diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2023-06-29 21:00:17 +0300 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2023-06-29 21:00:17 +0300 |
commit | 1b722407a13b7f8658d2e26917791f32805980a2 (patch) | |
tree | 30aab582725a46e42843d75e2eb9ce4151f0f3ed /include/uapi | |
parent | f8824e151fbfa0ac0a258015d606ea6f4a10251b (diff) | |
parent | 5ff2977b19769fd24b0cfbe7cbe4d5114b6106af (diff) | |
download | linux-1b722407a13b7f8658d2e26917791f32805980a2.tar.xz |
Merge tag 'drm-next-2023-06-29' of git://anongit.freedesktop.org/drm/drm
Pull drm updates from Dave Airlie:
"There is one set of patches to misc for a i915 gsc/mei proxy driver.
Otherwise it's mostly amdgpu/i915/msm, lots of hw enablement and lots
of refactoring.
core:
- replace strlcpy with strscpy
- EDID changes to support further conversion to struct drm_edid
- Move i915 DSC parameter code to common DRM helpers
- Add Colorspace functionality
aperture:
- ignore framebuffers with non-primary devices
fbdev:
- use fbdev i/o helpers
- add Kconfig options for fb_ops helpers
- use new fb io helpers directly in drivers
sysfs:
- export DRM connector ID
scheduler:
- Avoid an infinite loop
ttm:
- store function table in .rodata
- Add query for TTM mem limit
- Add NUMA awareness to pools
- Export ttm_pool_fini()
bridge:
- fsl-ldb: support i.MX6SX
- lt9211, lt9611: remove blanking packets
- tc358768: implement input bus formats, devm cleanups
- ti-snd65dsi86: implement wait_hpd_asserted
- analogix: fix endless probe loop
- samsung-dsim: support swapped clock, fix enabling, support var
clock
- display-connector: Add support for external power supply
- imx: Fix module linking
- tc358762: Support reset GPIO
panel:
- nt36523: Support Lenovo J606F
- st7703: Support Anbernic RG353V-V2
- InnoLux G070ACE-L01 support
- boe-tv101wum-nl6: Improve initialization
- sharp-ls043t1le001: Mode fixes
- simple: BOE EV121WXM-N10-1850, S6D7AA0
- Ampire AM-800480L1TMQW-T00H
- Rocktech RK043FN48H
- Starry himax83102-j02
- Starry ili9882t
amdgpu:
- add new ctx query flag to handle reset better
- add new query/set shadow buffer for rdna3
- DCN 3.2/3.1.x/3.0.x updates
- Enable DC_FP on loongarch
- PCIe fix for RDNA2
- improve DC FAMS/SubVP support for better power management
- partition support for lots of engines
- Take NUMA into account when allocating memory
- Add new DRM_AMDGPU_WERROR config parameter to help with CI
- Initial SMU13 overdrive support
- Add support for new colorspace KMS API
- W=1 fixes
amdkfd:
- Query TTM mem limit rather than hardcoding it
- GC 9.4.3 partition support
- Handle NUMA for partitions
- Add debugger interface for enabling gdb
- Add KFD event age tracking
radeon:
- Fix possible UAF
i915:
- new getparam for PXP support
- GSC/MEI proxy driver
- Meteorlake display enablement
- avoid clearing preallocated framebuffers with TTM
- implement framebuffer mmap support
- Disable sampler indirect state in bindless heap
- Enable fdinfo for GuC backends
- GuC loading and firmware table handling fixes
- Various refactors for multi-tile enablement
- Define MOCS and PAT tables for MTL
- GSC/MEI support for Meteorlake
- PMU multi-tile support
- Large driver kernel doc cleanup
- Allow VRR toggling and arbitrary refresh rates
- Support async flips on linear buffers on display ver 12+
- Expose CRTC CTM property on ILK/SNB/VLV
- New debugfs for display clock frequencies
- Hotplug refactoring
- Display refactoring
- I915_GEM_CREATE_EXT_SET_PAT for Mesa on Meteorlake
- Use large rings for compute contexts
- HuC loading for MTL
- Allow user to set cache at BO creation
- MTL powermanagement enhancements
- Switch to dedicated workqueues to stop using flush_scheduled_work()
- Move display runtime init under display/
- Remove 10bit gamma on desktop gen3 parts, they don't support it
habanalabs:
- uapi: return 0 for user queries if there was a h/w or f/w error
- Add pci health check when we lose connection with the firmware.
This can be used to distinguish between pci link down and firmware
getting stuck.
- Add more info to the error print when TPC interrupt occur.
- Firmware fixes
msm:
- Adreno A660 bindings
- SM8350 MDSS bindings fix
- Added support for DPU on sm6350 and sm6375 platforms
- Implemented tearcheck support to support vsync on SM150 and newer
platforms
- Enabled missing features (DSPP, DSC, split display) on sc8180x,
sc8280xp, sm8450
- Added support for DSI and 28nm DSI PHY on MSM8226 platform
- Added support for DSI on sm6350 and sm6375 platforms
- Added support for display controller on MSM8226 platform
- A690 GPU support
- Move cmdstream dumping out of fence signaling path
- a610 support
- Support for a6xx devices without GMU
nouveau:
- NULL ptr before deref fixes
armada:
- implement fbdev emulation as client
sun4i:
- fix mipi-dsi dotclock
- release clocks
vc4:
- rgb range toggle property
- BT601 / BT2020 HDMI support
vkms:
- convert to drmm helpers
- add reflection and rotation support
- fix rgb565 conversion
gma500:
- fix iomem access
shmobile:
- support renesas soc platform
- enable fbdev
mxsfb:
- Add support for i.MX93 LCDIF
stm:
- dsi: Use devm_ helper
- ltdc: Fix potential invalid pointer deref
renesas:
- Group drivers in renesas subdirectory to prepare for new platform
- Drop deprecated R-Car H3 ES1.x support
meson:
- Add support for MIPI DSI displays
virtio:
- add sync object support
mediatek:
- Add display binding document for MT6795"
* tag 'drm-next-2023-06-29' of git://anongit.freedesktop.org/drm/drm: (1791 commits)
drm/i915: Fix a NULL vs IS_ERR() bug
drm/i915: make i915_drm_client_fdinfo() reference conditional again
drm/i915/huc: Fix missing error code in intel_huc_init()
drm/i915/gsc: take a wakeref for the proxy-init-completion check
drm/msm/a6xx: Add A610 speedbin support
drm/msm/a6xx: Add A619_holi speedbin support
drm/msm/a6xx: Use adreno_is_aXYZ macros in speedbin matching
drm/msm/a6xx: Use "else if" in GPU speedbin rev matching
drm/msm/a6xx: Fix some A619 tunables
drm/msm/a6xx: Add A610 support
drm/msm/a6xx: Add support for A619_holi
drm/msm/adreno: Disable has_cached_coherent in GMU wrapper configurations
drm/msm/a6xx: Introduce GMU wrapper support
drm/msm/a6xx: Move CX GMU power counter enablement to hw_init
drm/msm/a6xx: Extend and explain UBWC config
drm/msm/a6xx: Remove both GBIF and RBBM GBIF halt on hw init
drm/msm/a6xx: Add a helper for software-resetting the GPU
drm/msm/a6xx: Improve a6xx_bus_clear_pending_transactions()
drm/msm/a6xx: Move a6xx_bus_clear_pending_transactions to a6xx_gpu
drm/msm/a6xx: Move force keepalive vote removal to a6xx_gmu_force_off()
...
