2 * Copyright © 2015 Intel Corporation
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
28 #include <sys/sysinfo.h>
32 #include "drm-uapi/drm_fourcc.h"
34 #include "anv_private.h"
35 #include "util/strtod.h"
36 #include "util/debug.h"
37 #include "util/build_id.h"
38 #include "util/disk_cache.h"
39 #include "util/mesa-sha1.h"
40 #include "util/os_file.h"
41 #include "util/u_atomic.h"
42 #include "util/u_string.h"
45 #include "common/gen_defines.h"
46 #include "compiler/glsl_types.h"
48 #include "genxml/gen7_pack.h"
50 /* This is probably far to big but it reflects the max size used for messages
51 * in OpenGLs KHR_debug.
53 #define MAX_DEBUG_MESSAGE_LENGTH 4096
56 compiler_debug_log(void *data
, const char *fmt
, ...)
58 char str
[MAX_DEBUG_MESSAGE_LENGTH
];
59 struct anv_device
*device
= (struct anv_device
*)data
;
61 if (list_empty(&device
->instance
->debug_report_callbacks
.callbacks
))
66 (void) vsnprintf(str
, MAX_DEBUG_MESSAGE_LENGTH
, fmt
, args
);
69 vk_debug_report(&device
->instance
->debug_report_callbacks
,
70 VK_DEBUG_REPORT_DEBUG_BIT_EXT
,
71 VK_DEBUG_REPORT_OBJECT_TYPE_UNKNOWN_EXT
,
76 compiler_perf_log(void *data
, const char *fmt
, ...)
81 if (unlikely(INTEL_DEBUG
& DEBUG_PERF
))
82 intel_logd_v(fmt
, args
);
88 anv_compute_heap_size(int fd
, uint64_t gtt_size
)
90 /* Query the total ram from the system */
94 uint64_t total_ram
= (uint64_t)info
.totalram
* (uint64_t)info
.mem_unit
;
96 /* We don't want to burn too much ram with the GPU. If the user has 4GiB
97 * or less, we use at most half. If they have more than 4GiB, we use 3/4.
99 uint64_t available_ram
;
100 if (total_ram
<= 4ull * 1024ull * 1024ull * 1024ull)
101 available_ram
= total_ram
/ 2;
103 available_ram
= total_ram
* 3 / 4;
105 /* We also want to leave some padding for things we allocate in the driver,
106 * so don't go over 3/4 of the GTT either.
108 uint64_t available_gtt
= gtt_size
* 3 / 4;
110 return MIN2(available_ram
, available_gtt
);
114 anv_physical_device_init_heaps(struct anv_physical_device
*device
, int fd
)
117 if (anv_gem_get_context_param(fd
, 0, I915_CONTEXT_PARAM_GTT_SIZE
,
119 /* If, for whatever reason, we can't actually get the GTT size from the
120 * kernel (too old?) fall back to the aperture size.
122 anv_perf_warn(NULL
, NULL
,
123 "Failed to get I915_CONTEXT_PARAM_GTT_SIZE: %m");
125 if (anv_gem_get_aperture(fd
, >t_size
) == -1) {
126 return vk_errorf(NULL
, NULL
, VK_ERROR_INITIALIZATION_FAILED
,
127 "failed to get aperture size: %m");
131 device
->supports_48bit_addresses
= (device
->info
.gen
>= 8) &&
132 gtt_size
> (4ULL << 30 /* GiB */);
134 uint64_t heap_size
= anv_compute_heap_size(fd
, gtt_size
);
136 if (heap_size
> (2ull << 30) && !device
->supports_48bit_addresses
) {
137 /* When running with an overridden PCI ID, we may get a GTT size from
138 * the kernel that is greater than 2 GiB but the execbuf check for 48bit
139 * address support can still fail. Just clamp the address space size to
140 * 2 GiB if we don't have 48-bit support.
142 intel_logw("%s:%d: The kernel reported a GTT size larger than 2 GiB but "
143 "not support for 48-bit addresses",
145 heap_size
= 2ull << 30;
148 if (heap_size
<= 3ull * (1ull << 30)) {
149 /* In this case, everything fits nicely into the 32-bit address space,
150 * so there's no need for supporting 48bit addresses on client-allocated
153 device
->memory
.heap_count
= 1;
154 device
->memory
.heaps
[0] = (struct anv_memory_heap
) {
155 .vma_start
= LOW_HEAP_MIN_ADDRESS
,
156 .vma_size
= LOW_HEAP_SIZE
,
158 .flags
= VK_MEMORY_HEAP_DEVICE_LOCAL_BIT
,
159 .supports_48bit_addresses
= false,
162 /* Not everything will fit nicely into a 32-bit address space. In this
163 * case we need a 64-bit heap. Advertise a small 32-bit heap and a
164 * larger 48-bit heap. If we're in this case, then we have a total heap
165 * size larger than 3GiB which most likely means they have 8 GiB of
166 * video memory and so carving off 1 GiB for the 32-bit heap should be
169 const uint64_t heap_size_32bit
= 1ull << 30;
170 const uint64_t heap_size_48bit
= heap_size
- heap_size_32bit
;
172 assert(device
->supports_48bit_addresses
);
174 device
->memory
.heap_count
= 2;
175 device
->memory
.heaps
[0] = (struct anv_memory_heap
) {
176 .vma_start
= HIGH_HEAP_MIN_ADDRESS
,
177 /* Leave the last 4GiB out of the high vma range, so that no state
178 * base address + size can overflow 48 bits. For more information see
179 * the comment about Wa32bitGeneralStateOffset in anv_allocator.c
181 .vma_size
= gtt_size
- (1ull << 32) - HIGH_HEAP_MIN_ADDRESS
,
182 .size
= heap_size_48bit
,
183 .flags
= VK_MEMORY_HEAP_DEVICE_LOCAL_BIT
,
184 .supports_48bit_addresses
= true,
186 device
->memory
.heaps
[1] = (struct anv_memory_heap
) {
187 .vma_start
= LOW_HEAP_MIN_ADDRESS
,
188 .vma_size
= LOW_HEAP_SIZE
,
189 .size
= heap_size_32bit
,
190 .flags
= VK_MEMORY_HEAP_DEVICE_LOCAL_BIT
,
191 .supports_48bit_addresses
= false,
195 uint32_t type_count
= 0;
196 for (uint32_t heap
= 0; heap
< device
->memory
.heap_count
; heap
++) {
197 uint32_t valid_buffer_usage
= ~0;
199 /* There appears to be a hardware issue in the VF cache where it only
200 * considers the bottom 32 bits of memory addresses. If you happen to
201 * have two vertex buffers which get placed exactly 4 GiB apart and use
202 * them in back-to-back draw calls, you can get collisions. In order to
203 * solve this problem, we require vertex and index buffers be bound to
204 * memory allocated out of the 32-bit heap.
206 if (device
->memory
.heaps
[heap
].supports_48bit_addresses
) {
207 valid_buffer_usage
&= ~(VK_BUFFER_USAGE_INDEX_BUFFER_BIT
|
208 VK_BUFFER_USAGE_VERTEX_BUFFER_BIT
);
211 if (device
->info
.has_llc
) {
212 /* Big core GPUs share LLC with the CPU and thus one memory type can be
213 * both cached and coherent at the same time.
215 device
->memory
.types
[type_count
++] = (struct anv_memory_type
) {
216 .propertyFlags
= VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT
|
217 VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT
|
218 VK_MEMORY_PROPERTY_HOST_COHERENT_BIT
|
219 VK_MEMORY_PROPERTY_HOST_CACHED_BIT
,
221 .valid_buffer_usage
= valid_buffer_usage
,
224 /* The spec requires that we expose a host-visible, coherent memory
225 * type, but Atom GPUs don't share LLC. Thus we offer two memory types
226 * to give the application a choice between cached, but not coherent and
227 * coherent but uncached (WC though).
