->dm_irq_params.window_param.update_win = false; spin_unlock_irq(&drm_dev->event_lock); return 0; @@ -3096,7 +3102,7 @@ static int crc_win_x_start_get(void *data, u64 *val) struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc); spin_lock_irq(&drm_dev->event_lock); - *val = acrtc->dm_irq_params.crc_window.x_start; + *val = acrtc->dm_irq_params.window_param.x_start; spin_unlock_irq(&drm_dev->event_lock); return 0; @@ -3116,8 +3122,8 @@ static int crc_win_y_start_set(void *data, u64 val) struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc); spin_lock_irq(&drm_dev->event_lock); - acrtc->dm_irq_params.crc_window.y_start = (uint16_t) val; - acrtc->dm_irq_params.crc_window.update_win = false; + acrtc->dm_irq_params.window_param.y_start = (uint16_t) val; + acrtc->dm_irq_params.window_param.update_win = false; spin_unlock_irq(&drm_dev->event_lock); return 0; @@ -3133,7 +3139,7 @@ static int crc_win_y_start_get(void *data, u64 *val) struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc); spin_lock_irq(&drm_dev->event_lock); - *val = acrtc->dm_irq_params.crc_window.y_start; + *val = acrtc->dm_irq_params.window_param.y_start; spin_unlock_irq(&drm_dev->event_lock); return 0; @@ -3152,8 +3158,8 @@ static int crc_win_x_end_set(void *data, u64 val) struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc); spin_lock_irq(&drm_dev->event_lock); - acrtc->dm_irq_params.crc_window.x_end = (uint16_t) val; - acrtc->dm_irq_params.crc_window.update_win = false; + acrtc->dm_irq_params.window_param.x_end = (uint16_t) val; + acrtc->dm_irq_params.window_param.update_win = false; spin_unlock_irq(&drm_dev->event_lock); return 0; @@ -3169,7 +3175,7 @@ static int crc_win_x_end_get(void *data, u64 *val) struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc); spin_lock_irq(&drm_dev->event_lock); - *val = acrtc->dm_irq_params.crc_window.x_end; + *val = acrtc->dm_irq_params.window_param.x_end; spin_unlock_irq(&drm_dev->event_lock); return 0; @@ -3188,8 +3194,8 @@ static int crc_win_y_end_set(void *data, u64 val) struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc); spin_lock_irq(&drm_dev->event_lock); - acrtc->dm_irq_params.crc_window.y_end = (uint16_t) val; - acrtc->dm_irq_params.crc_window.update_win = false; + acrtc->dm_irq_params.window_param.y_end = (uint16_t) val; + acrtc->dm_irq_params.window_param.update_win = false; spin_unlock_irq(&drm_dev->event_lock); return 0; @@ -3205,7 +3211,7 @@ static int crc_win_y_end_get(void *data, u64 *val) struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc); spin_lock_irq(&drm_dev->event_lock); - *val = acrtc->dm_irq_params.crc_window.y_end; + *val = acrtc->dm_irq_params.window_param.y_end; spin_unlock_irq(&drm_dev->event_lock); return 0; @@ -3228,31 +3234,38 @@ static int crc_win_update_set(void *data, u64 val) return 0; if (val) { + new_acrtc = to_amdgpu_crtc(new_crtc); + mutex_lock(&adev->dm.dc_lock); + /* PSR may write to OTG CRC window control register, + * so close it before starting secure_display. + */ + amdgpu_dm_psr_disable(new_acrtc->dm_irq_params.stream); + spin_lock_irq(&adev_to_drm(adev)->event_lock); spin_lock_irq(&crc_rd_wrk->crc_rd_work_lock); if (crc_rd_wrk->crtc) { old_crtc = crc_rd_wrk->crtc; old_acrtc = to_amdgpu_crtc(old_crtc); } - new_acrtc = to_amdgpu_crtc(new_crtc); if (old_crtc && old_crtc != new_crtc) { - old_acrtc->dm_irq_params.crc_window.activated = false; - old_acrtc->dm_irq_params.crc_window.update_win = false; - old_acrtc->dm_irq_params.crc_window.skip_frame_cnt = 0; + old_acrtc->dm_irq_params.window_param.activated = false; + old_acrtc->dm_irq_params.window_param.update_win = false; + old_acrtc->dm_irq_params.window_param.skip_frame_cnt = 0; - new_acrtc->dm_irq_params.crc_window.activated = true; - new_acrtc->dm_irq_params.crc_window.update_win = true; - new_acrtc->dm_irq_params.crc_window.skip_frame_cnt = 0; + new_acrtc->dm_irq_params.window_param.activated = true; + new_acrtc->dm_irq_params.window_param.update_win = true; + new_acrtc->dm_irq_params.window_param.skip_frame_cnt = 0; crc_rd_wrk->crtc = new_crtc; } else { - new_acrtc->dm_irq_params.crc_window.activated = true; - new_acrtc->dm_irq_params.crc_window.update_win = true; - new_acrtc->dm_irq_params.crc_window.skip_frame_cnt = 0; + new_acrtc->dm_irq_params.window_param.activated = true; + new_acrtc->dm_irq_params.window_param.update_win = true; + new_acrtc->dm_irq_params.window_param.skip_frame_cnt = 0; crc_rd_wrk->crtc = new_crtc; } spin_unlock_irq(&crc_rd_wrk->crc_rd_work_lock); spin_unlock_irq(&adev_to_drm(adev)->event_lock); + mutex_unlock(&adev->dm.dc_lock); } return 0; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c index 6202e31c7e3a..a7fd98f57f94 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c @@ -495,7 +495,7 @@ static void update_config(void *handle, struct cp_psp_stream_config *config) link->dp.mst_enabled = config->mst_enabled; link->dp.usb4_enabled = config->usb4_enabled; display->adjust.disable = MOD_HDCP_DISPLAY_DISABLE_AUTHENTICATION; - link->adjust.auth_delay = 3; + link->adjust.auth_delay = 0; link->adjust.hdcp1.disable = 0; conn_state = aconnector->base.state; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c index f0b01c8dc4a6..e47098fa5aac 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c @@ -42,39 +42,6 @@ #include "dm_helpers.h" #include "ddc_service_types.h" -struct monitor_patch_info { - unsigned int manufacturer_id; - unsigned int product_id; - void (*patch_func)(struct dc_edid_caps *edid_caps, unsigned int param); - unsigned int patch_param; -}; -static void set_max_dsc_bpp_limit(struct dc_edid_caps *edid_caps, unsigned int param); - -static const struct monitor_patch_info monitor_patch_table[] = { -{0x6D1E, 0x5BBF, set_max_dsc_bpp_limit, 15}, -{0x6D1E, 0x5B9A, set_max_dsc_bpp_limit, 15}, -}; - -static void set_max_dsc_bpp_limit(struct dc_edid_caps *edid_caps, unsigned int param) -{ - if (edid_caps) - edid_caps->panel_patch.max_dsc_target_bpp_limit = param; -} - -static int amdgpu_dm_patch_edid_caps(struct dc_edid_caps *edid_caps) -{ - int i, ret = 0; - - for (i = 0; i < ARRAY_SIZE(monitor_patch_table); i++) - if ((edid_caps->manufacturer_id == monitor_patch_table[i].manufacturer_id) - && (edid_caps->product_id == monitor_patch_table[i].product_id)) { - monitor_patch_table[i].patch_func(edid_caps, monitor_patch_table[i].patch_param); - ret++; - } - - return ret; -} - /* dm_helpers_parse_edid_caps * * Parse edid caps @@ -149,8 +116,6 @@ enum dc_edid_status dm_helpers_parse_edid_caps( kfree(sads); kfree(sadb); - amdgpu_dm_patch_edid_caps(edid_caps); - return result; } @@ -1006,3 +971,11 @@ void dm_helpers_enable_periodic_detection(struct dc_context *ctx, bool enable) { /* TODO: add periodic detection implementation */ } + +void dm_helpers_dp_mst_update_branch_bandwidth( + struct dc_context *ctx, + struct dc_link *link) +{ + // TODO +} + diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq_params.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq_params.h index 79b5f9999fec..5c9303241aeb 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq_params.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq_params.h @@ -39,7 +39,7 @@ struct dm_irq_params { #ifdef CONFIG_DEBUG_FS enum amdgpu_dm_pipe_crc_source crc_src; #ifdef CONFIG_DRM_AMD_SECURE_DISPLAY - struct crc_window_parm crc_window; + struct crc_window_param window_param; #endif #endif }; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c index 6ff96b4bdda5..24d859ad4712 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c @@ -642,15 +642,18 @@ static void set_dsc_configs_from_fairness_vars(struct dsc_mst_fairness_params *p int count, int k) { + struct drm_connector *drm_connector; int i; for (i = 0; i < count; i++) { + drm_connector = ¶ms[i].aconnector->base; + memset(¶ms[i].timing->dsc_cfg, 0, sizeof(params[i].timing->dsc_cfg)); if (vars[i + k].dsc_enabled && dc_dsc_compute_config( params[i].sink->ctx->dc->res_pool->dscs[0], ¶ms[i].sink->dsc_caps.dsc_dec_caps, params[i].sink->ctx->dc->debug.dsc_min_slice_height_override, - params[i].sink->edid_caps.panel_patch.max_dsc_target_bpp_limit, + drm_connector->display_info.max_dsc_bpp, 0, params[i].timing, ¶ms[i].timing->dsc_cfg)) { @@ -692,24 +695,28 @@ static int bpp_x16_from_pbn(struct dsc_mst_fairness_params param, int pbn) struct dc_dsc_config dsc_config; u64 kbps; + struct drm_connector *drm_connector = ¶m.aconnector->base; + uint32_t max_dsc_target_bpp_limit_override = + drm_connector->display_info.max_dsc_bpp; + kbps = div_u64((u64)pbn * 994 * 8 * 54, 64); dc_dsc_compute_config( param.sink->ctx->dc->res_pool->dscs[0], ¶m.sink->dsc_caps.dsc_dec_caps, param.sink->ctx->dc->debug.dsc_min_slice_height_override, - param.sink->edid_caps.panel_patch.max_dsc_target_bpp_limit, + max_dsc_target_bpp_limit_override, (int) kbps, param.timing, &dsc_config); return dsc_config.bits_per_pixel; } -static bool increase_dsc_bpp(struct drm_atomic_state *state, - struct drm_dp_mst_topology_state *mst_state, - struct dc_link *dc_link, - struct dsc_mst_fairness_params *params, - struct dsc_mst_fairness_vars *vars, - int count, - int k) +static int increase_dsc_bpp(struct drm_atomic_state *state, + struct drm_dp_mst_topology_state *mst_state, + struct dc_link *dc_link, + struct dsc_mst_fairness_params *params, + struct dsc_mst_fairness_vars *vars, + int count, + int k) { int i; bool bpp_increased[MAX_PIPES]; @@ -719,6 +726,7 @@ static bool increase_dsc_bpp(struct drm_atomic_state *state, int remaining_to_increase = 0; int link_timeslots_used; int fair_pbn_alloc; + int ret = 0; for (i = 0; i < count; i++) { if (vars[i + k].dsc_enabled) { @@ -757,52 +765,60 @@ static bool increase_dsc_bpp(struct drm_atomic_state *state, if (initial_slack[next_index] > fair_pbn_alloc) { vars[next_index].pbn += fair_pbn_alloc; - if (drm_dp_atomic_find_time_slots(state, - params[next_index].port->mgr, - params[next_index].port, - vars[next_index].pbn) < 0) - return false; - if (!drm_dp_mst_atomic_check(state)) { + ret = drm_dp_atomic_find_time_slots(state, + params[next_index].port->mgr, + params[next_index].port, + vars[next_index].pbn); + if (ret < 0) + return ret; + + ret = drm_dp_mst_atomic_check(state); + if (ret == 0) { vars[next_index].bpp_x16 = bpp_x16_from_pbn(params[next_index], vars[next_index].pbn); } else { vars[next_index].pbn -= fair_pbn_alloc; - if (drm_dp_atomic_find_time_slots(state, - params[next_index].port->mgr, - params[next_index].port, - vars[next_index].pbn) < 0) - return false; + ret = drm_dp_atomic_find_time_slots(state, + params[next_index].port->mgr, + params[next_index].port, + vars[next_index].pbn); + if (ret < 0) + return ret; } } else { vars[next_index].pbn += initial_slack[next_index]; - if (drm_dp_atomic_find_time_slots(state, - params[next_index].port->mgr, - params[next_index].port, - vars[next_index].pbn) < 0) - return false; - if (!drm_dp_mst_atomic_check(state)) { + ret = drm_dp_atomic_find_time_slots(state, + params[next_index].port->mgr, + params[next_index].port, + vars[next_index].pbn); + if (ret < 0) + return ret; + + ret = drm_dp_mst_atomic_check(state); + if (ret == 0) { vars[next_index].bpp_x16 = params[next_index].bw_range.max_target_bpp_x16; } else { vars[next_index].pbn -= initial_slack[next_index]; - if (drm_dp_atomic_find_time_slots(state, - params[next_index].port->mgr, - params[next_index].port, - vars[next_index].pbn) < 0) - return false; + ret = drm_dp_atomic_find_time_slots(state, + params[next_index].port->mgr, + params[next_index].port, + vars[next_index].pbn); + if (ret < 0) + return ret; } } bpp_increased[next_index] = true; remaining_to_increase--; } - return true; + return 0; } -static bool try_disable_dsc(struct drm_atomic_state *state, - struct dc_link *dc_link, - struct dsc_mst_fairness_params *params, - struct dsc_mst_fairness_vars *vars, - int count, - int k) +static int try_disable_dsc(struct drm_atomic_state *state, + struct dc_link *dc_link, + struct dsc_mst_fairness_params *params, + struct dsc_mst_fairness_vars *vars, + int count, + int k) { int i; bool tried[MAX_PIPES]; @@ -810,6 +826,7 @@ static bool try_disable_dsc(struct drm_atomic_state *state, int max_kbps_increase; int next_index; int remaining_to_try = 0; + int ret; for (i = 0; i < count; i++) { if (vars[i + k].dsc_enabled @@ -840,49 +857,52 @@ static bool try_disable_dsc(struct drm_atomic_state *state, break; vars[next_index].pbn = kbps_to_peak_pbn(params[next_index].bw_range.stream_kbps); - if (drm_dp_atomic_find_time_slots(state, - params[next_index].port->mgr, - params[next_index].port, - vars[next_index].pbn) < 0) - return false; + ret = drm_dp_atomic_find_time_slots(state, + params[next_index].port->mgr, + params[next_index].port, + vars[next_index].pbn); + if (ret < 0) + return ret; - if (!drm_dp_mst_atomic_check(state)) { + ret = drm_dp_mst_atomic_check(state); + if (ret == 0) { vars[next_index].dsc_enabled = false; vars[next_index].bpp_x16 = 0; } else { vars[next_index].pbn = kbps_to_peak_pbn(params[next_index].bw_range.max_kbps); - if (drm_dp_atomic_find_time_slots(state, - params[next_index].port->mgr, - params[next_index].port, - vars[next_index].pbn) < 0) - return false; + ret = drm_dp_atomic_find_time_slots(state, + params[next_index].port->mgr, + params[next_index].port, + vars[next_index].pbn); + if (ret < 0) + return ret; } tried[next_index] = true; remaining_to_try--; } - return true; + return 0; } -static bool compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, - struct dc_state *dc_state, - struct dc_link *dc_link, - struct dsc_mst_fairness_vars *vars, - struct drm_dp_mst_topology_mgr *mgr, - int *link_vars_start_index) +static int compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, + struct dc_state *dc_state, + struct dc_link *dc_link, + struct dsc_mst_fairness_vars *vars, + struct drm_dp_mst_topology_mgr *mgr, + int *link_vars_start_index) { struct dc_stream_state *stream; struct dsc_mst_fairness_params params[MAX_PIPES]; struct amdgpu_dm_connector *aconnector; struct drm_dp_mst_topology_state *mst_state = drm_atomic_get_mst_topology_state(state, mgr); int count = 0; - int i, k; + int i, k, ret; bool debugfs_overwrite = false; memset(params, 0, sizeof(params)); if (IS_ERR(mst_state)) - return false; + return PTR_ERR(mst_state); mst_state->pbn_div = dm_mst_get_pbn_divider(dc_link); #if defined(CONFIG_DRM_AMD_DC_DCN) @@ -933,7 +953,7 @@ static bool compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, if (count == 0) { ASSERT(0); - return true; + return 0; } /* k is start index of vars for current phy link used by mst hub */ @@ -947,13 +967,17 @@ static bool compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.stream_kbps); vars[i + k].dsc_enabled = false; vars[i + k].bpp_x16 = 0; - if (drm_dp_atomic_find_time_slots(state, params[i].port->mgr, params[i].port, - vars[i + k].pbn) < 0) - return false; + ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr, params[i].port, + vars[i + k].pbn); + if (ret < 0) + return ret; } - if (!drm_dp_mst_atomic_check(state) && !debugfs_overwrite) { + ret = drm_dp_mst_atomic_check(state); + if (ret == 0 && !debugfs_overwrite) { set_dsc_configs_from_fairness_vars(params, vars, count, k); - return true; + return 0; + } else if (ret != -ENOSPC) { + return ret; } /* Try max compression */ @@ -962,31 +986,36 @@ static bool compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.min_kbps); vars[i + k].dsc_enabled = true; vars[i + k].bpp_x16 = params[i].bw_range.min_target_bpp_x16; - if (drm_dp_atomic_find_time_slots(state, params[i].port->mgr, - params[i].port, vars[i + k].pbn) < 0) - return false; + ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr, + params[i].port, vars[i + k].pbn); + if (ret < 0) + return ret; } else { vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.stream_kbps); vars[i + k].dsc_enabled = false; vars[i + k].bpp_x16 = 0; - if (drm_dp_atomic_find_time_slots(state, params[i].port->mgr, - params[i].port, vars[i + k].pbn) < 0) - return false; + ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr, + params[i].port, vars[i + k].pbn); + if (ret < 0) + return ret; } } - if (drm_dp_mst_atomic_check(state)) - return false; + ret = drm_dp_mst_atomic_check(state); + if (ret != 0) + return ret; /* Optimize degree of compression */ - if (!increase_dsc_bpp(state, mst_state, dc_link, params, vars, count, k)) - return false; + ret = increase_dsc_bpp(state, mst_state, dc_link, params, vars, count, k); + if (ret < 0) + return ret; - if (!try_disable_dsc(state, dc_link, params, vars, count, k)) - return false; + ret = try_disable_dsc(state, dc_link, params, vars, count, k); + if (ret < 0) + return ret; set_dsc_configs_from_fairness_vars(params, vars, count, k); - return true; + return 0; } static bool is_dsc_need_re_compute( @@ -1087,15 +1116,17 @@ static bool is_dsc_need_re_compute( return is_dsc_need_re_compute; } -bool compute_mst_dsc_configs_for_state(struct drm_atomic_state *state, - struct dc_state *dc_state, - struct dsc_mst_fairness_vars *vars) +int compute_mst_dsc_configs_for_state(struct drm_atomic_state *state, + struct dc_state *dc_state, + struct dsc_mst_fairness_vars *vars) { int i, j; struct dc_stream_state *stream; bool computed_streams[MAX_PIPES]; struct amdgpu_dm_connector *aconnector; + struct drm_dp_mst_topology_mgr *mst_mgr; int link_vars_start_index = 0; + int ret = 0; for (i = 0; i < dc_state->stream_count; i++) computed_streams[i] = false; @@ -1108,7 +1139,7 @@ bool compute_mst_dsc_configs_for_state(struct drm_atomic_state *state, aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context; - if (!aconnector || !aconnector->dc_sink) + if (!aconnector || !aconnector->dc_sink || !aconnector->port) continue; if (!aconnector->dc_sink->dsc_caps.dsc_dec_caps.is_dsc_supported) @@ -1118,19 +1149,16 @@ bool compute_mst_dsc_configs_for_state(struct drm_atomic_state *state, continue; if (dcn20_remove_stream_from_ctx(stream->ctx->dc, dc_state, stream) != DC_OK) - return false; + return -EINVAL; if (!is_dsc_need_re_compute(state, dc_state, stream->link)) continue; - mutex_lock(&aconnector->mst_mgr.lock); - if (!compute_mst_dsc_configs_for_link(state, dc_state, stream->link, vars, - &aconnector->mst_mgr, - &link_vars_start_index)) { - mutex_unlock(&aconnector->mst_mgr.lock); - return false; - } - mutex_unlock(&aconnector->mst_mgr.lock); + mst_mgr = aconnector->port->mgr; + ret = compute_mst_dsc_configs_for_link(state, dc_state, stream->link, vars, mst_mgr, + &link_vars_start_index); + if (ret != 0) + return ret; for (j = 0; j < dc_state->stream_count; j++) { if (dc_state->streams[j]->link == stream->link) @@ -1143,22 +1171,23 @@ bool compute_mst_dsc_configs_for_state(struct drm_atomic_state *state, if (stream->timing.flags.DSC == 1) if (dc_stream_add_dsc_to_resource(stream->ctx->dc, dc_state, stream) != DC_OK) - return false; + return -EINVAL; } - return true; + return ret; } -static bool - pre_compute_mst_dsc_configs_for_state(struct drm_atomic_state *state, - struct dc_state *dc_state, - struct dsc_mst_fairness_vars *vars) +static int pre_compute_mst_dsc_configs_for_state(struct drm_atomic_state *state, + struct dc_state *dc_state, + struct dsc_mst_fairness_vars *vars) { int i, j; struct dc_stream_state *stream; bool computed_streams[MAX_PIPES]; struct amdgpu_dm_connector *aconnector; + struct drm_dp_mst_topology_mgr *mst_mgr; int link_vars_start_index = 0; + int ret; for (i = 0; i < dc_state->stream_count; i++) computed_streams[i] = false; @@ -1171,7 +1200,7 @@ static bool aconnector = (struct amdgpu_dm_connector *)stream->dm_stream_context; - if (!aconnector || !aconnector->dc_sink) + if (!aconnector || !aconnector->dc_sink || !aconnector->port) continue; if (!aconnector->dc_sink->dsc_caps.dsc_dec_caps.is_dsc_supported) @@ -1183,14 +1212,11 @@ static bool if (!is_dsc_need_re_compute(state, dc_state, stream->link)) continue; - mutex_lock(&aconnector->mst_mgr.lock); - if (!compute_mst_dsc_configs_for_link(state, dc_state, stream->link, vars, - &aconnector->mst_mgr, - &link_vars_start_index)) { - mutex_unlock(&aconnector->mst_mgr.lock); - return false; - } - mutex_unlock(&aconnector->mst_mgr.lock); + mst_mgr = aconnector->port->mgr; + ret = compute_mst_dsc_configs_for_link(state, dc_state, stream->link, vars, mst_mgr, + &link_vars_start_index); + if (ret != 0) + return ret; for (j = 0; j < dc_state->stream_count; j++) { if (dc_state->streams[j]->link == stream->link) @@ -1198,7 +1224,7 @@ static bool } } - return true; + return ret; } static int find_crtc_index_in_state_by_stream(struct drm_atomic_state *state, @@ -1253,9 +1279,9 @@ static bool is_dsc_precompute_needed(struct drm_atomic_state *state) return ret; } -bool pre_validate_dsc(struct drm_atomic_state *state, - struct dm_atomic_state **dm_state_ptr, - struct dsc_mst_fairness_vars *vars) +int pre_validate_dsc(struct drm_atomic_state *state, + struct dm_atomic_state **dm_state_ptr, + struct dsc_mst_fairness_vars *vars) { int i; struct dm_atomic_state *dm_state; @@ -1264,11 +1290,12 @@ bool pre_validate_dsc(struct drm_atomic_state *state, if (!is_dsc_precompute_needed(state)) { DRM_INFO_ONCE("DSC precompute is not needed.\n"); - return true; + return 0; } - if (dm_atomic_get_state(state, dm_state_ptr)) { + ret = dm_atomic_get_state(state, dm_state_ptr); + if (ret != 0) { DRM_INFO_ONCE("dm_atomic_get_state() failed\n"); - return false; + return ret; } dm_state = *dm_state_ptr; @@ -1280,7 +1307,7 @@ bool pre_validate_dsc(struct drm_atomic_state *state, local_dc_state = kmemdup(dm_state->context, sizeof(struct dc_state), GFP_KERNEL); if (!local_dc_state) - return false; + return -ENOMEM; for (i = 0; i < local_dc_state->stream_count; i++) { struct dc_stream_state *stream = dm_state->context->streams[i]; @@ -1316,9 +1343,9 @@ bool pre_validate_dsc(struct drm_atomic_state *state, if (ret != 0) goto clean_exit; - if (!pre_compute_mst_dsc_configs_for_state(state, local_dc_state, vars)) { + ret = pre_compute_mst_dsc_configs_for_state(state, local_dc_state, vars); + if (ret != 0) { DRM_INFO_ONCE("pre_compute_mst_dsc_configs_for_state() failed\n"); - ret = -EINVAL; goto clean_exit; } @@ -1349,7 +1376,7 @@ clean_exit: kfree(local_dc_state); - return (ret == 0); + return ret; } static unsigned int kbps_from_pbn(unsigned int pbn) @@ -1392,6 +1419,7 @@ enum dc_status dm_dp_mst_is_port_support_mode( unsigned int upper_link_bw_in_kbps = 0, down_link_bw_in_kbps = 0; unsigned int max_compressed_bw_in_kbps = 0; struct dc_dsc_bw_range bw_range = {0}; + struct drm_dp_mst_topology_mgr *mst_mgr; /* * check if the mode could be supported if DSC pass-through is supported @@ -1400,7 +1428,8 @@ enum dc_status dm_dp_mst_is_port_support_mode( */ if (is_dsc_common_config_possible(stream, &bw_range) && aconnector->port->passthrough_aux) { - mutex_lock(&aconnector->mst_mgr.lock); + mst_mgr = aconnector->port->mgr; + mutex_lock(&mst_mgr->lock); cur_link_settings = stream->link->verified_link_cap; @@ -1413,7 +1442,7 @@ enum dc_status dm_dp_mst_is_port_support_mode( end_to_end_bw_in_kbps = min(upper_link_bw_in_kbps, down_link_bw_in_kbps); - mutex_unlock(&aconnector->mst_mgr.lock); + mutex_unlock(&mst_mgr->lock); /* * use the maximum dsc compression bandwidth as the required diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.h index b92a7c5671aa..97fd70df531b 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.h @@ -53,15 +53,15 @@ struct dsc_mst_fairness_vars { struct amdgpu_dm_connector *aconnector; }; -bool compute_mst_dsc_configs_for_state(struct drm_atomic_state *state, - struct dc_state *dc_state, - struct dsc_mst_fairness_vars *vars); +int compute_mst_dsc_configs_for_state(struct drm_atomic_state *state, + struct dc_state *dc_state, + struct dsc_mst_fairness_vars *vars); bool needs_dsc_aux_workaround(struct dc_link *link); -bool pre_validate_dsc(struct drm_atomic_state *state, - struct dm_atomic_state **dm_state_ptr, - struct dsc_mst_fairness_vars *vars); +int pre_validate_dsc(struct drm_atomic_state *state, + struct dm_atomic_state **dm_state_ptr, + struct dsc_mst_fairness_vars *vars); enum dc_status dm_dp_mst_is_port_support_mode( struct amdgpu_dm_connector *aconnector, diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c index dfd3be49eac8..e6854f7270a6 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c @@ -1369,7 +1369,7 @@ static bool dm_plane_format_mod_supported(struct drm_plane *plane, { struct amdgpu_device *adev = drm_to_adev(plane->dev); const struct drm_format_info *info = drm_format_info(format); - struct hw_asic_id asic_id = adev->dm.dc->ctx->asic_id; + int i; enum dm_micro_swizzle microtile = modifier_gfx9_swizzle_mode(modifier) & 3; @@ -1386,49 +1386,13 @@ static bool dm_plane_format_mod_supported(struct drm_plane *plane, return true; } - /* check if swizzle mode is supported by this version of DCN */ - switch (asic_id.chip_family) { - case FAMILY_SI: - case FAMILY_CI: - case FAMILY_KV: - case FAMILY_CZ: - case FAMILY_VI: - /* asics before AI does not have modifier support */ - return false; - case FAMILY_AI: - case FAMILY_RV: - case FAMILY_NV: - case FAMILY_VGH: - case FAMILY_YELLOW_CARP: - case AMDGPU_FAMILY_GC_10_3_6: - case AMDGPU_FAMILY_GC_10_3_7: - switch (AMD_FMT_MOD_GET(TILE, modifier)) { - case AMD_FMT_MOD_TILE_GFX9_64K_R_X: - case AMD_FMT_MOD_TILE_GFX9_64K_D_X: - case AMD_FMT_MOD_TILE_GFX9_64K_S_X: - case AMD_FMT_MOD_TILE_GFX9_64K_D: - return true; - default: - return false; - } - break; - case AMDGPU_FAMILY_GC_11_0_0: - case AMDGPU_FAMILY_GC_11_0_1: - switch (AMD_FMT_MOD_GET(TILE, modifier)) { - case AMD_FMT_MOD_TILE_GFX11_256K_R_X: - case AMD_FMT_MOD_TILE_GFX9_64K_R_X: - case AMD_FMT_MOD_TILE_GFX9_64K_D_X: - case AMD_FMT_MOD_TILE_GFX9_64K_S_X: - case AMD_FMT_MOD_TILE_GFX9_64K_D: - return true; - default: - return false; - } - break; - default: - ASSERT(0); /* Unknown asic */ - break; + /* Check that the modifier is on the list of the plane's supported modifiers. */ + for (i = 0; i < plane->modifier_count; i++) { + if (modifier == plane->modifiers[i]) + break; } + if (i == plane->modifier_count) + return false; /* * For D swizzle the canonical modifier depends on the bpp, so check diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_trace.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_trace.h index d3bc9dc21771..0f580ea37576 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_trace.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_trace.h @@ -37,6 +37,7 @@ #include <drm/drm_framebuffer.h> #include <drm/drm_encoder.h> #include <drm/drm_atomic.h> +#include "dcn10/dcn10_optc.h" #include "dc/inc/core_types.h" @@ -662,6 +663,69 @@ TRACE_EVENT(dcn_fpu, ) ); +TRACE_EVENT(dcn_optc_lock_unlock_state, + TP_PROTO(const struct optc *optc_state, int instance, bool lock, const char *function, const int line), + TP_ARGS(optc_state, instance, lock, function, line), + + TP_STRUCT__entry( + __field(const char *, function) + __field(int, instance) + __field(bool, lock) + __field(int, line) + __field(int, opp_count) + __field(int, max_h_total) + __field(int, max_v_total) + __field(int, min_h_blank) + __field(int, min_h_sync_width) + __field(int, min_v_sync_width) + __field(int, min_v_blank) + __field(int, min_v_blank_interlace) + __field(int, vstartup_start) + __field(int, vupdate_offset) + __field(int, vupdate_width) + __field(int, vready_offset) + ), + TP_fast_assign( + __entry->function = function; + __entry->instance = instance; + __entry->lock = lock; + __entry->line = line; + __entry->opp_count = optc_state->opp_count; + __entry->max_h_total = optc_state->max_h_total; + __entry->max_v_total = optc_state->max_v_total; + __entry->min_h_blank = optc_state->min_h_blank; + __entry->min_h_sync_width = optc_state->min_h_sync_width; + __entry->min_v_sync_width = optc_state->min_v_sync_width; + __entry->min_v_blank = optc_state->min_v_blank; + __entry->min_v_blank_interlace = optc_state->min_v_blank_interlace; + __entry->vstartup_start = optc_state->vstartup_start; + __entry->vupdate_offset = optc_state->vupdate_offset; + __entry->vupdate_width = optc_state->vupdate_width; + __entry->vready_offset = optc_state->vupdate_offset; + ), + TP_printk("%s: %s()+%d: optc_instance=%d opp_count=%d max_h_total=%d max_v_total=%d " + "min_h_blank=%d min_h_sync_width=%d min_v_sync_width=%d min_v_blank=%d " + "min_v_blank_interlace=%d vstartup_start=%d vupdate_offset=%d vupdate_width=%d " + "vready_offset=%d", + __entry->lock ? "Lock" : "Unlock", + __entry->function, + __entry->line, + __entry->instance, + __entry->opp_count, + __entry->max_h_total, + __entry->max_v_total, + __entry->min_h_blank, + __entry->min_h_sync_width, + __entry->min_v_sync_width, + __entry->min_v_blank, + __entry->min_v_blank_interlace, + __entry->vstartup_start, + __entry->vupdate_offset, + __entry->vupdate_width, + __entry->vready_offset + ) +); + #endif /* _AMDGPU_DM_TRACE_H_ */ #undef TRACE_INCLUDE_PATH diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/dc_fpu.c b/drivers/gpu/drm/amd/display/amdgpu_dm/dc_fpu.c index ab0c6d191038..1743ca0a3641 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/dc_fpu.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/dc_fpu.c @@ -31,6 +31,8 @@ #elif defined(CONFIG_PPC64) #include <asm/switch_to.h> #include <asm/cputable.h> +#elif defined(CONFIG_ARM64) +#include <asm/neon.h> #endif /** @@ -99,6 +101,8 @@ void dc_fpu_begin(const char *function_name, const int line) preempt_disable(); enable_kernel_fp(); } +#elif defined(CONFIG_ARM64) + kernel_neon_begin(); #endif } @@ -136,6 +140,8 @@ void dc_fpu_end(const char *function_name, const int line) disable_kernel_fp(); preempt_enable(); } +#elif defined(CONFIG_ARM64) + kernel_neon_end(); #endif } diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c index 9b8ea6e9a2b9..a1a00f432168 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c @@ -138,7 +138,9 @@ static uint8_t get_number_of_objects(struct bios_parser *bp, uint32_t offset) uint32_t object_table_offset = bp->object_info_tbl_offset + offset; - table = GET_IMAGE(ATOM_OBJECT_TABLE, object_table_offset); + table = ((ATOM_OBJECT_TABLE *) bios_get_image(&bp->base, + object_table_offset, + struct_size(table, asObjects, 1))); if (!table) return 0; @@ -166,8 +168,9 @@ static struct graphics_object_id bios_parser_get_connector_id( uint32_t connector_table_offset = bp->object_info_tbl_offset + le16_to_cpu(bp->object_info_tbl.v1_1->usConnectorObjectTableOffset); - ATOM_OBJECT_TABLE *tbl = - GET_IMAGE(ATOM_OBJECT_TABLE, connector_table_offset); + ATOM_OBJECT_TABLE *tbl = ((ATOM_OBJECT_TABLE *) bios_get_image(&bp->base, + connector_table_offset, + struct_size(tbl, asObjects, 1))); if (!tbl) { dm_error("Can't get connector table from atom bios.\n"); @@ -662,8 +665,9 @@ static enum bp_result get_ss_info_v3_1( if (!DATA_TABLES(ASIC_InternalSS_Info)) return BP_RESULT_UNSUPPORTED; - ss_table_header_include = GET_IMAGE(ATOM_ASIC_INTERNAL_SS_INFO_V3, - DATA_TABLES(ASIC_InternalSS_Info)); + ss_table_header_include = ((ATOM_ASIC_INTERNAL_SS_INFO_V3 *) bios_get_image(&bp->base, + DATA_TABLES(ASIC_InternalSS_Info), + struct_size(ss_table_header_include, asSpreadSpectrum, 1))); table_size = (le16_to_cpu(ss_table_header_include->sHeader.usStructureSize) - sizeof(ATOM_COMMON_TABLE_HEADER)) @@ -1029,8 +1033,10 @@ static enum bp_result get_ss_info_from_internal_ss_info_tbl_V2_1( if (!DATA_TABLES(ASIC_InternalSS_Info)) return result; - header = GET_IMAGE(ATOM_ASIC_INTERNAL_SS_INFO_V2, - DATA_TABLES(ASIC_InternalSS_Info)); + header = ((ATOM_ASIC_INTERNAL_SS_INFO_V2 *) bios_get_image( + &bp->base, + DATA_TABLES(ASIC_InternalSS_Info), + struct_size(header, asSpreadSpectrum, 1))); memset(info, 0, sizeof(struct spread_spectrum_info)); @@ -1709,8 +1715,10 @@ static uint32_t get_ss_entry_number_from_internal_ss_info_tbl_v2_1( if (!DATA_TABLES(ASIC_InternalSS_Info)) return 0; - header_include = GET_IMAGE(ATOM_ASIC_INTERNAL_SS_INFO_V2, - DATA_TABLES(ASIC_InternalSS_Info)); + header_include = ((ATOM_ASIC_INTERNAL_SS_INFO_V2 *) bios_get_image( + &bp->base, + DATA_TABLES(ASIC_InternalSS_Info), + struct_size(header_include, asSpreadSpectrum, 1))); size = (le16_to_cpu(header_include->sHeader.usStructureSize) - sizeof(ATOM_COMMON_TABLE_HEADER)) @@ -1746,8 +1754,9 @@ static uint32_t get_ss_entry_number_from_internal_ss_info_tbl_V3_1( if (!DATA_TABLES(ASIC_InternalSS_Info)) return number; - header_include = GET_IMAGE(ATOM_ASIC_INTERNAL_SS_INFO_V3, - DATA_TABLES(ASIC_InternalSS_Info)); + header_include = ((ATOM_ASIC_INTERNAL_SS_INFO_V3 *) bios_get_image(&bp->base, + DATA_TABLES(ASIC_InternalSS_Info), + struct_size(header_include, asSpreadSpectrum, 1))); size = (le16_to_cpu(header_include->sHeader.usStructureSize) - sizeof(ATOM_COMMON_TABLE_HEADER)) / sizeof(ATOM_ASIC_SS_ASSIGNMENT_V3); @@ -1789,11 +1798,13 @@ static enum bp_result bios_parser_get_gpio_pin_info( if (!DATA_TABLES(GPIO_Pin_LUT)) return BP_RESULT_BADBIOSTABLE; - header = GET_IMAGE(ATOM_GPIO_PIN_LUT, DATA_TABLES(GPIO_Pin_LUT)); + header = ((ATOM_GPIO_PIN_LUT *) bios_get_image(&bp->base, + DATA_TABLES(GPIO_Pin_LUT), + struct_size(header, asGPIO_Pin, 1))); if (!header) return BP_RESULT_BADBIOSTABLE; - if (sizeof(ATOM_COMMON_TABLE_HEADER) + sizeof(ATOM_GPIO_PIN_LUT) + if (sizeof(ATOM_COMMON_TABLE_HEADER) + struct_size(header, asGPIO_Pin, 1) > le16_to_cpu(header->sHeader.usStructureSize)) return BP_RESULT_BADBIOSTABLE; @@ -1978,7 +1989,8 @@ static ATOM_OBJECT *get_bios_object(struct bios_parser *bp, offset += bp->object_info_tbl_offset; - tbl = GET_IMAGE(ATOM_OBJECT_TABLE, offset); + tbl = ((ATOM_OBJECT_TABLE *) bios_get_image(&bp->base, offset, + struct_size(tbl, asObjects, 1))); if (!tbl) return NULL; @@ -2600,8 +2612,7 @@ static enum bp_result update_slot_layout_info( for (;;) { - record_header = (ATOM_COMMON_RECORD_HEADER *) - GET_IMAGE(ATOM_COMMON_RECORD_HEADER, record_offset); + record_header = GET_IMAGE(ATOM_COMMON_RECORD_HEADER, record_offset); if (record_header == NULL) { result = BP_RESULT_BADBIOSTABLE; break; @@ -2615,7 +2626,7 @@ static enum bp_result update_slot_layout_info( if (record_header->ucRecordType == ATOM_BRACKET_LAYOUT_RECORD_TYPE && - sizeof(ATOM_BRACKET_LAYOUT_RECORD) + struct_size(record, asConnInfo, 1) <= record_header->ucRecordSize) { record = (ATOM_BRACKET_LAYOUT_RECORD *) (record_header); @@ -2709,8 +2720,9 @@ static enum bp_result get_bracket_layout_record( genericTableOffset = bp->object_info_tbl_offset + bp->object_info_tbl.v1_3->usMiscObjectTableOffset; - object_table = (ATOM_OBJECT_TABLE *) - GET_IMAGE(ATOM_OBJECT_TABLE, genericTableOffset); + object_table = ((ATOM_OBJECT_TABLE *) bios_get_image(&bp->base, + genericTableOffset, + struct_size(object_table, asObjects, 1))); if (!object_table) return BP_RESULT_FAILURE; diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c index ee0456b5e14e..074e70a5c458 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c @@ -462,6 +462,7 @@ static enum bp_result get_gpio_i2c_info( uint32_t count = 0; unsigned int table_index = 0; bool find_valid = false; + struct atom_gpio_pin_assignment *pin; if (!info) return BP_RESULT_BADINPUT; @@ -489,20 +490,17 @@ static enum bp_result get_gpio_i2c_info( - sizeof(struct atom_common_table_header)) / sizeof(struct atom_gpio_pin_assignment); + pin = (struct atom_gpio_pin_assignment *) header->gpio_pin; + for (table_index = 0; table_index < count; table_index++) { - if (((record->i2c_id & I2C_HW_CAP) == ( - header->gpio_pin[table_index].gpio_id & - I2C_HW_CAP)) && - ((record->i2c_id & I2C_HW_ENGINE_ID_MASK) == - (header->gpio_pin[table_index].gpio_id & - I2C_HW_ENGINE_ID_MASK)) && - ((record->i2c_id & I2C_HW_LANE_MUX) == - (header->gpio_pin[table_index].gpio_id & - I2C_HW_LANE_MUX))) { + if (((record->i2c_id & I2C_HW_CAP) == (pin->gpio_id & I2C_HW_CAP)) && + ((record->i2c_id & I2C_HW_ENGINE_ID_MASK) == (pin->gpio_id & I2C_HW_ENGINE_ID_MASK)) && + ((record->i2c_id & I2C_HW_LANE_MUX) == (pin->gpio_id & I2C_HW_LANE_MUX))) { /* still valid */ find_valid = true; break; } + pin = (struct atom_gpio_pin_assignment *)((uint8_t *)pin + sizeof(struct atom_gpio_pin_assignment)); } /* If we don't find the entry that we are looking for then @@ -2393,6 +2391,26 @@ static enum bp_result get_vram_info_v25( return result; } +static enum bp_result get_vram_info_v30( + struct bios_parser *bp, + struct dc_vram_info *info) +{ + struct atom_vram_info_header_v3_0 *info_v30; + enum bp_result result = BP_RESULT_OK; + + info_v30 = GET_IMAGE(struct atom_vram_info_header_v3_0, + DATA_TABLES(vram_info)); + + if (info_v30 == NULL) + return BP_RESULT_BADBIOSTABLE; + + info->num_chans = info_v30->channel_num; + info->dram_channel_width_bytes = (1 << info_v30->channel_width) / 8; + + return result; +} + + /* * get_integrated_info_v11 * @@ -3060,6 +3078,16 @@ static enum bp_result bios_parser_get_vram_info( } break; + case 3: + switch (revision.minor) { + case 0: + result = get_vram_info_v30(bp, info); + break; + default: + break; + } + break; + default: return result; } diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr_vbios_smu.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr_vbios_smu.h index 3e5df27aa96f..1ce19d875358 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr_vbios_smu.h +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn21/rn_clk_mgr_vbios_smu.h @@ -26,6 +26,8 @@ #ifndef DAL_DC_RN_CLK_MGR_VBIOS_SMU_H_ #define DAL_DC_RN_CLK_MGR_VBIOS_SMU_H_ +enum dcn_pwr_state; + int rn_vbios_smu_get_smu_version(struct clk_mgr_internal *clk_mgr); int rn_vbios_smu_set_dispclk(struct clk_mgr_internal *clk_mgr, int requested_dispclk_khz); int rn_vbios_smu_set_dprefclk(struct clk_mgr_internal *clk_mgr); @@ -33,7 +35,7 @@ int rn_vbios_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int reque int rn_vbios_smu_set_min_deep_sleep_dcfclk(struct clk_mgr_internal *clk_mgr, int requested_min_ds_dcfclk_khz); void rn_vbios_smu_set_phyclk(struct clk_mgr_internal *clk_mgr, int requested_phyclk_khz); int rn_vbios_smu_set_dppclk(struct clk_mgr_internal *clk_mgr, int requested_dpp_khz); -void rn_vbios_smu_set_dcn_low_power_state(struct clk_mgr_internal *clk_mgr, int display_count); +void rn_vbios_smu_set_dcn_low_power_state(struct clk_mgr_internal *clk_mgr, enum dcn_pwr_state); void rn_vbios_smu_enable_48mhz_tmdp_refclk_pwrdwn(struct clk_mgr_internal *clk_mgr, bool enable); void rn_vbios_smu_enable_pme_wa(struct clk_mgr_internal *clk_mgr); int rn_vbios_smu_is_periodic_retraining_disabled(struct clk_mgr_internal *clk_mgr); diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.c index c1eaf571407a..1c0569b1dc8f 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn31/dcn31_clk_mgr.c @@ -609,8 +609,10 @@ static void dcn31_clk_mgr_helper_populate_bw_params(struct clk_mgr_internal *clk } bw_params->vram_type = bios_info->memory_type; - bw_params->num_channels = bios_info->ma_channel_number; + bw_params->dram_channel_width_bytes = bios_info->memory_type == 0x22 ? 8 : 4; + //bw_params->dram_channel_width_bytes = dc->ctx->asic_id.vram_width; + bw_params->num_channels = bios_info->ma_channel_number ? bios_info->ma_channel_number : 4; for (i = 0; i < WM_SET_COUNT; i++) { bw_params->wm_table.entries[i].wm_inst = i; diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.c index 1131c6d73f6c..20a06c04e4a1 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.c @@ -363,32 +363,32 @@ static struct wm_table ddr5_wm_table = { .wm_inst = WM_A, .wm_type = WM_TYPE_PSTATE_CHG, .pstate_latency_us = 11.72, - .sr_exit_time_us = 9, - .sr_enter_plus_exit_time_us = 11, + .sr_exit_time_us = 12.5, + .sr_enter_plus_exit_time_us = 14.5, .valid = true, }, { .wm_inst = WM_B, .wm_type = WM_TYPE_PSTATE_CHG, .pstate_latency_us = 11.72, - .sr_exit_time_us = 9, - .sr_enter_plus_exit_time_us = 11, + .sr_exit_time_us = 12.5, + .sr_enter_plus_exit_time_us = 14.5, .valid = true, }, { .wm_inst = WM_C, .wm_type = WM_TYPE_PSTATE_CHG, .pstate_latency_us = 11.72, - .sr_exit_time_us = 9, - .sr_enter_plus_exit_time_us = 11, + .sr_exit_time_us = 12.5, + .sr_enter_plus_exit_time_us = 14.5, .valid = true, }, { .wm_inst = WM_D, .wm_type = WM_TYPE_PSTATE_CHG, .pstate_latency_us = 11.72, - .sr_exit_time_us = 9, - .sr_enter_plus_exit_time_us = 11, + .sr_exit_time_us = 12.5, + .sr_enter_plus_exit_time_us = 14.5, .valid = true, }, } @@ -400,32 +400,32 @@ static struct wm_table lpddr5_wm_table = { .wm_inst = WM_A, .wm_type = WM_TYPE_PSTATE_CHG, .pstate_latency_us = 11.65333, - .sr_exit_time_us = 11.5, - .sr_enter_plus_exit_time_us = 14.5, + .sr_exit_time_us = 16.5, + .sr_enter_plus_exit_time_us = 18.5, .valid = true, }, { .wm_inst = WM_B, .wm_type = WM_TYPE_PSTATE_CHG, .pstate_latency_us = 11.65333, - .sr_exit_time_us = 11.5, - .sr_enter_plus_exit_time_us = 14.5, + .sr_exit_time_us = 16.5, + .sr_enter_plus_exit_time_us = 18.5, .valid = true, }, { .wm_inst = WM_C, .wm_type = WM_TYPE_PSTATE_CHG, .pstate_latency_us = 11.65333, - .sr_exit_time_us = 11.5, - .sr_enter_plus_exit_time_us = 14.5, + .sr_exit_time_us = 16.5, + .sr_enter_plus_exit_time_us = 18.5, .valid = true, }, { .wm_inst = WM_D, .wm_type = WM_TYPE_PSTATE_CHG, .pstate_latency_us = 11.65333, - .sr_exit_time_us = 11.5, - .sr_enter_plus_exit_time_us = 14.5, + .sr_exit_time_us = 16.5, + .sr_enter_plus_exit_time_us = 18.5, .valid = true, }, } diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_smu.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_smu.c index ef0795b14a1f..2db595672a46 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_smu.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_smu.c @@ -123,9 +123,10 @@ static int dcn314_smu_send_msg_with_param(struct clk_mgr_internal *clk_mgr, uint32_t result; result = dcn314_smu_wait_for_response(clk_mgr, 10, 200000); - ASSERT(result == VBIOSSMC_Result_OK); - smu_print("SMU response after wait: %d\n", result); + if (result != VBIOSSMC_Result_OK) + smu_print("SMU Response was not OK. SMU response after wait received is: %d\n", + result); if (result == VBIOSSMC_Status_BUSY) return -1; @@ -216,6 +217,12 @@ int dcn314_smu_set_hard_min_dcfclk(struct clk_mgr_internal *clk_mgr, int request VBIOSSMC_MSG_SetHardMinDcfclkByFreq, khz_to_mhz_ceil(requested_dcfclk_khz)); +#ifdef DBG + smu_print("actual_dcfclk_set_mhz %d is set to : %d\n", + actual_dcfclk_set_mhz, + actual_dcfclk_set_mhz * 1000); +#endif + return actual_dcfclk_set_mhz * 1000; } diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c index 893991a0eb97..07edd9777edf 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c @@ -458,19 +458,6 @@ static void dcn315_get_dpm_table_from_smu(struct clk_mgr_internal *clk_mgr, dcn315_smu_transfer_dpm_table_smu_2_dram(clk_mgr); } -static uint32_t find_max_clk_value(const uint32_t clocks[], uint32_t num_clocks) -{ - uint32_t max = 0; - int i; - - for (i = 0; i < num_clocks; ++i) { - if (clocks[i] > max) - max = clocks[i]; - } - - return max; -} - static void dcn315_clk_mgr_helper_populate_bw_params( struct clk_mgr_internal *clk_mgr, struct integrated_info *bios_info, @@ -478,29 +465,21 @@ static void dcn315_clk_mgr_helper_populate_bw_params( { int i; struct clk_bw_params *bw_params = clk_mgr->base.bw_params; - uint32_t max_pstate = 0, max_fclk = 0, min_pstate = 0; + uint32_t max_pstate = clock_table->NumDfPstatesEnabled - 1; struct clk_limit_table_entry def_max = bw_params->clk_table.entries[bw_params->clk_table.num_entries - 1]; - /* Find highest fclk pstate */ - for (i = 0; i < clock_table->NumDfPstatesEnabled; i++) { - if (clock_table->DfPstateTable[i].FClk > max_fclk) { - max_fclk = clock_table->DfPstateTable[i].FClk; - max_pstate = i; - } - } - /* For 315 we want to base clock table on dcfclk, need at least one entry regardless of pmfw table */ for (i = 0; i < clock_table->NumDcfClkLevelsEnabled; i++) { int j; - uint32_t min_fclk = clock_table->DfPstateTable[0].FClk; - for (j = 1; j < clock_table->NumDfPstatesEnabled; j++) { - if (clock_table->DfPstateTable[j].Voltage <= clock_table->SocVoltage[i] - && clock_table->DfPstateTable[j].FClk < min_fclk) { - min_fclk = clock_table->DfPstateTable[j].FClk; - min_pstate = j; - } + /* DF table is sorted with clocks decreasing */ + for (j = clock_table->NumDfPstatesEnabled - 2; j >= 0; j--) { + if (clock_table->DfPstateTable[j].Voltage <= clock_table->SocVoltage[i]) + max_pstate = j; } + /* Max DCFCLK should match up with max pstate */ + if (i == clock_table->NumDcfClkLevelsEnabled - 1) + max_pstate = 0; /* First search defaults for the clocks we don't read using closest lower or equal default dcfclk */ for (j = bw_params->clk_table.num_entries - 1; j > 0; j--) @@ -511,9 +490,9 @@ static void dcn315_clk_mgr_helper_populate_bw_params( bw_params->clk_table.entries[i].dtbclk_mhz = bw_params->clk_table.entries[j].dtbclk_mhz; /* Now update clocks we do read */ - bw_params->clk_table.entries[i].fclk_mhz = min_fclk; - bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[min_pstate].MemClk; - bw_params->clk_table.entries[i].voltage = clock_table->DfPstateTable[min_pstate].Voltage; + bw_params->clk_table.entries[i].fclk_mhz = clock_table->DfPstateTable[max_pstate].FClk; + bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[max_pstate].MemClk; + bw_params->clk_table.entries[i].voltage = clock_table->SocVoltage[i]; bw_params->clk_table.entries[i].dcfclk_mhz = clock_table->DcfClocks[i]; bw_params->clk_table.entries[i].socclk_mhz = clock_table->SocClocks[i]; bw_params->clk_table.entries[i].dispclk_mhz = clock_table->DispClocks[i]; @@ -521,25 +500,16 @@ static void dcn315_clk_mgr_helper_populate_bw_params( bw_params->clk_table.entries[i].wck_ratio = 1; } - /* Make sure to include at least one entry and highest pstate */ - if (max_pstate != min_pstate || i == 0) { - bw_params->clk_table.entries[i].fclk_mhz = max_fclk; - bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[max_pstate].MemClk; - bw_params->clk_table.entries[i].voltage = clock_table->DfPstateTable[max_pstate].Voltage; - bw_params->clk_table.entries[i].dcfclk_mhz = find_max_clk_value(clock_table->DcfClocks, NUM_DCFCLK_DPM_LEVELS); + /* Make sure to include at least one entry */ + if (i == 0) { + bw_params->clk_table.entries[i].fclk_mhz = clock_table->DfPstateTable[0].FClk; + bw_params->clk_table.entries[i].memclk_mhz = clock_table->DfPstateTable[0].MemClk; + bw_params->clk_table.entries[i].voltage = clock_table->DfPstateTable[0].Voltage; + bw_params->clk_table.entries[i].dcfclk_mhz = clock_table->DcfClocks[0]; bw_params->clk_table.entries[i].wck_ratio = 1; i++; } - bw_params->clk_table.num_entries = i--; - - /* Make sure all highest clocks are included*/ - bw_params->clk_table.entries[i].socclk_mhz = find_max_clk_value(clock_table->SocClocks, NUM_SOCCLK_DPM_LEVELS); - bw_params->clk_table.entries[i].dispclk_mhz = find_max_clk_value(clock_table->DispClocks, NUM_DISPCLK_DPM_LEVELS); - bw_params->clk_table.entries[i].dppclk_mhz = find_max_clk_value(clock_table->DppClocks, NUM_DPPCLK_DPM_LEVELS); - ASSERT(clock_table->DcfClocks[i] == find_max_clk_value(clock_table->DcfClocks, NUM_DCFCLK_DPM_LEVELS)); - bw_params->clk_table.entries[i].phyclk_mhz = def_max.phyclk_mhz; - bw_params->clk_table.entries[i].phyclk_d18_mhz = def_max.phyclk_d18_mhz; - bw_params->clk_table.entries[i].dtbclk_mhz = def_max.dtbclk_mhz; + bw_params->clk_table.num_entries = i; /* Set any 0 clocks to max default setting. Not an issue for * power since we aren't doing switching in such case anyway @@ -565,6 +535,11 @@ static void dcn315_clk_mgr_helper_populate_bw_params( if (!bw_params->clk_table.entries[i].dtbclk_mhz) bw_params->clk_table.entries[i].dtbclk_mhz = def_max.dtbclk_mhz; } + + /* Make sure all highest default clocks are included*/ + ASSERT(bw_params->clk_table.entries[i-1].phyclk_mhz == def_max.phyclk_mhz); + ASSERT(bw_params->clk_table.entries[i-1].phyclk_d18_mhz == def_max.phyclk_d18_mhz); + ASSERT(bw_params->clk_table.entries[i-1].dtbclk_mhz == def_max.dtbclk_mhz); ASSERT(bw_params->clk_table.entries[i-1].dcfclk_mhz); bw_params->vram_type = bios_info->memory_type; bw_params->num_channels = bios_info->ma_channel_number; diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c index 187f5b27fdc8..3edc81e2d417 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c @@ -553,6 +553,7 @@ static void dcn316_clk_mgr_helper_populate_bw_params( bw_params->vram_type = bios_info->memory_type; bw_params->num_channels = bios_info->ma_channel_number; + bw_params->dram_channel_width_bytes = bios_info->memory_type == 0x22 ? 8 : 4; for (i = 0; i < WM_SET_COUNT; i++) { bw_params->wm_table.entries[i].wm_inst = i; diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c index 1c612ccf1944..6f77d8e538ab 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn32/dcn32_clk_mgr.c @@ -157,6 +157,7 @@ void dcn32_init_clocks(struct clk_mgr *clk_mgr_base) struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base); unsigned int num_levels; struct clk_limit_num_entries *num_entries_per_clk = &clk_mgr_base->bw_params->clk_table.num_entries_per_clk; + unsigned int i; memset(&(clk_mgr_base->clks), 0, sizeof(struct dc_clocks)); clk_mgr_base->clks.p_state_change_support = true; @@ -205,18 +206,17 @@ void dcn32_init_clocks(struct clk_mgr *clk_mgr_base) clk_mgr->dpm_present = true; if (clk_mgr_base->ctx->dc->debug.min_disp_clk_khz) { - unsigned int i; - for (i = 0; i < num_levels; i++) if (clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz < khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_disp_clk_khz)) clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz = khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_disp_clk_khz); } + for (i = 0; i < num_levels; i++) + if (clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz > 1950) + clk_mgr_base->bw_params->clk_table.entries[i].dispclk_mhz = 1950; if (clk_mgr_base->ctx->dc->debug.min_dpp_clk_khz) { - unsigned int i; - for (i = 0; i < num_levels; i++) if (clk_mgr_base->bw_params->clk_table.entries[i].dppclk_mhz < khz_to_mhz_ceil(clk_mgr_base->ctx->dc->debug.min_dpp_clk_khz)) @@ -669,6 +669,9 @@ static void dcn32_get_memclk_states_from_smu(struct clk_mgr *clk_mgr_base) &clk_mgr_base->bw_params->clk_table.entries[0].memclk_mhz, &num_entries_per_clk->num_memclk_levels); + /* memclk must have at least one level */ + num_entries_per_clk->num_memclk_levels = num_entries_per_clk->num_memclk_levels ? num_entries_per_clk->num_memclk_levels : 1; + dcn32_init_single_clock(clk_mgr, PPCLK_FCLK, &clk_mgr_base->bw_params->clk_table.entries[0].fclk_mhz, &num_entries_per_clk->num_fclk_levels); diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 997ab031f816..1c3de3a1671e 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -135,9 +135,7 @@ static const char DC_BUILD_ID[] = "production-build"; * one or two (in the pipe-split case). */ -/******************************************************************************* - * Private functions - ******************************************************************************/ +/* Private functions */ static inline void elevate_update_type(enum surface_update_type *original, enum surface_update_type new) { @@ -401,9 +399,6 @@ bool dc_stream_adjust_vmin_vmax(struct dc *dc, { int i; - if (memcmp(adjust, &stream->adjust, sizeof(struct dc_crtc_timing_adjust)) == 0) - return true; - stream->adjust.v_total_max = adjust->v_total_max; stream->adjust.v_total_mid = adjust->v_total_mid; stream->adjust.v_total_mid_frame_num = adjust->v_total_mid_frame_num; @@ -424,18 +419,14 @@ bool dc_stream_adjust_vmin_vmax(struct dc *dc, } /** - ***************************************************************************** - * Function: dc_stream_get_last_vrr_vtotal + * dc_stream_get_last_used_drr_vtotal - dc_stream_get_last_vrr_vtotal * - * @brief - * Looks up the pipe context of dc_stream_state and gets the - * last VTOTAL used by DRR (Dynamic Refresh Rate) + * @dc: [in] dc reference + * @stream: [in] Initial dc stream state + * @adjust: [in] Updated parameters for vertical_total_min and * - * @param [in] dc: dc reference - * @param [in] stream: Initial dc stream state - * @param [in] adjust: Updated parameters for vertical_total_min and - * vertical_total_max - ***************************************************************************** + * Looks up the pipe context of dc_stream_state and gets the last VTOTAL used + * by DRR (Dynamic Refresh Rate) */ bool dc_stream_get_last_used_drr_vtotal(struct dc *dc, struct dc_stream_state *stream, @@ -491,86 +482,79 @@ bool dc_stream_get_crtc_position(struct dc *dc, } #if defined(CONFIG_DRM_AMD_SECURE_DISPLAY) -bool dc_stream_forward_dmcu_crc_window(struct dc *dc, struct dc_stream_state *stream, - struct crc_params *crc_window) +static inline void +dc_stream_forward_dmub_crc_window(struct dc_dmub_srv *dmub_srv, + struct rect *rect, struct otg_phy_mux *mux_mapping, bool is_stop) { - int i; - struct dmcu *dmcu = dc->res_pool->dmcu; - struct pipe_ctx *pipe; - struct crc_region tmp_win, *crc_win; - struct otg_phy_mux mapping_tmp, *mux_mapping; - - /*crc window can't be null*/ - if (!crc_window) - return false; - - if ((dmcu != NULL && dmcu->funcs->is_dmcu_initialized(dmcu))) { - crc_win = &tmp_win; - mux_mapping = &mapping_tmp; - /*set crc window*/ - tmp_win.x_start = crc_window->windowa_x_start; - tmp_win.y_start = crc_window->windowa_y_start; - tmp_win.x_end = crc_window->windowa_x_end; - tmp_win.y_end = crc_window->windowa_y_end; - - for (i = 0; i < MAX_PIPES; i++) { - pipe = &dc->current_state->res_ctx.pipe_ctx[i]; - if (pipe->stream == stream && !pipe->top_pipe && !pipe->prev_odm_pipe) - break; - } - - /* Stream not found */ - if (i == MAX_PIPES) - return false; - + union dmub_rb_cmd cmd = {0}; - /*set mux routing info*/ - mapping_tmp.phy_output_num = stream->link->link_enc_hw_inst; - mapping_tmp.otg_output_num = pipe->stream_res.tg->inst; + cmd.secure_display.roi_info.phy_id = mux_mapping->phy_output_num; + cmd.secure_display.roi_info.otg_id = mux_mapping->otg_output_num; - dmcu->funcs->forward_crc_window(dmcu, crc_win, mux_mapping); + if (is_stop) { + cmd.secure_display.header.type = DMUB_CMD__SECURE_DISPLAY; + cmd.secure_display.header.sub_type = DMUB_CMD__SECURE_DISPLAY_CRC_STOP_UPDATE; } else { - DC_LOG_DC("dmcu is not initialized"); - return false; + cmd.secure_display.header.type = DMUB_CMD__SECURE_DISPLAY; + cmd.secure_display.header.sub_type = DMUB_CMD__SECURE_DISPLAY_CRC_WIN_NOTIFY; + cmd.secure_display.roi_info.x_start = rect->x; + cmd.secure_display.roi_info.y_start = rect->y; + cmd.secure_display.roi_info.x_end = rect->x + rect->width; + cmd.secure_display.roi_info.y_end = rect->y + rect->height; } - return true; + dc_dmub_srv_cmd_queue(dmub_srv, &cmd); + dc_dmub_srv_cmd_execute(dmub_srv); } -bool dc_stream_stop_dmcu_crc_win_update(struct dc *dc, struct dc_stream_state *stream) +static inline void +dc_stream_forward_dmcu_crc_window(struct dmcu *dmcu, + struct rect *rect, struct otg_phy_mux *mux_mapping, bool is_stop) { - int i; - struct dmcu *dmcu = dc->res_pool->dmcu; - struct pipe_ctx *pipe; - struct otg_phy_mux mapping_tmp, *mux_mapping; + if (is_stop) + dmcu->funcs->stop_crc_win_update(dmcu, mux_mapping); + else + dmcu->funcs->forward_crc_window(dmcu, rect, mux_mapping); +} - if ((dmcu != NULL && dmcu->funcs->is_dmcu_initialized(dmcu))) { - mux_mapping = &mapping_tmp; +bool +dc_stream_forward_crc_window(struct dc *dc, + struct rect *rect, struct dc_stream_state *stream, bool is_stop) +{ + struct dmcu *dmcu; + struct dc_dmub_srv *dmub_srv; + struct otg_phy_mux mux_mapping; + struct pipe_ctx *pipe; + int i; - for (i = 0; i < MAX_PIPES; i++) { - pipe = &dc->current_state->res_ctx.pipe_ctx[i]; - if (pipe->stream == stream && !pipe->top_pipe && !pipe->prev_odm_pipe) - break; - } + for (i = 0; i < MAX_PIPES; i++) { + pipe = &dc->current_state->res_ctx.pipe_ctx[i]; + if (pipe->stream == stream && !pipe->top_pipe && !pipe->prev_odm_pipe) + break; + } - /* Stream not found */ - if (i == MAX_PIPES) - return false; + /* Stream not found */ + if (i == MAX_PIPES) + return false; + mux_mapping.phy_output_num = stream->link->link_enc_hw_inst; + mux_mapping.otg_output_num = pipe->stream_res.tg->inst; - /*set mux routing info*/ - mapping_tmp.phy_output_num = stream->link->link_enc_hw_inst; - mapping_tmp.otg_output_num = pipe->stream_res.tg->inst; + dmcu = dc->res_pool->dmcu; + dmub_srv = dc->ctx->dmub_srv; - dmcu->funcs->stop_crc_win_update(dmcu, mux_mapping); - } else { - DC_LOG_DC("dmcu is not initialized"); + /* forward to dmub */ + if (dmub_srv) + dc_stream_forward_dmub_crc_window(dmub_srv, rect, &mux_mapping, is_stop); + /* forward to dmcu */ + else if (dmcu && dmcu->funcs->is_dmcu_initialized(dmcu)) + dc_stream_forward_dmcu_crc_window(dmcu, rect, &mux_mapping, is_stop); + else return false; - } return true; } -#endif +#endif /* CONFIG_DRM_AMD_SECURE_DISPLAY */ /** * dc_stream_configure_crc() - Configure CRC capture for the given stream. @@ -1070,6 +1054,8 @@ static void disable_dangling_plane(struct dc *dc, struct dc_state *context) int i, j; struct dc_state *dangling_context = dc_create_state(dc); struct dc_state *current_ctx; + struct pipe_ctx *pipe; + struct timing_generator *tg; if (dangling_context == NULL) return; @@ -1112,6 +1098,18 @@ static void disable_dangling_plane(struct dc *dc, struct dc_state *context) } if (should_disable && old_stream) { + pipe = &dc->current_state->res_ctx.pipe_ctx[i]; + tg = pipe->stream_res.tg; + /* When disabling plane for a phantom pipe, we must turn on the + * phantom OTG so the disable programming gets the double buffer + * update. Otherwise the pipe will be left in a partially disabled + * state that can result in underflow or hang when enabling it + * again for different use. + */ + if (old_stream->mall_stream_config.type == SUBVP_PHANTOM) { + if (tg->funcs->enable_crtc) + tg->funcs->enable_crtc(tg); + } dc_rem_all_planes_for_stream(dc, old_stream, dangling_context); disable_all_writeback_pipes_for_stream(dc, old_stream, dangling_context); @@ -1127,6 +1125,15 @@ static void disable_dangling_plane(struct dc *dc, struct dc_state *context) dc->hwss.interdependent_update_lock(dc, dc->current_state, false); dc->hwss.post_unlock_program_front_end(dc, dangling_context); } + /* We need to put the phantom OTG back into it's default (disabled) state or we + * can get corruption when transition from one SubVP config to a different one. + * The OTG is set to disable on falling edge of VUPDATE so the plane disable + * will still get it's double buffer update. + */ + if (old_stream->mall_stream_config.type == SUBVP_PHANTOM) { + if (tg->funcs->disable_phantom_crtc) + tg->funcs->disable_phantom_crtc(tg); + } } } @@ -1185,7 +1192,7 @@ static void disable_vbios_mode_if_required( if (pix_clk_100hz != requested_pix_clk_100hz) { core_link_disable_stream(pipe); - pipe->stream->dpms_off = false; + pipe->stream->dpms_off = true; } } } @@ -1219,9 +1226,7 @@ static void wait_for_no_pipes_pending(struct dc *dc, struct dc_state *context) PERF_TRACE(); } -/******************************************************************************* - * Public functions - ******************************************************************************/ +/* Public functions */ struct dc *dc_create(const struct dc_init_data *init_params) { @@ -1488,17 +1493,19 @@ static void program_timing_sync( } } -static bool context_changed( - struct dc *dc, - struct dc_state *context) +static bool streams_changed(struct dc *dc, + struct dc_stream_state *streams[], + uint8_t stream_count) { uint8_t i; - if (context->stream_count != dc->current_state->stream_count) + if (stream_count != dc->current_state->stream_count) return true; for (i = 0; i < dc->current_state->stream_count; i++) { - if (dc->current_state->streams[i] != context->streams[i]) + if (dc->current_state->streams[i] != streams[i]) + return true; + if (!streams[i]->link->link_state_valid) return true; } @@ -1722,8 +1729,13 @@ void dc_z10_save_init(struct dc *dc) dc->hwss.z10_save_init(dc); } -/* - * Applies given context to HW and copy it into current context. +/** + * dc_commit_state_no_check - Apply context to the hardware + * + * @dc: DC object with the current status to be updated + * @context: New state that will become the current status at the end of this function + * + * Applies given context to the hardware and copy it into current context. * It's up to the user to release the src context afterwards. */ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *context) @@ -1760,6 +1772,12 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c context->stream_count == 0) dc->hwss.prepare_bandwidth(dc, context); + /* When SubVP is active, all HW programming must be done while + * SubVP lock is acquired + */ + if (dc->hwss.subvp_pipe_control_lock) + dc->hwss.subvp_pipe_control_lock(dc, context, true, true, NULL, subvp_prev_use); + if (dc->debug.enable_double_buffered_dsc_pg_support) dc->hwss.update_dsc_pg(dc, context, false); @@ -1787,9 +1805,6 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c dc->hwss.wait_for_mpcc_disconnect(dc, dc->res_pool, pipe); } - if (dc->hwss.subvp_pipe_control_lock) - dc->hwss.subvp_pipe_control_lock(dc, context, true, true, NULL, subvp_prev_use); - result = dc->hwss.apply_ctx_to_hw(dc, context); if (result != DC_OK) { @@ -1888,12 +1903,108 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c return result; } +/** + * dc_commit_streams - Commit current stream state + * + * @dc: DC object with the commit state to be configured in the hardware + * @streams: Array with a list of stream state + * @stream_count: Total of streams + * + * Function responsible for commit streams change to the hardware. + * + * Return: + * Return DC_OK if everything work as expected, otherwise, return a dc_status + * code. + */ +enum dc_status dc_commit_streams(struct dc *dc, + struct dc_stream_state *streams[], + uint8_t stream_count) +{ + int i, j; + struct dc_state *context; + enum dc_status res = DC_OK; + struct dc_validation_set set[MAX_STREAMS] = {0}; + + if (dc->ctx->dce_environment == DCE_ENV_VIRTUAL_HW) + return res; + + if (!streams_changed(dc, streams, stream_count)) + return res; + + DC_LOG_DC("%s: %d streams\n", __func__, stream_count); + + for (i = 0; i < stream_count; i++) { + struct dc_stream_state *stream = streams[i]; + struct dc_stream_status *status = dc_stream_get_status(stream); + + dc_stream_log(dc, stream); + + set[i].stream = stream; + + if (status) { + set[i].plane_count = status->plane_count; + for (j = 0; j < status->plane_count; j++) + set[i].plane_states[j] = status->plane_states[j]; + } + } + + context = dc_create_state(dc); + if (!context) + goto context_alloc_fail; + + dc_resource_state_copy_construct_current(dc, context); + + res = dc_validate_with_context(dc, set, stream_count, context, false); + if (res != DC_OK) { + BREAK_TO_DEBUGGER(); + goto fail; + } + + res = dc_commit_state_no_check(dc, context); + + for (i = 0; i < stream_count; i++) { + for (j = 0; j < context->stream_count; j++) { + if (streams[i]->stream_id == context->streams[j]->stream_id) + streams[i]->out.otg_offset = context->stream_status[j].primary_otg_inst; + + if (dc_is_embedded_signal(streams[i]->signal)) { + struct dc_stream_status *status = dc_stream_get_status_from_state(context, streams[i]); + + if (dc->hwss.is_abm_supported) + status->is_abm_supported = dc->hwss.is_abm_supported(dc, context, streams[i]); + else + status->is_abm_supported = true; + } + } + } + +fail: + dc_release_state(context); + +context_alloc_fail: + + DC_LOG_DC("%s Finished.\n", __func__); + + return (res == DC_OK); +} + +/* TODO: When the transition to the new commit sequence is done, remove this + * function in favor of dc_commit_streams. */ bool dc_commit_state(struct dc *dc, struct dc_state *context) { enum dc_status result = DC_ERROR_UNEXPECTED; int i; - if (!context_changed(dc, context)) + /* TODO: Since change commit sequence can have a huge impact, + * we decided to only enable it for DCN3x. However, as soon as + * we get more confident about this change we'll need to enable + * the new sequence for all ASICs. */ + if (dc->ctx->dce_version >= DCN_VERSION_3_2) { + result = dc_commit_streams(dc, context->streams, context->stream_count); + return result == DC_OK; + } + + if (!streams_changed(dc, context->streams, context->stream_count)) return DC_OK; DC_LOG_DC("%s: %d streams\n", @@ -3297,22 +3408,6 @@ static void commit_planes_for_stream(struct dc *dc, dc->hwss.pipe_control_lock(dc, top_pipe_to_program, true); } - if (update_type != UPDATE_TYPE_FAST) { - for (i = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *new_pipe = &context->res_ctx.pipe_ctx[i]; - - if ((new_pipe->stream && new_pipe->stream->mall_stream_config.type == SUBVP_PHANTOM) || - subvp_prev_use) { - // If old context or new context has phantom pipes, apply - // the phantom timings now. We can't change the phantom - // pipe configuration safely without driver acquiring - // the DMCUB lock first. - dc->hwss.apply_ctx_to_hw(dc, context); - break; - } - } - } - dc_dmub_update_dirty_rect(dc, surface_count, stream, srf_updates, context); if (update_type != UPDATE_TYPE_FAST) { @@ -3370,6 +3465,24 @@ static void commit_planes_for_stream(struct dc *dc, return; } + if (update_type != UPDATE_TYPE_FAST) { + for (j = 0; j < dc->res_pool->pipe_count; j++) { + struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j]; + + if (dc->debug.visual_confirm == VISUAL_CONFIRM_SUBVP && + pipe_ctx->stream && pipe_ctx->plane_state) { + /* Only update visual confirm for SUBVP here. + * The bar appears on all pipes, so we need to update the bar on all displays, + * so the information doesn't get stale. + */ + struct mpcc_blnd_cfg blnd_cfg = { 0 }; + + dc->hwss.update_visual_confirm_color(dc, pipe_ctx, &blnd_cfg.black_color, + pipe_ctx->plane_res.hubp->inst); + } + } + } + if (!IS_DIAG_DC(dc->ctx->dce_environment)) { for (i = 0; i < surface_count; i++) { struct dc_plane_state *plane_state = srf_updates[i].surface; @@ -3487,7 +3600,6 @@ static void commit_planes_for_stream(struct dc *dc, dc->hwss.update_plane_addr(dc, pipe_ctx); } } - } if (should_lock_all_pipes && dc->hwss.interdependent_update_lock) { @@ -3524,6 +3636,44 @@ static void commit_planes_for_stream(struct dc *dc, top_pipe_to_program->stream_res.tg); } + /* For phantom pipe OTG enable, it has to be done after any previous pipe + * that was in use has already been programmed at gotten its double buffer + * update for "disable". + */ + if (update_type != UPDATE_TYPE_FAST) { + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; + struct pipe_ctx *old_pipe = &dc->current_state->res_ctx.pipe_ctx[i]; + + /* If an active, non-phantom pipe is being transitioned into a phantom + * pipe, wait for the double buffer update to complete first before we do + * ANY phantom pipe programming. + */ + if (pipe->stream && pipe->stream->mall_stream_config.type == SUBVP_PHANTOM && + old_pipe->stream && old_pipe->stream->mall_stream_config.type != SUBVP_PHANTOM) { + old_pipe->stream_res.tg->funcs->wait_for_state( + old_pipe->stream_res.tg, + CRTC_STATE_VBLANK); + old_pipe->stream_res.tg->funcs->wait_for_state( + old_pipe->stream_res.tg, + CRTC_STATE_VACTIVE); + } + } + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *new_pipe = &context->res_ctx.pipe_ctx[i]; + + if ((new_pipe->stream && new_pipe->stream->mall_stream_config.type == SUBVP_PHANTOM) || + subvp_prev_use) { + // If old context or new context has phantom pipes, apply + // the phantom timings now. We can't change the phantom + // pipe configuration safely without driver acquiring + // the DMCUB lock first. + dc->hwss.apply_ctx_to_hw(dc, context); + break; + } + } + } + if (update_type != UPDATE_TYPE_FAST) dc->hwss.post_unlock_program_front_end(dc, context); if (update_type != UPDATE_TYPE_FAST) @@ -3563,10 +3713,24 @@ static void commit_planes_for_stream(struct dc *dc, } } -/* Determines if the incoming context requires a applying transition state with unnecessary - * pipe splitting and ODM disabled, due to hardware limitations. In a case where - * the OPP associated with an MPCC might change due to plane additions, this function +/** + * could_mpcc_tree_change_for_active_pipes - Check if an OPP associated with MPCC might change + * + * @dc: Used to get the current state status + * @stream: Target stream, which we want to remove the attached planes + * @surface_count: Number of surface update + * @is_plane_addition: [in] Fill out with true if it is a plane addition case + * + * DCN32x and newer support a feature named Dynamic ODM which can conflict with + * the MPO if used simultaneously in some specific configurations (e.g., + * 4k@144). This function checks if the incoming context requires applying a + * transition state with unnecessary pipe splitting and ODM disabled to + * circumvent our hardware limitations to prevent this edge case. If the OPP + * associated with an MPCC might change due to plane additions, this function * returns true. + * + * Return: + * Return true if OPP and MPCC might change, otherwise, return false. */ static bool could_mpcc_tree_change_for_active_pipes(struct dc *dc, struct dc_stream_state *stream, @@ -3576,7 +3740,6 @@ static bool could_mpcc_tree_change_for_active_pipes(struct dc *dc, struct dc_stream_status *cur_stream_status = stream_get_status(dc->current_state, stream); bool force_minimal_pipe_splitting = false; - uint32_t i; *is_plane_addition = false; @@ -3608,39 +3771,41 @@ static bool could_mpcc_tree_change_for_active_pipes(struct dc *dc, } } - /* For SubVP pipe split case when adding MPO video - * we need to add a minimal transition. In this case - * there will be 2 streams (1 main stream, 1 phantom - * stream). + /* For SubVP when adding or removing planes we need to add a minimal transition + * (even when disabling all planes). Whenever disabling a phantom pipe, we + * must use the minimal transition path to disable the pipe correctly. */ - if (cur_stream_status && - dc->current_state->stream_count == 2 && - stream->mall_stream_config.type == SUBVP_MAIN) { - bool is_pipe_split = false; - - for (i = 0; i < dc->res_pool->pipe_count; i++) { - if (dc->current_state->res_ctx.pipe_ctx[i].stream == stream && - (dc->current_state->res_ctx.pipe_ctx[i].bottom_pipe || - dc->current_state->res_ctx.pipe_ctx[i].next_odm_pipe)) { - is_pipe_split = true; - break; - } - } - + if (cur_stream_status && stream->mall_stream_config.type == SUBVP_MAIN) { /* determine if minimal transition is required due to SubVP*/ - if (surface_count > 0 && is_pipe_split) { - if (cur_stream_status->plane_count > surface_count) { - force_minimal_pipe_splitting = true; - } else if (cur_stream_status->plane_count < surface_count) { - force_minimal_pipe_splitting = true; - *is_plane_addition = true; - } + if (cur_stream_status->plane_count > surface_count) { + force_minimal_pipe_splitting = true; + } else if (cur_stream_status->plane_count < surface_count) { + force_minimal_pipe_splitting = true; + *is_plane_addition = true; } } return force_minimal_pipe_splitting; } +/** + * commit_minimal_transition_state - Create a transition pipe split state + * + * @dc: Used to get the current state status + * @transition_base_context: New transition state + * + * In some specific configurations, such as pipe split on multi-display with + * MPO and/or Dynamic ODM, removing a plane may cause unsupported pipe + * programming when moving to new planes. To mitigate those types of problems, + * this function adds a transition state that minimizes pipe usage before + * programming the new configuration. When adding a new plane, the current + * state requires the least pipes, so it is applied without splitting. When + * removing a plane, the new state requires the least pipes, so it is applied + * without splitting. + * + * Return: + * Return false if something is wrong in the transition state. + */ static bool commit_minimal_transition_state(struct dc *dc, struct dc_state *transition_base_context) { @@ -3650,9 +3815,48 @@ static bool commit_minimal_transition_state(struct dc *dc, bool temp_subvp_policy; enum dc_status ret = DC_ERROR_UNEXPECTED; unsigned int i, j; + unsigned int pipe_in_use = 0; + bool subvp_in_use = false; if (!transition_context) return false; + /* Setup: + * Store the current ODM and MPC config in some temp variables to be + * restored after we commit the transition state. + */ + + /* check current pipes in use*/ + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe = &transition_base_context->res_ctx.pipe_ctx[i]; + + if (pipe->plane_state) + pipe_in_use++; + } + + /* If SubVP is enabled and we are adding or removing planes from any main subvp + * pipe, we must use the minimal transition. + */ + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe = &dc->current_state->res_ctx.pipe_ctx[i]; + + if (pipe->stream && pipe->stream->mall_stream_config.type == SUBVP_PHANTOM) { + subvp_in_use = true; + break; + } + } + + /* When the OS add a new surface if we have been used all of pipes with odm combine + * and mpc split feature, it need use commit_minimal_transition_state to transition safely. + * After OS exit MPO, it will back to use odm and mpc split with all of pipes, we need + * call it again. Otherwise return true to skip. + * + * Reduce the scenarios to use dc_commit_state_no_check in the stage of flip. Especially + * enter/exit MPO when DCN still have enough resources. + */ + if (pipe_in_use != dc->res_pool->pipe_count && !subvp_in_use) { + dc_release_state(transition_context); + return true; + } if (!dc->config.is_vmin_only_asic) { tmp_mpc_policy = dc->debug.pipe_split_policy; @@ -3667,7 +3871,7 @@ static bool commit_minimal_transition_state(struct dc *dc, dc_resource_state_copy_construct(transition_base_context, transition_context); - //commit minimal state + /* commit minimal state */ if (dc->res_pool->funcs->validate_bandwidth(dc, transition_context, false)) { for (i = 0; i < transition_context->stream_count; i++) { struct dc_stream_status *stream_status = &transition_context->stream_status[i]; @@ -3685,10 +3889,12 @@ static bool commit_minimal_transition_state(struct dc *dc, ret = dc_commit_state_no_check(dc, transition_context); } - /*always release as dc_commit_state_no_check retains in good case*/ + /* always release as dc_commit_state_no_check retains in good case */ dc_release_state(transition_context); - /*restore previous pipe split and odm policy*/ + /* TearDown: + * Restore original configuration for ODM and MPO. + */ if (!dc->config.is_vmin_only_asic) dc->debug.pipe_split_policy = tmp_mpc_policy; @@ -3696,12 +3902,12 @@ static bool commit_minimal_transition_state(struct dc *dc, dc->debug.force_disable_subvp = temp_subvp_policy; if (ret != DC_OK) { - /*this should never happen*/ + /* this should never happen */ BREAK_TO_DEBUGGER(); return false; } - /*force full surface update*/ + /* force full surface update */ for (i = 0; i < dc->current_state->stream_count; i++) { for (j = 0; j < dc->current_state->stream_status[i].plane_count; j++) { dc->current_state->stream_status[i].plane_states[j]->update_flags.raw = 0xFFFFFFFF; @@ -3806,6 +4012,18 @@ void dc_commit_updates_for_stream(struct dc *dc, struct dc_context *dc_ctx = dc->ctx; int i, j; + /* TODO: Since change commit sequence can have a huge impact, + * we decided to only enable it for DCN3x. However, as soon as + * we get more confident about this change we'll need to enable + * the new sequence for all ASICs. + */ + if (dc->ctx->dce_version >= DCN_VERSION_3_2) { + dc_update_planes_and_stream(dc, srf_updates, + surface_count, stream, + stream_update); + return; + } + stream_status = dc_stream_get_status(stream); context = dc->current_state; @@ -4387,21 +4605,17 @@ void dc_mclk_switch_using_fw_based_vblank_stretch_shut_down(struct dc *dc) dc->current_state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching_shut_down = true; } -/* - ***************************************************************************** - * Function: dc_is_dmub_outbox_supported - +/** + * dc_is_dmub_outbox_supported - Check if DMUB firmware support outbox notification * - * @brief - * Checks whether DMUB FW supports outbox notifications, if supported - * DM should register outbox interrupt prior to actually enabling interrupts - * via dc_enable_dmub_outbox + * @dc: [in] dc structure * - * @param - * [in] dc: dc structure + * Checks whether DMUB FW supports outbox notifications, if supported DM + * should register outbox interrupt prior to actually enabling interrupts + * via dc_enable_dmub_outbox * - * @return - * True if DMUB FW supports outbox notifications, False otherwise - ***************************************************************************** + * Return: + * True if DMUB FW supports outbox notifications, False otherwise */ bool dc_is_dmub_outbox_supported(struct dc *dc) { @@ -4419,21 +4633,17 @@ bool dc_is_dmub_outbox_supported(struct dc *dc) return dc->debug.enable_dmub_aux_for_legacy_ddc; } -/* - ***************************************************************************** - * Function: dc_enable_dmub_notifications +/** + * dc_enable_dmub_notifications - Check if dmub fw supports outbox * - * @brief - * Calls dc_is_dmub_outbox_supported to check if dmub fw supports outbox - * notifications. All DMs shall switch to dc_is_dmub_outbox_supported. - * This API shall be removed after switching. + * @dc: [in] dc structure * - * @param - * [in] dc: dc structure + * Calls dc_is_dmub_outbox_supported to check if dmub fw supports outbox + * notifications. All DMs shall switch to dc_is_dmub_outbox_supported. This + * API shall be removed after switching. * - * @return - * True if DMUB FW supports outbox notifications, False otherwise - ***************************************************************************** + * Return: + * True if DMUB FW supports outbox notifications, False otherwise */ bool dc_enable_dmub_notifications(struct dc *dc) { @@ -4441,18 +4651,11 @@ bool dc_enable_dmub_notifications(struct dc *dc) } /** - ***************************************************************************** - * Function: dc_enable_dmub_outbox + * dc_enable_dmub_outbox - Enables DMUB unsolicited notification * - * @brief - * Enables DMUB unsolicited notifications to x86 via outbox + * dc: [in] dc structure * - * @param - * [in] dc: dc structure - * - * @return - * None - ***************************************************************************** + * Enables DMUB unsolicited notifications to x86 via outbox. */ void dc_enable_dmub_outbox(struct dc *dc) { @@ -4553,21 +4756,17 @@ uint8_t get_link_index_from_dpia_port_index(const struct dc *dc, } /** - ***************************************************************************** - * Function: dc_process_dmub_set_config_async + * dc_process_dmub_set_config_async - Submits set_config command * - * @brief - * Submits set_config command to dmub via inbox message + * @dc: [in] dc structure + * @link_index: [in] link_index: link index + * @payload: [in] aux payload + * @notify: [out] set_config immediate reply * - * @param - * [in] dc: dc structure - * [in] link_index: link index - * [in] payload: aux payload - * [out] notify: set_config immediate reply + * Submits set_config command to dmub via inbox message. * - * @return - * True if successful, False if failure - ***************************************************************************** + * Return: + * True if successful, False if failure */ bool dc_process_dmub_set_config_async(struct dc *dc, uint32_t link_index, @@ -4603,21 +4802,17 @@ bool dc_process_dmub_set_config_async(struct dc *dc, } /** - ***************************************************************************** - * Function: dc_process_dmub_set_mst_slots + * dc_process_dmub_set_mst_slots - Submits MST solt allocation * - * @brief - * Submits mst slot allocation command to dmub via inbox message + * @dc: [in] dc structure + * @link_index: [in] link index + * @mst_alloc_slots: [in] mst slots to be allotted + * @mst_slots_in_use: [out] mst slots in use returned in failure case * - * @param - * [in] dc: dc structure - * [in] link_index: link index - * [in] mst_alloc_slots: mst slots to be allotted - * [out] mst_slots_in_use: mst slots in use returned in failure case + * Submits mst slot allocation command to dmub via inbox message * - * @return - * DC_OK if successful, DC_ERROR if failure - ***************************************************************************** + * Return: + * DC_OK if successful, DC_ERROR if failure */ enum dc_status dc_process_dmub_set_mst_slots(const struct dc *dc, uint32_t link_index, @@ -4657,19 +4852,12 @@ enum dc_status dc_process_dmub_set_mst_slots(const struct dc *dc, } /** - ***************************************************************************** - * Function: dc_process_dmub_dpia_hpd_int_enable + * dc_process_dmub_dpia_hpd_int_enable - Submits DPIA DPD interruption * - * @brief - * Submits dpia hpd int enable command to dmub via inbox message + * @dc [in]: dc structure + * @hpd_int_enable [in]: 1 for hpd int enable, 0 to disable * - * @param - * [in] dc: dc structure - * [in] hpd_int_enable: 1 for hpd int enable, 0 to disable - * - * @return - * None - ***************************************************************************** + * Submits dpia hpd int enable command to dmub via inbox message */ void dc_process_dmub_dpia_hpd_int_enable(const struct dc *dc, uint32_t hpd_int_enable) @@ -4698,16 +4886,13 @@ void dc_disable_accelerated_mode(struct dc *dc) /** - ***************************************************************************** - * dc_notify_vsync_int_state() - notifies vsync enable/disable state + * dc_notify_vsync_int_state - notifies vsync enable/disable state * @dc: dc structure - * @stream: stream where vsync int state changed - * @enable: whether vsync is enabled or disabled - * - * Called when vsync is enabled/disabled - * Will notify DMUB to start/stop ABM interrupts after steady state is reached + * @stream: stream where vsync int state changed + * @enable: whether vsync is enabled or disabled * - ***************************************************************************** + * Called when vsync is enabled/disabled Will notify DMUB to start/stop ABM + * interrupts after steady state is reached. */ void dc_notify_vsync_int_state(struct dc *dc, struct dc_stream_state *stream, bool enable) { @@ -4749,14 +4934,18 @@ void dc_notify_vsync_int_state(struct dc *dc, struct dc_stream_state *stream, bo if (pipe->stream_res.abm && pipe->stream_res.abm->funcs->set_abm_pause) pipe->stream_res.abm->funcs->set_abm_pause(pipe->stream_res.abm, !enable, i, pipe->stream_res.tg->inst); } -/* - * dc_extended_blank_supported: Decide whether extended blank is supported + +/** + * dc_extended_blank_supported 0 Decide whether extended blank is supported + * + * @dc: [in] Current DC state * - * Extended blank is a freesync optimization feature to be enabled in the future. - * During the extra vblank period gained from freesync, we have the ability to enter z9/z10. + * Extended blank is a freesync optimization feature to be enabled in the + * future. During the extra vblank period gained from freesync, we have the + * ability to enter z9/z10. * - * @param [in] dc: Current DC state - * @return: Indicate whether extended blank is supported (true or false) + * Return: + * Indicate whether extended blank is supported (true or false) */ bool dc_extended_blank_supported(struct dc *dc) { diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c index 7c2e3b8dc26a..471078fc3900 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c @@ -366,6 +366,7 @@ void get_hdr_visual_confirm_color( struct tg_color *color) { uint32_t color_value = MAX_TG_COLOR_VALUE; + bool is_sdr = false; /* Determine the overscan color based on the top-most (desktop) plane's context */ struct pipe_ctx *top_pipe_ctx = pipe_ctx; @@ -382,7 +383,8 @@ void get_hdr_visual_confirm_color( /* FreeSync 2 ARGB2101010 - set border color to pink */ color->color_r_cr = color_value; color->color_b_cb = color_value; - } + } else + is_sdr = true; break; case PIXEL_FORMAT_FP16: if (top_pipe_ctx->stream->out_transfer_func->tf == TRANSFER_FUNCTION_PQ) { @@ -391,14 +393,19 @@ void get_hdr_visual_confirm_color( } else if (top_pipe_ctx->stream->out_transfer_func->tf == TRANSFER_FUNCTION_GAMMA22) { /* FreeSync 2 HDR - set border color to green */ color->color_g_y = color_value; - } + } else + is_sdr = true; break; default: + is_sdr = true; + break; + } + + if (is_sdr) { /* SDR - set border color to Gray */ color->color_r_cr = color_value/2; color->color_b_cb = color_value/2; color->color_g_y = color_value/2; - break; } } diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link.c b/drivers/gpu/drm/amd/display/dc/core/dc_link.c index d7b1ace6328a..5304e9daf90a 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link.c @@ -4229,6 +4229,7 @@ static void fpga_dp_hpo_enable_link_and_stream(struct dc_state *state, struct pi link_hwss->ext.set_throttled_vcp_size(pipe_ctx, avg_time_slots_per_mtp); dc->hwss.unblank_stream(pipe_ctx, &stream->link->cur_link_settings); + dc->hwss.enable_audio_stream(pipe_ctx); } void core_link_enable_stream( @@ -4308,10 +4309,7 @@ void core_link_enable_stream( /* Still enable stream features & audio on seamless boot for DP external displays */ if (pipe_ctx->stream->signal == SIGNAL_TYPE_DISPLAY_PORT) { enable_stream_features(pipe_ctx); - if (pipe_ctx->stream_res.audio != NULL) { - pipe_ctx->stream_res.stream_enc->funcs->dp_audio_enable(pipe_ctx->stream_res.stream_enc); - dc->hwss.enable_audio_stream(pipe_ctx); - } + dc->hwss.enable_audio_stream(pipe_ctx); } #if defined(CONFIG_DRM_AMD_DC_HDCP) @@ -4665,6 +4663,10 @@ void dc_link_set_preferred_training_settings(struct dc *dc, link->preferred_link_setting.link_rate = LINK_RATE_UNKNOWN; } + if (link->connector_signal == SIGNAL_TYPE_DISPLAY_PORT && + link->type == dc_connection_mst_branch) + dm_helpers_dp_mst_update_branch_bandwidth(dc->ctx, link); + /* Retrain now, or wait until next stream update to apply */ if (skip_immediate_retrain == false) dc_link_set_preferred_link_settings(dc, &link->preferred_link_setting, link); diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_ddc.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_ddc.c index 651231387043..ce8d6a54ca54 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link_ddc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_ddc.c @@ -82,6 +82,7 @@ struct dp_hdmi_dongle_signature_data { #define HDMI_SCDC_STATUS_FLAGS 0x40 #define HDMI_SCDC_ERR_DETECT 0x50 #define HDMI_SCDC_TEST_CONFIG 0xC0 +#define HDMI_SCDC_DEVICE_ID 0xD3 union hdmi_scdc_update_read_data { uint8_t byte[2]; diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c index 1254d38f1778..dedd1246ce58 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_dp.c @@ -1912,7 +1912,7 @@ enum dc_status dpcd_configure_lttpr_mode(struct dc_link *link, struct link_train return status; } -static void dpcd_exit_training_mode(struct dc_link *link) +static void dpcd_exit_training_mode(struct dc_link *link, enum dp_link_encoding encoding) { uint8_t sink_status = 0; uint8_t i; @@ -1920,12 +1920,14 @@ static void dpcd_exit_training_mode(struct dc_link *link) /* clear training pattern set */ dpcd_set_training_pattern(link, DP_TRAINING_PATTERN_VIDEOIDLE); - /* poll for intra-hop disable */ - for (i = 0; i < 10; i++) { - if ((core_link_read_dpcd(link, DP_SINK_STATUS, &sink_status, 1) == DC_OK) && - (sink_status & DP_INTRA_HOP_AUX_REPLY_INDICATION) == 0) - break; - udelay(1000); + if (encoding == DP_128b_132b_ENCODING) { + /* poll for intra-hop disable */ + for (i = 0; i < 10; i++) { + if ((core_link_read_dpcd(link, DP_SINK_STATUS, &sink_status, 1) == DC_OK) && + (sink_status & DP_INTRA_HOP_AUX_REPLY_INDICATION) == 0) + break; + udelay(1000); + } } } @@ -2649,7 +2651,7 @@ enum link_training_result dc_link_dp_perform_link_training( <_settings); /* reset previous training states */ - dpcd_exit_training_mode(link); + dpcd_exit_training_mode(link, encoding); /* configure link prior to entering training mode */ dpcd_configure_lttpr_mode(link, <_settings); @@ -2670,7 +2672,7 @@ enum link_training_result dc_link_dp_perform_link_training( ASSERT(0); /* exit training mode */ - dpcd_exit_training_mode(link); + dpcd_exit_training_mode(link, encoding); /* switch to video idle */ if ((status == LINK_TRAINING_SUCCESS) || !skip_video_pattern) @@ -2771,8 +2773,11 @@ bool perform_link_training_with_retries( /* Update verified link settings to current one * Because DPIA LT might fallback to lower link setting. */ - link->verified_link_cap.link_rate = link->cur_link_settings.link_rate; - link->verified_link_cap.lane_count = link->cur_link_settings.lane_count; + if (stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST) { + link->verified_link_cap.link_rate = link->cur_link_settings.link_rate; + link->verified_link_cap.lane_count = link->cur_link_settings.lane_count; + dm_helpers_dp_mst_update_branch_bandwidth(link->ctx, link); + } } } else { status = dc_link_dp_perform_link_training(link, @@ -3020,7 +3025,7 @@ static enum dc_link_rate get_lttpr_max_link_rate(struct dc_link *link) static enum dc_link_rate get_cable_max_link_rate(struct dc_link *link) { - enum dc_link_rate cable_max_link_rate = LINK_RATE_HIGH3; + enum dc_link_rate cable_max_link_rate = LINK_RATE_UNKNOWN; if (link->dpcd_caps.cable_id.bits.UHBR10_20_CAPABILITY & DP_UHBR20) cable_max_link_rate = LINK_RATE_UHBR20; @@ -3083,15 +3088,29 @@ struct dc_link_settings dp_get_max_link_cap(struct dc_link *link) max_link_cap.link_spread = link->reported_link_cap.link_spread; - /* Lower link settings based on cable attributes */ + /* Lower link settings based on cable attributes + * Cable ID is a DP2 feature to identify max certified link rate that + * a cable can carry. The cable identification method requires both + * cable and display hardware support. Since the specs comes late, it is + * anticipated that the first round of DP2 cables and displays may not + * be fully compatible to reliably return cable ID data. Therefore the + * decision of our cable id policy is that if the cable can return non + * zero cable id data, we will take cable's link rate capability into + * account. However if we get zero data, the cable link rate capability + * is considered inconclusive. In this case, we will not take cable's + * capability into account to avoid of over limiting hardware capability + * from users. The max overall link rate capability is still determined + * after actual dp pre-training. Cable id is considered as an auxiliary + * method of determining max link bandwidth capability. + */ cable_max_link_rate = get_cable_max_link_rate(link); if (!link->dc->debug.ignore_cable_id && + cable_max_link_rate != LINK_RATE_UNKNOWN && cable_max_link_rate < max_link_cap.link_rate) max_link_cap.link_rate = cable_max_link_rate; - /* - * account for lttpr repeaters cap + /* account for lttpr repeaters cap * notes: repeaters do not snoop in the DPRX Capabilities addresses (3.6.3). */ if (dp_is_lttpr_present(link)) { @@ -4540,9 +4559,19 @@ void dc_link_dp_handle_link_loss(struct dc_link *link) for (i = 0; i < MAX_PIPES; i++) { pipe_ctx = &link->dc->current_state->res_ctx.pipe_ctx[i]; - if (pipe_ctx && pipe_ctx->stream && !pipe_ctx->stream->dpms_off && - pipe_ctx->stream->link == link && !pipe_ctx->prev_odm_pipe) + if (pipe_ctx && pipe_ctx->stream && !pipe_ctx->stream->dpms_off + && pipe_ctx->stream->link == link && !pipe_ctx->prev_odm_pipe) { + // Always use max settings here for DP 1.4a LL Compliance CTS + if (link->is_automated) { + pipe_ctx->link_config.dp_link_settings.lane_count = + link->verified_link_cap.lane_count; + pipe_ctx->link_config.dp_link_settings.link_rate = + link->verified_link_cap.link_rate; + pipe_ctx->link_config.dp_link_settings.link_spread = + link->verified_link_cap.link_spread; + } core_link_enable_stream(link->dc->current_state, pipe_ctx); + } } } @@ -4583,6 +4612,8 @@ bool dc_link_handle_hpd_rx_irq(struct dc_link *link, union hpd_irq_data *out_hpd } if (hpd_irq_dpcd_data.bytes.device_service_irq.bits.AUTOMATED_TEST) { + // Workaround for DP 1.4a LL Compliance CTS as USB4 has to share encoders unlike DP and USBC + link->is_automated = true; device_service_clear.bits.AUTOMATED_TEST = 1; core_link_write_dpcd( link, @@ -5031,7 +5062,7 @@ static bool dpcd_read_sink_ext_caps(struct dc_link *link) return true; } -bool dp_retrieve_lttpr_cap(struct dc_link *link) +enum dc_status dp_retrieve_lttpr_cap(struct dc_link *link) { uint8_t lttpr_dpcd_data[8]; enum dc_status status = DC_ERROR_UNEXPECTED; @@ -5099,7 +5130,7 @@ bool dp_retrieve_lttpr_cap(struct dc_link *link) CONN_DATA_DETECT(link, lttpr_dpcd_data, sizeof(lttpr_dpcd_data), "LTTPR Caps: "); DC_LOG_DC("is_lttpr_present = %d\n", is_lttpr_present); - return is_lttpr_present; + return status; } bool dp_is_lttpr_present(struct dc_link *link) @@ -5227,76 +5258,45 @@ static void retrieve_cable_id(struct dc_link *link) &link->dpcd_caps.cable_id, &usbc_cable_id); } -/* DPRX may take some time to respond to AUX messages after HPD asserted. - * If AUX read unsuccessful, try to wake unresponsive DPRX by toggling DPCD SET_POWER (0x600). - */ -static enum dc_status wa_try_to_wake_dprx(struct dc_link *link, uint64_t timeout_ms) +static enum dc_status wake_up_aux_channel(struct dc_link *link) { enum dc_status status = DC_ERROR_UNEXPECTED; - uint8_t dpcd_data = 0; - uint64_t start_ts = 0; - uint64_t current_ts = 0; - uint64_t time_taken_ms = 0; - enum dc_connection_type type = dc_connection_none; - bool lttpr_present; - bool vbios_lttpr_interop = link->dc->caps.vbios_lttpr_aware; + uint32_t aux_channel_retry_cnt = 0; + uint8_t dpcd_power_state = '\0'; - lttpr_present = dp_is_lttpr_present(link) || - (!vbios_lttpr_interop || !link->dc->caps.extended_aux_timeout_support); - DC_LOG_DC("lttpr_present = %d.\n", lttpr_present ? 1 : 0); + while (status != DC_OK && aux_channel_retry_cnt < 10) { + status = core_link_read_dpcd(link, DP_SET_POWER, + &dpcd_power_state, sizeof(dpcd_power_state)); - /* Issue an AUX read to test DPRX responsiveness. If LTTPR is supported the first read is expected to - * be to determine LTTPR capabilities. Otherwise trying to read power state should be an innocuous AUX read. - */ - if (lttpr_present) - status = core_link_read_dpcd( - link, - DP_LT_TUNABLE_PHY_REPEATER_FIELD_DATA_STRUCTURE_REV, - &dpcd_data, - sizeof(dpcd_data)); - else - status = core_link_read_dpcd( - link, - DP_SET_POWER, - &dpcd_data, - sizeof(dpcd_data)); + /* Delay 1 ms if AUX CH is in power down state. Based on spec + * section 2.3.1.2, if AUX CH may be powered down due to + * write to DPCD 600h = 2. Sink AUX CH is monitoring differential + * signal and may need up to 1 ms before being able to reply. + */ + if (status != DC_OK || dpcd_power_state == DP_SET_POWER_D3) { + udelay(1000); + aux_channel_retry_cnt++; + } + } if (status != DC_OK) { - DC_LOG_WARNING("%s: Read DPCD LTTPR_CAP failed - try to toggle DPCD SET_POWER for %lld ms.", - __func__, - timeout_ms); - start_ts = dm_get_timestamp(link->ctx); - - do { - if (!dc_link_detect_sink(link, &type) || type == dc_connection_none) - break; - - dpcd_data = DP_SET_POWER_D3; - status = core_link_write_dpcd( - link, - DP_SET_POWER, - &dpcd_data, - sizeof(dpcd_data)); - - dpcd_data = DP_SET_POWER_D0; - status = core_link_write_dpcd( - link, - DP_SET_POWER, - &dpcd_data, - sizeof(dpcd_data)); - - current_ts = dm_get_timestamp(link->ctx); - time_taken_ms = div_u64(dm_get_elapse_time_in_ns(link->ctx, current_ts, start_ts), 1000000); - } while (status != DC_OK && time_taken_ms < timeout_ms); + dpcd_power_state = DP_SET_POWER_D0; + status = core_link_write_dpcd( + link, + DP_SET_POWER, + &dpcd_power_state, + sizeof(dpcd_power_state)); - DC_LOG_WARNING("%s: DPCD SET_POWER %s after %lld ms%s", - __func__, - (status == DC_OK) ? "succeeded" : "failed", - time_taken_ms, - (type == dc_connection_none) ? ". Unplugged." : "."); + dpcd_power_state = DP_SET_POWER_D3; + status = core_link_write_dpcd( + link, + DP_SET_POWER, + &dpcd_power_state, + sizeof(dpcd_power_state)); + return DC_ERROR_UNEXPECTED; } - return status; + return DC_OK; } static bool retrieve_link_cap(struct dc_link *link) @@ -5308,7 +5308,6 @@ static bool retrieve_link_cap(struct dc_link *link) /*Only need to read 1 byte starting from DP_DPRX_FEATURE_ENUMERATION_LIST. */ uint8_t dpcd_dprx_data = '\0'; - uint8_t dpcd_power_state = '\0'; struct dp_device_vendor_id sink_id; union down_stream_port_count down_strm_port_count; @@ -5316,11 +5315,9 @@ static bool retrieve_link_cap(struct dc_link *link) union dp_downstream_port_present ds_port = { 0 }; enum dc_status status = DC_ERROR_UNEXPECTED; uint32_t read_dpcd_retry_cnt = 3; - uint32_t aux_channel_retry_cnt = 0; int i; struct dp_sink_hw_fw_revision dp_hw_fw_revision; const uint32_t post_oui_delay = 30; // 30ms - bool is_lttpr_present = false; memset(dpcd_data, '\0', sizeof(dpcd_data)); memset(&down_strm_port_count, @@ -5335,51 +5332,17 @@ static bool retrieve_link_cap(struct dc_link *link) dc_link_aux_try_to_configure_timeout(link->ddc, LINK_AUX_DEFAULT_LTTPR_TIMEOUT_PERIOD); - /* Try to ensure AUX channel active before proceeding. */ - if (link->dc->debug.aux_wake_wa.bits.enable_wa) { - uint64_t timeout_ms = link->dc->debug.aux_wake_wa.bits.timeout_ms; + status = dp_retrieve_lttpr_cap(link); - if (link->dc->debug.aux_wake_wa.bits.use_default_timeout) - timeout_ms = LINK_AUX_WAKE_TIMEOUT_MS; - status = wa_try_to_wake_dprx(link, timeout_ms); - } - - while (status != DC_OK && aux_channel_retry_cnt < 10) { - status = core_link_read_dpcd(link, DP_SET_POWER, - &dpcd_power_state, sizeof(dpcd_power_state)); - - /* Delay 1 ms if AUX CH is in power down state. Based on spec - * section 2.3.1.2, if AUX CH may be powered down due to - * write to DPCD 600h = 2. Sink AUX CH is monitoring differential - * signal and may need up to 1 ms before being able to reply. - */ - if (status != DC_OK || dpcd_power_state == DP_SET_POWER_D3) { - udelay(1000); - aux_channel_retry_cnt++; - } - } - - /* If aux channel is not active, return false and trigger another detect*/ if (status != DC_OK) { - dpcd_power_state = DP_SET_POWER_D0; - status = core_link_write_dpcd( - link, - DP_SET_POWER, - &dpcd_power_state, - sizeof(dpcd_power_state)); - - dpcd_power_state = DP_SET_POWER_D3; - status = core_link_write_dpcd( - link, - DP_SET_POWER, - &dpcd_power_state, - sizeof(dpcd_power_state)); - return false; + status = wake_up_aux_channel(link); + if (status == DC_OK) + dp_retrieve_lttpr_cap(link); + else + return false; } - is_lttpr_present = dp_retrieve_lttpr_cap(link); - - if (is_lttpr_present) + if (dp_is_lttpr_present(link)) configure_lttpr_mode_transparent(link); /* Read DP tunneling information. */ @@ -5406,7 +5369,7 @@ static bool retrieve_link_cap(struct dc_link *link) return false; } - if (!is_lttpr_present) + if (!dp_is_lttpr_present(link)) dc_link_aux_try_to_configure_timeout(link->ddc, LINK_AUX_DEFAULT_TIMEOUT_PERIOD); { @@ -7294,6 +7257,7 @@ void dp_retrain_link_dp_test(struct dc_link *link, struct pipe_ctx *pipes = &link->dc->current_state->res_ctx.pipe_ctx[0]; unsigned int i; + bool do_fallback = false; for (i = 0; i < MAX_PIPES; i++) { @@ -7326,32 +7290,23 @@ void dp_retrain_link_dp_test(struct dc_link *link, memset(&link->cur_link_settings, 0, sizeof(link->cur_link_settings)); + if (link->ep_type == DISPLAY_ENDPOINT_USB4_DPIA) + do_fallback = true; + perform_link_training_with_retries( link_setting, skip_video_pattern, LINK_TRAINING_ATTEMPTS, &pipes[i], SIGNAL_TYPE_DISPLAY_PORT, - false); + do_fallback); link->dc->hwss.enable_stream(&pipes[i]); link->dc->hwss.unblank_stream(&pipes[i], link_setting); - if (pipes[i].stream_res.audio) { - /* notify audio driver for - * audio modes of monitor */ - pipes[i].stream_res.audio->funcs->az_enable( - pipes[i].stream_res.audio); - - /* un-mute audio */ - /* TODO: audio should be per stream rather than - * per link */ - pipes[i].stream_res.stream_enc->funcs-> - audio_mute_control( - pipes[i].stream_res.stream_enc, false); - } + link->dc->hwss.enable_audio_stream(&pipes[i]); } } } diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_dpia.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_dpia.c index 74e36b34d3f7..d130d58ac08e 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link_dpia.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_dpia.c @@ -791,10 +791,14 @@ static enum link_training_result dpia_training_eq_transparent( } if (dp_is_ch_eq_done(lane_count, dpcd_lane_status) && - dp_is_symbol_locked(link->cur_link_settings.lane_count, dpcd_lane_status) && - dp_is_interlane_aligned(dpcd_lane_status_updated)) { - result = LINK_TRAINING_SUCCESS; - break; + dp_is_symbol_locked(link->cur_link_settings.lane_count, dpcd_lane_status)) { + /* Take into consideration corner case for DP 1.4a LL Compliance CTS as USB4 + * has to share encoders unlike DP and USBC + */ + if (dp_is_interlane_aligned(dpcd_lane_status_updated) || (link->is_automated && retries_eq)) { + result = LINK_TRAINING_SUCCESS; + break; + } } /* Update VS/PE. */ @@ -1008,7 +1012,8 @@ enum link_training_result dc_link_dpia_perform_link_training( */ if (result == LINK_TRAINING_SUCCESS) { msleep(5); - result = dp_check_link_loss_status(link, <_settings); + if (!link->is_automated) + result = dp_check_link_loss_status(link, <_settings); } else if (result == LINK_TRAINING_ABORT) { dpia_training_abort(link, <_settings, repeater_id); } else { diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c index fd8db482e56f..da164685547d 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c @@ -1768,6 +1768,17 @@ bool dc_remove_plane_from_context( return true; } +/** + * dc_rem_all_planes_for_stream - Remove planes attached to the target stream. + * + * @dc: Current dc state. + * @stream: Target stream, which we want to remove the attached plans. + * @context: New context. + * + * Return: + * Return true if DC was able to remove all planes from the target + * stream, otherwise, return false. + */ bool dc_rem_all_planes_for_stream( const struct dc *dc, struct dc_stream_state *stream, @@ -2562,9 +2573,12 @@ enum dc_status resource_map_pool_resources( /** * dc_resource_state_copy_construct_current() - Creates a new dc_state from existing state - * Is a shallow copy. Increments refcounts on existing streams and planes. + * * @dc: copy out of dc->current_state * @dst_ctx: copy into this + * + * This function makes a shallow copy of the current DC state and increments + * refcounts on existing streams and planes. */ void dc_resource_state_copy_construct_current( const struct dc *dc, @@ -2593,15 +2607,241 @@ bool dc_resource_is_dsc_encoding_supported(const struct dc *dc) return dc->res_pool->res_cap->num_dsc > 0; } +static bool planes_changed_for_existing_stream(struct dc_state *context, + struct dc_stream_state *stream, + const struct dc_validation_set set[], + int set_count) +{ + int i, j; + struct dc_stream_status *stream_status = NULL; + + for (i = 0; i < context->stream_count; i++) { + if (context->streams[i] == stream) { + stream_status = &context->stream_status[i]; + break; + } + } + + if (!stream_status) + ASSERT(0); + + for (i = 0; i < set_count; i++) + if (set[i].stream == stream) + break; + + if (i == set_count) + ASSERT(0); + + if (set[i].plane_count != stream_status->plane_count) + return true; + + for (j = 0; j < set[i].plane_count; j++) + if (set[i].plane_states[j] != stream_status->plane_states[j]) + return true; + + return false; +} /** - * dc_validate_global_state() - Determine if HW can support a given state - * Checks HW resource availability and bandwidth requirement. + * dc_validate_with_context - Validate and update the potential new stream in the context object + * + * @dc: Used to get the current state status + * @set: An array of dc_validation_set with all the current streams reference + * @set_count: Total of streams + * @context: New context + * @fast_validate: Enable or disable fast validation + * + * This function updates the potential new stream in the context object. It + * creates multiple lists for the add, remove, and unchanged streams. In + * particular, if the unchanged streams have a plane that changed, it is + * necessary to remove all planes from the unchanged streams. In summary, this + * function is responsible for validating the new context. + * + * Return: + * In case of success, return DC_OK (1), otherwise, return a DC error. + */ +enum dc_status dc_validate_with_context(struct dc *dc, + const struct dc_validation_set set[], + int set_count, + struct dc_state *context, + bool fast_validate) +{ + struct dc_stream_state *unchanged_streams[MAX_PIPES] = { 0 }; + struct dc_stream_state *del_streams[MAX_PIPES] = { 0 }; + struct dc_stream_state *add_streams[MAX_PIPES] = { 0 }; + int old_stream_count = context->stream_count; + enum dc_status res = DC_ERROR_UNEXPECTED; + int unchanged_streams_count = 0; + int del_streams_count = 0; + int add_streams_count = 0; + bool found = false; + int i, j, k; + + DC_LOGGER_INIT(dc->ctx->logger); + + /* First build a list of streams to be remove from current context */ + for (i = 0; i < old_stream_count; i++) { + struct dc_stream_state *stream = context->streams[i]; + + for (j = 0; j < set_count; j++) { + if (stream == set[j].stream) { + found = true; + break; + } + } + + if (!found) + del_streams[del_streams_count++] = stream; + + found = false; + } + + /* Second, build a list of new streams */ + for (i = 0; i < set_count; i++) { + struct dc_stream_state *stream = set[i].stream; + + for (j = 0; j < old_stream_count; j++) { + if (stream == context->streams[j]) { + found = true; + break; + } + } + + if (!found) + add_streams[add_streams_count++] = stream; + + found = false; + } + + /* Build a list of unchanged streams which is necessary for handling + * planes change such as added, removed, and updated. + */ + for (i = 0; i < set_count; i++) { + /* Check if stream is part of the delete list */ + for (j = 0; j < del_streams_count; j++) { + if (set[i].stream == del_streams[j]) { + found = true; + break; + } + } + + if (!found) { + /* Check if stream is part of the add list */ + for (j = 0; j < add_streams_count; j++) { + if (set[i].stream == add_streams[j]) { + found = true; + break; + } + } + } + + if (!found) + unchanged_streams[unchanged_streams_count++] = set[i].stream; + + found = false; + } + + /* Remove all planes for unchanged streams if planes changed */ + for (i = 0; i < unchanged_streams_count; i++) { + if (planes_changed_for_existing_stream(context, + unchanged_streams[i], + set, + set_count)) { + if (!dc_rem_all_planes_for_stream(dc, + unchanged_streams[i], + context)) { + res = DC_FAIL_DETACH_SURFACES; + goto fail; + } + } + } + + /* Remove all planes for removed streams and then remove the streams */ + for (i = 0; i < del_streams_count; i++) { + /* Need to cpy the dwb data from the old stream in order to efc to work */ + if (del_streams[i]->num_wb_info > 0) { + for (j = 0; j < add_streams_count; j++) { + if (del_streams[i]->sink == add_streams[j]->sink) { + add_streams[j]->num_wb_info = del_streams[i]->num_wb_info; + for (k = 0; k < del_streams[i]->num_wb_info; k++) + add_streams[j]->writeback_info[k] = del_streams[i]->writeback_info[k]; + } + } + } + + if (!dc_rem_all_planes_for_stream(dc, del_streams[i], context)) { + res = DC_FAIL_DETACH_SURFACES; + goto fail; + } + + res = dc_remove_stream_from_ctx(dc, context, del_streams[i]); + if (res != DC_OK) + goto fail; + } + + /* Swap seamless boot stream to pipe 0 (if needed) to ensure pipe_ctx + * matches. This may change in the future if seamless_boot_stream can be + * multiple. + */ + for (i = 0; i < add_streams_count; i++) { + mark_seamless_boot_stream(dc, add_streams[i]); + if (add_streams[i]->apply_seamless_boot_optimization && i != 0) { + struct dc_stream_state *temp = add_streams[0]; + + add_streams[0] = add_streams[i]; + add_streams[i] = temp; + break; + } + } + + /* Add new streams and then add all planes for the new stream */ + for (i = 0; i < add_streams_count; i++) { + calculate_phy_pix_clks(add_streams[i]); + res = dc_add_stream_to_ctx(dc, context, add_streams[i]); + if (res != DC_OK) + goto fail; + + if (!add_all_planes_for_stream(dc, add_streams[i], set, set_count, context)) { + res = DC_FAIL_ATTACH_SURFACES; + goto fail; + } + } + + /* Add all planes for unchanged streams if planes changed */ + for (i = 0; i < unchanged_streams_count; i++) { + if (planes_changed_for_existing_stream(context, + unchanged_streams[i], + set, + set_count)) { + if (!add_all_planes_for_stream(dc, unchanged_streams[i], set, set_count, context)) { + res = DC_FAIL_ATTACH_SURFACES; + goto fail; + } + } + } + + res = dc_validate_global_state(dc, context, fast_validate); + +fail: + if (res != DC_OK) + DC_LOG_WARNING("%s:resource validation failed, dc_status:%d\n", + __func__, + res); + + return res; +} + +/** + * dc_validate_global_state() - Determine if hardware can support a given state + * * @dc: dc struct for this driver * @new_ctx: state to be validated * @fast_validate: set to true if only yes/no to support matters * - * Return: DC_OK if the result can be programmed. Otherwise, an error code. + * Checks hardware resource availability and bandwidth requirement. + * + * Return: + * DC_OK if the result can be programmed. Otherwise, an error code. */ enum dc_status dc_validate_global_state( struct dc *dc, @@ -2789,6 +3029,12 @@ static void set_avi_info_frame( hdmi_info.bits.C0_C1 = COLORIMETRY_EXTENDED; } + if (pixel_encoding && color_space == COLOR_SPACE_2020_YCBCR && + stream->out_transfer_func->tf == TRANSFER_FUNCTION_GAMMA22) { + hdmi_info.bits.EC0_EC2 = 0; + hdmi_info.bits.C0_C1 = COLORIMETRY_ITU709; + } + /* TODO: un-hardcode aspect ratio */ aspect = stream->timing.aspect_ratio; @@ -3734,4 +3980,4 @@ bool dc_resource_acquire_secondary_pipe_for_mpc_odm( } return true; -}
\ No newline at end of file +} diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index bfc5474c0f4c..57fc9193c770 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -47,7 +47,7 @@ struct aux_payload; struct set_config_cmd_payload; struct dmub_notification; -#define DC_VER "3.2.207" +#define DC_VER "3.2.212" #define MAX_SURFACES 3 #define MAX_PLANES 6 @@ -56,9 +56,7 @@ struct dmub_notification; #define MIN_VIEWPORT_SIZE 12 #define MAX_NUM_EDP 2 -/******************************************************************************* - * Display Core Interfaces - ******************************************************************************/ +/* Display Core Interfaces */ struct dc_versions { const char *dc_ver; struct dmcu_version dmcu_version; @@ -263,6 +261,7 @@ struct dc_caps { uint32_t cache_line_size; uint32_t cache_num_ways; uint16_t subvp_fw_processing_delay_us; + uint8_t subvp_drr_max_vblank_margin_us; uint16_t subvp_prefetch_end_to_mall_start_us; uint8_t subvp_swath_height_margin_lines; // subvp start line must be aligned to 2 x swath height uint16_t subvp_pstate_allow_width_us; @@ -395,6 +394,7 @@ struct dc_config { bool disable_dmcu; bool enable_4to1MPC; bool enable_windowed_mpo_odm; + bool forceHBR2CP2520; // Used for switching between test patterns TPS4 and CP2520 uint32_t allow_edp_hotplug_detection; bool clamp_min_dcfclk; uint64_t vblank_alignment_dto_params; @@ -408,6 +408,7 @@ struct dc_config { bool use_default_clock_table; bool force_bios_enable_lttpr; uint8_t force_bios_fixed_vs; + int sdpif_request_limit_words_per_umc; }; @@ -457,15 +458,15 @@ enum pipe_split_policy { MPC_SPLIT_DYNAMIC = 0, /** - * @MPC_SPLIT_DYNAMIC: Avoid pipe split, which means that DC will not + * @MPC_SPLIT_AVOID: Avoid pipe split, which means that DC will not * try any sort of split optimization. */ MPC_SPLIT_AVOID = 1, /** - * @MPC_SPLIT_DYNAMIC: With this option, DC will only try to optimize - * the pipe utilization when using a single display; if the user - * connects to a second display, DC will avoid pipe split. + * @MPC_SPLIT_AVOID_MULT_DISP: With this option, DC will only try to + * optimize the pipe utilization when using a single display; if the + * user connects to a second display, DC will avoid pipe split. */ MPC_SPLIT_AVOID_MULT_DISP = 2, }; @@ -494,9 +495,12 @@ enum dcn_zstate_support_state { DCN_ZSTATE_SUPPORT_ALLOW_Z10_ONLY, DCN_ZSTATE_SUPPORT_DISALLOW, }; -/* - * For any clocks that may differ per pipe - * only the max is stored in this structure + +/** + * struct dc_clocks - DC pipe clocks + * + * For any clocks that may differ per pipe only the max is stored in this + * structure */ struct dc_clocks { int dispclk_khz; @@ -523,6 +527,16 @@ struct dc_clocks { bool prev_p_state_change_support; bool fclk_prev_p_state_change_support; int num_ways; + + /* + * @fw_based_mclk_switching + * + * DC has a mechanism that leverage the variable refresh rate to switch + * memory clock in cases that we have a large latency to achieve the + * memory clock change and a short vblank window. DC has some + * requirements to enable this feature, and this field describes if the + * system support or not such a feature. + */ bool fw_based_mclk_switching; bool fw_based_mclk_switching_shut_down; int prev_num_ways; @@ -764,7 +778,6 @@ struct dc_debug_options { bool disable_mem_low_power; bool pstate_enabled; bool disable_dmcu; - bool disable_psr; bool force_abm_enable; bool disable_stereo_support; bool vsr_support; @@ -852,6 +865,8 @@ struct dc_debug_options { bool enable_double_buffered_dsc_pg_support; bool enable_dp_dig_pixel_rate_div_policy; enum lttpr_mode lttpr_mode_override; + unsigned int dsc_delay_factor_wa_x1000; + unsigned int min_prefetch_in_strobe_ns; }; struct gpu_info_soc_bounding_box_v1_0; @@ -988,9 +1003,7 @@ void dc_init_callbacks(struct dc *dc, void dc_deinit_callbacks(struct dc *dc); void dc_destroy(struct dc **dc); -/******************************************************************************* - * Surface Interfaces - ******************************************************************************/ +/* Surface Interfaces */ enum { TRANSFER_FUNC_POINTS = 1025 @@ -1269,12 +1282,23 @@ void dc_post_update_surfaces_to_stream( #include "dc_stream.h" -/* - * Structure to store surface/stream associations for validation +/** + * struct dc_validation_set - Struct to store surface/stream associations for validation */ struct dc_validation_set { + /** + * @stream: Stream state properties + */ struct dc_stream_state *stream; + + /** + * @plane_state: Surface state + */ struct dc_plane_state *plane_states[MAX_SURFACES]; + + /** + * @plane_count: Total of active planes + */ uint8_t plane_count; }; @@ -1286,6 +1310,12 @@ enum dc_status dc_validate_plane(struct dc *dc, const struct dc_plane_state *pla void get_clock_requirements_for_state(struct dc_state *state, struct AsicStateEx *info); +enum dc_status dc_validate_with_context(struct dc *dc, + const struct dc_validation_set set[], + int set_count, + struct dc_state *context, + bool fast_validate); + bool dc_set_generic_gpio_for_stereo(bool enable, struct gpio_service *gpio_service); @@ -1321,15 +1351,12 @@ void dc_resource_state_destruct(struct dc_state *context); bool dc_resource_is_dsc_encoding_supported(const struct dc *dc); -/* - * TODO update to make it about validation sets - * Set up streams and links associated to drive sinks - * The streams parameter is an absolute set of all active streams. - * - * After this call: - * Phy, Encoder, Timing Generator are programmed and enabled. - * New streams are enabled with blank stream; no memory read. - */ +enum dc_status dc_commit_streams(struct dc *dc, + struct dc_stream_state *streams[], + uint8_t stream_count); + +/* TODO: When the transition to the new commit sequence is done, remove this + * function in favor of dc_commit_streams. */ bool dc_commit_state(struct dc *dc, struct dc_state *context); struct dc_state *dc_create_state(struct dc *dc); @@ -1337,9 +1364,7 @@ struct dc_state *dc_copy_state(struct dc_state *src_ctx); void dc_retain_state(struct dc_state *context); void dc_release_state(struct dc_state *context); -/******************************************************************************* - * Link Interfaces - ******************************************************************************/ +/* Link Interfaces */ struct dpcd_caps { union dpcd_rev dpcd_rev; @@ -1441,9 +1466,7 @@ struct hdcp_caps { uint32_t dc_get_opp_for_plane(struct dc *dc, struct dc_plane_state *plane); -/******************************************************************************* - * Sink Interfaces - A sink corresponds to a display output device - ******************************************************************************/ +/* Sink Interfaces - A sink corresponds to a display output device */ struct dc_container_id { // 128bit GUID in binary form @@ -1526,9 +1549,7 @@ struct dc_cursor { }; -/******************************************************************************* - * Interrupt interfaces - ******************************************************************************/ +/* Interrupt interfaces */ enum dc_irq_source dc_interrupt_to_irq_source( struct dc *dc, uint32_t src_id, @@ -1540,9 +1561,7 @@ enum dc_irq_source dc_get_hpd_irq_source_at_index( void dc_notify_vsync_int_state(struct dc *dc, struct dc_stream_state *stream, bool enable); -/******************************************************************************* - * Power Interfaces - ******************************************************************************/ +/* Power Interfaces */ void dc_set_power_state( struct dc *dc, @@ -1615,14 +1634,10 @@ enum dc_status dc_process_dmub_set_mst_slots(const struct dc *dc, void dc_process_dmub_dpia_hpd_int_enable(const struct dc *dc, uint32_t hpd_int_enable); -/******************************************************************************* - * DSC Interfaces - ******************************************************************************/ +/* DSC Interfaces */ #include "dc_dsc.h" -/******************************************************************************* - * Disable acc mode Interfaces - ******************************************************************************/ +/* Disable acc mode Interfaces */ void dc_disable_accelerated_mode(struct dc *dc); #endif /* DC_INTERFACE_H_ */ diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c index 0541e87e4f38..097556f7b32c 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c @@ -423,25 +423,20 @@ void dc_dmub_srv_get_visual_confirm_color_cmd(struct dc *dc, struct pipe_ctx *pi #ifdef CONFIG_DRM_AMD_DC_DCN /** - * *********************************************************************************************** - * populate_subvp_cmd_drr_info: Helper to populate DRR pipe info for the DMCUB subvp command + * populate_subvp_cmd_drr_info - Helper to populate DRR pipe info for the DMCUB subvp command * - * Populate the DMCUB SubVP command with DRR pipe info. All the information required for calculating - * the SubVP + DRR microschedule is populated here. + * @dc: [in] current dc state + * @subvp_pipe: [in] pipe_ctx for the SubVP pipe + * @vblank_pipe: [in] pipe_ctx for the DRR pipe + * @pipe_data: [in] Pipe data which stores the VBLANK/DRR info + * + * Populate the DMCUB SubVP command with DRR pipe info. All the information + * required for calculating the SubVP + DRR microschedule is populated here. * * High level algorithm: * 1. Get timing for SubVP pipe, phantom pipe, and DRR pipe * 2. Calculate the min and max vtotal which supports SubVP + DRR microschedule * 3. Populate the drr_info with the min and max supported vtotal values - * - * @param [in] dc: current dc state - * @param [in] subvp_pipe: pipe_ctx for the SubVP pipe - * @param [in] vblank_pipe: pipe_ctx for the DRR pipe - * @param [in] pipe_data: Pipe data which stores the VBLANK/DRR info - * - * @return: void - * - * *********************************************************************************************** */ static void populate_subvp_cmd_drr_info(struct dc *dc, struct pipe_ctx *subvp_pipe, @@ -482,33 +477,37 @@ static void populate_subvp_cmd_drr_info(struct dc *dc, (((uint64_t)main_timing->pix_clk_100hz * 100))); drr_active_us = div64_u64(((uint64_t)drr_timing->v_addressable * drr_timing->h_total * 1000000), (((uint64_t)drr_timing->pix_clk_100hz * 100))); - max_drr_vblank_us = div64_u64((subvp_active_us - prefetch_us - drr_active_us), 2) + drr_active_us; - max_drr_mallregion_us = subvp_active_us - prefetch_us - mall_region_us; + max_drr_vblank_us = div64_u64((subvp_active_us - prefetch_us - + dc->caps.subvp_fw_processing_delay_us - drr_active_us), 2) + drr_active_us; + max_drr_mallregion_us = subvp_active_us - prefetch_us - mall_region_us - dc->caps.subvp_fw_processing_delay_us; max_drr_supported_us = max_drr_vblank_us > max_drr_mallregion_us ? max_drr_vblank_us : max_drr_mallregion_us; max_vtotal_supported = div64_u64(((uint64_t)drr_timing->pix_clk_100hz * 100 * max_drr_supported_us), (((uint64_t)drr_timing->h_total * 1000000))); + /* When calculating the max vtotal supported for SubVP + DRR cases, add + * margin due to possible rounding errors (being off by 1 line in the + * FW calculation can incorrectly push the P-State switch to wait 1 frame + * longer). + */ + max_vtotal_supported = max_vtotal_supported - dc->caps.subvp_drr_max_vblank_margin_us; + pipe_data->pipe_config.vblank_data.drr_info.min_vtotal_supported = min_vtotal_supported; pipe_data->pipe_config.vblank_data.drr_info.max_vtotal_supported = max_vtotal_supported; } /** - * *********************************************************************************************** - * populate_subvp_cmd_vblank_pipe_info: Helper to populate VBLANK pipe info for the DMUB subvp command - * - * Populate the DMCUB SubVP command with VBLANK pipe info. All the information required to calculate - * the microschedule for SubVP + VBLANK case is stored in the pipe_data (subvp_data and vblank_data). - * Also check if the VBLANK pipe is a DRR display -- if it is make a call to populate drr_info. - * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * @param [in] cmd: DMUB cmd to be populated with SubVP info - * @param [in] vblank_pipe: pipe_ctx for the VBLANK pipe - * @param [in] cmd_pipe_index: index for the pipe array in DMCUB SubVP cmd + * populate_subvp_cmd_vblank_pipe_info - Helper to populate VBLANK pipe info for the DMUB subvp command * - * @return: void + * @dc: [in] current dc state + * @context: [in] new dc state + * @cmd: [in] DMUB cmd to be populated with SubVP info + * @vblank_pipe: [in] pipe_ctx for the VBLANK pipe + * @cmd_pipe_index: [in] index for the pipe array in DMCUB SubVP cmd * - * *********************************************************************************************** + * Populate the DMCUB SubVP command with VBLANK pipe info. All the information + * required to calculate the microschedule for SubVP + VBLANK case is stored in + * the pipe_data (subvp_data and vblank_data). Also check if the VBLANK pipe + * is a DRR display -- if it is make a call to populate drr_info. */ static void populate_subvp_cmd_vblank_pipe_info(struct dc *dc, struct dc_state *context, @@ -551,22 +550,18 @@ static void populate_subvp_cmd_vblank_pipe_info(struct dc *dc, } /** - * *********************************************************************************************** - * update_subvp_prefetch_end_to_mall_start: Helper for SubVP + SubVP case + * update_subvp_prefetch_end_to_mall_start - Helper for SubVP + SubVP case * - * For SubVP + SubVP, we use a single vertical interrupt to start the microschedule for both - * SubVP pipes. In order for this to work correctly, the MALL REGION of both SubVP pipes must - * start at the same time. This function lengthens the prefetch end to mall start delay of the - * SubVP pipe that has the shorter prefetch so that both MALL REGION's will start at the same time. + * @dc: [in] current dc state + * @context: [in] new dc state + * @cmd: [in] DMUB cmd to be populated with SubVP info + * @subvp_pipes: [in] Array of SubVP pipes (should always be length 2) * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * @param [in] cmd: DMUB cmd to be populated with SubVP info - * @param [in] subvp_pipes: Array of SubVP pipes (should always be length 2) - * - * @return: void - * - * *********************************************************************************************** + * For SubVP + SubVP, we use a single vertical interrupt to start the + * microschedule for both SubVP pipes. In order for this to work correctly, the + * MALL REGION of both SubVP pipes must start at the same time. This function + * lengthens the prefetch end to mall start delay of the SubVP pipe that has + * the shorter prefetch so that both MALL REGION's will start at the same time. */ static void update_subvp_prefetch_end_to_mall_start(struct dc *dc, struct dc_state *context, @@ -608,22 +603,17 @@ static void update_subvp_prefetch_end_to_mall_start(struct dc *dc, } /** - * *************************************************************************************** - * setup_subvp_dmub_command: Helper to populate the SubVP pipe info for the DMUB subvp command - * - * Populate the DMCUB SubVP command with SubVP pipe info. All the information required to - * calculate the microschedule for the SubVP pipe is stored in the pipe_data of the DMCUB - * SubVP command. - * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * @param [in] cmd: DMUB cmd to be populated with SubVP info - * @param [in] subvp_pipe: pipe_ctx for the SubVP pipe - * @param [in] cmd_pipe_index: index for the pipe array in DMCUB SubVP cmd + * populate_subvp_cmd_pipe_info - Helper to populate the SubVP pipe info for the DMUB subvp command * - * @return: void + * @dc: [in] current dc state + * @context: [in] new dc state + * @cmd: [in] DMUB cmd to be populated with SubVP info + * @subvp_pipe: [in] pipe_ctx for the SubVP pipe + * @cmd_pipe_index: [in] index for the pipe array in DMCUB SubVP cmd * - * *************************************************************************************** + * Populate the DMCUB SubVP command with SubVP pipe info. All the information + * required to calculate the microschedule for the SubVP pipe is stored in the + * pipe_data of the DMCUB SubVP command. */ static void populate_subvp_cmd_pipe_info(struct dc *dc, struct dc_state *context, @@ -703,19 +693,14 @@ static void populate_subvp_cmd_pipe_info(struct dc *dc, } /** - * *************************************************************************************** - * dc_dmub_setup_subvp_dmub_command: Populate the DMCUB SubVP command + * dc_dmub_setup_subvp_dmub_command - Populate the DMCUB SubVP command * - * This function loops through each pipe and populates the DMUB - * SubVP CMD info based on the pipe (e.g. SubVP, VBLANK). + * @dc: [in] current dc state + * @context: [in] new dc state + * @cmd: [in] DMUB cmd to be populated with SubVP info * - * @param [in] dc: current dc state - * @param [in] context: new dc state - * @param [in] cmd: DMUB cmd to be populated with SubVP info - * - * @return: void - * - * *************************************************************************************** + * This function loops through each pipe and populates the DMUB SubVP CMD info + * based on the pipe (e.g. SubVP, VBLANK). */ void dc_dmub_setup_subvp_dmub_command(struct dc *dc, struct dc_state *context, @@ -882,11 +867,59 @@ void dc_dmub_srv_log_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv) diag_data.is_cw6_enabled); } +static bool dc_can_pipe_disable_cursor(struct pipe_ctx *pipe_ctx) +{ + struct pipe_ctx *test_pipe, *split_pipe; + const struct scaler_data *scl_data = &pipe_ctx->plane_res.scl_data; + struct rect r1 = scl_data->recout, r2, r2_half; + int r1_r = r1.x + r1.width, r1_b = r1.y + r1.height, r2_r, r2_b; + int cur_layer = pipe_ctx->plane_state->layer_index; + + /** + * Disable the cursor if there's another pipe above this with a + * plane that contains this pipe's viewport to prevent double cursor + * and incorrect scaling artifacts. + */ + for (test_pipe = pipe_ctx->top_pipe; test_pipe; + test_pipe = test_pipe->top_pipe) { + // Skip invisible layer and pipe-split plane on same layer + if (!test_pipe->plane_state->visible || test_pipe->plane_state->layer_index == cur_layer) + continue; + + r2 = test_pipe->plane_res.scl_data.recout; + r2_r = r2.x + r2.width; + r2_b = r2.y + r2.height; + split_pipe = test_pipe; + + /** + * There is another half plane on same layer because of + * pipe-split, merge together per same height. + */ + for (split_pipe = pipe_ctx->top_pipe; split_pipe; + split_pipe = split_pipe->top_pipe) + if (split_pipe->plane_state->layer_index == test_pipe->plane_state->layer_index) { + r2_half = split_pipe->plane_res.scl_data.recout; + r2.x = (r2_half.x < r2.x) ? r2_half.x : r2.x; + r2.width = r2.width + r2_half.width; + r2_r = r2.x + r2.width; + break; + } + + if (r1.x >= r2.x && r1.y >= r2.y && r1_r <= r2_r && r1_b <= r2_b) + return true; + } + + return false; +} + static bool dc_dmub_should_update_cursor_data(struct pipe_ctx *pipe_ctx) { if (pipe_ctx->plane_state != NULL) { if (pipe_ctx->plane_state->address.type == PLN_ADDR_TYPE_VIDEO_PROGRESSIVE) return false; + + if (dc_can_pipe_disable_cursor(pipe_ctx)) + return false; } if ((pipe_ctx->stream->link->psr_settings.psr_version == DC_PSR_VERSION_SU_1 || @@ -962,19 +995,14 @@ static void dc_build_cursor_attribute_update_payload1( } /** - * *************************************************************************************** - * dc_send_update_cursor_info_to_dmu: Populate the DMCUB Cursor update info command - * - * This function would store the cursor related information and pass it into dmub + * dc_send_update_cursor_info_to_dmu - Populate the DMCUB Cursor update info command * - * @param [in] pCtx: pipe context - * @param [in] pipe_idx: pipe index + * @pCtx: [in] pipe context + * @pipe_idx: [in] pipe index * - * @return: void - * - * *************************************************************************************** + * This function would store the cursor related information and pass it into + * dmub */ - void dc_send_update_cursor_info_to_dmu( struct pipe_ctx *pCtx, uint8_t pipe_idx) { diff --git a/drivers/gpu/drm/amd/display/dc/dc_link.h b/drivers/gpu/drm/amd/display/dc/dc_link.h index caf0c7af2d0b..dc6afe33bca2 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_link.h +++ b/drivers/gpu/drm/amd/display/dc/dc_link.h @@ -117,7 +117,7 @@ struct psr_settings { * Add a struct dc_panel_config under dc_link */ struct dc_panel_config { - // extra panel power sequence parameters + /* extra panel power sequence parameters */ struct pps { unsigned int extra_t3_ms; unsigned int extra_t7_ms; @@ -127,13 +127,21 @@ struct dc_panel_config { unsigned int extra_t12_ms; unsigned int extra_post_OUI_ms; } pps; - // ABM + /* PSR */ + struct psr { + bool disable_psr; + bool disallow_psrsu; + bool rc_disable; + bool rc_allow_static_screen; + bool rc_allow_fullscreen_VPB; + } psr; + /* ABM */ struct varib { unsigned int varibright_feature_enable; unsigned int def_varibright_level; unsigned int abm_config_setting; } varib; - // edp DSC + /* edp DSC */ struct dsc { bool disable_dsc_edp; unsigned int force_dsc_edp_policy; @@ -158,6 +166,14 @@ struct dc_link { enum dc_irq_source irq_source_hpd_rx;/* aka DP Short Pulse */ bool is_hpd_filter_disabled; bool dp_ss_off; + + /** + * @link_state_valid: + * + * If there is no link and local sink, this variable should be set to + * false. Otherwise, it should be set to true; usually, the function + * core_link_enable_stream sets this field to true. + */ bool link_state_valid; bool aux_access_disabled; bool sync_lt_in_progress; @@ -168,6 +184,7 @@ struct dc_link { bool is_dig_mapping_flexible; bool hpd_status; /* HPD status of link without physical HPD pin. */ bool is_hpd_pending; /* Indicates a new received hpd */ + bool is_automated; /* Indicates automated testing */ bool edp_sink_present; diff --git a/drivers/gpu/drm/amd/display/dc/dc_stream.h b/drivers/gpu/drm/amd/display/dc/dc_stream.h index 9e6025c98db9..f4dfd3a49b68 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_stream.h +++ b/drivers/gpu/drm/amd/display/dc/dc_stream.h @@ -41,6 +41,10 @@ struct timing_sync_info { struct dc_stream_status { int primary_otg_inst; int stream_enc_inst; + + /** + * @plane_count: Total of planes attached to a single stream + */ int plane_count; int audio_inst; struct timing_sync_info timing_sync_info; @@ -197,7 +201,18 @@ struct dc_stream_state { bool use_vsc_sdp_for_colorimetry; bool ignore_msa_timing_param; + /** + * @allow_freesync: + * + * It say if Freesync is enabled or not. + */ bool allow_freesync; + + /** + * @vrr_active_variable: + * + * It describes if VRR is in use. + */ bool vrr_active_variable; bool freesync_on_desktop; @@ -517,10 +532,10 @@ bool dc_stream_get_crtc_position(struct dc *dc, unsigned int *nom_v_pos); #if defined(CONFIG_DRM_AMD_SECURE_DISPLAY) -bool dc_stream_forward_dmcu_crc_window(struct dc *dc, struct dc_stream_state *stream, - struct crc_params *crc_window); -bool dc_stream_stop_dmcu_crc_win_update(struct dc *dc, - struct dc_stream_state *stream); +bool dc_stream_forward_crc_window(struct dc *dc, + struct rect *rect, + struct dc_stream_state *stream, + bool is_stop); #endif bool dc_stream_configure_crc(struct dc *dc, diff --git a/drivers/gpu/drm/amd/display/dc/dc_trace.h b/drivers/gpu/drm/amd/display/dc/dc_trace.h index c711797e5c9e..bbec308a3a5e 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_trace.h +++ b/drivers/gpu/drm/amd/display/dc/dc_trace.h @@ -40,3 +40,5 @@ #define TRACE_DCN_FPU(begin, function, line, ref_count) \ trace_dcn_fpu(begin, function, line, ref_count) +#define TRACE_OPTC_LOCK_UNLOCK_STATE(optc, inst, lock) \ + trace_dcn_optc_lock_unlock_state(optc, inst, lock, __func__, __LINE__) diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h index ad9041472cca..dc78e2404b48 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_types.h @@ -993,4 +993,11 @@ struct display_endpoint_id { enum display_endpoint_type ep_type; }; +#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY) +struct otg_phy_mux { + uint8_t phy_output_num; + uint8_t otg_output_num; +}; +#endif + #endif /* DC_TYPES_H_ */ diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c index fbb19e253f50..d3cc5ec46956 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c @@ -927,19 +927,20 @@ static bool dcn10_recv_edid_cea_ack(struct dmcu *dmcu, int *offset) #if defined(CONFIG_DRM_AMD_SECURE_DISPLAY) static void dcn10_forward_crc_window(struct dmcu *dmcu, - struct crc_region *crc_win, + struct rect *rect, struct otg_phy_mux *mux_mapping) { struct dce_dmcu *dmcu_dce = TO_DCE_DMCU(dmcu); unsigned int dmcu_max_retry_on_wait_reg_ready = 801; unsigned int dmcu_wait_reg_ready_interval = 100; unsigned int crc_start = 0, crc_end = 0, otg_phy_mux = 0; + int x_start, y_start, x_end, y_end; /* If microcontroller is not running, do nothing */ if (dmcu->dmcu_state != DMCU_RUNNING) return; - if (!crc_win) + if (!rect) return; /* waitDMCUReadyForCmd */ @@ -947,9 +948,14 @@ static void dcn10_forward_crc_window(struct dmcu *dmcu, dmcu_wait_reg_ready_interval, dmcu_max_retry_on_wait_reg_ready); + x_start = rect->x; + y_start = rect->y; + x_end = x_start + rect->width; + y_end = y_start + rect->height; + /* build up nitification data */ - crc_start = (((unsigned int) crc_win->x_start) << 16) | crc_win->y_start; - crc_end = (((unsigned int) crc_win->x_end) << 16) | crc_win->y_end; + crc_start = (((unsigned int) x_start) << 16) | y_start; + crc_end = (((unsigned int) x_end) << 16) | y_end; otg_phy_mux = (((unsigned int) mux_mapping->otg_output_num) << 16) | mux_mapping->phy_output_num; diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c index bec5e9f787fc..2d3201b77d6a 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c @@ -399,7 +399,11 @@ static bool dmub_psr_copy_settings(struct dmub_psr *dmub, link->psr_settings.force_ffu_mode = 0; copy_settings_data->force_ffu_mode = link->psr_settings.force_ffu_mode; - if (link->fec_state == dc_link_fec_enabled && + if (((link->dpcd_caps.fec_cap.bits.FEC_CAPABLE && + !link->dc->debug.disable_fec) && + (link->dpcd_caps.dsc_caps.dsc_basic_caps.fields.dsc_support.DSC_SUPPORT && + !link->panel_config.dsc.disable_dsc_edp && + link->dc->caps.edp_dsc_support)) && link->dpcd_caps.sink_dev_id == DP_DEVICE_ID_38EC11 && (!memcmp(link->dpcd_caps.sink_dev_id_str, DP_SINK_DEVICE_STR_ID_1, sizeof(DP_SINK_DEVICE_STR_ID_1)) || @@ -409,6 +413,11 @@ static bool dmub_psr_copy_settings(struct dmub_psr *dmub, else copy_settings_data->debug.bitfields.force_wakeup_by_tps3 = 0; + //WA for PSR1 on specific TCON, require frame delay for frame re-lock + copy_settings_data->relock_delay_frame_cnt = 0; + if (link->dpcd_caps.sink_dev_id == DP_BRANCH_DEVICE_ID_001CF8) + copy_settings_data->relock_delay_frame_cnt = 2; + dc_dmub_srv_cmd_queue(dc->dmub_srv, &cmd); dc_dmub_srv_cmd_execute(dc->dmub_srv); dc_dmub_srv_wait_idle(dc->dmub_srv); diff --git a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c index d260eaa1509e..c5380ce70653 100644 --- a/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dce110/dce110_hw_sequencer.c @@ -688,16 +688,6 @@ void dce110_enable_stream(struct pipe_ctx *pipe_ctx) early_control = lane_count; tg->funcs->set_early_control(tg, early_control); - - /* enable audio only within mode set */ - if (pipe_ctx->stream_res.audio != NULL) { - if (dc_is_dp_signal(pipe_ctx->stream->signal)) - pipe_ctx->stream_res.stream_enc->funcs->dp_audio_enable(pipe_ctx->stream_res.stream_enc); - } - - - - } static enum bp_result link_transmitter_control( @@ -1081,12 +1071,14 @@ void dce110_enable_audio_stream(struct pipe_ctx *pipe_ctx) struct dc *dc; struct clk_mgr *clk_mgr; unsigned int i, num_audio = 1; + const struct link_hwss *link_hwss; if (!pipe_ctx->stream) return; dc = pipe_ctx->stream->ctx->dc; clk_mgr = dc->clk_mgr; + link_hwss = get_link_hwss(pipe_ctx->stream->link, &pipe_ctx->link_res); if (pipe_ctx->stream_res.audio && pipe_ctx->stream_res.audio->enabled == true) return; @@ -1103,56 +1095,35 @@ void dce110_enable_audio_stream(struct pipe_ctx *pipe_ctx) if (num_audio >= 1 && clk_mgr->funcs->enable_pme_wa) /*this is the first audio. apply the PME w/a in order to wake AZ from D3*/ clk_mgr->funcs->enable_pme_wa(clk_mgr); - /* un-mute audio */ - /* TODO: audio should be per stream rather than per link */ - if (is_dp_128b_132b_signal(pipe_ctx)) - pipe_ctx->stream_res.hpo_dp_stream_enc->funcs->audio_mute_control( - pipe_ctx->stream_res.hpo_dp_stream_enc, false); - else - pipe_ctx->stream_res.stream_enc->funcs->audio_mute_control( - pipe_ctx->stream_res.stream_enc, false); + + link_hwss->enable_audio_packet(pipe_ctx); + if (pipe_ctx->stream_res.audio) pipe_ctx->stream_res.audio->enabled = true; } - - if (dc_is_dp_signal(pipe_ctx->stream->signal)) - dp_source_sequence_trace(pipe_ctx->stream->link, DPCD_SOURCE_SEQ_AFTER_ENABLE_AUDIO_STREAM); } void dce110_disable_audio_stream(struct pipe_ctx *pipe_ctx) { struct dc *dc; struct clk_mgr *clk_mgr; + const struct link_hwss *link_hwss; if (!pipe_ctx || !pipe_ctx->stream) return; dc = pipe_ctx->stream->ctx->dc; clk_mgr = dc->clk_mgr; + link_hwss = get_link_hwss(pipe_ctx->stream->link, &pipe_ctx->link_res); if (pipe_ctx->stream_res.audio && pipe_ctx->stream_res.audio->enabled == false) return; - if (is_dp_128b_132b_signal(pipe_ctx)) - pipe_ctx->stream_res.hpo_dp_stream_enc->funcs->audio_mute_control( - pipe_ctx->stream_res.hpo_dp_stream_enc, true); - else - pipe_ctx->stream_res.stream_enc->funcs->audio_mute_control( - pipe_ctx->stream_res.stream_enc, true); + link_hwss->disable_audio_packet(pipe_ctx); + if (pipe_ctx->stream_res.audio) { pipe_ctx->stream_res.audio->enabled = false; - if (dc_is_dp_signal(pipe_ctx->stream->signal)) - if (is_dp_128b_132b_signal(pipe_ctx)) - pipe_ctx->stream_res.hpo_dp_stream_enc->funcs->dp_audio_disable( - pipe_ctx->stream_res.hpo_dp_stream_enc); - else - pipe_ctx->stream_res.stream_enc->funcs->dp_audio_disable( - pipe_ctx->stream_res.stream_enc); - else - pipe_ctx->stream_res.stream_enc->funcs->hdmi_audio_disable( - pipe_ctx->stream_res.stream_enc); - if (clk_mgr->funcs->enable_pme_wa) /*this is the first audio. apply the PME w/a in order to wake AZ from D3*/ clk_mgr->funcs->enable_pme_wa(clk_mgr); @@ -1163,9 +1134,6 @@ void dce110_disable_audio_stream(struct pipe_ctx *pipe_ctx) * stream->stream_engine_id); */ } - - if (dc_is_dp_signal(pipe_ctx->stream->signal)) - dp_source_sequence_trace(pipe_ctx->stream->link, DPCD_SOURCE_SEQ_AFTER_DISABLE_AUDIO_STREAM); } void dce110_disable_stream(struct pipe_ctx *pipe_ctx) @@ -1487,6 +1455,9 @@ static enum dc_status apply_single_controller_ctx_to_hw( unsigned int event_triggers = 0; struct pipe_ctx *odm_pipe = pipe_ctx->next_odm_pipe; struct dce_hwseq *hws = dc->hwseq; + const struct link_hwss *link_hwss = get_link_hwss( + link, &pipe_ctx->link_res); + if (hws->funcs.disable_stream_gating) { hws->funcs.disable_stream_gating(dc, pipe_ctx); @@ -1497,23 +1468,8 @@ static enum dc_status apply_single_controller_ctx_to_hw( build_audio_output(context, pipe_ctx, &audio_output); - if (dc_is_dp_signal(pipe_ctx->stream->signal)) - if (is_dp_128b_132b_signal(pipe_ctx)) - pipe_ctx->stream_res.hpo_dp_stream_enc->funcs->dp_audio_setup( - pipe_ctx->stream_res.hpo_dp_stream_enc, - pipe_ctx->stream_res.audio->inst, - &pipe_ctx->stream->audio_info); - else - pipe_ctx->stream_res.stream_enc->funcs->dp_audio_setup( - pipe_ctx->stream_res.stream_enc, - pipe_ctx->stream_res.audio->inst, - &pipe_ctx->stream->audio_info); - else - pipe_ctx->stream_res.stream_enc->funcs->hdmi_audio_setup( - pipe_ctx->stream_res.stream_enc, - pipe_ctx->stream_res.audio->inst, - &pipe_ctx->stream->audio_info, - &audio_output.crtc_info); + link_hwss->setup_audio_output(pipe_ctx, &audio_output, + pipe_ctx->stream_res.audio->inst); pipe_ctx->stream_res.audio->funcs->az_configure( pipe_ctx->stream_res.audio, diff --git a/drivers/gpu/drm/amd/display/dc/dce60/dce60_resource.c b/drivers/gpu/drm/amd/display/dc/dce60/dce60_resource.c index fc6aa098bda0..8db9f7514466 100644 --- a/drivers/gpu/drm/amd/display/dc/dce60/dce60_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dce60/dce60_resource.c @@ -1128,6 +1128,7 @@ struct resource_pool *dce60_create_resource_pool( if (dce60_construct(num_virtual_links, dc, pool)) return &pool->base; + kfree(pool); BREAK_TO_DEBUGGER(); return NULL; } @@ -1325,6 +1326,7 @@ struct resource_pool *dce61_create_resource_pool( if (dce61_construct(num_virtual_links, dc, pool)) return &pool->base; + kfree(pool); BREAK_TO_DEBUGGER(); return NULL; } @@ -1518,6 +1520,7 @@ struct resource_pool *dce64_create_resource_pool( if (dce64_construct(num_virtual_links, dc, pool)) return &pool->base; + kfree(pool); BREAK_TO_DEBUGGER(); return NULL; } diff --git a/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c b/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c index b28025960050..5825e6f412bd 100644 --- a/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dce80/dce80_resource.c @@ -1137,6 +1137,7 @@ struct resource_pool *dce80_create_resource_pool( if (dce80_construct(num_virtual_links, dc, pool)) return &pool->base; + kfree(pool); BREAK_TO_DEBUGGER(); return NULL; } @@ -1336,6 +1337,7 @@ struct resource_pool *dce81_create_resource_pool( if (dce81_construct(num_virtual_links, dc, pool)) return &pool->base; + kfree(pool); BREAK_TO_DEBUGGER(); return NULL; } diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.h b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.h index e48fd044f572..ba1c0621f0f8 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.h +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hubbub.h @@ -171,6 +171,7 @@ struct dcn_hubbub_registers { uint32_t DCHUBBUB_ARB_FCLK_PSTATE_CHANGE_WATERMARK_B; uint32_t DCHUBBUB_ARB_FCLK_PSTATE_CHANGE_WATERMARK_C; uint32_t DCHUBBUB_ARB_FCLK_PSTATE_CHANGE_WATERMARK_D; + uint32_t SDPIF_REQUEST_RATE_LIMIT; }; #define HUBBUB_REG_FIELD_LIST_DCN32(type) \ @@ -360,7 +361,8 @@ struct dcn_hubbub_registers { type DCHUBBUB_ARB_ALLOW_SR_ENTER_WATERMARK_Z8_C;\ type DCHUBBUB_ARB_ALLOW_SR_EXIT_WATERMARK_Z8_C;\ type DCHUBBUB_ARB_ALLOW_SR_ENTER_WATERMARK_Z8_D;\ - type DCHUBBUB_ARB_ALLOW_SR_EXIT_WATERMARK_Z8_D + type DCHUBBUB_ARB_ALLOW_SR_EXIT_WATERMARK_Z8_D;\ + type SDPIF_REQUEST_RATE_LIMIT struct dcn_hubbub_shift { diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c index 11e4c4e46947..587b04b8712d 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c @@ -97,10 +97,12 @@ void dcn10_lock_all_pipes(struct dc *dc, bool lock) { struct pipe_ctx *pipe_ctx; + struct pipe_ctx *old_pipe_ctx; struct timing_generator *tg; int i; for (i = 0; i < dc->res_pool->pipe_count; i++) { + old_pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; pipe_ctx = &context->res_ctx.pipe_ctx[i]; tg = pipe_ctx->stream_res.tg; @@ -110,7 +112,7 @@ void dcn10_lock_all_pipes(struct dc *dc, */ if (pipe_ctx->top_pipe || !pipe_ctx->stream || - !pipe_ctx->plane_state || + (!pipe_ctx->plane_state && !old_pipe_ctx->plane_state) || !tg->funcs->is_tg_enabled(tg)) continue; diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c index 33d780218790..c9e53dc49c92 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c @@ -27,6 +27,7 @@ #include "reg_helper.h" #include "dcn10_optc.h" #include "dc.h" +#include "dc_trace.h" #define REG(reg)\ optc1->tg_regs->reg @@ -657,6 +658,8 @@ void optc1_lock(struct timing_generator *optc) REG_WAIT(OTG_MASTER_UPDATE_LOCK, UPDATE_LOCK_STATUS, 1, 1, 10); + + TRACE_OPTC_LOCK_UNLOCK_STATE(optc1, optc->inst, true); } void optc1_unlock(struct timing_generator *optc) @@ -665,6 +668,8 @@ void optc1_unlock(struct timing_generator *optc) REG_SET(OTG_MASTER_UPDATE_LOCK, 0, OTG_MASTER_UPDATE_LOCK, 0); + + TRACE_OPTC_LOCK_UNLOCK_STATE(optc1, optc->inst, false); } void optc1_get_position(struct timing_generator *optc, diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c index 56d30baf12df..6bfac8088ab0 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_resource.c @@ -1295,47 +1295,6 @@ static uint32_t read_pipe_fuses(struct dc_context *ctx) return value; } -/* - * Some architectures don't support soft-float (e.g. aarch64), on those - * this function has to be called with hardfloat enabled, make sure not - * to inline it so whatever fp stuff is done stays inside - */ -static noinline void dcn10_resource_construct_fp( - struct dc *dc) -{ - if (dc->ctx->dce_version == DCN_VERSION_1_01) { - struct dcn_soc_bounding_box *dcn_soc = dc->dcn_soc; - struct dcn_ip_params *dcn_ip = dc->dcn_ip; - struct display_mode_lib *dml = &dc->dml; - - dml->ip.max_num_dpp = 3; - /* TODO how to handle 23.84? */ - dcn_soc->dram_clock_change_latency = 23; - dcn_ip->max_num_dpp = 3; - } - if (ASICREV_IS_RV1_F0(dc->ctx->asic_id.hw_internal_rev)) { - dc->dcn_soc->urgent_latency = 3; - dc->debug.disable_dmcu = true; - dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 41.60f; - } - - - dc->dcn_soc->number_of_channels = dc->ctx->asic_id.vram_width / ddr4_dram_width; - ASSERT(dc->dcn_soc->number_of_channels < 3); - if (dc->dcn_soc->number_of_channels == 0)/*old sbios bug*/ - dc->dcn_soc->number_of_channels = 2; - - if (dc->dcn_soc->number_of_channels == 1) { - dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 19.2f; - dc->dcn_soc->fabric_and_dram_bandwidth_vnom0p8 = 17.066f; - dc->dcn_soc->fabric_and_dram_bandwidth_vmid0p72 = 14.933f; - dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65 = 12.8f; - if (ASICREV_IS_RV1_F0(dc->ctx->asic_id.hw_internal_rev)) { - dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 20.80f; - } - } -} - static bool verify_clock_values(struct dm_pp_clock_levels_with_voltage *clks) { int i; @@ -1510,8 +1469,9 @@ static bool dcn10_resource_construct( memcpy(dc->dcn_ip, &dcn10_ip_defaults, sizeof(dcn10_ip_defaults)); memcpy(dc->dcn_soc, &dcn10_soc_defaults, sizeof(dcn10_soc_defaults)); - /* Other architectures we build for build this with soft-float */ + DC_FP_START(); dcn10_resource_construct_fp(dc); + DC_FP_END(); if (!dc->config.is_vmin_only_asic) if (ASICREV_IS_RAVEN2(dc->ctx->asic_id.hw_internal_rev)) diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb.h index a85ed228dfc2..a9dd9ae23ec9 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb.h +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_dwb.h @@ -27,204 +27,177 @@ #define TO_DCN20_DWBC(dwbc_base) \ container_of(dwbc_base, struct dcn20_dwbc, base) -/* DCN */ -#define BASE_INNER(seg) \ - DCE_BASE__INST0_SEG ## seg - -#define BASE(seg) \ - BASE_INNER(seg) - -#define SR(reg_name)\ - .reg_name = BASE(mm ## reg_name ## _BASE_IDX) + \ - mm ## reg_name - -#define SRI(reg_name, block, id)\ - .reg_name = BASE(mm ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ - mm ## block ## id ## _ ## reg_name - -#define SRI2(reg_name, block, id)\ - .reg_name = BASE(mm ## reg_name ## _BASE_IDX) + \ - mm ## reg_name - -#define SRII(reg_name, block, id)\ - .reg_name[id] = BASE(mm ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ - mm ## block ## id ## _ ## reg_name - -#define SF(reg_name, field_name, post_fix)\ - .field_name = reg_name ## __ ## field_name ## post_fix - - #define DWBC_COMMON_REG_LIST_DCN2_0(inst) \ - SRI2(WB_ENABLE, CNV, inst),\ - SRI2(WB_EC_CONFIG, CNV, inst),\ - SRI2(CNV_MODE, CNV, inst),\ - SRI2(CNV_WINDOW_START, CNV, inst),\ - SRI2(CNV_WINDOW_SIZE, CNV, inst),\ - SRI2(CNV_UPDATE, CNV, inst),\ - SRI2(CNV_SOURCE_SIZE, CNV, inst),\ - SRI2(CNV_TEST_CNTL, CNV, inst),\ - SRI2(CNV_TEST_CRC_RED, CNV, inst),\ - SRI2(CNV_TEST_CRC_GREEN, CNV, inst),\ - SRI2(CNV_TEST_CRC_BLUE, CNV, inst),\ - SRI2(WBSCL_COEF_RAM_SELECT, WBSCL, inst),\ - SRI2(WBSCL_COEF_RAM_TAP_DATA, WBSCL, inst),\ - SRI2(WBSCL_MODE, WBSCL, inst),\ - SRI2(WBSCL_TAP_CONTROL, WBSCL, inst),\ - SRI2(WBSCL_DEST_SIZE, WBSCL, inst),\ - SRI2(WBSCL_HORZ_FILTER_SCALE_RATIO, WBSCL, inst),\ - SRI2(WBSCL_HORZ_FILTER_INIT_Y_RGB, WBSCL, inst),\ - SRI2(WBSCL_HORZ_FILTER_INIT_CBCR, WBSCL, inst),\ - SRI2(WBSCL_VERT_FILTER_SCALE_RATIO, WBSCL, inst),\ - SRI2(WBSCL_VERT_FILTER_INIT_Y_RGB, WBSCL, inst),\ - SRI2(WBSCL_VERT_FILTER_INIT_CBCR, WBSCL, inst),\ - SRI2(WBSCL_ROUND_OFFSET, WBSCL, inst),\ - SRI2(WBSCL_OVERFLOW_STATUS, WBSCL, inst),\ - SRI2(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL, inst),\ - SRI2(WBSCL_TEST_CNTL, WBSCL, inst),\ - SRI2(WBSCL_TEST_CRC_RED, WBSCL, inst),\ - SRI2(WBSCL_TEST_CRC_GREEN, WBSCL, inst),\ - SRI2(WBSCL_TEST_CRC_BLUE, WBSCL, inst),\ - SRI2(WBSCL_BACKPRESSURE_CNT_EN, WBSCL, inst),\ - SRI2(WB_MCIF_BACKPRESSURE_CNT, WBSCL, inst),\ - SRI2(WBSCL_CLAMP_Y_RGB, WBSCL, inst),\ - SRI2(WBSCL_CLAMP_CBCR, WBSCL, inst),\ - SRI2(WBSCL_OUTSIDE_PIX_STRATEGY, WBSCL, inst),\ - SRI2(WBSCL_OUTSIDE_PIX_STRATEGY_CBCR, WBSCL, inst),\ - SRI2(WBSCL_DEBUG, WBSCL, inst),\ - SRI2(WBSCL_TEST_DEBUG_INDEX, WBSCL, inst),\ - SRI2(WBSCL_TEST_DEBUG_DATA, WBSCL, inst),\ - SRI2(WB_DEBUG_CTRL, CNV, inst),\ - SRI2(WB_DBG_MODE, CNV, inst),\ - SRI2(WB_HW_DEBUG, CNV, inst),\ - SRI2(CNV_TEST_DEBUG_INDEX, CNV, inst),\ - SRI2(CNV_TEST_DEBUG_DATA, CNV, inst),\ - SRI2(WB_SOFT_RESET, CNV, inst),\ - SRI2(WB_WARM_UP_MODE_CTL1, CNV, inst),\ - SRI2(WB_WARM_UP_MODE_CTL2, CNV, inst) + SRI2_DWB(WB_ENABLE, CNV, inst),\ + SRI2_DWB(WB_EC_CONFIG, CNV, inst),\ + SRI2_DWB(CNV_MODE, CNV, inst),\ + SRI2_DWB(CNV_WINDOW_START, CNV, inst),\ + SRI2_DWB(CNV_WINDOW_SIZE, CNV, inst),\ + SRI2_DWB(CNV_UPDATE, CNV, inst),\ + SRI2_DWB(CNV_SOURCE_SIZE, CNV, inst),\ + SRI2_DWB(CNV_TEST_CNTL, CNV, inst),\ + SRI2_DWB(CNV_TEST_CRC_RED, CNV, inst),\ + SRI2_DWB(CNV_TEST_CRC_GREEN, CNV, inst),\ + SRI2_DWB(CNV_TEST_CRC_BLUE, CNV, inst),\ + SRI2_DWB(WBSCL_COEF_RAM_SELECT, WBSCL, inst),\ + SRI2_DWB(WBSCL_COEF_RAM_TAP_DATA, WBSCL, inst),\ + SRI2_DWB(WBSCL_MODE, WBSCL, inst),\ + SRI2_DWB(WBSCL_TAP_CONTROL, WBSCL, inst),\ + SRI2_DWB(WBSCL_DEST_SIZE, WBSCL, inst),\ + SRI2_DWB(WBSCL_HORZ_FILTER_SCALE_RATIO, WBSCL, inst),\ + SRI2_DWB(WBSCL_HORZ_FILTER_INIT_Y_RGB, WBSCL, inst),\ + SRI2_DWB(WBSCL_HORZ_FILTER_INIT_CBCR, WBSCL, inst),\ + SRI2_DWB(WBSCL_VERT_FILTER_SCALE_RATIO, WBSCL, inst),\ + SRI2_DWB(WBSCL_VERT_FILTER_INIT_Y_RGB, WBSCL, inst),\ + SRI2_DWB(WBSCL_VERT_FILTER_INIT_CBCR, WBSCL, inst),\ + SRI2_DWB(WBSCL_ROUND_OFFSET, WBSCL, inst),\ + SRI2_DWB(WBSCL_OVERFLOW_STATUS, WBSCL, inst),\ + SRI2_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL, inst),\ + SRI2_DWB(WBSCL_TEST_CNTL, WBSCL, inst),\ + SRI2_DWB(WBSCL_TEST_CRC_RED, WBSCL, inst),\ + SRI2_DWB(WBSCL_TEST_CRC_GREEN, WBSCL, inst),\ + SRI2_DWB(WBSCL_TEST_CRC_BLUE, WBSCL, inst),\ + SRI2_DWB(WBSCL_BACKPRESSURE_CNT_EN, WBSCL, inst),\ + SRI2_DWB(WB_MCIF_BACKPRESSURE_CNT, WBSCL, inst),\ + SRI2_DWB(WBSCL_CLAMP_Y_RGB, WBSCL, inst),\ + SRI2_DWB(WBSCL_CLAMP_CBCR, WBSCL, inst),\ + SRI2_DWB(WBSCL_OUTSIDE_PIX_STRATEGY, WBSCL, inst),\ + SRI2_DWB(WBSCL_OUTSIDE_PIX_STRATEGY_CBCR, WBSCL, inst),\ + SRI2_DWB(WBSCL_DEBUG, WBSCL, inst),\ + SRI2_DWB(WBSCL_TEST_DEBUG_INDEX, WBSCL, inst),\ + SRI2_DWB(WBSCL_TEST_DEBUG_DATA, WBSCL, inst),\ + SRI2_DWB(WB_DEBUG_CTRL, CNV, inst),\ + SRI2_DWB(WB_DBG_MODE, CNV, inst),\ + SRI2_DWB(WB_HW_DEBUG, CNV, inst),\ + SRI2_DWB(CNV_TEST_DEBUG_INDEX, CNV, inst),\ + SRI2_DWB(CNV_TEST_DEBUG_DATA, CNV, inst),\ + SRI2_DWB(WB_SOFT_RESET, CNV, inst),\ + SRI2_DWB(WB_WARM_UP_MODE_CTL1, CNV, inst),\ + SRI2_DWB(WB_WARM_UP_MODE_CTL2, CNV, inst) #define DWBC_COMMON_MASK_SH_LIST_DCN2_0(mask_sh) \ - SF(WB_ENABLE, WB_ENABLE, mask_sh),\ - SF(WB_EC_CONFIG, DISPCLK_R_WB_GATE_DIS, mask_sh),\ - SF(WB_EC_CONFIG, DISPCLK_G_WB_GATE_DIS, mask_sh),\ - SF(WB_EC_CONFIG, DISPCLK_G_WBSCL_GATE_DIS, mask_sh),\ - SF(WB_EC_CONFIG, WB_TEST_CLK_SEL, mask_sh),\ - SF(WB_EC_CONFIG, WB_LB_LS_DIS, mask_sh),\ - SF(WB_EC_CONFIG, WB_LB_SD_DIS, mask_sh),\ - SF(WB_EC_CONFIG, WB_LUT_LS_DIS, mask_sh),\ - SF(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_MODE_SEL, mask_sh),\ - SF(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_DIS, mask_sh),\ - SF(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_FORCE, mask_sh),\ - SF(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_STATE, mask_sh),\ - SF(WB_EC_CONFIG, WB_RAM_PW_SAVE_MODE, mask_sh),\ - SF(WB_EC_CONFIG, WBSCL_LUT_MEM_PWR_STATE, mask_sh),\ - SF(CNV_MODE, CNV_OUT_BPC, mask_sh),\ - SF(CNV_MODE, CNV_FRAME_CAPTURE_RATE, mask_sh),\ - SF(CNV_MODE, CNV_WINDOW_CROP_EN, mask_sh),\ - SF(CNV_MODE, CNV_STEREO_TYPE, mask_sh),\ - SF(CNV_MODE, CNV_INTERLACED_MODE, mask_sh),\ - SF(CNV_MODE, CNV_EYE_SELECTION, mask_sh),\ - SF(CNV_MODE, CNV_STEREO_POLARITY, mask_sh),\ - SF(CNV_MODE, CNV_INTERLACED_FIELD_ORDER, mask_sh),\ - SF(CNV_MODE, CNV_STEREO_SPLIT, mask_sh),\ - SF(CNV_MODE, CNV_NEW_CONTENT, mask_sh),\ - SF(CNV_MODE, CNV_FRAME_CAPTURE_EN_CURRENT, mask_sh),\ - SF(CNV_MODE, CNV_FRAME_CAPTURE_EN, mask_sh),\ - SF(CNV_WINDOW_START, CNV_WINDOW_START_X, mask_sh),\ - SF(CNV_WINDOW_START, CNV_WINDOW_START_Y, mask_sh),\ - SF(CNV_WINDOW_SIZE, CNV_WINDOW_WIDTH, mask_sh),\ - SF(CNV_WINDOW_SIZE, CNV_WINDOW_HEIGHT, mask_sh),\ - SF(CNV_UPDATE, CNV_UPDATE_PENDING, mask_sh),\ - SF(CNV_UPDATE, CNV_UPDATE_TAKEN, mask_sh),\ - SF(CNV_UPDATE, CNV_UPDATE_LOCK, mask_sh),\ - SF(CNV_SOURCE_SIZE, CNV_SOURCE_WIDTH, mask_sh),\ - SF(CNV_SOURCE_SIZE, CNV_SOURCE_HEIGHT, mask_sh),\ - SF(CNV_TEST_CNTL, CNV_TEST_CRC_EN, mask_sh),\ - SF(CNV_TEST_CNTL, CNV_TEST_CRC_CONT_EN, mask_sh),\ - SF(CNV_TEST_CRC_RED, CNV_TEST_CRC_RED_MASK, mask_sh),\ - SF(CNV_TEST_CRC_RED, CNV_TEST_CRC_SIG_RED, mask_sh),\ - SF(CNV_TEST_CRC_GREEN, CNV_TEST_CRC_GREEN_MASK, mask_sh),\ - SF(CNV_TEST_CRC_GREEN, CNV_TEST_CRC_SIG_GREEN, mask_sh),\ - SF(CNV_TEST_CRC_BLUE, CNV_TEST_CRC_BLUE_MASK, mask_sh),\ - SF(CNV_TEST_CRC_BLUE, CNV_TEST_CRC_SIG_BLUE, mask_sh),\ - SF(WB_DEBUG_CTRL, WB_DEBUG_EN, mask_sh),\ - SF(WB_DEBUG_CTRL, WB_DEBUG_SEL, mask_sh),\ - SF(WB_DBG_MODE, WB_DBG_MODE_EN, mask_sh),\ - SF(WB_DBG_MODE, WB_DBG_DIN_FMT, mask_sh),\ - SF(WB_DBG_MODE, WB_DBG_36MODE, mask_sh),\ - SF(WB_DBG_MODE, WB_DBG_CMAP, mask_sh),\ - SF(WB_DBG_MODE, WB_DBG_PXLRATE_ERROR, mask_sh),\ - SF(WB_DBG_MODE, WB_DBG_SOURCE_WIDTH, mask_sh),\ - SF(WB_HW_DEBUG, WB_HW_DEBUG, mask_sh),\ - SF(WB_SOFT_RESET, WB_SOFT_RESET, mask_sh),\ - SF(CNV_TEST_DEBUG_INDEX, CNV_TEST_DEBUG_INDEX, mask_sh),\ - SF(CNV_TEST_DEBUG_INDEX, CNV_TEST_DEBUG_WRITE_EN, mask_sh),\ - SF(CNV_TEST_DEBUG_DATA, CNV_TEST_DEBUG_DATA, mask_sh),\ - SF(WBSCL_COEF_RAM_SELECT, WBSCL_COEF_RAM_TAP_PAIR_IDX, mask_sh),\ - SF(WBSCL_COEF_RAM_SELECT, WBSCL_COEF_RAM_PHASE, mask_sh),\ - SF(WBSCL_COEF_RAM_SELECT, WBSCL_COEF_RAM_FILTER_TYPE, mask_sh),\ - SF(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_EVEN_TAP_COEF, mask_sh),\ - SF(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_EVEN_TAP_COEF_EN, mask_sh),\ - SF(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_ODD_TAP_COEF, mask_sh),\ - SF(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_ODD_TAP_COEF_EN, mask_sh),\ - SF(WBSCL_MODE, WBSCL_MODE, mask_sh),\ - SF(WBSCL_MODE, WBSCL_OUT_BIT_DEPTH, mask_sh),\ - SF(WBSCL_TAP_CONTROL, WBSCL_V_NUM_OF_TAPS_Y_RGB, mask_sh),\ - SF(WBSCL_TAP_CONTROL, WBSCL_V_NUM_OF_TAPS_CBCR, mask_sh),\ - SF(WBSCL_TAP_CONTROL, WBSCL_H_NUM_OF_TAPS_Y_RGB, mask_sh),\ - SF(WBSCL_TAP_CONTROL, WBSCL_H_NUM_OF_TAPS_CBCR, mask_sh),\ - SF(WBSCL_DEST_SIZE, WBSCL_DEST_HEIGHT, mask_sh),\ - SF(WBSCL_DEST_SIZE, WBSCL_DEST_WIDTH, mask_sh),\ - SF(WBSCL_HORZ_FILTER_SCALE_RATIO, WBSCL_H_SCALE_RATIO, mask_sh),\ - SF(WBSCL_HORZ_FILTER_INIT_Y_RGB, WBSCL_H_INIT_FRAC_Y_RGB, mask_sh),\ - SF(WBSCL_HORZ_FILTER_INIT_Y_RGB, WBSCL_H_INIT_INT_Y_RGB, mask_sh),\ - SF(WBSCL_HORZ_FILTER_INIT_CBCR, WBSCL_H_INIT_FRAC_CBCR, mask_sh),\ - SF(WBSCL_HORZ_FILTER_INIT_CBCR, WBSCL_H_INIT_INT_CBCR, mask_sh),\ - SF(WBSCL_VERT_FILTER_SCALE_RATIO, WBSCL_V_SCALE_RATIO, mask_sh),\ - SF(WBSCL_VERT_FILTER_INIT_Y_RGB, WBSCL_V_INIT_FRAC_Y_RGB, mask_sh),\ - SF(WBSCL_VERT_FILTER_INIT_Y_RGB, WBSCL_V_INIT_INT_Y_RGB, mask_sh),\ - SF(WBSCL_VERT_FILTER_INIT_CBCR, WBSCL_V_INIT_FRAC_CBCR, mask_sh),\ - SF(WBSCL_VERT_FILTER_INIT_CBCR, WBSCL_V_INIT_INT_CBCR, mask_sh),\ - SF(WBSCL_ROUND_OFFSET, WBSCL_ROUND_OFFSET_Y_RGB, mask_sh),\ - SF(WBSCL_ROUND_OFFSET, WBSCL_ROUND_OFFSET_CBCR, mask_sh),\ - SF(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_FLAG, mask_sh),\ - SF(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_ACK, mask_sh),\ - SF(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_MASK, mask_sh),\ - SF(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_INT_STATUS, mask_sh),\ - SF(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_INT_TYPE, mask_sh),\ - SF(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_FLAG, mask_sh),\ - SF(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_ACK, mask_sh),\ - SF(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_MASK, mask_sh),\ - SF(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_INT_STATUS, mask_sh),\ - SF(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_INT_TYPE, mask_sh),\ - SF(WBSCL_TEST_CNTL, WBSCL_TEST_CRC_EN, mask_sh),\ - SF(WBSCL_TEST_CNTL, WBSCL_TEST_CRC_CONT_EN, mask_sh),\ - SF(WBSCL_TEST_CRC_RED, WBSCL_TEST_CRC_RED_MASK, mask_sh),\ - SF(WBSCL_TEST_CRC_RED, WBSCL_TEST_CRC_SIG_RED, mask_sh),\ - SF(WBSCL_TEST_CRC_GREEN, WBSCL_TEST_CRC_GREEN_MASK, mask_sh),\ - SF(WBSCL_TEST_CRC_GREEN, WBSCL_TEST_CRC_SIG_GREEN, mask_sh),\ - SF(WBSCL_TEST_CRC_BLUE, WBSCL_TEST_CRC_BLUE_MASK, mask_sh),\ - SF(WBSCL_TEST_CRC_BLUE, WBSCL_TEST_CRC_SIG_BLUE, mask_sh),\ - SF(WBSCL_BACKPRESSURE_CNT_EN, WBSCL_BACKPRESSURE_CNT_EN, mask_sh),\ - SF(WB_MCIF_BACKPRESSURE_CNT, WB_MCIF_Y_MAX_BACKPRESSURE, mask_sh),\ - SF(WB_MCIF_BACKPRESSURE_CNT, WB_MCIF_C_MAX_BACKPRESSURE, mask_sh),\ - SF(WBSCL_CLAMP_Y_RGB, WBSCL_CLAMP_UPPER_Y_RGB, mask_sh),\ - SF(WBSCL_CLAMP_Y_RGB, WBSCL_CLAMP_LOWER_Y_RGB, mask_sh),\ - SF(WBSCL_CLAMP_CBCR, WBSCL_CLAMP_UPPER_CBCR, mask_sh),\ - SF(WBSCL_CLAMP_CBCR, WBSCL_CLAMP_LOWER_CBCR, mask_sh),\ - SF(WBSCL_OUTSIDE_PIX_STRATEGY, WBSCL_OUTSIDE_PIX_STRATEGY, mask_sh),\ - SF(WBSCL_OUTSIDE_PIX_STRATEGY, WBSCL_BLACK_COLOR_G_Y, mask_sh),\ - SF(WBSCL_OUTSIDE_PIX_STRATEGY_CBCR, WBSCL_BLACK_COLOR_B_CB, mask_sh),\ - SF(WBSCL_OUTSIDE_PIX_STRATEGY_CBCR, WBSCL_BLACK_COLOR_R_CR, mask_sh),\ - SF(WBSCL_DEBUG, WBSCL_DEBUG, mask_sh),\ - SF(WBSCL_TEST_DEBUG_INDEX, WBSCL_TEST_DEBUG_INDEX, mask_sh),\ - SF(WBSCL_TEST_DEBUG_INDEX, WBSCL_TEST_DEBUG_WRITE_EN, mask_sh),\ - SF(WBSCL_TEST_DEBUG_DATA, WBSCL_TEST_DEBUG_DATA, mask_sh),\ - SF(WB_WARM_UP_MODE_CTL1, WIDTH_WARMUP, mask_sh),\ - SF(WB_WARM_UP_MODE_CTL1, HEIGHT_WARMUP, mask_sh),\ - SF(WB_WARM_UP_MODE_CTL1, GMC_WARM_UP_ENABLE, mask_sh),\ - SF(WB_WARM_UP_MODE_CTL2, DATA_VALUE_WARMUP, mask_sh),\ - SF(WB_WARM_UP_MODE_CTL2, MODE_WARMUP, mask_sh),\ - SF(WB_WARM_UP_MODE_CTL2, DATA_DEPTH_WARMUP, mask_sh) + SF_DWB(WB_ENABLE, WB_ENABLE, mask_sh),\ + SF_DWB(WB_EC_CONFIG, DISPCLK_R_WB_GATE_DIS, mask_sh),\ + SF_DWB(WB_EC_CONFIG, DISPCLK_G_WB_GATE_DIS, mask_sh),\ + SF_DWB(WB_EC_CONFIG, DISPCLK_G_WBSCL_GATE_DIS, mask_sh),\ + SF_DWB(WB_EC_CONFIG, WB_TEST_CLK_SEL, mask_sh),\ + SF_DWB(WB_EC_CONFIG, WB_LB_LS_DIS, mask_sh),\ + SF_DWB(WB_EC_CONFIG, WB_LB_SD_DIS, mask_sh),\ + SF_DWB(WB_EC_CONFIG, WB_LUT_LS_DIS, mask_sh),\ + SF_DWB(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_MODE_SEL, mask_sh),\ + SF_DWB(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_DIS, mask_sh),\ + SF_DWB(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_FORCE, mask_sh),\ + SF_DWB(WB_EC_CONFIG, WBSCL_LB_MEM_PWR_STATE, mask_sh),\ + SF_DWB(WB_EC_CONFIG, WB_RAM_PW_SAVE_MODE, mask_sh),\ + SF_DWB(WB_EC_CONFIG, WBSCL_LUT_MEM_PWR_STATE, mask_sh),\ + SF_DWB(CNV_MODE, CNV_OUT_BPC, mask_sh),\ + SF_DWB(CNV_MODE, CNV_FRAME_CAPTURE_RATE, mask_sh),\ + SF_DWB(CNV_MODE, CNV_WINDOW_CROP_EN, mask_sh),\ + SF_DWB(CNV_MODE, CNV_STEREO_TYPE, mask_sh),\ + SF_DWB(CNV_MODE, CNV_INTERLACED_MODE, mask_sh),\ + SF_DWB(CNV_MODE, CNV_EYE_SELECTION, mask_sh),\ + SF_DWB(CNV_MODE, CNV_STEREO_POLARITY, mask_sh),\ + SF_DWB(CNV_MODE, CNV_INTERLACED_FIELD_ORDER, mask_sh),\ + SF_DWB(CNV_MODE, CNV_STEREO_SPLIT, mask_sh),\ + SF_DWB(CNV_MODE, CNV_NEW_CONTENT, mask_sh),\ + SF_DWB(CNV_MODE, CNV_FRAME_CAPTURE_EN_CURRENT, mask_sh),\ + SF_DWB(CNV_MODE, CNV_FRAME_CAPTURE_EN, mask_sh),\ + SF_DWB(CNV_WINDOW_START, CNV_WINDOW_START_X, mask_sh),\ + SF_DWB(CNV_WINDOW_START, CNV_WINDOW_START_Y, mask_sh),\ + SF_DWB(CNV_WINDOW_SIZE, CNV_WINDOW_WIDTH, mask_sh),\ + SF_DWB(CNV_WINDOW_SIZE, CNV_WINDOW_HEIGHT, mask_sh),\ + SF_DWB(CNV_UPDATE, CNV_UPDATE_PENDING, mask_sh),\ + SF_DWB(CNV_UPDATE, CNV_UPDATE_TAKEN, mask_sh),\ + SF_DWB(CNV_UPDATE, CNV_UPDATE_LOCK, mask_sh),\ + SF_DWB(CNV_SOURCE_SIZE, CNV_SOURCE_WIDTH, mask_sh),\ + SF_DWB(CNV_SOURCE_SIZE, CNV_SOURCE_HEIGHT, mask_sh),\ + SF_DWB(CNV_TEST_CNTL, CNV_TEST_CRC_EN, mask_sh),\ + SF_DWB(CNV_TEST_CNTL, CNV_TEST_CRC_CONT_EN, mask_sh),\ + SF_DWB(CNV_TEST_CRC_RED, CNV_TEST_CRC_RED_MASK, mask_sh),\ + SF_DWB(CNV_TEST_CRC_RED, CNV_TEST_CRC_SIG_RED, mask_sh),\ + SF_DWB(CNV_TEST_CRC_GREEN, CNV_TEST_CRC_GREEN_MASK, mask_sh),\ + SF_DWB(CNV_TEST_CRC_GREEN, CNV_TEST_CRC_SIG_GREEN, mask_sh),\ + SF_DWB(CNV_TEST_CRC_BLUE, CNV_TEST_CRC_BLUE_MASK, mask_sh),\ + SF_DWB(CNV_TEST_CRC_BLUE, CNV_TEST_CRC_SIG_BLUE, mask_sh),\ + SF_DWB(WB_DEBUG_CTRL, WB_DEBUG_EN, mask_sh),\ + SF_DWB(WB_DEBUG_CTRL, WB_DEBUG_SEL, mask_sh),\ + SF_DWB(WB_DBG_MODE, WB_DBG_MODE_EN, mask_sh),\ + SF_DWB(WB_DBG_MODE, WB_DBG_DIN_FMT, mask_sh),\ + SF_DWB(WB_DBG_MODE, WB_DBG_36MODE, mask_sh),\ + SF_DWB(WB_DBG_MODE, WB_DBG_CMAP, mask_sh),\ + SF_DWB(WB_DBG_MODE, WB_DBG_PXLRATE_ERROR, mask_sh),\ + SF_DWB(WB_DBG_MODE, WB_DBG_SOURCE_WIDTH, mask_sh),\ + SF_DWB(WB_HW_DEBUG, WB_HW_DEBUG, mask_sh),\ + SF_DWB(WB_SOFT_RESET, WB_SOFT_RESET, mask_sh),\ + SF_DWB(CNV_TEST_DEBUG_INDEX, CNV_TEST_DEBUG_INDEX, mask_sh),\ + SF_DWB(CNV_TEST_DEBUG_INDEX, CNV_TEST_DEBUG_WRITE_EN, mask_sh),\ + SF_DWB(CNV_TEST_DEBUG_DATA, CNV_TEST_DEBUG_DATA, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_SELECT, WBSCL_COEF_RAM_TAP_PAIR_IDX, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_SELECT, WBSCL_COEF_RAM_PHASE, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_SELECT, WBSCL_COEF_RAM_FILTER_TYPE, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_EVEN_TAP_COEF, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_EVEN_TAP_COEF_EN, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_ODD_TAP_COEF, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_TAP_DATA, WBSCL_COEF_RAM_ODD_TAP_COEF_EN, mask_sh),\ + SF_DWB(WBSCL_MODE, WBSCL_MODE, mask_sh),\ + SF_DWB(WBSCL_MODE, WBSCL_OUT_BIT_DEPTH, mask_sh),\ + SF_DWB(WBSCL_TAP_CONTROL, WBSCL_V_NUM_OF_TAPS_Y_RGB, mask_sh),\ + SF_DWB(WBSCL_TAP_CONTROL, WBSCL_V_NUM_OF_TAPS_CBCR, mask_sh),\ + SF_DWB(WBSCL_TAP_CONTROL, WBSCL_H_NUM_OF_TAPS_Y_RGB, mask_sh),\ + SF_DWB(WBSCL_TAP_CONTROL, WBSCL_H_NUM_OF_TAPS_CBCR, mask_sh),\ + SF_DWB(WBSCL_DEST_SIZE, WBSCL_DEST_HEIGHT, mask_sh),\ + SF_DWB(WBSCL_DEST_SIZE, WBSCL_DEST_WIDTH, mask_sh),\ + SF_DWB(WBSCL_HORZ_FILTER_SCALE_RATIO, WBSCL_H_SCALE_RATIO, mask_sh),\ + SF_DWB(WBSCL_HORZ_FILTER_INIT_Y_RGB, WBSCL_H_INIT_FRAC_Y_RGB, mask_sh),\ + SF_DWB(WBSCL_HORZ_FILTER_INIT_Y_RGB, WBSCL_H_INIT_INT_Y_RGB, mask_sh),\ + SF_DWB(WBSCL_HORZ_FILTER_INIT_CBCR, WBSCL_H_INIT_FRAC_CBCR, mask_sh),\ + SF_DWB(WBSCL_HORZ_FILTER_INIT_CBCR, WBSCL_H_INIT_INT_CBCR, mask_sh),\ + SF_DWB(WBSCL_VERT_FILTER_SCALE_RATIO, WBSCL_V_SCALE_RATIO, mask_sh),\ + SF_DWB(WBSCL_VERT_FILTER_INIT_Y_RGB, WBSCL_V_INIT_FRAC_Y_RGB, mask_sh),\ + SF_DWB(WBSCL_VERT_FILTER_INIT_Y_RGB, WBSCL_V_INIT_INT_Y_RGB, mask_sh),\ + SF_DWB(WBSCL_VERT_FILTER_INIT_CBCR, WBSCL_V_INIT_FRAC_CBCR, mask_sh),\ + SF_DWB(WBSCL_VERT_FILTER_INIT_CBCR, WBSCL_V_INIT_INT_CBCR, mask_sh),\ + SF_DWB(WBSCL_ROUND_OFFSET, WBSCL_ROUND_OFFSET_Y_RGB, mask_sh),\ + SF_DWB(WBSCL_ROUND_OFFSET, WBSCL_ROUND_OFFSET_CBCR, mask_sh),\ + SF_DWB(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_FLAG, mask_sh),\ + SF_DWB(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_ACK, mask_sh),\ + SF_DWB(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_MASK, mask_sh),\ + SF_DWB(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_INT_STATUS, mask_sh),\ + SF_DWB(WBSCL_OVERFLOW_STATUS, WBSCL_DATA_OVERFLOW_INT_TYPE, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_FLAG, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_ACK, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_MASK, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_INT_STATUS, mask_sh),\ + SF_DWB(WBSCL_COEF_RAM_CONFLICT_STATUS, WBSCL_HOST_CONFLICT_INT_TYPE, mask_sh),\ + SF_DWB(WBSCL_TEST_CNTL, WBSCL_TEST_CRC_EN, mask_sh),\ + SF_DWB(WBSCL_TEST_CNTL, WBSCL_TEST_CRC_CONT_EN, mask_sh),\ + SF_DWB(WBSCL_TEST_CRC_RED, WBSCL_TEST_CRC_RED_MASK, mask_sh),\ + SF_DWB(WBSCL_TEST_CRC_RED, WBSCL_TEST_CRC_SIG_RED, mask_sh),\ + SF_DWB(WBSCL_TEST_CRC_GREEN, WBSCL_TEST_CRC_GREEN_MASK, mask_sh),\ + SF_DWB(WBSCL_TEST_CRC_GREEN, WBSCL_TEST_CRC_SIG_GREEN, mask_sh),\ + SF_DWB(WBSCL_TEST_CRC_BLUE, WBSCL_TEST_CRC_BLUE_MASK, mask_sh),\ + SF_DWB(WBSCL_TEST_CRC_BLUE, WBSCL_TEST_CRC_SIG_BLUE, mask_sh),\ + SF_DWB(WBSCL_BACKPRESSURE_CNT_EN, WBSCL_BACKPRESSURE_CNT_EN, mask_sh),\ + SF_DWB(WB_MCIF_BACKPRESSURE_CNT, WB_MCIF_Y_MAX_BACKPRESSURE, mask_sh),\ + SF_DWB(WB_MCIF_BACKPRESSURE_CNT, WB_MCIF_C_MAX_BACKPRESSURE, mask_sh),\ + SF_DWB(WBSCL_CLAMP_Y_RGB, WBSCL_CLAMP_UPPER_Y_RGB, mask_sh),\ + SF_DWB(WBSCL_CLAMP_Y_RGB, WBSCL_CLAMP_LOWER_Y_RGB, mask_sh),\ + SF_DWB(WBSCL_CLAMP_CBCR, WBSCL_CLAMP_UPPER_CBCR, mask_sh),\ + SF_DWB(WBSCL_CLAMP_CBCR, WBSCL_CLAMP_LOWER_CBCR, mask_sh),\ + SF_DWB(WBSCL_OUTSIDE_PIX_STRATEGY, WBSCL_OUTSIDE_PIX_STRATEGY, mask_sh),\ + SF_DWB(WBSCL_OUTSIDE_PIX_STRATEGY, WBSCL_BLACK_COLOR_G_Y, mask_sh),\ + SF_DWB(WBSCL_OUTSIDE_PIX_STRATEGY_CBCR, WBSCL_BLACK_COLOR_B_CB, mask_sh),\ + SF_DWB(WBSCL_OUTSIDE_PIX_STRATEGY_CBCR, WBSCL_BLACK_COLOR_R_CR, mask_sh),\ + SF_DWB(WBSCL_DEBUG, WBSCL_DEBUG, mask_sh),\ + SF_DWB(WBSCL_TEST_DEBUG_INDEX, WBSCL_TEST_DEBUG_INDEX, mask_sh),\ + SF_DWB(WBSCL_TEST_DEBUG_INDEX, WBSCL_TEST_DEBUG_WRITE_EN, mask_sh),\ + SF_DWB(WBSCL_TEST_DEBUG_DATA, WBSCL_TEST_DEBUG_DATA, mask_sh),\ + SF_DWB(WB_WARM_UP_MODE_CTL1, WIDTH_WARMUP, mask_sh),\ + SF_DWB(WB_WARM_UP_MODE_CTL1, HEIGHT_WARMUP, mask_sh),\ + SF_DWB(WB_WARM_UP_MODE_CTL1, GMC_WARM_UP_ENABLE, mask_sh),\ + SF_DWB(WB_WARM_UP_MODE_CTL2, DATA_VALUE_WARMUP, mask_sh),\ + SF_DWB(WB_WARM_UP_MODE_CTL2, MODE_WARMUP, mask_sh),\ + SF_DWB(WB_WARM_UP_MODE_CTL2, DATA_DEPTH_WARMUP, mask_sh) #define DWBC_REG_FIELD_LIST_DCN2_0(type) \ type WB_ENABLE;\ diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.c index 4996d2810edb..938dba5249d4 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hubp.c @@ -623,6 +623,10 @@ void hubp2_cursor_set_attributes( hubp->att.size.bits.width = attr->width; hubp->att.size.bits.height = attr->height; hubp->att.cur_ctl.bits.mode = attr->color_format; + + hubp->cur_rect.w = attr->width; + hubp->cur_rect.h = attr->height; + hubp->att.cur_ctl.bits.pitch = hw_pitch; hubp->att.cur_ctl.bits.line_per_chunk = lpc; hubp->att.cur_ctl.bits.cur_2x_magnify = attr->attribute_flags.bits.ENABLE_MAGNIFICATION; diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c index d732b6f031a1..3f3d4daa6294 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c @@ -1079,6 +1079,29 @@ void dcn20_blank_pixel_data( 0); } + if (!blank && dc->debug.enable_single_display_2to1_odm_policy) { + /* when exiting dynamic ODM need to reinit DPG state for unused pipes */ + struct pipe_ctx *old_odm_pipe = dc->current_state->res_ctx.pipe_ctx[pipe_ctx->pipe_idx].next_odm_pipe; + + odm_pipe = pipe_ctx->next_odm_pipe; + + while (old_odm_pipe) { + if (!odm_pipe || old_odm_pipe->pipe_idx != odm_pipe->pipe_idx) + dc->hwss.set_disp_pattern_generator(dc, + old_odm_pipe, + CONTROLLER_DP_TEST_PATTERN_VIDEOMODE, + CONTROLLER_DP_COLOR_SPACE_UDEFINED, + COLOR_DEPTH_888, + NULL, + 0, + 0, + 0); + old_odm_pipe = old_odm_pipe->next_odm_pipe; + if (odm_pipe) + odm_pipe = odm_pipe->next_odm_pipe; + } + } + if (!blank) if (stream_res->abm) { dc->hwss.set_pipe(pipe_ctx); @@ -1270,16 +1293,6 @@ void dcn20_pipe_control_lock( lock, &hw_locks, &inst_flags); - } else if (pipe->stream && pipe->stream->mall_stream_config.type == SUBVP_MAIN) { - union dmub_inbox0_cmd_lock_hw hw_lock_cmd = { 0 }; - hw_lock_cmd.bits.command_code = DMUB_INBOX0_CMD__HW_LOCK; - hw_lock_cmd.bits.hw_lock_client = HW_LOCK_CLIENT_DRIVER; - hw_lock_cmd.bits.lock_pipe = 1; - hw_lock_cmd.bits.otg_inst = pipe->stream_res.tg->inst; - hw_lock_cmd.bits.lock = lock; - if (!lock) - hw_lock_cmd.bits.should_release = 1; - dmub_hw_lock_mgr_inbox0_cmd(dc->ctx->dmub_srv, hw_lock_cmd); } else if (pipe->plane_state != NULL && pipe->plane_state->triplebuffer_flips) { if (lock) pipe->stream_res.tg->funcs->triplebuffer_lock(pipe->stream_res.tg); @@ -1297,6 +1310,19 @@ static void dcn20_detect_pipe_changes(struct pipe_ctx *old_pipe, struct pipe_ctx { new_pipe->update_flags.raw = 0; + /* If non-phantom pipe is being transitioned to a phantom pipe, + * set disable and return immediately. This is because the pipe + * that was previously in use must be fully disabled before we + * can "enable" it as a phantom pipe (since the OTG will certainly + * be different). The post_unlock sequence will set the correct + * update flags to enable the phantom pipe. + */ + if (old_pipe->plane_state && !old_pipe->plane_state->is_phantom && + new_pipe->plane_state && new_pipe->plane_state->is_phantom) { + new_pipe->update_flags.bits.disable = 1; + return; + } + /* Exit on unchanged, unused pipe */ if (!old_pipe->plane_state && !new_pipe->plane_state) return; @@ -1650,10 +1676,8 @@ static void dcn20_program_pipe( pipe_ctx->pipe_dlg_param.vupdate_width); if (pipe_ctx->stream->mall_stream_config.type != SUBVP_PHANTOM) { - pipe_ctx->stream_res.tg->funcs->wait_for_state( - pipe_ctx->stream_res.tg, CRTC_STATE_VBLANK); - pipe_ctx->stream_res.tg->funcs->wait_for_state( - pipe_ctx->stream_res.tg, CRTC_STATE_VACTIVE); + pipe_ctx->stream_res.tg->funcs->wait_for_state(pipe_ctx->stream_res.tg, CRTC_STATE_VBLANK); + pipe_ctx->stream_res.tg->funcs->wait_for_state(pipe_ctx->stream_res.tg, CRTC_STATE_VACTIVE); } pipe_ctx->stream_res.tg->funcs->set_vtg_params( @@ -1823,6 +1847,17 @@ void dcn20_program_front_end_for_ctx( context->stream_status[0].plane_count > 1) { pipe->plane_res.hubp->funcs->hubp_wait_pipe_read_start(pipe->plane_res.hubp); } + + /* when dynamic ODM is active, pipes must be reconfigured when all planes are + * disabled, as some transitions will leave software and hardware state + * mismatched. + */ + if (dc->debug.enable_single_display_2to1_odm_policy && + pipe->stream && + pipe->update_flags.bits.disable && + !pipe->prev_odm_pipe && + hws->funcs.update_odm) + hws->funcs.update_odm(dc, context, pipe); } } @@ -1856,27 +1891,7 @@ void dcn20_post_unlock_program_front_end( for (j = 0; j < TIMEOUT_FOR_PIPE_ENABLE_MS*1000 && hubp->funcs->hubp_is_flip_pending(hubp); j++) - mdelay(1); - } - } - - for (i = 0; i < dc->res_pool->pipe_count; i++) { - struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; - struct pipe_ctx *old_pipe = &dc->current_state->res_ctx.pipe_ctx[i]; - - /* If an active, non-phantom pipe is being transitioned into a phantom - * pipe, wait for the double buffer update to complete first before we do - * phantom pipe programming (HUBP_VTG_SEL updates right away so that can - * cause issues). - */ - if (pipe->stream && pipe->stream->mall_stream_config.type == SUBVP_PHANTOM && - old_pipe->stream && old_pipe->stream->mall_stream_config.type != SUBVP_PHANTOM) { - old_pipe->stream_res.tg->funcs->wait_for_state( - old_pipe->stream_res.tg, - CRTC_STATE_VBLANK); - old_pipe->stream_res.tg->funcs->wait_for_state( - old_pipe->stream_res.tg, - CRTC_STATE_VACTIVE); + udelay(1); } } @@ -1891,6 +1906,11 @@ void dcn20_post_unlock_program_front_end( */ while (pipe) { if (pipe->stream && pipe->stream->mall_stream_config.type == SUBVP_PHANTOM) { + /* When turning on the phantom pipe we want to run through the + * entire enable sequence, so apply all the "enable" flags. + */ + if (dc->hwss.apply_update_flags_for_phantom) + dc->hwss.apply_update_flags_for_phantom(pipe); if (dc->hwss.update_phantom_vp_position) dc->hwss.update_phantom_vp_position(dc, context, pipe); dcn20_program_pipe(dc, pipe, context); @@ -2611,14 +2631,6 @@ void dcn20_enable_stream(struct pipe_ctx *pipe_ctx) if (dc->hwseq->funcs.set_pixels_per_cycle) dc->hwseq->funcs.set_pixels_per_cycle(pipe_ctx); - - /* enable audio only within mode set */ - if (pipe_ctx->stream_res.audio != NULL) { - if (is_dp_128b_132b_signal(pipe_ctx)) - pipe_ctx->stream_res.hpo_dp_stream_enc->funcs->dp_audio_enable(pipe_ctx->stream_res.hpo_dp_stream_enc); - else if (dc_is_dp_signal(pipe_ctx->stream->signal)) - pipe_ctx->stream_res.stream_enc->funcs->dp_audio_enable(pipe_ctx->stream_res.stream_enc); - } } void dcn20_program_dmdata_engine(struct pipe_ctx *pipe_ctx) diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mmhubbub.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mmhubbub.h index 7bcee5894d2e..5ab32aa51e13 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mmhubbub.h +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_mmhubbub.h @@ -29,13 +29,6 @@ #define TO_DCN20_MMHUBBUB(mcif_wb_base) \ container_of(mcif_wb_base, struct dcn20_mmhubbub, base) -/* DCN */ -#define BASE_INNER(seg) \ - DCE_BASE__INST0_SEG ## seg - -#define BASE(seg) \ - BASE_INNER(seg) - #define MCIF_WB_COMMON_REG_LIST_DCN2_0(inst) \ SRI(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB, inst),\ SRI(MCIF_WB_BUFMGR_CUR_LINE_R, MCIF_WB, inst),\ diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c index 8224b9bf01d1..8a0dd0d7134b 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c @@ -124,8 +124,6 @@ enum dcn20_clk_src_array_id { * macros to expend register list macro defined in HW object header file */ /* DCN */ -/* TODO awful hack. fixup dcn20_dwb.h */ -#undef BASE_INNER #define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg #define BASE(seg) BASE_INNER(seg) @@ -138,6 +136,15 @@ enum dcn20_clk_src_array_id { .reg_name = BASE(mm ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ mm ## block ## id ## _ ## reg_name +#define SRI2_DWB(reg_name, block, id)\ + .reg_name = BASE(mm ## reg_name ## _BASE_IDX) + \ + mm ## reg_name +#define SF_DWB(reg_name, field_name, post_fix)\ + .field_name = reg_name ## __ ## field_name ## post_fix + +#define SF_DWB2(reg_name, block, id, field_name, post_fix) \ + .field_name = reg_name ## __ ## field_name ## post_fix + #define SRIR(var_name, reg_name, block, id)\ .var_name = BASE(mm ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ mm ## block ## id ## _ ## reg_name @@ -1454,6 +1461,22 @@ enum dc_status dcn20_remove_stream_from_ctx(struct dc *dc, struct dc_state *new_ return result; } +/** + * dcn20_split_stream_for_odm - Check if stream can be splited for ODM + * + * @dc: DC object with resource pool info required for pipe split + * @res_ctx: Persistent state of resources + * @prev_odm_pipe: Reference to the previous ODM pipe + * @next_odm_pipe: Reference to the next ODM pipe + * + * This function takes a logically active pipe and a logically free pipe and + * halves all the scaling parameters that need to be halved while populating + * the free pipe with the required resources and configuring the next/previous + * ODM pipe pointers. + * + * Return: + * Return true if split stream for ODM is possible, otherwise, return false. + */ bool dcn20_split_stream_for_odm( const struct dc *dc, struct resource_context *res_ctx, diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.h b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.h index f1ef46e8da5b..e7a1b7fa2cce 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.h +++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_vmid.h @@ -28,12 +28,6 @@ #include "vmid.h" -#define BASE_INNER(seg) \ - DCE_BASE__INST0_SEG ## seg - -#define BASE(seg) \ - BASE_INNER(seg) - #define DCN20_VMID_REG_LIST(id)\ SRI(CNTL, DCN_VM_CONTEXT, id),\ SRI(PAGE_TABLE_BASE_ADDR_HI32, DCN_VM_CONTEXT, id),\ diff --git a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c index 887081472c0d..fbcf0afeae0d 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn21/dcn21_resource.c @@ -94,8 +94,6 @@ * macros to expend register list macro defined in HW object header file */ /* DCN */ -/* TODO awful hack. fixup dcn20_dwb.h */ -#undef BASE_INNER #define BASE_INNER(seg) DMU_BASE__INST0_SEG ## seg #define BASE(seg) BASE_INNER(seg) @@ -671,12 +669,15 @@ static const struct dc_debug_options debug_defaults_diags = { .disable_pplib_wm_range = true, .disable_stutter = true, .disable_48mhz_pwrdwn = true, - .disable_psr = true, .enable_tri_buf = true, .use_max_lb = true }; static const struct dc_panel_config panel_config_defaults = { + .psr = { + .disable_psr = false, + .disallow_psrsu = false, + }, .ilr = { .optimize_edp_link_rate = true, }, diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_dwb.h b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_dwb.h index 1010930cf071..fc00ec0a0881 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_dwb.h +++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_dwb.h @@ -27,21 +27,6 @@ #define TO_DCN30_DWBC(dwbc_base) \ container_of(dwbc_base, struct dcn30_dwbc, base) -/* DCN */ -#define BASE_INNER(seg) \ - DCE_BASE__INST0_SEG ## seg - -#define BASE(seg) \ - BASE_INNER(seg) - -#define SF_DWB(reg_name, block, id, field_name, post_fix)\ - .field_name = block ## id ## _ ## reg_name ## __ ## field_name ## post_fix - - /* set field name */ -#define SF_DWB2(reg_name, block, id, field_name, post_fix)\ - .field_name = reg_name ## __ ## field_name ## post_fix - - #define DWBC_COMMON_REG_LIST_DCN30(inst) \ SR(DWB_ENABLE_CLK_CTRL),\ SR(DWB_MEM_PWR_CTRL),\ diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_mmhubbub.h b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_mmhubbub.h index 7446e54bf5aa..376620a8f02f 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_mmhubbub.h +++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_mmhubbub.h @@ -31,13 +31,6 @@ #define TO_DCN30_MMHUBBUB(mcif_wb_base) \ container_of(mcif_wb_base, struct dcn30_mmhubbub, base) -/* DCN */ -#define BASE_INNER(seg) \ - DCE_BASE__INST0_SEG ## seg - -#define BASE(seg) \ - BASE_INNER(seg) - #define MCIF_WB_COMMON_REG_LIST_DCN3_0(inst) \ SRI(MCIF_WB_BUFMGR_SW_CONTROL, MCIF_WB, inst),\ SRI(MCIF_WB_BUFMGR_STATUS, MCIF_WB, inst),\ diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_optc.c b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_optc.c index 892d3c4d01a1..867d60151aeb 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_optc.c +++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_optc.c @@ -30,6 +30,7 @@ #include "dc_dmub_srv.h" #include "dml/dcn30/dcn30_fpu.h" +#include "dc_trace.h" #define REG(reg)\ optc1->tg_regs->reg @@ -58,6 +59,8 @@ void optc3_triplebuffer_lock(struct timing_generator *optc) REG_WAIT(OTG_MASTER_UPDATE_LOCK, UPDATE_LOCK_STATUS, 1, 1, 10); + + TRACE_OPTC_LOCK_UNLOCK_STATE(optc1, optc->inst, true); } void optc3_lock_doublebuffer_enable(struct timing_generator *optc) @@ -93,6 +96,8 @@ void optc3_lock_doublebuffer_enable(struct timing_generator *optc) MASTER_UPDATE_LOCK_VUPDATE_KEEPOUT_START_OFFSET, 0, MASTER_UPDATE_LOCK_VUPDATE_KEEPOUT_END_OFFSET, 100, OTG_MASTER_UPDATE_LOCK_VUPDATE_KEEPOUT_EN, 1); + + TRACE_OPTC_LOCK_UNLOCK_STATE(optc1, optc->inst, true); } void optc3_lock_doublebuffer_disable(struct timing_generator *optc) @@ -108,6 +113,8 @@ void optc3_lock_doublebuffer_disable(struct timing_generator *optc) REG_UPDATE(OTG_GLOBAL_CONTROL2, GLOBAL_UPDATE_LOCK_EN, 0); REG_UPDATE(OTG_GLOBAL_CONTROL0, MASTER_UPDATE_LOCK_DB_EN, 0); + + TRACE_OPTC_LOCK_UNLOCK_STATE(optc1, optc->inst, true); } void optc3_lock(struct timing_generator *optc) @@ -122,6 +129,8 @@ void optc3_lock(struct timing_generator *optc) REG_WAIT(OTG_MASTER_UPDATE_LOCK, UPDATE_LOCK_STATUS, 1, 1, 10); + + TRACE_OPTC_LOCK_UNLOCK_STATE(optc1, optc->inst, true); } void optc3_set_out_mux(struct timing_generator *optc, enum otg_out_mux_dest dest) diff --git a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c index 020f512e9690..c18c52a60100 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c @@ -108,8 +108,6 @@ enum dcn30_clk_src_array_id { */ /* DCN */ -/* TODO awful hack. fixup dcn20_dwb.h */ -#undef BASE_INNER #define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg #define BASE(seg) BASE_INNER(seg) @@ -142,6 +140,9 @@ enum dcn30_clk_src_array_id { .reg_name[id] = BASE(mm ## block ## id ## _ ## temp_name ## _BASE_IDX) + \ mm ## block ## id ## _ ## temp_name +#define SF_DWB2(reg_name, block, id, field_name, post_fix) \ + .field_name = reg_name ## __ ## field_name ## post_fix + #define DCCG_SRII(reg_name, block, id)\ .block ## _ ## reg_name[id] = BASE(mm ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ mm ## block ## id ## _ ## reg_name @@ -723,7 +724,6 @@ static const struct dc_debug_options debug_defaults_drv = { .underflow_assert_delay_us = 0xFFFFFFFF, .dwb_fi_phase = -1, // -1 = disable, .dmub_command_table = true, - .disable_psr = false, .use_max_lb = true, .exit_idle_opt_for_cursor_updates = true }; @@ -742,11 +742,17 @@ static const struct dc_debug_options debug_defaults_diags = { .scl_reset_length10 = true, .dwb_fi_phase = -1, // -1 = disable .dmub_command_table = true, - .disable_psr = true, .enable_tri_buf = true, .use_max_lb = true }; +static const struct dc_panel_config panel_config_defaults = { + .psr = { + .disable_psr = false, + .disallow_psrsu = false, + }, +}; + static void dcn30_dpp_destroy(struct dpp **dpp) { kfree(TO_DCN20_DPP(*dpp)); @@ -1323,6 +1329,7 @@ static struct clock_source *dcn30_clock_source_create( return &clk_src->base; } + kfree(clk_src); BREAK_TO_DEBUGGER(); return NULL; } @@ -2212,6 +2219,11 @@ void dcn30_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params } } +static void dcn30_get_panel_config_defaults(struct dc_panel_config *panel_config) +{ + *panel_config = panel_config_defaults; +} + static const struct resource_funcs dcn30_res_pool_funcs = { .destroy = dcn30_destroy_resource_pool, .link_enc_create = dcn30_link_encoder_create, @@ -2231,6 +2243,7 @@ static const struct resource_funcs dcn30_res_pool_funcs = { .release_post_bldn_3dlut = dcn30_release_post_bldn_3dlut, .update_bw_bounding_box = dcn30_update_bw_bounding_box, .patch_unknown_plane_state = dcn20_patch_unknown_plane_state, + .get_panel_config_defaults = dcn30_get_panel_config_defaults, }; #define CTX ctx diff --git a/drivers/gpu/drm/amd/display/dc/dcn301/dcn301_resource.c b/drivers/gpu/drm/amd/display/dc/dcn301/dcn301_resource.c index f04595b750ab..480145f09246 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn301/dcn301_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn301/dcn301_resource.c @@ -107,8 +107,6 @@ enum dcn301_clk_src_array_id { */ /* DCN */ -/* TODO awful hack. fixup dcn20_dwb.h */ -#undef BASE_INNER #define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg #define BASE(seg) BASE_INNER(seg) @@ -146,6 +144,9 @@ enum dcn301_clk_src_array_id { .reg_name[id] = BASE(mm ## block ## id ## _ ## temp_name ## _BASE_IDX) + \ mm ## block ## id ## _ ## temp_name +#define SF_DWB2(reg_name, block, id, field_name, post_fix) \ + .field_name = reg_name ## __ ## field_name ## post_fix + #define DCCG_SRII(reg_name, block, id)\ .block ## _ ## reg_name[id] = BASE(mm ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ mm ## block ## id ## _ ## reg_name @@ -1288,6 +1289,7 @@ static struct clock_source *dcn301_clock_source_create( return &clk_src->base; } + kfree(clk_src); BREAK_TO_DEBUGGER(); return NULL; } diff --git a/drivers/gpu/drm/amd/display/dc/dcn302/dcn302_resource.c b/drivers/gpu/drm/amd/display/dc/dcn302/dcn302_resource.c index b925b6ddde5a..7d11c2a43cbe 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn302/dcn302_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn302/dcn302_resource.c @@ -112,10 +112,16 @@ static const struct dc_debug_options debug_defaults_diags = { .dwb_fi_phase = -1, // -1 = disable .dmub_command_table = true, .enable_tri_buf = true, - .disable_psr = true, .use_max_lb = true }; +static const struct dc_panel_config panel_config_defaults = { + .psr = { + .disable_psr = false, + .disallow_psrsu = false, + }, +}; + enum dcn302_clk_src_array_id { DCN302_CLK_SRC_PLL0, DCN302_CLK_SRC_PLL1, @@ -177,7 +183,6 @@ static const struct dc_plane_cap plane_cap = { mm ## reg_name /* DCN */ -#undef BASE_INNER #define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg #define BASE(seg) BASE_INNER(seg) @@ -210,6 +215,9 @@ static const struct dc_plane_cap plane_cap = { .reg_name[id] = BASE(mm ## block ## id ## _ ## temp_name ## _BASE_IDX) + \ mm ## block ## id ## _ ## temp_name +#define SF_DWB2(reg_name, block, id, field_name, post_fix) \ + .field_name = reg_name ## __ ## field_name ## post_fix + #define SRII_MPC_RMU(reg_name, block, id)\ .RMU##_##reg_name[id] = BASE(mm ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ mm ## block ## id ## _ ## reg_name @@ -458,6 +466,7 @@ static struct clock_source *dcn302_clock_source_create(struct dc_context *ctx, s return &clk_src->base; } + kfree(clk_src); BREAK_TO_DEBUGGER(); return NULL; } @@ -1132,6 +1141,11 @@ void dcn302_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_param DC_FP_END(); } +static void dcn302_get_panel_config_defaults(struct dc_panel_config *panel_config) +{ + *panel_config = panel_config_defaults; +} + static struct resource_funcs dcn302_res_pool_funcs = { .destroy = dcn302_destroy_resource_pool, .link_enc_create = dcn302_link_encoder_create, @@ -1151,6 +1165,7 @@ static struct resource_funcs dcn302_res_pool_funcs = { .release_post_bldn_3dlut = dcn30_release_post_bldn_3dlut, .update_bw_bounding_box = dcn302_update_bw_bounding_box, .patch_unknown_plane_state = dcn20_patch_unknown_plane_state, + .get_panel_config_defaults = dcn302_get_panel_config_defaults, }; static struct dc_cap_funcs cap_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/dcn303/dcn303_resource.c b/drivers/gpu/drm/amd/display/dc/dcn303/dcn303_resource.c index 527d5c902878..92393b04cc44 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn303/dcn303_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn303/dcn303_resource.c @@ -96,7 +96,13 @@ static const struct dc_debug_options debug_defaults_diags = { .dwb_fi_phase = -1, // -1 = disable .dmub_command_table = true, .enable_tri_buf = true, - .disable_psr = true, +}; + +static const struct dc_panel_config panel_config_defaults = { + .psr = { + .disable_psr = false, + .disallow_psrsu = false, + }, }; enum dcn303_clk_src_array_id { @@ -156,7 +162,6 @@ static const struct dc_plane_cap plane_cap = { mm ## reg_name /* DCN */ -#undef BASE_INNER #define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg #define BASE(seg) BASE_INNER(seg) @@ -189,6 +194,9 @@ static const struct dc_plane_cap plane_cap = { .reg_name[id] = BASE(mm ## block ## id ## _ ## temp_name ## _BASE_IDX) + \ mm ## block ## id ## _ ## temp_name +#define SF_DWB2(reg_name, block, id, field_name, post_fix) \ + .field_name = reg_name ## __ ## field_name ## post_fix + #define SRII_MPC_RMU(reg_name, block, id)\ .RMU##_##reg_name[id] = BASE(mm ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ mm ## block ## id ## _ ## reg_name @@ -425,6 +433,7 @@ static struct clock_source *dcn303_clock_source_create(struct dc_context *ctx, s return &clk_src->base; } + kfree(clk_src); BREAK_TO_DEBUGGER(); return NULL; } @@ -1055,6 +1064,10 @@ static void dcn303_destroy_resource_pool(struct resource_pool **pool) *pool = NULL; } +static void dcn303_get_panel_config_defaults(struct dc_panel_config *panel_config) +{ + *panel_config = panel_config_defaults; +} void dcn303_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params) { @@ -1082,6 +1095,7 @@ static struct resource_funcs dcn303_res_pool_funcs = { .release_post_bldn_3dlut = dcn30_release_post_bldn_3dlut, .update_bw_bounding_box = dcn303_update_bw_bounding_box, .patch_unknown_plane_state = dcn20_patch_unknown_plane_state, + .get_panel_config_defaults = dcn303_get_panel_config_defaults, }; static struct dc_cap_funcs cap_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.c index de5e18c2a3ac..24e9ff65434d 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.c +++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.c @@ -134,23 +134,10 @@ static void apg31_se_audio_setup( /* Disable forced mem power off */ REG_UPDATE(APG_MEM_PWR, APG_MEM_PWR_FORCE, 0); - - apg31_enable(apg); -} - -static void apg31_audio_mute_control( - struct apg *apg, - bool mute) -{ - if (mute) - apg31_disable(apg); - else - apg31_enable(apg); } static struct apg_funcs dcn31_apg_funcs = { .se_audio_setup = apg31_se_audio_setup, - .audio_mute_control = apg31_audio_mute_control, .enable_apg = apg31_enable, .disable_apg = apg31_disable, }; diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.h b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.h index 24f568e120d8..1b81f6773c53 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.h +++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_apg.h @@ -84,10 +84,6 @@ struct apg_funcs { unsigned int az_inst, struct audio_info *audio_info); - void (*audio_mute_control)( - struct apg *apg, - bool mute); - void (*enable_apg)( struct apg *apg); diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hpo_dp_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hpo_dp_stream_encoder.c index 814f401db3b3..16639bd03adf 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hpo_dp_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hpo_dp_stream_encoder.c @@ -600,14 +600,6 @@ static void dcn31_hpo_dp_stream_enc_map_stream_to_link( } } -static void dcn31_hpo_dp_stream_enc_mute_control( - struct hpo_dp_stream_encoder *enc, - bool mute) -{ - ASSERT(enc->apg); - enc->apg->funcs->audio_mute_control(enc->apg, mute); -} - static void dcn31_hpo_dp_stream_enc_audio_setup( struct hpo_dp_stream_encoder *enc, unsigned int az_inst, @@ -726,7 +718,6 @@ static const struct hpo_dp_stream_encoder_funcs dcn30_str_enc_funcs = { .stop_dp_info_packets = dcn31_hpo_dp_stream_enc_stop_dp_info_packets, .dp_set_dsc_pps_info_packet = dcn31_hpo_dp_stream_enc_set_dsc_pps_info_packet, .map_stream_to_link = dcn31_hpo_dp_stream_enc_map_stream_to_link, - .audio_mute_control = dcn31_hpo_dp_stream_enc_mute_control, .dp_audio_setup = dcn31_hpo_dp_stream_enc_audio_setup, .dp_audio_enable = dcn31_hpo_dp_stream_enc_audio_enable, .dp_audio_disable = dcn31_hpo_dp_stream_enc_audio_disable, diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hubp.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hubp.c index 84e1486f3d51..39a57bcd7866 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hubp.c @@ -87,6 +87,7 @@ static struct hubp_funcs dcn31_hubp_funcs = { .hubp_init = hubp3_init, .set_unbounded_requesting = hubp31_set_unbounded_requesting, .hubp_soft_reset = hubp31_soft_reset, + .hubp_set_flip_int = hubp1_set_flip_int, .hubp_in_blank = hubp1_in_blank, .program_extended_blank = hubp31_program_extended_blank, }; diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hwseq.c index bdf101547484..165c920ca776 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_hwseq.c @@ -89,7 +89,8 @@ static void enable_memory_low_power(struct dc *dc) REG_UPDATE(MMHUBBUB_MEM_PWR_CNTL, VGA_MEM_PWR_FORCE, 1); } - if (dc->debug.enable_mem_low_power.bits.mpc) + if (dc->debug.enable_mem_low_power.bits.mpc && + dc->res_pool->mpc->funcs->set_mpc_mem_lp_mode) dc->res_pool->mpc->funcs->set_mpc_mem_lp_mode(dc->res_pool->mpc); @@ -141,7 +142,8 @@ void dcn31_init_hw(struct dc *dc) if (!dcb->funcs->is_accelerated_mode(dcb)) { hws->funcs.bios_golden_init(dc); - hws->funcs.disable_vga(dc->hwseq); + if (hws->funcs.disable_vga) + hws->funcs.disable_vga(dc->hwseq); } // Initialize the dccg if (res_pool->dccg->funcs->dccg_init) diff --git a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c index fddc21a5a04c..3ca517dcc82d 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c @@ -119,8 +119,6 @@ enum dcn31_clk_src_array_id { */ /* DCN */ -/* TODO awful hack. fixup dcn20_dwb.h */ -#undef BASE_INNER #define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg #define BASE(seg) BASE_INNER(seg) @@ -153,6 +151,9 @@ enum dcn31_clk_src_array_id { .reg_name[id] = BASE(reg ## block ## id ## _ ## temp_name ## _BASE_IDX) + \ reg ## block ## id ## _ ## temp_name +#define SF_DWB2(reg_name, block, id, field_name, post_fix) \ + .field_name = reg_name ## __ ## field_name ## post_fix + #define DCCG_SRII(reg_name, block, id)\ .block ## _ ## reg_name[id] = BASE(reg ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ reg ## block ## id ## _ ## reg_name @@ -911,6 +912,10 @@ static const struct dc_debug_options debug_defaults_diags = { }; static const struct dc_panel_config panel_config_defaults = { + .psr = { + .disable_psr = false, + .disallow_psrsu = false, + }, .ilr = { .optimize_edp_link_rate = true, }, @@ -1625,6 +1630,7 @@ static struct clock_source *dcn31_clock_source_create( return &clk_src->base; } + kfree(clk_src); BREAK_TO_DEBUGGER(); return NULL; } @@ -1634,6 +1640,31 @@ static bool is_dual_plane(enum surface_pixel_format format) return format >= SURFACE_PIXEL_FORMAT_VIDEO_BEGIN || format == SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA; } +int dcn31x_populate_dml_pipes_from_context(struct dc *dc, + struct dc_state *context, + display_e2e_pipe_params_st *pipes, + bool fast_validate) +{ + uint32_t pipe_cnt; + int i; + + dc_assert_fp_enabled(); + + pipe_cnt = dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + + for (i = 0; i < pipe_cnt; i++) { + pipes[i].pipe.src.gpuvm = 1; + if (dc->debug.dml_hostvm_override == DML_HOSTVM_NO_OVERRIDE) { + //pipes[pipe_cnt].pipe.src.hostvm = dc->res_pool->hubbub->riommu_active; + pipes[i].pipe.src.hostvm = dc->vm_pa_config.is_hvm_enabled; + } else if (dc->debug.dml_hostvm_override == DML_HOSTVM_OVERRIDE_FALSE) + pipes[i].pipe.src.hostvm = false; + else if (dc->debug.dml_hostvm_override == DML_HOSTVM_OVERRIDE_TRUE) + pipes[i].pipe.src.hostvm = true; + } + return pipe_cnt; +} + int dcn31_populate_dml_pipes_from_context( struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, @@ -1645,7 +1676,7 @@ int dcn31_populate_dml_pipes_from_context( bool upscaled = false; DC_FP_START(); - dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); DC_FP_END(); for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { @@ -1675,12 +1706,6 @@ int dcn31_populate_dml_pipes_from_context( dcn31_zero_pipe_dcc_fraction(pipes, pipe_cnt); DC_FP_END(); - if (dc->debug.dml_hostvm_override == DML_HOSTVM_NO_OVERRIDE) - pipes[pipe_cnt].pipe.src.hostvm = dc->res_pool->hubbub->riommu_active; - else if (dc->debug.dml_hostvm_override == DML_HOSTVM_OVERRIDE_FALSE) - pipes[pipe_cnt].pipe.src.hostvm = false; - else if (dc->debug.dml_hostvm_override == DML_HOSTVM_OVERRIDE_TRUE) - pipes[pipe_cnt].pipe.src.hostvm = true; if (pipes[pipe_cnt].dout.dsc_enable) { switch (timing->display_color_depth) { @@ -1898,6 +1923,8 @@ static bool dcn31_resource_construct( dc->caps.max_slave_rgb_planes = 2; dc->caps.post_blend_color_processing = true; dc->caps.force_dp_tps4_for_cp2520 = true; + if (dc->config.forceHBR2CP2520) + dc->caps.force_dp_tps4_for_cp2520 = false; dc->caps.dp_hpo = true; dc->caps.dp_hdmi21_pcon_support = true; dc->caps.edp_dsc_support = true; diff --git a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.c b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.c index 1bd7e0f327d8..367cb6e6d074 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.c @@ -97,7 +97,7 @@ static void dccg314_set_pixel_rate_div( enum pixel_rate_div cur_k1 = PIXEL_RATE_DIV_NA, cur_k2 = PIXEL_RATE_DIV_NA; dccg314_get_pixel_rate_div(dccg, otg_inst, &cur_k1, &cur_k2); - if (k1 == PIXEL_RATE_DIV_NA || k2 == PIXEL_RATE_DIV_NA || (k1 == cur_k1 && k2 == cur_k2)) + if (k1 == cur_k1 && k2 == cur_k2) return; switch (otg_inst) { diff --git a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dio_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dio_stream_encoder.c index 7e773bf7b895..38842f938bed 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dio_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dio_stream_encoder.c @@ -49,18 +49,30 @@ #define CTX \ enc1->base.ctx +static void enc314_reset_fifo(struct stream_encoder *enc, bool reset) +{ + struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc); + uint32_t reset_val = reset ? 1 : 0; + uint32_t is_symclk_on; + + REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_RESET, reset_val); + REG_GET(DIG_FE_CNTL, DIG_SYMCLK_FE_ON, &is_symclk_on); + + if (is_symclk_on) + REG_WAIT(DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, reset_val, 10, 5000); + else + udelay(10); +} static void enc314_enable_fifo(struct stream_encoder *enc) { struct dcn10_stream_encoder *enc1 = DCN10STRENC_FROM_STRENC(enc); - /* TODO: Confirm if we need to wait for DIG_SYMCLK_FE_ON */ - REG_WAIT(DIG_FE_CNTL, DIG_SYMCLK_FE_ON, 1, 10, 5000); REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_READ_START_LEVEL, 0x7); - REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_RESET, 1); - REG_WAIT(DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, 1, 10, 5000); - REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_RESET, 0); - REG_WAIT(DIG_FIFO_CTRL0, DIG_FIFO_RESET_DONE, 0, 10, 5000); + + enc314_reset_fifo(enc, true); + enc314_reset_fifo(enc, false); + REG_UPDATE(DIG_FIFO_CTRL0, DIG_FIFO_ENABLE, 1); } diff --git a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_optc.c b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_optc.c index 47eb162f1a75..f246aab23050 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_optc.c +++ b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_optc.c @@ -237,11 +237,10 @@ static struct timing_generator_funcs dcn314_tg_funcs = { .clear_optc_underflow = optc1_clear_optc_underflow, .setup_global_swap_lock = NULL, .get_crc = optc1_get_crc, - .configure_crc = optc2_configure_crc, + .configure_crc = optc1_configure_crc, .set_dsc_config = optc3_set_dsc_config, .get_dsc_status = optc2_get_dsc_status, .set_dwb_source = NULL, - .set_odm_bypass = optc3_set_odm_bypass, .set_odm_combine = optc314_set_odm_combine, .get_optc_source = optc2_get_optc_source, .set_out_mux = optc3_set_out_mux, diff --git a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_resource.c b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_resource.c index d0ad72caead2..4fffc7bb8088 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_resource.c @@ -184,6 +184,9 @@ enum dcn31_clk_src_array_id { .reg_name[id] = BASE(reg ## block ## id ## _ ## temp_name ## _BASE_IDX) + \ reg ## block ## id ## _ ## temp_name +#define SF_DWB2(reg_name, block, id, field_name, post_fix) \ + .field_name = reg_name ## __ ## field_name ## post_fix + #define DCCG_SRII(reg_name, block, id)\ .block ## _ ## reg_name[id] = BASE(reg ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ reg ## block ## id ## _ ## reg_name @@ -847,7 +850,7 @@ static const struct resource_caps res_cap_dcn314 = { .num_ddc = 5, .num_vmid = 16, .num_mpc_3dlut = 2, - .num_dsc = 3, + .num_dsc = 4, }; static const struct dc_plane_cap plane_cap = { @@ -937,6 +940,10 @@ static const struct dc_debug_options debug_defaults_diags = { }; static const struct dc_panel_config panel_config_defaults = { + .psr = { + .disable_psr = false, + .disallow_psrsu = false, + }, .ilr = { .optimize_edp_link_rate = true, }, @@ -1766,6 +1773,8 @@ static bool dcn314_resource_construct( dc->caps.max_slave_rgb_planes = 2; dc->caps.post_blend_color_processing = true; dc->caps.force_dp_tps4_for_cp2520 = true; + if (dc->config.forceHBR2CP2520) + dc->caps.force_dp_tps4_for_cp2520 = false; dc->caps.dp_hpo = true; dc->caps.dp_hdmi21_pcon_support = true; dc->caps.edp_dsc_support = true; diff --git a/drivers/gpu/drm/amd/display/dc/dcn315/dcn315_resource.c b/drivers/gpu/drm/amd/display/dc/dcn315/dcn315_resource.c index 58746c437554..7887078c5f64 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn315/dcn315_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn315/dcn315_resource.c @@ -151,8 +151,6 @@ enum dcn31_clk_src_array_id { */ /* DCN */ -/* TODO awful hack. fixup dcn20_dwb.h */ -#undef BASE_INNER #define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg #define BASE(seg) BASE_INNER(seg) @@ -185,6 +183,9 @@ enum dcn31_clk_src_array_id { .reg_name[id] = BASE(reg ## block ## id ## _ ## temp_name ## _BASE_IDX) + \ reg ## block ## id ## _ ## temp_name +#define SF_DWB2(reg_name, block, id, field_name, post_fix) \ + .field_name = reg_name ## __ ## field_name ## post_fix + #define DCCG_SRII(reg_name, block, id)\ .block ## _ ## reg_name[id] = BASE(reg ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ reg ## block ## id ## _ ## reg_name @@ -907,6 +908,10 @@ static const struct dc_debug_options debug_defaults_diags = { }; static const struct dc_panel_config panel_config_defaults = { + .psr = { + .disable_psr = false, + .disallow_psrsu = false, + }, .ilr = { .optimize_edp_link_rate = true, }, @@ -1623,6 +1628,7 @@ static struct clock_source *dcn31_clock_source_create( return &clk_src->base; } + kfree(clk_src); BREAK_TO_DEBUGGER(); return NULL; } @@ -1643,7 +1649,7 @@ static int dcn315_populate_dml_pipes_from_context( const int max_usable_det = context->bw_ctx.dml.ip.config_return_buffer_size_in_kbytes - DCN3_15_MIN_COMPBUF_SIZE_KB; DC_FP_START(); - dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); DC_FP_END(); for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { @@ -1662,7 +1668,6 @@ static int dcn315_populate_dml_pipes_from_context( pipes[pipe_cnt].pipe.src.immediate_flip = true; pipes[pipe_cnt].pipe.src.unbounded_req_mode = false; - pipes[pipe_cnt].pipe.src.gpuvm = true; pipes[pipe_cnt].pipe.dest.vfront_porch = timing->v_front_porch; pipes[pipe_cnt].pipe.src.dcc_rate = 3; pipes[pipe_cnt].dout.dsc_input_bpc = 0; @@ -1703,7 +1708,9 @@ static int dcn315_populate_dml_pipes_from_context( dc->config.enable_4to1MPC = true; context->bw_ctx.dml.ip.det_buffer_size_kbytes = (max_usable_det / DCN3_15_CRB_SEGMENT_SIZE_KB / 4) * DCN3_15_CRB_SEGMENT_SIZE_KB; - } else if (!is_dual_plane(pipe->plane_state->format) && pipe->plane_state->src_rect.width <= 5120) { + } else if (!is_dual_plane(pipe->plane_state->format) + && pipe->plane_state->src_rect.width <= 5120 + && pipe->stream->timing.pix_clk_100hz < dcn_get_max_non_odm_pix_rate_100hz(&dc->dml.soc)) { /* Limit to 5k max to avoid forced pipe split when there is not enough detile for swath */ context->bw_ctx.dml.ip.det_buffer_size_kbytes = 192; pipes[0].pipe.src.unbounded_req_mode = true; @@ -1779,6 +1786,8 @@ static bool dcn315_resource_construct( dc->caps.max_slave_rgb_planes = 2; dc->caps.post_blend_color_processing = true; dc->caps.force_dp_tps4_for_cp2520 = true; + if (dc->config.forceHBR2CP2520) + dc->caps.force_dp_tps4_for_cp2520 = false; dc->caps.dp_hpo = true; dc->caps.dp_hdmi21_pcon_support = true; dc->caps.edp_dsc_support = true; diff --git a/drivers/gpu/drm/amd/display/dc/dcn316/dcn316_resource.c b/drivers/gpu/drm/amd/display/dc/dcn316/dcn316_resource.c index 6b40a11ac83a..b4d5076e124c 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn316/dcn316_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn316/dcn316_resource.c @@ -142,8 +142,6 @@ enum dcn31_clk_src_array_id { */ /* DCN */ -/* TODO awful hack. fixup dcn20_dwb.h */ -#undef BASE_INNER #define BASE_INNER(seg) DCN_BASE__INST0_SEG ## seg #define BASE(seg) BASE_INNER(seg) @@ -176,6 +174,9 @@ enum dcn31_clk_src_array_id { .reg_name[id] = BASE(reg ## block ## id ## _ ## temp_name ## _BASE_IDX) + \ reg ## block ## id ## _ ## temp_name +#define SF_DWB2(reg_name, block, id, field_name, post_fix) \ + .field_name = reg_name ## __ ## field_name ## post_fix + #define DCCG_SRII(reg_name, block, id)\ .block ## _ ## reg_name[id] = BASE(reg ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ reg ## block ## id ## _ ## reg_name @@ -906,6 +907,10 @@ static const struct dc_debug_options debug_defaults_diags = { }; static const struct dc_panel_config panel_config_defaults = { + .psr = { + .disable_psr = false, + .disallow_psrsu = false, + }, .ilr = { .optimize_edp_link_rate = true, }, @@ -1646,7 +1651,7 @@ static int dcn316_populate_dml_pipes_from_context( const int max_usable_det = context->bw_ctx.dml.ip.config_return_buffer_size_in_kbytes - DCN3_16_MIN_COMPBUF_SIZE_KB; DC_FP_START(); - dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); DC_FP_END(); for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { @@ -1665,7 +1670,6 @@ static int dcn316_populate_dml_pipes_from_context( pipes[pipe_cnt].pipe.src.immediate_flip = true; pipes[pipe_cnt].pipe.src.unbounded_req_mode = false; - pipes[pipe_cnt].pipe.src.gpuvm = true; pipes[pipe_cnt].pipe.dest.vfront_porch = timing->v_front_porch; pipes[pipe_cnt].pipe.src.dcc_rate = 3; pipes[pipe_cnt].dout.dsc_input_bpc = 0; @@ -1781,6 +1785,8 @@ static bool dcn316_resource_construct( dc->caps.max_slave_rgb_planes = 2; dc->caps.post_blend_color_processing = true; dc->caps.force_dp_tps4_for_cp2520 = true; + if (dc->config.forceHBR2CP2520) + dc->caps.force_dp_tps4_for_cp2520 = false; dc->caps.dp_hpo = true; dc->caps.dp_hdmi21_pcon_support = true; dc->caps.edp_dsc_support = true; diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hubbub.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hubbub.c index 9fbb72369c10..5947c2cb0f30 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hubbub.c @@ -41,6 +41,10 @@ #define FN(reg_name, field_name) \ hubbub2->shifts->field_name, hubbub2->masks->field_name +/** + * @DCN32_CRB_SEGMENT_SIZE_KB: Maximum Configurable Return Buffer size for + * DCN32 + */ #define DCN32_CRB_SEGMENT_SIZE_KB 64 static void dcn32_init_crb(struct hubbub *hubbub) @@ -68,6 +72,23 @@ static void dcn32_init_crb(struct hubbub *hubbub) REG_UPDATE(DCHUBBUB_DEBUG_CTRL_0, DET_DEPTH, 0x47F); } +void hubbub32_set_request_limit(struct hubbub *hubbub, int memory_channel_count, int words_per_channel) +{ + struct dcn20_hubbub *hubbub2 = TO_DCN20_HUBBUB(hubbub); + + uint32_t request_limit = 3 * memory_channel_count * words_per_channel / 4; + + ASSERT((request_limit & (~0xFFF)) == 0); //field is only 24 bits long + ASSERT(request_limit > 0); //field is only 24 bits long + + if (request_limit > 0xFFF) + request_limit = 0xFFF; + + if (request_limit > 0) + REG_UPDATE(SDPIF_REQUEST_RATE_LIMIT, SDPIF_REQUEST_RATE_LIMIT, request_limit); +} + + void dcn32_program_det_size(struct hubbub *hubbub, int hubp_inst, unsigned int det_buffer_size_in_kbyte) { struct dcn20_hubbub *hubbub2 = TO_DCN20_HUBBUB(hubbub); @@ -945,6 +966,7 @@ static const struct hubbub_funcs hubbub32_funcs = { .init_crb = dcn32_init_crb, .hubbub_read_state = hubbub2_read_state, .force_usr_retraining_allow = hubbub32_force_usr_retraining_allow, + .set_request_limit = hubbub32_set_request_limit }; void hubbub32_construct(struct dcn20_hubbub *hubbub2, diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hubbub.h b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hubbub.h index cda94e0e31bf..786f9ce07f92 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hubbub.h +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hubbub.h @@ -82,7 +82,8 @@ SR(DCN_VM_FAULT_ADDR_MSB),\ SR(DCN_VM_FAULT_ADDR_LSB),\ SR(DCN_VM_FAULT_CNTL),\ - SR(DCN_VM_FAULT_STATUS) + SR(DCN_VM_FAULT_STATUS),\ + SR(SDPIF_REQUEST_RATE_LIMIT) #define HUBBUB_MASK_SH_LIST_DCN32(mask_sh)\ HUBBUB_SF(DCHUBBUB_GLOBAL_TIMER_CNTL, DCHUBBUB_GLOBAL_TIMER_ENABLE, mask_sh), \ @@ -159,7 +160,8 @@ HUBBUB_SF(DCN_VM_FAULT_STATUS, DCN_VM_ERROR_VMID, mask_sh), \ HUBBUB_SF(DCN_VM_FAULT_STATUS, DCN_VM_ERROR_TABLE_LEVEL, mask_sh), \ HUBBUB_SF(DCN_VM_FAULT_STATUS, DCN_VM_ERROR_PIPE, mask_sh), \ - HUBBUB_SF(DCN_VM_FAULT_STATUS, DCN_VM_ERROR_INTERRUPT_STATUS, mask_sh) + HUBBUB_SF(DCN_VM_FAULT_STATUS, DCN_VM_ERROR_INTERRUPT_STATUS, mask_sh),\ + HUBBUB_SF(SDPIF_REQUEST_RATE_LIMIT, SDPIF_REQUEST_RATE_LIMIT, mask_sh) bool hubbub32_program_urgent_watermarks( struct hubbub *hubbub, @@ -200,4 +202,6 @@ void hubbub32_construct(struct dcn20_hubbub *hubbub2, int pixel_chunk_size_kb, int config_return_buffer_size_kb); +void hubbub32_set_request_limit(struct hubbub *hubbub, int umc_count, int words_per_umc); + #endif diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c index cf5bd9713f54..763311ffb967 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c @@ -283,8 +283,7 @@ static uint32_t dcn32_calculate_cab_allocation(struct dc *dc, struct dc_state *c using the max for calculation */ if (hubp->curs_attr.width > 0) { - // Round cursor width to next multiple of 64 - cursor_size = (((hubp->curs_attr.width + 63) / 64) * 64) * hubp->curs_attr.height; + cursor_size = hubp->curs_attr.pitch * hubp->curs_attr.height; switch (pipe->stream->cursor_attributes.color_format) { case CURSOR_MODE_MONO: @@ -309,9 +308,9 @@ static uint32_t dcn32_calculate_cab_allocation(struct dc *dc, struct dc_state *c cursor_size > 16384) { /* cursor_num_mblk = CEILING(num_cursors*cursor_width*cursor_width*cursor_Bpe/mblk_bytes, 1) */ - cache_lines_used += (((hubp->curs_attr.width * hubp->curs_attr.height * cursor_bpp + - DCN3_2_MALL_MBLK_SIZE_BYTES - 1) / DCN3_2_MALL_MBLK_SIZE_BYTES) * - DCN3_2_MALL_MBLK_SIZE_BYTES) / dc->caps.cache_line_size + 2; + cache_lines_used += (((cursor_size + DCN3_2_MALL_MBLK_SIZE_BYTES - 1) / + DCN3_2_MALL_MBLK_SIZE_BYTES) * DCN3_2_MALL_MBLK_SIZE_BYTES) / + dc->caps.cache_line_size + 2; } break; } @@ -727,10 +726,7 @@ void dcn32_update_mall_sel(struct dc *dc, struct dc_state *context) struct hubp *hubp = pipe->plane_res.hubp; if (pipe->stream && pipe->plane_state && hubp && hubp->funcs->hubp_update_mall_sel) { - //Round cursor width up to next multiple of 64 - int cursor_width = ((hubp->curs_attr.width + 63) / 64) * 64; - int cursor_height = hubp->curs_attr.height; - int cursor_size = cursor_width * cursor_height; + int cursor_size = hubp->curs_attr.pitch * hubp->curs_attr.height; switch (hubp->curs_attr.color_format) { case CURSOR_MODE_MONO: @@ -984,15 +980,14 @@ void dcn32_init_hw(struct dc *dc) if (dc->res_pool->hubbub->funcs->init_crb) dc->res_pool->hubbub->funcs->init_crb(dc->res_pool->hubbub); + if (dc->res_pool->hubbub->funcs->set_request_limit && dc->config.sdpif_request_limit_words_per_umc > 0) + dc->res_pool->hubbub->funcs->set_request_limit(dc->res_pool->hubbub, dc->ctx->dc_bios->vram_info.num_chans, dc->config.sdpif_request_limit_words_per_umc); + // Get DMCUB capabilities if (dc->ctx->dmub_srv) { dc_dmub_srv_query_caps_cmd(dc->ctx->dmub_srv->dmub); dc->caps.dmub_caps.psr = dc->ctx->dmub_srv->dmub->feature_caps.psr; } - - /* Enable support for ODM and windowed MPO if policy flag is set */ - if (dc->debug.enable_single_display_2to1_odm_policy) - dc->config.enable_windowed_mpo_odm = true; } static int calc_mpc_flow_ctrl_cnt(const struct dc_stream_state *stream, @@ -1186,7 +1181,7 @@ unsigned int dcn32_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsign *k2_div = PIXEL_RATE_DIV_BY_2; else *k2_div = PIXEL_RATE_DIV_BY_4; - } else if (dc_is_dp_signal(pipe_ctx->stream->signal)) { + } else if (dc_is_dp_signal(pipe_ctx->stream->signal) || dc_is_virtual_signal(pipe_ctx->stream->signal)) { if (two_pix_per_container) { *k1_div = PIXEL_RATE_DIV_BY_1; *k2_div = PIXEL_RATE_DIV_BY_2; @@ -1371,6 +1366,33 @@ void dcn32_update_phantom_vp_position(struct dc *dc, } } +/* Treat the phantom pipe as if it needs to be fully enabled. + * If the pipe was previously in use but not phantom, it would + * have been disabled earlier in the sequence so we need to run + * the full enable sequence. + */ +void dcn32_apply_update_flags_for_phantom(struct pipe_ctx *phantom_pipe) +{ + phantom_pipe->update_flags.raw = 0; + if (phantom_pipe->stream && phantom_pipe->stream->mall_stream_config.type == SUBVP_PHANTOM) { + if (phantom_pipe->stream && phantom_pipe->plane_state) { + phantom_pipe->update_flags.bits.enable = 1; + phantom_pipe->update_flags.bits.mpcc = 1; + phantom_pipe->update_flags.bits.dppclk = 1; + phantom_pipe->update_flags.bits.hubp_interdependent = 1; + phantom_pipe->update_flags.bits.hubp_rq_dlg_ttu = 1; + phantom_pipe->update_flags.bits.gamut_remap = 1; + phantom_pipe->update_flags.bits.scaler = 1; + phantom_pipe->update_flags.bits.viewport = 1; + phantom_pipe->update_flags.bits.det_size = 1; + if (!phantom_pipe->top_pipe && !phantom_pipe->prev_odm_pipe) { + phantom_pipe->update_flags.bits.odm = 1; + phantom_pipe->update_flags.bits.global_sync = 1; + } + } + } +} + bool dcn32_dsc_pg_status( struct dce_hwseq *hws, unsigned int dsc_inst) diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.h b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.h index ac3657a5b9ea..7de36529cf99 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.h @@ -92,6 +92,8 @@ void dcn32_update_phantom_vp_position(struct dc *dc, struct dc_state *context, struct pipe_ctx *phantom_pipe); +void dcn32_apply_update_flags_for_phantom(struct pipe_ctx *phantom_pipe); + bool dcn32_dsc_pg_status( struct dce_hwseq *hws, unsigned int dsc_inst); diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_init.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_init.c index 45a949ba6f3f..dc4649458567 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_init.c +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_init.c @@ -110,6 +110,7 @@ static const struct hw_sequencer_funcs dcn32_funcs = { .update_visual_confirm_color = dcn20_update_visual_confirm_color, .update_phantom_vp_position = dcn32_update_phantom_vp_position, .update_dsc_pg = dcn32_update_dsc_pg, + .apply_update_flags_for_phantom = dcn32_apply_update_flags_for_phantom, }; static const struct hwseq_private_funcs dcn32_private_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_mmhubbub.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_mmhubbub.c index 41b0baf8e183..c3b089ba511a 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_mmhubbub.c +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_mmhubbub.c @@ -211,7 +211,7 @@ static void mmhubbub32_config_mcif_arb(struct mcif_wb *mcif_wb, REG_UPDATE(MCIF_WB_ARBITRATION_CONTROL, MCIF_WB_CLIENT_ARBITRATION_SLICE, params->arbitration_slice); } -const struct mcif_wb_funcs dcn32_mmhubbub_funcs = { +static const struct mcif_wb_funcs dcn32_mmhubbub_funcs = { .warmup_mcif = mmhubbub32_warmup_mcif, .enable_mcif = mmhubbub2_enable_mcif, .disable_mcif = mmhubbub2_disable_mcif, diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_mpc.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_mpc.c index 4edd0655965b..206a5ddbaf6d 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_mpc.c +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_mpc.c @@ -982,7 +982,7 @@ static bool mpc32_program_3dlut( return true; } -const struct mpc_funcs dcn32_mpc_funcs = { +static const struct mpc_funcs dcn32_mpc_funcs = { .read_mpcc_state = mpc1_read_mpcc_state, .insert_plane = mpc1_insert_plane, .remove_mpcc = mpc1_remove_mpcc, diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_optc.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_optc.c index 2b33eeb213e2..2ee798965bc2 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_optc.c +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_optc.c @@ -167,6 +167,13 @@ static void optc32_phantom_crtc_post_enable(struct timing_generator *optc) REG_WAIT(OTG_CLOCK_CONTROL, OTG_BUSY, 0, 1, 100000); } +static void optc32_disable_phantom_otg(struct timing_generator *optc) +{ + struct optc *optc1 = DCN10TG_FROM_TG(optc); + + REG_UPDATE(OTG_CONTROL, OTG_MASTER_EN, 0); +} + static void optc32_set_odm_bypass(struct timing_generator *optc, const struct dc_crtc_timing *dc_crtc_timing) { @@ -260,6 +267,7 @@ static struct timing_generator_funcs dcn32_tg_funcs = { .enable_crtc = optc32_enable_crtc, .disable_crtc = optc32_disable_crtc, .phantom_crtc_post_enable = optc32_phantom_crtc_post_enable, + .disable_phantom_crtc = optc32_disable_phantom_otg, /* used by enable_timing_synchronization. Not need for FPGA */ .is_counter_moving = optc1_is_counter_moving, .get_position = optc1_get_position, diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c index a88dd7b3d1c1..cdeff6de725d 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c @@ -106,8 +106,6 @@ enum dcn32_clk_src_array_id { */ /* DCN */ -/* TODO awful hack. fixup dcn20_dwb.h */ -#undef BASE_INNER #define BASE_INNER(seg) ctx->dcn_reg_offsets[seg] #define BASE(seg) BASE_INNER(seg) @@ -167,6 +165,9 @@ enum dcn32_clk_src_array_id { REG_STRUCT.reg_name[id] = BASE(reg ## block ## id ## _ ## temp_name ## _BASE_IDX) + \ reg ## block ## id ## _ ## temp_name +#define SF_DWB2(reg_name, block, id, field_name, post_fix) \ + .field_name = reg_name ## __ ## field_name ## post_fix + #define DCCG_SRII(reg_name, block, id)\ REG_STRUCT.block ## _ ## reg_name[id] = BASE(reg ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ reg ## block ## id ## _ ## reg_name @@ -724,6 +725,7 @@ static const struct dc_debug_options debug_defaults_drv = { .enable_dp_dig_pixel_rate_div_policy = 1, .allow_sw_cursor_fallback = false, .alloc_extra_way_for_cursor = true, + .min_prefetch_in_strobe_ns = 60000, // 60us }; static const struct dc_debug_options debug_defaults_diags = { @@ -829,6 +831,7 @@ static struct clock_source *dcn32_clock_source_create( return &clk_src->base; } + kfree(clk_src); BREAK_TO_DEBUGGER(); return NULL; } @@ -1678,7 +1681,7 @@ static void dcn32_enable_phantom_plane(struct dc *dc, /* Shadow pipe has small viewport. */ phantom_plane->clip_rect.y = 0; - phantom_plane->clip_rect.height = phantom_stream->timing.v_addressable; + phantom_plane->clip_rect.height = phantom_stream->src.height; phantom_plane->is_phantom = true; @@ -1918,8 +1921,9 @@ int dcn32_populate_dml_pipes_from_context( timing = &pipe->stream->timing; pipes[pipe_cnt].pipe.src.gpuvm = true; - pipes[pipe_cnt].pipe.src.dcc_fraction_of_zs_req_luma = 0; - pipes[pipe_cnt].pipe.src.dcc_fraction_of_zs_req_chroma = 0; + DC_FP_START(); + dcn32_zero_pipe_dcc_fraction(pipes, pipe_cnt); + DC_FP_END(); pipes[pipe_cnt].pipe.dest.vfront_porch = timing->v_front_porch; pipes[pipe_cnt].pipe.src.gpuvm_min_page_size_kbytes = 256; // according to spreadsheet pipes[pipe_cnt].pipe.src.unbounded_req_mode = false; @@ -2115,6 +2119,7 @@ static bool dcn32_resource_construct( dc->caps.cache_num_ways = 16; dc->caps.max_cab_allocation_bytes = 67108864; // 64MB = 1024 * 1024 * 64 dc->caps.subvp_fw_processing_delay_us = 15; + dc->caps.subvp_drr_max_vblank_margin_us = 40; dc->caps.subvp_prefetch_end_to_mall_start_us = 15; dc->caps.subvp_swath_height_margin_lines = 16; dc->caps.subvp_pstate_allow_width_us = 20; @@ -2125,6 +2130,8 @@ static bool dcn32_resource_construct( dc->caps.max_slave_rgb_planes = 2; dc->caps.post_blend_color_processing = true; dc->caps.force_dp_tps4_for_cp2520 = true; + if (dc->config.forceHBR2CP2520) + dc->caps.force_dp_tps4_for_cp2520 = false; dc->caps.dp_hpo = true; dc->caps.dp_hdmi21_pcon_support = true; dc->caps.edp_dsc_support = true; @@ -2408,6 +2415,9 @@ static bool dcn32_resource_construct( pool->base.oem_device = NULL; } + if (ASICREV_IS_GC_11_0_3(dc->ctx->asic_id.hw_internal_rev) && (dc->config.sdpif_request_limit_words_per_umc == 0)) + dc->config.sdpif_request_limit_words_per_umc = 16; + DC_FP_END(); return true; diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.h b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.h index f76120e67c16..f6bc9bd5da31 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.h +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.h @@ -1244,7 +1244,8 @@ void dcn32_restore_mall_state(struct dc *dc, SR(DCHUBBUB_ARB_FCLK_PSTATE_CHANGE_WATERMARK_C), \ SR(DCHUBBUB_ARB_FCLK_PSTATE_CHANGE_WATERMARK_D), \ SR(DCN_VM_FAULT_ADDR_MSB), SR(DCN_VM_FAULT_ADDR_LSB), \ - SR(DCN_VM_FAULT_CNTL), SR(DCN_VM_FAULT_STATUS) \ + SR(DCN_VM_FAULT_CNTL), SR(DCN_VM_FAULT_STATUS), \ + SR(SDPIF_REQUEST_RATE_LIMIT) \ ) /* DCCG */ diff --git a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource_helpers.c b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource_helpers.c index d51d0c40ae5b..b03a7814e96d 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource_helpers.c +++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource_helpers.c @@ -200,7 +200,7 @@ bool dcn32_all_pipes_have_stream_and_plane(struct dc *dc, struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; if (!pipe->stream) - return false; + continue; if (!pipe->plane_state) return false; diff --git a/drivers/gpu/drm/amd/display/dc/dcn321/dcn321_resource.c b/drivers/gpu/drm/amd/display/dc/dcn321/dcn321_resource.c index 61087f2385a9..6c79a47b6336 100644 --- a/drivers/gpu/drm/amd/display/dc/dcn321/dcn321_resource.c +++ b/drivers/gpu/drm/amd/display/dc/dcn321/dcn321_resource.c @@ -109,8 +109,6 @@ enum dcn321_clk_src_array_id { */ /* DCN */ -/* TODO awful hack. fixup dcn20_dwb.h */ -#undef BASE_INNER #define BASE_INNER(seg) ctx->dcn_reg_offsets[seg] #define BASE(seg) BASE_INNER(seg) @@ -174,6 +172,9 @@ enum dcn321_clk_src_array_id { REG_STRUCT.block ## _ ## reg_name[id] = BASE(reg ## block ## id ## _ ## reg_name ## _BASE_IDX) + \ reg ## block ## id ## _ ## reg_name +#define SF_DWB2(reg_name, block, id, field_name, post_fix) \ + .field_name = reg_name ## __ ## field_name ## post_fix + #define VUPDATE_SRII(reg_name, block, id)\ REG_STRUCT.reg_name[id] = BASE(reg ## reg_name ## _ ## block ## id ## _BASE_IDX) + \ reg ## reg_name ## _ ## block ## id @@ -722,6 +723,7 @@ static const struct dc_debug_options debug_defaults_drv = { .enable_dp_dig_pixel_rate_div_policy = 1, .allow_sw_cursor_fallback = false, .alloc_extra_way_for_cursor = true, + .min_prefetch_in_strobe_ns = 60000, // 60us }; static const struct dc_debug_options debug_defaults_diags = { @@ -828,6 +830,7 @@ static struct clock_source *dcn321_clock_source_create( return &clk_src->base; } + kfree(clk_src); BREAK_TO_DEBUGGER(); return NULL; } @@ -1703,6 +1706,7 @@ static bool dcn321_resource_construct( dc->caps.cache_num_ways = 16; dc->caps.max_cab_allocation_bytes = 33554432; // 32MB = 1024 * 1024 * 32 dc->caps.subvp_fw_processing_delay_us = 15; + dc->caps.subvp_drr_max_vblank_margin_us = 40; dc->caps.subvp_prefetch_end_to_mall_start_us = 15; dc->caps.subvp_swath_height_margin_lines = 16; dc->caps.subvp_pstate_allow_width_us = 20; diff --git a/drivers/gpu/drm/amd/display/dc/dm_helpers.h b/drivers/gpu/drm/amd/display/dc/dm_helpers.h index e3e5c39895a3..af1c50ed905a 100644 --- a/drivers/gpu/drm/amd/display/dc/dm_helpers.h +++ b/drivers/gpu/drm/amd/display/dc/dm_helpers.h @@ -116,6 +116,11 @@ bool dm_helpers_dp_mst_start_top_mgr( bool dm_helpers_dp_mst_stop_top_mgr( struct dc_context *ctx, struct dc_link *link); + +void dm_helpers_dp_mst_update_branch_bandwidth( + struct dc_context *ctx, + struct dc_link *link); + /** * OS specific aux read callback. */ diff --git a/drivers/gpu/drm/amd/display/dc/dml/Makefile b/drivers/gpu/drm/amd/display/dc/dml/Makefile index ca7d24000621..0ecea87cf48f 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/Makefile +++ b/drivers/gpu/drm/amd/display/dc/dml/Makefile @@ -33,6 +33,10 @@ ifdef CONFIG_PPC64 dml_ccflags := -mhard-float -maltivec endif +ifdef CONFIG_ARM64 +dml_rcflags := -mgeneral-regs-only +endif + ifdef CONFIG_CC_IS_GCC ifneq ($(call gcc-min-version, 70100),y) IS_OLD_GCC = 1 @@ -55,8 +59,6 @@ frame_warn_flag := -Wframe-larger-than=2048 endif CFLAGS_$(AMDDALPATH)/dc/dml/display_mode_lib.o := $(dml_ccflags) - -ifdef CONFIG_DRM_AMD_DC_DCN CFLAGS_$(AMDDALPATH)/dc/dml/display_mode_vba.o := $(dml_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml/dcn10/dcn10_fpu.o := $(dml_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml/dcn20/dcn20_fpu.o := $(dml_ccflags) @@ -88,7 +90,6 @@ CFLAGS_$(AMDDALPATH)/dc/dml/calcs/dcn_calcs.o := $(dml_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml/calcs/dcn_calc_auto.o := $(dml_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml/calcs/dcn_calc_math.o := $(dml_ccflags) -Wno-tautological-compare CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/display_mode_vba.o := $(dml_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn2x/dcn2x.o := $(dml_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/display_mode_vba_20.o := $(dml_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/display_rq_dlg_calc_20.o := $(dml_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/display_mode_vba_20v2.o := $(dml_rcflags) @@ -105,7 +106,18 @@ CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn32/display_mode_vba_util_32.o := $(dml_rcf CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn301/dcn301_fpu.o := $(dml_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/display_mode_lib.o := $(dml_rcflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dsc/rc_calc_fpu.o := $(dml_rcflags) -endif +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn10/dcn10_fpu.o := $(dml_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn20/dcn20_fpu.o := $(dml_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn314/display_mode_vba_314.o := $(dml_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn314/display_rq_dlg_calc_314.o := $(dml_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn314/dcn314_fpu.o := $(dml_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn30/dcn30_fpu.o := $(dml_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn32/dcn32_fpu.o := $(dml_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn321/dcn321_fpu.o := $(dml_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn31/dcn31_fpu.o := $(dml_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn302/dcn302_fpu.o := $(dml_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dcn303/dcn303_fpu.o := $(dml_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/calcs/dcn_calc_math.o := $(dml_rcflags) CFLAGS_$(AMDDALPATH)/dc/dml/dml1_display_rq_dlg_calc.o := $(dml_ccflags) CFLAGS_$(AMDDALPATH)/dc/dml/display_rq_dlg_helpers.o := $(dml_ccflags) CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml/dml1_display_rq_dlg_calc.o := $(dml_rcflags) diff --git a/drivers/gpu/drm/amd/display/dc/dml/dc_features.h b/drivers/gpu/drm/amd/display/dc/dml/dc_features.h index 74e86732e301..2cbdd75429ff 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dc_features.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dc_features.h @@ -29,6 +29,13 @@ #define DC__PRESENT 1 #define DC__PRESENT__1 1 #define DC__NUM_DPP 4 + +/** + * @DC__VOLTAGE_STATES: + * + * Define the maximum amount of states supported by the ASIC. Every ASIC has a + * specific number of states; this macro defines the maximum number of states. + */ #define DC__VOLTAGE_STATES 20 #define DC__NUM_DPP__4 1 #define DC__NUM_DPP__0_PRESENT 1 diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.c index 99644d896222..c5e84190c17a 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.c @@ -27,6 +27,8 @@ #include "dcn10/dcn10_resource.h" #include "dcn10_fpu.h" +#include "resource.h" +#include "amdgpu_dm/dc_fpu.h" /** * DOC: DCN10 FPU manipulation Overview @@ -121,3 +123,37 @@ struct _vcs_dpi_soc_bounding_box_st dcn1_0_soc = { .writeback_dram_clock_change_latency_us = 23.0, .return_bus_width_bytes = 64, }; + +void dcn10_resource_construct_fp(struct dc *dc) +{ + dc_assert_fp_enabled(); + if (dc->ctx->dce_version == DCN_VERSION_1_01) { + struct dcn_soc_bounding_box *dcn_soc = dc->dcn_soc; + struct dcn_ip_params *dcn_ip = dc->dcn_ip; + struct display_mode_lib *dml = &dc->dml; + + dml->ip.max_num_dpp = 3; + /* TODO how to handle 23.84? */ + dcn_soc->dram_clock_change_latency = 23; + dcn_ip->max_num_dpp = 3; + } + if (ASICREV_IS_RV1_F0(dc->ctx->asic_id.hw_internal_rev)) { + dc->dcn_soc->urgent_latency = 3; + dc->debug.disable_dmcu = true; + dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 41.60f; + } + + dc->dcn_soc->number_of_channels = dc->ctx->asic_id.vram_width / ddr4_dram_width; + ASSERT(dc->dcn_soc->number_of_channels < 3); + if (dc->dcn_soc->number_of_channels == 0)/*old sbios bug*/ + dc->dcn_soc->number_of_channels = 2; + + if (dc->dcn_soc->number_of_channels == 1) { + dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 19.2f; + dc->dcn_soc->fabric_and_dram_bandwidth_vnom0p8 = 17.066f; + dc->dcn_soc->fabric_and_dram_bandwidth_vmid0p72 = 14.933f; + dc->dcn_soc->fabric_and_dram_bandwidth_vmin0p65 = 12.8f; + if (ASICREV_IS_RV1_F0(dc->ctx->asic_id.hw_internal_rev)) + dc->dcn_soc->fabric_and_dram_bandwidth_vmax0p9 = 20.80f; + } +} diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.h index e74ed4b4ce5b..63219ecd8478 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn10/dcn10_fpu.h @@ -27,4 +27,6 @@ #ifndef __DCN10_FPU_H__ #define __DCN10_FPU_H__ +void dcn10_resource_construct_fp(struct dc *dc); + #endif /* __DCN20_FPU_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c index d680f1c5b69f..75dbb7ee193b 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c @@ -989,7 +989,7 @@ static enum dcn_zstate_support_state decide_zstate_support(struct dc *dc, struc if (context->bw_ctx.dml.vba.StutterPeriod > 5000.0 || optimized_min_dst_y_next_start_us > 5000) return DCN_ZSTATE_SUPPORT_ALLOW; - else if (link->psr_settings.psr_version == DC_PSR_VERSION_1 && !dc->debug.disable_psr) + else if (link->psr_settings.psr_version == DC_PSR_VERSION_1 && !link->panel_config.psr.disable_psr) return DCN_ZSTATE_SUPPORT_ALLOW_Z10_ONLY; else return DCN_ZSTATE_SUPPORT_DISALLOW; @@ -1228,6 +1228,7 @@ int dcn20_populate_dml_pipes_from_context( pipes[pipe_cnt].pipe.src.dcc = false; pipes[pipe_cnt].pipe.src.dcc_rate = 1; pipes[pipe_cnt].pipe.dest.synchronized_vblank_all_planes = synchronized_vblank; + pipes[pipe_cnt].pipe.dest.synchronize_timings = synchronized_vblank; pipes[pipe_cnt].pipe.dest.hblank_start = timing->h_total - timing->h_front_porch; pipes[pipe_cnt].pipe.dest.hblank_end = pipes[pipe_cnt].pipe.dest.hblank_start - timing->h_addressable @@ -1295,6 +1296,8 @@ int dcn20_populate_dml_pipes_from_context( case SIGNAL_TYPE_DISPLAY_PORT_MST: case SIGNAL_TYPE_DISPLAY_PORT: pipes[pipe_cnt].dout.output_type = dm_dp; + if (is_dp_128b_132b_signal(&res_ctx->pipe_ctx[i])) + pipes[pipe_cnt].dout.output_type = dm_dp2p0; break; case SIGNAL_TYPE_EDP: pipes[pipe_cnt].dout.output_type = dm_edp; diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn30/dcn30_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn30/dcn30_fpu.c index e1e92daba668..d4c0f9cdac8e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn30/dcn30_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn30/dcn30_fpu.c @@ -520,9 +520,7 @@ void dcn30_fpu_calculate_wm_and_dlg( pipe_idx++; } - DC_FP_START(); dcn20_calculate_dlg_params(dc, context, pipes, pipe_cnt, vlevel); - DC_FP_END(); if (!pstate_en) /* Restore full p-state latency */ diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.c index 7dd0845d1bd9..12b23bd50e19 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.c @@ -483,7 +483,7 @@ void dcn31_calculate_wm_and_dlg_fp( int pipe_cnt, int vlevel) { - int i, pipe_idx, active_dpp_count = 0; + int i, pipe_idx, active_hubp_count = 0; double dcfclk = context->bw_ctx.dml.vba.DCFCLKState[vlevel][context->bw_ctx.dml.vba.maxMpcComb]; dc_assert_fp_enabled(); @@ -529,7 +529,7 @@ void dcn31_calculate_wm_and_dlg_fp( continue; if (context->res_ctx.pipe_ctx[i].plane_state) - active_dpp_count++; + active_hubp_count++; pipes[pipe_idx].clks_cfg.dispclk_mhz = get_dispclk_calculated(&context->bw_ctx.dml, pipes, pipe_cnt); pipes[pipe_idx].clks_cfg.dppclk_mhz = get_dppclk_calculated(&context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx); @@ -547,9 +547,19 @@ void dcn31_calculate_wm_and_dlg_fp( } dcn20_calculate_dlg_params(dc, context, pipes, pipe_cnt, vlevel); - /* For 31x apu pstate change is only supported if possible in vactive or if there are no active dpps */ + /* For 31x apu pstate change is only supported if possible in vactive*/ context->bw_ctx.bw.dcn.clk.p_state_change_support = - context->bw_ctx.dml.vba.DRAMClockChangeSupport[vlevel][context->bw_ctx.dml.vba.maxMpcComb] == dm_dram_clock_change_vactive || !active_dpp_count; + context->bw_ctx.dml.vba.DRAMClockChangeSupport[vlevel][context->bw_ctx.dml.vba.maxMpcComb] == dm_dram_clock_change_vactive; + /* If DCN isn't making memory requests we can allow pstate change and lower clocks */ + if (!active_hubp_count) { + context->bw_ctx.bw.dcn.clk.socclk_khz = 0; + context->bw_ctx.bw.dcn.clk.dppclk_khz = 0; + context->bw_ctx.bw.dcn.clk.dcfclk_khz = 0; + context->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz = 0; + context->bw_ctx.bw.dcn.clk.dramclk_khz = 0; + context->bw_ctx.bw.dcn.clk.fclk_khz = 0; + context->bw_ctx.bw.dcn.clk.p_state_change_support = true; + } } void dcn31_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params) @@ -797,3 +807,8 @@ void dcn316_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_param else dml_init_instance(&dc->dml, &dcn3_16_soc, &dcn3_16_ip, DML_PROJECT_DCN31_FPGA); } + +int dcn_get_max_non_odm_pix_rate_100hz(struct _vcs_dpi_soc_bounding_box_st *soc) +{ + return soc->clock_limits[0].dispclk_mhz * 10000.0 / (1.0 + soc->dcn_downspread_percent / 100.0); +} diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h index fd58b2561ec9..687d3522cc33 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn31/dcn31_fpu.h @@ -46,5 +46,10 @@ void dcn31_calculate_wm_and_dlg_fp( void dcn31_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params); void dcn315_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params); void dcn316_update_bw_bounding_box(struct dc *dc, struct clk_bw_params *bw_params); +int dcn_get_max_non_odm_pix_rate_100hz(struct _vcs_dpi_soc_bounding_box_st *soc); +int dcn31x_populate_dml_pipes_from_context(struct dc *dc, + struct dc_state *context, + display_e2e_pipe_params_st *pipes, + bool fast_validate); #endif /* __DCN31_FPU_H__*/ diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c b/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c index b612edb14417..4e45c6d9ecdc 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn31/display_mode_vba_31.c @@ -1052,17 +1052,16 @@ static bool CalculatePrefetchSchedule( else bytes_pp = myPipe->BytePerPixelY + myPipe->BytePerPixelC; /*rev 99*/ - prefetch_bw_pr = dml_min(1, bytes_pp * myPipe->PixelClock / (double) myPipe->DPPPerPlane); + prefetch_bw_pr = bytes_pp * myPipe->PixelClock / (double) myPipe->DPPPerPlane; + prefetch_bw_pr = dml_min(1, myPipe->VRatio) * prefetch_bw_pr; max_Tsw = dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) * LineTime; prefetch_sw_bytes = PrefetchSourceLinesY * swath_width_luma_ub * myPipe->BytePerPixelY + PrefetchSourceLinesC * swath_width_chroma_ub * myPipe->BytePerPixelC; - prefetch_bw_oto = dml_max(bytes_pp * myPipe->PixelClock / myPipe->DPPPerPlane, prefetch_sw_bytes / (dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) * LineTime)); prefetch_bw_oto = dml_max(prefetch_bw_pr, prefetch_sw_bytes / max_Tsw); min_Lsw = dml_max(1, dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) / max_vratio_pre); Lsw_oto = dml_ceil(4 * dml_max(prefetch_sw_bytes / prefetch_bw_oto / LineTime, min_Lsw), 1) / 4; Tsw_oto = Lsw_oto * LineTime; - prefetch_bw_oto = (PrefetchSourceLinesY * swath_width_luma_ub * myPipe->BytePerPixelY + PrefetchSourceLinesC * swath_width_chroma_ub * myPipe->BytePerPixelC) / Tsw_oto; #ifdef __DML_VBA_DEBUG__ dml_print("DML: HTotal: %d\n", myPipe->HTotal); @@ -5361,6 +5360,58 @@ void dml31_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l v->ModeSupport[i][j] = true; } else { v->ModeSupport[i][j] = false; +#ifdef __DML_VBA_DEBUG__ + if (v->ScaleRatioAndTapsSupport == false) + dml_print("DML SUPPORT: ScaleRatioAndTapsSupport failed"); + if (v->SourceFormatPixelAndScanSupport == false) + dml_print("DML SUPPORT: SourceFormatPixelAndScanSupport failed"); + if (v->ViewportSizeSupport[i][j] == false) + dml_print("DML SUPPORT: ViewportSizeSupport failed"); + if (v->LinkCapacitySupport[i] == false) + dml_print("DML SUPPORT: LinkCapacitySupport failed"); + if (v->ODMCombine4To1SupportCheckOK[i] == false) + dml_print("DML SUPPORT: DSC422NativeNotSupported failed"); + if (v->NotEnoughDSCUnits[i] == true) + dml_print("DML SUPPORT: NotEnoughDSCUnits"); + if (v->DTBCLKRequiredMoreThanSupported[i] == true) + dml_print("DML SUPPORT: DTBCLKRequiredMoreThanSupported"); + if (v->ROBSupport[i][j] == false) + dml_print("DML SUPPORT: ROBSupport failed"); + if (v->DISPCLK_DPPCLK_Support[i][j] == false) + dml_print("DML SUPPORT: DISPCLK_DPPCLK_Support failed"); + if (v->TotalAvailablePipesSupport[i][j] == false) + dml_print("DML SUPPORT: DSC422NativeNotSupported failed"); + if (EnoughWritebackUnits == false) + dml_print("DML SUPPORT: DSC422NativeNotSupported failed"); + if (v->WritebackLatencySupport == false) + dml_print("DML SUPPORT: WritebackLatencySupport failed"); + if (v->WritebackScaleRatioAndTapsSupport == false) + dml_print("DML SUPPORT: DSC422NativeNotSupported "); + if (v->CursorSupport == false) + dml_print("DML SUPPORT: DSC422NativeNotSupported failed"); + if (v->PitchSupport == false) + dml_print("DML SUPPORT: PitchSupport failed"); + if (ViewportExceedsSurface == true) + dml_print("DML SUPPORT: ViewportExceedsSurface failed"); + if (v->PrefetchSupported[i][j] == false) + dml_print("DML SUPPORT: PrefetchSupported failed"); + if (v->DynamicMetadataSupported[i][j] == false) + dml_print("DML SUPPORT: DSC422NativeNotSupported failed"); + if (v->TotalVerticalActiveBandwidthSupport[i][j] == false) + dml_print("DML SUPPORT: TotalVerticalActiveBandwidthSupport failed"); + if (v->VRatioInPrefetchSupported[i][j] == false) + dml_print("DML SUPPORT: VRatioInPrefetchSupported failed"); + if (v->PTEBufferSizeNotExceeded[i][j] == false) + dml_print("DML SUPPORT: PTEBufferSizeNotExceeded failed"); + if (v->NonsupportedDSCInputBPC == true) + dml_print("DML SUPPORT: NonsupportedDSCInputBPC failed"); + if (!