Diffstat (limited to 'include/uapi')
-rw-r--r-- | include/uapi/drm/amdgpu_drm.h | 22 | ||||
-rw-r--r-- | include/uapi/drm/drm_fourcc.h | 43 | ||||
-rw-r--r-- | include/uapi/drm/drm_mode.h | 5 | ||||
-rw-r--r-- | include/uapi/drm/habanalabs_accel.h | 10 | ||||
-rw-r--r-- | include/uapi/drm/i915_drm.h | 95 | ||||
-rw-r--r-- | include/uapi/linux/kfd_ioctl.h | 682 | ||||
-rw-r--r-- | include/uapi/linux/kfd_sysfs.h | 15 |
7 files changed, 867 insertions, 5 deletions
diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h index b6eb90df5d05..79b14828d542 100644 --- a/include/uapi/drm/amdgpu_drm.h +++ b/include/uapi/drm/amdgpu_drm.h @@ -245,6 +245,8 @@ union drm_amdgpu_bo_list { /* indicate some errors are detected by RAS */ #define AMDGPU_CTX_QUERY2_FLAGS_RAS_CE (1<<3) #define AMDGPU_CTX_QUERY2_FLAGS_RAS_UE (1<<4) +/* indicate that the reset hasn't completed yet */ +#define AMDGPU_CTX_QUERY2_FLAGS_RESET_IN_PROGRESS (1<<5) /* Context priority level */ #define AMDGPU_CTX_PRIORITY_UNSET -2048 @@ -592,6 +594,7 @@ struct drm_amdgpu_gem_va { #define AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES 0x07 #define AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_WAIT 0x08 #define AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_SIGNAL 0x09 +#define AMDGPU_CHUNK_ID_CP_GFX_SHADOW 0x0a struct drm_amdgpu_cs_chunk { __u32 chunk_id; @@ -708,6 +711,15 @@ struct drm_amdgpu_cs_chunk_data { }; }; +#define AMDGPU_CS_CHUNK_CP_GFX_SHADOW_FLAGS_INIT_SHADOW 0x1 + +struct drm_amdgpu_cs_chunk_cp_gfx_shadow { + __u64 shadow_va; + __u64 csa_va; + __u64 gds_va; + __u64 flags; +}; + /* * Query h/w info: Flag that this is integrated (a.h.a. fusion) GPU * @@ -876,6 +888,8 @@ struct drm_amdgpu_cs_chunk_data { #define AMDGPU_INFO_VIDEO_CAPS_DECODE 0 /* Subquery id: Encode */ #define AMDGPU_INFO_VIDEO_CAPS_ENCODE 1 +/* Query the max number of IBs per gang per submission */ +#define AMDGPU_INFO_MAX_IBS 0x22 #define AMDGPU_INFO_MMR_SE_INDEX_SHIFT 0 #define AMDGPU_INFO_MMR_SE_INDEX_MASK 0xff @@ -1126,6 +1140,14 @@ struct drm_amdgpu_info_device { __u64 mall_size; /* AKA infinity cache */ /* high 32 bits of the rb pipes mask */ __u32 enabled_rb_pipes_mask_hi; + /* shadow area size for gfx11 */ + __u32 shadow_size; + /* shadow area base virtual alignment for gfx11 */ + __u32 shadow_alignment; + /* context save area size for gfx11 */ + __u32 csa_size; + /* context save area base virtual alignment for gfx11 */ + __u32 csa_alignment; }; struct drm_amdgpu_info_hw_ip { diff --git a/include/uapi/drm/drm_fourcc.h b/include/uapi/drm/drm_fourcc.h index de703c6be969..8db7fd3f743e 100644 --- a/include/uapi/drm/drm_fourcc.h +++ b/include/uapi/drm/drm_fourcc.h @@ -658,6 +658,49 @@ extern "C" { #define I915_FORMAT_MOD_4_TILED_DG2_RC_CCS_CC fourcc_mod_code(INTEL, 12) /* + * Intel Color Control Surfaces (CCS) for display ver. 14 render compression. + * + * The main surface is tile4 and at plane index 0, the CCS is linear and + * at index 1. A 64B CCS cache line corresponds to an area of 4x1 tiles in + * main surface. In other words, 4 bits in CCS map to a main surface cache + * line pair. The main surface pitch is required to be a multiple of four + * tile4 widths. + */ +#define I915_FORMAT_MOD_4_TILED_MTL_RC_CCS fourcc_mod_code(INTEL, 13) + +/* + * Intel Color Control Surfaces (CCS) for display ver. 14 media compression + * + * The main surface is tile4 and at plane index 0, the CCS is linear and + * at index 1. A 64B CCS cache line corresponds to an area of 4x1 tiles in + * main surface. In other words, 4 bits in CCS map to a main surface cache + * line pair. The main surface pitch is required to be a multiple of four + * tile4 widths. For semi-planar formats like NV12, CCS planes follow the + * Y and UV planes i.e., planes 0 and 1 are used for Y and UV surfaces, + * planes 2 and 3 for the respective CCS. + */ +#define I915_FORMAT_MOD_4_TILED_MTL_MC_CCS fourcc_mod_code(INTEL, 14) + +/* + * Intel Color Control Surface with Clear Color (CCS) for display ver. 14 render + * compression. + * + * The main surface is tile4 and is at plane index 0 whereas CCS is linear + * and at index 1. The clear color is stored at index 2, and the pitch should + * be ignored. The clear color structure is 256 bits. The first 128 bits + * represents Raw Clear Color Red, Green, Blue and Alpha color each represented + * by 32 bits. The raw clear color is consumed by the 3d engine and generates + * the converted clear color of size 64 bits. The first 32 bits store the Lower + * Converted Clear Color value and the next 32 bits store the Higher Converted + * Clear Color value when applicable. The Converted Clear Color values are + * consumed by the DE. The last 64 bits are used to store Color Discard Enable + * and Depth Clear Value Valid which are ignored by the DE. A CCS cache line + * corresponds to an area of 4x1 tiles in the main surface. The main surface + * pitch is required to be a multiple of 4 tile widths. + */ +#define I915_FORMAT_MOD_4_TILED_MTL_RC_CCS_CC fourcc_mod_code(INTEL, 15) + +/* * Tiled, NV12MT, grouped in 64 (pixels) x 32 (lines) -sized macroblocks * * Macroblocks are laid in a Z-shape, and each pixel data is following the diff --git a/include/uapi/drm/drm_mode.h b/include/uapi/drm/drm_mode.h index 46becedf5b2f..43691058d28f 100644 --- a/include/uapi/drm/drm_mode.h +++ b/include/uapi/drm/drm_mode.h @@ -834,6 +834,11 @@ struct drm_color_ctm { /* * Conversion matrix in S31.32 sign-magnitude * (not two's complement!) format. + * + * out matrix in + * |R| |0 1 2| |R| + * |G| = |3 4 5| x |G| + * |B| |6 7 8| |B| */ __u64 matrix[9]; }; diff --git a/include/uapi/drm/habanalabs_accel.h b/include/uapi/drm/habanalabs_accel.h