History

Merge tag 'drm-next-2022-03-24' of git://anongit.freedesktop.org/drm/drm

Pull drm updates from Dave Airlie: "Lots of work all over, Intel improving DG2 support, amdkfd CRIU support, msm new hw support, and faster fbdev support. dma-buf: - rename dma-buf-map to iosys-map core: - move buddy allocator to core - add pci/platform init macros - improve EDID parser deep color handling - EDID timing type 7 support - add GPD Win Max quirk - add yes/no helpers to string_helpers - flatten syncobj chains - add nomodeset support to lots of drivers - improve fb-helper clipping support - add default property value interface fbdev: - improve fbdev ops speed ttm: - add a backpointer from ttm bo->ttm resource dp: - move displayport headers - add a dp helper module bridge: - anx7625 atomic support, HDCP support panel: - split out panel-lvds and lvds bindings - find panels in OF subnodes privacy: - add chromeos privacy screen support fb: - hot unplug fw fb on forced removal simpledrm: - request region instead of marking ioresource busy - add panel oreintation property udmabuf: - fix oops with 0 pages amdgpu: - power management code cleanup - Enable freesync video mode by default - RAS code cleanup - Improve VRAM access for debug using SDMA - SR-IOV rework special register access and fixes - profiling power state request ioctl - expose IP discovery via sysfs - Cyan skillfish updates - GC 10.3.7, SDMA 5.2.7, DCN 3.1.6 updates - expose benchmark tests via debugfs - add module param to disable XGMI for testing - GPU reset debugfs register dumping support amdkfd: - CRIU support - SDMA queue fixes radeon: - UVD suspend fix - iMac backlight fix i915: - minimal parallel submission for execlists - DG2-G12 subplatform added - DG2 programming workarounds - DG2 accelerated migration support - flat CCS and CCS engine support for XeHP - initial small BAR support - drop fake LMEM support - ADL-N PCH support - bigjoiner updates - introduce VMA resources and async unbinding - register definitions cleanups - multi-FBC refactoring - DG1 OPROM over SPI support - ADL-N platform enabling - opregion mailbox #5 support - DP MST ESI improvements - drm device based logging - async flip optimisation for DG2 - CPU arch abstraction fixes - improve GuC ADS init to work on aarch64 - tweak TTM LRU priority hint - GuC 69.0.3 support - remove short term execbuf pins nouveau: - higher DP/eDP bitrates - backlight fixes msm: - dpu + dp support for sc8180x - dp support for sm8350 - dpu + dsi support for qcm2290 - 10nm dsi phy tuning support - bridge support for dp encoder - gpu support for additional 7c3 SKUs ingenic: - HDMI support for JZ4780 - aux channel EDID support ast: - AST2600 support - add wide screen support - create DP/DVI connectors omapdrm: - fix implicit dma_buf fencing vc4: - add CSC + full range support - better display firmware handoff panfrost: - add initial dual-core GPU support stm: - new revision support - fb handover support mediatek: - transfer display binding document to yaml format. - add mt8195 display device binding. - allow commands to be sent during video mode. - add wait_for_event for crtc disable by cmdq. tegra: - YUV format support rcar-du: - LVDS support for M3-W+ (R8A77961) exynos: - BGR pixel format for FIMD device" * tag 'drm-next-2022-03-24' of git://anongit.freedesktop.org/drm/drm: (1529 commits) drm/i915/display: Do not re-enable PSR after it was marked as not reliable drm/i915/display: Fix HPD short pulse handling for eDP drm/amdgpu: Use drm_mode_copy() drm/radeon: Use drm_mode_copy() drm/amdgpu: Use ternary operator in `vcn_v1_0_start()` drm/amdgpu: Remove pointless on stack mode copies drm/amd/pm: fix indenting in __smu_cmn_reg_print_error() drm/amdgpu/dc: fix typos in comments drm/amdgpu: fix typos in comments drm/amd/pm: fix typos in comments drm/amdgpu: Add stolen reserved memory for MI25 SRIOV. drm/amdgpu: Merge get_reserved_allocation to get_vbios_allocations. drm/amdkfd: evict svm bo worker handle error drm/amdgpu/vcn: fix vcn ring test failure in igt reload test drm/amdgpu: only allow secure submission on rings which support that drm/amdgpu: fixed the warnings reported by kernel test robot drm/amd/display: 3.2.177 drm/amd/display: [FW Promotion] Release 0.0.108.0 drm/amd/display: Add save/restore PANEL_PWRSEQ_REF_DIV2 drm/amd/display: Wait for hubp read line for Pollock ...

