diff options
Diffstat (limited to 'include/uapi/drm')
| -rw-r--r-- | include/uapi/drm/amdgpu_drm.h | 396 | ||||
| -rw-r--r-- | include/uapi/drm/amdxdna_accel.h | 252 | ||||
| -rw-r--r-- | include/uapi/drm/asahi_drm.h | 1194 | ||||
| -rw-r--r-- | include/uapi/drm/drm.h | 82 | ||||
| -rw-r--r-- | include/uapi/drm/drm_fourcc.h | 142 | ||||
| -rw-r--r-- | include/uapi/drm/drm_mode.h | 276 | ||||
| -rw-r--r-- | include/uapi/drm/drm_ras.h | 49 | ||||
| -rw-r--r-- | include/uapi/drm/ethosu_accel.h | 261 | ||||
| -rw-r--r-- | include/uapi/drm/ivpu_accel.h | 81 | ||||
| -rw-r--r-- | include/uapi/drm/msm_drm.h | 150 | ||||
| -rw-r--r-- | include/uapi/drm/nouveau_drm.h | 66 | ||||
| -rw-r--r-- | include/uapi/drm/nova_drm.h | 101 | ||||
| -rw-r--r-- | include/uapi/drm/panfrost_drm.h | 239 | ||||
| -rw-r--r-- | include/uapi/drm/panthor_drm.h | 283 | ||||
| -rw-r--r-- | include/uapi/drm/rocket_accel.h | 192 | ||||
| -rw-r--r-- | include/uapi/drm/v3d_drm.h | 2 | ||||
| -rw-r--r-- | include/uapi/drm/virtgpu_drm.h | 6 | ||||
| -rw-r--r-- | include/uapi/drm/xe_drm.h | 684 |
18 files changed, 4350 insertions, 106 deletions
diff --git a/include/uapi/drm/amdgpu_drm.h b/include/uapi/drm/amdgpu_drm.h index 25d5c6e90a99..9f3090db2f16 100644 --- a/include/uapi/drm/amdgpu_drm.h +++ b/include/uapi/drm/amdgpu_drm.h @@ -54,6 +54,10 @@ extern "C" { #define DRM_AMDGPU_VM 0x13 #define DRM_AMDGPU_FENCE_TO_HANDLE 0x14 #define DRM_AMDGPU_SCHED 0x15 +#define DRM_AMDGPU_USERQ 0x16 +#define DRM_AMDGPU_USERQ_SIGNAL 0x17 +#define DRM_AMDGPU_USERQ_WAIT 0x18 +#define DRM_AMDGPU_GEM_LIST_HANDLES 0x19 #define DRM_IOCTL_AMDGPU_GEM_CREATE DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_GEM_CREATE, union drm_amdgpu_gem_create) #define DRM_IOCTL_AMDGPU_GEM_MMAP DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_GEM_MMAP, union drm_amdgpu_gem_mmap) @@ -71,6 +75,10 @@ extern "C" { #define DRM_IOCTL_AMDGPU_VM DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_VM, union drm_amdgpu_vm) #define DRM_IOCTL_AMDGPU_FENCE_TO_HANDLE DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_FENCE_TO_HANDLE, union drm_amdgpu_fence_to_handle) #define DRM_IOCTL_AMDGPU_SCHED DRM_IOW(DRM_COMMAND_BASE + DRM_AMDGPU_SCHED, union drm_amdgpu_sched) +#define DRM_IOCTL_AMDGPU_USERQ DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_USERQ, union drm_amdgpu_userq) +#define DRM_IOCTL_AMDGPU_USERQ_SIGNAL DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_USERQ_SIGNAL, struct drm_amdgpu_userq_signal) +#define DRM_IOCTL_AMDGPU_USERQ_WAIT DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_USERQ_WAIT, struct drm_amdgpu_userq_wait) +#define DRM_IOCTL_AMDGPU_GEM_LIST_HANDLES DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDGPU_GEM_LIST_HANDLES, struct drm_amdgpu_gem_list_handles) /** * DOC: memory domains @@ -110,7 +118,7 @@ extern "C" { AMDGPU_GEM_DOMAIN_VRAM | \ AMDGPU_GEM_DOMAIN_GDS | \ AMDGPU_GEM_DOMAIN_GWS | \ - AMDGPU_GEM_DOMAIN_OA | \ + AMDGPU_GEM_DOMAIN_OA | \ AMDGPU_GEM_DOMAIN_DOORBELL) /* Flag that CPU access will be required for the case of VRAM domain */ @@ -319,6 +327,263 @@ union drm_amdgpu_ctx { union drm_amdgpu_ctx_out out; }; +/* user queue IOCTL operations */ +#define AMDGPU_USERQ_OP_CREATE 1 +#define AMDGPU_USERQ_OP_FREE 2 + +/* queue priority levels */ +/* low < normal low < normal high < high */ +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_MASK 0x3 +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_SHIFT 0 +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_NORMAL_LOW 0 +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_LOW 1 +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_NORMAL_HIGH 2 +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_HIGH 3 /* admin only */ +/* for queues that need access to protected content */ +#define AMDGPU_USERQ_CREATE_FLAGS_QUEUE_SECURE (1 << 2) + +/* + * This structure is a container to pass input configuration + * info for all supported userqueue related operations. + * For operation AMDGPU_USERQ_OP_CREATE: user is expected + * to set all fields, excep the parameter 'queue_id'. + * For operation AMDGPU_USERQ_OP_FREE: the only input parameter expected + * to be set is 'queue_id', eveything else is ignored. + */ +struct drm_amdgpu_userq_in { + /** AMDGPU_USERQ_OP_* */ + __u32 op; + /** Queue id passed for operation USERQ_OP_FREE */ + __u32 queue_id; + /** the target GPU engine to execute workload (AMDGPU_HW_IP_*) */ + __u32 ip_type; + /** + * @doorbell_handle: the handle of doorbell GEM object + * associated with this userqueue client. + */ + __u32 doorbell_handle; + /** + * @doorbell_offset: 32-bit offset of the doorbell in the doorbell bo. + * Kernel will generate absolute doorbell offset using doorbell_handle + * and doorbell_offset in the doorbell bo. + */ + __u32 doorbell_offset; + /** + * @flags: flags used for queue parameters + */ + __u32 flags; + /** + * @queue_va: Virtual address of the GPU memory which holds the queue + * object. The queue holds the workload packets. + */ + __u64 queue_va; + /** + * @queue_size: Size of the queue in bytes, this needs to be 256-byte + * aligned. + */ + __u64 queue_size; + /** + * @rptr_va : Virtual address of the GPU memory which holds the ring RPTR. + * This object must be at least 8 byte in size and aligned to 8-byte offset. + */ + __u64 rptr_va; + /** + * @wptr_va : Virtual address of the GPU memory which holds the ring WPTR. + * This object must be at least 8 byte in size and aligned to 8-byte offset. + * + * Queue, RPTR and WPTR can come from the same object, as long as the size + * and alignment related requirements are met. + */ + __u64 wptr_va; + /** + * @mqd: MQD (memory queue descriptor) is a set of parameters which allow + * the GPU to uniquely define and identify a usermode queue. + * + * MQD data can be of different size for different GPU IP/engine and + * their respective versions/revisions, so this points to a __u64 * + * which holds IP specific MQD of this usermode queue. + */ + __u64 mqd; + /** + * @size: size of MQD data in bytes, it must match the MQD structure + * size of the respective engine/revision defined in UAPI for ex, for + * gfx11 workloads, size = sizeof(drm_amdgpu_userq_mqd_gfx11). + */ + __u64 mqd_size; +}; + +/* The structure to carry output of userqueue ops */ +struct drm_amdgpu_userq_out { + /** + * For operation AMDGPU_USERQ_OP_CREATE: This field contains a unique + * queue ID to represent the newly created userqueue in the system, otherwise + * it should be ignored. + */ + __u32 queue_id; + __u32 _pad; +}; + +union drm_amdgpu_userq { + struct drm_amdgpu_userq_in in; + struct drm_amdgpu_userq_out out; +}; + +/* GFX V11 IP specific MQD parameters */ +struct drm_amdgpu_userq_mqd_gfx11 { + /** + * @shadow_va: Virtual address of the GPU memory to hold the shadow buffer. + * Use AMDGPU_INFO_IOCTL to find the exact size of the object. + */ + __u64 shadow_va; + /** + * @csa_va: Virtual address of the GPU memory to hold the CSA buffer. + * Use AMDGPU_INFO_IOCTL to find the exact size of the object. + */ + __u64 csa_va; +}; + +/* GFX V11 SDMA IP specific MQD parameters */ +struct drm_amdgpu_userq_mqd_sdma_gfx11 { + /** + * @csa_va: Virtual address of the GPU memory to hold the CSA buffer. + * This must be a from a separate GPU object, and use AMDGPU_INFO IOCTL + * to get the size. + */ + __u64 csa_va; +}; + +/* GFX V11 Compute IP specific MQD parameters */ +struct drm_amdgpu_userq_mqd_compute_gfx11 { + /** + * @eop_va: Virtual address of the GPU memory to hold the EOP buffer. + * This must be a from a separate GPU object, and use AMDGPU_INFO IOCTL + * to get the size. + */ + __u64 eop_va; +}; + +/* userq signal/wait ioctl */ +struct drm_amdgpu_userq_signal { + /** + * @queue_id: Queue handle used by the userq fence creation function + * to retrieve the WPTR. + */ + __u32 queue_id; + __u32 pad; + /** + * @syncobj_handles: The list of syncobj handles submitted by the user queue + * job to be signaled. + */ + __u64 syncobj_handles; + /** + * @num_syncobj_handles: A count that represents the number of syncobj handles in + * @syncobj_handles. + */ + __u16 num_syncobj_handles; + __u16 pad0; + __u32 pad1; + /** + * @bo_read_handles: The list of BO handles that the submitted user queue job + * is using for read only. This will update BO fences in the kernel. + */ + __u64 bo_read_handles; + /** + * @bo_write_handles: The list of BO handles that the submitted user queue job + * is using for write only. This will update BO fences in the kernel. + */ + __u64 bo_write_handles; + /** + * @num_bo_read_handles: A count that represents the number of read BO handles in + * @bo_read_handles. + */ + __u32 num_bo_read_handles; + /** + * @num_bo_write_handles: A count that represents the number of write BO handles in + * @bo_write_handles. + */ + __u32 num_bo_write_handles; +}; + +struct drm_amdgpu_userq_fence_info { + /** + * @va: A gpu address allocated for each queue which stores the + * read pointer (RPTR) value. + */ + __u64 va; + /** + * @value: A 64 bit value represents the write pointer (WPTR) of the + * queue commands which compared with the RPTR value to signal the + * fences. + */ + __u64 value; +}; + +struct drm_amdgpu_userq_wait { + /** + * @waitq_id: Queue handle used by the userq wait IOCTL to retrieve the + * wait queue and maintain the fence driver references in it. + */ + __u32 waitq_id; + __u32 pad; + /** + * @syncobj_handles: The list of syncobj handles submitted by the user queue + * job to get the va/value pairs. + */ + __u64 syncobj_handles; + /** + * @syncobj_timeline_handles: The list of timeline syncobj handles submitted by + * the user queue job to get the va/value pairs at given @syncobj_timeline_points. + */ + __u64 syncobj_timeline_handles; + /** + * @syncobj_timeline_points: The list of timeline syncobj points submitted by the + * user queue job for the corresponding @syncobj_timeline_handles. + */ + __u64 syncobj_timeline_points; + /** + * @bo_read_handles: The list of read BO handles submitted by the user queue + * job to get the va/value pairs. + */ + __u64 bo_read_handles; + /** + * @bo_write_handles: The list of write BO handles submitted by the user queue + * job to get the va/value pairs. + */ + __u64 bo_write_handles; + /** + * @num_syncobj_timeline_handles: A count that represents the number of timeline + * syncobj handles in @syncobj_timeline_handles. + */ + __u16 num_syncobj_timeline_handles; + /** + * @num_fences: This field can be used both as input and output. As input it defines + * the maximum number of fences that can be returned and as output it will specify + * how many fences were actually returned from the ioctl. + */ + __u16 num_fences; + /** + * @num_syncobj_handles: A count that represents the number of syncobj handles in + * @syncobj_handles. + */ + __u16 num_syncobj_handles; + __u16 pad0; + /** + * @num_bo_read_handles: A count that represents the number of read BO handles in + * @bo_read_handles. + */ + __u32 num_bo_read_handles; + /** + * @num_bo_write_handles: A count that represents the number of write BO handles in + * @bo_write_handles. + */ + __u32 num_bo_write_handles; + /** + * @out_fences: The field is a return value from the ioctl containing the list of + * address/value pairs to wait for. + */ + __u64 out_fences; +}; + /* vm ioctl */ #define AMDGPU_VM_OP_RESERVE_VMID 1 #define AMDGPU_VM_OP_UNRESERVE_VMID 2 @@ -540,6 +805,21 @@ union drm_amdgpu_wait_fences { #define AMDGPU_GEM_OP_GET_GEM_CREATE_INFO 0 #define AMDGPU_GEM_OP_SET_PLACEMENT 1 +#define AMDGPU_GEM_OP_GET_MAPPING_INFO 2 + +struct drm_amdgpu_gem_vm_entry { + /* Start of mapping (in bytes) */ + __u64 addr; + + /* Size of mapping (in bytes) */ + __u64 size; + + /* Mapping offset */ + __u64 offset; + + /* flags needed to recreate mapping */ + __u64 flags; +}; /* Sets or returns a value associated with a buffer. */ struct drm_amdgpu_gem_op { @@ -547,8 +827,44 @@ struct drm_amdgpu_gem_op { __u32 handle; /** AMDGPU_GEM_OP_* */ __u32 op; - /** Input or return value */ + /** Input or return value. For MAPPING_INFO op: pointer to array of struct drm_amdgpu_gem_vm_entry */ __u64 value; + /** For MAPPING_INFO op: number of mappings (in/out) */ + __u32 num_entries; + + __u32 padding; +}; + +#define AMDGPU_GEM_LIST_HANDLES_FLAG_IS_IMPORT (1 << 0) + +struct drm_amdgpu_gem_list_handles { + /* User pointer to array of drm_amdgpu_gem_bo_info_entry */ + __u64 entries; + + /* Size of entries buffer / Number of handles in process (if larger than size of buffer, must retry) */ + __u32 num_entries; + + __u32 padding; +}; + +struct drm_amdgpu_gem_list_handles_entry { + /* gem handle of buffer object */ + __u32 gem_handle; + + /* Currently just one flag: IS_IMPORT */ + __u32 flags; + + /* Size of bo */ + __u64 size; + + /* Preferred domains for GEM_CREATE */ + __u64 preferred_domains; + + /* GEM_CREATE flags for re-creation of buffer */ + __u64 alloc_flags; + + /* physical start_addr alignment in bytes for some HW requirements */ + __u64 alignment; }; #define AMDGPU_VA_OP_MAP 1 @@ -566,7 +882,7 @@ struct drm_amdgpu_gem_op { #define AMDGPU_VM_PAGE_WRITEABLE (1 << 2) /* executable mapping, new for VI */ #define AMDGPU_VM_PAGE_EXECUTABLE (1 << 3) -/* partially resident texture */ +/* unmapped page of partially resident textures */ #define AMDGPU_VM_PAGE_PRT (1 << 4) /* MTYPE flags use bit 5 to 8 */ #define AMDGPU_VM_MTYPE_MASK (0xf << 5) @@ -599,6 +915,19 @@ struct drm_amdgpu_gem_va { __u64 offset_in_bo; /** Specify mapping size. Must be correctly aligned. */ __u64 map_size; + /** + * vm_timeline_point is a sequence number used to add new timeline point. + */ + __u64 vm_timeline_point; + /** + * The vm page table update fence is installed in given vm_timeline_syncobj_out + * at vm_timeline_point. + */ + __u32 vm_timeline_syncobj_out; + /** the number of syncobj handles in @input_fence_syncobj_handles */ + __u32 num_syncobj_handles; + /** Array of sync object handle to wait for given input fences */ + __u64 input_fence_syncobj_handles; }; #define AMDGPU_HW_IP_GFX 0 @@ -758,10 +1087,11 @@ struct drm_amdgpu_cs_chunk_cp_gfx_shadow { * Query h/w info: Flag that this is integrated (a.h.a. fusion) GPU * */ -#define AMDGPU_IDS_FLAGS_FUSION 0x1 -#define AMDGPU_IDS_FLAGS_PREEMPTION 0x2 -#define AMDGPU_IDS_FLAGS_TMZ 0x4 -#define AMDGPU_IDS_FLAGS_CONFORMANT_TRUNC_COORD 0x8 +#define AMDGPU_IDS_FLAGS_FUSION 0x01 +#define AMDGPU_IDS_FLAGS_PREEMPTION 0x02 +#define AMDGPU_IDS_FLAGS_TMZ 0x04 +#define AMDGPU_IDS_FLAGS_CONFORMANT_TRUNC_COORD 0x08 +#define AMDGPU_IDS_FLAGS_GANG_SUBMIT 0x10 /* * Query h/w info: Flag identifying VF/PF/PT mode @@ -940,6 +1270,8 @@ struct drm_amdgpu_cs_chunk_cp_gfx_shadow { #define AMDGPU_INFO_MAX_IBS 0x22 /* query last page fault info */ #define AMDGPU_INFO_GPUVM_FAULT 0x23 +/* query FW object size and alignment */ +#define AMDGPU_INFO_UQ_FW_AREAS 0x24 #define AMDGPU_INFO_MMR_SE_INDEX_SHIFT 0 #define AMDGPU_INFO_MMR_SE_INDEX_MASK 0xff @@ -1093,6 +1425,8 @@ struct drm_amdgpu_info_vbios { #define AMDGPU_VRAM_TYPE_DDR5 10 #define AMDGPU_VRAM_TYPE_LPDDR4 11 #define AMDGPU_VRAM_TYPE_LPDDR5 12 +#define AMDGPU_VRAM_TYPE_HBM3E 13 +#define AMDGPU_VRAM_TYPE_HBM4 14 struct drm_amdgpu_info_device { /** PCI Device ID */ @@ -1198,6 +1532,9 @@ struct drm_amdgpu_info_device { __u32 csa_size; /* context save area base virtual alignment for gfx11 */ __u32 csa_alignment; + /* Userq IP mask (1 << AMDGPU_HW_IP_*) */ + __u32 userq_ip_mask; + __u32 pad; }; struct drm_amdgpu_info_hw_ip { @@ -1214,6 +1551,8 @@ struct drm_amdgpu_info_hw_ip { __u32 available_rings; /** version info: bits 23:16 major, 15:8 minor, 7:0 revision */ __u32 ip_discovery_version; + /* Userq available slots */ + __u32 userq_num_slots; }; struct drm_amdgpu_info_num_handles { @@ -1279,6 +1618,39 @@ struct drm_amdgpu_info_gpuvm_fault { __u32 vmhub; }; +struct drm_amdgpu_info_uq_metadata_gfx { + /* 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_uq_metadata_compute { + /* EOP size for gfx11 */ + __u32 eop_size; + /* EOP base virtual alignment for gfx11 */ + __u32 eop_alignment; +}; + +struct drm_amdgpu_info_uq_metadata_sdma { + /* context save area size for sdma6 */ + __u32 csa_size; + /* context save area base virtual alignment for sdma6 */ + __u32 csa_alignment; +}; + +struct drm_amdgpu_info_uq_metadata { + union { + struct drm_amdgpu_info_uq_metadata_gfx gfx; + struct drm_amdgpu_info_uq_metadata_compute compute; + struct drm_amdgpu_info_uq_metadata_sdma sdma; + }; +}; + /* * Supported GPU families */ @@ -1298,17 +1670,9 @@ struct drm_amdgpu_info_gpuvm_fault { #define AMDGPU_FAMILY_GC_10_3_6 149 /* GC 10.3.6 */ #define AMDGPU_FAMILY_GC_10_3_7 151 /* GC 10.3.7 */ #define AMDGPU_FAMILY_GC_11_5_0 150 /* GC 11.5.0 */ +#define AMDGPU_FAMILY_GC_11_5_4 154 /* GC 11.5.4 */ #define AMDGPU_FAMILY_GC_12_0_0 152 /* GC 12.0.0 */ -/* FIXME wrong namespace! */ -struct drm_color_ctm_3x4 { - /* - * Conversion matrix with 3x4 dimensions in S31.32 sign-magnitude - * (not two's complement!) format. - */ - __u64 matrix[12]; -}; - #if defined(__cplusplus) } #endif diff --git a/include/uapi/drm/amdxdna_accel.h b/include/uapi/drm/amdxdna_accel.h index a706ead39082..61d3686fa3b1 100644 --- a/include/uapi/drm/amdxdna_accel.h +++ b/include/uapi/drm/amdxdna_accel.h @@ -19,6 +19,14 @@ extern "C" { #define AMDXDNA_INVALID_BO_HANDLE 0 #define AMDXDNA_INVALID_FENCE_HANDLE 0 +/* + * Define hardware context priority + */ +#define AMDXDNA_QOS_REALTIME_PRIORITY 0x100 +#define AMDXDNA_QOS_HIGH_PRIORITY 0x180 +#define AMDXDNA_QOS_NORMAL_PRIORITY 0x200 +#define AMDXDNA_QOS_LOW_PRIORITY 0x280 + enum amdxdna_device_type { AMDXDNA_DEV_TYPE_UNKNOWN = -1, AMDXDNA_DEV_TYPE_KMQ, @@ -34,6 +42,7 @@ enum amdxdna_drm_ioctl_id { DRM_AMDXDNA_EXEC_CMD, DRM_AMDXDNA_GET_INFO, DRM_AMDXDNA_SET_STATE, + DRM_AMDXDNA_GET_ARRAY = 10, }; /** @@ -147,10 +156,36 @@ struct amdxdna_drm_config_hwctx { enum amdxdna_bo_type { AMDXDNA_BO_INVALID = 0, - AMDXDNA_BO_SHMEM, - AMDXDNA_BO_DEV_HEAP, - AMDXDNA_BO_DEV, - AMDXDNA_BO_CMD, + AMDXDNA_BO_SHMEM = 1, /* Be compatible with legacy application code. */ + AMDXDNA_BO_SHARE = 1, + AMDXDNA_BO_DEV_HEAP = 2, + AMDXDNA_BO_DEV = 3, + AMDXDNA_BO_CMD = 4, +}; + +/** + * struct amdxdna_drm_va_entry + * @vaddr: Virtual address. + * @len: Size of entry. + */ +struct amdxdna_drm_va_entry { + __u64 vaddr; + __u64 len; +}; + +/** + * struct amdxdna_drm_va_tbl + * @dmabuf_fd: The fd of dmabuf. + * @num_entries: Number of va entries. + * @va_entries: Array of va entries. + * + * The input can be either a dmabuf fd or a virtual address entry table. + * When dmabuf_fd is used, num_entries must be zero. + */ +struct amdxdna_drm_va_tbl { + __s32 dmabuf_fd; + __u32 num_entries; + struct amdxdna_drm_va_entry va_entries[]; }; /** @@ -319,7 +354,8 @@ struct amdxdna_drm_query_clock_metadata { }; enum amdxdna_sensor_type { - AMDXDNA_SENSOR_TYPE_POWER + AMDXDNA_SENSOR_TYPE_POWER, + AMDXDNA_SENSOR_TYPE_COLUMN_UTILIZATION }; /** @@ -416,6 +452,52 @@ enum amdxdna_drm_get_param { DRM_AMDXDNA_QUERY_HW_CONTEXTS, DRM_AMDXDNA_QUERY_FIRMWARE_VERSION = 8, DRM_AMDXDNA_GET_POWER_MODE, + DRM_AMDXDNA_QUERY_TELEMETRY, + DRM_AMDXDNA_GET_FORCE_PREEMPT_STATE, + DRM_AMDXDNA_QUERY_RESOURCE_INFO, + DRM_AMDXDNA_GET_FRAME_BOUNDARY_PREEMPT_STATE, +}; + +/** + * struct amdxdna_drm_get_resource_info - Get resource information + */ +struct amdxdna_drm_get_resource_info { + /** @npu_clk_max: max H-Clocks */ + __u64 npu_clk_max; + /** @npu_tops_max: max TOPs */ + __u64 npu_tops_max; + /** @npu_task_max: max number of tasks */ + __u64 npu_task_max; + /** @npu_tops_curr: current TOPs */ + __u64 npu_tops_curr; + /** @npu_task_curr: current number of tasks */ + __u64 npu_task_curr; +}; + +/** + * struct amdxdna_drm_attribute_state - State of an attribute + */ +struct amdxdna_drm_attribute_state { + /** @state: enabled or disabled */ + __u8 state; + /** @pad: MBZ */ + __u8 pad[7]; +}; + +/** + * struct amdxdna_drm_query_telemetry_header - Telemetry data header + */ +struct amdxdna_drm_query_telemetry_header { + /** @major: Firmware telemetry interface major version number */ + __u32 major; + /** @minor: Firmware telemetry interface minor version number */ + __u32 minor; + /** @type: Telemetry query type */ + __u32 type; + /** @map_num_elements: Total number of elements in the map table */ + __u32 map_num_elements; + /** @map: Element map */ + __u32 map[]; }; /** @@ -430,10 +512,166 @@ struct amdxdna_drm_get_info { __u64 buffer; /* in/out */ }; +#define AMDXDNA_HWCTX_STATE_IDLE 0 +#define AMDXDNA_HWCTX_STATE_ACTIVE 1 + +/** + * struct amdxdna_drm_hwctx_entry - The hardware context array entry + */ +struct amdxdna_drm_hwctx_entry { + /** @context_id: Context ID. */ + __u32 context_id; + /** @start_col: Start AIE array column assigned to context. */ + __u32 start_col; + /** @num_col: Number of AIE array columns assigned to context. */ + __u32 num_col; + /** @hwctx_id: The real hardware context id. */ + __u32 hwctx_id; + /** @pid: ID of process which created this context. */ + __s64 pid; + /** @command_submissions: Number of commands submitted. */ + __u64 command_submissions; + /** @command_completions: Number of commands completed. */ + __u64 command_completions; + /** @migrations: Number of times been migrated. */ + __u64 migrations; + /** @preemptions: Number of times been preempted. */ + __u64 preemptions; + /** @errors: Number of errors happened. */ + __u64 errors; + /** @priority: Context priority. */ + __u64 priority; + /** @heap_usage: Usage of device heap buffer. */ + __u64 heap_usage; + /** @suspensions: Number of times been suspended. */ + __u64 suspensions; + /** + * @state: Context state. + * %AMDXDNA_HWCTX_STATE_IDLE + * %AMDXDNA_HWCTX_STATE_ACTIVE + */ + __u32 state; + /** @pasid: PASID been bound. */ + __u32 pasid; + /** @gops: Giga operations per second. */ + __u32 gops; + /** @fps: Frames per second. */ + __u32 fps; + /** @dma_bandwidth: DMA bandwidth. */ + __u32 dma_bandwidth; + /** @latency: Frame response latency. */ + __u32 latency; + /** @frame_exec_time: Frame execution time. */ + __u32 frame_exec_time; + /** @txn_op_idx: Index of last control code executed. */ + __u32 txn_op_idx; + /** @ctx_pc: Program counter. */ + __u32 ctx_pc; + /** @fatal_error_type: Fatal error type if context crashes. */ + __u32 fatal_error_type; + /** @fatal_error_exception_type: Firmware exception type. */ + __u32 fatal_error_exception_type; + /** @fatal_error_exception_pc: Firmware exception program counter. */ + __u32 fatal_error_exception_pc; + /** @fatal_error_app_module: Exception module name. */ + __u32 fatal_error_app_module; + /** @pad: Structure pad. */ + __u32 pad; +}; + +/** + * struct amdxdna_async_error - XDNA async error structure + */ +struct amdxdna_async_error { + /** @err_code: Error code. */ + __u64 err_code; + /** @ts_us: Timestamp. */ + __u64 ts_us; + /** @ex_err_code: Extra error code */ + __u64 ex_err_code; +}; + +/** + * struct amdxdna_drm_bo_usage - all types of BO usage + * BOs managed by XRT/SHIM/driver is counted as internal. + * Others are counted as external which are managed by applications. + * + * Among all types of BOs: + * AMDXDNA_BO_DEV_HEAP - is counted for internal. + * AMDXDNA_BO_SHARE - is counted for external. + * AMDXDNA_BO_CMD - is counted for internal. + * AMDXDNA_BO_DEV - is counted by heap_usage only, not internal + * or external. It does not add to the total memory + * footprint since its mem comes from heap which is + * already counted as internal. + */ +struct amdxdna_drm_bo_usage { + /** @pid: The ID of the process to query from. */ + __s64 pid; + /** @total_usage: Total BO size used by process. */ + __u64 total_usage; + /** @internal_usage: Total internal BO size used by process. */ + __u64 internal_usage; + /** @heap_usage: Total device BO size used by process. */ + __u64 heap_usage; +}; + +/* + * Supported params in struct amdxdna_drm_get_array + */ +#define DRM_AMDXDNA_HW_CONTEXT_ALL 0 +#define DRM_AMDXDNA_HW_LAST_ASYNC_ERR 2 +#define DRM_AMDXDNA_BO_USAGE 6 + +/** + * struct amdxdna_drm_get_array - Get information array. + */ +struct amdxdna_drm_get_array { + /** + * @param: + * + * Supported params: + * + * %DRM_AMDXDNA_HW_CONTEXT_ALL: + * Returns all created hardware contexts. + * + * %DRM_AMDXDNA_HW_LAST_ASYNC_ERR: + * Returns last async error. + * + * %DRM_AMDXDNA_BO_USAGE: + * Returns usage of heap/internal/external BOs. + */ + __u32 param; + /** + * @element_size: + * + * Specifies maximum element size and returns the actual element size. + */ + __u32 element_size; + /** + * @num_element: + * + * Specifies maximum number of elements and returns the actual number + * of elements. + */ + __u32 num_element; /* in/out */ + /** @pad: MBZ */ + __u32 pad; + /** + * @buffer: + * + * Specifies the match conditions and returns the matched information + * array. + */ + __u64 buffer; +}; + enum amdxdna_drm_set_param { DRM_AMDXDNA_SET_POWER_MODE, DRM_AMDXDNA_WRITE_AIE_MEM, DRM_AMDXDNA_WRITE_AIE_REG, + DRM_AMDXDNA_SET_FORCE_PREEMPT, + DRM_AMDXDNA_SET_FRAME_BOUNDARY_PREEMPT, }; /** @@ -494,6 +732,10 @@ struct amdxdna_drm_set_power_mode { DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDXDNA_SET_STATE, \ struct amdxdna_drm_set_state) +#define DRM_IOCTL_AMDXDNA_GET_ARRAY \ + DRM_IOWR(DRM_COMMAND_BASE + DRM_AMDXDNA_GET_ARRAY, \ + struct amdxdna_drm_get_array) + #if defined(__cplusplus) } /* extern c end */ #endif diff --git a/include/uapi/drm/asahi_drm.h b/include/uapi/drm/asahi_drm.h new file mode 100644 index 000000000000..de67f1c603af --- /dev/null +++ b/include/uapi/drm/asahi_drm.h @@ -0,0 +1,1194 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright (C) The Asahi Linux Contributors + * Copyright (C) 2018-2023 Collabora Ltd. + * Copyright (C) 2014-2018 Broadcom + */ +#ifndef _ASAHI_DRM_H_ +#define _ASAHI_DRM_H_ + +#include "drm.h" + +#if defined(__cplusplus) +extern "C" { +#endif + +/** + * DOC: Introduction to the Asahi UAPI + * + * This documentation describes the Asahi IOCTLs. + * + * Just a few generic rules about the data passed to the Asahi IOCTLs (cribbed + * from Panthor): + * + * - Structures must be aligned on 64-bit/8-byte. If the object is not + * naturally aligned, a padding field must be added. + * - Fields must be explicitly aligned to their natural type alignment with + * pad[0..N] fields. + * - All padding fields will be checked by the driver to make sure they are + * zeroed. + * - Flags can be added, but not removed/replaced. + * - New fields can be added to the main structures (the structures + * directly passed to the ioctl). Those fields can be added at the end of + * the structure, or replace existing padding fields. Any new field being + * added must preserve the behavior that existed before those fields were + * added when a value of zero is passed. + * - New fields can be added to indirect objects (objects pointed by the + * main structure), iff those objects are passed a size to reflect the + * size known by the userspace driver (see + * drm_asahi_cmd_header::size). + * - If the kernel driver is too old to know some fields, those will be + * ignored if zero, and otherwise rejected (and so will be zero on output). + * - If userspace is too old to know some fields, those will be zeroed + * (input) before the structure is parsed by the kernel driver. + * - Each new flag/field addition must come with a driver version update so + * the userspace driver doesn't have to guess which flags are supported. + * - Structures should not contain unions, as this would defeat the + * extensibility of such structures. + * - IOCTLs can't be removed or replaced. New IOCTL IDs should be placed + * at the end of the drm_asahi_ioctl_id enum. + */ + +/** + * enum drm_asahi_ioctl_id - IOCTL IDs + * + * Place new ioctls at the end, don't re-order, don't replace or remove entries. + * + * These IDs are not meant to be used directly. Use the DRM_IOCTL_ASAHI_xxx + * definitions instead. + */ +enum drm_asahi_ioctl_id { + /** @DRM_ASAHI_GET_PARAMS: Query device properties. */ + DRM_ASAHI_GET_PARAMS = 0, + + /** @DRM_ASAHI_GET_TIME: Query device time. */ + DRM_ASAHI_GET_TIME, + + /** @DRM_ASAHI_VM_CREATE: Create a GPU VM address space. */ + DRM_ASAHI_VM_CREATE, + + /** @DRM_ASAHI_VM_DESTROY: Destroy a VM. */ + DRM_ASAHI_VM_DESTROY, + + /** @DRM_ASAHI_VM_BIND: Bind/unbind memory to a VM. */ + DRM_ASAHI_VM_BIND, + + /** @DRM_ASAHI_GEM_CREATE: Create a buffer object. */ + DRM_ASAHI_GEM_CREATE, + + /** + * @DRM_ASAHI_GEM_MMAP_OFFSET: Get offset to pass to mmap() to map a + * given GEM handle. + */ + DRM_ASAHI_GEM_MMAP_OFFSET, + + /** @DRM_ASAHI_GEM_BIND_OBJECT: Bind memory as a special object */ + DRM_ASAHI_GEM_BIND_OBJECT, + + /** @DRM_ASAHI_QUEUE_CREATE: Create a scheduling queue. */ + DRM_ASAHI_QUEUE_CREATE, + + /** @DRM_ASAHI_QUEUE_DESTROY: Destroy a scheduling queue. */ + DRM_ASAHI_QUEUE_DESTROY, + + /** @DRM_ASAHI_SUBMIT: Submit commands to a queue. */ + DRM_ASAHI_SUBMIT, +}; + +#define DRM_ASAHI_MAX_CLUSTERS 64 + +/** + * struct drm_asahi_params_global - Global parameters. + * + * This struct may be queried by drm_asahi_get_params. + */ +struct drm_asahi_params_global { + /** @features: Feature bits from drm_asahi_feature */ + __u64 features; + + /** @gpu_generation: GPU generation, e.g. 13 for G13G */ + __u32 gpu_generation; + + /** @gpu_variant: GPU variant as a character, e.g. 'C' for G13C */ + __u32 gpu_variant; + + /** + * @gpu_revision: GPU revision in BCD, e.g. 0x00 for 'A0' or + * 0x21 for 'C1' + */ + __u32 gpu_revision; + + /** @chip_id: Chip ID in BCD, e.g. 0x8103 for T8103 */ + __u32 chip_id; + + /** @num_dies: Number of dies in the SoC */ + __u32 num_dies; + + /** @num_clusters_total: Number of GPU clusters (across all dies) */ + __u32 num_clusters_total; + + /** + * @num_cores_per_cluster: Number of logical cores per cluster + * (including inactive/nonexistent) + */ + __u32 num_cores_per_cluster; + + /** @max_frequency_khz: Maximum GPU core clock frequency */ + __u32 max_frequency_khz; + + /** @core_masks: Bitmask of present/enabled cores per cluster */ + __u64 core_masks[DRM_ASAHI_MAX_CLUSTERS]; + + /** + * @vm_start: VM range start VMA. Together with @vm_end, this defines + * the window of valid GPU VAs. Userspace is expected to subdivide VAs + * out of this window. + * + * This window contains all virtual addresses that userspace needs to + * know about. There may be kernel-internal GPU VAs outside this range, + * but that detail is not relevant here. + */ + __u64 vm_start; + + /** @vm_end: VM range end VMA */ + __u64 vm_end; + + /** + * @vm_kernel_min_size: Minimum kernel VMA window size. + * + * When creating a VM, userspace is required to carve out a section of + * virtual addresses (within the range given by @vm_start and + * @vm_end). The kernel will allocate various internal structures + * within the specified VA range. + * + * Allowing userspace to choose the VA range for the kernel, rather than + * the kernel reserving VAs and requiring userspace to cope, can assist + * in implementing SVM. + */ + __u64 vm_kernel_min_size; + + /** + * @max_commands_per_submission: Maximum number of supported commands + * per submission. This mirrors firmware limits. Userspace must split up + * larger command buffers, which may require inserting additional + * synchronization. + */ + __u32 max_commands_per_submission; + + /** + * @max_attachments: Maximum number of drm_asahi_attachment's per + * command + */ + __u32 max_attachments; + + /** + * @command_timestamp_frequency_hz: Timebase frequency for timestamps + * written during command execution, specified via drm_asahi_timestamp + * structures. As this rate is controlled by the firmware, it is a + * queryable parameter. + * + * Userspace must divide by this frequency to convert timestamps to + * seconds, rather than hardcoding a particular firmware's rate. + */ + __u64 command_timestamp_frequency_hz; +}; + +/** + * enum drm_asahi_feature - Feature bits + * + * This covers only features that userspace cannot infer from the architecture + * version. Most features don't need to be here. + */ +enum drm_asahi_feature { + /** + * @DRM_ASAHI_FEATURE_SOFT_FAULTS: GPU has "soft fault" enabled. Shader + * loads of unmapped memory will return zero. Shader stores to unmapped + * memory will be silently discarded. Note that only shader load/store + * is affected. Other hardware units are not affected, notably including + * texture sampling. + * + * Soft fault is set when initializing the GPU and cannot be runtime + * toggled. Therefore, it is exposed as a feature bit and not a + * userspace-settable flag on the VM. When soft fault is enabled, + * userspace can speculate memory accesses more aggressively. + */ + DRM_ASAHI_FEATURE_SOFT_FAULTS = (1UL) << 0, +}; + +/** + * struct drm_asahi_get_params - Arguments passed to DRM_IOCTL_ASAHI_GET_PARAMS + */ +struct drm_asahi_get_params { + /** @param_group: Parameter group to fetch (MBZ) */ + __u32 param_group; + + /** @pad: MBZ */ + __u32 pad; + + /** @pointer: User pointer to write parameter struct */ + __u64 pointer; + + /** + * @size: Size of the user buffer. In case of older userspace, this may + * be less than sizeof(struct drm_asahi_params_global). The kernel will + * not write past the length specified here, allowing extensibility. + */ + __u64 size; +}; + +/** + * struct drm_asahi_vm_create - Arguments passed to DRM_IOCTL_ASAHI_VM_CREATE + */ +struct drm_asahi_vm_create { + /** + * @kernel_start: Start of the kernel-reserved address range. See + * drm_asahi_params_global::vm_kernel_min_size. + * + * Both @kernel_start and @kernel_end must be within the range of + * valid VAs given by drm_asahi_params_global::vm_start and + * drm_asahi_params_global::vm_end. The size of the kernel range + * (@kernel_end - @kernel_start) must be at least + * drm_asahi_params_global::vm_kernel_min_size. + * + * Userspace must not bind any memory on this VM into this reserved + * range, it is for kernel use only. + */ + __u64 kernel_start; + + /** + * @kernel_end: End of the kernel-reserved address range. See + * @kernel_start. + */ + __u64 kernel_end; + + /** @vm_id: Returned VM ID */ + __u32 vm_id; + + /** @pad: MBZ */ + __u32 pad; +}; + +/** + * struct drm_asahi_vm_destroy - Arguments passed to DRM_IOCTL_ASAHI_VM_DESTROY + */ +struct drm_asahi_vm_destroy { + /** @vm_id: VM ID to be destroyed */ + __u32 vm_id; + + /** @pad: MBZ */ + __u32 pad; +}; + +/** + * enum drm_asahi_gem_flags - Flags for GEM creation + */ +enum drm_asahi_gem_flags { + /** + * @DRM_ASAHI_GEM_WRITEBACK: BO should be CPU-mapped as writeback. + * + * Map as writeback instead of write-combine. This optimizes for CPU + * reads. + */ + DRM_ASAHI_GEM_WRITEBACK = (1L << 0), + + /** + * @DRM_ASAHI_GEM_VM_PRIVATE: BO is private to this GPU VM (no exports). + */ + DRM_ASAHI_GEM_VM_PRIVATE = (1L << 1), +}; + +/** + * struct drm_asahi_gem_create - Arguments passed to DRM_IOCTL_ASAHI_GEM_CREATE + */ +struct drm_asahi_gem_create { + /** @size: Size of the BO */ + __u64 size; + + /** @flags: Combination of drm_asahi_gem_flags flags. */ + __u32 flags; + + /** + * @vm_id: VM ID to assign to the BO, if DRM_ASAHI_GEM_VM_PRIVATE is set + */ + __u32 vm_id; + + /** @handle: Returned GEM handle for the BO */ + __u32 handle; + + /** @pad: MBZ */ + __u32 pad; +}; + +/** + * struct drm_asahi_gem_mmap_offset - Arguments passed to + * DRM_IOCTL_ASAHI_GEM_MMAP_OFFSET + */ +struct drm_asahi_gem_mmap_offset { + /** @handle: Handle for the object being mapped. */ + __u32 handle; + + /** @flags: Must be zero */ + __u32 flags; + + /** @offset: The fake offset to use for subsequent mmap call */ + __u64 offset; +}; + +/** + * enum drm_asahi_bind_flags - Flags for GEM binding + */ +enum drm_asahi_bind_flags { + /** + * @DRM_ASAHI_BIND_UNBIND: Instead of binding a GEM object to the range, + * simply unbind the GPU VMA range. + */ + DRM_ASAHI_BIND_UNBIND = (1L << 0), + + /** @DRM_ASAHI_BIND_READ: Map BO with GPU read permission */ + DRM_ASAHI_BIND_READ = (1L << 1), + + /** @DRM_ASAHI_BIND_WRITE: Map BO with GPU write permission */ + DRM_ASAHI_BIND_WRITE = (1L << 2), + + /** + * @DRM_ASAHI_BIND_SINGLE_PAGE: Map a single page of the BO repeatedly + * across the VA range. + * + * This is useful to fill a VA range with scratch pages or zero pages. + * It is intended as a mechanism to accelerate sparse. + */ + DRM_ASAHI_BIND_SINGLE_PAGE = (1L << 3), +}; + +/** + * struct drm_asahi_gem_bind_op - Description of a single GEM bind operation. + */ +struct drm_asahi_gem_bind_op { + /** @flags: Combination of drm_asahi_bind_flags flags. */ + __u32 flags; + + /** @handle: GEM object to bind (except for UNBIND) */ + __u32 handle; + + /** + * @offset: Offset into the object (except for UNBIND). + * + * For a regular bind, this is the beginning of the region of the GEM + * object to bind. + * + * For a single-page bind, this is the offset to the single page that + * will be repeatedly bound. + * + * Must be page-size aligned. + */ + __u64 offset; + + /** + * @range: Number of bytes to bind/unbind to @addr. + * + * Must be page-size aligned. + */ + __u64 range; + + /** + * @addr: Address to bind to. + * + * Must be page-size aligned. + */ + __u64 addr; +}; + +/** + * struct drm_asahi_vm_bind - Arguments passed to + * DRM_IOCTL_ASAHI_VM_BIND + */ +struct drm_asahi_vm_bind { + /** @vm_id: The ID of the VM to bind to */ + __u32 vm_id; + + /** @num_binds: number of binds in this IOCTL. */ + __u32 num_binds; + + /** + * @stride: Stride in bytes between consecutive binds. This allows + * extensibility of drm_asahi_gem_bind_op. + */ + __u32 stride; + + /** @pad: MBZ */ + __u32 pad; + + /** + * @userptr: User pointer to an array of @num_binds structures of type + * @drm_asahi_gem_bind_op and size @stride bytes. + */ + __u64 userptr; +}; + +/** + * enum drm_asahi_bind_object_op - Special object bind operation + */ +enum drm_asahi_bind_object_op { + /** @DRM_ASAHI_BIND_OBJECT_OP_BIND: Bind a BO as a special GPU object */ + DRM_ASAHI_BIND_OBJECT_OP_BIND = 0, + + /** @DRM_ASAHI_BIND_OBJECT_OP_UNBIND: Unbind a special GPU object */ + DRM_ASAHI_BIND_OBJECT_OP_UNBIND = 1, +}; + +/** + * enum drm_asahi_bind_object_flags - Special object bind flags + */ +enum drm_asahi_bind_object_flags { + /** + * @DRM_ASAHI_BIND_OBJECT_USAGE_TIMESTAMPS: Map a BO as a timestamp + * buffer. + */ + DRM_ASAHI_BIND_OBJECT_USAGE_TIMESTAMPS = (1L << 0), +}; + +/** + * struct drm_asahi_gem_bind_object - Arguments passed to + * DRM_IOCTL_ASAHI_GEM_BIND_OBJECT + */ +struct drm_asahi_gem_bind_object { + /** @op: Bind operation (enum drm_asahi_bind_object_op) */ + __u32 op; + + /** @flags: Combination of drm_asahi_bind_object_flags flags. */ + __u32 flags; + + /** @handle: GEM object to bind/unbind (BIND) */ + __u32 handle; + + /** @vm_id: The ID of the VM to operate on (MBZ currently) */ + __u32 vm_id; + + /** @offset: Offset into the object (BIND only) */ + __u64 offset; + + /** @range: Number of bytes to bind/unbind (BIND only) */ + __u64 range; + + /** @object_handle: Object handle (out for BIND, in for UNBIND) */ + __u32 object_handle; + + /** @pad: MBZ */ + __u32 pad; +}; + +/** + * enum drm_asahi_cmd_type - Command type + */ +enum drm_asahi_cmd_type { + /** + * @DRM_ASAHI_CMD_RENDER: Render command, executing on the render + * subqueue. Combined vertex and fragment operation. + * + * Followed by a @drm_asahi_cmd_render payload. + */ + DRM_ASAHI_CMD_RENDER = 0, + + /** + * @DRM_ASAHI_CMD_COMPUTE: Compute command on the compute subqueue. + * + * Followed by a @drm_asahi_cmd_compute payload. + */ + DRM_ASAHI_CMD_COMPUTE = 1, + + /** + * @DRM_ASAHI_SET_VERTEX_ATTACHMENTS: Software command to set + * attachments for subsequent vertex shaders in the same submit. + * + * Followed by (possibly multiple) @drm_asahi_attachment payloads. + */ + DRM_ASAHI_SET_VERTEX_ATTACHMENTS = 2, + + /** + * @DRM_ASAHI_SET_FRAGMENT_ATTACHMENTS: Software command to set + * attachments for subsequent fragment shaders in the same submit. + * + * Followed by (possibly multiple) @drm_asahi_attachment payloads. + */ + DRM_ASAHI_SET_FRAGMENT_ATTACHMENTS = 3, + + /** + * @DRM_ASAHI_SET_COMPUTE_ATTACHMENTS: Software command to set + * attachments for subsequent compute shaders in the same submit. + * + * Followed by (possibly multiple) @drm_asahi_attachment payloads. + */ + DRM_ASAHI_SET_COMPUTE_ATTACHMENTS = 4, +}; + +/** + * enum drm_asahi_priority - Scheduling queue priority. + * + * These priorities are forwarded to the firmware to influence firmware + * scheduling. The exact policy is ultimately decided by firmware, but + * these enums allow userspace to communicate the intentions. + */ +enum drm_asahi_priority { + /** @DRM_ASAHI_PRIORITY_LOW: Low priority queue. */ + DRM_ASAHI_PRIORITY_LOW = 0, + + /** @DRM_ASAHI_PRIORITY_MEDIUM: Medium priority queue. */ + DRM_ASAHI_PRIORITY_MEDIUM = 1, + + /** + * @DRM_ASAHI_PRIORITY_HIGH: High priority queue. + * + * Reserved for future extension. + */ + DRM_ASAHI_PRIORITY_HIGH = 2, + + /** + * @DRM_ASAHI_PRIORITY_REALTIME: Real-time priority queue. + * + * Reserved for future extension. + */ + DRM_ASAHI_PRIORITY_REALTIME = 3, +}; + +/** + * struct drm_asahi_queue_create - Arguments passed to + * DRM_IOCTL_ASAHI_QUEUE_CREATE + */ +struct drm_asahi_queue_create { + /** @flags: MBZ */ + __u32 flags; + + /** @vm_id: The ID of the VM this queue is bound to */ + __u32 vm_id; + + /** @priority: One of drm_asahi_priority */ + __u32 priority; + + /** @queue_id: The returned queue ID */ + __u32 queue_id; + + /** + * @usc_exec_base: GPU base address for all USC binaries (shaders) on + * this queue. USC addresses are 32-bit relative to this 64-bit base. + * + * This sets the following registers on all queue commands: + * + * USC_EXEC_BASE_TA (vertex) + * USC_EXEC_BASE_ISP (fragment) + * USC_EXEC_BASE_CP (compute) + * + * While the hardware lets us configure these independently per command, + * we do not have a use case for this. Instead, we expect userspace to + * fix a 4GiB VA carveout for USC memory and pass its base address here. + */ + __u64 usc_exec_base; +}; + +/** + * struct drm_asahi_queue_destroy - Arguments passed to + * DRM_IOCTL_ASAHI_QUEUE_DESTROY + */ +struct drm_asahi_queue_destroy { + /** @queue_id: The queue ID to be destroyed */ + __u32 queue_id; + + /** @pad: MBZ */ + __u32 pad; +}; + +/** + * enum drm_asahi_sync_type - Sync item type + */ +enum drm_asahi_sync_type { + /** @DRM_ASAHI_SYNC_SYNCOBJ: Binary sync object */ + DRM_ASAHI_SYNC_SYNCOBJ = 0, + + /** @DRM_ASAHI_SYNC_TIMELINE_SYNCOBJ: Timeline sync object */ + DRM_ASAHI_SYNC_TIMELINE_SYNCOBJ = 1, +}; + +/** + * struct drm_asahi_sync - Sync item + */ +struct drm_asahi_sync { + /** @sync_type: One of drm_asahi_sync_type */ + __u32 sync_type; + + /** @handle: The sync object handle */ + __u32 handle; + + /** @timeline_value: Timeline value for timeline sync objects */ + __u64 timeline_value; +}; + +/** + * define DRM_ASAHI_BARRIER_NONE - Command index for no barrier + * + * This special value may be passed in to drm_asahi_command::vdm_barrier or + * drm_asahi_command::cdm_barrier to indicate that the respective subqueue + * should not wait on any previous work. + */ +#define DRM_ASAHI_BARRIER_NONE (0xFFFFu) + +/** + * struct drm_asahi_cmd_header - Top level command structure + * + * This struct is core to the command buffer definition and therefore is not + * extensible. + */ +struct