229 device
->memory
.types
[type_count
++] = (struct anv_memory_type
) {
230 .propertyFlags
= VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT
|
231 VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT
|
232 VK_MEMORY_PROPERTY_HOST_COHERENT_BIT
,
234 .valid_buffer_usage
= valid_buffer_usage
,
236 device
->memory
.types
[type_count
++] = (struct anv_memory_type
) {
237 .propertyFlags
= VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT
|
238 VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT
|
239 VK_MEMORY_PROPERTY_HOST_CACHED_BIT
,
241 .valid_buffer_usage
= valid_buffer_usage
,
245 device
->memory
.type_count
= type_count
;
251 anv_physical_device_init_uuids(struct anv_physical_device
*device
)
253 const struct build_id_note
*note
=
254 build_id_find_nhdr_for_addr(anv_physical_device_init_uuids
);
256 return vk_errorf(device
->instance
, device
,
257 VK_ERROR_INITIALIZATION_FAILED
,
258 "Failed to find build-id");
261 unsigned build_id_len
= build_id_length(note
);
262 if (build_id_len
< 20) {
263 return vk_errorf(device
->instance
, device
,
264 VK_ERROR_INITIALIZATION_FAILED
,
265 "build-id too short. It needs to be a SHA");
268 memcpy(device
->driver_build_sha1
, build_id_data(note
), 20);
270 struct mesa_sha1 sha1_ctx
;
272 STATIC_ASSERT(VK_UUID_SIZE
<= sizeof(sha1
));
274 /* The pipeline cache UUID is used for determining when a pipeline cache is
275 * invalid. It needs both a driver build and the PCI ID of the device.
277 _mesa_sha1_init(&sha1_ctx
);
278 _mesa_sha1_update(&sha1_ctx
, build_id_data(note
), build_id_len
);
279 _mesa_sha1_update(&sha1_ctx
, &device
->chipset_id
,
280 sizeof(device
->chipset_id
));
281 _mesa_sha1_update(&sha1_ctx
, &device
->always_use_bindless
,
282 sizeof(device
->always_use_bindless
));
283 _mesa_sha1_update(&sha1_ctx
, &device
->has_a64_buffer_access
,
284 sizeof(device
->has_a64_buffer_access
));
285 _mesa_sha1_update(&sha1_ctx
, &device
->has_bindless_images
,
286 sizeof(device
->has_bindless_images
));
287 _mesa_sha1_update(&sha1_ctx
, &device
->has_bindless_samplers
,
288 sizeof(device
->has_bindless_samplers
));
289 _mesa_sha1_final(&sha1_ctx
, sha1
);
290 memcpy(device
->pipeline_cache_uuid
, sha1
, VK_UUID_SIZE
);
292 /* The driver UUID is used for determining sharability of images and memory
293 * between two Vulkan instances in separate processes. People who want to
294 * share memory need to also check the device UUID (below) so all this
295 * needs to be is the build-id.
297 memcpy(device
->driver_uuid
, build_id_data(note
), VK_UUID_SIZE
);
299 /* The device UUID uniquely identifies the given device within the machine.
300 * Since we never have more than one device, this doesn't need to be a real
301 * UUID. However, on the off-chance that someone tries to use this to
302 * cache pre-tiled images or something of the like, we use the PCI ID and
303 * some bits of ISL info to ensure that this is safe.
305 _mesa_sha1_init(&sha1_ctx
);
306 _mesa_sha1_update(&sha1_ctx
, &device
->chipset_id
,
307 sizeof(device
->chipset_id
));
308 _mesa_sha1_update(&sha1_ctx
, &device
->isl_dev
.has_bit6_swizzling
,
309 sizeof(device
->isl_dev
.has_bit6_swizzling
));
310 _mesa_sha1_final(&sha1_ctx
, sha1
);
311 memcpy(device
->device_uuid
, sha1
, VK_UUID_SIZE
);
317 anv_physical_device_init_disk_cache(struct anv_physical_device
*device
)
319 #ifdef ENABLE_SHADER_CACHE
321 MAYBE_UNUSED
int len
= snprintf(renderer
, sizeof(renderer
), "anv_%04x",
323 assert(len
== sizeof(renderer
) - 2);
326 _mesa_sha1_format(timestamp
, device
->driver_build_sha1
);
328 const uint64_t driver_flags
=
329 brw_get_compiler_config_value(device
->compiler
);
330 device
->disk_cache
= disk_cache_create(renderer
, timestamp
, driver_flags
);
332 device
->disk_cache
= NULL
;
337 anv_physical_device_free_disk_cache(struct anv_physical_device
*device
)
339 #ifdef ENABLE_SHADER_CACHE
340 if (device
->disk_cache
)
341 disk_cache_destroy(device
->disk_cache
);
343 assert(device
->disk_cache
== NULL
);
348 get_available_system_memory()
350 char *meminfo
= os_read_file("/proc/meminfo");
354 char *str
= strstr(meminfo
, "MemAvailable:");
360 uint64_t kb_mem_available
;
361 if (sscanf(str
, "MemAvailable: %" PRIx64
, &kb_mem_available
) == 1) {
363 return kb_mem_available
<< 10;
371 anv_physical_device_init(struct anv_physical_device
*device
,
372 struct anv_instance
*instance
,
373 drmDevicePtr drm_device
)
375 const char *primary_path
= drm_device
->nodes
[DRM_NODE_PRIMARY
];
376 const char *path
= drm_device
->nodes
[DRM_NODE_RENDER
];
381 brw_process_intel_debug_variable();
383 fd
= open(path
, O_RDWR
| O_CLOEXEC
);
385 return vk_error(VK_ERROR_INCOMPATIBLE_DRIVER
);
387 device
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
388 device
->instance
= instance
;
390 assert(strlen(path
) < ARRAY_SIZE(device
->path
));
391 snprintf(device
->path
, ARRAY_SIZE(device
->path
), "%s", path
);
393 device
->no_hw
= getenv("INTEL_NO_HW") != NULL
;
395 const int pci_id_override
= gen_get_pci_device_id_override();
396 if (pci_id_override
< 0) {
397 device
->chipset_id
= anv_gem_get_param(fd
, I915_PARAM_CHIPSET_ID
);
398 if (!device
->chipset_id
) {
399 result
= vk_error(VK_ERROR_INCOMPATIBLE_DRIVER
);
403 device
->chipset_id
= pci_id_override
;
404 device
->no_hw
= true;
407 device
->pci_info
.domain
= drm_device
->businfo
.pci
->domain
;
408 device
->pci_info
.bus
= drm_device
->businfo
.pci
->bus
;
409 device
->pci_info
.device
= drm_device
->businfo
.pci
->dev
;
410 device
->pci_info
.function
= drm_device
->businfo
.pci
->func
;
412 device
->name
= gen_get_device_name(device
->chipset_id
);
413 if (!gen_get_device_info(device
->chipset_id
, &device
->info
)) {
414 result
= vk_error(VK_ERROR_INCOMPATIBLE_DRIVER
);
418 if (device
->info
.is_haswell
) {
419 intel_logw("Haswell Vulkan support is incomplete");
420 } else if (device
->info
.gen
== 7 && !device
->info
.is_baytrail
) {
421 intel_logw("Ivy Bridge Vulkan support is incomplete");
422 } else if (device
->info
.gen
== 7 && device
->info
.is_baytrail
) {
423 intel_logw("Bay Trail Vulkan support is incomplete");
424 } else if (device
->info
.gen
>= 8 && device
->info
.gen
<= 11) {
425 /* Gen8-11 fully supported */
427 result
= vk_errorf(device
->instance
, device
,
428 VK_ERROR_INCOMPATIBLE_DRIVER
,
429 "Vulkan not yet supported on %s", device
->name
);
433 device
->cmd_parser_version
= -1;
434 if (device
->info
.gen
== 7) {
435 device
->cmd_parser_version
=
436 anv_gem_get_param(fd
, I915_PARAM_CMD_PARSER_VERSION
);
437 if (device
->cmd_parser_version
== -1) {
438 result
= vk_errorf(device
->instance
, device
,
439 VK_ERROR_INITIALIZATION_FAILED
,
440 "failed to get command parser version");
445 if (!anv_gem_get_param(fd
, I915_PARAM_HAS_WAIT_TIMEOUT
)) {
446 result
= vk_errorf(device
->instance
, device
,
447 VK_ERROR_INITIALIZATION_FAILED
,
448 "kernel missing gem wait");
452 if (!anv_gem_get_param(fd
, I915_PARAM_HAS_EXECBUF2
)) {
453 result
= vk_errorf(device
->instance
, device
,
454 VK_ERROR_INITIALIZATION_FAILED
,
455 "kernel missing execbuf2");
459 if (!device
->info
.has_llc
&&
460 anv_gem_get_param(fd
, I915_PARAM_MMAP_VERSION
) < 1) {
461 result
= vk_errorf(device
->instance
, device
,
462 VK_ERROR_INITIALIZATION_FAILED
,
463 "kernel missing wc mmap");
467 result
= anv_physical_device_init_heaps(device
, fd
);
468 if (result
!= VK_SUCCESS
)
471 device
->has_exec_async
= anv_gem_get_param(fd
, I915_PARAM_HAS_EXEC_ASYNC
);
472 device
->has_exec_capture
= anv_gem_get_param(fd
, I915_PARAM_HAS_EXEC_CAPTURE
);
473 device
->has_exec_fence
= anv_gem_get_param(fd
, I915_PARAM_HAS_EXEC_FENCE
);
474 device
->has_syncobj
= anv_gem_get_param(fd
, I915_PARAM_HAS_EXEC_FENCE_ARRAY
);
475 device
->has_syncobj_wait
= device
->has_syncobj
&&
476 anv_gem_supports_syncobj_wait(fd
);
477 device
->has_context_priority
= anv_gem_has_context_priority(fd
);
479 device
->use_softpin
= anv_gem_get_param(fd
, I915_PARAM_HAS_EXEC_SOFTPIN
)
480 && device
->supports_48bit_addresses
;
482 device
->has_context_isolation
=
483 anv_gem_get_param(fd
, I915_PARAM_HAS_CONTEXT_ISOLATION
);
485 device
->always_use_bindless
=
486 env_var_as_boolean("ANV_ALWAYS_BINDLESS", false);
488 /* We first got the A64 messages on broadwell and we can only use them if
489 * we can pass addresses directly into the shader which requires softpin.