Linus Torvalds committed 4 years ago

Commit b14ffae2 parents 52deda9 c6e90a11466 files · +487504 −33843Browse files

Showing the changes to include/uapi (6 of 1466 files). Show all

Unified · Split

6 files
  1. include/uapi/drm/amdgpu_drm.h +20 −1
  2. include/uapi/drm/drm_mode.h +60 −28
  3. include/uapi/drm/i915_drm.h +40 −5
  4. include/uapi/drm/msm_drm.h +19 −13
  5. include/uapi/drm/panfrost_drm.h +2 −2
  6. include/uapi/linux/kfd_ioctl.h +91 −7
include/uapi/drm/amdgpu_drm.h+20−1
@@ -206,6 +206,8 @@ union drm_amdgpu_bo_list { · 205 unmodified lines
206206 #define AMDGPU_CTX_OP_FREE_CTX 2
207207 #define AMDGPU_CTX_OP_QUERY_STATE 3
208208 #define AMDGPU_CTX_OP_QUERY_STATE2 4
209+#define AMDGPU_CTX_OP_GET_STABLE_PSTATE 5
210+#define AMDGPU_CTX_OP_SET_STABLE_PSTATE 6
209211
210212 /* GPU reset status */
211213 #define AMDGPU_CTX_NO_RESET 0
@@ -238,10 +240,18 @@ union drm_amdgpu_bo_list { · 26 unmodified lines
238240 #define AMDGPU_CTX_PRIORITY_HIGH 512
239241 #define AMDGPU_CTX_PRIORITY_VERY_HIGH 1023
240242
243+/* select a stable profiling pstate for perfmon tools */
244+#define AMDGPU_CTX_STABLE_PSTATE_FLAGS_MASK 0xf
245+#define AMDGPU_CTX_STABLE_PSTATE_NONE 0
246+#define AMDGPU_CTX_STABLE_PSTATE_STANDARD 1
247+#define AMDGPU_CTX_STABLE_PSTATE_MIN_SCLK 2
248+#define AMDGPU_CTX_STABLE_PSTATE_MIN_MCLK 3
249+#define AMDGPU_CTX_STABLE_PSTATE_PEAK 4
250+
241251 struct drm_amdgpu_ctx_in {
242252 /** AMDGPU_CTX_OP_* */
243253 __u32 op;
244− /** For future use, no flags defined so far */
254+ /** Flags */
245255 __u32 flags;
246256 __u32 ctx_id;
247257 /** AMDGPU_CTX_PRIORITY_* */
@@ -262,6 +272,11 @@ union drm_amdgpu_ctx_out { · 14 unmodified lines
262272 /** Reset status since the last call of the ioctl. */
263273 __u32 reset_status;
264274 } state;
275+
276+ struct {
277+ __u32 flags;
278+ __u32 _pad;
279+ } pstate;
265280 };
266281
267282 union drm_amdgpu_ctx {
@@ -728,6 +743,8 @@ struct drm_amdgpu_cs_chunk_data { · 460 unmodified lines
728743 #define AMDGPU_INFO_FW_DMCUB 0x14
729744 /* Subquery id: Query TOC firmware version */
730745 #define AMDGPU_INFO_FW_TOC 0x15
746+ /* Subquery id: Query CAP firmware version */
747+ #define AMDGPU_INFO_FW_CAP 0x16
731748
732749 /* number of bytes moved for TTM migration */
733750 #define AMDGPU_INFO_NUM_BYTES_MOVED 0x0f
@@ -1134,6 +1151,8 @@ struct drm_amdgpu_info_video_caps { · 400 unmodified lines
11341151 #define AMDGPU_FAMILY_NV 143 /* Navi10 */
11351152 #define AMDGPU_FAMILY_VGH 144 /* Van Gogh */
11361153 #define AMDGPU_FAMILY_YC 146 /* Yellow Carp */
1154+#define AMDGPU_FAMILY_GC_10_3_6 149 /* GC 10.3.6 */
1155+#define AMDGPU_FAMILY_GC_10_3_7 151 /* GC 10.3.7 */
11371156
11381157 #if defined(__cplusplus)
11391158 }
include/uapi/drm/drm_mode.h+60−28
@@ -663,41 +663,73 @@ struct drm_mode_fb_cmd { · 662 unmodified lines
663663 #define DRM_MODE_FB_INTERLACED (1<<0) /* for interlaced framebuffers */
664664 #define DRM_MODE_FB_MODIFIERS (1<<1) /* enables ->modifer[] */
665665
666+/**