drm_asahi_cmd_header { + /** @cmd_type: One of drm_asahi_cmd_type */ + __u16 cmd_type; + + /** + * @size: Size of this command, not including this header. + * + * For hardware commands, this enables extensibility of commands without + * requiring extra command types. Passing a command that is shorter + * than expected is explicitly allowed for backwards-compatibility. + * Truncated fields will be zeroed. + * + * For the synthetic attachment setting commands, this implicitly + * encodes the number of attachments. These commands take multiple + * fixed-size @drm_asahi_attachment structures as their payload, so size + * equals number of attachments * sizeof(struct drm_asahi_attachment). + */ + __u16 size; + + /** + * @vdm_barrier: VDM (render) command index to wait on. + * + * Barriers are indices relative to the beginning of a given submit. A + * barrier of 0 waits on commands submitted to the respective subqueue + * in previous submit ioctls. A barrier of N waits on N previous + * commands on the subqueue within the current submit ioctl. As a + * special case, passing @DRM_ASAHI_BARRIER_NONE avoids waiting on any + * commands in the subqueue. + * + * Examples: + * + * 0: This waits on all previous work. + * + * NONE: This does not wait for anything on this subqueue. + * + * 1: This waits on the first render command in the submit. + * This is valid only if there are multiple render commands in the + * same submit. + * + * Barriers are valid only for hardware commands. Synthetic software + * commands to set attachments must pass NONE here. + */ + __u16 vdm_barrier; + + /** + * @cdm_barrier: CDM (compute) command index to wait on. + * + * See @vdm_barrier, and replace VDM/render with CDM/compute. + */ + __u16 cdm_barrier; +}; + +/** + * struct drm_asahi_submit - Arguments passed to DRM_IOCTL_ASAHI_SUBMIT + */ +struct drm_asahi_submit { + /** + * @syncs: An optional pointer to an array of drm_asahi_sync. The first + * @in_sync_count elements are in-syncs, then the remaining + * @out_sync_count elements are out-syncs. Using a single array with + * explicit partitioning simplifies handling. + */ + __u64 syncs; + + /** + * @cmdbuf: Pointer to the command buffer to submit. + * + * This is a flat command buffer. By design, it contains no CPU + * pointers, which makes it suitable for a virtgpu wire protocol without + * requiring any serializing/deserializing step. + * + * It consists of a series of commands. Each command begins with a + * fixed-size @drm_asahi_cmd_header header and is followed by a + * variable-length payload according to the type and size in the header. + * + * The combined count of "real" hardware commands must be nonzero and at + * most drm_asahi_params_global::max_commands_per_submission. + */ + __u64 cmdbuf; + + /** @flags: Flags for command submission (MBZ) */ + __u32 flags; + + /** @queue_id: The queue ID to be submitted to */ + __u32 queue_id; + + /** + * @in_sync_count: Number of sync objects to wait on before starting + * this job. + */ + __u32 in_sync_count; + + /** + * @out_sync_count: Number of sync objects to signal upon completion of + * this job. + */ + __u32 out_sync_count; + + /** @cmdbuf_size: Command buffer size in bytes */ + __u32 cmdbuf_size; + + /** @pad: MBZ */ + __u32 pad; +}; + +/** + * struct drm_asahi_attachment - Describe an "attachment". + * + * Attachments are any memory written by shaders, notably including render + * target attachments written by the end-of-tile program. This is purely a hint + * about the accessed memory regions. It is optional to specify, which is + * fortunate as it cannot be specified precisely with bindless access anyway. + * But where possible, it's probably a good idea for userspace to include these + * hints, forwarded to the firmware. + * + * This struct is implicitly sized and therefore is not extensible. + */ +struct drm_asahi_attachment { + /** @pointer: Base address of the attachment */ + __u64 pointer; + + /** @size: Size of the attachment in bytes */ + __u64 size; + + /** @pad: MBZ */ + __u32 pad; + + /** @flags: MBZ */ + __u32 flags; +}; + +enum drm_asahi_render_flags { + /** + * @DRM_ASAHI_RENDER_VERTEX_SCRATCH: A vertex stage shader uses scratch + * memory. + */ + DRM_ASAHI_RENDER_VERTEX_SCRATCH = (1U << 0), + + /** + * @DRM_ASAHI_RENDER_PROCESS_EMPTY_TILES: Process even empty tiles. + * This must be set when clearing render targets. + */ + DRM_ASAHI_RENDER_PROCESS_EMPTY_TILES = (1U << 1), + + /** + * @DRM_ASAHI_RENDER_NO_VERTEX_CLUSTERING: Run vertex stage on a single + * cluster (on multi-cluster GPUs) + * + * This harms performance but can workaround certain sync/coherency + * bugs, and therefore is useful for debugging. + */ + DRM_ASAHI_RENDER_NO_VERTEX_CLUSTERING = (1U << 2), + + /** + * @DRM_ASAHI_RENDER_DBIAS_IS_INT: Use integer depth bias formula. + * + * Graphics specifications contain two alternate formulas for depth + * bias, a float formula used with floating-point depth buffers and an + * integer formula using with unorm depth buffers. This flag specifies + * that the integer formula should be used. If omitted, the float + * formula is used instead. + * + * This corresponds to bit 18 of the relevant hardware control register, + * so we match that here for efficiency. + */ + DRM_ASAHI_RENDER_DBIAS_IS_INT = (1U << 18), +}; + +/** + * struct drm_asahi_zls_buffer - Describe a depth or stencil buffer. + * + * These fields correspond to hardware registers in the ZLS (Z Load/Store) unit. + * There are three hardware registers for each field respectively for loads, + * stores, and partial renders. In practice, it makes sense to set all to the + * same values, except in exceptional cases not yet implemented in userspace, so + * we do not duplicate here for simplicity/efficiency. + * + * This struct is embedded in other structs and therefore is not extensible. + */ +struct drm_asahi_zls_buffer { + /** @base: Base address of the buffer */ + __u64 base; + + /** + * @comp_base: If the load buffer is compressed, address of the + * compression metadata section. + */ + __u64 comp_base; + + /** + * @stride: If layered rendering is enabled, the number of bytes + * between each layer of the buffer. + */ + __u32 stride; + + /** + * @comp_stride: If layered rendering is enabled, the number of bytes + * between each layer of the compression metadata. + */ + __u32 comp_stride; +}; + +/** + * struct drm_asahi_timestamp - Describe a timestamp write. + * + * The firmware can optionally write the GPU timestamp at render pass + * granularities, but it needs to be mapped specially via + * DRM_IOCTL_ASAHI_GEM_BIND_OBJECT. This structure therefore describes where to + * write as a handle-offset pair, rather than a GPU address like normal. + * + * This struct is embedded in other structs and therefore is not extensible. + */ +struct drm_asahi_timestamp { + /** + * @handle: Handle of the timestamp buffer, or 0 to skip this + * timestamp. If nonzero, this must equal the value returned in + * drm_asahi_gem_bind_object::object_handle. + */ + __u32 handle; + + /** @offset: Offset to write into the timestamp buffer */ + __u32 offset; +}; + +/** + * struct drm_asahi_timestamps - Describe timestamp writes. + * + * Each operation that can be timestamped, can be timestamped at the start and + * end. Therefore, drm_asahi_timestamp structs always come in pairs, bundled + * together into drm_asahi_timestamps. + * + * This struct is embedded in other structs and therefore is not extensible. + */ +struct drm_asahi_timestamps { + /** @start: Timestamp recorded at the start of the operation */ + struct drm_asahi_timestamp start; + + /** @end: Timestamp recorded at the end of the operation */ + struct drm_asahi_timestamp end; +}; + +/** + * struct drm_asahi_helper_program - Describe helper program configuration. + * + * The helper program is a compute-like kernel required for various hardware + * functionality. Its most important role is dynamically allocating + * scratch/stack memory for individual subgroups, by partitioning a static + * allocation shared for the whole device. It is supplied by userspace via + * drm_asahi_helper_program and internally dispatched by the hardware as needed. + * + * This struct is embedded in other structs and therefore is not extensible. + */ +struct drm_asahi_helper_program { + /** + * @binary: USC address to the helper program binary. This is a tagged + * pointer with configuration in the bottom bits. + */ + __u32 binary; + + /** @cfg: Additional configuration bits for the helper program. */ + __u32 cfg; + + /** + * @data: Data passed to the helper program. This value is not + * interpreted by the kernel, firmware, or hardware in any way. It is + * simply a sideband for userspace, set with the submit ioctl and read + * via special registers inside the helper program. + * + * In practice, userspace will pass a 64-bit GPU VA here pointing to the + * actual arguments, which presumably don't fit in 64-bits. + */ + __u64 data; +}; + +/** + * struct drm_asahi_bg_eot - Describe a background or end-of-tile program. + * + * The background and end-of-tile programs are dispatched by the hardware at the + * beginning and end of rendering. As the hardware "tilebuffer" is simply local + * memory, these programs are necessary to implement API-level render targets. + * The fragment-like background program is responsible for loading either the + * clear colour or the existing render target contents, while the compute-like + * end-of-tile program stores the tilebuffer contents to memory. + * + * This struct is embedded in other structs and therefore is not extensible. + */ +struct drm_asahi_bg_eot { + /** + * @usc: USC address of the hardware USC words binding resources + * (including images and uniforms) and the program itself. Note this is + * an additional layer of indirection compared to the helper program, + * avoiding the need for a sideband for data. This is a tagged pointer + * with additional configuration in the bottom bits. + */ + __u32 usc; + + /** + * @rsrc_spec: Resource specifier for the program. This is a packed + * hardware data structure describing the required number of registers, + * uniforms, bound textures, and bound samplers. + */ + __u32 rsrc_spec; +}; + +/** + * struct drm_asahi_cmd_render - Command to submit 3D + * + * This command submits a single render pass. The hardware control stream may + * include many draws and subpasses, but within the command, the framebuffer + * dimensions and attachments are fixed. + * + * The hardware requires the firmware to set a large number of Control Registers + * setting up state at render pass granularity before each command rendering 3D. + * The firmware bundles this state into data structures. Unfortunately, we + * cannot expose either any of that directly to userspace, because the + * kernel-firmware ABI is not stable. Although we can guarantee the firmware + * updates in tandem with the kernel, we cannot break old userspace when + * upgrading the firmware and kernel. Therefore, we need to abstract well the + * data structures to avoid tying our hands with future firmwares. + * + * The bulk of drm_asahi_cmd_render therefore consists of values of hardware + * control registers, marshalled via the firmware interface. + * + * The framebuffer/tilebuffer dimensions are also specified here. In addition to + * being passed to the firmware/hardware, the kernel requires these dimensions + * to calculate various essential tiling-related data structures. It is + * unfortunate that our submits are heavier than on vendors with saner + * hardware-software interfaces. The upshot is all of this information is + * readily available to userspace with all current APIs. + * + * It looks odd - but it's not overly burdensome and it ensures we can remain + * compatible with old userspace. + */ +struct drm_asahi_cmd_render { + /** @flags: Combination of drm_asahi_render_flags flags. */ + __u32 flags; + + /** + * @isp_zls_pixels: ISP_ZLS_PIXELS register value. This contains the + * depth/stencil width/height, which may differ from the framebuffer + * width/height. + */ + __u32 isp_zls_pixels; + + /** + * @vdm_ctrl_stream_base: VDM_CTRL_STREAM_BASE register value. GPU + * address to the beginning of the VDM control stream. + */ + __u64 vdm_ctrl_stream_base; + + /** @vertex_helper: Helper program used for the vertex shader */ + struct drm_asahi_helper_program vertex_helper; + + /** @fragment_helper: Helper program used for the fragment shader */ + struct drm_asahi_helper_program fragment_helper; + + /** + * @isp_scissor_base: ISP_SCISSOR_BASE register value. GPU address of an + * array of scissor descriptors indexed in the render pass. + */ + __u64 isp_scissor_base; + + /** + * @isp_dbias_base: ISP_DBIAS_BASE register value. GPU address of an + * array of depth bias values indexed in the render pass. + */ + __u64 isp_dbias_base; + + /** + * @isp_oclqry_base: ISP_OCLQRY_BASE register value. GPU address of an + * array of occlusion query results written by the render pass. + */ + __u64 isp_oclqry_base; + + /** @depth: Depth buffer */ + struct drm_asahi_zls_buffer depth; + + /** @stencil: Stencil buffer */ + struct drm_asahi_zls_buffer stencil; + + /** @zls_ctrl: ZLS_CTRL register value */ + __u64 zls_ctrl; + + /** @ppp_multisamplectl: PPP_MULTISAMPLECTL register value */ + __u64 ppp_multisamplectl; + + /** + * @sampler_heap: Base address of the sampler heap. This heap is used + * for both vertex shaders and fragment shaders. The registers are + * per-stage, but there is no known use case for separate heaps. + */ + __u64 sampler_heap; + + /** @ppp_ctrl: PPP_CTRL register value */ + __u32 ppp_ctrl; + + /** @width_px: Framebuffer width in pixels */ + __u16 width_px; + + /** @height_px: Framebuffer height in pixels */ + __u16 height_px; + + /** @layers: Number of layers in the framebuffer */ + __u16 layers; + + /** @sampler_count: Number of samplers in the sampler heap. */ + __u16 sampler_count; + + /** @utile_width_px: Width of a logical tilebuffer tile in pixels */ + __u8 utile_width_px; + + /** @utile_height_px: Height of a logical tilebuffer tile in pixels */ + __u8 utile_height_px; + + /** @samples: # of samples in the framebuffer. Must be 1, 2, or 4. */ + __u8 samples; + + /** @sample_size_B: # of bytes in the tilebuffer required per sample. */ + __u8 sample_size_B; + + /** + * @isp_merge_upper_x: 32-bit float used in the hardware triangle + * merging. Calculate as: tan(60 deg) * width. + * + * Making these values UAPI avoids requiring floating-point calculations + * in the kernel in the hot path. + */ + __u32 isp_merge_upper_x; + + /** + * @isp_merge_upper_y: 32-bit float. Calculate as: tan(60 deg) * height. + * See @isp_merge_upper_x. + */ + __u32 isp_merge_upper_y; + + /** @bg: Background program run for each tile at the start */ + struct drm_asahi_bg_eot bg; + + /** @eot: End-of-tile program ran for each tile at the end */ + struct drm_asahi_bg_eot eot; + + /** + * @partial_bg: Background program ran at the start of each tile when + * resuming the render pass during a partial render. + */ + struct drm_asahi_bg_eot partial_bg; + + /** + * @partial_eot: End-of-tile program ran at the end of each tile when + * pausing the render pass during a partial render. + */ + struct drm_asahi_bg_eot partial_eot; + + /** + * @isp_bgobjdepth: ISP_BGOBJDEPTH register value. This is the depth + * buffer clear value, encoded in the depth buffer's format: either a + * 32-bit float or a 16-bit unorm (with upper bits zeroed). + */ + __u32 isp_bgobjdepth; + + /** + * @isp_bgobjvals: ISP_BGOBJVALS register value. The bottom 8-bits + * contain the stencil buffer clear value. + */ + __u32 isp_bgobjvals; + + /** @ts_vtx: Timestamps for the vertex portion of the render */ + struct drm_asahi_timestamps ts_vtx; + + /** @ts_frag: Timestamps for the fragment portion of the render */ + struct drm_asahi_timestamps ts_frag; +}; + +/** + * struct drm_asahi_cmd_compute - Command to submit compute + * + * This command submits a control stream consisting of compute dispatches. There + * is essentially no limit on how many compute dispatches may be included in a + * single compute command, although timestamps are at command granularity. + */ +struct drm_asahi_cmd_compute { + /** @flags: MBZ */ + __u32 flags; + + /** @sampler_count: Number of samplers in the sampler heap. */ + __u32 sampler_count; + + /** + * @cdm_ctrl_stream_base: CDM_CTRL_STREAM_BASE register value. GPU + * address to the beginning of the CDM control stream. + */ + __u64 cdm_ctrl_stream_base; + + /** + * @cdm_ctrl_stream_end: GPU base address to the end of the hardware + * control stream. Note this only considers the first contiguous segment + * of the control stream, as the stream might jump elsewhere. + */ + __u64 cdm_ctrl_stream_end; + + /** @sampler_heap: Base address of the sampler heap. */ + __u64 sampler_heap; + + /** @helper: Helper program used for this compute command */ + struct drm_asahi_helper_program helper; + + /** @ts: Timestamps for the compute command */ + struct drm_asahi_timestamps ts; +}; + +/** + * struct drm_asahi_get_time - Arguments passed to DRM_IOCTL_ASAHI_GET_TIME + */ +struct drm_asahi_get_time { + /** @flags: MBZ. */ + __u64 flags; + + /** @gpu_timestamp: On return, the GPU timestamp in nanoseconds. */ + __u64 gpu_timestamp; +}; + +/** + * DRM_IOCTL_ASAHI() - Build an Asahi IOCTL number + * @__access: Access type. Must be R, W or RW. + * @__id: One of the DRM_ASAHI_xxx id. + * @__type: Suffix of the type being passed to the IOCTL. + * + * Don't use this macro directly, use the DRM_IOCTL_ASAHI_xxx + * values instead. + * + * Return: An IOCTL number to be passed to ioctl() from userspace. + */ +#define DRM_IOCTL_ASAHI(__access, __id, __type) \ + DRM_IO ## __access(DRM_COMMAND_BASE + DRM_ASAHI_ ## __id, \ + struct drm_asahi_ ## __type) + +/* Note: this is an enum so that it can be resolved by Rust bindgen. */ +enum { + DRM_IOCTL_ASAHI_GET_PARAMS = DRM_IOCTL_ASAHI(W, GET_PARAMS, get_params), + DRM_IOCTL_ASAHI_GET_TIME = DRM_IOCTL_ASAHI(WR, GET_TIME, get_time), + DRM_IOCTL_ASAHI_VM_CREATE = DRM_IOCTL_ASAHI(WR, VM_CREATE, vm_create), + DRM_IOCTL_ASAHI_VM_DESTROY = DRM_IOCTL_ASAHI(W, VM_DESTROY, vm_destroy), + DRM_IOCTL_ASAHI_VM_BIND = DRM_IOCTL_ASAHI(W, VM_BIND, vm_bind), + DRM_IOCTL_ASAHI_GEM_CREATE = DRM_IOCTL_ASAHI(WR, GEM_CREATE, gem_create), + DRM_IOCTL_ASAHI_GEM_MMAP_OFFSET = DRM_IOCTL_ASAHI(WR, GEM_MMAP_OFFSET, gem_mmap_offset), + DRM_IOCTL_ASAHI_GEM_BIND_OBJECT = DRM_IOCTL_ASAHI(WR, GEM_BIND_OBJECT, gem_bind_object), + DRM_IOCTL_ASAHI_QUEUE_CREATE = DRM_IOCTL_ASAHI(WR, QUEUE_CREATE, queue_create), + DRM_IOCTL_ASAHI_QUEUE_DESTROY = DRM_IOCTL_ASAHI(W, QUEUE_DESTROY, queue_destroy), + DRM_IOCTL_ASAHI_SUBMIT = DRM_IOCTL_ASAHI(W, SUBMIT, submit), +}; + +#if defined(__cplusplus) +} +#endif + +#endif /* _ASAHI_DRM_H_ */ diff --git a/include/uapi/drm/drm.h b/include/uapi/drm/drm.h index 7fba37b94401..27cc159c1d27 100644 --- a/include/uapi/drm/drm.h +++ b/include/uapi/drm/drm.h @@ -597,35 +597,66 @@ struct drm_set_version { int drm_dd_minor; }; -/* DRM_IOCTL_GEM_CLOSE ioctl argument type */ +/** + * struct drm_gem_close - Argument for &DRM_IOCTL_GEM_CLOSE ioctl. + * @handle: Handle of the object to be closed. + * @pad: Padding. + * + * Releases the handle to an mm object. + */ struct drm_gem_close { - /** Handle of the object to be closed. */ __u32 handle; __u32 pad; }; -/* DRM_IOCTL_GEM_FLINK ioctl argument type */ +/** + * struct drm_gem_flink - Argument for &DRM_IOCTL_GEM_FLINK ioctl. + * @handle: Handle for the object being named. + * @name: Returned global name. + * + * Create a global name for an object, returning the name. + * + * Note that the name does not hold a reference; when the object + * is freed, the name goes away. + */ struct drm_gem_flink { - /** Handle for the object being named */ __u32 handle; - - /** Returned global name */ __u32 name; }; -/* DRM_IOCTL_GEM_OPEN ioctl argument type */ +/** + * struct drm_gem_open - Argument for &DRM_IOCTL_GEM_OPEN ioctl. + * @name: Name of object being opened. + * @handle: Returned handle for the object. + * @size: Returned size of the object + * + * Open an object using the global name, returning a handle and the size. + * + * This handle (of course) holds a reference to the object, so the object + * will not go away until the handle is deleted. + */ struct drm_gem_open { - /** Name of object being opened */ __u32 name; - - /** Returned handle for the object */ __u32 handle; - - /** Returned size of the object */ __u64 size; }; /** + * struct drm_gem_change_handle - Argument for &DRM_IOCTL_GEM_CHANGE_HANDLE ioctl. + * @handle: The handle of a gem object. + * @new_handle: An available gem handle. + * + * This ioctl changes the handle of a GEM object to the specified one. + * The new handle must be unused. On success the old handle is closed + * and all further IOCTL should refer to the new handle only. + * Calls to DRM_IOCTL_PRIME_FD_TO_HANDLE will return the new handle. + */ +struct drm_gem_change_handle { + __u32 handle; + __u32 new_handle; +}; + +/** * DRM_CAP_DUMB_BUFFER * * If set to 1, the driver supports creating dumb buffers via the @@ -875,6 +906,21 @@ struct drm_get_cap { */ #define DRM_CLIENT_CAP_CURSOR_PLANE_HOTSPOT 6 +/** + * DRM_CLIENT_CAP_PLANE_COLOR_PIPELINE + * + * If set to 1 the DRM core will allow setting the COLOR_PIPELINE + * property on a &drm_plane, as well as drm_colorop properties. + * + * Setting of these plane properties will be rejected when this client + * cap is set: + * - COLOR_ENCODING + * - COLOR_RANGE + * + * The client must enable &DRM_CLIENT_CAP_ATOMIC first. + */ +#define DRM_CLIENT_CAP_PLANE_COLOR_PIPELINE 7 + /* DRM_IOCTL_SET_CLIENT_CAP ioctl argument type */ struct drm_set_client_cap { __u64 capability; @@ -905,13 +951,17 @@ struct drm_syncobj_destroy { }; #define DRM_SYNCOBJ_FD_TO_HANDLE_FLAGS_IMPORT_SYNC_FILE (1 << 0) +#define DRM_SYNCOBJ_FD_TO_HANDLE_FLAGS_TIMELINE (1 << 1) #define DRM_SYNCOBJ_HANDLE_TO_FD_FLAGS_EXPORT_SYNC_FILE (1 << 0) +#define DRM_SYNCOBJ_HANDLE_TO_FD_FLAGS_TIMELINE (1 << 1) struct drm_syncobj_handle { __u32 handle; __u32 flags; __s32 fd; __u32 pad; + + __u64 point; }; struct drm_syncobj_transfer { @@ -1305,6 +1355,14 @@ extern "C" { */ #define DRM_IOCTL_SET_CLIENT_NAME DRM_IOWR(0xD1, struct drm_set_client_name) +/** + * DRM_IOCTL_GEM_CHANGE_HANDLE - Move an object to a different handle + * + * Some applications (notably CRIU) need objects to have specific gem handles. + * This ioctl changes the object at one gem handle to use a new gem handle. + */ +#define DRM_IOCTL_GEM_CHANGE_HANDLE DRM_IOWR(0xD2, struct drm_gem_change_handle) + /* * Device specific ioctls should only be in their respective headers * The device specific ioctl range is from 0x40 to 0x9f. diff --git a/include/uapi/drm/drm_fourcc.h b/include/uapi/drm/drm_fourcc.h index e41a3cec6a9e..ac66fa93b5a3 100644 --- a/include/uapi/drm/drm_fourcc.h +++ b/include/uapi/drm/drm_fourcc.h @@ -210,6 +210,10 @@ extern "C" { #define DRM_FORMAT_RGBA1010102 fourcc_code('R', 'A', '3', '0') /* [31:0] R:G:B:A 10:10:10:2 little endian */ #define DRM_FORMAT_BGRA1010102 fourcc_code('B', 'A', '3', '0') /* [31:0] B:G:R:A 10:10:10:2 little endian */ +/* 48 bpp RGB */ +#define DRM_FORMAT_RGB161616 fourcc_code('R', 'G', '4', '8') /* [47:0] R:G:B 16:16:16 little endian */ +#define DRM_FORMAT_BGR161616 fourcc_code('B', 'G', '4', '8') /* [47:0] B:G:R 16:16:16 little endian */ + /* 64 bpp RGB */ #define DRM_FORMAT_XRGB16161616 fourcc_code('X', 'R', '4', '8') /* [63:0] x:R:G:B 16:16:16:16 little endian */ #define DRM_FORMAT_XBGR16161616 fourcc_code('X', 'B', '4', '8') /* [63:0] x:B:G:R 16:16:16:16 little endian */ @@ -218,7 +222,7 @@ extern "C" { #define DRM_FORMAT_ABGR16161616 fourcc_code('A', 'B', '4', '8') /* [63:0] A:B:G:R 16:16:16:16 little endian */ /* - * Floating point 64bpp RGB + * Half-Floating point - 16b/component * IEEE 754-2008 binary16 half-precision float * [15:0] sign:exponent:mantissa 1:5:10 */ @@ -228,6 +232,20 @@ extern "C" { #define DRM_FORMAT_ARGB16161616F fourcc_code('A', 'R', '4', 'H') /* [63:0] A:R:G:B 16:16:16:16 little endian */ #define DRM_FORMAT_ABGR16161616F fourcc_code('A', 'B', '4', 'H') /* [63:0] A:B:G:R 16:16:16:16 little endian */ +#define DRM_FORMAT_R16F fourcc_code('R', ' ', ' ', 'H') /* [15:0] R 16 little endian */ +#define DRM_FORMAT_GR1616F fourcc_code('G', 'R', ' ', 'H') /* [31:0] G:R 16:16 little endian */ +#define DRM_FORMAT_BGR161616F fourcc_code('B', 'G', 'R', 'H') /* [47:0] B:G:R 16:16:16 little endian */ + +/* + * Floating point - 32b/component + * IEEE 754-2008 binary32 float + * [31:0] sign:exponent:mantissa 1:8:23 + */ +#define DRM_FORMAT_R32F fourcc_code('R', ' ', ' ', 'F') /* [31:0] R 32 little endian */ +#define DRM_FORMAT_GR3232F fourcc_code('G', 'R', ' ', 'F') /* [63:0] R:G 32:32 little endian */ +#define DRM_FORMAT_BGR323232F fourcc_code('B', 'G', 'R', 'F') /* [95:0] R:G:B 32:32:32 little endian */ +#define DRM_FORMAT_ABGR32323232F fourcc_code('A', 'B', '8', 'F') /* [127:0] R:G:B:A 32:32:32:32 little endian */ + /* * RGBA format with 10-bit components packed in 64-bit per pixel, with 6 bits * of unused padding per component: @@ -378,6 +396,42 @@ extern "C" { #define DRM_FORMAT_Q401 fourcc_code('Q', '4', '0', '1') /* + * 3 plane YCbCr LSB aligned + * In order to use these formats in a similar fashion to MSB aligned ones + * implementation can multiply the values by 2^6=64. For that reason the padding + * must only contain zeros. + * index 0 = Y plane, [15:0] z:Y [6:10] little endian + * index 1 = Cb plane, [15:0] z:Cb [6:10] little endian + * index 2 = Cr plane, [15:0] z:Cr [6:10] little endian + */ +#define DRM_FORMAT_S010 fourcc_code('S', '0', '1', '0') /* 2x2 subsampled Cb (1) and Cr (2) planes 10 bits per channel */ +#define DRM_FORMAT_S210 fourcc_code('S', '2', '1', '0') /* 2x1 subsampled Cb (1) and Cr (2) planes 10 bits per channel */ +#define DRM_FORMAT_S410 fourcc_code('S', '4', '1', '0') /* non-subsampled Cb (1) and Cr (2) planes 10 bits per channel */ + +/* + * 3 plane YCbCr LSB aligned + * In order to use these formats in a similar fashion to MSB aligned ones + * implementation can multiply the values by 2^4=16. For that reason the padding + * must only contain zeros. + * index 0 = Y plane, [15:0] z:Y [4:12] little endian + * index 1 = Cb plane, [15:0] z:Cb [4:12] little endian + * index 2 = Cr plane, [15:0] z:Cr [4:12] little endian + */ +#define DRM_FORMAT_S012 fourcc_code('S', '0', '1', '2') /* 2x2 subsampled Cb (1) and Cr (2) planes 12 bits per channel */ +#define DRM_FORMAT_S212 fourcc_code('S', '2', '1', '2') /* 2x1 subsampled Cb (1) and Cr (2) planes 12 bits per channel */ +#define DRM_FORMAT_S412 fourcc_code('S', '4', '1', '2') /* non-subsampled Cb (1) and Cr (2) planes 12 bits per channel */ + +/* + * 3 plane YCbCr + * index 0 = Y plane, [15:0] Y little endian + * index 1 = Cb plane, [15:0] Cb little endian + * index 2 = Cr plane, [15:0] Cr little endian + */ +#define DRM_FORMAT_S016 fourcc_code('S', '0', '1', '6') /* 2x2 subsampled Cb (1) and Cr (2) planes 16 bits per channel */ +#define DRM_FORMAT_S216 fourcc_code('S', '2', '1', '6') /* 2x1 subsampled Cb (1) and Cr (2) planes 16 bits per channel */ +#define DRM_FORMAT_S416 fourcc_code('S', '4', '1', '6') /* non-subsampled Cb (1) and Cr (2) planes 16 bits per channel */ + +/* * 3 plane YCbCr * index 0: Y plane, [7:0] Y * index 1: Cb plane, [7:0] Cb @@ -422,6 +476,7 @@ extern "C" { #define DRM_FORMAT_MOD_VENDOR_ALLWINNER 0x09 #define DRM_FORMAT_MOD_VENDOR_AMLOGIC 0x0a #define DRM_FORMAT_MOD_VENDOR_MTK 0x0b +#define DRM_FORMAT_MOD_VENDOR_APPLE 0x0c /* add more to the end as needed */ @@ -924,14 +979,20 @@ extern "C" { * 2 = Gob Height 8, Turing+ Page Kind mapping * 3 = Reserved for future use. * - * 22:22 s Sector layout. On Tegra GPUs prior to Xavier, there is a further - * bit remapping step that occurs at an even lower level than the - * page kind and block linear swizzles. This causes the layout of - * surfaces mapped in those SOC's GPUs to be incompatible with the - * equivalent mapping on other GPUs in the same system. - * - * 0 = Tegra K1 - Tegra Parker/TX2 Layout. - * 1 = Desktop GPU and Tegra Xavier+ Layout + * 22:22 s Sector layout. There is a further bit remapping step that occurs + * 26:27 at an even lower level than the page kind and block linear + * swizzles. This causes the bit arrangement of surfaces in memory + * to differ subtly, and prevents direct sharing of surfaces between + * GPUs with different layouts. + * + * 0 = Tegra K1 - Tegra Parker/TX2 Layout + * 1 = Pre-GB20x, GB20x 32+ bpp, GB10, Tegra Xavier-Orin Layout + * 2 = GB20x(Blackwell 2)+ 8 bpp surface layout + * 3 = GB20x(Blackwell 2)+ 16 bpp surface layout + * 4 = Reserved for future use. + * 5 = Reserved for future use. + * 6 = Reserved for future use. + * 7 = Reserved for future use. * * 25:23 c Lossless Framebuffer Compression type. * @@ -946,7 +1007,7 @@ extern "C" { * 6 = Reserved for future use * 7 = Reserved for future use * - * 55:25 - Reserved for future use. Must be zero. + * 55:28 - Reserved for future use. Must be zero. */ #define DRM_FORMAT_MOD_NVIDIA_BLOCK_LINEAR_2D(c, s, g, k, h) \ fourcc_mod_code(NVIDIA, (0x10 | \ @@ -954,6 +1015,7 @@ extern "C" { (((k) & 0xff) << 12) | \ (((g) & 0x3) << 20) | \ (((s) & 0x1) << 22) | \ + (((s) & 0x6) << 25) | \ (((c) & 0x7) << 23))) /* To grandfather in prior block linear format modifiers to the above layout, @@ -1361,6 +1423,22 @@ drm_fourcc_canonicalize_nvidia_format_mod(__u64 modifier) DRM_FORMAT_MOD_ARM_CODE(DRM_FORMAT_MOD_ARM_TYPE_MISC, 1ULL) /* + * ARM 64k interleaved modifier + * + * This is used by ARM Mali v10+ GPUs. With this modifier, the plane is divided + * into 64k byte 1:1 or 2:1 -sided tiles. The 64k tiles are laid out linearly. + * Each 64k tile is divided into blocks of 16x16 texel blocks, which are + * themselves laid out linearly within a 64k tile. Then within each 16x16 + * block, texel blocks are laid out according to U order, similar to + * 16X16_BLOCK_U_INTERLEAVED. + * + * Note that unlike 16X16_BLOCK_U_INTERLEAVED, the layout does not change + * depending on whether a format is compressed or not. + */ +#define DRM_FORMAT_MOD_ARM_INTERLEAVED_64K \ + DRM_FORMAT_MOD_ARM_CODE(DRM_FORMAT_MOD_ARM_TYPE_MISC, 2ULL) + +/* * Allwinner tiled modifier * * This tiling mode is implemented by the VPU found on all Allwinner platforms, @@ -1495,6 +1573,50 @@ drm_fourcc_canonicalize_nvidia_format_mod(__u64 modifier) #define DRM_FORMAT_MOD_MTK_16L_32S_TILE DRM_FORMAT_MOD_MTK(MTK_FMT_MOD_TILE_16L32S) /* + * Apple GPU-tiled layouts. + * + * Apple GPUs support nonlinear tilings with optional lossless compression. + * + * GPU-tiled images are divided into 16KiB tiles: + * + * Bytes per pixel Tile size + * --------------- --------- + * 1 128x128 + * 2 128x64 + * 4 64x64 + * 8 64x32 + * 16 32x32 + * + * Tiles are raster-order. Pixels within a tile are interleaved (Morton order). + * + * Compressed images pad the body to 128-bytes and are immediately followed by a + * metadata section. The metadata section rounds the image dimensions to + * powers-of-two and contains 8 bytes for each 16x16 compression subtile. + * Subtiles are interleaved (Morton order). + * + * All images are 128-byte aligned. + * + * These layouts fundamentally do not have meaningful strides. No matter how we + * specify strides for these layouts, userspace unaware of Apple image layouts + * will be unable to use correctly the specified stride for any purpose. + * Userspace aware of the image layouts do not use strides. The most "correct" + * convention would be setting the image stride to 0. Unfortunately, some + * software assumes the stride is at least (width * bytes per pixel). We + * therefore require that stride equals (width * bytes per pixel). Since the + * stride is arbitrary here, we pick the simplest convention. + * + * Although containing two sections, compressed image layouts are treated in + * software as a single plane. This is modelled after AFBC, a similar + * scheme. Attempting to separate the sections to be "explicit" in DRM would + * only generate more confusion, as software does not treat the image this way. + * + * For detailed information on the hardware image layouts, see + * https://docs.mesa3d.org/drivers/asahi.html#image-layouts + */ +#define DRM_FORMAT_MOD_APPLE_GPU_TILED fourcc_mod_code(APPLE, 1) +#define DRM_FORMAT_MOD_APPLE_GPU_TILED_COMPRESSED fourcc_mod_code(APPLE, 2) + +/* * AMD modifiers * * Memory layout: diff --git a/include/uapi/drm/drm_mode.h b/include/uapi/drm/drm_mode.h index c082810c08a8..a4bdc4bd11bc 100644 --- a/include/uapi/drm/drm_mode.h +++ b/include/uapi/drm/drm_mode.h @@ -27,6 +27,9 @@ #ifndef _DRM_MODE_H #define _DRM_MODE_H +#include <linux/bits.h> +#include <linux/const.h> + #include "drm.h" #if defined(__cplusplus) @@ -166,6 +169,10 @@ extern "C" { #define DRM_MODE_LINK_STATUS_GOOD 0 #define DRM_MODE_LINK_STATUS_BAD 1 +/* Panel type property */ +#define DRM_MODE_PANEL_TYPE_UNKNOWN 0 +#define DRM_MODE_PANEL_TYPE_OLED 1 + /* * DRM_MODE_ROTATE_<degrees> * @@ -629,6 +636,7 @@ struct drm_mode_connector_set_property { #define DRM_MODE_OBJECT_FB 0xfbfbfbfb #define DRM_MODE_OBJECT_BLOB 0xbbbbbbbb #define DRM_MODE_OBJECT_PLANE 0xeeeeeeee +#define DRM_MODE_OBJECT_COLOROP 0xfafafafa #define DRM_MODE_OBJECT_ANY 0 struct drm_mode_obj_get_properties { @@ -846,6 +854,20 @@ struct drm_color_ctm { __u64 matrix[9]; }; +struct drm_color_ctm_3x4 { + /* + * Conversion matrix with 3x4 dimensions in S31.32 sign-magnitude + * (not two's complement!) format. + * + * out matrix in + * |R| |0 1 2 3 | | R | + * |G| = |4 5 6 7 | x | G | + * |B| |8 9 10 11| | B | + * |1.0| + */ + __u64 matrix[12]; +}; + struct drm_color_lut { /* * Values are mapped linearly to 0.0 - 1.0 range, with 0x0 == 0.0 and @@ -857,6 +879,125 @@ struct drm_color_lut { __u16 reserved; }; +/* + * struct drm_color_lut32 + * + * 32-bit per channel color LUT entry, similar to drm_color_lut. + */ +struct drm_color_lut32 { + __u32 red; + __u32 green; + __u32 blue; + __u32 reserved; +}; + +/** + * enum drm_colorop_type - Type of color operation + * + * drm_colorops can be of many different types. Each type behaves differently + * and defines a different set of properties. This enum defines all types and + * gives a high-level description. + */ +enum drm_colorop_type { + /** + * @DRM_COLOROP_1D_CURVE: + * + * enum string "1D Curve" + * + * A 1D curve that is being applied to all color channels. The + * curve is specified via the CURVE_1D_TYPE colorop property. + */ + DRM_COLOROP_1D_CURVE, + + /** + * @DRM_COLOROP_1D_LUT: + * + * enum string "1D LUT" + * + * A simple 1D LUT of uniformly spaced &drm_color_lut32 entries, + * packed into a blob via the DATA property. The driver's + * expected LUT size is advertised via the SIZE property. + * + * The DATA blob is an array of struct drm_color_lut32 with size + * of "size". + */ + DRM_COLOROP_1D_LUT, + + /** + * @DRM_COLOROP_CTM_3X4: + * + * enum string "3x4 Matrix" + * + * A 3x4 matrix. Its values are specified via the + * &drm_color_ctm_3x4 struct provided via the DATA property. + * + * The DATA blob is a float[12]: + * out matrix in + * | R | | 0 1 2 3 | | R | + * | G | = | 4 5 6 7 | x | G | + * | B | | 8 9 10 12 | | B | + */ + DRM_COLOROP_CTM_3X4, + + /** + * @DRM_COLOROP_MULTIPLIER: + * + * enum string "Multiplier" + * + * A simple multiplier, applied to all color values. The + * multiplier is specified as a S31.32 via the MULTIPLIER + * property. + */ + DRM_COLOROP_MULTIPLIER, + + /** + * @DRM_COLOROP_3D_LUT: + * + * enum string "3D LUT" + * + * A 3D LUT of &drm_color_lut32 entries, + * packed into a blob via the DATA property. The driver's expected + * LUT size is advertised via the SIZE property, i.e., a 3D LUT with + * 17x17x17 entries will have SIZE set to 17. + * + * The DATA blob is a 3D array of struct drm_color_lut32 with dimension + * length of "size". + * The LUT elements are traversed like so: + * + * for B in range 0..n + * for G in range 0..n + * for R in range 0..n + * index = R + n * (G + n * B) + * color = lut3d[index] + */ + DRM_COLOROP_3D_LUT, +}; + +/** + * enum drm_colorop_lut3d_interpolation_type - type of 3DLUT interpolation + */ +enum drm_colorop_lut3d_interpolation_type { + /** + * @DRM_COLOROP_LUT3D_INTERPOLATION_TETRAHEDRAL: + * + * Tetrahedral 3DLUT interpolation + */ + DRM_COLOROP_LUT3D_INTERPOLATION_TETRAHEDRAL, +}; + +/** + * enum drm_colorop_lut1d_interpolation_type - type of interpolation for 1D LUTs + */ +enum drm_colorop_lut1d_interpolation_type { + /** + * @DRM_COLOROP_LUT1D_INTERPOLATION_LINEAR: + * + * Linear interpolation. Values between points of the LUT will be + * linearly interpolated. + */ + DRM_COLOROP_LUT1D_INTERPOLATION_LINEAR, +}; + /** * struct drm_plane_size_hint - Plane size hints * @width: The width of the plane in pixel @@ -962,6 +1103,14 @@ struct hdr_output_metadata { * Request that the kernel sends back a vblank event (see * struct drm_event_vblank) with the &DRM_EVENT_FLIP_COMPLETE type when the * page-flip is done. + * + * When used with atomic uAPI, one event will be delivered per CRTC included in + * the atomic commit. A CRTC is included in an atomic commit if one of its + * properties is set, or if a property is set on a connector or plane linked + * via the CRTC_ID property to the CRTC. At least one CRTC must be included, + * and all pulled in CRTCs must be either previously or newly powered on (in + * other words, a powered off CRTC which stays off cannot be included in the + * atomic commit). */ #define DRM_MODE_PAGE_FLIP_EVENT 0x01 /** @@ -1058,7 +1207,7 @@ struct drm_mode_crtc_page_flip_target { * struct drm_mode_create_dumb - Create a KMS dumb buffer for scanout. * @height: buffer height in pixels * @width: buffer width in pixels - * @bpp: bits per pixel + * @bpp: color mode * @flags: must be zero * @handle: buffer object handle * @pitch: number of bytes between two consecutive lines @@ -1066,6 +1215,54 @@ struct drm_mode_crtc_page_flip_target { * * User-space fills @height, @width, @bpp and @flags. If the IOCTL succeeds, * the kernel fills @handle, @pitch and @size. + * + * The value of @bpp is a color-mode number describing a specific format + * or a variant thereof. The value often corresponds to the number of bits + * per pixel for most modes, although there are exceptions. Each color mode + * maps to a DRM format plus a number of modes with similar pixel layout. + * Framebuffer layout is always linear. + * + * Support for all modes and formats is optional. Even if dumb-buffer + * creation with a certain color mode succeeds, it is not guaranteed that + * the DRM driver supports any of the related formats. Most drivers support + * a color mode of 32 with a format of DRM_FORMAT_XRGB8888 on their primary + * plane. + * + * +------------+------------------------+------------------------+ + * | Color mode | Framebuffer format | Compatible formats | + * +============+========================+========================+ + * | 32 | * DRM_FORMAT_XRGB8888 | * DRM_FORMAT_BGRX8888 | + * | | | * DRM_FORMAT_RGBX8888 | + * | | | * DRM_FORMAT_XBGR8888 | + * +------------+------------------------+------------------------+ + * | 24 | * DRM_FORMAT_RGB888 | * DRM_FORMAT_BGR888 | + * +------------+------------------------+------------------------+ + * | 16 | * DRM_FORMAT_RGB565 | * DRM_FORMAT_BGR565 | + * +------------+------------------------+------------------------+ + * | 15 | * DRM_FORMAT_XRGB1555 | * DRM_FORMAT_BGRX1555 | + * | | | * DRM_FORMAT_RGBX1555 | + * | | | * DRM_FORMAT_XBGR1555 | + * +------------+------------------------+------------------------+ + * | 8 | * DRM_FORMAT_C8 | * DRM_FORMAT_D8 | + * | | | * DRM_FORMAT_R8 | + * +------------+------------------------+------------------------+ + * | 4 | * DRM_FORMAT_C4 | * DRM_FORMAT_D4 | + * | | | * DRM_FORMAT_R4 | + * +------------+------------------------+------------------------+ + * | 2 | * DRM_FORMAT_C2 | * DRM_FORMAT_D2 | + * | | | * DRM_FORMAT_R2 | + * +------------+------------------------+------------------------+ + * | 1 | * DRM_FORMAT_C1 | * DRM_FORMAT_D1 | + * | | | * DRM_FORMAT_R1 | + * +------------+------------------------+------------------------+ + * + * Color modes of 10, 12, 15, 30 and 64 are only supported for use by + * legacy user space. Please don't use them in new code. Other modes + * are not support. + * + * Do not attempt to allocate anything but linear framebuffer memory + * with single-plane RGB data. Allocation of other framebuffer + * layouts requires dedicated ioctls in the respective DRM driver. */ struct drm_mode_create_dumb { __u32 height; @@ -1355,6 +1552,83 @@ struct drm_mode_closefb { __u32 pad; }; +/* + * Put 16-bit ARGB values into a standard 64-bit representation that can be + * used for ioctl parameters, inter-driver communication, etc. + * + * If the component values being provided contain less than 16 bits of + * precision, use a conversion ratio to get a better color approximation. + * The ratio is computed as (2^16 - 1) / (2^bpc - 1), where bpc and 16 are + * the input and output precision, respectively. + * Also note bpc must be greater than 0. + */ +#define __DRM_ARGB64_PREP(c, shift) \ + (((__u64)(c) & __GENMASK(15, 0)) << (shift)) + +#define __DRM_ARGB64_PREP_BPC(c, shift, bpc) \ +({ \ + __u16 mask = __GENMASK((bpc) - 1, 0); \ + __u16 conv = __KERNEL_DIV_ROUND_CLOSEST((mask & (c)) * \ + __GENMASK(15, 0), mask);\ + __DRM_ARGB64_PREP(conv, shift); \ +}) + +#define DRM_ARGB64_PREP(alpha, red, green, blue) \ +( \ + __DRM_ARGB64_PREP(alpha, 48) | \ + __DRM_ARGB64_PREP(red, 32) | \ + __DRM_ARGB64_PREP(green, 16) | \ + __DRM_ARGB64_PREP(blue, 0) \ +) + +#define DRM_ARGB64_PREP_BPC(alpha, red, green, blue, bpc) \ +({ \ + __typeof__(bpc) __bpc = bpc; \ + __DRM_ARGB64_PREP_BPC(alpha, 48, __bpc) | \ + __DRM_ARGB64_PREP_BPC(red, 32, __bpc) | \ + __DRM_ARGB64_PREP_BPC(green, 16, __bpc) | \ + __DRM_ARGB64_PREP_BPC(blue, 0, __bpc); \ +}) + +/* + * Extract the specified color component from a standard 64-bit ARGB value. + * + * If the requested precision is less than 16 bits, make use of a conversion + * ratio calculated as (2^bpc - 1) / (2^16 - 1), where bpc and 16 are the + * output and input precision, respectively. + * + * If speed is more important than accuracy, use DRM_ARGB64_GET*_BPCS() + * instead of DRM_ARGB64_GET*_BPC() in order to replace the expensive + * division with a simple bit right-shift operation. + */ +#define __DRM_ARGB64_GET(c, shift) \ + ((__u16)(((__u64)(c) >> (shift)) & __GENMASK(15, 0))) + +#define __DRM_ARGB64_GET_BPC(c, shift, bpc) \ +({ \ + __u16 comp = __DRM_ARGB64_GET(c, shift); \ + __KERNEL_DIV_ROUND_CLOSEST(comp * __GENMASK((bpc) - 1, 0), \ + __GENMASK(15, 0)); \ +}) + +#define __DRM_ARGB64_GET_BPCS(c, shift, bpc) \ + (__DRM_ARGB64_GET(c, shift) >> (16 - (bpc))) + +#define DRM_ARGB64_GETA(c) __DRM_ARGB64_GET(c, 48) +#define DRM_ARGB64_GETR(c) __DRM_ARGB64_GET(c, 32) +#define DRM_ARGB64_GETG(c) __DRM_ARGB64_GET(c, 16) +#define DRM_ARGB64_GETB(c) __DRM_ARGB64_GET(c, 0) + +#define DRM_ARGB64_GETA_BPC(c, bpc) __DRM_ARGB64_GET_BPC(c, 48, bpc) +#define DRM_ARGB64_GETR_BPC(c, bpc) __DRM_ARGB64_GET_BPC(c, 32, bpc) +#define DRM_ARGB64_GETG_BPC(c, bpc) __DRM_ARGB64_GET_BPC(c, 16, bpc) +#define