491 device
->has_a64_buffer_access
= device
->info
.gen
>= 8 &&
494 /* We first get bindless image access on Skylake and we can only really do
495 * it if we don't have any relocations so we need softpin.
497 device
->has_bindless_images
= device
->info
.gen
>= 9 &&
500 /* We've had bindless samplers since Ivy Bridge (forever in Vulkan terms)
501 * because it's just a matter of setting the sampler address in the sample
502 * message header. However, we've not bothered to wire it up for vec4 so
503 * we leave it disabled on gen7.
505 device
->has_bindless_samplers
= device
->info
.gen
>= 8;
507 device
->has_mem_available
= get_available_system_memory() != 0;
509 /* Starting with Gen10, the timestamp frequency of the command streamer may
510 * vary from one part to another. We can query the value from the kernel.
512 if (device
->info
.gen
>= 10) {
513 int timestamp_frequency
=
514 anv_gem_get_param(fd
, I915_PARAM_CS_TIMESTAMP_FREQUENCY
);
516 if (timestamp_frequency
< 0)
517 intel_logw("Kernel 4.16-rc1+ required to properly query CS timestamp frequency");
519 device
->info
.timestamp_frequency
= timestamp_frequency
;
522 /* GENs prior to 8 do not support EU/Subslice info */
523 if (device
->info
.gen
>= 8) {
524 device
->subslice_total
= anv_gem_get_param(fd
, I915_PARAM_SUBSLICE_TOTAL
);
525 device
->eu_total
= anv_gem_get_param(fd
, I915_PARAM_EU_TOTAL
);
527 /* Without this information, we cannot get the right Braswell
528 * brandstrings, and we have to use conservative numbers for GPGPU on
529 * many platforms, but otherwise, things will just work.
531 if (device
->subslice_total
< 1 || device
->eu_total
< 1) {
532 intel_logw("Kernel 4.1 required to properly query GPU properties");
534 } else if (device
->info
.gen
== 7) {
535 device
->subslice_total
= 1 << (device
->info
.gt
- 1);
538 if (device
->info
.is_cherryview
&&
539 device
->subslice_total
> 0 && device
->eu_total
> 0) {
540 /* Logical CS threads = EUs per subslice * num threads per EU */
541 uint32_t max_cs_threads
=
542 device
->eu_total
/ device
->subslice_total
* device
->info
.num_thread_per_eu
;
544 /* Fuse configurations may give more threads than expected, never less. */
545 if (max_cs_threads
> device
->info
.max_cs_threads
)
546 device
->info
.max_cs_threads
= max_cs_threads
;
549 device
->compiler
= brw_compiler_create(NULL
, &device
->info
);
550 if (device
->compiler
== NULL
) {
551 result
= vk_error(VK_ERROR_OUT_OF_HOST_MEMORY
);
554 device
->compiler
->shader_debug_log
= compiler_debug_log
;
555 device
->compiler
->shader_perf_log
= compiler_perf_log
;
556 device
->compiler
->supports_pull_constants
= false;
557 device
->compiler
->constant_buffer_0_is_relative
=
558 device
->info
.gen
< 8 || !device
->has_context_isolation
;
559 device
->compiler
->supports_shader_constants
= true;
561 /* Broadwell PRM says:
563 * "Before Gen8, there was a historical configuration control field to
564 * swizzle address bit[6] for in X/Y tiling modes. This was set in three
565 * different places: TILECTL[1:0], ARB_MODE[5:4], and
566 * DISP_ARB_CTL[14:13].
568 * For Gen8 and subsequent generations, the swizzle fields are all
569 * reserved, and the CPU's memory controller performs all address
570 * swizzling modifications."
573 device
->info
.gen
< 8 && anv_gem_get_bit6_swizzle(fd
, I915_TILING_X
);
575 isl_device_init(&device
->isl_dev
, &device
->info
, swizzled
);
577 result
= anv_physical_device_init_uuids(device
);
578 if (result
!= VK_SUCCESS
)
581 anv_physical_device_init_disk_cache(device
);
583 if (instance
->enabled_extensions
.KHR_display
) {
584 master_fd
= open(primary_path
, O_RDWR
| O_CLOEXEC
);
585 if (master_fd
>= 0) {
586 /* prod the device with a GETPARAM call which will fail if
587 * we don't have permission to even render on this device
589 if (anv_gem_get_param(master_fd
, I915_PARAM_CHIPSET_ID
) == 0) {
595 device
->master_fd
= master_fd
;
597 result
= anv_init_wsi(device
);
598 if (result
!= VK_SUCCESS
) {
599 ralloc_free(device
->compiler
);
600 anv_physical_device_free_disk_cache(device
);
604 anv_physical_device_get_supported_extensions(device
,
605 &device
->supported_extensions
);
608 device
->local_fd
= fd
;
620 anv_physical_device_finish(struct anv_physical_device
*device
)
622 anv_finish_wsi(device
);
623 anv_physical_device_free_disk_cache(device
);
624 ralloc_free(device
->compiler
);
625 close(device
->local_fd
);
626 if (device
->master_fd
>= 0)
627 close(device
->master_fd
);
631 default_alloc_func(void *pUserData
, size_t size
, size_t align
,
632 VkSystemAllocationScope allocationScope
)
638 default_realloc_func(void *pUserData
, void *pOriginal
, size_t size
,
639 size_t align
, VkSystemAllocationScope allocationScope
)
641 return realloc(pOriginal
, size
);
645 default_free_func(void *pUserData
, void *pMemory
)
650 static const VkAllocationCallbacks default_alloc
= {
652 .pfnAllocation
= default_alloc_func
,
653 .pfnReallocation
= default_realloc_func
,
654 .pfnFree
= default_free_func
,
657 VkResult
anv_EnumerateInstanceExtensionProperties(
658 const char* pLayerName
,
659 uint32_t* pPropertyCount
,
660 VkExtensionProperties
* pProperties
)
662 VK_OUTARRAY_MAKE(out
, pProperties
, pPropertyCount
);
664 for (int i
= 0; i
< ANV_INSTANCE_EXTENSION_COUNT
; i
++) {
665 if (anv_instance_extensions_supported
.extensions
[i
]) {
666 vk_outarray_append(&out
, prop
) {
667 *prop
= anv_instance_extensions
[i
];
672 return vk_outarray_status(&out
);
675 VkResult
anv_CreateInstance(
676 const VkInstanceCreateInfo
* pCreateInfo
,
677 const VkAllocationCallbacks
* pAllocator
,
678 VkInstance
* pInstance
)
680 struct anv_instance
*instance
;
683 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_INSTANCE_CREATE_INFO
);
685 struct anv_instance_extension_table enabled_extensions
= {};
686 for (uint32_t i
= 0; i
< pCreateInfo
->enabledExtensionCount
; i
++) {
688 for (idx
= 0; idx
< ANV_INSTANCE_EXTENSION_COUNT
; idx
++) {
689 if (strcmp(pCreateInfo
->ppEnabledExtensionNames
[i
],
690 anv_instance_extensions
[idx
].extensionName
) == 0)
694 if (idx
>= ANV_INSTANCE_EXTENSION_COUNT
)
695 return vk_error(VK_ERROR_EXTENSION_NOT_PRESENT
);
697 if (!anv_instance_extensions_supported
.extensions
[idx
])
698 return vk_error(VK_ERROR_EXTENSION_NOT_PRESENT
);
700 enabled_extensions
.extensions
[idx
] = true;
703 instance
= vk_alloc2(&default_alloc
, pAllocator
, sizeof(*instance
), 8,
704 VK_SYSTEM_ALLOCATION_SCOPE_INSTANCE
);
706 return vk_error(VK_ERROR_OUT_OF_HOST_MEMORY
);
708 instance
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
711 instance
->alloc
= *pAllocator
;
713 instance
->alloc
= default_alloc
;
715 instance
->app_info
= (struct anv_app_info
) { .api_version
= 0 };
716 if (pCreateInfo
->pApplicationInfo
) {
717 const VkApplicationInfo
*app
= pCreateInfo
->pApplicationInfo
;
719 instance
->app_info
.app_name
=
720 vk_strdup(&instance
->alloc
, app
->pApplicationName
,
721 VK_SYSTEM_ALLOCATION_SCOPE_INSTANCE
);
722 instance
->app_info
.app_version
= app
->applicationVersion
;
724 instance
->app_info
.engine_name
=
725 vk_strdup(&instance
->alloc
, app
->pEngineName
,
726 VK_SYSTEM_ALLOCATION_SCOPE_INSTANCE
);
727 instance
->app_info
.engine_version
= app
->engineVersion
;
729 instance
->app_info
.api_version
= app
->apiVersion
;
732 if (instance
->app_info
.api_version
== 0)
733 instance
->app_info
.api_version
= VK_API_VERSION_1_0
;
735 instance
->enabled_extensions
= enabled_extensions
;
737 for (unsigned i
= 0; i
< ARRAY_SIZE(instance
->dispatch
.entrypoints
); i
++) {
738 /* Vulkan requires that entrypoints for extensions which have not been
739 * enabled must not be advertised.