667+ * struct drm_mode_fb_cmd2 - Frame-buffer metadata.
668+ *
669+ * This struct holds frame-buffer metadata. There are two ways to use it:
670+ *
671+ * - User-space can fill this struct and perform a &DRM_IOCTL_MODE_ADDFB2
672+ * ioctl to register a new frame-buffer. The new frame-buffer object ID will
673+ * be set by the kernel in @fb_id.
674+ * - User-space can set @fb_id and perform a &DRM_IOCTL_MODE_GETFB2 ioctl to
675+ * fetch metadata about an existing frame-buffer.
676+ *
677+ * In case of planar formats, this struct allows up to 4 buffer objects with
678+ * offsets and pitches per plane. The pitch and offset order is dictated by the
679+ * format FourCC as defined by ``drm_fourcc.h``, e.g. NV12 is described as:
680+ *
681+ * YUV 4:2:0 image with a plane of 8 bit Y samples followed by an
682+ * interleaved U/V plane containing 8 bit 2x2 subsampled colour difference
683+ * samples.
684+ *
685+ * So it would consist of a Y plane at ``offsets[0]`` and a UV plane at
686+ * ``offsets[1]``.
687+ *
688+ * To accommodate tiled, compressed, etc formats, a modifier can be specified.
689+ * For more information see the "Format Modifiers" section. Note that even
690+ * though it looks like we have a modifier per-plane, we in fact do not. The
691+ * modifier for each plane must be identical. Thus all combinations of
692+ * different data layouts for multi-plane formats must be enumerated as
693+ * separate modifiers.
694+ *
695+ * All of the entries in @handles, @pitches, @offsets and @modifier must be
696+ * zero when unused. Warning, for @offsets and @modifier zero can't be used to
697+ * figure out whether the entry is used or not since it's a valid value (a zero
698+ * offset is common, and a zero modifier is &DRM_FORMAT_MOD_LINEAR).
699+ */
666700 struct drm_mode_fb_cmd2 {
701+ /** @fb_id: Object ID of the frame-buffer. */
667702 __u32 fb_id;
703+ /** @width: Width of the frame-buffer. */
668704 __u32 width;
705+ /** @height: Height of the frame-buffer. */
669706 __u32 height;
670− __u32 pixel_format; /* fourcc code from drm_fourcc.h */
671− __u32 flags; /* see above flags */
707+ /**
708+ * @pixel_format: FourCC format code, see ``DRM_FORMAT_*`` constants in
709+ * ``drm_fourcc.h``.
710+ */
711+ __u32 pixel_format;
712+ /**
713+ * @flags: Frame-buffer flags (see &DRM_MODE_FB_INTERLACED and
714+ * &DRM_MODE_FB_MODIFIERS).
715+ */
716+ __u32 flags;
672717
673− /*
674− * In case of planar formats, this ioctl allows up to 4
675− * buffer objects with offsets and pitches per plane.
676− * The pitch and offset order is dictated by the fourcc,
677− * e.g. NV12 (https://fourcc.org/yuv.php#NV12) is described as:
678− *
679− * YUV 4:2:0 image with a plane of 8 bit Y samples
680− * followed by an interleaved U/V plane containing
681− * 8 bit 2x2 subsampled colour difference samples.
682− *
683− * So it would consist of Y as offsets[0] and UV as
684− * offsets[1]. Note that offsets[0] will generally
685− * be 0 (but this is not required).
686− *