DRM_ARGB64_GETB_BPC(c, bpc) __DRM_ARGB64_GET_BPC(c, 0, bpc) + +#define DRM_ARGB64_GETA_BPCS(c, bpc) __DRM_ARGB64_GET_BPCS(c, 48, bpc) +#define DRM_ARGB64_GETR_BPCS(c, bpc) __DRM_ARGB64_GET_BPCS(c, 32, bpc) +#define DRM_ARGB64_GETG_BPCS(c, bpc) __DRM_ARGB64_GET_BPCS(c, 16, bpc) +#define DRM_ARGB64_GETB_BPCS(c, bpc) __DRM_ARGB64_GET_BPCS(c, 0, bpc) + #if defined(__cplusplus) } #endif diff --git a/include/uapi/drm/drm_ras.h b/include/uapi/drm/drm_ras.h new file mode 100644 index 000000000000..5f40fa5b869d --- /dev/null +++ b/include/uapi/drm/drm_ras.h @@ -0,0 +1,49 @@ +/* SPDX-License-Identifier: ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) */ +/* Do not edit directly, auto-generated from: */ +/* Documentation/netlink/specs/drm_ras.yaml */ +/* YNL-GEN uapi header */ +/* To regenerate run: tools/net/ynl/ynl-regen.sh */ + +#ifndef _UAPI_LINUX_DRM_RAS_H +#define _UAPI_LINUX_DRM_RAS_H + +#define DRM_RAS_FAMILY_NAME "drm-ras" +#define DRM_RAS_FAMILY_VERSION 1 + +/* + * Type of the node. Currently, only error-counter nodes are supported, which + * expose reliability counters for a hardware/software component. + */ +enum drm_ras_node_type { + DRM_RAS_NODE_TYPE_ERROR_COUNTER = 1, +}; + +enum { + DRM_RAS_A_NODE_ATTRS_NODE_ID = 1, + DRM_RAS_A_NODE_ATTRS_DEVICE_NAME, + DRM_RAS_A_NODE_ATTRS_NODE_NAME, + DRM_RAS_A_NODE_ATTRS_NODE_TYPE, + + __DRM_RAS_A_NODE_ATTRS_MAX, + DRM_RAS_A_NODE_ATTRS_MAX = (__DRM_RAS_A_NODE_ATTRS_MAX - 1) +}; + +enum { + DRM_RAS_A_ERROR_COUNTER_ATTRS_NODE_ID = 1, + DRM_RAS_A_ERROR_COUNTER_ATTRS_ERROR_ID, + DRM_RAS_A_ERROR_COUNTER_ATTRS_ERROR_NAME, + DRM_RAS_A_ERROR_COUNTER_ATTRS_ERROR_VALUE, + + __DRM_RAS_A_ERROR_COUNTER_ATTRS_MAX, + DRM_RAS_A_ERROR_COUNTER_ATTRS_MAX = (__DRM_RAS_A_ERROR_COUNTER_ATTRS_MAX - 1) +}; + +enum { + DRM_RAS_CMD_LIST_NODES = 1, + DRM_RAS_CMD_GET_ERROR_COUNTER, + + __DRM_RAS_CMD_MAX, + DRM_RAS_CMD_MAX = (__DRM_RAS_CMD_MAX - 1) +}; + +#endif /* _UAPI_LINUX_DRM_RAS_H */ diff --git a/include/uapi/drm/ethosu_accel.h b/include/uapi/drm/ethosu_accel.h new file mode 100644 index 000000000000..af78bb4686d7 --- /dev/null +++ b/include/uapi/drm/ethosu_accel.h @@ -0,0 +1,261 @@ +/* SPDX-License-Identifier: MIT */ +/* Copyright (C) 2025 Arm, Ltd. */ +#ifndef _ETHOSU_DRM_H_ +#define _ETHOSU_DRM_H_ + +#include "drm.h" + +#if defined(__cplusplus) +extern "C" { +#endif + +/** + * DOC: IOCTL IDs + * + * enum drm_ethosu_ioctl_id - IOCTL IDs + * + * Place new ioctls at the end, don't re-order, don't replace or remove entries. + * + * These IDs are not meant to be used directly. Use the DRM_IOCTL_ETHOSU_xxx + * definitions instead. + */ +enum drm_ethosu_ioctl_id { + /** @DRM_ETHOSU_DEV_QUERY: Query device information. */ + DRM_ETHOSU_DEV_QUERY = 0, + + /** @DRM_ETHOSU_BO_CREATE: Create a buffer object. */ + DRM_ETHOSU_BO_CREATE, + + /** @DRM_ETHOSU_BO_WAIT: Wait on a buffer object's fence. */ + DRM_ETHOSU_BO_WAIT, + + /** + * @DRM_ETHOSU_BO_MMAP_OFFSET: Get the file offset to pass to + * mmap to map a GEM object. + */ + DRM_ETHOSU_BO_MMAP_OFFSET, + + /** + * @DRM_ETHOSU_CMDSTREAM_BO_CREATE: Create a command stream buffer + * object. + */ + DRM_ETHOSU_CMDSTREAM_BO_CREATE, + + /** @DRM_ETHOSU_SUBMIT: Submit a job and BOs to run. */ + DRM_ETHOSU_SUBMIT, +}; + +/** + * DOC: IOCTL arguments + */ + +/** + * enum drm_ethosu_dev_query_type - Query type + * + * Place new types at the end, don't re-order, don't remove or replace. + */ +enum drm_ethosu_dev_query_type { + /** @DRM_ETHOSU_DEV_QUERY_NPU_INFO: Query NPU information. */ + DRM_ETHOSU_DEV_QUERY_NPU_INFO = 0, +}; + +/** + * struct drm_ethosu_gpu_info - NPU information + * + * Structure grouping all queryable information relating to the NPU. + */ +struct drm_ethosu_npu_info { + /** @id : NPU ID. */ + __u32 id; +#define DRM_ETHOSU_ARCH_MAJOR(x) ((x) >> 28) +#define DRM_ETHOSU_ARCH_MINOR(x) (((x) >> 20) & 0xff) +#define DRM_ETHOSU_ARCH_PATCH(x) (((x) >> 16) & 0xf) +#define DRM_ETHOSU_PRODUCT_MAJOR(x) (((x) >> 12) & 0xf) +#define DRM_ETHOSU_VERSION_MAJOR(x) (((x) >> 8) & 0xf) +#define DRM_ETHOSU_VERSION_MINOR(x) (((x) >> 4) & 0xff) +#define DRM_ETHOSU_VERSION_STATUS(x) ((x) & 0xf) + + /** @gpu_rev: GPU revision. */ + __u32 config; + + __u32 sram_size; +}; + +/** + * struct drm_ethosu_dev_query - Arguments passed to DRM_ETHOSU_IOCTL_DEV_QUERY + */ +struct drm_ethosu_dev_query { + /** @type: the query type (see drm_ethosu_dev_query_type). */ + __u32 type; + + /** + * @size: size of the type being queried. + * + * If pointer is NULL, size is updated by the driver to provide the + * output structure size. If pointer is not NULL, the driver will + * only copy min(size, actual_structure_size) bytes to the pointer, + * and update the size accordingly. This allows us to extend query + * types without breaking userspace. + */ + __u32 size; + + /** + * @pointer: user pointer to a query type struct. + * + * Pointer can be NULL, in which case, nothing is copied, but the + * actual structure size is returned. If not NULL, it must point to + * a location that's large enough to hold size bytes. + */ + __u64 pointer; +}; + +/** + * enum drm_ethosu_bo_flags - Buffer object flags, passed at creation time. + */ +enum drm_ethosu_bo_flags { + /** + * @DRM_ETHOSU_BO_NO_MMAP: The buffer object will never be CPU-mapped + * in userspace. + */ + DRM_ETHOSU_BO_NO_MMAP = (1 << 0), +}; + +/** + * struct drm_ethosu_bo_create - Arguments passed to DRM_IOCTL_ETHOSU_BO_CREATE. + */ +struct drm_ethosu_bo_create { + /** + * @size: Requested size for the object + * + * The (page-aligned) allocated size for the object will be returned. + */ + __u64 size; + + /** + * @flags: Flags. Must be a combination of drm_ethosu_bo_flags flags. + */ + __u32 flags; + + /** + * @handle: Returned handle for the object. + * + * Object handles are nonzero. + */ + __u32 handle; +}; + +/** + * struct drm_ethosu_bo_mmap_offset - Arguments passed to DRM_IOCTL_ETHOSU_BO_MMAP_OFFSET. + */ +struct drm_ethosu_bo_mmap_offset { + /** @handle: Handle of the object we want an mmap offset for. */ + __u32 handle; + + /** @pad: MBZ. */ + __u32 pad; + + /** @offset: The fake offset to use for subsequent mmap calls. */ + __u64 offset; +}; + +/** + * struct drm_ethosu_wait_bo - ioctl argument for waiting for + * completion of the last DRM_ETHOSU_SUBMIT on a BO. + * + * This is useful for cases where multiple processes might be + * rendering to a BO and you want to wait for all rendering to be + * completed. + */ +struct drm_ethosu_bo_wait { + __u32 handle; + __u32 pad; + __s64 timeout_ns; /* absolute */ +}; + +struct drm_ethosu_cmdstream_bo_create { + /* Size of the data argument. */ + __u32 size; + + /* Flags, currently must be 0. */ + __u32 flags; + + /* Pointer to the data. */ + __u64 data; + + /** Returned GEM handle for the BO. */ + __u32 handle; + + /* Pad, must be 0. */ + __u32 pad; +}; + +/** + * struct drm_ethosu_job - A job to be run on the NPU + * + * The kernel will schedule the execution of this job taking into account its + * dependencies with other jobs. All tasks in the same job will be executed + * sequentially on the same core, to benefit from memory residency in SRAM. + */ +struct drm_ethosu_job { + /** Input: BO handle for cmdstream. */ + __u32 cmd_bo; + + /** Input: Amount of SRAM to use. */ + __u32 sram_size; + +#define ETHOSU_MAX_REGIONS 8 + /** Input: Array of BO handles for each region. */ + __u32 region_bo_handles[ETHOSU_MAX_REGIONS]; +}; + +/** + * struct drm_ethosu_submit - ioctl argument for submitting commands to the NPU. + * + * The kernel will schedule the execution of these jobs in dependency order. + */ +struct drm_ethosu_submit { + /** Input: Pointer to an array of struct drm_ethosu_job. */ + __u64 jobs; + + /** Input: Number of jobs passed in. */ + __u32 job_count; + + /** Reserved, must be zero. */ + __u32 pad; +}; + +/** + * DRM_IOCTL_ETHOSU() - Build a ethosu IOCTL number + * @__access: Access type. Must be R, W or RW. + * @__id: One of the DRM_ETHOSU_xxx id. + * @__type: Suffix of the type being passed to the IOCTL. + * + * Don't use this macro directly, use the DRM_IOCTL_ETHOSU_xxx + * values instead. + * + * Return: An IOCTL number to be passed to ioctl() from userspace. + */ +#define DRM_IOCTL_ETHOSU(__access, __id, __type) \ + DRM_IO ## __access(DRM_COMMAND_BASE + DRM_ETHOSU_ ## __id, \ + struct drm_ethosu_ ## __type) + +enum { + DRM_IOCTL_ETHOSU_DEV_QUERY = + DRM_IOCTL_ETHOSU(WR, DEV_QUERY, dev_query), + DRM_IOCTL_ETHOSU_BO_CREATE = + DRM_IOCTL_ETHOSU(WR, BO_CREATE, bo_create), + DRM_IOCTL_ETHOSU_BO_WAIT = + DRM_IOCTL_ETHOSU(WR, BO_WAIT, bo_wait), + DRM_IOCTL_ETHOSU_BO_MMAP_OFFSET = + DRM_IOCTL_ETHOSU(WR, BO_MMAP_OFFSET, bo_mmap_offset), + DRM_IOCTL_ETHOSU_CMDSTREAM_BO_CREATE = + DRM_IOCTL_ETHOSU(WR, CMDSTREAM_BO_CREATE, cmdstream_bo_create), + DRM_IOCTL_ETHOSU_SUBMIT = + DRM_IOCTL_ETHOSU(WR, SUBMIT, submit), +}; + +#if defined(__cplusplus) +} +#endif + +#endif /* _ETHOSU_DRM_H_ */ diff --git a/include/uapi/drm/ivpu_accel.h b/include/uapi/drm/ivpu_accel.h index 746c43bd3eb6..264505d54f93 100644 --- a/include/uapi/drm/ivpu_accel.h +++ b/include/uapi/drm/ivpu_accel.h @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: GPL-2.0-only WITH Linux-syscall-note */ /* - * Copyright (C) 2020-2024 Intel Corporation + * Copyright (C) 2020-2025 Intel Corporation */ #ifndef __UAPI_IVPU_DRM_H__ @@ -25,6 +25,7 @@ extern "C" { #define DRM_IVPU_CMDQ_CREATE 0x0b #define DRM_IVPU_CMDQ_DESTROY 0x0c #define DRM_IVPU_CMDQ_SUBMIT 0x0d +#define DRM_IVPU_BO_CREATE_FROM_USERPTR 0x0e #define DRM_IOCTL_IVPU_GET_PARAM \ DRM_IOWR(DRM_COMMAND_BASE + DRM_IVPU_GET_PARAM, struct drm_ivpu_param) @@ -69,6 +70,10 @@ extern "C" { #define DRM_IOCTL_IVPU_CMDQ_SUBMIT \ DRM_IOW(DRM_COMMAND_BASE + DRM_IVPU_CMDQ_SUBMIT, struct drm_ivpu_cmdq_submit) +#define DRM_IOCTL_IVPU_BO_CREATE_FROM_USERPTR \ + DRM_IOWR(DRM_COMMAND_BASE + DRM_IVPU_BO_CREATE_FROM_USERPTR, \ + struct drm_ivpu_bo_create_from_userptr) + /** * DOC: contexts * @@ -90,6 +95,7 @@ extern "C" { #define DRM_IVPU_PARAM_TILE_CONFIG 11 #define DRM_IVPU_PARAM_SKU 12 #define DRM_IVPU_PARAM_CAPABILITIES 13 +#define DRM_IVPU_PARAM_PREEMPT_BUFFER_SIZE 14 #define DRM_IVPU_PLATFORM_TYPE_SILICON 0 @@ -126,6 +132,13 @@ extern "C" { * command queue destroy and submit job on specific command queue. */ #define DRM_IVPU_CAP_MANAGE_CMDQ 3 +/** + * DRM_IVPU_CAP_BO_CREATE_FROM_USERPTR + * + * Driver supports creating buffer objects from user space memory pointers. + * This allows creating GEM buffers from existing user memory regions. + */ +#define DRM_IVPU_CAP_BO_CREATE_FROM_USERPTR 4 /** * struct drm_ivpu_param - Get/Set VPU parameters @@ -147,7 +160,7 @@ struct drm_ivpu_param { * platform type when executing on a simulator or emulator (read-only) * * %DRM_IVPU_PARAM_CORE_CLOCK_RATE: - * Current PLL frequency (read-only) + * Maximum frequency of the NPU data processing unit clock (read-only) * * %DRM_IVPU_PARAM_NUM_CONTEXTS: * Maximum number of simultaneously existing contexts (read-only) @@ -176,6 +189,9 @@ struct drm_ivpu_param { * * %DRM_IVPU_PARAM_CAPABILITIES: * Supported capabilities (read-only) + * + * %DRM_IVPU_PARAM_PREEMPT_BUFFER_SIZE: + * Size of the preemption buffer (read-only) */ __u32 param; @@ -190,6 +206,7 @@ struct drm_ivpu_param { #define DRM_IVPU_BO_HIGH_MEM DRM_IVPU_BO_SHAVE_MEM #define DRM_IVPU_BO_MAPPABLE 0x00000002 #define DRM_IVPU_BO_DMA_MEM 0x00000004 +#define DRM_IVPU_BO_READ_ONLY 0x00000008 #define DRM_IVPU_BO_CACHED 0x00000000 #define DRM_IVPU_BO_UNCACHED 0x00010000 @@ -200,6 +217,7 @@ struct drm_ivpu_param { (DRM_IVPU_BO_HIGH_MEM | \ DRM_IVPU_BO_MAPPABLE | \ DRM_IVPU_BO_DMA_MEM | \ + DRM_IVPU_BO_READ_ONLY | \ DRM_IVPU_BO_CACHE_MASK) /** @@ -252,6 +270,44 @@ struct drm_ivpu_bo_create { }; /** + * struct drm_ivpu_bo_create_from_userptr - Create dma-buf from user pointer + * + * Create a GEM buffer object from a user pointer to a memory region. + */ +struct drm_ivpu_bo_create_from_userptr { + /** @user_ptr: User pointer to memory region (must be page aligned) */ + __u64 user_ptr; + + /** @size: Size of the memory region in bytes (must be page aligned) */ + __u64 size; + + /** + * @flags: + * + * Supported flags: + * + * %DRM_IVPU_BO_HIGH_MEM: + * + * Allocate VPU address from >4GB range. + * + * %DRM_IVPU_BO_DMA_MEM: + * + * Allocate from DMA memory range accessible by hardware DMA. + * + * %DRM_IVPU_BO_READ_ONLY: + * + * Allocate as a read-only buffer object. + */ + __u32 flags; + + /** @handle: Returned GEM object handle */ + __u32 handle; + + /** @vpu_addr: Returned VPU virtual address */ + __u64 vpu_addr; +}; + +/** * struct drm_ivpu_bo_info - Query buffer object info */ struct drm_ivpu_bo_info { @@ -371,6 +427,13 @@ struct drm_ivpu_cmdq_submit { * to be executed. The offset has to be 8-byte aligned. */ __u32 commands_offset; + /** + * @preempt_buffer_index: + * + * Index of the preemption buffer in the buffers_ptr array. + */ + __u32 preempt_buffer_index; + __u32 reserved; }; /* drm_ivpu_bo_wait job status codes */ @@ -445,6 +508,9 @@ struct drm_ivpu_metric_streamer_get_data { __u64 data_size; }; +/* Command queue flags */ +#define DRM_IVPU_CMDQ_FLAG_TURBO 0x00000001 + /** * struct drm_ivpu_cmdq_create - Create command queue for job submission */ @@ -462,6 +528,17 @@ struct drm_ivpu_cmdq_create { * %DRM_IVPU_JOB_PRIORITY_REALTIME */ __u32 priority; + /** + * @flags: + * + * Supported flags: + * + * %DRM_IVPU_CMDQ_FLAG_TURBO + * + * Enable low-latency mode for the command queue. The NPU will maximize performance + * when executing jobs from such queue at the cost of increased power usage. + */ + __u32 flags; }; /** diff --git a/include/uapi/drm/msm_drm.h b/include/uapi/drm/msm_drm.h index 2342cb90857e..b99098792371 100644 --- a/include/uapi/drm/msm_drm.h +++ b/include/uapi/drm/msm_drm.h @@ -91,6 +91,33 @@ struct drm_msm_timespec { #define MSM_PARAM_UBWC_SWIZZLE 0x12 /* RO */ #define MSM_PARAM_MACROTILE_MODE 0x13 /* RO */ #define MSM_PARAM_UCHE_TRAP_BASE 0x14 /* RO */ +/* PRR (Partially Resident Region) is required for sparse residency: */ +#define MSM_PARAM_HAS_PRR 0x15 /* RO */ +/* MSM_PARAM_EN_VM_BIND is set to 1 to enable VM_BIND ops. + * + * With VM_BIND enabled, userspace is required to allocate iova and use the + * VM_BIND ops for map/unmap ioctls. MSM_INFO_SET_IOVA and MSM_INFO_GET_IOVA + * will be rejected. (The latter does not have a sensible meaning when a BO + * can have multiple and/or partial mappings.) + * + * With VM_BIND enabled, userspace does not include a submit_bo table in the + * SUBMIT ioctl (this will be rejected), the resident set is determined by + * the the VM_BIND ops. + * + * Enabling VM_BIND will fail on devices which do not have per-process pgtables. + * And it is not allowed to disable VM_BIND once it has been enabled. + * + * Enabling VM_BIND should be done (attempted) prior to allocating any BOs or + * submitqueues of type MSM_SUBMITQUEUE_VM_BIND. + * + * Relatedly, when VM_BIND mode is enabled, the kernel will not try to recover + * from GPU faults or failed async VM_BIND ops, in particular because it is + * difficult to communicate to userspace which op failed so that userspace + * could rewind and try again. When the VM is marked unusable, the SUBMIT + * ioctl will throw -EPIPE. + */ +#define MSM_PARAM_EN_VM_BIND 0x16 /* WO, once */ +#define MSM_PARAM_AQE 0x17 /* RO */ /* For backwards compat. The original support for preemption was based on * a single ring per priority level so # of priority levels equals the # @@ -114,6 +141,19 @@ struct drm_msm_param { #define MSM_BO_SCANOUT 0x00000001 /* scanout capable */ #define MSM_BO_GPU_READONLY 0x00000002 +/* Private buffers do not need to be explicitly listed in the SUBMIT + * ioctl, unless referenced by a drm_msm_gem_submit_cmd. Private + * buffers may NOT be imported/exported or used for scanout (or any + * other situation where buffers can be indefinitely pinned, but + * cases other than scanout are all kernel owned BOs which are not + * visible to userspace). + * + * In exchange for those constraints, all private BOs associated with + * a single context (drm_file) share a single dma_resv, and if there + * has been no eviction since the last submit, there are no per-BO + * bookeeping to do, significantly cutting the SUBMIT overhead. + */ +#define MSM_BO_NO_SHARE 0x00000004 #define MSM_BO_CACHE_MASK 0x000f0000 /* cache modes */ #define MSM_BO_CACHED 0x00010000 @@ -123,6 +163,7 @@ struct drm_msm_param { #define MSM_BO_FLAGS (MSM_BO_SCANOUT | \ MSM_BO_GPU_READONLY | \ + MSM_BO_NO_SHARE | \ MSM_BO_CACHE_MASK) struct drm_msm_gem_new { @@ -180,6 +221,17 @@ struct drm_msm_gem_cpu_fini { * Cmdstream Submission: */ +#define MSM_SYNCOBJ_RESET 0x00000001 /* Reset syncobj after wait. */ +#define MSM_SYNCOBJ_FLAGS ( \ + MSM_SYNCOBJ_RESET | \ + 0) + +struct drm_msm_syncobj { + __u32 handle; /* in, syncobj handle. */ + __u32 flags; /* in, from MSM_SUBMIT_SYNCOBJ_FLAGS */ + __u64 point; /* in, timepoint for timeline syncobjs. */ +}; + /* The value written into the cmdstream is logically: * * ((relocbuf->gpuaddr + reloc_offset) << shift) | or @@ -221,7 +273,10 @@ struct drm_msm_gem_submit_cmd { __u32 size; /* in, cmdstream size */ __u32 pad; __u32 nr_relocs; /* in, number of submit_reloc's */ - __u64 relocs; /* in, ptr to array of submit_reloc's */ + union { + __u64 relocs; /* in, ptr to array of submit_reloc's */ + __u64 iova; /* cmdstream address (for VM_BIND contexts) */ + }; }; /* Each buffer referenced elsewhere in the cmdstream submit (ie. the @@ -269,17 +324,6 @@ struct drm_msm_gem_submit_bo { MSM_SUBMIT_FENCE_SN_IN | \ 0) -#define MSM_SUBMIT_SYNCOBJ_RESET 0x00000001 /* Reset syncobj after wait. */ -#define MSM_SUBMIT_SYNCOBJ_FLAGS ( \ - MSM_SUBMIT_SYNCOBJ_RESET | \ - 0) - -struct drm_msm_gem_submit_syncobj { - __u32 handle; /* in, syncobj handle. */ - __u32 flags; /* in, from MSM_SUBMIT_SYNCOBJ_FLAGS */ - __u64 point; /* in, timepoint for timeline syncobjs. */ -}; - /* Each cmdstream submit consists of a table of buffers involved, and * one or more cmdstream buffers. This allows for conditional execution * (context-restore), and IB buffers needed for per tile/bin draw cmds. @@ -293,13 +337,80 @@ struct drm_msm_gem_submit { __u64 cmds; /* in, ptr to array of submit_cmd's */ __s32 fence_fd; /* in/out fence fd (see MSM_SUBMIT_FENCE_FD_IN/OUT) */ __u32 queueid; /* in, submitqueue id */ - __u64 in_syncobjs; /* in, ptr to array of drm_msm_gem_submit_syncobj */ - __u64 out_syncobjs; /* in, ptr to array of drm_msm_gem_submit_syncobj */ + __u64 in_syncobjs; /* in, ptr to array of drm_msm_syncobj */ + __u64 out_syncobjs; /* in, ptr to array of drm_msm_syncobj */ __u32 nr_in_syncobjs; /* in, number of entries in in_syncobj */ __u32 nr_out_syncobjs; /* in, number of entries in out_syncobj. */ __u32 syncobj_stride; /* in, stride of syncobj arrays. */ __u32 pad; /*in, reserved for future use, always 0. */ +}; +#define MSM_VM_BIND_OP_UNMAP 0 +#define MSM_VM_BIND_OP_MAP 1 +#define MSM_VM_BIND_OP_MAP_NULL 2 + +#define MSM_VM_BIND_OP_DUMP 1 +#define MSM_VM_BIND_OP_FLAGS ( \ + MSM_VM_BIND_OP_DUMP | \ + 0) + +/** + * struct drm_msm_vm_bind_op - bind/unbind op to run + */ +struct drm_msm_vm_bind_op { + /** @op: one of MSM_VM_BIND_OP_x */ + __u32 op; + /** @handle: GEM object handle, MBZ for UNMAP or MAP_NULL */ + __u32 handle; + /** @obj_offset: Offset into GEM object, MBZ for UNMAP or MAP_NULL */ + __u64 obj_offset; + /** @iova: Address to operate on */ + __u64 iova; + /** @range: Number of bites to to map/unmap */ + __u64 range; + /** @flags: Bitmask of MSM_VM_BIND_OP_FLAG_x */ + __u32 flags; + /** @pad: MBZ */ + __u32 pad; +}; + +#define MSM_VM_BIND_FENCE_FD_IN 0x00000001 +#define MSM_VM_BIND_FENCE_FD_OUT 0x00000002 +#define MSM_VM_BIND_FLAGS ( \ + MSM_VM_BIND_FENCE_FD_IN | \ + MSM_VM_BIND_FENCE_FD_OUT | \ + 0) + +/** + * struct drm_msm_vm_bind - Input of &DRM_IOCTL_MSM_VM_BIND + */ +struct drm_msm_vm_bind { + /** @flags: in, bitmask of MSM_VM_BIND_x */ + __u32 flags; + /** @nr_ops: the number of bind ops in this ioctl */ + __u32 nr_ops; + /** @fence_fd: in/out fence fd (see MSM_VM_BIND_FENCE_FD_IN/OUT) */ + __s32 fence_fd; + /** @queue_id: in, submitqueue id */ + __u32 queue_id; + /** @in_syncobjs: in, ptr to array of drm_msm_gem_syncobj */ + __u64 in_syncobjs; + /** @out_syncobjs: in, ptr to array of drm_msm_gem_syncobj */ + __u64 out_syncobjs; + /** @nr_in_syncobjs: in, number of entries in in_syncobj */ + __u32 nr_in_syncobjs; + /** @nr_out_syncobjs: in, number of entries in out_syncobj */ + __u32 nr_out_syncobjs; + /** @syncobj_stride: in, stride of syncobj arrays */ + __u32 syncobj_stride; + /** @op_stride: sizeof each struct drm_msm_vm_bind_op in @ops */ + __u32 op_stride; + union { + /** @op: used if num_ops == 1 */ + struct drm_msm_vm_bind_op op; + /** @ops: userptr to array of drm_msm_vm_bind_op if num_ops > 1 */ + __u64 ops; + }; }; #define MSM_WAIT_FENCE_BOOST 0x00000001 @@ -345,12 +456,19 @@ struct drm_msm_gem_madvise { /* * Draw queues allow the user to set specific submission parameter. Command * submissions specify a specific submitqueue to use. ID 0 is reserved for - * backwards compatibility as a "default" submitqueue + * backwards compatibility as a "default" submitqueue. + * + * Because VM_BIND async updates happen on the CPU, they must run on a + * virtual queue created with the flag MSM_SUBMITQUEUE_VM_BIND. If we had + * a way to do pgtable updates on the GPU, we could drop this restriction. */ #define MSM_SUBMITQUEUE_ALLOW_PREEMPT 0x00000001 +#define MSM_SUBMITQUEUE_VM_BIND 0x00000002 /* virtual queue for VM_BIND ops */ + #define MSM_SUBMITQUEUE_FLAGS ( \ MSM_SUBMITQUEUE_ALLOW_PREEMPT | \ + MSM_SUBMITQUEUE_VM_BIND | \ 0) /* @@ -388,6 +506,7 @@ struct