((v->HostVMEnable == false + && v->ImmediateFlipRequirement[0] != dm_immediate_flip_required) + || v->ImmediateFlipSupportedForState[i][j] == true)) + dml_print("DML SUPPORT: ImmediateFlipRequirement failed"); + if (FMTBufferExceeded == true) + dml_print("DML SUPPORT: FMTBufferExceeded failed"); +#endif } } } diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c index cf420ad2b8dc..1dd51c4b6804 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn314/dcn314_fpu.c @@ -29,6 +29,7 @@ #include "dcn31/dcn31_hubbub.h" #include "dcn314_fpu.h" #include "dml/dcn20/dcn20_fpu.h" +#include "dml/dcn31/dcn31_fpu.h" #include "dml/display_mode_vba.h" struct _vcs_dpi_ip_params_st dcn3_14_ip = { @@ -146,8 +147,8 @@ struct _vcs_dpi_soc_bounding_box_st dcn3_14_soc = { }, }, .num_states = 5, - .sr_exit_time_us = 9.0, - .sr_enter_plus_exit_time_us = 11.0, + .sr_exit_time_us = 16.5, + .sr_enter_plus_exit_time_us = 18.5, .sr_exit_z8_time_us = 442.0, .sr_enter_plus_exit_z8_time_us = 560.0, .writeback_latency_us = 12.0, @@ -264,11 +265,8 @@ void dcn314_update_bw_bounding_box_fpu(struct dc *dc, struct clk_bw_params *bw_p dc->dml.soc.dispclk_dppclk_vco_speed_mhz = max_dispclk_mhz * 2; } - if ((int)(dcn3_14_soc.dram_clock_change_latency_us * 1000) - != dc->debug.dram_clock_change_latency_ns - && dc->debug.dram_clock_change_latency_ns) { - dcn3_14_soc.dram_clock_change_latency_us = dc->debug.dram_clock_change_latency_ns / 1000; - } + dcn20_patch_bounding_box(dc, &dcn3_14_soc); + if (!IS_FPGA_MAXIMUS_DC(dc->ctx->dce_environment)) dml_init_instance(&dc->dml, &dcn3_14_soc, &dcn3_14_ip, DML_PROJECT_DCN314); else @@ -291,7 +289,7 @@ int dcn314_populate_dml_pipes_from_context_fpu(struct dc *dc, struct dc_state *c dc_assert_fp_enabled(); - dcn20_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); + dcn31x_populate_dml_pipes_from_context(dc, context, pipes, fast_validate); for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) { struct dc_crtc_timing *timing; @@ -318,8 +316,6 @@ int dcn314_populate_dml_pipes_from_context_fpu(struct dc *dc, struct dc_state *c pipes[pipe_cnt].pipe.src.immediate_flip = true; pipes[pipe_cnt].pipe.src.unbounded_req_mode = false; - pipes[pipe_cnt].pipe.src.hostvm = dc->res_pool->hubbub->riommu_active; - pipes[pipe_cnt].pipe.src.gpuvm = true; pipes[pipe_cnt].pipe.src.dcc_fraction_of_zs_req_luma = 0; pipes[pipe_cnt].pipe.src.dcc_fraction_of_zs_req_chroma = 0; pipes[pipe_cnt].pipe.dest.vfront_porch = timing->v_front_porch; diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c b/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c index 0d12fd079cd6..41f0b4c1c72f 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c @@ -1074,17 +1074,16 @@ static bool CalculatePrefetchSchedule( else bytes_pp = myPipe->BytePerPixelY + myPipe->BytePerPixelC; /*rev 99*/ - prefetch_bw_pr = dml_min(1, bytes_pp * myPipe->PixelClock / (double) myPipe->DPPPerPlane); + prefetch_bw_pr = bytes_pp * myPipe->PixelClock / (double) myPipe->DPPPerPlane; + prefetch_bw_pr = dml_min(1, myPipe->VRatio) * prefetch_bw_pr; max_Tsw = dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) * LineTime; prefetch_sw_bytes = PrefetchSourceLinesY * swath_width_luma_ub * myPipe->BytePerPixelY + PrefetchSourceLinesC * swath_width_chroma_ub * myPipe->BytePerPixelC; - prefetch_bw_oto = dml_max(bytes_pp * myPipe->PixelClock / myPipe->DPPPerPlane, prefetch_sw_bytes / (dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) * LineTime)); prefetch_bw_oto = dml_max(prefetch_bw_pr, prefetch_sw_bytes / max_Tsw); min_Lsw = dml_max(1, dml_max(PrefetchSourceLinesY, PrefetchSourceLinesC) / max_vratio_pre); Lsw_oto = dml_ceil(4 * dml_max(prefetch_sw_bytes / prefetch_bw_oto / LineTime, min_Lsw), 1) / 4; Tsw_oto = Lsw_oto * LineTime; - prefetch_bw_oto = (PrefetchSourceLinesY * swath_width_luma_ub * myPipe->BytePerPixelY + PrefetchSourceLinesC * swath_width_chroma_ub * myPipe->BytePerPixelC) / Tsw_oto; #ifdef __DML_VBA_DEBUG__ dml_print("DML: HTotal: %d\n", myPipe->HTotal); diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c index 819de0f11012..97b333b230d1 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c @@ -531,9 +531,12 @@ void dcn32_set_phantom_stream_timing(struct dc *dc, unsigned int i, pipe_idx; struct pipe_ctx *pipe; uint32_t phantom_vactive, phantom_bp, pstate_width_fw_delay_lines; + unsigned int num_dpp; unsigned int vlevel = context->bw_ctx.dml.vba.VoltageLevel; unsigned int dcfclk = context->bw_ctx.dml.vba.DCFCLKState[vlevel][context->bw_ctx.dml.vba.maxMpcComb]; unsigned int socclk = context->bw_ctx.dml.vba.SOCCLKPerState[vlevel]; + struct vba_vars_st *vba = &context->bw_ctx.dml.vba; + struct dc_stream_state *main_stream = ref_pipe->stream; dc_assert_fp_enabled(); @@ -569,13 +572,23 @@ void dcn32_set_phantom_stream_timing(struct dc *dc, phantom_vactive = get_subviewport_lines_needed_in_mall(&context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx) + pstate_width_fw_delay_lines + dc->caps.subvp_swath_height_margin_lines; + // W/A for DCC corruption with certain high resolution timings. + // Determing if pipesplit is used. If so, add meta_row_height to the phantom vactive. + num_dpp = vba->NoOfDPP[vba->VoltageLevel][vba->maxMpcComb][vba->pipe_plane[pipe_idx]]; + phantom_vactive += num_dpp > 1 ? vba->meta_row_height[vba->pipe_plane[pipe_idx]] : 0; + // For backporch of phantom pipe, use vstartup of the main pipe phantom_bp = get_vstartup(&context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx); phantom_stream->dst.y = 0; phantom_stream->dst.height = phantom_vactive; + /* When scaling, DML provides the end to end required number of lines for MALL. + * dst.height is always correct for this case, but src.height is not which causes a + * delta between main and phantom pipe scaling outputs. Need to adjust src.height on + * phantom for this case. + */ phantom_stream->src.y = 0; - phantom_stream->src.height = phantom_vactive; + phantom_stream->src.height = (double)phantom_vactive * (double)main_stream->src.height / (double)main_stream->dst.height; phantom_stream->timing.v_addressable = phantom_vactive; phantom_stream->timing.v_front_porch = 1; @@ -1191,9 +1204,7 @@ static void dcn32_full_validate_bw_helper(struct dc *dc, } } else { // Most populate phantom DLG params before programming hardware / timing for phantom pipe - DC_FP_START(); dcn32_helper_populate_phantom_dlg_params(dc, context, pipes, *pipe_cnt); - DC_FP_END(); /* Call validate_apply_pipe_split flags after calling DML getters for * phantom dlg params, or some of the VBA params indicating pipe split @@ -1230,7 +1241,7 @@ static void dcn32_calculate_dlg_params(struct dc *dc, struct dc_state *context, display_e2e_pipe_params_st *pipes, int pipe_cnt, int vlevel) { - int i, pipe_idx; + int i, pipe_idx, active_hubp_count = 0; bool usr_retraining_support = false; bool unbounded_req_enabled = false; @@ -1275,6 +1286,8 @@ static void dcn32_calculate_dlg_params(struct dc *dc, struct dc_state *context, for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { if (!context->res_ctx.pipe_ctx[i].stream) continue; + if (context->res_ctx.pipe_ctx[i].plane_state) + active_hubp_count++; pipes[pipe_idx].pipe.dest.vstartup_start = get_vstartup(&context->bw_ctx.dml, pipes, pipe_cnt, pipe_idx); pipes[pipe_idx].pipe.dest.vupdate_offset = get_vupdate_offset(&context->bw_ctx.dml, pipes, pipe_cnt, @@ -1300,6 +1313,16 @@ static void dcn32_calculate_dlg_params(struct dc *dc, struct dc_state *context, context->res_ctx.pipe_ctx[i].pipe_dlg_param = pipes[pipe_idx].pipe.dest; pipe_idx++; } + /* If DCN isn't making memory requests we can allow pstate change and lower clocks */ + if (!active_hubp_count) { + context->bw_ctx.bw.dcn.clk.socclk_khz = 0; + context->bw_ctx.bw.dcn.clk.dppclk_khz = 0; + context->bw_ctx.bw.dcn.clk.dcfclk_khz = 0; + context->bw_ctx.bw.dcn.clk.dcfclk_deep_sleep_khz = 0; + context->bw_ctx.bw.dcn.clk.dramclk_khz = 0; + context->bw_ctx.bw.dcn.clk.fclk_khz = 0; + context->bw_ctx.bw.dcn.clk.p_state_change_support = true; + } /*save a original dppclock copy*/ context->bw_ctx.bw.dcn.clk.bw_dppclk_khz = context->bw_ctx.bw.dcn.clk.dppclk_khz; context->bw_ctx.bw.dcn.clk.bw_dispclk_khz = context->bw_ctx.bw.dcn.clk.dispclk_khz; @@ -1494,11 +1517,8 @@ bool dcn32_internal_validate_bw(struct dc *dc, dml_log_pipe_params(&context->bw_ctx.dml, pipes, pipe_cnt); - if (!fast_validate) { - DC_FP_START(); + if (!fast_validate) dcn32_full_validate_bw_helper(dc, context, pipes, &vlevel, split, merge, &pipe_cnt); - DC_FP_END(); - } if (fast_validate || (dc->debug.dml_disallow_alternate_prefetch_modes && @@ -1734,6 +1754,9 @@ bool dcn32_internal_validate_bw(struct dc *dc, } if (repopulate_pipes) { + int flag_max_mpc_comb = vba->maxMpcComb; + int flag_vlevel = vlevel; + pipe_cnt = dc->res_pool->funcs->populate_dml_pipes(dc, context, pipes, fast_validate); /* repopulate_pipes = 1 means the pipes were either split or merged. In this case @@ -1741,10 +1764,28 @@ bool dcn32_internal_validate_bw(struct dc *dc, * ensure all the params are calculated correctly. We do not need to run the * pipe split check again after this call (pipes are already split / merged). * */ - if (!fast_validate) { - context->bw_ctx.dml.soc.allow_for_pstate_or_stutter_in_vblank_final = - dm_prefetch_support_uclk_fclk_and_stutter_if_possible; - vlevel = dml_get_voltage_level(&context->bw_ctx.dml, pipes, pipe_cnt); + context->bw_ctx.dml.soc.allow_for_pstate_or_stutter_in_vblank_final = + dm_prefetch_support_uclk_fclk_and_stutter_if_possible; + vlevel = dml_get_voltage_level(&context->bw_ctx.dml, pipes, pipe_cnt); + if (vlevel == context->bw_ctx.dml.soc.num_states) { + /* failed after DET size changes */ + goto validate_fail; + } else if (flag_max_mpc_comb == 0 && + flag_max_mpc_comb != context->bw_ctx.dml.vba.maxMpcComb) { + /* check the context constructed with pipe split flags is still valid*/ + bool flags_valid = false; + for (int i = flag_vlevel; i < context->bw_ctx.dml.soc.num_states; i++) { + if (vba->ModeSupport[i][flag_max_mpc_comb]) { + vba->maxMpcComb = flag_max_mpc_comb; + vba->VoltageLevel = i; + vlevel = i; + flags_valid = true; + } + } + + /* this should never happen */ + if (!flags_valid) + goto validate_fail; } } *vlevel_out = vlevel; @@ -1803,6 +1844,12 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context, */ context->bw_ctx.dml.soc.dram_clock_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.pstate_latency_us; + /* For DCN32/321 need to validate with fclk pstate change latency equal to dummy so + * prefetch is scheduled correctly to account for dummy pstate. + */ + if (dummy_latency_index == 0) + context->bw_ctx.dml.soc.fclk_change_latency_us = + dc->clk_mgr->bw_params->dummy_pstate_table[dummy_latency_index].dummy_pstate_latency_us; dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, false); maxMpcComb = context->bw_ctx.dml.vba.maxMpcComb; dcfclk_from_fw_based_mclk_switching = context->bw_ctx.dml.vba.DCFCLKState[vlevel][context->bw_ctx.dml.vba.maxMpcComb]; @@ -1990,6 +2037,10 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context, context->perf_params.stutter_period_us = context->bw_ctx.dml.vba.StutterPeriod; + if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching && dummy_latency_index == 0) + context->bw_ctx.dml.soc.fclk_change_latency_us = + dc->clk_mgr->bw_params->dummy_pstate_table[dummy_latency_index].dummy_pstate_latency_us; + dcn32_calculate_dlg_params(dc, context, pipes, pipe_cnt, vlevel); if (!pstate_en) @@ -1997,8 +2048,12 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context, context->bw_ctx.dml.soc.dram_clock_change_latency_us = dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.pstate_latency_us; - if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching) + if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching) { dcn30_setup_mclk_switch_using_fw_based_vblank_stretch(dc, context); + if (dummy_latency_index == 0) + context->bw_ctx.dml.soc.fclk_change_latency_us = + dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.fclk_change_latency_us; + } } static void dcn32_get_optimal_dcfclk_fclk_for_uclk(unsigned int uclk_mts, @@ -2145,9 +2200,7 @@ static int build_synthetic_soc_states(struct clk_bw_params *bw_params, entry.fabricclk_mhz = 0; entry.dram_speed_mts = 0; - DC_FP_START(); insert_entry_into_table_sorted(table, num_entries, &entry); - DC_FP_END(); } // Insert the max DCFCLK @@ -2155,9 +2208,7 @@ static int build_synthetic_soc_states(struct clk_bw_params *bw_params, entry.fabricclk_mhz = 0; entry.dram_speed_mts = 0; - DC_FP_START(); insert_entry_into_table_sorted(table, num_entries, &entry); - DC_FP_END(); // Insert the UCLK DPMS for (i = 0; i < num_uclk_dpms; i++) { @@ -2165,9 +2216,7 @@ static int build_synthetic_soc_states(struct clk_bw_params *bw_params, entry.fabricclk_mhz = 0; entry.dram_speed_mts = bw_params->clk_table.entries[i].memclk_mhz * 16; - DC_FP_START(); insert_entry_into_table_sorted(table, num_entries, &entry); - DC_FP_END(); } // If FCLK is coarse grained, insert individual DPMs. @@ -2177,9 +2226,7 @@ static int build_synthetic_soc_states(struct clk_bw_params *bw_params, entry.fabricclk_mhz = bw_params->clk_table.entries[i].fclk_mhz; entry.dram_speed_mts = 0; - DC_FP_START(); insert_entry_into_table_sorted(table, num_entries, &entry); - DC_FP_END(); } } // If FCLK fine grained, only insert max @@ -2188,9 +2235,7 @@ static int build_synthetic_soc_states(struct clk_bw_params *bw_params, entry.fabricclk_mhz = max_fclk_mhz; entry.dram_speed_mts = 0; - DC_FP_START(); insert_entry_into_table_sorted(table, num_entries, &entry); - DC_FP_END(); } // At this point, the table contains all "points of interest" based on @@ -2359,9 +2404,13 @@ void dcn32_update_bw_bounding_box_fpu(struct dc *dc, struct clk_bw_params *bw_pa if (dc->ctx->dc_bios->vram_info.dram_channel_width_bytes) dcn3_2_soc.dram_channel_width_bytes = dc->ctx->dc_bios->vram_info.dram_channel_width_bytes; - } + /* DML DSC delay factor workaround */ + dcn3_2_ip.dsc_delay_factor_wa = dc->debug.dsc_delay_factor_wa_x1000 / 1000.0; + + dcn3_2_ip.min_prefetch_in_strobe_us = dc->debug.min_prefetch_in_strobe_ns / 1000.0; + /* Override dispclk_dppclk_vco_speed_mhz from Clk Mgr */ dcn3_2_soc.dispclk_dppclk_vco_speed_mhz = dc->clk_mgr->dentist_vco_freq_khz / 1000.0; dc->dml.soc.dispclk_dppclk_vco_speed_mhz = dc->clk_mgr->dentist_vco_freq_khz / 1000.0; @@ -2521,3 +2570,11 @@ void dcn32_update_bw_bounding_box_fpu(struct dc *dc, struct clk_bw_params *bw_pa } } +void dcn32_zero_pipe_dcc_fraction(display_e2e_pipe_params_st *pipes, + int pipe_cnt) +{ + dc_assert_fp_enabled(); + + pipes[pipe_cnt].pipe.src.dcc_fraction_of_zs_req_luma = 0; + pipes[pipe_cnt].pipe.src.dcc_fraction_of_zs_req_chroma = 0; +} diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h index 3a3dc2ce4c73..ab010e7e840b 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.h @@ -73,4 +73,7 @@ int dcn32_find_dummy_latency_index_for_fw_based_mclk_switch(struct dc *dc, void dcn32_patch_dpm_table(struct clk_bw_params *bw_params); +void dcn32_zero_pipe_dcc_fraction(display_e2e_pipe_params_st *pipes, + int pipe_cnt); + #endif diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c index 5b91660a6496..e5c8f6a71b5b 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.c @@ -364,10 +364,11 @@ static void DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerforman for (k = 0; k < mode_lib->vba.NumberOfActiveSurfaces; ++k) { v->DSCDelay[k] = dml32_DSCDelayRequirement(mode_lib->vba.DSCEnabled[k], mode_lib->vba.ODMCombineEnabled[k], mode_lib->vba.DSCInputBitPerComponent[k], - mode_lib->vba.OutputBpp[k], mode_lib->vba.HActive[k], mode_lib->vba.HTotal[k], + mode_lib->vba.OutputBppPerState[mode_lib->vba.VoltageLevel][k], + mode_lib->vba.HActive[k], mode_lib->vba.HTotal[k], mode_lib->vba.NumberOfDSCSlices[k], mode_lib->vba.OutputFormat[k], mode_lib->vba.Output[k], mode_lib->vba.PixelClock[k], - mode_lib->vba.PixelClockBackEnd[k]); + mode_lib->vba.PixelClockBackEnd[k], mode_lib->vba.ip.dsc_delay_factor_wa); } for (k = 0; k < mode_lib->vba.NumberOfActiveSurfaces; ++k) @@ -717,6 +718,8 @@ static void DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerforman do { MaxTotalRDBandwidth = 0; + DestinationLineTimesForPrefetchLessThan2 = false; + VRatioPrefetchMoreThanMax = false; #ifdef __DML_VBA_DEBUG__ dml_print("DML::%s: Start loop: VStartup = %d\n", __func__, mode_lib->vba.VStartupLines); #endif @@ -785,6 +788,8 @@ static void DISPCLKDPPCLKDCFCLKDeepSleepPrefetchParametersWatermarksAndPerforman v->SwathHeightY[k], v->SwathHeightC[k], TWait, + v->DRAMSpeedPerState[mode_lib->vba.VoltageLevel] <= MEM_STROBE_FREQ_MHZ ? + mode_lib->vba.ip.min_prefetch_in_strobe_us : 0, /* Output */ &v->DSTXAfterScaler[k], &v->DSTYAfterScaler[k], @@ -1627,7 +1632,7 @@ static void mode_support_configuration(struct vba_vars_st *v, && !mode_lib->vba.MSOOrODMSplitWithNonDPLink && !mode_lib->vba.NotEnoughLanesForMSO && mode_lib->vba.LinkCapacitySupport[i] == true && !mode_lib->vba.P2IWith420 - && !mode_lib->vba.DSCOnlyIfNecessaryWithBPP + //&& !mode_lib->vba.DSCOnlyIfNecessaryWithBPP && !mode_lib->vba.DSC422NativeNotSupported && !mode_lib->vba.MPCCombineMethodIncompatible && mode_lib->vba.ODMCombine2To1SupportCheckOK[i] == true @@ -2281,7 +2286,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l && (mode_lib->vba.Output[k] == dm_dp || mode_lib->vba.Output[k] == dm_dp2p0 || mode_lib->vba.Output[k] == dm_edp || mode_lib->vba.Output[k] == dm_hdmi) - && mode_lib->vba.OutputBppPerState[i][k] == 0) { + && mode_lib->vba.OutputBppPerState[i][k] == 0 && (mode_lib->vba.UsesMALLForPStateChange[k] != dm_use_mall_pstate_change_phantom_pipe)) { mode_lib->vba.LinkCapacitySupport[i] = false; } } @@ -2475,7 +2480,8 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l mode_lib->vba.OutputBppPerState[i][k], mode_lib->vba.HActive[k], mode_lib->vba.HTotal[k], mode_lib->vba.NumberOfDSCSlices[k], mode_lib->vba.OutputFormat[k], mode_lib->vba.Output[k], - mode_lib->vba.PixelClock[k], mode_lib->vba.PixelClockBackEnd[k]); + mode_lib->vba.PixelClock[k], mode_lib->vba.PixelClockBackEnd[k], + mode_lib->vba.ip.dsc_delay_factor_wa); } for (k = 0; k <= mode_lib->vba.NumberOfActiveSurfaces - 1; k++) { @@ -3190,6 +3196,7 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l mode_lib->vba.FCLKChangeLatency, mode_lib->vba.UrgLatency[i], mode_lib->vba.SREnterPlusExitTime); + memset(&v->dummy_vars.dml32_ModeSupportAndSystemConfigurationFull, 0, sizeof(DmlPipe)); v->dummy_vars.dml32_ModeSupportAndSystemConfigurationFull.myPipe.Dppclk = mode_lib->vba.RequiredDPPCLK[i][j][k]; v->dummy_vars.dml32_ModeSupportAndSystemConfigurationFull.myPipe.Dispclk = mode_lib->vba.RequiredDISPCLK[i][j]; v->dummy_vars.dml32_ModeSupportAndSystemConfigurationFull.myPipe.PixelClock = mode_lib->vba.PixelClock[k]; @@ -3242,6 +3249,8 @@ void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_l v->swath_width_chroma_ub_this_state[k], v->SwathHeightYThisState[k], v->SwathHeightCThisState[k], v->TWait, + v->DRAMSpeedPerState[i] <= MEM_STROBE_FREQ_MHZ ? + mode_lib->vba.ip.min_prefetch_in_strobe_us : 0, /* Output */ &v->dummy_vars.dml32_ModeSupportAndSystemConfigurationFull.DSTXAfterScaler[k], diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.h b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.h index c62e0991358b..c8b28c83ddf4 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_32.h @@ -46,9 +46,14 @@ // Prefetch schedule max vratio #define __DML_MAX_VRATIO_PRE__ 4.0 +#define __DML_VBA_MAX_DST_Y_PRE__ 63.75 + #define BPP_INVALID 0 #define BPP_BLENDED_PIPE 0xffffffff +#define MEM_STROBE_FREQ_MHZ 1600 +#define MEM_STROBE_MAX_DELIVERY_TIME_US 60.0 + struct display_mode_lib; void dml32_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_lib); diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.c b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.c index ad66e241f9ae..debe46b24a3e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.c @@ -1726,7 +1726,8 @@ unsigned int dml32_DSCDelayRequirement(bool DSCEnabled, enum output_format_class OutputFormat, enum output_encoder_class Output, double PixelClock, - double PixelClockBackEnd) + double PixelClockBackEnd, + double dsc_delay_factor_wa) { unsigned int DSCDelayRequirement_val; @@ -1746,7 +1747,7 @@ unsigned int dml32_DSCDelayRequirement(bool DSCEnabled, } DSCDelayRequirement_val = DSCDelayRequirement_val + (HTotal - HActive) * - dml_ceil(DSCDelayRequirement_val / HActive, 1); + dml_ceil((double)DSCDelayRequirement_val / HActive, 1); DSCDelayRequirement_val = DSCDelayRequirement_val * PixelClock / PixelClockBackEnd; @@ -1764,7 +1765,7 @@ unsigned int dml32_DSCDelayRequirement(bool DSCEnabled, dml_print("DML::%s: DSCDelayRequirement_val = %d\n", __func__, DSCDelayRequirement_val); #endif - return DSCDelayRequirement_val; + return dml_ceil(DSCDelayRequirement_val * dsc_delay_factor_wa, 1); } void dml32_CalculateSurfaceSizeInMall( @@ -3416,6 +3417,7 @@ bool dml32_CalculatePrefetchSchedule( unsigned int SwathHeightY, unsigned int SwathHeightC, double TWait, + double TPreReq, /* Output */ double *DSTXAfterScaler, double *DSTYAfterScaler, @@ -3666,6 +3668,7 @@ bool dml32_CalculatePrefetchSchedule( dst_y_prefetch_equ = VStartup - (*TSetup + dml_max(TWait + TCalc, *Tdmdl)) / LineTime - (*DSTYAfterScaler + (double) *DSTXAfterScaler / (double) myPipe->HTotal); + dst_y_prefetch_equ = dml_min(dst_y_prefetch_equ, __DML_VBA_MAX_DST_Y_PRE__); #ifdef __DML_VBA_DEBUG__ dml_print("DML::%s: HTotal = %d\n", __func__, myPipe->HTotal); dml_print("DML::%s: min_Lsw = %f\n", __func__, min_Lsw); @@ -3725,7 +3728,8 @@ bool dml32_CalculatePrefetchSchedule( *VRatioPrefetchY = 0; *VRatioPrefetchC = 0; *RequiredPrefetchPixDataBWLuma = 0; - if (dst_y_prefetch_equ > 1) { + if (dst_y_prefetch_equ > 1 && + (Tpre_rounded >= TPreReq || dst_y_prefetch_equ == __DML_VBA_MAX_DST_Y_PRE__)) { double PrefetchBandwidth1; double PrefetchBandwidth2; double PrefetchBandwidth3; @@ -3871,7 +3875,11 @@ bool dml32_CalculatePrefetchSchedule( } if (dst_y_prefetch_oto < dst_y_prefetch_equ) { - *DestinationLinesForPrefetch = dst_y_prefetch_oto; + if (dst_y_prefetch_oto * LineTime < TPreReq) { + *DestinationLinesForPrefetch = dst_y_prefetch_equ; + } else { + *DestinationLinesForPrefetch = dst_y_prefetch_oto; + } TimeForFetchingMetaPTE = Tvm_oto; TimeForFetchingRowInVBlank = Tr0_oto; *PrefetchBandwidth = prefetch_bw_oto; @@ -4396,7 +4404,7 @@ void dml32_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport( if (v->NumberOfActiveSurfaces > 1) { ActiveClockChangeLatencyHidingY = ActiveClockChangeLatencyHidingY - - (1 - 1 / v->NumberOfActiveSurfaces) * SwathHeightY[k] * v->HTotal[k] + - (1.0 - 1.0 / v->NumberOfActiveSurfaces) * SwathHeightY[k] * v->HTotal[k] / v->PixelClock[k] / v->VRatio[k]; } diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.h b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.h index 55cead0d4237..3989c2a28fae 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_mode_vba_util_32.h @@ -30,7 +30,7 @@ #include "os_types.h" #include "../dc_features.h" #include "../display_mode_structs.h" -#include "dml/display_mode_vba.h" +#include "../display_mode_vba.h" unsigned int dml32_dscceComputeDelay( unsigned int bpc, @@ -327,7 +327,8 @@ unsigned int dml32_DSCDelayRequirement(bool DSCEnabled, enum output_format_class OutputFormat, enum output_encoder_class Output, double PixelClock, - double PixelClockBackEnd); + double PixelClockBackEnd, + double dsc_delay_factor_wa); void dml32_CalculateSurfaceSizeInMall( unsigned int NumberOfActiveSurfaces, @@ -742,6 +743,7 @@ bool dml32_CalculatePrefetchSchedule( unsigned int SwathHeightY, unsigned int SwathHeightC, double TWait, + double TPreReq, /* Output */ double *DSTXAfterScaler, double *DSTYAfterScaler, diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_rq_dlg_calc_32.c b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_rq_dlg_calc_32.c index a1276f6b9581..395ae8761980 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_rq_dlg_calc_32.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/display_rq_dlg_calc_32.c @@ -291,8 +291,8 @@ void dml32_rq_dlg_get_dlg_reg(struct display_mode_lib *mode_lib, dml_print("DML_DLG: %s: vready_after_vcount0 = %d\n", __func__, dlg_regs->vready_after_vcount0); - dst_x_after_scaler = get_dst_x_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx); - dst_y_after_scaler = get_dst_y_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx); + dst_x_after_scaler = dml_ceil(get_dst_x_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx), 1); + dst_y_after_scaler = dml_ceil(get_dst_y_after_scaler(mode_lib, e2e_pipe_param, num_pipes, pipe_idx), 1); // do some adjustment on the dst_after scaler to account for odm combine mode dml_print("DML_DLG: %s: input dst_x_after_scaler = %d\n", __func__, dst_x_after_scaler); diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn321/dcn321_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn321/dcn321_fpu.c index dd90f241e906..432b4ecd01a7 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn321/dcn321_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn321/dcn321_fpu.c @@ -29,6 +29,7 @@ #include "dcn321_fpu.h" #include "dcn32/dcn32_resource.h" #include "dcn321/dcn321_resource.h" +#include "dml/dcn32/display_mode_vba_util_32.h" #define DCN3_2_DEFAULT_DET_SIZE 256 @@ -119,15 +120,15 @@ struct _vcs_dpi_soc_bounding_box_st dcn3_21_soc = { }, }, .num_states = 1, - .sr_exit_time_us = 12.36, - .sr_enter_plus_exit_time_us = 16.72, + .sr_exit_time_us = 19.95, + .sr_enter_plus_exit_time_us = 24.36, .sr_exit_z8_time_us = 285.0, .sr_enter_plus_exit_z8_time_us = 320, .writeback_latency_us = 12.0, .round_trip_ping_latency_dcfclk_cycles = 263, - .urgent_latency_pixel_data_only_us = 4.0, - .urgent_latency_pixel_mixed_with_vm_data_us = 4.0, - .urgent_latency_vm_data_only_us = 4.0, + .urgent_latency_pixel_data_only_us = 9.35, + .urgent_latency_pixel_mixed_with_vm_data_us = 9.35, + .urgent_latency_vm_data_only_us = 9.35, .fclk_change_latency_us = 20, .usr_retraining_latency_us = 2, .smn_latency_us = 2, @@ -538,9 +539,13 @@ void dcn321_update_bw_bounding_box_fpu(struct dc *dc, struct clk_bw_params *bw_p if (dc->ctx->dc_bios->vram_info.dram_channel_width_bytes) dcn3_21_soc.dram_channel_width_bytes = dc->ctx->dc_bios->vram_info.dram_channel_width_bytes; - } + /* DML DSC delay factor workaround */ + dcn3_21_ip.dsc_delay_factor_wa = dc->debug.dsc_delay_factor_wa_x1000 / 1000.0; + + dcn3_21_ip.min_prefetch_in_strobe_us = dc->debug.min_prefetch_in_strobe_ns / 1000.0; + /* Override dispclk_dppclk_vco_speed_mhz from Clk Mgr */ dcn3_21_soc.dispclk_dppclk_vco_speed_mhz = dc->clk_mgr->dentist_vco_freq_khz / 1000.0; dc->dml.soc.dispclk_dppclk_vco_speed_mhz = dc->clk_mgr->dentist_vco_freq_khz / 1000.0; diff --git a/drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h b/drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h index f394b3f3922a..0bffae95f3a2 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h +++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h @@ -105,14 +105,39 @@ enum source_macro_tile_size { enum cursor_bpp { dm_cur_2bit = 0, dm_cur_32bit = 1, dm_cur_64bit = 2 }; + +/** + * @enum clock_change_support - It represents possible reasons to change the DRAM clock. + * + * DC may change the DRAM clock during its execution, and this enum tracks all + * the available methods. Note that every ASIC has their specific way to deal + * with these clock switch. + */ enum clock_change_support { + /** + * @dm_dram_clock_change_uninitialized: If you see this, we might have + * a code initialization issue + */ dm_dram_clock_change_uninitialized = 0, + + /** + * @dm_dram_clock_change_vactive: Support DRAM switch in VActive + */ dm_dram_clock_change_vactive, + + /** + * @dm_dram_clock_change_vblank: Support DRAM switch in VBlank + */ dm_dram_clock_change_vblank, + dm_dram_clock_change_vactive_w_mall_full_frame, dm_dram_clock_change_vactive_w_mall_sub_vp, dm_dram_clock_change_vblank_w_mall_full_frame, dm_dram_clock_change_vblank_w_mall_sub_vp, + + /** + * @dm_dram_clock_change_unsupported: Do not support DRAM switch + */ dm_dram_clock_change_unsupported }; diff --git a/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h b/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h index f33a8879b05a..64d602e6412f 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h +++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_structs.h @@ -364,6 +364,10 @@ struct _vcs_dpi_ip_params_st { unsigned int max_num_dp2p0_outputs; unsigned int max_num_dp2p0_streams; unsigned int VBlankNomDefaultUS; + + /* DM workarounds */ + double dsc_delay_factor_wa; // TODO: Remove after implementing root cause fix + double min_prefetch_in_strobe_us; }; struct _vcs_dpi_display_xfc_params_st { diff --git a/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.c b/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.c index 03924aed8d5c..8e6585dab20e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.c +++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.c @@ -625,7 +625,7 @@ static void fetch_pipe_params(struct display_mode_lib *mode_lib) mode_lib->vba.skip_dio_check[mode_lib->vba.NumberOfActivePlanes] = dout->is_virtual; - if (!dout->dsc_enable) + if (dout->dsc_enable) mode_lib->vba.ForcedOutputLinkBPP[mode_lib->vba.NumberOfActivePlanes] = dout->output_bpp; else mode_lib->vba.ForcedOutputLinkBPP[mode_lib->vba.NumberOfActivePlanes] = 0.0; diff --git a/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.h b/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.h index 630f3395e90a..d46aa4817e70 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.h +++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_vba.h @@ -419,6 +419,15 @@ struct vba_vars_st { double MinPixelChunkSizeBytes; unsigned int DCCMetaBufferSizeBytes; // Pipe/Plane Parameters + + /** @VoltageLevel: + * Every ASIC has a fixed number of DPM states, and some devices might + * have some particular voltage configuration that does not map + * directly to the DPM states. This field tells how many states the + * target device supports; even though this field combines the DPM and + * special SOC voltages, it mostly matches the total number of DPM + * states. + */ int VoltageLevel; double FabricClock; double DRAMSpeed; diff --git a/drivers/gpu/drm/amd/display/dc/dml/dsc/qp_tables.h b/drivers/gpu/drm/amd/display/dc/dml/dsc/qp_tables.h index e5fac9f4181d..dcff0dd2b6a1 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dsc/qp_tables.h +++ b/drivers/gpu/drm/amd/display/dc/dml/dsc/qp_tables.h @@ -25,7 +25,7 @@ */ -const qp_table qp_table_422_10bpc_min = { +static const qp_table qp_table_422_10bpc_min = { { 6, { 0, 4, 5, 6, 6, 6, 6, 7, 7, 8, 9, 9, 9, 12, 16} }, { 6.5, { 0, 4, 5, 6, 6, 6, 6, 7, 7, 8, 9, 9, 9, 12, 16} }, { 7, { 0, 4, 5, 6, 6, 6, 6, 7, 7, 7, 9, 9, 9, 11, 15} }, @@ -58,7 +58,7 @@ const qp_table qp_table_422_10bpc_min = { }; -const qp_table qp_table_444_8bpc_max = { +static const qp_table qp_table_444_8bpc_max = { { 6, { 4, 6, 8, 8, 9, 9, 9, 10, 11, 12, 12, 12, 12, 13, 15} }, { 6.5, { 4, 6, 7, 8, 8, 8, 9, 10, 11, 11, 12, 12, 12, 13, 15} }, { 7, { 4, 5, 7, 7, 8, 8, 8, 9, 10, 11, 11, 12, 12, 13, 14} }, @@ -99,7 +99,7 @@ const qp_table qp_table_444_8bpc_max = { }; -const qp_table qp_table_420_12bpc_max = { +static const qp_table qp_table_420_12bpc_max = { { 4, {11, 12, 13, 14, 15, 15, 15, 16, 17, 18, 18, 19, 19, 21, 22} }, { 4.5, {10, 11, 12, 13, 14, 15, 15, 16, 17, 18, 18, 19, 19, 20, 21} }, { 5, { 9, 11, 12, 13, 14, 15, 15, 16, 17, 17, 18, 18, 19, 20, 21} }, @@ -132,7 +132,7 @@ const qp_table qp_table_420_12bpc_max = { }; -const qp_table qp_table_444_10bpc_min = { +static const qp_table qp_table_444_10bpc_min = { { 6, { 0, 4, 7, 7, 9, 9, 9, 9, 9, 10, 10, 10, 10, 12, 18} }, { 6.5, { 0, 4, 6, 7, 8, 8, 9, 9, 9, 9, 10, 10, 10, 12, 18} }, { 7, { 0, 4, 6, 6, 8, 8, 8, 8, 8, 9, 9, 10, 10, 12, 17} }, @@ -185,7 +185,7 @@ const qp_table qp_table_444_10bpc_min = { }; -const qp_table qp_table_420_8bpc_max = { +static const qp_table qp_table_420_8bpc_max = { { 4, { 4, 4, 5, 6, 7, 7, 7, 8, 9, 10, 10, 11, 11, 13, 14} }, { 4.5, { 4, 4, 5, 6, 7, 7, 7, 8, 9, 10, 10, 11, 11, 12, 13} }, { 5, { 3, 4, 5, 6, 7, 7, 7, 8, 9, 9, 10, 10, 11, 12, 13} }, @@ -206,7 +206,7 @@ const qp_table qp_table_420_8bpc_max = { }; -const qp_table qp_table_444_8bpc_min = { +static const qp_table qp_table_444_8bpc_min = { { 6, { 0, 1, 3, 3, 5, 5, 5, 5, 5, 6, 6, 6, 6, 9, 14} }, { 6.5, { 0, 1, 2, 3, 4, 4, 5, 5, 5, 5, 6, 6, 6, 9, 14} }, { 7, { 0, 0, 2, 2, 4, 4, 4, 4, 4, 5, 5, 6, 6, 9, 13} }, @@ -247,7 +247,7 @@ const qp_table qp_table_444_8bpc_min = { }; -const qp_table qp_table_444_12bpc_min = { +static const qp_table qp_table_444_12bpc_min = { { 6, { 0, 5, 11, 11, 13, 13, 13, 13, 13, 14, 14, 14, 14, 17, 22} }, { 6.5, { 0, 5, 10, 11, 12, 12, 13, 13, 13, 13, 14, 14, 14, 17, 22} }, { 7, { 0, 5, 10, 10, 12, 12, 12, 12, 12, 13, 13, 14, 14, 17, 21} }, @@ -312,7 +312,7 @@ const qp_table qp_table_444_12bpc_min = { }; -const qp_table qp_table_420_12bpc_min = { +static const qp_table qp_table_420_12bpc_min = { { 4, { 0, 4, 9, 10, 11, 11, 11, 11, 11, 11, 13, 13, 13, 15, 21} }, { 4.5, { 0, 4, 8, 9, 10, 11, 11, 11, 11, 11, 13, 13, 13, 15, 20} }, { 5, { 0, 4, 8, 9, 10, 11, 11, 11, 11, 11, 13, 13, 13, 15, 20} }, @@ -345,7 +345,7 @@ const qp_table qp_table_420_12bpc_min = { }; -const qp_table qp_table_422_12bpc_min = { +static const qp_table qp_table_422_12bpc_min = { { 6, { 0, 4, 9, 10, 11, 11, 11, 11, 11, 11, 13, 13, 13, 16, 20} }, { 6.5, { 0, 4, 9, 10, 11, 11, 11, 11, 11, 11, 13, 13, 13, 16, 20} }, { 7, { 0, 4, 9, 10, 11, 11, 11, 11, 11, 11, 13, 13, 13, 15, 19} }, @@ -386,7 +386,7 @@ const qp_table qp_table_422_12bpc_min = { }; -const qp_table qp_table_422_12bpc_max = { +static const qp_table qp_table_422_12bpc_max = { { 6, {12, 12, 13, 14, 15, 15, 15, 16, 17, 18, 18, 19, 19, 20, 21} }, { 6.5, {12, 12, 13, 14, 15, 15, 15, 16, 17, 18, 18, 19, 19, 20, 21} }, { 7, {11, 12, 13, 14, 15, 15, 15, 16, 17, 17, 18, 18, 19, 19, 20} }, @@ -427,7 +427,7 @@ const qp_table qp_table_422_12bpc_max = { }; -const qp_table qp_table_444_12bpc_max = { +static const qp_table qp_table_444_12bpc_max = { { 6, {12, 14, 16, 16, 17, 17, 17, 18, 19, 20, 20, 20, 20, 21, 23} }, { 6.5, {12, 14, 15, 16, 16, 16, 17, 18, 19, 19, 20, 20, 20, 21, 23} }, { 7, {12, 13, 15, 15, 16, 16, 16, 17, 18, 19, 19, 20, 20, 21, 22} }, @@ -492,7 +492,7 @@ const qp_table qp_table_444_12bpc_max = { }; -const qp_table qp_table_420_8bpc_min = { +static const qp_table qp_table_420_8bpc_min = { { 4, { 0, 0, 1, 1, 3, 3, 3, 3, 3, 3, 5, 5, 5, 9, 13} }, { 4.5, { 0, 0, 1, 1, 3, 3, 3, 3, 3, 3, 5, 5, 5, 8, 12} }, { 5, { 0, 0, 1, 1, 3, 3, 3, 3, 3, 3, 5, 5, 5, 8, 12} }, @@ -513,7 +513,7 @@ const qp_table qp_table_420_8bpc_min = { }; -const qp_table qp_table_422_8bpc_min = { +static const qp_table qp_table_422_8bpc_min = { { 6, { 0, 0, 1, 2, 3, 3, 3, 3, 3, 3, 5, 5, 5, 8, 12} }, { 6.5, { 0, 0, 1, 2, 3, 3, 3, 3, 3, 3, 5, 5, 5, 8, 12} }, { 7, { 0, 0, 1, 2, 3, 3, 3, 3, 3, 3, 5, 5, 5, 7, 11} }, @@ -538,7 +538,7 @@ const qp_table qp_table_422_8bpc_min = { }; -const qp_table qp_table_422_10bpc_max = { +static const qp_table qp_table_422_10bpc_max = { { 6, { 8, 8, 9, 10, 11, 11, 11, 12, 13, 14, 14, 15, 15, 16, 17} }, { 6.5, { 8, 8, 9, 10, 11, 11, 11, 12, 13, 14, 14, 15, 15, 16, 17} }, { 7, { 7, 8, 9, 10, 11, 11, 11, 12, 13, 13, 14, 14, 15, 15, 16} }, @@ -571,7 +571,7 @@ const qp_table qp_table_422_10bpc_max = { }; -const qp_table qp_table_420_10bpc_max = { +static const qp_table qp_table_420_10bpc_max = { { 4, { 8, 8, 9, 10, 11, 11, 11, 12, 13, 14, 14, 15, 15, 17, 18} }, { 4.5, { 8, 8, 9, 10, 11, 11, 11, 12, 13, 14, 14, 15, 15, 16, 17} }, { 5, { 7, 8, 9, 10, 11, 11, 11, 12, 13, 13, 14, 14, 15, 16, 17} }, @@ -598,7 +598,7 @@ const qp_table qp_table_420_10bpc_max = { }; -const qp_table qp_table_420_10bpc_min = { +static const qp_table qp_table_420_10bpc_min = { { 4, { 0, 4, 4, 5, 7, 7, 7, 7, 7, 7, 9, 9, 9, 13, 17} }, { 4.5, { 0, 4, 4, 5, 7, 7, 7, 7, 7, 7, 9, 9, 9, 12, 16} }, { 5, { 0, 4, 4, 5, 7, 7, 7, 7, 7, 7, 9, 9, 9, 12, 16} }, @@ -625,7 +625,7 @@ const qp_table qp_table_420_10bpc_min = { }; -const qp_table qp_table_444_10bpc_max = { +static const qp_table qp_table_444_10bpc_max = { { 6, { 8, 10, 12, 12, 13, 13, 13, 14, 15, 16, 16, 16, 16, 17, 19} }, { 6.5, { 8, 10, 11, 12, 12, 12, 13, 14, 15, 15, 16, 16, 16, 17, 19} }, { 7, { 8, 9, 11, 11, 12, 12, 12, 13, 14, 15, 15, 16, 16, 17, 18} }, @@ -678,7 +678,7 @@ const qp_table qp_table_444_10bpc_max = { }; -const qp_table qp_table_422_8bpc_max = { +static const qp_table qp_table_422_8bpc_max = { { 6, { 4, 4, 5, 6, 7, 7, 7, 8, 9, 10, 10, 11, 11, 12, 13} }, { 6.5, { 4, 4, 5, 6, 7, 7, 7, 8, 9, 10, 10, 11, 11, 12, 13} }, { 7, { 3, 4, 5, 6, 7, 7, 7, 8, 9, 9, 10, 10, 11, 11, 12} }, diff --git a/drivers/gpu/drm/amd/display/dc/gpio/dcn32/hw_factory_dcn32.c b/drivers/gpu/drm/amd/display/dc/gpio/dcn32/hw_factory_dcn32.c index d635b73af46f..0ea52ba5ac82 100644 --- a/drivers/gpu/drm/amd/display/dc/gpio/dcn32/hw_factory_dcn32.c +++ b/drivers/gpu/drm/amd/display/dc/gpio/dcn32/hw_factory_dcn32.c @@ -108,6 +108,13 @@ static const struct ddc_registers ddc_data_regs_dcn[] = { ddc_data_regs_dcn2(4), ddc_data_regs_dcn2(5), { + // add a dummy entry for cases no such port + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,}, + .ddc_setup = 0, + .phy_aux_cntl = 0, + .dc_gpio_aux_ctrl_5 = 0 + }, + { DDC_GPIO_VGA_REG_LIST(DATA), .ddc_setup = 0, .phy_aux_cntl = 0, @@ -122,6 +129,13 @@ static const struct ddc_registers ddc_clk_regs_dcn[] = { ddc_clk_regs_dcn2(4), ddc_clk_regs_dcn2(5), { + // add a dummy entry for cases no such port + {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,}, + .ddc_setup = 0, + .phy_aux_cntl = 0, + .dc_gpio_aux_ctrl_5 = 0 + }, + { DDC_GPIO_VGA_REG_LIST(CLK), .ddc_setup = 0, .phy_aux_cntl = 0, diff --git a/drivers/gpu/drm/amd/display/dc/gpio/hw_ddc.c b/drivers/gpu/drm/amd/display/dc/gpio/hw_ddc.c index 6fd38cdd68c0..525bc8881950 100644 --- a/drivers/gpu/drm/amd/display/dc/gpio/hw_ddc.c +++ b/drivers/gpu/drm/amd/display/dc/gpio/hw_ddc.c @@ -94,11 +94,14 @@ static enum gpio_result set_config( * is required for detection of AUX mode */ if (hw_gpio->base.en != GPIO_DDC_LINE_VIP_PAD) { if (!ddc_data_pd_en || !ddc_clk_pd_en) { - - REG_SET_2(gpio.MASK_reg, regval, + if (hw_gpio->base.en == GPIO_DDC_LINE_DDC_VGA) { + // bit 4 of mask has different usage in some cases + REG_SET(gpio.MASK_reg, regval, DC_GPIO_DDC1DATA_PD_EN, 1); + } else { + REG_SET_2(gpio.MASK_reg, regval, DC_GPIO_DDC1DATA_PD_EN, 1, DC_GPIO_DDC1CLK_PD_EN, 1); - + } if (config_data->type == GPIO_CONFIG_TYPE_I2C_AUX_DUAL_MODE) msleep(3); diff --git a/drivers/gpu/drm/amd/display/dc/inc/core_types.h b/drivers/gpu/drm/amd/display/dc/inc/core_types.h index 9498105c98ab..85495ef026f5 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/core_types.h +++ b/drivers/gpu/drm/amd/display/dc/inc/core_types.h @@ -115,6 +115,13 @@ struct resource_funcs { int vlevel); void (*update_soc_for_wm_a)( struct dc *dc, struct dc_state *context); + + /** + * @populate_dml_pipes - Populate pipe data struct + * + * Returns: + * Total of pipes available in the specific ASIC. + */ int (*populate_dml_pipes)( struct dc *dc, struct dc_state *context, @@ -413,7 +420,10 @@ struct pipe_ctx { struct pll_settings pll_settings; - /* link config records software decision for what link config should be + /** + * @link_config: + * + * link config records software decision for what link config should be * enabled given current link capability and stream during hw resource * mapping. This is to decouple the dependency on link capability during * dc commit or update. @@ -507,33 +517,62 @@ struct bw_context { union bw_output bw; struct display_mode_lib dml; }; + /** - * struct dc_state - The full description of a state requested by a user - * - * @streams: Stream properties - * @stream_status: The planes on a given stream - * @res_ctx: Persistent state of resources - * @bw_ctx: The output from bandwidth and watermark calculations and the DML - * @pp_display_cfg: PowerPlay clocks and settings - * @dcn_bw_vars: non-stack memory to support bandwidth calculations - * + * struct dc_state - The full description of a state requested by users */ struct dc_state { + /** + * @streams: Stream state properties + */ struct dc_stream_state *streams[MAX_PIPES]; + + /** + * @stream_status: Planes status on a given stream + */ struct dc_stream_status stream_status[MAX_PIPES]; + + /** + * @stream_count: Total of streams in use + */ uint8_t stream_count; uint8_t stream_mask; + /** + * @res_ctx: Persistent state of resources + */ struct resource_context res_ctx; + /** + * @bw_ctx: The output from bandwidth and watermark calculations and the DML + * + * Each context must have its own instance of VBA, and in order to + * initialize and obtain IP and SOC, the base DML instance from DC is + * initially copied into every context. + */ struct bw_context bw_ctx; - /* Note: these are big structures, do *not* put on stack! */ + /** + * @pp_display_cfg: PowerPlay clocks and settings + * Note: this is a big struct, do *not* put on stack! + */ struct dm_pp_display_configuration pp_display_cfg; + + /** + * @dcn_bw_vars: non-stack memory to support bandwidth calculations + * Note: this is a big struct, do *not* put on stack! + */ struct dcn_bw_internal_vars dcn_bw_vars; struct clk_mgr *clk_mgr; + /** + * @refcount: refcount reference + * + * Notice that dc_state is used around the code to capture the current + * context, so we need to pass it everywhere. That's why we want to use + * kref in this struct. + */ struct kref refcount; struct { diff --git a/drivers/gpu/drm/amd/display/dc/inc/dc_link_dp.h b/drivers/gpu/drm/amd/display/dc/inc/dc_link_dp.h index b304d450b038..e8d8c5cb1309 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/dc_link_dp.h +++ b/drivers/gpu/drm/amd/display/dc/inc/dc_link_dp.h @@ -193,7 +193,7 @@ enum dc_status dpcd_configure_lttpr_mode( struct link_training_settings *lt_settings); enum dp_link_encoding dp_get_link_encoding_format(const struct dc_link_settings *link_settings); -bool dp_retrieve_lttpr_cap(struct dc_link *link); +enum dc_status dp_retrieve_lttpr_cap(struct dc_link *link); bool dp_is_lttpr_present(struct dc_link *link); enum lttpr_mode dp_decide_lttpr_mode(struct dc_link *link, struct dc_link_settings *link_setting); void dp_get_lttpr_mode_override(struct dc_link *link, enum lttpr_mode *override); diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h index e7571c6f5ead..f2e1fcb668fb 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h @@ -167,10 +167,26 @@ struct hubbub_funcs { void (*force_pstate_change_control)(struct hubbub *hubbub, bool force, bool allow); void (*init_watermarks)(struct hubbub *hubbub); + + /** + * @program_det_size: + * + * DE-Tile buffers (DET) is a memory that is used to convert the tiled + * data into linear, which the rest of the display can use to generate + * the graphics output. One of the main features of this component is + * that each pipe has a configurable DET buffer which means that when a + * pipe is not enabled, the device can assign the memory to other + * enabled pipes to try to be more efficient. + * + * DET logic is handled by dchubbub. Some ASICs provide a feature named + * Configurable Return Buffer (CRB) segments which can be allocated to + * compressed or detiled buffers. + */ void (*program_det_size)(struct hubbub *hubbub, int hubp_inst, unsigned det_buffer_size_in_kbyte); void (*program_compbuf_size)(struct hubbub *hubbub, unsigned compbuf_size_kb, bool safe_to_increase); void (*init_crb)(struct hubbub *hubbub); void (*force_usr_retraining_allow)(struct hubbub *hubbub, bool allow); + void (*set_request_limit)(struct hubbub *hubbub, int memory_channel_count, int words_per_channel); }; struct hubbub { diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dmcu.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dmcu.h index 8df2765cce78..de3113ecbc77 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dmcu.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dmcu.h @@ -56,20 +56,6 @@ struct dmcu { bool auto_load_dmcu; }; -#if defined(CONFIG_DRM_AMD_SECURE_DISPLAY) -struct crc_region { - uint16_t x_start; - uint16_t y_start; - uint16_t x_end; - uint16_t y_end; -}; - -struct otg_phy_mux { - uint8_t phy_output_num; - uint8_t otg_output_num; -}; -#endif - struct dmcu_funcs { bool (*dmcu_init)(struct dmcu *dmcu); bool (*load_iram)(struct dmcu *dmcu, @@ -100,7 +86,7 @@ struct dmcu_funcs { bool (*recv_edid_cea_ack)(struct dmcu *dmcu, int *offset); #if defined(CONFIG_DRM_AMD_SECURE_DISPLAY) void (*forward_crc_window)(struct dmcu *dmcu, - struct crc_region *crc_win, + struct rect *rect, struct otg_phy_mux *mux_mapping); void (*stop_crc_win_update)(struct dmcu *dmcu, struct otg_phy_mux *mux_mapping); diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h b/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h index cd2be729846b..a819f0f97c5f 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h @@ -35,6 +35,13 @@ ******************************************************************************/ #define MAX_AUDIOS 7 + +/** + * @MAX_PIPES: + * + * Every ASIC support a fixed number of pipes; MAX_PIPES defines a large number + * to be used inside loops and for determining array sizes. + */ #define MAX_PIPES 6 #define MAX_DIG_LINK_ENCODERS 7 #define MAX_DWB_PIPES 1 diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h b/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h index 42afa1952890..42db4b7b79fd 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/stream_encoder.h @@ -243,6 +243,9 @@ struct stream_encoder_funcs { uint32_t hubp_requestor_id, enum dynamic_metadata_mode dmdata_mode); + /** + * @dp_set_odm_combine: Sets up DP stream encoder for ODM. + */ void (*dp_set_odm_combine)( struct stream_encoder *enc, bool odm_combine); @@ -317,9 +320,6 @@ struct hpo_dp_stream_encoder_funcs { uint32_t stream_enc_inst, uint32_t link_enc_inst); - void (*audio_mute_control)( - struct hpo_dp_stream_encoder *enc, bool mute); - void (*dp_audio_setup)( struct hpo_dp_stream_encoder *enc, unsigned int az_inst, diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h index 25a1df45b264..0e42e721dd15 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h @@ -185,6 +185,7 @@ struct timing_generator_funcs { #ifdef CONFIG_DRM_AMD_DC_DCN void (*phantom_crtc_post_enable)(struct timing_generator *tg); #endif + void (*disable_phantom_crtc)(struct timing_generator *tg); bool (*immediate_disable_crtc)(struct timing_generator *tg); bool (*is_counter_moving)(struct timing_generator *tg); void (*get_position)(struct timing_generator *tg, @@ -301,6 +302,11 @@ struct timing_generator_funcs { void (*get_dsc_status)(struct timing_generator *optc, uint32_t *dsc_mode); void (*set_odm_bypass)(struct timing_generator *optc, const struct dc_crtc_timing *dc_crtc_timing); + + /** + * @set_odm_combine: Set up the ODM block to read from the correct + * OPP(s) and turn on/off ODM memory. + */ void (*set_odm_combine)(struct timing_generator *optc, int *opp_id, int opp_cnt, struct dc_crtc_timing *timing); void (*set_h_timing_div_manual_mode)(struct timing_generator *optc, bool manual_mode); diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h index d04b68dad413..c43523f9ff6d 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h @@ -263,6 +263,7 @@ struct hw_sequencer_funcs { void (*update_phantom_vp_position)(struct dc *dc, struct dc_state *context, struct pipe_ctx *phantom_pipe); + void (*apply_update_flags_for_phantom)(struct pipe_ctx *phantom_pipe); void (*commit_subvp_config)(struct dc *dc, struct dc_state *context); void (*subvp_pipe_control_lock)(struct dc *dc, diff --git a/drivers/gpu/drm/amd/display/dc/inc/link_hwss.h b/drivers/gpu/drm/amd/display/dc/inc/link_hwss.h index 89964c980b87..0f69946cce9f 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/link_hwss.h +++ b/drivers/gpu/drm/amd/display/dc/inc/link_hwss.h @@ -38,6 +38,7 @@ struct link_resource; struct pipe_ctx; struct encoder_set_dp_phy_pattern_param; struct link_mst_stream_allocation_table; +struct audio_output; struct link_hwss_ext { /* function pointers below may require to check for NULL if caller @@ -79,6 +80,10 @@ struct link_hwss { void (*disable_link_output)(struct dc_link *link, const struct link_resource *link_res, enum signal_type signal); + void (*setup_audio_output)(struct pipe_ctx *pipe_ctx, + struct audio_output *audio_output, uint32_t audio_inst); + void (*enable_audio_packet)(struct pipe_ctx *pipe_ctx); + void (*disable_audio_packet)(struct pipe_ctx *pipe_ctx); }; #endif /* __DC_LINK_HWSS_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/irq/dcn201/irq_service_dcn201.c b/drivers/gpu/drm/amd/display/dc/irq/dcn201/irq_service_dcn201.c index 45f99351a0ab..5f4f6dd79511 100644 --- a/drivers/gpu/drm/amd/display/dc/irq/dcn201/irq_service_dcn201.c +++ b/drivers/gpu/drm/amd/display/dc/irq/dcn201/irq_service_dcn201.c @@ -28,20 +28,19 @@ #include "include/logger_interface.h" #include "../dce110/irq_service_dce110.h" +#include "irq_service_dcn201.h" #include "dcn/dcn_2_0_3_offset.h" #include "dcn/dcn_2_0_3_sh_mask.h" #include "cyan_skillfish_ip_offset.h" #include "soc15_hw_ip.h" - -#include "irq_service_dcn201.h" - #include "ivsrcid/dcn/irqsrcs_dcn_1_0.h" -static enum dc_irq_source to_dal_irq_source_dcn201(struct irq_service *irq_service, - uint32_t src_id, - uint32_t ext_id) +enum dc_irq_source to_dal_irq_source_dcn201( + struct irq_service *irq_service, + uint32_t src_id, + uint32_t ext_id) { switch (src_id) { case DCN_1_0__SRCID__DC_D1_OTG_VSTARTUP: @@ -79,7 +78,6 @@ static enum dc_irq_source to_dal_irq_source_dcn201(struct irq_service *irq_servi default: return DC_IRQ_SOURCE_INVALID; } - return DC_IRQ_SOURCE_INVALID; } static bool hpd_ack( @@ -138,6 +136,11 @@ static const struct irq_source_info_funcs vupdate_no_lock_irq_info_funcs = { .ack = NULL }; +static const struct irq_source_info_funcs dmub_outbox_irq_info_funcs = { + .set = NULL, + .ack = NULL +}; + #undef BASE_INNER #define BASE_INNER(seg) DMU_BASE__INST0_SEG ## seg diff --git a/drivers/gpu/drm/amd/display/dc/irq/dcn201/irq_service_dcn201.h b/drivers/gpu/drm/amd/display/dc/irq/dcn201/irq_service_dcn201.h index 8e27c5e219a3..0cfd2f2d62e8 100644 --- a/drivers/gpu/drm/amd/display/dc/irq/dcn201/irq_service_dcn201.h +++ b/drivers/gpu/drm/amd/display/dc/irq/dcn201/irq_service_dcn201.h @@ -1,5 +1,5 @@ /* - * Copyright 2018 Advanced Micro Devices, Inc. + * Copyright 2022 Advanced Micro Devices, Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), diff --git a/drivers/gpu/drm/amd/display/dc/irq/irq_service.c b/drivers/gpu/drm/amd/display/dc/irq/irq_service.c index 7bad39bba86b..d100edaedbbb 100644 --- a/drivers/gpu/drm/amd/display/dc/irq/irq_service.c +++ b/drivers/gpu/drm/amd/display/dc/irq/irq_service.c @@ -112,8 +112,15 @@ bool dal_irq_service_set( dal_irq_service_ack(irq_service, source); - if (info->funcs && info->funcs->set) + if (info->funcs && info->funcs->set) { + if (info->funcs->set == dal_irq_service_dummy_set) { + DC_LOG_WARNING("%s: src: %d, st: %d\n", __func__, + source, enable); + ASSERT(0); + } + return info->funcs->set(irq_service, info, enable); + } dal_irq_service_set_generic(irq_service, info, enable); @@ -146,8 +153,14 @@ bool dal_irq_service_ack( return false; } - if (info->funcs && info->funcs->ack) + if (info->funcs && info->funcs->ack) { + if (info->funcs->ack == dal_irq_service_dummy_ack) { + DC_LOG_WARNING("%s: src: %d\n", __func__, source); + ASSERT(0); + } + return info->funcs->ack(irq_service, info); + } dal_irq_service_ack_generic(irq_service, info); diff --git a/drivers/gpu/drm/amd/display/dc/link/link_hwss_dio.c b/drivers/gpu/drm/amd/display/dc/link/link_hwss_dio.c index 4227adbc646a..33148b753c03 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_hwss_dio.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_hwss_dio.c @@ -170,11 +170,63 @@ static void update_dio_stream_allocation_table(struct dc_link *link, link_enc->funcs->update_mst_stream_allocation_table(link_enc, table); } +void setup_dio_audio_output(struct pipe_ctx *pipe_ctx, + struct audio_output *audio_output, uint32_t audio_inst) +{ + if (dc_is_dp_signal(pipe_ctx->stream->signal)) + pipe_ctx->stream_res.stream_enc->funcs->dp_audio_setup( + pipe_ctx->stream_res.stream_enc, + audio_inst, + &pipe_ctx->stream->audio_info); + else + pipe_ctx->stream_res.stream_enc->funcs->hdmi_audio_setup( + pipe_ctx->stream_res.stream_enc, + audio_inst, + &pipe_ctx->stream->audio_info, + &audio_output->crtc_info); +} + +void enable_dio_audio_packet(struct pipe_ctx *pipe_ctx) +{ + if (dc_is_dp_signal(pipe_ctx->stream->signal)) + pipe_ctx->stream_res.stream_enc->funcs->dp_audio_enable( + pipe_ctx->stream_res.stream_enc); + + pipe_ctx->stream_res.stream_enc->funcs->audio_mute_control( + pipe_ctx->stream_res.stream_enc, false); + + if (dc_is_dp_signal(pipe_ctx->stream->signal)) + dp_source_sequence_trace(pipe_ctx->stream->link, + DPCD_SOURCE_SEQ_AFTER_ENABLE_AUDIO_STREAM); +} + +void disable_dio_audio_packet(struct pipe_ctx *pipe_ctx) +{ + pipe_ctx->stream_res.stream_enc->funcs->audio_mute_control( + pipe_ctx->stream_res.stream_enc, true); + + if (pipe_ctx->stream_res.audio) { + if (dc_is_dp_signal(pipe_ctx->stream->signal)) + pipe_ctx->stream_res.stream_enc->funcs->dp_audio_disable( + pipe_ctx->stream_res.stream_enc); + else + pipe_ctx->stream_res.stream_enc->funcs->hdmi_audio_disable( + pipe_ctx->stream_res.stream_enc); + } + + if (dc_is_dp_signal(pipe_ctx->stream->signal)) + dp_source_sequence_trace(pipe_ctx->stream->link, + DPCD_SOURCE_SEQ_AFTER_DISABLE_AUDIO_STREAM); +} + static const struct link_hwss dio_link_hwss = { .setup_stream_encoder = setup_dio_stream_encoder, .reset_stream_encoder = reset_dio_stream_encoder, .setup_stream_attribute = setup_dio_stream_attribute, .disable_link_output = disable_dio_link_output, + .setup_audio_output = setup_dio_audio_output, + .enable_audio_packet = enable_dio_audio_packet, + .disable_audio_packet = disable_dio_audio_packet, .ext = { .set_throttled_vcp_size = set_dio_throttled_vcp_size, .enable_dp_link_output = enable_dio_dp_link_output, diff --git a/drivers/gpu/drm/amd/display/dc/link/link_hwss_dio.h b/drivers/gpu/drm/amd/display/dc/link/link_hwss_dio.h index 126d37f847a1..9a108c3d7831 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_hwss_dio.h +++ b/drivers/gpu/drm/amd/display/dc/link/link_hwss_dio.h @@ -50,5 +50,9 @@ void set_dio_dp_lane_settings(struct dc_link *link, const struct link_resource *link_res, const struct dc_link_settings *link_settings, const struct dc_lane_settings lane_settings[LANE_COUNT_DP_MAX]); +void setup_dio_audio_output(struct pipe_ctx *pipe_ctx, + struct audio_output *audio_output, uint32_t audio_inst); +void enable_dio_audio_packet(struct pipe_ctx *pipe_ctx); +void disable_dio_audio_packet(struct pipe_ctx *pipe_ctx); #endif /* __LINK_HWSS_DIO_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/link/link_hwss_dpia.c b/drivers/gpu/drm/amd/display/dc/link/link_hwss_dpia.c index 64f7ea6a9aa3..861f3cd5b356 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_hwss_dpia.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_hwss_dpia.c @@ -57,6 +57,9 @@ static const struct link_hwss dpia_link_hwss = { .reset_stream_encoder = reset_dio_stream_encoder, .setup_stream_attribute = setup_dio_stream_attribute, .disable_link_output = disable_dio_link_output, + .setup_audio_output = setup_dio_audio_output, + .enable_audio_packet = enable_dio_audio_packet, + .disable_audio_packet = disable_dio_audio_packet, .ext = { .set_throttled_vcp_size = set_dio_throttled_vcp_size, .enable_dp_link_output = enable_dio_dp_link_output, diff --git a/drivers/gpu/drm/amd/display/dc/link/link_hwss_hpo_dp.c b/drivers/gpu/drm/amd/display/dc/link/link_hwss_hpo_dp.c index 153a88381f2c..2f46e1ac4ce0 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_hwss_hpo_dp.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_hwss_hpo_dp.c @@ -262,11 +262,36 @@ static void update_hpo_dp_stream_allocation_table(struct dc_link *link, table); } +static void setup_hpo_dp_audio_output(struct pipe_ctx *pipe_ctx, + struct audio_output *audio_output, uint32_t audio_inst) +{ + pipe_ctx->stream_res.hpo_dp_stream_enc->funcs->dp_audio_setup( + pipe_ctx->stream_res.hpo_dp_stream_enc, + audio_inst, + &pipe_ctx->stream->audio_info); +} + +static void enable_hpo_dp_audio_packet(struct pipe_ctx *pipe_ctx) +{ + pipe_ctx->stream_res.hpo_dp_stream_enc->funcs->dp_audio_enable( + pipe_ctx->stream_res.hpo_dp_stream_enc); +} + +static void disable_hpo_dp_audio_packet(struct pipe_ctx *pipe_ctx) +{ + if (pipe_ctx->stream_res.audio) + pipe_ctx->stream_res.hpo_dp_stream_enc->funcs->dp_audio_disable( + pipe_ctx->stream_res.hpo_dp_stream_enc); +} + static const struct link_hwss hpo_dp_link_hwss = { .setup_stream_encoder = setup_hpo_dp_stream_encoder, .reset_stream_encoder = reset_hpo_dp_stream_encoder, .setup_stream_attribute = setup_hpo_dp_stream_attribute, .disable_link_output = disable_hpo_dp_link_output, + .setup_audio_output = setup_hpo_dp_audio_output, + .enable_audio_packet = enable_hpo_dp_audio_packet, + .disable_audio_packet = disable_hpo_dp_audio_packet, .ext = { .set_throttled_vcp_size = set_hpo_dp_throttled_vcp_size, .set_hblank_min_symbol_width = set_hpo_dp_hblank_min_symbol_width, diff --git a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h index 7a8f61517424..795d8811af9a 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h @@ -225,6 +225,12 @@ union dmub_psr_debug_flags { * Use TPS3 signal when restore main link. */ uint32_t force_wakeup_by_tps3 : 1; + + /** + * Back to back flip, therefore cannot power down PHY + */ + uint32_t back_to_back_flip : 1; + } bitfields; /** @@ -401,8 +407,9 @@ union dmub_fw_boot_options { uint32_t gpint_scratch8: 1; /* 1 if GPINT is in scratch8*/ uint32_t usb4_cm_version: 1; /**< 1 CM support */ uint32_t dpia_hpd_int_enable_supported: 1; /* 1 if dpia hpd int enable supported */ + uint32_t usb4_dpia_bw_alloc_supported: 1; /* 1 if USB4 dpia BW allocation supported */ - uint32_t reserved : 16; /**< reserved */ + uint32_t reserved : 15; /**< reserved */ } bits; /**< boot bits */ uint32_t all; /**< 32-bit access to bits */ }; @@ -731,6 +738,11 @@ enum dmub_cmd_type { */ /** + * Command type used for all SECURE_DISPLAY commands. + */ + DMUB_CMD__SECURE_DISPLAY = 85, + + /** * Command type used to set DPIA HPD interrupt state */ DMUB_CMD__DPIA_HPD_INT_ENABLE = 86, @@ -1866,9 +1878,13 @@ struct dmub_cmd_psr_copy_settings_data { */ uint8_t use_phy_fsm; /** + * frame delay for frame re-lock + */ + uint8_t relock_delay_frame_cnt; + /** * Explicit padding to 2 byte boundary. */ - uint8_t pad3[2]; + uint8_t pad3; }; /** @@ -3144,6 +3160,33 @@ struct dmub_rb_cmd_get_usbc_cable_id { }; /** + * Command type of a DMUB_CMD__SECURE_DISPLAY command + */ +enum dmub_cmd_secure_display_type { + DMUB_CMD__SECURE_DISPLAY_TEST_CMD = 0, /* test command to only check if inbox message works */ + DMUB_CMD__SECURE_DISPLAY_CRC_STOP_UPDATE, + DMUB_CMD__SECURE_DISPLAY_CRC_WIN_NOTIFY +}; + +/** + * Definition of a DMUB_CMD__SECURE_DISPLAY command + */ +struct dmub_rb_cmd_secure_display { + struct dmub_cmd_header header; + /** + * Data passed from driver to dmub firmware. + */ + struct dmub_cmd_roi_info { + uint16_t x_start; + uint16_t x_end; + uint16_t y_start; + uint16_t y_end; + uint8_t otg_id; + uint8_t phy_id; + } roi_info; +}; + +/** * union dmub_rb_cmd - DMUB inbox command. */ union dmub_rb_cmd { @@ -3348,6 +3391,11 @@ union dmub_rb_cmd { */ struct dmub_rb_cmd_query_hpd_state query_hpd; /** + * Definition of a DMUB_CMD__SECURE_DISPLAY command. + */ + struct dmub_rb_cmd_secure_display secure_display; + + /** * Definition of a DMUB_CMD__DPIA_HPD_INT_ENABLE command. */ struct dmub_rb_cmd_dpia_hpd_int_enable dpia_hpd_int_enable; diff --git a/drivers/gpu/drm/amd/display/modules/color/color_gamma.c b/drivers/gpu/drm/amd/display/modules/color/color_gamma.c index 447a0ec9cbe2..f6034213c700 100644 --- a/drivers/gpu/drm/amd/display/modules/color/color_gamma.c +++ b/drivers/gpu/drm/amd/display/modules/color/color_gamma.c @@ -61,7 +61,7 @@ static const int32_t numerator01[] = { 31308, 180000, 0, 0, 0}; static const int32_t numerator02[] = { 12920, 4500, 0, 0, 0}; static const int32_t numerator03[] = { 55, 99, 0, 0, 0}; static const int32_t numerator04[] = { 55, 99, 0, 0, 0}; -static const int32_t numerator05[] = { 2400, 2200, 2200, 2400, 2600}; +static const int32_t numerator05[] = { 2400, 2222, 2200, 2400, 2600}; /* one-time setup of X points */ void setup_x_points_distribution(void) diff --git a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c index 0f39ab9dc5b4..c2e00f7b8381 100644 --- a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c +++ b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c @@ -688,10 +688,10 @@ static void build_vrr_infopacket_fs2_data(enum color_transfer_func app_tf, if (app_tf != TRANSFER_FUNC_UNKNOWN) { infopacket->valid = true; - infopacket->sb[6] |= 0x08; // PB6 = [Bit 3 = Native Color Active] - - if (app_tf == TRANSFER_FUNC_GAMMA_22) { - infopacket->sb[9] |= 0x04; // PB6 = [Bit 2 = Gamma 2.2 EOTF Active] + if (app_tf != TRANSFER_FUNC_PQ2084) { + infopacket->sb[6] |= 0x08; // PB6 = [Bit 3 = Native Color Active] + if (app_tf == TRANSFER_FUNC_GAMMA_22) + infopacket->sb[9] |= 0x04; // PB6 = [Bit 2 = Gamma 2.2 EOTF Active] } } } diff --git a/drivers/gpu/drm/amd/display/modules/inc/mod_info_packet.h b/drivers/gpu/drm/amd/display/modules/inc/mod_info_packet.h index 1d8b746b02f2..edf5845f6a1f 100644 --- a/drivers/gpu/drm/amd/display/modules/inc/mod_info_packet.h +++ b/drivers/gpu/drm/amd/display/modules/inc/mod_info_packet.h @@ -35,7 +35,8 @@ struct mod_vrr_params; void mod_build_vsc_infopacket(const struct dc_stream_state *stream, struct dc_info_packet *info_packet, - enum dc_color_space cs); + enum dc_color_space cs, + enum color_transfer_func tf); void mod_build_hf_vsif_infopacket(const struct dc_stream_state *stream, struct dc_info_packet *info_packet); diff --git a/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c b/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c index 27ceba9d6d65..69691058ab89 100644 --- a/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c +++ b/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c @@ -132,7 +132,8 @@ enum ColorimetryYCCDP { void mod_build_vsc_infopacket(const struct dc_stream_state *stream, struct dc_info_packet *info_packet, - enum dc_color_space cs) + enum dc_color_space cs, + enum color_transfer_func tf) { unsigned int vsc_packet_revision = vsc_packet_undefined; unsigned int i; @@ -382,6 +383,9 @@ void mod_build_vsc_infopacket(const struct dc_stream_state *stream, colorimetryFormat = ColorimetryYCC_DP_AdobeYCC; else if (cs == COLOR_SPACE_2020_YCBCR) colorimetryFormat = ColorimetryYCC_DP_ITU2020YCbCr; + + if (cs == COLOR_SPACE_2020_YCBCR && tf == TRANSFER_FUNC_GAMMA_22) + colorimetryFormat = ColorimetryYCC_DP_ITU709; break; default: diff --git a/drivers/gpu/drm/amd/include/atombios.h b/drivers/gpu/drm/amd/include/atombios.h index 15943bc21bc5..4dc738c51771 100644 --- a/drivers/gpu/drm/amd/include/atombios.h +++ b/drivers/gpu/drm/amd/include/atombios.h @@ -4107,7 +4107,7 @@ typedef struct _ATOM_FAKE_EDID_PATCH_RECORD { UCHAR ucRecordType; UCHAR ucFakeEDIDLength; // = 128 means EDID length is 128 bytes, otherwise the EDID length = ucFakeEDIDLength*128 - UCHAR ucFakeEDIDString[1]; // This actually has ucFakeEdidLength elements. + UCHAR ucFakeEDIDString[]; // This actually has ucFakeEdidLength elements. } ATOM_FAKE_EDID_PATCH_RECORD; typedef struct _ATOM_PANEL_RESOLUTION_PATCH_RECORD @@ -4386,7 +4386,7 @@ typedef struct _ATOM_GPIO_PIN_ASSIGNMENT typedef struct _ATOM_GPIO_PIN_LUT { ATOM_COMMON_TABLE_HEADER sHeader; - ATOM_GPIO_PIN_ASSIGNMENT asGPIO_Pin[1]; + ATOM_GPIO_PIN_ASSIGNMENT asGPIO_Pin[]; }ATOM_GPIO_PIN_LUT; /****************************************************************************/ @@ -4513,7 +4513,7 @@ typedef struct _ATOM_DISPLAY_OBJECT_PATH USHORT usSize; //the size of ATOM_DISPLAY_OBJECT_PATH USHORT usConnObjectId; //Connector Object ID USHORT usGPUObjectId; //GPU ID - USHORT usGraphicObjIds[1]; //1st Encoder Obj source from GPU to last Graphic Obj destinate to connector. + USHORT usGraphicObjIds[]; //1st Encoder Obj source from GPU to last Graphic Obj destinate to connector. }ATOM_DISPLAY_OBJECT_PATH; typedef struct _ATOM_DISPLAY_EXTERNAL_OBJECT_PATH @@ -4530,7 +4530,7 @@ typedef struct _ATOM_DISPLAY_OBJECT_PATH_TABLE UCHAR ucNumOfDispPath; UCHAR ucVersion; UCHAR ucPadding[2]; - ATOM_DISPLAY_OBJECT_PATH asDispPath[1]; + ATOM_DISPLAY_OBJECT_PATH asDispPath[]; }ATOM_DISPLAY_OBJECT_PATH_TABLE; typedef struct _ATOM_OBJECT //each object has this structure @@ -4545,7 +4545,7 @@ typedef struct _ATOM_OBJECT_TABLE //Above 4 object table { UCHAR ucNumberOfObjects; UCHAR ucPadding[3]; - ATOM_OBJECT asObjects[1]; + ATOM_OBJECT asObjects[]; }ATOM_OBJECT_TABLE; typedef struct _ATOM_SRC_DST_TABLE_FOR_ONE_OBJECT //usSrcDstTableOffset pointing to this structure @@ -4733,7 +4733,7 @@ typedef struct _ATOM_CONNECTOR_DEVICE_TAG_RECORD ATOM_COMMON_RECORD_HEADER sheader; UCHAR ucNumberOfDevice; UCHAR ucReserved; - ATOM_CONNECTOR_DEVICE_TAG asDeviceTag[1]; //This Id is same as "ATOM_DEVICE_XXX_SUPPORT", 1 is only for allocation + ATOM_CONNECTOR_DEVICE_TAG asDeviceTag[]; //This Id is same as "ATOM_DEVICE_XXX_SUPPORT" }ATOM_CONNECTOR_DEVICE_TAG_RECORD; @@ -4793,7 +4793,7 @@ typedef struct _ATOM_OBJECT_GPIO_CNTL_RECORD ATOM_COMMON_RECORD_HEADER sheader; UCHAR ucFlags; // Future expnadibility UCHAR ucNumberOfPins; // Number of GPIO pins used to control the object - ATOM_GPIO_PIN_CONTROL_PAIR asGpio[1]; // the real gpio pin pair determined by number of pins ucNumberOfPins + ATOM_GPIO_PIN_CONTROL_PAIR asGpio[]; // the real gpio pin pair determined by number of pins ucNumberOfPins }ATOM_OBJECT_GPIO_CNTL_RECORD; //Definitions for GPIO pin state @@ -4982,7 +4982,7 @@ typedef struct _ATOM_BRACKET_LAYOUT_RECORD UCHAR ucWidth; UCHAR ucConnNum; UCHAR ucReserved; - ATOM_CONNECTOR_LAYOUT_INFO asConnInfo[1]; + ATOM_CONNECTOR_LAYOUT_INFO asConnInfo[]; }ATOM_BRACKET_LAYOUT_RECORD; @@ -5146,7 +5146,7 @@ typedef struct _ATOM_I2C_VOLTAGE_OBJECT_V3 UCHAR ucVoltageControlOffset; UCHAR ucVoltageControlFlag; // Bit0: 0 - One byte data; 1 - Two byte data UCHAR ulReserved[3]; - VOLTAGE_LUT_ENTRY asVolI2cLut[1]; // end with 0xff + VOLTAGE_LUT_ENTRY asVolI2cLut[]; // end with 0xff }ATOM_I2C_VOLTAGE_OBJECT_V3; // ATOM_I2C_VOLTAGE_OBJECT_V3.ucVoltageControlFlag @@ -5161,7 +5161,7 @@ typedef struct _ATOM_GPIO_VOLTAGE_OBJECT_V3 UCHAR ucPhaseDelay; // phase delay in unit of micro second UCHAR ucReserved; ULONG ulGpioMaskVal; // GPIO Mask value - VOLTAGE_LUT_ENTRY_V2 asVolGpioLut[1]; + VOLTAGE_LUT_ENTRY_V2 asVolGpioLut[]; }ATOM_GPIO_VOLTAGE_OBJECT_V3; typedef struct _ATOM_LEAKAGE_VOLTAGE_OBJECT_V3 @@ -5171,7 +5171,7 @@ typedef struct _ATOM_LEAKAGE_VOLTAGE_OBJECT_V3 UCHAR ucLeakageEntryNum; // indicate the entry number of LeakageId/Voltage Lut table UCHAR ucReserved[2]; ULONG ulMaxVoltageLevel; - LEAKAGE_VOLTAGE_LUT_ENTRY_V2 asLeakageIdLut[1]; + LEAKAGE_VOLTAGE_LUT_ENTRY_V2 asLeakageIdLut[]; }ATOM_LEAKAGE_VOLTAGE_OBJECT_V3; @@ -6599,7 +6599,7 @@ typedef struct _ATOM_FUSION_SYSTEM_INFO_V3 typedef struct _ATOM_I2C_DATA_RECORD { UCHAR ucNunberOfBytes; //Indicates how many bytes SW needs to write to the external ASIC for one block, besides to "Start" and "Stop" - UCHAR ucI2CData[1]; //I2C data in bytes, should be less than 16 bytes usually + UCHAR ucI2CData[]; //I2C data in bytes, should be less than 16 bytes usually }ATOM_I2C_DATA_RECORD; @@ -6610,14 +6610,14 @@ typedef struct _ATOM_I2C_DEVICE_SETUP_INFO UCHAR ucSSChipID; //SS chip being used UCHAR ucSSChipSlaveAddr; //Slave Address to set up this SS chip UCHAR ucNumOfI2CDataRecords; //number of data block - ATOM_I2C_DATA_RECORD asI2CData[1]; + ATOM_I2C_DATA_RECORD asI2CData[]; }ATOM_I2C_DEVICE_SETUP_INFO; //========================================================================================== typedef struct _ATOM_ASIC_MVDD_INFO { ATOM_COMMON_TABLE_HEADER sHeader; - ATOM_I2C_DEVICE_SETUP_INFO asI2CSetup[1]; + ATOM_I2C_DEVICE_SETUP_INFO asI2CSetup[]; }ATOM_ASIC_MVDD_INFO; //========================================================================================== @@ -6679,7 +6679,7 @@ typedef struct _ATOM_ASIC_INTERNAL_SS_INFO typedef struct _ATOM_ASIC_INTERNAL_SS_INFO_V2 { ATOM_COMMON_TABLE_HEADER sHeader; - ATOM_ASIC_SS_ASSIGNMENT_V2 asSpreadSpectrum[1]; //this is point only. + ATOM_ASIC_SS_ASSIGNMENT_V2 asSpreadSpectrum[]; //this is point only. }ATOM_ASIC_INTERNAL_SS_INFO_V2; typedef struct _ATOM_ASIC_SS_ASSIGNMENT_V3 @@ -6701,7 +6701,7 @@ typedef struct _ATOM_ASIC_SS_ASSIGNMENT_V3 typedef struct _ATOM_ASIC_INTERNAL_SS_INFO_V3 { ATOM_COMMON_TABLE_HEADER sHeader; - ATOM_ASIC_SS_ASSIGNMENT_V3 asSpreadSpectrum[1]; //this is pointer only. + ATOM_ASIC_SS_ASSIGNMENT_V3 asSpreadSpectrum[]; //this is pointer only. }ATOM_ASIC_INTERNAL_SS_INFO_V3; diff --git a/drivers/gpu/drm/amd/include/atomfirmware.h b/drivers/gpu/drm/amd/include/atomfirmware.h index ff855cb21d3f..bbe1337a8cee 100644 --- a/drivers/gpu/drm/amd/include/atomfirmware.h +++ b/drivers/gpu/drm/amd/include/atomfirmware.h @@ -705,20 +705,65 @@ struct atom_gpio_pin_lut_v2_1 }; -/* - *************************************************************************** - Data Table vram_usagebyfirmware structure - *************************************************************************** -*/ +/* + * VBIOS/PRE-OS always reserve a FB region at the top of frame buffer. driver should not write + * access that region. driver can allocate their own reservation region as long as it does not + * overlap firwmare's reservation region. + * if (pre-NV1X) atom data table firmwareInfoTable version < 3.3: + * in this case, atom data table vram_usagebyfirmwareTable version always <= 2.1 + * if VBIOS/UEFI GOP is posted: + * VBIOS/UEFIGOP update used_by_firmware_in_kb = total reserved size by VBIOS + * update start_address_in_kb = total_mem_size_in_kb - used_by_firmware_in_kb; + * ( total_mem_size_in_kb = reg(CONFIG_MEMSIZE)<<10) + * driver can allocate driver reservation region under firmware reservation, + * used_by_driver_in_kb = driver reservation size + * driver reservation start address = (start_address_in_kb - used_by_driver_in_kb) + * Comment1[hchan]: There is only one reservation at the beginning of the FB reserved by + * host driver. Host driver would overwrite the table with the following + * used_by_firmware_in_kb = total reserved size for pf-vf info exchange and + * set SRIOV_MSG_SHARE_RESERVATION mask start_address_in_kb = 0 + * else there is no VBIOS reservation region: + * driver must allocate driver reservation region at top of FB. + * driver set used_by_driver_in_kb = driver reservation size + * driver reservation start address = (total_mem_size_in_kb - used_by_driver_in_kb) + * same as Comment1 + * else (NV1X and after): + * if VBIOS/UEFI GOP is posted: + * VBIOS/UEFIGOP update: + * used_by_firmware_in_kb = atom_firmware_Info_v3_3.fw_reserved_size_in_kb; + * start_address_in_kb = total_mem_size_in_kb - used_by_firmware_in_kb; + * (total_mem_size_in_kb = reg(CONFIG_MEMSIZE)<<10) + * if vram_usagebyfirmwareTable version <= 2.1: + * driver can allocate driver reservation region under firmware reservation, + * driver set used_by_driver_in_kb = driver reservation size + * driver reservation start address = start_address_in_kb - used_by_driver_in_kb + * same as Comment1 + * else driver can: + * allocate it reservation any place as long as it does overlap pre-OS FW reservation area + * set used_by_driver_region0_in_kb = driver reservation size + * set driver_region0_start_address_in_kb = driver reservation region start address + * Comment2[hchan]: Host driver can set used_by_firmware_in_kb and start_address_in_kb to + * zero as the reservation for VF as it doesn’t exist. And Host driver should also + * update atom_firmware_Info table to remove the same VBIOS reservation as well. + */ struct vram_usagebyfirmware_v2_1 { - struct atom_common_table_header table_header; - uint32_t start_address_in_kb; - uint16_t used_by_firmware_in_kb; - uint16_t used_by_driver_in_kb; + struct atom_common_table_header table_header; + uint32_t start_address_in_kb; + uint16_t used_by_firmware_in_kb; + uint16_t used_by_driver_in_kb; }; +struct vram_usagebyfirmware_v2_2 { + struct atom_common_table_header table_header; + uint32_t fw_region_start_address_in_kb; + uint16_t used_by_firmware_in_kb; + uint16_t reserved; + uint32_t driver_region0_start_address_in_kb; + uint32_t used_by_driver_region0_in_kb; + uint32_t reserved32[7]; +}; /* *************************************************************************** diff --git a/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_4_0.h b/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_4_0.h index a81138c9e491..03cfa0517df2 100644 --- a/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_4_0.h +++ b/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_4_0.h @@ -38,4 +38,7 @@ #define VCN_4_0__SRCID__JPEG6_DECODE 174 // 0xae JRBC6 Decode interrupt #define VCN_4_0__SRCID__JPEG7_DECODE 175 // 0xaf JRBC7 Decode interrupt +#define VCN_4_0__SRCID_UVD_POISON 160 +#define VCN_4_0__SRCID_DJPEG0_POISON 161 +#define VCN_4_0__SRCID_EJPEG0_POISON 162 #endif diff --git a/drivers/gpu/drm/amd/include/kgd_pp_interface.h b/drivers/gpu/drm/amd/include/kgd_pp_interface.h index a40ead44778a..d18162e9ed1d 100644 --- a/drivers/gpu/drm/amd/include/kgd_pp_interface.h +++ b/drivers/gpu/drm/amd/include/kgd_pp_interface.h @@ -354,7 +354,8 @@ struct amd_pm_funcs { int (*get_power_profile_mode)(void *handle, char *buf); int (*set_power_profile_mode)(void *handle, long *input, uint32_t size); int (*set_fine_grain_clk_vol)(void *handle, uint32_t type, long *input, uint32_t size); - int (*odn_edit_dpm_table)(void *handle, uint32_t type, long *input, uint32_t size); + int (*odn_edit_dpm_table)(void *handle, enum PP_OD_DPM_TABLE_COMMAND type, + long *input, uint32_t size); int (*set_mp1_state)(void *handle, enum pp_mp1_state mp1_state); int (*smu_i2c_bus_access)(void *handle, bool acquire); int (*gfx_state_change_set)(void *handle, uint32_t state); diff --git a/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c b/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c index ec055858eb95..304190d5c9d2 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c +++ b/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c @@ -838,7 +838,8 @@ static int pp_set_fine_grain_clk_vol(void *handle, uint32_t type, long *input, u return hwmgr->hwmgr_func->set_fine_grain_clk_vol(hwmgr, type, input, size); } -static int pp_odn_edit_dpm_table(void *handle, uint32_t type, long *input, uint32_t size) +static int pp_odn_edit_dpm_table(void *handle, enum PP_OD_DPM_TABLE_COMMAND type, + long *input, uint32_t size) { struct pp_hwmgr *hwmgr = handle; @@ -1507,7 +1508,7 @@ static void pp_pm_compute_clocks(void *handle) struct pp_hwmgr *hwmgr = handle; struct amdgpu_device *adev = hwmgr->adev; - if (!amdgpu_device_has_dc_support(adev)) { + if (!adev->dc_enabled) { amdgpu_dpm_get_active_displays(adev); adev->pm.pm_display_cfg.num_display = adev->pm.dpm.new_active_crtc_count; adev->pm.pm_display_cfg.vrefresh = amdgpu_dpm_get_vrefresh(adev); diff --git a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/pp_psm.c b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/pp_psm.c index 67d7da0b6fed..1d829402cd2e 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/pp_psm.c +++ b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/pp_psm.c @@ -75,8 +75,10 @@ int psm_init_power_state_table(struct pp_hwmgr *hwmgr) for (i = 0; i < table_entries; i++) { result = hwmgr->hwmgr_func->get_pp_table_entry(hwmgr, i, state); if (result) { + kfree(hwmgr->current_ps); kfree(hwmgr->request_ps); kfree(hwmgr->ps); + hwmgr->current_ps = NULL; hwmgr->request_ps = NULL; hwmgr->ps = NULL; return -EINVAL; diff --git a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/vega10_thermal.c b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/vega10_thermal.c index dad3e3741a4e..190af79f3236 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/vega10_thermal.c +++ b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/vega10_thermal.c @@ -67,22 +67,21 @@ int vega10_fan_ctrl_get_fan_speed_info(struct pp_hwmgr *hwmgr, int vega10_fan_ctrl_get_fan_speed_pwm(struct pp_hwmgr *hwmgr, uint32_t *speed) { - uint32_t current_rpm; - uint32_t percent = 0; - - if (hwmgr->thermal_controller.fanInfo.bNoFan) - return 0; + struct amdgpu_device *adev = hwmgr->adev; + uint32_t duty100, duty; + uint64_t tmp64; - if (vega10_get_current_rpm(hwmgr, ¤t_rpm)) - return -1; + duty100 = REG_GET_FIELD(RREG32_SOC15(THM, 0, mmCG_FDO_CTRL1), + CG_FDO_CTRL1, FMAX_DUTY100); + duty = REG_GET_FIELD(RREG32_SOC15(THM, 0, mmCG_THERMAL_STATUS), + CG_THERMAL_STATUS, FDO_PWM_DUTY); - if (hwmgr->thermal_controller. - advanceFanControlParameters.usMaxFanRPM != 0) - percent = current_rpm * 255 / - hwmgr->thermal_controller. - advanceFanControlParameters.usMaxFanRPM; + if (!duty100) + return -EINVAL; - *speed = MIN(percent, 255); + tmp64 = (uint64_t)duty * 255; + do_div(tmp64, duty100); + *speed = MIN((uint32_t)tmp64, 255); return 0; } diff --git a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/vega20_hwmgr.c b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/vega20_hwmgr.c index 97b3ad369046..b30684c84e20 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/vega20_hwmgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/vega20_hwmgr.c @@ -2961,7 +2961,8 @@ static int vega20_odn_edit_dpm_table(struct pp_hwmgr *hwmgr, data->od8_settings.od8_settings_array; OverDriveTable_t *od_table = &(data->smc_state_table.overdrive_table); - int32_t input_index, input_clk, input_vol, i; + int32_t input_clk, input_vol, i; + uint32_t input_index; int od8_id; int ret; diff --git a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c index 4fe75dd2b329..20e5f66f853f 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c @@ -1156,22 +1156,21 @@ static int smu_smc_hw_setup(struct smu_context *smu) uint64_t features_supported; int ret = 0; - if (adev->in_suspend && smu_is_dpm_running(smu)) { - dev_info(adev->dev, "dpm has been enabled\n"); - /* this is needed specifically */ - switch (adev->ip_versions[MP1_HWIP][0]) { - case IP_VERSION(11, 0, 7): - case IP_VERSION(11, 0, 11): - case IP_VERSION(11, 5, 0): - case IP_VERSION(11, 0, 12): + switch (adev->ip_versions[MP1_HWIP][0]) { + case IP_VERSION(11, 0, 7): + case IP_VERSION(11, 0, 11): + case IP_VERSION(11, 5, 0): + case IP_VERSION(11, 0, 12): + if (adev->in_suspend && smu_is_dpm_running(smu)) { + dev_info(adev->dev, "dpm has been enabled\n"); ret = smu_system_features_control(smu, true); if (ret) dev_err(adev->dev, "Failed system features control!\n"); - break; - default: - break; + return ret; } - return ret; + break; + default: + break; } ret = smu_init_display_count(smu, 0); @@ -1449,6 +1448,7 @@ static int smu_disable_dpms(struct smu_context *smu) switch (adev->ip_versions[MP1_HWIP][0]) { case IP_VERSION(13, 0, 0): case IP_VERSION(13, 0, 7): + case IP_VERSION(13, 0, 10): return 0; default: break; @@ -1517,7 +1517,7 @@ static int smu_disable_dpms(struct smu_context *smu) } if (adev->ip_versions[GC_HWIP][0] >= IP_VERSION(9, 4, 2) && - adev->gfx.rlc.funcs->stop) + !amdgpu_sriov_vf(adev) && adev->gfx.rlc.funcs->stop) adev->gfx.rlc.funcs->stop(adev); return ret; diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h index e2fa3b066b96..44bbf17e4bef 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h @@ -568,6 +568,10 @@ struct smu_context u32 param_reg; u32 msg_reg; u32 resp_reg; + + u32 debug_param_reg; + u32 debug_msg_reg; + u32 debug_resp_reg; }; struct i2c_adapter; @@ -1388,6 +1392,14 @@ enum smu_cmn2asic_mapping_type { CMN2ASIC_MAPPING_WORKLOAD, }; +enum smu_baco_seq { + BACO_SEQ_BACO = 0, + BACO_SEQ_MSR, + BACO_SEQ_BAMACO, + BACO_SEQ_ULPS, + BACO_SEQ_COUNT, +}; + #define MSG_MAP(msg, index, valid_in_vf) \ [SMU_MSG_##msg] = {1, (index), (valid_in_vf)} diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu13_driver_if_v13_0_0.h b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu13_driver_if_v13_0_0.h index 063f4a737605..b76f0f7e4299 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu13_driver_if_v13_0_0.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu13_driver_if_v13_0_0.h @@ -25,7 +25,7 @@ #define SMU13_DRIVER_IF_V13_0_0_H //Increment this version if SkuTable_t or BoardTable_t change -#define PPTABLE_VERSION 0x24 +#define PPTABLE_VERSION 0x26 #define NUM_GFXCLK_DPM_LEVELS 16 #define NUM_SOCCLK_DPM_LEVELS 8 @@ -109,6 +109,22 @@ #define FEATURE_SPARE_63_BIT 63 #define NUM_FEATURES 64 +#define ALLOWED_FEATURE_CTRL_DEFAULT 0xFFFFFFFFFFFFFFFFULL +#define ALLOWED_FEATURE_CTRL_SCPM ((1 << FEATURE_DPM_GFXCLK_BIT) | \ + (1 << FEATURE_DPM_GFX_POWER_OPTIMIZER_BIT) | \ + (1 << FEATURE_DPM_UCLK_BIT) | \ + (1 << FEATURE_DPM_FCLK_BIT) | \ + (1 << FEATURE_DPM_SOCCLK_BIT) | \ + (1 << FEATURE_DPM_MP0CLK_BIT) | \ + (1 << FEATURE_DPM_LINK_BIT) | \ + (1 << FEATURE_DPM_DCN_BIT) | \ + (1 << FEATURE_DS_GFXCLK_BIT) | \ + (1 << FEATURE_DS_SOCCLK_BIT) | \ + (1 << FEATURE_DS_FCLK_BIT) | \ + (1 << FEATURE_DS_LCLK_BIT) | \ + (1 << FEATURE_DS_DCFCLK_BIT) | \ + (1 << FEATURE_DS_UCLK_BIT)) + //For use with feature control messages typedef enum { FEATURE_PWR_ALL, @@ -133,6 +149,7 @@ typedef enum { #define DEBUG_OVERRIDE_DISABLE_DFLL 0x00000200 #define DEBUG_OVERRIDE_ENABLE_RLC_VF_BRINGUP_MODE 0x00000400 #define DEBUG_OVERRIDE_DFLL_MASTER_MODE 0x00000800 +#define DEBUG_OVERRIDE_ENABLE_PROFILING_MODE 0x00001000 // VR Mapping Bit Defines #define VR_MAPPING_VR_SELECT_MASK 0x01 @@ -262,15 +279,15 @@ typedef enum { } I2cControllerPort_e; typedef enum { - I2C_CONTROLLER_NAME_VR_GFX = 0, - I2C_CONTROLLER_NAME_VR_SOC, - I2C_CONTROLLER_NAME_VR_VMEMP, - I2C_CONTROLLER_NAME_VR_VDDIO, - I2C_CONTROLLER_NAME_LIQUID0, - I2C_CONTROLLER_NAME_LIQUID1, - I2C_CONTROLLER_NAME_PLX, - I2C_CONTROLLER_NAME_OTHER, - I2C_CONTROLLER_NAME_COUNT, + I2C_CONTROLLER_NAME_VR_GFX = 0, + I2C_CONTROLLER_NAME_VR_SOC, + I2C_CONTROLLER_NAME_VR_VMEMP, + I2C_CONTROLLER_NAME_VR_VDDIO, + I2C_CONTROLLER_NAME_LIQUID0, + I2C_CONTROLLER_NAME_LIQUID1, + I2C_CONTROLLER_NAME_PLX, + I2C_CONTROLLER_NAME_FAN_INTAKE, + I2C_CONTROLLER_NAME_COUNT, } I2cControllerName_e; typedef enum { @@ -282,16 +299,17 @@ typedef enum { I2C_CONTROLLER_THROTTLER_LIQUID0, I2C_CONTROLLER_THROTTLER_LIQUID1, I2C_CONTROLLER_THROTTLER_PLX, + I2C_CONTROLLER_THROTTLER_FAN_INTAKE, I2C_CONTROLLER_THROTTLER_INA3221, I2C_CONTROLLER_THROTTLER_COUNT, } I2cControllerThrottler_e; typedef enum { - I2C_CONTROLLER_PROTOCOL_VR_XPDE132G5, - I2C_CONTROLLER_PROTOCOL_VR_IR35217, - I2C_CONTROLLER_PROTOCOL_TMP_TMP102A, - I2C_CONTROLLER_PROTOCOL_INA3221, - I2C_CONTROLLER_PROTOCOL_COUNT, + I2C_CONTROLLER_PROTOCOL_VR_XPDE132G5, + I2C_CONTROLLER_PROTOCOL_VR_IR35217, + I2C_CONTROLLER_PROTOCOL_TMP_MAX31875, + I2C_CONTROLLER_PROTOCOL_INA3221, + I2C_CONTROLLER_PROTOCOL_COUNT, } I2cControllerProtocol_e; typedef struct { @@ -658,13 +676,20 @@ typedef struct { #define PP_NUM_OD_VF_CURVE_POINTS PP_NUM_RTAVFS_PWL_ZONES + 1 +typedef enum { + FAN_MODE_AUTO = 0, + FAN_MODE_MANUAL_LINEAR, +} FanMode_e; typedef struct { uint32_t FeatureCtrlMask; //Voltage control int16_t VoltageOffsetPerZoneBoundary[PP_NUM_OD_VF_CURVE_POINTS]; - uint16_t reserved[2]; + uint16_t VddGfxVmax; // in mV + + uint8_t IdlePwrSavingFeaturesCtrl; + uint8_t RuntimePwrSavingFeaturesCtrl; //Frequency changes int16_t GfxclkFmin; // MHz @@ -674,7 +699,7 @@ typedef struct { //PPT int16_t Ppt; // % - int16_t reserved1; + int16_t Tdc; //Fan control uint8_t FanLinearPwmPoints[NUM_OD_FAN_MAX_POINTS]; @@ -701,16 +726,19 @@ typedef struct { uint32_t FeatureCtrlMask; int16_t VoltageOffsetPerZoneBoundary; - uint16_t reserved[2]; + uint16_t VddGfxVmax; // in mV + + uint8_t IdlePwrSavingFeaturesCtrl; + uint8_t RuntimePwrSavingFeaturesCtrl; - uint16_t GfxclkFmin; // MHz - uint16_t GfxclkFmax; // MHz + int16_t GfxclkFmin; // MHz + int16_t GfxclkFmax; // MHz uint16_t UclkFmin; // MHz uint16_t UclkFmax; // MHz //PPT int16_t Ppt; // % - int16_t reserved1; + int16_t Tdc; uint8_t FanLinearPwmPoints; uint8_t FanLinearTempPoints; @@ -857,7 +885,8 @@ typedef struct { uint16_t FanStartTempMin; uint16_t FanStartTempMax; - uint32_t Spare[12]; + uint16_t PowerMinPpt0[POWER_SOURCE_COUNT]; + uint32_t Spare[11]; } MsgLimits_t; @@ -1041,7 +1070,17 @@ typedef struct { uint32_t GfxoffSpare[15]; // GFX GPO - uint32_t GfxGpoSpare[16]; + uint32_t DfllBtcMasterScalerM; + int32_t DfllBtcMasterScalerB; + uint32_t DfllBtcSlaveScalerM; + int32_t DfllBtcSlaveScalerB; + + uint32_t DfllPccAsWaitCtrl; //GDFLL_AS_WAIT_CTRL_PCC register value to be passed to RLC msg + uint32_t DfllPccAsStepCtrl; //GDFLL_AS_STEP_CTRL_PCC register value to be passed to RLC msg + + uint32_t DfllL2FrequencyBoostM; //Unitless (float) + uint32_t DfllL2FrequencyBoostB; //In MHz (integer) + uint32_t GfxGpoSpare[8]; // GFX DCS @@ -1114,12 +1153,14 @@ typedef struct { uint16_t IntakeTempHighIntakeAcousticLimit; uint16_t IntakeTempAcouticLimitReleaseRate; - uint16_t FanStalledTempLimitOffset; + int16_t FanAbnormalTempLimitOffset; uint16_t FanStalledTriggerRpm; - uint16_t FanAbnormalTriggerRpm; - uint16_t FanPadding; + uint16_t FanAbnormalTriggerRpmCoeff; + uint16_t FanAbnormalDetectionEnable; - uint32_t FanSpare[14]; + uint8_t FanIntakeSensorSupport; + uint8_t FanIntakePadding[3]; + uint32_t FanSpare[13]; // SECTION: VDD_GFX AVFS @@ -1198,8 +1239,13 @@ typedef struct { int16_t TotalBoardPowerM; int16_t TotalBoardPowerB; + //PMFW-11158 + QuadraticInt_t qFeffCoeffGameClock[POWER_SOURCE_COUNT]; + QuadraticInt_t qFeffCoeffBaseClock[POWER_SOURCE_COUNT]; + QuadraticInt_t qFeffCoeffBoostClock[POWER_SOURCE_COUNT]; + // SECTION: Sku Reserved - uint32_t Spare[61]; + uint32_t Spare[43]; // Padding for MMHUB - do not modify this uint32_t MmHubPadding[8]; @@ -1288,8 +1334,11 @@ typedef struct { uint32_t PostVoltageSetBacoDelay; // in microseconds. Amount of time FW will wait after power good is established or PSI0 command is issued uint32_t BacoEntryDelay; // in milliseconds. Amount of time FW will wait to trigger BACO entry after receiving entry notification from OS + uint8_t FuseWritePowerMuxPresent; + uint8_t FuseWritePadding[3]; + // SECTION: Board Reserved - uint32_t BoardSpare[64]; + uint32_t BoardSpare[63]; // SECTION: Structure Padding @@ -1381,7 +1430,7 @@ typedef struct { uint16_t AverageTotalBoardPower; uint16_t AvgTemperature[TEMP_COUNT]; - uint16_t TempPadding; + uint16_t AvgTemperatureFanIntake; uint8_t PcieRate ; uint8_t PcieWidth ; @@ -1550,5 +1599,7 @@ typedef struct { #define IH_INTERRUPT_CONTEXT_ID_AUDIO_D0 0x5 #define IH_INTERRUPT_CONTEXT_ID_AUDIO_D3 0x6 #define IH_INTERRUPT_CONTEXT_ID_THERMAL_THROTTLING 0x7 +#define IH_INTERRUPT_CONTEXT_ID_FAN_ABNORMAL 0x8 +#define IH_INTERRUPT_CONTEXT_ID_FAN_RECOVERY 0x9 #endif diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_4_ppsmc.h b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_4_ppsmc.h index d9b0cd752200..f4d6c07b56ea 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_4_ppsmc.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_4_ppsmc.h @@ -54,14 +54,14 @@ #define PPSMC_MSG_TestMessage 0x01 ///< To check if PMFW is alive and responding. Requirement specified by PMFW team #define PPSMC_MSG_GetPmfwVersion 0x02 ///< Get PMFW version #define PPSMC_MSG_GetDriverIfVersion 0x03 ///< Get PMFW_DRIVER_IF version -#define PPSMC_MSG_EnableGfxOff 0x04 ///< Enable GFXOFF -#define PPSMC_MSG_DisableGfxOff 0x05 ///< Disable GFXOFF +#define PPSMC_MSG_SPARE0 0x04 ///< SPARE +#define PPSMC_MSG_SPARE1 0x05 ///< SPARE #define PPSMC_MSG_PowerDownVcn 0x06 ///< Power down VCN #define PPSMC_MSG_PowerUpVcn 0x07 ///< Power up VCN; VCN is power gated by default #define PPSMC_MSG_SetHardMinVcn 0x08 ///< For wireless display #define PPSMC_MSG_SetSoftMinGfxclk 0x09 ///< Set SoftMin for GFXCLK, argument is frequency in MHz -#define PPSMC_MSG_ActiveProcessNotify 0x0A ///< Needs update -#define PPSMC_MSG_ForcePowerDownGfx 0x0B ///< Force power down GFX, i.e. enter GFXOFF +#define PPSMC_MSG_SPARE2 0x0A ///< SPARE +#define PPSMC_MSG_SPARE3 0x0B ///< SPARE #define PPSMC_MSG_PrepareMp1ForUnload 0x0C ///< Prepare PMFW for GFX driver unload #define PPSMC_MSG_SetDriverDramAddrHigh 0x0D ///< Set high 32 bits of DRAM address for Driver table transfer #define PPSMC_MSG_SetDriverDramAddrLow 0x0E ///< Set low 32 bits of DRAM address for Driver table transfer @@ -73,8 +73,7 @@ #define PPSMC_MSG_SetSoftMinFclk 0x14 ///< Set hard min for FCLK #define PPSMC_MSG_SetSoftMinVcn 0x15 ///< Set soft min for VCN clocks (VCLK and DCLK) - -#define PPSMC_MSG_EnableGfxImu 0x16 ///< Needs update +#define PPSMC_MSG_EnableGfxImu 0x16 ///< Enable GFX IMU #define PPSMC_MSG_GetGfxclkFrequency 0x17 ///< Get GFX clock frequency #define PPSMC_MSG_GetFclkFrequency 0x18 ///< Get FCLK frequency @@ -102,8 +101,8 @@ #define PPSMC_MSG_SetHardMinIspxclkByFreq 0x2C ///< Set HardMin by frequency for ISPXCLK #define PPSMC_MSG_PowerDownUmsch 0x2D ///< Power down VCN.UMSCH (aka VSCH) scheduler #define PPSMC_MSG_PowerUpUmsch 0x2E ///< Power up VCN.UMSCH (aka VSCH) scheduler -#define PPSMC_Message_IspStutterOn_MmhubPgDis 0x2F ///< ISP StutterOn mmHub PgDis -#define PPSMC_Message_IspStutterOff_MmhubPgEn 0x30 ///< ISP StufferOff mmHub PgEn +#define PPSMC_MSG_IspStutterOn_MmhubPgDis 0x2F ///< ISP StutterOn mmHub PgDis +#define PPSMC_MSG_IspStutterOff_MmhubPgEn 0x30 ///< ISP StufferOff mmHub PgEn #define PPSMC_Message_Count 0x31 ///< Total number of PPSMC messages /** @}*/ diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v11_0.h b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v11_0.h index a9215494dcdd..d466db6f0ad4 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v11_0.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v11_0.h @@ -147,14 +147,6 @@ struct smu_11_5_power_context { uint32_t max_fast_ppt_limit; }; -enum smu_v11_0_baco_seq { - BACO_SEQ_BACO = 0, - BACO_SEQ_MSR, - BACO_SEQ_BAMACO, - BACO_SEQ_ULPS, - BACO_SEQ_COUNT, -}; - #if defined(SWSMU_CODE_LAYER_L2) || defined(SWSMU_CODE_LAYER_L3) int smu_v11_0_init_microcode(struct smu_context *smu); @@ -257,7 +249,7 @@ int smu_v11_0_baco_enter(struct smu_context *smu); int smu_v11_0_baco_exit(struct smu_context *smu); int smu_v11_0_baco_set_armd3_sequence(struct smu_context *smu, - enum smu_v11_0_baco_seq baco_seq); + enum smu_baco_seq baco_seq); int smu_v11_0_mode1_reset(struct smu_context *smu); diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h index 8f72202aea8e..b7f4569aff2a 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_v13_0.h @@ -30,7 +30,7 @@ #define SMU13_DRIVER_IF_VERSION_ALDE 0x08 #define SMU13_DRIVER_IF_VERSION_SMU_V13_0_4 0x07 #define SMU13_DRIVER_IF_VERSION_SMU_V13_0_5 0x04 -#define SMU13_DRIVER_IF_VERSION_SMU_V13_0_0 0x30 +#define SMU13_DRIVER_IF_VERSION_SMU_V13_0_0_10 0x32 #define SMU13_DRIVER_IF_VERSION_SMU_V13_0_7 0x2C #define SMU13_DRIVER_IF_VERSION_SMU_V13_0_10 0x1D @@ -124,14 +124,6 @@ struct smu_13_0_power_context { enum smu_13_0_power_state power_state; }; -enum smu_v13_0_baco_seq { - BACO_SEQ_BACO = 0, - BACO_SEQ_MSR, - BACO_SEQ_BAMACO, - BACO_SEQ_ULPS, - BACO_SEQ_COUNT, -}; - #if defined(SWSMU_CODE_LAYER_L2) || defined(SWSMU_CODE_LAYER_L3) int smu_v13_0_init_microcode(struct smu_context *smu); @@ -218,6 +210,9 @@ int smu_v13_0_set_azalia_d3_pme(struct smu_context *smu); int smu_v13_0_get_max_sustainable_clocks_by_dc(struct smu_context *smu, struct pp_smu_nv_clock_table *max_clocks); +int smu_v13_0_baco_set_armd3_sequence(struct smu_context *smu, + enum smu_baco_seq baco_seq); + bool smu_v13_0_baco_is_support(struct smu_context *smu); enum smu_baco_state smu_v13_0_baco_get_state(struct smu_context *smu); diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c index 74996a8fb671..697e98a0a20a 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c @@ -377,7 +377,13 @@ static void sienna_cichlid_check_bxco_support(struct smu_context *smu) if (((adev->pdev->device == 0x73A1) && (adev->pdev->revision == 0x00)) || ((adev->pdev->device == 0x73BF) && - (adev->pdev->revision == 0xCF))) + (adev->pdev->revision == 0xCF)) || + ((adev->pdev->device == 0x7422) && + (adev->pdev->revision == 0x00)) || + ((adev->pdev->device == 0x73A3) && + (adev->pdev->revision == 0x00)) || + ((adev->pdev->device == 0x73E3) && + (adev->pdev->revision == 0x00))) smu_baco->platform_support = false; } diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c index dccbd9f70723..70b560737687 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c @@ -1576,7 +1576,7 @@ int smu_v11_0_set_azalia_d3_pme(struct smu_context *smu) } int smu_v11_0_baco_set_armd3_sequence(struct smu_context *smu, - enum smu_v11_0_baco_seq baco_seq) + enum smu_baco_seq baco_seq) { return smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_ArmD3, baco_seq, NULL); } diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c index c4552ade8d44..89f0f6eb19f3 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c @@ -289,7 +289,8 @@ int smu_v13_0_check_fw_version(struct smu_context *smu) smu->smc_driver_if_version = SMU13_DRIVER_IF_VERSION_ALDE; break; case IP_VERSION(13, 0, 0): - smu->smc_driver_if_version = SMU13_DRIVER_IF_VERSION_SMU_V13_0_0; + case IP_VERSION(13, 0, 10): + smu->smc_driver_if_version = SMU13_DRIVER_IF_VERSION_SMU_V13_0_0_10; break; case IP_VERSION(13, 0, 7): smu->smc_driver_if_version = SMU13_DRIVER_IF_VERSION_SMU_V13_0_7; @@ -305,9 +306,6 @@ int smu_v13_0_check_fw_version(struct smu_context *smu) case IP_VERSION(13, 0, 5): smu->smc_driver_if_version = SMU13_DRIVER_IF_VERSION_SMU_V13_0_5; break; - case IP_VERSION(13, 0, 10): - smu->smc_driver_if_version = SMU13_DRIVER_IF_VERSION_SMU_V13_0_10; - break; default: dev_err(adev->dev, "smu unsupported IP version: 0x%x.\n", adev->ip_versions[MP1_HWIP][0]); @@ -842,6 +840,7 @@ int smu_v13_0_gfx_off_control(struct smu_context *smu, bool enable) case IP_VERSION(13, 0, 5): case IP_VERSION(13, 0, 7): case IP_VERSION(13, 0, 8): + case IP_VERSION(13, 0, 10): if (!(adev->pm.pp_feature & PP_GFXOFF_MASK)) return 0; if (enable) @@ -2231,6 +2230,15 @@ int smu_v13_0_gfx_ulv_control(struct smu_context *smu, return ret; } +int smu_v13_0_baco_set_armd3_sequence(struct smu_context *smu, + enum smu_baco_seq baco_seq) +{ + return smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_ArmD3, + baco_seq, + NULL); +} + bool smu_v13_0_baco_is_support(struct smu_context *smu) { struct smu_baco_context *smu_baco = &smu->smu_baco; diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c index 29529328152d..5bcb61f77e41 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c @@ -70,6 +70,26 @@ #define MP0_MP1_DATA_REGION_SIZE_COMBOPPTABLE 0x4000 +#define mmMP1_SMN_C2PMSG_66 0x0282 +#define mmMP1_SMN_C2PMSG_66_BASE_IDX 0 + +#define mmMP1_SMN_C2PMSG_82 0x0292 +#define mmMP1_SMN_C2PMSG_82_BASE_IDX 0 + +#define mmMP1_SMN_C2PMSG_90 0x029a +#define mmMP1_SMN_C2PMSG_90_BASE_IDX 0 + +#define mmMP1_SMN_C2PMSG_75 0x028b +#define mmMP1_SMN_C2PMSG_75_BASE_IDX 0 + +#define mmMP1_SMN_C2PMSG_53 0x0275 +#define mmMP1_SMN_C2PMSG_53_BASE_IDX 0 + +#define mmMP1_SMN_C2PMSG_54 0x0276 +#define mmMP1_SMN_C2PMSG_54_BASE_IDX 0 + +#define DEBUGSMC_MSG_Mode1Reset 2 + static struct cmn2asic_msg_mapping smu_v13_0_0_message_map[SMU_MSG_MAX_COUNT] = { MSG_MAP(TestMessage, PPSMC_MSG_TestMessage, 1), MSG_MAP(GetSmuVersion, PPSMC_MSG_GetSmuVersion, 1), @@ -120,6 +140,7 @@ static struct cmn2asic_msg_mapping smu_v13_0_0_message_map[SMU_MSG_MAX_COUNT] = MSG_MAP(Mode1Reset, PPSMC_MSG_Mode1Reset, 0), MSG_MAP(PrepareMp1ForUnload, PPSMC_MSG_PrepareMp1ForUnload, 0), MSG_MAP(DFCstateControl, PPSMC_MSG_SetExternalClientDfCstateAllow, 0), + MSG_MAP(ArmD3, PPSMC_MSG_ArmD3, 0), }; static struct cmn2asic_mapping smu_v13_0_0_clk_map[SMU_CLK_COUNT] = { @@ -1566,6 +1587,31 @@ static int smu_v13_0_0_set_power_profile_mode(struct smu_context *smu, NULL); } +static int smu_v13_0_0_baco_enter(struct smu_context *smu) +{ + struct smu_baco_context *smu_baco = &smu->smu_baco; + struct amdgpu_device *adev = smu->adev; + + if (adev->in_runpm && smu_cmn_is_audio_func_enabled(adev)) + return smu_v13_0_baco_set_armd3_sequence(smu, + smu_baco->maco_support ? BACO_SEQ_BAMACO : BACO_SEQ_BACO); + else + return smu_v13_0_baco_enter(smu); +} + +static int smu_v13_0_0_baco_exit(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + + if (adev->in_runpm && smu_cmn_is_audio_func_enabled(adev)) { + /* Wait for PMFW handling for the Dstate change */ + usleep_range(10000, 11000); + return smu_v13_0_baco_set_armd3_sequence(smu, BACO_SEQ_ULPS); + } else { + return smu_v13_0_baco_exit(smu); + } +} + static bool smu_v13_0_0_is_mode1_reset_supported(struct smu_context *smu) { struct amdgpu_device *adev = smu->adev; @@ -1763,6 +1809,35 @@ static int smu_v13_0_0_set_df_cstate(struct smu_context *smu, NULL); } +static int smu_v13_0_0_mode1_reset(struct smu_context *smu) +{ + int ret; + struct amdgpu_device *adev = smu->adev; + + if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 10)) + ret = smu_cmn_send_debug_smc_msg(smu, DEBUGSMC_MSG_Mode1Reset); + else + ret = smu_cmn_send_smc_msg(smu, SMU_MSG_Mode1Reset, NULL); + + if (!ret) + msleep(SMU13_MODE1_RESET_WAIT_TIME_IN_MS); + + return ret; +} + +static void smu_v13_0_0_set_smu_mailbox_registers(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + + smu->param_reg = SOC15_REG_OFFSET(MP1, 0, mmMP1_SMN_C2PMSG_82); + smu->msg_reg = SOC15_REG_OFFSET(MP1, 0, mmMP1_SMN_C2PMSG_66); + smu->resp_reg = SOC15_REG_OFFSET(MP1, 0, mmMP1_SMN_C2PMSG_90); + + smu->debug_param_reg = SOC15_REG_OFFSET(MP1, 0, mmMP1_SMN_C2PMSG_53); + smu->debug_msg_reg = SOC15_REG_OFFSET(MP1, 0, mmMP1_SMN_C2PMSG_75); + smu->debug_resp_reg = SOC15_REG_OFFSET(MP1, 0, mmMP1_SMN_C2PMSG_54); +} + static const struct pptable_funcs smu_v13_0_0_ppt_funcs = { .get_allowed_feature_mask = smu_v13_0_0_get_allowed_feature_mask, .set_default_dpm_table = smu_v13_0_0_set_default_dpm_table, @@ -1827,10 +1902,10 @@ static const struct pptable_funcs smu_v13_0_0_ppt_funcs = { .baco_is_support = smu_v13_0_baco_is_support, .baco_get_state = smu_v13_0_baco_get_state, .baco_set_state = smu_v13_0_baco_set_state, - .baco_enter = smu_v13_0_baco_enter, - .baco_exit = smu_v13_0_baco_exit, + .baco_enter = smu_v13_0_0_baco_enter, + .baco_exit = smu_v13_0_0_baco_exit, .mode1_reset_is_support = smu_v13_0_0_is_mode1_reset_supported, - .mode1_reset = smu_v13_0_mode1_reset, + .mode1_reset = smu_v13_0_0_mode1_reset, .set_mp1_state = smu_v13_0_0_set_mp1_state, .set_df_cstate = smu_v13_0_0_set_df_cstate, }; @@ -1844,5 +1919,5 @@ void smu_v13_0_0_set_ppt_funcs(struct smu_context *smu) smu->table_map = smu_v13_0_0_table_map; smu->pwr_src_map = smu_v13_0_0_pwr_src_map; smu->workload_map = smu_v13_0_0_workload_map; - smu_v13_0_set_smu_mailbox_registers(smu); + smu_v13_0_0_set_smu_mailbox_registers(smu); } diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c index c4102cfb734c..d74debc584f8 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c @@ -122,6 +122,7 @@ static struct cmn2asic_msg_mapping smu_v13_0_7_message_map[SMU_MSG_MAX_COUNT] = MSG_MAP(PrepareMp1ForUnload, PPSMC_MSG_PrepareMp1ForUnload, 0), MSG_MAP(SetMGpuFanBoostLimitRpm, PPSMC_MSG_SetMGpuFanBoostLimitRpm, 0), MSG_MAP(DFCstateControl, PPSMC_MSG_SetExternalClientDfCstateAllow, 0), + MSG_MAP(ArmD3, PPSMC_MSG_ArmD3, 0), }; static struct cmn2asic_mapping smu_v13_0_7_clk_map[SMU_CLK_COUNT] = { @@ -1578,6 +1579,31 @@ static int smu_v13_0_7_set_mp1_state(struct smu_context *smu, return ret; } +static int smu_v13_0_7_baco_enter(struct smu_context *smu) +{ + struct smu_baco_context *smu_baco = &smu->smu_baco; + struct amdgpu_device *adev = smu->adev; + + if (adev->in_runpm && smu_cmn_is_audio_func_enabled(adev)) + return smu_v13_0_baco_set_armd3_sequence(smu, + smu_baco->maco_support ? BACO_SEQ_BAMACO : BACO_SEQ_BACO); + else + return smu_v13_0_baco_enter(smu); +} + +static int smu_v13_0_7_baco_exit(struct smu_context *smu) +{ + struct amdgpu_device *adev = smu->adev; + + if (adev->in_runpm && smu_cmn_is_audio_func_enabled(adev)) { + /* Wait for PMFW handling for the Dstate change */ + usleep_range(10000, 11000); + return smu_v13_0_baco_set_armd3_sequence(smu, BACO_SEQ_ULPS); + } else { + return smu_v13_0_baco_exit(smu); + } +} + static bool smu_v13_0_7_is_mode1_reset_supported(struct smu_context *smu) { struct amdgpu_device *adev = smu->adev; @@ -1655,8 +1681,8 @@ static const struct pptable_funcs smu_v13_0_7_ppt_funcs = { .baco_is_support = smu_v13_0_baco_is_support, .baco_get_state = smu_v13_0_baco_get_state, .baco_set_state = smu_v13_0_baco_set_state, - .baco_enter = smu_v13_0_baco_enter, - .baco_exit = smu_v13_0_baco_exit, + .baco_enter = smu_v13_0_7_baco_enter, + .baco_exit = smu_v13_0_7_baco_exit, .mode1_reset_is_support = smu_v13_0_7_is_mode1_reset_supported, .mode1_reset = smu_v13_0_mode1_reset, .set_mp1_state = smu_v13_0_7_set_mp1_state, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c index e4f8f90ac5aa..768b6e7dbd77 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c @@ -233,6 +233,18 @@ static void __smu_cmn_send_msg(struct smu_context *smu, WREG32(smu->msg_reg, msg); } +static int __smu_cmn_send_debug_msg(struct smu_context *smu, + u32 msg, + u32 param) +{ + struct amdgpu_device *adev = smu->adev; + + WREG32(smu->debug_param_reg, param); + WREG32(smu->debug_msg_reg, msg); + WREG32(smu->debug_resp_reg, 0); + + return 0; +} /** * smu_cmn_send_msg_without_waiting -- send the message; don't wait for status * @smu: pointer to an SMU context @@ -386,6 +398,12 @@ int smu_cmn_send_smc_msg(struct smu_context *smu, read_arg); } +int smu_cmn_send_debug_smc_msg(struct smu_context *smu, + uint32_t msg) +{ + return __smu_cmn_send_debug_msg(smu, msg, 0); +} + int smu_cmn_to_asic_specific_index(struct smu_context *smu, enum smu_cmn2asic_mapping_type type, uint32_t index) diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h index 1526ce09c399..f82cf76dd3a4 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h @@ -42,6 +42,9 @@ int smu_cmn_send_smc_msg(struct smu_context *smu, enum smu_message_type msg, uint32_t *read_arg); +int smu_cmn_send_debug_smc_msg(struct smu_context *smu, + uint32_t msg); + int smu_cmn_wait_for_response(struct smu_context *smu); int smu_cmn_to_asic_specific_index(struct smu_context *smu, diff --git a/drivers/gpu/drm/bridge/parade-ps8640.c b/drivers/gpu/drm/bridge/parade-ps8640.c index 5be6562c2a19..6a614e54b383 100644 --- a/drivers/gpu/drm/bridge/parade-ps8640.c +++ b/drivers/gpu/drm/bridge/parade-ps8640.c @@ -105,6 +105,7 @@ struct ps8640 { struct gpio_desc *gpio_powerdown; struct device_link *link; bool pre_enabled; + bool need_post_hpd_delay; }; static const struct regmap_config ps8640_regmap_config[] = { @@ -173,14 +174,31 @@ static int _ps8640_wait_hpd_asserted(struct ps8640 *ps_bridge, unsigned long wai { struct regmap *map = ps_bridge->regmap[PAGE2_TOP_CNTL]; int status; + int ret; /* * Apparently something about the firmware in the chip signals that * HPD goes high by reporting GPIO9 as high (even though HPD isn't * actually connected to GPIO9). */ - return regmap_read_poll_timeout(map, PAGE2_GPIO_H, status, - status & PS_GPIO9, wait_us / 10, wait_us); + ret = regmap_read_poll_timeout(map, PAGE2_GPIO_H, status, + status & PS_GPIO9, wait_us / 10, wait_us); + + /* + * The first time we see HPD go high after a reset we delay an extra + * 50 ms. The best guess is that the MCU is doing "stuff" during this + * time (maybe talking to the panel) and we don't want to interrupt it. + * + * No locking is done around "need_post_hpd_delay". If we're here we + * know we're holding a PM Runtime reference and the only other place + * that touches this is PM Runtime resume. + */ + if (!ret && ps_bridge->need_post_hpd_delay) { + ps_bridge->need_post_hpd_delay = false; + msleep(50); + } + + return ret; } static int ps8640_wait_hpd_asserted(struct drm_dp_aux *aux, unsigned long wait_us) @@ -388,6 +406,9 @@ static int __maybe_unused ps8640_resume(struct device *dev) msleep(50); gpiod_set_value(ps_bridge->gpio_reset, 0); + /* We just reset things, so we need a delay after the first HPD */ + ps_bridge->need_post_hpd_delay = true; + /* * Mystery 200 ms delay for the "MCU to be ready". It's unclear if * this is truly necessary since the MCU will already signal that diff --git a/drivers/gpu/drm/display/drm_dp_dual_mode_helper.c b/drivers/gpu/drm/display/drm_dp_dual_mode_helper.c index 3ea53bb67d3b..bd61e20770a5 100644 --- a/drivers/gpu/drm/display/drm_dp_dual_mode_helper.c +++ b/drivers/gpu/drm/display/drm_dp_dual_mode_helper.c @@ -63,23 +63,45 @@ ssize_t drm_dp_dual_mode_read(struct i2c_adapter *adapter, u8 offset, void *buffer, size_t size) { + u8 zero = 0; + char *tmpbuf = NULL; + /* + * As sub-addressing is not supported by all adaptors, + * always explicitly read from the start and discard + * any bytes that come before the requested offset. + * This way, no matter whether the adaptor supports it + * or not, we'll end up reading the proper data. + */ struct i2c_msg msgs[] = { { .addr = DP_DUAL_MODE_SLAVE_ADDRESS, .flags = 0, .len = 1, - .buf = &offset, + .buf = &zero, }, { .addr = DP_DUAL_MODE_SLAVE_ADDRESS, .flags = I2C_M_RD, - .len = size, + .len = size + offset, .buf = buffer, }, }; int ret; + if (offset) { + tmpbuf = kmalloc(size + offset, GFP_KERNEL); + if (!tmpbuf) + return -ENOMEM; + + msgs[1].buf = tmpbuf; + } + ret = i2c_transfer(adapter, msgs, ARRAY_SIZE(msgs)); + if (tmpbuf) + memcpy(buffer, tmpbuf + offset, size); + + kfree(tmpbuf); + if (ret < 0) return ret; if (ret != ARRAY_SIZE(msgs)) @@ -208,18 +230,6 @@ enum drm_dp_dual_mode_type drm_dp_dual_mode_detect(const struct drm_device *dev, if (ret) return DRM_DP_DUAL_MODE_UNKNOWN; - /* - * Sigh. Some (maybe all?) type 1 adaptors are broken and ack - * the offset but ignore it, and instead they just always return - * data from the start of the HDMI ID buffer. So for a broken - * type 1 HDMI adaptor a single byte read will always give us - * 0x44, and for a type 1 DVI adaptor it should give 0x00 - * (assuming it implements any registers). Fortunately neither - * of those values will match the type 2 signature of the - * DP_DUAL_MODE_ADAPTOR_ID register so we can proceed with - * the type 2 adaptor detection safely even in the presence - * of broken type 1 adaptors. - */ ret = drm_dp_dual_mode_read(adapter, DP_DUAL_MODE_ADAPTOR_ID, &adaptor_id, sizeof(adaptor_id)); drm_dbg_kms(dev, "DP dual mode adaptor ID: %02x (err %zd)\n", adaptor_id, ret); @@ -233,11 +243,10 @@ enum drm_dp_dual_mode_type drm_dp_dual_mode_detect(const struct drm_device *dev, return DRM_DP_DUAL_MODE_TYPE2_DVI; } /* - * If neither a proper type 1 ID nor a broken type 1 adaptor - * as described above, assume type 1, but let the user know - * that we may have misdetected the type. + * If not a proper type 1 ID, still assume type 1, but let + * the user know that we may have misdetected the type. */ - if (!is_type1_adaptor(adaptor_id) && adaptor_id != hdmi_id[0]) + if (!is_type1_adaptor(adaptor_id)) drm_err(dev, "Unexpected DP dual mode adaptor ID %02x\n", adaptor_id); } @@ -343,10 +352,8 @@ EXPORT_SYMBOL(drm_dp_dual_mode_get_tmds_output); * @enable: enable (as opposed to disable) the TMDS output buffers * * Set the state of the TMDS output buffers in the adaptor. For - * type2 this is set via the DP_DUAL_MODE_TMDS_OEN register. As - * some type 1 adaptors have problems with registers (see comments - * in drm_dp_dual_mode_detect()) we avoid touching the register, - * making this function a no-op on type 1 adaptors. + * type2 this is set via the DP_DUAL_MODE_TMDS_OEN register. + * Type1 adaptors do not support any register writes. * * Returns: * 0 on success, negative error code on failure diff --git a/drivers/gpu/drm/display/drm_dp_mst_topology.c b/drivers/gpu/drm/display/drm_dp_mst_topology.c index ecd22c038c8c..51a46689cda7 100644 --- a/drivers/gpu/drm/display/drm_dp_mst_topology.c +++ b/drivers/gpu/drm/display/drm_dp_mst_topology.c @@ -5186,7 +5186,7 @@ int drm_dp_mst_add_affected_dsc_crtcs(struct drm_atomic_state *state, struct drm mst_state = drm_atomic_get_mst_topology_state(state, mgr); if (IS_ERR(mst_state)) - return -EINVAL; + return PTR_ERR(mst_state); list_for_each_entry(pos, &mst_state->payloads, next) { diff --git a/drivers/gpu/drm/drm_drv.c b/drivers/gpu/drm/drm_drv.c index 8214a0b1ab7f..203bf8d6c34c 100644 --- a/drivers/gpu/drm/drm_drv.c +++ b/drivers/gpu/drm/drm_drv.c @@ -615,7 +615,7 @@ static int drm_dev_init(struct drm_device *dev, mutex_init(&dev->clientlist_mutex); mutex_init(&dev->master_mutex); - ret = drmm_add_action(dev, drm_dev_init_release, NULL); + ret = drmm_add_action_or_reset(dev, drm_dev_init_release, NULL); if (ret) return ret; diff --git a/drivers/gpu/drm/drm_edid.c b/drivers/gpu/drm/drm_edid.c index 1b0b0f85af29..3841aba17abd 100644 --- a/drivers/gpu/drm/drm_edid.c +++ b/drivers/gpu/drm/drm_edid.c @@ -87,6 +87,8 @@ static int oui(u8 first, u8 second, u8 third) #define EDID_QUIRK_FORCE_10BPC (1 << 11) /* Non desktop display (i.e. HMD) */ #define EDID_QUIRK_NON_DESKTOP (1 << 12) +/* Cap the DSC target bitrate to 15bpp */ +#define EDID_QUIRK_CAP_DSC_15BPP (1 << 13) #define MICROSOFT_IEEE_OUI 0xca125c @@ -147,6 +149,12 @@ static const struct edid_quirk { EDID_QUIRK('F', 'C', 'M', 13600, EDID_QUIRK_PREFER_LARGE_75 | EDID_QUIRK_DETAILED_IN_CM), + /* LG 27GP950 */ + EDID_QUIRK('G', 'S', 'M', 0x5bbf, EDID_QUIRK_CAP_DSC_15BPP), + + /* LG 27GN950 */ + EDID_QUIRK('G', 'S', 'M', 0x5b9a, EDID_QUIRK_CAP_DSC_15BPP), + /* LGD panel of HP zBook 17 G2, eDP 10 bpc, but reports unknown bpc */ EDID_QUIRK('L', 'G', 'D', 764, EDID_QUIRK_FORCE_10BPC), @@ -6365,6 +6373,7 @@ static void drm_reset_display_info(struct drm_connector *connector) info->mso_stream_count = 0; info->mso_pixel_overlap = 0; + info->max_dsc_bpp = 0; } static u32 update_display_info(struct drm_connector *connector, @@ -6454,6 +6463,9 @@ out: info->non_desktop = true; } + if (quirks & EDID_QUIRK_CAP_DSC_15BPP) + info->max_dsc_bpp = 15; + return quirks; } diff --git a/drivers/gpu/drm/drm_format_helper.c b/drivers/gpu/drm/drm_format_helper.c index 653a5821dd53..74ff33c2ddaa 100644 --- a/drivers/gpu/drm/drm_format_helper.c +++ b/drivers/gpu/drm/drm_format_helper.c @@ -817,6 +817,38 @@ static bool is_listed_fourcc(const uint32_t *fourccs, size_t nfourccs, uint32_t return false; } +static const uint32_t conv_from_xrgb8888[] = { + DRM_FORMAT_XRGB8888, + DRM_FORMAT_ARGB8888, + DRM_FORMAT_XRGB2101010, + DRM_FORMAT_ARGB2101010, + DRM_FORMAT_RGB565, + DRM_FORMAT_RGB888, +}; + +static const uint32_t conv_from_rgb565_888[] = { + DRM_FORMAT_XRGB8888, + DRM_FORMAT_ARGB8888, +}; + +static bool is_conversion_supported(uint32_t from, uint32_t to) +{ + switch (from) { + case DRM_FORMAT_XRGB8888: + case DRM_FORMAT_ARGB8888: + return is_listed_fourcc(conv_from_xrgb8888, ARRAY_SIZE(conv_from_xrgb8888), to); + case DRM_FORMAT_RGB565: + case DRM_FORMAT_RGB888: + return is_listed_fourcc(conv_from_rgb565_888, ARRAY_SIZE(conv_from_rgb565_888), to); + case DRM_FORMAT_XRGB2101010: + return to == DRM_FORMAT_ARGB2101010; + case DRM_FORMAT_ARGB2101010: + return to == DRM_FORMAT_XRGB2101010; + default: + return false; + } +} + /** * drm_fb_build_fourcc_list - Filters a list of supported color formats against * the device's native formats @@ -837,7 +869,9 @@ static bool is_listed_fourcc(const uint32_t *fourccs, size_t nfourccs, uint32_t * be handed over to drm_universal_plane_init() et al. Native formats * will go before emulated formats. Other heuristics might be applied * to optimize the order. Formats near the beginning of the list are - * usually preferred over formats near the end of the list. + * usually preferred over formats near the end of the list. Formats + * without conversion helpers will be skipped. New drivers should only + * pass in XRGB8888 and avoid exposing additional emulated formats. * * Returns: * The number of color-formats 4CC codes returned in @fourccs_out. @@ -849,7 +883,7 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, { u32 *fourccs = fourccs_out; const u32 *fourccs_end = fourccs_out + nfourccs_out; - bool found_native = false; + uint32_t native_format = 0; size_t i; /* @@ -868,27 +902,19 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, drm_dbg_kms(dev, "adding native format %p4cc\n", &fourcc); - if (!found_native) - found_native = is_listed_fourcc(driver_fourccs, driver_nfourccs, fourcc); + /* + * There should only be one native format with the current API. + * This API needs to be refactored to correctly support arbitrary + * sets of native formats, since it needs to report which native + * format to use for each emulated format. + */ + if (!native_format) + native_format = fourcc; *fourccs = fourcc; ++fourccs; } /* - * The plane's atomic_update helper converts the framebuffer's color format - * to a native format when copying to device memory. - * - * If there is not a single format supported by both, device and - * driver, the native formats are likely not supported by the conversion - * helpers. Therefore *only* support the native formats and add a - * conversion helper ASAP. - */ - if (!found_native) { - drm_warn(dev, "Format conversion helpers required to add extra formats.\n"); - goto out; - } - - /* * The extra formats, emulated by the driver, go second. */ @@ -900,6 +926,9 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, } else if (fourccs == fourccs_end) { drm_warn(dev, "Ignoring emulated format %p4cc\n", &fourcc); continue; /* end of available output buffer */ + } else if (!is_conversion_supported(fourcc, native_format)) { + drm_dbg_kms(dev, "Unsupported emulated format %p4cc\n", &fourcc); + continue; /* format is not supported for conversion */ } drm_dbg_kms(dev, "adding emulated format %p4cc\n", &fourcc); @@ -908,7 +937,6 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, ++fourccs; } -out: return fourccs - fourccs_out; } EXPORT_SYMBOL(drm_fb_build_fourcc_list); diff --git a/drivers/gpu/drm/drm_internal.h b/drivers/gpu/drm/drm_internal.h index 7bb98e6a446d..5ea5e260118c 100644 --- a/drivers/gpu/drm/drm_internal.h +++ b/drivers/gpu/drm/drm_internal.h @@ -104,7 +104,8 @@ static inline void drm_vblank_flush_worker(struct drm_vblank_crtc *vblank) static inline void drm_vblank_destroy_worker(struct drm_vblank_crtc *vblank) { - kthread_destroy_worker(vblank->worker); + if (vblank->worker) + kthread_destroy_worker(vblank->worker); } int drm_vblank_worker_init(struct drm_vblank_crtc *vblank); diff --git a/drivers/gpu/drm/drm_panel_orientation_quirks.c b/drivers/gpu/drm/drm_panel_orientation_quirks.c index 8a0c0e0bb5bd..52d8800a8ab8 100644 --- a/drivers/gpu/drm/drm_panel_orientation_quirks.c +++ b/drivers/gpu/drm/drm_panel_orientation_quirks.c @@ -134,6 +134,12 @@ static const struct dmi_system_id orientation_data[] = { DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "One S1003"), }, .driver_data = (void *)&lcd800x1280_rightside_up, + }, { /* Acer Switch V 10 (SW5-017) */ + .matches = { + DMI_EXACT_MATCH(DMI_SYS_VENDOR, "Acer"), + DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "SW5-017"), + }, + .driver_data = (void *)&lcd800x1280_rightside_up, }, { /* Anbernic Win600 */ .matches = { DMI_EXACT_MATCH(DMI_BOARD_VENDOR, "Anbernic"), @@ -319,6 +325,12 @@ static const struct dmi_system_id orientation_data[] = { DMI_MATCH(DMI_BIOS_VERSION, "BLADE_21"), }, .driver_data = (void *)&lcd1200x1920_rightside_up, + }, { /* Nanote UMPC-01 */ + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "RWC CO.,LTD"), + DMI_MATCH(DMI_PRODUCT_NAME, "UMPC-01"), + }, + .driver_data = (void *)&lcd1200x1920_rightside_up, }, { /* OneGX1 Pro */ .matches = { DMI_EXACT_MATCH(DMI_SYS_VENDOR, "SYSTEM_MANUFACTURER"), diff --git a/drivers/gpu/drm/etnaviv/etnaviv_dump.c b/drivers/gpu/drm/etnaviv/etnaviv_dump.c index f418e0b75772..44b5f3c35aab 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_dump.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_dump.c @@ -83,10 +83,15 @@ static void etnaviv_core_dump_registers(struct core_dump_iterator *iter, { struct etnaviv_dump_registers *reg = iter->data; unsigned int i; + u32 read_addr; for (i = 0; i < ARRAY_SIZE(etnaviv_dump_registers); i++, reg++) { + read_addr = etnaviv_dump_registers[i]; + if (read_addr >= VIVS_PM_POWER_CONTROLS && + read_addr <= VIVS_PM_PULSE_EATER) + read_addr = gpu_fix_power_address(gpu, read_addr); reg->reg = cpu_to_le32(etnaviv_dump_registers[i]); - reg->value = cpu_to_le32(gpu_read(gpu, etnaviv_dump_registers[i])); + reg->value = cpu_to_le32(gpu_read(gpu, read_addr)); } etnaviv_core_dump_header(iter, ETDUMP_BUF_REG, reg); diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gem.c b/drivers/gpu/drm/etnaviv/etnaviv_gem.c index cc386f8a7116..68e4446a94ad 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_gem.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_gem.c @@ -130,7 +130,7 @@ static int etnaviv_gem_mmap_obj(struct etnaviv_gem_object *etnaviv_obj, { pgprot_t vm_page_prot; - vma->vm_flags |= VM_IO | VM_MIXEDMAP | VM_DONTEXPAND | VM_DONTDUMP; + vma->vm_flags |= VM_PFNMAP | VM_DONTEXPAND | VM_DONTDUMP; vm_page_prot = vm_get_page_prot(vma->vm_flags); @@ -165,7 +165,8 @@ static vm_fault_t etnaviv_gem_fault(struct vm_fault *vmf) struct vm_area_struct *vma = vmf->vma; struct drm_gem_object *obj = vma->vm_private_data; struct etnaviv_gem_object *etnaviv_obj = to_etnaviv_bo(obj); - struct page **pages, *page; + struct page **pages; + unsigned long pfn; pgoff_t pgoff; int err; @@ -189,12 +190,12 @@ static vm_fault_t etnaviv_gem_fault(struct vm_fault *vmf) /* We don't use vmf->pgoff since that has the fake offset: */ pgoff = (vmf->address - vma->vm_start) >> PAGE_SHIFT; - page = pages[pgoff]; + pfn = page_to_pfn(pages[pgoff]); VERB("Inserting %p pfn %lx, pa %lx", (void *)vmf->address, - page_to_pfn(page), page_to_pfn(page) << PAGE_SHIFT); + pfn, pfn << PAGE_SHIFT); - return vmf_insert_page(vma, vmf->address, page); + return vmf_insert_pfn(vma, vmf->address, pfn); } int etnaviv_gem_mmap_offset(struct drm_gem_object *obj, u64 *offset) @@ -258,7 +259,12 @@ struct etnaviv_vram_mapping *etnaviv_gem_mapping_get( if (mapping->use == 0) { mutex_lock(&mmu_context->lock); if (mapping->context == mmu_context) - mapping->use += 1; + if (va && mapping->iova != va) { + etnaviv_iommu_reap_mapping(mapping); + mapping = NULL; + } else { + mapping->use += 1; + } else mapping = NULL; mutex_unlock(&mmu_context->lock); @@ -504,7 +510,6 @@ void etnaviv_gem_free_object(struct drm_gem_object *obj) kfree(mapping); } - drm_gem_free_mmap_offset(obj); etnaviv_obj->ops->release(etnaviv_obj); drm_gem_object_release(obj); diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gem.h b/drivers/gpu/drm/etnaviv/etnaviv_gem.h index 63688e6e4580..baa81cbf701a 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_gem.h +++ b/drivers/gpu/drm/etnaviv/etnaviv_gem.h @@ -96,6 +96,7 @@ struct etnaviv_gem_submit { int out_fence_id; struct list_head node; /* GPU active submit list */ struct etnaviv_cmdbuf cmdbuf; + struct pid *pid; /* submitting process */ bool runtime_resumed; u32 exec_state; u32 flags; diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c b/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c index 1ac916b24891..1491159d0d20 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_gem_submit.c @@ -399,6 +399,9 @@ static void submit_cleanup(struct kref *kref) mutex_unlock(&submit->gpu->fence_lock); dma_fence_put(submit->out_fence); } + + put_pid(submit->pid); + kfree(submit->pmrs); kfree(submit); } @@ -422,6 +425,7 @@ int etnaviv_ioctl_gem_submit(struct drm_device *dev, void *data, struct sync_file *sync_file = NULL; struct ww_acquire_ctx ticket; int out_fence_fd = -1; + struct pid *pid = get_pid(task_pid(current)); void *stream; int ret; @@ -519,6 +523,8 @@ int etnaviv_ioctl_gem_submit(struct drm_device *dev, void *data, goto err_submit_ww_acquire; } + submit->pid = pid; + ret = etnaviv_cmdbuf_init(priv->cmdbuf_suballoc, &submit->cmdbuf, ALIGN(args->stream_size, 8) + 8); if (ret) diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c index 37018bc55810..51320eeebfcf 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_gpu.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_gpu.c @@ -416,6 +416,12 @@ static void etnaviv_hw_identify(struct etnaviv_gpu *gpu) if (gpu->identity.model == chipModel_GC700) gpu->identity.features &= ~chipFeatures_FAST_CLEAR; + /* These models/revisions don't have the 2D pipe bit */ + if ((gpu->identity.model == chipModel_GC500 && + gpu->identity.revision <= 2) || + gpu->identity.model == chipModel_GC300) + gpu->identity.features |= chipFeatures_PIPE_2D; + if ((gpu->identity.model == chipModel_GC500 && gpu->identity.revision < 2) || (gpu->identity.model == chipModel_GC300 && @@ -449,8 +455,9 @@ static void etnaviv_hw_identify(struct etnaviv_gpu *gpu) gpu_read(gpu, VIVS_HI_CHIP_MINOR_FEATURE_5); } - /* GC600 idle register reports zero bits where modules aren't present */ - if (gpu->identity.model == chipModel_GC600) + /* GC600/300 idle register reports zero bits where modules aren't present */ + if (gpu->identity.model == chipModel_GC600 || + gpu->identity.model == chipModel_GC300) gpu->idle_mask = VIVS_HI_IDLE_STATE_TX | VIVS_HI_IDLE_STATE_RA | VIVS_HI_IDLE_STATE_SE | @@ -583,7 +590,7 @@ static void etnaviv_gpu_enable_mlcg(struct etnaviv_gpu *gpu) u32 pmc, ppc; /* enable clock gating */ - ppc = gpu_read(gpu, VIVS_PM_POWER_CONTROLS); + ppc = gpu_read_power(gpu, VIVS_PM_POWER_CONTROLS); ppc |= VIVS_PM_POWER_CONTROLS_ENABLE_MODULE_CLOCK_GATING; /* Disable stall module clock gating for 4.3.0.1 and 4.3.0.2 revs */ @@ -591,9 +598,9 @@ static void etnaviv_gpu_enable_mlcg(struct etnaviv_gpu *gpu) gpu->identity.revision == 0x4302) ppc |= VIVS_PM_POWER_CONTROLS_DISABLE_STALL_MODULE_CLOCK_GATING; - gpu_write(gpu, VIVS_PM_POWER_CONTROLS, ppc); + gpu_write_power(gpu, VIVS_PM_POWER_CONTROLS, ppc); - pmc = gpu_read(gpu, VIVS_PM_MODULE_CONTROLS); + pmc = gpu_read_power(gpu, VIVS_PM_MODULE_CONTROLS); /* Disable PA clock gating for GC400+ without bugfix except for GC420 */ if (gpu->identity.model >= chipModel_GC400 && @@ -616,19 +623,20 @@ static void etnaviv_gpu_enable_mlcg(struct etnaviv_gpu *gpu) /* Disable TX clock gating on affected core revisions. */ if (etnaviv_is_model_rev(gpu, GC4000, 0x5222) || - etnaviv_is_model_rev(gpu, GC2000, 0x5108)) + etnaviv_is_model_rev(gpu, GC2000, 0x5108) || + etnaviv_is_model_rev(gpu, GC2000, 0x6202) || + etnaviv_is_model_rev(gpu, GC2000, 0x6203)) pmc |= VIVS_PM_MODULE_CONTROLS_DISABLE_MODULE_CLOCK_GATING_TX; - /* Disable SE, RA and TX clock gating on affected core revisions. */ + /* Disable SE and RA clock gating on affected core revisions. */ if (etnaviv_is_model_rev(gpu, GC7000, 0x6202)) pmc |= VIVS_PM_MODULE_CONTROLS_DISABLE_MODULE_CLOCK_GATING_SE | - VIVS_PM_MODULE_CONTROLS_DISABLE_MODULE_CLOCK_GATING_RA | - VIVS_PM_MODULE_CONTROLS_DISABLE_MODULE_CLOCK_GATING_TX; + VIVS_PM_MODULE_CONTROLS_DISABLE_MODULE_CLOCK_GATING_RA; pmc |= VIVS_PM_MODULE_CONTROLS_DISABLE_MODULE_CLOCK_GATING_RA_HZ; pmc |= VIVS_PM_MODULE_CONTROLS_DISABLE_MODULE_CLOCK_GATING_RA_EZ; - gpu_write(gpu, VIVS_PM_MODULE_CONTROLS, pmc); + gpu_write_power(gpu, VIVS_PM_MODULE_CONTROLS, pmc); } void etnaviv_gpu_start_fe(struct etnaviv_gpu *gpu, u32 address, u16 prefetch) @@ -688,11 +696,11 @@ static void etnaviv_gpu_setup_pulse_eater(struct etnaviv_gpu *gpu) (gpu->identity.features & chipFeatures_PIPE_3D)) { /* Performance fix: disable internal DFS */ - pulse_eater = gpu_read(gpu, VIVS_PM_PULSE_EATER); + pulse_eater = gpu_read_power(gpu, VIVS_PM_PULSE_EATER); pulse_eater |= BIT(18); } - gpu_write(gpu, VIVS_PM_PULSE_EATER, pulse_eater); + gpu_write_power(gpu, VIVS_PM_PULSE_EATER, pulse_eater); } static void etnaviv_gpu_hw_init(struct etnaviv_gpu *gpu) @@ -1045,12 +1053,28 @@ pm_put: } #endif -void etnaviv_gpu_recover_hang(struct etnaviv_gpu *gpu) +void etnaviv_gpu_recover_hang(struct etnaviv_gem_submit *submit) { + struct etnaviv_gpu *gpu = submit->gpu; + char *comm = NULL, *cmd = NULL; + struct task_struct *task; unsigned int i; dev_err(gpu->dev, "recover hung GPU!\n"); + task = get_pid_task(submit->pid, PIDTYPE_PID); + if (task) { + comm = kstrdup(task->comm, GFP_KERNEL); + cmd = kstrdup_quotable_cmdline(task, GFP_KERNEL); + put_task_struct(task); + } + + if (comm && cmd) + dev_err(gpu->dev, "offending task: %s (%s)\n", comm, cmd); + + kfree(cmd); + kfree(comm); + if (pm_runtime_get_sync(gpu->dev) < 0) goto pm_put; @@ -1294,9 +1318,9 @@ static void sync_point_perfmon_sample_pre(struct etnaviv_gpu *gpu, u32 val; /* disable clock gating */ - val = gpu_read(gpu, VIVS_PM_POWER_CONTROLS); + val = gpu_read_power(gpu, VIVS_PM_POWER_CONTROLS); val &= ~VIVS_PM_POWER_CONTROLS_ENABLE_MODULE_CLOCK_GATING; - gpu_write(gpu, VIVS_PM_POWER_CONTROLS, val); + gpu_write_power(gpu, VIVS_PM_POWER_CONTROLS, val); /* enable debug register */ val = gpu_read(gpu, VIVS_HI_CLOCK_CONTROL); @@ -1327,9 +1351,9 @@ static void sync_point_perfmon_sample_post(struct etnaviv_gpu *gpu, gpu_write(gpu, VIVS_HI_CLOCK_CONTROL, val); /* enable clock gating */ - val = gpu_read(gpu, VIVS_PM_POWER_CONTROLS); + val = gpu_read_power(gpu, VIVS_PM_POWER_CONTROLS); val |= VIVS_PM_POWER_CONTROLS_ENABLE_MODULE_CLOCK_GATING; - gpu_write(gpu, VIVS_PM_POWER_CONTROLS, val); + gpu_write_power(gpu, VIVS_PM_POWER_CONTROLS, val); } diff --git a/drivers/gpu/drm/etnaviv/etnaviv_gpu.h b/drivers/gpu/drm/etnaviv/etnaviv_gpu.h index 85eddd492774..f1204b070fb8 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_gpu.h +++ b/drivers/gpu/drm/etnaviv/etnaviv_gpu.h @@ -10,6 +10,7 @@ #include "etnaviv_gem.h" #include "etnaviv_mmu.h" #include "etnaviv_drv.h" +#include "common.xml.h" struct etnaviv_gem_submit; struct etnaviv_vram_mapping; @@ -159,6 +160,26 @@ static inline u32 gpu_read(struct etnaviv_gpu *gpu, u32 reg) return readl(gpu->mmio + reg); } +static inline u32 gpu_fix_power_address(struct etnaviv_gpu *gpu, u32 reg) +{ + /* Power registers in GC300 < 2.0 are offset by 0x100 */ + if (gpu->identity.model == chipModel_GC300 && + gpu->identity.revision < 0x2000) + reg += 0x100; + + return reg; +} + +static inline void gpu_write_power(struct etnaviv_gpu *gpu, u32 reg, u32 data) +{ + writel(data, gpu->mmio + gpu_fix_power_address(gpu, reg)); +} + +static inline u32 gpu_read_power(struct etnaviv_gpu *gpu, u32 reg) +{ + return readl(gpu->mmio + gpu_fix_power_address(gpu, reg)); +} + int etnaviv_gpu_get_param(struct etnaviv_gpu *gpu, u32 param, u64 *value); int etnaviv_gpu_init(struct etnaviv_gpu *gpu); @@ -168,7 +189,7 @@ bool etnaviv_fill_identity_from_hwdb(struct etnaviv_gpu *gpu); int etnaviv_gpu_debugfs(struct etnaviv_gpu *gpu, struct seq_file *m); #endif -void etnaviv_gpu_recover_hang(struct etnaviv_gpu *gpu); +void etnaviv_gpu_recover_hang(struct etnaviv_gem_submit *submit); void etnaviv_gpu_retire(struct etnaviv_gpu *gpu); int etnaviv_gpu_wait_fence_interruptible(struct etnaviv_gpu *gpu, u32 fence, struct drm_etnaviv_timespec *timeout); diff --git a/drivers/gpu/drm/etnaviv/etnaviv_hwdb.c b/drivers/gpu/drm/etnaviv/etnaviv_hwdb.c index f2fc645c7956..57f334e24189 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_hwdb.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_hwdb.c @@ -70,6 +70,37 @@ static const struct etnaviv_chip_identity etnaviv_chip_identities[] = { }, { .model = 0x7000, + .revision = 0x6203, + .product_id = 0x70003, + .customer_id = 0x4, + .eco_id = 0, + .stream_count = 16, + .register_max = 64, + .thread_count = 512, + .shader_core_count = 2, + .vertex_cache_size = 16, + .vertex_output_buffer_size = 1024, + .pixel_pipes = 1, + .instruction_count = 512, + .num_constants = 320, + .buffer_size = 0, + .varyings_count = 16, + .features = 0xe0287c8d, + .minor_features0 = 0xc1589eff, + .minor_features1 = 0xfefbfad9, + .minor_features2 = 0xeb9d4fbf, + .minor_features3 = 0xedfffced, + .minor_features4 = 0xdb0dafc7, + .minor_features5 = 0x3b5ac333, + .minor_features6 = 0xfcce6000, + .minor_features7 = 0xfffbfa6f, + .minor_features8 = 0x00e10ef3, + .minor_features9 = 0x00c8003c, + .minor_features10 = 0x00004040, + .minor_features11 = 0x00000024, + }, + { + .model = 0x7000, .revision = 0x6204, .product_id = ~0U, .customer_id = ~0U, diff --git a/drivers/gpu/drm/etnaviv/etnaviv_mmu.c b/drivers/gpu/drm/etnaviv/etnaviv_mmu.c index dc1aa738c4f1..67bdce5326c6 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_mmu.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_mmu.c @@ -80,10 +80,10 @@ static int etnaviv_iommu_map(struct etnaviv_iommu_context *context, u32 iova, return -EINVAL; for_each_sgtable_dma_sg(sgt, sg, i) { - u32 pa = sg_dma_address(sg) - sg->offset; + phys_addr_t pa = sg_dma_address(sg) - sg->offset; size_t bytes = sg_dma_len(sg) + sg->offset; - VERB("map[%d]: %08x %08x(%zx)", i, iova, pa, bytes); + VERB("map[%d]: %08x %pap(%zx)", i, iova, &pa, bytes); ret = etnaviv_context_map(context, da, pa, bytes, prot); if (ret) @@ -135,6 +135,19 @@ static void etnaviv_iommu_remove_mapping(struct etnaviv_iommu_context *context, drm_mm_remove_node(&mapping->vram_node); } +void etnaviv_iommu_reap_mapping(struct etnaviv_vram_mapping *mapping) +{ + struct etnaviv_iommu_context *context = mapping->context; + + lockdep_assert_held(&context->lock); + WARN_ON(mapping->use); + + etnaviv_iommu_remove_mapping(context, mapping); + etnaviv_iommu_context_put(mapping->context); + mapping->context = NULL; + list_del_init(&mapping->mmu_node); +} + static int etnaviv_iommu_find_iova(struct etnaviv_iommu_context *context, struct drm_mm_node *node, size_t size) { @@ -202,10 +215,7 @@ static int etnaviv_iommu_find_iova(struct etnaviv_iommu_context *context, * this mapping. */ list_for_each_entry_safe(m, n, &list, scan_node) { - etnaviv_iommu_remove_mapping(context, m); - etnaviv_iommu_context_put(m->context); - m->context = NULL; - list_del_init(&m->mmu_node); + etnaviv_iommu_reap_mapping(m); list_del_init(&m->scan_node); } @@ -257,10 +267,7 @@ static int etnaviv_iommu_insert_exact(struct etnaviv_iommu_context *context, } list_for_each_entry_safe(m, n, &scan_list, scan_node) { - etnaviv_iommu_remove_mapping(context, m); - etnaviv_iommu_context_put(m->context); - m->context = NULL; - list_del_init(&m->mmu_node); + etnaviv_iommu_reap_mapping(m); list_del_init(&m->scan_node); } diff --git a/drivers/gpu/drm/etnaviv/etnaviv_mmu.h b/drivers/gpu/drm/etnaviv/etnaviv_mmu.h index e4a0b7d09c2e..c01a147f0dfd 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_mmu.h +++ b/drivers/gpu/drm/etnaviv/etnaviv_mmu.h @@ -91,6 +91,7 @@ int etnaviv_iommu_map_gem(struct etnaviv_iommu_context *context, struct etnaviv_vram_mapping *mapping, u64 va); void etnaviv_iommu_unmap_gem(struct etnaviv_iommu_context *context, struct etnaviv_vram_mapping *mapping); +void etnaviv_iommu_reap_mapping(struct etnaviv_vram_mapping *mapping); int etnaviv_iommu_get_suballoc_va(struct etnaviv_iommu_context *ctx, struct etnaviv_vram_mapping *mapping, diff --git a/drivers/gpu/drm/etnaviv/etnaviv_sched.c b/drivers/gpu/drm/etnaviv/etnaviv_sched.c index 72e2553fbc98..d29f467eee13 100644 --- a/drivers/gpu/drm/etnaviv/etnaviv_sched.c +++ b/drivers/gpu/drm/etnaviv/etnaviv_sched.c @@ -67,7 +67,7 @@ static enum drm_gpu_sched_stat etnaviv_sched_timedout_job(struct drm_sched_job /* get the GPU back into the init state */ etnaviv_core_dump(submit); - etnaviv_gpu_recover_hang(gpu); + etnaviv_gpu_recover_hang(submit); drm_sched_resubmit_jobs(&gpu->sched); diff --git a/drivers/gpu/drm/i915/Makefile b/drivers/gpu/drm/i915/Makefile index 51704b54317c..01974b82d205 100644 --- a/drivers/gpu/drm/i915/Makefile +++ b/drivers/gpu/drm/i915/Makefile @@ -127,9 +127,11 @@ gt-y += \ gt/intel_sseu.o \ gt/intel_sseu_debugfs.o \ gt/intel_timeline.o \ + gt/intel_wopcm.o \ gt/intel_workarounds.o \ gt/shmem_utils.o \ gt/sysfs_engines.o + # x86 intel-gtt module support gt-$(CONFIG_X86) += gt/intel_ggtt_gmch.o # autogenerated null render state @@ -183,8 +185,7 @@ i915-y += \ i915_trace_points.o \ i915_ttm_buddy_manager.o \ i915_vma.o \ - i915_vma_resource.o \ - intel_wopcm.o + i915_vma_resource.o # general-purpose microcontroller (GuC) support i915-y += gt/uc/intel_uc.o \ @@ -247,6 +248,7 @@ i915-y += \ display/intel_global_state.o \ display/intel_hdcp.o \ display/intel_hotplug.o \ + display/intel_hti.o \ display/intel_lpe_audio.o \ display/intel_modeset_verify.o \ display/intel_modeset_setup.o \ diff --git a/drivers/gpu/drm/i915/display/g4x_dp.c b/drivers/gpu/drm/i915/display/g4x_dp.c index e3e3d27ffb53..3593938dcd87 100644 --- a/drivers/gpu/drm/i915/display/g4x_dp.c +++ b/drivers/gpu/drm/i915/display/g4x_dp.c @@ -8,6 +8,7 @@ #include <linux/string_helpers.h> #include "g4x_dp.h" +#include "i915_reg.h" #include "intel_audio.h" #include "intel_backlight.h" #include "intel_connector.h" diff --git a/drivers/gpu/drm/i915/display/g4x_dp.h b/drivers/gpu/drm/i915/display/g4x_dp.h index e1f50263a725..a38b3e1e01d3 100644 --- a/drivers/gpu/drm/i915/display/g4x_dp.h +++ b/drivers/gpu/drm/i915/display/g4x_dp.h @@ -8,7 +8,7 @@ #include <linux/types.h> -#include "i915_reg.h" +#include "i915_reg_defs.h" enum pipe; enum port; diff --git a/drivers/gpu/drm/i915/display/g4x_hdmi.c b/drivers/gpu/drm/i915/display/g4x_hdmi.c index 8aadf96fa5e9..121caeaa409b 100644 --- a/drivers/gpu/drm/i915/display/g4x_hdmi.c +++ b/drivers/gpu/drm/i915/display/g4x_hdmi.c @@ -6,6 +6,7 @@ */ #include "g4x_hdmi.h" +#include "i915_reg.h" #include "intel_audio.h" #include "intel_connector.h" #include "intel_crtc.h" @@ -78,6 +79,18 @@ static bool intel_hdmi_get_hw_state(struct intel_encoder *encoder, return ret; } +static int g4x_hdmi_compute_config(struct intel_encoder *encoder, + struct intel_crtc_state *crtc_state, + struct drm_connector_state *conn_state) +{ + struct drm_i915_private *i915 = to_i915(encoder->base.dev); + + if (HAS_PCH_SPLIT(i915)) + crtc_state->has_pch_encoder = true; + + return intel_hdmi_compute_config(encoder, crtc_state, conn_state); +} + static void intel_hdmi_get_config(struct intel_encoder *encoder, struct intel_crtc_state *pipe_config) { @@ -543,7 +556,7 @@ void g4x_hdmi_init(struct drm_i915_private *dev_priv, "HDMI %c", port_name(port)); intel_encoder->hotplug = intel_hdmi_hotplug; - intel_encoder->compute_config = intel_hdmi_compute_config; + intel_encoder->compute_config = g4x_hdmi_compute_config; if (HAS_PCH_SPLIT(dev_priv)) { intel_encoder->disable = pch_disable_hdmi; intel_encoder->post_disable = pch_post_disable_hdmi; diff --git a/drivers/gpu/drm/i915/display/i9xx_plane.c b/drivers/gpu/drm/i915/display/i9xx_plane.c index 5afbe3e98ee8..ecaeb7dc196b 100644 --- a/drivers/gpu/drm/i915/display/i9xx_plane.c +++ b/drivers/gpu/drm/i915/display/i9xx_plane.c @@ -8,6 +8,9 @@ #include <drm/drm_blend.h> #include <drm/drm_fourcc.h> +#include "i915_irq.h" +#include "i915_reg.h" +#include "i9xx_plane.h" #include "intel_atomic.h" #include "intel_atomic_plane.h" #include "intel_de.h" @@ -15,7 +18,6 @@ #include "intel_fb.h" #include "intel_fbc.h" #include "intel_sprite.h" -#include "i9xx_plane.h" /* Primary plane formats for gen <= 3 */ static const u32 i8xx_primary_formats[] = { diff --git a/drivers/gpu/drm/i915/display/icl_dsi.c b/drivers/gpu/drm/i915/display/icl_dsi.c index e05e7cd6c412..d16b30a2dded 100644 --- a/drivers/gpu/drm/i915/display/icl_dsi.c +++ b/drivers/gpu/drm/i915/display/icl_dsi.c @@ -29,6 +29,7 @@ #include <drm/drm_atomic_helper.h> #include <drm/drm_mipi_dsi.h> +#include "i915_reg.h" #include "icl_dsi.h" #include "icl_dsi_regs.h" #include "intel_atomic.h" diff --git a/drivers/gpu/drm/i915/display/icl_dsi_regs.h b/drivers/gpu/drm/i915/display/icl_dsi_regs.h index f78f28b8dd94..d4845ac65acc 100644 --- a/drivers/gpu/drm/i915/display/icl_dsi_regs.h +++ b/drivers/gpu/drm/i915/display/icl_dsi_regs.h @@ -6,7 +6,7 @@ #ifndef __ICL_DSI_REGS_H__ #define __ICL_DSI_REGS_H__ -#include "i915_reg_defs.h" +#include "intel_display_reg_defs.h" /* Gen11 DSI */ #define _MMIO_DSI(tc, dsi0, dsi1) _MMIO_TRANS((tc) - TRANSCODER_DSI_0, \ diff --git a/drivers/gpu/drm/i915/display/intel_atomic_plane.c b/drivers/gpu/drm/i915/display/intel_atomic_plane.c index bcf0239b9533..10e1fc9d0698 100644 --- a/drivers/gpu/drm/i915/display/intel_atomic_plane.c +++ b/drivers/gpu/drm/i915/display/intel_atomic_plane.c @@ -757,7 +757,7 @@ void intel_plane_update_noarm(struct intel_plane *plane, { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - trace_intel_plane_update_noarm(&plane->base, crtc); + trace_intel_plane_update_noarm(plane, crtc); if (plane->update_noarm) plane->update_noarm(plane, crtc_state, plane_state); @@ -769,7 +769,7 @@ void intel_plane_update_arm(struct intel_plane *plane, { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - trace_intel_plane_update_arm(&plane->base, crtc); + trace_intel_plane_update_arm(plane, crtc); if (crtc_state->do_async_flip && plane->async_flip) plane->async_flip(plane, crtc_state, plane_state, true); @@ -782,7 +782,7 @@ void intel_plane_disable_arm(struct intel_plane *plane, { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - trace_intel_plane_disable_arm(&plane->base, crtc); + trace_intel_plane_disable_arm(plane, crtc); plane->disable_arm(plane, crtc_state); } diff --git a/drivers/gpu/drm/i915/display/intel_audio.c b/drivers/gpu/drm/i915/display/intel_audio.c index c3176c9c89a6..98c3322b4549 100644 --- a/drivers/gpu/drm/i915/display/intel_audio.c +++ b/drivers/gpu/drm/i915/display/intel_audio.c @@ -838,8 +838,8 @@ void intel_audio_codec_enable(struct intel_encoder *encoder, if (i915->display.funcs.audio) i915->display.funcs.audio->audio_codec_enable(encoder, - crtc_state, - conn_state); + crtc_state, + conn_state); mutex_lock(&i915->display.audio.mutex); encoder->audio_connector = connector; @@ -854,7 +854,7 @@ void intel_audio_codec_enable(struct intel_encoder *encoder, if (!intel_crtc_has_type(crtc_state, INTEL_OUTPUT_DP_MST)) pipe = -1; acomp->base.audio_ops->pin_eld_notify(acomp->base.audio_ops->audio_ptr, - (int) port, (int) pipe); + (int)port, (int)pipe); } intel_lpe_audio_notify(i915, pipe, port, connector->eld, @@ -891,8 +891,8 @@ void intel_audio_codec_disable(struct intel_encoder *encoder, if (i915->display.funcs.audio) i915->display.funcs.audio->audio_codec_disable(encoder, - old_crtc_state, - old_conn_state); + old_crtc_state, + old_conn_state); mutex_lock(&i915->display.audio.mutex); encoder->audio_connector = NULL; @@ -905,7 +905,7 @@ void intel_audio_codec_disable(struct intel_encoder *encoder, if (!intel_crtc_has_type(old_crtc_state, INTEL_OUTPUT_DP_MST)) pipe = -1; acomp->base.audio_ops->pin_eld_notify(acomp->base.audio_ops->audio_ptr, - (int) port, (int) pipe); + (int)port, (int)pipe); } intel_lpe_audio_notify(i915, pipe, port, NULL, 0, false); @@ -1129,10 +1129,10 @@ static int i915_audio_component_get_cdclk_freq(struct device *kdev) static struct intel_encoder *get_saved_enc(struct drm_i915_private *i915, int port, int pipe) { - struct intel_encoder *encoder; - /* MST */ if (pipe >= 0) { + struct intel_encoder *encoder; + if (drm_WARN_ON(&i915->drm, pipe >= ARRAY_SIZE(i915->display.audio.encoder_map))) return NULL; @@ -1143,7 +1143,7 @@ static struct intel_encoder *get_saved_enc(struct drm_i915_private *i915, * MST or not. So it will poll all the port & pipe * combinations */ - if (encoder != NULL && encoder->port == port && + if (encoder && encoder->port == port && encoder->type == INTEL_OUTPUT_DP_MST) return encoder; } @@ -1153,14 +1153,12 @@ static struct intel_encoder *get_saved_enc(struct drm_i915_private *i915, return NULL; for_each_pipe(i915, pipe) { - encoder = i915->display.audio.encoder_map[pipe]; - if (encoder == NULL) - continue; + struct intel_encoder *encoder; - if (encoder->type == INTEL_OUTPUT_DP_MST) - continue; + encoder = i915->display.audio.encoder_map[pipe]; - if (port == encoder->port) + if (encoder && encoder->port == port && + encoder->type != INTEL_OUTPUT_DP_MST) return encoder; } diff --git a/drivers/gpu/drm/i915/display/intel_audio_regs.h b/drivers/gpu/drm/i915/display/intel_audio_regs.h index 4f432c2eb543..616e7b1275c4 100644 --- a/drivers/gpu/drm/i915/display/intel_audio_regs.h +++ b/drivers/gpu/drm/i915/display/intel_audio_regs.h @@ -6,7 +6,7 @@ #ifndef __INTEL_AUDIO_REGS_H__ #define __INTEL_AUDIO_REGS_H__ -#include "i915_reg_defs.h" +#include "intel_display_reg_defs.h" #define G4X_AUD_CNTL_ST _MMIO(0x620B4) #define G4X_ELD_VALID REG_BIT(14) diff --git a/drivers/gpu/drm/i915/display/intel_backlight.c b/drivers/gpu/drm/i915/display/intel_backlight.c index beba39a38c87..71af88a70461 100644 --- a/drivers/gpu/drm/i915/display/intel_backlight.c +++ b/drivers/gpu/drm/i915/display/intel_backlight.c @@ -10,6 +10,7 @@ #include <acpi/video.h> +#include "i915_reg.h" #include "intel_backlight.h" #include "intel_backlight_regs.h" #include "intel_connector.h" diff --git a/drivers/gpu/drm/i915/display/intel_backlight_regs.h b/drivers/gpu/drm/i915/display/intel_backlight_regs.h index 50c1210f6d5d..344eb8096bd2 100644 --- a/drivers/gpu/drm/i915/display/intel_backlight_regs.h +++ b/drivers/gpu/drm/i915/display/intel_backlight_regs.h @@ -6,7 +6,7 @@ #ifndef __INTEL_BACKLIGHT_REGS_H__ #define __INTEL_BACKLIGHT_REGS_H__ -#include "i915_reg_defs.h" +#include "intel_display_reg_defs.h" #define _VLV_BLC_PWM_CTL2_A (DISPLAY_MMIO_BASE(dev_priv) + 0x61250) #define _VLV_BLC_PWM_CTL2_B (DISPLAY_MMIO_BASE(dev_priv) + 0x61350) diff --git a/drivers/gpu/drm/i915/display/intel_bw.c b/drivers/gpu/drm/i915/display/intel_bw.c index 4ace026b29bd..1c236f02b380 100644 --- a/drivers/gpu/drm/i915/display/intel_bw.c +++ b/drivers/gpu/drm/i915/display/intel_bw.c @@ -439,7 +439,8 @@ static int tgl_get_bw_info(struct drm_i915_private *dev_priv, const struct intel return ret; } - if (dram_info->type == INTEL_DRAM_LPDDR4 || dram_info->type == INTEL_DRAM_LPDDR5) + if (DISPLAY_VER(dev_priv) < 14 && + (dram_info->type == INTEL_DRAM_LPDDR4 || dram_info->type == INTEL_DRAM_LPDDR5)) num_channels *= 2; qi.deinterleave = qi.deinterleave ? : DIV_ROUND_UP(num_channels, is_y_tile ? 4 : 2); diff --git a/drivers/gpu/drm/i915/display/intel_cdclk.c b/drivers/gpu/drm/i915/display/intel_cdclk.c index eada931cb1c8..b74e36d76013 100644 --- a/drivers/gpu/drm/i915/display/intel_cdclk.c +++ b/drivers/gpu/drm/i915/display/intel_cdclk.c @@ -24,6 +24,7 @@ #include <linux/time.h> #include "hsw_ips.h" +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_atomic_plane.h" #include "intel_audio.h" @@ -2755,7 +2756,7 @@ int intel_modeset_calc_cdclk(struct intel_atomic_state *state) if (IS_ERR(crtc_state)) return PTR_ERR(crtc_state); - if (drm_atomic_crtc_needs_modeset(&crtc_state->uapi)) + if (intel_crtc_needs_modeset(crtc_state)) pipe = INVALID_PIPE; } diff --git a/drivers/gpu/drm/i915/display/intel_color.c b/drivers/gpu/drm/i915/display/intel_color.c index 4bb113c39f4b..250e83f1f5ac 100644 --- a/drivers/gpu/drm/i915/display/intel_color.c +++ b/drivers/gpu/drm/i915/display/intel_color.c @@ -22,6 +22,7 @@ * */ +#include "i915_reg.h" #include "intel_color.h" #include "intel_de.h" #include "intel_display_types.h" @@ -184,31 +185,31 @@ static void ilk_update_pipe_csc(struct intel_crtc *crtc, const u16 coeff[9], const u16 postoff[3]) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); enum pipe pipe = crtc->pipe; - intel_de_write_fw(dev_priv, PIPE_CSC_PREOFF_HI(pipe), preoff[0]); - intel_de_write_fw(dev_priv, PIPE_CSC_PREOFF_ME(pipe), preoff[1]); - intel_de_write_fw(dev_priv, PIPE_CSC_PREOFF_LO(pipe), preoff[2]); + intel_de_write_fw(i915, PIPE_CSC_PREOFF_HI(pipe), preoff[0]); + intel_de_write_fw(i915, PIPE_CSC_PREOFF_ME(pipe), preoff[1]); + intel_de_write_fw(i915, PIPE_CSC_PREOFF_LO(pipe), preoff[2]); - intel_de_write_fw(dev_priv, PIPE_CSC_COEFF_RY_GY(pipe), + intel_de_write_fw(i915, PIPE_CSC_COEFF_RY_GY(pipe), coeff[0] << 16 | coeff[1]); - intel_de_write_fw(dev_priv, PIPE_CSC_COEFF_BY(pipe), coeff[2] << 16); + intel_de_write_fw(i915, PIPE_CSC_COEFF_BY(pipe), coeff[2] << 16); - intel_de_write_fw(dev_priv, PIPE_CSC_COEFF_RU_GU(pipe), + intel_de_write_fw(i915, PIPE_CSC_COEFF_RU_GU(pipe), coeff[3] << 16 | coeff[4]); - intel_de_write_fw(dev_priv, PIPE_CSC_COEFF_BU(pipe), coeff[5] << 16); + intel_de_write_fw(i915, PIPE_CSC_COEFF_BU(pipe), coeff[5] << 16); - intel_de_write_fw(dev_priv, PIPE_CSC_COEFF_RV_GV(pipe), + intel_de_write_fw(i915, PIPE_CSC_COEFF_RV_GV(pipe), coeff[6] << 16 | coeff[7]); - intel_de_write_fw(dev_priv, PIPE_CSC_COEFF_BV(pipe), coeff[8] << 16); + intel_de_write_fw(i915, PIPE_CSC_COEFF_BV(pipe), coeff[8] << 16); - if (DISPLAY_VER(dev_priv) >= 7) { - intel_de_write_fw(dev_priv, PIPE_CSC_POSTOFF_HI(pipe), + if (DISPLAY_VER(i915) >= 7) { + intel_de_write_fw(i915, PIPE_CSC_POSTOFF_HI(pipe), postoff[0]); - intel_de_write_fw(dev_priv, PIPE_CSC_POSTOFF_ME(pipe), + intel_de_write_fw(i915, PIPE_CSC_POSTOFF_ME(pipe), postoff[1]); - intel_de_write_fw(dev_priv, PIPE_CSC_POSTOFF_LO(pipe), + intel_de_write_fw(i915, PIPE_CSC_POSTOFF_LO(pipe), postoff[2]); } } @@ -218,55 +219,55 @@ static void icl_update_output_csc(struct intel_crtc *crtc, const u16 coeff[9], const u16 postoff[3]) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); enum pipe pipe = crtc->pipe; - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_PREOFF_HI(pipe), preoff[0]); - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_PREOFF_ME(pipe), preoff[1]); - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_PREOFF_LO(pipe), preoff[2]); + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_PREOFF_HI(pipe), preoff[0]); + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_PREOFF_ME(pipe), preoff[1]); + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_PREOFF_LO(pipe), preoff[2]); - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_COEFF_RY_GY(pipe), + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_COEFF_RY_GY(pipe), coeff[0] << 16 | coeff[1]); - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_COEFF_BY(pipe), + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_COEFF_BY(pipe), coeff[2] << 16); - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_COEFF_RU_GU(pipe), + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_COEFF_RU_GU(pipe), coeff[3] << 16 | coeff[4]); - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_COEFF_BU(pipe), + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_COEFF_BU(pipe), coeff[5] << 16); - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_COEFF_RV_GV(pipe), + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_COEFF_RV_GV(pipe), coeff[6] << 16 | coeff[7]); - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_COEFF_BV(pipe), + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_COEFF_BV(pipe), coeff[8] << 16); - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_POSTOFF_HI(pipe), postoff[0]); - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_POSTOFF_ME(pipe), postoff[1]); - intel_de_write_fw(dev_priv, PIPE_CSC_OUTPUT_POSTOFF_LO(pipe), postoff[2]); + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_POSTOFF_HI(pipe), postoff[0]); + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_POSTOFF_ME(pipe), postoff[1]); + intel_de_write_fw(i915, PIPE_CSC_OUTPUT_POSTOFF_LO(pipe), postoff[2]); } static bool ilk_csc_limited_range(const struct intel_crtc_state *crtc_state) { - struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); + struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev); /* * FIXME if there's a gamma LUT after the CSC, we should * do the range compression using the gamma LUT instead. */ return crtc_state->limited_color_range && - (IS_HASWELL(dev_priv) || IS_BROADWELL(dev_priv) || - IS_DISPLAY_VER(dev_priv, 9, 10)); + (IS_HASWELL(i915) || IS_BROADWELL(i915) || + IS_DISPLAY_VER(i915, 9, 10)); } static void ilk_csc_convert_ctm(const struct intel_crtc_state *crtc_state, - u16 coeffs[9]) + u16 coeffs[9], bool limited_color_range) { const struct drm_color_ctm *ctm = crtc_state->hw.ctm->data; const u64 *input; u64 temp[9]; int i; - if (ilk_csc_limited_range(crtc_state)) + if (limited_color_range) input = ctm_mult_by_limited(temp, ctm->matrix); else input = ctm->matrix; @@ -313,13 +314,13 @@ static void ilk_csc_convert_ctm(const struct intel_crtc_state *crtc_state, static void ilk_load_csc_matrix(const struct intel_crtc_state *crtc_state) { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); bool limited_color_range = ilk_csc_limited_range(crtc_state); if (crtc_state->hw.ctm) { u16 coeff[9]; - ilk_csc_convert_ctm(crtc_state, coeff); + ilk_csc_convert_ctm(crtc_state, coeff, limited_color_range); ilk_update_pipe_csc(crtc, ilk_csc_off_zero, coeff, limited_color_range ? ilk_csc_postoff_limited_range : @@ -339,7 +340,7 @@ static void ilk_load_csc_matrix(const struct intel_crtc_state *crtc_state) * LUT is needed but CSC is not we need to load an * identity matrix. */ - drm_WARN_ON(&dev_priv->drm, !IS_GEMINILAKE(dev_priv)); + drm_WARN_ON(&i915->drm, !IS_GEMINILAKE(i915)); ilk_update_pipe_csc(crtc, ilk_csc_off_zero, ilk_csc_coeff_identity, @@ -354,7 +355,7 @@ static void icl_load_csc_matrix(const struct intel_crtc_state *crtc_state) if (crtc_state->hw.ctm) { u16 coeff[9]; - ilk_csc_convert_ctm(crtc_state, coeff); + ilk_csc_convert_ctm(crtc_state, coeff, false); ilk_update_pipe_csc(crtc, ilk_csc_off_zero, coeff, ilk_csc_off_zero); } @@ -373,7 +374,7 @@ static void icl_load_csc_matrix(const struct intel_crtc_state *crtc_state) static void chv_load_cgm_csc(struct intel_crtc *crtc, const struct drm_property_blob *blob) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); const struct drm_color_ctm *ctm = blob->data; enum pipe pipe = crtc->pipe; u16 coeffs[9]; @@ -397,15 +398,15 @@ static void chv_load_cgm_csc(struct intel_crtc *crtc, coeffs[i] |= (abs_coeff >> 20) & 0xfff; } - intel_de_write_fw(dev_priv, CGM_PIPE_CSC_COEFF01(pipe), + intel_de_write_fw(i915, CGM_PIPE_CSC_COEFF01(pipe), coeffs[1] << 16 | coeffs[0]); - intel_de_write_fw(dev_priv, CGM_PIPE_CSC_COEFF23(pipe), + intel_de_write_fw(i915, CGM_PIPE_CSC_COEFF23(pipe), coeffs[3] << 16 | coeffs[2]); - intel_de_write_fw(dev_priv, CGM_PIPE_CSC_COEFF45(pipe), + intel_de_write_fw(i915, CGM_PIPE_CSC_COEFF45(pipe), coeffs[5] << 16 | coeffs[4]); - intel_de_write_fw(dev_priv, CGM_PIPE_CSC_COEFF67(pipe), + intel_de_write_fw(i915, CGM_PIPE_CSC_COEFF67(pipe), coeffs[7] << 16 | coeffs[6]); - intel_de_write_fw(dev_priv, CGM_PIPE_CSC_COEFF8(pipe), + intel_de_write_fw(i915, CGM_PIPE_CSC_COEFF8(pipe), coeffs[8]); } @@ -424,32 +425,32 @@ static u32 intel_color_lut_pack(u32 val, int bit_precision) static u32 i9xx_lut_8(const struct drm_color_lut *color) { - return drm_color_lut_extract(color->red, 8) << 16 | - drm_color_lut_extract(color->green, 8) << 8 | - drm_color_lut_extract(color->blue, 8); + return REG_FIELD_PREP(PALETTE_RED_MASK, drm_color_lut_extract(color->red, 8)) | + REG_FIELD_PREP(PALETTE_GREEN_MASK, drm_color_lut_extract(color->green, 8)) | + REG_FIELD_PREP(PALETTE_BLUE_MASK, drm_color_lut_extract(color->blue, 8)); } static void i9xx_lut_8_pack(struct drm_color_lut *entry, u32 val) { - entry->red = intel_color_lut_pack(REG_FIELD_GET(LGC_PALETTE_RED_MASK, val), 8); - entry->green = intel_color_lut_pack(REG_FIELD_GET(LGC_PALETTE_GREEN_MASK, val), 8); - entry->blue = intel_color_lut_pack(REG_FIELD_GET(LGC_PALETTE_BLUE_MASK, val), 8); + entry->red = intel_color_lut_pack(REG_FIELD_GET(PALETTE_RED_MASK, val), 8); + entry->green = intel_color_lut_pack(REG_FIELD_GET(PALETTE_GREEN_MASK, val), 8); + entry->blue = intel_color_lut_pack(REG_FIELD_GET(PALETTE_BLUE_MASK, val), 8); } /* i965+ "10.6" bit interpolated format "even DW" (low 8 bits) */ static u32 i965_lut_10p6_ldw(const struct drm_color_lut *color) { - return (color->red & 0xff) << 16 | - (color->green & 0xff) << 8 | - (color->blue & 0xff); + return REG_FIELD_PREP(PALETTE_RED_MASK, color->red & 0xff) | + REG_FIELD_PREP(PALETTE_GREEN_MASK, color->green & 0xff) | + REG_FIELD_PREP(PALETTE_BLUE_MASK, color->blue & 0xff); } /* i965+ "10.6" interpolated format "odd DW" (high 8 bits) */ static u32 i965_lut_10p6_udw(const struct drm_color_lut *color) { - return (color->red >> 8) << 16 | - (color->green >> 8) << 8 | - (color->blue >> 8); + return REG_FIELD_PREP(PALETTE_RED_MASK, color->red >> 8) | + REG_FIELD_PREP(PALETTE_GREEN_MASK, color->green >> 8) | + REG_FIELD_PREP(PALETTE_BLUE_MASK, color->blue >> 8); } static void i965_lut_10p6_pack(struct drm_color_lut *entry, u32 ldw, u32 udw) @@ -470,26 +471,42 @@ static u16 i965_lut_11p6_max_pack(u32 val) static u32 ilk_lut_10(const struct drm_color_lut *color) { - return drm_color_lut_extract(color->red, 10) << 20 | - drm_color_lut_extract(color->green, 10) << 10 | - drm_color_lut_extract(color->blue, 10); + return REG_FIELD_PREP(PREC_PALETTE_10_RED_MASK, drm_color_lut_extract(color->red, 10)) | + REG_FIELD_PREP(PREC_PALETTE_10_GREEN_MASK, drm_color_lut_extract(color->green, 10)) | + REG_FIELD_PREP(PREC_PALETTE_10_BLUE_MASK, drm_color_lut_extract(color->blue, 10)); } static void ilk_lut_10_pack(struct drm_color_lut *entry, u32 val) { - entry->red = intel_color_lut_pack(REG_FIELD_GET(PREC_PALETTE_RED_MASK, val), 10); - entry->green = intel_color_lut_pack(REG_FIELD_GET(PREC_PALETTE_GREEN_MASK, val), 10); - entry->blue = intel_color_lut_pack(REG_FIELD_GET(PREC_PALETTE_BLUE_MASK, val), 10); + entry->red = intel_color_lut_pack(REG_FIELD_GET(PREC_PALETTE_10_RED_MASK, val), 10); + entry->green = intel_color_lut_pack(REG_FIELD_GET(PREC_PALETTE_10_GREEN_MASK, val), 10); + entry->blue = intel_color_lut_pack(REG_FIELD_GET(PREC_PALETTE_10_BLUE_MASK, val), 10); } -static void icl_lut_multi_seg_pack(struct drm_color_lut *entry, u32 ldw, u32 udw) +/* ilk+ "12.4" interpolated format (low 6 bits) */ +static u32 ilk_lut_12p4_ldw(const struct drm_color_lut *color) { - entry->red = REG_FIELD_GET(PAL_PREC_MULTI_SEG_RED_UDW_MASK, udw) << 6 | - REG_FIELD_GET(PAL_PREC_MULTI_SEG_RED_LDW_MASK, ldw); - entry->green = REG_FIELD_GET(PAL_PREC_MULTI_SEG_GREEN_UDW_MASK, udw) << 6 | - REG_FIELD_GET(PAL_PREC_MULTI_SEG_GREEN_LDW_MASK, ldw); - entry->blue = REG_FIELD_GET(PAL_PREC_MULTI_SEG_BLUE_UDW_MASK, udw) << 6 | - REG_FIELD_GET(PAL_PREC_MULTI_SEG_BLUE_LDW_MASK, ldw); + return REG_FIELD_PREP(PREC_PALETTE_12P4_RED_LDW_MASK, color->red & 0x3f) | + REG_FIELD_PREP(PREC_PALETTE_12P4_GREEN_LDW_MASK, color->green & 0x3f) | + REG_FIELD_PREP(PREC_PALETTE_12P4_BLUE_LDW_MASK, color->blue & 0x3f); +} + +/* ilk+ "12.4" interpolated format (high 10 bits) */ +static u32 ilk_lut_12p4_udw(const struct drm_color_lut *color) +{ + return REG_FIELD_PREP(PREC_PALETTE_12P4_RED_UDW_MASK, color->red >> 6) | + REG_FIELD_PREP(PREC_PALETTE_12P4_GREEN_UDW_MASK, color->green >> 6) | + REG_FIELD_PREP(PREC_PALETTE_12P4_BLUE_UDW_MASK, color->blue >> 6); +} + +static void ilk_lut_12p4_pack(struct drm_color_lut *entry, u32 ldw, u32 udw) +{ + entry->red = REG_FIELD_GET(PREC_PALETTE_12P4_RED_UDW_MASK, udw) << 6 | + REG_FIELD_GET(PREC_PALETTE_12P4_RED_LDW_MASK, ldw); + entry->green = REG_FIELD_GET(PREC_PALETTE_12P4_GREEN_UDW_MASK, udw) << 6 | + REG_FIELD_GET(PREC_PALETTE_12P4_GREEN_LDW_MASK, ldw); + entry->blue = REG_FIELD_GET(PREC_PALETTE_12P4_BLUE_UDW_MASK, udw) << 6 | + REG_FIELD_GET(PREC_PALETTE_12P4_BLUE_LDW_MASK, ldw); } static void icl_color_commit_noarm(const struct intel_crtc_state *crtc_state) @@ -511,31 +528,31 @@ static void i9xx_color_commit_arm(const struct intel_crtc_state *crtc_state) static void ilk_color_commit_arm(const struct intel_crtc_state *crtc_state) { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); /* update PIPECONF GAMMA_MODE */ ilk_set_pipeconf(crtc_state); - intel_de_write_fw(dev_priv, PIPE_CSC_MODE(crtc->pipe), + intel_de_write_fw(i915, PIPE_CSC_MODE(crtc->pipe), crtc_state->csc_mode); } static void hsw_color_commit_arm(const struct intel_crtc_state *crtc_state) { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); - intel_de_write(dev_priv, GAMMA_MODE(crtc->pipe), + intel_de_write(i915, GAMMA_MODE(crtc->pipe), crtc_state->gamma_mode); - intel_de_write_fw(dev_priv, PIPE_CSC_MODE(crtc->pipe), + intel_de_write_fw(i915, PIPE_CSC_MODE(crtc->pipe), crtc_state->csc_mode); } static void skl_color_commit_arm(const struct intel_crtc_state *crtc_state) { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); enum pipe pipe = crtc->pipe; u32 val = 0; @@ -548,12 +565,12 @@ static void skl_color_commit_arm(const struct intel_crtc_state *crtc_state) val |= SKL_BOTTOM_COLOR_GAMMA_ENABLE; if (crtc_state->csc_enable) val |= SKL_BOTTOM_COLOR_CSC_ENABLE; - intel_de_write(dev_priv, SKL_BOTTOM_COLOR(pipe), val); + intel_de_write(i915, SKL_BOTTOM_COLOR(pipe), val); - intel_de_write(dev_priv, GAMMA_MODE(crtc->pipe), + intel_de_write(i915, GAMMA_MODE(crtc->pipe), crtc_state->gamma_mode); - intel_de_write_fw(dev_priv, PIPE_CSC_MODE(crtc->pipe), + intel_de_write_fw(i915, PIPE_CSC_MODE(crtc->pipe), crtc_state->csc_mode); } @@ -565,7 +582,7 @@ create_linear_lut(struct drm_i915_private *i915, int lut_size) int i; blob = drm_property_create_blob(&i915->drm, - sizeof(struct drm_color_lut) * lut_size, + sizeof(lut[0]) * lut_size, NULL); if (IS_ERR(blob)) return blob; @@ -583,6 +600,30 @@ create_linear_lut(struct drm_i915_private *i915, int lut_size) return blob; } +static struct drm_property_blob * +create_resized_lut(struct drm_i915_private *i915, + const struct drm_property_blob *blob_in, int lut_out_size) +{ + int i, lut_in_size = drm_color_lut_size(blob_in); + struct drm_property_blob *blob_out; + const struct drm_color_lut *lut_in; + struct drm_color_lut *lut_out; + + blob_out = drm_property_create_blob(&i915->drm, + sizeof(lut_out[0]) * lut_out_size, + NULL); + if (IS_ERR(blob_out)) + return blob_out; + + lut_in = blob_in->data; + lut_out = blob_out->data; + + for (i = 0; i < lut_out_size; i++) + lut_out[i] = lut_in[i * (lut_in_size - 1) / (lut_out_size - 1)]; + + return blob_out; +} + static void i9xx_load_lut_8(struct intel_crtc *crtc, const struct drm_property_blob *blob) { @@ -643,7 +684,7 @@ static void i965_load_luts(const struct intel_crtc_state *crtc_state) static void ilk_load_lut_8(struct intel_crtc *crtc, const struct drm_property_blob *blob) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); const struct drm_color_lut *lut; enum pipe pipe = crtc->pipe; int i; @@ -654,20 +695,20 @@ static void ilk_load_lut_8(struct intel_crtc *crtc, lut = blob->data; for (i = 0; i < 256; i++) - intel_de_write_fw(dev_priv, LGC_PALETTE(pipe, i), + intel_de_write_fw(i915, LGC_PALETTE(pipe, i), i9xx_lut_8(&lut[i])); } static void ilk_load_lut_10(struct intel_crtc *crtc, const struct drm_property_blob *blob) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); const struct drm_color_lut *lut = blob->data; int i, lut_size = drm_color_lut_size(blob); enum pipe pipe = crtc->pipe; for (i = 0; i < lut_size; i++) - intel_de_write_fw(dev_priv, PREC_PALETTE(pipe, i), + intel_de_write_fw(i915, PREC_PALETTE(pipe, i), ilk_lut_10(&lut[i])); } @@ -708,27 +749,22 @@ static void ivb_load_lut_10(struct intel_crtc *crtc, const struct drm_property_blob *blob, u32 prec_index) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - int hw_lut_size = ivb_lut_10_size(prec_index); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); const struct drm_color_lut *lut = blob->data; int i, lut_size = drm_color_lut_size(blob); enum pipe pipe = crtc->pipe; - for (i = 0; i < hw_lut_size; i++) { - /* We discard half the user entries in split gamma mode */ - const struct drm_color_lut *entry = - &lut[i * (lut_size - 1) / (hw_lut_size - 1)]; - - intel_de_write_fw(dev_priv, PREC_PAL_INDEX(pipe), prec_index++); - intel_de_write_fw(dev_priv, PREC_PAL_DATA(pipe), - ilk_lut_10(entry)); + for (i = 0; i < lut_size; i++) { + intel_de_write_fw(i915, PREC_PAL_INDEX(pipe), prec_index++); + intel_de_write_fw(i915, PREC_PAL_DATA(pipe), + ilk_lut_10(&lut[i])); } /* * Reset the index, otherwise it prevents the legacy palette to be * written properly. */ - intel_de_write_fw(dev_priv, PREC_PAL_INDEX(pipe), 0); + intel_de_write_fw(i915, PREC_PAL_INDEX(pipe), 0); } /* On BDW+ the index auto increment mode actually works */ @@ -736,55 +772,45 @@ static void bdw_load_lut_10(struct intel_crtc *crtc, const struct drm_property_blob *blob, u32 prec_index) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - int hw_lut_size = ivb_lut_10_size(prec_index); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); const struct drm_color_lut *lut = blob->data; int i, lut_size = drm_color_lut_size(blob); enum pipe pipe = crtc->pipe; - intel_de_write_fw(dev_priv, PREC_PAL_INDEX(pipe), + intel_de_write_fw(i915, PREC_PAL_INDEX(pipe), prec_index | PAL_PREC_AUTO_INCREMENT); - for (i = 0; i < hw_lut_size; i++) { - /* We discard half the user entries in split gamma mode */ - const struct drm_color_lut *entry = - &lut[i * (lut_size - 1) / (hw_lut_size - 1)]; - - intel_de_write_fw(dev_priv, PREC_PAL_DATA(pipe), - ilk_lut_10(entry)); - } + for (i = 0; i < lut_size; i++) + intel_de_write_fw(i915, PREC_PAL_DATA(pipe), + ilk_lut_10(&lut[i])); /* * Reset the index, otherwise it prevents the legacy palette to be * written properly. */ - intel_de_write_fw(dev_priv, PREC_PAL_INDEX(pipe), 0); + intel_de_write_fw(i915, PREC_PAL_INDEX(pipe), 0); } static void ivb_load_lut_ext_max(const struct intel_crtc_state *crtc_state) { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); enum pipe pipe = crtc->pipe; /* Program the max register to clamp values > 1.0. */ intel_dsb_reg_write(crtc_state, PREC_PAL_EXT_GC_MAX(pipe, 0), 1 << 16); intel_dsb_reg_write(crtc_state, PREC_PAL_EXT_GC_MAX(pipe, 1), 1 << 16); intel_dsb_reg_write(crtc_state, PREC_PAL_EXT_GC_MAX(pipe, 2), 1 << 16); +} - /* - * Program the gc max 2 register to clamp values > 1.0. - * ToDo: Extend the ABI to be able to program values - * from 3.0 to 7.0 - */ - if (DISPLAY_VER(dev_priv) >= 10) { - intel_dsb_reg_write(crtc_state, PREC_PAL_EXT2_GC_MAX(pipe, 0), - 1 << 16); - intel_dsb_reg_write(crtc_state, PREC_PAL_EXT2_GC_MAX(pipe, 1), - 1 << 16); - intel_dsb_reg_write(crtc_state, PREC_PAL_EXT2_GC_MAX(pipe, 2), - 1 << 16); - } +static void glk_load_lut_ext2_max(const struct intel_crtc_state *crtc_state) +{ + struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); + enum pipe pipe = crtc->pipe; + + /* Program the max register to clamp values > 1.0. */ + intel_dsb_reg_write(crtc_state, PREC_PAL_EXT2_GC_MAX(pipe, 0), 1 << 16); + intel_dsb_reg_write(crtc_state, PREC_PAL_EXT2_GC_MAX(pipe, 1), 1 << 16); + intel_dsb_reg_write(crtc_state, PREC_PAL_EXT2_GC_MAX(pipe, 2), 1 << 16); } static void ivb_load_luts(const struct intel_crtc_state *crtc_state) @@ -858,7 +884,7 @@ static void glk_load_degamma_lut(const struct intel_crtc_state *crtc_state, const struct drm_property_blob *blob) { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); const struct drm_color_lut *lut = blob->data; int i, lut_size = drm_color_lut_size(blob); enum pipe pipe = crtc->pipe; @@ -868,8 +894,8 @@ static void glk_load_degamma_lut(const struct intel_crtc_state *crtc_state, * ignore the index bits, so we need to reset it to index 0 * separately. */ - intel_de_write_fw(dev_priv, PRE_CSC_GAMC_INDEX(pipe), 0); - intel_de_write_fw(dev_priv, PRE_CSC_GAMC_INDEX(pipe), + intel_de_write_fw(i915, PRE_CSC_GAMC_INDEX(pipe), 0); + intel_de_write_fw(i915, PRE_CSC_GAMC_INDEX(pipe), PRE_CSC_GAMC_AUTO_INCREMENT); for (i = 0; i < lut_size; i++) { @@ -886,15 +912,15 @@ static void glk_load_degamma_lut(const struct intel_crtc_state *crtc_state, * ToDo: Extend to max 7.0. Enable 32 bit input value * as compared to just 16 to achieve this. */ - intel_de_write_fw(dev_priv, PRE_CSC_GAMC_DATA(pipe), + intel_de_write_fw(i915, PRE_CSC_GAMC_DATA(pipe), lut[i].green); } /* Clamp values > 1.0. */ - while (i++ < glk_degamma_lut_size(dev_priv)) - intel_de_write_fw(dev_priv, PRE_CSC_GAMC_DATA(pipe), 1 << 16); + while (i++ < glk_degamma_lut_size(i915)) + intel_de_write_fw(i915, PRE_CSC_GAMC_DATA(pipe), 1 << 16); - intel_de_write_fw(dev_priv, PRE_CSC_GAMC_INDEX(pipe), 0); + intel_de_write_fw(i915, PRE_CSC_GAMC_INDEX(pipe), 0); } static void glk_load_luts(const struct intel_crtc_state *crtc_state) @@ -913,6 +939,7 @@ static void glk_load_luts(const struct intel_crtc_state *crtc_state) case GAMMA_MODE_MODE_10BIT: bdw_load_lut_10(crtc, post_csc_lut, PAL_PREC_INDEX_VALUE(0)); ivb_load_lut_ext_max(crtc_state); + glk_load_lut_ext2_max(crtc_state); break; default: MISSING_CASE(crtc_state->gamma_mode); @@ -920,23 +947,9 @@ static void glk_load_luts(const struct intel_crtc_state *crtc_state) } } -/* ilk+ "12.4" interpolated format (high 10 bits) */ -static u32 ilk_lut_12p4_udw(const struct drm_color_lut *color) -{ - return (color->red >> 6) << 20 | (color->green >> 6) << 10 | - (color->blue >> 6); -} - -/* ilk+ "12.4" interpolated format (low 6 bits) */ -static u32 ilk_lut_12p4_ldw(const struct drm_color_lut *color) -{ - return (color->red & 0x3f) << 24 | (color->green & 0x3f) << 14 | - (color->blue & 0x3f) << 4; -} - static void -icl_load_gcmax(const struct intel_crtc_state *crtc_state, - const struct drm_color_lut *color) +ivb_load_lut_max(const struct intel_crtc_state *crtc_state, + const struct drm_color_lut *color) { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); enum pipe pipe = crtc->pipe; @@ -1028,8 +1041,7 @@ icl_program_gamma_multi_segment(const struct intel_crtc_state *crtc_state) /* The last entry in the LUT is to be programmed in GCMAX */ entry = &lut[256 * 8 * 128]; - icl_load_gcmax(crtc_state, entry); - ivb_load_lut_ext_max(crtc_state); + ivb_load_lut_max(crtc_state, entry); } static void icl_load_luts(const struct intel_crtc_state *crtc_state) @@ -1048,10 +1060,13 @@ static void icl_load_luts(const struct intel_crtc_state *crtc_state) case GAMMA_MODE_MODE_12BIT_MULTI_SEGMENTED: icl_program_gamma_superfine_segment(crtc_state); icl_program_gamma_multi_segment(crtc_state); + ivb_load_lut_ext_max(crtc_state); + glk_load_lut_ext2_max(crtc_state); break; case GAMMA_MODE_MODE_10BIT: bdw_load_lut_10(crtc, post_csc_lut, PAL_PREC_INDEX_VALUE(0)); ivb_load_lut_ext_max(crtc_state); + glk_load_lut_ext2_max(crtc_state); break; default: MISSING_CASE(crtc_state->gamma_mode); @@ -1063,61 +1078,61 @@ static void icl_load_luts(const struct intel_crtc_state *crtc_state) static u32 chv_cgm_degamma_ldw(const struct drm_color_lut *color) { - return drm_color_lut_extract(color->green, 14) << 16 | - drm_color_lut_extract(color->blue, 14); + return REG_FIELD_PREP(CGM_PIPE_DEGAMMA_GREEN_LDW_MASK, drm_color_lut_extract(color->green, 14)) | + REG_FIELD_PREP(CGM_PIPE_DEGAMMA_BLUE_LDW_MASK, drm_color_lut_extract(color->blue, 14)); } static u32 chv_cgm_degamma_udw(const struct drm_color_lut *color) { - return drm_color_lut_extract(color->red, 14); + return REG_FIELD_PREP(CGM_PIPE_DEGAMMA_RED_UDW_MASK, drm_color_lut_extract(color->red, 14)); } static void chv_load_cgm_degamma(struct intel_crtc *crtc, const struct drm_property_blob *blob) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); const struct drm_color_lut *lut = blob->data; int i, lut_size = drm_color_lut_size(blob); enum pipe pipe = crtc->pipe; for (i = 0; i < lut_size; i++) { - intel_de_write_fw(dev_priv, CGM_PIPE_DEGAMMA(pipe, i, 0), + intel_de_write_fw(i915, CGM_PIPE_DEGAMMA(pipe, i, 0), chv_cgm_degamma_ldw(&lut[i])); - intel_de_write_fw(dev_priv, CGM_PIPE_DEGAMMA(pipe, i, 1), + intel_de_write_fw(i915, CGM_PIPE_DEGAMMA(pipe, i, 1), chv_cgm_degamma_udw(&lut[i])); } } static u32 chv_cgm_gamma_ldw(const struct drm_color_lut *color) { - return drm_color_lut_extract(color->green, 10) << 16 | - drm_color_lut_extract(color->blue, 10); + return REG_FIELD_PREP(CGM_PIPE_GAMMA_GREEN_LDW_MASK, drm_color_lut_extract(color->green, 10)) | + REG_FIELD_PREP(CGM_PIPE_GAMMA_BLUE_LDW_MASK, drm_color_lut_extract(color->blue, 10)); } static u32 chv_cgm_gamma_udw(const struct drm_color_lut *color) { - return drm_color_lut_extract(color->red, 10); + return REG_FIELD_PREP(CGM_PIPE_GAMMA_RED_UDW_MASK, drm_color_lut_extract(color->red, 10)); } static void chv_cgm_gamma_pack(struct drm_color_lut *entry, u32 ldw, u32 udw) { - entry->green = intel_color_lut_pack(REG_FIELD_GET(CGM_PIPE_GAMMA_GREEN_MASK, ldw), 10); - entry->blue = intel_color_lut_pack(REG_FIELD_GET(CGM_PIPE_GAMMA_BLUE_MASK, ldw), 10); - entry->red = intel_color_lut_pack(REG_FIELD_GET(CGM_PIPE_GAMMA_RED_MASK, udw), 10); + entry->green = intel_color_lut_pack(REG_FIELD_GET(CGM_PIPE_GAMMA_GREEN_LDW_MASK, ldw), 10); + entry->blue = intel_color_lut_pack(REG_FIELD_GET(CGM_PIPE_GAMMA_BLUE_LDW_MASK, ldw), 10); + entry->red = intel_color_lut_pack(REG_FIELD_GET(CGM_PIPE_GAMMA_RED_UDW_MASK, udw), 10); } static void chv_load_cgm_gamma(struct intel_crtc *crtc, const struct drm_property_blob *blob) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); const struct drm_color_lut *lut = blob->data; int i, lut_size = drm_color_lut_size(blob); enum pipe pipe = crtc->pipe; for (i = 0; i < lut_size; i++) { - intel_de_write_fw(dev_priv, CGM_PIPE_GAMMA(pipe, i, 0), + intel_de_write_fw(i915, CGM_PIPE_GAMMA(pipe, i, 0), chv_cgm_gamma_ldw(&lut[i])); - intel_de_write_fw(dev_priv, CGM_PIPE_GAMMA(pipe, i, 1), + intel_de_write_fw(i915, CGM_PIPE_GAMMA(pipe, i, 1), chv_cgm_gamma_udw(&lut[i])); } } @@ -1125,7 +1140,7 @@ static void chv_load_cgm_gamma(struct intel_crtc *crtc, static void chv_load_luts(const struct intel_crtc_state *crtc_state) { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); const struct drm_property_blob *pre_csc_lut = crtc_state->pre_csc_lut; const struct drm_property_blob *post_csc_lut = crtc_state->post_csc_lut; const struct drm_property_blob *ctm = crtc_state->hw.ctm; @@ -1141,30 +1156,30 @@ static void chv_load_luts(const struct intel_crtc_state *crtc_state) else i965_load_luts(crtc_state); - intel_de_write_fw(dev_priv, CGM_PIPE_MODE(crtc->pipe), + intel_de_write_fw(i915, CGM_PIPE_MODE(crtc->pipe), crtc_state->cgm_mode); } void intel_color_load_luts(const struct intel_crtc_state *crtc_state) { - struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); + struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev); - dev_priv->display.funcs.color->load_luts(crtc_state); + i915->display.funcs.color->load_luts(crtc_state); } void intel_color_commit_noarm(const struct intel_crtc_state *crtc_state) { - struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); + struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev); - if (dev_priv->display.funcs.color->color_commit_noarm) - dev_priv->display.funcs.color->color_commit_noarm(crtc_state); + if (i915->display.funcs.color->color_commit_noarm) + i915->display.funcs.color->color_commit_noarm(crtc_state); } void intel_color_commit_arm(const struct intel_crtc_state *crtc_state) { - struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); + struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev); - dev_priv->display.funcs.color->color_commit_arm(crtc_state); + i915->display.funcs.color->color_commit_arm(crtc_state); } static bool intel_can_preload_luts(const struct intel_crtc_state *new_crtc_state) @@ -1200,23 +1215,23 @@ static bool chv_can_preload_luts(const struct intel_crtc_state *new_crtc_state) int intel_color_check(struct intel_crtc_state *crtc_state) { - struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); + struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev); - return dev_priv->display.funcs.color->color_check(crtc_state); + return i915->display.funcs.color->color_check(crtc_state); } void intel_color_get_config(struct intel_crtc_state *crtc_state) { - struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); + struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev); - if (dev_priv->display.funcs.color->read_luts) - dev_priv->display.funcs.color->read_luts(crtc_state); + if (i915->display.funcs.color->read_luts) + i915->display.funcs.color->read_luts(crtc_state); } static bool need_plane_update(struct intel_plane *plane, const struct intel_crtc_state *crtc_state) { - struct drm_i915_private *dev_priv = to_i915(plane->base.dev); + struct drm_i915_private *i915 = to_i915(plane->base.dev); /* * On pre-SKL the pipe gamma enable and pipe csc enable for @@ -1224,7 +1239,7 @@ static bool need_plane_update(struct intel_plane *plane, * We have to reconfigure that even if the plane is inactive. */ return crtc_state->active_planes & BIT(plane->id) || - (DISPLAY_VER(dev_priv) < 9 && + (DISPLAY_VER(i915) < 9 && plane->id == PLANE_PRIMARY); } @@ -1232,7 +1247,7 @@ static int intel_color_add_affected_planes(struct intel_crtc_state *new_crtc_state) { struct intel_crtc *crtc = to_intel_crtc(new_crtc_state->uapi.crtc); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); struct intel_atomic_state *state = to_intel_atomic_state(new_crtc_state->uapi.state); const struct intel_crtc_state *old_crtc_state = @@ -1240,14 +1255,14 @@ intel_color_add_affected_planes(struct intel_crtc_state *new_crtc_state) struct intel_plane *plane; if (!new_crtc_state->hw.active || - drm_atomic_crtc_needs_modeset(&new_crtc_state->uapi)) + intel_crtc_needs_modeset(new_crtc_state)) return 0; if (new_crtc_state->gamma_enable == old_crtc_state->gamma_enable && new_crtc_state->csc_enable == old_crtc_state->csc_enable) return 0; - for_each_intel_plane_on_crtc(&dev_priv->drm, crtc, plane) { + for_each_intel_plane_on_crtc(&i915->drm, crtc, plane) { struct intel_plane_state *plane_state; if (!need_plane_update(plane, new_crtc_state)) @@ -1260,7 +1275,7 @@ intel_color_add_affected_planes(struct intel_crtc_state *new_crtc_state) new_crtc_state->update_planes |= BIT(plane->id); /* plane control register changes blocked by CxSR */ - if (HAS_GMCH(dev_priv)) + if (HAS_GMCH(i915)) new_crtc_state->disable_cxsr = true; } @@ -1286,7 +1301,7 @@ static int check_lut_size(const struct drm_property_blob *lut, int expected) static int check_luts(const struct intel_crtc_state *crtc_state) { - struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); + struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev); const struct drm_property_blob *gamma_lut = crtc_state->hw.gamma_lut; const struct drm_property_blob *degamma_lut = crtc_state->hw.degamma_lut; int gamma_length, degamma_length; @@ -1298,15 +1313,15 @@ static int check_luts(const struct intel_crtc_state *crtc_state) /* C8 relies on its palette being stored in the legacy LUT */ if (crtc_state->c8_planes) { - drm_dbg_kms(&dev_priv->drm, + drm_dbg_kms(&i915->drm, "C8 pixelformat requires the legacy LUT\n"); return -EINVAL; } - degamma_length = INTEL_INFO(dev_priv)->display.color.degamma_lut_size; - gamma_length = INTEL_INFO(dev_priv)->display.color.gamma_lut_size; - degamma_tests = INTEL_INFO(dev_priv)->display.color.degamma_lut_tests; - gamma_tests = INTEL_INFO(dev_priv)->display.color.gamma_lut_tests; + degamma_length = INTEL_INFO(i915)->display.color.degamma_lut_size; + gamma_length = INTEL_INFO(i915)->display.color.gamma_lut_size; + degamma_tests = INTEL_INFO(i915)->display.color.degamma_lut_tests; + gamma_tests = INTEL_INFO(i915)->display.color.gamma_lut_tests; if (check_lut_size(degamma_lut, degamma_length) || check_lut_size(gamma_lut, gamma_length)) @@ -1344,7 +1359,7 @@ void intel_color_assert_luts(const struct intel_crtc_state *crtc_state) crtc_state->pre_csc_lut != i915->display.color.glk_linear_degamma_lut); drm_WARN_ON(&i915->drm, crtc_state->post_csc_lut != crtc_state->hw.gamma_lut); - } else { + } else if (crtc_state->gamma_mode != GAMMA_MODE_MODE_SPLIT) { drm_WARN_ON(&i915->drm, crtc_state->pre_csc_lut != crtc_state->hw.degamma_lut && crtc_state->pre_csc_lut != crtc_state->hw.gamma_lut); @@ -1443,6 +1458,20 @@ static int chv_color_check(struct intel_crtc_state *crtc_state) return 0; } +static bool ilk_gamma_enable(const struct intel_crtc_state *crtc_state) +{ + return (crtc_state->hw.gamma_lut || + crtc_state->hw.degamma_lut) && + !crtc_state->c8_planes; +} + +static bool ilk_csc_enable(const struct intel_crtc_state *crtc_state) +{ + return crtc_state->output_format != INTEL_OUTPUT_FORMAT_RGB || + ilk_csc_limited_range(crtc_state) || + crtc_state->hw.ctm; +} + static u32 ilk_gamma_mode(const struct intel_crtc_state *crtc_state) { if (!crtc_state->gamma_enable || @@ -1488,22 +1517,29 @@ static void ilk_assign_luts(struct intel_crtc_state *crtc_state) static int ilk_color_check(struct intel_crtc_state *crtc_state) { + struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev); int ret; ret = check_luts(crtc_state); if (ret) return ret; - crtc_state->gamma_enable = - crtc_state->hw.gamma_lut && - !crtc_state->c8_planes; + if (crtc_state->hw.degamma_lut && crtc_state->hw.gamma_lut) { + drm_dbg_kms(&i915->drm, + "Degamma and gamma together are not possible\n"); + return -EINVAL; + } - /* - * We don't expose the ctm on ilk/snb currently, also RGB - * limited range output is handled by the hw automagically. - */ - crtc_state->csc_enable = - crtc_state->output_format != INTEL_OUTPUT_FORMAT_RGB; + if (crtc_state->output_format != INTEL_OUTPUT_FORMAT_RGB && + crtc_state->hw.ctm) { + drm_dbg_kms(&i915->drm, + "YCbCr and CTM together are not possible\n"); + return -EINVAL; + } + + crtc_state->gamma_enable = ilk_gamma_enable(crtc_state); + + crtc_state->csc_enable = ilk_csc_enable(crtc_state); crtc_state->gamma_mode = ilk_gamma_mode(crtc_state); @@ -1522,14 +1558,10 @@ static int ilk_color_check(struct intel_crtc_state *crtc_state) static u32 ivb_gamma_mode(const struct intel_crtc_state *crtc_state) { - if (!crtc_state->gamma_enable || - crtc_state_is_legacy_gamma(crtc_state)) - return GAMMA_MODE_MODE_8BIT; - else if (crtc_state->hw.gamma_lut && - crtc_state->hw.degamma_lut) + if (crtc_state->hw.degamma_lut && crtc_state->hw.gamma_lut) return GAMMA_MODE_MODE_SPLIT; - else - return GAMMA_MODE_MODE_10BIT; + + return ilk_gamma_mode(crtc_state); } static u32 ivb_csc_mode(const struct intel_crtc_state *crtc_state) @@ -1548,10 +1580,41 @@ static u32 ivb_csc_mode(const struct intel_crtc_state *crtc_state) return CSC_POSITION_BEFORE_GAMMA; } +static int ivb_assign_luts(struct intel_crtc_state *crtc_state) +{ + struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev); + struct drm_property_blob *degamma_lut, *gamma_lut; + + if (crtc_state->gamma_mode != GAMMA_MODE_MODE_SPLIT) { + ilk_assign_luts(crtc_state); + return 0; + } + + drm_WARN_ON(&i915->drm, drm_color_lut_size(crtc_state->hw.degamma_lut) != 1024); + drm_WARN_ON(&i915->drm, drm_color_lut_size(crtc_state->hw.gamma_lut) != 1024); + + degamma_lut = create_resized_lut(i915, crtc_state->hw.degamma_lut, 512); + if (IS_ERR(degamma_lut)) + return PTR_ERR(degamma_lut); + + gamma_lut = create_resized_lut(i915, crtc_state->hw.gamma_lut, 512); + if (IS_ERR(gamma_lut)) { + drm_property_blob_put(degamma_lut); + return PTR_ERR(gamma_lut); + } + + drm_property_replace_blob(&crtc_state->pre_csc_lut, degamma_lut); + drm_property_replace_blob(&crtc_state->post_csc_lut, gamma_lut); + + drm_property_blob_put(degamma_lut); + drm_property_blob_put(gamma_lut); + + return 0; +} + static int ivb_color_check(struct intel_crtc_state *crtc_state) { - struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); - bool limited_color_range = ilk_csc_limited_range(crtc_state); + struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev); int ret; ret = check_luts(crtc_state); @@ -1560,19 +1623,21 @@ static int ivb_color_check(struct intel_crtc_state *crtc_state) if (crtc_state->output_format != INTEL_OUTPUT_FORMAT_RGB && crtc_state->hw.ctm) { - drm_dbg_kms(&dev_priv->drm, - "YCBCR and CTM together are not possible\n"); + drm_dbg_kms(&i915->drm, + "YCbCr and CTM together are not possible\n"); return -EINVAL; } - crtc_state->gamma_enable = - (crtc_state->hw.gamma_lut || - crtc_state->hw.degamma_lut) && - !crtc_state->c8_planes; + if (crtc_state->output_format != INTEL_OUTPUT_FORMAT_RGB && + crtc_state->hw.degamma_lut && crtc_state->hw.gamma_lut) { + drm_dbg_kms(&i915->drm, + "YCbCr and degamma+gamma together are not possible\n"); + return -EINVAL; + } - crtc_state->csc_enable = - crtc_state->output_format != INTEL_OUTPUT_FORMAT_RGB || - crtc_state->hw.ctm || limited_color_range; + crtc_state->gamma_enable = ilk_gamma_enable(crtc_state); + + crtc_state->csc_enable = ilk_csc_enable(crtc_state); crtc_state->gamma_mode = ivb_gamma_mode(crtc_state); @@ -1582,7 +1647,9 @@ static int ivb_color_check(struct intel_crtc_state *crtc_state) if (ret) return ret; - ilk_assign_luts(crtc_state); + ret = ivb_assign_luts(crtc_state); + if (ret) + return ret; crtc_state->preload_luts = intel_can_preload_luts(crtc_state); @@ -1617,7 +1684,7 @@ static void glk_assign_luts(struct intel_crtc_state *crtc_state) static int glk_color_check(struct intel_crtc_state *crtc_state) { - struct drm_i915_private *dev_priv = to_i915(crtc_state->uapi.crtc->dev); + struct drm_i915_private *i915 = to_i915(crtc_state->uapi.crtc->dev); int ret; ret = check_luts(crtc_state); @@ -1626,8 +1693,15 @@ static int glk_color_check(struct intel_crtc_state *crtc_state) if (crtc_state->output_format != INTEL_OUTPUT_FORMAT_RGB && crtc_state->hw.ctm) { - drm_dbg_kms(&dev_priv->drm, - "YCBCR and CTM together are not possible\n"); + drm_dbg_kms(&i915->drm, + "YCbCr and CTM together are not possible\n"); + return -EINVAL; + } + + if (crtc_state->output_format != INTEL_OUTPUT_FORMAT_RGB && + crtc_state->hw.degamma_lut && crtc_state->hw.gamma_lut) { + drm_dbg_kms(&i915->drm, + "YCbCr and degamma+gamma together are not possible\n"); return -EINVAL; } @@ -1798,19 +1872,19 @@ static int icl_gamma_precision(const struct intel_crtc_state *crtc_state) int intel_color_get_gamma_bit_precision(const struct intel_crtc_state *crtc_state) { struct intel_crtc *crtc = to_intel_crtc(crtc_state->uapi.crtc); - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); - if (HAS_GMCH(dev_priv)) { - if (IS_CHERRYVIEW(dev_priv)) + if (HAS_GMCH(i915)) { + if (IS_CHERRYVIEW(i915)) return chv_gamma_precision(crtc_state); else return i9xx_gamma_precision(crtc_state); } else { - if (DISPLAY_VER(dev_priv) >= 11) + if (DISPLAY_VER(i915) >= 11) return icl_gamma_precision(crtc_state); - else if (DISPLAY_VER(dev_priv) == 10) + else if (DISPLAY_VER(i915) == 10) return glk_gamma_precision(crtc_state); - else if (IS_IRONLAKE(dev_priv)) + else if (IS_IRONLAKE(i915)) return ilk_gamma_precision(crtc_state); } @@ -1895,7 +1969,7 @@ static struct drm_property_blob *i9xx_read_lut_8(struct intel_crtc *crtc) int i; blob = drm_property_create_blob(&dev_priv->drm, - sizeof(struct drm_color_lut) * LEGACY_LUT_LENGTH, + sizeof(lut[0]) * LEGACY_LUT_LENGTH, NULL); if (IS_ERR(blob)) return NULL; @@ -1930,7 +2004,7 @@ static struct drm_property_blob *i965_read_lut_10p6(struct intel_crtc *crtc) struct drm_color_lut *lut; blob = drm_property_create_blob(&dev_priv->drm, - sizeof(struct drm_color_lut) * lut_size, + sizeof(lut[0]) * lut_size, NULL); if (IS_ERR(blob)) return NULL; @@ -1966,14 +2040,14 @@ static void i965_read_luts(struct intel_crtc_state *crtc_state) static struct drm_property_blob *chv_read_cgm_gamma(struct intel_crtc *crtc) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - int i, lut_size = INTEL_INFO(dev_priv)->display.color.gamma_lut_size; + struct drm_i915_private *i915 = to_i915(crtc->base.dev); + int i, lut_size = INTEL_INFO(i915)->display.color.gamma_lut_size; enum pipe pipe = crtc->pipe; struct drm_property_blob *blob; struct drm_color_lut *lut; - blob = drm_property_create_blob(&dev_priv->drm, - sizeof(struct drm_color_lut) * lut_size, + blob = drm_property_create_blob(&i915->drm, + sizeof(lut[0]) * lut_size, NULL); if (IS_ERR(blob)) return NULL; @@ -1981,8 +2055,8 @@ static struct drm_property_blob *chv_read_cgm_gamma(struct intel_crtc *crtc) lut = blob->data; for (i = 0; i < lut_size; i++) { - u32 ldw = intel_de_read_fw(dev_priv, CGM_PIPE_GAMMA(pipe, i, 0)); - u32 udw = intel_de_read_fw(dev_priv, CGM_PIPE_GAMMA(pipe, i, 1)); + u32 ldw = intel_de_read_fw(i915, CGM_PIPE_GAMMA(pipe, i, 0)); + u32 udw = intel_de_read_fw(i915, CGM_PIPE_GAMMA(pipe, i, 1)); chv_cgm_gamma_pack(&lut[i], ldw, udw); } @@ -2002,14 +2076,14 @@ static void chv_read_luts(struct intel_crtc_state *crtc_state) static struct drm_property_blob *ilk_read_lut_8(struct intel_crtc *crtc) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); enum pipe pipe = crtc->pipe; struct drm_property_blob *blob; struct drm_color_lut *lut; int i; - blob = drm_property_create_blob(&dev_priv->drm, - sizeof(struct drm_color_lut) * LEGACY_LUT_LENGTH, + blob = drm_property_create_blob(&i915->drm, + sizeof(lut[0]) * LEGACY_LUT_LENGTH, NULL); if (IS_ERR(blob)) return NULL; @@ -2017,7 +2091,7 @@ static struct drm_property_blob *ilk_read_lut_8(struct intel_crtc *crtc) lut = blob->data; for (i = 0; i < LEGACY_LUT_LENGTH; i++) { - u32 val = intel_de_read_fw(dev_priv, LGC_PALETTE(pipe, i)); + u32 val = intel_de_read_fw(i915, LGC_PALETTE(pipe, i)); i9xx_lut_8_pack(&lut[i], val); } @@ -2027,14 +2101,14 @@ static struct drm_property_blob *ilk_read_lut_8(struct intel_crtc *crtc) static struct drm_property_blob *ilk_read_lut_10(struct intel_crtc *crtc) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - int i, lut_size = INTEL_INFO(dev_priv)->display.color.gamma_lut_size; + struct drm_i915_private *i915 = to_i915(crtc->base.dev); + int i, lut_size = INTEL_INFO(i915)->display.color.gamma_lut_size; enum pipe pipe = crtc->pipe; struct drm_property_blob *blob; struct drm_color_lut *lut; - blob = drm_property_create_blob(&dev_priv->drm, - sizeof(struct drm_color_lut) * lut_size, + blob = drm_property_create_blob(&i915->drm, + sizeof(lut[0]) * lut_size, NULL); if (IS_ERR(blob)) return NULL; @@ -2042,7 +2116,7 @@ static struct drm_property_blob *ilk_read_lut_10(struct intel_crtc *crtc) lut = blob->data; for (i = 0; i < lut_size; i++) { - u32 val = intel_de_read_fw(dev_priv, PREC_PALETTE(pipe, i)); + u32 val = intel_de_read_fw(i915, PREC_PALETTE(pipe, i)); ilk_lut_10_pack(&lut[i], val); } @@ -2077,33 +2151,33 @@ static void ilk_read_luts(struct intel_crtc_state *crtc_state) static struct drm_property_blob *bdw_read_lut_10(struct intel_crtc *crtc, u32 prec_index) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); + struct drm_i915_private *i915 = to_i915(crtc->base.dev); int i, hw_lut_size = ivb_lut_10_size(prec_index); - int lut_size = INTEL_INFO(dev_priv)->display.color.gamma_lut_size; + int lut_size = INTEL_INFO(i915)->display.color.gamma_lut_size; enum pipe pipe = crtc->pipe; struct drm_property_blob *blob; struct drm_color_lut *lut; - drm_WARN_ON(&dev_priv->drm, lut_size != hw_lut_size); + drm_WARN_ON(&i915->drm, lut_size != hw_lut_size); - blob = drm_property_create_blob(&dev_priv->drm, - sizeof(struct drm_color_lut) * lut_size, + blob = drm_property_create_blob(&i915->drm, + sizeof(lut[0]) * lut_size, NULL); if (IS_ERR(blob)) return NULL; lut = blob->data; - intel_de_write_fw(dev_priv, PREC_PAL_INDEX(pipe), + intel_de_write_fw(i915, PREC_PAL_INDEX(pipe), prec_index | PAL_PREC_AUTO_INCREMENT); for (i = 0; i < lut_size; i++) { - u32 val = intel_de_read_fw(dev_priv, PREC_PAL_DATA(pipe)); + u32 val = intel_de_read_fw(i915, PREC_PAL_DATA(pipe)); ilk_lut_10_pack(&lut[i], val); } - intel_de_write_fw(dev_priv, PREC_PAL_INDEX(pipe), 0); + intel_de_write_fw(i915, PREC_PAL_INDEX(pipe), 0); return blob; } @@ -2131,31 +2205,31 @@ static void glk_read_luts(struct intel_crtc_state *crtc_state) static struct drm_property_blob * icl_read_lut_multi_segment(struct intel_crtc *crtc) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - int i, lut_size = INTEL_INFO(dev_priv)->display.color.gamma_lut_size; + struct drm_i915_private *i915 = to_i915(crtc->base.dev); + int i, lut_size = INTEL_INFO(i915)->display.color.gamma_lut_size; enum pipe pipe = crtc->pipe; struct drm_property_blob *blob; struct drm_color_lut *lut; - blob = drm_property_create_blob(&dev_priv->drm, - sizeof(struct drm_color_lut) * lut_size, + blob = drm_property_create_blob(&i915->drm, + sizeof(lut[0]) * lut_size, NULL); if (IS_ERR(blob)) return NULL; lut = blob->data; - intel_de_write_fw(dev_priv, PREC_PAL_MULTI_SEG_INDEX(pipe), + intel_de_write_fw(i915, PREC_PAL_MULTI_SEG_INDEX(pipe), PAL_PREC_AUTO_INCREMENT); for (i = 0; i < 9; i++) { - u32 ldw = intel_de_read_fw(dev_priv, PREC_PAL_MULTI_SEG_DATA(pipe)); - u32 udw = intel_de_read_fw(dev_priv, PREC_PAL_MULTI_SEG_DATA(pipe)); + u32 ldw = intel_de_read_fw(i915, PREC_PAL_MULTI_SEG_DATA(pipe)); + u32 udw = intel_de_read_fw(i915, PREC_PAL_MULTI_SEG_DATA(pipe)); - icl_lut_multi_seg_pack(&lut[i], ldw, udw); + ilk_lut_12p4_pack(&lut[i], ldw, udw); } - intel_de_write_fw(dev_priv, PREC_PAL_MULTI_SEG_INDEX(pipe), 0); + intel_de_write_fw(i915, PREC_PAL_MULTI_SEG_INDEX(pipe), 0); /* * FIXME readouts from PAL_PREC_DATA register aren't giving @@ -2268,15 +2342,15 @@ static const struct intel_color_funcs ilk_color_funcs = { void intel_color_crtc_init(struct intel_crtc *crtc) { - struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - bool has_ctm = INTEL_INFO(dev_priv)->display.color.degamma_lut_size != 0; + struct drm_i915_private *i915 = to_i915(crtc->base.dev); + bool has_ctm = INTEL_INFO(i915)->display.color.degamma_lut_size != 0; drm_mode_crtc_set_gamma_size(&crtc->base, 256); drm_crtc_enable_color_mgmt(&crtc->base, - INTEL_INFO(dev_priv)->display.color.degamma_lut_size, + INTEL_INFO(i915)->display.color.degamma_lut_size, has_ctm, - INTEL_INFO(dev_priv)->display.color.gamma_lut_size); + INTEL_INFO(i915)->display.color.gamma_lut_size); } int intel_color_init(struct drm_i915_private *i915) diff --git a/drivers/gpu/drm/i915/display/intel_combo_phy.c b/drivers/gpu/drm/i915/display/intel_combo_phy.c index 71d7aece1dc6..8b870b2dd4f9 100644 --- a/drivers/gpu/drm/i915/display/intel_combo_phy.c +++ b/drivers/gpu/drm/i915/display/intel_combo_phy.c @@ -3,6 +3,7 @@ * Copyright © 2018 Intel Corporation */ +#include "i915_reg.h" #include "intel_combo_phy.h" #include "intel_combo_phy_regs.h" #include "intel_de.h" diff --git a/drivers/gpu/drm/i915/display/intel_connector.h b/drivers/gpu/drm/i915/display/intel_connector.h index 7d7b588d2286..9d2bc261b204 100644 --- a/drivers/gpu/drm/i915/display/intel_connector.h +++ b/drivers/gpu/drm/i915/display/intel_connector.h @@ -6,7 +6,7 @@ #ifndef __INTEL_CONNECTOR_H__ #define __INTEL_CONNECTOR_H__ -#include "intel_display.h" +#include <linux/types.h> struct drm_connector; struct edid; diff --git a/drivers/gpu/drm/i915/display/intel_crt.c b/drivers/gpu/drm/i915/display/intel_crt.c index 94d0a5e1dd03..797ad9489f7e 100644 --- a/drivers/gpu/drm/i915/display/intel_crt.c +++ b/drivers/gpu/drm/i915/display/intel_crt.c @@ -34,6 +34,8 @@ #include <drm/drm_probe_helper.h> #include "i915_drv.h" +#include "i915_irq.h" +#include "i915_reg.h" #include "intel_connector.h" #include "intel_crt.h" #include "intel_crtc.h" diff --git a/drivers/gpu/drm/i915/display/intel_cursor.c b/drivers/gpu/drm/i915/display/intel_cursor.c index 96422c98656a..d190fa0d393b 100644 --- a/drivers/gpu/drm/i915/display/intel_cursor.c +++ b/drivers/gpu/drm/i915/display/intel_cursor.c @@ -10,12 +10,13 @@ #include <drm/drm_damage_helper.h> #include <drm/drm_fourcc.h> +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_atomic_plane.h" #include "intel_cursor.h" #include "intel_de.h" -#include "intel_display_types.h" #include "intel_display.h" +#include "intel_display_types.h" #include "intel_fb.h" #include "intel_fb_pin.h" #include "intel_frontbuffer.h" diff --git a/drivers/gpu/drm/i915/display/intel_ddi.c b/drivers/gpu/drm/i915/display/intel_ddi.c index e95bde5cf060..0f1ec2a98cc8 100644 --- a/drivers/gpu/drm/i915/display/intel_ddi.c +++ b/drivers/gpu/drm/i915/display/intel_ddi.c @@ -31,6 +31,7 @@ #include <drm/drm_privacy_screen_consumer.h> #include "i915_drv.h" +#include "i915_reg.h" #include "intel_audio.h" #include "intel_audio_regs.h" #include "intel_backlight.h" @@ -56,6 +57,7 @@ #include "intel_hdcp.h" #include "intel_hdmi.h" #include "intel_hotplug.h" +#include "intel_hti.h" #include "intel_lspcon.h" #include "intel_mg_phy_regs.h" #include "intel_pps.h" @@ -846,22 +848,65 @@ bool intel_ddi_get_hw_state(struct intel_encoder *encoder, } static enum intel_display_power_domain -intel_ddi_main_link_aux_domain(struct intel_digital_port *dig_port) +intel_ddi_main_link_aux_domain(struct intel_digital_port *dig_port, + const struct intel_crtc_state *crtc_state) { - /* ICL+ HW requires corresponding AUX IOs to be powered up for PSR with + struct drm_i915_private *i915 = to_i915(dig_port->base.base.dev); + enum phy phy = intel_port_to_phy(i915, dig_port->base.port); + + /* + * ICL+ HW requires corresponding AUX IOs to be powered up for PSR with * DC states enabled at the same time, while for driver initiated AUX * transfers we need the same AUX IOs to be powered but with DC states - * disabled. Accordingly use the AUX power domain here which leaves DC - * states enabled. - * However, for non-A AUX ports the corresponding non-EDP transcoders - * would have already enabled power well 2 and DC_OFF. This means we can - * acquire a wider POWER_DOMAIN_AUX_{B,C,D,F} reference instead of a - * specific AUX_IO reference without powering up any extra wells. - * Note that PSR is enabled only on Port A even though this function - * returns the correct domain for other ports too. + * disabled. Accordingly use the AUX_IO_<port> power domain here which + * leaves DC states enabled. + * + * Before MTL TypeC PHYs (in all TypeC modes and both DP/HDMI) also require + * AUX IO to be enabled, but all these require DC_OFF to be enabled as + * well, so we can acquire a wider AUX_<port> power domain reference + * instead of a specific AUX_IO_<port> reference without powering up any + * extra wells. */ - return dig_port->aux_ch == AUX_CH_A ? POWER_DOMAIN_AUX_IO_A : - intel_aux_power_domain(dig_port); + if (intel_encoder_can_psr(&dig_port->base)) + return intel_display_power_aux_io_domain(i915, dig_port->aux_ch); + else if (DISPLAY_VER(i915) < 14 && + (intel_crtc_has_dp_encoder(crtc_state) || + intel_phy_is_tc(i915, phy))) + return intel_aux_power_domain(dig_port); + else + return POWER_DOMAIN_INVALID; +} + +static void +main_link_aux_power_domain_get(struct intel_digital_port *dig_port, + const struct intel_crtc_state *crtc_state) +{ + struct drm_i915_private *i915 = to_i915(dig_port->base.base.dev); + enum intel_display_power_domain domain = + intel_ddi_main_link_aux_domain(dig_port, crtc_state); + + drm_WARN_ON(&i915->drm, dig_port->aux_wakeref); + + if (domain == POWER_DOMAIN_INVALID) + return; + + dig_port->aux_wakeref = intel_display_power_get(i915, domain); +} + +static void +main_link_aux_power_domain_put(struct intel_digital_port *dig_port, + const struct intel_crtc_state *crtc_state) +{ + struct drm_i915_private *i915 = to_i915(dig_port->base.base.dev); + enum intel_display_power_domain domain = + intel_ddi_main_link_aux_domain(dig_port, crtc_state); + intel_wakeref_t wf; + + wf = fetch_and_zero(&dig_port->aux_wakeref); + if (!wf) + return; + + intel_display_power_put(i915, domain, wf); } static void intel_ddi_get_power_domains(struct intel_encoder *encoder, @@ -869,7 +914,6 @@ static void intel_ddi_get_power_domains(struct intel_encoder *encoder, { struct drm_i915_private *dev_priv = to_i915(encoder->base.dev); struct intel_digital_port *dig_port; - enum phy phy = intel_port_to_phy(dev_priv, encoder->port); /* * TODO: Add support for MST encoders. Atm, the following should never @@ -888,17 +932,7 @@ static void intel_ddi_get_power_domains(struct intel_encoder *encoder, dig_port->ddi_io_power_domain); } - /* - * AUX power is only needed for (e)DP mode, and for HDMI mode on TC - * ports. - */ - if (intel_crtc_has_dp_encoder(crtc_state) || - intel_phy_is_tc(dev_priv, phy)) { - drm_WARN_ON(&dev_priv->drm, dig_port->aux_wakeref); - dig_port->aux_wakeref = - intel_display_power_get(dev_priv, - intel_ddi_main_link_aux_domain(dig_port)); - } + main_link_aux_power_domain_get(dig_port, crtc_state); } void intel_ddi_enable_pipe_clock(struct intel_encoder *encoder, @@ -2737,10 +2771,7 @@ static void intel_ddi_post_disable(struct intel_atomic_state *state, intel_ddi_post_disable_dp(state, encoder, old_crtc_state, old_conn_state); - if (intel_crtc_has_dp_encoder(old_crtc_state) || is_tc_port) - intel_display_power_put(dev_priv, - intel_ddi_main_link_aux_domain(dig_port), - fetch_and_zero(&dig_port->aux_wakeref)); + main_link_aux_power_domain_put(dig_port, old_crtc_state); if (is_tc_port) intel_tc_port_put_link(dig_port); @@ -3061,12 +3092,7 @@ intel_ddi_pre_pll_enable(struct intel_atomic_state *state, if (is_tc_port) intel_tc_port_get_link(dig_port, crtc_state->lane_count); - if (intel_crtc_has_dp_encoder(crtc_state) || is_tc_port) { - drm_WARN_ON(&dev_priv->drm, dig_port->aux_wakeref); - dig_port->aux_wakeref = - intel_display_power_get(dev_priv, - intel_ddi_main_link_aux_domain(dig_port)); - } + main_link_aux_power_domain_get(dig_port, crtc_state); if (is_tc_port && !intel_tc_port_in_tbt_alt_mode(dig_port)) /* @@ -4113,12 +4139,6 @@ intel_ddi_max_lanes(struct intel_digital_port *dig_port) return max_lanes; } -static bool hti_uses_phy(struct drm_i915_private *i915, enum phy phy) -{ - return i915->hti_state & HDPORT_ENABLED && - i915->hti_state & HDPORT_DDI_USED(phy); -} - static enum hpd_pin xelpd_hpd_pin(struct drm_i915_private *dev_priv, enum port port) { @@ -4247,7 +4267,7 @@ void intel_ddi_init(struct drm_i915_private *dev_priv, enum port port) * driver. In that case we should skip initializing the corresponding * outputs. */ - if (hti_uses_phy(dev_priv, phy)) { + if (intel_hti_uses_phy(dev_priv, phy)) { drm_dbg_kms(&dev_priv->drm, "PORT %c / PHY %c reserved by HTI\n", port_name(port), phy_name(phy)); return; diff --git a/drivers/gpu/drm/i915/display/intel_display.c b/drivers/gpu/drm/i915/display/intel_display.c index b9393f9fc764..b3e23708d194 100644 --- a/drivers/gpu/drm/i915/display/intel_display.c +++ b/drivers/gpu/drm/i915/display/intel_display.c @@ -76,6 +76,7 @@ #include "g4x_hdmi.h" #include "hsw_ips.h" #include "i915_drv.h" +#include "i915_reg.h" #include "i915_utils.h" #include "icl_dsi.h" #include "intel_acpi.h" @@ -90,6 +91,7 @@ #include "intel_display_types.h" #include "intel_dmc.h" #include "intel_dp_link_training.h" +#include "intel_dpio_phy.h" #include "intel_dpt.h" #include "intel_dsb.h" #include "intel_fbc.h" @@ -99,6 +101,7 @@ #include "intel_frontbuffer.h" #include "intel_hdcp.h" #include "intel_hotplug.h" +#include "intel_hti.h" #include "intel_modeset_verify.h" #include "intel_modeset_setup.h" #include "intel_overlay.h" @@ -888,7 +891,7 @@ static bool gpu_reset_clobbers_display(struct drm_i915_private *dev_priv) void intel_display_prepare_reset(struct drm_i915_private *dev_priv) { - struct drm_modeset_acquire_ctx *ctx = &dev_priv->reset_ctx; + struct drm_modeset_acquire_ctx *ctx = &dev_priv->display.restore.reset_ctx; struct drm_atomic_state *state; int ret; @@ -944,13 +947,13 @@ void intel_display_prepare_reset(struct drm_i915_private *dev_priv) return; } - dev_priv->modeset_restore_state = state; + dev_priv->display.restore.modeset_state = state; state->acquire_ctx = ctx; } void intel_display_finish_reset(struct drm_i915_private *i915) { - struct drm_modeset_acquire_ctx *ctx = &i915->reset_ctx; + struct drm_modeset_acquire_ctx *ctx = &i915->display.restore.reset_ctx; struct drm_atomic_state *state; int ret; @@ -961,7 +964,7 @@ void intel_display_finish_reset(struct drm_i915_private *i915) if (!test_bit(I915_RESET_MODESET, &to_gt(i915)->reset.flags)) return; - state = fetch_and_zero(&i915->modeset_restore_state); + state = fetch_and_zero(&i915->display.restore.modeset_state); if (!state) goto unlock; @@ -2441,7 +2444,7 @@ int intel_display_suspend(struct drm_device *dev) drm_err(&dev_priv->drm, "Suspending crtc's failed with %i\n", ret); else - dev_priv->modeset_restore_state = state; + dev_priv->display.restore.modeset_state = state; return ret; } @@ -4048,20 +4051,19 @@ static bool hsw_get_pipe_config(struct intel_crtc *crtc, struct intel_crtc_state *pipe_config) { struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); - struct intel_display_power_domain_set power_domain_set = { }; bool active; u32 tmp; - if (!intel_display_power_get_in_set_if_enabled(dev_priv, &power_domain_set, + if (!intel_display_power_get_in_set_if_enabled(dev_priv, &crtc->hw_readout_power_domains, POWER_DOMAIN_PIPE(crtc->pipe))) return false; pipe_config->shared_dpll = NULL; - active = hsw_get_transcoder_state(crtc, pipe_config, &power_domain_set); + active = hsw_get_transcoder_state(crtc, pipe_config, &crtc->hw_readout_power_domains); if ((IS_GEMINILAKE(dev_priv) || IS_BROXTON(dev_priv)) && - bxt_get_dsi_transcoder_state(crtc, pipe_config, &power_domain_set)) { + bxt_get_dsi_transcoder_state(crtc, pipe_config, &crtc->hw_readout_power_domains)) { drm_WARN_ON(&dev_priv->drm, active); active = true; } @@ -4120,7 +4122,7 @@ static bool hsw_get_pipe_config(struct intel_crtc *crtc, pipe_config->ips_linetime = REG_FIELD_GET(HSW_IPS_LINETIME_MASK, tmp); - if (intel_display_power_get_in_set_if_enabled(dev_priv, &power_domain_set, + if (intel_display_power_get_in_set_if_enabled(dev_priv, &crtc->hw_readout_power_domains, POWER_DOMAIN_PIPE_PANEL_FITTER(crtc->pipe))) { if (DISPLAY_VER(dev_priv) >= 9) skl_get_pfit_config(pipe_config); @@ -4151,7 +4153,7 @@ static bool hsw_get_pipe_config(struct intel_crtc *crtc, } out: - intel_display_power_put_all_in_set(dev_priv, &power_domain_set); + intel_display_power_put_all_in_set(dev_priv, &crtc->hw_readout_power_domains); return active; } @@ -5930,7 +5932,7 @@ int intel_modeset_all_pipes(struct intel_atomic_state *state, return PTR_ERR(crtc_state); if (!crtc_state->hw.active || - drm_atomic_crtc_needs_modeset(&crtc_state->uapi)) + intel_crtc_needs_modeset(crtc_state)) continue; drm_dbg_kms(&dev_priv->drm, "[CRTC:%d:%s] Full modeset due to %s\n", @@ -8569,7 +8571,7 @@ static void intel_mode_config_init(struct drm_i915_private *i915) struct drm_mode_config *mode_config = &i915->drm.mode_config; drm_mode_config_init(&i915->drm); - INIT_LIST_HEAD(&i915->global_obj_list); + INIT_LIST_HEAD(&i915->display.global.obj_list); mode_config->min_width = 0; mode_config->min_height = 0; @@ -8734,12 +8736,7 @@ int intel_modeset_init_nogem(struct drm_i915_private *i915) if (i915->display.cdclk.max_cdclk_freq == 0) intel_update_max_cdclk(i915); - /* - * If the platform has HTI, we need to find out whether it has reserved - * any display resources before we create our display outputs. - */ - if (INTEL_INFO(i915)->display.has_hti) - i915->hti_state = intel_de_read(i915, HDPORT_STATE); + intel_hti_init(i915); /* Just disable it once at startup */ intel_vga_disable(i915); @@ -8902,14 +8899,14 @@ void i830_disable_pipe(struct drm_i915_private *dev_priv, enum pipe pipe) void intel_display_resume(struct drm_device *dev) { struct drm_i915_private *i915 = to_i915(dev); - struct drm_atomic_state *state = i915->modeset_restore_state; + struct drm_atomic_state *state = i915->display.restore.modeset_state; struct drm_modeset_acquire_ctx ctx; int ret; if (!HAS_DISPLAY(i915)) return; - i915->modeset_restore_state = NULL; + i915->display.restore.modeset_state = NULL; if (state) state->acquire_ctx = &ctx; diff --git a/drivers/gpu/drm/i915/display/intel_display.h b/drivers/gpu/drm/i915/display/intel_display.h index c803330a276d..714030136b7f 100644 --- a/drivers/gpu/drm/i915/display/intel_display.h +++ b/drivers/gpu/drm/i915/display/intel_display.h @@ -53,6 +53,7 @@ struct intel_digital_port; struct intel_dp; struct intel_encoder; struct intel_initial_plane_config; +struct intel_link_m_n; struct intel_load_detect_pipe; struct intel_plane; struct intel_plane_state; @@ -61,24 +62,6 @@ struct intel_remapped_info; struct intel_rotation_info; struct pci_dev; -enum i915_gpio { - GPIOA, - GPIOB, - GPIOC, - GPIOD, - GPIOE, - GPIOF, - GPIOG, - GPIOH, - __GPIOI_UNUSED, - GPIOJ, - GPIOK, - GPIOL, - GPIOM, - GPION, - GPIOO, -}; - /* * Keep the pipe enum values fixed: the code assumes that PIPE_A=0, the * rest have consecutive values and match the enum values of transcoders @@ -279,17 +262,6 @@ enum tc_port_mode { TC_PORT_LEGACY, }; -enum dpio_channel { - DPIO_CH0, - DPIO_CH1 -}; - -enum dpio_phy { - DPIO_PHY0, - DPIO_PHY1, - DPIO_PHY2, -}; - enum aux_ch { AUX_CH_A, AUX_CH_B, @@ -316,15 +288,6 @@ enum aux_ch { #define aux_ch_name(a) ((a) + 'A') -/* Used by dp and fdi links */ -struct intel_link_m_n { - u32 tu; - u32 data_m; - u32 data_n; - u32 link_m; - u32 link_n; -}; - enum phy { PHY_NONE = -1, diff --git a/drivers/gpu/drm/i915/display/intel_display_core.h b/drivers/gpu/drm/i915/display/intel_display_core.h index 337d8e08ba43..57ddce3ba02b 100644 --- a/drivers/gpu/drm/i915/display/intel_display_core.h +++ b/drivers/gpu/drm/i915/display/intel_display_core.h @@ -14,6 +14,7 @@ #include <linux/workqueue.h> #include <drm/drm_connector.h> +#include <drm/drm_modeset_lock.h> #include "intel_cdclk.h" #include "intel_display.h" @@ -345,6 +346,10 @@ struct intel_display { } fdi; struct { + struct list_head obj_list; + } global; + + struct { /* * Base address of where the gmbus and gpio blocks are located * (either on PCH or on SoC for platforms without PCH). @@ -371,6 +376,16 @@ struct intel_display { } hdcp; struct { + /* + * HTI (aka HDPORT) state read during initial hw readout. Most + * platforms don't have HTI, so this will just stay 0. Those + * that do will use this later to figure out which PLLs and PHYs + * are unavailable for driver usage. + */ + u32 state; + } hti; + + struct { struct i915_power_domains domains; /* Shadow for DISPLAY_PHY_CONTROL which can't be safely read */ @@ -397,6 +412,12 @@ struct intel_display { } quirks; struct { + /* restore state for suspend/resume and display reset */ + struct drm_atomic_state *modeset_state; + struct drm_modeset_acquire_ctx reset_ctx; + } restore; + + struct { enum { I915_SAGV_UNKNOWN = 0, I915_SAGV_DISABLED, diff --git a/drivers/gpu/drm/i915/display/intel_display_debugfs.c b/drivers/gpu/drm/i915/display/intel_display_debugfs.c index cfc056a05bbf..7bcd90384a46 100644 --- a/drivers/gpu/drm/i915/display/intel_display_debugfs.c +++ b/drivers/gpu/drm/i915/display/intel_display_debugfs.c @@ -9,6 +9,8 @@ #include <drm/drm_fourcc.h> #include "i915_debugfs.h" +#include "i915_irq.h" +#include "i915_reg.h" #include "intel_de.h" #include "intel_display_debugfs.h" #include "intel_display_power.h" diff --git a/drivers/gpu/drm/i915/display/intel_display_power.c b/drivers/gpu/drm/i915/display/intel_display_power.c index 4c1de91e56ff..3adba64937de 100644 --- a/drivers/gpu/drm/i915/display/intel_display_power.c +++ b/drivers/gpu/drm/i915/display/intel_display_power.c @@ -129,6 +129,18 @@ intel_display_power_domain_str(enum intel_display_power_domain domain) return "AUDIO_MMIO"; case POWER_DOMAIN_AUDIO_PLAYBACK: return "AUDIO_PLAYBACK"; + case POWER_DOMAIN_AUX_IO_A: + return "AUX_IO_A"; + case POWER_DOMAIN_AUX_IO_B: + return "AUX_IO_B"; + case POWER_DOMAIN_AUX_IO_C: + return "AUX_IO_C"; + case POWER_DOMAIN_AUX_IO_D: + return "AUX_IO_D"; + case POWER_DOMAIN_AUX_IO_E: + return "AUX_IO_E"; + case POWER_DOMAIN_AUX_IO_F: + return "AUX_IO_F"; case POWER_DOMAIN_AUX_A: return "AUX_A"; case POWER_DOMAIN_AUX_B: @@ -153,8 +165,6 @@ intel_display_power_domain_str(enum intel_display_power_domain domain) return "AUX_USBC5"; case POWER_DOMAIN_AUX_USBC6: return "AUX_USBC6"; - case POWER_DOMAIN_AUX_IO_A: - return "AUX_IO_A"; case POWER_DOMAIN_AUX_TBT1: return "AUX_TBT1"; case POWER_DOMAIN_AUX_TBT2: @@ -2289,6 +2299,7 @@ struct intel_ddi_port_domains { enum intel_display_power_domain ddi_lanes; enum intel_display_power_domain ddi_io; + enum intel_display_power_domain aux_io; enum intel_display_power_domain aux_legacy_usbc; enum intel_display_power_domain aux_tbt; }; @@ -2303,6 +2314,7 @@ i9xx_port_domains[] = { .ddi_lanes = POWER_DOMAIN_PORT_DDI_LANES_A, .ddi_io = POWER_DOMAIN_PORT_DDI_IO_A, + .aux_io = POWER_DOMAIN_AUX_IO_A, .aux_legacy_usbc = POWER_DOMAIN_AUX_A, .aux_tbt = POWER_DOMAIN_INVALID, }, @@ -2318,6 +2330,7 @@ d11_port_domains[] = { .ddi_lanes = POWER_DOMAIN_PORT_DDI_LANES_A, .ddi_io = POWER_DOMAIN_PORT_DDI_IO_A, + .aux_io = POWER_DOMAIN_AUX_IO_A, .aux_legacy_usbc = POWER_DOMAIN_AUX_A, .aux_tbt = POWER_DOMAIN_INVALID, }, { @@ -2328,6 +2341,7 @@ d11_port_domains[] = { .ddi_lanes = POWER_DOMAIN_PORT_DDI_LANES_C, .ddi_io = POWER_DOMAIN_PORT_DDI_IO_C, + .aux_io = POWER_DOMAIN_AUX_IO_C, .aux_legacy_usbc = POWER_DOMAIN_AUX_C, .aux_tbt = POWER_DOMAIN_AUX_TBT1, }, @@ -2343,6 +2357,7 @@ d12_port_domains[] = { .ddi_lanes = POWER_DOMAIN_PORT_DDI_LANES_A, .ddi_io = POWER_DOMAIN_PORT_DDI_IO_A, + .aux_io = POWER_DOMAIN_AUX_IO_A, .aux_legacy_usbc = POWER_DOMAIN_AUX_A, .aux_tbt = POWER_DOMAIN_INVALID, }, { @@ -2353,6 +2368,7 @@ d12_port_domains[] = { .ddi_lanes = POWER_DOMAIN_PORT_DDI_LANES_TC1, .ddi_io = POWER_DOMAIN_PORT_DDI_IO_TC1, + .aux_io = POWER_DOMAIN_INVALID, .aux_legacy_usbc = POWER_DOMAIN_AUX_USBC1, .aux_tbt = POWER_DOMAIN_AUX_TBT1, }, @@ -2368,6 +2384,7 @@ d13_port_domains[] = { .ddi_lanes = POWER_DOMAIN_PORT_DDI_LANES_A, .ddi_io = POWER_DOMAIN_PORT_DDI_IO_A, + .aux_io = POWER_DOMAIN_AUX_IO_A, .aux_legacy_usbc = POWER_DOMAIN_AUX_A, .aux_tbt = POWER_DOMAIN_INVALID, }, { @@ -2378,6 +2395,7 @@ d13_port_domains[] = { .ddi_lanes = POWER_DOMAIN_PORT_DDI_LANES_TC1, .ddi_io = POWER_DOMAIN_PORT_DDI_IO_TC1, + .aux_io = POWER_DOMAIN_INVALID, .aux_legacy_usbc = POWER_DOMAIN_AUX_USBC1, .aux_tbt = POWER_DOMAIN_AUX_TBT1, }, { @@ -2388,6 +2406,7 @@ d13_port_domains[] = { .ddi_lanes = POWER_DOMAIN_PORT_DDI_LANES_D, .ddi_io = POWER_DOMAIN_PORT_DDI_IO_D, + .aux_io = POWER_DOMAIN_AUX_IO_D, .aux_legacy_usbc = POWER_DOMAIN_AUX_D, .aux_tbt = POWER_DOMAIN_INVALID, }, @@ -2433,7 +2452,7 @@ intel_display_power_ddi_io_domain(struct drm_i915_private *i915, enum port port) { const struct intel_ddi_port_domains *domains = intel_port_domains_for_port(i915, port); - if (drm_WARN_ON(&i915->drm, !domains) || domains->ddi_io == POWER_DOMAIN_INVALID) + if (drm_WARN_ON(&i915->drm, !domains || domains->ddi_io == POWER_DOMAIN_INVALID)) return POWER_DOMAIN_PORT_DDI_IO_A; return domains->ddi_io + (int)(port - domains->port_start); @@ -2444,7 +2463,7 @@ intel_display_power_ddi_lanes_domain(struct drm_i915_private *i915, enum port po { const struct intel_ddi_port_domains *domains = intel_port_domains_for_port(i915, port); - if (drm_WARN_ON(&i915->drm, !domains) || domains->ddi_lanes == POWER_DOMAIN_INVALID) + if (drm_WARN_ON(&i915->drm, !domains || domains->ddi_lanes == POWER_DOMAIN_INVALID)) return POWER_DOMAIN_PORT_DDI_LANES_A; return domains->ddi_lanes + (int)(port - domains->port_start); @@ -2466,11 +2485,22 @@ intel_port_domains_for_aux_ch(struct drm_i915_private *i915, enum aux_ch aux_ch) } enum intel_display_power_domain +intel_display_power_aux_io_domain(struct drm_i915_private *i915, enum aux_ch aux_ch) +{ + const struct intel_ddi_port_domains *domains = intel_port_domains_for_aux_ch(i915, aux_ch); + + if (drm_WARN_ON(&i915->drm, !domains || domains->aux_io == POWER_DOMAIN_INVALID)) + return POWER_DOMAIN_AUX_IO_A; + + return domains->aux_io + (int)(aux_ch - domains->aux_ch_start); +} + +enum intel_display_power_domain intel_display_power_legacy_aux_domain(struct drm_i915_private *i915, enum aux_ch aux_ch) { const struct intel_ddi_port_domains *domains = intel_port_domains_for_aux_ch(i915, aux_ch); - if (drm_WARN_ON(&i915->drm, !domains) || domains->aux_legacy_usbc == POWER_DOMAIN_INVALID) + if (drm_WARN_ON(&i915->drm, !domains || domains->aux_legacy_usbc == POWER_DOMAIN_INVALID)) return POWER_DOMAIN_AUX_A; return domains->aux_legacy_usbc + (int)(aux_ch - domains->aux_ch_start); @@ -2481,7 +2511,7 @@ intel_display_power_tbt_aux_domain(struct drm_i915_private *i915, enum aux_ch au { const struct intel_ddi_port_domains *domains = intel_port_domains_for_aux_ch(i915, aux_ch); - if (drm_WARN_ON(&i915->drm, !domains) || domains->aux_tbt == POWER_DOMAIN_INVALID) + if (drm_WARN_ON(&i915->drm, !domains || domains->aux_tbt == POWER_DOMAIN_INVALID)) return POWER_DOMAIN_AUX_TBT1; return domains->aux_tbt + (int)(aux_ch - domains->aux_ch_start); diff --git a/drivers/gpu/drm/i915/display/intel_display_power.h b/drivers/gpu/drm/i915/display/intel_display_power.h index 7136ea3f233e..2154d900b1aa 100644 --- a/drivers/gpu/drm/i915/display/intel_display_power.h +++ b/drivers/gpu/drm/i915/display/intel_display_power.h @@ -6,11 +6,12 @@ #ifndef __INTEL_DISPLAY_POWER_H__ #define __INTEL_DISPLAY_POWER_H__ -#include "intel_runtime_pm.h" +#include "intel_wakeref.h" enum aux_ch; enum dpio_channel; enum dpio_phy; +enum i915_drm_suspend_mode; enum port; struct drm_i915_private; struct i915_power_well; @@ -76,6 +77,14 @@ enum intel_display_power_domain { POWER_DOMAIN_VGA, POWER_DOMAIN_AUDIO_MMIO, POWER_DOMAIN_AUDIO_PLAYBACK, + + POWER_DOMAIN_AUX_IO_A, + POWER_DOMAIN_AUX_IO_B, + POWER_DOMAIN_AUX_IO_C, + POWER_DOMAIN_AUX_IO_D, + POWER_DOMAIN_AUX_IO_E, + POWER_DOMAIN_AUX_IO_F, + POWER_DOMAIN_AUX_A, POWER_DOMAIN_AUX_B, POWER_DOMAIN_AUX_C, @@ -90,8 +99,6 @@ enum intel_display_power_domain { POWER_DOMAIN_AUX_USBC5, POWER_DOMAIN_AUX_USBC6, - POWER_DOMAIN_AUX_IO_A, - POWER_DOMAIN_AUX_TBT1, POWER_DOMAIN_AUX_TBT2, POWER_DOMAIN_AUX_TBT3, @@ -249,6 +256,8 @@ intel_display_power_ddi_lanes_domain(struct drm_i915_private *i915, enum port po enum intel_display_power_domain intel_display_power_ddi_io_domain(struct drm_i915_private *i915, enum port port); enum intel_display_power_domain +intel_display_power_aux_io_domain(struct drm_i915_private *i915, enum aux_ch aux_ch); +enum intel_display_power_domain intel_display_power_legacy_aux_domain(struct drm_i915_private *i915, enum aux_ch aux_ch); enum intel_display_power_domain intel_display_power_tbt_aux_domain(struct drm_i915_private *i915, enum aux_ch aux_ch); diff --git a/drivers/gpu/drm/i915/display/intel_display_power_map.c b/drivers/gpu/drm/i915/display/intel_display_power_map.c index dc04afc6cc8f..f5d66ca85b19 100644 --- a/drivers/gpu/drm/i915/display/intel_display_power_map.c +++ b/drivers/gpu/drm/i915/display/intel_display_power_map.c @@ -170,6 +170,8 @@ I915_DECL_PW_DOMAINS(vlv_pwdoms_display, POWER_DOMAIN_VGA, POWER_DOMAIN_AUDIO_MMIO, POWER_DOMAIN_AUDIO_PLAYBACK, + POWER_DOMAIN_AUX_IO_B, + POWER_DOMAIN_AUX_IO_C, POWER_DOMAIN_AUX_B, POWER_DOMAIN_AUX_C, POWER_DOMAIN_GMBUS, @@ -179,6 +181,8 @@ I915_DECL_PW_DOMAINS(vlv_pwdoms_dpio_cmn_bc, POWER_DOMAIN_PORT_DDI_LANES_B, POWER_DOMAIN_PORT_DDI_LANES_C, POWER_DOMAIN_PORT_CRT, + POWER_DOMAIN_AUX_IO_B, + POWER_DOMAIN_AUX_IO_C, POWER_DOMAIN_AUX_B, POWER_DOMAIN_AUX_C, POWER_DOMAIN_INIT); @@ -186,6 +190,8 @@ I915_DECL_PW_DOMAINS(vlv_pwdoms_dpio_cmn_bc, I915_DECL_PW_DOMAINS(vlv_pwdoms_dpio_tx_bc_lanes, POWER_DOMAIN_PORT_DDI_LANES_B, POWER_DOMAIN_PORT_DDI_LANES_C, + POWER_DOMAIN_AUX_IO_B, + POWER_DOMAIN_AUX_IO_C, POWER_DOMAIN_AUX_B, POWER_DOMAIN_AUX_C, POWER_DOMAIN_INIT); @@ -243,6 +249,9 @@ I915_DECL_PW_DOMAINS(chv_pwdoms_display, POWER_DOMAIN_VGA, POWER_DOMAIN_AUDIO_MMIO, POWER_DOMAIN_AUDIO_PLAYBACK, + POWER_DOMAIN_AUX_IO_B, + POWER_DOMAIN_AUX_IO_C, + POWER_DOMAIN_AUX_IO_D, POWER_DOMAIN_AUX_B, POWER_DOMAIN_AUX_C, POWER_DOMAIN_AUX_D, @@ -252,12 +261,15 @@ I915_DECL_PW_DOMAINS(chv_pwdoms_display, I915_DECL_PW_DOMAINS(chv_pwdoms_dpio_cmn_bc, POWER_DOMAIN_PORT_DDI_LANES_B, POWER_DOMAIN_PORT_DDI_LANES_C, + POWER_DOMAIN_AUX_IO_B, + POWER_DOMAIN_AUX_IO_C, POWER_DOMAIN_AUX_B, POWER_DOMAIN_AUX_C, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(chv_pwdoms_dpio_cmn_d, POWER_DOMAIN_PORT_DDI_LANES_D, + POWER_DOMAIN_AUX_IO_D, POWER_DOMAIN_AUX_D, POWER_DOMAIN_INIT); @@ -305,6 +317,9 @@ static const struct i915_power_well_desc_list chv_power_wells[] = { POWER_DOMAIN_VGA, \ POWER_DOMAIN_AUDIO_MMIO, \ POWER_DOMAIN_AUDIO_PLAYBACK, \ + POWER_DOMAIN_AUX_IO_B, \ + POWER_DOMAIN_AUX_IO_C, \ + POWER_DOMAIN_AUX_IO_D, \ POWER_DOMAIN_AUX_B, \ POWER_DOMAIN_AUX_C, \ POWER_DOMAIN_AUX_D @@ -318,6 +333,7 @@ I915_DECL_PW_DOMAINS(skl_pwdoms_dc_off, POWER_DOMAIN_AUX_A, POWER_DOMAIN_MODESET, POWER_DOMAIN_GT_IRQ, + POWER_DOMAIN_DC_OFF, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(skl_pwdoms_ddi_io_a_e, @@ -407,6 +423,8 @@ static const struct i915_power_well_desc_list skl_power_wells[] = { POWER_DOMAIN_VGA, \ POWER_DOMAIN_AUDIO_MMIO, \ POWER_DOMAIN_AUDIO_PLAYBACK, \ + POWER_DOMAIN_AUX_IO_B, \ + POWER_DOMAIN_AUX_IO_C, \ POWER_DOMAIN_AUX_B, \ POWER_DOMAIN_AUX_C @@ -420,16 +438,20 @@ I915_DECL_PW_DOMAINS(bxt_pwdoms_dc_off, POWER_DOMAIN_GMBUS, POWER_DOMAIN_MODESET, POWER_DOMAIN_GT_IRQ, + POWER_DOMAIN_DC_OFF, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(bxt_pwdoms_dpio_cmn_a, POWER_DOMAIN_PORT_DDI_LANES_A, + POWER_DOMAIN_AUX_IO_A, POWER_DOMAIN_AUX_A, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(bxt_pwdoms_dpio_cmn_bc, POWER_DOMAIN_PORT_DDI_LANES_B, POWER_DOMAIN_PORT_DDI_LANES_C, + POWER_DOMAIN_AUX_IO_B, + POWER_DOMAIN_AUX_IO_C, POWER_DOMAIN_AUX_B, POWER_DOMAIN_AUX_C, POWER_DOMAIN_INIT); @@ -483,6 +505,8 @@ static const struct i915_power_well_desc_list bxt_power_wells[] = { POWER_DOMAIN_VGA, \ POWER_DOMAIN_AUDIO_MMIO, \ POWER_DOMAIN_AUDIO_PLAYBACK, \ + POWER_DOMAIN_AUX_IO_B, \ + POWER_DOMAIN_AUX_IO_C, \ POWER_DOMAIN_AUX_B, \ POWER_DOMAIN_AUX_C @@ -496,6 +520,7 @@ I915_DECL_PW_DOMAINS(glk_pwdoms_dc_off, POWER_DOMAIN_GMBUS, POWER_DOMAIN_MODESET, POWER_DOMAIN_GT_IRQ, + POWER_DOMAIN_DC_OFF, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(glk_pwdoms_ddi_io_a, POWER_DOMAIN_PORT_DDI_IO_A); @@ -504,29 +529,34 @@ I915_DECL_PW_DOMAINS(glk_pwdoms_ddi_io_c, POWER_DOMAIN_PORT_DDI_IO_C); I915_DECL_PW_DOMAINS(glk_pwdoms_dpio_cmn_a, POWER_DOMAIN_PORT_DDI_LANES_A, + POWER_DOMAIN_AUX_IO_A, POWER_DOMAIN_AUX_A, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(glk_pwdoms_dpio_cmn_b, POWER_DOMAIN_PORT_DDI_LANES_B, + POWER_DOMAIN_AUX_IO_B, POWER_DOMAIN_AUX_B, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(glk_pwdoms_dpio_cmn_c, POWER_DOMAIN_PORT_DDI_LANES_C, + POWER_DOMAIN_AUX_IO_C, POWER_DOMAIN_AUX_C, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(glk_pwdoms_aux_a, - POWER_DOMAIN_AUX_A, POWER_DOMAIN_AUX_IO_A, + POWER_DOMAIN_AUX_A, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(glk_pwdoms_aux_b, + POWER_DOMAIN_AUX_IO_B, POWER_DOMAIN_AUX_B, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(glk_pwdoms_aux_c, + POWER_DOMAIN_AUX_IO_C, POWER_DOMAIN_AUX_C, POWER_DOMAIN_INIT); @@ -617,6 +647,11 @@ I915_DECL_PW_DOMAINS(icl_pwdoms_pw_4, POWER_DOMAIN_VGA, \ POWER_DOMAIN_AUDIO_MMIO, \ POWER_DOMAIN_AUDIO_PLAYBACK, \ + POWER_DOMAIN_AUX_IO_B, \ + POWER_DOMAIN_AUX_IO_C, \ + POWER_DOMAIN_AUX_IO_D, \ + POWER_DOMAIN_AUX_IO_E, \ + POWER_DOMAIN_AUX_IO_F, \ POWER_DOMAIN_AUX_B, \ POWER_DOMAIN_AUX_C, \ POWER_DOMAIN_AUX_D, \ @@ -658,13 +693,23 @@ I915_DECL_PW_DOMAINS(icl_pwdoms_ddi_io_e, POWER_DOMAIN_PORT_DDI_IO_E); I915_DECL_PW_DOMAINS(icl_pwdoms_ddi_io_f, POWER_DOMAIN_PORT_DDI_IO_F); I915_DECL_PW_DOMAINS(icl_pwdoms_aux_a, - POWER_DOMAIN_AUX_A, - POWER_DOMAIN_AUX_IO_A); -I915_DECL_PW_DOMAINS(icl_pwdoms_aux_b, POWER_DOMAIN_AUX_B); -I915_DECL_PW_DOMAINS(icl_pwdoms_aux_c, POWER_DOMAIN_AUX_C); -I915_DECL_PW_DOMAINS(icl_pwdoms_aux_d, POWER_DOMAIN_AUX_D); -I915_DECL_PW_DOMAINS(icl_pwdoms_aux_e, POWER_DOMAIN_AUX_E); -I915_DECL_PW_DOMAINS(icl_pwdoms_aux_f, POWER_DOMAIN_AUX_F); + POWER_DOMAIN_AUX_IO_A, + POWER_DOMAIN_AUX_A); +I915_DECL_PW_DOMAINS(icl_pwdoms_aux_b, + POWER_DOMAIN_AUX_IO_B, + POWER_DOMAIN_AUX_B); +I915_DECL_PW_DOMAINS(icl_pwdoms_aux_c, + POWER_DOMAIN_AUX_IO_C, + POWER_DOMAIN_AUX_C); +I915_DECL_PW_DOMAINS(icl_pwdoms_aux_d, + POWER_DOMAIN_AUX_IO_D, + POWER_DOMAIN_AUX_D); +I915_DECL_PW_DOMAINS(icl_pwdoms_aux_e, + POWER_DOMAIN_AUX_IO_E, + POWER_DOMAIN_AUX_E); +I915_DECL_PW_DOMAINS(icl_pwdoms_aux_f, + POWER_DOMAIN_AUX_IO_F, + POWER_DOMAIN_AUX_F); I915_DECL_PW_DOMAINS(icl_pwdoms_aux_tbt1, POWER_DOMAIN_AUX_TBT1); I915_DECL_PW_DOMAINS(icl_pwdoms_aux_tbt2, POWER_DOMAIN_AUX_TBT2); I915_DECL_PW_DOMAINS(icl_pwdoms_aux_tbt3, POWER_DOMAIN_AUX_TBT3); @@ -816,6 +861,7 @@ I915_DECL_PW_DOMAINS(tgl_pwdoms_dc_off, POWER_DOMAIN_AUX_B, POWER_DOMAIN_AUX_C, POWER_DOMAIN_MODESET, + POWER_DOMAIN_DC_OFF, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(tgl_pwdoms_ddi_io_tc1, POWER_DOMAIN_PORT_DDI_IO_TC1); @@ -1012,6 +1058,7 @@ I915_DECL_PW_DOMAINS(rkl_pwdoms_dc_off, POWER_DOMAIN_AUX_A, POWER_DOMAIN_AUX_B, POWER_DOMAIN_MODESET, + POWER_DOMAIN_DC_OFF, POWER_DOMAIN_INIT); static const struct i915_power_well_desc rkl_power_wells_main[] = { @@ -1094,6 +1141,7 @@ I915_DECL_PW_DOMAINS(dg1_pwdoms_dc_off, POWER_DOMAIN_AUX_A, POWER_DOMAIN_AUX_B, POWER_DOMAIN_MODESET, + POWER_DOMAIN_DC_OFF, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(dg1_pwdoms_pw_2, @@ -1215,6 +1263,9 @@ I915_DECL_PW_DOMAINS(xelpd_pwdoms_pw_a, POWER_DOMAIN_PORT_DDI_LANES_TC4, \ POWER_DOMAIN_VGA, \ POWER_DOMAIN_AUDIO_PLAYBACK, \ + POWER_DOMAIN_AUX_IO_C, \ + POWER_DOMAIN_AUX_IO_D, \ + POWER_DOMAIN_AUX_IO_E, \ POWER_DOMAIN_AUX_C, \ POWER_DOMAIN_AUX_D, \ POWER_DOMAIN_AUX_E, \ @@ -1255,6 +1306,7 @@ I915_DECL_PW_DOMAINS(xelpd_pwdoms_dc_off, POWER_DOMAIN_AUX_A, POWER_DOMAIN_AUX_B, POWER_DOMAIN_MODESET, + POWER_DOMAIN_DC_OFF, POWER_DOMAIN_INIT); static const struct i915_power_well_desc xelpd_power_wells_main[] = { @@ -1376,6 +1428,7 @@ I915_DECL_PW_DOMAINS(xelpdp_pwdoms_dc_off, POWER_DOMAIN_MODESET, POWER_DOMAIN_AUX_A, POWER_DOMAIN_AUX_B, + POWER_DOMAIN_DC_OFF, POWER_DOMAIN_INIT); I915_DECL_PW_DOMAINS(xelpdp_pwdoms_aux_tc1, diff --git a/drivers/gpu/drm/i915/display/intel_display_power_well.h b/drivers/gpu/drm/i915/display/intel_display_power_well.h index e13b521e322a..ba7cb977e7c7 100644 --- a/drivers/gpu/drm/i915/display/intel_display_power_well.h +++ b/drivers/gpu/drm/i915/display/intel_display_power_well.h @@ -7,8 +7,8 @@ #include <linux/types.h> -#include "intel_display.h" #include "intel_display_power.h" +#include "intel_dpio_phy.h" struct drm_i915_private; struct i915_power_well; diff --git a/drivers/gpu/drm/i915/display/intel_display_reg_defs.h b/drivers/gpu/drm/i915/display/intel_display_reg_defs.h new file mode 100644 index 000000000000..02605418ff08 --- /dev/null +++ b/drivers/gpu/drm/i915/display/intel_display_reg_defs.h @@ -0,0 +1,53 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright © 2022 Intel Corporation + */ + +#ifndef __INTEL_DISPLAY_REG_DEFS_H__ +#define __INTEL_DISPLAY_REG_DEFS_H__ + +#include "i915_reg_defs.h" + +#define DISPLAY_MMIO_BASE(dev_priv) (INTEL_INFO(dev_priv)->display.mmio_offset) + +#define VLV_DISPLAY_BASE 0x180000 + +/* + * Named helper wrappers around _PICK_EVEN() and _PICK(). + */ +#define _PIPE(pipe, a, b) _PICK_EVEN(pipe, a, b) +#define _PLANE(plane, a, b) _PICK_EVEN(plane, a, b) +#define _TRANS(tran, a, b) _PICK_EVEN(tran, a, b) +#define _PORT(port, a, b) _PICK_EVEN(port, a, b) +#define _PLL(pll, a, b) _PICK_EVEN(pll, a, b) +#define _PHY(phy, a, b) _PICK_EVEN(phy, a, b) + +#define _MMIO_PIPE(pipe, a, b) _MMIO(_PIPE(pipe, a, b)) +#define _MMIO_PLANE(plane, a, b) _MMIO(_PLANE(plane, a, b)) +#define _MMIO_TRANS(tran, a, b) _MMIO(_TRANS(tran, a, b)) +#define _MMIO_PORT(port, a, b) _MMIO(_PORT(port, a, b)) +#define _MMIO_PLL(pll, a, b) _MMIO(_PLL(pll, a, b)) +#define _MMIO_PHY(phy, a, b) _MMIO(_PHY(phy, a, b)) + +#define _PHY3(phy, ...) _PICK(phy, __VA_ARGS__) + +#define _MMIO_PIPE3(pipe, a, b, c) _MMIO(_PICK(pipe, a, b, c)) +#define _MMIO_PORT3(pipe, a, b, c) _MMIO(_PICK(pipe, a, b, c)) +#define _MMIO_PHY3(phy, a, b, c) _MMIO(_PHY3(phy, a, b, c)) +#define _MMIO_PLL3(pll, ...) _MMIO(_PICK(pll, __VA_ARGS__)) + +/* + * Device info offset array based helpers for groups of registers with unevenly + * spaced base offsets. + */ +#define _MMIO_PIPE2(pipe, reg) _MMIO(INTEL_INFO(dev_priv)->display.pipe_offsets[(pipe)] - \ + INTEL_INFO(dev_priv)->display.pipe_offsets[PIPE_A] + \ + DISPLAY_MMIO_BASE(dev_priv) + (reg)) +#define _MMIO_TRANS2(tran, reg) _MMIO(INTEL_INFO(dev_priv)->display.trans_offsets[(tran)] - \ + INTEL_INFO(dev_priv)->display.trans_offsets[TRANSCODER_A] + \ + DISPLAY_MMIO_BASE(dev_priv) + (reg)) +#define _MMIO_CURSOR2(pipe, reg) _MMIO(INTEL_INFO(dev_priv)->display.cursor_offsets[(pipe)] - \ + INTEL_INFO(dev_priv)->display.cursor_offsets[PIPE_A] + \ + DISPLAY_MMIO_BASE(dev_priv) + (reg)) + +#endif /* __INTEL_DISPLAY_REG_DEFS_H__ */ diff --git a/drivers/gpu/drm/i915/display/intel_display_trace.h b/drivers/gpu/drm/i915/display/intel_display_trace.h index 2dd5a4b7f5d8..725aba3fa531 100644 --- a/drivers/gpu/drm/i915/display/intel_display_trace.h +++ b/drivers/gpu/drm/i915/display/intel_display_trace.h @@ -18,11 +18,15 @@ #include "intel_crtc.h" #include "intel_display_types.h" +#define __dev_name_i915(i915) dev_name((i915)->drm.dev) +#define __dev_name_kms(obj) dev_name((obj)->base.dev->dev) + TRACE_EVENT(intel_pipe_enable, TP_PROTO(struct intel_crtc *crtc), TP_ARGS(crtc), TP_STRUCT__entry( + __string(dev, __dev_name_kms(crtc)) __array(u32, frame, 3) __array(u32, scanline, 3) __field(enum pipe, pipe) @@ -30,6 +34,7 @@ TRACE_EVENT(intel_pipe_enable, TP_fast_assign( struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); struct intel_crtc *it__; + __assign_str(dev, __dev_name_kms(crtc)); for_each_intel_crtc(&dev_priv->drm, it__) { __entry->frame[it__->pipe] = intel_crtc_get_vblank_counter(it__); __entry->scanline[it__->pipe] = intel_get_crtc_scanline(it__); @@ -37,8 +42,8 @@ TRACE_EVENT(intel_pipe_enable, __entry->pipe = crtc->pipe; ), - TP_printk("pipe %c enable, pipe A: frame=%u, scanline=%u, pipe B: frame=%u, scanline=%u, pipe C: frame=%u, scanline=%u", - pipe_name(__entry->pipe), + TP_printk("dev %s, pipe %c enable, pipe A: frame=%u, scanline=%u, pipe B: frame=%u, scanline=%u, pipe C: frame=%u, scanline=%u", + __get_str(dev), pipe_name(__entry->pipe), __entry->frame[PIPE_A], __entry->scanline[PIPE_A], __entry->frame[PIPE_B], __entry->scanline[PIPE_B], __entry->frame[PIPE_C], __entry->scanline[PIPE_C]) @@ -49,6 +54,7 @@ TRACE_EVENT(intel_pipe_disable, TP_ARGS(crtc), TP_STRUCT__entry( + __string(dev, __dev_name_kms(crtc)) __array(u32, frame, 3) __array(u32, scanline, 3) __field(enum pipe, pipe) @@ -57,6 +63,7 @@ TRACE_EVENT(intel_pipe_disable, TP_fast_assign( struct drm_i915_private *dev_priv = to_i915(crtc->base.dev); struct intel_crtc *it__; + __assign_str(dev, __dev_name_kms(crtc)); for_each_intel_crtc(&dev_priv->drm, it__) { __entry->frame[it__->pipe] = intel_crtc_get_vblank_counter(it__); __entry->scanline[it__->pipe] = intel_get_crtc_scanline(it__); @@ -64,8 +71,8 @@ TRACE_EVENT(intel_pipe_disable, __entry->pipe = crtc->pipe; ), - TP_printk("pipe %c disable, pipe A: frame=%u, scanline=%u, pipe B: frame=%u, scanline=%u, pipe C: frame=%u, scanline=%u", - pipe_name(__entry->pipe), + TP_printk("dev %s, pipe %c disable, pipe A: frame=%u, scanline=%u, pipe B: frame=%u, scanline=%u, pipe C: frame=%u, scanline=%u", + __get_str(dev), pipe_name(__entry->pipe), __entry->frame[PIPE_A], __entry->scanline[PIPE_A], __entry->frame[PIPE_B], __entry->scanline[PIPE_B], __entry->frame[PIPE_C], __entry->scanline[PIPE_C]) @@ -76,6 +83,7 @@ TRACE_EVENT(intel_pipe_crc, TP_ARGS(crtc, crcs), TP_STRUCT__entry( + __string(dev, __dev_name_kms(crtc)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) @@ -83,16 +91,19 @@ TRACE_EVENT(intel_pipe_crc, ), TP_fast_assign( + __assign_str(dev, __dev_name_kms(crtc)); __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); memcpy(__entry->crcs, crcs, sizeof(__entry->crcs)); ), - TP_printk("pipe %c, frame=%u, scanline=%u crc=%08x %08x %08x %08x %08x", - pipe_name(__entry->pipe), __entry->frame, __entry->scanline, - __entry->crcs[0], __entry->crcs[1], __entry->crcs[2], - __entry->crcs[3], __entry->crcs[4]) + TP_printk("dev %s, pipe %c, frame=%u, scanline=%u crc=%08x %08x %08x %08x %08x", + __get_str(dev), pipe_name(__entry->pipe), + __entry->frame, __entry->scanline, + __entry->crcs[0], __entry->crcs[1], + __entry->crcs[2], __entry->crcs[3], + __entry->crcs[4]) ); TRACE_EVENT(intel_cpu_fifo_underrun, @@ -100,6 +111,7 @@ TRACE_EVENT(intel_cpu_fifo_underrun, TP_ARGS(dev_priv, pipe), TP_STRUCT__entry( + __string(dev, __dev_name_i915(dev_priv)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) @@ -107,13 +119,14 @@ TRACE_EVENT(intel_cpu_fifo_underrun, TP_fast_assign( struct intel_crtc *crtc = intel_crtc_for_pipe(dev_priv, pipe); + __assign_str(dev, __dev_name_kms(crtc)); __entry->pipe = pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); ), - TP_printk("pipe %c, frame=%u, scanline=%u", - pipe_name(__entry->pipe), + TP_printk("dev %s, pipe %c, frame=%u, scanline=%u", + __get_str(dev), pipe_name(__entry->pipe), __entry->frame, __entry->scanline) ); @@ -122,6 +135,7 @@ TRACE_EVENT(intel_pch_fifo_underrun, TP_ARGS(dev_priv, pch_transcoder), TP_STRUCT__entry( + __string(dev, __dev_name_i915(dev_priv)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) @@ -130,13 +144,14 @@ TRACE_EVENT(intel_pch_fifo_underrun, TP_fast_assign( enum pipe pipe = pch_transcoder; struct intel_crtc *crtc = intel_crtc_for_pipe(dev_priv, pipe); + __assign_str(dev, __dev_name_i915(dev_priv)); __entry->pipe = pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); ), - TP_printk("pch transcoder %c, frame=%u, scanline=%u", - pipe_name(__entry->pipe), + TP_printk("dev %s, pch transcoder %c, frame=%u, scanline=%u", + __get_str(dev), pipe_name(__entry->pipe), __entry->frame, __entry->scanline) ); @@ -145,6 +160,7 @@ TRACE_EVENT(intel_memory_cxsr, TP_ARGS(dev_priv, old, new), TP_STRUCT__entry( + __string(dev, __dev_name_i915(dev_priv)) __array(u32, frame, 3) __array(u32, scanline, 3) __field(bool, old) @@ -153,6 +169,7 @@ TRACE_EVENT(intel_memory_cxsr, TP_fast_assign( struct intel_crtc *crtc; + __assign_str(dev, __dev_name_i915(dev_priv)); for_each_intel_crtc(&dev_priv->drm, crtc) { __entry->frame[crtc->pipe] = intel_crtc_get_vblank_counter(crtc); __entry->scanline[crtc->pipe] = intel_get_crtc_scanline(crtc); @@ -161,8 +178,8 @@ TRACE_EVENT(intel_memory_cxsr, __entry->new = new; ), - TP_printk("%s->%s, pipe A: frame=%u, scanline=%u, pipe B: frame=%u, scanline=%u, pipe C: frame=%u, scanline=%u", - str_on_off(__entry->old), str_on_off(__entry->new), + TP_printk("dev %s, cxsr %s->%s, pipe A: frame=%u, scanline=%u, pipe B: frame=%u, scanline=%u, pipe C: frame=%u, scanline=%u", + __get_str(dev), str_on_off(__entry->old), str_on_off(__entry->new), __entry->frame[PIPE_A], __entry->scanline[PIPE_A], __entry->frame[PIPE_B], __entry->scanline[PIPE_B], __entry->frame[PIPE_C], __entry->scanline[PIPE_C]) @@ -173,6 +190,7 @@ TRACE_EVENT(g4x_wm, TP_ARGS(crtc, wm), TP_STRUCT__entry( + __string(dev, __dev_name_kms(crtc)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) @@ -191,6 +209,7 @@ TRACE_EVENT(g4x_wm, ), TP_fast_assign( + __assign_str(dev, __dev_name_kms(crtc)); __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); @@ -208,8 +227,9 @@ TRACE_EVENT(g4x_wm, __entry->fbc = wm->fbc_en; ), - TP_printk("pipe %c, frame=%u, scanline=%u, wm %d/%d/%d, sr %s/%d/%d/%d, hpll %s/%d/%d/%d, fbc %s", - pipe_name(__entry->pipe), __entry->frame, __entry->scanline, + TP_printk("dev %s, pipe %c, frame=%u, scanline=%u, wm %d/%d/%d, sr %s/%d/%d/%d, hpll %s/%d/%d/%d, fbc %s", + __get_str(dev), pipe_name(__entry->pipe), + __entry->frame, __entry->scanline, __entry->primary, __entry->sprite, __entry->cursor, str_yes_no(__entry->cxsr), __entry->sr_plane, __entry->sr_cursor, __entry->sr_fbc, str_yes_no(__entry->hpll), __entry->hpll_plane, __entry->hpll_cursor, __entry->hpll_fbc, @@ -221,6 +241,7 @@ TRACE_EVENT(vlv_wm, TP_ARGS(crtc, wm), TP_STRUCT__entry( + __string(dev, __dev_name_kms(crtc)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) @@ -235,6 +256,7 @@ TRACE_EVENT(vlv_wm, ), TP_fast_assign( + __assign_str(dev, __dev_name_kms(crtc)); __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); @@ -248,9 +270,10 @@ TRACE_EVENT(vlv_wm, __entry->sr_cursor = wm->sr.cursor; ), - TP_printk("pipe %c, frame=%u, scanline=%u, level=%d, cxsr=%d, wm %d/%d/%d/%d, sr %d/%d", - pipe_name(__entry->pipe), __entry->frame, - __entry->scanline, __entry->level, __entry->cxsr, + TP_printk("dev %s, pipe %c, frame=%u, scanline=%u, level=%d, cxsr=%d, wm %d/%d/%d/%d, sr %d/%d", + __get_str(dev), pipe_name(__entry->pipe), + __entry->frame, __entry->scanline, + __entry->level, __entry->cxsr, __entry->primary, __entry->sprite0, __entry->sprite1, __entry->cursor, __entry->sr_plane, __entry->sr_cursor) ); @@ -260,6 +283,7 @@ TRACE_EVENT(vlv_fifo_size, TP_ARGS(crtc, sprite0_start, sprite1_start, fifo_size), TP_STRUCT__entry( + __string(dev, __dev_name_kms(crtc)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) @@ -269,6 +293,7 @@ TRACE_EVENT(vlv_fifo_size, ), TP_fast_assign( + __assign_str(dev, __dev_name_kms(crtc)); __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); @@ -277,90 +302,96 @@ TRACE_EVENT(vlv_fifo_size, __entry->fifo_size = fifo_size; ), - TP_printk("pipe %c, frame=%u, scanline=%u, %d/%d/%d", - pipe_name(__entry->pipe), __entry->frame, - __entry->scanline, __entry->sprite0_start, - __entry->sprite1_start, __entry->fifo_size) + TP_printk("dev %s, pipe %c, frame=%u, scanline=%u, %d/%d/%d", + __get_str(dev), pipe_name(__entry->pipe), + __entry->frame, __entry->scanline, + __entry->sprite0_start, __entry->sprite1_start, __entry->fifo_size) ); TRACE_EVENT(intel_plane_update_noarm, - TP_PROTO(struct drm_plane *plane, struct intel_crtc *crtc), + TP_PROTO(struct intel_plane *plane, struct intel_crtc *crtc), TP_ARGS(plane, crtc), TP_STRUCT__entry( + __string(dev, __dev_name_kms(plane)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) __array(int, src, 4) __array(int, dst, 4) - __string(name, plane->name) + __string(name, plane->base.name) ), TP_fast_assign( - __assign_str(name, plane->name); + __assign_str(dev, __dev_name_kms(plane)); + __assign_str(name, plane->base.name); __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); - memcpy(__entry->src, &plane->state->src, sizeof(__entry->src)); - memcpy(__entry->dst, &plane->state->dst, sizeof(__entry->dst)); + memcpy(__entry->src, &plane->base.state->src, sizeof(__entry->src)); + memcpy(__entry->dst, &plane->base.state->dst, sizeof(__entry->dst)); ), - TP_printk("pipe %c, plane %s, frame=%u, scanline=%u, " DRM_RECT_FP_FMT " -> " DRM_RECT_FMT, - pipe_name(__entry->pipe), __get_str(name), + TP_printk("dev %s, pipe %c, plane %s, frame=%u, scanline=%u, " DRM_RECT_FP_FMT " -> " DRM_RECT_FMT, + __get_str(dev), pipe_name(__entry->pipe), __get_str(name), __entry->frame, __entry->scanline, DRM_RECT_FP_ARG((const struct drm_rect *)__entry->src), DRM_RECT_ARG((const struct drm_rect *)__entry->dst)) ); TRACE_EVENT(intel_plane_update_arm, - TP_PROTO(struct drm_plane *plane, struct intel_crtc *crtc), + TP_PROTO(struct intel_plane *plane, struct intel_crtc *crtc), TP_ARGS(plane, crtc), TP_STRUCT__entry( + __string(dev, __dev_name_kms(plane)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) __array(int, src, 4) __array(int, dst, 4) - __string(name, plane->name) + __string(name, plane->base.name) ), TP_fast_assign( - __assign_str(name, plane->name); + __assign_str(dev, __dev_name_kms(plane)); + __assign_str(name, plane->base.name); __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); - memcpy(__entry->src, &plane->state->src, sizeof(__entry->src)); - memcpy(__entry->dst, &plane->state->dst, sizeof(__entry->dst)); + memcpy(__entry->src, &plane->base.state->src, sizeof(__entry->src)); + memcpy(__entry->dst, &plane->base.state->dst, sizeof(__entry->dst)); ), - TP_printk("pipe %c, plane %s, frame=%u, scanline=%u, " DRM_RECT_FP_FMT " -> " DRM_RECT_FMT, - pipe_name(__entry->pipe), __get_str(name), + TP_printk("dev %s, pipe %c, plane %s, frame=%u, scanline=%u, " DRM_RECT_FP_FMT " -> " DRM_RECT_FMT, + __get_str(dev), pipe_name(__entry->pipe), __get_str(name), __entry->frame, __entry->scanline, DRM_RECT_FP_ARG((const struct drm_rect *)__entry->src), DRM_RECT_ARG((const struct drm_rect *)__entry->dst)) ); TRACE_EVENT(intel_plane_disable_arm, - TP_PROTO(struct drm_plane *plane, struct intel_crtc *crtc), + TP_PROTO(struct intel_plane *plane, struct intel_crtc *crtc), TP_ARGS(plane, crtc), TP_STRUCT__entry( + __string(dev, __dev_name_kms(plane)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) - __string(name, plane->name) + __string(name, plane->base.name) ), TP_fast_assign( - __assign_str(name, plane->name); + __assign_str(dev, __dev_name_kms(plane)); + __assign_str(name, plane->base.name); __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); ), - TP_printk("pipe %c, plane %s, frame=%u, scanline=%u", - pipe_name(__entry->pipe), __get_str(name), + TP_printk("dev %s, pipe %c, plane %s, frame=%u, scanline=%u", + __get_str(dev), pipe_name(__entry->pipe), __get_str(name), __entry->frame, __entry->scanline) ); @@ -369,6 +400,8 @@ TRACE_EVENT(intel_fbc_activate, TP_ARGS(plane), TP_STRUCT__entry( + __string(dev, __dev_name_kms(plane)) + __string(name, plane->base.name) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) @@ -377,13 +410,16 @@ TRACE_EVENT(intel_fbc_activate, TP_fast_assign( struct intel_crtc *crtc = intel_crtc_for_pipe(to_i915(plane->base.dev), plane->pipe); + __assign_str(dev, __dev_name_kms(plane)); + __assign_str(name, plane->base.name) __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); ), - TP_printk("pipe %c, frame=%u, scanline=%u", - pipe_name(__entry->pipe), __entry->frame, __entry->scanline) + TP_printk("dev %s, pipe %c, plane %s, frame=%u, scanline=%u", + __get_str(dev), pipe_name(__entry->pipe), __get_str(name), + __entry->frame, __entry->scanline) ); TRACE_EVENT(intel_fbc_deactivate, @@ -391,6 +427,8 @@ TRACE_EVENT(intel_fbc_deactivate, TP_ARGS(plane), TP_STRUCT__entry( + __string(dev, __dev_name_kms(plane)) + __string(name, plane->base.name) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) @@ -399,13 +437,16 @@ TRACE_EVENT(intel_fbc_deactivate, TP_fast_assign( struct intel_crtc *crtc = intel_crtc_for_pipe(to_i915(plane->base.dev), plane->pipe); + __assign_str(dev, __dev_name_kms(plane)); + __assign_str(name, plane->base.name) __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); ), - TP_printk("pipe %c, frame=%u, scanline=%u", - pipe_name(__entry->pipe), __entry->frame, __entry->scanline) + TP_printk("dev %s, pipe %c, plane %s, frame=%u, scanline=%u", + __get_str(dev), pipe_name(__entry->pipe), __get_str(name), + __entry->frame, __entry->scanline) ); TRACE_EVENT(intel_fbc_nuke, @@ -413,6 +454,8 @@ TRACE_EVENT(intel_fbc_nuke, TP_ARGS(plane), TP_STRUCT__entry( + __string(dev, __dev_name_kms(plane)) + __string(name, plane->base.name) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) @@ -421,13 +464,16 @@ TRACE_EVENT(intel_fbc_nuke, TP_fast_assign( struct intel_crtc *crtc = intel_crtc_for_pipe(to_i915(plane->base.dev), plane->pipe); + __assign_str(dev, __dev_name_kms(plane)); + __assign_str(name, plane->base.name) __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); ), - TP_printk("pipe %c, frame=%u, scanline=%u", - pipe_name(__entry->pipe), __entry->frame, __entry->scanline) + TP_printk("dev %s, pipe %c, plane %s, frame=%u, scanline=%u", + __get_str(dev), pipe_name(__entry->pipe), __get_str(name), + __entry->frame, __entry->scanline) ); TRACE_EVENT(intel_crtc_vblank_work_start, @@ -435,20 +481,22 @@ TRACE_EVENT(intel_crtc_vblank_work_start, TP_ARGS(crtc), TP_STRUCT__entry( + __string(dev, __dev_name_kms(crtc)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) ), TP_fast_assign( + __assign_str(dev, __dev_name_kms(crtc)); __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); ), - TP_printk("pipe %c, frame=%u, scanline=%u", - pipe_name(__entry->pipe), __entry->frame, - __entry->scanline) + TP_printk("dev %s, pipe %c, frame=%u, scanline=%u", + __get_str(dev), pipe_name(__entry->pipe), + __entry->frame, __entry->scanline) ); TRACE_EVENT(intel_crtc_vblank_work_end, @@ -456,20 +504,22 @@ TRACE_EVENT(intel_crtc_vblank_work_end, TP_ARGS(crtc), TP_STRUCT__entry( + __string(dev, __dev_name_kms(crtc)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) ), TP_fast_assign( + __assign_str(dev, __dev_name_kms(crtc)); __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); ), - TP_printk("pipe %c, frame=%u, scanline=%u", - pipe_name(__entry->pipe), __entry->frame, - __entry->scanline) + TP_printk("dev %s, pipe %c, frame=%u, scanline=%u", + __get_str(dev), pipe_name(__entry->pipe), + __entry->frame, __entry->scanline) ); TRACE_EVENT(intel_pipe_update_start, @@ -477,6 +527,7 @@ TRACE_EVENT(intel_pipe_update_start, TP_ARGS(crtc), TP_STRUCT__entry( + __string(dev, __dev_name_kms(crtc)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) @@ -485,6 +536,7 @@ TRACE_EVENT(intel_pipe_update_start, ), TP_fast_assign( + __assign_str(dev, __dev_name_kms(crtc)); __entry->pipe = crtc->pipe; __entry->frame = intel_crtc_get_vblank_counter(crtc); __entry->scanline = intel_get_crtc_scanline(crtc); @@ -492,9 +544,10 @@ TRACE_EVENT(intel_pipe_update_start, __entry->max = crtc->debug.max_vbl; ), - TP_printk("pipe %c, frame=%u, scanline=%u, min=%u, max=%u", - pipe_name(__entry->pipe), __entry->frame, - __entry->scanline, __entry->min, __entry->max) + TP_printk("dev %s, pipe %c, frame=%u, scanline=%u, min=%u, max=%u", + __get_str(dev), pipe_name(__entry->pipe), + __entry->frame, __entry->scanline, + __entry->min, __entry->max) ); TRACE_EVENT(intel_pipe_update_vblank_evaded, @@ -502,6 +555,7 @@ TRACE_EVENT(intel_pipe_update_vblank_evaded, TP_ARGS(crtc), TP_STRUCT__entry( + __string(dev, __dev_name_kms(crtc)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) @@ -510,6 +564,7 @@ TRACE_EVENT(intel_pipe_update_vblank_evaded, ), TP_fast_assign( + __assign_str(dev, __dev_name_kms(crtc)); __entry->pipe = crtc->pipe; __entry->frame = crtc->debug.start_vbl_count; __entry->scanline = crtc->debug.scanline_start; @@ -517,9 +572,10 @@ TRACE_EVENT(intel_pipe_update_vblank_evaded, __entry->max = crtc->debug.max_vbl; ), - TP_printk("pipe %c, frame=%u, scanline=%u, min=%u, max=%u", - pipe_name(__entry->pipe), __entry->frame, - __entry->scanline, __entry->min, __entry->max) + TP_printk("dev %s, pipe %c, frame=%u, scanline=%u, min=%u, max=%u", + __get_str(dev), pipe_name(__entry->pipe), + __entry->frame, __entry->scanline, + __entry->min, __entry->max) ); TRACE_EVENT(intel_pipe_update_end, @@ -527,56 +583,64 @@ TRACE_EVENT(intel_pipe_update_end, TP_ARGS(crtc, frame, scanline_end), TP_STRUCT__entry( + __string(dev, __dev_name_kms(crtc)) __field(enum pipe, pipe) __field(u32, frame) __field(u32, scanline) ), TP_fast_assign( + __assign_str(dev, __dev_name_kms(crtc)); __entry->pipe = crtc->pipe; __entry->frame = frame; __entry->scanline = scanline_end; ), - TP_printk("pipe %c, frame=%u, scanline=%u", - pipe_name(__entry->pipe), __entry->frame, - __entry->scanline) + TP_printk("dev %s, pipe %c, frame=%u, scanline=%u", + __get_str(dev), pipe_name(__entry->pipe), + __entry->frame, __entry->scanline) ); TRACE_EVENT(intel_frontbuffer_invalidate, - TP_PROTO(unsigned int frontbuffer_bits, unsigned int origin), - TP_ARGS(frontbuffer_bits, origin), + TP_PROTO(struct drm_i915_private *i915, + unsigned int frontbuffer_bits, unsigned int origin), + TP_ARGS(i915, frontbuffer_bits, origin), TP_STRUCT__entry( + __string(dev, __dev_name_i915(i915)) __field(unsigned int, frontbuffer_bits) __field(unsigned int, origin) ), TP_fast_assign( + __assign_str(dev, __dev_name_i915(i915)); __entry->frontbuffer_bits = frontbuffer_bits; __entry->origin = origin; ), - TP_printk("frontbuffer_bits=0x%08x, origin=%u", - __entry->frontbuffer_bits, __entry->origin) + TP_printk("dev %s, frontbuffer_bits=0x%08x, origin=%u", + __get_str(dev), __entry->frontbuffer_bits, __entry->origin) ); TRACE_EVENT(intel_frontbuffer_flush, - TP_PROTO(unsigned int frontbuffer_bits, unsigned int origin), - TP_ARGS(frontbuffer_bits, origin), + TP_PROTO(struct drm_i915_private *i915, + unsigned int frontbuffer_bits, unsigned int origin), + TP_ARGS(i915, frontbuffer_bits, origin), TP_STRUCT__entry( + __string(dev, __dev_name_i915(i915)) __field(unsigned int, frontbuffer_bits) __field(unsigned int, origin) ), TP_fast_assign( + __assign_str(dev, __dev_name_i915(i915)); __entry->frontbuffer_bits = frontbuffer_bits; __entry->origin = origin; ), - TP_printk("frontbuffer_bits=0x%08x, origin=%u", - __entry->frontbuffer_bits, __entry->origin) + TP_printk("dev %s, frontbuffer_bits=0x%08x, origin=%u", + __get_str(dev), __entry->frontbuffer_bits, __entry->origin) ); #endif /* __INTEL_DISPLAY_TRACE_H__ */ diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h b/drivers/gpu/drm/i915/display/intel_display_types.h index 7f18c052ec16..f07395065a69 100644 --- a/drivers/gpu/drm/i915/display/intel_display_types.h +++ b/drivers/gpu/drm/i915/display/intel_display_types.h @@ -969,6 +969,15 @@ struct intel_mpllb_state { u32 mpllb_sscstep; }; +/* Used by dp and fdi links */ +struct intel_link_m_n { + u32 tu; + u32 data_m; + u32 data_n; + u32 link_m; + u32 link_n; +}; + struct intel_crtc_state { /* * uapi (drm) state. This is the software state shown to userspace. @@ -1366,6 +1375,7 @@ struct intel_crtc { u16 vmax_vblank_start; struct intel_display_power_domain_set enabled_power_domains; + struct intel_display_power_domain_set hw_readout_power_domains; struct intel_overlay *overlay; struct intel_crtc_state *config; @@ -1803,51 +1813,6 @@ struct intel_dp_mst_encoder { struct intel_connector *connector; }; -static inline enum dpio_channel -vlv_dig_port_to_channel(struct intel_digital_port *dig_port) -{ - switch (dig_port->base.port) { - default: - MISSING_CASE(dig_port->base.port); - fallthrough; - case PORT_B: - case PORT_D: - return DPIO_CH0; - case PORT_C: - return DPIO_CH1; - } -} - -static inline enum dpio_phy -vlv_dig_port_to_phy(struct intel_digital_port *dig_port) -{ - switch (dig_port->base.port) { - default: - MISSING_CASE(dig_port->base.port); - fallthrough; - case PORT_B: - case PORT_C: - return DPIO_PHY0; - case PORT_D: - return DPIO_PHY1; - } -} - -static inline enum dpio_channel -vlv_pipe_to_channel(enum pipe pipe) -{ - switch (pipe) { - default: - MISSING_CASE(pipe); - fallthrough; - case PIPE_A: - case PIPE_C: - return DPIO_CH0; - case PIPE_B: - return DPIO_CH1; - } -} - struct intel_load_detect_pipe { struct drm_atomic_state *restore_state; }; diff --git a/drivers/gpu/drm/i915/display/intel_dp.c b/drivers/gpu/drm/i915/display/intel_dp.c index 7400d6b4c587..67089711d9e2 100644 --- a/drivers/gpu/drm/i915/display/intel_dp.c +++ b/drivers/gpu/drm/i915/display/intel_dp.c @@ -46,6 +46,7 @@ #include "g4x_dp.h" #include "i915_debugfs.h" #include "i915_drv.h" +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_audio.h" #include "intel_backlight.h" @@ -4876,6 +4877,12 @@ void intel_dp_encoder_flush_work(struct drm_encoder *encoder) intel_pps_vdd_off_sync(intel_dp); + /* + * Ensure power off delay is respected on module remove, so that we can + * reduce delays at driver probe. See pps_init_timestamps(). + */ + intel_pps_wait_power_cycle(intel_dp); + intel_dp_aux_fini(intel_dp); } diff --git a/drivers/gpu/drm/i915/display/intel_dp_aux.c b/drivers/gpu/drm/i915/display/intel_dp_aux.c index 48c375c65a41..664bebdecea7 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_aux.c +++ b/drivers/gpu/drm/i915/display/intel_dp_aux.c @@ -4,6 +4,7 @@ */ #include "i915_drv.h" +#include "i915_reg.h" #include "i915_trace.h" #include "intel_display_types.h" #include "intel_dp_aux.h" diff --git a/drivers/gpu/drm/i915/display/intel_dp_hdcp.c b/drivers/gpu/drm/i915/display/intel_dp_hdcp.c index 35360dd543ac..e0c177161407 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_hdcp.c +++ b/drivers/gpu/drm/i915/display/intel_dp_hdcp.c @@ -11,6 +11,7 @@ #include <drm/display/drm_hdcp_helper.h> #include <drm/drm_print.h> +#include "i915_reg.h" #include "intel_ddi.h" #include "intel_de.h" #include "intel_display_types.h" diff --git a/drivers/gpu/drm/i915/display/intel_dp_mst.c b/drivers/gpu/drm/i915/display/intel_dp_mst.c index cd4e61026d98..4077a979a924 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_mst.c +++ b/drivers/gpu/drm/i915/display/intel_dp_mst.c @@ -29,6 +29,7 @@ #include <drm/drm_probe_helper.h> #include "i915_drv.h" +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_audio.h" #include "intel_connector.h" diff --git a/drivers/gpu/drm/i915/display/intel_dpio_phy.c b/drivers/gpu/drm/i915/display/intel_dpio_phy.c index 8732b8722ed7..7eb7440b3180 100644 --- a/drivers/gpu/drm/i915/display/intel_dpio_phy.c +++ b/drivers/gpu/drm/i915/display/intel_dpio_phy.c @@ -21,6 +21,7 @@ * DEALINGS IN THE SOFTWARE. */ +#include "i915_reg.h" #include "intel_ddi.h" #include "intel_ddi_buf_trans.h" #include "intel_de.h" @@ -655,6 +656,48 @@ bxt_ddi_phy_get_lane_lat_optim_mask(struct intel_encoder *encoder) return mask; } +enum dpio_channel vlv_dig_port_to_channel(struct intel_digital_port *dig_port) +{ + switch (dig_port->base.port) { + default: + MISSING_CASE(dig_port->base.port); + fallthrough; + case PORT_B: + case PORT_D: + return DPIO_CH0; + case PORT_C: + return DPIO_CH1; + } +} + +enum dpio_phy vlv_dig_port_to_phy(struct intel_digital_port *dig_port) +{ + switch (dig_port->base.port) { + default: + MISSING_CASE(dig_port->base.port); + fallthrough; + case PORT_B: + case PORT_C: + return DPIO_PHY0; + case PORT_D: + return DPIO_PHY1; + } +} + +enum dpio_channel vlv_pipe_to_channel(enum pipe pipe) +{ + switch (pipe) { + default: + MISSING_CASE(pipe); + fallthrough; + case PIPE_A: + case PIPE_C: + return DPIO_CH0; + case PIPE_B: + return DPIO_CH1; + } +} + void chv_set_phy_signal_level(struct intel_encoder *encoder, const struct intel_crtc_state *crtc_state, u32 deemph_reg_value, u32 margin_reg_value, diff --git a/drivers/gpu/drm/i915/display/intel_dpio_phy.h b/drivers/gpu/drm/i915/display/intel_dpio_phy.h index 9c3d008e8e1a..9c7725dacb47 100644 --- a/drivers/gpu/drm/i915/display/intel_dpio_phy.h +++ b/drivers/gpu/drm/i915/display/intel_dpio_phy.h @@ -8,13 +8,24 @@ #include <linux/types.h> -enum dpio_channel; -enum dpio_phy; +enum pipe; enum port; struct drm_i915_private; struct intel_crtc_state; +struct intel_digital_port; struct intel_encoder; +enum dpio_channel { + DPIO_CH0, + DPIO_CH1, +}; + +enum dpio_phy { + DPIO_PHY0, + DPIO_PHY1, + DPIO_PHY2, +}; + void bxt_port_to_phy_channel(struct drm_i915_private *dev_priv, enum port port, enum dpio_phy *phy, enum dpio_channel *ch); void bxt_ddi_phy_set_signal_levels(struct intel_encoder *encoder, @@ -30,6 +41,10 @@ void bxt_ddi_phy_set_lane_optim_mask(struct intel_encoder *encoder, u8 lane_lat_optim_mask); u8 bxt_ddi_phy_get_lane_lat_optim_mask(struct intel_encoder *encoder); +enum dpio_channel vlv_dig_port_to_channel(struct intel_digital_port *dig_port); +enum dpio_phy vlv_dig_port_to_phy(struct intel_digital_port *dig_port); +enum dpio_channel vlv_pipe_to_channel(enum pipe pipe); + void chv_set_phy_signal_level(struct intel_encoder *encoder, const struct intel_crtc_state *crtc_state, u32 deemph_reg_value, u32 margin_reg_value, diff --git a/drivers/gpu/drm/i915/display/intel_dpll.c b/drivers/gpu/drm/i915/display/intel_dpll.c index b15ba78d64d6..c236aafe9be0 100644 --- a/drivers/gpu/drm/i915/display/intel_dpll.c +++ b/drivers/gpu/drm/i915/display/intel_dpll.c @@ -6,10 +6,12 @@ #include <linux/kernel.h> #include <linux/string_helpers.h> +#include "i915_reg.h" #include "intel_crtc.h" #include "intel_de.h" #include "intel_display.h" #include "intel_display_types.h" +#include "intel_dpio_phy.h" #include "intel_dpll.h" #include "intel_lvds.h" #include "intel_panel.h" diff --git a/drivers/gpu/drm/i915/display/intel_dpll_mgr.c b/drivers/gpu/drm/i915/display/intel_dpll_mgr.c index 7c6c094a0a01..1974eb580ed1 100644 --- a/drivers/gpu/drm/i915/display/intel_dpll_mgr.c +++ b/drivers/gpu/drm/i915/display/intel_dpll_mgr.c @@ -23,6 +23,7 @@ #include <linux/string_helpers.h> +#include "i915_reg.h" #include "intel_de.h" #include "intel_display_types.h" #include "intel_dkl_phy.h" @@ -30,6 +31,7 @@ #include "intel_dpio_phy.h" #include "intel_dpll.h" #include "intel_dpll_mgr.h" +#include "intel_hti.h" #include "intel_mg_phy_regs.h" #include "intel_pch_refclk.h" #include "intel_tc.h" @@ -3163,14 +3165,6 @@ static void icl_update_active_dpll(struct intel_atomic_state *state, icl_set_active_port_dpll(crtc_state, port_dpll_id); } -static u32 intel_get_hti_plls(struct drm_i915_private *i915) -{ - if (!