687− * To accommodate tiled, compressed, etc formats, a
688− * modifier can be specified. The default value of zero
689− * indicates "native" format as specified by the fourcc.
690− * Vendor specific modifier token. Note that even though
691− * it looks like we have a modifier per-plane, we in fact
692− * do not. The modifier for each plane must be identical.
693− * Thus all combinations of different data layouts for
694− * multi plane formats must be enumerated as separate
695− * modifiers.
718+ /**
719+ * @handles: GEM buffer handle, one per plane. Set to 0 if the plane is
720+ * unused. The same handle can be used for multiple planes.
696721 */
697722 __u32 handles[4];
698− __u32 pitches[4]; /* pitch for each plane */
699− __u32 offsets[4]; /* offset of each plane */
700− __u64 modifier[4]; /* ie, tiling, compress */
723+ /** @pitches: Pitch (aka. stride) in bytes, one per plane. */
724+ __u32 pitches[4];
725+ /** @offsets: Offset into the buffer in bytes, one per plane. */
726+ __u32 offsets[4];
727+ /**
728+ * @modifier: Format modifier, one per plane. See ``DRM_FORMAT_MOD_*``
729+ * constants in ``drm_fourcc.h``. All planes must use the same
730+ * modifier. Ignored unless &DRM_MODE_FB_MODIFIERS is set in @flags.
731+ */
732+ __u64 modifier[4];
701733 };
702734
703735 #define DRM_MODE_FB_DIRTY_ANNOTATE_COPY 0x01
include/uapi/drm/i915_drm.h+40−5
@@ -1118,10 +1118,16 @@ struct drm_i915_gem_exec_object2 { · 1117 unmodified lines
11181118 /**
11191119 * When the EXEC_OBJECT_PINNED flag is specified this is populated by
11201120 * the user with the GTT offset at which this object will be pinned.
1121+ *
11211122 * When the I915_EXEC_NO_RELOC flag is specified this must contain the
11221123 * presumed_offset of the object.
1124+ *
11231125 * During execbuffer2 the kernel populates it with the value of the
11241126 * current GTT offset of the object, for future presumed_offset writes.
1127+ *
1128+ * See struct drm_i915_gem_create_ext for the rules when dealing with
1129+ * alignment restrictions with I915_MEMORY_CLASS_DEVICE, on devices with
1130+ * minimum page sizes, like DG2.
11251131 */
11261132 __u64 offset;
11271133
@@ -3144,11 +3150,40 @@ struct drm_i915_gem_create_ext { · 2016 unmodified lines
31443150 *
31453151 * The (page-aligned) allocated size for the object will be returned.
31463152 *
3147− * Note that for some devices we have might have further minimum
3148− * page-size restrictions(larger than 4K), like for device local-memory.
3149− * However in general the final size here should always reflect any
3150− * rounding up, if for example using the I915_GEM_CREATE_EXT_MEMORY_REGIONS
3151− * extension to place the object in device local-memory.
3153+ *
3154+ * DG2 64K min page size implications:
3155+ *
3156+ * On discrete platforms, starting from DG2, we have to contend with GTT
3157+ * page size restrictions when dealing with I915_MEMORY_CLASS_DEVICE
3158+ * objects. Specifically the hardware only supports 64K or larger GTT