drm_msm_submitqueue_query { #define DRM_MSM_SUBMITQUEUE_NEW 0x0A #define DRM_MSM_SUBMITQUEUE_CLOSE 0x0B #define DRM_MSM_SUBMITQUEUE_QUERY 0x0C +#define DRM_MSM_VM_BIND 0x0D #define DRM_IOCTL_MSM_GET_PARAM DRM_IOWR(DRM_COMMAND_BASE + DRM_MSM_GET_PARAM, struct drm_msm_param) #define DRM_IOCTL_MSM_SET_PARAM DRM_IOW (DRM_COMMAND_BASE + DRM_MSM_SET_PARAM, struct drm_msm_param) @@ -401,6 +520,7 @@ struct drm_msm_submitqueue_query { #define DRM_IOCTL_MSM_SUBMITQUEUE_NEW DRM_IOWR(DRM_COMMAND_BASE + DRM_MSM_SUBMITQUEUE_NEW, struct drm_msm_submitqueue) #define DRM_IOCTL_MSM_SUBMITQUEUE_CLOSE DRM_IOW (DRM_COMMAND_BASE + DRM_MSM_SUBMITQUEUE_CLOSE, __u32) #define DRM_IOCTL_MSM_SUBMITQUEUE_QUERY DRM_IOW (DRM_COMMAND_BASE + DRM_MSM_SUBMITQUEUE_QUERY, struct drm_msm_submitqueue_query) +#define DRM_IOCTL_MSM_VM_BIND DRM_IOWR(DRM_COMMAND_BASE + DRM_MSM_VM_BIND, struct drm_msm_vm_bind) #if defined(__cplusplus) } diff --git a/include/uapi/drm/nouveau_drm.h b/include/uapi/drm/nouveau_drm.h index dd87f8f30793..1fa82fa6af38 100644 --- a/include/uapi/drm/nouveau_drm.h +++ b/include/uapi/drm/nouveau_drm.h @@ -432,6 +432,69 @@ struct drm_nouveau_exec { __u64 push_ptr; }; +struct drm_nouveau_get_zcull_info { + /** + * @width_align_pixels: required alignment for region widths, in pixels + * (typically #TPC's * 16). + */ + __u32 width_align_pixels; + /** + * @height_align_pixels: required alignment for region heights, in + * pixels (typically 32). + */ + __u32 height_align_pixels; + /** + * @pixel_squares_by_aliquots: the pixel area covered by an aliquot + * (typically #Zcull_banks * 16 * 16). + */ + __u32 pixel_squares_by_aliquots; + /** + * @aliquot_total: the total aliquot pool available in hardware + */ + __u32 aliquot_total; + /** + * @zcull_region_byte_multiplier: the size of an aliquot in bytes, which + * is used for save/restore operations on a region + */ + __u32 zcull_region_byte_multiplier; + /** + * @zcull_region_header_size: the region header size in bytes, which is + * used for save/restore operations on a region + */ + __u32 zcull_region_header_size; + /** + * @zcull_subregion_header_size: the subregion header size in bytes, + * which is used for save/restore operations on a region + */ + __u32 zcull_subregion_header_size; + /** + * @subregion_count: the total number of subregions the hardware + * supports + */ + __u32 subregion_count; + /** + * @subregion_width_align_pixels: required alignment for subregion + * widths, in pixels (typically #TPC's * 16). + */ + __u32 subregion_width_align_pixels; + /** + * @subregion_height_align_pixels: required alignment for subregion + * heights, in pixels + */ + __u32 subregion_height_align_pixels; + + /** + * @ctxsw_size: the size, in bytes, of a zcull context switching region. + * Will be zero if the kernel does not support zcull context switching. + */ + __u32 ctxsw_size; + /** + * @ctxsw_align: the alignment, in bytes, of a zcull context switching + * region + */ + __u32 ctxsw_align; +}; + #define DRM_NOUVEAU_GETPARAM 0x00 #define DRM_NOUVEAU_SETPARAM 0x01 /* deprecated */ #define DRM_NOUVEAU_CHANNEL_ALLOC 0x02 @@ -445,6 +508,7 @@ struct drm_nouveau_exec { #define DRM_NOUVEAU_VM_INIT 0x10 #define DRM_NOUVEAU_VM_BIND 0x11 #define DRM_NOUVEAU_EXEC 0x12 +#define DRM_NOUVEAU_GET_ZCULL_INFO 0x13 #define DRM_NOUVEAU_GEM_NEW 0x40 #define DRM_NOUVEAU_GEM_PUSHBUF 0x41 #define DRM_NOUVEAU_GEM_CPU_PREP 0x42 @@ -513,6 +577,8 @@ struct drm_nouveau_svm_bind { #define DRM_IOCTL_NOUVEAU_VM_INIT DRM_IOWR(DRM_COMMAND_BASE + DRM_NOUVEAU_VM_INIT, struct drm_nouveau_vm_init) #define DRM_IOCTL_NOUVEAU_VM_BIND DRM_IOWR(DRM_COMMAND_BASE + DRM_NOUVEAU_VM_BIND, struct drm_nouveau_vm_bind) #define DRM_IOCTL_NOUVEAU_EXEC DRM_IOWR(DRM_COMMAND_BASE + DRM_NOUVEAU_EXEC, struct drm_nouveau_exec) + +#define DRM_IOCTL_NOUVEAU_GET_ZCULL_INFO DRM_IOR (DRM_COMMAND_BASE + DRM_NOUVEAU_GET_ZCULL_INFO, struct drm_nouveau_get_zcull_info) #if defined(__cplusplus) } #endif diff --git a/include/uapi/drm/nova_drm.h b/include/uapi/drm/nova_drm.h new file mode 100644 index 000000000000..3ca90ed9d2bb --- /dev/null +++ b/include/uapi/drm/nova_drm.h @@ -0,0 +1,101 @@ +/* SPDX-License-Identifier: MIT */ + +#ifndef __NOVA_DRM_H__ +#define __NOVA_DRM_H__ + +#include "drm.h" + +/* DISCLAIMER: Do not use, this is not a stable uAPI. + * + * This uAPI serves only testing purposes as long as this driver is still in + * development. It is required to implement and test infrastructure which is + * upstreamed in the context of this driver. See also [1]. + * + * [1] https://lore.kernel.org/dri-devel/Zfsj0_tb-0-tNrJy@cassiopeiae/T/#u + */ + +#if defined(__cplusplus) +extern "C" { +#endif + +/* + * NOVA_GETPARAM_VRAM_BAR_SIZE + * + * Query the VRAM BAR size in bytes. + */ +#define NOVA_GETPARAM_VRAM_BAR_SIZE 0x1 + +/** + * struct drm_nova_getparam - query GPU and driver metadata + */ +struct drm_nova_getparam { + /** + * @param: The identifier of the parameter to query. + */ + __u64 param; + + /** + * @value: The value for the specified parameter. + */ + __u64 value; +}; + +/** + * struct drm_nova_gem_create - create a new DRM GEM object + */ +struct drm_nova_gem_create { + /** + * @handle: The handle of the new DRM GEM object. + */ + __u32 handle; + + /** + * @pad: 32 bit padding, should be 0. + */ + __u32 pad; + + /** + * @size: The size of the new DRM GEM object. + */ + __u64 size; +}; + +/** + * struct drm_nova_gem_info - query DRM GEM object metadata + */ +struct drm_nova_gem_info { + /** + * @handle: The handle of the DRM GEM object to query. + */ + __u32 handle; + + /** + * @pad: 32 bit padding, should be 0. + */ + __u32 pad; + + /** + * @size: The size of the DRM GEM obejct. + */ + __u64 size; +}; + +#define DRM_NOVA_GETPARAM 0x00 +#define DRM_NOVA_GEM_CREATE 0x01 +#define DRM_NOVA_GEM_INFO 0x02 + +/* Note: this is an enum so that it can be resolved by Rust bindgen. */ +enum { + DRM_IOCTL_NOVA_GETPARAM = DRM_IOWR(DRM_COMMAND_BASE + DRM_NOVA_GETPARAM, + struct drm_nova_getparam), + DRM_IOCTL_NOVA_GEM_CREATE = DRM_IOWR(DRM_COMMAND_BASE + DRM_NOVA_GEM_CREATE, + struct drm_nova_gem_create), + DRM_IOCTL_NOVA_GEM_INFO = DRM_IOWR(DRM_COMMAND_BASE + DRM_NOVA_GEM_INFO, + struct drm_nova_gem_info), +}; + +#if defined(__cplusplus) +} +#endif + +#endif /* __NOVA_DRM_H__ */ diff --git a/include/uapi/drm/panfrost_drm.h b/include/uapi/drm/panfrost_drm.h index 568724be6628..50d5337f35ef 100644 --- a/include/uapi/drm/panfrost_drm.h +++ b/include/uapi/drm/panfrost_drm.h @@ -21,6 +21,11 @@ extern "C" { #define DRM_PANFROST_PERFCNT_ENABLE 0x06 #define DRM_PANFROST_PERFCNT_DUMP 0x07 #define DRM_PANFROST_MADVISE 0x08 +#define DRM_PANFROST_SET_LABEL_BO 0x09 +#define DRM_PANFROST_JM_CTX_CREATE 0x0a +#define DRM_PANFROST_JM_CTX_DESTROY 0x0b +#define DRM_PANFROST_SYNC_BO 0x0c +#define DRM_PANFROST_QUERY_BO_INFO 0x0d #define DRM_IOCTL_PANFROST_SUBMIT DRM_IOW(DRM_COMMAND_BASE + DRM_PANFROST_SUBMIT, struct drm_panfrost_submit) #define DRM_IOCTL_PANFROST_WAIT_BO DRM_IOW(DRM_COMMAND_BASE + DRM_PANFROST_WAIT_BO, struct drm_panfrost_wait_bo) @@ -29,6 +34,11 @@ extern "C" { #define DRM_IOCTL_PANFROST_GET_PARAM DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_GET_PARAM, struct drm_panfrost_get_param) #define DRM_IOCTL_PANFROST_GET_BO_OFFSET DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_GET_BO_OFFSET, struct drm_panfrost_get_bo_offset) #define DRM_IOCTL_PANFROST_MADVISE DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_MADVISE, struct drm_panfrost_madvise) +#define DRM_IOCTL_PANFROST_SET_LABEL_BO DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_SET_LABEL_BO, struct drm_panfrost_set_label_bo) +#define DRM_IOCTL_PANFROST_JM_CTX_CREATE DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_JM_CTX_CREATE, struct drm_panfrost_jm_ctx_create) +#define DRM_IOCTL_PANFROST_JM_CTX_DESTROY DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_JM_CTX_DESTROY, struct drm_panfrost_jm_ctx_destroy) +#define DRM_IOCTL_PANFROST_SYNC_BO DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_SYNC_BO, struct drm_panfrost_sync_bo) +#define DRM_IOCTL_PANFROST_QUERY_BO_INFO DRM_IOWR(DRM_COMMAND_BASE + DRM_PANFROST_QUERY_BO_INFO, struct drm_panfrost_query_bo_info) /* * Unstable ioctl(s): only exposed when the unsafe unstable_ioctls module @@ -48,27 +58,47 @@ extern "C" { * This asks the kernel to have the GPU execute a render command list. */ struct drm_panfrost_submit { - - /** Address to GPU mapping of job descriptor */ + /** + * @jc: Address to GPU mapping of job descriptor + */ __u64 jc; - - /** An optional array of sync objects to wait on before starting this job. */ + /** + * @in_syncs: An optional array of sync objects to wait on + * before starting this job. + */ __u64 in_syncs; - - /** Number of sync objects to wait on before starting this job. */ + /** + * @in_sync_count: Number of sync objects to wait on before + * starting this job. + */ __u32 in_sync_count; - - /** An optional sync object to place the completion fence in. */ + /** + * @out_sync: An optional sync object to place the completion fence in. + */ __u32 out_sync; - - /** Pointer to a u32 array of the BOs that are referenced by the job. */ + /** + * @bo_handles: Pointer to a u32 array of the BOs that are + * referenced by the job. + */ __u64 bo_handles; - - /** Number of BO handles passed in (size is that times 4). */ + /** + * @bo_handle_count: Number of BO handles passed in (size is + * that times 4). + */ __u32 bo_handle_count; - - /** A combination of PANFROST_JD_REQ_* */ + /** + * @requirements: A combination of PANFROST_JD_REQ_* + */ __u32 requirements; + /** + * @jm_ctx_handle: JM context handle. Zero if you want to use the + * default context. + */ + __u32 jm_ctx_handle; + /** + * @pad: Padding field. Must be zero. + */ + __u32 pad; }; /** @@ -80,14 +110,26 @@ struct drm_panfrost_submit { * completed. */ struct drm_panfrost_wait_bo { + /** + * @handle: Handle for the object to wait for. + */ __u32 handle; + /** + * @pad: Padding, must be zero-filled. + */ __u32 pad; - __s64 timeout_ns; /* absolute */ + /** + * @timeout_ns: absolute number of nanoseconds to wait. + */ + __s64 timeout_ns; }; -/* Valid flags to pass to drm_panfrost_create_bo */ +/* Valid flags to pass to drm_panfrost_create_bo. + * PANFROST_BO_WB_MMAP can't be set if PANFROST_BO_HEAP is. + */ #define PANFROST_BO_NOEXEC 1 #define PANFROST_BO_HEAP 2 +#define PANFROST_BO_WB_MMAP 4 /** * struct drm_panfrost_create_bo - ioctl argument for creating Panfrost BOs. @@ -95,16 +137,26 @@ struct drm_panfrost_wait_bo { * The flags argument is a bit mask of PANFROST_BO_* flags. */ struct drm_panfrost_create_bo { + /** + * @size: size of shmem/BO area to create (bytes) + */ __u32 size; + /** + * @flags: see PANFROST_BO_* flags + */ __u32 flags; - /** Returned GEM handle for the BO. */ + /** + * @handle: Returned GEM handle for the BO. + */ __u32 handle; - /* Pad, must be zero-filled. */ + /** + * @pad: Padding, must be zero-filled. + */ __u32 pad; /** - * Returned offset for the BO in the GPU address space. This offset - * is private to the DRM fd and is valid for the lifetime of the GEM - * handle. + * @offset: Returned offset for the BO in the GPU address space. + * This offset is private to the DRM fd and is valid for the + * lifetime of the GEM handle. * * This offset value will always be nonzero, since various HW * units treat 0 specially. @@ -124,10 +176,17 @@ struct drm_panfrost_create_bo { * used in a future extension. */ struct drm_panfrost_mmap_bo { - /** Handle for the object being mapped. */ + /** + * @handle: Handle for the object being mapped. + */ __u32 handle; + /** + * @flags: currently not used (should be zero) + */ __u32 flags; - /** offset into the drm node to use for subsequent mmap call. */ + /** + * @offset: offset into the drm node to use for subsequent mmap call. + */ __u64 offset; }; @@ -175,6 +234,14 @@ enum drm_panfrost_param { DRM_PANFROST_PARAM_AFBC_FEATURES, DRM_PANFROST_PARAM_SYSTEM_TIMESTAMP, DRM_PANFROST_PARAM_SYSTEM_TIMESTAMP_FREQUENCY, + DRM_PANFROST_PARAM_ALLOWED_JM_CTX_PRIORITIES, + DRM_PANFROST_PARAM_SELECTED_COHERENCY, +}; + +enum drm_panfrost_gpu_coherency { + DRM_PANFROST_GPU_COHERENCY_ACE_LITE = 0, + DRM_PANFROST_GPU_COHERENCY_ACE = 1, + DRM_PANFROST_GPU_COHERENCY_NONE = 31, }; struct drm_panfrost_get_param { @@ -183,7 +250,7 @@ struct drm_panfrost_get_param { __u64 value; }; -/** +/* * Returns the offset for the BO in the GPU address space for this DRM fd. * This is the same value returned by drm_panfrost_create_bo, if that was called * from this DRM fd. @@ -227,6 +294,87 @@ struct drm_panfrost_madvise { __u32 retained; /* out, whether backing store still exists */ }; +/** + * struct drm_panfrost_set_label_bo - ioctl argument for labelling Panfrost BOs. + */ +struct drm_panfrost_set_label_bo { + /** + * @handle: Handle of the buffer object to label. + */ + __u32 handle; + /** + * @pad: Must be zero. + */ + __u32 pad; + /** + * @label: User pointer to a NUL-terminated string + * + * Length cannot be greater than 4096. + * NULL is permitted and means clear the label. + */ + __u64 label; +}; + +/* Valid flags to pass to drm_panfrost_bo_sync_op */ +#define PANFROST_BO_SYNC_CPU_CACHE_FLUSH 0 +#define PANFROST_BO_SYNC_CPU_CACHE_FLUSH_AND_INVALIDATE 1 + +/** + * struct drm_panthor_bo_flush_map_op - BO map sync op + */ +struct drm_panfrost_bo_sync_op { + /** @handle: Handle of the buffer object to sync. */ + __u32 handle; + + /** @type: Type of sync operation. */ + __u32 type; + + /** + * @offset: Offset into the BO at which the sync range starts. + * + * This will be rounded down to the nearest cache line as needed. + */ + __u32 offset; + + /** + * @size: Size of the range to sync + * + * @size + @offset will be rounded up to the nearest cache line as + * needed. + */ + __u32 size; +}; + +/** + * struct drm_panfrost_sync_bo - ioctl argument for syncing BO maps + */ +struct drm_panfrost_sync_bo { + /** Array of struct drm_panfrost_bo_sync_op */ + __u64 ops; + + /** Number of BO sync ops */ + __u32 op_count; + + __u32 pad; +}; + +/** BO comes from a different subsystem. */ +#define DRM_PANFROST_BO_IS_IMPORTED (1 << 0) + +struct drm_panfrost_query_bo_info { + /** Handle of the object being queried. */ + __u32 handle; + + /** Extra flags that are not coming from the BO_CREATE ioctl(). */ + __u32 extra_flags; + + /** Flags passed at creation time. */ + __u32 create_flags; + + /** Will be zero on return. */ + __u32 pad; +}; + /* Definitions for coredump decoding in user space */ #define PANFROSTDUMP_MAJOR 1 #define PANFROSTDUMP_MINOR 0 @@ -278,6 +426,49 @@ struct panfrost_dump_registers { __u32 value; }; +enum drm_panfrost_jm_ctx_priority { + /** + * @PANFROST_JM_CTX_PRIORITY_LOW: Low priority context. + */ + PANFROST_JM_CTX_PRIORITY_LOW = 0, + + /** + * @PANFROST_JM_CTX_PRIORITY_MEDIUM: Medium priority context. + */ + PANFROST_JM_CTX_PRIORITY_MEDIUM, + + /** + * @PANFROST_JM_CTX_PRIORITY_HIGH: High priority context. + * + * Requires CAP_SYS_NICE or DRM_MASTER. + */ + PANFROST_JM_CTX_PRIORITY_HIGH, +}; + +struct drm_panfrost_jm_ctx_create { + /** + * @handle: Handle of the created JM context + */ + __u32 handle; + /** + * @priority: Context priority (see enum drm_panfrost_jm_ctx_priority). + */ + __u32 priority; +}; + +struct drm_panfrost_jm_ctx_destroy { + /** + * @handle: Handle of the JM context to destroy. + * + * Must be a valid context handle returned by DRM_IOCTL_PANTHOR_JM_CTX_CREATE. + */ + __u32 handle; + /** + * @pad: Padding field, must be zero. + */ + __u32 pad; +}; + #if defined(__cplusplus) } #endif diff --git a/include/uapi/drm/panthor_drm.h b/include/uapi/drm/panthor_drm.h index 97e2c4510e69..0e455d91e77d 100644 --- a/include/uapi/drm/panthor_drm.h +++ b/include/uapi/drm/panthor_drm.h @@ -127,6 +127,33 @@ enum drm_panthor_ioctl_id { /** @DRM_PANTHOR_TILER_HEAP_DESTROY: Destroy a tiler heap. */ DRM_PANTHOR_TILER_HEAP_DESTROY, + + /** @DRM_PANTHOR_BO_SET_LABEL: Label a BO. */ + DRM_PANTHOR_BO_SET_LABEL, + + /** + * @DRM_PANTHOR_SET_USER_MMIO_OFFSET: Set the offset to use as the user MMIO offset. + * + * The default behavior is to pick the MMIO offset based on the size of the pgoff_t + * type seen by the process that manipulates the FD, such that a 32-bit process can + * always map the user MMIO ranges. But this approach doesn't work well for emulators + * like FEX, where the emulator is an 64-bit binary which might be executing 32-bit + * code. In that case, the kernel thinks it's the 64-bit process and assumes + * DRM_PANTHOR_USER_MMIO_OFFSET_64BIT is in use, but the UMD library expects + * DRM_PANTHOR_USER_MMIO_OFFSET_32BIT, because it can't mmap() anything above the + * pgoff_t size. + */ + DRM_PANTHOR_SET_USER_MMIO_OFFSET, + + /** @DRM_PANTHOR_BO_SYNC: Sync BO data to/from the device */ + DRM_PANTHOR_BO_SYNC, + + /** + * @DRM_PANTHOR_BO_QUERY_INFO: Query information about a BO. + * + * This is useful for imported BOs. + */ + DRM_PANTHOR_BO_QUERY_INFO, }; /** @@ -229,6 +256,26 @@ enum drm_panthor_dev_query_type { }; /** + * enum drm_panthor_gpu_coherency: Type of GPU coherency + */ +enum drm_panthor_gpu_coherency { + /** + * @DRM_PANTHOR_GPU_COHERENCY_ACE_LITE: ACE Lite coherency. + */ + DRM_PANTHOR_GPU_COHERENCY_ACE_LITE = 0, + + /** + * @DRM_PANTHOR_GPU_COHERENCY_ACE: ACE coherency. + */ + DRM_PANTHOR_GPU_COHERENCY_ACE = 1, + + /** + * @DRM_PANTHOR_GPU_COHERENCY_NONE: No coherency. + */ + DRM_PANTHOR_GPU_COHERENCY_NONE = 31, +}; + +/** * struct drm_panthor_gpu_info - GPU information * * Structure grouping all queryable information relating to the GPU. @@ -284,7 +331,16 @@ struct drm_panthor_gpu_info { */ __u32 thread_max_barrier_size; - /** @coherency_features: Coherency features. */ + /** + * @coherency_features: Coherency features. + * + * Combination of drm_panthor_gpu_coherency flags. + * + * Note that this is just what the coherency protocols supported by the + * GPU, but the actual coherency in place depends on the SoC + * integration and is reflected by + * drm_panthor_gpu_info::selected_coherency. + */ __u32 coherency_features; /** @texture_features: Texture features. */ @@ -293,6 +349,13 @@ struct drm_panthor_gpu_info { /** @as_present: Bitmask encoding the number of address-space exposed by the MMU. */ __u32 as_present; + /** + * @selected_coherency: Coherency selected for this device. + * + * One of drm_panthor_gpu_coherency. + */ + __u32 selected_coherency; + /** @shader_present: Bitmask encoding the shader cores exposed by the GPU. */ __u64 shader_present; @@ -307,6 +370,9 @@ struct drm_panthor_gpu_info { /** @pad: MBZ. */ __u32 pad; + + /** @gpu_features: Bitmask describing supported GPU-wide features */ + __u64 gpu_features; }; /** @@ -344,6 +410,38 @@ struct drm_panthor_csif_info { }; /** + * enum drm_panthor_timestamp_info_flags - drm_panthor_timestamp_info.flags + */ +enum drm_panthor_timestamp_info_flags { + /** @DRM_PANTHOR_TIMESTAMP_GPU: Query GPU time. */ + DRM_PANTHOR_TIMESTAMP_GPU = 1 << 0, + + /** @DRM_PANTHOR_TIMESTAMP_CPU_NONE: Don't query CPU time. */ + DRM_PANTHOR_TIMESTAMP_CPU_NONE = 0 << 1, + + /** @DRM_PANTHOR_TIMESTAMP_CPU_MONOTONIC: Query CPU time using CLOCK_MONOTONIC. */ + DRM_PANTHOR_TIMESTAMP_CPU_MONOTONIC = 1 << 1, + + /** @DRM_PANTHOR_TIMESTAMP_CPU_MONOTONIC_RAW: Query CPU time using CLOCK_MONOTONIC_RAW. */ + DRM_PANTHOR_TIMESTAMP_CPU_MONOTONIC_RAW = 2 << 1, + + /** @DRM_PANTHOR_TIMESTAMP_CPU_TYPE_MASK: Space reserved for CPU clock type. */ + DRM_PANTHOR_TIMESTAMP_CPU_TYPE_MASK = 7 << 1, + + /** @DRM_PANTHOR_TIMESTAMP_GPU_OFFSET: Query GPU offset. */ + DRM_PANTHOR_TIMESTAMP_GPU_OFFSET = 1 << 4, + + /** @DRM_PANTHOR_TIMESTAMP_GPU_CYCLE_COUNT: Query GPU cycle count. */ + DRM_PANTHOR_TIMESTAMP_GPU_CYCLE_COUNT = 1 << 5, + + /** @DRM_PANTHOR_TIMESTAMP_FREQ: Query timestamp frequency. */ + DRM_PANTHOR_TIMESTAMP_FREQ = 1 << 6, + + /** @DRM_PANTHOR_TIMESTAMP_DURATION: Return duration of time query. */ + DRM_PANTHOR_TIMESTAMP_DURATION = 1 << 7, +}; + +/** * struct drm_panthor_timestamp_info - Timestamp information * * Structure grouping all queryable information relating to the GPU timestamp. @@ -355,11 +453,38 @@ struct drm_panthor_timestamp_info { */ __u64 timestamp_frequency; - /** @current_timestamp: The current timestamp. */ + /** @current_timestamp: The current GPU timestamp. */ __u64 current_timestamp; - /** @timestamp_offset: The offset of the timestamp timer. */ + /** @timestamp_offset: The offset of the GPU timestamp timer. */ __u64 timestamp_offset; + + /** + * @flags: Bitmask of drm_panthor_timestamp_info_flags. + * + * If set to 0, then it is interpreted as: + * DRM_PANTHOR_TIMESTAMP_GPU | + * DRM_PANTHOR_TIMESTAMP_GPU_OFFSET | + * DRM_PANTHOR_TIMESTAMP_FREQ + * + * Note: these flags are exclusive to each other (only one can be used): + * - DRM_PANTHOR_TIMESTAMP_CPU_NONE + * - DRM_PANTHOR_TIMESTAMP_CPU_MONOTONIC + * - DRM_PANTHOR_TIMESTAMP_CPU_MONOTONIC_RAW + */ + __u32 flags; + + /** @duration_nsec: Duration of time query. */ + __u32 duration_nsec; + + /** @cycle_count: Value of GPU_CYCLE_COUNT. */ + __u64 cycle_count; + + /** @cpu_timestamp_sec: Seconds part of CPU timestamp. */ + __u64 cpu_timestamp_sec; + + /** @cpu_timestamp_nsec: Nanseconds part of CPU timestamp. */ + __u64 cpu_timestamp_nsec; }; /** @@ -615,6 +740,15 @@ struct drm_panthor_vm_get_state { enum drm_panthor_bo_flags { /** @DRM_PANTHOR_BO_NO_MMAP: The buffer object will never be CPU-mapped in userspace. */ DRM_PANTHOR_BO_NO_MMAP = (1 << 0), + + /** + * @DRM_PANTHOR_BO_WB_MMAP: Force "Write-Back Cacheable" CPU mapping. + * + * CPU map the buffer object in userspace by forcing the "Write-Back + * Cacheable" cacheability attribute. The mapping otherwise uses the + * "Non-Cacheable" attribute