diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c')
-rw-r--r-- | drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c | 255 |
1 files changed, 114 insertions, 141 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 3ebbddb63705..e1531d97f486 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -379,41 +379,6 @@ static const uint32_t ecc_umc_mcumc_ctrl_mask_addrs[] = { (0x001d43e0 + 0x00001800), }; -static const uint32_t ecc_umc_mcumc_status_addrs[] = { - (0x000143c2 + 0x00000000), - (0x000143c2 + 0x00000800), - (0x000143c2 + 0x00001000), - (0x000143c2 + 0x00001800), - (0x000543c2 + 0x00000000), - (0x000543c2 + 0x00000800), - (0x000543c2 + 0x00001000), - (0x000543c2 + 0x00001800), - (0x000943c2 + 0x00000000), - (0x000943c2 + 0x00000800), - (0x000943c2 + 0x00001000), - (0x000943c2 + 0x00001800), - (0x000d43c2 + 0x00000000), - (0x000d43c2 + 0x00000800), - (0x000d43c2 + 0x00001000), - (0x000d43c2 + 0x00001800), - (0x001143c2 + 0x00000000), - (0x001143c2 + 0x00000800), - (0x001143c2 + 0x00001000), - (0x001143c2 + 0x00001800), - (0x001543c2 + 0x00000000), - (0x001543c2 + 0x00000800), - (0x001543c2 + 0x00001000), - (0x001543c2 + 0x00001800), - (0x001943c2 + 0x00000000), - (0x001943c2 + 0x00000800), - (0x001943c2 + 0x00001000), - (0x001943c2 + 0x00001800), - (0x001d43c2 + 0x00000000), - (0x001d43c2 + 0x00000800), - (0x001d43c2 + 0x00001000), - (0x001d43c2 + 0x00001800), -}; - static int gmc_v9_0_ecc_interrupt_state(struct amdgpu_device *adev, struct amdgpu_irq_src *src, unsigned type, @@ -502,6 +467,7 @@ static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev, WREG32(reg, tmp); } } + break; default: break; } @@ -510,122 +476,136 @@ static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev, } static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, - struct amdgpu_irq_src *source, - struct amdgpu_iv_entry *entry) + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) { - struct amdgpu_vmhub *hub; bool retry_fault = !!(entry->src_data[1] & 0x80); uint32_t status = 0, cid = 0, rw = 0; - u64 addr; - char hub_name[10]; + struct amdgpu_task_info task_info; + struct amdgpu_vmhub *hub; const char *mmhub_cid; + const char *hub_name; + u64 addr; addr = (u64)entry->src_data[0] << 12; addr |= ((u64)entry->src_data[1] & 0xf) << 44; - if (retry_fault && amdgpu_gmc_filter_faults(adev, addr, entry->pasid, - entry->timestamp)) - return 1; /* This also prevents sending it to KFD */ + if (retry_fault) { + /* Returning 1 here also prevents sending the IV to the KFD */ + + /* Process it onyl if it's the first fault for this address */ + if (entry->ih != &adev->irq.ih_soft && + amdgpu_gmc_filter_faults(adev, addr, entry->pasid, + entry->timestamp)) + return 1; + + /* Delegate it to a different ring if the hardware hasn't + * already done it. + */ + if (in_interrupt()) { + amdgpu_irq_delegate(adev, entry, 8); + return 1; + } + + /* Try to handle the recoverable page faults by filling page + * tables + */ + if (amdgpu_vm_handle_fault(adev, entry->pasid, addr)) + return 1; + } + + if (!printk_ratelimit()) + return 0; if (entry->client_id == SOC15_IH_CLIENTID_VMC) { - snprintf(hub_name, sizeof(hub_name), "mmhub0"); + hub_name = "mmhub0"; hub = &adev->vmhub[AMDGPU_MMHUB_0]; } else if (entry->client_id == SOC15_IH_CLIENTID_VMC1) { - snprintf(hub_name, sizeof(hub_name), "mmhub1"); + hub_name = "mmhub1"; hub = &adev->vmhub[AMDGPU_MMHUB_1]; } else { - snprintf(hub_name, sizeof(hub_name), "gfxhub0"); + hub_name = "gfxhub0"; hub = &adev->vmhub[AMDGPU_GFXHUB_0]; } - /* If it's the first fault for this address, process it normally */ - if (retry_fault && !in_interrupt() && - amdgpu_vm_handle_fault(adev, entry->pasid, addr)) - return 1; /* This also prevents sending it to KFD */ + memset(&task_info, 0, sizeof(struct amdgpu_task_info)); + amdgpu_vm_get_task_info(adev, entry->pasid, &task_info); - if (!amdgpu_sriov_vf(adev)) { - /* - * Issue a dummy read to wait for the status register to - * be updated to avoid reading an incorrect value due to - * the new fast GRBM interface. - */ - if (entry->vmid_src == AMDGPU_GFXHUB_0) - RREG32(hub->vm_l2_pro_fault_status); - - status = RREG32(hub->vm_l2_pro_fault_status); - cid = REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, CID); - rw = REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, RW); - WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1); - } + dev_err(adev->dev, + "[%s] %s page fault (src_id:%u ring:%u vmid:%u " + "pasid:%u, for process %s pid %d thread %s pid %d)\n", + hub_name, retry_fault ? "retry" : "no-retry", + entry->src_id, entry->ring_id, entry->vmid, + entry->pasid, task_info.process_name, task_info.tgid, + task_info.task_name, task_info.pid); + dev_err(adev->dev, " in page starting at address 0x%012llx from client %d\n", + addr, entry->client_id); - if (printk_ratelimit()) { - struct amdgpu_task_info task_info; - - memset(&task_info, 0, sizeof(struct amdgpu_task_info)); - amdgpu_vm_get_task_info(adev, entry->pasid, &task_info); - - dev_err(adev->dev, - "[%s] %s page fault (src_id:%u ring:%u vmid:%u " - "pasid:%u, for process %s pid %d thread %s pid %d)\n", - hub_name, retry_fault ? "retry" : "no-retry", - entry->src_id, entry->ring_id, entry->vmid, - entry->pasid, task_info.process_name, task_info.tgid, - task_info.task_name, task_info.pid); - dev_err(adev->dev, " in page starting at address 0x%016llx from client %d\n", - addr, entry->client_id); - if (!amdgpu_sriov_vf(adev)) { - dev_err(adev->dev, - "VM_L2_PROTECTION_FAULT_STATUS:0x%08X\n", - status); - if (hub == &adev->vmhub[AMDGPU_GFXHUB_0]) { - dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n", - cid >= ARRAY_SIZE(gfxhub_client_ids) ? "unknown" : gfxhub_client_ids[cid], - cid); - } else { - switch (adev->asic_type) { - case CHIP_VEGA10: - mmhub_cid = mmhub_client_ids_vega10[cid][rw]; - break; - case CHIP_VEGA12: - mmhub_cid = mmhub_client_ids_vega12[cid][rw]; - break; - case CHIP_VEGA20: - mmhub_cid = mmhub_client_ids_vega20[cid][rw]; - break; - case CHIP_ARCTURUS: - mmhub_cid = mmhub_client_ids_arcturus[cid][rw]; - break; - case CHIP_RAVEN: - mmhub_cid = mmhub_client_ids_raven[cid][rw]; - break; - case CHIP_RENOIR: - mmhub_cid = mmhub_client_ids_renoir[cid][rw]; - break; - default: - mmhub_cid = NULL; - break; - } - dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n", - mmhub_cid ? mmhub_cid : "unknown", cid); - } - dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n", - REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS)); - dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n", - REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR)); - dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n", - REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS)); - dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n", - REG_GET_FIELD(status, - VM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR)); - dev_err(adev->dev, "\t