index d9ef1b151d04..e6436f3e8ea6 100644 --- a/include/uapi/drm/habanalabs_accel.h +++ b/include/uapi/drm/habanalabs_accel.h @@ -787,18 +787,28 @@ enum hl_server_type { * The address which accessing it caused the razwi. * Razwi initiator. * Razwi cause, was it a page fault or MMU access error. + * May return 0 even though no new data is available, in that case + * timestamp will be 0. * HL_INFO_DEV_MEM_ALLOC_PAGE_SIZES - Retrieve valid page sizes for device memory allocation * HL_INFO_SECURED_ATTESTATION - Retrieve attestation report of the boot. * HL_INFO_REGISTER_EVENTFD - Register eventfd for event notifications. * HL_INFO_UNREGISTER_EVENTFD - Unregister eventfd * HL_INFO_GET_EVENTS - Retrieve the last occurred events * HL_INFO_UNDEFINED_OPCODE_EVENT - Retrieve last undefined opcode error information. + * May return 0 even though no new data is available, in that case + * timestamp will be 0. * HL_INFO_ENGINE_STATUS - Retrieve the status of all the h/w engines in the asic. * HL_INFO_PAGE_FAULT_EVENT - Retrieve parameters of captured page fault. + * May return 0 even though no new data is available, in that case + * timestamp will be 0. * HL_INFO_USER_MAPPINGS - Retrieve user mappings, captured after page fault event. * HL_INFO_FW_GENERIC_REQ - Send generic request to FW. * HL_INFO_HW_ERR_EVENT - Retrieve information on the reported HW error. + * May return 0 even though no new data is available, in that case + * timestamp will be 0. * HL_INFO_FW_ERR_EVENT - Retrieve information on the reported FW error. + * May return 0 even though no new data is available, in that case + * timestamp will be 0. */ #define HL_INFO_HW_IP_INFO 0 #define HL_INFO_HW_EVENTS 1 diff --git a/include/uapi/drm/i915_drm.h b/include/uapi/drm/i915_drm.h index dba7c5a5b25e..7000e5910a1d 100644 --- a/include/uapi/drm/i915_drm.h +++ b/include/uapi/drm/i915_drm.h @@ -280,7 +280,16 @@ enum drm_i915_pmu_engine_sample { #define I915_PMU_ENGINE_SEMA(class, instance) \ __I915_PMU_ENGINE(class, instance, I915_SAMPLE_SEMA) -#define __I915_PMU_OTHER(x) (__I915_PMU_ENGINE(0xff, 0xff, 0xf) + 1 + (x)) +/* + * Top 4 bits of every non-engine counter are GT id. + */ +#define __I915_PMU_GT_SHIFT (60) + +#define ___I915_PMU_OTHER(gt, x) \ + (((__u64)__I915_PMU_ENGINE(0xff, 0xff, 0xf) + 1 + (x)) | \ + ((__u64)(gt) << __I915_PMU_GT_SHIFT)) + +#define __I915_PMU_OTHER(x) ___I915_PMU_OTHER(0, x) #define I915_PMU_ACTUAL_FREQUENCY __I915_PMU_OTHER(0) #define I915_PMU_REQUESTED_FREQUENCY __I915_PMU_OTHER(1) @@ -290,6 +299,12 @@ enum drm_i915_pmu_engine_sample { #define I915_PMU_LAST /* Deprecated - do not use */ I915_PMU_RC6_RESIDENCY +#define __I915_PMU_ACTUAL_FREQUENCY(gt) ___I915_PMU_OTHER(gt, 0) +#define __I915_PMU_REQUESTED_FREQUENCY(gt) ___I915_PMU_OTHER(gt, 1) +#define __I915_PMU_INTERRUPTS(gt) ___I915_PMU_OTHER(gt, 2) +#define __I915_PMU_RC6_RESIDENCY(gt) ___I915_PMU_OTHER(gt, 3) +#define __I915_PMU_SOFTWARE_GT_AWAKE_TIME(gt) ___I915_PMU_OTHER(gt, 4) + /* Each region is a minimum of 16k, and there are at most 255 of them. */ #define I915_NR_TEX_REGIONS 255 /* table size 2k - maximum due to use @@ -659,7 +674,8 @@ typedef struct drm_i915_irq_wait { * If the IOCTL is successful, the returned parameter will be set to one of the * following values: * * 0 if HuC firmware load is not complete, - * * 1 if HuC firmware is authenticated and running. + * * 1 if HuC firmware is loaded and fully authenticated, + * * 2 if HuC firmware is loaded and authenticated for clear media only */ #define I915_PARAM_HUC_STATUS 42 @@ -771,6 +787,25 @@ typedef struct drm_i915_irq_wait { */ #define I915_PARAM_OA_TIMESTAMP_FREQUENCY 57 +/* + * Query the status of PXP support in i915. + * + * The query can fail in the following scenarios with the listed error codes: + * -ENODEV = PXP support is not available on the GPU device or in the + * kernel due to missing component drivers or kernel configs. + * + * If the IOCTL is successful, the returned parameter will be set to one of + * the following values: + * 1 = PXP feature is supported and is ready for use. + * 2 = PXP feature is supported but should be ready soon (pending + * initialization of non-i915 system dependencies). + * + * NOTE: When param is supported (positive return values), user space should + * still refer to the GEM PXP context-creation UAPI header specs to be + * aware of possible failure due to system state machine at the time. + */ +#define I915_PARAM_PXP_STATUS 58 + /* Must be kept compact -- no holes and well documented */ /** @@ -2096,6 +2131,21 @@ struct drm_i915_gem_context_param { * * -ENODEV: feature not available * -EPERM: trying to mark a recoverable or not bannable context as protected + * -ENXIO: A dependency such as a component driver or firmware is not yet + * loaded so user space may need to attempt again. Depending on the + * device, this error may be reported if protected context creation is + * attempted very early after kernel start because the internal timeout + * waiting for such dependencies is not guaranteed to be larger than + * required (numbers differ depending on system and kernel config): + * - ADL/RPL: dependencies may take up to 3 seconds from kernel start + * while context creation internal timeout is 250 milisecs + * - MTL: dependencies may take up to 8 seconds from kernel start + * while context creation internal timeout is 250 milisecs + * NOTE: such dependencies happen once, so a subsequent call to create a + * protected context after a prior successful call will not experience + * such timeouts and will not return -ENXIO (unless the driver is reloaded, + * or, depending on the device, resumes from a suspended