if the GPU is not IO coherent. + */ + DRM_PANTHOR_BO_WB_MMAP = (1 << 1), }; /** @@ -978,6 +1112,141 @@ struct drm_panthor_tiler_heap_destroy { }; /** + * struct drm_panthor_bo_set_label - Arguments passed to DRM_IOCTL_PANTHOR_BO_SET_LABEL + */ +struct drm_panthor_bo_set_label { + /** @handle: Handle of the buffer object to label. */ + __u32 handle; + + /** @pad: MBZ. */ + __u32 pad; + + /** + * @label: User pointer to a NUL-terminated string + * + * Length cannot be greater than 4096 + */ + __u64 label; +}; + +/** + * struct drm_panthor_set_user_mmio_offset - Arguments passed to + * DRM_IOCTL_PANTHOR_SET_USER_MMIO_OFFSET + * + * This ioctl is only really useful if you want to support userspace + * CPU emulation environments where the size of an unsigned long differs + * between the host and the guest architectures. + */ +struct drm_panthor_set_user_mmio_offset { + /** + * @offset: User MMIO offset to use. + * + * Must be either DRM_PANTHOR_USER_MMIO_OFFSET_32BIT or + * DRM_PANTHOR_USER_MMIO_OFFSET_64BIT. + * + * Use DRM_PANTHOR_USER_MMIO_OFFSET (which selects OFFSET_32BIT or + * OFFSET_64BIT based on the size of an unsigned long) unless you + * have a very good reason to overrule this decision. + */ + __u64 offset; +}; + +/** + * enum drm_panthor_bo_sync_op_type - BO sync type + */ +enum drm_panthor_bo_sync_op_type { + /** @DRM_PANTHOR_BO_SYNC_CPU_CACHE_FLUSH: Flush CPU caches. */ + DRM_PANTHOR_BO_SYNC_CPU_CACHE_FLUSH = 0, + + /** @DRM_PANTHOR_BO_SYNC_CPU_CACHE_FLUSH_AND_INVALIDATE: Flush and invalidate CPU caches. */ + DRM_PANTHOR_BO_SYNC_CPU_CACHE_FLUSH_AND_INVALIDATE = 1, +}; + +/** + * struct drm_panthor_bo_sync_op - BO map sync op + */ +struct drm_panthor_bo_sync_op { + /** @handle: Handle of the buffer object to sync. */ + __u32 handle; + + /** @type: Type of operation. */ + __u32 type; + + /** + * @offset: Offset into the BO at which the sync range starts. + * + * This will be rounded down to the nearest cache line as needed. + */ + __u64 offset; + + /** + * @size: Size of the range to sync + * + * @size + @offset will be rounded up to the nearest cache line as + * needed. + */ + __u64 size; +}; + +/** + * struct drm_panthor_bo_sync - BO map sync request + */ +struct drm_panthor_bo_sync { + /** + * @ops: Array of struct drm_panthor_bo_sync_op sync operations. + */ + struct drm_panthor_obj_array ops; +}; + +/** + * enum drm_panthor_bo_extra_flags - Set of flags returned on a BO_QUERY_INFO request + * + * Those are flags reflecting BO properties that are not directly coming from the flags + * passed are creation time, or information on BOs that were imported from other drivers. + */ +enum drm_panthor_bo_extra_flags { + /** + * @DRM_PANTHOR_BO_IS_IMPORTED: BO has been imported from an external driver. + * + * Note that imported dma-buf handles are not flagged as imported if they + * where exported by panthor. Only buffers that are coming from other drivers + * (dma heaps, other GPUs, display controllers, V4L, ...). + * + * It's also important to note that all imported BOs are mapped cached and can't + * be considered IO-coherent even if the GPU is. This means they require explicit + * syncs that must go through the DRM_PANTHOR_BO_SYNC ioctl (userland cache + * maintenance is not allowed in that case, because extra operations might be + * needed to make changes visible to the CPU/device, like buffer migration when the + * exporter is a GPU with its own VRAM). + */ + DRM_PANTHOR_BO_IS_IMPORTED = (1 << 0), +}; + +/** + * struct drm_panthor_bo_query_info - Query BO info + */ +struct drm_panthor_bo_query_info { + /** @handle: Handle of the buffer object to query flags on. */ + __u32 handle; + + /** + * @extra_flags: Combination of enum drm_panthor_bo_extra_flags flags. + */ + __u32 extra_flags; + + /** + * @create_flags: Flags passed at creation time. + * + * Combination of enum drm_panthor_bo_flags flags. + * Will be zero if the buffer comes from a different driver. + */ + __u32 create_flags; + + /** @pad: Will be zero on return. */ + __u32 pad; +}; + +/** * DRM_IOCTL_PANTHOR() - Build a Panthor IOCTL number * @__access: Access type. Must be R, W or RW. * @__id: One of the DRM_PANTHOR_xxx id. @@ -1019,6 +1288,14 @@ enum { DRM_IOCTL_PANTHOR(WR, TILER_HEAP_CREATE, tiler_heap_create), DRM_IOCTL_PANTHOR_TILER_HEAP_DESTROY = DRM_IOCTL_PANTHOR(WR, TILER_HEAP_DESTROY, tiler_heap_destroy), + DRM_IOCTL_PANTHOR_BO_SET_LABEL = + DRM_IOCTL_PANTHOR(WR, BO_SET_LABEL, bo_set_label), + DRM_IOCTL_PANTHOR_SET_USER_MMIO_OFFSET = + DRM_IOCTL_PANTHOR(WR, SET_USER_MMIO_OFFSET, set_user_mmio_offset), + DRM_IOCTL_PANTHOR_BO_SYNC = + DRM_IOCTL_PANTHOR(WR, BO_SYNC, bo_sync), + DRM_IOCTL_PANTHOR_BO_QUERY_INFO = + DRM_IOCTL_PANTHOR(WR, BO_QUERY_INFO, bo_query_info), }; #if defined(__cplusplus) diff --git a/include/uapi/drm/rocket_accel.h b/include/uapi/drm/rocket_accel.h new file mode 100644 index 000000000000..d0685e372b79 --- /dev/null +++ b/include/uapi/drm/rocket_accel.h @@ -0,0 +1,192 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright © 2024 Tomeu Vizoso + */ +#ifndef __DRM_UAPI_ROCKET_ACCEL_H__ +#define __DRM_UAPI_ROCKET_ACCEL_H__ + +#include "drm.h" + +#if defined(__cplusplus) +extern "C" { +#endif + +#define DRM_ROCKET_CREATE_BO 0x00 +#define DRM_ROCKET_SUBMIT 0x01 +#define DRM_ROCKET_PREP_BO 0x02 +#define DRM_ROCKET_FINI_BO 0x03 + +#define DRM_IOCTL_ROCKET_CREATE_BO DRM_IOWR(DRM_COMMAND_BASE + DRM_ROCKET_CREATE_BO, struct drm_rocket_create_bo) +#define DRM_IOCTL_ROCKET_SUBMIT DRM_IOW(DRM_COMMAND_BASE + DRM_ROCKET_SUBMIT, struct drm_rocket_submit) +#define DRM_IOCTL_ROCKET_PREP_BO DRM_IOW(DRM_COMMAND_BASE + DRM_ROCKET_PREP_BO, struct drm_rocket_prep_bo) +#define DRM_IOCTL_ROCKET_FINI_BO DRM_IOW(DRM_COMMAND_BASE + DRM_ROCKET_FINI_BO, struct drm_rocket_fini_bo) + +/** + * struct drm_rocket_create_bo - ioctl argument for creating Rocket BOs. + * + */ +struct drm_rocket_create_bo { + /** + * @size: Input: Size of the requested BO. + */ + __u32 size; + + /** + * @handle: Output: GEM handle for the BO. + */ + __u32 handle; + + /** + * @dma_address: Output: DMA address for the BO in the NPU address + * space. This address is private to the DRM fd and is valid for + * the lifetime of the GEM handle. + */ + __u64 dma_address; + + /** + * @offset: Output: Offset into the drm node to use for subsequent + * mmap call. + */ + __u64 offset; +}; + +/** + * struct drm_rocket_prep_bo - ioctl argument for starting CPU ownership of the BO. + * + * Takes care of waiting for any NPU jobs that might still use the NPU and performs cache + * synchronization. + */ +struct drm_rocket_prep_bo { + /** + * @handle: Input: GEM handle of the buffer object. + */ + __u32 handle; + + /** + * @reserved: Reserved, must be zero. + */ + __u32 reserved; + + /** + * @timeout_ns: Input: Amount of time to wait for NPU jobs. + */ + __s64 timeout_ns; +}; + +/** + * struct drm_rocket_fini_bo - ioctl argument for finishing CPU ownership of the BO. + * + * Synchronize caches for NPU access. + */ +struct drm_rocket_fini_bo { + /** + * @handle: Input: GEM handle of the buffer object. + */ + __u32 handle; + + /** + * @reserved: Reserved, must be zero. + */ + __u32 reserved; +}; + +/** + * struct drm_rocket_task - A task to be run on the NPU + * + * A task is the smallest unit of work that can be run on the NPU. + */ +struct drm_rocket_task { + /** + * @regcmd: Input: DMA address to NPU mapping of register command buffer + */ + __u32 regcmd; + + /** + * @regcmd_count: Input: Number of commands in the register command + * buffer + */ + __u32 regcmd_count; +}; + +/** + * struct drm_rocket_job - A job to be run on the NPU + * + * The kernel will schedule the execution of this job taking into account its + * dependencies with other jobs. All tasks in the same job will be executed + * sequentially on the same core, to benefit from memory residency in SRAM. + */ +struct drm_rocket_job { + /** + * @tasks: Input: Pointer to an array of struct drm_rocket_task. + */ + __u64 tasks; + + /** + * @in_bo_handles: Input: Pointer to a u32 array of the BOs that + * are read by the job. + */ + __u64 in_bo_handles; + + /** + * @out_bo_handles: Input: Pointer to a u32 array of the BOs that + * are written to by the job. + */ + __u64 out_bo_handles; + + /** + * @task_count: Input: Number of tasks passed in. + */ + __u32 task_count; + + /** + * @task_struct_size: Input: Size in bytes of the structs in the + * @tasks field. + */ + __u32 task_struct_size; + + /** + * @in_bo_handle_count: Input: Number of input BO handles passed in + * (size is that times 4). + */ + __u32 in_bo_handle_count; + + /** + * @out_bo_handle_count: Input: Number of output BO handles passed in + * (size is that times 4). + */ + __u32 out_bo_handle_count; +}; + +/** + * struct drm_rocket_submit - ioctl argument for submitting commands to the NPU. + * + * The kernel will schedule the execution of these jobs in dependency order. + */ +struct drm_rocket_submit { + /** + * @jobs: Input: Pointer to an array of struct drm_rocket_job. + */ + __u64 jobs; + + /** + * @job_count: Input: Number of jobs passed in. + */ + __u32 job_count; + + /** + * @job_struct_size: Input: Size in bytes of the structs in the + * @jobs field. + */ + __u32 job_struct_size; + + /** + * @reserved: Reserved, must be zero. + */ + __u64 reserved; +}; + +#if defined(__cplusplus) +} +#endif + +#endif /* __DRM_UAPI_ROCKET_ACCEL_H__ */ diff --git a/include/uapi/drm/v3d_drm.h b/include/uapi/drm/v3d_drm.h index dbbc404d2b3d..d9b01f4c3a04 100644 --- a/include/uapi/drm/v3d_drm.h +++ b/include/uapi/drm/v3d_drm.h @@ -294,6 +294,8 @@ enum drm_v3d_param { DRM_V3D_PARAM_SUPPORTS_CPU_QUEUE, DRM_V3D_PARAM_MAX_PERF_COUNTERS, DRM_V3D_PARAM_SUPPORTS_SUPER_PAGES, + DRM_V3D_PARAM_GLOBAL_RESET_COUNTER, + DRM_V3D_PARAM_CONTEXT_RESET_COUNTER, }; struct drm_v3d_get_param { diff --git a/include/uapi/drm/virtgpu_drm.h b/include/uapi/drm/virtgpu_drm.h index c2ce71987e9b..9debb320c34b 100644 --- a/include/uapi/drm/virtgpu_drm.h +++ b/include/uapi/drm/virtgpu_drm.h @@ -163,6 +163,12 @@ struct drm_virtgpu_3d_wait { __u32 flags; }; +#define VIRTGPU_DRM_CAPSET_VIRGL 1 +#define VIRTGPU_DRM_CAPSET_VIRGL2 2 +#define VIRTGPU_DRM_CAPSET_GFXSTREAM_VULKAN 3 +#define VIRTGPU_DRM_CAPSET_VENUS 4 +#define VIRTGPU_DRM_CAPSET_CROSS_DOMAIN 5 +#define VIRTGPU_DRM_CAPSET_DRM 6 struct drm_virtgpu_get_caps { __u32 cap_set_id; __u32 cap_set_ver; diff --git a/include/uapi/drm/xe_drm.h b/include/uapi/drm/xe_drm.h index 616916985e3f..ae2fda23ce7c 100644 --- a/include/uapi/drm/xe_drm.h +++ b/include/uapi/drm/xe_drm.h @@ -81,6 +81,9 @@ extern "C" { * - &DRM_IOCTL_XE_EXEC * - &DRM_IOCTL_XE_WAIT_USER_FENCE * - &DRM_IOCTL_XE_OBSERVATION + * - &DRM_IOCTL_XE_MADVISE + * - &DRM_IOCTL_XE_VM_QUERY_MEM_RANGE_ATTRS + * - &DRM_IOCTL_XE_VM_GET_PROPERTY */ /* @@ -102,6 +105,10 @@ extern "C" { #define DRM_XE_EXEC 0x09 #define DRM_XE_WAIT_USER_FENCE 0x0a #define DRM_XE_OBSERVATION 0x0b +#define DRM_XE_MADVISE 0x0c +#define DRM_XE_VM_QUERY_MEM_RANGE_ATTRS 0x0d +#define DRM_XE_EXEC_QUEUE_SET_PROPERTY 0x0e +#define DRM_XE_VM_GET_PROPERTY 0x0f /* Must be kept compact -- no holes */ @@ -117,6 +124,10 @@ extern "C" { #define DRM_IOCTL_XE_EXEC DRM_IOW(DRM_COMMAND_BASE + DRM_XE_EXEC, struct drm_xe_exec) #define DRM_IOCTL_XE_WAIT_USER_FENCE DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_WAIT_USER_FENCE, struct drm_xe_wait_user_fence) #define DRM_IOCTL_XE_OBSERVATION DRM_IOW(DRM_COMMAND_BASE + DRM_XE_OBSERVATION, struct drm_xe_observation_param) +#define DRM_IOCTL_XE_MADVISE DRM_IOW(DRM_COMMAND_BASE + DRM_XE_MADVISE, struct drm_xe_madvise) +#define DRM_IOCTL_XE_VM_QUERY_MEM_RANGE_ATTRS DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_VM_QUERY_MEM_RANGE_ATTRS, struct drm_xe_vm_query_mem_range_attr) +#define DRM_IOCTL_XE_EXEC_QUEUE_SET_PROPERTY DRM_IOW(DRM_COMMAND_BASE + DRM_XE_EXEC_QUEUE_SET_PROPERTY, struct drm_xe_exec_queue_set_property) +#define DRM_IOCTL_XE_VM_GET_PROPERTY DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_VM_GET_PROPERTY, struct drm_xe_vm_get_property) /** * DOC: Xe IOCTL Extensions @@ -204,8 +215,12 @@ struct drm_xe_ext_set_property { /** @pad: MBZ */ __u32 pad; - /** @value: property value */ - __u64 value; + union { + /** @value: property value */ + __u64 value; + /** @ptr: pointer to user value */ + __u64 ptr; + }; /** @reserved: Reserved */ __u64 reserved[2]; @@ -323,10 +338,6 @@ struct drm_xe_mem_region { __u64 total_size; /** * @used: Estimate of the memory used in bytes for this region. - * - * Requires CAP_PERFMON or CAP_SYS_ADMIN to get reliable - * accounting. Without this the value here will always equal - * zero. */ __u64 used; /** @@ -351,9 +362,7 @@ struct drm_xe_mem_region { * @cpu_visible_used: Estimate of CPU visible memory used, in * bytes. * - * Requires CAP_PERFMON or CAP_SYS_ADMIN to get reliable - * accounting. Without this the value here will always equal - * zero. Note this is only currently tracked for + * Note this is only currently tracked for * DRM_XE_MEM_REGION_CLASS_VRAM regions (for other types the value * here will always be zero). */ @@ -397,6 +406,12 @@ struct drm_xe_query_mem_regions { * has low latency hint support * - %DRM_XE_QUERY_CONFIG_FLAG_HAS_CPU_ADDR_MIRROR - Flag is set if the * device has CPU address mirroring support + * - %DRM_XE_QUERY_CONFIG_FLAG_HAS_NO_COMPRESSION_HINT - Flag is set if the + * device supports the userspace hint %DRM_XE_GEM_CREATE_FLAG_NO_COMPRESSION. + * This is exposed only on Xe2+. + * - %DRM_XE_QUERY_CONFIG_FLAG_HAS_DISABLE_STATE_CACHE_PERF_FIX - Flag is set + * if a queue can be creaed with + * %DRM_XE_EXEC_QUEUE_SET_DISABLE_STATE_CACHE_PERF_FIX * - %DRM_XE_QUERY_CONFIG_MIN_ALIGNMENT - Minimal memory alignment * required by this device, typically SZ_4K or SZ_64K * - %DRM_XE_QUERY_CONFIG_VA_BITS - Maximum bits of a virtual address @@ -415,6 +430,9 @@ struct drm_xe_query_config { #define DRM_XE_QUERY_CONFIG_FLAG_HAS_VRAM (1 << 0) #define DRM_XE_QUERY_CONFIG_FLAG_HAS_LOW_LATENCY (1 << 1) #define DRM_XE_QUERY_CONFIG_FLAG_HAS_CPU_ADDR_MIRROR (1 << 2) + #define DRM_XE_QUERY_CONFIG_FLAG_HAS_NO_COMPRESSION_HINT (1 << 3) + #define DRM_XE_QUERY_CONFIG_FLAG_HAS_DISABLE_STATE_CACHE_PERF_FIX (1 << 4) + #define DRM_XE_QUERY_CONFIG_FLAG_HAS_PURGING_SUPPORT (1 << 5) #define DRM_XE_QUERY_CONFIG_MIN_ALIGNMENT 2 #define DRM_XE_QUERY_CONFIG_VA_BITS 3 #define DRM_XE_QUERY_CONFIG_MAX_EXEC_QUEUE_PRIORITY 4 @@ -760,8 +778,16 @@ struct drm_xe_device_query { * gem creation * * The @flags can be: - * - %DRM_XE_GEM_CREATE_FLAG_DEFER_BACKING - * - %DRM_XE_GEM_CREATE_FLAG_SCANOUT + * - %DRM_XE_GEM_CREATE_FLAG_DEFER_BACKING - Modify the GEM object + * allocation strategy by deferring physical memory allocation + * until the object is either bound to a virtual memory region via + * VM_BIND or accessed by the CPU. As a result, no backing memory is + * reserved at the time of GEM object creation. + * - %DRM_XE_GEM_CREATE_FLAG_SCANOUT - Indicates that the GEM object is + * intended for scanout via the display engine. When set, kernel ensures + * that the allocation is placed in a memory region compatible with the + * display engine requirements. This may impose restrictions on tiling, + * alignment, and memory placement to guarantee proper display functionality. * - %DRM_XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM - When using VRAM as a * possible placement, ensure that the corresponding VRAM allocation * will always use the CPU accessible part of VRAM. This is important @@ -777,6 +803,17 @@ struct drm_xe_device_query { * need to use VRAM for display surfaces, therefore the kernel requires * setting this flag for such objects, otherwise an error is thrown on * small-bar systems. + * - %DRM_XE_GEM_CREATE_FLAG_NO_COMPRESSION - Allows userspace to + * hint that compression (CCS) should be disabled for the buffer being + * created. This can avoid unnecessary memory operations and CCS state + * management. + * On pre-Xe2 platforms, this flag is currently rejected as compression + * control is not supported via PAT index. On Xe2+ platforms, compression + * is controlled via PAT entries. If this flag is set, the driver will reject + * any VM bind that requests a PAT index enabling compression for this BO. + * Note: On dGPU platforms, there is currently no change in behavior with + * this flag, but future improvements may leverage it. The current benefit is + * primarily applicable to iGPU platforms. * * @cpu_caching supports the following values: * - %DRM_XE_GEM_CPU_CACHING_WB - Allocate the pages with write-back @@ -823,6 +860,7 @@ struct drm_xe_gem_create { #define DRM_XE_GEM_CREATE_FLAG_DEFER_BACKING (1 << 0) #define DRM_XE_GEM_CREATE_FLAG_SCANOUT (1 << 1) #define DRM_XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM (1 << 2) +#define DRM_XE_GEM_CREATE_FLAG_NO_COMPRESSION (1 << 3) /** * @flags: Flags, currently a mask of memory instances of where BO can * be placed @@ -917,13 +955,17 @@ struct drm_xe_gem_mmap_offset { * struct drm_xe_vm_create - Input of &DRM_IOCTL_XE_VM_CREATE * * The @flags can be: - * - %DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE + * - %DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE - Map the whole virtual address + * space of the VM to scratch page. A vm_bind would overwrite the scratch + * page mapping. This flag is mutually exclusive with the + * %DRM_XE_VM_CREATE_FLAG_FAULT_MODE flag, with an exception of on x2 and + * xe3 platform. * - %DRM_XE_VM_CREATE_FLAG_LR_MODE - An LR, or Long Running VM accepts * exec submissions to its exec_queues that don't have an upper time * limit on the job execution time. But exec submissions to these - * don't allow any of the flags DRM_XE_SYNC_FLAG_SYNCOBJ, - * DRM_XE_SYNC_FLAG_TIMELINE_SYNCOBJ, DRM_XE_SYNC_FLAG_DMA_BUF, - * used as out-syncobjs, that is, together with DRM_XE_SYNC_FLAG_SIGNAL. + * don't allow any of the sync types DRM_XE_SYNC_TYPE_SYNCOBJ, + * DRM_XE_SYNC_TYPE_TIMELINE_SYNCOBJ, used as out-syncobjs, that is, + * together with sync flag DRM_XE_SYNC_FLAG_SIGNAL. * LR VMs can be created in recoverable page-fault mode using * DRM_XE_VM_CREATE_FLAG_FAULT_MODE, if the device supports it. * If that flag is omitted, the UMD can not rely on the slightly @@ -935,6 +977,11 @@ struct drm_xe_gem_mmap_offset { * demand when accessed, and also allows per-VM overcommit of memory. * The xe driver internally uses recoverable pagefaults to implement * this. + * - %DRM_XE_VM_CREATE_FLAG_NO_VM_OVERCOMMIT - Requires also + * DRM_XE_VM_CREATE_FLAG_FAULT_MODE. This disallows per-VM overcommit + * but only during a &DRM_IOCTL_XE_VM_BIND operation with the + * %DRM_XE_VM_BIND_FLAG_IMMEDIATE flag set. This may be useful for + * user-space naively probing the amount of available memory. */ struct drm_xe_vm_create { /** @extensions: Pointer to the first extension struct, if any */ @@ -943,6 +990,7 @@ struct drm_xe_vm_create { #define DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE (1 << 0) #define DRM_XE_VM_CREATE_FLAG_LR_MODE (1 << 1) #define DRM_XE_VM_CREATE_FLAG_FAULT_MODE (1 << 2) +#define DRM_XE_VM_CREATE_FLAG_NO_VM_OVERCOMMIT (1 << 3) /** @flags: Flags */ __u32 flags; @@ -999,6 +1047,31 @@ struct drm_xe_vm_destroy { * valid on VMs with DRM_XE_VM_CREATE_FLAG_FAULT_MODE set. The CPU address * mirror flag are only valid for DRM_XE_VM_BIND_OP_MAP operations, the BO * handle MBZ, and the BO offset MBZ. + * - %DRM_XE_VM_BIND_FLAG_MADVISE_AUTORESET - Can be used in combination with + * %DRM_XE_VM_BIND_FLAG_CPU_ADDR_MIRROR to reset madvises when the underlying + * CPU address space range is unmapped (typically with munmap(2) or brk(2)). + * The madvise values set with &DRM_IOCTL_XE_MADVISE are reset to the values + * that were present immediately after the &DRM_IOCTL_XE_VM_BIND. + * The reset GPU virtual address range is the intersection of the range bound + * using &DRM_IOCTL_XE_VM_BIND and the virtual CPU address space range + * unmapped. + * This functionality is present to mimic the behaviour of CPU address space + * madvises set using madvise(2), which are typically reset on unmap. + * Note: free(3) may or may not call munmap(2) and/or brk(2), and may thus + * not invoke autoreset. Neither will stack variables going out of scope. + * Therefore it's recommended to always explicitly reset the madvises when + * freeing the memory backing a region used in a &DRM_IOCTL_XE_MADVISE call. + * - %DRM_XE_VM_BIND_FLAG_DECOMPRESS - Request on-device decompression for a MAP. + * When set on a MAP bind operation, request the driver schedule an on-device + * in-place decompression (via the migrate/resolve path) for the GPU mapping + * created by this bind. Only valid for DRM_XE_VM_BIND_OP_MAP; usage on + * other ops is rejected. The bind's pat_index must select the device's + * "no-compression" PAT. Only meaningful for VRAM-backed BOs on devices that + * support Flat CCS and the required HW generation XE2+. + * + * The @prefetch_mem_region_instance for %DRM_XE_VM_BIND_OP_PREFETCH can also be: + * - %DRM_XE_CONSULT_MEM_ADVISE_PREF_LOC, which ensures prefetching occurs in + * the memory region advised by madvise. */ struct drm_xe_vm_bind_op { /** @extensions: Pointer to the first extension struct, if any */ @@ -1045,7 +1118,9 @@ struct drm_xe_vm_bind_op { * incoherent GT access is possible. * * Note: For userptr and externally imported dma-buf the kernel expects - * either 1WAY or 2WAY for the @pat_index. + * either 1WAY or 2WAY for the @pat_index. Starting from NVL-P, for + * userptr, svm, madvise and externally imported dma-buf the kernel expects + * either 2WAY or 1WAY and XA @pat_index. * * For DRM_XE_VM_BIND_FLAG_NULL bindings there are no KMD restrictions * on the @pat_index. For