3159+ * page sizes for such memory. The kernel will already ensure that all
3160+ * I915_MEMORY_CLASS_DEVICE memory is allocated using 64K or larger page
3161+ * sizes underneath.
3162+ *
3163+ * Note that the returned size here will always reflect any required
3164+ * rounding up done by the kernel, i.e 4K will now become 64K on devices
3165+ * such as DG2.
3166+ *
3167+ * Special DG2 GTT address alignment requirement:
3168+ *
3169+ * The GTT alignment will also need to be at least 2M for such objects.
3170+ *
3171+ * Note that due to how the hardware implements 64K GTT page support, we
3172+ * have some further complications:
3173+ *
3174+ * 1) The entire PDE (which covers a 2MB virtual address range), must
3175+ * contain only 64K PTEs, i.e mixing 4K and 64K PTEs in the same
3176+ * PDE is forbidden by the hardware.
3177+ *
3178+ * 2) We still need to support 4K PTEs for I915_MEMORY_CLASS_SYSTEM
3179+ * objects.
3180+ *
3181+ * To keep things simple for userland, we mandate that any GTT mappings
3182+ * must be aligned to and rounded up to 2MB. The kernel will internally
3183+ * pad them out to the next 2MB boundary. As this only wastes virtual
3184+ * address space and avoids userland having to copy any needlessly
3185+ * complicated PDE sharing scheme (coloring) and only affects DG2, this
3186+ * is deemed to be a good compromise.
31523187 */
31533188 __u64 size;
31543189 /**
include/uapi/drm/msm_drm.h+19−13
@@ -67,16 +67,21 @@ struct drm_msm_timespec { · 66 unmodified lines
6767 __s64 tv_nsec; /* nanoseconds */
6868 };
6969
70−#define MSM_PARAM_GPU_ID 0x01
71−#define MSM_PARAM_GMEM_SIZE 0x02
72−#define MSM_PARAM_CHIP_ID 0x03
73−#define MSM_PARAM_MAX_FREQ 0x04
74−#define MSM_PARAM_TIMESTAMP 0x05
75−#define MSM_PARAM_GMEM_BASE 0x06
76−#define MSM_PARAM_PRIORITIES 0x07 /* The # of priority levels */
77−#define MSM_PARAM_PP_PGTABLE 0x08 /* => 1 for per-process pagetables, else 0 */
78−#define MSM_PARAM_FAULTS 0x09
79−#define MSM_PARAM_SUSPENDS 0x0a
70+/* Below "RO" indicates a read-only param, "WO" indicates write-only, and
71+ * "RW" indicates a param that can be both read (GET_PARAM) and written
72+ * (SET_PARAM)
73+ */
74+#define MSM_PARAM_GPU_ID 0x01 /* RO */
75+#define MSM_PARAM_GMEM_SIZE 0x02 /* RO */
76+#define MSM_PARAM_CHIP_ID 0x03 /* RO */
77+#define MSM_PARAM_MAX_FREQ 0x04 /* RO */
78+#define MSM_PARAM_TIMESTAMP 0x05 /* RO */
79+#define MSM_PARAM_GMEM_BASE 0x06 /* RO */
80+#define MSM_PARAM_PRIORITIES 0x07 /* RO: The # of priority levels */
81+#define MSM_PARAM_PP_PGTABLE 0x08 /* RO: Deprecated, always returns zero */
82+#define MSM_PARAM_FAULTS 0x09 /* RO */
83+#define MSM_PARAM_SUSPENDS 0x0a /* RO */
84+#define MSM_PARAM_SYSPROF 0x0b /* WO: 1 preserves perfcntrs, 2 also disables suspend */
8085
8186 /* For backwards compat. The original support for preemption was based on
8287 * a single ring per priority level so # of priority levels equals the #
@@ -227,6 +232,7 @@ struct drm_msm_gem_submit_bo { · 144 unmodified lines
