OpenHarmony/third_party_mesa3d
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

intel: Sync xe_drm.h take 2 part 3

Browse Source 
Sync xe_drm.h with commit ac7b89571d80 ("drm/xe/uapi: Kill exec_queue_set_property").

Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26360>
This commit is contained in:
José Roberto de Souza
2023-11-24 10:51:39 -08:00
committed by Marge Bot
parent d1109f67bb
commit 1f0a9f853c
7 changed files with 195 additions and 167 deletions
Download Patch File Download Diff File Expand all files Collapse all files

229
include/drm-uapi/xe_drm.h
View File

@@ -103,12 +103,11 @@ struct xe_user_extension {
#define DRM_XE_VM_CREATE 0x03 #define DRM_XE_VM_CREATE 0x03
#define DRM_XE_VM_DESTROY 0x04 #define DRM_XE_VM_DESTROY 0x04
#define DRM_XE_VM_BIND 0x05 #define DRM_XE_VM_BIND 0x05
#define DRM_XE_EXEC 0x06 #define DRM_XE_EXEC_QUEUE_CREATE 0x06
#define DRM_XE_EXEC_QUEUE_CREATE 0x07 #define DRM_XE_EXEC_QUEUE_DESTROY 0x07
#define DRM_XE_EXEC_QUEUE_DESTROY 0x08 #define DRM_XE_EXEC_QUEUE_GET_PROPERTY 0x08
#define DRM_XE_EXEC_QUEUE_SET_PROPERTY 0x09 #define DRM_XE_EXEC 0x09
#define DRM_XE_EXEC_QUEUE_GET_PROPERTY 0x0a #define DRM_XE_WAIT_USER_FENCE 0x0a
#define DRM_XE_WAIT_USER_FENCE 0x0b
/* Must be kept compact -- no holes */ /* Must be kept compact -- no holes */
#define DRM_IOCTL_XE_DEVICE_QUERY DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_DEVICE_QUERY, struct drm_xe_device_query) #define DRM_IOCTL_XE_DEVICE_QUERY DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_DEVICE_QUERY, struct drm_xe_device_query)
@@ -117,14 +116,21 @@ struct xe_user_extension {
#define DRM_IOCTL_XE_VM_CREATE DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_VM_CREATE, struct drm_xe_vm_create) #define DRM_IOCTL_XE_VM_CREATE DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_VM_CREATE, struct drm_xe_vm_create)
#define DRM_IOCTL_XE_VM_DESTROY DRM_IOW(DRM_COMMAND_BASE + DRM_XE_VM_DESTROY, struct drm_xe_vm_destroy) #define DRM_IOCTL_XE_VM_DESTROY DRM_IOW(DRM_COMMAND_BASE + DRM_XE_VM_DESTROY, struct drm_xe_vm_destroy)
#define DRM_IOCTL_XE_VM_BIND DRM_IOW(DRM_COMMAND_BASE + DRM_XE_VM_BIND, struct drm_xe_vm_bind) #define DRM_IOCTL_XE_VM_BIND DRM_IOW(DRM_COMMAND_BASE + DRM_XE_VM_BIND, struct drm_xe_vm_bind)
#define DRM_IOCTL_XE_EXEC DRM_IOW(DRM_COMMAND_BASE + DRM_XE_EXEC, struct drm_xe_exec)
#define DRM_IOCTL_XE_EXEC_QUEUE_CREATE DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_EXEC_QUEUE_CREATE, struct drm_xe_exec_queue_create) #define DRM_IOCTL_XE_EXEC_QUEUE_CREATE DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_EXEC_QUEUE_CREATE, struct drm_xe_exec_queue_create)
#define DRM_IOCTL_XE_EXEC_QUEUE_DESTROY DRM_IOW(DRM_COMMAND_BASE + DRM_XE_EXEC_QUEUE_DESTROY, struct drm_xe_exec_queue_destroy) #define DRM_IOCTL_XE_EXEC_QUEUE_DESTROY DRM_IOW(DRM_COMMAND_BASE + DRM_XE_EXEC_QUEUE_DESTROY, struct drm_xe_exec_queue_destroy)
#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_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_EXEC_QUEUE_GET_PROPERTY DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_EXEC_QUEUE_GET_PROPERTY, struct drm_xe_exec_queue_get_property) #define DRM_IOCTL_XE_EXEC_QUEUE_GET_PROPERTY DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_EXEC_QUEUE_GET_PROPERTY, struct drm_xe_exec_queue_get_property)
#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_WAIT_USER_FENCE DRM_IOWR(DRM_COMMAND_BASE + DRM_XE_WAIT_USER_FENCE, struct drm_xe_wait_user_fence)
/** struct drm_xe_engine_class_instance - instance of an engine class */ /**
* struct drm_xe_engine_class_instance - instance of an engine class
*
* It is returned as part of the @drm_xe_engine, but it also is used as
* the input of engine selection for both @drm_xe_exec_queue_create and
* @drm_xe_query_engine_cycles
*
*/
struct drm_xe_engine_class_instance { struct drm_xe_engine_class_instance {
#define DRM_XE_ENGINE_CLASS_RENDER 0 #define DRM_XE_ENGINE_CLASS_RENDER 0
#define DRM_XE_ENGINE_CLASS_COPY 1 #define DRM_XE_ENGINE_CLASS_COPY 1