state). + * -EIO: The firmware did not succeed in creating the protected context. */ #define I915_CONTEXT_PARAM_PROTECTED_CONTENT 0xd /* Must be kept compact -- no holes and well documented */ @@ -3630,9 +3680,13 @@ struct drm_i915_gem_create_ext { * * For I915_GEM_CREATE_EXT_PROTECTED_CONTENT usage see * struct drm_i915_gem_create_ext_protected_content. + * + * For I915_GEM_CREATE_EXT_SET_PAT usage see + * struct drm_i915_gem_create_ext_set_pat. */ #define I915_GEM_CREATE_EXT_MEMORY_REGIONS 0 #define I915_GEM_CREATE_EXT_PROTECTED_CONTENT 1 +#define I915_GEM_CREATE_EXT_SET_PAT 2 __u64 extensions; }; @@ -3747,6 +3801,43 @@ struct drm_i915_gem_create_ext_protected_content { __u32 flags; }; +/** + * struct drm_i915_gem_create_ext_set_pat - The + * I915_GEM_CREATE_EXT_SET_PAT extension. + * + * If this extension is provided, the specified caching policy (PAT index) is + * applied to the buffer object. + * + * Below is an example on how to create an object with specific caching policy: + * + * .. code-block:: C + * + * struct drm_i915_gem_create_ext_set_pat set_pat_ext = { + * .base = { .name = I915_GEM_CREATE_EXT_SET_PAT }, + * .pat_index = 0, + * }; + * struct drm_i915_gem_create_ext create_ext = { + * .size = PAGE_SIZE, + * .extensions = (uintptr_t)&set_pat_ext, + * }; + * + * int err = ioctl(fd, DRM_IOCTL_I915_GEM_CREATE_EXT, &create_ext); + * if (err) ... + */ +struct drm_i915_gem_create_ext_set_pat { + /** @base: Extension link. See struct i915_user_extension. */ + struct i915_user_extension base; + /** + * @pat_index: PAT index to be set + * PAT index is a bit field in Page Table Entry to control caching + * behaviors for GPU accesses. The definition of PAT index is + * platform dependent and can be found in hardware specifications, + */ + __u32 pat_index; + /** @rsvd: reserved for future use */ + __u32 rsvd; +}; + /* ID of the protected content session managed by i915 when PXP is active */ #define I915_PROTECTED_CONTENT_DEFAULT_SESSION 0xf diff --git a/include/uapi/linux/kfd_ioctl.h b/include/uapi/linux/kfd_ioctl.h index 2da5c3ad71bd..eeb2fdcbdcb7 100644 --- a/include/uapi/linux/kfd_ioctl.h +++ b/include/uapi/linux/kfd_ioctl.h @@ -38,9 +38,11 @@ * - 1.10 - Add SMI profiler event log * - 1.11 - Add unified memory for ctx save/restore area * - 1.12 - Add DMA buf export ioctl + * - 1.13 - Add debugger API + * - 1.14 - Update kfd_event_data */ #define KFD_IOCTL_MAJOR_VERSION 1 -#define KFD_IOCTL_MINOR_VERSION 12 +#define KFD_IOCTL_MINOR_VERSION 14 struct kfd_ioctl_get_version_args { __u32 major_version; /* from KFD */ @@ -110,6 +112,32 @@ struct kfd_ioctl_get_available_memory_args { __u32 pad; }; +struct kfd_dbg_device_info_entry { + __u64 exception_status; + __u64 lds_base; + __u64 lds_limit; + __u64 scratch_base; + __u64 scratch_limit; + __u64 gpuvm_base; + __u64 gpuvm_limit; + __u32 gpu_id; + __u32 location_id; + __u32 vendor_id; + __u32 device_id; + __u32 revision_id; + __u32 subsystem_vendor_id; + __u32 subsystem_device_id; + __u32 fw_version; + __u32 gfx_target_version; + __u32 simd_count; + __u32 max_waves_per_simd; + __u32 array_count; + __u32 simd_arrays_per_engine; + __u32 num_xcc; + __u32 capability; + __u32 debug_prop; +}; + /* For kfd_ioctl_set_memory_policy_args.default_policy and alternate_policy */ #define KFD_IOC_CACHE_POLICY_COHERENT 0 #define KFD_IOC_CACHE_POLICY_NONCOHERENT 1 @@ -293,12 +321,20 @@ struct kfd_hsa_hw_exception_data { __u32 gpu_id; }; +/* hsa signal event data */ +struct kfd_hsa_signal_event_data { + __u64 last_event_age; /* to and from KFD */ +}; + /* Event data */ struct kfd_event_data { union { + /* From KFD */ struct kfd_hsa_memory_exception_data memory_exception_data; struct kfd_hsa_hw_exception_data hw_exception_data; - }; /* From KFD */ + /* To and From KFD */ + struct kfd_hsa_signal_event_data signal_event_data; + }; __u64 kfd_event_data_ext; /* pointer to an extension structure for future exception types */ __u32 event_id; /* to KFD */ @@ -773,6 +809,640 @@ struct kfd_ioctl_set_xnack_mode_args { __s32 xnack_enabled; }; +/* Wave launch override modes */ +enum kfd_dbg_trap_override_mode { + KFD_DBG_TRAP_OVERRIDE_OR = 0, + KFD_DBG_TRAP_OVERRIDE_REPLACE = 1 +}; + +/* Wave launch overrides */ +enum kfd_dbg_trap_mask { + KFD_DBG_TRAP_MASK_FP_INVALID = 1, + KFD_DBG_TRAP_MASK_FP_INPUT_DENORMAL = 2, + KFD_DBG_TRAP_MASK_FP_DIVIDE_BY_ZERO = 4, + KFD_DBG_TRAP_MASK_FP_OVERFLOW = 8, + KFD_DBG_TRAP_MASK_FP_UNDERFLOW = 16, + KFD_DBG_TRAP_MASK_FP_INEXACT = 32, + KFD_DBG_TRAP_MASK_INT_DIVIDE_BY_ZERO = 64, + KFD_DBG_TRAP_MASK_DBG_ADDRESS_WATCH = 128, + KFD_DBG_TRAP_MASK_DBG_MEMORY_VIOLATION = 256, + KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_START = (1 << 30), + KFD_DBG_TRAP_MASK_TRAP_ON_WAVE_END = (1 << 31) +}; + +/* Wave launch modes */ +enum kfd_dbg_trap_wave_launch_mode { + KFD_DBG_TRAP_WAVE_LAUNCH_MODE_NORMAL = 0, + KFD_DBG_TRAP_WAVE_LAUNCH_MODE_HALT = 1, + KFD_DBG_TRAP_WAVE_LAUNCH_MODE_DEBUG = 3 +}; + +/* Address watch modes */ +enum kfd_dbg_trap_address_watch_mode { + KFD_DBG_TRAP_ADDRESS_WATCH_MODE_READ = 0, + KFD_DBG_TRAP_ADDRESS_WATCH_MODE_NONREAD = 1, + KFD_DBG_TRAP_ADDRESS_WATCH_MODE_ATOMIC = 2, + KFD_DBG_TRAP_ADDRESS_WATCH_MODE_ALL = 3 +}; + +/* Additional wave settings */ +enum kfd_dbg_trap_flags { + KFD_DBG_TRAP_FLAG_SINGLE_MEM_OP = 1, +}; + +/* Trap exceptions */ +enum kfd_dbg_trap_exception_code { + EC_NONE = 0, + /* per queue */ + EC_QUEUE_WAVE_ABORT = 1, + EC_QUEUE_WAVE_TRAP = 2, + EC_QUEUE_WAVE_MATH_ERROR = 3, + EC_QUEUE_WAVE_ILLEGAL_INSTRUCTION = 4, + EC_QUEUE_WAVE_MEMORY_VIOLATION = 5, + EC_QUEUE_WAVE_APERTURE_VIOLATION = 6, + EC_QUEUE_PACKET_DISPATCH_DIM_INVALID = 16, + EC_QUEUE_PACKET_DISPATCH_GROUP_SEGMENT_SIZE_INVALID = 17, + EC_QUEUE_PACKET_DISPATCH_CODE_INVALID = 18, + EC_QUEUE_PACKET_RESERVED = 19, + EC_QUEUE_PACKET_UNSUPPORTED = 20, + EC_QUEUE_PACKET_DISPATCH_WORK_GROUP_SIZE_INVALID = 21, + EC_QUEUE_PACKET_DISPATCH_REGISTER_INVALID = 22, + EC_QUEUE_PACKET_VENDOR_UNSUPPORTED = 23, + EC_QUEUE_PREEMPTION_ERROR = 30, + EC_QUEUE_NEW = 31, + /* per device */ + EC_DEVICE_QUEUE_DELETE = 32, + EC_DEVICE_MEMORY_VIOLATION = 33, + EC_DEVICE_RAS_ERROR = 34, + EC_DEVICE_FATAL_HALT = 35, + EC_DEVICE_NEW = 36, + /* per process */ + EC_PROCESS_RUNTIME = 48, + EC_PROCESS_DEVICE_REMOVE = 49, + EC_MAX +}; + +/* Mask generated by ecode in kfd_dbg_trap_exception_code */ +#define KFD_EC_MASK(ecode) (1ULL << (ecode - 1)) + +/* Masks for exception code type checks below */ +#define KFD_EC_MASK_QUEUE (KFD_EC_MASK(EC_QUEUE_WAVE_ABORT) | \ + KFD_EC_MASK(EC_QUEUE_WAVE_TRAP) | \ + KFD_EC_MASK(EC_QUEUE_WAVE_MATH_ERROR) | \ + KFD_EC_MASK(EC_QUEUE_WAVE_ILLEGAL_INSTRUCTION) | \ + KFD_EC_MASK(EC_QUEUE_WAVE_MEMORY_VIOLATION) | \ + KFD_EC_MASK(EC_QUEUE_WAVE_APERTURE_VIOLATION) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_DISPATCH_DIM_INVALID) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_DISPATCH_GROUP_SEGMENT_SIZE_INVALID) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_DISPATCH_CODE_INVALID) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_RESERVED) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_UNSUPPORTED) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_DISPATCH_WORK_GROUP_SIZE_INVALID) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_DISPATCH_REGISTER_INVALID) | \ + KFD_EC_MASK(EC_QUEUE_PACKET_VENDOR_UNSUPPORTED) | \ + KFD_EC_MASK(EC_QUEUE_PREEMPTION_ERROR) | \ + KFD_EC_MASK(EC_QUEUE_NEW)) +#define KFD_EC_MASK_DEVICE (KFD_EC_MASK(EC_DEVICE_QUEUE_DELETE) | \ + KFD_EC_MASK(EC_DEVICE_RAS_ERROR) | \ + KFD_EC_MASK(EC_DEVICE_FATAL_HALT) | \ + KFD_EC_MASK(EC_DEVICE_MEMORY_VIOLATION) | \ + KFD_EC_MASK(EC_DEVICE_NEW)) +#define KFD_EC_MASK_PROCESS (KFD_EC_MASK(EC_PROCESS_RUNTIME) | \ + KFD_EC_MASK(EC_PROCESS_DEVICE_REMOVE)) + +/* Checks for exception code types for KFD search */ +#define KFD_DBG_EC_TYPE_IS_QUEUE(ecode) \ + (!!(KFD_EC_MASK(ecode) & KFD_EC_MASK_QUEUE)) +#define KFD_DBG_EC_TYPE_IS_DEVICE(ecode) \ + (!!(KFD_EC_MASK(ecode) & KFD_EC_MASK_DEVICE)) +#define KFD_DBG_EC_TYPE_IS_PROCESS(ecode) \ + (!!(KFD_EC_MASK(ecode) & KFD_EC_MASK_PROCESS)) + + +/* Runtime enable states */ +enum kfd_dbg_runtime_state { + DEBUG_RUNTIME_STATE_DISABLED = 0, + DEBUG_RUNTIME_STATE_ENABLED = 1, + DEBUG_RUNTIME_STATE_ENABLED_BUSY = 2, + DEBUG_RUNTIME_STATE_ENABLED_ERROR = 3 +}; + +/* Runtime enable status */ +struct kfd_runtime_info { + __u64 r_debug; + __u32 runtime_state; + __u32 ttmp_setup; +}; + +/* Enable modes for runtime enable */ +#define KFD_RUNTIME_ENABLE_MODE_ENABLE_MASK 1 +#define KFD_RUNTIME_ENABLE_MODE_TTMP_SAVE_MASK 2 + +/** + * kfd_ioctl_runtime_enable_args - Arguments for runtime enable + * + * Coordinates debug exception signalling and debug device enablement with runtime. + * + * @r_debug - pointer to user struct for sharing information between ROCr and the debuggger + * @mode_mask - mask to set mode + * KFD_RUNTIME_ENABLE_MODE_ENABLE_MASK - enable runtime for debugging, otherwise disable + * KFD_RUNTIME_ENABLE_MODE_TTMP_SAVE_MASK - enable trap temporary setup (ignore on disable) + * @capabilities_mask - mask to notify runtime on what KFD supports + * + * Return - 0 on SUCCESS. + * - EBUSY if runtime enable call already pending. + * - EEXIST if user queues already active prior to call. + * If process is debug enabled, runtime enable will enable debug devices and + * wait for debugger process to send runtime exception EC_PROCESS_RUNTIME + * to unblock - see kfd_ioctl_dbg_trap_args. + * + */ +struct kfd_ioctl_runtime_enable_args { + __u64 r_debug; + __u32 mode_mask; + __u32 capabilities_mask; +}; + +/* Queue information */ +struct kfd_queue_snapshot_entry { + __u64 exception_status; + __u64 ring_base_address; + __u64 write_pointer_address; + __u64 read_pointer_address; + __u64 ctx_save_restore_address; + __u32 queue_id; + __u32 gpu_id; + __u32 ring_size; + __u32 queue_type; + __u32 ctx_save_restore_area_size; + __u32 reserved; +}; + +/* Queue status return for suspend/resume */ +#define KFD_DBG_QUEUE_ERROR_BIT 30 +#define KFD_DBG_QUEUE_INVALID_BIT 31 +#define KFD_DBG_QUEUE_ERROR_MASK (1 << KFD_DBG_QUEUE_ERROR_BIT) +#define KFD_DBG_QUEUE_INVALID_MASK (1 << KFD_DBG_QUEUE_INVALID_BIT) + +/* Context save area header information */ +struct kfd_context_save_area_header { + struct { + __u32 control_stack_offset; + __u32 control_stack_size; + __u32 wave_state_offset; + __u32 wave_state_size; + } wave_state; + __u32 debug_offset; + __u32 debug_size; + __u64 err_payload_addr; + __u32 err_event_id; + __u32 reserved1; +}; + +/* + * Debug operations + * + * For specifics on usage and return values, see documentation per operation + * below. Otherwise, generic error returns apply: + * - ESRCH if the process to debug does not exist. + * + * - EINVAL (with KFD_IOC_DBG_TRAP_ENABLE exempt) if operation + * KFD_IOC_DBG_TRAP_ENABLE has not succeeded prior. + * Also returns this error if GPU hardware scheduling is not supported. + * + * - EPERM (with KFD_IOC_DBG_TRAP_DISABLE exempt) if target process is not + * PTRACE_ATTACHED. KFD_IOC_DBG_TRAP_DISABLE is exempt to allow + * clean up of debug mode as long as process is debug enabled. + * + * - EACCES if any DBG_HW_OP (debug hardware operation) is requested when + * AMDKFD_IOC_RUNTIME_ENABLE has not succeeded prior. + * + * - ENODEV if any GPU does not support debugging on a DBG_HW_OP call. + * + * - Other errors