(i915->hti_state & HDPORT_ENABLED)) - return 0; - - return REG_FIELD_GET(HDPORT_DPLL_USED_MASK, i915->hti_state); -} - static int icl_compute_combo_phy_dpll(struct intel_atomic_state *state, struct intel_crtc *crtc) { @@ -3245,7 +3239,7 @@ static int icl_get_combo_phy_dpll(struct intel_atomic_state *state, } /* Eliminate DPLLs from consideration if reserved by HTI */ - dpll_mask &= ~intel_get_hti_plls(dev_priv); + dpll_mask &= ~intel_hti_dpll_mask(dev_priv); port_dpll->pll = intel_find_shared_dpll(state, crtc, &port_dpll->hw_state, diff --git a/drivers/gpu/drm/i915/display/intel_drrs.c b/drivers/gpu/drm/i915/display/intel_drrs.c index e27408efaae2..5b9e44443814 100644 --- a/drivers/gpu/drm/i915/display/intel_drrs.c +++ b/drivers/gpu/drm/i915/display/intel_drrs.c @@ -4,6 +4,7 @@ */ #include "i915_drv.h" +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_de.h" #include "intel_display_types.h" diff --git a/drivers/gpu/drm/i915/display/intel_dsb.c b/drivers/gpu/drm/i915/display/intel_dsb.c index fc9c3e41c333..1e1c6107d51b 100644 --- a/drivers/gpu/drm/i915/display/intel_dsb.c +++ b/drivers/gpu/drm/i915/display/intel_dsb.c @@ -7,6 +7,7 @@ #include "gem/i915_gem_internal.h" #include "i915_drv.h" +#include "i915_reg.h" #include "intel_de.h" #include "intel_display_types.h" #include "intel_dsb.h" diff --git a/drivers/gpu/drm/i915/display/intel_dvo.c b/drivers/gpu/drm/i915/display/intel_dvo.c index 595087288922..c86f9890754d 100644 --- a/drivers/gpu/drm/i915/display/intel_dvo.c +++ b/drivers/gpu/drm/i915/display/intel_dvo.c @@ -32,6 +32,7 @@ #include <drm/drm_crtc.h> #include "i915_drv.h" +#include "i915_reg.h" #include "intel_connector.h" #include "intel_de.h" #include "intel_display_types.h" diff --git a/drivers/gpu/drm/i915/display/intel_dvo_dev.h b/drivers/gpu/drm/i915/display/intel_dvo_dev.h index 50205f064d93..ecff7b190856 100644 --- a/drivers/gpu/drm/i915/display/intel_dvo_dev.h +++ b/drivers/gpu/drm/i915/display/intel_dvo_dev.h @@ -23,12 +23,12 @@ #ifndef __INTEL_DVO_DEV_H__ #define __INTEL_DVO_DEV_H__ -#include <linux/i2c.h> - -#include <drm/drm_crtc.h> - #include "i915_reg_defs.h" +enum drm_connector_status; +struct drm_display_mode; +struct i2c_adapter; + struct intel_dvo_device { const char *name; int type; diff --git a/drivers/gpu/drm/i915/display/intel_fbc.c b/drivers/gpu/drm/i915/display/intel_fbc.c index 3f24f326b989..b5ee5ea0d010 100644 --- a/drivers/gpu/drm/i915/display/intel_fbc.c +++ b/drivers/gpu/drm/i915/display/intel_fbc.c @@ -1183,7 +1183,7 @@ static bool intel_fbc_can_flip_nuke(struct intel_atomic_state *state, const struct drm_framebuffer *old_fb = old_plane_state->hw.fb; const struct drm_framebuffer *new_fb = new_plane_state->hw.fb; - if (drm_atomic_crtc_needs_modeset(&new_crtc_state->uapi)) + if (intel_crtc_needs_modeset(new_crtc_state)) return false; if (!intel_fbc_is_ok(old_plane_state) || diff --git a/drivers/gpu/drm/i915/display/intel_fdi.c b/drivers/gpu/drm/i915/display/intel_fdi.c index 7f47e5c85c81..063f1da4f229 100644 --- a/drivers/gpu/drm/i915/display/intel_fdi.c +++ b/drivers/gpu/drm/i915/display/intel_fdi.c @@ -5,6 +5,7 @@ #include <linux/string_helpers.h> +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_crtc.h" #include "intel_ddi.h" diff --git a/drivers/gpu/drm/i915/display/intel_fifo_underrun.h b/drivers/gpu/drm/i915/display/intel_fifo_underrun.h index e04f22ac1f49..2e47d7d3c101 100644 --- a/drivers/gpu/drm/i915/display/intel_fifo_underrun.h +++ b/drivers/gpu/drm/i915/display/intel_fifo_underrun.h @@ -8,9 +8,8 @@ #include <linux/types.h> -#include "intel_display.h" - struct drm_i915_private; +enum pipe; bool intel_set_cpu_fifo_underrun_reporting(struct drm_i915_private *dev_priv, enum pipe pipe, bool enable); diff --git a/drivers/gpu/drm/i915/display/intel_frontbuffer.c b/drivers/gpu/drm/i915/display/intel_frontbuffer.c index d80e3e8a9b01..17a7aa8b28c2 100644 --- a/drivers/gpu/drm/i915/display/intel_frontbuffer.c +++ b/drivers/gpu/drm/i915/display/intel_frontbuffer.c @@ -88,7 +88,7 @@ static void frontbuffer_flush(struct drm_i915_private *i915, if (!frontbuffer_bits) return; - trace_intel_frontbuffer_flush(frontbuffer_bits, origin); + trace_intel_frontbuffer_flush(i915, frontbuffer_bits, origin); might_sleep(); intel_drrs_flush(i915, frontbuffer_bits); @@ -176,7 +176,7 @@ void __intel_fb_invalidate(struct intel_frontbuffer *front, spin_unlock(&i915->display.fb_tracking.lock); } - trace_intel_frontbuffer_invalidate(frontbuffer_bits, origin); + trace_intel_frontbuffer_invalidate(i915, frontbuffer_bits, origin); might_sleep(); intel_psr_invalidate(i915, frontbuffer_bits, origin); diff --git a/drivers/gpu/drm/i915/display/intel_global_state.c b/drivers/gpu/drm/i915/display/intel_global_state.c index 7a19215ad844..02b593b1e2ea 100644 --- a/drivers/gpu/drm/i915/display/intel_global_state.c +++ b/drivers/gpu/drm/i915/display/intel_global_state.c @@ -45,14 +45,14 @@ void intel_atomic_global_obj_init(struct drm_i915_private *dev_priv, obj->state = state; obj->funcs = funcs; - list_add_tail(&obj->head, &dev_priv->global_obj_list); + list_add_tail(&obj->head, &dev_priv->display.global.obj_list); } void intel_atomic_global_obj_cleanup(struct drm_i915_private *dev_priv) { struct intel_global_obj *obj, *next; - list_for_each_entry_safe(obj, next, &dev_priv->global_obj_list, head) { + list_for_each_entry_safe(obj, next, &dev_priv->display.global.obj_list, head) { list_del(&obj->head); drm_WARN_ON(&dev_priv->drm, kref_read(&obj->state->ref) != 1); diff --git a/drivers/gpu/drm/i915/display/intel_global_state.h b/drivers/gpu/drm/i915/display/intel_global_state.h index 1f16fa3073c9..f01ee0bb3e5a 100644 --- a/drivers/gpu/drm/i915/display/intel_global_state.h +++ b/drivers/gpu/drm/i915/display/intel_global_state.h @@ -27,7 +27,7 @@ struct intel_global_obj { }; #define intel_for_each_global_obj(obj, dev_priv) \ - list_for_each_entry(obj, &(dev_priv)->global_obj_list, head) + list_for_each_entry(obj, &(dev_priv)->display.global.obj_list, head) #define for_each_new_global_obj_in_state(__state, obj, new_obj_state, __i) \ for ((__i) = 0; \ diff --git a/drivers/gpu/drm/i915/display/intel_gmbus.c b/drivers/gpu/drm/i915/display/intel_gmbus.c index 74443f57f62d..a5840a28a69d 100644 --- a/drivers/gpu/drm/i915/display/intel_gmbus.c +++ b/drivers/gpu/drm/i915/display/intel_gmbus.c @@ -34,6 +34,8 @@ #include <drm/display/drm_hdcp_helper.h> #include "i915_drv.h" +#include "i915_irq.h" +#include "i915_reg.h" #include "intel_de.h" #include "intel_display_types.h" #include "intel_gmbus.h" @@ -49,9 +51,27 @@ struct intel_gmbus { struct drm_i915_private *i915; }; +enum gmbus_gpio { + GPIOA, + GPIOB, + GPIOC, + GPIOD, + GPIOE, + GPIOF, + GPIOG, + GPIOH, + __GPIOI_UNUSED, + GPIOJ, + GPIOK, + GPIOL, + GPIOM, + GPION, + GPIOO, +}; + struct gmbus_pin { const char *name; - enum i915_gpio gpio; + enum gmbus_gpio gpio; }; /* Map gmbus pin pairs to names and registers. */ diff --git a/drivers/gpu/drm/i915/display/intel_hdcp_regs.h b/drivers/gpu/drm/i915/display/intel_hdcp_regs.h index 2a3733e8966c..8023c85c7fa0 100644 --- a/drivers/gpu/drm/i915/display/intel_hdcp_regs.h +++ b/drivers/gpu/drm/i915/display/intel_hdcp_regs.h @@ -6,7 +6,7 @@ #ifndef __INTEL_HDCP_REGS_H__ #define __INTEL_HDCP_REGS_H__ -#include "i915_reg_defs.h" +#include "intel_display_reg_defs.h" /* HDCP Key Registers */ #define HDCP_KEY_CONF _MMIO(0x66c00) diff --git a/drivers/gpu/drm/i915/display/intel_hdmi.c b/drivers/gpu/drm/i915/display/intel_hdmi.c index 02f8374ea51f..bac85d88054f 100644 --- a/drivers/gpu/drm/i915/display/intel_hdmi.c +++ b/drivers/gpu/drm/i915/display/intel_hdmi.c @@ -42,6 +42,7 @@ #include "i915_debugfs.h" #include "i915_drv.h" +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_connector.h" #include "intel_ddi.h" @@ -2057,13 +2058,6 @@ static bool hdmi_bpc_possible(const struct intel_crtc_state *crtc_state, int bpc if (!intel_hdmi_source_bpc_possible(dev_priv, bpc)) return false; - /* - * HDMI deep color affects the clocks, so it's only possible - * when not cloning with other encoder types. - */ - if (bpc > 8 && crtc_state->output_types != BIT(INTEL_OUTPUT_HDMI)) - return false; - /* Display Wa_1405510057:icl,ehl */ if (intel_hdmi_is_ycbcr420(crtc_state) && bpc == 10 && DISPLAY_VER(dev_priv) == 11 && @@ -2190,9 +2184,13 @@ static bool intel_hdmi_has_audio(struct intel_encoder *encoder, } static enum intel_output_format -intel_hdmi_output_format(struct intel_connector *connector, +intel_hdmi_output_format(const struct intel_crtc_state *crtc_state, + struct intel_connector *connector, bool ycbcr_420_output) { + if (!crtc_state->has_hdmi_sink) + return INTEL_OUTPUT_FORMAT_RGB; + if (connector->base.ycbcr_420_allowed && ycbcr_420_output) return INTEL_OUTPUT_FORMAT_YCBCR420; else @@ -2211,7 +2209,8 @@ static int intel_hdmi_compute_output_format(struct intel_encoder *encoder, bool ycbcr_420_only = drm_mode_is_420_only(info, adjusted_mode); int ret; - crtc_state->output_format = intel_hdmi_output_format(connector, ycbcr_420_only); + crtc_state->output_format = + intel_hdmi_output_format(crtc_state, connector, ycbcr_420_only); if (ycbcr_420_only && !intel_hdmi_is_ycbcr420(crtc_state)) { drm_dbg_kms(&i915->drm, @@ -2226,13 +2225,19 @@ static int intel_hdmi_compute_output_format(struct intel_encoder *encoder, !drm_mode_is_420_also(info, adjusted_mode)) return ret; - crtc_state->output_format = intel_hdmi_output_format(connector, true); + crtc_state->output_format = intel_hdmi_output_format(crtc_state, connector, true); ret = intel_hdmi_compute_clock(encoder, crtc_state, respect_downstream_limits); } return ret; } +static bool intel_hdmi_is_cloned(const struct intel_crtc_state *crtc_state) +{ + return crtc_state->uapi.encoder_mask && + !is_power_of_2(crtc_state->uapi.encoder_mask); +} + int intel_hdmi_compute_config(struct intel_encoder *encoder, struct intel_crtc_state *pipe_config, struct drm_connector_state *conn_state) @@ -2248,8 +2253,9 @@ int intel_hdmi_compute_config(struct intel_encoder *encoder, return -EINVAL; pipe_config->output_format = INTEL_OUTPUT_FORMAT_RGB; - pipe_config->has_hdmi_sink = intel_has_hdmi_sink(intel_hdmi, - conn_state); + pipe_config->has_hdmi_sink = + intel_has_hdmi_sink(intel_hdmi, conn_state) && + !intel_hdmi_is_cloned(pipe_config); if (pipe_config->has_hdmi_sink) pipe_config->has_infoframe = true; @@ -2257,9 +2263,6 @@ int intel_hdmi_compute_config(struct intel_encoder *encoder, if (adjusted_mode->flags & DRM_MODE_FLAG_DBLCLK) pipe_config->pixel_multiplier = 2; - if (HAS_PCH_SPLIT(dev_priv) && !HAS_DDI(dev_priv)) - pipe_config->has_pch_encoder = true; - pipe_config->has_audio = intel_hdmi_has_audio(encoder, pipe_config, conn_state); diff --git a/drivers/gpu/drm/i915/display/intel_hdmi.h b/drivers/gpu/drm/i915/display/intel_hdmi.h index 93f65a917c36..774dda2376ed 100644 --- a/drivers/gpu/drm/i915/display/intel_hdmi.h +++ b/drivers/gpu/drm/i915/display/intel_hdmi.h @@ -6,20 +6,20 @@ #ifndef __INTEL_HDMI_H__ #define __INTEL_HDMI_H__ -#include <linux/hdmi.h> #include <linux/types.h> +enum hdmi_infoframe_type; +enum port; struct drm_connector; +struct drm_connector_state; struct drm_encoder; struct drm_i915_private; struct intel_connector; +struct intel_crtc_state; struct intel_digital_port; struct intel_encoder; -struct intel_crtc_state; struct intel_hdmi; -struct drm_connector_state; union hdmi_infoframe; -enum port; void intel_hdmi_init_connector(struct intel_digital_port *dig_port, struct intel_connector *intel_connector); diff --git a/drivers/gpu/drm/i915/display/intel_hti.c b/drivers/gpu/drm/i915/display/intel_hti.c new file mode 100644 index 000000000000..12a1f4ce1a77 --- /dev/null +++ b/drivers/gpu/drm/i915/display/intel_hti.c @@ -0,0 +1,38 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright © 2022 Intel Corporation + */ + +#include "i915_drv.h" +#include "intel_de.h" +#include "intel_display.h" +#include "intel_hti.h" +#include "intel_hti_regs.h" + +void intel_hti_init(struct drm_i915_private *i915) +{ + /* + * If the platform has HTI, we need to find out whether it has reserved + * any display resources before we create our display outputs. + */ + if (INTEL_INFO(i915)->display.has_hti) + i915->display.hti.state = intel_de_read(i915, HDPORT_STATE); +} + +bool intel_hti_uses_phy(struct drm_i915_private *i915, enum phy phy) +{ + return i915->display.hti.state & HDPORT_ENABLED && + i915->display.hti.state & HDPORT_DDI_USED(phy); +} + +u32 intel_hti_dpll_mask(struct drm_i915_private *i915) +{ + if (!(i915->display.hti.state & HDPORT_ENABLED)) + return 0; + + /* + * Note: This is subtle. The values must coincide with what's defined + * for the platform. + */ + return REG_FIELD_GET(HDPORT_DPLL_USED_MASK, i915->display.hti.state); +} diff --git a/drivers/gpu/drm/i915/display/intel_hti.h b/drivers/gpu/drm/i915/display/intel_hti.h new file mode 100644 index 000000000000..2893d6668657 --- /dev/null +++ b/drivers/gpu/drm/i915/display/intel_hti.h @@ -0,0 +1,18 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright © 2022 Intel Corporation + */ + +#ifndef __INTEL_HTI_H__ +#define __INTEL_HTI_H__ + +#include <linux/types.h> + +struct drm_i915_private; +enum phy; + +void intel_hti_init(struct drm_i915_private *i915); +bool intel_hti_uses_phy(struct drm_i915_private *i915, enum phy phy); +u32 intel_hti_dpll_mask(struct drm_i915_private *i915); + +#endif /* __INTEL_HTI_H__ */ diff --git a/drivers/gpu/drm/i915/display/intel_hti_regs.h b/drivers/gpu/drm/i915/display/intel_hti_regs.h new file mode 100644 index 000000000000..e206f2837fc8 --- /dev/null +++ b/drivers/gpu/drm/i915/display/intel_hti_regs.h @@ -0,0 +1,16 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright © 2022 Intel Corporation + */ + +#ifndef __INTEL_HTI_REGS_H__ +#define __INTEL_HTI_REGS_H__ + +#include "i915_reg_defs.h" + +#define HDPORT_STATE _MMIO(0x45050) +#define HDPORT_DPLL_USED_MASK REG_GENMASK(15, 12) +#define HDPORT_DDI_USED(phy) REG_BIT(2 * (phy) + 1) +#define HDPORT_ENABLED REG_BIT(0) + +#endif /* __INTEL_HTI_REGS_H__ */ diff --git a/drivers/gpu/drm/i915/display/intel_lpe_audio.c b/drivers/gpu/drm/i915/display/intel_lpe_audio.c index 22ca8754ea96..8aaaef4d7856 100644 --- a/drivers/gpu/drm/i915/display/intel_lpe_audio.c +++ b/drivers/gpu/drm/i915/display/intel_lpe_audio.c @@ -71,6 +71,8 @@ #include <drm/intel_lpe_audio.h> #include "i915_drv.h" +#include "i915_irq.h" +#include "i915_reg.h" #include "intel_de.h" #include "intel_lpe_audio.h" #include "intel_pci_config.h" diff --git a/drivers/gpu/drm/i915/display/intel_lspcon.c b/drivers/gpu/drm/i915/display/intel_lspcon.c index 15d59de8810e..9ff1c0b223ad 100644 --- a/drivers/gpu/drm/i915/display/intel_lspcon.c +++ b/drivers/gpu/drm/i915/display/intel_lspcon.c @@ -28,6 +28,7 @@ #include <drm/drm_atomic_helper.h> #include <drm/drm_edid.h> +#include "i915_reg.h" #include "intel_de.h" #include "intel_display_types.h" #include "intel_dp.h" diff --git a/drivers/gpu/drm/i915/display/intel_lvds.c b/drivers/gpu/drm/i915/display/intel_lvds.c index 246787bbf5ef..7bf1bdfd03ec 100644 --- a/drivers/gpu/drm/i915/display/intel_lvds.c +++ b/drivers/gpu/drm/i915/display/intel_lvds.c @@ -39,6 +39,7 @@ #include <drm/drm_edid.h> #include "i915_drv.h" +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_backlight.h" #include "intel_connector.h" diff --git a/drivers/gpu/drm/i915/display/intel_mg_phy_regs.h b/drivers/gpu/drm/i915/display/intel_mg_phy_regs.h index 07978f8d5fb7..0e8248bce52d 100644 --- a/drivers/gpu/drm/i915/display/intel_mg_phy_regs.h +++ b/drivers/gpu/drm/i915/display/intel_mg_phy_regs.h @@ -6,7 +6,7 @@ #ifndef __INTEL_MG_PHY_REGS__ #define __INTEL_MG_PHY_REGS__ -#include "i915_reg_defs.h" +#include "intel_display_reg_defs.h" #define MG_PHY_PORT_LN(ln, tc_port, ln0p1, ln0p2, ln1p1) \ _MMIO(_PORT(tc_port, ln0p1, ln0p2) + (ln) * ((ln1p1) - (ln0p1))) diff --git a/drivers/gpu/drm/i915/display/intel_modeset_setup.c b/drivers/gpu/drm/i915/display/intel_modeset_setup.c index 9d8ca230be39..96395bfbd41d 100644 --- a/drivers/gpu/drm/i915/display/intel_modeset_setup.c +++ b/drivers/gpu/drm/i915/display/intel_modeset_setup.c @@ -10,6 +10,7 @@ #include <drm/drm_atomic_state_helper.h> #include "i915_drv.h" +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_bw.h" #include "intel_color.h" diff --git a/drivers/gpu/drm/i915/display/intel_panel.c b/drivers/gpu/drm/i915/display/intel_panel.c index 69ce77711b7c..1640726bfbf6 100644 --- a/drivers/gpu/drm/i915/display/intel_panel.c +++ b/drivers/gpu/drm/i915/display/intel_panel.c @@ -31,6 +31,7 @@ #include <linux/kernel.h> #include <linux/pwm.h> +#include "i915_reg.h" #include "intel_backlight.h" #include "intel_connector.h" #include "intel_de.h" diff --git a/drivers/gpu/drm/i915/display/intel_pch_display.c b/drivers/gpu/drm/i915/display/intel_pch_display.c index 837152dca063..cecc0d007cf3 100644 --- a/drivers/gpu/drm/i915/display/intel_pch_display.c +++ b/drivers/gpu/drm/i915/display/intel_pch_display.c @@ -4,6 +4,7 @@ */ #include "g4x_dp.h" +#include "i915_reg.h" #include "intel_crt.h" #include "intel_de.h" #include "intel_display_types.h" diff --git a/drivers/gpu/drm/i915/display/intel_pch_refclk.c b/drivers/gpu/drm/i915/display/intel_pch_refclk.c index a66097cdc1e0..08a94365b7d1 100644 --- a/drivers/gpu/drm/i915/display/intel_pch_refclk.c +++ b/drivers/gpu/drm/i915/display/intel_pch_refclk.c @@ -3,6 +3,7 @@ * Copyright © 2021 Intel Corporation */ +#include "i915_reg.h" #include "intel_de.h" #include "intel_display_types.h" #include "intel_panel.h" diff --git a/drivers/gpu/drm/i915/display/intel_pipe_crc.c b/drivers/gpu/drm/i915/display/intel_pipe_crc.c index 1c74388c60d7..e9774670e3f6 100644 --- a/drivers/gpu/drm/i915/display/intel_pipe_crc.c +++ b/drivers/gpu/drm/i915/display/intel_pipe_crc.c @@ -24,11 +24,12 @@ * */ -#include <linux/circ_buf.h> #include <linux/ctype.h> #include <linux/debugfs.h> #include <linux/seq_file.h> +#include "i915_irq.h" +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_de.h" #include "intel_display_types.h" diff --git a/drivers/gpu/drm/i915/display/intel_pps.c b/drivers/gpu/drm/i915/display/intel_pps.c index 21944f5bf3a8..9bbf41a076f7 100644 --- a/drivers/gpu/drm/i915/display/intel_pps.c +++ b/drivers/gpu/drm/i915/display/intel_pps.c @@ -5,10 +5,12 @@ #include "g4x_dp.h" #include "i915_drv.h" +#include "i915_reg.h" #include "intel_de.h" #include "intel_display_power_well.h" #include "intel_display_types.h" #include "intel_dp.h" +#include "intel_dpio_phy.h" #include "intel_dpll.h" #include "intel_lvds.h" #include "intel_pps.h" @@ -1098,7 +1100,13 @@ bool intel_pps_have_panel_power_or_vdd(struct intel_dp *intel_dp) static void pps_init_timestamps(struct intel_dp *intel_dp) { - intel_dp->pps.panel_power_off_time = ktime_get_boottime(); + /* + * Initialize panel power off time to 0, assuming panel power could have + * been toggled between kernel boot and now only by a previously loaded + * and removed i915, which has already ensured sufficient power off + * delay at module remove. + */ + intel_dp->pps.panel_power_off_time = 0; intel_dp->pps.last_power_on = jiffies; intel_dp->pps.last_backlight_off = jiffies; } diff --git a/drivers/gpu/drm/i915/display/intel_psr.c b/drivers/gpu/drm/i915/display/intel_psr.c index 904a1049eff3..5b678916e6db 100644 --- a/drivers/gpu/drm/i915/display/intel_psr.c +++ b/drivers/gpu/drm/i915/display/intel_psr.c @@ -27,6 +27,7 @@ #include "display/intel_dp.h" #include "i915_drv.h" +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_crtc.h" #include "intel_de.h" @@ -779,6 +780,7 @@ static bool psr2_granularity_check(struct intel_dp *intel_dp, struct intel_crtc_state *crtc_state) { struct drm_i915_private *dev_priv = dp_to_i915(intel_dp); + const struct drm_dsc_config *vdsc_cfg = &crtc_state->dsc.config; const int crtc_hdisplay = crtc_state->hw.adjusted_mode.crtc_hdisplay; const int crtc_vdisplay = crtc_state->hw.adjusted_mode.crtc_vdisplay; u16 y_granularity = 0; @@ -809,6 +811,10 @@ static bool psr2_granularity_check(struct intel_dp *intel_dp, if (y_granularity == 0 || crtc_vdisplay % y_granularity) return false; + if (crtc_state->dsc.compression_enable && + vdsc_cfg->slice_height % y_granularity) + return false; + crtc_state->su_y_granularity = y_granularity; return true; } @@ -1470,7 +1476,8 @@ unlock: static u32 man_trk_ctl_enable_bit_get(struct drm_i915_private *dev_priv) { - return IS_ALDERLAKE_P(dev_priv) ? 0 : PSR2_MAN_TRK_CTL_ENABLE; + return IS_ALDERLAKE_P(dev_priv) || DISPLAY_VER(dev_priv) >= 14 ? 0 : + PSR2_MAN_TRK_CTL_ENABLE; } static u32 man_trk_ctl_single_full_frame_bit_get(struct drm_i915_private *dev_priv) @@ -1482,14 +1489,14 @@ static u32 man_trk_ctl_single_full_frame_bit_get(struct drm_i915_private *dev_pr static u32 man_trk_ctl_partial_frame_bit_get(struct drm_i915_private *dev_priv) { - return IS_ALDERLAKE_P(dev_priv) ? + return IS_ALDERLAKE_P(dev_priv) || DISPLAY_VER(dev_priv) >= 14 ? ADLP_PSR2_MAN_TRK_CTL_SF_PARTIAL_FRAME_UPDATE : PSR2_MAN_TRK_CTL_SF_PARTIAL_FRAME_UPDATE; } static u32 man_trk_ctl_continuos_full_frame(struct drm_i915_private *dev_priv) { - return IS_ALDERLAKE_P(dev_priv) ? + return IS_ALDERLAKE_P(dev_priv) || DISPLAY_VER(dev_priv) >= 14 ? ADLP_PSR2_MAN_TRK_CTL_SF_CONTINUOS_FULL_FRAME : PSR2_MAN_TRK_CTL_SF_CONTINUOS_FULL_FRAME; } @@ -1678,9 +1685,6 @@ static void intel_psr2_sel_fetch_pipe_alignment(const struct intel_crtc_state *c pipe_clip->y1 -= pipe_clip->y1 % y_alignment; if (pipe_clip->y2 % y_alignment) pipe_clip->y2 = ((pipe_clip->y2 / y_alignment) + 1) * y_alignment; - - if (IS_ALDERLAKE_P(dev_priv) && crtc_state->dsc.compression_enable) - drm_warn(&dev_priv->drm, "Missing PSR2 sel fetch alignment with DSC\n"); } /* @@ -2209,8 +2213,11 @@ static void _psr_invalidate_handle(struct intel_dp *intel_dp) if (intel_dp->psr.psr2_sel_fetch_enabled) { u32 val; - if (intel_dp->psr.psr2_sel_fetch_cff_enabled) + if (intel_dp->psr.psr2_sel_fetch_cff_enabled) { + /* Send one update otherwise lag is observed in screen */ + intel_de_write(dev_priv, CURSURFLIVE(intel_dp->psr.pipe), 0); return; + } val = man_trk_ctl_enable_bit_get(dev_priv) | man_trk_ctl_partial_frame_bit_get(dev_priv) | diff --git a/drivers/gpu/drm/i915/display/intel_sdvo.c b/drivers/gpu/drm/i915/display/intel_sdvo.c index 48b7b1aa37b2..329b9d9af667 100644 --- a/drivers/gpu/drm/i915/display/intel_sdvo.c +++ b/drivers/gpu/drm/i915/display/intel_sdvo.c @@ -37,6 +37,7 @@ #include <drm/drm_edid.h> #include "i915_drv.h" +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_connector.h" #include "intel_crtc.h" diff --git a/drivers/gpu/drm/i915/display/intel_snps_phy.c b/drivers/gpu/drm/i915/display/intel_snps_phy.c index 937cefd6f78f..c799e891f8b5 100644 --- a/drivers/gpu/drm/i915/display/intel_snps_phy.c +++ b/drivers/gpu/drm/i915/display/intel_snps_phy.c @@ -5,6 +5,7 @@ #include <linux/util_macros.h> +#include "i915_reg.h" #include "intel_ddi.h" #include "intel_ddi_buf_trans.h" #include "intel_de.h" diff --git a/drivers/gpu/drm/i915/display/intel_snps_phy_regs.h b/drivers/gpu/drm/i915/display/intel_snps_phy_regs.h index 0543465aaf14..a04d692169d4 100644 --- a/drivers/gpu/drm/i915/display/intel_snps_phy_regs.h +++ b/drivers/gpu/drm/i915/display/intel_snps_phy_regs.h @@ -6,7 +6,7 @@ #ifndef __INTEL_SNPS_PHY_REGS__ #define __INTEL_SNPS_PHY_REGS__ -#include "i915_reg_defs.h" +#include "intel_display_reg_defs.h" #define _SNPS_PHY_A_BASE 0x168000 #define _SNPS_PHY_B_BASE 0x169000 diff --git a/drivers/gpu/drm/i915/display/intel_sprite.c b/drivers/gpu/drm/i915/display/intel_sprite.c index 7649c50b5445..e6b4d24b9cd0 100644 --- a/drivers/gpu/drm/i915/display/intel_sprite.c +++ b/drivers/gpu/drm/i915/display/intel_sprite.c @@ -42,6 +42,7 @@ #include <drm/drm_rect.h> #include "i915_drv.h" +#include "i915_reg.h" #include "i915_vgpu.h" #include "i9xx_plane.h" #include "intel_atomic_plane.h" diff --git a/drivers/gpu/drm/i915/display/intel_sprite.h b/drivers/gpu/drm/i915/display/intel_sprite.h index 4f63e4967731..4635c7ad23f9 100644 --- a/drivers/gpu/drm/i915/display/intel_sprite.h +++ b/drivers/gpu/drm/i915/display/intel_sprite.h @@ -8,14 +8,13 @@ #include <linux/types.h> -#include "intel_display.h" - struct drm_device; struct drm_display_mode; struct drm_file; struct drm_i915_private; struct intel_crtc_state; struct intel_plane_state; +enum pipe; /* * FIXME: We should instead only take spinlocks once for the entire update @@ -34,12 +33,6 @@ int intel_sprite_set_colorkey_ioctl(struct drm_device *dev, void *data, int intel_plane_check_src_coordinates(struct intel_plane_state *plane_state); int chv_plane_check_rotation(const struct intel_plane_state *plane_state); -static inline u8 icl_hdr_plane_mask(void) -{ - return BIT(PLANE_PRIMARY) | - BIT(PLANE_SPRITE0) | BIT(PLANE_SPRITE1); -} - int ivb_plane_min_cdclk(const struct intel_crtc_state *crtc_state, const struct intel_plane_state *plane_state); int hsw_plane_min_cdclk(const struct intel_crtc_state *crtc_state, diff --git a/drivers/gpu/drm/i915/display/intel_tv.c b/drivers/gpu/drm/i915/display/intel_tv.c index cf7d5c1ab406..4d2101ca1692 100644 --- a/drivers/gpu/drm/i915/display/intel_tv.c +++ b/drivers/gpu/drm/i915/display/intel_tv.c @@ -35,6 +35,8 @@ #include <drm/drm_edid.h> #include "i915_drv.h" +#include "i915_reg.h" +#include "i915_irq.h" #include "intel_connector.h" #include "intel_crtc.h" #include "intel_de.h" diff --git a/drivers/gpu/drm/i915/display/intel_vdsc.c b/drivers/gpu/drm/i915/display/intel_vdsc.c index 269f9792390d..9d3b77b41b5c 100644 --- a/drivers/gpu/drm/i915/display/intel_vdsc.c +++ b/drivers/gpu/drm/i915/display/intel_vdsc.c @@ -10,6 +10,7 @@ #include <drm/display/drm_dsc_helper.h> #include "i915_drv.h" +#include "i915_reg.h" #include "intel_crtc.h" #include "intel_de.h" #include "intel_display_types.h" diff --git a/drivers/gpu/drm/i915/display/intel_vga.c b/drivers/gpu/drm/i915/display/intel_vga.c index b5d058404c14..a69bfcac9a94 100644 --- a/drivers/gpu/drm/i915/display/intel_vga.c +++ b/drivers/gpu/drm/i915/display/intel_vga.c @@ -10,6 +10,7 @@ #include <video/vga.h> #include "i915_drv.h" +#include "i915_reg.h" #include "intel_de.h" #include "intel_vga.h" diff --git a/drivers/gpu/drm/i915/display/intel_vrr.c b/drivers/gpu/drm/i915/display/intel_vrr.c index 5eac99021875..7b1357e82b69 100644 --- a/drivers/gpu/drm/i915/display/intel_vrr.c +++ b/drivers/gpu/drm/i915/display/intel_vrr.c @@ -5,6 +5,7 @@ */ #include "i915_drv.h" +#include "i915_reg.h" #include "intel_de.h" #include "intel_display_types.h" #include "intel_vrr.h" diff --git a/drivers/gpu/drm/i915/display/skl_scaler.c b/drivers/gpu/drm/i915/display/skl_scaler.c index 4092679be21e..d7390067b7d4 100644 --- a/drivers/gpu/drm/i915/display/skl_scaler.c +++ b/drivers/gpu/drm/i915/display/skl_scaler.c @@ -2,6 +2,8 @@ /* * Copyright © 2020 Intel Corporation */ + +#include "i915_reg.h" #include "intel_de.h" #include "intel_display_types.h" #include "intel_fb.h" diff --git a/drivers/gpu/drm/i915/display/skl_universal_plane.c b/drivers/gpu/drm/i915/display/skl_universal_plane.c index 7cb713043408..76490cc59d8f 100644 --- a/drivers/gpu/drm/i915/display/skl_universal_plane.c +++ b/drivers/gpu/drm/i915/display/skl_universal_plane.c @@ -9,6 +9,8 @@ #include <drm/drm_fourcc.h> #include "i915_drv.h" +#include "i915_irq.h" +#include "i915_reg.h" #include "intel_atomic_plane.h" #include "intel_de.h" #include "intel_display_types.h" @@ -246,6 +248,11 @@ bool icl_is_nv12_y_plane(struct drm_i915_private *dev_priv, icl_nv12_y_plane_mask(dev_priv) & BIT(plane_id); } +u8 icl_hdr_plane_mask(void) +{ + return BIT(PLANE_PRIMARY) | BIT(PLANE_SPRITE0) | BIT(PLANE_SPRITE1); +} + bool icl_is_hdr_plane(struct drm_i915_private *dev_priv, enum plane_id plane_id) { return DISPLAY_VER(dev_priv) >= 11 && diff --git a/drivers/gpu/drm/i915/display/skl_universal_plane.h b/drivers/gpu/drm/i915/display/skl_universal_plane.h index 351040b64dc7..be64c201f9b3 100644 --- a/drivers/gpu/drm/i915/display/skl_universal_plane.h +++ b/drivers/gpu/drm/i915/display/skl_universal_plane.h @@ -30,6 +30,7 @@ int skl_calc_main_surface_offset(const struct intel_plane_state *plane_state, bool icl_is_nv12_y_plane(struct drm_i915_private *dev_priv, enum plane_id plane_id); +u8 icl_hdr_plane_mask(void); bool icl_is_hdr_plane(struct drm_i915_private *dev_priv, enum plane_id plane_id); #endif diff --git a/drivers/gpu/drm/i915/display/skl_watermark.c b/drivers/gpu/drm/i915/display/skl_watermark.c index d58e667016e4..e0766d1be966 100644 --- a/drivers/gpu/drm/i915/display/skl_watermark.c +++ b/drivers/gpu/drm/i915/display/skl_watermark.c @@ -2744,7 +2744,7 @@ static int skl_wm_add_affected_planes(struct intel_atomic_state *state, * power well the hardware state will go out of sync * with the software state. */ - if (!drm_atomic_crtc_needs_modeset(&new_crtc_state->uapi) && + if (!intel_crtc_needs_modeset(new_crtc_state) && skl_plane_selected_wm_equals(plane, &old_crtc_state->wm.skl.optimal, &new_crtc_state->wm.skl.optimal)) diff --git a/drivers/gpu/drm/i915/display/vlv_dsi.c b/drivers/gpu/drm/i915/display/vlv_dsi.c index 5a741ea4505f..84481030883a 100644 --- a/drivers/gpu/drm/i915/display/vlv_dsi.c +++ b/drivers/gpu/drm/i915/display/vlv_dsi.c @@ -31,6 +31,7 @@ #include <drm/drm_mipi_dsi.h> #include "i915_drv.h" +#include "i915_reg.h" #include "intel_atomic.h" #include "intel_backlight.h" #include "intel_connector.h" diff --git a/drivers/gpu/drm/i915/display/vlv_dsi_regs.h b/drivers/gpu/drm/i915/display/vlv_dsi_regs.h index e065b8f2ee08..abbe427e462e 100644 --- a/drivers/gpu/drm/i915/display/vlv_dsi_regs.h +++ b/drivers/gpu/drm/i915/display/vlv_dsi_regs.h @@ -6,7 +6,7 @@ #ifndef __VLV_DSI_REGS_H__ #define __VLV_DSI_REGS_H__ -#include "i915_reg_defs.h" +#include "intel_display_reg_defs.h" #define VLV_MIPI_BASE VLV_DISPLAY_BASE #define BXT_MIPI_BASE 0x60000 diff --git a/drivers/gpu/drm/i915/gem/i915_gem_context.c b/drivers/gpu/drm/i915/gem/i915_gem_context.c index 01402f3c58f6..7f2831efc798 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_context.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_context.c @@ -546,7 +546,7 @@ set_proto_ctx_engines_bond(struct i915_user_extension __user *base, void *data) } if (intel_engine_uses_guc(master)) { - DRM_DEBUG("bonding extension not supported with GuC submission"); + drm_dbg(&i915->drm, "bonding extension not supported with GuC submission"); return -ENODEV; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c b/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c index 9322ac29008b..fd556a076d05 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_dmabuf.c @@ -240,7 +240,6 @@ static int i915_gem_object_get_pages_dmabuf(struct drm_i915_gem_object *obj) { struct drm_i915_private *i915 = to_i915(obj->base.dev); struct sg_table *sgt; - unsigned int sg_page_sizes; assert_object_held(obj); @@ -264,8 +263,7 @@ static int i915_gem_object_get_pages_dmabuf(struct drm_i915_gem_object *obj) (!HAS_LLC(i915) && !IS_DG1(i915))) wbinvd_on_all_cpus(); - sg_page_sizes = i915_sg_dma_sizes(sgt->sgl); - __i915_gem_object_set_pages(obj, sgt, sg_page_sizes); + __i915_gem_object_set_pages(obj, sgt); return 0; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 1160723c9d2d..29e9e8d5b6fe 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -30,6 +30,7 @@ #include "i915_gem_context.h" #include "i915_gem_evict.h" #include "i915_gem_ioctls.h" +#include "i915_reg.h" #include "i915_trace.h" #include "i915_user_extensions.h" @@ -53,13 +54,13 @@ enum { #define DBG_FORCE_RELOC 0 /* choose one of the above! */ }; -/* __EXEC_OBJECT_NO_RESERVE is BIT(31), defined in i915_vma.h */ -#define __EXEC_OBJECT_HAS_PIN BIT(30) -#define __EXEC_OBJECT_HAS_FENCE BIT(29) -#define __EXEC_OBJECT_USERPTR_INIT BIT(28) -#define __EXEC_OBJECT_NEEDS_MAP BIT(27) -#define __EXEC_OBJECT_NEEDS_BIAS BIT(26) -#define __EXEC_OBJECT_INTERNAL_FLAGS (~0u << 26) /* all of the above + */ +/* __EXEC_OBJECT_ flags > BIT(29) defined in i915_vma.h */ +#define __EXEC_OBJECT_HAS_PIN BIT(29) +#define __EXEC_OBJECT_HAS_FENCE BIT(28) +#define __EXEC_OBJECT_USERPTR_INIT BIT(27) +#define __EXEC_OBJECT_NEEDS_MAP BIT(26) +#define __EXEC_OBJECT_NEEDS_BIAS BIT(25) +#define __EXEC_OBJECT_INTERNAL_FLAGS (~0u << 25) /* all of the above + */ #define __EXEC_OBJECT_RESERVED (__EXEC_OBJECT_HAS_PIN | __EXEC_OBJECT_HAS_FENCE) #define __EXEC_HAS_RELOC BIT(31) @@ -2101,7 +2102,8 @@ static int eb_move_to_gpu(struct i915_execbuffer *eb) eb->composite_fence ? eb->composite_fence : &eb->requests[j]->fence, - flags | __EXEC_OBJECT_NO_RESERVE); + flags | __EXEC_OBJECT_NO_RESERVE | + __EXEC_OBJECT_NO_REQUEST_AWAIT); } } @@ -2148,7 +2150,8 @@ err_skip: return err; } -static int i915_gem_check_execbuffer(struct drm_i915_gem_execbuffer2 *exec) +static int i915_gem_check_execbuffer(struct drm_i915_private *i915, + struct drm_i915_gem_execbuffer2 *exec) { if (exec->flags & __I915_EXEC_ILLEGAL_FLAGS) return -EINVAL; @@ -2161,7 +2164,7 @@ static int i915_gem_check_execbuffer(struct drm_i915_gem_execbuffer2 *exec) } if (exec->DR4 == 0xffffffff) { - DRM_DEBUG("UXA submitting garbage DR4, fixing up\n"); + drm_dbg(&i915->drm, "UXA submitting garbage DR4, fixing up\n"); exec->DR4 = 0; } if (exec->DR1 || exec->DR4) @@ -2799,7 +2802,8 @@ add_timeline_fence_array(struct i915_execbuffer *eb, syncobj = drm_syncobj_find(eb->file, user_fence.handle); if (!syncobj) { - DRM_DEBUG("Invalid syncobj handle provided\n"); + drm_dbg(&eb->i915->drm, + "Invalid syncobj handle provided\n"); return -ENOENT; } @@ -2807,7 +2811,8 @@ add_timeline_fence_array(struct i915_execbuffer *eb, if (!fence && user_fence.flags && !(user_fence.flags & I915_EXEC_FENCE_SIGNAL)) { - DRM_DEBUG("Syncobj handle has no fence\n"); + drm_dbg(&eb->i915->drm, + "Syncobj handle has no fence\n"); drm_syncobj_put(syncobj); return -EINVAL; } @@ -2816,7 +2821,9 @@ add_timeline_fence_array(struct i915_execbuffer *eb, err = dma_fence_chain_find_seqno(&fence, point); if (err && !(user_fence.flags & I915_EXEC_FENCE_SIGNAL)) { - DRM_DEBUG("Syncobj handle missing requested point %llu\n", point); + drm_dbg(&eb->i915->drm, + "Syncobj handle missing requested point %llu\n", + point); dma_fence_put(fence); drm_syncobj_put(syncobj); return err; @@ -2842,7 +2849,8 @@ add_timeline_fence_array(struct i915_execbuffer *eb, * 0) would break the timeline. */ if (user_fence.flags & I915_EXEC_FENCE_WAIT) { - DRM_DEBUG("Trying to wait & signal the same timeline point.\n"); + drm_dbg(&eb->i915->drm, + "Trying to wait & signal the same timeline point.\n"); dma_fence_put(fence); drm_syncobj_put(syncobj); return -EINVAL; @@ -2913,14 +2921,16 @@ static int add_fence_array(struct i915_execbuffer *eb) syncobj = drm_syncobj_find(eb->file, user_fence.handle); if (!syncobj) { - DRM_DEBUG("Invalid syncobj handle provided\n"); + drm_dbg(&eb->i915->drm, + "Invalid syncobj handle provided\n"); return -ENOENT; } if (user_fence.flags & I915_EXEC_FENCE_WAIT) { fence = drm_syncobj_fence_get(syncobj); if (!fence) { - DRM_DEBUG("Syncobj handle has no fence\n"); + drm_dbg(&eb->i915->drm, + "Syncobj handle has no fence\n"); drm_syncobj_put(syncobj); return -EINVAL; } @@ -3515,7 +3525,7 @@ i915_gem_execbuffer2_ioctl(struct drm_device *dev, void *data, return -EINVAL; } - err = i915_gem_check_execbuffer(args); + err = i915_gem_check_execbuffer(i915, args); if (err) return err; diff --git a/drivers/gpu/drm/i915/gem/i915_gem_internal.c b/drivers/gpu/drm/i915/gem/i915_gem_internal.c index 629acb403a2c..f66bcefc09ec 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_internal.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_internal.c @@ -35,7 +35,6 @@ static int i915_gem_object_get_pages_internal(struct drm_i915_gem_object *obj) struct drm_i915_private *i915 = to_i915(obj->base.dev); struct sg_table *st; struct scatterlist *sg; - unsigned int sg_page_sizes; unsigned int npages; int max_order = MAX_ORDER; unsigned int max_segment; @@ -64,7 +63,6 @@ create_st: sg = st->sgl; st->nents = 0; - sg_page_sizes = 0; do { int order = min(fls(npages) - 1, max_order); @@ -83,7 +81,6 @@ create_st: } while (1); sg_set_page(sg, page, PAGE_SIZE << order, 0); - sg_page_sizes |= PAGE_SIZE << order; st->nents++; npages -= 1 << order; @@ -105,7 +102,7 @@ create_st: goto err; } - __i915_gem_object_set_pages(obj, st, sg_page_sizes); + __i915_gem_object_set_pages(obj, st); return 0; diff --git a/drivers/gpu/drm/i915/gem/i915_gem_mman.c b/drivers/gpu/drm/i915/gem/i915_gem_mman.c index e63329bc8065..c29efdef8313 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_mman.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_mman.c @@ -330,7 +330,7 @@ retry: if (ret) goto err_rpm; - ret = intel_gt_reset_trylock(ggtt->vm.gt, &srcu); + ret = intel_gt_reset_lock_interruptible(ggtt->vm.gt, &srcu); if (ret) goto err_pages; diff --git a/drivers/gpu/drm/i915/gem/i915_gem_object.h b/drivers/gpu/drm/i915/gem/i915_gem_object.h index 6b9ecff42bb5..3db53769864c 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_object.h +++ b/drivers/gpu/drm/i915/gem/i915_gem_object.h @@ -403,8 +403,7 @@ i915_gem_object_get_dma_address(struct drm_i915_gem_object *obj, unsigned long n); void __i915_gem_object_set_pages(struct drm_i915_gem_object *obj, - struct sg_table *pages, - unsigned int sg_page_sizes); + struct sg_table *pages); int ____i915_gem_object_get_pages(struct drm_i915_gem_object *obj); int __i915_gem_object_get_pages(struct drm_i915_gem_object *obj); diff --git a/drivers/gpu/drm/i915/gem/i915_gem_pages.c b/drivers/gpu/drm/i915/gem/i915_gem_pages.c index 16f845663ff2..05a27723ebb8 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_pages.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_pages.c @@ -16,8 +16,7 @@ #include "i915_gem_mman.h" void __i915_gem_object_set_pages(struct drm_i915_gem_object *obj, - struct sg_table *pages, - unsigned int sg_page_sizes) + struct sg_table *pages) { struct drm_i915_private *i915 = to_i915(obj->base.dev); unsigned long supported = RUNTIME_INFO(i915)->page_sizes; @@ -45,8 +44,8 @@ void __i915_gem_object_set_pages(struct drm_i915_gem_object *obj, obj->mm.pages = pages; - GEM_BUG_ON(!sg_page_sizes); - obj->mm.page_sizes.phys = sg_page_sizes; + obj->mm.page_sizes.phys = i915_sg_dma_sizes(pages->sgl); + GEM_BUG_ON(!obj->mm.page_sizes.phys); /* * Calculate the supported page-sizes which fit into the given diff --git a/drivers/gpu/drm/i915/gem/i915_gem_phys.c b/drivers/gpu/drm/i915/gem/i915_gem_phys.c index 0d0e46dae559..68453572275b 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_phys.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_phys.c @@ -79,7 +79,7 @@ static int i915_gem_object_get_pages_phys(struct drm_i915_gem_object *obj) /* We're no longer struct page backed */ obj->mem_flags &= ~I915_BO_FLAG_STRUCT_PAGE; - __i915_gem_object_set_pages(obj, st, sg->length); + __i915_gem_object_set_pages(obj, st); return 0; @@ -209,11 +209,8 @@ static int i915_gem_object_shmem_to_phys(struct drm_i915_gem_object *obj) return 0; err_xfer: - if (!IS_ERR_OR_NULL(pages)) { - unsigned int sg_page_sizes = i915_sg_dma_sizes(pages->sgl); - - __i915_gem_object_set_pages(obj, pages, sg_page_sizes); - } + if (!IS_ERR_OR_NULL(pages)) + __i915_gem_object_set_pages(obj, pages); return err; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_shmem.c b/drivers/gpu/drm/i915/gem/i915_gem_shmem.c index 2f7804492cd5..9c759df700ca 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_shmem.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_shmem.c @@ -247,7 +247,7 @@ rebuild_st: if (i915_gem_object_can_bypass_llc(obj)) obj->cache_dirty = true; - __i915_gem_object_set_pages(obj, st, i915_sg_dma_sizes(st->sgl)); + __i915_gem_object_set_pages(obj, st); return 0; diff --git a/drivers/gpu/drm/i915/gem/i915_gem_stolen.c b/drivers/gpu/drm/i915/gem/i915_gem_stolen.c index 0c70711818ed..bc9521078807 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_stolen.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_stolen.c @@ -628,7 +628,7 @@ static int i915_gem_object_get_pages_stolen(struct drm_i915_gem_object *obj) sg_dma_len(pages->sgl), POISON_INUSE); - __i915_gem_object_set_pages(obj, pages, obj->stolen->size); + __i915_gem_object_set_pages(obj, pages); return 0; } diff --git a/drivers/gpu/drm/i915/gem/i915_gem_ttm.c b/drivers/gpu/drm/i915/gem/i915_gem_ttm.c index 25129af70f70..1e50fb0d6bfc 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_ttm.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_ttm.c @@ -603,6 +603,10 @@ static int i915_ttm_truncate(struct drm_i915_gem_object *obj) WARN_ON_ONCE(obj->mm.madv == I915_MADV_WILLNEED); + err = ttm_bo_wait(bo, true, false); + if (err) + return err; + err = i915_ttm_move_notify(bo); if (err) return err; @@ -815,8 +819,7 @@ static int __i915_ttm_get_pages(struct drm_i915_gem_object *obj, GEM_BUG_ON(obj->mm.rsgt); obj->mm.rsgt = rsgt; - __i915_gem_object_set_pages(obj, &rsgt->table, - i915_sg_dma_sizes(rsgt->table.sgl)); + __i915_gem_object_set_pages(obj, &rsgt->table); } GEM_BUG_ON(bo->ttm && ((obj->base.size >> PAGE_SHIFT) < bo->ttm->num_pages)); @@ -1031,9 +1034,6 @@ static vm_fault_t vm_fault_ttm(struct vm_fault *vmf) vm_fault_t ret; int idx; - if (i915_ttm_is_ghost_object(bo)) - return VM_FAULT_SIGBUS; - /* Sanity check that we allow writing into this object */ if (unlikely(i915_gem_object_is_readonly(obj) && area->vm_flags & VM_WRITE)) @@ -1048,9 +1048,6 @@ static vm_fault_t vm_fault_ttm(struct vm_fault *vmf) return VM_FAULT_SIGBUS; } - if (i915_ttm_cpu_maps_iomem(bo->resource)) - wakeref = intel_runtime_pm_get(&to_i915(obj->base.dev)->runtime_pm); - if (!i915_ttm_resource_mappable(bo->resource)) { int err = -ENODEV; int i; @@ -1078,6 +1075,9 @@ static vm_fault_t vm_fault_ttm(struct vm_fault *vmf) } } + if (i915_ttm_cpu_maps_iomem(bo->resource)) + wakeref = intel_runtime_pm_get(&to_i915(obj->base.dev)->runtime_pm); + if (drm_dev_enter(dev, &idx)) { ret = ttm_bo_vm_fault_reserved(vmf, vmf->vma->vm_page_prot, TTM_BO_VM_NUM_PREFAULT); @@ -1098,6 +1098,8 @@ static vm_fault_t vm_fault_ttm(struct vm_fault *vmf) spin_lock(&to_i915(obj->base.dev)->runtime_pm.lmem_userfault_lock); list_add(&obj->userfault_link, &to_i915(obj->base.dev)->runtime_pm.lmem_userfault_list); spin_unlock(&to_i915(obj->base.dev)->runtime_pm.lmem_userfault_lock); + + GEM_WARN_ON(!i915_ttm_cpu_maps_iomem(bo->resource)); } if (wakeref & CONFIG_DRM_I915_USERFAULT_AUTOSUSPEND) @@ -1180,6 +1182,8 @@ static void i915_ttm_unmap_virtual(struct drm_i915_gem_object *obj) } } + GEM_WARN_ON(obj->userfault_count); + ttm_bo_unmap_virtual(i915_gem_to_ttm(obj)); if (wakeref) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c index 1b1a22716722..9348b1804d53 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_userptr.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_userptr.c @@ -131,7 +131,6 @@ static int i915_gem_userptr_get_pages(struct drm_i915_gem_object *obj) const unsigned long num_pages = obj->base.size >> PAGE_SHIFT; unsigned int max_segment = i915_sg_segment_size(obj->base.dev->dev); struct sg_table *st; - unsigned int sg_page_sizes; struct page **pvec; int ret; @@ -170,8 +169,7 @@ alloc_table: if (i915_gem_object_can_bypass_llc(obj)) obj->cache_dirty = true; - sg_page_sizes = i915_sg_dma_sizes(st->sgl); - __i915_gem_object_set_pages(obj, st, sg_page_sizes); + __i915_gem_object_set_pages(obj, st); return 0; @@ -427,9 +425,10 @@ probe_range(struct mm_struct *mm, unsigned long addr, unsigned long len) { VMA_ITERATOR(vmi, mm, addr); struct vm_area_struct *vma; + unsigned long end = addr + len; mmap_read_lock(mm); - for_each_vma_range(vmi, vma, addr + len) { + for_each_vma_range(vmi, vma, end) { /* Check for holes, note that we also update the addr below */ if (vma->vm_start > addr) break; @@ -441,7 +440,7 @@ probe_range(struct mm_struct *mm, unsigned long addr, unsigned long len) } mmap_read_unlock(mm); - if (vma) + if (vma || addr < end) return -EFAULT; return 0; } diff --git a/drivers/gpu/drm/i915/gem/selftests/huge_gem_object.c b/drivers/gpu/drm/i915/gem/selftests/huge_gem_object.c index f963b8e1e37b..cbd9b624a788 100644 --- a/drivers/gpu/drm/i915/gem/selftests/huge_gem_object.c +++ b/drivers/gpu/drm/i915/gem/selftests/huge_gem_object.c @@ -68,7 +68,7 @@ static int huge_get_pages(struct drm_i915_gem_object *obj) if (i915_gem_gtt_prepare_pages(obj, pages)) goto err; - __i915_gem_object_set_pages(obj, pages, PAGE_SIZE); + __i915_gem_object_set_pages(obj, pages); return 0; diff --git a/drivers/gpu/drm/i915/gem/selftests/huge_pages.c b/drivers/gpu/drm/i915/gem/selftests/huge_pages.c index 0cb99e75b0bc..beaf27e09e8a 100644 --- a/drivers/gpu/drm/i915/gem/selftests/huge_pages.c +++ b/drivers/gpu/drm/i915/gem/selftests/huge_pages.c @@ -136,7 +136,7 @@ static int get_huge_pages(struct drm_i915_gem_object *obj) goto err; GEM_BUG_ON(sg_page_sizes != obj->mm.page_mask); - __i915_gem_object_set_pages(obj, st, sg_page_sizes); + __i915_gem_object_set_pages(obj, st); return 0; @@ -210,7 +210,6 @@ static int fake_get_huge_pages(struct drm_i915_gem_object *obj) const u64 max_len = rounddown_pow_of_two(UINT_MAX); struct sg_table *st; struct scatterlist *sg; - unsigned int sg_page_sizes; u64 rem; st = kmalloc(sizeof(*st), GFP); @@ -226,7 +225,6 @@ static int fake_get_huge_pages(struct drm_i915_gem_object *obj) rem = obj->base.size; sg = st->sgl; st->nents = 0; - sg_page_sizes = 0; do { unsigned int page_size = get_largest_page_size(i915, rem); unsigned int len = min(page_size * div_u64(rem, page_size), @@ -239,8 +237,6 @@ static int fake_get_huge_pages(struct drm_i915_gem_object *obj) sg_dma_len(sg) = len; sg_dma_address(sg) = page_size; - sg_page_sizes |= len; - st->nents++; rem -= len; @@ -254,7 +250,7 @@ static int fake_get_huge_pages(struct drm_i915_gem_object *obj) i915_sg_trim(st); - __i915_gem_object_set_pages(obj, st, sg_page_sizes); + __i915_gem_object_set_pages(obj, st); return 0; } @@ -286,7 +282,7 @@ static int fake_get_huge_pages_single(struct drm_i915_gem_object *obj) sg_dma_len(sg) = obj->base.size; sg_dma_address(sg) = page_size; - __i915_gem_object_set_pages(obj, st, sg->length); + __i915_gem_object_set_pages(obj, st); return 0; #undef GFP diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c index 9a6a6b5b722b..692a16914ca0 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_client_blt.c @@ -13,6 +13,7 @@ #include "gt/intel_gt_regs.h" #include "gem/i915_gem_lmem.h" +#include "gem/selftests/igt_gem_utils.h" #include "selftests/igt_flush_test.h" #include "selftests/mock_drm.h" #include "selftests/i915_random.h" @@ -457,21 +458,6 @@ static int verify_buffer(const struct tiled_blits *t, return ret; } -static int move_to_active(struct i915_vma *vma, - struct i915_request *rq, - unsigned int flags) -{ - int err; - - i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, false); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, flags); - i915_vma_unlock(vma); - - return err; -} - static int pin_buffer(struct i915_vma *vma, u64 addr) { int err; @@ -525,11 +511,11 @@ tiled_blit(struct tiled_blits *t, goto err_bb; } - err = move_to_active(t->batch, rq, 0); + err = igt_vma_move_to_active_unlocked(t->batch, rq, 0); if (!err) - err = move_to_active(src->vma, rq, 0); + err = igt_vma_move_to_active_unlocked(src->vma, rq, 0); if (!err) - err = move_to_active(dst->vma, rq, 0); + err = igt_vma_move_to_active_unlocked(dst->vma, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, t->batch->node.start, diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c index a666d7e610f5..c228fe4aba50 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_coherency.c @@ -239,9 +239,7 @@ static int gpu_set(struct context *ctx, unsigned long offset, u32 v) } intel_ring_advance(rq, cs); - err = i915_request_await_object(rq, vma->obj, true); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); out_rq: i915_request_add(rq); diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c index d8864444432b..a0ff51d71d07 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_context.c @@ -984,15 +984,11 @@ retry: goto err_batch; } - err = i915_request_await_object(rq, batch->obj, false); - if (err == 0) - err = i915_vma_move_to_active(batch, rq, 0); + err = i915_vma_move_to_active(batch, rq, 0); if (err) goto skip_request; - err = i915_request_await_object(rq, vma->obj, true); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); if (err) goto skip_request; @@ -1553,9 +1549,7 @@ static int write_to_scratch(struct i915_gem_context *ctx, } i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, false); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, 0); + err = i915_vma_move_to_active(vma, rq, 0); i915_vma_unlock(vma); if (err) goto skip_request; @@ -1689,9 +1683,7 @@ static int read_from_scratch(struct i915_gem_context *ctx, } i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, true); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); i915_vma_unlock(vma); if (err) goto skip_request; diff --git a/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c b/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c index 1cae24349a96..3f658d5717d8 100644 --- a/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c +++ b/drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c @@ -17,6 +17,7 @@ #include "gt/intel_gt.h" #include "gt/intel_gt_pm.h" #include "gt/intel_migrate.h" +#include "i915_reg.h" #include "i915_ttm_buddy_manager.h" #include "huge_gem_object.h" @@ -565,10 +566,8 @@ retry: goto err_unpin; } - err = i915_request_await_object(rq, vma->obj, true); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, - EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(vma, rq, + EXEC_OBJECT_WRITE); i915_request_add(rq); err_unpin: @@ -1608,9 +1607,7 @@ retry: goto out_unpin; } - err = i915_request_await_object(rq, vma->obj, false); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, 0); + err = i915_vma_move_to_active(vma, rq, 0); err = engine->emit_bb_start(rq, vma->node.start, 0, 0); i915_request_get(rq); diff --git a/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.c b/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.c index 3c55e77b0f1b..374b10ac430e 100644 --- a/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.c +++ b/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.c @@ -131,17 +131,13 @@ int igt_gpu_fill_dw(struct intel_context *ce, } i915_vma_lock(batch); - err = i915_request_await_object(rq, batch->obj, false); - if (err == 0) - err = i915_vma_move_to_active(batch, rq, 0); + err = i915_vma_move_to_active(batch, rq, 0); i915_vma_unlock(batch); if (err) goto skip_request; i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, true); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); i915_vma_unlock(vma); if (err) goto skip_request; diff --git a/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.h b/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.h index 4221cf84d175..1379fbc14431 100644 --- a/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.h +++ b/drivers/gpu/drm/i915/gem/selftests/igt_gem_utils.h @@ -9,6 +9,8 @@ #include <linux/types.h> +#include "i915_vma.h" + struct i915_request; struct i915_gem_context; struct i915_vma; @@ -29,4 +31,16 @@ int igt_gpu_fill_dw(struct intel_context *ce, struct i915_vma *vma, u64 offset, unsigned long count, u32 val); +static inline int __must_check +igt_vma_move_to_active_unlocked(struct i915_vma *vma, struct i915_request *rq, + unsigned int flags) +{ + int err; + + i915_vma_lock(vma); + err = _i915_vma_move_to_active(vma, rq, &rq->fence, flags); + i915_vma_unlock(vma); + return err; +} + #endif /* __IGT_GEM_UTILS_H__ */ diff --git a/drivers/gpu/drm/i915/gt/intel_engine_cs.c b/drivers/gpu/drm/i915/gt/intel_engine_cs.c index 3b7d750ad054..c33e0d72d670 100644 --- a/drivers/gpu/drm/i915/gt/intel_engine_cs.c +++ b/drivers/gpu/drm/i915/gt/intel_engine_cs.c @@ -13,6 +13,8 @@ #include "i915_cmd_parser.h" #include "i915_drv.h" +#include "i915_irq.h" +#include "i915_reg.h" #include "intel_breadcrumbs.h" #include "intel_context.h" #include "intel_engine.h" @@ -244,6 +246,13 @@ static const struct engine_info intel_engines[] = { { .graphics_ver = 12, .base = GEN12_COMPUTE3_RING_BASE } } }, + [GSC0] = { + .class = OTHER_CLASS, + .instance = OTHER_GSC_INSTANCE, + .mmio_bases = { + { .graphics_ver = 12, .base = MTL_GSC_RING_BASE } + } + }, }; /** @@ -324,6 +333,7 @@ u32 intel_engine_context_size(struct intel_gt *gt, u8 class) case VIDEO_DECODE_CLASS: case VIDEO_ENHANCEMENT_CLASS: case COPY_ENGINE_CLASS: + case OTHER_CLASS: if (GRAPHICS_VER(gt->i915) < 8) return 0; return GEN8_LR_CONTEXT_OTHER_SIZE; @@ -415,6 +425,7 @@ static u32 get_reset_domain(u8 ver, enum intel_engine_id id) [CCS1] = GEN11_GRDOM_RENDER, [CCS2] = GEN11_GRDOM_RENDER, [CCS3] = GEN11_GRDOM_RENDER, + [GSC0] = GEN12_GRDOM_GSC, }; GEM_BUG_ON(id >= ARRAY_SIZE(engine_reset_domains) || !engine_reset_domains[id]); diff --git a/drivers/gpu/drm/i915/gt/intel_engine_types.h b/drivers/gpu/drm/i915/gt/intel_engine_types.h index 6b5d4ea22b67..4fd54fb8810f 100644 --- a/drivers/gpu/drm/i915/gt/intel_engine_types.h +++ b/drivers/gpu/drm/i915/gt/intel_engine_types.h @@ -136,6 +136,7 @@ enum intel_engine_id { CCS2, CCS3, #define _CCS(n) (CCS0 + (n)) + GSC0, I915_NUM_ENGINES #define INVALID_ENGINE ((enum intel_engine_id)-1) }; diff --git a/drivers/gpu/drm/i915/gt/intel_engine_user.c b/drivers/gpu/drm/i915/gt/intel_engine_user.c index 46a174f8aa00..cd4f1b126f75 100644 --- a/drivers/gpu/drm/i915/gt/intel_engine_user.c +++ b/drivers/gpu/drm/i915/gt/intel_engine_user.c @@ -140,6 +140,7 @@ const char *intel_engine_class_repr(u8 class) [COPY_ENGINE_CLASS] = "bcs", [VIDEO_DECODE_CLASS] = "vcs", [VIDEO_ENHANCEMENT_CLASS] = "vecs", + [OTHER_CLASS] = "other", [COMPUTE_CLASS] = "ccs", }; @@ -190,6 +191,15 @@ static void add_legacy_ring(struct legacy_ring *ring, ring->instance++; } +static void engine_rename(struct intel_engine_cs *engine, const char *name, u16 instance) +{ + char old[sizeof(engine->name)]; + + memcpy(old, engine->name, sizeof(engine->name)); + scnprintf(engine->name, sizeof(engine->name), "%s%u", name, instance); + drm_dbg(&engine->i915->drm, "renamed %s to %s\n", old, engine->name); +} + void intel_engines_driver_register(struct drm_i915_private *i915) { struct legacy_ring ring = {}; @@ -205,11 +215,19 @@ void intel_engines_driver_register(struct drm_i915_private *i915) struct intel_engine_cs *engine = container_of((struct rb_node *)it, typeof(*engine), uabi_node); - char old[sizeof(engine->name)]; if (intel_gt_has_unrecoverable_error(engine->gt)) continue; /* ignore incomplete engines */ + /* + * We don't want to expose the GSC engine to the users, but we + * still rename it so it is easier to identify in the debug logs + */ + if (engine->id == GSC0) { + engine_rename(engine, "gsc", 0); + continue; + } + GEM_BUG_ON(engine->class >= ARRAY_SIZE(uabi_classes)); engine->uabi_class = uabi_classes[engine->class]; @@ -219,11 +237,9 @@ void intel_engines_driver_register(struct drm_i915_private *i915) i915->engine_uabi_class_count[engine->uabi_class]++; /* Replace the internal name with the final user facing name */ - memcpy(old, engine->name, sizeof(engine->name)); - scnprintf(engine->name, sizeof(engine->name), "%s%u", - intel_engine_class_repr(engine->class), - engine->uabi_instance); - DRM_DEBUG_DRIVER("renamed %s to %s\n", old, engine->name); + engine_rename(engine, + intel_engine_class_repr(engine->class), + engine->uabi_instance); rb_link_node(&engine->uabi_node, prev, p); rb_insert_color(&engine->uabi_node, &i915->uabi_engines); diff --git a/drivers/gpu/drm/i915/gt/intel_execlists_submission.c b/drivers/gpu/drm/i915/gt/intel_execlists_submission.c index 0187bc72310d..49a8f10d76c7 100644 --- a/drivers/gpu/drm/i915/gt/intel_execlists_submission.c +++ b/drivers/gpu/drm/i915/gt/intel_execlists_submission.c @@ -110,6 +110,7 @@ #include <linux/string_helpers.h> #include "i915_drv.h" +#include "i915_reg.h" #include "i915_trace.h" #include "i915_vgpu.h" #include "gen8_engine_cs.h" @@ -3921,6 +3922,7 @@ static struct intel_context * execlists_create_virtual(struct intel_engine_cs **siblings, unsigned int count, unsigned long flags) { + struct drm_i915_private *i915 = siblings[0]->i915; struct virtual_engine *ve; unsigned int n; int err; @@ -3929,7 +3931,7 @@ execlists_create_virtual(struct intel_engine_cs **siblings, unsigned int count, if (!ve) return ERR_PTR(-ENOMEM); - ve->base.i915 = siblings[0]->i915; + ve->base.i915 = i915; ve->base.gt = siblings[0]->gt; ve->base.uncore = siblings[0]->uncore; ve->base.id = -1; @@ -3988,8 +3990,9 @@ execlists_create_virtual(struct intel_engine_cs **siblings, unsigned int count, GEM_BUG_ON(!is_power_of_2(sibling->mask)); if (sibling->mask & ve->base.mask) { - DRM_DEBUG("duplicate %s entry in load balancer\n", - sibling->name); + drm_dbg(&i915->drm, + "duplicate %s entry in load balancer\n", + sibling->name); err = -EINVAL; goto err_put; } @@ -4023,8 +4026,9 @@ execlists_create_virtual(struct intel_engine_cs **siblings, unsigned int count, */ if (ve->base.class != OTHER_CLASS) { if (ve->base.class != sibling->class) { - DRM_DEBUG("invalid mixing of engine class, sibling %d, already %d\n", - sibling->class, ve->base.class); + drm_dbg(&i915->drm, + "invalid mixing of engine class, sibling %d, already %d\n", + sibling->class, ve->base.class); err = -EINVAL; goto err_put; } diff --git a/drivers/gpu/drm/i915/gt/intel_ggtt.c b/drivers/gpu/drm/i915/gt/intel_ggtt.c index 2518cebbf931..8145851ad23d 100644 --- a/drivers/gpu/drm/i915/gt/intel_ggtt.c +++ b/drivers/gpu/drm/i915/gt/intel_ggtt.c @@ -560,7 +560,7 @@ static int init_ggtt(struct i915_ggtt *ggtt) * why. */ ggtt->pin_bias = max_t(u32, I915_GTT_PAGE_SIZE, - intel_wopcm_guc_size(&ggtt->vm.i915->wopcm)); + intel_wopcm_guc_size(&ggtt->vm.gt->wopcm)); ret = intel_vgt_balloon(ggtt); if (ret) diff --git a/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c b/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c index ea775e601686..995082d45cb2 100644 --- a/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c +++ b/drivers/gpu/drm/i915/gt/intel_ggtt_fencing.c @@ -816,8 +816,8 @@ i915_gem_object_save_bit_17_swizzle(struct drm_i915_gem_object *obj, if (obj->bit_17 == NULL) { obj->bit_17 = bitmap_zalloc(page_count, GFP_KERNEL); if (obj->bit_17 == NULL) { - DRM_ERROR("Failed to allocate memory for bit 17 " - "record\n"); + drm_err(&to_i915(obj->base.dev)->drm, + "Failed to allocate memory for bit 17 record\n"); return; } } diff --git a/drivers/gpu/drm/i915/gt/intel_gt.c b/drivers/gpu/drm/i915/gt/intel_gt.c index 8e914c4066ed..0325f071046c 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt.c +++ b/drivers/gpu/drm/i915/gt/intel_gt.c @@ -12,6 +12,7 @@ #include "i915_drv.h" #include "i915_perf_oa_regs.h" +#include "i915_reg.h" #include "intel_context.h" #include "intel_engine_pm.h" #include "intel_engine_regs.h" @@ -54,6 +55,7 @@ void intel_gt_common_init_early(struct intel_gt *gt) seqcount_mutex_init(>->tlb.seqno, >->tlb.invalidate_lock); intel_gt_pm_init_early(gt); + intel_wopcm_init_early(>->wopcm); intel_uc_init_early(>->uc); intel_rps_init_early(>->rps); } @@ -190,7 +192,7 @@ int intel_gt_init_hw(struct intel_gt *gt) ret = i915_ppgtt_init_hw(gt); if (ret) { - DRM_ERROR("Enabling PPGTT failed (%d)\n", ret); + drm_err(&i915->drm, "Enabling PPGTT failed (%d)\n", ret); goto out; } @@ -262,7 +264,7 @@ intel_gt_clear_error_registers(struct intel_gt *gt, * some errors might have become stuck, * mask them. */ - DRM_DEBUG_DRIVER("EIR stuck: 0x%08x, masking\n", eir); + drm_dbg(>->i915->drm, "EIR stuck: 0x%08x, masking\n", eir); rmw_set(uncore, EMR, eir); intel_uncore_write(uncore, GEN2_IIR, I915_MASTER_ERROR_INTERRUPT); diff --git a/drivers/gpu/drm/i915/gt/intel_gt_irq.c b/drivers/gpu/drm/i915/gt/intel_gt_irq.c index f26882fdc24c..6f6b9e04d916 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt_irq.c +++ b/drivers/gpu/drm/i915/gt/intel_gt_irq.c @@ -17,6 +17,9 @@ static void guc_irq_handler(struct intel_guc *guc, u16 iir) { + if (unlikely(!guc->interrupts.enabled)) + return; + if (iir & GUC_INTR_GUC2HOST) intel_guc_to_host_event_handler(guc); } @@ -44,8 +47,9 @@ gen11_gt_engine_identity(struct intel_gt *gt, !time_after32(local_clock() >> 10, timeout_ts)); if (unlikely(!(ident & GEN11_INTR_DATA_VALID))) { - DRM_ERROR("INTR_IDENTITY_REG%u:%u 0x%08x not valid!\n", - bank, bit, ident); + drm_err(>->i915->drm, + "INTR_IDENTITY_REG%u:%u 0x%08x not valid!\n", + bank, bit, ident); return 0; } @@ -81,35 +85,27 @@ gen11_other_irq_handler(struct intel_gt *gt, const u8 instance, instance, iir); } -static void -gen11_engine_irq_handler(struct intel_gt *gt, const u8 class, - const u8 instance, const u16 iir) +static struct intel_gt *pick_gt(struct intel_gt *gt, u8 class, u8 instance) { - struct intel_engine_cs *engine; - - /* - * Platforms with standalone media have their media engines in another - * GT. - */ - if (MEDIA_VER(gt->i915) >= 13 && - (class == VIDEO_DECODE_CLASS || class == VIDEO_ENHANCEMENT_CLASS)) { - if (!gt->i915->media_gt) - goto err; + struct intel_gt *media_gt = gt->i915->media_gt; - gt = gt->i915->media_gt; + /* we expect the non-media gt to be passed in */ + GEM_BUG_ON(gt == media_gt); + + if (!media_gt) + return gt; + + switch (class) { + case VIDEO_DECODE_CLASS: + case VIDEO_ENHANCEMENT_CLASS: + return media_gt; + case OTHER_CLASS: + if (instance == OTHER_GSC_INSTANCE && HAS_ENGINE(media_gt, GSC0)) + return media_gt; + fallthrough; + default: + return gt; } - - if (instance <= MAX_ENGINE_INSTANCE) - engine = gt->engine_class[class][instance]; - else - engine = NULL; - - if (likely(engine)) - return intel_engine_cs_irq(engine, iir); - -err: - WARN_ONCE(1, "unhandled engine interrupt class=0x%x, instance=0x%x\n", - class, instance); } static void @@ -122,8 +118,17 @@ gen11_gt_identity_handler(struct intel_gt *gt, const u32 identity) if (unlikely(!intr)) return; - if (class <= COPY_ENGINE_CLASS || class == COMPUTE_CLASS) - return gen11_engine_irq_handler(gt, class, instance, intr); + /* + * Platforms with standalone media have the media and GSC engines in + * another GT. + */ + gt = pick_gt(gt, class, instance); + + if (class <= MAX_ENGINE_CLASS && instance <= MAX_ENGINE_INSTANCE) { + struct intel_engine_cs *engine = gt->engine_class[class][instance]; + if (engine) + return intel_engine_cs_irq(engine, intr); + } if (class == OTHER_CLASS) return gen11_other_irq_handler(gt, instance, intr); @@ -206,7 +211,7 @@ void gen11_gt_irq_reset(struct intel_gt *gt) intel_uncore_write(uncore, GEN11_VCS_VECS_INTR_ENABLE, 0); if (CCS_MASK(gt)) intel_uncore_write(uncore, GEN12_CCS_RSVD_INTR_ENABLE, 0); - if (HAS_HECI_GSC(gt->i915)) + if (HAS_HECI_GSC(gt->i915) || HAS_ENGINE(gt, GSC0)) intel_uncore_write(uncore, GEN11_GUNIT_CSME_INTR_ENABLE, 0); /* Restore masks irqs on RCS, BCS, VCS and VECS engines. */ @@ -233,7 +238,7 @@ void gen11_gt_irq_reset(struct intel_gt *gt) intel_uncore_write(uncore, GEN12_CCS0_CCS1_INTR_MASK, ~0); if (HAS_ENGINE(gt, CCS2) || HAS_ENGINE(gt, CCS3)) intel_uncore_write(uncore, GEN12_CCS2_CCS3_INTR_MASK, ~0); - if (HAS_HECI_GSC(gt->i915)) + if (HAS_HECI_GSC(gt->i915) || HAS_ENGINE(gt, GSC0)) intel_uncore_write(uncore, GEN11_GUNIT_CSME_INTR_MASK, ~0); intel_uncore_write(uncore, GEN11_GPM_WGBOXPERF_INTR_ENABLE, 0); @@ -249,7 +254,8 @@ void gen11_gt_irq_postinstall(struct intel_gt *gt) { struct intel_uncore *uncore = gt->uncore; u32 irqs = GT_RENDER_USER_INTERRUPT; - const u32 gsc_mask = GSC_IRQ_INTF(0) | GSC_IRQ_INTF(1); + u32 guc_mask = intel_uc_wants_guc(>->uc) ? GUC_INTR_GUC2HOST : 0; + u32 gsc_mask = 0; u32 dmask; u32 smask; @@ -261,6 +267,11 @@ void gen11_gt_irq_postinstall(struct intel_gt *gt) dmask = irqs << 16 | irqs; smask = irqs << 16; + if (HAS_ENGINE(gt, GSC0)) + gsc_mask = irqs; + else if (HAS_HECI_GSC(gt->i915)) + gsc_mask = GSC_IRQ_INTF(0) | GSC_IRQ_INTF(1); + BUILD_BUG_ON(irqs & 0xffff0000); /* Enable RCS, BCS, VCS and VECS class interrupts. */ @@ -268,9 +279,8 @@ void gen11_gt_irq_postinstall(struct intel_gt *gt) intel_uncore_write(uncore, GEN11_VCS_VECS_INTR_ENABLE, dmask); if (CCS_MASK(gt)) intel_uncore_write(uncore, GEN12_CCS_RSVD_INTR_ENABLE, smask); - if (HAS_HECI_GSC(gt->i915)) - intel_uncore_write(uncore, GEN11_GUNIT_CSME_INTR_ENABLE, - gsc_mask); + if (gsc_mask) + intel_uncore_write(uncore, GEN11_GUNIT_CSME_INTR_ENABLE, gsc_mask); /* Unmask irqs on RCS, BCS, VCS and VECS engines. */ intel_uncore_write(uncore, GEN11_RCS0_RSVD_INTR_MASK, ~smask); @@ -296,9 +306,22 @@ void gen11_gt_irq_postinstall(struct intel_gt *gt) intel_uncore_write(uncore, GEN12_CCS0_CCS1_INTR_MASK, ~dmask); if (HAS_ENGINE(gt, CCS2) || HAS_ENGINE(gt, CCS3)) intel_uncore_write(uncore, GEN12_CCS2_CCS3_INTR_MASK, ~dmask); - if (HAS_HECI_GSC(gt->i915)) + if (gsc_mask) intel_uncore_write(uncore, GEN11_GUNIT_CSME_INTR_MASK, ~gsc_mask); + if (guc_mask) { + /* the enable bit is common for both GTs but the masks are separate */ + u32 mask = gt->type == GT_MEDIA ? + REG_FIELD_PREP(ENGINE0_MASK, guc_mask) : + REG_FIELD_PREP(ENGINE1_MASK, guc_mask); + + intel_uncore_write(uncore, GEN11_GUC_SG_INTR_ENABLE, + REG_FIELD_PREP(ENGINE1_MASK, guc_mask)); + + /* we might not be the first GT to write this reg */ + intel_uncore_rmw(uncore, MTL_GUC_MGUC_INTR_MASK, mask, 0); + } + /* * RPS interrupts will get enabled/disabled on demand when RPS itself * is enabled/disabled. @@ -307,10 +330,6 @@ void gen11_gt_irq_postinstall(struct intel_gt *gt) gt->pm_imr = ~gt->pm_ier; intel_uncore_write(uncore, GEN11_GPM_WGBOXPERF_INTR_ENABLE, 0); intel_uncore_write(uncore, GEN11_GPM_WGBOXPERF_INTR_MASK, ~0); - - /* Same thing for GuC interrupts */ - intel_uncore_write(uncore, GEN11_GUC_SG_INTR_ENABLE, 0); - intel_uncore_write(uncore, GEN11_GUC_SG_INTR_MASK, ~0); } void gen5_gt_irq_handler(struct intel_gt *gt, u32 gt_iir) @@ -359,7 +378,8 @@ void gen6_gt_irq_handler(struct intel_gt *gt, u32 gt_iir) if (gt_iir & (GT_BLT_CS_ERROR_INTERRUPT | GT_BSD_CS_ERROR_INTERRUPT | GT_CS_MASTER_ERROR_INTERRUPT)) - DRM_DEBUG("Command parser error, gt_iir 0x%08x\n", gt_iir); + drm_dbg(>->i915->drm, "Command parser error, gt_iir 0x%08x\n", + gt_iir); if (gt_iir & GT_PARITY_ERROR(gt->i915)) gen7_parity_error_irq_handler(gt, gt_iir); diff --git a/drivers/gpu/drm/i915/gt/intel_gt_pm.c b/drivers/gpu/drm/i915/gt/intel_gt_pm.c index f553e2173bda..16db85fab0b1 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt_pm.c +++ b/drivers/gpu/drm/i915/gt/intel_gt_pm.c @@ -7,6 +7,7 @@ #include <linux/suspend.h> #include "i915_drv.h" +#include "i915_irq.h" #include "i915_params.h" #include "intel_context.h" #include "intel_engine_pm.h" @@ -19,10 +20,31 @@ #include "intel_rc6.h" #include "intel_rps.h" #include "intel_wakeref.h" +#include "intel_pcode.h" #include "pxp/intel_pxp_pm.h" #define I915_GT_SUSPEND_IDLE_TIMEOUT (HZ / 2) +static void mtl_media_busy(struct intel_gt *gt) +{ + /* Wa_14017073508: mtl */ + if (IS_MTL_GRAPHICS_STEP(gt->i915, P, STEP_A0, STEP_B0) && + gt->type == GT_MEDIA) + snb_pcode_write_p(gt->uncore, PCODE_MBOX_GT_STATE, + PCODE_MBOX_GT_STATE_MEDIA_BUSY, + PCODE_MBOX_GT_STATE_DOMAIN_MEDIA, 0); +} + +static void mtl_media_idle(struct intel_gt *gt) +{ + /* Wa_14017073508: mtl */ + if (IS_MTL_GRAPHICS_STEP(gt->i915, P, STEP_A0, STEP_B0) && + gt->type == GT_MEDIA) + snb_pcode_write_p(gt->uncore, PCODE_MBOX_GT_STATE, + PCODE_MBOX_GT_STATE_MEDIA_NOT_BUSY, + PCODE_MBOX_GT_STATE_DOMAIN_MEDIA, 0); +} + static void user_forcewake(struct intel_gt *gt, bool suspend) { int count = atomic_read(>->user_wakeref); @@ -70,6 +92,9 @@ static int __gt_unpark(struct intel_wakeref *wf) GT_TRACE(gt, "\n"); + /* Wa_14017073508: mtl */ + mtl_media_busy(gt); + /* * It seems that the DMC likes to transition between the DC states a lot * when there are no connected displays (no active power domains) during @@ -119,6 +144,9 @@ static int __gt_park(struct intel_wakeref *wf) GEM_BUG_ON(!wakeref); intel_display_power_put_async(i915, POWER_DOMAIN_GT_IRQ, wakeref); + /* Wa_14017073508: mtl */ + mtl_media_idle(gt); + return 0; } diff --git a/drivers/gpu/drm/i915/gt/intel_gt_pm_debugfs.c b/drivers/gpu/drm/i915/gt/intel_gt_pm_debugfs.c index 40d0a3be42ac..83df4cd5e06c 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt_pm_debugfs.c +++ b/drivers/gpu/drm/i915/gt/intel_gt_pm_debugfs.c @@ -83,19 +83,6 @@ static int fw_domains_show(struct seq_file *m, void *data) } DEFINE_INTEL_GT_DEBUGFS_ATTRIBUTE(fw_domains); -static void print_rc6_res(struct seq_file *m, - const char *title, - const i915_reg_t reg) -{ - struct intel_gt *gt = m->private; - intel_wakeref_t wakeref; - - with_intel_runtime_pm(gt->uncore->rpm, wakeref) - seq_printf(m, "%s %u (%llu us)\n", title, - intel_uncore_read(gt->uncore, reg), - intel_rc6_residency_us(>->rc6, reg)); -} - static int vlv_drpc(struct seq_file *m) { struct intel_gt *gt = m->private; @@ -115,8 +102,8 @@ static int vlv_drpc(struct seq_file *m) seq_printf(m, "Media Power Well: %s\n", (pw_status & VLV_GTLC_PW_MEDIA_STATUS_MASK) ? "Up" : "Down"); - print_rc6_res(m, "Render RC6 residency since boot:", GEN6_GT_GFX_RC6); - print_rc6_res(m, "Media RC6 residency since boot:", VLV_GT_MEDIA_RC6); + intel_rc6_print_residency(m, "Render RC6 residency since boot:", INTEL_RC6_RES_RC6); + intel_rc6_print_residency(m, "Media RC6 residency since boot:", INTEL_RC6_RES_VLV_MEDIA); return fw_domains_show(m, NULL); } @@ -192,11 +179,11 @@ static int gen6_drpc(struct seq_file *m) } /* Not exactly sure what this is */ - print_rc6_res(m, "RC6 \"Locked to RPn\" residency since boot:", - GEN6_GT_GFX_RC6_LOCKED); - print_rc6_res(m, "RC6 residency since boot:", GEN6_GT_GFX_RC6); - print_rc6_res(m, "RC6+ residency since boot:", GEN6_GT_GFX_RC6p); - print_rc6_res(m, "RC6++ residency since boot:", GEN6_GT_GFX_RC6pp); + intel_rc6_print_residency(m, "RC6 \"Locked to RPn\" residency since boot:", + INTEL_RC6_RES_RC6_LOCKED); + intel_rc6_print_residency(m, "RC6 residency since boot:", INTEL_RC6_RES_RC6); + intel_rc6_print_residency(m, "RC6+ residency since boot:", INTEL_RC6_RES_RC6p); + intel_rc6_print_residency(m, "RC6++ residency since boot:", INTEL_RC6_RES_RC6pp); if (GRAPHICS_VER(i915) <= 7) { seq_printf(m, "RC6 voltage: %dmV\n", @@ -269,6 +256,61 @@ static int ilk_drpc(struct seq_file *m) return 0; } +static int mtl_drpc(struct seq_file *m) +{ + struct intel_gt *gt = m->private; + struct intel_uncore *uncore = gt->uncore; + u32 gt_core_status, rcctl1, mt_fwake_req; + u32 mtl_powergate_enable = 0, mtl_powergate_status = 0; + + mt_fwake_req = intel_uncore_read_fw(uncore, FORCEWAKE_MT); + gt_core_status = intel_uncore_read(uncore, MTL_MIRROR_TARGET_WP1); + + rcctl1 = intel_uncore_read(uncore, GEN6_RC_CONTROL); + mtl_powergate_enable = intel_uncore_read(uncore, GEN9_PG_ENABLE); + mtl_powergate_status = intel_uncore_read(uncore, + GEN9_PWRGT_DOMAIN_STATUS); + + seq_printf(m, "RC6 Enabled: %s\n", + str_yes_no(rcctl1 & GEN6_RC_CTL_RC6_ENABLE)); + if (gt->type == GT_MEDIA) { + seq_printf(m, "Media Well Gating Enabled: %s\n", + str_yes_no(mtl_powergate_enable & GEN9_MEDIA_PG_ENABLE)); + } else { + seq_printf(m, "Render Well Gating Enabled: %s\n", + str_yes_no(mtl_powergate_enable & GEN9_RENDER_PG_ENABLE)); + } + + seq_puts(m, "Current RC state: "); + switch (REG_FIELD_GET(MTL_CC_MASK, gt_core_status)) { + case MTL_CC0: + seq_puts(m, "RC0\n"); + break; + case MTL_CC6: + seq_puts(m, "RC6\n"); + break; + default: + MISSING_CASE(REG_FIELD_GET(MTL_CC_MASK, gt_core_status)); + seq_puts(m, "Unknown\n"); + break; + } + + seq_printf(m, "Multi-threaded Forcewake Request: 0x%x\n", mt_fwake_req); + if (gt->type == GT_MEDIA) + seq_printf(m, "Media Power Well: %s\n", + (mtl_powergate_status & + GEN9_PWRGT_MEDIA_STATUS_MASK) ? "Up" : "Down"); + else + seq_printf(m, "Render Power Well: %s\n", + (mtl_powergate_status & + GEN9_PWRGT_RENDER_STATUS_MASK) ? "Up" : "Down"); + + /* Works for both render and media gt's */ + intel_rc6_print_residency(m, "RC6 residency since boot:", INTEL_RC6_RES_RC6); + + return fw_domains_show(m, NULL); +} + static int drpc_show(struct seq_file *m, void *unused) { struct intel_gt *gt = m->private; @@ -277,7 +319,9 @@ static int drpc_show(struct seq_file *m, void *unused) int err = -ENODEV; with_intel_runtime_pm(gt->uncore->rpm, wakeref) { - if (IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915)) + if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 70)) + err = mtl_drpc(m); + else if (IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915)) err = vlv_drpc(m); else if (GRAPHICS_VER(i915) >= 6) err = gen6_drpc(m); @@ -307,7 +351,7 @@ void intel_gt_pm_frequency_dump(struct intel_gt *gt, struct drm_printer *p) drm_printf(p, "Current VID: %d\n", (rgvstat & MEMSTAT_VID_MASK) >> MEMSTAT_VID_SHIFT); drm_printf(p, "Current P-state: %d\n", - (rgvstat & MEMSTAT_PSTATE_MASK) >> MEMSTAT_PSTATE_SHIFT); + REG_FIELD_GET(MEMSTAT_PSTATE_MASK, rgvstat)); } else if (IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915)) { u32 rpmodectl, freq_sts; diff --git a/drivers/gpu/drm/i915/gt/intel_gt_regs.h b/drivers/gpu/drm/i915/gt/intel_gt_regs.h index 70177d3f2e94..c3cd92691795 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt_regs.h +++ b/drivers/gpu/drm/i915/gt/intel_gt_regs.h @@ -7,6 +7,7 @@ #define __INTEL_GT_REGS__ #include "i915_reg_defs.h" +#include "display/intel_display_reg_defs.h" /* VLV_DISPLAY_BASE */ #define MCR_REG(offset) ((const i915_mcr_reg_t){ .reg = (offset) }) @@ -21,6 +22,13 @@ */ #define PERF_REG(offset) _MMIO(offset) +/* MTL workpoint reg to get core C state and actual freq of 3D, SAMedia */ +#define MTL_MIRROR_TARGET_WP1 _MMIO(0xc60) +#define MTL_CAGF_MASK REG_GENMASK(8, 0) +#define MTL_CC0 0x0 +#define MTL_CC6 0x3 +#define MTL_CC_MASK REG_GENMASK(12, 9) + /* RPM unit config (Gen8+) */ #define RPM_CONFIG0 _MMIO(0xd00) #define GEN9_RPM_CONFIG0_CRYSTAL_CLOCK_FREQ_SHIFT 3 @@ -643,6 +651,7 @@ #define XEHPC_GRDOM_BLT3 REG_BIT(26) #define XEHPC_GRDOM_BLT2 REG_BIT(25) #define XEHPC_GRDOM_BLT1 REG_BIT(24) +#define GEN12_GRDOM_GSC REG_BIT(21) #define GEN11_GRDOM_SFC3 REG_BIT(20) #define GEN11_GRDOM_SFC2 REG_BIT(19) #define GEN11_GRDOM_SFC1 REG_BIT(18) @@ -798,12 +807,9 @@ #define GEN6_RP_DOWN_TIMEOUT _MMIO(0xa010) #define GEN6_RP_INTERRUPT_LIMITS _MMIO(0xa014) #define GEN6_RPSTAT1 _MMIO(0xa01c) -#define GEN6_CAGF_SHIFT 8 -#define HSW_CAGF_SHIFT 7 -#define GEN9_CAGF_SHIFT 23 -#define GEN6_CAGF_MASK (0x7f << GEN6_CAGF_SHIFT) -#define HSW_CAGF_MASK (0x7f << HSW_CAGF_SHIFT) -#define GEN9_CAGF_MASK (0x1ff << GEN9_CAGF_SHIFT) +#define GEN6_CAGF_MASK REG_GENMASK(14, 8) +#define HSW_CAGF_MASK REG_GENMASK(13, 7) +#define GEN9_CAGF_MASK REG_GENMASK(31, 23) #define GEN6_RP_CONTROL _MMIO(0xa024) #define GEN6_RP_MEDIA_TURBO (1 << 11) #define GEN6_RP_MEDIA_MODE_MASK (3 << 9) @@ -1375,8 +1381,7 @@ #define MEMSTAT_ILK _MMIO(0x111f8) #define MEMSTAT_VID_MASK 0x7f00 #define MEMSTAT_VID_SHIFT 8 -#define MEMSTAT_PSTATE_MASK 0x00f8 -#define MEMSTAT_PSTATE_SHIFT 3 +#define MEMSTAT_PSTATE_MASK REG_GENMASK(7, 3) #define MEMSTAT_MON_ACTV (1 << 2) #define MEMSTAT_SRC_CTL_MASK 0x0003 #define MEMSTAT_SRC_CTL_CORE 0 @@ -1517,6 +1522,8 @@ #define FORCEWAKE_MEDIA_VLV _MMIO(0x1300b8) #define FORCEWAKE_ACK_MEDIA_VLV _MMIO(0x1300bc) +#define MTL_MEDIA_MC6 _MMIO(0x138048) + #define GEN6_GT_THREAD_STATUS_REG _MMIO(0x13805c) #define GEN6_GT_THREAD_STATUS_CORE_MASK 0x7 @@ -1548,11 +1555,13 @@ #define GEN12_RPSTAT1 _MMIO(0x1381b4) #define GEN12_VOLTAGE_MASK REG_GENMASK(10, 0) +#define GEN12_CAGF_MASK REG_GENMASK(19, 11) #define GEN11_GT_INTR_DW(x) _MMIO(0x190018 + ((x) * 4)) #define GEN11_CSME (31) #define GEN11_GUNIT (28) #define GEN11_GUC (25) +#define MTL_MGUC (24) #define GEN11_WDPERF (20) #define GEN11_KCR (19) #define GEN11_GTPM (16) @@ -1607,6 +1616,7 @@ #define GEN11_VECS0_VECS1_INTR_MASK _MMIO(0x1900d0) #define GEN12_VECS2_VECS3_INTR_MASK _MMIO(0x1900d4) #define GEN11_GUC_SG_INTR_MASK _MMIO(0x1900e8) +#define MTL_GUC_MGUC_INTR_MASK _MMIO(0x1900e8) /* MTL+ */ #define GEN11_GPM_WGBOXPERF_INTR_MASK _MMIO(0x1900ec) #define GEN11_CRYPTO_RSVD_INTR_MASK _MMIO(0x1900f0) #define GEN11_GUNIT_CSME_INTR_MASK _MMIO(0x1900f4) diff --git a/drivers/gpu/drm/i915/gt/intel_gt_sysfs_pm.c b/drivers/gpu/drm/i915/gt/intel_gt_sysfs_pm.c index 2b5f05b31187..cf71305ad586 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt_sysfs_pm.c +++ b/drivers/gpu/drm/i915/gt/intel_gt_sysfs_pm.c @@ -165,13 +165,13 @@ sysfs_gt_attribute_r_func(struct kobject *kobj, struct attribute *attr, INTEL_GT_ATTR_RO(_name) #ifdef CONFIG_PM -static u32 get_residency(struct intel_gt *gt, i915_reg_t reg) +static u32 get_residency(struct intel_gt *gt, enum intel_rc6_res_type id) { intel_wakeref_t wakeref; u64 res = 0; with_intel_runtime_pm(gt->uncore->rpm, wakeref) - res = intel_rc6_residency_us(>->rc6, reg); + res = intel_rc6_residency_us(>->rc6, id); return DIV_ROUND_CLOSEST_ULL(res, 1000); } @@ -210,22 +210,22 @@ static ssize_t rc6_enable_dev_show(struct device *dev, static u32 __rc6_residency_ms_show(struct intel_gt *gt) { - return get_residency(gt, GEN6_GT_GFX_RC6); + return get_residency(gt, INTEL_RC6_RES_RC6); } static u32 __rc6p_residency_ms_show(struct intel_gt *gt) { - return get_residency(gt, GEN6_GT_GFX_RC6p); + return get_residency(gt, INTEL_RC6_RES_RC6p); } static u32 __rc6pp_residency_ms_show(struct intel_gt *gt) { - return get_residency(gt, GEN6_GT_GFX_RC6pp); + return get_residency(gt, INTEL_RC6_RES_RC6pp); } static u32 __media_rc6_residency_ms_show(struct intel_gt *gt) { - return get_residency(gt, VLV_GT_MEDIA_RC6); + return get_residency(gt, INTEL_RC6_RES_VLV_MEDIA); } INTEL_GT_SYSFS_SHOW_MIN(rc6_residency_ms); diff --git a/drivers/gpu/drm/i915/gt/intel_gt_types.h b/drivers/gpu/drm/i915/gt/intel_gt_types.h index a0cc73b401ef..c1d9cd255e06 100644 --- a/drivers/gpu/drm/i915/gt/intel_gt_types.h +++ b/drivers/gpu/drm/i915/gt/intel_gt_types.h @@ -31,6 +31,7 @@ #include "intel_migrate_types.h" #include "intel_wakeref.h" #include "pxp/intel_pxp_types.h" +#include "intel_wopcm.h" struct drm_i915_private; struct i915_ggtt; @@ -101,6 +102,7 @@ struct intel_gt { struct intel_uc uc; struct intel_gsc gsc; + struct intel_wopcm wopcm; struct { /* Serialize global tlb invalidations */ diff --git a/drivers/gpu/drm/i915/gt/intel_gtt.c b/drivers/gpu/drm/i915/gt/intel_gtt.c index e82a9d763e57..2ba3983984b9 100644 --- a/drivers/gpu/drm/i915/gt/intel_gtt.c +++ b/drivers/gpu/drm/i915/gt/intel_gtt.c @@ -12,6 +12,7 @@ #include "gem/i915_gem_internal.h" #include "gem/i915_gem_lmem.h" +#include "i915_reg.h" #include "i915_trace.h" #include "i915_utils.h" #include "intel_gt.h" diff --git a/drivers/gpu/drm/i915/gt/intel_rc6.c b/drivers/gpu/drm/i915/gt/intel_rc6.c index f8d0523f4c18..2ee4051e4d96 100644 --- a/drivers/gpu/drm/i915/gt/intel_rc6.c +++ b/drivers/gpu/drm/i915/gt/intel_rc6.c @@ -551,6 +551,23 @@ static void __intel_rc6_disable(struct intel_rc6 *rc6) intel_uncore_forcewake_put(uncore, FORCEWAKE_ALL); } +static void rc6_res_reg_init(struct intel_rc6 *rc6) +{ + memset(rc6->res_reg, INVALID_MMIO_REG.reg, sizeof(rc6->res_reg)); + + switch (rc6_to_gt(rc6)->type) { + case GT_MEDIA: + rc6->res_reg[INTEL_RC6_RES_RC6] = MTL_MEDIA_MC6; + break; + default: + rc6->res_reg[INTEL_RC6_RES_RC6_LOCKED] = GEN6_GT_GFX_RC6_LOCKED; + rc6->res_reg[INTEL_RC6_RES_RC6] = GEN6_GT_GFX_RC6; + rc6->res_reg[INTEL_RC6_RES_RC6p] = GEN6_GT_GFX_RC6p; + rc6->res_reg[INTEL_RC6_RES_RC6pp] = GEN6_GT_GFX_RC6pp; + break; + } +} + void intel_rc6_init(struct intel_rc6 *rc6) { struct drm_i915_private *i915 = rc6_to_i915(rc6); @@ -562,6 +579,8 @@ void intel_rc6_init(struct intel_rc6 *rc6) if (!rc6_supported(rc6)) return; + rc6_res_reg_init(rc6); + if (IS_CHERRYVIEW(i915)) err = chv_rc6_init(rc6); else if (IS_VALLEYVIEW(i915)) @@ -736,31 +755,19 @@ static u64 vlv_residency_raw(struct intel_uncore *uncore, const i915_reg_t reg) return lower | (u64)upper << 8; } -u64 intel_rc6_residency_ns(struct intel_rc6 *rc6, const i915_reg_t reg) +u64 intel_rc6_residency_ns(struct intel_rc6 *rc6, enum intel_rc6_res_type id) { struct drm_i915_private *i915 = rc6_to_i915(rc6); struct intel_uncore *uncore = rc6_to_uncore(rc6); u64 time_hw, prev_hw, overflow_hw; + i915_reg_t reg = rc6->res_reg[id]; unsigned int fw_domains; unsigned long flags; - unsigned int i; u32 mul, div; if (!rc6->supported) return 0; - /* - * Store previous hw counter values for counter wrap-around handling. - * - * There are only four interesting registers and they live next to each - * other so we can use the relative address, compared to the smallest - * one as the index into driver storage. - */ - i = (i915_mmio_reg_offset(reg) - - i915_mmio_reg_offset(GEN6_GT_GFX_RC6_LOCKED)) / sizeof(u32); - if (drm_WARN_ON_ONCE(&i915->drm, i >= ARRAY_SIZE(rc6->cur_residency))) - return 0; - fw_domains = intel_uncore_forcewake_for_reg(uncore, reg, FW_REG_READ); spin_lock_irqsave(&uncore->lock, flags); @@ -789,11 +796,11 @@ u64 intel_rc6_residency_ns(struct intel_rc6 *rc6, const i915_reg_t reg) /* * Counter wrap handling. * - * But relying on a sufficient frequency of queries otherwise counters - * can still wrap. + * Store previous hw counter values for counter wrap-around handling. But + * relying on a sufficient frequency of queries otherwise counters can still wrap. */ - prev_hw = rc6->prev_hw_residency[i]; - rc6->prev_hw_residency[i] = time_hw; + prev_hw = rc6->prev_hw_residency[id]; + rc6->prev_hw_residency[id] = time_hw; /* RC6 delta from last sample. */ if (time_hw >= prev_hw) @@ -802,8 +809,8 @@ u64 intel_rc6_residency_ns(struct intel_rc6 *rc6, const i915_reg_t reg) time_hw += overflow_hw - prev_hw; /* Add delta to RC6 extended raw driver copy. */ - time_hw += rc6->cur_residency[i]; - rc6->cur_residency[i] = time_hw; + time_hw += rc6->cur_residency[id]; + rc6->cur_residency[id] = time_hw; intel_uncore_forcewake_put__locked(uncore, fw_domains); spin_unlock_irqrestore(&uncore->lock, flags); @@ -811,9 +818,22 @@ u64 intel_rc6_residency_ns(struct intel_rc6 *rc6, const i915_reg_t reg) return mul_u64_u32_div(time_hw, mul, div); } -u64 intel_rc6_residency_us(struct intel_rc6 *rc6, i915_reg_t reg) +u64 intel_rc6_residency_us(struct intel_rc6 *rc6, enum intel_rc6_res_type id) +{ + return DIV_ROUND_UP_ULL(intel_rc6_residency_ns(rc6, id), 1000); +} + +void intel_rc6_print_residency(struct seq_file *m, const char *title, + enum intel_rc6_res_type id) { - return DIV_ROUND_UP_ULL(intel_rc6_residency_ns(rc6, reg), 1000); + struct intel_gt *gt = m->private; + i915_reg_t reg = gt->rc6.res_reg[id]; + intel_wakeref_t wakeref; + + with_intel_runtime_pm(gt->uncore->rpm, wakeref) + seq_printf(m, "%s %u (%llu us)\n", title, + intel_uncore_read(gt->uncore, reg), + intel_rc6_residency_us(>->rc6, id)); } #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST) diff --git a/drivers/gpu/drm/i915/gt/intel_rc6.h b/drivers/gpu/drm/i915/gt/intel_rc6.h index b6fea71afc22..456fa668a276 100644 --- a/drivers/gpu/drm/i915/gt/intel_rc6.h +++ b/drivers/gpu/drm/i915/gt/intel_rc6.h @@ -6,10 +6,11 @@ #ifndef INTEL_RC6_H #define INTEL_RC6_H -#include "i915_reg_defs.h" +#include <linux/types.h> -struct intel_engine_cs; +enum intel_rc6_res_type; struct intel_rc6; +struct seq_file; void intel_rc6_init(struct intel_rc6 *rc6); void intel_rc6_fini(struct intel_rc6 *rc6); @@ -21,7 +22,9 @@ void intel_rc6_sanitize(struct intel_rc6 *rc6); void intel_rc6_enable(struct intel_rc6 *rc6); void intel_rc6_disable(struct intel_rc6 *rc6); -u64 intel_rc6_residency_ns(struct intel_rc6 *rc6, i915_reg_t reg); -u64 intel_rc6_residency_us(struct intel_rc6 *rc6, i915_reg_t reg); +u64 intel_rc6_residency_ns(struct intel_rc6 *rc6, enum intel_rc6_res_type id); +u64 intel_rc6_residency_us(struct intel_rc6 *rc6, enum intel_rc6_res_type id); +void intel_rc6_print_residency(struct seq_file *m, const char *title, + enum intel_rc6_res_type id); #endif /* INTEL_RC6_H */ diff --git a/drivers/gpu/drm/i915/gt/intel_rc6_types.h b/drivers/gpu/drm/i915/gt/intel_rc6_types.h index e747492b2f46..fa23c4dce00b 100644 --- a/drivers/gpu/drm/i915/gt/intel_rc6_types.h +++ b/drivers/gpu/drm/i915/gt/intel_rc6_types.h @@ -13,9 +13,20 @@ struct drm_i915_gem_object; +/* RC6 residency types */ +enum intel_rc6_res_type { + INTEL_RC6_RES_RC6_LOCKED, + INTEL_RC6_RES_RC6, + INTEL_RC6_RES_RC6p, + INTEL_RC6_RES_RC6pp, + INTEL_RC6_RES_MAX, + INTEL_RC6_RES_VLV_MEDIA = INTEL_RC6_RES_RC6p, +}; + struct intel_rc6 { - u64 prev_hw_residency[4]; - u64 cur_residency[4]; + i915_reg_t res_reg[INTEL_RC6_RES_MAX]; + u64 prev_hw_residency[INTEL_RC6_RES_MAX]; + u64 cur_residency[INTEL_RC6_RES_MAX]; u32 ctl_enable; diff --git a/drivers/gpu/drm/i915/gt/intel_renderstate.c b/drivers/gpu/drm/i915/gt/intel_renderstate.c index 5121e6dc2fa5..9c1ae070ee7b 100644 --- a/drivers/gpu/drm/i915/gt/intel_renderstate.c +++ b/drivers/gpu/drm/i915/gt/intel_renderstate.c @@ -215,9 +215,7 @@ int intel_renderstate_emit(struct intel_renderstate *so, if (!so->vma) return 0; - err = i915_request_await_object(rq, so->vma->obj, false); - if (err == 0) - err = i915_vma_move_to_active(so->vma, rq, 0); + err = i915_vma_move_to_active(so->vma, rq, 0); if (err) return err; diff --git a/drivers/gpu/drm/i915/gt/intel_reset.c b/drivers/gpu/drm/i915/gt/intel_reset.c index 3159df6cdd49..24736ebee17c 100644 --- a/drivers/gpu/drm/i915/gt/intel_reset.c +++ b/drivers/gpu/drm/i915/gt/intel_reset.c @@ -1407,15 +1407,19 @@ out: intel_runtime_pm_put(gt->uncore->rpm, wakeref); } -int intel_gt_reset_trylock(struct intel_gt *gt, int *srcu) +static int _intel_gt_reset_lock(struct intel_gt *gt, int *srcu, bool retry) { might_lock(>->reset.backoff_srcu); - might_sleep(); + if (retry) + might_sleep(); rcu_read_lock(); while (test_bit(I915_RESET_BACKOFF, >->reset.flags)) { rcu_read_unlock(); + if (!retry) + return -EBUSY; + if (wait_event_interruptible(gt->reset.queue, !test_bit(I915_RESET_BACKOFF, >->reset.flags))) @@ -1429,6 +1433,16 @@ int intel_gt_reset_trylock(struct intel_gt *gt, int *srcu) return 0; } +int intel_gt_reset_trylock(struct intel_gt *gt, int *srcu) +{ + return _intel_gt_reset_lock(gt, srcu, false); +} + +int intel_gt_reset_lock_interruptible(struct intel_gt *gt, int *srcu) +{ + return _intel_gt_reset_lock(gt, srcu, true); +} + void intel_gt_reset_unlock(struct intel_gt *gt, int tag) __releases(>->reset.backoff_srcu) { diff --git a/drivers/gpu/drm/i915/gt/intel_reset.h b/drivers/gpu/drm/i915/gt/intel_reset.h index adc734e67387..25c975b6e8fc 100644 --- a/drivers/gpu/drm/i915/gt/intel_reset.h +++ b/drivers/gpu/drm/i915/gt/intel_reset.h @@ -39,6 +39,7 @@ int __intel_engine_reset_bh(struct intel_engine_cs *engine, void __i915_request_reset(struct i915_request *rq, bool guilty); int __must_check intel_gt_reset_trylock(struct intel_gt *gt, int *srcu); +int __must_check intel_gt_reset_lock_interruptible(struct intel_gt *gt, int *srcu); void intel_gt_reset_unlock(struct intel_gt *gt, int tag); void intel_gt_set_wedged(struct intel_gt *gt); diff --git a/drivers/gpu/drm/i915/gt/intel_ring_submission.c b/drivers/gpu/drm/i915/gt/intel_ring_submission.c index d5d6f1fadcae..356c787e11d3 100644 --- a/drivers/gpu/drm/i915/gt/intel_ring_submission.c +++ b/drivers/gpu/drm/i915/gt/intel_ring_submission.c @@ -12,7 +12,9 @@ #include "gen6_ppgtt.h" #include "gen7_renderclear.h" #include "i915_drv.h" +#include "i915_irq.h" #include "i915_mitigations.h" +#include "i915_reg.h" #include "intel_breadcrumbs.h" #include "intel_context.h" #include "intel_engine_regs.h" diff --git a/drivers/gpu/drm/i915/gt/intel_rps.c b/drivers/gpu/drm/i915/gt/intel_rps.c index 6c34a83c24b3..9ad3bc7201cb 100644 --- a/drivers/gpu/drm/i915/gt/intel_rps.c +++ b/drivers/gpu/drm/i915/gt/intel_rps.c @@ -430,7 +430,8 @@ static int __gen5_rps_set(struct intel_rps *rps, u8 val) rgvswctl = intel_uncore_read16(uncore, MEMSWCTL); if (rgvswctl & MEMCTL_CMD_STS) { - DRM_DEBUG("gpu busy, RCS change rejected\n"); + drm_dbg(&rps_to_i915(rps)->drm, + "gpu busy, RCS change rejected\n"); return -EBUSY; /* still busy with another command */ } @@ -1953,7 +1954,8 @@ void gen6_rps_irq_handler(struct intel_rps *rps, u32 pm_iir) intel_engine_cs_irq(gt->engine[VECS0], pm_iir >> 10); if (pm_iir & PM_VEBOX_CS_ERROR_INTERRUPT) - DRM_DEBUG("Command parser error, pm_iir 0x%08x\n", pm_iir); + drm_dbg(&rps_to_i915(rps)->drm, + "Command parser error, pm_iir 0x%08x\n", pm_iir); } void gen5_rps_irq_handler(struct intel_rps *rps) @@ -2072,22 +2074,45 @@ void intel_rps_sanitize(struct intel_rps *rps) rps_disable_interrupts(rps); } +u32 intel_rps_read_rpstat_fw(struct intel_rps *rps) +{ + struct drm_i915_private *i915 = rps_to_i915(rps); + i915_reg_t rpstat; + + rpstat = (GRAPHICS_VER(i915) >= 12) ? GEN12_RPSTAT1 : GEN6_RPSTAT1; + + return intel_uncore_read_fw(rps_to_gt(rps)->uncore, rpstat); +} + +u32 intel_rps_read_rpstat(struct intel_rps *rps) +{ + struct drm_i915_private *i915 = rps_to_i915(rps); + i915_reg_t rpstat; + + rpstat = (GRAPHICS_VER(i915) >= 12) ? GEN12_RPSTAT1 : GEN6_RPSTAT1; + + return intel_uncore_read(rps_to_gt(rps)->uncore, rpstat); +} + u32 intel_rps_get_cagf(struct intel_rps *rps, u32 rpstat) { struct drm_i915_private *i915 = rps_to_i915(rps); u32 cagf; - if (IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915)) - cagf = (rpstat >> 8) & 0xff; + if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 70)) + cagf = REG_FIELD_GET(MTL_CAGF_MASK, rpstat); + else if (GRAPHICS_VER(i915) >= 12) + cagf = REG_FIELD_GET(GEN12_CAGF_MASK, rpstat); + else if (IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915)) + cagf = REG_FIELD_GET(RPE_MASK, rpstat); else if (GRAPHICS_VER(i915) >= 9) - cagf = (rpstat & GEN9_CAGF_MASK) >> GEN9_CAGF_SHIFT; + cagf = REG_FIELD_GET(GEN9_CAGF_MASK, rpstat); else if (IS_HASWELL(i915) || IS_BROADWELL(i915)) - cagf = (rpstat & HSW_CAGF_MASK) >> HSW_CAGF_SHIFT; + cagf = REG_FIELD_GET(HSW_CAGF_MASK, rpstat); else if (GRAPHICS_VER(i915) >= 6) - cagf = (rpstat & GEN6_CAGF_MASK) >> GEN6_CAGF_SHIFT; + cagf = REG_FIELD_GET(GEN6_CAGF_MASK, rpstat); else - cagf = gen5_invert_freq(rps, (rpstat & MEMSTAT_PSTATE_MASK) >> - MEMSTAT_PSTATE_SHIFT); + cagf = gen5_invert_freq(rps, REG_FIELD_GET(MEMSTAT_PSTATE_MASK, rpstat)); return cagf; } @@ -2098,7 +2123,15 @@ static u32 read_cagf(struct intel_rps *rps) struct intel_uncore *uncore = rps_to_uncore(rps); u32 freq; - if (IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915)) { + /* + * For Gen12+ reading freq from HW does not need a forcewake and + * registers will return 0 freq when GT is in RC6 + */ + if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 70)) { + freq = intel_uncore_read(uncore, MTL_MIRROR_TARGET_WP1); + } else if (GRAPHICS_VER(i915) >= 12) { + freq = intel_uncore_read(uncore, GEN12_RPSTAT1); + } else if (IS_VALLEYVIEW(i915) || IS_CHERRYVIEW(i915)) { vlv_punit_get(i915); freq = vlv_punit_read(i915, PUNIT_REG_GPU_FREQ_STS); vlv_punit_put(i915); @@ -2264,7 +2297,7 @@ static void rps_frequency_dump(struct intel_rps *rps, struct drm_printer *p) rpinclimit = intel_uncore_read(uncore, GEN6_RP_UP_THRESHOLD); rpdeclimit = intel_uncore_read(uncore, GEN6_RP_DOWN_THRESHOLD); - rpstat = intel_uncore_read(uncore, GEN6_RPSTAT1); + rpstat = intel_rps_read_rpstat(rps); rpcurupei = intel_uncore_read(uncore, GEN6_RP_CUR_UP_EI) & GEN6_CURICONT_MASK; rpcurup = intel_uncore_read(uncore, GEN6_RP_CUR_UP) & GEN6_CURBSYTAVG_MASK; rpprevup = intel_uncore_read(uncore, GEN6_RP_PREV_UP) & GEN6_CURBSYTAVG_MASK; @@ -2399,7 +2432,7 @@ static void slpc_frequency_dump(struct intel_rps *rps, struct drm_printer *p) drm_printf(p, "PM MASK=0x%08x\n", pm_mask); drm_printf(p, "pm_intrmsk_mbz: 0x%08x\n", rps->pm_intrmsk_mbz); - drm_printf(p, "RPSTAT1: 0x%08x\n", intel_uncore_read(uncore, GEN6_RPSTAT1)); + drm_printf(p, "RPSTAT1: 0x%08x\n", intel_rps_read_rpstat(rps)); drm_printf(p, "RPNSWREQ: %dMHz\n", intel_rps_get_requested_frequency(rps)); drm_printf(p, "Lowest (RPN) frequency: %dMHz\n", intel_gpu_freq(rps, caps.min_freq)); diff --git a/drivers/gpu/drm/i915/gt/intel_rps.h b/drivers/gpu/drm/i915/gt/intel_rps.h index 110300dfd438..9e1cad9ba0e9 100644 --- a/drivers/gpu/drm/i915/gt/intel_rps.h +++ b/drivers/gpu/drm/i915/gt/intel_rps.h @@ -48,6 +48,8 @@ u32 intel_rps_get_rp1_frequency(struct intel_rps *rps); u32 intel_rps_get_rpn_frequency(struct intel_rps *rps); u32 intel_rps_read_punit_req(struct intel_rps *rps); u32 intel_rps_read_punit_req_frequency(struct intel_rps *rps); +u32 intel_rps_read_rpstat(struct intel_rps *rps); +u32 intel_rps_read_rpstat_fw(struct intel_rps *rps); void gen6_rps_get_freq_caps(struct intel_rps *rps, struct intel_rps_freq_caps *caps); void intel_rps_raise_unslice(struct intel_rps *rps); void intel_rps_lower_unslice(struct intel_rps *rps); diff --git a/drivers/gpu/drm/i915/intel_wopcm.c b/drivers/gpu/drm/i915/gt/intel_wopcm.c index 322fb9eeb880..7ebbcc191c2d 100644 --- a/drivers/gpu/drm/i915/intel_wopcm.c +++ b/drivers/gpu/drm/i915/gt/intel_wopcm.c @@ -64,9 +64,9 @@ #define GEN9_GUC_FW_RESERVED SZ_128K #define GEN9_GUC_WOPCM_OFFSET (GUC_WOPCM_RESERVED + GEN9_GUC_FW_RESERVED) -static inline struct drm_i915_private *wopcm_to_i915(struct intel_wopcm *wopcm) +static inline struct intel_gt *wopcm_to_gt(struct intel_wopcm *wopcm) { - return container_of(wopcm, struct drm_i915_private, wopcm); + return container_of(wopcm, struct intel_gt, wopcm); } /** @@ -77,7 +77,8 @@ static inline struct drm_i915_private *wopcm_to_i915(struct intel_wopcm *wopcm) */ void intel_wopcm_init_early(struct intel_wopcm *wopcm) { - struct drm_i915_private *i915 = wopcm_to_i915(wopcm); + struct intel_gt *gt = wopcm_to_gt(wopcm); + struct drm_i915_private *i915 = gt->i915; if (!HAS_GT_UC(i915)) return; @@ -157,10 +158,11 @@ static bool check_hw_restrictions(struct drm_i915_private *i915, return true; } -static bool __check_layout(struct drm_i915_private *i915, u32 wopcm_size, +static bool __check_layout(struct intel_gt *gt, u32 wopcm_size, u32 guc_wopcm_base, u32 guc_wopcm_size, u32 guc_fw_size, u32 huc_fw_size) { + struct drm_i915_private *i915 = gt->i915; const u32 ctx_rsvd = context_reserved_size(i915); u32 size; @@ -181,12 +183,14 @@ static bool __check_layout(struct drm_i915_private *i915, u32 wopcm_size, return false; } - size = huc_fw_size + WOPCM_RESERVED_SIZE; - if (unlikely(guc_wopcm_base < size)) { - drm_err(&i915->drm, "WOPCM: no space for %s: %uK < %uK\n", - intel_uc_fw_type_repr(INTEL_UC_FW_TYPE_HUC), - guc_wopcm_base / SZ_1K, size / SZ_1K); - return false; + if (intel_uc_supports_huc(>->uc)) { + size = huc_fw_size + WOPCM_RESERVED_SIZE; + if (unlikely(guc_wopcm_base < size)) { + drm_err(&i915->drm, "WOPCM: no space for %s: %uK < %uK\n", + intel_uc_fw_type_repr(INTEL_UC_FW_TYPE_HUC), + guc_wopcm_base / SZ_1K, size / SZ_1K); + return false; + } } return check_hw_restrictions(i915, guc_wopcm_base, guc_wopcm_size, @@ -228,8 +232,8 @@ static bool __wopcm_regs_writable(struct intel_uncore *uncore) */ void intel_wopcm_init(struct intel_wopcm *wopcm) { - struct drm_i915_private *i915 = wopcm_to_i915(wopcm); - struct intel_gt *gt = to_gt(i915); + struct intel_gt *gt = wopcm_to_gt(wopcm); + struct drm_i915_private *i915 = gt->i915; u32 guc_fw_size = intel_uc_fw_get_upload_size(>->uc.guc.fw); u32 huc_fw_size = intel_uc_fw_get_upload_size(>->uc.huc.fw); u32 ctx_rsvd = context_reserved_size(i915); @@ -275,6 +279,19 @@ void intel_wopcm_init(struct intel_wopcm *wopcm) } /* + * On platforms with a media GT, the WOPCM is partitioned between the + * two GTs, so we would have to take that into account when doing the + * math below. There is also a new section reserved for the GSC context + * that would have to be factored in. However, all platforms with a + * media GT also have GuC depriv enabled, so the WOPCM regs are + * pre-locked and therefore we don't have to do the math ourselves. + */ + if (unlikely(i915->media_gt)) { + drm_err(&i915->drm, "Unlocked WOPCM regs with media GT\n"); + return; + } + + /* * Aligned value of guc_wopcm_base will determine available WOPCM space * for HuC firmware and mandatory reserved area. */ @@ -295,7 +312,7 @@ void intel_wopcm_init(struct intel_wopcm *wopcm) guc_wopcm_base / SZ_1K, guc_wopcm_size / SZ_1K); check: - if (__check_layout(i915, wopcm_size, guc_wopcm_base, guc_wopcm_size, + if (__check_layout(gt, wopcm_size, guc_wopcm_base, guc_wopcm_size, guc_fw_size, huc_fw_size)) { wopcm->guc.base = guc_wopcm_base; wopcm->guc.size = guc_wopcm_size; diff --git a/drivers/gpu/drm/i915/intel_wopcm.h b/drivers/gpu/drm/i915/gt/intel_wopcm.h index 17d6aa86008a..17d6aa86008a 100644 --- a/drivers/gpu/drm/i915/intel_wopcm.h +++ b/drivers/gpu/drm/i915/gt/intel_wopcm.h diff --git a/drivers/gpu/drm/i915/gt/intel_workarounds.c b/drivers/gpu/drm/i915/gt/intel_workarounds.c index 3cdf5c24dbc5..1bd8d63ad4f3 100644 --- a/drivers/gpu/drm/i915/gt/intel_workarounds.c +++ b/drivers/gpu/drm/i915/gt/intel_workarounds.c @@ -4,6 +4,7 @@ */ #include "i915_drv.h" +#include "i915_reg.h" #include "intel_context.h" #include "intel_engine_pm.h" #include "intel_engine_regs.h" @@ -17,46 +18,68 @@ /** * DOC: Hardware workarounds * - * This file is intended as a central place to implement most [1]_ of the - * required workarounds for hardware to work as originally intended. They fall - * in five basic categories depending on how/when they are applied: + * Hardware workarounds are register programming documented to be executed in + * the driver that fall outside of the normal programming sequences for a + * platform. There are some basic categories of workarounds, depending on + * how/when they are applied: * - * - Workarounds that touch registers that are saved/restored to/from the HW - * context image. The list is emitted (via Load Register Immediate commands) - * everytime a new context is created. - * - GT workarounds. The list of these WAs is applied whenever these registers - * revert to default values (on GPU reset, suspend/resume [2]_, etc..). - * - Display workarounds. The list is applied during display clock-gating - * initialization. - * - Workarounds that whitelist a privileged register, so that UMDs can manage - * them directly. This is just a special case of a MMMIO workaround (as we - * write the list of these to/be-whitelisted registers to some special HW - * registers). - * - Workaround batchbuffers, that get executed automatically by the hardware - * on every HW context restore. + * - Context workarounds: workarounds that touch registers that are + * saved/restored to/from the HW context image. The list is emitted (via Load + * Register Immediate commands) once when initializing the device and saved in + * the default context. That default context is then used on every context + * creation to have a "primed golden context", i.e. a context image that + * already contains the changes needed to all the registers. * - * .. [1] Please notice that there are other WAs that, due to their nature, - * cannot be applied from a central place. Those are peppered around the rest - * of the code, as needed. + * - Engine workarounds: the list of these WAs is applied whenever the specific + * engine is reset. It's also possible that a set of engine classes share a + * common power domain and they are reset together. This happens on some + * platforms with render and compute engines. In this case (at least) one of + * them need to keeep the workaround programming: the approach taken in the + * driver is to tie those workarounds to the first compute/render engine that + * is registered. When executing with GuC submission, engine resets are + * outside of kernel driver control, hence the list of registers involved in + * written once, on engine initialization, and then passed to GuC, that + * saves/restores their values before/after the reset takes place. See + * ``drivers/gpu/drm/i915/gt/uc/intel_guc_ads.c`` for reference. * - * .. [2] Technically, some registers are powercontext saved & restored, so they - * survive a suspend/resume. In practice, writing them again is not too - * costly and simplifies things. We can revisit this in the future. + * - GT workarounds: the list of these WAs is applied whenever these registers + * revert to their default values: on GPU reset, suspend/resume [1]_, etc. + * + * - Register whitelist: some workarounds need to be implemented in userspace, + * but need to touch privileged registers. The whitelist in the kernel + * instructs the hardware to allow the access to happen. From the kernel side, + * this is just a special case of a MMIO workaround (as we write the list of + * these to/be-whitelisted registers to some special HW registers). + * + * - Workaround batchbuffers: buffers that get executed automatically by the + * hardware on every HW context restore. These buffers are created and + * programmed in the default context so the hardware always go through those + * programming sequences when switching contexts. The support for workaround + * batchbuffers is enabled these hardware mechanisms: + * + * #. INDIRECT_CTX: A batchbuffer and an offset are provided in the default + * context, pointing the hardware to jump to that location when that offset + * is reached in the context restore. Workaround batchbuffer in the driver + * currently uses this mechanism for all platforms. * - * Layout - * ~~~~~~ + * #. BB_PER_CTX_PTR: A batchbuffer is provided in the default context, + * pointing the hardware to a buffer to continue executing after the + * engine registers are restored in a context restore sequence. This is + * currently not used in the driver. * - * Keep things in this file ordered by WA type, as per the above (context, GT, - * display, register whitelist, batchbuffer). Then, inside each type, keep the - * following order: + * - Other: There are WAs that, due to their nature, cannot be applied from a + * central place. Those are peppered around the rest of the code, as needed. + * Workarounds related to the display IP are the main example. * - * - Infrastructure functions and macros - * - WAs per platform in standard gen/chrono order - * - Public functions to init or apply the given workaround type. + * .. [1] Technically, some registers are powercontext saved & restored, so they + * survive a suspend/resume. In practice, writing them again is not too + * costly and simplifies things, so it's the approach taken in the driver. */ -static void wa_init_start(struct i915_wa_list *wal, const char *name, const char *engine_name) +static void wa_init_start(struct i915_wa_list *wal, struct intel_gt *gt, + const char *name, const char *engine_name) { + wal->gt = gt; wal->name = name; wal->engine_name = engine_name; } @@ -80,13 +103,14 @@ static void wa_init_finish(struct i915_wa_list *wal) if (!wal->count) return; - DRM_DEBUG_DRIVER("Initialized %u %s workarounds on %s\n", - wal->wa_count, wal->name, wal->engine_name); + drm_dbg(&wal->gt->i915->drm, "Initialized %u %s workarounds on %s\n", + wal->wa_count, wal->name, wal->engine_name); } static void _wa_add(struct i915_wa_list *wal, const struct i915_wa *wa) { unsigned int addr = i915_mmio_reg_offset(wa->reg); + struct drm_i915_private *i915 = wal->gt->i915; unsigned int start = 0, end = wal->count; const unsigned int grow = WA_LIST_CHUNK; struct i915_wa *wa_; @@ -99,7 +123,7 @@ static void _wa_add(struct i915_wa_list *wal, const struct i915_wa *wa) list = kmalloc_array(ALIGN(wal->count + 1, grow), sizeof(*wa), GFP_KERNEL); if (!list) { - DRM_ERROR("No space for workaround init!\n"); + drm_err(&i915->drm, "No space for workaround init!\n"); return; } @@ -122,9 +146,10 @@ static void _wa_add(struct i915_wa_list *wal, const struct i915_wa *wa) wa_ = &wal->list[mid]; if ((wa->clr | wa_->clr) && !(wa->clr & ~wa_->clr)) { - DRM_ERROR("Discarding overwritten w/a for reg %04x (clear: %08x, set: %08x)\n", - i915_mmio_reg_offset(wa_->reg), - wa_->clr, wa_->set); + drm_err(&i915->drm, + "Discarding overwritten w/a for reg %04x (clear: %08x, set: %08x)\n", + i915_mmio_reg_offset(wa_->reg), + wa_->clr, wa_->set); wa_->set &= ~wa->clr; } @@ -826,7 +851,7 @@ __intel_engine_init_ctx_wa(struct intel_engine_cs *engine, { struct drm_i915_private *i915 = engine->i915; - wa_init_start(wal, name, engine->name); + wa_init_start(wal, engine->gt, name, engine->name); /* Applies to all engines */ /* @@ -1676,7 +1701,7 @@ void intel_gt_init_workarounds(struct intel_gt *gt) { struct i915_wa_list *wal = >->wa_list; - wa_init_start(wal, "GT", "global"); + wa_init_start(wal, gt, "GT", "global"); gt_init_workarounds(gt, wal); wa_init_finish(wal); } @@ -1698,12 +1723,14 @@ wal_get_fw_for_rmw(struct intel_uncore *uncore, const struct i915_wa_list *wal) } static bool -wa_verify(const struct i915_wa *wa, u32 cur, const char *name, const char *from) +wa_verify(struct intel_gt *gt, const struct i915_wa *wa, u32 cur, + const char *name, const char *from) { if ((cur ^ wa->set) & wa->read) { - DRM_ERROR("%s workaround lost on %s! (reg[%x]=0x%x, relevant bits were 0x%x vs expected 0x%x)\n", - name, from, i915_mmio_reg_offset(wa->reg), - cur, cur & wa->read, wa->set & wa->read); + drm_err(>->i915->drm, + "%s workaround lost on %s! (reg[%x]=0x%x, relevant bits were 0x%x vs expected 0x%x)\n", + name, from, i915_mmio_reg_offset(wa->reg), + cur, cur & wa->read, wa->set & wa->read); return false; } @@ -1711,9 +1738,9 @@ wa_verify(const struct i915_wa *wa, u32 cur, const char *name, const char *from) return true; } -static void -wa_list_apply(struct intel_gt *gt, const struct i915_wa_list *wal) +static void wa_list_apply(const struct i915_wa_list *wal) { + struct intel_gt *gt = wal->gt; struct intel_uncore *uncore = gt->uncore; enum forcewake_domains fw; unsigned long flags; @@ -1749,7 +1776,7 @@ wa_list_apply(struct intel_gt *gt, const struct i915_wa_list *wal) intel_gt_mcr_read_any_fw(gt, wa->mcr_reg) : intel_uncore_read_fw(uncore, wa->reg); - wa_verify(wa, val, wal->name, "application"); + wa_verify(gt, wa, val, wal->name, "application"); } } @@ -1759,7 +1786,7 @@ wa_list_apply(struct intel_gt *gt, const struct i915_wa_list *wal) void intel_gt_apply_workarounds(struct intel_gt *gt) { - wa_list_apply(gt, >->wa_list); + wa_list_apply(>->wa_list); } static bool wa_list_verify(struct intel_gt *gt, @@ -1779,7 +1806,7 @@ static bool wa_list_verify(struct intel_gt *gt, intel_uncore_forcewake_get__locked(uncore, fw); for (i = 0, wa = wal->list; i < wal->count; i++, wa++) - ok &= wa_verify(wa, wa->is_mcr ? + ok &= wa_verify(wal->gt, wa, wa->is_mcr ? intel_gt_mcr_read_any_fw(gt, wa->mcr_reg) : intel_uncore_read_fw(uncore, wa->reg), wal->name, from); @@ -2127,7 +2154,7 @@ void intel_engine_init_whitelist(struct intel_engine_cs *engine) struct drm_i915_private *i915 = engine->i915; struct i915_wa_list *w = &engine->whitelist; - wa_init_start(w, "whitelist", engine->name); + wa_init_start(w, engine->gt, "whitelist", engine->name); if (IS_PONTEVECCHIO(i915)) pvc_whitelist_build(engine); @@ -3012,14 +3039,14 @@ void intel_engine_init_workarounds(struct intel_engine_cs *engine) if (GRAPHICS_VER(engine->i915) < 4) return; - wa_init_start(wal, "engine", engine->name); + wa_init_start(wal, engine->gt, "engine", engine->name); engine_init_workarounds(engine, wal); wa_init_finish(wal); } void intel_engine_apply_workarounds(struct intel_engine_cs *engine) { - wa_list_apply(engine->gt, &engine->wa_list); + wa_list_apply(&engine->wa_list); } static const struct i915_range mcr_ranges_gen8[] = { @@ -3163,9 +3190,7 @@ retry: goto err_vma; } - err = i915_request_await_object(rq, vma->obj, true); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); if (err == 0) err = wa_list_srm(rq, wal, vma); @@ -3193,7 +3218,7 @@ retry: if (mcr_range(rq->engine->i915, i915_mmio_reg_offset(wa->reg))) continue; - if (!wa_verify(wa, results[i], wal->name, from)) + if (!wa_verify(wal->gt, wa, results[i], wal->name, from)) err = -ENXIO; } diff --git a/drivers/gpu/drm/i915/gt/intel_workarounds_types.h b/drivers/gpu/drm/i915/gt/intel_workarounds_types.h index 7c8b01d00043..e14188120e66 100644 --- a/drivers/gpu/drm/i915/gt/intel_workarounds_types.h +++ b/drivers/gpu/drm/i915/gt/intel_workarounds_types.h @@ -10,6 +10,8 @@ #include "i915_reg_defs.h" +struct intel_gt; + struct i915_wa { union { i915_reg_t reg; @@ -24,6 +26,7 @@ struct i915_wa { }; struct i915_wa_list { + struct intel_gt *gt; const char *name; const char *engine_name; struct i915_wa *list; diff --git a/drivers/gpu/drm/i915/gt/selftest_execlists.c b/drivers/gpu/drm/i915/gt/selftest_execlists.c index 2c7c053a8808..ab2e9a6a2452 100644 --- a/drivers/gpu/drm/i915/gt/selftest_execlists.c +++ b/drivers/gpu/drm/i915/gt/selftest_execlists.c @@ -2764,9 +2764,7 @@ static int create_gang(struct intel_engine_cs *engine, i915_request_get(rq); i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, false); - if (!err) - err = i915_vma_move_to_active(vma, rq, 0); + err = i915_vma_move_to_active(vma, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, vma->node.start, @@ -3180,15 +3178,11 @@ create_gpr_client(struct intel_engine_cs *engine, } i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, false); - if (!err) - err = i915_vma_move_to_active(vma, rq, 0); + err = i915_vma_move_to_active(vma, rq, 0); i915_vma_unlock(vma); i915_vma_lock(batch); if (!err) - err = i915_request_await_object(rq, batch->obj, false); - if (!err) err = i915_vma_move_to_active(batch, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, @@ -3521,9 +3515,7 @@ static int smoke_submit(struct preempt_smoke *smoke, if (vma) { i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, false); - if (!err) - err = i915_vma_move_to_active(vma, rq, 0); + err = i915_vma_move_to_active(vma, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, vma->node.start, diff --git a/drivers/gpu/drm/i915/gt/selftest_hangcheck.c b/drivers/gpu/drm/i915/gt/selftest_hangcheck.c index 71263058a7b0..bc05ef48c194 100644 --- a/drivers/gpu/drm/i915/gt/selftest_hangcheck.c +++ b/drivers/gpu/drm/i915/gt/selftest_hangcheck.c @@ -99,22 +99,6 @@ static u64 hws_address(const struct i915_vma *hws, return hws->node.start + offset_in_page(sizeof(u32)*rq->fence.context); } -static int move_to_active(struct i915_vma *vma, - struct i915_request *rq, - unsigned int flags) -{ - int err; - - i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, - flags & EXEC_OBJECT_WRITE); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, flags); - i915_vma_unlock(vma); - - return err; -} - static struct i915_request * hang_create_request(struct hang *h, struct intel_engine_cs *engine) { @@ -175,11 +159,11 @@ hang_create_request(struct hang *h, struct intel_engine_cs *engine) goto unpin_hws; } - err = move_to_active(vma, rq, 0); + err = igt_vma_move_to_active_unlocked(vma, rq, 0); if (err) goto cancel_rq; - err = move_to_active(hws, rq, 0); + err = igt_vma_move_to_active_unlocked(hws, rq, 0); if (err) goto cancel_rq; @@ -1519,18 +1503,9 @@ static int __igt_reset_evict_vma(struct intel_gt *gt, } } - i915_vma_lock(arg.vma); - err = i915_request_await_object(rq, arg.vma->obj, - flags & EXEC_OBJECT_WRITE); - if (err == 0) { - err = i915_vma_move_to_active(arg.vma, rq, flags); - if (err) - pr_err("[%s] Move to active failed: %d!\n", engine->name, err); - } else { - pr_err("[%s] Request await failed: %d!\n", engine->name, err); - } - - i915_vma_unlock(arg.vma); + err = igt_vma_move_to_active_unlocked(arg.vma, rq, flags); + if (err) + pr_err("[%s] Move to active failed: %d!\n", engine->name, err); if (flags & EXEC_OBJECT_NEEDS_FENCE) i915_vma_unpin_fence(arg.vma); diff --git a/drivers/gpu/drm/i915/gt/selftest_lrc.c b/drivers/gpu/drm/i915/gt/selftest_lrc.c index 82d3f8058995..7c56ffd2c659 100644 --- a/drivers/gpu/drm/i915/gt/selftest_lrc.c +++ b/drivers/gpu/drm/i915/gt/selftest_lrc.c @@ -452,9 +452,7 @@ retry: *cs++ = i915_ggtt_offset(scratch) + RING_TAIL_IDX * sizeof(u32); *cs++ = 0; - err = i915_request_await_object(rq, scratch->obj, true); - if (!err) - err = i915_vma_move_to_active(scratch, rq, EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(scratch, rq, EXEC_OBJECT_WRITE); i915_request_get(rq); i915_request_add(rq); @@ -602,9 +600,7 @@ __gpr_read(struct intel_context *ce, struct i915_vma *scratch, u32 *slot) } i915_vma_lock(scratch); - err = i915_request_await_object(rq, scratch->obj, true); - if (!err) - err = i915_vma_move_to_active(scratch, rq, EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(scratch, rq, EXEC_OBJECT_WRITE); i915_vma_unlock(scratch); i915_request_get(rq); @@ -1053,21 +1049,6 @@ store_context(struct intel_context *ce, struct i915_vma *scratch) return batch; } -static int move_to_active(struct i915_request *rq, - struct i915_vma *vma, - unsigned int flags) -{ - int err; - - i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, flags); - if (!err) - err = i915_vma_move_to_active(vma, rq, flags); - i915_vma_unlock(vma); - - return err; -} - static struct i915_request * record_registers(struct intel_context *ce, struct i915_vma *before, @@ -1093,19 +1074,19 @@ record_registers(struct intel_context *ce, if (IS_ERR(rq)) goto err_after; - err = move_to_active(rq, before, EXEC_OBJECT_WRITE); + err = igt_vma_move_to_active_unlocked(before, rq, EXEC_OBJECT_WRITE); if (err) goto err_rq; - err = move_to_active(rq, b_before, 0); + err = igt_vma_move_to_active_unlocked(b_before, rq, 0); if (err) goto err_rq; - err = move_to_active(rq, after, EXEC_OBJECT_WRITE); + err = igt_vma_move_to_active_unlocked(after, rq, EXEC_OBJECT_WRITE); if (err) goto err_rq; - err = move_to_active(rq, b_after, 0); + err = igt_vma_move_to_active_unlocked(b_after, rq, 0); if (err) goto err_rq; @@ -1243,7 +1224,7 @@ static int poison_registers(struct intel_context *ce, u32 poison, u32 *sema) goto err_batch; } - err = move_to_active(rq, batch, 0); + err = igt_vma_move_to_active_unlocked(batch, rq, 0); if (err) goto err_rq; diff --git a/drivers/gpu/drm/i915/gt/selftest_mocs.c b/drivers/gpu/drm/i915/gt/selftest_mocs.c index c1d861333c44..f27cc28608d4 100644 --- a/drivers/gpu/drm/i915/gt/selftest_mocs.c +++ b/drivers/gpu/drm/i915/gt/selftest_mocs.c @@ -7,6 +7,7 @@ #include "gt/intel_gpu_commands.h" #include "i915_selftest.h" +#include "gem/selftests/igt_gem_utils.h" #include "gem/selftests/mock_context.h" #include "selftests/igt_reset.h" #include "selftests/igt_spinner.h" @@ -228,9 +229,7 @@ static int check_mocs_engine(struct live_mocs *arg, return PTR_ERR(rq); i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, true); - if (!err) - err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); i915_vma_unlock(vma); /* Read the mocs tables back using SRM */ diff --git a/drivers/gpu/drm/i915/gt/selftest_rc6.c b/drivers/gpu/drm/i915/gt/selftest_rc6.c index 8c70b7e12074..2ceeadecc639 100644 --- a/drivers/gpu/drm/i915/gt/selftest_rc6.c +++ b/drivers/gpu/drm/i915/gt/selftest_rc6.c @@ -19,11 +19,11 @@ static u64 rc6_residency(struct intel_rc6 *rc6) /* XXX VLV_GT_MEDIA_RC6? */ - result = intel_rc6_residency_ns(rc6, GEN6_GT_GFX_RC6); + result = intel_rc6_residency_ns(rc6, INTEL_RC6_RES_RC6); if (HAS_RC6p(rc6_to_i915(rc6))) - result += intel_rc6_residency_ns(rc6, GEN6_GT_GFX_RC6p); + result += intel_rc6_residency_ns(rc6, INTEL_RC6_RES_RC6p); if (HAS_RC6pp(rc6_to_i915(rc6))) - result += intel_rc6_residency_ns(rc6, GEN6_GT_GFX_RC6pp); + result += intel_rc6_residency_ns(rc6, INTEL_RC6_RES_RC6pp); return result; } diff --git a/drivers/gpu/drm/i915/gt/selftest_rps.c b/drivers/gpu/drm/i915/gt/selftest_rps.c index 99a372486fb7..39f1b7564170 100644 --- a/drivers/gpu/drm/i915/gt/selftest_rps.c +++ b/drivers/gpu/drm/i915/gt/selftest_rps.c @@ -652,9 +652,7 @@ int live_rps_frequency_cs(void *arg) goto err_vma; } - err = i915_request_await_object(rq, vma->obj, false); - if (!err) - err = i915_vma_move_to_active(vma, rq, 0); + err = i915_vma_move_to_active(vma, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, vma->node.start, @@ -793,9 +791,7 @@ int live_rps_frequency_srm(void *arg) goto err_vma; } - err = i915_request_await_object(rq, vma->obj, false); - if (!err) - err = i915_vma_move_to_active(vma, rq, 0); + err = i915_vma_move_to_active(vma, rq, 0); if (!err) err = rq->engine->emit_bb_start(rq, vma->node.start, diff --git a/drivers/gpu/drm/i915/gt/selftest_slpc.c b/drivers/gpu/drm/i915/gt/selftest_slpc.c index 82ec95a299f6..bd44ce73a504 100644 --- a/drivers/gpu/drm/i915/gt/selftest_slpc.c +++ b/drivers/gpu/drm/i915/gt/selftest_slpc.c @@ -13,6 +13,14 @@ enum test_type { VARY_MAX, MAX_GRANTED, SLPC_POWER, + TILE_INTERACTION, +}; + +struct slpc_thread { + struct kthread_worker *worker; + struct kthread_work work; + struct intel_gt *gt; + int result; }; static int slpc_set_min_freq(struct intel_guc_slpc *slpc, u32 freq) @@ -212,7 +220,8 @@ static int max_granted_freq(struct intel_guc_slpc *slpc, struct intel_rps *rps, *max_act_freq = intel_rps_read_actual_frequency(rps); if (*max_act_freq != slpc->rp0_freq) { /* Check if there was some throttling by pcode */ - perf_limit_reasons = intel_uncore_read(gt->uncore, GT0_PERF_LIMIT_REASONS); + perf_limit_reasons = intel_uncore_read(gt->uncore, + intel_gt_perf_limit_reasons_reg(gt)); /* If not, this is an error */ if (!(perf_limit_reasons & GT0_PERF_LIMIT_REASONS_MASK)) { @@ -310,9 +319,10 @@ static int run_test(struct intel_gt *gt, int test_type) break; case MAX_GRANTED: + case TILE_INTERACTION: /* Media engines have a different RP0 */ - if (engine->class == VIDEO_DECODE_CLASS || - engine->class == VIDEO_ENHANCEMENT_CLASS) { + if (gt->type != GT_MEDIA && (engine->class == VIDEO_DECODE_CLASS || + engine->class == VIDEO_ENHANCEMENT_CLASS)) { igt_spinner_end(&spin); st_engine_heartbeat_enable(engine); err = 0; @@ -335,7 +345,8 @@ static int run_test(struct intel_gt *gt, int test_type) if (max_act_freq <= slpc->min_freq) { pr_err("Actual freq did not rise above min\n"); pr_err("Perf Limit Reasons: 0x%x\n", - intel_uncore_read(gt->uncore, GT0_PERF_LIMIT_REASONS)); + intel_uncore_read(gt->uncore, + intel_gt_perf_limit_reasons_reg(gt))); err = -EINVAL; } } @@ -426,6 +437,56 @@ static int live_slpc_power(void *arg) return ret; } +static void slpc_spinner_thread(struct kthread_work *work) +{ + struct slpc_thread *thread = container_of(work, typeof(*thread), work); + + thread->result = run_test(thread->gt, TILE_INTERACTION); +} + +static int live_slpc_tile_interaction(void *arg) +{ + struct drm_i915_private *i915 = arg; + struct intel_gt *gt; + struct slpc_thread *threads; + int i = 0, ret = 0; + + threads = kcalloc(I915_MAX_GT, sizeof(*threads), GFP_KERNEL); + if (!threads) + return -ENOMEM; + + for_each_gt(gt, i915, i) { + threads[i].worker = kthread_create_worker(0, "igt/slpc_parallel:%d", gt->info.id); + + if (IS_ERR(threads[i].worker)) { + ret = PTR_ERR(threads[i].worker); + break; + } + + threads[i].gt = gt; + kthread_init_work(&threads[i].work, slpc_spinner_thread); + kthread_queue_work(threads[i].worker, &threads[i].work); + } + + for_each_gt(gt, i915, i) { + int status; + + if (IS_ERR_OR_NULL(threads[i].worker)) + continue; + + kthread_flush_work(&threads[i].work); + status = READ_ONCE(threads[i].result); + if (status && !ret) { + pr_err("%s GT %d failed ", __func__, gt->info.id); + ret = status; + } + kthread_destroy_worker(threads[i].worker); + } + + kfree(threads); + return ret; +} + int intel_slpc_live_selftests(struct drm_i915_private *i915) { static const struct i915_subtest tests[] = { @@ -433,6 +494,7 @@ int intel_slpc_live_selftests(struct drm_i915_private *i915) SUBTEST(live_slpc_vary_min), SUBTEST(live_slpc_max_granted), SUBTEST(live_slpc_power), + SUBTEST(live_slpc_tile_interaction), }; struct intel_gt *gt; diff --git a/drivers/gpu/drm/i915/gt/selftest_workarounds.c b/drivers/gpu/drm/i915/gt/selftest_workarounds.c index 21b1edc052f8..96e3861706d6 100644 --- a/drivers/gpu/drm/i915/gt/selftest_workarounds.c +++ b/drivers/gpu/drm/i915/gt/selftest_workarounds.c @@ -66,14 +66,14 @@ reference_lists_init(struct intel_gt *gt, struct wa_lists *lists) memset(lists, 0, sizeof(*lists)); - wa_init_start(&lists->gt_wa_list, "GT_REF", "global"); + wa_init_start(&lists->gt_wa_list, gt, "GT_REF", "global"); gt_init_workarounds(gt, &lists->gt_wa_list); wa_init_finish(&lists->gt_wa_list); for_each_engine(engine, gt, id) { struct i915_wa_list *wal = &lists->engine[id].wa_list; - wa_init_start(wal, "REF", engine->name); + wa_init_start(wal, gt, "REF", engine->name); engine_init_workarounds(engine, wal); wa_init_finish(wal); @@ -139,9 +139,7 @@ read_nonprivs(struct intel_context *ce) } i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, true); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(vma, rq, EXEC_OBJECT_WRITE); i915_vma_unlock(vma); if (err) goto err_req; @@ -632,16 +630,12 @@ retry: goto err_request; } - err = i915_request_await_object(rq, batch->obj, false); - if (err == 0) - err = i915_vma_move_to_active(batch, rq, 0); + err = i915_vma_move_to_active(batch, rq, 0); if (err) goto err_request; - err = i915_request_await_object(rq, scratch->obj, true); - if (err == 0) - err = i915_vma_move_to_active(scratch, rq, - EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(scratch, rq, + EXEC_OBJECT_WRITE); if (err) goto err_request; @@ -860,9 +854,7 @@ static int read_whitelisted_registers(struct intel_context *ce, return PTR_ERR(rq); i915_vma_lock(results); - err = i915_request_await_object(rq, results->obj, true); - if (err == 0) - err = i915_vma_move_to_active(results, rq, EXEC_OBJECT_WRITE); + err = i915_vma_move_to_active(results, rq, EXEC_OBJECT_WRITE); i915_vma_unlock(results); if (err) goto err_req; @@ -944,9 +936,7 @@ static int scrub_whitelisted_registers(struct intel_context *ce) } i915_vma_lock(batch); - err = i915_request_await_object(rq, batch->obj, false); - if (err == 0) - err = i915_vma_move_to_active(batch, rq, 0); + err = i915_vma_move_to_active(batch, rq, 0); i915_vma_unlock(batch); if (err) goto err_request; diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.c b/drivers/gpu/drm/i915/gt/uc/intel_guc.c index 27b09ba1d295..52aede324788 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.c @@ -98,6 +98,8 @@ static void gen9_enable_guc_interrupts(struct intel_guc *guc) gt->pm_guc_events); gen6_gt_pm_enable_irq(gt, gt->pm_guc_events); spin_unlock_irq(gt->irq_lock); + + guc->interrupts.enabled = true; } static void gen9_disable_guc_interrupts(struct intel_guc *guc) @@ -105,6 +107,7 @@ static void gen9_disable_guc_interrupts(struct intel_guc *guc) struct intel_gt *gt = guc_to_gt(guc); assert_rpm_wakelock_held(>->i915->runtime_pm); + guc->interrupts.enabled = false; spin_lock_irq(gt->irq_lock); @@ -116,39 +119,39 @@ static void gen9_disable_guc_interrupts(struct intel_guc *guc) gen9_reset_guc_interrupts(guc); } +static bool __gen11_reset_guc_interrupts(struct intel_gt *gt) +{ + u32 irq = gt->type == GT_MEDIA ? MTL_MGUC : GEN11_GUC; + + lockdep_assert_held(gt->irq_lock); + return gen11_gt_reset_one_iir(gt, 0, irq); +} + static void gen11_reset_guc_interrupts(struct intel_guc *guc) { struct intel_gt *gt = guc_to_gt(guc); spin_lock_irq(gt->irq_lock); - gen11_gt_reset_one_iir(gt, 0, GEN11_GUC); + __gen11_reset_guc_interrupts(gt); spin_unlock_irq(gt->irq_lock); } static void gen11_enable_guc_interrupts(struct intel_guc *guc) { struct intel_gt *gt = guc_to_gt(guc); - u32 events = REG_FIELD_PREP(ENGINE1_MASK, GUC_INTR_GUC2HOST); spin_lock_irq(gt->irq_lock); - WARN_ON_ONCE(gen11_gt_reset_one_iir(gt, 0, GEN11_GUC)); - intel_uncore_write(gt->uncore, - GEN11_GUC_SG_INTR_ENABLE, events); - intel_uncore_write(gt->uncore, - GEN11_GUC_SG_INTR_MASK, ~events); + __gen11_reset_guc_interrupts(gt); spin_unlock_irq(gt->irq_lock); + + guc->interrupts.enabled = true; } static void gen11_disable_guc_interrupts(struct intel_guc *guc) { struct intel_gt *gt = guc_to_gt(guc); - spin_lock_irq(gt->irq_lock); - - intel_uncore_write(gt->uncore, GEN11_GUC_SG_INTR_MASK, ~0); - intel_uncore_write(gt->uncore, GEN11_GUC_SG_INTR_ENABLE, 0); - - spin_unlock_irq(gt->irq_lock); + guc->interrupts.enabled = false; intel_synchronize_irq(gt->i915); gen11_reset_guc_interrupts(guc); @@ -156,7 +159,8 @@ static void gen11_disable_guc_interrupts(struct intel_guc *guc) void intel_guc_init_early(struct intel_guc *guc) { - struct drm_i915_private *i915 = guc_to_gt(guc)->i915; + struct intel_gt *gt = guc_to_gt(guc); + struct drm_i915_private *i915 = gt->i915; intel_uc_fw_init_early(&guc->fw, INTEL_UC_FW_TYPE_GUC); intel_guc_ct_init_early(&guc->ct); @@ -168,12 +172,17 @@ void intel_guc_init_early(struct intel_guc *guc) mutex_init(&guc->send_mutex); spin_lock_init(&guc->irq_lock); if (GRAPHICS_VER(i915) >= 11) { - guc->notify_reg = GEN11_GUC_HOST_INTERRUPT; guc->interrupts.reset = gen11_reset_guc_interrupts; guc->interrupts.enable = gen11_enable_guc_interrupts; guc->interrupts.disable = gen11_disable_guc_interrupts; - guc->send_regs.base = - i915_mmio_reg_offset(GEN11_SOFT_SCRATCH(0)); + if (gt->type == GT_MEDIA) { + guc->notify_reg = MEDIA_GUC_HOST_INTERRUPT; + guc->send_regs.base = i915_mmio_reg_offset(MEDIA_SOFT_SCRATCH(0)); + } else { + guc->notify_reg = GEN11_GUC_HOST_INTERRUPT; + guc->send_regs.base = i915_mmio_reg_offset(GEN11_SOFT_SCRATCH(0)); + } + guc->send_regs.count = GEN11_SOFT_SCRATCH_COUNT; } else { @@ -871,14 +880,14 @@ void intel_guc_load_status(struct intel_guc *guc, struct drm_printer *p) u32 status = intel_uncore_read(uncore, GUC_STATUS); u32 i; - drm_printf(p, "\nGuC status 0x%08x:\n", status); + drm_printf(p, "GuC status 0x%08x:\n", status); drm_printf(p, "\tBootrom status = 0x%x\n", (status & GS_BOOTROM_MASK) >> GS_BOOTROM_SHIFT); drm_printf(p, "\tuKernel status = 0x%x\n", (status & GS_UKERNEL_MASK) >> GS_UKERNEL_SHIFT); drm_printf(p, "\tMIA Core status = 0x%x\n", (status & GS_MIA_MASK) >> GS_MIA_SHIFT); - drm_puts(p, "\nScratch registers:\n"); + drm_puts(p, "Scratch registers:\n"); for (i = 0; i < 16; i++) { drm_printf(p, "\t%2d: \t0x%x\n", i, intel_uncore_read(uncore, SOFT_SCRATCH(i))); diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.h b/drivers/gpu/drm/i915/gt/uc/intel_guc.h index 357873ef692b..1bb3f9829286 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc.h +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.h @@ -78,6 +78,7 @@ struct intel_guc { /** @interrupts: pointers to GuC interrupt-managing functions. */ struct { + bool enabled; void (*reset)(struct intel_guc *guc); void (*enable)(struct intel_guc *guc); void (*disable)(struct intel_guc *guc); @@ -330,9 +331,11 @@ retry: return err; } +/* Only call this from the interrupt handler code */ static inline void intel_guc_to_host_event_handler(struct intel_guc *guc) { - intel_guc_ct_event_handler(&guc->ct); + if (guc->interrupts.enabled) + intel_guc_ct_event_handler(&guc->ct); } /* GuC addresses above GUC_GGTT_TOP also don't map through the GTT */ diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ads.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ads.c index a419d60166c8..a7f737c4792e 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ads.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ads.c @@ -488,6 +488,11 @@ static void fill_engine_enable_masks(struct intel_gt *gt, info_map_write(info_map, engine_enabled_masks[GUC_BLITTER_CLASS], BCS_MASK(gt)); info_map_write(info_map, engine_enabled_masks[GUC_VIDEO_CLASS], VDBOX_MASK(gt)); info_map_write(info_map, engine_enabled_masks[GUC_VIDEOENHANCE_CLASS], VEBOX_MASK(gt)); + + /* The GSC engine is an instance (6) of OTHER_CLASS */ + if (gt->engine[GSC0]) + info_map_write(info_map, engine_enabled_masks[GUC_GSC_OTHER_CLASS], + BIT(gt->engine[GSC0]->instance)); } #define LR_HW_CONTEXT_SIZE (80 * sizeof(u32)) @@ -529,9 +534,6 @@ static int guc_prep_golden_context(struct intel_guc *guc) } for (engine_class = 0; engine_class <= MAX_ENGINE_CLASS; ++engine_class) { - if (engine_class == OTHER_CLASS) - continue; - guc_class = engine_class_to_guc_class(engine_class); if (!info_map_read(&info_map, engine_enabled_masks[guc_class])) @@ -609,9 +611,6 @@ static void guc_init_golden_context(struct intel_guc *guc) addr_ggtt = intel_guc_ggtt_offset(guc, guc->ads_vma) + offset; for (engine_class = 0; engine_class <= MAX_ENGINE_CLASS; ++engine_class) { - if (engine_class == OTHER_CLASS) - continue; - guc_class = engine_class_to_guc_class(engine_class); if (!ads_blob_read(guc, system_info.engine_enabled_masks[guc_class])) continue; diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c index 4e6dca707d94..1d49a7ec0bd8 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_capture.c @@ -132,6 +132,11 @@ static const struct __guc_mmio_reg_descr xe_lpd_blt_inst_regs[] = { COMMON_BASE_ENGINE_INSTANCE, }; +/* XE_LPD - GSC Per-Engine-Instance */ +static const struct __guc_mmio_reg_descr xe_lpd_gsc_inst_regs[] = { + COMMON_BASE_ENGINE_INSTANCE, +}; + /* GEN9 - Global */ static const struct __guc_mmio_reg_descr default_global_regs[] = { COMMON_BASE_GLOBAL, @@ -177,6 +182,8 @@ static struct __guc_mmio_reg_descr_group default_lists[] = { MAKE_REGLIST(xe_lpd_vec_inst_regs, PF, ENGINE_INSTANCE, GUC_VIDEOENHANCE_CLASS), MAKE_REGLIST(empty_regs_list, PF, ENGINE_CLASS, GUC_BLITTER_CLASS), MAKE_REGLIST(xe_lpd_blt_inst_regs, PF, ENGINE_INSTANCE, GUC_BLITTER_CLASS), + MAKE_REGLIST(empty_regs_list, PF, ENGINE_CLASS, GUC_GSC_OTHER_CLASS), + MAKE_REGLIST(xe_lpd_gsc_inst_regs, PF, ENGINE_INSTANCE, GUC_GSC_OTHER_CLASS), {} }; @@ -192,6 +199,8 @@ static const struct __guc_mmio_reg_descr_group xe_lpd_lists[] = { MAKE_REGLIST(xe_lpd_vec_inst_regs, PF, ENGINE_INSTANCE, GUC_VIDEOENHANCE_CLASS), MAKE_REGLIST(empty_regs_list, PF, ENGINE_CLASS, GUC_BLITTER_CLASS), MAKE_REGLIST(xe_lpd_blt_inst_regs, PF, ENGINE_INSTANCE, GUC_BLITTER_CLASS), + MAKE_REGLIST(empty_regs_list, PF, ENGINE_CLASS, GUC_GSC_OTHER_CLASS), + MAKE_REGLIST(xe_lpd_gsc_inst_regs, PF, ENGINE_INSTANCE, GUC_GSC_OTHER_CLASS), {} }; @@ -454,6 +463,8 @@ __stringify_engclass(u32 class) return "Blitter"; case GUC_COMPUTE_CLASS: return "Compute"; + case GUC_GSC_OTHER_CLASS: + return "GSC-Other"; default: break; } diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h index 968ebd79dce7..4ae5fc2f6002 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h @@ -47,7 +47,8 @@ #define GUC_VIDEOENHANCE_CLASS 2 #define GUC_BLITTER_CLASS 3 #define GUC_COMPUTE_CLASS 4 -#define GUC_LAST_ENGINE_CLASS GUC_COMPUTE_CLASS +#define GUC_GSC_OTHER_CLASS 5 +#define GUC_LAST_ENGINE_CLASS GUC_GSC_OTHER_CLASS #define GUC_MAX_ENGINE_CLASSES 16 #define GUC_MAX_INSTANCES_PER_CLASS 32 @@ -169,6 +170,7 @@ static u8 engine_class_guc_class_map[] = { [COPY_ENGINE_CLASS] = GUC_BLITTER_CLASS, [VIDEO_DECODE_CLASS] = GUC_VIDEO_CLASS, [VIDEO_ENHANCEMENT_CLASS] = GUC_VIDEOENHANCE_CLASS, + [OTHER_CLASS] = GUC_GSC_OTHER_CLASS, [COMPUTE_CLASS] = GUC_COMPUTE_CLASS, }; @@ -178,12 +180,13 @@ static u8 guc_class_engine_class_map[] = { [GUC_VIDEO_CLASS] = VIDEO_DECODE_CLASS, [GUC_VIDEOENHANCE_CLASS] = VIDEO_ENHANCEMENT_CLASS, [GUC_COMPUTE_CLASS] = COMPUTE_CLASS, + [GUC_GSC_OTHER_CLASS] = OTHER_CLASS, }; static inline u8 engine_class_to_guc_class(u8 class) { BUILD_BUG_ON(ARRAY_SIZE(engine_class_guc_class_map) != MAX_ENGINE_CLASS + 1); - GEM_BUG_ON(class > MAX_ENGINE_CLASS || class == OTHER_CLASS); + GEM_BUG_ON(class > MAX_ENGINE_CLASS); return engine_class_guc_class_map[class]; } diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_rc.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_rc.c index 8f8dd05835c5..b5855091cf6a 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_rc.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_rc.c @@ -11,9 +11,20 @@ static bool __guc_rc_supported(struct intel_guc *guc) { + struct intel_gt *gt = guc_to_gt(guc); + + /* + * Wa_14017073508: mtl + * Do not enable gucrc to avoid additional interrupts which + * may disrupt pcode wa. + */ + if (IS_MTL_GRAPHICS_STEP(gt->i915, P, STEP_A0, STEP_B0) && + gt->type == GT_MEDIA) + return false; + /* GuC RC is unavailable for pre-Gen12 */ return guc->submission_supported && - GRAPHICS_VER(guc_to_gt(guc)->i915) >= 12; + GRAPHICS_VER(gt->i915) >= 12; } static bool __guc_rc_selected(struct intel_guc *guc) diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_reg.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_reg.h index a7092f711e9c..9915de32e894 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_reg.h +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_reg.h @@ -36,6 +36,7 @@ #define SOFT_SCRATCH_COUNT 16 #define GEN11_SOFT_SCRATCH(n) _MMIO(0x190240 + (n) * 4) +#define MEDIA_SOFT_SCRATCH(n) _MMIO(0x190310 + (n) * 4) #define GEN11_SOFT_SCRATCH_COUNT 4 #define UOS_RSA_SCRATCH(i) _MMIO(0xc200 + (i) * 4) @@ -101,6 +102,7 @@ #define GUC_SEND_INTERRUPT _MMIO(0xc4c8) #define GUC_SEND_TRIGGER (1<<0) #define GEN11_GUC_HOST_INTERRUPT _MMIO(0x1901f0) +#define MEDIA_GUC_HOST_INTERRUPT _MMIO(0x190304) #define GEN12_GUC_SEM_INTR_ENABLES _MMIO(0xc71c) #define GUC_SEM_INTR_ROUTE_TO_GUC BIT(31) diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c index 4ccb29f9ac55..0a42f1807f52 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c @@ -30,6 +30,7 @@ #include "intel_guc_submission.h" #include "i915_drv.h" +#include "i915_reg.h" #include "i915_trace.h" /** @@ -1401,7 +1402,9 @@ static void guc_timestamp_ping(struct work_struct *wrk) /* * Synchronize with gt reset to make sure the worker does not - * corrupt the engine/guc stats. + * corrupt the engine/guc stats. NB: can't actually block waiting + * for a reset to complete as the reset requires flushing out + * this worker thread if started. So waiting would deadlock. */ ret = intel_gt_reset_trylock(gt, &srcu); if (ret) @@ -4111,6 +4114,9 @@ static inline void guc_kernel_context_pin(struct intel_guc *guc, if (context_guc_id_invalid(ce)) pin_guc_id(guc, ce); + if (!test_bit(CONTEXT_GUC_INIT, &ce->flags)) + guc_context_init(ce); + try_context_registration(ce, true); } @@ -4901,7 +4907,7 @@ void intel_guc_submission_print_info(struct intel_guc *guc, drm_printf(p, "GuC Number Outstanding Submission G2H: %u\n", atomic_read(&guc->outstanding_submission_g2h)); - drm_printf(p, "GuC tasklet count: %u\n\n", + drm_printf(p, "GuC tasklet count: %u\n", atomic_read(&sched_engine->tasklet.count)); spin_lock_irqsave(&sched_engine->lock, flags); @@ -4949,7 +4955,7 @@ static inline void guc_log_context(struct drm_printer *p, atomic_read(&ce->pin_count)); drm_printf(p, "\t\tGuC ID Ref Count: %u\n", atomic_read(&ce->guc_id.ref)); - drm_printf(p, "\t\tSchedule State: 0x%x\n\n", + drm_printf(p, "\t\tSchedule State: 0x%x\n", ce->guc_state.sched_state); } @@ -4978,7 +4984,7 @@ void intel_guc_submission_print_context_info(struct intel_guc *guc, READ_ONCE(*ce->parallel.guc.wq_head)); drm_printf(p, "\t\tWQI Tail: %u\n", READ_ONCE(*ce->parallel.guc.wq_tail)); - drm_printf(p, "\t\tWQI Status: %u\n\n", + drm_printf(p, "\t\tWQI Status: %u\n", READ_ONCE(*ce->parallel.guc.wq_status)); } @@ -4986,7 +4992,7 @@ void intel_guc_submission_print_context_info(struct intel_guc *guc, emit_bb_start_parent_no_preempt_mid_batch) { u8 i; - drm_printf(p, "\t\tChildren Go: %u\n\n", + drm_printf(p, "\t\tChildren Go: %u\n", get_children_go_value(ce)); for (i = 0; i < ce->parallel.number_children; ++i) drm_printf(p, "\t\tChildren Join: %u\n", diff --git a/drivers/gpu/drm/i915/gt/uc/intel_huc.c b/drivers/gpu/drm/i915/gt/uc/intel_huc.c index fbc8bae14f76..be855811d85d 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_huc.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_huc.c @@ -211,12 +211,41 @@ void intel_huc_unregister_gsc_notifier(struct intel_huc *huc, struct bus_type *b huc->delayed_load.nb.notifier_call = NULL; } +static bool vcs_supported(struct intel_gt *gt) +{ + intel_engine_mask_t mask = gt->info.engine_mask; + + /* + * We reach here from i915_driver_early_probe for the primary GT before + * its engine mask is set, so we use the device info engine mask for it; + * this means we're not taking VCS fusing into account, but if the + * primary GT supports VCS engines we expect at least one of them to + * remain unfused so we're fine. + * For other GTs we expect the GT-specific mask to be set before we + * call this function. + */ + GEM_BUG_ON(!gt_is_root(gt) && !gt->info.engine_mask); + + if (gt_is_root(gt)) + mask = RUNTIME_INFO(gt->i915)->platform_engine_mask; + else + mask = gt->info.engine_mask; + + return __ENGINE_INSTANCES_MASK(mask, VCS0, I915_MAX_VCS); +} + void intel_huc_init_early(struct intel_huc *huc) { struct drm_i915_private *i915 = huc_to_gt(huc)->i915; + struct intel_gt *gt = huc_to_gt(huc); intel_uc_fw_init_early(&huc->fw, INTEL_UC_FW_TYPE_HUC); + if (!vcs_supported(gt)) { + intel_uc_fw_change_status(&huc->fw, INTEL_UC_FIRMWARE_NOT_SUPPORTED); + return; + } + if (GRAPHICS_VER(i915) >= 11) { huc->status.reg = GEN11_HUC_KERNEL_LOAD_INFO; huc->status.mask = HUC_LOAD_SUCCESSFUL; diff --git a/drivers/gpu/drm/i915/gt/uc/intel_uc.c b/drivers/gpu/drm/i915/gt/uc/intel_uc.c index dbd048b77e19..1d28286e6f06 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_uc.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_uc.c @@ -357,8 +357,8 @@ static int uc_init_wopcm(struct intel_uc *uc) { struct intel_gt *gt = uc_to_gt(uc); struct intel_uncore *uncore = gt->uncore; - u32 base = intel_wopcm_guc_base(>->i915->wopcm); - u32 size = intel_wopcm_guc_size(>->i915->wopcm); + u32 base = intel_wopcm_guc_base(>->wopcm); + u32 size = intel_wopcm_guc_size(>->wopcm); u32 huc_agent = intel_uc_uses_huc(uc) ? HUC_LOADING_AGENT_GUC : 0; u32 mask; int err; @@ -636,8 +636,10 @@ void intel_uc_runtime_suspend(struct intel_uc *uc) { struct intel_guc *guc = &uc->guc; - if (!intel_guc_is_ready(guc)) + if (!intel_guc_is_ready(guc)) { + guc->interrupts.enabled = false; return; + } /* * Wait for any outstanding CTB before tearing down communication /w the @@ -657,8 +659,10 @@ void intel_uc_suspend(struct intel_uc *uc) intel_wakeref_t wakeref; int err; - if (!intel_guc_is_ready(guc)) + if (!intel_guc_is_ready(guc)) { + guc->interrupts.enabled = false; return; + } with_intel_runtime_pm(&uc_to_gt(uc)->i915->runtime_pm, wakeref) { err = intel_guc_suspend(guc); diff --git a/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.c b/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.c index de2843dc1307..0c80ba51a4bd 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.c @@ -478,10 +478,11 @@ static int check_gsc_manifest(const struct firmware *fw, return 0; } -static int check_ccs_header(struct drm_i915_private *i915, +static int check_ccs_header(struct intel_gt *gt, const struct firmware *fw, struct intel_uc_fw *uc_fw) { + struct drm_i915_private *i915 = gt->i915; struct uc_css_header *css; size_t size; @@ -523,10 +524,10 @@ static int check_ccs_header(struct drm_i915_private *i915, /* Sanity check whether this fw is not larger than whole WOPCM memory */ size = __intel_uc_fw_get_upload_size(uc_fw); - if (unlikely(size >= i915->wopcm.size)) { + if (unlikely(size >= gt->wopcm.size)) { drm_warn(&i915->drm, "%s firmware %s: invalid size: %zu > %zu\n", intel_uc_fw_type_repr(uc_fw->type), uc_fw->file_selected.path, - size, (size_t)i915->wopcm.size); + size, (size_t)gt->wopcm.size); return -E2BIG; } @@ -554,7 +555,8 @@ static int check_ccs_header(struct drm_i915_private *i915, */ int intel_uc_fw_fetch(struct intel_uc_fw *uc_fw) { - struct drm_i915_private *i915 = __uc_fw_to_gt(uc_fw)->i915; + struct intel_gt *gt = __uc_fw_to_gt(uc_fw); + struct drm_i915_private *i915 = gt->i915; struct intel_uc_fw_file file_ideal; struct device *dev = i915->drm.dev; struct drm_i915_gem_object *obj; @@ -562,7 +564,7 @@ int intel_uc_fw_fetch(struct intel_uc_fw *uc_fw) bool old_ver = false; int err; - GEM_BUG_ON(!i915->wopcm.size); + GEM_BUG_ON(!gt->wopcm.size); GEM_BUG_ON(!intel_uc_fw_is_enabled(uc_fw)); err = i915_inject_probe_error(i915, -ENXIO); @@ -575,6 +577,17 @@ int intel_uc_fw_fetch(struct intel_uc_fw *uc_fw) err = firmware_request_nowarn(&fw, uc_fw->file_selected.path, dev); memcpy(&file_ideal, &uc_fw->file_wanted, sizeof(file_ideal)); + if (!err && fw->size > INTEL_UC_RSVD_GGTT_PER_FW) { + drm_err(&i915->drm, + "%s firmware %s: size (%zuKB) exceeds max supported size (%uKB)\n", + intel_uc_fw_type_repr(uc_fw->type), uc_fw->file_selected.path, + fw->size / SZ_1K, INTEL_UC_RSVD_GGTT_PER_FW / SZ_1K); + + /* try to find another blob to load */ + release_firmware(fw); + err = -ENOENT; + } + /* Any error is terminal if overriding. Don't bother searching for older versions */ if (err && intel_uc_fw_is_overridden(uc_fw)) goto fail; @@ -604,7 +617,7 @@ int intel_uc_fw_fetch(struct intel_uc_fw *uc_fw) if (uc_fw->loaded_via_gsc) err = check_gsc_manifest(fw, uc_fw); else - err = check_ccs_header(i915, fw, uc_fw); + err = check_ccs_header(gt, fw, uc_fw); if (err) goto fail; @@ -677,14 +690,30 @@ fail: static u32 uc_fw_ggtt_offset(struct intel_uc_fw *uc_fw) { - struct i915_ggtt *ggtt = __uc_fw_to_gt(uc_fw)->ggtt; + struct intel_gt *gt = __uc_fw_to_gt(uc_fw); + struct i915_ggtt *ggtt = gt->ggtt; struct drm_mm_node *node = &ggtt->uc_fw; + u32 offset = uc_fw->type * INTEL_UC_RSVD_GGTT_PER_FW; + + /* + * The media GT shares the GGTT with the root GT, which means that + * we need to use different offsets for the binaries on the media GT. + * To keep the math simple, we use 8MB for the root tile and 8MB for + * the media one. This will need to be updated if we ever have more + * than 1 media GT. + */ + BUILD_BUG_ON(INTEL_UC_FW_NUM_TYPES * INTEL_UC_RSVD_GGTT_PER_FW > SZ_8M); + GEM_BUG_ON(gt->type == GT_MEDIA && gt->info.id > 1); + if (gt->type == GT_MEDIA) + offset += SZ_8M; GEM_BUG_ON(!drm_mm_node_allocated(node)); GEM_BUG_ON(upper_32_bits(node->start)); GEM_BUG_ON(upper_32_bits(node->start + node->size - 1)); + GEM_BUG_ON(offset + uc_fw->obj->base.size > node->size); + GEM_BUG_ON(uc_fw->obj->base.size > INTEL_UC_RSVD_GGTT_PER_FW); - return lower_32_bits(node->start); + return lower_32_bits(node->start + offset); } static void uc_fw_bind_ggtt(struct intel_uc_fw *uc_fw) @@ -699,7 +728,6 @@ static void uc_fw_bind_ggtt(struct intel_uc_fw *uc_fw) dummy->bi.pages = obj->mm.pages; GEM_BUG_ON(!i915_gem_object_has_pinned_pages(obj)); - GEM_BUG_ON(dummy->node_size > ggtt->uc_fw.size); /* uc_fw->obj cache domains were not controlled across suspend */ if (i915_gem_object_has_struct_page(obj)) diff --git a/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.h b/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.h index cb586f7df270..bc898ba5355d 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.h +++ b/drivers/gpu/drm/i915/gt/uc/intel_uc_fw.h @@ -6,6 +6,7 @@ #ifndef _INTEL_UC_FW_H_ #define _INTEL_UC_FW_H_ +#include <linux/sizes.h> #include <linux/types.h> #include "intel_uc_fw_abi.h" #include "intel_device_info.h" @@ -114,6 +115,19 @@ struct intel_uc_fw { (uc)->fw.file_selected.minor_ver, \ (uc)->fw.file_selected.patch_ver)) +/* + * When we load the uC binaries, we pin them in a reserved section at the top of + * the GGTT, which is ~18 MBs. On multi-GT systems where the GTs share the GGTT, + * we also need to make sure that each binary is pinned to a unique location + * during load, because the different GT can go through the FW load at the same + * time (see uc_fw_ggtt_offset() for details). + * Given that the available space is much greater than what is required by the + * binaries, to keep things simple instead of dynamically partitioning the + * reserved section to make space for all the blobs we can just reserve a static + * chunk for each binary. + */ +#define INTEL_UC_RSVD_GGTT_PER_FW SZ_2M + #ifdef CONFIG_DRM_I915_DEBUG_GUC void intel_uc_fw_change_status(struct intel_uc_fw *uc_fw, enum intel_uc_fw_status status); diff --git a/drivers/gpu/drm/i915/gt/uc/selftest_guc_hangcheck.c b/drivers/gpu/drm/i915/gt/uc/selftest_guc_hangcheck.c index 01f8cd3c3134..d91b58f70403 100644 --- a/drivers/gpu/drm/i915/gt/uc/selftest_guc_hangcheck.c +++ b/drivers/gpu/drm/i915/gt/uc/selftest_guc_hangcheck.c @@ -35,11 +35,14 @@ static int intel_hang_guc(void *arg) struct i915_request *rq; intel_wakeref_t wakeref; struct i915_gpu_error *global = >->i915->gpu_error; - struct intel_engine_cs *engine; + struct intel_engine_cs *engine = intel_selftest_find_any_engine(gt); unsigned int reset_count; u32 guc_status; u32 old_beat; + if (!engine) + return 0; + ctx = kernel_context(gt->i915, NULL); if (IS_ERR(ctx)) { drm_err(>->i915->drm, "Failed get kernel context: %ld\n", PTR_ERR(ctx)); @@ -48,14 +51,13 @@ static int intel_hang_guc(void *arg) wakeref = intel_runtime_pm_get(gt->uncore->rpm); - ce = intel_context_create(gt->engine[BCS0]); + ce = intel_context_create(engine); if (IS_ERR(ce)) { ret = PTR_ERR(ce); drm_err(>->i915->drm, "Failed to create spinner request: %d\n", ret); goto err; } - engine = ce->engine; reset_count = i915_reset_count(global); old_beat = engine->props.heartbeat_interval_ms; diff --git a/drivers/gpu/drm/i915/gvt/cfg_space.c b/drivers/gpu/drm/i915/gvt/cfg_space.c index 357c5b65e097..9bafac1eaf48 100644 --- a/drivers/gpu/drm/i915/gvt/cfg_space.c +++ b/drivers/gpu/drm/i915/gvt/cfg_space.c @@ -244,7 +244,7 @@ static void emulate_pci_bar_write(struct intel_vgpu *vgpu, unsigned int offset, } /** - * intel_vgpu_emulate_cfg_read - emulate vGPU configuration space write + * intel_vgpu_emulate_cfg_write - emulate vGPU configuration space write * @vgpu: target vgpu * @offset: offset * @p_data: write data ptr diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.c b/drivers/gpu/drm/i915/gvt/cmd_parser.c index de13f102d4fd..0ebf5fbf0e39 100644 --- a/drivers/gpu/drm/i915/gvt/cmd_parser.c +++ b/drivers/gpu/drm/i915/gvt/cmd_parser.c @@ -37,6 +37,7 @@ #include <linux/slab.h> #include "i915_drv.h" +#include "i915_reg.h" #include "gt/intel_engine_regs.h" #include "gt/intel_gpu_commands.h" #include "gt/intel_gt_regs.h" diff --git a/drivers/gpu/drm/i915/gvt/display.c b/drivers/gpu/drm/i915/gvt/display.c index c7722c818b4d..c033249e73f4 100644 --- a/drivers/gpu/drm/i915/gvt/display.c +++ b/drivers/gpu/drm/i915/gvt/display.c @@ -36,6 +36,8 @@ #include "i915_reg.h" #include "gvt.h" +#include "display/intel_dpio_phy.h" + static int get_edp_pipe(struct intel_vgpu *vgpu) { u32 data = vgpu_vreg(vgpu, _TRANS_DDI_FUNC_CTL_EDP); diff --git a/drivers/gpu/drm/i915/gvt/dmabuf.c b/drivers/gpu/drm/i915/gvt/dmabuf.c index 01e54b45c5c1..355f1c0e8664 100644 --- a/drivers/gpu/drm/i915/gvt/dmabuf.c +++ b/drivers/gpu/drm/i915/gvt/dmabuf.c @@ -88,7 +88,7 @@ static int vgpu_gem_get_pages( sg_dma_address(sg) = dma_addr; } - __i915_gem_object_set_pages(obj, st, PAGE_SIZE); + __i915_gem_object_set_pages(obj, st); out: if (ret) { dma_addr_t dma_addr; diff --git a/drivers/gpu/drm/i915/gvt/dmabuf.h b/drivers/gpu/drm/i915/gvt/dmabuf.h index 5f8f03fb1d1b..3dcdb6570eda 100644 --- a/drivers/gpu/drm/i915/gvt/dmabuf.h +++ b/drivers/gpu/drm/i915/gvt/dmabuf.h @@ -48,7 +48,7 @@ struct intel_vgpu_fb_info { struct intel_vgpu_dmabuf_obj *obj; }; -/** +/* * struct intel_vgpu_dmabuf_obj- Intel vGPU device buffer object */ struct intel_vgpu_dmabuf_obj { diff --git a/drivers/gpu/drm/i915/gvt/firmware.c b/drivers/gpu/drm/i915/gvt/firmware.c index 54fe442238c6..a683c22d5b64 100644 --- a/drivers/gpu/drm/i915/gvt/firmware.c +++ b/drivers/gpu/drm/i915/gvt/firmware.c @@ -104,7 +104,7 @@ static int expose_firmware_sysfs(struct intel_gvt *gvt) memcpy(p, gvt->firmware.mmio, info->mmio_size); - crc32_start = offsetof(struct gvt_firmware_header, crc32) + 4; + crc32_start = offsetof(struct gvt_firmware_header, version); h->crc32 = crc32_le(0, firmware + crc32_start, size - crc32_start); firmware_attr.size = size; diff --git a/drivers/gpu/drm/i915/gvt/gtt.c b/drivers/gpu/drm/i915/gvt/gtt.c index ce0eb03709c3..51e5e8fb505b 100644 --- a/drivers/gpu/drm/i915/gvt/gtt.c +++ b/drivers/gpu/drm/i915/gvt/gtt.c @@ -282,11 +282,6 @@ static inline int get_next_pt_type(int type) return gtt_type_table[type].next_pt_type; } -static inline int get_pt_type(int type) -{ - return gtt_type_table[type].pt_type; -} - static inline int get_entry_type(int type) { return gtt_type_table[type].entry_type; @@ -2785,7 +2780,7 @@ int intel_gvt_init_gtt(struct intel_gvt *gvt) * intel_gvt_clean_gtt - clean up mm components of a GVT device * @gvt: GVT device * - * This function is called at the driver unloading stage, to clean up the + * This function is called at the driver unloading stage, to clean up * the mm components of a GVT device. * */ diff --git a/drivers/gpu/drm/i915/gvt/gvt.h b/drivers/gpu/drm/i915/gvt/gvt.h index dbf8d7470b2c..62823c0e13ab 100644 --- a/drivers/gpu/drm/i915/gvt/gvt.h +++ b/drivers/gpu/drm/i915/gvt/gvt.h @@ -227,8 +227,6 @@ struct intel_vgpu { unsigned long nr_cache_entries; struct mutex cache_lock; - atomic_t released; - struct kvm_page_track_notifier_node track_node; #define NR_BKT (1 << 18) struct hlist_head ptable[NR_BKT]; diff --git a/drivers/gpu/drm/i915/gvt/handlers.c b/drivers/gpu/drm/i915/gvt/handlers.c index 1cb388484bf0..735fc83e7026 100644 --- a/drivers/gpu/drm/i915/gvt/handlers.c +++ b/drivers/gpu/drm/i915/gvt/handlers.c @@ -43,6 +43,7 @@ #include "intel_mchbar_regs.h" #include "display/intel_display_types.h" #include "display/intel_dmc_regs.h" +#include "display/intel_dpio_phy.h" #include "display/intel_fbc.h" #include "display/vlv_dsi_pll_regs.h" #include "gt/intel_gt_regs.h" diff --git a/drivers/gpu/drm/i915/gvt/kvmgt.c b/drivers/gpu/drm/i915/gvt/kvmgt.c index 7a45e5360caf..897b6fdbbaed 100644 --- a/drivers/gpu/drm/i915/gvt/kvmgt.c +++ b/drivers/gpu/drm/i915/gvt/kvmgt.c @@ -684,7 +684,6 @@ static int intel_vgpu_open_device(struct vfio_device *vfio_dev) intel_gvt_activate_vgpu(vgpu); - atomic_set(&vgpu->released, 0); return 0; } @@ -706,9 +705,6 @@ static void intel_vgpu_close_device(struct vfio_device *vfio_dev) if (!vgpu->attached) return; - if (atomic_cmpxchg(&vgpu->released, 0, 1)) - return; - intel_gvt_release_vgpu(vgpu); debugfs_remove(debugfs_lookup(KVMGT_DEBUGFS_FILENAME, vgpu->debugfs)); diff --git a/drivers/gpu/drm/i915/gvt/mmio.c b/drivers/gpu/drm/i915/gvt/mmio.c index 9acc00505fde..5b5def6ddef7 100644 --- a/drivers/gpu/drm/i915/gvt/mmio.c +++ b/drivers/gpu/drm/i915/gvt/mmio.c @@ -37,6 +37,7 @@ #include "i915_reg.h" #include "gvt.h" +#include "display/intel_dpio_phy.h" #include "gt/intel_gt_regs.h" /** diff --git a/drivers/gpu/drm/i915/gvt/mmio_context.c b/drivers/gpu/drm/i915/gvt/mmio_context.c index 200c1162daa3..490e8ae51228 100644 --- a/drivers/gpu/drm/i915/gvt/mmio_context.c +++ b/drivers/gpu/drm/i915/gvt/mmio_context.c @@ -34,6 +34,7 @@ */ #include "i915_drv.h" +#include "i915_reg.h" #include "gt/intel_context.h" #include "gt/intel_engine_regs.h" #include "gt/intel_gpu_commands.h" diff --git a/drivers/gpu/drm/i915/gvt/page_track.c b/drivers/gpu/drm/i915/gvt/page_track.c index 3375b51c75f1..df34e73cba41 100644 --- a/drivers/gpu/drm/i915/gvt/page_track.c +++ b/drivers/gpu/drm/i915/gvt/page_track.c @@ -120,7 +120,7 @@ int intel_vgpu_enable_page_track(struct intel_vgpu *vgpu, unsigned long gfn) } /** - * intel_vgpu_enable_page_track - cancel write-protection on guest page + * intel_vgpu_disable_page_track - cancel write-protection on guest page * @vgpu: a vGPU * @gfn: the gfn of guest page * diff --git a/drivers/gpu/drm/i915/gvt/scheduler.c b/drivers/gpu/drm/i915/gvt/scheduler.c index d6fe94cd0fdb..9cd8fcbf7cad 100644 --- a/drivers/gpu/drm/i915/gvt/scheduler.c +++ b/drivers/gpu/drm/i915/gvt/scheduler.c @@ -570,9 +570,8 @@ retry: if (gmadr_bytes == 8) bb->bb_start_cmd_va[2] = 0; - ret = i915_vma_move_to_active(bb->vma, - workload->req, - 0); + ret = i915_vma_move_to_active(bb->vma, workload->req, + __EXEC_OBJECT_NO_REQUEST_AWAIT); if (ret) goto err; diff --git a/drivers/gpu/drm/i915/gvt/vgpu.c b/drivers/gpu/drm/i915/gvt/vgpu.c index 56c71474008a..3c529c2705dd 100644 --- a/drivers/gpu/drm/i915/gvt/vgpu.c +++ b/drivers/gpu/drm/i915/gvt/vgpu.c @@ -158,7 +158,7 @@ void intel_gvt_clean_vgpu_types(struct intel_gvt *gvt) } /** - * intel_gvt_active_vgpu - activate a virtual GPU + * intel_gvt_activate_vgpu - activate a virtual GPU * @vgpu: virtual GPU * * This function is called when user wants to activate a virtual GPU. @@ -172,7 +172,7 @@ void intel_gvt_activate_vgpu(struct intel_vgpu *vgpu) } /** - * intel_gvt_deactive_vgpu - deactivate a virtual GPU + * intel_gvt_deactivate_vgpu - deactivate a virtual GPU * @vgpu: virtual GPU * * This function is called when user wants to deactivate a virtual GPU. @@ -295,7 +295,7 @@ out_free_vgpu: } /** - * intel_gvt_destroy_vgpu - destroy an idle virtual GPU + * intel_gvt_destroy_idle_vgpu - destroy an idle virtual GPU * @vgpu: virtual GPU * * This function is called when user wants to destroy an idle virtual GPU. diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c index ae987e92251d..6c7ac73b69a5 100644 --- a/drivers/gpu/drm/i915/i915_debugfs.c +++ b/drivers/gpu/drm/i915/i915_debugfs.c @@ -688,8 +688,8 @@ i915_drop_caches_set(void *data, u64 val) unsigned int flags; int ret; - DRM_DEBUG("Dropping caches: 0x%08llx [0x%08llx]\n", - val, val & DROP_ALL); + drm_dbg(&i915->drm, "Dropping caches: 0x%08llx [0x%08llx]\n", + val, val & DROP_ALL); ret = gt_drop_caches(to_gt(i915), val); if (ret) diff --git a/drivers/gpu/drm/i915/i915_driver.c b/drivers/gpu/drm/i915/i915_driver.c index c3d43f9b1e45..69103ae37779 100644 --- a/drivers/gpu/drm/i915/i915_driver.c +++ b/drivers/gpu/drm/i915/i915_driver.c @@ -372,8 +372,6 @@ static int i915_driver_early_probe(struct drm_i915_private *dev_priv) if (ret) goto err_ttm; - intel_wopcm_init_early(&dev_priv->wopcm); - ret = intel_root_gt_init_early(dev_priv); if (ret < 0) goto err_rootgt; diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index 05b3300cc4ed..a380db36d52c 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -62,7 +62,6 @@ #include "intel_runtime_pm.h" #include "intel_step.h" #include "intel_uncore.h" -#include "intel_wopcm.h" struct drm_i915_clock_gating_funcs; struct drm_i915_gem_object; @@ -235,8 +234,6 @@ struct drm_i915_private { struct intel_gvt *gvt; - struct intel_wopcm wopcm; - struct pci_dev *bridge_dev; struct rb_root uabi_engines; @@ -287,28 +284,13 @@ struct drm_i915_private { unsigned long gem_quirks; - struct drm_atomic_state *modeset_restore_state; - struct drm_modeset_acquire_ctx reset_ctx; - struct i915_gem_mm mm; - /* Kernel Modesetting */ - - struct list_head global_obj_list; - bool mchbar_need_disable; struct intel_l3_parity l3_parity; /* - * HTI (aka HDPORT) state read during initial hw readout. Most - * platforms don't have HTI, so this will just stay 0. Those that do - * will use this later to figure out which PLLs and PHYs are unavailable - * for driver usage. - */ - u32 hti_state; - - /* * edram size in MB. * Cannot be determined by PCIID. You must always read a register. */ @@ -740,6 +722,10 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915, #define IS_XEHPSDV_GRAPHICS_STEP(__i915, since, until) \ (IS_XEHPSDV(__i915) && IS_GRAPHICS_STEP(__i915, since, until)) +#define IS_MTL_GRAPHICS_STEP(__i915, variant, since, until) \ + (IS_SUBPLATFORM(__i915, INTEL_METEORLAKE, INTEL_SUBPLATFORM_##variant) && \ + IS_GRAPHICS_STEP(__i915, since, until)) + /* * DG2 hardware steppings are a bit unusual. The hardware design was forked to * create three variants (G10, G11, and G12) which each have distinct @@ -778,12 +764,15 @@ IS_SUBPLATFORM(const struct drm_i915_private *i915, #define __HAS_ENGINE(engine_mask, id) ((engine_mask) & BIT(id)) #define HAS_ENGINE(gt, id) __HAS_ENGINE((gt)->info.engine_mask, id) -#define ENGINE_INSTANCES_MASK(gt, first, count) ({ \ +#define __ENGINE_INSTANCES_MASK(mask, first, count) ({ \ unsigned int first__ = (first); \ unsigned int count__ = (count); \ - ((gt)->info.engine_mask & \ - GENMASK(first__ + count__ - 1, first__)) >> first__; \ + ((mask) & GENMASK(first__ + count__ - 1, first__)) >> first__; \ }) + +#define ENGINE_INSTANCES_MASK(gt, first, count) \ + __ENGINE_INSTANCES_MASK((gt)->info.engine_mask, first, count) + #define RCS_MASK(gt) \ ENGINE_INSTANCES_MASK(gt, RCS0, I915_MAX_RCS) #define BCS_MASK(gt) \ diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 299f94a9fb87..8468ca9885fd 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -1140,8 +1140,10 @@ int i915_gem_init(struct drm_i915_private *dev_priv) if (ret) return ret; - intel_uc_fetch_firmwares(&to_gt(dev_priv)->uc); - intel_wopcm_init(&dev_priv->wopcm); + for_each_gt(gt, dev_priv, i) { + intel_uc_fetch_firmwares(>->uc); + intel_wopcm_init(>->wopcm); + } ret = i915_init_ggtt(dev_priv); if (ret) { @@ -1286,7 +1288,7 @@ int i915_gem_open(struct drm_i915_private *i915, struct drm_file *file) struct i915_drm_client *client; int ret = -ENOMEM; - DRM_DEBUG("\n"); + drm_dbg(&i915->drm, "\n"); file_priv = kzalloc(sizeof(*file_priv), GFP_KERNEL); if (!file_priv) diff --git a/drivers/gpu/drm/i915/i915_getparam.c b/drivers/gpu/drm/i915/i915_getparam.c index 3047e80e1163..61ef2d9cfa62 100644 --- a/drivers/gpu/drm/i915/i915_getparam.c +++ b/drivers/gpu/drm/i915/i915_getparam.c @@ -179,7 +179,7 @@ int i915_getparam_ioctl(struct drm_device *dev, void *data, value = i915_perf_oa_timestamp_frequency(i915); break; default: - DRM_DEBUG("Unknown parameter %d\n", param->param); + drm_dbg(&i915->drm, "Unknown parameter %d\n", param->param); return -EINVAL; } diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c index f2d53edcd2ee..9d5d5a397b64 100644 --- a/drivers/gpu/drm/i915/i915_gpu_error.c +++ b/drivers/gpu/drm/i915/i915_gpu_error.c @@ -55,6 +55,7 @@ #include "i915_drv.h" #include "i915_gpu_error.h" #include "i915_memcpy.h" +#include "i915_reg.h" #include "i915_scatterlist.h" #include "i915_utils.h" diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c index d68859866bf2..edfe363af838 100644 --- a/drivers/gpu/drm/i915/i915_irq.c +++ b/drivers/gpu/drm/i915/i915_irq.c @@ -28,7 +28,6 @@ #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt -#include <linux/circ_buf.h> #include <linux/slab.h> #include <linux/sysrq.h> @@ -248,7 +247,7 @@ void gen3_irq_reset(struct intel_uncore *uncore, i915_reg_t imr, intel_uncore_posting_read(uncore, iir); } -void gen2_irq_reset(struct intel_uncore *uncore) +static void gen2_irq_reset(struct intel_uncore *uncore) { intel_uncore_write16(uncore, GEN2_IMR, 0xffff); intel_uncore_posting_read16(uncore, GEN2_IMR); @@ -309,8 +308,8 @@ void gen3_irq_init(struct intel_uncore *uncore, intel_uncore_posting_read(uncore, imr); } -void gen2_irq_init(struct intel_uncore *uncore, - u32 imr_val, u32 ier_val) +static void gen2_irq_init(struct intel_uncore *uncore, + u32 imr_val, u32 ier_val) { gen2_assert_iir_is_zero(uncore); @@ -1086,8 +1085,9 @@ static void ivb_parity_work(struct work_struct *work) kobject_uevent_env(&dev_priv->drm.primary->kdev->kobj, KOBJ_CHANGE, parity_event); - DRM_DEBUG("Parity error: Slice = %d, Row = %d, Bank = %d, Sub bank = %d.\n", - slice, row, bank, subbank); + drm_dbg(&dev_priv->drm, + "Parity error: Slice = %d, Row = %d, Bank = %d, Sub bank = %d.\n", + slice, row, bank, subbank); kfree(parity_event[4]); kfree(parity_event[3]); @@ -2774,7 +2774,8 @@ static irqreturn_t dg1_irq_handler(int irq, void *arg) master_ctl = raw_reg_read(regs, GEN11_GFX_MSTR_IRQ); raw_reg_write(regs, GEN11_GFX_MSTR_IRQ, master_ctl); } else { - DRM_ERROR("Tile not supported: 0x%08x\n", master_tile_ctl); + drm_err(&i915->drm, "Tile not supported: 0x%08x\n", + master_tile_ctl); dg1_master_intr_enable(regs); return IRQ_NONE; } @@ -3871,7 +3872,7 @@ static void i8xx_irq_reset(struct drm_i915_private *dev_priv) i9xx_pipestat_irq_reset(dev_priv); - GEN2_IRQ_RESET(uncore); + gen2_irq_reset(uncore); dev_priv->irq_mask = ~0u; } @@ -3897,7 +3898,7 @@ static void i8xx_irq_postinstall(struct drm_i915_private *dev_priv) I915_MASTER_ERROR_INTERRUPT | I915_USER_INTERRUPT; - GEN2_IRQ_INIT(uncore, dev_priv->irq_mask, enable_mask); + gen2_irq_init(uncore, dev_priv->irq_mask, enable_mask); /* Interrupt setup is already guaranteed to be single-threaded, this is * just to make the assert_spin_locked check happy. */ @@ -3940,7 +3941,7 @@ static void i8xx_error_irq_ack(struct drm_i915_private *i915, static void i8xx_error_irq_handler(struct drm_i915_private *dev_priv, u16 eir, u16 eir_stuck) { - DRM_DEBUG("Master Error: EIR 0x%04x\n", eir); + drm_dbg(&dev_priv->drm, "Master Error: EIR 0x%04x\n", eir); if (eir_stuck) drm_dbg(&dev_priv->drm, "EIR stuck: 0x%04x, masked\n", @@ -3975,7 +3976,7 @@ static void i9xx_error_irq_ack(struct drm_i915_private *dev_priv, static void i9xx_error_irq_handler(struct drm_i915_private *dev_priv, u32 eir, u32 eir_stuck) { - DRM_DEBUG("Master Error, EIR 0x%08x\n", eir); + drm_dbg(&dev_priv->drm, "Master Error, EIR 0x%08x\n", eir); if (eir_stuck) drm_dbg(&dev_priv->drm, "EIR stuck: 0x%08x, masked\n", diff --git a/drivers/gpu/drm/i915/i915_irq.h b/drivers/gpu/drm/i915/i915_irq.h index 82639d9d7e82..9b004fc3444e 100644 --- a/drivers/gpu/drm/i915/i915_irq.h +++ b/drivers/gpu/drm/i915/i915_irq.h @@ -90,12 +90,9 @@ void i965_disable_vblank(struct drm_crtc *crtc); void ilk_disable_vblank(struct drm_crtc *crtc); void bdw_disable_vblank(struct drm_crtc *crtc); -void gen2_irq_reset(struct intel_uncore *uncore); void gen3_irq_reset(struct intel_uncore *uncore, i915_reg_t imr, i915_reg_t iir, i915_reg_t ier); -void gen2_irq_init(struct intel_uncore *uncore, - u32 imr_val, u32 ier_val); void gen3_irq_init(struct intel_uncore *uncore, i915_reg_t imr, u32 imr_val, i915_reg_t ier, u32 ier_val, @@ -111,9 +108,6 @@ void gen3_irq_init(struct intel_uncore *uncore, #define GEN3_IRQ_RESET(uncore, type) \ gen3_irq_reset((uncore), type##IMR, type##IIR, type##IER) -#define GEN2_IRQ_RESET(uncore) \ - gen2_irq_reset(uncore) - #define GEN8_IRQ_INIT_NDX(uncore, type, which, imr_val, ier_val) \ ({ \ unsigned int which_ = which; \ @@ -129,7 +123,4 @@ void gen3_irq_init(struct intel_uncore *uncore, type##IER, ier_val, \ type##IIR) -#define GEN2_IRQ_INIT(uncore, imr_val, ier_val) \ - gen2_irq_init((uncore), imr_val, ier_val) - #endif /* __I915_IRQ_H__ */ diff --git a/drivers/gpu/drm/i915/i915_pci.c b/drivers/gpu/drm/i915/i915_pci.c index 211913be40ce..6da9784fe4a2 100644 --- a/drivers/gpu/drm/i915/i915_pci.c +++ b/drivers/gpu/drm/i915/i915_pci.c @@ -1078,7 +1078,6 @@ static const struct intel_device_info dg2_info = { XE_LPD_FEATURES, .__runtime.cpu_transcoder_mask = BIT(TRANSCODER_A) | BIT(TRANSCODER_B) | BIT(TRANSCODER_C) | BIT(TRANSCODER_D), - .require_force_probe = 1, }; static const struct intel_device_info ats_m_info = { @@ -1146,6 +1145,7 @@ static const struct intel_device_info mtl_info = { .extra_gt_list = xelpmp_extra_gt, .has_flat_ccs = 0, .has_gmd_id = 1, + .has_guc_deprivilege = 1, .has_mslice_steering = 0, .has_snoop = 1, .__runtime.memory_regions = REGION_SMEM | REGION_STOLEN_LMEM, diff --git a/drivers/gpu/drm/i915/i915_perf.c b/drivers/gpu/drm/i915/i915_perf.c index 0dd597a7a11f..00e09bb18b13 100644 --- a/drivers/gpu/drm/i915/i915_perf.c +++ b/drivers/gpu/drm/i915/i915_perf.c @@ -215,6 +215,7 @@ #include "i915_file_private.h" #include "i915_perf.h" #include "i915_perf_oa_regs.h" +#include "i915_reg.h" /* HW requires this to be a power of two, between 128k and 16M, though driver * is currently generally designed assuming the largest 16M size is used such @@ -530,9 +531,9 @@ static bool oa_buffer_check_unlocked(struct i915_perf_stream *stream) if (OA_TAKEN(hw_tail, tail) > report_size && __ratelimit(&stream->perf->tail_pointer_race)) - DRM_NOTE("unlanded report(s) head=0x%x " - "tail=0x%x hw_tail=0x%x\n", - head, tail, hw_tail); + drm_notice(&stream->uncore->i915->drm, + "unlanded report(s) head=0x%x tail=0x%x hw_tail=0x%x\n", + head, tail, hw_tail); stream->oa_buffer.tail = gtt_offset + tail; stream->oa_buffer.aging_tail = gtt_offset + hw_tail; @@ -1015,7 +1016,8 @@ static int gen7_append_oa_reports(struct i915_perf_stream *stream, */ if (report32[0] == 0) { if (__ratelimit(&stream->perf->spurious_report_rs)) - DRM_NOTE("Skipping spurious, invalid OA report\n"); + drm_notice(&uncore->i915->drm, + "Skipping spurious, invalid OA report\n"); continue; } @@ -1602,8 +1604,9 @@ static void i915_oa_stream_destroy(struct i915_perf_stream *stream) free_noa_wait(stream); if (perf->spurious_report_rs.missed) { - DRM_NOTE("%d spurious OA report notices suppressed due to ratelimiting\n", - perf->spurious_report_rs.missed); + drm_notice(>->i915->drm, + "%d spurious OA report notices suppressed due to ratelimiting\n", + perf->spurious_report_rs.missed); } } @@ -2251,9 +2254,7 @@ retry: goto err_add_request; } - err = i915_request_await_object(rq, vma->obj, 0); - if (!err) - err = i915_vma_move_to_active(vma, rq, 0); + err = i915_vma_move_to_active(vma, rq, 0); if (err) goto err_add_request; diff --git a/drivers/gpu/drm/i915/i915_perf_types.h b/drivers/gpu/drm/i915/i915_perf_types.h index e0c96b44eda8..ca150b7af3f2 100644 --- a/drivers/gpu/drm/i915/i915_perf_types.h +++ b/drivers/gpu/drm/i915/i915_perf_types.h @@ -146,8 +146,8 @@ struct i915_perf_stream { */ struct intel_engine_cs *engine; - /* - * Lock associated with operations on stream + /** + * @lock: Lock associated with operations on stream */ struct mutex lock; diff --git a/drivers/gpu/drm/i915/i915_pmu.c b/drivers/gpu/drm/i915/i915_pmu.c index 958b37123bf1..52531ab28c5f 100644 --- a/drivers/gpu/drm/i915/i915_pmu.c +++ b/drivers/gpu/drm/i915/i915_pmu.c @@ -148,13 +148,13 @@ static u64 __get_rc6(struct intel_gt *gt) struct drm_i915_private *i915 = gt->i915; u64 val; - val = intel_rc6_residency_ns(>->rc6, GEN6_GT_GFX_RC6); + val = intel_rc6_residency_ns(>->rc6, INTEL_RC6_RES_RC6); if (HAS_RC6p(i915)) - val += intel_rc6_residency_ns(>->rc6, GEN6_GT_GFX_RC6p); + val += intel_rc6_residency_ns(>->rc6, INTEL_RC6_RES_RC6p); if (HAS_RC6pp(i915)) - val += intel_rc6_residency_ns(>->rc6, GEN6_GT_GFX_RC6pp); + val += intel_rc6_residency_ns(>->rc6, INTEL_RC6_RES_RC6pp); return val; } @@ -371,7 +371,6 @@ static void frequency_sample(struct intel_gt *gt, unsigned int period_ns) { struct drm_i915_private *i915 = gt->i915; - struct intel_uncore *uncore = gt->uncore; struct i915_pmu *pmu = &i915->pmu; struct intel_rps *rps = >->rps; @@ -394,7 +393,7 @@ frequency_sample(struct intel_gt *gt, unsigned int period_ns) * case we assume the system is running at the intended * frequency. Fortunately, the read should rarely fail! */ - val = intel_uncore_read_fw(uncore, GEN6_RPSTAT1); + val = intel_rps_read_rpstat_fw(rps); if (val) val = intel_rps_get_cagf(rps, val); else diff --git a/drivers/gpu/drm/i915/i915_query.c b/drivers/gpu/drm/i915/i915_query.c index 6ec9c9fb7b0d..00871ef99792 100644 --- a/drivers/gpu/drm/i915/i915_query.c +++ b/drivers/gpu/drm/i915/i915_query.c @@ -250,8 +250,9 @@ static int query_perf_config_data(struct drm_i915_private *i915, return total_size; if (query_item->length < total_size) { - DRM_DEBUG("Invalid query config data item size=%u expected=%u\n", - query_item->length, total_size); + drm_dbg(&i915->drm, + "Invalid query config data item size=%u expected=%u\n", + query_item->length, total_size); return -EINVAL; } @@ -418,9 +419,10 @@ static int query_perf_config_list(struct drm_i915_private *i915, } while (n_configs > alloc); if (query_item->length < sizeof_perf_config_list(n_configs)) { - DRM_DEBUG("Invalid query config list item size=%u expected=%zu\n", - query_item->length, - sizeof_perf_config_list(n_configs)); + drm_dbg(&i915->drm, + "Invalid query config list item size=%u expected=%zu\n", + query_item->length, + sizeof_perf_config_list(n_configs)); kfree(oa_config_ids); return -EINVAL; } diff --git a/drivers/gpu/drm/i915/i915_reg.h b/drivers/gpu/drm/i915/i915_reg.h index 1c0da50c0dc7..8e1892d14774 100644 --- a/drivers/gpu/drm/i915/i915_reg.h +++ b/drivers/gpu/drm/i915/i915_reg.h @@ -26,6 +26,7 @@ #define _I915_REG_H_ #include "i915_reg_defs.h" +#include "display/intel_display_reg_defs.h" /** * DOC: The i915 register macro definition style guide @@ -115,75 +116,6 @@ * #define GEN8_BAR _MMIO(0xb888) */ -#define DISPLAY_MMIO_BASE(dev_priv) (INTEL_INFO(dev_priv)->display.mmio_offset) - -/* - * Given the first two numbers __a and __b of arbitrarily many evenly spaced - * numbers, pick the 0-based __index'th value. - * - * Always prefer this over _PICK() if the numbers are evenly spaced. - */ -#define _PICK_EVEN(__index, __a, __b) ((__a) + (__index) * ((__b) - (__a))) - -/* - * Given the arbitrary numbers in varargs, pick the 0-based __index'th number. - * - * Always prefer _PICK_EVEN() over this if the numbers are evenly spaced. - */ -#define _PICK(__index, ...) (((const u32 []){ __VA_ARGS__ })[__index]) - -/* - * Named helper wrappers around _PICK_EVEN() and _PICK(). - */ -#define _PIPE(pipe, a, b) _PICK_EVEN(pipe, a, b) -#define _PLANE(plane, a, b) _PICK_EVEN(plane, a, b) -#define _TRANS(tran, a, b) _PICK_EVEN(tran, a, b) -#define _PORT(port, a, b) _PICK_EVEN(port, a, b) -#define _PLL(pll, a, b) _PICK_EVEN(pll, a, b) -#define _PHY(phy, a, b) _PICK_EVEN(phy, a, b) - -#define _MMIO_PIPE(pipe, a, b) _MMIO(_PIPE(pipe, a, b)) -#define _MMIO_PLANE(plane, a, b) _MMIO(_PLANE(plane, a, b)) -#define _MMIO_TRANS(tran, a, b) _MMIO(_TRANS(tran, a, b)) -#define _MMIO_PORT(port, a, b) _MMIO(_PORT(port, a, b)) -#define _MMIO_PLL(pll, a, b) _MMIO(_PLL(pll, a, b)) -#define _MMIO_PHY(phy, a, b) _MMIO(_PHY(phy, a, b)) - -#define _PHY3(phy, ...) _PICK(phy, __VA_ARGS__) - -#define _MMIO_PIPE3(pipe, a, b, c) _MMIO(_PICK(pipe, a, b, c)) -#define _MMIO_PORT3(pipe, a, b, c) _MMIO(_PICK(pipe, a, b, c)) -#define _MMIO_PHY3(phy, a, b, c) _MMIO(_PHY3(phy, a, b, c)) -#define _MMIO_PLL3(pll, ...) _MMIO(_PICK(pll, __VA_ARGS__)) - - -/* - * Device info offset array based helpers for groups of registers with unevenly - * spaced base offsets. - */ -#define _MMIO_PIPE2(pipe, reg) _MMIO(INTEL_INFO(dev_priv)->display.pipe_offsets[(pipe)] - \ - INTEL_INFO(dev_priv)->display.pipe_offsets[PIPE_A] + \ - DISPLAY_MMIO_BASE(dev_priv) + (reg)) -#define _MMIO_TRANS2(tran, reg) _MMIO(INTEL_INFO(dev_priv)->display.trans_offsets[(tran)] - \ - INTEL_INFO(dev_priv)->display.trans_offsets[TRANSCODER_A] + \ - DISPLAY_MMIO_BASE(dev_priv) + (reg)) -#define _MMIO_CURSOR2(pipe, reg) _MMIO(INTEL_INFO(dev_priv)->display.cursor_offsets[(pipe)] - \ - INTEL_INFO(dev_priv)->display.cursor_offsets[PIPE_A] + \ - DISPLAY_MMIO_BASE(dev_priv) + (reg)) - -#define __MASKED_FIELD(mask, value) ((mask) << 16 | (value)) -#define _MASKED_FIELD(mask, value) ({ \ - if (__builtin_constant_p(mask)) \ - BUILD_BUG_ON_MSG(((mask) & 0xffff0000), "Incorrect mask"); \ - if (__builtin_constant_p(value)) \ - BUILD_BUG_ON_MSG((value) & 0xffff0000, "Incorrect value"); \ - if (__builtin_constant_p(mask) && __builtin_constant_p(value)) \ - BUILD_BUG_ON_MSG((value) & ~(mask), \ - "Incorrect value for mask"); \ - __MASKED_FIELD(mask, value); }) -#define _MASKED_BIT_ENABLE(a) ({ typeof(a) _a = (a); _MASKED_FIELD(_a, _a); }) -#define _MASKED_BIT_DISABLE(a) (_MASKED_FIELD((a), 0)) - #define GU_CNTL _MMIO(0x101010) #define LMEM_INIT REG_BIT(7) @@ -970,6 +902,7 @@ #define GEN11_VEBOX2_RING_BASE 0x1d8000 #define XEHP_VEBOX3_RING_BASE 0x1e8000 #define XEHP_VEBOX4_RING_BASE 0x1f8000 +#define MTL_GSC_RING_BASE 0x11a000 #define GEN12_COMPUTE0_RING_BASE 0x1a000 #define GEN12_COMPUTE1_RING_BASE 0x1c000 #define GEN12_COMPUTE2_RING_BASE 0x1e000 @@ -1147,11 +1080,6 @@ #define MBUS_JOIN_PIPE_SELECT(pipe) REG_FIELD_PREP(MBUS_JOIN_PIPE_SELECT_MASK, pipe) #define MBUS_JOIN_PIPE_SELECT_NONE MBUS_JOIN_PIPE_SELECT(7) -#define HDPORT_STATE _MMIO(0x45050) -#define HDPORT_DPLL_USED_MASK REG_GENMASK(15, 12) -#define HDPORT_DDI_USED(phy) REG_BIT(2 * (phy) + 1) -#define HDPORT_ENABLED REG_BIT(0) - /* Make render/texture TLB fetches lower priorty than associated data * fetches. This is not turned on by default */ @@ -1781,9 +1709,10 @@ #define _PALETTE_A 0xa000 #define _PALETTE_B 0xa800 #define _CHV_PALETTE_C 0xc000 -#define PALETTE_RED_MASK REG_GENMASK(23, 16) -#define PALETTE_GREEN_MASK REG_GENMASK(15, 8) -#define PALETTE_BLUE_MASK REG_GENMASK(7, 0) +/* 8bit mode / i965+ 10.6 interpolated mode ldw/udw */ +#define PALETTE_RED_MASK REG_GENMASK(23, 16) +#define PALETTE_GREEN_MASK REG_GENMASK(15, 8) +#define PALETTE_BLUE_MASK REG_GENMASK(7, 0) #define PALETTE(pipe, i) _MMIO(DISPLAY_MMIO_BASE(dev_priv) + \ _PICK((pipe), _PALETTE_A, \ _PALETTE_B, _CHV_PALETTE_C) + \ @@ -3758,9 +3687,10 @@ /* Skylake+ pipe bottom (background) color */ #define _SKL_BOTTOM_COLOR_A 0x70034 +#define _SKL_BOTTOM_COLOR_B 0x71034 #define SKL_BOTTOM_COLOR_GAMMA_ENABLE REG_BIT(31) #define SKL_BOTTOM_COLOR_CSC_ENABLE REG_BIT(30) -#define SKL_BOTTOM_COLOR(pipe) _MMIO_PIPE2(pipe, _SKL_BOTTOM_COLOR_A) +#define SKL_BOTTOM_COLOR(pipe) _MMIO_PIPE(pipe, _SKL_BOTTOM_COLOR_A, _SKL_BOTTOM_COLOR_B) #define _ICL_PIPE_A_STATUS 0x70058 #define ICL_PIPESTATUS(pipe) _MMIO_PIPE2(pipe, _ICL_PIPE_A_STATUS) @@ -5378,17 +5308,24 @@ /* legacy palette */ #define _LGC_PALETTE_A 0x4a000 #define _LGC_PALETTE_B 0x4a800 -#define LGC_PALETTE_RED_MASK REG_GENMASK(23, 16) -#define LGC_PALETTE_GREEN_MASK REG_GENMASK(15, 8) -#define LGC_PALETTE_BLUE_MASK REG_GENMASK(7, 0) +/* see PALETTE_* for the bits */ #define LGC_PALETTE(pipe, i) _MMIO(_PIPE(pipe, _LGC_PALETTE_A, _LGC_PALETTE_B) + (i) * 4) /* ilk/snb precision palette */ #define _PREC_PALETTE_A 0x4b000 #define _PREC_PALETTE_B 0x4c000 -#define PREC_PALETTE_RED_MASK REG_GENMASK(29, 20) -#define PREC_PALETTE_GREEN_MASK REG_GENMASK(19, 10) -#define PREC_PALETTE_BLUE_MASK REG_GENMASK(9, 0) +/* 10bit mode */ +#define PREC_PALETTE_10_RED_MASK REG_GENMASK(29, 20) +#define PREC_PALETTE_10_GREEN_MASK REG_GENMASK(19, 10) +#define PREC_PALETTE_10_BLUE_MASK REG_GENMASK(9, 0) +/* 12.4 interpolated mode ldw */ +#define PREC_PALETTE_12P4_RED_LDW_MASK REG_GENMASK(29, 24) +#define PREC_PALETTE_12P4_GREEN_LDW_MASK REG_GENMASK(19, 14) +#define PREC_PALETTE_12P4_BLUE_LDW_MASK REG_GENMASK(9, 4) +/* 12.4 interpolated mode udw */ +#define PREC_PALETTE_12P4_RED_UDW_MASK REG_GENMASK(29, 20) +#define PREC_PALETTE_12P4_GREEN_UDW_MASK REG_GENMASK(19, 10) +#define PREC_PALETTE_12P4_BLUE_UDW_MASK REG_GENMASK(9, 0) #define PREC_PALETTE(pipe, i) _MMIO(_PIPE(pipe, _PREC_PALETTE_A, _PREC_PALETTE_B) + (i) * 4) #define _PREC_PIPEAGCMAX 0x4d000 @@ -6678,6 +6615,15 @@ /* XEHP_PCODE_FREQUENCY_CONFIG param2 */ #define PCODE_MBOX_DOMAIN_NONE 0x0 #define PCODE_MBOX_DOMAIN_MEDIAFF 0x3 + +/* Wa_14017210380: mtl */ +#define PCODE_MBOX_GT_STATE 0x50 +/* sub-commands (param1) */ +#define PCODE_MBOX_GT_STATE_MEDIA_BUSY 0x1 +#define PCODE_MBOX_GT_STATE_MEDIA_NOT_BUSY 0x2 +/* param2 */ +#define PCODE_MBOX_GT_STATE_DOMAIN_MEDIA 0x1 + #define GEN6_PCODE_DATA _MMIO(0x138128) #define GEN6_PCODE_FREQ_IA_RATIO_SHIFT 8 #define GEN6_PCODE_FREQ_RING_RATIO_SHIFT 16 @@ -7618,12 +7564,10 @@ enum skl_power_gate { #define _PAL_PREC_DATA_A 0x4A404 #define _PAL_PREC_DATA_B 0x4AC04 #define _PAL_PREC_DATA_C 0x4B404 +/* see PREC_PALETTE_* for the bits */ #define _PAL_PREC_GC_MAX_A 0x4A410 #define _PAL_PREC_GC_MAX_B 0x4AC10 #define _PAL_PREC_GC_MAX_C 0x4B410 -#define PREC_PAL_DATA_RED_MASK REG_GENMASK(29, 20) -#define PREC_PAL_DATA_GREEN_MASK REG_GENMASK(19, 10) -#define PREC_PAL_DATA_BLUE_MASK REG_GENMASK(9, 0) #define _PAL_PREC_EXT_GC_MAX_A 0x4A420 #define _PAL_PREC_EXT_GC_MAX_B 0x4AC20 #define _PAL_PREC_EXT_GC_MAX_C 0x4B420 @@ -7656,12 +7600,7 @@ enum skl_power_gate { #define _PAL_PREC_MULTI_SEG_DATA_A 0x4A40C #define _PAL_PREC_MULTI_SEG_DATA_B 0x4AC0C -#define PAL_PREC_MULTI_SEG_RED_LDW_MASK REG_GENMASK(29, 24) -#define PAL_PREC_MULTI_SEG_RED_UDW_MASK REG_GENMASK(29, 20) -#define PAL_PREC_MULTI_SEG_GREEN_LDW_MASK REG_GENMASK(19, 14) -#define PAL_PREC_MULTI_SEG_GREEN_UDW_MASK REG_GENMASK(19, 10) -#define PAL_PREC_MULTI_SEG_BLUE_LDW_MASK REG_GENMASK(9, 4) -#define PAL_PREC_MULTI_SEG_BLUE_UDW_MASK REG_GENMASK(9, 0) +/* see PREC_PALETTE_12P4_* for the bits */ #define PREC_PAL_MULTI_SEG_INDEX(pipe) _MMIO_PIPE(pipe, \ _PAL_PREC_MULTI_SEG_INDEX_A, \ @@ -7722,13 +7661,17 @@ enum skl_power_gate { #define _CGM_PIPE_A_CSC_COEFF67 (VLV_DISPLAY_BASE + 0x6790C) #define _CGM_PIPE_A_CSC_COEFF8 (VLV_DISPLAY_BASE + 0x67910) #define _CGM_PIPE_A_DEGAMMA (VLV_DISPLAY_BASE + 0x66000) -#define CGM_PIPE_DEGAMMA_RED_MASK REG_GENMASK(13, 0) -#define CGM_PIPE_DEGAMMA_GREEN_MASK REG_GENMASK(29, 16) -#define CGM_PIPE_DEGAMMA_BLUE_MASK REG_GENMASK(13, 0) +/* cgm degamma ldw */ +#define CGM_PIPE_DEGAMMA_GREEN_LDW_MASK REG_GENMASK(29, 16) +#define CGM_PIPE_DEGAMMA_BLUE_LDW_MASK REG_GENMASK(13, 0) +/* cgm degamma udw */ +#define CGM_PIPE_DEGAMMA_RED_UDW_MASK REG_GENMASK(13, 0) #define _CGM_PIPE_A_GAMMA (VLV_DISPLAY_BASE + 0x67000) -#define CGM_PIPE_GAMMA_RED_MASK REG_GENMASK(9, 0) -#define CGM_PIPE_GAMMA_GREEN_MASK REG_GENMASK(25, 16) -#define CGM_PIPE_GAMMA_BLUE_MASK REG_GENMASK(9, 0) +/* cgm gamma ldw */ +#define CGM_PIPE_GAMMA_GREEN_LDW_MASK REG_GENMASK(25, 16) +#define CGM_PIPE_GAMMA_BLUE_LDW_MASK REG_GENMASK(9, 0) +/* cgm gamma udw */ +#define CGM_PIPE_GAMMA_RED_UDW_MASK REG_GENMASK(9, 0) #define _CGM_PIPE_A_MODE (VLV_DISPLAY_BASE + 0x67A00) #define CGM_PIPE_MODE_GAMMA (1 << 2) #define CGM_PIPE_MODE_CSC (1 << 1) diff --git a/drivers/gpu/drm/i915/i915_reg_defs.h b/drivers/gpu/drm/i915/i915_reg_defs.h index f1859046a9c4..be43580a6979 100644 --- a/drivers/gpu/drm/i915/i915_reg_defs.h +++ b/drivers/gpu/drm/i915/i915_reg_defs.h @@ -98,6 +98,34 @@ */ #define REG_FIELD_GET64(__mask, __val) ((u64)FIELD_GET(__mask, __val)) +#define __MASKED_FIELD(mask, value) ((mask) << 16 | (value)) +#define _MASKED_FIELD(mask, value) ({ \ + if (__builtin_constant_p(mask)) \ + BUILD_BUG_ON_MSG(((mask) & 0xffff0000), "Incorrect mask"); \ + if (__builtin_constant_p(value)) \ + BUILD_BUG_ON_MSG((value) & 0xffff0000, "Incorrect value"); \ + if (__builtin_constant_p(mask) && __builtin_constant_p(value)) \ + BUILD_BUG_ON_MSG((value) & ~(mask), \ + "Incorrect value for mask"); \ + __MASKED_FIELD(mask, value); }) +#define _MASKED_BIT_ENABLE(a) ({ typeof(a) _a = (a); _MASKED_FIELD(_a, _a); }) +#define _MASKED_BIT_DISABLE(a) (_MASKED_FIELD((a), 0)) + +/* + * Given the first two numbers __a and __b of arbitrarily many evenly spaced + * numbers, pick the 0-based __index'th value. + * + * Always prefer this over _PICK() if the numbers are evenly spaced. + */ +#define _PICK_EVEN(__index, __a, __b) ((__a) + (__index) * ((__b) - (__a))) + +/* + * Given the arbitrary numbers in varargs, pick the 0-based __index'th number. + * + * Always prefer _PICK_EVEN() over this if the numbers are evenly spaced. + */ +#define _PICK(__index, ...) (((const u32 []){ __VA_ARGS__ })[__index]) + typedef struct { u32 reg; } i915_reg_t; @@ -120,6 +148,4 @@ typedef struct { #define i915_mmio_reg_equal(a, b) (i915_mmio_reg_offset(a) == i915_mmio_reg_offset(b)) #define i915_mmio_reg_valid(r) (!i915_mmio_reg_equal(r, INVALID_MMIO_REG)) -#define VLV_DISPLAY_BASE 0x180000 - #endif /* __I915_REG_DEFS__ */ diff --git a/drivers/gpu/drm/i915/i915_sysfs.c b/drivers/gpu/drm/i915/i915_sysfs.c index 1e2750210831..595e8b574990 100644 --- a/drivers/gpu/drm/i915/i915_sysfs.c +++ b/drivers/gpu/drm/i915/i915_sysfs.c @@ -218,7 +218,8 @@ static const struct bin_attribute error_state_attr = { static void i915_setup_error_capture(struct device *kdev) { if (sysfs_create_bin_file(&kdev->kobj, &error_state_attr)) - DRM_ERROR("error_state sysfs setup failed\n"); + drm_err(&kdev_minor_to_i915(kdev)->drm, + "error_state sysfs setup failed\n"); } static void i915_teardown_error_capture(struct device *kdev) diff --git a/drivers/gpu/drm/i915/i915_trace.h b/drivers/gpu/drm/i915/i915_trace.h index c70a02517e02..f6f9228a1351 100644 --- a/drivers/gpu/drm/i915/i915_trace.h +++ b/drivers/gpu/drm/i915/i915_trace.h @@ -15,7 +15,6 @@ #include "gt/intel_engine.h" #include "i915_drv.h" -#include "i915_irq.h" /* object tracking */ diff --git a/drivers/gpu/drm/i915/i915_vma.c b/drivers/gpu/drm/i915/i915_vma.c index c39488eb9eeb..703fee6b5f75 100644 --- a/drivers/gpu/drm/i915/i915_vma.c +++ b/drivers/gpu/drm/i915/i915_vma.c @@ -73,14 +73,16 @@ static void vma_print_allocator(struct i915_vma *vma, const char *reason) char buf[512]; if (!vma->node.stack) { - DRM_DEBUG_DRIVER("vma.node [%08llx + %08llx] %s: unknown owner\n", - vma->node.start, vma->node.size, reason); + drm_dbg(&to_i915(vma->obj->base.dev)->drm, + "vma.node [%08llx + %08llx] %s: unknown owner\n", + vma->node.start, vma->node.size, reason); return; } stack_depot_snprint(vma->node.stack, buf, sizeof(buf), 0); - DRM_DEBUG_DRIVER("vma.node [%08llx + %08llx] %s: inserted at %s\n", - vma->node.start, vma->node.size, reason, buf); + drm_dbg(&to_i915(vma->obj->base.dev)->drm, + "vma.node [%08llx + %08llx] %s: inserted at %s\n", + vma->node.start, vma->node.size, reason, buf); } #else @@ -782,9 +784,9 @@ i915_vma_insert(struct i915_vma *vma, struct i915_gem_ww_ctx *ww, * attempt to find space. */ if (size > end) { - DRM_DEBUG("Attempting to bind an object larger than the aperture: request=%llu > %s aperture=%llu\n", - size, flags & PIN_MAPPABLE ? "mappable" : "total", - end); + drm_dbg(&to_i915(vma->obj->base.dev)->drm, + "Attempting to bind an object larger than the aperture: request=%llu > %s aperture=%llu\n", + size, flags & PIN_MAPPABLE ? "mappable" : "total", end); return -ENOSPC; } @@ -1842,6 +1844,11 @@ int _i915_vma_move_to_active(struct i915_vma *vma, GEM_BUG_ON(!vma->pages); + if (!(flags & __EXEC_OBJECT_NO_REQUEST_AWAIT)) { + err = i915_request_await_object(rq, vma->obj, flags & EXEC_OBJECT_WRITE); + if (unlikely(err)) + return err; + } err = __i915_vma_move_to_active(vma, rq); if (unlikely(err)) return err; diff --git a/drivers/gpu/drm/i915/i915_vma.h b/drivers/gpu/drm/i915/i915_vma.h index aecd9c64486b..0757977a489b 100644 --- a/drivers/gpu/drm/i915/i915_vma.h +++ b/drivers/gpu/drm/i915/i915_vma.h @@ -55,6 +55,7 @@ static inline bool i915_vma_is_active(const struct i915_vma *vma) /* do not reserve memory to prevent deadlocks */ #define __EXEC_OBJECT_NO_RESERVE BIT(31) +#define __EXEC_OBJECT_NO_REQUEST_AWAIT BIT(30) int __must_check _i915_vma_move_to_active(struct i915_vma *vma, struct i915_request *rq, diff --git a/drivers/gpu/drm/i915/intel_device_info.c b/drivers/gpu/drm/i915/intel_device_info.c index 1dc1fb29a776..849baf6c3b3c 100644 --- a/drivers/gpu/drm/i915/intel_device_info.c +++ b/drivers/gpu/drm/i915/intel_device_info.c @@ -30,9 +30,10 @@ #include "display/intel_cdclk.h" #include "display/intel_de.h" #include "gt/intel_gt_regs.h" -#include "intel_device_info.h" #include "i915_drv.h" +#include "i915_reg.h" #include "i915_utils.h" +#include "intel_device_info.h" #define PLATFORM_NAME(x) [INTEL_##x] = #x static const char * const platform_names[] = { @@ -488,7 +489,7 @@ void intel_device_info_runtime_init(struct drm_i915_private *dev_priv) if (DISPLAY_VER(dev_priv) >= 11 && (dfsm & ICL_DFSM_DMC_DISABLE)) runtime->has_dmc = 0; - if (DISPLAY_VER(dev_priv) >= 10 && + if (IS_DISPLAY_VER(dev_priv, 10, 12) && (dfsm & GLK_DFSM_DISPLAY_DSC_DISABLE)) runtime->has_dsc = 0; } diff --git a/drivers/gpu/drm/i915/intel_dram.c b/drivers/gpu/drm/i915/intel_dram.c index 2403ccd52c74..bba8cb6e8ae4 100644 --- a/drivers/gpu/drm/i915/intel_dram.c +++ b/drivers/gpu/drm/i915/intel_dram.c @@ -471,8 +471,7 @@ static int xelpdp_get_dram_info(struct drm_i915_private *i915) u32 val = intel_uncore_read(&i915->uncore, MTL_MEM_SS_INFO_GLOBAL); struct dram_info *dram_info = &i915->dram_info; - val = REG_FIELD_GET(MTL_DDR_TYPE_MASK, val); - switch (val) { + switch (REG_FIELD_GET(MTL_DDR_TYPE_MASK, val)) { case 0: dram_info->type = INTEL_DRAM_DDR4; break; diff --git a/drivers/gpu/drm/i915/intel_gvt_mmio_table.c b/drivers/gpu/drm/i915/intel_gvt_mmio_table.c index 638b77d64bf4..ce6b3c3b636a 100644 --- a/drivers/gpu/drm/i915/intel_gvt_mmio_table.c +++ b/drivers/gpu/drm/i915/intel_gvt_mmio_table.c @@ -6,6 +6,7 @@ #include "display/intel_audio_regs.h" #include "display/intel_backlight_regs.h" #include "display/intel_dmc_regs.h" +#include "display/intel_dpio_phy.h" #include "display/vlv_dsi_pll_regs.h" #include "gt/intel_gt_regs.h" #include "gvt/gvt.h" diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c index ee34e2785636..73c88b1c9545 100644 --- a/drivers/gpu/drm/i915/intel_pm.c +++ b/drivers/gpu/drm/i915/intel_pm.c @@ -1426,7 +1426,7 @@ static int g4x_compute_intermediate_wm(struct intel_atomic_state *state, enum plane_id plane_id; if (!new_crtc_state->hw.active || - drm_atomic_crtc_needs_modeset(&new_crtc_state->uapi)) { + intel_crtc_needs_modeset(new_crtc_state)) { *intermediate = *optimal; intermediate->cxsr = false; @@ -1914,7 +1914,6 @@ static int vlv_compute_pipe_wm(struct intel_atomic_state *state, { struct intel_crtc_state *crtc_state = intel_atomic_get_new_crtc_state(state, crtc); - bool needs_modeset = drm_atomic_crtc_needs_modeset(&crtc_state->uapi); const struct intel_plane_state *old_plane_state; const struct intel_plane_state *new_plane_state; struct intel_plane *plane; @@ -1941,7 +1940,7 @@ static int vlv_compute_pipe_wm(struct intel_atomic_state *state, * FIFO setting we took over from the BIOS even if there * are no active planes on the crtc. */ - if (needs_modeset) + if (intel_crtc_needs_modeset(crtc_state)) dirty = ~0; if (!dirty) @@ -1961,7 +1960,7 @@ static int vlv_compute_pipe_wm(struct intel_atomic_state *state, if (ret) return ret; - if (needs_modeset || + if (intel_crtc_needs_modeset(crtc_state) || memcmp(old_fifo_state, new_fifo_state, sizeof(*new_fifo_state)) != 0) crtc_state->fifo_changed = true; @@ -2084,7 +2083,7 @@ static int vlv_compute_intermediate_wm(struct intel_atomic_state *state, int level; if (!new_crtc_state->hw.active || - drm_atomic_crtc_needs_modeset(&new_crtc_state->uapi)) { + intel_crtc_needs_modeset(new_crtc_state)) { *intermediate = *optimal; intermediate->cxsr = false; @@ -3142,7 +3141,7 @@ static int ilk_compute_intermediate_wm(struct intel_atomic_state *state, */ *a = new_crtc_state->wm.ilk.optimal; if (!new_crtc_state->hw.active || - drm_atomic_crtc_needs_modeset(&new_crtc_state->uapi) || + intel_crtc_needs_modeset(new_crtc_state) || state->skip_intermediate_wm) return 0; diff --git a/drivers/gpu/drm/i915/intel_step.c b/drivers/gpu/drm/i915/intel_step.c index 75d7a86c60c0..84a6fe736a3b 100644 --- a/drivers/gpu/drm/i915/intel_step.c +++ b/drivers/gpu/drm/i915/intel_step.c @@ -131,6 +131,10 @@ static const struct intel_step_info adls_rpls_revids[] = { [0xC] = { COMMON_GT_MEDIA_STEP(D0), .display_step = STEP_C0 }, }; +static const struct intel_step_info adlp_rplp_revids[] = { + [0x4] = { COMMON_GT_MEDIA_STEP(C0), .display_step = STEP_E0 }, +}; + static const struct intel_step_info adlp_n_revids[] = { [0x0] = { COMMON_GT_MEDIA_STEP(A0), .display_step = STEP_D0 }, }; @@ -187,6 +191,9 @@ void intel_step_init(struct drm_i915_private *i915) } else if (IS_ADLP_N(i915)) { revids = adlp_n_revids; size = ARRAY_SIZE(adlp_n_revids); + } else if (IS_ADLP_RPLP(i915)) { + revids = adlp_rplp_revids; + size = ARRAY_SIZE(adlp_rplp_revids); } else if (IS_ALDERLAKE_P(i915)) { revids = adlp_revids; size = ARRAY_SIZE(adlp_revids); diff --git a/drivers/gpu/drm/i915/intel_uncore.c b/drivers/gpu/drm/i915/intel_uncore.c index 2a3e2869fe71..8006a6c61466 100644 --- a/drivers/gpu/drm/i915/intel_uncore.c +++ b/drivers/gpu/drm/i915/intel_uncore.c @@ -29,6 +29,7 @@ #include "i915_drv.h" #include "i915_iosf_mbi.h" +#include "i915_reg.h" #include "i915_trace.h" #include "i915_vgpu.h" #include "intel_pm.h" @@ -178,8 +179,9 @@ static inline void fw_domain_wait_ack_clear(const struct intel_uncore_forcewake_domain *d) { if (wait_ack_clear(d, FORCEWAKE_KERNEL)) { - DRM_ERROR("%s: timed out waiting for forcewake ack to clear.\n", - intel_uncore_forcewake_domain_to_str(d->id)); + drm_err(&d->uncore->i915->drm, + "%s: timed out waiting for forcewake ack to clear.\n", + intel_uncore_forcewake_domain_to_str(d->id)); add_taint_for_CI(d->uncore->i915, TAINT_WARN); /* CI now unreliable */ } } @@ -226,11 +228,12 @@ fw_domain_wait_ack_with_fallback(const struct intel_uncore_forcewake_domain *d, fw_clear(d, FORCEWAKE_KERNEL_FALLBACK); } while (!ack_detected && pass++ < 10); - DRM_DEBUG_DRIVER("%s had to use fallback to %s ack, 0x%x (passes %u)\n", - intel_uncore_forcewake_domain_to_str(d->id), - type == ACK_SET ? "set" : "clear", - fw_ack(d), - pass); + drm_dbg(&d->uncore->i915->drm, + "%s had to use fallback to %s ack, 0x%x (passes %u)\n", + intel_uncore_forcewake_domain_to_str(d->id), + type == ACK_SET ? "set" : "clear", + fw_ack(d), + pass); return ack_detected ? 0 : -ETIMEDOUT; } @@ -255,8 +258,9 @@ static inline void fw_domain_wait_ack_set(const struct intel_uncore_forcewake_domain *d) { if (wait_ack_set(d, FORCEWAKE_KERNEL)) { - DRM_ERROR("%s: timed out waiting for forcewake ack request.\n", - intel_uncore_forcewake_domain_to_str(d->id)); + drm_err(&d->uncore->i915->drm, + "%s: timed out waiting for forcewake ack request.\n", + intel_uncore_forcewake_domain_to_str(d->id)); add_taint_for_CI(d->uncore->i915, TAINT_WARN); /* CI now unreliable */ } } diff --git a/drivers/gpu/drm/i915/pxp/intel_pxp_cmd_interface_42.h b/drivers/gpu/drm/i915/pxp/intel_pxp_cmd_interface_42.h new file mode 100644 index 000000000000..739f9072fa5f --- /dev/null +++ b/drivers/gpu/drm/i915/pxp/intel_pxp_cmd_interface_42.h @@ -0,0 +1,28 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright(c) 2020, Intel Corporation. All rights reserved. + */ + +#ifndef __INTEL_PXP_FW_INTERFACE_42_H__ +#define __INTEL_PXP_FW_INTERFACE_42_H__ + +#include <linux/types.h> +#include "intel_pxp_cmd_interface_cmn.h" + +/* PXP-Opcode for Init Session */ +#define PXP42_CMDID_INIT_SESSION 0x1e + +/* PXP-Input-Packet: Init Session (Arb-Session) */ +struct pxp42_create_arb_in { + struct pxp_cmd_header header; + u32 protection_mode; +#define PXP42_ARB_SESSION_MODE_HEAVY 0x2 + u32 session_id; +} __packed; + +/* PXP-Output-Packet: Init Session */ +struct pxp42_create_arb_out { + struct pxp_cmd_header header; +} __packed; + +#endif /* __INTEL_PXP_FW_INTERFACE_42_H__ */ diff --git a/drivers/gpu/drm/i915/pxp/intel_pxp_cmd_interface_43.h b/drivers/gpu/drm/i915/pxp/intel_pxp_cmd_interface_43.h new file mode 100644 index 000000000000..ad67e3f49c20 --- /dev/null +++ b/drivers/gpu/drm/i915/pxp/intel_pxp_cmd_interface_43.h @@ -0,0 +1,26 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright(c) 2022, Intel Corporation. All rights reserved. + */ + +#ifndef __INTEL_PXP_FW_INTERFACE_43_H__ +#define __INTEL_PXP_FW_INTERFACE_43_H__ + +#include <linux/types.h> +#include "intel_pxp_cmd_interface_cmn.h" + +/* PXP-Cmd-Op definitions */ +#define PXP43_CMDID_START_HUC_AUTH 0x0000003A + +/* PXP-Input-Packet: HUC-Authentication */ +struct pxp43_start_huc_auth_in { + struct pxp_cmd_header header; + __le64 huc_base_address; +} __packed; + +/* PXP-Output-Packet: HUC-Authentication */ +struct pxp43_start_huc_auth_out { + struct pxp_cmd_header header; +} __packed; + +#endif /* __INTEL_PXP_FW_INTERFACE_43_H__ */ diff --git a/drivers/gpu/drm/i915/pxp/intel_pxp_cmd_interface_cmn.h b/drivers/gpu/drm/i915/pxp/intel_pxp_cmd_interface_cmn.h new file mode 100644 index 000000000000..c2f23394f9b8 --- /dev/null +++ b/drivers/gpu/drm/i915/pxp/intel_pxp_cmd_interface_cmn.h @@ -0,0 +1,35 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright(c) 2022, Intel Corporation. All rights reserved. + */ + +#ifndef __INTEL_PXP_FW_INTERFACE_CMN_H__ +#define __INTEL_PXP_FW_INTERFACE_CMN_H__ + +#include <linux/types.h> + +#define PXP_APIVER(x, y) (((x) & 0xFFFF) << 16 | ((y) & 0xFFFF)) + +/* + * there are a lot of status codes for PXP, but we only define the cross-API + * common ones that we actually can handle in the kernel driver. Other failure + * codes should be printed to error msg for debug. + */ +enum pxp_status { + PXP_STATUS_SUCCESS = 0x0, + PXP_STATUS_OP_NOT_PERMITTED = 0x4013 +}; + +/* Common PXP FW message header */ +struct pxp_cmd_header { + u32 api_version; + u32 command_id; + union { + u32 status; /* out */ + u32 stream_id; /* in */ + }; + /* Length of the message (excluding the header) */ + u32 buffer_len; +} __packed; + +#endif /* __INTEL_PXP_FW_INTERFACE_CMN_H__ */ diff --git a/drivers/gpu/drm/i915/pxp/intel_pxp_huc.c b/drivers/gpu/drm/i915/pxp/intel_pxp_huc.c index 7ec36d94e758..2e1165522950 100644 --- a/drivers/gpu/drm/i915/pxp/intel_pxp_huc.c +++ b/drivers/gpu/drm/i915/pxp/intel_pxp_huc.c @@ -3,7 +3,8 @@ * Copyright(c) 2021-2022, Intel Corporation. All rights reserved. */ -#include "drm/i915_drm.h" +#include <drm/i915_drm.h> + #include "i915_drv.h" #include "gem/i915_gem_region.h" @@ -13,14 +14,14 @@ #include "intel_pxp_huc.h" #include "intel_pxp_tee.h" #include "intel_pxp_types.h" -#include "intel_pxp_tee_interface.h" +#include "intel_pxp_cmd_interface_43.h" int intel_pxp_huc_load_and_auth(struct intel_pxp *pxp) { struct intel_gt *gt = pxp_to_gt(pxp); struct intel_huc *huc = >->uc.huc; - struct pxp_tee_start_huc_auth_in huc_in = {0}; - struct pxp_tee_start_huc_auth_out huc_out = {0}; + struct pxp43_start_huc_auth_in huc_in = {0}; + struct pxp43_start_huc_auth_out huc_out = {0}; dma_addr_t huc_phys_addr; u8 client_id = 0; u8 fence_id = 0; @@ -32,8 +33,8 @@ int intel_pxp_huc_load_and_auth(struct intel_pxp *pxp) huc_phys_addr = i915_gem_object_get_dma_address(huc->fw.obj, 0); /* write the PXP message into the lmem (the sg list) */ - huc_in.header.api_version = PXP_TEE_43_APIVER; - huc_in.header.command_id = PXP_TEE_43_START_HUC_AUTH; + huc_in.header.api_version = PXP_APIVER(4, 3); + huc_in.header.command_id = PXP43_CMDID_START_HUC_AUTH; huc_in.header.status = 0; huc_in.header.buffer_len = sizeof(huc_in.huc_base_address); huc_in.huc_base_address = huc_phys_addr; diff --git a/drivers/gpu/drm/i915/pxp/intel_pxp_tee.c b/drivers/gpu/drm/i915/pxp/intel_pxp_tee.c index 052fd2f9a583..b0c9170b1395 100644 --- a/drivers/gpu/drm/i915/pxp/intel_pxp_tee.c +++ b/drivers/gpu/drm/i915/pxp/intel_pxp_tee.c @@ -14,7 +14,7 @@ #include "intel_pxp.h" #include "intel_pxp_session.h" #include "intel_pxp_tee.h" -#include "intel_pxp_tee_interface.h" +#include "intel_pxp_cmd_interface_42.h" #include "intel_pxp_huc.h" static inline struct intel_pxp *i915_dev_to_pxp(struct device *i915_kdev) @@ -286,14 +286,14 @@ int intel_pxp_tee_cmd_create_arb_session(struct intel_pxp *pxp, int arb_session_id) { struct drm_i915_private *i915 = pxp_to_gt(pxp)->i915; - struct pxp_tee_create_arb_in msg_in = {0}; - struct pxp_tee_create_arb_out msg_out = {0}; + struct pxp42_create_arb_in msg_in = {0}; + struct pxp42_create_arb_out msg_out = {0}; int ret; - msg_in.header.api_version = PXP_TEE_APIVER; - msg_in.header.command_id = PXP_TEE_ARB_CMDID; + msg_in.header.api_version = PXP_APIVER(4, 2); + msg_in.header.command_id = PXP42_CMDID_INIT_SESSION; msg_in.header.buffer_len = sizeof(msg_in) - sizeof(msg_in.header); - msg_in.protection_mode = PXP_TEE_ARB_PROTECTION_MODE; + msg_in.protection_mode = PXP42_ARB_SESSION_MODE_HEAVY; msg_in.session_id = arb_session_id; ret = intel_pxp_tee_io_message(pxp, diff --git a/drivers/gpu/drm/i915/pxp/intel_pxp_tee_interface.h b/drivers/gpu/drm/i915/pxp/intel_pxp_tee_interface.h deleted file mode 100644 index 7edc1760f142..000000000000 --- a/drivers/gpu/drm/i915/pxp/intel_pxp_tee_interface.h +++ /dev/null @@ -1,57 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -/* - * Copyright(c) 2020-2022, Intel Corporation. All rights reserved. - */ - -#ifndef __INTEL_PXP_TEE_INTERFACE_H__ -#define __INTEL_PXP_TEE_INTERFACE_H__ - -#include <linux/types.h> - -#define PXP_TEE_APIVER 0x40002 -#define PXP_TEE_43_APIVER 0x00040003 -#define PXP_TEE_ARB_CMDID 0x1e -#define PXP_TEE_ARB_PROTECTION_MODE 0x2 -#define PXP_TEE_43_START_HUC_AUTH 0x0000003A - -/* - * there are a lot of status codes for PXP, but we only define the ones we - * actually can handle in the driver. other failure codes will be printed to - * error msg for debug. - */ -enum pxp_status { - PXP_STATUS_SUCCESS = 0x0, - PXP_STATUS_OP_NOT_PERMITTED = 0x4013 -}; - -/* PXP TEE message header */ -struct pxp_tee_cmd_header { - u32 api_version; - u32 command_id; - u32 status; - /* Length of the message (excluding the header) */ - u32 buffer_len; -} __packed; - -/* PXP TEE message input to create a arbitrary session */ -struct pxp_tee_create_arb_in { - struct pxp_tee_cmd_header header; - u32 protection_mode; - u32 session_id; -} __packed; - -/* PXP TEE message output to create a arbitrary session */ -struct pxp_tee_create_arb_out { - struct pxp_tee_cmd_header header; -} __packed; - -struct pxp_tee_start_huc_auth_in { - struct pxp_tee_cmd_header header; - __le64 huc_base_address; -}; - -struct pxp_tee_start_huc_auth_out { - struct pxp_tee_cmd_header header; -}; - -#endif /* __INTEL_PXP_TEE_INTERFACE_H__ */ diff --git a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c index 27c733b00976..eae7d947d7de 100644 --- a/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c +++ b/drivers/gpu/drm/i915/selftests/i915_gem_gtt.c @@ -61,7 +61,6 @@ static int fake_get_pages(struct drm_i915_gem_object *obj) #define PFN_BIAS 0x1000 struct sg_table *pages; struct scatterlist *sg; - unsigned int sg_page_sizes; typeof(obj->base.size) rem; pages = kmalloc(sizeof(*pages), GFP); @@ -74,7 +73,6 @@ static int fake_get_pages(struct drm_i915_gem_object *obj) return -ENOMEM; } - sg_page_sizes = 0; rem = obj->base.size; for (sg = pages->sgl; sg; sg = sg_next(sg)) { unsigned long len = min_t(typeof(rem), rem, BIT(31)); @@ -83,13 +81,12 @@ static int fake_get_pages(struct drm_i915_gem_object *obj) sg_set_page(sg, pfn_to_page(PFN_BIAS), len, 0); sg_dma_address(sg) = page_to_phys(sg_page(sg)); sg_dma_len(sg) = len; - sg_page_sizes |= len; rem -= len; } GEM_BUG_ON(rem); - __i915_gem_object_set_pages(obj, pages, sg_page_sizes); + __i915_gem_object_set_pages(obj, pages); return 0; #undef GFP diff --git a/drivers/gpu/drm/i915/selftests/i915_request.c b/drivers/gpu/drm/i915/selftests/i915_request.c index a46350c37e9d..0daa8669181d 100644 --- a/drivers/gpu/drm/i915/selftests/i915_request.c +++ b/drivers/gpu/drm/i915/selftests/i915_request.c @@ -1223,9 +1223,7 @@ static int live_all_engines(void *arg) goto out_request; } - err = i915_request_await_object(request[idx], batch->obj, 0); - if (err == 0) - err = i915_vma_move_to_active(batch, request[idx], 0); + err = i915_vma_move_to_active(batch, request[idx], 0); GEM_BUG_ON(err); err = engine->emit_bb_start(request[idx], @@ -1352,10 +1350,7 @@ static int live_sequential_engines(void *arg) } } - err = i915_request_await_object(request[idx], - batch->obj, false); - if (err == 0) - err = i915_vma_move_to_active(batch, request[idx], 0); + err = i915_vma_move_to_active(batch, request[idx], 0); GEM_BUG_ON(err); err = engine->emit_bb_start(request[idx], @@ -1710,7 +1705,8 @@ static int live_breadcrumbs_smoketest(void *arg) { struct drm_i915_private *i915 = arg; const unsigned int nengines = num_uabi_engines(i915); - const unsigned int ncpus = num_online_cpus(); + const unsigned int ncpus = /* saturate with nengines * ncpus */ + max_t(int, 2, DIV_ROUND_UP(num_online_cpus(), nengines)); unsigned long num_waits, num_fences; struct intel_engine_cs *engine; struct smoke_thread *threads; @@ -1782,7 +1778,7 @@ static int live_breadcrumbs_smoketest(void *arg) goto out_flush; } /* One ring interleaved between requests from all cpus */ - smoke[idx].max_batch /= num_online_cpus() + 1; + smoke[idx].max_batch /= ncpus + 1; pr_debug("Limiting batches to %d requests on %s\n", smoke[idx].max_batch, engine->name); diff --git a/drivers/gpu/drm/i915/selftests/igt_spinner.c b/drivers/gpu/drm/i915/selftests/igt_spinner.c index 0c22594ae274..16978ac59797 100644 --- a/drivers/gpu/drm/i915/selftests/igt_spinner.c +++ b/drivers/gpu/drm/i915/selftests/igt_spinner.c @@ -119,22 +119,6 @@ static u64 hws_address(const struct i915_vma *hws, return hws->node.start + seqno_offset(rq->fence.context); } -static int move_to_active(struct i915_vma *vma, - struct i915_request *rq, - unsigned int flags) -{ - int err; - - i915_vma_lock(vma); - err = i915_request_await_object(rq, vma->obj, - flags & EXEC_OBJECT_WRITE); - if (err == 0) - err = i915_vma_move_to_active(vma, rq, flags); - i915_vma_unlock(vma); - - return err; -} - struct i915_request * igt_spinner_create_request(struct igt_spinner *spin, struct intel_context *ce, @@ -165,11 +149,11 @@ igt_spinner_create_request(struct igt_spinner *spin, if (IS_ERR(rq)) return ERR_CAST(rq); - err = move_to_active(vma, rq, 0); + err = igt_vma_move_to_active_unlocked(vma, rq, 0); if (err) goto cancel_rq; - err = move_to_active(hws, rq, 0); + err = igt_vma_move_to_active_unlocked(hws, rq, 0); if (err) goto cancel_rq; diff --git a/drivers/gpu/drm/i915/selftests/mock_region.c b/drivers/gpu/drm/i915/selftests/mock_region.c index bac21fe84ca5..6324eb32d4dd 100644 --- a/drivers/gpu/drm/i915/selftests/mock_region.c +++ b/drivers/gpu/drm/i915/selftests/mock_region.c @@ -41,7 +41,7 @@ static int mock_region_get_pages(struct drm_i915_gem_object *obj) } pages = &obj->mm.rsgt->table; - __i915_gem_object_set_pages(obj, pages, i915_sg_dma_sizes(pages->sgl)); + __i915_gem_object_set_pages(obj, pages); return 0; diff --git a/drivers/gpu/drm/i915/vlv_sideband.c b/drivers/gpu/drm/i915/vlv_sideband.c index c26001300ebd..6eea6e1a99c0 100644 --- a/drivers/gpu/drm/i915/vlv_sideband.c +++ b/drivers/gpu/drm/i915/vlv_sideband.c @@ -8,6 +8,8 @@ #include "i915_reg.h" #include "vlv_sideband.h" +#include "display/intel_dpio_phy.h" + /* * IOSF sideband, see VLV2_SidebandMsg_HAS.docx and * VLV_VLV2_PUNIT_HAS_0.8.docx diff --git a/drivers/gpu/drm/imx/Kconfig b/drivers/gpu/drm/imx/Kconfig index 975de4ff7313..fd5b2471fdf0 100644 --- a/drivers/gpu/drm/imx/Kconfig +++ b/drivers/gpu/drm/imx/Kconfig @@ -4,7 +4,6 @@ config DRM_IMX select DRM_KMS_HELPER select VIDEOMODE_HELPERS select DRM_GEM_DMA_HELPER - select DRM_KMS_HELPER depends on DRM && (ARCH_MXC || ARCH_MULTIPLATFORM || COMPILE_TEST) depends on IMX_IPUV3_CORE help diff --git a/drivers/gpu/drm/imx/imx-tve.c b/drivers/gpu/drm/imx/imx-tve.c index d64ebd2cf15e..d6832f506322 100644 --- a/drivers/gpu/drm/imx/imx-tve.c +++ b/drivers/gpu/drm/imx/imx-tve.c @@ -217,8 +217,9 @@ static int imx_tve_connector_get_modes(struct drm_connector *connector) return ret; } -static int imx_tve_connector_mode_valid(struct drm_connector *connector, - struct drm_display_mode *mode) +static enum drm_mode_status +imx_tve_connector_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) { struct imx_tve *tve = con_to_tve(connector); unsigned long rate; diff --git a/drivers/gpu/drm/lima/lima_devfreq.c b/drivers/gpu/drm/lima/lima_devfreq.c index 011be7ff51e1..bc8fb4e38d0a 100644 --- a/drivers/gpu/drm/lima/lima_devfreq.c +++ b/drivers/gpu/drm/lima/lima_devfreq.c @@ -112,11 +112,6 @@ int lima_devfreq_init(struct lima_device *ldev) unsigned long cur_freq; int ret; const char *regulator_names[] = { "mali", NULL }; - const char *clk_names[] = { "core", NULL }; - struct dev_pm_opp_config config = { - .regulator_names = regulator_names, - .clk_names = clk_names, - }; if (!device_property_present(dev, "operating-points-v2")) /* Optional, continue without devfreq */ @@ -124,7 +119,15 @@ int lima_devfreq_init(struct lima_device *ldev) spin_lock_init(&ldevfreq->lock); - ret = devm_pm_opp_set_config(dev, &config); + /* + * clkname is set separately so it is not affected by the optional + * regulator setting which may return error. + */ + ret = devm_pm_opp_set_clkname(dev, "core"); + if (ret) + return ret; + + ret = devm_pm_opp_set_regulators(dev, regulator_names); if (ret) { /* Continue if the optional regulator is missing */ if (ret != -ENODEV) diff --git a/drivers/gpu/drm/msm/Kconfig b/drivers/gpu/drm/msm/Kconfig index 4e0cbd682725..3c9dfdb0b328 100644 --- a/drivers/gpu/drm/msm/Kconfig +++ b/drivers/gpu/drm/msm/Kconfig @@ -155,7 +155,7 @@ config DRM_MSM_HDMI Compile in support for the HDMI output MSM DRM driver. It can be a primary or a secondary display on device. Note that this is used only for the direct HDMI output. If the device outputs HDMI data - throught some kind of DSI-to-HDMI bridge, this option can be disabled. + through some kind of DSI-to-HDMI bridge, this option can be disabled. config DRM_MSM_HDMI_HDCP bool "Enable HDMI HDCP support in MSM DRM driver" diff --git a/drivers/gpu/drm/msm/adreno/a6xx_gpu_state.c b/drivers/gpu/drm/msm/adreno/a6xx_gpu_state.c index 55f443328d8e..a5c3d1ed255a 100644 --- a/drivers/gpu/drm/msm/adreno/a6xx_gpu_state.c +++ b/drivers/gpu/drm/msm/adreno/a6xx_gpu_state.c @@ -91,7 +91,7 @@ struct a6xx_state_memobj { static void *state_kcalloc(struct a6xx_gpu_state *a6xx_state, int nr, size_t objsize) { struct a6xx_state_memobj *obj = - kzalloc((nr * objsize) + sizeof(*obj), GFP_KERNEL); + kvzalloc((nr * objsize) + sizeof(*obj), GFP_KERNEL); if (!obj) return NULL; @@ -813,6 +813,9 @@ static struct msm_gpu_state_bo *a6xx_snapshot_gmu_bo( { struct msm_gpu_state_bo *snapshot; + if (!bo->size) + return NULL; + snapshot = state_kcalloc(a6xx_state, 1, sizeof(*snapshot)); if (!snapshot) return NULL; @@ -1040,8 +1043,13 @@ static void a6xx_gpu_state_destroy(struct kref *kref) if (a6xx_state->gmu_hfi) kvfree(a6xx_state->gmu_hfi->data); - list_for_each_entry_safe(obj, tmp, &a6xx_state->objs, node) - kfree(obj); + if (a6xx_state->gmu_debug) + kvfree(a6xx_state->gmu_debug->data); + + list_for_each_entry_safe(obj, tmp, &a6xx_state->objs, node) { + list_del(&obj->node); + kvfree(obj); + } adreno_gpu_state_destroy(state); kfree(a6xx_state); diff --git a/drivers/gpu/drm/msm/adreno/adreno_device.c b/drivers/gpu/drm/msm/adreno/adreno_device.c index 24b489b6129a..628806423f7d 100644 --- a/drivers/gpu/drm/msm/adreno/adreno_device.c +++ b/drivers/gpu/drm/msm/adreno/adreno_device.c @@ -679,6 +679,9 @@ static int adreno_system_suspend(struct device *dev) struct msm_gpu *gpu = dev_to_gpu(dev); int remaining, ret; + if (!gpu) + return 0; + suspend_scheduler(gpu); remaining = wait_event_timeout(gpu->retire_event, @@ -700,7 +703,12 @@ out: static int adreno_system_resume(struct device *dev) { - resume_scheduler(dev_to_gpu(dev)); + struct msm_gpu *gpu = dev_to_gpu(dev); + + if (!gpu) + return 0; + + resume_scheduler(gpu); return pm_runtime_force_resume(dev); } diff --git a/drivers/gpu/drm/msm/adreno/adreno_gpu.c b/drivers/gpu/drm/msm/adreno/adreno_gpu.c index 382fb7f9e497..5a0e8491cd3a 100644 --- a/drivers/gpu/drm/msm/adreno/adreno_gpu.c +++ b/drivers/gpu/drm/msm/adreno/adreno_gpu.c @@ -729,7 +729,12 @@ static char *adreno_gpu_ascii85_encode(u32 *src, size_t len) return buf; } -/* len is expected to be in bytes */ +/* len is expected to be in bytes + * + * WARNING: *ptr should be allocated with kvmalloc or friends. It can be free'd + * with kvfree() and replaced with a newly kvmalloc'd buffer on the first call + * when the unencoded raw data is encoded + */ void adreno_show_object(struct drm_printer *p, void **ptr, int len, bool *encoded) { diff --git a/drivers/gpu/drm/msm/disp/mdp4/mdp4_lvds_connector.c b/drivers/gpu/drm/msm/disp/mdp4/mdp4_lvds_connector.c index 7288041dd86a..7444b75c4215 100644 --- a/drivers/gpu/drm/msm/disp/mdp4/mdp4_lvds_connector.c +++ b/drivers/gpu/drm/msm/disp/mdp4/mdp4_lvds_connector.c @@ -56,8 +56,9 @@ static int mdp4_lvds_connector_get_modes(struct drm_connector *connector) return ret; } -static int mdp4_lvds_connector_mode_valid(struct drm_connector *connector, - struct drm_display_mode *mode) +static enum drm_mode_status +mdp4_lvds_connector_mode_valid(struct drm_connector *connector, + struct drm_display_mode *mode) { struct mdp4_lvds_connector *mdp4_lvds_connector = to_mdp4_lvds_connector(connector); diff --git a/drivers/gpu/drm/msm/dp/dp_ctrl.c b/drivers/gpu/drm/msm/dp/dp_ctrl.c index 3854c9f1f7e9..dd26ca651a05 100644 --- a/drivers/gpu/drm/msm/dp/dp_ctrl.c +++ b/drivers/gpu/drm/msm/dp/dp_ctrl.c @@ -1243,8 +1243,7 @@ static int dp_ctrl_link_train(struct dp_ctrl_private *ctrl, { int ret = 0; const u8 *dpcd = ctrl->panel->dpcd; - u8 encoding = DP_SET_ANSI_8B10B; - u8 ssc; + u8 encoding[] = { 0, DP_SET_ANSI_8B10B }; u8 assr; struct dp_link_info link_info = {0}; @@ -1256,13 +1255,11 @@ static int dp_ctrl_link_train(struct dp_ctrl_private *ctrl, dp_aux_link_configure(ctrl->aux, &link_info); - if (drm_dp_max_downspread(dpcd)) { - ssc = DP_SPREAD_AMP_0_5; - drm_dp_dpcd_write(ctrl->aux, DP_DOWNSPREAD_CTRL, &ssc, 1); - } + if (drm_dp_max_downspread(dpcd)) + encoding[0] |= DP_SPREAD_AMP_0_5; - drm_dp_dpcd_write(ctrl->aux, DP_MAIN_LINK_CHANNEL_CODING_SET, - &encoding, 1); + /* config DOWNSPREAD_CTRL and MAIN_LINK_CHANNEL_CODING_SET */ + drm_dp_dpcd_write(ctrl->aux, DP_DOWNSPREAD_CTRL, encoding, 2); if (drm_dp_alternate_scrambler_reset_cap(dpcd)) { assr = DP_ALTERNATE_SCRAMBLER_RESET_ENABLE; diff --git a/drivers/gpu/drm/msm/dp/dp_display.c b/drivers/gpu/drm/msm/dp/dp_display.c index 39c9e55f07cc..c9d9b384ddd0 100644 --- a/drivers/gpu/drm/msm/dp/dp_display.c +++ b/drivers/gpu/drm/msm/dp/dp_display.c @@ -1249,7 +1249,7 @@ int dp_display_request_irq(struct msm_dp *dp_display) return -EINVAL; } - rc = devm_request_irq(&dp->pdev->dev, dp->irq, + rc = devm_request_irq(dp_display->drm_dev->dev, dp->irq, dp_display_irq_handler, IRQF_TRIGGER_HIGH, "dp_display_isr", dp); if (rc < 0) { @@ -1528,6 +1528,11 @@ void msm_dp_debugfs_init(struct msm_dp *dp_display, struct drm_minor *minor) } } +static void of_dp_aux_depopulate_bus_void(void *data) +{ + of_dp_aux_depopulate_bus(data); +} + static int dp_display_get_next_bridge(struct msm_dp *dp) { int rc; @@ -1552,10 +1557,16 @@ static int dp_display_get_next_bridge(struct msm_dp *dp) * panel driver is probed asynchronously but is the best we * can do without a bigger driver reorganization. */ - rc = devm_of_dp_aux_populate_ep_devices(dp_priv->aux); + rc = of_dp_aux_populate_bus(dp_priv->aux, NULL); of_node_put(aux_bus); if (rc) goto error; + + rc = devm_add_action_or_reset(dp->drm_dev->dev, + of_dp_aux_depopulate_bus_void, + dp_priv->aux); + if (rc) + goto error; } else if (dp->is_edp) { DRM_ERROR("eDP aux_bus not found\n"); return -ENODEV; @@ -1568,7 +1579,7 @@ static int dp_display_get_next_bridge(struct msm_dp *dp) * For DisplayPort interfaces external bridges are optional, so * silently ignore an error if one is not present (-ENODEV). */ - rc = dp_parser_find_next_bridge(dp_priv->parser); + rc = devm_dp_parser_find_next_bridge(dp->drm_dev->dev, dp_priv->parser); if (!dp->is_edp && rc == -ENODEV) return 0; @@ -1597,6 +1608,12 @@ int msm_dp_modeset_init(struct msm_dp *dp_display, struct drm_device *dev, return -EINVAL; priv = dev->dev_private; + + if (priv->num_bridges == ARRAY_SIZE(priv->bridges)) { + DRM_DEV_ERROR(dev->dev, "too many bridges\n"); + return -ENOSPC; + } + dp_display->drm_dev = dev; dp_priv = container_of(dp_display, struct dp_display_private, dp_display); diff --git a/drivers/gpu/drm/msm/dp/dp_drm.c b/drivers/gpu/drm/msm/dp/dp_drm.c index 6df25f7662e7..6db82f9b03af 100644 --- a/drivers/gpu/drm/msm/dp/dp_drm.c +++ b/drivers/gpu/drm/msm/dp/dp_drm.c @@ -31,6 +31,36 @@ static enum drm_connector_status dp_bridge_detect(struct drm_bridge *bridge) connector_status_disconnected; } +static int dp_bridge_atomic_check(struct drm_bridge *bridge, + struct drm_bridge_state *bridge_state, + struct drm_crtc_state *crtc_state, + struct drm_connector_state *conn_state) +{ + struct msm_dp *dp; + + dp = to_dp_bridge(bridge)->dp_display; + + drm_dbg_dp(dp->drm_dev, "is_connected = %s\n", + (dp->is_connected) ? "true" : "false"); + + /* + * There is no protection in the DRM framework to check if the display + * pipeline has been already disabled before trying to disable it again. + * Hence if the sink is unplugged, the pipeline gets disabled, but the + * crtc->active is still true. Any attempt to set the mode or manually + * disable this encoder will result in the crash. + * + * TODO: add support for telling the DRM subsystem that the pipeline is + * disabled by the hardware and thus all access to it should be forbidden. + * After that this piece of code can be removed. + */ + if (bridge->ops & DRM_BRIDGE_OP_HPD) + return (dp->is_connected) ? 