@@ -145,6 +151,33 @@ struct drm_xe_engine_class_instance {
__u16 pad; __u16 pad;
}; };
/**
* struct drm_xe_engine - describe hardware engine
*/
struct drm_xe_engine {
/** @instance: The @drm_xe_engine_class_instance */
struct drm_xe_engine_class_instance instance;
/** @reserved: Reserved */
__u64 reserved[3];
};
/**
* struct drm_xe_query_engines - describe engines
*
* If a query is made with a struct @drm_xe_device_query where .query
* is equal to %DRM_XE_DEVICE_QUERY_ENGINES, then the reply uses an array of
* struct @drm_xe_query_engines in .data.
*/
struct drm_xe_query_engines {
/** @num_engines: number of engines returned in @engines */
__u32 num_engines;
/** @pad: MBZ */
__u32 pad;
/** @engines: The returned engines for this device */
struct drm_xe_engine engines[];
};
/** /**
* enum drm_xe_memory_class - Supported memory classes. * enum drm_xe_memory_class - Supported memory classes.
*/ */
@@ -160,10 +193,10 @@ enum drm_xe_memory_class {
}; };
/** /**
* struct drm_xe_query_mem_region - Describes some region as known to * struct drm_xe_mem_region - Describes some region as known to
* the driver. * the driver.
*/ */
struct drm_xe_query_mem_region { struct drm_xe_mem_region {
/** /**
* @mem_class: The memory class describing this region. * @mem_class: The memory class describing this region.
* *
@@ -177,18 +210,18 @@ struct drm_xe_query_mem_region {
* a unique pair. * a unique pair.
*/ */
__u16 instance; __u16 instance;
/** @pad: MBZ */
__u32 pad;
/** /**
* @min_page_size: Min page-size in bytes for this region. * @min_page_size: Min page-size in bytes for this region.
* *
* When the kernel allocates memory for this region, the * When the kernel allocates memory for this region, the
* underlying pages will be at least @min_page_size in size. * underlying pages will be at least @min_page_size in size.
* * Buffer objects with an allowable placement in this region must be
* Important note: When userspace allocates a GTT address which * created with a size aligned to this value.
* can point to memory allocated from this region, it must also * GPU virtual address mappings of (parts of) buffer objects that
* respect this minimum alignment. This is enforced by the * may be placed in this region must also have their GPU virtual
* kernel. * address and range aligned to this value.
* Affected IOCTLS will return %-EINVAL if alignment restrictions are
* not met.
*/ */
__u32 min_page_size; __u32 min_page_size;
/** /**
@@ -244,8 +277,8 @@ struct drm_xe_query_mem_region {
* in .data. struct drm_xe_query_engine_cycles is allocated by the user and * in .data. struct drm_xe_query_engine_cycles is allocated by the user and
* .data points to this allocated structure. * .data points to this allocated structure.
* *
* The query returns the engine cycles and the frequency that can * The query returns the engine cycles, which along with GT's @reference_clock,
* be used to calculate the engine timestamp. In addition the * can be used to calculate the engine timestamp. In addition the
* query returns a set of cpu timestamps that indicate when the command * query returns a set of cpu timestamps that indicate when the command
* streamer cycle count was captured. * streamer cycle count was captured.
*/ */
@@ -273,9 +306,6 @@ struct drm_xe_query_engine_cycles {
*/ */
__u64 engine_cycles; __u64 engine_cycles;
/** @engine_frequency: Frequency of the engine cycles in Hz. */
__u64 engine_frequency;
/** /**
* @cpu_timestamp: CPU timestamp in ns. The timestamp is captured before * @cpu_timestamp: CPU timestamp in ns. The timestamp is captured before
* reading the engine_cycles register using the reference clockid set by the * reading the engine_cycles register using the reference clockid set by the
@@ -298,12 +328,12 @@ struct drm_xe_query_engine_cycles {
* struct drm_xe_query_mem_regions in .data. * struct drm_xe_query_mem_regions in .data.
*/ */
struct drm_xe_query_mem_regions { struct drm_xe_query_mem_regions {