227232 #define MSM_SUBMIT_SUDO 0x10000000 /* run submitted cmds from RB */
228233 #define MSM_SUBMIT_SYNCOBJ_IN 0x08000000 /* enable input syncobj */
229234 #define MSM_SUBMIT_SYNCOBJ_OUT 0x04000000 /* enable output syncobj */
235+#define MSM_SUBMIT_FENCE_SN_IN 0x02000000 /* userspace passes in seqno fence */
230236 #define MSM_SUBMIT_FLAGS ( \
231237 MSM_SUBMIT_NO_IMPLICIT | \
232238 MSM_SUBMIT_FENCE_FD_IN | \
@@ -234,6 +240,7 @@ struct drm_msm_gem_submit_bo { · 1 unmodified line
234240 MSM_SUBMIT_SUDO | \
235241 MSM_SUBMIT_SYNCOBJ_IN | \
236242 MSM_SUBMIT_SYNCOBJ_OUT | \
243+ MSM_SUBMIT_FENCE_SN_IN | \
237244 0)
238245
239246 #define MSM_SUBMIT_SYNCOBJ_RESET 0x00000001 /* Reset syncobj after wait. */
@@ -253,7 +260,7 @@ struct drm_msm_gem_submit_syncobj { · 13 unmodified lines
253260 */
254261 struct drm_msm_gem_submit {
255262 __u32 flags; /* MSM_PIPE_x | MSM_SUBMIT_x */
256− __u32 fence; /* out */
263+ __u32 fence; /* out (or in with MSM_SUBMIT_FENCE_SN_IN flag) */
257264 __u32 nr_bos; /* in, number of submit_bo's */
258265 __u32 nr_cmds; /* in, number of submit_cmd's */
259266 __u64 bos; /* in, ptr to array of submit_bo's */
@@ -333,9 +340,7 @@ struct drm_msm_submitqueue_query { · 73 unmodified lines
333340 };
334341
335342 #define DRM_MSM_GET_PARAM 0x00
336−/* placeholder:
337343 #define DRM_MSM_SET_PARAM 0x01
338− */
339344 #define DRM_MSM_GEM_NEW 0x02
340345 #define DRM_MSM_GEM_INFO 0x03
341346 #define DRM_MSM_GEM_CPU_PREP 0x04
@@ -351,6 +356,7 @@ struct drm_msm_submitqueue_query { · 9 unmodified lines
351356 #define DRM_MSM_SUBMITQUEUE_QUERY 0x0C
352357
353358 #define DRM_IOCTL_MSM_GET_PARAM DRM_IOWR(DRM_COMMAND_BASE + DRM_MSM_GET_PARAM, struct drm_msm_param)
359+#define DRM_IOCTL_MSM_SET_PARAM DRM_IOW (DRM_COMMAND_BASE + DRM_MSM_SET_PARAM, struct drm_msm_param)
354360 #define DRM_IOCTL_MSM_GEM_NEW DRM_IOWR(DRM_COMMAND_BASE + DRM_MSM_GEM_NEW, struct drm_msm_gem_new)
355361 #define DRM_IOCTL_MSM_GEM_INFO DRM_IOWR(DRM_COMMAND_BASE + DRM_MSM_GEM_INFO, struct drm_msm_gem_info)
356362 #define DRM_IOCTL_MSM_GEM_CPU_PREP DRM_IOW (DRM_COMMAND_BASE + DRM_MSM_GEM_CPU_PREP, struct drm_msm_gem_cpu_prep)
include/uapi/drm/panfrost_drm.h+2−2
@@ -84,14 +84,14 @@ struct drm_panfrost_wait_bo { · 83 unmodified lines
8484 __s64 timeout_ns; /* absolute */
8585 };
8686
87+/* Valid flags to pass to drm_panfrost_create_bo */
8788 #define PANFROST_BO_NOEXEC 1
8889 #define PANFROST_BO_HEAP 2
8990
9091 /**
9192 * struct drm_panfrost_create_bo - ioctl argument for creating Panfrost BOs.
9293 *
93− * There are currently no values for the flags argument, but it may be
94− * used in a future extension.
94+ * The flags argument is a bit mask of PANFROST_BO_* flags.
9595 */
9696 struct drm_panfrost_create_bo {
9797 __u32 size;
include/uapi/linux/kfd_ioctl.h+91−7
@@ -32,9 +32,11 @@ · 31 unmodified lines
3232 * - 1.4 - Indicate new SRAM EDC bit in device properties
3333 * - 1.5 - Add SVM API
3434 * - 1.6 - Query clear flags in SVM get_attr API
35+ * - 1.7 - Checkpoint Restore (CRIU) API