0 : -ENOTCONN; + + return 0; +} + + /** * dp_bridge_get_modes - callback to add drm modes via drm_mode_probed_add() * @bridge: Poiner to drm bridge @@ -61,6 +91,9 @@ static int dp_bridge_get_modes(struct drm_bridge *bridge, struct drm_connector * } static const struct drm_bridge_funcs dp_bridge_ops = { + .atomic_duplicate_state = drm_atomic_helper_bridge_duplicate_state, + .atomic_destroy_state = drm_atomic_helper_bridge_destroy_state, + .atomic_reset = drm_atomic_helper_bridge_reset, .enable = dp_bridge_enable, .disable = dp_bridge_disable, .post_disable = dp_bridge_post_disable, @@ -68,6 +101,7 @@ static const struct drm_bridge_funcs dp_bridge_ops = { .mode_valid = dp_bridge_mode_valid, .get_modes = dp_bridge_get_modes, .detect = dp_bridge_detect, + .atomic_check = dp_bridge_atomic_check, }; struct drm_bridge *dp_bridge_init(struct msm_dp *dp_display, struct drm_device *dev, diff --git a/drivers/gpu/drm/msm/dp/dp_parser.c b/drivers/gpu/drm/msm/dp/dp_parser.c index dd732215d55b..dcbe893d66d7 100644 --- a/drivers/gpu/drm/msm/dp/dp_parser.c +++ b/drivers/gpu/drm/msm/dp/dp_parser.c @@ -240,12 +240,12 @@ static int dp_parser_clock(struct dp_parser *parser) return 0; } -int dp_parser_find_next_bridge(struct dp_parser *parser) +int devm_dp_parser_find_next_bridge(struct device *dev, struct dp_parser *parser) { - struct device *dev = &parser->pdev->dev; + struct platform_device *pdev = parser->pdev; struct drm_bridge *bridge; - bridge = devm_drm_of_get_bridge(dev, dev->of_node, 1, 0); + bridge = devm_drm_of_get_bridge(dev, pdev->dev.of_node, 1, 0); if (IS_ERR(bridge)) return PTR_ERR(bridge); diff --git a/drivers/gpu/drm/msm/dp/dp_parser.h b/drivers/gpu/drm/msm/dp/dp_parser.h index 866c1a82bf1a..d30ab773db46 100644 --- a/drivers/gpu/drm/msm/dp/dp_parser.h +++ b/drivers/gpu/drm/msm/dp/dp_parser.h @@ -138,8 +138,9 @@ struct dp_parser { struct dp_parser *dp_parser_get(struct platform_device *pdev); /** - * dp_parser_find_next_bridge() - find an additional bridge to DP + * devm_dp_parser_find_next_bridge() - find an additional bridge to DP * + * @dev: device to tie bridge lifetime to * @parser: dp_parser data from client * * This function is used to find any additional bridge attached to @@ -147,6 +148,6 @@ struct dp_parser *dp_parser_get(struct platform_device *pdev); * * Return: 0 if able to get the bridge, otherwise negative errno for failure. */ -int dp_parser_find_next_bridge(struct dp_parser *parser); +int devm_dp_parser_find_next_bridge(struct device *dev, struct dp_parser *parser); #endif diff --git a/drivers/gpu/drm/msm/dsi/dsi.c b/drivers/gpu/drm/msm/dsi/dsi.c index 39bbabb5daf6..8a95c744972a 100644 --- a/drivers/gpu/drm/msm/dsi/dsi.c +++ b/drivers/gpu/drm/msm/dsi/dsi.c @@ -218,6 +218,12 @@ int msm_dsi_modeset_init(struct msm_dsi *msm_dsi, struct drm_device *dev, return -EINVAL; priv = dev->dev_private; + + if (priv->num_bridges == ARRAY_SIZE(priv->bridges)) { + DRM_DEV_ERROR(dev->dev, "too many bridges\n"); + return -ENOSPC; + } + msm_dsi->dev = dev; ret = msm_dsi_host_modeset_init(msm_dsi->host, dev); diff --git a/drivers/gpu/drm/msm/hdmi/hdmi.c b/drivers/gpu/drm/msm/hdmi/hdmi.c index 93fe61b86967..f28fb21e3891 100644 --- a/drivers/gpu/drm/msm/hdmi/hdmi.c +++ b/drivers/gpu/drm/msm/hdmi/hdmi.c @@ -300,6 +300,11 @@ int msm_hdmi_modeset_init(struct hdmi *hdmi, struct platform_device *pdev = hdmi->pdev; int ret; + if (priv->num_bridges == ARRAY_SIZE(priv->bridges)) { + DRM_DEV_ERROR(dev->dev, "too many bridges\n"); + return -ENOSPC; + } + hdmi->dev = dev; hdmi->encoder = encoder; @@ -339,7 +344,7 @@ int msm_hdmi_modeset_init(struct hdmi *hdmi, goto fail; } - ret = devm_request_irq(&pdev->dev, hdmi->irq, + ret = devm_request_irq(dev->dev, hdmi->irq, msm_hdmi_irq, IRQF_TRIGGER_HIGH, "hdmi_isr", hdmi); if (ret < 0) { diff --git a/drivers/gpu/drm/msm/msm_drv.c b/drivers/gpu/drm/msm/msm_drv.c index 28034c21f6bc..105b5b48e828 100644 --- a/drivers/gpu/drm/msm/msm_drv.c +++ b/drivers/gpu/drm/msm/msm_drv.c @@ -247,6 +247,7 @@ static int msm_drm_uninit(struct device *dev) for (i = 0; i < priv->num_bridges; i++) drm_bridge_remove(priv->bridges[i]); + priv->num_bridges = 0; pm_runtime_get_sync(dev); msm_irq_uninstall(ddev); diff --git a/drivers/gpu/drm/msm/msm_gem_submit.c b/drivers/gpu/drm/msm/msm_gem_submit.c index 5599d93ec0d2..45a3e5cadc7d 100644 --- a/drivers/gpu/drm/msm/msm_gem_submit.c +++ b/drivers/gpu/drm/msm/msm_gem_submit.c @@ -501,11 +501,11 @@ out: */ static void submit_cleanup(struct msm_gem_submit *submit, bool error) { - unsigned cleanup_flags = BO_LOCKED | BO_OBJ_PINNED; + unsigned cleanup_flags = BO_LOCKED; unsigned i; if (error) - cleanup_flags |= BO_VMA_PINNED; + cleanup_flags |= BO_VMA_PINNED | BO_OBJ_PINNED; for (i = 0; i < submit->nr_bos; i++) { struct msm_gem_object *msm_obj = submit->bos[i].obj; @@ -706,7 +706,7 @@ int msm_ioctl_gem_submit(struct drm_device *dev, void *data, struct msm_drm_private *priv = dev->dev_private; struct drm_msm_gem_submit *args = data; struct msm_file_private *ctx = file->driver_priv; - struct msm_gem_submit *submit = NULL; + struct msm_gem_submit *submit; struct msm_gpu *gpu = priv->gpu; struct msm_gpu_submitqueue *queue; struct msm_ringbuffer *ring; @@ -946,8 +946,7 @@ out_unlock: put_unused_fd(out_fence_fd); mutex_unlock(&queue->lock); out_post_unlock: - if (submit) - msm_gem_submit_put(submit); + msm_gem_submit_put(submit); if (!IS_ERR_OR_NULL(post_deps)) { for (i = 0; i < args->nr_out_syncobjs; ++i) { kfree(post_deps[i].chain); diff --git a/drivers/gpu/drm/msm/msm_gpu.c b/drivers/gpu/drm/msm/msm_gpu.c index 0098ee8438aa..021f4e29b613 100644 --- a/drivers/gpu/drm/msm/msm_gpu.c +++ b/drivers/gpu/drm/msm/msm_gpu.c @@ -997,4 +997,6 @@ void msm_gpu_cleanup(struct msm_gpu *gpu) } msm_devfreq_cleanup(gpu); + + platform_set_drvdata(gpu->pdev, NULL); } diff --git a/drivers/gpu/drm/msm/msm_gpu.h b/drivers/gpu/drm/msm/msm_gpu.h index ff911e7305ce..58a72e6b1400 100644 --- a/drivers/gpu/drm/msm/msm_gpu.h +++ b/drivers/gpu/drm/msm/msm_gpu.h @@ -280,6 +280,10 @@ struct msm_gpu { static inline struct msm_gpu *dev_to_gpu(struct device *dev) { struct adreno_smmu_priv *adreno_smmu = dev_get_drvdata(dev); + + if (!adreno_smmu) + return NULL; + return container_of(adreno_smmu, struct msm_gpu, adreno_smmu); } diff --git a/drivers/gpu/drm/msm/msm_ringbuffer.c b/drivers/gpu/drm/msm/msm_ringbuffer.c index cad4c3525f0b..57a8e9564540 100644 --- a/drivers/gpu/drm/msm/msm_ringbuffer.c +++ b/drivers/gpu/drm/msm/msm_ringbuffer.c @@ -25,7 +25,8 @@ static struct dma_fence *msm_job_run(struct drm_sched_job *job) msm_gem_lock(obj); msm_gem_unpin_vma_fenced(submit->bos[i].vma, fctx); - submit->bos[i].flags &= ~BO_VMA_PINNED; + msm_gem_unpin_locked(obj); + submit->bos[i].flags &= ~(BO_VMA_PINNED | BO_OBJ_PINNED); msm_gem_unlock(obj); } diff --git a/drivers/gpu/drm/nouveau/Kbuild b/drivers/gpu/drm/nouveau/Kbuild index 60586fb8275e..5e5617006da5 100644 --- a/drivers/gpu/drm/nouveau/Kbuild +++ b/drivers/gpu/drm/nouveau/Kbuild @@ -54,10 +54,6 @@ nouveau-y += nouveau_bios.o nouveau-y += nouveau_connector.o nouveau-y += nouveau_display.o nouveau-y += nouveau_dp.o -nouveau-y += nouveau_fbcon.o -nouveau-y += nv04_fbcon.o -nouveau-y += nv50_fbcon.o -nouveau-y += nvc0_fbcon.o include $(src)/dispnv04/Kbuild include $(src)/dispnv50/Kbuild diff --git a/drivers/gpu/drm/nouveau/dispnv04/crtc.c b/drivers/gpu/drm/nouveau/dispnv04/crtc.c index ee92d576d277..0e0f117bc70b 100644 --- a/drivers/gpu/drm/nouveau/dispnv04/crtc.c +++ b/drivers/gpu/drm/nouveau/dispnv04/crtc.c @@ -23,6 +23,7 @@ * DEALINGS IN THE SOFTWARE. */ #include <drm/drm_crtc_helper.h> +#include <drm/drm_fb_helper.h> #include <drm/drm_fourcc.h> #include <drm/drm_plane_helper.h> #include <drm/drm_vblank.h> @@ -37,7 +38,6 @@ #include "nouveau_crtc.h" #include "hw.h" #include "nvreg.h" -#include "nouveau_fbcon.h" #include "disp.h" #include "nouveau_dma.h" @@ -761,7 +761,8 @@ static void nv_crtc_destroy(struct drm_crtc *crtc) nouveau_bo_unmap(nv_crtc->cursor.nvbo); nouveau_bo_unpin(nv_crtc->cursor.nvbo); nouveau_bo_ref(NULL, &nv_crtc->cursor.nvbo); - nvif_notify_dtor(&nv_crtc->vblank); + nvif_event_dtor(&nv_crtc->vblank); + nvif_head_dtor(&nv_crtc->head); kfree(nv_crtc); } @@ -914,14 +915,6 @@ nv04_crtc_mode_set_base_atomic(struct drm_crtc *crtc, struct drm_framebuffer *fb, int x, int y, enum mode_set_atomic state) { - struct nouveau_drm *drm = nouveau_drm(crtc->dev); - struct drm_device *dev = drm->dev; - - if (state == ENTER_ATOMIC_MODE_SET) - nouveau_fbcon_accel_save_disable(dev); - else - nouveau_fbcon_accel_restore(dev); - return nv04_crtc_do_mode_set_base(crtc, fb, x, y, true); } @@ -1080,10 +1073,10 @@ nv04_finish_page_flip(struct nouveau_channel *chan, } int -nv04_flip_complete(struct nvif_notify *notify) +nv04_flip_complete(struct nvif_event *event, void *argv, u32 argc) { - struct nouveau_cli *cli = (void *)notify->object->client; - struct nouveau_drm *drm = cli->drm; + struct nv04_display *disp = container_of(event, typeof(*disp), flip); + struct nouveau_drm *drm = disp->drm; struct nouveau_channel *chan = drm->channel; struct nv04_page_flip_state state; @@ -1094,7 +1087,7 @@ nv04_flip_complete(struct nvif_notify *notify) state.bpp / 8); } - return NVIF_NOTIFY_KEEP; + return NVIF_EVENT_KEEP; } static int @@ -1279,13 +1272,13 @@ static const struct drm_plane_funcs nv04_primary_plane_funcs = { DRM_PLANE_NON_ATOMIC_FUNCS, }; -static int nv04_crtc_vblank_handler(struct nvif_notify *notify) +static int +nv04_crtc_vblank_handler(struct nvif_event *event, void *repv, u32 repc) { - struct nouveau_crtc *nv_crtc = - container_of(notify, struct nouveau_crtc, vblank); + struct nouveau_crtc *nv_crtc = container_of(event, struct nouveau_crtc, vblank); drm_crtc_handle_vblank(&nv_crtc->base); - return NVIF_NOTIFY_KEEP; + return NVIF_EVENT_KEEP; } int @@ -1341,14 +1334,10 @@ nv04_crtc_create(struct drm_device *dev, int crtc_num) nv04_cursor_init(nv_crtc); - ret = nvif_notify_ctor(&disp->disp.object, "kmsVbl", nv04_crtc_vblank_handler, - false, NV04_DISP_NTFY_VBLANK, - &(struct nvif_notify_head_req_v0) { - .head = nv_crtc->index, - }, - sizeof(struct nvif_notify_head_req_v0), - sizeof(struct nvif_notify_head_rep_v0), - &nv_crtc->vblank); + ret = nvif_head_ctor(&disp->disp, nv_crtc->base.name, nv_crtc->index, &nv_crtc->head); + if (ret) + return ret; - return ret; + return nvif_head_vblank_event_ctor(&nv_crtc->head, "kmsVbl", nv04_crtc_vblank_handler, + false, &nv_crtc->vblank); } diff --git a/drivers/gpu/drm/nouveau/dispnv04/disp.c b/drivers/gpu/drm/nouveau/dispnv04/disp.c index 99fee4d8cd31..e9ac3fb27ff7 100644 --- a/drivers/gpu/drm/nouveau/dispnv04/disp.c +++ b/drivers/gpu/drm/nouveau/dispnv04/disp.c @@ -61,7 +61,7 @@ nv04_display_fini(struct drm_device *dev, bool runtime, bool suspend) struct drm_crtc *crtc; /* Disable flip completion events. */ - nvif_notify_put(&disp->flip); + nvif_event_block(&disp->flip); /* Disable vblank interrupts. */ NVWriteCRTC(dev, 0, NV_PCRTC_INTR_EN_0, 0); @@ -121,7 +121,7 @@ nv04_display_init(struct drm_device *dev, bool resume, bool runtime) encoder->enc_save(&encoder->base.base); /* Enable flip completion events. */ - nvif_notify_get(&disp->flip); + nvif_event_allow(&disp->flip); if (!resume) return 0; @@ -202,7 +202,7 @@ nv04_display_destroy(struct drm_device *dev) nouveau_hw_save_vga_fonts(dev, 0); - nvif_notify_dtor(&disp->flip); + nvif_event_dtor(&disp->flip); nouveau_display(dev)->priv = NULL; vfree(disp); @@ -227,6 +227,8 @@ nv04_display_create(struct drm_device *dev) if (!disp) return -ENOMEM; + disp->drm = drm; + nvif_object_map(&drm->client.device.object, NULL, 0); nouveau_display(dev)->priv = disp; @@ -239,9 +241,10 @@ nv04_display_create(struct drm_device *dev) /* Request page flip completion event. */ if (drm->channel) { - nvif_notify_ctor(&drm->channel->nvsw, "kmsFlip", nv04_flip_complete, - false, NV04_NVSW_NTFY_UEVENT, - NULL, 0, 0, &disp->flip); + ret = nvif_event_ctor(&drm->channel->nvsw, "kmsFlip", 0, nv04_flip_complete, + true, NULL, 0, &disp->flip); + if (ret) + return ret; } nouveau_hw_save_vga_fonts(dev, 1); diff --git a/drivers/gpu/drm/nouveau/dispnv04/disp.h b/drivers/gpu/drm/nouveau/dispnv04/disp.h index f0a24126641a..11a6663758ec 100644 --- a/drivers/gpu/drm/nouveau/dispnv04/disp.h +++ b/drivers/gpu/drm/nouveau/dispnv04/disp.h @@ -6,6 +6,8 @@ #include "nouveau_display.h" +#include <nvif/event.h> + struct nouveau_encoder; enum nv04_fp_display_regs { @@ -84,7 +86,8 @@ struct nv04_display { uint32_t saved_vga_font[4][16384]; uint32_t dac_users[4]; struct nouveau_bo *image[2]; - struct nvif_notify flip; + struct nvif_event flip; + struct nouveau_drm *drm; }; static inline struct nv04_display * @@ -179,5 +182,5 @@ nouveau_bios_run_init_table(struct drm_device *dev, u16 table, ); } -int nv04_flip_complete(struct nvif_notify *); +int nv04_flip_complete(struct nvif_event *, void *, u32); #endif diff --git a/drivers/gpu/drm/nouveau/dispnv50/crc.c b/drivers/gpu/drm/nouveau/dispnv50/crc.c index b834e8a9ae77..9c942fbd836d 100644 --- a/drivers/gpu/drm/nouveau/dispnv50/crc.c +++ b/drivers/gpu/drm/nouveau/dispnv50/crc.c @@ -463,7 +463,7 @@ void nv50_crc_atomic_set(struct nv50_head *head, if (!outp) return; - func->set_src(head, outp->or, nv50_crc_source_type(outp, asyh->crc.src), + func->set_src(head, outp->outp.or.id, nv50_crc_source_type(outp, asyh->crc.src), &crc->ctx[crc->ctx_idx]); } diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c b/drivers/gpu/drm/nouveau/dispnv50/disp.c index aa94f8e284dd..edcb2529b402 100644 --- a/drivers/gpu/drm/nouveau/dispnv50/disp.c +++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c @@ -46,8 +46,8 @@ #include <nvif/class.h> #include <nvif/cl0002.h> -#include <nvif/cl5070.h> #include <nvif/event.h> +#include <nvif/if0012.h> #include <nvif/if0014.h> #include <nvif/timer.h> @@ -64,7 +64,6 @@ #include "nouveau_connector.h" #include "nouveau_encoder.h" #include "nouveau_fence.h" -#include "nouveau_fbcon.h" #include <subdev/bios/dp.h> @@ -317,52 +316,6 @@ nv50_outp_dump_caps(struct nouveau_drm *drm, outp->base.base.name, outp->caps.dp_interlace); } -static void -nv50_outp_release(struct nouveau_encoder *nv_encoder) -{ - struct nv50_disp *disp = nv50_disp(nv_encoder->base.base.dev); - struct { - struct nv50_disp_mthd_v1 base; - } args = { - .base.version = 1, - .base.method = NV50_DISP_MTHD_V1_RELEASE, - .base.hasht = nv_encoder->dcb->hasht, - .base.hashm = nv_encoder->dcb->hashm, - }; - - nvif_mthd(&disp->disp->object, 0, &args, sizeof(args)); - nv_encoder->or = -1; - nv_encoder->link = 0; -} - -static int -nv50_outp_acquire(struct nouveau_encoder *nv_encoder, bool hda) -{ - struct nouveau_drm *drm = nouveau_drm(nv_encoder->base.base.dev); - struct nv50_disp *disp = nv50_disp(drm->dev); - struct { - struct nv50_disp_mthd_v1 base; - struct nv50_disp_acquire_v0 info; - } args = { - .base.version = 1, - .base.method = NV50_DISP_MTHD_V1_ACQUIRE, - .base.hasht = nv_encoder->dcb->hasht, - .base.hashm = nv_encoder->dcb->hashm, - .info.hda = hda, - }; - int ret; - - ret = nvif_mthd(&disp->disp->object, 0, &args, sizeof(args)); - if (ret) { - NV_ERROR(drm, "error acquiring output path: %d\n", ret); - return ret; - } - - nv_encoder->or = args.info.or; - nv_encoder->link = args.info.link; - return 0; -} - static int nv50_outp_atomic_check_view(struct drm_encoder *encoder, struct drm_crtc_state *crtc_state, @@ -489,9 +442,9 @@ nv50_dac_atomic_disable(struct drm_encoder *encoder, struct drm_atomic_state *st struct nv50_core *core = nv50_disp(encoder->dev)->core; const u32 ctrl = NVDEF(NV507D, DAC_SET_CONTROL, OWNER, NONE); - core->func->dac->ctrl(core, nv_encoder->or, ctrl, NULL); + core->func->dac->ctrl(core, nv_encoder->outp.or.id, ctrl, NULL); nv_encoder->crtc = NULL; - nv50_outp_release(nv_encoder); + nvif_outp_release(&nv_encoder->outp); } static void @@ -516,9 +469,9 @@ nv50_dac_atomic_enable(struct drm_encoder *encoder, struct drm_atomic_state *sta ctrl |= NVDEF(NV507D, DAC_SET_CONTROL, PROTOCOL, RGB_CRT); - nv50_outp_acquire(nv_encoder, false); + nvif_outp_acquire_rgb_crt(&nv_encoder->outp); - core->func->dac->ctrl(core, nv_encoder->or, ctrl, asyh); + core->func->dac->ctrl(core, nv_encoder->outp.or.id, ctrl, asyh); asyh->or.depth = 0; nv_encoder->crtc = &nv_crtc->base; @@ -634,7 +587,7 @@ nv50_audio_component_get_eld(struct device *kdev, int port, int dev_id, nv_connector = nouveau_connector(nv_encoder->audio.connector); nv_crtc = nouveau_crtc(nv_encoder->crtc); - if (!nv_crtc || nv_encoder->or != port || nv_crtc->index != dev_id) + if (!nv_crtc || nv_encoder->outp.or.id != port || nv_crtc->index != dev_id) continue; *enabled = nv_encoder->audio.enabled; @@ -718,33 +671,37 @@ nv50_audio_component_fini(struct nouveau_drm *drm) /****************************************************************************** * Audio *****************************************************************************/ +static bool +nv50_audio_supported(struct drm_encoder *encoder) +{ + struct nv50_disp *disp = nv50_disp(encoder->dev); + + if (disp->disp->object.oclass <= GT200_DISP || + disp->disp->object.oclass == GT206_DISP) + return false; + + return true; +} + static void nv50_audio_disable(struct drm_encoder *encoder, struct nouveau_crtc *nv_crtc) { struct nouveau_drm *drm = nouveau_drm(encoder->dev); struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); - struct nv50_disp *disp = nv50_disp(encoder->dev); - struct { - struct nv50_disp_mthd_v1 base; - struct nv50_disp_sor_hda_eld_v0 eld; - } args = { - .base.version = 1, - .base.method = NV50_DISP_MTHD_V1_SOR_HDA_ELD, - .base.hasht = nv_encoder->dcb->hasht, - .base.hashm = (0xf0ff & nv_encoder->dcb->hashm) | - (0x0100 << nv_crtc->index), - }; + struct nvif_outp *outp = &nv_encoder->outp; + + if (!nv50_audio_supported(encoder)) + return; mutex_lock(&drm->audio.lock); if (nv_encoder->audio.enabled) { nv_encoder->audio.enabled = false; nv_encoder->audio.connector = NULL; - nvif_mthd(&disp->disp->object, 0, &args, sizeof(args)); + nvif_outp_hda_eld(&nv_encoder->outp, nv_crtc->index, NULL, 0); } mutex_unlock(&drm->audio.lock); - nv50_audio_component_eld_notify(drm->audio.component, nv_encoder->or, - nv_crtc->index); + nv50_audio_component_eld_notify(drm->audio.component, outp->or.id, nv_crtc->index); } static void @@ -754,159 +711,101 @@ nv50_audio_enable(struct drm_encoder *encoder, struct nouveau_crtc *nv_crtc, { struct nouveau_drm *drm = nouveau_drm(encoder->dev); struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); - struct nv50_disp *disp = nv50_disp(encoder->dev); - struct __packed { - struct { - struct nv50_disp_mthd_v1 mthd; - struct nv50_disp_sor_hda_eld_v0 eld; - } base; - u8 data[sizeof(nv_connector->base.eld)]; - } args = { - .base.mthd.version = 1, - .base.mthd.method = NV50_DISP_MTHD_V1_SOR_HDA_ELD, - .base.mthd.hasht = nv_encoder->dcb->hasht, - .base.mthd.hashm = (0xf0ff & nv_encoder->dcb->hashm) | - (0x0100 << nv_crtc->index), - }; - - if (!drm_detect_monitor_audio(nv_connector->edid)) + struct nvif_outp *outp = &nv_encoder->outp; + + if (!nv50_audio_supported(encoder) || !drm_detect_monitor_audio(nv_connector->edid)) return; mutex_lock(&drm->audio.lock); - memcpy(args.data, nv_connector->base.eld, sizeof(args.data)); - - nvif_mthd(&disp->disp->object, 0, &args, - sizeof(args.base) + drm_eld_size(args.data)); + nvif_outp_hda_eld(&nv_encoder->outp, nv_crtc->index, nv_connector->base.eld, + drm_eld_size(nv_connector->base.eld)); nv_encoder->audio.enabled = true; nv_encoder->audio.connector = &nv_connector->base; mutex_unlock(&drm->audio.lock); - nv50_audio_component_eld_notify(drm->audio.component, nv_encoder->or, - nv_crtc->index); + nv50_audio_component_eld_notify(drm->audio.component, outp->or.id, nv_crtc->index); } /****************************************************************************** * HDMI *****************************************************************************/ static void -nv50_hdmi_disable(struct drm_encoder *encoder, struct nouveau_crtc *nv_crtc) -{ - struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); - struct nv50_disp *disp = nv50_disp(encoder->dev); - struct { - struct nv50_disp_mthd_v1 base; - struct nv50_disp_sor_hdmi_pwr_v0 pwr; - } args = { - .base.version = 1, - .base.method = NV50_DISP_MTHD_V1_SOR_HDMI_PWR, - .base.hasht = nv_encoder->dcb->hasht, - .base.hashm = (0xf0ff & nv_encoder->dcb->hashm) | - (0x0100 << nv_crtc->index), - }; - - nvif_mthd(&disp->disp->object, 0, &args, sizeof(args)); -} - -static void nv50_hdmi_enable(struct drm_encoder *encoder, struct nouveau_crtc *nv_crtc, struct nouveau_connector *nv_connector, struct drm_atomic_state *state, - struct drm_display_mode *mode) + struct drm_display_mode *mode, bool hda) { struct nouveau_drm *drm = nouveau_drm(encoder->dev); struct nouveau_encoder *nv_encoder = nouveau_encoder(encoder); - struct nv50_disp *disp = nv50_disp(encoder->dev); - struct { - struct nv50_disp_mthd_v1 base; - struct nv50_disp_sor_hdmi_pwr_v0 pwr; - u8 infoframes[2 * 17]; /* two frames, up to 17 bytes each */ - } args = { - .base.version = 1, - .base.method = NV50_DISP_MTHD_V1_SOR_HDMI_PWR, - .base.hasht = nv_encoder->dcb->hasht, - .base.hashm = (0xf0ff & nv_encoder->dcb->hashm) | - (0x0100 << nv_crtc->index), - .pwr.state = 1, - .pwr.rekey = 56, /* binary driver, and tegra, constant */ - }; - struct drm_hdmi_info *hdmi; + struct drm_hdmi_info *hdmi = &nv_connector->base.display_info.hdmi; + union hdmi_infoframe infoframe = { 0 }; + const u8 rekey = 56; /* binary driver, and tegra, constant */ + u8 scdc = 0; u32 max_ac_packet; - union hdmi_infoframe avi_frame; - union hdmi_infoframe vendor_frame; - bool high_tmds_clock_ratio = false, scrambling = false; - u8 config; - int ret; - int size; - - if (!drm_detect_hdmi_monitor(nv_connector->edid)) - return; - - hdmi = &nv_connector->base.display_info.hdmi; - - ret = drm_hdmi_avi_infoframe_from_display_mode(&avi_frame.avi, - &nv_connector->base, mode); - if (!ret) { - drm_hdmi_avi_infoframe_quant_range(&avi_frame.avi, - &nv_connector->base, mode, - HDMI_QUANTIZATION_RANGE_FULL); - /* We have an AVI InfoFrame, populate it to the display */ - args.pwr.avi_infoframe_length - = hdmi_infoframe_pack(&avi_frame, args.infoframes, 17); - } - - ret = drm_hdmi_vendor_infoframe_from_display_mode(&vendor_frame.vendor.hdmi, - &nv_connector->base, mode); - if (!ret) { - /* We have a Vendor InfoFrame, populate it to the display */ - args.pwr.vendor_infoframe_length - = hdmi_infoframe_pack(&vendor_frame, - args.infoframes - + args.pwr.avi_infoframe_length, - 17); - } + struct { + struct nvif_outp_infoframe_v0 infoframe; + u8 data[17]; + } args = { 0 }; + int ret, size; max_ac_packet = mode->htotal - mode->hdisplay; - max_ac_packet -= args.pwr.rekey; + max_ac_packet -= rekey; max_ac_packet -= 18; /* constant from tegra */ - args.pwr.max_ac_packet = max_ac_packet / 32; + max_ac_packet /= 32; if (hdmi->scdc.scrambling.supported) { - high_tmds_clock_ratio = mode->clock > 340000; - scrambling = high_tmds_clock_ratio || - hdmi->scdc.scrambling.low_rates; - } + const bool high_tmds_clock_ratio = mode->clock > 340000; - args.pwr.scdc = - NV50_DISP_SOR_HDMI_PWR_V0_SCDC_SCRAMBLE * scrambling | - NV50_DISP_SOR_HDMI_PWR_V0_SCDC_DIV_BY_4 * high_tmds_clock_ratio; + ret = drm_scdc_readb(nv_encoder->i2c, SCDC_TMDS_CONFIG, &scdc); + if (ret < 0) { + NV_ERROR(drm, "Failure to read SCDC_TMDS_CONFIG: %d\n", ret); + return; + } - size = sizeof(args.base) - + sizeof(args.pwr) - + args.pwr.avi_infoframe_length - + args.pwr.vendor_infoframe_length; - nvif_mthd(&disp->disp->object, 0, &args, size); + scdc &= ~(SCDC_TMDS_BIT_CLOCK_RATIO_BY_40 | SCDC_SCRAMBLING_ENABLE); + if (high_tmds_clock_ratio || hdmi->scdc.scrambling.low_rates) + scdc |= SCDC_SCRAMBLING_ENABLE; + if (high_tmds_clock_ratio) + scdc |= SCDC_TMDS_BIT_CLOCK_RATIO_BY_40; - nv50_audio_enable(encoder, nv_crtc, nv_connector, state, mode); + ret = drm_scdc_writeb(nv_encoder->i2c, SCDC_TMDS_CONFIG, scdc); + if (ret < 0) + NV_ERROR(drm, "Failure to write SCDC_TMDS_CONFIG = 0x%02x: %d\n", + scdc, ret); + } - /* If SCDC is supported by the downstream monitor, update - * divider / scrambling settings to what we programmed above. - */ - if (!hdmi->scdc.scrambling.supported) + ret = nvif_outp_acquire_tmds(&nv_encoder->outp, nv_crtc->index, true, + max_ac_packet, rekey, scdc, hda); + if (ret) return; - ret = drm_scdc_readb(nv_encoder->i2c, SCDC_TMDS_CONFIG, &config); - if (ret < 0) { - NV_ERROR(drm, "Failure to read SCDC_TMDS_CONFIG: %d\n", ret); - return; + /* AVI InfoFrame. */ + args.infoframe.version = 0; + args.infoframe.head = nv_crtc->index; + + if (!drm_hdmi_avi_infoframe_from_display_mode(&infoframe.avi, &nv_connector->base, mode)) { + drm_hdmi_avi_infoframe_quant_range(&infoframe.avi, &nv_connector->base, mode, + HDMI_QUANTIZATION_RANGE_FULL); + + size = hdmi_infoframe_pack(&infoframe, args.data, ARRAY_SIZE(args.data)); + } else { + size = 0; } - config &= ~(SCDC_TMDS_BIT_CLOCK_RATIO_BY_40 | SCDC_SCRAMBLING_ENABLE); - config |= SCDC_TMDS_BIT_CLOCK_RATIO_BY_40 * high_tmds_clock_ratio; - config |= SCDC_SCRAMBLING_ENABLE * scrambling; - ret = drm_scdc_writeb(nv_encoder->i2c, SCDC_TMDS_CONFIG, config); - if (ret < 0) - NV_ERROR(drm, "Failure to write SCDC_TMDS_CONFIG = 0x%02x: %d\n", - config, ret); + + nvif_outp_infoframe(&nv_encoder->outp, NVIF_OUTP_INFOFRAME_V0_AVI, &args.infoframe, size); + + /* Vendor InfoFrame. */ + memset(&args.data, 0, sizeof(args.data)); + if (!drm_hdmi_vendor_infoframe_from_display_mode(&infoframe.vendor.hdmi, + &nv_connector->base, mode)) + size = hdmi_infoframe_pack(&infoframe, args.data, ARRAY_SIZE(args.data)); + else + size = 0; + + nvif_outp_infoframe(&nv_encoder->outp, NVIF_OUTP_INFOFRAME_V0_VSI, &args.infoframe, size); + + nv50_audio_enable(encoder, nv_crtc, nv_connector, state, mode); } /****************************************************************************** @@ -979,16 +878,6 @@ nv50_msto_prepare(struct drm_atomic_state *state, struct nv50_mstc *mstc = msto->mstc; struct nv50_mstm *mstm = mstc->mstm; struct drm_dp_mst_atomic_payload *payload; - struct { - struct nv50_disp_mthd_v1 base; - struct nv50_disp_sor_dp_mst_vcpi_v0 vcpi; - } args = { - .base.version = 1, - .base.method = NV50_DISP_MTHD_V1_SOR_DP_MST_VCPI, - .base.hasht = mstm->outp->dcb->hasht, - .base.hashm = (0xf0ff & mstm->outp->dcb->hashm) | - (0x0100 << msto->head->base.index), - }; NV_ATOMIC(drm, "%s: msto prepare\n", msto->encoder.name); @@ -997,22 +886,16 @@ nv50_msto_prepare(struct drm_atomic_state *state, // TODO: Figure out if we want to do a better job of handling VCPI allocation failures here? if (msto->disabled) { drm_dp_remove_payload(mgr, mst_state, payload); + + nvif_outp_dp_mst_vcpi(&mstm->outp->outp, msto->head->base.index, 0, 0, 0, 0); } else { if (msto->enabled) drm_dp_add_payload_part1(mgr, mst_state, payload); - args.vcpi.start_slot = payload->vc_start_slot; - args.vcpi.num_slots = payload->time_slots; - args.vcpi.pbn = payload->pbn; - args.vcpi.aligned_pbn = payload->time_slots * mst_state->pbn_div; + nvif_outp_dp_mst_vcpi(&mstm->outp->outp, msto->head->base.index, + payload->vc_start_slot, payload->time_slots, + payload->pbn, payload->time_slots * mst_state->pbn_div); } - - NV_ATOMIC(drm, "%s: %s: %02x %02x %04x %04x\n", - msto->encoder.name, msto->head->base.base.name, - args.vcpi.start_slot, args.vcpi.num_slots, - args.vcpi.pbn, args.vcpi.aligned_pbn); - - nvif_mthd(&drm->display->disp.object, 0, &args, sizeof(args)); } static int @@ -1107,10 +990,12 @@ nv50_msto_atomic_enable(struct drm_encoder *encoder, struct drm_atomic_state *st if (WARN_ON(!mstc)) return; - if (!mstm->links++) - nv50_outp_acquire(mstm->outp, false /*XXX: MST audio.*/); + if (!mstm->links++) { + /*XXX: MST audio. */ + nvif_outp_acquire_dp(&mstm->outp->outp, mstm->outp->dp.dpcd, 0, 0, false, true); + } - if (mstm->outp->link & 1) + if (mstm->outp->outp.or.link & 1) proto = NV917D_SOR_SET_CONTROL_PROTOCOL_DP_A; else proto = NV917D_SOR_SET_CONTROL_PROTOCOL_DP_B; @@ -1405,7 +1290,7 @@ nv50_mstm_prepare(struct drm_atomic_state *state, if (mstm->disabled) { if (!mstm->links) - nv50_outp_release(mstm->outp); + nvif_outp_release(&mstm->outp->outp); mstm->disabled = false; } } @@ -1473,26 +1358,6 @@ nv50_mstm_remove(struct nv50_mstm *mstm) drm_dp_mst_topology_mgr_set_mst(&mstm->mgr, false); } -static int -nv50_mstm_enable(struct nv50_mstm *mstm, int state) -{ - struct nouveau_encoder *outp = mstm->outp; - struct { - struct nv50_disp_mthd_v1 base; - struct nv50_disp_sor_dp_mst_link_v0 mst; - } args = { - .base.version = 1, - .base.method = NV50_DISP_MTHD_V1_SOR_DP_MST_LINK, - .base.hasht = outp->dcb->hasht, - .base.hashm = outp->dcb->hashm, - .mst.state = state, - }; - struct nouveau_drm *drm = nouveau_drm(outp->base.base.dev); - struct nvif_object *disp = &drm->display->disp.object; - - return nvif_mthd(disp, 0, &args, sizeof(args)); -} - int nv50_mstm_detect(struct nouveau_encoder *outp) { @@ -1513,15 +1378,9 @@ nv50_mstm_detect(struct nouveau_encoder *outp) return ret; /* And start enabling */ - ret = nv50_mstm_enable(mstm, true); - if (ret) - return ret; - ret = drm_dp_mst_topology_mgr_set_mst(&mstm->mgr, true); - if (ret) { - nv50_mstm_enable(mstm, false); + if (ret) return ret; - } mstm->is_mst = true; return 1; @@ -1623,7 +1482,7 @@ nv50_sor_update(struct nouveau_encoder *nv_encoder, u8 head, asyh->or.depth = depth; } - core->func->sor->ctrl(core, nv_encoder->or, nv_encoder->ctrl, asyh); + core->func->sor->ctrl(core, nv_encoder->outp.or.id, nv_encoder->ctrl, asyh); } /* TODO: Should we extend this to PWM-only backlights? @@ -1666,8 +1525,7 @@ nv50_sor_atomic_disable(struct drm_encoder *encoder, struct drm_atomic_state *st nv_encoder->update(nv_encoder, nv_crtc->index, NULL, 0, 0); nv50_audio_disable(encoder, nv_crtc); - nv50_hdmi_disable(&nv_encoder->base.base, nv_crtc); - nv50_outp_release(nv_encoder); + nvif_outp_release(&nv_encoder->outp); nv_encoder->crtc = NULL; } @@ -1679,16 +1537,8 @@ nv50_sor_atomic_enable(struct drm_encoder *encoder, struct drm_atomic_state *sta struct nv50_head_atom *asyh = nv50_head_atom(drm_atomic_get_new_crtc_state(state, &nv_crtc->base)); struct drm_display_mode *mode = &asyh->state.adjusted_mode; - struct { - struct nv50_disp_mthd_v1 base; - struct nv50_disp_sor_lvds_script_v0 lvds; - } lvds = { - .base.version = 1, - .base.method = NV50_DISP_MTHD_V1_SOR_LVDS_SCRIPT, - .base.hasht = nv_encoder->dcb->hasht, - .base.hashm = nv_encoder->dcb->hashm, - }; struct nv50_disp *disp = nv50_disp(encoder->dev); + struct nvif_outp *outp = &nv_encoder->outp; struct drm_device *dev = encoder->dev; struct nouveau_drm *drm = nouveau_drm(dev); struct nouveau_connector *nv_connector; @@ -1696,7 +1546,7 @@ nv50_sor_atomic_enable(struct drm_encoder *encoder, struct drm_atomic_state *sta struct nouveau_backlight *backlight; #endif struct nvbios *bios = &drm->vbios; - bool hda = false; + bool lvds_dual = false, lvds_8bpc = false, hda = false; u8 proto = NV507D_SOR_SET_CONTROL_PROTOCOL_CUSTOM; u8 depth = NV837D_SOR_SET_CONTROL_PIXEL_DEPTH_DEFAULT; @@ -1707,11 +1557,16 @@ nv50_sor_atomic_enable(struct drm_encoder *encoder, struct drm_atomic_state *sta disp->disp->object.oclass >= GF110_DISP) && drm_detect_monitor_audio(nv_connector->edid)) hda = true; - nv50_outp_acquire(nv_encoder, hda); switch (nv_encoder->dcb->type) { case DCB_OUTPUT_TMDS: - if (nv_encoder->link & 1) { + if (disp->disp->object.oclass == NV50_DISP || + !drm_detect_hdmi_monitor(nv_connector->edid)) + nvif_outp_acquire_tmds(outp, nv_crtc->index, false, 0, 0, 0, false); + else + nv50_hdmi_enable(encoder, nv_crtc, nv_connector, state, mode, hda); + + if (nv_encoder->outp.or.link & 1) { proto = NV507D_SOR_SET_CONTROL_PROTOCOL_SINGLE_TMDS_A; /* Only enable dual-link if: * - Need to (i.e. rate > 165MHz) @@ -1726,44 +1581,41 @@ nv50_sor_atomic_enable(struct drm_encoder *encoder, struct drm_atomic_state *sta } else { proto = NV507D_SOR_SET_CONTROL_PROTOCOL_SINGLE_TMDS_B; } - - nv50_hdmi_enable(&nv_encoder->base.base, nv_crtc, nv_connector, state, mode); break; case DCB_OUTPUT_LVDS: proto = NV507D_SOR_SET_CONTROL_PROTOCOL_LVDS_CUSTOM; if (bios->fp_no_ddc) { - if (bios->fp.dual_link) - lvds.lvds.script |= 0x0100; - if (bios->fp.if_is_24bit) - lvds.lvds.script |= 0x0200; + lvds_dual = bios->fp.dual_link; + lvds_8bpc = bios->fp.if_is_24bit; } else { if (nv_connector->type == DCB_CONNECTOR_LVDS_SPWG) { if (((u8 *)nv_connector->edid)[121] == 2) - lvds.lvds.script |= 0x0100; + lvds_dual = true; } else if (mode->clock >= bios->fp.duallink_transition_clk) { - lvds.lvds.script |= 0x0100; + lvds_dual = true; } - if (lvds.lvds.script & 0x0100) { + if (lvds_dual) { if (bios->fp.strapless_is_24bit & 2) - lvds.lvds.script |= 0x0200; + lvds_8bpc = true; } else { if (bios->fp.strapless_is_24bit & 1) - lvds.lvds.script |= 0x0200; + lvds_8bpc = true; } if (asyh->or.bpc == 8) - lvds.lvds.script |= 0x0200; + lvds_8bpc = true; } - nvif_mthd(&disp->disp->object, 0, &lvds, sizeof(lvds)); + nvif_outp_acquire_lvds(&nv_encoder->outp, lvds_dual, lvds_8bpc); break; case DCB_OUTPUT_DP: + nvif_outp_acquire_dp(&nv_encoder->outp, nv_encoder->dp.dpcd, 0, 0, hda, false); depth = nv50_dp_bpc_to_depth(asyh->or.bpc); - if (nv_encoder->link & 1) + if (nv_encoder->outp.or.link & 1) proto = NV887D_SOR_SET_CONTROL_PROTOCOL_DP_A; else proto = NV887D_SOR_SET_CONTROL_PROTOCOL_DP_B; @@ -1921,9 +1773,9 @@ nv50_pior_atomic_disable(struct drm_encoder *encoder, struct drm_atomic_state *s struct nv50_core *core = nv50_disp(encoder->dev)->core; const u32 ctrl = NVDEF(NV507D, PIOR_SET_CONTROL, OWNER, NONE); - core->func->pior->ctrl(core, nv_encoder->or, ctrl, NULL); + core->func->pior->ctrl(core, nv_encoder->outp.or.id, ctrl, NULL); nv_encoder->crtc = NULL; - nv50_outp_release(nv_encoder); + nvif_outp_release(&nv_encoder->outp); } static void @@ -1944,8 +1796,6 @@ nv50_pior_atomic_enable(struct drm_encoder *encoder, struct drm_atomic_state *st break; } - nv50_outp_acquire(nv_encoder, false); - switch (asyh->or.bpc) { case 10: asyh->or.depth = NV837D_PIOR_SET_CONTROL_PIXEL_DEPTH_BPP_30_444; break; case 8: asyh->or.depth = NV837D_PIOR_SET_CONTROL_PIXEL_DEPTH_BPP_24_444; break; @@ -1955,15 +1805,19 @@ nv50_pior_atomic_enable(struct drm_encoder *encoder, struct drm_atomic_state *st switch (nv_encoder->dcb->type) { case DCB_OUTPUT_TMDS: + ctrl |= NVDEF(NV507D, PIOR_SET_CONTROL, PROTOCOL, EXT_TMDS_ENC); + nvif_outp_acquire_tmds(&nv_encoder->outp, false, false, 0, 0, 0, false); + break; case DCB_OUTPUT_DP: ctrl |= NVDEF(NV507D, PIOR_SET_CONTROL, PROTOCOL, EXT_TMDS_ENC); + nvif_outp_acquire_dp(&nv_encoder->outp, nv_encoder->dp.dpcd, 0, 0, false, false); break; default: BUG(); break; } - core->func->pior->ctrl(core, nv_encoder->or, ctrl, asyh); + core->func->pior->ctrl(core, nv_encoder->outp.or.id, ctrl, asyh); nv_encoder->crtc = &nv_crtc->base; } @@ -2587,7 +2441,7 @@ nv50_disp_atomic_state_alloc(struct drm_device *dev) static const struct drm_mode_config_funcs nv50_disp_func = { .fb_create = nouveau_user_framebuffer_create, - .output_poll_changed = nouveau_fbcon_output_poll_changed, + .output_poll_changed = drm_fb_helper_output_poll_changed, .atomic_check = nv50_disp_atomic_check, .atomic_commit = nv50_disp_atomic_commit, .atomic_state_alloc = nv50_disp_atomic_state_alloc, diff --git a/drivers/gpu/drm/nouveau/dispnv50/head.c b/drivers/gpu/drm/nouveau/dispnv50/head.c index c3c57be54e1c..f006e56e1e08 100644 --- a/drivers/gpu/drm/nouveau/dispnv50/head.c +++ b/drivers/gpu/drm/nouveau/dispnv50/head.c @@ -517,7 +517,8 @@ nv50_head_destroy(struct drm_crtc *crtc) { struct nv50_head *head = nv50_head(crtc); - nvif_notify_dtor(&head->base.vblank); + nvif_event_dtor(&head->base.vblank); + nvif_head_dtor(&head->base.head); nv50_lut_fini(&head->olut); drm_crtc_cleanup(crtc); kfree(head); @@ -554,15 +555,15 @@ nvd9_head_func = { .late_register = nv50_head_late_register, }; -static int nv50_head_vblank_handler(struct nvif_notify *notify) +static int +nv50_head_vblank_handler(struct nvif_event *event, void *repv, u32 repc) { - struct nouveau_crtc *nv_crtc = - container_of(notify, struct nouveau_crtc, vblank); + struct nouveau_crtc *nv_crtc = container_of(event, struct nouveau_crtc, vblank); if (drm_crtc_handle_vblank(&nv_crtc->base)) nv50_crc_handle_vblank(nv50_head(&nv_crtc->base)); - return NVIF_NOTIFY_KEEP; + return NVIF_EVENT_KEEP; } struct nv50_head * @@ -624,14 +625,12 @@ nv50_head_create(struct drm_device *dev, int index) } } - ret = nvif_notify_ctor(&disp->disp->object, "kmsVbl", nv50_head_vblank_handler, - false, NV04_DISP_NTFY_VBLANK, - &(struct nvif_notify_head_req_v0) { - .head = nv_crtc->index, - }, - sizeof(struct nvif_notify_head_req_v0), - sizeof(struct nvif_notify_head_rep_v0), - &nv_crtc->vblank); + ret = nvif_head_ctor(disp->disp, head->base.base.name, head->base.index, &head->base.head); + if (ret) + return ERR_PTR(ret); + + ret = nvif_head_vblank_event_ctor(&head->base.head, "kmsVbl", nv50_head_vblank_handler, + false, &nv_crtc->vblank); if (ret) return ERR_PTR(ret); diff --git a/drivers/gpu/drm/nouveau/include/nvfw/acr.h b/drivers/gpu/drm/nouveau/include/nvfw/acr.h index e65d6a8db104..6f19560bc54b 100644 --- a/drivers/gpu/drm/nouveau/include/nvfw/acr.h +++ b/drivers/gpu/drm/nouveau/include/nvfw/acr.h @@ -39,6 +39,23 @@ struct wpr_header_v1 { void wpr_header_v1_dump(struct nvkm_subdev *, const struct wpr_header_v1 *); +struct wpr_generic_header { +#define WPR_GENERIC_HEADER_ID_LSF_UCODE_DESC 1 +#define WPR_GENERIC_HEADER_ID_LSF_WPR_HEADER 2 +#define WPR_GENERIC_HEADER_ID_LSF_SHARED_SUB_WPR 3 +#define WPR_GENERIC_HEADER_ID_LSF_LSB_HEADER 4 + u16 identifier; + u16 version; + u32 size; +}; + +struct wpr_header_v2 { + struct wpr_generic_header hdr; + struct wpr_header_v1 wpr; +}; + +void wpr_header_v2_dump(struct nvkm_subdev *, const struct wpr_header_v2 *); + struct lsf_signature { u8 prd_keys[2][16]; u8 dbg_keys[2][16]; @@ -89,6 +106,74 @@ struct lsb_header_v1 { void lsb_header_v1_dump(struct nvkm_subdev *, struct lsb_header_v1 *); +struct lsb_header_v2 { + struct wpr_generic_header hdr; + struct lsf_signature_v2 { + struct wpr_generic_header hdr; + u32 falcon_id; + u8 prd_present; + u8 dbg_present; + u16 reserved; + u32 sig_size; + u8 prod_sig[2][384 + 128]; + u8 debug_sig[2][384 + 128]; + u16 sig_algo_ver; + u16 sig_algo; + u16 hash_algo_ver; + u16 hash_algo; + u32 sig_algo_padding_type; + u8 depmap[11 * 2 * 4]; + u32 depmap_count; + u8 supports_versioning; + u8 pad[3]; + u32 ls_ucode_version; + u32 ls_ucode_id; + u32 ucode_ls_encrypted; + u32 ls_eng_algo_type; + u32 ls_eng_algo_ver; + u8 ls_enc_iv[16]; + u8 rsvd[36]; + } signature; + u32 ucode_off; + u32 ucode_size; + u32 data_size; + u32 bl_code_size; + u32 bl_imem_off; + u32 bl_data_off; + u32 bl_data_size; + u32 rsvd0; + u32 app_code_off; + u32 app_code_size; + u32 app_data_off; + u32 app_data_size; + u32 app_imem_offset; + u32 app_dmem_offset; + u32 flags; + u32 monitor_code_offset; + u32 monitor_data_offset; + u32 manifest_offset; + struct hs_fmc_params { + u8 hs_fmc; + u8 padding[3]; + u16 pkc_algo; + u16 pkc_algo_version; + u32 engid_mask; + u32 ucode_id; + u32 fuse_ver; + u8 pkc_signature[384 + 128]; + u8 pkc_key[2048]; + u8 rsvd[4]; + } hs_fmc_params; + struct hs_ovl_sig_blob_params { + u8 hs_ovl_sig_blob_present; + u32 hs_ovl_sig_blob_offset; + u32 hs_ovl_sig_blob_size; + } hs_ovl_sig_blob_params; + u8 rsvd[20]; +}; + +void lsb_header_v2_dump(struct nvkm_subdev *, struct lsb_header_v2 *); + struct flcn_acr_desc { union { u8 reserved_dmem[0x200]; diff --git a/drivers/gpu/drm/nouveau/include/nvfw/hs.h b/drivers/gpu/drm/nouveau/include/nvfw/hs.h index b53bbc4cd130..8c4cd08a7b5f 100644 --- a/drivers/gpu/drm/nouveau/include/nvfw/hs.h +++ b/drivers/gpu/drm/nouveau/include/nvfw/hs.h @@ -17,6 +17,20 @@ struct nvfw_hs_header { const struct nvfw_hs_header *nvfw_hs_header(struct nvkm_subdev *, const void *); +struct nvfw_hs_header_v2 { + u32 sig_prod_offset; + u32 sig_prod_size; + u32 patch_loc; + u32 patch_sig; + u32 meta_data_offset; + u32 meta_data_size; + u32 num_sig; + u32 header_offset; + u32 header_size; +}; + +const struct nvfw_hs_header_v2 *nvfw_hs_header_v2(struct nvkm_subdev *, const void *); + struct nvfw_hs_load_header { u32 non_sec_code_off; u32 non_sec_code_size; @@ -28,4 +42,18 @@ struct nvfw_hs_load_header { const struct nvfw_hs_load_header * nvfw_hs_load_header(struct nvkm_subdev *, const void *); + +struct nvfw_hs_load_header_v2 { + u32 os_code_offset; + u32 os_code_size; + u32 os_data_offset; + u32 os_data_size; + u32 num_apps; + struct { + u32 offset; + u32 size; + } app[0]; +}; + +const struct nvfw_hs_load_header_v2 *nvfw_hs_load_header_v2(struct nvkm_subdev *, const void *); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvfw/ls.h b/drivers/gpu/drm/nouveau/include/nvfw/ls.h index f63692a2a16c..d531121bfa35 100644 --- a/drivers/gpu/drm/nouveau/include/nvfw/ls.h +++ b/drivers/gpu/drm/nouveau/include/nvfw/ls.h @@ -50,4 +50,55 @@ struct nvfw_ls_desc_v1 { const struct nvfw_ls_desc_v1 * nvfw_ls_desc_v1(struct nvkm_subdev *, const void *); + +struct nvfw_ls_desc_v2 { + u32 descriptor_size; + u32 image_size; + u32 tools_version; + u32 app_version; + char date[64]; + u32 secure_bootloader; + u32 bootloader_start_offset; + u32 bootloader_size; + u32 bootloader_imem_offset; + u32 bootloader_entry_point; + u32 app_start_offset; + u32 app_size; + u32 app_imem_offset; + u32 app_imem_entry; + u32 app_dmem_offset; + u32 app_resident_code_offset; + u32 app_resident_code_size; + u32 app_resident_data_offset; + u32 app_resident_data_size; + u32 nb_imem_overlays; + u32 nb_dmem_overlays; + struct { + u32 start; + u32 size; + } load_ovl[64]; +}; + +const struct nvfw_ls_desc_v2 *nvfw_ls_desc_v2(struct nvkm_subdev *, const void *); + +struct nvfw_ls_hsbl_bin_hdr { + u32 bin_magic; + u32 bin_ver; + u32 bin_size; + u32 header_offset; +}; + +const struct nvfw_ls_hsbl_bin_hdr *nvfw_ls_hsbl_bin_hdr(struct nvkm_subdev *, const void *); + +struct nvfw_ls_hsbl_hdr { + u32 sig_prod_offset; + u32 sig_prod_size; + u32 patch_loc; + u32 patch_sig; + u32 meta_data_offset; + u32 meta_data_size; + u32 num_sig; +}; + +const struct nvfw_ls_hsbl_hdr *nvfw_ls_hsbl_hdr(struct nvkm_subdev *, const void *); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvfw/sec2.h b/drivers/gpu/drm/nouveau/include/nvfw/sec2.h index 9a37ad4179cb..b3331d679c4e 100644 --- a/drivers/gpu/drm/nouveau/include/nvfw/sec2.h +++ b/drivers/gpu/drm/nouveau/include/nvfw/sec2.h @@ -10,6 +10,7 @@ struct nv_sec2_args { }; #define NV_SEC2_UNIT_INIT 0x01 +#define NV_SEC2_UNIT_UNLOAD 0x06 #define NV_SEC2_UNIT_ACR 0x08 struct nv_sec2_init_msg { @@ -33,6 +34,29 @@ struct nv_sec2_init_msg { u16 sw_managed_area_size; }; +struct nv_sec2_init_msg_v1 { + struct nvfw_falcon_msg hdr; +#define NV_SEC2_INIT_MSG_INIT 0x00 + u8 msg_type; + + u8 num_queues; + u16 os_debug_entry_point; + + struct { + u32 offset; + u16 size; + u8 index; +#define NV_SEC2_INIT_MSG_QUEUE_ID_CMDQ 0x00 +#define NV_SEC2_INIT_MSG_QUEUE_ID_MSGQ 0x01 + u8 id; + } queue_info[2]; + + u32 sw_managed_area_offset; + u16 sw_managed_area_size; + + u32 unkn[8]; +}; + struct nv_sec2_acr_cmd { struct nvfw_falcon_cmd hdr; #define NV_SEC2_ACR_CMD_BOOTSTRAP_FALCON 0x00 @@ -57,4 +81,25 @@ struct nv_sec2_acr_bootstrap_falcon_msg { u32 error_code; u32 falcon_id; }; + +#define NV_SEC2_UNIT_V2_INIT 0x01 +#define NV_SEC2_UNIT_V2_UNLOAD 0x05 +#define NV_SEC2_UNIT_V2_ACR 0x07 + +struct nv_sec2_acr_bootstrap_falcon_cmd_v1 { + struct nv_sec2_acr_cmd cmd; +#define NV_SEC2_ACR_BOOTSTRAP_FALCON_FLAGS_RESET_YES 0x00000000 +#define NV_SEC2_ACR_BOOTSTRAP_FALCON_FLAGS_RESET_NO 0x00000001 + u32 flags; + u32 falcon_id; + u32 unkn08; + u32 unkn0c; +}; + +struct nv_sec2_acr_bootstrap_falcon_msg_v1 { + struct nv_sec2_acr_msg msg; + u32 error_code; + u32 falcon_id; + u32 unkn08; +}; #endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/cl0046.h b/drivers/gpu/drm/nouveau/include/nvif/cl0046.h index d490d401870a..eca7c3950654 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/cl0046.h +++ b/drivers/gpu/drm/nouveau/include/nvif/cl0046.h @@ -2,28 +2,5 @@ #ifndef __NVIF_CL0046_H__ #define __NVIF_CL0046_H__ -#define NV04_DISP_NTFY_VBLANK 0x00 #define NV04_DISP_NTFY_CONN 0x01 - -struct nv04_disp_mthd_v0 { - __u8 version; -#define NV04_DISP_SCANOUTPOS 0x00 - __u8 method; - __u8 head; - __u8 pad03[5]; -}; - -struct nv04_disp_scanoutpos_v0 { - __u8 version; - __u8 pad01[7]; - __s64 time[2]; - __u16 vblanks; - __u16 vblanke; - __u16 vtotal; - __u16 vline; - __u16 hblanks; - __u16 hblanke; - __u16 htotal; - __u16 hline; -}; #endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/cl006b.h b/drivers/gpu/drm/nouveau/include/nvif/cl006b.h deleted file mode 100644 index c960c449e430..000000000000 --- a/drivers/gpu/drm/nouveau/include/nvif/cl006b.h +++ /dev/null @@ -1,12 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NVIF_CL006B_H__ -#define __NVIF_CL006B_H__ - -struct nv03_channel_dma_v0 { - __u8 version; - __u8 chid; - __u8 pad02[2]; - __u32 offset; - __u64 pushbuf; -}; -#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/cl0080.h b/drivers/gpu/drm/nouveau/include/nvif/cl0080.h index 59759c4fb62e..8b5a240d57e4 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/cl0080.h +++ b/drivers/gpu/drm/nouveau/include/nvif/cl0080.h @@ -68,7 +68,7 @@ struct nv_device_time_v0 { /* Returns the number of available runlists. */ #define NV_DEVICE_HOST_RUNLISTS NV_DEVICE_HOST(0x00000000) -/* Returns the number of available channels. */ +/* Returns the number of available channels (0 if per-runlist). */ #define NV_DEVICE_HOST_CHANNELS NV_DEVICE_HOST(0x00000001) /* Returns a mask of available engine types on runlist(data). */ @@ -90,4 +90,6 @@ struct nv_device_time_v0 { #define NV_DEVICE_HOST_RUNLIST_ENGINES_SEC2 0x00004000 #define NV_DEVICE_HOST_RUNLIST_ENGINES_NVDEC 0x00008000 #define NV_DEVICE_HOST_RUNLIST_ENGINES_NVENC 0x00010000 +/* Returns the number of available channels on runlist(data). */ +#define NV_DEVICE_HOST_RUNLIST_CHANNELS NV_DEVICE_HOST(0x00000101) #endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/cl506e.h b/drivers/gpu/drm/nouveau/include/nvif/cl506e.h deleted file mode 100644 index 9df289c7a84f..000000000000 --- a/drivers/gpu/drm/nouveau/include/nvif/cl506e.h +++ /dev/null @@ -1,13 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NVIF_CL506E_H__ -#define __NVIF_CL506E_H__ - -struct nv50_channel_dma_v0 { - __u8 version; - __u8 chid; - __u8 pad02[6]; - __u64 vmm; - __u64 pushbuf; - __u64 offset; -}; -#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/cl506f.h b/drivers/gpu/drm/nouveau/include/nvif/cl506f.h deleted file mode 100644 index 327c96a994bb..000000000000 --- a/drivers/gpu/drm/nouveau/include/nvif/cl506f.h +++ /dev/null @@ -1,14 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NVIF_CL506F_H__ -#define __NVIF_CL506F_H__ - -struct nv50_channel_gpfifo_v0 { - __u8 version; - __u8 chid; - __u8 pad02[2]; - __u32 ilength; - __u64 ioffset; - __u64 pushbuf; - __u64 vmm; -}; -#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/cl5070.h b/drivers/gpu/drm/nouveau/include/nvif/cl5070.h deleted file mode 100644 index 56affb606adf..000000000000 --- a/drivers/gpu/drm/nouveau/include/nvif/cl5070.h +++ /dev/null @@ -1,92 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NVIF_CL5070_H__ -#define __NVIF_CL5070_H__ - -#define NV50_DISP_MTHD 0x00 - -struct nv50_disp_mthd_v0 { - __u8 version; -#define NV50_DISP_SCANOUTPOS 0x00 - __u8 method; - __u8 head; - __u8 pad03[5]; -}; - -struct nv50_disp_scanoutpos_v0 { - __u8 version; - __u8 pad01[7]; - __s64 time[2]; - __u16 vblanks; - __u16 vblanke; - __u16 vtotal; - __u16 vline; - __u16 hblanks; - __u16 hblanke; - __u16 htotal; - __u16 hline; -}; - -struct nv50_disp_mthd_v1 { - __u8 version; -#define NV50_DISP_MTHD_V1_ACQUIRE 0x01 -#define NV50_DISP_MTHD_V1_RELEASE 0x02 -#define NV50_DISP_MTHD_V1_SOR_HDA_ELD 0x21 -#define NV50_DISP_MTHD_V1_SOR_HDMI_PWR 0x22 -#define NV50_DISP_MTHD_V1_SOR_LVDS_SCRIPT 0x23 -#define NV50_DISP_MTHD_V1_SOR_DP_MST_LINK 0x25 -#define NV50_DISP_MTHD_V1_SOR_DP_MST_VCPI 0x26 - __u8 method; - __u16 hasht; - __u16 hashm; - __u8 pad06[2]; -}; - -struct nv50_disp_acquire_v0 { - __u8 version; - __u8 or; - __u8 link; - __u8 hda; - __u8 pad04[4]; -}; - -struct nv50_disp_sor_hda_eld_v0 { - __u8 version; - __u8 pad01[7]; - __u8 data[]; -}; - -struct nv50_disp_sor_hdmi_pwr_v0 { - __u8 version; - __u8 state; - __u8 max_ac_packet; - __u8 rekey; - __u8 avi_infoframe_length; - __u8 vendor_infoframe_length; -#define NV50_DISP_SOR_HDMI_PWR_V0_SCDC_SCRAMBLE (1 << 0) -#define NV50_DISP_SOR_HDMI_PWR_V0_SCDC_DIV_BY_4 (1 << 1) - __u8 scdc; - __u8 pad07[1]; -}; - -struct nv50_disp_sor_lvds_script_v0 { - __u8 version; - __u8 pad01[1]; - __u16 script; - __u8 pad04[4]; -}; - -struct nv50_disp_sor_dp_mst_link_v0 { - __u8 version; - __u8 state; - __u8 pad02[6]; -}; - -struct nv50_disp_sor_dp_mst_vcpi_v0 { - __u8 version; - __u8 pad01[1]; - __u8 start_slot; - __u8 num_slots; - __u16 pbn; - __u16 aligned_pbn; -}; -#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/cl826e.h b/drivers/gpu/drm/nouveau/include/nvif/cl826e.h deleted file mode 100644 index 1b6496d31580..000000000000 --- a/drivers/gpu/drm/nouveau/include/nvif/cl826e.h +++ /dev/null @@ -1,15 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NVIF_CL826E_H__ -#define __NVIF_CL826E_H__ - -struct g82_channel_dma_v0 { - __u8 version; - __u8 chid; - __u8 pad02[6]; - __u64 vmm; - __u64 pushbuf; - __u64 offset; -}; - -#define NV826E_V0_NTFY_NON_STALL_INTERRUPT 0x00 -#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/cl826f.h b/drivers/gpu/drm/nouveau/include/nvif/cl826f.h deleted file mode 100644 index 148602264a76..000000000000 --- a/drivers/gpu/drm/nouveau/include/nvif/cl826f.h +++ /dev/null @@ -1,16 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NVIF_CL826F_H__ -#define __NVIF_CL826F_H__ - -struct g82_channel_gpfifo_v0 { - __u8 version; - __u8 chid; - __u8 pad02[2]; - __u32 ilength; - __u64 ioffset; - __u64 pushbuf; - __u64 vmm; -}; - -#define NV826F_V0_NTFY_NON_STALL_INTERRUPT 0x00 -#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/cl906f.h b/drivers/gpu/drm/nouveau/include/nvif/cl906f.h deleted file mode 100644 index 3823d6891b55..000000000000 --- a/drivers/gpu/drm/nouveau/include/nvif/cl906f.h +++ /dev/null @@ -1,16 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NVIF_CL906F_H__ -#define __NVIF_CL906F_H__ - -struct fermi_channel_gpfifo_v0 { - __u8 version; - __u8 chid; - __u8 pad02[2]; - __u32 ilength; - __u64 ioffset; - __u64 vmm; -}; - -#define NV906F_V0_NTFY_NON_STALL_INTERRUPT 0x00 -#define NV906F_V0_NTFY_KILLED 0x01 -#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/cla06f.h b/drivers/gpu/drm/nouveau/include/nvif/cla06f.h deleted file mode 100644 index cfa18f1fbf83..000000000000 --- a/drivers/gpu/drm/nouveau/include/nvif/cla06f.h +++ /dev/null @@ -1,18 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NVIF_CLA06F_H__ -#define __NVIF_CLA06F_H__ - -struct kepler_channel_gpfifo_a_v0 { - __u8 version; - __u8 priv; - __u16 chid; - __u32 ilength; - __u64 ioffset; - __u64 runlist; - __u64 vmm; - __u64 inst; -}; - -#define NVA06F_V0_NTFY_NON_STALL_INTERRUPT 0x00 -#define NVA06F_V0_NTFY_KILLED 0x01 -#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/class.h b/drivers/gpu/drm/nouveau/include/nvif/class.h index 8641db649f48..ad1e5de84e80 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/class.h +++ b/drivers/gpu/drm/nouveau/include/nvif/class.h @@ -32,11 +32,17 @@ #define NVIF_CLASS_VMM_GM200 /* ifb00d.h */ 0x8000b00d #define NVIF_CLASS_VMM_GP100 /* ifc00d.h */ 0x8000c00d +#define NVIF_CLASS_EVENT /* if000e.h */ 0x8000000e + #define NVIF_CLASS_DISP /* if0010.h */ 0x80000010 #define NVIF_CLASS_CONN /* if0011.h */ 0x80000011 #define NVIF_CLASS_OUTP /* if0012.h */ 0x80000012 +#define NVIF_CLASS_HEAD /* if0013.h */ 0x80000013 #define NVIF_CLASS_DISP_CHAN /* if0014.h */ 0x80000014 +#define NVIF_CLASS_CHAN /* if0020.h */ 0x80000020 +#define NVIF_CLASS_CGRP /* if0021.h */ 0x80000021 + /* the below match nvidia-assigned (either in hw, or sw) class numbers */ #define NV_NULL_CLASS 0x00000030 @@ -58,25 +64,30 @@ #define NV04_DISP /* cl0046.h */ 0x00000046 #define VOLTA_USERMODE_A 0x0000c361 +#define TURING_USERMODE_A 0x0000c461 +#define AMPERE_USERMODE_A 0x0000c561 #define MAXWELL_FAULT_BUFFER_A /* clb069.h */ 0x0000b069 #define VOLTA_FAULT_BUFFER_A /* clb069.h */ 0x0000c369 -#define NV03_CHANNEL_DMA /* cl506b.h */ 0x0000006b -#define NV10_CHANNEL_DMA /* cl506b.h */ 0x0000006e -#define NV17_CHANNEL_DMA /* cl506b.h */ 0x0000176e -#define NV40_CHANNEL_DMA /* cl506b.h */ 0x0000406e - -#define NV50_CHANNEL_GPFIFO /* cl506f.h */ 0x0000506f -#define G82_CHANNEL_GPFIFO /* cl826f.h */ 0x0000826f -#define FERMI_CHANNEL_GPFIFO /* cl906f.h */ 0x0000906f -#define KEPLER_CHANNEL_GPFIFO_A /* cla06f.h */ 0x0000a06f -#define KEPLER_CHANNEL_GPFIFO_B /* cla06f.h */ 0x0000a16f -#define MAXWELL_CHANNEL_GPFIFO_A /* cla06f.h */ 0x0000b06f -#define PASCAL_CHANNEL_GPFIFO_A /* cla06f.h */ 0x0000c06f -#define VOLTA_CHANNEL_GPFIFO_A /* clc36f.h */ 0x0000c36f -#define TURING_CHANNEL_GPFIFO_A /* clc36f.h */ 0x0000c46f -#define AMPERE_CHANNEL_GPFIFO_B /* clc36f.h */ 0x0000c76f +#define NV03_CHANNEL_DMA /* if0020.h */ 0x0000006b +#define NV10_CHANNEL_DMA /* if0020.h */ 0x0000006e +#define NV17_CHANNEL_DMA /* if0020.h */ 0x0000176e +#define NV40_CHANNEL_DMA /* if0020.h */ 0x0000406e + +#define KEPLER_CHANNEL_GROUP_A /* if0021.h */ 0x0000a06c + +#define NV50_CHANNEL_GPFIFO /* if0020.h */ 0x0000506f +#define G82_CHANNEL_GPFIFO /* if0020.h */ 0x0000826f +#define FERMI_CHANNEL_GPFIFO /* if0020.h */ 0x0000906f +#define KEPLER_CHANNEL_GPFIFO_A /* if0020.h */ 0x0000a06f +#define KEPLER_CHANNEL_GPFIFO_B /* if0020.h */ 0x0000a16f +#define MAXWELL_CHANNEL_GPFIFO_A /* if0020.h */ 0x0000b06f +#define PASCAL_CHANNEL_GPFIFO_A /* if0020.h */ 0x0000c06f +#define VOLTA_CHANNEL_GPFIFO_A /* if0020.h */ 0x0000c36f +#define TURING_CHANNEL_GPFIFO_A /* if0020.h */ 0x0000c46f +#define AMPERE_CHANNEL_GPFIFO_A /* if0020.h */ 0x0000c56f +#define AMPERE_CHANNEL_GPFIFO_B /* if0020.h */ 0x0000c76f #define NV50_DISP /* if0010.h */ 0x00005070 #define G82_DISP /* if0010.h */ 0x00008270 @@ -179,6 +190,8 @@ #define TURING_A /* cl9097.h */ 0x0000c597 +#define AMPERE_B /* cl9097.h */ 0x0000c797 + #define NV74_BSP 0x000074b0 #define GT212_MSVLD 0x000085b1 @@ -206,6 +219,7 @@ #define PASCAL_DMA_COPY_B 0x0000c1b5 #define VOLTA_DMA_COPY_A 0x0000c3b5 #define TURING_DMA_COPY_A 0x0000c5b5 +#define AMPERE_DMA_COPY_A 0x0000c6b5 #define AMPERE_DMA_COPY_B 0x0000c7b5 #define FERMI_DECOMPRESS 0x000090b8 @@ -222,6 +236,7 @@ #define PASCAL_COMPUTE_B 0x0000c1c0 #define VOLTA_COMPUTE_A 0x0000c3c0 #define TURING_COMPUTE_A 0x0000c5c0 +#define AMPERE_COMPUTE_B 0x0000c7c0 #define NV74_CIPHER 0x000074c1 #endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/clb069.h b/drivers/gpu/drm/nouveau/include/nvif/clb069.h index eef5d0227bab..d7689de35ab2 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/clb069.h +++ b/drivers/gpu/drm/nouveau/include/nvif/clb069.h @@ -8,5 +8,8 @@ struct nvif_clb069_v0 { __u32 put; }; -#define NVB069_V0_NTFY_FAULT 0x00 +union nvif_clb069_event_args { + struct nvif_clb069_event_vn { + } vn; +}; #endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/clc36f.h b/drivers/gpu/drm/nouveau/include/nvif/clc36f.h deleted file mode 100644 index f66885891238..000000000000 --- a/drivers/gpu/drm/nouveau/include/nvif/clc36f.h +++ /dev/null @@ -1,19 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NVIF_CLC36F_H__ -#define __NVIF_CLC36F_H__ - -struct volta_channel_gpfifo_a_v0 { - __u8 version; - __u8 priv; - __u16 chid; - __u32 ilength; - __u64 ioffset; - __u64 runlist; - __u64 vmm; - __u64 inst; - __u32 token; -}; - -#define NVC36F_V0_NTFY_NON_STALL_INTERRUPT 0x00 -#define NVC36F_V0_NTFY_KILLED 0x01 -#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/conn.h b/drivers/gpu/drm/nouveau/include/nvif/conn.h index f72a8f138f47..dc355e1dfafa 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/conn.h +++ b/drivers/gpu/drm/nouveau/include/nvif/conn.h @@ -2,6 +2,7 @@ #ifndef __NVIF_CONN_H__ #define __NVIF_CONN_H__ #include <nvif/object.h> +#include <nvif/event.h> struct nvif_disp; struct nvif_conn { @@ -11,8 +12,17 @@ struct nvif_conn { int nvif_conn_ctor(struct nvif_disp *, const char *name, int id, struct nvif_conn *); void nvif_conn_dtor(struct nvif_conn *); +static inline int +nvif_conn_id(struct nvif_conn *conn) +{ + return conn->object.handle; +} + #define NVIF_CONN_HPD_STATUS_UNSUPPORTED 0 /* negative if query fails */ #define NVIF_CONN_HPD_STATUS_NOT_PRESENT 1 #define NVIF_CONN_HPD_STATUS_PRESENT 2 int nvif_conn_hpd_status(struct nvif_conn *); + +int nvif_conn_event_ctor(struct nvif_conn *, const char *name, nvif_event_func, u8 types, + struct nvif_event *); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/disp.h b/drivers/gpu/drm/nouveau/include/nvif/disp.h index 742632ad3bea..56eb7293e01c 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/disp.h +++ b/drivers/gpu/drm/nouveau/include/nvif/disp.h @@ -7,6 +7,7 @@ struct nvif_disp { struct nvif_object object; unsigned long conn_mask; unsigned long outp_mask; + unsigned long head_mask; }; int nvif_disp_ctor(struct nvif_device *, const char *name, s32 oclass, diff --git a/drivers/gpu/drm/nouveau/include/nvif/event.h b/drivers/gpu/drm/nouveau/include/nvif/event.h index a6b1ee4f10ca..68bf6635841f 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/event.h +++ b/drivers/gpu/drm/nouveau/include/nvif/event.h @@ -1,63 +1,36 @@ /* SPDX-License-Identifier: MIT */ #ifndef __NVIF_EVENT_H__ #define __NVIF_EVENT_H__ - -struct nvif_notify_req_v0 { - __u8 version; - __u8 reply; - __u8 pad02[5]; -#define NVIF_NOTIFY_V0_ROUTE_NVIF 0x00 - __u8 route; - __u64 token; /* must be unique */ - __u8 data[]; /* request data (below) */ -}; - -struct nvif_notify_rep_v0 { - __u8 version; - __u8 pad01[6]; - __u8 route; - __u64 token; - __u8 data[]; /* reply data (below) */ -}; - -struct nvif_notify_head_req_v0 { - /* nvif_notify_req ... */ - __u8 version; - __u8 head; - __u8 pad02[6]; -}; - -struct nvif_notify_head_rep_v0 { - /* nvif_notify_rep ... */ - __u8 version; - __u8 pad01[7]; -}; - -struct nvif_notify_conn_req_v0 { - /* nvif_notify_req ... */ - __u8 version; -#define NVIF_NOTIFY_CONN_V0_PLUG 0x01 -#define NVIF_NOTIFY_CONN_V0_UNPLUG 0x02 -#define NVIF_NOTIFY_CONN_V0_IRQ 0x04 -#define NVIF_NOTIFY_CONN_V0_ANY 0x07 - __u8 mask; - __u8 conn; - __u8 pad03[5]; -}; - -struct nvif_notify_conn_rep_v0 { - /* nvif_notify_rep ... */ - __u8 version; - __u8 mask; - __u8 pad02[6]; -}; - -struct nvif_notify_uevent_req { - /* nvif_notify_req ... */ -}; - -struct nvif_notify_uevent_rep { - /* nvif_notify_rep ... */ -}; - +#include <nvif/object.h> +#include <nvif/if000e.h> +struct nvif_event; + +#define NVIF_EVENT_KEEP 0 +#define NVIF_EVENT_DROP 1 +typedef int (*nvif_event_func)(struct nvif_event *, void *repv, u32 repc); + +struct nvif_event { + struct nvif_object object; + nvif_event_func func; +}; + +static inline bool +nvif_event_constructed(struct nvif_event *event) +{ + return nvif_object_constructed(&event->object); +} + +int nvif_event_ctor_(struct nvif_object *, const char *, u32, nvif_event_func, bool, + struct nvif_event_v0 *, u32, bool, struct nvif_event *); + +static inline int +nvif_event_ctor(struct nvif_object *parent, const char *name, u32 handle, nvif_event_func func, + bool wait, struct nvif_event_v0 *args, u32 argc, struct nvif_event *event) +{ + return nvif_event_ctor_(parent, name, handle, func, wait, args, argc, true, event); +} + +void nvif_event_dtor(struct nvif_event *); +int nvif_event_allow(struct nvif_event *); +int nvif_event_block(struct nvif_event *); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/head.h b/drivers/gpu/drm/nouveau/include/nvif/head.h new file mode 100644 index 000000000000..3ec36999e956 --- /dev/null +++ b/drivers/gpu/drm/nouveau/include/nvif/head.h @@ -0,0 +1,23 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __NVIF_HEAD_H__ +#define __NVIF_HEAD_H__ +#include <nvif/object.h> +#include <nvif/event.h> +struct nvif_disp; + +struct nvif_head { + struct nvif_object object; +}; + +int nvif_head_ctor(struct nvif_disp *, const char *name, int id, struct nvif_head *); +void nvif_head_dtor(struct nvif_head *); + +static inline int +nvif_head_id(struct nvif_head *head) +{ + return head->object.handle; +} + +int nvif_head_vblank_event_ctor(struct nvif_head *, const char *name, nvif_event_func, bool wait, + struct nvif_event *); +#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/if0004.h b/drivers/gpu/drm/nouveau/include/nvif/if0004.h index d324c73c27fb..1d916a137941 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/if0004.h +++ b/drivers/gpu/drm/nouveau/include/nvif/if0004.h @@ -2,7 +2,10 @@ #ifndef __NVIF_IF0004_H__ #define __NVIF_IF0004_H__ -#define NV04_NVSW_NTFY_UEVENT 0x00 +union nv04_nvsw_event_args { + struct nv04_nvsw_event_vn { + } vn; +}; #define NV04_NVSW_GET_REF 0x00 diff --git a/drivers/gpu/drm/nouveau/include/nvif/if000e.h b/drivers/gpu/drm/nouveau/include/nvif/if000e.h new file mode 100644 index 000000000000..90a936cb1766 --- /dev/null +++ b/drivers/gpu/drm/nouveau/include/nvif/if000e.h @@ -0,0 +1,26 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __NVIF_IF000E_H__ +#define __NVIF_IF000E_H__ + +union nvif_event_args { + struct nvif_event_v0 { + __u8 version; + __u8 wait; + __u8 pad02[6]; + __u8 data[]; + } v0; +}; + +#define NVIF_EVENT_V0_ALLOW 0x00 +#define NVIF_EVENT_V0_BLOCK 0x01 + +union nvif_event_allow_args { + struct nvif_event_allow_vn { + } vn; +}; + +union nvif_event_block_args { + struct nvif_event_block_vn { + } vn; +}; +#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/if0010.h b/drivers/gpu/drm/nouveau/include/nvif/if0010.h index fc236ef28965..4c835bbe6fe3 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/if0010.h +++ b/drivers/gpu/drm/nouveau/include/nvif/if0010.h @@ -8,6 +8,7 @@ union nvif_disp_args { __u8 pad01[3]; __u32 conn_mask; __u32 outp_mask; + __u32 head_mask; } v0; }; #endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/if0011.h b/drivers/gpu/drm/nouveau/include/nvif/if0011.h index 04ba6581f840..69b0b779f942 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/if0011.h +++ b/drivers/gpu/drm/nouveau/include/nvif/if0011.h @@ -10,6 +10,17 @@ union nvif_conn_args { } v0; }; +union nvif_conn_event_args { + struct nvif_conn_event_v0 { + __u8 version; +#define NVIF_CONN_EVENT_V0_PLUG 0x01 +#define NVIF_CONN_EVENT_V0_UNPLUG 0x02 +#define NVIF_CONN_EVENT_V0_IRQ 0x04 + __u8 types; + __u8 pad02[6]; + } v0; +}; + #define NVIF_CONN_V0_HPD_STATUS 0x00000000 union nvif_conn_hpd_status_args { diff --git a/drivers/gpu/drm/nouveau/include/nvif/if0012.h b/drivers/gpu/drm/nouveau/include/nvif/if0012.h index 243bd35d942f..eb99d84eb844 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/if0012.h +++ b/drivers/gpu/drm/nouveau/include/nvif/if0012.h @@ -11,6 +11,13 @@ union nvif_outp_args { }; #define NVIF_OUTP_V0_LOAD_DETECT 0x00 +#define NVIF_OUTP_V0_ACQUIRE 0x01 +#define NVIF_OUTP_V0_RELEASE 0x02 +#define NVIF_OUTP_V0_INFOFRAME 0x03 +#define NVIF_OUTP_V0_HDA_ELD 0x04 +#define NVIF_OUTP_V0_DP_AUX_PWR 0x05 +#define NVIF_OUTP_V0_DP_RETRAIN 0x06 +#define NVIF_OUTP_V0_DP_MST_VCPI 0x07 union nvif_outp_load_detect_args { struct nvif_outp_load_detect_v0 { @@ -20,4 +27,95 @@ union nvif_outp_load_detect_args { __u32 data; /*TODO: move vbios loadval parsing into nvkm */ } v0; }; + +union nvif_outp_acquire_args { + struct nvif_outp_acquire_v0 { + __u8 version; +#define NVIF_OUTP_ACQUIRE_V0_RGB_CRT 0x00 +#define NVIF_OUTP_ACQUIRE_V0_TV 0x01 +#define NVIF_OUTP_ACQUIRE_V0_TMDS 0x02 +#define NVIF_OUTP_ACQUIRE_V0_LVDS 0x03 +#define NVIF_OUTP_ACQUIRE_V0_DP 0x04 + __u8 proto; + __u8 or; + __u8 link; + __u8 pad04[4]; + union { + struct { + __u8 head; + __u8 hdmi; + __u8 hdmi_max_ac_packet; + __u8 hdmi_rekey; +#define NVIF_OUTP_ACQUIRE_V0_TMDS_HDMI_SCDC_SCRAMBLE (1 << 0) +#define NVIF_OUTP_ACQUIRE_V0_TMDS_HDMI_SCDC_DIV_BY_4 (1 << 1) + __u8 hdmi_scdc; + __u8 hdmi_hda; + __u8 pad06[2]; + } tmds; + struct { + __u8 dual; + __u8 bpc8; + __u8 pad02[6]; + } lvds; + struct { + __u8 link_nr; /* 0 = highest possible. */ + __u8 link_bw; /* 0 = highest possible, DP BW code otherwise. */ + __u8 hda; + __u8 mst; + __u8 pad04[4]; + __u8 dpcd[16]; + } dp; + }; + } v0; +}; + +union nvif_outp_release_args { + struct nvif_outp_release_vn { + } vn; +}; + +union nvif_outp_infoframe_args { + struct nvif_outp_infoframe_v0 { + __u8 version; +#define NVIF_OUTP_INFOFRAME_V0_AVI 0 +#define NVIF_OUTP_INFOFRAME_V0_VSI 1 + __u8 type; + __u8 head; + __u8 pad03[5]; + __u8 data[]; + } v0; +}; + +union nvif_outp_hda_eld_args { + struct nvif_outp_hda_eld_v0 { + __u8 version; + __u8 head; + __u8 pad02[6]; + __u8 data[]; + } v0; +}; + +union nvif_outp_dp_aux_pwr_args { + struct nvif_outp_dp_aux_pwr_v0 { + __u8 version; + __u8 state; + __u8 pad02[6]; + } v0; +}; + +union nvif_outp_dp_retrain_args { + struct nvif_outp_dp_retrain_vn { + } vn; +}; + +union nvif_outp_dp_mst_vcpi_args { + struct nvif_outp_dp_mst_vcpi_v0 { + __u8 version; + __u8 head; + __u8 start_slot; + __u8 num_slots; + __u16 pbn; + __u16 aligned_pbn; + } v0; +}; #endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/if0013.h b/drivers/gpu/drm/nouveau/include/nvif/if0013.h new file mode 100644 index 000000000000..6756c7467ae4 --- /dev/null +++ b/drivers/gpu/drm/nouveau/include/nvif/if0013.h @@ -0,0 +1,35 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __NVIF_IF0013_H__ +#define __NVIF_IF0013_H__ + +union nvif_head_args { + struct nvif_head_v0 { + __u8 version; + __u8 id; + __u8 pad02[6]; + } v0; +}; + +union nvif_head_event_args { + struct nvif_head_event_vn { + } vn; +}; + +#define NVIF_HEAD_V0_SCANOUTPOS 0x00 + +union nvif_head_scanoutpos_args { + struct nvif_head_scanoutpos_v0 { + __u8 version; + __u8 pad01[7]; + __s64 time[2]; + __u16 vblanks; + __u16 vblanke; + __u16 vtotal; + __u16 vline; + __u16 hblanks; + __u16 hblanke; + __u16 htotal; + __u16 hline; + } v0; +}; +#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/if0020.h b/drivers/gpu/drm/nouveau/include/nvif/if0020.h new file mode 100644 index 000000000000..085e0ae8a450 --- /dev/null +++ b/drivers/gpu/drm/nouveau/include/nvif/if0020.h @@ -0,0 +1,45 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __NVIF_IF0020_H__ +#define __NVIF_IF0020_H__ + +union nvif_chan_args { + struct nvif_chan_v0 { + __u8 version; + __u8 namelen; + __u8 runlist; + __u8 runq; + __u8 priv; + __u8 pad05; + __u16 devm; + __u64 vmm; + + __u64 ctxdma; + __u64 offset; + __u64 length; + + __u64 huserd; + __u64 ouserd; + + __u32 token; + __u16 chid; + __u8 pad3e; +#define NVIF_CHAN_V0_INST_APER_VRAM 0 +#define NVIF_CHAN_V0_INST_APER_HOST 1 +#define NVIF_CHAN_V0_INST_APER_NCOH 2 +#define NVIF_CHAN_V0_INST_APER_INST 0xff + __u8 aper; + __u64 inst; + + __u8 name[]; + } v0; +}; + +union nvif_chan_event_args { + struct nvif_chan_event_v0 { + __u8 version; +#define NVIF_CHAN_EVENT_V0_NON_STALL_INTR 0x00 +#define NVIF_CHAN_EVENT_V0_KILLED 0x01 + __u8 type; + } v0; +}; +#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/if0021.h b/drivers/gpu/drm/nouveau/include/nvif/if0021.h new file mode 100644 index 000000000000..5013def90455 --- /dev/null +++ b/drivers/gpu/drm/nouveau/include/nvif/if0021.h @@ -0,0 +1,16 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __NVIF_IF0021_H__ +#define __NVIF_IF0021_H__ + +union nvif_cgrp_args { + struct nvif_cgrp_v0 { + __u8 version; + __u8 namelen; + __u8 runlist; + __u8 pad03[3]; + __u16 cgid; + __u64 vmm; + __u8 name[]; + } v0; +}; +#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/ioctl.h b/drivers/gpu/drm/nouveau/include/nvif/ioctl.h index 886c63fe753f..4e047bb1fc07 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/ioctl.h +++ b/drivers/gpu/drm/nouveau/include/nvif/ioctl.h @@ -15,10 +15,6 @@ struct nvif_ioctl_v0 { #define NVIF_IOCTL_V0_WR 0x06 #define NVIF_IOCTL_V0_MAP 0x07 #define NVIF_IOCTL_V0_UNMAP 0x08 -#define NVIF_IOCTL_V0_NTFY_NEW 0x09 -#define NVIF_IOCTL_V0_NTFY_DEL 0x0a -#define NVIF_IOCTL_V0_NTFY_GET 0x0b -#define NVIF_IOCTL_V0_NTFY_PUT 0x0c __u8 type; __u8 pad02[4]; #define NVIF_IOCTL_V0_OWNER_NVIF 0x00 @@ -63,6 +59,14 @@ struct nvif_ioctl_new_v0 { struct nvif_ioctl_del { }; +struct nvif_ioctl_mthd_v0 { + /* nvif_ioctl ... */ + __u8 version; + __u8 method; + __u8 pad02[6]; + __u8 data[]; /* method data (class.h) */ +}; + struct nvif_ioctl_rd_v0 { /* nvif_ioctl ... */ __u8 version; @@ -95,43 +99,4 @@ struct nvif_ioctl_map_v0 { struct nvif_ioctl_unmap { }; - -struct nvif_ioctl_ntfy_new_v0 { - /* nvif_ioctl ... */ - __u8 version; - __u8 event; - __u8 index; - __u8 pad03[5]; - __u8 data[]; /* event request data (event.h) */ -}; - -struct nvif_ioctl_ntfy_del_v0 { - /* nvif_ioctl ... */ - __u8 version; - __u8 index; - __u8 pad02[6]; -}; - -struct nvif_ioctl_ntfy_get_v0 { - /* nvif_ioctl ... */ - __u8 version; - __u8 index; - __u8 pad02[6]; -}; - -struct nvif_ioctl_ntfy_put_v0 { - /* nvif_ioctl ... */ - __u8 version; - __u8 index; - __u8 pad02[6]; -}; - -struct nvif_ioctl_mthd_v0 { - /* nvif_ioctl ... */ - __u8 version; - __u8 method; - __u8 pad02[6]; - __u8 data[]; /* method data (class.h) */ -}; - #endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/notify.h b/drivers/gpu/drm/nouveau/include/nvif/notify.h deleted file mode 100644 index 39f6b7ee1719..000000000000 --- a/drivers/gpu/drm/nouveau/include/nvif/notify.h +++ /dev/null @@ -1,35 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NVIF_NOTIFY_H__ -#define __NVIF_NOTIFY_H__ - -struct nvif_notify { - struct nvif_object *object; - const char *name; - int index; - -#define NVIF_NOTIFY_USER 0 -#define NVIF_NOTIFY_WORK 1 - unsigned long flags; - atomic_t putcnt; - void (*dtor)(struct nvif_notify *); -#define NVIF_NOTIFY_DROP 0 -#define NVIF_NOTIFY_KEEP 1 - int (*func)(struct nvif_notify *); - - /* this is const for a *very* good reason - the data might be on the - * stack from an irq handler. if you're not nvif/notify.c then you - * should probably think twice before casting it away... - */ - const void *data; - u32 size; - struct work_struct work; -}; - -int nvif_notify_ctor(struct nvif_object *, const char *name, - int (*func)(struct nvif_notify *), bool work, u8 type, - void *data, u32 size, u32 reply, struct nvif_notify *); -int nvif_notify_dtor(struct nvif_notify *); -int nvif_notify_get(struct nvif_notify *); -int nvif_notify_put(struct nvif_notify *); -int nvif_notify(const void *, u32, const void *, u32); -#endif diff --git a/drivers/gpu/drm/nouveau/include/nvif/outp.h b/drivers/gpu/drm/nouveau/include/nvif/outp.h index 0d6aa07a9184..45daadec3c0c 100644 --- a/drivers/gpu/drm/nouveau/include/nvif/outp.h +++ b/drivers/gpu/drm/nouveau/include/nvif/outp.h @@ -2,13 +2,32 @@ #ifndef __NVIF_OUTP_H__ #define __NVIF_OUTP_H__ #include <nvif/object.h> +#include <nvif/if0012.h> struct nvif_disp; struct nvif_outp { struct nvif_object object; + + struct { + int id; + int link; + } or; }; int nvif_outp_ctor(struct nvif_disp *, const char *name, int id, struct nvif_outp *); void nvif_outp_dtor(struct nvif_outp *); int nvif_outp_load_detect(struct nvif_outp *, u32 loadval); +int nvif_outp_acquire_rgb_crt(struct nvif_outp *); +int nvif_outp_acquire_tmds(struct nvif_outp *, int head, + bool hdmi, u8 max_ac_packet, u8 rekey, u8 scdc, bool hda); +int nvif_outp_acquire_lvds(struct nvif_outp *, bool dual, bool bpc8); +int nvif_outp_acquire_dp(struct nvif_outp *, u8 dpcd[16], + int link_nr, int link_bw, bool hda, bool mst); +void nvif_outp_release(struct nvif_outp *); +int nvif_outp_infoframe(struct nvif_outp *, u8 type, struct nvif_outp_infoframe_v0 *, u32 size); +int nvif_outp_hda_eld(struct nvif_outp *, int head, void *data, u32 size); +int nvif_outp_dp_aux_pwr(struct nvif_outp *, bool enable); +int nvif_outp_dp_retrain(struct nvif_outp *); +int nvif_outp_dp_mst_vcpi(struct nvif_outp *, int head, + u8 start_slot, u8 num_slots, u16 pbn, u16 aligned_pbn); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/client.h b/drivers/gpu/drm/nouveau/include/nvkm/core/client.h index 2f86606e708c..0d9fc741a719 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/client.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/client.h @@ -10,28 +10,19 @@ struct nvkm_client { u64 device; u32 debug; - struct nvkm_client_notify *notify[32]; struct rb_root objroot; void *data; - int (*ntfy)(const void *, u32, const void *, u32); + int (*event)(u64 token, void *argv, u32 argc); struct list_head umem; spinlock_t lock; }; -int nvkm_client_new(const char *name, u64 device, const char *cfg, - const char *dbg, - int (*)(const void *, u32, const void *, u32), - struct nvkm_client **); +int nvkm_client_new(const char *name, u64 device, const char *cfg, const char *dbg, + int (*)(u64, void *, u32), struct nvkm_client **); struct nvkm_client *nvkm_client_search(struct nvkm_client *, u64 handle); -int nvkm_client_notify_new(struct nvkm_object *, struct nvkm_event *, - void *data, u32 size); -int nvkm_client_notify_del(struct nvkm_client *, int index); -int nvkm_client_notify_get(struct nvkm_client *, int index); -int nvkm_client_notify_put(struct nvkm_client *, int index); - /* logging for client-facing objects */ #define nvif_printk(o,l,p,f,a...) do { \ const struct nvkm_object *_object = (o); \ diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/device.h b/drivers/gpu/drm/nouveau/include/nvkm/core/device.h index efede1f11e1d..f65b5009acf7 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/device.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/device.h @@ -2,6 +2,7 @@ #ifndef __NVKM_DEVICE_H__ #define __NVKM_DEVICE_H__ #include <core/oclass.h> +#include <core/intr.h> enum nvkm_subdev_type; enum nvkm_device_type { @@ -60,6 +61,16 @@ struct nvkm_device { #undef NVKM_LAYOUT_INST #undef NVKM_LAYOUT_ONCE struct list_head subdev; + + struct { + struct list_head intr; + struct list_head prio[NVKM_INTR_PRIO_NR]; + spinlock_t lock; + int irq; + bool alloc; + bool armed; + bool legacy_done; + } intr; }; struct nvkm_subdev *nvkm_device_subdev(struct nvkm_device *, int type, int inst); @@ -72,6 +83,7 @@ struct nvkm_device_func { int (*preinit)(struct nvkm_device *); int (*init)(struct nvkm_device *); void (*fini)(struct nvkm_device *, bool suspend); + int (*irq)(struct nvkm_device *); resource_size_t (*resource_addr)(struct nvkm_device *, unsigned bar); resource_size_t (*resource_size)(struct nvkm_device *, unsigned bar); bool cpu_coherent; diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/engine.h b/drivers/gpu/drm/nouveau/include/nvkm/core/engine.h index e58923b67d74..b67b9c1a6b4e 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/engine.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/engine.h @@ -12,12 +12,6 @@ struct nvkm_engine { const struct nvkm_engine_func *func; struct nvkm_subdev subdev; spinlock_t lock; - - struct { - refcount_t refcount; - struct mutex mutex; - bool enabled; - } use; }; struct nvkm_engine_func { @@ -27,6 +21,7 @@ struct nvkm_engine_func { int (*info)(struct nvkm_engine *, u64 mthd, u64 *data); int (*init)(struct nvkm_engine *); int (*fini)(struct nvkm_engine *, bool suspend); + int (*reset)(struct nvkm_engine *); void (*intr)(struct nvkm_engine *); void (*tile)(struct nvkm_engine *, int region, struct nvkm_fb_tile *); bool (*chsw_load)(struct nvkm_engine *); @@ -54,6 +49,7 @@ int nvkm_engine_new_(const struct nvkm_engine_func *, struct nvkm_device *, struct nvkm_engine *nvkm_engine_ref(struct nvkm_engine *); void nvkm_engine_unref(struct nvkm_engine **); +int nvkm_engine_reset(struct nvkm_engine *); void nvkm_engine_tile(struct nvkm_engine *, int region); bool nvkm_engine_chsw_load(struct nvkm_engine *); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/event.h b/drivers/gpu/drm/nouveau/include/nvkm/core/event.h index a7a413f07a78..82b267c11147 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/event.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/event.h @@ -2,34 +2,76 @@ #ifndef __NVKM_EVENT_H__ #define __NVKM_EVENT_H__ #include <core/os.h> -struct nvkm_notify; struct nvkm_object; +struct nvkm_oclass; +struct nvkm_uevent; struct nvkm_event { const struct nvkm_event_func *func; + struct nvkm_subdev *subdev; int types_nr; int index_nr; spinlock_t refs_lock; spinlock_t list_lock; - struct list_head list; int *refs; + + struct list_head ntfy; }; struct nvkm_event_func { - int (*ctor)(struct nvkm_object *, void *data, u32 size, - struct nvkm_notify *); - void (*send)(void *data, u32 size, struct nvkm_notify *); void (*init)(struct nvkm_event *, int type, int index); void (*fini)(struct nvkm_event *, int type, int index); }; -int nvkm_event_init(const struct nvkm_event_func *func, int types_nr, - int index_nr, struct nvkm_event *); +int __nvkm_event_init(const struct nvkm_event_func *func, struct nvkm_subdev *, int types_nr, + int index_nr, struct nvkm_event *); + +/* Each nvkm_event needs its own lockdep class due to inter-dependencies, to + * prevent lockdep false-positives. + * + * Inlining the spinlock initialisation ensures each is unique. + */ +static __always_inline int +nvkm_event_init(const struct nvkm_event_func *func, struct nvkm_subdev *subdev, + int types_nr, int index_nr, struct nvkm_event *event) +{ + spin_lock_init(&event->refs_lock); + spin_lock_init(&event->list_lock); + return __nvkm_event_init(func, subdev, types_nr, index_nr, event); +} + void nvkm_event_fini(struct nvkm_event *); -void nvkm_event_get(struct nvkm_event *, u32 types, int index); -void nvkm_event_put(struct nvkm_event *, u32 types, int index); -void nvkm_event_send(struct nvkm_event *, u32 types, int index, - void *data, u32 size); + +#define NVKM_EVENT_KEEP 0 +#define NVKM_EVENT_DROP 1 +struct nvkm_event_ntfy; +typedef int (*nvkm_event_func)(struct nvkm_event_ntfy *, u32 bits); + +struct nvkm_event_ntfy { + struct nvkm_event *event; + int id; + u32 bits; + bool wait; + nvkm_event_func func; + + atomic_t allowed; + bool running; + + struct list_head head; +}; + +void nvkm_event_ntfy(struct nvkm_event *, int id, u32 bits); +bool nvkm_event_ntfy_valid(struct nvkm_event *, int id, u32 bits); +void nvkm_event_ntfy_add(struct nvkm_event *, int id, u32 bits, bool wait, nvkm_event_func, + struct nvkm_event_ntfy *); +void nvkm_event_ntfy_del(struct nvkm_event_ntfy *); +void nvkm_event_ntfy_allow(struct nvkm_event_ntfy *); +void nvkm_event_ntfy_block(struct nvkm_event_ntfy *); + +typedef int (*nvkm_uevent_func)(struct nvkm_object *, u64 token, u32 bits); + +int nvkm_uevent_new(const struct nvkm_oclass *, void *argv, u32 argc, struct nvkm_object **); +int nvkm_uevent_add(struct nvkm_uevent *, struct nvkm_event *, int id, u32 bits, nvkm_uevent_func); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/falcon.h b/drivers/gpu/drm/nouveau/include/nvkm/core/falcon.h index fd9a3f9a518e..b857cf142c4a 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/falcon.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/falcon.h @@ -1,34 +1,166 @@ #ifndef __NVKM_FALCON_H__ #define __NVKM_FALCON_H__ +#include <core/firmware.h> #include <engine/falcon.h> +enum nvkm_falcon_mem { + IMEM, + DMEM, + EMEM, +}; + +static inline const char * +nvkm_falcon_mem(enum nvkm_falcon_mem mem) +{ + switch (mem) { + case IMEM: return "imem"; + case DMEM: return "dmem"; + case EMEM: return "emem"; + default: + WARN_ON(1); + return "?mem"; + } +} + +struct nvkm_falcon_func_pio { + int min; + int max; + void (*wr_init)(struct nvkm_falcon *, u8 port, bool sec, u32 mem_base); + void (*wr)(struct nvkm_falcon *, u8 port, const u8 *img, int len, u16 tag); + void (*rd_init)(struct nvkm_falcon *, u8 port, u32 mem_base); + void (*rd)(struct nvkm_falcon *, u8 port, const u8 *img, int len); +}; + +struct nvkm_falcon_func_dma { + int (*init)(struct nvkm_falcon *, u64 dma_addr, int xfer_len, + enum nvkm_falcon_mem, bool sec, u32 *cmd); + void (*xfer)(struct nvkm_falcon *, u32 mem_base, u32 dma_base, u32 cmd); + bool (*done)(struct nvkm_falcon *); +}; + int nvkm_falcon_ctor(const struct nvkm_falcon_func *, struct nvkm_subdev *owner, const char *name, u32 addr, struct nvkm_falcon *); void nvkm_falcon_dtor(struct nvkm_falcon *); +int nvkm_falcon_reset(struct nvkm_falcon *); +int nvkm_falcon_pio_wr(struct nvkm_falcon *, const u8 *img, u32 img_base, u8 port, + enum nvkm_falcon_mem mem_type, u32 mem_base, int len, u16 tag, bool sec); +int nvkm_falcon_pio_rd(struct nvkm_falcon *, u8 port, enum nvkm_falcon_mem type, u32 mem_base, + const u8 *img, u32 img_base, int len); +int nvkm_falcon_dma_wr(struct nvkm_falcon *, const u8 *img, u64 dma_addr, u32 dma_base, + enum nvkm_falcon_mem mem_type, u32 mem_base, int len, bool sec); + +int gm200_flcn_reset_wait_mem_scrubbing(struct nvkm_falcon *); +int gm200_flcn_disable(struct nvkm_falcon *); +int gm200_flcn_enable(struct nvkm_falcon *); +void gm200_flcn_bind_inst(struct nvkm_falcon *, int, u64); +int gm200_flcn_bind_stat(struct nvkm_falcon *, bool); +extern const struct nvkm_falcon_func_pio gm200_flcn_imem_pio; +extern const struct nvkm_falcon_func_pio gm200_flcn_dmem_pio; +void gm200_flcn_tracepc(struct nvkm_falcon *); + +int gp102_flcn_reset_eng(struct nvkm_falcon *); +extern const struct nvkm_falcon_func_pio gp102_flcn_emem_pio; + +int ga102_flcn_select(struct nvkm_falcon *); +int ga102_flcn_reset_prep(struct nvkm_falcon *); +int ga102_flcn_reset_wait_mem_scrubbing(struct nvkm_falcon *); +extern const struct nvkm_falcon_func_dma ga102_flcn_dma; void nvkm_falcon_v1_load_imem(struct nvkm_falcon *, void *, u32, u32, u16, u8, bool); void nvkm_falcon_v1_load_dmem(struct nvkm_falcon *, void *, u32, u32, u8); -void nvkm_falcon_v1_read_dmem(struct nvkm_falcon *, u32, u32, u8, void *); -void nvkm_falcon_v1_bind_context(struct nvkm_falcon *, struct nvkm_memory *); -int nvkm_falcon_v1_wait_for_halt(struct nvkm_falcon *, u32); -int nvkm_falcon_v1_clear_interrupt(struct nvkm_falcon *, u32); -void nvkm_falcon_v1_set_start_addr(struct nvkm_falcon *, u32 start_addr); void nvkm_falcon_v1_start(struct nvkm_falcon *); -int nvkm_falcon_v1_enable(struct nvkm_falcon *); -void nvkm_falcon_v1_disable(struct nvkm_falcon *); -void gp102_sec2_flcn_bind_context(struct nvkm_falcon *, struct nvkm_memory *); -int gp102_sec2_flcn_enable(struct nvkm_falcon *); +#define FLCN_PRINTK(f,l,p,fmt,a...) ({ \ + if ((f)->owner->name != (f)->name) \ + nvkm_printk___((f)->owner, (f)->user, NV_DBG_##l, p, "%s:"fmt, (f)->name, ##a); \ + else \ + nvkm_printk___((f)->owner, (f)->user, NV_DBG_##l, p, fmt, ##a); \ +}) +#define FLCN_DBG(f,fmt,a...) FLCN_PRINTK((f), DEBUG, info, " "fmt"\n", ##a) +#define FLCN_ERR(f,fmt,a...) FLCN_PRINTK((f), ERROR, err, " "fmt"\n", ##a) +#define FLCN_ERRON(f,c,fmt,a...) \ + ({ bool _cond = (c); _cond ? FLCN_ERR(f, fmt, ##a) : FLCN_DBG(f, fmt, ##a); _cond; }) + + +struct nvkm_falcon_fw { + const struct nvkm_falcon_fw_func { + int (*signature)(struct nvkm_falcon_fw *, u32 *sig_base_src); + int (*reset)(struct nvkm_falcon_fw *); + int (*setup)(struct nvkm_falcon_fw *); + int (*load)(struct nvkm_falcon_fw *); + int (*load_bld)(struct nvkm_falcon_fw *); + int (*boot)(struct nvkm_falcon_fw *, + u32 *mbox0, u32 *mbox1, u32 mbox0_ok, u32 irqsclr); + } *func; + struct nvkm_firmware fw; + + u32 sig_base_prd; + u32 sig_base_dbg; + u32 sig_base_img; + u32 sig_size; + int sig_nr; + u8 *sigs; + u32 fuse_ver; + u32 engine_id; + u32 ucode_id; + + u32 nmem_base_img; + u32 nmem_base; + u32 nmem_size; + + u32 imem_base_img; + u32 imem_base; + u32 imem_size; + + u32 dmem_base_img; + u32 dmem_base; + u32 dmem_size; + u32 dmem_sign; + + u8 *boot; + u32 boot_size; + u32 boot_addr; + + struct nvkm_falcon *falcon; + struct nvkm_memory *inst; + struct nvkm_vmm *vmm; + struct nvkm_vma *vma; +}; + +int nvkm_falcon_fw_ctor(const struct nvkm_falcon_fw_func *, const char *name, struct nvkm_device *, + bool bl, const void *src, u32 len, struct nvkm_falcon *, + struct nvkm_falcon_fw *); +int nvkm_falcon_fw_ctor_hs(const struct nvkm_falcon_fw_func *, const char *name, + struct nvkm_subdev *, const char *bl, const char *img, int ver, + struct nvkm_falcon *falcon, struct nvkm_falcon_fw *fw); +int nvkm_falcon_fw_ctor_hs_v2(const struct nvkm_falcon_fw_func *, const char *name, + struct nvkm_subdev *, const char *img, int ver, struct nvkm_falcon *, + struct nvkm_falcon_fw *); +int nvkm_falcon_fw_sign(struct nvkm_falcon_fw *, u32 sig_base_img, u32 sig_size, const u8 *sigs, + int sig_nr_prd, u32 sig_base_prd, int sig_nr_dbg, u32 sig_base_dbg); +int nvkm_falcon_fw_patch(struct nvkm_falcon_fw *); +void nvkm_falcon_fw_dtor(struct nvkm_falcon_fw *); +int nvkm_falcon_fw_oneinit(struct nvkm_falcon_fw *, struct nvkm_falcon *, struct nvkm_vmm *, + struct nvkm_memory *inst); +int nvkm_falcon_fw_boot(struct nvkm_falcon_fw *, struct nvkm_subdev *user, + bool release, u32 *pmbox0, u32 *pmbox1, u32 mbox0_ok, u32 irqsclr); + +extern const struct nvkm_falcon_fw_func gm200_flcn_fw; +int gm200_flcn_fw_signature(struct nvkm_falcon_fw *, u32 *); +int gm200_flcn_fw_reset(struct nvkm_falcon_fw *); +int gm200_flcn_fw_load(struct nvkm_falcon_fw *); +int gm200_flcn_fw_boot(struct nvkm_falcon_fw *, u32 *, u32 *, u32, u32); + +int ga100_flcn_fw_signature(struct nvkm_falcon_fw *, u32 *); + +extern const struct nvkm_falcon_fw_func ga102_flcn_fw; +int ga102_flcn_fw_load(struct nvkm_falcon_fw *); +int ga102_flcn_fw_boot(struct nvkm_falcon_fw *, u32 *, u32 *, u32, u32); -#define FLCN_PRINTK(t,f,fmt,a...) do { \ - if ((f)->owner->name != (f)->name) \ - nvkm_##t((f)->owner, "%s: "fmt"\n", (f)->name, ##a); \ - else \ - nvkm_##t((f)->owner, fmt"\n", ##a); \ -} while(0) -#define FLCN_DBG(f,fmt,a...) FLCN_PRINTK(debug, (f), fmt, ##a) -#define FLCN_ERR(f,fmt,a...) FLCN_PRINTK(error, (f), fmt, ##a) +#define FLCNFW_PRINTK(f,l,p,fmt,a...) FLCN_PRINTK((f)->falcon, l, p, "%s: "fmt, (f)->fw.name, ##a) +#define FLCNFW_DBG(f,fmt,a...) FLCNFW_PRINTK((f), DEBUG, info, fmt"\n", ##a) +#define FLCNFW_ERR(f,fmt,a...) FLCNFW_PRINTK((f), ERROR, err, fmt"\n", ##a) /** * struct nvfw_falcon_msg - header for all messages @@ -72,6 +204,7 @@ int nvkm_falcon_msgq_new(struct nvkm_falcon_qmgr *, const char *name, void nvkm_falcon_msgq_del(struct nvkm_falcon_msgq **); void nvkm_falcon_msgq_init(struct nvkm_falcon_msgq *, u32 index, u32 offset, u32 size); +bool nvkm_falcon_msgq_empty(struct nvkm_falcon_msgq *); int nvkm_falcon_msgq_recv_initmsg(struct nvkm_falcon_msgq *, void *, u32 size); void nvkm_falcon_msgq_recv(struct nvkm_falcon_msgq *); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/firmware.h b/drivers/gpu/drm/nouveau/include/nvkm/core/firmware.h index 85bcb80f6873..d4e507e252b1 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/firmware.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/firmware.h @@ -1,9 +1,34 @@ /* SPDX-License-Identifier: MIT */ #ifndef __NVKM_FIRMWARE_H__ #define __NVKM_FIRMWARE_H__ +#include <core/memory.h> #include <core/option.h> #include <core/subdev.h> +struct nvkm_firmware { + const struct nvkm_firmware_func { + enum nvkm_firmware_type { + NVKM_FIRMWARE_IMG_RAM, + NVKM_FIRMWARE_IMG_DMA, + } type; + } *func; + const char *name; + struct nvkm_device *device; + + int len; + u8 *img; + u64 phys; + + struct nvkm_firmware_mem { + struct nvkm_memory memory; + struct scatterlist sgl; + } mem; +}; + +int nvkm_firmware_ctor(const struct nvkm_firmware_func *, const char *name, struct nvkm_device *, + const void *ptr, int len, struct nvkm_firmware *); +void nvkm_firmware_dtor(struct nvkm_firmware *); + int nvkm_firmware_get(const struct nvkm_subdev *, const char *fwname, int ver, const struct firmware **); void nvkm_firmware_put(const struct firmware *); diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/intr.h b/drivers/gpu/drm/nouveau/include/nvkm/core/intr.h new file mode 100644 index 000000000000..a003d6a544b0 --- /dev/null +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/intr.h @@ -0,0 +1,73 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __NVKM_INTR_H__ +#define __NVKM_INTR_H__ +#include <core/os.h> +struct nvkm_device; +struct nvkm_subdev; + +enum nvkm_intr_prio { + NVKM_INTR_PRIO_VBLANK = 0, + NVKM_INTR_PRIO_NORMAL, + NVKM_INTR_PRIO_NR +}; + +enum nvkm_intr_type { + NVKM_INTR_SUBDEV = -1, /* lookup vector by requesting subdev, in mapping table. */ + NVKM_INTR_VECTOR_0 = 0, +}; + +struct nvkm_intr { + const struct nvkm_intr_func { + bool (*pending)(struct nvkm_intr *); + void (*unarm)(struct nvkm_intr *); + void (*rearm)(struct nvkm_intr *); + void (*block)(struct nvkm_intr *, int leaf, u32 mask); + void (*allow)(struct nvkm_intr *, int leaf, u32 mask); + void (*reset)(struct nvkm_intr *, int leaf, u32 mask); + } *func; + const struct nvkm_intr_data { + int type; /* enum nvkm_subdev_type (+ve), enum nvkm_intr_type (-ve) */ + int inst; + int leaf; + u32 mask; /* 0-terminated. */ + bool legacy; /* auto-create "legacy" nvkm_subdev_intr() handler */ + } *data; + + struct nvkm_subdev *subdev; + int leaves; + u32 *stat; + u32 *mask; + + struct list_head head; +}; + +void nvkm_intr_ctor(struct nvkm_device *); +void nvkm_intr_dtor(struct nvkm_device *); +int nvkm_intr_install(struct nvkm_device *); +void nvkm_intr_unarm(struct nvkm_device *); +void nvkm_intr_rearm(struct nvkm_device *); + +int nvkm_intr_add(const struct nvkm_intr_func *, const struct nvkm_intr_data *, + struct nvkm_subdev *, int leaves, struct nvkm_intr *); +void nvkm_intr_block(struct nvkm_subdev *, enum nvkm_intr_type); +void nvkm_intr_allow(struct nvkm_subdev *, enum nvkm_intr_type); + +struct nvkm_inth; +typedef irqreturn_t (*nvkm_inth_func)(struct nvkm_inth *); + +struct nvkm_inth { + struct nvkm_intr *intr; + int leaf; + u32 mask; + nvkm_inth_func func; + + atomic_t allowed; + + struct list_head head; +}; + +int nvkm_inth_add(struct nvkm_intr *, enum nvkm_intr_type, enum nvkm_intr_prio, + struct nvkm_subdev *, nvkm_inth_func, struct nvkm_inth *); +void nvkm_inth_allow(struct nvkm_inth *); +void nvkm_inth_block(struct nvkm_inth *); +#endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/layout.h b/drivers/gpu/drm/nouveau/include/nvkm/core/layout.h index 7afe1579b20f..58108dea5aeb 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/layout.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/layout.h @@ -1,8 +1,10 @@ /* SPDX-License-Identifier: MIT */ +NVKM_LAYOUT_ONCE(NVKM_SUBDEV_TOP , struct nvkm_top , top) +NVKM_LAYOUT_ONCE(NVKM_SUBDEV_GSP , struct nvkm_gsp , gsp) +NVKM_LAYOUT_ONCE(NVKM_SUBDEV_VFN , struct nvkm_vfn , vfn) NVKM_LAYOUT_ONCE(NVKM_SUBDEV_PCI , struct nvkm_pci , pci) NVKM_LAYOUT_ONCE(NVKM_SUBDEV_VBIOS , struct nvkm_bios , bios) NVKM_LAYOUT_ONCE(NVKM_SUBDEV_DEVINIT , struct nvkm_devinit , devinit) -NVKM_LAYOUT_ONCE(NVKM_SUBDEV_TOP , struct nvkm_top , top) NVKM_LAYOUT_ONCE(NVKM_SUBDEV_PRIVRING, struct nvkm_subdev , privring) NVKM_LAYOUT_ONCE(NVKM_SUBDEV_GPIO , struct nvkm_gpio , gpio) NVKM_LAYOUT_ONCE(NVKM_SUBDEV_I2C , struct nvkm_i2c , i2c) @@ -23,7 +25,6 @@ NVKM_LAYOUT_ONCE(NVKM_SUBDEV_VOLT , struct nvkm_volt , volt) NVKM_LAYOUT_ONCE(NVKM_SUBDEV_ICCSENSE, struct nvkm_iccsense, iccsense) NVKM_LAYOUT_ONCE(NVKM_SUBDEV_THERM , struct nvkm_therm , therm) NVKM_LAYOUT_ONCE(NVKM_SUBDEV_CLK , struct nvkm_clk , clk) -NVKM_LAYOUT_ONCE(NVKM_SUBDEV_GSP , struct nvkm_gsp , gsp) NVKM_LAYOUT_INST(NVKM_SUBDEV_IOCTRL , struct nvkm_subdev , ioctrl, 3) NVKM_LAYOUT_ONCE(NVKM_SUBDEV_FLA , struct nvkm_subdev , fla) diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/memory.h b/drivers/gpu/drm/nouveau/include/nvkm/core/memory.h index 74d3f1a809d7..d3b6a68ddda3 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/memory.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/memory.h @@ -37,6 +37,7 @@ struct nvkm_memory_func { void (*release)(struct nvkm_memory *); int (*map)(struct nvkm_memory *, u64 offset, struct nvkm_vmm *, struct nvkm_vma *, void *argv, u32 argc); + int (*kmap)(struct nvkm_memory *, struct nvkm_memory **); }; struct nvkm_memory_ptrs { @@ -63,6 +64,7 @@ void nvkm_memory_tags_put(struct nvkm_memory *, struct nvkm_device *, #define nvkm_memory_boot(p,v) (p)->func->boot((p),(v)) #define nvkm_memory_map(p,o,vm,va,av,ac) \ (p)->func->map((p),(o),(vm),(va),(av),(ac)) +#define nvkm_memory_kmap(p,i) ((p)->func->kmap ? (p)->func->kmap((p), (i)) : -ENOSYS) /* accessor macros - kmap()/done() must bracket use of the other accessor * macros to guarantee correct behaviour across all chipsets diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/notify.h b/drivers/gpu/drm/nouveau/include/nvkm/core/notify.h deleted file mode 100644 index 3d358a66db3a..000000000000 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/notify.h +++ /dev/null @@ -1,39 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NVKM_NOTIFY_H__ -#define __NVKM_NOTIFY_H__ -#include <core/os.h> -struct nvkm_object; - -struct nvkm_notify { - struct nvkm_event *event; - struct list_head head; -#define NVKM_NOTIFY_USER 0 -#define NVKM_NOTIFY_WORK 1 - unsigned long flags; - int block; -#define NVKM_NOTIFY_DROP 0 -#define NVKM_NOTIFY_KEEP 1 - int (*func)(struct nvkm_notify *); - - /* set by nvkm_event ctor */ - u32 types; - int index; - u32 size; - - struct work_struct work; - /* this is const for a *very* good reason - the data might be on the - * stack from an irq handler. if you're not core/notify.c then you - * should probably think twice before casting it away... - */ - const void *data; -}; - -int nvkm_notify_init(struct nvkm_object *, struct nvkm_event *, - int (*func)(struct nvkm_notify *), bool work, - void *data, u32 size, u32 reply, - struct nvkm_notify *); -void nvkm_notify_fini(struct nvkm_notify *); -void nvkm_notify_get(struct nvkm_notify *); -void nvkm_notify_put(struct nvkm_notify *); -void nvkm_notify_send(struct nvkm_notify *, void *data, u32 size); -#endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/object.h b/drivers/gpu/drm/nouveau/include/nvkm/core/object.h index 7efcd5d2f2ff..ed1f66360782 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/object.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/object.h @@ -4,6 +4,7 @@ #include <core/oclass.h> struct nvkm_event; struct nvkm_gpuobj; +struct nvkm_uevent; struct nvkm_object { const struct nvkm_object_func *func; @@ -43,6 +44,7 @@ struct nvkm_object_func { int (*bind)(struct nvkm_object *, struct nvkm_gpuobj *, int align, struct nvkm_gpuobj **); int (*sclass)(struct nvkm_object *, int index, struct nvkm_oclass *); + int (*uevent)(struct nvkm_object *, void *argv, u32 argc, struct nvkm_uevent *); }; void nvkm_object_ctor(const struct nvkm_object_func *, diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/os.h b/drivers/gpu/drm/nouveau/include/nvkm/core/os.h index d7ba3205207f..4486d9862849 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/os.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/os.h @@ -34,4 +34,24 @@ nvkm_blob_dtor(struct nvkm_blob *blob) blob->data = NULL; blob->size = 0; } + +#define nvkm_list_find_next(p,h,m,c) ({ \ + typeof(p) _p = NULL; \ + list_for_each_entry_continue(p, (h), m) { \ + if (c) { \ + _p = p; \ + break; \ + } \ + } \ + _p; \ +}) +#define nvkm_list_find(p,h,m,c) \ + (p = container_of((h), typeof(*p), m), nvkm_list_find_next(p, (h), m, (c))) +#define nvkm_list_foreach(p,h,m,c) \ + for (p = nvkm_list_find(p, (h), m, (c)); p; p = nvkm_list_find_next(p, (h), m, (c))) + +/*FIXME: remove after */ +#define nvkm_fifo_chan nvkm_chan +#define nvkm_fifo_chan_func nvkm_chan_func +#define nvkm_fifo_cgrp nvkm_cgrp #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/subdev.h b/drivers/gpu/drm/nouveau/include/nvkm/core/subdev.h index 96113c8bee8c..bce6e1ba09ea 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/subdev.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/subdev.h @@ -17,10 +17,19 @@ struct nvkm_subdev { struct nvkm_device *device; enum nvkm_subdev_type type; int inst; + char name[16]; u32 debug; - struct list_head head; + struct { + refcount_t refcount; + struct mutex mutex; + bool enabled; + } use; + + struct nvkm_inth inth; + + struct list_head head; void **pself; bool oneinit; }; @@ -38,22 +47,41 @@ struct nvkm_subdev_func { extern const char *nvkm_subdev_type[NVKM_SUBDEV_NR]; int nvkm_subdev_new_(const struct nvkm_subdev_func *, struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_subdev **); -void nvkm_subdev_ctor(const struct nvkm_subdev_func *, struct nvkm_device *, - enum nvkm_subdev_type, int inst, struct nvkm_subdev *); +void __nvkm_subdev_ctor(const struct nvkm_subdev_func *, struct nvkm_device *, + enum nvkm_subdev_type, int inst, struct nvkm_subdev *); + +static inline void +nvkm_subdev_ctor(const struct nvkm_subdev_func *func, struct nvkm_device *device, + enum nvkm_subdev_type type, int inst, struct nvkm_subdev *subdev) +{ + __nvkm_subdev_ctor(func, device, type, inst, subdev); + mutex_init(&subdev->use.mutex); +} + void nvkm_subdev_disable(struct nvkm_device *, enum nvkm_subdev_type, int inst); void nvkm_subdev_del(struct nvkm_subdev **); +int nvkm_subdev_ref(struct nvkm_subdev *); +void nvkm_subdev_unref(struct nvkm_subdev *); int nvkm_subdev_preinit(struct nvkm_subdev *); +int nvkm_subdev_oneinit(struct nvkm_subdev *); int nvkm_subdev_init(struct nvkm_subdev *); int nvkm_subdev_fini(struct nvkm_subdev *, bool suspend); int nvkm_subdev_info(struct nvkm_subdev *, u64, u64 *); void nvkm_subdev_intr(struct nvkm_subdev *); /* subdev logging */ -#define nvkm_printk_(s,l,p,f,a...) do { \ - const struct nvkm_subdev *_subdev = (s); \ - if (CONFIG_NOUVEAU_DEBUG >= (l) && _subdev->debug >= (l)) \ - dev_##p(_subdev->device->dev, "%s: "f, _subdev->name, ##a); \ +#define nvkm_printk_ok(s,u,l) \ + ((CONFIG_NOUVEAU_DEBUG >= (l)) && ((s)->debug >= (l) || ((u) && (u)->debug >= (l)))) +#define nvkm_printk___(s,u,l,p,f,a...) do { \ + if (nvkm_printk_ok((s), (u), (l))) { \ + if ((u) && (u) != (s)) \ + dev_##p((s)->device->dev, "%s(%s):"f, (s)->name, (u)->name, ##a); \ + else \ + dev_##p((s)->device->dev, "%s:"f, (s)->name, ##a); \ + } \ } while(0) +#define nvkm_printk__(s,l,p,f,a...) nvkm_printk___((s), (s), (l), p, f, ##a) +#define nvkm_printk_(s,l,p,f,a...) nvkm_printk__((s), (l), p, " "f, ##a) #define nvkm_printk(s,l,p,f,a...) nvkm_printk_((s), NV_DBG_##l, p, f, ##a) #define nvkm_fatal(s,f,a...) nvkm_printk((s), FATAL, crit, f, ##a) #define nvkm_error(s,f,a...) nvkm_printk((s), ERROR, err, f, ##a) diff --git a/drivers/gpu/drm/nouveau/include/nvkm/core/tegra.h b/drivers/gpu/drm/nouveau/include/nvkm/core/tegra.h index 924009dd2bb0..ccee53d4e4ec 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/core/tegra.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/core/tegra.h @@ -8,7 +8,6 @@ struct nvkm_device_tegra { const struct nvkm_device_tegra_func *func; struct nvkm_device device; struct platform_device *pdev; - int irq; struct reset_control *rst; struct clk *clk; diff --git a/drivers/gpu/drm/nouveau/include/nvkm/engine/ce.h b/drivers/gpu/drm/nouveau/include/nvkm/engine/ce.h index cfd2da8e66fe..b616a1e8ca02 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/engine/ce.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/engine/ce.h @@ -12,4 +12,6 @@ int gp100_ce_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct n int gp102_ce_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_engine **); int gv100_ce_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_engine **); int tu102_ce_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_engine **); +int ga100_ce_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_engine **); +int ga102_ce_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_engine **); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/engine/disp.h b/drivers/gpu/drm/nouveau/include/nvkm/engine/disp.h index 8b5d8a434be8..ad9aef2df48f 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/engine/disp.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/engine/disp.h @@ -16,6 +16,7 @@ struct nvkm_disp { struct list_head conns; struct nvkm_event hpd; +#define NVKM_DISP_HEAD_EVENT_VBLANK BIT(0) struct nvkm_event vblank; struct { @@ -31,13 +32,7 @@ struct nvkm_disp { struct { unsigned long mask; int nr; - } wndw, head, dac; - - struct { - unsigned long mask; - int nr; - u32 lvdsconf; - } sor; + } wndw, head, dac, sor; struct { unsigned long mask; diff --git a/drivers/gpu/drm/nouveau/include/nvkm/engine/falcon.h b/drivers/gpu/drm/nouveau/include/nvkm/engine/falcon.h index b593407b9e36..cd86d9198e4a 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/engine/falcon.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/engine/falcon.h @@ -16,15 +16,16 @@ enum nvkm_falcon_dmaidx { struct nvkm_falcon { const struct nvkm_falcon_func *func; - const struct nvkm_subdev *owner; + struct nvkm_subdev *owner; const char *name; u32 addr; + u32 addr2; struct mutex mutex; struct mutex dmem_mutex; bool oneinit; - const struct nvkm_subdev *user; + struct nvkm_subdev *user; u8 version; u8 secret; @@ -50,13 +51,42 @@ struct nvkm_falcon { struct nvkm_engine engine; }; -int nvkm_falcon_get(struct nvkm_falcon *, const struct nvkm_subdev *); -void nvkm_falcon_put(struct nvkm_falcon *, const struct nvkm_subdev *); +int nvkm_falcon_get(struct nvkm_falcon *, struct nvkm_subdev *); +void nvkm_falcon_put(struct nvkm_falcon *, struct nvkm_subdev *); int nvkm_falcon_new_(const struct nvkm_falcon_func *, struct nvkm_device *, enum nvkm_subdev_type, int inst, bool enable, u32 addr, struct nvkm_engine **); struct nvkm_falcon_func { + int (*disable)(struct nvkm_falcon *); + int (*enable)(struct nvkm_falcon *); + int (*select)(struct nvkm_falcon *); + u32 addr2; + bool reset_pmc; + int (*reset_eng)(struct nvkm_falcon *); + int (*reset_prep)(struct nvkm_falcon *); + int (*reset_wait_mem_scrubbing)(struct nvkm_falcon *); + + u32 debug; + void (*bind_inst)(struct nvkm_falcon *, int target, u64 addr); + int (*bind_stat)(struct nvkm_falcon *, bool intr); + bool bind_intr; + + const struct nvkm_falcon_func_pio *imem_pio; + const struct nvkm_falcon_func_dma *imem_dma; + + const struct nvkm_falcon_func_pio *dmem_pio; + const struct nvkm_falcon_func_dma *dmem_dma; + + u32 emem_addr; + const struct nvkm_falcon_func_pio *emem_pio; + + struct { + u32 head; + u32 tail; + u32 stride; + } cmdq, msgq; + struct { u32 *data; u32 size; @@ -66,29 +96,11 @@ struct nvkm_falcon_func { u32 size; } data; void (*init)(struct nvkm_falcon *); - void (*intr)(struct nvkm_falcon *, struct nvkm_fifo_chan *); - - u32 debug; - u32 fbif; + void (*intr)(struct nvkm_falcon *, struct nvkm_chan *); void (*load_imem)(struct nvkm_falcon *, void *, u32, u32, u16, u8, bool); void (*load_dmem)(struct nvkm_falcon *, void *, u32, u32, u8); - void (*read_dmem)(struct nvkm_falcon *, u32, u32, u8, void *); - u32 emem_addr; - void (*bind_context)(struct nvkm_falcon *, struct nvkm_memory *); - int (*wait_for_halt)(struct nvkm_falcon *, u32); - int (*clear_interrupt)(struct nvkm_falcon *, u32); - void (*set_start_addr)(struct nvkm_falcon *, u32 start_addr); void (*start)(struct nvkm_falcon *); - int (*enable)(struct nvkm_falcon *falcon); - void (*disable)(struct nvkm_falcon *falcon); - int (*reset)(struct nvkm_falcon *); - - struct { - u32 head; - u32 tail; - u32 stride; - } cmdq, msgq; struct nvkm_sclass sclass[]; }; @@ -116,13 +128,5 @@ nvkm_falcon_mask(struct nvkm_falcon *falcon, u32 addr, u32 mask, u32 val) void nvkm_falcon_load_imem(struct nvkm_falcon *, void *, u32, u32, u16, u8, bool); void nvkm_falcon_load_dmem(struct nvkm_falcon *, void *, u32, u32, u8); -void nvkm_falcon_read_dmem(struct nvkm_falcon *, u32, u32, u8, void *); -void nvkm_falcon_bind_context(struct nvkm_falcon *, struct nvkm_memory *); -void nvkm_falcon_set_start_addr(struct nvkm_falcon *, u32); void nvkm_falcon_start(struct nvkm_falcon *); -int nvkm_falcon_wait_for_halt(struct nvkm_falcon *, u32); -int nvkm_falcon_clear_interrupt(struct nvkm_falcon *, u32); -int nvkm_falcon_enable(struct nvkm_falcon *); -void nvkm_falcon_disable(struct nvkm_falcon *); -int nvkm_falcon_reset(struct nvkm_falcon *); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/engine/fifo.h b/drivers/gpu/drm/nouveau/include/nvkm/engine/fifo.h index 15099913504d..221abd6c4310 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/engine/fifo.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/engine/fifo.h @@ -6,56 +6,76 @@ #include <core/event.h> struct nvkm_fault_data; -#define NVKM_FIFO_CHID_NR 4096 #define NVKM_FIFO_ENGN_NR 16 -struct nvkm_fifo_engn { - struct nvkm_object *object; - int refcount; - int usecount; -}; - -struct nvkm_fifo_chan { - const struct nvkm_fifo_chan_func *func; - struct nvkm_fifo *fifo; - u32 engm; - struct nvkm_object object; +struct nvkm_chan { + const struct nvkm_chan_func *func; + char name[64]; + struct nvkm_cgrp *cgrp; + int runq; - struct list_head head; - u16 chid; struct nvkm_gpuobj *inst; - struct nvkm_gpuobj *push; struct nvkm_vmm *vmm; - u64 addr; - u32 size; + struct nvkm_gpuobj *push; + int id; + + struct { + struct nvkm_memory *mem; + u32 base; + } userd; + + u32 ramfc_offset; + struct nvkm_gpuobj *ramfc; + struct nvkm_gpuobj *cache; + struct nvkm_gpuobj *eng; + struct nvkm_gpuobj *pgd; + struct nvkm_ramht *ramht; + + spinlock_t lock; + atomic_t blocked; + atomic_t errored; - struct nvkm_fifo_engn engn[NVKM_FIFO_ENGN_NR]; + struct list_head cctxs; + struct list_head head; }; +struct nvkm_chan *nvkm_chan_get_chid(struct nvkm_engine *, int id, unsigned long *irqflags); +struct nvkm_chan *nvkm_chan_get_inst(struct nvkm_engine *, u64 inst, unsigned long *irqflags); +void nvkm_chan_put(struct nvkm_chan **, unsigned long irqflags); + struct nvkm_fifo { const struct nvkm_fifo_func *func; struct nvkm_engine engine; - DECLARE_BITMAP(mask, NVKM_FIFO_CHID_NR); - int nr; - struct list_head chan; + struct nvkm_chid *chid; + struct nvkm_chid *cgid; + + struct list_head runqs; + struct list_head runls; + + struct { +#define NVKM_FIFO_NONSTALL_EVENT BIT(0) + struct nvkm_event event; + struct nvkm_inth intr; + } nonstall; + + struct { + u32 chan_msec; + } timeout; + + struct { + struct nvkm_memory *mem; + struct nvkm_vma *bar1; + } userd; + spinlock_t lock; struct mutex mutex; - - struct nvkm_event uevent; /* async user trigger */ - struct nvkm_event kevent; /* channel killed */ }; void nvkm_fifo_fault(struct nvkm_fifo *, struct nvkm_fault_data *); void nvkm_fifo_pause(struct nvkm_fifo *, unsigned long *); void nvkm_fifo_start(struct nvkm_fifo *, unsigned long *); - -void nvkm_fifo_chan_put(struct nvkm_fifo *, unsigned long flags, - struct nvkm_fifo_chan **); -struct nvkm_fifo_chan * -nvkm_fifo_chan_inst(struct nvkm_fifo *, u64 inst, unsigned long *flags); -struct nvkm_fifo_chan * -nvkm_fifo_chan_chid(struct nvkm_fifo *, int chid, unsigned long *flags); +bool nvkm_fifo_ctxsw_in_progress(struct nvkm_engine *); int nv04_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int nv10_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); @@ -63,6 +83,7 @@ int nv17_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct int nv40_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int nv50_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int g84_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); +int g98_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int gf100_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int gk104_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int gk110_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); @@ -70,10 +91,9 @@ int gk208_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct int gk20a_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int gm107_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int gm200_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); -int gm20b_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int gp100_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); -int gp10b_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int gv100_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int tu102_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); +int ga100_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); int ga102_fifo_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_fifo **); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/engine/gr.h b/drivers/gpu/drm/nouveau/include/nvkm/engine/gr.h index b28b752ffaa2..a2333cfe6955 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/engine/gr.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/engine/gr.h @@ -54,4 +54,5 @@ int gp108_gr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct n int gp10b_gr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_gr **); int gv100_gr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_gr **); int tu102_gr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_gr **); +int ga102_gr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_gr **); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/engine/nvdec.h b/drivers/gpu/drm/nouveau/include/nvkm/engine/nvdec.h index 97bd3092f68a..9baf197ac833 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/engine/nvdec.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/engine/nvdec.h @@ -12,4 +12,5 @@ struct nvkm_nvdec { }; int gm107_nvdec_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_nvdec **); +int ga102_nvdec_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_nvdec **); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/engine/sec2.h b/drivers/gpu/drm/nouveau/include/nvkm/engine/sec2.h index 06264c840eae..8d48fb20fa54 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/engine/sec2.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/engine/sec2.h @@ -10,15 +10,18 @@ struct nvkm_sec2 { struct nvkm_engine engine; struct nvkm_falcon falcon; + atomic_t running; + atomic_t initmsg; + struct nvkm_falcon_qmgr *qmgr; struct nvkm_falcon_cmdq *cmdq; struct nvkm_falcon_msgq *msgq; struct work_struct work; - bool initmsg_received; }; int gp102_sec2_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_sec2 **); int gp108_sec2_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_sec2 **); int tu102_sec2_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_sec2 **); +int ga102_sec2_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_sec2 **); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/acr.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/acr.h index c0b254f7f0b5..73d2a6ae9ab2 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/acr.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/acr.h @@ -36,7 +36,7 @@ struct nvkm_acr { const struct nvkm_acr_func *func; struct nvkm_subdev subdev; - struct list_head hsfw, hsf; + struct list_head hsfw; struct list_head lsfw, lsf; u64 managed_falcons; @@ -50,6 +50,7 @@ struct nvkm_acr { struct nvkm_vmm *vmm; bool done; + struct nvkm_acr_lsf *rtos; const struct firmware *wpr_fw; bool wpr_comp; @@ -64,7 +65,9 @@ int gm20b_acr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct int gp102_acr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_acr **); int gp108_acr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_acr **); int gp10b_acr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_acr **); +int gv100_acr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_acr **); int tu102_acr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_acr **); +int ga102_acr_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_acr **); struct nvkm_acr_lsfw { const struct nvkm_acr_lsf_func *func; @@ -77,6 +80,7 @@ struct nvkm_acr_lsfw { const struct firmware *sig; + bool secure_bootloader; u32 bootloader_size; u32 bootloader_imem_offset; @@ -87,10 +91,19 @@ struct nvkm_acr_lsfw { u32 app_resident_code_size; u32 app_resident_data_offset; u32 app_resident_data_size; + u32 app_imem_offset; + u32 app_dmem_offset; u32 ucode_size; u32 data_size; + u32 fuse_ver; + u32 engine_id; + u32 ucode_id; + u32 sig_size; + u32 sig_nr; + u8 *sigs; + struct { u32 lsb; u32 img; @@ -105,10 +118,10 @@ struct nvkm_acr_lsf_func { #define NVKM_ACR_LSF_DMACTL_REQ_CTX 0x00000004 #define NVKM_ACR_LSF_FORCE_PRIV_LOAD 0x00000008 u32 flags; + u32 bl_entry; u32 bld_size; void (*bld_write)(struct nvkm_acr *, u32 bld, struct nvkm_acr_lsfw *); void (*bld_patch)(struct nvkm_acr *, u32 bld, s64 adjust); - int (*boot)(struct nvkm_falcon *); u64 bootstrap_falcons; int (*bootstrap_falcon)(struct nvkm_falcon *, enum nvkm_acr_lsf_id); int (*bootstrap_multiple_falcons)(struct nvkm_falcon *, u32 mask); @@ -122,8 +135,20 @@ int nvkm_acr_lsfw_load_sig_image_desc_v1(struct nvkm_subdev *, struct nvkm_falcon *, enum nvkm_acr_lsf_id, const char *path, int ver, const struct nvkm_acr_lsf_func *); + +int +nvkm_acr_lsfw_load_sig_image_desc_v2(struct nvkm_subdev *, struct nvkm_falcon *, + enum nvkm_acr_lsf_id, const char *path, + int ver, const struct nvkm_acr_lsf_func *); + int nvkm_acr_lsfw_load_bl_inst_data_sig(struct nvkm_subdev *, struct nvkm_falcon *, enum nvkm_acr_lsf_id, const char *path, int ver, const struct nvkm_acr_lsf_func *); + +int +nvkm_acr_lsfw_load_bl_sig_net(struct nvkm_subdev *, struct nvkm_falcon *, + enum nvkm_acr_lsf_id, const char *path, + int ver, const struct nvkm_acr_lsf_func *, + const void *, u32, const void *, u32); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/fault.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/fault.h index 9c78f072d62b..e40bbf378a8d 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/fault.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/fault.h @@ -2,18 +2,21 @@ #define __NVKM_FAULT_H__ #include <core/subdev.h> #include <core/event.h> -#include <core/notify.h> struct nvkm_fault { const struct nvkm_fault_func *func; struct nvkm_subdev subdev; + struct nvkm_inth info_fault; + struct nvkm_fault_buffer *buffer[2]; int buffer_nr; +#define NVKM_FAULT_BUFFER_EVENT_PENDING BIT(0) struct nvkm_event event; - struct nvkm_notify nrpfb; + struct nvkm_event_ntfy nrpfb; + struct work_struct nrpfb_work; struct nvkm_device_oclass user; }; diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/fb.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/fb.h index ef6a6297148c..40768373cdd9 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/fb.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/fb.h @@ -35,6 +35,11 @@ struct nvkm_fb { struct nvkm_blob vpr_scrubber; + struct { + struct page *flush_page; + dma_addr_t flush_page_addr; + } sysmem; + struct nvkm_ram *ram; struct { @@ -53,6 +58,8 @@ struct nvkm_fb { struct nvkm_memory *mmu_wr; }; +int nvkm_fb_mem_unlock(struct nvkm_fb *); + void nvkm_fb_tile_init(struct nvkm_fb *, int region, u32 addr, u32 size, u32 pitch, u32 flags, struct nvkm_fb_tile *); void nvkm_fb_tile_fini(struct nvkm_fb *, int region, struct nvkm_fb_tile *); diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gpio.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gpio.h index 0e46ea1fe972..537c4fc58b4f 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gpio.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gpio.h @@ -8,9 +8,6 @@ #include <subdev/bios/gpio.h> struct nvkm_gpio_ntfy_req { -#define NVKM_GPIO_HI 0x01 -#define NVKM_GPIO_LO 0x02 -#define NVKM_GPIO_TOGGLED 0x03 u8 mask; u8 line; }; @@ -23,6 +20,9 @@ struct nvkm_gpio { const struct nvkm_gpio_func *func; struct nvkm_subdev subdev; +#define NVKM_GPIO_HI BIT(0) +#define NVKM_GPIO_LO BIT(1) +#define NVKM_GPIO_TOGGLED (NVKM_GPIO_HI | NVKM_GPIO_LO) struct nvkm_event event; }; diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h index cf42a59d4e58..72619d7df73e 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/gsp.h @@ -5,9 +5,12 @@ #include <core/falcon.h> struct nvkm_gsp { + const struct nvkm_gsp_func *func; struct nvkm_subdev subdev; + struct nvkm_falcon falcon; }; int gv100_gsp_new(struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_gsp **); +int ga102_gsp_new(struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_gsp **); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/i2c.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/i2c.h index 146e13292203..40a1065ae626 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/i2c.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/i2c.h @@ -7,20 +7,6 @@ #include <subdev/bios.h> #include <subdev/bios/i2c.h> -struct nvkm_i2c_ntfy_req { -#define NVKM_I2C_PLUG 0x01 -#define NVKM_I2C_UNPLUG 0x02 -#define NVKM_I2C_IRQ 0x04 -#define NVKM_I2C_DONE 0x08 -#define NVKM_I2C_ANY 0x0f - u8 mask; - u8 port; -}; - -struct nvkm_i2c_ntfy_rep { - u8 mask; -}; - struct nvkm_i2c_bus_probe { struct i2c_board_info dev; u8 udelay; /* set to 0 to use the standard delay */ @@ -79,6 +65,11 @@ struct nvkm_i2c { struct list_head bus; struct list_head aux; +#define NVKM_I2C_PLUG BIT(0) +#define NVKM_I2C_UNPLUG BIT(1) +#define NVKM_I2C_IRQ BIT(2) +#define NVKM_I2C_DONE BIT(3) +#define NVKM_I2C_ANY (NVKM_I2C_PLUG | NVKM_I2C_UNPLUG | NVKM_I2C_IRQ | NVKM_I2C_DONE) struct nvkm_event event; }; diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/instmem.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/instmem.h index f967b97d163c..fcdaefc99fe8 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/instmem.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/instmem.h @@ -28,7 +28,7 @@ u32 nvkm_instmem_rd32(struct nvkm_instmem *, u32 addr); void nvkm_instmem_wr32(struct nvkm_instmem *, u32 addr, u32 data); int nvkm_instobj_new(struct nvkm_instmem *, u32 size, u32 align, bool zero, struct nvkm_memory **); - +int nvkm_instobj_wrap(struct nvkm_device *, struct nvkm_memory *, struct nvkm_memory **); int nv04_instmem_new(struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_instmem **); int nv40_instmem_new(struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_instmem **); diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/ltc.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/ltc.h index d32a326a9290..64294042ec07 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/ltc.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/ltc.h @@ -4,7 +4,8 @@ #include <core/subdev.h> #include <core/mm.h> -#define NVKM_LTC_MAX_ZBC_CNT 16 +#define NVKM_LTC_MAX_ZBC_COLOR_CNT 32 +#define NVKM_LTC_MAX_ZBC_DEPTH_CNT 16 struct nvkm_ltc { const struct nvkm_ltc_func *func; @@ -18,11 +19,13 @@ struct nvkm_ltc { u32 tag_base; struct nvkm_memory *tag_ram; - int zbc_min; - int zbc_max; - u32 zbc_color[NVKM_LTC_MAX_ZBC_CNT][4]; - u32 zbc_depth[NVKM_LTC_MAX_ZBC_CNT]; - u32 zbc_stencil[NVKM_LTC_MAX_ZBC_CNT]; + int zbc_color_min; + int zbc_color_max; + u32 zbc_color[NVKM_LTC_MAX_ZBC_COLOR_CNT][4]; + int zbc_depth_min; + int zbc_depth_max; + u32 zbc_depth[NVKM_LTC_MAX_ZBC_DEPTH_CNT]; + u32 zbc_stencil[NVKM_LTC_MAX_ZBC_DEPTH_CNT]; }; void nvkm_ltc_tags_clear(struct nvkm_device *, u32 first, u32 count); @@ -41,4 +44,5 @@ int gm200_ltc_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct int gp100_ltc_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_ltc **); int gp102_ltc_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_ltc **); int gp10b_ltc_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_ltc **); +int ga102_ltc_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_ltc **); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/mc.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/mc.h index cb86a56e68d4..127ac545e4b2 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/mc.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/mc.h @@ -6,15 +6,14 @@ struct nvkm_mc { const struct nvkm_mc_func *func; struct nvkm_subdev subdev; + + struct nvkm_intr intr; }; void nvkm_mc_enable(struct nvkm_device *, enum nvkm_subdev_type, int); void nvkm_mc_disable(struct nvkm_device *, enum nvkm_subdev_type, int); bool nvkm_mc_enabled(struct nvkm_device *, enum nvkm_subdev_type, int); void nvkm_mc_reset(struct nvkm_device *, enum nvkm_subdev_type, int); -void nvkm_mc_intr(struct nvkm_device *, bool *handled); -void nvkm_mc_intr_unarm(struct nvkm_device *); -void nvkm_mc_intr_rearm(struct nvkm_device *); void nvkm_mc_intr_mask(struct nvkm_device *, enum nvkm_subdev_type, int, bool enable); void nvkm_mc_unk260(struct nvkm_device *, u32 data); @@ -31,6 +30,5 @@ int gk104_mc_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct n int gk20a_mc_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_mc **); int gp100_mc_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_mc **); int gp10b_mc_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_mc **); -int tu102_mc_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_mc **); int ga100_mc_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_mc **); #endif diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/pci.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/pci.h index 74c19bdfb757..3c103101d5fc 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/pci.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/pci.h @@ -13,7 +13,6 @@ struct nvkm_pci { const struct nvkm_pci_func *func; struct nvkm_subdev subdev; struct pci_dev *pdev; - int irq; struct { struct agp_bridge_data *bridge; @@ -38,6 +37,7 @@ void nvkm_pci_wr08(struct nvkm_pci *, u16 addr, u8 data); void nvkm_pci_wr32(struct nvkm_pci *, u16 addr, u32 data); u32 nvkm_pci_mask(struct nvkm_pci *, u16 addr, u32 mask, u32 value); void nvkm_pci_rom_shadow(struct nvkm_pci *, bool shadow); +void nvkm_pci_msi_rearm(struct nvkm_device *); int nv04_pci_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_pci **); int nv40_pci_new(struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_pci **); diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/top.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/top.h index ee75c5524c43..73e717b980b8 100644 --- a/drivers/gpu/drm/nouveau/include/nvkm/subdev/top.h +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/top.h @@ -21,6 +21,7 @@ struct nvkm_top_device { struct list_head head; }; +int nvkm_top_parse(struct nvkm_device *); u32 nvkm_top_addr(struct nvkm_device *, enum nvkm_subdev_type, int); u32 nvkm_top_reset(struct nvkm_device *, enum nvkm_subdev_type, int); u32 nvkm_top_intr_mask(struct nvkm_device *, enum nvkm_subdev_type, int); diff --git a/drivers/gpu/drm/nouveau/include/nvkm/subdev/vfn.h b/drivers/gpu/drm/nouveau/include/nvkm/subdev/vfn.h new file mode 100644 index 000000000000..cc6d0796c265 --- /dev/null +++ b/drivers/gpu/drm/nouveau/include/nvkm/subdev/vfn.h @@ -0,0 +1,23 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __NVKM_VFN_H__ +#define __NVKM_VFN_H__ +#include <core/subdev.h> + +struct nvkm_vfn { + const struct nvkm_vfn_func *func; + struct nvkm_subdev subdev; + + struct { + u32 priv; + u32 user; + } addr; + + struct nvkm_intr intr; + + struct nvkm_device_oclass user; +}; + +int gv100_vfn_new(struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_vfn **); +int tu102_vfn_new(struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_vfn **); +int ga100_vfn_new(struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_vfn **); +#endif diff --git a/drivers/gpu/drm/nouveau/nouveau_abi16.c b/drivers/gpu/drm/nouveau/nouveau_abi16.c index 5bee655e7e63..82dab51d8aeb 100644 --- a/drivers/gpu/drm/nouveau/nouveau_abi16.c +++ b/drivers/gpu/drm/nouveau/nouveau_abi16.c @@ -27,7 +27,6 @@ #include <nvif/ioctl.h> #include <nvif/class.h> #include <nvif/cl0002.h> -#include <nvif/cla06f.h> #include <nvif/unpack.h> #include "nouveau_drv.h" @@ -253,7 +252,7 @@ nouveau_abi16_ioctl_channel_alloc(ABI16_IOCTL_ARGS) struct nouveau_abi16 *abi16 = nouveau_abi16_get(file_priv); struct nouveau_abi16_chan *chan; struct nvif_device *device; - u64 engine; + u64 engine, runm; int ret; if (unlikely(!abi16)) @@ -263,6 +262,7 @@ nouveau_abi16_ioctl_channel_alloc(ABI16_IOCTL_ARGS) return nouveau_abi16_put(abi16, -ENODEV); device = &abi16->device; + engine = NV_DEVICE_HOST_RUNLIST_ENGINES_GR; /* hack to allow channel engine type specification on kepler */ if (device->info.family >= NV_DEVICE_INFO_V0_KEPLER) { @@ -276,19 +276,18 @@ nouveau_abi16_ioctl_channel_alloc(ABI16_IOCTL_ARGS) default: return nouveau_abi16_put(abi16, -ENOSYS); } - } else { - engine = NV_DEVICE_HOST_RUNLIST_ENGINES_GR; - } - if (engine != NV_DEVICE_HOST_RUNLIST_ENGINES_CE) - engine = nvif_fifo_runlist(device, engine); - else - engine = nvif_fifo_runlist_ce(device); - init->fb_ctxdma_handle = engine; - init->tt_ctxdma_handle = 0; + init->fb_ctxdma_handle = 0; + init->tt_ctxdma_handle = 0; + } } - if (init->fb_ctxdma_handle == ~0 || init->tt_ctxdma_handle == ~0) + if (engine != NV_DEVICE_HOST_RUNLIST_ENGINES_CE) + runm = nvif_fifo_runlist(device, engine); + else + runm = nvif_fifo_runlist_ce(device); + + if (!runm || init->fb_ctxdma_handle == ~0 || init->tt_ctxdma_handle == ~0) return nouveau_abi16_put(abi16, -EINVAL); /* allocate "abi16 channel" data and make up a handle for it */ @@ -300,8 +299,8 @@ nouveau_abi16_ioctl_channel_alloc(ABI16_IOCTL_ARGS) list_add(&chan->head, &abi16->channels); /* create channel object and initialise dma and fence management */ - ret = nouveau_channel_new(drm, device, init->fb_ctxdma_handle, - init->tt_ctxdma_handle, false, &chan->chan); + ret = nouveau_channel_new(drm, device, false, runm, init->fb_ctxdma_handle, + init->tt_ctxdma_handle, &chan->chan); if (ret) goto done; diff --git a/drivers/gpu/drm/nouveau/nouveau_bo.c b/drivers/gpu/drm/nouveau/nouveau_bo.c index 813937ad1dc2..a11871e3119c 100644 --- a/drivers/gpu/drm/nouveau/nouveau_bo.c +++ b/drivers/gpu/drm/nouveau/nouveau_bo.c @@ -856,6 +856,9 @@ nouveau_bo_move_init(struct nouveau_drm *drm) int (*init)(struct nouveau_channel *, u32 handle); } _methods[] = { { "COPY", 4, 0xc7b5, nve0_bo_move_copy, nve0_bo_move_init }, + { "GRCE", 0, 0xc7b5, nve0_bo_move_copy, nvc0_bo_move_init }, + { "COPY", 4, 0xc6b5, nve0_bo_move_copy, nve0_bo_move_init }, + { "GRCE", 0, 0xc6b5, nve0_bo_move_copy, nvc0_bo_move_init }, { "COPY", 4, 0xc5b5, nve0_bo_move_copy, nve0_bo_move_init }, { "GRCE", 0, 0xc5b5, nve0_bo_move_copy, nvc0_bo_move_init }, { "COPY", 4, 0xc3b5, nve0_bo_move_copy, nve0_bo_move_init }, diff --git a/drivers/gpu/drm/nouveau/nouveau_chan.c b/drivers/gpu/drm/nouveau/nouveau_chan.c index 48dea5d0c580..e648ecd0c1a0 100644 --- a/drivers/gpu/drm/nouveau/nouveau_chan.c +++ b/drivers/gpu/drm/nouveau/nouveau_chan.c @@ -25,12 +25,7 @@ #include <nvif/class.h> #include <nvif/cl0002.h> -#include <nvif/cl006b.h> -#include <nvif/cl506f.h> -#include <nvif/cl906f.h> -#include <nvif/cla06f.h> -#include <nvif/clc36f.h> -#include <nvif/ioctl.h> +#include <nvif/if0020.h> #include "nouveau_drv.h" #include "nouveau_dma.h" @@ -46,15 +41,17 @@ int nouveau_vram_pushbuf; module_param_named(vram_pushbuf, nouveau_vram_pushbuf, int, 0400); static int -nouveau_channel_killed(struct nvif_notify *ntfy) +nouveau_channel_killed(struct nvif_event *event, void *repv, u32 repc) { - struct nouveau_channel *chan = container_of(ntfy, typeof(*chan), kill); + struct nouveau_channel *chan = container_of(event, typeof(*chan), kill); struct nouveau_cli *cli = (void *)chan->user.client; + NV_PRINTK(warn, cli, "channel %d killed!\n", chan->chid); atomic_set(&chan->killed, 1); if (chan->fence) nouveau_fence_context_kill(chan->fence, -ENODEV); - return NVIF_NOTIFY_DROP; + + return NVIF_EVENT_DROP; } int @@ -96,8 +93,9 @@ nouveau_channel_del(struct nouveau_channel **pchan) nvif_object_dtor(&chan->nvsw); nvif_object_dtor(&chan->gart); nvif_object_dtor(&chan->vram); - nvif_notify_dtor(&chan->kill); + nvif_event_dtor(&chan->kill); nvif_object_dtor(&chan->user); + nvif_mem_dtor(&chan->mem_userd); nvif_object_dtor(&chan->push.ctxdma); nouveau_vma_del(&chan->push.vma); nouveau_bo_unmap(chan->push.buffer); @@ -247,134 +245,113 @@ nouveau_channel_prep(struct nouveau_drm *drm, struct nvif_device *device, } static int -nouveau_channel_ind(struct nouveau_drm *drm, struct nvif_device *device, - u64 runlist, bool priv, struct nouveau_channel **pchan) +nouveau_channel_ctor(struct nouveau_drm *drm, struct nvif_device *device, bool priv, u64 runm, + struct nouveau_channel **pchan) { - static const u16 oclasses[] = { AMPERE_CHANNEL_GPFIFO_B, - TURING_CHANNEL_GPFIFO_A, - VOLTA_CHANNEL_GPFIFO_A, - PASCAL_CHANNEL_GPFIFO_A, - MAXWELL_CHANNEL_GPFIFO_A, - KEPLER_CHANNEL_GPFIFO_B, - KEPLER_CHANNEL_GPFIFO_A, - FERMI_CHANNEL_GPFIFO, - G82_CHANNEL_GPFIFO, - NV50_CHANNEL_GPFIFO, - 0 }; - const u16 *oclass = oclasses; - union { - struct nv50_channel_gpfifo_v0 nv50; - struct fermi_channel_gpfifo_v0 fermi; - struct kepler_channel_gpfifo_a_v0 kepler; - struct volta_channel_gpfifo_a_v0 volta; + static const struct { + s32 oclass; + int version; + } hosts[] = { + { AMPERE_CHANNEL_GPFIFO_B, 0 }, + { AMPERE_CHANNEL_GPFIFO_A, 0 }, + { TURING_CHANNEL_GPFIFO_A, 0 }, + { VOLTA_CHANNEL_GPFIFO_A, 0 }, + { PASCAL_CHANNEL_GPFIFO_A, 0 }, + { MAXWELL_CHANNEL_GPFIFO_A, 0 }, + { KEPLER_CHANNEL_GPFIFO_B, 0 }, + { KEPLER_CHANNEL_GPFIFO_A, 0 }, + { FERMI_CHANNEL_GPFIFO , 0 }, + { G82_CHANNEL_GPFIFO , 0 }, + { NV50_CHANNEL_GPFIFO , 0 }, + { NV40_CHANNEL_DMA , 0 }, + { NV17_CHANNEL_DMA , 0 }, + { NV10_CHANNEL_DMA , 0 }, + { NV03_CHANNEL_DMA , 0 }, + {} + }; + struct { + struct nvif_chan_v0 chan; + char name[TASK_COMM_LEN+16]; } args; + struct nouveau_cli *cli = (void *)device->object.client; struct nouveau_channel *chan; - u32 size; - int ret; + const u64 plength = 0x10000; + const u64 ioffset = plength; + const u64 ilength = 0x02000; + char name[TASK_COMM_LEN]; + int cid, ret; + u64 size; + + cid = nvif_mclass(&device->object, hosts); + if (cid < 0) + return cid; + + if (hosts[cid].oclass < NV50_CHANNEL_GPFIFO) + size = plength; + else + size = ioffset + ilength; /* allocate dma push buffer */ - ret = nouveau_channel_prep(drm, device, 0x12000, &chan); + ret = nouveau_channel_prep(drm, device, size, &chan); *pchan = chan; if (ret) return ret; /* create channel object */ - do { - if (oclass[0] >= VOLTA_CHANNEL_GPFIFO_A) { - args.volta.version = 0; - args.volta.ilength = 0x02000; - args.volta.ioffset = 0x10000 + chan->push.addr; - args.volta.runlist = runlist; - args.volta.vmm = nvif_handle(&chan->vmm->vmm.object); - args.volta.priv = priv; - size = sizeof(args.volta); - } else - if (oclass[0] >= KEPLER_CHANNEL_GPFIFO_A) { - args.kepler.version = 0; - args.kepler.ilength = 0x02000; - args.kepler.ioffset = 0x10000 + chan->push.addr; - args.kepler.runlist = runlist; - args.kepler.vmm = nvif_handle(&chan->vmm->vmm.object); - args.kepler.priv = priv; - size = sizeof(args.kepler); - } else - if (oclass[0] >= FERMI_CHANNEL_GPFIFO) { - args.fermi.version = 0; - args.fermi.ilength = 0x02000; - args.fermi.ioffset = 0x10000 + chan->push.addr; - args.fermi.vmm = nvif_handle(&chan->vmm->vmm.object); - size = sizeof(args.fermi); - } else { - args.nv50.version = 0; - args.nv50.ilength = 0x02000; - args.nv50.ioffset = 0x10000 + chan->push.addr; - args.nv50.pushbuf = nvif_handle(&chan->push.ctxdma); - args.nv50.vmm = nvif_handle(&chan->vmm->vmm.object); - size = sizeof(args.nv50); - } - - ret = nvif_object_ctor(&device->object, "abi16ChanUser", 0, - *oclass++, &args, size, &chan->user); - if (ret == 0) { - if (chan->user.oclass >= VOLTA_CHANNEL_GPFIFO_A) { - chan->chid = args.volta.chid; - chan->inst = args.volta.inst; - chan->token = args.volta.token; - } else - if (chan->user.oclass >= KEPLER_CHANNEL_GPFIFO_A) { - chan->chid = args.kepler.chid; - chan->inst = args.kepler.inst; - } else - if (chan->user.oclass >= FERMI_CHANNEL_GPFIFO) { - chan->chid = args.fermi.chid; - } else { - chan->chid = args.nv50.chid; - } + args.chan.version = 0; + args.chan.namelen = sizeof(args.name); + args.chan.runlist = __ffs64(runm); + args.chan.runq = 0; + args.chan.priv = priv; + args.chan.devm = BIT(0); + if (hosts[cid].oclass < NV50_CHANNEL_GPFIFO) { + args.chan.vmm = 0; + args.chan.ctxdma = nvif_handle(&chan->push.ctxdma); + args.chan.offset = chan->push.addr; + args.chan.length = 0; + } else { + args.chan.vmm = nvif_handle(&chan->vmm->vmm.object); + if (hosts[cid].oclass < FERMI_CHANNEL_GPFIFO) + args.chan.ctxdma = nvif_handle(&chan->push.ctxdma); + else + args.chan.ctxdma = 0; + args.chan.offset = ioffset + chan->push.addr; + args.chan.length = ilength; + } + args.chan.huserd = 0; + args.chan.ouserd = 0; + + /* allocate userd */ + if (hosts[cid].oclass >= VOLTA_CHANNEL_GPFIFO_A) { + ret = nvif_mem_ctor(&cli->mmu, "abi16ChanUSERD", NVIF_CLASS_MEM_GF100, + NVIF_MEM_VRAM | NVIF_MEM_COHERENT | NVIF_MEM_MAPPABLE, + 0, PAGE_SIZE, NULL, 0, &chan->mem_userd); + if (ret) return ret; - } - } while (*oclass); - nouveau_channel_del(pchan); - return ret; -} + args.chan.huserd = nvif_handle(&chan->mem_userd.object); + args.chan.ouserd = 0; -static int -nouveau_channel_dma(struct nouveau_drm *drm, struct nvif_device *device, - struct nouveau_channel **pchan) -{ - static const u16 oclasses[] = { NV40_CHANNEL_DMA, - NV17_CHANNEL_DMA, - NV10_CHANNEL_DMA, - NV03_CHANNEL_DMA, - 0 }; - const u16 *oclass = oclasses; - struct nv03_channel_dma_v0 args; - struct nouveau_channel *chan; - int ret; + chan->userd = &chan->mem_userd.object; + } else { + chan->userd = &chan->user; + } - /* allocate dma push buffer */ - ret = nouveau_channel_prep(drm, device, 0x10000, &chan); - *pchan = chan; - if (ret) - return ret; + get_task_comm(name, current); + snprintf(args.name, sizeof(args.name), "%s[%d]", name, task_pid_nr(current)); - /* create channel object */ - args.version = 0; - args.pushbuf = nvif_handle(&chan->push.ctxdma); - args.offset = chan->push.addr; - - do { - ret = nvif_object_ctor(&device->object, "abi16ChanUser", 0, - *oclass++, &args, sizeof(args), - &chan->user); - if (ret == 0) { - chan->chid = args.chid; - return ret; - } - } while (ret && *oclass); + ret = nvif_object_ctor(&device->object, "abi16ChanUser", 0, hosts[cid].oclass, + &args, sizeof(args), &chan->user); + if (ret) { + nouveau_channel_del(pchan); + return ret; + } - nouveau_channel_del(pchan); - return ret; + chan->runlist = args.chan.runlist; + chan->chid = args.chan.chid; + chan->inst = args.chan.inst; + chan->token = args.chan.token; + return 0; } static int @@ -385,18 +362,24 @@ nouveau_channel_init(struct nouveau_channel *chan, u32 vram, u32 gart) struct nv_dma_v0 args = {}; int ret, i; - ret = nvif_object_map(&chan->user, NULL, 0); + ret = nvif_object_map(chan->userd, NULL, 0); if (ret) return ret; - if (chan->user.oclass >= FERMI_CHANNEL_GPFIFO && - chan->user.oclass < AMPERE_CHANNEL_GPFIFO_B) { - ret = nvif_notify_ctor(&chan->user, "abi16ChanKilled", - nouveau_channel_killed, - true, NV906F_V0_NTFY_KILLED, - NULL, 0, 0, &chan->kill); + if (chan->user.oclass >= FERMI_CHANNEL_GPFIFO) { + struct { + struct nvif_event_v0 base; + struct nvif_chan_event_v0 host; + } args; + + args.host.version = 0; + args.host.type = NVIF_CHAN_EVENT_V0_KILLED; + + ret = nvif_event_ctor(&chan->user, "abi16ChanKilled", chan->chid, + nouveau_channel_killed, false, + &args.base, sizeof(args), &chan->kill); if (ret == 0) - ret = nvif_notify_get(&chan->kill); + ret = nvif_event_allow(&chan->kill); if (ret) { NV_ERROR(drm, "Failed to request channel kill " "notification: %d\n", ret); @@ -503,24 +486,18 @@ nouveau_channel_init(struct nouveau_channel *chan, u32 vram, u32 gart) int nouveau_channel_new(struct nouveau_drm *drm, struct nvif_device *device, - u32 arg0, u32 arg1, bool priv, - struct nouveau_channel **pchan) + bool priv, u64 runm, u32 vram, u32 gart, struct nouveau_channel **pchan) { struct nouveau_cli *cli = (void *)device->object.client; int ret; - /* hack until fencenv50 is fixed, and agp access relaxed */ - ret = nouveau_channel_ind(drm, device, arg0, priv, pchan); + ret = nouveau_channel_ctor(drm, device, priv, runm, pchan); if (ret) { - NV_PRINTK(dbg, cli, "ib channel create, %d\n", ret); - ret = nouveau_channel_dma(drm, device, pchan); - if (ret) { - NV_PRINTK(dbg, cli, "dma channel create, %d\n", ret); - return ret; - } + NV_PRINTK(dbg, cli, "channel create, %d\n", ret); + return ret; } - ret = nouveau_channel_init(*pchan, arg0, arg1); + ret = nouveau_channel_init(*pchan, vram, gart); if (ret) { NV_PRINTK(err, cli, "channel failed to initialise, %d\n", ret); nouveau_channel_del(pchan); @@ -534,6 +511,12 @@ nouveau_channel_new(struct nouveau_drm *drm, struct nvif_device *device, return ret; } +void +nouveau_channels_fini(struct nouveau_drm *drm) +{ + kfree(drm->runl); +} + int nouveau_channels_init(struct nouveau_drm *drm) { @@ -541,20 +524,53 @@ nouveau_channels_init(struct nouveau_drm *drm) struct nv_device_info_v1 m; struct { struct nv_device_info_v1_data channels; + struct nv_device_info_v1_data runlists; } v; } args = { .m.version = 1, .m.count = sizeof(args.v) / sizeof(args.v.channels), .v.channels.mthd = NV_DEVICE_HOST_CHANNELS, + .v.runlists.mthd = NV_DEVICE_HOST_RUNLISTS, }; struct nvif_object *device = &drm->client.device.object; - int ret; + int ret, i; ret = nvif_object_mthd(device, NV_DEVICE_V0_INFO, &args, sizeof(args)); - if (ret || args.v.channels.mthd == NV_DEVICE_INFO_INVALID) + if (ret || + args.v.runlists.mthd == NV_DEVICE_INFO_INVALID || !args.v.runlists.data || + args.v.channels.mthd == NV_DEVICE_INFO_INVALID) return -ENODEV; - drm->chan.nr = args.v.channels.data; - drm->chan.context_base = dma_fence_context_alloc(drm->chan.nr); + drm->chan_nr = drm->chan_total = args.v.channels.data; + drm->runl_nr = fls64(args.v.runlists.data); + drm->runl = kcalloc(drm->runl_nr, sizeof(*drm->runl), GFP_KERNEL); + if (!drm->runl) + return -ENOMEM; + + if (drm->chan_nr == 0) { + for (i = 0; i < drm->runl_nr; i++) { + if (!(args.v.runlists.data & BIT(i))) + continue; + + args.v.channels.mthd = NV_DEVICE_HOST_RUNLIST_CHANNELS; + args.v.channels.data = i; + + ret = nvif_object_mthd(device, NV_DEVICE_V0_INFO, &args, sizeof(args)); + if (ret || args.v.channels.mthd == NV_DEVICE_INFO_INVALID) + return -ENODEV; + + drm->runl[i].chan_nr = args.v.channels.data; + drm->runl[i].chan_id_base = drm->chan_total; + drm->runl[i].context_base = dma_fence_context_alloc(drm->runl[i].chan_nr); + + drm->chan_total += drm->runl[i].chan_nr; + } + } else { + drm->runl[0].context_base = dma_fence_context_alloc(drm->chan_nr); + for (i = 1; i < drm->runl_nr; i++) + drm->runl[i].context_base = drm->runl[0].context_base; + + } + return 0; } diff --git a/drivers/gpu/drm/nouveau/nouveau_chan.h b/drivers/gpu/drm/nouveau/nouveau_chan.h index 98ba9d27e6b4..e06a8ffed31a 100644 --- a/drivers/gpu/drm/nouveau/nouveau_chan.h +++ b/drivers/gpu/drm/nouveau/nouveau_chan.h @@ -2,7 +2,7 @@ #ifndef __NOUVEAU_CHAN_H__ #define __NOUVEAU_CHAN_H__ #include <nvif/object.h> -#include <nvif/notify.h> +#include <nvif/event.h> #include <nvif/push.h> struct nvif_device; @@ -16,6 +16,10 @@ struct nouveau_channel { struct nouveau_drm *drm; struct nouveau_vmm *vmm; + struct nvif_mem mem_userd; + struct nvif_object *userd; + + int runlist; int chid; u64 inst; u32 token; @@ -50,15 +54,15 @@ struct nouveau_channel { struct nvif_object user; - struct nvif_notify kill; + struct nvif_event kill; atomic_t killed; }; int nouveau_channels_init(struct nouveau_drm *); +void nouveau_channels_fini(struct nouveau_drm *); -int nouveau_channel_new(struct nouveau_drm *, struct nvif_device *, - u32 arg0, u32 arg1, bool priv, - struct nouveau_channel **); +int nouveau_channel_new(struct nouveau_drm *, struct nvif_device *, bool priv, u64 runm, + u32 vram, u32 gart, struct nouveau_channel **); void nouveau_channel_del(struct nouveau_channel **); int nouveau_channel_idle(struct nouveau_channel *); diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.c b/drivers/gpu/drm/nouveau/nouveau_connector.c index 1991bbb1d05c..086b66b60d91 100644 --- a/drivers/gpu/drm/nouveau/nouveau_connector.c +++ b/drivers/gpu/drm/nouveau/nouveau_connector.c @@ -47,8 +47,7 @@ #include "nouveau_crtc.h" #include <nvif/class.h> -#include <nvif/cl0046.h> -#include <nvif/event.h> +#include <nvif/if0011.h> struct drm_display_mode * nouveau_conn_native_mode(struct drm_connector *connector) @@ -396,7 +395,8 @@ static void nouveau_connector_destroy(struct drm_connector *connector) { struct nouveau_connector *nv_connector = nouveau_connector(connector); - nvif_notify_dtor(&nv_connector->hpd); + nvif_event_dtor(&nv_connector->irq); + nvif_event_dtor(&nv_connector->hpd); kfree(nv_connector->edid); drm_connector_unregister(connector); drm_connector_cleanup(connector); @@ -1162,39 +1162,38 @@ nouveau_connector_funcs_lvds = { }; void -nouveau_connector_hpd(struct drm_connector *connector) +nouveau_connector_hpd(struct nouveau_connector *nv_connector, u64 bits) { - struct nouveau_drm *drm = nouveau_drm(connector->dev); - u32 mask = drm_connector_mask(connector); + struct nouveau_drm *drm = nouveau_drm(nv_connector->base.dev); + u32 mask = drm_connector_mask(&nv_connector->base); + unsigned long flags; - mutex_lock(&drm->hpd_lock); + spin_lock_irqsave(&drm->hpd_lock, flags); if (!(drm->hpd_pending & mask)) { + nv_connector->hpd_pending |= bits; drm->hpd_pending |= mask; schedule_work(&drm->hpd_work); } - mutex_unlock(&drm->hpd_lock); + spin_unlock_irqrestore(&drm->hpd_lock, flags); } static int -nouveau_connector_hotplug(struct nvif_notify *notify) +nouveau_connector_irq(struct nvif_event *event, void *repv, u32 repc) { - struct nouveau_connector *nv_connector = - container_of(notify, typeof(*nv_connector), hpd); - struct drm_connector *connector = &nv_connector->base; - struct drm_device *dev = connector->dev; - struct nouveau_drm *drm = nouveau_drm(dev); - const struct nvif_notify_conn_rep_v0 *rep = notify->data; - bool plugged = (rep->mask != NVIF_NOTIFY_CONN_V0_UNPLUG); + struct nouveau_connector *nv_connector = container_of(event, typeof(*nv_connector), irq); - if (rep->mask & NVIF_NOTIFY_CONN_V0_IRQ) { - nouveau_dp_irq(drm, nv_connector); - return NVIF_NOTIFY_KEEP; - } + schedule_work(&nv_connector->irq_work); + return NVIF_EVENT_KEEP; +} - NV_DEBUG(drm, "%splugged %s\n", plugged ? "" : "un", connector->name); - nouveau_connector_hpd(connector); +static int +nouveau_connector_hotplug(struct nvif_event *event, void *repv, u32 repc) +{ + struct nouveau_connector *nv_connector = container_of(event, typeof(*nv_connector), hpd); + struct nvif_conn_event_v0 *rep = repv; - return NVIF_NOTIFY_KEEP; + nouveau_connector_hpd(nv_connector, rep->types); + return NVIF_EVENT_KEEP; } static ssize_t @@ -1290,6 +1289,7 @@ nouveau_connector_create(struct drm_device *dev, connector = &nv_connector->base; nv_connector->index = index; + INIT_WORK(&nv_connector->irq_work, nouveau_dp_irq); /* attempt to parse vbios connector type and hotplug gpio */ nv_connector->dcb = olddcb_conn(dev, index); @@ -1401,6 +1401,7 @@ nouveau_connector_create(struct drm_device *dev, drm_connector_init(dev, connector, funcs, type); drm_connector_helper_add(connector, &nouveau_connector_helper_funcs); + connector->polled = DRM_CONNECTOR_POLL_CONNECT; if (nv_connector->dcb && (disp->disp.conn_mask & BIT(nv_connector->index))) { ret = nvif_conn_ctor(&disp->disp, nv_connector->base.name, nv_connector->index, @@ -1409,6 +1410,25 @@ nouveau_connector_create(struct drm_device *dev, kfree(nv_connector); return ERR_PTR(ret); } + + ret = nvif_conn_event_ctor(&nv_connector->conn, "kmsHotplug", + nouveau_connector_hotplug, + NVIF_CONN_EVENT_V0_PLUG | NVIF_CONN_EVENT_V0_UNPLUG, + &nv_connector->hpd); + if (ret == 0) + connector->polled = DRM_CONNECTOR_POLL_HPD; + + if (nv_connector->aux.transfer) { + ret = nvif_conn_event_ctor(&nv_connector->conn, "kmsDpIrq", + nouveau_connector_irq, NVIF_CONN_EVENT_V0_IRQ, + &nv_connector->irq); + if (ret) { + nvif_event_dtor(&nv_connector->hpd); + nvif_conn_dtor(&nv_connector->conn); + kfree(nv_connector); + return ERR_PTR(ret); + } + } } connector->funcs->reset(connector); @@ -1452,21 +1472,6 @@ nouveau_connector_create(struct drm_device *dev, break; } - ret = nvif_notify_ctor(&disp->disp.object, "kmsHotplug", - nouveau_connector_hotplug, - true, NV04_DISP_NTFY_CONN, - &(struct nvif_notify_conn_req_v0) { - .mask = NVIF_NOTIFY_CONN_V0_ANY, - .conn = index, - }, - sizeof(struct nvif_notify_conn_req_v0), - sizeof(struct nvif_notify_conn_rep_v0), - &nv_connector->hpd); - if (ret) - connector->polled = DRM_CONNECTOR_POLL_CONNECT; - else - connector->polled = DRM_CONNECTOR_POLL_HPD; - drm_connector_register(connector); return connector; } diff --git a/drivers/gpu/drm/nouveau/nouveau_connector.h b/drivers/gpu/drm/nouveau/nouveau_connector.h index f4e17ff68bf9..35bcb541722b 100644 --- a/drivers/gpu/drm/nouveau/nouveau_connector.h +++ b/drivers/gpu/drm/nouveau/nouveau_connector.h @@ -27,7 +27,7 @@ #ifndef __NOUVEAU_CONNECTOR_H__ #define __NOUVEAU_CONNECTOR_H__ #include <nvif/conn.h> -#include <nvif/notify.h> +#include <nvif/event.h> #include <nvhw/class/cl507d.h> #include <nvhw/class/cl907d.h> @@ -124,7 +124,10 @@ struct nouveau_connector { u8 *dcb; struct nvif_conn conn; - struct nvif_notify hpd; + u64 hpd_pending; + struct nvif_event hpd; + struct nvif_event irq; + struct work_struct irq_work; struct drm_dp_aux aux; @@ -198,7 +201,7 @@ nouveau_crtc_connector_get(struct nouveau_crtc *nv_crtc) struct drm_connector * nouveau_connector_create(struct drm_device *, const struct dcb_output *); -void nouveau_connector_hpd(struct drm_connector *connector); +void nouveau_connector_hpd(struct nouveau_connector *, u64 bits); extern int nouveau_tv_disable; extern int nouveau_ignorelid; diff --git a/drivers/gpu/drm/nouveau/nouveau_crtc.h b/drivers/gpu/drm/nouveau/nouveau_crtc.h index 7f63be2ec35d..c717f664a7b8 100644 --- a/drivers/gpu/drm/nouveau/nouveau_crtc.h +++ b/drivers/gpu/drm/nouveau/nouveau_crtc.h @@ -26,16 +26,17 @@ #ifndef __NOUVEAU_CRTC_H__ #define __NOUVEAU_CRTC_H__ - #include <drm/drm_crtc.h> -#include <nvif/notify.h> +#include <nvif/head.h> +#include <nvif/event.h> struct nouveau_crtc { struct drm_crtc base; + struct nvif_head head; int index; - struct nvif_notify vblank; + struct nvif_event vblank; uint32_t dpms_saved_fp_control; uint32_t fp_users; diff --git a/drivers/gpu/drm/nouveau/nouveau_display.c b/drivers/gpu/drm/nouveau/nouveau_display.c index 2e97186090c8..ec3ffff487fc 100644 --- a/drivers/gpu/drm/nouveau/nouveau_display.c +++ b/drivers/gpu/drm/nouveau/nouveau_display.c @@ -35,15 +35,14 @@ #include <drm/drm_probe_helper.h> #include <drm/drm_vblank.h> -#include "nouveau_fbcon.h" #include "nouveau_crtc.h" #include "nouveau_gem.h" #include "nouveau_connector.h" #include "nv50_display.h" #include <nvif/class.h> -#include <nvif/cl0046.h> -#include <nvif/event.h> +#include <nvif/if0011.h> +#include <nvif/if0013.h> #include <dispnv50/crc.h> int @@ -52,7 +51,7 @@ nouveau_display_vblank_enable(struct drm_crtc *crtc) struct nouveau_crtc *nv_crtc; nv_crtc = nouveau_crtc(crtc); - nvif_notify_get(&nv_crtc->vblank); + nvif_event_allow(&nv_crtc->vblank); return 0; } @@ -63,7 +62,7 @@ nouveau_display_vblank_disable(struct drm_crtc *crtc) struct nouveau_crtc *nv_crtc; nv_crtc = nouveau_crtc(crtc); - nvif_notify_put(&nv_crtc->vblank); + nvif_event_block(&nv_crtc->vblank); } static inline int @@ -84,24 +83,20 @@ static bool nouveau_display_scanoutpos_head(struct drm_crtc *crtc, int *vpos, int *hpos, ktime_t *stime, ktime_t *etime) { - struct { - struct nv04_disp_mthd_v0 base; - struct nv04_disp_scanoutpos_v0 scan; - } args = { - .base.method = NV04_DISP_SCANOUTPOS, - .base.head = nouveau_crtc(crtc)->index, - }; - struct nouveau_display *disp = nouveau_display(crtc->dev); struct drm_vblank_crtc *vblank = &crtc->dev->vblank[drm_crtc_index(crtc)]; + struct nvif_head *head = &nouveau_crtc(crtc)->head; + struct nvif_head_scanoutpos_v0 args; int retry = 20; bool ret = false; + args.version = 0; + do { - ret = nvif_mthd(&disp->disp.object, 0, &args, sizeof(args)); + ret = nvif_mthd(&head->object, NVIF_HEAD_V0_SCANOUTPOS, &args, sizeof(args)); if (ret != 0) return false; - if (args.scan.vline) { + if (args.vline) { ret = true; break; } @@ -109,11 +104,10 @@ nouveau_display_scanoutpos_head(struct drm_crtc *crtc, int *vpos, int *hpos, if (retry) ndelay(vblank->linedur_ns); } while (retry--); - *hpos = args.scan.hline; - *vpos = calc(args.scan.vblanks, args.scan.vblanke, - args.scan.vtotal, args.scan.vline); - if (stime) *stime = ns_to_ktime(args.scan.time[0]); - if (etime) *etime = ns_to_ktime(args.scan.time[1]); + *hpos = args.hline; + *vpos = calc(args.vblanks, args.vblanke, args.vtotal, args.vline); + if (stime) *stime = ns_to_ktime(args.time[0]); + if (etime) *etime = ns_to_ktime(args.time[1]); return ret; } @@ -397,7 +391,7 @@ nouveau_user_framebuffer_create(struct drm_device *dev, static const struct drm_mode_config_funcs nouveau_mode_config_funcs = { .fb_create = nouveau_user_framebuffer_create, - .output_poll_changed = nouveau_fbcon_output_poll_changed, + .output_poll_changed = drm_fb_helper_output_poll_changed, }; @@ -456,9 +450,9 @@ nouveau_display_hpd_resume(struct drm_device *dev) { struct nouveau_drm *drm = nouveau_drm(dev); - mutex_lock(&drm->hpd_lock); + spin_lock_irq(&drm->hpd_lock); drm->hpd_pending = ~0; - mutex_unlock(&drm->hpd_lock); + spin_unlock_irq(&drm->hpd_lock); schedule_work(&drm->hpd_work); } @@ -475,10 +469,10 @@ nouveau_display_hpd_work(struct work_struct *work) pm_runtime_get_sync(dev->dev); - mutex_lock(&drm->hpd_lock); + spin_lock_irq(&drm->hpd_lock); pending = drm->hpd_pending; drm->hpd_pending = 0; - mutex_unlock(&drm->hpd_lock); + spin_unlock_irq(&drm->hpd_lock); /* Nothing to do, exit early without updating the last busy counter */ if (!pending) @@ -488,14 +482,30 @@ nouveau_display_hpd_work(struct work_struct *work) drm_connector_list_iter_begin(dev, &conn_iter); nouveau_for_each_non_mst_connector_iter(connector, &conn_iter) { + struct nouveau_connector *nv_connector = nouveau_connector(connector); enum drm_connector_status old_status = connector->status; - u64 old_epoch_counter = connector->epoch_counter; + u64 bits, old_epoch_counter = connector->epoch_counter; if (!(pending & drm_connector_mask(connector))) continue; - connector->status = drm_helper_probe_detect(connector, NULL, - false); + spin_lock_irq(&drm->hpd_lock); + bits = nv_connector->hpd_pending; + nv_connector->hpd_pending = 0; + spin_unlock_irq(&drm->hpd_lock); + + drm_dbg_kms(dev, "[CONNECTOR:%d:%s] plug:%d unplug:%d irq:%d\n", + connector->base.id, connector->name, + !!(bits & NVIF_CONN_EVENT_V0_PLUG), + !!(bits & NVIF_CONN_EVENT_V0_UNPLUG), + !!(bits & NVIF_CONN_EVENT_V0_IRQ)); + + if (bits & NVIF_CONN_EVENT_V0_IRQ) { + if (nouveau_dp_link_check(nv_connector)) + continue; + } + + connector->status = drm_helper_probe_detect(connector, NULL, false); if (old_epoch_counter == connector->epoch_counter) continue; @@ -573,7 +583,8 @@ nouveau_display_init(struct drm_device *dev, bool resume, bool runtime) drm_connector_list_iter_begin(dev, &conn_iter); nouveau_for_each_non_mst_connector_iter(connector, &conn_iter) { struct nouveau_connector *conn = nouveau_connector(connector); - nvif_notify_get(&conn->hpd); + nvif_event_allow(&conn->hpd); + nvif_event_allow(&conn->irq); } drm_connector_list_iter_end(&conn_iter); @@ -608,7 +619,8 @@ nouveau_display_fini(struct drm_device *dev, bool suspend, bool runtime) drm_connector_list_iter_begin(dev, &conn_iter); nouveau_for_each_non_mst_connector_iter(connector, &conn_iter) { struct nouveau_connector *conn = nouveau_connector(connector); - nvif_notify_put(&conn->hpd); + nvif_event_block(&conn->irq); + nvif_event_block(&conn->hpd); } drm_connector_list_iter_end(&conn_iter); @@ -732,7 +744,7 @@ nouveau_display_create(struct drm_device *dev) } INIT_WORK(&drm->hpd_work, nouveau_display_hpd_work); - mutex_init(&drm->hpd_lock); + spin_lock_init(&drm->hpd_lock); #ifdef CONFIG_ACPI drm->acpi_nb.notifier_call = nouveau_display_acpi_ntfy; register_acpi_notifier(&drm->acpi_nb); @@ -766,8 +778,7 @@ nouveau_display_destroy(struct drm_device *dev) nvif_disp_dtor(&disp->disp); - nouveau_drm(dev)->display = NULL; - mutex_destroy(&drm->hpd_lock); + drm->display = NULL; kfree(disp); } @@ -776,6 +787,9 @@ nouveau_display_suspend(struct drm_device *dev, bool runtime) { struct nouveau_display *disp = nouveau_display(dev); + /* Disable console. */ + drm_fb_helper_set_suspend_unlocked(dev->fb_helper, true); + if (drm_drv_uses_atomic_modeset(dev)) { if (!runtime) { disp->suspend = drm_atomic_helper_suspend(dev); @@ -803,8 +817,10 @@ nouveau_display_resume(struct drm_device *dev, bool runtime) drm_atomic_helper_resume(dev, disp->suspend); disp->suspend = NULL; } - return; } + + /* Enable console. */ + drm_fb_helper_set_suspend_unlocked(dev->fb_helper, false); } int diff --git a/drivers/gpu/drm/nouveau/nouveau_dma.c b/drivers/gpu/drm/nouveau/nouveau_dma.c index ddb75d80bc53..b90cac6d5772 100644 --- a/drivers/gpu/drm/nouveau/nouveau_dma.c +++ b/drivers/gpu/drm/nouveau/nouveau_dma.c @@ -42,9 +42,9 @@ READ_GET(struct nouveau_channel *chan, uint64_t *prev_get, int *timeout) { uint64_t val; - val = nvif_rd32(&chan->user, chan->user_get); + val = nvif_rd32(chan->userd, chan->user_get); if (chan->user_get_hi) - val |= (uint64_t)nvif_rd32(&chan->user, chan->user_get_hi) << 32; + val |= (uint64_t)nvif_rd32(chan->userd, chan->user_get_hi) << 32; /* reset counter as long as GET is still advancing, this is * to avoid misdetecting a GPU lockup if the GPU happens to @@ -86,7 +86,7 @@ nv50_dma_push(struct nouveau_channel *chan, u64 offset, int length) /* Flush writes. */ nouveau_bo_rd32(pb, 0); - nvif_wr32(&chan->user, 0x8c, chan->dma.ib_put); + nvif_wr32(chan->userd, 0x8c, chan->dma.ib_put); if (user->func && user->func->doorbell) user->func->doorbell(user, chan->token); chan->dma.ib_free--; @@ -98,7 +98,7 @@ nv50_dma_push_wait(struct nouveau_channel *chan, int count) uint32_t cnt = 0, prev_get = 0; while (chan->dma.ib_free < count) { - uint32_t get = nvif_rd32(&chan->user, 0x88); + uint32_t get = nvif_rd32(chan->userd, 0x88); if (get != prev_get) { prev_get = get; cnt = 0; diff --git a/drivers/gpu/drm/nouveau/nouveau_dp.c b/drivers/gpu/drm/nouveau/nouveau_dp.c index 20db8ea1a0ba..e00876f92aee 100644 --- a/drivers/gpu/drm/nouveau/nouveau_dp.c +++ b/drivers/gpu/drm/nouveau/nouveau_dp.c @@ -29,8 +29,7 @@ #include "nouveau_encoder.h" #include "nouveau_crtc.h" -#include <nvif/class.h> -#include <nvif/cl5070.h> +#include <nvif/if0011.h> MODULE_PARM_DESC(mst, "Enable DisplayPort multi-stream (default: enabled)"); static int nouveau_mst = 1; @@ -140,12 +139,17 @@ nouveau_dp_detect(struct nouveau_connector *nv_connector, * TODO: look into checking this before probing I2C to detect DVI/HDMI */ hpd = nvif_conn_hpd_status(&nv_connector->conn); - if (hpd == NVIF_CONN_HPD_STATUS_NOT_PRESENT) + if (hpd == NVIF_CONN_HPD_STATUS_NOT_PRESENT) { + nvif_outp_dp_aux_pwr(&nv_encoder->outp, false); goto out; + } + nvif_outp_dp_aux_pwr(&nv_encoder->outp, true); status = nouveau_dp_probe_dpcd(nv_connector, nv_encoder); - if (status == connector_status_disconnected) + if (status == connector_status_disconnected) { + nvif_outp_dp_aux_pwr(&nv_encoder->outp, false); goto out; + } /* If we're in MST mode, we're done here */ if (mstm && mstm->can_mst && mstm->is_mst) { @@ -193,6 +197,7 @@ nouveau_dp_detect(struct nouveau_connector *nv_connector, ret = NOUVEAU_DP_MST; goto out; } else if (ret != 0) { + nvif_outp_dp_aux_pwr(&nv_encoder->outp, false); goto out; } } @@ -206,14 +211,28 @@ out: return ret; } -void nouveau_dp_irq(struct nouveau_drm *drm, - struct nouveau_connector *nv_connector) +bool +nouveau_dp_link_check(struct nouveau_connector *nv_connector) +{ + struct nouveau_encoder *nv_encoder = find_encoder(&nv_connector->base, DCB_OUTPUT_DP); + + if (!nv_encoder || nv_encoder->outp.or.id < 0) + return true; + + return nvif_outp_dp_retrain(&nv_encoder->outp) == 0; +} + +void +nouveau_dp_irq(struct work_struct *work) { + struct nouveau_connector *nv_connector = + container_of(work, typeof(*nv_connector), irq_work); struct drm_connector *connector = &nv_connector->base; struct nouveau_encoder *outp = find_encoder(connector, DCB_OUTPUT_DP); + struct nouveau_drm *drm = nouveau_drm(outp->base.base.dev); struct nv50_mstm *mstm; + u64 hpd = 0; int ret; - bool send_hpd = false; if (!outp) return; @@ -225,14 +244,14 @@ void nouveau_dp_irq(struct nouveau_drm *drm, if (mstm && mstm->is_mst) { if (!nv50_mstm_service(drm, nv_connector, mstm)) - send_hpd = true; + hpd |= NVIF_CONN_EVENT_V0_UNPLUG; } else { drm_dp_cec_irq(&nv_connector->aux); if (nouveau_dp_has_sink_count(connector, outp)) { ret = drm_dp_read_sink_count(&nv_connector->aux); if (ret != outp->dp.sink_count) - send_hpd = true; + hpd |= NVIF_CONN_EVENT_V0_PLUG; if (ret >= 0) outp->dp.sink_count = ret; } @@ -240,8 +259,7 @@ void nouveau_dp_irq(struct nouveau_drm *drm, mutex_unlock(&outp->dp.hpd_irq_lock); - if (send_hpd) - nouveau_connector_hpd(connector); + nouveau_connector_hpd(nv_connector, NVIF_CONN_EVENT_V0_IRQ | hpd); } /* TODO: diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c index fd99ec0f4257..80f154b6adab 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drm.c +++ b/drivers/gpu/drm/nouveau/nouveau_drm.c @@ -33,6 +33,8 @@ #include <drm/drm_aperture.h> #include <drm/drm_crtc_helper.h> #include <drm/drm_drv.h> +#include <drm/drm_fb_helper.h> +#include <drm/drm_fbdev_generic.h> #include <drm/drm_gem_ttm_helper.h> #include <drm/drm_ioctl.h> #include <drm/drm_vblank.h> @@ -49,7 +51,6 @@ #include <nvif/class.h> #include <nvif/cl0002.h> -#include <nvif/cla06f.h> #include "nouveau_drv.h" #include "nouveau_dma.h" @@ -62,7 +63,6 @@ #include "nouveau_bios.h" #include "nouveau_ioctl.h" #include "nouveau_abi16.h" -#include "nouveau_fbcon.h" #include "nouveau_fence.h" #include "nouveau_debugfs.h" #include "nouveau_usif.h" @@ -316,28 +316,19 @@ static void nouveau_accel_ce_init(struct nouveau_drm *drm) { struct nvif_device *device = &drm->client.device; + u64 runm; int ret = 0; /* Allocate channel that has access to a (preferably async) copy * engine, to use for TTM buffer moves. */ - if (device->info.family >= NV_DEVICE_INFO_V0_KEPLER) { - ret = nouveau_channel_new(drm, device, - nvif_fifo_runlist_ce(device), 0, - true, &drm->cechan); - } else - if (device->info.chipset >= 0xa3 && - device->info.chipset != 0xaa && - device->info.chipset != 0xac) { - /* Prior to Kepler, there's only a single runlist, so all - * engines can be accessed from any channel. - * - * We still want to use a separate channel though. - */ - ret = nouveau_channel_new(drm, device, NvDmaFB, NvDmaTT, false, - &drm->cechan); + runm = nvif_fifo_runlist_ce(device); + if (!runm) { + NV_DEBUG(drm, "no ce runlist\n"); + return; } + ret = nouveau_channel_new(drm, device, false, runm, NvDmaFB, NvDmaTT, &drm->cechan); if (ret) NV_ERROR(drm, "failed to create ce channel, %d\n", ret); } @@ -355,23 +346,17 @@ static void nouveau_accel_gr_init(struct nouveau_drm *drm) { struct nvif_device *device = &drm->client.device; - u32 arg0, arg1; + u64 runm; int ret; - if (device->info.family >= NV_DEVICE_INFO_V0_AMPERE) - return; - /* Allocate channel that has access to the graphics engine. */ - if (device->info.family >= NV_DEVICE_INFO_V0_KEPLER) { - arg0 = nvif_fifo_runlist(device, NV_DEVICE_HOST_RUNLIST_ENGINES_GR); - arg1 = 1; - } else { - arg0 = NvDmaFB; - arg1 = NvDmaTT; + runm = nvif_fifo_runlist(device, NV_DEVICE_HOST_RUNLIST_ENGINES_GR); + if (!runm) { + NV_DEBUG(drm, "no gr runlist\n"); + return; } - ret = nouveau_channel_new(drm, device, arg0, arg1, false, - &drm->channel); + ret = nouveau_channel_new(drm, device, false, runm, NvDmaFB, NvDmaTT, &drm->channel); if (ret) { NV_ERROR(drm, "failed to create kernel channel, %d\n", ret); nouveau_accel_gr_fini(drm); @@ -436,6 +421,7 @@ nouveau_accel_fini(struct nouveau_drm *drm) nouveau_accel_gr_fini(drm); if (drm->fence) nouveau_fence(drm)->dtor(drm); + nouveau_channels_fini(drm); } static void @@ -485,6 +471,7 @@ nouveau_accel_init(struct nouveau_drm *drm) case PASCAL_CHANNEL_GPFIFO_A: case VOLTA_CHANNEL_GPFIFO_A: case TURING_CHANNEL_GPFIFO_A: + case AMPERE_CHANNEL_GPFIFO_A: case AMPERE_CHANNEL_GPFIFO_B: ret = nvc0_fence_create(drm); break; @@ -611,7 +598,6 @@ nouveau_drm_device_init(struct drm_device *dev) nouveau_hwmon_init(dev); nouveau_svm_init(drm); nouveau_dmem_init(drm); - nouveau_fbcon_init(dev); nouveau_led_init(dev); if (nouveau_pmops_runtime()) { @@ -655,7 +641,6 @@ nouveau_drm_device_fini(struct drm_device *dev) } nouveau_led_fini(dev); - nouveau_fbcon_fini(dev); nouveau_dmem_fini(drm); nouveau_svm_fini(drm); nouveau_hwmon_fini(dev); @@ -809,6 +794,11 @@ static int nouveau_drm_probe(struct pci_dev *pdev, if (ret) goto fail_drm_dev_init; + if (nouveau_drm(drm_dev)->client.device.info.ram_size <= 32 * 1024 * 1024) + drm_fbdev_generic_setup(drm_dev, 8); + else + drm_fbdev_generic_setup(drm_dev, 32); + quirk_broken_nv_runpm(pdev); return 0; @@ -865,8 +855,6 @@ nouveau_do_suspend(struct drm_device *dev, bool runtime) nouveau_led_suspend(dev); if (dev->mode_config.num_crtc) { - NV_DEBUG(drm, "suspending console...\n"); - nouveau_fbcon_set_suspend(dev, 1); NV_DEBUG(drm, "suspending display...\n"); ret = nouveau_display_suspend(dev, runtime); if (ret) @@ -940,8 +928,6 @@ nouveau_do_resume(struct drm_device *dev, bool runtime) if (dev->mode_config.num_crtc) { NV_DEBUG(drm, "resuming display...\n"); nouveau_display_resume(dev, runtime); - NV_DEBUG(drm, "resuming console...\n"); - nouveau_fbcon_set_suspend(dev, 0); } nouveau_led_resume(dev); @@ -1296,7 +1282,6 @@ static void nouveau_display_options(void) DRM_DEBUG_DRIVER("... tv_disable : %d\n", nouveau_tv_disable); DRM_DEBUG_DRIVER("... ignorelid : %d\n", nouveau_ignorelid); DRM_DEBUG_DRIVER("... duallink : %d\n", nouveau_duallink); - DRM_DEBUG_DRIVER("... nofbaccel : %d\n", nouveau_nofbaccel); DRM_DEBUG_DRIVER("... config : %s\n", nouveau_config); DRM_DEBUG_DRIVER("... debug : %s\n", nouveau_debug); DRM_DEBUG_DRIVER("... noaccel : %d\n", nouveau_noaccel); diff --git a/drivers/gpu/drm/nouveau/nouveau_drv.h b/drivers/gpu/drm/nouveau/nouveau_drv.h index 84df5ddae4d0..d6dd07bfa64a 100644 --- a/drivers/gpu/drm/nouveau/nouveau_drv.h +++ b/drivers/gpu/drm/nouveau/nouveau_drv.h @@ -78,11 +78,6 @@ enum nouveau_drm_object_route { NVDRM_OBJECT_ANY = NVIF_IOCTL_V0_OWNER_ANY, }; -enum nouveau_drm_notify_route { - NVDRM_NOTIFY_NVIF = 0, - NVDRM_NOTIFY_USIF -}; - enum nouveau_drm_handle { NVDRM_CHAN = 0xcccc0000, /* |= client chid */ NVDRM_NVSW = 0x55550000, @@ -179,16 +174,19 @@ struct nouveau_drm { void *fence; /* Global channel management. */ + int chan_total; /* Number of channels across all runlists. */ + int chan_nr; /* 0 if per-runlist CHIDs. */ + int runl_nr; struct { - int nr; + int chan_nr; + int chan_id_base; u64 context_base; - } chan; + } *runl; /* context for accelerated drm-internal operations */ struct nouveau_channel *cechan; struct nouveau_channel *channel; struct nvkm_gpuobj *notify; - struct nouveau_fbdev *fbcon; struct nvif_object ntfy; /* nv10-nv40 tiling regions */ @@ -201,10 +199,8 @@ struct nouveau_drm { struct nvbios vbios; struct nouveau_display *display; struct work_struct hpd_work; - struct mutex hpd_lock; + spinlock_t hpd_lock; u32 hpd_pending; - struct work_struct fbcon_work; - int fbcon_new_state; #ifdef CONFIG_ACPI struct notifier_block acpi_nb; #endif diff --git a/drivers/gpu/drm/nouveau/nouveau_encoder.h b/drivers/gpu/drm/nouveau/nouveau_encoder.h index b72e5783a00f..70c1ad6c4d9d 100644 --- a/drivers/gpu/drm/nouveau/nouveau_encoder.h +++ b/drivers/gpu/drm/nouveau/nouveau_encoder.h @@ -48,7 +48,6 @@ struct nouveau_encoder { struct dcb_output *dcb; struct nvif_outp outp; int or; - int link; struct i2c_adapter *i2c; struct nvkm_i2c_aux *aux; @@ -142,8 +141,8 @@ enum nouveau_dp_status { }; int nouveau_dp_detect(struct nouveau_connector *, struct nouveau_encoder *); -void nouveau_dp_irq(struct nouveau_drm *drm, - struct nouveau_connector *nv_connector); +bool nouveau_dp_link_check(struct nouveau_connector *); +void nouveau_dp_irq(struct work_struct *); enum drm_mode_status nv50_dp_mode_valid(struct drm_connector *, struct nouveau_encoder *, const struct drm_display_mode *, diff --git a/drivers/gpu/drm/nouveau/nouveau_fbcon.h b/drivers/gpu/drm/nouveau/nouveau_fbcon.h deleted file mode 100644 index 1796d8824580..000000000000 --- a/drivers/gpu/drm/nouveau/nouveau_fbcon.h +++ /dev/null @@ -1,82 +0,0 @@ -/* - * Copyright (C) 2008 Maarten Maathuis. - * All Rights Reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining - * a copy of this software and associated documentation files (the - * "Software"), to deal in the Software without restriction, including - * without limitation the rights to use, copy, modify, merge, publish, - * distribute, sublicense, and/or sell copies of the Software, and to - * permit persons to whom the Software is furnished to do so, subject to - * the following conditions: - * - * The above copyright notice and this permission notice (including the - * next paragraph) shall be included in all copies or substantial - * portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. - * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE - * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION - * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION - * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * - */ - -#ifndef __NOUVEAU_FBCON_H__ -#define __NOUVEAU_FBCON_H__ - -#include <drm/drm_fb_helper.h> - -#include "nouveau_display.h" - -struct nouveau_vma; - -struct nouveau_fbdev { - struct drm_fb_helper helper; /* must be first */ - unsigned int saved_flags; - struct nvif_object surf2d; - struct nvif_object clip; - struct nvif_object rop; - struct nvif_object patt; - struct nvif_object gdi; - struct nvif_object blit; - struct nvif_object twod; - struct nouveau_vma *vma; - - struct mutex hotplug_lock; - bool hotplug_waiting; -}; - -void nouveau_fbcon_restore(void); - -int nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region); -int nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect); -int nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image); -int nv04_fbcon_accel_init(struct fb_info *info); - -int nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect); -int nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region); -int nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image); -int nv50_fbcon_accel_init(struct fb_info *info); - -int nvc0_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect); -int nvc0_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region); -int nvc0_fbcon_imageblit(struct fb_info *info, const struct fb_image *image); -int nvc0_fbcon_accel_init(struct fb_info *info); - -void nouveau_fbcon_gpu_lockup(struct fb_info *info); - -int nouveau_fbcon_init(struct drm_device *dev); -void nouveau_fbcon_fini(struct drm_device *dev); -void nouveau_fbcon_set_suspend(struct drm_device *dev, int state); -void nouveau_fbcon_accel_save_disable(struct drm_device *dev); -void nouveau_fbcon_accel_restore(struct drm_device *dev); - -void nouveau_fbcon_output_poll_changed(struct drm_device *dev); -void nouveau_fbcon_hotplug_resume(struct nouveau_fbdev *fbcon); -extern int nouveau_nofbaccel; - -#endif /* __NV50_FBCON_H__ */ - diff --git a/drivers/gpu/drm/nouveau/nouveau_fence.c b/drivers/gpu/drm/nouveau/nouveau_fence.c index abcac7db4347..ee5e9d40c166 100644 --- a/drivers/gpu/drm/nouveau/nouveau_fence.c +++ b/drivers/gpu/drm/nouveau/nouveau_fence.c @@ -29,9 +29,7 @@ #include <linux/sched/signal.h> #include <trace/events/dma_fence.h> -#include <nvif/cl826e.h> -#include <nvif/notify.h> -#include <nvif/event.h> +#include <nvif/if0020.h> #include "nouveau_drv.h" #include "nouveau_dma.h" @@ -79,10 +77,6 @@ nouveau_local_fence(struct dma_fence *fence, struct nouveau_drm *drm) fence->ops != &nouveau_fence_ops_uevent) return NULL; - if (fence->context < drm->chan.context_base || - fence->context >= drm->chan.context_base + drm->chan.nr) - return NULL; - return from_fence(fence); } @@ -90,8 +84,9 @@ void nouveau_fence_context_kill(struct nouveau_fence_chan *fctx, int error) { struct nouveau_fence *fence; + unsigned long flags; - spin_lock_irq(&fctx->lock); + spin_lock_irqsave(&fctx->lock, flags); while (!list_empty(&fctx->pending)) { fence = list_entry(fctx->pending.next, typeof(*fence), head); @@ -99,16 +94,16 @@ nouveau_fence_context_kill(struct nouveau_fence_chan *fctx, int error) dma_fence_set_error(&fence->base, error); if (nouveau_fence_signal(fence)) - nvif_notify_put(&fctx->notify); + nvif_event_block(&fctx->event); } - spin_unlock_irq(&fctx->lock); + spin_unlock_irqrestore(&fctx->lock, flags); } void nouveau_fence_context_del(struct nouveau_fence_chan *fctx) { nouveau_fence_context_kill(fctx, 0); - nvif_notify_dtor(&fctx->notify); + nvif_event_dtor(&fctx->event); fctx->dead = 1; /* @@ -150,12 +145,11 @@ nouveau_fence_update(struct nouveau_channel *chan, struct nouveau_fence_chan *fc } static int -nouveau_fence_wait_uevent_handler(struct nvif_notify *notify) +nouveau_fence_wait_uevent_handler(struct nvif_event *event, void *repv, u32 repc) { - struct nouveau_fence_chan *fctx = - container_of(notify, typeof(*fctx), notify); + struct nouveau_fence_chan *fctx = container_of(event, typeof(*fctx), event); unsigned long flags; - int ret = NVIF_NOTIFY_KEEP; + int ret = NVIF_EVENT_KEEP; spin_lock_irqsave(&fctx->lock, flags); if (!list_empty(&fctx->pending)) { @@ -165,7 +159,7 @@ nouveau_fence_wait_uevent_handler(struct nvif_notify *notify) fence = list_entry(fctx->pending.next, typeof(*fence), head); chan = rcu_dereference_protected(fence->channel, lockdep_is_held(&fctx->lock)); if (nouveau_fence_update(chan, fctx)) - ret = NVIF_NOTIFY_DROP; + ret = NVIF_EVENT_DROP; } spin_unlock_irqrestore(&fctx->lock, flags); @@ -177,12 +171,16 @@ nouveau_fence_context_new(struct nouveau_channel *chan, struct nouveau_fence_cha { struct nouveau_fence_priv *priv = (void*)chan->drm->fence; struct nouveau_cli *cli = (void *)chan->user.client; + struct { + struct nvif_event_v0 base; + struct nvif_chan_event_v0 host; + } args; int ret; INIT_LIST_HEAD(&fctx->flip); INIT_LIST_HEAD(&fctx->pending); spin_lock_init(&fctx->lock); - fctx->context = chan->drm->chan.context_base + chan->chid; + fctx->context = chan->drm->runl[chan->runlist].context_base + chan->chid; if (chan == chan->drm->cechan) strcpy(fctx->name, "copy engine channel"); @@ -195,13 +193,12 @@ nouveau_fence_context_new(struct nouveau_channel *chan, struct nouveau_fence_cha if (!priv->uevent) return; - ret = nvif_notify_ctor(&chan->user, "fenceNonStallIntr", - nouveau_fence_wait_uevent_handler, - false, NV826E_V0_NTFY_NON_STALL_INTERRUPT, - &(struct nvif_notify_uevent_req) { }, - sizeof(struct nvif_notify_uevent_req), - sizeof(struct nvif_notify_uevent_rep), - &fctx->notify); + args.host.version = 0; + args.host.type = NVIF_CHAN_EVENT_V0_NON_STALL_INTR; + + ret = nvif_event_ctor(&chan->user, "fenceNonStallIntr", (chan->runlist << 16) | chan->chid, + nouveau_fence_wait_uevent_handler, false, + &args.base, sizeof(args), &fctx->event); WARN_ON(ret); } @@ -230,7 +227,7 @@ nouveau_fence_emit(struct nouveau_fence *fence, struct nouveau_channel *chan) spin_lock_irq(&fctx->lock); if (nouveau_fence_update(chan, fctx)) - nvif_notify_put(&fctx->notify); + nvif_event_block(&fctx->event); list_add_tail(&fence->head, &fctx->pending); spin_unlock_irq(&fctx->lock); @@ -254,7 +251,7 @@ nouveau_fence_done(struct nouveau_fence *fence) spin_lock_irqsave(&fctx->lock, flags); chan = rcu_dereference_protected(fence->channel, lockdep_is_held(&fctx->lock)); if (chan && nouveau_fence_update(chan, fctx)) - nvif_notify_put(&fctx->notify); + nvif_event_block(&fctx->event); spin_unlock_irqrestore(&fctx->lock, flags); } return dma_fence_is_signaled(&fence->base); @@ -505,13 +502,13 @@ static bool nouveau_fence_enable_signaling(struct dma_fence *f) bool ret; if (!fctx->notify_ref++) - nvif_notify_get(&fctx->notify); + nvif_event_allow(&fctx->event); ret = nouveau_fence_no_signaling(f); if (ret) set_bit(DMA_FENCE_FLAG_USER_BITS, &fence->base.flags); else if (!--fctx->notify_ref) - nvif_notify_put(&fctx->notify); + nvif_event_block(&fctx->event); return ret; } diff --git a/drivers/gpu/drm/nouveau/nouveau_fence.h b/drivers/gpu/drm/nouveau/nouveau_fence.h index 4887caa69c65..0ca2bc85adf6 100644 --- a/drivers/gpu/drm/nouveau/nouveau_fence.h +++ b/drivers/gpu/drm/nouveau/nouveau_fence.h @@ -3,7 +3,7 @@ #define __NOUVEAU_FENCE_H__ #include <linux/dma-fence.h> -#include <nvif/notify.h> +#include <nvif/event.h> struct nouveau_drm; struct nouveau_bo; @@ -44,7 +44,7 @@ struct nouveau_fence_chan { u32 context; char name[32]; - struct nvif_notify notify; + struct nvif_event event; int notify_ref, dead; }; diff --git a/drivers/gpu/drm/nouveau/nouveau_nvif.c b/drivers/gpu/drm/nouveau/nouveau_nvif.c index df0fe58ca3ab..1d49ebdfd5dc 100644 --- a/drivers/gpu/drm/nouveau/nouveau_nvif.c +++ b/drivers/gpu/drm/nouveau/nouveau_nvif.c @@ -27,12 +27,10 @@ ******************************************************************************/ #include <core/client.h> -#include <core/notify.h> #include <core/ioctl.h> #include <nvif/client.h> #include <nvif/driver.h> -#include <nvif/notify.h> #include <nvif/event.h> #include <nvif/ioctl.h> @@ -72,10 +70,23 @@ nvkm_client_suspend(void *priv) } static int +nvkm_client_event(u64 token, void *repv, u32 repc) +{ + struct nvif_object *object = (void *)(unsigned long)token; + struct nvif_event *event = container_of(object, typeof(*event), object); + + if (event->func(event, repv, repc) == NVIF_EVENT_KEEP) + return NVKM_EVENT_KEEP; + + return NVKM_EVENT_DROP; +} + +static int nvkm_client_driver_init(const char *name, u64 device, const char *cfg, const char *dbg, void **ppriv) { - return nvkm_client_new(name, device, cfg, dbg, nvif_notify, (struct nvkm_client **)ppriv); + return nvkm_client_new(name, device, cfg, dbg, nvkm_client_event, + (struct nvkm_client **)ppriv); } const struct nvif_driver diff --git a/drivers/gpu/drm/nouveau/nouveau_svm.c b/drivers/gpu/drm/nouveau/nouveau_svm.c index 31a5b81ee9fc..a74ba8d84ba7 100644 --- a/drivers/gpu/drm/nouveau/nouveau_svm.c +++ b/drivers/gpu/drm/nouveau/nouveau_svm.c @@ -24,7 +24,7 @@ #include "nouveau_chan.h" #include "nouveau_dmem.h" -#include <nvif/notify.h> +#include <nvif/event.h> #include <nvif/object.h> #include <nvif/vmm.h> @@ -51,7 +51,8 @@ struct nouveau_svm { u32 putaddr; u32 get; u32 put; - struct nvif_notify notify; + struct nvif_event notify; + struct work_struct work; struct nouveau_svm_fault { u64 inst; @@ -711,13 +712,11 @@ out: return ret; } -static int -nouveau_svm_fault(struct nvif_notify *notify) +static void +nouveau_svm_fault(struct work_struct *work) { - struct nouveau_svm_fault_buffer *buffer = - container_of(notify, typeof(*buffer), notify); - struct nouveau_svm *svm = - container_of(buffer, typeof(*svm), buffer[buffer->id]); + struct nouveau_svm_fault_buffer *buffer = container_of(work, typeof(*buffer), work); + struct nouveau_svm *svm = container_of(buffer, typeof(*svm), buffer[buffer->id]); struct nvif_object *device = &svm->drm->client.device.object; struct nouveau_svmm *svmm; struct { @@ -737,7 +736,7 @@ nouveau_svm_fault(struct nvif_notify *notify) buffer->put = nvif_rd32(device, buffer->putaddr); buffer->get = nvif_rd32(device, buffer->getaddr); if (buffer->get == buffer->put) - return NVIF_NOTIFY_KEEP; + return; } buffer->fault_nr = 0; @@ -881,7 +880,15 @@ nouveau_svm_fault(struct nvif_notify *notify) /* Issue fault replay to the GPU. */ if (replay) nouveau_svm_fault_replay(svm); - return NVIF_NOTIFY_KEEP; +} + +static int +nouveau_svm_event(struct nvif_event *event, void *argv, u32 argc) +{ + struct nouveau_svm_fault_buffer *buffer = container_of(event, typeof(*buffer), notify); + + schedule_work(&buffer->work); + return NVIF_EVENT_KEEP; } static struct nouveau_pfnmap_args * @@ -936,7 +943,9 @@ static void nouveau_svm_fault_buffer_fini(struct nouveau_svm *svm, int id) { struct nouveau_svm_fault_buffer *buffer = &svm->buffer[id]; - nvif_notify_put(&buffer->notify); + + nvif_event_block(&buffer->notify); + flush_work(&buffer->work); } static int @@ -944,10 +953,12 @@ nouveau_svm_fault_buffer_init(struct nouveau_svm *svm, int id) { struct nouveau_svm_fault_buffer *buffer = &svm->buffer[id]; struct nvif_object *device = &svm->drm->client.device.object; + buffer->get = nvif_rd32(device, buffer->getaddr); buffer->put = nvif_rd32(device, buffer->putaddr); SVM_DBG(svm, "get %08x put %08x (init)", buffer->get, buffer->put); - return nvif_notify_get(&buffer->notify); + + return nvif_event_allow(&buffer->notify); } static void @@ -956,15 +967,18 @@ nouveau_svm_fault_buffer_dtor(struct nouveau_svm *svm, int id) struct nouveau_svm_fault_buffer *buffer = &svm->buffer[id]; int i; + if (!nvif_object_constructed(&buffer->object)) + return; + + nouveau_svm_fault_buffer_fini(svm, id); + if (buffer->fault) { for (i = 0; buffer->fault[i] && i < buffer->entries; i++) kfree(buffer->fault[i]); kvfree(buffer->fault); } - nouveau_svm_fault_buffer_fini(svm, id); - - nvif_notify_dtor(&buffer->notify); + nvif_event_dtor(&buffer->notify); nvif_object_dtor(&buffer->object); } @@ -990,10 +1004,10 @@ nouveau_svm_fault_buffer_ctor(struct nouveau_svm *svm, s32 oclass, int id) buffer->entries = args.entries; buffer->getaddr = args.get; buffer->putaddr = args.put; + INIT_WORK(&buffer->work, nouveau_svm_fault); - ret = nvif_notify_ctor(&buffer->object, "svmFault", nouveau_svm_fault, - true, NVB069_V0_NTFY_FAULT, NULL, 0, 0, - &buffer->notify); + ret = nvif_event_ctor(&buffer->object, "svmFault", id, nouveau_svm_event, true, NULL, 0, + &buffer->notify); if (ret) return ret; diff --git a/drivers/gpu/drm/nouveau/nouveau_usif.c b/drivers/gpu/drm/nouveau/nouveau_usif.c index 36df6840c099..002d1479ba89 100644 --- a/drivers/gpu/drm/nouveau/nouveau_usif.c +++ b/drivers/gpu/drm/nouveau/nouveau_usif.c @@ -151,12 +151,6 @@ usif_ioctl(struct drm_file *filp, void __user *user, u32 argc) case NVIF_IOCTL_V0_NEW: ret = usif_object_new(filp, data, size, argv, argc, abi16); break; - case NVIF_IOCTL_V0_NTFY_NEW: - case NVIF_IOCTL_V0_NTFY_DEL: - case NVIF_IOCTL_V0_NTFY_GET: - case NVIF_IOCTL_V0_NTFY_PUT: - ret = -ENOSYS; - break; default: ret = nvif_client_ioctl(client, argv, argc); break; diff --git a/drivers/gpu/drm/nouveau/nouveau_vga.c b/drivers/gpu/drm/nouveau/nouveau_vga.c index 60cd8c0463df..789393b94291 100644 --- a/drivers/gpu/drm/nouveau/nouveau_vga.c +++ b/drivers/gpu/drm/nouveau/nouveau_vga.c @@ -7,7 +7,6 @@ #include "nouveau_drv.h" #include "nouveau_acpi.h" -#include "nouveau_fbcon.h" #include "nouveau_vga.h" static unsigned int diff --git a/drivers/gpu/drm/nouveau/nv04_fbcon.c b/drivers/gpu/drm/nouveau/nv04_fbcon.c deleted file mode 100644 index c30b8dacd86b..000000000000 --- a/drivers/gpu/drm/nouveau/nv04_fbcon.c +++ /dev/null @@ -1,259 +0,0 @@ -/* - * Copyright 2009 Ben Skeggs - * Copyright 2008 Stuart Bennett - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice (including the next - * paragraph) shall be included in all copies or substantial portions of the - * Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER - * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING - * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER - * DEALINGS IN THE SOFTWARE. - */ -#define NVIF_DEBUG_PRINT_DISABLE -#include "nouveau_drv.h" -#include "nouveau_dma.h" -#include "nouveau_fbcon.h" - -#include <nvif/push006c.h> - -int -nv04_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct nouveau_drm *drm = nouveau_drm(nfbdev->helper.dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_push *push = chan->chan.push; - int ret; - - ret = PUSH_WAIT(push, 4); - if (ret) - return ret; - - PUSH_NVSQ(push, NV05F, 0x0300, (region->sy << 16) | region->sx, - 0x0304, (region->dy << 16) | region->dx, - 0x0308, (region->height << 16) | region->width); - PUSH_KICK(push); - return 0; -} - -int -nv04_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct nouveau_drm *drm = nouveau_drm(nfbdev->helper.dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_push *push = chan->chan.push; - int ret; - - ret = PUSH_WAIT(push, 7); - if (ret) - return ret; - - PUSH_NVSQ(push, NV04A, 0x02fc, (rect->rop != ROP_COPY) ? 1 : 3); - if (info->fix.visual == FB_VISUAL_TRUECOLOR || - info->fix.visual == FB_VISUAL_DIRECTCOLOR) - PUSH_NVSQ(push, NV04A, 0x03fc, ((uint32_t *)info->pseudo_palette)[rect->color]); - else - PUSH_NVSQ(push, NV04A, 0x03fc, rect->color); - PUSH_NVSQ(push, NV04A, 0x0400, (rect->dx << 16) | rect->dy, - 0x0404, (rect->width << 16) | rect->height); - PUSH_KICK(push); - return 0; -} - -int -nv04_fbcon_imageblit(struct fb_info *info, const struct fb_image *image) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct nouveau_drm *drm = nouveau_drm(nfbdev->helper.dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_push *push = chan->chan.push; - uint32_t fg; - uint32_t bg; - uint32_t dsize; - uint32_t *data = (uint32_t *)image->data; - int ret; - - if (image->depth != 1) - return -ENODEV; - - ret = PUSH_WAIT(push, 8); - if (ret) - return ret; - - if (info->fix.visual == FB_VISUAL_TRUECOLOR || - info->fix.visual == FB_VISUAL_DIRECTCOLOR) { - fg = ((uint32_t *) info->pseudo_palette)[image->fg_color]; - bg = ((uint32_t *) info->pseudo_palette)[image->bg_color]; - } else { - fg = image->fg_color; - bg = image->bg_color; - } - - PUSH_NVSQ(push, NV04A, 0x0be4, (image->dy << 16) | (image->dx & 0xffff), - 0x0be8, ((image->dy + image->height) << 16) | - ((image->dx + image->width) & 0xffff), - 0x0bec, bg, - 0x0bf0, fg, - 0x0bf4, (image->height << 16) | ALIGN(image->width, 8), - 0x0bf8, (image->height << 16) | image->width, - 0x0bfc, (image->dy << 16) | (image->dx & 0xffff)); - - dsize = ALIGN(ALIGN(image->width, 8) * image->height, 32) >> 5; - while (dsize) { - int iter_len = dsize > 128 ? 128 : dsize; - - ret = PUSH_WAIT(push, iter_len + 1); - if (ret) - return ret; - - PUSH_NVSQ(push, NV04A, 0x0c00, data, iter_len); - data += iter_len; - dsize -= iter_len; - } - - PUSH_KICK(push); - return 0; -} - -int -nv04_fbcon_accel_init(struct fb_info *info) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct drm_device *dev = nfbdev->helper.dev; - struct nouveau_drm *drm = nouveau_drm(dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_device *device = &drm->client.device; - struct nvif_push *push = chan->chan.push; - struct nvkm_device *nvkm_device = nvxx_device(&drm->client.device); - resource_size_t fb_base = nvkm_device->func->resource_addr(nvkm_device, 1); - int surface_fmt, pattern_fmt, rect_fmt; - int ret; - - switch (info->var.bits_per_pixel) { - case 8: - surface_fmt = 1; - pattern_fmt = 3; - rect_fmt = 3; - break; - case 16: - surface_fmt = 4; - pattern_fmt = 1; - rect_fmt = 1; - break; - case 32: - switch (info->var.transp.length) { - case 0: /* depth 24 */ - case 8: /* depth 32 */ - break; - default: - return -EINVAL; - } - - surface_fmt = 6; - pattern_fmt = 3; - rect_fmt = 3; - break; - default: - return -EINVAL; - } - - ret = nvif_object_ctor(&chan->user, "fbconCtxSurf2d", 0x0062, - device->info.family >= NV_DEVICE_INFO_V0_CELSIUS ? - 0x0062 : 0x0042, NULL, 0, &nfbdev->surf2d); - if (ret) - return ret; - - ret = nvif_object_ctor(&chan->user, "fbconCtxClip", 0x0019, 0x0019, - NULL, 0, &nfbdev->clip); - if (ret) - return ret; - - ret = nvif_object_ctor(&chan->user, "fbconCtxRop", 0x0043, 0x0043, - NULL, 0, &nfbdev->rop); - if (ret) - return ret; - - ret = nvif_object_ctor(&chan->user, "fbconCtxPatt", 0x0044, 0x0044, - NULL, 0, &nfbdev->patt); - if (ret) - return ret; - - ret = nvif_object_ctor(&chan->user, "fbconGdiRectText", 0x004a, 0x004a, - NULL, 0, &nfbdev->gdi); - if (ret) - return ret; - - ret = nvif_object_ctor(&chan->user, "fbconImageBlit", 0x005f, - device->info.chipset >= 0x11 ? 0x009f : 0x005f, - NULL, 0, &nfbdev->blit); - if (ret) - return ret; - - if (PUSH_WAIT(push, 49 + (device->info.chipset >= 0x11 ? 4 : 0))) { - nouveau_fbcon_gpu_lockup(info); - return 0; - } - - PUSH_NVSQ(push, NV042, 0x0000, nfbdev->surf2d.handle); - PUSH_NVSQ(push, NV042, 0x0184, chan->vram.handle, - 0x0188, chan->vram.handle); - PUSH_NVSQ(push, NV042, 0x0300, surface_fmt, - 0x0304, info->fix.line_length | (info->fix.line_length << 16), - 0x0308, info->fix.smem_start - fb_base, - 0x030c, info->fix.smem_start - fb_base); - - PUSH_NVSQ(push, NV043, 0x0000, nfbdev->rop.handle); - PUSH_NVSQ(push, NV043, 0x0300, 0x55); - - PUSH_NVSQ(push, NV044, 0x0000, nfbdev->patt.handle); - PUSH_NVSQ(push, NV044, 0x0300, pattern_fmt, -#ifdef __BIG_ENDIAN - 0x0304, 2, -#else - 0x0304, 1, -#endif - 0x0308, 0, - 0x030c, 1, - 0x0310, ~0, - 0x0314, ~0, - 0x0318, ~0, - 0x031c, ~0); - - PUSH_NVSQ(push, NV019, 0x0000, nfbdev->clip.handle); - PUSH_NVSQ(push, NV019, 0x0300, 0, - 0x0304, (info->var.yres_virtual << 16) | info->var.xres_virtual); - - PUSH_NVSQ(push, NV05F, 0x0000, nfbdev->blit.handle); - PUSH_NVSQ(push, NV05F, 0x019c, nfbdev->surf2d.handle); - PUSH_NVSQ(push, NV05F, 0x02fc, 3); - if (nfbdev->blit.oclass == 0x009f) { - PUSH_NVSQ(push, NV09F, 0x0120, 0, - 0x0124, 1, - 0x0128, 2); - } - - PUSH_NVSQ(push, NV04A, 0x0000, nfbdev->gdi.handle); - PUSH_NVSQ(push, NV04A, 0x0198, nfbdev->surf2d.handle); - PUSH_NVSQ(push, NV04A, 0x0188, nfbdev->patt.handle, - 0x018c, nfbdev->rop.handle); - PUSH_NVSQ(push, NV04A, 0x0304, 1); - PUSH_NVSQ(push, NV04A, 0x0300, rect_fmt); - PUSH_NVSQ(push, NV04A, 0x02fc, 3); - - PUSH_KICK(push); - return 0; -} - diff --git a/drivers/gpu/drm/nouveau/nv50_fbcon.c b/drivers/gpu/drm/nouveau/nv50_fbcon.c deleted file mode 100644 index 71f92e4750f9..000000000000 --- a/drivers/gpu/drm/nouveau/nv50_fbcon.c +++ /dev/null @@ -1,299 +0,0 @@ -/* - * Copyright 2010 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#define NVIF_DEBUG_PRINT_DISABLE -#include "nouveau_drv.h" -#include "nouveau_dma.h" -#include "nouveau_fbcon.h" -#include "nouveau_vmm.h" - -#include <nvif/push206e.h> - -#include <nvhw/class/cl502d.h> - -int -nv50_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct nouveau_drm *drm = nouveau_drm(nfbdev->helper.dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_push *push = chan->chan.push; - u32 colour; - int ret; - - if (info->fix.visual == FB_VISUAL_TRUECOLOR || - info->fix.visual == FB_VISUAL_DIRECTCOLOR) - colour = ((uint32_t *)info->pseudo_palette)[rect->color]; - else - colour = rect->color; - - ret = PUSH_WAIT(push, rect->rop == ROP_COPY ? 7 : 11); - if (ret) - return ret; - - if (rect->rop != ROP_COPY) { - PUSH_MTHD(push, NV502D, SET_OPERATION, - NVDEF(NV502D, SET_OPERATION, V, ROP_AND)); - } - - PUSH_MTHD(push, NV502D, SET_RENDER_SOLID_PRIM_COLOR, colour); - - PUSH_MTHD(push, NV502D, RENDER_SOLID_PRIM_POINT_SET_X(0), rect->dx, - RENDER_SOLID_PRIM_POINT_Y(0), rect->dy, - RENDER_SOLID_PRIM_POINT_SET_X(1), rect->dx + rect->width, - RENDER_SOLID_PRIM_POINT_Y(1), rect->dy + rect->height); - - if (rect->rop != ROP_COPY) { - PUSH_MTHD(push, NV502D, SET_OPERATION, - NVDEF(NV502D, SET_OPERATION, V, SRCCOPY)); - } - - PUSH_KICK(push); - return 0; -} - -int -nv50_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct nouveau_drm *drm = nouveau_drm(nfbdev->helper.dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_push *push = chan->chan.push; - int ret; - - ret = PUSH_WAIT(push, 12); - if (ret) - return ret; - - PUSH_MTHD(push, NV502D, WAIT_FOR_IDLE, 0); - - PUSH_MTHD(push, NV502D, SET_PIXELS_FROM_MEMORY_DST_X0, region->dx, - SET_PIXELS_FROM_MEMORY_DST_Y0, region->dy, - SET_PIXELS_FROM_MEMORY_DST_WIDTH, region->width, - SET_PIXELS_FROM_MEMORY_DST_HEIGHT, region->height); - - PUSH_MTHD(push, NV502D, SET_PIXELS_FROM_MEMORY_SRC_X0_FRAC, 0, - SET_PIXELS_FROM_MEMORY_SRC_X0_INT, region->sx, - SET_PIXELS_FROM_MEMORY_SRC_Y0_FRAC, 0, - PIXELS_FROM_MEMORY_SRC_Y0_INT, region->sy); - PUSH_KICK(push); - return 0; -} - -int -nv50_fbcon_imageblit(struct fb_info *info, const struct fb_image *image) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct nouveau_drm *drm = nouveau_drm(nfbdev->helper.dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_push *push = chan->chan.push; - uint32_t dwords, *data = (uint32_t *)image->data; - uint32_t mask = ~(~0 >> (32 - info->var.bits_per_pixel)); - uint32_t *palette = info->pseudo_palette, bg, fg; - int ret; - - if (image->depth != 1) - return -ENODEV; - - if (info->fix.visual == FB_VISUAL_TRUECOLOR || - info->fix.visual == FB_VISUAL_DIRECTCOLOR) { - bg = palette[image->bg_color] | mask; - fg = palette[image->fg_color] | mask; - } else { - bg = image->bg_color; - fg = image->fg_color; - } - - ret = PUSH_WAIT(push, 11); - if (ret) - return ret; - - PUSH_MTHD(push, NV502D, SET_PIXELS_FROM_CPU_COLOR0, bg, - SET_PIXELS_FROM_CPU_COLOR1, fg); - - PUSH_MTHD(push, NV502D, SET_PIXELS_FROM_CPU_SRC_WIDTH, image->width, - SET_PIXELS_FROM_CPU_SRC_HEIGHT, image->height); - - PUSH_MTHD(push, NV502D, SET_PIXELS_FROM_CPU_DST_X0_FRAC, 0, - SET_PIXELS_FROM_CPU_DST_X0_INT, image->dx, - SET_PIXELS_FROM_CPU_DST_Y0_FRAC, 0, - SET_PIXELS_FROM_CPU_DST_Y0_INT, image->dy); - - dwords = ALIGN(ALIGN(image->width, 8) * image->height, 32) >> 5; - while (dwords) { - int count = dwords > 2047 ? 2047 : dwords; - - ret = PUSH_WAIT(push, count + 1); - if (ret) - return ret; - - dwords -= count; - - PUSH_NINC(push, NV502D, PIXELS_FROM_CPU_DATA, data, count); - data += count; - } - - PUSH_KICK(push); - return 0; -} - -int -nv50_fbcon_accel_init(struct fb_info *info) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct drm_device *dev = nfbdev->helper.dev; - struct nouveau_drm *drm = nouveau_drm(dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_push *push = chan->chan.push; - int ret, format; - - switch (info->var.bits_per_pixel) { - case 8: - format = NV502D_SET_DST_FORMAT_V_Y8; - break; - case 15: - format = NV502D_SET_DST_FORMAT_V_X1R5G5B5; - break; - case 16: - format = NV502D_SET_DST_FORMAT_V_R5G6B5; - break; - case 32: - switch (info->var.transp.length) { - case 0: /* depth 24 */ - case 8: /* depth 32, just use 24.. */ - format = NV502D_SET_DST_FORMAT_V_X8R8G8B8; - break; - case 2: /* depth 30 */ - format = NV502D_SET_DST_FORMAT_V_A2B10G10R10; - break; - default: - return -EINVAL; - } - break; - default: - return -EINVAL; - } - - ret = nvif_object_ctor(&chan->user, "fbconTwoD", 0x502d, 0x502d, - NULL, 0, &nfbdev->twod); - if (ret) - return ret; - - ret = PUSH_WAIT(push, 56); - if (ret) { - nouveau_fbcon_gpu_lockup(info); - return ret; - } - - PUSH_MTHD(push, NV502D, SET_OBJECT, nfbdev->twod.handle); - PUSH_MTHD(push, NV502D, SET_DST_CONTEXT_DMA, chan->vram.handle, - SET_SRC_CONTEXT_DMA, chan->vram.handle, - SET_SEMAPHORE_CONTEXT_DMA, chan->vram.handle); - - PUSH_MTHD(push, NV502D, SET_DST_FORMAT, - NVVAL(NV502D, SET_DST_FORMAT, V, format), - - SET_DST_MEMORY_LAYOUT, - NVDEF(NV502D, SET_DST_MEMORY_LAYOUT, V, PITCH)); - - PUSH_MTHD(push, NV502D, SET_DST_PITCH, info->fix.line_length, - SET_DST_WIDTH, info->var.xres_virtual, - SET_DST_HEIGHT, info->var.yres_virtual, - - SET_DST_OFFSET_UPPER, - NVVAL(NV502D, SET_DST_OFFSET_UPPER, V, upper_32_bits(nfbdev->vma->addr)), - - SET_DST_OFFSET_LOWER, - NVVAL(NV502D, SET_DST_OFFSET_LOWER, V, lower_32_bits(nfbdev->vma->addr))); - - PUSH_MTHD(push, NV502D, SET_SRC_FORMAT, - NVVAL(NV502D, SET_SRC_FORMAT, V, format), - - SET_SRC_MEMORY_LAYOUT, - NVDEF(NV502D, SET_SRC_MEMORY_LAYOUT, V, PITCH)); - - PUSH_MTHD(push, NV502D, SET_SRC_PITCH, info->fix.line_length, - SET_SRC_WIDTH, info->var.xres_virtual, - SET_SRC_HEIGHT, info->var.yres_virtual, - - SET_SRC_OFFSET_UPPER, - NVVAL(NV502D, SET_SRC_OFFSET_UPPER, V, upper_32_bits(nfbdev->vma->addr)), - - SET_SRC_OFFSET_LOWER, - NVVAL(NV502D, SET_SRC_OFFSET_LOWER, V, lower_32_bits(nfbdev->vma->addr))); - - PUSH_MTHD(push, NV502D, SET_CLIP_ENABLE, - NVDEF(NV502D, SET_CLIP_ENABLE, V, FALSE)); - - PUSH_MTHD(push, NV502D, SET_ROP, - NVVAL(NV502D, SET_ROP, V, 0x55)); - - PUSH_MTHD(push, NV502D, SET_OPERATION, - NVDEF(NV502D, SET_OPERATION, V, SRCCOPY)); - - PUSH_MTHD(push, NV502D, SET_MONOCHROME_PATTERN_COLOR_FORMAT, - NVDEF(NV502D, SET_MONOCHROME_PATTERN_COLOR_FORMAT, V, A8R8G8B8), - - SET_MONOCHROME_PATTERN_FORMAT, - NVDEF(NV502D, SET_MONOCHROME_PATTERN_FORMAT, V, LE_M1)); - - PUSH_MTHD(push, NV502D, RENDER_SOLID_PRIM_MODE, - NVDEF(NV502D, RENDER_SOLID_PRIM_MODE, V, RECTS), - - SET_RENDER_SOLID_PRIM_COLOR_FORMAT, - NVVAL(NV502D, SET_RENDER_SOLID_PRIM_COLOR_FORMAT, V, format)); - - PUSH_MTHD(push, NV502D, SET_PIXELS_FROM_CPU_DATA_TYPE, - NVDEF(NV502D, SET_PIXELS_FROM_CPU_DATA_TYPE, V, INDEX), - - SET_PIXELS_FROM_CPU_COLOR_FORMAT, - NVVAL(NV502D, SET_PIXELS_FROM_CPU_COLOR_FORMAT, V, format), - - SET_PIXELS_FROM_CPU_INDEX_FORMAT, - NVDEF(NV502D, SET_PIXELS_FROM_CPU_INDEX_FORMAT, V, I1), - - SET_PIXELS_FROM_CPU_MONO_FORMAT, - NVDEF(NV502D, SET_PIXELS_FROM_CPU_MONO_FORMAT, V, CGA6_M1), - - SET_PIXELS_FROM_CPU_WRAP, - NVDEF(NV502D, SET_PIXELS_FROM_CPU_WRAP, V, WRAP_BYTE)); - - PUSH_MTHD(push, NV502D, SET_PIXELS_FROM_CPU_MONO_OPACITY, - NVDEF(NV502D, SET_PIXELS_FROM_CPU_MONO_OPACITY, V, OPAQUE)); - - PUSH_MTHD(push, NV502D, SET_PIXELS_FROM_CPU_DX_DU_FRAC, 0, - SET_PIXELS_FROM_CPU_DX_DU_INT, 1, - SET_PIXELS_FROM_CPU_DY_DV_FRAC, 0, - SET_PIXELS_FROM_CPU_DY_DV_INT, 1); - - PUSH_MTHD(push, NV502D, SET_PIXELS_FROM_MEMORY_SAFE_OVERLAP, - NVDEF(NV502D, SET_PIXELS_FROM_MEMORY_SAFE_OVERLAP, V, TRUE)); - - PUSH_MTHD(push, NV502D, SET_PIXELS_FROM_MEMORY_DU_DX_FRAC, 0, - SET_PIXELS_FROM_MEMORY_DU_DX_INT, 1, - SET_PIXELS_FROM_MEMORY_DV_DY_FRAC, 0, - SET_PIXELS_FROM_MEMORY_DV_DY_INT, 1); - PUSH_KICK(push); - return 0; -} - diff --git a/drivers/gpu/drm/nouveau/nv84_fence.c b/drivers/gpu/drm/nouveau/nv84_fence.c index c3526a8622e3..812b8c62eeba 100644 --- a/drivers/gpu/drm/nouveau/nv84_fence.c +++ b/drivers/gpu/drm/nouveau/nv84_fence.c @@ -76,12 +76,18 @@ nv84_fence_sync32(struct nouveau_channel *chan, u64 virtual, u32 sequence) return ret; } +static inline u32 +nv84_fence_chid(struct nouveau_channel *chan) +{ + return chan->drm->runl[chan->runlist].chan_id_base + chan->chid; +} + static int nv84_fence_emit(struct nouveau_fence *fence) { struct nouveau_channel *chan = fence->channel; struct nv84_fence_chan *fctx = chan->fence; - u64 addr = fctx->vma->addr + chan->chid * 16; + u64 addr = fctx->vma->addr + nv84_fence_chid(chan) * 16; return fctx->base.emit32(chan, addr, fence->base.seqno); } @@ -91,7 +97,7 @@ nv84_fence_sync(struct nouveau_fence *fence, struct nouveau_channel *prev, struct nouveau_channel *chan) { struct nv84_fence_chan *fctx = chan->fence; - u64 addr = fctx->vma->addr + prev->chid * 16; + u64 addr = fctx->vma->addr + nv84_fence_chid(prev) * 16; return fctx->base.sync32(chan, addr, fence->base.seqno); } @@ -100,7 +106,7 @@ static u32 nv84_fence_read(struct nouveau_channel *chan) { struct nv84_fence_priv *priv = chan->drm->fence; - return nouveau_bo_rd32(priv->bo, chan->chid * 16/4); + return nouveau_bo_rd32(priv->bo, nv84_fence_chid(chan) * 16/4); } static void @@ -109,7 +115,7 @@ nv84_fence_context_del(struct nouveau_channel *chan) struct nv84_fence_priv *priv = chan->drm->fence; struct nv84_fence_chan *fctx = chan->fence; - nouveau_bo_wr32(priv->bo, chan->chid * 16 / 4, fctx->base.sequence); + nouveau_bo_wr32(priv->bo, nv84_fence_chid(chan) * 16 / 4, fctx->base.sequence); mutex_lock(&priv->mutex); nouveau_vma_del(&fctx->vma); mutex_unlock(&priv->mutex); @@ -152,9 +158,9 @@ nv84_fence_suspend(struct nouveau_drm *drm) struct nv84_fence_priv *priv = drm->fence; int i; - priv->suspend = vmalloc(array_size(sizeof(u32), drm->chan.nr)); + priv->suspend = vmalloc(array_size(sizeof(u32), drm->chan_total)); if (priv->suspend) { - for (i = 0; i < drm->chan.nr; i++) + for (i = 0; i < drm->chan_total; i++) priv->suspend[i] = nouveau_bo_rd32(priv->bo, i*4); } @@ -168,7 +174,7 @@ nv84_fence_resume(struct nouveau_drm *drm) int i; if (priv->suspend) { - for (i = 0; i < drm->chan.nr; i++) + for (i = 0; i < drm->chan_total; i++) nouveau_bo_wr32(priv->bo, i*4, priv->suspend[i]); vfree(priv->suspend); priv->suspend = NULL; @@ -204,7 +210,7 @@ nv84_fence_create(struct nouveau_drm *drm) priv->base.context_new = nv84_fence_context_new; priv->base.context_del = nv84_fence_context_del; - priv->base.uevent = drm->client.device.info.family < NV_DEVICE_INFO_V0_AMPERE; + priv->base.uevent = true; mutex_init(&priv->mutex); @@ -216,7 +222,7 @@ nv84_fence_create(struct nouveau_drm *drm) * will lose CPU/GPU coherency! */ NOUVEAU_GEM_DOMAIN_GART | NOUVEAU_GEM_DOMAIN_COHERENT; - ret = nouveau_bo_new(&drm->client, 16 * drm->chan.nr, 0, + ret = nouveau_bo_new(&drm->client, 16 * drm->chan_total, 0, domain, 0, 0, NULL, NULL, &priv->bo); if (ret == 0) { ret = nouveau_bo_pin(priv->bo, domain, false); diff --git a/drivers/gpu/drm/nouveau/nvc0_fbcon.c b/drivers/gpu/drm/nouveau/nvc0_fbcon.c deleted file mode 100644 index 7908a1a3e00f..000000000000 --- a/drivers/gpu/drm/nouveau/nvc0_fbcon.c +++ /dev/null @@ -1,297 +0,0 @@ -/* - * Copyright 2010 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#define NVIF_DEBUG_PRINT_DISABLE -#include "nouveau_drv.h" -#include "nouveau_dma.h" -#include "nouveau_fbcon.h" -#include "nouveau_vmm.h" - -#include <nvif/push906f.h> - -#include <nvhw/class/cl902d.h> - -int -nvc0_fbcon_fillrect(struct fb_info *info, const struct fb_fillrect *rect) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct nouveau_drm *drm = nouveau_drm(nfbdev->helper.dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_push *push = chan->chan.push; - u32 colour; - int ret; - - if (info->fix.visual == FB_VISUAL_TRUECOLOR || - info->fix.visual == FB_VISUAL_DIRECTCOLOR) - colour = ((uint32_t *)info->pseudo_palette)[rect->color]; - else - colour = rect->color; - - ret = PUSH_WAIT(push, rect->rop == ROP_COPY ? 7 : 9); - if (ret) - return ret; - - if (rect->rop != ROP_COPY) { - PUSH_IMMD(push, NV902D, SET_OPERATION, - NVDEF(NV902D, SET_OPERATION, V, ROP_AND)); - } - - PUSH_MTHD(push, NV902D, SET_RENDER_SOLID_PRIM_COLOR, colour); - - PUSH_MTHD(push, NV902D, RENDER_SOLID_PRIM_POINT_SET_X(0), rect->dx, - RENDER_SOLID_PRIM_POINT_Y(0), rect->dy, - RENDER_SOLID_PRIM_POINT_SET_X(1), rect->dx + rect->width, - RENDER_SOLID_PRIM_POINT_Y(1), rect->dy + rect->height); - - if (rect->rop != ROP_COPY) { - PUSH_IMMD(push, NV902D, SET_OPERATION, - NVDEF(NV902D, SET_OPERATION, V, SRCCOPY)); - } - - PUSH_KICK(push); - return 0; -} - -int -nvc0_fbcon_copyarea(struct fb_info *info, const struct fb_copyarea *region) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct nouveau_drm *drm = nouveau_drm(nfbdev->helper.dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_push *push = chan->chan.push; - int ret; - - ret = PUSH_WAIT(push, 11); - if (ret) - return ret; - - PUSH_IMMD(push, NV902D, WAIT_FOR_IDLE, 0); - - PUSH_MTHD(push, NV902D, SET_PIXELS_FROM_MEMORY_DST_X0, region->dx, - SET_PIXELS_FROM_MEMORY_DST_Y0, region->dy, - SET_PIXELS_FROM_MEMORY_DST_WIDTH, region->width, - SET_PIXELS_FROM_MEMORY_DST_HEIGHT, region->height); - - PUSH_MTHD(push, NV902D, SET_PIXELS_FROM_MEMORY_SRC_X0_FRAC, 0, - SET_PIXELS_FROM_MEMORY_SRC_X0_INT, region->sx, - SET_PIXELS_FROM_MEMORY_SRC_Y0_FRAC, 0, - PIXELS_FROM_MEMORY_SRC_Y0_INT, region->sy); - PUSH_KICK(push); - return 0; -} - -int -nvc0_fbcon_imageblit(struct fb_info *info, const struct fb_image *image) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct nouveau_drm *drm = nouveau_drm(nfbdev->helper.dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_push *push = chan->chan.push; - uint32_t dwords, *data = (uint32_t *)image->data; - uint32_t mask = ~(~0 >> (32 - info->var.bits_per_pixel)); - uint32_t *palette = info->pseudo_palette, bg, fg; - int ret; - - if (image->depth != 1) - return -ENODEV; - - if (info->fix.visual == FB_VISUAL_TRUECOLOR || - info->fix.visual == FB_VISUAL_DIRECTCOLOR) { - bg = palette[image->bg_color] | mask; - fg = palette[image->fg_color] | mask; - } else { - bg = image->bg_color; - fg = image->fg_color; - } - - ret = PUSH_WAIT(push, 11); - if (ret) - return ret; - - PUSH_MTHD(push, NV902D, SET_PIXELS_FROM_CPU_COLOR0, bg, - SET_PIXELS_FROM_CPU_COLOR1, fg); - - PUSH_MTHD(push, NV902D, SET_PIXELS_FROM_CPU_SRC_WIDTH, image->width, - SET_PIXELS_FROM_CPU_SRC_HEIGHT, image->height); - - PUSH_MTHD(push, NV902D, SET_PIXELS_FROM_CPU_DST_X0_FRAC, 0, - SET_PIXELS_FROM_CPU_DST_X0_INT, image->dx, - SET_PIXELS_FROM_CPU_DST_Y0_FRAC, 0, - SET_PIXELS_FROM_CPU_DST_Y0_INT, image->dy); - - dwords = ALIGN(ALIGN(image->width, 8) * image->height, 32) >> 5; - while (dwords) { - int count = dwords > 2047 ? 2047 : dwords; - - ret = PUSH_WAIT(push, count + 1); - if (ret) - return ret; - - dwords -= count; - - PUSH_NINC(push, NV902D, PIXELS_FROM_CPU_DATA, data, count); - data += count; - } - - PUSH_KICK(push); - return 0; -} - -int -nvc0_fbcon_accel_init(struct fb_info *info) -{ - struct nouveau_fbdev *nfbdev = info->par; - struct drm_device *dev = nfbdev->helper.dev; - struct nouveau_drm *drm = nouveau_drm(dev); - struct nouveau_channel *chan = drm->channel; - struct nvif_push *push = chan->chan.push; - int ret, format; - - ret = nvif_object_ctor(&chan->user, "fbconTwoD", 0x902d, 0x902d, - NULL, 0, &nfbdev->twod); - if (ret) - return ret; - - switch (info->var.bits_per_pixel) { - case 8: - format = NV902D_SET_DST_FORMAT_V_Y8; - break; - case 15: - format = NV902D_SET_DST_FORMAT_V_X1R5G5B5; - break; - case 16: - format = NV902D_SET_DST_FORMAT_V_R5G6B5; - break; - case 32: - switch (info->var.transp.length) { - case 0: /* depth 24 */ - case 8: /* depth 32, just use 24.. */ - format = NV902D_SET_DST_FORMAT_V_X8R8G8B8; - break; - case 2: /* depth 30 */ - format = NV902D_SET_DST_FORMAT_V_A2B10G10R10; - break; - default: - return -EINVAL; - } - break; - default: - return -EINVAL; - } - - ret = PUSH_WAIT(push, 52); - if (ret) { - WARN_ON(1); - nouveau_fbcon_gpu_lockup(info); - return ret; - } - - PUSH_MTHD(push, NV902D, SET_OBJECT, nfbdev->twod.handle); - - PUSH_MTHD(push, NV902D, SET_DST_FORMAT, - NVVAL(NV902D, SET_DST_FORMAT, V, format), - - SET_DST_MEMORY_LAYOUT, - NVDEF(NV902D, SET_DST_MEMORY_LAYOUT, V, PITCH)); - - PUSH_MTHD(push, NV902D, SET_DST_PITCH, info->fix.line_length, - SET_DST_WIDTH, info->var.xres_virtual, - SET_DST_HEIGHT, info->var.yres_virtual, - - SET_DST_OFFSET_UPPER, - NVVAL(NV902D, SET_DST_OFFSET_UPPER, V, upper_32_bits(nfbdev->vma->addr)), - - SET_DST_OFFSET_LOWER, - NVVAL(NV902D, SET_DST_OFFSET_LOWER, V, lower_32_bits(nfbdev->vma->addr))); - - PUSH_MTHD(push, NV902D, SET_SRC_FORMAT, - NVVAL(NV902D, SET_SRC_FORMAT, V, format), - - SET_SRC_MEMORY_LAYOUT, - NVDEF(NV902D, SET_SRC_MEMORY_LAYOUT, V, PITCH)); - - PUSH_MTHD(push, NV902D, SET_SRC_PITCH, info->fix.line_length, - SET_SRC_WIDTH, info->var.xres_virtual, - SET_SRC_HEIGHT, info->var.yres_virtual, - - SET_SRC_OFFSET_UPPER, - NVVAL(NV902D, SET_SRC_OFFSET_UPPER, V, upper_32_bits(nfbdev->vma->addr)), - - SET_SRC_OFFSET_LOWER, - NVVAL(NV902D, SET_SRC_OFFSET_LOWER, V, lower_32_bits(nfbdev->vma->addr))); - - PUSH_IMMD(push, NV902D, SET_CLIP_ENABLE, - NVDEF(NV902D, SET_CLIP_ENABLE, V, FALSE)); - - PUSH_IMMD(push, NV902D, SET_ROP, - NVVAL(NV902D, SET_ROP, V, 0x55)); - - PUSH_IMMD(push, NV902D, SET_OPERATION, - NVDEF(NV902D, SET_OPERATION, V, SRCCOPY)); - - PUSH_MTHD(push, NV902D, SET_MONOCHROME_PATTERN_COLOR_FORMAT, - NVDEF(NV902D, SET_MONOCHROME_PATTERN_COLOR_FORMAT, V, A8R8G8B8), - - SET_MONOCHROME_PATTERN_FORMAT, - NVDEF(NV902D, SET_MONOCHROME_PATTERN_FORMAT, V, LE_M1)); - - PUSH_MTHD(push, NV902D, RENDER_SOLID_PRIM_MODE, - NVDEF(NV902D, RENDER_SOLID_PRIM_MODE, V, RECTS), - - SET_RENDER_SOLID_PRIM_COLOR_FORMAT, - NVVAL(NV902D, SET_RENDER_SOLID_PRIM_COLOR_FORMAT, V, format)); - - PUSH_MTHD(push, NV902D, SET_PIXELS_FROM_CPU_DATA_TYPE, - NVDEF(NV902D, SET_PIXELS_FROM_CPU_DATA_TYPE, V, INDEX), - - SET_PIXELS_FROM_CPU_COLOR_FORMAT, - NVVAL(NV902D, SET_PIXELS_FROM_CPU_COLOR_FORMAT, V, format), - - SET_PIXELS_FROM_CPU_INDEX_FORMAT, - NVDEF(NV902D, SET_PIXELS_FROM_CPU_INDEX_FORMAT, V, I1), - - SET_PIXELS_FROM_CPU_MONO_FORMAT, - NVDEF(NV902D, SET_PIXELS_FROM_CPU_MONO_FORMAT, V, CGA6_M1), - - SET_PIXELS_FROM_CPU_WRAP, - NVDEF(NV902D, SET_PIXELS_FROM_CPU_WRAP, V, WRAP_BYTE)); - - PUSH_IMMD(push, NV902D, SET_PIXELS_FROM_CPU_MONO_OPACITY, - NVDEF(NV902D, SET_PIXELS_FROM_CPU_MONO_OPACITY, V, OPAQUE)); - - PUSH_MTHD(push, NV902D, SET_PIXELS_FROM_CPU_DX_DU_FRAC, 0, - SET_PIXELS_FROM_CPU_DX_DU_INT, 1, - SET_PIXELS_FROM_CPU_DY_DV_FRAC, 0, - SET_PIXELS_FROM_CPU_DY_DV_INT, 1); - - PUSH_IMMD(push, NV902D, SET_PIXELS_FROM_MEMORY_SAFE_OVERLAP, - NVDEF(NV902D, SET_PIXELS_FROM_MEMORY_SAFE_OVERLAP, V, TRUE)); - - PUSH_MTHD(push, NV902D, SET_PIXELS_FROM_MEMORY_DU_DX_FRAC, 0, - SET_PIXELS_FROM_MEMORY_DU_DX_INT, 1, - SET_PIXELS_FROM_MEMORY_DV_DY_FRAC, 0, - SET_PIXELS_FROM_MEMORY_DV_DY_INT, 1); - PUSH_KICK(push); - return 0; -} - diff --git a/drivers/gpu/drm/nouveau/nvif/Kbuild b/drivers/gpu/drm/nouveau/nvif/Kbuild index 6abc4bc42e35..b7963a39dd91 100644 --- a/drivers/gpu/drm/nouveau/nvif/Kbuild +++ b/drivers/gpu/drm/nouveau/nvif/Kbuild @@ -5,10 +5,11 @@ nvif-y += nvif/conn.o nvif-y += nvif/device.o nvif-y += nvif/disp.o nvif-y += nvif/driver.o +nvif-y += nvif/event.o nvif-y += nvif/fifo.o +nvif-y += nvif/head.o nvif-y += nvif/mem.o nvif-y += nvif/mmu.o -nvif-y += nvif/notify.o nvif-y += nvif/outp.o nvif-y += nvif/timer.o nvif-y += nvif/vmm.o diff --git a/drivers/gpu/drm/nouveau/nvif/conn.c b/drivers/gpu/drm/nouveau/nvif/conn.c index 4ce935d58c90..a3cf91aeae2d 100644 --- a/drivers/gpu/drm/nouveau/nvif/conn.c +++ b/drivers/gpu/drm/nouveau/nvif/conn.c @@ -27,6 +27,25 @@ #include <nvif/if0011.h> int +nvif_conn_event_ctor(struct nvif_conn *conn, const char *name, nvif_event_func func, u8 types, + struct nvif_event *event) +{ + struct { + struct nvif_event_v0 base; + struct nvif_conn_event_v0 conn; + } args; + int ret; + + args.conn.version = 0; + args.conn.types = types; + + ret = nvif_event_ctor_(&conn->object, name ?: "nvifConnHpd", nvif_conn_id(conn), + func, true, &args.base, sizeof(args), false, event); + NVIF_DEBUG(&conn->object, "[NEW EVENT:HPD types:%02x]", types); + return ret; +} + +int nvif_conn_hpd_status(struct nvif_conn *conn) { struct nvif_conn_hpd_status_v0 args; diff --git a/drivers/gpu/drm/nouveau/nvif/disp.c b/drivers/gpu/drm/nouveau/nvif/disp.c index 926b0c04b1e8..09915f2715af 100644 --- a/drivers/gpu/drm/nouveau/nvif/disp.c +++ b/drivers/gpu/drm/nouveau/nvif/disp.c @@ -72,9 +72,10 @@ nvif_disp_ctor(struct nvif_device *device, const char *name, s32 oclass, struct if (ret) return ret; - NVIF_DEBUG(&disp->object, "[NEW] conn_mask:%08x outp_mask:%08x", - args.conn_mask, args.outp_mask); + NVIF_DEBUG(&disp->object, "[NEW] conn_mask:%08x outp_mask:%08x head_mask:%08x", + args.conn_mask, args.outp_mask, args.head_mask); disp->conn_mask = args.conn_mask; disp->outp_mask = args.outp_mask; + disp->head_mask = args.head_mask; return 0; } diff --git a/drivers/gpu/drm/nouveau/nvif/event.c b/drivers/gpu/drm/nouveau/nvif/event.c new file mode 100644 index 000000000000..61ff4d6eba9f --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvif/event.c @@ -0,0 +1,81 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include <nvif/event.h> +#include <nvif/printf.h> + +#include <nvif/class.h> +#include <nvif/if000e.h> + +int +nvif_event_block(struct nvif_event *event) +{ + if (nvif_event_constructed(event)) { + int ret = nvif_mthd(&event->object, NVIF_EVENT_V0_BLOCK, NULL, 0); + NVIF_ERRON(ret, &event->object, "[BLOCK]"); + return ret; + } + return 0; +} + +int +nvif_event_allow(struct nvif_event *event) +{ + if (nvif_event_constructed(event)) { + int ret = nvif_mthd(&event->object, NVIF_EVENT_V0_ALLOW, NULL, 0); + NVIF_ERRON(ret, &event->object, "[ALLOW]"); + return ret; + } + return 0; +} + +void +nvif_event_dtor(struct nvif_event *event) +{ + nvif_object_dtor(&event->object); +} + +int +nvif_event_ctor_(struct nvif_object *parent, const char *name, u32 handle, nvif_event_func func, + bool wait, struct nvif_event_v0 *args, u32 argc, bool warn, + struct nvif_event *event) +{ + struct nvif_event_v0 _args; + int ret; + + if (!args) { + args = &_args; + argc = sizeof(_args); + } + + args->version = 0; + args->wait = wait; + + ret = nvif_object_ctor(parent, name ?: "nvifEvent", handle, + NVIF_CLASS_EVENT, args, argc, &event->object); + NVIF_ERRON(ret && warn, parent, "[NEW EVENT wait:%d size:%zd]", + args->wait, argc - sizeof(*args)); + if (ret) + return ret; + + event->func = func; + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nvif/head.c b/drivers/gpu/drm/nouveau/nvif/head.c new file mode 100644 index 000000000000..f00e01d232db --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvif/head.c @@ -0,0 +1,58 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include <nvif/head.h> +#include <nvif/disp.h> +#include <nvif/printf.h> + +#include <nvif/class.h> +#include <nvif/if0013.h> + +int +nvif_head_vblank_event_ctor(struct nvif_head *head, const char *name, nvif_event_func func, + bool wait, struct nvif_event *event) +{ + int ret = nvif_event_ctor(&head->object, name ?: "nvifHeadVBlank", nvif_head_id(head), + func, wait, NULL, 0, event); + NVIF_ERRON(ret, &head->object, "[NEW EVENT:VBLANK]"); + return ret; +} + +void +nvif_head_dtor(struct nvif_head *head) +{ + nvif_object_dtor(&head->object); +} + +int +nvif_head_ctor(struct nvif_disp *disp, const char *name, int id, struct nvif_head *head) +{ + struct nvif_head_v0 args; + int ret; + + args.version = 0; + args.id = id; + + ret = nvif_object_ctor(&disp->object, name ? name : "nvifHead", id, NVIF_CLASS_HEAD, + &args, sizeof(args), &head->object); + NVIF_ERRON(ret, &disp->object, "[NEW head id:%d]", args.id); + return ret; +} diff --git a/drivers/gpu/drm/nouveau/nvif/notify.c b/drivers/gpu/drm/nouveau/nvif/notify.c deleted file mode 100644 index 143c8dc6889e..000000000000 --- a/drivers/gpu/drm/nouveau/nvif/notify.c +++ /dev/null @@ -1,210 +0,0 @@ -/* - * Copyright 2014 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs <bskeggs@redhat.com> - */ - -#include <nvif/client.h> -#include <nvif/driver.h> -#include <nvif/notify.h> -#include <nvif/object.h> -#include <nvif/ioctl.h> -#include <nvif/event.h> - -static inline int -nvif_notify_put_(struct nvif_notify *notify) -{ - struct nvif_object *object = notify->object; - struct { - struct nvif_ioctl_v0 ioctl; - struct nvif_ioctl_ntfy_put_v0 ntfy; - } args = { - .ioctl.type = NVIF_IOCTL_V0_NTFY_PUT, - .ntfy.index = notify->index, - }; - - if (atomic_inc_return(¬ify->putcnt) != 1) - return 0; - - return nvif_object_ioctl(object, &args, sizeof(args), NULL); -} - -int -nvif_notify_put(struct nvif_notify *notify) -{ - if (likely(notify->object) && - test_and_clear_bit(NVIF_NOTIFY_USER, ¬ify->flags)) { - int ret = nvif_notify_put_(notify); - if (test_bit(NVIF_NOTIFY_WORK, ¬ify->flags)) - flush_work(¬ify->work); - return ret; - } - return 0; -} - -static inline int -nvif_notify_get_(struct nvif_notify *notify) -{ - struct nvif_object *object = notify->object; - struct { - struct nvif_ioctl_v0 ioctl; - struct nvif_ioctl_ntfy_get_v0 ntfy; - } args = { - .ioctl.type = NVIF_IOCTL_V0_NTFY_GET, - .ntfy.index = notify->index, - }; - - if (atomic_dec_return(¬ify->putcnt) != 0) - return 0; - - return nvif_object_ioctl(object, &args, sizeof(args), NULL); -} - -int -nvif_notify_get(struct nvif_notify *notify) -{ - if (likely(notify->object) && - !test_and_set_bit(NVIF_NOTIFY_USER, ¬ify->flags)) - return nvif_notify_get_(notify); - return 0; -} - -static inline int -nvif_notify_func(struct nvif_notify *notify, bool keep) -{ - int ret = notify->func(notify); - if (ret == NVIF_NOTIFY_KEEP || - !test_and_clear_bit(NVIF_NOTIFY_USER, ¬ify->flags)) { - if (!keep) - atomic_dec(¬ify->putcnt); - else - nvif_notify_get_(notify); - } - return ret; -} - -static void -nvif_notify_work(struct work_struct *work) -{ - struct nvif_notify *notify = container_of(work, typeof(*notify), work); - nvif_notify_func(notify, true); -} - -int -nvif_notify(const void *header, u32 length, const void *data, u32 size) -{ - struct nvif_notify *notify = NULL; - const union { - struct nvif_notify_rep_v0 v0; - } *args = header; - int ret = NVIF_NOTIFY_DROP; - - if (length == sizeof(args->v0) && args->v0.version == 0) { - if (WARN_ON(args->v0.route)) - return NVIF_NOTIFY_DROP; - notify = (void *)(unsigned long)args->v0.token; - } - - if (!WARN_ON(notify == NULL)) { - struct nvif_client *client = notify->object->client; - if (!WARN_ON(notify->size != size)) { - atomic_inc(¬ify->putcnt); - if (test_bit(NVIF_NOTIFY_WORK, ¬ify->flags)) { - memcpy((void *)notify->data, data, size); - schedule_work(¬ify->work); - return NVIF_NOTIFY_DROP; - } - notify->data = data; - ret = nvif_notify_func(notify, client->driver->keep); - notify->data = NULL; - } - } - - return ret; -} - -int -nvif_notify_dtor(struct nvif_notify *notify) -{ - struct nvif_object *object = notify->object; - struct { - struct nvif_ioctl_v0 ioctl; - struct nvif_ioctl_ntfy_del_v0 ntfy; - } args = { - .ioctl.type = NVIF_IOCTL_V0_NTFY_DEL, - .ntfy.index = notify->index, - }; - int ret = nvif_notify_put(notify); - if (ret >= 0 && object) { - ret = nvif_object_ioctl(object, &args, sizeof(args), NULL); - notify->object = NULL; - kfree((void *)notify->data); - } - return ret; -} - -int -nvif_notify_ctor(struct nvif_object *object, const char *name, - int (*func)(struct nvif_notify *), bool work, u8 event, - void *data, u32 size, u32 reply, struct nvif_notify *notify) -{ - struct { - struct nvif_ioctl_v0 ioctl; - struct nvif_ioctl_ntfy_new_v0 ntfy; - struct nvif_notify_req_v0 req; - } *args; - int ret = -ENOMEM; - - notify->object = object; - notify->name = name ? name : "nvifNotify"; - notify->flags = 0; - atomic_set(¬ify->putcnt, 1); - notify->func = func; - notify->data = NULL; - notify->size = reply; - if (work) { - INIT_WORK(¬ify->work, nvif_notify_work); - set_bit(NVIF_NOTIFY_WORK, ¬ify->flags); - notify->data = kmalloc(notify->size, GFP_KERNEL); - if (!notify->data) - goto done; - } - - if (!(args = kmalloc(sizeof(*args) + size, GFP_KERNEL))) - goto done; - args->ioctl.version = 0; - args->ioctl.type = NVIF_IOCTL_V0_NTFY_NEW; - args->ntfy.version = 0; - args->ntfy.event = event; - args->req.version = 0; - args->req.reply = notify->size; - args->req.route = 0; - args->req.token = (unsigned long)(void *)notify; - - memcpy(args->req.data, data, size); - ret = nvif_object_ioctl(object, args, sizeof(*args) + size, NULL); - notify->index = args->ntfy.index; - kfree(args); -done: - if (ret) - nvif_notify_dtor(notify); - return ret; -} diff --git a/drivers/gpu/drm/nouveau/nvif/outp.c b/drivers/gpu/drm/nouveau/nvif/outp.c index 7bfe91a8d6f9..7da39f1eae9f 100644 --- a/drivers/gpu/drm/nouveau/nvif/outp.c +++ b/drivers/gpu/drm/nouveau/nvif/outp.c @@ -24,7 +24,177 @@ #include <nvif/printf.h> #include <nvif/class.h> -#include <nvif/if0012.h> + +int +nvif_outp_dp_mst_vcpi(struct nvif_outp *outp, int head, + u8 start_slot, u8 num_slots, u16 pbn, u16 aligned_pbn) +{ + struct nvif_outp_dp_mst_vcpi_v0 args; + int ret; + + args.version = 0; + args.head = head; + args.start_slot = start_slot; + args.num_slots = num_slots; + args.pbn = pbn; + args.aligned_pbn = aligned_pbn; + + ret = nvif_object_mthd(&outp->object, NVIF_OUTP_V0_DP_MST_VCPI, &args, sizeof(args)); + NVIF_ERRON(ret, &outp->object, + "[DP_MST_VCPI head:%d start_slot:%02x num_slots:%02x pbn:%04x aligned_pbn:%04x]", + args.head, args.start_slot, args.num_slots, args.pbn, args.aligned_pbn); + return ret; +} + +int +nvif_outp_dp_retrain(struct nvif_outp *outp) +{ + int ret = nvif_object_mthd(&outp->object, NVIF_OUTP_V0_DP_RETRAIN, NULL, 0); + NVIF_ERRON(ret, &outp->object, "[DP_RETRAIN]"); + return ret; +} + +int +nvif_outp_dp_aux_pwr(struct nvif_outp *outp, bool enable) +{ + struct nvif_outp_dp_aux_pwr_v0 args; + int ret; + + args.version = 0; + args.state = enable; + + ret = nvif_object_mthd(&outp->object, NVIF_OUTP_V0_DP_AUX_PWR, &args, sizeof(args)); + NVIF_ERRON(ret, &outp->object, "[DP_AUX_PWR state:%d]", args.state); + return ret; +} + +int +nvif_outp_hda_eld(struct nvif_outp *outp, int head, void *data, u32 size) +{ + struct { + struct nvif_outp_hda_eld_v0 mthd; + u8 data[128]; + } args; + int ret; + + if (WARN_ON(size > ARRAY_SIZE(args.data))) + return -EINVAL; + + args.mthd.version = 0; + args.mthd.head = head; + + memcpy(args.data, data, size); + ret = nvif_mthd(&outp->object, NVIF_OUTP_V0_HDA_ELD, &args, sizeof(args.mthd) + size); + NVIF_ERRON(ret, &outp->object, "[HDA_ELD head:%d size:%d]", head, size); + return ret; +} + +int +nvif_outp_infoframe(struct nvif_outp *outp, u8 type, struct nvif_outp_infoframe_v0 *args, u32 size) +{ + int ret; + + args->type = type; + + ret = nvif_mthd(&outp->object, NVIF_OUTP_V0_INFOFRAME, args, sizeof(*args) + size); + NVIF_ERRON(ret, &outp->object, "[INFOFRAME type:%d size:%d]", type, size); + return ret; +} + +void +nvif_outp_release(struct nvif_outp *outp) +{ + int ret = nvif_mthd(&outp->object, NVIF_OUTP_V0_RELEASE, NULL, 0); + NVIF_ERRON(ret, &outp->object, "[RELEASE]"); + outp->or.id = -1; +} + +static inline int +nvif_outp_acquire(struct nvif_outp *outp, u8 proto, struct nvif_outp_acquire_v0 *args) +{ + int ret; + + args->version = 0; + args->proto = proto; + + ret = nvif_mthd(&outp->object, NVIF_OUTP_V0_ACQUIRE, args, sizeof(*args)); + if (ret) + return ret; + + outp->or.id = args->or; + outp->or.link = args->link; + return 0; +} + +int +nvif_outp_acquire_dp(struct nvif_outp *outp, u8 dpcd[16], + int link_nr, int link_bw, bool hda, bool mst) +{ + struct nvif_outp_acquire_v0 args; + int ret; + + args.dp.link_nr = link_nr; + args.dp.link_bw = link_bw; + args.dp.hda = hda; + args.dp.mst = mst; + memcpy(args.dp.dpcd, dpcd, sizeof(args.dp.dpcd)); + + ret = nvif_outp_acquire(outp, NVIF_OUTP_ACQUIRE_V0_DP, &args); + NVIF_ERRON(ret, &outp->object, + "[ACQUIRE proto:DP link_nr:%d link_bw:%02x hda:%d mst:%d] or:%d link:%d", + args.dp.link_nr, args.dp.link_bw, args.dp.hda, args.dp.mst, args.or, args.link); + return ret; +} + +int +nvif_outp_acquire_lvds(struct nvif_outp *outp, bool dual, bool bpc8) +{ + struct nvif_outp_acquire_v0 args; + int ret; + + args.lvds.dual = dual; + args.lvds.bpc8 = bpc8; + + ret = nvif_outp_acquire(outp, NVIF_OUTP_ACQUIRE_V0_LVDS, &args); + NVIF_ERRON(ret, &outp->object, + "[ACQUIRE proto:LVDS dual:%d 8bpc:%d] or:%d link:%d", + args.lvds.dual, args.lvds.bpc8, args.or, args.link); + return ret; +} + +int +nvif_outp_acquire_tmds(struct nvif_outp *outp, int head, + bool hdmi, u8 max_ac_packet, u8 rekey, u8 scdc, bool hda) +{ + struct nvif_outp_acquire_v0 args; + int ret; + + args.tmds.head = head; + args.tmds.hdmi = hdmi; + args.tmds.hdmi_max_ac_packet = max_ac_packet; + args.tmds.hdmi_rekey = rekey; + args.tmds.hdmi_scdc = scdc; + args.tmds.hdmi_hda = hda; + + ret = nvif_outp_acquire(outp, NVIF_OUTP_ACQUIRE_V0_TMDS, &args); + NVIF_ERRON(ret, &outp->object, + "[ACQUIRE proto:TMDS head:%d hdmi:%d max_ac_packet:%d rekey:%d scdc:%d hda:%d]" + " or:%d link:%d", args.tmds.head, args.tmds.hdmi, args.tmds.hdmi_max_ac_packet, + args.tmds.hdmi_rekey, args.tmds.hdmi_scdc, args.tmds.hdmi_hda, + args.or, args.link); + return ret; +} + +int +nvif_outp_acquire_rgb_crt(struct nvif_outp *outp) +{ + struct nvif_outp_acquire_v0 args; + int ret; + + ret = nvif_outp_acquire(outp, NVIF_OUTP_ACQUIRE_V0_RGB_CRT, &args); + NVIF_ERRON(ret, &outp->object, "[ACQUIRE proto:RGB_CRT] or:%d", args.or); + return ret; +} int nvif_outp_load_detect(struct nvif_outp *outp, u32 loadval) @@ -58,5 +228,9 @@ nvif_outp_ctor(struct nvif_disp *disp, const char *name, int id, struct nvif_out ret = nvif_object_ctor(&disp->object, name ?: "nvifOutp", id, NVIF_CLASS_OUTP, &args, sizeof(args), &outp->object); NVIF_ERRON(ret, &disp->object, "[NEW outp id:%d]", id); - return ret; + if (ret) + return ret; + + outp->or.id = -1; + return 0; } diff --git a/drivers/gpu/drm/nouveau/nvif/user.c b/drivers/gpu/drm/nouveau/nvif/user.c index d89f5b67b304..b648a5e036af 100644 --- a/drivers/gpu/drm/nouveau/nvif/user.c +++ b/drivers/gpu/drm/nouveau/nvif/user.c @@ -41,7 +41,9 @@ nvif_user_ctor(struct nvif_device *device, const char *name) int version; const struct nvif_user_func *func; } users[] = { - { VOLTA_USERMODE_A, -1, &nvif_userc361 }, + { AMPERE_USERMODE_A, -1, &nvif_userc361 }, + { TURING_USERMODE_A, -1, &nvif_userc361 }, + { VOLTA_USERMODE_A, -1, &nvif_userc361 }, {} }; int cid, ret; diff --git a/drivers/gpu/drm/nouveau/nvkm/core/Kbuild b/drivers/gpu/drm/nouveau/nvkm/core/Kbuild index 2b471ab585b4..e40712023c73 100644 --- a/drivers/gpu/drm/nouveau/nvkm/core/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/core/Kbuild @@ -5,12 +5,13 @@ nvkm-y += nvkm/core/enum.o nvkm-y += nvkm/core/event.o nvkm-y += nvkm/core/firmware.o nvkm-y += nvkm/core/gpuobj.o +nvkm-y += nvkm/core/intr.o nvkm-y += nvkm/core/ioctl.o nvkm-y += nvkm/core/memory.o nvkm-y += nvkm/core/mm.o -nvkm-y += nvkm/core/notify.o nvkm-y += nvkm/core/object.o nvkm-y += nvkm/core/oproxy.o nvkm-y += nvkm/core/option.o nvkm-y += nvkm/core/ramht.o nvkm-y += nvkm/core/subdev.o +nvkm-y += nvkm/core/uevent.o diff --git a/drivers/gpu/drm/nouveau/nvkm/core/client.c b/drivers/gpu/drm/nouveau/nvkm/core/client.c index 0c8c55c73b12..ebdeb8eb9e77 100644 --- a/drivers/gpu/drm/nouveau/nvkm/core/client.c +++ b/drivers/gpu/drm/nouveau/nvkm/core/client.c @@ -23,7 +23,6 @@ */ #include <core/client.h> #include <core/device.h> -#include <core/notify.h> #include <core/option.h> #include <nvif/class.h> @@ -44,7 +43,7 @@ nvkm_uclient_new(const struct nvkm_oclass *oclass, void *argv, u32 argc, if (!(ret = nvif_unpack(ret, &argv, &argc, args->v0, 0, 0, false))){ args->v0.name[sizeof(args->v0.name) - 1] = 0; ret = nvkm_client_new(args->v0.name, args->v0.device, NULL, - NULL, oclass->client->ntfy, &client); + NULL, oclass->client->event, &client); if (ret) return ret; } else @@ -68,113 +67,6 @@ nvkm_uclient_sclass = { .ctor = nvkm_uclient_new, }; -struct nvkm_client_notify { - struct nvkm_client *client; - struct nvkm_notify n; - u8 version; - u8 size; - union { - struct nvif_notify_rep_v0 v0; - } rep; -}; - -static int -nvkm_client_notify(struct nvkm_notify *n) -{ - struct nvkm_client_notify *notify = container_of(n, typeof(*notify), n); - struct nvkm_client *client = notify->client; - return client->ntfy(¬ify->rep, notify->size, n->data, n->size); -} - -int -nvkm_client_notify_put(struct nvkm_client *client, int index) -{ - if (index < ARRAY_SIZE(client->notify)) { - if (client->notify[index]) { - nvkm_notify_put(&client->notify[index]->n); - return 0; - } - } - return -ENOENT; -} - -int -nvkm_client_notify_get(struct nvkm_client *client, int index) -{ - if (index < ARRAY_SIZE(client->notify)) { - if (client->notify[index]) { - nvkm_notify_get(&client->notify[index]->n); - return 0; - } - } - return -ENOENT; -} - -int -nvkm_client_notify_del(struct nvkm_client *client, int index) -{ - if (index < ARRAY_SIZE(client->notify)) { - if (client->notify[index]) { - nvkm_notify_fini(&client->notify[index]->n); - kfree(client->notify[index]); - client->notify[index] = NULL; - return 0; - } - } - return -ENOENT; -} - -int -nvkm_client_notify_new(struct nvkm_object *object, - struct nvkm_event *event, void *data, u32 size) -{ - struct nvkm_client *client = object->client; - struct nvkm_client_notify *notify; - union { - struct nvif_notify_req_v0 v0; - } *req = data; - u8 index, reply; - int ret = -ENOSYS; - - for (index = 0; index < ARRAY_SIZE(client->notify); index++) { - if (!client->notify[index]) - break; - } - - if (index == ARRAY_SIZE(client->notify)) - return -ENOSPC; - - notify = kzalloc(sizeof(*notify), GFP_KERNEL); - if (!notify) - return -ENOMEM; - - nvif_ioctl(object, "notify new size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, req->v0, 0, 0, true))) { - nvif_ioctl(object, "notify new vers %d reply %d route %02x " - "token %llx\n", req->v0.version, - req->v0.reply, req->v0.route, req->v0.token); - notify->version = req->v0.version; - notify->size = sizeof(notify->rep.v0); - notify->rep.v0.version = req->v0.version; - notify->rep.v0.route = req->v0.route; - notify->rep.v0.token = req->v0.token; - reply = req->v0.reply; - } - - if (ret == 0) { - ret = nvkm_notify_init(object, event, nvkm_client_notify, - false, data, size, reply, ¬ify->n); - if (ret == 0) { - client->notify[index] = notify; - notify->client = client; - return index; - } - } - - kfree(notify); - return ret; -} - static const struct nvkm_object_func nvkm_client; struct nvkm_client * nvkm_client_search(struct nvkm_client *client, u64 handle) @@ -255,23 +147,13 @@ nvkm_client_child_get(struct nvkm_object *object, int index, static int nvkm_client_fini(struct nvkm_object *object, bool suspend) { - struct nvkm_client *client = nvkm_client(object); - const char *name[2] = { "fini", "suspend" }; - int i; - nvif_debug(object, "%s notify\n", name[suspend]); - for (i = 0; i < ARRAY_SIZE(client->notify); i++) - nvkm_client_notify_put(client, i); return 0; } static void * nvkm_client_dtor(struct nvkm_object *object) { - struct nvkm_client *client = nvkm_client(object); - int i; - for (i = 0; i < ARRAY_SIZE(client->notify); i++) - nvkm_client_notify_del(client, i); - return client; + return nvkm_client(object); } static const struct nvkm_object_func @@ -283,10 +165,8 @@ nvkm_client = { }; int -nvkm_client_new(const char *name, u64 device, const char *cfg, - const char *dbg, - int (*ntfy)(const void *, u32, const void *, u32), - struct nvkm_client **pclient) +nvkm_client_new(const char *name, u64 device, const char *cfg, const char *dbg, + int (*event)(u64, void *, u32), struct nvkm_client **pclient) { struct nvkm_oclass oclass = { .base = nvkm_uclient_sclass }; struct nvkm_client *client; @@ -300,7 +180,7 @@ nvkm_client_new(const char *name, u64 device, const char *cfg, client->device = device; client->debug = nvkm_dbgopt(dbg, "CLIENT"); client->objroot = RB_ROOT; - client->ntfy = ntfy; + client->event = event; INIT_LIST_HEAD(&client->umem); spin_lock_init(&client->lock); return 0; diff --git a/drivers/gpu/drm/nouveau/nvkm/core/engine.c b/drivers/gpu/drm/nouveau/nvkm/core/engine.c index e41a39ae1597..36a31e9eea22 100644 --- a/drivers/gpu/drm/nouveau/nvkm/core/engine.c +++ b/drivers/gpu/drm/nouveau/nvkm/core/engine.c @@ -35,16 +35,23 @@ nvkm_engine_chsw_load(struct nvkm_engine *engine) return false; } +int +nvkm_engine_reset(struct nvkm_engine *engine) +{ + if (engine->func->reset) + return engine->func->reset(engine); + + nvkm_subdev_fini(&engine->subdev, false); + return nvkm_subdev_init(&engine->subdev); +} + void nvkm_engine_unref(struct nvkm_engine **pengine) { struct nvkm_engine *engine = *pengine; + if (engine) { - if (refcount_dec_and_mutex_lock(&engine->use.refcount, &engine->use.mutex)) { - nvkm_subdev_fini(&engine->subdev, false); - engine->use.enabled = false; - mutex_unlock(&engine->use.mutex); - } + nvkm_subdev_unref(&engine->subdev); *pengine = NULL; } } @@ -53,21 +60,13 @@ struct nvkm_engine * nvkm_engine_ref(struct nvkm_engine *engine) { int ret; + if (engine) { - if (!refcount_inc_not_zero(&engine->use.refcount)) { - mutex_lock(&engine->use.mutex); - if (!refcount_inc_not_zero(&engine->use.refcount)) { - engine->use.enabled = true; - if ((ret = nvkm_subdev_init(&engine->subdev))) { - engine->use.enabled = false; - mutex_unlock(&engine->use.mutex); - return ERR_PTR(ret); - } - refcount_set(&engine->use.refcount, 1); - } - mutex_unlock(&engine->use.mutex); - } + ret = nvkm_subdev_ref(&engine->subdev); + if (ret) + return ERR_PTR(ret); } + return engine; } @@ -91,14 +90,10 @@ static int nvkm_engine_info(struct nvkm_subdev *subdev, u64 mthd, u64 *data) { struct nvkm_engine *engine = nvkm_engine(subdev); - if (engine->func->info) { - if (!IS_ERR((engine = nvkm_engine_ref(engine)))) { - int ret = engine->func->info(engine, mthd, data); - nvkm_engine_unref(&engine); - return ret; - } - return PTR_ERR(engine); - } + + if (engine->func->info) + return engine->func->info(engine, mthd, data); + return -ENOSYS; } @@ -117,26 +112,6 @@ nvkm_engine_init(struct nvkm_subdev *subdev) struct nvkm_engine *engine = nvkm_engine(subdev); struct nvkm_fb *fb = subdev->device->fb; int ret = 0, i; - s64 time; - - if (!engine->use.enabled) { - nvkm_trace(subdev, "init skipped, engine has no users\n"); - return ret; - } - - if (engine->func->oneinit && !engine->subdev.oneinit) { - nvkm_trace(subdev, "one-time init running...\n"); - time = ktime_to_us(ktime_get()); - ret = engine->func->oneinit(engine); - if (ret) { - nvkm_trace(subdev, "one-time init failed, %d\n", ret); - return ret; - } - - engine->subdev.oneinit = true; - time = ktime_to_us(ktime_get()) - time; - nvkm_trace(subdev, "one-time init completed in %lldus\n", time); - } if (engine->func->init) ret = engine->func->init(engine); @@ -147,6 +122,17 @@ nvkm_engine_init(struct nvkm_subdev *subdev) } static int +nvkm_engine_oneinit(struct nvkm_subdev *subdev) +{ + struct nvkm_engine *engine = nvkm_engine(subdev); + + if (engine->func->oneinit) + return engine->func->oneinit(engine); + + return 0; +} + +static int nvkm_engine_preinit(struct nvkm_subdev *subdev) { struct nvkm_engine *engine = nvkm_engine(subdev); @@ -161,7 +147,6 @@ nvkm_engine_dtor(struct nvkm_subdev *subdev) struct nvkm_engine *engine = nvkm_engine(subdev); if (engine->func->dtor) return engine->func->dtor(engine); - mutex_destroy(&engine->use.mutex); return engine; } @@ -169,6 +154,7 @@ const struct nvkm_subdev_func nvkm_engine = { .dtor = nvkm_engine_dtor, .preinit = nvkm_engine_preinit, + .oneinit = nvkm_engine_oneinit, .init = nvkm_engine_init, .fini = nvkm_engine_fini, .info = nvkm_engine_info, @@ -179,10 +165,9 @@ int nvkm_engine_ctor(const struct nvkm_engine_func *func, struct nvkm_device *device, enum nvkm_subdev_type type, int inst, bool enable, struct nvkm_engine *engine) { - nvkm_subdev_ctor(&nvkm_engine, device, type, inst, &engine->subdev); engine->func = func; - refcount_set(&engine->use.refcount, 0); - mutex_init(&engine->use.mutex); + nvkm_subdev_ctor(&nvkm_engine, device, type, inst, &engine->subdev); + refcount_set(&engine->subdev.use.refcount, 0); if (!nvkm_boolopt(device->cfgopt, engine->subdev.name, enable)) { nvkm_debug(&engine->subdev, "disabled\n"); diff --git a/drivers/gpu/drm/nouveau/nvkm/core/event.c b/drivers/gpu/drm/nouveau/nvkm/core/event.c index 006618d77aa4..a6c877135598 100644 --- a/drivers/gpu/drm/nouveau/nvkm/core/event.c +++ b/drivers/gpu/drm/nouveau/nvkm/core/event.c @@ -20,54 +20,171 @@ * OTHER DEALINGS IN THE SOFTWARE. */ #include <core/event.h> -#include <core/notify.h> +#include <core/subdev.h> -void +static void nvkm_event_put(struct nvkm_event *event, u32 types, int index) { assert_spin_locked(&event->refs_lock); + + nvkm_trace(event->subdev, "event: decr %08x on %d\n", types, index); + while (types) { int type = __ffs(types); types &= ~(1 << type); if (--event->refs[index * event->types_nr + type] == 0) { + nvkm_trace(event->subdev, "event: blocking %d on %d\n", type, index); if (event->func->fini) event->func->fini(event, 1 << type, index); } } } -void +static void nvkm_event_get(struct nvkm_event *event, u32 types, int index) { assert_spin_locked(&event->refs_lock); + + nvkm_trace(event->subdev, "event: incr %08x on %d\n", types, index); + while (types) { int type = __ffs(types); types &= ~(1 << type); if (++event->refs[index * event->types_nr + type] == 1) { + nvkm_trace(event->subdev, "event: allowing %d on %d\n", type, index); if (event->func->init) event->func->init(event, 1 << type, index); } } } +static void +nvkm_event_ntfy_state(struct nvkm_event_ntfy *ntfy) +{ + struct nvkm_event *event = ntfy->event; + unsigned long flags; + + nvkm_trace(event->subdev, "event: ntfy state changed\n"); + spin_lock_irqsave(&event->refs_lock, flags); + + if (atomic_read(&ntfy->allowed) != ntfy->running) { + if (ntfy->running) { + nvkm_event_put(ntfy->event, ntfy->bits, ntfy->id); + ntfy->running = false; + } else { + nvkm_event_get(ntfy->event, ntfy->bits, ntfy->id); + ntfy->running = true; + } + } + + spin_unlock_irqrestore(&event->refs_lock, flags); +} + +static void +nvkm_event_ntfy_remove(struct nvkm_event_ntfy *ntfy) +{ + spin_lock_irq(&ntfy->event->list_lock); + list_del_init(&ntfy->head); + spin_unlock_irq(&ntfy->event->list_lock); +} + +static void +nvkm_event_ntfy_insert(struct nvkm_event_ntfy *ntfy) +{ + spin_lock_irq(&ntfy->event->list_lock); + list_add_tail(&ntfy->head, &ntfy->event->ntfy); + spin_unlock_irq(&ntfy->event->list_lock); +} + +static void +nvkm_event_ntfy_block_(struct nvkm_event_ntfy *ntfy, bool wait) +{ + struct nvkm_subdev *subdev = ntfy->event->subdev; + + nvkm_trace(subdev, "event: ntfy block %08x on %d wait:%d\n", ntfy->bits, ntfy->id, wait); + + if (atomic_xchg(&ntfy->allowed, 0) == 1) { + nvkm_event_ntfy_state(ntfy); + if (wait) + nvkm_event_ntfy_remove(ntfy); + } +} + void -nvkm_event_send(struct nvkm_event *event, u32 types, int index, - void *data, u32 size) +nvkm_event_ntfy_block(struct nvkm_event_ntfy *ntfy) { - struct nvkm_notify *notify; + if (ntfy->event) + nvkm_event_ntfy_block_(ntfy, ntfy->wait); +} + +void +nvkm_event_ntfy_allow(struct nvkm_event_ntfy *ntfy) +{ + nvkm_trace(ntfy->event->subdev, "event: ntfy allow %08x on %d\n", ntfy->bits, ntfy->id); + + if (atomic_xchg(&ntfy->allowed, 1) == 0) { + nvkm_event_ntfy_state(ntfy); + if (ntfy->wait) + nvkm_event_ntfy_insert(ntfy); + } +} + +void +nvkm_event_ntfy_del(struct nvkm_event_ntfy *ntfy) +{ + struct nvkm_event *event = ntfy->event; + + if (!event) + return; + + nvkm_trace(event->subdev, "event: ntfy del %08x on %d\n", ntfy->bits, ntfy->id); + + nvkm_event_ntfy_block_(ntfy, false); + nvkm_event_ntfy_remove(ntfy); + ntfy->event = NULL; +} + +void +nvkm_event_ntfy_add(struct nvkm_event *event, int id, u32 bits, bool wait, nvkm_event_func func, + struct nvkm_event_ntfy *ntfy) +{ + nvkm_trace(event->subdev, "event: ntfy add %08x on %d wait:%d\n", id, bits, wait); + + ntfy->event = event; + ntfy->id = id; + ntfy->bits = bits; + ntfy->wait = wait; + ntfy->func = func; + atomic_set(&ntfy->allowed, 0); + ntfy->running = false; + INIT_LIST_HEAD(&ntfy->head); + if (!ntfy->wait) + nvkm_event_ntfy_insert(ntfy); +} + +bool +nvkm_event_ntfy_valid(struct nvkm_event *event, int id, u32 bits) +{ + return true; +} + +void +nvkm_event_ntfy(struct nvkm_event *event, int id, u32 bits) +{ + struct nvkm_event_ntfy *ntfy, *ntmp; unsigned long flags; - if (!event->refs || WARN_ON(index >= event->index_nr)) + if (!event->refs || WARN_ON(id >= event->index_nr)) return; + nvkm_trace(event->subdev, "event: ntfy %08x on %d\n", bits, id); spin_lock_irqsave(&event->list_lock, flags); - list_for_each_entry(notify, &event->list, head) { - if (notify->index == index && (notify->types & types)) { - if (event->func->send) { - event->func->send(data, size, notify); - continue; - } - nvkm_notify_send(notify, data, size); + + list_for_each_entry_safe(ntfy, ntmp, &event->ntfy, head) { + if (ntfy->id == id && ntfy->bits & bits) { + if (atomic_read(&ntfy->allowed)) + ntfy->func(ntfy, ntfy->bits & bits); } } + spin_unlock_irqrestore(&event->list_lock, flags); } @@ -81,20 +198,17 @@ nvkm_event_fini(struct nvkm_event *event) } int -nvkm_event_init(const struct nvkm_event_func *func, int types_nr, int index_nr, - struct nvkm_event *event) +__nvkm_event_init(const struct nvkm_event_func *func, struct nvkm_subdev *subdev, + int types_nr, int index_nr, struct nvkm_event *event) { - event->refs = kzalloc(array3_size(index_nr, types_nr, - sizeof(*event->refs)), - GFP_KERNEL); + event->refs = kzalloc(array3_size(index_nr, types_nr, sizeof(*event->refs)), GFP_KERNEL); if (!event->refs) return -ENOMEM; event->func = func; + event->subdev = subdev; event->types_nr = types_nr; event->index_nr = index_nr; - spin_lock_init(&event->refs_lock); - spin_lock_init(&event->list_lock); - INIT_LIST_HEAD(&event->list); + INIT_LIST_HEAD(&event->ntfy); return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/core/firmware.c b/drivers/gpu/drm/nouveau/nvkm/core/firmware.c index ca1f8463cff5..fcf2a002f6cb 100644 --- a/drivers/gpu/drm/nouveau/nvkm/core/firmware.c +++ b/drivers/gpu/drm/nouveau/nvkm/core/firmware.c @@ -22,6 +22,9 @@ #include <core/device.h> #include <core/firmware.h> +#include <subdev/fb.h> +#include <subdev/mmu.h> + int nvkm_firmware_load_name(const struct nvkm_subdev *subdev, const char *base, const char *name, int ver, const struct firmware **pfw) @@ -107,3 +110,127 @@ nvkm_firmware_put(const struct firmware *fw) { release_firmware(fw); } + +#define nvkm_firmware_mem(p) container_of((p), struct nvkm_firmware, mem.memory) + +static int +nvkm_firmware_mem_map(struct nvkm_memory *memory, u64 offset, struct nvkm_vmm *vmm, + struct nvkm_vma *vma, void *argv, u32 argc) +{ + struct nvkm_firmware *fw = nvkm_firmware_mem(memory); + struct nvkm_vmm_map map = { + .memory = &fw->mem.memory, + .offset = offset, + .sgl = &fw->mem.sgl, + }; + + if (WARN_ON(fw->func->type != NVKM_FIRMWARE_IMG_DMA)) + return -ENOSYS; + + return nvkm_vmm_map(vmm, vma, argv, argc, &map); +} + +static u64 +nvkm_firmware_mem_size(struct nvkm_memory *memory) +{ + return sg_dma_len(&nvkm_firmware_mem(memory)->mem.sgl); +} + +static u64 +nvkm_firmware_mem_addr(struct nvkm_memory *memory) +{ + return nvkm_firmware_mem(memory)->phys; +} + +static u8 +nvkm_firmware_mem_page(struct nvkm_memory *memory) +{ + return PAGE_SHIFT; +} + +static enum nvkm_memory_target +nvkm_firmware_mem_target(struct nvkm_memory *memory) +{ + return NVKM_MEM_TARGET_HOST; +} + +static void * +nvkm_firmware_mem_dtor(struct nvkm_memory *memory) +{ + return NULL; +} + +static const struct nvkm_memory_func +nvkm_firmware_mem = { + .dtor = nvkm_firmware_mem_dtor, + .target = nvkm_firmware_mem_target, + .page = nvkm_firmware_mem_page, + .addr = nvkm_firmware_mem_addr, + .size = nvkm_firmware_mem_size, + .map = nvkm_firmware_mem_map, +}; + +void +nvkm_firmware_dtor(struct nvkm_firmware *fw) +{ + struct nvkm_memory *memory = &fw->mem.memory; + + if (!fw->img) + return; + + switch (fw->func->type) { + case NVKM_FIRMWARE_IMG_RAM: + kfree(fw->img); + break; + case NVKM_FIRMWARE_IMG_DMA: + nvkm_memory_unref(&memory); + dma_free_coherent(fw->device->dev, sg_dma_len(&fw->mem.sgl), fw->img, fw->phys); + break; + default: + WARN_ON(1); + break; + } + + fw->img = NULL; +} + +int +nvkm_firmware_ctor(const struct nvkm_firmware_func *func, const char *name, + struct nvkm_device *device, const void *src, int len, struct nvkm_firmware *fw) +{ + fw->func = func; + fw->name = name; + fw->device = device; + fw->len = len; + + switch (fw->func->type) { + case NVKM_FIRMWARE_IMG_RAM: + fw->img = kmemdup(src, fw->len, GFP_KERNEL); + break; + case NVKM_FIRMWARE_IMG_DMA: { + dma_addr_t addr; + + len = ALIGN(fw->len, PAGE_SIZE); + + fw->img = dma_alloc_coherent(fw->device->dev, len, &addr, GFP_KERNEL); + if (fw->img) { + memcpy(fw->img, src, fw->len); + fw->phys = addr; + } + + sg_init_one(&fw->mem.sgl, fw->img, len); + sg_dma_address(&fw->mem.sgl) = fw->phys; + sg_dma_len(&fw->mem.sgl) = len; + } + break; + default: + WARN_ON(1); + return -EINVAL; + } + + if (!fw->img) + return -ENOMEM; + + nvkm_memory_ctor(&nvkm_firmware_mem, &fw->mem.memory); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/core/intr.c b/drivers/gpu/drm/nouveau/nvkm/core/intr.c new file mode 100644 index 000000000000..e20b7ca218c3 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/core/intr.c @@ -0,0 +1,442 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include <core/intr.h> +#include <core/device.h> +#include <core/subdev.h> +#include <subdev/pci.h> +#include <subdev/top.h> + +static int +nvkm_intr_xlat(struct nvkm_subdev *subdev, struct nvkm_intr *intr, + enum nvkm_intr_type type, int *leaf, u32 *mask) +{ + struct nvkm_device *device = subdev->device; + + if (type < NVKM_INTR_VECTOR_0) { + if (type == NVKM_INTR_SUBDEV) { + const struct nvkm_intr_data *data = intr->data; + struct nvkm_top_device *tdev; + + while (data && data->mask) { + if (data->type == NVKM_SUBDEV_TOP) { + list_for_each_entry(tdev, &device->top->device, head) { + if (tdev->intr >= 0 && + tdev->type == subdev->type && + tdev->inst == subdev->inst) { + if (data->mask & BIT(tdev->intr)) { + *leaf = data->leaf; + *mask = BIT(tdev->intr); + return 0; + } + } + } + } else + if (data->type == subdev->type && data->inst == subdev->inst) { + *leaf = data->leaf; + *mask = data->mask; + return 0; + } + + data++; + } + } else { + return -ENOSYS; + } + } else { + if (type < intr->leaves * sizeof(*intr->stat) * 8) { + *leaf = type / 32; + *mask = BIT(type % 32); + return 0; + } + } + + return -EINVAL; +} + +static struct nvkm_intr * +nvkm_intr_find(struct nvkm_subdev *subdev, enum nvkm_intr_type type, int *leaf, u32 *mask) +{ + struct nvkm_intr *intr; + int ret; + + list_for_each_entry(intr, &subdev->device->intr.intr, head) { + ret = nvkm_intr_xlat(subdev, intr, type, leaf, mask); + if (ret == 0) + return intr; + } + + return NULL; +} + +static void +nvkm_intr_allow_locked(struct nvkm_intr *intr, int leaf, u32 mask) +{ + intr->mask[leaf] |= mask; + if (intr->func->allow) { + if (intr->func->reset) + intr->func->reset(intr, leaf, mask); + intr->func->allow(intr, leaf, mask); + } +} + +void +nvkm_intr_allow(struct nvkm_subdev *subdev, enum nvkm_intr_type type) +{ + struct nvkm_device *device = subdev->device; + struct nvkm_intr *intr; + unsigned long flags; + int leaf; + u32 mask; + + intr = nvkm_intr_find(subdev, type, &leaf, &mask); + if (intr) { + nvkm_debug(intr->subdev, "intr %d/%08x allowed by %s\n", leaf, mask, subdev->name); + spin_lock_irqsave(&device->intr.lock, flags); + nvkm_intr_allow_locked(intr, leaf, mask); + spin_unlock_irqrestore(&device->intr.lock, flags); + } +} + +static void +nvkm_intr_block_locked(struct nvkm_intr *intr, int leaf, u32 mask) +{ + intr->mask[leaf] &= ~mask; + if (intr->func->block) + intr->func->block(intr, leaf, mask); +} + +void +nvkm_intr_block(struct nvkm_subdev *subdev, enum nvkm_intr_type type) +{ + struct nvkm_device *device = subdev->device; + struct nvkm_intr *intr; + unsigned long flags; + int leaf; + u32 mask; + + intr = nvkm_intr_find(subdev, type, &leaf, &mask); + if (intr) { + nvkm_debug(intr->subdev, "intr %d/%08x blocked by %s\n", leaf, mask, subdev->name); + spin_lock_irqsave(&device->intr.lock, flags); + nvkm_intr_block_locked(intr, leaf, mask); + spin_unlock_irqrestore(&device->intr.lock, flags); + } +} + +static void +nvkm_intr_rearm_locked(struct nvkm_device *device) +{ + struct nvkm_intr *intr; + + list_for_each_entry(intr, &device->intr.intr, head) + intr->func->rearm(intr); +} + +static void +nvkm_intr_unarm_locked(struct nvkm_device *device) +{ + struct nvkm_intr *intr; + + list_for_each_entry(intr, &device->intr.intr, head) + intr->func->unarm(intr); +} + +static irqreturn_t +nvkm_intr(int irq, void *arg) +{ + struct nvkm_device *device = arg; + struct nvkm_intr *intr; + struct nvkm_inth *inth; + irqreturn_t ret = IRQ_NONE; + bool pending = false; + int prio, leaf; + + /* Disable all top-level interrupt sources, and re-arm MSI interrupts. */ + spin_lock(&device->intr.lock); + if (!device->intr.armed) + goto done_unlock; + + nvkm_intr_unarm_locked(device); + nvkm_pci_msi_rearm(device); + + /* Fetch pending interrupt masks. */ + list_for_each_entry(intr, &device->intr.intr, head) { + if (intr->func->pending(intr)) + pending = true; + } + + if (!pending) + goto done; + + /* Check that GPU is still on the bus by reading NV_PMC_BOOT_0. */ + if (WARN_ON(nvkm_rd32(device, 0x000000) == 0xffffffff)) + goto done; + + /* Execute handlers. */ + for (prio = 0; prio < ARRAY_SIZE(device->intr.prio); prio++) { + list_for_each_entry(inth, &device->intr.prio[prio], head) { + struct nvkm_intr *intr = inth->intr; + + if (intr->stat[inth->leaf] & inth->mask) { + if (atomic_read(&inth->allowed)) { + if (intr->func->reset) + intr->func->reset(intr, inth->leaf, inth->mask); + if (inth->func(inth) == IRQ_HANDLED) + ret = IRQ_HANDLED; + } + } + } + } + + /* Nothing handled? Some debugging/protection from IRQ storms is in order... */ + if (ret == IRQ_NONE) { + list_for_each_entry(intr, &device->intr.intr, head) { + for (leaf = 0; leaf < intr->leaves; leaf++) { + if (intr->stat[leaf]) { + nvkm_warn(intr->subdev, "intr%d: %08x\n", + leaf, intr->stat[leaf]); + nvkm_intr_block_locked(intr, leaf, intr->stat[leaf]); + } + } + } + } + +done: + /* Re-enable all top-level interrupt sources. */ + nvkm_intr_rearm_locked(device); +done_unlock: + spin_unlock(&device->intr.lock); + return ret; +} + +int +nvkm_intr_add(const struct nvkm_intr_func *func, const struct nvkm_intr_data *data, + struct nvkm_subdev *subdev, int leaves, struct nvkm_intr *intr) +{ + struct nvkm_device *device = subdev->device; + int i; + + intr->func = func; + intr->data = data; + intr->subdev = subdev; + intr->leaves = leaves; + intr->stat = kcalloc(leaves, sizeof(*intr->stat), GFP_KERNEL); + intr->mask = kcalloc(leaves, sizeof(*intr->mask), GFP_KERNEL); + if (!intr->stat || !intr->mask) { + kfree(intr->stat); + return -ENOMEM; + } + + if (intr->subdev->debug >= NV_DBG_DEBUG) { + for (i = 0; i < intr->leaves; i++) + intr->mask[i] = ~0; + } + + spin_lock_irq(&device->intr.lock); + list_add_tail(&intr->head, &device->intr.intr); + spin_unlock_irq(&device->intr.lock); + return 0; +} + +static irqreturn_t +nvkm_intr_subdev(struct nvkm_inth *inth) +{ + struct nvkm_subdev *subdev = container_of(inth, typeof(*subdev), inth); + + nvkm_subdev_intr(subdev); + return IRQ_HANDLED; +} + +static void +nvkm_intr_subdev_add_dev(struct nvkm_intr *intr, enum nvkm_subdev_type type, int inst) +{ + struct nvkm_subdev *subdev; + enum nvkm_intr_prio prio; + int ret; + + subdev = nvkm_device_subdev(intr->subdev->device, type, inst); + if (!subdev || !subdev->func->intr) + return; + + if (type == NVKM_ENGINE_DISP) + prio = NVKM_INTR_PRIO_VBLANK; + else + prio = NVKM_INTR_PRIO_NORMAL; + + ret = nvkm_inth_add(intr, NVKM_INTR_SUBDEV, prio, subdev, nvkm_intr_subdev, &subdev->inth); + if (WARN_ON(ret)) + return; + + nvkm_inth_allow(&subdev->inth); +} + +static void +nvkm_intr_subdev_add(struct nvkm_intr *intr) +{ + const struct nvkm_intr_data *data; + struct nvkm_device *device = intr->subdev->device; + struct nvkm_top_device *tdev; + + for (data = intr->data; data && data->mask; data++) { + if (data->legacy) { + if (data->type == NVKM_SUBDEV_TOP) { + list_for_each_entry(tdev, &device->top->device, head) { + if (tdev->intr < 0 || !(data->mask & BIT(tdev->intr))) + continue; + + nvkm_intr_subdev_add_dev(intr, tdev->type, tdev->inst); + } + } else { + nvkm_intr_subdev_add_dev(intr, data->type, data->inst); + } + } + } +} + +void +nvkm_intr_rearm(struct nvkm_device *device) +{ + struct nvkm_intr *intr; + int i; + + if (unlikely(!device->intr.legacy_done)) { + list_for_each_entry(intr, &device->intr.intr, head) + nvkm_intr_subdev_add(intr); + device->intr.legacy_done = true; + } + + spin_lock_irq(&device->intr.lock); + list_for_each_entry(intr, &device->intr.intr, head) { + for (i = 0; intr->func->block && i < intr->leaves; i++) { + intr->func->block(intr, i, ~0); + intr->func->allow(intr, i, intr->mask[i]); + } + } + + nvkm_intr_rearm_locked(device); + device->intr.armed = true; + spin_unlock_irq(&device->intr.lock); +} + +void +nvkm_intr_unarm(struct nvkm_device *device) +{ + spin_lock_irq(&device->intr.lock); + nvkm_intr_unarm_locked(device); + device->intr.armed = false; + spin_unlock_irq(&device->intr.lock); +} + +int +nvkm_intr_install(struct nvkm_device *device) +{ + int ret; + + device->intr.irq = device->func->irq(device); + if (device->intr.irq < 0) + return device->intr.irq; + + ret = request_irq(device->intr.irq, nvkm_intr, IRQF_SHARED, "nvkm", device); + if (ret) + return ret; + + device->intr.alloc = true; + return 0; +} + +void +nvkm_intr_dtor(struct nvkm_device *device) +{ + struct nvkm_intr *intr, *intt; + + list_for_each_entry_safe(intr, intt, &device->intr.intr, head) { + list_del(&intr->head); + kfree(intr->mask); + kfree(intr->stat); + } + + if (device->intr.alloc) + free_irq(device->intr.irq, device); +} + +void +nvkm_intr_ctor(struct nvkm_device *device) +{ + int i; + + INIT_LIST_HEAD(&device->intr.intr); + for (i = 0; i < ARRAY_SIZE(device->intr.prio); i++) + INIT_LIST_HEAD(&device->intr.prio[i]); + + spin_lock_init(&device->intr.lock); + device->intr.armed = false; +} + +void +nvkm_inth_block(struct nvkm_inth *inth) +{ + if (unlikely(!inth->intr)) + return; + + atomic_set(&inth->allowed, 0); +} + +void +nvkm_inth_allow(struct nvkm_inth *inth) +{ + struct nvkm_intr *intr = inth->intr; + unsigned long flags; + + if (unlikely(!inth->intr)) + return; + + spin_lock_irqsave(&intr->subdev->device->intr.lock, flags); + if (!atomic_xchg(&inth->allowed, 1)) { + if ((intr->mask[inth->leaf] & inth->mask) != inth->mask) + nvkm_intr_allow_locked(intr, inth->leaf, inth->mask); + } + spin_unlock_irqrestore(&intr->subdev->device->intr.lock, flags); +} + +int +nvkm_inth_add(struct nvkm_intr *intr, enum nvkm_intr_type type, enum nvkm_intr_prio prio, + struct nvkm_subdev *subdev, nvkm_inth_func func, struct nvkm_inth *inth) +{ + struct nvkm_device *device = subdev->device; + int ret; + + if (WARN_ON(inth->mask)) + return -EBUSY; + + ret = nvkm_intr_xlat(subdev, intr, type, &inth->leaf, &inth->mask); + if (ret) + return ret; + + nvkm_debug(intr->subdev, "intr %d/%08x requested by %s\n", + inth->leaf, inth->mask, subdev->name); + + inth->intr = intr; + inth->func = func; + atomic_set(&inth->allowed, 0); + list_add_tail(&inth->head, &device->intr.prio[prio]); + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/core/ioctl.c b/drivers/gpu/drm/nouveau/nvkm/core/ioctl.c index 45f920da89af..0b33287e43a7 100644 --- a/drivers/gpu/drm/nouveau/nvkm/core/ioctl.c +++ b/drivers/gpu/drm/nouveau/nvkm/core/ioctl.c @@ -47,6 +47,26 @@ nvkm_ioctl_nop(struct nvkm_client *client, return ret; } +#include <nvif/class.h> + +static int +nvkm_ioctl_sclass_(struct nvkm_object *object, int index, struct nvkm_oclass *oclass) +{ + if ( object->func->uevent && + !object->func->uevent(object, NULL, 0, NULL) && index-- == 0) { + oclass->ctor = nvkm_uevent_new; + oclass->base.minver = 0; + oclass->base.maxver = 0; + oclass->base.oclass = NVIF_CLASS_EVENT; + return 0; + } + + if (object->func->sclass) + return object->func->sclass(object, index, oclass); + + return -ENOSYS; +} + static int nvkm_ioctl_sclass(struct nvkm_client *client, struct nvkm_object *object, void *data, u32 size) @@ -64,8 +84,7 @@ nvkm_ioctl_sclass(struct nvkm_client *client, if (size != args->v0.count * sizeof(args->v0.oclass[0])) return -EINVAL; - while (object->func->sclass && - object->func->sclass(object, i, &oclass) >= 0) { + while (nvkm_ioctl_sclass_(object, i, &oclass) >= 0) { if (i < args->v0.count) { args->v0.oclass[i].oclass = oclass.base.oclass; args->v0.oclass[i].minver = oclass.base.minver; @@ -100,7 +119,7 @@ nvkm_ioctl_new(struct nvkm_client *client, } else return ret; - if (!parent->func->sclass) { + if (!parent->func->sclass && !parent->func->uevent) { nvif_ioctl(parent, "cannot have children\n"); return -EINVAL; } @@ -113,7 +132,7 @@ nvkm_ioctl_new(struct nvkm_client *client, oclass.object = args->v0.object; oclass.client = client; oclass.parent = parent; - ret = parent->func->sclass(parent, i++, &oclass); + ret = nvkm_ioctl_sclass_(parent, i++, &oclass); if (ret) return ret; } while (oclass.base.oclass != args->v0.oclass); @@ -294,90 +313,6 @@ nvkm_ioctl_unmap(struct nvkm_client *client, return ret; } -static int -nvkm_ioctl_ntfy_new(struct nvkm_client *client, - struct nvkm_object *object, void *data, u32 size) -{ - union { - struct nvif_ioctl_ntfy_new_v0 v0; - } *args = data; - struct nvkm_event *event; - int ret = -ENOSYS; - - nvif_ioctl(object, "ntfy new size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, true))) { - nvif_ioctl(object, "ntfy new vers %d event %02x\n", - args->v0.version, args->v0.event); - ret = nvkm_object_ntfy(object, args->v0.event, &event); - if (ret == 0) { - ret = nvkm_client_notify_new(object, event, data, size); - if (ret >= 0) { - args->v0.index = ret; - ret = 0; - } - } - } - - return ret; -} - -static int -nvkm_ioctl_ntfy_del(struct nvkm_client *client, - struct nvkm_object *object, void *data, u32 size) -{ - union { - struct nvif_ioctl_ntfy_del_v0 v0; - } *args = data; - int ret = -ENOSYS; - - nvif_ioctl(object, "ntfy del size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(object, "ntfy del vers %d index %d\n", - args->v0.version, args->v0.index); - ret = nvkm_client_notify_del(client, args->v0.index); - } - - return ret; -} - -static int -nvkm_ioctl_ntfy_get(struct nvkm_client *client, - struct nvkm_object *object, void *data, u32 size) -{ - union { - struct nvif_ioctl_ntfy_get_v0 v0; - } *args = data; - int ret = -ENOSYS; - - nvif_ioctl(object, "ntfy get size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(object, "ntfy get vers %d index %d\n", - args->v0.version, args->v0.index); - ret = nvkm_client_notify_get(client, args->v0.index); - } - - return ret; -} - -static int -nvkm_ioctl_ntfy_put(struct nvkm_client *client, - struct nvkm_object *object, void *data, u32 size) -{ - union { - struct nvif_ioctl_ntfy_put_v0 v0; - } *args = data; - int ret = -ENOSYS; - - nvif_ioctl(object, "ntfy put size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(object, "ntfy put vers %d index %d\n", - args->v0.version, args->v0.index); - ret = nvkm_client_notify_put(client, args->v0.index); - } - - return ret; -} - static struct { int version; int (*func)(struct nvkm_client *, struct nvkm_object *, void *, u32); @@ -392,10 +327,6 @@ nvkm_ioctl_v0[] = { { 0x00, nvkm_ioctl_wr }, { 0x00, nvkm_ioctl_map }, { 0x00, nvkm_ioctl_unmap }, - { 0x00, nvkm_ioctl_ntfy_new }, - { 0x00, nvkm_ioctl_ntfy_del }, - { 0x00, nvkm_ioctl_ntfy_get }, - { 0x00, nvkm_ioctl_ntfy_put }, }; static int diff --git a/drivers/gpu/drm/nouveau/nvkm/core/notify.c b/drivers/gpu/drm/nouveau/nvkm/core/notify.c deleted file mode 100644 index 023610d01458..