/** @num_regions: number of memory regions returned in @regions */ /** @num_mem_regions: number of memory regions returned in @mem_regions */
__u32 num_regions; __u32 num_mem_regions;
/** @pad: MBZ */ /** @pad: MBZ */
__u32 pad; __u32 pad;
/** @regions: The returned regions for this device */ /** @mem_regions: The returned memory regions for this device */
struct drm_xe_query_mem_region regions[]; struct drm_xe_mem_region mem_regions[];
}; };
/** /**
@@ -332,22 +362,26 @@ struct drm_xe_query_config {
}; };
/** /**
* struct drm_xe_query_gt - describe an individual GT. * struct drm_xe_gt - describe an individual GT.
* *
* To be used with drm_xe_query_gt_list, which will return a list with all the * To be used with drm_xe_query_gt_list, which will return a list with all the
* existing GT individual descriptions. * existing GT individual descriptions.
* Graphics Technology (GT) is a subset of a GPU/tile that is responsible for * Graphics Technology (GT) is a subset of a GPU/tile that is responsible for
* implementing graphics and/or media operations. * implementing graphics and/or media operations.
*/ */
struct drm_xe_query_gt { struct drm_xe_gt {
#define DRM_XE_QUERY_GT_TYPE_MAIN 0 #define DRM_XE_QUERY_GT_TYPE_MAIN 0
#define DRM_XE_QUERY_GT_TYPE_MEDIA 1 #define DRM_XE_QUERY_GT_TYPE_MEDIA 1
/** @type: GT type: Main or Media */ /** @type: GT type: Main or Media */
__u16 type; __u16 type;
/** @tile_id: Tile ID where this GT lives (Information only) */
__u16 tile_id;
/** @gt_id: Unique ID of this GT within the PCI Device */ /** @gt_id: Unique ID of this GT within the PCI Device */
__u16 gt_id; __u16 gt_id;
/** @clock_freq: A clock frequency for timestamp */ /** @pad: MBZ */
__u32 clock_freq; __u16 pad[3];
/** @reference_clock: A clock frequency for timestamp */
__u32 reference_clock;
/** /**
* @near_mem_regions: Bit mask of instances from * @near_mem_regions: Bit mask of instances from
* drm_xe_query_mem_regions that are nearest to the current engines * drm_xe_query_mem_regions that are nearest to the current engines
@@ -379,7 +413,7 @@ struct drm_xe_query_gt_list {
/** @pad: MBZ */ /** @pad: MBZ */
__u32 pad; __u32 pad;
/** @gt_list: The GT list returned for this device */ /** @gt_list: The GT list returned for this device */
struct drm_xe_query_gt gt_list[]; struct drm_xe_gt gt_list[];
}; };
/** /**
@@ -442,7 +476,7 @@ struct drm_xe_query_topology_mask {
* *
* .. code-block:: C * .. code-block:: C
* *
* struct drm_xe_engine_class_instance *hwe; * struct drm_xe_query_engines *engines;
* struct drm_xe_device_query query = { * struct drm_xe_device_query query = {
* .extensions = 0, * .extensions = 0,
* .query = DRM_XE_DEVICE_QUERY_ENGINES, * .query = DRM_XE_DEVICE_QUERY_ENGINES,
@@ -450,20 +484,24 @@ struct drm_xe_query_topology_mask {
* .data = 0, * .data = 0,
* }; * };
* ioctl(fd, DRM_IOCTL_XE_DEVICE_QUERY, &query); * ioctl(fd, DRM_IOCTL_XE_DEVICE_QUERY, &query);
* hwe = malloc(query.size); * engines = malloc(query.size);
* query.data = (uintptr_t)hwe; * query.data = (uintptr_t)engines;
* ioctl(fd, DRM_IOCTL_XE_DEVICE_QUERY, &query); * ioctl(fd, DRM_IOCTL_XE_DEVICE_QUERY, &query);
* int num_engines = query.size / sizeof(*hwe); * for (int i = 0; i < engines->num_engines; i++) {
* for (int i = 0; i < num_engines; i++) {
* printf("Engine %d: %s\n", i, * printf("Engine %d: %s\n", i,
* hwe[i].engine_class == DRM_XE_ENGINE_CLASS_RENDER ? "RENDER": * engines->engines[i].instance.engine_class ==
* hwe[i].engine_class == DRM_XE_ENGINE_CLASS_COPY ? "COPY": * DRM_XE_ENGINE_CLASS_RENDER ? "RENDER":
* hwe[i].engine_class == DRM_XE_ENGINE_CLASS_VIDEO_DECODE ? "VIDEO_DECODE": * engines->engines[i].instance.engine_class ==
* hwe[i].engine_class == DRM_XE_ENGINE_CLASS_VIDEO_ENHANCE ? "VIDEO_ENHANCE": * DRM_XE_ENGINE_CLASS_COPY ? "COPY":
* hwe[i].engine_class == DRM_XE_ENGINE_CLASS_COMPUTE ? "COMPUTE": * engines->engines[i].instance.engine_class ==
* DRM_XE_ENGINE_CLASS_VIDEO_DECODE ? "VIDEO_DECODE":