741 if (!anv_instance_entrypoint_is_enabled(i
, instance
->app_info
.api_version
,
742 &instance
->enabled_extensions
)) {
743 instance
->dispatch
.entrypoints
[i
] = NULL
;
745 instance
->dispatch
.entrypoints
[i
] =
746 anv_instance_dispatch_table
.entrypoints
[i
];
750 for (unsigned i
= 0; i
< ARRAY_SIZE(instance
->device_dispatch
.entrypoints
); i
++) {
751 /* Vulkan requires that entrypoints for extensions which have not been
752 * enabled must not be advertised.
754 if (!anv_device_entrypoint_is_enabled(i
, instance
->app_info
.api_version
,
755 &instance
->enabled_extensions
, NULL
)) {
756 instance
->device_dispatch
.entrypoints
[i
] = NULL
;
758 instance
->device_dispatch
.entrypoints
[i
] =
759 anv_device_dispatch_table
.entrypoints
[i
];
763 instance
->physicalDeviceCount
= -1;
765 result
= vk_debug_report_instance_init(&instance
->debug_report_callbacks
);
766 if (result
!= VK_SUCCESS
) {
767 vk_free2(&default_alloc
, pAllocator
, instance
);
768 return vk_error(result
);
771 instance
->pipeline_cache_enabled
=
772 env_var_as_boolean("ANV_ENABLE_PIPELINE_CACHE", true);
775 glsl_type_singleton_init_or_ref();
777 VG(VALGRIND_CREATE_MEMPOOL(instance
, 0, false));
779 *pInstance
= anv_instance_to_handle(instance
);
784 void anv_DestroyInstance(
785 VkInstance _instance
,
786 const VkAllocationCallbacks
* pAllocator
)
788 ANV_FROM_HANDLE(anv_instance
, instance
, _instance
);
793 if (instance
->physicalDeviceCount
> 0) {
794 /* We support at most one physical device. */
795 assert(instance
->physicalDeviceCount
== 1);
796 anv_physical_device_finish(&instance
->physicalDevice
);
799 vk_free(&instance
->alloc
, (char *)instance
->app_info
.app_name
);
800 vk_free(&instance
->alloc
, (char *)instance
->app_info
.engine_name
);
802 VG(VALGRIND_DESTROY_MEMPOOL(instance
));
804 vk_debug_report_instance_destroy(&instance
->debug_report_callbacks
);
806 glsl_type_singleton_decref();
809 vk_free(&instance
->alloc
, instance
);
813 anv_enumerate_devices(struct anv_instance
*instance
)
815 /* TODO: Check for more devices ? */
816 drmDevicePtr devices
[8];
817 VkResult result
= VK_ERROR_INCOMPATIBLE_DRIVER
;
820 instance
->physicalDeviceCount
= 0;
822 max_devices
= drmGetDevices2(0, devices
, ARRAY_SIZE(devices
));
824 return VK_ERROR_INCOMPATIBLE_DRIVER
;
826 for (unsigned i
= 0; i
< (unsigned)max_devices
; i
++) {
827 if (devices
[i
]->available_nodes
& 1 << DRM_NODE_RENDER
&&
828 devices
[i
]->bustype
== DRM_BUS_PCI
&&
829 devices
[i
]->deviceinfo
.pci
->vendor_id
== 0x8086) {
831 result
= anv_physical_device_init(&instance
->physicalDevice
,
832 instance
, devices
[i
]);
833 if (result
!= VK_ERROR_INCOMPATIBLE_DRIVER
)
837 drmFreeDevices(devices
, max_devices
);
839 if (result
== VK_SUCCESS
)
840 instance
->physicalDeviceCount
= 1;
846 anv_instance_ensure_physical_device(struct anv_instance
*instance
)
848 if (instance
->physicalDeviceCount
< 0) {
849 VkResult result
= anv_enumerate_devices(instance
);
850 if (result
!= VK_SUCCESS
&&
851 result
!= VK_ERROR_INCOMPATIBLE_DRIVER
)
858 VkResult
anv_EnumeratePhysicalDevices(
859 VkInstance _instance
,
860 uint32_t* pPhysicalDeviceCount
,
861 VkPhysicalDevice
* pPhysicalDevices
)
863 ANV_FROM_HANDLE(anv_instance
, instance
, _instance
);
864 VK_OUTARRAY_MAKE(out
, pPhysicalDevices
, pPhysicalDeviceCount
);
866 VkResult result
= anv_instance_ensure_physical_device(instance
);
867 if (result
!= VK_SUCCESS
)
870 if (instance
->physicalDeviceCount
== 0)
873 assert(instance
->physicalDeviceCount
== 1);
874 vk_outarray_append(&out
, i
) {
875 *i
= anv_physical_device_to_handle(&instance
->physicalDevice
);
878 return vk_outarray_status(&out
);
881 VkResult
anv_EnumeratePhysicalDeviceGroups(
882 VkInstance _instance
,
883 uint32_t* pPhysicalDeviceGroupCount
,
884 VkPhysicalDeviceGroupProperties
* pPhysicalDeviceGroupProperties
)
886 ANV_FROM_HANDLE(anv_instance
, instance
, _instance
);
887 VK_OUTARRAY_MAKE(out
, pPhysicalDeviceGroupProperties
,
888 pPhysicalDeviceGroupCount
);
890 VkResult result
= anv_instance_ensure_physical_device(instance
);
891 if (result
!= VK_SUCCESS
)
894 if (instance
->physicalDeviceCount
== 0)
897 assert(instance
->physicalDeviceCount
== 1);
899 vk_outarray_append(&out
, p
) {
900 p
->physicalDeviceCount
= 1;
901 memset(p
->physicalDevices
, 0, sizeof(p
->physicalDevices
));
902 p
->physicalDevices
[0] =
903 anv_physical_device_to_handle(&instance
->physicalDevice
);
904 p
->subsetAllocation
= false;
906 vk_foreach_struct(ext
, p
->pNext
)
907 anv_debug_ignored_stype(ext
->sType
);
910 return vk_outarray_status(&out
);
913 void anv_GetPhysicalDeviceFeatures(
914 VkPhysicalDevice physicalDevice
,
915 VkPhysicalDeviceFeatures
* pFeatures
)
917 ANV_FROM_HANDLE(anv_physical_device
, pdevice
, physicalDevice
);
919 *pFeatures
= (VkPhysicalDeviceFeatures
) {
920 .robustBufferAccess
= true,
921 .fullDrawIndexUint32
= true,
922 .imageCubeArray
= true,
923 .independentBlend
= true,
924 .geometryShader
= true,
925 .tessellationShader
= true,
926 .sampleRateShading
= true,
927 .dualSrcBlend
= true,
929 .multiDrawIndirect
= true,
930 .drawIndirectFirstInstance
= true,
932 .depthBiasClamp
= true,
933 .fillModeNonSolid
= true,
934 .depthBounds
= false,
938 .multiViewport
= true,
939 .samplerAnisotropy
= true,
940 .textureCompressionETC2
= pdevice
->info
.gen
>= 8 ||
941 pdevice
->info
.is_baytrail
,
942 .textureCompressionASTC_LDR
= pdevice
->info
.gen
>= 9, /* FINISHME CHV */
943 .textureCompressionBC
= true,
944 .occlusionQueryPrecise
= true,
945 .pipelineStatisticsQuery
= true,
946 .fragmentStoresAndAtomics
= true,
947 .shaderTessellationAndGeometryPointSize
= true,
948 .shaderImageGatherExtended
= true,
949 .shaderStorageImageExtendedFormats
= true,
950 .shaderStorageImageMultisample
= false,
951 .shaderStorageImageReadWithoutFormat
= false,
952 .shaderStorageImageWriteWithoutFormat
= true,
953 .shaderUniformBufferArrayDynamicIndexing
= true,
954 .shaderSampledImageArrayDynamicIndexing
= true,
955 .shaderStorageBufferArrayDynamicIndexing
= true,
956 .shaderStorageImageArrayDynamicIndexing
= true,
957 .shaderClipDistance
= true,
958 .shaderCullDistance
= true,
959 .shaderFloat64
= pdevice
->info
.gen
>= 8 &&
960 pdevice
->info
.has_64bit_types
,
961 .shaderInt64
= pdevice
->info
.gen
>= 8 &&
962 pdevice
->info
.has_64bit_types
,
963 .shaderInt16
= pdevice
->info
.gen
>= 8,
964 .shaderResourceMinLod
= pdevice
->info
.gen
>= 9,
965 .variableMultisampleRate
= true,
966 .inheritedQueries
= true,
969 /* We can't do image stores in vec4 shaders */
970 pFeatures
->vertexPipelineStoresAndAtomics
=
971 pdevice
->compiler
->scalar_stage
[MESA_SHADER_VERTEX
] &&
972 pdevice
->compiler
->scalar_stage
[MESA_SHADER_GEOMETRY
];
974 struct anv_app_info
*app_info
= &pdevice
->instance
->app_info
;
976 /* The new DOOM and Wolfenstein games require depthBounds without
977 * checking for it. They seem to run fine without it so just claim it's
978 * there and accept the consequences.