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/core/notify.c +++ /dev/null @@ -1,163 +0,0 @@ -/* - * Copyright 2014 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs <bskeggs@redhat.com> - */ -#include <core/notify.h> -#include <core/event.h> - -static inline void -nvkm_notify_put_locked(struct nvkm_notify *notify) -{ - if (notify->block++ == 0) - nvkm_event_put(notify->event, notify->types, notify->index); -} - -void -nvkm_notify_put(struct nvkm_notify *notify) -{ - struct nvkm_event *event = notify->event; - unsigned long flags; - if (likely(event) && - test_and_clear_bit(NVKM_NOTIFY_USER, ¬ify->flags)) { - spin_lock_irqsave(&event->refs_lock, flags); - nvkm_notify_put_locked(notify); - spin_unlock_irqrestore(&event->refs_lock, flags); - if (test_bit(NVKM_NOTIFY_WORK, ¬ify->flags)) - flush_work(¬ify->work); - } -} - -static inline void -nvkm_notify_get_locked(struct nvkm_notify *notify) -{ - if (--notify->block == 0) - nvkm_event_get(notify->event, notify->types, notify->index); -} - -void -nvkm_notify_get(struct nvkm_notify *notify) -{ - struct nvkm_event *event = notify->event; - unsigned long flags; - if (likely(event) && - !test_and_set_bit(NVKM_NOTIFY_USER, ¬ify->flags)) { - spin_lock_irqsave(&event->refs_lock, flags); - nvkm_notify_get_locked(notify); - spin_unlock_irqrestore(&event->refs_lock, flags); - } -} - -static inline void -nvkm_notify_func(struct nvkm_notify *notify) -{ - struct nvkm_event *event = notify->event; - int ret = notify->func(notify); - unsigned long flags; - if ((ret == NVKM_NOTIFY_KEEP) || - !test_and_clear_bit(NVKM_NOTIFY_USER, ¬ify->flags)) { - spin_lock_irqsave(&event->refs_lock, flags); - nvkm_notify_get_locked(notify); - spin_unlock_irqrestore(&event->refs_lock, flags); - } -} - -static void -nvkm_notify_work(struct work_struct *work) -{ - struct nvkm_notify *notify = container_of(work, typeof(*notify), work); - nvkm_notify_func(notify); -} - -void -nvkm_notify_send(struct nvkm_notify *notify, void *data, u32 size) -{ - struct nvkm_event *event = notify->event; - unsigned long flags; - - assert_spin_locked(&event->list_lock); - BUG_ON(size != notify->size); - - spin_lock_irqsave(&event->refs_lock, flags); - if (notify->block) { - spin_unlock_irqrestore(&event->refs_lock, flags); - return; - } - nvkm_notify_put_locked(notify); - spin_unlock_irqrestore(&event->refs_lock, flags); - - if (test_bit(NVKM_NOTIFY_WORK, ¬ify->flags)) { - memcpy((void *)notify->data, data, size); - schedule_work(¬ify->work); - } else { - notify->data = data; - nvkm_notify_func(notify); - notify->data = NULL; - } -} - -void -nvkm_notify_fini(struct nvkm_notify *notify) -{ - unsigned long flags; - if (notify->event) { - nvkm_notify_put(notify); - spin_lock_irqsave(¬ify->event->list_lock, flags); - list_del(¬ify->head); - spin_unlock_irqrestore(¬ify->event->list_lock, flags); - kfree((void *)notify->data); - notify->event = NULL; - } -} - -int -nvkm_notify_init(struct nvkm_object *object, struct nvkm_event *event, - int (*func)(struct nvkm_notify *), bool work, - void *data, u32 size, u32 reply, - struct nvkm_notify *notify) -{ - unsigned long flags; - int ret = -ENODEV; - if ((notify->event = event), event->refs) { - ret = event->func->ctor(object, data, size, notify); - if (ret == 0 && (ret = -EINVAL, notify->size == reply)) { - notify->flags = 0; - notify->block = 1; - notify->func = func; - notify->data = NULL; - if (ret = 0, work) { - INIT_WORK(¬ify->work, nvkm_notify_work); - set_bit(NVKM_NOTIFY_WORK, ¬ify->flags); - notify->data = kmalloc(reply, GFP_KERNEL); - if (!notify->data) - ret = -ENOMEM; - } - } - if (ret == 0) { - spin_lock_irqsave(&event->list_lock, flags); - list_add_tail(¬ify->head, &event->list); - spin_unlock_irqrestore(&event->list_lock, flags); - } - } - if (ret) - notify->event = NULL; - return ret; -} diff --git a/drivers/gpu/drm/nouveau/nvkm/core/oproxy.c b/drivers/gpu/drm/nouveau/nvkm/core/oproxy.c index 16299837a296..3385528da650 100644 --- a/drivers/gpu/drm/nouveau/nvkm/core/oproxy.c +++ b/drivers/gpu/drm/nouveau/nvkm/core/oproxy.c @@ -47,7 +47,12 @@ nvkm_oproxy_map(struct nvkm_object *object, void *argv, u32 argc, static int nvkm_oproxy_unmap(struct nvkm_object *object) { - return nvkm_object_unmap(nvkm_oproxy(object)->object); + struct nvkm_oproxy *oproxy = nvkm_oproxy(object); + + if (unlikely(!oproxy->object)) + return 0; + + return nvkm_object_unmap(oproxy->object); } static int @@ -106,6 +111,18 @@ nvkm_oproxy_sclass(struct nvkm_object *object, int index, } static int +nvkm_oproxy_uevent(struct nvkm_object *object, void *argv, u32 argc, + struct nvkm_uevent *uevent) +{ + struct nvkm_oproxy *oproxy = nvkm_oproxy(object); + + if (!oproxy->object->func->uevent) + return -ENOSYS; + + return oproxy->object->func->uevent(oproxy->object, argv, argc, uevent); +} + +static int nvkm_oproxy_fini(struct nvkm_object *object, bool suspend) { struct nvkm_oproxy *oproxy = nvkm_oproxy(object); @@ -188,6 +205,7 @@ nvkm_oproxy_func = { .wr32 = nvkm_oproxy_wr32, .bind = nvkm_oproxy_bind, .sclass = nvkm_oproxy_sclass, + .uevent = nvkm_oproxy_uevent, }; void diff --git a/drivers/gpu/drm/nouveau/nvkm/core/subdev.c b/drivers/gpu/drm/nouveau/nvkm/core/subdev.c index a74b7acb6832..6c20e827a069 100644 --- a/drivers/gpu/drm/nouveau/nvkm/core/subdev.c +++ b/drivers/gpu/drm/nouveau/nvkm/core/subdev.c @@ -54,7 +54,7 @@ int nvkm_subdev_fini(struct nvkm_subdev *subdev, bool suspend) { struct nvkm_device *device = subdev->device; - const char *action = suspend ? "suspend" : "fini"; + const char *action = suspend ? "suspend" : subdev->use.enabled ? "fini" : "reset"; s64 time; nvkm_trace(subdev, "%s running...\n", action); @@ -68,6 +68,7 @@ nvkm_subdev_fini(struct nvkm_subdev *subdev, bool suspend) return ret; } } + subdev->use.enabled = false; nvkm_mc_reset(device, subdev->type, subdev->inst); @@ -97,30 +98,49 @@ nvkm_subdev_preinit(struct nvkm_subdev *subdev) return 0; } -int -nvkm_subdev_init(struct nvkm_subdev *subdev) +static int +nvkm_subdev_oneinit_(struct nvkm_subdev *subdev) { s64 time; int ret; - nvkm_trace(subdev, "init running...\n"); + if (!subdev->func->oneinit || subdev->oneinit) + return 0; + + nvkm_trace(subdev, "one-time init running...\n"); time = ktime_to_us(ktime_get()); + ret = subdev->func->oneinit(subdev); + if (ret) { + nvkm_error(subdev, "one-time init failed, %d\n", ret); + return ret; + } - if (subdev->func->oneinit && !subdev->oneinit) { - s64 time; - nvkm_trace(subdev, "one-time init running...\n"); - time = ktime_to_us(ktime_get()); - ret = subdev->func->oneinit(subdev); - if (ret) { - nvkm_error(subdev, "one-time init failed, %d\n", ret); - return ret; - } + subdev->oneinit = true; + time = ktime_to_us(ktime_get()) - time; + nvkm_trace(subdev, "one-time init completed in %lldus\n", time); + return 0; +} - subdev->oneinit = true; - time = ktime_to_us(ktime_get()) - time; - nvkm_trace(subdev, "one-time init completed in %lldus\n", time); +static int +nvkm_subdev_init_(struct nvkm_subdev *subdev) +{ + s64 time; + int ret; + + if (subdev->use.enabled) { + nvkm_trace(subdev, "init skipped, already running\n"); + return 0; } + nvkm_trace(subdev, "init running...\n"); + time = ktime_to_us(ktime_get()); + + ret = nvkm_subdev_oneinit_(subdev); + if (ret) + return ret; + + subdev->use.enabled = true; + if (subdev->func->init) { ret = subdev->func->init(subdev); if (ret) { @@ -134,6 +154,64 @@ nvkm_subdev_init(struct nvkm_subdev *subdev) return 0; } +int +nvkm_subdev_init(struct nvkm_subdev *subdev) +{ + int ret; + + mutex_lock(&subdev->use.mutex); + if (refcount_read(&subdev->use.refcount) == 0) { + nvkm_trace(subdev, "init skipped, no users\n"); + mutex_unlock(&subdev->use.mutex); + return 0; + } + + ret = nvkm_subdev_init_(subdev); + mutex_unlock(&subdev->use.mutex); + return ret; +} + +int +nvkm_subdev_oneinit(struct nvkm_subdev *subdev) +{ + int ret; + + mutex_lock(&subdev->use.mutex); + ret = nvkm_subdev_oneinit_(subdev); + mutex_unlock(&subdev->use.mutex); + return ret; +} + +void +nvkm_subdev_unref(struct nvkm_subdev *subdev) +{ + if (refcount_dec_and_mutex_lock(&subdev->use.refcount, &subdev->use.mutex)) { + nvkm_subdev_fini(subdev, false); + mutex_unlock(&subdev->use.mutex); + } +} + +int +nvkm_subdev_ref(struct nvkm_subdev *subdev) +{ + int ret; + + if (subdev && !refcount_inc_not_zero(&subdev->use.refcount)) { + mutex_lock(&subdev->use.mutex); + if (!refcount_inc_not_zero(&subdev->use.refcount)) { + if ((ret = nvkm_subdev_init_(subdev))) { + mutex_unlock(&subdev->use.mutex); + return ret; + } + + refcount_set(&subdev->use.refcount, 1); + } + mutex_unlock(&subdev->use.mutex); + } + + return 0; +} + void nvkm_subdev_del(struct nvkm_subdev **psubdev) { @@ -146,6 +224,7 @@ nvkm_subdev_del(struct nvkm_subdev **psubdev) list_del(&subdev->head); if (subdev->func->dtor) *psubdev = subdev->func->dtor(subdev); + mutex_destroy(&subdev->use.mutex); time = ktime_to_us(ktime_get()) - time; nvkm_trace(subdev, "destroy completed in %lldus\n", time); kfree(*psubdev); @@ -167,8 +246,8 @@ nvkm_subdev_disable(struct nvkm_device *device, enum nvkm_subdev_type type, int } void -nvkm_subdev_ctor(const struct nvkm_subdev_func *func, struct nvkm_device *device, - enum nvkm_subdev_type type, int inst, struct nvkm_subdev *subdev) +__nvkm_subdev_ctor(const struct nvkm_subdev_func *func, struct nvkm_device *device, + enum nvkm_subdev_type type, int inst, struct nvkm_subdev *subdev) { subdev->func = func; subdev->device = device; @@ -180,6 +259,8 @@ nvkm_subdev_ctor(const struct nvkm_subdev_func *func, struct nvkm_device *device else strscpy(subdev->name, nvkm_subdev_type[type], sizeof(subdev->name)); subdev->debug = nvkm_dbgopt(device->dbgopt, subdev->name); + + refcount_set(&subdev->use.refcount, 1); list_add_tail(&subdev->head, &device->subdev); } diff --git a/drivers/gpu/drm/nouveau/nvkm/core/uevent.c b/drivers/gpu/drm/nouveau/nvkm/core/uevent.c new file mode 100644 index 000000000000..ba9d9edaec75 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/core/uevent.c @@ -0,0 +1,157 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#define nvkm_uevent(p) container_of((p), struct nvkm_uevent, object) +#include <core/event.h> +#include <core/client.h> + +#include <nvif/if000e.h> + +struct nvkm_uevent { + struct nvkm_object object; + struct nvkm_object *parent; + nvkm_uevent_func func; + bool wait; + + struct nvkm_event_ntfy ntfy; + atomic_t allowed; +}; + +static int +nvkm_uevent_mthd_block(struct nvkm_uevent *uevent, union nvif_event_block_args *args, u32 argc) +{ + if (argc != sizeof(args->vn)) + return -ENOSYS; + + nvkm_event_ntfy_block(&uevent->ntfy); + atomic_set(&uevent->allowed, 0); + return 0; +} + +static int +nvkm_uevent_mthd_allow(struct nvkm_uevent *uevent, union nvif_event_allow_args *args, u32 argc) +{ + if (argc != sizeof(args->vn)) + return -ENOSYS; + + nvkm_event_ntfy_allow(&uevent->ntfy); + atomic_set(&uevent->allowed, 1); + return 0; +} + +static int +nvkm_uevent_mthd(struct nvkm_object *object, u32 mthd, void *argv, u32 argc) +{ + struct nvkm_uevent *uevent = nvkm_uevent(object); + + switch (mthd) { + case NVIF_EVENT_V0_ALLOW: return nvkm_uevent_mthd_allow(uevent, argv, argc); + case NVIF_EVENT_V0_BLOCK: return nvkm_uevent_mthd_block(uevent, argv, argc); + default: + break; + } + + return -EINVAL; +} + +static int +nvkm_uevent_fini(struct nvkm_object *object, bool suspend) +{ + struct nvkm_uevent *uevent = nvkm_uevent(object); + + nvkm_event_ntfy_block(&uevent->ntfy); + return 0; +} + +static int +nvkm_uevent_init(struct nvkm_object *object) +{ + struct nvkm_uevent *uevent = nvkm_uevent(object); + + if (atomic_read(&uevent->allowed)) + nvkm_event_ntfy_allow(&uevent->ntfy); + + return 0; +} + +static void * +nvkm_uevent_dtor(struct nvkm_object *object) +{ + struct nvkm_uevent *uevent = nvkm_uevent(object); + + nvkm_event_ntfy_del(&uevent->ntfy); + return uevent; +} + +static const struct nvkm_object_func +nvkm_uevent = { + .dtor = nvkm_uevent_dtor, + .init = nvkm_uevent_init, + .fini = nvkm_uevent_fini, + .mthd = nvkm_uevent_mthd, +}; + +static int +nvkm_uevent_ntfy(struct nvkm_event_ntfy *ntfy, u32 bits) +{ + struct nvkm_uevent *uevent = container_of(ntfy, typeof(*uevent), ntfy); + struct nvkm_client *client = uevent->object.client; + + if (uevent->func) + return uevent->func(uevent->parent, uevent->object.token, bits); + + return client->event(uevent->object.token, NULL, 0); +} + +int +nvkm_uevent_add(struct nvkm_uevent *uevent, struct nvkm_event *event, int id, u32 bits, + nvkm_uevent_func func) +{ + if (WARN_ON(uevent->func)) + return -EBUSY; + + nvkm_event_ntfy_add(event, id, bits, uevent->wait, nvkm_uevent_ntfy, &uevent->ntfy); + uevent->func = func; + return 0; +} + +int +nvkm_uevent_new(const struct nvkm_oclass *oclass, void *argv, u32 argc, + struct nvkm_object **pobject) +{ + struct nvkm_object *parent = oclass->parent; + struct nvkm_uevent *uevent; + union nvif_event_args *args = argv; + + if (argc < sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + + if (!(uevent = kzalloc(sizeof(*uevent), GFP_KERNEL))) + return -ENOMEM; + *pobject = &uevent->object; + + nvkm_object_ctor(&nvkm_uevent, oclass, &uevent->object); + uevent->parent = parent; + uevent->func = NULL; + uevent->wait = args->v0.wait; + uevent->ntfy.event = NULL; + return parent->func->uevent(parent, &args->v0.data, argc - sizeof(args->v0), uevent); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/ce/Kbuild b/drivers/gpu/drm/nouveau/nvkm/engine/ce/Kbuild index ba88613e1e46..8bf1635ffabc 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/ce/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/engine/ce/Kbuild @@ -8,3 +8,5 @@ nvkm-y += nvkm/engine/ce/gp100.o nvkm-y += nvkm/engine/ce/gp102.o nvkm-y += nvkm/engine/ce/gv100.o nvkm-y += nvkm/engine/ce/tu102.o +nvkm-y += nvkm/engine/ce/ga100.o +nvkm-y += nvkm/engine/ce/ga102.o diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/ce/ga100.c b/drivers/gpu/drm/nouveau/nvkm/engine/ce/ga100.c new file mode 100644 index 000000000000..6648ed62daa6 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/ce/ga100.c @@ -0,0 +1,82 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +#include <subdev/vfn.h> + +#include <nvif/class.h> + +static irqreturn_t +ga100_ce_intr(struct nvkm_inth *inth) +{ + struct nvkm_subdev *subdev = container_of(inth, typeof(*subdev), inth); + + /*TODO*/ + nvkm_error(subdev, "intr\n"); + return IRQ_NONE; +} + +int +ga100_ce_fini(struct nvkm_engine *engine, bool suspend) +{ + nvkm_inth_block(&engine->subdev.inth); + return 0; +} + +int +ga100_ce_init(struct nvkm_engine *engine) +{ + nvkm_inth_allow(&engine->subdev.inth); + return 0; +} + +int +ga100_ce_oneinit(struct nvkm_engine *engine) +{ + struct nvkm_subdev *subdev = &engine->subdev; + struct nvkm_device *device = subdev->device; + u32 vector; + + vector = nvkm_rd32(device, 0x10442c + (subdev->inst * 0x80)) & 0x00000fff; + + return nvkm_inth_add(&device->vfn->intr, vector, NVKM_INTR_PRIO_NORMAL, + subdev, ga100_ce_intr, &subdev->inth); +} + +static const struct nvkm_engine_func +ga100_ce = { + .oneinit = ga100_ce_oneinit, + .init = ga100_ce_init, + .fini = ga100_ce_fini, + .cclass = &gv100_ce_cclass, + .sclass = { + { -1, -1, AMPERE_DMA_COPY_A }, + {} + } +}; + +int +ga100_ce_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, + struct nvkm_engine **pengine) +{ + return nvkm_engine_new_(&ga100_ce, device, type, inst, true, pengine); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/usergv100.c b/drivers/gpu/drm/nouveau/nvkm/engine/ce/ga102.c index 3dc3b8b312de..9f3448ad625f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/usergv100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/ce/ga102.c @@ -1,5 +1,5 @@ /* - * Copyright 2018 Red Hat Inc. + * Copyright 2021 Red Hat Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), @@ -19,27 +19,26 @@ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. */ -#include "user.h" +#include "priv.h" -static int -gv100_fifo_user_map(struct nvkm_object *object, void *argv, u32 argc, - enum nvkm_object_map *type, u64 *addr, u64 *size) -{ - struct nvkm_device *device = object->engine->subdev.device; - *addr = 0x810000 + device->func->resource_addr(device, 0); - *size = 0x010000; - *type = NVKM_OBJECT_MAP_IO; - return 0; -} +#include <nvif/class.h> -static const struct nvkm_object_func -gv100_fifo_user = { - .map = gv100_fifo_user_map, +static const struct nvkm_engine_func +ga102_ce = { + .oneinit = ga100_ce_oneinit, + .init = ga100_ce_init, + .fini = ga100_ce_fini, + .cclass = &gv100_ce_cclass, + .sclass = { + { -1, -1, AMPERE_DMA_COPY_A }, + { -1, -1, AMPERE_DMA_COPY_B }, + {} + } }; int -gv100_fifo_user_new(const struct nvkm_oclass *oclass, void *argv, u32 argc, - struct nvkm_object **pobject) +ga102_ce_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, + struct nvkm_engine **pengine) { - return nvkm_object_new_(&gv100_fifo_user, oclass, argv, argc, pobject); + return nvkm_engine_new_(&ga102_ce, device, type, inst, true, pengine); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/ce/gt215.c b/drivers/gpu/drm/nouveau/nvkm/engine/ce/gt215.c index 09a112af2f89..c9bf6305c3ec 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/ce/gt215.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/ce/gt215.c @@ -40,7 +40,7 @@ gt215_ce_isr_error_name[] = { }; void -gt215_ce_intr(struct nvkm_falcon *ce, struct nvkm_fifo_chan *chan) +gt215_ce_intr(struct nvkm_falcon *ce, struct nvkm_chan *chan) { struct nvkm_subdev *subdev = &ce->engine.subdev; struct nvkm_device *device = subdev->device; @@ -55,9 +55,9 @@ gt215_ce_intr(struct nvkm_falcon *ce, struct nvkm_fifo_chan *chan) nvkm_error(subdev, "DISPATCH_ERROR %04x [%s] ch %d [%010llx %s] " "subc %d mthd %04x data %08x\n", ssta, - en ? en->name : "", chan ? chan->chid : -1, + en ? en->name : "", chan ? chan->id : -1, chan ? chan->inst->addr : 0, - chan ? chan->object.client->name : "unknown", + chan ? chan->name : "unknown", subc, mthd, data); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/ce/priv.h b/drivers/gpu/drm/nouveau/nvkm/engine/ce/priv.h index cd53b93664d6..c4c046916fa6 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/ce/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/ce/priv.h @@ -8,4 +8,8 @@ void gk104_ce_intr(struct nvkm_engine *); void gp100_ce_intr(struct nvkm_engine *); extern const struct nvkm_object_func gv100_ce_cclass; + +int ga100_ce_oneinit(struct nvkm_engine *); +int ga100_ce_init(struct nvkm_engine *); +int ga100_ce_fini(struct nvkm_engine *, bool); #endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/cipher/g84.c b/drivers/gpu/drm/nouveau/nvkm/engine/cipher/g84.c index be2a7181dc15..caca4f639895 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/cipher/g84.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/cipher/g84.c @@ -81,8 +81,7 @@ g84_cipher_intr(struct nvkm_engine *cipher) { struct nvkm_subdev *subdev = &cipher->subdev; struct nvkm_device *device = subdev->device; - struct nvkm_fifo *fifo = device->fifo; - struct nvkm_fifo_chan *chan; + struct nvkm_chan *chan; u32 stat = nvkm_rd32(device, 0x102130); u32 mthd = nvkm_rd32(device, 0x102190); u32 data = nvkm_rd32(device, 0x102194); @@ -90,16 +89,16 @@ g84_cipher_intr(struct nvkm_engine *cipher) unsigned long flags; char msg[128]; - chan = nvkm_fifo_chan_inst(fifo, (u64)inst << 12, &flags); + chan = nvkm_chan_get_inst(cipher, (u64)inst << 12, &flags); if (stat) { nvkm_snprintbf(msg, sizeof(msg), g84_cipher_intr_mask, stat); nvkm_error(subdev, "%08x [%s] ch %d [%010llx %s] " "mthd %04x data %08x\n", stat, msg, - chan ? chan->chid : -1, (u64)inst << 12, - chan ? chan->object.client->name : "unknown", + chan ? chan->id : -1, (u64)inst << 12, + chan ? chan->name : "unknown", mthd, data); } - nvkm_fifo_chan_put(fifo, flags, &chan); + nvkm_chan_put(&chan, flags); nvkm_wr32(device, 0x102130, stat); nvkm_wr32(device, 0x10200c, 0x10); diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c index d8cf71fb0512..364fea320cb3 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/base.c @@ -1095,7 +1095,7 @@ nv98_chipset = { .volt = { 0x00000001, nv40_volt_new }, .disp = { 0x00000001, g94_disp_new }, .dma = { 0x00000001, nv50_dma_new }, - .fifo = { 0x00000001, g84_fifo_new }, + .fifo = { 0x00000001, g98_fifo_new }, .gr = { 0x00000001, g84_gr_new }, .mspdec = { 0x00000001, g98_mspdec_new }, .msppp = { 0x00000001, g98_msppp_new }, @@ -1161,7 +1161,7 @@ nva3_chipset = { .ce = { 0x00000001, gt215_ce_new }, .disp = { 0x00000001, gt215_disp_new }, .dma = { 0x00000001, nv50_dma_new }, - .fifo = { 0x00000001, g84_fifo_new }, + .fifo = { 0x00000001, g98_fifo_new }, .gr = { 0x00000001, gt215_gr_new }, .mpeg = { 0x00000001, g84_mpeg_new }, .mspdec = { 0x00000001, gt215_mspdec_new }, @@ -1195,7 +1195,7 @@ nva5_chipset = { .ce = { 0x00000001, gt215_ce_new }, .disp = { 0x00000001, gt215_disp_new }, .dma = { 0x00000001, nv50_dma_new }, - .fifo = { 0x00000001, g84_fifo_new }, + .fifo = { 0x00000001, g98_fifo_new }, .gr = { 0x00000001, gt215_gr_new }, .mspdec = { 0x00000001, gt215_mspdec_new }, .msppp = { 0x00000001, gt215_msppp_new }, @@ -1228,7 +1228,7 @@ nva8_chipset = { .ce = { 0x00000001, gt215_ce_new }, .disp = { 0x00000001, gt215_disp_new }, .dma = { 0x00000001, nv50_dma_new }, - .fifo = { 0x00000001, g84_fifo_new }, + .fifo = { 0x00000001, g98_fifo_new }, .gr = { 0x00000001, gt215_gr_new }, .mspdec = { 0x00000001, gt215_mspdec_new }, .msppp = { 0x00000001, gt215_msppp_new }, @@ -1259,7 +1259,7 @@ nvaa_chipset = { .volt = { 0x00000001, nv40_volt_new }, .disp = { 0x00000001, mcp77_disp_new }, .dma = { 0x00000001, nv50_dma_new }, - .fifo = { 0x00000001, g84_fifo_new }, + .fifo = { 0x00000001, g98_fifo_new }, .gr = { 0x00000001, gt200_gr_new }, .mspdec = { 0x00000001, g98_mspdec_new }, .msppp = { 0x00000001, g98_msppp_new }, @@ -1291,7 +1291,7 @@ nvac_chipset = { .volt = { 0x00000001, nv40_volt_new }, .disp = { 0x00000001, mcp77_disp_new }, .dma = { 0x00000001, nv50_dma_new }, - .fifo = { 0x00000001, g84_fifo_new }, + .fifo = { 0x00000001, g98_fifo_new }, .gr = { 0x00000001, mcp79_gr_new }, .mspdec = { 0x00000001, g98_mspdec_new }, .msppp = { 0x00000001, g98_msppp_new }, @@ -1325,7 +1325,7 @@ nvaf_chipset = { .ce = { 0x00000001, gt215_ce_new }, .disp = { 0x00000001, mcp89_disp_new }, .dma = { 0x00000001, nv50_dma_new }, - .fifo = { 0x00000001, g84_fifo_new }, + .fifo = { 0x00000001, g98_fifo_new }, .gr = { 0x00000001, mcp89_gr_new }, .mspdec = { 0x00000001, gt215_mspdec_new }, .msppp = { 0x00000001, gt215_msppp_new }, @@ -2130,7 +2130,7 @@ nv12b_chipset = { .volt = { 0x00000001, gm20b_volt_new }, .ce = { 0x00000004, gm200_ce_new }, .dma = { 0x00000001, gf119_dma_new }, - .fifo = { 0x00000001, gm20b_fifo_new }, + .fifo = { 0x00000001, gm200_fifo_new }, .gr = { 0x00000001, gm20b_gr_new }, .sw = { 0x00000001, gf100_sw_new }, }; @@ -2356,7 +2356,7 @@ nv13b_chipset = { .top = { 0x00000001, gk104_top_new }, .ce = { 0x00000001, gp100_ce_new }, .dma = { 0x00000001, gf119_dma_new }, - .fifo = { 0x00000001, gp10b_fifo_new }, + .fifo = { 0x00000001, gp100_fifo_new }, .gr = { 0x00000001, gp10b_gr_new }, .sw = { 0x00000001, gf100_sw_new }, }; @@ -2364,7 +2364,7 @@ nv13b_chipset = { static const struct nvkm_device_chip nv140_chipset = { .name = "GV100", - .acr = { 0x00000001, gp108_acr_new }, + .acr = { 0x00000001, gv100_acr_new }, .bar = { 0x00000001, gm107_bar_new }, .bios = { 0x00000001, nvkm_bios_new }, .bus = { 0x00000001, gf100_bus_new }, @@ -2385,6 +2385,7 @@ nv140_chipset = { .therm = { 0x00000001, gp100_therm_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, gk104_top_new }, + .vfn = { 0x00000001, gv100_vfn_new }, .ce = { 0x000001ff, gv100_ce_new }, .disp = { 0x00000001, gv100_disp_new }, .dma = { 0x00000001, gv100_dma_new }, @@ -2411,7 +2412,7 @@ nv162_chipset = { .i2c = { 0x00000001, gm200_i2c_new }, .imem = { 0x00000001, nv50_instmem_new }, .ltc = { 0x00000001, gp102_ltc_new }, - .mc = { 0x00000001, tu102_mc_new }, + .mc = { 0x00000001, gp100_mc_new }, .mmu = { 0x00000001, tu102_mmu_new }, .pci = { 0x00000001, gp100_pci_new }, .pmu = { 0x00000001, gp102_pmu_new }, @@ -2419,6 +2420,7 @@ nv162_chipset = { .therm = { 0x00000001, gp100_therm_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, gk104_top_new }, + .vfn = { 0x00000001, tu102_vfn_new }, .ce = { 0x0000001f, tu102_ce_new }, .disp = { 0x00000001, tu102_disp_new }, .dma = { 0x00000001, gv100_dma_new }, @@ -2445,7 +2447,7 @@ nv164_chipset = { .i2c = { 0x00000001, gm200_i2c_new }, .imem = { 0x00000001, nv50_instmem_new }, .ltc = { 0x00000001, gp102_ltc_new }, - .mc = { 0x00000001, tu102_mc_new }, + .mc = { 0x00000001, gp100_mc_new }, .mmu = { 0x00000001, tu102_mmu_new }, .pci = { 0x00000001, gp100_pci_new }, .pmu = { 0x00000001, gp102_pmu_new }, @@ -2453,6 +2455,7 @@ nv164_chipset = { .therm = { 0x00000001, gp100_therm_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, gk104_top_new }, + .vfn = { 0x00000001, tu102_vfn_new }, .ce = { 0x0000001f, tu102_ce_new }, .disp = { 0x00000001, tu102_disp_new }, .dma = { 0x00000001, gv100_dma_new }, @@ -2479,7 +2482,7 @@ nv166_chipset = { .i2c = { 0x00000001, gm200_i2c_new }, .imem = { 0x00000001, nv50_instmem_new }, .ltc = { 0x00000001, gp102_ltc_new }, - .mc = { 0x00000001, tu102_mc_new }, + .mc = { 0x00000001, gp100_mc_new }, .mmu = { 0x00000001, tu102_mmu_new }, .pci = { 0x00000001, gp100_pci_new }, .pmu = { 0x00000001, gp102_pmu_new }, @@ -2487,6 +2490,7 @@ nv166_chipset = { .therm = { 0x00000001, gp100_therm_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, gk104_top_new }, + .vfn = { 0x00000001, tu102_vfn_new }, .ce = { 0x0000001f, tu102_ce_new }, .disp = { 0x00000001, tu102_disp_new }, .dma = { 0x00000001, gv100_dma_new }, @@ -2513,7 +2517,7 @@ nv167_chipset = { .i2c = { 0x00000001, gm200_i2c_new }, .imem = { 0x00000001, nv50_instmem_new }, .ltc = { 0x00000001, gp102_ltc_new }, - .mc = { 0x00000001, tu102_mc_new }, + .mc = { 0x00000001, gp100_mc_new }, .mmu = { 0x00000001, tu102_mmu_new }, .pci = { 0x00000001, gp100_pci_new }, .pmu = { 0x00000001, gp102_pmu_new }, @@ -2521,6 +2525,7 @@ nv167_chipset = { .therm = { 0x00000001, gp100_therm_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, gk104_top_new }, + .vfn = { 0x00000001, tu102_vfn_new }, .ce = { 0x0000001f, tu102_ce_new }, .disp = { 0x00000001, tu102_disp_new }, .dma = { 0x00000001, gv100_dma_new }, @@ -2547,7 +2552,7 @@ nv168_chipset = { .i2c = { 0x00000001, gm200_i2c_new }, .imem = { 0x00000001, nv50_instmem_new }, .ltc = { 0x00000001, gp102_ltc_new }, - .mc = { 0x00000001, tu102_mc_new }, + .mc = { 0x00000001, gp100_mc_new }, .mmu = { 0x00000001, tu102_mmu_new }, .pci = { 0x00000001, gp100_pci_new }, .pmu = { 0x00000001, gp102_pmu_new }, @@ -2555,6 +2560,7 @@ nv168_chipset = { .therm = { 0x00000001, gp100_therm_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, gk104_top_new }, + .vfn = { 0x00000001, tu102_vfn_new }, .ce = { 0x0000001f, tu102_ce_new }, .disp = { 0x00000001, tu102_disp_new }, .dma = { 0x00000001, gv100_dma_new }, @@ -2571,6 +2577,7 @@ nv170_chipset = { .bar = { 0x00000001, tu102_bar_new }, .bios = { 0x00000001, nvkm_bios_new }, .devinit = { 0x00000001, ga100_devinit_new }, + .fault = { 0x00000001, tu102_fault_new }, .fb = { 0x00000001, ga100_fb_new }, .gpio = { 0x00000001, gk104_gpio_new }, .i2c = { 0x00000001, gm200_i2c_new }, @@ -2581,111 +2588,159 @@ nv170_chipset = { .privring = { 0x00000001, gm200_privring_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, ga100_top_new }, + .vfn = { 0x00000001, ga100_vfn_new }, + .ce = { 0x000003ff, ga100_ce_new }, + .fifo = { 0x00000001, ga100_fifo_new }, }; static const struct nvkm_device_chip nv172_chipset = { .name = "GA102", + .acr = { 0x00000001, ga102_acr_new }, .bar = { 0x00000001, tu102_bar_new }, .bios = { 0x00000001, nvkm_bios_new }, .devinit = { 0x00000001, ga100_devinit_new }, + .fault = { 0x00000001, tu102_fault_new }, .fb = { 0x00000001, ga102_fb_new }, .gpio = { 0x00000001, ga102_gpio_new }, + .gsp = { 0x00000001, ga102_gsp_new }, .i2c = { 0x00000001, gm200_i2c_new }, .imem = { 0x00000001, nv50_instmem_new }, + .ltc = { 0x00000001, ga102_ltc_new }, .mc = { 0x00000001, ga100_mc_new }, .mmu = { 0x00000001, tu102_mmu_new }, .pci = { 0x00000001, gp100_pci_new }, .privring = { 0x00000001, gm200_privring_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, ga100_top_new }, + .vfn = { 0x00000001, ga100_vfn_new }, + .ce = { 0x0000001f, ga102_ce_new }, .disp = { 0x00000001, ga102_disp_new }, .dma = { 0x00000001, gv100_dma_new }, .fifo = { 0x00000001, ga102_fifo_new }, + .gr = { 0x00000001, ga102_gr_new }, + .nvdec = { 0x00000001, ga102_nvdec_new }, + .sec2 = { 0x00000001, ga102_sec2_new }, }; static const struct nvkm_device_chip nv173_chipset = { .name = "GA103", + .acr = { 0x00000001, ga102_acr_new }, .bar = { 0x00000001, tu102_bar_new }, .bios = { 0x00000001, nvkm_bios_new }, .devinit = { 0x00000001, ga100_devinit_new }, + .fault = { 0x00000001, tu102_fault_new }, .fb = { 0x00000001, ga102_fb_new }, .gpio = { 0x00000001, ga102_gpio_new }, + .gsp = { 0x00000001, ga102_gsp_new }, .i2c = { 0x00000001, gm200_i2c_new }, .imem = { 0x00000001, nv50_instmem_new }, + .ltc = { 0x00000001, ga102_ltc_new }, .mc = { 0x00000001, ga100_mc_new }, .mmu = { 0x00000001, tu102_mmu_new }, .pci = { 0x00000001, gp100_pci_new }, .privring = { 0x00000001, gm200_privring_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, ga100_top_new }, + .vfn = { 0x00000001, ga100_vfn_new }, + .ce = { 0x0000001f, ga102_ce_new }, .disp = { 0x00000001, ga102_disp_new }, .dma = { 0x00000001, gv100_dma_new }, .fifo = { 0x00000001, ga102_fifo_new }, + .gr = { 0x00000001, ga102_gr_new }, + .nvdec = { 0x00000001, ga102_nvdec_new }, + .sec2 = { 0x00000001, ga102_sec2_new }, }; static const struct nvkm_device_chip nv174_chipset = { .name = "GA104", + .acr = { 0x00000001, ga102_acr_new }, .bar = { 0x00000001, tu102_bar_new }, .bios = { 0x00000001, nvkm_bios_new }, .devinit = { 0x00000001, ga100_devinit_new }, + .fault = { 0x00000001, tu102_fault_new }, .fb = { 0x00000001, ga102_fb_new }, .gpio = { 0x00000001, ga102_gpio_new }, + .gsp = { 0x00000001, ga102_gsp_new }, .i2c = { 0x00000001, gm200_i2c_new }, .imem = { 0x00000001, nv50_instmem_new }, + .ltc = { 0x00000001, ga102_ltc_new }, .mc = { 0x00000001, ga100_mc_new }, .mmu = { 0x00000001, tu102_mmu_new }, .pci = { 0x00000001, gp100_pci_new }, .privring = { 0x00000001, gm200_privring_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, ga100_top_new }, + .vfn = { 0x00000001, ga100_vfn_new }, + .ce = { 0x0000001f, ga102_ce_new }, .disp = { 0x00000001, ga102_disp_new }, .dma = { 0x00000001, gv100_dma_new }, .fifo = { 0x00000001, ga102_fifo_new }, + .gr = { 0x00000001, ga102_gr_new }, + .nvdec = { 0x00000001, ga102_nvdec_new }, + .sec2 = { 0x00000001, ga102_sec2_new }, }; static const struct nvkm_device_chip nv176_chipset = { .name = "GA106", + .acr = { 0x00000001, ga102_acr_new }, .bar = { 0x00000001, tu102_bar_new }, .bios = { 0x00000001, nvkm_bios_new }, .devinit = { 0x00000001, ga100_devinit_new }, + .fault = { 0x00000001, tu102_fault_new }, .fb = { 0x00000001, ga102_fb_new }, .gpio = { 0x00000001, ga102_gpio_new }, + .gsp = { 0x00000001, ga102_gsp_new }, .i2c = { 0x00000001, gm200_i2c_new }, .imem = { 0x00000001, nv50_instmem_new }, + .ltc = { 0x00000001, ga102_ltc_new }, .mc = { 0x00000001, ga100_mc_new }, .mmu = { 0x00000001, tu102_mmu_new }, .pci = { 0x00000001, gp100_pci_new }, .privring = { 0x00000001, gm200_privring_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, ga100_top_new }, + .vfn = { 0x00000001, ga100_vfn_new }, + .ce = { 0x0000001f, ga102_ce_new }, .disp = { 0x00000001, ga102_disp_new }, .dma = { 0x00000001, gv100_dma_new }, .fifo = { 0x00000001, ga102_fifo_new }, + .gr = { 0x00000001, ga102_gr_new }, + .nvdec = { 0x00000001, ga102_nvdec_new }, + .sec2 = { 0x00000001, ga102_sec2_new }, }; static const struct nvkm_device_chip nv177_chipset = { .name = "GA107", + .acr = { 0x00000001, ga102_acr_new }, .bar = { 0x00000001, tu102_bar_new }, .bios = { 0x00000001, nvkm_bios_new }, .devinit = { 0x00000001, ga100_devinit_new }, + .fault = { 0x00000001, tu102_fault_new }, .fb = { 0x00000001, ga102_fb_new }, .gpio = { 0x00000001, ga102_gpio_new }, + .gsp = { 0x00000001, ga102_gsp_new }, .i2c = { 0x00000001, gm200_i2c_new }, .imem = { 0x00000001, nv50_instmem_new }, + .ltc = { 0x00000001, ga102_ltc_new }, .mc = { 0x00000001, ga100_mc_new }, .mmu = { 0x00000001, tu102_mmu_new }, .pci = { 0x00000001, gp100_pci_new }, .privring = { 0x00000001, gm200_privring_new }, .timer = { 0x00000001, gk20a_timer_new }, .top = { 0x00000001, ga100_top_new }, + .vfn = { 0x00000001, ga100_vfn_new }, + .ce = { 0x0000001f, ga102_ce_new }, .disp = { 0x00000001, ga102_disp_new }, .dma = { 0x00000001, gv100_dma_new }, .fifo = { 0x00000001, ga102_fifo_new }, + .gr = { 0x00000001, ga102_gr_new }, + .nvdec = { 0x00000001, ga102_nvdec_new }, + .sec2 = { 0x00000001, ga102_sec2_new }, }; struct nvkm_subdev * @@ -2734,6 +2789,8 @@ nvkm_device_fini(struct nvkm_device *device, bool suspend) if (device->func->fini) device->func->fini(device, suspend); + nvkm_intr_unarm(device); + time = ktime_to_us(ktime_get()) - time; nvdev_trace(device, "%s completed in %lldus...\n", action, time); return 0; @@ -2759,6 +2816,8 @@ nvkm_device_preinit(struct nvkm_device *device) nvdev_trace(device, "preinit running...\n"); time = ktime_to_us(ktime_get()); + nvkm_intr_unarm(device); + if (device->func->preinit) { ret = device->func->preinit(device); if (ret) @@ -2775,6 +2834,14 @@ nvkm_device_preinit(struct nvkm_device *device) if (ret) goto fail; + ret = nvkm_top_parse(device); + if (ret) + goto fail; + + ret = nvkm_fb_mem_unlock(device->fb); + if (ret) + goto fail; + time = ktime_to_us(ktime_get()) - time; nvdev_trace(device, "preinit completed in %lldus\n", time); return 0; @@ -2800,6 +2867,8 @@ nvkm_device_init(struct nvkm_device *device) nvdev_trace(device, "init running...\n"); time = ktime_to_us(ktime_get()); + nvkm_intr_rearm(device); + if (device->func->init) { ret = device->func->init(device); if (ret) @@ -2837,6 +2906,8 @@ nvkm_device_del(struct nvkm_device **pdevice) if (device) { mutex_lock(&nv_devices_mutex); + nvkm_intr_dtor(device); + list_for_each_entry_safe_reverse(subdev, subtmp, &device->subdev, head) nvkm_subdev_del(&subdev); @@ -3144,6 +3215,7 @@ nvkm_device_ctor(const struct nvkm_device_func *func, device->name = device->chip->name; mutex_init(&device->mutex); + nvkm_intr_ctor(device); #define NVKM_LAYOUT_ONCE(type,data,ptr) \ if (device->chip->ptr.inst && (subdev_mask & (BIT_ULL(type)))) { \ @@ -3185,7 +3257,7 @@ nvkm_device_ctor(const struct nvkm_device_func *func, #undef NVKM_LAYOUT_INST #undef NVKM_LAYOUT_ONCE - ret = 0; + ret = nvkm_intr_install(device); done: if (device->pri && (!mmio || ret)) { iounmap(device->pri); diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/pci.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/pci.c index f302d2b5782a..abccb2bb68a6 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/pci.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/pci.c @@ -1574,6 +1574,12 @@ nvkm_device_pci_resource_size(struct nvkm_device *device, unsigned bar) return pci_resource_len(pdev->pdev, bar); } +static int +nvkm_device_pci_irq(struct nvkm_device *device) +{ + return nvkm_device_pci(device)->pdev->irq; +} + static void nvkm_device_pci_fini(struct nvkm_device *device, bool suspend) { @@ -1612,6 +1618,7 @@ nvkm_device_pci_func = { .dtor = nvkm_device_pci_dtor, .preinit = nvkm_device_pci_preinit, .fini = nvkm_device_pci_fini, + .irq = nvkm_device_pci_irq, .resource_addr = nvkm_device_pci_resource_addr, .resource_size = nvkm_device_pci_resource_size, .cpu_coherent = !IS_ENABLED(CONFIG_ARM), diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/priv.h b/drivers/gpu/drm/nouveau/nvkm/engine/device/priv.h index 93949b3c7214..24faaac15891 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/priv.h @@ -27,6 +27,7 @@ #include <subdev/therm.h> #include <subdev/timer.h> #include <subdev/top.h> +#include <subdev/vfn.h> #include <subdev/volt.h> #include <engine/bsp.h> diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c index ac9e122586bc..87caa4a72921 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/tegra.c @@ -206,45 +206,12 @@ nvkm_device_tegra_resource_size(struct nvkm_device *device, unsigned bar) return res ? resource_size(res) : 0; } -static irqreturn_t -nvkm_device_tegra_intr(int irq, void *arg) -{ - struct nvkm_device_tegra *tdev = arg; - struct nvkm_device *device = &tdev->device; - bool handled = false; - nvkm_mc_intr_unarm(device); - nvkm_mc_intr(device, &handled); - nvkm_mc_intr_rearm(device); - return handled ? IRQ_HANDLED : IRQ_NONE; -} - -static void -nvkm_device_tegra_fini(struct nvkm_device *device, bool suspend) -{ - struct nvkm_device_tegra *tdev = nvkm_device_tegra(device); - if (tdev->irq) { - free_irq(tdev->irq, tdev); - tdev->irq = 0; - } -} - static int -nvkm_device_tegra_init(struct nvkm_device *device) +nvkm_device_tegra_irq(struct nvkm_device *device) { struct nvkm_device_tegra *tdev = nvkm_device_tegra(device); - int irq, ret; - - irq = platform_get_irq_byname(tdev->pdev, "stall"); - if (irq < 0) - return irq; - ret = request_irq(irq, nvkm_device_tegra_intr, - IRQF_SHARED, "nvkm", tdev); - if (ret) - return ret; - - tdev->irq = irq; - return 0; + return platform_get_irq_byname(tdev->pdev, "stall"); } static void * @@ -260,8 +227,7 @@ static const struct nvkm_device_func nvkm_device_tegra_func = { .tegra = nvkm_device_tegra, .dtor = nvkm_device_tegra_dtor, - .init = nvkm_device_tegra_init, - .fini = nvkm_device_tegra_fini, + .irq = nvkm_device_tegra_irq, .resource_addr = nvkm_device_tegra_resource_addr, .resource_size = nvkm_device_tegra_resource_size, .cpu_coherent = false, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/device/user.c b/drivers/gpu/drm/nouveau/nvkm/engine/device/user.c index 45f509c11c36..9b39ec341615 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/device/user.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/device/user.c @@ -342,6 +342,8 @@ nvkm_udevice_child_get(struct nvkm_object *object, int index, sclass = &device->mmu->user; else if (device->fault && index-- == 0) sclass = &device->fault->user; + else if (device->vfn && index-- == 0) + sclass = &device->vfn->user; else return -EINVAL; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/Kbuild b/drivers/gpu/drm/nouveau/nvkm/engine/disp/Kbuild index 600072a904be..e1aecd3fe96c 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/Kbuild @@ -28,9 +28,7 @@ nvkm-y += nvkm/engine/disp/gv100.o nvkm-y += nvkm/engine/disp/tu102.o nvkm-y += nvkm/engine/disp/ga102.o -nvkm-y += nvkm/engine/disp/rootnv04.o -nvkm-y += nvkm/engine/disp/rootnv50.o - nvkm-y += nvkm/engine/disp/udisp.o nvkm-y += nvkm/engine/disp/uconn.o nvkm-y += nvkm/engine/disp/uoutp.o +nvkm-y += nvkm/engine/disp/uhead.o diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/base.c index 65c99d948b68..73104b59f97f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/base.c @@ -29,7 +29,6 @@ #include "outp.h" #include <core/client.h> -#include <core/notify.h> #include <core/ramht.h> #include <subdev/bios.h> #include <subdev/bios/dcb.h> @@ -57,32 +56,8 @@ nvkm_disp_vblank_init(struct nvkm_event *event, int type, int id) head->func->vblank_get(head); } -static int -nvkm_disp_vblank_ctor(struct nvkm_object *object, void *data, u32 size, - struct nvkm_notify *notify) -{ - struct nvkm_disp *disp = - container_of(notify->event, typeof(*disp), vblank); - union { - struct nvif_notify_head_req_v0 v0; - } *req = data; - int ret = -ENOSYS; - - if (!(ret = nvif_unpack(ret, &data, &size, req->v0, 0, 0, false))) { - notify->size = sizeof(struct nvif_notify_head_rep_v0); - if (ret = -ENXIO, req->v0.head <= disp->vblank.index_nr) { - notify->types = 1; - notify->index = req->v0.head; - return 0; - } - } - - return ret; -} - static const struct nvkm_event_func nvkm_disp_vblank_func = { - .ctor = nvkm_disp_vblank_ctor, .init = nvkm_disp_vblank_init, .fini = nvkm_disp_vblank_fini, }; @@ -90,59 +65,7 @@ nvkm_disp_vblank_func = { void nvkm_disp_vblank(struct nvkm_disp *disp, int head) { - struct nvif_notify_head_rep_v0 rep = {}; - nvkm_event_send(&disp->vblank, 1, head, &rep, sizeof(rep)); -} - -static int -nvkm_disp_hpd_ctor(struct nvkm_object *object, void *data, u32 size, - struct nvkm_notify *notify) -{ - struct nvkm_disp *disp = - container_of(notify->event, typeof(*disp), hpd); - union { - struct nvif_notify_conn_req_v0 v0; - } *req = data; - struct nvkm_outp *outp; - int ret = -ENOSYS; - - if (!(ret = nvif_unpack(ret, &data, &size, req->v0, 0, 0, false))) { - notify->size = sizeof(struct nvif_notify_conn_rep_v0); - list_for_each_entry(outp, &disp->outps, head) { - if (ret = -ENXIO, outp->conn->index == req->v0.conn) { - if (ret = -ENODEV, outp->conn->hpd.event) { - notify->types = req->v0.mask; - notify->index = req->v0.conn; - ret = 0; - } - break; - } - } - } - - return ret; -} - -static const struct nvkm_event_func -nvkm_disp_hpd_func = { - .ctor = nvkm_disp_hpd_ctor -}; - -int -nvkm_disp_ntfy(struct nvkm_object *object, u32 type, struct nvkm_event **event) -{ - struct nvkm_disp *disp = nvkm_disp(object->engine); - switch (type) { - case NV04_DISP_NTFY_VBLANK: - *event = &disp->vblank; - return 0; - case NV04_DISP_NTFY_CONN: - *event = &disp->hpd; - return 0; - default: - break; - } - return -EINVAL; + nvkm_event_ntfy(&disp->vblank, head, NVKM_DISP_HEAD_EVENT_VBLANK); } static int @@ -343,9 +266,7 @@ nvkm_disp_oneinit(struct nvkm_engine *engine) /* Apparently we need to create a new one! */ ret = nvkm_conn_new(disp, i, &connE, &outp->conn); if (ret) { - nvkm_error(&disp->engine.subdev, - "failed to create outp %d conn: %d\n", - outp->index, ret); + nvkm_error(subdev, "failed to create outp %d conn: %d\n", outp->index, ret); nvkm_conn_del(&outp->conn); list_del(&outp->head); nvkm_outp_del(&outp); @@ -355,10 +276,6 @@ nvkm_disp_oneinit(struct nvkm_engine *engine) list_add_tail(&outp->conn->head, &disp->conns); } - ret = nvkm_event_init(&nvkm_disp_hpd_func, 3, hpd, &disp->hpd); - if (ret) - return ret; - if (disp->func->oneinit) { ret = disp->func->oneinit(disp); if (ret) @@ -382,7 +299,7 @@ nvkm_disp_oneinit(struct nvkm_engine *engine) list_for_each_entry(head, &disp->heads, head) i = max(i, head->id + 1); - return nvkm_event_init(&nvkm_disp_vblank_func, 1, i, &disp->vblank); + return nvkm_event_init(&nvkm_disp_vblank_func, subdev, 1, i, &disp->vblank); } static void * @@ -406,7 +323,6 @@ nvkm_disp_dtor(struct nvkm_engine *engine) } nvkm_event_fini(&disp->vblank); - nvkm_event_fini(&disp->hpd); while (!list_empty(&disp->conns)) { conn = list_first_entry(&disp->conns, typeof(*conn), head); @@ -473,5 +389,6 @@ nvkm_disp_new_(const struct nvkm_disp_func *func, struct nvkm_device *device, mutex_init(&disp->super.mutex); } - return nvkm_event_init(func->uevent, 1, ARRAY_SIZE(disp->chan), &disp->uevent); + return nvkm_event_init(func->uevent, &disp->engine.subdev, 1, ARRAY_SIZE(disp->chan), + &disp->uevent); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/conn.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/conn.c index 7ed11801a3ae..fbdae1137864 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/conn.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/conn.c @@ -29,38 +29,14 @@ #include <nvif/event.h> -static int -nvkm_conn_hpd(struct nvkm_notify *notify) -{ - struct nvkm_conn *conn = container_of(notify, typeof(*conn), hpd); - struct nvkm_disp *disp = conn->disp; - struct nvkm_gpio *gpio = disp->engine.subdev.device->gpio; - const struct nvkm_gpio_ntfy_rep *line = notify->data; - struct nvif_notify_conn_rep_v0 rep; - int index = conn->index; - - CONN_DBG(conn, "HPD: %d", line->mask); - - if (!nvkm_gpio_get(gpio, 0, DCB_GPIO_UNUSED, conn->hpd.index)) - rep.mask = NVIF_NOTIFY_CONN_V0_UNPLUG; - else - rep.mask = NVIF_NOTIFY_CONN_V0_PLUG; - rep.version = 0; - - nvkm_event_send(&disp->hpd, rep.mask, index, &rep, sizeof(rep)); - return NVKM_NOTIFY_KEEP; -} - void nvkm_conn_fini(struct nvkm_conn *conn) { - nvkm_notify_put(&conn->hpd); } void nvkm_conn_init(struct nvkm_conn *conn) { - nvkm_notify_get(&conn->hpd); } void @@ -68,7 +44,6 @@ nvkm_conn_del(struct nvkm_conn **pconn) { struct nvkm_conn *conn = *pconn; if (conn) { - nvkm_notify_fini(&conn->hpd); kfree(*pconn); *pconn = NULL; } @@ -106,20 +81,6 @@ nvkm_conn_ctor(struct nvkm_disp *disp, int index, struct nvbios_connE *info, } conn->info.hpd = func.line; - - ret = nvkm_notify_init(NULL, &gpio->event, nvkm_conn_hpd, - true, &(struct nvkm_gpio_ntfy_req) { - .mask = NVKM_GPIO_TOGGLED, - .line = func.line, - }, - sizeof(struct nvkm_gpio_ntfy_req), - sizeof(struct nvkm_gpio_ntfy_rep), - &conn->hpd); - if (ret) { - CONN_ERR(conn, "func %02x failed, %d", info->hpd, ret); - } else { - CONN_DBG(conn, "func %02x (HPD)", info->hpd); - } } } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/conn.h b/drivers/gpu/drm/nouveau/nvkm/engine/disp/conn.h index f109634ce5ca..a0600e72b0ec 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/conn.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/conn.h @@ -3,7 +3,6 @@ #define __NVKM_DISP_CONN_H__ #include "priv.h" -#include <core/notify.h> #include <subdev/bios.h> #include <subdev/bios/conn.h> @@ -12,8 +11,6 @@ struct nvkm_conn { int index; struct nvbios_connE info; - struct nvkm_notify hpd; - struct list_head head; struct nvkm_object object; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c index c1b3206f27e6..40c8ea43c42f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.c @@ -274,70 +274,17 @@ nvkm_dp_train_cr(struct lt_state *lt) } static int -nvkm_dp_train_links(struct nvkm_outp *outp, int rate) +nvkm_dp_train_link(struct nvkm_outp *outp, int rate) { struct nvkm_ior *ior = outp->ior; - struct nvkm_disp *disp = outp->disp; - struct nvkm_subdev *subdev = &disp->engine.subdev; - struct nvkm_bios *bios = subdev->device->bios; struct lt_state lt = { .outp = outp, + .pc2 = outp->dp.dpcd[DPCD_RC02] & DPCD_RC02_TPS3_SUPPORTED, }; - u32 lnkcmp; u8 sink[2], data; int ret; - OUTP_DBG(outp, "training %d x %d MB/s", ior->dp.nr, ior->dp.bw * 27); - - /* Intersect misc. capabilities of the OR and sink. */ - if (disp->engine.subdev.device->chipset < 0x110) - outp->dp.dpcd[DPCD_RC03] &= ~DPCD_RC03_TPS4_SUPPORTED; - if (disp->engine.subdev.device->chipset < 0xd0) - outp->dp.dpcd[DPCD_RC02] &= ~DPCD_RC02_TPS3_SUPPORTED; - lt.pc2 = outp->dp.dpcd[DPCD_RC02] & DPCD_RC02_TPS3_SUPPORTED; - - if (AMPERE_IED_HACK(disp) && (lnkcmp = lt.outp->dp.info.script[0])) { - /* Execute BeforeLinkTraining script from DP Info table. */ - while (ior->dp.bw < nvbios_rd08(bios, lnkcmp)) - lnkcmp += 3; - lnkcmp = nvbios_rd16(bios, lnkcmp + 1); - - nvbios_init(&outp->disp->engine.subdev, lnkcmp, - init.outp = &outp->info; - init.or = ior->id; - init.link = ior->asy.link; - ); - } - - /* Set desired link configuration on the source. */ - if ((lnkcmp = lt.outp->dp.info.lnkcmp)) { - if (outp->dp.version < 0x30) { - while ((ior->dp.bw * 2700) < nvbios_rd16(bios, lnkcmp)) - lnkcmp += 4; - lnkcmp = nvbios_rd16(bios, lnkcmp + 2); - } else { - while (ior->dp.bw < nvbios_rd08(bios, lnkcmp)) - lnkcmp += 3; - lnkcmp = nvbios_rd16(bios, lnkcmp + 1); - } - - nvbios_init(subdev, lnkcmp, - init.outp = &outp->info; - init.or = ior->id; - init.link = ior->asy.link; - ); - } - - ret = ior->func->dp->links(ior, outp->dp.aux); - if (ret) { - if (ret < 0) { - OUTP_ERR(outp, "train failed with %d", ret); - return ret; - } - return 0; - } - - ior->func->dp->power(ior, ior->dp.nr); + OUTP_DBG(outp, "training %dx%02x", ior->dp.nr, ior->dp.bw); /* Select LTTPR non-transparent mode if we have a valid configuration, * use transparent mode otherwise. @@ -393,6 +340,71 @@ nvkm_dp_train_links(struct nvkm_outp *outp, int rate) return ret; } +static int +nvkm_dp_train_links(struct nvkm_outp *outp, int rate) +{ + struct nvkm_ior *ior = outp->ior; + struct nvkm_disp *disp = outp->disp; + struct nvkm_subdev *subdev = &disp->engine.subdev; + struct nvkm_bios *bios = subdev->device->bios; + u32 lnkcmp; + int ret; + + OUTP_DBG(outp, "programming link for %dx%02x", ior->dp.nr, ior->dp.bw); + + /* Intersect misc. capabilities of the OR and sink. */ + if (disp->engine.subdev.device->chipset < 0x110) + outp->dp.dpcd[DPCD_RC03] &= ~DPCD_RC03_TPS4_SUPPORTED; + if (disp->engine.subdev.device->chipset < 0xd0) + outp->dp.dpcd[DPCD_RC02] &= ~DPCD_RC02_TPS3_SUPPORTED; + + if (AMPERE_IED_HACK(disp) && (lnkcmp = outp->dp.info.script[0])) { + /* Execute BeforeLinkTraining script from DP Info table. */ + while (ior->dp.bw < nvbios_rd08(bios, lnkcmp)) + lnkcmp += 3; + lnkcmp = nvbios_rd16(bios, lnkcmp + 1); + + nvbios_init(&outp->disp->engine.subdev, lnkcmp, + init.outp = &outp->info; + init.or = ior->id; + init.link = ior->asy.link; + ); + } + + /* Set desired link configuration on the source. */ + if ((lnkcmp = outp->dp.info.lnkcmp)) { + if (outp->dp.version < 0x30) { + while ((ior->dp.bw * 2700) < nvbios_rd16(bios, lnkcmp)) + lnkcmp += 4; + lnkcmp = nvbios_rd16(bios, lnkcmp + 2); + } else { + while (ior->dp.bw < nvbios_rd08(bios, lnkcmp)) + lnkcmp += 3; + lnkcmp = nvbios_rd16(bios, lnkcmp + 1); + } + + nvbios_init(subdev, lnkcmp, + init.outp = &outp->info; + init.or = ior->id; + init.link = ior->asy.link; + ); + } + + ret = ior->func->dp->links(ior, outp->dp.aux); + if (ret) { + if (ret < 0) { + OUTP_ERR(outp, "train failed with %d", ret); + return ret; + } + return 0; + } + + ior->func->dp->power(ior, ior->dp.nr); + + /* Attempt to train the link in this configuration. */ + return nvkm_dp_train_link(outp, rate); +} + static void nvkm_dp_train_fini(struct nvkm_outp *outp) { @@ -439,6 +451,16 @@ nvkm_dp_train(struct nvkm_outp *outp, u32 dataKBps) int ret = -EINVAL, nr, rate; u8 pwr; + /* Retraining link? Skip source configuration, it can mess up the active modeset. */ + if (atomic_read(&outp->dp.lt.done)) { + for (rate = 0; rate < outp->dp.rates; rate++) { + if (outp->dp.rate[rate].rate == ior->dp.bw * 27000) + return nvkm_dp_train_link(outp, ret); + } + WARN_ON(1); + return -EINVAL; + } + /* Ensure sink is not in a low-power state. */ if (!nvkm_rdaux(outp->dp.aux, DPCD_SC00, &pwr, 1)) { if ((pwr & DPCD_SC00_SET_POWER) != DPCD_SC00_SET_POWER_D0) { @@ -455,6 +477,21 @@ nvkm_dp_train(struct nvkm_outp *outp, u32 dataKBps) /* Link training. */ OUTP_DBG(outp, "training"); nvkm_dp_train_init(outp); + + /* Validate and train at configuration requested (if any) on ACQUIRE. */ + if (outp->dp.lt.nr) { + for (nr = outp->dp.links; ret < 0 && nr; nr >>= 1) { + for (rate = 0; nr == outp->dp.lt.nr && rate < outp->dp.rates; rate++) { + if (outp->dp.rate[rate].rate / 27000 == outp->dp.lt.bw) { + ior->dp.bw = outp->dp.rate[rate].rate / 27000; + ior->dp.nr = nr; + ret = nvkm_dp_train_links(outp, rate); + } + } + } + } + + /* Otherwise, loop through all valid link configurations that support the data rate. */ for (nr = outp->dp.links; ret < 0 && nr; nr >>= 1) { for (rate = 0; ret < 0 && rate < outp->dp.rates; rate++) { if (outp->dp.rate[rate].rate * nr >= dataKBps || WARN_ON(!ior->dp.nr)) { @@ -465,6 +502,8 @@ nvkm_dp_train(struct nvkm_outp *outp, u32 dataKBps) } } } + + /* Finish up. */ nvkm_dp_train_fini(outp); if (ret < 0) OUTP_ERR(outp, "training failed"); @@ -595,18 +634,38 @@ nvkm_dp_enable_supported_link_rates(struct nvkm_outp *outp) return outp->dp.rates != 0; } -static bool -nvkm_dp_enable(struct nvkm_outp *outp, bool enable) +void +nvkm_dp_enable(struct nvkm_outp *outp, bool auxpwr) { + struct nvkm_gpio *gpio = outp->disp->engine.subdev.device->gpio; struct nvkm_i2c_aux *aux = outp->dp.aux; - if (enable) { - if (!outp->dp.present) { - OUTP_DBG(outp, "aux power -> always"); - nvkm_i2c_aux_monitor(aux, true); - outp->dp.present = true; + if (auxpwr && !outp->dp.aux_pwr) { + /* eDP panels need powering on by us (if the VBIOS doesn't default it + * to on) before doing any AUX channel transactions. LVDS panel power + * is handled by the SOR itself, and not required for LVDS DDC. + */ + if (outp->conn->info.type == DCB_CONNECTOR_eDP) { + int power = nvkm_gpio_get(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff); + if (power == 0) { + nvkm_gpio_set(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff, 1); + outp->dp.aux_pwr_pu = true; + } + + /* We delay here unconditionally, even if already powered, + * because some laptop panels having a significant resume + * delay before the panel begins responding. + * + * This is likely a bit of a hack, but no better idea for + * handling this at the moment. + */ + msleep(300); } + OUTP_DBG(outp, "aux power -> always"); + nvkm_i2c_aux_monitor(aux, true); + outp->dp.aux_pwr = true; + /* Detect any LTTPRs before reading DPCD receiver caps. */ if (!nvkm_rdaux(aux, DPCD_LTTPR_REV, outp->dp.lttpr, sizeof(outp->dp.lttpr)) && outp->dp.lttpr[0] >= 0x14 && outp->dp.lttpr[2]) { @@ -659,96 +718,41 @@ nvkm_dp_enable(struct nvkm_outp *outp, bool enable) outp->dp.rates++; } } - - return true; } - } - - if (outp->dp.present) { + } else + if (!auxpwr && outp->dp.aux_pwr) { OUTP_DBG(outp, "aux power -> demand"); nvkm_i2c_aux_monitor(aux, false); - outp->dp.present = false; - } - - atomic_set(&outp->dp.lt.done, 0); - return false; -} - -static int -nvkm_dp_hpd(struct nvkm_notify *notify) -{ - const struct nvkm_i2c_ntfy_rep *line = notify->data; - struct nvkm_outp *outp = container_of(notify, typeof(*outp), dp.hpd); - struct nvkm_conn *conn = outp->conn; - struct nvkm_disp *disp = outp->disp; - struct nvif_notify_conn_rep_v0 rep = {}; + outp->dp.aux_pwr = false; + atomic_set(&outp->dp.lt.done, 0); - OUTP_DBG(outp, "HPD: %d", line->mask); - if (line->mask & NVKM_I2C_IRQ) { - if (atomic_read(&outp->dp.lt.done)) - outp->func->acquire(outp); - rep.mask |= NVIF_NOTIFY_CONN_V0_IRQ; - } else { - nvkm_dp_enable(outp, true); + /* Restore eDP panel GPIO to its prior state if we changed it, as + * it could potentially interfere with other outputs. + */ + if (outp->conn->info.type == DCB_CONNECTOR_eDP) { + if (outp->dp.aux_pwr_pu) { + nvkm_gpio_set(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff, 0); + outp->dp.aux_pwr_pu = false; + } + } } - - if (line->mask & NVKM_I2C_UNPLUG) - rep.mask |= NVIF_NOTIFY_CONN_V0_UNPLUG; - if (line->mask & NVKM_I2C_PLUG) - rep.mask |= NVIF_NOTIFY_CONN_V0_PLUG; - - nvkm_event_send(&disp->hpd, rep.mask, conn->index, &rep, sizeof(rep)); - return NVKM_NOTIFY_KEEP; } static void nvkm_dp_fini(struct nvkm_outp *outp) { - nvkm_notify_put(&outp->dp.hpd); nvkm_dp_enable(outp, false); } static void nvkm_dp_init(struct nvkm_outp *outp) { - struct nvkm_gpio *gpio = outp->disp->engine.subdev.device->gpio; - - nvkm_notify_put(&outp->conn->hpd); - - /* eDP panels need powering on by us (if the VBIOS doesn't default it - * to on) before doing any AUX channel transactions. LVDS panel power - * is handled by the SOR itself, and not required for LVDS DDC. - */ - if (outp->conn->info.type == DCB_CONNECTOR_eDP) { - int power = nvkm_gpio_get(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff); - if (power == 0) - nvkm_gpio_set(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff, 1); - - /* We delay here unconditionally, even if already powered, - * because some laptop panels having a significant resume - * delay before the panel begins responding. - * - * This is likely a bit of a hack, but no better idea for - * handling this at the moment. - */ - msleep(300); - - /* If the eDP panel can't be detected, we need to restore - * the panel power GPIO to avoid breaking another output. - */ - if (!nvkm_dp_enable(outp, true) && power == 0) - nvkm_gpio_set(gpio, 0, DCB_GPIO_PANEL_POWER, 0xff, 0); - } else { - nvkm_dp_enable(outp, true); - } - - nvkm_notify_get(&outp->dp.hpd); + nvkm_dp_enable(outp, outp->dp.enabled); } static void * nvkm_dp_dtor(struct nvkm_outp *outp) { - nvkm_notify_fini(&outp->dp.hpd); return outp; } @@ -797,21 +801,6 @@ nvkm_dp_new(struct nvkm_disp *disp, int index, struct dcb_output *dcbE, struct n OUTP_DBG(outp, "bios dp %02x %02x %02x %02x", outp->dp.version, hdr, cnt, len); - /* hotplug detect, replaces gpio-based mechanism with aux events */ - ret = nvkm_notify_init(NULL, &i2c->event, nvkm_dp_hpd, true, - &(struct nvkm_i2c_ntfy_req) { - .mask = NVKM_I2C_PLUG | NVKM_I2C_UNPLUG | - NVKM_I2C_IRQ, - .port = outp->dp.aux->id, - }, - sizeof(struct nvkm_i2c_ntfy_req), - sizeof(struct nvkm_i2c_ntfy_rep), - &outp->dp.hpd); - if (ret) { - OUTP_ERR(outp, "error monitoring aux hpd: %d", ret); - return ret; - } - mutex_init(&outp->dp.mutex); atomic_set(&outp->dp.lt.done, 0); return 0; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.h b/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.h index 1d86baa6a424..9a6be43916bc 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/dp.h @@ -6,6 +6,7 @@ int nvkm_dp_new(struct nvkm_disp *, int index, struct dcb_output *, struct nvkm_outp **); void nvkm_dp_disable(struct nvkm_outp *, struct nvkm_ior *); +void nvkm_dp_enable(struct nvkm_outp *, bool auxpwr); /* DPCD Receiver Capabilities */ #define DPCD_RC00_DPCD_REV 0x00000 diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/g84.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/g84.c index 4966a51af3d7..23ae451ba473 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/g84.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/g84.c @@ -29,9 +29,54 @@ #include <nvif/class.h> -void -g84_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, - u8 rekey, u8 *avi, u8 avi_size, u8 *vendor, u8 vendor_size) +static void +g84_sor_hdmi_infoframe_vsi(struct nvkm_ior *ior, int head, void *data, u32 size) +{ + struct nvkm_device *device = ior->disp->engine.subdev.device; + struct packed_hdmi_infoframe vsi; + const u32 hoff = head * 0x800; + + nvkm_mask(device, 0x61653c + hoff, 0x00010001, 0x00010000); + if (!size) + return; + + pack_hdmi_infoframe(&vsi, data, size); + + nvkm_wr32(device, 0x616544 + hoff, vsi.header); + nvkm_wr32(device, 0x616548 + hoff, vsi.subpack0_low); + nvkm_wr32(device, 0x61654c + hoff, vsi.subpack0_high); + /* Is there a second (or up to fourth?) set of subpack registers here? */ + /* nvkm_wr32(device, 0x616550 + hoff, vsi.subpack1_low); */ + /* nvkm_wr32(device, 0x616554 + hoff, vsi.subpack1_high); */ + + nvkm_mask(device, 0x61653c + hoff, 0x00010001, 0x00010001); +} + +static void +g84_sor_hdmi_infoframe_avi(struct nvkm_ior *ior, int head, void *data, u32 size) +{ + struct nvkm_device *device = ior->disp->engine.subdev.device; + struct packed_hdmi_infoframe avi; + const u32 hoff = head * 0x800; + + pack_hdmi_infoframe(&avi, data, size); + + nvkm_mask(device, 0x616520 + hoff, 0x00000001, 0x00000000); + if (!size) + return; + + nvkm_wr32(device, 0x616528 + hoff, avi.header); + nvkm_wr32(device, 0x61652c + hoff, avi.subpack0_low); + nvkm_wr32(device, 0x616530 + hoff, avi.subpack0_high); + nvkm_wr32(device, 0x616534 + hoff, avi.subpack1_low); + nvkm_wr32(device, 0x616538 + hoff, avi.subpack1_high); + + nvkm_mask(device, 0x616520 + hoff, 0x00000001, 0x00000001); +} + + +static void +g84_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, u8 rekey) { struct nvkm_device *device = ior->disp->engine.subdev.device; const u32 ctrl = 0x40000000 * enable | @@ -39,31 +84,13 @@ g84_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, max_ac_packet << 16 | rekey; const u32 hoff = head * 0x800; - struct packed_hdmi_infoframe avi_infoframe; - struct packed_hdmi_infoframe vendor_infoframe; - - pack_hdmi_infoframe(&avi_infoframe, avi, avi_size); - pack_hdmi_infoframe(&vendor_infoframe, vendor, vendor_size); if (!(ctrl & 0x40000000)) { nvkm_mask(device, 0x6165a4 + hoff, 0x40000000, 0x00000000); - nvkm_mask(device, 0x61653c + hoff, 0x00000001, 0x00000000); - nvkm_mask(device, 0x616520 + hoff, 0x00000001, 0x00000000); nvkm_mask(device, 0x616500 + hoff, 0x00000001, 0x00000000); return; } - /* AVI InfoFrame */ - nvkm_mask(device, 0x616520 + hoff, 0x00000001, 0x00000000); - if (avi_size) { - nvkm_wr32(device, 0x616528 + hoff, avi_infoframe.header); - nvkm_wr32(device, 0x61652c + hoff, avi_infoframe.subpack0_low); - nvkm_wr32(device, 0x616530 + hoff, avi_infoframe.subpack0_high); - nvkm_wr32(device, 0x616534 + hoff, avi_infoframe.subpack1_low); - nvkm_wr32(device, 0x616538 + hoff, avi_infoframe.subpack1_high); - nvkm_mask(device, 0x616520 + hoff, 0x00000001, 0x00000001); - } - /* Audio InfoFrame */ nvkm_mask(device, 0x616500 + hoff, 0x00000001, 0x00000000); nvkm_wr32(device, 0x616508 + hoff, 0x000a0184); @@ -71,17 +98,6 @@ g84_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, nvkm_wr32(device, 0x616510 + hoff, 0x00000000); nvkm_mask(device, 0x616500 + hoff, 0x00000001, 0x00000001); - /* Vendor InfoFrame */ - nvkm_mask(device, 0x61653c + hoff, 0x00010001, 0x00010000); - if (vendor_size) { - nvkm_wr32(device, 0x616544 + hoff, vendor_infoframe.header); - nvkm_wr32(device, 0x616548 + hoff, vendor_infoframe.subpack0_low); - nvkm_wr32(device, 0x61654c + hoff, vendor_infoframe.subpack0_high); - /* Is there a second (or up to fourth?) set of subpack registers here? */ - /* nvkm_wr32(device, 0x616550 + hoff, vendor_infoframe->subpack1_low); */ - /* nvkm_wr32(device, 0x616554 + hoff, vendor_infoframe->subpack1_high); */ - nvkm_mask(device, 0x61653c + hoff, 0x00010001, 0x00010001); - } nvkm_mask(device, 0x6165d0 + hoff, 0x00070001, 0x00010001); /* SPARE, HW_CTS */ nvkm_mask(device, 0x616568 + hoff, 0x00010101, 0x00000000); /* ACR_CTRL, ?? */ @@ -96,14 +112,19 @@ g84_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, nvkm_mask(device, 0x6165a4 + hoff, 0x5f1f007f, ctrl); } +const struct nvkm_ior_func_hdmi +g84_sor_hdmi = { + .ctrl = g84_sor_hdmi_ctrl, + .infoframe_avi = g84_sor_hdmi_infoframe_avi, + .infoframe_vsi = g84_sor_hdmi_infoframe_vsi, +}; + static const struct nvkm_ior_func g84_sor = { .state = nv50_sor_state, .power = nv50_sor_power, .clock = nv50_sor_clock, - .hdmi = { - .ctrl = g84_sor_hdmi_ctrl, - }, + .hdmi = &g84_sor_hdmi, }; int diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/ga102.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/ga102.c index 7489d0d7fce0..52099b75f52a 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/ga102.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/ga102.c @@ -105,10 +105,7 @@ ga102_sor = { .state = gv100_sor_state, .power = nv50_sor_power, .clock = ga102_sor_clock, - .hdmi = { - .ctrl = gv100_sor_hdmi_ctrl, - .scdc = gm200_sor_hdmi_scdc, - }, + .hdmi = &gv100_sor_hdmi, .dp = &ga102_sor_dp, .hda = &gv100_sor_hda, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gf119.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gf119.c index 39822f1b5b95..a48e9bdf4cd0 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gf119.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gf119.c @@ -202,19 +202,61 @@ gf119_sor_dp = { }; static void -gf119_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, - u8 rekey, u8 *avi, u8 avi_size, u8 *vendor, u8 vendor_size) +gf119_sor_hdmi_infoframe_vsi(struct nvkm_ior *ior, int head, void *data, u32 size) +{ + struct nvkm_device *device = ior->disp->engine.subdev.device; + struct packed_hdmi_infoframe vsi; + const u32 hoff = head * 0x800; + + pack_hdmi_infoframe(&vsi, data, size); + + nvkm_mask(device, 0x616730 + hoff, 0x00010001, 0x00010000); + if (!size) + return; + + /* + * These appear to be the audio infoframe registers, + * but no other set of infoframe registers has yet + * been found. + */ + nvkm_wr32(device, 0x616738 + hoff, vsi.header); + nvkm_wr32(device, 0x61673c + hoff, vsi.subpack0_low); + nvkm_wr32(device, 0x616740 + hoff, vsi.subpack0_high); + /* Is there a second (or further?) set of subpack registers here? */ + + nvkm_mask(device, 0x616730 + hoff, 0x00000001, 0x00000001); +} + +static void +gf119_sor_hdmi_infoframe_avi(struct nvkm_ior *ior, int head, void *data, u32 size) +{ + struct nvkm_device *device = ior->disp->engine.subdev.device; + struct packed_hdmi_infoframe avi; + const u32 hoff = head * 0x800; + + pack_hdmi_infoframe(&avi, data, size); + + nvkm_mask(device, 0x616714 + hoff, 0x00000001, 0x00000000); + if (!size) + return; + + nvkm_wr32(device, 0x61671c + hoff, avi.header); + nvkm_wr32(device, 0x616720 + hoff, avi.subpack0_low); + nvkm_wr32(device, 0x616724 + hoff, avi.subpack0_high); + nvkm_wr32(device, 0x616728 + hoff, avi.subpack1_low); + nvkm_wr32(device, 0x61672c + hoff, avi.subpack1_high); + + nvkm_mask(device, 0x616714 + hoff, 0x00000001, 0x00000001); +} + +static void +gf119_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, u8 rekey) { struct nvkm_device *device = ior->disp->engine.subdev.device; const u32 ctrl = 0x40000000 * enable | max_ac_packet << 16 | rekey; const u32 hoff = head * 0x800; - struct packed_hdmi_infoframe avi_infoframe; - struct packed_hdmi_infoframe vendor_infoframe; - - pack_hdmi_infoframe(&avi_infoframe, avi, avi_size); - pack_hdmi_infoframe(&vendor_infoframe, vendor, vendor_size); if (!(ctrl & 0x40000000)) { nvkm_mask(device, 0x616798 + hoff, 0x40000000, 0x00000000); @@ -224,32 +266,6 @@ gf119_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe return; } - /* AVI InfoFrame */ - nvkm_mask(device, 0x616714 + hoff, 0x00000001, 0x00000000); - if (avi_size) { - nvkm_wr32(device, 0x61671c + hoff, avi_infoframe.header); - nvkm_wr32(device, 0x616720 + hoff, avi_infoframe.subpack0_low); - nvkm_wr32(device, 0x616724 + hoff, avi_infoframe.subpack0_high); - nvkm_wr32(device, 0x616728 + hoff, avi_infoframe.subpack1_low); - nvkm_wr32(device, 0x61672c + hoff, avi_infoframe.subpack1_high); - nvkm_mask(device, 0x616714 + hoff, 0x00000001, 0x00000001); - } - - /* GENERIC(?) / Vendor InfoFrame? */ - nvkm_mask(device, 0x616730 + hoff, 0x00010001, 0x00010000); - if (vendor_size) { - /* - * These appear to be the audio infoframe registers, - * but no other set of infoframe registers has yet - * been found. - */ - nvkm_wr32(device, 0x616738 + hoff, vendor_infoframe.header); - nvkm_wr32(device, 0x61673c + hoff, vendor_infoframe.subpack0_low); - nvkm_wr32(device, 0x616740 + hoff, vendor_infoframe.subpack0_high); - /* Is there a second (or further?) set of subpack registers here? */ - nvkm_mask(device, 0x616730 + hoff, 0x00000001, 0x00000001); - } - /* ??? InfoFrame? */ nvkm_mask(device, 0x6167a4 + hoff, 0x00000001, 0x00000000); nvkm_wr32(device, 0x6167ac + hoff, 0x00000010); @@ -259,6 +275,13 @@ gf119_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe nvkm_mask(device, 0x616798 + hoff, 0x401f007f, ctrl); } +static const struct nvkm_ior_func_hdmi +gf119_sor_hdmi = { + .ctrl = gf119_sor_hdmi_ctrl, + .infoframe_avi = gf119_sor_hdmi_infoframe_avi, + .infoframe_vsi = gf119_sor_hdmi_infoframe_vsi, +}; + void gf119_sor_clock(struct nvkm_ior *sor) { @@ -305,9 +328,7 @@ gf119_sor = { .state = gf119_sor_state, .power = nv50_sor_power, .clock = gf119_sor_clock, - .hdmi = { - .ctrl = gf119_sor_hdmi_ctrl, - }, + .hdmi = &gf119_sor_hdmi, .dp = &gf119_sor_dp, .hda = &gf119_sor_hda, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gk104.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gk104.c index 7248e9ec835e..876a21a0cebb 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gk104.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gk104.c @@ -30,8 +30,51 @@ #include <nvif/class.h> void -gk104_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, - u8 rekey, u8 *avi, u8 avi_size, u8 *vendor, u8 vendor_size) +gk104_sor_hdmi_infoframe_vsi(struct nvkm_ior *ior, int head, void *data, u32 size) +{ + struct nvkm_device *device = ior->disp->engine.subdev.device; + struct packed_hdmi_infoframe vsi; + const u32 hoff = head * 0x400; + + pack_hdmi_infoframe(&vsi, data, size); + + /* GENERIC(?) / Vendor InfoFrame? */ + nvkm_mask(device, 0x690100 + hoff, 0x00010001, 0x00000000); + if (!size) + return; + + nvkm_wr32(device, 0x690108 + hoff, vsi.header); + nvkm_wr32(device, 0x69010c + hoff, vsi.subpack0_low); + nvkm_wr32(device, 0x690110 + hoff, vsi.subpack0_high); + /* Is there a second (or further?) set of subpack registers here? */ + nvkm_mask(device, 0x690100 + hoff, 0x00000001, 0x00000001); +} + +void +gk104_sor_hdmi_infoframe_avi(struct nvkm_ior *ior, int head, void *data, u32 size) +{ + struct nvkm_device *device = ior->disp->engine.subdev.device; + struct packed_hdmi_infoframe avi; + const u32 hoff = head * 0x400; + + pack_hdmi_infoframe(&avi, data, size); + + /* AVI InfoFrame */ + nvkm_mask(device, 0x690000 + hoff, 0x00000001, 0x00000000); + if (!size) + return; + + nvkm_wr32(device, 0x690008 + hoff, avi.header); + nvkm_wr32(device, 0x69000c + hoff, avi.subpack0_low); + nvkm_wr32(device, 0x690010 + hoff, avi.subpack0_high); + nvkm_wr32(device, 0x690014 + hoff, avi.subpack1_low); + nvkm_wr32(device, 0x690018 + hoff, avi.subpack1_high); + + nvkm_mask(device, 0x690000 + hoff, 0x00000001, 0x00000001); +} + +void +gk104_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, u8 rekey) { struct nvkm_device *device = ior->disp->engine.subdev.device; const u32 ctrl = 0x40000000 * enable | @@ -39,11 +82,6 @@ gk104_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe rekey; const u32 hoff = head * 0x800; const u32 hdmi = head * 0x400; - struct packed_hdmi_infoframe avi_infoframe; - struct packed_hdmi_infoframe vendor_infoframe; - - pack_hdmi_infoframe(&avi_infoframe, avi, avi_size); - pack_hdmi_infoframe(&vendor_infoframe, vendor, vendor_size); if (!(ctrl & 0x40000000)) { nvkm_mask(device, 0x616798 + hoff, 0x40000000, 0x00000000); @@ -53,28 +91,6 @@ gk104_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe return; } - /* AVI InfoFrame */ - nvkm_mask(device, 0x690000 + hdmi, 0x00000001, 0x00000000); - if (avi_size) { - nvkm_wr32(device, 0x690008 + hdmi, avi_infoframe.header); - nvkm_wr32(device, 0x69000c + hdmi, avi_infoframe.subpack0_low); - nvkm_wr32(device, 0x690010 + hdmi, avi_infoframe.subpack0_high); - nvkm_wr32(device, 0x690014 + hdmi, avi_infoframe.subpack1_low); - nvkm_wr32(device, 0x690018 + hdmi, avi_infoframe.subpack1_high); - nvkm_mask(device, 0x690000 + hdmi, 0x00000001, 0x00000001); - } - - /* GENERIC(?) / Vendor InfoFrame? */ - nvkm_mask(device, 0x690100 + hdmi, 0x00010001, 0x00000000); - if (vendor_size) { - nvkm_wr32(device, 0x690108 + hdmi, vendor_infoframe.header); - nvkm_wr32(device, 0x69010c + hdmi, vendor_infoframe.subpack0_low); - nvkm_wr32(device, 0x690110 + hdmi, vendor_infoframe.subpack0_high); - /* Is there a second (or further?) set of subpack registers here? */ - nvkm_mask(device, 0x690100 + hdmi, 0x00000001, 0x00000001); - } - - /* ??? InfoFrame? */ nvkm_mask(device, 0x6900c0 + hdmi, 0x00000001, 0x00000000); nvkm_wr32(device, 0x6900cc + hdmi, 0x00000010); @@ -87,14 +103,19 @@ gk104_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe nvkm_mask(device, 0x616798 + hoff, 0x401f007f, ctrl); } +const struct nvkm_ior_func_hdmi +gk104_sor_hdmi = { + .ctrl = gk104_sor_hdmi_ctrl, + .infoframe_avi = gk104_sor_hdmi_infoframe_avi, + .infoframe_vsi = gk104_sor_hdmi_infoframe_vsi, +}; + static const struct nvkm_ior_func gk104_sor = { .state = gf119_sor_state, .power = nv50_sor_power, .clock = gf119_sor_clock, - .hdmi = { - .ctrl = gk104_sor_hdmi_ctrl, - }, + .hdmi = &gk104_sor_hdmi, .dp = &gf119_sor_dp, .hda = &gf119_sor_hda, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gm107.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gm107.c index 9e9ef49bd8ac..b4d8e868616f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gm107.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gm107.c @@ -70,9 +70,7 @@ gm107_sor = { .state = gf119_sor_state, .power = nv50_sor_power, .clock = gf119_sor_clock, - .hdmi = { - .ctrl = gk104_sor_hdmi_ctrl, - }, + .hdmi = &gk104_sor_hdmi, .dp = &gm107_sor_dp, .hda = &gf119_sor_hda, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gm200.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gm200.c index 4ecc8f98af6e..562ebae57d44 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gm200.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gm200.c @@ -79,6 +79,14 @@ gm200_sor_hdmi_scdc(struct nvkm_ior *ior, u8 scdc) ior->tmds.high_speed = !!(scdc & 0x2); } +const struct nvkm_ior_func_hdmi +gm200_sor_hdmi = { + .ctrl = gk104_sor_hdmi_ctrl, + .scdc = gm200_sor_hdmi_scdc, + .infoframe_avi = gk104_sor_hdmi_infoframe_avi, + .infoframe_vsi = gk104_sor_hdmi_infoframe_vsi, +}; + void gm200_sor_route_set(struct nvkm_outp *outp, struct nvkm_ior *ior) { @@ -131,10 +139,7 @@ gm200_sor = { .state = gf119_sor_state, .power = nv50_sor_power, .clock = gf119_sor_clock, - .hdmi = { - .ctrl = gk104_sor_hdmi_ctrl, - .scdc = gm200_sor_hdmi_scdc, - }, + .hdmi = &gm200_sor_hdmi, .dp = &gm200_sor_dp, .hda = &gf119_sor_hda, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gp100.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gp100.c index 7172a9dfd89b..7f1eb4332040 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gp100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gp100.c @@ -37,10 +37,7 @@ gp100_sor = { .state = gf119_sor_state, .power = nv50_sor_power, .clock = gf119_sor_clock, - .hdmi = { - .ctrl = gk104_sor_hdmi_ctrl, - .scdc = gm200_sor_hdmi_scdc, - }, + .hdmi = &gm200_sor_hdmi, .dp = &gm200_sor_dp, .hda = &gf119_sor_hda, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gt215.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gt215.c index 70c49e7af9cf..a2c7c6f83dcd 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gt215.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gt215.c @@ -92,9 +92,53 @@ gt215_sor_dp = { .watermark = g94_sor_dp_watermark, }; -void -gt215_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, - u8 rekey, u8 *avi, u8 avi_size, u8 *vendor, u8 vendor_size) +static void +gt215_sor_hdmi_infoframe_vsi(struct nvkm_ior *ior, int head, void *data, u32 size) +{ + struct nvkm_device *device = ior->disp->engine.subdev.device; + struct packed_hdmi_infoframe vsi; + const u32 soff = nv50_ior_base(ior); + + pack_hdmi_infoframe(&vsi, data, size); + + nvkm_mask(device, 0x61c53c + soff, 0x00010001, 0x00010000); + if (!size) + return; + + nvkm_wr32(device, 0x61c544 + soff, vsi.header); + nvkm_wr32(device, 0x61c548 + soff, vsi.subpack0_low); + nvkm_wr32(device, 0x61c54c + soff, vsi.subpack0_high); + /* Is there a second (or up to fourth?) set of subpack registers here? */ + /* nvkm_wr32(device, 0x61c550 + soff, vsi.subpack1_low); */ + /* nvkm_wr32(device, 0x61c554 + soff, vsi.subpack1_high); */ + + nvkm_mask(device, 0x61c53c + soff, 0x00010001, 0x00010001); +} + +static void +gt215_sor_hdmi_infoframe_avi(struct nvkm_ior *ior, int head, void *data, u32 size) +{ + struct nvkm_device *device = ior->disp->engine.subdev.device; + struct packed_hdmi_infoframe avi; + const u32 soff = nv50_ior_base(ior); + + pack_hdmi_infoframe(&avi, data, size); + + nvkm_mask(device, 0x61c520 + soff, 0x00000001, 0x00000000); + if (size) + return; + + nvkm_wr32(device, 0x61c528 + soff, avi.header); + nvkm_wr32(device, 0x61c52c + soff, avi.subpack0_low); + nvkm_wr32(device, 0x61c530 + soff, avi.subpack0_high); + nvkm_wr32(device, 0x61c534 + soff, avi.subpack1_low); + nvkm_wr32(device, 0x61c538 + soff, avi.subpack1_high); + + nvkm_mask(device, 0x61c520 + soff, 0x00000001, 0x00000001); +} + +static void +gt215_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, u8 rekey) { struct nvkm_device *device = ior->disp->engine.subdev.device; const u32 ctrl = 0x40000000 * enable | @@ -102,11 +146,6 @@ gt215_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe max_ac_packet << 16 | rekey; const u32 soff = nv50_ior_base(ior); - struct packed_hdmi_infoframe avi_infoframe; - struct packed_hdmi_infoframe vendor_infoframe; - - pack_hdmi_infoframe(&avi_infoframe, avi, avi_size); - pack_hdmi_infoframe(&vendor_infoframe, vendor, vendor_size); if (!(ctrl & 0x40000000)) { nvkm_mask(device, 0x61c5a4 + soff, 0x40000000, 0x00000000); @@ -116,17 +155,6 @@ gt215_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe return; } - /* AVI InfoFrame */ - nvkm_mask(device, 0x61c520 + soff, 0x00000001, 0x00000000); - if (avi_size) { - nvkm_wr32(device, 0x61c528 + soff, avi_infoframe.header); - nvkm_wr32(device, 0x61c52c + soff, avi_infoframe.subpack0_low); - nvkm_wr32(device, 0x61c530 + soff, avi_infoframe.subpack0_high); - nvkm_wr32(device, 0x61c534 + soff, avi_infoframe.subpack1_low); - nvkm_wr32(device, 0x61c538 + soff, avi_infoframe.subpack1_high); - nvkm_mask(device, 0x61c520 + soff, 0x00000001, 0x00000001); - } - /* Audio InfoFrame */ nvkm_mask(device, 0x61c500 + soff, 0x00000001, 0x00000000); nvkm_wr32(device, 0x61c508 + soff, 0x000a0184); @@ -134,18 +162,6 @@ gt215_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe nvkm_wr32(device, 0x61c510 + soff, 0x00000000); nvkm_mask(device, 0x61c500 + soff, 0x00000001, 0x00000001); - /* Vendor InfoFrame */ - nvkm_mask(device, 0x61c53c + soff, 0x00010001, 0x00010000); - if (vendor_size) { - nvkm_wr32(device, 0x61c544 + soff, vendor_infoframe.header); - nvkm_wr32(device, 0x61c548 + soff, vendor_infoframe.subpack0_low); - nvkm_wr32(device, 0x61c54c + soff, vendor_infoframe.subpack0_high); - /* Is there a second (or up to fourth?) set of subpack registers here? */ - /* nvkm_wr32(device, 0x61c550 + soff, vendor_infoframe.subpack1_low); */ - /* nvkm_wr32(device, 0x61c554 + soff, vendor_infoframe.subpack1_high); */ - nvkm_mask(device, 0x61c53c + soff, 0x00010001, 0x00010001); - } - nvkm_mask(device, 0x61c5d0 + soff, 0x00070001, 0x00010001); /* SPARE, HW_CTS */ nvkm_mask(device, 0x61c568 + soff, 0x00010101, 0x00000000); /* ACR_CTRL, ?? */ nvkm_mask(device, 0x61c578 + soff, 0x80000000, 0x80000000); /* ACR_0441_ENABLE */ @@ -159,14 +175,19 @@ gt215_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe nvkm_mask(device, 0x61c5a4 + soff, 0x5f1f007f, ctrl); } +const struct nvkm_ior_func_hdmi +gt215_sor_hdmi = { + .ctrl = gt215_sor_hdmi_ctrl, + .infoframe_avi = gt215_sor_hdmi_infoframe_avi, + .infoframe_vsi = gt215_sor_hdmi_infoframe_vsi, +}; + static const struct nvkm_ior_func gt215_sor = { .state = g94_sor_state, .power = nv50_sor_power, .clock = nv50_sor_clock, - .hdmi = { - .ctrl = gt215_sor_hdmi_ctrl, - }, + .hdmi = >215_sor_hdmi, .dp = >215_sor_dp, .hda = >215_sor_hda, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gv100.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gv100.c index 6b9d49270fa7..115d0997fd62 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/gv100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/gv100.c @@ -96,9 +96,54 @@ gv100_sor_dp = { .watermark = gv100_sor_dp_watermark, }; -void -gv100_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, - u8 rekey, u8 *avi, u8 avi_size, u8 *vendor, u8 vendor_size) +static void +gv100_sor_hdmi_infoframe_vsi(struct nvkm_ior *ior, int head, void *data, u32 size) +{ + struct nvkm_device *device = ior->disp->engine.subdev.device; + struct packed_hdmi_infoframe vsi; + const u32 hoff = head * 0x400; + + pack_hdmi_infoframe(&vsi, data, size); + + nvkm_mask(device, 0x6f0100 + hoff, 0x00010001, 0x00000000); + if (!size) + return; + + nvkm_wr32(device, 0x6f0108 + hoff, vsi.header); + nvkm_wr32(device, 0x6f010c + hoff, vsi.subpack0_low); + nvkm_wr32(device, 0x6f0110 + hoff, vsi.subpack0_high); + nvkm_wr32(device, 0x6f0114 + hoff, 0x00000000); + nvkm_wr32(device, 0x6f0118 + hoff, 0x00000000); + nvkm_wr32(device, 0x6f011c + hoff, 0x00000000); + nvkm_wr32(device, 0x6f0120 + hoff, 0x00000000); + nvkm_wr32(device, 0x6f0124 + hoff, 0x00000000); + nvkm_mask(device, 0x6f0100 + hoff, 0x00000001, 0x00000001); +} + +static void +gv100_sor_hdmi_infoframe_avi(struct nvkm_ior *ior, int head, void *data, u32 size) +{ + struct nvkm_device *device = ior->disp->engine.subdev.device; + struct packed_hdmi_infoframe avi; + const u32 hoff = head * 0x400; + + pack_hdmi_infoframe(&avi, data, size); + + nvkm_mask(device, 0x6f0000 + hoff, 0x00000001, 0x00000000); + if (!size) + return; + + nvkm_wr32(device, 0x6f0008 + hoff, avi.header); + nvkm_wr32(device, 0x6f000c + hoff, avi.subpack0_low); + nvkm_wr32(device, 0x6f0010 + hoff, avi.subpack0_high); + nvkm_wr32(device, 0x6f0014 + hoff, avi.subpack1_low); + nvkm_wr32(device, 0x6f0018 + hoff, avi.subpack1_high); + + nvkm_mask(device, 0x6f0000 + hoff, 0x00000001, 0x00000001); +} + +static void +gv100_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packet, u8 rekey) { struct nvkm_device *device = ior->disp->engine.subdev.device; const u32 ctrl = 0x40000000 * enable | @@ -106,11 +151,6 @@ gv100_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe rekey; const u32 hoff = head * 0x800; const u32 hdmi = head * 0x400; - struct packed_hdmi_infoframe avi_infoframe; - struct packed_hdmi_infoframe vendor_infoframe; - - pack_hdmi_infoframe(&avi_infoframe, avi, avi_size); - pack_hdmi_infoframe(&vendor_infoframe, vendor, vendor_size); if (!(ctrl & 0x40000000)) { nvkm_mask(device, 0x6165c0 + hoff, 0x40000000, 0x00000000); @@ -120,32 +160,6 @@ gv100_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe return; } - /* AVI InfoFrame (AVI). */ - nvkm_mask(device, 0x6f0000 + hdmi, 0x00000001, 0x00000000); - if (avi_size) { - nvkm_wr32(device, 0x6f0008 + hdmi, avi_infoframe.header); - nvkm_wr32(device, 0x6f000c + hdmi, avi_infoframe.subpack0_low); - nvkm_wr32(device, 0x6f0010 + hdmi, avi_infoframe.subpack0_high); - nvkm_wr32(device, 0x6f0014 + hdmi, avi_infoframe.subpack1_low); - nvkm_wr32(device, 0x6f0018 + hdmi, avi_infoframe.subpack1_high); - nvkm_mask(device, 0x6f0000 + hdmi, 0x00000001, 0x00000001); - } - - /* Vendor-specific InfoFrame (VSI). */ - nvkm_mask(device, 0x6f0100 + hdmi, 0x00010001, 0x00000000); - if (vendor_size) { - nvkm_wr32(device, 0x6f0108 + hdmi, vendor_infoframe.header); - nvkm_wr32(device, 0x6f010c + hdmi, vendor_infoframe.subpack0_low); - nvkm_wr32(device, 0x6f0110 + hdmi, vendor_infoframe.subpack0_high); - nvkm_wr32(device, 0x6f0114 + hdmi, 0x00000000); - nvkm_wr32(device, 0x6f0118 + hdmi, 0x00000000); - nvkm_wr32(device, 0x6f011c + hdmi, 0x00000000); - nvkm_wr32(device, 0x6f0120 + hdmi, 0x00000000); - nvkm_wr32(device, 0x6f0124 + hdmi, 0x00000000); - nvkm_mask(device, 0x6f0100 + hdmi, 0x00000001, 0x00000001); - } - - /* General Control (GCP). */ nvkm_mask(device, 0x6f00c0 + hdmi, 0x00000001, 0x00000000); nvkm_wr32(device, 0x6f00cc + hdmi, 0x00000010); @@ -158,6 +172,14 @@ gv100_sor_hdmi_ctrl(struct nvkm_ior *ior, int head, bool enable, u8 max_ac_packe nvkm_mask(device, 0x6165c0 + hoff, 0x401f007f, ctrl); } +const struct nvkm_ior_func_hdmi +gv100_sor_hdmi = { + .ctrl = gv100_sor_hdmi_ctrl, + .scdc = gm200_sor_hdmi_scdc, + .infoframe_avi = gv100_sor_hdmi_infoframe_avi, + .infoframe_vsi = gv100_sor_hdmi_infoframe_vsi, +}; + void gv100_sor_state(struct nvkm_ior *sor, struct nvkm_ior_state *state) { @@ -190,10 +212,7 @@ gv100_sor = { .state = gv100_sor_state, .power = nv50_sor_power, .clock = gf119_sor_clock, - .hdmi = { - .ctrl = gv100_sor_hdmi_ctrl, - .scdc = gm200_sor_hdmi_scdc, - }, + .hdmi = &gv100_sor_hdmi, .dp = &gv100_sor_dp, .hda = &gv100_sor_hda, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/head.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/head.c index 83152c26fe3e..7f5d13d13c94 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/head.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/head.c @@ -39,44 +39,6 @@ nvkm_head_find(struct nvkm_disp *disp, int id) return NULL; } -int -nvkm_head_mthd_scanoutpos(struct nvkm_object *object, - struct nvkm_head *head, void *data, u32 size) -{ - union { - struct nv04_disp_scanoutpos_v0 v0; - } *args = data; - int ret = -ENOSYS; - - nvif_ioctl(object, "head scanoutpos size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(object, "head scanoutpos vers %d\n", - args->v0.version); - - head->func->state(head, &head->arm); - args->v0.vtotal = head->arm.vtotal; - args->v0.vblanks = head->arm.vblanks; - args->v0.vblanke = head->arm.vblanke; - args->v0.htotal = head->arm.htotal; - args->v0.hblanks = head->arm.hblanks; - args->v0.hblanke = head->arm.hblanke; - - /* We don't support reading htotal/vtotal on pre-NV50 VGA, - * so we have to give up and trigger the timestamping - * fallback in the drm core. - */ - if (!args->v0.vtotal || !args->v0.htotal) - return -ENOTSUPP; - - args->v0.time[0] = ktime_to_ns(ktime_get()); - head->func->rgpos(head, &args->v0.hline, &args->v0.vline); - args->v0.time[1] = ktime_to_ns(ktime_get()); - } else - return ret; - - return 0; -} - void nvkm_head_del(struct nvkm_head **phead) { diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/head.h b/drivers/gpu/drm/nouveau/nvkm/engine/disp/head.h index 84a2989193cf..856252bf559a 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/head.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/head.h @@ -1,6 +1,7 @@ /* SPDX-License-Identifier: MIT */ #ifndef __NVKM_DISP_HEAD_H__ #define __NVKM_DISP_HEAD_H__ +#include <nvif/object.h> #include "priv.h" struct nvkm_head { @@ -26,12 +27,12 @@ struct nvkm_head { u8 depth; } or; } arm, asy; + + struct nvkm_object object; }; int nvkm_head_new_(const struct nvkm_head_func *, struct nvkm_disp *, int id); void nvkm_head_del(struct nvkm_head **); -int nvkm_head_mthd_scanoutpos(struct nvkm_object *, - struct nvkm_head *, void *, u32); struct nvkm_head *nvkm_head_find(struct nvkm_disp *, int id); struct nvkm_head_func { diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/ior.h b/drivers/gpu/drm/nouveau/nvkm/engine/disp/ior.h index 671c4674ffcc..da1b1a626ef2 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/ior.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/ior.h @@ -63,12 +63,12 @@ struct nvkm_ior_func { void (*war_2)(struct nvkm_ior *); void (*war_3)(struct nvkm_ior *); - struct { - void (*ctrl)(struct nvkm_ior *, int head, bool enable, - u8 max_ac_packet, u8 rekey, u8 *avi, u8 avi_size, - u8 *vendor, u8 vendor_size); + const struct nvkm_ior_func_hdmi { + void (*ctrl)(struct nvkm_ior *, int head, bool enable, u8 max_ac_packet, u8 rekey); void (*scdc)(struct nvkm_ior *, u8 scdc); - } hdmi; + void (*infoframe_avi)(struct nvkm_ior *, int head, void *data, u32 size); + void (*infoframe_vsi)(struct nvkm_ior *, int head, void *data, u32 size); + } *hdmi; const struct nvkm_ior_func_dp { u8 lanes[4]; @@ -124,9 +124,10 @@ void nv50_sor_power(struct nvkm_ior *, bool, bool, bool, bool, bool); void nv50_sor_clock(struct nvkm_ior *); int g84_sor_new(struct nvkm_disp *, int); -void g84_sor_hdmi_ctrl(struct nvkm_ior *, int, bool, u8, u8, u8 *, u8 , u8 *, u8); +extern const struct nvkm_ior_func_hdmi g84_sor_hdmi; int g94_sor_cnt(struct nvkm_disp *, unsigned long *); + void g94_sor_state(struct nvkm_ior *, struct nvkm_ior_state *); extern const struct nvkm_ior_func_dp g94_sor_dp; int g94_sor_dp_links(struct nvkm_ior *, struct nvkm_i2c_aux *); @@ -137,7 +138,7 @@ void g94_sor_dp_audio_sym(struct nvkm_ior *, int, u16, u32); void g94_sor_dp_activesym(struct nvkm_ior *, int, u8, u8, u8, u8); void g94_sor_dp_watermark(struct nvkm_ior *, int, u8); -void gt215_sor_hdmi_ctrl(struct nvkm_ior *, int, bool, u8, u8, u8 *, u8 , u8 *, u8); +extern const struct nvkm_ior_func_hdmi gt215_sor_hdmi; void gt215_sor_dp_audio(struct nvkm_ior *, int, bool); extern const struct nvkm_ior_func_hda gt215_sor_hda; @@ -156,12 +157,16 @@ void gf119_sor_hda_hpd(struct nvkm_ior *, int, bool); void gf119_sor_hda_eld(struct nvkm_ior *, int, u8 *, u8); int gk104_sor_new(struct nvkm_disp *, int); -void gk104_sor_hdmi_ctrl(struct nvkm_ior *, int, bool, u8, u8, u8 *, u8 , u8 *, u8); +extern const struct nvkm_ior_func_hdmi gk104_sor_hdmi; +void gk104_sor_hdmi_ctrl(struct nvkm_ior *, int, bool, u8, u8); +void gk104_sor_hdmi_infoframe_avi(struct nvkm_ior *, int, void *, u32); +void gk104_sor_hdmi_infoframe_vsi(struct nvkm_ior *, int, void *, u32); void gm107_sor_dp_pattern(struct nvkm_ior *, int); void gm200_sor_route_set(struct nvkm_outp *, struct nvkm_ior *); int gm200_sor_route_get(struct nvkm_outp *, int *); +extern const struct nvkm_ior_func_hdmi gm200_sor_hdmi; void gm200_sor_hdmi_scdc(struct nvkm_ior *, u8); extern const struct nvkm_ior_func_dp gm200_sor_dp; void gm200_sor_dp_drive(struct nvkm_ior *, int, int, int, int, int); @@ -170,7 +175,7 @@ int gp100_sor_new(struct nvkm_disp *, int); int gv100_sor_cnt(struct nvkm_disp *, unsigned long *); void gv100_sor_state(struct nvkm_ior *, struct nvkm_ior_state *); -void gv100_sor_hdmi_ctrl(struct nvkm_ior *, int, bool, u8, u8, u8 *, u8 , u8 *, u8); +extern const struct nvkm_ior_func_hdmi gv100_sor_hdmi; void gv100_sor_dp_audio(struct nvkm_ior *, int, bool); void gv100_sor_dp_audio_sym(struct nvkm_ior *, int, u16, u32); void gv100_sor_dp_watermark(struct nvkm_ior *, int, u8); diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/mcp77.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/mcp77.c index 916b1d477b0b..841e3b69fcaf 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/mcp77.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/mcp77.c @@ -31,9 +31,7 @@ mcp77_sor = { .state = g94_sor_state, .power = nv50_sor_power, .clock = nv50_sor_clock, - .hdmi = { - .ctrl = g84_sor_hdmi_ctrl, - }, + .hdmi = &g84_sor_hdmi, .dp = &g94_sor_dp, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/mcp89.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/mcp89.c index a5a0b9439374..f96ba4752655 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/mcp89.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/mcp89.c @@ -44,9 +44,7 @@ mcp89_sor = { .state = g94_sor_state, .power = nv50_sor_power, .clock = nv50_sor_clock, - .hdmi = { - .ctrl = gt215_sor_hdmi_ctrl, - }, + .hdmi = >215_sor_hdmi, .dp = &mcp89_sor_dp, .hda = >215_sor_hda, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/nv50.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/nv50.c index a46e13cc9ff1..be8116802960 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/nv50.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/nv50.c @@ -503,7 +503,7 @@ nv50_disp_chan_uevent_init(struct nvkm_event *event, int types, int index) void nv50_disp_chan_uevent_send(struct nvkm_disp *disp, int chid) { - nvkm_event_send(&disp->uevent, NVKM_DISP_EVENT_CHAN_AWAKEN, chid, NULL, 0); + nvkm_event_ntfy(&disp->uevent, chid, NVKM_DISP_EVENT_CHAN_AWAKEN); } const struct nvkm_event_func @@ -1238,6 +1238,8 @@ nv50_disp_super_2_2(struct nvkm_disp *disp, struct nvkm_head *head) if (!ior) return; + outp = ior->asy.outp; + /* For some reason, NVIDIA decided not to: * * A) Give dual-link LVDS a separate EVO protocol, like for TMDS. @@ -1247,13 +1249,13 @@ nv50_disp_super_2_2(struct nvkm_disp *disp, struct nvkm_head *head) * Override the values we usually read from HW with the same * data we pass though an ioctl instead. */ - if (ior->type == SOR && ior->asy.proto == LVDS) { - head->asy.or.depth = (disp->sor.lvdsconf & 0x0200) ? 24 : 18; - ior->asy.link = (disp->sor.lvdsconf & 0x0100) ? 3 : 1; + if (outp && ior->type == SOR && ior->asy.proto == LVDS) { + head->asy.or.depth = outp->lvds.bpc8 ? 24 : 18; + ior->asy.link = outp->lvds.dual ? 3 : 1; } /* Handle any link training, etc. */ - if ((outp = ior->asy.outp) && outp->func->acquire) + if (outp && outp->func->acquire) outp->func->acquire(outp); /* Execute OnInt2 IED script. */ diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h index 3f3924c41957..b7631c1ab242 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/outp.h @@ -2,7 +2,6 @@ #ifndef __NVKM_DISP_OUTP_H__ #define __NVKM_DISP_OUTP_H__ #include "priv.h" -#include <core/notify.h> #include <subdev/bios.h> #include <subdev/bios/dcb.h> @@ -28,13 +27,19 @@ struct nvkm_outp { union { struct { + bool dual; + bool bpc8; + } lvds; + + struct { struct nvbios_dpout info; u8 version; struct nvkm_i2c_aux *aux; - struct nvkm_notify hpd; - bool present; + bool enabled; + bool aux_pwr; + bool aux_pwr_pu; u8 lttpr[6]; u8 lttprs; u8 dpcd[16]; @@ -49,12 +54,17 @@ struct nvkm_outp { struct mutex mutex; struct { atomic_t done; + u8 nr; + u8 bw; bool mst; } lt; } dp; }; struct nvkm_object object; + struct { + struct nvkm_head *head; + } asy; }; int nvkm_outp_new_(const struct nvkm_outp_func *, struct nvkm_disp *, int index, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/priv.h b/drivers/gpu/drm/nouveau/nvkm/engine/disp/priv.h index cb25dfe849f0..ec5292a8f3c8 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/priv.h @@ -42,10 +42,6 @@ struct nvkm_disp_func { } user[]; }; -int nvkm_disp_ntfy(struct nvkm_object *, u32, struct nvkm_event **); -int nv04_disp_mthd(struct nvkm_object *, u32, void *, u32); -int nv50_disp_root_mthd_(struct nvkm_object *, u32, void *, u32); - int nv50_disp_oneinit(struct nvkm_disp *); int nv50_disp_init(struct nvkm_disp *); void nv50_disp_fini(struct nvkm_disp *); @@ -86,4 +82,5 @@ extern const struct nvkm_event_func gv100_disp_chan_uevent; int nvkm_udisp_new(const struct nvkm_oclass *, void *, u32, struct nvkm_object **); int nvkm_uconn_new(const struct nvkm_oclass *, void *, u32, struct nvkm_object **); int nvkm_uoutp_new(const struct nvkm_oclass *, void *, u32, struct nvkm_object **); +int nvkm_uhead_new(const struct nvkm_oclass *, void *, u32, struct nvkm_object **); #endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/rootnv50.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/rootnv50.c deleted file mode 100644 index 0af45ccd140c..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/rootnv50.c +++ /dev/null @@ -1,250 +0,0 @@ -/* - * Copyright 2012 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#include "chan.h" -#include "head.h" -#include "ior.h" -#include "outp.h" - -#include <core/client.h> - -#include <nvif/class.h> -#include <nvif/cl5070.h> -#include <nvif/unpack.h> - -int -nv50_disp_root_mthd_(struct nvkm_object *object, u32 mthd, void *data, u32 size) -{ - union { - struct nv50_disp_mthd_v0 v0; - struct nv50_disp_mthd_v1 v1; - } *args = data; - struct nvkm_disp *disp = nvkm_udisp(object); - struct nvkm_outp *temp, *outp = NULL; - struct nvkm_head *head; - u16 type, mask = 0; - int hidx, ret = -ENOSYS; - - if (mthd != NV50_DISP_MTHD) - return -EINVAL; - - nvif_ioctl(object, "disp mthd size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, true))) { - nvif_ioctl(object, "disp mthd vers %d mthd %02x head %d\n", - args->v0.version, args->v0.method, args->v0.head); - mthd = args->v0.method; - hidx = args->v0.head; - } else - if (!(ret = nvif_unpack(ret, &data, &size, args->v1, 1, 1, true))) { - nvif_ioctl(object, "disp mthd vers %d mthd %02x " - "type %04x mask %04x\n", - args->v1.version, args->v1.method, - args->v1.hasht, args->v1.hashm); - mthd = args->v1.method; - type = args->v1.hasht; - mask = args->v1.hashm; - hidx = ffs((mask >> 8) & 0x0f) - 1; - } else - return ret; - - if (!(head = nvkm_head_find(disp, hidx))) - return -ENXIO; - - if (mask) { - list_for_each_entry(temp, &disp->outps, head) { - if ((temp->info.hasht == type) && - (temp->info.hashm & mask) == mask) { - outp = temp; - break; - } - } - if (outp == NULL) - return -ENXIO; - } - - switch (mthd) { - case NV50_DISP_SCANOUTPOS: { - return nvkm_head_mthd_scanoutpos(object, head, data, size); - } - default: - break; - } - - switch (mthd * !!outp) { - case NV50_DISP_MTHD_V1_ACQUIRE: { - union { - struct nv50_disp_acquire_v0 v0; - } *args = data; - int ret = -ENOSYS; - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - ret = nvkm_outp_acquire(outp, NVKM_OUTP_USER, args->v0.hda); - if (ret == 0) { - args->v0.or = outp->ior->id; - args->v0.link = outp->ior->asy.link; - } - } - return ret; - } - break; - case NV50_DISP_MTHD_V1_RELEASE: - nvkm_outp_release(outp, NVKM_OUTP_USER); - return 0; - case NV50_DISP_MTHD_V1_SOR_HDA_ELD: { - union { - struct nv50_disp_sor_hda_eld_v0 v0; - } *args = data; - struct nvkm_ior *ior = outp->ior; - int ret = -ENOSYS; - - nvif_ioctl(object, "disp sor hda eld size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, true))) { - nvif_ioctl(object, "disp sor hda eld vers %d\n", - args->v0.version); - if (size > 0x60) - return -E2BIG; - } else - return ret; - - if (!ior->hda) - return -ENODEV; - - if (size && args->v0.data[0]) { - if (outp->info.type == DCB_OUTPUT_DP) - ior->func->dp->audio(ior, hidx, true); - ior->func->hda->hpd(ior, hidx, true); - ior->func->hda->eld(ior, hidx, data, size); - } else { - if (outp->info.type == DCB_OUTPUT_DP) - ior->func->dp->audio(ior, hidx, false); - ior->func->hda->hpd(ior, hidx, false); - } - - return 0; - } - break; - case NV50_DISP_MTHD_V1_SOR_HDMI_PWR: { - union { - struct nv50_disp_sor_hdmi_pwr_v0 v0; - } *args = data; - u8 *vendor, vendor_size; - u8 *avi, avi_size; - int ret = -ENOSYS; - - nvif_ioctl(object, "disp sor hdmi ctrl size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, true))) { - nvif_ioctl(object, "disp sor hdmi ctrl vers %d state %d " - "max_ac_packet %d rekey %d scdc %d\n", - args->v0.version, args->v0.state, - args->v0.max_ac_packet, args->v0.rekey, - args->v0.scdc); - if (args->v0.max_ac_packet > 0x1f || args->v0.rekey > 0x7f) - return -EINVAL; - if ((args->v0.avi_infoframe_length - + args->v0.vendor_infoframe_length) > size) - return -EINVAL; - else - if ((args->v0.avi_infoframe_length - + args->v0.vendor_infoframe_length) < size) - return -E2BIG; - avi = data; - avi_size = args->v0.avi_infoframe_length; - vendor = avi + avi_size; - vendor_size = args->v0.vendor_infoframe_length; - } else - return ret; - - if (!outp->ior->func->hdmi.ctrl) - return -ENODEV; - - outp->ior->func->hdmi.ctrl(outp->ior, hidx, args->v0.state, - args->v0.max_ac_packet, - args->v0.rekey, avi, avi_size, - vendor, vendor_size); - - if (outp->ior->func->hdmi.scdc) - outp->ior->func->hdmi.scdc(outp->ior, args->v0.scdc); - - return 0; - } - break; - case NV50_DISP_MTHD_V1_SOR_LVDS_SCRIPT: { - union { - struct nv50_disp_sor_lvds_script_v0 v0; - } *args = data; - int ret = -ENOSYS; - nvif_ioctl(object, "disp sor lvds script size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(object, "disp sor lvds script " - "vers %d name %04x\n", - args->v0.version, args->v0.script); - disp->sor.lvdsconf = args->v0.script; - return 0; - } else - return ret; - } - break; - case NV50_DISP_MTHD_V1_SOR_DP_MST_LINK: { - union { - struct nv50_disp_sor_dp_mst_link_v0 v0; - } *args = data; - int ret = -ENOSYS; - nvif_ioctl(object, "disp sor dp mst link size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(object, "disp sor dp mst link vers %d state %d\n", - args->v0.version, args->v0.state); - outp->dp.lt.mst = !!args->v0.state; - return 0; - } else - return ret; - } - break; - case NV50_DISP_MTHD_V1_SOR_DP_MST_VCPI: { - union { - struct nv50_disp_sor_dp_mst_vcpi_v0 v0; - } *args = data; - int ret = -ENOSYS; - nvif_ioctl(object, "disp sor dp mst vcpi size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(object, "disp sor dp mst vcpi vers %d " - "slot %02x/%02x pbn %04x/%04x\n", - args->v0.version, args->v0.start_slot, - args->v0.num_slots, args->v0.pbn, - args->v0.aligned_pbn); - if (!outp->ior->func->dp->vcpi) - return -ENODEV; - outp->ior->func->dp->vcpi(outp->ior, hidx, - args->v0.start_slot, - args->v0.num_slots, - args->v0.pbn, - args->v0.aligned_pbn); - return 0; - } else - return ret; - } - break; - default: - break; - } - - return -EINVAL; -} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/tu102.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/tu102.c index e4ad1a6f6c88..f5242a672279 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/tu102.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/tu102.c @@ -88,10 +88,7 @@ tu102_sor = { .state = gv100_sor_state, .power = nv50_sor_power, .clock = gf119_sor_clock, - .hdmi = { - .ctrl = gv100_sor_hdmi_ctrl, - .scdc = gm200_sor_hdmi_scdc, - }, + .hdmi = &gv100_sor_hdmi, .dp = &tu102_sor_dp, .hda = &gv100_sor_hda, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/uconn.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/uconn.c index fd9f18144c26..dad942be6679 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/uconn.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/uconn.c @@ -21,12 +21,86 @@ */ #define nvkm_uconn(p) container_of((p), struct nvkm_conn, object) #include "conn.h" +#include "outp.h" +#include <core/client.h> +#include <core/event.h> #include <subdev/gpio.h> +#include <subdev/i2c.h> #include <nvif/if0011.h> static int +nvkm_uconn_uevent_aux(struct nvkm_object *object, u64 token, u32 bits) +{ + union nvif_conn_event_args args; + + args.v0.version = 0; + args.v0.types = 0; + if (bits & NVKM_I2C_PLUG) + args.v0.types |= NVIF_CONN_EVENT_V0_PLUG; + if (bits & NVKM_I2C_UNPLUG) + args.v0.types |= NVIF_CONN_EVENT_V0_UNPLUG; + if (bits & NVKM_I2C_IRQ) + args.v0.types |= NVIF_CONN_EVENT_V0_IRQ; + + return object->client->event(token, &args, sizeof(args.v0)); +} + +static int +nvkm_uconn_uevent_gpio(struct nvkm_object *object, u64 token, u32 bits) +{ + union nvif_conn_event_args args; + + args.v0.version = 0; + args.v0.types = 0; + if (bits & NVKM_GPIO_HI) + args.v0.types |= NVIF_CONN_EVENT_V0_PLUG; + if (bits & NVKM_GPIO_LO) + args.v0.types |= NVIF_CONN_EVENT_V0_UNPLUG; + + return object->client->event(token, &args, sizeof(args.v0)); +} + +static int +nvkm_uconn_uevent(struct nvkm_object *object, void *argv, u32 argc, struct nvkm_uevent *uevent) +{ + struct nvkm_conn *conn = nvkm_uconn(object); + struct nvkm_device *device = conn->disp->engine.subdev.device; + struct nvkm_outp *outp; + union nvif_conn_event_args *args = argv; + u64 bits = 0; + + if (!uevent) { + if (conn->info.hpd == DCB_GPIO_UNUSED) + return -ENOSYS; + return 0; + } + + if (argc != sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + + list_for_each_entry(outp, &conn->disp->outps, head) { + if (outp->info.connector == conn->index && outp->dp.aux) { + if (args->v0.types & NVIF_CONN_EVENT_V0_PLUG ) bits |= NVKM_I2C_PLUG; + if (args->v0.types & NVIF_CONN_EVENT_V0_UNPLUG) bits |= NVKM_I2C_UNPLUG; + if (args->v0.types & NVIF_CONN_EVENT_V0_IRQ ) bits |= NVKM_I2C_IRQ; + + return nvkm_uevent_add(uevent, &device->i2c->event, outp->dp.aux->id, bits, + nvkm_uconn_uevent_aux); + } + } + + if (args->v0.types & NVIF_CONN_EVENT_V0_PLUG ) bits |= NVKM_GPIO_HI; + if (args->v0.types & NVIF_CONN_EVENT_V0_UNPLUG) bits |= NVKM_GPIO_LO; + if (args->v0.types & NVIF_CONN_EVENT_V0_IRQ) + return -EINVAL; + + return nvkm_uevent_add(uevent, &device->gpio->event, conn->info.hpd, bits, + nvkm_uconn_uevent_gpio); +} + +static int nvkm_uconn_mthd_hpd_status(struct nvkm_conn *conn, void *argv, u32 argc) { struct nvkm_gpio *gpio = conn->disp->engine.subdev.device->gpio; @@ -82,6 +156,7 @@ static const struct nvkm_object_func nvkm_uconn = { .dtor = nvkm_uconn_dtor, .mthd = nvkm_uconn_mthd, + .uevent = nvkm_uconn_uevent, }; int diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/udisp.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/udisp.c index 0841e7ce0343..0268d1d75805 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/udisp.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/udisp.c @@ -21,6 +21,7 @@ */ #include "priv.h" #include "conn.h" +#include "head.h" #include "outp.h" #include <nvif/class.h> @@ -43,6 +44,12 @@ nvkm_udisp_sclass(struct nvkm_object *object, int index, struct nvkm_oclass *scl return 0; } + if (index-- == 0) { + sclass->base = (struct nvkm_sclass) { 0, 0, NVIF_CLASS_HEAD }; + sclass->ctor = nvkm_uhead_new; + return 0; + } + if (disp->func->user[index].ctor) { sclass->base = disp->func->user[index].base; sclass->ctor = disp->func->user[index].ctor; @@ -52,17 +59,6 @@ nvkm_udisp_sclass(struct nvkm_object *object, int index, struct nvkm_oclass *scl return -EINVAL; } -static int -nvkm_udisp_mthd(struct nvkm_object *object, u32 mthd, void *argv, u32 argc) -{ - struct nvkm_disp *disp = nvkm_udisp(object); - - if (disp->engine.subdev.device->card_type >= NV_50) - return nv50_disp_root_mthd_(object, mthd, argv, argc); - - return nv04_disp_mthd(object, mthd, argv, argc); -} - static void * nvkm_udisp_dtor(struct nvkm_object *object) { @@ -78,8 +74,6 @@ nvkm_udisp_dtor(struct nvkm_object *object) static const struct nvkm_object_func nvkm_udisp = { .dtor = nvkm_udisp_dtor, - .mthd = nvkm_udisp_mthd, - .ntfy = nvkm_disp_ntfy, .sclass = nvkm_udisp_sclass, }; @@ -89,6 +83,7 @@ nvkm_udisp_new(const struct nvkm_oclass *oclass, void *argv, u32 argc, struct nv struct nvkm_disp *disp = nvkm_disp(oclass->engine); struct nvkm_conn *conn; struct nvkm_outp *outp; + struct nvkm_head *head; union nvif_disp_args *args = argv; if (argc != sizeof(args->v0) || args->v0.version != 0) @@ -111,5 +106,9 @@ nvkm_udisp_new(const struct nvkm_oclass *oclass, void *argv, u32 argc, struct nv list_for_each_entry(outp, &disp->outps, head) args->v0.outp_mask |= BIT(outp->index); + args->v0.head_mask = 0; + list_for_each_entry(head, &disp->heads, head) + args->v0.head_mask |= BIT(head->id); + return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/uhead.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/uhead.c new file mode 100644 index 000000000000..f072cec16040 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/uhead.c @@ -0,0 +1,127 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#define nvkm_uhead(p) container_of((p), struct nvkm_head, object) +#include "head.h" +#include <core/event.h> + +#include <nvif/if0013.h> + +#include <nvif/event.h> + +static int +nvkm_uhead_uevent(struct nvkm_object *object, void *argv, u32 argc, struct nvkm_uevent *uevent) +{ + struct nvkm_head *head = nvkm_uhead(object); + union nvif_head_event_args *args = argv; + + if (!uevent) + return 0; + if (argc != sizeof(args->vn)) + return -ENOSYS; + + return nvkm_uevent_add(uevent, &head->disp->vblank, head->id, + NVKM_DISP_HEAD_EVENT_VBLANK, NULL); +} + +static int +nvkm_uhead_mthd_scanoutpos(struct nvkm_head *head, void *argv, u32 argc) +{ + union nvif_head_scanoutpos_args *args = argv; + + if (argc != sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + + head->func->state(head, &head->arm); + args->v0.vtotal = head->arm.vtotal; + args->v0.vblanks = head->arm.vblanks; + args->v0.vblanke = head->arm.vblanke; + args->v0.htotal = head->arm.htotal; + args->v0.hblanks = head->arm.hblanks; + args->v0.hblanke = head->arm.hblanke; + + /* We don't support reading htotal/vtotal on pre-NV50 VGA, + * so we have to give up and trigger the timestamping + * fallback in the drm core. + */ + if (!args->v0.vtotal || !args->v0.htotal) + return -ENOTSUPP; + + args->v0.time[0] = ktime_to_ns(ktime_get()); + head->func->rgpos(head, &args->v0.hline, &args->v0.vline); + args->v0.time[1] = ktime_to_ns(ktime_get()); + return 0; +} + +static int +nvkm_uhead_mthd(struct nvkm_object *object, u32 mthd, void *argv, u32 argc) +{ + struct nvkm_head *head = nvkm_uhead(object); + + switch (mthd) { + case NVIF_HEAD_V0_SCANOUTPOS: return nvkm_uhead_mthd_scanoutpos(head, argv, argc); + default: + return -EINVAL; + } +} + +static void * +nvkm_uhead_dtor(struct nvkm_object *object) +{ + struct nvkm_head *head = nvkm_uhead(object); + struct nvkm_disp *disp = head->disp; + + spin_lock(&disp->client.lock); + head->object.func = NULL; + spin_unlock(&disp->client.lock); + return NULL; +} + +static const struct nvkm_object_func +nvkm_uhead = { + .dtor = nvkm_uhead_dtor, + .mthd = nvkm_uhead_mthd, + .uevent = nvkm_uhead_uevent, +}; + +int +nvkm_uhead_new(const struct nvkm_oclass *oclass, void *argv, u32 argc, struct nvkm_object **pobject) +{ + struct nvkm_disp *disp = nvkm_udisp(oclass->parent); + struct nvkm_head *head; + union nvif_head_args *args = argv; + int ret; + + if (argc != sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + if (!(head = nvkm_head_find(disp, args->v0.id))) + return -EINVAL; + + ret = -EBUSY; + spin_lock(&disp->client.lock); + if (!head->object.func) { + nvkm_object_ctor(&nvkm_uhead, oclass, &head->object); + *pobject = &head->object; + ret = 0; + } + spin_unlock(&disp->client.lock); + return ret; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/uoutp.c b/drivers/gpu/drm/nouveau/nvkm/engine/disp/uoutp.c index abedb3e86361..4f0ca709c85a 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/uoutp.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/disp/uoutp.c @@ -21,11 +21,238 @@ */ #define nvkm_uoutp(p) container_of((p), struct nvkm_outp, object) #include "outp.h" +#include "dp.h" +#include "head.h" #include "ior.h" #include <nvif/if0012.h> static int +nvkm_uoutp_mthd_dp_mst_vcpi(struct nvkm_outp *outp, void *argv, u32 argc) +{ + struct nvkm_ior *ior = outp->ior; + union nvif_outp_dp_mst_vcpi_args *args = argv; + + if (argc != sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + if (!ior->func->dp || !ior->func->dp->vcpi || !nvkm_head_find(outp->disp, args->v0.head)) + return -EINVAL; + + ior->func->dp->vcpi(ior, args->v0.head, args->v0.start_slot, args->v0.num_slots, + args->v0.pbn, args->v0.aligned_pbn); + return 0; +} + +static int +nvkm_uoutp_mthd_dp_retrain(struct nvkm_outp *outp, void *argv, u32 argc) +{ + union nvif_outp_dp_retrain_args *args = argv; + + if (argc != sizeof(args->vn)) + return -ENOSYS; + + if (!atomic_read(&outp->dp.lt.done)) + return 0; + + return outp->func->acquire(outp); +} + +static int +nvkm_uoutp_mthd_dp_aux_pwr(struct nvkm_outp *outp, void *argv, u32 argc) +{ + union nvif_outp_dp_aux_pwr_args *args = argv; + + if (argc != sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + + outp->dp.enabled = !!args->v0.state; + nvkm_dp_enable(outp, outp->dp.enabled); + return 0; +} + +static int +nvkm_uoutp_mthd_hda_eld(struct nvkm_outp *outp, void *argv, u32 argc) +{ + struct nvkm_ior *ior = outp->ior; + union nvif_outp_hda_eld_args *args = argv; + + if (argc < sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + argc -= sizeof(args->v0); + + if (!ior->hda || !nvkm_head_find(outp->disp, args->v0.head)) + return -EINVAL; + if (argc > 0x60) + return -E2BIG; + + if (argc && args->v0.data[0]) { + if (outp->info.type == DCB_OUTPUT_DP) + ior->func->dp->audio(ior, args->v0.head, true); + ior->func->hda->hpd(ior, args->v0.head, true); + ior->func->hda->eld(ior, args->v0.head, args->v0.data, argc); + } else { + if (outp->info.type == DCB_OUTPUT_DP) + ior->func->dp->audio(ior, args->v0.head, false); + ior->func->hda->hpd(ior, args->v0.head, false); + } + + return 0; +} + +static int +nvkm_uoutp_mthd_infoframe(struct nvkm_outp *outp, void *argv, u32 argc) +{ + struct nvkm_ior *ior = outp->ior; + union nvif_outp_infoframe_args *args = argv; + ssize_t size = argc - sizeof(*args); + + if (argc < sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + if (!nvkm_head_find(outp->disp, args->v0.head)) + return -EINVAL; + + switch (ior->func->hdmi ? args->v0.type : 0xff) { + case NVIF_OUTP_INFOFRAME_V0_AVI: + ior->func->hdmi->infoframe_avi(ior, args->v0.head, &args->v0.data, size); + return 0; + case NVIF_OUTP_INFOFRAME_V0_VSI: + ior->func->hdmi->infoframe_vsi(ior, args->v0.head, &args->v0.data, size); + return 0; + default: + break; + } + + return -EINVAL; +} + +static int +nvkm_uoutp_mthd_release(struct nvkm_outp *outp, void *argv, u32 argc) +{ + struct nvkm_head *head = outp->asy.head; + struct nvkm_ior *ior = outp->ior; + union nvif_outp_release_args *args = argv; + + if (argc != sizeof(args->vn)) + return -ENOSYS; + + if (ior->func->hdmi && head) { + ior->func->hdmi->infoframe_avi(ior, head->id, NULL, 0); + ior->func->hdmi->infoframe_vsi(ior, head->id, NULL, 0); + ior->func->hdmi->ctrl(ior, head->id, false, 0, 0); + } + + nvkm_outp_release(outp, NVKM_OUTP_USER); + return 0; +} + +static int +nvkm_uoutp_mthd_acquire_dp(struct nvkm_outp *outp, u8 dpcd[16], + u8 link_nr, u8 link_bw, bool hda, bool mst) +{ + int ret; + + ret = nvkm_outp_acquire(outp, NVKM_OUTP_USER, hda); + if (ret) + return ret; + + memcpy(outp->dp.dpcd, dpcd, sizeof(outp->dp.dpcd)); + outp->dp.lt.nr = link_nr; + outp->dp.lt.bw = link_bw; + outp->dp.lt.mst = mst; + return 0; +} + +static int +nvkm_uoutp_mthd_acquire_tmds(struct nvkm_outp *outp, u8 head, u8 hdmi, u8 hdmi_max_ac_packet, + u8 hdmi_rekey, u8 hdmi_scdc, u8 hdmi_hda) +{ + struct nvkm_ior *ior; + int ret; + + if (!(outp->asy.head = nvkm_head_find(outp->disp, head))) + return -EINVAL; + + ret = nvkm_outp_acquire(outp, NVKM_OUTP_USER, hdmi && hdmi_hda); + if (ret) + return ret; + + ior = outp->ior; + + if (hdmi) { + if (!ior->func->hdmi || + hdmi_max_ac_packet > 0x1f || hdmi_rekey > 0x7f || + (hdmi_scdc && !ior->func->hdmi->scdc)) { + nvkm_outp_release(outp, NVKM_OUTP_USER); + return -EINVAL; + } + + ior->func->hdmi->ctrl(ior, head, hdmi, hdmi_max_ac_packet, hdmi_rekey); + if (ior->func->hdmi->scdc) + ior->func->hdmi->scdc(ior, hdmi_scdc); + } + + return 0; +} + +static int +nvkm_uoutp_mthd_acquire_lvds(struct nvkm_outp *outp, bool dual, bool bpc8) +{ + if (outp->info.type != DCB_OUTPUT_LVDS) + return -EINVAL; + + outp->lvds.dual = dual; + outp->lvds.bpc8 = bpc8; + + return nvkm_outp_acquire(outp, NVKM_OUTP_USER, false); +} + +static int +nvkm_uoutp_mthd_acquire(struct nvkm_outp *outp, void *argv, u32 argc) +{ + union nvif_outp_acquire_args *args = argv; + int ret; + + if (argc != sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + if (outp->ior) + return -EBUSY; + + switch (args->v0.proto) { + case NVIF_OUTP_ACQUIRE_V0_RGB_CRT: + ret = nvkm_outp_acquire(outp, NVKM_OUTP_USER, false); + break; + case NVIF_OUTP_ACQUIRE_V0_TMDS: + ret = nvkm_uoutp_mthd_acquire_tmds(outp, args->v0.tmds.head, + args->v0.tmds.hdmi, + args->v0.tmds.hdmi_max_ac_packet, + args->v0.tmds.hdmi_rekey, + args->v0.tmds.hdmi_scdc, + args->v0.tmds.hdmi_hda); + break; + case NVIF_OUTP_ACQUIRE_V0_LVDS: + ret = nvkm_uoutp_mthd_acquire_lvds(outp, args->v0.lvds.dual, args->v0.lvds.bpc8); + break; + case NVIF_OUTP_ACQUIRE_V0_DP: + ret = nvkm_uoutp_mthd_acquire_dp(outp, args->v0.dp.dpcd, + args->v0.dp.link_nr, + args->v0.dp.link_bw, + args->v0.dp.hda != 0, + args->v0.dp.mst != 0); + break; + default: + ret = -EINVAL; + break; + } + + if (ret) + return ret; + + args->v0.or = outp->ior->id; + args->v0.link = outp->ior->asy.link; + return 0; +} + +static int nvkm_uoutp_mthd_load_detect(struct nvkm_outp *outp, void *argv, u32 argc) { union nvif_outp_load_detect_args *args = argv; @@ -49,10 +276,28 @@ nvkm_uoutp_mthd_load_detect(struct nvkm_outp *outp, void *argv, u32 argc) } static int +nvkm_uoutp_mthd_acquired(struct nvkm_outp *outp, u32 mthd, void *argv, u32 argc) +{ + switch (mthd) { + case NVIF_OUTP_V0_RELEASE : return nvkm_uoutp_mthd_release (outp, argv, argc); + case NVIF_OUTP_V0_INFOFRAME : return nvkm_uoutp_mthd_infoframe (outp, argv, argc); + case NVIF_OUTP_V0_HDA_ELD : return nvkm_uoutp_mthd_hda_eld (outp, argv, argc); + case NVIF_OUTP_V0_DP_RETRAIN : return nvkm_uoutp_mthd_dp_retrain (outp, argv, argc); + case NVIF_OUTP_V0_DP_MST_VCPI: return nvkm_uoutp_mthd_dp_mst_vcpi(outp, argv, argc); + default: + break; + } + + return -EINVAL; +} + +static int nvkm_uoutp_mthd_noacquire(struct nvkm_outp *outp, u32 mthd, void *argv, u32 argc) { switch (mthd) { case NVIF_OUTP_V0_LOAD_DETECT: return nvkm_uoutp_mthd_load_detect(outp, argv, argc); + case NVIF_OUTP_V0_ACQUIRE : return nvkm_uoutp_mthd_acquire (outp, argv, argc); + case NVIF_OUTP_V0_DP_AUX_PWR : return nvkm_uoutp_mthd_dp_aux_pwr (outp, argv, argc); default: break; } @@ -73,6 +318,11 @@ nvkm_uoutp_mthd(struct nvkm_object *object, u32 mthd, void *argv, u32 argc) if (ret <= 0) goto done; + if (outp->ior) + ret = nvkm_uoutp_mthd_acquired(outp, mthd, argv, argc); + else + ret = -EIO; + done: mutex_unlock(&disp->super.mutex); return ret; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/falcon.c b/drivers/gpu/drm/nouveau/nvkm/engine/falcon.c index 43b7dec45179..d619b40a42c3 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/falcon.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/falcon.c @@ -65,10 +65,10 @@ nvkm_falcon_intr(struct nvkm_engine *engine) u32 dest = nvkm_rd32(device, base + 0x01c); u32 intr = nvkm_rd32(device, base + 0x008) & dest & ~(dest >> 16); u32 inst = nvkm_rd32(device, base + 0x050) & 0x3fffffff; - struct nvkm_fifo_chan *chan; + struct nvkm_chan *chan; unsigned long flags; - chan = nvkm_fifo_chan_inst(device->fifo, (u64)inst << 12, &flags); + chan = nvkm_chan_get_inst(engine, (u64)inst << 12, &flags); if (intr & 0x00000040) { if (falcon->func->intr) { @@ -89,7 +89,7 @@ nvkm_falcon_intr(struct nvkm_engine *engine) nvkm_wr32(device, base + 0x004, intr); } - nvkm_fifo_chan_put(device->fifo, flags, &chan); + nvkm_chan_put(&chan, flags); } static int diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/Kbuild b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/Kbuild index 5e831d347a95..5a074b9970ab 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/Kbuild @@ -1,11 +1,18 @@ # SPDX-License-Identifier: MIT nvkm-y += nvkm/engine/fifo/base.o +nvkm-y += nvkm/engine/fifo/cgrp.o +nvkm-y += nvkm/engine/fifo/chan.o +nvkm-y += nvkm/engine/fifo/chid.o +nvkm-y += nvkm/engine/fifo/runl.o +nvkm-y += nvkm/engine/fifo/runq.o + nvkm-y += nvkm/engine/fifo/nv04.o nvkm-y += nvkm/engine/fifo/nv10.o nvkm-y += nvkm/engine/fifo/nv17.o nvkm-y += nvkm/engine/fifo/nv40.o nvkm-y += nvkm/engine/fifo/nv50.o nvkm-y += nvkm/engine/fifo/g84.o +nvkm-y += nvkm/engine/fifo/g98.o nvkm-y += nvkm/engine/fifo/gf100.o nvkm-y += nvkm/engine/fifo/gk104.o nvkm-y += nvkm/engine/fifo/gk110.o @@ -13,28 +20,11 @@ nvkm-y += nvkm/engine/fifo/gk208.o nvkm-y += nvkm/engine/fifo/gk20a.o nvkm-y += nvkm/engine/fifo/gm107.o nvkm-y += nvkm/engine/fifo/gm200.o -nvkm-y += nvkm/engine/fifo/gm20b.o nvkm-y += nvkm/engine/fifo/gp100.o -nvkm-y += nvkm/engine/fifo/gp10b.o nvkm-y += nvkm/engine/fifo/gv100.o nvkm-y += nvkm/engine/fifo/tu102.o +nvkm-y += nvkm/engine/fifo/ga100.o nvkm-y += nvkm/engine/fifo/ga102.o -nvkm-y += nvkm/engine/fifo/chan.o -nvkm-y += nvkm/engine/fifo/channv50.o -nvkm-y += nvkm/engine/fifo/chang84.o - -nvkm-y += nvkm/engine/fifo/dmanv04.o -nvkm-y += nvkm/engine/fifo/dmanv10.o -nvkm-y += nvkm/engine/fifo/dmanv17.o -nvkm-y += nvkm/engine/fifo/dmanv40.o - -nvkm-y += nvkm/engine/fifo/gpfifonv50.o -nvkm-y += nvkm/engine/fifo/gpfifog84.o -nvkm-y += nvkm/engine/fifo/gpfifogf100.o -nvkm-y += nvkm/engine/fifo/gpfifogk104.o -nvkm-y += nvkm/engine/fifo/gpfifogv100.o -nvkm-y += nvkm/engine/fifo/gpfifotu102.o - -nvkm-y += nvkm/engine/fifo/usergv100.o -nvkm-y += nvkm/engine/fifo/usertu102.o +nvkm-y += nvkm/engine/fifo/ucgrp.o +nvkm-y += nvkm/engine/fifo/uchan.o diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/base.c index 58b8df75fc40..5ea9a2ff0663 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/base.c @@ -23,25 +23,32 @@ */ #include "priv.h" #include "chan.h" +#include "chid.h" +#include "runl.h" +#include "runq.h" -#include <core/client.h> #include <core/gpuobj.h> -#include <core/notify.h> +#include <subdev/bar.h> #include <subdev/mc.h> +#include <subdev/mmu.h> -#include <nvif/event.h> #include <nvif/cl0080.h> #include <nvif/unpack.h> -void -nvkm_fifo_recover_chan(struct nvkm_fifo *fifo, int chid) +bool +nvkm_fifo_ctxsw_in_progress(struct nvkm_engine *engine) { - unsigned long flags; - if (WARN_ON(!fifo->func->recover_chan)) - return; - spin_lock_irqsave(&fifo->lock, flags); - fifo->func->recover_chan(fifo, chid); - spin_unlock_irqrestore(&fifo->lock, flags); + struct nvkm_runl *runl; + struct nvkm_engn *engn; + + nvkm_runl_foreach(runl, engine->subdev.device->fifo) { + nvkm_runl_foreach_engn(engn, runl) { + if (engn->engine == engine) + return engn->func->chsw ? engn->func->chsw(engn) : false; + } + } + + return false; } void @@ -59,160 +66,23 @@ nvkm_fifo_start(struct nvkm_fifo *fifo, unsigned long *flags) void nvkm_fifo_fault(struct nvkm_fifo *fifo, struct nvkm_fault_data *info) { - return fifo->func->fault(fifo, info); -} - -void -nvkm_fifo_chan_put(struct nvkm_fifo *fifo, unsigned long flags, - struct nvkm_fifo_chan **pchan) -{ - struct nvkm_fifo_chan *chan = *pchan; - if (likely(chan)) { - *pchan = NULL; - spin_unlock_irqrestore(&fifo->lock, flags); - } -} - -struct nvkm_fifo_chan * -nvkm_fifo_chan_inst_locked(struct nvkm_fifo *fifo, u64 inst) -{ - struct nvkm_fifo_chan *chan; - list_for_each_entry(chan, &fifo->chan, head) { - if (chan->inst->addr == inst) { - list_del(&chan->head); - list_add(&chan->head, &fifo->chan); - return chan; - } - } - return NULL; -} - -struct nvkm_fifo_chan * -nvkm_fifo_chan_inst(struct nvkm_fifo *fifo, u64 inst, unsigned long *rflags) -{ - struct nvkm_fifo_chan *chan; - unsigned long flags; - spin_lock_irqsave(&fifo->lock, flags); - if ((chan = nvkm_fifo_chan_inst_locked(fifo, inst))) { - *rflags = flags; - return chan; - } - spin_unlock_irqrestore(&fifo->lock, flags); - return NULL; -} - -struct nvkm_fifo_chan * -nvkm_fifo_chan_chid(struct nvkm_fifo *fifo, int chid, unsigned long *rflags) -{ - struct nvkm_fifo_chan *chan; - unsigned long flags; - spin_lock_irqsave(&fifo->lock, flags); - list_for_each_entry(chan, &fifo->chan, head) { - if (chan->chid == chid) { - list_del(&chan->head); - list_add(&chan->head, &fifo->chan); - *rflags = flags; - return chan; - } - } - spin_unlock_irqrestore(&fifo->lock, flags); - return NULL; -} - -void -nvkm_fifo_kevent(struct nvkm_fifo *fifo, int chid) -{ - nvkm_event_send(&fifo->kevent, 1, chid, NULL, 0); -} - -static int -nvkm_fifo_kevent_ctor(struct nvkm_object *object, void *data, u32 size, - struct nvkm_notify *notify) -{ - struct nvkm_fifo_chan *chan = nvkm_fifo_chan(object); - if (size == 0) { - notify->size = 0; - notify->types = 1; - notify->index = chan->chid; - return 0; - } - return -ENOSYS; -} - -static const struct nvkm_event_func -nvkm_fifo_kevent_func = { - .ctor = nvkm_fifo_kevent_ctor, -}; - -static void -nvkm_fifo_uevent_fini(struct nvkm_event *event, int type, int index) -{ - struct nvkm_fifo *fifo = container_of(event, typeof(*fifo), uevent); - fifo->func->uevent_fini(fifo); -} - -static void -nvkm_fifo_uevent_init(struct nvkm_event *event, int type, int index) -{ - struct nvkm_fifo *fifo = container_of(event, typeof(*fifo), uevent); - fifo->func->uevent_init(fifo); -} - -static int -nvkm_fifo_uevent_ctor(struct nvkm_object *object, void *data, u32 size, - struct nvkm_notify *notify) -{ - union { - struct nvif_notify_uevent_req none; - } *req = data; - int ret = -ENOSYS; - - if (!(ret = nvif_unvers(ret, &data, &size, req->none))) { - notify->size = sizeof(struct nvif_notify_uevent_rep); - notify->types = 1; - notify->index = 0; - } - - return ret; -} - -static const struct nvkm_event_func -nvkm_fifo_uevent_func = { - .ctor = nvkm_fifo_uevent_ctor, - .init = nvkm_fifo_uevent_init, - .fini = nvkm_fifo_uevent_fini, -}; - -void -nvkm_fifo_uevent(struct nvkm_fifo *fifo) -{ - struct nvif_notify_uevent_rep rep = { - }; - nvkm_event_send(&fifo->uevent, 1, 0, &rep, sizeof(rep)); + return fifo->func->mmu_fault->recover(fifo, info); } static int -nvkm_fifo_class_new_(struct nvkm_device *device, - const struct nvkm_oclass *oclass, void *data, u32 size, - struct nvkm_object **pobject) +nvkm_fifo_class_new(struct nvkm_device *device, const struct nvkm_oclass *oclass, + void *argv, u32 argc, struct nvkm_object **pobject) { struct nvkm_fifo *fifo = nvkm_fifo(oclass->engine); - return fifo->func->class_new(fifo, oclass, data, size, pobject); -} -static const struct nvkm_device_oclass -nvkm_fifo_class_ = { - .ctor = nvkm_fifo_class_new_, -}; + if (oclass->engn == &fifo->func->cgrp.user) + return nvkm_ucgrp_new(fifo, oclass, argv, argc, pobject); -static int -nvkm_fifo_class_new(struct nvkm_device *device, - const struct nvkm_oclass *oclass, void *data, u32 size, - struct nvkm_object **pobject) -{ - const struct nvkm_fifo_chan_oclass *sclass = oclass->engn; - struct nvkm_fifo *fifo = nvkm_fifo(oclass->engine); - return sclass->ctor(fifo, oclass, data, size, pobject); + if (oclass->engn == &fifo->func->chan.user) + return nvkm_uchan_new(fifo, NULL, oclass, argv, argc, pobject); + + WARN_ON(1); + return -ENOSYS; } static const struct nvkm_device_oclass @@ -221,24 +91,28 @@ nvkm_fifo_class = { }; static int -nvkm_fifo_class_get(struct nvkm_oclass *oclass, int index, - const struct nvkm_device_oclass **class) +nvkm_fifo_class_get(struct nvkm_oclass *oclass, int index, const struct nvkm_device_oclass **class) { struct nvkm_fifo *fifo = nvkm_fifo(oclass->engine); - const struct nvkm_fifo_chan_oclass *sclass; + const struct nvkm_fifo_func_cgrp *cgrp = &fifo->func->cgrp; + const struct nvkm_fifo_func_chan *chan = &fifo->func->chan; int c = 0; - if (fifo->func->class_get) { - int ret = fifo->func->class_get(fifo, index, oclass); - if (ret == 0) - *class = &nvkm_fifo_class_; - return ret; + /* *_CHANNEL_GROUP_* */ + if (cgrp->user.oclass) { + if (c++ == index) { + oclass->base = cgrp->user; + oclass->engn = &fifo->func->cgrp.user; + *class = &nvkm_fifo_class; + return 0; + } } - while ((sclass = fifo->func->chan[c])) { + /* *_CHANNEL_DMA, *_CHANNEL_GPFIFO_* */ + if (chan->user.oclass) { if (c++ == index) { - oclass->base = sclass->base; - oclass->engn = sclass; + oclass->base = chan->user; + oclass->engn = &fifo->func->chan.user; *class = &nvkm_fifo_class; return 0; } @@ -247,19 +121,47 @@ nvkm_fifo_class_get(struct nvkm_oclass *oclass, int index, return c; } -static void -nvkm_fifo_intr(struct nvkm_engine *engine) +static int +nvkm_fifo_fini(struct nvkm_engine *engine, bool suspend) { struct nvkm_fifo *fifo = nvkm_fifo(engine); - fifo->func->intr(fifo); + struct nvkm_runl *runl; + + nvkm_inth_block(&fifo->engine.subdev.inth); + + nvkm_runl_foreach(runl, fifo) + nvkm_runl_fini(runl); + + return 0; } static int -nvkm_fifo_fini(struct nvkm_engine *engine, bool suspend) +nvkm_fifo_init(struct nvkm_engine *engine) { struct nvkm_fifo *fifo = nvkm_fifo(engine); - if (fifo->func->fini) - fifo->func->fini(fifo); + struct nvkm_runq *runq; + struct nvkm_runl *runl; + u32 mask = 0; + + if (fifo->func->init_pbdmas) { + nvkm_runq_foreach(runq, fifo) + mask |= BIT(runq->id); + + fifo->func->init_pbdmas(fifo, mask); + + nvkm_runq_foreach(runq, fifo) + runq->func->init(runq); + } + + nvkm_runl_foreach(runl, fifo) { + if (runl->func->init) + runl->func->init(runl); + } + + if (fifo->func->init) + fifo->func->init(fifo); + + nvkm_inth_allow(&fifo->engine.subdev.inth); return 0; } @@ -267,22 +169,146 @@ static int nvkm_fifo_info(struct nvkm_engine *engine, u64 mthd, u64 *data) { struct nvkm_fifo *fifo = nvkm_fifo(engine); + struct nvkm_runl *runl; + struct nvkm_engn *engn; + int ret; + + ret = nvkm_subdev_oneinit(&fifo->engine.subdev); + if (ret) + return ret; + switch (mthd) { - case NV_DEVICE_HOST_CHANNELS: *data = fifo->nr; return 0; + case NV_DEVICE_HOST_CHANNELS: *data = fifo->chid ? fifo->chid->nr : 0; return 0; + case NV_DEVICE_HOST_RUNLISTS: + *data = 0; + nvkm_runl_foreach(runl, fifo) + *data |= BIT(runl->id); + return 0; + case NV_DEVICE_HOST_RUNLIST_ENGINES: + runl = nvkm_runl_get(fifo, *data, 0); + if (runl) { + *data = 0; + nvkm_runl_foreach_engn(engn, runl) { +#define CASE(n) case NVKM_ENGINE_##n: *data |= NV_DEVICE_HOST_RUNLIST_ENGINES_##n; break + switch (engn->engine->subdev.type) { + case NVKM_ENGINE_DMAOBJ: + break; + CASE(SW ); + CASE(GR ); + CASE(MPEG ); + CASE(ME ); + CASE(CIPHER); + CASE(BSP ); + CASE(VP ); + CASE(CE ); + CASE(SEC ); + CASE(MSVLD ); + CASE(MSPDEC); + CASE(MSPPP ); + CASE(MSENC ); + CASE(VIC ); + CASE(SEC2 ); + CASE(NVDEC ); + CASE(NVENC ); + default: + WARN_ON(1); + break; + } +#undef CASE + } + return 0; + } + return -EINVAL; + case NV_DEVICE_HOST_RUNLIST_CHANNELS: + if (!fifo->chid) { + runl = nvkm_runl_get(fifo, *data, 0); + if (runl) { + *data = runl->chid->nr; + return 0; + } + } + return -EINVAL; default: - if (fifo->func->info) - return fifo->func->info(fifo, mthd, data); break; } + return -ENOSYS; } static int nvkm_fifo_oneinit(struct nvkm_engine *engine) { + struct nvkm_subdev *subdev = &engine->subdev; + struct nvkm_device *device = subdev->device; struct nvkm_fifo *fifo = nvkm_fifo(engine); - if (fifo->func->oneinit) - return fifo->func->oneinit(fifo); + struct nvkm_runl *runl; + struct nvkm_engn *engn; + int ret, nr, i; + + /* Initialise CHID/CGID allocator(s) on GPUs where they aren't per-runlist. */ + if (fifo->func->chid_nr) { + ret = fifo->func->chid_ctor(fifo, fifo->func->chid_nr(fifo)); + if (ret) + return ret; + } + + /* Create runqueues for each PBDMA. */ + if (fifo->func->runq_nr) { + for (nr = fifo->func->runq_nr(fifo), i = 0; i < nr; i++) { + if (!nvkm_runq_new(fifo, i)) + return -ENOMEM; + } + } + + /* Create runlists. */ + ret = fifo->func->runl_ctor(fifo); + if (ret) + return ret; + + nvkm_runl_foreach(runl, fifo) { + RUNL_DEBUG(runl, "chan:%06x", runl->chan); + nvkm_runl_foreach_engn(engn, runl) { + ENGN_DEBUG(engn, ""); + } + } + + /* Register interrupt handler. */ + if (fifo->func->intr) { + ret = nvkm_inth_add(&device->mc->intr, NVKM_INTR_SUBDEV, NVKM_INTR_PRIO_NORMAL, + subdev, fifo->func->intr, &subdev->inth); + if (ret) { + nvkm_error(subdev, "intr %d\n", ret); + return ret; + } + } + + /* Initialise non-stall intr handling. */ + if (fifo->func->nonstall_ctor) { + ret = fifo->func->nonstall_ctor(fifo); + if (ret) { + nvkm_error(subdev, "nonstall %d\n", ret); + } + } + + /* Allocate USERD + BAR1 polling area. */ + if (fifo->func->chan.func->userd->bar == 1) { + struct nvkm_vmm *bar1 = nvkm_bar_bar1_vmm(device); + + ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, fifo->chid->nr * + fifo->func->chan.func->userd->size, 0, true, + &fifo->userd.mem); + if (ret) + return ret; + + ret = nvkm_vmm_get(bar1, 12, nvkm_memory_size(fifo->userd.mem), &fifo->userd.bar1); + if (ret) + return ret; + + ret = nvkm_memory_map(fifo->userd.mem, 0, bar1, fifo->userd.bar1, NULL, 0); + if (ret) + return ret; + } + return 0; } @@ -292,25 +318,28 @@ nvkm_fifo_preinit(struct nvkm_engine *engine) nvkm_mc_reset(engine->subdev.device, NVKM_ENGINE_FIFO, 0); } -static int -nvkm_fifo_init(struct nvkm_engine *engine) -{ - struct nvkm_fifo *fifo = nvkm_fifo(engine); - fifo->func->init(fifo); - return 0; -} - static void * nvkm_fifo_dtor(struct nvkm_engine *engine) { struct nvkm_fifo *fifo = nvkm_fifo(engine); - void *data = fifo; - if (fifo->func->dtor) - data = fifo->func->dtor(fifo); - nvkm_event_fini(&fifo->kevent); - nvkm_event_fini(&fifo->uevent); + struct nvkm_runl *runl, *runt; + struct nvkm_runq *runq, *rtmp; + + if (fifo->userd.bar1) + nvkm_vmm_put(nvkm_bar_bar1_vmm(engine->subdev.device), &fifo->userd.bar1); + nvkm_memory_unref(&fifo->userd.mem); + + list_for_each_entry_safe(runl, runt, &fifo->runls, head) + nvkm_runl_del(runl); + list_for_each_entry_safe(runq, rtmp, &fifo->runqs, head) + nvkm_runq_del(runq); + + nvkm_chid_unref(&fifo->cgid); + nvkm_chid_unref(&fifo->chid); + + nvkm_event_fini(&fifo->nonstall.event); mutex_destroy(&fifo->mutex); - return data; + return fifo; } static const struct nvkm_engine_func @@ -321,37 +350,40 @@ nvkm_fifo = { .info = nvkm_fifo_info, .init = nvkm_fifo_init, .fini = nvkm_fifo_fini, - .intr = nvkm_fifo_intr, .base.sclass = nvkm_fifo_class_get, }; int -nvkm_fifo_ctor(const struct nvkm_fifo_func *func, struct nvkm_device *device, - enum nvkm_subdev_type type, int inst, int nr, struct nvkm_fifo *fifo) +nvkm_fifo_new_(const struct nvkm_fifo_func *func, struct nvkm_device *device, + enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { + struct nvkm_fifo *fifo; int ret; + if (!(fifo = *pfifo = kzalloc(sizeof(*fifo), GFP_KERNEL))) + return -ENOMEM; + fifo->func = func; - INIT_LIST_HEAD(&fifo->chan); + INIT_LIST_HEAD(&fifo->runqs); + INIT_LIST_HEAD(&fifo->runls); + /*TODO: Needs to be >CTXSW_TIMEOUT, so RC can recover before this is hit. + * CTXSW_TIMEOUT HW default seems to differ between GPUs, so just a + * large number for now until we support changing it. + */ + fifo->timeout.chan_msec = 10000; spin_lock_init(&fifo->lock); mutex_init(&fifo->mutex); - if (WARN_ON(fifo->nr > NVKM_FIFO_CHID_NR)) - fifo->nr = NVKM_FIFO_CHID_NR; - else - fifo->nr = nr; - bitmap_clear(fifo->mask, 0, fifo->nr); - ret = nvkm_engine_ctor(&nvkm_fifo, device, type, inst, true, &fifo->engine); if (ret) return ret; - if (func->uevent_init) { - ret = nvkm_event_init(&nvkm_fifo_uevent_func, 1, 1, - &fifo->uevent); + if (func->nonstall) { + ret = nvkm_event_init(func->nonstall, &fifo->engine.subdev, 1, 1, + &fifo->nonstall.event); if (ret) return ret; } - return nvkm_event_init(&nvkm_fifo_kevent_func, 1, nr, &fifo->kevent); + return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/cgrp.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/cgrp.c new file mode 100644 index 000000000000..ea53fb3d5d06 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/cgrp.c @@ -0,0 +1,252 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "cgrp.h" +#include "chan.h" +#include "chid.h" +#include "runl.h" +#include "priv.h" + +#include <core/gpuobj.h> +#include <subdev/mmu.h> + +static void +nvkm_cgrp_ectx_put(struct nvkm_cgrp *cgrp, struct nvkm_ectx **pectx) +{ + struct nvkm_ectx *ectx = *pectx; + + if (ectx) { + struct nvkm_engn *engn = ectx->engn; + + if (refcount_dec_and_test(&ectx->refs)) { + CGRP_TRACE(cgrp, "dtor ectx %d[%s]", engn->id, engn->engine->subdev.name); + nvkm_object_del(&ectx->object); + list_del(&ectx->head); + kfree(ectx); + } + + *pectx = NULL; + } +} + +static int +nvkm_cgrp_ectx_get(struct nvkm_cgrp *cgrp, struct nvkm_engn *engn, struct nvkm_ectx **pectx, + struct nvkm_chan *chan, struct nvkm_client *client) +{ + struct nvkm_engine *engine = engn->engine; + struct nvkm_oclass cclass = { + .client = client, + .engine = engine, + }; + struct nvkm_ectx *ectx; + int ret = 0; + + /* Look for an existing context for this engine in the channel group. */ + ectx = nvkm_list_find(ectx, &cgrp->ectxs, head, ectx->engn == engn); + if (ectx) { + refcount_inc(&ectx->refs); + *pectx = ectx; + return 0; + } + + /* Nope - create a fresh one. */ + CGRP_TRACE(cgrp, "ctor ectx %d[%s]", engn->id, engn->engine->subdev.name); + if (!(ectx = *pectx = kzalloc(sizeof(*ectx), GFP_KERNEL))) + return -ENOMEM; + + ectx->engn = engn; + refcount_set(&ectx->refs, 1); + refcount_set(&ectx->uses, 0); + list_add_tail(&ectx->head, &cgrp->ectxs); + + /* Allocate the HW structures. */ + if (engine->func->fifo.cclass) + ret = engine->func->fifo.cclass(chan, &cclass, &ectx->object); + else if (engine->func->cclass) + ret = nvkm_object_new_(engine->func->cclass, &cclass, NULL, 0, &ectx->object); + + if (ret) + nvkm_cgrp_ectx_put(cgrp, pectx); + + return ret; +} + +void +nvkm_cgrp_vctx_put(struct nvkm_cgrp *cgrp, struct nvkm_vctx **pvctx) +{ + struct nvkm_vctx *vctx = *pvctx; + + if (vctx) { + struct nvkm_engn *engn = vctx->ectx->engn; + + if (refcount_dec_and_test(&vctx->refs)) { + CGRP_TRACE(cgrp, "dtor vctx %d[%s]", engn->id, engn->engine->subdev.name); + nvkm_vmm_put(vctx->vmm, &vctx->vma); + nvkm_gpuobj_del(&vctx->inst); + + nvkm_cgrp_ectx_put(cgrp, &vctx->ectx); + if (vctx->vmm) { + atomic_dec(&vctx->vmm->engref[engn->engine->subdev.type]); + nvkm_vmm_unref(&vctx->vmm); + } + list_del(&vctx->head); + kfree(vctx); + } + + *pvctx = NULL; + } +} + +int +nvkm_cgrp_vctx_get(struct nvkm_cgrp *cgrp, struct nvkm_engn *engn, struct nvkm_chan *chan, + struct nvkm_vctx **pvctx, struct nvkm_client *client) +{ + struct nvkm_ectx *ectx; + struct nvkm_vctx *vctx; + int ret; + + /* Look for an existing sub-context for this engine+VEID in the channel group. */ + vctx = nvkm_list_find(vctx, &cgrp->vctxs, head, + vctx->ectx->engn == engn && vctx->vmm == chan->vmm); + if (vctx) { + refcount_inc(&vctx->refs); + *pvctx = vctx; + return 0; + } + + /* Nope - create a fresh one. But, context first. */ + ret = nvkm_cgrp_ectx_get(cgrp, engn, &ectx, chan, client); + if (ret) { + CGRP_ERROR(cgrp, "ectx %d[%s]: %d", engn->id, engn->engine->subdev.name, ret); + return ret; + } + + /* Now, create the sub-context. */ + CGRP_TRACE(cgrp, "ctor vctx %d[%s]", engn->id, engn->engine->subdev.name); + if (!(vctx = *pvctx = kzalloc(sizeof(*vctx), GFP_KERNEL))) { + nvkm_cgrp_ectx_put(cgrp, &ectx); + return -ENOMEM; + } + + vctx->ectx = ectx; + vctx->vmm = nvkm_vmm_ref(chan->vmm); + refcount_set(&vctx->refs, 1); + list_add_tail(&vctx->head, &cgrp->vctxs); + + /* MMU on some GPUs needs to know engine usage for TLB invalidation. */ + if (vctx->vmm) + atomic_inc(&vctx->vmm->engref[engn->engine->subdev.type]); + + /* Allocate the HW structures. */ + if (engn->func->bind) { + ret = nvkm_object_bind(vctx->ectx->object, NULL, 0, &vctx->inst); + if (ret == 0 && engn->func->ctor) + ret = engn->func->ctor(engn, vctx); + } + + if (ret) + nvkm_cgrp_vctx_put(cgrp, pvctx); + + return ret; +} + +static void +nvkm_cgrp_del(struct kref *kref) +{ + struct nvkm_cgrp *cgrp = container_of(kref, typeof(*cgrp), kref); + struct nvkm_runl *runl = cgrp->runl; + + if (runl->cgid) + nvkm_chid_put(runl->cgid, cgrp->id, &cgrp->lock); + + mutex_destroy(&cgrp->mutex); + nvkm_vmm_unref(&cgrp->vmm); + kfree(cgrp); +} + +void +nvkm_cgrp_unref(struct nvkm_cgrp **pcgrp) +{ + struct nvkm_cgrp *cgrp = *pcgrp; + + if (!cgrp) + return; + + kref_put(&cgrp->kref, nvkm_cgrp_del); + *pcgrp = NULL; +} + +struct nvkm_cgrp * +nvkm_cgrp_ref(struct nvkm_cgrp *cgrp) +{ + if (cgrp) + kref_get(&cgrp->kref); + + return cgrp; +} + +void +nvkm_cgrp_put(struct nvkm_cgrp **pcgrp, unsigned long irqflags) +{ + struct nvkm_cgrp *cgrp = *pcgrp; + + if (!cgrp) + return; + + *pcgrp = NULL; + spin_unlock_irqrestore(&cgrp->lock, irqflags); +} + +int +nvkm_cgrp_new(struct nvkm_runl *runl, const char *name, struct nvkm_vmm *vmm, bool hw, + struct nvkm_cgrp **pcgrp) +{ + struct nvkm_cgrp *cgrp; + + if (!(cgrp = *pcgrp = kmalloc(sizeof(*cgrp), GFP_KERNEL))) + return -ENOMEM; + + cgrp->func = runl->fifo->func->cgrp.func; + strscpy(cgrp->name, name, sizeof(cgrp->name)); + cgrp->runl = runl; + cgrp->vmm = nvkm_vmm_ref(vmm); + cgrp->hw = hw; + cgrp->id = -1; + kref_init(&cgrp->kref); + INIT_LIST_HEAD(&cgrp->chans); + cgrp->chan_nr = 0; + spin_lock_init(&cgrp->lock); + INIT_LIST_HEAD(&cgrp->ectxs); + INIT_LIST_HEAD(&cgrp->vctxs); + mutex_init(&cgrp->mutex); + atomic_set(&cgrp->rc, NVKM_CGRP_RC_NONE); + + if (runl->cgid) { + cgrp->id = nvkm_chid_get(runl->cgid, cgrp); + if (cgrp->id < 0) { + RUNL_ERROR(runl, "!cgids"); + nvkm_cgrp_unref(pcgrp); + return -ENOSPC; + } + } + + return 0; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/cgrp.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/cgrp.h index d0ac60b06720..5f6abd59a6ff 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/cgrp.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/cgrp.h @@ -1,11 +1,75 @@ -#ifndef __NVKM_FIFO_CGRP_H__ -#define __NVKM_FIFO_CGRP_H__ -#include "priv.h" +/* SPDX-License-Identifier: MIT */ +#ifndef __NVKM_CGRP_H__ +#define __NVKM_CGRP_H__ +#include <core/os.h> +struct nvkm_chan; +struct nvkm_client; + +struct nvkm_vctx { + struct nvkm_ectx *ectx; + struct nvkm_vmm *vmm; + refcount_t refs; + + struct nvkm_gpuobj *inst; + struct nvkm_vma *vma; -struct nvkm_fifo_cgrp { - int id; struct list_head head; - struct list_head chan; +}; + +struct nvkm_ectx { + struct nvkm_engn *engn; + refcount_t refs; + refcount_t uses; + + struct nvkm_object *object; + + struct list_head head; +}; + +struct nvkm_cgrp { + const struct nvkm_cgrp_func { + void (*preempt)(struct nvkm_cgrp *); + } *func; + char name[64]; + struct nvkm_runl *runl; + struct nvkm_vmm *vmm; + bool hw; + int id; + struct kref kref; + + struct list_head chans; int chan_nr; + + spinlock_t lock; /* protects irq handler channel (group) lookup */ + + struct list_head ectxs; + struct list_head vctxs; + struct mutex mutex; + +#define NVKM_CGRP_RC_NONE 0 +#define NVKM_CGRP_RC_PENDING 1 +#define NVKM_CGRP_RC_RUNNING 2 + atomic_t rc; + + struct list_head head; }; + +int nvkm_cgrp_new(struct nvkm_runl *, const char *name, struct nvkm_vmm *, bool hw, + struct nvkm_cgrp **); +struct nvkm_cgrp *nvkm_cgrp_ref(struct nvkm_cgrp *); +void nvkm_cgrp_unref(struct nvkm_cgrp **); +int nvkm_cgrp_vctx_get(struct nvkm_cgrp *, struct nvkm_engn *, struct nvkm_chan *, + struct nvkm_vctx **, struct nvkm_client *); +void nvkm_cgrp_vctx_put(struct nvkm_cgrp *, struct nvkm_vctx **); + +void nvkm_cgrp_put(struct nvkm_cgrp **, unsigned long irqflags); + +#define nvkm_cgrp_foreach_chan(chan,cgrp) list_for_each_entry((chan), &(cgrp)->chans, head) +#define nvkm_cgrp_foreach_chan_safe(chan,ctmp,cgrp) \ + list_for_each_entry_safe((chan), (ctmp), &(cgrp)->chans, head) + +#define CGRP_PRCLI(c,l,p,f,a...) RUNL_PRINT((c)->runl, l, p, "%04x:[%s]"f, (c)->id, (c)->name, ##a) +#define CGRP_PRINT(c,l,p,f,a...) RUNL_PRINT((c)->runl, l, p, "%04x:"f, (c)->id, ##a) +#define CGRP_ERROR(c,f,a...) CGRP_PRCLI((c), ERROR, err, " "f"\n", ##a) +#define CGRP_TRACE(c,f,a...) CGRP_PRINT((c), TRACE, info, " "f"\n", ##a) #endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.c index 2e7f32cebf2a..b7c9d6115bce 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.c @@ -22,285 +22,265 @@ * Authors: Ben Skeggs */ #include "chan.h" +#include "chid.h" +#include "cgrp.h" +#include "chid.h" +#include "runl.h" +#include "priv.h" -#include <core/client.h> -#include <core/gpuobj.h> -#include <core/oproxy.h> +#include <core/ramht.h> #include <subdev/mmu.h> #include <engine/dma.h> -struct nvkm_fifo_chan_object { - struct nvkm_oproxy oproxy; - struct nvkm_fifo_chan *chan; - int hash; +#include <nvif/if0020.h> + +const struct nvkm_event_func +nvkm_chan_event = { }; -static struct nvkm_fifo_engn * -nvkm_fifo_chan_engn(struct nvkm_fifo_chan *chan, struct nvkm_engine *engine) +void +nvkm_chan_cctx_bind(struct nvkm_chan *chan, struct nvkm_engn *engn, struct nvkm_cctx *cctx) { - int engi = chan->fifo->func->engine_id(chan->fifo, engine); - if (engi >= 0) - return &chan->engn[engi]; - return NULL; + struct nvkm_cgrp *cgrp = chan->cgrp; + struct nvkm_runl *runl = cgrp->runl; + struct nvkm_engine *engine = engn->engine; + + if (!engn->func->bind) + return; + + CHAN_TRACE(chan, "%sbind cctx %d[%s]", cctx ? "" : "un", engn->id, engine->subdev.name); + + /* Prevent any channel in channel group from being rescheduled, kick them + * off host and any engine(s) they're loaded on. + */ + if (cgrp->hw) + nvkm_runl_block(runl); + else + nvkm_chan_block(chan); + nvkm_chan_preempt(chan, true); + + /* Update context pointer. */ + engn->func->bind(engn, cctx, chan); + + /* Resume normal operation. */ + if (cgrp->hw) + nvkm_runl_allow(runl); + else + nvkm_chan_allow(chan); } -static int -nvkm_fifo_chan_child_fini(struct nvkm_oproxy *base, bool suspend) +void +nvkm_chan_cctx_put(struct nvkm_chan *chan, struct nvkm_cctx **pcctx) { - struct nvkm_fifo_chan_object *object = - container_of(base, typeof(*object), oproxy); - struct nvkm_engine *engine = object->oproxy.object->engine; - struct nvkm_fifo_chan *chan = object->chan; - struct nvkm_fifo_engn *engn = nvkm_fifo_chan_engn(chan, engine); - const char *name = engine->subdev.name; - int ret = 0; - - if (--engn->usecount) - return 0; + struct nvkm_cctx *cctx = *pcctx; - if (chan->func->engine_fini) { - ret = chan->func->engine_fini(chan, engine, suspend); - if (ret) { - nvif_error(&chan->object, - "detach %s failed, %d\n", name, ret); - return ret; + if (cctx) { + struct nvkm_engn *engn = cctx->vctx->ectx->engn; + + if (refcount_dec_and_mutex_lock(&cctx->refs, &chan->cgrp->mutex)) { + CHAN_TRACE(chan, "dtor cctx %d[%s]", engn->id, engn->engine->subdev.name); + nvkm_cgrp_vctx_put(chan->cgrp, &cctx->vctx); + list_del(&cctx->head); + kfree(cctx); + mutex_unlock(&chan->cgrp->mutex); } - } - if (engn->object) { - ret = nvkm_object_fini(engn->object, suspend); - if (ret && suspend) - return ret; + *pcctx = NULL; } - - nvif_trace(&chan->object, "detached %s\n", name); - return ret; } -static int -nvkm_fifo_chan_child_init(struct nvkm_oproxy *base) +int +nvkm_chan_cctx_get(struct nvkm_chan *chan, struct nvkm_engn *engn, struct nvkm_cctx **pcctx, + struct nvkm_client *client) { - struct nvkm_fifo_chan_object *object = - container_of(base, typeof(*object), oproxy); - struct nvkm_engine *engine = object->oproxy.object->engine; - struct nvkm_fifo_chan *chan = object->chan; - struct nvkm_fifo_engn *engn = nvkm_fifo_chan_engn(chan, engine); - const char *name = engine->subdev.name; + struct nvkm_cgrp *cgrp = chan->cgrp; + struct nvkm_vctx *vctx; + struct nvkm_cctx *cctx; int ret; - if (engn->usecount++) + /* Look for an existing channel context for this engine+VEID. */ + mutex_lock(&cgrp->mutex); + cctx = nvkm_list_find(cctx, &chan->cctxs, head, + cctx->vctx->ectx->engn == engn && cctx->vctx->vmm == chan->vmm); + if (cctx) { + refcount_inc(&cctx->refs); + *pcctx = cctx; + mutex_unlock(&chan->cgrp->mutex); return 0; + } - if (engn->object) { - ret = nvkm_object_init(engn->object); - if (ret) - return ret; + /* Nope - create a fresh one. But, sub-context first. */ + ret = nvkm_cgrp_vctx_get(cgrp, engn, chan, &vctx, client); + if (ret) { + CHAN_ERROR(chan, "vctx %d[%s]: %d", engn->id, engn->engine->subdev.name, ret); + goto done; } - if (chan->func->engine_init) { - ret = chan->func->engine_init(chan, engine); - if (ret) { - nvif_error(&chan->object, - "attach %s failed, %d\n", name, ret); - return ret; - } + /* Now, create the channel context - to track engine binding. */ + CHAN_TRACE(chan, "ctor cctx %d[%s]", engn->id, engn->engine->subdev.name); + if (!(cctx = *pcctx = kzalloc(sizeof(*cctx), GFP_KERNEL))) { + nvkm_cgrp_vctx_put(cgrp, &vctx); + ret = -ENOMEM; + goto done; } - nvif_trace(&chan->object, "attached %s\n", name); - return 0; + cctx->vctx = vctx; + refcount_set(&cctx->refs, 1); + refcount_set(&cctx->uses, 0); + list_add_tail(&cctx->head, &chan->cctxs); +done: + mutex_unlock(&cgrp->mutex); + return ret; } -static void -nvkm_fifo_chan_child_del(struct nvkm_oproxy *base) +int +nvkm_chan_preempt_locked(struct nvkm_chan *chan, bool wait) { - struct nvkm_fifo_chan_object *object = - container_of(base, typeof(*object), oproxy); - struct nvkm_engine *engine = object->oproxy.base.engine; - struct nvkm_fifo_chan *chan = object->chan; - struct nvkm_fifo_engn *engn = nvkm_fifo_chan_engn(chan, engine); - - if (chan->func->object_dtor) - chan->func->object_dtor(chan, object->hash); - - if (!--engn->refcount) { - if (chan->func->engine_dtor) - chan->func->engine_dtor(chan, engine); - nvkm_object_del(&engn->object); - if (chan->vmm) - atomic_dec(&chan->vmm->engref[engine->subdev.type]); - } -} + struct nvkm_runl *runl = chan->cgrp->runl; -static const struct nvkm_oproxy_func -nvkm_fifo_chan_child_func = { - .dtor[0] = nvkm_fifo_chan_child_del, - .init[0] = nvkm_fifo_chan_child_init, - .fini[0] = nvkm_fifo_chan_child_fini, -}; + CHAN_TRACE(chan, "preempt"); + chan->func->preempt(chan); + if (!wait) + return 0; + + return nvkm_runl_preempt_wait(runl); +} -static int -nvkm_fifo_chan_child_new(const struct nvkm_oclass *oclass, void *data, u32 size, - struct nvkm_object **pobject) +int +nvkm_chan_preempt(struct nvkm_chan *chan, bool wait) { - struct nvkm_engine *engine = oclass->engine; - struct nvkm_fifo_chan *chan = nvkm_fifo_chan(oclass->parent); - struct nvkm_fifo_engn *engn = nvkm_fifo_chan_engn(chan, engine); - struct nvkm_fifo_chan_object *object; - int ret = 0; + int ret; - if (!(object = kzalloc(sizeof(*object), GFP_KERNEL))) - return -ENOMEM; - nvkm_oproxy_ctor(&nvkm_fifo_chan_child_func, oclass, &object->oproxy); - object->chan = chan; - *pobject = &object->oproxy.base; - - if (!engn->refcount++) { - struct nvkm_oclass cclass = { - .client = oclass->client, - .engine = oclass->engine, - }; - - if (chan->vmm) - atomic_inc(&chan->vmm->engref[engine->subdev.type]); - - if (engine->func->fifo.cclass) { - ret = engine->func->fifo.cclass(chan, &cclass, - &engn->object); - } else - if (engine->func->cclass) { - ret = nvkm_object_new_(engine->func->cclass, &cclass, - NULL, 0, &engn->object); - } - if (ret) - return ret; + if (!chan->func->preempt) + return 0; - if (chan->func->engine_ctor) { - ret = chan->func->engine_ctor(chan, oclass->engine, - engn->object); - if (ret) - return ret; - } - } + mutex_lock(&chan->cgrp->runl->mutex); + ret = nvkm_chan_preempt_locked(chan, wait); + mutex_unlock(&chan->cgrp->runl->mutex); + return ret; +} - ret = oclass->base.ctor(&(const struct nvkm_oclass) { - .base = oclass->base, - .engn = oclass->engn, - .handle = oclass->handle, - .object = oclass->object, - .client = oclass->client, - .parent = engn->object ? - engn->object : - oclass->parent, - .engine = engine, - }, data, size, &object->oproxy.object); - if (ret) - return ret; +void +nvkm_chan_remove_locked(struct nvkm_chan *chan) +{ + struct nvkm_cgrp *cgrp = chan->cgrp; + struct nvkm_runl *runl = cgrp->runl; - if (chan->func->object_ctor) { - object->hash = - chan->func->object_ctor(chan, object->oproxy.object); - if (object->hash < 0) - return object->hash; - } + if (list_empty(&chan->head)) + return; - return 0; + CHAN_TRACE(chan, "remove"); + if (!--cgrp->chan_nr) { + runl->cgrp_nr--; + list_del(&cgrp->head); + } + runl->chan_nr--; + list_del_init(&chan->head); + atomic_set(&runl->changed, 1); } -static int -nvkm_fifo_chan_child_get(struct nvkm_object *object, int index, - struct nvkm_oclass *oclass) +void +nvkm_chan_remove(struct nvkm_chan *chan, bool preempt) { - struct nvkm_fifo_chan *chan = nvkm_fifo_chan(object); - struct nvkm_fifo *fifo = chan->fifo; - struct nvkm_engine *engine; - u32 engm = chan->engm; - int engi, ret, c; - - for (; c = 0, engi = __ffs(engm), engm; engm &= ~(1ULL << engi)) { - if (!(engine = fifo->func->id_engine(fifo, engi))) - continue; - oclass->engine = engine; - oclass->base.oclass = 0; - - if (engine->func->fifo.sclass) { - ret = engine->func->fifo.sclass(oclass, index); - if (oclass->base.oclass) { - if (!oclass->base.ctor) - oclass->base.ctor = nvkm_object_new; - oclass->ctor = nvkm_fifo_chan_child_new; - return 0; - } + struct nvkm_runl *runl = chan->cgrp->runl; + + mutex_lock(&runl->mutex); + if (preempt && chan->func->preempt) + nvkm_chan_preempt_locked(chan, true); + nvkm_chan_remove_locked(chan); + nvkm_runl_update_locked(runl, true); + mutex_unlock(&runl->mutex); +} - index -= ret; - continue; - } +void +nvkm_chan_insert(struct nvkm_chan *chan) +{ + struct nvkm_cgrp *cgrp = chan->cgrp; + struct nvkm_runl *runl = cgrp->runl; - while (engine->func->sclass[c].oclass) { - if (c++ == index) { - oclass->base = engine->func->sclass[index]; - if (!oclass->base.ctor) - oclass->base.ctor = nvkm_object_new; - oclass->ctor = nvkm_fifo_chan_child_new; - return 0; - } - } - index -= c; + mutex_lock(&runl->mutex); + if (WARN_ON(!list_empty(&chan->head))) { + mutex_unlock(&runl->mutex); + return; } - return -EINVAL; + CHAN_TRACE(chan, "insert"); + list_add_tail(&chan->head, &cgrp->chans); + runl->chan_nr++; + if (!cgrp->chan_nr++) { + list_add_tail(&cgrp->head, &cgrp->runl->cgrps); + runl->cgrp_nr++; + } + atomic_set(&runl->changed, 1); + nvkm_runl_update_locked(runl, true); + mutex_unlock(&runl->mutex); } -static int -nvkm_fifo_chan_ntfy(struct nvkm_object *object, u32 type, - struct nvkm_event **pevent) +static void +nvkm_chan_block_locked(struct nvkm_chan *chan) { - struct nvkm_fifo_chan *chan = nvkm_fifo_chan(object); - if (chan->func->ntfy) - return chan->func->ntfy(chan, type, pevent); - return -ENODEV; + CHAN_TRACE(chan, "block %d", atomic_read(&chan->blocked)); + if (atomic_inc_return(&chan->blocked) == 1) + chan->func->stop(chan); } -static int -nvkm_fifo_chan_map(struct nvkm_object *object, void *argv, u32 argc, - enum nvkm_object_map *type, u64 *addr, u64 *size) +void +nvkm_chan_error(struct nvkm_chan *chan, bool preempt) { - struct nvkm_fifo_chan *chan = nvkm_fifo_chan(object); - *type = NVKM_OBJECT_MAP_IO; - *addr = chan->addr; - *size = chan->size; - return 0; + unsigned long flags; + + spin_lock_irqsave(&chan->lock, flags); + if (atomic_inc_return(&chan->errored) == 1) { + CHAN_ERROR(chan, "errored - disabling channel"); + nvkm_chan_block_locked(chan); + if (preempt) + chan->func->preempt(chan); + nvkm_event_ntfy(&chan->cgrp->runl->chid->event, chan->id, NVKM_CHAN_EVENT_ERRORED); + } + spin_unlock_irqrestore(&chan->lock, flags); } -static int -nvkm_fifo_chan_fini(struct nvkm_object *object, bool suspend) +void +nvkm_chan_block(struct nvkm_chan *chan) { - struct nvkm_fifo_chan *chan = nvkm_fifo_chan(object); - chan->func->fini(chan); - return 0; + spin_lock_irq(&chan->lock); + nvkm_chan_block_locked(chan); + spin_unlock_irq(&chan->lock); } -static int -nvkm_fifo_chan_init(struct nvkm_object *object) +void +nvkm_chan_allow(struct nvkm_chan *chan) { - struct nvkm_fifo_chan *chan = nvkm_fifo_chan(object); - chan->func->init(chan); - return 0; + spin_lock_irq(&chan->lock); + CHAN_TRACE(chan, "allow %d", atomic_read(&chan->blocked)); + if (atomic_dec_and_test(&chan->blocked)) + chan->func->start(chan); + spin_unlock_irq(&chan->lock); } -static void * -nvkm_fifo_chan_dtor(struct nvkm_object *object) +void +nvkm_chan_del(struct nvkm_chan **pchan) { - struct nvkm_fifo_chan *chan = nvkm_fifo_chan(object); - struct nvkm_fifo *fifo = chan->fifo; - void *data = chan->func->dtor(chan); - unsigned long flags; + struct nvkm_chan *chan = *pchan; + + if (!chan) + return; + + if (chan->func->ramfc->clear) + chan->func->ramfc->clear(chan); - spin_lock_irqsave(&fifo->lock, flags); - if (!list_empty(&chan->head)) { - __clear_bit(chan->chid, fifo->mask); - list_del(&chan->head); + nvkm_ramht_del(&chan->ramht); + nvkm_gpuobj_del(&chan->pgd); + nvkm_gpuobj_del(&chan->eng); + nvkm_gpuobj_del(&chan->cache); + nvkm_gpuobj_del(&chan->ramfc); + + nvkm_memory_unref(&chan->userd.mem); + + if (chan->cgrp) { + nvkm_chid_put(chan->cgrp->runl->chid, chan->id, &chan->cgrp->lock); + nvkm_cgrp_unref(&chan->cgrp); } - spin_unlock_irqrestore(&fifo->lock, flags); if (chan->vmm) { nvkm_vmm_part(chan->vmm, chan->inst->memory); @@ -309,85 +289,192 @@ nvkm_fifo_chan_dtor(struct nvkm_object *object) nvkm_gpuobj_del(&chan->push); nvkm_gpuobj_del(&chan->inst); - return data; + kfree(chan); } -static const struct nvkm_object_func -nvkm_fifo_chan_func = { - .dtor = nvkm_fifo_chan_dtor, - .init = nvkm_fifo_chan_init, - .fini = nvkm_fifo_chan_fini, - .ntfy = nvkm_fifo_chan_ntfy, - .map = nvkm_fifo_chan_map, - .sclass = nvkm_fifo_chan_child_get, -}; +void +nvkm_chan_put(struct nvkm_chan **pchan, unsigned long irqflags) +{ + struct nvkm_chan *chan = *pchan; + + if (!chan) + return; + + *pchan = NULL; + spin_unlock_irqrestore(&chan->cgrp->lock, irqflags); +} + +struct nvkm_chan * +nvkm_chan_get_inst(struct nvkm_engine *engine, u64 inst, unsigned long *pirqflags) +{ + struct nvkm_fifo *fifo = engine->subdev.device->fifo; + struct nvkm_runl *runl; + struct nvkm_engn *engn; + struct nvkm_chan *chan; + + nvkm_runl_foreach(runl, fifo) { + nvkm_runl_foreach_engn(engn, runl) { + if (engine == &fifo->engine || engn->engine == engine) { + chan = nvkm_runl_chan_get_inst(runl, inst, pirqflags); + if (chan || engn->engine == engine) + return chan; + } + } + } + + return NULL; +} + +struct nvkm_chan * +nvkm_chan_get_chid(struct nvkm_engine *engine, int id, unsigned long *pirqflags) +{ + struct nvkm_fifo *fifo = engine->subdev.device->fifo; + struct nvkm_runl *runl; + struct nvkm_engn *engn; + + nvkm_runl_foreach(runl, fifo) { + nvkm_runl_foreach_engn(engn, runl) { + if (fifo->chid || engn->engine == engine) + return nvkm_runl_chan_get_chid(runl, id, pirqflags); + } + } + + return NULL; +} int -nvkm_fifo_chan_ctor(const struct nvkm_fifo_chan_func *func, - struct nvkm_fifo *fifo, u32 size, u32 align, bool zero, - u64 hvmm, u64 push, u32 engm, int bar, u32 base, - u32 user, const struct nvkm_oclass *oclass, - struct nvkm_fifo_chan *chan) +nvkm_chan_new_(const struct nvkm_chan_func *func, struct nvkm_runl *runl, int runq, + struct nvkm_cgrp *cgrp, const char *name, bool priv, u32 devm, struct nvkm_vmm *vmm, + struct nvkm_dmaobj *dmaobj, u64 offset, u64 length, + struct nvkm_memory *userd, u64 ouserd, struct nvkm_chan **pchan) { - struct nvkm_client *client = oclass->client; + struct nvkm_fifo *fifo = runl->fifo; struct nvkm_device *device = fifo->engine.subdev.device; - struct nvkm_dmaobj *dmaobj; - unsigned long flags; + struct nvkm_chan *chan; int ret; - nvkm_object_ctor(&nvkm_fifo_chan_func, oclass, &chan->object); + /* Validate arguments against class requirements. */ + if ((runq && runq >= runl->func->runqs) || + (!func->inst->vmm != !vmm) || + ((func->userd->bar < 0) == !userd) || + (!func->ramfc->ctxdma != !dmaobj) || + ((func->ramfc->devm < devm) && devm != BIT(0)) || + (!func->ramfc->priv && priv)) { + RUNL_DEBUG(runl, "args runq:%d:%d vmm:%d:%p userd:%d:%p " + "push:%d:%p devm:%08x:%08x priv:%d:%d", + runl->func->runqs, runq, func->inst->vmm, vmm, + func->userd->bar < 0, userd, func->ramfc->ctxdma, dmaobj, + func->ramfc->devm, devm, func->ramfc->priv, priv); + return -EINVAL; + } + + if (!(chan = *pchan = kzalloc(sizeof(*chan), GFP_KERNEL))) + return -ENOMEM; + chan->func = func; - chan->fifo = fifo; - chan->engm = engm; + strscpy(chan->name, name, sizeof(chan->name)); + chan->runq = runq; + chan->id = -1; + spin_lock_init(&chan->lock); + atomic_set(&chan->blocked, 1); + atomic_set(&chan->errored, 0); + INIT_LIST_HEAD(&chan->cctxs); INIT_LIST_HEAD(&chan->head); - /* instance memory */ - ret = nvkm_gpuobj_new(device, size, align, zero, NULL, &chan->inst); - if (ret) - return ret; + /* Join channel group. + * + * GK110 and newer support channel groups (aka TSGs), where individual channels + * share a timeslice, and, engine context(s). + * + * As such, engine contexts are tracked in nvkm_cgrp and we need them even when + * channels aren't in an API channel group, and on HW that doesn't support TSGs. + */ + if (!cgrp) { + ret = nvkm_cgrp_new(runl, chan->name, vmm, fifo->func->cgrp.force, &chan->cgrp); + if (ret) { + RUNL_DEBUG(runl, "cgrp %d", ret); + return ret; + } - /* allocate push buffer ctxdma instance */ - if (push) { - dmaobj = nvkm_dmaobj_search(client, push); - if (IS_ERR(dmaobj)) - return PTR_ERR(dmaobj); + cgrp = chan->cgrp; + } else { + if (cgrp->runl != runl || cgrp->vmm != vmm) { + RUNL_DEBUG(runl, "cgrp %d %d", cgrp->runl != runl, cgrp->vmm != vmm); + return -EINVAL; + } - ret = nvkm_object_bind(&dmaobj->object, chan->inst, -16, - &chan->push); - if (ret) - return ret; + chan->cgrp = nvkm_cgrp_ref(cgrp); } - /* channel address space */ - if (hvmm) { - struct nvkm_vmm *vmm = nvkm_uvmm_search(client, hvmm); - if (IS_ERR(vmm)) - return PTR_ERR(vmm); + /* Allocate instance block. */ + ret = nvkm_gpuobj_new(device, func->inst->size, 0x1000, func->inst->zero, NULL, + &chan->inst); + if (ret) { + RUNL_DEBUG(runl, "inst %d", ret); + return ret; + } - if (vmm->mmu != device->mmu) + /* Initialise virtual address-space. */ + if (func->inst->vmm) { + if (WARN_ON(vmm->mmu != device->mmu)) return -EINVAL; ret = nvkm_vmm_join(vmm, chan->inst->memory); - if (ret) + if (ret) { + RUNL_DEBUG(runl, "vmm %d", ret); return ret; + } chan->vmm = nvkm_vmm_ref(vmm); } - /* allocate channel id */ - spin_lock_irqsave(&fifo->lock, flags); - chan->chid = find_first_zero_bit(fifo->mask, NVKM_FIFO_CHID_NR); - if (chan->chid >= NVKM_FIFO_CHID_NR) { - spin_unlock_irqrestore(&fifo->lock, flags); + /* Allocate HW ctxdma for push buffer. */ + if (func->ramfc->ctxdma) { + ret = nvkm_object_bind(&dmaobj->object, chan->inst, -16, &chan->push); + if (ret) { + RUNL_DEBUG(runl, "bind %d", ret); + return ret; + } + } + + /* Allocate channel ID. */ + chan->id = nvkm_chid_get(runl->chid, chan); + if (chan->id < 0) { + RUNL_ERROR(runl, "!chids"); return -ENOSPC; } - list_add(&chan->head, &fifo->chan); - __set_bit(chan->chid, fifo->mask); - spin_unlock_irqrestore(&fifo->lock, flags); - - /* determine address of this channel's user registers */ - chan->addr = device->func->resource_addr(device, bar) + - base + user * chan->chid; - chan->size = user; + + if (cgrp->id < 0) + cgrp->id = chan->id; + + /* Initialise USERD. */ + if (func->userd->bar < 0) { + if (ouserd + chan->func->userd->size >= nvkm_memory_size(userd)) { + RUNL_DEBUG(runl, "ouserd %llx", ouserd); + return -EINVAL; + } + + ret = nvkm_memory_kmap(userd, &chan->userd.mem); + if (ret) { + RUNL_DEBUG(runl, "userd %d", ret); + return ret; + } + + chan->userd.base = ouserd; + } else { + chan->userd.mem = nvkm_memory_ref(fifo->userd.mem); + chan->userd.base = chan->id * chan->func->userd->size; + } + + if (chan->func->userd->clear) + chan->func->userd->clear(chan); + + /* Initialise RAMFC. */ + ret = chan->func->ramfc->write(chan, offset, length, devm, priv); + if (ret) { + RUNL_DEBUG(runl, "ramfc %d", ret); + return ret; + } + return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.h index e53504354841..85b94f699128 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chan.h @@ -1,35 +1,78 @@ /* SPDX-License-Identifier: MIT */ -#ifndef __NVKM_FIFO_CHAN_H__ -#define __NVKM_FIFO_CHAN_H__ -#define nvkm_fifo_chan(p) container_of((p), struct nvkm_fifo_chan, object) -#include "priv.h" - -struct nvkm_fifo_chan_func { - void *(*dtor)(struct nvkm_fifo_chan *); - void (*init)(struct nvkm_fifo_chan *); - void (*fini)(struct nvkm_fifo_chan *); - int (*ntfy)(struct nvkm_fifo_chan *, u32 type, struct nvkm_event **); - int (*engine_ctor)(struct nvkm_fifo_chan *, struct nvkm_engine *, - struct nvkm_object *); - void (*engine_dtor)(struct nvkm_fifo_chan *, struct nvkm_engine *); - int (*engine_init)(struct nvkm_fifo_chan *, struct nvkm_engine *); - int (*engine_fini)(struct nvkm_fifo_chan *, struct nvkm_engine *, - bool suspend); - int (*object_ctor)(struct nvkm_fifo_chan *, struct nvkm_object *); - void (*object_dtor)(struct nvkm_fifo_chan *, int); - u32 (*submit_token)(struct nvkm_fifo_chan *); +#ifndef __NVKM_CHAN_H__ +#define __NVKM_CHAN_H__ +#include <engine/fifo.h> +struct nvkm_dmaobj; +struct nvkm_engn; +struct nvkm_runl; + +extern const struct nvkm_event_func nvkm_chan_event; + +struct nvkm_cctx { + struct nvkm_vctx *vctx; + refcount_t refs; + refcount_t uses; + + struct list_head head; }; -int nvkm_fifo_chan_ctor(const struct nvkm_fifo_chan_func *, struct nvkm_fifo *, - u32 size, u32 align, bool zero, u64 vm, u64 push, - u32 engm, int bar, u32 base, u32 user, - const struct nvkm_oclass *, struct nvkm_fifo_chan *); +struct nvkm_chan_func { + const struct nvkm_chan_func_inst { + u32 size; + bool zero; + bool vmm; + } *inst; -struct nvkm_fifo_chan_oclass { - int (*ctor)(struct nvkm_fifo *, const struct nvkm_oclass *, - void *data, u32 size, struct nvkm_object **); - struct nvkm_sclass base; + const struct nvkm_chan_func_userd { + int bar; + u32 base; + u32 size; + void (*clear)(struct nvkm_chan *); + } *userd; + + const struct nvkm_chan_func_ramfc { + const struct nvkm_ramfc_layout { + unsigned bits:6; + unsigned ctxs:5; + unsigned ctxp:8; + unsigned regs:5; + unsigned regp; + } *layout; + int (*write)(struct nvkm_chan *, u64 offset, u64 length, u32 devm, bool priv); + void (*clear)(struct nvkm_chan *); + bool ctxdma; + u32 devm; + bool priv; + } *ramfc; + + void (*bind)(struct nvkm_chan *); + void (*unbind)(struct nvkm_chan *); + void (*start)(struct nvkm_chan *); + void (*stop)(struct nvkm_chan *); + void (*preempt)(struct nvkm_chan *); + u32 (*doorbell_handle)(struct nvkm_chan *); }; -int gf100_fifo_chan_ntfy(struct nvkm_fifo_chan *, u32, struct nvkm_event **); +int nvkm_chan_new_(const struct nvkm_chan_func *, struct nvkm_runl *, int runq, struct nvkm_cgrp *, + const char *name, bool priv, u32 devm, struct nvkm_vmm *, struct nvkm_dmaobj *, + u64 offset, u64 length, struct nvkm_memory *userd, u64 userd_bar1, + struct nvkm_chan **); +void nvkm_chan_del(struct nvkm_chan **); +void nvkm_chan_allow(struct nvkm_chan *); +void nvkm_chan_block(struct nvkm_chan *); +void nvkm_chan_error(struct nvkm_chan *, bool preempt); +void nvkm_chan_insert(struct nvkm_chan *); +void nvkm_chan_remove(struct nvkm_chan *, bool preempt); +void nvkm_chan_remove_locked(struct nvkm_chan *); +int nvkm_chan_preempt(struct nvkm_chan *, bool wait); +int nvkm_chan_preempt_locked(struct nvkm_chan *, bool wait); +int nvkm_chan_cctx_get(struct nvkm_chan *, struct nvkm_engn *, struct nvkm_cctx **, + struct nvkm_client * /*TODO: remove need for this */); +void nvkm_chan_cctx_put(struct nvkm_chan *, struct nvkm_cctx **); +void nvkm_chan_cctx_bind(struct nvkm_chan *, struct nvkm_engn *, struct nvkm_cctx *); + +#define CHAN_PRCLI(c,l,p,f,a...) CGRP_PRINT((c)->cgrp, l, p, "%04x:[%s]"f, (c)->id, (c)->name, ##a) +#define CHAN_PRINT(c,l,p,f,a...) CGRP_PRINT((c)->cgrp, l, p, "%04x:"f, (c)->id, ##a) +#define CHAN_ERROR(c,f,a...) CHAN_PRCLI((c), ERROR, err, " "f"\n", ##a) +#define CHAN_TRACE(c,f,a...) CHAN_PRINT((c), TRACE, info, " "f"\n", ##a) #endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chang84.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chang84.c deleted file mode 100644 index 3492c561f2cf..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chang84.c +++ /dev/null @@ -1,263 +0,0 @@ -/* - * Copyright 2012 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#include "channv50.h" - -#include <core/client.h> -#include <core/ramht.h> -#include <subdev/mmu.h> -#include <subdev/timer.h> - -#include <nvif/cl826e.h> - -static int -g84_fifo_chan_ntfy(struct nvkm_fifo_chan *chan, u32 type, - struct nvkm_event **pevent) -{ - switch (type) { - case NV826E_V0_NTFY_NON_STALL_INTERRUPT: - *pevent = &chan->fifo->uevent; - return 0; - default: - break; - } - return -EINVAL; -} - -static int -g84_fifo_chan_engine_addr(struct nvkm_engine *engine) -{ - switch (engine->subdev.type) { - case NVKM_ENGINE_DMAOBJ: - case NVKM_ENGINE_SW : return -1; - case NVKM_ENGINE_GR : return 0x0020; - case NVKM_ENGINE_VP : - case NVKM_ENGINE_MSPDEC: return 0x0040; - case NVKM_ENGINE_MPEG : - case NVKM_ENGINE_MSPPP : return 0x0060; - case NVKM_ENGINE_BSP : - case NVKM_ENGINE_MSVLD : return 0x0080; - case NVKM_ENGINE_CIPHER: - case NVKM_ENGINE_SEC : return 0x00a0; - case NVKM_ENGINE_CE : return 0x00c0; - default: - WARN_ON(1); - return -1; - } -} - -static int -g84_fifo_chan_engine_fini(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine, bool suspend) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - struct nv50_fifo *fifo = chan->fifo; - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - u32 engn, save; - int offset; - bool done; - - offset = g84_fifo_chan_engine_addr(engine); - if (offset < 0) - return 0; - - engn = fifo->base.func->engine_id(&fifo->base, engine) - 1; - save = nvkm_mask(device, 0x002520, 0x0000003f, 1 << engn); - nvkm_wr32(device, 0x0032fc, chan->base.inst->addr >> 12); - done = nvkm_msec(device, 2000, - if (nvkm_rd32(device, 0x0032fc) != 0xffffffff) - break; - ) >= 0; - nvkm_wr32(device, 0x002520, save); - if (!done) { - nvkm_error(subdev, "channel %d [%s] unload timeout\n", - chan->base.chid, chan->base.object.client->name); - if (suspend) - return -EBUSY; - } - - nvkm_kmap(chan->eng); - nvkm_wo32(chan->eng, offset + 0x00, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x04, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x08, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x0c, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x10, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x14, 0x00000000); - nvkm_done(chan->eng); - return 0; -} - - -static int -g84_fifo_chan_engine_init(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - struct nvkm_gpuobj *engn = *nv50_fifo_chan_engine(chan, engine); - u64 limit, start; - int offset; - - offset = g84_fifo_chan_engine_addr(engine); - if (offset < 0) - return 0; - limit = engn->addr + engn->size - 1; - start = engn->addr; - - nvkm_kmap(chan->eng); - nvkm_wo32(chan->eng, offset + 0x00, 0x00190000); - nvkm_wo32(chan->eng, offset + 0x04, lower_32_bits(limit)); - nvkm_wo32(chan->eng, offset + 0x08, lower_32_bits(start)); - nvkm_wo32(chan->eng, offset + 0x0c, upper_32_bits(limit) << 24 | - upper_32_bits(start)); - nvkm_wo32(chan->eng, offset + 0x10, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x14, 0x00000000); - nvkm_done(chan->eng); - return 0; -} - -static int -g84_fifo_chan_engine_ctor(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine, - struct nvkm_object *object) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - - if (g84_fifo_chan_engine_addr(engine) < 0) - return 0; - - return nvkm_object_bind(object, NULL, 0, nv50_fifo_chan_engine(chan, engine)); -} - -static int -g84_fifo_chan_object_ctor(struct nvkm_fifo_chan *base, - struct nvkm_object *object) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - u32 handle = object->handle; - u32 context; - - switch (object->engine->subdev.type) { - case NVKM_ENGINE_DMAOBJ: - case NVKM_ENGINE_SW : context = 0x00000000; break; - case NVKM_ENGINE_GR : context = 0x00100000; break; - case NVKM_ENGINE_MPEG : - case NVKM_ENGINE_MSPPP : context = 0x00200000; break; - case NVKM_ENGINE_ME : - case NVKM_ENGINE_CE : context = 0x00300000; break; - case NVKM_ENGINE_VP : - case NVKM_ENGINE_MSPDEC: context = 0x00400000; break; - case NVKM_ENGINE_CIPHER: - case NVKM_ENGINE_SEC : - case NVKM_ENGINE_VIC : context = 0x00500000; break; - case NVKM_ENGINE_BSP : - case NVKM_ENGINE_MSVLD : context = 0x00600000; break; - default: - WARN_ON(1); - return -EINVAL; - } - - return nvkm_ramht_insert(chan->ramht, object, 0, 4, handle, context); -} - -static void -g84_fifo_chan_init(struct nvkm_fifo_chan *base) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - struct nv50_fifo *fifo = chan->fifo; - struct nvkm_device *device = fifo->base.engine.subdev.device; - u64 addr = chan->ramfc->addr >> 8; - u32 chid = chan->base.chid; - - nvkm_wr32(device, 0x002600 + (chid * 4), 0x80000000 | addr); - nv50_fifo_runlist_update(fifo); -} - -static const struct nvkm_fifo_chan_func -g84_fifo_chan_func = { - .dtor = nv50_fifo_chan_dtor, - .init = g84_fifo_chan_init, - .fini = nv50_fifo_chan_fini, - .ntfy = g84_fifo_chan_ntfy, - .engine_ctor = g84_fifo_chan_engine_ctor, - .engine_dtor = nv50_fifo_chan_engine_dtor, - .engine_init = g84_fifo_chan_engine_init, - .engine_fini = g84_fifo_chan_engine_fini, - .object_ctor = g84_fifo_chan_object_ctor, - .object_dtor = nv50_fifo_chan_object_dtor, -}; - -int -g84_fifo_chan_ctor(struct nv50_fifo *fifo, u64 vmm, u64 push, - const struct nvkm_oclass *oclass, - struct nv50_fifo_chan *chan) -{ - struct nvkm_device *device = fifo->base.engine.subdev.device; - int ret; - - if (!vmm) - return -EINVAL; - - ret = nvkm_fifo_chan_ctor(&g84_fifo_chan_func, &fifo->base, - 0x10000, 0x1000, false, vmm, push, - BIT(G84_FIFO_ENGN_SW) | - BIT(G84_FIFO_ENGN_GR) | - BIT(G84_FIFO_ENGN_MPEG) | - BIT(G84_FIFO_ENGN_MSPPP) | - BIT(G84_FIFO_ENGN_ME) | - BIT(G84_FIFO_ENGN_CE0) | - BIT(G84_FIFO_ENGN_VP) | - BIT(G84_FIFO_ENGN_MSPDEC) | - BIT(G84_FIFO_ENGN_CIPHER) | - BIT(G84_FIFO_ENGN_SEC) | - BIT(G84_FIFO_ENGN_VIC) | - BIT(G84_FIFO_ENGN_BSP) | - BIT(G84_FIFO_ENGN_MSVLD) | - BIT(G84_FIFO_ENGN_DMA), - 0, 0xc00000, 0x2000, oclass, &chan->base); - chan->fifo = fifo; - if (ret) - return ret; - - ret = nvkm_gpuobj_new(device, 0x0200, 0, true, chan->base.inst, - &chan->eng); - if (ret) - return ret; - - ret = nvkm_gpuobj_new(device, 0x4000, 0, false, chan->base.inst, - &chan->pgd); - if (ret) - return ret; - - ret = nvkm_gpuobj_new(device, 0x1000, 0x400, true, chan->base.inst, - &chan->cache); - if (ret) - return ret; - - ret = nvkm_gpuobj_new(device, 0x100, 0x100, true, chan->base.inst, - &chan->ramfc); - if (ret) - return ret; - - return nvkm_ramht_new(device, 0x8000, 16, chan->base.inst, &chan->ramht); -} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/changf100.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/changf100.h deleted file mode 100644 index f7ac1061fa84..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/changf100.h +++ /dev/null @@ -1,29 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __GF100_FIFO_CHAN_H__ -#define __GF100_FIFO_CHAN_H__ -#define gf100_fifo_chan(p) container_of((p), struct gf100_fifo_chan, base) -#include "chan.h" -#include "gf100.h" - -struct gf100_fifo_chan { - struct nvkm_fifo_chan base; - struct gf100_fifo *fifo; - - struct list_head head; - bool killed; - -#define GF100_FIFO_ENGN_GR 0 -#define GF100_FIFO_ENGN_MSPDEC 1 -#define GF100_FIFO_ENGN_MSPPP 2 -#define GF100_FIFO_ENGN_MSVLD 3 -#define GF100_FIFO_ENGN_CE0 4 -#define GF100_FIFO_ENGN_CE1 5 -#define GF100_FIFO_ENGN_SW 15 - struct gf100_fifo_engn { - struct nvkm_gpuobj *inst; - struct nvkm_vma *vma; - } engn[NVKM_FIFO_ENGN_NR]; -}; - -extern const struct nvkm_fifo_chan_oclass gf100_fifo_gpfifo_oclass; -#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/changk104.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/changk104.h deleted file mode 100644 index 9713daee6c76..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/changk104.h +++ /dev/null @@ -1,52 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __GK104_FIFO_CHAN_H__ -#define __GK104_FIFO_CHAN_H__ -#define gk104_fifo_chan(p) container_of((p), struct gk104_fifo_chan, base) -#include "chan.h" -#include "gk104.h" - -struct gk104_fifo_chan { - struct nvkm_fifo_chan base; - struct gk104_fifo *fifo; - int runl; - - struct nvkm_fifo_cgrp *cgrp; - struct list_head head; - bool killed; - -#define GK104_FIFO_ENGN_SW 15 - struct gk104_fifo_engn { - struct nvkm_gpuobj *inst; - struct nvkm_vma *vma; - } engn[NVKM_FIFO_ENGN_NR]; -}; - -extern const struct nvkm_fifo_chan_func gk104_fifo_gpfifo_func; - -int gk104_fifo_gpfifo_new(struct gk104_fifo *, const struct nvkm_oclass *, - void *data, u32 size, struct nvkm_object **); -void *gk104_fifo_gpfifo_dtor(struct nvkm_fifo_chan *); -void gk104_fifo_gpfifo_init(struct nvkm_fifo_chan *); -void gk104_fifo_gpfifo_fini(struct nvkm_fifo_chan *); -struct gk104_fifo_engn *gk104_fifo_gpfifo_engine(struct gk104_fifo_chan *, struct nvkm_engine *); -int gk104_fifo_gpfifo_engine_ctor(struct nvkm_fifo_chan *, struct nvkm_engine *, - struct nvkm_object *); -void gk104_fifo_gpfifo_engine_dtor(struct nvkm_fifo_chan *, - struct nvkm_engine *); -int gk104_fifo_gpfifo_kick(struct gk104_fifo_chan *); -int gk104_fifo_gpfifo_kick_locked(struct gk104_fifo_chan *); - -int gv100_fifo_gpfifo_new(struct gk104_fifo *, const struct nvkm_oclass *, - void *data, u32 size, struct nvkm_object **); -int gv100_fifo_gpfifo_new_(const struct nvkm_fifo_chan_func *, - struct gk104_fifo *, u64 *, u16 *, u64, u64, u64, - u64 *, bool, u32 *, const struct nvkm_oclass *, - struct nvkm_object **); -int gv100_fifo_gpfifo_engine_init(struct nvkm_fifo_chan *, - struct nvkm_engine *); -int gv100_fifo_gpfifo_engine_fini(struct nvkm_fifo_chan *, - struct nvkm_engine *, bool); - -int tu102_fifo_gpfifo_new(struct gk104_fifo *, const struct nvkm_oclass *, - void *data, u32 size, struct nvkm_object **); -#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/channv04.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/channv04.h deleted file mode 100644 index 727bc8976b40..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/channv04.h +++ /dev/null @@ -1,29 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NV04_FIFO_CHAN_H__ -#define __NV04_FIFO_CHAN_H__ -#define nv04_fifo_chan(p) container_of((p), struct nv04_fifo_chan, base) -#include "chan.h" -#include "nv04.h" - -struct nv04_fifo_chan { - struct nvkm_fifo_chan base; - struct nv04_fifo *fifo; - u32 ramfc; -#define NV04_FIFO_ENGN_SW 0 -#define NV04_FIFO_ENGN_GR 1 -#define NV04_FIFO_ENGN_MPEG 2 -#define NV04_FIFO_ENGN_DMA 3 - struct nvkm_gpuobj *engn[NVKM_FIFO_ENGN_NR]; -}; - -extern const struct nvkm_fifo_chan_func nv04_fifo_dma_func; -void *nv04_fifo_dma_dtor(struct nvkm_fifo_chan *); -void nv04_fifo_dma_init(struct nvkm_fifo_chan *); -void nv04_fifo_dma_fini(struct nvkm_fifo_chan *); -void nv04_fifo_dma_object_dtor(struct nvkm_fifo_chan *, int); - -extern const struct nvkm_fifo_chan_oclass nv04_fifo_dma_oclass; -extern const struct nvkm_fifo_chan_oclass nv10_fifo_dma_oclass; -extern const struct nvkm_fifo_chan_oclass nv17_fifo_dma_oclass; -extern const struct nvkm_fifo_chan_oclass nv40_fifo_dma_oclass; -#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/channv50.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/channv50.c deleted file mode 100644 index c44d7c81dd52..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/channv50.c +++ /dev/null @@ -1,276 +0,0 @@ -/* - * Copyright 2012 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#include "channv50.h" - -#include <core/client.h> -#include <core/ramht.h> -#include <subdev/mmu.h> -#include <subdev/timer.h> - -static int -nv50_fifo_chan_engine_addr(struct nvkm_engine *engine) -{ - switch (engine->subdev.type) { - case NVKM_ENGINE_DMAOBJ: - case NVKM_ENGINE_SW : return -1; - case NVKM_ENGINE_GR : return 0x0000; - case NVKM_ENGINE_MPEG : return 0x0060; - default: - WARN_ON(1); - return -1; - } -} - -struct nvkm_gpuobj ** -nv50_fifo_chan_engine(struct nv50_fifo_chan *chan, struct nvkm_engine *engine) -{ - int engi = chan->base.fifo->func->engine_id(chan->base.fifo, engine); - if (engi >= 0) - return &chan->engn[engi]; - return NULL; -} - -static int -nv50_fifo_chan_engine_fini(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine, bool suspend) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - struct nv50_fifo *fifo = chan->fifo; - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - int offset, ret = 0; - u32 me; - - offset = nv50_fifo_chan_engine_addr(engine); - if (offset < 0) - return 0; - - /* HW bug workaround: - * - * PFIFO will hang forever if the connected engines don't report - * that they've processed the context switch request. - * - * In order for the kickoff to work, we need to ensure all the - * connected engines are in a state where they can answer. - * - * Newer chipsets don't seem to suffer from this issue, and well, - * there's also a "ignore these engines" bitmask reg we can use - * if we hit the issue there.. - */ - me = nvkm_mask(device, 0x00b860, 0x00000001, 0x00000001); - - /* do the kickoff... */ - nvkm_wr32(device, 0x0032fc, chan->base.inst->addr >> 12); - if (nvkm_msec(device, 2000, - if (nvkm_rd32(device, 0x0032fc) != 0xffffffff) - break; - ) < 0) { - nvkm_error(subdev, "channel %d [%s] unload timeout\n", - chan->base.chid, chan->base.object.client->name); - if (suspend) - ret = -EBUSY; - } - nvkm_wr32(device, 0x00b860, me); - - if (ret == 0) { - nvkm_kmap(chan->eng); - nvkm_wo32(chan->eng, offset + 0x00, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x04, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x08, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x0c, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x10, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x14, 0x00000000); - nvkm_done(chan->eng); - } - - return ret; -} - -static int -nv50_fifo_chan_engine_init(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - struct nvkm_gpuobj *engn = *nv50_fifo_chan_engine(chan, engine); - u64 limit, start; - int offset; - - offset = nv50_fifo_chan_engine_addr(engine); - if (offset < 0) - return 0; - limit = engn->addr + engn->size - 1; - start = engn->addr; - - nvkm_kmap(chan->eng); - nvkm_wo32(chan->eng, offset + 0x00, 0x00190000); - nvkm_wo32(chan->eng, offset + 0x04, lower_32_bits(limit)); - nvkm_wo32(chan->eng, offset + 0x08, lower_32_bits(start)); - nvkm_wo32(chan->eng, offset + 0x0c, upper_32_bits(limit) << 24 | - upper_32_bits(start)); - nvkm_wo32(chan->eng, offset + 0x10, 0x00000000); - nvkm_wo32(chan->eng, offset + 0x14, 0x00000000); - nvkm_done(chan->eng); - return 0; -} - -void -nv50_fifo_chan_engine_dtor(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - nvkm_gpuobj_del(nv50_fifo_chan_engine(chan, engine)); -} - -static int -nv50_fifo_chan_engine_ctor(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine, - struct nvkm_object *object) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - - if (nv50_fifo_chan_engine_addr(engine) < 0) - return 0; - - return nvkm_object_bind(object, NULL, 0, nv50_fifo_chan_engine(chan, engine)); -} - -void -nv50_fifo_chan_object_dtor(struct nvkm_fifo_chan *base, int cookie) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - nvkm_ramht_remove(chan->ramht, cookie); -} - -static int -nv50_fifo_chan_object_ctor(struct nvkm_fifo_chan *base, - struct nvkm_object *object) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - u32 handle = object->handle; - u32 context; - - switch (object->engine->subdev.type) { - case NVKM_ENGINE_DMAOBJ: - case NVKM_ENGINE_SW : context = 0x00000000; break; - case NVKM_ENGINE_GR : context = 0x00100000; break; - case NVKM_ENGINE_MPEG : context = 0x00200000; break; - default: - WARN_ON(1); - return -EINVAL; - } - - return nvkm_ramht_insert(chan->ramht, object, 0, 4, handle, context); -} - -void -nv50_fifo_chan_fini(struct nvkm_fifo_chan *base) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - struct nv50_fifo *fifo = chan->fifo; - struct nvkm_device *device = fifo->base.engine.subdev.device; - u32 chid = chan->base.chid; - - /* remove channel from runlist, fifo will unload context */ - nvkm_mask(device, 0x002600 + (chid * 4), 0x80000000, 0x00000000); - nv50_fifo_runlist_update(fifo); - nvkm_wr32(device, 0x002600 + (chid * 4), 0x00000000); -} - -static void -nv50_fifo_chan_init(struct nvkm_fifo_chan *base) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - struct nv50_fifo *fifo = chan->fifo; - struct nvkm_device *device = fifo->base.engine.subdev.device; - u64 addr = chan->ramfc->addr >> 12; - u32 chid = chan->base.chid; - - nvkm_wr32(device, 0x002600 + (chid * 4), 0x80000000 | addr); - nv50_fifo_runlist_update(fifo); -} - -void * -nv50_fifo_chan_dtor(struct nvkm_fifo_chan *base) -{ - struct nv50_fifo_chan *chan = nv50_fifo_chan(base); - nvkm_ramht_del(&chan->ramht); - nvkm_gpuobj_del(&chan->pgd); - nvkm_gpuobj_del(&chan->eng); - nvkm_gpuobj_del(&chan->cache); - nvkm_gpuobj_del(&chan->ramfc); - return chan; -} - -static const struct nvkm_fifo_chan_func -nv50_fifo_chan_func = { - .dtor = nv50_fifo_chan_dtor, - .init = nv50_fifo_chan_init, - .fini = nv50_fifo_chan_fini, - .engine_ctor = nv50_fifo_chan_engine_ctor, - .engine_dtor = nv50_fifo_chan_engine_dtor, - .engine_init = nv50_fifo_chan_engine_init, - .engine_fini = nv50_fifo_chan_engine_fini, - .object_ctor = nv50_fifo_chan_object_ctor, - .object_dtor = nv50_fifo_chan_object_dtor, -}; - -int -nv50_fifo_chan_ctor(struct nv50_fifo *fifo, u64 vmm, u64 push, - const struct nvkm_oclass *oclass, - struct nv50_fifo_chan *chan) -{ - struct nvkm_device *device = fifo->base.engine.subdev.device; - int ret; - - if (!vmm) - return -EINVAL; - - ret = nvkm_fifo_chan_ctor(&nv50_fifo_chan_func, &fifo->base, - 0x10000, 0x1000, false, vmm, push, - BIT(NV50_FIFO_ENGN_SW) | - BIT(NV50_FIFO_ENGN_GR) | - BIT(NV50_FIFO_ENGN_MPEG) | - BIT(NV50_FIFO_ENGN_DMA), - 0, 0xc00000, 0x2000, oclass, &chan->base); - chan->fifo = fifo; - if (ret) - return ret; - - ret = nvkm_gpuobj_new(device, 0x0200, 0x1000, true, chan->base.inst, - &chan->ramfc); - if (ret) - return ret; - - ret = nvkm_gpuobj_new(device, 0x1200, 0, true, chan->base.inst, - &chan->eng); - if (ret) - return ret; - - ret = nvkm_gpuobj_new(device, 0x4000, 0, false, chan->base.inst, - &chan->pgd); - if (ret) - return ret; - - return nvkm_ramht_new(device, 0x8000, 16, chan->base.inst, &chan->ramht); -} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/channv50.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/channv50.h deleted file mode 100644 index 3a95730d7ff5..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/channv50.h +++ /dev/null @@ -1,53 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NV50_FIFO_CHAN_H__ -#define __NV50_FIFO_CHAN_H__ -#define nv50_fifo_chan(p) container_of((p), struct nv50_fifo_chan, base) -#include "chan.h" -#include "nv50.h" - -struct nv50_fifo_chan { - struct nv50_fifo *fifo; - struct nvkm_fifo_chan base; - - struct nvkm_gpuobj *ramfc; - struct nvkm_gpuobj *cache; - struct nvkm_gpuobj *eng; - struct nvkm_gpuobj *pgd; - struct nvkm_ramht *ramht; - -#define NV50_FIFO_ENGN_SW 0 -#define NV50_FIFO_ENGN_GR 1 -#define NV50_FIFO_ENGN_MPEG 2 -#define NV50_FIFO_ENGN_DMA 3 - -#define G84_FIFO_ENGN_SW 0 -#define G84_FIFO_ENGN_GR 1 -#define G84_FIFO_ENGN_MPEG 2 -#define G84_FIFO_ENGN_MSPPP 2 -#define G84_FIFO_ENGN_ME 3 -#define G84_FIFO_ENGN_CE0 3 -#define G84_FIFO_ENGN_VP 4 -#define G84_FIFO_ENGN_MSPDEC 4 -#define G84_FIFO_ENGN_CIPHER 5 -#define G84_FIFO_ENGN_SEC 5 -#define G84_FIFO_ENGN_VIC 5 -#define G84_FIFO_ENGN_BSP 6 -#define G84_FIFO_ENGN_MSVLD 6 -#define G84_FIFO_ENGN_DMA 7 - struct nvkm_gpuobj *engn[NVKM_FIFO_ENGN_NR]; -}; - -int nv50_fifo_chan_ctor(struct nv50_fifo *, u64 vmm, u64 push, - const struct nvkm_oclass *, struct nv50_fifo_chan *); -void *nv50_fifo_chan_dtor(struct nvkm_fifo_chan *); -void nv50_fifo_chan_fini(struct nvkm_fifo_chan *); -struct nvkm_gpuobj **nv50_fifo_chan_engine(struct nv50_fifo_chan *, struct nvkm_engine *); -void nv50_fifo_chan_engine_dtor(struct nvkm_fifo_chan *, struct nvkm_engine *); -void nv50_fifo_chan_object_dtor(struct nvkm_fifo_chan *, int); - -int g84_fifo_chan_ctor(struct nv50_fifo *, u64 vmm, u64 push, - const struct nvkm_oclass *, struct nv50_fifo_chan *); - -extern const struct nvkm_fifo_chan_oclass nv50_fifo_gpfifo_oclass; -extern const struct nvkm_fifo_chan_oclass g84_fifo_gpfifo_oclass; -#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chid.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chid.c new file mode 100644 index 000000000000..23944d95efd5 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chid.c @@ -0,0 +1,111 @@ +/* + * Copyright 2020 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "chid.h" + +void +nvkm_chid_put(struct nvkm_chid *chid, int id, spinlock_t *data_lock) +{ + if (id >= 0) { + spin_lock_irq(&chid->lock); + spin_lock(data_lock); + chid->data[id] = NULL; + spin_unlock(data_lock); + clear_bit(id, chid->used); + spin_unlock_irq(&chid->lock); + } +} + +int +nvkm_chid_get(struct nvkm_chid *chid, void *data) +{ + int id = -1, cid; + + spin_lock_irq(&chid->lock); + cid = find_first_zero_bit(chid->used, chid->nr); + if (cid < chid->nr) { + set_bit(cid, chid->used); + chid->data[cid] = data; + id = cid; + } + spin_unlock_irq(&chid->lock); + return id; +} + +static void +nvkm_chid_del(struct kref *kref) +{ + struct nvkm_chid *chid = container_of(kref, typeof(*chid), kref); + + nvkm_event_fini(&chid->event); + + kvfree(chid->data); + kfree(chid); +} + +void +nvkm_chid_unref(struct nvkm_chid **pchid) +{ + struct nvkm_chid *chid = *pchid; + + if (!chid) + return; + + kref_put(&chid->kref, nvkm_chid_del); + *pchid = NULL; +} + +struct nvkm_chid * +nvkm_chid_ref(struct nvkm_chid *chid) +{ + if (chid) + kref_get(&chid->kref); + + return chid; +} + +int +nvkm_chid_new(const struct nvkm_event_func *func, struct nvkm_subdev *subdev, + int nr, int first, int count, struct nvkm_chid **pchid) +{ + struct nvkm_chid *chid; + int id; + + if (!(chid = *pchid = kzalloc(struct_size(chid, used, nr), GFP_KERNEL))) + return -ENOMEM; + + kref_init(&chid->kref); + chid->nr = nr; + chid->mask = chid->nr - 1; + spin_lock_init(&chid->lock); + + if (!(chid->data = kvzalloc(sizeof(*chid->data) * nr, GFP_KERNEL))) { + nvkm_chid_unref(pchid); + return -ENOMEM; + } + + for (id = 0; id < first; id++) + __set_bit(id, chid->used); + for (id = first + count; id < nr; id++) + __set_bit(id, chid->used); + + return nvkm_event_init(func, subdev, 1, nr, &chid->event); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chid.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chid.h new file mode 100644 index 000000000000..2a42efb18401 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/chid.h @@ -0,0 +1,25 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __NVKM_CHID_H__ +#define __NVKM_CHID_H__ +#include <core/event.h> + +struct nvkm_chid { + struct kref kref; + int nr; + u32 mask; + + struct nvkm_event event; + + void **data; + + spinlock_t lock; + unsigned long used[]; +}; + +int nvkm_chid_new(const struct nvkm_event_func *, struct nvkm_subdev *, + int nr, int first, int count, struct nvkm_chid **pchid); +struct nvkm_chid *nvkm_chid_ref(struct nvkm_chid *); +void nvkm_chid_unref(struct nvkm_chid **); +int nvkm_chid_get(struct nvkm_chid *, void *data); +void nvkm_chid_put(struct nvkm_chid *, int id, spinlock_t *data_lock); +#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv04.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv04.c deleted file mode 100644 index dbcdc5fab990..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv04.c +++ /dev/null @@ -1,226 +0,0 @@ -/* - * Copyright 2012 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#include "channv04.h" -#include "regsnv04.h" - -#include <core/client.h> -#include <core/ramht.h> -#include <subdev/instmem.h> - -#include <nvif/class.h> -#include <nvif/cl006b.h> -#include <nvif/unpack.h> - -void -nv04_fifo_dma_object_dtor(struct nvkm_fifo_chan *base, int cookie) -{ - struct nv04_fifo_chan *chan = nv04_fifo_chan(base); - struct nvkm_instmem *imem = chan->fifo->base.engine.subdev.device->imem; - - mutex_lock(&chan->fifo->base.mutex); - nvkm_ramht_remove(imem->ramht, cookie); - mutex_unlock(&chan->fifo->base.mutex); -} - -static int -nv04_fifo_dma_object_ctor(struct nvkm_fifo_chan *base, - struct nvkm_object *object) -{ - struct nv04_fifo_chan *chan = nv04_fifo_chan(base); - struct nvkm_instmem *imem = chan->fifo->base.engine.subdev.device->imem; - u32 context = 0x80000000 | chan->base.chid << 24; - u32 handle = object->handle; - int hash; - - switch (object->engine->subdev.type) { - case NVKM_ENGINE_DMAOBJ: - case NVKM_ENGINE_SW : context |= 0x00000000; break; - case NVKM_ENGINE_GR : context |= 0x00010000; break; - case NVKM_ENGINE_MPEG : context |= 0x00020000; break; - default: - WARN_ON(1); - return -EINVAL; - } - - mutex_lock(&chan->fifo->base.mutex); - hash = nvkm_ramht_insert(imem->ramht, object, chan->base.chid, 4, - handle, context); - mutex_unlock(&chan->fifo->base.mutex); - return hash; -} - -void -nv04_fifo_dma_fini(struct nvkm_fifo_chan *base) -{ - struct nv04_fifo_chan *chan = nv04_fifo_chan(base); - struct nv04_fifo *fifo = chan->fifo; - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_memory *fctx = device->imem->ramfc; - const struct nv04_fifo_ramfc *c; - unsigned long flags; - u32 mask = fifo->base.nr - 1; - u32 data = chan->ramfc; - u32 chid; - - /* prevent fifo context switches */ - spin_lock_irqsave(&fifo->base.lock, flags); - nvkm_wr32(device, NV03_PFIFO_CACHES, 0); - - /* if this channel is active, replace it with a null context */ - chid = nvkm_rd32(device, NV03_PFIFO_CACHE1_PUSH1) & mask; - if (chid == chan->base.chid) { - nvkm_mask(device, NV04_PFIFO_CACHE1_DMA_PUSH, 0x00000001, 0); - nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH0, 0); - nvkm_mask(device, NV04_PFIFO_CACHE1_PULL0, 0x00000001, 0); - - c = fifo->ramfc; - nvkm_kmap(fctx); - do { - u32 rm = ((1ULL << c->bits) - 1) << c->regs; - u32 cm = ((1ULL << c->bits) - 1) << c->ctxs; - u32 rv = (nvkm_rd32(device, c->regp) & rm) >> c->regs; - u32 cv = (nvkm_ro32(fctx, c->ctxp + data) & ~cm); - nvkm_wo32(fctx, c->ctxp + data, cv | (rv << c->ctxs)); - } while ((++c)->bits); - nvkm_done(fctx); - - c = fifo->ramfc; - do { - nvkm_wr32(device, c->regp, 0x00000000); - } while ((++c)->bits); - - nvkm_wr32(device, NV03_PFIFO_CACHE1_GET, 0); - nvkm_wr32(device, NV03_PFIFO_CACHE1_PUT, 0); - nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH1, mask); - nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH0, 1); - nvkm_wr32(device, NV04_PFIFO_CACHE1_PULL0, 1); - } - - /* restore normal operation, after disabling dma mode */ - nvkm_mask(device, NV04_PFIFO_MODE, 1 << chan->base.chid, 0); - nvkm_wr32(device, NV03_PFIFO_CACHES, 1); - spin_unlock_irqrestore(&fifo->base.lock, flags); -} - -void -nv04_fifo_dma_init(struct nvkm_fifo_chan *base) -{ - struct nv04_fifo_chan *chan = nv04_fifo_chan(base); - struct nv04_fifo *fifo = chan->fifo; - struct nvkm_device *device = fifo->base.engine.subdev.device; - u32 mask = 1 << chan->base.chid; - unsigned long flags; - spin_lock_irqsave(&fifo->base.lock, flags); - nvkm_mask(device, NV04_PFIFO_MODE, mask, mask); - spin_unlock_irqrestore(&fifo->base.lock, flags); -} - -void * -nv04_fifo_dma_dtor(struct nvkm_fifo_chan *base) -{ - struct nv04_fifo_chan *chan = nv04_fifo_chan(base); - struct nv04_fifo *fifo = chan->fifo; - struct nvkm_instmem *imem = fifo->base.engine.subdev.device->imem; - const struct nv04_fifo_ramfc *c = fifo->ramfc; - - nvkm_kmap(imem->ramfc); - do { - nvkm_wo32(imem->ramfc, chan->ramfc + c->ctxp, 0x00000000); - } while ((++c)->bits); - nvkm_done(imem->ramfc); - return chan; -} - -const struct nvkm_fifo_chan_func -nv04_fifo_dma_func = { - .dtor = nv04_fifo_dma_dtor, - .init = nv04_fifo_dma_init, - .fini = nv04_fifo_dma_fini, - .object_ctor = nv04_fifo_dma_object_ctor, - .object_dtor = nv04_fifo_dma_object_dtor, -}; - -static int -nv04_fifo_dma_new(struct nvkm_fifo *base, const struct nvkm_oclass *oclass, - void *data, u32 size, struct nvkm_object **pobject) -{ - struct nvkm_object *parent = oclass->parent; - union { - struct nv03_channel_dma_v0 v0; - } *args = data; - struct nv04_fifo *fifo = nv04_fifo(base); - struct nv04_fifo_chan *chan = NULL; - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_instmem *imem = device->imem; - int ret = -ENOSYS; - - nvif_ioctl(parent, "create channel dma size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(parent, "create channel dma vers %d pushbuf %llx " - "offset %08x\n", args->v0.version, - args->v0.pushbuf, args->v0.offset); - if (!args->v0.pushbuf) - return -EINVAL; - } else - return ret; - - if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) - return -ENOMEM; - *pobject = &chan->base.object; - - ret = nvkm_fifo_chan_ctor(&nv04_fifo_dma_func, &fifo->base, - 0x1000, 0x1000, false, 0, args->v0.pushbuf, - BIT(NV04_FIFO_ENGN_SW) | - BIT(NV04_FIFO_ENGN_GR) | - BIT(NV04_FIFO_ENGN_DMA), - 0, 0x800000, 0x10000, oclass, &chan->base); - chan->fifo = fifo; - if (ret) - return ret; - - args->v0.chid = chan->base.chid; - chan->ramfc = chan->base.chid * 32; - - nvkm_kmap(imem->ramfc); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x00, args->v0.offset); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x04, args->v0.offset); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x08, chan->base.push->addr >> 4); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x10, - NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES | - NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES | -#ifdef __BIG_ENDIAN - NV_PFIFO_CACHE1_BIG_ENDIAN | -#endif - NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8); - nvkm_done(imem->ramfc); - return 0; -} - -const struct nvkm_fifo_chan_oclass -nv04_fifo_dma_oclass = { - .base.oclass = NV03_CHANNEL_DMA, - .base.minver = 0, - .base.maxver = 0, - .ctor = nv04_fifo_dma_new, -}; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv10.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv10.c deleted file mode 100644 index 07d80d54a07c..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv10.c +++ /dev/null @@ -1,97 +0,0 @@ -/* - * Copyright 2012 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#include "channv04.h" -#include "regsnv04.h" - -#include <core/client.h> -#include <core/gpuobj.h> -#include <subdev/instmem.h> - -#include <nvif/class.h> -#include <nvif/cl006b.h> -#include <nvif/unpack.h> - -static int -nv10_fifo_dma_new(struct nvkm_fifo *base, const struct nvkm_oclass *oclass, - void *data, u32 size, struct nvkm_object **pobject) -{ - struct nvkm_object *parent = oclass->parent; - union { - struct nv03_channel_dma_v0 v0; - } *args = data; - struct nv04_fifo *fifo = nv04_fifo(base); - struct nv04_fifo_chan *chan = NULL; - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_instmem *imem = device->imem; - int ret = -ENOSYS; - - nvif_ioctl(parent, "create channel dma size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(parent, "create channel dma vers %d pushbuf %llx " - "offset %08x\n", args->v0.version, - args->v0.pushbuf, args->v0.offset); - if (!args->v0.pushbuf) - return -EINVAL; - } else - return ret; - - if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) - return -ENOMEM; - *pobject = &chan->base.object; - - ret = nvkm_fifo_chan_ctor(&nv04_fifo_dma_func, &fifo->base, - 0x1000, 0x1000, false, 0, args->v0.pushbuf, - BIT(NV04_FIFO_ENGN_SW) | - BIT(NV04_FIFO_ENGN_GR) | - BIT(NV04_FIFO_ENGN_DMA), - 0, 0x800000, 0x10000, oclass, &chan->base); - chan->fifo = fifo; - if (ret) - return ret; - - args->v0.chid = chan->base.chid; - chan->ramfc = chan->base.chid * 32; - - nvkm_kmap(imem->ramfc); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x00, args->v0.offset); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x04, args->v0.offset); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x0c, chan->base.push->addr >> 4); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x14, - NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES | - NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES | -#ifdef __BIG_ENDIAN - NV_PFIFO_CACHE1_BIG_ENDIAN | -#endif - NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8); - nvkm_done(imem->ramfc); - return 0; -} - -const struct nvkm_fifo_chan_oclass -nv10_fifo_dma_oclass = { - .base.oclass = NV10_CHANNEL_DMA, - .base.minver = 0, - .base.maxver = 0, - .ctor = nv10_fifo_dma_new, -}; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv17.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv17.c deleted file mode 100644 index edd70a114218..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv17.c +++ /dev/null @@ -1,98 +0,0 @@ -/* - * Copyright 2012 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#include "channv04.h" -#include "regsnv04.h" - -#include <core/client.h> -#include <core/gpuobj.h> -#include <subdev/instmem.h> - -#include <nvif/class.h> -#include <nvif/cl006b.h> -#include <nvif/unpack.h> - -static int -nv17_fifo_dma_new(struct nvkm_fifo *base, const struct nvkm_oclass *oclass, - void *data, u32 size, struct nvkm_object **pobject) -{ - struct nvkm_object *parent = oclass->parent; - union { - struct nv03_channel_dma_v0 v0; - } *args = data; - struct nv04_fifo *fifo = nv04_fifo(base); - struct nv04_fifo_chan *chan = NULL; - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_instmem *imem = device->imem; - int ret = -ENOSYS; - - nvif_ioctl(parent, "create channel dma size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(parent, "create channel dma vers %d pushbuf %llx " - "offset %08x\n", args->v0.version, - args->v0.pushbuf, args->v0.offset); - if (!args->v0.pushbuf) - return -EINVAL; - } else - return ret; - - if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) - return -ENOMEM; - *pobject = &chan->base.object; - - ret = nvkm_fifo_chan_ctor(&nv04_fifo_dma_func, &fifo->base, - 0x1000, 0x1000, false, 0, args->v0.pushbuf, - BIT(NV04_FIFO_ENGN_SW) | - BIT(NV04_FIFO_ENGN_GR) | - BIT(NV04_FIFO_ENGN_MPEG) | /* NV31- */ - BIT(NV04_FIFO_ENGN_DMA), - 0, 0x800000, 0x10000, oclass, &chan->base); - chan->fifo = fifo; - if (ret) - return ret; - - args->v0.chid = chan->base.chid; - chan->ramfc = chan->base.chid * 64; - - nvkm_kmap(imem->ramfc); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x00, args->v0.offset); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x04, args->v0.offset); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x0c, chan->base.push->addr >> 4); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x14, - NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES | - NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES | -#ifdef __BIG_ENDIAN - NV_PFIFO_CACHE1_BIG_ENDIAN | -#endif - NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8); - nvkm_done(imem->ramfc); - return 0; -} - -const struct nvkm_fifo_chan_oclass -nv17_fifo_dma_oclass = { - .base.oclass = NV17_CHANNEL_DMA, - .base.minver = 0, - .base.maxver = 0, - .ctor = nv17_fifo_dma_new, -}; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv40.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv40.c deleted file mode 100644 index 0411fb908457..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/dmanv40.c +++ /dev/null @@ -1,254 +0,0 @@ -/* - * Copyright 2012 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#include "channv04.h" -#include "regsnv04.h" - -#include <core/client.h> -#include <core/ramht.h> -#include <subdev/instmem.h> - -#include <nvif/class.h> -#include <nvif/cl006b.h> -#include <nvif/unpack.h> - -static bool -nv40_fifo_dma_engine(struct nvkm_engine *engine, u32 *reg, u32 *ctx) -{ - switch (engine->subdev.type) { - case NVKM_ENGINE_DMAOBJ: - case NVKM_ENGINE_SW: - return false; - case NVKM_ENGINE_GR: - *reg = 0x0032e0; - *ctx = 0x38; - return true; - case NVKM_ENGINE_MPEG: - if (engine->subdev.device->chipset < 0x44) - return false; - *reg = 0x00330c; - *ctx = 0x54; - return true; - default: - WARN_ON(1); - return false; - } -} - -static struct nvkm_gpuobj ** -nv40_fifo_dma_engn(struct nv04_fifo_chan *chan, struct nvkm_engine *engine) -{ - int engi = chan->base.fifo->func->engine_id(chan->base.fifo, engine); - if (engi >= 0) - return &chan->engn[engi]; - return NULL; -} - -static int -nv40_fifo_dma_engine_fini(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine, bool suspend) -{ - struct nv04_fifo_chan *chan = nv04_fifo_chan(base); - struct nv04_fifo *fifo = chan->fifo; - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_instmem *imem = device->imem; - unsigned long flags; - u32 reg, ctx; - int chid; - - if (!nv40_fifo_dma_engine(engine, ®, &ctx)) - return 0; - - spin_lock_irqsave(&fifo->base.lock, flags); - nvkm_mask(device, 0x002500, 0x00000001, 0x00000000); - - chid = nvkm_rd32(device, 0x003204) & (fifo->base.nr - 1); - if (chid == chan->base.chid) - nvkm_wr32(device, reg, 0x00000000); - nvkm_kmap(imem->ramfc); - nvkm_wo32(imem->ramfc, chan->ramfc + ctx, 0x00000000); - nvkm_done(imem->ramfc); - - nvkm_mask(device, 0x002500, 0x00000001, 0x00000001); - spin_unlock_irqrestore(&fifo->base.lock, flags); - return 0; -} - -static int -nv40_fifo_dma_engine_init(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine) -{ - struct nv04_fifo_chan *chan = nv04_fifo_chan(base); - struct nv04_fifo *fifo = chan->fifo; - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_instmem *imem = device->imem; - unsigned long flags; - u32 inst, reg, ctx; - int chid; - - if (!nv40_fifo_dma_engine(engine, ®, &ctx)) - return 0; - inst = (*nv40_fifo_dma_engn(chan, engine))->addr >> 4; - - spin_lock_irqsave(&fifo->base.lock, flags); - nvkm_mask(device, 0x002500, 0x00000001, 0x00000000); - - chid = nvkm_rd32(device, 0x003204) & (fifo->base.nr - 1); - if (chid == chan->base.chid) - nvkm_wr32(device, reg, inst); - nvkm_kmap(imem->ramfc); - nvkm_wo32(imem->ramfc, chan->ramfc + ctx, inst); - nvkm_done(imem->ramfc); - - nvkm_mask(device, 0x002500, 0x00000001, 0x00000001); - spin_unlock_irqrestore(&fifo->base.lock, flags); - return 0; -} - -static void -nv40_fifo_dma_engine_dtor(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine) -{ - struct nv04_fifo_chan *chan = nv04_fifo_chan(base); - nvkm_gpuobj_del(nv40_fifo_dma_engn(chan, engine)); -} - -static int -nv40_fifo_dma_engine_ctor(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine, - struct nvkm_object *object) -{ - struct nv04_fifo_chan *chan = nv04_fifo_chan(base); - u32 reg, ctx; - - if (!nv40_fifo_dma_engine(engine, ®, &ctx)) - return 0; - - return nvkm_object_bind(object, NULL, 0, nv40_fifo_dma_engn(chan, engine)); -} - -static int -nv40_fifo_dma_object_ctor(struct nvkm_fifo_chan *base, - struct nvkm_object *object) -{ - struct nv04_fifo_chan *chan = nv04_fifo_chan(base); - struct nvkm_instmem *imem = chan->fifo->base.engine.subdev.device->imem; - u32 context = chan->base.chid << 23; - u32 handle = object->handle; - int hash; - - switch (object->engine->subdev.type) { - case NVKM_ENGINE_DMAOBJ: - case NVKM_ENGINE_SW : context |= 0x00000000; break; - case NVKM_ENGINE_GR : context |= 0x00100000; break; - case NVKM_ENGINE_MPEG : context |= 0x00200000; break; - default: - WARN_ON(1); - return -EINVAL; - } - - mutex_lock(&chan->fifo->base.mutex); - hash = nvkm_ramht_insert(imem->ramht, object, chan->base.chid, 4, - handle, context); - mutex_unlock(&chan->fifo->base.mutex); - return hash; -} - -static const struct nvkm_fifo_chan_func -nv40_fifo_dma_func = { - .dtor = nv04_fifo_dma_dtor, - .init = nv04_fifo_dma_init, - .fini = nv04_fifo_dma_fini, - .engine_ctor = nv40_fifo_dma_engine_ctor, - .engine_dtor = nv40_fifo_dma_engine_dtor, - .engine_init = nv40_fifo_dma_engine_init, - .engine_fini = nv40_fifo_dma_engine_fini, - .object_ctor = nv40_fifo_dma_object_ctor, - .object_dtor = nv04_fifo_dma_object_dtor, -}; - -static int -nv40_fifo_dma_new(struct nvkm_fifo *base, const struct nvkm_oclass *oclass, - void *data, u32 size, struct nvkm_object **pobject) -{ - struct nvkm_object *parent = oclass->parent; - union { - struct nv03_channel_dma_v0 v0; - } *args = data; - struct nv04_fifo *fifo = nv04_fifo(base); - struct nv04_fifo_chan *chan = NULL; - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_instmem *imem = device->imem; - int ret = -ENOSYS; - - nvif_ioctl(parent, "create channel dma size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(parent, "create channel dma vers %d pushbuf %llx " - "offset %08x\n", args->v0.version, - args->v0.pushbuf, args->v0.offset); - if (!args->v0.pushbuf) - return -EINVAL; - } else - return ret; - - if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) - return -ENOMEM; - *pobject = &chan->base.object; - - ret = nvkm_fifo_chan_ctor(&nv40_fifo_dma_func, &fifo->base, - 0x1000, 0x1000, false, 0, args->v0.pushbuf, - BIT(NV04_FIFO_ENGN_SW) | - BIT(NV04_FIFO_ENGN_GR) | - BIT(NV04_FIFO_ENGN_MPEG) | - BIT(NV04_FIFO_ENGN_DMA), - 0, 0xc00000, 0x1000, oclass, &chan->base); - chan->fifo = fifo; - if (ret) - return ret; - - args->v0.chid = chan->base.chid; - chan->ramfc = chan->base.chid * 128; - - nvkm_kmap(imem->ramfc); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x00, args->v0.offset); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x04, args->v0.offset); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x0c, chan->base.push->addr >> 4); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x18, 0x30000000 | - NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES | - NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES | -#ifdef __BIG_ENDIAN - NV_PFIFO_CACHE1_BIG_ENDIAN | -#endif - NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8); - nvkm_wo32(imem->ramfc, chan->ramfc + 0x3c, 0x0001ffff); - nvkm_done(imem->ramfc); - return 0; -} - -const struct nvkm_fifo_chan_oclass -nv40_fifo_dma_oclass = { - .base.oclass = NV40_CHANNEL_DMA, - .base.minver = 0, - .base.maxver = 0, - .ctor = nv40_fifo_dma_new, -}; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/g84.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/g84.c index 3885c3830b94..6b229a3fbd97 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/g84.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/g84.c @@ -21,112 +21,211 @@ * * Authors: Ben Skeggs */ -#include "nv50.h" -#include "channv50.h" +#include "priv.h" +#include "cgrp.h" +#include "chan.h" +#include "runl.h" + +#include <core/ramht.h> +#include <subdev/timer.h> + +#include <nvif/class.h> static void -g84_fifo_uevent_fini(struct nvkm_fifo *fifo) +g84_chan_bind(struct nvkm_chan *chan) { - struct nvkm_device *device = fifo->engine.subdev.device; - nvkm_mask(device, 0x002140, 0x40000000, 0x00000000); + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; + + nvkm_wr32(device, 0x002600 + (chan->id * 4), chan->ramfc->addr >> 8); } -static void -g84_fifo_uevent_init(struct nvkm_fifo *fifo) +static int +g84_chan_ramfc_write(struct nvkm_chan *chan, u64 offset, u64 length, u32 devm, bool priv) { - struct nvkm_device *device = fifo->engine.subdev.device; - nvkm_mask(device, 0x002140, 0x40000000, 0x40000000); + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; + const u32 limit2 = ilog2(length / 8); + int ret; + + ret = nvkm_gpuobj_new(device, 0x0200, 0, true, chan->inst, &chan->eng); + if (ret) + return ret; + + ret = nvkm_gpuobj_new(device, 0x4000, 0, false, chan->inst, &chan->pgd); + if (ret) + return ret; + + ret = nvkm_gpuobj_new(device, 0x1000, 0x400, true, chan->inst, &chan->cache); + if (ret) + return ret; + + ret = nvkm_gpuobj_new(device, 0x100, 0x100, true, chan->inst, &chan->ramfc); + if (ret) + return ret; + + ret = nvkm_ramht_new(device, 0x8000, 16, chan->inst, &chan->ramht); + if (ret) + return ret; + + nvkm_kmap(chan->ramfc); + nvkm_wo32(chan->ramfc, 0x3c, 0x403f6078); + nvkm_wo32(chan->ramfc, 0x44, 0x01003fff); + nvkm_wo32(chan->ramfc, 0x48, chan->push->node->offset >> 4); + nvkm_wo32(chan->ramfc, 0x50, lower_32_bits(offset)); + nvkm_wo32(chan->ramfc, 0x54, upper_32_bits(offset) | (limit2 << 16)); + nvkm_wo32(chan->ramfc, 0x60, 0x7fffffff); + nvkm_wo32(chan->ramfc, 0x78, 0x00000000); + nvkm_wo32(chan->ramfc, 0x7c, 0x30000000 | devm); + nvkm_wo32(chan->ramfc, 0x80, ((chan->ramht->bits - 9) << 27) | + (4 << 24) /* SEARCH_FULL */ | + (chan->ramht->gpuobj->node->offset >> 4)); + nvkm_wo32(chan->ramfc, 0x88, chan->cache->addr >> 10); + nvkm_wo32(chan->ramfc, 0x98, chan->inst->addr >> 12); + nvkm_done(chan->ramfc); + return 0; } -static struct nvkm_engine * -g84_fifo_id_engine(struct nvkm_fifo *fifo, int engi) +static const struct nvkm_chan_func_ramfc +g84_chan_ramfc = { + .write = g84_chan_ramfc_write, + .ctxdma = true, + .devm = 0xfff, +}; + +const struct nvkm_chan_func +g84_chan = { + .inst = &nv50_chan_inst, + .userd = &nv50_chan_userd, + .ramfc = &g84_chan_ramfc, + .bind = g84_chan_bind, + .unbind = nv50_chan_unbind, + .start = nv50_chan_start, + .stop = nv50_chan_stop, +}; + +static void +g84_ectx_bind(struct nvkm_engn *engn, struct nvkm_cctx *cctx, struct nvkm_chan *chan) { - struct nvkm_device *device = fifo->engine.subdev.device; - struct nvkm_engine *engine; - enum nvkm_subdev_type type; - - switch (engi) { - case G84_FIFO_ENGN_SW : type = NVKM_ENGINE_SW; break; - case G84_FIFO_ENGN_GR : type = NVKM_ENGINE_GR; break; - case G84_FIFO_ENGN_MPEG : - if ((engine = nvkm_device_engine(device, NVKM_ENGINE_MSPPP, 0))) - return engine; - type = NVKM_ENGINE_MPEG; - break; - case G84_FIFO_ENGN_ME : - if ((engine = nvkm_device_engine(device, NVKM_ENGINE_CE, 0))) - return engine; - type = NVKM_ENGINE_ME; - break; - case G84_FIFO_ENGN_VP : - if ((engine = nvkm_device_engine(device, NVKM_ENGINE_MSPDEC, 0))) - return engine; - type = NVKM_ENGINE_VP; - break; - case G84_FIFO_ENGN_CIPHER: - if ((engine = nvkm_device_engine(device, NVKM_ENGINE_VIC, 0))) - return engine; - if ((engine = nvkm_device_engine(device, NVKM_ENGINE_SEC, 0))) - return engine; - type = NVKM_ENGINE_CIPHER; - break; - case G84_FIFO_ENGN_BSP : - if ((engine = nvkm_device_engine(device, NVKM_ENGINE_MSVLD, 0))) - return engine; - type = NVKM_ENGINE_BSP; - break; - case G84_FIFO_ENGN_DMA : type = NVKM_ENGINE_DMAOBJ; break; + struct nvkm_subdev *subdev = &chan->cgrp->runl->fifo->engine.subdev; + struct nvkm_device *device = subdev->device; + u64 start = 0, limit = 0; + u32 flags = 0, ptr0, save; + + switch (engn->engine->subdev.type) { + case NVKM_ENGINE_GR : ptr0 = 0x0020; break; + case NVKM_ENGINE_VP : + case NVKM_ENGINE_MSPDEC: ptr0 = 0x0040; break; + case NVKM_ENGINE_MPEG : + case NVKM_ENGINE_MSPPP : ptr0 = 0x0060; break; + case NVKM_ENGINE_BSP : + case NVKM_ENGINE_MSVLD : ptr0 = 0x0080; break; + case NVKM_ENGINE_CIPHER: + case NVKM_ENGINE_SEC : ptr0 = 0x00a0; break; + case NVKM_ENGINE_CE : ptr0 = 0x00c0; break; default: WARN_ON(1); - return NULL; + return; + } + + if (!cctx) { + save = nvkm_mask(device, 0x002520, 0x0000003f, BIT(engn->id - 1)); + nvkm_wr32(device, 0x0032fc, chan->inst->addr >> 12); + nvkm_msec(device, 2000, + if (nvkm_rd32(device, 0x0032fc) != 0xffffffff) + break; + ); + nvkm_wr32(device, 0x002520, save); + } else { + flags = 0x00190000; + start = cctx->vctx->inst->addr; + limit = start + cctx->vctx->inst->size - 1; } - return nvkm_device_engine(fifo->engine.subdev.device, type, 0); + nvkm_kmap(chan->eng); + nvkm_wo32(chan->eng, ptr0 + 0x00, flags); + nvkm_wo32(chan->eng, ptr0 + 0x04, lower_32_bits(limit)); + nvkm_wo32(chan->eng, ptr0 + 0x08, lower_32_bits(start)); + nvkm_wo32(chan->eng, ptr0 + 0x0c, upper_32_bits(limit) << 24 | + lower_32_bits(start)); + nvkm_wo32(chan->eng, ptr0 + 0x10, 0x00000000); + nvkm_wo32(chan->eng, ptr0 + 0x14, 0x00000000); + nvkm_done(chan->eng); } +const struct nvkm_engn_func +g84_engn = { + .bind = g84_ectx_bind, + .ramht_add = nv50_eobj_ramht_add, + .ramht_del = nv50_eobj_ramht_del, +}; + +static void +g84_fifo_nonstall_block(struct nvkm_event *event, int type, int index) +{ + struct nvkm_fifo *fifo = container_of(event, typeof(*fifo), nonstall.event); + unsigned long flags; + + spin_lock_irqsave(&fifo->lock, flags); + nvkm_mask(fifo->engine.subdev.device, 0x002140, 0x40000000, 0x00000000); + spin_unlock_irqrestore(&fifo->lock, flags); +} + +static void +g84_fifo_nonstall_allow(struct nvkm_event *event, int type, int index) +{ + struct nvkm_fifo *fifo = container_of(event, typeof(*fifo), nonstall.event); + unsigned long flags; + + spin_lock_irqsave(&fifo->lock, flags); + nvkm_mask(fifo->engine.subdev.device, 0x002140, 0x40000000, 0x40000000); + spin_unlock_irqrestore(&fifo->lock, flags); +} + +const struct nvkm_event_func +g84_fifo_nonstall = { + .init = g84_fifo_nonstall_allow, + .fini = g84_fifo_nonstall_block, +}; + static int -g84_fifo_engine_id(struct nvkm_fifo *base, struct nvkm_engine *engine) +g84_fifo_runl_ctor(struct nvkm_fifo *fifo) { - switch (engine->subdev.type) { - case NVKM_ENGINE_SW : return G84_FIFO_ENGN_SW; - case NVKM_ENGINE_GR : return G84_FIFO_ENGN_GR; - case NVKM_ENGINE_MPEG : - case NVKM_ENGINE_MSPPP : return G84_FIFO_ENGN_MPEG; - case NVKM_ENGINE_CE : return G84_FIFO_ENGN_CE0; - case NVKM_ENGINE_VP : - case NVKM_ENGINE_MSPDEC: return G84_FIFO_ENGN_VP; - case NVKM_ENGINE_CIPHER: - case NVKM_ENGINE_SEC : return G84_FIFO_ENGN_CIPHER; - case NVKM_ENGINE_BSP : - case NVKM_ENGINE_MSVLD : return G84_FIFO_ENGN_BSP; - case NVKM_ENGINE_DMAOBJ: return G84_FIFO_ENGN_DMA; - default: - WARN_ON(1); - return -1; - } + struct nvkm_runl *runl; + + runl = nvkm_runl_new(fifo, 0, 0, 0); + if (IS_ERR(runl)) + return PTR_ERR(runl); + + nvkm_runl_add(runl, 0, fifo->func->engn_sw, NVKM_ENGINE_SW, 0); + nvkm_runl_add(runl, 0, fifo->func->engn_sw, NVKM_ENGINE_DMAOBJ, 0); + nvkm_runl_add(runl, 1, fifo->func->engn, NVKM_ENGINE_GR, 0); + nvkm_runl_add(runl, 2, fifo->func->engn, NVKM_ENGINE_MPEG, 0); + nvkm_runl_add(runl, 3, fifo->func->engn, NVKM_ENGINE_ME, 0); + nvkm_runl_add(runl, 4, fifo->func->engn, NVKM_ENGINE_VP, 0); + nvkm_runl_add(runl, 5, fifo->func->engn, NVKM_ENGINE_CIPHER, 0); + nvkm_runl_add(runl, 6, fifo->func->engn, NVKM_ENGINE_BSP, 0); + return 0; } static const struct nvkm_fifo_func g84_fifo = { - .dtor = nv50_fifo_dtor, - .oneinit = nv50_fifo_oneinit, + .chid_nr = nv50_fifo_chid_nr, + .chid_ctor = nv50_fifo_chid_ctor, + .runl_ctor = g84_fifo_runl_ctor, .init = nv50_fifo_init, .intr = nv04_fifo_intr, - .engine_id = g84_fifo_engine_id, - .id_engine = g84_fifo_id_engine, .pause = nv04_fifo_pause, .start = nv04_fifo_start, - .uevent_init = g84_fifo_uevent_init, - .uevent_fini = g84_fifo_uevent_fini, - .chan = { - &g84_fifo_gpfifo_oclass, - NULL - }, + .nonstall = &g84_fifo_nonstall, + .runl = &nv50_runl, + .engn = &g84_engn, + .engn_sw = &nv50_engn_sw, + .cgrp = {{ }, &nv04_cgrp }, + .chan = {{ 0, 0, G82_CHANNEL_GPFIFO }, &g84_chan }, }; int g84_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return nv50_fifo_new_(&g84_fifo, device, type, inst, pfifo); + return nvkm_fifo_new_(&g84_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/g98.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/g98.c new file mode 100644 index 000000000000..c6ca050c38bf --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/g98.c @@ -0,0 +1,70 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" +#include "chan.h" +#include "runl.h" + +#include <nvif/class.h> + +static int +g98_fifo_runl_ctor(struct nvkm_fifo *fifo) +{ + struct nvkm_runl *runl; + + runl = nvkm_runl_new(fifo, 0, 0, 0); + if (IS_ERR(runl)) + return PTR_ERR(runl); + + nvkm_runl_add(runl, 0, fifo->func->engn_sw, NVKM_ENGINE_SW, 0); + nvkm_runl_add(runl, 0, fifo->func->engn_sw, NVKM_ENGINE_DMAOBJ, 0); + nvkm_runl_add(runl, 1, fifo->func->engn, NVKM_ENGINE_GR, 0); + nvkm_runl_add(runl, 2, fifo->func->engn, NVKM_ENGINE_MSPPP, 0); + nvkm_runl_add(runl, 3, fifo->func->engn, NVKM_ENGINE_CE, 0); + nvkm_runl_add(runl, 4, fifo->func->engn, NVKM_ENGINE_MSPDEC, 0); + nvkm_runl_add(runl, 5, fifo->func->engn, NVKM_ENGINE_SEC, 0); + nvkm_runl_add(runl, 6, fifo->func->engn, NVKM_ENGINE_MSVLD, 0); + return 0; +} + +static const struct nvkm_fifo_func +g98_fifo = { + .chid_nr = nv50_fifo_chid_nr, + .chid_ctor = nv50_fifo_chid_ctor, + .runl_ctor = g98_fifo_runl_ctor, + .init = nv50_fifo_init, + .intr = nv04_fifo_intr, + .pause = nv04_fifo_pause, + .start = nv04_fifo_start, + .nonstall = &g84_fifo_nonstall, + .runl = &nv50_runl, + .engn = &g84_engn, + .engn_sw = &nv50_engn_sw, + .cgrp = {{ }, &nv04_cgrp }, + .chan = {{ 0, 0, G82_CHANNEL_GPFIFO }, &g84_chan }, +}; + +int +g98_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, + struct nvkm_fifo **pfifo) +{ + return nvkm_fifo_new_(&g98_fifo, device, type, inst, pfifo); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/ga100.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/ga100.c new file mode 100644 index 000000000000..12a5d99d5e77 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/ga100.c @@ -0,0 +1,550 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" +#include "cgrp.h" +#include "chan.h" +#include "chid.h" +#include "runl.h" +#include "runq.h" + +#include <core/gpuobj.h> +#include <subdev/top.h> +#include <subdev/vfn.h> + +#include <nvif/class.h> + +/*TODO: allocate? */ +#define GA100_FIFO_NONSTALL_VECTOR 0 + +static u32 +ga100_chan_doorbell_handle(struct nvkm_chan *chan) +{ + return (chan->cgrp->runl->doorbell << 16) | chan->id; +} + +static void +ga100_chan_stop(struct nvkm_chan *chan) +{ + struct nvkm_runl *runl = chan->cgrp->runl; + + nvkm_wr32(runl->fifo->engine.subdev.device, runl->chan + (chan->id * 4), 0x00000003); +} + +static void +ga100_chan_start(struct nvkm_chan *chan) +{ + struct nvkm_runl *runl = chan->cgrp->runl; + struct nvkm_device *device = runl->fifo->engine.subdev.device; + const int gfid = 0; + + nvkm_wr32(device, runl->chan + (chan->id * 4), 0x00000002); + nvkm_wr32(device, runl->addr + 0x0090, (gfid << 16) | chan->id); /* INTERNAL_DOORBELL. */ +} + +static void +ga100_chan_unbind(struct nvkm_chan *chan) +{ + struct nvkm_runl *runl = chan->cgrp->runl; + + nvkm_wr32(runl->fifo->engine.subdev.device, runl->chan + (chan->id * 4), 0xffffffff); +} + +static int +ga100_chan_ramfc_write(struct nvkm_chan *chan, u64 offset, u64 length, u32 devm, bool priv) +{ + const u32 limit2 = ilog2(length / 8); + + nvkm_kmap(chan->inst); + nvkm_wo32(chan->inst, 0x010, 0x0000face); + nvkm_wo32(chan->inst, 0x030, 0x7ffff902); + nvkm_wo32(chan->inst, 0x048, lower_32_bits(offset)); + nvkm_wo32(chan->inst, 0x04c, upper_32_bits(offset) | (limit2 << 16)); + nvkm_wo32(chan->inst, 0x084, 0x20400000); + nvkm_wo32(chan->inst, 0x094, 0x30000000 | devm); + nvkm_wo32(chan->inst, 0x0e4, priv ? 0x00000020 : 0x00000000); + nvkm_wo32(chan->inst, 0x0e8, chan->id); + nvkm_wo32(chan->inst, 0x0f4, 0x00001000 | (priv ? 0x00000100 : 0x00000000)); + nvkm_wo32(chan->inst, 0x0f8, 0x80000000 | GA100_FIFO_NONSTALL_VECTOR); + nvkm_mo32(chan->inst, 0x218, 0x00000000, 0x00000000); + nvkm_done(chan->inst); + return 0; +} + +static const struct nvkm_chan_func_ramfc +ga100_chan_ramfc = { + .write = ga100_chan_ramfc_write, + .devm = 0xfff, + .priv = true, +}; + +const struct nvkm_chan_func +ga100_chan = { + .inst = &gf100_chan_inst, + .userd = &gv100_chan_userd, + .ramfc = &ga100_chan_ramfc, + .unbind = ga100_chan_unbind, + .start = ga100_chan_start, + .stop = ga100_chan_stop, + .preempt = gk110_chan_preempt, + .doorbell_handle = ga100_chan_doorbell_handle, +}; + +static void +ga100_cgrp_preempt(struct nvkm_cgrp *cgrp) +{ + struct nvkm_runl *runl = cgrp->runl; + + nvkm_wr32(runl->fifo->engine.subdev.device, runl->addr + 0x098, 0x01000000 | cgrp->id); +} + +const struct nvkm_cgrp_func +ga100_cgrp = { + .preempt = ga100_cgrp_preempt, +}; + +static int +ga100_engn_cxid(struct nvkm_engn *engn, bool *cgid) +{ + struct nvkm_runl *runl = engn->runl; + struct nvkm_device *device = runl->fifo->engine.subdev.device; + u32 stat = nvkm_rd32(device, runl->addr + 0x200 + engn->id * 0x40); + + ENGN_DEBUG(engn, "status %08x", stat); + *cgid = true; + + switch ((stat & 0x0000e000) >> 13) { + case 0 /* INVALID */: return -ENODEV; + case 1 /* VALID */: + case 5 /* SAVE */: return (stat & 0x00000fff); + case 6 /* LOAD */: return (stat & 0x0fff0000) >> 16; + case 7 /* SWITCH */: + if (nvkm_engine_chsw_load(engn->engine)) + return (stat & 0x0fff0000) >> 16; + return (stat & 0x00000fff); + default: + WARN_ON(1); + break; + } + + return -ENODEV; +} + +const struct nvkm_engn_func +ga100_engn = { + .cxid = ga100_engn_cxid, + .ctor = gk104_ectx_ctor, + .bind = gv100_ectx_bind, +}; + +const struct nvkm_engn_func +ga100_engn_ce = { + .cxid = ga100_engn_cxid, + .ctor = gv100_ectx_ce_ctor, + .bind = gv100_ectx_ce_bind, +}; + +static bool +ga100_runq_idle(struct nvkm_runq *runq) +{ + struct nvkm_device *device = runq->fifo->engine.subdev.device; + + return !(nvkm_rd32(device, 0x04015c + (runq->id * 0x800)) & 0x0000e000); +} + +static bool +ga100_runq_intr_1(struct nvkm_runq *runq, struct nvkm_runl *runl) +{ + struct nvkm_device *device = runq->fifo->engine.subdev.device; + u32 inte = nvkm_rd32(device, 0x040180 + (runq->id * 0x800)); + u32 intr = nvkm_rd32(device, 0x040148 + (runq->id * 0x800)); + u32 stat = intr & inte; + + if (!stat) { + RUNQ_DEBUG(runq, "inte1 %08x %08x", intr, inte); + return false; + } + + if (stat & 0x80000000) { + u32 chid = nvkm_rd32(device, 0x040120 + (runq->id * 0x0800)) & runl->chid->mask; + struct nvkm_chan *chan; + unsigned long flags; + + RUNQ_ERROR(runq, "CTXNOTVALID chid:%d", chid); + chan = nvkm_runl_chan_get_chid(runl, chid, &flags); + if (chan) { + nvkm_chan_error(chan, true); + nvkm_chan_put(&chan, flags); + } + + nvkm_mask(device, 0x0400ac + (runq->id * 0x800), 0x00030000, 0x00030000); + stat &= ~0x80000000; + } + + if (stat) { + RUNQ_ERROR(runq, "intr1 %08x", stat); + nvkm_wr32(device, 0x0401a0 + (runq->id * 0x800), stat); + } + + nvkm_wr32(device, 0x040148 + (runq->id * 0x800), intr); + return true; +} + +static bool +ga100_runq_intr_0(struct nvkm_runq *runq, struct nvkm_runl *runl) +{ + struct nvkm_device *device = runq->fifo->engine.subdev.device; + u32 inte = nvkm_rd32(device, 0x040170 + (runq->id * 0x800)); + u32 intr = nvkm_rd32(device, 0x040108 + (runq->id * 0x800)); + u32 stat = intr & inte; + + if (!stat) { + RUNQ_DEBUG(runq, "inte0 %08x %08x", intr, inte); + return false; + } + + /*TODO: expand on this when fixing up gf100's version. */ + if (stat & 0xc6afe000) { + u32 chid = nvkm_rd32(device, 0x040120 + (runq->id * 0x0800)) & runl->chid->mask; + struct nvkm_chan *chan; + unsigned long flags; + + RUNQ_ERROR(runq, "intr0 %08x", stat); + chan = nvkm_runl_chan_get_chid(runl, chid, &flags); + if (chan) { + nvkm_chan_error(chan, true); + nvkm_chan_put(&chan, flags); + } + + stat &= ~0xc6afe000; + } + + if (stat) { + RUNQ_ERROR(runq, "intr0 %08x", stat); + nvkm_wr32(device, 0x040190 + (runq->id * 0x800), stat); + } + + nvkm_wr32(device, 0x040108 + (runq->id * 0x800), intr); + return true; +} + +static bool +ga100_runq_intr(struct nvkm_runq *runq, struct nvkm_runl *runl) +{ + bool intr0 = ga100_runq_intr_0(runq, runl); + bool intr1 = ga100_runq_intr_1(runq, runl); + + return intr0 || intr1; +} + +static void +ga100_runq_init(struct nvkm_runq *runq) +{ + struct nvkm_device *device = runq->fifo->engine.subdev.device; + + nvkm_wr32(device, 0x040108 + (runq->id * 0x800), 0xffffffff); /* INTR_0 */ + nvkm_wr32(device, 0x040148 + (runq->id * 0x800), 0xffffffff); /* INTR_1 */ + nvkm_wr32(device, 0x040170 + (runq->id * 0x800), 0xffffffff); /* INTR_0_EN_SET_TREE */ + nvkm_wr32(device, 0x040180 + (runq->id * 0x800), 0xffffffff); /* INTR_1_EN_SET_TREE */ +} + +const struct nvkm_runq_func +ga100_runq = { + .init = ga100_runq_init, + .intr = ga100_runq_intr, + .idle = ga100_runq_idle, +}; + +static bool +ga100_runl_preempt_pending(struct nvkm_runl *runl) +{ + return nvkm_rd32(runl->fifo->engine.subdev.device, runl->addr + 0x098) & 0x00100000; +} + +static void +ga100_runl_preempt(struct nvkm_runl *runl) +{ + nvkm_wr32(runl->fifo->engine.subdev.device, runl->addr + 0x098, 0x00000000); +} + +static void +ga100_runl_allow(struct nvkm_runl *runl, u32 engm) +{ + nvkm_mask(runl->fifo->engine.subdev.device, runl->addr + 0x094, 0x00000001, 0x00000000); +} + +static void +ga100_runl_block(struct nvkm_runl *runl, u32 engm) +{ + nvkm_mask(runl->fifo->engine.subdev.device, runl->addr + 0x094, 0x00000001, 0x00000001); +} + +static bool +ga100_runl_pending(struct nvkm_runl *runl) +{ + struct nvkm_device *device = runl->fifo->engine.subdev.device; + + return nvkm_rd32(device, runl->addr + 0x08c) & 0x00008000; +} + +static void +ga100_runl_commit(struct nvkm_runl *runl, struct nvkm_memory *memory, u32 start, int count) +{ + struct nvkm_device *device = runl->fifo->engine.subdev.device; + u64 addr = nvkm_memory_addr(memory) + start; + + nvkm_wr32(device, runl->addr + 0x080, lower_32_bits(addr)); + nvkm_wr32(device, runl->addr + 0x084, upper_32_bits(addr)); + nvkm_wr32(device, runl->addr + 0x088, count); +} + +static irqreturn_t +ga100_runl_intr(struct nvkm_inth *inth) +{ + struct nvkm_runl *runl = container_of(inth, typeof(*runl), inth); + struct nvkm_engn *engn; + struct nvkm_device *device = runl->fifo->engine.subdev.device; + u32 inte = nvkm_rd32(device, runl->addr + 0x120); + u32 intr = nvkm_rd32(device, runl->addr + 0x100); + u32 stat = intr & inte; + u32 info; + + if (!stat) { + RUNL_DEBUG(runl, "inte %08x %08x", intr, inte); + return IRQ_NONE; + } + + if (stat & 0x00000007) { + nvkm_runl_foreach_engn_cond(engn, runl, stat & BIT(engn->id)) { + info = nvkm_rd32(device, runl->addr + 0x224 + (engn->id * 0x40)); + + tu102_fifo_intr_ctxsw_timeout_info(engn, info); + + nvkm_wr32(device, runl->addr + 0x100, BIT(engn->id)); + stat &= ~BIT(engn->id); + } + } + + if (stat & 0x00000300) { + nvkm_wr32(device, runl->addr + 0x100, stat & 0x00000300); + stat &= ~0x00000300; + } + + if (stat & 0x00010000) { + if (runl->runq[0]) { + if (runl->runq[0]->func->intr(runl->runq[0], runl)) + stat &= ~0x00010000; + } + } + + if (stat & 0x00020000) { + if (runl->runq[1]) { + if (runl->runq[1]->func->intr(runl->runq[1], runl)) + stat &= ~0x00020000; + } + } + + if (stat) { + RUNL_ERROR(runl, "intr %08x", stat); + nvkm_wr32(device, runl->addr + 0x140, stat); + } + + nvkm_wr32(device, runl->addr + 0x180, 0x00000001); + return IRQ_HANDLED; +} + +static void +ga100_runl_fini(struct nvkm_runl *runl) +{ + nvkm_mask(runl->fifo->engine.subdev.device, runl->addr + 0x300, 0x80000000, 0x00000000); + nvkm_inth_block(&runl->inth); +} + +static void +ga100_runl_init(struct nvkm_runl *runl) +{ + struct nvkm_fifo *fifo = runl->fifo; + struct nvkm_runq *runq; + struct nvkm_device *device = fifo->engine.subdev.device; + int i; + + /* Submit NULL runlist and preempt. */ + nvkm_wr32(device, runl->addr + 0x088, 0x00000000); + runl->func->preempt(runl); + + /* Enable doorbell. */ + nvkm_mask(device, runl->addr + 0x300, 0x80000000, 0x80000000); + + nvkm_wr32(device, runl->addr + 0x100, 0xffffffff); /* INTR_0 */ + nvkm_wr32(device, runl->addr + 0x140, 0xffffffff); /* INTR_0_EN_CLEAR_TREE(0) */ + nvkm_wr32(device, runl->addr + 0x120, 0x000f1307); /* INTR_0_EN_SET_TREE(0) */ + nvkm_wr32(device, runl->addr + 0x148, 0xffffffff); /* INTR_0_EN_CLEAR_TREE(1) */ + nvkm_wr32(device, runl->addr + 0x128, 0x00000000); /* INTR_0_EN_SET_TREE(1) */ + + /* Init PBDMA(s). */ + for (i = 0; i < runl->runq_nr; i++) { + runq = runl->runq[i]; + runq->func->init(runq); + } + + nvkm_inth_allow(&runl->inth); +} + +const struct nvkm_runl_func +ga100_runl = { + .init = ga100_runl_init, + .fini = ga100_runl_fini, + .size = 16, + .update = nv50_runl_update, + .insert_cgrp = gv100_runl_insert_cgrp, + .insert_chan = gv100_runl_insert_chan, + .commit = ga100_runl_commit, + .wait = nv50_runl_wait, + .pending = ga100_runl_pending, + .block = ga100_runl_block, + .allow = ga100_runl_allow, + .preempt = ga100_runl_preempt, + .preempt_pending = ga100_runl_preempt_pending, +}; + +static int +ga100_runl_new(struct nvkm_fifo *fifo, int id, u32 addr, struct nvkm_runl **prunl) +{ + struct nvkm_device *device = fifo->engine.subdev.device; + struct nvkm_runl *runl; + u32 chcfg = nvkm_rd32(device, addr + 0x004); + u32 chnum = 1 << (chcfg & 0x0000000f); + u32 chaddr = (chcfg & 0xfffffff0); + u32 dbcfg = nvkm_rd32(device, addr + 0x008); + u32 vector = nvkm_rd32(device, addr + 0x160); + int i, ret; + + runl = *prunl = nvkm_runl_new(fifo, id, addr, chnum); + if (IS_ERR(runl)) + return PTR_ERR(runl); + + for (i = 0; i < 2; i++) { + u32 pbcfg = nvkm_rd32(device, addr + 0x010 + (i * 0x04)); + if (pbcfg & 0x80000000) { + runl->runq[runl->runq_nr] = + nvkm_runq_new(fifo, ((pbcfg & 0x03fffc00) - 0x040000) / 0x800); + if (!runl->runq[runl->runq_nr]) + return -ENOMEM; + + runl->runq_nr++; + } + } + + ret = nvkm_inth_add(&device->vfn->intr, vector & 0x00000fff, NVKM_INTR_PRIO_NORMAL, + &fifo->engine.subdev, ga100_runl_intr, &runl->inth); + if (ret) + return ret; + + runl->chan = chaddr; + runl->doorbell = dbcfg >> 16; + return 0; +} + +static irqreturn_t +ga100_fifo_nonstall_intr(struct nvkm_inth *inth) +{ + struct nvkm_fifo *fifo = container_of(inth, typeof(*fifo), nonstall.intr); + + nvkm_event_ntfy(&fifo->nonstall.event, 0, NVKM_FIFO_NONSTALL_EVENT); + return IRQ_HANDLED; +} + +static void +ga100_fifo_nonstall_block(struct nvkm_event *event, int type, int index) +{ + struct nvkm_fifo *fifo = container_of(event, typeof(*fifo), nonstall.event); + + nvkm_inth_block(&fifo->nonstall.intr); +} + +static void +ga100_fifo_nonstall_allow(struct nvkm_event *event, int type, int index) +{ + struct nvkm_fifo *fifo = container_of(event, typeof(*fifo), nonstall.event); + + nvkm_inth_allow(&fifo->nonstall.intr); +} + +const struct nvkm_event_func +ga100_fifo_nonstall = { + .init = ga100_fifo_nonstall_allow, + .fini = ga100_fifo_nonstall_block, +}; + +int +ga100_fifo_nonstall_ctor(struct nvkm_fifo *fifo) +{ + return nvkm_inth_add(&fifo->engine.subdev.device->vfn->intr, GA100_FIFO_NONSTALL_VECTOR, + NVKM_INTR_PRIO_NORMAL, &fifo->engine.subdev, ga100_fifo_nonstall_intr, + &fifo->nonstall.intr); +} + +int +ga100_fifo_runl_ctor(struct nvkm_fifo *fifo) +{ + struct nvkm_device *device = fifo->engine.subdev.device; + struct nvkm_top_device *tdev; + struct nvkm_runl *runl; + int id = 0, ret; + + nvkm_list_foreach(tdev, &device->top->device, head, tdev->runlist >= 0) { + runl = nvkm_runl_get(fifo, -1, tdev->runlist); + if (!runl) { + ret = ga100_runl_new(fifo, id++, tdev->runlist, &runl); + if (ret) + return ret; + } + + if (tdev->engine < 0) + continue; + + nvkm_runl_add(runl, tdev->engine, (tdev->type == NVKM_ENGINE_CE) ? + fifo->func->engn_ce : fifo->func->engn, tdev->type, tdev->inst); + } + + return 0; +} + +static const struct nvkm_fifo_func +ga100_fifo = { + .runl_ctor = ga100_fifo_runl_ctor, + .mmu_fault = &tu102_fifo_mmu_fault, + .nonstall_ctor = ga100_fifo_nonstall_ctor, + .nonstall = &ga100_fifo_nonstall, + .runl = &ga100_runl, + .runq = &ga100_runq, + .engn = &ga100_engn, + .engn_ce = &ga100_engn_ce, + .cgrp = {{ 0, 0, KEPLER_CHANNEL_GROUP_A }, &ga100_cgrp, .force = true }, + .chan = {{ 0, 0, AMPERE_CHANNEL_GPFIFO_A }, &ga100_chan }, +}; + +int +ga100_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, + struct nvkm_fifo **pfifo) +{ + return nvkm_fifo_new_(&ga100_fifo, device, type, inst, pfifo); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/ga102.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/ga102.c index c630dbd2911a..2cdf5da339b6 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/ga102.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/ga102.c @@ -19,293 +19,27 @@ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. */ -#define ga102_fifo(p) container_of((p), struct ga102_fifo, base.engine) -#define ga102_chan(p) container_of((p), struct ga102_chan, object) -#include <engine/fifo.h> -#include "user.h" +#include "priv.h" -#include <core/memory.h> -#include <subdev/mmu.h> -#include <subdev/timer.h> -#include <subdev/top.h> - -#include <nvif/cl0080.h> -#include <nvif/clc36f.h> #include <nvif/class.h> -struct ga102_fifo { - struct nvkm_fifo base; -}; - -struct ga102_chan { - struct nvkm_object object; - - struct { - u32 runl; - u32 chan; - } ctrl; - - struct nvkm_memory *mthd; - struct nvkm_memory *inst; - struct nvkm_memory *user; - struct nvkm_memory *runl; - - struct nvkm_vmm *vmm; -}; - -static int -ga102_chan_sclass(struct nvkm_object *object, int index, struct nvkm_oclass *oclass) -{ - if (index == 0) { - oclass->ctor = nvkm_object_new; - oclass->base = (struct nvkm_sclass) { -1, -1, AMPERE_DMA_COPY_B }; - return 0; - } - - return -EINVAL; -} - -static int -ga102_chan_map(struct nvkm_object *object, void *argv, u32 argc, - enum nvkm_object_map *type, u64 *addr, u64 *size) -{ - struct ga102_chan *chan = ga102_chan(object); - struct nvkm_device *device = chan->object.engine->subdev.device; - u64 bar2 = nvkm_memory_bar2(chan->user); - - if (bar2 == ~0ULL) - return -EFAULT; - - *type = NVKM_OBJECT_MAP_IO; - *addr = device->func->resource_addr(device, 3) + bar2; - *size = 0x1000; - return 0; -} - -static int -ga102_chan_fini(struct nvkm_object *object, bool suspend) -{ - struct ga102_chan *chan = ga102_chan(object); - struct nvkm_device *device = chan->object.engine->subdev.device; - - nvkm_wr32(device, chan->ctrl.chan, 0x00000003); - - nvkm_wr32(device, chan->ctrl.runl + 0x098, 0x01000000); - nvkm_msec(device, 2000, - if (!