* engines->engines[i].instance.engine_class ==
* DRM_XE_ENGINE_CLASS_VIDEO_ENHANCE ? "VIDEO_ENHANCE":
* engines->engines[i].instance.engine_class ==
* DRM_XE_ENGINE_CLASS_COMPUTE ? "COMPUTE":
* "UNKNOWN"); * "UNKNOWN");
* } * }
* free(hwe); * free(engines);
*/ */
struct drm_xe_device_query { struct drm_xe_device_query {
/** @extensions: Pointer to the first extension struct, if any */ /** @extensions: Pointer to the first extension struct, if any */
@@ -494,14 +532,16 @@ struct drm_xe_gem_create {
__u64 extensions; __u64 extensions;
/** /**
* @size: Requested size for the object * @size: Size of the object to be created, must match region
* * (system or vram) minimum alignment (&min_page_size).
* The (page-aligned) allocated size for the object will be returned.
*/ */
__u64 size; __u64 size;
#define DRM_XE_GEM_CREATE_FLAG_DEFER_BACKING (0x1 << 24) /** @placement: A mask of memory instances of where BO can be placed. */
#define DRM_XE_GEM_CREATE_FLAG_SCANOUT (0x1 << 25) __u32 placement;
#define DRM_XE_GEM_CREATE_FLAG_DEFER_BACKING (1 << 0)
#define DRM_XE_GEM_CREATE_FLAG_SCANOUT (1 << 1)
/* /*
* When using VRAM as a possible placement, ensure that the corresponding 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 * allocation will always use the CPU accessible part of VRAM. This is important
@@ -517,7 +557,7 @@ struct drm_xe_gem_create {
* display surfaces, therefore the kernel requires setting this flag for such * display surfaces, therefore the kernel requires setting this flag for such
* objects, otherwise an error is thrown on small-bar systems. * objects, otherwise an error is thrown on small-bar systems.
*/ */
#define DRM_XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM (0x1 << 26) #define DRM_XE_GEM_CREATE_FLAG_NEEDS_VISIBLE_VRAM (1 << 2)
/** /**
* @flags: Flags, currently a mask of memory instances of where BO can * @flags: Flags, currently a mask of memory instances of where BO can
* be placed * be placed
@@ -559,7 +599,7 @@ struct drm_xe_gem_create {
#define DRM_XE_GEM_CPU_CACHING_WC 2 #define DRM_XE_GEM_CPU_CACHING_WC 2
__u16 cpu_caching; __u16 cpu_caching;
/** @pad: MBZ */ /** @pad: MBZ */
__u16 pad; __u16 pad[3];
/** @reserved: Reserved */ /** @reserved: Reserved */
__u64 reserved[2]; __u64 reserved[2];
@@ -601,13 +641,33 @@ struct drm_xe_ext_set_property {
}; };
struct drm_xe_vm_create { struct drm_xe_vm_create {
#define DRM_XE_VM_EXTENSION_SET_PROPERTY 0
/** @extensions: Pointer to the first extension struct, if any */ /** @extensions: Pointer to the first extension struct, if any */
__u64 extensions; __u64 extensions;
#define DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE (1 << 0) #define DRM_XE_VM_CREATE_FLAG_SCRATCH_PAGE (1 << 0)
#define DRM_XE_VM_CREATE_FLAG_COMPUTE_MODE (1 << 1) /*
* 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.
* 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
* different per-VM overcommit semantics that are enabled by
* DRM_XE_VM_CREATE_FLAG_FAULT_MODE (see below), but KMD may
* still enable recoverable pagefaults if supported by the device.
*/
#define DRM_XE_VM_CREATE_FLAG_LR_MODE (1 << 1)
#define DRM_XE_VM_CREATE_FLAG_ASYNC_DEFAULT (1 << 2) #define DRM_XE_VM_CREATE_FLAG_ASYNC_DEFAULT (1 << 2)
/*
* DRM_XE_VM_CREATE_FLAG_FAULT_MODE requires also
* DRM_XE_VM_CREATE_FLAG_LR_MODE. It allows memory to be allocated
* on demand when accessed, and also allows per-VM overcommit of memory.
* The xe driver internally uses recoverable pagefaults to implement
* this.
*/
#define DRM_XE_VM_CREATE_FLAG_FAULT_MODE (1 << 3) #define DRM_XE_VM_CREATE_FLAG_FAULT_MODE (1 << 3)
/** @flags: Flags */ /** @flags: Flags */
__u32 flags; __u32 flags;
@@ -631,6 +691,9 @@ struct drm_xe_vm_destroy {
}; };
struct drm_xe_vm_bind_op { struct drm_xe_vm_bind_op {
/** @extensions: Pointer to the first extension struct, if any */
__u64 extensions;
/** /**
* @obj: GEM object to operate on, MBZ for MAP_USERPTR, MBZ for UNMAP * @obj: GEM object to operate on, MBZ for MAP_USERPTR, MBZ for UNMAP