980 if (app_info
->engine_name
&& strcmp(app_info
->engine_name
, "idTech") == 0)
981 pFeatures
->depthBounds
= true;
984 void anv_GetPhysicalDeviceFeatures2(
985 VkPhysicalDevice physicalDevice
,
986 VkPhysicalDeviceFeatures2
* pFeatures
)
988 ANV_FROM_HANDLE(anv_physical_device
, pdevice
, physicalDevice
);
989 anv_GetPhysicalDeviceFeatures(physicalDevice
, &pFeatures
->features
);
991 vk_foreach_struct(ext
, pFeatures
->pNext
) {
992 switch (ext
->sType
) {
993 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_8BIT_STORAGE_FEATURES_KHR
: {
994 VkPhysicalDevice8BitStorageFeaturesKHR
*features
=
995 (VkPhysicalDevice8BitStorageFeaturesKHR
*)ext
;
996 features
->storageBuffer8BitAccess
= pdevice
->info
.gen
>= 8;
997 features
->uniformAndStorageBuffer8BitAccess
= pdevice
->info
.gen
>= 8;
998 features
->storagePushConstant8
= pdevice
->info
.gen
>= 8;
1002 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_16BIT_STORAGE_FEATURES
: {
1003 VkPhysicalDevice16BitStorageFeatures
*features
=
1004 (VkPhysicalDevice16BitStorageFeatures
*)ext
;
1005 features
->storageBuffer16BitAccess
= pdevice
->info
.gen
>= 8;
1006 features
->uniformAndStorageBuffer16BitAccess
= pdevice
->info
.gen
>= 8;
1007 features
->storagePushConstant16
= pdevice
->info
.gen
>= 8;
1008 features
->storageInputOutput16
= false;
1012 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_BUFFER_DEVICE_ADDRESS_FEATURES_EXT
: {
1013 VkPhysicalDeviceBufferDeviceAddressFeaturesEXT
*features
= (void *)ext
;
1014 features
->bufferDeviceAddress
= pdevice
->has_a64_buffer_access
;
1015 features
->bufferDeviceAddressCaptureReplay
= false;
1016 features
->bufferDeviceAddressMultiDevice
= false;
1020 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_COMPUTE_SHADER_DERIVATIVES_FEATURES_NV
: {
1021 VkPhysicalDeviceComputeShaderDerivativesFeaturesNV
*features
=
1022 (VkPhysicalDeviceComputeShaderDerivativesFeaturesNV
*)ext
;
1023 features
->computeDerivativeGroupQuads
= true;
1024 features
->computeDerivativeGroupLinear
= true;
1028 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_CONDITIONAL_RENDERING_FEATURES_EXT
: {
1029 VkPhysicalDeviceConditionalRenderingFeaturesEXT
*features
=
1030 (VkPhysicalDeviceConditionalRenderingFeaturesEXT
*)ext
;
1031 features
->conditionalRendering
= pdevice
->info
.gen
>= 8 ||
1032 pdevice
->info
.is_haswell
;
1033 features
->inheritedConditionalRendering
= pdevice
->info
.gen
>= 8 ||
1034 pdevice
->info
.is_haswell
;
1038 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DEPTH_CLIP_ENABLE_FEATURES_EXT
: {
1039 VkPhysicalDeviceDepthClipEnableFeaturesEXT
*features
=
1040 (VkPhysicalDeviceDepthClipEnableFeaturesEXT
*)ext
;
1041 features
->depthClipEnable
= true;
1045 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_FLOAT16_INT8_FEATURES_KHR
: {
1046 VkPhysicalDeviceFloat16Int8FeaturesKHR
*features
= (void *)ext
;
1047 features
->shaderFloat16
= pdevice
->info
.gen
>= 8;
1048 features
->shaderInt8
= pdevice
->info
.gen
>= 8;
1052 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_FRAGMENT_SHADER_INTERLOCK_FEATURES_EXT
: {
1053 VkPhysicalDeviceFragmentShaderInterlockFeaturesEXT
*features
=
1054 (VkPhysicalDeviceFragmentShaderInterlockFeaturesEXT
*)ext
;
1055 features
->fragmentShaderSampleInterlock
= pdevice
->info
.gen
>= 9;
1056 features
->fragmentShaderPixelInterlock
= pdevice
->info
.gen
>= 9;
1057 features
->fragmentShaderShadingRateInterlock
= false;
1061 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_HOST_QUERY_RESET_FEATURES_EXT
: {
1062 VkPhysicalDeviceHostQueryResetFeaturesEXT
*features
=
1063 (VkPhysicalDeviceHostQueryResetFeaturesEXT
*)ext
;
1064 features
->hostQueryReset
= true;
1068 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DESCRIPTOR_INDEXING_FEATURES_EXT
: {
1069 VkPhysicalDeviceDescriptorIndexingFeaturesEXT
*features
=
1070 (VkPhysicalDeviceDescriptorIndexingFeaturesEXT
*)ext
;
1071 features
->shaderInputAttachmentArrayDynamicIndexing
= false;
1072 features
->shaderUniformTexelBufferArrayDynamicIndexing
= true;
1073 features
->shaderStorageTexelBufferArrayDynamicIndexing
= true;
1074 features
->shaderUniformBufferArrayNonUniformIndexing
= false;
1075 features
->shaderSampledImageArrayNonUniformIndexing
= true;
1076 features
->shaderStorageBufferArrayNonUniformIndexing
= true;
1077 features
->shaderStorageImageArrayNonUniformIndexing
= true;
1078 features
->shaderInputAttachmentArrayNonUniformIndexing
= false;
1079 features
->shaderUniformTexelBufferArrayNonUniformIndexing
= true;
1080 features
->shaderStorageTexelBufferArrayNonUniformIndexing
= true;
1081 features
->descriptorBindingUniformBufferUpdateAfterBind
= false;
1082 features
->descriptorBindingSampledImageUpdateAfterBind
= true;
1083 features
->descriptorBindingStorageImageUpdateAfterBind
= true;
1084 features
->descriptorBindingStorageBufferUpdateAfterBind
= true;
1085 features
->descriptorBindingUniformTexelBufferUpdateAfterBind
= true;
1086 features
->descriptorBindingStorageTexelBufferUpdateAfterBind
= true;