may be returned when a DBG_HW_OP occurs while the GPU + * is in a fatal state. + * + */ +enum kfd_dbg_trap_operations { + KFD_IOC_DBG_TRAP_ENABLE = 0, + KFD_IOC_DBG_TRAP_DISABLE = 1, + KFD_IOC_DBG_TRAP_SEND_RUNTIME_EVENT = 2, + KFD_IOC_DBG_TRAP_SET_EXCEPTIONS_ENABLED = 3, + KFD_IOC_DBG_TRAP_SET_WAVE_LAUNCH_OVERRIDE = 4, /* DBG_HW_OP */ + KFD_IOC_DBG_TRAP_SET_WAVE_LAUNCH_MODE = 5, /* DBG_HW_OP */ + KFD_IOC_DBG_TRAP_SUSPEND_QUEUES = 6, /* DBG_HW_OP */ + KFD_IOC_DBG_TRAP_RESUME_QUEUES = 7, /* DBG_HW_OP */ + KFD_IOC_DBG_TRAP_SET_NODE_ADDRESS_WATCH = 8, /* DBG_HW_OP */ + KFD_IOC_DBG_TRAP_CLEAR_NODE_ADDRESS_WATCH = 9, /* DBG_HW_OP */ + KFD_IOC_DBG_TRAP_SET_FLAGS = 10, + KFD_IOC_DBG_TRAP_QUERY_DEBUG_EVENT = 11, + KFD_IOC_DBG_TRAP_QUERY_EXCEPTION_INFO = 12, + KFD_IOC_DBG_TRAP_GET_QUEUE_SNAPSHOT = 13, + KFD_IOC_DBG_TRAP_GET_DEVICE_SNAPSHOT = 14 +}; + +/** + * kfd_ioctl_dbg_trap_enable_args + * + * Arguments for KFD_IOC_DBG_TRAP_ENABLE. + * + * Enables debug session for target process. Call @op KFD_IOC_DBG_TRAP_DISABLE in + * kfd_ioctl_dbg_trap_args to disable debug session. + * + * @exception_mask (IN) - exceptions to raise to the debugger + * @rinfo_ptr (IN) - pointer to runtime info buffer (see kfd_runtime_info) + * @rinfo_size (IN/OUT) - size of runtime info buffer in bytes + * @dbg_fd (IN) - fd the KFD will nofify the debugger with of raised + * exceptions set in exception_mask. + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on SUCCESS. + * Copies KFD saved kfd_runtime_info to @rinfo_ptr on enable. + * Size of kfd_runtime saved by the KFD returned to @rinfo_size. + * - EBADF if KFD cannot get a reference to dbg_fd. + * - EFAULT if KFD cannot copy runtime info to rinfo_ptr. + * - EINVAL if target process is already debug enabled. + * + */ +struct kfd_ioctl_dbg_trap_enable_args { + __u64 exception_mask; + __u64 rinfo_ptr; + __u32 rinfo_size; + __u32 dbg_fd; +}; + +/** + * kfd_ioctl_dbg_trap_send_runtime_event_args + * + * + * Arguments for KFD_IOC_DBG_TRAP_SEND_RUNTIME_EVENT. + * Raises exceptions to runtime. + * + * @exception_mask (IN) - exceptions to raise to runtime + * @gpu_id (IN) - target device id + * @queue_id (IN) - target queue id + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on SUCCESS. + * - ENODEV if gpu_id not found. + * If exception_mask contains EC_PROCESS_RUNTIME, unblocks pending + * AMDKFD_IOC_RUNTIME_ENABLE call - see kfd_ioctl_runtime_enable_args. + * All other exceptions are raised to runtime through err_payload_addr. + * See kfd_context_save_area_header. + */ +struct kfd_ioctl_dbg_trap_send_runtime_event_args { + __u64 exception_mask; + __u32 gpu_id; + __u32 queue_id; +}; + +/** + * kfd_ioctl_dbg_trap_set_exceptions_enabled_args + * + * Arguments for KFD_IOC_SET_EXCEPTIONS_ENABLED + * Set new exceptions to be raised to the debugger. + * + * @exception_mask (IN) - new exceptions to raise the debugger + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on SUCCESS. + */ +struct kfd_ioctl_dbg_trap_set_exceptions_enabled_args { + __u64 exception_mask; +}; + +/** + * kfd_ioctl_dbg_trap_set_wave_launch_override_args + * + * Arguments for KFD_IOC_DBG_TRAP_SET_WAVE_LAUNCH_OVERRIDE + * Enable HW exceptions to raise trap. + * + * @override_mode (IN) - see kfd_dbg_trap_override_mode + * @enable_mask (IN/OUT) - reference kfd_dbg_trap_mask. + * IN is the override modes requested to be enabled. + * OUT is referenced in Return below. + * @support_request_mask (IN/OUT) - reference kfd_dbg_trap_mask. + * IN is the override modes requested for support check. + * OUT is referenced in Return below. + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on SUCCESS. + * Previous enablement is returned in @enable_mask. + * Actual override support is returned in @support_request_mask. + * - EINVAL if override mode is not supported. + * - EACCES if trap support requested is not actually supported. + * i.e. enable_mask (IN) is not a subset of support_request_mask (OUT). + * Otherwise it is considered a generic error (see kfd_dbg_trap_operations). + */ +struct kfd_ioctl_dbg_trap_set_wave_launch_override_args { + __u32 override_mode; + __u32 enable_mask; + __u32 support_request_mask; + __u32 pad; +}; + +/** + * kfd_ioctl_dbg_trap_set_wave_launch_mode_args + * + * Arguments for KFD_IOC_DBG_TRAP_SET_WAVE_LAUNCH_MODE + * Set wave launch mode. + * + * @mode (IN) - see kfd_dbg_trap_wave_launch_mode + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on SUCCESS. + */ +struct kfd_ioctl_dbg_trap_set_wave_launch_mode_args { + __u32 launch_mode; + __u32 pad; +}; + +/** + * kfd_ioctl_dbg_trap_suspend_queues_ags + * + * Arguments for KFD_IOC_DBG_TRAP_SUSPEND_QUEUES + * Suspend queues. + * + * @exception_mask (IN) - raised exceptions to clear + * @queue_array_ptr (IN) - pointer to array of queue ids (u32 per queue id) + * to suspend + * @num_queues (IN) - number of queues to suspend in @queue_array_ptr + * @grace_period (IN) - wave time allowance before preemption + * per 1K GPU clock cycle unit + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Destruction of a suspended queue is blocked until the queue is + * resumed. This allows the debugger to access queue information and + * the its context save area without running into a race condition on + * queue destruction. + * Automatically copies per queue context save area header information + * into the save area base + * (see kfd_queue_snapshot_entry and kfd_context_save_area_header). + * + * Return - Number of queues suspended on SUCCESS. + * . KFD_DBG_QUEUE_ERROR_MASK and KFD_DBG_QUEUE_INVALID_MASK masked + * for each queue id in @queue_array_ptr array reports unsuccessful + * suspend reason. + * KFD_DBG_QUEUE_ERROR_MASK = HW failure. + * KFD_DBG_QUEUE_INVALID_MASK = queue does not exist, is new or + * is being destroyed. + */ +struct kfd_ioctl_dbg_trap_suspend_queues_args { + __u64 exception_mask; + __u64 queue_array_ptr; + __u32 num_queues; + __u32 grace_period; +}; + +/** + * kfd_ioctl_dbg_trap_resume_queues_args + * + * Arguments for KFD_IOC_DBG_TRAP_RESUME_QUEUES + * Resume queues. + * + * @queue_array_ptr (IN) - pointer to array of queue ids (u32 per queue id) + * to resume + * @num_queues (IN) - number of queues to resume in @queue_array_ptr + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - Number of queues resumed on SUCCESS. + * KFD_DBG_QUEUE_ERROR_MASK and KFD_DBG_QUEUE_INVALID_MASK mask + * for each queue id in @queue_array_ptr array reports unsuccessful + * resume reason. + * KFD_DBG_QUEUE_ERROR_MASK = HW failure. + * KFD_DBG_QUEUE_INVALID_MASK = queue does not exist. + */ +struct kfd_ioctl_dbg_trap_resume_queues_args { + __u64 queue_array_ptr; + __u32 num_queues; + __u32 pad; +}; + +/** + * kfd_ioctl_dbg_trap_set_node_address_watch_args + * + * Arguments for KFD_IOC_DBG_TRAP_SET_NODE_ADDRESS_WATCH + * Sets address watch for device. + * + * @address (IN) - watch address to set + * @mode (IN) - see kfd_dbg_trap_address_watch_mode + * @mask (IN) - watch address mask + * @gpu_id (IN) - target gpu to set watch point + * @id (OUT) - watch id allocated + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on SUCCESS. + * Allocated watch ID returned to @id. + * - ENODEV if gpu_id not found. + * - ENOMEM if watch IDs can be allocated + */ +struct kfd_ioctl_dbg_trap_set_node_address_watch_args { + __u64 address; + __u32 mode; + __u32 mask; + __u32 gpu_id; + __u32 id; +}; + +/** + * kfd_ioctl_dbg_trap_clear_node_address_watch_args + * + * Arguments for KFD_IOC_DBG_TRAP_CLEAR_NODE_ADDRESS_WATCH + * Clear address watch for device. + * + * @gpu_id (IN) - target device to clear watch point + * @id (IN) - allocated watch id to clear + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on SUCCESS. + * - ENODEV if gpu_id not found. + * - EINVAL if watch ID has not been allocated. + */ +struct kfd_ioctl_dbg_trap_clear_node_address_watch_args { + __u32 gpu_id; + __u32 id; +}; + +/** + * kfd_ioctl_dbg_trap_set_flags_args + * + * Arguments for KFD_IOC_DBG_TRAP_SET_FLAGS + * Sets flags for wave behaviour. + * + * @flags (IN/OUT) - IN = flags to enable, OUT = flags previously enabled + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on SUCCESS. + * - EACCESS if any debug device does not allow flag options. + */ +struct kfd_ioctl_dbg_trap_set_flags_args { + __u32 flags; + __u32 pad; +}; + +/** + * kfd_ioctl_dbg_trap_query_debug_event_args + * + * Arguments for KFD_IOC_DBG_TRAP_QUERY_DEBUG_EVENT + * + * Find one or more raised exceptions. This function can return multiple + * exceptions from a single queue or a single device with one call. To find + * all raised exceptions, this function must be called repeatedly until it + * returns -EAGAIN. Returned exceptions can optionally be cleared by + * setting the corresponding bit in the @exception_mask input parameter. + * However, clearing an exception prevents retrieving further information + * about it with KFD_IOC_DBG_TRAP_QUERY_EXCEPTION_INFO. + * + * @exception_mask (IN/OUT) - exception to clear (IN) and raised (OUT) + * @gpu_id (OUT) - gpu id of exceptions raised + * @queue_id (OUT) - queue id of exceptions raised + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on raised exception found + * Raised exceptions found are returned in @exception mask + * with reported source id returned in @gpu_id or @queue_id. + * - EAGAIN if no raised exception has been found + */ +struct kfd_ioctl_dbg_trap_query_debug_event_args { + __u64 exception_mask; + __u32 gpu_id; + __u32 queue_id; +}; + +/** + * kfd_ioctl_dbg_trap_query_exception_info_args + * + * Arguments KFD_IOC_DBG_TRAP_QUERY_EXCEPTION_INFO + * Get additional info on raised exception. + * + * @info_ptr (IN) - pointer to exception info buffer to copy to + * @info_size (IN/OUT) - exception info buffer size (bytes) + * @source_id (IN) - target gpu or queue id + * @exception_code (IN) - target exception + * @clear_exception (IN) - clear raised @exception_code exception + * (0 = false, 1 = true) + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on SUCCESS. + * If @exception_code is EC_DEVICE_MEMORY_VIOLATION, copy @info_size(OUT) + * bytes of memory exception data to @info_ptr. + * If @exception_code is EC_PROCESS_RUNTIME, copy saved + * kfd_runtime_info to @info_ptr. + * Actual required @info_ptr size (bytes) is returned in @info_size. + */ +struct kfd_ioctl_dbg_trap_query_exception_info_args { + __u64 info_ptr; + __u32 info_size; + __u32 source_id; + __u32 exception_code; + __u32 clear_exception; +}; + +/** + * kfd_ioctl_dbg_trap_get_queue_snapshot_args + * + * Arguments KFD_IOC_DBG_TRAP_GET_QUEUE_SNAPSHOT + * Get queue information. + * + * @exception_mask (IN) - exceptions raised to clear + * @snapshot_buf_ptr (IN) - queue snapshot entry buffer (see kfd_queue_snapshot_entry) + * @num_queues (IN/OUT) - number of queue snapshot entries + * The debugger specifies the size of the array allocated in @num_queues. + * KFD returns the number of queues that actually existed. If this is + * larger than the size specified by the debugger, KFD will not overflow + * the array allocated by the debugger. + * + * @entry_size (IN/OUT) - size per entry in bytes + * The debugger specifies sizeof(struct kfd_queue_snapshot_entry) in + * @entry_size. KFD returns the number of bytes actually populated per + * entry. The debugger should use the KFD_IOCTL_MINOR_VERSION to