*/ */
@@ -698,12 +761,6 @@ struct drm_xe_vm_bind_op {
/** @addr: Address to operate on, MBZ for UNMAP_ALL */ /** @addr: Address to operate on, MBZ for UNMAP_ALL */
__u64 addr; __u64 addr;
/**
* @tile_mask: Mask for which tiles to create binds for, 0 == All tiles,
* only applies to creating new VMAs
*/
__u64 tile_mask;
#define DRM_XE_VM_BIND_OP_MAP 0x0 #define DRM_XE_VM_BIND_OP_MAP 0x0
#define DRM_XE_VM_BIND_OP_UNMAP 0x1 #define DRM_XE_VM_BIND_OP_UNMAP 0x1
#define DRM_XE_VM_BIND_OP_MAP_USERPTR 0x2 #define DRM_XE_VM_BIND_OP_MAP_USERPTR 0x2
@@ -737,8 +794,11 @@ struct drm_xe_vm_bind_op {
*/ */
__u32 prefetch_mem_region_instance; __u32 prefetch_mem_region_instance;
/** @pad: MBZ */
__u32 pad2;
/** @reserved: Reserved */ /** @reserved: Reserved */
__u64 reserved[2]; __u64 reserved[3];
}; };
struct drm_xe_vm_bind { struct drm_xe_vm_bind {
@@ -755,12 +815,12 @@ struct drm_xe_vm_bind {
*/ */
__u32 exec_queue_id; __u32 exec_queue_id;
/** @num_binds: number of binds in this IOCTL */
__u32 num_binds;
/** @pad: MBZ */ /** @pad: MBZ */
__u32 pad; __u32 pad;
/** @num_binds: number of binds in this IOCTL */
__u32 num_binds;
union { union {
/** @bind: used if num_binds == 1 */ /** @bind: used if num_binds == 1 */
struct drm_xe_vm_bind_op bind; struct drm_xe_vm_bind_op bind;
@@ -772,12 +832,12 @@ struct drm_xe_vm_bind {
__u64 vector_of_binds; __u64 vector_of_binds;
}; };
/** @pad: MBZ */
__u32 pad2;
/** @num_syncs: amount of syncs to wait on */ /** @num_syncs: amount of syncs to wait on */
__u32 num_syncs; __u32 num_syncs;
/** @pad2: MBZ */
__u32 pad2;
/** @syncs: pointer to struct drm_xe_sync array */ /** @syncs: pointer to struct drm_xe_sync array */
__u64 syncs; __u64 syncs;
@@ -799,18 +859,8 @@ struct drm_xe_vm_bind {
/* Monitor 64MB contiguous region with 2M sub-granularity */ /* Monitor 64MB contiguous region with 2M sub-granularity */
#define DRM_XE_ACC_GRANULARITY_64M 3 #define DRM_XE_ACC_GRANULARITY_64M 3
/** struct drm_xe_exec_queue_create {
* struct drm_xe_exec_queue_set_property - exec queue set property #define DRM_XE_EXEC_QUEUE_EXTENSION_SET_PROPERTY 0
*
* Same namespace for extensions as drm_xe_exec_queue_create
*/
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;
#define DRM_XE_EXEC_QUEUE_SET_PROPERTY_PRIORITY 0 #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_TIMESLICE 1
#define DRM_XE_EXEC_QUEUE_SET_PROPERTY_PREEMPTION_TIMEOUT 2 #define DRM_XE_EXEC_QUEUE_SET_PROPERTY_PREEMPTION_TIMEOUT 2
@@ -819,18 +869,7 @@ struct drm_xe_exec_queue_set_property {
#define DRM_XE_EXEC_QUEUE_SET_PROPERTY_ACC_TRIGGER 5 #define DRM_XE_EXEC_QUEUE_SET_PROPERTY_ACC_TRIGGER 5
#define DRM_XE_EXEC_QUEUE_SET_PROPERTY_ACC_NOTIFY 6 #define DRM_XE_EXEC_QUEUE_SET_PROPERTY_ACC_NOTIFY 6
#define DRM_XE_EXEC_QUEUE_SET_PROPERTY_ACC_GRANULARITY 7 #define DRM_XE_EXEC_QUEUE_SET_PROPERTY_ACC_GRANULARITY 7
/** @property: property to set */
__u32 property;
/** @value: property value */
__u64 value;
/** @reserved: Reserved */
__u64 reserved[2];
};
struct drm_xe_exec_queue_create {
#define DRM_XE_EXEC_QUEUE_EXTENSION_SET_PROPERTY 0
/** @extensions: Pointer to the first extension struct, if any */ /** @extensions: Pointer to the first extension struct, if any */
__u64 extensions; __u64 extensions;
@@ -895,15 +934,15 @@ struct drm_xe_sync {
/** @extensions: Pointer to the first extension struct, if any */ /** @extensions: Pointer to the first extension struct, if any */
__u64 extensions; __u64 extensions;
#define DRM_XE_SYNC_FLAG_SYNCOBJ 0x0 #define DRM_XE_SYNC_TYPE_SYNCOBJ 0x0
#define DRM_XE_SYNC_FLAG_TIMELINE_SYNCOBJ 0x1 #define DRM_XE_SYNC_TYPE_TIMELINE_SYNCOBJ 0x1
#define DRM_XE_SYNC_FLAG_DMA_BUF 0x2 #define DRM_XE_SYNC_TYPE_USER_FENCE 0x2
#define DRM_XE_SYNC_FLAG_USER_FENCE 0x3 /** @type: Type of the this sync object */
#define DRM_XE_SYNC_FLAG_SIGNAL 0x10 __u32 type;
__u32 flags;
/** @pad: MBZ */ #define DRM_XE_SYNC_FLAG_SIGNAL (1 << 0)
__u32 pad; /** @flags: Sync Flags */
__u32 flags;
union { union {
__u32 handle; __u32 handle;