1087 features
->descriptorBindingUpdateUnusedWhilePending
= true;
1088 features
->descriptorBindingPartiallyBound
= true;
1089 features
->descriptorBindingVariableDescriptorCount
= false;
1090 features
->runtimeDescriptorArray
= true;
1094 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_INLINE_UNIFORM_BLOCK_FEATURES_EXT
: {
1095 VkPhysicalDeviceInlineUniformBlockFeaturesEXT
*features
=
1096 (VkPhysicalDeviceInlineUniformBlockFeaturesEXT
*)ext
;
1097 features
->inlineUniformBlock
= true;
1098 features
->descriptorBindingInlineUniformBlockUpdateAfterBind
= true;
1102 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MULTIVIEW_FEATURES
: {
1103 VkPhysicalDeviceMultiviewFeatures
*features
=
1104 (VkPhysicalDeviceMultiviewFeatures
*)ext
;
1105 features
->multiview
= true;
1106 features
->multiviewGeometryShader
= true;
1107 features
->multiviewTessellationShader
= true;
1111 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_IMAGELESS_FRAMEBUFFER_FEATURES_KHR
: {
1112 VkPhysicalDeviceImagelessFramebufferFeaturesKHR
*features
=
1113 (VkPhysicalDeviceImagelessFramebufferFeaturesKHR
*)ext
;
1114 features
->imagelessFramebuffer
= true;
1118 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROTECTED_MEMORY_FEATURES
: {
1119 VkPhysicalDeviceProtectedMemoryFeatures
*features
= (void *)ext
;
1120 features
->protectedMemory
= false;
1124 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SAMPLER_YCBCR_CONVERSION_FEATURES
: {
1125 VkPhysicalDeviceSamplerYcbcrConversionFeatures
*features
=
1126 (VkPhysicalDeviceSamplerYcbcrConversionFeatures
*) ext
;
1127 features
->samplerYcbcrConversion
= true;
1131 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SCALAR_BLOCK_LAYOUT_FEATURES_EXT
: {
1132 VkPhysicalDeviceScalarBlockLayoutFeaturesEXT
*features
=
1133 (VkPhysicalDeviceScalarBlockLayoutFeaturesEXT
*)ext
;
1134 features
->scalarBlockLayout
= true;
1138 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_ATOMIC_INT64_FEATURES_KHR
: {
1139 VkPhysicalDeviceShaderAtomicInt64FeaturesKHR
*features
= (void *)ext
;
1140 features
->shaderBufferInt64Atomics
=
1141 pdevice
->info
.gen
>= 9 && pdevice
->use_softpin
;
1142 features
->shaderSharedInt64Atomics
= VK_FALSE
;
1146 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_DEMOTE_TO_HELPER_INVOCATION_FEATURES_EXT
: {
1147 VkPhysicalDeviceShaderDemoteToHelperInvocationFeaturesEXT
*features
= (void *)ext
;
1148 features
->shaderDemoteToHelperInvocation
= true;
1152 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_DRAW_PARAMETERS_FEATURES
: {
1153 VkPhysicalDeviceShaderDrawParametersFeatures
*features
= (void *)ext
;
1154 features
->shaderDrawParameters
= true;
1158 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TEXEL_BUFFER_ALIGNMENT_FEATURES_EXT
: {
1159 VkPhysicalDeviceTexelBufferAlignmentFeaturesEXT
*features
=
1160 (VkPhysicalDeviceTexelBufferAlignmentFeaturesEXT
*)ext
;
1161 features
->texelBufferAlignment
= true;
1165 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VARIABLE_POINTERS_FEATURES
: {
1166 VkPhysicalDeviceVariablePointersFeatures
*features
= (void *)ext
;
1167 features
->variablePointersStorageBuffer
= true;
1168 features
->variablePointers
= true;
1172 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TRANSFORM_FEEDBACK_FEATURES_EXT
: {
1173 VkPhysicalDeviceTransformFeedbackFeaturesEXT
*features
=
1174 (VkPhysicalDeviceTransformFeedbackFeaturesEXT
*)ext
;
1175 features
->transformFeedback
= true;
1176 features
->geometryStreams
= true;
1180 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_UNIFORM_BUFFER_STANDARD_LAYOUT_FEATURES_KHR
: {
1181 VkPhysicalDeviceUniformBufferStandardLayoutFeaturesKHR
*features
=
1182 (VkPhysicalDeviceUniformBufferStandardLayoutFeaturesKHR
*)ext
;
1183 features
->uniformBufferStandardLayout
= true;
1187 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VERTEX_ATTRIBUTE_DIVISOR_FEATURES_EXT
: {
1188 VkPhysicalDeviceVertexAttributeDivisorFeaturesEXT
*features
=
1189 (VkPhysicalDeviceVertexAttributeDivisorFeaturesEXT
*)ext
;
1190 features
->vertexAttributeInstanceRateDivisor
= true;
1191 features
->vertexAttributeInstanceRateZeroDivisor
= true;
1195 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_YCBCR_IMAGE_ARRAYS_FEATURES_EXT
: {
1196 VkPhysicalDeviceYcbcrImageArraysFeaturesEXT
*features
=
1197 (VkPhysicalDeviceYcbcrImageArraysFeaturesEXT
*)ext
;
1198 features
->ycbcrImageArrays
= true;
1203 anv_debug_ignored_stype(ext
->sType
);
1209 #define MAX_PER_STAGE_DESCRIPTOR_UNIFORM_BUFFERS 64
1211 #define MAX_PER_STAGE_DESCRIPTOR_INPUT_ATTACHMENTS 64
1212 #define MAX_DESCRIPTOR_SET_INPUT_ATTACHMENTS 256
1214 void anv_GetPhysicalDeviceProperties(
1215 VkPhysicalDevice physicalDevice
,
1216 VkPhysicalDeviceProperties
* pProperties
)
1218 ANV_FROM_HANDLE(anv_physical_device
, pdevice
, physicalDevice
);
1219 const struct gen_device_info
*devinfo
= &pdevice
->info
;
1221 /* See assertions made when programming the buffer surface state. */
1222 const uint32_t max_raw_buffer_sz
= devinfo
->gen
>= 7 ?