such mappings there is no actual memory being @@ -1101,9 +1176,12 @@ struct drm_xe_vm_bind_op { #define DRM_XE_VM_BIND_FLAG_DUMPABLE (1 << 3) #define DRM_XE_VM_BIND_FLAG_CHECK_PXP (1 << 4) #define DRM_XE_VM_BIND_FLAG_CPU_ADDR_MIRROR (1 << 5) +#define DRM_XE_VM_BIND_FLAG_MADVISE_AUTORESET (1 << 6) +#define DRM_XE_VM_BIND_FLAG_DECOMPRESS (1 << 7) /** @flags: Bind flags */ __u32 flags; +#define DRM_XE_CONSULT_MEM_ADVISE_PREF_LOC -1 /** * @prefetch_mem_region_instance: Memory region to prefetch VMA to. * It is a region instance, not a mask. @@ -1189,6 +1267,89 @@ struct drm_xe_vm_bind { __u64 reserved[2]; }; +/** struct xe_vm_fault - Describes faults for %DRM_XE_VM_GET_PROPERTY_FAULTS */ +struct xe_vm_fault { + /** @address: Canonical address of the fault */ + __u64 address; + /** @address_precision: Precision of faulted address */ + __u32 address_precision; + /** @access_type: Type of address access that resulted in fault */ +#define FAULT_ACCESS_TYPE_READ 0 +#define FAULT_ACCESS_TYPE_WRITE 1 +#define FAULT_ACCESS_TYPE_ATOMIC 2 + __u8 access_type; + /** @fault_type: Type of fault reported */ +#define FAULT_TYPE_NOT_PRESENT 0 +#define FAULT_TYPE_WRITE_ACCESS 1 +#define FAULT_TYPE_ATOMIC_ACCESS 2 + __u8 fault_type; + /** @fault_level: fault level of the fault */ +#define FAULT_LEVEL_PTE 0 +#define FAULT_LEVEL_PDE 1 +#define FAULT_LEVEL_PDP 2 +#define FAULT_LEVEL_PML4 3 +#define FAULT_LEVEL_PML5 4 + __u8 fault_level; + /** @pad: MBZ */ + __u8 pad; + /** @reserved: MBZ */ + __u64 reserved[4]; +}; + +/** + * struct drm_xe_vm_get_property - Input of &DRM_IOCTL_XE_VM_GET_PROPERTY + * + * The user provides a VM and a property to query among DRM_XE_VM_GET_PROPERTY_*, + * and sets the values in the vm_id and property members, respectively. This + * determines both the VM to get the property of, as well as the property to + * report. + * + * If size is set to 0, the driver fills it with the required size for the + * requested property. The user is expected here to allocate memory for the + * property structure and to provide a pointer to the allocated memory using the + * data member. For some properties, this may be zero, in which case, the + * value of the property will be saved to the value member and size will remain + * zero on return. + * + * If size is not zero, then the IOCTL will attempt to copy the requested + * property into the data member. + * + * The IOCTL will return -ENOENT if the VM could not be identified from the + * provided VM ID, or -EINVAL if the IOCTL fails for any other reason, such as + * providing an invalid size for the given property or if the property data + * could not be copied to the memory allocated to the data member. + * + * The property member can be: + * - %DRM_XE_VM_GET_PROPERTY_FAULTS + */ +struct drm_xe_vm_get_property { + /** @extensions: Pointer to the first extension struct, if any */ + __u64 extensions; + + /** @vm_id: The ID of the VM to query the properties of */ + __u32 vm_id; + +#define DRM_XE_VM_GET_PROPERTY_FAULTS 0 + /** @property: property to get */ + __u32 property; + + /** @size: Size to allocate for @data */ + __u32 size; + + /** @pad: MBZ */ + __u32 pad; + + union { + /** @data: Pointer to user-defined array of flexible size and type */ + __u64 data; + /** @value: Return value for scalar queries */ + __u64 value; + }; + + /** @reserved: MBZ */ + __u64 reserved[3]; +}; + /** * struct drm_xe_exec_queue_create - Input of &DRM_IOCTL_XE_EXEC_QUEUE_CREATE * @@ -1206,9 +1367,28 @@ struct drm_xe_vm_bind { * there is no need to explicitly set that. When a queue of type * %DRM_XE_PXP_TYPE_HWDRM is created, the PXP default HWDRM session * (%XE_PXP_HWDRM_DEFAULT_SESSION) will be started, if isn't already running. + * The user is expected to query the PXP status via the query ioctl (see + * %DRM_XE_DEVICE_QUERY_PXP_STATUS) and to wait for PXP to be ready before + * attempting to create a queue with this property. When a queue is created + * before PXP is ready, the ioctl will return -EBUSY if init is still in + * progress or -EIO if init failed. * Given that going into a power-saving state kills PXP HWDRM sessions, * runtime PM will be blocked while queues of this type are alive. * All PXP queues will be killed if a PXP invalidation event occurs. + * - %DRM_XE_EXEC_QUEUE_SET_PROPERTY_MULTI_GROUP - Create a multi-queue group + * or add secondary queues to a multi-queue group. + * If the extension's 'value' field has %DRM_XE_MULTI_GROUP_CREATE flag set, + * then a new multi-queue group is created with this queue as the primary queue + * (Q0). Otherwise, the queue gets added to the multi-queue group whose primary + * queue's exec_queue_id is specified in the lower 32 bits of the 'value' field. + * All the other non-relevant bits of extension's 'value' field while adding the + * primary or the secondary queues of the group must be set to 0. + * - %DRM_XE_EXEC_QUEUE_SET_PROPERTY_MULTI_QUEUE_PRIORITY - Set the queue + * priority within the multi-queue group. Current valid priority values are 0–2 + * (default is 1), with higher values indicating higher priority. + * - %DRM_XE_EXEC_QUEUE_SET_DISABLE_STATE_CACHE_PERF_FIX - Set the queue to + * enable render color cache keying on BTP+BTI instead of just BTI + * (only valid for render queues). * * The example below shows how to use @drm_xe_exec_queue_create to create * a simple exec_queue (no parallel submission) of class @@ -1249,6 +1429,11 @@ struct drm_xe_exec_queue_create { #define DRM_XE_EXEC_QUEUE_SET_PROPERTY_PRIORITY 0 #define DRM_XE_EXEC_QUEUE_SET_PROPERTY_TIMESLICE 1 #define DRM_XE_EXEC_QUEUE_SET_PROPERTY_PXP_TYPE 2 +#define DRM_XE_EXEC_QUEUE_SET_HANG_REPLAY_STATE 3 +#define DRM_XE_EXEC_QUEUE_SET_PROPERTY_MULTI_GROUP 4 +#define DRM_XE_MULTI_GROUP_CREATE (1ull << 63) +#define DRM_XE_EXEC_QUEUE_SET_PROPERTY_MULTI_QUEUE_PRIORITY 5 +#define DRM_XE_EXEC_QUEUE_SET_DISABLE_STATE_CACHE_PERF_FIX 6 /** @extensions: Pointer to the first extension struct, if any */ __u64 extensions; @@ -1385,7 +1570,7 @@ struct drm_xe_sync { /** * @timeline_value: Input for the timeline sync object. Needs to be - * different than 0 when used with %DRM_XE_SYNC_FLAG_TIMELINE_SYNCOBJ. + * different than 0 when used with %DRM_XE_SYNC_TYPE_TIMELINE_SYNCOBJ. */ __u64 timeline_value; @@ -1420,6 +1605,7 @@ struct drm_xe_exec { /** @exec_queue_id: Exec queue ID for the batch buffer */ __u32 exec_queue_id; +#define DRM_XE_MAX_SYNCS 1024 /** @num_syncs: Amount of struct drm_xe_sync in array. */ __u32 num_syncs; @@ -1608,6 +1794,12 @@ enum drm_xe_oa_unit_type { /** @DRM_XE_OA_UNIT_TYPE_OAM: OAM OA unit */ DRM_XE_OA_UNIT_TYPE_OAM, + + /** @DRM_XE_OA_UNIT_TYPE_OAM_SAG: OAM_SAG OA unit */ + DRM_XE_OA_UNIT_TYPE_OAM_SAG, + + /** @DRM_XE_OA_UNIT_TYPE_MERT: MERT OA unit */ + DRM_XE_OA_UNIT_TYPE_MERT, }; /** @@ -1629,12 +1821,20 @@ struct drm_xe_oa_unit { #define DRM_XE_OA_CAPS_SYNCS (1 << 1) #define DRM_XE_OA_CAPS_OA_BUFFER_SIZE (1 << 2) #define DRM_XE_OA_CAPS_WAIT_NUM_REPORTS (1 << 3) +#define DRM_XE_OA_CAPS_OAM (1 << 4) +#define DRM_XE_OA_CAPS_OA_UNIT_GT_ID (1 << 5) /** @oa_timestamp_freq: OA timestamp freq */ __u64 oa_timestamp_freq; + /** @gt_id: gt id for this OA unit */ + __u16 gt_id; + + /** @reserved1: MBZ */ + __u16 reserved1[3]; + /** @reserved: MBZ */ - __u64 reserved[4]; + __u64 reserved[3]; /** @num_engines: number of engines in @eci array */ __u64 num_engines; @@ -1961,6 +2161,454 @@ struct drm_xe_query_eu_stall { __u64 sampling_rates[]; }; +/** + * struct drm_xe_madvise - Input of &DRM_IOCTL_XE_MADVISE + * + * This structure is used to set memory attributes for a virtual address range + * in a VM. The type of attribute is specified by @type, and the corresponding + * union member is used to provide additional parameters for @type. + * + * Supported attribute types: + * - DRM_XE_MEM_RANGE_ATTR_PREFERRED_LOC: Set preferred memory location. + * - DRM_XE_MEM_RANGE_ATTR_ATOMIC: Set atomic access policy. + * - DRM_XE_MEM_RANGE_ATTR_PAT: Set page attribute table index. + * - DRM_XE_VMA_ATTR_PURGEABLE_STATE: Set purgeable state for BOs. + * + * Example: + * + * .. code-block:: C + * + * struct drm_xe_madvise madvise = { + * .vm_id = vm_id, + * .start = 0x100000, + * .range = 0x2000, + * .type = DRM_XE_MEM_RANGE_ATTR_ATOMIC, + * .atomic_val = DRM_XE_ATOMIC_DEVICE, + * }; + * + * ioctl(fd, DRM_IOCTL_XE_MADVISE, &madvise); + * + */ +struct drm_xe_madvise { + /** @extensions: Pointer to the first extension struct, if any */ + __u64 extensions; + + /** @start: start of the virtual address range */ + __u64 start; + + /** @range: size of the virtual address range */ + __u64 range; + + /** @vm_id: vm_id of the virtual range */ + __u32 vm_id; + +#define DRM_XE_MEM_RANGE_ATTR_PREFERRED_LOC 0 +#define DRM_XE_MEM_RANGE_ATTR_ATOMIC 1 +#define DRM_XE_MEM_RANGE_ATTR_PAT 2 +#define DRM_XE_VMA_ATTR_PURGEABLE_STATE 3 + /** @type: type of attribute */ + __u32 type; + + union { + /** + * @preferred_mem_loc: preferred memory location + * + * Used when @type == DRM_XE_MEM_RANGE_ATTR_PREFERRED_LOC + * + * Supported values for @preferred_mem_loc.devmem_fd: + * - DRM_XE_PREFERRED_LOC_DEFAULT_DEVICE: set vram of fault tile as preferred loc + * - DRM_XE_PREFERRED_LOC_DEFAULT_SYSTEM: set smem as preferred loc + * + * Supported values for @preferred_mem_loc.migration_policy: + * - DRM_XE_MIGRATE_ALL_PAGES + * - DRM_XE_MIGRATE_ONLY_SYSTEM_PAGES + */ + struct { +#define DRM_XE_PREFERRED_LOC_DEFAULT_DEVICE 0 +#define DRM_XE_PREFERRED_LOC_DEFAULT_SYSTEM -1 + /** + * @preferred_mem_loc.devmem_fd: + * Device file-descriptor of the device where the + * preferred memory is located, or one of the + * above special values. Please also see + * @preferred_mem_loc.region_instance below. + */ + __u32 devmem_fd; + +#define DRM_XE_MIGRATE_ALL_PAGES 0 +#define DRM_XE_MIGRATE_ONLY_SYSTEM_PAGES 1 + /** @preferred_mem_loc.migration_policy: Page migration policy */ + __u16 migration_policy; + + /** + * @preferred_mem_loc.region_instance : Region instance. + * MBZ if @devmem_fd <= &DRM_XE_PREFERRED_LOC_DEFAULT_DEVICE. + * Otherwise should point to the desired device + * VRAM instance of the device indicated by + * @preferred_mem_loc.devmem_fd. + */ + __u16 region_instance; + + /** @preferred_mem_loc.reserved : Reserved */ + __u64 reserved; + } preferred_mem_loc; + + /** + * @atomic: Atomic access policy + * + * Used when @type == DRM_XE_MEM_RANGE_ATTR_ATOMIC. + * + * Supported values for @atomic.val: + * - DRM_XE_ATOMIC_UNDEFINED: Undefined or default behaviour. + * Support both GPU and CPU atomic operations for system allocator. + * Support GPU atomic operations for normal(bo) allocator. + * - DRM_XE_ATOMIC_DEVICE: Support GPU atomic operations. + * - DRM_XE_ATOMIC_GLOBAL: Support both GPU and CPU atomic operations. + * - DRM_XE_ATOMIC_CPU: Support CPU atomic only, no GPU atomics supported. + */ + struct { +#define DRM_XE_ATOMIC_UNDEFINED 0 +#define DRM_XE_ATOMIC_DEVICE 1 +#define DRM_XE_ATOMIC_GLOBAL 2 +#define DRM_XE_ATOMIC_CPU 3 + /** @atomic.val: value of atomic operation */ + __u32 val; + + /** @atomic.pad: MBZ */ + __u32 pad; + + /** @atomic.reserved: Reserved */ + __u64 reserved; + } atomic; + + /** + * @pat_index: Page attribute table index + * + * Used when @type == DRM_XE_MEM_RANGE_ATTR_PAT. + */ + struct { + /** @pat_index.val: PAT index value */ + __u32 val; + + /** @pat_index.pad: MBZ */ + __u32 pad; + + /** @pat_index.reserved: Reserved */ + __u64 reserved; + } pat_index; + + /** + * @purge_state_val: Purgeable state configuration + * + * Used when @type == DRM_XE_VMA_ATTR_PURGEABLE_STATE. + * + * Configures the purgeable state of buffer objects in the specified + * virtual address range. This allows applications to hint to the kernel + * about bo's usage patterns for better memory management. + * + * By default all VMAs are in WILLNEED state. + * + * Supported values for @purge_state_val.val: + * - DRM_XE_VMA_PURGEABLE_STATE_WILLNEED (0): Marks BO as needed. + * If the BO was previously purged, the kernel sets the __u32 at + * @retained_ptr to 0 (backing store lost) so the application knows + * it must recreate the BO. + * + * - DRM_XE_VMA_PURGEABLE_STATE_DONTNEED (1): Marks BO as not currently + * needed. Kernel may purge it under memory pressure to reclaim memory. + * Only applies to non-shared BOs. The kernel sets the __u32 at + * @retained_ptr to 1 if the backing store still exists (not yet purged), + * or 0 if it was already purged. + * + * Important: Once marked as DONTNEED, touching the BO's memory + * is undefined behavior. It may succeed temporarily (before the + * kernel purges the backing store) but will suddenly fail once + * the BO transitions to PURGED state. + * + * To transition back: use WILLNEED and check @retained_ptr — + * if 0, backing store was lost and the BO must be recreated. + * + * The following operations are blocked in DONTNEED state to + * prevent the BO from being re-mapped after madvise: + * - New mmap() calls: Fail with -EBUSY + * - VM_BIND operations: Fail with -EBUSY + * - New dma-buf exports: Fail with -EBUSY + * - CPU page faults (existing mmap): Fail with SIGBUS + * - GPU page faults (fault-mode VMs): Fail with -EACCES + */ + struct { +#define DRM_XE_VMA_PURGEABLE_STATE_WILLNEED 0 +#define DRM_XE_VMA_PURGEABLE_STATE_DONTNEED 1 + /** @purge_state_val.val: value for DRM_XE_VMA_ATTR_PURGEABLE_STATE */ + __u32 val; + + /** @purge_state_val.pad: MBZ */ + __u32 pad; + /** + * @purge_state_val.retained_ptr: Pointer to a __u32 output + * field for backing store status. + * + * Userspace must initialize the __u32 value at this address + * to 0 before the ioctl. Kernel writes a __u32 after the + * operation: + * - 1 if backing store exists (not purged) + * - 0 if backing store was purged + * + * If userspace fails to initialize to 0, ioctl returns -EINVAL. + * This ensures a safe default (0 = assume purged) if kernel + * cannot write the result. + * + * Similar to i915's drm_i915_gem_madvise.retained field. + */ + __u64 retained_ptr; + } purge_state_val; + }; + + /** @reserved: Reserved */ + __u64 reserved[2]; +}; + +/** + * struct drm_xe_mem_range_attr - Output of &DRM_IOCTL_XE_VM_QUERY_MEM_RANGES_ATTRS + * + * This structure is provided by userspace and filled by KMD in response to the + * DRM_IOCTL_XE_VM_QUERY_MEM_RANGES_ATTRS ioctl. It describes memory attributes of + * a memory ranges within a user specified address range in a VM. + * + * The structure includes information such as atomic access policy, + * page attribute table (PAT) index, and preferred memory location. + * Userspace allocates an array of these structures and passes a pointer to the + * ioctl to retrieve attributes for each memory ranges + * + * @extensions: Pointer to the first extension struct, if any + * @start: Start address of the memory range + * @end: End address of the virtual memory range + * + */ +struct drm_xe_mem_range_attr { + /** @extensions: Pointer to the first extension struct, if any */ + __u64 extensions; + + /** @start: start of the memory range */ + __u64 start; + + /** @end: end of the memory range */ + __u64 end; + + /** @preferred_mem_loc: preferred memory location */ + struct { + /** @preferred_mem_loc.devmem_fd: fd for preferred loc */ + __u32 devmem_fd; + + /** @preferred_mem_loc.migration_policy: Page migration policy */ + __u32 migration_policy; + } preferred_mem_loc; + + /** @atomic: Atomic access policy */ + struct { + /** @atomic.val: atomic attribute */ + __u32 val; + + /** @atomic.reserved: Reserved */ + __u32 reserved; + } atomic; + + /** @pat_index: Page attribute table index */ + struct { + /** @pat_index.val: PAT index */ + __u32 val; + + /** @pat_index.reserved: Reserved */ + __u32 reserved; + } pat_index; + + /** @reserved: Reserved */ + __u64 reserved[2]; +}; + +/** + * struct drm_xe_vm_query_mem_range_attr - Input of &DRM_IOCTL_XE_VM_QUERY_MEM_ATTRIBUTES + * + * This structure is used to query memory attributes of memory regions + * within a user specified address range in a VM. It provides detailed + * information about each memory range, including atomic access policy, + * page attribute table (PAT) index, and preferred memory location. + * + * Userspace first calls the ioctl with @num_mem_ranges = 0, + * @sizeof_mem_ranges_attr = 0 and @vector_of_vma_mem_attr = NULL to retrieve + * the number of memory regions and size of each memory range attribute. + * Then, it allocates a buffer of that size and calls the ioctl again to fill + * the buffer with memory range attributes. + * + * If second call fails with -ENOSPC, it means memory ranges changed between + * first call and now, retry IOCTL again with @num_mem_ranges = 0, + * @sizeof_mem_ranges_attr = 0 and @vector_of_vma_mem_attr = NULL followed by + * Second ioctl call. + * + * Example: + * + * .. code-block:: C + * + * struct drm_xe_vm_query_mem_range_attr query = { + * .vm_id = vm_id, + * .start = 0x100000, + * .range = 0x2000, + * }; + * + * // First ioctl call to get num of mem regions and sizeof each attribute + * ioctl(fd, DRM_IOCTL_XE_VM_QUERY_MEM_RANGE_ATTRS, &query); + * + * // Allocate buffer for the memory region attributes + * void *ptr = malloc(query.num_mem_ranges * query.sizeof_mem_range_attr); + * void *ptr_start = ptr; + * + * query.vector_of_mem_attr = (uintptr_t)ptr; + * + * // Second ioctl call to actually fill the memory attributes + * ioctl(fd, DRM_IOCTL_XE_VM_QUERY_MEM_RANGE_ATTRS, &query); + * + * // Iterate over the returned memory region attributes + * for (unsigned int i = 0; i < query.num_mem_ranges; ++i) { + * struct drm_xe_mem_range_attr *attr = (struct drm_xe_mem_range_attr *)ptr; + * + * // Do something with attr + * + * // Move pointer by one entry + * ptr += query.sizeof_mem_range_attr; + * } + * + * free(ptr_start); + */ +struct drm_xe_vm_query_mem_range_attr { + /** @extensions: Pointer to the first extension struct, if any */ + __u64 extensions; + + /** @vm_id: vm_id of the virtual range */ + __u32 vm_id; + + /** @num_mem_ranges: number of mem_ranges in range */ + __u32 num_mem_ranges; + + /** @start: start of the virtual address range */ + __u64 start; + + /** @range: size of the virtual address range */ + __u64 range; + + /** @sizeof_mem_range_attr: size of struct drm_xe_mem_range_attr */ + __u64 sizeof_mem_range_attr; + + /** @vector_of_mem_attr: userptr to array of struct drm_xe_mem_range_attr */ + __u64 vector_of_mem_attr; + + /** @reserved: Reserved */ + __u64 reserved[2]; + +}; + +/** + * struct drm_xe_exec_queue_set_property - exec queue set property + * + * Sets execution queue properties dynamically. + * Currently only %DRM_XE_EXEC_QUEUE_SET_PROPERTY_MULTI_QUEUE_PRIORITY + * property can be dynamically set. + */ +struct drm_xe_exec_queue_set_property { + /** @extensions: Pointer to the first extension struct, if any */ + __u64 extensions; + + /** @exec_queue_id: Exec queue ID */ + __u32 exec_queue_id; + + /** @property: property to set */ + __u32 property; + + /** @value: property value */ + __u64 value; + + /** @reserved: Reserved */ + __u64 reserved[2]; +}; + +/** + * DOC: Xe DRM RAS + * + * The enums and strings defined below map to the attributes of the DRM RAS Netlink Interface. + * Refer to Documentation/netlink/specs/drm_ras.yaml for complete interface specification. + * + * Node Registration + * ================= + * + * The driver registers DRM RAS nodes for each error severity level. + * enum drm_xe_ras_error_severity defines the node-id, while DRM_XE_RAS_ERROR_SEVERITY_NAMES maps + * node-id to node-name. + * + * Error Classification + * ==================== + * + * Each node contains a list of error counters. Each error is identified by a error-id and + * an error-name. enum drm_xe_ras_error_component defines the error-id, while + * DRM_XE_RAS_ERROR_COMPONENT_NAMES maps error-id to error-name. + * + * User Interface + * ============== + * + * To retrieve error values of a error counter, userspace applications should + * follow the below steps: + * + * 1. Use command LIST_NODES to enumerate all available nodes + * 2. Select node by node-id or node-name + * 3. Use command GET_ERROR_COUNTERS to list errors of specific node + * 4. Query specific error values using either error-id or error-name + * + * .. code-block:: C + * + * // Lookup tables for ID-to-name resolution + * static const char *nodes[] = DRM_XE_RAS_ERROR_SEVERITY_NAMES; + * static const char *errors[] = DRM_XE_RAS_ERROR_COMPONENT_NAMES; + * + */ + +/** + * enum drm_xe_ras_error_severity - DRM RAS error severity. + */ +enum drm_xe_ras_error_severity { + /** @DRM_XE_RAS_ERR_SEV_CORRECTABLE: Correctable Error */ + DRM_XE_RAS_ERR_SEV_CORRECTABLE = 0, + /** @DRM_XE_RAS_ERR_SEV_UNCORRECTABLE: Uncorrectable Error */ + DRM_XE_RAS_ERR_SEV_UNCORRECTABLE, + /** @DRM_XE_RAS_ERR_SEV_MAX: Max severity */ + DRM_XE_RAS_ERR_SEV_MAX /* non-ABI */ +}; + +/** + * enum drm_xe_ras_error_component - DRM RAS error component. + */ +enum drm_xe_ras_error_component { + /** @DRM_XE_RAS_ERR_COMP_CORE_COMPUTE: Core Compute Error */ + DRM_XE_RAS_ERR_COMP_CORE_COMPUTE = 1, + /** @DRM_XE_RAS_ERR_COMP_SOC_INTERNAL: SoC Internal Error */ + DRM_XE_RAS_ERR_COMP_SOC_INTERNAL, + /** @DRM_XE_RAS_ERR_COMP_MAX: Max Error */ + DRM_XE_RAS_ERR_COMP_MAX /* non-ABI */ +}; + +/* + * Error severity to name mapping. + */ +#define DRM_XE_RAS_ERROR_SEVERITY_NAMES { \ + [DRM_XE_RAS_ERR_SEV_CORRECTABLE] = "correctable-errors", \ + [DRM_XE_RAS_ERR_SEV_UNCORRECTABLE] = "uncorrectable-errors", \ +} + +/* + * Error component to name mapping. + */ +#define DRM_XE_RAS_ERROR_COMPONENT_NAMES { \ + [DRM_XE_RAS_ERR_COMP_CORE_COMPUTE] = "core-compute", \ + [DRM_XE_RAS_ERR_COMP_SOC_INTERNAL] = "soc-internal" \ +} + #if defined(__cplusplus) } #endif |