7
src/gallium/drivers/iris/xe/iris_kmd_backend.c
View File

@@ -63,7 +63,7 @@ xe_gem_create(struct iris_bufmgr *bufmgr,
.flags = flags, .flags = flags,
}; };
for (uint16_t i = 0; i < regions_count; i++) for (uint16_t i = 0; i < regions_count; i++)
gem_create.flags |= BITFIELD_BIT(regions[i]->instance); gem_create.placement |= BITFIELD_BIT(regions[i]->instance);
const struct intel_device_info *devinfo = iris_bufmgr_get_device_info(bufmgr); const struct intel_device_info *devinfo = iris_bufmgr_get_device_info(bufmgr);
const struct intel_device_info_pat_entry *pat_entry; const struct intel_device_info_pat_entry *pat_entry;
@@ -350,13 +350,12 @@ xe_batch_submit(struct iris_batch *batch)
util_dynarray_foreach(&batch->exec_fences, struct iris_batch_fence, util_dynarray_foreach(&batch->exec_fences, struct iris_batch_fence,
fence) { fence) {
uint32_t flags = DRM_XE_SYNC_FLAG_SYNCOBJ;
if (fence->flags & IRIS_BATCH_FENCE_SIGNAL) if (fence->flags & IRIS_BATCH_FENCE_SIGNAL)
flags |= DRM_XE_SYNC_FLAG_SIGNAL; syncs[i].flags = DRM_XE_SYNC_FLAG_SIGNAL;
syncs[i].handle = fence->handle; syncs[i].handle = fence->handle;
syncs[i].flags = flags; syncs[i].type = DRM_XE_SYNC_TYPE_SYNCOBJ;
i++; i++;
} }
} }

14
src/intel/common/xe/intel_engine.c
View File

@@ -70,25 +70,23 @@ intel_engine_class_to_xe(enum intel_engine_class intel)
struct intel_query_engine_info * struct intel_query_engine_info *
xe_engine_get_info(int fd) xe_engine_get_info(int fd)
{ {
struct drm_xe_engine_class_instance *xe_engines; struct drm_xe_query_engines *xe_engines;
uint32_t len;
xe_engines = xe_device_query_alloc_fetch(fd, DRM_XE_DEVICE_QUERY_ENGINES, &len); xe_engines = xe_device_query_alloc_fetch(fd, DRM_XE_DEVICE_QUERY_ENGINES, NULL);
if (!xe_engines) if (!xe_engines)
return NULL; return NULL;
const uint32_t engines_count = len / sizeof(*xe_engines);
struct intel_query_engine_info *intel_engines_info; struct intel_query_engine_info *intel_engines_info;
intel_engines_info = calloc(1, sizeof(*intel_engines_info) + intel_engines_info = calloc(1, sizeof(*intel_engines_info) +
sizeof(*intel_engines_info->engines) * sizeof(*intel_engines_info->engines) *
engines_count); xe_engines->num_engines);
if (!intel_engines_info) { if (!intel_engines_info) {
goto error_free_xe_engines; goto error_free_xe_engines;
return NULL; return NULL;
} }
for (uint32_t i = 0; i < engines_count; i++) { for (uint32_t i = 0; i < xe_engines->num_engines; i++) {
struct drm_xe_engine_class_instance *xe_engine = &xe_engines[i]; struct drm_xe_engine_class_instance *xe_engine = &xe_engines->engines[i].instance;
struct intel_engine_class_instance *intel_engine = &intel_engines_info->engines[i]; struct intel_engine_class_instance *intel_engine = &intel_engines_info->engines[i];
intel_engine->engine_class = xe_engine_class_to_intel(xe_engine->engine_class); intel_engine->engine_class = xe_engine_class_to_intel(xe_engine->engine_class);
@@ -96,7 +94,7 @@ xe_engine_get_info(int fd)
intel_engine->gt_id = xe_engine->gt_id; intel_engine->gt_id = xe_engine->gt_id;
} }
intel_engines_info->num_engines = engines_count; intel_engines_info->num_engines = xe_engines->num_engines;
free(xe_engines); free(xe_engines);
return intel_engines_info; return intel_engines_info;