1223 (1ul << 30) : (1ul << 27);
1225 const uint32_t max_ssbos
= pdevice
->has_a64_buffer_access
? UINT16_MAX
: 64;
1226 const uint32_t max_textures
=
1227 pdevice
->has_bindless_images
? UINT16_MAX
: 128;
1228 const uint32_t max_samplers
=
1229 pdevice
->has_bindless_samplers
? UINT16_MAX
:
1230 (devinfo
->gen
>= 8 || devinfo
->is_haswell
) ? 128 : 16;
1231 const uint32_t max_images
=
1232 pdevice
->has_bindless_images
? UINT16_MAX
: MAX_IMAGES
;
1234 /* The moment we have anything bindless, claim a high per-stage limit */
1235 const uint32_t max_per_stage
=
1236 pdevice
->has_a64_buffer_access
? UINT32_MAX
:
1237 MAX_BINDING_TABLE_SIZE
- MAX_RTS
;
1239 VkSampleCountFlags sample_counts
=
1240 isl_device_get_sample_counts(&pdevice
->isl_dev
);
1243 VkPhysicalDeviceLimits limits
= {
1244 .maxImageDimension1D
= (1 << 14),
1245 .maxImageDimension2D
= (1 << 14),
1246 .maxImageDimension3D
= (1 << 11),
1247 .maxImageDimensionCube
= (1 << 14),
1248 .maxImageArrayLayers
= (1 << 11),
1249 .maxTexelBufferElements
= 128 * 1024 * 1024,
1250 .maxUniformBufferRange
= (1ul << 27),
1251 .maxStorageBufferRange
= max_raw_buffer_sz
,
1252 .maxPushConstantsSize
= MAX_PUSH_CONSTANTS_SIZE
,
1253 .maxMemoryAllocationCount
= UINT32_MAX
,
1254 .maxSamplerAllocationCount
= 64 * 1024,
1255 .bufferImageGranularity
= 64, /* A cache line */
1256 .sparseAddressSpaceSize
= 0,
1257 .maxBoundDescriptorSets
= MAX_SETS
,
1258 .maxPerStageDescriptorSamplers
= max_samplers
,
1259 .maxPerStageDescriptorUniformBuffers
= MAX_PER_STAGE_DESCRIPTOR_UNIFORM_BUFFERS
,
1260 .maxPerStageDescriptorStorageBuffers
= max_ssbos
,
1261 .maxPerStageDescriptorSampledImages
= max_textures
,
1262 .maxPerStageDescriptorStorageImages
= max_images
,
1263 .maxPerStageDescriptorInputAttachments
= MAX_PER_STAGE_DESCRIPTOR_INPUT_ATTACHMENTS
,
1264 .maxPerStageResources
= max_per_stage
,
1265 .maxDescriptorSetSamplers
= 6 * max_samplers
, /* number of stages * maxPerStageDescriptorSamplers */
1266 .maxDescriptorSetUniformBuffers
= 6 * MAX_PER_STAGE_DESCRIPTOR_UNIFORM_BUFFERS
, /* number of stages * maxPerStageDescriptorUniformBuffers */
1267 .maxDescriptorSetUniformBuffersDynamic
= MAX_DYNAMIC_BUFFERS
/ 2,
1268 .maxDescriptorSetStorageBuffers
= 6 * max_ssbos
, /* number of stages * maxPerStageDescriptorStorageBuffers */
1269 .maxDescriptorSetStorageBuffersDynamic
= MAX_DYNAMIC_BUFFERS
/ 2,
1270 .maxDescriptorSetSampledImages
= 6 * max_textures
, /* number of stages * maxPerStageDescriptorSampledImages */
1271 .maxDescriptorSetStorageImages
= 6 * max_images
, /* number of stages * maxPerStageDescriptorStorageImages */
1272 .maxDescriptorSetInputAttachments
= MAX_DESCRIPTOR_SET_INPUT_ATTACHMENTS
,
1273 .maxVertexInputAttributes
= MAX_VBS
,
1274 .maxVertexInputBindings
= MAX_VBS
,
1275 .maxVertexInputAttributeOffset
= 2047,
1276 .maxVertexInputBindingStride
= 2048,
1277 .maxVertexOutputComponents
= 128,
1278 .maxTessellationGenerationLevel
= 64,
1279 .maxTessellationPatchSize
= 32,
1280 .maxTessellationControlPerVertexInputComponents
= 128,
1281 .maxTessellationControlPerVertexOutputComponents
= 128,
1282 .maxTessellationControlPerPatchOutputComponents
= 128,
1283 .maxTessellationControlTotalOutputComponents
= 2048,
1284 .maxTessellationEvaluationInputComponents
= 128,
1285 .maxTessellationEvaluationOutputComponents
= 128,
1286 .maxGeometryShaderInvocations
= 32,
1287 .maxGeometryInputComponents
= 64,
1288 .maxGeometryOutputComponents
= 128,
1289 .maxGeometryOutputVertices
= 256,
1290 .maxGeometryTotalOutputComponents
= 1024,
1291 .maxFragmentInputComponents
= 116, /* 128 components - (PSIZ, CLIP_DIST0, CLIP_DIST1) */
1292 .maxFragmentOutputAttachments
= 8,
1293 .maxFragmentDualSrcAttachments
= 1,
1294 .maxFragmentCombinedOutputResources
= 8,
1295 .maxComputeSharedMemorySize
= 64 * 1024,
1296 .maxComputeWorkGroupCount
= { 65535, 65535, 65535 },
1297 .maxComputeWorkGroupInvocations
= 32 * devinfo
->max_cs_threads
,
1298 .maxComputeWorkGroupSize
= {
1299 16 * devinfo
->max_cs_threads
,
1300 16 * devinfo
->max_cs_threads
,
1301 16 * devinfo
->max_cs_threads
,
1303 .subPixelPrecisionBits
= 8,
1304 .subTexelPrecisionBits
= 8,
1305 .mipmapPrecisionBits
= 8,
1306 .maxDrawIndexedIndexValue
= UINT32_MAX
,
1307 .maxDrawIndirectCount
= UINT32_MAX
,
1308 .maxSamplerLodBias
= 16,
1309 .maxSamplerAnisotropy
= 16,
1310 .maxViewports
= MAX_VIEWPORTS
,
1311 .maxViewportDimensions
= { (1 << 14), (1 << 14) },
1312 .viewportBoundsRange
= { INT16_MIN
, INT16_MAX
},
1313 .viewportSubPixelBits
= 13, /* We take a float? */
1314 .minMemoryMapAlignment
= 4096, /* A page */
1315 /* The dataport requires texel alignment so we need to assume a worst
1316 * case of R32G32B32A32 which is 16 bytes.
1318 .minTexelBufferOffsetAlignment
= 16,
1319 /* We need 16 for UBO block reads to work and 32 for push UBOs */
1320 .minUniformBufferOffsetAlignment
= 32,
1321 .minStorageBufferOffsetAlignment
= 4,
1322 .minTexelOffset
= -8,
1323 .maxTexelOffset
= 7,
1324 .minTexelGatherOffset
= -32,
1325 .maxTexelGatherOffset
= 31,
1326 .minInterpolationOffset
= -0.5,
1327 .maxInterpolationOffset
= 0.4375,
1328 .subPixelInterpolationOffsetBits
= 4,
1329 .maxFramebufferWidth
= (1 << 14),
1330 .maxFramebufferHeight
= (1 << 14),
1331 .maxFramebufferLayers
= (1 << 11),
1332 .framebufferColorSampleCounts
= sample_counts
,
1333 .framebufferDepthSampleCounts
= sample_counts
,
1334 .framebufferStencilSampleCounts
= sample_counts
,
1335 .framebufferNoAttachmentsSampleCounts
= sample_counts
,
1336 .maxColorAttachments
= MAX_RTS
,
1337 .sampledImageColorSampleCounts
= sample_counts
,
1338 .sampledImageIntegerSampleCounts
= VK_SAMPLE_COUNT_1_BIT
,
1339 .sampledImageDepthSampleCounts
= sample_counts
,
1340 .sampledImageStencilSampleCounts
= sample_counts
,
1341 .storageImageSampleCounts
= VK_SAMPLE_COUNT_1_BIT
,
1342 .maxSampleMaskWords
= 1,
1343 .timestampComputeAndGraphics
= true,
1344 .timestampPeriod
= 1000000000.0 / devinfo
->timestamp_frequency
,
1345 .maxClipDistances
= 8,
1346 .maxCullDistances
= 8,
1347 .maxCombinedClipAndCullDistances
= 8,
1348 .discreteQueuePriorities
= 2,
1349 .pointSizeRange
= { 0.125, 255.875 },
1350 .lineWidthRange
= { 0.0, 7.9921875 },
1351 .pointSizeGranularity
= (1.0 / 8.0),
1352 .lineWidthGranularity
= (1.0 / 128.0),
1353 .strictLines
= false, /* FINISHME */
1354 .standardSampleLocations
= true,
1355 .optimalBufferCopyOffsetAlignment
= 128,
1356 .optimalBufferCopyRowPitchAlignment
= 128,
1357 .nonCoherentAtomSize
= 64,
1360 *pProperties
= (VkPhysicalDeviceProperties
) {
1361 .apiVersion
= anv_physical_device_api_version(pdevice
),
1362 .driverVersion
= vk_get_driver_version(),
1364 .deviceID
= pdevice
->chipset_id
,
1365 .deviceType
= VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU
,
1367 .sparseProperties
= {0}, /* Broadwell doesn't do sparse. */
1370 snprintf(pProperties
->deviceName
, sizeof(pProperties
->deviceName
),
1371 "%s", pdevice
->name
);
1372 memcpy(pProperties
->pipelineCacheUUID
,
1373 pdevice
->pipeline_cache_uuid
, VK_UUID_SIZE
);
1376 void anv_GetPhysicalDeviceProperties2(
1377 VkPhysicalDevice physicalDevice
,
1378 VkPhysicalDeviceProperties2
* pProperties
)
1380 ANV_FROM_HANDLE(anv_physical_device
, pdevice
, physicalDevice
);
1382 anv_GetPhysicalDeviceProperties(physicalDevice
, &pProperties
->properties
);
1384 vk_foreach_struct(ext
, pProperties
->pNext
) {
1385 switch (ext
->sType
) {
1386 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DEPTH_STENCIL_RESOLVE_PROPERTIES_KHR
: {
1387 VkPhysicalDeviceDepthStencilResolvePropertiesKHR
*props
=
1388 (VkPhysicalDeviceDepthStencilResolvePropertiesKHR
*)ext
;
1390 /* We support all of the depth resolve modes */
1391 props
->supportedDepthResolveModes
=
1392 VK_RESOLVE_MODE_SAMPLE_ZERO_BIT_KHR
|
1393 VK_RESOLVE_MODE_AVERAGE_BIT_KHR
|
1394 VK_RESOLVE_MODE_MIN_BIT_KHR
|
1395 VK_RESOLVE_MODE_MAX_BIT_KHR
;
1397 /* Average doesn't make sense for stencil so we don't support that */
1398 props
->supportedStencilResolveModes
=
1399 VK_RESOLVE_MODE_SAMPLE_ZERO_BIT_KHR
;
1400 if (pdevice
->info
.gen
>= 8) {
1401 /* The advanced stencil resolve modes currently require stencil
1402 * sampling be supported by the hardware.