(nvkm_rd32(device, chan->ctrl.runl + 0x098) & 0x00100000)) - break; - ); - - nvkm_wr32(device, chan->ctrl.runl + 0x088, 0); - - nvkm_wr32(device, chan->ctrl.chan, 0xffffffff); - return 0; -} - -static int -ga102_chan_init(struct nvkm_object *object) -{ - struct ga102_chan *chan = ga102_chan(object); - struct nvkm_device *device = chan->object.engine->subdev.device; - - nvkm_mask(device, chan->ctrl.runl + 0x300, 0x80000000, 0x80000000); - - nvkm_wr32(device, chan->ctrl.runl + 0x080, lower_32_bits(nvkm_memory_addr(chan->runl))); - nvkm_wr32(device, chan->ctrl.runl + 0x084, upper_32_bits(nvkm_memory_addr(chan->runl))); - nvkm_wr32(device, chan->ctrl.runl + 0x088, 2); - - nvkm_wr32(device, chan->ctrl.chan, 0x00000002); - nvkm_wr32(device, chan->ctrl.runl + 0x0090, 0); - return 0; -} - -static void * -ga102_chan_dtor(struct nvkm_object *object) -{ - struct ga102_chan *chan = ga102_chan(object); - - if (chan->vmm) { - nvkm_vmm_part(chan->vmm, chan->inst); - nvkm_vmm_unref(&chan->vmm); - } - - nvkm_memory_unref(&chan->runl); - nvkm_memory_unref(&chan->user); - nvkm_memory_unref(&chan->inst); - nvkm_memory_unref(&chan->mthd); - return chan; -} - -static const struct nvkm_object_func -ga102_chan = { - .dtor = ga102_chan_dtor, - .init = ga102_chan_init, - .fini = ga102_chan_fini, - .map = ga102_chan_map, - .sclass = ga102_chan_sclass, -}; - -static int -ga102_chan_new(struct nvkm_device *device, - const struct nvkm_oclass *oclass, void *argv, u32 argc, struct nvkm_object **pobject) -{ - struct volta_channel_gpfifo_a_v0 *args = argv; - struct nvkm_top_device *tdev; - struct nvkm_vmm *vmm; - struct ga102_chan *chan; - int ret; - - if (argc != sizeof(*args)) - return -ENOSYS; - - vmm = nvkm_uvmm_search(oclass->client, args->vmm); - if (IS_ERR(vmm)) - return PTR_ERR(vmm); - - if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) - return -ENOMEM; - - nvkm_object_ctor(&ga102_chan, oclass, &chan->object); - *pobject = &chan->object; - - list_for_each_entry(tdev, &device->top->device, head) { - if (tdev->type == NVKM_ENGINE_CE) { - chan->ctrl.runl = tdev->runlist; - break; - } - } - - if (!chan->ctrl.runl) - return -ENODEV; - - chan->ctrl.chan = nvkm_rd32(device, chan->ctrl.runl + 0x004) & 0xfffffff0; - - args->chid = 0; - args->inst = 0; - args->token = nvkm_rd32(device, chan->ctrl.runl + 0x008) & 0xffff0000; - - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 0x1000, 0x1000, true, &chan->mthd); - if (ret) - return ret; - - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 0x1000, 0x1000, true, &chan->inst); - if (ret) - return ret; - - nvkm_kmap(chan->inst); - nvkm_wo32(chan->inst, 0x010, 0x0000face); - nvkm_wo32(chan->inst, 0x030, 0x7ffff902); - nvkm_wo32(chan->inst, 0x048, lower_32_bits(args->ioffset)); - nvkm_wo32(chan->inst, 0x04c, upper_32_bits(args->ioffset) | - (order_base_2(args->ilength / 8) << 16)); - nvkm_wo32(chan->inst, 0x084, 0x20400000); - nvkm_wo32(chan->inst, 0x094, 0x30000001); - nvkm_wo32(chan->inst, 0x0ac, 0x00020000); - nvkm_wo32(chan->inst, 0x0e4, 0x00000000); - nvkm_wo32(chan->inst, 0x0e8, 0); - nvkm_wo32(chan->inst, 0x0f4, 0x00001000); - nvkm_wo32(chan->inst, 0x0f8, 0x10003080); - nvkm_mo32(chan->inst, 0x218, 0x00000000, 0x00000000); - nvkm_wo32(chan->inst, 0x220, lower_32_bits(nvkm_memory_bar2(chan->mthd))); - nvkm_wo32(chan->inst, 0x224, upper_32_bits(nvkm_memory_bar2(chan->mthd))); - nvkm_done(chan->inst); - - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 0x1000, 0x1000, true, &chan->user); - if (ret) - return ret; - - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 0x1000, 0x1000, true, &chan->runl); - if (ret) - return ret; - - nvkm_kmap(chan->runl); - nvkm_wo32(chan->runl, 0x00, 0x80030001); - nvkm_wo32(chan->runl, 0x04, 1); - nvkm_wo32(chan->runl, 0x08, 0); - nvkm_wo32(chan->runl, 0x0c, 0x00000000); - nvkm_wo32(chan->runl, 0x10, lower_32_bits(nvkm_memory_addr(chan->user))); - nvkm_wo32(chan->runl, 0x14, upper_32_bits(nvkm_memory_addr(chan->user))); - nvkm_wo32(chan->runl, 0x18, lower_32_bits(nvkm_memory_addr(chan->inst))); - nvkm_wo32(chan->runl, 0x1c, upper_32_bits(nvkm_memory_addr(chan->inst))); - nvkm_done(chan->runl); - - ret = nvkm_vmm_join(vmm, chan->inst); - if (ret) - return ret; - - chan->vmm = nvkm_vmm_ref(vmm); - return 0; -} - -static const struct nvkm_device_oclass -ga102_chan_oclass = { - .ctor = ga102_chan_new, -}; - -static int -ga102_user_new(struct nvkm_device *device, - const struct nvkm_oclass *oclass, void *argv, u32 argc, struct nvkm_object **pobject) -{ - return tu102_fifo_user_new(oclass, argv, argc, pobject); -} - -static const struct nvkm_device_oclass -ga102_user_oclass = { - .ctor = ga102_user_new, -}; - -static int -ga102_fifo_sclass(struct nvkm_oclass *oclass, int index, const struct nvkm_device_oclass **class) -{ - if (index == 0) { - oclass->base = (struct nvkm_sclass) { -1, -1, VOLTA_USERMODE_A }; - *class = &ga102_user_oclass; - return 0; - } else - if (index == 1) { - oclass->base = (struct nvkm_sclass) { 0, 0, AMPERE_CHANNEL_GPFIFO_B }; - *class = &ga102_chan_oclass; - return 0; - } - - return 2; -} - -static int -ga102_fifo_info(struct nvkm_engine *engine, u64 mthd, u64 *data) -{ - switch (mthd) { - case NV_DEVICE_HOST_CHANNELS: *data = 1; return 0; - default: - break; - } - - return -ENOSYS; -} - -static void * -ga102_fifo_dtor(struct nvkm_engine *engine) -{ - return ga102_fifo(engine); -} - -static const struct nvkm_engine_func +static const struct nvkm_fifo_func ga102_fifo = { - .dtor = ga102_fifo_dtor, - .info = ga102_fifo_info, - .base.sclass = ga102_fifo_sclass, + .runl_ctor = ga100_fifo_runl_ctor, + .mmu_fault = &tu102_fifo_mmu_fault, + .nonstall_ctor = ga100_fifo_nonstall_ctor, + .nonstall = &ga100_fifo_nonstall, + .runl = &ga100_runl, + .runq = &ga100_runq, + .engn = &ga100_engn, + .engn_ce = &ga100_engn_ce, + .cgrp = {{ 0, 0, KEPLER_CHANNEL_GROUP_A }, &ga100_cgrp, .force = true }, + .chan = {{ 0, 0, AMPERE_CHANNEL_GPFIFO_B }, &ga100_chan }, }; int ga102_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - struct ga102_fifo *fifo; - - if (!(fifo = kzalloc(sizeof(*fifo), GFP_KERNEL))) - return -ENOMEM; - - nvkm_engine_ctor(&ga102_fifo, device, type, inst, true, &fifo->base.engine); - *pfifo = &fifo->base; - return 0; + return nvkm_fifo_new_(&ga102_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gf100.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gf100.c index 8b4f36b3e34b..5bb65258c36d 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gf100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gf100.c @@ -21,186 +21,456 @@ * * Authors: Ben Skeggs */ -#include "gf100.h" -#include "changf100.h" +#include "priv.h" +#include "cgrp.h" +#include "chan.h" +#include "chid.h" +#include "runl.h" +#include "runq.h" -#include <core/client.h> -#include <core/enum.h> #include <core/gpuobj.h> #include <subdev/bar.h> #include <subdev/fault.h> +#include <subdev/mc.h> +#include <subdev/mmu.h> #include <engine/sw.h> #include <nvif/class.h> +void +gf100_chan_preempt(struct nvkm_chan *chan) +{ + nvkm_wr32(chan->cgrp->runl->fifo->engine.subdev.device, 0x002634, chan->id); +} + static void -gf100_fifo_uevent_init(struct nvkm_fifo *fifo) +gf100_chan_stop(struct nvkm_chan *chan) { - struct nvkm_device *device = fifo->engine.subdev.device; - nvkm_mask(device, 0x002140, 0x80000000, 0x80000000); + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; + + nvkm_mask(device, 0x003004 + (chan->id * 8), 0x00000001, 0x00000000); } static void -gf100_fifo_uevent_fini(struct nvkm_fifo *fifo) +gf100_chan_start(struct nvkm_chan *chan) { + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; + + nvkm_wr32(device, 0x003004 + (chan->id * 8), 0x001f0001); +} + +static void gf100_fifo_intr_engine(struct nvkm_fifo *); + +static void +gf100_chan_unbind(struct nvkm_chan *chan) +{ + struct nvkm_fifo *fifo = chan->cgrp->runl->fifo; struct nvkm_device *device = fifo->engine.subdev.device; - nvkm_mask(device, 0x002140, 0x80000000, 0x00000000); + + /*TODO: Is this cargo-culted, or necessary? RM does *something* here... Why? */ + gf100_fifo_intr_engine(fifo); + + nvkm_wr32(device, 0x003000 + (chan->id * 8), 0x00000000); } +static void +gf100_chan_bind(struct nvkm_chan *chan) +{ + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; + + nvkm_wr32(device, 0x003000 + (chan->id * 8), 0xc0000000 | chan->inst->addr >> 12); +} + +static int +gf100_chan_ramfc_write(struct nvkm_chan *chan, u64 offset, u64 length, u32 devm, bool priv) +{ + const u64 userd = nvkm_memory_addr(chan->userd.mem) + chan->userd.base; + const u32 limit2 = ilog2(length / 8); + + nvkm_kmap(chan->inst); + nvkm_wo32(chan->inst, 0x08, lower_32_bits(userd)); + nvkm_wo32(chan->inst, 0x0c, upper_32_bits(userd)); + nvkm_wo32(chan->inst, 0x10, 0x0000face); + nvkm_wo32(chan->inst, 0x30, 0xfffff902); + nvkm_wo32(chan->inst, 0x48, lower_32_bits(offset)); + nvkm_wo32(chan->inst, 0x4c, upper_32_bits(offset) | (limit2 << 16)); + nvkm_wo32(chan->inst, 0x54, 0x00000002); + nvkm_wo32(chan->inst, 0x84, 0x20400000); + nvkm_wo32(chan->inst, 0x94, 0x30000000 | devm); + nvkm_wo32(chan->inst, 0x9c, 0x00000100); + nvkm_wo32(chan->inst, 0xa4, 0x1f1f1f1f); + nvkm_wo32(chan->inst, 0xa8, 0x1f1f1f1f); + nvkm_wo32(chan->inst, 0xac, 0x0000001f); + nvkm_wo32(chan->inst, 0xb8, 0xf8000000); + nvkm_wo32(chan->inst, 0xf8, 0x10003080); /* 0x002310 */ + nvkm_wo32(chan->inst, 0xfc, 0x10000010); /* 0x002350 */ + nvkm_done(chan->inst); + return 0; +} + +static const struct nvkm_chan_func_ramfc +gf100_chan_ramfc = { + .write = gf100_chan_ramfc_write, + .devm = 0xfff, +}; + void -gf100_fifo_runlist_commit(struct gf100_fifo *fifo) +gf100_chan_userd_clear(struct nvkm_chan *chan) { - struct gf100_fifo_chan *chan; - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - struct nvkm_memory *cur; - int nr = 0; - int target; + nvkm_kmap(chan->userd.mem); + nvkm_wo32(chan->userd.mem, chan->userd.base + 0x040, 0x00000000); + nvkm_wo32(chan->userd.mem, chan->userd.base + 0x044, 0x00000000); + nvkm_wo32(chan->userd.mem, chan->userd.base + 0x048, 0x00000000); + nvkm_wo32(chan->userd.mem, chan->userd.base + 0x04c, 0x00000000); + nvkm_wo32(chan->userd.mem, chan->userd.base + 0x050, 0x00000000); + nvkm_wo32(chan->userd.mem, chan->userd.base + 0x058, 0x00000000); + nvkm_wo32(chan->userd.mem, chan->userd.base + 0x05c, 0x00000000); + nvkm_wo32(chan->userd.mem, chan->userd.base + 0x060, 0x00000000); + nvkm_wo32(chan->userd.mem, chan->userd.base + 0x088, 0x00000000); + nvkm_wo32(chan->userd.mem, chan->userd.base + 0x08c, 0x00000000); + nvkm_done(chan->userd.mem); +} - mutex_lock(&fifo->base.mutex); - cur = fifo->runlist.mem[fifo->runlist.active]; - fifo->runlist.active = !fifo->runlist.active; +static const struct nvkm_chan_func_userd +gf100_chan_userd = { + .bar = 1, + .size = 0x1000, + .clear = gf100_chan_userd_clear, +}; - nvkm_kmap(cur); - list_for_each_entry(chan, &fifo->chan, head) { - nvkm_wo32(cur, (nr * 8) + 0, chan->base.chid); - nvkm_wo32(cur, (nr * 8) + 4, 0x00000004); - nr++; - } - nvkm_done(cur); +const struct nvkm_chan_func_inst +gf100_chan_inst = { + .size = 0x1000, + .zero = true, + .vmm = true, +}; - switch (nvkm_memory_target(cur)) { - case NVKM_MEM_TARGET_VRAM: target = 0; break; - case NVKM_MEM_TARGET_NCOH: target = 3; break; +static const struct nvkm_chan_func +gf100_chan = { + .inst = &gf100_chan_inst, + .userd = &gf100_chan_userd, + .ramfc = &gf100_chan_ramfc, + .bind = gf100_chan_bind, + .unbind = gf100_chan_unbind, + .start = gf100_chan_start, + .stop = gf100_chan_stop, + .preempt = gf100_chan_preempt, +}; + +static void +gf100_ectx_bind(struct nvkm_engn *engn, struct nvkm_cctx *cctx, struct nvkm_chan *chan) +{ + u64 addr = 0ULL; + u32 ptr0; + + switch (engn->engine->subdev.type) { + case NVKM_ENGINE_SW : return; + case NVKM_ENGINE_GR : ptr0 = 0x0210; break; + case NVKM_ENGINE_CE : ptr0 = 0x0230 + (engn->engine->subdev.inst * 0x10); break; + case NVKM_ENGINE_MSPDEC: ptr0 = 0x0250; break; + case NVKM_ENGINE_MSPPP : ptr0 = 0x0260; break; + case NVKM_ENGINE_MSVLD : ptr0 = 0x0270; break; default: - mutex_unlock(&fifo->base.mutex); WARN_ON(1); return; } - nvkm_wr32(device, 0x002270, (nvkm_memory_addr(cur) >> 12) | - (target << 28)); - nvkm_wr32(device, 0x002274, 0x01f00000 | nr); + if (cctx) { + addr = cctx->vctx->vma->addr; + addr |= 4ULL; + } - if (wait_event_timeout(fifo->runlist.wait, - !(nvkm_rd32(device, 0x00227c) & 0x00100000), - msecs_to_jiffies(2000)) == 0) - nvkm_error(subdev, "runlist update timeout\n"); - mutex_unlock(&fifo->base.mutex); + nvkm_kmap(chan->inst); + nvkm_wo32(chan->inst, ptr0 + 0, lower_32_bits(addr)); + nvkm_wo32(chan->inst, ptr0 + 4, upper_32_bits(addr)); + nvkm_done(chan->inst); } -void -gf100_fifo_runlist_remove(struct gf100_fifo *fifo, struct gf100_fifo_chan *chan) +static int +gf100_ectx_ctor(struct nvkm_engn *engn, struct nvkm_vctx *vctx) +{ + int ret; + + ret = nvkm_vmm_get(vctx->vmm, 12, vctx->inst->size, &vctx->vma); + if (ret) + return ret; + + return nvkm_memory_map(vctx->inst, 0, vctx->vmm, vctx->vma, NULL, 0); +} + +bool +gf100_engn_mmu_fault_triggered(struct nvkm_engn *engn) { - mutex_lock(&fifo->base.mutex); - list_del_init(&chan->head); - mutex_unlock(&fifo->base.mutex); + struct nvkm_runl *runl = engn->runl; + struct nvkm_fifo *fifo = runl->fifo; + struct nvkm_device *device = fifo->engine.subdev.device; + u32 data = nvkm_rd32(device, 0x002a30 + (engn->id * 4)); + + ENGN_DEBUG(engn, "%08x: mmu fault triggered", data); + if (!(data & 0x00000100)) + return false; + + spin_lock(&fifo->lock); + nvkm_mask(device, 0x002a30 + (engn->id * 4), 0x00000100, 0x00000000); + if (atomic_dec_and_test(&runl->rc_triggered)) + nvkm_mask(device, 0x002140, 0x00000100, 0x00000100); + spin_unlock(&fifo->lock); + return true; } void -gf100_fifo_runlist_insert(struct gf100_fifo *fifo, struct gf100_fifo_chan *chan) +gf100_engn_mmu_fault_trigger(struct nvkm_engn *engn) { - mutex_lock(&fifo->base.mutex); - list_add_tail(&chan->head, &fifo->chan); - mutex_unlock(&fifo->base.mutex); + struct nvkm_runl *runl = engn->runl; + struct nvkm_fifo *fifo = runl->fifo; + struct nvkm_device *device = fifo->engine.subdev.device; + + ENGN_DEBUG(engn, "triggering mmu fault on 0x%02x", engn->fault); + spin_lock(&fifo->lock); + if (atomic_inc_return(&runl->rc_triggered) == 1) + nvkm_mask(device, 0x002140, 0x00000100, 0x00000000); + nvkm_wr32(device, 0x002100, 0x00000100); + nvkm_wr32(device, 0x002a30 + (engn->id * 4), 0x00000100 | engn->fault); + spin_unlock(&fifo->lock); } -static struct nvkm_engine * -gf100_fifo_id_engine(struct nvkm_fifo *fifo, int engi) +/*TODO: clean all this up. */ +struct gf100_engn_status { + bool busy; + bool save; + bool unk0; + bool unk1; + u8 chid; +}; + +static void +gf100_engn_status(struct nvkm_engn *engn, struct gf100_engn_status *status) { - enum nvkm_subdev_type type; - int inst; + u32 stat = nvkm_rd32(engn->engine->subdev.device, 0x002640 + (engn->id * 4)); - switch (engi) { - case GF100_FIFO_ENGN_GR : type = NVKM_ENGINE_GR ; inst = 0; break; - case GF100_FIFO_ENGN_MSPDEC: type = NVKM_ENGINE_MSPDEC; inst = 0; break; - case GF100_FIFO_ENGN_MSPPP : type = NVKM_ENGINE_MSPPP ; inst = 0; break; - case GF100_FIFO_ENGN_MSVLD : type = NVKM_ENGINE_MSVLD ; inst = 0; break; - case GF100_FIFO_ENGN_CE0 : type = NVKM_ENGINE_CE ; inst = 0; break; - case GF100_FIFO_ENGN_CE1 : type = NVKM_ENGINE_CE ; inst = 1; break; - case GF100_FIFO_ENGN_SW : type = NVKM_ENGINE_SW ; inst = 0; break; - default: - WARN_ON(1); - return NULL; - } + status->busy = (stat & 0x10000000); + status->save = (stat & 0x00100000); + status->unk0 = (stat & 0x00004000); + status->unk1 = (stat & 0x00001000); + status->chid = (stat & 0x0000007f); - return nvkm_device_engine(fifo->engine.subdev.device, type, inst); + ENGN_DEBUG(engn, "%08x: busy %d save %d unk0 %d unk1 %d chid %d", + stat, status->busy, status->save, status->unk0, status->unk1, status->chid); } static int -gf100_fifo_engine_id(struct nvkm_fifo *base, struct nvkm_engine *engine) -{ - switch (engine->subdev.type) { - case NVKM_ENGINE_GR : return GF100_FIFO_ENGN_GR; - case NVKM_ENGINE_MSPDEC: return GF100_FIFO_ENGN_MSPDEC; - case NVKM_ENGINE_MSPPP : return GF100_FIFO_ENGN_MSPPP; - case NVKM_ENGINE_MSVLD : return GF100_FIFO_ENGN_MSVLD; - case NVKM_ENGINE_CE : return GF100_FIFO_ENGN_CE0 + engine->subdev.inst; - case NVKM_ENGINE_SW : return GF100_FIFO_ENGN_SW; - default: - WARN_ON(1); - return -1; +gf100_engn_cxid(struct nvkm_engn *engn, bool *cgid) +{ + struct gf100_engn_status status; + + gf100_engn_status(engn, &status); + if (status.busy) { + *cgid = false; + return status.chid; } + + return -ENODEV; } -static void -gf100_fifo_recover_work(struct work_struct *w) +static bool +gf100_engn_chsw(struct nvkm_engn *engn) { - struct gf100_fifo *fifo = container_of(w, typeof(*fifo), recover.work); - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_engine *engine; - unsigned long flags; - u32 engm, engn, todo; + struct gf100_engn_status status; + + gf100_engn_status(engn, &status); + if (status.busy && (status.unk0 || status.unk1)) + return true; - spin_lock_irqsave(&fifo->base.lock, flags); - engm = fifo->recover.mask; - fifo->recover.mask = 0ULL; - spin_unlock_irqrestore(&fifo->base.lock, flags); + return false; +} - nvkm_mask(device, 0x002630, engm, engm); +static const struct nvkm_engn_func +gf100_engn = { + .chsw = gf100_engn_chsw, + .cxid = gf100_engn_cxid, + .mmu_fault_trigger = gf100_engn_mmu_fault_trigger, + .mmu_fault_triggered = gf100_engn_mmu_fault_triggered, + .ctor = gf100_ectx_ctor, + .bind = gf100_ectx_bind, +}; - for (todo = engm; engn = __ffs(todo), todo; todo &= ~BIT_ULL(engn)) { - if ((engine = gf100_fifo_id_engine(&fifo->base, engn))) { - nvkm_subdev_fini(&engine->subdev, false); - WARN_ON(nvkm_subdev_init(&engine->subdev)); +const struct nvkm_engn_func +gf100_engn_sw = { +}; + +static const struct nvkm_bitfield +gf100_runq_intr_0_names[] = { +/* { 0x00008000, "" } seen with null ib push */ + { 0x00200000, "ILLEGAL_MTHD" }, + { 0x00800000, "EMPTY_SUBC" }, + {} +}; + +bool +gf100_runq_intr(struct nvkm_runq *runq, struct nvkm_runl *null) +{ + struct nvkm_subdev *subdev = &runq->fifo->engine.subdev; + struct nvkm_device *device = subdev->device; + u32 mask = nvkm_rd32(device, 0x04010c + (runq->id * 0x2000)); + u32 stat = nvkm_rd32(device, 0x040108 + (runq->id * 0x2000)) & mask; + u32 addr = nvkm_rd32(device, 0x0400c0 + (runq->id * 0x2000)); + u32 data = nvkm_rd32(device, 0x0400c4 + (runq->id * 0x2000)); + u32 chid = nvkm_rd32(device, 0x040120 + (runq->id * 0x2000)) & runq->fifo->chid->mask; + u32 subc = (addr & 0x00070000) >> 16; + u32 mthd = (addr & 0x00003ffc); + u32 show = stat; + struct nvkm_chan *chan; + unsigned long flags; + char msg[128]; + + if (stat & 0x00800000) { + if (device->sw) { + if (nvkm_sw_mthd(device->sw, chid, subc, mthd, data)) + show &= ~0x00800000; } } - gf100_fifo_runlist_commit(fifo); - nvkm_wr32(device, 0x00262c, engm); - nvkm_mask(device, 0x002630, engm, 0x00000000); + if (show) { + nvkm_snprintbf(msg, sizeof(msg), runq->func->intr_0_names, show); + chan = nvkm_chan_get_chid(&runq->fifo->engine, chid, &flags); + nvkm_error(subdev, "PBDMA%d: %08x [%s] ch %d [%010llx %s] " + "subc %d mthd %04x data %08x\n", + runq->id, show, msg, chid, chan ? chan->inst->addr : 0, + chan ? chan->name : "unknown", subc, mthd, data); + + /*TODO: use proper procedure for clearing each exception / debug output */ + if ((stat & 0xc67fe000) && chan) + nvkm_chan_error(chan, true); + nvkm_chan_put(&chan, flags); + } + + nvkm_wr32(device, 0x0400c0 + (runq->id * 0x2000), 0x80600008); + nvkm_wr32(device, 0x040108 + (runq->id * 0x2000), stat); + return true; +} + +void +gf100_runq_init(struct nvkm_runq *runq) +{ + struct nvkm_device *device = runq->fifo->engine.subdev.device; + + nvkm_mask(device, 0x04013c + (runq->id * 0x2000), 0x10000100, 0x00000000); + nvkm_wr32(device, 0x040108 + (runq->id * 0x2000), 0xffffffff); /* INTR */ + nvkm_wr32(device, 0x04010c + (runq->id * 0x2000), 0xfffffeff); /* INTREN */ +} + +static const struct nvkm_runq_func +gf100_runq = { + .init = gf100_runq_init, + .intr = gf100_runq_intr, + .intr_0_names = gf100_runq_intr_0_names, +}; + +bool +gf100_runl_preempt_pending(struct nvkm_runl *runl) +{ + return nvkm_rd32(runl->fifo->engine.subdev.device, 0x002634) & 0x00100000; } static void -gf100_fifo_recover(struct gf100_fifo *fifo, struct nvkm_engine *engine, - struct gf100_fifo_chan *chan) +gf100_runl_fault_clear(struct nvkm_runl *runl) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - u32 chid = chan->base.chid; - int engi = gf100_fifo_engine_id(&fifo->base, engine); + nvkm_mask(runl->fifo->engine.subdev.device, 0x00262c, 0x00000000, 0x00000000); +} + +static void +gf100_runl_allow(struct nvkm_runl *runl, u32 engm) +{ + nvkm_mask(runl->fifo->engine.subdev.device, 0x002630, engm, 0x00000000); +} + +static void +gf100_runl_block(struct nvkm_runl *runl, u32 engm) +{ + nvkm_mask(runl->fifo->engine.subdev.device, 0x002630, engm, engm); +} + +static bool +gf100_runl_pending(struct nvkm_runl *runl) +{ + return nvkm_rd32(runl->fifo->engine.subdev.device, 0x00227c) & 0x00100000; +} + +static void +gf100_runl_commit(struct nvkm_runl *runl, struct nvkm_memory *memory, u32 start, int count) +{ + struct nvkm_device *device = runl->fifo->engine.subdev.device; + u64 addr = nvkm_memory_addr(memory) + start; + int target; + + switch (nvkm_memory_target(memory)) { + case NVKM_MEM_TARGET_VRAM: target = 0; break; + case NVKM_MEM_TARGET_NCOH: target = 3; break; + default: + WARN_ON(1); + return; + } + + nvkm_wr32(device, 0x002270, (target << 28) | (addr >> 12)); + nvkm_wr32(device, 0x002274, 0x01f00000 | count); +} + +static void +gf100_runl_insert_chan(struct nvkm_chan *chan, struct nvkm_memory *memory, u64 offset) +{ + nvkm_wo32(memory, offset + 0, chan->id); + nvkm_wo32(memory, offset + 4, 0x00000004); +} - nvkm_error(subdev, "%s engine fault on channel %d, recovering...\n", - engine->subdev.name, chid); - assert_spin_locked(&fifo->base.lock); +static const struct nvkm_runl_func +gf100_runl = { + .size = 8, + .update = nv50_runl_update, + .insert_chan = gf100_runl_insert_chan, + .commit = gf100_runl_commit, + .wait = nv50_runl_wait, + .pending = gf100_runl_pending, + .block = gf100_runl_block, + .allow = gf100_runl_allow, + .fault_clear = gf100_runl_fault_clear, + .preempt_pending = gf100_runl_preempt_pending, +}; - nvkm_mask(device, 0x003004 + (chid * 0x08), 0x00000001, 0x00000000); - list_del_init(&chan->head); - chan->killed = true; +static void +gf100_fifo_nonstall_allow(struct nvkm_event *event, int type, int index) +{ + struct nvkm_fifo *fifo = container_of(event, typeof(*fifo), nonstall.event); + unsigned long flags; - if (engi >= 0 && engi != GF100_FIFO_ENGN_SW) - fifo->recover.mask |= BIT(engi); - schedule_work(&fifo->recover.work); - nvkm_fifo_kevent(&fifo->base, chid); + spin_lock_irqsave(&fifo->lock, flags); + nvkm_mask(fifo->engine.subdev.device, 0x002140, 0x80000000, 0x80000000); + spin_unlock_irqrestore(&fifo->lock, flags); } +void +gf100_fifo_nonstall_block(struct nvkm_event *event, int type, int index) +{ + struct nvkm_fifo *fifo = container_of(event, typeof(*fifo), nonstall.event); + unsigned long flags; + + spin_lock_irqsave(&fifo->lock, flags); + nvkm_mask(fifo->engine.subdev.device, 0x002140, 0x80000000, 0x00000000); + spin_unlock_irqrestore(&fifo->lock, flags); +} + +const struct nvkm_event_func +gf100_fifo_nonstall = { + .init = gf100_fifo_nonstall_allow, + .fini = gf100_fifo_nonstall_block, +}; + static const struct nvkm_enum -gf100_fifo_fault_engine[] = { +gf100_fifo_mmu_fault_engine[] = { { 0x00, "PGRAPH", NULL, NVKM_ENGINE_GR }, { 0x03, "PEEPHOLE", NULL, NVKM_ENGINE_IFB }, { 0x04, "BAR1", NULL, NVKM_SUBDEV_BAR }, { 0x05, "BAR3", NULL, NVKM_SUBDEV_INSTMEM }, - { 0x07, "PFIFO", NULL, NVKM_ENGINE_FIFO }, + { 0x07, "PFIFO" }, { 0x10, "PMSVLD", NULL, NVKM_ENGINE_MSVLD }, { 0x11, "PMSPPP", NULL, NVKM_ENGINE_MSPPP }, { 0x13, "PCOUNTER" }, @@ -212,7 +482,7 @@ gf100_fifo_fault_engine[] = { }; static const struct nvkm_enum -gf100_fifo_fault_reason[] = { +gf100_fifo_mmu_fault_reason[] = { { 0x00, "PT_NOT_PRESENT" }, { 0x01, "PT_TOO_SHORT" }, { 0x02, "PAGE_NOT_PRESENT" }, @@ -226,7 +496,7 @@ gf100_fifo_fault_reason[] = { }; static const struct nvkm_enum -gf100_fifo_fault_hubclient[] = { +gf100_fifo_mmu_fault_hubclient[] = { { 0x01, "PCOPY0" }, { 0x02, "PCOPY1" }, { 0x04, "DISPATCH" }, @@ -245,7 +515,7 @@ gf100_fifo_fault_hubclient[] = { }; static const struct nvkm_enum -gf100_fifo_fault_gpcclient[] = { +gf100_fifo_mmu_fault_gpcclient[] = { { 0x01, "TEX" }, { 0x0c, "ESETUP" }, { 0x0e, "CTXCTL" }, @@ -253,29 +523,55 @@ gf100_fifo_fault_gpcclient[] = { {} }; -static void -gf100_fifo_fault(struct nvkm_fifo *base, struct nvkm_fault_data *info) +const struct nvkm_enum +gf100_fifo_mmu_fault_access[] = { + { 0x00, "READ" }, + { 0x01, "WRITE" }, + {} +}; + +void +gf100_fifo_mmu_fault_recover(struct nvkm_fifo *fifo, struct nvkm_fault_data *info) { - struct gf100_fifo *fifo = gf100_fifo(base); - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; + struct nvkm_subdev *subdev = &fifo->engine.subdev; struct nvkm_device *device = subdev->device; - const struct nvkm_enum *er, *eu, *ec; + const struct nvkm_enum *er, *ee, *ec, *ea; struct nvkm_engine *engine = NULL; - struct nvkm_fifo_chan *chan; + struct nvkm_runl *runl; + struct nvkm_engn *engn; + struct nvkm_chan *chan; unsigned long flags; - char gpcid[8] = ""; + char ct[8] = "HUB/"; + + /* Lookup engine by MMU fault ID. */ + nvkm_runl_foreach(runl, fifo) { + engn = nvkm_runl_find_engn(engn, runl, engn->fault == info->engine); + if (engn) { + /* Fault triggered by CTXSW_TIMEOUT recovery procedure. */ + if (engn->func->mmu_fault_triggered && + engn->func->mmu_fault_triggered(engn)) { + nvkm_runl_rc_engn(runl, engn); + return; + } + + engine = engn->engine; + break; + } + } - er = nvkm_enum_find(gf100_fifo_fault_reason, info->reason); - eu = nvkm_enum_find(gf100_fifo_fault_engine, info->engine); + er = nvkm_enum_find(fifo->func->mmu_fault->reason, info->reason); + ee = nvkm_enum_find(fifo->func->mmu_fault->engine, info->engine); if (info->hub) { - ec = nvkm_enum_find(gf100_fifo_fault_hubclient, info->client); + ec = nvkm_enum_find(fifo->func->mmu_fault->hubclient, info->client); } else { - ec = nvkm_enum_find(gf100_fifo_fault_gpcclient, info->client); - snprintf(gpcid, sizeof(gpcid), "GPC%d/", info->gpc); + ec = nvkm_enum_find(fifo->func->mmu_fault->gpcclient, info->client); + snprintf(ct, sizeof(ct), "GPC%d/", info->gpc); } + ea = nvkm_enum_find(fifo->func->mmu_fault->access, info->access); - if (eu && eu->data2) { - switch (eu->data2) { + /* Handle BAR faults. */ + if (ee && ee->data2) { + switch (ee->data2) { case NVKM_SUBDEV_BAR: nvkm_bar_bar1_reset(device); break; @@ -286,77 +582,104 @@ gf100_fifo_fault(struct nvkm_fifo *base, struct nvkm_fault_data *info) nvkm_mask(device, 0x001718, 0x00000000, 0x00000000); break; default: - engine = nvkm_device_engine(device, eu->data2, eu->inst); break; } } - chan = nvkm_fifo_chan_inst(&fifo->base, info->inst, &flags); + chan = nvkm_chan_get_inst(&fifo->engine, info->inst, &flags); nvkm_error(subdev, - "%s fault at %010llx engine %02x [%s] client %02x [%s%s] " - "reason %02x [%s] on channel %d [%010llx %s]\n", - info->access ? "write" : "read", info->addr, - info->engine, eu ? eu->name : "", - info->client, gpcid, ec ? ec->name : "", - info->reason, er ? er->name : "", chan ? chan->chid : -1, - info->inst, chan ? chan->object.client->name : "unknown"); - - if (engine && chan) - gf100_fifo_recover(fifo, engine, (void *)chan); - nvkm_fifo_chan_put(&fifo->base, flags, &chan); + "fault %02x [%s] at %016llx engine %02x [%s] client %02x " + "[%s%s] reason %02x [%s] on channel %d [%010llx %s]\n", + info->access, ea ? ea->name : "", info->addr, + info->engine, ee ? ee->name : engine ? engine->subdev.name : "", + info->client, ct, ec ? ec->name : "", + info->reason, er ? er->name : "", + chan ? chan->id : -1, info->inst, chan ? chan->name : "unknown"); + + /* Handle host/engine faults. */ + if (chan) + nvkm_runl_rc_cgrp(chan->cgrp); + + nvkm_chan_put(&chan, flags); } -static const struct nvkm_enum -gf100_fifo_sched_reason[] = { - { 0x0a, "CTXSW_TIMEOUT" }, - {} +static const struct nvkm_fifo_func_mmu_fault +gf100_fifo_mmu_fault = { + .recover = gf100_fifo_mmu_fault_recover, + .access = gf100_fifo_mmu_fault_access, + .engine = gf100_fifo_mmu_fault_engine, + .reason = gf100_fifo_mmu_fault_reason, + .hubclient = gf100_fifo_mmu_fault_hubclient, + .gpcclient = gf100_fifo_mmu_fault_gpcclient, }; -static void -gf100_fifo_intr_sched_ctxsw(struct gf100_fifo *fifo) +void +gf100_fifo_intr_ctxsw_timeout(struct nvkm_fifo *fifo, u32 engm) { - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_engine *engine; - struct gf100_fifo_chan *chan; - unsigned long flags; - u32 engn; - - spin_lock_irqsave(&fifo->base.lock, flags); - for (engn = 0; engn < 6; engn++) { - u32 stat = nvkm_rd32(device, 0x002640 + (engn * 0x04)); - u32 busy = (stat & 0x80000000); - u32 save = (stat & 0x00100000); /* maybe? */ - u32 unk0 = (stat & 0x00040000); - u32 unk1 = (stat & 0x00001000); - u32 chid = (stat & 0x0000007f); - (void)save; - - if (busy && unk0 && unk1) { - list_for_each_entry(chan, &fifo->chan, head) { - if (chan->base.chid == chid) { - engine = gf100_fifo_id_engine(&fifo->base, engn); - if (!engine) - break; - gf100_fifo_recover(fifo, engine, chan); - break; + struct nvkm_runl *runl; + struct nvkm_engn *engn, *engn2; + bool cgid, cgid2; + int id, id2; + + nvkm_runl_foreach(runl, fifo) { + /* Stop the runlist, and go through all engines serving it. */ + nvkm_runl_block(runl); + nvkm_runl_foreach_engn_cond(engn, runl, engm & BIT(engn->id)) { + /* Determine what channel (group) the engine is on. */ + id = engn->func->cxid(engn, &cgid); + if (id >= 0) { + /* Trigger MMU fault on any engine(s) on that channel (group). */ + nvkm_runl_foreach_engn_cond(engn2, runl, engn2->func->cxid) { + id2 = engn2->func->cxid(engn2, &cgid2); + if (cgid2 == cgid && id2 == id) + engn2->func->mmu_fault_trigger(engn2); } } } + nvkm_runl_allow(runl); /* HW will keep runlist blocked via ERROR_SCHED_DISABLE. */ } - spin_unlock_irqrestore(&fifo->base.lock, flags); } static void -gf100_fifo_intr_sched(struct gf100_fifo *fifo) +gf100_fifo_intr_sched_ctxsw(struct nvkm_fifo *fifo) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; + struct nvkm_runl *runl; + struct nvkm_engn *engn; + u32 engm = 0; + + /* Look for any engines that are busy, and awaiting chsw ack. */ + nvkm_runl_foreach(runl, fifo) { + nvkm_runl_foreach_engn_cond(engn, runl, engn->func->chsw) { + if (WARN_ON(engn->fault < 0) || !engn->func->chsw(engn)) + continue; + + engm |= BIT(engn->id); + } + } + + if (!engm) + return; + + fifo->func->intr_ctxsw_timeout(fifo, engm); +} + +static const struct nvkm_enum +gf100_fifo_intr_sched_names[] = { + { 0x0a, "CTXSW_TIMEOUT" }, + {} +}; + +void +gf100_fifo_intr_sched(struct nvkm_fifo *fifo) +{ + struct nvkm_subdev *subdev = &fifo->engine.subdev; struct nvkm_device *device = subdev->device; u32 intr = nvkm_rd32(device, 0x00254c); u32 code = intr & 0x000000ff; const struct nvkm_enum *en; - en = nvkm_enum_find(gf100_fifo_sched_reason, code); + en = nvkm_enum_find(gf100_fifo_intr_sched_names, code); nvkm_error(subdev, "SCHED_ERROR %02x [%s]\n", code, en ? en->name : ""); @@ -370,7 +693,7 @@ gf100_fifo_intr_sched(struct gf100_fifo *fifo) } void -gf100_fifo_intr_fault(struct nvkm_fifo *fifo, int unit) +gf100_fifo_intr_mmu_fault_unit(struct nvkm_fifo *fifo, int unit) { struct nvkm_device *device = fifo->engine.subdev.device; u32 inst = nvkm_rd32(device, 0x002800 + (unit * 0x10)); @@ -393,61 +716,45 @@ gf100_fifo_intr_fault(struct nvkm_fifo *fifo, int unit) nvkm_fifo_fault(fifo, &info); } -static const struct nvkm_bitfield -gf100_fifo_pbdma_intr[] = { -/* { 0x00008000, "" } seen with null ib push */ - { 0x00200000, "ILLEGAL_MTHD" }, - { 0x00800000, "EMPTY_SUBC" }, - {} -}; - -static void -gf100_fifo_intr_pbdma(struct gf100_fifo *fifo, int unit) +void +gf100_fifo_intr_mmu_fault(struct nvkm_fifo *fifo) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - u32 stat = nvkm_rd32(device, 0x040108 + (unit * 0x2000)); - u32 addr = nvkm_rd32(device, 0x0400c0 + (unit * 0x2000)); - u32 data = nvkm_rd32(device, 0x0400c4 + (unit * 0x2000)); - u32 chid = nvkm_rd32(device, 0x040120 + (unit * 0x2000)) & 0x7f; - u32 subc = (addr & 0x00070000) >> 16; - u32 mthd = (addr & 0x00003ffc); - struct nvkm_fifo_chan *chan; - unsigned long flags; - u32 show= stat; - char msg[128]; + struct nvkm_device *device = fifo->engine.subdev.device; + unsigned long mask = nvkm_rd32(device, 0x00259c); + int unit; - if (stat & 0x00800000) { - if (device->sw) { - if (nvkm_sw_mthd(device->sw, chid, subc, mthd, data)) - show &= ~0x00800000; - } + for_each_set_bit(unit, &mask, 32) { + fifo->func->intr_mmu_fault_unit(fifo, unit); + nvkm_wr32(device, 0x00259c, BIT(unit)); } +} - if (show) { - nvkm_snprintbf(msg, sizeof(msg), gf100_fifo_pbdma_intr, show); - chan = nvkm_fifo_chan_chid(&fifo->base, chid, &flags); - nvkm_error(subdev, "PBDMA%d: %08x [%s] ch %d [%010llx %s] " - "subc %d mthd %04x data %08x\n", - unit, show, msg, chid, chan ? chan->inst->addr : 0, - chan ? chan->object.client->name : "unknown", - subc, mthd, data); - nvkm_fifo_chan_put(&fifo->base, flags, &chan); +bool +gf100_fifo_intr_pbdma(struct nvkm_fifo *fifo) +{ + struct nvkm_device *device = fifo->engine.subdev.device; + struct nvkm_runq *runq; + u32 mask = nvkm_rd32(device, 0x0025a0); + bool handled = false; + + nvkm_runq_foreach_cond(runq, fifo, mask & BIT(runq->id)) { + if (runq->func->intr(runq, NULL)) + handled = true; + + nvkm_wr32(device, 0x0025a0, BIT(runq->id)); } - nvkm_wr32(device, 0x0400c0 + (unit * 0x2000), 0x80600008); - nvkm_wr32(device, 0x040108 + (unit * 0x2000), stat); + return handled; } static void -gf100_fifo_intr_runlist(struct gf100_fifo *fifo) +gf100_fifo_intr_runlist(struct nvkm_fifo *fifo) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; + struct nvkm_subdev *subdev = &fifo->engine.subdev; struct nvkm_device *device = subdev->device; u32 intr = nvkm_rd32(device, 0x002a00); if (intr & 0x10000000) { - wake_up(&fifo->runlist.wait); nvkm_wr32(device, 0x002a00, 0x10000000); intr &= ~0x10000000; } @@ -459,9 +766,9 @@ gf100_fifo_intr_runlist(struct gf100_fifo *fifo) } static void -gf100_fifo_intr_engine_unit(struct gf100_fifo *fifo, int engn) +gf100_fifo_intr_engine_unit(struct nvkm_fifo *fifo, int engn) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; + struct nvkm_subdev *subdev = &fifo->engine.subdev; struct nvkm_device *device = subdev->device; u32 intr = nvkm_rd32(device, 0x0025a8 + (engn * 0x04)); u32 inte = nvkm_rd32(device, 0x002628); @@ -472,22 +779,22 @@ gf100_fifo_intr_engine_unit(struct gf100_fifo *fifo, int engn) for (unkn = 0; unkn < 8; unkn++) { u32 ints = (intr >> (unkn * 0x04)) & inte; if (ints & 0x1) { - nvkm_fifo_uevent(&fifo->base); + nvkm_event_ntfy(&fifo->nonstall.event, 0, NVKM_FIFO_NONSTALL_EVENT); ints &= ~1; } if (ints) { - nvkm_error(subdev, "ENGINE %d %d %01x", - engn, unkn, ints); + nvkm_error(subdev, "ENGINE %d %d %01x", engn, unkn, ints); nvkm_mask(device, 0x002628, ints, 0); } } } -void -gf100_fifo_intr_engine(struct gf100_fifo *fifo) +static void +gf100_fifo_intr_engine(struct nvkm_fifo *fifo) { - struct nvkm_device *device = fifo->base.engine.subdev.device; + struct nvkm_device *device = fifo->engine.subdev.device; u32 mask = nvkm_rd32(device, 0x0025a4); + while (mask) { u32 unit = __ffs(mask); gf100_fifo_intr_engine_unit(fifo, unit); @@ -495,11 +802,11 @@ gf100_fifo_intr_engine(struct gf100_fifo *fifo) } } -static void -gf100_fifo_intr(struct nvkm_fifo *base) +static irqreturn_t +gf100_fifo_intr(struct nvkm_inth *inth) { - struct gf100_fifo *fifo = gf100_fifo(base); - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; + struct nvkm_fifo *fifo = container_of(inth, typeof(*fifo), engine.subdev.inth); + struct nvkm_subdev *subdev = &fifo->engine.subdev; struct nvkm_device *device = subdev->device; u32 mask = nvkm_rd32(device, 0x002140); u32 stat = nvkm_rd32(device, 0x002100) & mask; @@ -532,25 +839,13 @@ gf100_fifo_intr(struct nvkm_fifo *base) } if (stat & 0x10000000) { - u32 mask = nvkm_rd32(device, 0x00259c); - while (mask) { - u32 unit = __ffs(mask); - gf100_fifo_intr_fault(&fifo->base, unit); - nvkm_wr32(device, 0x00259c, (1 << unit)); - mask &= ~(1 << unit); - } + gf100_fifo_intr_mmu_fault(fifo); stat &= ~0x10000000; } if (stat & 0x20000000) { - u32 mask = nvkm_rd32(device, 0x0025a0); - while (mask) { - u32 unit = __ffs(mask); - gf100_fifo_intr_pbdma(fifo, unit); - nvkm_wr32(device, 0x0025a0, (1 << unit)); - mask &= ~(1 << unit); - } - stat &= ~0x20000000; + if (gf100_fifo_intr_pbdma(fifo)) + stat &= ~0x20000000; } if (stat & 0x40000000) { @@ -565,71 +860,26 @@ gf100_fifo_intr(struct nvkm_fifo *base) if (stat) { nvkm_error(subdev, "INTR %08x\n", stat); + spin_lock(&fifo->lock); nvkm_mask(device, 0x002140, stat, 0x00000000); + spin_unlock(&fifo->lock); nvkm_wr32(device, 0x002100, stat); } -} - -static int -gf100_fifo_oneinit(struct nvkm_fifo *base) -{ - struct gf100_fifo *fifo = gf100_fifo(base); - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - struct nvkm_vmm *bar = nvkm_bar_bar1_vmm(device); - int ret; - - /* Determine number of PBDMAs by checking valid enable bits. */ - nvkm_wr32(device, 0x002204, 0xffffffff); - fifo->pbdma_nr = hweight32(nvkm_rd32(device, 0x002204)); - nvkm_debug(subdev, "%d PBDMA(s)\n", fifo->pbdma_nr); - - - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 0x1000, 0x1000, - false, &fifo->runlist.mem[0]); - if (ret) - return ret; - - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 0x1000, 0x1000, - false, &fifo->runlist.mem[1]); - if (ret) - return ret; - init_waitqueue_head(&fifo->runlist.wait); - - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 128 * 0x1000, - 0x1000, false, &fifo->user.mem); - if (ret) - return ret; - - ret = nvkm_vmm_get(bar, 12, nvkm_memory_size(fifo->user.mem), - &fifo->user.bar); - if (ret) - return ret; - - return nvkm_memory_map(fifo->user.mem, 0, bar, fifo->user.bar, NULL, 0); -} - -static void -gf100_fifo_fini(struct nvkm_fifo *base) -{ - struct gf100_fifo *fifo = gf100_fifo(base); - flush_work(&fifo->recover.work); + return IRQ_HANDLED; } static void -gf100_fifo_init(struct nvkm_fifo *base) +gf100_fifo_init_pbdmas(struct nvkm_fifo *fifo, u32 mask) { - struct gf100_fifo *fifo = gf100_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; - int i; + struct nvkm_device *device = fifo->engine.subdev.device; /* Enable PBDMAs. */ - nvkm_wr32(device, 0x000204, (1 << fifo->pbdma_nr) - 1); - nvkm_wr32(device, 0x002204, (1 << fifo->pbdma_nr) - 1); + nvkm_wr32(device, 0x000204, mask); + nvkm_wr32(device, 0x002204, mask); /* Assign engines to PBDMAs. */ - if (fifo->pbdma_nr >= 3) { + if ((mask & 7) == 7) { nvkm_wr32(device, 0x002208, ~(1 << 0)); /* PGRAPH */ nvkm_wr32(device, 0x00220c, ~(1 << 1)); /* PVP */ nvkm_wr32(device, 0x002210, ~(1 << 1)); /* PMSPP */ @@ -638,62 +888,82 @@ gf100_fifo_init(struct nvkm_fifo *base) nvkm_wr32(device, 0x00221c, ~(1 << 1)); /* PCE1 */ } - /* PBDMA[n] */ - for (i = 0; i < fifo->pbdma_nr; i++) { - nvkm_mask(device, 0x04013c + (i * 0x2000), 0x10000100, 0x00000000); - nvkm_wr32(device, 0x040108 + (i * 0x2000), 0xffffffff); /* INTR */ - nvkm_wr32(device, 0x04010c + (i * 0x2000), 0xfffffeff); /* INTREN */ - } + nvkm_mask(device, 0x002a04, 0xbfffffff, 0xbfffffff); +} + +static void +gf100_fifo_init(struct nvkm_fifo *fifo) +{ + struct nvkm_device *device = fifo->engine.subdev.device; nvkm_mask(device, 0x002200, 0x00000001, 0x00000001); - nvkm_wr32(device, 0x002254, 0x10000000 | fifo->user.bar->addr >> 12); + nvkm_wr32(device, 0x002254, 0x10000000 | fifo->userd.bar1->addr >> 12); nvkm_wr32(device, 0x002100, 0xffffffff); nvkm_wr32(device, 0x002140, 0x7fffffff); nvkm_wr32(device, 0x002628, 0x00000001); /* ENGINE_INTR_EN */ } -static void * -gf100_fifo_dtor(struct nvkm_fifo *base) +static int +gf100_fifo_runl_ctor(struct nvkm_fifo *fifo) +{ + struct nvkm_runl *runl; + + runl = nvkm_runl_new(fifo, 0, 0, 0); + if (IS_ERR(runl)) + return PTR_ERR(runl); + + nvkm_runl_add(runl, 0, fifo->func->engn, NVKM_ENGINE_GR, 0); + nvkm_runl_add(runl, 1, fifo->func->engn, NVKM_ENGINE_MSPDEC, 0); + nvkm_runl_add(runl, 2, fifo->func->engn, NVKM_ENGINE_MSPPP, 0); + nvkm_runl_add(runl, 3, fifo->func->engn, NVKM_ENGINE_MSVLD, 0); + nvkm_runl_add(runl, 4, fifo->func->engn, NVKM_ENGINE_CE, 0); + nvkm_runl_add(runl, 5, fifo->func->engn, NVKM_ENGINE_CE, 1); + nvkm_runl_add(runl, 15, &gf100_engn_sw, NVKM_ENGINE_SW, 0); + return 0; +} + +int +gf100_fifo_runq_nr(struct nvkm_fifo *fifo) +{ + struct nvkm_device *device = fifo->engine.subdev.device; + u32 save; + + /* Determine number of PBDMAs by checking valid enable bits. */ + save = nvkm_mask(device, 0x000204, 0xffffffff, 0xffffffff); + save = nvkm_mask(device, 0x000204, 0xffffffff, save); + return hweight32(save); +} + +int +gf100_fifo_chid_ctor(struct nvkm_fifo *fifo, int nr) { - struct gf100_fifo *fifo = gf100_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; - nvkm_vmm_put(nvkm_bar_bar1_vmm(device), &fifo->user.bar); - nvkm_memory_unref(&fifo->user.mem); - nvkm_memory_unref(&fifo->runlist.mem[0]); - nvkm_memory_unref(&fifo->runlist.mem[1]); - return fifo; + return nvkm_chid_new(&nvkm_chan_event, &fifo->engine.subdev, nr, 0, nr, &fifo->chid); } static const struct nvkm_fifo_func gf100_fifo = { - .dtor = gf100_fifo_dtor, - .oneinit = gf100_fifo_oneinit, + .chid_nr = nv50_fifo_chid_nr, + .chid_ctor = gf100_fifo_chid_ctor, + .runq_nr = gf100_fifo_runq_nr, + .runl_ctor = gf100_fifo_runl_ctor, .init = gf100_fifo_init, - .fini = gf100_fifo_fini, + .init_pbdmas = gf100_fifo_init_pbdmas, .intr = gf100_fifo_intr, - .fault = gf100_fifo_fault, - .engine_id = gf100_fifo_engine_id, - .id_engine = gf100_fifo_id_engine, - .uevent_init = gf100_fifo_uevent_init, - .uevent_fini = gf100_fifo_uevent_fini, - .chan = { - &gf100_fifo_gpfifo_oclass, - NULL - }, + .intr_mmu_fault_unit = gf100_fifo_intr_mmu_fault_unit, + .intr_ctxsw_timeout = gf100_fifo_intr_ctxsw_timeout, + .mmu_fault = &gf100_fifo_mmu_fault, + .nonstall = &gf100_fifo_nonstall, + .runl = &gf100_runl, + .runq = &gf100_runq, + .engn = &gf100_engn, + .cgrp = {{ }, &nv04_cgrp }, + .chan = {{ 0, 0, FERMI_CHANNEL_GPFIFO }, &gf100_chan }, }; int gf100_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - struct gf100_fifo *fifo; - - if (!(fifo = kzalloc(sizeof(*fifo), GFP_KERNEL))) - return -ENOMEM; - INIT_LIST_HEAD(&fifo->chan); - INIT_WORK(&fifo->recover.work, gf100_fifo_recover_work); - *pfifo = &fifo->base; - - return nvkm_fifo_ctor(&gf100_fifo, device, type, inst, 128, &fifo->base); + return nvkm_fifo_new_(&gf100_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gf100.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gf100.h deleted file mode 100644 index b8642490eb2f..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gf100.h +++ /dev/null @@ -1,38 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __GF100_FIFO_H__ -#define __GF100_FIFO_H__ -#define gf100_fifo(p) container_of((p), struct gf100_fifo, base) -#include "priv.h" - -#include <subdev/mmu.h> - -struct gf100_fifo_chan; -struct gf100_fifo { - struct nvkm_fifo base; - - struct list_head chan; - - struct { - struct work_struct work; - u64 mask; - } recover; - - int pbdma_nr; - - struct { - struct nvkm_memory *mem[2]; - int active; - wait_queue_head_t wait; - } runlist; - - struct { - struct nvkm_memory *mem; - struct nvkm_vma *bar; - } user; -}; - -void gf100_fifo_intr_engine(struct gf100_fifo *); -void gf100_fifo_runlist_insert(struct gf100_fifo *, struct gf100_fifo_chan *); -void gf100_fifo_runlist_remove(struct gf100_fifo *, struct gf100_fifo_chan *); -void gf100_fifo_runlist_commit(struct gf100_fifo *); -#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c index e771bd519ee2..d8a4d773a58c 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.c @@ -21,643 +21,318 @@ * * Authors: Ben Skeggs */ -#include "gk104.h" +#include "priv.h" #include "cgrp.h" -#include "changk104.h" +#include "chan.h" +#include "chid.h" +#include "runl.h" +#include "runq.h" -#include <core/client.h> #include <core/gpuobj.h> -#include <subdev/bar.h> -#include <subdev/fault.h> -#include <subdev/timer.h> +#include <subdev/mc.h> +#include <subdev/mmu.h> #include <subdev/top.h> -#include <engine/sw.h> #include <nvif/class.h> -#include <nvif/cl0080.h> +#include <nvif/if900d.h> void -gk104_fifo_engine_status(struct gk104_fifo *fifo, int engn, - struct gk104_fifo_engine_status *status) +gk104_chan_stop(struct nvkm_chan *chan) { - struct nvkm_engine *engine = fifo->engine[engn].engine; - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - u32 stat = nvkm_rd32(device, 0x002640 + (engn * 0x08)); - - status->busy = !!(stat & 0x80000000); - status->faulted = !!(stat & 0x40000000); - status->next.tsg = !!(stat & 0x10000000); - status->next.id = (stat & 0x0fff0000) >> 16; - status->chsw = !!(stat & 0x00008000); - status->save = !!(stat & 0x00004000); - status->load = !!(stat & 0x00002000); - status->prev.tsg = !!(stat & 0x00001000); - status->prev.id = (stat & 0x00000fff); - status->chan = NULL; - - if (status->busy && status->chsw) { - if (status->load && status->save) { - if (engine && nvkm_engine_chsw_load(engine)) - status->chan = &status->next; - else - status->chan = &status->prev; - } else - if (status->load) { - status->chan = &status->next; - } else { - status->chan = &status->prev; - } - } else - if (status->load) { - status->chan = &status->prev; - } + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; - nvkm_debug(subdev, "engine %02d: busy %d faulted %d chsw %d " - "save %d load %d %sid %d%s-> %sid %d%s\n", - engn, status->busy, status->faulted, - status->chsw, status->save, status->load, - status->prev.tsg ? "tsg" : "ch", status->prev.id, - status->chan == &status->prev ? "*" : " ", - status->next.tsg ? "tsg" : "ch", status->next.id, - status->chan == &status->next ? "*" : " "); -} - -int -gk104_fifo_class_new(struct nvkm_fifo *base, const struct nvkm_oclass *oclass, - void *argv, u32 argc, struct nvkm_object **pobject) -{ - struct gk104_fifo *fifo = gk104_fifo(base); - if (oclass->engn == &fifo->func->chan) { - const struct gk104_fifo_chan_user *user = oclass->engn; - return user->ctor(fifo, oclass, argv, argc, pobject); - } else - if (oclass->engn == &fifo->func->user) { - const struct gk104_fifo_user_user *user = oclass->engn; - return user->ctor(oclass, argv, argc, pobject); - } - WARN_ON(1); - return -EINVAL; -} - -int -gk104_fifo_class_get(struct nvkm_fifo *base, int index, - struct nvkm_oclass *oclass) -{ - struct gk104_fifo *fifo = gk104_fifo(base); - int c = 0; - - if (fifo->func->user.ctor && c++ == index) { - oclass->base = fifo->func->user.user; - oclass->engn = &fifo->func->user; - return 0; - } - - if (fifo->func->chan.ctor && c++ == index) { - oclass->base = fifo->func->chan.user; - oclass->engn = &fifo->func->chan; - return 0; - } - - return c; + nvkm_mask(device, 0x800004 + (chan->id * 8), 0x00000800, 0x00000800); } void -gk104_fifo_uevent_fini(struct nvkm_fifo *fifo) +gk104_chan_start(struct nvkm_chan *chan) { - struct nvkm_device *device = fifo->engine.subdev.device; - nvkm_mask(device, 0x002140, 0x80000000, 0x00000000); -} + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; -void -gk104_fifo_uevent_init(struct nvkm_fifo *fifo) -{ - struct nvkm_device *device = fifo->engine.subdev.device; - nvkm_mask(device, 0x002140, 0x80000000, 0x80000000); + nvkm_mask(device, 0x800004 + (chan->id * 8), 0x00000400, 0x00000400); } void -gk104_fifo_runlist_commit(struct gk104_fifo *fifo, int runl, - struct nvkm_memory *mem, int nr) +gk104_chan_unbind(struct nvkm_chan *chan) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - int target; + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; - switch (nvkm_memory_target(mem)) { - case NVKM_MEM_TARGET_VRAM: target = 0; break; - case NVKM_MEM_TARGET_NCOH: target = 3; break; - default: - WARN_ON(1); - return; - } - - nvkm_wr32(device, 0x002270, (nvkm_memory_addr(mem) >> 12) | - (target << 28)); - nvkm_wr32(device, 0x002274, (runl << 20) | nr); - - if (nvkm_msec(device, 2000, - if (!(nvkm_rd32(device, 0x002284 + (runl * 0x08)) & 0x00100000)) - break; - ) < 0) - nvkm_error(subdev, "runlist %d update timeout\n", runl); + nvkm_wr32(device, 0x800000 + (chan->id * 8), 0x00000000); } void -gk104_fifo_runlist_update(struct gk104_fifo *fifo, int runl) +gk104_chan_bind_inst(struct nvkm_chan *chan) { - const struct gk104_fifo_runlist_func *func = fifo->func->runlist; - struct gk104_fifo_chan *chan; - struct nvkm_memory *mem; - struct nvkm_fifo_cgrp *cgrp; - int nr = 0; - - mutex_lock(&fifo->base.mutex); - mem = fifo->runlist[runl].mem[fifo->runlist[runl].next]; - fifo->runlist[runl].next = !fifo->runlist[runl].next; - - nvkm_kmap(mem); - list_for_each_entry(chan, &fifo->runlist[runl].chan, head) { - func->chan(chan, mem, nr++ * func->size); - } - - list_for_each_entry(cgrp, &fifo->runlist[runl].cgrp, head) { - func->cgrp(cgrp, mem, nr++ * func->size); - list_for_each_entry(chan, &cgrp->chan, head) { - func->chan(chan, mem, nr++ * func->size); - } - } - nvkm_done(mem); + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; - func->commit(fifo, runl, mem, nr); - mutex_unlock(&fifo->base.mutex); + nvkm_wr32(device, 0x800000 + (chan->id * 8), 0x80000000 | chan->inst->addr >> 12); } void -gk104_fifo_runlist_remove(struct gk104_fifo *fifo, struct gk104_fifo_chan *chan) -{ - struct nvkm_fifo_cgrp *cgrp = chan->cgrp; - mutex_lock(&fifo->base.mutex); - if (!list_empty(&chan->head)) { - list_del_init(&chan->head); - if (cgrp && !--cgrp->chan_nr) - list_del_init(&cgrp->head); - } - mutex_unlock(&fifo->base.mutex); -} - -void -gk104_fifo_runlist_insert(struct gk104_fifo *fifo, struct gk104_fifo_chan *chan) -{ - struct nvkm_fifo_cgrp *cgrp = chan->cgrp; - mutex_lock(&fifo->base.mutex); - if (cgrp) { - if (!cgrp->chan_nr++) - list_add_tail(&cgrp->head, &fifo->runlist[chan->runl].cgrp); - list_add_tail(&chan->head, &cgrp->chan); - } else { - list_add_tail(&chan->head, &fifo->runlist[chan->runl].chan); - } - mutex_unlock(&fifo->base.mutex); -} - -void -gk104_fifo_runlist_chan(struct gk104_fifo_chan *chan, - struct nvkm_memory *memory, u32 offset) -{ - nvkm_wo32(memory, offset + 0, chan->base.chid); - nvkm_wo32(memory, offset + 4, 0x00000000); -} - -const struct gk104_fifo_runlist_func -gk104_fifo_runlist = { - .size = 8, - .chan = gk104_fifo_runlist_chan, - .commit = gk104_fifo_runlist_commit, +gk104_chan_bind(struct nvkm_chan *chan) +{ + struct nvkm_runl *runl = chan->cgrp->runl; + struct nvkm_device *device = runl->fifo->engine.subdev.device; + + nvkm_mask(device, 0x800004 + (chan->id * 8), 0x000f0000, runl->id << 16); + gk104_chan_bind_inst(chan); +} + +static int +gk104_chan_ramfc_write(struct nvkm_chan *chan, u64 offset, u64 length, u32 devm, bool priv) +{ + const u64 userd = nvkm_memory_addr(chan->userd.mem) + chan->userd.base; + const u32 limit2 = ilog2(length / 8); + + nvkm_kmap(chan->inst); + nvkm_wo32(chan->inst, 0x08, lower_32_bits(userd)); + nvkm_wo32(chan->inst, 0x0c, upper_32_bits(userd)); + nvkm_wo32(chan->inst, 0x10, 0x0000face); + nvkm_wo32(chan->inst, 0x30, 0xfffff902); + nvkm_wo32(chan->inst, 0x48, lower_32_bits(offset)); + nvkm_wo32(chan->inst, 0x4c, upper_32_bits(offset) | (limit2 << 16)); + nvkm_wo32(chan->inst, 0x84, 0x20400000); + nvkm_wo32(chan->inst, 0x94, 0x30000000 | devm); + nvkm_wo32(chan->inst, 0x9c, 0x00000100); + nvkm_wo32(chan->inst, 0xac, 0x0000001f); + nvkm_wo32(chan->inst, 0xe4, priv ? 0x00000020 : 0x00000000); + nvkm_wo32(chan->inst, 0xe8, chan->id); + nvkm_wo32(chan->inst, 0xb8, 0xf8000000); + nvkm_wo32(chan->inst, 0xf8, 0x10003080); /* 0x002310 */ + nvkm_wo32(chan->inst, 0xfc, 0x10000010); /* 0x002350 */ + nvkm_done(chan->inst); + return 0; +} + +const struct nvkm_chan_func_ramfc +gk104_chan_ramfc = { + .write = gk104_chan_ramfc_write, + .devm = 0xfff, + .priv = true, }; -void -gk104_fifo_pbdma_init(struct gk104_fifo *fifo) -{ - struct nvkm_device *device = fifo->base.engine.subdev.device; - nvkm_wr32(device, 0x000204, (1 << fifo->pbdma_nr) - 1); -} - -int -gk104_fifo_pbdma_nr(struct gk104_fifo *fifo) -{ - struct nvkm_device *device = fifo->base.engine.subdev.device; - /* Determine number of PBDMAs by checking valid enable bits. */ - nvkm_wr32(device, 0x000204, 0xffffffff); - return hweight32(nvkm_rd32(device, 0x000204)); -} - -const struct gk104_fifo_pbdma_func -gk104_fifo_pbdma = { - .nr = gk104_fifo_pbdma_nr, - .init = gk104_fifo_pbdma_init, +const struct nvkm_chan_func_userd +gk104_chan_userd = { + .bar = 1, + .size = 0x200, + .clear = gf100_chan_userd_clear, }; -struct nvkm_engine * -gk104_fifo_id_engine(struct nvkm_fifo *base, int engi) -{ - if (engi == GK104_FIFO_ENGN_SW) - return nvkm_device_engine(base->engine.subdev.device, NVKM_ENGINE_SW, 0); - - return gk104_fifo(base)->engine[engi].engine; -} - -int -gk104_fifo_engine_id(struct nvkm_fifo *base, struct nvkm_engine *engine) -{ - struct gk104_fifo *fifo = gk104_fifo(base); - int engn; +static const struct nvkm_chan_func +gk104_chan = { + .inst = &gf100_chan_inst, + .userd = &gk104_chan_userd, + .ramfc = &gk104_chan_ramfc, + .bind = gk104_chan_bind, + .unbind = gk104_chan_unbind, + .start = gk104_chan_start, + .stop = gk104_chan_stop, + .preempt = gf100_chan_preempt, +}; - if (engine->subdev.type == NVKM_ENGINE_SW) - return GK104_FIFO_ENGN_SW; +static void +gk104_ectx_bind(struct nvkm_engn *engn, struct nvkm_cctx *cctx, struct nvkm_chan *chan) +{ + u32 ptr0, ptr1 = 0; + u64 addr = 0ULL; + + switch (engn->engine->subdev.type) { + case NVKM_ENGINE_SW : return; + case NVKM_ENGINE_GR : ptr0 = 0x0210; break; + case NVKM_ENGINE_SEC : ptr0 = 0x0220; break; + case NVKM_ENGINE_MSPDEC: ptr0 = 0x0250; break; + case NVKM_ENGINE_MSPPP : ptr0 = 0x0260; break; + case NVKM_ENGINE_MSVLD : ptr0 = 0x0270; break; + case NVKM_ENGINE_VIC : ptr0 = 0x0280; break; + case NVKM_ENGINE_MSENC : ptr0 = 0x0290; break; + case NVKM_ENGINE_NVDEC : + ptr1 = 0x0270; + ptr0 = 0x0210; + break; + case NVKM_ENGINE_NVENC : + if (!engn->engine->subdev.inst) + ptr1 = 0x0290; + ptr0 = 0x0210; + break; + default: + WARN_ON(1); + return; + } - for (engn = 0; engn < fifo->engine_nr && engine; engn++) { - if (fifo->engine[engn].engine == engine) - return engn; + if (cctx) { + addr = cctx->vctx->vma->addr; + addr |= 4ULL; } - WARN_ON(1); - return -1; + nvkm_kmap(chan->inst); + nvkm_wo32(chan->inst, ptr0 + 0, lower_32_bits(addr)); + nvkm_wo32(chan->inst, ptr0 + 4, upper_32_bits(addr)); + if (ptr1) { + nvkm_wo32(chan->inst, ptr1 + 0, lower_32_bits(addr)); + nvkm_wo32(chan->inst, ptr1 + 4, upper_32_bits(addr)); + } + nvkm_done(chan->inst); } -static void -gk104_fifo_recover_work(struct work_struct *w) +int +gk104_ectx_ctor(struct nvkm_engn *engn, struct nvkm_vctx *vctx) { - struct gk104_fifo *fifo = container_of(w, typeof(*fifo), recover.work); - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_engine *engine; - unsigned long flags; - u32 engm, runm, todo; - int engn, runl; - - spin_lock_irqsave(&fifo->base.lock, flags); - runm = fifo->recover.runm; - engm = fifo->recover.engm; - fifo->recover.engm = 0; - fifo->recover.runm = 0; - spin_unlock_irqrestore(&fifo->base.lock, flags); - - nvkm_mask(device, 0x002630, runm, runm); - - for (todo = engm; engn = __ffs(todo), todo; todo &= ~BIT(engn)) { - if ((engine = fifo->engine[engn].engine)) { - nvkm_subdev_fini(&engine->subdev, false); - WARN_ON(nvkm_subdev_init(&engine->subdev)); - } - } + struct gf100_vmm_map_v0 args = { .priv = 1 }; + int ret; - for (todo = runm; runl = __ffs(todo), todo; todo &= ~BIT(runl)) - gk104_fifo_runlist_update(fifo, runl); + ret = nvkm_vmm_get(vctx->vmm, 12, vctx->inst->size, &vctx->vma); + if (ret) + return ret; - nvkm_wr32(device, 0x00262c, runm); - nvkm_mask(device, 0x002630, runm, 0x00000000); + return nvkm_memory_map(vctx->inst, 0, vctx->vmm, vctx->vma, &args, sizeof(args)); } -static void gk104_fifo_recover_engn(struct gk104_fifo *fifo, int engn); +/*TODO: clean this up */ +struct gk104_engn_status { + bool busy; + bool faulted; + bool chsw; + bool save; + bool load; + struct { + bool tsg; + u32 id; + } prev, next, *chan; +}; static void -gk104_fifo_recover_runl(struct gk104_fifo *fifo, int runl) -{ - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - const u32 runm = BIT(runl); - - assert_spin_locked(&fifo->base.lock); - if (fifo->recover.runm & runm) - return; - fifo->recover.runm |= runm; - - /* Block runlist to prevent channel assignment(s) from changing. */ - nvkm_mask(device, 0x002630, runm, runm); - - /* Schedule recovery. */ - nvkm_warn(subdev, "runlist %d: scheduled for recovery\n", runl); - schedule_work(&fifo->recover.work); -} - -static struct gk104_fifo_chan * -gk104_fifo_recover_chid(struct gk104_fifo *fifo, int runl, int chid) +gk104_engn_status(struct nvkm_engn *engn, struct gk104_engn_status *status) { - struct gk104_fifo_chan *chan; - struct nvkm_fifo_cgrp *cgrp; + u32 stat = nvkm_rd32(engn->runl->fifo->engine.subdev.device, 0x002640 + (engn->id * 0x08)); - list_for_each_entry(chan, &fifo->runlist[runl].chan, head) { - if (chan->base.chid == chid) { - list_del_init(&chan->head); - return chan; - } - } + status->busy = !!(stat & 0x80000000); + status->faulted = !!(stat & 0x40000000); + status->next.tsg = !!(stat & 0x10000000); + status->next.id = (stat & 0x0fff0000) >> 16; + status->chsw = !!(stat & 0x00008000); + status->save = !!(stat & 0x00004000); + status->load = !!(stat & 0x00002000); + status->prev.tsg = !!(stat & 0x00001000); + status->prev.id = (stat & 0x00000fff); + status->chan = NULL; - list_for_each_entry(cgrp, &fifo->runlist[runl].cgrp, head) { - if (cgrp->id == chid) { - chan = list_first_entry(&cgrp->chan, typeof(*chan), head); - list_del_init(&chan->head); - if (!--cgrp->chan_nr) - list_del_init(&cgrp->head); - return chan; + if (status->busy && status->chsw) { + if (status->load && status->save) { + if (nvkm_engine_chsw_load(engn->engine)) + status->chan = &status->next; + else + status->chan = &status->prev; + } else + if (status->load) { + status->chan = &status->next; + } else { + status->chan = &status->prev; } + } else + if (status->load) { + status->chan = &status->prev; } - return NULL; -} - -static void -gk104_fifo_recover_chan(struct nvkm_fifo *base, int chid) -{ - struct gk104_fifo *fifo = gk104_fifo(base); - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - const u32 stat = nvkm_rd32(device, 0x800004 + (chid * 0x08)); - const u32 runl = (stat & 0x000f0000) >> 16; - const bool used = (stat & 0x00000001); - unsigned long engn, engm = fifo->runlist[runl].engm; - struct gk104_fifo_chan *chan; - - assert_spin_locked(&fifo->base.lock); - if (!used) - return; - - /* Lookup SW state for channel, and mark it as dead. */ - chan = gk104_fifo_recover_chid(fifo, runl, chid); - if (chan) { - chan->killed = true; - nvkm_fifo_kevent(&fifo->base, chid); - } - - /* Disable channel. */ - nvkm_wr32(device, 0x800004 + (chid * 0x08), stat | 0x00000800); - nvkm_warn(subdev, "channel %d: killed\n", chid); - - /* Block channel assignments from changing during recovery. */ - gk104_fifo_recover_runl(fifo, runl); - - /* Schedule recovery for any engines the channel is on. */ - for_each_set_bit(engn, &engm, fifo->engine_nr) { - struct gk104_fifo_engine_status status; - gk104_fifo_engine_status(fifo, engn, &status); - if (!status.chan || status.chan->id != chid) - continue; - gk104_fifo_recover_engn(fifo, engn); - } + ENGN_DEBUG(engn, "%08x: busy %d faulted %d chsw %d save %d load %d %sid %d%s-> %sid %d%s", + stat, status->busy, status->faulted, status->chsw, status->save, status->load, + status->prev.tsg ? "tsg" : "ch", status->prev.id, + status->chan == &status->prev ? "*" : " ", + status->next.tsg ? "tsg" : "ch", status->next.id, + status->chan == &status->next ? "*" : " "); } -static void -gk104_fifo_recover_engn(struct gk104_fifo *fifo, int engn) +int +gk104_engn_cxid(struct nvkm_engn *engn, bool *cgid) { - struct nvkm_engine *engine = fifo->engine[engn].engine; - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - const u32 runl = fifo->engine[engn].runl; - const u32 engm = BIT(engn); - struct gk104_fifo_engine_status status; - int mmui = -1; - - assert_spin_locked(&fifo->base.lock); - if (fifo->recover.engm & engm) - return; - fifo->recover.engm |= engm; - - /* Block channel assignments from changing during recovery. */ - gk104_fifo_recover_runl(fifo, runl); + struct gk104_engn_status status; - /* Determine which channel (if any) is currently on the engine. */ - gk104_fifo_engine_status(fifo, engn, &status); + gk104_engn_status(engn, &status); if (status.chan) { - /* The channel is not longer viable, kill it. */ - gk104_fifo_recover_chan(&fifo->base, status.chan->id); + *cgid = status.chan->tsg; + return status.chan->id; } - /* Determine MMU fault ID for the engine, if we're not being - * called from the fault handler already. - */ - if (!status.faulted && engine) { - mmui = nvkm_top_fault_id(device, engine->subdev.type, engine->subdev.inst); - if (mmui < 0) { - const struct nvkm_enum *en = fifo->func->fault.engine; - for (; en && en->name; en++) { - if (en->data2 == engine->subdev.type && - en->inst == engine->subdev.inst) { - mmui = en->value; - break; - } - } - } - WARN_ON(mmui < 0); - } - - /* Trigger a MMU fault for the engine. - * - * No good idea why this is needed, but nvgpu does something similar, - * and it makes recovery from CTXSW_TIMEOUT a lot more reliable. - */ - if (mmui >= 0) { - nvkm_wr32(device, 0x002a30 + (engn * 0x04), 0x00000100 | mmui); - - /* Wait for fault to trigger. */ - nvkm_msec(device, 2000, - gk104_fifo_engine_status(fifo, engn, &status); - if (status.faulted) - break; - ); - - /* Release MMU fault trigger, and ACK the fault. */ - nvkm_wr32(device, 0x002a30 + (engn * 0x04), 0x00000000); - nvkm_wr32(device, 0x00259c, BIT(mmui)); - nvkm_wr32(device, 0x002100, 0x10000000); - } - - /* Schedule recovery. */ - nvkm_warn(subdev, "engine %d: scheduled for recovery\n", engn); - schedule_work(&fifo->recover.work); + return -ENODEV; } -static void -gk104_fifo_fault(struct nvkm_fifo *base, struct nvkm_fault_data *info) +bool +gk104_engn_chsw(struct nvkm_engn *engn) { - struct gk104_fifo *fifo = gk104_fifo(base); - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - const struct nvkm_enum *er, *ee, *ec, *ea; - struct nvkm_engine *engine = NULL; - struct nvkm_fifo_chan *chan; - unsigned long flags; - const char *en = ""; - char ct[8] = "HUB/"; - - er = nvkm_enum_find(fifo->func->fault.reason, info->reason); - ee = nvkm_enum_find(fifo->func->fault.engine, info->engine); - if (info->hub) { - ec = nvkm_enum_find(fifo->func->fault.hubclient, info->client); - } else { - ec = nvkm_enum_find(fifo->func->fault.gpcclient, info->client); - snprintf(ct, sizeof(ct), "GPC%d/", info->gpc); - } - ea = nvkm_enum_find(fifo->func->fault.access, info->access); + struct gk104_engn_status status; - if (ee && ee->data2) { - switch (ee->data2) { - case NVKM_SUBDEV_BAR: - nvkm_bar_bar1_reset(device); - break; - case NVKM_SUBDEV_INSTMEM: - nvkm_bar_bar2_reset(device); - break; - case NVKM_ENGINE_IFB: - nvkm_mask(device, 0x001718, 0x00000000, 0x00000000); - break; - default: - engine = nvkm_device_engine(device, ee->data2, 0); - break; - } - } - - if (ee == NULL) { - struct nvkm_subdev *subdev = nvkm_top_fault(device, info->engine); - if (subdev) { - if (subdev->func == &nvkm_engine) - engine = container_of(subdev, typeof(*engine), subdev); - en = engine->subdev.name; - } - } else { - en = ee->name; - } - - spin_lock_irqsave(&fifo->base.lock, flags); - chan = nvkm_fifo_chan_inst_locked(&fifo->base, info->inst); - - nvkm_error(subdev, - "fault %02x [%s] at %016llx engine %02x [%s] client %02x " - "[%s%s] reason %02x [%s] on channel %d [%010llx %s]\n", - info->access, ea ? ea->name : "", info->addr, - info->engine, ee ? ee->name : en, - info->client, ct, ec ? ec->name : "", - info->reason, er ? er->name : "", chan ? chan->chid : -1, - info->inst, chan ? chan->object.client->name : "unknown"); - - /* Kill the channel that caused the fault. */ - if (chan) - gk104_fifo_recover_chan(&fifo->base, chan->chid); - - /* Channel recovery will probably have already done this for the - * correct engine(s), but just in case we can't find the channel - * information... - */ - if (engine) { - int engn = fifo->base.func->engine_id(&fifo->base, engine); - if (engn >= 0 && engn != GK104_FIFO_ENGN_SW) - gk104_fifo_recover_engn(fifo, engn); - } + gk104_engn_status(engn, &status); + if (status.busy && status.chsw) + return true; - spin_unlock_irqrestore(&fifo->base.lock, flags); + return false; } -static const struct nvkm_enum -gk104_fifo_bind_reason[] = { - { 0x01, "BIND_NOT_UNBOUND" }, - { 0x02, "SNOOP_WITHOUT_BAR1" }, - { 0x03, "UNBIND_WHILE_RUNNING" }, - { 0x05, "INVALID_RUNLIST" }, - { 0x06, "INVALID_CTX_TGT" }, - { 0x0b, "UNBIND_WHILE_PARKED" }, - {} +const struct nvkm_engn_func +gk104_engn = { + .chsw = gk104_engn_chsw, + .cxid = gk104_engn_cxid, + .mmu_fault_trigger = gf100_engn_mmu_fault_trigger, + .mmu_fault_triggered = gf100_engn_mmu_fault_triggered, + .ctor = gk104_ectx_ctor, + .bind = gk104_ectx_bind, }; -void -gk104_fifo_intr_bind(struct gk104_fifo *fifo) +const struct nvkm_engn_func +gk104_engn_ce = { + .chsw = gk104_engn_chsw, + .cxid = gk104_engn_cxid, + .mmu_fault_trigger = gf100_engn_mmu_fault_trigger, + .mmu_fault_triggered = gf100_engn_mmu_fault_triggered, +}; + +bool +gk104_runq_idle(struct nvkm_runq *runq) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - u32 intr = nvkm_rd32(device, 0x00252c); - u32 code = intr & 0x000000ff; - const struct nvkm_enum *en = - nvkm_enum_find(gk104_fifo_bind_reason, code); + struct nvkm_device *device = runq->fifo->engine.subdev.device; - nvkm_error(subdev, "BIND_ERROR %02x [%s]\n", code, en ? en->name : ""); + return !(nvkm_rd32(device, 0x003080 + (runq->id * 4)) & 0x0000e000); } -static const struct nvkm_enum -gk104_fifo_sched_reason[] = { - { 0x0a, "CTXSW_TIMEOUT" }, +static const struct nvkm_bitfield +gk104_runq_intr_1_names[] = { + { 0x00000001, "HCE_RE_ILLEGAL_OP" }, + { 0x00000002, "HCE_RE_ALIGNB" }, + { 0x00000004, "HCE_PRIV" }, + { 0x00000008, "HCE_ILLEGAL_MTHD" }, + { 0x00000010, "HCE_ILLEGAL_CLASS" }, {} }; -static void -gk104_fifo_intr_sched_ctxsw(struct gk104_fifo *fifo) +static bool +gk104_runq_intr_1(struct nvkm_runq *runq) { - struct nvkm_device *device = fifo->base.engine.subdev.device; - unsigned long flags, engm = 0; - u32 engn; - - /* We need to ACK the SCHED_ERROR here, and prevent it reasserting, - * as MMU_FAULT cannot be triggered while it's pending. - */ - spin_lock_irqsave(&fifo->base.lock, flags); - nvkm_mask(device, 0x002140, 0x00000100, 0x00000000); - nvkm_wr32(device, 0x002100, 0x00000100); - - for (engn = 0; engn < fifo->engine_nr; engn++) { - struct gk104_fifo_engine_status status; - - gk104_fifo_engine_status(fifo, engn, &status); - if (!status.busy || !status.chsw) - continue; - - engm |= BIT(engn); - } - - for_each_set_bit(engn, &engm, fifo->engine_nr) - gk104_fifo_recover_engn(fifo, engn); - - nvkm_mask(device, 0x002140, 0x00000100, 0x00000100); - spin_unlock_irqrestore(&fifo->base.lock, flags); -} - -static void -gk104_fifo_intr_sched(struct gk104_fifo *fifo) -{ - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; + struct nvkm_subdev *subdev = &runq->fifo->engine.subdev; struct nvkm_device *device = subdev->device; - u32 intr = nvkm_rd32(device, 0x00254c); - u32 code = intr & 0x000000ff; - const struct nvkm_enum *en = - nvkm_enum_find(gk104_fifo_sched_reason, code); - - nvkm_error(subdev, "SCHED_ERROR %02x [%s]\n", code, en ? en->name : ""); + u32 mask = nvkm_rd32(device, 0x04014c + (runq->id * 0x2000)); + u32 stat = nvkm_rd32(device, 0x040148 + (runq->id * 0x2000)) & mask; + u32 chid = nvkm_rd32(device, 0x040120 + (runq->id * 0x2000)) & 0xfff; + char msg[128]; - switch (code) { - case 0x0a: - gk104_fifo_intr_sched_ctxsw(fifo); - break; - default: - break; + if (stat & 0x80000000) { + if (runq->func->intr_1_ctxnotvalid && + runq->func->intr_1_ctxnotvalid(runq, chid)) + stat &= ~0x80000000; } -} -void -gk104_fifo_intr_chsw(struct gk104_fifo *fifo) -{ - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - u32 stat = nvkm_rd32(device, 0x00256c); - nvkm_error(subdev, "CHSW_ERROR %08x\n", stat); - nvkm_wr32(device, 0x00256c, stat); -} + if (stat) { + nvkm_snprintbf(msg, sizeof(msg), gk104_runq_intr_1_names, stat); + nvkm_error(subdev, "PBDMA%d: %08x [%s] ch %d %08x %08x\n", + runq->id, stat, msg, chid, + nvkm_rd32(device, 0x040150 + (runq->id * 0x2000)), + nvkm_rd32(device, 0x040154 + (runq->id * 0x2000))); + } -void -gk104_fifo_intr_dropped_fault(struct gk104_fifo *fifo) -{ - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - u32 stat = nvkm_rd32(device, 0x00259c); - nvkm_error(subdev, "DROPPED_MMU_FAULT %08x\n", stat); + nvkm_wr32(device, 0x040148 + (runq->id * 0x2000), stat); + return true; } -static const struct nvkm_bitfield gk104_fifo_pbdma_intr_0[] = { +const struct nvkm_bitfield +gk104_runq_intr_0_names[] = { { 0x00000001, "MEMREQ" }, { 0x00000002, "MEMACK_TIMEOUT" }, { 0x00000004, "MEMACK_EXTRA" }, @@ -691,430 +366,111 @@ static const struct nvkm_bitfield gk104_fifo_pbdma_intr_0[] = { {} }; -void -gk104_fifo_intr_pbdma_0(struct gk104_fifo *fifo, int unit) +bool +gk104_runq_intr(struct nvkm_runq *runq, struct nvkm_runl *null) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - u32 mask = nvkm_rd32(device, 0x04010c + (unit * 0x2000)); - u32 stat = nvkm_rd32(device, 0x040108 + (unit * 0x2000)) & mask; - u32 addr = nvkm_rd32(device, 0x0400c0 + (unit * 0x2000)); - u32 data = nvkm_rd32(device, 0x0400c4 + (unit * 0x2000)); - u32 chid = nvkm_rd32(device, 0x040120 + (unit * 0x2000)) & 0xfff; - u32 subc = (addr & 0x00070000) >> 16; - u32 mthd = (addr & 0x00003ffc); - u32 show = stat; - struct nvkm_fifo_chan *chan; - unsigned long flags; - char msg[128]; - - if (stat & 0x00800000) { - if (device->sw) { - if (nvkm_sw_mthd(device->sw, chid, subc, mthd, data)) - show &= ~0x00800000; - } - } - - nvkm_wr32(device, 0x0400c0 + (unit * 0x2000), 0x80600008); - - if (show) { - nvkm_snprintbf(msg, sizeof(msg), gk104_fifo_pbdma_intr_0, show); - chan = nvkm_fifo_chan_chid(&fifo->base, chid, &flags); - nvkm_error(subdev, "PBDMA%d: %08x [%s] ch %d [%010llx %s] " - "subc %d mthd %04x data %08x\n", - unit, show, msg, chid, chan ? chan->inst->addr : 0, - chan ? chan->object.client->name : "unknown", - subc, mthd, data); - nvkm_fifo_chan_put(&fifo->base, flags, &chan); - } + bool intr0 = gf100_runq_intr(runq, NULL); + bool intr1 = gk104_runq_intr_1(runq); - nvkm_wr32(device, 0x040108 + (unit * 0x2000), stat); + return intr0 || intr1; } -static const struct nvkm_bitfield gk104_fifo_pbdma_intr_1[] = { - { 0x00000001, "HCE_RE_ILLEGAL_OP" }, - { 0x00000002, "HCE_RE_ALIGNB" }, - { 0x00000004, "HCE_PRIV" }, - { 0x00000008, "HCE_ILLEGAL_MTHD" }, - { 0x00000010, "HCE_ILLEGAL_CLASS" }, - {} -}; - void -gk104_fifo_intr_pbdma_1(struct gk104_fifo *fifo, int unit) +gk104_runq_init(struct nvkm_runq *runq) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - u32 mask = nvkm_rd32(device, 0x04014c + (unit * 0x2000)); - u32 stat = nvkm_rd32(device, 0x040148 + (unit * 0x2000)) & mask; - u32 chid = nvkm_rd32(device, 0x040120 + (unit * 0x2000)) & 0xfff; - char msg[128]; + struct nvkm_device *device = runq->fifo->engine.subdev.device; - if (stat) { - nvkm_snprintbf(msg, sizeof(msg), gk104_fifo_pbdma_intr_1, stat); - nvkm_error(subdev, "PBDMA%d: %08x [%s] ch %d %08x %08x\n", - unit, stat, msg, chid, - nvkm_rd32(device, 0x040150 + (unit * 0x2000)), - nvkm_rd32(device, 0x040154 + (unit * 0x2000))); - } + gf100_runq_init(runq); - nvkm_wr32(device, 0x040148 + (unit * 0x2000), stat); + nvkm_wr32(device, 0x040148 + (runq->id * 0x2000), 0xffffffff); /* HCE.INTR */ + nvkm_wr32(device, 0x04014c + (runq->id * 0x2000), 0xffffffff); /* HCE.INTREN */ } -void -gk104_fifo_intr_runlist(struct gk104_fifo *fifo) +static u32 +gk104_runq_runm(struct nvkm_runq *runq) { - struct nvkm_device *device = fifo->base.engine.subdev.device; - u32 mask = nvkm_rd32(device, 0x002a00); - while (mask) { - int runl = __ffs(mask); - wake_up(&fifo->runlist[runl].wait); - nvkm_wr32(device, 0x002a00, 1 << runl); - mask &= ~(1 << runl); - } + return nvkm_rd32(runq->fifo->engine.subdev.device, 0x002390 + (runq->id * 0x04)); } +const struct nvkm_runq_func +gk104_runq = { + .init = gk104_runq_init, + .intr = gk104_runq_intr, + .intr_0_names = gk104_runq_intr_0_names, + .idle = gk104_runq_idle, +}; + void -gk104_fifo_intr_engine(struct gk104_fifo *fifo) +gk104_runl_fault_clear(struct nvkm_runl *runl) { - nvkm_fifo_uevent(&fifo->base); -} - -static void -gk104_fifo_intr(struct nvkm_fifo *base) -{ - struct gk104_fifo *fifo = gk104_fifo(base); - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - u32 mask = nvkm_rd32(device, 0x002140); - u32 stat = nvkm_rd32(device, 0x002100) & mask; - - if (stat & 0x00000001) { - gk104_fifo_intr_bind(fifo); - nvkm_wr32(device, 0x002100, 0x00000001); - stat &= ~0x00000001; - } - - if (stat & 0x00000010) { - nvkm_error(subdev, "PIO_ERROR\n"); - nvkm_wr32(device, 0x002100, 0x00000010); - stat &= ~0x00000010; - } - - if (stat & 0x00000100) { - gk104_fifo_intr_sched(fifo); - nvkm_wr32(device, 0x002100, 0x00000100); - stat &= ~0x00000100; - } - - if (stat & 0x00010000) { - gk104_fifo_intr_chsw(fifo); - nvkm_wr32(device, 0x002100, 0x00010000); - stat &= ~0x00010000; - } - - if (stat & 0x00800000) { - nvkm_error(subdev, "FB_FLUSH_TIMEOUT\n"); - nvkm_wr32(device, 0x002100, 0x00800000); - stat &= ~0x00800000; - } - - if (stat & 0x01000000) { - nvkm_error(subdev, "LB_ERROR\n"); - nvkm_wr32(device, 0x002100, 0x01000000); - stat &= ~0x01000000; - } - - if (stat & 0x08000000) { - gk104_fifo_intr_dropped_fault(fifo); - nvkm_wr32(device, 0x002100, 0x08000000); - stat &= ~0x08000000; - } - - if (stat & 0x10000000) { - u32 mask = nvkm_rd32(device, 0x00259c); - while (mask) { - u32 unit = __ffs(mask); - fifo->func->intr.fault(&fifo->base, unit); - nvkm_wr32(device, 0x00259c, (1 << unit)); - mask &= ~(1 << unit); - } - stat &= ~0x10000000; - } - - if (stat & 0x20000000) { - u32 mask = nvkm_rd32(device, 0x0025a0); - while (mask) { - u32 unit = __ffs(mask); - gk104_fifo_intr_pbdma_0(fifo, unit); - gk104_fifo_intr_pbdma_1(fifo, unit); - nvkm_wr32(device, 0x0025a0, (1 << unit)); - mask &= ~(1 << unit); - } - stat &= ~0x20000000; - } - - if (stat & 0x40000000) { - gk104_fifo_intr_runlist(fifo); - stat &= ~0x40000000; - } - - if (stat & 0x80000000) { - nvkm_wr32(device, 0x002100, 0x80000000); - gk104_fifo_intr_engine(fifo); - stat &= ~0x80000000; - } - - if (stat) { - nvkm_error(subdev, "INTR %08x\n", stat); - nvkm_mask(device, 0x002140, stat, 0x00000000); - nvkm_wr32(device, 0x002100, stat); - } + nvkm_wr32(runl->fifo->engine.subdev.device, 0x00262c, BIT(runl->id)); } void -gk104_fifo_fini(struct nvkm_fifo *base) +gk104_runl_allow(struct nvkm_runl *runl, u32 engm) { - struct gk104_fifo *fifo = gk104_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; - flush_work(&fifo->recover.work); - /* allow mmu fault interrupts, even when we're not using fifo */ - nvkm_mask(device, 0x002140, 0x10000000, 0x10000000); + nvkm_mask(runl->fifo->engine.subdev.device, 0x002630, BIT(runl->id), 0x00000000); } -int -gk104_fifo_info(struct nvkm_fifo *base, u64 mthd, u64 *data) +void +gk104_runl_block(struct nvkm_runl *runl, u32 engm) { - struct gk104_fifo *fifo = gk104_fifo(base); - switch (mthd) { - case NV_DEVICE_HOST_RUNLISTS: - *data = (1ULL << fifo->runlist_nr) - 1; - return 0; - case NV_DEVICE_HOST_RUNLIST_ENGINES: { - if (*data < fifo->runlist_nr) { - unsigned long engm = fifo->runlist[*data].engm; - struct nvkm_engine *engine; - int engn; - *data = 0; - for_each_set_bit(engn, &engm, fifo->engine_nr) { - if ((engine = fifo->engine[engn].engine)) { -#define CASE(n) case NVKM_ENGINE_##n: *data |= NV_DEVICE_HOST_RUNLIST_ENGINES_##n; break - switch (engine->subdev.type) { - CASE(SW ); - CASE(GR ); - CASE(MPEG ); - CASE(ME ); - CASE(CIPHER); - CASE(BSP ); - CASE(VP ); - CASE(CE ); - CASE(SEC ); - CASE(MSVLD ); - CASE(MSPDEC); - CASE(MSPPP ); - CASE(MSENC ); - CASE(VIC ); - CASE(SEC2 ); - CASE(NVDEC ); - CASE(NVENC ); - default: - WARN_ON(1); - break; - } - } - } - return 0; - } - } - return -EINVAL; - default: - return -EINVAL; - } + nvkm_mask(runl->fifo->engine.subdev.device, 0x002630, BIT(runl->id), BIT(runl->id)); } -int -gk104_fifo_oneinit(struct nvkm_fifo *base) +bool +gk104_runl_pending(struct nvkm_runl *runl) { - struct gk104_fifo *fifo = gk104_fifo(base); - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - struct nvkm_vmm *bar = nvkm_bar_bar1_vmm(device); - struct nvkm_top_device *tdev; - int pbid, ret, i, j; - u32 *map; - - fifo->pbdma_nr = fifo->func->pbdma->nr(fifo); - nvkm_debug(subdev, "%d PBDMA(s)\n", fifo->pbdma_nr); - - /* Read PBDMA->runlist(s) mapping from HW. */ - if (!(map = kcalloc(fifo->pbdma_nr, sizeof(*map), GFP_KERNEL))) - return -ENOMEM; + struct nvkm_device *device = runl->fifo->engine.subdev.device; - for (i = 0; i < fifo->pbdma_nr; i++) - map[i] = nvkm_rd32(device, 0x002390 + (i * 0x04)); - - /* Determine runlist configuration from topology device info. */ - list_for_each_entry(tdev, &device->top->device, head) { - const int engn = tdev->engine; - char _en[16], *en; - - if (engn < 0) - continue; - - /* Determine which PBDMA handles requests for this engine. */ - for (j = 0, pbid = -1; j < fifo->pbdma_nr; j++) { - if (map[j] & BIT(tdev->runlist)) { - pbid = j; - break; - } - } - - fifo->engine[engn].engine = nvkm_device_engine(device, tdev->type, tdev->inst); - if (!fifo->engine[engn].engine) { - snprintf(_en, sizeof(_en), "%s, %d", - nvkm_subdev_type[tdev->type], tdev->inst); - en = _en; - } else { - en = fifo->engine[engn].engine->subdev.name; - } - - nvkm_debug(subdev, "engine %2d: runlist %2d pbdma %2d (%s)\n", - tdev->engine, tdev->runlist, pbid, en); - - fifo->engine[engn].runl = tdev->runlist; - fifo->engine[engn].pbid = pbid; - fifo->engine_nr = max(fifo->engine_nr, engn + 1); - fifo->runlist[tdev->runlist].engm |= BIT(engn); - fifo->runlist[tdev->runlist].engm_sw |= BIT(engn); - if (tdev->type == NVKM_ENGINE_GR) - fifo->runlist[tdev->runlist].engm_sw |= BIT(GK104_FIFO_ENGN_SW); - fifo->runlist_nr = max(fifo->runlist_nr, tdev->runlist + 1); - } - - kfree(map); - - for (i = 0; i < fifo->runlist_nr; i++) { - for (j = 0; j < ARRAY_SIZE(fifo->runlist[i].mem); j++) { - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, - fifo->base.nr * 2/* TSG+chan */ * - fifo->func->runlist->size, - 0x1000, false, - &fifo->runlist[i].mem[j]); - if (ret) - return ret; - } - - init_waitqueue_head(&fifo->runlist[i].wait); - INIT_LIST_HEAD(&fifo->runlist[i].cgrp); - INIT_LIST_HEAD(&fifo->runlist[i].chan); - } - - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, - fifo->base.nr * 0x200, 0x1000, true, - &fifo->user.mem); - if (ret) - return ret; - - ret = nvkm_vmm_get(bar, 12, nvkm_memory_size(fifo->user.mem), - &fifo->user.bar); - if (ret) - return ret; - - return nvkm_memory_map(fifo->user.mem, 0, bar, fifo->user.bar, NULL, 0); + return nvkm_rd32(device, 0x002284 + (runl->id * 0x08)) & 0x00100000; } void -gk104_fifo_init(struct nvkm_fifo *base) +gk104_runl_commit(struct nvkm_runl *runl, struct nvkm_memory *memory, u32 start, int count) { - struct gk104_fifo *fifo = gk104_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; - int i; - - /* Enable PBDMAs. */ - fifo->func->pbdma->init(fifo); - - /* PBDMA[n] */ - for (i = 0; i < fifo->pbdma_nr; i++) { - nvkm_mask(device, 0x04013c + (i * 0x2000), 0x10000100, 0x00000000); - nvkm_wr32(device, 0x040108 + (i * 0x2000), 0xffffffff); /* INTR */ - nvkm_wr32(device, 0x04010c + (i * 0x2000), 0xfffffeff); /* INTREN */ - } - - /* PBDMA[n].HCE */ - for (i = 0; i < fifo->pbdma_nr; i++) { - nvkm_wr32(device, 0x040148 + (i * 0x2000), 0xffffffff); /* INTR */ - nvkm_wr32(device, 0x04014c + (i * 0x2000), 0xffffffff); /* INTREN */ - } - - nvkm_wr32(device, 0x002254, 0x10000000 | fifo->user.bar->addr >> 12); - - if (fifo->func->pbdma->init_timeout) - fifo->func->pbdma->init_timeout(fifo); - - nvkm_wr32(device, 0x002100, 0xffffffff); - nvkm_wr32(device, 0x002140, 0x7fffffff); -} - -void * -gk104_fifo_dtor(struct nvkm_fifo *base) -{ - struct gk104_fifo *fifo = gk104_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; - int i; - - nvkm_vmm_put(nvkm_bar_bar1_vmm(device), &fifo->user.bar); - nvkm_memory_unref(&fifo->user.mem); + struct nvkm_fifo *fifo = runl->fifo; + struct nvkm_device *device = fifo->engine.subdev.device; + u64 addr = nvkm_memory_addr(memory) + start; + int target; - for (i = 0; i < fifo->runlist_nr; i++) { - nvkm_memory_unref(&fifo->runlist[i].mem[1]); - nvkm_memory_unref(&fifo->runlist[i].mem[0]); + switch (nvkm_memory_target(memory)) { + case NVKM_MEM_TARGET_VRAM: target = 0; break; + case NVKM_MEM_TARGET_NCOH: target = 3; break; + default: + WARN_ON(1); + return; } - return fifo; + spin_lock_irq(&fifo->lock); + nvkm_wr32(device, 0x002270, (target << 28) | (addr >> 12)); + nvkm_wr32(device, 0x002274, (runl->id << 20) | count); + spin_unlock_irq(&fifo->lock); } -static const struct nvkm_fifo_func -gk104_fifo_ = { - .dtor = gk104_fifo_dtor, - .oneinit = gk104_fifo_oneinit, - .info = gk104_fifo_info, - .init = gk104_fifo_init, - .fini = gk104_fifo_fini, - .intr = gk104_fifo_intr, - .fault = gk104_fifo_fault, - .engine_id = gk104_fifo_engine_id, - .id_engine = gk104_fifo_id_engine, - .uevent_init = gk104_fifo_uevent_init, - .uevent_fini = gk104_fifo_uevent_fini, - .recover_chan = gk104_fifo_recover_chan, - .class_get = gk104_fifo_class_get, - .class_new = gk104_fifo_class_new, -}; - -int -gk104_fifo_new_(const struct gk104_fifo_func *func, struct nvkm_device *device, - enum nvkm_subdev_type type, int inst, int nr, struct nvkm_fifo **pfifo) +void +gk104_runl_insert_chan(struct nvkm_chan *chan, struct nvkm_memory *memory, u64 offset) { - struct gk104_fifo *fifo; - - if (!(fifo = kzalloc(sizeof(*fifo), GFP_KERNEL))) - return -ENOMEM; - fifo->func = func; - INIT_WORK(&fifo->recover.work, gk104_fifo_recover_work); - *pfifo = &fifo->base; - - return nvkm_fifo_ctor(&gk104_fifo_, device, type, inst, nr, &fifo->base); + nvkm_wo32(memory, offset + 0, chan->id); + nvkm_wo32(memory, offset + 4, 0x00000000); } -const struct nvkm_enum -gk104_fifo_fault_access[] = { - { 0x0, "READ" }, - { 0x1, "WRITE" }, - {} +static const struct nvkm_runl_func +gk104_runl = { + .size = 8, + .update = nv50_runl_update, + .insert_chan = gk104_runl_insert_chan, + .commit = gk104_runl_commit, + .wait = nv50_runl_wait, + .pending = gk104_runl_pending, + .block = gk104_runl_block, + .allow = gk104_runl_allow, + .fault_clear = gk104_runl_fault_clear, + .preempt_pending = gf100_runl_preempt_pending, }; -const struct nvkm_enum -gk104_fifo_fault_engine[] = { +static const struct nvkm_enum +gk104_fifo_mmu_fault_engine[] = { { 0x00, "GR", NULL, NVKM_ENGINE_GR }, { 0x01, "DISPLAY" }, { 0x02, "CAPTURE" }, @@ -1122,14 +478,14 @@ gk104_fifo_fault_engine[] = { { 0x04, "BAR1", NULL, NVKM_SUBDEV_BAR }, { 0x05, "BAR2", NULL, NVKM_SUBDEV_INSTMEM }, { 0x06, "SCHED" }, - { 0x07, "HOST0", NULL, NVKM_ENGINE_FIFO }, - { 0x08, "HOST1", NULL, NVKM_ENGINE_FIFO }, - { 0x09, "HOST2", NULL, NVKM_ENGINE_FIFO }, - { 0x0a, "HOST3", NULL, NVKM_ENGINE_FIFO }, - { 0x0b, "HOST4", NULL, NVKM_ENGINE_FIFO }, - { 0x0c, "HOST5", NULL, NVKM_ENGINE_FIFO }, - { 0x0d, "HOST6", NULL, NVKM_ENGINE_FIFO }, - { 0x0e, "HOST7", NULL, NVKM_ENGINE_FIFO }, + { 0x07, "HOST0" }, + { 0x08, "HOST1" }, + { 0x09, "HOST2" }, + { 0x0a, "HOST3" }, + { 0x0b, "HOST4" }, + { 0x0c, "HOST5" }, + { 0x0d, "HOST6" }, + { 0x0e, "HOST7" }, { 0x0f, "HOSTSR" }, { 0x10, "MSVLD", NULL, NVKM_ENGINE_MSVLD }, { 0x11, "MSPPP", NULL, NVKM_ENGINE_MSPPP }, @@ -1145,7 +501,7 @@ gk104_fifo_fault_engine[] = { }; const struct nvkm_enum -gk104_fifo_fault_reason[] = { +gk104_fifo_mmu_fault_reason[] = { { 0x00, "PDE" }, { 0x01, "PDE_SIZE" }, { 0x02, "PTE" }, @@ -1166,7 +522,7 @@ gk104_fifo_fault_reason[] = { }; const struct nvkm_enum -gk104_fifo_fault_hubclient[] = { +gk104_fifo_mmu_fault_hubclient[] = { { 0x00, "VIP" }, { 0x01, "CE0" }, { 0x02, "CE1" }, @@ -1203,7 +559,7 @@ gk104_fifo_fault_hubclient[] = { }; const struct nvkm_enum -gk104_fifo_fault_gpcclient[] = { +gk104_fifo_mmu_fault_gpcclient[] = { { 0x00, "L1_0" }, { 0x01, "T1_0" }, { 0x02, "PE_0" }, { 0x03, "L1_1" }, { 0x04, "T1_1" }, { 0x05, "PE_1" }, { 0x06, "L1_2" }, { 0x07, "T1_2" }, { 0x08, "PE_2" }, @@ -1228,22 +584,250 @@ gk104_fifo_fault_gpcclient[] = { {} }; -static const struct gk104_fifo_func +const struct nvkm_fifo_func_mmu_fault +gk104_fifo_mmu_fault = { + .recover = gf100_fifo_mmu_fault_recover, + .access = gf100_fifo_mmu_fault_access, + .engine = gk104_fifo_mmu_fault_engine, + .reason = gk104_fifo_mmu_fault_reason, + .hubclient = gk104_fifo_mmu_fault_hubclient, + .gpcclient = gk104_fifo_mmu_fault_gpcclient, +}; + +static const struct nvkm_enum +gk104_fifo_intr_bind_reason[] = { + { 0x01, "BIND_NOT_UNBOUND" }, + { 0x02, "SNOOP_WITHOUT_BAR1" }, + { 0x03, "UNBIND_WHILE_RUNNING" }, + { 0x05, "INVALID_RUNLIST" }, + { 0x06, "INVALID_CTX_TGT" }, + { 0x0b, "UNBIND_WHILE_PARKED" }, + {} +}; + +void +gk104_fifo_intr_bind(struct nvkm_fifo *fifo) +{ + struct nvkm_subdev *subdev = &fifo->engine.subdev; + u32 intr = nvkm_rd32(subdev->device, 0x00252c); + u32 code = intr & 0x000000ff; + const struct nvkm_enum *en = nvkm_enum_find(gk104_fifo_intr_bind_reason, code); + + nvkm_error(subdev, "BIND_ERROR %02x [%s]\n", code, en ? en->name : ""); +} + +void +gk104_fifo_intr_chsw(struct nvkm_fifo *fifo) +{ + struct nvkm_subdev *subdev = &fifo->engine.subdev; + struct nvkm_device *device = subdev->device; + u32 stat = nvkm_rd32(device, 0x00256c); + + nvkm_error(subdev, "CHSW_ERROR %08x\n", stat); + nvkm_wr32(device, 0x00256c, stat); +} + +static void +gk104_fifo_intr_dropped_fault(struct nvkm_fifo *fifo) +{ + struct nvkm_subdev *subdev = &fifo->engine.subdev; + u32 stat = nvkm_rd32(subdev->device, 0x00259c); + + nvkm_error(subdev, "DROPPED_MMU_FAULT %08x\n", stat); +} + +void +gk104_fifo_intr_runlist(struct nvkm_fifo *fifo) +{ + struct nvkm_device *device = fifo->engine.subdev.device; + struct nvkm_runl *runl; + u32 mask = nvkm_rd32(device, 0x002a00); + + nvkm_runl_foreach_cond(runl, fifo, mask & BIT(runl->id)) { + nvkm_wr32(device, 0x002a00, BIT(runl->id)); + } +} + +irqreturn_t +gk104_fifo_intr(struct nvkm_inth *inth) +{ + struct nvkm_fifo *fifo = container_of(inth, typeof(*fifo), engine.subdev.inth); + struct nvkm_subdev *subdev = &fifo->engine.subdev; + struct nvkm_device *device = subdev->device; + u32 mask = nvkm_rd32(device, 0x002140); + u32 stat = nvkm_rd32(device, 0x002100) & mask; + + if (stat & 0x00000001) { + gk104_fifo_intr_bind(fifo); + nvkm_wr32(device, 0x002100, 0x00000001); + stat &= ~0x00000001; + } + + if (stat & 0x00000010) { + nvkm_error(subdev, "PIO_ERROR\n"); + nvkm_wr32(device, 0x002100, 0x00000010); + stat &= ~0x00000010; + } + + if (stat & 0x00000100) { + gf100_fifo_intr_sched(fifo); + nvkm_wr32(device, 0x002100, 0x00000100); + stat &= ~0x00000100; + } + + if (stat & 0x00010000) { + gk104_fifo_intr_chsw(fifo); + nvkm_wr32(device, 0x002100, 0x00010000); + stat &= ~0x00010000; + } + + if (stat & 0x00800000) { + nvkm_error(subdev, "FB_FLUSH_TIMEOUT\n"); + nvkm_wr32(device, 0x002100, 0x00800000); + stat &= ~0x00800000; + } + + if (stat & 0x01000000) { + nvkm_error(subdev, "LB_ERROR\n"); + nvkm_wr32(device, 0x002100, 0x01000000); + stat &= ~0x01000000; + } + + if (stat & 0x08000000) { + gk104_fifo_intr_dropped_fault(fifo); + nvkm_wr32(device, 0x002100, 0x08000000); + stat &= ~0x08000000; + } + + if (stat & 0x10000000) { + gf100_fifo_intr_mmu_fault(fifo); + stat &= ~0x10000000; + } + + if (stat & 0x20000000) { + if (gf100_fifo_intr_pbdma(fifo)) + stat &= ~0x20000000; + } + + if (stat & 0x40000000) { + gk104_fifo_intr_runlist(fifo); + stat &= ~0x40000000; + } + + if (stat & 0x80000000) { + nvkm_wr32(device, 0x002100, 0x80000000); + nvkm_event_ntfy(&fifo->nonstall.event, 0, NVKM_FIFO_NONSTALL_EVENT); + stat &= ~0x80000000; + } + + if (stat) { + nvkm_error(subdev, "INTR %08x\n", stat); + spin_lock(&fifo->lock); + nvkm_mask(device, 0x002140, stat, 0x00000000); + spin_unlock(&fifo->lock); + nvkm_wr32(device, 0x002100, stat); + } + + return IRQ_HANDLED; +} + +void +gk104_fifo_init_pbdmas(struct nvkm_fifo *fifo, u32 mask) +{ + struct nvkm_device *device = fifo->engine.subdev.device; + + nvkm_wr32(device, 0x000204, mask); + nvkm_mask(device, 0x002a04, 0xbfffffff, 0xbfffffff); +} + +void +gk104_fifo_init(struct nvkm_fifo *fifo) +{ + struct nvkm_device *device = fifo->engine.subdev.device; + + if (fifo->func->chan.func->userd->bar == 1) + nvkm_wr32(device, 0x002254, 0x10000000 | fifo->userd.bar1->addr >> 12); + + nvkm_wr32(device, 0x002100, 0xffffffff); + nvkm_wr32(device, 0x002140, 0x7fffffff); +} + +int +gk104_fifo_runl_ctor(struct nvkm_fifo *fifo) +{ + struct nvkm_device *device = fifo->engine.subdev.device; + struct nvkm_top_device *tdev; + struct nvkm_runl *runl; + struct nvkm_runq *runq; + const struct nvkm_engn_func *func; + + nvkm_list_foreach(tdev, &device->top->device, head, tdev->runlist >= 0) { + runl = nvkm_runl_get(fifo, tdev->runlist, tdev->runlist); + if (!runl) { + runl = nvkm_runl_new(fifo, tdev->runlist, tdev->runlist, 0); + if (IS_ERR(runl)) + return PTR_ERR(runl); + + nvkm_runq_foreach_cond(runq, fifo, gk104_runq_runm(runq) & BIT(runl->id)) { + if (WARN_ON(runl->runq_nr == ARRAY_SIZE(runl->runq))) + return -ENOMEM; + + runl->runq[runl->runq_nr++] = runq; + } + + } + + if (tdev->engine < 0) + continue; + + switch (tdev->type) { + case NVKM_ENGINE_CE: + func = fifo->func->engn_ce; + break; + case NVKM_ENGINE_GR: + nvkm_runl_add(runl, 15, &gf100_engn_sw, NVKM_ENGINE_SW, 0); + fallthrough; + default: + func = fifo->func->engn; + break; + } + + nvkm_runl_add(runl, tdev->engine, func, tdev->type, tdev->inst); + } + + return 0; +} + +int +gk104_fifo_chid_nr(struct nvkm_fifo *fifo) +{ + return 4096; +} + +static const struct nvkm_fifo_func gk104_fifo = { - .intr.fault = gf100_fifo_intr_fault, - .pbdma = &gk104_fifo_pbdma, - .fault.access = gk104_fifo_fault_access, - .fault.engine = gk104_fifo_fault_engine, - .fault.reason = gk104_fifo_fault_reason, - .fault.hubclient = gk104_fifo_fault_hubclient, - .fault.gpcclient = gk104_fifo_fault_gpcclient, - .runlist = &gk104_fifo_runlist, - .chan = {{0,0,KEPLER_CHANNEL_GPFIFO_A}, gk104_fifo_gpfifo_new }, + .chid_nr = gk104_fifo_chid_nr, + .chid_ctor = gf100_fifo_chid_ctor, + .runq_nr = gf100_fifo_runq_nr, + .runl_ctor = gk104_fifo_runl_ctor, + .init = gk104_fifo_init, + .init_pbdmas = gk104_fifo_init_pbdmas, + .intr = gk104_fifo_intr, + .intr_mmu_fault_unit = gf100_fifo_intr_mmu_fault_unit, + .intr_ctxsw_timeout = gf100_fifo_intr_ctxsw_timeout, + .mmu_fault = &gk104_fifo_mmu_fault, + .nonstall = &gf100_fifo_nonstall, + .runl = &gk104_runl, + .runq = &gk104_runq, + .engn = &gk104_engn, + .engn_ce = &gk104_engn_ce, + .cgrp = {{ }, &nv04_cgrp }, + .chan = {{ 0, 0, KEPLER_CHANNEL_GPFIFO_A }, &gk104_chan }, }; int gk104_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return gk104_fifo_new_(&gk104_fifo, device, type, inst, 4096, pfifo); + return nvkm_fifo_new_(&gk104_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.h deleted file mode 100644 index f2d12ae73944..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk104.h +++ /dev/null @@ -1,168 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __GK104_FIFO_H__ -#define __GK104_FIFO_H__ -#define gk104_fifo(p) container_of((p), struct gk104_fifo, base) -#include "priv.h" -struct nvkm_fifo_cgrp; - -#include <core/enum.h> -#include <subdev/mmu.h> - -struct gk104_fifo_chan; -struct gk104_fifo { - const struct gk104_fifo_func *func; - struct nvkm_fifo base; - - struct { - struct work_struct work; - u32 engm; - u32 runm; - } recover; - - int pbdma_nr; - - struct { - struct nvkm_engine *engine; - int runl; - int pbid; - } engine[16]; - int engine_nr; - - struct { - struct nvkm_memory *mem[2]; - int next; - wait_queue_head_t wait; - struct list_head cgrp; - struct list_head chan; - u32 engm; - u32 engm_sw; - } runlist[16]; - int runlist_nr; - - struct { - struct nvkm_memory *mem; - struct nvkm_vma *bar; - } user; -}; - -struct gk104_fifo_func { - struct { - void (*fault)(struct nvkm_fifo *, int unit); - } intr; - - const struct gk104_fifo_pbdma_func { - int (*nr)(struct gk104_fifo *); - void (*init)(struct gk104_fifo *); - void (*init_timeout)(struct gk104_fifo *); - } *pbdma; - - struct { - const struct nvkm_enum *access; - const struct nvkm_enum *engine; - const struct nvkm_enum *reason; - const struct nvkm_enum *hubclient; - const struct nvkm_enum *gpcclient; - } fault; - - const struct gk104_fifo_runlist_func { - u8 size; - void (*cgrp)(struct nvkm_fifo_cgrp *, - struct nvkm_memory *, u32 offset); - void (*chan)(struct gk104_fifo_chan *, - struct nvkm_memory *, u32 offset); - void (*commit)(struct gk104_fifo *, int runl, - struct nvkm_memory *, int entries); - } *runlist; - - struct gk104_fifo_user_user { - struct nvkm_sclass user; - int (*ctor)(const struct nvkm_oclass *, void *, u32, - struct nvkm_object **); - } user; - - struct gk104_fifo_chan_user { - struct nvkm_sclass user; - int (*ctor)(struct gk104_fifo *, const struct nvkm_oclass *, - void *, u32, struct nvkm_object **); - } chan; - bool cgrp_force; -}; - -struct gk104_fifo_engine_status { - bool busy; - bool faulted; - bool chsw; - bool save; - bool load; - struct { - bool tsg; - u32 id; - } prev, next, *chan; -}; - -int gk104_fifo_new_(const struct gk104_fifo_func *, struct nvkm_device *, enum nvkm_subdev_type, - int index, int nr, struct nvkm_fifo **); -void gk104_fifo_runlist_insert(struct gk104_fifo *, struct gk104_fifo_chan *); -void gk104_fifo_runlist_remove(struct gk104_fifo *, struct gk104_fifo_chan *); -void gk104_fifo_runlist_update(struct gk104_fifo *, int runl); -void gk104_fifo_engine_status(struct gk104_fifo *fifo, int engn, - struct gk104_fifo_engine_status *status); -void gk104_fifo_intr_bind(struct gk104_fifo *fifo); -void gk104_fifo_intr_chsw(struct gk104_fifo *fifo); -void gk104_fifo_intr_dropped_fault(struct gk104_fifo *fifo); -void gk104_fifo_intr_pbdma_0(struct gk104_fifo *fifo, int unit); -void gk104_fifo_intr_pbdma_1(struct gk104_fifo *fifo, int unit); -void gk104_fifo_intr_runlist(struct gk104_fifo *fifo); -void gk104_fifo_intr_engine(struct gk104_fifo *fifo); -void *gk104_fifo_dtor(struct nvkm_fifo *base); -int gk104_fifo_oneinit(struct nvkm_fifo *base); -int gk104_fifo_info(struct nvkm_fifo *base, u64 mthd, u64 *data); -void gk104_fifo_init(struct nvkm_fifo *base); -void gk104_fifo_fini(struct nvkm_fifo *base); -int gk104_fifo_class_new(struct nvkm_fifo *base, const struct nvkm_oclass *oclass, - void *argv, u32 argc, struct nvkm_object **pobject); -int gk104_fifo_class_get(struct nvkm_fifo *base, int index, - struct nvkm_oclass *oclass); -void gk104_fifo_uevent_fini(struct nvkm_fifo *fifo); -void gk104_fifo_uevent_init(struct nvkm_fifo *fifo); - -extern const struct gk104_fifo_pbdma_func gk104_fifo_pbdma; -int gk104_fifo_pbdma_nr(struct gk104_fifo *); -void gk104_fifo_pbdma_init(struct gk104_fifo *); -extern const struct nvkm_enum gk104_fifo_fault_access[]; -extern const struct nvkm_enum gk104_fifo_fault_engine[]; -extern const struct nvkm_enum gk104_fifo_fault_reason[]; -extern const struct nvkm_enum gk104_fifo_fault_hubclient[]; -extern const struct nvkm_enum gk104_fifo_fault_gpcclient[]; -extern const struct gk104_fifo_runlist_func gk104_fifo_runlist; -void gk104_fifo_runlist_chan(struct gk104_fifo_chan *, - struct nvkm_memory *, u32); -void gk104_fifo_runlist_commit(struct gk104_fifo *, int runl, - struct nvkm_memory *, int); - -extern const struct gk104_fifo_runlist_func gk110_fifo_runlist; -void gk110_fifo_runlist_cgrp(struct nvkm_fifo_cgrp *, - struct nvkm_memory *, u32); - -extern const struct gk104_fifo_pbdma_func gk208_fifo_pbdma; -void gk208_fifo_pbdma_init_timeout(struct gk104_fifo *); - -void gm107_fifo_intr_fault(struct nvkm_fifo *, int); -extern const struct nvkm_enum gm107_fifo_fault_engine[]; -extern const struct gk104_fifo_runlist_func gm107_fifo_runlist; - -extern const struct gk104_fifo_pbdma_func gm200_fifo_pbdma; -int gm200_fifo_pbdma_nr(struct gk104_fifo *); - -void gp100_fifo_intr_fault(struct nvkm_fifo *, int); -extern const struct nvkm_enum gp100_fifo_fault_engine[]; - -extern const struct nvkm_enum gv100_fifo_fault_access[]; -extern const struct nvkm_enum gv100_fifo_fault_reason[]; -extern const struct nvkm_enum gv100_fifo_fault_hubclient[]; -extern const struct nvkm_enum gv100_fifo_fault_gpcclient[]; -void gv100_fifo_runlist_cgrp(struct nvkm_fifo_cgrp *, - struct nvkm_memory *, u32); -void gv100_fifo_runlist_chan(struct gk104_fifo_chan *, - struct nvkm_memory *, u32); -#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk110.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk110.c index 915278c7e012..a8ff21cf7712 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk110.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk110.c @@ -21,47 +21,112 @@ * * Authors: Ben Skeggs */ -#include "gk104.h" +#include "priv.h" #include "cgrp.h" -#include "changk104.h" +#include "chan.h" +#include "chid.h" +#include "runl.h" #include <core/memory.h> +#include <subdev/timer.h> #include <nvif/class.h> void -gk110_fifo_runlist_cgrp(struct nvkm_fifo_cgrp *cgrp, - struct nvkm_memory *memory, u32 offset) +gk110_chan_preempt(struct nvkm_chan *chan) +{ + struct nvkm_cgrp *cgrp = chan->cgrp; + + if (cgrp->hw) { + cgrp->func->preempt(cgrp); + return; + } + + gf100_chan_preempt(chan); +} + +const struct nvkm_chan_func +gk110_chan = { + .inst = &gf100_chan_inst, + .userd = &gk104_chan_userd, + .ramfc = &gk104_chan_ramfc, + .bind = gk104_chan_bind, + .unbind = gk104_chan_unbind, + .start = gk104_chan_start, + .stop = gk104_chan_stop, + .preempt = gk110_chan_preempt, +}; + +static void +gk110_cgrp_preempt(struct nvkm_cgrp *cgrp) +{ + nvkm_wr32(cgrp->runl->fifo->engine.subdev.device, 0x002634, 0x01000000 | cgrp->id); +} + +const struct nvkm_cgrp_func +gk110_cgrp = { + .preempt = gk110_cgrp_preempt, +}; + +void +gk110_runl_insert_cgrp(struct nvkm_cgrp *cgrp, struct nvkm_memory *memory, u64 offset) { nvkm_wo32(memory, offset + 0, (cgrp->chan_nr << 26) | (128 << 18) | (3 << 14) | 0x00002000 | cgrp->id); nvkm_wo32(memory, offset + 4, 0x00000000); } -const struct gk104_fifo_runlist_func -gk110_fifo_runlist = { +const struct nvkm_runl_func +gk110_runl = { .size = 8, - .cgrp = gk110_fifo_runlist_cgrp, - .chan = gk104_fifo_runlist_chan, - .commit = gk104_fifo_runlist_commit, + .update = nv50_runl_update, + .insert_cgrp = gk110_runl_insert_cgrp, + .insert_chan = gk104_runl_insert_chan, + .commit = gk104_runl_commit, + .wait = nv50_runl_wait, + .pending = gk104_runl_pending, + .block = gk104_runl_block, + .allow = gk104_runl_allow, + .fault_clear = gk104_runl_fault_clear, + .preempt_pending = gf100_runl_preempt_pending, }; -static const struct gk104_fifo_func +int +gk110_fifo_chid_ctor(struct nvkm_fifo *fifo, int nr) +{ + int ret; + + ret = nvkm_chid_new(&nvkm_chan_event, &fifo->engine.subdev, nr, 0, nr, &fifo->cgid); + if (ret) + return ret; + + return gf100_fifo_chid_ctor(fifo, nr); +} + +static const struct nvkm_fifo_func gk110_fifo = { - .intr.fault = gf100_fifo_intr_fault, - .pbdma = &gk104_fifo_pbdma, - .fault.access = gk104_fifo_fault_access, - .fault.engine = gk104_fifo_fault_engine, - .fault.reason = gk104_fifo_fault_reason, - .fault.hubclient = gk104_fifo_fault_hubclient, - .fault.gpcclient = gk104_fifo_fault_gpcclient, - .runlist = &gk110_fifo_runlist, - .chan = {{0,0,KEPLER_CHANNEL_GPFIFO_B}, gk104_fifo_gpfifo_new }, + .chid_nr = gk104_fifo_chid_nr, + .chid_ctor = gk110_fifo_chid_ctor, + .runq_nr = gf100_fifo_runq_nr, + .runl_ctor = gk104_fifo_runl_ctor, + .init = gk104_fifo_init, + .init_pbdmas = gk104_fifo_init_pbdmas, + .intr = gk104_fifo_intr, + .intr_mmu_fault_unit = gf100_fifo_intr_mmu_fault_unit, + .intr_ctxsw_timeout = gf100_fifo_intr_ctxsw_timeout, + .mmu_fault = &gk104_fifo_mmu_fault, + .nonstall = &gf100_fifo_nonstall, + .runl = &gk110_runl, + .runq = &gk104_runq, + .engn = &gk104_engn, + .engn_ce = &gk104_engn_ce, + .cgrp = {{ 0, 0, KEPLER_CHANNEL_GROUP_A }, &gk110_cgrp }, + .chan = {{ 0, 0, KEPLER_CHANNEL_GPFIFO_B }, &gk110_chan }, }; int gk110_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return gk104_fifo_new_(&gk110_fifo, device, type, inst, 4096, pfifo); + return nvkm_fifo_new_(&gk110_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk208.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk208.c index cb703693de52..8fa2b0be141a 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk208.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk208.c @@ -21,44 +21,57 @@ * * Authors: Ben Skeggs */ -#include "gk104.h" -#include "changk104.h" +#include "priv.h" +#include "runq.h" #include <nvif/class.h> void -gk208_fifo_pbdma_init_timeout(struct gk104_fifo *fifo) +gk208_runq_init(struct nvkm_runq *runq) { - struct nvkm_device *device = fifo->base.engine.subdev.device; - int i; + gk104_runq_init(runq); - for (i = 0; i < fifo->pbdma_nr; i++) - nvkm_wr32(device, 0x04012c + (i * 0x2000), 0x0000ffff); + nvkm_wr32(runq->fifo->engine.subdev.device, 0x04012c + (runq->id * 0x2000), 0x000f4240); } -const struct gk104_fifo_pbdma_func -gk208_fifo_pbdma = { - .nr = gk104_fifo_pbdma_nr, - .init = gk104_fifo_pbdma_init, - .init_timeout = gk208_fifo_pbdma_init_timeout, +const struct nvkm_runq_func +gk208_runq = { + .init = gk208_runq_init, + .intr = gk104_runq_intr, + .intr_0_names = gk104_runq_intr_0_names, + .idle = gk104_runq_idle, }; -static const struct gk104_fifo_func +static int +gk208_fifo_chid_nr(struct nvkm_fifo *fifo) +{ + return 1024; +} + +static const struct nvkm_fifo_func gk208_fifo = { - .intr.fault = gf100_fifo_intr_fault, - .pbdma = &gk208_fifo_pbdma, - .fault.access = gk104_fifo_fault_access, - .fault.engine = gk104_fifo_fault_engine, - .fault.reason = gk104_fifo_fault_reason, - .fault.hubclient = gk104_fifo_fault_hubclient, - .fault.gpcclient = gk104_fifo_fault_gpcclient, - .runlist = &gk110_fifo_runlist, - .chan = {{0,0,KEPLER_CHANNEL_GPFIFO_A}, gk104_fifo_gpfifo_new }, + .chid_nr = gk208_fifo_chid_nr, + .chid_ctor = gk110_fifo_chid_ctor, + .runq_nr = gf100_fifo_runq_nr, + .runl_ctor = gk104_fifo_runl_ctor, + .init = gk104_fifo_init, + .init_pbdmas = gk104_fifo_init_pbdmas, + .intr = gk104_fifo_intr, + .intr_mmu_fault_unit = gf100_fifo_intr_mmu_fault_unit, + .intr_ctxsw_timeout = gf100_fifo_intr_ctxsw_timeout, + .mmu_fault = &gk104_fifo_mmu_fault, + .nonstall = &gf100_fifo_nonstall, + .runl = &gk110_runl, + .runq = &gk208_runq, + .engn = &gk104_engn, + .engn_ce = &gk104_engn_ce, + .cgrp = {{ 0, 0, KEPLER_CHANNEL_GROUP_A }, &gk110_cgrp }, + .chan = {{ 0, 0, KEPLER_CHANNEL_GPFIFO_A }, &gk110_chan }, }; int gk208_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return gk104_fifo_new_(&gk208_fifo, device, type, inst, 1024, pfifo); + return nvkm_fifo_new_(&gk208_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk20a.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk20a.c index 6e35cf44c640..b63ca836130f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk20a.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gk20a.c @@ -19,27 +19,34 @@ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER * DEALINGS IN THE SOFTWARE. */ -#include "gk104.h" -#include "changk104.h" +#include "priv.h" #include <nvif/class.h> -static const struct gk104_fifo_func +static const struct nvkm_fifo_func gk20a_fifo = { - .intr.fault = gf100_fifo_intr_fault, - .pbdma = &gk208_fifo_pbdma, - .fault.access = gk104_fifo_fault_access, - .fault.engine = gk104_fifo_fault_engine, - .fault.reason = gk104_fifo_fault_reason, - .fault.hubclient = gk104_fifo_fault_hubclient, - .fault.gpcclient = gk104_fifo_fault_gpcclient, - .runlist = &gk110_fifo_runlist, - .chan = {{0,0,KEPLER_CHANNEL_GPFIFO_A}, gk104_fifo_gpfifo_new }, + .chid_nr = nv50_fifo_chid_nr, + .chid_ctor = gk110_fifo_chid_ctor, + .runq_nr = gf100_fifo_runq_nr, + .runl_ctor = gk104_fifo_runl_ctor, + .init = gk104_fifo_init, + .init_pbdmas = gk104_fifo_init_pbdmas, + .intr = gk104_fifo_intr, + .intr_mmu_fault_unit = gf100_fifo_intr_mmu_fault_unit, + .intr_ctxsw_timeout = gf100_fifo_intr_ctxsw_timeout, + .mmu_fault = &gk104_fifo_mmu_fault, + .nonstall = &gf100_fifo_nonstall, + .runl = &gk110_runl, + .runq = &gk208_runq, + .engn = &gk104_engn, + .engn_ce = &gk104_engn_ce, + .cgrp = {{ }, &gk110_cgrp }, + .chan = {{ 0, 0, KEPLER_CHANNEL_GPFIFO_A }, &gk110_chan }, }; int gk20a_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return gk104_fifo_new_(&gk20a_fifo, device, type, inst, 128, pfifo); + return nvkm_fifo_new_(&gk20a_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gm107.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gm107.c index 7af6e687d474..5ba60021b510 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gm107.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gm107.c @@ -21,46 +21,65 @@ * * Authors: Ben Skeggs */ -#include "gk104.h" -#include "changk104.h" +#include "priv.h" +#include "chan.h" +#include "runl.h" #include <core/gpuobj.h> #include <subdev/fault.h> #include <nvif/class.h> +const struct nvkm_chan_func +gm107_chan = { + .inst = &gf100_chan_inst, + .userd = &gk104_chan_userd, + .ramfc = &gk104_chan_ramfc, + .bind = gk104_chan_bind_inst, + .unbind = gk104_chan_unbind, + .start = gk104_chan_start, + .stop = gk104_chan_stop, + .preempt = gk110_chan_preempt, +}; + static void -gm107_fifo_runlist_chan(struct gk104_fifo_chan *chan, - struct nvkm_memory *memory, u32 offset) +gm107_runl_insert_chan(struct nvkm_chan *chan, struct nvkm_memory *memory, u64 offset) { - nvkm_wo32(memory, offset + 0, chan->base.chid); - nvkm_wo32(memory, offset + 4, chan->base.inst->addr >> 12); + nvkm_wo32(memory, offset + 0, chan->id); + nvkm_wo32(memory, offset + 4, chan->inst->addr >> 12); } -const struct gk104_fifo_runlist_func -gm107_fifo_runlist = { +const struct nvkm_runl_func +gm107_runl = { .size = 8, - .cgrp = gk110_fifo_runlist_cgrp, - .chan = gm107_fifo_runlist_chan, - .commit = gk104_fifo_runlist_commit, + .update = nv50_runl_update, + .insert_cgrp = gk110_runl_insert_cgrp, + .insert_chan = gm107_runl_insert_chan, + .commit = gk104_runl_commit, + .wait = nv50_runl_wait, + .pending = gk104_runl_pending, + .block = gk104_runl_block, + .allow = gk104_runl_allow, + .fault_clear = gk104_runl_fault_clear, + .preempt_pending = gf100_runl_preempt_pending, }; -const struct nvkm_enum -gm107_fifo_fault_engine[] = { +static const struct nvkm_enum +gm107_fifo_mmu_fault_engine[] = { { 0x01, "DISPLAY" }, { 0x02, "CAPTURE" }, { 0x03, "IFB", NULL, NVKM_ENGINE_IFB }, { 0x04, "BAR1", NULL, NVKM_SUBDEV_BAR }, { 0x05, "BAR2", NULL, NVKM_SUBDEV_INSTMEM }, { 0x06, "SCHED" }, - { 0x07, "HOST0", NULL, NVKM_ENGINE_FIFO }, - { 0x08, "HOST1", NULL, NVKM_ENGINE_FIFO }, - { 0x09, "HOST2", NULL, NVKM_ENGINE_FIFO }, - { 0x0a, "HOST3", NULL, NVKM_ENGINE_FIFO }, - { 0x0b, "HOST4", NULL, NVKM_ENGINE_FIFO }, - { 0x0c, "HOST5", NULL, NVKM_ENGINE_FIFO }, - { 0x0d, "HOST6", NULL, NVKM_ENGINE_FIFO }, - { 0x0e, "HOST7", NULL, NVKM_ENGINE_FIFO }, + { 0x07, "HOST0" }, + { 0x08, "HOST1" }, + { 0x09, "HOST2" }, + { 0x0a, "HOST3" }, + { 0x0b, "HOST4" }, + { 0x0c, "HOST5" }, + { 0x0d, "HOST6" }, + { 0x0e, "HOST7" }, { 0x0f, "HOSTSR" }, { 0x13, "PERF" }, { 0x17, "PMU" }, @@ -68,8 +87,18 @@ gm107_fifo_fault_engine[] = { {} }; +const struct nvkm_fifo_func_mmu_fault +gm107_fifo_mmu_fault = { + .recover = gf100_fifo_mmu_fault_recover, + .access = gf100_fifo_mmu_fault_access, + .engine = gm107_fifo_mmu_fault_engine, + .reason = gk104_fifo_mmu_fault_reason, + .hubclient = gk104_fifo_mmu_fault_hubclient, + .gpcclient = gk104_fifo_mmu_fault_gpcclient, +}; + void -gm107_fifo_intr_fault(struct nvkm_fifo *fifo, int unit) +gm107_fifo_intr_mmu_fault_unit(struct nvkm_fifo *fifo, int unit) { struct nvkm_device *device = fifo->engine.subdev.device; u32 inst = nvkm_rd32(device, 0x002800 + (unit * 0x10)); @@ -92,22 +121,36 @@ gm107_fifo_intr_fault(struct nvkm_fifo *fifo, int unit) nvkm_fifo_fault(fifo, &info); } -static const struct gk104_fifo_func +static int +gm107_fifo_chid_nr(struct nvkm_fifo *fifo) +{ + return 2048; +} + +static const struct nvkm_fifo_func gm107_fifo = { - .intr.fault = gm107_fifo_intr_fault, - .pbdma = &gk208_fifo_pbdma, - .fault.access = gk104_fifo_fault_access, - .fault.engine = gm107_fifo_fault_engine, - .fault.reason = gk104_fifo_fault_reason, - .fault.hubclient = gk104_fifo_fault_hubclient, - .fault.gpcclient = gk104_fifo_fault_gpcclient, - .runlist = &gm107_fifo_runlist, - .chan = {{0,0,KEPLER_CHANNEL_GPFIFO_B}, gk104_fifo_gpfifo_new }, + .chid_nr = gm107_fifo_chid_nr, + .chid_ctor = gk110_fifo_chid_ctor, + .runq_nr = gf100_fifo_runq_nr, + .runl_ctor = gk104_fifo_runl_ctor, + .init = gk104_fifo_init, + .init_pbdmas = gk104_fifo_init_pbdmas, + .intr = gk104_fifo_intr, + .intr_mmu_fault_unit = gm107_fifo_intr_mmu_fault_unit, + .intr_ctxsw_timeout = gf100_fifo_intr_ctxsw_timeout, + .mmu_fault = &gm107_fifo_mmu_fault, + .nonstall = &gf100_fifo_nonstall, + .runl = &gm107_runl, + .runq = &gk208_runq, + .engn = &gk104_engn, + .engn_ce = &gk104_engn_ce, + .cgrp = {{ 0, 0, KEPLER_CHANNEL_GROUP_A }, &gk110_cgrp }, + .chan = {{ 0, 0, KEPLER_CHANNEL_GPFIFO_B }, &gm107_chan }, }; int gm107_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return gk104_fifo_new_(&gm107_fifo, device, type, inst, 2048, pfifo); + return nvkm_fifo_new_(&gm107_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gm200.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gm200.c index 573658cb6c73..d92d1ac39191 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gm200.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gm200.c @@ -21,41 +21,46 @@ * * Authors: Ben Skeggs */ -#include "gk104.h" -#include "changk104.h" +#include "priv.h" #include <nvif/class.h> int -gm200_fifo_pbdma_nr(struct gk104_fifo *fifo) +gm200_fifo_runq_nr(struct nvkm_fifo *fifo) { - struct nvkm_device *device = fifo->base.engine.subdev.device; - return nvkm_rd32(device, 0x002004) & 0x000000ff; + return nvkm_rd32(fifo->engine.subdev.device, 0x002004) & 0x000000ff; } -const struct gk104_fifo_pbdma_func -gm200_fifo_pbdma = { - .nr = gm200_fifo_pbdma_nr, - .init = gk104_fifo_pbdma_init, - .init_timeout = gk208_fifo_pbdma_init_timeout, -}; +int +gm200_fifo_chid_nr(struct nvkm_fifo *fifo) +{ + return nvkm_rd32(fifo->engine.subdev.device, 0x002008); +} -static const struct gk104_fifo_func +static const struct nvkm_fifo_func gm200_fifo = { - .intr.fault = gm107_fifo_intr_fault, - .pbdma = &gm200_fifo_pbdma, - .fault.access = gk104_fifo_fault_access, - .fault.engine = gm107_fifo_fault_engine, - .fault.reason = gk104_fifo_fault_reason, - .fault.hubclient = gk104_fifo_fault_hubclient, - .fault.gpcclient = gk104_fifo_fault_gpcclient, - .runlist = &gm107_fifo_runlist, - .chan = {{0,0,MAXWELL_CHANNEL_GPFIFO_A}, gk104_fifo_gpfifo_new }, + .chid_nr = gm200_fifo_chid_nr, + .chid_ctor = gk110_fifo_chid_ctor, + .runq_nr = gm200_fifo_runq_nr, + .runl_ctor = gk104_fifo_runl_ctor, + .init = gk104_fifo_init, + .init_pbdmas = gk104_fifo_init_pbdmas, + .intr = gk104_fifo_intr, + .intr_mmu_fault_unit = gm107_fifo_intr_mmu_fault_unit, + .intr_ctxsw_timeout = gf100_fifo_intr_ctxsw_timeout, + .mmu_fault = &gm107_fifo_mmu_fault, + .nonstall = &gf100_fifo_nonstall, + .runl = &gm107_runl, + .runq = &gk208_runq, + .engn = &gk104_engn, + .engn_ce = &gk104_engn_ce, + .cgrp = {{ 0, 0, KEPLER_CHANNEL_GROUP_A }, &gk110_cgrp }, + .chan = {{ 0, 0, MAXWELL_CHANNEL_GPFIFO_A }, &gm107_chan }, }; int gm200_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return gk104_fifo_new_(&gm200_fifo, device, type, inst, 4096, pfifo); + return nvkm_fifo_new_(&gm200_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gm20b.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gm20b.c deleted file mode 100644 index 556c97e54f14..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gm20b.c +++ /dev/null @@ -1,45 +0,0 @@ -/* - * Copyright (c) 2015, NVIDIA CORPORATION. All rights reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER - * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING - * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER - * DEALINGS IN THE SOFTWARE. - */ -#include "gk104.h" -#include "changk104.h" - -#include <nvif/class.h> - -static const struct gk104_fifo_func -gm20b_fifo = { - .intr.fault = gm107_fifo_intr_fault, - .pbdma = &gm200_fifo_pbdma, - .fault.access = gk104_fifo_fault_access, - .fault.engine = gm107_fifo_fault_engine, - .fault.reason = gk104_fifo_fault_reason, - .fault.hubclient = gk104_fifo_fault_hubclient, - .fault.gpcclient = gk104_fifo_fault_gpcclient, - .runlist = &gm107_fifo_runlist, - .chan = {{0,0,MAXWELL_CHANNEL_GPFIFO_A}, gk104_fifo_gpfifo_new }, -}; - -int -gm20b_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, - struct nvkm_fifo **pfifo) -{ - return gk104_fifo_new_(&gm20b_fifo, device, type, inst, 512, pfifo); -} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gp100.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gp100.c index 6b46b6b65b87..65bdb6a7d517 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gp100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gp100.c @@ -21,30 +21,54 @@ * * Authors: Ben Skeggs */ -#include "gk104.h" -#include "changk104.h" +#include "priv.h" +#include "runl.h" +#include <core/gpuobj.h> #include <subdev/fault.h> #include <nvif/class.h> -const struct nvkm_enum -gp100_fifo_fault_engine[] = { +static void +gp100_runl_insert_chan(struct nvkm_chan *chan, struct nvkm_memory *memory, u64 offset) +{ + nvkm_wo32(memory, offset + 0, chan->id | chan->runq << 14); + nvkm_wo32(memory, offset + 4, chan->inst->addr >> 12); +} + +static const struct nvkm_runl_func +gp100_runl = { + .runqs = 2, + .size = 8, + .update = nv50_runl_update, + .insert_cgrp = gk110_runl_insert_cgrp, + .insert_chan = gp100_runl_insert_chan, + .commit = gk104_runl_commit, + .wait = nv50_runl_wait, + .pending = gk104_runl_pending, + .block = gk104_runl_block, + .allow = gk104_runl_allow, + .fault_clear = gk104_runl_fault_clear, + .preempt_pending = gf100_runl_preempt_pending, +}; + +static const struct nvkm_enum +gp100_fifo_mmu_fault_engine[] = { { 0x01, "DISPLAY" }, { 0x03, "IFB", NULL, NVKM_ENGINE_IFB }, { 0x04, "BAR1", NULL, NVKM_SUBDEV_BAR }, { 0x05, "BAR2", NULL, NVKM_SUBDEV_INSTMEM }, - { 0x06, "HOST0", NULL, NVKM_ENGINE_FIFO }, - { 0x07, "HOST1", NULL, NVKM_ENGINE_FIFO }, - { 0x08, "HOST2", NULL, NVKM_ENGINE_FIFO }, - { 0x09, "HOST3", NULL, NVKM_ENGINE_FIFO }, - { 0x0a, "HOST4", NULL, NVKM_ENGINE_FIFO }, - { 0x0b, "HOST5", NULL, NVKM_ENGINE_FIFO }, - { 0x0c, "HOST6", NULL, NVKM_ENGINE_FIFO }, - { 0x0d, "HOST7", NULL, NVKM_ENGINE_FIFO }, - { 0x0e, "HOST8", NULL, NVKM_ENGINE_FIFO }, - { 0x0f, "HOST9", NULL, NVKM_ENGINE_FIFO }, - { 0x10, "HOST10", NULL, NVKM_ENGINE_FIFO }, + { 0x06, "HOST0" }, + { 0x07, "HOST1" }, + { 0x08, "HOST2" }, + { 0x09, "HOST3" }, + { 0x0a, "HOST4" }, + { 0x0b, "HOST5" }, + { 0x0c, "HOST6" }, + { 0x0d, "HOST7" }, + { 0x0e, "HOST8" }, + { 0x0f, "HOST9" }, + { 0x10, "HOST10" }, { 0x13, "PERF" }, { 0x17, "PMU" }, { 0x18, "PTP" }, @@ -52,8 +76,18 @@ gp100_fifo_fault_engine[] = { {} }; -void -gp100_fifo_intr_fault(struct nvkm_fifo *fifo, int unit) +static const struct nvkm_fifo_func_mmu_fault +gp100_fifo_mmu_fault = { + .recover = gf100_fifo_mmu_fault_recover, + .access = gf100_fifo_mmu_fault_access, + .engine = gp100_fifo_mmu_fault_engine, + .reason = gk104_fifo_mmu_fault_reason, + .hubclient = gk104_fifo_mmu_fault_hubclient, + .gpcclient = gk104_fifo_mmu_fault_gpcclient, +}; + +static void +gp100_fifo_intr_mmu_fault_unit(struct nvkm_fifo *fifo, int unit) { struct nvkm_device *device = fifo->engine.subdev.device; u32 inst = nvkm_rd32(device, 0x002800 + (unit * 0x10)); @@ -76,23 +110,30 @@ gp100_fifo_intr_fault(struct nvkm_fifo *fifo, int unit) nvkm_fifo_fault(fifo, &info); } -static const struct gk104_fifo_func +static const struct nvkm_fifo_func gp100_fifo = { - .intr.fault = gp100_fifo_intr_fault, - .pbdma = &gm200_fifo_pbdma, - .fault.access = gk104_fifo_fault_access, - .fault.engine = gp100_fifo_fault_engine, - .fault.reason = gk104_fifo_fault_reason, - .fault.hubclient = gk104_fifo_fault_hubclient, - .fault.gpcclient = gk104_fifo_fault_gpcclient, - .runlist = &gm107_fifo_runlist, - .chan = {{0,0,PASCAL_CHANNEL_GPFIFO_A}, gk104_fifo_gpfifo_new }, - .cgrp_force = true, + .chid_nr = gm200_fifo_chid_nr, + .chid_ctor = gk110_fifo_chid_ctor, + .runq_nr = gm200_fifo_runq_nr, + .runl_ctor = gk104_fifo_runl_ctor, + .init = gk104_fifo_init, + .init_pbdmas = gk104_fifo_init_pbdmas, + .intr = gk104_fifo_intr, + .intr_mmu_fault_unit = gp100_fifo_intr_mmu_fault_unit, + .intr_ctxsw_timeout = gf100_fifo_intr_ctxsw_timeout, + .mmu_fault = &gp100_fifo_mmu_fault, + .nonstall = &gf100_fifo_nonstall, + .runl = &gp100_runl, + .runq = &gk208_runq, + .engn = &gk104_engn, + .engn_ce = &gk104_engn_ce, + .cgrp = {{ 0, 0, KEPLER_CHANNEL_GROUP_A }, &gk110_cgrp, .force = true }, + .chan = {{ 0, 0, PASCAL_CHANNEL_GPFIFO_A }, &gm107_chan }, }; int gp100_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return gk104_fifo_new_(&gp100_fifo, device, type, inst, 4096, pfifo); + return nvkm_fifo_new_(&gp100_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gp10b.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gp10b.c deleted file mode 100644 index 7a5929cb4d29..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gp10b.c +++ /dev/null @@ -1,46 +0,0 @@ -/* - * Copyright (c) 2017, NVIDIA CORPORATION. All rights reserved. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER - * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING - * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER - * DEALINGS IN THE SOFTWARE. - */ -#include "gk104.h" -#include "changk104.h" - -#include <nvif/class.h> - -static const struct gk104_fifo_func -gp10b_fifo = { - .intr.fault = gp100_fifo_intr_fault, - .pbdma = &gm200_fifo_pbdma, - .fault.access = gk104_fifo_fault_access, - .fault.engine = gp100_fifo_fault_engine, - .fault.reason = gk104_fifo_fault_reason, - .fault.hubclient = gk104_fifo_fault_hubclient, - .fault.gpcclient = gk104_fifo_fault_gpcclient, - .runlist = &gm107_fifo_runlist, - .chan = {{0,0,PASCAL_CHANNEL_GPFIFO_A}, gk104_fifo_gpfifo_new }, - .cgrp_force = true, -}; - -int -gp10b_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, - struct nvkm_fifo **pfifo) -{ - return gk104_fifo_new_(&gp10b_fifo, device, type, inst, 512, pfifo); -} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifog84.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifog84.c deleted file mode 100644 index 2121f517b1dd..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifog84.c +++ /dev/null @@ -1,95 +0,0 @@ -/* - * Copyright 2012 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#include "channv50.h" - -#include <core/client.h> -#include <core/ramht.h> - -#include <nvif/class.h> -#include <nvif/cl826f.h> -#include <nvif/unpack.h> - -static int -g84_fifo_gpfifo_new(struct nvkm_fifo *base, const struct nvkm_oclass *oclass, - void *data, u32 size, struct nvkm_object **pobject) -{ - struct nvkm_object *parent = oclass->parent; - union { - struct g82_channel_gpfifo_v0 v0; - } *args = data; - struct nv50_fifo *fifo = nv50_fifo(base); - struct nv50_fifo_chan *chan; - u64 ioffset, ilength; - int ret = -ENOSYS; - - nvif_ioctl(parent, "create channel gpfifo size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(parent, "create channel gpfifo vers %d vmm %llx " - "pushbuf %llx ioffset %016llx " - "ilength %08x\n", - args->v0.version, args->v0.vmm, args->v0.pushbuf, - args->v0.ioffset, args->v0.ilength); - if (!args->v0.pushbuf) - return -EINVAL; - } else - return ret; - - if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) - return -ENOMEM; - *pobject = &chan->base.object; - - ret = g84_fifo_chan_ctor(fifo, args->v0.vmm, args->v0.pushbuf, - oclass, chan); - if (ret) - return ret; - - args->v0.chid = chan->base.chid; - ioffset = args->v0.ioffset; - ilength = order_base_2(args->v0.ilength / 8); - - nvkm_kmap(chan->ramfc); - nvkm_wo32(chan->ramfc, 0x3c, 0x403f6078); - nvkm_wo32(chan->ramfc, 0x44, 0x01003fff); - nvkm_wo32(chan->ramfc, 0x48, chan->base.push->node->offset >> 4); - nvkm_wo32(chan->ramfc, 0x50, lower_32_bits(ioffset)); - nvkm_wo32(chan->ramfc, 0x54, upper_32_bits(ioffset) | (ilength << 16)); - nvkm_wo32(chan->ramfc, 0x60, 0x7fffffff); - nvkm_wo32(chan->ramfc, 0x78, 0x00000000); - nvkm_wo32(chan->ramfc, 0x7c, 0x30000001); - nvkm_wo32(chan->ramfc, 0x80, ((chan->ramht->bits - 9) << 27) | - (4 << 24) /* SEARCH_FULL */ | - (chan->ramht->gpuobj->node->offset >> 4)); - nvkm_wo32(chan->ramfc, 0x88, chan->cache->addr >> 10); - nvkm_wo32(chan->ramfc, 0x98, chan->base.inst->addr >> 12); - nvkm_done(chan->ramfc); - return 0; -} - -const struct nvkm_fifo_chan_oclass -g84_fifo_gpfifo_oclass = { - .base.oclass = G82_CHANNEL_GPFIFO, - .base.minver = 0, - .base.maxver = 0, - .ctor = g84_fifo_gpfifo_new, -}; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogf100.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogf100.c deleted file mode 100644 index 4e78bbe3b94b..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogf100.c +++ /dev/null @@ -1,308 +0,0 @@ -/* - * Copyright 2012 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#include "changf100.h" - -#include <core/client.h> -#include <core/gpuobj.h> -#include <subdev/fb.h> -#include <subdev/timer.h> - -#include <nvif/class.h> -#include <nvif/cl906f.h> -#include <nvif/unpack.h> - -int -gf100_fifo_chan_ntfy(struct nvkm_fifo_chan *chan, u32 type, - struct nvkm_event **pevent) -{ - switch (type) { - case NV906F_V0_NTFY_NON_STALL_INTERRUPT: - *pevent = &chan->fifo->uevent; - return 0; - case NV906F_V0_NTFY_KILLED: - *pevent = &chan->fifo->kevent; - return 0; - default: - break; - } - return -EINVAL; -} - -static u32 -gf100_fifo_gpfifo_engine_addr(struct nvkm_engine *engine) -{ - switch (engine->subdev.type) { - case NVKM_ENGINE_SW : return 0; - case NVKM_ENGINE_GR : return 0x0210; - case NVKM_ENGINE_CE : return 0x0230 + (engine->subdev.inst * 0x10); - case NVKM_ENGINE_MSPDEC: return 0x0250; - case NVKM_ENGINE_MSPPP : return 0x0260; - case NVKM_ENGINE_MSVLD : return 0x0270; - default: - WARN_ON(1); - return 0; - } -} - -static struct gf100_fifo_engn * -gf100_fifo_gpfifo_engine(struct gf100_fifo_chan *chan, struct nvkm_engine *engine) -{ - int engi = chan->base.fifo->func->engine_id(chan->base.fifo, engine); - if (engi >= 0) - return &chan->engn[engi]; - return NULL; -} - -static int -gf100_fifo_gpfifo_engine_fini(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine, bool suspend) -{ - const u32 offset = gf100_fifo_gpfifo_engine_addr(engine); - struct gf100_fifo_chan *chan = gf100_fifo_chan(base); - struct nvkm_subdev *subdev = &chan->fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - struct nvkm_gpuobj *inst = chan->base.inst; - int ret = 0; - - mutex_lock(&chan->fifo->base.mutex); - nvkm_wr32(device, 0x002634, chan->base.chid); - if (nvkm_msec(device, 2000, - if (nvkm_rd32(device, 0x002634) == chan->base.chid) - break; - ) < 0) { - nvkm_error(subdev, "channel %d [%s] kick timeout\n", - chan->base.chid, chan->base.object.client->name); - ret = -ETIMEDOUT; - } - mutex_unlock(&chan->fifo->base.mutex); - - if (ret && suspend) - return ret; - - if (offset) { - nvkm_kmap(inst); - nvkm_wo32(inst, offset + 0x00, 0x00000000); - nvkm_wo32(inst, offset + 0x04, 0x00000000); - nvkm_done(inst); - } - - return ret; -} - -static int -gf100_fifo_gpfifo_engine_init(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine) -{ - const u32 offset = gf100_fifo_gpfifo_engine_addr(engine); - struct gf100_fifo_chan *chan = gf100_fifo_chan(base); - struct gf100_fifo_engn *engn = gf100_fifo_gpfifo_engine(chan, engine); - struct nvkm_gpuobj *inst = chan->base.inst; - - if (offset) { - nvkm_kmap(inst); - nvkm_wo32(inst, offset + 0x00, lower_32_bits(engn->vma->addr) | 4); - nvkm_wo32(inst, offset + 0x04, upper_32_bits(engn->vma->addr)); - nvkm_done(inst); - } - - return 0; -} - -static void -gf100_fifo_gpfifo_engine_dtor(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine) -{ - struct gf100_fifo_chan *chan = gf100_fifo_chan(base); - struct gf100_fifo_engn *engn = gf100_fifo_gpfifo_engine(chan, engine); - nvkm_vmm_put(chan->base.vmm, &engn->vma); - nvkm_gpuobj_del(&engn->inst); -} - -static int -gf100_fifo_gpfifo_engine_ctor(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine, - struct nvkm_object *object) -{ - struct gf100_fifo_chan *chan = gf100_fifo_chan(base); - struct gf100_fifo_engn *engn = gf100_fifo_gpfifo_engine(chan, engine); - int ret; - - if (!gf100_fifo_gpfifo_engine_addr(engine)) - return 0; - - ret = nvkm_object_bind(object, NULL, 0, &engn->inst); - if (ret) - return ret; - - ret = nvkm_vmm_get(chan->base.vmm, 12, engn->inst->size, &engn->vma); - if (ret) - return ret; - - return nvkm_memory_map(engn->inst, 0, chan->base.vmm, engn->vma, NULL, 0); -} - -static void -gf100_fifo_gpfifo_fini(struct nvkm_fifo_chan *base) -{ - struct gf100_fifo_chan *chan = gf100_fifo_chan(base); - struct gf100_fifo *fifo = chan->fifo; - struct nvkm_device *device = fifo->base.engine.subdev.device; - u32 coff = chan->base.chid * 8; - - if (!list_empty(&chan->head) && !chan->killed) { - gf100_fifo_runlist_remove(fifo, chan); - nvkm_mask(device, 0x003004 + coff, 0x00000001, 0x00000000); - gf100_fifo_runlist_commit(fifo); - } - - gf100_fifo_intr_engine(fifo); - - nvkm_wr32(device, 0x003000 + coff, 0x00000000); -} - -static void -gf100_fifo_gpfifo_init(struct nvkm_fifo_chan *base) -{ - struct gf100_fifo_chan *chan = gf100_fifo_chan(base); - struct gf100_fifo *fifo = chan->fifo; - struct nvkm_device *device = fifo->base.engine.subdev.device; - u32 addr = chan->base.inst->addr >> 12; - u32 coff = chan->base.chid * 8; - - nvkm_wr32(device, 0x003000 + coff, 0xc0000000 | addr); - - if (list_empty(&chan->head) && !chan->killed) { - gf100_fifo_runlist_insert(fifo, chan); - nvkm_wr32(device, 0x003004 + coff, 0x001f0001); - gf100_fifo_runlist_commit(fifo); - } -} - -static void * -gf100_fifo_gpfifo_dtor(struct nvkm_fifo_chan *base) -{ - return gf100_fifo_chan(base); -} - -static const struct nvkm_fifo_chan_func -gf100_fifo_gpfifo_func = { - .dtor = gf100_fifo_gpfifo_dtor, - .init = gf100_fifo_gpfifo_init, - .fini = gf100_fifo_gpfifo_fini, - .ntfy = gf100_fifo_chan_ntfy, - .engine_ctor = gf100_fifo_gpfifo_engine_ctor, - .engine_dtor = gf100_fifo_gpfifo_engine_dtor, - .engine_init = gf100_fifo_gpfifo_engine_init, - .engine_fini = gf100_fifo_gpfifo_engine_fini, -}; - -static int -gf100_fifo_gpfifo_new(struct nvkm_fifo *base, const struct nvkm_oclass *oclass, - void *data, u32 size, struct nvkm_object **pobject) -{ - union { - struct fermi_channel_gpfifo_v0 v0; - } *args = data; - struct gf100_fifo *fifo = gf100_fifo(base); - struct nvkm_object *parent = oclass->parent; - struct gf100_fifo_chan *chan; - u64 usermem, ioffset, ilength; - int ret = -ENOSYS, i; - - nvif_ioctl(parent, "create channel gpfifo size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(parent, "create channel gpfifo vers %d vmm %llx " - "ioffset %016llx ilength %08x\n", - args->v0.version, args->v0.vmm, args->v0.ioffset, - args->v0.ilength); - if (!args->v0.vmm) - return -EINVAL; - } else - return ret; - - /* allocate channel */ - if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) - return -ENOMEM; - *pobject = &chan->base.object; - chan->fifo = fifo; - INIT_LIST_HEAD(&chan->head); - - ret = nvkm_fifo_chan_ctor(&gf100_fifo_gpfifo_func, &fifo->base, - 0x1000, 0x1000, true, args->v0.vmm, 0, - BIT(GF100_FIFO_ENGN_GR) | - BIT(GF100_FIFO_ENGN_MSPDEC) | - BIT(GF100_FIFO_ENGN_MSPPP) | - BIT(GF100_FIFO_ENGN_MSVLD) | - BIT(GF100_FIFO_ENGN_CE0) | - BIT(GF100_FIFO_ENGN_CE1) | - BIT(GF100_FIFO_ENGN_SW), - 1, fifo->user.bar->addr, 0x1000, - oclass, &chan->base); - if (ret) - return ret; - - args->v0.chid = chan->base.chid; - - /* clear channel control registers */ - - usermem = chan->base.chid * 0x1000; - ioffset = args->v0.ioffset; - ilength = order_base_2(args->v0.ilength / 8); - - nvkm_kmap(fifo->user.mem); - for (i = 0; i < 0x1000; i += 4) - nvkm_wo32(fifo->user.mem, usermem + i, 0x00000000); - nvkm_done(fifo->user.mem); - usermem = nvkm_memory_addr(fifo->user.mem) + usermem; - - /* RAMFC */ - nvkm_kmap(chan->base.inst); - nvkm_wo32(chan->base.inst, 0x08, lower_32_bits(usermem)); - nvkm_wo32(chan->base.inst, 0x0c, upper_32_bits(usermem)); - nvkm_wo32(chan->base.inst, 0x10, 0x0000face); - nvkm_wo32(chan->base.inst, 0x30, 0xfffff902); - nvkm_wo32(chan->base.inst, 0x48, lower_32_bits(ioffset)); - nvkm_wo32(chan->base.inst, 0x4c, upper_32_bits(ioffset) | - (ilength << 16)); - nvkm_wo32(chan->base.inst, 0x54, 0x00000002); - nvkm_wo32(chan->base.inst, 0x84, 0x20400000); - nvkm_wo32(chan->base.inst, 0x94, 0x30000001); - nvkm_wo32(chan->base.inst, 0x9c, 0x00000100); - nvkm_wo32(chan->base.inst, 0xa4, 0x1f1f1f1f); - nvkm_wo32(chan->base.inst, 0xa8, 0x1f1f1f1f); - nvkm_wo32(chan->base.inst, 0xac, 0x0000001f); - nvkm_wo32(chan->base.inst, 0xb8, 0xf8000000); - nvkm_wo32(chan->base.inst, 0xf8, 0x10003080); /* 0x002310 */ - nvkm_wo32(chan->base.inst, 0xfc, 0x10000010); /* 0x002350 */ - nvkm_done(chan->base.inst); - return 0; -} - -const struct nvkm_fifo_chan_oclass -gf100_fifo_gpfifo_oclass = { - .base.oclass = FERMI_CHANNEL_GPFIFO, - .base.minver = 0, - .base.maxver = 0, - .ctor = gf100_fifo_gpfifo_new, -}; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogk104.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogk104.c deleted file mode 100644 index 80456ec70e8a..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogk104.c +++ /dev/null @@ -1,361 +0,0 @@ -/* - * Copyright 2012 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#include "changk104.h" -#include "cgrp.h" - -#include <core/client.h> -#include <core/gpuobj.h> -#include <subdev/fb.h> -#include <subdev/mmu.h> -#include <subdev/timer.h> - -#include <nvif/class.h> -#include <nvif/cla06f.h> -#include <nvif/unpack.h> - -int -gk104_fifo_gpfifo_kick_locked(struct gk104_fifo_chan *chan) -{ - struct gk104_fifo *fifo = chan->fifo; - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - struct nvkm_client *client = chan->base.object.client; - struct nvkm_fifo_cgrp *cgrp = chan->cgrp; - int ret = 0; - - if (cgrp) - nvkm_wr32(device, 0x002634, cgrp->id | 0x01000000); - else - nvkm_wr32(device, 0x002634, chan->base.chid); - if (nvkm_msec(device, 2000, - if (!(nvkm_rd32(device, 0x002634) & 0x00100000)) - break; - ) < 0) { - nvkm_error(subdev, "%s %d [%s] kick timeout\n", - cgrp ? "tsg" : "channel", - cgrp ? cgrp->id : chan->base.chid, client->name); - nvkm_fifo_recover_chan(&fifo->base, chan->base.chid); - ret = -ETIMEDOUT; - } - return ret; -} - -int -gk104_fifo_gpfifo_kick(struct gk104_fifo_chan *chan) -{ - int ret; - mutex_lock(&chan->base.fifo->mutex); - ret = gk104_fifo_gpfifo_kick_locked(chan); - mutex_unlock(&chan->base.fifo->mutex); - return ret; -} - -static u32 -gk104_fifo_gpfifo_engine_addr(struct nvkm_engine *engine) -{ - switch (engine->subdev.type) { - case NVKM_ENGINE_SW : - case NVKM_ENGINE_CE : return 0; - case NVKM_ENGINE_GR : return 0x0210; - case NVKM_ENGINE_SEC : return 0x0220; - case NVKM_ENGINE_MSPDEC: return 0x0250; - case NVKM_ENGINE_MSPPP : return 0x0260; - case NVKM_ENGINE_MSVLD : return 0x0270; - case NVKM_ENGINE_VIC : return 0x0280; - case NVKM_ENGINE_MSENC : return 0x0290; - case NVKM_ENGINE_NVDEC : return 0x02100270; - case NVKM_ENGINE_NVENC : - if (engine->subdev.inst) - return 0x0210; - return 0x02100290; - default: - WARN_ON(1); - return 0; - } -} - -struct gk104_fifo_engn * -gk104_fifo_gpfifo_engine(struct gk104_fifo_chan *chan, struct nvkm_engine *engine) -{ - int engi = chan->base.fifo->func->engine_id(chan->base.fifo, engine); - if (engi >= 0) - return &chan->engn[engi]; - return NULL; -} - -static int -gk104_fifo_gpfifo_engine_fini(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine, bool suspend) -{ - struct gk104_fifo_chan *chan = gk104_fifo_chan(base); - struct nvkm_gpuobj *inst = chan->base.inst; - u32 offset = gk104_fifo_gpfifo_engine_addr(engine); - int ret; - - ret = gk104_fifo_gpfifo_kick(chan); - if (ret && suspend) - return ret; - - if (offset) { - nvkm_kmap(inst); - nvkm_wo32(inst, (offset & 0xffff) + 0x00, 0x00000000); - nvkm_wo32(inst, (offset & 0xffff) + 0x04, 0x00000000); - if ((offset >>= 16)) { - nvkm_wo32(inst, offset + 0x00, 0x00000000); - nvkm_wo32(inst, offset + 0x04, 0x00000000); - } - nvkm_done(inst); - } - - return ret; -} - -static int -gk104_fifo_gpfifo_engine_init(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine) -{ - struct gk104_fifo_chan *chan = gk104_fifo_chan(base); - struct gk104_fifo_engn *engn = gk104_fifo_gpfifo_engine(chan, engine); - struct nvkm_gpuobj *inst = chan->base.inst; - u32 offset = gk104_fifo_gpfifo_engine_addr(engine); - - if (offset) { - u32 datalo = lower_32_bits(engn->vma->addr) | 0x00000004; - u32 datahi = upper_32_bits(engn->vma->addr); - nvkm_kmap(inst); - nvkm_wo32(inst, (offset & 0xffff) + 0x00, datalo); - nvkm_wo32(inst, (offset & 0xffff) + 0x04, datahi); - if ((offset >>= 16)) { - nvkm_wo32(inst, offset + 0x00, datalo); - nvkm_wo32(inst, offset + 0x04, datahi); - } - nvkm_done(inst); - } - - return 0; -} - -void -gk104_fifo_gpfifo_engine_dtor(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine) -{ - struct gk104_fifo_chan *chan = gk104_fifo_chan(base); - struct gk104_fifo_engn *engn = gk104_fifo_gpfifo_engine(chan, engine); - nvkm_vmm_put(chan->base.vmm, &engn->vma); - nvkm_gpuobj_del(&engn->inst); -} - -int -gk104_fifo_gpfifo_engine_ctor(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine, - struct nvkm_object *object) -{ - struct gk104_fifo_chan *chan = gk104_fifo_chan(base); - struct gk104_fifo_engn *engn = gk104_fifo_gpfifo_engine(chan, engine); - int ret; - - if (!gk104_fifo_gpfifo_engine_addr(engine)) { - if (engine->subdev.type != NVKM_ENGINE_CE || - engine->subdev.device->card_type < GV100) - return 0; - } - - ret = nvkm_object_bind(object, NULL, 0, &engn->inst); - if (ret) - return ret; - - if (!gk104_fifo_gpfifo_engine_addr(engine)) - return 0; - - ret = nvkm_vmm_get(chan->base.vmm, 12, engn->inst->size, &engn->vma); - if (ret) - return ret; - - return nvkm_memory_map(engn->inst, 0, chan->base.vmm, engn->vma, NULL, 0); -} - -void -gk104_fifo_gpfifo_fini(struct nvkm_fifo_chan *base) -{ - struct gk104_fifo_chan *chan = gk104_fifo_chan(base); - struct gk104_fifo *fifo = chan->fifo; - struct nvkm_device *device = fifo->base.engine.subdev.device; - u32 coff = chan->base.chid * 8; - - if (!list_empty(&chan->head)) { - gk104_fifo_runlist_remove(fifo, chan); - nvkm_mask(device, 0x800004 + coff, 0x00000800, 0x00000800); - gk104_fifo_gpfifo_kick(chan); - gk104_fifo_runlist_update(fifo, chan->runl); - } - - nvkm_wr32(device, 0x800000 + coff, 0x00000000); -} - -void -gk104_fifo_gpfifo_init(struct nvkm_fifo_chan *base) -{ - struct gk104_fifo_chan *chan = gk104_fifo_chan(base); - struct gk104_fifo *fifo = chan->fifo; - struct nvkm_device *device = fifo->base.engine.subdev.device; - u32 addr = chan->base.inst->addr >> 12; - u32 coff = chan->base.chid * 8; - - nvkm_mask(device, 0x800004 + coff, 0x000f0000, chan->runl << 16); - nvkm_wr32(device, 0x800000 + coff, 0x80000000 | addr); - - if (list_empty(&chan->head) && !chan->killed) { - gk104_fifo_runlist_insert(fifo, chan); - nvkm_mask(device, 0x800004 + coff, 0x00000400, 0x00000400); - gk104_fifo_runlist_update(fifo, chan->runl); - nvkm_mask(device, 0x800004 + coff, 0x00000400, 0x00000400); - } -} - -void * -gk104_fifo_gpfifo_dtor(struct nvkm_fifo_chan *base) -{ - struct gk104_fifo_chan *chan = gk104_fifo_chan(base); - kfree(chan->cgrp); - return chan; -} - -const struct nvkm_fifo_chan_func -gk104_fifo_gpfifo_func = { - .dtor = gk104_fifo_gpfifo_dtor, - .init = gk104_fifo_gpfifo_init, - .fini = gk104_fifo_gpfifo_fini, - .ntfy = gf100_fifo_chan_ntfy, - .engine_ctor = gk104_fifo_gpfifo_engine_ctor, - .engine_dtor = gk104_fifo_gpfifo_engine_dtor, - .engine_init = gk104_fifo_gpfifo_engine_init, - .engine_fini = gk104_fifo_gpfifo_engine_fini, -}; - -static int -gk104_fifo_gpfifo_new_(struct gk104_fifo *fifo, u64 *runlists, u16 *chid, - u64 vmm, u64 ioffset, u64 ilength, u64 *inst, bool priv, - const struct nvkm_oclass *oclass, - struct nvkm_object **pobject) -{ - struct gk104_fifo_chan *chan; - int runlist = ffs(*runlists) -1, ret, i; - u64 usermem; - - if (!vmm || runlist < 0 || runlist >= fifo->runlist_nr) - return -EINVAL; - *runlists = BIT_ULL(runlist); - - /* Allocate the channel. */ - if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) - return -ENOMEM; - *pobject = &chan->base.object; - chan->fifo = fifo; - chan->runl = runlist; - INIT_LIST_HEAD(&chan->head); - - ret = nvkm_fifo_chan_ctor(&gk104_fifo_gpfifo_func, &fifo->base, - 0x1000, 0x1000, true, vmm, 0, fifo->runlist[runlist].engm_sw, - 1, fifo->user.bar->addr, 0x200, - oclass, &chan->base); - if (ret) - return ret; - - *chid = chan->base.chid; - *inst = chan->base.inst->addr; - - /* Hack to support GPUs where even individual channels should be - * part of a channel group. - */ - if (fifo->func->cgrp_force) { - if (!(chan->cgrp = kmalloc(sizeof(*chan->cgrp), GFP_KERNEL))) - return -ENOMEM; - chan->cgrp->id = chan->base.chid; - INIT_LIST_HEAD(&chan->cgrp->head); - INIT_LIST_HEAD(&chan->cgrp->chan); - chan->cgrp->chan_nr = 0; - } - - /* Clear channel control registers. */ - usermem = chan->base.chid * 0x200; - ilength = order_base_2(ilength / 8); - - nvkm_kmap(fifo->user.mem); - for (i = 0; i < 0x200; i += 4) - nvkm_wo32(fifo->user.mem, usermem + i, 0x00000000); - nvkm_done(fifo->user.mem); - usermem = nvkm_memory_addr(fifo->user.mem) + usermem; - - /* RAMFC */ - nvkm_kmap(chan->base.inst); - nvkm_wo32(chan->base.inst, 0x08, lower_32_bits(usermem)); - nvkm_wo32(chan->base.inst, 0x0c, upper_32_bits(usermem)); - nvkm_wo32(chan->base.inst, 0x10, 0x0000face); - nvkm_wo32(chan->base.inst, 0x30, 0xfffff902); - nvkm_wo32(chan->base.inst, 0x48, lower_32_bits(ioffset)); - nvkm_wo32(chan->base.inst, 0x4c, upper_32_bits(ioffset) | - (ilength << 16)); - nvkm_wo32(chan->base.inst, 0x84, 0x20400000); - nvkm_wo32(chan->base.inst, 0x94, 0x30000001); - nvkm_wo32(chan->base.inst, 0x9c, 0x00000100); - nvkm_wo32(chan->base.inst, 0xac, 0x0000001f); - nvkm_wo32(chan->base.inst, 0xe4, priv ? 0x00000020 : 0x00000000); - nvkm_wo32(chan->base.inst, 0xe8, chan->base.chid); - nvkm_wo32(chan->base.inst, 0xb8, 0xf8000000); - nvkm_wo32(chan->base.inst, 0xf8, 0x10003080); /* 0x002310 */ - nvkm_wo32(chan->base.inst, 0xfc, 0x10000010); /* 0x002350 */ - nvkm_done(chan->base.inst); - return 0; -} - -int -gk104_fifo_gpfifo_new(struct gk104_fifo *fifo, const struct nvkm_oclass *oclass, - void *data, u32 size, struct nvkm_object **pobject) -{ - struct nvkm_object *parent = oclass->parent; - union { - struct kepler_channel_gpfifo_a_v0 v0; - } *args = data; - int ret = -ENOSYS; - - nvif_ioctl(parent, "create channel gpfifo size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(parent, "create channel gpfifo vers %d vmm %llx " - "ioffset %016llx ilength %08x " - "runlist %016llx priv %d\n", - args->v0.version, args->v0.vmm, args->v0.ioffset, - args->v0.ilength, args->v0.runlist, args->v0.priv); - return gk104_fifo_gpfifo_new_(fifo, - &args->v0.runlist, - &args->v0.chid, - args->v0.vmm, - args->v0.ioffset, - args->v0.ilength, - &args->v0.inst, - args->v0.priv, - oclass, pobject); - } - - return ret; -} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogv100.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogv100.c deleted file mode 100644 index 428f9b41165c..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifogv100.c +++ /dev/null @@ -1,241 +0,0 @@ -/* - * Copyright 2018 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - */ -#include "changk104.h" -#include "cgrp.h" - -#include <core/client.h> -#include <core/gpuobj.h> - -#include <nvif/clc36f.h> -#include <nvif/unpack.h> - -static u32 -gv100_fifo_gpfifo_submit_token(struct nvkm_fifo_chan *chan) -{ - return chan->chid; -} - -static int -gv100_fifo_gpfifo_engine_valid(struct gk104_fifo_chan *chan, bool ce, bool valid) -{ - struct nvkm_subdev *subdev = &chan->base.fifo->engine.subdev; - struct nvkm_device *device = subdev->device; - const u32 mask = ce ? 0x00020000 : 0x00010000; - const u32 data = valid ? mask : 0x00000000; - int ret; - - /* Block runlist to prevent the channel from being rescheduled. */ - mutex_lock(&chan->fifo->base.mutex); - nvkm_mask(device, 0x002630, BIT(chan->runl), BIT(chan->runl)); - - /* Preempt the channel. */ - ret = gk104_fifo_gpfifo_kick_locked(chan); - if (ret == 0) { - /* Update engine context validity. */ - nvkm_kmap(chan->base.inst); - nvkm_mo32(chan->base.inst, 0x0ac, mask, data); - nvkm_done(chan->base.inst); - } - - /* Resume runlist. */ - nvkm_mask(device, 0x002630, BIT(chan->runl), 0); - mutex_unlock(&chan->fifo->base.mutex); - return ret; -} - -int -gv100_fifo_gpfifo_engine_fini(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine, bool suspend) -{ - struct gk104_fifo_chan *chan = gk104_fifo_chan(base); - struct nvkm_gpuobj *inst = chan->base.inst; - int ret; - - if (engine->subdev.type == NVKM_ENGINE_CE) { - ret = gv100_fifo_gpfifo_engine_valid(chan, true, false); - if (ret && suspend) - return ret; - - nvkm_kmap(inst); - nvkm_wo32(chan->base.inst, 0x220, 0x00000000); - nvkm_wo32(chan->base.inst, 0x224, 0x00000000); - nvkm_done(inst); - return ret; - } - - ret = gv100_fifo_gpfifo_engine_valid(chan, false, false); - if (ret && suspend) - return ret; - - nvkm_kmap(inst); - nvkm_wo32(inst, 0x0210, 0x00000000); - nvkm_wo32(inst, 0x0214, 0x00000000); - nvkm_done(inst); - return ret; -} - -int -gv100_fifo_gpfifo_engine_init(struct nvkm_fifo_chan *base, - struct nvkm_engine *engine) -{ - struct gk104_fifo_chan *chan = gk104_fifo_chan(base); - struct gk104_fifo_engn *engn = gk104_fifo_gpfifo_engine(chan, engine); - struct nvkm_gpuobj *inst = chan->base.inst; - - if (engine->subdev.type == NVKM_ENGINE_CE) { - const u64 bar2 = nvkm_memory_bar2(engn->inst->memory); - - nvkm_kmap(inst); - nvkm_wo32(chan->base.inst, 0x220, lower_32_bits(bar2)); - nvkm_wo32(chan->base.inst, 0x224, upper_32_bits(bar2)); - nvkm_done(inst); - - return gv100_fifo_gpfifo_engine_valid(chan, true, true); - } - - nvkm_kmap(inst); - nvkm_wo32(inst, 0x210, lower_32_bits(engn->vma->addr) | 0x00000004); - nvkm_wo32(inst, 0x214, upper_32_bits(engn->vma->addr)); - nvkm_done(inst); - - return gv100_fifo_gpfifo_engine_valid(chan, false, true); -} - -static const struct nvkm_fifo_chan_func -gv100_fifo_gpfifo = { - .dtor = gk104_fifo_gpfifo_dtor, - .init = gk104_fifo_gpfifo_init, - .fini = gk104_fifo_gpfifo_fini, - .ntfy = gf100_fifo_chan_ntfy, - .engine_ctor = gk104_fifo_gpfifo_engine_ctor, - .engine_dtor = gk104_fifo_gpfifo_engine_dtor, - .engine_init = gv100_fifo_gpfifo_engine_init, - .engine_fini = gv100_fifo_gpfifo_engine_fini, - .submit_token = gv100_fifo_gpfifo_submit_token, -}; - -int -gv100_fifo_gpfifo_new_(const struct nvkm_fifo_chan_func *func, - struct gk104_fifo *fifo, u64 *runlists, u16 *chid, - u64 vmm, u64 ioffset, u64 ilength, u64 *inst, bool priv, - u32 *token, const struct nvkm_oclass *oclass, - struct nvkm_object **pobject) -{ - struct gk104_fifo_chan *chan; - int runlist = ffs(*runlists) -1, ret, i; - u64 usermem; - - if (!vmm || runlist < 0 || runlist >= fifo->runlist_nr) - return -EINVAL; - *runlists = BIT_ULL(runlist); - - /* Allocate the channel. */ - if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) - return -ENOMEM; - *pobject = &chan->base.object; - chan->fifo = fifo; - chan->runl = runlist; - INIT_LIST_HEAD(&chan->head); - - ret = nvkm_fifo_chan_ctor(func, &fifo->base, 0x1000, 0x1000, true, vmm, - 0, fifo->runlist[runlist].engm, 1, fifo->user.bar->addr, 0x200, - oclass, &chan->base); - if (ret) - return ret; - - *chid = chan->base.chid; - *inst = chan->base.inst->addr; - *token = chan->base.func->submit_token(&chan->base); - - /* Hack to support GPUs where even individual channels should be - * part of a channel group. - */ - if (fifo->func->cgrp_force) { - if (!(chan->cgrp = kmalloc(sizeof(*chan->cgrp), GFP_KERNEL))) - return -ENOMEM; - chan->cgrp->id = chan->base.chid; - INIT_LIST_HEAD(&chan->cgrp->head); - INIT_LIST_HEAD(&chan->cgrp->chan); - chan->cgrp->chan_nr = 0; - } - - /* Clear channel control registers. */ - usermem = chan->base.chid * 0x200; - ilength = order_base_2(ilength / 8); - - nvkm_kmap(fifo->user.mem); - for (i = 0; i < 0x200; i += 4) - nvkm_wo32(fifo->user.mem, usermem + i, 0x00000000); - nvkm_done(fifo->user.mem); - usermem = nvkm_memory_addr(fifo->user.mem) + usermem; - - /* RAMFC */ - nvkm_kmap(chan->base.inst); - nvkm_wo32(chan->base.inst, 0x008, lower_32_bits(usermem)); - nvkm_wo32(chan->base.inst, 0x00c, upper_32_bits(usermem)); - nvkm_wo32(chan->base.inst, 0x010, 0x0000face); - nvkm_wo32(chan->base.inst, 0x030, 0x7ffff902); - nvkm_wo32(chan->base.inst, 0x048, lower_32_bits(ioffset)); - nvkm_wo32(chan->base.inst, 0x04c, upper_32_bits(ioffset) | - (ilength << 16)); - nvkm_wo32(chan->base.inst, 0x084, 0x20400000); - nvkm_wo32(chan->base.inst, 0x094, 0x30000001); - nvkm_wo32(chan->base.inst, 0x0e4, priv ? 0x00000020 : 0x00000000); - nvkm_wo32(chan->base.inst, 0x0e8, chan->base.chid); - nvkm_wo32(chan->base.inst, 0x0f4, 0x00001000); - nvkm_wo32(chan->base.inst, 0x0f8, 0x10003080); - nvkm_mo32(chan->base.inst, 0x218, 0x00000000, 0x00000000); - nvkm_done(chan->base.inst); - return 0; -} - -int -gv100_fifo_gpfifo_new(struct gk104_fifo *fifo, const struct nvkm_oclass *oclass, - void *data, u32 size, struct nvkm_object **pobject) -{ - struct nvkm_object *parent = oclass->parent; - union { - struct volta_channel_gpfifo_a_v0 v0; - } *args = data; - int ret = -ENOSYS; - - nvif_ioctl(parent, "create channel gpfifo size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(parent, "create channel gpfifo vers %d vmm %llx " - "ioffset %016llx ilength %08x " - "runlist %016llx priv %d\n", - args->v0.version, args->v0.vmm, args->v0.ioffset, - args->v0.ilength, args->v0.runlist, args->v0.priv); - return gv100_fifo_gpfifo_new_(&gv100_fifo_gpfifo, fifo, - &args->v0.runlist, - &args->v0.chid, - args->v0.vmm, - args->v0.ioffset, - args->v0.ilength, - &args->v0.inst, - args->v0.priv, - &args->v0.token, - oclass, pobject); - } - - return ret; -} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifonv50.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifonv50.c deleted file mode 100644 index d8f28ec1e4a8..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifonv50.c +++ /dev/null @@ -1,93 +0,0 @@ -/* - * Copyright 2012 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs - */ -#include "channv50.h" - -#include <core/client.h> -#include <core/ramht.h> - -#include <nvif/class.h> -#include <nvif/cl506f.h> -#include <nvif/unpack.h> - -static int -nv50_fifo_gpfifo_new(struct nvkm_fifo *base, const struct nvkm_oclass *oclass, - void *data, u32 size, struct nvkm_object **pobject) -{ - struct nvkm_object *parent = oclass->parent; - union { - struct nv50_channel_gpfifo_v0 v0; - } *args = data; - struct nv50_fifo *fifo = nv50_fifo(base); - struct nv50_fifo_chan *chan; - u64 ioffset, ilength; - int ret = -ENOSYS; - - nvif_ioctl(parent, "create channel gpfifo size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(parent, "create channel gpfifo vers %d vmm %llx " - "pushbuf %llx ioffset %016llx " - "ilength %08x\n", - args->v0.version, args->v0.vmm, args->v0.pushbuf, - args->v0.ioffset, args->v0.ilength); - if (!args->v0.pushbuf) - return -EINVAL; - } else - return ret; - - if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) - return -ENOMEM; - *pobject = &chan->base.object; - - ret = nv50_fifo_chan_ctor(fifo, args->v0.vmm, args->v0.pushbuf, - oclass, chan); - if (ret) - return ret; - - args->v0.chid = chan->base.chid; - ioffset = args->v0.ioffset; - ilength = order_base_2(args->v0.ilength / 8); - - nvkm_kmap(chan->ramfc); - nvkm_wo32(chan->ramfc, 0x3c, 0x403f6078); - nvkm_wo32(chan->ramfc, 0x44, 0x01003fff); - nvkm_wo32(chan->ramfc, 0x48, chan->base.push->node->offset >> 4); - nvkm_wo32(chan->ramfc, 0x50, lower_32_bits(ioffset)); - nvkm_wo32(chan->ramfc, 0x54, upper_32_bits(ioffset) | (ilength << 16)); - nvkm_wo32(chan->ramfc, 0x60, 0x7fffffff); - nvkm_wo32(chan->ramfc, 0x78, 0x00000000); - nvkm_wo32(chan->ramfc, 0x7c, 0x30000001); - nvkm_wo32(chan->ramfc, 0x80, ((chan->ramht->bits - 9) << 27) | - (4 << 24) /* SEARCH_FULL */ | - (chan->ramht->gpuobj->node->offset >> 4)); - nvkm_done(chan->ramfc); - return 0; -} - -const struct nvkm_fifo_chan_oclass -nv50_fifo_gpfifo_oclass = { - .base.oclass = NV50_CHANNEL_GPFIFO, - .base.minver = 0, - .base.maxver = 0, - .ctor = nv50_fifo_gpfifo_new, -}; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifotu102.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifotu102.c deleted file mode 100644 index 99aafa103a31..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gpfifotu102.c +++ /dev/null @@ -1,81 +0,0 @@ -/* - * Copyright 2018 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - */ -#include "changk104.h" -#include "cgrp.h" - -#include <core/client.h> -#include <core/gpuobj.h> - -#include <nvif/clc36f.h> -#include <nvif/unpack.h> - -static u32 -tu102_fifo_gpfifo_submit_token(struct nvkm_fifo_chan *base) -{ - struct gk104_fifo_chan *chan = gk104_fifo_chan(base); - return (chan->runl << 16) | chan->base.chid; -} - -static const struct nvkm_fifo_chan_func -tu102_fifo_gpfifo = { - .dtor = gk104_fifo_gpfifo_dtor, - .init = gk104_fifo_gpfifo_init, - .fini = gk104_fifo_gpfifo_fini, - .ntfy = gf100_fifo_chan_ntfy, - .engine_ctor = gk104_fifo_gpfifo_engine_ctor, - .engine_dtor = gk104_fifo_gpfifo_engine_dtor, - .engine_init = gv100_fifo_gpfifo_engine_init, - .engine_fini = gv100_fifo_gpfifo_engine_fini, - .submit_token = tu102_fifo_gpfifo_submit_token, -}; - -int -tu102_fifo_gpfifo_new(struct gk104_fifo *fifo, const struct nvkm_oclass *oclass, - void *data, u32 size, struct nvkm_object **pobject) -{ - struct nvkm_object *parent = oclass->parent; - union { - struct volta_channel_gpfifo_a_v0 v0; - } *args = data; - int ret = -ENOSYS; - - nvif_ioctl(parent, "create channel gpfifo size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { - nvif_ioctl(parent, "create channel gpfifo vers %d vmm %llx " - "ioffset %016llx ilength %08x " - "runlist %016llx priv %d\n", - args->v0.version, args->v0.vmm, args->v0.ioffset, - args->v0.ilength, args->v0.runlist, args->v0.priv); - return gv100_fifo_gpfifo_new_(&tu102_fifo_gpfifo, fifo, - &args->v0.runlist, - &args->v0.chid, - args->v0.vmm, - args->v0.ioffset, - args->v0.ilength, - &args->v0.inst, - args->v0.priv, - &args->v0.token, - oclass, pobject); - } - - return ret; -} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gv100.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gv100.c index faf0fe9f704c..33066c8cdc64 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gv100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/gv100.c @@ -19,32 +19,180 @@ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. */ -#include "gk104.h" +#include "priv.h" +#include "chan.h" +#include "chid.h" #include "cgrp.h" -#include "changk104.h" -#include "user.h" +#include "runl.h" +#include "runq.h" #include <core/gpuobj.h> +#include <subdev/mmu.h> #include <nvif/class.h> +static u32 +gv100_chan_doorbell_handle(struct nvkm_chan *chan) +{ + return chan->id; +} + +static int +gv100_chan_ramfc_write(struct nvkm_chan *chan, u64 offset, u64 length, u32 devm, bool priv) +{ + const u64 userd = nvkm_memory_addr(chan->userd.mem) + chan->userd.base; + const u32 limit2 = ilog2(length / 8); + + nvkm_kmap(chan->inst); + nvkm_wo32(chan->inst, 0x008, lower_32_bits(userd)); + nvkm_wo32(chan->inst, 0x00c, upper_32_bits(userd)); + nvkm_wo32(chan->inst, 0x010, 0x0000face); + nvkm_wo32(chan->inst, 0x030, 0x7ffff902); + nvkm_wo32(chan->inst, 0x048, lower_32_bits(offset)); + nvkm_wo32(chan->inst, 0x04c, upper_32_bits(offset) | (limit2 << 16)); + nvkm_wo32(chan->inst, 0x084, 0x20400000); + nvkm_wo32(chan->inst, 0x094, 0x30000000 | devm); + nvkm_wo32(chan->inst, 0x0e4, priv ? 0x00000020 : 0x00000000); + nvkm_wo32(chan->inst, 0x0e8, chan->id); + nvkm_wo32(chan->inst, 0x0f4, 0x00001000 | (priv ? 0x00000100 : 0x00000000)); + nvkm_wo32(chan->inst, 0x0f8, 0x10003080); + nvkm_mo32(chan->inst, 0x218, 0x00000000, 0x00000000); + nvkm_done(chan->inst); + return 0; +} + +const struct nvkm_chan_func_ramfc +gv100_chan_ramfc = { + .write = gv100_chan_ramfc_write, + .devm = 0xfff, + .priv = true, +}; + +const struct nvkm_chan_func_userd +gv100_chan_userd = { + .bar = -1, + .size = 0x200, + .clear = gf100_chan_userd_clear, +}; + +static const struct nvkm_chan_func +gv100_chan = { + .inst = &gf100_chan_inst, + .userd = &gv100_chan_userd, + .ramfc = &gv100_chan_ramfc, + .bind = gk104_chan_bind_inst, + .unbind = gk104_chan_unbind, + .start = gk104_chan_start, + .stop = gk104_chan_stop, + .preempt = gk110_chan_preempt, + .doorbell_handle = gv100_chan_doorbell_handle, +}; + +void +gv100_ectx_bind(struct nvkm_engn *engn, struct nvkm_cctx *cctx, struct nvkm_chan *chan) +{ + u64 addr = 0ULL; + + if (cctx) { + addr = cctx->vctx->vma->addr; + addr |= 4ULL; + } + + nvkm_kmap(chan->inst); + nvkm_wo32(chan->inst, 0x210, lower_32_bits(addr)); + nvkm_wo32(chan->inst, 0x214, upper_32_bits(addr)); + nvkm_mo32(chan->inst, 0x0ac, 0x00010000, cctx ? 0x00010000 : 0x00000000); + nvkm_done(chan->inst); +} + +const struct nvkm_engn_func +gv100_engn = { + .chsw = gk104_engn_chsw, + .cxid = gk104_engn_cxid, + .ctor = gk104_ectx_ctor, + .bind = gv100_ectx_bind, +}; + +void +gv100_ectx_ce_bind(struct nvkm_engn *engn, struct nvkm_cctx *cctx, struct nvkm_chan *chan) +{ + const u64 bar2 = cctx ? nvkm_memory_bar2(cctx->vctx->inst->memory) : 0ULL; + + nvkm_kmap(chan->inst); + nvkm_wo32(chan->inst, 0x220, lower_32_bits(bar2)); + nvkm_wo32(chan->inst, 0x224, upper_32_bits(bar2)); + nvkm_mo32(chan->inst, 0x0ac, 0x00020000, cctx ? 0x00020000 : 0x00000000); + nvkm_done(chan->inst); +} + +int +gv100_ectx_ce_ctor(struct nvkm_engn *engn, struct nvkm_vctx *vctx) +{ + if (nvkm_memory_bar2(vctx->inst->memory) == ~0ULL) + return -EFAULT; + + return 0; +} + +const struct nvkm_engn_func +gv100_engn_ce = { + .chsw = gk104_engn_chsw, + .cxid = gk104_engn_cxid, + .ctor = gv100_ectx_ce_ctor, + .bind = gv100_ectx_ce_bind, +}; + +static bool +gv100_runq_intr_1_ctxnotvalid(struct nvkm_runq *runq, int chid) +{ + struct nvkm_fifo *fifo = runq->fifo; + struct nvkm_device *device = fifo->engine.subdev.device; + struct nvkm_chan *chan; + unsigned long flags; + + RUNQ_ERROR(runq, "CTXNOTVALID chid:%d", chid); + + chan = nvkm_chan_get_chid(&fifo->engine, chid, &flags); + if (WARN_ON_ONCE(!chan)) + return false; + + nvkm_chan_error(chan, true); + nvkm_chan_put(&chan, flags); + + nvkm_mask(device, 0x0400ac + (runq->id * 0x2000), 0x00030000, 0x00030000); + nvkm_wr32(device, 0x040148 + (runq->id * 0x2000), 0x80000000); + return true; +} + +const struct nvkm_runq_func +gv100_runq = { + .init = gk208_runq_init, + .intr = gk104_runq_intr, + .intr_0_names = gk104_runq_intr_0_names, + .intr_1_ctxnotvalid = gv100_runq_intr_1_ctxnotvalid, + .idle = gk104_runq_idle, +}; + +void +gv100_runl_preempt(struct nvkm_runl *runl) +{ + nvkm_wr32(runl->fifo->engine.subdev.device, 0x002638, BIT(runl->id)); +} + void -gv100_fifo_runlist_chan(struct gk104_fifo_chan *chan, - struct nvkm_memory *memory, u32 offset) +gv100_runl_insert_chan(struct nvkm_chan *chan, struct nvkm_memory *memory, u64 offset) { - struct nvkm_memory *usermem = chan->fifo->user.mem; - const u64 user = nvkm_memory_addr(usermem) + (chan->base.chid * 0x200); - const u64 inst = chan->base.inst->addr; + const u64 user = nvkm_memory_addr(chan->userd.mem) + chan->userd.base; + const u64 inst = chan->inst->addr; - nvkm_wo32(memory, offset + 0x0, lower_32_bits(user)); + nvkm_wo32(memory, offset + 0x0, lower_32_bits(user) | chan->runq << 1); nvkm_wo32(memory, offset + 0x4, upper_32_bits(user)); - nvkm_wo32(memory, offset + 0x8, lower_32_bits(inst) | chan->base.chid); + nvkm_wo32(memory, offset + 0x8, lower_32_bits(inst) | chan->id); nvkm_wo32(memory, offset + 0xc, upper_32_bits(inst)); } void -gv100_fifo_runlist_cgrp(struct nvkm_fifo_cgrp *cgrp, - struct nvkm_memory *memory, u32 offset) +gv100_runl_insert_cgrp(struct nvkm_cgrp *cgrp, struct nvkm_memory *memory, u64 offset) { nvkm_wo32(memory, offset + 0x0, (128 << 24) | (3 << 16) | 0x00000001); nvkm_wo32(memory, offset + 0x4, cgrp->chan_nr); @@ -52,16 +200,24 @@ gv100_fifo_runlist_cgrp(struct nvkm_fifo_cgrp *cgrp, nvkm_wo32(memory, offset + 0xc, 0x00000000); } -static const struct gk104_fifo_runlist_func -gv100_fifo_runlist = { +static const struct nvkm_runl_func +gv100_runl = { + .runqs = 2, .size = 16, - .cgrp = gv100_fifo_runlist_cgrp, - .chan = gv100_fifo_runlist_chan, - .commit = gk104_fifo_runlist_commit, + .update = nv50_runl_update, + .insert_cgrp = gv100_runl_insert_cgrp, + .insert_chan = gv100_runl_insert_chan, + .commit = gk104_runl_commit, + .wait = nv50_runl_wait, + .pending = gk104_runl_pending, + .block = gk104_runl_block, + .allow = gk104_runl_allow, + .preempt = gv100_runl_preempt, + .preempt_pending = gf100_runl_preempt_pending, }; const struct nvkm_enum -gv100_fifo_fault_gpcclient[] = { +gv100_fifo_mmu_fault_gpcclient[] = { { 0x00, "T1_0" }, { 0x01, "T1_1" }, { 0x02, "T1_2" }, @@ -163,7 +319,7 @@ gv100_fifo_fault_gpcclient[] = { }; const struct nvkm_enum -gv100_fifo_fault_hubclient[] = { +gv100_fifo_mmu_fault_hubclient[] = { { 0x00, "VIP" }, { 0x01, "CE0" }, { 0x02, "CE1" }, @@ -225,7 +381,7 @@ gv100_fifo_fault_hubclient[] = { }; const struct nvkm_enum -gv100_fifo_fault_reason[] = { +gv100_fifo_mmu_fault_reason[] = { { 0x00, "PDE" }, { 0x01, "PDE_SIZE" }, { 0x02, "PTE" }, @@ -246,7 +402,7 @@ gv100_fifo_fault_reason[] = { }; static const struct nvkm_enum -gv100_fifo_fault_engine[] = { +gv100_fifo_mmu_fault_engine[] = { { 0x01, "DISPLAY" }, { 0x03, "PTP" }, { 0x04, "BAR1", NULL, NVKM_SUBDEV_BAR }, @@ -273,7 +429,7 @@ gv100_fifo_fault_engine[] = { }; const struct nvkm_enum -gv100_fifo_fault_access[] = { +gv100_fifo_mmu_fault_access[] = { { 0x0, "VIRT_READ" }, { 0x1, "VIRT_WRITE" }, { 0x2, "VIRT_ATOMIC" }, @@ -286,23 +442,51 @@ gv100_fifo_fault_access[] = { {} }; -static const struct gk104_fifo_func +static const struct nvkm_fifo_func_mmu_fault +gv100_fifo_mmu_fault = { + .recover = gf100_fifo_mmu_fault_recover, + .access = gv100_fifo_mmu_fault_access, + .engine = gv100_fifo_mmu_fault_engine, + .reason = gv100_fifo_mmu_fault_reason, + .hubclient = gv100_fifo_mmu_fault_hubclient, + .gpcclient = gv100_fifo_mmu_fault_gpcclient, +}; + +static void +gv100_fifo_intr_ctxsw_timeout(struct nvkm_fifo *fifo, u32 engm) +{ + struct nvkm_runl *runl; + struct nvkm_engn *engn; + + nvkm_runl_foreach(runl, fifo) { + nvkm_runl_foreach_engn_cond(engn, runl, engm & BIT(engn->id)) + nvkm_runl_rc_engn(runl, engn); + } +} + +static const struct nvkm_fifo_func gv100_fifo = { - .pbdma = &gm200_fifo_pbdma, - .fault.access = gv100_fifo_fault_access, - .fault.engine = gv100_fifo_fault_engine, - .fault.reason = gv100_fifo_fault_reason, - .fault.hubclient = gv100_fifo_fault_hubclient, - .fault.gpcclient = gv100_fifo_fault_gpcclient, - .runlist = &gv100_fifo_runlist, - .user = {{-1,-1,VOLTA_USERMODE_A }, gv100_fifo_user_new }, - .chan = {{ 0, 0,VOLTA_CHANNEL_GPFIFO_A}, gv100_fifo_gpfifo_new }, - .cgrp_force = true, + .chid_nr = gm200_fifo_chid_nr, + .chid_ctor = gk110_fifo_chid_ctor, + .runq_nr = gm200_fifo_runq_nr, + .runl_ctor = gk104_fifo_runl_ctor, + .init = gk104_fifo_init, + .init_pbdmas = gk104_fifo_init_pbdmas, + .intr = gk104_fifo_intr, + .intr_ctxsw_timeout = gv100_fifo_intr_ctxsw_timeout, + .mmu_fault = &gv100_fifo_mmu_fault, + .nonstall = &gf100_fifo_nonstall, + .runl = &gv100_runl, + .runq = &gv100_runq, + .engn = &gv100_engn, + .engn_ce = &gv100_engn_ce, + .cgrp = {{ 0, 0, KEPLER_CHANNEL_GROUP_A }, &gk110_cgrp, .force = true }, + .chan = {{ 0, 0, VOLTA_CHANNEL_GPFIFO_A }, &gv100_chan }, }; int gv100_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return gk104_fifo_new_(&gv100_fifo, device, type, inst, 4096, pfifo); + return nvkm_fifo_new_(&gv100_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv04.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv04.c index c6730c124769..674faf002b20 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv04.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv04.c @@ -21,38 +21,201 @@ * * Authors: Ben Skeggs */ -#include "nv04.h" -#include "channv04.h" +#include "priv.h" +#include "cgrp.h" +#include "chan.h" +#include "chid.h" +#include "runl.h" + #include "regsnv04.h" -#include <core/client.h> #include <core/ramht.h> #include <subdev/instmem.h> +#include <subdev/mc.h> #include <subdev/timer.h> #include <engine/sw.h> -static const struct nv04_fifo_ramfc -nv04_fifo_ramfc[] = { - { 32, 0, 0x00, 0, NV04_PFIFO_CACHE1_DMA_PUT }, - { 32, 0, 0x04, 0, NV04_PFIFO_CACHE1_DMA_GET }, - { 16, 0, 0x08, 0, NV04_PFIFO_CACHE1_DMA_INSTANCE }, - { 16, 16, 0x08, 0, NV04_PFIFO_CACHE1_DMA_DCOUNT }, - { 32, 0, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_STATE }, - { 32, 0, 0x10, 0, NV04_PFIFO_CACHE1_DMA_FETCH }, - { 32, 0, 0x14, 0, NV04_PFIFO_CACHE1_ENGINE }, - { 32, 0, 0x18, 0, NV04_PFIFO_CACHE1_PULL1 }, - {} +#include <nvif/class.h> + +void +nv04_chan_stop(struct nvkm_chan *chan) +{ + struct nvkm_fifo *fifo = chan->cgrp->runl->fifo; + struct nvkm_device *device = fifo->engine.subdev.device; + struct nvkm_memory *fctx = device->imem->ramfc; + const struct nvkm_ramfc_layout *c; + unsigned long flags; + u32 data = chan->ramfc_offset; + u32 chid; + + /* prevent fifo context switches */ + spin_lock_irqsave(&fifo->lock, flags); + nvkm_wr32(device, NV03_PFIFO_CACHES, 0); + + /* if this channel is active, replace it with a null context */ + chid = nvkm_rd32(device, NV03_PFIFO_CACHE1_PUSH1) & fifo->chid->mask; + if (chid == chan->id) { + nvkm_mask(device, NV04_PFIFO_CACHE1_DMA_PUSH, 0x00000001, 0); + nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH0, 0); + nvkm_mask(device, NV04_PFIFO_CACHE1_PULL0, 0x00000001, 0); + + c = chan->func->ramfc->layout; + nvkm_kmap(fctx); + do { + u32 rm = ((1ULL << c->bits) - 1) << c->regs; + u32 cm = ((1ULL << c->bits) - 1) << c->ctxs; + u32 rv = (nvkm_rd32(device, c->regp) & rm) >> c->regs; + u32 cv = (nvkm_ro32(fctx, c->ctxp + data) & ~cm); + nvkm_wo32(fctx, c->ctxp + data, cv | (rv << c->ctxs)); + } while ((++c)->bits); + nvkm_done(fctx); + + c = chan->func->ramfc->layout; + do { + nvkm_wr32(device, c->regp, 0x00000000); + } while ((++c)->bits); + + nvkm_wr32(device, NV03_PFIFO_CACHE1_GET, 0); + nvkm_wr32(device, NV03_PFIFO_CACHE1_PUT, 0); + nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH1, fifo->chid->mask); + nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH0, 1); + nvkm_wr32(device, NV04_PFIFO_CACHE1_PULL0, 1); + } + + /* restore normal operation, after disabling dma mode */ + nvkm_mask(device, NV04_PFIFO_MODE, BIT(chan->id), 0); + nvkm_wr32(device, NV03_PFIFO_CACHES, 1); + spin_unlock_irqrestore(&fifo->lock, flags); +} + +void +nv04_chan_start(struct nvkm_chan *chan) +{ + struct nvkm_fifo *fifo = chan->cgrp->runl->fifo; + unsigned long flags; + + spin_lock_irqsave(&fifo->lock, flags); + nvkm_mask(fifo->engine.subdev.device, NV04_PFIFO_MODE, BIT(chan->id), BIT(chan->id)); + spin_unlock_irqrestore(&fifo->lock, flags); +} + +void +nv04_chan_ramfc_clear(struct nvkm_chan *chan) +{ + struct nvkm_memory *ramfc = chan->cgrp->runl->fifo->engine.subdev.device->imem->ramfc; + const struct nvkm_ramfc_layout *c = chan->func->ramfc->layout; + + nvkm_kmap(ramfc); + do { + nvkm_wo32(ramfc, chan->ramfc_offset + c->ctxp, 0x00000000); + } while ((++c)->bits); + nvkm_done(ramfc); +} + +static int +nv04_chan_ramfc_write(struct nvkm_chan *chan, u64 offset, u64 length, u32 devm, bool priv) +{ + struct nvkm_memory *ramfc = chan->cgrp->runl->fifo->engine.subdev.device->imem->ramfc; + const u32 base = chan->id * 32; + + chan->ramfc_offset = base; + + nvkm_kmap(ramfc); + nvkm_wo32(ramfc, base + 0x00, offset); + nvkm_wo32(ramfc, base + 0x04, offset); + nvkm_wo32(ramfc, base + 0x08, chan->push->addr >> 4); + nvkm_wo32(ramfc, base + 0x10, NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES | + NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES | +#ifdef __BIG_ENDIAN + NV_PFIFO_CACHE1_BIG_ENDIAN | +#endif + NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8); + nvkm_done(ramfc); + return 0; +} + +static const struct nvkm_chan_func_ramfc +nv04_chan_ramfc = { + .layout = (const struct nvkm_ramfc_layout[]) { + { 32, 0, 0x00, 0, NV04_PFIFO_CACHE1_DMA_PUT }, + { 32, 0, 0x04, 0, NV04_PFIFO_CACHE1_DMA_GET }, + { 16, 0, 0x08, 0, NV04_PFIFO_CACHE1_DMA_INSTANCE }, + { 16, 16, 0x08, 0, NV04_PFIFO_CACHE1_DMA_DCOUNT }, + { 32, 0, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_STATE }, + { 32, 0, 0x10, 0, NV04_PFIFO_CACHE1_DMA_FETCH }, + { 32, 0, 0x14, 0, NV04_PFIFO_CACHE1_ENGINE }, + { 32, 0, 0x18, 0, NV04_PFIFO_CACHE1_PULL1 }, + {} + }, + .write = nv04_chan_ramfc_write, + .clear = nv04_chan_ramfc_clear, + .ctxdma = true, +}; + +const struct nvkm_chan_func_userd +nv04_chan_userd = { + .bar = 0, + .base = 0x800000, + .size = 0x010000, +}; + +const struct nvkm_chan_func_inst +nv04_chan_inst = { + .size = 0x1000, +}; + +static const struct nvkm_chan_func +nv04_chan = { + .inst = &nv04_chan_inst, + .userd = &nv04_chan_userd, + .ramfc = &nv04_chan_ramfc, + .start = nv04_chan_start, + .stop = nv04_chan_stop, +}; + +const struct nvkm_cgrp_func +nv04_cgrp = { +}; + +void +nv04_eobj_ramht_del(struct nvkm_chan *chan, int hash) +{ + struct nvkm_fifo *fifo = chan->cgrp->runl->fifo; + struct nvkm_instmem *imem = fifo->engine.subdev.device->imem; + + mutex_lock(&fifo->mutex); + nvkm_ramht_remove(imem->ramht, hash); + mutex_unlock(&fifo->mutex); +} + +static int +nv04_eobj_ramht_add(struct nvkm_engn *engn, struct nvkm_object *eobj, struct nvkm_chan *chan) +{ + struct nvkm_fifo *fifo = chan->cgrp->runl->fifo; + struct nvkm_instmem *imem = fifo->engine.subdev.device->imem; + u32 context = 0x80000000 | chan->id << 24 | engn->id << 16; + int hash; + + mutex_lock(&fifo->mutex); + hash = nvkm_ramht_insert(imem->ramht, eobj, chan->id, 4, eobj->handle, context); + mutex_unlock(&fifo->mutex); + return hash; +} + +const struct nvkm_engn_func +nv04_engn = { + .ramht_add = nv04_eobj_ramht_add, + .ramht_del = nv04_eobj_ramht_del, }; void -nv04_fifo_pause(struct nvkm_fifo *base, unsigned long *pflags) -__acquires(fifo->base.lock) +nv04_fifo_pause(struct nvkm_fifo *fifo, unsigned long *pflags) +__acquires(fifo->lock) { - struct nv04_fifo *fifo = nv04_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; + struct nvkm_device *device = fifo->engine.subdev.device; unsigned long flags; - spin_lock_irqsave(&fifo->base.lock, flags); + spin_lock_irqsave(&fifo->lock, flags); *pflags = flags; nvkm_wr32(device, NV03_PFIFO_CACHES, 0x00000000); @@ -81,50 +244,21 @@ __acquires(fifo->base.lock) } void -nv04_fifo_start(struct nvkm_fifo *base, unsigned long *pflags) -__releases(fifo->base.lock) +nv04_fifo_start(struct nvkm_fifo *fifo, unsigned long *pflags) +__releases(fifo->lock) { - struct nv04_fifo *fifo = nv04_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; + struct nvkm_device *device = fifo->engine.subdev.device; unsigned long flags = *pflags; nvkm_mask(device, NV04_PFIFO_CACHE1_PULL0, 0x00000001, 0x00000001); nvkm_wr32(device, NV03_PFIFO_CACHES, 0x00000001); - spin_unlock_irqrestore(&fifo->base.lock, flags); + spin_unlock_irqrestore(&fifo->lock, flags); } -struct nvkm_engine * -nv04_fifo_id_engine(struct nvkm_fifo *fifo, int engi) -{ - enum nvkm_subdev_type type; - - switch (engi) { - case NV04_FIFO_ENGN_SW : type = NVKM_ENGINE_SW; break; - case NV04_FIFO_ENGN_GR : type = NVKM_ENGINE_GR; break; - case NV04_FIFO_ENGN_MPEG: type = NVKM_ENGINE_MPEG; break; - case NV04_FIFO_ENGN_DMA : type = NVKM_ENGINE_DMAOBJ; break; - default: - WARN_ON(1); - return NULL; - } - - return nvkm_device_engine(fifo->engine.subdev.device, type, 0); -} - -int -nv04_fifo_engine_id(struct nvkm_fifo *base, struct nvkm_engine *engine) -{ - switch (engine->subdev.type) { - case NVKM_ENGINE_SW : return NV04_FIFO_ENGN_SW; - case NVKM_ENGINE_GR : return NV04_FIFO_ENGN_GR; - case NVKM_ENGINE_MPEG : return NV04_FIFO_ENGN_MPEG; - case NVKM_ENGINE_DMAOBJ: return NV04_FIFO_ENGN_DMA; - default: - WARN_ON(1); - return 0; - } -} +const struct nvkm_runl_func +nv04_runl = { +}; static const char * nv_dma_state_err(u32 state) @@ -166,11 +300,11 @@ nv04_fifo_swmthd(struct nvkm_device *device, u32 chid, u32 addr, u32 data) } static void -nv04_fifo_cache_error(struct nv04_fifo *fifo, u32 chid, u32 get) +nv04_fifo_intr_cache_error(struct nvkm_fifo *fifo, u32 chid, u32 get) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; + struct nvkm_subdev *subdev = &fifo->engine.subdev; struct nvkm_device *device = subdev->device; - struct nvkm_fifo_chan *chan; + struct nvkm_chan *chan; unsigned long flags; u32 pull0 = nvkm_rd32(device, 0x003250); u32 mthd, data; @@ -193,12 +327,12 @@ nv04_fifo_cache_error(struct nv04_fifo *fifo, u32 chid, u32 get) if (!(pull0 & 0x00000100) || !nv04_fifo_swmthd(device, chid, mthd, data)) { - chan = nvkm_fifo_chan_chid(&fifo->base, chid, &flags); + chan = nvkm_chan_get_chid(&fifo->engine, chid, &flags); nvkm_error(subdev, "CACHE_ERROR - " "ch %d [%s] subc %d mthd %04x data %08x\n", - chid, chan ? chan->object.client->name : "unknown", + chid, chan ? chan->name : "unknown", (mthd >> 13) & 7, mthd & 0x1ffc, data); - nvkm_fifo_chan_put(&fifo->base, flags, &chan); + nvkm_chan_put(&chan, flags); } nvkm_wr32(device, NV04_PFIFO_CACHE1_DMA_PUSH, 0); @@ -217,20 +351,20 @@ nv04_fifo_cache_error(struct nv04_fifo *fifo, u32 chid, u32 get) } static void -nv04_fifo_dma_pusher(struct nv04_fifo *fifo, u32 chid) +nv04_fifo_intr_dma_pusher(struct nvkm_fifo *fifo, u32 chid) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; + struct nvkm_subdev *subdev = &fifo->engine.subdev; struct nvkm_device *device = subdev->device; u32 dma_get = nvkm_rd32(device, 0x003244); u32 dma_put = nvkm_rd32(device, 0x003240); u32 push = nvkm_rd32(device, 0x003220); u32 state = nvkm_rd32(device, 0x003228); - struct nvkm_fifo_chan *chan; + struct nvkm_chan *chan; unsigned long flags; const char *name; - chan = nvkm_fifo_chan_chid(&fifo->base, chid, &flags); - name = chan ? chan->object.client->name : "unknown"; + chan = nvkm_chan_get_chid(&fifo->engine, chid, &flags); + name = chan ? chan->name : "unknown"; if (device->card_type == NV_50) { u32 ho_get = nvkm_rd32(device, 0x003328); u32 ho_put = nvkm_rd32(device, 0x003320); @@ -261,18 +395,18 @@ nv04_fifo_dma_pusher(struct nv04_fifo *fifo, u32 chid) if (dma_get != dma_put) nvkm_wr32(device, 0x003244, dma_put); } - nvkm_fifo_chan_put(&fifo->base, flags, &chan); + nvkm_chan_put(&chan, flags); nvkm_wr32(device, 0x003228, 0x00000000); nvkm_wr32(device, 0x003220, 0x00000001); nvkm_wr32(device, 0x002100, NV_PFIFO_INTR_DMA_PUSHER); } -void -nv04_fifo_intr(struct nvkm_fifo *base) +irqreturn_t +nv04_fifo_intr(struct nvkm_inth *inth) { - struct nv04_fifo *fifo = nv04_fifo(base); - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; + struct nvkm_fifo *fifo = container_of(inth, typeof(*fifo), engine.subdev.inth); + struct nvkm_subdev *subdev = &fifo->engine.subdev; struct nvkm_device *device = subdev->device; u32 mask = nvkm_rd32(device, NV03_PFIFO_INTR_EN_0); u32 stat = nvkm_rd32(device, NV03_PFIFO_INTR_0) & mask; @@ -281,16 +415,16 @@ nv04_fifo_intr(struct nvkm_fifo *base) reassign = nvkm_rd32(device, NV03_PFIFO_CACHES) & 1; nvkm_wr32(device, NV03_PFIFO_CACHES, 0); - chid = nvkm_rd32(device, NV03_PFIFO_CACHE1_PUSH1) & (fifo->base.nr - 1); + chid = nvkm_rd32(device, NV03_PFIFO_CACHE1_PUSH1) & fifo->chid->mask; get = nvkm_rd32(device, NV03_PFIFO_CACHE1_GET); if (stat & NV_PFIFO_INTR_CACHE_ERROR) { - nv04_fifo_cache_error(fifo, chid, get); + nv04_fifo_intr_cache_error(fifo, chid, get); stat &= ~NV_PFIFO_INTR_CACHE_ERROR; } if (stat & NV_PFIFO_INTR_DMA_PUSHER) { - nv04_fifo_dma_pusher(fifo, chid); + nv04_fifo_intr_dma_pusher(fifo, chid); stat &= ~NV_PFIFO_INTR_DMA_PUSHER; } @@ -313,7 +447,7 @@ nv04_fifo_intr(struct nvkm_fifo *base) if (stat & 0x40000000) { nvkm_wr32(device, 0x002100, 0x40000000); - nvkm_fifo_uevent(&fifo->base); + nvkm_event_ntfy(&fifo->nonstall.event, 0, NVKM_FIFO_NONSTALL_EVENT); stat &= ~0x40000000; } } @@ -325,13 +459,13 @@ nv04_fifo_intr(struct nvkm_fifo *base) } nvkm_wr32(device, NV03_PFIFO_CACHES, reassign); + return IRQ_HANDLED; } void -nv04_fifo_init(struct nvkm_fifo *base) +nv04_fifo_init(struct nvkm_fifo *fifo) { - struct nv04_fifo *fifo = nv04_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; + struct nvkm_device *device = fifo->engine.subdev.device; struct nvkm_instmem *imem = device->imem; struct nvkm_ramht *ramht = imem->ramht; struct nvkm_memory *ramro = imem->ramro; @@ -346,7 +480,7 @@ nv04_fifo_init(struct nvkm_fifo *base) nvkm_wr32(device, NV03_PFIFO_RAMRO, nvkm_memory_addr(ramro) >> 8); nvkm_wr32(device, NV03_PFIFO_RAMFC, nvkm_memory_addr(ramfc) >> 8); - nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH1, fifo->base.nr - 1); + nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH1, fifo->chid->mask); nvkm_wr32(device, NV03_PFIFO_INTR_0, 0xffffffff); nvkm_wr32(device, NV03_PFIFO_INTR_EN_0, 0xffffffff); @@ -357,43 +491,53 @@ nv04_fifo_init(struct nvkm_fifo *base) } int -nv04_fifo_new_(const struct nvkm_fifo_func *func, struct nvkm_device *device, - enum nvkm_subdev_type type, int inst, int nr, const struct nv04_fifo_ramfc *ramfc, - struct nvkm_fifo **pfifo) +nv04_fifo_runl_ctor(struct nvkm_fifo *fifo) { - struct nv04_fifo *fifo; - int ret; - - if (!(fifo = kzalloc(sizeof(*fifo), GFP_KERNEL))) - return -ENOMEM; - fifo->ramfc = ramfc; - *pfifo = &fifo->base; + struct nvkm_runl *runl; - ret = nvkm_fifo_ctor(func, device, type, inst, nr, &fifo->base); - if (ret) - return ret; + runl = nvkm_runl_new(fifo, 0, 0, 0); + if (IS_ERR(runl)) + return PTR_ERR(runl); - set_bit(nr - 1, fifo->base.mask); /* inactive channel */ + nvkm_runl_add(runl, 0, fifo->func->engn_sw, NVKM_ENGINE_SW, 0); + nvkm_runl_add(runl, 0, fifo->func->engn_sw, NVKM_ENGINE_DMAOBJ, 0); + nvkm_runl_add(runl, 1, fifo->func->engn , NVKM_ENGINE_GR, 0); + nvkm_runl_add(runl, 2, fifo->func->engn , NVKM_ENGINE_MPEG, 0); /* NV31- */ return 0; } +int +nv04_fifo_chid_ctor(struct nvkm_fifo *fifo, int nr) +{ + /* The last CHID is reserved by HW as a "channel invalid" marker. */ + return nvkm_chid_new(&nvkm_chan_event, &fifo->engine.subdev, nr, 0, nr - 1, &fifo->chid); +} + +static int +nv04_fifo_chid_nr(struct nvkm_fifo *fifo) +{ + return 16; +} + static const struct nvkm_fifo_func nv04_fifo = { + .chid_nr = nv04_fifo_chid_nr, + .chid_ctor = nv04_fifo_chid_ctor, + .runl_ctor = nv04_fifo_runl_ctor, .init = nv04_fifo_init, .intr = nv04_fifo_intr, - .engine_id = nv04_fifo_engine_id, - .id_engine = nv04_fifo_id_engine, .pause = nv04_fifo_pause, .start = nv04_fifo_start, - .chan = { - &nv04_fifo_dma_oclass, - NULL - }, + .runl = &nv04_runl, + .engn = &nv04_engn, + .engn_sw = &nv04_engn, + .cgrp = {{ }, &nv04_cgrp }, + .chan = {{ 0, 0, NV03_CHANNEL_DMA }, &nv04_chan }, }; int nv04_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return nv04_fifo_new_(&nv04_fifo, device, type, inst, 16, nv04_fifo_ramfc, pfifo); + return nvkm_fifo_new_(&nv04_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv04.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv04.h deleted file mode 100644 index 3f23bcde4a54..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv04.h +++ /dev/null @@ -1,23 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NV04_FIFO_H__ -#define __NV04_FIFO_H__ -#define nv04_fifo(p) container_of((p), struct nv04_fifo, base) -#include "priv.h" - -struct nv04_fifo_ramfc { - unsigned bits:6; - unsigned ctxs:5; - unsigned ctxp:8; - unsigned regs:5; - unsigned regp; -}; - -struct nv04_fifo { - struct nvkm_fifo base; - const struct nv04_fifo_ramfc *ramfc; -}; - -int nv04_fifo_new_(const struct nvkm_fifo_func *, struct nvkm_device *, enum nvkm_subdev_type, int, - int nr, const struct nv04_fifo_ramfc *, struct nvkm_fifo **); -void nv04_fifo_init(struct nvkm_fifo *); -#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv10.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv10.c index f8887f0f2f82..a4bcf6b0a7e2 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv10.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv10.c @@ -21,41 +21,93 @@ * * Authors: Ben Skeggs */ -#include "nv04.h" -#include "channv04.h" +#include "priv.h" +#include "cgrp.h" +#include "chan.h" +#include "runl.h" + +#include <core/gpuobj.h> +#include <subdev/instmem.h> + #include "regsnv04.h" -static const struct nv04_fifo_ramfc -nv10_fifo_ramfc[] = { - { 32, 0, 0x00, 0, NV04_PFIFO_CACHE1_DMA_PUT }, - { 32, 0, 0x04, 0, NV04_PFIFO_CACHE1_DMA_GET }, - { 32, 0, 0x08, 0, NV10_PFIFO_CACHE1_REF_CNT }, - { 16, 0, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_INSTANCE }, - { 16, 16, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_DCOUNT }, - { 32, 0, 0x10, 0, NV04_PFIFO_CACHE1_DMA_STATE }, - { 32, 0, 0x14, 0, NV04_PFIFO_CACHE1_DMA_FETCH }, - { 32, 0, 0x18, 0, NV04_PFIFO_CACHE1_ENGINE }, - { 32, 0, 0x1c, 0, NV04_PFIFO_CACHE1_PULL1 }, - {} +#include <nvif/class.h> + +static int +nv10_chan_ramfc_write(struct nvkm_chan *chan, u64 offset, u64 length, u32 devm, bool priv) +{ + struct nvkm_memory *ramfc = chan->cgrp->runl->fifo->engine.subdev.device->imem->ramfc; + const u32 base = chan->id * 32; + + chan->ramfc_offset = base; + + nvkm_kmap(ramfc); + nvkm_wo32(ramfc, base + 0x00, offset); + nvkm_wo32(ramfc, base + 0x04, offset); + nvkm_wo32(ramfc, base + 0x0c, chan->push->addr >> 4); + nvkm_wo32(ramfc, base + 0x14, NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES | + NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES | +#ifdef __BIG_ENDIAN + NV_PFIFO_CACHE1_BIG_ENDIAN | +#endif + NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8); + nvkm_done(ramfc); + return 0; +} + +static const struct nvkm_chan_func_ramfc +nv10_chan_ramfc = { + .layout = (const struct nvkm_ramfc_layout[]) { + { 32, 0, 0x00, 0, NV04_PFIFO_CACHE1_DMA_PUT }, + { 32, 0, 0x04, 0, NV04_PFIFO_CACHE1_DMA_GET }, + { 32, 0, 0x08, 0, NV10_PFIFO_CACHE1_REF_CNT }, + { 16, 0, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_INSTANCE }, + { 16, 16, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_DCOUNT }, + { 32, 0, 0x10, 0, NV04_PFIFO_CACHE1_DMA_STATE }, + { 32, 0, 0x14, 0, NV04_PFIFO_CACHE1_DMA_FETCH }, + { 32, 0, 0x18, 0, NV04_PFIFO_CACHE1_ENGINE }, + { 32, 0, 0x1c, 0, NV04_PFIFO_CACHE1_PULL1 }, + {} + }, + .write = nv10_chan_ramfc_write, + .clear = nv04_chan_ramfc_clear, + .ctxdma = true, +}; + +static const struct nvkm_chan_func +nv10_chan = { + .inst = &nv04_chan_inst, + .userd = &nv04_chan_userd, + .ramfc = &nv10_chan_ramfc, + .start = nv04_chan_start, + .stop = nv04_chan_stop, }; +int +nv10_fifo_chid_nr(struct nvkm_fifo *fifo) +{ + return 32; +} + static const struct nvkm_fifo_func nv10_fifo = { + .chid_nr = nv10_fifo_chid_nr, + .chid_ctor = nv04_fifo_chid_ctor, + .runl_ctor = nv04_fifo_runl_ctor, .init = nv04_fifo_init, .intr = nv04_fifo_intr, - .engine_id = nv04_fifo_engine_id, - .id_engine = nv04_fifo_id_engine, .pause = nv04_fifo_pause, .start = nv04_fifo_start, - .chan = { - &nv10_fifo_dma_oclass, - NULL - }, + .runl = &nv04_runl, + .engn = &nv04_engn, + .engn_sw = &nv04_engn, + .cgrp = {{ }, &nv04_cgrp }, + .chan = {{ 0, 0, NV10_CHANNEL_DMA }, &nv10_chan }, }; int nv10_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return nv04_fifo_new_(&nv10_fifo, device, type, inst, 32, nv10_fifo_ramfc, pfifo); + return nvkm_fifo_new_(&nv10_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv17.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv17.c index 3f94c7b5b054..c70f44fd4f3b 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv17.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv17.c @@ -21,37 +21,78 @@ * * Authors: Ben Skeggs */ -#include "nv04.h" -#include "channv04.h" +#include "priv.h" +#include "cgrp.h" +#include "chan.h" +#include "chid.h" +#include "runl.h" + #include "regsnv04.h" #include <core/ramht.h> #include <subdev/instmem.h> -static const struct nv04_fifo_ramfc -nv17_fifo_ramfc[] = { - { 32, 0, 0x00, 0, NV04_PFIFO_CACHE1_DMA_PUT }, - { 32, 0, 0x04, 0, NV04_PFIFO_CACHE1_DMA_GET }, - { 32, 0, 0x08, 0, NV10_PFIFO_CACHE1_REF_CNT }, - { 16, 0, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_INSTANCE }, - { 16, 16, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_DCOUNT }, - { 32, 0, 0x10, 0, NV04_PFIFO_CACHE1_DMA_STATE }, - { 32, 0, 0x14, 0, NV04_PFIFO_CACHE1_DMA_FETCH }, - { 32, 0, 0x18, 0, NV04_PFIFO_CACHE1_ENGINE }, - { 32, 0, 0x1c, 0, NV04_PFIFO_CACHE1_PULL1 }, - { 32, 0, 0x20, 0, NV10_PFIFO_CACHE1_ACQUIRE_VALUE }, - { 32, 0, 0x24, 0, NV10_PFIFO_CACHE1_ACQUIRE_TIMESTAMP }, - { 32, 0, 0x28, 0, NV10_PFIFO_CACHE1_ACQUIRE_TIMEOUT }, - { 32, 0, 0x2c, 0, NV10_PFIFO_CACHE1_SEMAPHORE }, - { 32, 0, 0x30, 0, NV10_PFIFO_CACHE1_DMA_SUBROUTINE }, - {} +#include <nvif/class.h> + +static int +nv17_chan_ramfc_write(struct nvkm_chan *chan, u64 offset, u64 length, u32 devm, bool priv) +{ + struct nvkm_memory *ramfc = chan->cgrp->runl->fifo->engine.subdev.device->imem->ramfc; + const u32 base = chan->id * 64; + + chan->ramfc_offset = base; + + nvkm_kmap(ramfc); + nvkm_wo32(ramfc, base + 0x00, offset); + nvkm_wo32(ramfc, base + 0x04, offset); + nvkm_wo32(ramfc, base + 0x0c, chan->push->addr >> 4); + nvkm_wo32(ramfc, base + 0x14, NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES | + NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES | +#ifdef __BIG_ENDIAN + NV_PFIFO_CACHE1_BIG_ENDIAN | +#endif + NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8); + nvkm_done(ramfc); + return 0; +} + +static const struct nvkm_chan_func_ramfc +nv17_chan_ramfc = { + .layout = (const struct nvkm_ramfc_layout[]) { + { 32, 0, 0x00, 0, NV04_PFIFO_CACHE1_DMA_PUT }, + { 32, 0, 0x04, 0, NV04_PFIFO_CACHE1_DMA_GET }, + { 32, 0, 0x08, 0, NV10_PFIFO_CACHE1_REF_CNT }, + { 16, 0, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_INSTANCE }, + { 16, 16, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_DCOUNT }, + { 32, 0, 0x10, 0, NV04_PFIFO_CACHE1_DMA_STATE }, + { 32, 0, 0x14, 0, NV04_PFIFO_CACHE1_DMA_FETCH }, + { 32, 0, 0x18, 0, NV04_PFIFO_CACHE1_ENGINE }, + { 32, 0, 0x1c, 0, NV04_PFIFO_CACHE1_PULL1 }, + { 32, 0, 0x20, 0, NV10_PFIFO_CACHE1_ACQUIRE_VALUE }, + { 32, 0, 0x24, 0, NV10_PFIFO_CACHE1_ACQUIRE_TIMESTAMP }, + { 32, 0, 0x28, 0, NV10_PFIFO_CACHE1_ACQUIRE_TIMEOUT }, + { 32, 0, 0x2c, 0, NV10_PFIFO_CACHE1_SEMAPHORE }, + { 32, 0, 0x30, 0, NV10_PFIFO_CACHE1_DMA_SUBROUTINE }, + {} + }, + .write = nv17_chan_ramfc_write, + .clear = nv04_chan_ramfc_clear, + .ctxdma = true, +}; + +static const struct nvkm_chan_func +nv17_chan = { + .inst = &nv04_chan_inst, + .userd = &nv04_chan_userd, + .ramfc = &nv17_chan_ramfc, + .start = nv04_chan_start, + .stop = nv04_chan_stop, }; static void -nv17_fifo_init(struct nvkm_fifo *base) +nv17_fifo_init(struct nvkm_fifo *fifo) { - struct nv04_fifo *fifo = nv04_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; + struct nvkm_device *device = fifo->engine.subdev.device; struct nvkm_instmem *imem = device->imem; struct nvkm_ramht *ramht = imem->ramht; struct nvkm_memory *ramro = imem->ramro; @@ -67,7 +108,7 @@ nv17_fifo_init(struct nvkm_fifo *base) nvkm_wr32(device, NV03_PFIFO_RAMFC, nvkm_memory_addr(ramfc) >> 8 | 0x00010000); - nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH1, fifo->base.nr - 1); + nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH1, fifo->chid->mask); nvkm_wr32(device, NV03_PFIFO_INTR_0, 0xffffffff); nvkm_wr32(device, NV03_PFIFO_INTR_EN_0, 0xffffffff); @@ -79,21 +120,23 @@ nv17_fifo_init(struct nvkm_fifo *base) static const struct nvkm_fifo_func nv17_fifo = { + .chid_nr = nv10_fifo_chid_nr, + .chid_ctor = nv04_fifo_chid_ctor, + .runl_ctor = nv04_fifo_runl_ctor, .init = nv17_fifo_init, .intr = nv04_fifo_intr, - .engine_id = nv04_fifo_engine_id, - .id_engine = nv04_fifo_id_engine, .pause = nv04_fifo_pause, .start = nv04_fifo_start, - .chan = { - &nv17_fifo_dma_oclass, - NULL - }, + .runl = &nv04_runl, + .engn = &nv04_engn, + .engn_sw = &nv04_engn, + .cgrp = {{ }, &nv04_cgrp }, + .chan = {{ 0, 0, NV17_CHANNEL_DMA }, &nv17_chan }, }; int nv17_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return nv04_fifo_new_(&nv17_fifo, device, type, inst, 32, nv17_fifo_ramfc, pfifo); + return nvkm_fifo_new_(&nv17_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv40.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv40.c index f9ea46809bc0..e50a94b6d7f8 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv40.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv40.c @@ -21,46 +21,166 @@ * * Authors: Ben Skeggs */ -#include "nv04.h" -#include "channv04.h" +#include "priv.h" +#include "cgrp.h" +#include "chan.h" +#include "chid.h" +#include "runl.h" + #include "regsnv04.h" #include <core/ramht.h> #include <subdev/fb.h> #include <subdev/instmem.h> -static const struct nv04_fifo_ramfc -nv40_fifo_ramfc[] = { - { 32, 0, 0x00, 0, NV04_PFIFO_CACHE1_DMA_PUT }, - { 32, 0, 0x04, 0, NV04_PFIFO_CACHE1_DMA_GET }, - { 32, 0, 0x08, 0, NV10_PFIFO_CACHE1_REF_CNT }, - { 32, 0, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_INSTANCE }, - { 32, 0, 0x10, 0, NV04_PFIFO_CACHE1_DMA_DCOUNT }, - { 32, 0, 0x14, 0, NV04_PFIFO_CACHE1_DMA_STATE }, - { 28, 0, 0x18, 0, NV04_PFIFO_CACHE1_DMA_FETCH }, - { 2, 28, 0x18, 28, 0x002058 }, - { 32, 0, 0x1c, 0, NV04_PFIFO_CACHE1_ENGINE }, - { 32, 0, 0x20, 0, NV04_PFIFO_CACHE1_PULL1 }, - { 32, 0, 0x24, 0, NV10_PFIFO_CACHE1_ACQUIRE_VALUE }, - { 32, 0, 0x28, 0, NV10_PFIFO_CACHE1_ACQUIRE_TIMESTAMP }, - { 32, 0, 0x2c, 0, NV10_PFIFO_CACHE1_ACQUIRE_TIMEOUT }, - { 32, 0, 0x30, 0, NV10_PFIFO_CACHE1_SEMAPHORE }, - { 32, 0, 0x34, 0, NV10_PFIFO_CACHE1_DMA_SUBROUTINE }, - { 32, 0, 0x38, 0, NV40_PFIFO_GRCTX_INSTANCE }, - { 17, 0, 0x3c, 0, NV04_PFIFO_DMA_TIMESLICE }, - { 32, 0, 0x40, 0, 0x0032e4 }, - { 32, 0, 0x44, 0, 0x0032e8 }, - { 32, 0, 0x4c, 0, 0x002088 }, - { 32, 0, 0x50, 0, 0x003300 }, - { 32, 0, 0x54, 0, 0x00330c }, - {} +#include <nvif/class.h> + +static int +nv40_chan_ramfc_write(struct nvkm_chan *chan, u64 offset, u64 length, u32 devm, bool priv) +{ + struct nvkm_memory *ramfc = chan->cgrp->runl->fifo->engine.subdev.device->imem->ramfc; + const u32 base = chan->id * 128; + + chan->ramfc_offset = base; + + nvkm_kmap(ramfc); + nvkm_wo32(ramfc, base + 0x00, offset); + nvkm_wo32(ramfc, base + 0x04, offset); + nvkm_wo32(ramfc, base + 0x0c, chan->push->addr >> 4); + nvkm_wo32(ramfc, base + 0x18, 0x30000000 | + NV_PFIFO_CACHE1_DMA_FETCH_TRIG_128_BYTES | + NV_PFIFO_CACHE1_DMA_FETCH_SIZE_128_BYTES | +#ifdef __BIG_ENDIAN + NV_PFIFO_CACHE1_BIG_ENDIAN | +#endif + NV_PFIFO_CACHE1_DMA_FETCH_MAX_REQS_8); + nvkm_wo32(ramfc, base + 0x3c, 0x0001ffff); + nvkm_done(ramfc); + return 0; +} + +static const struct nvkm_chan_func_ramfc +nv40_chan_ramfc = { + .layout = (const struct nvkm_ramfc_layout[]) { + { 32, 0, 0x00, 0, NV04_PFIFO_CACHE1_DMA_PUT }, + { 32, 0, 0x04, 0, NV04_PFIFO_CACHE1_DMA_GET }, + { 32, 0, 0x08, 0, NV10_PFIFO_CACHE1_REF_CNT }, + { 32, 0, 0x0c, 0, NV04_PFIFO_CACHE1_DMA_INSTANCE }, + { 32, 0, 0x10, 0, NV04_PFIFO_CACHE1_DMA_DCOUNT }, + { 32, 0, 0x14, 0, NV04_PFIFO_CACHE1_DMA_STATE }, + { 28, 0, 0x18, 0, NV04_PFIFO_CACHE1_DMA_FETCH }, + { 2, 28, 0x18, 28, 0x002058 }, + { 32, 0, 0x1c, 0, NV04_PFIFO_CACHE1_ENGINE }, + { 32, 0, 0x20, 0, NV04_PFIFO_CACHE1_PULL1 }, + { 32, 0, 0x24, 0, NV10_PFIFO_CACHE1_ACQUIRE_VALUE }, + { 32, 0, 0x28, 0, NV10_PFIFO_CACHE1_ACQUIRE_TIMESTAMP }, + { 32, 0, 0x2c, 0, NV10_PFIFO_CACHE1_ACQUIRE_TIMEOUT }, + { 32, 0, 0x30, 0, NV10_PFIFO_CACHE1_SEMAPHORE }, + { 32, 0, 0x34, 0, NV10_PFIFO_CACHE1_DMA_SUBROUTINE }, + { 32, 0, 0x38, 0, NV40_PFIFO_GRCTX_INSTANCE }, + { 17, 0, 0x3c, 0, NV04_PFIFO_DMA_TIMESLICE }, + { 32, 0, 0x40, 0, 0x0032e4 }, + { 32, 0, 0x44, 0, 0x0032e8 }, + { 32, 0, 0x4c, 0, 0x002088 }, + { 32, 0, 0x50, 0, 0x003300 }, + { 32, 0, 0x54, 0, 0x00330c }, + {} + }, + .write = nv40_chan_ramfc_write, + .clear = nv04_chan_ramfc_clear, + .ctxdma = true, +}; + +static const struct nvkm_chan_func_userd +nv40_chan_userd = { + .bar = 0, + .base = 0xc00000, + .size = 0x001000, +}; + +static const struct nvkm_chan_func +nv40_chan = { + .inst = &nv04_chan_inst, + .userd = &nv40_chan_userd, + .ramfc = &nv40_chan_ramfc, + .start = nv04_chan_start, + .stop = nv04_chan_stop, }; +static int +nv40_eobj_ramht_add(struct nvkm_engn *engn, struct nvkm_object *eobj, struct nvkm_chan *chan) +{ + struct nvkm_fifo *fifo = chan->cgrp->runl->fifo; + struct nvkm_instmem *imem = fifo->engine.subdev.device->imem; + u32 context = chan->id << 23 | engn->id << 20; + int hash; + + mutex_lock(&fifo->mutex); + hash = nvkm_ramht_insert(imem->ramht, eobj, chan->id, 4, eobj->handle, context); + mutex_unlock(&fifo->mutex); + return hash; +} + static void -nv40_fifo_init(struct nvkm_fifo *base) +nv40_ectx_bind(struct nvkm_engn *engn, struct nvkm_cctx *cctx, struct nvkm_chan *chan) { - struct nv04_fifo *fifo = nv04_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; + struct nvkm_fifo *fifo = chan->cgrp->runl->fifo; + struct nvkm_device *device = fifo->engine.subdev.device; + struct nvkm_memory *ramfc = device->imem->ramfc; + u32 inst = 0x00000000, reg, ctx; + int chid; + + switch (engn->engine->subdev.type) { + case NVKM_ENGINE_GR: + reg = 0x0032e0; + ctx = 0x38; + break; + case NVKM_ENGINE_MPEG: + if (WARN_ON(device->chipset < 0x44)) + return; + reg = 0x00330c; + ctx = 0x54; + break; + default: + WARN_ON(1); + return; + } + + if (cctx) + inst = cctx->vctx->inst->addr >> 4; + + spin_lock_irq(&fifo->lock); + nvkm_mask(device, 0x002500, 0x00000001, 0x00000000); + + chid = nvkm_rd32(device, 0x003204) & (fifo->chid->nr - 1); + if (chid == chan->id) + nvkm_wr32(device, reg, inst); + + nvkm_kmap(ramfc); + nvkm_wo32(ramfc, chan->ramfc_offset + ctx, inst); + nvkm_done(ramfc); + + nvkm_mask(device, 0x002500, 0x00000001, 0x00000001); + spin_unlock_irq(&fifo->lock); +} + +static const struct nvkm_engn_func +nv40_engn = { + .bind = nv40_ectx_bind, + .ramht_add = nv40_eobj_ramht_add, + .ramht_del = nv04_eobj_ramht_del, +}; + +static const struct nvkm_engn_func +nv40_engn_sw = { + .ramht_add = nv40_eobj_ramht_add, + .ramht_del = nv04_eobj_ramht_del, +}; + +static void +nv40_fifo_init(struct nvkm_fifo *fifo) +{ + struct nvkm_device *device = fifo->engine.subdev.device; struct nvkm_fb *fb = device->fb; struct nvkm_instmem *imem = device->imem; struct nvkm_ramht *ramht = imem->ramht; @@ -98,7 +218,7 @@ nv40_fifo_init(struct nvkm_fifo *base) break; } - nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH1, fifo->base.nr - 1); + nvkm_wr32(device, NV03_PFIFO_CACHE1_PUSH1, fifo->chid->mask); nvkm_wr32(device, NV03_PFIFO_INTR_0, 0xffffffff); nvkm_wr32(device, NV03_PFIFO_INTR_EN_0, 0xffffffff); @@ -110,21 +230,23 @@ nv40_fifo_init(struct nvkm_fifo *base) static const struct nvkm_fifo_func nv40_fifo = { + .chid_nr = nv10_fifo_chid_nr, + .chid_ctor = nv04_fifo_chid_ctor, + .runl_ctor = nv04_fifo_runl_ctor, .init = nv40_fifo_init, .intr = nv04_fifo_intr, - .engine_id = nv04_fifo_engine_id, - .id_engine = nv04_fifo_id_engine, .pause = nv04_fifo_pause, .start = nv04_fifo_start, - .chan = { - &nv40_fifo_dma_oclass, - NULL - }, + .runl = &nv04_runl, + .engn = &nv40_engn, + .engn_sw = &nv40_engn_sw, + .cgrp = {{ }, &nv04_cgrp }, + .chan = {{ 0, 0, NV40_CHANNEL_DMA }, &nv40_chan }, }; int nv40_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return nv04_fifo_new_(&nv40_fifo, device, type, inst, 32, nv40_fifo_ramfc, pfifo); + return nvkm_fifo_new_(&nv40_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv50.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv50.c index a08742cf425a..954b5f3a7d57 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv50.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv50.c @@ -21,62 +21,325 @@ * * Authors: Ben Skeggs */ -#include "nv50.h" -#include "channv50.h" +#include "priv.h" +#include "cgrp.h" +#include "chan.h" +#include "chid.h" +#include "runl.h" -#include <core/gpuobj.h> +#include <core/ramht.h> +#include <subdev/timer.h> -static void -nv50_fifo_runlist_update_locked(struct nv50_fifo *fifo) +#include <nvif/class.h> + +void +nv50_eobj_ramht_del(struct nvkm_chan *chan, int hash) { - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_memory *cur; - int i, p; + nvkm_ramht_remove(chan->ramht, hash); +} - cur = fifo->runlist[fifo->cur_runlist]; - fifo->cur_runlist = !fifo->cur_runlist; +int +nv50_eobj_ramht_add(struct nvkm_engn *engn, struct nvkm_object *eobj, struct nvkm_chan *chan) +{ + return nvkm_ramht_insert(chan->ramht, eobj, 0, 4, eobj->handle, engn->id << 20); +} - nvkm_kmap(cur); - for (i = 0, p = 0; i < fifo->base.nr; i++) { - if (nvkm_rd32(device, 0x002600 + (i * 4)) & 0x80000000) - nvkm_wo32(cur, p++ * 4, i); - } - nvkm_done(cur); +void +nv50_chan_stop(struct nvkm_chan *chan) +{ + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; - nvkm_wr32(device, 0x0032f4, nvkm_memory_addr(cur) >> 12); - nvkm_wr32(device, 0x0032ec, p); - nvkm_wr32(device, 0x002500, 0x00000101); + nvkm_mask(device, 0x002600 + (chan->id * 4), 0x80000000, 0x00000000); } void -nv50_fifo_runlist_update(struct nv50_fifo *fifo) +nv50_chan_start(struct nvkm_chan *chan) { - mutex_lock(&fifo->base.mutex); - nv50_fifo_runlist_update_locked(fifo); - mutex_unlock(&fifo->base.mutex); + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; + + nvkm_mask(device, 0x002600 + (chan->id * 4), 0x80000000, 0x80000000); } -int -nv50_fifo_oneinit(struct nvkm_fifo *base) +void +nv50_chan_unbind(struct nvkm_chan *chan) { - struct nv50_fifo *fifo = nv50_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; + + nvkm_wr32(device, 0x002600 + (chan->id * 4), 0x00000000); +} + +static void +nv50_chan_bind(struct nvkm_chan *chan) +{ + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; + + nvkm_wr32(device, 0x002600 + (chan->id * 4), chan->ramfc->addr >> 12); +} + +static int +nv50_chan_ramfc_write(struct nvkm_chan *chan, u64 offset, u64 length, u32 devm, bool priv) +{ + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; + const u32 limit2 = ilog2(length / 8); int ret; - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 128 * 4, 0x1000, - false, &fifo->runlist[0]); + ret = nvkm_gpuobj_new(device, 0x0200, 0x1000, true, chan->inst, &chan->ramfc); + if (ret) + return ret; + + ret = nvkm_gpuobj_new(device, 0x1200, 0, true, chan->inst, &chan->eng); if (ret) return ret; - return nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 128 * 4, 0x1000, - false, &fifo->runlist[1]); + ret = nvkm_gpuobj_new(device, 0x4000, 0, false, chan->inst, &chan->pgd); + if (ret) + return ret; + + ret = nvkm_ramht_new(device, 0x8000, 16, chan->inst, &chan->ramht); + if (ret) + return ret; + + nvkm_kmap(chan->ramfc); + nvkm_wo32(chan->ramfc, 0x3c, 0x403f6078); + nvkm_wo32(chan->ramfc, 0x44, 0x01003fff); + nvkm_wo32(chan->ramfc, 0x48, chan->push->node->offset >> 4); + nvkm_wo32(chan->ramfc, 0x50, lower_32_bits(offset)); + nvkm_wo32(chan->ramfc, 0x54, upper_32_bits(offset) | (limit2 << 16)); + nvkm_wo32(chan->ramfc, 0x60, 0x7fffffff); + nvkm_wo32(chan->ramfc, 0x78, 0x00000000); + nvkm_wo32(chan->ramfc, 0x7c, 0x30000000 | devm); + nvkm_wo32(chan->ramfc, 0x80, ((chan->ramht->bits - 9) << 27) | + (4 << 24) /* SEARCH_FULL */ | + (chan->ramht->gpuobj->node->offset >> 4)); + nvkm_done(chan->ramfc); + return 0; +} + +static const struct nvkm_chan_func_ramfc +nv50_chan_ramfc = { + .write = nv50_chan_ramfc_write, + .ctxdma = true, + .devm = 0xfff, +}; + +const struct nvkm_chan_func_userd +nv50_chan_userd = { + .bar = 0, + .base = 0xc00000, + .size = 0x002000, +}; + +const struct nvkm_chan_func_inst +nv50_chan_inst = { + .size = 0x10000, + .vmm = true, +}; + +static const struct nvkm_chan_func +nv50_chan = { + .inst = &nv50_chan_inst, + .userd = &nv50_chan_userd, + .ramfc = &nv50_chan_ramfc, + .bind = nv50_chan_bind, + .unbind = nv50_chan_unbind, + .start = nv50_chan_start, + .stop = nv50_chan_stop, +}; + +static void +nv50_ectx_bind(struct nvkm_engn *engn, struct nvkm_cctx *cctx, struct nvkm_chan *chan) +{ + struct nvkm_subdev *subdev = &chan->cgrp->runl->fifo->engine.subdev; + struct nvkm_device *device = subdev->device; + u64 start = 0, limit = 0; + u32 flags = 0, ptr0, save; + + switch (engn->engine->subdev.type) { + case NVKM_ENGINE_GR : ptr0 = 0x0000; break; + case NVKM_ENGINE_MPEG : ptr0 = 0x0060; break; + default: + WARN_ON(1); + return; + } + + if (!cctx) { + /* HW bug workaround: + * + * PFIFO will hang forever if the connected engines don't report + * that they've processed the context switch request. + * + * In order for the kickoff to work, we need to ensure all the + * connected engines are in a state where they can answer. + * + * Newer chipsets don't seem to suffer from this issue, and well, + * there's also a "ignore these engines" bitmask reg we can use + * if we hit the issue there.. + */ + save = nvkm_mask(device, 0x00b860, 0x00000001, 0x00000001); + + /* Tell engines to save out contexts. */ + nvkm_wr32(device, 0x0032fc, chan->inst->addr >> 12); + nvkm_msec(device, 2000, + if (nvkm_rd32(device, 0x0032fc) != 0xffffffff) + break; + ); + nvkm_wr32(device, 0x00b860, save); + } else { + flags = 0x00190000; + start = cctx->vctx->inst->addr; + limit = start + cctx->vctx->inst->size - 1; + } + + nvkm_kmap(chan->eng); + nvkm_wo32(chan->eng, ptr0 + 0x00, flags); + nvkm_wo32(chan->eng, ptr0 + 0x04, lower_32_bits(limit)); + nvkm_wo32(chan->eng, ptr0 + 0x08, lower_32_bits(start)); + nvkm_wo32(chan->eng, ptr0 + 0x0c, upper_32_bits(limit) << 24 | + lower_32_bits(start)); + nvkm_wo32(chan->eng, ptr0 + 0x10, 0x00000000); + nvkm_wo32(chan->eng, ptr0 + 0x14, 0x00000000); + nvkm_done(chan->eng); } +static const struct nvkm_engn_func +nv50_engn = { + .bind = nv50_ectx_bind, + .ramht_add = nv50_eobj_ramht_add, + .ramht_del = nv50_eobj_ramht_del, +}; + +const struct nvkm_engn_func +nv50_engn_sw = { + .ramht_add = nv50_eobj_ramht_add, + .ramht_del = nv50_eobj_ramht_del, +}; + +static bool +nv50_runl_pending(struct nvkm_runl *runl) +{ + return nvkm_rd32(runl->fifo->engine.subdev.device, 0x0032ec) & 0x00000100; +} + +int +nv50_runl_wait(struct nvkm_runl *runl) +{ + struct nvkm_fifo *fifo = runl->fifo; + + nvkm_msec(fifo->engine.subdev.device, fifo->timeout.chan_msec, + if (!nvkm_runl_update_pending(runl)) + return 0; + usleep_range(1, 2); + ); + + return -ETIMEDOUT; +} + +static void +nv50_runl_commit(struct nvkm_runl *runl, struct nvkm_memory *memory, u32 start, int count) +{ + struct nvkm_device *device = runl->fifo->engine.subdev.device; + u64 addr = nvkm_memory_addr(memory) + start; + + nvkm_wr32(device, 0x0032f4, addr >> 12); + nvkm_wr32(device, 0x0032ec, count); +} + +static void +nv50_runl_insert_chan(struct nvkm_chan *chan, struct nvkm_memory *memory, u64 offset) +{ + nvkm_wo32(memory, offset, chan->id); +} + +static struct nvkm_memory * +nv50_runl_alloc(struct nvkm_runl *runl, u32 *offset) +{ + const u32 segment = ALIGN((runl->cgrp_nr + runl->chan_nr) * runl->func->size, 0x1000); + const u32 maxsize = (runl->cgid ? runl->cgid->nr : 0) + runl->chid->nr; + int ret; + + if (unlikely(!runl->mem)) { + ret = nvkm_memory_new(runl->fifo->engine.subdev.device, NVKM_MEM_TARGET_INST, + maxsize * 2 * runl->func->size, 0, false, &runl->mem); + if (ret) { + RUNL_ERROR(runl, "alloc %d\n", ret); + return ERR_PTR(ret); + } + } else { + if (runl->offset + segment >= nvkm_memory_size(runl->mem)) { + ret = runl->func->wait(runl); + if (ret) { + RUNL_DEBUG(runl, "rewind timeout"); + return ERR_PTR(ret); + } + + runl->offset = 0; + } + } + + *offset = runl->offset; + runl->offset += segment; + return runl->mem; +} + +int +nv50_runl_update(struct nvkm_runl *runl) +{ + struct nvkm_memory *memory; + struct nvkm_cgrp *cgrp; + struct nvkm_chan *chan; + u32 start, offset, count; + + /*TODO: prio, interleaving. */ + + RUNL_TRACE(runl, "RAMRL: update cgrps:%d chans:%d", runl->cgrp_nr, runl->chan_nr); + memory = nv50_runl_alloc(runl, &start); + if (IS_ERR(memory)) + return PTR_ERR(memory); + + RUNL_TRACE(runl, "RAMRL: update start:%08x", start); + offset = start; + + nvkm_kmap(memory); + nvkm_runl_foreach_cgrp(cgrp, runl) { + if (cgrp->hw) { + CGRP_TRACE(cgrp, " RAMRL+%08x: chans:%d", offset, cgrp->chan_nr); + runl->func->insert_cgrp(cgrp, memory, offset); + offset += runl->func->size; + } + + nvkm_cgrp_foreach_chan(chan, cgrp) { + CHAN_TRACE(chan, "RAMRL+%08x: [%s]", offset, chan->name); + runl->func->insert_chan(chan, memory, offset); + offset += runl->func->size; + } + } + nvkm_done(memory); + + /*TODO: look into using features on newer HW to guarantee forward progress. */ + list_rotate_left(&runl->cgrps); + + count = (offset - start) / runl->func->size; + RUNL_TRACE(runl, "RAMRL: commit start:%08x count:%d", start, count); + + runl->func->commit(runl, memory, start, count); + return 0; +} + +const struct nvkm_runl_func +nv50_runl = { + .size = 4, + .update = nv50_runl_update, + .insert_chan = nv50_runl_insert_chan, + .commit = nv50_runl_commit, + .wait = nv50_runl_wait, + .pending = nv50_runl_pending, +}; + void -nv50_fifo_init(struct nvkm_fifo *base) +nv50_fifo_init(struct nvkm_fifo *fifo) { - struct nv50_fifo *fifo = nv50_fifo(base); - struct nvkm_device *device = fifo->base.engine.subdev.device; + struct nvkm_runl *runl = nvkm_runl_first(fifo); + struct nvkm_device *device = fifo->engine.subdev.device; int i; nvkm_mask(device, 0x000200, 0x00000100, 0x00000000); @@ -89,61 +352,47 @@ nv50_fifo_init(struct nvkm_fifo *base) for (i = 0; i < 128; i++) nvkm_wr32(device, 0x002600 + (i * 4), 0x00000000); - nv50_fifo_runlist_update_locked(fifo); + + atomic_set(&runl->changed, 1); + runl->func->update(runl); nvkm_wr32(device, 0x003200, 0x00000001); nvkm_wr32(device, 0x003250, 0x00000001); nvkm_wr32(device, 0x002500, 0x00000001); } -void * -nv50_fifo_dtor(struct nvkm_fifo *base) +int +nv50_fifo_chid_ctor(struct nvkm_fifo *fifo, int nr) { - struct nv50_fifo *fifo = nv50_fifo(base); - nvkm_memory_unref(&fifo->runlist[1]); - nvkm_memory_unref(&fifo->runlist[0]); - return fifo; + /* CHID 0 is unusable (some kind of PIO channel?), 127 is "channel invalid". */ + return nvkm_chid_new(&nvkm_chan_event, &fifo->engine.subdev, nr, 1, nr - 2, &fifo->chid); } int -nv50_fifo_new_(const struct nvkm_fifo_func *func, struct nvkm_device *device, - enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) +nv50_fifo_chid_nr(struct nvkm_fifo *fifo) { - struct nv50_fifo *fifo; - int ret; - - if (!(fifo = kzalloc(sizeof(*fifo), GFP_KERNEL))) - return -ENOMEM; - *pfifo = &fifo->base; - - ret = nvkm_fifo_ctor(func, device, type, inst, 128, &fifo->base); - if (ret) - return ret; - - set_bit(0, fifo->base.mask); /* PIO channel */ - set_bit(127, fifo->base.mask); /* inactive channel */ - return 0; + return 128; } static const struct nvkm_fifo_func nv50_fifo = { - .dtor = nv50_fifo_dtor, - .oneinit = nv50_fifo_oneinit, + .chid_nr = nv50_fifo_chid_nr, + .chid_ctor = nv50_fifo_chid_ctor, + .runl_ctor = nv04_fifo_runl_ctor, .init = nv50_fifo_init, .intr = nv04_fifo_intr, - .engine_id = nv04_fifo_engine_id, - .id_engine = nv04_fifo_id_engine, .pause = nv04_fifo_pause, .start = nv04_fifo_start, - .chan = { - &nv50_fifo_gpfifo_oclass, - NULL - }, + .runl = &nv50_runl, + .engn = &nv50_engn, + .engn_sw = &nv50_engn_sw, + .cgrp = {{ }, &nv04_cgrp }, + .chan = {{ 0, 0, NV50_CHANNEL_GPFIFO }, &nv50_chan }, }; int nv50_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - return nv50_fifo_new_(&nv50_fifo, device, type, inst, pfifo); + return nvkm_fifo_new_(&nv50_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv50.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv50.h deleted file mode 100644 index 0111e7e5a4e3..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/nv50.h +++ /dev/null @@ -1,20 +0,0 @@ -/* SPDX-License-Identifier: MIT */ -#ifndef __NV50_FIFO_H__ -#define __NV50_FIFO_H__ -#define nv50_fifo(p) container_of((p), struct nv50_fifo, base) -#include "priv.h" - -struct nv50_fifo { - struct nvkm_fifo base; - struct nvkm_memory *runlist[2]; - int cur_runlist; -}; - -int nv50_fifo_new_(const struct nvkm_fifo_func *, struct nvkm_device *, enum nvkm_subdev_type, int, - struct nvkm_fifo **); - -void *nv50_fifo_dtor(struct nvkm_fifo *); -int nv50_fifo_oneinit(struct nvkm_fifo *); -void nv50_fifo_init(struct nvkm_fifo *); -void nv50_fifo_runlist_update(struct nv50_fifo *); -#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/priv.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/priv.h index 79cec57647f0..4d448be19224 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/priv.h @@ -3,46 +3,207 @@ #define __NVKM_FIFO_PRIV_H__ #define nvkm_fifo(p) container_of((p), struct nvkm_fifo, engine) #include <engine/fifo.h> +#include <core/enum.h> +struct nvkm_cctx; +struct nvkm_cgrp; +struct nvkm_engn; +struct nvkm_memory; +struct nvkm_runl; +struct nvkm_runq; +struct nvkm_vctx; -int nvkm_fifo_ctor(const struct nvkm_fifo_func *, struct nvkm_device *, enum nvkm_subdev_type, int, - int nr, struct nvkm_fifo *); -void nvkm_fifo_uevent(struct nvkm_fifo *); -void nvkm_fifo_kevent(struct nvkm_fifo *, int chid); -void nvkm_fifo_recover_chan(struct nvkm_fifo *, int chid); - -struct nvkm_fifo_chan * -nvkm_fifo_chan_inst_locked(struct nvkm_fifo *, u64 inst); - -struct nvkm_fifo_chan_oclass; struct nvkm_fifo_func { - void *(*dtor)(struct nvkm_fifo *); - int (*oneinit)(struct nvkm_fifo *); - int (*info)(struct nvkm_fifo *, u64 mthd, u64 *data); + int (*chid_nr)(struct nvkm_fifo *); + int (*chid_ctor)(struct nvkm_fifo *, int nr); + int (*runq_nr)(struct nvkm_fifo *); + int (*runl_ctor)(struct nvkm_fifo *); + void (*init)(struct nvkm_fifo *); - void (*fini)(struct nvkm_fifo *); - void (*intr)(struct nvkm_fifo *); - void (*fault)(struct nvkm_fifo *, struct nvkm_fault_data *); - int (*engine_id)(struct nvkm_fifo *, struct nvkm_engine *); - struct nvkm_engine *(*id_engine)(struct nvkm_fifo *, int engi); + void (*init_pbdmas)(struct nvkm_fifo *, u32 mask); + + irqreturn_t (*intr)(struct nvkm_inth *); + void (*intr_mmu_fault_unit)(struct nvkm_fifo *, int unit); + void (*intr_ctxsw_timeout)(struct nvkm_fifo *, u32 engm); + + const struct nvkm_fifo_func_mmu_fault { + void (*recover)(struct nvkm_fifo *, struct nvkm_fault_data *); + const struct nvkm_enum *access; + const struct nvkm_enum *engine; + const struct nvkm_enum *reason; + const struct nvkm_enum *hubclient; + const struct nvkm_enum *gpcclient; + } *mmu_fault; + void (*pause)(struct nvkm_fifo *, unsigned long *); void (*start)(struct nvkm_fifo *, unsigned long *); - void (*uevent_init)(struct nvkm_fifo *); - void (*uevent_fini)(struct nvkm_fifo *); - void (*recover_chan)(struct nvkm_fifo *, int chid); - int (*class_get)(struct nvkm_fifo *, int index, struct nvkm_oclass *); - int (*class_new)(struct nvkm_fifo *, const struct nvkm_oclass *, - void *, u32, struct nvkm_object **); - const struct nvkm_fifo_chan_oclass *chan[]; + + int (*nonstall_ctor)(struct nvkm_fifo *); + const struct nvkm_event_func *nonstall; + + const struct nvkm_runl_func *runl; + const struct nvkm_runq_func *runq; + const struct nvkm_engn_func *engn; + const struct nvkm_engn_func *engn_sw; + const struct nvkm_engn_func *engn_ce; + + struct nvkm_fifo_func_cgrp { + struct nvkm_sclass user; + const struct nvkm_cgrp_func *func; + bool force; + } cgrp; + + struct nvkm_fifo_func_chan { + struct nvkm_sclass user; + const struct nvkm_chan_func *func; + } chan; }; -void nv04_fifo_intr(struct nvkm_fifo *); -int nv04_fifo_engine_id(struct nvkm_fifo *, struct nvkm_engine *); -struct nvkm_engine *nv04_fifo_id_engine(struct nvkm_fifo *, int); +int nvkm_fifo_new_(const struct nvkm_fifo_func *, struct nvkm_device *, enum nvkm_subdev_type, int, + struct nvkm_fifo **); + +int nv04_fifo_chid_ctor(struct nvkm_fifo *, int); +int nv04_fifo_runl_ctor(struct nvkm_fifo *); +void nv04_fifo_init(struct nvkm_fifo *); +irqreturn_t nv04_fifo_intr(struct nvkm_inth *); void nv04_fifo_pause(struct nvkm_fifo *, unsigned long *); void nv04_fifo_start(struct nvkm_fifo *, unsigned long *); +extern const struct nvkm_runl_func nv04_runl; +extern const struct nvkm_engn_func nv04_engn; +extern const struct nvkm_cgrp_func nv04_cgrp; +extern const struct nvkm_chan_func_inst nv04_chan_inst; +extern const struct nvkm_chan_func_userd nv04_chan_userd; +void nv04_chan_ramfc_clear(struct nvkm_chan *); +void nv04_chan_start(struct nvkm_chan *); +void nv04_chan_stop(struct nvkm_chan *); +void nv04_eobj_ramht_del(struct nvkm_chan *, int); + +int nv10_fifo_chid_nr(struct nvkm_fifo *); + +int nv50_fifo_chid_nr(struct nvkm_fifo *); +int nv50_fifo_chid_ctor(struct nvkm_fifo *, int); +void nv50_fifo_init(struct nvkm_fifo *); +extern const struct nvkm_runl_func nv50_runl; +int nv50_runl_update(struct nvkm_runl *); +int nv50_runl_wait(struct nvkm_runl *); +extern const struct nvkm_engn_func nv50_engn_sw; +extern const struct nvkm_chan_func_inst nv50_chan_inst; +extern const struct nvkm_chan_func_userd nv50_chan_userd; +void nv50_chan_unbind(struct nvkm_chan *); +void nv50_chan_start(struct nvkm_chan *); +void nv50_chan_stop(struct nvkm_chan *); +void nv50_chan_preempt(struct nvkm_chan *); +int nv50_eobj_ramht_add(struct nvkm_engn *, struct nvkm_object *, struct nvkm_chan *); +void nv50_eobj_ramht_del(struct nvkm_chan *, int); + +extern const struct nvkm_event_func g84_fifo_nonstall; +extern const struct nvkm_engn_func g84_engn; +extern const struct nvkm_chan_func g84_chan; + +int gf100_fifo_chid_ctor(struct nvkm_fifo *, int); +int gf100_fifo_runq_nr(struct nvkm_fifo *); +bool gf100_fifo_intr_pbdma(struct nvkm_fifo *); +void gf100_fifo_intr_mmu_fault(struct nvkm_fifo *); +void gf100_fifo_intr_mmu_fault_unit(struct nvkm_fifo *, int); +void gf100_fifo_intr_sched(struct nvkm_fifo *); +void gf100_fifo_intr_ctxsw_timeout(struct nvkm_fifo *, u32); +void gf100_fifo_mmu_fault_recover(struct nvkm_fifo *, struct nvkm_fault_data *); +extern const struct nvkm_enum gf100_fifo_mmu_fault_access[]; +extern const struct nvkm_event_func gf100_fifo_nonstall; +bool gf100_runl_preempt_pending(struct nvkm_runl *); +void gf100_runq_init(struct nvkm_runq *); +bool gf100_runq_intr(struct nvkm_runq *, struct nvkm_runl *); +void gf100_engn_mmu_fault_trigger(struct nvkm_engn *); +bool gf100_engn_mmu_fault_triggered(struct nvkm_engn *); +extern const struct nvkm_engn_func gf100_engn_sw; +extern const struct nvkm_chan_func_inst gf100_chan_inst; +void gf100_chan_userd_clear(struct nvkm_chan *); +void gf100_chan_preempt(struct nvkm_chan *); + +int gk104_fifo_chid_nr(struct nvkm_fifo *); +int gk104_fifo_runl_ctor(struct nvkm_fifo *); +void gk104_fifo_init(struct nvkm_fifo *); +void gk104_fifo_init_pbdmas(struct nvkm_fifo *, u32); +irqreturn_t gk104_fifo_intr(struct nvkm_inth *); +void gk104_fifo_intr_runlist(struct nvkm_fifo *); +void gk104_fifo_intr_chsw(struct nvkm_fifo *); +void gk104_fifo_intr_bind(struct nvkm_fifo *); +extern const struct nvkm_fifo_func_mmu_fault gk104_fifo_mmu_fault; +extern const struct nvkm_enum gk104_fifo_mmu_fault_reason[]; +extern const struct nvkm_enum gk104_fifo_mmu_fault_hubclient[]; +extern const struct nvkm_enum gk104_fifo_mmu_fault_gpcclient[]; +void gk104_runl_insert_chan(struct nvkm_chan *, struct nvkm_memory *, u64); +void gk104_runl_commit(struct nvkm_runl *, struct nvkm_memory *, u32, int); +bool gk104_runl_pending(struct nvkm_runl *); +void gk104_runl_block(struct nvkm_runl *, u32); +void gk104_runl_allow(struct nvkm_runl *, u32); +void gk104_runl_fault_clear(struct nvkm_runl *); +extern const struct nvkm_runq_func gk104_runq; +void gk104_runq_init(struct nvkm_runq *); +bool gk104_runq_intr(struct nvkm_runq *, struct nvkm_runl *); +extern const struct nvkm_bitfield gk104_runq_intr_0_names[]; +bool gk104_runq_idle(struct nvkm_runq *); +extern const struct nvkm_engn_func gk104_engn; +bool gk104_engn_chsw(struct nvkm_engn *); +int gk104_engn_cxid(struct nvkm_engn *, bool *cgid); +int gk104_ectx_ctor(struct nvkm_engn *, struct nvkm_vctx *); +extern const struct nvkm_engn_func gk104_engn_ce; +extern const struct nvkm_chan_func_userd gk104_chan_userd; +extern const struct nvkm_chan_func_ramfc gk104_chan_ramfc; +void gk104_chan_bind(struct nvkm_chan *); +void gk104_chan_bind_inst(struct nvkm_chan *); +void gk104_chan_unbind(struct nvkm_chan *); +void gk104_chan_start(struct nvkm_chan *); +void gk104_chan_stop(struct nvkm_chan *); + +int gk110_fifo_chid_ctor(struct nvkm_fifo *, int); +extern const struct nvkm_runl_func gk110_runl; +extern const struct nvkm_cgrp_func gk110_cgrp; +void gk110_runl_insert_cgrp(struct nvkm_cgrp *, struct nvkm_memory *, u64); +extern const struct nvkm_chan_func gk110_chan; +void gk110_chan_preempt(struct nvkm_chan *); + +extern const struct nvkm_runq_func gk208_runq; +void gk208_runq_init(struct nvkm_runq *); + +void gm107_fifo_intr_mmu_fault_unit(struct nvkm_fifo *, int); +extern const struct nvkm_fifo_func_mmu_fault gm107_fifo_mmu_fault; +extern const struct nvkm_runl_func gm107_runl; +extern const struct nvkm_chan_func gm107_chan; + +int gm200_fifo_chid_nr(struct nvkm_fifo *); +int gm200_fifo_runq_nr(struct nvkm_fifo *); + +extern const struct nvkm_enum gv100_fifo_mmu_fault_access[]; +extern const struct nvkm_enum gv100_fifo_mmu_fault_reason[]; +extern const struct nvkm_enum gv100_fifo_mmu_fault_hubclient[]; +extern const struct nvkm_enum gv100_fifo_mmu_fault_gpcclient[]; +void gv100_runl_insert_cgrp(struct nvkm_cgrp *, struct nvkm_memory *, u64); +void gv100_runl_insert_chan(struct nvkm_chan *, struct nvkm_memory *, u64); +void gv100_runl_preempt(struct nvkm_runl *); +extern const struct nvkm_runq_func gv100_runq; +extern const struct nvkm_engn_func gv100_engn; +void gv100_ectx_bind(struct nvkm_engn *, struct nvkm_cctx *, struct nvkm_chan *); +extern const struct nvkm_engn_func gv100_engn_ce; +int gv100_ectx_ce_ctor(struct nvkm_engn *, struct nvkm_vctx *); +void gv100_ectx_ce_bind(struct nvkm_engn *, struct nvkm_cctx *, struct nvkm_chan *); +extern const struct nvkm_chan_func_userd gv100_chan_userd; +extern const struct nvkm_chan_func_ramfc gv100_chan_ramfc; + +void tu102_fifo_intr_ctxsw_timeout_info(struct nvkm_engn *, u32 info); +extern const struct nvkm_fifo_func_mmu_fault tu102_fifo_mmu_fault; -void gf100_fifo_intr_fault(struct nvkm_fifo *, int); +int ga100_fifo_runl_ctor(struct nvkm_fifo *); +int ga100_fifo_nonstall_ctor(struct nvkm_fifo *); +extern const struct nvkm_event_func ga100_fifo_nonstall; +extern const struct nvkm_runl_func ga100_runl; +extern const struct nvkm_runq_func ga100_runq; +extern const struct nvkm_engn_func ga100_engn; +extern const struct nvkm_engn_func ga100_engn_ce; +extern const struct nvkm_cgrp_func ga100_cgrp; +extern const struct nvkm_chan_func ga100_chan; -int gk104_fifo_engine_id(struct nvkm_fifo *, struct nvkm_engine *); -struct nvkm_engine *gk104_fifo_id_engine(struct nvkm_fifo *, int); +int nvkm_uchan_new(struct nvkm_fifo *, struct nvkm_cgrp *, const struct nvkm_oclass *, + void *argv, u32 argc, struct nvkm_object **); +int nvkm_ucgrp_new(struct nvkm_fifo *, const struct nvkm_oclass *, void *argv, u32 argc, + struct nvkm_object **); #endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runl.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runl.c new file mode 100644 index 000000000000..b5836cbc29aa --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runl.c @@ -0,0 +1,430 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "runl.h" +#include "cgrp.h" +#include "chan.h" +#include "chid.h" +#include "priv.h" +#include "runq.h" + +#include <core/gpuobj.h> +#include <subdev/timer.h> +#include <subdev/top.h> + +struct nvkm_cgrp * +nvkm_engn_cgrp_get(struct nvkm_engn *engn, unsigned long *pirqflags) +{ + struct nvkm_cgrp *cgrp = NULL; + struct nvkm_chan *chan; + bool cgid; + int id; + + id = engn->func->cxid(engn, &cgid); + if (id < 0) + return NULL; + + if (!cgid) { + chan = nvkm_runl_chan_get_chid(engn->runl, id, pirqflags); + if (chan) + cgrp = chan->cgrp; + } else { + cgrp = nvkm_runl_cgrp_get_cgid(engn->runl, id, pirqflags); + } + + WARN_ON(!cgrp); + return cgrp; +} + +static void +nvkm_runl_rc(struct nvkm_runl *runl) +{ + struct nvkm_fifo *fifo = runl->fifo; + struct nvkm_cgrp *cgrp, *gtmp; + struct nvkm_chan *chan, *ctmp; + struct nvkm_engn *engn; + unsigned long flags; + int rc, state, i; + bool reset; + + /* Runlist is blocked before scheduling recovery - fetch count. */ + BUG_ON(!mutex_is_locked(&runl->mutex)); + rc = atomic_xchg(&runl->rc_pending, 0); + if (!rc) + return; + + /* Look for channel groups flagged for RC. */ + nvkm_runl_foreach_cgrp_safe(cgrp, gtmp, runl) { + state = atomic_cmpxchg(&cgrp->rc, NVKM_CGRP_RC_PENDING, NVKM_CGRP_RC_RUNNING); + if (state == NVKM_CGRP_RC_PENDING) { + /* Disable all channels in them, and remove from runlist. */ + nvkm_cgrp_foreach_chan_safe(chan, ctmp, cgrp) { + nvkm_chan_error(chan, false); + nvkm_chan_remove_locked(chan); + } + } + } + + /* On GPUs with runlist preempt, wait for PBDMA(s) servicing runlist to go idle. */ + if (runl->func->preempt) { + for (i = 0; i < runl->runq_nr; i++) { + struct nvkm_runq *runq = runl->runq[i]; + + if (runq) { + nvkm_msec(fifo->engine.subdev.device, 2000, + if (runq->func->idle(runq)) + break; + ); + } + } + } + + /* Look for engines that are still on flagged channel groups - reset them. */ + nvkm_runl_foreach_engn_cond(engn, runl, engn->func->cxid) { + cgrp = nvkm_engn_cgrp_get(engn, &flags); + if (!cgrp) { + ENGN_DEBUG(engn, "cxid not valid"); + continue; + } + + reset = atomic_read(&cgrp->rc) == NVKM_CGRP_RC_RUNNING; + nvkm_cgrp_put(&cgrp, flags); + if (!reset) { + ENGN_DEBUG(engn, "cxid not in recovery"); + continue; + } + + ENGN_DEBUG(engn, "resetting..."); + /*TODO: can we do something less of a potential catastrophe on failure? */ + WARN_ON(nvkm_engine_reset(engn->engine)); + } + + /* Submit runlist update, and clear any remaining exception state. */ + runl->func->update(runl); + if (runl->func->fault_clear) + runl->func->fault_clear(runl); + + /* Unblock runlist processing. */ + while (rc--) + nvkm_runl_allow(runl); + runl->func->wait(runl); +} + +static void +nvkm_runl_rc_runl(struct nvkm_runl *runl) +{ + RUNL_ERROR(runl, "rc scheduled"); + + nvkm_runl_block(runl); + if (runl->func->preempt) + runl->func->preempt(runl); + + atomic_inc(&runl->rc_pending); + schedule_work(&runl->work); +} + +void +nvkm_runl_rc_cgrp(struct nvkm_cgrp *cgrp) +{ + if (atomic_cmpxchg(&cgrp->rc, NVKM_CGRP_RC_NONE, NVKM_CGRP_RC_PENDING) != NVKM_CGRP_RC_NONE) + return; + + CGRP_ERROR(cgrp, "rc scheduled"); + nvkm_runl_rc_runl(cgrp->runl); +} + +void +nvkm_runl_rc_engn(struct nvkm_runl *runl, struct nvkm_engn *engn) +{ + struct nvkm_cgrp *cgrp; + unsigned long flags; + + /* Lookup channel group currently on engine. */ + cgrp = nvkm_engn_cgrp_get(engn, &flags); + if (!cgrp) { + ENGN_DEBUG(engn, "rc skipped, not on channel"); + return; + } + + nvkm_runl_rc_cgrp(cgrp); + nvkm_cgrp_put(&cgrp, flags); +} + +static void +nvkm_runl_work(struct work_struct *work) +{ + struct nvkm_runl *runl = container_of(work, typeof(*runl), work); + + mutex_lock(&runl->mutex); + nvkm_runl_rc(runl); + mutex_unlock(&runl->mutex); + +} + +struct nvkm_chan * +nvkm_runl_chan_get_inst(struct nvkm_runl *runl, u64 inst, unsigned long *pirqflags) +{ + struct nvkm_chid *chid = runl->chid; + struct nvkm_chan *chan; + unsigned long flags; + int id; + + spin_lock_irqsave(&chid->lock, flags); + for_each_set_bit(id, chid->used, chid->nr) { + chan = chid->data[id]; + if (likely(chan)) { + if (chan->inst->addr == inst) { + spin_lock(&chan->cgrp->lock); + *pirqflags = flags; + spin_unlock(&chid->lock); + return chan; + } + } + } + spin_unlock_irqrestore(&chid->lock, flags); + return NULL; +} + +struct nvkm_chan * +nvkm_runl_chan_get_chid(struct nvkm_runl *runl, int id, unsigned long *pirqflags) +{ + struct nvkm_chid *chid = runl->chid; + struct nvkm_chan *chan; + unsigned long flags; + + spin_lock_irqsave(&chid->lock, flags); + if (!WARN_ON(id >= chid->nr)) { + chan = chid->data[id]; + if (likely(chan)) { + spin_lock(&chan->cgrp->lock); + *pirqflags = flags; + spin_unlock(&chid->lock); + return chan; + } + } + spin_unlock_irqrestore(&chid->lock, flags); + return NULL; +} + +struct nvkm_cgrp * +nvkm_runl_cgrp_get_cgid(struct nvkm_runl *runl, int id, unsigned long *pirqflags) +{ + struct nvkm_chid *cgid = runl->cgid; + struct nvkm_cgrp *cgrp; + unsigned long flags; + + spin_lock_irqsave(&cgid->lock, flags); + if (!WARN_ON(id >= cgid->nr)) { + cgrp = cgid->data[id]; + if (likely(cgrp)) { + spin_lock(&cgrp->lock); + *pirqflags = flags; + spin_unlock(&cgid->lock); + return cgrp; + } + } + spin_unlock_irqrestore(&cgid->lock, flags); + return NULL; +} + +int +nvkm_runl_preempt_wait(struct nvkm_runl *runl) +{ + return nvkm_msec(runl->fifo->engine.subdev.device, runl->fifo->timeout.chan_msec, + if (!runl->func->preempt_pending(runl)) + break; + + nvkm_runl_rc(runl); + usleep_range(1, 2); + ) < 0 ? -ETIMEDOUT : 0; +} + +bool +nvkm_runl_update_pending(struct nvkm_runl *runl) +{ + if (!runl->func->pending(runl)) + return false; + + nvkm_runl_rc(runl); + return true; +} + +void +nvkm_runl_update_locked(struct nvkm_runl *runl, bool wait) +{ + if (atomic_xchg(&runl->changed, 0) && runl->func->update) { + runl->func->update(runl); + if (wait) + runl->func->wait(runl); + } +} + +void +nvkm_runl_allow(struct nvkm_runl *runl) +{ + struct nvkm_fifo *fifo = runl->fifo; + unsigned long flags; + + spin_lock_irqsave(&fifo->lock, flags); + if (!--runl->blocked) { + RUNL_TRACE(runl, "running"); + runl->func->allow(runl, ~0); + } + spin_unlock_irqrestore(&fifo->lock, flags); +} + +void +nvkm_runl_block(struct nvkm_runl *runl) +{ + struct nvkm_fifo *fifo = runl->fifo; + unsigned long flags; + + spin_lock_irqsave(&fifo->lock, flags); + if (!runl->blocked++) { + RUNL_TRACE(runl, "stopped"); + runl->func->block(runl, ~0); + } + spin_unlock_irqrestore(&fifo->lock, flags); +} + +void +nvkm_runl_fini(struct nvkm_runl *runl) +{ + if (runl->func->fini) + runl->func->fini(runl); + + flush_work(&runl->work); +} + +void +nvkm_runl_del(struct nvkm_runl *runl) +{ + struct nvkm_engn *engn, *engt; + + nvkm_memory_unref(&runl->mem); + + list_for_each_entry_safe(engn, engt, &runl->engns, head) { + list_del(&engn->head); + kfree(engn); + } + + nvkm_chid_unref(&runl->chid); + nvkm_chid_unref(&runl->cgid); + + list_del(&runl->head); + mutex_destroy(&runl->mutex); + kfree(runl); +} + +struct nvkm_engn * +nvkm_runl_add(struct nvkm_runl *runl, int engi, const struct nvkm_engn_func *func, + enum nvkm_subdev_type type, int inst) +{ + struct nvkm_fifo *fifo = runl->fifo; + struct nvkm_device *device = fifo->engine.subdev.device; + struct nvkm_engine *engine; + struct nvkm_engn *engn; + + engine = nvkm_device_engine(device, type, inst); + if (!engine) { + RUNL_DEBUG(runl, "engn %d.%d[%s] not found", engi, inst, nvkm_subdev_type[type]); + return NULL; + } + + if (!(engn = kzalloc(sizeof(*engn), GFP_KERNEL))) + return NULL; + + engn->func = func; + engn->runl = runl; + engn->id = engi; + engn->engine = engine; + engn->fault = -1; + list_add_tail(&engn->head, &runl->engns); + + /* Lookup MMU engine ID for fault handling. */ + if (device->top) + engn->fault = nvkm_top_fault_id(device, engine->subdev.type, engine->subdev.inst); + + if (engn->fault < 0 && fifo->func->mmu_fault) { + const struct nvkm_enum *map = fifo->func->mmu_fault->engine; + + while (map->name) { + if (map->data2 == engine->subdev.type && map->inst == engine->subdev.inst) { + engn->fault = map->value; + break; + } + map++; + } + } + + return engn; +} + +struct nvkm_runl * +nvkm_runl_get(struct nvkm_fifo *fifo, int runi, u32 addr) +{ + struct nvkm_runl *runl; + + nvkm_runl_foreach(runl, fifo) { + if ((runi >= 0 && runl->id == runi) || (runi < 0 && runl->addr == addr)) + return runl; + } + + return NULL; +} + +struct nvkm_runl * +nvkm_runl_new(struct nvkm_fifo *fifo, int runi, u32 addr, int id_nr) +{ + struct nvkm_subdev *subdev = &fifo->engine.subdev; + struct nvkm_runl *runl; + int ret; + + if (!(runl = kzalloc(sizeof(*runl), GFP_KERNEL))) + return NULL; + + runl->func = fifo->func->runl; + runl->fifo = fifo; + runl->id = runi; + runl->addr = addr; + INIT_LIST_HEAD(&runl->engns); + INIT_LIST_HEAD(&runl->cgrps); + atomic_set(&runl->changed, 0); + mutex_init(&runl->mutex); + INIT_WORK(&runl->work, nvkm_runl_work); + atomic_set(&runl->rc_triggered, 0); + atomic_set(&runl->rc_pending, 0); + list_add_tail(&runl->head, &fifo->runls); + + if (!fifo->chid) { + if ((ret = nvkm_chid_new(&nvkm_chan_event, subdev, id_nr, 0, id_nr, &runl->cgid)) || + (ret = nvkm_chid_new(&nvkm_chan_event, subdev, id_nr, 0, id_nr, &runl->chid))) { + RUNL_ERROR(runl, "cgid/chid: %d", ret); + nvkm_runl_del(runl); + return NULL; + } + } else { + runl->cgid = nvkm_chid_ref(fifo->cgid); + runl->chid = nvkm_chid_ref(fifo->chid); + } + + return runl; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runl.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runl.h new file mode 100644 index 000000000000..c93d21bb7bd5 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runl.h @@ -0,0 +1,125 @@ +#ifndef __NVKM_RUNL_H__ +#define __NVKM_RUNL_H__ +#include <core/intr.h> +struct nvkm_cctx; +struct nvkm_cgrp; +struct nvkm_chan; +struct nvkm_memory; +struct nvkm_object; +struct nvkm_vctx; +enum nvkm_subdev_type; + +struct nvkm_engn { + const struct nvkm_engn_func { + bool (*chsw)(struct nvkm_engn *); + int (*cxid)(struct nvkm_engn *, bool *cgid); + void (*mmu_fault_trigger)(struct nvkm_engn *); + bool (*mmu_fault_triggered)(struct nvkm_engn *); + int (*ctor)(struct nvkm_engn *, struct nvkm_vctx *); + void (*bind)(struct nvkm_engn *, struct nvkm_cctx *, struct nvkm_chan *); + int (*ramht_add)(struct nvkm_engn *, struct nvkm_object *, struct nvkm_chan *); + void (*ramht_del)(struct nvkm_chan *, int hash); + } *func; + struct nvkm_runl *runl; + int id; + + struct nvkm_engine *engine; + + int fault; + + struct list_head head; +}; + +#define ENGN_PRINT(e,l,p,f,a...) \ + RUNL_PRINT((e)->runl, l, p, "%02d[%8s]:"f, (e)->id, (e)->engine->subdev.name, ##a) +#define ENGN_DEBUG(e,f,a...) ENGN_PRINT((e), DEBUG, info, " "f"\n", ##a) + +struct nvkm_runl { + const struct nvkm_runl_func { + void (*init)(struct nvkm_runl *); + void (*fini)(struct nvkm_runl *); + int runqs; + u8 size; + int (*update)(struct nvkm_runl *); + void (*insert_cgrp)(struct nvkm_cgrp *, struct nvkm_memory *, u64 offset); + void (*insert_chan)(struct nvkm_chan *, struct nvkm_memory *, u64 offset); + void (*commit)(struct nvkm_runl *, struct nvkm_memory *, u32 start, int count); + int (*wait)(struct nvkm_runl *); + bool (*pending)(struct nvkm_runl *); + void (*block)(struct nvkm_runl *, u32 engm); + void (*allow)(struct nvkm_runl *, u32 engm); + void (*fault_clear)(struct nvkm_runl *); + void (*preempt)(struct nvkm_runl *); + bool (*preempt_pending)(struct nvkm_runl *); + } *func; + struct nvkm_fifo *fifo; + int id; + u32 addr; + u32 chan; + u16 doorbell; + + struct nvkm_chid *cgid; +#define NVKM_CHAN_EVENT_ERRORED BIT(0) + struct nvkm_chid *chid; + + struct list_head engns; + + struct nvkm_runq *runq[2]; + int runq_nr; + + struct nvkm_inth inth; + + struct list_head cgrps; + int cgrp_nr; + int chan_nr; + atomic_t changed; + struct nvkm_memory *mem; + u32 offset; + struct mutex mutex; + + int blocked; + + struct work_struct work; + atomic_t rc_triggered; + atomic_t rc_pending; + + struct list_head head; +}; + +struct nvkm_runl *nvkm_runl_new(struct nvkm_fifo *, int runi, u32 addr, int id_nr); +struct nvkm_runl *nvkm_runl_get(struct nvkm_fifo *, int runi, u32 addr); +struct nvkm_engn *nvkm_runl_add(struct nvkm_runl *, int engi, const struct nvkm_engn_func *, + enum nvkm_subdev_type, int inst); +void nvkm_runl_del(struct nvkm_runl *); +void nvkm_runl_fini(struct nvkm_runl *); +void nvkm_runl_block(struct nvkm_runl *); +void nvkm_runl_allow(struct nvkm_runl *); +void nvkm_runl_update_locked(struct nvkm_runl *, bool wait); +bool nvkm_runl_update_pending(struct nvkm_runl *); +int nvkm_runl_preempt_wait(struct nvkm_runl *); + +void nvkm_runl_rc_engn(struct nvkm_runl *, struct nvkm_engn *); +void nvkm_runl_rc_cgrp(struct nvkm_cgrp *); + +struct nvkm_cgrp *nvkm_runl_cgrp_get_cgid(struct nvkm_runl *, int cgid, unsigned long *irqflags); +struct nvkm_chan *nvkm_runl_chan_get_chid(struct nvkm_runl *, int chid, unsigned long *irqflags); +struct nvkm_chan *nvkm_runl_chan_get_inst(struct nvkm_runl *, u64 inst, unsigned long *irqflags); + +#define nvkm_runl_find_engn(engn,runl,cond) nvkm_list_find(engn, &(runl)->engns, head, (cond)) + +#define nvkm_runl_first(fifo) list_first_entry(&(fifo)->runls, struct nvkm_runl, head) +#define nvkm_runl_foreach(runl,fifo) list_for_each_entry((runl), &(fifo)->runls, head) +#define nvkm_runl_foreach_cond(runl,fifo,cond) nvkm_list_foreach(runl, &(fifo)->runls, head, (cond)) +#define nvkm_runl_foreach_engn(engn,runl) list_for_each_entry((engn), &(runl)->engns, head) +#define nvkm_runl_foreach_engn_cond(engn,runl,cond) \ + nvkm_list_foreach(engn, &(runl)->engns, head, (cond)) +#define nvkm_runl_foreach_cgrp(cgrp,runl) list_for_each_entry((cgrp), &(runl)->cgrps, head) +#define nvkm_runl_foreach_cgrp_safe(cgrp,gtmp,runl) \ + list_for_each_entry_safe((cgrp), (gtmp), &(runl)->cgrps, head) + +#define RUNL_PRINT(r,l,p,f,a...) \ + nvkm_printk__(&(r)->fifo->engine.subdev, NV_DBG_##l, p, "%06x:"f, (r)->addr, ##a) +#define RUNL_ERROR(r,f,a...) RUNL_PRINT((r), ERROR, err, " "f"\n", ##a) +#define RUNL_DEBUG(r,f,a...) RUNL_PRINT((r), DEBUG, info, " "f"\n", ##a) +#define RUNL_TRACE(r,f,a...) RUNL_PRINT((r), TRACE, info, " "f"\n", ##a) +#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runq.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runq.c new file mode 100644 index 000000000000..33bcf5fb3ef0 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runq.c @@ -0,0 +1,45 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "runq.h" +#include "priv.h" + +void +nvkm_runq_del(struct nvkm_runq *runq) +{ + list_del(&runq->head); + kfree(runq); +} + +struct nvkm_runq * +nvkm_runq_new(struct nvkm_fifo *fifo, int pbid) +{ + struct nvkm_runq *runq; + + if (!(runq = kzalloc(sizeof(*runq), GFP_KERNEL))) + return NULL; + + runq->func = fifo->func->runq; + runq->fifo = fifo; + runq->id = pbid; + list_add_tail(&runq->head, &fifo->runqs); + return runq; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runq.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runq.h new file mode 100644 index 000000000000..2cb4836e8b31 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/runq.h @@ -0,0 +1,31 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __NVKM_RUNQ_H__ +#define __NVKM_RUNQ_H__ +#include <core/os.h> +struct nvkm_runl; + +struct nvkm_runq { + const struct nvkm_runq_func { + void (*init)(struct nvkm_runq *); + bool (*intr)(struct nvkm_runq *, struct nvkm_runl *); + const struct nvkm_bitfield *intr_0_names; + bool (*intr_1_ctxnotvalid)(struct nvkm_runq *, int chid); + bool (*idle)(struct nvkm_runq *); + } *func; + struct nvkm_fifo *fifo; + int id; + + struct list_head head; +}; + +struct nvkm_runq *nvkm_runq_new(struct nvkm_fifo *, int pbid); +void nvkm_runq_del(struct nvkm_runq *); + +#define nvkm_runq_foreach(runq,fifo) list_for_each_entry((runq), &(fifo)->runqs, head) +#define nvkm_runq_foreach_cond(runq,fifo,cond) nvkm_list_foreach(runq, &(fifo)->runqs, head, (cond)) + +#define RUNQ_PRINT(r,l,p,f,a...) \ + nvkm_printk__(&(r)->fifo->engine.subdev, NV_DBG_##l, p, "PBDMA%d:"f, (r)->id, ##a) +#define RUNQ_ERROR(r,f,a...) RUNQ_PRINT((r), ERROR, err, " "f"\n", ##a) +#define RUNQ_DEBUG(r,f,a...) RUNQ_PRINT((r), DEBUG, info, " "f"\n", ##a) +#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/tu102.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/tu102.c index 260b197f81bc..ea9e151dbb48 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/tu102.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/tu102.c @@ -19,46 +19,83 @@ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. */ -#include "gk104.h" +#include "priv.h" #include "cgrp.h" -#include "changk104.h" -#include "user.h" +#include "chan.h" +#include "runl.h" -#include <core/client.h> -#include <core/gpuobj.h> -#include <subdev/bar.h> -#include <subdev/fault.h> -#include <subdev/top.h> -#include <subdev/timer.h> -#include <engine/sw.h> +#include <core/memory.h> +#include <subdev/mc.h> +#include <subdev/vfn.h> #include <nvif/class.h> +static u32 +tu102_chan_doorbell_handle(struct nvkm_chan *chan) +{ + return (chan->cgrp->runl->id << 16) | chan->id; +} + static void -tu102_fifo_runlist_commit(struct gk104_fifo *fifo, int runl, - struct nvkm_memory *mem, int nr) +tu102_chan_start(struct nvkm_chan *chan) { - struct nvkm_device *device = fifo->base.engine.subdev.device; - u64 addr = nvkm_memory_addr(mem); - /*XXX: target? */ + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; + + gk104_chan_start(chan); + nvkm_wr32(device, device->vfn->addr.user + 0x0090, chan->func->doorbell_handle(chan)); +} + +static const struct nvkm_chan_func +tu102_chan = { + .inst = &gf100_chan_inst, + .userd = &gv100_chan_userd, + .ramfc = &gv100_chan_ramfc, + .bind = gk104_chan_bind_inst, + .unbind = gk104_chan_unbind, + .start = tu102_chan_start, + .stop = gk104_chan_stop, + .preempt = gk110_chan_preempt, + .doorbell_handle = tu102_chan_doorbell_handle, +}; + +static bool +tu102_runl_pending(struct nvkm_runl *runl) +{ + struct nvkm_device *device = runl->fifo->engine.subdev.device; + + return nvkm_rd32(device, 0x002b0c + (runl->id * 0x10)) & 0x00008000; +} - nvkm_wr32(device, 0x002b00 + (runl * 0x10), lower_32_bits(addr)); - nvkm_wr32(device, 0x002b04 + (runl * 0x10), upper_32_bits(addr)); - nvkm_wr32(device, 0x002b08 + (runl * 0x10), nr); +static void +tu102_runl_commit(struct nvkm_runl *runl, struct nvkm_memory *memory, u32 start, int count) +{ + struct nvkm_device *device = runl->fifo->engine.subdev.device; + u64 addr = nvkm_memory_addr(memory) + start; + /*XXX: target? */ - /*XXX: how to wait? can you even wait? */ + nvkm_wr32(device, 0x002b00 + (runl->id * 0x10), lower_32_bits(addr)); + nvkm_wr32(device, 0x002b04 + (runl->id * 0x10), upper_32_bits(addr)); + nvkm_wr32(device, 0x002b08 + (runl->id * 0x10), count); } -static const struct gk104_fifo_runlist_func -tu102_fifo_runlist = { +static const struct nvkm_runl_func +tu102_runl = { + .runqs = 2, .size = 16, - .cgrp = gv100_fifo_runlist_cgrp, - .chan = gv100_fifo_runlist_chan, - .commit = tu102_fifo_runlist_commit, + .update = nv50_runl_update, + .insert_cgrp = gv100_runl_insert_cgrp, + .insert_chan = gv100_runl_insert_chan, + .commit = tu102_runl_commit, + .wait = nv50_runl_wait, + .pending = tu102_runl_pending, + .block = gk104_runl_block, + .allow = gk104_runl_allow, + .preempt = gv100_runl_preempt, + .preempt_pending = gf100_runl_preempt_pending, }; static const struct nvkm_enum -tu102_fifo_fault_engine[] = { +tu102_fifo_mmu_fault_engine[] = { { 0x01, "DISPLAY" }, { 0x03, "PTP" }, { 0x06, "PWR_PMU" }, @@ -85,305 +122,82 @@ tu102_fifo_fault_engine[] = { {} }; -static void -tu102_fifo_pbdma_init(struct gk104_fifo *fifo) -{ - struct nvkm_device *device = fifo->base.engine.subdev.device; - const u32 mask = (1 << fifo->pbdma_nr) - 1; - /*XXX: this is a bit of a guess at this point in time. */ - nvkm_mask(device, 0xb65000, 0x80000fff, 0x80000000 | mask); -} - -static const struct gk104_fifo_pbdma_func -tu102_fifo_pbdma = { - .nr = gm200_fifo_pbdma_nr, - .init = tu102_fifo_pbdma_init, - .init_timeout = gk208_fifo_pbdma_init_timeout, -}; - -static const struct gk104_fifo_func -tu102_fifo = { - .pbdma = &tu102_fifo_pbdma, - .fault.access = gv100_fifo_fault_access, - .fault.engine = tu102_fifo_fault_engine, - .fault.reason = gv100_fifo_fault_reason, - .fault.hubclient = gv100_fifo_fault_hubclient, - .fault.gpcclient = gv100_fifo_fault_gpcclient, - .runlist = &tu102_fifo_runlist, - .user = {{-1,-1,VOLTA_USERMODE_A }, tu102_fifo_user_new }, - .chan = {{ 0, 0,TURING_CHANNEL_GPFIFO_A}, tu102_fifo_gpfifo_new }, - .cgrp_force = true, +const struct nvkm_fifo_func_mmu_fault +tu102_fifo_mmu_fault = { + .recover = gf100_fifo_mmu_fault_recover, + .access = gv100_fifo_mmu_fault_access, + .engine = tu102_fifo_mmu_fault_engine, + .reason = gv100_fifo_mmu_fault_reason, + .hubclient = gv100_fifo_mmu_fault_hubclient, + .gpcclient = gv100_fifo_mmu_fault_gpcclient, }; -static void -tu102_fifo_recover_work(struct work_struct *w) +void +tu102_fifo_intr_ctxsw_timeout_info(struct nvkm_engn *engn, u32 info) { - struct gk104_fifo *fifo = container_of(w, typeof(*fifo), recover.work); - struct nvkm_device *device = fifo->base.engine.subdev.device; - struct nvkm_engine *engine; + struct nvkm_runl *runl = engn->runl; + struct nvkm_cgrp *cgrp; unsigned long flags; - u32 engm, runm, todo; - int engn, runl; - - spin_lock_irqsave(&fifo->base.lock, flags); - runm = fifo->recover.runm; - engm = fifo->recover.engm; - fifo->recover.engm = 0; - fifo->recover.runm = 0; - spin_unlock_irqrestore(&fifo->base.lock, flags); - - nvkm_mask(device, 0x002630, runm, runm); - - for (todo = engm; engn = __ffs(todo), todo; todo &= ~BIT(engn)) { - if ((engine = fifo->engine[engn].engine)) { - nvkm_subdev_fini(&engine->subdev, false); - WARN_ON(nvkm_subdev_init(&engine->subdev)); - } - } - - for (todo = runm; runl = __ffs(todo), todo; todo &= ~BIT(runl)) - gk104_fifo_runlist_update(fifo, runl); - - nvkm_mask(device, 0x002630, runm, 0x00000000); -} - -static void tu102_fifo_recover_engn(struct gk104_fifo *fifo, int engn); - -static void -tu102_fifo_recover_runl(struct gk104_fifo *fifo, int runl) -{ - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - const u32 runm = BIT(runl); - - assert_spin_locked(&fifo->base.lock); - if (fifo->recover.runm & runm) - return; - fifo->recover.runm |= runm; - - /* Block runlist to prevent channel assignment(s) from changing. */ - nvkm_mask(device, 0x002630, runm, runm); - - /* Schedule recovery. */ - nvkm_warn(subdev, "runlist %d: scheduled for recovery\n", runl); - schedule_work(&fifo->recover.work); -} - -static struct gk104_fifo_chan * -tu102_fifo_recover_chid(struct gk104_fifo *fifo, int runl, int chid) -{ - struct gk104_fifo_chan *chan; - struct nvkm_fifo_cgrp *cgrp; - - list_for_each_entry(chan, &fifo->runlist[runl].chan, head) { - if (chan->base.chid == chid) { - list_del_init(&chan->head); - return chan; - } - } - - list_for_each_entry(cgrp, &fifo->runlist[runl].cgrp, head) { - if (cgrp->id == chid) { - chan = list_first_entry(&cgrp->chan, typeof(*chan), head); - list_del_init(&chan->head); - if (!