6
src/intel/dev/xe/intel_device_info.c
View File

@@ -85,8 +85,8 @@ intel_device_info_xe_query_regions(int fd, struct intel_device_info *devinfo,
if (!regions) if (!regions)
return false; return false;
for (int i = 0; i < regions->num_regions; i++) { for (int i = 0; i < regions->num_mem_regions; i++) {
struct drm_xe_query_mem_region *region = &regions->regions[i]; struct drm_xe_mem_region *region = &regions->mem_regions[i];
switch (region->mem_class) { switch (region->mem_class) {
case DRM_XE_MEM_REGION_CLASS_SYSMEM: { case DRM_XE_MEM_REGION_CLASS_SYSMEM: {
@@ -139,7 +139,7 @@ xe_query_gts(int fd, struct intel_device_info *devinfo)
for (uint32_t i = 0; i < gt_list->num_gt; i++) { for (uint32_t i = 0; i < gt_list->num_gt; i++) {
if (gt_list->gt_list[i].type == DRM_XE_QUERY_GT_TYPE_MAIN) if (gt_list->gt_list[i].type == DRM_XE_QUERY_GT_TYPE_MAIN)
devinfo->timestamp_frequency = gt_list->gt_list[i].clock_freq; devinfo->timestamp_frequency = gt_list->gt_list[i].reference_clock;
} }
free(gt_list); free(gt_list);

12
src/intel/vulkan/xe/anv_batch_chain.c
View File

@@ -45,7 +45,8 @@ xe_execute_simple_batch(struct anv_queue *queue,
return vk_errorf(device, VK_ERROR_UNKNOWN, "Unable to create sync obj"); return vk_errorf(device, VK_ERROR_UNKNOWN, "Unable to create sync obj");
struct drm_xe_sync sync = { struct drm_xe_sync sync = {
.flags = DRM_XE_SYNC_FLAG_SYNCOBJ | DRM_XE_SYNC_FLAG_SIGNAL, .type = DRM_XE_SYNC_TYPE_SYNCOBJ,
.flags = DRM_XE_SYNC_FLAG_SIGNAL,
.handle = syncobj_handle, .handle = syncobj_handle,
}; };
struct drm_xe_exec exec = { struct drm_xe_exec exec = {
@@ -91,14 +92,14 @@ xe_exec_fill_sync(struct drm_xe_sync *xe_sync, struct vk_sync *vk_sync,
xe_sync->handle = syncobj->syncobj; xe_sync->handle = syncobj->syncobj;
if (value) { if (value) {
xe_sync->flags |= DRM_XE_SYNC_FLAG_TIMELINE_SYNCOBJ; xe_sync->type = DRM_XE_SYNC_TYPE_TIMELINE_SYNCOBJ;
xe_sync->timeline_value = value; xe_sync->timeline_value = value;
} else { } else {
xe_sync->flags |= DRM_XE_SYNC_FLAG_SYNCOBJ; xe_sync->type = DRM_XE_SYNC_TYPE_SYNCOBJ;
} }
if (signal) if (signal)
xe_sync->flags |= DRM_XE_SYNC_FLAG_SIGNAL; xe_sync->flags = DRM_XE_SYNC_FLAG_SIGNAL;
} }
static VkResult static VkResult
@@ -193,7 +194,8 @@ xe_execute_trtt_batch(struct anv_sparse_submission *submit,
VkResult result; VkResult result;
struct drm_xe_sync extra_sync = { struct drm_xe_sync extra_sync = {
.flags = DRM_XE_SYNC_FLAG_TIMELINE_SYNCOBJ | DRM_XE_SYNC_FLAG_SIGNAL, .type = DRM_XE_SYNC_TYPE_TIMELINE_SYNCOBJ,
.flags = DRM_XE_SYNC_FLAG_SIGNAL,
.handle = trtt->timeline_handle, .handle = trtt->timeline_handle,
.timeline_value = trtt_bbo->timeline_val, .timeline_value = trtt_bbo->timeline_val,
}; };

3
src/intel/vulkan/xe/anv_kmd_backend.c
View File

@@ -63,7 +63,7 @@ xe_gem_create(struct anv_device *device,
.flags = flags, .flags = flags,
}; };
for (uint16_t i = 0; i < regions_count; i++) for (uint16_t i = 0; i < regions_count; i++)
gem_create.flags |= BITFIELD_BIT(regions[i]->instance); gem_create.placement |= BITFIELD_BIT(regions[i]->instance);
const struct intel_device_info_pat_entry *pat_entry = const struct intel_device_info_pat_entry *pat_entry =
anv_device_get_pat_entry(device, alloc_flags); anv_device_get_pat_entry(device, alloc_flags);
@@ -147,7 +147,6 @@ xe_vm_bind_op(struct anv_device *device,
.obj_offset = bind->bo_offset, .obj_offset = bind->bo_offset,
.range = bind->size, .range = bind->size,
.addr = intel_48b_address(bind->address), .addr = intel_48b_address(bind->address),
.tile_mask = 0,
.op = DRM_XE_VM_BIND_OP_UNMAP, .op = DRM_XE_VM_BIND_OP_UNMAP,
.flags = 0, .flags = 0,
.prefetch_mem_region_instance = 0, .prefetch_mem_region_instance = 0,