36+ * - 1.8 - CRIU - Support for SDMA transfers with GTT BOs
3537 */
3638 #define KFD_IOCTL_MAJOR_VERSION 1
37−#define KFD_IOCTL_MINOR_VERSION 6
39+#define KFD_IOCTL_MINOR_VERSION 8
3840
3941 struct kfd_ioctl_get_version_args {
4042 __u32 major_version; /* from KFD */
@@ -194,6 +196,8 @@ struct kfd_ioctl_dbg_wave_control_args { · 153 unmodified lines
194196 __u32 buf_size_in_bytes; /*including gpu_id and buf_size */
195197 };
196198
199+#define KFD_INVALID_FD 0xffffffff
200+
197201 /* Matching HSA_EVENTTYPE */
198202 #define KFD_IOC_EVENT_SIGNAL 0
199203 #define KFD_IOC_EVENT_NODECHANGE 1
@@ -462,12 +466,89 @@ enum kfd_smi_event { · 262 unmodified lines
462466 };
463467
464468 #define KFD_SMI_EVENT_MASK_FROM_INDEX(i) (1ULL << ((i) - 1))
469+#define KFD_SMI_EVENT_MSG_SIZE 96
465470
466471 struct kfd_ioctl_smi_events_args {
467472 __u32 gpuid; /* to KFD */
468473 __u32 anon_fd; /* from KFD */
469474 };
470475
476+/**************************************************************************************************
477+ * CRIU IOCTLs (Checkpoint Restore In Userspace)
478+ *
479+ * When checkpointing a process, the userspace application will perform:
480+ * 1. PROCESS_INFO op to determine current process information. This pauses execution and evicts
481+ * all the queues.
482+ * 2. CHECKPOINT op to checkpoint process contents (BOs, queues, events, svm-ranges)
483+ * 3. UNPAUSE op to un-evict all the queues
484+ *
485+ * When restoring a process, the CRIU userspace application will perform:
486+ *
487+ * 1. RESTORE op to restore process contents
488+ * 2. RESUME op to start the process
489+ *
490+ * Note: Queues are forced into an evicted state after a successful PROCESS_INFO. User
491+ * application needs to perform an UNPAUSE operation after calling PROCESS_INFO.
492+ */
493+
494+enum kfd_criu_op {
495+ KFD_CRIU_OP_PROCESS_INFO,
496+ KFD_CRIU_OP_CHECKPOINT,
497+ KFD_CRIU_OP_UNPAUSE,
498+ KFD_CRIU_OP_RESTORE,
499+ KFD_CRIU_OP_RESUME,
500+};
501+
502+/**
503+ * kfd_ioctl_criu_args - Arguments perform CRIU operation
504+ * @devices: [in/out] User pointer to memory location for devices information.
505+ * This is an array of type kfd_criu_device_bucket.
506+ * @bos: [in/out] User pointer to memory location for BOs information
507+ * This is an array of type kfd_criu_bo_bucket.
508+ * @priv_data: [in/out] User pointer to memory location for private data
509+ * @priv_data_size: [in/out] Size of priv_data in bytes
510+ * @num_devices: [in/out] Number of GPUs used by process. Size of @devices array.
511+ * @num_bos [in/out] Number of BOs used by process. Size of @bos array.
512+ * @num_objects: [in/out] Number of objects used by process. Objects are opaque to
513+ * user application.
514+ * @pid: [in/out] PID of the process being checkpointed
515+ * @op [in] Type of operation (kfd_criu_op)
516+ *
517+ * Return: 0 on success, -errno on failure
518+ */
519+struct kfd_ioctl_criu_args {
520+ __u64 devices; /* Used during ops: CHECKPOINT, RESTORE */