--cgrp->chan_nr) - list_del_init(&cgrp->head); - return chan; - } - } - - return NULL; -} -static void -tu102_fifo_recover_chan(struct nvkm_fifo *base, int chid) -{ - struct gk104_fifo *fifo = gk104_fifo(base); - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - const u32 stat = nvkm_rd32(device, 0x800004 + (chid * 0x08)); - const u32 runl = (stat & 0x000f0000) >> 16; - const bool used = (stat & 0x00000001); - unsigned long engn, engm = fifo->runlist[runl].engm; - struct gk104_fifo_chan *chan; - - assert_spin_locked(&fifo->base.lock); - if (!used) + /* Check that engine hasn't become unstuck since timeout raised. */ + ENGN_DEBUG(engn, "CTXSW_TIMEOUT %08x", info); + if (info & 0xc0000000) return; - /* Lookup SW state for channel, and mark it as dead. */ - chan = tu102_fifo_recover_chid(fifo, runl, chid); - if (chan) { - chan->killed = true; - nvkm_fifo_kevent(&fifo->base, chid); - } - - /* Disable channel. */ - nvkm_wr32(device, 0x800004 + (chid * 0x08), stat | 0x00000800); - nvkm_warn(subdev, "channel %d: killed\n", chid); - - /* Block channel assignments from changing during recovery. */ - tu102_fifo_recover_runl(fifo, runl); - - /* Schedule recovery for any engines the channel is on. */ - for_each_set_bit(engn, &engm, fifo->engine_nr) { - struct gk104_fifo_engine_status status; - - gk104_fifo_engine_status(fifo, engn, &status); - if (!status.chan || status.chan->id != chid) - continue; - tu102_fifo_recover_engn(fifo, engn); + /* Determine channel group the engine is stuck on, and schedule recovery. */ + switch (info & 0x0000c000) { + case 0x00004000: /* LOAD */ + cgrp = nvkm_runl_cgrp_get_cgid(runl, info & 0x3fff0000, &flags); + break; + case 0x00008000: /* SAVE */ + case 0x0000c000: /* SWITCH */ + cgrp = nvkm_runl_cgrp_get_cgid(runl, info & 0x00003fff, &flags); + break; + default: + cgrp = NULL; + break; } -} - -static void -tu102_fifo_recover_engn(struct gk104_fifo *fifo, int engn) -{ - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - const u32 runl = fifo->engine[engn].runl; - const u32 engm = BIT(engn); - struct gk104_fifo_engine_status status; - - assert_spin_locked(&fifo->base.lock); - if (fifo->recover.engm & engm) - return; - fifo->recover.engm |= engm; - /* Block channel assignments from changing during recovery. */ - tu102_fifo_recover_runl(fifo, runl); - - /* Determine which channel (if any) is currently on the engine. */ - gk104_fifo_engine_status(fifo, engn, &status); - if (status.chan) { - /* The channel is not longer viable, kill it. */ - tu102_fifo_recover_chan(&fifo->base, status.chan->id); + if (!WARN_ON(!cgrp)) { + nvkm_runl_rc_cgrp(cgrp); + nvkm_cgrp_put(&cgrp, flags); } - - /* Preempt the runlist */ - nvkm_wr32(device, 0x2638, BIT(runl)); - - /* Schedule recovery. */ - nvkm_warn(subdev, "engine %d: scheduled for recovery\n", engn); - schedule_work(&fifo->recover.work); } static void -tu102_fifo_fault(struct nvkm_fifo *base, struct nvkm_fault_data *info) +tu102_fifo_intr_ctxsw_timeout(struct nvkm_fifo *fifo) { - struct gk104_fifo *fifo = gk104_fifo(base); - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - const struct nvkm_enum *er, *ee, *ec, *ea; - struct nvkm_engine *engine = NULL; - struct nvkm_fifo_chan *chan; - unsigned long flags; - const char *en = ""; - char ct[8] = "HUB/"; - int engn; - - er = nvkm_enum_find(fifo->func->fault.reason, info->reason); - ee = nvkm_enum_find(fifo->func->fault.engine, info->engine); - if (info->hub) { - ec = nvkm_enum_find(fifo->func->fault.hubclient, info->client); - } else { - ec = nvkm_enum_find(fifo->func->fault.gpcclient, info->client); - snprintf(ct, sizeof(ct), "GPC%d/", info->gpc); - } - ea = nvkm_enum_find(fifo->func->fault.access, info->access); - - if (ee && ee->data2) { - switch (ee->data2) { - case NVKM_SUBDEV_BAR: - nvkm_bar_bar1_reset(device); - break; - case NVKM_SUBDEV_INSTMEM: - nvkm_bar_bar2_reset(device); - break; - case NVKM_ENGINE_IFB: - nvkm_mask(device, 0x001718, 0x00000000, 0x00000000); - break; - default: - engine = nvkm_device_engine(device, ee->data2, 0); - break; - } - } - - if (ee == NULL) { - struct nvkm_subdev *subdev = nvkm_top_fault(device, info->engine); - if (subdev) { - if (subdev->func == &nvkm_engine) - engine = container_of(subdev, typeof(*engine), subdev); - en = engine->subdev.name; + struct nvkm_device *device = fifo->engine.subdev.device; + struct nvkm_runl *runl; + struct nvkm_engn *engn; + u32 engm = nvkm_rd32(device, 0x002a30); + u32 info; + + nvkm_runl_foreach(runl, fifo) { + nvkm_runl_foreach_engn_cond(engn, runl, engm & BIT(engn->id)) { + info = nvkm_rd32(device, 0x003200 + (engn->id * 4)); + tu102_fifo_intr_ctxsw_timeout_info(engn, info); } - } else { - en = ee->name; } - spin_lock_irqsave(&fifo->base.lock, flags); - chan = nvkm_fifo_chan_inst_locked(&fifo->base, info->inst); - - nvkm_error(subdev, - "fault %02x [%s] at %016llx engine %02x [%s] client %02x " - "[%s%s] reason %02x [%s] on channel %d [%010llx %s]\n", - info->access, ea ? ea->name : "", info->addr, - info->engine, ee ? ee->name : en, - info->client, ct, ec ? ec->name : "", - info->reason, er ? er->name : "", chan ? chan->chid : -1, - info->inst, chan ? chan->object.client->name : "unknown"); - - /* Kill the channel that caused the fault. */ - if (chan) - tu102_fifo_recover_chan(&fifo->base, chan->chid); - - /* Channel recovery will probably have already done this for the - * correct engine(s), but just in case we can't find the channel - * information... - */ - for (engn = 0; engn < fifo->engine_nr && engine; engn++) { - if (fifo->engine[engn].engine == engine) { - tu102_fifo_recover_engn(fifo, engn); - break; - } - } - - spin_unlock_irqrestore(&fifo->base.lock, flags); -} - -static void -tu102_fifo_intr_ctxsw_timeout(struct gk104_fifo *fifo) -{ - struct nvkm_device *device = fifo->base.engine.subdev.device; - unsigned long flags, engm; - u32 engn; - - spin_lock_irqsave(&fifo->base.lock, flags); - - engm = nvkm_rd32(device, 0x2a30); - nvkm_wr32(device, 0x2a30, engm); - - for_each_set_bit(engn, &engm, 32) - tu102_fifo_recover_engn(fifo, engn); - - spin_unlock_irqrestore(&fifo->base.lock, flags); + nvkm_wr32(device, 0x002a30, engm); } static void -tu102_fifo_intr_sched(struct gk104_fifo *fifo) +tu102_fifo_intr_sched(struct nvkm_fifo *fifo) { - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; - struct nvkm_device *device = subdev->device; - u32 intr = nvkm_rd32(device, 0x00254c); + struct nvkm_subdev *subdev = &fifo->engine.subdev; + u32 intr = nvkm_rd32(subdev->device, 0x00254c); u32 code = intr & 0x000000ff; nvkm_error(subdev, "SCHED_ERROR %02x\n", code); } -static void -tu102_fifo_intr(struct nvkm_fifo *base) +static irqreturn_t +tu102_fifo_intr(struct nvkm_inth *inth) { - struct gk104_fifo *fifo = gk104_fifo(base); - struct nvkm_subdev *subdev = &fifo->base.engine.subdev; + struct nvkm_fifo *fifo = container_of(inth, typeof(*fifo), engine.subdev.inth); + struct nvkm_subdev *subdev = &fifo->engine.subdev; struct nvkm_device *device = subdev->device; u32 mask = nvkm_rd32(device, 0x002140); u32 stat = nvkm_rd32(device, 0x002100) & mask; @@ -412,17 +226,8 @@ tu102_fifo_intr(struct nvkm_fifo *base) } if (stat & 0x20000000) { - u32 mask = nvkm_rd32(device, 0x0025a0); - - while (mask) { - u32 unit = __ffs(mask); - - gk104_fifo_intr_pbdma_0(fifo, unit); - gk104_fifo_intr_pbdma_1(fifo, unit); - nvkm_wr32(device, 0x0025a0, (1 << unit)); - mask &= ~(1 << unit); - } - stat &= ~0x20000000; + if (gf100_fifo_intr_pbdma(fifo)) + stat &= ~0x20000000; } if (stat & 0x40000000) { @@ -432,46 +237,50 @@ tu102_fifo_intr(struct nvkm_fifo *base) if (stat & 0x80000000) { nvkm_wr32(device, 0x002100, 0x80000000); - gk104_fifo_intr_engine(fifo); + nvkm_event_ntfy(&fifo->nonstall.event, 0, NVKM_FIFO_NONSTALL_EVENT); stat &= ~0x80000000; } if (stat) { nvkm_error(subdev, "INTR %08x\n", stat); + spin_lock(&fifo->lock); nvkm_mask(device, 0x002140, stat, 0x00000000); + spin_unlock(&fifo->lock); nvkm_wr32(device, 0x002100, stat); } + + return IRQ_HANDLED; +} + +static void +tu102_fifo_init_pbdmas(struct nvkm_fifo *fifo, u32 mask) +{ + /* Not directly related to PBDMAs, but, enables doorbell to function. */ + nvkm_mask(fifo->engine.subdev.device, 0xb65000, 0x80000000, 0x80000000); } static const struct nvkm_fifo_func -tu102_fifo_ = { - .dtor = gk104_fifo_dtor, - .oneinit = gk104_fifo_oneinit, - .info = gk104_fifo_info, +tu102_fifo = { + .chid_nr = gm200_fifo_chid_nr, + .chid_ctor = gk110_fifo_chid_ctor, + .runq_nr = gm200_fifo_runq_nr, + .runl_ctor = gk104_fifo_runl_ctor, .init = gk104_fifo_init, - .fini = gk104_fifo_fini, + .init_pbdmas = tu102_fifo_init_pbdmas, .intr = tu102_fifo_intr, - .fault = tu102_fifo_fault, - .engine_id = gk104_fifo_engine_id, - .id_engine = gk104_fifo_id_engine, - .uevent_init = gk104_fifo_uevent_init, - .uevent_fini = gk104_fifo_uevent_fini, - .recover_chan = tu102_fifo_recover_chan, - .class_get = gk104_fifo_class_get, - .class_new = gk104_fifo_class_new, + .mmu_fault = &tu102_fifo_mmu_fault, + .nonstall = &gf100_fifo_nonstall, + .runl = &tu102_runl, + .runq = &gv100_runq, + .engn = &gv100_engn, + .engn_ce = &gv100_engn_ce, + .cgrp = {{ 0, 0, KEPLER_CHANNEL_GROUP_A }, &gk110_cgrp, .force = true }, + .chan = {{ 0, 0, TURING_CHANNEL_GPFIFO_A }, &tu102_chan }, }; int tu102_fifo_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fifo **pfifo) { - struct gk104_fifo *fifo; - - if (!(fifo = kzalloc(sizeof(*fifo), GFP_KERNEL))) - return -ENOMEM; - fifo->func = &tu102_fifo; - INIT_WORK(&fifo->recover.work, tu102_fifo_recover_work); - *pfifo = &fifo->base; - - return nvkm_fifo_ctor(&tu102_fifo_, device, type, inst, 4096, &fifo->base); + return nvkm_fifo_new_(&tu102_fifo, device, type, inst, pfifo); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/ucgrp.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/ucgrp.c new file mode 100644 index 000000000000..52c594dfb1b8 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/ucgrp.c @@ -0,0 +1,125 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#define nvkm_ucgrp(p) container_of((p), struct nvkm_ucgrp, object) +#include "priv.h" +#include "cgrp.h" +#include "runl.h" + +#include <subdev/mmu.h> + +#include <nvif/if0021.h> + +struct nvkm_ucgrp { + struct nvkm_object object; + struct nvkm_cgrp *cgrp; +}; + +static int +nvkm_ucgrp_chan_new(const struct nvkm_oclass *oclass, void *argv, u32 argc, + struct nvkm_object **pobject) +{ + struct nvkm_cgrp *cgrp = nvkm_ucgrp(oclass->parent)->cgrp; + + return nvkm_uchan_new(cgrp->runl->fifo, cgrp, oclass, argv, argc, pobject); +} + +static int +nvkm_ucgrp_sclass(struct nvkm_object *object, int index, struct nvkm_oclass *oclass) +{ + struct nvkm_cgrp *cgrp = nvkm_ucgrp(object)->cgrp; + struct nvkm_fifo *fifo = cgrp->runl->fifo; + const struct nvkm_fifo_func_chan *chan = &fifo->func->chan; + int c = 0; + + /* *_CHANNEL_GPFIFO_* */ + if (chan->user.oclass) { + if (c++ == index) { + oclass->base = chan->user; + oclass->ctor = nvkm_ucgrp_chan_new; + return 0; + } + } + + return -EINVAL; +} + +static void * +nvkm_ucgrp_dtor(struct nvkm_object *object) +{ + struct nvkm_ucgrp *ucgrp = nvkm_ucgrp(object); + + nvkm_cgrp_unref(&ucgrp->cgrp); + return ucgrp; +} + +static const struct nvkm_object_func +nvkm_ucgrp = { + .dtor = nvkm_ucgrp_dtor, + .sclass = nvkm_ucgrp_sclass, +}; + +int +nvkm_ucgrp_new(struct nvkm_fifo *fifo, const struct nvkm_oclass *oclass, void *argv, u32 argc, + struct nvkm_object **pobject) +{ + union nvif_cgrp_args *args = argv; + struct nvkm_runl *runl; + struct nvkm_vmm *vmm; + struct nvkm_ucgrp *ucgrp; + int ret; + + if (argc < sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + argc -= sizeof(args->v0); + + if (args->v0.namelen != argc) + return -EINVAL; + + /* Lookup objects referenced in args. */ + runl = nvkm_runl_get(fifo, args->v0.runlist, 0); + if (!runl) + return -EINVAL; + + vmm = nvkm_uvmm_search(oclass->client, args->v0.vmm); + if (IS_ERR(vmm)) + return PTR_ERR(vmm); + + /* Allocate channel group. */ + if (!(ucgrp = kzalloc(sizeof(*ucgrp), GFP_KERNEL))) { + ret = -ENOMEM; + goto done; + } + + nvkm_object_ctor(&nvkm_ucgrp, oclass, &ucgrp->object); + *pobject = &ucgrp->object; + + ret = nvkm_cgrp_new(runl, args->v0.name, vmm, true, &ucgrp->cgrp); + if (ret) + goto done; + + /* Return channel group info to caller. */ + args->v0.cgid = ucgrp->cgrp->id; + +done: + nvkm_vmm_unref(&vmm); + return ret; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/uchan.c b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/uchan.c new file mode 100644 index 000000000000..1dac95ae7b43 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/uchan.c @@ -0,0 +1,409 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#define nvkm_uchan(p) container_of((p), struct nvkm_uchan, object) +#include "priv.h" +#include "cgrp.h" +#include "chan.h" +#include "chid.h" +#include "runl.h" + +#include <core/gpuobj.h> +#include <core/oproxy.h> +#include <subdev/mmu.h> +#include <engine/dma.h> + +#include <nvif/if0020.h> + +struct nvkm_uchan { + struct nvkm_object object; + struct nvkm_chan *chan; +}; + +static int +nvkm_uchan_uevent(struct nvkm_object *object, void *argv, u32 argc, struct nvkm_uevent *uevent) +{ + struct nvkm_chan *chan = nvkm_uchan(object)->chan; + struct nvkm_runl *runl = chan->cgrp->runl; + union nvif_chan_event_args *args = argv; + + if (!uevent) + return 0; + if (argc != sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + + switch (args->v0.type) { + case NVIF_CHAN_EVENT_V0_NON_STALL_INTR: + return nvkm_uevent_add(uevent, &runl->fifo->nonstall.event, 0, + NVKM_FIFO_NONSTALL_EVENT, NULL); + case NVIF_CHAN_EVENT_V0_KILLED: + return nvkm_uevent_add(uevent, &runl->chid->event, chan->id, + NVKM_CHAN_EVENT_ERRORED, NULL); + default: + break; + } + + return -ENOSYS; +} + +struct nvkm_uobj { + struct nvkm_oproxy oproxy; + struct nvkm_chan *chan; + struct nvkm_cctx *cctx; + int hash; +}; + +static int +nvkm_uchan_object_fini_1(struct nvkm_oproxy *oproxy, bool suspend) +{ + struct nvkm_uobj *uobj = container_of(oproxy, typeof(*uobj), oproxy); + struct nvkm_chan *chan = uobj->chan; + struct nvkm_cctx *cctx = uobj->cctx; + struct nvkm_ectx *ectx = cctx->vctx->ectx; + + if (!ectx->object) + return 0; + + /* Unbind engine context from channel, if no longer required. */ + if (refcount_dec_and_mutex_lock(&cctx->uses, &chan->cgrp->mutex)) { + nvkm_chan_cctx_bind(chan, ectx->engn, NULL); + + if (refcount_dec_and_test(&ectx->uses)) + nvkm_object_fini(ectx->object, false); + mutex_unlock(&chan->cgrp->mutex); + } + + return 0; +} + +static int +nvkm_uchan_object_init_0(struct nvkm_oproxy *oproxy) +{ + struct nvkm_uobj *uobj = container_of(oproxy, typeof(*uobj), oproxy); + struct nvkm_chan *chan = uobj->chan; + struct nvkm_cctx *cctx = uobj->cctx; + struct nvkm_ectx *ectx = cctx->vctx->ectx; + int ret = 0; + + if (!ectx->object) + return 0; + + /* Bind engine context to channel, if it hasn't been already. */ + if (!refcount_inc_not_zero(&cctx->uses)) { + mutex_lock(&chan->cgrp->mutex); + if (!refcount_inc_not_zero(&cctx->uses)) { + if (!refcount_inc_not_zero(&ectx->uses)) { + ret = nvkm_object_init(ectx->object); + if (ret == 0) + refcount_set(&ectx->uses, 1); + } + + if (ret == 0) { + nvkm_chan_cctx_bind(chan, ectx->engn, cctx); + refcount_set(&cctx->uses, 1); + } + } + mutex_unlock(&chan->cgrp->mutex); + } + + return ret; +} + +static void +nvkm_uchan_object_dtor(struct nvkm_oproxy *oproxy) +{ + struct nvkm_uobj *uobj = container_of(oproxy, typeof(*uobj), oproxy); + struct nvkm_engn *engn; + + if (!uobj->cctx) + return; + + engn = uobj->cctx->vctx->ectx->engn; + if (engn->func->ramht_del) + engn->func->ramht_del(uobj->chan, uobj->hash); + + nvkm_chan_cctx_put(uobj->chan, &uobj->cctx); +} + +static const struct nvkm_oproxy_func +nvkm_uchan_object = { + .dtor[1] = nvkm_uchan_object_dtor, + .init[0] = nvkm_uchan_object_init_0, + .fini[1] = nvkm_uchan_object_fini_1, +}; + +static int +nvkm_uchan_object_new(const struct nvkm_oclass *oclass, void *argv, u32 argc, + struct nvkm_object **pobject) +{ + struct nvkm_chan *chan = nvkm_uchan(oclass->parent)->chan; + struct nvkm_cgrp *cgrp = chan->cgrp; + struct nvkm_engn *engn; + struct nvkm_uobj *uobj; + int ret; + + /* Lookup host engine state for target engine. */ + engn = nvkm_runl_find_engn(engn, cgrp->runl, engn->engine == oclass->engine); + if (WARN_ON(!engn)) + return -EINVAL; + + /* Allocate SW object. */ + if (!(uobj = kzalloc(sizeof(*uobj), GFP_KERNEL))) + return -ENOMEM; + + nvkm_oproxy_ctor(&nvkm_uchan_object, oclass, &uobj->oproxy); + uobj->chan = chan; + *pobject = &uobj->oproxy.base; + + /* Ref. channel context for target engine.*/ + ret = nvkm_chan_cctx_get(chan, engn, &uobj->cctx, oclass->client); + if (ret) + return ret; + + /* Allocate HW object. */ + ret = oclass->base.ctor(&(const struct nvkm_oclass) { + .base = oclass->base, + .engn = oclass->engn, + .handle = oclass->handle, + .object = oclass->object, + .client = oclass->client, + .parent = uobj->cctx->vctx->ectx->object ?: oclass->parent, + .engine = engn->engine, + }, argv, argc, &uobj->oproxy.object); + if (ret) + return ret; + + if (engn->func->ramht_add) { + uobj->hash = engn->func->ramht_add(engn, uobj->oproxy.object, uobj->chan); + if (uobj->hash < 0) + return uobj->hash; + } + + return 0; +} + +static int +nvkm_uchan_sclass(struct nvkm_object *object, int index, struct nvkm_oclass *oclass) +{ + struct nvkm_chan *chan = nvkm_uchan(object)->chan; + struct nvkm_engn *engn; + int ret, runq = 0; + + nvkm_runl_foreach_engn(engn, chan->cgrp->runl) { + struct nvkm_engine *engine = engn->engine; + int c = 0; + + /* Each runqueue, on runlists with multiple, has its own LCE. */ + if (engn->runl->func->runqs) { + if (engine->subdev.type == NVKM_ENGINE_CE) { + if (chan->runq != runq++) + continue; + } + } + + oclass->engine = engine; + oclass->base.oclass = 0; + + if (engine->func->fifo.sclass) { + ret = engine->func->fifo.sclass(oclass, index); + if (oclass->base.oclass) { + if (!oclass->base.ctor) + oclass->base.ctor = nvkm_object_new; + oclass->ctor = nvkm_uchan_object_new; + return 0; + } + + index -= ret; + continue; + } + + while (engine->func->sclass[c].oclass) { + if (c++ == index) { + oclass->base = engine->func->sclass[index]; + if (!oclass->base.ctor) + oclass->base.ctor = nvkm_object_new; + oclass->ctor = nvkm_uchan_object_new; + return 0; + } + } + + index -= c; + } + + return -EINVAL; +} + +static int +nvkm_uchan_map(struct nvkm_object *object, void *argv, u32 argc, + enum nvkm_object_map *type, u64 *addr, u64 *size) +{ + struct nvkm_chan *chan = nvkm_uchan(object)->chan; + struct nvkm_device *device = chan->cgrp->runl->fifo->engine.subdev.device; + + if (chan->func->userd->bar < 0) + return -ENOSYS; + + *type = NVKM_OBJECT_MAP_IO; + *addr = device->func->resource_addr(device, chan->func->userd->bar) + + chan->func->userd->base + chan->userd.base; + *size = chan->func->userd->size; + return 0; +} + +static int +nvkm_uchan_fini(struct nvkm_object *object, bool suspend) +{ + struct nvkm_chan *chan = nvkm_uchan(object)->chan; + + nvkm_chan_block(chan); + nvkm_chan_remove(chan, true); + + if (chan->func->unbind) + chan->func->unbind(chan); + + return 0; +} + +static int +nvkm_uchan_init(struct nvkm_object *object) +{ + struct nvkm_chan *chan = nvkm_uchan(object)->chan; + + if (atomic_read(&chan->errored)) + return 0; + + if (chan->func->bind) + chan->func->bind(chan); + + nvkm_chan_allow(chan); + nvkm_chan_insert(chan); + return 0; +} + +static void * +nvkm_uchan_dtor(struct nvkm_object *object) +{ + struct nvkm_uchan *uchan = nvkm_uchan(object); + + nvkm_chan_del(&uchan->chan); + return uchan; +} + +static const struct nvkm_object_func +nvkm_uchan = { + .dtor = nvkm_uchan_dtor, + .init = nvkm_uchan_init, + .fini = nvkm_uchan_fini, + .map = nvkm_uchan_map, + .sclass = nvkm_uchan_sclass, + .uevent = nvkm_uchan_uevent, +}; + +int +nvkm_uchan_new(struct nvkm_fifo *fifo, struct nvkm_cgrp *cgrp, const struct nvkm_oclass *oclass, + void *argv, u32 argc, struct nvkm_object **pobject) +{ + union nvif_chan_args *args = argv; + struct nvkm_runl *runl; + struct nvkm_vmm *vmm = NULL; + struct nvkm_dmaobj *ctxdma = NULL; + struct nvkm_memory *userd = NULL; + struct nvkm_uchan *uchan; + struct nvkm_chan *chan; + int ret; + + if (argc < sizeof(args->v0) || args->v0.version != 0) + return -ENOSYS; + argc -= sizeof(args->v0); + + if (args->v0.namelen != argc) + return -EINVAL; + + /* Lookup objects referenced in args. */ + runl = nvkm_runl_get(fifo, args->v0.runlist, 0); + if (!runl) + return -EINVAL; + + if (args->v0.vmm) { + vmm = nvkm_uvmm_search(oclass->client, args->v0.vmm); + if (IS_ERR(vmm)) + return PTR_ERR(vmm); + } + + if (args->v0.ctxdma) { + ctxdma = nvkm_dmaobj_search(oclass->client, args->v0.ctxdma); + if (IS_ERR(ctxdma)) { + ret = PTR_ERR(ctxdma); + goto done; + } + } + + if (args->v0.huserd) { + userd = nvkm_umem_search(oclass->client, args->v0.huserd); + if (IS_ERR(userd)) { + ret = PTR_ERR(userd); + userd = NULL; + goto done; + } + } + + /* Allocate channel. */ + if (!(uchan = kzalloc(sizeof(*uchan), GFP_KERNEL))) { + ret = -ENOMEM; + goto done; + } + + nvkm_object_ctor(&nvkm_uchan, oclass, &uchan->object); + *pobject = &uchan->object; + + ret = nvkm_chan_new_(fifo->func->chan.func, runl, args->v0.runq, cgrp, args->v0.name, + args->v0.priv != 0, args->v0.devm, vmm, ctxdma, args->v0.offset, + args->v0.length, userd, args->v0.ouserd, &uchan->chan); + if (ret) + goto done; + + chan = uchan->chan; + + /* Return channel info to caller. */ + if (chan->func->doorbell_handle) + args->v0.token = chan->func->doorbell_handle(chan); + else + args->v0.token = ~0; + + args->v0.chid = chan->id; + + switch (nvkm_memory_target(chan->inst->memory)) { + case NVKM_MEM_TARGET_INST: args->v0.aper = NVIF_CHAN_V0_INST_APER_INST; break; + case NVKM_MEM_TARGET_VRAM: args->v0.aper = NVIF_CHAN_V0_INST_APER_VRAM; break; + case NVKM_MEM_TARGET_HOST: args->v0.aper = NVIF_CHAN_V0_INST_APER_HOST; break; + case NVKM_MEM_TARGET_NCOH: args->v0.aper = NVIF_CHAN_V0_INST_APER_NCOH; break; + default: + WARN_ON(1); + ret = -EFAULT; + break; + } + + args->v0.inst = nvkm_memory_addr(chan->inst->memory); +done: + nvkm_memory_unref(&userd); + nvkm_vmm_unref(&vmm); + return ret; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/user.h b/drivers/gpu/drm/nouveau/nvkm/engine/fifo/user.h deleted file mode 100644 index 54a3a3092cc0..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/user.h +++ /dev/null @@ -1,8 +0,0 @@ -#ifndef __NVKM_FIFO_USER_H__ -#define __NVKM_FIFO_USER_H__ -#include "priv.h" -int gv100_fifo_user_new(const struct nvkm_oclass *, void *, u32, - struct nvkm_object **); -int tu102_fifo_user_new(const struct nvkm_oclass *, void *, u32, - struct nvkm_object **); -#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/Kbuild b/drivers/gpu/drm/nouveau/nvkm/engine/gr/Kbuild index 558c86fd8e82..b5418f05ccd8 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/Kbuild @@ -40,6 +40,7 @@ nvkm-y += nvkm/engine/gr/gp108.o nvkm-y += nvkm/engine/gr/gp10b.o nvkm-y += nvkm/engine/gr/gv100.o nvkm-y += nvkm/engine/gr/tu102.o +nvkm-y += nvkm/engine/gr/ga102.o nvkm-y += nvkm/engine/gr/ctxnv40.o nvkm-y += nvkm/engine/gr/ctxnv50.o @@ -63,3 +64,4 @@ nvkm-y += nvkm/engine/gr/ctxgp104.o nvkm-y += nvkm/engine/gr/ctxgp107.o nvkm-y += nvkm/engine/gr/ctxgv100.o nvkm-y += nvkm/engine/gr/ctxtu102.o +nvkm-y += nvkm/engine/gr/ctxga102.o diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/base.c index 61759f54406e..71b824e6da9d 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/base.c @@ -136,6 +136,17 @@ nvkm_gr_oneinit(struct nvkm_engine *engine) } static int +nvkm_gr_reset(struct nvkm_engine *engine) +{ + struct nvkm_gr *gr = nvkm_gr(engine); + + if (gr->func->reset) + return gr->func->reset(gr); + + return -ENOSYS; +} + +static int nvkm_gr_init(struct nvkm_engine *engine) { struct nvkm_gr *gr = nvkm_gr(engine); @@ -166,6 +177,7 @@ nvkm_gr = { .oneinit = nvkm_gr_oneinit, .init = nvkm_gr_init, .fini = nvkm_gr_fini, + .reset = nvkm_gr_reset, .intr = nvkm_gr_intr, .tile = nvkm_gr_tile, .chsw_load = nvkm_gr_chsw_load, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxga102.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxga102.c new file mode 100644 index 000000000000..11461adf5036 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxga102.c @@ -0,0 +1,77 @@ +/* + * Copyright 2019 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "ctxgf100.h" + +static void +ga102_grctx_generate_sm_id(struct gf100_gr *gr, int gpc, int tpc, int sm) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + + tpc = gv100_gr_nonpes_aware_tpc(gr, gpc, tpc); + + nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x608), sm); +} + +static void +ga102_grctx_generate_unkn(struct gf100_gr *gr) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + + nvkm_mask(device, 0x41980c, 0x00000010, 0x00000010); + nvkm_mask(device, 0x41be08, 0x00000004, 0x00000004); +} + +static void +ga102_grctx_generate_r419ea8(struct gf100_gr *gr) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + + nvkm_wr32(device, 0x419ea8, nvkm_rd32(device, 0x504728) | 0x08000000); +} + +const struct gf100_grctx_func +ga102_grctx = { + .main = gf100_grctx_generate_main, + .unkn = ga102_grctx_generate_unkn, + .bundle = gm107_grctx_generate_bundle, + .bundle_size = 0x3000, + .bundle_min_gpm_fifo_depth = 0x180, + .bundle_token_limit = 0x1140, + .pagepool = gp100_grctx_generate_pagepool, + .pagepool_size = 0x20000, + .attrib_cb_size = gp102_grctx_generate_attrib_cb_size, + .attrib_cb = gv100_grctx_generate_attrib_cb, + .attrib = gv100_grctx_generate_attrib, + .attrib_nr_max = 0x800, + .attrib_nr = 0x4a1, + .alpha_nr_max = 0xc00, + .alpha_nr = 0x800, + .unknown_size = 0x80000, + .unknown = tu102_grctx_generate_unknown, + .gfxp_nr = 0xd28, + .sm_id = ga102_grctx_generate_sm_id, + .skip_pd_num_tpc_per_gpc = true, + .rop_mapping = gv100_grctx_generate_rop_mapping, + .r406500 = gm200_grctx_generate_r406500, + .r400088 = gv100_grctx_generate_r400088, + .r419ea8 = ga102_grctx_generate_r419ea8, +}; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.c index 297915719bf2..cb390e0134a2 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.c @@ -26,6 +26,7 @@ #include <subdev/fb.h> #include <subdev/mc.h> #include <subdev/timer.h> +#include <engine/fifo.h> /******************************************************************************* * PGRAPH context register lists @@ -990,43 +991,16 @@ gf100_grctx_pack_tpc[] = { * PGRAPH context implementation ******************************************************************************/ -int -gf100_grctx_mmio_data(struct gf100_grctx *info, u32 size, u32 align, bool priv) -{ - if (info->data) { - info->buffer[info->buffer_nr] = round_up(info->addr, align); - info->addr = info->buffer[info->buffer_nr] + size; - info->data->size = size; - info->data->align = align; - info->data->priv = priv; - info->data++; - return info->buffer_nr++; - } - return -1; -} - void -gf100_grctx_mmio_item(struct gf100_grctx *info, u32 addr, u32 data, - int shift, int buffer) +gf100_grctx_patch_wr32(struct gf100_gr_chan *chan, u32 addr, u32 data) { - struct nvkm_device *device = info->gr->base.engine.subdev.device; - if (info->data) { - if (shift >= 0) { - info->mmio->addr = addr; - info->mmio->data = data; - info->mmio->shift = shift; - info->mmio->buffer = buffer; - if (buffer >= 0) - data |= info->buffer[buffer] >> shift; - info->mmio++; - } else - return; - } else { - if (buffer >= 0) - return; + if (unlikely(!chan->mmio)) { + nvkm_wr32(chan->gr->base.engine.subdev.device, addr, data); + return; } - nvkm_wr32(device, addr, data); + nvkm_wo32(chan->mmio, chan->mmio_nr++ * 4, addr); + nvkm_wo32(chan->mmio, chan->mmio_nr++ * 4, data); } void @@ -1037,56 +1011,60 @@ gf100_grctx_generate_r419cb8(struct gf100_gr *gr) } void -gf100_grctx_generate_bundle(struct gf100_grctx *info) +gf100_grctx_generate_bundle(struct gf100_gr_chan *chan, u64 addr, u32 size) { - const struct gf100_grctx_func *grctx = info->gr->func->grctx; - const int s = 8; - const int b = mmio_vram(info, grctx->bundle_size, (1 << s), true); - mmio_refn(info, 0x408004, 0x00000000, s, b); - mmio_wr32(info, 0x408008, 0x80000000 | (grctx->bundle_size >> s)); - mmio_refn(info, 0x418808, 0x00000000, s, b); - mmio_wr32(info, 0x41880c, 0x80000000 | (grctx->bundle_size >> s)); + gf100_grctx_patch_wr32(chan, 0x408004, addr >> 8); + gf100_grctx_patch_wr32(chan, 0x408008, 0x80000000 | (size >> 8)); + gf100_grctx_patch_wr32(chan, 0x418808, addr >> 8); + gf100_grctx_patch_wr32(chan, 0x41880c, 0x80000000 | (size >> 8)); } void -gf100_grctx_generate_pagepool(struct gf100_grctx *info) +gf100_grctx_generate_pagepool(struct gf100_gr_chan *chan, u64 addr) { - const struct gf100_grctx_func *grctx = info->gr->func->grctx; - const int s = 8; - const int b = mmio_vram(info, grctx->pagepool_size, (1 << s), true); - mmio_refn(info, 0x40800c, 0x00000000, s, b); - mmio_wr32(info, 0x408010, 0x80000000); - mmio_refn(info, 0x419004, 0x00000000, s, b); - mmio_wr32(info, 0x419008, 0x00000000); + gf100_grctx_patch_wr32(chan, 0x40800c, addr >> 8); + gf100_grctx_patch_wr32(chan, 0x408010, 0x80000000); + gf100_grctx_patch_wr32(chan, 0x419004, addr >> 8); + gf100_grctx_patch_wr32(chan, 0x419008, 0x00000000); } void -gf100_grctx_generate_attrib(struct gf100_grctx *info) +gf100_grctx_generate_attrib(struct gf100_gr_chan *chan) { - struct gf100_gr *gr = info->gr; + struct gf100_gr *gr = chan->gr; const struct gf100_grctx_func *grctx = gr->func->grctx; const u32 attrib = grctx->attrib_nr; - const u32 size = 0x20 * (grctx->attrib_nr_max + grctx->alpha_nr_max); - const int s = 12; - const int b = mmio_vram(info, size * gr->tpc_total, (1 << s), false); int gpc, tpc; u32 bo = 0; - mmio_refn(info, 0x418810, 0x80000000, s, b); - mmio_refn(info, 0x419848, 0x10000000, s, b); - mmio_wr32(info, 0x405830, (attrib << 16)); + gf100_grctx_patch_wr32(chan, 0x405830, (attrib << 16)); for (gpc = 0; gpc < gr->gpc_nr; gpc++) { for (tpc = 0; tpc < gr->tpc_nr[gpc]; tpc++) { const u32 o = TPC_UNIT(gpc, tpc, 0x0520); - mmio_skip(info, o, (attrib << 16) | ++bo); - mmio_wr32(info, o, (attrib << 16) | --bo); + + gf100_grctx_patch_wr32(chan, o, (attrib << 16) | bo); bo += grctx->attrib_nr_max; } } } void +gf100_grctx_generate_attrib_cb(struct gf100_gr_chan *chan, u64 addr, u32 size) +{ + gf100_grctx_patch_wr32(chan, 0x418810, 0x80000000 | addr >> 12); + gf100_grctx_patch_wr32(chan, 0x419848, 0x10000000 | addr >> 12); +} + +u32 +gf100_grctx_generate_attrib_cb_size(struct gf100_gr *gr) +{ + const struct gf100_grctx_func *grctx = gr->func->grctx; + + return 0x20 * (grctx->attrib_nr_max + grctx->alpha_nr_max) * gr->tpc_total; +} + +void gf100_grctx_generate_unkn(struct gf100_gr *gr) { } @@ -1361,8 +1339,9 @@ gf100_grctx_generate_floorsweep(struct gf100_gr *gr) } void -gf100_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info) +gf100_grctx_generate_main(struct gf100_gr_chan *chan) { + struct gf100_gr *gr = chan->gr; struct nvkm_device *device = gr->base.engine.subdev.device; const struct gf100_grctx_func *grctx = gr->func->grctx; u32 idle_timeout; @@ -1380,15 +1359,23 @@ gf100_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info) gf100_gr_mmio(gr, gr->sw_ctx); } + if (gr->func->init_419bd8) + gr->func->init_419bd8(gr); + if (grctx->r419ea8) + grctx->r419ea8(gr); + gf100_gr_wait_idle(gr); idle_timeout = nvkm_mask(device, 0x404154, 0xffffffff, 0x00000000); - grctx->pagepool(info); - grctx->bundle(info); - grctx->attrib(info); + grctx->pagepool(chan, chan->pagepool->addr); + grctx->bundle(chan, chan->bundle_cb->addr, grctx->bundle_size); + grctx->attrib_cb(chan, chan->attrib_cb->addr, grctx->attrib_cb_size(gr)); + grctx->attrib(chan); if (grctx->patch_ltc) - grctx->patch_ltc(info); + grctx->patch_ltc(chan); + if (grctx->unknown_size) + grctx->unknown(chan, chan->unknown->addr, grctx->unknown_size); grctx->unkn(gr); gf100_grctx_generate_floorsweep(gr); @@ -1396,12 +1383,23 @@ gf100_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info) gf100_gr_wait_idle(gr); if (grctx->r400088) grctx->r400088(gr, false); + if (gr->bundle) gf100_gr_icmd(gr, gr->bundle); else gf100_gr_icmd(gr, grctx->icmd); - if (grctx->sw_veid_bundle_init) + + if (gr->bundle_veid) + gf100_gr_icmd(gr, gr->bundle_veid); + else gf100_gr_icmd(gr, grctx->sw_veid_bundle_init); + + if (gr->bundle64) + gf100_gr_icmd(gr, gr->bundle64); + else + if (grctx->sw_bundle64_init) + gf100_gr_icmd(gr, grctx->sw_bundle64_init); + if (grctx->r400088) grctx->r400088(gr, true); nvkm_wr32(device, 0x404154, idle_timeout); @@ -1428,21 +1426,20 @@ gf100_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info) grctx->r408840(gr); if (grctx->r419c0c) grctx->r419c0c(gr); + + gf100_gr_wait_idle(gr); } #define CB_RESERVED 0x80000 int -gf100_grctx_generate(struct gf100_gr *gr) +gf100_grctx_generate(struct gf100_gr *gr, struct gf100_gr_chan *chan, struct nvkm_gpuobj *inst) { const struct gf100_grctx_func *grctx = gr->func->grctx; struct nvkm_subdev *subdev = &gr->base.engine.subdev; struct nvkm_device *device = subdev->device; - struct nvkm_memory *inst = NULL; struct nvkm_memory *data = NULL; - struct nvkm_vmm *vmm = NULL; struct nvkm_vma *ctx = NULL; - struct gf100_grctx info; int ret, i; u64 addr; @@ -1457,72 +1454,47 @@ gf100_grctx_generate(struct gf100_gr *gr) grctx->unkn88c(gr, true); /* Reset FECS. */ - nvkm_wr32(device, 0x409614, 0x00000070); - nvkm_usec(device, 10, NVKM_DELAY); - nvkm_mask(device, 0x409614, 0x00000700, 0x00000700); - nvkm_usec(device, 10, NVKM_DELAY); - nvkm_rd32(device, 0x409614); + gr->func->fecs.reset(gr); if (grctx->unkn88c) grctx->unkn88c(gr, false); /* NV_PGRAPH_FE_PWR_MODE_AUTO. */ nvkm_wr32(device, 0x404170, 0x00000010); + nvkm_msec(device, 2000, + if (!(nvkm_rd32(device, 0x404170) & 0x00000010)) + break; + ); /* Init SCC RAM. */ nvkm_wr32(device, 0x40802c, 0x00000001); - /* Allocate memory to for a "channel", which we'll use to generate - * the default context values. - */ - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, - 0x1000, 0x1000, true, &inst); - if (ret) - goto done; - - ret = nvkm_vmm_new(device, 0, 0, NULL, 0, NULL, "grctx", &vmm); - if (ret) - goto done; - - vmm->debug = subdev->debug; - - ret = nvkm_vmm_join(vmm, inst); - if (ret) - goto done; - + /* Allocate memory to store context, and dummy global context buffers. */ ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, CB_RESERVED + gr->size, 0, true, &data); if (ret) goto done; - ret = nvkm_vmm_get(vmm, 0, nvkm_memory_size(data), &ctx); + ret = nvkm_vmm_get(chan->vmm, 0, nvkm_memory_size(data), &ctx); if (ret) goto done; - ret = nvkm_memory_map(data, 0, vmm, ctx, NULL, 0); + ret = nvkm_memory_map(data, 0, chan->vmm, ctx, NULL, 0); if (ret) goto done; - /* Setup context pointer. */ nvkm_kmap(inst); nvkm_wo32(inst, 0x0210, lower_32_bits(ctx->addr + CB_RESERVED) | 4); nvkm_wo32(inst, 0x0214, upper_32_bits(ctx->addr + CB_RESERVED)); nvkm_done(inst); - /* Setup default state for mmio list construction. */ - info.gr = gr; - info.data = gr->mmio_data; - info.mmio = gr->mmio_list; - info.addr = ctx->addr; - info.buffer_nr = 0; - /* Make channel current. */ - addr = nvkm_memory_addr(inst) >> 12; + addr = inst->addr >> 12; if (gr->firmware) { ret = gf100_gr_fecs_bind_pointer(gr, 0x80000000 | addr); if (ret) - goto done; + goto done_inst; nvkm_kmap(data); nvkm_wo32(data, 0x1c, 1); @@ -1540,19 +1512,27 @@ gf100_grctx_generate(struct gf100_gr *gr) ); } - grctx->main(gr, &info); + grctx->main(chan); - /* Trigger a context unload by unsetting the "next channel valid" bit - * and faking a context switch interrupt. - */ - nvkm_mask(device, 0x409b04, 0x80000000, 0x00000000); - nvkm_wr32(device, 0x409000, 0x00000100); - if (nvkm_msec(device, 2000, - if (!(nvkm_rd32(device, 0x409b00) & 0x80000000)) - break; - ) < 0) { - ret = -EBUSY; - goto done; + if (!gr->firmware) { + /* Trigger a context unload by unsetting the "next channel valid" bit + * and faking a context switch interrupt. + */ + nvkm_mask(device, 0x409b04, 0x80000000, 0x00000000); + nvkm_wr32(device, 0x409000, 0x00000100); + if (nvkm_msec(device, 2000, + if (!(nvkm_rd32(device, 0x409b00) & 0x80000000)) + break; + ) < 0) { + ret = -EBUSY; + goto done_inst; + } + } else { + ret = gf100_gr_fecs_wfi_golden_save(gr, 0x80000000 | addr); + if (ret) + goto done_inst; + + nvkm_mask(device, 0x409b00, 0x80000000, 0x00000000); } gr->data = kmalloc(gr->size, GFP_KERNEL); @@ -1566,12 +1546,14 @@ gf100_grctx_generate(struct gf100_gr *gr) ret = -ENOMEM; } +done_inst: + nvkm_kmap(inst); + nvkm_wo32(inst, 0x0210, 0); + nvkm_wo32(inst, 0x0214, 0); + nvkm_done(inst); done: - nvkm_vmm_put(vmm, &ctx); + nvkm_vmm_put(chan->vmm, &ctx); nvkm_memory_unref(&data); - nvkm_vmm_part(vmm, inst); - nvkm_vmm_unref(&vmm); - nvkm_memory_unref(&inst); return ret; } @@ -1590,6 +1572,8 @@ gf100_grctx = { .bundle_size = 0x1800, .pagepool = gf100_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gf100_grctx_generate_attrib_cb, .attrib = gf100_grctx_generate_attrib, .attrib_nr_max = 0x324, .attrib_nr = 0x218, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.h b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.h index 32bbddc0993e..00dbeda7e346 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.h @@ -3,27 +3,12 @@ #define __NVKM_GRCTX_NVC0_H__ #include "gf100.h" -struct gf100_grctx { - struct gf100_gr *gr; - struct gf100_gr_data *data; - struct gf100_gr_mmio *mmio; - int buffer_nr; - u64 buffer[4]; - u64 addr; -}; - -int gf100_grctx_mmio_data(struct gf100_grctx *, u32 size, u32 align, bool priv); -void gf100_grctx_mmio_item(struct gf100_grctx *, u32 addr, u32 data, int s, int); - -#define mmio_vram(a,b,c,d) gf100_grctx_mmio_data((a), (b), (c), (d)) -#define mmio_refn(a,b,c,d,e) gf100_grctx_mmio_item((a), (b), (c), (d), (e)) -#define mmio_skip(a,b,c) mmio_refn((a), (b), (c), -1, -1) -#define mmio_wr32(a,b,c) mmio_refn((a), (b), (c), 0, -1) +void gf100_grctx_patch_wr32(struct gf100_gr_chan *, u32 addr, u32 data); struct gf100_grctx_func { void (*unkn88c)(struct gf100_gr *, bool on); /* main context generation function */ - void (*main)(struct gf100_gr *, struct gf100_grctx *); + void (*main)(struct gf100_gr_chan *); /* context-specific modify-on-first-load list generation function */ void (*unkn)(struct gf100_gr *); /* mmio context data */ @@ -37,23 +22,29 @@ struct gf100_grctx_func { const struct gf100_gr_pack *icmd; const struct gf100_gr_pack *mthd; const struct gf100_gr_pack *sw_veid_bundle_init; + const struct gf100_gr_pack *sw_bundle64_init; /* bundle circular buffer */ - void (*bundle)(struct gf100_grctx *); + void (*bundle)(struct gf100_gr_chan *, u64 addr, u32 size); u32 bundle_size; u32 bundle_min_gpm_fifo_depth; u32 bundle_token_limit; /* pagepool */ - void (*pagepool)(struct gf100_grctx *); + void (*pagepool)(struct gf100_gr_chan *, u64 addr); u32 pagepool_size; /* attribute(/alpha) circular buffer */ - void (*attrib)(struct gf100_grctx *); + u32 (*attrib_cb_size)(struct gf100_gr *); + void (*attrib_cb)(struct gf100_gr_chan *, u64 addr, u32 size); + void (*attrib)(struct gf100_gr_chan *); u32 attrib_nr_max; u32 attrib_nr; u32 alpha_nr_max; u32 alpha_nr; u32 gfxp_nr; + /* some other context buffer */ + void (*unknown)(struct gf100_gr_chan *, u64 addr, u32 size); + u32 unknown_size; /* other patch buffer stuff */ - void (*patch_ltc)(struct gf100_grctx *); + void (*patch_ltc)(struct gf100_gr_chan *); /* floorsweeping */ void (*sm_id)(struct gf100_gr *, int gpc, int tpc, int sm); void (*tpc_nr)(struct gf100_gr *, int gpc); @@ -78,14 +69,17 @@ struct gf100_grctx_func { void (*r419a3c)(struct gf100_gr *); void (*r408840)(struct gf100_gr *); void (*r419c0c)(struct gf100_gr *); + void (*r419ea8)(struct gf100_gr *); }; extern const struct gf100_grctx_func gf100_grctx; -int gf100_grctx_generate(struct gf100_gr *); -void gf100_grctx_generate_main(struct gf100_gr *, struct gf100_grctx *); -void gf100_grctx_generate_bundle(struct gf100_grctx *); -void gf100_grctx_generate_pagepool(struct gf100_grctx *); -void gf100_grctx_generate_attrib(struct gf100_grctx *); +int gf100_grctx_generate(struct gf100_gr *, struct gf100_gr_chan *, struct nvkm_gpuobj *inst); +void gf100_grctx_generate_main(struct gf100_gr_chan *); +void gf100_grctx_generate_pagepool(struct gf100_gr_chan *, u64); +void gf100_grctx_generate_bundle(struct gf100_gr_chan *, u64, u32); +u32 gf100_grctx_generate_attrib_cb_size(struct gf100_gr *); +void gf100_grctx_generate_attrib_cb(struct gf100_gr_chan *, u64, u32); +void gf100_grctx_generate_attrib(struct gf100_gr_chan *); void gf100_grctx_generate_unkn(struct gf100_gr *); void gf100_grctx_generate_floorsweep(struct gf100_gr *); void gf100_grctx_generate_sm_id(struct gf100_gr *, int, int, int); @@ -97,14 +91,14 @@ void gf100_grctx_generate_max_ways_evict(struct gf100_gr *); void gf100_grctx_generate_r419cb8(struct gf100_gr *); extern const struct gf100_grctx_func gf108_grctx; -void gf108_grctx_generate_attrib(struct gf100_grctx *); +void gf108_grctx_generate_attrib(struct gf100_gr_chan *); void gf108_grctx_generate_unkn(struct gf100_gr *); extern const struct gf100_grctx_func gf104_grctx; extern const struct gf100_grctx_func gf110_grctx; extern const struct gf100_grctx_func gf117_grctx; -void gf117_grctx_generate_attrib(struct gf100_grctx *); +void gf117_grctx_generate_attrib(struct gf100_gr_chan *); void gf117_grctx_generate_rop_mapping(struct gf100_gr *); void gf117_grctx_generate_dist_skip_table(struct gf100_gr *); @@ -115,9 +109,9 @@ void gk104_grctx_generate_alpha_beta_tables(struct gf100_gr *); void gk104_grctx_generate_gpc_tpc_nr(struct gf100_gr *); extern const struct gf100_grctx_func gk20a_grctx; -void gk104_grctx_generate_bundle(struct gf100_grctx *); -void gk104_grctx_generate_pagepool(struct gf100_grctx *); -void gk104_grctx_generate_patch_ltc(struct gf100_grctx *); +void gk104_grctx_generate_pagepool(struct gf100_gr_chan *, u64); +void gk104_grctx_generate_bundle(struct gf100_gr_chan *, u64, u32); +void gk104_grctx_generate_patch_ltc(struct gf100_gr_chan *); void gk104_grctx_generate_unkn(struct gf100_gr *); void gk104_grctx_generate_r418800(struct gf100_gr *); @@ -128,9 +122,10 @@ extern const struct gf100_grctx_func gk110b_grctx; extern const struct gf100_grctx_func gk208_grctx; extern const struct gf100_grctx_func gm107_grctx; -void gm107_grctx_generate_bundle(struct gf100_grctx *); -void gm107_grctx_generate_pagepool(struct gf100_grctx *); -void gm107_grctx_generate_attrib(struct gf100_grctx *); +void gm107_grctx_generate_pagepool(struct gf100_gr_chan *, u64); +void gm107_grctx_generate_bundle(struct gf100_gr_chan *, u64, u32); +void gm107_grctx_generate_attrib_cb(struct gf100_gr_chan *, u64, u32); +void gm107_grctx_generate_attrib(struct gf100_gr_chan *); void gm107_grctx_generate_sm_id(struct gf100_gr *, int, int, int); extern const struct gf100_grctx_func gm200_grctx; @@ -143,11 +138,13 @@ void gm200_grctx_generate_r419a3c(struct gf100_gr *); extern const struct gf100_grctx_func gm20b_grctx; extern const struct gf100_grctx_func gp100_grctx; -void gp100_grctx_generate_pagepool(struct gf100_grctx *); +void gp100_grctx_generate_pagepool(struct gf100_gr_chan *, u64); +void gp100_grctx_generate_attrib_cb(struct gf100_gr_chan *, u64, u32); void gp100_grctx_generate_smid_config(struct gf100_gr *); extern const struct gf100_grctx_func gp102_grctx; -void gp102_grctx_generate_attrib(struct gf100_grctx *); +u32 gp102_grctx_generate_attrib_cb_size(struct gf100_gr *); +void gp102_grctx_generate_attrib(struct gf100_gr_chan *); extern const struct gf100_grctx_func gp104_grctx; @@ -158,11 +155,15 @@ extern const struct gf100_grctx_func gv100_grctx; extern const struct gf100_grctx_func tu102_grctx; void gv100_grctx_unkn88c(struct gf100_gr *, bool); void gv100_grctx_generate_unkn(struct gf100_gr *); -extern const struct gf100_gr_init gv100_grctx_init_sw_veid_bundle_init_0[]; -void gv100_grctx_generate_attrib(struct gf100_grctx *); +void gv100_grctx_generate_attrib_cb(struct gf100_gr_chan *, u64, u32); +void gv100_grctx_generate_attrib(struct gf100_gr_chan *); void gv100_grctx_generate_rop_mapping(struct gf100_gr *); void gv100_grctx_generate_r400088(struct gf100_gr *, bool); +void tu102_grctx_generate_unknown(struct gf100_gr_chan *, u64, u32); + +extern const struct gf100_grctx_func ga102_grctx; + /* context init value lists */ extern const struct gf100_gr_pack gf100_grctx_pack_icmd[]; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf104.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf104.c index 7a0564b6e3c7..ba63a3b46518 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf104.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf104.c @@ -94,6 +94,8 @@ gf104_grctx = { .bundle_size = 0x1800, .pagepool = gf100_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gf100_grctx_generate_attrib_cb, .attrib = gf100_grctx_generate_attrib, .attrib_nr_max = 0x324, .attrib_nr = 0x218, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf108.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf108.c index dda2c32e6232..0bc2eab6ad98 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf108.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf108.c @@ -733,25 +733,20 @@ gf108_grctx_pack_tpc[] = { ******************************************************************************/ void -gf108_grctx_generate_attrib(struct gf100_grctx *info) +gf108_grctx_generate_attrib(struct gf100_gr_chan *chan) { - struct gf100_gr *gr = info->gr; + struct gf100_gr *gr = chan->gr; const struct gf100_grctx_func *grctx = gr->func->grctx; const u32 alpha = grctx->alpha_nr; const u32 beta = grctx->attrib_nr; - const u32 size = 0x20 * (grctx->attrib_nr_max + grctx->alpha_nr_max); - const int s = 12; - const int b = mmio_vram(info, size * gr->tpc_total, (1 << s), false); const int timeslice_mode = 1; const int max_batches = 0xffff; u32 bo = 0; u32 ao = bo + grctx->attrib_nr_max * gr->tpc_total; int gpc, tpc; - mmio_refn(info, 0x418810, 0x80000000, s, b); - mmio_refn(info, 0x419848, 0x10000000, s, b); - mmio_wr32(info, 0x405830, (beta << 16) | alpha); - mmio_wr32(info, 0x4064c4, ((alpha / 4) << 16) | max_batches); + gf100_grctx_patch_wr32(chan, 0x405830, (beta << 16) | alpha); + gf100_grctx_patch_wr32(chan, 0x4064c4, ((alpha / 4) << 16) | max_batches); for (gpc = 0; gpc < gr->gpc_nr; gpc++) { for (tpc = 0; tpc < gr->tpc_nr[gpc]; tpc++) { @@ -759,10 +754,10 @@ gf108_grctx_generate_attrib(struct gf100_grctx *info) const u32 b = beta; const u32 t = timeslice_mode; const u32 o = TPC_UNIT(gpc, tpc, 0x500); - mmio_skip(info, o + 0x20, (t << 28) | (b << 16) | ++bo); - mmio_wr32(info, o + 0x20, (t << 28) | (b << 16) | --bo); + + gf100_grctx_patch_wr32(chan, o + 0x20, (t << 28) | (b << 16) | bo); bo += grctx->attrib_nr_max; - mmio_wr32(info, o + 0x44, (a << 16) | ao); + gf100_grctx_patch_wr32(chan, o + 0x44, (a << 16) | ao); ao += grctx->alpha_nr_max; } } @@ -795,6 +790,8 @@ gf108_grctx = { .bundle_size = 0x1800, .pagepool = gf100_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gf100_grctx_generate_attrib_cb, .attrib = gf108_grctx_generate_attrib, .attrib_nr_max = 0x324, .attrib_nr = 0x218, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf110.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf110.c index f5cca5e6a4f2..64b723b0afb5 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf110.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf110.c @@ -342,6 +342,8 @@ gf110_grctx = { .bundle_size = 0x1800, .pagepool = gf100_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gf100_grctx_generate_attrib_cb, .attrib = gf100_grctx_generate_attrib, .attrib_nr_max = 0x324, .attrib_nr = 0x218, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf117.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf117.c index 276c282d19aa..e34c5da2a9ff 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf117.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf117.c @@ -241,38 +241,34 @@ gf117_grctx_generate_rop_mapping(struct gf100_gr *gr) } void -gf117_grctx_generate_attrib(struct gf100_grctx *info) +gf117_grctx_generate_attrib(struct gf100_gr_chan *chan) { - struct gf100_gr *gr = info->gr; + struct gf100_gr *gr = chan->gr; const struct gf100_grctx_func *grctx = gr->func->grctx; const u32 alpha = grctx->alpha_nr; const u32 beta = grctx->attrib_nr; - const u32 size = 0x20 * (grctx->attrib_nr_max + grctx->alpha_nr_max); - const int s = 12; - const int b = mmio_vram(info, size * gr->tpc_total, (1 << s), false); const int timeslice_mode = 1; const int max_batches = 0xffff; u32 bo = 0; u32 ao = bo + grctx->attrib_nr_max * gr->tpc_total; int gpc, ppc; - mmio_refn(info, 0x418810, 0x80000000, s, b); - mmio_refn(info, 0x419848, 0x10000000, s, b); - mmio_wr32(info, 0x405830, (beta << 16) | alpha); - mmio_wr32(info, 0x4064c4, ((alpha / 4) << 16) | max_batches); + gf100_grctx_patch_wr32(chan, 0x405830, (beta << 16) | alpha); + gf100_grctx_patch_wr32(chan, 0x4064c4, ((alpha / 4) << 16) | max_batches); for (gpc = 0; gpc < gr->gpc_nr; gpc++) { - for (ppc = 0; ppc < gr->ppc_nr[gpc]; ppc++) { + for (ppc = 0; ppc < gr->func->ppc_nr; ppc++) { const u32 a = alpha * gr->ppc_tpc_nr[gpc][ppc]; const u32 b = beta * gr->ppc_tpc_nr[gpc][ppc]; const u32 t = timeslice_mode; const u32 o = PPC_UNIT(gpc, ppc, 0); + if (!(gr->ppc_mask[gpc] & (1 << ppc))) continue; - mmio_skip(info, o + 0xc0, (t << 28) | (b << 16) | ++bo); - mmio_wr32(info, o + 0xc0, (t << 28) | (b << 16) | --bo); + + gf100_grctx_patch_wr32(chan, o + 0xc0, (t << 28) | (b << 16) | bo); bo += grctx->attrib_nr_max * gr->ppc_tpc_nr[gpc][ppc]; - mmio_wr32(info, o + 0xe4, (a << 16) | ao); + gf100_grctx_patch_wr32(chan, o + 0xe4, (a << 16) | ao); ao += grctx->alpha_nr_max * gr->ppc_tpc_nr[gpc][ppc]; } } @@ -294,6 +290,8 @@ gf117_grctx = { .bundle_size = 0x1800, .pagepool = gf100_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gf100_grctx_generate_attrib_cb, .attrib = gf117_grctx_generate_attrib, .attrib_nr_max = 0x324, .attrib_nr = 0x218, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf119.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf119.c index 0cfe46366af6..426ad1b8d426 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf119.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf119.c @@ -510,6 +510,8 @@ gf119_grctx = { .bundle_size = 0x1800, .pagepool = gf100_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gf100_grctx_generate_attrib_cb, .attrib = gf108_grctx_generate_attrib, .attrib_nr_max = 0x324, .attrib_nr = 0x218, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk104.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk104.c index 304e9d268bad..94233d0119df 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk104.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk104.c @@ -861,43 +861,33 @@ gk104_grctx_generate_r418800(struct gf100_gr *gr) } void -gk104_grctx_generate_patch_ltc(struct gf100_grctx *info) +gk104_grctx_generate_patch_ltc(struct gf100_gr_chan *chan) { - struct nvkm_device *device = info->gr->base.engine.subdev.device; + struct nvkm_device *device = chan->gr->base.engine.subdev.device; u32 data0 = nvkm_rd32(device, 0x17e91c); u32 data1 = nvkm_rd32(device, 0x17e920); + /*XXX: Figure out how to modify this correctly! */ - mmio_wr32(info, 0x17e91c, data0); - mmio_wr32(info, 0x17e920, data1); + gf100_grctx_patch_wr32(chan, 0x17e91c, data0); + gf100_grctx_patch_wr32(chan, 0x17e920, data1); } void -gk104_grctx_generate_bundle(struct gf100_grctx *info) +gk104_grctx_generate_bundle(struct gf100_gr_chan *chan, u64 addr, u32 size) { - const struct gf100_grctx_func *grctx = info->gr->func->grctx; - const u32 state_limit = min(grctx->bundle_min_gpm_fifo_depth, - grctx->bundle_size / 0x20); + const struct gf100_grctx_func *grctx = chan->gr->func->grctx; + const u32 state_limit = min(grctx->bundle_min_gpm_fifo_depth, size / 0x20); const u32 token_limit = grctx->bundle_token_limit; - const int s = 8; - const int b = mmio_vram(info, grctx->bundle_size, (1 << s), true); - mmio_refn(info, 0x408004, 0x00000000, s, b); - mmio_wr32(info, 0x408008, 0x80000000 | (grctx->bundle_size >> s)); - mmio_refn(info, 0x418808, 0x00000000, s, b); - mmio_wr32(info, 0x41880c, 0x80000000 | (grctx->bundle_size >> s)); - mmio_wr32(info, 0x4064c8, (state_limit << 16) | token_limit); + + gf100_grctx_generate_bundle(chan, addr, size); + gf100_grctx_patch_wr32(chan, 0x4064c8, (state_limit << 16) | token_limit); } void -gk104_grctx_generate_pagepool(struct gf100_grctx *info) +gk104_grctx_generate_pagepool(struct gf100_gr_chan *chan, u64 addr) { - const struct gf100_grctx_func *grctx = info->gr->func->grctx; - const int s = 8; - const int b = mmio_vram(info, grctx->pagepool_size, (1 << s), true); - mmio_refn(info, 0x40800c, 0x00000000, s, b); - mmio_wr32(info, 0x408010, 0x80000000); - mmio_refn(info, 0x419004, 0x00000000, s, b); - mmio_wr32(info, 0x419008, 0x00000000); - mmio_wr32(info, 0x4064cc, 0x80000000); + gf100_grctx_generate_pagepool(chan, addr); + gf100_grctx_patch_wr32(chan, 0x4064cc, 0x80000000); } void @@ -991,6 +981,8 @@ gk104_grctx = { .bundle_token_limit = 0x600, .pagepool = gk104_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gf100_grctx_generate_attrib_cb, .attrib = gf117_grctx_generate_attrib, .attrib_nr_max = 0x324, .attrib_nr = 0x218, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk110.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk110.c index 86547cfc38dc..4391458e1fb2 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk110.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk110.c @@ -838,6 +838,8 @@ gk110_grctx = { .bundle_token_limit = 0x7c0, .pagepool = gk104_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gf100_grctx_generate_attrib_cb, .attrib = gf117_grctx_generate_attrib, .attrib_nr_max = 0x324, .attrib_nr = 0x218, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk110b.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk110b.c index ebb947bd1446..7b9a34f9ec3c 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk110b.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk110b.c @@ -87,6 +87,8 @@ gk110b_grctx = { .bundle_token_limit = 0x600, .pagepool = gk104_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gf100_grctx_generate_attrib_cb, .attrib = gf117_grctx_generate_attrib, .attrib_nr_max = 0x324, .attrib_nr = 0x218, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk208.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk208.c index 4d40512b5c99..c78d07a8bb7d 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk208.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk208.c @@ -553,6 +553,8 @@ gk208_grctx = { .bundle_token_limit = 0x200, .pagepool = gk104_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gf100_grctx_generate_attrib_cb, .attrib = gf117_grctx_generate_attrib, .attrib_nr_max = 0x324, .attrib_nr = 0x218, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk20a.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk20a.c index c0d36bc601f9..ac5fdcb5cd3f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk20a.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgk20a.c @@ -25,8 +25,9 @@ #include <subdev/mc.h> static void -gk20a_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info) +gk20a_grctx_generate_main(struct gf100_gr_chan *chan) { + struct gf100_gr *gr = chan->gr; struct nvkm_device *device = gr->base.engine.subdev.device; const struct gf100_grctx_func *grctx = gr->func->grctx; u32 idle_timeout; @@ -38,7 +39,8 @@ gk20a_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info) idle_timeout = nvkm_mask(device, 0x404154, 0xffffffff, 0x00000000); - grctx->attrib(info); + grctx->attrib_cb(chan, chan->attrib_cb->addr, grctx->attrib_cb_size(gr)); + grctx->attrib(chan); grctx->unkn(gr); @@ -60,8 +62,8 @@ gk20a_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info) gf100_gr_wait_idle(gr); gf100_gr_icmd(gr, gr->bundle); - grctx->pagepool(info); - grctx->bundle(info); + grctx->pagepool(chan, chan->pagepool->addr); + grctx->bundle(chan, chan->bundle_cb->addr, grctx->bundle_size); } const struct gf100_grctx_func @@ -74,6 +76,8 @@ gk20a_grctx = { .bundle_token_limit = 0x100, .pagepool = gk104_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gf100_grctx_generate_attrib_cb, .attrib = gf117_grctx_generate_attrib, .attrib_nr_max = 0x240, .attrib_nr = 0x240, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm107.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm107.c index 0b3964e6b36e..beac66eb2a80 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm107.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm107.c @@ -876,75 +876,70 @@ gm107_grctx_generate_r419e00(struct gf100_gr *gr) } void -gm107_grctx_generate_bundle(struct gf100_grctx *info) +gm107_grctx_generate_bundle(struct gf100_gr_chan *chan, u64 addr, u32 size) { - const struct gf100_grctx_func *grctx = info->gr->func->grctx; - const u32 state_limit = min(grctx->bundle_min_gpm_fifo_depth, - grctx->bundle_size / 0x20); + const struct gf100_grctx_func *grctx = chan->gr->func->grctx; + const u32 state_limit = min(grctx->bundle_min_gpm_fifo_depth, size / 0x20); const u32 token_limit = grctx->bundle_token_limit; - const int s = 8; - const int b = mmio_vram(info, grctx->bundle_size, (1 << s), true); - mmio_refn(info, 0x408004, 0x00000000, s, b); - mmio_wr32(info, 0x408008, 0x80000000 | (grctx->bundle_size >> s)); - mmio_refn(info, 0x418e24, 0x00000000, s, b); - mmio_wr32(info, 0x418e28, 0x80000000 | (grctx->bundle_size >> s)); - mmio_wr32(info, 0x4064c8, (state_limit << 16) | token_limit); + + gf100_grctx_patch_wr32(chan, 0x408004, addr >> 8); + gf100_grctx_patch_wr32(chan, 0x408008, 0x80000000 | (size >> 8)); + gf100_grctx_patch_wr32(chan, 0x418e24, addr >> 8); + gf100_grctx_patch_wr32(chan, 0x418e28, 0x80000000 | (size >> 8)); + gf100_grctx_patch_wr32(chan, 0x4064c8, (state_limit << 16) | token_limit); } void -gm107_grctx_generate_pagepool(struct gf100_grctx *info) +gm107_grctx_generate_pagepool(struct gf100_gr_chan *chan, u64 addr) { - const struct gf100_grctx_func *grctx = info->gr->func->grctx; - const int s = 8; - const int b = mmio_vram(info, grctx->pagepool_size, (1 << s), true); - mmio_refn(info, 0x40800c, 0x00000000, s, b); - mmio_wr32(info, 0x408010, 0x80000000); - mmio_refn(info, 0x419004, 0x00000000, s, b); - mmio_wr32(info, 0x419008, 0x00000000); - mmio_wr32(info, 0x4064cc, 0x80000000); - mmio_wr32(info, 0x418e30, 0x80000000); /* guess at it being related */ + gk104_grctx_generate_pagepool(chan, addr); + gf100_grctx_patch_wr32(chan, 0x418e30, 0x80000000); } void -gm107_grctx_generate_attrib(struct gf100_grctx *info) +gm107_grctx_generate_attrib(struct gf100_gr_chan *chan) { - struct gf100_gr *gr = info->gr; + struct gf100_gr *gr = chan->gr; const struct gf100_grctx_func *grctx = gr->func->grctx; const u32 alpha = grctx->alpha_nr; const u32 attrib = grctx->attrib_nr; - const u32 size = 0x20 * (grctx->attrib_nr_max + grctx->alpha_nr_max); - const int s = 12; - const int b = mmio_vram(info, size * gr->tpc_total, (1 << s), false); const int max_batches = 0xffff; u32 bo = 0; u32 ao = bo + grctx->attrib_nr_max * gr->tpc_total; int gpc, ppc, n = 0; - mmio_refn(info, 0x418810, 0x80000000, s, b); - mmio_refn(info, 0x419848, 0x10000000, s, b); - mmio_refn(info, 0x419c2c, 0x10000000, s, b); - mmio_wr32(info, 0x405830, (attrib << 16) | alpha); - mmio_wr32(info, 0x4064c4, ((alpha / 4) << 16) | max_batches); + gf100_grctx_patch_wr32(chan, 0x405830, (attrib << 16) | alpha); + gf100_grctx_patch_wr32(chan, 0x4064c4, ((alpha / 4) << 16) | max_batches); for (gpc = 0; gpc < gr->gpc_nr; gpc++) { - for (ppc = 0; ppc < gr->ppc_nr[gpc]; ppc++, n++) { + for (ppc = 0; ppc < gr->func->ppc_nr; ppc++, n++) { const u32 as = alpha * gr->ppc_tpc_nr[gpc][ppc]; const u32 bs = attrib * gr->ppc_tpc_nr[gpc][ppc]; const u32 u = 0x418ea0 + (n * 0x04); const u32 o = PPC_UNIT(gpc, ppc, 0); + if (!(gr->ppc_mask[gpc] & (1 << ppc))) continue; - mmio_wr32(info, o + 0xc0, bs); - mmio_wr32(info, o + 0xf4, bo); + + gf100_grctx_patch_wr32(chan, o + 0xc0, bs); + gf100_grctx_patch_wr32(chan, o + 0xf4, bo); bo += grctx->attrib_nr_max * gr->ppc_tpc_nr[gpc][ppc]; - mmio_wr32(info, o + 0xe4, as); - mmio_wr32(info, o + 0xf8, ao); + gf100_grctx_patch_wr32(chan, o + 0xe4, as); + gf100_grctx_patch_wr32(chan, o + 0xf8, ao); ao += grctx->alpha_nr_max * gr->ppc_tpc_nr[gpc][ppc]; - mmio_wr32(info, u, ((bs / 3) << 16) | bs); + gf100_grctx_patch_wr32(chan, u, ((bs / 3) << 16) | bs); } } } +void +gm107_grctx_generate_attrib_cb(struct gf100_gr_chan *chan, u64 addr, u32 size) +{ + gf100_grctx_generate_attrib_cb(chan, addr, size); + + gf100_grctx_patch_wr32(chan, 0x419c2c, 0x10000000 | addr >> 12); +} + static void gm107_grctx_generate_r406500(struct gf100_gr *gr) { @@ -978,6 +973,8 @@ gm107_grctx = { .bundle_token_limit = 0x2c0, .pagepool = gm107_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gm107_grctx_generate_attrib_cb, .attrib = gm107_grctx_generate_attrib, .attrib_nr_max = 0xff0, .attrib_nr = 0xaa0, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm200.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm200.c index 013d05a0f0f6..175da8ac656c 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm200.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm200.c @@ -87,7 +87,7 @@ gm200_grctx_generate_dist_skip_table(struct gf100_gr *gr) int gpc, ppc, i; for (gpc = 0; gpc < gr->gpc_nr; gpc++) { - for (ppc = 0; ppc < gr->ppc_nr[gpc]; ppc++) { + for (ppc = 0; ppc < gr->func->ppc_nr; ppc++) { u8 ppc_tpcs = gr->ppc_tpc_nr[gpc][ppc]; u8 ppc_tpcm = gr->ppc_tpc_mask[gpc][ppc]; while (ppc_tpcs-- > gr->ppc_tpc_min) @@ -111,6 +111,8 @@ gm200_grctx = { .bundle_token_limit = 0x780, .pagepool = gm107_grctx_generate_pagepool, .pagepool_size = 0x20000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gm107_grctx_generate_attrib_cb, .attrib = gm107_grctx_generate_attrib, .attrib_nr_max = 0x600, .attrib_nr = 0x400, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm20b.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm20b.c index 6b92f8aa18a3..b8edccfada58 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm20b.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm20b.c @@ -22,8 +22,9 @@ #include "ctxgf100.h" static void -gm20b_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info) +gm20b_grctx_generate_main(struct gf100_gr_chan *chan) { + struct gf100_gr *gr = chan->gr; struct nvkm_device *device = gr->base.engine.subdev.device; const struct gf100_grctx_func *grctx = gr->func->grctx; u32 idle_timeout; @@ -35,7 +36,8 @@ gm20b_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info) idle_timeout = nvkm_mask(device, 0x404154, 0xffffffff, 0x00000000); - grctx->attrib(info); + grctx->attrib_cb(chan, chan->attrib_cb->addr, grctx->attrib_cb_size(gr)); + grctx->attrib(chan); grctx->unkn(gr); @@ -63,8 +65,8 @@ gm20b_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info) gf100_gr_wait_idle(gr); gf100_gr_icmd(gr, gr->bundle); - grctx->pagepool(info); - grctx->bundle(info); + grctx->pagepool(chan, chan->pagepool->addr); + grctx->bundle(chan, chan->bundle_cb->addr, grctx->bundle_size); } const struct gf100_grctx_func @@ -77,6 +79,8 @@ gm20b_grctx = { .bundle_token_limit = 0x1c0, .pagepool = gm107_grctx_generate_pagepool, .pagepool_size = 0x8000, + .attrib_cb_size = gf100_grctx_generate_attrib_cb_size, + .attrib_cb = gm107_grctx_generate_attrib_cb, .attrib = gm107_grctx_generate_attrib, .attrib_nr_max = 0x600, .attrib_nr = 0x400, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp100.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp100.c index 0b3326262e12..8485aaeae7a9 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp100.c @@ -30,66 +30,76 @@ ******************************************************************************/ void -gp100_grctx_generate_pagepool(struct gf100_grctx *info) +gp100_grctx_generate_pagepool(struct gf100_gr_chan *chan, u64 addr) { - const struct gf100_grctx_func *grctx = info->gr->func->grctx; - const int s = 8; - const int b = mmio_vram(info, grctx->pagepool_size, (1 << s), true); - mmio_refn(info, 0x40800c, 0x00000000, s, b); - mmio_wr32(info, 0x408010, 0x8007d800); - mmio_refn(info, 0x419004, 0x00000000, s, b); - mmio_wr32(info, 0x419008, 0x00000000); + gf100_grctx_patch_wr32(chan, 0x40800c, addr >> 8); + gf100_grctx_patch_wr32(chan, 0x408010, 0x8007d800); + gf100_grctx_patch_wr32(chan, 0x419004, addr >> 8); + gf100_grctx_patch_wr32(chan, 0x419008, 0x00000000); } static void -gp100_grctx_generate_attrib(struct gf100_grctx *info) +gp100_grctx_generate_attrib(struct gf100_gr_chan *chan) { - struct gf100_gr *gr = info->gr; + struct gf100_gr *gr = chan->gr; const struct gf100_grctx_func *grctx = gr->func->grctx; const u32 alpha = grctx->alpha_nr; const u32 attrib = grctx->attrib_nr; - const int s = 12; const int max_batches = 0xffff; u32 size = grctx->alpha_nr_max * gr->tpc_total; u32 ao = 0; u32 bo = ao + size; - int gpc, ppc, b, n = 0; + int gpc, ppc, n = 0; - for (gpc = 0; gpc < gr->gpc_nr; gpc++) - size += grctx->attrib_nr_max * gr->ppc_nr[gpc] * gr->ppc_tpc_max; - size = ((size * 0x20) + 128) & ~127; - b = mmio_vram(info, size, (1 << s), false); - - mmio_refn(info, 0x418810, 0x80000000, s, b); - mmio_refn(info, 0x419848, 0x10000000, s, b); - mmio_refn(info, 0x419c2c, 0x10000000, s, b); - mmio_refn(info, 0x419b00, 0x00000000, s, b); - mmio_wr32(info, 0x419b04, 0x80000000 | size >> 7); - mmio_wr32(info, 0x405830, attrib); - mmio_wr32(info, 0x40585c, alpha); - mmio_wr32(info, 0x4064c4, ((alpha / 4) << 16) | max_batches); + gf100_grctx_patch_wr32(chan, 0x405830, attrib); + gf100_grctx_patch_wr32(chan, 0x40585c, alpha); + gf100_grctx_patch_wr32(chan, 0x4064c4, ((alpha / 4) << 16) | max_batches); for (gpc = 0; gpc < gr->gpc_nr; gpc++) { - for (ppc = 0; ppc < gr->ppc_nr[gpc]; ppc++, n++) { + for (ppc = 0; ppc < gr->func->ppc_nr; ppc++, n++) { const u32 as = alpha * gr->ppc_tpc_nr[gpc][ppc]; const u32 bs = attrib * gr->ppc_tpc_max; const u32 u = 0x418ea0 + (n * 0x04); const u32 o = PPC_UNIT(gpc, ppc, 0); + if (!(gr->ppc_mask[gpc] & (1 << ppc))) continue; - mmio_wr32(info, o + 0xc0, bs); - mmio_wr32(info, o + 0xf4, bo); - mmio_wr32(info, o + 0xf0, bs); + + gf100_grctx_patch_wr32(chan, o + 0xc0, bs); + gf100_grctx_patch_wr32(chan, o + 0xf4, bo); + gf100_grctx_patch_wr32(chan, o + 0xf0, bs); bo += grctx->attrib_nr_max * gr->ppc_tpc_max; - mmio_wr32(info, o + 0xe4, as); - mmio_wr32(info, o + 0xf8, ao); + gf100_grctx_patch_wr32(chan, o + 0xe4, as); + gf100_grctx_patch_wr32(chan, o + 0xf8, ao); ao += grctx->alpha_nr_max * gr->ppc_tpc_nr[gpc][ppc]; - mmio_wr32(info, u, bs); + gf100_grctx_patch_wr32(chan, u, bs); } } - mmio_wr32(info, 0x418eec, 0x00000000); - mmio_wr32(info, 0x41befc, 0x00000000); + gf100_grctx_patch_wr32(chan, 0x418eec, 0x00000000); + gf100_grctx_patch_wr32(chan, 0x41befc, 0x00000000); +} + +void +gp100_grctx_generate_attrib_cb(struct gf100_gr_chan *chan, u64 addr, u32 size) +{ + gm107_grctx_generate_attrib_cb(chan, addr, size); + + gf100_grctx_patch_wr32(chan, 0x419b00, 0x00000000 | addr >> 12); + gf100_grctx_patch_wr32(chan, 0x419b04, 0x80000000 | size >> 7); +} + +static u32 +gp100_grctx_generate_attrib_cb_size(struct gf100_gr *gr) +{ + const struct gf100_grctx_func *grctx = gr->func->grctx; + u32 size = grctx->alpha_nr_max * gr->tpc_total; + int gpc; + + for (gpc = 0; gpc < gr->gpc_nr; gpc++) + size += grctx->attrib_nr_max * gr->func->ppc_nr * gr->ppc_tpc_max; + + return ((size * 0x20) + 128) & ~127; } void @@ -123,6 +133,8 @@ gp100_grctx = { .bundle_token_limit = 0x1080, .pagepool = gp100_grctx_generate_pagepool, .pagepool_size = 0x20000, + .attrib_cb_size = gp100_grctx_generate_attrib_cb_size, + .attrib_cb = gp100_grctx_generate_attrib_cb, .attrib = gp100_grctx_generate_attrib, .attrib_nr_max = 0x660, .attrib_nr = 0x440, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp102.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp102.c index daee17bf7d0d..7537979a5492 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp102.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp102.c @@ -37,58 +37,62 @@ gp102_grctx_generate_r408840(struct gf100_gr *gr) } void -gp102_grctx_generate_attrib(struct gf100_grctx *info) +gp102_grctx_generate_attrib(struct gf100_gr_chan *chan) { - struct gf100_gr *gr = info->gr; + struct gf100_gr *gr = chan->gr; const struct gf100_grctx_func *grctx = gr->func->grctx; const u32 alpha = grctx->alpha_nr; const u32 attrib = grctx->attrib_nr; const u32 gfxp = grctx->gfxp_nr; - const int s = 12; const int max_batches = 0xffff; u32 size = grctx->alpha_nr_max * gr->tpc_total; u32 ao = 0; u32 bo = ao + size; - int gpc, ppc, b, n = 0; + int gpc, ppc, n = 0; - for (gpc = 0; gpc < gr->gpc_nr; gpc++) - size += grctx->gfxp_nr * gr->ppc_nr[gpc] * gr->ppc_tpc_max; - size = ((size * 0x20) + 128) & ~127; - b = mmio_vram(info, size, (1 << s), false); - - mmio_refn(info, 0x418810, 0x80000000, s, b); - mmio_refn(info, 0x419848, 0x10000000, s, b); - mmio_refn(info, 0x419c2c, 0x10000000, s, b); - mmio_refn(info, 0x419b00, 0x00000000, s, b); - mmio_wr32(info, 0x419b04, 0x80000000 | size >> 7); - mmio_wr32(info, 0x405830, attrib); - mmio_wr32(info, 0x40585c, alpha); - mmio_wr32(info, 0x4064c4, ((alpha / 4) << 16) | max_batches); + gf100_grctx_patch_wr32(chan, 0x405830, attrib); + gf100_grctx_patch_wr32(chan, 0x40585c, alpha); + gf100_grctx_patch_wr32(chan, 0x4064c4, ((alpha / 4) << 16) | max_batches); for (gpc = 0; gpc < gr->gpc_nr; gpc++) { - for (ppc = 0; ppc < gr->ppc_nr[gpc]; ppc++, n++) { + for (ppc = 0; ppc < gr->func->ppc_nr; ppc++, n++) { const u32 as = alpha * gr->ppc_tpc_nr[gpc][ppc]; const u32 bs = attrib * gr->ppc_tpc_max; const u32 gs = gfxp * gr->ppc_tpc_max; const u32 u = 0x418ea0 + (n * 0x04); const u32 o = PPC_UNIT(gpc, ppc, 0); const u32 p = GPC_UNIT(gpc, 0xc44 + (ppc * 4)); + if (!(gr->ppc_mask[gpc] & (1 << ppc))) continue; - mmio_wr32(info, o + 0xc0, gs); - mmio_wr32(info, p, bs); - mmio_wr32(info, o + 0xf4, bo); - mmio_wr32(info, o + 0xf0, bs); + + gf100_grctx_patch_wr32(chan, o + 0xc0, gs); + gf100_grctx_patch_wr32(chan, p, bs); + gf100_grctx_patch_wr32(chan, o + 0xf4, bo); + gf100_grctx_patch_wr32(chan, o + 0xf0, bs); bo += gs; - mmio_wr32(info, o + 0xe4, as); - mmio_wr32(info, o + 0xf8, ao); + gf100_grctx_patch_wr32(chan, o + 0xe4, as); + gf100_grctx_patch_wr32(chan, o + 0xf8, ao); ao += grctx->alpha_nr_max * gr->ppc_tpc_nr[gpc][ppc]; - mmio_wr32(info, u, bs); + gf100_grctx_patch_wr32(chan, u, bs); } } - mmio_wr32(info, 0x4181e4, 0x00000100); - mmio_wr32(info, 0x41befc, 0x00000100); + gf100_grctx_patch_wr32(chan, 0x4181e4, 0x00000100); + gf100_grctx_patch_wr32(chan, 0x41befc, 0x00000100); +} + +u32 +gp102_grctx_generate_attrib_cb_size(struct gf100_gr *gr) +{ + const struct gf100_grctx_func *grctx = gr->func->grctx; + u32 size = grctx->alpha_nr_max * gr->tpc_total; + int gpc; + + for (gpc = 0; gpc < gr->gpc_nr; gpc++) + size += grctx->gfxp_nr * gr->func->ppc_nr * gr->ppc_tpc_max; + + return ((size * 0x20) + 127) & ~127; } const struct gf100_grctx_func @@ -101,6 +105,8 @@ gp102_grctx = { .bundle_token_limit = 0x900, .pagepool = gp100_grctx_generate_pagepool, .pagepool_size = 0x20000, + .attrib_cb_size = gp102_grctx_generate_attrib_cb_size, + .attrib_cb = gp100_grctx_generate_attrib_cb, .attrib = gp102_grctx_generate_attrib, .attrib_nr_max = 0x4b0, .attrib_nr = 0x320, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp104.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp104.c index 3b85e3d326b2..90b5f793e567 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp104.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp104.c @@ -31,6 +31,8 @@ gp104_grctx = { .bundle_token_limit = 0x900, .pagepool = gp100_grctx_generate_pagepool, .pagepool_size = 0x20000, + .attrib_cb_size = gp102_grctx_generate_attrib_cb_size, + .attrib_cb = gp100_grctx_generate_attrib_cb, .attrib = gp102_grctx_generate_attrib, .attrib_nr_max = 0x4b0, .attrib_nr = 0x320, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp107.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp107.c index 5060c5ee5ce0..d191761a0471 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp107.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgp107.c @@ -39,6 +39,8 @@ gp107_grctx = { .bundle_token_limit = 0x300, .pagepool = gp100_grctx_generate_pagepool, .pagepool_size = 0x20000, + .attrib_cb_size = gp102_grctx_generate_attrib_cb_size, + .attrib_cb = gp100_grctx_generate_attrib_cb, .attrib = gp102_grctx_generate_attrib, .attrib_nr_max = 0x15de, .attrib_nr = 0x540, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgv100.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgv100.c index 39553d55d3f3..957ea9d6bad4 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgv100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgv100.c @@ -25,7 +25,7 @@ * PGRAPH context implementation ******************************************************************************/ -const struct gf100_gr_init +static const struct gf100_gr_init gv100_grctx_init_sw_veid_bundle_init_0[] = { { 0x00001000, 64, 0x00100000, 0x00000008 }, { 0x00000941, 64, 0x00100000, 0x00000000 }, @@ -59,67 +59,70 @@ gv100_grctx_pack_sw_veid_bundle_init[] = { }; void -gv100_grctx_generate_attrib(struct gf100_grctx *info) +gv100_grctx_generate_attrib(struct gf100_gr_chan *chan) { - struct gf100_gr *gr = info->gr; + struct gf100_gr *gr = chan->gr; const struct gf100_grctx_func *grctx = gr->func->grctx; const u32 alpha = grctx->alpha_nr; const u32 attrib = grctx->attrib_nr; const u32 gfxp = grctx->gfxp_nr; - const int s = 12; + const int max_batches = 0xffff; u32 size = grctx->alpha_nr_max * gr->tpc_total; u32 ao = 0; u32 bo = ao + size; - int gpc, ppc, b, n = 0; + int gpc, ppc, n = 0; - for (gpc = 0; gpc < gr->gpc_nr; gpc++) - size += grctx->gfxp_nr * gr->ppc_nr[gpc] * gr->ppc_tpc_max; - size = ((size * 0x20) + 127) & ~127; - b = mmio_vram(info, size, (1 << s), false); - - mmio_refn(info, 0x418810, 0x80000000, s, b); - mmio_refn(info, 0x419848, 0x10000000, s, b); - mmio_refn(info, 0x419c2c, 0x10000000, s, b); - mmio_refn(info, 0x419e00, 0x00000000, s, b); - mmio_wr32(info, 0x419e04, 0x80000000 | size >> 7); - mmio_wr32(info, 0x405830, attrib); - mmio_wr32(info, 0x40585c, alpha); + gf100_grctx_patch_wr32(chan, 0x405830, attrib); + gf100_grctx_patch_wr32(chan, 0x40585c, alpha); + gf100_grctx_patch_wr32(chan, 0x4064c4, ((alpha / 4) << 16) | max_batches); for (gpc = 0; gpc < gr->gpc_nr; gpc++) { - for (ppc = 0; ppc < gr->ppc_nr[gpc]; ppc++, n++) { + for (ppc = 0; ppc < gr->func->ppc_nr; ppc++, n++) { const u32 as = alpha * gr->ppc_tpc_nr[gpc][ppc]; const u32 bs = attrib * gr->ppc_tpc_max; const u32 gs = gfxp * gr->ppc_tpc_max; const u32 u = 0x418ea0 + (n * 0x04); const u32 o = PPC_UNIT(gpc, ppc, 0); + if (!(gr->ppc_mask[gpc] & (1 << ppc))) continue; - mmio_wr32(info, o + 0xc0, gs); - mmio_wr32(info, o + 0xf4, bo); - mmio_wr32(info, o + 0xf0, bs); + + gf100_grctx_patch_wr32(chan, o + 0xc0, gs); + gf100_grctx_patch_wr32(chan, o + 0xf4, bo); + gf100_grctx_patch_wr32(chan, o + 0xf0, bs); bo += gs; - mmio_wr32(info, o + 0xe4, as); - mmio_wr32(info, o + 0xf8, ao); + gf100_grctx_patch_wr32(chan, o + 0xe4, as); + gf100_grctx_patch_wr32(chan, o + 0xf8, ao); ao += grctx->alpha_nr_max * gr->ppc_tpc_nr[gpc][ppc]; - mmio_wr32(info, u, bs); + gf100_grctx_patch_wr32(chan, u, bs); } } - mmio_wr32(info, 0x4181e4, 0x00000100); - mmio_wr32(info, 0x41befc, 0x00000100); + gf100_grctx_patch_wr32(chan, 0x4181e4, 0x00000100); + gf100_grctx_patch_wr32(chan, 0x41befc, 0x00000100); +} + +void +gv100_grctx_generate_attrib_cb(struct gf100_gr_chan *chan, u64 addr, u32 size) +{ + gm107_grctx_generate_attrib_cb(chan, addr, size); + + gf100_grctx_patch_wr32(chan, 0x419e00, 0x00000000 | addr >> 12); + gf100_grctx_patch_wr32(chan, 0x419e04, 0x80000000 | size >> 7); } void gv100_grctx_generate_rop_mapping(struct gf100_gr *gr) { struct nvkm_device *device = gr->base.engine.subdev.device; + const u32 mapregs = DIV_ROUND_UP(gr->func->gpc_nr * gr->func->tpc_nr, 6); u32 data; int i, j; /* Pack tile map into register format. */ nvkm_wr32(device, 0x418bb8, (gr->tpc_total << 8) | gr->screen_tile_row_offset); - for (i = 0; i < 11; i++) { + for (i = 0; i < mapregs; i++) { for (data = 0, j = 0; j < 6; j++) data |= (gr->tile[i * 6 + j] & 0x1f) << (j * 5); nvkm_wr32(device, 0x418b08 + (i * 4), data); @@ -157,6 +160,9 @@ static void gv100_grctx_generate_sm_id(struct gf100_gr *gr, int gpc, int tpc, int sm) { struct nvkm_device *device = gr->base.engine.subdev.device; + + tpc = gv100_gr_nonpes_aware_tpc(gr, gpc, tpc); + nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x608), sm); nvkm_wr32(device, GPC_UNIT(gpc, 0x0c10 + tpc * 4), sm); nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x088), sm); @@ -198,6 +204,8 @@ gv100_grctx = { .bundle_token_limit = 0x1680, .pagepool = gp100_grctx_generate_pagepool, .pagepool_size = 0x20000, + .attrib_cb_size = gp102_grctx_generate_attrib_cb_size, + .attrib_cb = gv100_grctx_generate_attrib_cb, .attrib = gv100_grctx_generate_attrib, .attrib_nr_max = 0x6c0, .attrib_nr = 0x480, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxtu102.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxtu102.c index 2299ca07d04a..542ab0c78be6 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxtu102.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxtu102.c @@ -34,6 +34,9 @@ static void tu102_grctx_generate_sm_id(struct gf100_gr *gr, int gpc, int tpc, int sm) { struct nvkm_device *device = gr->base.engine.subdev.device; + + tpc = gv100_gr_nonpes_aware_tpc(gr, gpc, tpc); + nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x608), sm); nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x088), sm); } @@ -47,42 +50,38 @@ tu102_grctx_init_unknown_bundle_init_0[] = { }; static const struct gf100_gr_pack -tu102_grctx_pack_sw_veid_bundle_init[] = { - { gv100_grctx_init_sw_veid_bundle_init_0 }, - { tu102_grctx_init_unknown_bundle_init_0 }, +tu102_grctx_pack_sw_bundle64_init[] = { + { tu102_grctx_init_unknown_bundle_init_0, .type = 64 }, {} }; -static void -tu102_grctx_generate_attrib(struct gf100_grctx *info) +void +tu102_grctx_generate_unknown(struct gf100_gr_chan *chan, u64 addr, u32 size) { - const u64 size = 0x80000; /*XXX: educated guess */ - const int s = 8; - const int b = mmio_vram(info, size, (1 << s), true); - - gv100_grctx_generate_attrib(info); - - mmio_refn(info, 0x408070, 0x00000000, s, b); - mmio_wr32(info, 0x408074, size >> s); /*XXX: guess */ - mmio_refn(info, 0x419034, 0x00000000, s, b); - mmio_wr32(info, 0x408078, 0x00000000); + gf100_grctx_patch_wr32(chan, 0x408070, addr >> 8); + gf100_grctx_patch_wr32(chan, 0x408074, size >> 8); /*XXX: guess */ + gf100_grctx_patch_wr32(chan, 0x419034, addr >> 8); + gf100_grctx_patch_wr32(chan, 0x408078, 0x00000000); } const struct gf100_grctx_func tu102_grctx = { - .unkn88c = gv100_grctx_unkn88c, .main = gf100_grctx_generate_main, .unkn = gv100_grctx_generate_unkn, - .sw_veid_bundle_init = tu102_grctx_pack_sw_veid_bundle_init, + .sw_bundle64_init = tu102_grctx_pack_sw_bundle64_init, .bundle = gm107_grctx_generate_bundle, .bundle_size = 0x3000, .bundle_min_gpm_fifo_depth = 0x180, .bundle_token_limit = 0xa80, .pagepool = gp100_grctx_generate_pagepool, .pagepool_size = 0x20000, - .attrib = tu102_grctx_generate_attrib, + .attrib_cb_size = gp102_grctx_generate_attrib_cb_size, + .attrib_cb = gv100_grctx_generate_attrib_cb, + .attrib = gv100_grctx_generate_attrib, .attrib_nr_max = 0x800, .attrib_nr = 0x700, + .unknown_size = 0x80000, + .unknown = tu102_grctx_generate_unknown, .alpha_nr_max = 0xc00, .alpha_nr = 0x800, .gfxp_nr = 0xfa8, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ga102.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ga102.c new file mode 100644 index 000000000000..a5b5ac2755a2 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ga102.c @@ -0,0 +1,347 @@ +/* + * Copyright 2019 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "gf100.h" +#include "ctxgf100.h" + +#include <core/firmware.h> +#include <subdev/acr.h> +#include <subdev/timer.h> +#include <subdev/vfn.h> + +#include <nvfw/flcn.h> + +#include <nvif/class.h> + +static void +ga102_gr_zbc_clear_color(struct gf100_gr *gr, int zbc) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + u32 invalid[] = { 0, 0, 0, 0 }, *color; + + if (gr->zbc_color[zbc].format) + color = gr->zbc_color[zbc].l2; + else + color = invalid; + + nvkm_mask(device, 0x41bcb4, 0x0000001f, zbc); + nvkm_wr32(device, 0x41bcec, color[0]); + nvkm_wr32(device, 0x41bcf0, color[1]); + nvkm_wr32(device, 0x41bcf4, color[2]); + nvkm_wr32(device, 0x41bcf8, color[3]); +} + +static const struct gf100_gr_func_zbc +ga102_gr_zbc = { + .clear_color = ga102_gr_zbc_clear_color, + .clear_depth = gp100_gr_zbc_clear_depth, + .stencil_get = gp102_gr_zbc_stencil_get, + .clear_stencil = gp102_gr_zbc_clear_stencil, +}; + +static void +ga102_gr_gpccs_reset(struct gf100_gr *gr) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + + nvkm_wr32(device, 0x41a610, 0x00000000); + nvkm_msec(device, 1, NVKM_DELAY); + nvkm_wr32(device, 0x41a610, 0x00000001); +} + +static const struct nvkm_acr_lsf_func +ga102_gr_gpccs_acr = { + .flags = NVKM_ACR_LSF_FORCE_PRIV_LOAD, + .bl_entry = 0x3400, + .bld_size = sizeof(struct flcn_bl_dmem_desc_v2), + .bld_write = gp108_gr_acr_bld_write, + .bld_patch = gp108_gr_acr_bld_patch, +}; + +static void +ga102_gr_fecs_reset(struct gf100_gr *gr) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + + nvkm_wr32(device, 0x409614, 0x00000010); + nvkm_wr32(device, 0x41a614, 0x00000020); + nvkm_usec(device, 10, NVKM_DELAY); + nvkm_wr32(device, 0x409614, 0x00000110); + nvkm_wr32(device, 0x41a614, 0x00000a20); + nvkm_usec(device, 10, NVKM_DELAY); + nvkm_rd32(device, 0x409614); + nvkm_rd32(device, 0x41a614); +} + +static const struct nvkm_acr_lsf_func +ga102_gr_fecs_acr = { + .bl_entry = 0x7e00, + .bld_size = sizeof(struct flcn_bl_dmem_desc_v2), + .bld_write = gp108_gr_acr_bld_write, + .bld_patch = gp108_gr_acr_bld_patch, +}; + +static void +ga102_gr_init_rop_exceptions(struct gf100_gr *gr) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + + nvkm_wr32(device, 0x41bcbc, 0x40000000); + nvkm_wr32(device, 0x41bc38, 0x40000000); + nvkm_wr32(device, 0x41ac94, nvkm_rd32(device, 0x502c94)); +} + +static void +ga102_gr_init_40a790(struct gf100_gr *gr) +{ + nvkm_wr32(gr->base.engine.subdev.device, 0x40a790, 0xc0000000); +} + +static void +ga102_gr_init_gpc_mmu(struct gf100_gr *gr) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + + nvkm_wr32(device, 0x418880, nvkm_rd32(device, 0x100c80) & 0xf8001fff); + nvkm_wr32(device, 0x418894, 0x00000000); + + nvkm_wr32(device, 0x4188b4, nvkm_rd32(device, 0x100cc8)); + nvkm_wr32(device, 0x4188b8, nvkm_rd32(device, 0x100ccc)); + nvkm_wr32(device, 0x4188b0, nvkm_rd32(device, 0x100cc4)); +} + +static struct nvkm_intr * +ga102_gr_oneinit_intr(struct gf100_gr *gr, enum nvkm_intr_type *pvector) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + + *pvector = nvkm_rd32(device, 0x400154) & 0x00000fff; + return &device->vfn->intr; +} + +static const struct gf100_gr_func +ga102_gr = { + .oneinit_intr = ga102_gr_oneinit_intr, + .oneinit_tiles = gm200_gr_oneinit_tiles, + .oneinit_sm_id = gv100_gr_oneinit_sm_id, + .init = gf100_gr_init, + .init_419bd8 = gv100_gr_init_419bd8, + .init_gpc_mmu = ga102_gr_init_gpc_mmu, + .init_vsc_stream_master = gk104_gr_init_vsc_stream_master, + .init_zcull = tu102_gr_init_zcull, + .init_num_active_ltcs = gf100_gr_init_num_active_ltcs, + .init_swdx_pes_mask = gp102_gr_init_swdx_pes_mask, + .init_fs = tu102_gr_init_fs, + .init_fecs_exceptions = tu102_gr_init_fecs_exceptions, + .init_40a790 = ga102_gr_init_40a790, + .init_ds_hww_esr_2 = gm200_gr_init_ds_hww_esr_2, + .init_sked_hww_esr = gk104_gr_init_sked_hww_esr, + .init_ppc_exceptions = gk104_gr_init_ppc_exceptions, + .init_504430 = gv100_gr_init_504430, + .init_shader_exceptions = gv100_gr_init_shader_exceptions, + .init_rop_exceptions = ga102_gr_init_rop_exceptions, + .init_4188a4 = gv100_gr_init_4188a4, + .trap_mp = gv100_gr_trap_mp, + .fecs.reset = ga102_gr_fecs_reset, + .gpccs.reset = ga102_gr_gpccs_reset, + .rops = gm200_gr_rops, + .gpc_nr = 7, + .tpc_nr = 6, + .ppc_nr = 3, + .grctx = &ga102_grctx, + .zbc = &ga102_gr_zbc, + .sclass = { + { -1, -1, FERMI_TWOD_A }, + { -1, -1, KEPLER_INLINE_TO_MEMORY_B }, + { -1, -1, AMPERE_B, &gf100_fermi }, + { -1, -1, AMPERE_COMPUTE_B }, + {} + } +}; + +MODULE_FIRMWARE("nvidia/ga102/gr/fecs_bl.bin"); +MODULE_FIRMWARE("nvidia/ga102/gr/fecs_sig.bin"); +MODULE_FIRMWARE("nvidia/ga102/gr/gpccs_bl.bin"); +MODULE_FIRMWARE("nvidia/ga102/gr/gpccs_sig.bin"); +MODULE_FIRMWARE("nvidia/ga102/gr/NET_img.bin"); + +MODULE_FIRMWARE("nvidia/ga103/gr/fecs_bl.bin"); +MODULE_FIRMWARE("nvidia/ga103/gr/fecs_sig.bin"); +MODULE_FIRMWARE("nvidia/ga103/gr/gpccs_bl.bin"); +MODULE_FIRMWARE("nvidia/ga103/gr/gpccs_sig.bin"); +MODULE_FIRMWARE("nvidia/ga103/gr/NET_img.bin"); + +MODULE_FIRMWARE("nvidia/ga104/gr/fecs_bl.bin"); +MODULE_FIRMWARE("nvidia/ga104/gr/fecs_sig.bin"); +MODULE_FIRMWARE("nvidia/ga104/gr/gpccs_bl.bin"); +MODULE_FIRMWARE("nvidia/ga104/gr/gpccs_sig.bin"); +MODULE_FIRMWARE("nvidia/ga104/gr/NET_img.bin"); + +MODULE_FIRMWARE("nvidia/ga106/gr/fecs_bl.bin"); +MODULE_FIRMWARE("nvidia/ga106/gr/fecs_sig.bin"); +MODULE_FIRMWARE("nvidia/ga106/gr/gpccs_bl.bin"); +MODULE_FIRMWARE("nvidia/ga106/gr/gpccs_sig.bin"); +MODULE_FIRMWARE("nvidia/ga106/gr/NET_img.bin"); + +MODULE_FIRMWARE("nvidia/ga107/gr/fecs_bl.bin"); +MODULE_FIRMWARE("nvidia/ga107/gr/fecs_sig.bin"); +MODULE_FIRMWARE("nvidia/ga107/gr/gpccs_bl.bin"); +MODULE_FIRMWARE("nvidia/ga107/gr/gpccs_sig.bin"); +MODULE_FIRMWARE("nvidia/ga107/gr/NET_img.bin"); + +struct netlist_region { + u32 region_id; + u32 data_size; + u32 data_offset; +}; + +struct netlist_image_header { + u32 version; + u32 regions; +}; + +struct netlist_image { + struct netlist_image_header header; + struct netlist_region regions[]; +}; + +struct netlist_av64 { + u32 addr; + u32 data_hi; + u32 data_lo; +}; + +static int +ga102_gr_av64_to_init(struct nvkm_blob *blob, struct gf100_gr_pack **ppack) +{ + struct gf100_gr_init *init; + struct gf100_gr_pack *pack; + int nent; + int i; + + nent = (blob->size / sizeof(struct netlist_av64)); + pack = vzalloc((sizeof(*pack) * 2) + (sizeof(*init) * (nent + 1))); + if (!pack) + return -ENOMEM; + + init = (void *)(pack + 2); + pack[0].init = init; + pack[0].type = 64; + + for (i = 0; i < nent; i++) { + struct gf100_gr_init *ent = &init[i]; + struct netlist_av64 *av = &((struct netlist_av64 *)blob->data)[i]; + + ent->addr = av->addr; + ent->data = ((u64)av->data_hi << 32) | av->data_lo; + ent->count = 1; + ent->pitch = 1; + } + + *ppack = pack; + return 0; +} + +static int +ga102_gr_load(struct gf100_gr *gr, int ver, const struct gf100_gr_fwif *fwif) +{ + struct nvkm_subdev *subdev = &gr->base.engine.subdev; + const struct firmware *fw; + const struct netlist_image *net; + const struct netlist_region *fecs_inst = NULL; + const struct netlist_region *fecs_data = NULL; + const struct netlist_region *gpccs_inst = NULL; + const struct netlist_region *gpccs_data = NULL; + int ret, i; + + ret = nvkm_firmware_get(subdev, "gr/NET_img", 0, &fw); + if (ret) + return ret; + + net = (const void *)fw->data; + nvkm_debug(subdev, "netlist version %d, %d regions\n", + net->header.version, net->header.regions); + + for (i = 0; i < net->header.regions; i++) { + const struct netlist_region *reg = &net->regions[i]; + struct nvkm_blob blob = { + .data = (void *)fw->data + reg->data_offset, + .size = reg->data_size, + }; + + nvkm_debug(subdev, "\t%2d: %08x %08x\n", + reg->region_id, reg->data_offset, reg->data_size); + + switch (reg->region_id) { + case 0: fecs_data = reg; break; + case 1: fecs_inst = reg; break; + case 2: gpccs_data = reg; break; + case 3: gpccs_inst = reg; break; + case 4: gk20a_gr_av_to_init(&blob, &gr->bundle); break; + case 5: gk20a_gr_aiv_to_init(&blob, &gr->sw_ctx); break; + case 7: gk20a_gr_av_to_method(&blob, &gr->method); break; + case 28: tu102_gr_av_to_init_veid(&blob, &gr->bundle_veid); break; + case 34: ga102_gr_av64_to_init(&blob, &gr->bundle64); break; + case 48: gk20a_gr_av_to_init(&blob, &gr->sw_nonctx1); break; + case 49: gk20a_gr_av_to_init(&blob, &gr->sw_nonctx2); break; + case 50: gk20a_gr_av_to_init(&blob, &gr->sw_nonctx3); break; + case 51: gk20a_gr_av_to_init(&blob, &gr->sw_nonctx4); break; + default: + break; + } + } + + ret = nvkm_acr_lsfw_load_bl_sig_net(subdev, &gr->fecs.falcon, NVKM_ACR_LSF_FECS, + "gr/fecs_", ver, fwif->fecs, + fw->data + fecs_inst->data_offset, + fecs_inst->data_size, + fw->data + fecs_data->data_offset, + fecs_data->data_size); + if (ret) + return ret; + + ret = nvkm_acr_lsfw_load_bl_sig_net(subdev, &gr->gpccs.falcon, NVKM_ACR_LSF_GPCCS, + "gr/gpccs_", ver, fwif->gpccs, + fw->data + gpccs_inst->data_offset, + gpccs_inst->data_size, + fw->data + gpccs_data->data_offset, + gpccs_data->data_size); + if (ret) + return ret; + + gr->firmware = true; + + nvkm_firmware_put(fw); + return 0; +} + +static const struct gf100_gr_fwif +ga102_gr_fwif[] = { + { 0, ga102_gr_load, &ga102_gr, &ga102_gr_fecs_acr, &ga102_gr_gpccs_acr }, + { -1, gm200_gr_nofw }, + {} +}; + +int +ga102_gr_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_gr **pgr) +{ + return gf100_gr_new_(ga102_gr_fwif, device, type, inst, pgr); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c index f16eabf4f642..5f20079c3660 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c @@ -67,7 +67,7 @@ gf100_gr_zbc_color_get(struct gf100_gr *gr, int format, struct nvkm_ltc *ltc = gr->base.engine.subdev.device->ltc; int zbc = -ENOSPC, i; - for (i = ltc->zbc_min; i <= ltc->zbc_max; i++) { + for (i = ltc->zbc_color_min; i <= ltc->zbc_color_max; i++) { if (gr->zbc_color[i].format) { if (gr->zbc_color[i].format != format) continue; @@ -114,7 +114,7 @@ gf100_gr_zbc_depth_get(struct gf100_gr *gr, int format, struct nvkm_ltc *ltc = gr->base.engine.subdev.device->ltc; int zbc = -ENOSPC, i; - for (i = ltc->zbc_min; i <= ltc->zbc_max; i++) { + for (i = ltc->zbc_depth_min; i <= ltc->zbc_depth_max; i++) { if (gr->zbc_depth[i].format) { if (gr->zbc_depth[i].format != format) continue; @@ -355,15 +355,14 @@ static void * gf100_gr_chan_dtor(struct nvkm_object *object) { struct gf100_gr_chan *chan = gf100_gr_chan(object); - int i; - - for (i = 0; i < ARRAY_SIZE(chan->data); i++) { - nvkm_vmm_put(chan->vmm, &chan->data[i].vma); - nvkm_memory_unref(&chan->data[i].mem); - } nvkm_vmm_put(chan->vmm, &chan->mmio_vma); nvkm_memory_unref(&chan->mmio); + + nvkm_vmm_put(chan->vmm, &chan->attrib_cb); + nvkm_vmm_put(chan->vmm, &chan->unknown); + nvkm_vmm_put(chan->vmm, &chan->bundle_cb); + nvkm_vmm_put(chan->vmm, &chan->pagepool); nvkm_vmm_unref(&chan->vmm); return chan; } @@ -380,12 +379,10 @@ gf100_gr_chan_new(struct nvkm_gr *base, struct nvkm_fifo_chan *fifoch, struct nvkm_object **pobject) { struct gf100_gr *gr = gf100_gr(base); - struct gf100_gr_data *data = gr->mmio_data; - struct gf100_gr_mmio *mmio = gr->mmio_list; struct gf100_gr_chan *chan; struct gf100_vmm_map_v0 args = { .priv = 1 }; struct nvkm_device *device = gr->base.engine.subdev.device; - int ret, i; + int ret; if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) return -ENOMEM; @@ -394,63 +391,91 @@ gf100_gr_chan_new(struct nvkm_gr *base, struct nvkm_fifo_chan *fifoch, chan->vmm = nvkm_vmm_ref(fifoch->vmm); *pobject = &chan->object; - /* allocate memory for a "mmio list" buffer that's used by the HUB - * fuc to modify some per-context register settings on first load - * of the context. - */ - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 0x1000, 0x100, - false, &chan->mmio); + /* Map pagepool. */ + ret = nvkm_vmm_get(chan->vmm, 12, nvkm_memory_size(gr->pagepool), &chan->pagepool); if (ret) return ret; - ret = nvkm_vmm_get(fifoch->vmm, 12, 0x1000, &chan->mmio_vma); + ret = nvkm_memory_map(gr->pagepool, 0, chan->vmm, chan->pagepool, &args, sizeof(args)); if (ret) return ret; - ret = nvkm_memory_map(chan->mmio, 0, fifoch->vmm, - chan->mmio_vma, &args, sizeof(args)); + /* Map bundle circular buffer. */ + ret = nvkm_vmm_get(chan->vmm, 12, nvkm_memory_size(gr->bundle_cb), &chan->bundle_cb); + if (ret) + return ret; + + ret = nvkm_memory_map(gr->bundle_cb, 0, chan->vmm, chan->bundle_cb, &args, sizeof(args)); + if (ret) + return ret; + + /* Map attribute circular buffer. */ + ret = nvkm_vmm_get(chan->vmm, 12, nvkm_memory_size(gr->attrib_cb), &chan->attrib_cb); if (ret) return ret; - /* allocate buffers referenced by mmio list */ - for (i = 0; data->size && i < ARRAY_SIZE(gr->mmio_data); i++) { - ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, - data->size, data->align, false, - &chan->data[i].mem); + if (device->card_type < GP100) { + ret = nvkm_memory_map(gr->attrib_cb, 0, chan->vmm, chan->attrib_cb, NULL, 0); if (ret) return ret; - - ret = nvkm_vmm_get(fifoch->vmm, 12, - nvkm_memory_size(chan->data[i].mem), - &chan->data[i].vma); + } else { + ret = nvkm_memory_map(gr->attrib_cb, 0, chan->vmm, chan->attrib_cb, + &args, sizeof(args));; if (ret) return ret; + } - args.priv = data->priv; + /* Map some context buffer of unknown purpose. */ + if (gr->func->grctx->unknown_size) { + ret = nvkm_vmm_get(chan->vmm, 12, nvkm_memory_size(gr->unknown), &chan->unknown); + if (ret) + return ret; - ret = nvkm_memory_map(chan->data[i].mem, 0, chan->vmm, - chan->data[i].vma, &args, sizeof(args)); + ret = nvkm_memory_map(gr->unknown, 0, chan->vmm, chan->unknown, + &args, sizeof(args)); if (ret) return ret; + } - data++; + /* Generate golden context image. */ + mutex_lock(&gr->fecs.mutex); + if (gr->data == NULL) { + ret = gf100_grctx_generate(gr, chan, fifoch->inst); + if (ret) { + nvkm_error(&base->engine.subdev, "failed to construct context\n"); + return ret; + } } + mutex_unlock(&gr->fecs.mutex); - /* finally, fill in the mmio list and point the context at it */ - nvkm_kmap(chan->mmio); - for (i = 0; mmio->addr && i < ARRAY_SIZE(gr->mmio_list); i++) { - u32 addr = mmio->addr; - u32 data = mmio->data; + /* allocate memory for a "mmio list" buffer that's used by the HUB + * fuc to modify some per-context register settings on first load + * of the context. + */ + ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 0x1000, 0x100, + false, &chan->mmio); + if (ret) + return ret; - if (mmio->buffer >= 0) { - u64 info = chan->data[mmio->buffer].vma->addr; - data |= info >> mmio->shift; - } + ret = nvkm_vmm_get(fifoch->vmm, 12, 0x1000, &chan->mmio_vma); + if (ret) + return ret; - nvkm_wo32(chan->mmio, chan->mmio_nr++ * 4, addr); - nvkm_wo32(chan->mmio, chan->mmio_nr++ * 4, data); - mmio++; - } + ret = nvkm_memory_map(chan->mmio, 0, fifoch->vmm, + chan->mmio_vma, &args, sizeof(args)); + if (ret) + return ret; + + /* finally, fill in the mmio list and point the context at it */ + nvkm_kmap(chan->mmio); + gr->func->grctx->pagepool(chan, chan->pagepool->addr); + gr->func->grctx->bundle(chan, chan->bundle_cb->addr, gr->func->grctx->bundle_size); + gr->func->grctx->attrib_cb(chan, chan->attrib_cb->addr, gr->func->grctx->attrib_cb_size(gr)); + gr->func->grctx->attrib(chan); + if (gr->func->grctx->patch_ltc) + gr->func->grctx->patch_ltc(chan); + if (gr->func->grctx->unknown_size) + gr->func->grctx->unknown(chan, chan->unknown->addr, gr->func->grctx->unknown_size); nvkm_done(chan->mmio); return 0; } @@ -727,7 +752,7 @@ gf100_gr_fecs_ctrl_ctxsw(struct gf100_gr *gr, u32 mthd) struct nvkm_device *device = gr->base.engine.subdev.device; nvkm_wr32(device, 0x409804, 0xffffffff); - nvkm_wr32(device, 0x409840, 0xffffffff); + nvkm_wr32(device, 0x409800, 0x00000000); nvkm_wr32(device, 0x409500, 0xffffffff); nvkm_wr32(device, 0x409504, mthd); nvkm_msec(device, 2000, @@ -771,12 +796,45 @@ gf100_gr_fecs_stop_ctxsw(struct nvkm_gr *base) return ret; } +static int +gf100_gr_fecs_halt_pipeline(struct gf100_gr *gr) +{ + int ret = 0; + + if (gr->firmware) { + mutex_lock(&gr->fecs.mutex); + ret = gf100_gr_fecs_ctrl_ctxsw(gr, 0x04); + mutex_unlock(&gr->fecs.mutex); + } + + return ret; +} + +int +gf100_gr_fecs_wfi_golden_save(struct gf100_gr *gr, u32 inst) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + + nvkm_mask(device, 0x409800, 0x00000003, 0x00000000); + nvkm_wr32(device, 0x409500, inst); + nvkm_wr32(device, 0x409504, 0x00000009); + nvkm_msec(device, 2000, + u32 stat = nvkm_rd32(device, 0x409800); + if (stat & 0x00000002) + return -EIO; + if (stat & 0x00000001) + return 0; + ); + + return -ETIMEDOUT; +} + int gf100_gr_fecs_bind_pointer(struct gf100_gr *gr, u32 inst) { struct nvkm_device *device = gr->base.engine.subdev.device; - nvkm_wr32(device, 0x409840, 0x00000030); + nvkm_mask(device, 0x409800, 0x00000030, 0x00000000); nvkm_wr32(device, 0x409500, inst); nvkm_wr32(device, 0x409504, 0x00000003); nvkm_msec(device, 2000, @@ -867,7 +925,7 @@ gf100_gr_fecs_discover_pm_image_size(struct gf100_gr *gr, u32 *psize) { struct nvkm_device *device = gr->base.engine.subdev.device; - nvkm_wr32(device, 0x409840, 0xffffffff); + nvkm_wr32(device, 0x409800, 0x00000000); nvkm_wr32(device, 0x409500, 0x00000000); nvkm_wr32(device, 0x409504, 0x00000025); nvkm_msec(device, 2000, @@ -883,7 +941,7 @@ gf100_gr_fecs_discover_zcull_image_size(struct gf100_gr *gr, u32 *psize) { struct nvkm_device *device = gr->base.engine.subdev.device; - nvkm_wr32(device, 0x409840, 0xffffffff); + nvkm_wr32(device, 0x409800, 0x00000000); nvkm_wr32(device, 0x409500, 0x00000000); nvkm_wr32(device, 0x409504, 0x00000016); nvkm_msec(device, 2000, @@ -899,7 +957,7 @@ gf100_gr_fecs_discover_image_size(struct gf100_gr *gr, u32 *psize) { struct nvkm_device *device = gr->base.engine.subdev.device; - nvkm_wr32(device, 0x409840, 0xffffffff); + nvkm_wr32(device, 0x409800, 0x00000000); nvkm_wr32(device, 0x409500, 0x00000000); nvkm_wr32(device, 0x409504, 0x00000010); nvkm_msec(device, 2000, @@ -915,7 +973,7 @@ gf100_gr_fecs_set_watchdog_timeout(struct gf100_gr *gr, u32 timeout) { struct nvkm_device *device = gr->base.engine.subdev.device; - nvkm_wr32(device, 0x409840, 0xffffffff); + nvkm_wr32(device, 0x409800, 0x00000000); nvkm_wr32(device, 0x409500, timeout); nvkm_wr32(device, 0x409504, 0x00000021); } @@ -955,7 +1013,7 @@ gf100_gr_zbc_init(struct gf100_gr *gr) const u32 f32_1[] = { 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000 }; struct nvkm_ltc *ltc = gr->base.engine.subdev.device->ltc; - int index, c = ltc->zbc_min, d = ltc->zbc_min, s = ltc->zbc_min; + int index, c = ltc->zbc_color_min, d = ltc->zbc_depth_min, s = ltc->zbc_depth_min; if (!gr->zbc_color[0].format) { gf100_gr_zbc_color_get(gr, 1, & zero[0], &zero[4]); c++; @@ -971,13 +1029,13 @@ gf100_gr_zbc_init(struct gf100_gr *gr) } } - for (index = c; index <= ltc->zbc_max; index++) + for (index = c; index <= ltc->zbc_color_max; index++) gr->func->zbc->clear_color(gr, index); - for (index = d; index <= ltc->zbc_max; index++) + for (index = d; index <= ltc->zbc_depth_max; index++) gr->func->zbc->clear_depth(gr, index); if (gr->func->zbc->clear_stencil) { - for (index = s; index <= ltc->zbc_max; index++) + for (index = s; index <= ltc->zbc_depth_max; index++) gr->func->zbc->clear_stencil(gr, index); } } @@ -1003,7 +1061,7 @@ gf100_gr_wait_idle(struct gf100_gr *gr) nvkm_rd32(device, 0x400700); gr_enabled = nvkm_rd32(device, 0x200) & 0x1000; - ctxsw_active = nvkm_rd32(device, 0x2640) & 0x8000; + ctxsw_active = nvkm_fifo_ctxsw_in_progress(&gr->base.engine); gr_busy = nvkm_rd32(device, 0x40060c) & 0x1; if (!gr_enabled || (!gr_busy && !ctxsw_active)) @@ -1039,7 +1097,7 @@ gf100_gr_icmd(struct gf100_gr *gr, const struct gf100_gr_pack *p) struct nvkm_device *device = gr->base.engine.subdev.device; const struct gf100_gr_pack *pack; const struct gf100_gr_init *init; - u32 data = 0; + u64 data = 0; nvkm_wr32(device, 0x400208, 0x80000000); @@ -1049,6 +1107,8 @@ gf100_gr_icmd(struct gf100_gr *gr, const struct gf100_gr_pack *p) if ((pack == p && init == p->init) || data != init->data) { nvkm_wr32(device, 0x400204, init->data); + if (pack->type == 64) + nvkm_wr32(device, 0x40020c, upper_32_bits(init->data)); data = init->data; } @@ -1542,13 +1602,13 @@ gf100_gr_ctxctl_isr(struct gf100_gr *gr) } } -static void -gf100_gr_intr(struct nvkm_gr *base) +static irqreturn_t +gf100_gr_intr(struct nvkm_inth *inth) { - struct gf100_gr *gr = gf100_gr(base); + struct gf100_gr *gr = container_of(inth, typeof(*gr), base.engine.subdev.inth); struct nvkm_subdev *subdev = &gr->base.engine.subdev; struct nvkm_device *device = subdev->device; - struct nvkm_fifo_chan *chan; + struct nvkm_chan *chan; unsigned long flags; u64 inst = nvkm_rd32(device, 0x409b00) & 0x0fffffff; u32 stat = nvkm_rd32(device, 0x400100); @@ -1561,10 +1621,10 @@ gf100_gr_intr(struct nvkm_gr *base) const char *name = "unknown"; int chid = -1; - chan = nvkm_fifo_chan_inst(device->fifo, (u64)inst << 12, &flags); + chan = nvkm_chan_get_inst(&gr->base.engine, (u64)inst << 12, &flags); if (chan) { - name = chan->object.client->name; - chid = chan->chid; + name = chan->name; + chid = chan->id; } if (device->card_type < NV_E0 || subc < 4) @@ -1631,7 +1691,8 @@ gf100_gr_intr(struct nvkm_gr *base) } nvkm_wr32(device, 0x400500, 0x00010001); - nvkm_fifo_chan_put(device->fifo, flags, &chan); + nvkm_chan_put(&chan, flags); + return IRQ_HANDLED; } static void @@ -1721,7 +1782,7 @@ gf100_gr_init_ctxctl_ext(struct gf100_gr *gr) nvkm_mc_unk260(device, 1); /* start both of them running */ - nvkm_wr32(device, 0x409840, 0xffffffff); + nvkm_wr32(device, 0x409800, 0x00000000); nvkm_wr32(device, 0x41a10c, 0x00000000); nvkm_wr32(device, 0x40910c, 0x00000000); @@ -1763,15 +1824,6 @@ gf100_gr_init_ctxctl_ext(struct gf100_gr *gr) return ret; } - /* Generate golden context image. */ - if (gr->data == NULL) { - int ret = gf100_grctx_generate(gr); - if (ret) { - nvkm_error(subdev, "failed to construct context\n"); - return ret; - } - } - return 0; } @@ -1823,14 +1875,6 @@ gf100_gr_init_ctxctl_int(struct gf100_gr *gr) } gr->size = nvkm_rd32(device, 0x409804); - if (gr->data == NULL) { - int ret = gf100_grctx_generate(gr); - if (ret) { - nvkm_error(subdev, "failed to construct context\n"); - return ret; - } - } - return 0; } @@ -1847,10 +1891,11 @@ gf100_gr_init_ctxctl(struct gf100_gr *gr) return ret; } -void +int gf100_gr_oneinit_sm_id(struct gf100_gr *gr) { int tpc, gpc; + for (tpc = 0; tpc < gr->tpc_max; tpc++) { for (gpc = 0; gpc < gr->gpc_nr; gpc++) { if (tpc < gr->tpc_nr[gpc]) { @@ -1860,6 +1905,8 @@ gf100_gr_oneinit_sm_id(struct gf100_gr *gr) } } } + + return 0; } void @@ -1944,7 +1991,17 @@ gf100_gr_oneinit(struct nvkm_gr *base) struct gf100_gr *gr = gf100_gr(base); struct nvkm_subdev *subdev = &gr->base.engine.subdev; struct nvkm_device *device = subdev->device; - int i, j; + struct nvkm_intr *intr = &device->mc->intr; + enum nvkm_intr_type intr_type = NVKM_INTR_SUBDEV; + int ret, i, j; + + if (gr->func->oneinit_intr) + intr = gr->func->oneinit_intr(gr, &intr_type); + + ret = nvkm_inth_add(intr, intr_type, NVKM_INTR_PRIO_NORMAL, &gr->base.engine.subdev, + gf100_gr_intr, &gr->base.engine.subdev.inth); + if (ret) + return ret; nvkm_pmu_pgob(device->pmu, false); @@ -1954,12 +2011,14 @@ gf100_gr_oneinit(struct nvkm_gr *base) gr->tpc_nr[i] = nvkm_rd32(device, GPC_UNIT(i, 0x2608)); gr->tpc_max = max(gr->tpc_max, gr->tpc_nr[i]); gr->tpc_total += gr->tpc_nr[i]; - gr->ppc_nr[i] = gr->func->ppc_nr; - for (j = 0; j < gr->ppc_nr[i]; j++) { + for (j = 0; j < gr->func->ppc_nr; j++) { gr->ppc_tpc_mask[i][j] = nvkm_rd32(device, GPC_UNIT(i, 0x0c30 + (j * 4))); if (gr->ppc_tpc_mask[i][j] == 0) continue; + + gr->ppc_nr[i]++; + gr->ppc_mask[i] |= (1 << j); gr->ppc_tpc_nr[i][j] = hweight8(gr->ppc_tpc_mask[i][j]); if (gr->ppc_tpc_min == 0 || @@ -1968,12 +2027,37 @@ gf100_gr_oneinit(struct nvkm_gr *base) if (gr->ppc_tpc_max < gr->ppc_tpc_nr[i][j]) gr->ppc_tpc_max = gr->ppc_tpc_nr[i][j]; } + + gr->ppc_total += gr->ppc_nr[i]; + } + + /* Allocate global context buffers. */ + ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, gr->func->grctx->pagepool_size, + 0x100, false, &gr->pagepool); + if (ret) + return ret; + + ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, gr->func->grctx->bundle_size, + 0x100, false, &gr->bundle_cb); + if (ret) + return ret; + + ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, gr->func->grctx->attrib_cb_size(gr), + 0x1000, false, &gr->attrib_cb); + if (ret) + return ret; + + if (gr->func->grctx->unknown_size) { + ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, gr->func->grctx->unknown_size, + 0x100, false, &gr->unknown); + if (ret) + return ret; } memset(gr->tile, 0xff, sizeof(gr->tile)); gr->func->oneinit_tiles(gr); - gr->func->oneinit_sm_id(gr); - return 0; + + return gr->func->oneinit_sm_id(gr); } static int @@ -1983,7 +2067,7 @@ gf100_gr_init_(struct nvkm_gr *base) struct nvkm_subdev *subdev = &base->engine.subdev; struct nvkm_device *device = subdev->device; bool reset = device->chipset == 0x137 || device->chipset == 0x138; - u32 ret; + int ret; /* On certain GP107/GP108 boards, we trigger a weird issue where * GR will stop responding to PRI accesses after we've asked the @@ -2019,7 +2103,12 @@ gf100_gr_init_(struct nvkm_gr *base) if (ret) return ret; - return gr->func->init(gr); + ret = gr->func->init(gr); + if (ret) + return ret; + + nvkm_inth_allow(&subdev->inth); + return 0; } static int @@ -2027,6 +2116,9 @@ gf100_gr_fini(struct nvkm_gr *base, bool suspend) { struct gf100_gr *gr = gf100_gr(base); struct nvkm_subdev *subdev = &gr->base.engine.subdev; + + nvkm_inth_block(&subdev->inth); + nvkm_falcon_put(&gr->gpccs.falcon, subdev); nvkm_falcon_put(&gr->fecs.falcon, subdev); return 0; @@ -2039,6 +2131,11 @@ gf100_gr_dtor(struct nvkm_gr *base) kfree(gr->data); + nvkm_memory_unref(&gr->unknown); + nvkm_memory_unref(&gr->attrib_cb); + nvkm_memory_unref(&gr->bundle_cb); + nvkm_memory_unref(&gr->pagepool); + nvkm_falcon_dtor(&gr->gpccs.falcon); nvkm_falcon_dtor(&gr->fecs.falcon); @@ -2047,81 +2144,27 @@ gf100_gr_dtor(struct nvkm_gr *base) nvkm_blob_dtor(&gr->gpccs.inst); nvkm_blob_dtor(&gr->gpccs.data); + vfree(gr->bundle64); + vfree(gr->bundle_veid); vfree(gr->bundle); vfree(gr->method); vfree(gr->sw_ctx); vfree(gr->sw_nonctx); + vfree(gr->sw_nonctx1); + vfree(gr->sw_nonctx2); + vfree(gr->sw_nonctx3); + vfree(gr->sw_nonctx4); return gr; } -static const struct nvkm_gr_func -gf100_gr_ = { - .dtor = gf100_gr_dtor, - .oneinit = gf100_gr_oneinit, - .init = gf100_gr_init_, - .fini = gf100_gr_fini, - .intr = gf100_gr_intr, - .units = gf100_gr_units, - .chan_new = gf100_gr_chan_new, - .object_get = gf100_gr_object_get, - .chsw_load = gf100_gr_chsw_load, - .ctxsw.pause = gf100_gr_fecs_stop_ctxsw, - .ctxsw.resume = gf100_gr_fecs_start_ctxsw, - .ctxsw.inst = gf100_gr_ctxsw_inst, -}; - static const struct nvkm_falcon_func gf100_gr_flcn = { - .fbif = 0x600, .load_imem = nvkm_falcon_v1_load_imem, .load_dmem = nvkm_falcon_v1_load_dmem, - .read_dmem = nvkm_falcon_v1_read_dmem, - .bind_context = nvkm_falcon_v1_bind_context, - .wait_for_halt = nvkm_falcon_v1_wait_for_halt, - .clear_interrupt = nvkm_falcon_v1_clear_interrupt, - .set_start_addr = nvkm_falcon_v1_set_start_addr, .start = nvkm_falcon_v1_start, - .enable = nvkm_falcon_v1_enable, - .disable = nvkm_falcon_v1_disable, }; -int -gf100_gr_new_(const struct gf100_gr_fwif *fwif, struct nvkm_device *device, - enum nvkm_subdev_type type, int inst, struct nvkm_gr **pgr) -{ - struct gf100_gr *gr; - int ret; - - if (!(gr = kzalloc(sizeof(*gr), GFP_KERNEL))) - return -ENOMEM; - *pgr = &gr->base; - - ret = nvkm_gr_ctor(&gf100_gr_, device, type, inst, true, &gr->base); - if (ret) - return ret; - - fwif = nvkm_firmware_load(&gr->base.engine.subdev, fwif, "Gr", gr); - if (IS_ERR(fwif)) - return PTR_ERR(fwif); - - gr->func = fwif->func; - - ret = nvkm_falcon_ctor(&gf100_gr_flcn, &gr->base.engine.subdev, - "fecs", 0x409000, &gr->fecs.falcon); - if (ret) - return ret; - - mutex_init(&gr->fecs.mutex); - - ret = nvkm_falcon_ctor(&gf100_gr_flcn, &gr->base.engine.subdev, - "gpccs", 0x41a000, &gr->gpccs.falcon); - if (ret) - return ret; - - return 0; -} - void gf100_gr_init_num_tpc_per_gpc(struct gf100_gr *gr, bool pd, bool ds) { @@ -2146,6 +2189,29 @@ gf100_gr_init_400054(struct gf100_gr *gr) } void +gf100_gr_init_exception2(struct gf100_gr *gr) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + + nvkm_wr32(device, 0x40011c, 0xffffffff); + nvkm_wr32(device, 0x400134, 0xffffffff); +} + +void +gf100_gr_init_rop_exceptions(struct gf100_gr *gr) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + int rop; + + for (rop = 0; rop < gr->rop_nr; rop++) { + nvkm_wr32(device, ROP_UNIT(rop, 0x144), 0x40000000); + nvkm_wr32(device, ROP_UNIT(rop, 0x070), 0x40000000); + nvkm_wr32(device, ROP_UNIT(rop, 0x204), 0xffffffff); + nvkm_wr32(device, ROP_UNIT(rop, 0x208), 0xffffffff); + } +} + +void gf100_gr_init_shader_exceptions(struct gf100_gr *gr, int gpc, int tpc) { struct nvkm_device *device = gr->base.engine.subdev.device; @@ -2252,21 +2318,47 @@ gf100_gr_init_vsc_stream_master(struct gf100_gr *gr) nvkm_mask(device, TPC_UNIT(0, 0, 0x05c), 0x00000001, 0x00000001); } +static int +gf100_gr_reset(struct nvkm_gr *base) +{ + struct nvkm_subdev *subdev = &base->engine.subdev; + struct nvkm_device *device = subdev->device; + struct gf100_gr *gr = gf100_gr(base); + + nvkm_mask(device, 0x400500, 0x00000001, 0x00000000); + + WARN_ON(gf100_gr_fecs_halt_pipeline(gr)); + + subdev->func->fini(subdev, false); + nvkm_mc_disable(device, subdev->type, subdev->inst); + if (gr->func->gpccs.reset) + gr->func->gpccs.reset(gr); + + nvkm_mc_enable(device, subdev->type, subdev->inst); + return subdev->func->init(subdev); +} + int gf100_gr_init(struct gf100_gr *gr) { struct nvkm_device *device = gr->base.engine.subdev.device; - int gpc, tpc, rop; + int gpc, tpc; - if (gr->func->init_419bd8) - gr->func->init_419bd8(gr); + nvkm_mask(device, 0x400500, 0x00010001, 0x00000000); gr->func->init_gpc_mmu(gr); - if (gr->sw_nonctx) + if (gr->sw_nonctx1) { + gf100_gr_mmio(gr, gr->sw_nonctx1); + gf100_gr_mmio(gr, gr->sw_nonctx2); + gf100_gr_mmio(gr, gr->sw_nonctx3); + gf100_gr_mmio(gr, gr->sw_nonctx4); + } else + if (gr->sw_nonctx) { gf100_gr_mmio(gr, gr->sw_nonctx); - else + } else { gf100_gr_mmio(gr, gr->func->mmio); + } gf100_gr_wait_idle(gr); @@ -2298,6 +2390,10 @@ gf100_gr_init(struct gf100_gr *gr) nvkm_wr32(device, 0x400124, 0x00000002); gr->func->init_fecs_exceptions(gr); + + if (gr->func->init_40a790) + gr->func->init_40a790(gr); + if (gr->func->init_ds_hww_esr_2) gr->func->init_ds_hww_esr_2(gr); @@ -2346,19 +2442,14 @@ gf100_gr_init(struct gf100_gr *gr) nvkm_wr32(device, GPC_UNIT(gpc, 0x2c94), 0xffffffff); } - for (rop = 0; rop < gr->rop_nr; rop++) { - nvkm_wr32(device, ROP_UNIT(rop, 0x144), 0x40000000); - nvkm_wr32(device, ROP_UNIT(rop, 0x070), 0x40000000); - nvkm_wr32(device, ROP_UNIT(rop, 0x204), 0xffffffff); - nvkm_wr32(device, ROP_UNIT(rop, 0x208), 0xffffffff); - } + gr->func->init_rop_exceptions(gr); nvkm_wr32(device, 0x400108, 0xffffffff); nvkm_wr32(device, 0x400138, 0xffffffff); nvkm_wr32(device, 0x400118, 0xffffffff); nvkm_wr32(device, 0x400130, 0xffffffff); - nvkm_wr32(device, 0x40011c, 0xffffffff); - nvkm_wr32(device, 0x400134, 0xffffffff); + if (gr->func->init_exception2) + gr->func->init_exception2(gr); if (gr->func->init_400054) gr->func->init_400054(gr); @@ -2371,6 +2462,18 @@ gf100_gr_init(struct gf100_gr *gr) return gf100_gr_init_ctxctl(gr); } +void +gf100_gr_fecs_reset(struct gf100_gr *gr) +{ + struct nvkm_device *device = gr->base.engine.subdev.device; + + nvkm_wr32(device, 0x409614, 0x00000070); + nvkm_usec(device, 10, NVKM_DELAY); + nvkm_mask(device, 0x409614, 0x00000700, 0x00000700); + nvkm_usec(device, 10, NVKM_DELAY); + nvkm_rd32(device, 0x409614); +} + #include "fuc/hubgf100.fuc3.h" struct gf100_gr_ucode @@ -2391,6 +2494,22 @@ gf100_gr_gpccs_ucode = { .data.size = sizeof(gf100_grgpc_data), }; +static const struct nvkm_gr_func +gf100_gr_ = { + .dtor = gf100_gr_dtor, + .oneinit = gf100_gr_oneinit, + .init = gf100_gr_init_, + .fini = gf100_gr_fini, + .reset = gf100_gr_reset, + .units = gf100_gr_units, + .chan_new = gf100_gr_chan_new, + .object_get = gf100_gr_object_get, + .chsw_load = gf100_gr_chsw_load, + .ctxsw.pause = gf100_gr_fecs_stop_ctxsw, + .ctxsw.resume = gf100_gr_fecs_start_ctxsw, + .ctxsw.inst = gf100_gr_ctxsw_inst, +}; + static const struct gf100_gr_func gf100_gr = { .oneinit_tiles = gf100_gr_oneinit_tiles, @@ -2406,10 +2525,13 @@ gf100_gr = { .init_419eb4 = gf100_gr_init_419eb4, .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_shader_exceptions = gf100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gf100_gr_init_400054, .trap_mp = gf100_gr_trap_mp, .mmio = gf100_gr_pack_mmio, .fecs.ucode = &gf100_gr_fecs_ucode, + .fecs.reset = gf100_gr_fecs_reset, .gpccs.ucode = &gf100_gr_gpccs_ucode, .rops = gf100_gr_rops, .grctx = &gf100_grctx, @@ -2483,6 +2605,42 @@ gf100_gr_fwif[] = { }; int +gf100_gr_new_(const struct gf100_gr_fwif *fwif, struct nvkm_device *device, + enum nvkm_subdev_type type, int inst, struct nvkm_gr **pgr) +{ + struct gf100_gr *gr; + int ret; + + if (!(gr = kzalloc(sizeof(*gr), GFP_KERNEL))) + return -ENOMEM; + *pgr = &gr->base; + + ret = nvkm_gr_ctor(&gf100_gr_, device, type, inst, true, &gr->base); + if (ret) + return ret; + + fwif = nvkm_firmware_load(&gr->base.engine.subdev, fwif, "Gr", gr); + if (IS_ERR(fwif)) + return PTR_ERR(fwif); + + gr->func = fwif->func; + + ret = nvkm_falcon_ctor(&gf100_gr_flcn, &gr->base.engine.subdev, + "fecs", 0x409000, &gr->fecs.falcon); + if (ret) + return ret; + + mutex_init(&gr->fecs.mutex); + + ret = nvkm_falcon_ctor(&gf100_gr_flcn, &gr->base.engine.subdev, + "gpccs", 0x41a000, &gr->gpccs.falcon); + if (ret) + return ret; + + return 0; +} + +int gf100_gr_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_gr **pgr) { return gf100_gr_new_(gf100_gr_fwif, device, type, inst, pgr); diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.h b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.h index c0038f906135..94ca7ac16acf 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.h @@ -44,19 +44,6 @@ struct nvkm_acr_lsfw; #define PPC_UNIT(t, m, r) (0x503000 + (t) * 0x8000 + (m) * 0x200 + (r)) #define TPC_UNIT(t, m, r) (0x504000 + (t) * 0x8000 + (m) * 0x800 + (r)) -struct gf100_gr_data { - u32 size; - u32 align; - bool priv; -}; - -struct gf100_gr_mmio { - u32 addr; - u32 data; - u32 shift; - int buffer; -}; - struct gf100_gr_zbc_color { u32 format; u32 ds[4]; @@ -101,13 +88,19 @@ struct gf100_gr { * using hardcoded arrays. To be allocated with vzalloc(). */ struct gf100_gr_pack *sw_nonctx; + struct gf100_gr_pack *sw_nonctx1; + struct gf100_gr_pack *sw_nonctx2; + struct gf100_gr_pack *sw_nonctx3; + struct gf100_gr_pack *sw_nonctx4; struct gf100_gr_pack *sw_ctx; struct gf100_gr_pack *bundle; + struct gf100_gr_pack *bundle_veid; + struct gf100_gr_pack *bundle64; struct gf100_gr_pack *method; - struct gf100_gr_zbc_color zbc_color[NVKM_LTC_MAX_ZBC_CNT]; - struct gf100_gr_zbc_depth zbc_depth[NVKM_LTC_MAX_ZBC_CNT]; - struct gf100_gr_zbc_stencil zbc_stencil[NVKM_LTC_MAX_ZBC_CNT]; + struct gf100_gr_zbc_color zbc_color[NVKM_LTC_MAX_ZBC_COLOR_CNT]; + struct gf100_gr_zbc_depth zbc_depth[NVKM_LTC_MAX_ZBC_DEPTH_CNT]; + struct gf100_gr_zbc_stencil zbc_stencil[NVKM_LTC_MAX_ZBC_DEPTH_CNT]; u8 rop_nr; u8 gpc_nr; @@ -120,6 +113,12 @@ struct gf100_gr { u8 ppc_tpc_nr[GPC_MAX][4]; u8 ppc_tpc_min; u8 ppc_tpc_max; + u8 ppc_total; + + struct nvkm_memory *pagepool; + struct nvkm_memory *bundle_cb; + struct nvkm_memory *attrib_cb; + struct nvkm_memory *unknown; u8 screen_tile_row_offset; u8 tile[TPC_MAX]; @@ -130,8 +129,6 @@ struct gf100_gr { } sm[TPC_MAX]; u8 sm_nr; - struct gf100_gr_data mmio_data[4]; - struct gf100_gr_mmio mmio_list[4096/8]; u32 size; u32 *data; u32 size_zcull; @@ -139,6 +136,7 @@ struct gf100_gr { }; int gf100_gr_fecs_bind_pointer(struct gf100_gr *, u32 inst); +int gf100_gr_fecs_wfi_golden_save(struct gf100_gr *, u32 inst); struct gf100_gr_func_zbc { void (*clear_color)(struct gf100_gr *, int zbc); @@ -149,8 +147,9 @@ struct gf100_gr_func_zbc { }; struct gf100_gr_func { + struct nvkm_intr *(*oneinit_intr)(struct gf100_gr *, enum nvkm_intr_type *); void (*oneinit_tiles)(struct gf100_gr *); - void (*oneinit_sm_id)(struct gf100_gr *); + int (*oneinit_sm_id)(struct gf100_gr *); int (*init)(struct gf100_gr *); void (*init_419bd8)(struct gf100_gr *); void (*init_gpc_mmu)(struct gf100_gr *); @@ -164,6 +163,7 @@ struct gf100_gr_func { void (*init_swdx_pes_mask)(struct gf100_gr *); void (*init_fs)(struct gf100_gr *); void (*init_fecs_exceptions)(struct gf100_gr *); + void (*init_40a790)(struct gf100_gr *); void (*init_ds_hww_esr_2)(struct gf100_gr *); void (*init_40601c)(struct gf100_gr *); void (*init_sked_hww_esr)(struct gf100_gr *); @@ -174,6 +174,8 @@ struct gf100_gr_func { void (*init_tex_hww_esr)(struct gf100_gr *, int gpc, int tpc); void (*init_504430)(struct gf100_gr *, int gpc, int tpc); void (*init_shader_exceptions)(struct gf100_gr *, int gpc, int tpc); + void (*init_rop_exceptions)(struct gf100_gr *); + void (*init_exception2)(struct gf100_gr *); void (*init_400054)(struct gf100_gr *); void (*init_4188a4)(struct gf100_gr *); void (*trap_mp)(struct gf100_gr *, int gpc, int tpc); @@ -181,9 +183,11 @@ struct gf100_gr_func { const struct gf100_gr_pack *mmio; struct { struct gf100_gr_ucode *ucode; + void (*reset)(struct gf100_gr *); } fecs; struct { struct gf100_gr_ucode *ucode; + void (*reset)(struct gf100_gr *); } gpccs; int (*rops)(struct gf100_gr *); int gpc_nr; @@ -197,7 +201,7 @@ struct gf100_gr_func { int gf100_gr_rops(struct gf100_gr *); void gf100_gr_oneinit_tiles(struct gf100_gr *); -void gf100_gr_oneinit_sm_id(struct gf100_gr *); +int gf100_gr_oneinit_sm_id(struct gf100_gr *); int gf100_gr_init(struct gf100_gr *); void gf100_gr_init_vsc_stream_master(struct gf100_gr *); void gf100_gr_init_zcull(struct gf100_gr *); @@ -208,9 +212,12 @@ void gf100_gr_init_419cc0(struct gf100_gr *); void gf100_gr_init_419eb4(struct gf100_gr *); void gf100_gr_init_tex_hww_esr(struct gf100_gr *, int, int); void gf100_gr_init_shader_exceptions(struct gf100_gr *, int, int); +void gf100_gr_init_rop_exceptions(struct gf100_gr *); +void gf100_gr_init_exception2(struct gf100_gr *); void gf100_gr_init_400054(struct gf100_gr *); void gf100_gr_init_num_tpc_per_gpc(struct gf100_gr *, bool, bool); extern const struct gf100_gr_func_zbc gf100_gr_zbc; +void gf100_gr_fecs_reset(struct gf100_gr *); void gf117_gr_init_zcull(struct gf100_gr *); @@ -226,9 +233,13 @@ void gm107_gr_init_shader_exceptions(struct gf100_gr *, int, int); void gm107_gr_init_400054(struct gf100_gr *); int gk20a_gr_init(struct gf100_gr *); +int gk20a_gr_av_to_init_(struct nvkm_blob *, u8 count, u32 pitch, struct gf100_gr_pack **); +int gk20a_gr_av_to_init(struct nvkm_blob *, struct gf100_gr_pack **); +int gk20a_gr_aiv_to_init(struct nvkm_blob *, struct gf100_gr_pack **); +int gk20a_gr_av_to_method(struct nvkm_blob *, struct gf100_gr_pack **); void gm200_gr_oneinit_tiles(struct gf100_gr *); -void gm200_gr_oneinit_sm_id(struct gf100_gr *); +int gm200_gr_oneinit_sm_id(struct gf100_gr *); int gm200_gr_rops(struct gf100_gr *); void gm200_gr_init_num_active_ltcs(struct gf100_gr *); void gm200_gr_init_ds_hww_esr_2(struct gf100_gr *); @@ -242,14 +253,24 @@ extern const struct gf100_gr_func_zbc gp100_gr_zbc; void gp102_gr_init_swdx_pes_mask(struct gf100_gr *); extern const struct gf100_gr_func_zbc gp102_gr_zbc; +int gp102_gr_zbc_stencil_get(struct gf100_gr *, int, const u32, const u32); +void gp102_gr_zbc_clear_stencil(struct gf100_gr *, int); extern const struct gf100_gr_func gp107_gr; +int gv100_gr_oneinit_sm_id(struct gf100_gr *); +u32 gv100_gr_nonpes_aware_tpc(struct gf100_gr *gr, u32 gpc, u32 tpc); void gv100_gr_init_419bd8(struct gf100_gr *); void gv100_gr_init_504430(struct gf100_gr *, int, int); void gv100_gr_init_shader_exceptions(struct gf100_gr *, int, int); +void gv100_gr_init_4188a4(struct gf100_gr *); void gv100_gr_trap_mp(struct gf100_gr *, int, int); +int tu102_gr_av_to_init_veid(struct nvkm_blob *, struct gf100_gr_pack **); +void tu102_gr_init_zcull(struct gf100_gr *); +void tu102_gr_init_fs(struct gf100_gr *); +void tu102_gr_init_fecs_exceptions(struct gf100_gr *); + #define gf100_gr_chan(p) container_of((p), struct gf100_gr_chan, object) #include <core/object.h> @@ -258,14 +279,14 @@ struct gf100_gr_chan { struct gf100_gr *gr; struct nvkm_vmm *vmm; + struct nvkm_vma *pagepool; + struct nvkm_vma *bundle_cb; + struct nvkm_vma *attrib_cb; + struct nvkm_vma *unknown; + struct nvkm_memory *mmio; struct nvkm_vma *mmio_vma; int mmio_nr; - - struct { - struct nvkm_memory *mem; - struct nvkm_vma *vma; - } data[4]; }; void gf100_gr_ctxctl_debug(struct gf100_gr *); @@ -279,7 +300,7 @@ struct gf100_gr_init { u32 addr; u8 count; u32 pitch; - u32 data; + u64 data; }; struct gf100_gr_pack { @@ -403,6 +424,9 @@ int gf100_gr_load(struct gf100_gr *, int, const struct gf100_gr_fwif *); int gf100_gr_nofw(struct gf100_gr *, int, const struct gf100_gr_fwif *); int gk20a_gr_load_sw(struct gf100_gr *, const char *path, int ver); +int gk20a_gr_load_net(struct gf100_gr *, const char *, const char *, int, + int (*)(struct nvkm_blob *, struct gf100_gr_pack **), + struct gf100_gr_pack **); int gm200_gr_nofw(struct gf100_gr *, int, const struct gf100_gr_fwif *); int gm200_gr_load(struct gf100_gr *, int, const struct gf100_gr_fwif *); @@ -415,6 +439,8 @@ void gm20b_gr_acr_bld_patch(struct nvkm_acr *, u32, s64); extern const struct nvkm_acr_lsf_func gp108_gr_gpccs_acr; extern const struct nvkm_acr_lsf_func gp108_gr_fecs_acr; +void gp108_gr_acr_bld_write(struct nvkm_acr *, u32, struct nvkm_acr_lsfw *); +void gp108_gr_acr_bld_patch(struct nvkm_acr *, u32, s64); int gf100_gr_new_(const struct gf100_gr_fwif *, struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_gr **); diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf104.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf104.c index 3acd99c306f2..63bd29c22fe1 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf104.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf104.c @@ -127,10 +127,13 @@ gf104_gr = { .init_419eb4 = gf100_gr_init_419eb4, .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_shader_exceptions = gf100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gf100_gr_init_400054, .trap_mp = gf100_gr_trap_mp, .mmio = gf104_gr_pack_mmio, .fecs.ucode = &gf100_gr_fecs_ucode, + .fecs.reset = gf100_gr_fecs_reset, .gpccs.ucode = &gf100_gr_gpccs_ucode, .rops = gf100_gr_rops, .grctx = &gf104_grctx, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf108.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf108.c index ab3760e804b8..495a844f925f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf108.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf108.c @@ -125,10 +125,13 @@ gf108_gr = { .init_419eb4 = gf100_gr_init_419eb4, .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_shader_exceptions = gf100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gf100_gr_init_400054, .trap_mp = gf100_gr_trap_mp, .mmio = gf108_gr_pack_mmio, .fecs.ucode = &gf100_gr_fecs_ucode, + .fecs.reset = gf100_gr_fecs_reset, .gpccs.ucode = &gf100_gr_gpccs_ucode, .rops = gf100_gr_rops, .grctx = &gf108_grctx, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf110.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf110.c index 616e2def1865..70fad235d161 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf110.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf110.c @@ -99,10 +99,13 @@ gf110_gr = { .init_419eb4 = gf100_gr_init_419eb4, .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_shader_exceptions = gf100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gf100_gr_init_400054, .trap_mp = gf100_gr_trap_mp, .mmio = gf110_gr_pack_mmio, .fecs.ucode = &gf100_gr_fecs_ucode, + .fecs.reset = gf100_gr_fecs_reset, .gpccs.ucode = &gf100_gr_gpccs_ucode, .rops = gf100_gr_rops, .grctx = &gf110_grctx, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf117.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf117.c index 669e7536970e..f12728248048 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf117.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf117.c @@ -125,7 +125,9 @@ gf117_gr_init_zcull(struct gf100_gr *gr) { struct nvkm_device *device = gr->base.engine.subdev.device; const u32 magicgpc918 = DIV_ROUND_UP(0x00800000, gr->tpc_total); - const u8 tile_nr = ALIGN(gr->tpc_total, 32); + /*TODO: fill in litter vals for gf117-gm2xx */ + const u8 tile_nr = !gr->func->gpc_nr ? ALIGN(gr->tpc_total, 32) : + (gr->func->gpc_nr * gr->func->tpc_nr); u8 bank[GPC_MAX] = {}, gpc, i, j; u32 data; @@ -163,10 +165,13 @@ gf117_gr = { .init_419eb4 = gf100_gr_init_419eb4, .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_shader_exceptions = gf100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gf100_gr_init_400054, .trap_mp = gf100_gr_trap_mp, .mmio = gf117_gr_pack_mmio, .fecs.ucode = &gf117_gr_fecs_ucode, + .fecs.reset = gf100_gr_fecs_reset, .gpccs.ucode = &gf117_gr_gpccs_ucode, .rops = gf100_gr_rops, .ppc_nr = 1, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf119.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf119.c index 5b09bda8110c..75ceb514c06e 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf119.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf119.c @@ -190,10 +190,13 @@ gf119_gr = { .init_419eb4 = gf100_gr_init_419eb4, .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_shader_exceptions = gf100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gf100_gr_init_400054, .trap_mp = gf100_gr_trap_mp, .mmio = gf119_gr_pack_mmio, .fecs.ucode = &gf100_gr_fecs_ucode, + .fecs.reset = gf100_gr_fecs_reset, .gpccs.ucode = &gf100_gr_gpccs_ucode, .rops = gf100_gr_rops, .grctx = &gf119_grctx, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk104.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk104.c index b680eaa0f350..e53ade24ad23 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk104.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk104.c @@ -418,7 +418,7 @@ gk104_gr_init_ppc_exceptions(struct gf100_gr *gr) int gpc, ppc; for (gpc = 0; gpc < gr->gpc_nr; gpc++) { - for (ppc = 0; ppc < gr->ppc_nr[gpc]; ppc++) { + for (ppc = 0; ppc < gr->func->ppc_nr; ppc++) { if (!(gr->ppc_mask[gpc] & (1 << ppc))) continue; nvkm_wr32(device, PPC_UNIT(gpc, ppc, 0x038), 0xc0000000); @@ -470,10 +470,13 @@ gk104_gr = { .init_ppc_exceptions = gk104_gr_init_ppc_exceptions, .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_shader_exceptions = gf100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gf100_gr_init_400054, .trap_mp = gf100_gr_trap_mp, .mmio = gk104_gr_pack_mmio, .fecs.ucode = &gk104_gr_fecs_ucode, + .fecs.reset = gf100_gr_fecs_reset, .gpccs.ucode = &gk104_gr_gpccs_ucode, .rops = gf100_gr_rops, .ppc_nr = 1, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk110.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk110.c index 103e06a77e65..c7e1c5dbc6a9 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk110.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk110.c @@ -366,10 +366,13 @@ gk110_gr = { .init_ppc_exceptions = gk104_gr_init_ppc_exceptions, .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_shader_exceptions = gf100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gf100_gr_init_400054, .trap_mp = gf100_gr_trap_mp, .mmio = gk110_gr_pack_mmio, .fecs.ucode = &gk110_gr_fecs_ucode, + .fecs.reset = gf100_gr_fecs_reset, .gpccs.ucode = &gk110_gr_gpccs_ucode, .rops = gf100_gr_rops, .ppc_nr = 2, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk110b.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk110b.c index 034d0b11a17d..458abae571bf 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk110b.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk110b.c @@ -118,10 +118,13 @@ gk110b_gr = { .init_ppc_exceptions = gk104_gr_init_ppc_exceptions, .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_shader_exceptions = gf100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gf100_gr_init_400054, .trap_mp = gf100_gr_trap_mp, .mmio = gk110b_gr_pack_mmio, .fecs.ucode = &gk110_gr_fecs_ucode, + .fecs.reset = gf100_gr_fecs_reset, .gpccs.ucode = &gk110_gr_gpccs_ucode, .rops = gf100_gr_rops, .ppc_nr = 2, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk208.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk208.c index 116d682f9f96..d3f6b65c21d2 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk208.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk208.c @@ -176,10 +176,13 @@ gk208_gr = { .init_ppc_exceptions = gk104_gr_init_ppc_exceptions, .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_shader_exceptions = gf100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gf100_gr_init_400054, .trap_mp = gf100_gr_trap_mp, .mmio = gk208_gr_pack_mmio, .fecs.ucode = &gk208_gr_fecs_ucode, + .fecs.reset = gf100_gr_fecs_reset, .gpccs.ucode = &gk208_gr_gpccs_ucode, .rops = gf100_gr_rops, .ppc_nr = 1, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk20a.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk20a.c index be0b2cefd8e8..035ea213f543 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk20a.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gk20a.c @@ -33,47 +33,40 @@ struct gk20a_fw_av u32 data; }; -static int -gk20a_gr_av_to_init(struct gf100_gr *gr, const char *path, const char *name, - int ver, struct gf100_gr_pack **ppack) +int +gk20a_gr_av_to_init_(struct nvkm_blob *blob, u8 count, u32 pitch, struct gf100_gr_pack **ppack) { - struct nvkm_subdev *subdev = &gr->base.engine.subdev; - struct nvkm_blob blob; struct gf100_gr_init *init; struct gf100_gr_pack *pack; int nent; - int ret; int i; - ret = nvkm_firmware_load_blob(subdev, path, name, ver, &blob); - if (ret) - return ret; - - nent = (blob.size / sizeof(struct gk20a_fw_av)); + nent = (blob->size / sizeof(struct gk20a_fw_av)); pack = vzalloc((sizeof(*pack) * 2) + (sizeof(*init) * (nent + 1))); - if (!pack) { - ret = -ENOMEM; - goto end; - } + if (!pack) + return -ENOMEM; init = (void *)(pack + 2); pack[0].init = init; for (i = 0; i < nent; i++) { struct gf100_gr_init *ent = &init[i]; - struct gk20a_fw_av *av = &((struct gk20a_fw_av *)blob.data)[i]; + struct gk20a_fw_av *av = &((struct gk20a_fw_av *)blob->data)[i]; ent->addr = av->addr; ent->data = av->data; - ent->count = 1; - ent->pitch = 1; + ent->count = ((ent->addr & 0xffff) != 0xe100) ? count : 1; + ent->pitch = pitch; } *ppack = pack; + return 0; +} -end: - nvkm_blob_dtor(&blob); - return ret; +int +gk20a_gr_av_to_init(struct nvkm_blob *blob, struct gf100_gr_pack **ppack) +{ + return gk20a_gr_av_to_init_(blob, 1, 1, ppack); } struct gk20a_fw_aiv @@ -83,35 +76,25 @@ struct gk20a_fw_aiv u32 data; }; -static int -gk20a_gr_aiv_to_init(struct gf100_gr *gr, const char *path, const char *name, - int ver, struct gf100_gr_pack **ppack) +int +gk20a_gr_aiv_to_init(struct nvkm_blob *blob, struct gf100_gr_pack **ppack) { - struct nvkm_subdev *subdev = &gr->base.engine.subdev; - struct nvkm_blob blob; struct gf100_gr_init *init; struct gf100_gr_pack *pack; int nent; - int ret; int i; - ret = nvkm_firmware_load_blob(subdev, path, name, ver, &blob); - if (ret) - return ret; - - nent = (blob.size / sizeof(struct gk20a_fw_aiv)); + nent = (blob->size / sizeof(struct gk20a_fw_aiv)); pack = vzalloc((sizeof(*pack) * 2) + (sizeof(*init) * (nent + 1))); - if (!pack) { - ret = -ENOMEM; - goto end; - } + if (!pack) + return -ENOMEM; init = (void *)(pack + 2); pack[0].init = init; for (i = 0; i < nent; i++) { struct gf100_gr_init *ent = &init[i]; - struct gk20a_fw_aiv *av = &((struct gk20a_fw_aiv *)blob.data)[i]; + struct gk20a_fw_aiv *av = &((struct gk20a_fw_aiv *)blob->data)[i]; ent->addr = av->addr; ent->data = av->data; @@ -120,44 +103,30 @@ gk20a_gr_aiv_to_init(struct gf100_gr *gr, const char *path, const char *name, } *ppack = pack; - -end: - nvkm_blob_dtor(&blob); - return ret; + return 0; } -static int -gk20a_gr_av_to_method(struct gf100_gr *gr, const char *path, const char *name, - int ver, struct gf100_gr_pack **ppack) +int +gk20a_gr_av_to_method(struct nvkm_blob *blob, struct gf100_gr_pack **ppack) { - struct nvkm_subdev *subdev = &gr->base.engine.subdev; - struct nvkm_blob blob; struct gf100_gr_init *init; struct gf100_gr_pack *pack; /* We don't suppose we will initialize more than 16 classes here... */ static const unsigned int max_classes = 16; u32 classidx = 0, prevclass = 0; int nent; - int ret; int i; - ret = nvkm_firmware_load_blob(subdev, path, name, ver, &blob); - if (ret) - return ret; - - nent = (blob.size / sizeof(struct gk20a_fw_av)); - + nent = (blob->size / sizeof(struct gk20a_fw_av)); pack = vzalloc((sizeof(*pack) * (max_classes + 1)) + (sizeof(*init) * (nent + max_classes + 1))); - if (!pack) { - ret = -ENOMEM; - goto end; - } + if (!pack) + return -ENOMEM; init = (void *)(pack + max_classes + 1); for (i = 0; i < nent; i++, init++) { - struct gk20a_fw_av *av = &((struct gk20a_fw_av *)blob.data)[i]; + struct gk20a_fw_av *av = &((struct gk20a_fw_av *)blob->data)[i]; u32 class = av->addr & 0xffff; u32 addr = (av->addr & 0xffff0000) >> 14; @@ -169,8 +138,7 @@ gk20a_gr_av_to_method(struct gf100_gr *gr, const char *path, const char *name, prevclass = class; if (++classidx >= max_classes) { vfree(pack); - ret = -ENOSPC; - goto end; + return -ENOSPC; } } @@ -181,10 +149,7 @@ gk20a_gr_av_to_method(struct gf100_gr *gr, const char *path, const char *name, } *ppack = pack; - -end: - nvkm_blob_dtor(&blob); - return ret; + return 0; } static int @@ -294,6 +259,7 @@ gk20a_gr = { .init_rop_active_fbps = gk104_gr_init_rop_active_fbps, .trap_mp = gf100_gr_trap_mp, .set_hww_esr_report_mask = gk20a_gr_set_hww_esr_report_mask, + .fecs.reset = gf100_gr_fecs_reset, .rops = gf100_gr_rops, .ppc_nr = 1, .grctx = &gk20a_grctx, @@ -308,12 +274,29 @@ gk20a_gr = { }; int +gk20a_gr_load_net(struct gf100_gr *gr, const char *path, const char *name, int ver, + int (*load)(struct nvkm_blob *, struct gf100_gr_pack **), + struct gf100_gr_pack **ppack) +{ + struct nvkm_blob blob; + int ret; + + ret = nvkm_firmware_load_blob(&gr->base.engine.subdev, path, name, ver, &blob); + if (ret) + return ret; + + ret = load(&blob, ppack); + nvkm_blob_dtor(&blob); + return 0; +} + +int gk20a_gr_load_sw(struct gf100_gr *gr, const char *path, int ver) { - if (gk20a_gr_av_to_init(gr, path, "sw_nonctx", ver, &gr->sw_nonctx) || - gk20a_gr_aiv_to_init(gr, path, "sw_ctx", ver, &gr->sw_ctx) || - gk20a_gr_av_to_init(gr, path, "sw_bundle_init", ver, &gr->bundle) || - gk20a_gr_av_to_method(gr, path, "sw_method_init", ver, &gr->method)) + if (gk20a_gr_load_net(gr, path, "sw_nonctx", ver, gk20a_gr_av_to_init, &gr->sw_nonctx) || + gk20a_gr_load_net(gr, path, "sw_ctx", ver, gk20a_gr_aiv_to_init, &gr->sw_ctx) || + gk20a_gr_load_net(gr, path, "sw_bundle_init", ver, gk20a_gr_av_to_init, &gr->bundle) || + gk20a_gr_load_net(gr, path, "sw_method_init", ver, gk20a_gr_av_to_method, &gr->method)) return -ENOENT; return 0; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm107.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm107.c index 310987174cb5..797b828a943b 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm107.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm107.c @@ -411,10 +411,13 @@ gm107_gr = { .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_504430 = gm107_gr_init_504430, .init_shader_exceptions = gm107_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gm107_gr_init_400054, .trap_mp = gf100_gr_trap_mp, .mmio = gm107_gr_pack_mmio, .fecs.ucode = &gm107_gr_fecs_ucode, + .fecs.reset = gf100_gr_fecs_reset, .gpccs.ucode = &gm107_gr_gpccs_ucode, .rops = gf100_gr_rops, .ppc_nr = 2, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm200.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm200.c index 385cfd91b266..b5210b31c1b2 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm200.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm200.c @@ -148,11 +148,11 @@ gm200_gr_tile_map_2_8[] = { 0, 1, 1, 0, 0, 1, 1, 0, }; -void +int gm200_gr_oneinit_sm_id(struct gf100_gr *gr) { /*XXX: There's a different algorithm here I've not yet figured out. */ - gf100_gr_oneinit_sm_id(gr); + return gf100_gr_oneinit_sm_id(gr); } void @@ -199,8 +199,11 @@ gm200_gr = { .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_504430 = gm107_gr_init_504430, .init_shader_exceptions = gm107_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_400054 = gm107_gr_init_400054, .trap_mp = gf100_gr_trap_mp, + .fecs.reset = gf100_gr_fecs_reset, .rops = gm200_gr_rops, .tpc_nr = 4, .ppc_nr = 2, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm20b.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm20b.c index ec1c46e47e00..458cd1a00d3f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm20b.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gm20b.c @@ -123,6 +123,7 @@ gm20b_gr = { .init_rop_active_fbps = gk104_gr_init_rop_active_fbps, .trap_mp = gf100_gr_trap_mp, .set_hww_esr_report_mask = gm20b_gr_set_hww_esr_report_mask, + .fecs.reset = gf100_gr_fecs_reset, .rops = gm200_gr_rops, .ppc_nr = 1, .grctx = &gm20b_grctx, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp100.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp100.c index 0550dd6f46f1..851e743d2cab 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp100.c @@ -87,7 +87,7 @@ gp100_gr_init_419c9c(struct gf100_gr *gr) void gp100_gr_init_fecs_exceptions(struct gf100_gr *gr) { - nvkm_wr32(gr->base.engine.subdev.device, 0x409c24, 0x000f0002); + nvkm_wr32(gr->base.engine.subdev.device, 0x409c24, 0x000e0002); } void @@ -119,7 +119,10 @@ gp100_gr = { .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_504430 = gm107_gr_init_504430, .init_shader_exceptions = gp100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .trap_mp = gf100_gr_trap_mp, + .fecs.reset = gf100_gr_fecs_reset, .rops = gm200_gr_rops, .gpc_nr = 6, .tpc_nr = 5, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp102.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp102.c index 5b001f374be0..0e223b7b5f0e 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp102.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp102.c @@ -26,7 +26,7 @@ #include <nvif/class.h> -static void +void gp102_gr_zbc_clear_stencil(struct gf100_gr *gr, int zbc) { struct nvkm_device *device = gr->base.engine.subdev.device; @@ -40,14 +40,14 @@ gp102_gr_zbc_clear_stencil(struct gf100_gr *gr, int zbc) gr->zbc_stencil[zbc].format << ((znum % 4) * 7)); } -static int +int gp102_gr_zbc_stencil_get(struct gf100_gr *gr, int format, const u32 ds, const u32 l2) { struct nvkm_ltc *ltc = gr->base.engine.subdev.device->ltc; int zbc = -ENOSPC, i; - for (i = ltc->zbc_min; i <= ltc->zbc_max; i++) { + for (i = ltc->zbc_depth_min; i <= ltc->zbc_depth_max; i++) { if (gr->zbc_stencil[i].format) { if (gr->zbc_stencil[i].format != format) continue; @@ -115,7 +115,10 @@ gp102_gr = { .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_504430 = gm107_gr_init_504430, .init_shader_exceptions = gp100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .trap_mp = gf100_gr_trap_mp, + .fecs.reset = gf100_gr_fecs_reset, .rops = gm200_gr_rops, .gpc_nr = 6, .tpc_nr = 5, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp104.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp104.c index 2655574ec63b..6802cb9b199f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp104.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp104.c @@ -43,7 +43,10 @@ gp104_gr = { .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_504430 = gm107_gr_init_504430, .init_shader_exceptions = gp100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .trap_mp = gf100_gr_trap_mp, + .fecs.reset = gf100_gr_fecs_reset, .rops = gm200_gr_rops, .gpc_nr = 6, .tpc_nr = 5, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp107.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp107.c index adabc04d4f3a..cc2bb0d0a987 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp107.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp107.c @@ -45,7 +45,10 @@ gp107_gr = { .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_504430 = gm107_gr_init_504430, .init_shader_exceptions = gp100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .trap_mp = gf100_gr_trap_mp, + .fecs.reset = gf100_gr_fecs_reset, .rops = gm200_gr_rops, .gpc_nr = 2, .tpc_nr = 3, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp108.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp108.c index 7310f0466bb7..311f703439e4 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp108.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp108.c @@ -25,7 +25,7 @@ #include <nvfw/flcn.h> -static void +void gp108_gr_acr_bld_patch(struct nvkm_acr *acr, u32 bld, s64 adjust) { struct flcn_bl_dmem_desc_v2 hdr; @@ -36,7 +36,7 @@ gp108_gr_acr_bld_patch(struct nvkm_acr *acr, u32 bld, s64 adjust) flcn_bl_dmem_desc_v2_dump(&acr->subdev, &hdr); } -static void +void gp108_gr_acr_bld_write(struct nvkm_acr *acr, u32 bld, struct nvkm_acr_lsfw *lsfw) { diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp10b.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp10b.c index e13683b6e7b1..5008881ca079 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp10b.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gp10b.c @@ -55,7 +55,10 @@ gp10b_gr = { .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, .init_504430 = gm107_gr_init_504430, .init_shader_exceptions = gp100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .trap_mp = gf100_gr_trap_mp, + .fecs.reset = gf100_gr_fecs_reset, .rops = gm200_gr_rops, .gpc_nr = 1, .tpc_nr = 2, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gv100.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gv100.c index 4d043c1173ea..7f7404a76140 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gv100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gv100.c @@ -52,10 +52,11 @@ gv100_gr_trap_mp(struct gf100_gr *gr, int gpc, int tpc) gv100_gr_trap_sm(gr, gpc, tpc, 1); } -static void +void gv100_gr_init_4188a4(struct gf100_gr *gr) { struct nvkm_device *device = gr->base.engine.subdev.device; + nvkm_mask(device, 0x4188a4, 0x03000000, 0x03000000); } @@ -65,7 +66,6 @@ gv100_gr_init_shader_exceptions(struct gf100_gr *gr, int gpc, int tpc) struct nvkm_device *device = gr->base.engine.subdev.device; int sm; for (sm = 0; sm < 0x100; sm += 0x80) { - nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x728 + sm), 0x0085eb64); nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x610), 0x00000001); nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x72c + sm), 0x00000004); } @@ -85,10 +85,202 @@ gv100_gr_init_419bd8(struct gf100_gr *gr) nvkm_mask(device, 0x419bd8, 0x00000700, 0x00000000); } +u32 +gv100_gr_nonpes_aware_tpc(struct gf100_gr *gr, u32 gpc, u32 tpc) +{ + u32 pes, temp, tpc_new = 0; + + for (pes = 0; pes < gr->ppc_nr[gpc]; pes++) { + if (gr->ppc_tpc_mask[gpc][pes] & BIT(tpc)) + break; + + tpc_new += gr->ppc_tpc_nr[gpc][pes]; + } + + temp = (BIT(tpc) - 1) & gr->ppc_tpc_mask[gpc][pes]; + temp = hweight32(temp); + return tpc_new + temp; +} + +static int +gv100_gr_scg_estimate_perf(struct gf100_gr *gr, unsigned long *gpc_tpc_mask, + u32 disable_gpc, u32 disable_tpc, int *perf) +{ + const u32 scale_factor = 512UL; /* Use fx23.9 */ + const u32 pix_scale = 1024*1024UL; /* Pix perf in [29:20] */ + const u32 world_scale = 1024UL; /* World performance in [19:10] */ + const u32 tpc_scale = 1; /* TPC balancing in [9:0] */ + u32 scg_num_pes = 0; + u32 min_scg_gpc_pix_perf = scale_factor; /* Init perf as maximum */ + u32 average_tpcs = 0; /* Average of # of TPCs per GPC */ + u32 deviation; /* absolute diff between TPC# and average_tpcs, averaged across GPCs */ + u32 norm_tpc_deviation; /* deviation/max_tpc_per_gpc */ + u32 tpc_balance; + u32 scg_gpc_pix_perf; + u32 scg_world_perf; + u32 gpc; + u32 pes; + int diff; + bool tpc_removed_gpc = false; + bool tpc_removed_pes = false; + u32 max_tpc_gpc = 0; + u32 num_tpc_mask; + u32 *num_tpc_gpc; + int ret = -EINVAL; + + if (!(num_tpc_gpc = kcalloc(gr->gpc_nr, sizeof(*num_tpc_gpc), GFP_KERNEL))) + return -ENOMEM; + + /* Calculate pix-perf-reduction-rate per GPC and find bottleneck TPC */ + for (gpc = 0; gpc < gr->gpc_nr; gpc++) { + num_tpc_mask = gpc_tpc_mask[gpc]; + + if ((gpc == disable_gpc) && num_tpc_mask & BIT(disable_tpc)) { + /* Safety check if a TPC is removed twice */ + if (WARN_ON(tpc_removed_gpc)) + goto done; + + /* Remove logical TPC from set */ + num_tpc_mask &= ~BIT(disable_tpc); + tpc_removed_gpc = true; + } + + /* track balancing of tpcs across gpcs */ + num_tpc_gpc[gpc] = hweight32(num_tpc_mask); + average_tpcs += num_tpc_gpc[gpc]; + + /* save the maximum numer of gpcs */ + max_tpc_gpc = num_tpc_gpc[gpc] > max_tpc_gpc ? num_tpc_gpc[gpc] : max_tpc_gpc; + + /* + * Calculate ratio between TPC count and post-FS and post-SCG + * + * ratio represents relative throughput of the GPC + */ + scg_gpc_pix_perf = scale_factor * num_tpc_gpc[gpc] / gr->tpc_nr[gpc]; + if (min_scg_gpc_pix_perf > scg_gpc_pix_perf) + min_scg_gpc_pix_perf = scg_gpc_pix_perf; + + /* Calculate # of surviving PES */ + for (pes = 0; pes < gr->ppc_nr[gpc]; pes++) { + /* Count the number of TPC on the set */ + num_tpc_mask = gr->ppc_tpc_mask[gpc][pes] & gpc_tpc_mask[gpc]; + + if ((gpc == disable_gpc) && (num_tpc_mask & BIT(disable_tpc))) { + if (WARN_ON(tpc_removed_pes)) + goto done; + + num_tpc_mask &= ~BIT(disable_tpc); + tpc_removed_pes = true; + } + + if (hweight32(num_tpc_mask)) + scg_num_pes++; + } + } + + if (WARN_ON(!tpc_removed_gpc || !tpc_removed_pes)) + goto done; + + if (max_tpc_gpc == 0) { + *perf = 0; + goto done_ok; + } + + /* Now calculate perf */ + scg_world_perf = (scale_factor * scg_num_pes) / gr->ppc_total; + deviation = 0; + average_tpcs = scale_factor * average_tpcs / gr->gpc_nr; + for (gpc = 0; gpc < gr->gpc_nr; gpc++) { + diff = average_tpcs - scale_factor * num_tpc_gpc[gpc]; + if (diff < 0) + diff = -diff; + + deviation += diff; + } + + deviation /= gr->gpc_nr; + + norm_tpc_deviation = deviation / max_tpc_gpc; + + tpc_balance = scale_factor - norm_tpc_deviation; + + if ((tpc_balance > scale_factor) || + (scg_world_perf > scale_factor) || + (min_scg_gpc_pix_perf > scale_factor) || + (norm_tpc_deviation > scale_factor)) { + WARN_ON(1); + goto done; + } + + *perf = (pix_scale * min_scg_gpc_pix_perf) + + (world_scale * scg_world_perf) + + (tpc_scale * tpc_balance); +done_ok: + ret = 0; +done: + kfree(num_tpc_gpc); + return ret; +} + +int +gv100_gr_oneinit_sm_id(struct gf100_gr *gr) +{ + unsigned long *gpc_tpc_mask; + u32 *tpc_table, *gpc_table; + u32 gpc, tpc, pes, gtpc; + int perf, maxperf, ret = 0; + + gpc_tpc_mask = kcalloc(gr->gpc_nr, sizeof(*gpc_tpc_mask), GFP_KERNEL); + gpc_table = kcalloc(gr->tpc_total, sizeof(*gpc_table), GFP_KERNEL); + tpc_table = kcalloc(gr->tpc_total, sizeof(*tpc_table), GFP_KERNEL); + if (!gpc_table || !tpc_table || !gpc_tpc_mask) { + ret = -ENOMEM; + goto done; + } + + for (gpc = 0; gpc < gr->gpc_nr; gpc++) { + for (pes = 0; pes < gr->ppc_nr[gpc]; pes++) + gpc_tpc_mask[gpc] |= gr->ppc_tpc_mask[gpc][pes]; + } + + for (gtpc = 0; gtpc < gr->tpc_total; gtpc++) { + for (maxperf = -1, gpc = 0; gpc < gr->gpc_nr; gpc++) { + for_each_set_bit(tpc, &gpc_tpc_mask[gpc], gr->tpc_nr[gpc]) { + ret = gv100_gr_scg_estimate_perf(gr, gpc_tpc_mask, gpc, tpc, &perf); + if (ret) + goto done; + + /* nvgpu does ">=" here, but this gets us RM's numbers. */ + if (perf > maxperf) { + maxperf = perf; + gpc_table[gtpc] = gpc; + tpc_table[gtpc] = tpc; + } + } + } + + gpc_tpc_mask[gpc_table[gtpc]] &= ~BIT(tpc_table[gtpc]); + } + + /*TODO: build table for sm_per_tpc != 1, don't use yet, but might need later? */ + for (gtpc = 0; gtpc < gr->tpc_total; gtpc++) { + gr->sm[gtpc].gpc = gpc_table[gtpc]; + gr->sm[gtpc].tpc = tpc_table[gtpc]; + gr->sm_nr++; + } + +done: + kfree(gpc_table); + kfree(tpc_table); + kfree(gpc_tpc_mask); + return ret; +} + static const struct gf100_gr_func gv100_gr = { .oneinit_tiles = gm200_gr_oneinit_tiles, - .oneinit_sm_id = gm200_gr_oneinit_sm_id, + .oneinit_sm_id = gv100_gr_oneinit_sm_id, .init = gf100_gr_init, .init_419bd8 = gv100_gr_init_419bd8, .init_gpc_mmu = gm200_gr_init_gpc_mmu, @@ -103,11 +295,14 @@ gv100_gr = { .init_ppc_exceptions = gk104_gr_init_ppc_exceptions, .init_504430 = gv100_gr_init_504430, .init_shader_exceptions = gv100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, .init_4188a4 = gv100_gr_init_4188a4, .trap_mp = gv100_gr_trap_mp, + .fecs.reset = gf100_gr_fecs_reset, .rops = gm200_gr_rops, .gpc_nr = 6, - .tpc_nr = 5, + .tpc_nr = 7, .ppc_nr = 3, .grctx = &gv100_grctx, .zbc = &gp102_gr_zbc, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv04.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv04.c index 0bc1a238de43..81bd682c2102 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv04.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv04.c @@ -1192,7 +1192,7 @@ nv04_gr_chan_new(struct nvkm_gr *base, struct nvkm_fifo_chan *fifoch, return -ENOMEM; nvkm_object_ctor(&nv04_gr_chan, oclass, &chan->object); chan->gr = gr; - chan->chid = fifoch->chid; + chan->chid = fifoch->id; *pobject = &chan->object; *ctx_reg(chan, NV04_PGRAPH_DEBUG_3) = 0xfad4ff31; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv10.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv10.c index 942450b33bc6..7fe6e58f6bab 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv10.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv10.c @@ -1011,7 +1011,7 @@ nv10_gr_chan_new(struct nvkm_gr *base, struct nvkm_fifo_chan *fifoch, return -ENOMEM; nvkm_object_ctor(&nv10_gr_chan, oclass, &chan->object); chan->gr = gr; - chan->chid = fifoch->chid; + chan->chid = fifoch->id; *pobject = &chan->object; NV_WRITE_CTX(0x00400e88, 0x08000000); diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv20.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv20.c index 6bff10cee71b..75434f5de7ad 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv20.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv20.c @@ -83,7 +83,7 @@ nv20_gr_chan_new(struct nvkm_gr *base, struct nvkm_fifo_chan *fifoch, return -ENOMEM; nvkm_object_ctor(&nv20_gr_chan, oclass, &chan->object); chan->gr = gr; - chan->chid = fifoch->chid; + chan->chid = fifoch->id; *pobject = &chan->object; ret = nvkm_memory_new(gr->base.engine.subdev.device, @@ -182,7 +182,7 @@ nv20_gr_intr(struct nvkm_gr *base) struct nv20_gr *gr = nv20_gr(base); struct nvkm_subdev *subdev = &gr->base.engine.subdev; struct nvkm_device *device = subdev->device; - struct nvkm_fifo_chan *chan; + struct nvkm_chan *chan; u32 stat = nvkm_rd32(device, NV03_PGRAPH_INTR); u32 nsource = nvkm_rd32(device, NV03_PGRAPH_NSOURCE); u32 nstatus = nvkm_rd32(device, NV03_PGRAPH_NSTATUS); @@ -196,7 +196,7 @@ nv20_gr_intr(struct nvkm_gr *base) char msg[128], src[128], sta[128]; unsigned long flags; - chan = nvkm_fifo_chan_chid(device->fifo, chid, &flags); + chan = nvkm_chan_get_chid(&gr->base.engine, chid, &flags); nvkm_wr32(device, NV03_PGRAPH_INTR, stat); nvkm_wr32(device, NV04_PGRAPH_FIFO, 0x00000001); @@ -209,11 +209,11 @@ nv20_gr_intr(struct nvkm_gr *base) "nstatus %08x [%s] ch %d [%s] subc %d " "class %04x mthd %04x data %08x\n", show, msg, nsource, src, nstatus, sta, chid, - chan ? chan->object.client->name : "unknown", + chan ? chan->name : "unknown", subc, class, mthd, data); } - nvkm_fifo_chan_put(device->fifo, flags, &chan); + nvkm_chan_put(&chan, flags); } int diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv25.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv25.c index f3a56f17d94a..94685e4d4f87 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv25.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv25.c @@ -29,7 +29,7 @@ nv25_gr_chan_new(struct nvkm_gr *base, struct nvkm_fifo_chan *fifoch, return -ENOMEM; nvkm_object_ctor(&nv25_gr_chan, oclass, &chan->object); chan->gr = gr; - chan->chid = fifoch->chid; + chan->chid = fifoch->id; *pobject = &chan->object; ret = nvkm_memory_new(gr->base.engine.subdev.device, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv2a.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv2a.c index f268d2642d29..2d6273675291 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv2a.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv2a.c @@ -29,7 +29,7 @@ nv2a_gr_chan_new(struct nvkm_gr *base, struct nvkm_fifo_chan *fifoch, return -ENOMEM; nvkm_object_ctor(&nv2a_gr_chan, oclass, &chan->object); chan->gr = gr; - chan->chid = fifoch->chid; + chan->chid = fifoch->id; *pobject = &chan->object; ret = nvkm_memory_new(gr->base.engine.subdev.device, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv30.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv30.c index e5737cdf2fa1..647bd6fede04 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv30.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv30.c @@ -30,7 +30,7 @@ nv30_gr_chan_new(struct nvkm_gr *base, struct nvkm_fifo_chan *fifoch, return -ENOMEM; nvkm_object_ctor(&nv30_gr_chan, oclass, &chan->object); chan->gr = gr; - chan->chid = fifoch->chid; + chan->chid = fifoch->id; *pobject = &chan->object; ret = nvkm_memory_new(gr->base.engine.subdev.device, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv34.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv34.c index 1ab2da8ebf4e..2eae3fe4ef4e 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv34.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv34.c @@ -29,7 +29,7 @@ nv34_gr_chan_new(struct nvkm_gr *base, struct nvkm_fifo_chan *fifoch, return -ENOMEM; nvkm_object_ctor(&nv34_gr_chan, oclass, &chan->object); chan->gr = gr; - chan->chid = fifoch->chid; + chan->chid = fifoch->id; *pobject = &chan->object; ret = nvkm_memory_new(gr->base.engine.subdev.device, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv35.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv35.c index 591260f5676b..657d7cdba369 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv35.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv35.c @@ -29,7 +29,7 @@ nv35_gr_chan_new(struct nvkm_gr *base, struct nvkm_fifo_chan *fifoch, return -ENOMEM; nvkm_object_ctor(&nv35_gr_chan, oclass, &chan->object); chan->gr = gr; - chan->chid = fifoch->chid; + chan->chid = fifoch->id; *pobject = &chan->object; ret = nvkm_memory_new(gr->base.engine.subdev.device, diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv40.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv40.c index 67f3535ff97e..d2df097a6cf6 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv40.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv40.c @@ -275,8 +275,8 @@ nv40_gr_intr(struct nvkm_gr *base) "nstatus %08x [%s] ch %d [%08x %s] subc %d " "class %04x mthd %04x data %08x\n", show, msg, nsource, src, nstatus, sta, - chan ? chan->fifo->chid : -1, inst << 4, - chan ? chan->fifo->object.client->name : "unknown", + chan ? chan->fifo->id : -1, inst << 4, + chan ? chan->fifo->name : "unknown", subc, class, mthd, data); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv50.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv50.c index 563a10097e95..1ba18a8e380f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv50.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/nv50.c @@ -622,7 +622,7 @@ nv50_gr_intr(struct nvkm_gr *base) struct nv50_gr *gr = nv50_gr(base); struct nvkm_subdev *subdev = &gr->base.engine.subdev; struct nvkm_device *device = subdev->device; - struct nvkm_fifo_chan *chan; + struct nvkm_chan *chan; u32 stat = nvkm_rd32(device, 0x400100); u32 inst = nvkm_rd32(device, 0x40032c) & 0x0fffffff; u32 addr = nvkm_rd32(device, 0x400704); @@ -637,10 +637,10 @@ nv50_gr_intr(struct nvkm_gr *base) char msg[128]; int chid = -1; - chan = nvkm_fifo_chan_inst(device->fifo, (u64)inst << 12, &flags); + chan = nvkm_chan_get_inst(&gr->base.engine, (u64)inst << 12, &flags); if (chan) { - name = chan->object.client->name; - chid = chan->chid; + name = chan->name; + chid = chan->id; } if (show & 0x00100000) { @@ -672,7 +672,7 @@ nv50_gr_intr(struct nvkm_gr *base) if (nvkm_rd32(device, 0x400824) & (1 << 31)) nvkm_wr32(device, 0x400824, nvkm_rd32(device, 0x400824) & ~(1 << 31)); - nvkm_fifo_chan_put(device->fifo, flags, &chan); + nvkm_chan_put(&chan, flags); } int diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/priv.h b/drivers/gpu/drm/nouveau/nvkm/engine/gr/priv.h index 9b2c66e8be90..08d5c96e6458 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/priv.h @@ -17,6 +17,7 @@ struct nvkm_gr_func { int (*oneinit)(struct nvkm_gr *); int (*init)(struct nvkm_gr *); int (*fini)(struct nvkm_gr *, bool); + int (*reset)(struct nvkm_gr *); void (*intr)(struct nvkm_gr *); void (*tile)(struct nvkm_gr *, int region, struct nvkm_fb_tile *); int (*tlb_flush)(struct nvkm_gr *); diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/tu102.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/tu102.c index 1a8a21844e12..3b6c8100a242 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/tu102.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/tu102.c @@ -24,13 +24,13 @@ #include <nvif/class.h> -static void +void tu102_gr_init_fecs_exceptions(struct gf100_gr *gr) { - nvkm_wr32(gr->base.engine.subdev.device, 0x409c24, 0x006f0002); + nvkm_wr32(gr->base.engine.subdev.device, 0x409c24, 0x006e0003); } -static void +void tu102_gr_init_fs(struct gf100_gr *gr) { struct nvkm_device *device = gr->base.engine.subdev.device; @@ -40,20 +40,21 @@ tu102_gr_init_fs(struct gf100_gr *gr) gk104_grctx_generate_gpc_tpc_nr(gr); for (sm = 0; sm < gr->sm_nr; sm++) { - nvkm_wr32(device, GPC_UNIT(gr->sm[sm].gpc, 0x0c10 + - gr->sm[sm].tpc * 4), sm); + int tpc = gv100_gr_nonpes_aware_tpc(gr, gr->sm[sm].gpc, gr->sm[sm].tpc); + + nvkm_wr32(device, GPC_UNIT(gr->sm[sm].gpc, 0x0c10 + tpc * 4), sm); } gm200_grctx_generate_dist_skip_table(gr); gf100_gr_init_num_tpc_per_gpc(gr, true, true); } -static void +void tu102_gr_init_zcull(struct gf100_gr *gr) { struct nvkm_device *device = gr->base.engine.subdev.device; const u32 magicgpc918 = DIV_ROUND_UP(0x00800000, gr->tpc_total); - const u8 tile_nr = ALIGN(gr->tpc_total, 64); + const u8 tile_nr = gr->func->gpc_nr * gr->func->tpc_nr; u8 bank[GPC_MAX] = {}, gpc, i, j; u32 data; @@ -93,7 +94,7 @@ tu102_gr_init_gpc_mmu(struct gf100_gr *gr) static const struct gf100_gr_func tu102_gr = { .oneinit_tiles = gm200_gr_oneinit_tiles, - .oneinit_sm_id = gm200_gr_oneinit_sm_id, + .oneinit_sm_id = gv100_gr_oneinit_sm_id, .init = gf100_gr_init, .init_419bd8 = gv100_gr_init_419bd8, .init_gpc_mmu = tu102_gr_init_gpc_mmu, @@ -109,10 +110,14 @@ tu102_gr = { .init_ppc_exceptions = gk104_gr_init_ppc_exceptions, .init_504430 = gv100_gr_init_504430, .init_shader_exceptions = gv100_gr_init_shader_exceptions, + .init_rop_exceptions = gf100_gr_init_rop_exceptions, + .init_exception2 = gf100_gr_init_exception2, + .init_4188a4 = gv100_gr_init_4188a4, .trap_mp = gv100_gr_trap_mp, + .fecs.reset = gf100_gr_fecs_reset, .rops = gm200_gr_rops, .gpc_nr = 6, - .tpc_nr = 5, + .tpc_nr = 6, .ppc_nr = 3, .grctx = &tu102_grctx, .zbc = &gp102_gr_zbc, @@ -137,6 +142,7 @@ MODULE_FIRMWARE("nvidia/tu102/gr/sw_ctx.bin"); MODULE_FIRMWARE("nvidia/tu102/gr/sw_nonctx.bin"); MODULE_FIRMWARE("nvidia/tu102/gr/sw_bundle_init.bin"); MODULE_FIRMWARE("nvidia/tu102/gr/sw_method_init.bin"); +MODULE_FIRMWARE("nvidia/tu102/gr/sw_veid_bundle_init.bin"); MODULE_FIRMWARE("nvidia/tu104/gr/fecs_bl.bin"); MODULE_FIRMWARE("nvidia/tu104/gr/fecs_inst.bin"); @@ -150,6 +156,7 @@ MODULE_FIRMWARE("nvidia/tu104/gr/sw_ctx.bin"); MODULE_FIRMWARE("nvidia/tu104/gr/sw_nonctx.bin"); MODULE_FIRMWARE("nvidia/tu104/gr/sw_bundle_init.bin"); MODULE_FIRMWARE("nvidia/tu104/gr/sw_method_init.bin"); +MODULE_FIRMWARE("nvidia/tu104/gr/sw_veid_bundle_init.bin"); MODULE_FIRMWARE("nvidia/tu106/gr/fecs_bl.bin"); MODULE_FIRMWARE("nvidia/tu106/gr/fecs_inst.bin"); @@ -163,6 +170,7 @@ MODULE_FIRMWARE("nvidia/tu106/gr/sw_ctx.bin"); MODULE_FIRMWARE("nvidia/tu106/gr/sw_nonctx.bin"); MODULE_FIRMWARE("nvidia/tu106/gr/sw_bundle_init.bin"); MODULE_FIRMWARE("nvidia/tu106/gr/sw_method_init.bin"); +MODULE_FIRMWARE("nvidia/tu106/gr/sw_veid_bundle_init.bin"); MODULE_FIRMWARE("nvidia/tu117/gr/fecs_bl.bin"); MODULE_FIRMWARE("nvidia/tu117/gr/fecs_inst.bin"); @@ -176,6 +184,7 @@ MODULE_FIRMWARE("nvidia/tu117/gr/sw_ctx.bin"); MODULE_FIRMWARE("nvidia/tu117/gr/sw_nonctx.bin"); MODULE_FIRMWARE("nvidia/tu117/gr/sw_bundle_init.bin"); MODULE_FIRMWARE("nvidia/tu117/gr/sw_method_init.bin"); +MODULE_FIRMWARE("nvidia/tu117/gr/sw_veid_bundle_init.bin"); MODULE_FIRMWARE("nvidia/tu116/gr/fecs_bl.bin"); MODULE_FIRMWARE("nvidia/tu116/gr/fecs_inst.bin"); @@ -189,6 +198,26 @@ MODULE_FIRMWARE("nvidia/tu116/gr/sw_ctx.bin"); MODULE_FIRMWARE("nvidia/tu116/gr/sw_nonctx.bin"); MODULE_FIRMWARE("nvidia/tu116/gr/sw_bundle_init.bin"); MODULE_FIRMWARE("nvidia/tu116/gr/sw_method_init.bin"); +MODULE_FIRMWARE("nvidia/tu116/gr/sw_veid_bundle_init.bin"); + +int +tu102_gr_av_to_init_veid(struct nvkm_blob *blob, struct gf100_gr_pack **ppack) +{ + return gk20a_gr_av_to_init_(blob, 64, 0x00100000, ppack); +} + +int +tu102_gr_load(struct gf100_gr *gr, int ver, const struct gf100_gr_fwif *fwif) +{ + int ret; + + ret = gm200_gr_load(gr, ver, fwif); + if (ret) + return ret; + + return gk20a_gr_load_net(gr, "gr/", "sw_veid_bundle_init", ver, tu102_gr_av_to_init_veid, + &gr->bundle_veid); +} static const struct gf100_gr_fwif tu102_gr_fwif[] = { diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv31.c b/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv31.c index b1054db4c1b8..cb0c3991b2ad 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv31.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv31.c @@ -213,8 +213,8 @@ nv31_mpeg_intr(struct nvkm_engine *engine) if (show) { nvkm_error(subdev, "ch %d [%s] %08x %08x %08x %08x\n", - mpeg->chan ? mpeg->chan->fifo->chid : -1, - mpeg->chan ? mpeg->chan->object.client->name : + mpeg->chan ? mpeg->chan->fifo->id : -1, + mpeg->chan ? mpeg->chan->fifo->name : "unknown", stat, type, mthd, data); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv44.c b/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv44.c index 521ce43a2871..0890a279458e 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv44.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/mpeg/nv44.c @@ -182,8 +182,8 @@ nv44_mpeg_intr(struct nvkm_engine *engine) if (show) { nvkm_error(subdev, "ch %d [%08x %s] %08x %08x %08x %08x\n", - chan ? chan->fifo->chid : -1, inst << 4, - chan ? chan->object.client->name : "unknown", + chan ? chan->fifo->id : -1, inst << 4, + chan ? chan->fifo->name : "unknown", stat, type, mthd, data); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/Kbuild b/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/Kbuild index 9a0fd9812750..f05e79670d22 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/Kbuild @@ -1,3 +1,4 @@ # SPDX-License-Identifier: MIT nvkm-y += nvkm/engine/nvdec/base.o nvkm-y += nvkm/engine/nvdec/gm107.o +nvkm-y += nvkm/engine/nvdec/ga102.o diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/base.c index b0181cc5953b..1f6e3b32ba16 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/base.c @@ -37,7 +37,7 @@ nvkm_nvdec = { int nvkm_nvdec_new_(const struct nvkm_nvdec_fwif *fwif, struct nvkm_device *device, - enum nvkm_subdev_type type, int inst, struct nvkm_nvdec **pnvdec) + enum nvkm_subdev_type type, int inst, u32 addr, struct nvkm_nvdec **pnvdec) { struct nvkm_nvdec *nvdec; int ret; @@ -57,5 +57,5 @@ nvkm_nvdec_new_(const struct nvkm_nvdec_fwif *fwif, struct nvkm_device *device, nvdec->func = fwif->func; return nvkm_falcon_ctor(nvdec->func->flcn, &nvdec->engine.subdev, - nvdec->engine.subdev.name, 0, &nvdec->falcon); + nvdec->engine.subdev.name, addr, &nvdec->falcon); }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/ga102.c b/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/ga102.c new file mode 100644 index 000000000000..37d8c3c0f3ab --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/ga102.c @@ -0,0 +1,61 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +#include <subdev/mc.h> +#include <subdev/timer.h> + +static const struct nvkm_falcon_func +ga102_nvdec_flcn = { + .disable = gm200_flcn_disable, + .enable = gm200_flcn_enable, + .addr2 = 0x1c00, + .reset_pmc = true, + .reset_prep = ga102_flcn_reset_prep, + .reset_wait_mem_scrubbing = ga102_flcn_reset_wait_mem_scrubbing, + .imem_dma = &ga102_flcn_dma, + .dmem_dma = &ga102_flcn_dma, +}; + +static const struct nvkm_nvdec_func +ga102_nvdec = { + .flcn = &ga102_nvdec_flcn, +}; + +static int +ga102_nvdec_nofw(struct nvkm_nvdec *nvdec, int ver, const struct nvkm_nvdec_fwif *fwif) +{ + return 0; +} + +static const struct nvkm_nvdec_fwif +ga102_nvdec_fwif[] = { + { -1, ga102_nvdec_nofw, &ga102_nvdec }, + {} +}; + +int +ga102_nvdec_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, + struct nvkm_nvdec **pnvdec) +{ + return nvkm_nvdec_new_(ga102_nvdec_fwif, device, type, inst, 0x848000, pnvdec); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/gm107.c b/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/gm107.c index 8c44ce44a6d7..564f7e8960a2 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/gm107.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/gm107.c @@ -23,18 +23,13 @@ static const struct nvkm_falcon_func gm107_nvdec_flcn = { + .disable = gm200_flcn_disable, + .enable = gm200_flcn_enable, + .reset_pmc = true, + .reset_wait_mem_scrubbing = gm200_flcn_reset_wait_mem_scrubbing, .debug = 0xd00, - .fbif = 0x600, - .load_imem = nvkm_falcon_v1_load_imem, - .load_dmem = nvkm_falcon_v1_load_dmem, - .read_dmem = nvkm_falcon_v1_read_dmem, - .bind_context = nvkm_falcon_v1_bind_context, - .wait_for_halt = nvkm_falcon_v1_wait_for_halt, - .clear_interrupt = nvkm_falcon_v1_clear_interrupt, - .set_start_addr = nvkm_falcon_v1_set_start_addr, - .start = nvkm_falcon_v1_start, - .enable = nvkm_falcon_v1_enable, - .disable = nvkm_falcon_v1_disable, + .imem_pio = &gm200_flcn_imem_pio, + .dmem_pio = &gm200_flcn_dmem_pio, }; static const struct nvkm_nvdec_func @@ -59,5 +54,5 @@ int gm107_nvdec_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_nvdec **pnvdec) { - return nvkm_nvdec_new_(gm107_nvdec_fwif, device, type, inst, pnvdec); + return nvkm_nvdec_new_(gm107_nvdec_fwif, device, type, inst, 0, pnvdec); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/priv.h b/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/priv.h index 0920f6a887e2..61e1f7aaa509 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/nvdec/priv.h @@ -15,5 +15,5 @@ struct nvkm_nvdec_fwif { }; int nvkm_nvdec_new_(const struct nvkm_nvdec_fwif *fwif, struct nvkm_device *, - enum nvkm_subdev_type, int, struct nvkm_nvdec **); + enum nvkm_subdev_type, int, u32 addr, struct nvkm_nvdec **); #endif diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/nvenc/gm107.c b/drivers/gpu/drm/nouveau/nvkm/engine/nvenc/gm107.c index f44d41bf2034..ad27d8b97569 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/nvenc/gm107.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/nvenc/gm107.c @@ -24,17 +24,6 @@ static const struct nvkm_falcon_func gm107_nvenc_flcn = { - .fbif = 0x800, - .load_imem = nvkm_falcon_v1_load_imem, - .load_dmem = nvkm_falcon_v1_load_dmem, - .read_dmem = nvkm_falcon_v1_read_dmem, - .bind_context = nvkm_falcon_v1_bind_context, - .wait_for_halt = nvkm_falcon_v1_wait_for_halt, - .clear_interrupt = nvkm_falcon_v1_clear_interrupt, - .set_start_addr = nvkm_falcon_v1_set_start_addr, - .start = nvkm_falcon_v1_start, - .enable = nvkm_falcon_v1_enable, - .disable = nvkm_falcon_v1_disable, }; static const struct nvkm_nvenc_func diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sec/g98.c b/drivers/gpu/drm/nouveau/nvkm/engine/sec/g98.c index 1b87df03c823..c15b2cbf506b 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sec/g98.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sec/g98.c @@ -40,7 +40,7 @@ static const struct nvkm_enum g98_sec_isr_error_name[] = { }; static void -g98_sec_intr(struct nvkm_falcon *sec, struct nvkm_fifo_chan *chan) +g98_sec_intr(struct nvkm_falcon *sec, struct nvkm_chan *chan) { struct nvkm_subdev *subdev = &sec->engine.subdev; struct nvkm_device *device = subdev->device; @@ -54,9 +54,9 @@ g98_sec_intr(struct nvkm_falcon *sec, struct nvkm_fifo_chan *chan) nvkm_error(subdev, "DISPATCH_ERROR %04x [%s] ch %d [%010llx %s] " "subc %d mthd %04x data %08x\n", ssta, - en ? en->name : "UNKNOWN", chan ? chan->chid : -1, + en ? en->name : "UNKNOWN", chan ? chan->id : -1, chan ? chan->inst->addr : 0, - chan ? chan->object.client->name : "unknown", + chan ? chan->name : "unknown", subc, mthd, data); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sec2/Kbuild b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/Kbuild index 63cd2be3de08..19feadb1f67b 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sec2/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/Kbuild @@ -3,3 +3,4 @@ nvkm-y += nvkm/engine/sec2/base.o nvkm-y += nvkm/engine/sec2/gp102.o nvkm-y += nvkm/engine/sec2/gp108.o nvkm-y += nvkm/engine/sec2/tu102.o +nvkm-y += nvkm/engine/sec2/ga102.o diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sec2/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/base.c index 092c6d0b8e01..f2c60da5d1e8 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sec2/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/base.c @@ -22,53 +22,99 @@ #include "priv.h" #include <core/firmware.h> -#include <subdev/top.h> +#include <subdev/mc.h> +#include <subdev/timer.h> -static void -nvkm_sec2_recv(struct work_struct *work) +#include <nvfw/sec2.h> + +static int +nvkm_sec2_finimsg(void *priv, struct nvfw_falcon_msg *hdr) +{ + struct nvkm_sec2 *sec2 = priv; + + atomic_set(&sec2->running, 0); + return 0; +} + +static int +nvkm_sec2_fini(struct nvkm_engine *engine, bool suspend) { - struct nvkm_sec2 *sec2 = container_of(work, typeof(*sec2), work); + struct nvkm_sec2 *sec2 = nvkm_sec2(engine); + struct nvkm_subdev *subdev = &sec2->engine.subdev; + struct nvkm_falcon *falcon = &sec2->falcon; + struct nvkm_falcon_cmdq *cmdq = sec2->cmdq; + struct nvfw_falcon_cmd cmd = { + .unit_id = sec2->func->unit_unload, + .size = sizeof(cmd), + }; + int ret; - if (!sec2->initmsg_received) { - int ret = sec2->func->initmsg(sec2); - if (ret) { - nvkm_error(&sec2->engine.subdev, - "error parsing init message: %d\n", ret); - return; - } + if (!subdev->use.enabled) + return 0; - sec2->initmsg_received = true; + if (atomic_read(&sec2->initmsg) == 1) { + ret = nvkm_falcon_cmdq_send(cmdq, &cmd, nvkm_sec2_finimsg, sec2, + msecs_to_jiffies(1000)); + WARN_ON(ret); + + nvkm_msec(subdev->device, 2000, + if (nvkm_falcon_rd32(falcon, 0x100) & 0x00000010) + break; + ); } - nvkm_falcon_msgq_recv(sec2->msgq); + nvkm_inth_block(&subdev->inth); + + nvkm_falcon_cmdq_fini(cmdq); + falcon->func->disable(falcon); + nvkm_falcon_put(falcon, subdev); + return 0; } -static void -nvkm_sec2_intr(struct nvkm_engine *engine) +static int +nvkm_sec2_init(struct nvkm_engine *engine) { struct nvkm_sec2 *sec2 = nvkm_sec2(engine); - sec2->func->intr(sec2); + struct nvkm_subdev *subdev = &sec2->engine.subdev; + struct nvkm_falcon *falcon = &sec2->falcon; + int ret; + + ret = nvkm_falcon_get(falcon, subdev); + if (ret) + return ret; + + nvkm_falcon_wr32(falcon, 0x014, 0xffffffff); + atomic_set(&sec2->initmsg, 0); + atomic_set(&sec2->running, 1); + nvkm_inth_allow(&subdev->inth); + + nvkm_falcon_start(falcon); + return 0; } static int -nvkm_sec2_fini(struct nvkm_engine *engine, bool suspend) +nvkm_sec2_oneinit(struct nvkm_engine *engine) { struct nvkm_sec2 *sec2 = nvkm_sec2(engine); - - flush_work(&sec2->work); - - if (suspend) { - nvkm_falcon_cmdq_fini(sec2->cmdq); - sec2->initmsg_received = false; + struct nvkm_subdev *subdev = &sec2->engine.subdev; + struct nvkm_intr *intr = &sec2->engine.subdev.device->mc->intr; + enum nvkm_intr_type type = NVKM_INTR_SUBDEV; + + if (sec2->func->intr_vector) { + intr = sec2->func->intr_vector(sec2, &type); + if (IS_ERR(intr)) + return PTR_ERR(intr); } - return 0; + return nvkm_inth_add(intr, type, NVKM_INTR_PRIO_NORMAL, subdev, sec2->func->intr, + &subdev->inth); } static void * nvkm_sec2_dtor(struct nvkm_engine *engine) { struct nvkm_sec2 *sec2 = nvkm_sec2(engine); + nvkm_falcon_msgq_del(&sec2->msgq); nvkm_falcon_cmdq_del(&sec2->cmdq); nvkm_falcon_qmgr_del(&sec2->qmgr); @@ -79,8 +125,9 @@ nvkm_sec2_dtor(struct nvkm_engine *engine) static const struct nvkm_engine_func nvkm_sec2 = { .dtor = nvkm_sec2_dtor, + .oneinit = nvkm_sec2_oneinit, + .init = nvkm_sec2_init, .fini = nvkm_sec2_fini, - .intr = nvkm_sec2_intr, }; int @@ -113,6 +160,5 @@ nvkm_sec2_new_(const struct nvkm_sec2_fwif *fwif, struct nvkm_device *device, (ret = nvkm_falcon_msgq_new(sec2->qmgr, "msgq", &sec2->msgq))) return ret; - INIT_WORK(&sec2->work, nvkm_sec2_recv); return 0; }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sec2/ga102.c b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/ga102.c new file mode 100644 index 000000000000..945abb8156d7 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/ga102.c @@ -0,0 +1,197 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" +#include <subdev/acr.h> +#include <subdev/vfn.h> + +#include <nvfw/flcn.h> +#include <nvfw/sec2.h> + +static int +ga102_sec2_initmsg(struct nvkm_sec2 *sec2) +{ + struct nv_sec2_init_msg_v1 msg; + int ret, i; + + ret = nvkm_falcon_msgq_recv_initmsg(sec2->msgq, &msg, sizeof(msg)); + if (ret) + return ret; + + if (msg.hdr.unit_id != NV_SEC2_UNIT_INIT || + msg.msg_type != NV_SEC2_INIT_MSG_INIT) + return -EINVAL; + + for (i = 0; i < ARRAY_SIZE(msg.queue_info); i++) { + if (msg.queue_info[i].id == NV_SEC2_INIT_MSG_QUEUE_ID_MSGQ) { + nvkm_falcon_msgq_init(sec2->msgq, msg.queue_info[i].index, + msg.queue_info[i].offset, + msg.queue_info[i].size); + } else { + nvkm_falcon_cmdq_init(sec2->cmdq, msg.queue_info[i].index, + msg.queue_info[i].offset, + msg.queue_info[i].size); + } + } + + return 0; +} + +static struct nvkm_intr * +ga102_sec2_intr_vector(struct nvkm_sec2 *sec2, enum nvkm_intr_type *pvector) +{ + struct nvkm_device *device = sec2->engine.subdev.device; + struct nvkm_falcon *falcon = &sec2->falcon; + int ret; + + ret = ga102_flcn_select(falcon); + if (ret) + return ERR_PTR(ret); + + *pvector = nvkm_rd32(device, 0x8403e0) & 0x000000ff; + return &device->vfn->intr; +} + +static int +ga102_sec2_acr_bootstrap_falcon_callback(void *priv, struct nvfw_falcon_msg *hdr) +{ + struct nv_sec2_acr_bootstrap_falcon_msg_v1 *msg = + container_of(hdr, typeof(*msg), msg.hdr); + struct nvkm_subdev *subdev = priv; + const char *name = nvkm_acr_lsf_id(msg->falcon_id); + + if (msg->error_code) { + nvkm_error(subdev, "ACR_BOOTSTRAP_FALCON failed for falcon %d [%s]: %08x %08x\n", + msg->falcon_id, name, msg->error_code, msg->unkn08); + return -EINVAL; + } + + nvkm_debug(subdev, "%s booted\n", name); + return 0; +} + +static int +ga102_sec2_acr_bootstrap_falcon(struct nvkm_falcon *falcon, enum nvkm_acr_lsf_id id) +{ + struct nvkm_sec2 *sec2 = container_of(falcon, typeof(*sec2), falcon); + struct nv_sec2_acr_bootstrap_falcon_cmd_v1 cmd = { + .cmd.hdr.unit_id = sec2->func->unit_acr, + .cmd.hdr.size = sizeof(cmd), + .cmd.cmd_type = NV_SEC2_ACR_CMD_BOOTSTRAP_FALCON, + .flags = NV_SEC2_ACR_BOOTSTRAP_FALCON_FLAGS_RESET_YES, + .falcon_id = id, + }; + + return nvkm_falcon_cmdq_send(sec2->cmdq, &cmd.cmd.hdr, + ga102_sec2_acr_bootstrap_falcon_callback, + &sec2->engine.subdev, + msecs_to_jiffies(1000)); +} + +static const struct nvkm_acr_lsf_func +ga102_sec2_acr_0 = { + .bld_size = sizeof(struct flcn_bl_dmem_desc_v2), + .bld_write = gp102_sec2_acr_bld_write_1, + .bld_patch = gp102_sec2_acr_bld_patch_1, + .bootstrap_falcons = BIT_ULL(NVKM_ACR_LSF_FECS) | + BIT_ULL(NVKM_ACR_LSF_GPCCS) | + BIT_ULL(NVKM_ACR_LSF_SEC2), + .bootstrap_falcon = ga102_sec2_acr_bootstrap_falcon, +}; + +static const struct nvkm_falcon_func +ga102_sec2_flcn = { + .disable = gm200_flcn_disable, + .enable = gm200_flcn_enable, + .select = ga102_flcn_select, + .addr2 = 0x1000, + .reset_pmc = true, + .reset_eng = gp102_flcn_reset_eng, + .reset_prep = ga102_flcn_reset_prep, + .reset_wait_mem_scrubbing = ga102_flcn_reset_wait_mem_scrubbing, + .imem_dma = &ga102_flcn_dma, + .dmem_pio = &gm200_flcn_dmem_pio, + .dmem_dma = &ga102_flcn_dma, + .emem_addr = 0x01000000, + .emem_pio = &gp102_flcn_emem_pio, + .start = nvkm_falcon_v1_start, + .cmdq = { 0xc00, 0xc04, 8 }, + .msgq = { 0xc80, 0xc84, 8 }, +}; + +static const struct nvkm_sec2_func +ga102_sec2 = { + .flcn = &ga102_sec2_flcn, + .intr_vector = ga102_sec2_intr_vector, + .intr = gp102_sec2_intr, + .initmsg = ga102_sec2_initmsg, + .unit_acr = NV_SEC2_UNIT_V2_ACR, + .unit_unload = NV_SEC2_UNIT_V2_UNLOAD, +}; + +MODULE_FIRMWARE("nvidia/ga102/sec2/desc.bin"); +MODULE_FIRMWARE("nvidia/ga102/sec2/image.bin"); +MODULE_FIRMWARE("nvidia/ga102/sec2/sig.bin"); +MODULE_FIRMWARE("nvidia/ga102/sec2/hs_bl_sig.bin"); + +MODULE_FIRMWARE("nvidia/ga103/sec2/desc.bin"); +MODULE_FIRMWARE("nvidia/ga103/sec2/image.bin"); +MODULE_FIRMWARE("nvidia/ga103/sec2/sig.bin"); +MODULE_FIRMWARE("nvidia/ga103/sec2/hs_bl_sig.bin"); + +MODULE_FIRMWARE("nvidia/ga104/sec2/desc.bin"); +MODULE_FIRMWARE("nvidia/ga104/sec2/image.bin"); +MODULE_FIRMWARE("nvidia/ga104/sec2/sig.bin"); +MODULE_FIRMWARE("nvidia/ga104/sec2/hs_bl_sig.bin"); + +MODULE_FIRMWARE("nvidia/ga106/sec2/desc.bin"); +MODULE_FIRMWARE("nvidia/ga106/sec2/image.bin"); +MODULE_FIRMWARE("nvidia/ga106/sec2/sig.bin"); +MODULE_FIRMWARE("nvidia/ga106/sec2/hs_bl_sig.bin"); + +MODULE_FIRMWARE("nvidia/ga107/sec2/desc.bin"); +MODULE_FIRMWARE("nvidia/ga107/sec2/image.bin"); +MODULE_FIRMWARE("nvidia/ga107/sec2/sig.bin"); +MODULE_FIRMWARE("nvidia/ga107/sec2/hs_bl_sig.bin"); + +static int +ga102_sec2_load(struct nvkm_sec2 *sec2, int ver, + const struct nvkm_sec2_fwif *fwif) +{ + return nvkm_acr_lsfw_load_sig_image_desc_v2(&sec2->engine.subdev, &sec2->falcon, + NVKM_ACR_LSF_SEC2, "sec2/", ver, fwif->acr); +} + +static const struct nvkm_sec2_fwif +ga102_sec2_fwif[] = { + { 0, ga102_sec2_load, &ga102_sec2, &ga102_sec2_acr_0 }, + { -1, gp102_sec2_nofw, &ga102_sec2 } +}; + +int +ga102_sec2_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, + struct nvkm_sec2 **psec2) +{ + /* TOP info wasn't updated on Turing to reflect the PRI + * address change for some reason. We override it here. + */ + return nvkm_sec2_new_(ga102_sec2_fwif, device, type, inst, 0x840000, psec2); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sec2/gp102.c b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/gp102.c index 44e39f5743d5..c64013d10500 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sec2/gp102.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/gp102.c @@ -74,16 +74,6 @@ gp102_sec2_acr_bootstrap_falcon(struct nvkm_falcon *falcon, msecs_to_jiffies(1000)); } -static int -gp102_sec2_acr_boot(struct nvkm_falcon *falcon) -{ - struct nv_sec2_args args = {}; - nvkm_falcon_load_dmem(falcon, &args, - falcon->func->emem_addr, sizeof(args), 0); - nvkm_falcon_start(falcon); - return 0; -} - static void gp102_sec2_acr_bld_patch(struct nvkm_acr *acr, u32 bld, s64 adjust) { @@ -122,7 +112,6 @@ gp102_sec2_acr_0 = { .bld_size = sizeof(struct loader_config_v1), .bld_write = gp102_sec2_acr_bld_write, .bld_patch = gp102_sec2_acr_bld_patch, - .boot = gp102_sec2_acr_boot, .bootstrap_falcons = BIT_ULL(NVKM_ACR_LSF_FECS) | BIT_ULL(NVKM_ACR_LSF_GPCCS) | BIT_ULL(NVKM_ACR_LSF_SEC2), @@ -160,89 +149,68 @@ gp102_sec2_initmsg(struct nvkm_sec2 *sec2) return 0; } -void -gp102_sec2_intr(struct nvkm_sec2 *sec2) +irqreturn_t +gp102_sec2_intr(struct nvkm_inth *inth) { + struct nvkm_sec2 *sec2 = container_of(inth, typeof(*sec2), engine.subdev.inth); struct nvkm_subdev *subdev = &sec2->engine.subdev; struct nvkm_falcon *falcon = &sec2->falcon; u32 disp = nvkm_falcon_rd32(falcon, 0x01c); u32 intr = nvkm_falcon_rd32(falcon, 0x008) & disp & ~(disp >> 16); if (intr & 0x00000040) { - schedule_work(&sec2->work); + if (unlikely(atomic_read(&sec2->initmsg) == 0)) { + int ret = sec2->func->initmsg(sec2); + + if (ret) + nvkm_error(subdev, "error parsing init message: %d\n", ret); + + atomic_set(&sec2->initmsg, ret ?: 1); + } + + if (atomic_read(&sec2->initmsg) > 0) { + if (!nvkm_falcon_msgq_empty(sec2->msgq)) + nvkm_falcon_msgq_recv(sec2->msgq); + } + nvkm_falcon_wr32(falcon, 0x004, 0x00000040); intr &= ~0x00000040; } + if (intr & 0x00000010) { + if (atomic_read(&sec2->running)) { + FLCN_ERR(falcon, "halted"); + gm200_flcn_tracepc(falcon); + } + + nvkm_falcon_wr32(falcon, 0x004, 0x00000010); + intr &= ~0x00000010; + } + if (intr) { nvkm_error(subdev, "unhandled intr %08x\n", intr); nvkm_falcon_wr32(falcon, 0x004, intr); } -} -int -gp102_sec2_flcn_enable(struct nvkm_falcon *falcon) -{ - nvkm_falcon_mask(falcon, 0x3c0, 0x00000001, 0x00000001); - udelay(10); - nvkm_falcon_mask(falcon, 0x3c0, 0x00000001, 0x00000000); - return nvkm_falcon_v1_enable(falcon); -} - -void -gp102_sec2_flcn_bind_context(struct nvkm_falcon *falcon, - struct nvkm_memory *ctx) -{ - struct nvkm_device *device = falcon->owner->device; - - nvkm_falcon_v1_bind_context(falcon, ctx); - if (!ctx) - return; - - /* Not sure if this is a WAR for a HW issue, or some additional - * programming sequence that's needed to properly complete the - * context switch we trigger above. - * - * Fixes unreliability of booting the SEC2 RTOS on Quadro P620, - * particularly when resuming from suspend. - * - * Also removes the need for an odd workaround where we needed - * to program SEC2's FALCON_CPUCTL_ALIAS_STARTCPU twice before - * the SEC2 RTOS would begin executing. - */ - nvkm_msec(device, 10, - u32 irqstat = nvkm_falcon_rd32(falcon, 0x008); - u32 flcn0dc = nvkm_falcon_rd32(falcon, 0x0dc); - if ((irqstat & 0x00000008) && - (flcn0dc & 0x00007000) == 0x00005000) - break; - ); - - nvkm_falcon_mask(falcon, 0x004, 0x00000008, 0x00000008); - nvkm_falcon_mask(falcon, 0x058, 0x00000002, 0x00000002); - - nvkm_msec(device, 10, - u32 flcn0dc = nvkm_falcon_rd32(falcon, 0x0dc); - if ((flcn0dc & 0x00007000) == 0x00000000) - break; - ); + return IRQ_HANDLED; } static const struct nvkm_falcon_func gp102_sec2_flcn = { + .disable = gm200_flcn_disable, + .enable = gm200_flcn_enable, + .reset_pmc = true, + .reset_eng = gp102_flcn_reset_eng, + .reset_wait_mem_scrubbing = gm200_flcn_reset_wait_mem_scrubbing, .debug = 0x408, - .fbif = 0x600, - .load_imem = nvkm_falcon_v1_load_imem, - .load_dmem = nvkm_falcon_v1_load_dmem, - .read_dmem = nvkm_falcon_v1_read_dmem, + .bind_inst = gm200_flcn_bind_inst, + .bind_stat = gm200_flcn_bind_stat, + .bind_intr = true, + .imem_pio = &gm200_flcn_imem_pio, + .dmem_pio = &gm200_flcn_dmem_pio, .emem_addr = 0x01000000, - .bind_context = gp102_sec2_flcn_bind_context, - .wait_for_halt = nvkm_falcon_v1_wait_for_halt, - .clear_interrupt = nvkm_falcon_v1_clear_interrupt, - .set_start_addr = nvkm_falcon_v1_set_start_addr, + .emem_pio = &gp102_flcn_emem_pio, .start = nvkm_falcon_v1_start, - .enable = gp102_sec2_flcn_enable, - .disable = nvkm_falcon_v1_disable, .cmdq = { 0xa00, 0xa04, 8 }, .msgq = { 0xa30, 0xa34, 8 }, }; @@ -250,6 +218,7 @@ gp102_sec2_flcn = { const struct nvkm_sec2_func gp102_sec2 = { .flcn = &gp102_sec2_flcn, + .unit_unload = NV_SEC2_UNIT_UNLOAD, .unit_acr = NV_SEC2_UNIT_ACR, .intr = gp102_sec2_intr, .initmsg = gp102_sec2_initmsg, @@ -268,7 +237,7 @@ MODULE_FIRMWARE("nvidia/gp107/sec2/desc.bin"); MODULE_FIRMWARE("nvidia/gp107/sec2/image.bin"); MODULE_FIRMWARE("nvidia/gp107/sec2/sig.bin"); -static void +void gp102_sec2_acr_bld_patch_1(struct nvkm_acr *acr, u32 bld, s64 adjust) { struct flcn_bl_dmem_desc_v2 hdr; @@ -279,7 +248,7 @@ gp102_sec2_acr_bld_patch_1(struct nvkm_acr *acr, u32 bld, s64 adjust) flcn_bl_dmem_desc_v2_dump(&acr->subdev, &hdr); } -static void +void gp102_sec2_acr_bld_write_1(struct nvkm_acr *acr, u32 bld, struct nvkm_acr_lsfw *lsfw) { @@ -304,7 +273,6 @@ gp102_sec2_acr_1 = { .bld_size = sizeof(struct flcn_bl_dmem_desc_v2), .bld_write = gp102_sec2_acr_bld_write_1, .bld_patch = gp102_sec2_acr_bld_patch_1, - .boot = gp102_sec2_acr_boot, .bootstrap_falcons = BIT_ULL(NVKM_ACR_LSF_FECS) | BIT_ULL(NVKM_ACR_LSF_GPCCS) | BIT_ULL(NVKM_ACR_LSF_SEC2), diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sec2/priv.h b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/priv.h index af19229e885d..172d2705c199 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sec2/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/priv.h @@ -2,15 +2,18 @@ #ifndef __NVKM_SEC2_PRIV_H__ #define __NVKM_SEC2_PRIV_H__ #include <engine/sec2.h> +struct nvkm_acr_lsfw; struct nvkm_sec2_func { const struct nvkm_falcon_func *flcn; + u8 unit_unload; u8 unit_acr; - void (*intr)(struct nvkm_sec2 *); + struct nvkm_intr *(*intr_vector)(struct nvkm_sec2 *, enum nvkm_intr_type *); + irqreturn_t (*intr)(struct nvkm_inth *); int (*initmsg)(struct nvkm_sec2 *); }; -void gp102_sec2_intr(struct nvkm_sec2 *); +irqreturn_t gp102_sec2_intr(struct nvkm_inth *); int gp102_sec2_initmsg(struct nvkm_sec2 *); struct nvkm_sec2_fwif { @@ -24,6 +27,8 @@ int gp102_sec2_nofw(struct nvkm_sec2 *, int, const struct nvkm_sec2_fwif *); int gp102_sec2_load(struct nvkm_sec2 *, int, const struct nvkm_sec2_fwif *); extern const struct nvkm_sec2_func gp102_sec2; extern const struct nvkm_acr_lsf_func gp102_sec2_acr_1; +void gp102_sec2_acr_bld_write_1(struct nvkm_acr *, u32, struct nvkm_acr_lsfw *); +void gp102_sec2_acr_bld_patch_1(struct nvkm_acr *, u32, s64); int nvkm_sec2_new_(const struct nvkm_sec2_fwif *, struct nvkm_device *, enum nvkm_subdev_type, int, u32 addr, struct nvkm_sec2 **); diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sec2/tu102.c b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/tu102.c index f3faeb705575..0afc4b2fa529 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sec2/tu102.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sec2/tu102.c @@ -22,21 +22,24 @@ #include "priv.h" #include <subdev/acr.h> +#include <nvfw/sec2.h> + static const struct nvkm_falcon_func tu102_sec2_flcn = { + .disable = gm200_flcn_disable, + .enable = gm200_flcn_enable, + .reset_pmc = true, + .reset_eng = gp102_flcn_reset_eng, + .reset_wait_mem_scrubbing = gm200_flcn_reset_wait_mem_scrubbing, .debug = 0x408, - .fbif = 0x600, - .load_imem = nvkm_falcon_v1_load_imem, - .load_dmem = nvkm_falcon_v1_load_dmem, - .read_dmem = nvkm_falcon_v1_read_dmem, + .bind_inst = gm200_flcn_bind_inst, + .bind_stat = gm200_flcn_bind_stat, + .bind_intr = true, + .imem_pio = &gm200_flcn_imem_pio, + .dmem_pio = &gm200_flcn_dmem_pio, .emem_addr = 0x01000000, - .bind_context = gp102_sec2_flcn_bind_context, - .wait_for_halt = nvkm_falcon_v1_wait_for_halt, - .clear_interrupt = nvkm_falcon_v1_clear_interrupt, - .set_start_addr = nvkm_falcon_v1_set_start_addr, + .emem_pio = &gp102_flcn_emem_pio, .start = nvkm_falcon_v1_start, - .enable = nvkm_falcon_v1_enable, - .disable = nvkm_falcon_v1_disable, .cmdq = { 0xc00, 0xc04, 8 }, .msgq = { 0xc80, 0xc84, 8 }, }; @@ -44,7 +47,8 @@ tu102_sec2_flcn = { static const struct nvkm_sec2_func tu102_sec2 = { .flcn = &tu102_sec2_flcn, - .unit_acr = 0x07, + .unit_unload = NV_SEC2_UNIT_V2_UNLOAD, + .unit_acr = NV_SEC2_UNIT_V2_ACR, .intr = gp102_sec2_intr, .initmsg = gp102_sec2_initmsg, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sw/base.c b/drivers/gpu/drm/nouveau/nvkm/engine/sw/base.c index 14871d0bd746..a9d464db6974 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sw/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sw/base.c @@ -35,7 +35,7 @@ nvkm_sw_mthd(struct nvkm_sw *sw, int chid, int subc, u32 mthd, u32 data) spin_lock_irqsave(&sw->engine.lock, flags); list_for_each_entry(chan, &sw->chan, head) { - if (chan->fifo->chid == chid) { + if (chan->fifo->id == chid) { handled = nvkm_sw_chan_mthd(chan, subc, mthd, data); list_del(&chan->head); list_add(&chan->head, &sw->chan); diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sw/chan.c b/drivers/gpu/drm/nouveau/nvkm/engine/sw/chan.c index f28967065639..834b8cbed51d 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sw/chan.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sw/chan.c @@ -23,7 +23,6 @@ */ #include "chan.h" -#include <core/notify.h> #include <engine/fifo.h> #include <nvif/event.h> @@ -36,7 +35,7 @@ nvkm_sw_chan_mthd(struct nvkm_sw_chan *chan, int subc, u32 mthd, u32 data) case 0x0000: return true; case 0x0500: - nvkm_event_send(&chan->event, 1, 0, NULL, 0); + nvkm_event_ntfy(&chan->event, 0, NVKM_SW_CHAN_EVENT_PAGE_FLIP); return true; default: if (chan->func->mthd) @@ -46,27 +45,8 @@ nvkm_sw_chan_mthd(struct nvkm_sw_chan *chan, int subc, u32 mthd, u32 data) return false; } -static int -nvkm_sw_chan_event_ctor(struct nvkm_object *object, void *data, u32 size, - struct nvkm_notify *notify) -{ - union { - struct nvif_notify_uevent_req none; - } *req = data; - int ret = -ENOSYS; - - if (!(ret = nvif_unvers(ret, &data, &size, req->none))) { - notify->size = sizeof(struct nvif_notify_uevent_rep); - notify->types = 1; - notify->index = 0; - } - - return ret; -} - static const struct nvkm_event_func nvkm_sw_chan_event = { - .ctor = nvkm_sw_chan_event_ctor, }; static void * @@ -107,5 +87,5 @@ nvkm_sw_chan_ctor(const struct nvkm_sw_chan_func *func, struct nvkm_sw *sw, list_add(&chan->head, &sw->chan); spin_unlock_irqrestore(&sw->engine.lock, flags); - return nvkm_event_init(&nvkm_sw_chan_event, 1, 1, &chan->event); + return nvkm_event_init(&nvkm_sw_chan_event, &sw->engine.subdev, 1, 1, &chan->event); } diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sw/chan.h b/drivers/gpu/drm/nouveau/nvkm/engine/sw/chan.h index 32de53427aa4..67b2e5ea93d9 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sw/chan.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sw/chan.h @@ -14,6 +14,7 @@ struct nvkm_sw_chan { struct nvkm_fifo_chan *fifo; struct list_head head; +#define NVKM_SW_CHAN_EVENT_PAGE_FLIP BIT(0) struct nvkm_event event; }; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sw/gf100.c b/drivers/gpu/drm/nouveau/nvkm/engine/sw/gf100.c index 55abf839f29d..c3cf6f2ff86c 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sw/gf100.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sw/gf100.c @@ -36,10 +36,10 @@ ******************************************************************************/ static int -gf100_sw_chan_vblsem_release(struct nvkm_notify *notify) +gf100_sw_chan_vblsem_release(struct nvkm_event_ntfy *notify, u32 bits) { struct nv50_sw_chan *chan = - container_of(notify, typeof(*chan), vblank.notify[notify->index]); + container_of(notify, typeof(*chan), vblank.notify[notify->id]); struct nvkm_sw *sw = chan->base.sw; struct nvkm_device *device = sw->engine.subdev.device; u32 inst = chan->base.fifo->inst->addr >> 12; @@ -50,7 +50,7 @@ gf100_sw_chan_vblsem_release(struct nvkm_notify *notify) nvkm_wr32(device, 0x060010, lower_32_bits(chan->vblank.offset)); nvkm_wr32(device, 0x060014, chan->vblank.value); - return NVKM_NOTIFY_DROP; + return NVKM_EVENT_DROP; } static bool @@ -73,7 +73,7 @@ gf100_sw_chan_mthd(struct nvkm_sw_chan *base, int subc, u32 mthd, u32 data) return true; case 0x040c: if (data < device->disp->vblank.index_nr) { - nvkm_notify_get(&chan->vblank.notify[data]); + nvkm_event_ntfy_allow(&chan->vblank.notify[data]); return true; } break; @@ -120,16 +120,8 @@ gf100_sw_chan_new(struct nvkm_sw *sw, struct nvkm_fifo_chan *fifoch, return ret; for (i = 0; disp && i < disp->vblank.index_nr; i++) { - ret = nvkm_notify_init(NULL, &disp->vblank, - gf100_sw_chan_vblsem_release, false, - &(struct nvif_notify_head_req_v0) { - .head = i, - }, - sizeof(struct nvif_notify_head_req_v0), - sizeof(struct nvif_notify_head_rep_v0), - &chan->vblank.notify[i]); - if (ret) - return ret; + nvkm_event_ntfy_add(&disp->vblank, i, NVKM_DISP_HEAD_EVENT_VBLANK, true, + gf100_sw_chan_vblsem_release, &chan->vblank.notify[i]); } return 0; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sw/nv50.c b/drivers/gpu/drm/nouveau/nvkm/engine/sw/nv50.c index 1fdd094c8b7e..9d7a9b7d5be3 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sw/nv50.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sw/nv50.c @@ -36,10 +36,10 @@ ******************************************************************************/ static int -nv50_sw_chan_vblsem_release(struct nvkm_notify *notify) +nv50_sw_chan_vblsem_release(struct nvkm_event_ntfy *notify, u32 bits) { struct nv50_sw_chan *chan = - container_of(notify, typeof(*chan), vblank.notify[notify->index]); + container_of(notify, typeof(*chan), vblank.notify[notify->id]); struct nvkm_sw *sw = chan->base.sw; struct nvkm_device *device = sw->engine.subdev.device; @@ -55,7 +55,7 @@ nv50_sw_chan_vblsem_release(struct nvkm_notify *notify) nvkm_wr32(device, 0x060014, chan->vblank.value); } - return NVKM_NOTIFY_DROP; + return NVKM_EVENT_DROP; } static bool @@ -70,7 +70,7 @@ nv50_sw_chan_mthd(struct nvkm_sw_chan *base, int subc, u32 mthd, u32 data) case 0x0404: chan->vblank.value = data; return true; case 0x0408: if (data < device->disp->vblank.index_nr) { - nvkm_notify_get(&chan->vblank.notify[data]); + nvkm_event_ntfy_allow(&chan->vblank.notify[data]); return true; } break; @@ -85,8 +85,10 @@ nv50_sw_chan_dtor(struct nvkm_sw_chan *base) { struct nv50_sw_chan *chan = nv50_sw_chan(base); int i; + for (i = 0; i < ARRAY_SIZE(chan->vblank.notify); i++) - nvkm_notify_fini(&chan->vblank.notify[i]); + nvkm_event_ntfy_del(&chan->vblank.notify[i]); + return chan; } @@ -113,16 +115,8 @@ nv50_sw_chan_new(struct nvkm_sw *sw, struct nvkm_fifo_chan *fifoch, return ret; for (i = 0; disp && i < disp->vblank.index_nr; i++) { - ret = nvkm_notify_init(NULL, &disp->vblank, - nv50_sw_chan_vblsem_release, false, - &(struct nvif_notify_head_req_v0) { - .head = i, - }, - sizeof(struct nvif_notify_head_req_v0), - sizeof(struct nvif_notify_head_rep_v0), - &chan->vblank.notify[i]); - if (ret) - return ret; + nvkm_event_ntfy_add(&disp->vblank, i, NVKM_DISP_HEAD_EVENT_VBLANK, true, + nv50_sw_chan_vblsem_release, &chan->vblank.notify[i]); } return 0; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sw/nv50.h b/drivers/gpu/drm/nouveau/nvkm/engine/sw/nv50.h index 6d364d7b406a..b42289ce8826 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sw/nv50.h +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sw/nv50.h @@ -5,12 +5,12 @@ #include "priv.h" #include "chan.h" #include "nvsw.h" -#include <core/notify.h> +#include <core/event.h> struct nv50_sw_chan { struct nvkm_sw_chan base; struct { - struct nvkm_notify notify[4]; + struct nvkm_event_ntfy notify[4]; u32 ctxdma; u64 offset; u32 value; diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/sw/nvsw.c b/drivers/gpu/drm/nouveau/nvkm/engine/sw/nvsw.c index 33dd03fff3c4..f5affa1c8f34 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/sw/nvsw.c +++ b/drivers/gpu/drm/nouveau/nvkm/engine/sw/nvsw.c @@ -27,33 +27,34 @@ #include <nvif/if0004.h> static int -nvkm_nvsw_mthd_(struct nvkm_object *object, u32 mthd, void *data, u32 size) +nvkm_nvsw_uevent(struct nvkm_object *object, void *argv, u32 argc, struct nvkm_uevent *uevent) { - struct nvkm_nvsw *nvsw = nvkm_nvsw(object); - if (nvsw->func->mthd) - return nvsw->func->mthd(nvsw, mthd, data, size); - return -ENODEV; + union nv04_nvsw_event_args *args = argv; + + if (!uevent) + return 0; + if (argc != sizeof(args->vn)) + return -ENOSYS; + + return nvkm_uevent_add(uevent, &nvkm_nvsw(object)->chan->event, 0, + NVKM_SW_CHAN_EVENT_PAGE_FLIP, NULL); } static int -nvkm_nvsw_ntfy_(struct nvkm_object *object, u32 mthd, - struct nvkm_event **pevent) +nvkm_nvsw_mthd(struct nvkm_object *object, u32 mthd, void *data, u32 size) { struct nvkm_nvsw *nvsw = nvkm_nvsw(object); - switch (mthd) { - case NV04_NVSW_NTFY_UEVENT: - *pevent = &nvsw->chan->event; - return 0; - default: - break; - } - return -EINVAL; + + if (nvsw->func->mthd) + return nvsw->func->mthd(nvsw, mthd, data, size); + + return -ENODEV; } static const struct nvkm_object_func nvkm_nvsw_ = { - .mthd = nvkm_nvsw_mthd_, - .ntfy = nvkm_nvsw_ntfy_, + .mthd = nvkm_nvsw_mthd, + .uevent = nvkm_nvsw_uevent, }; int diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/Kbuild b/drivers/gpu/drm/nouveau/nvkm/falcon/Kbuild index d79d783904ee..9ffe7b921ccb 100644 --- a/drivers/gpu/drm/nouveau/nvkm/falcon/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/Kbuild @@ -1,6 +1,12 @@ # SPDX-License-Identifier: MIT nvkm-y += nvkm/falcon/base.o nvkm-y += nvkm/falcon/cmdq.o +nvkm-y += nvkm/falcon/fw.o nvkm-y += nvkm/falcon/msgq.o nvkm-y += nvkm/falcon/qmgr.o nvkm-y += nvkm/falcon/v1.o + +nvkm-y += nvkm/falcon/gm200.o +nvkm-y += nvkm/falcon/gp102.o +nvkm-y += nvkm/falcon/ga100.o +nvkm-y += nvkm/falcon/ga102.o diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/base.c b/drivers/gpu/drm/nouveau/nvkm/falcon/base.c index f3f90c1063dd..235149f73a69 100644 --- a/drivers/gpu/drm/nouveau/nvkm/falcon/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/base.c @@ -22,119 +22,217 @@ #include "priv.h" #include <subdev/mc.h> +#include <subdev/timer.h> #include <subdev/top.h> -void -nvkm_falcon_load_imem(struct nvkm_falcon *falcon, void *data, u32 start, - u32 size, u16 tag, u8 port, bool secure) +static const struct nvkm_falcon_func_dma * +nvkm_falcon_dma(struct nvkm_falcon *falcon, enum nvkm_falcon_mem *mem_type, u32 *mem_base) { - if (secure && !falcon->secret) { - nvkm_warn(falcon->user, - "writing with secure tag on a non-secure falcon!\n"); - return; + switch (*mem_type) { + case IMEM: return falcon->func->imem_dma; + case DMEM: return falcon->func->dmem_dma; + default: + return NULL; } - - falcon->func->load_imem(falcon, data, start, size, tag, port, - secure); } -void -nvkm_falcon_load_dmem(struct nvkm_falcon *falcon, void *data, u32 start, - u32 size, u8 port) +int +nvkm_falcon_dma_wr(struct nvkm_falcon *falcon, const u8 *img, u64 dma_addr, u32 dma_base, + enum nvkm_falcon_mem mem_type, u32 mem_base, int len, bool sec) { - mutex_lock(&falcon->dmem_mutex); - - falcon->func->load_dmem(falcon, data, start, size, port); + const struct nvkm_falcon_func_dma *dma = nvkm_falcon_dma(falcon, &mem_type, &mem_base); + const char *type = nvkm_falcon_mem(mem_type); + const int dmalen = 256; + u32 dma_start = 0; + u32 dst, src, cmd; + int ret, i; + + if (WARN_ON(!dma->xfer)) + return -EINVAL; + + if (mem_type == DMEM) { + dma_start = dma_base; + dma_addr += dma_base; + } - mutex_unlock(&falcon->dmem_mutex); -} + FLCN_DBG(falcon, "%s %08x <- %08x bytes at %08x (%010llx %08x)", + type, mem_base, len, dma_base, dma_addr - dma_base, dma_start); + if (WARN_ON(!len || (len & (dmalen - 1)))) + return -EINVAL; -void -nvkm_falcon_read_dmem(struct nvkm_falcon *falcon, u32 start, u32 size, u8 port, - void *data) -{ - mutex_lock(&falcon->dmem_mutex); + ret = dma->init(falcon, dma_addr, dmalen, mem_type, sec, &cmd); + if (ret) + return ret; - falcon->func->read_dmem(falcon, start, size, port, data); + dst = mem_base; + src = dma_base; + if (len) { + while (len >= dmalen) { + dma->xfer(falcon, dst, src - dma_start, cmd); + + if (img && nvkm_printk_ok(falcon->owner, falcon->user, NV_DBG_TRACE)) { + for (i = 0; i < dmalen; i += 4, mem_base += 4) { + const int w = 8, x = (i / 4) % w; + + if (x == 0) + printk(KERN_INFO "%s %08x <-", type, mem_base); + printk(KERN_CONT " %08x", *(u32 *)(img + src + i)); + if (x == (w - 1) || ((i + 4) == dmalen)) + printk(KERN_CONT " <- %08x+%08x", dma_base, + src + i - dma_base - (x * 4)); + if (i == (7 * 4)) + printk(KERN_CONT " *"); + } + } + + if (nvkm_msec(falcon->owner->device, 2000, + if (dma->done(falcon)) + break; + ) < 0) + return -ETIMEDOUT; + + src += dmalen; + dst += dmalen; + len -= dmalen; + } + WARN_ON(len); + } - mutex_unlock(&falcon->dmem_mutex); + return 0; } -void -nvkm_falcon_bind_context(struct nvkm_falcon *falcon, struct nvkm_memory *inst) +static const struct nvkm_falcon_func_pio * +nvkm_falcon_pio(struct nvkm_falcon *falcon, enum nvkm_falcon_mem *mem_type, u32 *mem_base) { - if (!falcon->func->bind_context) { - nvkm_error(falcon->user, - "Context binding not supported on this falcon!\n"); - return; + switch (*mem_type) { + case IMEM: + return falcon->func->imem_pio; + case DMEM: + if (!falcon->func->emem_addr || *mem_base < falcon->func->emem_addr) + return falcon->func->dmem_pio; + + *mem_base -= falcon->func->emem_addr; + fallthrough; + case EMEM: + return falcon->func->emem_pio; + default: + return NULL; } - - falcon->func->bind_context(falcon, inst); } -void -nvkm_falcon_set_start_addr(struct nvkm_falcon *falcon, u32 start_addr) +int +nvkm_falcon_pio_rd(struct nvkm_falcon *falcon, u8 port, enum nvkm_falcon_mem mem_type, u32 mem_base, + const u8 *img, u32 img_base, int len) { - falcon->func->set_start_addr(falcon, start_addr); -} + const struct nvkm_falcon_func_pio *pio = nvkm_falcon_pio(falcon, &mem_type, &mem_base); + const char *type = nvkm_falcon_mem(mem_type); + int xfer_len; + + if (WARN_ON(!pio || !pio->rd)) + return -EINVAL; + + FLCN_DBG(falcon, "%s %08x -> %08x bytes at %08x", type, mem_base, len, img_base); + if (WARN_ON(!len || (len & (pio->min - 1)))) + return -EINVAL; + + pio->rd_init(falcon, port, mem_base); + do { + xfer_len = min(len, pio->max); + pio->rd(falcon, port, img, xfer_len); + + if (nvkm_printk_ok(falcon->owner, falcon->user, NV_DBG_TRACE)) { + for (img_base = 0; img_base < xfer_len; img_base += 4, mem_base += 4) { + if (((img_base / 4) % 8) == 0) + printk(KERN_INFO "%s %08x ->", type, mem_base); + printk(KERN_CONT " %08x", *(u32 *)(img + img_base)); + } + } + + img += xfer_len; + len -= xfer_len; + } while (len); -void -nvkm_falcon_start(struct nvkm_falcon *falcon) -{ - falcon->func->start(falcon); + return 0; } int -nvkm_falcon_enable(struct nvkm_falcon *falcon) +nvkm_falcon_pio_wr(struct nvkm_falcon *falcon, const u8 *img, u32 img_base, u8 port, + enum nvkm_falcon_mem mem_type, u32 mem_base, int len, u16 tag, bool sec) { - struct nvkm_device *device = falcon->owner->device; - int ret; - - nvkm_mc_enable(device, falcon->owner->type, falcon->owner->inst); - ret = falcon->func->enable(falcon); - if (ret) { - nvkm_mc_disable(device, falcon->owner->type, falcon->owner->inst); - return ret; - } + const struct nvkm_falcon_func_pio *pio = nvkm_falcon_pio(falcon, &mem_type, &mem_base); + const char *type = nvkm_falcon_mem(mem_type); + int xfer_len; + + if (WARN_ON(!pio || !pio->wr)) + return -EINVAL; + + FLCN_DBG(falcon, "%s %08x <- %08x bytes at %08x", type, mem_base, len, img_base); + if (WARN_ON(!len || (len & (pio->min - 1)))) + return -EINVAL; + + pio->wr_init(falcon, port, sec, mem_base); + do { + xfer_len = min(len, pio->max); + pio->wr(falcon, port, img, xfer_len, tag++); + + if (nvkm_printk_ok(falcon->owner, falcon->user, NV_DBG_TRACE)) { + for (img_base = 0; img_base < xfer_len; img_base += 4, mem_base += 4) { + if (((img_base / 4) % 8) == 0) + printk(KERN_INFO "%s %08x <-", type, mem_base); + printk(KERN_CONT " %08x", *(u32 *)(img + img_base)); + if ((img_base / 4) == 7 && mem_type == IMEM) + printk(KERN_CONT " %04x", tag - 1); + } + } + + img += xfer_len; + len -= xfer_len; + } while (len); return 0; } void -nvkm_falcon_disable(struct nvkm_falcon *falcon) +nvkm_falcon_load_imem(struct nvkm_falcon *falcon, void *data, u32 start, + u32 size, u16 tag, u8 port, bool secure) { - struct nvkm_device *device = falcon->owner->device; - - /* already disabled, return or wait_idle will timeout */ - if (!nvkm_mc_enabled(device, falcon->owner->type, falcon->owner->inst)) + if (secure && !falcon->secret) { + nvkm_warn(falcon->user, + "writing with secure tag on a non-secure falcon!\n"); return; + } - falcon->func->disable(falcon); - - nvkm_mc_disable(device, falcon->owner->type, falcon->owner->inst); + falcon->func->load_imem(falcon, data, start, size, tag, port, + secure); } -int -nvkm_falcon_reset(struct nvkm_falcon *falcon) +void +nvkm_falcon_load_dmem(struct nvkm_falcon *falcon, void *data, u32 start, + u32 size, u8 port) { - if (!falcon->func->reset) { - nvkm_falcon_disable(falcon); - return nvkm_falcon_enable(falcon); - } + mutex_lock(&falcon->dmem_mutex); - return falcon->func->reset(falcon); + falcon->func->load_dmem(falcon, data, start, size, port); + + mutex_unlock(&falcon->dmem_mutex); } -int -nvkm_falcon_wait_for_halt(struct nvkm_falcon *falcon, u32 ms) +void +nvkm_falcon_start(struct nvkm_falcon *falcon) { - return falcon->func->wait_for_halt(falcon, ms); + falcon->func->start(falcon); } int -nvkm_falcon_clear_interrupt(struct nvkm_falcon *falcon, u32 mask) +nvkm_falcon_reset(struct nvkm_falcon *falcon) { - return falcon->func->clear_interrupt(falcon, mask); + int ret; + + ret = falcon->func->disable(falcon); + if (WARN_ON(ret)) + return ret; + + return nvkm_falcon_enable(falcon); } static int @@ -169,7 +267,7 @@ nvkm_falcon_oneinit(struct nvkm_falcon *falcon) } void -nvkm_falcon_put(struct nvkm_falcon *falcon, const struct nvkm_subdev *user) +nvkm_falcon_put(struct nvkm_falcon *falcon, struct nvkm_subdev *user) { if (unlikely(!falcon)) return; @@ -183,7 +281,7 @@ nvkm_falcon_put(struct nvkm_falcon *falcon, const struct nvkm_subdev *user) } int -nvkm_falcon_get(struct nvkm_falcon *falcon, const struct nvkm_subdev *user) +nvkm_falcon_get(struct nvkm_falcon *falcon, struct nvkm_subdev *user) { int ret = 0; @@ -217,6 +315,7 @@ nvkm_falcon_ctor(const struct nvkm_falcon_func *func, falcon->owner = subdev; falcon->name = name; falcon->addr = addr; + falcon->addr2 = func->addr2; mutex_init(&falcon->mutex); mutex_init(&falcon->dmem_mutex); return 0; diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/cmdq.c b/drivers/gpu/drm/nouveau/nvkm/falcon/cmdq.c index 44cf6a8862e1..211ebe7afac6 100644 --- a/drivers/gpu/drm/nouveau/nvkm/falcon/cmdq.c +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/cmdq.c @@ -51,7 +51,7 @@ static void nvkm_falcon_cmdq_push(struct nvkm_falcon_cmdq *cmdq, void *data, u32 size) { struct nvkm_falcon *falcon = cmdq->qmgr->falcon; - nvkm_falcon_load_dmem(falcon, data, cmdq->position, size, 0); + nvkm_falcon_pio_wr(falcon, data, 0, 0, DMEM, cmdq->position, size, 0, false); cmdq->position += ALIGN(size, QUEUE_ALIGNMENT); } diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/fw.c b/drivers/gpu/drm/nouveau/nvkm/falcon/fw.c new file mode 100644 index 000000000000..80a480b12174 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/fw.c @@ -0,0 +1,354 @@ +/* + * Copyright 2022 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +#include <core/memory.h> +#include <subdev/mmu.h> + +#include <nvfw/fw.h> +#include <nvfw/hs.h> + +int +nvkm_falcon_fw_patch(struct nvkm_falcon_fw *fw) +{ + struct nvkm_falcon *falcon = fw->falcon; + u32 sig_base_src = fw->sig_base_prd; + u32 src, dst, len, i; + int idx = 0; + + FLCNFW_DBG(fw, "patching sigs:%d size:%d", fw->sig_nr, fw->sig_size); + if (fw->func->signature) { + idx = fw->func->signature(fw, &sig_base_src); + if (idx < 0) + return idx; + } + + src = idx * fw->sig_size; + dst = fw->sig_base_img; + len = fw->sig_size / 4; + FLCNFW_DBG(fw, "patch idx:%d src:%08x dst:%08x", idx, sig_base_src + src, dst); + for (i = 0; i < len; i++) { + u32 sig = *(u32 *)(fw->sigs + src); + + if (nvkm_printk_ok(falcon->owner, falcon->user, NV_DBG_TRACE)) { + if (i % 8 == 0) + printk(KERN_INFO "sig -> %08x:", dst); + printk(KERN_CONT " %08x", sig); + } + + *(u32 *)(fw->fw.img + dst) = sig; + src += 4; + dst += 4; + } + + return 0; +} + +static void +nvkm_falcon_fw_dtor_sigs(struct nvkm_falcon_fw *fw) +{ + kfree(fw->sigs); + fw->sigs = NULL; +} + +int +nvkm_falcon_fw_boot(struct nvkm_falcon_fw *fw, struct nvkm_subdev *user, + bool release, u32 *pmbox0, u32 *pmbox1, u32 mbox0_ok, u32 irqsclr) +{ + struct nvkm_falcon *falcon = fw->falcon; + int ret; + + ret = nvkm_falcon_get(falcon, user); + if (ret) + return ret; + + if (fw->sigs) { + ret = nvkm_falcon_fw_patch(fw); + if (ret) + goto done; + + nvkm_falcon_fw_dtor_sigs(fw); + } + + FLCNFW_DBG(fw, "resetting"); + fw->func->reset(fw); + + FLCNFW_DBG(fw, "loading"); + if (fw->func->setup) { + ret = fw->func->setup(fw); + if (ret) + goto done; + } + + ret = fw->func->load(fw); + if (ret) + goto done; + + FLCNFW_DBG(fw, "booting"); + ret = fw->func->boot(fw, pmbox0, pmbox1, mbox0_ok, irqsclr); + if (ret) + FLCNFW_ERR(fw, "boot failed: %d", ret); + else + FLCNFW_DBG(fw, "booted"); + +done: + if (ret || release) + nvkm_falcon_put(falcon, user); + return ret; +} + +int +nvkm_falcon_fw_oneinit(struct nvkm_falcon_fw *fw, struct nvkm_falcon *falcon, + struct nvkm_vmm *vmm, struct nvkm_memory *inst) +{ + int ret; + + fw->falcon = falcon; + fw->vmm = nvkm_vmm_ref(vmm); + fw->inst = nvkm_memory_ref(inst); + + if (fw->boot) { + FLCN_DBG(falcon, "mapping %s fw", fw->fw.name); + ret = nvkm_vmm_get(fw->vmm, 12, nvkm_memory_size(&fw->fw.mem.memory), &fw->vma); + if (ret) { + FLCN_ERR(falcon, "get %d", ret); + return ret; + } + + ret = nvkm_memory_map(&fw->fw.mem.memory, 0, fw->vmm, fw->vma, NULL, 0); + if (ret) { + FLCN_ERR(falcon, "map %d", ret); + return ret; + } + } + + return 0; +} + +void +nvkm_falcon_fw_dtor(struct nvkm_falcon_fw *fw) +{ + nvkm_vmm_put(fw->vmm, &fw->vma); + nvkm_vmm_unref(&fw->vmm); + nvkm_memory_unref(&fw->inst); + nvkm_falcon_fw_dtor_sigs(fw); + nvkm_firmware_dtor(&fw->fw); +} + +static const struct nvkm_firmware_func +nvkm_falcon_fw_dma = { + .type = NVKM_FIRMWARE_IMG_DMA, +}; + +static const struct nvkm_firmware_func +nvkm_falcon_fw = { + .type = NVKM_FIRMWARE_IMG_RAM, +}; + +int +nvkm_falcon_fw_sign(struct nvkm_falcon_fw *fw, u32 sig_base_img, u32 sig_size, const u8 *sigs, + int sig_nr_prd, u32 sig_base_prd, int sig_nr_dbg, u32 sig_base_dbg) +{ + fw->sig_base_prd = sig_base_prd; + fw->sig_base_dbg = sig_base_dbg; + fw->sig_base_img = sig_base_img; + fw->sig_size = sig_size; + fw->sig_nr = sig_nr_prd + sig_nr_dbg; + + fw->sigs = kmalloc_array(fw->sig_nr, fw->sig_size, GFP_KERNEL); + if (!fw->sigs) + return -ENOMEM; + + memcpy(fw->sigs, sigs + sig_base_prd, sig_nr_prd * fw->sig_size); + if (sig_nr_dbg) + memcpy(fw->sigs + sig_size, sigs + sig_base_dbg, sig_nr_dbg * fw->sig_size); + + return 0; +} + +int +nvkm_falcon_fw_ctor(const struct nvkm_falcon_fw_func *func, const char *name, + struct nvkm_device *device, bool dma, const void *src, u32 len, + struct nvkm_falcon *falcon, struct nvkm_falcon_fw *fw) +{ + const struct nvkm_firmware_func *type = dma ? &nvkm_falcon_fw_dma : &nvkm_falcon_fw; + int ret; + + fw->func = func; + + ret = nvkm_firmware_ctor(type, name, device, src, len, &fw->fw); + if (ret) + return ret; + + return falcon ? nvkm_falcon_fw_oneinit(fw, falcon, NULL, NULL) : 0; +} + +int +nvkm_falcon_fw_ctor_hs(const struct nvkm_falcon_fw_func *func, const char *name, + struct nvkm_subdev *subdev, const char *bl, const char *img, int ver, + struct nvkm_falcon *falcon, struct nvkm_falcon_fw *fw) +{ + const struct firmware *blob; + const struct nvfw_bin_hdr *hdr; + const struct nvfw_hs_header *hshdr; + const struct nvfw_hs_load_header *lhdr; + const struct nvfw_bl_desc *desc; + u32 loc, sig; + int ret; + + ret = nvkm_firmware_load_name(subdev, img, "", ver, &blob); + if (ret) + return ret; + + hdr = nvfw_bin_hdr(subdev, blob->data); + hshdr = nvfw_hs_header(subdev, blob->data + hdr->header_offset); + + ret = nvkm_falcon_fw_ctor(func, name, subdev->device, bl != NULL, + blob->data + hdr->data_offset, hdr->data_size, falcon, fw); + if (ret) + goto done; + + /* Earlier FW releases by NVIDIA for Nouveau's use aren't in NVIDIA's + * standard format, and don't have the indirection seen in the 0x10de + * case. + */ + switch (hdr->bin_magic) { + case 0x000010de: + loc = *(u32 *)(blob->data + hshdr->patch_loc); + sig = *(u32 *)(blob->data + hshdr->patch_sig); + break; + case 0x3b1d14f0: + loc = hshdr->patch_loc; + sig = hshdr->patch_sig; + break; + default: + WARN_ON(1); + ret = -EINVAL; + goto done; + } + + ret = nvkm_falcon_fw_sign(fw, loc, hshdr->sig_prod_size, blob->data, + 1, hshdr->sig_prod_offset + sig, + 1, hshdr->sig_dbg_offset + sig); + if (ret) + goto done; + + lhdr = nvfw_hs_load_header(subdev, blob->data + hshdr->hdr_offset); + + fw->nmem_base_img = 0; + fw->nmem_base = lhdr->non_sec_code_off; + fw->nmem_size = lhdr->non_sec_code_size; + + fw->imem_base_img = lhdr->apps[0]; + fw->imem_base = ALIGN(lhdr->apps[0], 0x100); + fw->imem_size = lhdr->apps[lhdr->num_apps + 0]; + + fw->dmem_base_img = lhdr->data_dma_base; + fw->dmem_base = 0; + fw->dmem_size = lhdr->data_size; + fw->dmem_sign = loc - lhdr->data_dma_base; + + if (bl) { + nvkm_firmware_put(blob); + + ret = nvkm_firmware_load_name(subdev, bl, "", ver, &blob); + if (ret) + return ret; + + hdr = nvfw_bin_hdr(subdev, blob->data); + desc = nvfw_bl_desc(subdev, blob->data + hdr->header_offset); + + fw->boot_addr = desc->start_tag << 8; + fw->boot_size = desc->code_size; + fw->boot = kmemdup(blob->data + hdr->data_offset + desc->code_off, + fw->boot_size, GFP_KERNEL); + if (!fw->boot) + ret = -ENOMEM; + } else { + fw->boot_addr = fw->nmem_base; + } + +done: + if (ret) + nvkm_falcon_fw_dtor(fw); + + nvkm_firmware_put(blob); + return ret; +} + +int +nvkm_falcon_fw_ctor_hs_v2(const struct nvkm_falcon_fw_func *func, const char *name, + struct nvkm_subdev *subdev, const char *img, int ver, + struct nvkm_falcon *falcon, struct nvkm_falcon_fw *fw) +{ + const struct nvfw_bin_hdr *hdr; + const struct nvfw_hs_header_v2 *hshdr; + const struct nvfw_hs_load_header_v2 *lhdr; + const struct firmware *blob; + u32 loc, sig, cnt, *meta; + int ret; + + ret = nvkm_firmware_load_name(subdev, img, "", ver, &blob); + if (ret) + return ret; + + hdr = nvfw_bin_hdr(subdev, blob->data); + hshdr = nvfw_hs_header_v2(subdev, blob->data + hdr->header_offset); + meta = (u32 *)(blob->data + hshdr->meta_data_offset); + loc = *(u32 *)(blob->data + hshdr->patch_loc); + sig = *(u32 *)(blob->data + hshdr->patch_sig); + cnt = *(u32 *)(blob->data + hshdr->num_sig); + + ret = nvkm_falcon_fw_ctor(func, name, subdev->device, true, + blob->data + hdr->data_offset, hdr->data_size, falcon, fw); + if (ret) + goto done; + + ret = nvkm_falcon_fw_sign(fw, loc, hshdr->sig_prod_size / cnt, blob->data, + cnt, hshdr->sig_prod_offset + sig, 0, 0); + if (ret) + goto done; + + lhdr = nvfw_hs_load_header_v2(subdev, blob->data + hshdr->header_offset); + + fw->imem_base_img = lhdr->app[0].offset; + fw->imem_base = 0; + fw->imem_size = lhdr->app[0].size; + + fw->dmem_base_img = lhdr->os_data_offset; + fw->dmem_base = 0; + fw->dmem_size = lhdr->os_data_size; + fw->dmem_sign = loc - lhdr->os_data_offset; + + fw->boot_addr = lhdr->app[0].offset; + + fw->fuse_ver = meta[0]; + fw->engine_id = meta[1]; + fw->ucode_id = meta[2]; + +done: + if (ret) + nvkm_falcon_fw_dtor(fw); + + nvkm_firmware_put(blob); + return ret; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/ga100.c b/drivers/gpu/drm/nouveau/nvkm/falcon/ga100.c new file mode 100644 index 000000000000..49fd32943916 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/ga100.c @@ -0,0 +1,62 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +int +ga100_flcn_fw_signature(struct nvkm_falcon_fw *fw, u32 *src_base_src) +{ + struct nvkm_falcon *falcon = fw->falcon; + struct nvkm_device *device = falcon->owner->device; + u32 reg_fuse_version; + int idx; + + FLCN_DBG(falcon, "brom: %08x %08x", fw->engine_id, fw->ucode_id); + FLCN_DBG(falcon, "fuse_version: %d", fw->fuse_ver); + + if (fw->engine_id & 0x00000001) { + reg_fuse_version = nvkm_rd32(device, 0x824140 + (fw->ucode_id - 1) * 4); + } else + if (fw->engine_id & 0x00000004) { + reg_fuse_version = nvkm_rd32(device, 0x824100 + (fw->ucode_id - 1) * 4); + } else + if (fw->engine_id & 0x00000400) { + reg_fuse_version = nvkm_rd32(device, 0x8241c0 + (fw->ucode_id - 1) * 4); + } else { + WARN_ON(1); + return -ENOSYS; + } + + FLCN_DBG(falcon, "reg_fuse_version: %08x", reg_fuse_version); + if (reg_fuse_version) { + reg_fuse_version = fls(reg_fuse_version); + FLCN_DBG(falcon, "reg_fuse_version: %d", reg_fuse_version); + + if (WARN_ON(fw->fuse_ver < reg_fuse_version)) + return -EINVAL; + + idx = fw->fuse_ver - reg_fuse_version; + } else { + idx = fw->sig_nr - 1; + } + + return idx; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/ga102.c b/drivers/gpu/drm/nouveau/nvkm/falcon/ga102.c new file mode 100644 index 000000000000..0ff450fe3590 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/ga102.c @@ -0,0 +1,148 @@ +/* + * Copyright 2022 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +#include <subdev/mc.h> +#include <subdev/timer.h> + +static bool +ga102_flcn_dma_done(struct nvkm_falcon *falcon) +{ + return !!(nvkm_falcon_rd32(falcon, 0x118) & 0x00000002); +} + +static void +ga102_flcn_dma_xfer(struct nvkm_falcon *falcon, u32 mem_base, u32 dma_base, u32 cmd) +{ + nvkm_falcon_wr32(falcon, 0x114, mem_base); + nvkm_falcon_wr32(falcon, 0x11c, dma_base); + nvkm_falcon_wr32(falcon, 0x118, cmd); +} + +static int +ga102_flcn_dma_init(struct nvkm_falcon *falcon, u64 dma_addr, int xfer_len, + enum nvkm_falcon_mem mem_type, bool sec, u32 *cmd) +{ + *cmd = (ilog2(xfer_len) - 2) << 8; + if (mem_type == IMEM) + *cmd |= 0x00000010; + if (sec) + *cmd |= 0x00000004; + + nvkm_falcon_wr32(falcon, 0x110, dma_addr >> 8); + nvkm_falcon_wr32(falcon, 0x128, 0x00000000); + return 0; +} + +const struct nvkm_falcon_func_dma +ga102_flcn_dma = { + .init = ga102_flcn_dma_init, + .xfer = ga102_flcn_dma_xfer, + .done = ga102_flcn_dma_done, +}; + +int +ga102_flcn_reset_wait_mem_scrubbing(struct nvkm_falcon *falcon) +{ + nvkm_falcon_mask(falcon, 0x040, 0x00000000, 0x00000000); + + if (nvkm_msec(falcon->owner->device, 20, + if (!(nvkm_falcon_rd32(falcon, 0x0f4) & 0x00001000)) + break; + ) < 0) + return -ETIMEDOUT; + + return 0; +} + +int +ga102_flcn_reset_prep(struct nvkm_falcon *falcon) +{ + nvkm_falcon_rd32(falcon, 0x0f4); + + nvkm_usec(falcon->owner->device, 150, + if (nvkm_falcon_rd32(falcon, 0x0f4) & 0x80000000) + break; + _warn = false; + ); + + return 0; +} + +int +ga102_flcn_select(struct nvkm_falcon *falcon) +{ + if ((nvkm_falcon_rd32(falcon, falcon->addr2 + 0x668) & 0x00000010) != 0x00000000) { + nvkm_falcon_wr32(falcon, falcon->addr2 + 0x668, 0x00000000); + if (nvkm_msec(falcon->owner->device, 10, + if (nvkm_falcon_rd32(falcon, falcon->addr2 + 0x668) & 0x00000001) + break; + ) < 0) + return -ETIMEDOUT; + } + + return 0; +} + +int +ga102_flcn_fw_boot(struct nvkm_falcon_fw *fw, u32 *mbox0, u32 *mbox1, u32 mbox0_ok, u32 irqsclr) +{ + struct nvkm_falcon *falcon = fw->falcon; + + nvkm_falcon_wr32(falcon, falcon->addr2 + 0x210, fw->dmem_sign); + nvkm_falcon_wr32(falcon, falcon->addr2 + 0x19c, fw->engine_id); + nvkm_falcon_wr32(falcon, falcon->addr2 + 0x198, fw->ucode_id); + nvkm_falcon_wr32(falcon, falcon->addr2 + 0x180, 0x00000001); + + return gm200_flcn_fw_boot(fw, mbox0, mbox1, mbox0_ok, irqsclr); +} + +int +ga102_flcn_fw_load(struct nvkm_falcon_fw *fw) +{ + struct nvkm_falcon *falcon = fw->falcon; + int ret = 0; + + nvkm_falcon_mask(falcon, 0x624, 0x00000080, 0x00000080); + nvkm_falcon_wr32(falcon, 0x10c, 0x00000000); + nvkm_falcon_mask(falcon, 0x600, 0x00010007, (0 << 16) | (1 << 2) | 1); + + ret = nvkm_falcon_dma_wr(falcon, fw->fw.img, fw->fw.phys, fw->imem_base_img, + IMEM, fw->imem_base, fw->imem_size, true); + if (ret) + return ret; + + ret = nvkm_falcon_dma_wr(falcon, fw->fw.img, fw->fw.phys, fw->dmem_base_img, + DMEM, fw->dmem_base, fw->dmem_size, false); + if (ret) + return ret; + + return 0; +} + +const struct nvkm_falcon_fw_func +ga102_flcn_fw = { + .signature = ga100_flcn_fw_signature, + .reset = gm200_flcn_fw_reset, + .load = ga102_flcn_fw_load, + .boot = ga102_flcn_fw_boot, +}; diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/gm200.c b/drivers/gpu/drm/nouveau/nvkm/falcon/gm200.c new file mode 100644 index 000000000000..393ade9f7e6c --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/gm200.c @@ -0,0 +1,345 @@ +/* + * Copyright 2022 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +#include <core/memory.h> +#include <subdev/mc.h> +#include <subdev/timer.h> + +void +gm200_flcn_tracepc(struct nvkm_falcon *falcon) +{ + u32 sctl = nvkm_falcon_rd32(falcon, 0x240); + u32 tidx = nvkm_falcon_rd32(falcon, 0x148); + int nr = (tidx & 0x00ff0000) >> 16, sp, ip; + + FLCN_ERR(falcon, "TRACEPC SCTL %08x TIDX %08x", sctl, tidx); + for (sp = 0; sp < nr; sp++) { + nvkm_falcon_wr32(falcon, 0x148, sp); + ip = nvkm_falcon_rd32(falcon, 0x14c); + FLCN_ERR(falcon, "TRACEPC: %08x", ip); + } +} + +static void +gm200_flcn_pio_dmem_rd(struct nvkm_falcon *falcon, u8 port, const u8 *img, int len) +{ + while (len >= 4) { + *(u32 *)img = nvkm_falcon_rd32(falcon, 0x1c4 + (port * 8)); + img += 4; + len -= 4; + } +} + +static void +gm200_flcn_pio_dmem_rd_init(struct nvkm_falcon *falcon, u8 port, u32 dmem_base) +{ + nvkm_falcon_wr32(falcon, 0x1c0 + (port * 8), BIT(25) | dmem_base); +} + +static void +gm200_flcn_pio_dmem_wr(struct nvkm_falcon *falcon, u8 port, const u8 *img, int len, u16 tag) +{ + while (len >= 4) { + nvkm_falcon_wr32(falcon, 0x1c4 + (port * 8), *(u32 *)img); + img += 4; + len -= 4; + } +} + +static void +gm200_flcn_pio_dmem_wr_init(struct nvkm_falcon *falcon, u8 port, bool sec, u32 dmem_base) +{ + nvkm_falcon_wr32(falcon, 0x1c0 + (port * 8), BIT(24) | dmem_base); +} + +const struct nvkm_falcon_func_pio +gm200_flcn_dmem_pio = { + .min = 4, + .max = 0x100, + .wr_init = gm200_flcn_pio_dmem_wr_init, + .wr = gm200_flcn_pio_dmem_wr, + .rd_init = gm200_flcn_pio_dmem_rd_init, + .rd = gm200_flcn_pio_dmem_rd, +}; + +static void +gm200_flcn_pio_imem_wr_init(struct nvkm_falcon *falcon, u8 port, bool sec, u32 imem_base) +{ + nvkm_falcon_wr32(falcon, 0x180 + (port * 0x10), (sec ? BIT(28) : 0) | BIT(24) | imem_base); +} + +static void +gm200_flcn_pio_imem_wr(struct nvkm_falcon *falcon, u8 port, const u8 *img, int len, u16 tag) +{ + nvkm_falcon_wr32(falcon, 0x188 + (port * 0x10), tag++); + while (len >= 4) { + nvkm_falcon_wr32(falcon, 0x184 + (port * 0x10), *(u32 *)img); + img += 4; + len -= 4; + } +} + +const struct nvkm_falcon_func_pio +gm200_flcn_imem_pio = { + .min = 0x100, + .max = 0x100, + .wr_init = gm200_flcn_pio_imem_wr_init, + .wr = gm200_flcn_pio_imem_wr, +}; + +int +gm200_flcn_bind_stat(struct nvkm_falcon *falcon, bool intr) +{ + if (intr && !(nvkm_falcon_rd32(falcon, 0x008) & 0x00000008)) + return -1; + + return (nvkm_falcon_rd32(falcon, 0x0dc) & 0x00007000) >> 12; +} + +void +gm200_flcn_bind_inst(struct nvkm_falcon *falcon, int target, u64 addr) +{ + nvkm_falcon_mask(falcon, 0x604, 0x00000007, 0x00000000); /* DMAIDX_VIRT */ + nvkm_falcon_wr32(falcon, 0x054, (1 << 30) | (target << 28) | (addr >> 12)); + nvkm_falcon_mask(falcon, 0x090, 0x00010000, 0x00010000); + nvkm_falcon_mask(falcon, 0x0a4, 0x00000008, 0x00000008); +} + +int +gm200_flcn_reset_wait_mem_scrubbing(struct nvkm_falcon *falcon) +{ + nvkm_falcon_mask(falcon, 0x040, 0x00000000, 0x00000000); + + if (nvkm_msec(falcon->owner->device, 10, + if (!(nvkm_falcon_rd32(falcon, 0x10c) & 0x00000006)) + break; + ) < 0) + return -ETIMEDOUT; + + return 0; +} + +int +gm200_flcn_enable(struct nvkm_falcon *falcon) +{ + struct nvkm_device *device = falcon->owner->device; + int ret; + + if (falcon->func->reset_eng) { + ret = falcon->func->reset_eng(falcon); + if (ret) + return ret; + } + + if (falcon->func->select) { + ret = falcon->func->select(falcon); + if (ret) + return ret; + } + + if (falcon->func->reset_pmc) + nvkm_mc_enable(device, falcon->owner->type, falcon->owner->inst); + + ret = falcon->func->reset_wait_mem_scrubbing(falcon); + if (ret) + return ret; + + nvkm_falcon_wr32(falcon, 0x084, nvkm_rd32(device, 0x000000)); + return 0; +} + +int +gm200_flcn_disable(struct nvkm_falcon *falcon) +{ + struct nvkm_device *device = falcon->owner->device; + int ret; + + if (falcon->func->select) { + ret = falcon->func->select(falcon); + if (ret) + return ret; + } + + nvkm_falcon_mask(falcon, 0x048, 0x00000003, 0x00000000); + nvkm_falcon_wr32(falcon, 0x014, 0xffffffff); + + if (falcon->func->reset_pmc) { + if (falcon->func->reset_prep) { + ret = falcon->func->reset_prep(falcon); + if (ret) + return ret; + } + + nvkm_mc_disable(device, falcon->owner->type, falcon->owner->inst); + } + + if (falcon->func->reset_eng) { + ret = falcon->func->reset_eng(falcon); + if (ret) + return ret; + } + + return 0; +} + +int +gm200_flcn_fw_boot(struct nvkm_falcon_fw *fw, u32 *pmbox0, u32 *pmbox1, u32 mbox0_ok, u32 irqsclr) +{ + struct nvkm_falcon *falcon = fw->falcon; + u32 mbox0, mbox1; + int ret = 0; + + nvkm_falcon_wr32(falcon, 0x040, pmbox0 ? *pmbox0 : 0xcafebeef); + if (pmbox1) + nvkm_falcon_wr32(falcon, 0x044, *pmbox1); + + nvkm_falcon_wr32(falcon, 0x104, fw->boot_addr); + nvkm_falcon_wr32(falcon, 0x100, 0x00000002); + + if (nvkm_msec(falcon->owner->device, 2000, + if (nvkm_falcon_rd32(falcon, 0x100) & 0x00000010) + break; + ) < 0) + ret = -ETIMEDOUT; + + mbox0 = nvkm_falcon_rd32(falcon, 0x040); + mbox1 = nvkm_falcon_rd32(falcon, 0x044); + if (FLCN_ERRON(falcon, ret || mbox0 != mbox0_ok, "mbox %08x %08x", mbox0, mbox1)) + ret = ret ?: -EIO; + + if (irqsclr) + nvkm_falcon_mask(falcon, 0x004, 0xffffffff, irqsclr); + + return ret; +} + +int +gm200_flcn_fw_load(struct nvkm_falcon_fw *fw) +{ + struct nvkm_falcon *falcon = fw->falcon; + int target, ret; + + if (fw->inst) { + nvkm_falcon_mask(falcon, 0x048, 0x00000001, 0x00000001); + + switch (nvkm_memory_target(fw->inst)) { + case NVKM_MEM_TARGET_VRAM: target = 0; break; + case NVKM_MEM_TARGET_HOST: target = 2; break; + case NVKM_MEM_TARGET_NCOH: target = 3; break; + default: + WARN_ON(1); + return -EINVAL; + } + + falcon->func->bind_inst(falcon, target, nvkm_memory_addr(fw->inst)); + + if (nvkm_msec(falcon->owner->device, 10, + if (falcon->func->bind_stat(falcon, falcon->func->bind_intr) == 5) + break; + ) < 0) + return -ETIMEDOUT; + + nvkm_falcon_mask(falcon, 0x004, 0x00000008, 0x00000008); + nvkm_falcon_mask(falcon, 0x058, 0x00000002, 0x00000002); + + if (nvkm_msec(falcon->owner->device, 10, + if (falcon->func->bind_stat(falcon, false) == 0) + break; + ) < 0) + return -ETIMEDOUT; + } else { + nvkm_falcon_mask(falcon, 0x624, 0x00000080, 0x00000080); + nvkm_falcon_wr32(falcon, 0x10c, 0x00000000); + } + + if (fw->boot) { + switch (nvkm_memory_target(&fw->fw.mem.memory)) { + case NVKM_MEM_TARGET_VRAM: target = 4; break; + case NVKM_MEM_TARGET_HOST: target = 5; break; + case NVKM_MEM_TARGET_NCOH: target = 6; break; + default: + WARN_ON(1); + return -EINVAL; + } + + ret = nvkm_falcon_pio_wr(falcon, fw->boot, 0, 0, + IMEM, falcon->code.limit - fw->boot_size, fw->boot_size, + fw->boot_addr >> 8, false); + if (ret) + return ret; + + return fw->func->load_bld(fw); + } + + ret = nvkm_falcon_pio_wr(falcon, fw->fw.img + fw->nmem_base_img, fw->nmem_base_img, 0, + IMEM, fw->nmem_base, fw->nmem_size, fw->nmem_base >> 8, false); + if (ret) + return ret; + + ret = nvkm_falcon_pio_wr(falcon, fw->fw.img + fw->imem_base_img, fw->imem_base_img, 0, + IMEM, fw->imem_base, fw->imem_size, fw->imem_base >> 8, true); + if (ret) + return ret; + + ret = nvkm_falcon_pio_wr(falcon, fw->fw.img + fw->dmem_base_img, fw->dmem_base_img, 0, + DMEM, fw->dmem_base, fw->dmem_size, 0, false); + if (ret) + return ret; + + return 0; +} + +int +gm200_flcn_fw_reset(struct nvkm_falcon_fw *fw) +{ + return nvkm_falcon_reset(fw->falcon); +} + +int +gm200_flcn_fw_signature(struct nvkm_falcon_fw *fw, u32 *sig_base_src) +{ + struct nvkm_falcon *falcon = fw->falcon; + u32 addr = falcon->func->debug; + int ret = 0; + + if (addr) { + ret = nvkm_falcon_enable(falcon); + if (ret) + return ret; + + if (nvkm_falcon_rd32(falcon, addr) & 0x00100000) { + *sig_base_src = fw->sig_base_dbg; + return 1; + } + } + + return ret; +} + +const struct nvkm_falcon_fw_func +gm200_flcn_fw = { + .signature = gm200_flcn_fw_signature, + .reset = gm200_flcn_fw_reset, + .load = gm200_flcn_fw_load, + .boot = gm200_flcn_fw_boot, +}; diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/gp102.c b/drivers/gpu/drm/nouveau/nvkm/falcon/gp102.c new file mode 100644 index 000000000000..c774935f3077 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/gp102.c @@ -0,0 +1,82 @@ +/* + * Copyright 2022 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +static void +gp102_flcn_pio_emem_rd(struct nvkm_falcon *falcon, u8 port, const u8 *img, int len) +{ + while (len >= 4) { + *(u32 *)img = nvkm_falcon_rd32(falcon, 0xac4 + (port * 8)); + img += 4; + len -= 4; + } +} + +static void +gp102_flcn_pio_emem_rd_init(struct nvkm_falcon *falcon, u8 port, u32 dmem_base) +{ + nvkm_falcon_wr32(falcon, 0xac0 + (port * 8), BIT(25) | dmem_base); +} + +static void +gp102_flcn_pio_emem_wr(struct nvkm_falcon *falcon, u8 port, const u8 *img, int len, u16 tag) +{ + while (len >= 4) { + nvkm_falcon_wr32(falcon, 0xac4 + (port * 8), *(u32 *)img); + img += 4; + len -= 4; + } +} + +static void +gp102_flcn_pio_emem_wr_init(struct nvkm_falcon *falcon, u8 port, bool sec, u32 emem_base) +{ + nvkm_falcon_wr32(falcon, 0xac0 + (port * 8), BIT(24) | emem_base); +} + +const struct nvkm_falcon_func_pio +gp102_flcn_emem_pio = { + .min = 4, + .max = 0x100, + .wr_init = gp102_flcn_pio_emem_wr_init, + .wr = gp102_flcn_pio_emem_wr, + .rd_init = gp102_flcn_pio_emem_rd_init, + .rd = gp102_flcn_pio_emem_rd, +}; + +int +gp102_flcn_reset_eng(struct nvkm_falcon *falcon) +{ + int ret; + + if (falcon->func->reset_prep) { + ret = falcon->func->reset_prep(falcon); + if (ret) + return ret; + } + + nvkm_falcon_mask(falcon, 0x3c0, 0x00000001, 0x00000001); + udelay(10); + nvkm_falcon_mask(falcon, 0x3c0, 0x00000001, 0x00000000); + + return falcon->func->reset_wait_mem_scrubbing(falcon); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/msgq.c b/drivers/gpu/drm/nouveau/nvkm/falcon/msgq.c index e74371dffc76..16b246fda666 100644 --- a/drivers/gpu/drm/nouveau/nvkm/falcon/msgq.c +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/msgq.c @@ -25,7 +25,7 @@ static void nvkm_falcon_msgq_open(struct nvkm_falcon_msgq *msgq) { - mutex_lock(&msgq->mutex); + spin_lock(&msgq->lock); msgq->position = nvkm_falcon_rd32(msgq->qmgr->falcon, msgq->tail_reg); } @@ -37,10 +37,10 @@ nvkm_falcon_msgq_close(struct nvkm_falcon_msgq *msgq, bool commit) if (commit) nvkm_falcon_wr32(falcon, msgq->tail_reg, msgq->position); - mutex_unlock(&msgq->mutex); + spin_unlock(&msgq->lock); } -static bool +bool nvkm_falcon_msgq_empty(struct nvkm_falcon_msgq *msgq) { u32 head = nvkm_falcon_rd32(msgq->qmgr->falcon, msgq->head_reg); @@ -68,7 +68,7 @@ nvkm_falcon_msgq_pop(struct nvkm_falcon_msgq *msgq, void *data, u32 size) return -EINVAL; } - nvkm_falcon_read_dmem(falcon, tail, size, 0, data); + nvkm_falcon_pio_rd(falcon, 0, DMEM, tail, data, 0, size); msgq->position += ALIGN(size, QUEUE_ALIGNMENT); return 0; } @@ -208,6 +208,6 @@ nvkm_falcon_msgq_new(struct nvkm_falcon_qmgr *qmgr, const char *name, msgq->qmgr = qmgr; msgq->name = name; - mutex_init(&msgq->mutex); + spin_lock_init(&msgq->lock); return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/priv.h b/drivers/gpu/drm/nouveau/nvkm/falcon/priv.h index 466188752eb0..11a24b9c8569 100644 --- a/drivers/gpu/drm/nouveau/nvkm/falcon/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/priv.h @@ -2,4 +2,12 @@ #ifndef __NVKM_FALCON_PRIV_H__ #define __NVKM_FALCON_PRIV_H__ #include <core/falcon.h> + +static inline int +nvkm_falcon_enable(struct nvkm_falcon *falcon) +{ + if (falcon->func->enable) + return falcon->func->enable(falcon); + return 0; +} #endif diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/qmgr.h b/drivers/gpu/drm/nouveau/nvkm/falcon/qmgr.h index 976cb7b7aa99..79f0da9e749f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/falcon/qmgr.h +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/qmgr.h @@ -73,7 +73,7 @@ struct nvkm_falcon_cmdq { struct nvkm_falcon_msgq { struct nvkm_falcon_qmgr *qmgr; const char *name; - struct mutex mutex; + spinlock_t lock; u32 head_reg; u32 tail_reg; @@ -82,8 +82,7 @@ struct nvkm_falcon_msgq { u32 position; }; -#define FLCNQ_PRINTK(t,q,f,a...) \ - FLCN_PRINTK(t, (q)->qmgr->falcon, "%s: "f, (q)->name, ##a) -#define FLCNQ_DBG(q,f,a...) FLCNQ_PRINTK(debug, (q), f, ##a) -#define FLCNQ_ERR(q,f,a...) FLCNQ_PRINTK(error, (q), f, ##a) +#define FLCNQ_PRINTK(q,l,p,f,a...) FLCN_PRINTK((q)->qmgr->falcon, l, p, "%s: "f, (q)->name, ##a) +#define FLCNQ_DBG(q,f,a...) FLCNQ_PRINTK((q), DEBUG, info, f, ##a) +#define FLCNQ_ERR(q,f,a...) FLCNQ_PRINTK((q), ERROR, err, f, ##a) #endif diff --git a/drivers/gpu/drm/nouveau/nvkm/falcon/v1.c b/drivers/gpu/drm/nouveau/nvkm/falcon/v1.c index b0ee4c31414c..dd2ddc54ac60 100644 --- a/drivers/gpu/drm/nouveau/nvkm/falcon/v1.c +++ b/drivers/gpu/drm/nouveau/nvkm/falcon/v1.c @@ -64,44 +64,13 @@ nvkm_falcon_v1_load_imem(struct nvkm_falcon *falcon, void *data, u32 start, nvkm_falcon_wr32(falcon, 0x184 + (port * 16), 0); } -static void -nvkm_falcon_v1_load_emem(struct nvkm_falcon *falcon, void *data, u32 start, - u32 size, u8 port) -{ - u8 rem = size % 4; - int i; - - size -= rem; - - nvkm_falcon_wr32(falcon, 0xac0 + (port * 8), start | (0x1 << 24)); - for (i = 0; i < size / 4; i++) - nvkm_falcon_wr32(falcon, 0xac4 + (port * 8), ((u32 *)data)[i]); - - /* - * If size is not a multiple of 4, mask the last word to ensure garbage - * does not get written - */ - if (rem) { - u32 extra = ((u32 *)data)[i]; - - nvkm_falcon_wr32(falcon, 0xac4 + (port * 8), - extra & (BIT(rem * 8) - 1)); - } -} - void nvkm_falcon_v1_load_dmem(struct nvkm_falcon *falcon, void *data, u32 start, u32 size, u8 port) { - const struct nvkm_falcon_func *func = falcon->func; u8 rem = size % 4; int i; - if (func->emem_addr && start >= func->emem_addr) - return nvkm_falcon_v1_load_emem(falcon, data, - start - func->emem_addr, size, - port); - size -= rem; nvkm_falcon_wr32(falcon, 0x1c0 + (port * 8), start | (0x1 << 24)); @@ -120,113 +89,6 @@ nvkm_falcon_v1_load_dmem(struct nvkm_falcon *falcon, void *data, u32 start, } } -static void -nvkm_falcon_v1_read_emem(struct nvkm_falcon *falcon, u32 start, u32 size, - u8 port, void *data) -{ - u8 rem = size % 4; - int i; - - size -= rem; - - nvkm_falcon_wr32(falcon, 0xac0 + (port * 8), start | (0x1 << 25)); - for (i = 0; i < size / 4; i++) - ((u32 *)data)[i] = nvkm_falcon_rd32(falcon, 0xac4 + (port * 8)); - - /* - * If size is not a multiple of 4, mask the last word to ensure garbage - * does not get read - */ - if (rem) { - u32 extra = nvkm_falcon_rd32(falcon, 0xac4 + (port * 8)); - - for (i = size; i < size + rem; i++) { - ((u8 *)data)[i] = (u8)(extra & 0xff); - extra >>= 8; - } - } -} - -void -nvkm_falcon_v1_read_dmem(struct nvkm_falcon *falcon, u32 start, u32 size, - u8 port, void *data) -{ - const struct nvkm_falcon_func *func = falcon->func; - u8 rem = size % 4; - int i; - - if (func->emem_addr && start >= func->emem_addr) - return nvkm_falcon_v1_read_emem(falcon, start - func->emem_addr, - size, port, data); - - size -= rem; - - nvkm_falcon_wr32(falcon, 0x1c0 + (port * 8), start | (0x1 << 25)); - for (i = 0; i < size / 4; i++) - ((u32 *)data)[i] = nvkm_falcon_rd32(falcon, 0x1c4 + (port * 8)); - - /* - * If size is not a multiple of 4, mask the last word to ensure garbage - * does not get read - */ - if (rem) { - u32 extra = nvkm_falcon_rd32(falcon, 0x1c4 + (port * 8)); - - for (i = size; i < size + rem; i++) { - ((u8 *)data)[i] = (u8)(extra & 0xff); - extra >>= 8; - } - } -} - -void -nvkm_falcon_v1_bind_context(struct nvkm_falcon *falcon, struct nvkm_memory *ctx) -{ - const u32 fbif = falcon->func->fbif; - u32 inst_loc; - - /* disable instance block binding */ - if (ctx == NULL) { - nvkm_falcon_wr32(falcon, 0x10c, 0x0); - return; - } - - nvkm_falcon_wr32(falcon, 0x10c, 0x1); - - /* setup apertures - virtual */ - nvkm_falcon_wr32(falcon, fbif + 4 * FALCON_DMAIDX_UCODE, 0x4); - nvkm_falcon_wr32(falcon, fbif + 4 * FALCON_DMAIDX_VIRT, 0x0); - /* setup apertures - physical */ - nvkm_falcon_wr32(falcon, fbif + 4 * FALCON_DMAIDX_PHYS_VID, 0x4); - nvkm_falcon_wr32(falcon, fbif + 4 * FALCON_DMAIDX_PHYS_SYS_COH, 0x5); - nvkm_falcon_wr32(falcon, fbif + 4 * FALCON_DMAIDX_PHYS_SYS_NCOH, 0x6); - - /* Set context */ - switch (nvkm_memory_target(ctx)) { - case NVKM_MEM_TARGET_VRAM: inst_loc = 0; break; - case NVKM_MEM_TARGET_HOST: inst_loc = 2; break; - case NVKM_MEM_TARGET_NCOH: inst_loc = 3; break; - default: - WARN_ON(1); - return; - } - - /* Enable context */ - nvkm_falcon_mask(falcon, 0x048, 0x1, 0x1); - nvkm_falcon_wr32(falcon, 0x054, - ((nvkm_memory_addr(ctx) >> 12) & 0xfffffff) | - (inst_loc << 28) | (1 << 30)); - - nvkm_falcon_mask(falcon, 0x090, 0x10000, 0x10000); - nvkm_falcon_mask(falcon, 0x0a4, 0x8, 0x8); -} - -void -nvkm_falcon_v1_set_start_addr(struct nvkm_falcon *falcon, u32 start_addr) -{ - nvkm_falcon_wr32(falcon, 0x104, start_addr); -} - void nvkm_falcon_v1_start(struct nvkm_falcon *falcon) { @@ -237,75 +99,3 @@ nvkm_falcon_v1_start(struct nvkm_falcon *falcon) else nvkm_falcon_wr32(falcon, 0x100, 0x2); } - -int -nvkm_falcon_v1_wait_for_halt(struct nvkm_falcon *falcon, u32 ms) -{ - struct nvkm_device *device = falcon->owner->device; - int ret; - - ret = nvkm_wait_msec(device, ms, falcon->addr + 0x100, 0x10, 0x10); - if (ret < 0) - return ret; - - return 0; -} - -int -nvkm_falcon_v1_clear_interrupt(struct nvkm_falcon *falcon, u32 mask) -{ - struct nvkm_device *device = falcon->owner->device; - int ret; - - /* clear interrupt(s) */ - nvkm_falcon_mask(falcon, 0x004, mask, mask); - /* wait until interrupts are cleared */ - ret = nvkm_wait_msec(device, 10, falcon->addr + 0x008, mask, 0x0); - if (ret < 0) - return ret; - - return 0; -} - -static int -falcon_v1_wait_idle(struct nvkm_falcon *falcon) -{ - struct nvkm_device *device = falcon->owner->device; - int ret; - - ret = nvkm_wait_msec(device, 10, falcon->addr + 0x04c, 0xffff, 0x0); - if (ret < 0) - return ret; - - return 0; -} - -int -nvkm_falcon_v1_enable(struct nvkm_falcon *falcon) -{ - struct nvkm_device *device = falcon->owner->device; - int ret; - - ret = nvkm_wait_msec(device, 10, falcon->addr + 0x10c, 0x6, 0x0); - if (ret < 0) { - nvkm_error(falcon->user, "Falcon mem scrubbing timeout\n"); - return ret; - } - - ret = falcon_v1_wait_idle(falcon); - if (ret) - return ret; - - /* enable IRQs */ - nvkm_falcon_wr32(falcon, 0x010, 0xff); - - return 0; -} - -void -nvkm_falcon_v1_disable(struct nvkm_falcon *falcon) -{ - /* disable IRQs and wait for any previous code to complete */ - nvkm_falcon_wr32(falcon, 0x014, 0xff); - falcon_v1_wait_idle(falcon); -} diff --git a/drivers/gpu/drm/nouveau/nvkm/nvfw/acr.c b/drivers/gpu/drm/nouveau/nvkm/nvfw/acr.c index bef790ad8f2f..83a9c48bc58c 100644 --- a/drivers/gpu/drm/nouveau/nvkm/nvfw/acr.c +++ b/drivers/gpu/drm/nouveau/nvkm/nvfw/acr.c @@ -45,6 +45,47 @@ wpr_header_v1_dump(struct nvkm_subdev *subdev, const struct wpr_header_v1 *hdr) nvkm_debug(subdev, "\tstatus : %d\n", hdr->status); } +void +wpr_generic_header_dump(struct nvkm_subdev *subdev, const struct wpr_generic_header *hdr) +{ + nvkm_debug(subdev, "wprGenericHeader\n"); + nvkm_debug(subdev, "\tidentifier : %04x\n", hdr->identifier); + nvkm_debug(subdev, "\tversion : %04x\n", hdr->version); + nvkm_debug(subdev, "\tsize : %08x\n", hdr->size); +} + +void +wpr_header_v2_dump(struct nvkm_subdev *subdev, const struct wpr_header_v2 *hdr) +{ + wpr_generic_header_dump(subdev, &hdr->hdr); + wpr_header_v1_dump(subdev, &hdr->wpr); +} + +void +lsb_header_v2_dump(struct nvkm_subdev *subdev, struct lsb_header_v2 *hdr) +{ + wpr_generic_header_dump(subdev, &hdr->hdr); + nvkm_debug(subdev, "lsbHeader\n"); + nvkm_debug(subdev, "\tucodeOff : 0x%x\n", hdr->ucode_off); + nvkm_debug(subdev, "\tucodeSize : 0x%x\n", hdr->ucode_size); + nvkm_debug(subdev, "\tdataSize : 0x%x\n", hdr->data_size); + nvkm_debug(subdev, "\tblCodeSize : 0x%x\n", hdr->bl_code_size); + nvkm_debug(subdev, "\tblImemOff : 0x%x\n", hdr->bl_imem_off); + nvkm_debug(subdev, "\tblDataOff : 0x%x\n", hdr->bl_data_off); + nvkm_debug(subdev, "\tblDataSize : 0x%x\n", hdr->bl_data_size); + nvkm_debug(subdev, "\treserved0 : %08x\n", hdr->rsvd0); + nvkm_debug(subdev, "\tappCodeOff : 0x%x\n", hdr->app_code_off); + nvkm_debug(subdev, "\tappCodeSize : 0x%x\n", hdr->app_code_size); + nvkm_debug(subdev, "\tappDataOff : 0x%x\n", hdr->app_data_off); + nvkm_debug(subdev, "\tappDataSize : 0x%x\n", hdr->app_data_size); + nvkm_debug(subdev, "\tappImemOffset : 0x%x\n", hdr->app_imem_offset); + nvkm_debug(subdev, "\tappDmemOffset : 0x%x\n", hdr->app_dmem_offset); + nvkm_debug(subdev, "\tflags : 0x%x\n", hdr->flags); + nvkm_debug(subdev, "\tmonitorCodeOff: 0x%x\n", hdr->monitor_code_offset); + nvkm_debug(subdev, "\tmonitorDataOff: 0x%x\n", hdr->monitor_data_offset); + nvkm_debug(subdev, "\tmanifestOffset: 0x%x\n", hdr->manifest_offset); +} + static void lsb_header_tail_dump(struct nvkm_subdev *subdev, struct lsb_header_tail *hdr) { diff --git a/drivers/gpu/drm/nouveau/nvkm/nvfw/hs.c b/drivers/gpu/drm/nouveau/nvkm/nvfw/hs.c index 04ed77cb2eba..a7e0583401d0 100644 --- a/drivers/gpu/drm/nouveau/nvkm/nvfw/hs.c +++ b/drivers/gpu/drm/nouveau/nvkm/nvfw/hs.c @@ -38,6 +38,24 @@ nvfw_hs_header(struct nvkm_subdev *subdev, const void *data) return hdr; } +const struct nvfw_hs_header_v2 * +nvfw_hs_header_v2(struct nvkm_subdev *subdev, const void *data) +{ + const struct nvfw_hs_header_v2 *hdr = data; + + nvkm_debug(subdev, "hsHeader:\n"); + nvkm_debug(subdev, "\tsigProdOffset : 0x%x\n", hdr->sig_prod_offset); + nvkm_debug(subdev, "\tsigProdSize : 0x%x\n", hdr->sig_prod_size); + nvkm_debug(subdev, "\tpatchLoc : 0x%x\n", hdr->patch_loc); + nvkm_debug(subdev, "\tpatchSig : 0x%x\n", hdr->patch_sig); + nvkm_debug(subdev, "\tmetadataOffset : 0x%x\n", hdr->meta_data_offset); + nvkm_debug(subdev, "\tmetadataSize : 0x%x\n", hdr->meta_data_size); + nvkm_debug(subdev, "\tnumSig : 0x%x\n", hdr->num_sig); + nvkm_debug(subdev, "\theaderOffset : 0x%x\n", hdr->header_offset); + nvkm_debug(subdev, "\theaderSize : 0x%x\n", hdr->header_size); + return hdr; +} + const struct nvfw_hs_load_header * nvfw_hs_load_header(struct nvkm_subdev *subdev, const void *data) { @@ -60,3 +78,24 @@ nvfw_hs_load_header(struct nvkm_subdev *subdev, const void *data) return hdr; } + +const struct nvfw_hs_load_header_v2 * +nvfw_hs_load_header_v2(struct nvkm_subdev *subdev, const void *data) +{ + const struct nvfw_hs_load_header_v2 *hdr = data; + int i; + + nvkm_debug(subdev, "hsLoadHeader:\n"); + nvkm_debug(subdev, "\tosCodeOffset : 0x%x\n", hdr->os_code_offset); + nvkm_debug(subdev, "\tosCodeSize : 0x%x\n", hdr->os_code_size); + nvkm_debug(subdev, "\tosDataOffset : 0x%x\n", hdr->os_data_offset); + nvkm_debug(subdev, "\tosDataSize : 0x%x\n", hdr->os_data_size); + nvkm_debug(subdev, "\tnumApps : 0x%x\n", hdr->num_apps); + for (i = 0; i < hdr->num_apps; i++) { + nvkm_debug(subdev, + "\tApp[%d] : offset 0x%x size 0x%x\n", i, + hdr->app[i].offset, hdr->app[i].size); + } + + return hdr; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/nvfw/ls.c b/drivers/gpu/drm/nouveau/nvkm/nvfw/ls.c index b847f281ce97..45c3a6c5e088 100644 --- a/drivers/gpu/drm/nouveau/nvkm/nvfw/ls.c +++ b/drivers/gpu/drm/nouveau/nvkm/nvfw/ls.c @@ -106,3 +106,75 @@ nvfw_ls_desc_v1(struct nvkm_subdev *subdev, const void *data) return hdr; } + +const struct nvfw_ls_desc_v2 * +nvfw_ls_desc_v2(struct nvkm_subdev *subdev, const void *data) +{ + const struct nvfw_ls_desc_v2 *hdr = data; + char *date; + int i; + + nvkm_debug(subdev, "lsUcodeImgDesc:\n"); + nvkm_debug(subdev, "\tdescriptorSize : %d\n", hdr->descriptor_size); + nvkm_debug(subdev, "\timageSize : %d\n", hdr->image_size); + nvkm_debug(subdev, "\ttoolsVersion : 0x%x\n", hdr->tools_version); + nvkm_debug(subdev, "\tappVersion : 0x%x\n", hdr->app_version); + + date = kstrndup(hdr->date, sizeof(hdr->date), GFP_KERNEL); + nvkm_debug(subdev, "\tdate : %s\n", date); + kfree(date); + + nvkm_debug(subdev, "\tsecureBootloader : 0x%x\n", hdr->secure_bootloader); + nvkm_debug(subdev, "\tbootloaderStartOffset: 0x%x\n", hdr->bootloader_start_offset); + nvkm_debug(subdev, "\tbootloaderSize : 0x%x\n", hdr->bootloader_size); + nvkm_debug(subdev, "\tbootloaderImemOffset : 0x%x\n", hdr->bootloader_imem_offset); + nvkm_debug(subdev, "\tbootloaderEntryPoint : 0x%x\n", hdr->bootloader_entry_point); + + nvkm_debug(subdev, "\tappStartOffset : 0x%x\n", hdr->app_start_offset); + nvkm_debug(subdev, "\tappSize : 0x%x\n", hdr->app_size); + nvkm_debug(subdev, "\tappImemOffset : 0x%x\n", hdr->app_imem_offset); + nvkm_debug(subdev, "\tappImemEntry : 0x%x\n", hdr->app_imem_entry); + nvkm_debug(subdev, "\tappDmemOffset : 0x%x\n", hdr->app_dmem_offset); + nvkm_debug(subdev, "\tappResidentCodeOffset: 0x%x\n", hdr->app_resident_code_offset); + nvkm_debug(subdev, "\tappResidentCodeSize : 0x%x\n", hdr->app_resident_code_size); + nvkm_debug(subdev, "\tappResidentDataOffset: 0x%x\n", hdr->app_resident_data_offset); + nvkm_debug(subdev, "\tappResidentDataSize : 0x%x\n", hdr->app_resident_data_size); + + nvkm_debug(subdev, "\tnbImemOverlays : %d\n", hdr->nb_imem_overlays); + nvkm_debug(subdev, "\tnbDmemOverlays : %d\n", hdr->nb_dmem_overlays); + for (i = 0; i < ARRAY_SIZE(hdr->load_ovl); i++) { + nvkm_debug(subdev, "\tloadOvl[%d] : 0x%x %d\n", i, + hdr->load_ovl[i].start, hdr->load_ovl[i].size); + } + + return hdr; +} + +const struct nvfw_ls_hsbl_bin_hdr * +nvfw_ls_hsbl_bin_hdr(struct nvkm_subdev *subdev, const void *data) +{ + const struct nvfw_ls_hsbl_bin_hdr *hdr = data; + + nvkm_debug(subdev, "lsHsblBinHdr:\n"); + nvkm_debug(subdev, "\tbinMagic : 0x%08x\n", hdr->bin_magic); + nvkm_debug(subdev, "\tbinVer : %d\n", hdr->bin_ver); + nvkm_debug(subdev, "\tbinSize : %d\n", hdr->bin_size); + nvkm_debug(subdev, "\theaderOffset : 0x%x\n", hdr->header_offset); + return hdr; +} + +const struct nvfw_ls_hsbl_hdr * +nvfw_ls_hsbl_hdr(struct nvkm_subdev *subdev, const void *data) +{ + const struct nvfw_ls_hsbl_hdr *hdr = data; + + nvkm_debug(subdev, "lsHsblHdr:\n"); + nvkm_debug(subdev, "\tsigProdOffset : 0x%x\n", hdr->sig_prod_offset); + nvkm_debug(subdev, "\tsigProdSize : 0x%x\n", hdr->sig_prod_size); + nvkm_debug(subdev, "\tpatchLoc : 0x%x\n", hdr->patch_loc); + nvkm_debug(subdev, "\tpatchSig : 0x%x\n", hdr->patch_sig); + nvkm_debug(subdev, "\tmetadataOffset : 0x%x\n", hdr->meta_data_offset); + nvkm_debug(subdev, "\tmetadataSize : 0x%x\n", hdr->meta_data_size); + nvkm_debug(subdev, "\tnumSig : 0x%x\n", hdr->num_sig); + return hdr; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/Kbuild b/drivers/gpu/drm/nouveau/nvkm/subdev/Kbuild index 2cb24fff7e32..4c2f6fc4ef58 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/Kbuild @@ -23,4 +23,5 @@ include $(src)/nvkm/subdev/privring/Kbuild include $(src)/nvkm/subdev/therm/Kbuild include $(src)/nvkm/subdev/timer/Kbuild include $(src)/nvkm/subdev/top/Kbuild +include $(src)/nvkm/subdev/vfn/Kbuild include $(src)/nvkm/subdev/volt/Kbuild diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/Kbuild b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/Kbuild index 5b9f64a8957f..5731f35b11e1 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/Kbuild @@ -1,10 +1,12 @@ # SPDX-License-Identifier: MIT nvkm-y += nvkm/subdev/acr/base.o -nvkm-y += nvkm/subdev/acr/hsfw.o nvkm-y += nvkm/subdev/acr/lsfw.o nvkm-y += nvkm/subdev/acr/gm200.o nvkm-y += nvkm/subdev/acr/gm20b.o nvkm-y += nvkm/subdev/acr/gp102.o nvkm-y += nvkm/subdev/acr/gp108.o +nvkm-y += nvkm/subdev/acr/gv100.o nvkm-y += nvkm/subdev/acr/gp10b.o nvkm-y += nvkm/subdev/acr/tu102.o +nvkm-y += nvkm/subdev/acr/ga100.o +nvkm-y += nvkm/subdev/acr/ga102.o diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/base.c index af6cac696d43..795f3a649b12 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/base.c @@ -24,50 +24,63 @@ #include <core/firmware.h> #include <core/memory.h> #include <subdev/mmu.h> +#include <subdev/gsp.h> +#include <subdev/pmu.h> +#include <engine/sec2.h> +#include <engine/nvdec.h> -static struct nvkm_acr_hsf * -nvkm_acr_hsf_find(struct nvkm_acr *acr, const char *name) +static struct nvkm_acr_hsfw * +nvkm_acr_hsfw_find(struct nvkm_acr *acr, const char *name) { - struct nvkm_acr_hsf *hsf; - list_for_each_entry(hsf, &acr->hsf, head) { - if (!strcmp(hsf->name, name)) - return hsf; + struct nvkm_acr_hsfw *hsfw; + + list_for_each_entry(hsfw, &acr->hsfw, head) { + if (!strcmp(hsfw->fw.fw.name, name)) + return hsfw; } + return NULL; } int -nvkm_acr_hsf_boot(struct nvkm_acr *acr, const char *name) +nvkm_acr_hsfw_boot(struct nvkm_acr *acr, const char *name) { struct nvkm_subdev *subdev = &acr->subdev; - struct nvkm_acr_hsf *hsf; - int ret; + struct nvkm_acr_hsfw *hsfw; - hsf = nvkm_acr_hsf_find(acr, name); - if (!hsf) + hsfw = nvkm_acr_hsfw_find(acr, name); + if (!hsfw) return -EINVAL; - nvkm_debug(subdev, "executing %s binary\n", hsf->name); - ret = nvkm_falcon_get(hsf->falcon, subdev); - if (ret) - return ret; + return nvkm_falcon_fw_boot(&hsfw->fw, subdev, true, NULL, NULL, + hsfw->boot_mbox0, hsfw->intr_clear); +} - ret = hsf->func->boot(acr, hsf); - nvkm_falcon_put(hsf->falcon, subdev); - if (ret) { - nvkm_error(subdev, "%s binary failed\n", hsf->name); - return ret; +static struct nvkm_acr_lsf * +nvkm_acr_rtos(struct nvkm_acr *acr) +{ + struct nvkm_acr_lsf *lsf; + + if (acr) { + list_for_each_entry(lsf, &acr->lsf, head) { + if (lsf->func->bootstrap_falcon) + return lsf; + } } - nvkm_debug(subdev, "%s binary completed successfully\n", hsf->name); - return 0; + return NULL; } static void nvkm_acr_unload(struct nvkm_acr *acr) { if (acr->done) { - nvkm_acr_hsf_boot(acr, "unload"); + if (acr->rtos) { + nvkm_subdev_unref(acr->rtos->falcon->owner); + acr->rtos = NULL; + } + + nvkm_acr_hsfw_boot(acr, "unload"); acr->done = false; } } @@ -76,7 +89,7 @@ static int nvkm_acr_load(struct nvkm_acr *acr) { struct nvkm_subdev *subdev = &acr->subdev; - struct nvkm_acr_lsf *lsf; + struct nvkm_acr_lsf *rtos = nvkm_acr_rtos(acr); u64 start, limit; int ret; @@ -100,12 +113,12 @@ nvkm_acr_load(struct nvkm_acr *acr) acr->done = true; - list_for_each_entry(lsf, &acr->lsf, head) { - if (lsf->func->boot) { - ret = lsf->func->boot(lsf->falcon); - if (ret) - break; - } + if (rtos) { + ret = nvkm_subdev_ref(rtos->falcon->owner); + if (ret) + return ret; + + acr->rtos = rtos; } return ret; @@ -118,33 +131,17 @@ nvkm_acr_reload(struct nvkm_acr *acr) return nvkm_acr_load(acr); } -static struct nvkm_acr_lsf * -nvkm_acr_falcon(struct nvkm_device *device) -{ - struct nvkm_acr *acr = device->acr; - struct nvkm_acr_lsf *lsf; - - if (acr) { - list_for_each_entry(lsf, &acr->lsf, head) { - if (lsf->func->bootstrap_falcon) - return lsf; - } - } - - return NULL; -} - int nvkm_acr_bootstrap_falcons(struct nvkm_device *device, unsigned long mask) { - struct nvkm_acr_lsf *acrflcn = nvkm_acr_falcon(device); struct nvkm_acr *acr = device->acr; + struct nvkm_acr_lsf *rtos = nvkm_acr_rtos(acr); unsigned long id; /* If there's no LS FW managing bootstrapping of other LS falcons, * we depend on the HS firmware being able to do it instead. */ - if (!acrflcn) { + if (!rtos) { /* Which isn't possible everywhere... */ if ((mask & acr->func->bootstrap_falcons) == mask) { int ret = nvkm_acr_reload(acr); @@ -156,16 +153,14 @@ nvkm_acr_bootstrap_falcons(struct nvkm_device *device, unsigned long mask) return -ENOSYS; } - if ((mask & acrflcn->func->bootstrap_falcons) != mask) + if ((mask & rtos->func->bootstrap_falcons) != mask) return -ENOSYS; - if (acrflcn->func->bootstrap_multiple_falcons) { - return acrflcn->func-> - bootstrap_multiple_falcons(acrflcn->falcon, mask); - } + if (rtos->func->bootstrap_multiple_falcons) + return rtos->func->bootstrap_multiple_falcons(rtos->falcon, mask); for_each_set_bit(id, &mask, NVKM_ACR_LSF_NUM) { - int ret = acrflcn->func->bootstrap_falcon(acrflcn->falcon, id); + int ret = rtos->func->bootstrap_falcon(rtos->falcon, id); if (ret) return ret; } @@ -189,6 +184,9 @@ nvkm_acr_managed_falcon(struct nvkm_device *device, enum nvkm_acr_lsf_id id) static int nvkm_acr_fini(struct nvkm_subdev *subdev, bool suspend) { + if (!subdev->use.enabled) + return 0; + nvkm_acr_unload(nvkm_acr(subdev)); return 0; } @@ -196,17 +194,19 @@ nvkm_acr_fini(struct nvkm_subdev *subdev, bool suspend) static int nvkm_acr_init(struct nvkm_subdev *subdev) { - if (!nvkm_acr_falcon(subdev->device)) + struct nvkm_acr *acr = nvkm_acr(subdev); + + if (!nvkm_acr_rtos(acr)) return 0; - return nvkm_acr_load(nvkm_acr(subdev)); + return nvkm_acr_load(acr); } static void nvkm_acr_cleanup(struct nvkm_acr *acr) { nvkm_acr_lsfw_del_all(acr); - nvkm_acr_hsfw_del_all(acr); + nvkm_firmware_put(acr->wpr_fw); acr->wpr_fw = NULL; } @@ -218,7 +218,8 @@ nvkm_acr_oneinit(struct nvkm_subdev *subdev) struct nvkm_acr *acr = nvkm_acr(subdev); struct nvkm_acr_hsfw *hsfw; struct nvkm_acr_lsfw *lsfw, *lsft; - struct nvkm_acr_lsf *lsf; + struct nvkm_acr_lsf *lsf, *rtos; + struct nvkm_falcon *falcon; u32 wpr_size = 0; u64 falcons; int ret, i; @@ -260,10 +261,10 @@ nvkm_acr_oneinit(struct nvkm_subdev *subdev) } /* Ensure the falcon that'll provide ACR functions is booted first. */ - lsf = nvkm_acr_falcon(device); - if (lsf) { - falcons = lsf->func->bootstrap_falcons; - list_move(&lsf->head, &acr->lsf); + rtos = nvkm_acr_rtos(acr); + if (rtos) { + falcons = rtos->func->bootstrap_falcons; + list_move(&rtos->head, &acr->lsf); } else { falcons = acr->func->bootstrap_falcons; } @@ -301,7 +302,7 @@ nvkm_acr_oneinit(struct nvkm_subdev *subdev) nvkm_wobj(acr->wpr, 0, acr->wpr_fw->data, acr->wpr_fw->size); if (!acr->wpr_fw || acr->wpr_comp) - acr->func->wpr_build(acr, nvkm_acr_falcon(device)); + acr->func->wpr_build(acr, rtos); acr->func->wpr_patch(acr, (s64)acr->wpr_start - acr->wpr_prev); if (acr->wpr_fw && acr->wpr_comp) { @@ -336,8 +337,16 @@ nvkm_acr_oneinit(struct nvkm_subdev *subdev) /* Load HS firmware blobs into ACR VMM. */ list_for_each_entry(hsfw, &acr->hsfw, head) { - nvkm_debug(subdev, "loading %s fw\n", hsfw->name); - ret = hsfw->func->load(acr, hsfw); + switch (hsfw->falcon_id) { + case NVKM_ACR_HSF_PMU : falcon = &device->pmu->falcon; break; + case NVKM_ACR_HSF_SEC2: falcon = &device->sec2->falcon; break; + case NVKM_ACR_HSF_GSP : falcon = &device->gsp->falcon; break; + default: + WARN_ON(1); + return -EINVAL; + } + + ret = nvkm_falcon_fw_oneinit(&hsfw->fw, falcon, acr->vmm, acr->inst); if (ret) return ret; } @@ -351,15 +360,13 @@ static void * nvkm_acr_dtor(struct nvkm_subdev *subdev) { struct nvkm_acr *acr = nvkm_acr(subdev); - struct nvkm_acr_hsf *hsf, *hst; + struct nvkm_acr_hsfw *hsfw, *hsft; struct nvkm_acr_lsf *lsf, *lst; - list_for_each_entry_safe(hsf, hst, &acr->hsf, head) { - nvkm_vmm_put(acr->vmm, &hsf->vma); - nvkm_memory_unref(&hsf->ucode); - kfree(hsf->imem); - list_del(&hsf->head); - kfree(hsf); + list_for_each_entry_safe(hsfw, hsft, &acr->hsfw, head) { + nvkm_falcon_fw_dtor(&hsfw->fw); + list_del(&hsfw->head); + kfree(hsfw); } nvkm_vmm_part(acr->vmm, acr->inst); @@ -420,7 +427,6 @@ nvkm_acr_new_(const struct nvkm_acr_fwif *fwif, struct nvkm_device *device, nvkm_subdev_ctor(&nvkm_acr, device, type, inst, &acr->subdev); INIT_LIST_HEAD(&acr->hsfw); INIT_LIST_HEAD(&acr->lsfw); - INIT_LIST_HEAD(&acr->hsf); INIT_LIST_HEAD(&acr->lsf); fwif = nvkm_firmware_load(&acr->subdev, fwif, "Acr", acr); diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/ga100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/ga100.c new file mode 100644 index 000000000000..e3370c1551c0 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/ga100.c @@ -0,0 +1,49 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +void +ga100_acr_wpr_check(struct nvkm_acr *acr, u64 *start, u64 *limit) +{ + struct nvkm_device *device = acr->subdev.device; + + *start = (u64)(nvkm_rd32(device, 0x1fa81c) & 0xffffff00) << 8; + *limit = (u64)(nvkm_rd32(device, 0x1fa820) & 0xffffff00) << 8; + *limit = *limit + 0x20000; +} + +int +ga100_acr_hsfw_ctor(struct nvkm_acr *acr, const char *bl, const char *fw, + const char *name, int ver, const struct nvkm_acr_hsf_fwif *fwif) +{ + struct nvkm_acr_hsfw *hsfw; + + if (!