53
src/intel/vulkan/xe/anv_queue.c
View File

@@ -63,6 +63,22 @@ create_engine(struct anv_device *device,
&physical->queue.families[queue_family_index]; &physical->queue.families[queue_family_index];
const struct intel_query_engine_info *engines = physical->engine_info; const struct intel_query_engine_info *engines = physical->engine_info;
struct drm_xe_engine_class_instance *instances; struct drm_xe_engine_class_instance *instances;
const VkDeviceQueueGlobalPriorityCreateInfoKHR *queue_priority =
vk_find_struct_const(pCreateInfo->pNext,
DEVICE_QUEUE_GLOBAL_PRIORITY_CREATE_INFO_KHR);
const VkQueueGlobalPriorityKHR priority = queue_priority ?
queue_priority->globalPriority :
VK_QUEUE_GLOBAL_PRIORITY_MEDIUM_KHR;
/* As per spec, the driver implementation may deny requests to acquire
* a priority above the default priority (MEDIUM) if the caller does not
* have sufficient privileges. In this scenario VK_ERROR_NOT_PERMITTED_KHR
* is returned.
*/
if (physical->max_context_priority >= VK_QUEUE_GLOBAL_PRIORITY_MEDIUM_KHR) {
if (priority > physical->max_context_priority)
return vk_error(device, VK_ERROR_NOT_PERMITTED_KHR);
}
instances = vk_alloc(&device->vk.alloc, instances = vk_alloc(&device->vk.alloc,
sizeof(*instances) * queue_family->queueCount, 8, sizeof(*instances) * queue_family->queueCount, 8,
@@ -83,12 +99,18 @@ create_engine(struct anv_device *device,
} }
assert(device->vm_id != 0); assert(device->vm_id != 0);
struct drm_xe_ext_set_property ext = {
.base.name = DRM_XE_EXEC_QUEUE_EXTENSION_SET_PROPERTY,
.property = DRM_XE_EXEC_QUEUE_SET_PROPERTY_PRIORITY,
.value = anv_vk_priority_to_drm_sched_priority(priority),
};
struct drm_xe_exec_queue_create create = { struct drm_xe_exec_queue_create create = {
/* Allows KMD to pick one of those engines for the submission queue */ /* Allows KMD to pick one of those engines for the submission queue */
.instances = (uintptr_t)instances, .instances = (uintptr_t)instances,
.vm_id = device->vm_id, .vm_id = device->vm_id,
.width = 1, .width = 1,
.num_placements = count, .num_placements = count,
.extensions = (uintptr_t)&ext,
}; };
int ret = intel_ioctl(device->fd, DRM_IOCTL_XE_EXEC_QUEUE_CREATE, &create); int ret = intel_ioctl(device->fd, DRM_IOCTL_XE_EXEC_QUEUE_CREATE, &create);
vk_free(&device->vk.alloc, instances); vk_free(&device->vk.alloc, instances);
@@ -100,38 +122,7 @@ create_engine(struct anv_device *device,
else else
queue->exec_queue_id = create.exec_queue_id; queue->exec_queue_id = create.exec_queue_id;
const VkDeviceQueueGlobalPriorityCreateInfoKHR *queue_priority =
vk_find_struct_const(pCreateInfo->pNext,
DEVICE_QUEUE_GLOBAL_PRIORITY_CREATE_INFO_KHR);
const VkQueueGlobalPriorityKHR priority = queue_priority ?
queue_priority->globalPriority :
VK_QUEUE_GLOBAL_PRIORITY_MEDIUM_KHR;
/* As per spec, the driver implementation may deny requests to acquire
* a priority above the default priority (MEDIUM) if the caller does not
* have sufficient privileges. In this scenario VK_ERROR_NOT_PERMITTED_KHR
* is returned.
*/
if (physical->max_context_priority >= VK_QUEUE_GLOBAL_PRIORITY_MEDIUM_KHR) {
if (priority > physical->max_context_priority)
goto priority_error;
struct drm_xe_exec_queue_set_property exec_queue_property = {
.exec_queue_id = create.exec_queue_id,
.property = DRM_XE_EXEC_QUEUE_SET_PROPERTY_PRIORITY,
.value = anv_vk_priority_to_drm_sched_priority(priority),
};
ret = intel_ioctl(device->fd, DRM_IOCTL_XE_EXEC_QUEUE_SET_PROPERTY,
&exec_queue_property);
if (ret && priority > VK_QUEUE_GLOBAL_PRIORITY_MEDIUM_KHR)
goto priority_error;
}
return VK_SUCCESS; return VK_SUCCESS;
priority_error:
anv_xe_destroy_engine(device, queue);
return vk_error(device, VK_ERROR_NOT_PERMITTED_KHR);
} }
VkResult VkResult