determine, + * which fields in struct kfd_queue_snapshot_entry are valid. This allows + * growing the ABI in a backwards compatible manner. + * Note that entry_size(IN) should still be used to stride the snapshot buffer in the + * event that it's larger than actual kfd_queue_snapshot_entry. + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on SUCCESS. + * Copies @num_queues(IN) queue snapshot entries of size @entry_size(IN) + * into @snapshot_buf_ptr if @num_queues(IN) > 0. + * Otherwise return @num_queues(OUT) queue snapshot entries that exist. + */ +struct kfd_ioctl_dbg_trap_queue_snapshot_args { + __u64 exception_mask; + __u64 snapshot_buf_ptr; + __u32 num_queues; + __u32 entry_size; +}; + +/** + * kfd_ioctl_dbg_trap_get_device_snapshot_args + * + * Arguments for KFD_IOC_DBG_TRAP_GET_DEVICE_SNAPSHOT + * Get device information. + * + * @exception_mask (IN) - exceptions raised to clear + * @snapshot_buf_ptr (IN) - pointer to snapshot buffer (see kfd_dbg_device_info_entry) + * @num_devices (IN/OUT) - number of debug devices to snapshot + * The debugger specifies the size of the array allocated in @num_devices. + * KFD returns the number of devices that actually existed. If this is + * larger than the size specified by the debugger, KFD will not overflow + * the array allocated by the debugger. + * + * @entry_size (IN/OUT) - size per entry in bytes + * The debugger specifies sizeof(struct kfd_dbg_device_info_entry) in + * @entry_size. KFD returns the number of bytes actually populated. The + * debugger should use KFD_IOCTL_MINOR_VERSION to determine, which fields + * in struct kfd_dbg_device_info_entry are valid. This allows growing the + * ABI in a backwards compatible manner. + * Note that entry_size(IN) should still be used to stride the snapshot buffer in the + * event that it's larger than actual kfd_dbg_device_info_entry. + * + * Generic errors apply (see kfd_dbg_trap_operations). + * Return - 0 on SUCCESS. + * Copies @num_devices(IN) device snapshot entries of size @entry_size(IN) + * into @snapshot_buf_ptr if @num_devices(IN) > 0. + * Otherwise return @num_devices(OUT) queue snapshot entries that exist. + */ +struct kfd_ioctl_dbg_trap_device_snapshot_args { + __u64 exception_mask; + __u64 snapshot_buf_ptr; + __u32 num_devices; + __u32 entry_size; +}; + +/** + * kfd_ioctl_dbg_trap_args + * + * Arguments to debug target process. + * + * @pid - target process to debug + * @op - debug operation (see kfd_dbg_trap_operations) + * + * @op determines which union struct args to use. + * Refer to kern docs for each kfd_ioctl_dbg_trap_*_args struct. + */ +struct kfd_ioctl_dbg_trap_args { + __u32 pid; + __u32 op; + + union { + struct kfd_ioctl_dbg_trap_enable_args enable; + struct kfd_ioctl_dbg_trap_send_runtime_event_args send_runtime_event; + struct kfd_ioctl_dbg_trap_set_exceptions_enabled_args set_exceptions_enabled; + struct kfd_ioctl_dbg_trap_set_wave_launch_override_args launch_override; + struct kfd_ioctl_dbg_trap_set_wave_launch_mode_args launch_mode; + struct kfd_ioctl_dbg_trap_suspend_queues_args suspend_queues; + struct kfd_ioctl_dbg_trap_resume_queues_args resume_queues; + struct kfd_ioctl_dbg_trap_set_node_address_watch_args set_node_address_watch; + struct kfd_ioctl_dbg_trap_clear_node_address_watch_args clear_node_address_watch; + struct kfd_ioctl_dbg_trap_set_flags_args set_flags; + struct kfd_ioctl_dbg_trap_query_debug_event_args query_debug_event; + struct kfd_ioctl_dbg_trap_query_exception_info_args query_exception_info; + struct kfd_ioctl_dbg_trap_queue_snapshot_args queue_snapshot; + struct kfd_ioctl_dbg_trap_device_snapshot_args device_snapshot; + }; +}; + #define AMDKFD_IOCTL_BASE 'K' #define AMDKFD_IO(nr) _IO(AMDKFD_IOCTL_BASE, nr) #define AMDKFD_IOR(nr, type) _IOR(AMDKFD_IOCTL_BASE, nr, type) @@ -887,7 +1557,13 @@ struct kfd_ioctl_set_xnack_mode_args { #define AMDKFD_IOC_EXPORT_DMABUF \ AMDKFD_IOWR(0x24, struct kfd_ioctl_export_dmabuf_args) +#define AMDKFD_IOC_RUNTIME_ENABLE \ + AMDKFD_IOWR(0x25, struct kfd_ioctl_runtime_enable_args) + +#define AMDKFD_IOC_DBG_TRAP \ + AMDKFD_IOWR(0x26, struct kfd_ioctl_dbg_trap_args) + #define AMDKFD_COMMAND_START 0x01 -#define AMDKFD_COMMAND_END 0x25 +#define AMDKFD_COMMAND_END 0x27 #endif diff --git a/include/uapi/linux/kfd_sysfs.h b/include/uapi/linux/kfd_sysfs.h index 3e330f368917..a51b7331e0b4 100644 --- a/include/uapi/linux/kfd_sysfs.h +++ b/include/uapi/linux/kfd_sysfs.h @@ -43,6 +43,11 @@ #define HSA_CAP_DOORBELL_TYPE_2_0 0x2 #define HSA_CAP_AQL_QUEUE_DOUBLE_MAP 0x00004000 +#define HSA_CAP_TRAP_DEBUG_SUPPORT 0x00008000 +#define HSA_CAP_TRAP_DEBUG_WAVE_LAUNCH_TRAP_OVERRIDE_SUPPORTED 0x00010000 +#define HSA_CAP_TRAP_DEBUG_WAVE_LAUNCH_MODE_SUPPORTED 0x00020000 +#define HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED 0x00040000 + /* Old buggy user mode depends on this being 0 */ #define HSA_CAP_RESERVED_WAS_SRAM_EDCSUPPORTED 0x00080000 @@ -53,8 +58,18 @@ #define HSA_CAP_SRAM_EDCSUPPORTED 0x04000000 #define HSA_CAP_SVMAPI_SUPPORTED 0x08000000 #define HSA_CAP_FLAGS_COHERENTHOSTACCESS 0x10000000 +#define HSA_CAP_TRAP_DEBUG_FIRMWARE_SUPPORTED 0x20000000 #define HSA_CAP_RESERVED 0xe00f8000 +/* debug_prop bits in node properties */ +#define HSA_DBG_WATCH_ADDR_MASK_LO_BIT_MASK 0x0000000f +#define HSA_DBG_WATCH_ADDR_MASK_LO_BIT_SHIFT 0 +#define HSA_DBG_WATCH_ADDR_MASK_HI_BIT_MASK 0x000003f0 +#define HSA_DBG_WATCH_ADDR_MASK_HI_BIT_SHIFT 4 +#define HSA_DBG_DISPATCH_INFO_ALWAYS_VALID 0x00000400 +#define HSA_DBG_WATCHPOINTS_EXCLUSIVE 0x00000800 +#define HSA_DBG_RESERVED 0xfffffffffffff000ull + /* Heap types in memory properties */ #define HSA_MEM_HEAP_TYPE_SYSTEM 0 #define HSA_MEM_HEAP_TYPE_FB_PUBLIC 1 |