(hsfw = kzalloc(sizeof(*hsfw), GFP_KERNEL))) + return -ENOMEM; + + hsfw->falcon_id = fwif->falcon_id; + hsfw->boot_mbox0 = fwif->boot_mbox0; + hsfw->intr_clear = fwif->intr_clear; + list_add_tail(&hsfw->head, &acr->hsfw); + + return nvkm_falcon_fw_ctor_hs_v2(fwif->func, name, &acr->subdev, fw, ver, NULL, &hsfw->fw); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/ga102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/ga102.c new file mode 100644 index 000000000000..45dcf493e972 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/ga102.c @@ -0,0 +1,326 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +#include <nvfw/acr.h> + +static int +ga102_acr_wpr_patch(struct nvkm_acr *acr, s64 adjust) +{ + struct wpr_header_v2 hdr; + struct lsb_header_v2 *lsb; + struct nvkm_acr_lsfw *lsfw; + u32 offset = 0; + + lsb = kvmalloc(sizeof(*lsb), GFP_KERNEL); + if (!lsb) + return -ENOMEM; + + do { + nvkm_robj(acr->wpr, offset, &hdr, sizeof(hdr)); + wpr_header_v2_dump(&acr->subdev, &hdr); + + list_for_each_entry(lsfw, &acr->lsfw, head) { + if (lsfw->id != hdr.wpr.falcon_id) + continue; + + nvkm_robj(acr->wpr, hdr.wpr.lsb_offset, lsb, sizeof(*lsb)); + lsb_header_v2_dump(&acr->subdev, lsb); + + lsfw->func->bld_patch(acr, lsb->bl_data_off, adjust); + break; + } + + offset += sizeof(hdr); + } while (hdr.wpr.falcon_id != WPR_HEADER_V1_FALCON_ID_INVALID); + + kvfree(lsb); + return 0; +} + +static int +ga102_acr_wpr_build_lsb(struct nvkm_acr *acr, struct nvkm_acr_lsfw *lsfw) +{ + struct lsb_header_v2 *hdr; + int ret = 0; + + if (WARN_ON(lsfw->sig->size != sizeof(hdr->signature))) + return -EINVAL; + + hdr = kvzalloc(sizeof(*hdr), GFP_KERNEL); + if (!hdr) + return -ENOMEM; + + hdr->hdr.identifier = WPR_GENERIC_HEADER_ID_LSF_LSB_HEADER; + hdr->hdr.version = 2; + hdr->hdr.size = sizeof(*hdr); + + memcpy(&hdr->signature, lsfw->sig->data, lsfw->sig->size); + hdr->ucode_off = lsfw->offset.img; + hdr->ucode_size = lsfw->ucode_size; + hdr->data_size = lsfw->data_size; + hdr->bl_code_size = lsfw->bootloader_size; + hdr->bl_imem_off = lsfw->bootloader_imem_offset; + hdr->bl_data_off = lsfw->offset.bld; + hdr->bl_data_size = lsfw->bl_data_size; + hdr->app_code_off = lsfw->app_start_offset + lsfw->app_resident_code_offset; + hdr->app_code_size = ALIGN(lsfw->app_resident_code_size, 0x100); + hdr->app_data_off = lsfw->app_start_offset + lsfw->app_resident_data_offset; + hdr->app_data_size = ALIGN(lsfw->app_resident_data_size, 0x100); + hdr->app_imem_offset = lsfw->app_imem_offset; + hdr->app_dmem_offset = lsfw->app_dmem_offset; + hdr->flags = lsfw->func->flags; + hdr->monitor_code_offset = 0; + hdr->monitor_data_offset = 0; + hdr->manifest_offset = 0; + + if (lsfw->secure_bootloader) { + struct nvkm_falcon_fw fw = { + .fw.img = hdr->hs_fmc_params.pkc_signature, + .fw.name = "LSFW", + .func = &(const struct nvkm_falcon_fw_func) { + .signature = ga100_flcn_fw_signature, + }, + .sig_size = lsfw->sig_size, + .sig_nr = lsfw->sig_nr, + .sigs = lsfw->sigs, + .fuse_ver = lsfw->fuse_ver, + .engine_id = lsfw->engine_id, + .ucode_id = lsfw->ucode_id, + .falcon = lsfw->falcon, + + }; + + ret = nvkm_falcon_get(fw.falcon, &acr->subdev); + if (ret == 0) { + hdr->hs_fmc_params.hs_fmc = 1; + hdr->hs_fmc_params.pkc_algo = 0; + hdr->hs_fmc_params.pkc_algo_version = 1; + hdr->hs_fmc_params.engid_mask = lsfw->engine_id; + hdr->hs_fmc_params.ucode_id = lsfw->ucode_id; + hdr->hs_fmc_params.fuse_ver = lsfw->fuse_ver; + ret = nvkm_falcon_fw_patch(&fw); + nvkm_falcon_put(fw.falcon, &acr->subdev); + } + } + + nvkm_wobj(acr->wpr, lsfw->offset.lsb, hdr, sizeof(*hdr)); + kvfree(hdr); + return ret; +} + +static int +ga102_acr_wpr_build(struct nvkm_acr *acr, struct nvkm_acr_lsf *rtos) +{ + struct nvkm_acr_lsfw *lsfw; + struct wpr_header_v2 hdr; + u32 offset = 0; + int ret; + + /*XXX: shared sub-WPR headers, fill terminator for now. */ + nvkm_wo32(acr->wpr, 0x300, (2 << 16) | WPR_GENERIC_HEADER_ID_LSF_SHARED_SUB_WPR); + nvkm_wo32(acr->wpr, 0x304, 0x14); + nvkm_wo32(acr->wpr, 0x308, 0xffffffff); + nvkm_wo32(acr->wpr, 0x30c, 0); + nvkm_wo32(acr->wpr, 0x310, 0); + + /* Fill per-LSF structures. */ + list_for_each_entry(lsfw, &acr->lsfw, head) { + struct lsf_signature_v2 *sig = (void *)lsfw->sig->data; + + hdr.hdr.identifier = WPR_GENERIC_HEADER_ID_LSF_WPR_HEADER; + hdr.hdr.version = 2; + hdr.hdr.size = sizeof(hdr); + hdr.wpr.falcon_id = lsfw->id; + hdr.wpr.lsb_offset = lsfw->offset.lsb; + hdr.wpr.bootstrap_owner = NVKM_ACR_LSF_GSPLITE; + hdr.wpr.lazy_bootstrap = 1; + hdr.wpr.bin_version = sig->ls_ucode_version; + hdr.wpr.status = WPR_HEADER_V1_STATUS_COPY; + + /* Write WPR header. */ + nvkm_wobj(acr->wpr, offset, &hdr, sizeof(hdr)); + offset += sizeof(hdr); + + /* Write LSB header. */ + ret = ga102_acr_wpr_build_lsb(acr, lsfw); + if (ret) + return ret; + + /* Write ucode image. */ + nvkm_wobj(acr->wpr, lsfw->offset.img, + lsfw->img.data, + lsfw->img.size); + + /* Write bootloader data. */ + lsfw->func->bld_write(acr, lsfw->offset.bld, lsfw); + } + + /* Finalise WPR. */ + hdr.hdr.identifier = WPR_GENERIC_HEADER_ID_LSF_WPR_HEADER; + hdr.hdr.version = 2; + hdr.hdr.size = sizeof(hdr); + hdr.wpr.falcon_id = WPR_HEADER_V1_FALCON_ID_INVALID; + nvkm_wobj(acr->wpr, offset, &hdr, sizeof(hdr)); + return 0; +} + +static u32 +ga102_acr_wpr_layout(struct nvkm_acr *acr) +{ + struct nvkm_acr_lsfw *lsfw; + u32 wpr = 0; + + wpr += 21 /* MAX_LSF */ * sizeof(struct wpr_header_v2); + wpr = ALIGN(wpr, 256); + + wpr += 0x100; /* Shared sub-WPR headers. */ + + list_for_each_entry(lsfw, &acr->lsfw, head) { + wpr = ALIGN(wpr, 256); + lsfw->offset.lsb = wpr; + wpr += sizeof(struct lsb_header_v2); + + wpr = ALIGN(wpr, 4096); + lsfw->offset.img = wpr; + wpr += lsfw->img.size; + + wpr = ALIGN(wpr, 256); + lsfw->offset.bld = wpr; + lsfw->bl_data_size = ALIGN(lsfw->func->bld_size, 256); + wpr += lsfw->bl_data_size; + } + + return wpr; +} + +static int +ga102_acr_wpr_parse(struct nvkm_acr *acr) +{ + const struct wpr_header_v2 *hdr = (void *)acr->wpr_fw->data; + + while (hdr->wpr.falcon_id != WPR_HEADER_V1_FALCON_ID_INVALID) { + wpr_header_v2_dump(&acr->subdev, hdr); + if (!nvkm_acr_lsfw_add(NULL, acr, NULL, (hdr++)->wpr.falcon_id)) + return -ENOMEM; + } + + return 0; +} + +MODULE_FIRMWARE("nvidia/ga102/acr/ucode_unload.bin"); +MODULE_FIRMWARE("nvidia/ga103/acr/ucode_unload.bin"); +MODULE_FIRMWARE("nvidia/ga104/acr/ucode_unload.bin"); +MODULE_FIRMWARE("nvidia/ga106/acr/ucode_unload.bin"); +MODULE_FIRMWARE("nvidia/ga107/acr/ucode_unload.bin"); + +static const struct nvkm_acr_hsf_fwif +ga102_acr_unload_fwif[] = { + { 0, ga100_acr_hsfw_ctor, &ga102_flcn_fw, NVKM_ACR_HSF_SEC2 }, + {} +}; + +MODULE_FIRMWARE("nvidia/ga102/acr/ucode_asb.bin"); +MODULE_FIRMWARE("nvidia/ga103/acr/ucode_asb.bin"); +MODULE_FIRMWARE("nvidia/ga104/acr/ucode_asb.bin"); +MODULE_FIRMWARE("nvidia/ga106/acr/ucode_asb.bin"); +MODULE_FIRMWARE("nvidia/ga107/acr/ucode_asb.bin"); + +static const struct nvkm_acr_hsf_fwif +ga102_acr_asb_fwif[] = { + { 0, ga100_acr_hsfw_ctor, &ga102_flcn_fw, NVKM_ACR_HSF_GSP }, + {} +}; + +static const struct nvkm_falcon_fw_func +ga102_acr_ahesasc_0 = { + .signature = ga100_flcn_fw_signature, + .reset = gm200_flcn_fw_reset, + .setup = gp102_acr_load_setup, + .load = ga102_flcn_fw_load, + .boot = ga102_flcn_fw_boot, +}; + +MODULE_FIRMWARE("nvidia/ga102/acr/ucode_ahesasc.bin"); +MODULE_FIRMWARE("nvidia/ga103/acr/ucode_ahesasc.bin"); +MODULE_FIRMWARE("nvidia/ga104/acr/ucode_ahesasc.bin"); +MODULE_FIRMWARE("nvidia/ga106/acr/ucode_ahesasc.bin"); +MODULE_FIRMWARE("nvidia/ga107/acr/ucode_ahesasc.bin"); + +static const struct nvkm_acr_hsf_fwif +ga102_acr_ahesasc_fwif[] = { + { 0, ga100_acr_hsfw_ctor, &ga102_acr_ahesasc_0, NVKM_ACR_HSF_SEC2 }, + {} +}; + +static const struct nvkm_acr_func +ga102_acr = { + .ahesasc = ga102_acr_ahesasc_fwif, + .asb = ga102_acr_asb_fwif, + .unload = ga102_acr_unload_fwif, + .wpr_parse = ga102_acr_wpr_parse, + .wpr_layout = ga102_acr_wpr_layout, + .wpr_alloc = gp102_acr_wpr_alloc, + .wpr_patch = ga102_acr_wpr_patch, + .wpr_build = ga102_acr_wpr_build, + .wpr_check = ga100_acr_wpr_check, + .init = tu102_acr_init, +}; + +static int +ga102_acr_load(struct nvkm_acr *acr, int version, + const struct nvkm_acr_fwif *fwif) +{ + struct nvkm_subdev *subdev = &acr->subdev; + const struct nvkm_acr_hsf_fwif *hsfwif; + + hsfwif = nvkm_firmware_load(subdev, fwif->func->ahesasc, "AcrAHESASC", + acr, NULL, "acr/ucode_ahesasc", "AHESASC"); + if (IS_ERR(hsfwif)) + return PTR_ERR(hsfwif); + + hsfwif = nvkm_firmware_load(subdev, fwif->func->asb, "AcrASB", + acr, NULL, "acr/ucode_asb", "ASB"); + if (IS_ERR(hsfwif)) + return PTR_ERR(hsfwif); + + hsfwif = nvkm_firmware_load(subdev, fwif->func->unload, "AcrUnload", + acr, NULL, "acr/ucode_unload", "unload"); + if (IS_ERR(hsfwif)) + return PTR_ERR(hsfwif); + + return 0; +} + +static const struct nvkm_acr_fwif +ga102_acr_fwif[] = { + { 0, ga102_acr_load, &ga102_acr }, + { -1, gm200_acr_nofw, &gm200_acr }, + {} +}; + +int +ga102_acr_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, + struct nvkm_acr **pacr) +{ + return nvkm_acr_new_(ga102_acr_fwif, device, type, inst, pacr); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gm200.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gm200.c index 82b4c8e1457c..31079c947758 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gm200.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gm200.c @@ -46,7 +46,7 @@ gm200_acr_nofw(struct nvkm_acr *acr, int ver, const struct nvkm_acr_fwif *fwif) int gm200_acr_init(struct nvkm_acr *acr) { - return nvkm_acr_hsf_boot(acr, "load"); + return nvkm_acr_hsfw_boot(acr, "load"); } void @@ -61,7 +61,7 @@ gm200_acr_wpr_check(struct nvkm_acr *acr, u64 *start, u64 *limit) *limit = *limit + 0x20000; } -void +int gm200_acr_wpr_patch(struct nvkm_acr *acr, s64 adjust) { struct nvkm_subdev *subdev = &acr->subdev; @@ -86,6 +86,8 @@ gm200_acr_wpr_patch(struct nvkm_acr *acr, s64 adjust) } offset += sizeof(hdr); } while (hdr.falcon_id != WPR_HEADER_V0_FALCON_ID_INVALID); + + return 0; } void @@ -219,162 +221,50 @@ gm200_acr_wpr_parse(struct nvkm_acr *acr) return 0; } -void -gm200_acr_hsfw_bld(struct nvkm_acr *acr, struct nvkm_acr_hsf *hsf) +int +gm200_acr_hsfw_load_bld(struct nvkm_falcon_fw *fw) { struct flcn_bl_dmem_desc_v1 hsdesc = { .ctx_dma = FALCON_DMAIDX_VIRT, - .code_dma_base = hsf->vma->addr, - .non_sec_code_off = hsf->non_sec_addr, - .non_sec_code_size = hsf->non_sec_size, - .sec_code_off = hsf->sec_addr, - .sec_code_size = hsf->sec_size, + .code_dma_base = fw->vma->addr, + .non_sec_code_off = fw->nmem_base, + .non_sec_code_size = fw->nmem_size, + .sec_code_off = fw->imem_base, + .sec_code_size = fw->imem_size, .code_entry_point = 0, - .data_dma_base = hsf->vma->addr + hsf->data_addr, - .data_size = hsf->data_size, + .data_dma_base = fw->vma->addr + fw->dmem_base_img, + .data_size = fw->dmem_size, }; - flcn_bl_dmem_desc_v1_dump(&acr->subdev, &hsdesc); - - nvkm_falcon_load_dmem(hsf->falcon, &hsdesc, 0, sizeof(hsdesc), 0); -} - -int -gm200_acr_hsfw_boot(struct nvkm_acr *acr, struct nvkm_acr_hsf *hsf, - u32 intr_clear, u32 mbox0_ok) -{ - struct nvkm_subdev *subdev = &acr->subdev; - struct nvkm_device *device = subdev->device; - struct nvkm_falcon *falcon = hsf->falcon; - u32 mbox0, mbox1; - int ret; - - /* Reset falcon. */ - nvkm_falcon_reset(falcon); - nvkm_falcon_bind_context(falcon, acr->inst); + flcn_bl_dmem_desc_v1_dump(fw->falcon->user, &hsdesc); - /* Load bootloader into IMEM. */ - nvkm_falcon_load_imem(falcon, hsf->imem, - falcon->code.limit - hsf->imem_size, - hsf->imem_size, - hsf->imem_tag, - 0, false); - - /* Load bootloader data into DMEM. */ - hsf->func->bld(acr, hsf); - - /* Boot the falcon. */ - nvkm_mc_intr_mask(device, falcon->owner->type, falcon->owner->inst, false); - - nvkm_falcon_wr32(falcon, 0x040, 0xdeada5a5); - nvkm_falcon_set_start_addr(falcon, hsf->imem_tag << 8); - nvkm_falcon_start(falcon); - ret = nvkm_falcon_wait_for_halt(falcon, 100); - if (ret) - return ret; - - /* Check for successful completion. */ - mbox0 = nvkm_falcon_rd32(falcon, 0x040); - mbox1 = nvkm_falcon_rd32(falcon, 0x044); - nvkm_debug(subdev, "mailbox %08x %08x\n", mbox0, mbox1); - if (mbox0 && mbox0 != mbox0_ok) - return -EIO; - - nvkm_falcon_clear_interrupt(falcon, intr_clear); - nvkm_mc_intr_mask(device, falcon->owner->type, falcon->owner->inst, true); - return ret; + return nvkm_falcon_pio_wr(fw->falcon, (u8 *)&hsdesc, 0, 0, DMEM, 0, sizeof(hsdesc), 0, 0); } int -gm200_acr_hsfw_load(struct nvkm_acr *acr, struct nvkm_acr_hsfw *hsfw, - struct nvkm_falcon *falcon) +gm200_acr_hsfw_ctor(struct nvkm_acr *acr, const char *bl, const char *fw, const char *name, int ver, + const struct nvkm_acr_hsf_fwif *fwif) { - struct nvkm_subdev *subdev = &acr->subdev; - struct nvkm_acr_hsf *hsf; - int ret; - - /* Patch the appropriate signature (production/debug) into the FW - * image, as determined by the mode the falcon is in. - */ - ret = nvkm_falcon_get(falcon, subdev); - if (ret) - return ret; - - if (hsfw->sig.patch_loc) { - if (!falcon->debug) { - nvkm_debug(subdev, "patching production signature\n"); - memcpy(hsfw->image + hsfw->sig.patch_loc, - hsfw->sig.prod.data, - hsfw->sig.prod.size); - } else { - nvkm_debug(subdev, "patching debug signature\n"); - memcpy(hsfw->image + hsfw->sig.patch_loc, - hsfw->sig.dbg.data, - hsfw->sig.dbg.size); - } - } - - nvkm_falcon_put(falcon, subdev); + struct nvkm_acr_hsfw *hsfw; - if (!(hsf = kzalloc(sizeof(*hsf), GFP_KERNEL))) + if (!(hsfw = kzalloc(sizeof(*hsfw), GFP_KERNEL))) return -ENOMEM; - hsf->func = hsfw->func; - hsf->name = hsfw->name; - list_add_tail(&hsf->head, &acr->hsf); - - hsf->imem_size = hsfw->imem_size; - hsf->imem_tag = hsfw->imem_tag; - hsf->imem = kmemdup(hsfw->imem, hsfw->imem_size, GFP_KERNEL); - if (!hsf->imem) - return -ENOMEM; - - hsf->non_sec_addr = hsfw->non_sec_addr; - hsf->non_sec_size = hsfw->non_sec_size; - hsf->sec_addr = hsfw->sec_addr; - hsf->sec_size = hsfw->sec_size; - hsf->data_addr = hsfw->data_addr; - hsf->data_size = hsfw->data_size; - - /* Make the FW image accessible to the HS bootloader. */ - ret = nvkm_memory_new(subdev->device, NVKM_MEM_TARGET_INST, - hsfw->image_size, 0x1000, false, &hsf->ucode); - if (ret) - return ret; - - nvkm_kmap(hsf->ucode); - nvkm_wobj(hsf->ucode, 0, hsfw->image, hsfw->image_size); - nvkm_done(hsf->ucode); - - ret = nvkm_vmm_get(acr->vmm, 12, nvkm_memory_size(hsf->ucode), - &hsf->vma); - if (ret) - return ret; - - ret = nvkm_memory_map(hsf->ucode, 0, acr->vmm, hsf->vma, NULL, 0); - if (ret) - return ret; - hsf->falcon = falcon; - return 0; -} + hsfw->falcon_id = fwif->falcon_id; + hsfw->boot_mbox0 = fwif->boot_mbox0; + hsfw->intr_clear = fwif->intr_clear; + list_add_tail(&hsfw->head, &acr->hsfw); -int -gm200_acr_unload_boot(struct nvkm_acr *acr, struct nvkm_acr_hsf *hsf) -{ - return gm200_acr_hsfw_boot(acr, hsf, 0, 0x1d); -} - -int -gm200_acr_unload_load(struct nvkm_acr *acr, struct nvkm_acr_hsfw *hsfw) -{ - return gm200_acr_hsfw_load(acr, hsfw, &acr->subdev.device->pmu->falcon); + return nvkm_falcon_fw_ctor_hs(fwif->func, name, &acr->subdev, bl, fw, ver, NULL, &hsfw->fw); } -const struct nvkm_acr_hsf_func +const struct nvkm_falcon_fw_func gm200_acr_unload_0 = { - .load = gm200_acr_unload_load, - .boot = gm200_acr_unload_boot, - .bld = gm200_acr_hsfw_bld, + .signature = gm200_flcn_fw_signature, + .reset = gm200_flcn_fw_reset, + .load = gm200_flcn_fw_load, + .load_bld = gm200_acr_hsfw_load_bld, + .boot = gm200_flcn_fw_boot, }; MODULE_FIRMWARE("nvidia/gm200/acr/ucode_unload.bin"); @@ -384,20 +274,15 @@ MODULE_FIRMWARE("nvidia/gp100/acr/ucode_unload.bin"); static const struct nvkm_acr_hsf_fwif gm200_acr_unload_fwif[] = { - { 0, nvkm_acr_hsfw_load, &gm200_acr_unload_0 }, + { 0, gm200_acr_hsfw_ctor, &gm200_acr_unload_0, NVKM_ACR_HSF_PMU, 0, 0x00000010 }, {} }; -int -gm200_acr_load_boot(struct nvkm_acr *acr, struct nvkm_acr_hsf *hsf) -{ - return gm200_acr_hsfw_boot(acr, hsf, 0x10, 0); -} - static int -gm200_acr_load_load(struct nvkm_acr *acr, struct nvkm_acr_hsfw *hsfw) +gm200_acr_load_setup(struct nvkm_falcon_fw *fw) { - struct flcn_acr_desc *desc = (void *)&hsfw->image[hsfw->data_addr]; + struct flcn_acr_desc *desc = (void *)&fw->fw.img[fw->dmem_base_img]; + struct nvkm_acr *acr = fw->falcon->owner->device->acr; desc->wpr_region_id = 1; desc->regions.no_regions = 2; @@ -408,15 +293,17 @@ gm200_acr_load_load(struct nvkm_acr *acr, struct nvkm_acr_hsfw *hsfw) desc->regions.region_props[0].write_mask = 0xc; desc->regions.region_props[0].client_mask = 0x2; flcn_acr_desc_dump(&acr->subdev, desc); - - return gm200_acr_hsfw_load(acr, hsfw, &acr->subdev.device->pmu->falcon); + return 0; } -static const struct nvkm_acr_hsf_func +static const struct nvkm_falcon_fw_func gm200_acr_load_0 = { - .load = gm200_acr_load_load, - .boot = gm200_acr_load_boot, - .bld = gm200_acr_hsfw_bld, + .signature = gm200_flcn_fw_signature, + .reset = gm200_flcn_fw_reset, + .setup = gm200_acr_load_setup, + .load = gm200_flcn_fw_load, + .load_bld = gm200_acr_hsfw_load_bld, + .boot = gm200_flcn_fw_boot, }; MODULE_FIRMWARE("nvidia/gm200/acr/bl.bin"); @@ -433,7 +320,7 @@ MODULE_FIRMWARE("nvidia/gp100/acr/ucode_load.bin"); static const struct nvkm_acr_hsf_fwif gm200_acr_load_fwif[] = { - { 0, nvkm_acr_hsfw_load, &gm200_acr_load_0 }, + { 0, gm200_acr_hsfw_ctor, &gm200_acr_load_0, NVKM_ACR_HSF_PMU, 0, 0x00000010 }, {} }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gm20b.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gm20b.c index 54e996f2f630..ef5fb79128b1 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gm20b.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gm20b.c @@ -45,43 +45,47 @@ gm20b_acr_wpr_alloc(struct nvkm_acr *acr, u32 wpr_size) wpr_size, 0, true, &acr->wpr); } -static void -gm20b_acr_load_bld(struct nvkm_acr *acr, struct nvkm_acr_hsf *hsf) +static int +gm20b_acr_hsfw_load_bld(struct nvkm_falcon_fw *fw) { struct flcn_bl_dmem_desc hsdesc = { .ctx_dma = FALCON_DMAIDX_VIRT, - .code_dma_base = hsf->vma->addr >> 8, - .non_sec_code_off = hsf->non_sec_addr, - .non_sec_code_size = hsf->non_sec_size, - .sec_code_off = hsf->sec_addr, - .sec_code_size = hsf->sec_size, + .code_dma_base = fw->vma->addr >> 8, + .non_sec_code_off = fw->nmem_base, + .non_sec_code_size = fw->nmem_size, + .sec_code_off = fw->imem_base, + .sec_code_size = fw->imem_size, .code_entry_point = 0, - .data_dma_base = (hsf->vma->addr + hsf->data_addr) >> 8, - .data_size = hsf->data_size, + .data_dma_base = (fw->vma->addr + fw->dmem_base_img) >> 8, + .data_size = fw->dmem_size, }; - flcn_bl_dmem_desc_dump(&acr->subdev, &hsdesc); + flcn_bl_dmem_desc_dump(fw->falcon->user, &hsdesc); - nvkm_falcon_load_dmem(hsf->falcon, &hsdesc, 0, sizeof(hsdesc), 0); + return nvkm_falcon_pio_wr(fw->falcon, (u8 *)&hsdesc, 0, 0, DMEM, 0, sizeof(hsdesc), 0, 0); } + static int -gm20b_acr_load_load(struct nvkm_acr *acr, struct nvkm_acr_hsfw *hsfw) +gm20b_acr_load_setup(struct nvkm_falcon_fw *fw) { - struct flcn_acr_desc *desc = (void *)&hsfw->image[hsfw->data_addr]; + struct flcn_acr_desc *desc = (void *)&fw->fw.img[fw->dmem_base_img]; + struct nvkm_acr *acr = fw->falcon->owner->device->acr; desc->ucode_blob_base = nvkm_memory_addr(acr->wpr); desc->ucode_blob_size = nvkm_memory_size(acr->wpr); flcn_acr_desc_dump(&acr->subdev, desc); - - return gm200_acr_hsfw_load(acr, hsfw, &acr->subdev.device->pmu->falcon); + return 0; } -const struct nvkm_acr_hsf_func +const struct nvkm_falcon_fw_func gm20b_acr_load_0 = { - .load = gm20b_acr_load_load, - .boot = gm200_acr_load_boot, - .bld = gm20b_acr_load_bld, + .signature = gm200_flcn_fw_signature, + .reset = gm200_flcn_fw_reset, + .setup = gm20b_acr_load_setup, + .load = gm200_flcn_fw_load, + .load_bld = gm20b_acr_hsfw_load_bld, + .boot = gm200_flcn_fw_boot, }; #if IS_ENABLED(CONFIG_ARCH_TEGRA_210_SOC) @@ -91,7 +95,7 @@ MODULE_FIRMWARE("nvidia/gm20b/acr/ucode_load.bin"); static const struct nvkm_acr_hsf_fwif gm20b_acr_load_fwif[] = { - { 0, nvkm_acr_hsfw_load, &gm20b_acr_load_0 }, + { 0, gm200_acr_hsfw_ctor, &gm20b_acr_load_0, NVKM_ACR_HSF_PMU, 0, 0x10 }, {} }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp102.c index fd97a935a380..084f28449e52 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp102.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp102.c @@ -29,7 +29,7 @@ #include <nvfw/acr.h> #include <nvfw/flcn.h> -void +int gp102_acr_wpr_patch(struct nvkm_acr *acr, s64 adjust) { struct wpr_header_v1 hdr; @@ -54,6 +54,8 @@ gp102_acr_wpr_patch(struct nvkm_acr *acr, s64 adjust) offset += sizeof(hdr); } while (hdr.falcon_id != WPR_HEADER_V1_FALCON_ID_INVALID); + + return 0; } int @@ -187,14 +189,15 @@ MODULE_FIRMWARE("nvidia/gp107/acr/ucode_unload.bin"); static const struct nvkm_acr_hsf_fwif gp102_acr_unload_fwif[] = { - { 0, nvkm_acr_hsfw_load, &gm200_acr_unload_0 }, + { 0, gm200_acr_hsfw_ctor, &gm200_acr_unload_0, NVKM_ACR_HSF_PMU, 0x1d, 0x00000010 }, {} }; int -gp102_acr_load_load(struct nvkm_acr *acr, struct nvkm_acr_hsfw *hsfw) +gp102_acr_load_setup(struct nvkm_falcon_fw *fw) { - struct flcn_acr_desc_v1 *desc = (void *)&hsfw->image[hsfw->data_addr]; + struct flcn_acr_desc_v1 *desc = (void *)&fw->fw.img[fw->dmem_base_img]; + struct nvkm_acr *acr = fw->falcon->owner->device->acr; desc->wpr_region_id = 1; desc->regions.no_regions = 2; @@ -204,19 +207,19 @@ gp102_acr_load_load(struct nvkm_acr *acr, struct nvkm_acr_hsfw *hsfw) desc->regions.region_props[0].read_mask = 0xf; desc->regions.region_props[0].write_mask = 0xc; desc->regions.region_props[0].client_mask = 0x2; - desc->regions.region_props[0].shadow_mem_start_addr = - acr->shadow_start >> 8; + desc->regions.region_props[0].shadow_mem_start_addr = acr->shadow_start >> 8; flcn_acr_desc_v1_dump(&acr->subdev, desc); - - return gm200_acr_hsfw_load(acr, hsfw, - &acr->subdev.device->sec2->falcon); + return 0; } -static const struct nvkm_acr_hsf_func +static const struct nvkm_falcon_fw_func gp102_acr_load_0 = { - .load = gp102_acr_load_load, - .boot = gm200_acr_load_boot, - .bld = gm200_acr_hsfw_bld, + .signature = gm200_flcn_fw_signature, + .reset = gm200_flcn_fw_reset, + .setup = gp102_acr_load_setup, + .load = gm200_flcn_fw_load, + .load_bld = gm200_acr_hsfw_load_bld, + .boot = gm200_flcn_fw_boot, }; MODULE_FIRMWARE("nvidia/gp102/acr/bl.bin"); @@ -233,7 +236,7 @@ MODULE_FIRMWARE("nvidia/gp107/acr/ucode_load.bin"); static const struct nvkm_acr_hsf_fwif gp102_acr_load_fwif[] = { - { 0, nvkm_acr_hsfw_load, &gp102_acr_load_0 }, + { 0, gm200_acr_hsfw_ctor, &gp102_acr_load_0, NVKM_ACR_HSF_SEC2, 0, 0x00000010 }, {} }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp108.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp108.c index 373d638a2177..6ab9d4959c17 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp108.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp108.c @@ -25,63 +25,62 @@ #include <nvfw/flcn.h> -void -gp108_acr_hsfw_bld(struct nvkm_acr *acr, struct nvkm_acr_hsf *hsf) +int +gp108_acr_hsfw_load_bld(struct nvkm_falcon_fw *fw) { struct flcn_bl_dmem_desc_v2 hsdesc = { .ctx_dma = FALCON_DMAIDX_VIRT, - .code_dma_base = hsf->vma->addr, - .non_sec_code_off = hsf->non_sec_addr, - .non_sec_code_size = hsf->non_sec_size, - .sec_code_off = hsf->sec_addr, - .sec_code_size = hsf->sec_size, + .code_dma_base = fw->vma->addr, + .non_sec_code_off = fw->nmem_base, + .non_sec_code_size = fw->nmem_size, + .sec_code_off = fw->imem_base, + .sec_code_size = fw->imem_size, .code_entry_point = 0, - .data_dma_base = hsf->vma->addr + hsf->data_addr, - .data_size = hsf->data_size, + .data_dma_base = fw->vma->addr + fw->dmem_base_img, + .data_size = fw->dmem_size, .argc = 0, .argv = 0, }; - flcn_bl_dmem_desc_v2_dump(&acr->subdev, &hsdesc); + flcn_bl_dmem_desc_v2_dump(fw->falcon->user, &hsdesc); - nvkm_falcon_load_dmem(hsf->falcon, &hsdesc, 0, sizeof(hsdesc), 0); + return nvkm_falcon_pio_wr(fw->falcon, (u8 *)&hsdesc, 0, 0, DMEM, 0, sizeof(hsdesc), 0, 0); } -const struct nvkm_acr_hsf_func -gp108_acr_unload_0 = { - .load = gm200_acr_unload_load, - .boot = gm200_acr_unload_boot, - .bld = gp108_acr_hsfw_bld, +const struct nvkm_falcon_fw_func +gp108_acr_hsfw_0 = { + .signature = gm200_flcn_fw_signature, + .reset = gm200_flcn_fw_reset, + .load = gm200_flcn_fw_load, + .load_bld = gp108_acr_hsfw_load_bld, + .boot = gm200_flcn_fw_boot, }; MODULE_FIRMWARE("nvidia/gp108/acr/unload_bl.bin"); MODULE_FIRMWARE("nvidia/gp108/acr/ucode_unload.bin"); -MODULE_FIRMWARE("nvidia/gv100/acr/unload_bl.bin"); -MODULE_FIRMWARE("nvidia/gv100/acr/ucode_unload.bin"); - static const struct nvkm_acr_hsf_fwif gp108_acr_unload_fwif[] = { - { 0, nvkm_acr_hsfw_load, &gp108_acr_unload_0 }, + { 0, gm200_acr_hsfw_ctor, &gp108_acr_hsfw_0, NVKM_ACR_HSF_PMU, 0x1d, 0x00000010 }, {} }; -static const struct nvkm_acr_hsf_func +const struct nvkm_falcon_fw_func gp108_acr_load_0 = { - .load = gp102_acr_load_load, - .boot = gm200_acr_load_boot, - .bld = gp108_acr_hsfw_bld, + .signature = gm200_flcn_fw_signature, + .reset = gm200_flcn_fw_reset, + .setup = gp102_acr_load_setup, + .load = gm200_flcn_fw_load, + .load_bld = gp108_acr_hsfw_load_bld, + .boot = gm200_flcn_fw_boot, }; MODULE_FIRMWARE("nvidia/gp108/acr/bl.bin"); MODULE_FIRMWARE("nvidia/gp108/acr/ucode_load.bin"); -MODULE_FIRMWARE("nvidia/gv100/acr/bl.bin"); -MODULE_FIRMWARE("nvidia/gv100/acr/ucode_load.bin"); - static const struct nvkm_acr_hsf_fwif gp108_acr_load_fwif[] = { - { 0, nvkm_acr_hsfw_load, &gp108_acr_load_0 }, + { 0, gm200_acr_hsfw_ctor, &gp108_acr_load_0, NVKM_ACR_HSF_SEC2, 0, 0x00000010 }, {} }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp10b.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp10b.c index f03ba028867b..a3422ab6deab 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp10b.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gp10b.c @@ -28,7 +28,7 @@ MODULE_FIRMWARE("nvidia/gp10b/acr/ucode_load.bin"); static const struct nvkm_acr_hsf_fwif gp10b_acr_load_fwif[] = { - { 0, nvkm_acr_hsfw_load, &gm20b_acr_load_0 }, + { 0, gm200_acr_hsfw_ctor, &gm20b_acr_load_0, NVKM_ACR_HSF_PMU, 0, 0x00000010 }, {} }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gv100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gv100.c new file mode 100644 index 000000000000..4c5ca6b40027 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/gv100.c @@ -0,0 +1,67 @@ +/* + * Copyright 2022 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +MODULE_FIRMWARE("nvidia/gv100/acr/unload_bl.bin"); +MODULE_FIRMWARE("nvidia/gv100/acr/ucode_unload.bin"); + +static const struct nvkm_acr_hsf_fwif +gv100_acr_unload_fwif[] = { + { 0, gm200_acr_hsfw_ctor, &gp108_acr_hsfw_0, NVKM_ACR_HSF_PMU, 0, 0x00000000 }, + {} +}; + +MODULE_FIRMWARE("nvidia/gv100/acr/bl.bin"); +MODULE_FIRMWARE("nvidia/gv100/acr/ucode_load.bin"); + +static const struct nvkm_acr_hsf_fwif +gv100_acr_load_fwif[] = { + { 0, gm200_acr_hsfw_ctor, &gp108_acr_load_0, NVKM_ACR_HSF_SEC2, 0, 0x00000010 }, + {} +}; + +static const struct nvkm_acr_func +gv100_acr = { + .load = gv100_acr_load_fwif, + .unload = gv100_acr_unload_fwif, + .wpr_parse = gp102_acr_wpr_parse, + .wpr_layout = gp102_acr_wpr_layout, + .wpr_alloc = gp102_acr_wpr_alloc, + .wpr_build = gp102_acr_wpr_build, + .wpr_patch = gp102_acr_wpr_patch, + .wpr_check = gm200_acr_wpr_check, + .init = gm200_acr_init, +}; + +static const struct nvkm_acr_fwif +gv100_acr_fwif[] = { + { 0, gp102_acr_load, &gv100_acr }, + { -1, gm200_acr_nofw, &gm200_acr }, + {} +}; + +int +gv100_acr_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, + struct nvkm_acr **pacr) +{ + return nvkm_acr_new_(gv100_acr_fwif, device, type, inst, pacr); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/hsfw.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/hsfw.c deleted file mode 100644 index a6ea89a5d51a..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/hsfw.c +++ /dev/null @@ -1,177 +0,0 @@ -/* - * Copyright 2019 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - */ -#include "priv.h" - -#include <core/firmware.h> - -#include <nvfw/fw.h> -#include <nvfw/hs.h> - -static void -nvkm_acr_hsfw_del(struct nvkm_acr_hsfw *hsfw) -{ - list_del(&hsfw->head); - kfree(hsfw->imem); - kfree(hsfw->image); - kfree(hsfw->sig.prod.data); - kfree(hsfw->sig.dbg.data); - kfree(hsfw); -} - -void -nvkm_acr_hsfw_del_all(struct nvkm_acr *acr) -{ - struct nvkm_acr_hsfw *hsfw, *hsft; - list_for_each_entry_safe(hsfw, hsft, &acr->hsfw, head) { - nvkm_acr_hsfw_del(hsfw); - } -} - -static int -nvkm_acr_hsfw_load_image(struct nvkm_acr *acr, const char *name, int ver, - struct nvkm_acr_hsfw *hsfw) -{ - struct nvkm_subdev *subdev = &acr->subdev; - const struct firmware *fw; - const struct nvfw_bin_hdr *hdr; - const struct nvfw_hs_header *fwhdr; - const struct nvfw_hs_load_header *lhdr; - u32 loc, sig; - int ret; - - ret = nvkm_firmware_get(subdev, name, ver, &fw); - if (ret < 0) - return ret; - - hdr = nvfw_bin_hdr(subdev, fw->data); - fwhdr = nvfw_hs_header(subdev, fw->data + hdr->header_offset); - - /* Earlier FW releases by NVIDIA for Nouveau's use aren't in NVIDIA's - * standard format, and don't have the indirection seen in the 0x10de - * case. - */ - switch (hdr->bin_magic) { - case 0x000010de: - loc = *(u32 *)(fw->data + fwhdr->patch_loc); - sig = *(u32 *)(fw->data + fwhdr->patch_sig); - break; - case 0x3b1d14f0: - loc = fwhdr->patch_loc; - sig = fwhdr->patch_sig; - break; - default: - ret = -EINVAL; - goto done; - } - - lhdr = nvfw_hs_load_header(subdev, fw->data + fwhdr->hdr_offset); - - if (!(hsfw->image = kmalloc(hdr->data_size, GFP_KERNEL))) { - ret = -ENOMEM; - goto done; - } - - memcpy(hsfw->image, fw->data + hdr->data_offset, hdr->data_size); - hsfw->image_size = hdr->data_size; - hsfw->non_sec_addr = lhdr->non_sec_code_off; - hsfw->non_sec_size = lhdr->non_sec_code_size; - hsfw->sec_addr = lhdr->apps[0]; - hsfw->sec_size = lhdr->apps[lhdr->num_apps]; - hsfw->data_addr = lhdr->data_dma_base; - hsfw->data_size = lhdr->data_size; - - hsfw->sig.prod.size = fwhdr->sig_prod_size; - hsfw->sig.prod.data = kmemdup(fw->data + fwhdr->sig_prod_offset + sig, - hsfw->sig.prod.size, GFP_KERNEL); - if (!hsfw->sig.prod.data) { - ret = -ENOMEM; - goto done; - } - - hsfw->sig.dbg.size = fwhdr->sig_dbg_size; - hsfw->sig.dbg.data = kmemdup(fw->data + fwhdr->sig_dbg_offset + sig, - hsfw->sig.dbg.size, GFP_KERNEL); - if (!hsfw->sig.dbg.data) { - ret = -ENOMEM; - goto done; - } - - hsfw->sig.patch_loc = loc; -done: - nvkm_firmware_put(fw); - return ret; -} - -static int -nvkm_acr_hsfw_load_bl(struct nvkm_acr *acr, const char *name, int ver, - struct nvkm_acr_hsfw *hsfw) -{ - struct nvkm_subdev *subdev = &acr->subdev; - const struct nvfw_bin_hdr *hdr; - const struct nvfw_bl_desc *desc; - const struct firmware *fw; - u8 *data; - int ret; - - ret = nvkm_firmware_get(subdev, name, ver, &fw); - if (ret) - return ret; - - hdr = nvfw_bin_hdr(subdev, fw->data); - desc = nvfw_bl_desc(subdev, fw->data + hdr->header_offset); - data = (void *)fw->data + hdr->data_offset; - - hsfw->imem_size = desc->code_size; - hsfw->imem_tag = desc->start_tag; - hsfw->imem = kmemdup(data + desc->code_off, desc->code_size, GFP_KERNEL); - nvkm_firmware_put(fw); - if (!hsfw->imem) - return -ENOMEM; - else - return 0; -} - -int -nvkm_acr_hsfw_load(struct nvkm_acr *acr, const char *bl, const char *fw, - const char *name, int version, - const struct nvkm_acr_hsf_fwif *fwif) -{ - struct nvkm_acr_hsfw *hsfw; - int ret; - - if (!(hsfw = kzalloc(sizeof(*hsfw), GFP_KERNEL))) - return -ENOMEM; - - hsfw->func = fwif->func; - hsfw->name = name; - list_add_tail(&hsfw->head, &acr->hsfw); - - ret = nvkm_acr_hsfw_load_bl(acr, bl, version, hsfw); - if (ret) - goto done; - - ret = nvkm_acr_hsfw_load_image(acr, fw, version, hsfw); -done: - if (ret) - nvkm_acr_hsfw_del(hsfw); - return ret; -} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/lsfw.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/lsfw.c index 9b1cf6711ae9..f36a359d4531 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/lsfw.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/lsfw.c @@ -29,6 +29,7 @@ void nvkm_acr_lsfw_del(struct nvkm_acr_lsfw *lsfw) { nvkm_blob_dtor(&lsfw->img); + kfree(lsfw->sigs); nvkm_firmware_put(lsfw->sig); list_del(&lsfw->head); kfree(lsfw); @@ -177,6 +178,75 @@ nvkm_acr_lsfw_load_sig_image_desc_v1(struct nvkm_subdev *subdev, } int +nvkm_acr_lsfw_load_sig_image_desc_v2(struct nvkm_subdev *subdev, + struct nvkm_falcon *falcon, + enum nvkm_acr_lsf_id id, + const char *path, int ver, + const struct nvkm_acr_lsf_func *func) +{ + const struct firmware *fw; + struct nvkm_acr_lsfw *lsfw; + const struct nvfw_ls_desc_v2 *desc; + int ret = 0; + + lsfw = nvkm_acr_lsfw_load_sig_image_desc_(subdev, falcon, id, path, ver, func, &fw); + if (IS_ERR(lsfw)) + return PTR_ERR(lsfw); + + desc = nvfw_ls_desc_v2(subdev, fw->data); + + lsfw->secure_bootloader = desc->secure_bootloader; + lsfw->bootloader_size = ALIGN(desc->bootloader_size, 256); + lsfw->bootloader_imem_offset = desc->bootloader_imem_offset; + + lsfw->app_size = ALIGN(desc->app_size, 256); + lsfw->app_start_offset = desc->app_start_offset; + lsfw->app_imem_entry = desc->app_imem_entry; + lsfw->app_resident_code_offset = desc->app_resident_code_offset; + lsfw->app_resident_code_size = desc->app_resident_code_size; + lsfw->app_resident_data_offset = desc->app_resident_data_offset; + lsfw->app_resident_data_size = desc->app_resident_data_size; + lsfw->app_imem_offset = desc->app_imem_offset; + lsfw->app_dmem_offset = desc->app_dmem_offset; + + lsfw->ucode_size = ALIGN(lsfw->app_resident_data_offset, 256) + lsfw->bootloader_size; + lsfw->data_size = lsfw->app_size + lsfw->bootloader_size - lsfw->ucode_size; + + nvkm_firmware_put(fw); + + if (lsfw->secure_bootloader) { + const struct firmware *hsbl; + const struct nvfw_ls_hsbl_bin_hdr *hdr; + const struct nvfw_ls_hsbl_hdr *hshdr; + u32 loc, sig, cnt, *meta; + + ret = nvkm_firmware_load_name(subdev, path, "hs_bl_sig", ver, &hsbl); + if (ret) + return ret; + + hdr = nvfw_ls_hsbl_bin_hdr(subdev, hsbl->data); + hshdr = nvfw_ls_hsbl_hdr(subdev, hsbl->data + hdr->header_offset); + meta = (u32 *)(hsbl->data + hshdr->meta_data_offset); + loc = *(u32 *)(hsbl->data + hshdr->patch_loc); + sig = *(u32 *)(hsbl->data + hshdr->patch_sig); + cnt = *(u32 *)(hsbl->data + hshdr->num_sig); + + lsfw->fuse_ver = meta[0]; + lsfw->engine_id = meta[1]; + lsfw->ucode_id = meta[2]; + lsfw->sig_size = hshdr->sig_prod_size / cnt; + lsfw->sig_nr = cnt; + lsfw->sigs = kmemdup(hsbl->data + hshdr->sig_prod_offset + sig, + lsfw->sig_nr * lsfw->sig_size, GFP_KERNEL); + nvkm_firmware_put(hsbl); + if (!lsfw->sigs) + ret = -ENOMEM; + } + + return ret; +} + +int nvkm_acr_lsfw_load_bl_inst_data_sig(struct nvkm_subdev *subdev, struct nvkm_falcon *falcon, enum nvkm_acr_lsf_id id, @@ -251,3 +321,78 @@ done: nvkm_firmware_put(bl); return ret; } + +int +nvkm_acr_lsfw_load_bl_sig_net(struct nvkm_subdev *subdev, + struct nvkm_falcon *falcon, + enum nvkm_acr_lsf_id id, + const char *path, int ver, + const struct nvkm_acr_lsf_func *func, + const void *inst_data, u32 inst_size, + const void *data_data, u32 data_size) +{ + struct nvkm_acr *acr = subdev->device->acr; + struct nvkm_acr_lsfw *lsfw; + const struct firmware _inst = { .data = inst_data, .size = inst_size }; + const struct firmware _data = { .data = data_data, .size = data_size }; + const struct firmware *bl = NULL, *inst = &_inst, *data = &_data; + const struct { + int bin_magic; + int bin_version; + int bin_size; + int header_offset; + int header_size; + } *hdr; + u32 *bldata; + int ret; + + if (IS_ERR((lsfw = nvkm_acr_lsfw_add(func, acr, falcon, id)))) + return PTR_ERR(lsfw); + + ret = nvkm_firmware_load_name(subdev, path, "bl", ver, &bl); + if (ret) + goto done; + + hdr = (const void *)bl->data; + bldata = (void *)(bl->data + hdr->header_offset); + + ret = nvkm_firmware_load_name(subdev, path, "sig", ver, &lsfw->sig); + if (ret) + goto done; + + lsfw->bootloader_size = ALIGN(hdr->header_size, 256); + lsfw->bootloader_imem_offset = func->bl_entry; + + lsfw->app_start_offset = lsfw->bootloader_size; + lsfw->app_imem_entry = 0; + lsfw->app_resident_code_offset = 0; + lsfw->app_resident_code_size = ALIGN(inst->size, 256); + lsfw->app_resident_data_offset = lsfw->app_resident_code_size; + lsfw->app_resident_data_size = ALIGN(data->size, 256); + lsfw->app_imem_offset = 0; + lsfw->app_dmem_offset = 0; + lsfw->app_size = lsfw->app_resident_code_size + lsfw->app_resident_data_size; + + lsfw->img.size = lsfw->bootloader_size + lsfw->app_size; + if (!(lsfw->img.data = kzalloc(lsfw->img.size, GFP_KERNEL))) { + ret = -ENOMEM; + goto done; + } + + memcpy(lsfw->img.data, bldata, lsfw->bootloader_size); + memcpy(lsfw->img.data + lsfw->app_start_offset + + lsfw->app_resident_code_offset, inst->data, inst->size); + memcpy(lsfw->img.data + lsfw->app_start_offset + + lsfw->app_resident_data_offset, data->data, data->size); + + lsfw->ucode_size = ALIGN(lsfw->app_resident_data_offset, 256) + + lsfw->bootloader_size; + lsfw->data_size = lsfw->app_size + lsfw->bootloader_size - + lsfw->ucode_size; + +done: + if (ret) + nvkm_acr_lsfw_del(lsfw); + nvkm_firmware_put(bl); + return ret; +} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/priv.h b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/priv.h index c30b841c9d35..4881c8ba3880 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/priv.h @@ -24,7 +24,7 @@ struct nvkm_acr_func { u32 (*wpr_layout)(struct nvkm_acr *); int (*wpr_alloc)(struct nvkm_acr *, u32 wpr_size); int (*wpr_build)(struct nvkm_acr *, struct nvkm_acr_lsf *rtos); - void (*wpr_patch)(struct nvkm_acr *, s64 adjust); + int (*wpr_patch)(struct nvkm_acr *, s64 adjust); void (*wpr_check)(struct nvkm_acr *, u64 *start, u64 *limit); int (*init)(struct nvkm_acr *); void (*fini)(struct nvkm_acr *); @@ -35,7 +35,7 @@ extern const struct nvkm_acr_func gm200_acr; int gm200_acr_wpr_parse(struct nvkm_acr *); u32 gm200_acr_wpr_layout(struct nvkm_acr *); int gm200_acr_wpr_build(struct nvkm_acr *, struct nvkm_acr_lsf *); -void gm200_acr_wpr_patch(struct nvkm_acr *, s64); +int gm200_acr_wpr_patch(struct nvkm_acr *, s64); void gm200_acr_wpr_check(struct nvkm_acr *, u64 *, u64 *); void gm200_acr_wpr_build_lsb_tail(struct nvkm_acr_lsfw *, struct lsb_header_tail *); @@ -48,96 +48,60 @@ u32 gp102_acr_wpr_layout(struct nvkm_acr *); int gp102_acr_wpr_alloc(struct nvkm_acr *, u32 wpr_size); int gp102_acr_wpr_build(struct nvkm_acr *, struct nvkm_acr_lsf *); int gp102_acr_wpr_build_lsb(struct nvkm_acr *, struct nvkm_acr_lsfw *); -void gp102_acr_wpr_patch(struct nvkm_acr *, s64); +int gp102_acr_wpr_patch(struct nvkm_acr *, s64); + +int tu102_acr_init(struct nvkm_acr *); + +void ga100_acr_wpr_check(struct nvkm_acr *, u64 *, u64 *); struct nvkm_acr_hsfw { - const struct nvkm_acr_hsf_func *func; - const char *name; - struct list_head head; + struct nvkm_falcon_fw fw; + + enum nvkm_acr_hsf_id { + NVKM_ACR_HSF_PMU, + NVKM_ACR_HSF_SEC2, + NVKM_ACR_HSF_GSP, + } falcon_id; + u32 boot_mbox0; + u32 intr_clear; - u32 imem_size; - u32 imem_tag; - u32 *imem; - - u8 *image; - u32 image_size; - u32 non_sec_addr; - u32 non_sec_size; - u32 sec_addr; - u32 sec_size; - u32 data_addr; - u32 data_size; - - struct { - struct { - void *data; - u32 size; - } prod, dbg; - u32 patch_loc; - } sig; + struct list_head head; }; +int nvkm_acr_hsfw_boot(struct nvkm_acr *, const char *name); + struct nvkm_acr_hsf_fwif { int version; int (*load)(struct nvkm_acr *, const char *bl, const char *fw, const char *name, int version, const struct nvkm_acr_hsf_fwif *); - const struct nvkm_acr_hsf_func *func; -}; + const struct nvkm_falcon_fw_func *func; -int nvkm_acr_hsfw_load(struct nvkm_acr *, const char *, const char *, - const char *, int, const struct nvkm_acr_hsf_fwif *); -void nvkm_acr_hsfw_del_all(struct nvkm_acr *); - -struct nvkm_acr_hsf { - const struct nvkm_acr_hsf_func *func; - const char *name; - struct list_head head; - - u32 imem_size; - u32 imem_tag; - u32 *imem; - - u32 non_sec_addr; - u32 non_sec_size; - u32 sec_addr; - u32 sec_size; - u32 data_addr; - u32 data_size; - - struct nvkm_memory *ucode; - struct nvkm_vma *vma; - struct nvkm_falcon *falcon; + enum nvkm_acr_hsf_id falcon_id; + u32 boot_mbox0; + u32 intr_clear; }; -struct nvkm_acr_hsf_func { - int (*load)(struct nvkm_acr *, struct nvkm_acr_hsfw *); - int (*boot)(struct nvkm_acr *, struct nvkm_acr_hsf *); - void (*bld)(struct nvkm_acr *, struct nvkm_acr_hsf *); -}; -int gm200_acr_hsfw_load(struct nvkm_acr *, struct nvkm_acr_hsfw *, - struct nvkm_falcon *); -int gm200_acr_hsfw_boot(struct nvkm_acr *, struct nvkm_acr_hsf *, - u32 clear_intr, u32 mbox0_ok); +int gm200_acr_hsfw_ctor(struct nvkm_acr *, const char *, const char *, const char *, int, + const struct nvkm_acr_hsf_fwif *); +int gm200_acr_hsfw_load_bld(struct nvkm_falcon_fw *); +extern const struct nvkm_falcon_fw_func gm200_acr_unload_0; -int gm200_acr_load_boot(struct nvkm_acr *, struct nvkm_acr_hsf *); +extern const struct nvkm_falcon_fw_func gm20b_acr_load_0; -extern const struct nvkm_acr_hsf_func gm200_acr_unload_0; -int gm200_acr_unload_load(struct nvkm_acr *, struct nvkm_acr_hsfw *); -int gm200_acr_unload_boot(struct nvkm_acr *, struct nvkm_acr_hsf *); -void gm200_acr_hsfw_bld(struct nvkm_acr *, struct nvkm_acr_hsf *); +int gp102_acr_load_setup(struct nvkm_falcon_fw *); -extern const struct nvkm_acr_hsf_func gm20b_acr_load_0; +extern const struct nvkm_falcon_fw_func gp108_acr_load_0; -int gp102_acr_load_load(struct nvkm_acr *, struct nvkm_acr_hsfw *); +extern const struct nvkm_falcon_fw_func gp108_acr_hsfw_0; +int gp108_acr_hsfw_load_bld(struct nvkm_falcon_fw *); -extern const struct nvkm_acr_hsf_func gp108_acr_unload_0; -void gp108_acr_hsfw_bld(struct nvkm_acr *, struct nvkm_acr_hsf *); +int ga100_acr_hsfw_ctor(struct nvkm_acr *, const char *, const char *, const char *, int, + const struct nvkm_acr_hsf_fwif *); int nvkm_acr_new_(const struct nvkm_acr_fwif *, struct nvkm_device *, enum nvkm_subdev_type, int inst, struct nvkm_acr **); -int nvkm_acr_hsf_boot(struct nvkm_acr *, const char *name); struct nvkm_acr_lsf { const struct nvkm_acr_lsf_func *func; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/tu102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/tu102.c index 05a87e77525f..c22d551c0078 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/acr/tu102.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/acr/tu102.c @@ -29,14 +29,14 @@ #include <nvfw/acr.h> -static int +int tu102_acr_init(struct nvkm_acr *acr) { - int ret = nvkm_acr_hsf_boot(acr, "AHESASC"); + int ret = nvkm_acr_hsfw_boot(acr, "AHESASC"); if (ret) return ret; - return nvkm_acr_hsf_boot(acr, "ASB"); + return nvkm_acr_hsfw_boot(acr, "ASB"); } static int @@ -85,12 +85,6 @@ tu102_acr_wpr_build(struct nvkm_acr *acr, struct nvkm_acr_lsf *rtos) } static int -tu102_acr_hsfw_boot(struct nvkm_acr *acr, struct nvkm_acr_hsf *hsf) -{ - return gm200_acr_hsfw_boot(acr, hsf, 0, 0); -} - -static int tu102_acr_hsfw_nofw(struct nvkm_acr *acr, const char *bl, const char *fw, const char *name, int version, const struct nvkm_acr_hsf_fwif *fwif) @@ -115,24 +109,11 @@ MODULE_FIRMWARE("nvidia/tu117/acr/ucode_unload.bin"); static const struct nvkm_acr_hsf_fwif tu102_acr_unload_fwif[] = { - { 0, nvkm_acr_hsfw_load, &gp108_acr_unload_0 }, + { 0, gm200_acr_hsfw_ctor, &gp108_acr_hsfw_0, NVKM_ACR_HSF_PMU, 0, 0x00000000 }, { -1, tu102_acr_hsfw_nofw }, {} }; -static int -tu102_acr_asb_load(struct nvkm_acr *acr, struct nvkm_acr_hsfw *hsfw) -{ - return gm200_acr_hsfw_load(acr, hsfw, &acr->subdev.device->gsp->falcon); -} - -static const struct nvkm_acr_hsf_func -tu102_acr_asb_0 = { - .load = tu102_acr_asb_load, - .boot = tu102_acr_hsfw_boot, - .bld = gp108_acr_hsfw_bld, -}; - MODULE_FIRMWARE("nvidia/tu102/acr/ucode_asb.bin"); MODULE_FIRMWARE("nvidia/tu104/acr/ucode_asb.bin"); MODULE_FIRMWARE("nvidia/tu106/acr/ucode_asb.bin"); @@ -141,18 +122,11 @@ MODULE_FIRMWARE("nvidia/tu117/acr/ucode_asb.bin"); static const struct nvkm_acr_hsf_fwif tu102_acr_asb_fwif[] = { - { 0, nvkm_acr_hsfw_load, &tu102_acr_asb_0 }, + { 0, gm200_acr_hsfw_ctor, &gp108_acr_hsfw_0, NVKM_ACR_HSF_GSP, 0, 0x00000000 }, { -1, tu102_acr_hsfw_nofw }, {} }; -static const struct nvkm_acr_hsf_func -tu102_acr_ahesasc_0 = { - .load = gp102_acr_load_load, - .boot = tu102_acr_hsfw_boot, - .bld = gp108_acr_hsfw_bld, -}; - MODULE_FIRMWARE("nvidia/tu102/acr/bl.bin"); MODULE_FIRMWARE("nvidia/tu102/acr/ucode_ahesasc.bin"); @@ -170,7 +144,7 @@ MODULE_FIRMWARE("nvidia/tu117/acr/ucode_ahesasc.bin"); static const struct nvkm_acr_hsf_fwif tu102_acr_ahesasc_fwif[] = { - { 0, nvkm_acr_hsfw_load, &tu102_acr_ahesasc_0 }, + { 0, gm200_acr_hsfw_ctor, &gp108_acr_load_0, NVKM_ACR_HSF_SEC2, 0, 0x00000000 }, { -1, tu102_acr_hsfw_nofw }, {} }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm200.c b/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm200.c index a308b9bde449..f30718d7e61a 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm200.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/devinit/gm200.c @@ -26,6 +26,7 @@ #include <subdev/bios.h> #include <subdev/bios/bit.h> #include <subdev/bios/pmu.h> +#include <subdev/pmu.h> #include <subdev/timer.h> static void @@ -85,13 +86,18 @@ pmu_load(struct nv50_devinit *init, u8 type, bool post, struct nvkm_subdev *subdev = &init->base.subdev; struct nvkm_bios *bios = subdev->device->bios; struct nvbios_pmuR pmu; + int ret; if (!nvbios_pmuRm(bios, type, &pmu)) return -EINVAL; - if (!post) + if (!post || !subdev->device->pmu) return 0; + ret = nvkm_falcon_reset(&subdev->device->pmu->falcon); + if (ret) + return ret; + pmu_code(init, pmu.boot_addr_pmu, pmu.boot_addr, pmu.boot_size, false); pmu_code(init, pmu.code_addr_pmu, pmu.code_addr, pmu.code_size, true); pmu_data(init, pmu.data_addr_pmu, pmu.data_addr, pmu.data_size); diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/base.c index fd54fa504efa..b53ac9a2552f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/base.c @@ -22,7 +22,6 @@ #include "priv.h" #include <core/memory.h> -#include <core/notify.h> static void nvkm_fault_ntfy_fini(struct nvkm_event *event, int type, int index) @@ -38,23 +37,8 @@ nvkm_fault_ntfy_init(struct nvkm_event *event, int type, int index) fault->func->buffer.intr(fault->buffer[index], true); } -static int -nvkm_fault_ntfy_ctor(struct nvkm_object *object, void *argv, u32 argc, - struct nvkm_notify *notify) -{ - struct nvkm_fault_buffer *buffer = nvkm_fault_buffer(object); - if (argc == 0) { - notify->size = 0; - notify->types = 1; - notify->index = buffer->id; - return 0; - } - return -ENOSYS; -} - static const struct nvkm_event_func nvkm_fault_ntfy = { - .ctor = nvkm_fault_ntfy_ctor, .init = nvkm_fault_ntfy_init, .fini = nvkm_fault_ntfy_fini, }; @@ -130,8 +114,7 @@ nvkm_fault_oneinit(struct nvkm_subdev *subdev) } } - ret = nvkm_event_init(&nvkm_fault_ntfy, 1, fault->buffer_nr, - &fault->event); + ret = nvkm_event_init(&nvkm_fault_ntfy, subdev, 1, fault->buffer_nr, &fault->event); if (ret) return ret; @@ -146,7 +129,7 @@ nvkm_fault_dtor(struct nvkm_subdev *subdev) struct nvkm_fault *fault = nvkm_fault(subdev); int i; - nvkm_notify_fini(&fault->nrpfb); + nvkm_event_ntfy_del(&fault->nrpfb); nvkm_event_fini(&fault->event); for (i = 0; i < fault->buffer_nr; i++) { diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/gp100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/gp100.c index 6af7959e02ea..04c7526888bc 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/gp100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/gp100.c @@ -65,7 +65,7 @@ gp100_fault_buffer_info(struct nvkm_fault_buffer *buffer) void gp100_fault_intr(struct nvkm_fault *fault) { - nvkm_event_send(&fault->event, 1, 0, NULL, 0); + nvkm_event_ntfy(&fault->event, 0, NVKM_FAULT_BUFFER_EVENT_PENDING); } static const struct nvkm_fault_func diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/gv100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/gv100.c index cd9d2ade5ac7..8e34d40e7649 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/gv100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/gv100.c @@ -27,10 +27,12 @@ #include <nvif/class.h> -static void -gv100_fault_buffer_process(struct nvkm_fault_buffer *buffer) +void +gv100_fault_buffer_process(struct work_struct *work) { - struct nvkm_device *device = buffer->fault->subdev.device; + struct nvkm_fault *fault = container_of(work, typeof(*fault), nrpfb_work); + struct nvkm_fault_buffer *buffer = fault->buffer[0]; + struct nvkm_device *device = fault->subdev.device; struct nvkm_memory *mem = buffer->mem; u32 get = nvkm_rd32(device, buffer->get); u32 put = nvkm_rd32(device, buffer->put); @@ -115,11 +117,12 @@ gv100_fault_buffer_info(struct nvkm_fault_buffer *buffer) } static int -gv100_fault_ntfy_nrpfb(struct nvkm_notify *notify) +gv100_fault_ntfy_nrpfb(struct nvkm_event_ntfy *ntfy, u32 bits) { - struct nvkm_fault *fault = container_of(notify, typeof(*fault), nrpfb); - gv100_fault_buffer_process(fault->buffer[0]); - return NVKM_NOTIFY_KEEP; + struct nvkm_fault *fault = container_of(ntfy, typeof(*fault), nrpfb); + + schedule_work(&fault->nrpfb_work); + return NVKM_EVENT_KEEP; } static void @@ -163,14 +166,14 @@ gv100_fault_intr(struct nvkm_fault *fault) if (stat & 0x20000000) { if (fault->buffer[0]) { - nvkm_event_send(&fault->event, 1, 0, NULL, 0); + nvkm_event_ntfy(&fault->event, 0, NVKM_FAULT_BUFFER_EVENT_PENDING); stat &= ~0x20000000; } } if (stat & 0x08000000) { if (fault->buffer[1]) { - nvkm_event_send(&fault->event, 1, 1, NULL, 0); + nvkm_event_ntfy(&fault->event, 1, NVKM_FAULT_BUFFER_EVENT_PENDING); stat &= ~0x08000000; } } @@ -183,9 +186,12 @@ gv100_fault_intr(struct nvkm_fault *fault) static void gv100_fault_fini(struct nvkm_fault *fault) { - nvkm_notify_put(&fault->nrpfb); + nvkm_event_ntfy_block(&fault->nrpfb); + flush_work(&fault->nrpfb_work); + if (fault->buffer[0]) fault->func->buffer.fini(fault->buffer[0]); + nvkm_mask(fault->subdev.device, 0x100a34, 0x80000000, 0x80000000); } @@ -194,15 +200,15 @@ gv100_fault_init(struct nvkm_fault *fault) { nvkm_mask(fault->subdev.device, 0x100a2c, 0x80000000, 0x80000000); fault->func->buffer.init(fault->buffer[0]); - nvkm_notify_get(&fault->nrpfb); + nvkm_event_ntfy_allow(&fault->nrpfb); } int gv100_fault_oneinit(struct nvkm_fault *fault) { - return nvkm_notify_init(&fault->buffer[0]->object, &fault->event, - gv100_fault_ntfy_nrpfb, true, NULL, 0, 0, - &fault->nrpfb); + nvkm_event_ntfy_add(&fault->event, 0, NVKM_FAULT_BUFFER_EVENT_PENDING, true, + gv100_fault_ntfy_nrpfb, &fault->nrpfb); + return 0; } static const struct nvkm_fault_func @@ -231,5 +237,10 @@ int gv100_fault_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fault **pfault) { - return nvkm_fault_new_(&gv100_fault, device, type, inst, pfault); + int ret = nvkm_fault_new_(&gv100_fault, device, type, inst, pfault); + if (ret) + return ret; + + INIT_WORK(&(*pfault)->nrpfb_work, gv100_fault_buffer_process); + return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/priv.h b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/priv.h index 36681c347fb5..a5510332c402 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/priv.h @@ -16,6 +16,8 @@ struct nvkm_fault_buffer { u32 put; struct nvkm_memory *mem; u64 addr; + + struct nvkm_inth inth; }; int nvkm_fault_new_(const struct nvkm_fault_func *, struct nvkm_device *, enum nvkm_subdev_type, @@ -46,6 +48,7 @@ void gp100_fault_buffer_fini(struct nvkm_fault_buffer *); void gp100_fault_buffer_init(struct nvkm_fault_buffer *); u64 gp100_fault_buffer_pin(struct nvkm_fault_buffer *); void gp100_fault_buffer_info(struct nvkm_fault_buffer *); +void gv100_fault_buffer_process(struct work_struct *); void gp100_fault_intr(struct nvkm_fault *); u64 gp10b_fault_buffer_pin(struct nvkm_fault_buffer *); diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/tu102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/tu102.c index 91eb6729c84d..967efaddae28 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/tu102.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/tu102.c @@ -24,20 +24,27 @@ #include <core/memory.h> #include <subdev/mc.h> #include <subdev/mmu.h> +#include <subdev/vfn.h> #include <engine/fifo.h> #include <nvif/class.h> +static irqreturn_t +tu102_fault_buffer_notify(struct nvkm_inth *inth) +{ + struct nvkm_fault_buffer *buffer = container_of(inth, typeof(*buffer), inth); + + nvkm_event_ntfy(&buffer->fault->event, buffer->id, NVKM_FAULT_BUFFER_EVENT_PENDING); + return IRQ_HANDLED; +} + static void tu102_fault_buffer_intr(struct nvkm_fault_buffer *buffer, bool enable) { - /*XXX: Earlier versions of RM touched the old regs on Turing, - * which don't appear to actually work anymore, but newer - * versions of RM don't appear to touch anything at all.. - */ - struct nvkm_device *device = buffer->fault->subdev.device; - - nvkm_mc_intr_mask(device, NVKM_SUBDEV_FAULT, 0, enable); + if (enable) + nvkm_inth_allow(&buffer->inth); + else + nvkm_inth_block(&buffer->inth); } static void @@ -46,10 +53,6 @@ tu102_fault_buffer_fini(struct nvkm_fault_buffer *buffer) struct nvkm_device *device = buffer->fault->subdev.device; const u32 foff = buffer->id * 0x20; - /* Disable the fault interrupts */ - nvkm_wr32(device, 0xb81408, 0x1); - nvkm_wr32(device, 0xb81410, 0x10); - nvkm_mask(device, 0xb83010 + foff, 0x80000000, 0x00000000); } @@ -59,10 +62,6 @@ tu102_fault_buffer_init(struct nvkm_fault_buffer *buffer) struct nvkm_device *device = buffer->fault->subdev.device; const u32 foff = buffer->id * 0x20; - /* Enable the fault interrupts */ - nvkm_wr32(device, 0xb81208, 0x1); - nvkm_wr32(device, 0xb81210, 0x10); - nvkm_mask(device, 0xb83010 + foff, 0xc0000000, 0x40000000); nvkm_wr32(device, 0xb83004 + foff, upper_32_bits(buffer->addr)); nvkm_wr32(device, 0xb83000 + foff, lower_32_bits(buffer->addr)); @@ -82,9 +81,10 @@ tu102_fault_buffer_info(struct nvkm_fault_buffer *buffer) buffer->put = 0xb8300c + foff; } -static void -tu102_fault_intr_fault(struct nvkm_fault *fault) +static irqreturn_t +tu102_fault_info_fault(struct nvkm_inth *inth) { + struct nvkm_fault *fault = container_of(inth, typeof(*fault), info_fault); struct nvkm_subdev *subdev = &fault->subdev; struct nvkm_device *device = subdev->device; struct nvkm_fault_data info; @@ -106,70 +106,61 @@ tu102_fault_intr_fault(struct nvkm_fault *fault) info.reason = (info1 & 0x0000001f); nvkm_fifo_fault(device->fifo, &info); -} - -static void -tu102_fault_intr(struct nvkm_fault *fault) -{ - struct nvkm_subdev *subdev = &fault->subdev; - struct nvkm_device *device = subdev->device; - u32 stat = nvkm_rd32(device, 0xb83094); - - if (stat & 0x80000000) { - tu102_fault_intr_fault(fault); - nvkm_wr32(device, 0xb83094, 0x80000000); - stat &= ~0x80000000; - } - if (stat & 0x00000200) { - /* Clear the associated interrupt flag */ - nvkm_wr32(device, 0xb81010, 0x10); - - if (fault->buffer[0]) { - nvkm_event_send(&fault->event, 1, 0, NULL, 0); - stat &= ~0x00000200; - } - } - - /* Replayable MMU fault */ - if (stat & 0x00000100) { - /* Clear the associated interrupt flag */ - nvkm_wr32(device, 0xb81008, 0x1); - - if (fault->buffer[1]) { - nvkm_event_send(&fault->event, 1, 1, NULL, 0); - stat &= ~0x00000100; - } - } - - if (stat) { - nvkm_debug(subdev, "intr %08x\n", stat); - } + nvkm_wr32(device, 0xb83094, 0x80000000); + return IRQ_HANDLED; } static void tu102_fault_fini(struct nvkm_fault *fault) { - nvkm_notify_put(&fault->nrpfb); + nvkm_event_ntfy_block(&fault->nrpfb); + flush_work(&fault->nrpfb_work); + if (fault->buffer[0]) fault->func->buffer.fini(fault->buffer[0]); - /*XXX: disable priv faults */ + + nvkm_inth_block(&fault->info_fault); } static void tu102_fault_init(struct nvkm_fault *fault) { - /*XXX: enable priv faults */ + nvkm_inth_allow(&fault->info_fault); + fault->func->buffer.init(fault->buffer[0]); - nvkm_notify_get(&fault->nrpfb); + nvkm_event_ntfy_allow(&fault->nrpfb); +} + +static int +tu102_fault_oneinit(struct nvkm_fault *fault) +{ + struct nvkm_device *device = fault->subdev.device; + struct nvkm_intr *intr = &device->vfn->intr; + int ret, i; + + ret = nvkm_inth_add(intr, nvkm_rd32(device, 0x100ee0) & 0x0000ffff, + NVKM_INTR_PRIO_NORMAL, &fault->subdev, tu102_fault_info_fault, + &fault->info_fault); + if (ret) + return ret; + + for (i = 0; i < fault->buffer_nr; i++) { + ret = nvkm_inth_add(intr, nvkm_rd32(device, 0x100ee4 + (i * 4)) >> 16, + NVKM_INTR_PRIO_NORMAL, &fault->subdev, + tu102_fault_buffer_notify, &fault->buffer[i]->inth); + if (ret) + return ret; + } + + return gv100_fault_oneinit(fault); } static const struct nvkm_fault_func tu102_fault = { - .oneinit = gv100_fault_oneinit, + .oneinit = tu102_fault_oneinit, .init = tu102_fault_init, .fini = tu102_fault_fini, - .intr = tu102_fault_intr, .buffer.nr = 2, .buffer.entry_size = 32, .buffer.info = tu102_fault_buffer_info, @@ -184,5 +175,10 @@ int tu102_fault_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fault **pfault) { - return nvkm_fault_new_(&tu102_fault, device, type, inst, pfault); + int ret = nvkm_fault_new_(&tu102_fault, device, type, inst, pfault); + if (ret) + return ret; + + INIT_WORK(&(*pfault)->nrpfb_work, gv100_fault_buffer_process); + return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/user.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/user.c index ac835c9582fd..c123e5893d76 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fault/user.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fault/user.c @@ -22,12 +22,28 @@ #include "priv.h" #include <core/memory.h> +#include <core/event.h> #include <subdev/mmu.h> #include <nvif/clb069.h> #include <nvif/unpack.h> static int +nvkm_ufault_uevent(struct nvkm_object *object, void *argv, u32 argc, struct nvkm_uevent *uevent) +{ + struct nvkm_fault_buffer *buffer = nvkm_fault_buffer(object); + union nvif_clb069_event_args *args = argv; + + if (!uevent) + return 0; + if (argc != sizeof(args->vn)) + return -ENOSYS; + + return nvkm_uevent_add(uevent, &buffer->fault->event, buffer->id, + NVKM_FAULT_BUFFER_EVENT_PENDING, NULL); +} + +static int nvkm_ufault_map(struct nvkm_object *object, void *argv, u32 argc, enum nvkm_object_map *type, u64 *addr, u64 *size) { @@ -40,18 +56,6 @@ nvkm_ufault_map(struct nvkm_object *object, void *argv, u32 argc, } static int -nvkm_ufault_ntfy(struct nvkm_object *object, u32 type, - struct nvkm_event **pevent) -{ - struct nvkm_fault_buffer *buffer = nvkm_fault_buffer(object); - if (type == NVB069_V0_NTFY_FAULT) { - *pevent = &buffer->fault->event; - return 0; - } - return -EINVAL; -} - -static int nvkm_ufault_fini(struct nvkm_object *object, bool suspend) { struct nvkm_fault_buffer *buffer = nvkm_fault_buffer(object); @@ -78,8 +82,8 @@ nvkm_ufault = { .dtor = nvkm_ufault_dtor, .init = nvkm_ufault_init, .fini = nvkm_ufault_fini, - .ntfy = nvkm_ufault_ntfy, .map = nvkm_ufault_map, + .uevent = nvkm_ufault_uevent, }; int diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/base.c index 6faaea948fc4..bac7dcc4c2c1 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/base.c @@ -57,6 +57,15 @@ nvkm_fb_tile_prog(struct nvkm_fb *fb, int region, struct nvkm_fb_tile *tile) } } +static void +nvkm_fb_sysmem_flush_page_init(struct nvkm_device *device) +{ + struct nvkm_fb *fb = device->fb; + + if (fb->func->sysmem.flush_page_init) + fb->func->sysmem.flush_page_init(fb); +} + int nvkm_fb_bios_memtype(struct nvkm_bios *bios) { @@ -125,12 +134,20 @@ nvkm_fb_oneinit(struct nvkm_subdev *subdev) return nvkm_mm_init(&fb->tags.mm, 0, 0, tags, 1); } -static int -nvkm_fb_init_scrub_vpr(struct nvkm_fb *fb) +int +nvkm_fb_mem_unlock(struct nvkm_fb *fb) { struct nvkm_subdev *subdev = &fb->subdev; int ret; + if (!fb->func->vpr.scrub_required) + return 0; + + if (!fb->func->vpr.scrub_required(fb)) { + nvkm_debug(subdev, "VPR not locked\n"); + return 0; + } + nvkm_debug(subdev, "VPR locked, running scrubber binary\n"); if (!fb->vpr_scrubber.size) { @@ -168,6 +185,8 @@ nvkm_fb_init(struct nvkm_subdev *subdev) for (i = 0; i < fb->tile.regions; i++) fb->func->tile.prog(fb, i, &fb->tile.region[i]); + nvkm_fb_sysmem_flush_page_init(subdev->device); + if (fb->func->init) fb->func->init(fb); @@ -183,13 +202,13 @@ nvkm_fb_init(struct nvkm_subdev *subdev) if (fb->func->init_unkn) fb->func->init_unkn(fb); - if (fb->func->vpr.scrub_required && - fb->func->vpr.scrub_required(fb)) { - ret = nvkm_fb_init_scrub_vpr(fb); - if (ret) - return ret; - } + return 0; +} +static int +nvkm_fb_preinit(struct nvkm_subdev *subdev) +{ + nvkm_fb_sysmem_flush_page_init(subdev->device); return 0; } @@ -212,20 +231,28 @@ nvkm_fb_dtor(struct nvkm_subdev *subdev) nvkm_blob_dtor(&fb->vpr_scrubber); + if (fb->sysmem.flush_page) { + dma_unmap_page(subdev->device->dev, fb->sysmem.flush_page_addr, + PAGE_SIZE, DMA_BIDIRECTIONAL); + __free_page(fb->sysmem.flush_page); + } + if (fb->func->dtor) return fb->func->dtor(fb); + return fb; } static const struct nvkm_subdev_func nvkm_fb = { .dtor = nvkm_fb_dtor, + .preinit = nvkm_fb_preinit, .oneinit = nvkm_fb_oneinit, .init = nvkm_fb_init, .intr = nvkm_fb_intr, }; -void +int nvkm_fb_ctor(const struct nvkm_fb_func *func, struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fb *fb) { @@ -234,6 +261,19 @@ nvkm_fb_ctor(const struct nvkm_fb_func *func, struct nvkm_device *device, fb->tile.regions = fb->func->tile.regions; fb->page = nvkm_longopt(device->cfgopt, "NvFbBigPage", fb->func->default_bigpage); mutex_init(&fb->tags.mutex); + + if (func->sysmem.flush_page_init) { + fb->sysmem.flush_page = alloc_page(GFP_KERNEL | __GFP_ZERO); + if (!fb->sysmem.flush_page) + return -ENOMEM; + + fb->sysmem.flush_page_addr = dma_map_page(device->dev, fb->sysmem.flush_page, + 0, PAGE_SIZE, DMA_BIDIRECTIONAL); + if (dma_mapping_error(device->dev, fb->sysmem.flush_page_addr)) + return -EFAULT; + } + + return 0; } int @@ -242,6 +282,5 @@ nvkm_fb_new_(const struct nvkm_fb_func *func, struct nvkm_device *device, { if (!(*pfb = kzalloc(sizeof(**pfb), GFP_KERNEL))) return -ENOMEM; - nvkm_fb_ctor(func, device, type, inst, *pfb); - return 0; + return nvkm_fb_ctor(func, device, type, inst, *pfb); } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ga100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ga100.c index b47bebfbc26f..5098f219e3e6 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ga100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ga100.c @@ -26,9 +26,10 @@ static const struct nvkm_fb_func ga100_fb = { .dtor = gf100_fb_dtor, .oneinit = gf100_fb_oneinit, - .init = gp100_fb_init, + .init = gm200_fb_init, .init_page = gv100_fb_init_page, .init_unkn = gp100_fb_init_unkn, + .sysmem.flush_page_init = gf100_fb_sysmem_flush_page_init, .ram_new = gp100_ram_new, .default_bigpage = 16, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ga102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ga102.c index 6ea7908f0563..8b7c8ea5e8a5 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ga102.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ga102.c @@ -22,15 +22,42 @@ #include "gf100.h" #include "ram.h" +#include <engine/nvdec.h> + +static int +ga102_fb_vpr_scrub(struct nvkm_fb *fb) +{ + struct nvkm_falcon_fw fw = {}; + int ret; + + ret = nvkm_falcon_fw_ctor_hs_v2(&ga102_flcn_fw, "mem-unlock", &fb->subdev, "nvdec/scrubber", + 0, &fb->subdev.device->nvdec[0]->falcon, &fw); + if (ret) + return ret; + + ret = nvkm_falcon_fw_boot(&fw, &fb->subdev, true, NULL, NULL, 0, 0); + nvkm_falcon_fw_dtor(&fw); + return ret; +} + +static bool +ga102_fb_vpr_scrub_required(struct nvkm_fb *fb) +{ + return (nvkm_rd32(fb->subdev.device, 0x1fa80c) & 0x00000010) != 0; +} + static const struct nvkm_fb_func ga102_fb = { .dtor = gf100_fb_dtor, .oneinit = gf100_fb_oneinit, - .init = gp100_fb_init, + .init = gm200_fb_init, .init_page = gv100_fb_init_page, .init_unkn = gp100_fb_init_unkn, + .sysmem.flush_page_init = gf100_fb_sysmem_flush_page_init, .ram_new = ga102_ram_new, .default_bigpage = 16, + .vpr.scrub_required = ga102_fb_vpr_scrub_required, + .vpr.scrub = ga102_fb_vpr_scrub, }; int @@ -38,3 +65,9 @@ ga102_fb_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, s { return gp102_fb_new_(&ga102_fb, device, type, inst, pfb); } + +MODULE_FIRMWARE("nvidia/ga102/nvdec/scrubber.bin"); +MODULE_FIRMWARE("nvidia/ga103/nvdec/scrubber.bin"); +MODULE_FIRMWARE("nvidia/ga104/nvdec/scrubber.bin"); +MODULE_FIRMWARE("nvidia/ga106/nvdec/scrubber.bin"); +MODULE_FIRMWARE("nvidia/ga107/nvdec/scrubber.bin"); diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.c index 9dcc40f9ef79..07db9b397ac1 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.c @@ -61,14 +61,6 @@ gf100_fb_oneinit(struct nvkm_fb *base) if (ret) return ret; - fb->r100c10_page = alloc_page(GFP_KERNEL | __GFP_ZERO); - if (fb->r100c10_page) { - fb->r100c10 = dma_map_page(device->dev, fb->r100c10_page, 0, - PAGE_SIZE, DMA_BIDIRECTIONAL); - if (dma_mapping_error(device->dev, fb->r100c10)) - return -EFAULT; - } - return 0; } @@ -86,14 +78,17 @@ gf100_fb_init_page(struct nvkm_fb *fb) } void +gf100_fb_sysmem_flush_page_init(struct nvkm_fb *fb) +{ + nvkm_wr32(fb->subdev.device, 0x100c10, fb->sysmem.flush_page_addr >> 8); +} + +void gf100_fb_init(struct nvkm_fb *base) { struct gf100_fb *fb = gf100_fb(base); struct nvkm_device *device = fb->base.subdev.device; - if (fb->r100c10_page) - nvkm_wr32(device, 0x100c10, fb->r100c10 >> 8); - if (base->func->clkgate_pack) { nvkm_therm_clkgate_init(device->therm, base->func->clkgate_pack); @@ -104,13 +99,6 @@ void * gf100_fb_dtor(struct nvkm_fb *base) { struct gf100_fb *fb = gf100_fb(base); - struct nvkm_device *device = fb->base.subdev.device; - - if (fb->r100c10_page) { - dma_unmap_page(device->dev, fb->r100c10, PAGE_SIZE, - DMA_BIDIRECTIONAL); - __free_page(fb->r100c10_page); - } return fb; } @@ -136,6 +124,7 @@ gf100_fb = { .init = gf100_fb_init, .init_page = gf100_fb_init_page, .intr = gf100_fb_intr, + .sysmem.flush_page_init = gf100_fb_sysmem_flush_page_init, .ram_new = gf100_ram_new, .default_bigpage = 17, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.h b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.h index 0cac7b06acc8..77472b558591 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gf100.h @@ -6,8 +6,6 @@ struct gf100_fb { struct nvkm_fb base; - struct page *r100c10_page; - dma_addr_t r100c10; }; int gf100_fb_new_(const struct nvkm_fb_func *, struct nvkm_device *, enum nvkm_subdev_type, int, @@ -16,7 +14,5 @@ void *gf100_fb_dtor(struct nvkm_fb *); void gf100_fb_init(struct nvkm_fb *); void gf100_fb_intr(struct nvkm_fb *); -void gp100_fb_init(struct nvkm_fb *); - void gm200_fb_init(struct nvkm_fb *base); #endif diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gm200.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gm200.c index 5acf8d15d06f..fb02092a65eb 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gm200.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gm200.c @@ -46,9 +46,6 @@ gm200_fb_init(struct nvkm_fb *base) struct gf100_fb *fb = gf100_fb(base); struct nvkm_device *device = fb->base.subdev.device; - if (fb->r100c10_page) - nvkm_wr32(device, 0x100c10, fb->r100c10 >> 8); - nvkm_wr32(device, 0x100cc8, nvkm_memory_addr(fb->base.mmu_wr) >> 8); nvkm_wr32(device, 0x100ccc, nvkm_memory_addr(fb->base.mmu_rd) >> 8); nvkm_mask(device, 0x100cc4, 0x00060000, @@ -62,6 +59,7 @@ gm200_fb = { .init = gm200_fb_init, .init_page = gm200_fb_init_page, .intr = gf100_fb_intr, + .sysmem.flush_page_init = gf100_fb_sysmem_flush_page_init, .ram_new = gm200_ram_new, .default_bigpage = 0 /* per-instance. */, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gm20b.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gm20b.c index 86f61a3f2fea..50875af94c18 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gm20b.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gm20b.c @@ -30,6 +30,7 @@ gm20b_fb = { .init = gm200_fb_init, .init_page = gm200_fb_init_page, .intr = gf100_fb_intr, + .sysmem.flush_page_init = gf100_fb_sysmem_flush_page_init, .default_bigpage = 0 /* per-instance. */, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp100.c index 09e943edc362..110c08c94849 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp100.c @@ -44,29 +44,15 @@ gp100_fb_init_remapper(struct nvkm_fb *fb) nvkm_mask(device, 0x100c14, 0x00040000, 0x00000000); } -void -gp100_fb_init(struct nvkm_fb *base) -{ - struct gf100_fb *fb = gf100_fb(base); - struct nvkm_device *device = fb->base.subdev.device; - - if (fb->r100c10_page) - nvkm_wr32(device, 0x100c10, fb->r100c10 >> 8); - - nvkm_wr32(device, 0x100cc8, nvkm_memory_addr(fb->base.mmu_wr) >> 8); - nvkm_wr32(device, 0x100ccc, nvkm_memory_addr(fb->base.mmu_rd) >> 8); - nvkm_mask(device, 0x100cc4, 0x00060000, - min(nvkm_memory_size(fb->base.mmu_rd) >> 16, (u64)2) << 17); -} - static const struct nvkm_fb_func gp100_fb = { .dtor = gf100_fb_dtor, .oneinit = gf100_fb_oneinit, - .init = gp100_fb_init, + .init = gm200_fb_init, .init_remapper = gp100_fb_init_remapper, .init_page = gm200_fb_init_page, .init_unkn = gp100_fb_init_unkn, + .sysmem.flush_page_init = gf100_fb_sysmem_flush_page_init, .ram_new = gp100_ram_new, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp102.c index 0e78b3d734a0..2658481d575b 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp102.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp102.c @@ -24,71 +24,22 @@ #include "gf100.h" #include "ram.h" -#include <core/firmware.h> -#include <core/memory.h> -#include <nvfw/fw.h> -#include <nvfw/hs.h> #include <engine/nvdec.h> int gp102_fb_vpr_scrub(struct nvkm_fb *fb) { struct nvkm_subdev *subdev = &fb->subdev; - struct nvkm_device *device = subdev->device; - struct nvkm_falcon *falcon = &device->nvdec[0]->falcon; - struct nvkm_blob *blob = &fb->vpr_scrubber; - const struct nvfw_bin_hdr *hsbin_hdr; - const struct nvfw_hs_header *fw_hdr; - const struct nvfw_hs_load_header *lhdr; - void *scrub_data; - u32 patch_loc, patch_sig; + struct nvkm_falcon_fw fw = {}; int ret; - nvkm_falcon_get(falcon, subdev); - - hsbin_hdr = nvfw_bin_hdr(subdev, blob->data); - fw_hdr = nvfw_hs_header(subdev, blob->data + hsbin_hdr->header_offset); - lhdr = nvfw_hs_load_header(subdev, blob->data + fw_hdr->hdr_offset); - scrub_data = blob->data + hsbin_hdr->data_offset; - - patch_loc = *(u32 *)(blob->data + fw_hdr->patch_loc); - patch_sig = *(u32 *)(blob->data + fw_hdr->patch_sig); - if (falcon->debug) { - memcpy(scrub_data + patch_loc, - blob->data + fw_hdr->sig_dbg_offset + patch_sig, - fw_hdr->sig_dbg_size); - } else { - memcpy(scrub_data + patch_loc, - blob->data + fw_hdr->sig_prod_offset + patch_sig, - fw_hdr->sig_prod_size); - } - - nvkm_falcon_reset(falcon); - nvkm_falcon_bind_context(falcon, NULL); - - nvkm_falcon_load_imem(falcon, scrub_data, lhdr->non_sec_code_off, - lhdr->non_sec_code_size, - lhdr->non_sec_code_off >> 8, 0, false); - nvkm_falcon_load_imem(falcon, scrub_data + lhdr->apps[0], - ALIGN(lhdr->apps[0], 0x100), - lhdr->apps[1], - lhdr->apps[0] >> 8, 0, true); - nvkm_falcon_load_dmem(falcon, scrub_data + lhdr->data_dma_base, 0, - lhdr->data_size, 0); - - nvkm_falcon_set_start_addr(falcon, 0x0); - nvkm_falcon_start(falcon); - - ret = nvkm_falcon_wait_for_halt(falcon, 500); - if (ret < 0) { - ret = -ETIMEDOUT; - goto end; - } + ret = nvkm_falcon_fw_ctor_hs(&gm200_flcn_fw, "mem-unlock", subdev, NULL, + "nvdec/scrubber", 0, &subdev->device->nvdec[0]->falcon, &fw); + if (ret) + return ret; - /* put nvdec in clean state - without reset it will remain in HS mode */ - nvkm_falcon_reset(falcon); -end: - nvkm_falcon_put(falcon, subdev); + ret = nvkm_falcon_fw_boot(&fw, subdev, true, NULL, NULL, 0, 0x00000000); + nvkm_falcon_fw_dtor(&fw); return ret; } @@ -104,9 +55,10 @@ static const struct nvkm_fb_func gp102_fb = { .dtor = gf100_fb_dtor, .oneinit = gf100_fb_oneinit, - .init = gp100_fb_init, + .init = gm200_fb_init, .init_remapper = gp100_fb_init_remapper, .init_page = gm200_fb_init_page, + .sysmem.flush_page_init = gf100_fb_sysmem_flush_page_init, .vpr.scrub_required = gp102_fb_vpr_scrub_required, .vpr.scrub = gp102_fb_vpr_scrub, .ram_new = gp100_ram_new, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp10b.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp10b.c index 84c9815a6d48..a04a5f712019 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp10b.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gp10b.c @@ -28,6 +28,7 @@ gp10b_fb = { .init = gm200_fb_init, .init_page = gm200_fb_init_page, .intr = gf100_fb_intr, + .sysmem.flush_page_init = gf100_fb_sysmem_flush_page_init, }; int diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gv100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gv100.c index 63daa83ae12d..1f0126437c1a 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gv100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/gv100.c @@ -32,9 +32,10 @@ static const struct nvkm_fb_func gv100_fb = { .dtor = gf100_fb_dtor, .oneinit = gf100_fb_oneinit, - .init = gp100_fb_init, + .init = gm200_fb_init, .init_page = gv100_fb_init_page, .init_unkn = gp100_fb_init_unkn, + .sysmem.flush_page_init = gf100_fb_sysmem_flush_page_init, .vpr.scrub_required = gp102_fb_vpr_scrub_required, .vpr.scrub = gp102_fb_vpr_scrub, .ram_new = gp100_ram_new, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/nv50.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/nv50.c index 95fd8f834010..a6efbd913c13 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/nv50.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/nv50.c @@ -137,8 +137,7 @@ nv50_fb_intr(struct nvkm_fb *base) struct nv50_fb *fb = nv50_fb(base); struct nvkm_subdev *subdev = &fb->base.subdev; struct nvkm_device *device = subdev->device; - struct nvkm_fifo *fifo = device->fifo; - struct nvkm_fifo_chan *chan; + struct nvkm_chan *chan; const struct nvkm_enum *en, *re, *cl, *sc; u32 trap[6], idx, inst; u8 st0, st1, st2, st3; @@ -178,35 +177,18 @@ nv50_fb_intr(struct nvkm_fb *base) else if (en && en->data) sc = nvkm_enum_find(en->data, st3); else sc = NULL; - chan = nvkm_fifo_chan_inst(fifo, inst, &flags); + chan = nvkm_chan_get_inst(&device->fifo->engine, inst, &flags); nvkm_error(subdev, "trapped %s at %02x%04x%04x on channel %d [%08x %s] " "engine %02x [%s] client %02x [%s] " "subclient %02x [%s] reason %08x [%s]\n", (trap[5] & 0x00000100) ? "read" : "write", trap[5] & 0xff, trap[4] & 0xffff, trap[3] & 0xffff, - chan ? chan->chid : -1, inst, - chan ? chan->object.client->name : "unknown", + chan ? chan->id : -1, inst, + chan ? chan->name : "unknown", st0, en ? en->name : "", st2, cl ? cl->name : "", st3, sc ? sc->name : "", st1, re ? re->name : ""); - nvkm_fifo_chan_put(fifo, flags, &chan); -} - -static int -nv50_fb_oneinit(struct nvkm_fb *base) -{ - struct nv50_fb *fb = nv50_fb(base); - struct nvkm_device *device = fb->base.subdev.device; - - fb->r100c08_page = alloc_page(GFP_KERNEL | __GFP_ZERO); - if (fb->r100c08_page) { - fb->r100c08 = dma_map_page(device->dev, fb->r100c08_page, 0, - PAGE_SIZE, DMA_BIDIRECTIONAL); - if (dma_mapping_error(device->dev, fb->r100c08)) - return -EFAULT; - } - - return 0; + nvkm_chan_put(&chan, flags); } static void @@ -215,12 +197,6 @@ nv50_fb_init(struct nvkm_fb *base) struct nv50_fb *fb = nv50_fb(base); struct nvkm_device *device = fb->base.subdev.device; - /* Not a clue what this is exactly. Without pointing it at a - * scratch page, VRAM->GART blits with M2MF (as in DDX DFS) - * cause IOMMU "read from address 0" errors (rh#561267) - */ - nvkm_wr32(device, 0x100c08, fb->r100c08 >> 8); - /* This is needed to get meaningful information from 100c90 * on traps. No idea what these values mean exactly. */ nvkm_wr32(device, 0x100c90, fb->func->trap); @@ -235,17 +211,16 @@ nv50_fb_tags(struct nvkm_fb *base) return 0; } +static void +nv50_fb_sysmem_flush_page_init(struct nvkm_fb *fb) +{ + nvkm_wr32(fb->subdev.device, 0x100c08, fb->sysmem.flush_page_addr >> 8); +} + static void * nv50_fb_dtor(struct nvkm_fb *base) { struct nv50_fb *fb = nv50_fb(base); - struct nvkm_device *device = fb->base.subdev.device; - - if (fb->r100c08_page) { - dma_unmap_page(device->dev, fb->r100c08, PAGE_SIZE, - DMA_BIDIRECTIONAL); - __free_page(fb->r100c08_page); - } return fb; } @@ -254,9 +229,9 @@ static const struct nvkm_fb_func nv50_fb_ = { .dtor = nv50_fb_dtor, .tags = nv50_fb_tags, - .oneinit = nv50_fb_oneinit, .init = nv50_fb_init, .intr = nv50_fb_intr, + .sysmem.flush_page_init = nv50_fb_sysmem_flush_page_init, .ram_new = nv50_fb_ram_new, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/nv50.h b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/nv50.h index a5e673859a90..4f68bc4513a7 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/nv50.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/nv50.h @@ -7,8 +7,6 @@ struct nv50_fb { const struct nv50_fb_func *func; struct nvkm_fb base; - struct page *r100c08_page; - dma_addr_t r100c08; }; struct nv50_fb_func { diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/priv.h b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/priv.h index 3f1be9780c65..ac03eac0f261 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/priv.h @@ -16,6 +16,10 @@ struct nvkm_fb_func { void (*init_unkn)(struct nvkm_fb *); void (*intr)(struct nvkm_fb *); + struct nvkm_fb_func_sysmem { + void (*flush_page_init)(struct nvkm_fb *); + } sysmem; + struct { bool (*scrub_required)(struct nvkm_fb *); int (*scrub)(struct nvkm_fb *); @@ -37,8 +41,8 @@ struct nvkm_fb_func { const struct nvkm_therm_clkgate_pack *clkgate_pack; }; -void nvkm_fb_ctor(const struct nvkm_fb_func *, struct nvkm_device *device, - enum nvkm_subdev_type type, int inst, struct nvkm_fb *); +int nvkm_fb_ctor(const struct nvkm_fb_func *, struct nvkm_device *device, + enum nvkm_subdev_type type, int inst, struct nvkm_fb *); int nvkm_fb_new_(const struct nvkm_fb_func *, struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_fb **); int nvkm_fb_bios_memtype(struct nvkm_bios *); @@ -72,6 +76,7 @@ void nv46_fb_tile_init(struct nvkm_fb *, int i, u32 addr, u32 size, int gf100_fb_oneinit(struct nvkm_fb *); int gf100_fb_init_page(struct nvkm_fb *); +void gf100_fb_sysmem_flush_page_init(struct nvkm_fb *); int gm200_fb_init_page(struct nvkm_fb *); diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ram.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ram.c index 03b1bdb27770..5c34416cb637 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ram.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ram.c @@ -25,6 +25,7 @@ #include "ram.h" #include <core/memory.h> +#include <subdev/instmem.h> #include <subdev/mmu.h> struct nvkm_vram { @@ -35,6 +36,12 @@ struct nvkm_vram { }; static int +nvkm_vram_kmap(struct nvkm_memory *memory, struct nvkm_memory **pmemory) +{ + return nvkm_instobj_wrap(nvkm_vram(memory)->ram->fb->subdev.device, memory, pmemory); +} + +static int nvkm_vram_map(struct nvkm_memory *memory, u64 offset, struct nvkm_vmm *vmm, struct nvkm_vma *vma, void *argv, u32 argc) { @@ -98,6 +105,7 @@ nvkm_vram = { .addr = nvkm_vram_addr, .size = nvkm_vram_size, .map = nvkm_vram_map, + .kmap = nvkm_vram_kmap, }; int diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gpio/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gpio/base.c index 048bcc70c3f4..b196baa376dc 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gpio/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gpio/base.c @@ -24,7 +24,6 @@ #include "priv.h" #include <core/option.h> -#include <core/notify.h> static int nvkm_gpio_drive(struct nvkm_gpio *gpio, int idx, int line, int dir, int out) @@ -123,23 +122,8 @@ nvkm_gpio_intr_init(struct nvkm_event *event, int type, int index) gpio->func->intr_mask(gpio, type, 1 << index, 1 << index); } -static int -nvkm_gpio_intr_ctor(struct nvkm_object *object, void *data, u32 size, - struct nvkm_notify *notify) -{ - struct nvkm_gpio_ntfy_req *req = data; - if (!WARN_ON(size != sizeof(*req))) { - notify->size = sizeof(struct nvkm_gpio_ntfy_rep); - notify->types = req->mask; - notify->index = req->line; - return 0; - } - return -EINVAL; -} - static const struct nvkm_event_func nvkm_gpio_intr_func = { - .ctor = nvkm_gpio_intr_ctor, .init = nvkm_gpio_intr_init, .fini = nvkm_gpio_intr_fini, }; @@ -153,11 +137,9 @@ nvkm_gpio_intr(struct nvkm_subdev *subdev) gpio->func->intr_stat(gpio, &hi, &lo); for (i = 0; (hi | lo) && i < gpio->func->lines; i++) { - struct nvkm_gpio_ntfy_rep rep = { - .mask = (NVKM_GPIO_HI * !!(hi & (1 << i))) | - (NVKM_GPIO_LO * !!(lo & (1 << i))), - }; - nvkm_event_send(&gpio->event, rep.mask, i, &rep, sizeof(rep)); + u32 mask = (NVKM_GPIO_HI * !!(hi & (1 << i))) | + (NVKM_GPIO_LO * !!(lo & (1 << i))); + nvkm_event_ntfy(&gpio->event, i, mask); } } @@ -251,6 +233,5 @@ nvkm_gpio_new_(const struct nvkm_gpio_func *func, struct nvkm_device *device, nvkm_subdev_ctor(&nvkm_gpio, device, type, inst, &gpio->subdev); gpio->func = func; - return nvkm_event_init(&nvkm_gpio_intr_func, 2, func->lines, - &gpio->event); + return nvkm_event_init(&nvkm_gpio_intr_func, &gpio->subdev, 2, func->lines, &gpio->event); } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/Kbuild b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/Kbuild index 67cc3b320169..7f61a1ed158b 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/Kbuild @@ -1,3 +1,4 @@ # SPDX-License-Identifier: MIT nvkm-y += nvkm/subdev/gsp/base.o nvkm-y += nvkm/subdev/gsp/gv100.o +nvkm-y += nvkm/subdev/gsp/ga102.o diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/base.c index 22574886b819..591ac95c2669 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/base.c @@ -53,5 +53,7 @@ nvkm_gsp_new_(const struct nvkm_gsp_fwif *fwif, struct nvkm_device *device, if (IS_ERR(fwif)) return PTR_ERR(fwif); - return nvkm_falcon_ctor(fwif->flcn, &gsp->subdev, gsp->subdev.name, 0, &gsp->falcon); + gsp->func = fwif->func; + + return nvkm_falcon_ctor(gsp->func->flcn, &gsp->subdev, gsp->subdev.name, 0, &gsp->falcon); } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/ga102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/ga102.c new file mode 100644 index 000000000000..525267412c3e --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/ga102.c @@ -0,0 +1,59 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +static const struct nvkm_falcon_func +ga102_gsp_flcn = { + .disable = gm200_flcn_disable, + .enable = gm200_flcn_enable, + .select = ga102_flcn_select, + .addr2 = 0x1000, + .reset_eng = gp102_flcn_reset_eng, + .reset_prep = ga102_flcn_reset_prep, + .reset_wait_mem_scrubbing = ga102_flcn_reset_wait_mem_scrubbing, + .imem_dma = &ga102_flcn_dma, + .dmem_dma = &ga102_flcn_dma, +}; + +static const struct nvkm_gsp_func +ga102_gsp = { + .flcn = &ga102_gsp_flcn, +}; + +static int +ga102_gsp_nofw(struct nvkm_gsp *gsp, int ver, const struct nvkm_gsp_fwif *fwif) +{ + return 0; +} + +struct nvkm_gsp_fwif +ga102_gsps[] = { + { -1, ga102_gsp_nofw, &ga102_gsp }, + {} +}; + +int +ga102_gsp_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, + struct nvkm_gsp **pgsp) +{ + return nvkm_gsp_new_(ga102_gsps, device, type, inst, pgsp); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/gv100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/gv100.c index 6c4ef62a746a..da6a809cd317 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/gv100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/gv100.c @@ -23,17 +23,20 @@ static const struct nvkm_falcon_func gv100_gsp_flcn = { - .fbif = 0x600, - .load_imem = nvkm_falcon_v1_load_imem, - .load_dmem = nvkm_falcon_v1_load_dmem, - .read_dmem = nvkm_falcon_v1_read_dmem, - .bind_context = gp102_sec2_flcn_bind_context, - .wait_for_halt = nvkm_falcon_v1_wait_for_halt, - .clear_interrupt = nvkm_falcon_v1_clear_interrupt, - .set_start_addr = nvkm_falcon_v1_set_start_addr, - .start = nvkm_falcon_v1_start, - .enable = gp102_sec2_flcn_enable, - .disable = nvkm_falcon_v1_disable, + .disable = gm200_flcn_disable, + .enable = gm200_flcn_enable, + .reset_eng = gp102_flcn_reset_eng, + .reset_wait_mem_scrubbing = gm200_flcn_reset_wait_mem_scrubbing, + .bind_inst = gm200_flcn_bind_inst, + .bind_stat = gm200_flcn_bind_stat, + .bind_intr = true, + .imem_pio = &gm200_flcn_imem_pio, + .dmem_pio = &gm200_flcn_dmem_pio, +}; + +static const struct nvkm_gsp_func +gv100_gsp = { + .flcn = &gv100_gsp_flcn, }; static int @@ -43,8 +46,8 @@ gv100_gsp_nofw(struct nvkm_gsp *gsp, int ver, const struct nvkm_gsp_fwif *fwif) } static struct nvkm_gsp_fwif -gv100_gsp[] = { - { -1, gv100_gsp_nofw, &gv100_gsp_flcn }, +gv100_gsps[] = { + { -1, gv100_gsp_nofw, &gv100_gsp }, {} }; @@ -52,5 +55,5 @@ int gv100_gsp_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_gsp **pgsp) { - return nvkm_gsp_new_(gv100_gsp, device, type, inst, pgsp); + return nvkm_gsp_new_(gv100_gsps, device, type, inst, pgsp); } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/priv.h b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/priv.h index 19381ddd38d4..89749a40203c 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/priv.h @@ -4,10 +4,14 @@ #include <subdev/gsp.h> enum nvkm_acr_lsf_id; +struct nvkm_gsp_func { + const struct nvkm_falcon_func *flcn; +}; + struct nvkm_gsp_fwif { int version; int (*load)(struct nvkm_gsp *, int ver, const struct nvkm_gsp_fwif *); - const struct nvkm_falcon_func *flcn; + const struct nvkm_gsp_func *func; }; int nvkm_gsp_new_(const struct nvkm_gsp_fwif *, struct nvkm_device *, enum nvkm_subdev_type, int, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/i2c/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/i2c/base.c index cb5cb533d91c..976539de4220 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/i2c/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/i2c/base.c @@ -26,7 +26,6 @@ #include "bus.h" #include "pad.h" -#include <core/notify.h> #include <core/option.h> #include <subdev/bios.h> #include <subdev/bios/dcb.h> @@ -104,23 +103,8 @@ nvkm_i2c_intr_init(struct nvkm_event *event, int type, int id) i2c->func->aux_mask(i2c, type, aux->intr, aux->intr); } -static int -nvkm_i2c_intr_ctor(struct nvkm_object *object, void *data, u32 size, - struct nvkm_notify *notify) -{ - struct nvkm_i2c_ntfy_req *req = data; - if (!WARN_ON(size != sizeof(*req))) { - notify->size = sizeof(struct nvkm_i2c_ntfy_rep); - notify->types = req->mask; - notify->index = req->port; - return 0; - } - return -EINVAL; -} - static const struct nvkm_event_func nvkm_i2c_intr_func = { - .ctor = nvkm_i2c_intr_ctor, .init = nvkm_i2c_intr_init, .fini = nvkm_i2c_intr_fini, }; @@ -145,13 +129,8 @@ nvkm_i2c_intr(struct nvkm_subdev *subdev) if (lo & aux->intr) mask |= NVKM_I2C_UNPLUG; if (rq & aux->intr) mask |= NVKM_I2C_IRQ; if (tx & aux->intr) mask |= NVKM_I2C_DONE; - if (mask) { - struct nvkm_i2c_ntfy_rep rep = { - .mask = mask, - }; - nvkm_event_send(&i2c->event, rep.mask, aux->id, - &rep, sizeof(rep)); - } + if (mask) + nvkm_event_ntfy(&i2c->event, aux->id, mask); } } @@ -427,5 +406,5 @@ nvkm_i2c_new_(const struct nvkm_i2c_func *func, struct nvkm_device *device, } } - return nvkm_event_init(&nvkm_i2c_intr_func, 4, i, &i2c->event); + return nvkm_event_init(&nvkm_i2c_intr_func, &i2c->subdev, 4, i, &i2c->event); } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/base.c index cd8163a52bb6..e0e4f97be029 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/base.c @@ -90,6 +90,18 @@ nvkm_instobj_ctor(const struct nvkm_memory_func *func, } int +nvkm_instobj_wrap(struct nvkm_device *device, + struct nvkm_memory *memory, struct nvkm_memory **pmemory) +{ + struct nvkm_instmem *imem = device->imem; + + if (!imem->func->memory_wrap) + return -ENOSYS; + + return imem->func->memory_wrap(imem, memory, pmemory); +} + +int nvkm_instobj_new(struct nvkm_instmem *imem, u32 size, u32 align, bool zero, struct nvkm_memory **pmemory) { diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/nv50.c b/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/nv50.c index c51bac76174c..4b2d7465d22f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/nv50.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/nv50.c @@ -348,13 +348,11 @@ nv50_instobj_func = { }; static int -nv50_instobj_new(struct nvkm_instmem *base, u32 size, u32 align, bool zero, - struct nvkm_memory **pmemory) +nv50_instobj_wrap(struct nvkm_instmem *base, + struct nvkm_memory *memory, struct nvkm_memory **pmemory) { struct nv50_instmem *imem = nv50_instmem(base); struct nv50_instobj *iobj; - struct nvkm_device *device = imem->base.subdev.device; - u8 page = max(order_base_2(align), 12); if (!(iobj = kzalloc(sizeof(*iobj), GFP_KERNEL))) return -ENOMEM; @@ -365,7 +363,25 @@ nv50_instobj_new(struct nvkm_instmem *base, u32 size, u32 align, bool zero, refcount_set(&iobj->maps, 0); INIT_LIST_HEAD(&iobj->lru); - return nvkm_ram_get(device, 0, 1, page, size, true, true, &iobj->ram); + iobj->ram = nvkm_memory_ref(memory); + return 0; +} + +static int +nv50_instobj_new(struct nvkm_instmem *imem, u32 size, u32 align, bool zero, + struct nvkm_memory **pmemory) +{ + u8 page = max(order_base_2(align), 12); + struct nvkm_memory *ram; + int ret; + + ret = nvkm_ram_get(imem->subdev.device, 0, 1, page, size, true, true, &ram); + if (ret) + return ret; + + ret = nv50_instobj_wrap(imem, ram, pmemory); + nvkm_memory_unref(&ram); + return ret; } /****************************************************************************** @@ -382,6 +398,7 @@ static const struct nvkm_instmem_func nv50_instmem = { .fini = nv50_instmem_fini, .memory_new = nv50_instobj_new, + .memory_wrap = nv50_instobj_wrap, .zero = false, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/priv.h b/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/priv.h index 56c15e30a5dd..fe92986a3885 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/instmem/priv.h @@ -12,6 +12,7 @@ struct nvkm_instmem_func { void (*wr32)(struct nvkm_instmem *, u32 addr, u32 data); int (*memory_new)(struct nvkm_instmem *, u32 size, u32 align, bool zero, struct nvkm_memory **); + int (*memory_wrap)(struct nvkm_instmem *, struct nvkm_memory *, struct nvkm_memory **); bool zero; }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/Kbuild b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/Kbuild index 728d75010847..0d8a915d727e 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/Kbuild @@ -7,3 +7,4 @@ nvkm-y += nvkm/subdev/ltc/gm200.o nvkm-y += nvkm/subdev/ltc/gp100.o nvkm-y += nvkm/subdev/ltc/gp102.o nvkm-y += nvkm/subdev/ltc/gp10b.o +nvkm-y += nvkm/subdev/ltc/ga102.o diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/base.c index fa683c190795..f742a7b7b175 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/base.c @@ -97,8 +97,10 @@ nvkm_ltc_init(struct nvkm_subdev *subdev) struct nvkm_ltc *ltc = nvkm_ltc(subdev); int i; - for (i = ltc->zbc_min; i <= ltc->zbc_max; i++) { + for (i = ltc->zbc_color_min; i <= ltc->zbc_color_max; i++) ltc->func->zbc_clear_color(ltc, i, ltc->zbc_color[i]); + + for (i = ltc->zbc_depth_min; i <= ltc->zbc_depth_max; i++) { ltc->func->zbc_clear_depth(ltc, i, ltc->zbc_depth[i]); if (ltc->func->zbc_clear_stencil) ltc->func->zbc_clear_stencil(ltc, i, ltc->zbc_stencil[i]); @@ -137,7 +139,9 @@ nvkm_ltc_new_(const struct nvkm_ltc_func *func, struct nvkm_device *device, nvkm_subdev_ctor(&nvkm_ltc, device, type, inst, <c->subdev); ltc->func = func; mutex_init(<c->mutex); - ltc->zbc_min = 1; /* reserve 0 for disabled */ - ltc->zbc_max = min(func->zbc, NVKM_LTC_MAX_ZBC_CNT) - 1; + ltc->zbc_color_min = 1; /* reserve 0 for disabled */ + ltc->zbc_color_max = min(func->zbc_color, NVKM_LTC_MAX_ZBC_COLOR_CNT) - 1; + ltc->zbc_depth_min = 1; /* reserve 0 for disabled */ + ltc->zbc_depth_max = min(func->zbc_depth, NVKM_LTC_MAX_ZBC_DEPTH_CNT) - 1; return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/ga102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/ga102.c new file mode 100644 index 000000000000..159d9f8c95f3 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/ga102.c @@ -0,0 +1,57 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +static void +ga102_ltc_zbc_clear_color(struct nvkm_ltc *ltc, int i, const u32 color[4]) +{ + struct nvkm_device *device = ltc->subdev.device; + + nvkm_mask(device, 0x17e338, 0x0000001f, i); + nvkm_wr32(device, 0x17e33c, color[0]); + nvkm_wr32(device, 0x17e340, color[1]); + nvkm_wr32(device, 0x17e344, color[2]); + nvkm_wr32(device, 0x17e348, color[3]); +} + +static const struct nvkm_ltc_func +ga102_ltc = { + .oneinit = gp100_ltc_oneinit, + .init = gp100_ltc_init, + .intr = gp100_ltc_intr, + .cbc_clear = gm107_ltc_cbc_clear, + .cbc_wait = gm107_ltc_cbc_wait, + .zbc_color = 31, + .zbc_depth = 16, + .zbc_clear_color = ga102_ltc_zbc_clear_color, + .zbc_clear_depth = gm107_ltc_zbc_clear_depth, + .zbc_clear_stencil = gp102_ltc_zbc_clear_stencil, + .invalidate = gf100_ltc_invalidate, + .flush = gf100_ltc_flush, +}; + +int +ga102_ltc_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, + struct nvkm_ltc **pltc) +{ + return nvkm_ltc_new_(&ga102_ltc, device, type, inst, pltc); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gf100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gf100.c index fd8aeafc812d..de71ba3c9292 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gf100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gf100.c @@ -241,7 +241,8 @@ gf100_ltc = { .intr = gf100_ltc_intr, .cbc_clear = gf100_ltc_cbc_clear, .cbc_wait = gf100_ltc_cbc_wait, - .zbc = 16, + .zbc_color = 16, + .zbc_depth = 16, .zbc_clear_color = gf100_ltc_zbc_clear_color, .zbc_clear_depth = gf100_ltc_zbc_clear_depth, .invalidate = gf100_ltc_invalidate, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gk104.c b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gk104.c index 94aa09244d67..5d61e3c6ff59 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gk104.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gk104.c @@ -42,7 +42,8 @@ gk104_ltc = { .intr = gf100_ltc_intr, .cbc_clear = gf100_ltc_cbc_clear, .cbc_wait = gf100_ltc_cbc_wait, - .zbc = 16, + .zbc_color = 16, + .zbc_depth = 16, .zbc_clear_color = gf100_ltc_zbc_clear_color, .zbc_clear_depth = gf100_ltc_zbc_clear_depth, .invalidate = gf100_ltc_invalidate, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gm107.c b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gm107.c index 54d1d65d5a85..18685d849657 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gm107.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gm107.c @@ -137,7 +137,8 @@ gm107_ltc = { .intr = gm107_ltc_intr, .cbc_clear = gm107_ltc_cbc_clear, .cbc_wait = gm107_ltc_cbc_wait, - .zbc = 16, + .zbc_color = 16, + .zbc_depth = 16, .zbc_clear_color = gm107_ltc_zbc_clear_color, .zbc_clear_depth = gm107_ltc_zbc_clear_depth, .invalidate = gf100_ltc_invalidate, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gm200.c b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gm200.c index 8cfdbbdd8e8d..7a9464b9def5 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gm200.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gm200.c @@ -49,7 +49,8 @@ gm200_ltc = { .intr = gm107_ltc_intr, .cbc_clear = gm107_ltc_cbc_clear, .cbc_wait = gm107_ltc_cbc_wait, - .zbc = 16, + .zbc_color = 16, + .zbc_depth = 16, .zbc_clear_color = gm107_ltc_zbc_clear_color, .zbc_clear_depth = gm107_ltc_zbc_clear_depth, .invalidate = gf100_ltc_invalidate, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp100.c index a4a6cd9b435a..1a17a451754c 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp100.c @@ -61,7 +61,8 @@ gp100_ltc = { .intr = gp100_ltc_intr, .cbc_clear = gm107_ltc_cbc_clear, .cbc_wait = gm107_ltc_cbc_wait, - .zbc = 16, + .zbc_color = 16, + .zbc_depth = 16, .zbc_clear_color = gm107_ltc_zbc_clear_color, .zbc_clear_depth = gm107_ltc_zbc_clear_depth, .invalidate = gf100_ltc_invalidate, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp102.c index ff05d617e7f4..265a05fd5f6b 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp102.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp102.c @@ -36,7 +36,8 @@ gp102_ltc = { .intr = gp100_ltc_intr, .cbc_clear = gm107_ltc_cbc_clear, .cbc_wait = gm107_ltc_cbc_wait, - .zbc = 16, + .zbc_color = 16, + .zbc_depth = 16, .zbc_clear_color = gm107_ltc_zbc_clear_color, .zbc_clear_depth = gm107_ltc_zbc_clear_depth, .zbc_clear_stencil = gp102_ltc_zbc_clear_stencil, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c index dfebd796cb4b..e7e8fdf3adab 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/gp10b.c @@ -50,7 +50,8 @@ gp10b_ltc = { .intr = gp100_ltc_intr, .cbc_clear = gm107_ltc_cbc_clear, .cbc_wait = gm107_ltc_cbc_wait, - .zbc = 16, + .zbc_color = 16, + .zbc_depth = 16, .zbc_clear_color = gm107_ltc_zbc_clear_color, .zbc_clear_depth = gm107_ltc_zbc_clear_depth, .zbc_clear_stencil = gp102_ltc_zbc_clear_stencil, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/priv.h b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/priv.h index 2bebe139005d..134e90c9e861 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/ltc/priv.h @@ -16,7 +16,8 @@ struct nvkm_ltc_func { void (*cbc_clear)(struct nvkm_ltc *, u32 start, u32 limit); void (*cbc_wait)(struct nvkm_ltc *); - int zbc; + int zbc_color; + int zbc_depth; void (*zbc_clear_color)(struct nvkm_ltc *, int, const u32[4]); void (*zbc_clear_depth)(struct nvkm_ltc *, int, const u32); void (*zbc_clear_stencil)(struct nvkm_ltc *, int, const u32); diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/Kbuild b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/Kbuild index ac2b34e9ac6a..2a3255ced8b7 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/Kbuild +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/Kbuild @@ -13,5 +13,4 @@ nvkm-y += nvkm/subdev/mc/gk104.o nvkm-y += nvkm/subdev/mc/gk20a.o nvkm-y += nvkm/subdev/mc/gp100.o nvkm-y += nvkm/subdev/mc/gp10b.o -nvkm-y += nvkm/subdev/mc/tu102.o nvkm-y += nvkm/subdev/mc/ga100.o diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/base.c index 21c4af3f81d5..c85600ba69f9 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/base.c @@ -37,84 +37,14 @@ nvkm_mc_unk260(struct nvkm_device *device, u32 data) void nvkm_mc_intr_mask(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, bool en) { - struct nvkm_mc *mc = device->mc; - const struct nvkm_mc_map *map; - if (likely(mc) && mc->func->intr_mask) { - u32 mask = nvkm_top_intr_mask(device, type, inst); - for (map = mc->func->intr; !mask && map->stat; map++) { - if (map->type == type && map->inst == inst) - mask = map->stat; - } - mc->func->intr_mask(mc, mask, en ? mask : 0); - } -} - -void -nvkm_mc_intr_unarm(struct nvkm_device *device) -{ - struct nvkm_mc *mc = device->mc; - if (likely(mc)) - mc->func->intr_unarm(mc); -} - -void -nvkm_mc_intr_rearm(struct nvkm_device *device) -{ - struct nvkm_mc *mc = device->mc; - if (likely(mc)) - mc->func->intr_rearm(mc); -} - -static u32 -nvkm_mc_intr_stat(struct nvkm_mc *mc) -{ - u32 intr = mc->func->intr_stat(mc); - if (WARN_ON_ONCE(intr == 0xffffffff)) - intr = 0; /* likely fallen off the bus */ - return intr; -} - -void -nvkm_mc_intr(struct nvkm_device *device, bool *handled) -{ - struct nvkm_mc *mc = device->mc; - struct nvkm_top *top = device->top; - struct nvkm_top_device *tdev; - struct nvkm_subdev *subdev; - const struct nvkm_mc_map *map; - u32 stat, intr; - - if (unlikely(!mc)) - return; - - stat = intr = nvkm_mc_intr_stat(mc); - - if (top) { - list_for_each_entry(tdev, &top->device, head) { - if (tdev->intr >= 0 && (stat & BIT(tdev->intr))) { - subdev = nvkm_device_subdev(device, tdev->type, tdev->inst); - if (subdev) { - nvkm_subdev_intr(subdev); - stat &= ~BIT(tdev->intr); - if (!stat) - break; - } - } - } - } + struct nvkm_subdev *subdev = nvkm_device_subdev(device, type, inst); - for (map = mc->func->intr; map->stat; map++) { - if (intr & map->stat) { - subdev = nvkm_device_subdev(device, map->type, map->inst); - if (subdev) - nvkm_subdev_intr(subdev); - stat &= ~map->stat; - } + if (subdev) { + if (en) + nvkm_intr_allow(subdev, NVKM_INTR_SUBDEV); + else + nvkm_intr_block(subdev, NVKM_INTR_SUBDEV); } - - if (stat) - nvkm_error(&mc->subdev, "intr %08x\n", stat); - *handled = intr != 0; } static u32 @@ -143,9 +73,8 @@ nvkm_mc_reset(struct nvkm_device *device, enum nvkm_subdev_type type, int inst) { u64 pmc_enable = nvkm_mc_reset_mask(device, true, type, inst); if (pmc_enable) { - nvkm_mask(device, 0x000200, pmc_enable, 0x00000000); - nvkm_mask(device, 0x000200, pmc_enable, pmc_enable); - nvkm_rd32(device, 0x000200); + device->mc->func->device->disable(device->mc, pmc_enable); + device->mc->func->device->enable(device->mc, pmc_enable); } } @@ -154,17 +83,15 @@ nvkm_mc_disable(struct nvkm_device *device, enum nvkm_subdev_type type, int inst { u64 pmc_enable = nvkm_mc_reset_mask(device, false, type, inst); if (pmc_enable) - nvkm_mask(device, 0x000200, pmc_enable, 0x00000000); + device->mc->func->device->disable(device->mc, pmc_enable); } void nvkm_mc_enable(struct nvkm_device *device, enum nvkm_subdev_type type, int inst) { u64 pmc_enable = nvkm_mc_reset_mask(device, false, type, inst); - if (pmc_enable) { - nvkm_mask(device, 0x000200, pmc_enable, pmc_enable); - nvkm_rd32(device, 0x000200); - } + if (pmc_enable) + device->mc->func->device->enable(device->mc, pmc_enable); } bool @@ -172,16 +99,7 @@ nvkm_mc_enabled(struct nvkm_device *device, enum nvkm_subdev_type type, int inst { u64 pmc_enable = nvkm_mc_reset_mask(device, false, type, inst); - return (pmc_enable != 0) && - ((nvkm_rd32(device, 0x000200) & pmc_enable) == pmc_enable); -} - - -static int -nvkm_mc_fini(struct nvkm_subdev *subdev, bool suspend) -{ - nvkm_mc_intr_unarm(subdev->device); - return 0; + return (pmc_enable != 0) && device->mc->func->device->enabled(device->mc, pmc_enable); } static int @@ -190,7 +108,6 @@ nvkm_mc_init(struct nvkm_subdev *subdev) struct nvkm_mc *mc = nvkm_mc(subdev); if (mc->func->init) mc->func->init(mc); - nvkm_mc_intr_rearm(subdev->device); return 0; } @@ -204,24 +121,27 @@ static const struct nvkm_subdev_func nvkm_mc = { .dtor = nvkm_mc_dtor, .init = nvkm_mc_init, - .fini = nvkm_mc_fini, }; -void -nvkm_mc_ctor(const struct nvkm_mc_func *func, struct nvkm_device *device, - enum nvkm_subdev_type type, int inst, struct nvkm_mc *mc) -{ - nvkm_subdev_ctor(&nvkm_mc, device, type, inst, &mc->subdev); - mc->func = func; -} - int nvkm_mc_new_(const struct nvkm_mc_func *func, struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_mc **pmc) { struct nvkm_mc *mc; + int ret; + if (!(mc = *pmc = kzalloc(sizeof(*mc), GFP_KERNEL))) return -ENOMEM; - nvkm_mc_ctor(func, device, type, inst, *pmc); + + nvkm_subdev_ctor(&nvkm_mc, device, type, inst, &mc->subdev); + mc->func = func; + + if (mc->func->intr) { + ret = nvkm_intr_add(mc->func->intr, mc->func->intrs, &mc->subdev, + mc->func->intr_nonstall ? 2 : 1, &mc->intr); + if (ret) + return ret; + } + return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/g84.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/g84.c index 4cfc1c984006..f4ee99137b1f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/g84.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/g84.c @@ -34,30 +34,29 @@ g84_mc_reset[] = { {} }; -static const struct nvkm_mc_map -g84_mc_intr[] = { - { 0x04000000, NVKM_ENGINE_DISP }, - { 0x00020000, NVKM_ENGINE_VP }, - { 0x00008000, NVKM_ENGINE_BSP }, - { 0x00004000, NVKM_ENGINE_CIPHER }, - { 0x00001000, NVKM_ENGINE_GR }, - { 0x00000100, NVKM_ENGINE_FIFO }, - { 0x00000001, NVKM_ENGINE_MPEG }, - { 0x0002d101, NVKM_SUBDEV_FB }, - { 0x10000000, NVKM_SUBDEV_BUS }, - { 0x00200000, NVKM_SUBDEV_GPIO }, - { 0x00200000, NVKM_SUBDEV_I2C }, - { 0x00100000, NVKM_SUBDEV_TIMER }, +static const struct nvkm_intr_data +g84_mc_intrs[] = { + { NVKM_ENGINE_DISP , 0, 0, 0x04000000, true }, + { NVKM_ENGINE_VP , 0, 0, 0x00020000, true }, + { NVKM_ENGINE_BSP , 0, 0, 0x00008000, true }, + { NVKM_ENGINE_CIPHER, 0, 0, 0x00004000, true }, + { NVKM_ENGINE_GR , 0, 0, 0x00001000, true }, + { NVKM_ENGINE_FIFO , 0, 0, 0x00000100 }, + { NVKM_ENGINE_MPEG , 0, 0, 0x00000001, true }, + { NVKM_SUBDEV_FB , 0, 0, 0x0002d101, true }, + { NVKM_SUBDEV_BUS , 0, 0, 0x10000000, true }, + { NVKM_SUBDEV_GPIO , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_I2C , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_TIMER , 0, 0, 0x00100000, true }, {}, }; static const struct nvkm_mc_func g84_mc = { .init = nv50_mc_init, - .intr = g84_mc_intr, - .intr_unarm = nv04_mc_intr_unarm, - .intr_rearm = nv04_mc_intr_rearm, - .intr_stat = nv04_mc_intr_stat, + .intr = &nv04_mc_intr, + .intrs = g84_mc_intrs, + .device = &nv04_mc_device, .reset = g84_mc_reset, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/g98.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/g98.c index b7e58d75d894..f42684809f08 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/g98.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/g98.c @@ -34,30 +34,29 @@ g98_mc_reset[] = { {} }; -static const struct nvkm_mc_map -g98_mc_intr[] = { - { 0x04000000, NVKM_ENGINE_DISP }, - { 0x00020000, NVKM_ENGINE_MSPDEC }, - { 0x00008000, NVKM_ENGINE_MSVLD }, - { 0x00004000, NVKM_ENGINE_SEC }, - { 0x00001000, NVKM_ENGINE_GR }, - { 0x00000100, NVKM_ENGINE_FIFO }, - { 0x00000001, NVKM_ENGINE_MSPPP }, - { 0x0002d101, NVKM_SUBDEV_FB }, - { 0x10000000, NVKM_SUBDEV_BUS }, - { 0x00200000, NVKM_SUBDEV_GPIO }, - { 0x00200000, NVKM_SUBDEV_I2C }, - { 0x00100000, NVKM_SUBDEV_TIMER }, +static const struct nvkm_intr_data +g98_mc_intrs[] = { + { NVKM_ENGINE_DISP , 0, 0, 0x04000000, true }, + { NVKM_ENGINE_MSPDEC, 0, 0, 0x00020000, true }, + { NVKM_ENGINE_MSVLD , 0, 0, 0x00008000, true }, + { NVKM_ENGINE_SEC , 0, 0, 0x00004000, true }, + { NVKM_ENGINE_GR , 0, 0, 0x00001000, true }, + { NVKM_ENGINE_FIFO , 0, 0, 0x00000100 }, + { NVKM_ENGINE_MSPPP , 0, 0, 0x00000001, true }, + { NVKM_SUBDEV_FB , 0, 0, 0x0002d101, true }, + { NVKM_SUBDEV_BUS , 0, 0, 0x10000000, true }, + { NVKM_SUBDEV_GPIO , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_I2C , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_TIMER , 0, 0, 0x00100000, true }, {}, }; static const struct nvkm_mc_func g98_mc = { .init = nv50_mc_init, - .intr = g98_mc_intr, - .intr_unarm = nv04_mc_intr_unarm, - .intr_rearm = nv04_mc_intr_rearm, - .intr_stat = nv04_mc_intr_stat, + .intr = &nv04_mc_intr, + .intrs = g98_mc_intrs, + .device = &nv04_mc_device, .reset = g98_mc_reset, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/ga100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/ga100.c index 4105175dfccd..1e2eabec1a76 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/ga100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/ga100.c @@ -22,49 +22,51 @@ #include "priv.h" static void -ga100_mc_intr_unarm(struct nvkm_mc *mc) +ga100_mc_device_disable(struct nvkm_mc *mc, u32 mask) { - nvkm_wr32(mc->subdev.device, 0xb81610, 0x00000004); -} + struct nvkm_device *device = mc->subdev.device; -static void -ga100_mc_intr_rearm(struct nvkm_mc *mc) -{ - nvkm_wr32(mc->subdev.device, 0xb81608, 0x00000004); + nvkm_mask(device, 0x000600, mask, 0x00000000); + nvkm_rd32(device, 0x000600); + nvkm_rd32(device, 0x000600); } static void -ga100_mc_intr_mask(struct nvkm_mc *mc, u32 mask, u32 intr) +ga100_mc_device_enable(struct nvkm_mc *mc, u32 mask) { - nvkm_wr32(mc->subdev.device, 0xb81210, mask & intr ); - nvkm_wr32(mc->subdev.device, 0xb81410, mask & ~(mask & intr)); + struct nvkm_device *device = mc->subdev.device; + + nvkm_mask(device, 0x000600, mask, mask); + nvkm_rd32(device, 0x000600); + nvkm_rd32(device, 0x000600); } -static u32 -ga100_mc_intr_stat(struct nvkm_mc *mc) +static bool +ga100_mc_device_enabled(struct nvkm_mc *mc, u32 mask) { - u32 intr_top = nvkm_rd32(mc->subdev.device, 0xb81600), intr = 0x00000000; - if (intr_top & 0x00000004) - intr = nvkm_mask(mc->subdev.device, 0xb81010, 0x00000000, 0x00000000); - return intr; + return (nvkm_rd32(mc->subdev.device, 0x000600) & mask) == mask; } +const struct nvkm_mc_device_func +ga100_mc_device = { + .enabled = ga100_mc_device_enabled, + .enable = ga100_mc_device_enable, + .disable = ga100_mc_device_disable, +}; + static void ga100_mc_init(struct nvkm_mc *mc) { - nv50_mc_init(mc); - nvkm_wr32(mc->subdev.device, 0xb81210, 0xffffffff); + struct nvkm_device *device = mc->subdev.device; + + nvkm_wr32(device, 0x000200, 0xffffffff); + nvkm_wr32(device, 0x000600, 0xffffffff); } static const struct nvkm_mc_func ga100_mc = { .init = ga100_mc_init, - .intr = gp100_mc_intr, - .intr_unarm = ga100_mc_intr_unarm, - .intr_rearm = ga100_mc_intr_rearm, - .intr_mask = ga100_mc_intr_mask, - .intr_stat = ga100_mc_intr_stat, - .reset = gk104_mc_reset, + .device = &ga100_mc_device, }; int diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gf100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gf100.c index 3a589c6f7fad..ab1eaa37123a 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gf100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gf100.c @@ -36,64 +36,29 @@ gf100_mc_reset[] = { {} }; -static const struct nvkm_mc_map -gf100_mc_intr[] = { - { 0x04000000, NVKM_ENGINE_DISP }, - { 0x00020000, NVKM_ENGINE_MSPDEC }, - { 0x00008000, NVKM_ENGINE_MSVLD }, - { 0x00001000, NVKM_ENGINE_GR }, - { 0x00000100, NVKM_ENGINE_FIFO }, - { 0x00000040, NVKM_ENGINE_CE, 1 }, - { 0x00000020, NVKM_ENGINE_CE, 0 }, - { 0x00000001, NVKM_ENGINE_MSPPP }, - { 0x40000000, NVKM_SUBDEV_PRIVRING }, - { 0x10000000, NVKM_SUBDEV_BUS }, - { 0x08000000, NVKM_SUBDEV_FB }, - { 0x02000000, NVKM_SUBDEV_LTC }, - { 0x01000000, NVKM_SUBDEV_PMU }, - { 0x00200000, NVKM_SUBDEV_GPIO }, - { 0x00200000, NVKM_SUBDEV_I2C }, - { 0x00100000, NVKM_SUBDEV_TIMER }, - { 0x00040000, NVKM_SUBDEV_THERM }, - { 0x00002000, NVKM_SUBDEV_FB }, +static const struct nvkm_intr_data +gf100_mc_intrs[] = { + { NVKM_ENGINE_DISP , 0, 0, 0x04000000, true }, + { NVKM_ENGINE_MSPDEC , 0, 0, 0x00020000, true }, + { NVKM_ENGINE_MSVLD , 0, 0, 0x00008000, true }, + { NVKM_ENGINE_GR , 0, 0, 0x00001000 }, + { NVKM_ENGINE_FIFO , 0, 0, 0x00000100 }, + { NVKM_ENGINE_CE , 1, 0, 0x00000040, true }, + { NVKM_ENGINE_CE , 0, 0, 0x00000020, true }, + { NVKM_ENGINE_MSPPP , 0, 0, 0x00000001, true }, + { NVKM_SUBDEV_PRIVRING, 0, 0, 0x40000000, true }, + { NVKM_SUBDEV_BUS , 0, 0, 0x10000000, true }, + { NVKM_SUBDEV_FB , 0, 0, 0x08002000, true }, + { NVKM_SUBDEV_LTC , 0, 0, 0x02000000, true }, + { NVKM_SUBDEV_PMU , 0, 0, 0x01000000, true }, + { NVKM_SUBDEV_GPIO , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_I2C , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_TIMER , 0, 0, 0x00100000, true }, + { NVKM_SUBDEV_THERM , 0, 0, 0x00040000, true }, {}, }; void -gf100_mc_intr_unarm(struct nvkm_mc *mc) -{ - struct nvkm_device *device = mc->subdev.device; - nvkm_wr32(device, 0x000140, 0x00000000); - nvkm_wr32(device, 0x000144, 0x00000000); - nvkm_rd32(device, 0x000140); -} - -void -gf100_mc_intr_rearm(struct nvkm_mc *mc) -{ - struct nvkm_device *device = mc->subdev.device; - nvkm_wr32(device, 0x000140, 0x00000001); - nvkm_wr32(device, 0x000144, 0x00000001); -} - -u32 -gf100_mc_intr_stat(struct nvkm_mc *mc) -{ - struct nvkm_device *device = mc->subdev.device; - u32 intr0 = nvkm_rd32(device, 0x000100); - u32 intr1 = nvkm_rd32(device, 0x000104); - return intr0 | intr1; -} - -void -gf100_mc_intr_mask(struct nvkm_mc *mc, u32 mask, u32 stat) -{ - struct nvkm_device *device = mc->subdev.device; - nvkm_mask(device, 0x000640, mask, stat); - nvkm_mask(device, 0x000644, mask, stat); -} - -void gf100_mc_unk260(struct nvkm_mc *mc, u32 data) { nvkm_wr32(mc->subdev.device, 0x000260, data); @@ -102,12 +67,11 @@ gf100_mc_unk260(struct nvkm_mc *mc, u32 data) static const struct nvkm_mc_func gf100_mc = { .init = nv50_mc_init, - .intr = gf100_mc_intr, - .intr_unarm = gf100_mc_intr_unarm, - .intr_rearm = gf100_mc_intr_rearm, - .intr_mask = gf100_mc_intr_mask, - .intr_stat = gf100_mc_intr_stat, + .intr = >215_mc_intr, + .intrs = gf100_mc_intrs, + .intr_nonstall = true, .reset = gf100_mc_reset, + .device = &nv04_mc_device, .unk260 = gf100_mc_unk260, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gk104.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gk104.c index d9b9067fa93f..66829586a124 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gk104.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gk104.c @@ -30,32 +30,32 @@ gk104_mc_reset[] = { {} }; -const struct nvkm_mc_map -gk104_mc_intr[] = { - { 0x04000000, NVKM_ENGINE_DISP }, - { 0x00000100, NVKM_ENGINE_FIFO }, - { 0x40000000, NVKM_SUBDEV_PRIVRING }, - { 0x10000000, NVKM_SUBDEV_BUS }, - { 0x08000000, NVKM_SUBDEV_FB }, - { 0x02000000, NVKM_SUBDEV_LTC }, - { 0x01000000, NVKM_SUBDEV_PMU }, - { 0x00200000, NVKM_SUBDEV_GPIO }, - { 0x00200000, NVKM_SUBDEV_I2C }, - { 0x00100000, NVKM_SUBDEV_TIMER }, - { 0x00040000, NVKM_SUBDEV_THERM }, - { 0x00002000, NVKM_SUBDEV_FB }, +const struct nvkm_intr_data +gk104_mc_intrs[] = { + { NVKM_ENGINE_DISP , 0, 0, 0x04000000, true }, + { NVKM_ENGINE_FIFO , 0, 0, 0x00000100 }, + { NVKM_SUBDEV_PRIVRING, 0, 0, 0x40000000, true }, + { NVKM_SUBDEV_BUS , 0, 0, 0x10000000, true }, + { NVKM_SUBDEV_FB , 0, 0, 0x08002000, true }, + { NVKM_SUBDEV_LTC , 0, 0, 0x02000000, true }, + { NVKM_SUBDEV_PMU , 0, 0, 0x01000000, true }, + { NVKM_SUBDEV_GPIO , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_I2C , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_TIMER , 0, 0, 0x00100000, true }, + { NVKM_SUBDEV_THERM , 0, 0, 0x00040000, true }, + { NVKM_SUBDEV_TOP , 0, 0, 0x00001000 }, + { NVKM_SUBDEV_TOP , 0, 0, 0xffffefff, true }, {}, }; static const struct nvkm_mc_func gk104_mc = { .init = nv50_mc_init, - .intr = gk104_mc_intr, - .intr_unarm = gf100_mc_intr_unarm, - .intr_rearm = gf100_mc_intr_rearm, - .intr_mask = gf100_mc_intr_mask, - .intr_stat = gf100_mc_intr_stat, + .intr = >215_mc_intr, + .intrs = gk104_mc_intrs, + .intr_nonstall = true, .reset = gk104_mc_reset, + .device = &nv04_mc_device, .unk260 = gf100_mc_unk260, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gk20a.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gk20a.c index 03590292749a..d98a6563a411 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gk20a.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gk20a.c @@ -26,11 +26,10 @@ static const struct nvkm_mc_func gk20a_mc = { .init = nv50_mc_init, - .intr = gk104_mc_intr, - .intr_unarm = gf100_mc_intr_unarm, - .intr_rearm = gf100_mc_intr_rearm, - .intr_mask = gf100_mc_intr_mask, - .intr_stat = gf100_mc_intr_stat, + .intr = >215_mc_intr, + .intrs = gk104_mc_intrs, + .intr_nonstall = true, + .device = &nv04_mc_device, .reset = gk104_mc_reset, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gp100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gp100.c index 5fd1a0595c33..eb2ab03f4360 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gp100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gp100.c @@ -21,108 +21,82 @@ * * Authors: Ben Skeggs */ -#define gp100_mc(p) container_of((p), struct gp100_mc, base) #include "priv.h" -struct gp100_mc { - struct nvkm_mc base; - spinlock_t lock; - bool intr; - u32 mask; +const struct nvkm_intr_data +gp100_mc_intrs[] = { + { NVKM_ENGINE_DISP , 0, 0, 0x04000000, true }, + { NVKM_ENGINE_FIFO , 0, 0, 0x00000100 }, + { NVKM_SUBDEV_FAULT , 0, 0, 0x00000200, true }, + { NVKM_SUBDEV_PRIVRING, 0, 0, 0x40000000, true }, + { NVKM_SUBDEV_BUS , 0, 0, 0x10000000, true }, + { NVKM_SUBDEV_FB , 0, 0, 0x08002000, true }, + { NVKM_SUBDEV_LTC , 0, 0, 0x02000000, true }, + { NVKM_SUBDEV_PMU , 0, 0, 0x01000000, true }, + { NVKM_SUBDEV_GPIO , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_I2C , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_TIMER , 0, 0, 0x00100000, true }, + { NVKM_SUBDEV_THERM , 0, 0, 0x00040000, true }, + { NVKM_SUBDEV_TOP , 0, 0, 0x00009000 }, + { NVKM_SUBDEV_TOP , 0, 0, 0xffff6fff, true }, + {}, }; static void -gp100_mc_intr_update(struct gp100_mc *mc) +gp100_mc_intr_allow(struct nvkm_intr *intr, int leaf, u32 mask) { - struct nvkm_device *device = mc->base.subdev.device; - u32 mask = mc->intr ? mc->mask : 0, i; - for (i = 0; i < 2; i++) { - nvkm_wr32(device, 0x000180 + (i * 0x04), ~mask); - nvkm_wr32(device, 0x000160 + (i * 0x04), mask); - } + struct nvkm_mc *mc = container_of(intr, typeof(*mc), intr); + + nvkm_wr32(mc->subdev.device, 0x000160 + (leaf * 4), mask); } -void -gp100_mc_intr_unarm(struct nvkm_mc *base) +static void +gp100_mc_intr_block(struct nvkm_intr *intr, int leaf, u32 mask) { - struct gp100_mc *mc = gp100_mc(base); - unsigned long flags; - spin_lock_irqsave(&mc->lock, flags); - mc->intr = false; - gp100_mc_intr_update(mc); - spin_unlock_irqrestore(&mc->lock, flags); + struct nvkm_mc *mc = container_of(intr, typeof(*mc), intr); + + nvkm_wr32(mc->subdev.device, 0x000180 + (leaf * 4), mask); } -void -gp100_mc_intr_rearm(struct nvkm_mc *base) +static void +gp100_mc_intr_rearm(struct nvkm_intr *intr) { - struct gp100_mc *mc = gp100_mc(base); - unsigned long flags; - spin_lock_irqsave(&mc->lock, flags); - mc->intr = true; - gp100_mc_intr_update(mc); - spin_unlock_irqrestore(&mc->lock, flags); + int i; + + for (i = 0; i < intr->leaves; i++) + intr->func->allow(intr, i, intr->mask[i]); } -void -gp100_mc_intr_mask(struct nvkm_mc *base, u32 mask, u32 intr) +static void +gp100_mc_intr_unarm(struct nvkm_intr *intr) { - struct gp100_mc *mc = gp100_mc(base); - unsigned long flags; - spin_lock_irqsave(&mc->lock, flags); - mc->mask = (mc->mask & ~mask) | intr; - gp100_mc_intr_update(mc); - spin_unlock_irqrestore(&mc->lock, flags); + int i; + + for (i = 0; i < intr->leaves; i++) + intr->func->block(intr, i, 0xffffffff); } -const struct nvkm_mc_map -gp100_mc_intr[] = { - { 0x04000000, NVKM_ENGINE_DISP }, - { 0x00000100, NVKM_ENGINE_FIFO }, - { 0x00000200, NVKM_SUBDEV_FAULT }, - { 0x40000000, NVKM_SUBDEV_PRIVRING }, - { 0x10000000, NVKM_SUBDEV_BUS }, - { 0x08000000, NVKM_SUBDEV_FB }, - { 0x02000000, NVKM_SUBDEV_LTC }, - { 0x01000000, NVKM_SUBDEV_PMU }, - { 0x00200000, NVKM_SUBDEV_GPIO }, - { 0x00200000, NVKM_SUBDEV_I2C }, - { 0x00100000, NVKM_SUBDEV_TIMER }, - { 0x00040000, NVKM_SUBDEV_THERM }, - { 0x00002000, NVKM_SUBDEV_FB }, - {}, +const struct nvkm_intr_func +gp100_mc_intr = { + .pending = nv04_mc_intr_pending, + .unarm = gp100_mc_intr_unarm, + .rearm = gp100_mc_intr_rearm, + .block = gp100_mc_intr_block, + .allow = gp100_mc_intr_allow, }; static const struct nvkm_mc_func gp100_mc = { .init = nv50_mc_init, - .intr = gp100_mc_intr, - .intr_unarm = gp100_mc_intr_unarm, - .intr_rearm = gp100_mc_intr_rearm, - .intr_mask = gp100_mc_intr_mask, - .intr_stat = gf100_mc_intr_stat, + .intr = &gp100_mc_intr, + .intrs = gp100_mc_intrs, + .intr_nonstall = true, + .device = &nv04_mc_device, .reset = gk104_mc_reset, }; int -gp100_mc_new_(const struct nvkm_mc_func *func, struct nvkm_device *device, - enum nvkm_subdev_type type, int inst, struct nvkm_mc **pmc) -{ - struct gp100_mc *mc; - - if (!(mc = kzalloc(sizeof(*mc), GFP_KERNEL))) - return -ENOMEM; - nvkm_mc_ctor(func, device, type, inst, &mc->base); - *pmc = &mc->base; - - spin_lock_init(&mc->lock); - mc->intr = false; - mc->mask = 0x7fffffff; - return 0; -} - -int gp100_mc_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_mc **pmc) { - return gp100_mc_new_(&gp100_mc, device, type, inst, pmc); + return nvkm_mc_new_(&gp100_mc, device, type, inst, pmc); } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gp10b.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gp10b.c index dd581d030ced..9bed9c5ea5d3 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gp10b.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gp10b.c @@ -34,16 +34,15 @@ gp10b_mc_init(struct nvkm_mc *mc) static const struct nvkm_mc_func gp10b_mc = { .init = gp10b_mc_init, - .intr = gp100_mc_intr, - .intr_unarm = gp100_mc_intr_unarm, - .intr_rearm = gp100_mc_intr_rearm, - .intr_mask = gp100_mc_intr_mask, - .intr_stat = gf100_mc_intr_stat, + .intr = &gp100_mc_intr, + .intrs = gp100_mc_intrs, + .intr_nonstall = true, + .device = &nv04_mc_device, .reset = gk104_mc_reset, }; int gp10b_mc_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_mc **pmc) { - return gp100_mc_new_(&gp10b_mc, device, type, inst, pmc); + return nvkm_mc_new_(&gp10b_mc, device, type, inst, pmc); } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gt215.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gt215.c index 1b4d43531dba..3d61836e42a3 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gt215.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/gt215.c @@ -34,39 +34,56 @@ gt215_mc_reset[] = { {} }; -static const struct nvkm_mc_map -gt215_mc_intr[] = { - { 0x04000000, NVKM_ENGINE_DISP }, - { 0x00400000, NVKM_ENGINE_CE, 0 }, - { 0x00020000, NVKM_ENGINE_MSPDEC }, - { 0x00008000, NVKM_ENGINE_MSVLD }, - { 0x00001000, NVKM_ENGINE_GR }, - { 0x00000100, NVKM_ENGINE_FIFO }, - { 0x00000001, NVKM_ENGINE_MSPPP }, - { 0x00429101, NVKM_SUBDEV_FB }, - { 0x10000000, NVKM_SUBDEV_BUS }, - { 0x00200000, NVKM_SUBDEV_GPIO }, - { 0x00200000, NVKM_SUBDEV_I2C }, - { 0x00100000, NVKM_SUBDEV_TIMER }, - { 0x00080000, NVKM_SUBDEV_THERM }, - { 0x00040000, NVKM_SUBDEV_PMU }, +static const struct nvkm_intr_data +gt215_mc_intrs[] = { + { NVKM_ENGINE_DISP , 0, 0, 0x04000000, true }, + { NVKM_ENGINE_CE , 0, 0, 0x00400000, true }, + { NVKM_ENGINE_MSPDEC, 0, 0, 0x00020000, true }, + { NVKM_ENGINE_MSVLD , 0, 0, 0x00008000, true }, + { NVKM_ENGINE_GR , 0, 0, 0x00001000, true }, + { NVKM_ENGINE_FIFO , 0, 0, 0x00000100 }, + { NVKM_ENGINE_MSPPP , 0, 0, 0x00000001, true }, + { NVKM_SUBDEV_FB , 0, 0, 0x00429101, true }, + { NVKM_SUBDEV_BUS , 0, 0, 0x10000000, true }, + { NVKM_SUBDEV_GPIO , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_I2C , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_TIMER , 0, 0, 0x00100000, true }, + { NVKM_SUBDEV_THERM , 0, 0, 0x00080000, true }, + { NVKM_SUBDEV_PMU , 0, 0, 0x00040000, true }, {}, }; static void -gt215_mc_intr_mask(struct nvkm_mc *mc, u32 mask, u32 stat) +gt215_mc_intr_allow(struct nvkm_intr *intr, int leaf, u32 mask) +{ + struct nvkm_mc *mc = container_of(intr, typeof(*mc), intr); + + nvkm_mask(mc->subdev.device, 0x000640 + (leaf * 4), mask, mask); +} + +static void +gt215_mc_intr_block(struct nvkm_intr *intr, int leaf, u32 mask) { - nvkm_mask(mc->subdev.device, 0x000640, mask, stat); + struct nvkm_mc *mc = container_of(intr, typeof(*mc), intr); + + nvkm_mask(mc->subdev.device, 0x000640 + (leaf * 4), mask, 0); } +const struct nvkm_intr_func +gt215_mc_intr = { + .pending = nv04_mc_intr_pending, + .unarm = nv04_mc_intr_unarm, + .rearm = nv04_mc_intr_rearm, + .block = gt215_mc_intr_block, + .allow = gt215_mc_intr_allow, +}; + static const struct nvkm_mc_func gt215_mc = { .init = nv50_mc_init, - .intr = gt215_mc_intr, - .intr_unarm = nv04_mc_intr_unarm, - .intr_rearm = nv04_mc_intr_rearm, - .intr_mask = gt215_mc_intr_mask, - .intr_stat = nv04_mc_intr_stat, + .intr = &nv04_mc_intr, + .intrs = gt215_mc_intrs, + .device = &nv04_mc_device, .reset = gt215_mc_reset, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv04.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv04.c index bc0d09bafa99..8482a5550e5f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv04.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv04.c @@ -30,37 +30,89 @@ nv04_mc_reset[] = { {} }; -static const struct nvkm_mc_map -nv04_mc_intr[] = { - { 0x01010000, NVKM_ENGINE_DISP }, - { 0x00001000, NVKM_ENGINE_GR }, - { 0x00000100, NVKM_ENGINE_FIFO }, - { 0x10000000, NVKM_SUBDEV_BUS }, - { 0x00100000, NVKM_SUBDEV_TIMER }, +static void +nv04_mc_device_disable(struct nvkm_mc *mc, u32 mask) +{ + nvkm_mask(mc->subdev.device, 0x000200, mask, 0x00000000); +} + +static void +nv04_mc_device_enable(struct nvkm_mc *mc, u32 mask) +{ + struct nvkm_device *device = mc->subdev.device; + + nvkm_mask(device, 0x000200, mask, mask); + nvkm_rd32(device, 0x000200); +} + +static bool +nv04_mc_device_enabled(struct nvkm_mc *mc, u32 mask) +{ + return (nvkm_rd32(mc->subdev.device, 0x000200) & mask) == mask; +} + +const struct nvkm_mc_device_func +nv04_mc_device = { + .enabled = nv04_mc_device_enabled, + .enable = nv04_mc_device_enable, + .disable = nv04_mc_device_disable, +}; + +static const struct nvkm_intr_data +nv04_mc_intrs[] = { + { NVKM_ENGINE_DISP , 0, 0, 0x01010000, true }, + { NVKM_ENGINE_GR , 0, 0, 0x00001000, true }, + { NVKM_ENGINE_FIFO , 0, 0, 0x00000100 }, + { NVKM_SUBDEV_BUS , 0, 0, 0x10000000, true }, + { NVKM_SUBDEV_TIMER, 0, 0, 0x00100000, true }, {} }; void -nv04_mc_intr_unarm(struct nvkm_mc *mc) +nv04_mc_intr_rearm(struct nvkm_intr *intr) { - struct nvkm_device *device = mc->subdev.device; - nvkm_wr32(device, 0x000140, 0x00000000); - nvkm_rd32(device, 0x000140); + struct nvkm_mc *mc = container_of(intr, typeof(*mc), intr); + int leaf; + + for (leaf = 0; leaf < intr->leaves; leaf++) + nvkm_wr32(mc->subdev.device, 0x000140 + (leaf * 4), 0x00000001); } void -nv04_mc_intr_rearm(struct nvkm_mc *mc) +nv04_mc_intr_unarm(struct nvkm_intr *intr) { - struct nvkm_device *device = mc->subdev.device; - nvkm_wr32(device, 0x000140, 0x00000001); + struct nvkm_mc *mc = container_of(intr, typeof(*mc), intr); + int leaf; + + for (leaf = 0; leaf < intr->leaves; leaf++) + nvkm_wr32(mc->subdev.device, 0x000140 + (leaf * 4), 0x00000000); + + nvkm_rd32(mc->subdev.device, 0x000140); } -u32 -nv04_mc_intr_stat(struct nvkm_mc *mc) +bool +nv04_mc_intr_pending(struct nvkm_intr *intr) { - return nvkm_rd32(mc->subdev.device, 0x000100); + struct nvkm_mc *mc = container_of(intr, typeof(*mc), intr); + bool pending = false; + int leaf; + + for (leaf = 0; leaf < intr->leaves; leaf++) { + intr->stat[leaf] = nvkm_rd32(mc->subdev.device, 0x000100 + (leaf * 4)); + if (intr->stat[leaf]) + pending = true; + } + + return pending; } +const struct nvkm_intr_func +nv04_mc_intr = { + .pending = nv04_mc_intr_pending, + .unarm = nv04_mc_intr_unarm, + .rearm = nv04_mc_intr_rearm, +}; + void nv04_mc_init(struct nvkm_mc *mc) { @@ -72,10 +124,9 @@ nv04_mc_init(struct nvkm_mc *mc) static const struct nvkm_mc_func nv04_mc = { .init = nv04_mc_init, - .intr = nv04_mc_intr, - .intr_unarm = nv04_mc_intr_unarm, - .intr_rearm = nv04_mc_intr_rearm, - .intr_stat = nv04_mc_intr_stat, + .intr = &nv04_mc_intr, + .intrs = nv04_mc_intrs, + .device = &nv04_mc_device, .reset = nv04_mc_reset, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv11.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv11.c index ab59ca1ee068..6d6278f434a4 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv11.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv11.c @@ -23,23 +23,22 @@ */ #include "priv.h" -static const struct nvkm_mc_map -nv11_mc_intr[] = { - { 0x03010000, NVKM_ENGINE_DISP }, - { 0x00001000, NVKM_ENGINE_GR }, - { 0x00000100, NVKM_ENGINE_FIFO }, - { 0x10000000, NVKM_SUBDEV_BUS }, - { 0x00100000, NVKM_SUBDEV_TIMER }, +static const struct nvkm_intr_data +nv11_mc_intrs[] = { + { NVKM_ENGINE_DISP , 0, 0, 0x03010000, true }, + { NVKM_ENGINE_GR , 0, 0, 0x00001000, true }, + { NVKM_ENGINE_FIFO , 0, 0, 0x00000100 }, + { NVKM_SUBDEV_BUS , 0, 0, 0x10000000, true }, + { NVKM_SUBDEV_TIMER, 0, 0, 0x00100000, true }, {} }; static const struct nvkm_mc_func nv11_mc = { .init = nv04_mc_init, - .intr = nv11_mc_intr, - .intr_unarm = nv04_mc_intr_unarm, - .intr_rearm = nv04_mc_intr_rearm, - .intr_stat = nv04_mc_intr_stat, + .intr = &nv04_mc_intr, + .intrs = nv11_mc_intrs, + .device = &nv04_mc_device, .reset = nv04_mc_reset, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv17.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv17.c index 03d756e26e57..dbad7c111ceb 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv17.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv17.c @@ -31,24 +31,23 @@ nv17_mc_reset[] = { {} }; -const struct nvkm_mc_map -nv17_mc_intr[] = { - { 0x03010000, NVKM_ENGINE_DISP }, - { 0x00001000, NVKM_ENGINE_GR }, - { 0x00000100, NVKM_ENGINE_FIFO }, - { 0x00000001, NVKM_ENGINE_MPEG }, - { 0x10000000, NVKM_SUBDEV_BUS }, - { 0x00100000, NVKM_SUBDEV_TIMER }, +const struct nvkm_intr_data +nv17_mc_intrs[] = { + { NVKM_ENGINE_DISP , 0, 0, 0x03010000, true }, + { NVKM_ENGINE_GR , 0, 0, 0x00001000, true }, + { NVKM_ENGINE_FIFO , 0, 0, 0x00000100 }, + { NVKM_ENGINE_MPEG , 0, 0, 0x00000001, true }, + { NVKM_SUBDEV_BUS , 0, 0, 0x10000000, true }, + { NVKM_SUBDEV_TIMER, 0, 0, 0x00100000, true }, {} }; static const struct nvkm_mc_func nv17_mc = { .init = nv04_mc_init, - .intr = nv17_mc_intr, - .intr_unarm = nv04_mc_intr_unarm, - .intr_rearm = nv04_mc_intr_rearm, - .intr_stat = nv04_mc_intr_stat, + .intr = &nv04_mc_intr, + .intrs = nv17_mc_intrs, + .device = &nv04_mc_device, .reset = nv17_mc_reset, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv44.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv44.c index 95f65766e8b0..649a9fcc0a2f 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv44.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv44.c @@ -40,10 +40,9 @@ nv44_mc_init(struct nvkm_mc *mc) static const struct nvkm_mc_func nv44_mc = { .init = nv44_mc_init, - .intr = nv17_mc_intr, - .intr_unarm = nv04_mc_intr_unarm, - .intr_rearm = nv04_mc_intr_rearm, - .intr_stat = nv04_mc_intr_stat, + .intr = &nv04_mc_intr, + .intrs = nv17_mc_intrs, + .device = &nv04_mc_device, .reset = nv17_mc_reset, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv50.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv50.c index fce3613cdfa5..d41099d35690 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv50.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/nv50.c @@ -23,17 +23,17 @@ */ #include "priv.h" -static const struct nvkm_mc_map -nv50_mc_intr[] = { - { 0x04000000, NVKM_ENGINE_DISP }, - { 0x00001000, NVKM_ENGINE_GR }, - { 0x00000100, NVKM_ENGINE_FIFO }, - { 0x00000001, NVKM_ENGINE_MPEG }, - { 0x00001101, NVKM_SUBDEV_FB }, - { 0x10000000, NVKM_SUBDEV_BUS }, - { 0x00200000, NVKM_SUBDEV_GPIO }, - { 0x00200000, NVKM_SUBDEV_I2C }, - { 0x00100000, NVKM_SUBDEV_TIMER }, +static const struct nvkm_intr_data +nv50_mc_intrs[] = { + { NVKM_ENGINE_DISP , 0, 0, 0x04000000, true }, + { NVKM_ENGINE_GR , 0, 0, 0x00001000, true }, + { NVKM_ENGINE_FIFO , 0, 0, 0x00000100 }, + { NVKM_ENGINE_MPEG , 0, 0, 0x00000001, true }, + { NVKM_SUBDEV_FB , 0, 0, 0x00001101, true }, + { NVKM_SUBDEV_BUS , 0, 0, 0x10000000, true }, + { NVKM_SUBDEV_GPIO , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_I2C , 0, 0, 0x00200000, true }, + { NVKM_SUBDEV_TIMER, 0, 0, 0x00100000, true }, {}, }; @@ -47,10 +47,9 @@ nv50_mc_init(struct nvkm_mc *mc) static const struct nvkm_mc_func nv50_mc = { .init = nv50_mc_init, - .intr = nv50_mc_intr, - .intr_unarm = nv04_mc_intr_unarm, - .intr_rearm = nv04_mc_intr_rearm, - .intr_stat = nv04_mc_intr_stat, + .intr = &nv04_mc_intr, + .intrs = nv50_mc_intrs, + .device = &nv04_mc_device, .reset = nv17_mc_reset, }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/priv.h b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/priv.h index c8bcabb98f99..7f38d54b4bc2 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/priv.h @@ -4,8 +4,6 @@ #define nvkm_mc(p) container_of((p), struct nvkm_mc, subdev) #include <subdev/mc.h> -void nvkm_mc_ctor(const struct nvkm_mc_func *, struct nvkm_device *, enum nvkm_subdev_type, int, - struct nvkm_mc *); int nvkm_mc_new_(const struct nvkm_mc_func *, struct nvkm_device *, enum nvkm_subdev_type, int, struct nvkm_mc **); @@ -18,46 +16,44 @@ struct nvkm_mc_map { struct nvkm_mc_func { void (*init)(struct nvkm_mc *); - const struct nvkm_mc_map *intr; - /* disable reporting of interrupts to host */ - void (*intr_unarm)(struct nvkm_mc *); - /* enable reporting of interrupts to host */ - void (*intr_rearm)(struct nvkm_mc *); - /* (un)mask delivery of specific interrupts */ - void (*intr_mask)(struct nvkm_mc *, u32 mask, u32 stat); - /* retrieve pending interrupt mask (NV_PMC_INTR) */ - u32 (*intr_stat)(struct nvkm_mc *); + + const struct nvkm_intr_func *intr; + const struct nvkm_intr_data *intrs; + bool intr_nonstall; + + const struct nvkm_mc_device_func { + bool (*enabled)(struct nvkm_mc *, u32 mask); + void (*enable)(struct nvkm_mc *, u32 mask); + void (*disable)(struct nvkm_mc *, u32 mask); + } *device; + const struct nvkm_mc_map *reset; + void (*unk260)(struct nvkm_mc *, u32); }; void nv04_mc_init(struct nvkm_mc *); -void nv04_mc_intr_unarm(struct nvkm_mc *); -void nv04_mc_intr_rearm(struct nvkm_mc *); -u32 nv04_mc_intr_stat(struct nvkm_mc *); +extern const struct nvkm_intr_func nv04_mc_intr; +bool nv04_mc_intr_pending(struct nvkm_intr *); +void nv04_mc_intr_unarm(struct nvkm_intr *); +void nv04_mc_intr_rearm(struct nvkm_intr *); +extern const struct nvkm_mc_device_func nv04_mc_device; extern const struct nvkm_mc_map nv04_mc_reset[]; -extern const struct nvkm_mc_map nv17_mc_intr[]; +extern const struct nvkm_intr_data nv17_mc_intrs[]; extern const struct nvkm_mc_map nv17_mc_reset[]; void nv44_mc_init(struct nvkm_mc *); void nv50_mc_init(struct nvkm_mc *); -void gk104_mc_init(struct nvkm_mc *); -void gf100_mc_intr_unarm(struct nvkm_mc *); -void gf100_mc_intr_rearm(struct nvkm_mc *); -void gf100_mc_intr_mask(struct nvkm_mc *, u32, u32); -u32 gf100_mc_intr_stat(struct nvkm_mc *); +extern const struct nvkm_intr_func gt215_mc_intr; void gf100_mc_unk260(struct nvkm_mc *, u32); -void gp100_mc_intr_unarm(struct nvkm_mc *); -void gp100_mc_intr_rearm(struct nvkm_mc *); -void gp100_mc_intr_mask(struct nvkm_mc *, u32, u32); -int gp100_mc_new_(const struct nvkm_mc_func *, struct nvkm_device *, enum nvkm_subdev_type, int, - struct nvkm_mc **); -extern const struct nvkm_mc_map gk104_mc_intr[]; +void gk104_mc_init(struct nvkm_mc *); +extern const struct nvkm_intr_data gk104_mc_intrs[]; extern const struct nvkm_mc_map gk104_mc_reset[]; -extern const struct nvkm_mc_map gp100_mc_intr[]; +extern const struct nvkm_intr_func gp100_mc_intr; +extern const struct nvkm_intr_data gp100_mc_intrs[]; #endif diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/tu102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mc/tu102.c deleted file mode 100644 index a96084b34a78..000000000000 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mc/tu102.c +++ /dev/null @@ -1,136 +0,0 @@ -/* - * Copyright 2018 Red Hat Inc. - * - * Permission is hereby granted, free of charge, to any person obtaining a - * copy of this software and associated documentation files (the "Software"), - * to deal in the Software without restriction, including without limitation - * the rights to use, copy, modify, merge, publish, distribute, sublicense, - * and/or sell copies of the Software, and to permit persons to whom the - * Software is furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in - * all copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL - * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR - * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, - * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR - * OTHER DEALINGS IN THE SOFTWARE. - */ -#define tu102_mc(p) container_of((p), struct tu102_mc, base) -#include "priv.h" - -struct tu102_mc { - struct nvkm_mc base; - spinlock_t lock; - bool intr; - u32 mask; -}; - -static void -tu102_mc_intr_update(struct tu102_mc *mc) -{ - struct nvkm_device *device = mc->base.subdev.device; - u32 mask = mc->intr ? mc->mask : 0, i; - - for (i = 0; i < 2; i++) { - nvkm_wr32(device, 0x000180 + (i * 0x04), ~mask); - nvkm_wr32(device, 0x000160 + (i * 0x04), mask); - } - - if (mask & 0x00000200) - nvkm_wr32(device, 0xb81608, 0x6); - else - nvkm_wr32(device, 0xb81610, 0x6); -} - -static void -tu102_mc_intr_unarm(struct nvkm_mc *base) -{ - struct tu102_mc *mc = tu102_mc(base); - unsigned long flags; - - spin_lock_irqsave(&mc->lock, flags); - mc->intr = false; - tu102_mc_intr_update(mc); - spin_unlock_irqrestore(&mc->lock, flags); -} - -static void -tu102_mc_intr_rearm(struct nvkm_mc *base) -{ - struct tu102_mc *mc = tu102_mc(base); - unsigned long flags; - - spin_lock_irqsave(&mc->lock, flags); - mc->intr = true; - tu102_mc_intr_update(mc); - spin_unlock_irqrestore(&mc->lock, flags); -} - -static void -tu102_mc_intr_mask(struct nvkm_mc *base, u32 mask, u32 intr) -{ - struct tu102_mc *mc = tu102_mc(base); - unsigned long flags; - - spin_lock_irqsave(&mc->lock, flags); - mc->mask = (mc->mask & ~mask) | intr; - tu102_mc_intr_update(mc); - spin_unlock_irqrestore(&mc->lock, flags); -} - -static u32 -tu102_mc_intr_stat(struct nvkm_mc *mc) -{ - struct nvkm_device *device = mc->subdev.device; - u32 intr0 = nvkm_rd32(device, 0x000100); - u32 intr1 = nvkm_rd32(device, 0x000104); - u32 intr_top = nvkm_rd32(device, 0xb81600); - - /* Turing and above route the MMU fault interrupts via a different - * interrupt tree with different control registers. For the moment remap - * them back to the old PMC vector. - */ - if (intr_top & 0x00000006) - intr0 |= 0x00000200; - - return intr0 | intr1; -} - - -static const struct nvkm_mc_func -tu102_mc = { - .init = nv50_mc_init, - .intr = gp100_mc_intr, - .intr_unarm = tu102_mc_intr_unarm, - .intr_rearm = tu102_mc_intr_rearm, - .intr_mask = tu102_mc_intr_mask, - .intr_stat = tu102_mc_intr_stat, - .reset = gk104_mc_reset, -}; - -static int -tu102_mc_new_(const struct nvkm_mc_func *func, struct nvkm_device *device, - enum nvkm_subdev_type type, int inst, struct nvkm_mc **pmc) -{ - struct tu102_mc *mc; - - if (!(mc = kzalloc(sizeof(*mc), GFP_KERNEL))) - return -ENOMEM; - nvkm_mc_ctor(func, device, type, inst, &mc->base); - *pmc = &mc->base; - - spin_lock_init(&mc->lock); - mc->intr = false; - mc->mask = 0x7fffffff; - return 0; -} - -int -tu102_mc_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_mc **pmc) -{ - return tu102_mc_new_(&tu102_mc, device, type, inst, pmc); -} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/uvmm.c b/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/uvmm.c index 186b4e63e559..524cd3c0e3fe 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/uvmm.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/mmu/uvmm.c @@ -39,7 +39,7 @@ nvkm_uvmm_search(struct nvkm_client *client, u64 handle) if (IS_ERR(object)) return (void *)object; - return nvkm_uvmm(object)->vmm; + return nvkm_vmm_ref(nvkm_uvmm(object)->vmm); } static int diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pci/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/pci/base.c index a7d42ea8ba28..5a0de45d36ce 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pci/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pci/base.c @@ -26,7 +26,15 @@ #include <core/option.h> #include <core/pci.h> -#include <subdev/mc.h> + +void +nvkm_pci_msi_rearm(struct nvkm_device *device) +{ + struct nvkm_pci *pci = device->pci; + + if (pci && pci->msi) + pci->func->msi_rearm(pci); +} u32 nvkm_pci_rd32(struct nvkm_pci *pci, u16 addr) @@ -65,24 +73,6 @@ nvkm_pci_rom_shadow(struct nvkm_pci *pci, bool shadow) nvkm_pci_wr32(pci, 0x0050, data); } -static irqreturn_t -nvkm_pci_intr(int irq, void *arg) -{ - struct nvkm_pci *pci = arg; - struct nvkm_device *device = pci->subdev.device; - bool handled = false; - - if (pci->irq < 0) - return IRQ_HANDLED; - - nvkm_mc_intr_unarm(device); - if (pci->msi) - pci->func->msi_rearm(pci); - nvkm_mc_intr(device, &handled); - nvkm_mc_intr_rearm(device); - return handled ? IRQ_HANDLED : IRQ_NONE; -} - static int nvkm_pci_fini(struct nvkm_subdev *subdev, bool suspend) { @@ -107,7 +97,6 @@ static int nvkm_pci_oneinit(struct nvkm_subdev *subdev) { struct nvkm_pci *pci = nvkm_pci(subdev); - struct pci_dev *pdev = pci->pdev; int ret; if (pci_is_pcie(pci->pdev)) { @@ -116,11 +105,6 @@ nvkm_pci_oneinit(struct nvkm_subdev *subdev) return ret; } - ret = request_irq(pdev->irq, nvkm_pci_intr, IRQF_SHARED, "nvkm", pci); - if (ret) - return ret; - - pci->irq = pdev->irq; return 0; } @@ -157,15 +141,6 @@ nvkm_pci_dtor(struct nvkm_subdev *subdev) nvkm_agp_dtor(pci); - if (pci->irq >= 0) { - /* freq_irq() will call the handler, we use pci->irq == -1 - * to signal that it's been torn down and should be a noop. - */ - int irq = pci->irq; - pci->irq = -1; - free_irq(irq, pci); - } - if (pci->msi) pci_disable_msi(pci->pdev); @@ -192,7 +167,6 @@ nvkm_pci_new_(const struct nvkm_pci_func *func, struct nvkm_device *device, nvkm_subdev_ctor(&nvkm_pci_func, device, type, inst, &pci->subdev); pci->func = func; pci->pdev = device->func->pci(device)->pdev; - pci->irq = -1; pci->pcie.speed = -1; pci->pcie.width = -1; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/base.c index 455e95a89259..8f2f50ad4ded 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/base.c @@ -81,43 +81,12 @@ nvkm_pmu_fini(struct nvkm_subdev *subdev, bool suspend) { struct nvkm_pmu *pmu = nvkm_pmu(subdev); + if (!subdev->use.enabled) + return 0; + if (pmu->func->fini) pmu->func->fini(pmu); - flush_work(&pmu->recv.work); - - reinit_completion(&pmu->wpr_ready); - - nvkm_falcon_cmdq_fini(pmu->lpq); - nvkm_falcon_cmdq_fini(pmu->hpq); - pmu->initmsg_received = false; - return 0; -} - -static void -nvkm_pmu_reset(struct nvkm_pmu *pmu) -{ - struct nvkm_device *device = pmu->subdev.device; - - if (!pmu->func->enabled(pmu)) - return; - - /* Reset. */ - if (pmu->func->reset) - pmu->func->reset(pmu); - - /* Wait for IMEM/DMEM scrubbing to be complete. */ - nvkm_msec(device, 2000, - if (!(nvkm_rd32(device, 0x10a10c) & 0x00000006)) - break; - ); -} - -static int -nvkm_pmu_preinit(struct nvkm_subdev *subdev) -{ - struct nvkm_pmu *pmu = nvkm_pmu(subdev); - nvkm_pmu_reset(pmu); return 0; } @@ -125,22 +94,10 @@ static int nvkm_pmu_init(struct nvkm_subdev *subdev) { struct nvkm_pmu *pmu = nvkm_pmu(subdev); - struct nvkm_device *device = pmu->subdev.device; if (!pmu->func->init) return 0; - if (pmu->func->enabled(pmu)) { - /* Inhibit interrupts, and wait for idle. */ - nvkm_wr32(device, 0x10a014, 0x0000ffff); - nvkm_msec(device, 2000, - if (!nvkm_rd32(device, 0x10a04c)) - break; - ); - - nvkm_pmu_reset(pmu); - } - return pmu->func->init(pmu); } @@ -160,7 +117,6 @@ nvkm_pmu_dtor(struct nvkm_subdev *subdev) static const struct nvkm_subdev_func nvkm_pmu = { .dtor = nvkm_pmu_dtor, - .preinit = nvkm_pmu_preinit, .init = nvkm_pmu_init, .fini = nvkm_pmu_fini, .intr = nvkm_pmu_intr, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gk20a.c b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gk20a.c index a67a42e73f08..b5e52b35f5d0 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gk20a.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gk20a.c @@ -197,7 +197,6 @@ gk20a_dvfs_data= { static const struct nvkm_pmu_func gk20a_pmu = { .flcn = >215_pmu_flcn, - .enabled = gf100_pmu_enabled, .init = gk20a_pmu_init, .fini = gk20a_pmu_fini, .reset = gf100_pmu_reset, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gm200.c b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gm200.c index 40439e329aa9..7359991f94c2 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gm200.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gm200.c @@ -24,30 +24,36 @@ #include "priv.h" static int -gm200_pmu_flcn_reset(struct nvkm_falcon *falcon) +gm200_pmu_flcn_bind_stat(struct nvkm_falcon *falcon, bool intr) { - struct nvkm_pmu *pmu = container_of(falcon, typeof(*pmu), falcon); + nvkm_falcon_wr32(falcon, 0x200, 0x0000030e); + return (nvkm_falcon_rd32(falcon, 0x20c) & 0x00007000) >> 12; +} - nvkm_falcon_wr32(falcon, 0x014, 0x0000ffff); - pmu->func->reset(pmu); - return nvkm_falcon_enable(falcon); +void +gm200_pmu_flcn_bind_inst(struct nvkm_falcon *falcon, int target, u64 addr) +{ + nvkm_falcon_wr32(falcon, 0xe00, 4); /* DMAIDX_UCODE */ + nvkm_falcon_wr32(falcon, 0xe04, 0); /* DMAIDX_VIRT */ + nvkm_falcon_wr32(falcon, 0xe08, 4); /* DMAIDX_PHYS_VID */ + nvkm_falcon_wr32(falcon, 0xe0c, 5); /* DMAIDX_PHYS_SYS_COH */ + nvkm_falcon_wr32(falcon, 0xe10, 6); /* DMAIDX_PHYS_SYS_NCOH */ + nvkm_falcon_mask(falcon, 0x090, 0x00010000, 0x00010000); + nvkm_falcon_wr32(falcon, 0x480, (1 << 30) | (target << 28) | (addr >> 12)); } const struct nvkm_falcon_func gm200_pmu_flcn = { + .disable = gm200_flcn_disable, + .enable = gm200_flcn_enable, + .reset_pmc = true, + .reset_wait_mem_scrubbing = gm200_flcn_reset_wait_mem_scrubbing, .debug = 0xc08, - .fbif = 0xe00, - .load_imem = nvkm_falcon_v1_load_imem, - .load_dmem = nvkm_falcon_v1_load_dmem, - .read_dmem = nvkm_falcon_v1_read_dmem, - .bind_context = nvkm_falcon_v1_bind_context, - .wait_for_halt = nvkm_falcon_v1_wait_for_halt, - .clear_interrupt = nvkm_falcon_v1_clear_interrupt, - .set_start_addr = nvkm_falcon_v1_set_start_addr, + .bind_inst = gm200_pmu_flcn_bind_inst, + .bind_stat = gm200_pmu_flcn_bind_stat, + .imem_pio = &gm200_flcn_imem_pio, + .dmem_pio = &gm200_flcn_dmem_pio, .start = nvkm_falcon_v1_start, - .enable = nvkm_falcon_v1_enable, - .disable = nvkm_falcon_v1_disable, - .reset = gm200_pmu_flcn_reset, .cmdq = { 0x4a0, 0x4b0, 4 }, .msgq = { 0x4c8, 0x4cc, 0 }, }; @@ -55,11 +61,9 @@ gm200_pmu_flcn = { static const struct nvkm_pmu_func gm200_pmu = { .flcn = &gm200_pmu_flcn, - .enabled = gf100_pmu_enabled, .reset = gf100_pmu_reset, }; - int gm200_pmu_nofw(struct nvkm_pmu *pmu, int ver, const struct nvkm_pmu_fwif *fwif) { diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gm20b.c b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gm20b.c index 612310d5d481..a72403777329 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gm20b.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gm20b.c @@ -62,16 +62,6 @@ gm20b_pmu_acr_bootstrap_falcon(struct nvkm_falcon *falcon, return ret; } -int -gm20b_pmu_acr_boot(struct nvkm_falcon *falcon) -{ - struct nv_pmu_args args = { .secure_mode = true }; - const u32 addr_args = falcon->data.limit - sizeof(struct nv_pmu_args); - nvkm_falcon_load_dmem(falcon, &args, addr_args, sizeof(args), 0); - nvkm_falcon_start(falcon); - return 0; -} - void gm20b_pmu_acr_bld_patch(struct nvkm_acr *acr, u32 bld, s64 adjust) { @@ -125,7 +115,6 @@ gm20b_pmu_acr = { .bld_size = sizeof(struct loader_config), .bld_write = gm20b_pmu_acr_bld_write, .bld_patch = gm20b_pmu_acr_bld_patch, - .boot = gm20b_pmu_acr_boot, .bootstrap_falcons = BIT_ULL(NVKM_ACR_LSF_PMU) | BIT_ULL(NVKM_ACR_LSF_FECS) | BIT_ULL(NVKM_ACR_LSF_GPCCS), @@ -166,7 +155,7 @@ gm20b_pmu_acr_init_wpr(struct nvkm_pmu *pmu) gm20b_pmu_acr_init_wpr_callback, pmu, 0); } -int +static int gm20b_pmu_initmsg(struct nvkm_pmu *pmu) { struct nv_pmu_init_msg msg; @@ -192,14 +181,13 @@ gm20b_pmu_initmsg(struct nvkm_pmu *pmu) return gm20b_pmu_acr_init_wpr(pmu); } -void +static void gm20b_pmu_recv(struct nvkm_pmu *pmu) { if (!pmu->initmsg_received) { int ret = pmu->func->initmsg(pmu); if (ret) { - nvkm_error(&pmu->subdev, - "error parsing init message: %d\n", ret); + nvkm_error(&pmu->subdev, "error parsing init message: %d\n", ret); return; } @@ -209,10 +197,44 @@ gm20b_pmu_recv(struct nvkm_pmu *pmu) nvkm_falcon_msgq_recv(pmu->msgq); } -static const struct nvkm_pmu_func +static void +gm20b_pmu_fini(struct nvkm_pmu *pmu) +{ + /*TODO: shutdown RTOS. */ + + flush_work(&pmu->recv.work); + nvkm_falcon_cmdq_fini(pmu->lpq); + nvkm_falcon_cmdq_fini(pmu->hpq); + + reinit_completion(&pmu->wpr_ready); + + nvkm_falcon_put(&pmu->falcon, &pmu->subdev); +} + +static int +gm20b_pmu_init(struct nvkm_pmu *pmu) +{ + struct nvkm_falcon *falcon = &pmu->falcon; + struct nv_pmu_args args = { .secure_mode = true }; + u32 addr_args = falcon->data.limit - sizeof(args); + int ret; + + ret = nvkm_falcon_get(&pmu->falcon, &pmu->subdev); + if (ret) + return ret; + + pmu->initmsg_received = false; + + nvkm_falcon_load_dmem(falcon, &args, addr_args, sizeof(args), 0); + nvkm_falcon_start(falcon); + return 0; +} + +const struct nvkm_pmu_func gm20b_pmu = { .flcn = &gm200_pmu_flcn, - .enabled = gf100_pmu_enabled, + .init = gm20b_pmu_init, + .fini = gm20b_pmu_fini, .intr = gt215_pmu_intr, .recv = gm20b_pmu_recv, .initmsg = gm20b_pmu_initmsg, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gp102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gp102.c index 1a6f9c3af5ec..cd3148360996 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gp102.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gp102.c @@ -23,25 +23,25 @@ */ #include "priv.h" -void -gp102_pmu_reset(struct nvkm_pmu *pmu) -{ - struct nvkm_device *device = pmu->subdev.device; - nvkm_mask(device, 0x10a3c0, 0x00000001, 0x00000001); - nvkm_mask(device, 0x10a3c0, 0x00000001, 0x00000000); -} - -static bool -gp102_pmu_enabled(struct nvkm_pmu *pmu) -{ - return !(nvkm_rd32(pmu->subdev.device, 0x10a3c0) & 0x00000001); -} +static const struct nvkm_falcon_func +gp102_pmu_flcn = { + .disable = gm200_flcn_disable, + .enable = gm200_flcn_enable, + .reset_eng = gp102_flcn_reset_eng, + .reset_wait_mem_scrubbing = gm200_flcn_reset_wait_mem_scrubbing, + .debug = 0xc08, + .bind_inst = gm200_pmu_flcn_bind_inst, + .bind_stat = gm200_flcn_bind_stat, + .imem_pio = &gm200_flcn_imem_pio, + .dmem_pio = &gm200_flcn_dmem_pio, + .start = nvkm_falcon_v1_start, + .cmdq = { 0x4a0, 0x4b0, 4 }, + .msgq = { 0x4c8, 0x4cc, 0 }, +}; static const struct nvkm_pmu_func gp102_pmu = { - .flcn = &gm200_pmu_flcn, - .enabled = gp102_pmu_enabled, - .reset = gp102_pmu_reset, + .flcn = &gp102_pmu_flcn, }; static const struct nvkm_pmu_fwif diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gp10b.c b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gp10b.c index 94cfb1791af6..a6f410ba60bc 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gp10b.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gp10b.c @@ -68,7 +68,6 @@ gp10b_pmu_acr = { .bld_size = sizeof(struct loader_config), .bld_write = gm20b_pmu_acr_bld_write, .bld_patch = gm20b_pmu_acr_bld_patch, - .boot = gm20b_pmu_acr_boot, .bootstrap_falcons = BIT_ULL(NVKM_ACR_LSF_PMU) | BIT_ULL(NVKM_ACR_LSF_FECS) | BIT_ULL(NVKM_ACR_LSF_GPCCS), @@ -76,16 +75,6 @@ gp10b_pmu_acr = { .bootstrap_multiple_falcons = gp10b_pmu_acr_bootstrap_multiple_falcons, }; -static const struct nvkm_pmu_func -gp10b_pmu = { - .flcn = &gm200_pmu_flcn, - .enabled = gf100_pmu_enabled, - .intr = gt215_pmu_intr, - .recv = gm20b_pmu_recv, - .initmsg = gm20b_pmu_initmsg, - .reset = gp102_pmu_reset, -}; - #if IS_ENABLED(CONFIG_ARCH_TEGRA_210_SOC) MODULE_FIRMWARE("nvidia/gp10b/pmu/desc.bin"); MODULE_FIRMWARE("nvidia/gp10b/pmu/image.bin"); @@ -94,8 +83,8 @@ MODULE_FIRMWARE("nvidia/gp10b/pmu/sig.bin"); static const struct nvkm_pmu_fwif gp10b_pmu_fwif[] = { - { 0, gm20b_pmu_load, &gp10b_pmu, &gp10b_pmu_acr }, - { -1, gm200_pmu_nofw, &gp10b_pmu }, + { 0, gm20b_pmu_load, &gm20b_pmu, &gp10b_pmu_acr }, + { -1, gm200_pmu_nofw, &gm20b_pmu }, {} }; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gt215.c b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gt215.c index b0407b86bc10..32cee21ed858 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gt215.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/gt215.c @@ -178,12 +178,14 @@ void gt215_pmu_fini(struct nvkm_pmu *pmu) { nvkm_wr32(pmu->subdev.device, 0x10a014, 0x00000060); + flush_work(&pmu->recv.work); } static void gt215_pmu_reset(struct nvkm_pmu *pmu) { struct nvkm_device *device = pmu->subdev.device; + nvkm_mask(device, 0x022210, 0x00000001, 0x00000000); nvkm_mask(device, 0x022210, 0x00000001, 0x00000001); nvkm_rd32(device, 0x022210); @@ -201,6 +203,23 @@ gt215_pmu_init(struct nvkm_pmu *pmu) struct nvkm_device *device = pmu->subdev.device; int i; + /* Inhibit interrupts, and wait for idle. */ + if (pmu->func->enabled(pmu)) { + nvkm_wr32(device, 0x10a014, 0x0000ffff); + nvkm_msec(device, 2000, + if (!nvkm_rd32(device, 0x10a04c)) + break; + ); + } + + pmu->func->reset(pmu); + + /* Wait for IMEM/DMEM scrubbing to be complete. */ + nvkm_msec(device, 2000, + if (!(nvkm_rd32(device, 0x10a10c) & 0x00000006)) + break; + ); + /* upload data segment */ nvkm_wr32(device, 0x10a1c0, 0x01000000); for (i = 0; i < pmu->func->data.size / 4; i++) @@ -243,20 +262,6 @@ gt215_pmu_init(struct nvkm_pmu *pmu) const struct nvkm_falcon_func gt215_pmu_flcn = { - .debug = 0xc08, - .fbif = 0xe00, - .load_imem = nvkm_falcon_v1_load_imem, - .load_dmem = nvkm_falcon_v1_load_dmem, - .read_dmem = nvkm_falcon_v1_read_dmem, - .bind_context = nvkm_falcon_v1_bind_context, - .wait_for_halt = nvkm_falcon_v1_wait_for_halt, - .clear_interrupt = nvkm_falcon_v1_clear_interrupt, - .set_start_addr = nvkm_falcon_v1_set_start_addr, - .start = nvkm_falcon_v1_start, - .enable = nvkm_falcon_v1_enable, - .disable = nvkm_falcon_v1_disable, - .cmdq = { 0x4a0, 0x4b0, 4 }, - .msgq = { 0x4c8, 0x4cc, 0 }, }; static const struct nvkm_pmu_func diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/priv.h b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/priv.h index 21abf31f4442..2d0a8fa6f196 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/pmu/priv.h @@ -46,13 +46,12 @@ void gp102_pmu_reset(struct nvkm_pmu *pmu); void gk110_pmu_pgob(struct nvkm_pmu *, bool); extern const struct nvkm_falcon_func gm200_pmu_flcn; +void gm200_pmu_flcn_bind_inst(struct nvkm_falcon *, int, u64); +extern const struct nvkm_pmu_func gm20b_pmu; void gm20b_pmu_acr_bld_patch(struct nvkm_acr *, u32, s64); void gm20b_pmu_acr_bld_write(struct nvkm_acr *, u32, struct nvkm_acr_lsfw *); -int gm20b_pmu_acr_boot(struct nvkm_falcon *); int gm20b_pmu_acr_bootstrap_falcon(struct nvkm_falcon *, enum nvkm_acr_lsf_id); -void gm20b_pmu_recv(struct nvkm_pmu *); -int gm20b_pmu_initmsg(struct nvkm_pmu *); struct nvkm_pmu_fwif { int version; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/top/base.c b/drivers/gpu/drm/nouveau/nvkm/subdev/top/base.c index 28d0789f50fe..eb348dfc1d7a 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/top/base.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/top/base.c @@ -117,11 +117,15 @@ nvkm_top_fault(struct nvkm_device *device, int fault) return NULL; } -static int -nvkm_top_oneinit(struct nvkm_subdev *subdev) +int +nvkm_top_parse(struct nvkm_device *device) { - struct nvkm_top *top = nvkm_top(subdev); - return top->func->oneinit(top); + struct nvkm_top *top = device->top; + + if (!top || !list_empty(&top->device)) + return 0; + + return top->func->parse(top); } static void * @@ -141,7 +145,6 @@ nvkm_top_dtor(struct nvkm_subdev *subdev) static const struct nvkm_subdev_func nvkm_top = { .dtor = nvkm_top_dtor, - .oneinit = nvkm_top_oneinit, }; int diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/top/ga100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/top/ga100.c index c982d834c8d9..84790cf52b90 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/top/ga100.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/top/ga100.c @@ -22,7 +22,7 @@ #include "priv.h" static int -ga100_top_oneinit(struct nvkm_top *top) +ga100_top_parse(struct nvkm_top *top) { struct nvkm_subdev *subdev = &top->subdev; struct nvkm_device *device = subdev->device; @@ -97,7 +97,7 @@ ga100_top_oneinit(struct nvkm_top *top) static const struct nvkm_top_func ga100_top = { - .oneinit = ga100_top_oneinit, + .parse = ga100_top_parse, }; int diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/top/gk104.c b/drivers/gpu/drm/nouveau/nvkm/subdev/top/gk104.c index 4dcad97bd505..2bbba8244cbf 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/top/gk104.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/top/gk104.c @@ -24,7 +24,7 @@ #include "priv.h" static int -gk104_top_oneinit(struct nvkm_top *top) +gk104_top_parse(struct nvkm_top *top) { struct nvkm_subdev *subdev = &top->subdev; struct nvkm_device *device = subdev->device; @@ -108,7 +108,7 @@ gk104_top_oneinit(struct nvkm_top *top) static const struct nvkm_top_func gk104_top = { - .oneinit = gk104_top_oneinit, + .parse = gk104_top_parse, }; int diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/top/priv.h b/drivers/gpu/drm/nouveau/nvkm/subdev/top/priv.h index 8e103a836705..532be91d8fd9 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/top/priv.h +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/top/priv.h @@ -5,7 +5,7 @@ #include <subdev/top.h> struct nvkm_top_func { - int (*oneinit)(struct nvkm_top *); + int (*parse)(struct nvkm_top *); }; int nvkm_top_new_(const struct nvkm_top_func *, struct nvkm_device *, enum nvkm_subdev_type, int, diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/Kbuild b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/Kbuild new file mode 100644 index 000000000000..23cd21b40a25 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/Kbuild @@ -0,0 +1,6 @@ +# SPDX-License-Identifier: MIT +nvkm-y += nvkm/subdev/vfn/base.o +nvkm-y += nvkm/subdev/vfn/uvfn.o +nvkm-y += nvkm/subdev/vfn/gv100.o +nvkm-y += nvkm/subdev/vfn/tu102.o +nvkm-y += nvkm/subdev/vfn/ga100.o diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/disp/rootnv04.c b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/base.c index 9acaec5c271e..62e81d551f44 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/disp/rootnv04.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/base.c @@ -1,5 +1,5 @@ /* - * Copyright 2012 Red Hat Inc. + * Copyright 2021 Red Hat Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), @@ -18,45 +18,43 @@ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. - * - * Authors: Ben Skeggs */ #include "priv.h" -#include "head.h" -#include <core/client.h> +static void * +nvkm_vfn_dtor(struct nvkm_subdev *subdev) +{ + return nvkm_vfn(subdev); +} -#include <nvif/cl0046.h> -#include <nvif/unpack.h> +static const struct nvkm_subdev_func +nvkm_vfn = { + .dtor = nvkm_vfn_dtor, +}; int -nv04_disp_mthd(struct nvkm_object *object, u32 mthd, void *data, u32 size) +nvkm_vfn_new_(const struct nvkm_vfn_func *func, struct nvkm_device *device, + enum nvkm_subdev_type type, int inst, u32 addr, struct nvkm_vfn **pvfn) { - struct nvkm_disp *disp = nvkm_disp(object->engine); - union { - struct nv04_disp_mthd_v0 v0; - } *args = data; - struct nvkm_head *head; - int id, ret = -ENOSYS; + struct nvkm_vfn *vfn; + int ret; - nvif_ioctl(object, "disp mthd size %d\n", size); - if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, true))) { - nvif_ioctl(object, "disp mthd vers %d mthd %02x head %d\n", - args->v0.version, args->v0.method, args->v0.head); - mthd = args->v0.method; - id = args->v0.head; - } else - return ret; + if (!(vfn = *pvfn = kzalloc(sizeof(*vfn), GFP_KERNEL))) + return -ENOMEM; - if (!(head = nvkm_head_find(disp, id))) - return -ENXIO; + nvkm_subdev_ctor(&nvkm_vfn, device, type, inst, &vfn->subdev); + vfn->func = func; + vfn->addr.priv = addr; + vfn->addr.user = vfn->addr.priv + func->user.addr; - switch (mthd) { - case NV04_DISP_SCANOUTPOS: - return nvkm_head_mthd_scanoutpos(object, head, data, size); - default: - break; + if (vfn->func->intr) { + ret = nvkm_intr_add(vfn->func->intr, vfn->func->intrs, + &vfn->subdev, 8, &vfn->intr); + if (ret) + return ret; } - return -EINVAL; + vfn->user.ctor = nvkm_uvfn_new; + vfn->user.base = func->user.base; + return 0; } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/ga100.c b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/ga100.c new file mode 100644 index 000000000000..fd5c6931322d --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/ga100.c @@ -0,0 +1,47 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +#include <nvif/class.h> + +static const struct nvkm_intr_data +ga100_vfn_intrs[] = { + { NVKM_ENGINE_DISP , 0, 4, 0x04000000, true }, + { NVKM_SUBDEV_GPIO , 0, 4, 0x00200000, true }, + { NVKM_SUBDEV_I2C , 0, 4, 0x00200000, true }, + { NVKM_SUBDEV_PRIVRING, 0, 4, 0x40000000, true }, + {} +}; + +static const struct nvkm_vfn_func +ga100_vfn = { + .intr = &tu102_vfn_intr, + .intrs = ga100_vfn_intrs, + .user = { 0x030000, 0x010000, { -1, -1, AMPERE_USERMODE_A } }, +}; + +int +ga100_vfn_new(struct nvkm_device *device, + enum nvkm_subdev_type type, int inst, struct nvkm_vfn **pvfn) +{ + return nvkm_vfn_new_(&ga100_vfn, device, type, inst, 0xb80000, pvfn); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/usertu102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/gv100.c index 217268f8ccad..ddd39d714c4a 100644 --- a/drivers/gpu/drm/nouveau/nvkm/engine/fifo/usertu102.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/gv100.c @@ -1,5 +1,5 @@ /* - * Copyright 2018 Red Hat Inc. + * Copyright 2021 Red Hat Inc. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), @@ -19,27 +19,18 @@ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR * OTHER DEALINGS IN THE SOFTWARE. */ -#include "user.h" +#include "priv.h" -static int -tu102_fifo_user_map(struct nvkm_object *object, void *argv, u32 argc, - enum nvkm_object_map *type, u64 *addr, u64 *size) -{ - struct nvkm_device *device = object->engine->subdev.device; - *addr = 0xbb0000 + device->func->resource_addr(device, 0); - *size = 0x010000; - *type = NVKM_OBJECT_MAP_IO; - return 0; -} +#include <nvif/class.h> -static const struct nvkm_object_func -tu102_fifo_user = { - .map = tu102_fifo_user_map, +static const struct nvkm_vfn_func +gv100_vfn = { + .user = { 0x810000, 0x010000, { -1, -1, VOLTA_USERMODE_A } }, }; int -tu102_fifo_user_new(const struct nvkm_oclass *oclass, void *argv, u32 argc, - struct nvkm_object **pobject) +gv100_vfn_new(struct nvkm_device *device, + enum nvkm_subdev_type type, int inst, struct nvkm_vfn **pvfn) { - return nvkm_object_new_(&tu102_fifo_user, oclass, argv, argc, pobject); + return nvkm_vfn_new_(&gv100_vfn, device, type, inst, 0, pvfn); } diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/priv.h b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/priv.h new file mode 100644 index 000000000000..96d53c02041b --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/priv.h @@ -0,0 +1,25 @@ +/* SPDX-License-Identifier: MIT */ +#ifndef __NVKM_VFN_PRIV_H__ +#define __NVKM_VFN_PRIV_H__ +#define nvkm_vfn(p) container_of((p), struct nvkm_vfn, subdev) +#include <subdev/vfn.h> + +struct nvkm_vfn_func { + const struct nvkm_intr_func *intr; + const struct nvkm_intr_data *intrs; + + struct { + u32 addr; + u32 size; + const struct nvkm_sclass base; + } user; +}; + +int nvkm_vfn_new_(const struct nvkm_vfn_func *, struct nvkm_device *, enum nvkm_subdev_type, int, + u32 addr, struct nvkm_vfn **); + +extern const struct nvkm_intr_func tu102_vfn_intr; + +int nvkm_uvfn_new(struct nvkm_device *, const struct nvkm_oclass *, void *, u32, + struct nvkm_object **); +#endif diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/tu102.c b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/tu102.c new file mode 100644 index 000000000000..3d063fb5e136 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/tu102.c @@ -0,0 +1,108 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#include "priv.h" + +#include <nvif/class.h> + +static void +tu102_vfn_intr_reset(struct nvkm_intr *intr, int leaf, u32 mask) +{ + struct nvkm_vfn *vfn = container_of(intr, typeof(*vfn), intr); + + nvkm_wr32(vfn->subdev.device, vfn->addr.priv + 0x1000 + (leaf * 4), mask); +} + +static void +tu102_vfn_intr_allow(struct nvkm_intr *intr, int leaf, u32 mask) +{ + struct nvkm_vfn *vfn = container_of(intr, typeof(*vfn), intr); + + nvkm_wr32(vfn->subdev.device, vfn->addr.priv + 0x1200 + (leaf * 4), mask); +} + +static void +tu102_vfn_intr_block(struct nvkm_intr *intr, int leaf, u32 mask) +{ + struct nvkm_vfn *vfn = container_of(intr, typeof(*vfn), intr); + + nvkm_wr32(vfn->subdev.device, vfn->addr.priv + 0x1400 + (leaf * 4), mask); +} + +static void +tu102_vfn_intr_rearm(struct nvkm_intr *intr) +{ + struct nvkm_vfn *vfn = container_of(intr, typeof(*vfn), intr); + + nvkm_wr32(vfn->subdev.device, vfn->addr.priv + 0x1608, 0x0000000f); +} + +static void +tu102_vfn_intr_unarm(struct nvkm_intr *intr) +{ + struct nvkm_vfn *vfn = container_of(intr, typeof(*vfn), intr); + + nvkm_wr32(vfn->subdev.device, vfn->addr.priv + 0x1610, 0x0000000f); +} + +static bool +tu102_vfn_intr_pending(struct nvkm_intr *intr) +{ + struct nvkm_vfn *vfn = container_of(intr, typeof(*vfn), intr); + struct nvkm_device *device = vfn->subdev.device; + u32 intr_top = nvkm_rd32(device, vfn->addr.priv + 0x1600); + int pending = 0, leaf; + + for (leaf = 0; leaf < 8; leaf++) { + if (intr_top & BIT(leaf / 2)) { + intr->stat[leaf] = nvkm_rd32(device, vfn->addr.priv + 0x1000 + (leaf * 4)); + if (intr->stat[leaf]) + pending++; + } else { + intr->stat[leaf] = 0; + } + } + + return pending != 0; +} + +const struct nvkm_intr_func +tu102_vfn_intr = { + .pending = tu102_vfn_intr_pending, + .unarm = tu102_vfn_intr_unarm, + .rearm = tu102_vfn_intr_rearm, + .block = tu102_vfn_intr_block, + .allow = tu102_vfn_intr_allow, + .reset = tu102_vfn_intr_reset, +}; + +static const struct nvkm_vfn_func +tu102_vfn = { + .intr = &tu102_vfn_intr, + .user = { 0x030000, 0x010000, { -1, -1, TURING_USERMODE_A } }, +}; + +int +tu102_vfn_new(struct nvkm_device *device, + enum nvkm_subdev_type type, int inst, struct nvkm_vfn **pvfn) +{ + return nvkm_vfn_new_(&tu102_vfn, device, type, inst, 0xb80000, pvfn); +} diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/uvfn.c b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/uvfn.c new file mode 100644 index 000000000000..c5460a14c541 --- /dev/null +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/uvfn.c @@ -0,0 +1,67 @@ +/* + * Copyright 2021 Red Hat Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ +#define nvkm_uvfn(p) container_of((p), struct nvkm_uvfn, object) +#include "priv.h" + +#include <core/object.h> + +struct nvkm_uvfn { + struct nvkm_object object; + struct nvkm_vfn *vfn; +}; + +static int +nvkm_uvfn_map(struct nvkm_object *object, void *argv, u32 argc, + enum nvkm_object_map *type, u64 *addr, u64 *size) +{ + struct nvkm_vfn *vfn = nvkm_uvfn(object)->vfn; + struct nvkm_device *device = vfn->subdev.device; + + *addr = device->func->resource_addr(device, 0) + vfn->addr.user; + *size = vfn->func->user.size; + *type = NVKM_OBJECT_MAP_IO; + return 0; +} + +static const struct nvkm_object_func +nvkm_uvfn = { + .map = nvkm_uvfn_map, +}; + +int +nvkm_uvfn_new(struct nvkm_device *device, const struct nvkm_oclass *oclass, + void *argv, u32 argc, struct nvkm_object **pobject) +{ + struct nvkm_uvfn *uvfn; + + if (argc != 0) + return -ENOSYS; + + if (!(uvfn = kzalloc(sizeof(*uvfn), GFP_KERNEL))) + return -ENOMEM; + + nvkm_object_ctor(&nvkm_uvfn, oclass, &uvfn->object); + uvfn->vfn = device->vfn; + + *pobject = &uvfn->object; + return 0; +} diff --git a/drivers/gpu/drm/panel/panel-simple.c b/drivers/gpu/drm/panel/panel-simple.c index 2944228a8e2c..8a3b685c2fcc 100644 --- a/drivers/gpu/drm/panel/panel-simple.c +++ b/drivers/gpu/drm/panel/panel-simple.c @@ -2500,6 +2500,7 @@ static const struct display_timing logictechno_lt161010_2nh_timing = { static const struct panel_desc logictechno_lt161010_2nh = { .timings = &logictechno_lt161010_2nh_timing, .num_timings = 1, + .bpc = 6, .size = { .width = 154, .height = 86, @@ -2529,6 +2530,7 @@ static const struct display_timing logictechno_lt170410_2whc_timing = { static const struct panel_desc logictechno_lt170410_2whc = { .timings = &logictechno_lt170410_2whc_timing, .num_timings = 1, + .bpc = 8, .size = { .width = 217, .height = 136, diff --git a/drivers/gpu/drm/panfrost/panfrost_mmu.c b/drivers/gpu/drm/panfrost/panfrost_mmu.c index e246d914e7f6..4e83a1891f3e 100644 --- a/drivers/gpu/drm/panfrost/panfrost_mmu.c +++ b/drivers/gpu/drm/panfrost/panfrost_mmu.c @@ -250,13 +250,22 @@ void panfrost_mmu_reset(struct panfrost_device *pfdev) static size_t get_pgsize(u64 addr, size_t size, size_t *count) { + /* + * io-pgtable only operates on multiple pages within a single table + * entry, so we need to split at boundaries of the table size, i.e. + * the next block size up. The distance from address A to the next + * boundary of block size B is logically B - A % B, but in unsigned + * two's complement where B is a power of two we get the equivalence + * B - A % B == (B - A) % B == (n * B - A) % B, and choose n = 0 :) + */ size_t blk_offset = -addr % SZ_2M; if (blk_offset || size < SZ_2M) { *count = min_not_zero(blk_offset, size) / SZ_4K; return SZ_4K; } - *count = size / SZ_2M; + blk_offset = -addr % SZ_1G ?: SZ_1G; + *count = min(blk_offset, size) / SZ_2M; return SZ_2M; } diff --git a/drivers/gpu/drm/radeon/Kconfig b/drivers/gpu/drm/radeon/Kconfig index 52819e7f1fca..97a277f9a25e 100644 --- a/drivers/gpu/drm/radeon/Kconfig +++ b/drivers/gpu/drm/radeon/Kconfig @@ -1,4 +1,34 @@ # SPDX-License-Identifier: MIT + +config DRM_RADEON + tristate "ATI Radeon" + depends on DRM && PCI && MMU + depends on AGP || !AGP + select FW_LOADER + select DRM_DISPLAY_DP_HELPER + select DRM_DISPLAY_HELPER + select DRM_KMS_HELPER + select DRM_TTM + select DRM_TTM_HELPER + select SND_HDA_COMPONENT if SND_HDA_CORE + select POWER_SUPPLY + select HWMON + select BACKLIGHT_CLASS_DEVICE + select INTERVAL_TREE + # radeon depends on ACPI_VIDEO when ACPI is enabled, for select to work + # ACPI_VIDEO's dependencies must also be selected. + select INPUT if ACPI + select ACPI_VIDEO if ACPI + # On x86 ACPI_VIDEO also needs ACPI_WMI + select X86_PLATFORM_DEVICES if ACPI && X86 + select ACPI_WMI if ACPI && X86 + help + Choose this option if you have an ATI Radeon graphics card. There + are both PCI and AGP versions. You don't need to choose this to + run the Radeon in plain VGA mode. + + If M is selected, the module will be called radeon. + config DRM_RADEON_USERPTR bool "Always enable userptr support" depends on DRM_RADEON diff --git a/drivers/gpu/drm/radeon/atombios.h b/drivers/gpu/drm/radeon/atombios.h index da35a970fcc0..235e59b547a1 100644 --- a/drivers/gpu/drm/radeon/atombios.h +++ b/drivers/gpu/drm/radeon/atombios.h @@ -3615,7 +3615,7 @@ typedef struct _ATOM_FAKE_EDID_PATCH_RECORD { UCHAR ucRecordType; UCHAR ucFakeEDIDLength; - UCHAR ucFakeEDIDString[1]; // This actually has ucFakeEdidLength elements. + UCHAR ucFakeEDIDString[]; // This actually has ucFakeEdidLength elements. } ATOM_FAKE_EDID_PATCH_RECORD; typedef struct _ATOM_PANEL_RESOLUTION_PATCH_RECORD diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index 166c18d62f6d..2e7161acd443 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -79,6 +79,7 @@ #include <drm/ttm/ttm_execbuf_util.h> #include <drm/drm_gem.h> +#include <drm/drm_audio_component.h> #include "radeon_family.h" #include "radeon_mode.h" @@ -1796,6 +1797,9 @@ struct r600_audio { struct radeon_audio_funcs *hdmi_funcs; struct radeon_audio_funcs *dp_funcs; struct radeon_audio_basic_funcs *funcs; + struct drm_audio_component *component; + bool component_registered; + struct mutex component_mutex; }; /* @@ -2994,6 +2998,10 @@ void radeon_irq_kms_set_irq_n_enabled(struct radeon_device *rdev, bool enable, const char *name, unsigned n); +/* Audio component binding */ +void radeon_audio_component_init(struct radeon_device *rdev); +void radeon_audio_component_fini(struct radeon_device *rdev); + #include "radeon_object.h" #endif diff --git a/drivers/gpu/drm/radeon/radeon_atombios.c b/drivers/gpu/drm/radeon/radeon_atombios.c index 204127bad89c..4ad5a328d920 100644 --- a/drivers/gpu/drm/radeon/radeon_atombios.c +++ b/drivers/gpu/drm/radeon/radeon_atombios.c @@ -1727,8 +1727,11 @@ struct radeon_encoder_atom_dig *radeon_atombios_get_lvds_info(struct } } record += fake_edid_record->ucFakeEDIDLength ? - fake_edid_record->ucFakeEDIDLength + 2 : - sizeof(ATOM_FAKE_EDID_PATCH_RECORD); + struct_size(fake_edid_record, + ucFakeEDIDString, + fake_edid_record->ucFakeEDIDLength) : + /* empty fake edid record must be 3 bytes long */ + sizeof(ATOM_FAKE_EDID_PATCH_RECORD) + 1; break; case LCD_PANEL_RESOLUTION_RECORD_TYPE: panel_res_record = (ATOM_PANEL_RESOLUTION_PATCH_RECORD *)record; diff --git a/drivers/gpu/drm/radeon/radeon_audio.c b/drivers/gpu/drm/radeon/radeon_audio.c index 7c5e80d03fc9..d6ccaf24ee0c 100644 --- a/drivers/gpu/drm/radeon/radeon_audio.c +++ b/drivers/gpu/drm/radeon/radeon_audio.c @@ -23,6 +23,7 @@ */ #include <linux/gcd.h> +#include <linux/component.h> #include <drm/drm_crtc.h> #include "dce6_afmt.h" @@ -180,6 +181,8 @@ static struct radeon_audio_funcs dce6_dp_funcs = { .dpms = evergreen_dp_enable, }; +static void radeon_audio_component_notify(struct radeon_device *rdev, int port); + static void radeon_audio_enable(struct radeon_device *rdev, struct r600_audio_pin *pin, u8 enable_mask) { @@ -207,6 +210,8 @@ static void radeon_audio_enable(struct radeon_device *rdev, if (rdev->audio.funcs->enable) rdev->audio.funcs->enable(rdev, pin, enable_mask); + + radeon_audio_component_notify(rdev, pin->id); } static void radeon_audio_interface_init(struct radeon_device *rdev) @@ -721,3 +726,115 @@ unsigned int radeon_audio_decode_dfs_div(unsigned int div) else return 0; } + +/* + * Audio component support + */ +static void radeon_audio_component_notify(struct radeon_device *rdev, int port) +{ + struct drm_audio_component *acomp; + + mutex_lock(&rdev->audio.component_mutex); + acomp = rdev->audio.component; + if (acomp && acomp->audio_ops && acomp->audio_ops->pin_eld_notify) + acomp->audio_ops->pin_eld_notify(acomp->audio_ops->audio_ptr, + port, -1); + mutex_unlock(&rdev->audio.component_mutex); +} + +static int radeon_audio_component_get_eld(struct device *kdev, int port, + int pipe, bool *enabled, + unsigned char *buf, int max_bytes) +{ + struct drm_device *dev = dev_get_drvdata(kdev); + struct radeon_device *rdev = dev->dev_private; + struct drm_encoder *encoder; + struct radeon_encoder *radeon_encoder; + struct radeon_encoder_atom_dig *dig; + struct drm_connector *connector; + int ret = 0; + + *enabled = false; + if (!rdev->audio.enabled || !rdev->mode_info.mode_config_initialized) + return 0; + + list_for_each_entry(encoder, &rdev->ddev->mode_config.encoder_list, head) { + if (!radeon_encoder_is_digital(encoder)) + continue; + radeon_encoder = to_radeon_encoder(encoder); + dig = radeon_encoder->enc_priv; + if (!dig->pin || dig->pin->id != port) + continue; + connector = radeon_get_connector_for_encoder(encoder); + if (!connector) + continue; + *enabled = true; + ret = drm_eld_size(connector->eld); + memcpy(buf, connector->eld, min(max_bytes, ret)); + break; + } + + return ret; +} + +static const struct drm_audio_component_ops radeon_audio_component_ops = { + .get_eld = radeon_audio_component_get_eld, +}; + +static int radeon_audio_component_bind(struct device *kdev, + struct device *hda_kdev, void *data) +{ + struct drm_device *dev = dev_get_drvdata(kdev); + struct radeon_device *rdev = dev->dev_private; + struct drm_audio_component *acomp = data; + + if (WARN_ON(!device_link_add(hda_kdev, kdev, DL_FLAG_STATELESS))) + return -ENOMEM; + + mutex_lock(&rdev->audio.component_mutex); + acomp->ops = &radeon_audio_component_ops; + acomp->dev = kdev; + rdev->audio.component = acomp; + mutex_unlock(&rdev->audio.component_mutex); + + return 0; +} + +static void radeon_audio_component_unbind(struct device *kdev, + struct device *hda_kdev, void *data) +{ + struct drm_device *dev = dev_get_drvdata(kdev); + struct radeon_device *rdev = dev->dev_private; + struct drm_audio_component *acomp = data; + + device_link_remove(hda_kdev, kdev); + + mutex_lock(&rdev->audio.component_mutex); + rdev->audio.component = NULL; + acomp->ops = NULL; + acomp->dev = NULL; + mutex_unlock(&rdev->audio.component_mutex); +} + +static const struct component_ops radeon_audio_component_bind_ops = { + .bind = radeon_audio_component_bind, + .unbind = radeon_audio_component_unbind, +}; + +void radeon_audio_component_init(struct radeon_device *rdev) +{ + if (rdev->audio.component_registered || + !radeon_audio || !radeon_audio_chipset_supported(rdev)) + return; + + if (!component_add(rdev->dev, &radeon_audio_component_bind_ops)) + rdev->audio.component_registered = true; +} + +void radeon_audio_component_fini(struct radeon_device *rdev) +{ + if (rdev->audio.component_registered) { + component_del(rdev->dev, &radeon_audio_component_bind_ops); + rdev->audio.component_registered = false; + } +} diff --git a/drivers/gpu/drm/radeon/radeon_bios.c b/drivers/gpu/drm/radeon/radeon_bios.c index 33121655d50b..1d99c9a2b56e 100644 --- a/drivers/gpu/drm/radeon/radeon_bios.c +++ b/drivers/gpu/drm/radeon/radeon_bios.c @@ -612,13 +612,14 @@ static bool radeon_acpi_vfct_bios(struct radeon_device *rdev) acpi_size tbl_size; UEFI_ACPI_VFCT *vfct; unsigned offset; + bool r = false; if (!ACPI_SUCCESS(acpi_get_table("VFCT", 1, &hdr))) return false; tbl_size = hdr->length; if (tbl_size < sizeof(UEFI_ACPI_VFCT)) { DRM_ERROR("ACPI VFCT table present but broken (too short #1)\n"); - return false; + goto out; } vfct = (UEFI_ACPI_VFCT *)hdr; @@ -631,13 +632,13 @@ static bool radeon_acpi_vfct_bios(struct radeon_device *rdev) offset += sizeof(VFCT_IMAGE_HEADER); if (offset > tbl_size) { DRM_ERROR("ACPI VFCT image header truncated\n"); - return false; + goto out; } offset += vhdr->ImageLength; if (offset > tbl_size) { DRM_ERROR("ACPI VFCT image truncated\n"); - return false; + goto out; } if (vhdr->ImageLength && @@ -649,15 +650,18 @@ static bool radeon_acpi_vfct_bios(struct radeon_device *rdev) rdev->bios = kmemdup(&vbios->VbiosContent, vhdr->ImageLength, GFP_KERNEL); + if (rdev->bios) + r = true; - if (!rdev->bios) - return false; - return true; + goto out; } } DRM_ERROR("ACPI VFCT table present but broken (too short #2)\n"); - return false; + +out: + acpi_put_table(hdr); + return r; } #else static inline bool radeon_acpi_vfct_bios(struct radeon_device *rdev) diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c index a556b6be1137..6344454a7721 100644 --- a/drivers/gpu/drm/radeon/radeon_device.c +++ b/drivers/gpu/drm/radeon/radeon_device.c @@ -1207,7 +1207,7 @@ static void radeon_check_arguments(struct radeon_device *rdev) * @pdev: pci dev pointer * @state: vga_switcheroo state * - * Callback for the switcheroo driver. Suspends or resumes the + * Callback for the switcheroo driver. Suspends or resumes * the asics before or after it is powered up using ACPI methods. */ static void radeon_switcheroo_set_state(struct pci_dev *pdev, enum vga_switcheroo_state state) @@ -1312,6 +1312,7 @@ int radeon_device_init(struct radeon_device *rdev, mutex_init(&rdev->pm.mutex); mutex_init(&rdev->gpu_clock_mutex); mutex_init(&rdev->srbm_mutex); + mutex_init(&rdev->audio.component_mutex); init_rwsem(&rdev->pm.mclk_lock); init_rwsem(&rdev->exclusive_lock); init_waitqueue_head(&rdev->irq.vblank_queue); @@ -1451,6 +1452,8 @@ int radeon_device_init(struct radeon_device *rdev, goto failed; } + radeon_audio_component_init(rdev); + r = radeon_ib_ring_tests(rdev); if (r) DRM_ERROR("ib ring test failed (%d).\n", r); @@ -1513,6 +1516,7 @@ void radeon_device_fini(struct radeon_device *rdev) rdev->shutdown = true; /* evict vram memory */ radeon_bo_evict_vram(rdev); + radeon_audio_component_fini(rdev); radeon_fini(rdev); if (!pci_is_thunderbolt_attached(rdev->pdev)) vga_switcheroo_unregister_client(rdev->pdev); diff --git a/drivers/gpu/drm/radeon/radeon_ttm.c b/drivers/gpu/drm/radeon/radeon_ttm.c index fff48306c05f..30402b5ce4c5 100644 --- a/drivers/gpu/drm/radeon/radeon_ttm.c +++ b/drivers/gpu/drm/radeon/radeon_ttm.c @@ -869,11 +869,11 @@ static ssize_t radeon_ttm_gtt_read(struct file *f, char __user *buf, page = rdev->gart.pages[p]; if (page) { - ptr = kmap(page); + ptr = kmap_local_page(page); ptr += off; r = copy_to_user(buf, ptr, cur_size); - kunmap(rdev->gart.pages[p]); + kunmap_local(ptr); } else r = clear_user(buf, cur_size); diff --git a/drivers/gpu/drm/rcar-du/Kconfig b/drivers/gpu/drm/rcar-du/Kconfig index c959e8c6be7d..1065dca885ef 100644 --- a/drivers/gpu/drm/rcar-du/Kconfig +++ b/drivers/gpu/drm/rcar-du/Kconfig @@ -41,15 +41,26 @@ config DRM_RCAR_LVDS depends on DRM_RCAR_USE_LVDS select DRM_KMS_HELPER select DRM_PANEL - select OF_FLATTREE - select OF_OVERLAY + +config DRM_RCAR_USE_MIPI_DSI + bool "R-Car DU MIPI DSI Encoder Support" + depends on DRM_BRIDGE && OF + default DRM_RCAR_DU + help + Enable support for the R-Car Display Unit embedded MIPI DSI encoders. config DRM_RCAR_MIPI_DSI - tristate "R-Car DU MIPI DSI Encoder Support" - depends on DRM && DRM_BRIDGE && OF + def_tristate DRM_RCAR_DU + depends on DRM_RCAR_USE_MIPI_DSI + select DRM_MIPI_DSI + +config DRM_RZG2L_MIPI_DSI + tristate "RZ/G2L MIPI DSI Encoder Support" + depends on DRM_BRIDGE && OF + depends on ARCH_RENESAS || COMPILE_TEST select DRM_MIPI_DSI help - Enable support for the R-Car Display Unit embedded MIPI DSI encoders. + Enable support for the RZ/G2L Display Unit embedded MIPI DSI encoders. config DRM_RCAR_VSP bool "R-Car DU VSP Compositor Support" if ARM diff --git a/drivers/gpu/drm/rcar-du/Makefile b/drivers/gpu/drm/rcar-du/Makefile index 6f132325c8b7..b8f2c82651d9 100644 --- a/drivers/gpu/drm/rcar-du/Makefile +++ b/drivers/gpu/drm/rcar-du/Makefile @@ -14,3 +14,5 @@ obj-$(CONFIG_DRM_RCAR_DU) += rcar-du-drm.o obj-$(CONFIG_DRM_RCAR_DW_HDMI) += rcar_dw_hdmi.o obj-$(CONFIG_DRM_RCAR_LVDS) += rcar_lvds.o obj-$(CONFIG_DRM_RCAR_MIPI_DSI) += rcar_mipi_dsi.o + +obj-$(CONFIG_DRM_RZG2L_MIPI_DSI) += rzg2l_mipi_dsi.o diff --git a/drivers/gpu/drm/rcar-du/rzg2l_mipi_dsi.c b/drivers/gpu/drm/rcar-du/rzg2l_mipi_dsi.c new file mode 100644 index 000000000000..aa95b85a2964 --- /dev/null +++ b/drivers/gpu/drm/rcar-du/rzg2l_mipi_dsi.c @@ -0,0 +1,816 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * RZ/G2L MIPI DSI Encoder Driver + * + * Copyright (C) 2022 Renesas Electronics Corporation + */ +#include <linux/clk.h> +#include <linux/delay.h> +#include <linux/io.h> +#include <linux/iopoll.h> +#include <linux/module.h> +#include <linux/of.h> +#include <linux/of_device.h> +#include <linux/of_graph.h> +#include <linux/platform_device.h> +#include <linux/pm_runtime.h> +#include <linux/reset.h> +#include <linux/slab.h> + +#include <drm/drm_atomic.h> +#include <drm/drm_atomic_helper.h> +#include <drm/drm_bridge.h> +#include <drm/drm_mipi_dsi.h> +#include <drm/drm_of.h> +#include <drm/drm_panel.h> +#include <drm/drm_probe_helper.h> + +#include "rzg2l_mipi_dsi_regs.h" + +struct rzg2l_mipi_dsi { + struct device *dev; + void __iomem *mmio; + + struct reset_control *rstc; + struct reset_control *arstc; + struct reset_control *prstc; + + struct mipi_dsi_host host; + struct drm_bridge bridge; + struct drm_bridge *next_bridge; + + struct clk *vclk; + + enum mipi_dsi_pixel_format format; + unsigned int num_data_lanes; + unsigned int lanes; + unsigned long mode_flags; +}; + +static inline struct rzg2l_mipi_dsi * +bridge_to_rzg2l_mipi_dsi(struct drm_bridge *bridge) +{ + return container_of(bridge, struct rzg2l_mipi_dsi, bridge); +} + +static inline struct rzg2l_mipi_dsi * +host_to_rzg2l_mipi_dsi(struct mipi_dsi_host *host) +{ + return container_of(host, struct rzg2l_mipi_dsi, host); +} + +struct rzg2l_mipi_dsi_timings { + unsigned long hsfreq_max; + u32 t_init; + u32 tclk_prepare; + u32 ths_prepare; + u32 tclk_zero; + u32 tclk_pre; + u32 tclk_post; + u32 tclk_trail; + u32 ths_zero; + u32 ths_trail; + u32 ths_exit; + u32 tlpx; +}; + +static const struct rzg2l_mipi_dsi_timings rzg2l_mipi_dsi_global_timings[] = { + { + .hsfreq_max = 80000, + .t_init = 79801, + .tclk_prepare = 8, + .ths_prepare = 13, + .tclk_zero = 33, + .tclk_pre = 24, + .tclk_post = 94, + .tclk_trail = 10, + .ths_zero = 23, + .ths_trail = 17, + .ths_exit = 13, + .tlpx = 6, + }, + { + .hsfreq_max = 125000, + .t_init = 79801, + .tclk_prepare = 8, + .ths_prepare = 12, + .tclk_zero = 33, + .tclk_pre = 15, + .tclk_post = 94, + .tclk_trail = 10, + .ths_zero = 23, + .ths_trail = 17, + .ths_exit = 13, + .tlpx = 6, + }, + { + .hsfreq_max = 250000, + .t_init = 79801, + .tclk_prepare = 8, + .ths_prepare = 12, + .tclk_zero = 33, + .tclk_pre = 13, + .tclk_post = 94, + .tclk_trail = 10, + .ths_zero = 23, + .ths_trail = 16, + .ths_exit = 13, + .tlpx = 6, + }, + { + .hsfreq_max = 360000, + .t_init = 79801, + .tclk_prepare = 8, + .ths_prepare = 10, + .tclk_zero = 33, + .tclk_pre = 4, + .tclk_post = 35, + .tclk_trail = 7, + .ths_zero = 16, + .ths_trail = 9, + .ths_exit = 13, + .tlpx = 6, + }, + { + .hsfreq_max = 720000, + .t_init = 79801, + .tclk_prepare = 8, + .ths_prepare = 9, + .tclk_zero = 33, + .tclk_pre = 4, + .tclk_post = 35, + .tclk_trail = 7, + .ths_zero = 16, + .ths_trail = 9, + .ths_exit = 13, + .tlpx = 6, + }, + { + .hsfreq_max = 1500000, + .t_init = 79801, + .tclk_prepare = 8, + .ths_prepare = 9, + .tclk_zero = 33, + .tclk_pre = 4, + .tclk_post = 35, + .tclk_trail = 7, + .ths_zero = 16, + .ths_trail = 9, + .ths_exit = 13, + .tlpx = 6, + }, +}; + +static void rzg2l_mipi_dsi_phy_write(struct rzg2l_mipi_dsi *dsi, u32 reg, u32 data) +{ + iowrite32(data, dsi->mmio + reg); +} + +static void rzg2l_mipi_dsi_link_write(struct rzg2l_mipi_dsi *dsi, u32 reg, u32 data) +{ + iowrite32(data, dsi->mmio + LINK_REG_OFFSET + reg); +} + +static u32 rzg2l_mipi_dsi_phy_read(struct rzg2l_mipi_dsi *dsi, u32 reg) +{ + return ioread32(dsi->mmio + reg); +} + +static u32 rzg2l_mipi_dsi_link_read(struct rzg2l_mipi_dsi *dsi, u32 reg) +{ + return ioread32(dsi->mmio + LINK_REG_OFFSET + reg); +} + +/* ----------------------------------------------------------------------------- + * Hardware Setup + */ + +static int rzg2l_mipi_dsi_dphy_init(struct rzg2l_mipi_dsi *dsi, + unsigned long hsfreq) +{ + const struct rzg2l_mipi_dsi_timings *dphy_timings; + unsigned int i; + u32 dphyctrl0; + u32 dphytim0; + u32 dphytim1; + u32 dphytim2; + u32 dphytim3; + int ret; + + /* All DSI global operation timings are set with recommended setting */ + for (i = 0; i < ARRAY_SIZE(rzg2l_mipi_dsi_global_timings); ++i) { + dphy_timings = &rzg2l_mipi_dsi_global_timings[i]; + if (hsfreq <= dphy_timings->hsfreq_max) + break; + } + + /* Initializing DPHY before accessing LINK */ + dphyctrl0 = DSIDPHYCTRL0_CAL_EN_HSRX_OFS | DSIDPHYCTRL0_CMN_MASTER_EN | + DSIDPHYCTRL0_RE_VDD_DETVCCQLV18 | DSIDPHYCTRL0_EN_BGR; + + rzg2l_mipi_dsi_phy_write(dsi, DSIDPHYCTRL0, dphyctrl0); + usleep_range(20, 30); + + dphyctrl0 |= DSIDPHYCTRL0_EN_LDO1200; + rzg2l_mipi_dsi_phy_write(dsi, DSIDPHYCTRL0, dphyctrl0); + usleep_range(10, 20); + + dphytim0 = DSIDPHYTIM0_TCLK_MISS(0) | + DSIDPHYTIM0_T_INIT(dphy_timings->t_init); + dphytim1 = DSIDPHYTIM1_THS_PREPARE(dphy_timings->ths_prepare) | + DSIDPHYTIM1_TCLK_PREPARE(dphy_timings->tclk_prepare) | + DSIDPHYTIM1_THS_SETTLE(0) | + DSIDPHYTIM1_TCLK_SETTLE(0); + dphytim2 = DSIDPHYTIM2_TCLK_TRAIL(dphy_timings->tclk_trail) | + DSIDPHYTIM2_TCLK_POST(dphy_timings->tclk_post) | + DSIDPHYTIM2_TCLK_PRE(dphy_timings->tclk_pre) | + DSIDPHYTIM2_TCLK_ZERO(dphy_timings->tclk_zero); + dphytim3 = DSIDPHYTIM3_TLPX(dphy_timings->tlpx) | + DSIDPHYTIM3_THS_EXIT(dphy_timings->ths_exit) | + DSIDPHYTIM3_THS_TRAIL(dphy_timings->ths_trail) | + DSIDPHYTIM3_THS_ZERO(dphy_timings->ths_zero); + + rzg2l_mipi_dsi_phy_write(dsi, DSIDPHYTIM0, dphytim0); + rzg2l_mipi_dsi_phy_write(dsi, DSIDPHYTIM1, dphytim1); + rzg2l_mipi_dsi_phy_write(dsi, DSIDPHYTIM2, dphytim2); + rzg2l_mipi_dsi_phy_write(dsi, DSIDPHYTIM3, dphytim3); + + ret = reset_control_deassert(dsi->rstc); + if (ret < 0) + return ret; + + udelay(1); + + return 0; +} + +static void rzg2l_mipi_dsi_dphy_exit(struct rzg2l_mipi_dsi *dsi) +{ + u32 dphyctrl0; + + dphyctrl0 = rzg2l_mipi_dsi_phy_read(dsi, DSIDPHYCTRL0); + + dphyctrl0 &= ~(DSIDPHYCTRL0_EN_LDO1200 | DSIDPHYCTRL0_EN_BGR); + rzg2l_mipi_dsi_phy_write(dsi, DSIDPHYCTRL0, dphyctrl0); + + reset_control_assert(dsi->rstc); +} + +static int rzg2l_mipi_dsi_startup(struct rzg2l_mipi_dsi *dsi, + const struct drm_display_mode *mode) +{ + unsigned long hsfreq; + unsigned int bpp; + u32 txsetr; + u32 clstptsetr; + u32 lptrnstsetr; + u32 clkkpt; + u32 clkbfht; + u32 clkstpt; + u32 golpbkt; + int ret; + + /* + * Relationship between hsclk and vclk must follow + * vclk * bpp = hsclk * 8 * lanes + * where vclk: video clock (Hz) + * bpp: video pixel bit depth + * hsclk: DSI HS Byte clock frequency (Hz) + * lanes: number of data lanes + * + * hsclk(bit) = hsclk(byte) * 8 + */ + bpp = mipi_dsi_pixel_format_to_bpp(dsi->format); + hsfreq = (mode->clock * bpp * 8) / (8 * dsi->lanes); + + ret = pm_runtime_resume_and_get(dsi->dev); + if (ret < 0) + return ret; + + clk_set_rate(dsi->vclk, mode->clock * 1000); + + ret = rzg2l_mipi_dsi_dphy_init(dsi, hsfreq); + if (ret < 0) + goto err_phy; + + /* Enable Data lanes and Clock lanes */ + txsetr = TXSETR_DLEN | TXSETR_NUMLANEUSE(dsi->lanes - 1) | TXSETR_CLEN; + rzg2l_mipi_dsi_link_write(dsi, TXSETR, txsetr); + + /* + * Global timings characteristic depends on high speed Clock Frequency + * Currently MIPI DSI-IF just supports maximum FHD@60 with: + * - videoclock = 148.5 (MHz) + * - bpp: maximum 24bpp + * - data lanes: maximum 4 lanes + * Therefore maximum hsclk will be 891 Mbps. + */ + if (hsfreq > 445500) { + clkkpt = 12; + clkbfht = 15; + clkstpt = 48; + golpbkt = 75; + } else if (hsfreq > 250000) { + clkkpt = 7; + clkbfht = 8; + clkstpt = 27; + golpbkt = 40; + } else { + clkkpt = 8; + clkbfht = 6; + clkstpt = 24; + golpbkt = 29; + } + + clstptsetr = CLSTPTSETR_CLKKPT(clkkpt) | CLSTPTSETR_CLKBFHT(clkbfht) | + CLSTPTSETR_CLKSTPT(clkstpt); + rzg2l_mipi_dsi_link_write(dsi, CLSTPTSETR, clstptsetr); + + lptrnstsetr = LPTRNSTSETR_GOLPBKT(golpbkt); + rzg2l_mipi_dsi_link_write(dsi, LPTRNSTSETR, lptrnstsetr); + + return 0; + +err_phy: + rzg2l_mipi_dsi_dphy_exit(dsi); + pm_runtime_put(dsi->dev); + + return ret; +} + +static void rzg2l_mipi_dsi_stop(struct rzg2l_mipi_dsi *dsi) +{ + rzg2l_mipi_dsi_dphy_exit(dsi); + pm_runtime_put(dsi->dev); +} + +static void rzg2l_mipi_dsi_set_display_timing(struct rzg2l_mipi_dsi *dsi, + const struct drm_display_mode *mode) +{ + u32 vich1ppsetr; + u32 vich1vssetr; + u32 vich1vpsetr; + u32 vich1hssetr; + u32 vich1hpsetr; + int dsi_format; + u32 delay[2]; + u8 index; + + /* Configuration for Pixel Packet */ + dsi_format = mipi_dsi_pixel_format_to_bpp(dsi->format); + switch (dsi_format) { + case 24: + vich1ppsetr = VICH1PPSETR_DT_RGB24; + break; + case 18: + vich1ppsetr = VICH1PPSETR_DT_RGB18; + break; + } + + if ((dsi->mode_flags & MIPI_DSI_MODE_VIDEO_SYNC_PULSE) && + !(dsi->mode_flags & MIPI_DSI_MODE_VIDEO_BURST)) + vich1ppsetr |= VICH1PPSETR_TXESYNC_PULSE; + + rzg2l_mipi_dsi_link_write(dsi, VICH1PPSETR, vich1ppsetr); + + /* Configuration for Video Parameters */ + vich1vssetr = VICH1VSSETR_VACTIVE(mode->vdisplay) | + VICH1VSSETR_VSA(mode->vsync_end - mode->vsync_start); + vich1vssetr |= (mode->flags & DRM_MODE_FLAG_PVSYNC) ? + VICH1VSSETR_VSPOL_HIGH : VICH1VSSETR_VSPOL_LOW; + + vich1vpsetr = VICH1VPSETR_VFP(mode->vsync_start - mode->vdisplay) | + VICH1VPSETR_VBP(mode->vtotal - mode->vsync_end); + + vich1hssetr = VICH1HSSETR_HACTIVE(mode->hdisplay) | + VICH1HSSETR_HSA(mode->hsync_end - mode->hsync_start); + vich1hssetr |= (mode->flags & DRM_MODE_FLAG_PHSYNC) ? + VICH1HSSETR_HSPOL_HIGH : VICH1HSSETR_HSPOL_LOW; + + vich1hpsetr = VICH1HPSETR_HFP(mode->hsync_start - mode->hdisplay) | + VICH1HPSETR_HBP(mode->htotal - mode->hsync_end); + + rzg2l_mipi_dsi_link_write(dsi, VICH1VSSETR, vich1vssetr); + rzg2l_mipi_dsi_link_write(dsi, VICH1VPSETR, vich1vpsetr); + rzg2l_mipi_dsi_link_write(dsi, VICH1HSSETR, vich1hssetr); + rzg2l_mipi_dsi_link_write(dsi, VICH1HPSETR, vich1hpsetr); + + /* + * Configuration for Delay Value + * Delay value based on 2 ranges of video clock. + * 74.25MHz is videoclock of HD@60p or FHD@30p + */ + if (mode->clock > 74250) { + delay[0] = 231; + delay[1] = 216; + } else { + delay[0] = 220; + delay[1] = 212; + } + + if (dsi->mode_flags & MIPI_DSI_CLOCK_NON_CONTINUOUS) + index = 0; + else + index = 1; + + rzg2l_mipi_dsi_link_write(dsi, VICH1SET1R, + VICH1SET1R_DLY(delay[index])); +} + +static int rzg2l_mipi_dsi_start_hs_clock(struct rzg2l_mipi_dsi *dsi) +{ + bool is_clk_cont; + u32 hsclksetr; + u32 status; + int ret; + + is_clk_cont = !(dsi->mode_flags & MIPI_DSI_CLOCK_NON_CONTINUOUS); + + /* Start HS clock */ + hsclksetr = HSCLKSETR_HSCLKRUN_HS | (is_clk_cont ? + HSCLKSETR_HSCLKMODE_CONT : + HSCLKSETR_HSCLKMODE_NON_CONT); + rzg2l_mipi_dsi_link_write(dsi, HSCLKSETR, hsclksetr); + + if (is_clk_cont) { + ret = read_poll_timeout(rzg2l_mipi_dsi_link_read, status, + status & PLSR_CLLP2HS, + 2000, 20000, false, dsi, PLSR); + if (ret < 0) { + dev_err(dsi->dev, "failed to start HS clock\n"); + return ret; + } + } + + dev_dbg(dsi->dev, "Start High Speed Clock with %s clock mode", + is_clk_cont ? "continuous" : "non-continuous"); + + return 0; +} + +static int rzg2l_mipi_dsi_stop_hs_clock(struct rzg2l_mipi_dsi *dsi) +{ + bool is_clk_cont; + u32 status; + int ret; + + is_clk_cont = !(dsi->mode_flags & MIPI_DSI_CLOCK_NON_CONTINUOUS); + + /* Stop HS clock */ + rzg2l_mipi_dsi_link_write(dsi, HSCLKSETR, + is_clk_cont ? HSCLKSETR_HSCLKMODE_CONT : + HSCLKSETR_HSCLKMODE_NON_CONT); + + if (is_clk_cont) { + ret = read_poll_timeout(rzg2l_mipi_dsi_link_read, status, + status & PLSR_CLHS2LP, + 2000, 20000, false, dsi, PLSR); + if (ret < 0) { + dev_err(dsi->dev, "failed to stop HS clock\n"); + return ret; + } + } + + return 0; +} + +static int rzg2l_mipi_dsi_start_video(struct rzg2l_mipi_dsi *dsi) +{ + u32 vich1set0r; + u32 status; + int ret; + + /* Configuration for Blanking sequence and start video input*/ + vich1set0r = VICH1SET0R_HFPNOLP | VICH1SET0R_HBPNOLP | + VICH1SET0R_HSANOLP | VICH1SET0R_VSTART; + rzg2l_mipi_dsi_link_write(dsi, VICH1SET0R, vich1set0r); + + ret = read_poll_timeout(rzg2l_mipi_dsi_link_read, status, + status & VICH1SR_VIRDY, + 2000, 20000, false, dsi, VICH1SR); + if (ret < 0) + dev_err(dsi->dev, "Failed to start video signal input\n"); + + return ret; +} + +static int rzg2l_mipi_dsi_stop_video(struct rzg2l_mipi_dsi *dsi) +{ + u32 status; + int ret; + + rzg2l_mipi_dsi_link_write(dsi, VICH1SET0R, VICH1SET0R_VSTPAFT); + ret = read_poll_timeout(rzg2l_mipi_dsi_link_read, status, + (status & VICH1SR_STOP) && (!(status & VICH1SR_RUNNING)), + 2000, 20000, false, dsi, VICH1SR); + if (ret < 0) + goto err; + + ret = read_poll_timeout(rzg2l_mipi_dsi_link_read, status, + !(status & LINKSR_HSBUSY), + 2000, 20000, false, dsi, LINKSR); + if (ret < 0) + goto err; + + return 0; + +err: + dev_err(dsi->dev, "Failed to stop video signal input\n"); + return ret; +} + +/* ----------------------------------------------------------------------------- + * Bridge + */ + +static int rzg2l_mipi_dsi_attach(struct drm_bridge *bridge, + enum drm_bridge_attach_flags flags) +{ + struct rzg2l_mipi_dsi *dsi = bridge_to_rzg2l_mipi_dsi(bridge); + + return drm_bridge_attach(bridge->encoder, dsi->next_bridge, bridge, + flags); +} + +static void rzg2l_mipi_dsi_atomic_enable(struct drm_bridge *bridge, + struct drm_bridge_state *old_bridge_state) +{ + struct drm_atomic_state *state = old_bridge_state->base.state; + struct rzg2l_mipi_dsi *dsi = bridge_to_rzg2l_mipi_dsi(bridge); + const struct drm_display_mode *mode; + struct drm_connector *connector; + struct drm_crtc *crtc; + int ret; + + connector = drm_atomic_get_new_connector_for_encoder(state, bridge->encoder); + crtc = drm_atomic_get_new_connector_state(state, connector)->crtc; + mode = &drm_atomic_get_new_crtc_state(state, crtc)->adjusted_mode; + + ret = rzg2l_mipi_dsi_startup(dsi, mode); + if (ret < 0) + return; + + rzg2l_mipi_dsi_set_display_timing(dsi, mode); + + ret = rzg2l_mipi_dsi_start_hs_clock(dsi); + if (ret < 0) + goto err_stop; + + ret = rzg2l_mipi_dsi_start_video(dsi); + if (ret < 0) + goto err_stop_clock; + + return; + +err_stop_clock: + rzg2l_mipi_dsi_stop_hs_clock(dsi); +err_stop: + rzg2l_mipi_dsi_stop(dsi); +} + +static void rzg2l_mipi_dsi_atomic_disable(struct drm_bridge *bridge, + struct drm_bridge_state *old_bridge_state) +{ + struct rzg2l_mipi_dsi *dsi = bridge_to_rzg2l_mipi_dsi(bridge); + + rzg2l_mipi_dsi_stop_video(dsi); + rzg2l_mipi_dsi_stop_hs_clock(dsi); + rzg2l_mipi_dsi_stop(dsi); +} + +static enum drm_mode_status +rzg2l_mipi_dsi_bridge_mode_valid(struct drm_bridge *bridge, + const struct drm_display_info *info, + const struct drm_display_mode *mode) +{ + if (mode->clock > 148500) + return MODE_CLOCK_HIGH; + + return MODE_OK; +} + +static const struct drm_bridge_funcs rzg2l_mipi_dsi_bridge_ops = { + .attach = rzg2l_mipi_dsi_attach, + .atomic_duplicate_state = drm_atomic_helper_bridge_duplicate_state, + .atomic_destroy_state = drm_atomic_helper_bridge_destroy_state, + .atomic_reset = drm_atomic_helper_bridge_reset, + .atomic_enable = rzg2l_mipi_dsi_atomic_enable, + .atomic_disable = rzg2l_mipi_dsi_atomic_disable, + .mode_valid = rzg2l_mipi_dsi_bridge_mode_valid, +}; + +/* ----------------------------------------------------------------------------- + * Host setting + */ + +static int rzg2l_mipi_dsi_host_attach(struct mipi_dsi_host *host, + struct mipi_dsi_device *device) +{ + struct rzg2l_mipi_dsi *dsi = host_to_rzg2l_mipi_dsi(host); + int ret; + + if (device->lanes > dsi->num_data_lanes) { + dev_err(dsi->dev, + "Number of lines of device (%u) exceeds host (%u)\n", + device->lanes, dsi->num_data_lanes); + return -EINVAL; + } + + switch (mipi_dsi_pixel_format_to_bpp(device->format)) { + case 24: + case 18: + break; + default: + dev_err(dsi->dev, "Unsupported format 0x%04x\n", device->format); + return -EINVAL; + } + + dsi->lanes = device->lanes; + dsi->format = device->format; + dsi->mode_flags = device->mode_flags; + + dsi->next_bridge = devm_drm_of_get_bridge(dsi->dev, dsi->dev->of_node, + 1, 0); + if (IS_ERR(dsi->next_bridge)) { + ret = PTR_ERR(dsi->next_bridge); + dev_err(dsi->dev, "failed to get next bridge: %d\n", ret); + return ret; + } + + drm_bridge_add(&dsi->bridge); + + return 0; +} + +static int rzg2l_mipi_dsi_host_detach(struct mipi_dsi_host *host, + struct mipi_dsi_device *device) +{ + struct rzg2l_mipi_dsi *dsi = host_to_rzg2l_mipi_dsi(host); + + drm_bridge_remove(&dsi->bridge); + + return 0; +} + +static const struct mipi_dsi_host_ops rzg2l_mipi_dsi_host_ops = { + .attach = rzg2l_mipi_dsi_host_attach, + .detach = rzg2l_mipi_dsi_host_detach, +}; + +/* ----------------------------------------------------------------------------- + * Power Management + */ + +static int __maybe_unused rzg2l_mipi_pm_runtime_suspend(struct device *dev) +{ + struct rzg2l_mipi_dsi *dsi = dev_get_drvdata(dev); + + reset_control_assert(dsi->prstc); + reset_control_assert(dsi->arstc); + + return 0; +} + +static int __maybe_unused rzg2l_mipi_pm_runtime_resume(struct device *dev) +{ + struct rzg2l_mipi_dsi *dsi = dev_get_drvdata(dev); + int ret; + + ret = reset_control_deassert(dsi->arstc); + if (ret < 0) + return ret; + + ret = reset_control_deassert(dsi->prstc); + if (ret < 0) + reset_control_assert(dsi->arstc); + + return ret; +} + +static const struct dev_pm_ops rzg2l_mipi_pm_ops = { + SET_RUNTIME_PM_OPS(rzg2l_mipi_pm_runtime_suspend, rzg2l_mipi_pm_runtime_resume, NULL) +}; + +/* ----------------------------------------------------------------------------- + * Probe & Remove + */ + +static int rzg2l_mipi_dsi_probe(struct platform_device *pdev) +{ + unsigned int num_data_lanes; + struct rzg2l_mipi_dsi *dsi; + u32 txsetr; + int ret; + + dsi = devm_kzalloc(&pdev->dev, sizeof(*dsi), GFP_KERNEL); + if (!dsi) + return -ENOMEM; + + platform_set_drvdata(pdev, dsi); + dsi->dev = &pdev->dev; + + ret = drm_of_get_data_lanes_count_ep(dsi->dev->of_node, 1, 0, 1, 4); + if (ret < 0) + return dev_err_probe(dsi->dev, ret, + "missing or invalid data-lanes property\n"); + + num_data_lanes = ret; + + dsi->mmio = devm_platform_ioremap_resource(pdev, 0); + if (IS_ERR(dsi->mmio)) + return PTR_ERR(dsi->mmio); + + dsi->vclk = devm_clk_get(dsi->dev, "vclk"); + if (IS_ERR(dsi->vclk)) + return PTR_ERR(dsi->vclk); + + dsi->rstc = devm_reset_control_get_exclusive(dsi->dev, "rst"); + if (IS_ERR(dsi->rstc)) + return dev_err_probe(dsi->dev, PTR_ERR(dsi->rstc), + "failed to get rst\n"); + + dsi->arstc = devm_reset_control_get_exclusive(dsi->dev, "arst"); + if (IS_ERR(dsi->arstc)) + return dev_err_probe(&pdev->dev, PTR_ERR(dsi->arstc), + "failed to get arst\n"); + + dsi->prstc = devm_reset_control_get_exclusive(dsi->dev, "prst"); + if (IS_ERR(dsi->prstc)) + return dev_err_probe(dsi->dev, PTR_ERR(dsi->prstc), + "failed to get prst\n"); + + platform_set_drvdata(pdev, dsi); + + pm_runtime_enable(dsi->dev); + + ret = pm_runtime_resume_and_get(dsi->dev); + if (ret < 0) + goto err_pm_disable; + + /* + * TXSETR register can be read only after DPHY init. But during probe + * mode->clock and format are not available. So initialize DPHY with + * timing parameters for 80Mbps. + */ + ret = rzg2l_mipi_dsi_dphy_init(dsi, 80000); + if (ret < 0) + goto err_phy; + + txsetr = rzg2l_mipi_dsi_link_read(dsi, TXSETR); + dsi->num_data_lanes = min(((txsetr >> 16) & 3) + 1, num_data_lanes); + rzg2l_mipi_dsi_dphy_exit(dsi); + pm_runtime_put(dsi->dev); + + /* Initialize the DRM bridge. */ + dsi->bridge.funcs = &rzg2l_mipi_dsi_bridge_ops; + dsi->bridge.of_node = dsi->dev->of_node; + + /* Init host device */ + dsi->host.dev = dsi->dev; + dsi->host.ops = &rzg2l_mipi_dsi_host_ops; + ret = mipi_dsi_host_register(&dsi->host); + if (ret < 0) + goto err_pm_disable; + + return 0; + +err_phy: + rzg2l_mipi_dsi_dphy_exit(dsi); + pm_runtime_put(dsi->dev); +err_pm_disable: + pm_runtime_disable(dsi->dev); + return ret; +} + +static int rzg2l_mipi_dsi_remove(struct platform_device *pdev) +{ + struct rzg2l_mipi_dsi *dsi = platform_get_drvdata(pdev); + + mipi_dsi_host_unregister(&dsi->host); + pm_runtime_disable(&pdev->dev); + + return 0; +} + +static const struct of_device_id rzg2l_mipi_dsi_of_table[] = { + { .compatible = "renesas,rzg2l-mipi-dsi" }, + { /* sentinel */ } +}; + +MODULE_DEVICE_TABLE(of, rzg2l_mipi_dsi_of_table); + +static struct platform_driver rzg2l_mipi_dsi_platform_driver = { + .probe = rzg2l_mipi_dsi_probe, + .remove = rzg2l_mipi_dsi_remove, + .driver = { + .name = "rzg2l-mipi-dsi", + .pm = &rzg2l_mipi_pm_ops, + .of_match_table = rzg2l_mipi_dsi_of_table, + }, +}; + +module_platform_driver(rzg2l_mipi_dsi_platform_driver); + +MODULE_AUTHOR("Biju Das <biju.das.jz@bp.renesas.com>"); +MODULE_DESCRIPTION("Renesas RZ/G2L MIPI DSI Encoder Driver"); +MODULE_LICENSE("GPL"); diff --git a/drivers/gpu/drm/rcar-du/rzg2l_mipi_dsi_regs.h b/drivers/gpu/drm/rcar-du/rzg2l_mipi_dsi_regs.h new file mode 100644 index 000000000000..1dbc16ec64a4 --- /dev/null +++ b/drivers/gpu/drm/rcar-du/rzg2l_mipi_dsi_regs.h @@ -0,0 +1,151 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* + * RZ/G2L MIPI DSI Interface Registers Definitions + * + * Copyright (C) 2022 Renesas Electronics Corporation + */ + +#ifndef __RZG2L_MIPI_DSI_REGS_H__ +#define __RZG2L_MIPI_DSI_REGS_H__ + +#include <linux/bits.h> + +/* DPHY Registers */ +#define DSIDPHYCTRL0 0x00 +#define DSIDPHYCTRL0_CAL_EN_HSRX_OFS BIT(16) +#define DSIDPHYCTRL0_CMN_MASTER_EN BIT(8) +#define DSIDPHYCTRL0_RE_VDD_DETVCCQLV18 BIT(2) +#define DSIDPHYCTRL0_EN_LDO1200 BIT(1) +#define DSIDPHYCTRL0_EN_BGR BIT(0) + +#define DSIDPHYTIM0 0x04 +#define DSIDPHYTIM0_TCLK_MISS(x) ((x) << 24) +#define DSIDPHYTIM0_T_INIT(x) ((x) << 0) + +#define DSIDPHYTIM1 0x08 +#define DSIDPHYTIM1_THS_PREPARE(x) ((x) << 24) +#define DSIDPHYTIM1_TCLK_PREPARE(x) ((x) << 16) +#define DSIDPHYTIM1_THS_SETTLE(x) ((x) << 8) +#define DSIDPHYTIM1_TCLK_SETTLE(x) ((x) << 0) + +#define DSIDPHYTIM2 0x0c +#define DSIDPHYTIM2_TCLK_TRAIL(x) ((x) << 24) +#define DSIDPHYTIM2_TCLK_POST(x) ((x) << 16) +#define DSIDPHYTIM2_TCLK_PRE(x) ((x) << 8) +#define DSIDPHYTIM2_TCLK_ZERO(x) ((x) << 0) + +#define DSIDPHYTIM3 0x10 +#define DSIDPHYTIM3_TLPX(x) ((x) << 24) +#define DSIDPHYTIM3_THS_EXIT(x) ((x) << 16) +#define DSIDPHYTIM3_THS_TRAIL(x) ((x) << 8) +#define DSIDPHYTIM3_THS_ZERO(x) ((x) << 0) + +/* --------------------------------------------------------*/ +/* Link Registers */ +#define LINK_REG_OFFSET 0x10000 + +/* Link Status Register */ +#define LINKSR 0x10 +#define LINKSR_LPBUSY BIT(13) +#define LINKSR_HSBUSY BIT(12) +#define LINKSR_VICHRUN1 BIT(8) +#define LINKSR_SQCHRUN1 BIT(4) +#define LINKSR_SQCHRUN0 BIT(0) + +/* Tx Set Register */ +#define TXSETR 0x100 +#define TXSETR_NUMLANECAP (0x3 << 16) +#define TXSETR_DLEN (1 << 9) +#define TXSETR_CLEN (1 << 8) +#define TXSETR_NUMLANEUSE(x) (((x) & 0x3) << 0) + +/* HS Clock Set Register */ +#define HSCLKSETR 0x104 +#define HSCLKSETR_HSCLKMODE_CONT (1 << 1) +#define HSCLKSETR_HSCLKMODE_NON_CONT (0 << 1) +#define HSCLKSETR_HSCLKRUN_HS (1 << 0) +#define HSCLKSETR_HSCLKRUN_LP (0 << 0) + +/* Reset Control Register */ +#define RSTCR 0x110 +#define RSTCR_SWRST BIT(0) +#define RSTCR_FCETXSTP BIT(16) + +/* Reset Status Register */ +#define RSTSR 0x114 +#define RSTSR_DL0DIR (1 << 15) +#define RSTSR_DLSTPST (0xf << 8) +#define RSTSR_SWRSTV1 (1 << 4) +#define RSTSR_SWRSTIB (1 << 3) +#define RSTSR_SWRSTAPB (1 << 2) +#define RSTSR_SWRSTLP (1 << 1) +#define RSTSR_SWRSTHS (1 << 0) + +/* Clock Lane Stop Time Set Register */ +#define CLSTPTSETR 0x314 +#define CLSTPTSETR_CLKKPT(x) ((x) << 24) +#define CLSTPTSETR_CLKBFHT(x) ((x) << 16) +#define CLSTPTSETR_CLKSTPT(x) ((x) << 2) + +/* LP Transition Time Set Register */ +#define LPTRNSTSETR 0x318 +#define LPTRNSTSETR_GOLPBKT(x) ((x) << 0) + +/* Physical Lane Status Register */ +#define PLSR 0x320 +#define PLSR_CLHS2LP BIT(27) +#define PLSR_CLLP2HS BIT(26) + +/* Video-Input Channel 1 Set 0 Register */ +#define VICH1SET0R 0x400 +#define VICH1SET0R_VSEN BIT(12) +#define VICH1SET0R_HFPNOLP BIT(10) +#define VICH1SET0R_HBPNOLP BIT(9) +#define VICH1SET0R_HSANOLP BIT(8) +#define VICH1SET0R_VSTPAFT BIT(1) +#define VICH1SET0R_VSTART BIT(0) + +/* Video-Input Channel 1 Set 1 Register */ +#define VICH1SET1R 0x404 +#define VICH1SET1R_DLY(x) (((x) & 0xfff) << 2) + +/* Video-Input Channel 1 Status Register */ +#define VICH1SR 0x410 +#define VICH1SR_VIRDY BIT(3) +#define VICH1SR_RUNNING BIT(2) +#define VICH1SR_STOP BIT(1) +#define VICH1SR_START BIT(0) + +/* Video-Input Channel 1 Pixel Packet Set Register */ +#define VICH1PPSETR 0x420 +#define VICH1PPSETR_DT_RGB18 (0x1e << 16) +#define VICH1PPSETR_DT_RGB18_LS (0x2e << 16) +#define VICH1PPSETR_DT_RGB24 (0x3e << 16) +#define VICH1PPSETR_TXESYNC_PULSE (1 << 15) +#define VICH1PPSETR_VC(x) ((x) << 22) + +/* Video-Input Channel 1 Vertical Size Set Register */ +#define VICH1VSSETR 0x428 +#define VICH1VSSETR_VACTIVE(x) (((x) & 0x7fff) << 16) +#define VICH1VSSETR_VSPOL_LOW (1 << 15) +#define VICH1VSSETR_VSPOL_HIGH (0 << 15) +#define VICH1VSSETR_VSA(x) (((x) & 0xfff) << 0) + +/* Video-Input Channel 1 Vertical Porch Set Register */ +#define VICH1VPSETR 0x42c +#define VICH1VPSETR_VFP(x) (((x) & 0x1fff) << 16) +#define VICH1VPSETR_VBP(x) (((x) & 0x1fff) << 0) + +/* Video-Input Channel 1 Horizontal Size Set Register */ +#define VICH1HSSETR 0x430 +#define VICH1HSSETR_HACTIVE(x) (((x) & 0x7fff) << 16) +#define VICH1HSSETR_HSPOL_LOW (1 << 15) +#define VICH1HSSETR_HSPOL_HIGH (0 << 15) +#define VICH1HSSETR_HSA(x) (((x) & 0xfff) << 0) + +/* Video-Input Channel 1 Horizontal Porch Set Register */ +#define VICH1HPSETR 0x434 +#define VICH1HPSETR_HFP(x) (((x) & 0x1fff) << 16) +#define VICH1HPSETR_HBP(x) (((x) & 0x1fff) << 0) + +#endif /* __RZG2L_MIPI_DSI_REGS_H__ */ diff --git a/drivers/gpu/drm/rockchip/dw-mipi-dsi-rockchip.c b/drivers/gpu/drm/rockchip/dw-mipi-dsi-rockchip.c index 92b599b089f9..7901c3babc8c 100644 --- a/drivers/gpu/drm/rockchip/dw-mipi-dsi-rockchip.c +++ b/drivers/gpu/drm/rockchip/dw-mipi-dsi-rockchip.c @@ -746,7 +746,7 @@ static void dw_mipi_dsi_rockchip_config(struct dw_mipi_dsi_rockchip *dsi) static void dw_mipi_dsi_rockchip_set_lcdsel(struct dw_mipi_dsi_rockchip *dsi, int mux) { - if (dsi->cdata->lcdsel_grf_reg < 0) + if (dsi->cdata->lcdsel_grf_reg) regmap_write(dsi->grf_regmap, dsi->cdata->lcdsel_grf_reg, mux ? dsi->cdata->lcdsel_lit : dsi->cdata->lcdsel_big); } @@ -1045,23 +1045,31 @@ static int dw_mipi_dsi_rockchip_host_attach(void *priv_data, if (ret) { DRM_DEV_ERROR(dsi->dev, "Failed to register component: %d\n", ret); - return ret; + goto out; } second = dw_mipi_dsi_rockchip_find_second(dsi); - if (IS_ERR(second)) - return PTR_ERR(second); + if (IS_ERR(second)) { + ret = PTR_ERR(second); + goto out; + } if (second) { ret = component_add(second, &dw_mipi_dsi_rockchip_ops); if (ret) { DRM_DEV_ERROR(second, "Failed to register component: %d\n", ret); - return ret; + goto out; } } return 0; + +out: + mutex_lock(&dsi->usage_mutex); + dsi->usage_mode = DW_DSI_USAGE_IDLE; + mutex_unlock(&dsi->usage_mutex); + return ret; } static int dw_mipi_dsi_rockchip_host_detach(void *priv_data, @@ -1629,7 +1637,6 @@ static const struct rockchip_dw_dsi_chip_data rk3399_chip_data[] = { static const struct rockchip_dw_dsi_chip_data rk3568_chip_data[] = { { .reg = 0xfe060000, - .lcdsel_grf_reg = -1, .lanecfg1_grf_reg = RK3568_GRF_VO_CON2, .lanecfg1 = HIWORD_UPDATE(0, RK3568_DSI0_SKEWCALHS | RK3568_DSI0_FORCETXSTOPMODE | @@ -1639,7 +1646,6 @@ static const struct rockchip_dw_dsi_chip_data rk3568_chip_data[] = { }, { .reg = 0xfe070000, - .lcdsel_grf_reg = -1, .lanecfg1_grf_reg = RK3568_GRF_VO_CON3, .lanecfg1 = HIWORD_UPDATE(0, RK3568_DSI1_SKEWCALHS | RK3568_DSI1_FORCETXSTOPMODE | @@ -1675,5 +1681,11 @@ struct platform_driver dw_mipi_dsi_rockchip_driver = { .of_match_table = dw_mipi_dsi_rockchip_dt_ids, .pm = &dw_mipi_dsi_rockchip_pm_ops, .name = "dw-mipi-dsi-rockchip", + /* + * For dual-DSI display, one DSI pokes at the other DSI's + * drvdata in dw_mipi_dsi_rockchip_find_second(). This is not + * safe for asynchronous probe. + */ + .probe_type = PROBE_FORCE_SYNCHRONOUS, }, }; diff --git a/drivers/gpu/drm/rockchip/dw_hdmi-rockchip.c b/drivers/gpu/drm/rockchip/dw_hdmi-rockchip.c index c14f88893868..2f4b8f64cbad 100644 --- a/drivers/gpu/drm/rockchip/dw_hdmi-rockchip.c +++ b/drivers/gpu/drm/rockchip/dw_hdmi-rockchip.c @@ -565,7 +565,8 @@ static int dw_hdmi_rockchip_bind(struct device *dev, struct device *master, ret = rockchip_hdmi_parse_dt(hdmi); if (ret) { - DRM_DEV_ERROR(hdmi->dev, "Unable to parse OF data\n"); + if (ret != -EPROBE_DEFER) + DRM_DEV_ERROR(hdmi->dev, "Unable to parse OF data\n"); return ret; } diff --git a/drivers/gpu/drm/rockchip/rockchip_drm_gem.c b/drivers/gpu/drm/rockchip/rockchip_drm_gem.c index bf1120e0f573..6edb7c52cb3d 100644 --- a/drivers/gpu/drm/rockchip/rockchip_drm_gem.c +++ b/drivers/gpu/drm/rockchip/rockchip_drm_gem.c @@ -365,9 +365,12 @@ rockchip_gem_create_with_handle(struct drm_file *file_priv, { struct rockchip_gem_object *rk_obj; struct drm_gem_object *obj; + bool is_framebuffer; int ret; - rk_obj = rockchip_gem_create_object(drm, size, false); + is_framebuffer = drm->fb_helper && file_priv == drm->fb_helper->client.file; + + rk_obj = rockchip_gem_create_object(drm, size, is_framebuffer); if (IS_ERR(rk_obj)) return ERR_CAST(rk_obj); diff --git a/drivers/gpu/drm/rockchip/rockchip_drm_vop2.c b/drivers/gpu/drm/rockchip/rockchip_drm_vop2.c index ad87db2fcaf6..8cecf81a5ae0 100644 --- a/drivers/gpu/drm/rockchip/rockchip_drm_vop2.c +++ b/drivers/gpu/drm/rockchip/rockchip_drm_vop2.c @@ -877,10 +877,14 @@ static void vop2_crtc_atomic_disable(struct drm_crtc *crtc, { struct vop2_video_port *vp = to_vop2_video_port(crtc); struct vop2 *vop2 = vp->vop2; + struct drm_crtc_state *old_crtc_state; int ret; vop2_lock(vop2); + old_crtc_state = drm_atomic_get_old_crtc_state(state, crtc); + drm_atomic_helper_disable_planes_on_crtc(old_crtc_state, false); + drm_crtc_vblank_off(crtc); /* @@ -996,13 +1000,15 @@ static int vop2_plane_atomic_check(struct drm_plane *plane, static void vop2_plane_atomic_disable(struct drm_plane *plane, struct drm_atomic_state *state) { - struct drm_plane_state *old_pstate = drm_atomic_get_old_plane_state(state, plane); + struct drm_plane_state *old_pstate = NULL; struct vop2_win *win = to_vop2_win(plane); struct vop2 *vop2 = win->vop2; drm_dbg(vop2->drm, "%s disable\n", win->data->name); - if (!old_pstate->crtc) + if (state) + old_pstate = drm_atomic_get_old_plane_state(state, plane); + if (old_pstate && !old_pstate->crtc) return; vop2_win_disable(win); diff --git a/drivers/gpu/drm/scheduler/sched_main.c b/drivers/gpu/drm/scheduler/sched_main.c index 857ec20be9e8..fd22d753b4ed 100644 --- a/drivers/gpu/drm/scheduler/sched_main.c +++ b/drivers/gpu/drm/scheduler/sched_main.c @@ -417,27 +417,6 @@ static void drm_sched_job_timedout(struct work_struct *work) } } - /** - * drm_sched_increase_karma - Update sched_entity guilty flag - * - * @bad: The job guilty of time out - * - * Increment on every hang caused by the 'bad' job. If this exceeds the hang - * limit of the scheduler then the respective sched entity is marked guilty and - * jobs from it will not be scheduled further - */ -void drm_sched_increase_karma(struct drm_sched_job *bad) -{ - drm_sched_increase_karma_ext(bad, 1); -} -EXPORT_SYMBOL(drm_sched_increase_karma); - -void drm_sched_reset_karma(struct drm_sched_job *bad) -{ - drm_sched_increase_karma_ext(bad, 0); -} -EXPORT_SYMBOL(drm_sched_reset_karma); - /** * drm_sched_stop - stop the scheduler * @@ -579,31 +558,14 @@ EXPORT_SYMBOL(drm_sched_start); */ void drm_sched_resubmit_jobs(struct drm_gpu_scheduler *sched) { - drm_sched_resubmit_jobs_ext(sched, INT_MAX); -} -EXPORT_SYMBOL(drm_sched_resubmit_jobs); - -/** - * drm_sched_resubmit_jobs_ext - helper to relunch certain number of jobs from mirror ring list - * - * @sched: scheduler instance - * @max: job numbers to relaunch - * - */ -void drm_sched_resubmit_jobs_ext(struct drm_gpu_scheduler *sched, int max) -{ struct drm_sched_job *s_job, *tmp; uint64_t guilty_context; bool found_guilty = false; struct dma_fence *fence; - int i = 0; list_for_each_entry_safe(s_job, tmp, &sched->pending_list, list) { struct drm_sched_fence *s_fence = s_job->s_fence; - if (i >= max) - break; - if (!found_guilty && atomic_read(&s_job->karma) > sched->hang_limit) { found_guilty = true; guilty_context = s_job->s_fence->scheduled.context; @@ -613,7 +575,6 @@ void drm_sched_resubmit_jobs_ext(struct drm_gpu_scheduler *sched, int max) dma_fence_set_error(&s_fence->finished, -ECANCELED); fence = sched->ops->run_job(s_job); - i++; if (IS_ERR_OR_NULL(fence)) { if (IS_ERR(fence)) @@ -629,7 +590,7 @@ void drm_sched_resubmit_jobs_ext(struct drm_gpu_scheduler *sched, int max) } } } -EXPORT_SYMBOL(drm_sched_resubmit_jobs_ext); +EXPORT_SYMBOL(drm_sched_resubmit_jobs); /** * drm_sched_job_init - init a scheduler job @@ -1165,13 +1126,15 @@ void drm_sched_fini(struct drm_gpu_scheduler *sched) EXPORT_SYMBOL(drm_sched_fini); /** - * drm_sched_increase_karma_ext - Update sched_entity guilty flag + * drm_sched_increase_karma - Update sched_entity guilty flag * * @bad: The job guilty of time out - * @type: type for increase/reset karma * + * Increment on every hang caused by the 'bad' job. If this exceeds the hang + * limit of the scheduler then the respective sched entity is marked guilty and + * jobs from it will not be scheduled further */ -void drm_sched_increase_karma_ext(struct drm_sched_job *bad, int type) +void drm_sched_increase_karma(struct drm_sched_job *bad) { int i; struct drm_sched_entity *tmp; @@ -1183,10 +1146,7 @@ void drm_sched_increase_karma_ext(struct drm_sched_job *bad, int type) * corrupt but keep in mind that kernel jobs always considered good. */ if (bad->s_priority != DRM_SCHED_PRIORITY_KERNEL) { - if (type == 0) - atomic_set(&bad->karma, 0); - else if (type == 1) - atomic_inc(&bad->karma); + atomic_inc(&bad->karma); for (i = DRM_SCHED_PRIORITY_MIN; i < DRM_SCHED_PRIORITY_KERNEL; i++) { @@ -1197,7 +1157,7 @@ void drm_sched_increase_karma_ext(struct drm_sched_job *bad, int type) if (bad->s_fence->scheduled.context == entity->fence_context) { if (entity->guilty) - atomic_set(entity->guilty, type); + atomic_set(entity->guilty, 1); break; } } @@ -1207,4 +1167,4 @@ void drm_sched_increase_karma_ext(struct drm_sched_job *bad, int type) } } } -EXPORT_SYMBOL(drm_sched_increase_karma_ext); +EXPORT_SYMBOL(drm_sched_increase_karma); diff --git a/drivers/gpu/drm/tegra/drm.c b/drivers/gpu/drm/tegra/drm.c index 6748ec1e0005..a1f909dac89a 100644 --- a/drivers/gpu/drm/tegra/drm.c +++ b/drivers/gpu/drm/tegra/drm.c @@ -1093,6 +1093,10 @@ static bool host1x_drm_wants_iommu(struct host1x_device *dev) struct host1x *host1x = dev_get_drvdata(dev->dev.parent); struct iommu_domain *domain; + /* Our IOMMU usage policy doesn't currently play well with GART */ + if (of_machine_is_compatible("nvidia,tegra20")) + return false; + /* * If the Tegra DRM clients are backed by an IOMMU, push buffers are * likely to be allocated beyond the 32-bit boundary if sufficient diff --git a/drivers/gpu/drm/vc4/vc4_drv.c b/drivers/gpu/drm/vc4/vc4_drv.c index b66bf7aea632..5990d8f8c363 100644 --- a/drivers/gpu/drm/vc4/vc4_drv.c +++ b/drivers/gpu/drm/vc4/vc4_drv.c @@ -476,7 +476,12 @@ static int __init vc4_drm_register(void) if (ret) return ret; - return platform_driver_register(&vc4_platform_driver); + ret = platform_driver_register(&vc4_platform_driver); + if (ret) + platform_unregister_drivers(component_drivers, + ARRAY_SIZE(component_drivers)); + + return ret; } static void __exit vc4_drm_unregister(void) diff --git a/drivers/gpu/drm/vc4/vc4_hdmi.c b/drivers/gpu/drm/vc4/vc4_hdmi.c index 6b223a5fcf6f..12a00d644b61 100644 --- a/drivers/gpu/drm/vc4/vc4_hdmi.c +++ b/drivers/gpu/drm/vc4/vc4_hdmi.c @@ -349,27 +349,40 @@ static int vc4_hdmi_reset_link(struct drm_connector *connector, return 0; vc4_hdmi = connector_to_vc4_hdmi(connector); - if (!vc4_hdmi_supports_scrambling(vc4_hdmi)) + mutex_lock(&vc4_hdmi->mutex); + + if (!vc4_hdmi_supports_scrambling(vc4_hdmi)) { + mutex_unlock(&vc4_hdmi->mutex); return 0; + } scrambling_needed = vc4_hdmi_mode_needs_scrambling(&vc4_hdmi->saved_adjusted_mode, vc4_hdmi->output_bpc, vc4_hdmi->output_format); - if (!scrambling_needed) + if (!scrambling_needed) { + mutex_unlock(&vc4_hdmi->mutex); return 0; + } if (conn_state->commit && - !try_wait_for_completion(&conn_state->commit->hw_done)) + !try_wait_for_completion(&conn_state->commit->hw_done)) { + mutex_unlock(&vc4_hdmi->mutex); return 0; + } ret = drm_scdc_readb(connector->ddc, SCDC_TMDS_CONFIG, &config); if (ret < 0) { drm_err(drm, "Failed to read TMDS config: %d\n", ret); + mutex_unlock(&vc4_hdmi->mutex); return 0; } - if (!!(config & SCDC_SCRAMBLING_ENABLE) == scrambling_needed) + if (!!(config & SCDC_SCRAMBLING_ENABLE) == scrambling_needed) { + mutex_unlock(&vc4_hdmi->mutex); return 0; + } + + mutex_unlock(&vc4_hdmi->mutex); /* * HDMI 2.0 says that one should not send scrambled data @@ -397,9 +410,8 @@ static void vc4_hdmi_handle_hotplug(struct vc4_hdmi *vc4_hdmi, * .adap_enable, which leads to that funtion being called with * our mutex held. * - * A similar situation occurs with - * drm_atomic_helper_connector_hdmi_reset_link() that will call - * into our KMS hooks if the scrambling was enabled. + * A similar situation occurs with vc4_hdmi_reset_link() that + * will call into our KMS hooks if the scrambling was enabled. * * Concurrency isn't an issue at the moment since we don't share * any state with any of the other frameworks so we can ignore @@ -3169,9 +3181,16 @@ static int vc4_hdmi_init_resources(struct drm_device *drm, DRM_ERROR("Failed to get HDMI state machine clock\n"); return PTR_ERR(vc4_hdmi->hsm_clock); } + vc4_hdmi->audio_clock = vc4_hdmi->hsm_clock; vc4_hdmi->cec_clock = vc4_hdmi->hsm_clock; + vc4_hdmi->hsm_rpm_clock = devm_clk_get(dev, "hdmi"); + if (IS_ERR(vc4_hdmi->hsm_rpm_clock)) { + DRM_ERROR("Failed to get HDMI state machine clock\n"); + return PTR_ERR(vc4_hdmi->hsm_rpm_clock); + } + return 0; } @@ -3254,6 +3273,12 @@ static int vc5_hdmi_init_resources(struct drm_device *drm, return PTR_ERR(vc4_hdmi->hsm_clock); } + vc4_hdmi->hsm_rpm_clock = devm_clk_get(dev, "hdmi"); + if (IS_ERR(vc4_hdmi->hsm_rpm_clock)) { + DRM_ERROR("Failed to get HDMI state machine clock\n"); + return PTR_ERR(vc4_hdmi->hsm_rpm_clock); + } + vc4_hdmi->pixel_bvb_clock = devm_clk_get(dev, "bvb"); if (IS_ERR(vc4_hdmi->pixel_bvb_clock)) { DRM_ERROR("Failed to get pixel bvb clock\n"); @@ -3317,7 +3342,7 @@ static int vc4_hdmi_runtime_suspend(struct device *dev) { struct vc4_hdmi *vc4_hdmi = dev_get_drvdata(dev); - clk_disable_unprepare(vc4_hdmi->hsm_clock); + clk_disable_unprepare(vc4_hdmi->hsm_rpm_clock); return 0; } @@ -3335,11 +3360,11 @@ static int vc4_hdmi_runtime_resume(struct device *dev) * its frequency while the power domain is active so that it * keeps its rate. */ - ret = clk_set_min_rate(vc4_hdmi->hsm_clock, HSM_MIN_CLOCK_FREQ); + ret = clk_set_min_rate(vc4_hdmi->hsm_rpm_clock, HSM_MIN_CLOCK_FREQ); if (ret) return ret; - ret = clk_prepare_enable(vc4_hdmi->hsm_clock); + ret = clk_prepare_enable(vc4_hdmi->hsm_rpm_clock); if (ret) return ret; @@ -3352,7 +3377,7 @@ static int vc4_hdmi_runtime_resume(struct device *dev) * case, it will lead to a silent CPU stall. Let's make sure we * prevent such a case. */ - rate = clk_get_rate(vc4_hdmi->hsm_clock); + rate = clk_get_rate(vc4_hdmi->hsm_rpm_clock); if (!rate) { ret = -EINVAL; goto err_disable_clk; diff --git a/drivers/gpu/drm/vc4/vc4_hdmi.h b/drivers/gpu/drm/vc4/vc4_hdmi.h index e3619836ca17..dc3ccd8002a0 100644 --- a/drivers/gpu/drm/vc4/vc4_hdmi.h +++ b/drivers/gpu/drm/vc4/vc4_hdmi.h @@ -164,6 +164,7 @@ struct vc4_hdmi { struct clk *cec_clock; struct clk *pixel_clock; struct clk *hsm_clock; + struct clk *hsm_rpm_clock; struct clk *audio_clock; struct clk *pixel_bvb_clock; diff --git a/drivers/gpu/drm/vc4/vc4_kms.c b/drivers/gpu/drm/vc4/vc4_kms.c index 5c97642ed66a..8fbeecdf2ec4 100644 --- a/drivers/gpu/drm/vc4/vc4_kms.c +++ b/drivers/gpu/drm/vc4/vc4_kms.c @@ -197,8 +197,8 @@ vc4_hvs_get_new_global_state(struct drm_atomic_state *state) struct drm_private_state *priv_state; priv_state = drm_atomic_get_new_private_obj_state(state, &vc4->hvs_channels); - if (IS_ERR(priv_state)) - return ERR_CAST(priv_state); + if (!priv_state) + return ERR_PTR(-EINVAL); return to_vc4_hvs_state(priv_state); } @@ -210,8 +210,8 @@ vc4_hvs_get_old_global_state(struct drm_atomic_state *state) struct drm_private_state *priv_state; priv_state = drm_atomic_get_old_private_obj_state(state, &vc4->hvs_channels); - if (IS_ERR(priv_state)) - return ERR_CAST(priv_state); + if (!priv_state) + return ERR_PTR(-EINVAL); return to_vc4_hvs_state(priv_state); } diff --git a/drivers/gpu/drm/xlnx/zynqmp_dp.c b/drivers/gpu/drm/xlnx/zynqmp_dp.c index 7c9ae167eac7..0a7b466446fb 100644 --- a/drivers/gpu/drm/xlnx/zynqmp_dp.c +++ b/drivers/gpu/drm/xlnx/zynqmp_dp.c @@ -1362,9 +1362,10 @@ static void zynqmp_dp_bridge_detach(struct drm_bridge *bridge) zynqmp_dp_aux_cleanup(dp); } -static int zynqmp_dp_bridge_mode_valid(struct drm_bridge *bridge, - const struct drm_display_info *info, - const struct drm_display_mode *mode) +static enum drm_mode_status +zynqmp_dp_bridge_mode_valid(struct drm_bridge *bridge, + const struct drm_display_info *info, + const struct drm_display_mode *mode) { struct zynqmp_dp *dp = bridge_to_dp(bridge); int rate; diff --git a/drivers/gpu/host1x/dev.c b/drivers/gpu/host1x/dev.c index 0cd3f97e7e49..f60ea24db0ec 100644 --- a/drivers/gpu/host1x/dev.c +++ b/drivers/gpu/host1x/dev.c @@ -292,6 +292,10 @@ static void host1x_setup_virtualization_tables(struct host1x *host) static bool host1x_wants_iommu(struct host1x *host1x) { + /* Our IOMMU usage policy doesn't currently play well with GART */ + if (of_machine_is_compatible("nvidia,tegra20")) + return false; + /* * If we support addressing a maximum of 32 bits of physical memory * and if the host1x firewall is enabled, there's no need to enable |