521+ __u64 bos; /* Used during ops: CHECKPOINT, RESTORE */
522+ __u64 priv_data; /* Used during ops: CHECKPOINT, RESTORE */
523+ __u64 priv_data_size; /* Used during ops: PROCESS_INFO, RESTORE */
524+ __u32 num_devices; /* Used during ops: PROCESS_INFO, RESTORE */
525+ __u32 num_bos; /* Used during ops: PROCESS_INFO, RESTORE */
526+ __u32 num_objects; /* Used during ops: PROCESS_INFO, RESTORE */
527+ __u32 pid; /* Used during ops: PROCESS_INFO, RESUME */
528+ __u32 op;
529+};
530+
531+struct kfd_criu_device_bucket {
532+ __u32 user_gpu_id;
533+ __u32 actual_gpu_id;
534+ __u32 drm_fd;
535+ __u32 pad;
536+};
537+
538+struct kfd_criu_bo_bucket {
539+ __u64 addr;
540+ __u64 size;
541+ __u64 offset;
542+ __u64 restored_offset; /* During restore, updated offset for BO */
543+ __u32 gpu_id; /* This is the user_gpu_id */
544+ __u32 alloc_flags;
545+ __u32 dmabuf_fd;
546+ __u32 pad;
547+};
548+
549+/* CRIU IOCTLs - END */
550+/**************************************************************************************************/
551+
471552 /* Register offset inside the remapped mmio page
472553 */
473554 enum kfd_mmio_remap {
@@ -596,7 +677,7 @@ struct kfd_ioctl_svm_args { · 122 unmodified lines
596677 __u32 op;
597678 __u32 nattr;
598679 /* Variable length array of attributes */
599− struct kfd_ioctl_svm_attribute attrs[0];
680+ struct kfd_ioctl_svm_attribute attrs[];
600681 };
601682
602683 /**
@@ -679,16 +760,16 @@ struct kfd_ioctl_set_xnack_mode_args { · 76 unmodified lines
679760 #define AMDKFD_IOC_WAIT_EVENTS \
680761 AMDKFD_IOWR(0x0C, struct kfd_ioctl_wait_events_args)
681762
682−#define AMDKFD_IOC_DBG_REGISTER \
763+#define AMDKFD_IOC_DBG_REGISTER_DEPRECATED \
683764 AMDKFD_IOW(0x0D, struct kfd_ioctl_dbg_register_args)
684765
685−#define AMDKFD_IOC_DBG_UNREGISTER \
766+#define AMDKFD_IOC_DBG_UNREGISTER_DEPRECATED \
686767 AMDKFD_IOW(0x0E, struct kfd_ioctl_dbg_unregister_args)
687768
688−#define AMDKFD_IOC_DBG_ADDRESS_WATCH \
769+#define AMDKFD_IOC_DBG_ADDRESS_WATCH_DEPRECATED \
689770 AMDKFD_IOW(0x0F, struct kfd_ioctl_dbg_address_watch_args)
690771
691−#define AMDKFD_IOC_DBG_WAVE_CONTROL \
772+#define AMDKFD_IOC_DBG_WAVE_CONTROL_DEPRECATED \
692773 AMDKFD_IOW(0x10, struct kfd_ioctl_dbg_wave_control_args)
693774
694775 #define AMDKFD_IOC_SET_SCRATCH_BACKING_VA \
@@ -742,7 +823,10 @@ struct kfd_ioctl_set_xnack_mode_args { · 47 unmodified lines
742823 #define AMDKFD_IOC_SET_XNACK_MODE \
743824 AMDKFD_IOWR(0x21, struct kfd_ioctl_set_xnack_mode_args)
744825
826+#define AMDKFD_IOC_CRIU_OP \
827+ AMDKFD_IOWR(0x22, struct kfd_ioctl_criu_args)
828+
745829 #define AMDKFD_COMMAND_START 0x01
746−#define AMDKFD_COMMAND_END 0x22
830+#define AMDKFD_COMMAND_END 0x23
747831
748832 #endif

Keyboard shortcuts

General

Show keyboard shortcuts
?
Focus the current search
/

Issues and pull requests

Open filters
f

Issues, pull requests and Inbox

Select next item
j↓
Select previous item
k↑
Open selected item
Enter
Dismiss filters or clear selection
Esc

Repository code

Go to file
t