RW: 0x%x\n", rw); + if (amdgpu_sriov_vf(adev)) + return 0; + + /* + * Issue a dummy read to wait for the status register to + * be updated to avoid reading an incorrect value due to + * the new fast GRBM interface. + */ + if (entry->vmid_src == AMDGPU_GFXHUB_0) + RREG32(hub->vm_l2_pro_fault_status); + + status = RREG32(hub->vm_l2_pro_fault_status); + cid = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, CID); + rw = REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, RW); + WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1); + + + dev_err(adev->dev, + "VM_L2_PROTECTION_FAULT_STATUS:0x%08X\n", + status); + if (hub == &adev->vmhub[AMDGPU_GFXHUB_0]) { + dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n", + cid >= ARRAY_SIZE(gfxhub_client_ids) ? "unknown" : + gfxhub_client_ids[cid], + cid); + } else { + switch (adev->asic_type) { + case CHIP_VEGA10: + mmhub_cid = mmhub_client_ids_vega10[cid][rw]; + break; + case CHIP_VEGA12: + mmhub_cid = mmhub_client_ids_vega12[cid][rw]; + break; + case CHIP_VEGA20: + mmhub_cid = mmhub_client_ids_vega20[cid][rw]; + break; + case CHIP_ARCTURUS: + mmhub_cid = mmhub_client_ids_arcturus[cid][rw]; + break; + case CHIP_RAVEN: + mmhub_cid = mmhub_client_ids_raven[cid][rw]; + break; + case CHIP_RENOIR: + mmhub_cid = mmhub_client_ids_renoir[cid][rw]; + break; + default: + mmhub_cid = NULL; + break; } + dev_err(adev->dev, "\t Faulty UTCL2 client ID: %s (0x%x)\n", + mmhub_cid ? mmhub_cid : "unknown", cid); } - + dev_err(adev->dev, "\t MORE_FAULTS: 0x%lx\n", + REG_GET_FIELD(status, + VM_L2_PROTECTION_FAULT_STATUS, MORE_FAULTS)); + dev_err(adev->dev, "\t WALKER_ERROR: 0x%lx\n", + REG_GET_FIELD(status, + VM_L2_PROTECTION_FAULT_STATUS, WALKER_ERROR)); + dev_err(adev->dev, "\t PERMISSION_FAULTS: 0x%lx\n", + REG_GET_FIELD(status, + VM_L2_PROTECTION_FAULT_STATUS, PERMISSION_FAULTS)); + dev_err(adev->dev, "\t MAPPING_ERROR: 0x%lx\n", + REG_GET_FIELD(status, + VM_L2_PROTECTION_FAULT_STATUS, MAPPING_ERROR)); + dev_err(adev->dev, "\t RW: 0x%x\n", rw); return 0; } @@ -711,6 +691,7 @@ static bool gmc_v9_0_get_atc_vmid_pasid_mapping_info(struct amdgpu_device *adev, * * @adev: amdgpu_device pointer * @vmid: vm instance to flush + * @vmhub: which hub to flush * @flush_type: the flush type * * Flush the TLB for the requested page table using certain type. @@ -827,6 +808,8 @@ static void gmc_v9_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, * * @adev: amdgpu_device pointer * @pasid: pasid to be flush + * @flush_type: the flush type + * @all_hub: flush all hubs * * Flush the TLB for the requested pasid. */ @@ -1166,15 +1149,7 @@ static void gmc_v9_0_set_mmhub_funcs(struct amdgpu_device *adev) static void gmc_v9_0_set_gfxhub_funcs(struct amdgpu_device *adev) { - switch (adev->asic_type) { - case CHIP_ARCTURUS: - case CHIP_VEGA20: - adev->gfxhub.funcs = &gfxhub_v1_1_funcs; - break; - default: - adev->gfxhub.funcs = &gfxhub_v1_0_funcs; - break; - } + adev->gfxhub.funcs = &gfxhub_v1_0_funcs; } static int gmc_v9_0_early_init(void *handle) @@ -1202,8 +1177,6 @@ static int gmc_v9_0_late_init(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; int r; - amdgpu_bo_late_init(adev); - r = amdgpu_gmc_allocate_vm_inv_eng(adev); if (r) return r; |