1404 props
->supportedStencilResolveModes
|=
1405 VK_RESOLVE_MODE_MIN_BIT_KHR
|
1406 VK_RESOLVE_MODE_MAX_BIT_KHR
;
1409 props
->independentResolveNone
= true;
1410 props
->independentResolve
= true;
1414 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DESCRIPTOR_INDEXING_PROPERTIES_EXT
: {
1415 VkPhysicalDeviceDescriptorIndexingPropertiesEXT
*props
=
1416 (VkPhysicalDeviceDescriptorIndexingPropertiesEXT
*)ext
;
1418 /* It's a bit hard to exactly map our implementation to the limits
1419 * described here. The bindless surface handle in the extended
1420 * message descriptors is 20 bits and it's an index into the table of
1421 * RENDER_SURFACE_STATE structs that starts at bindless surface base
1422 * address. Given that most things consume two surface states per
1423 * view (general/sampled for textures and write-only/read-write for
1424 * images), we claim 2^19 things.
1426 * For SSBOs, we just use A64 messages so there is no real limit
1427 * there beyond the limit on the total size of a descriptor set.
1429 const unsigned max_bindless_views
= 1 << 19;
1431 props
->maxUpdateAfterBindDescriptorsInAllPools
= max_bindless_views
;
1432 props
->shaderUniformBufferArrayNonUniformIndexingNative
= false;
1433 props
->shaderSampledImageArrayNonUniformIndexingNative
= false;
1434 props
->shaderStorageBufferArrayNonUniformIndexingNative
= true;
1435 props
->shaderStorageImageArrayNonUniformIndexingNative
= false;
1436 props
->shaderInputAttachmentArrayNonUniformIndexingNative
= false;
1437 props
->robustBufferAccessUpdateAfterBind
= true;
1438 props
->quadDivergentImplicitLod
= false;
1439 props
->maxPerStageDescriptorUpdateAfterBindSamplers
= max_bindless_views
;
1440 props
->maxPerStageDescriptorUpdateAfterBindUniformBuffers
= MAX_PER_STAGE_DESCRIPTOR_UNIFORM_BUFFERS
;
1441 props
->maxPerStageDescriptorUpdateAfterBindStorageBuffers
= UINT32_MAX
;
1442 props
->maxPerStageDescriptorUpdateAfterBindSampledImages
= max_bindless_views
;
1443 props
->maxPerStageDescriptorUpdateAfterBindStorageImages
= max_bindless_views
;
1444 props
->maxPerStageDescriptorUpdateAfterBindInputAttachments
= MAX_PER_STAGE_DESCRIPTOR_INPUT_ATTACHMENTS
;
1445 props
->maxPerStageUpdateAfterBindResources
= UINT32_MAX
;
1446 props
->maxDescriptorSetUpdateAfterBindSamplers
= max_bindless_views
;
1447 props
->maxDescriptorSetUpdateAfterBindUniformBuffers
= 6 * MAX_PER_STAGE_DESCRIPTOR_UNIFORM_BUFFERS
;
1448 props
->maxDescriptorSetUpdateAfterBindUniformBuffersDynamic
= MAX_DYNAMIC_BUFFERS
/ 2;
1449 props
->maxDescriptorSetUpdateAfterBindStorageBuffers
= UINT32_MAX
;
1450 props
->maxDescriptorSetUpdateAfterBindStorageBuffersDynamic
= MAX_DYNAMIC_BUFFERS
/ 2;
1451 props
->maxDescriptorSetUpdateAfterBindSampledImages
= max_bindless_views
;
1452 props
->maxDescriptorSetUpdateAfterBindStorageImages
= max_bindless_views
;
1453 props
->maxDescriptorSetUpdateAfterBindInputAttachments
= MAX_DESCRIPTOR_SET_INPUT_ATTACHMENTS
;
1457 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DRIVER_PROPERTIES_KHR
: {
1458 VkPhysicalDeviceDriverPropertiesKHR
*driver_props
=
1459 (VkPhysicalDeviceDriverPropertiesKHR
*) ext
;
1461 driver_props
->driverID
= VK_DRIVER_ID_INTEL_OPEN_SOURCE_MESA_KHR
;
1462 snprintf(driver_props
->driverName
, VK_MAX_DRIVER_NAME_SIZE_KHR
,
1463 "Intel open-source Mesa driver");
1465 snprintf(driver_props
->driverInfo
, VK_MAX_DRIVER_INFO_SIZE_KHR
,
1466 "Mesa " PACKAGE_VERSION MESA_GIT_SHA1
);
1468 driver_props
->conformanceVersion
= (VkConformanceVersionKHR
) {
1477 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_EXTERNAL_MEMORY_HOST_PROPERTIES_EXT
: {
1478 VkPhysicalDeviceExternalMemoryHostPropertiesEXT
*props
=
1479 (VkPhysicalDeviceExternalMemoryHostPropertiesEXT
*) ext
;
1480 /* Userptr needs page aligned memory. */
1481 props
->minImportedHostPointerAlignment
= 4096;
1485 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_ID_PROPERTIES
: {
1486 VkPhysicalDeviceIDProperties
*id_props
=
1487 (VkPhysicalDeviceIDProperties
*)ext
;
1488 memcpy(id_props
->deviceUUID
, pdevice
->device_uuid
, VK_UUID_SIZE
);
1489 memcpy(id_props
->driverUUID
, pdevice
->driver_uuid
, VK_UUID_SIZE
);
1490 /* The LUID is for Windows. */
1491 id_props
->deviceLUIDValid
= false;
1495 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_INLINE_UNIFORM_BLOCK_PROPERTIES_EXT
: {
1496 VkPhysicalDeviceInlineUniformBlockPropertiesEXT
*props
=
1497 (VkPhysicalDeviceInlineUniformBlockPropertiesEXT
*)ext
;
1498 props
->maxInlineUniformBlockSize
= MAX_INLINE_UNIFORM_BLOCK_SIZE
;
1499 props
->maxPerStageDescriptorInlineUniformBlocks
=
1500 MAX_INLINE_UNIFORM_BLOCK_DESCRIPTORS
;
1501 props
->maxPerStageDescriptorUpdateAfterBindInlineUniformBlocks
=
1502 MAX_INLINE_UNIFORM_BLOCK_DESCRIPTORS
;
1503 props
->maxDescriptorSetInlineUniformBlocks
=
1504 MAX_INLINE_UNIFORM_BLOCK_DESCRIPTORS
;
1505 props
->maxDescriptorSetUpdateAfterBindInlineUniformBlocks
=
1506 MAX_INLINE_UNIFORM_BLOCK_DESCRIPTORS
;
1510 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MAINTENANCE_3_PROPERTIES
: {
1511 VkPhysicalDeviceMaintenance3Properties
*props
=
1512 (VkPhysicalDeviceMaintenance3Properties
*)ext
;
1513 /* This value doesn't matter for us today as our per-stage
1514 * descriptors are the real limit.
1516 props
->maxPerSetDescriptors
= 1024;
1517 props
->maxMemoryAllocationSize
= MAX_MEMORY_ALLOCATION_SIZE
;
1521 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MULTIVIEW_PROPERTIES
: {
1522 VkPhysicalDeviceMultiviewProperties
*properties
=
1523 (VkPhysicalDeviceMultiviewProperties
*)ext
;
1524 properties
->maxMultiviewViewCount
= 16;
1525 properties
->maxMultiviewInstanceIndex
= UINT32_MAX
/ 16;
1529 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PCI_BUS_INFO_PROPERTIES_EXT
: {
1530 VkPhysicalDevicePCIBusInfoPropertiesEXT
*properties
=
1531 (VkPhysicalDevicePCIBusInfoPropertiesEXT
*)ext
;
1532 properties
->pciDomain
= pdevice
->pci_info
.domain
;
1533 properties
->pciBus
= pdevice
->pci_info
.bus
;
1534 properties
->pciDevice
= pdevice
->pci_info
.device
;
1535 properties
->pciFunction
= pdevice
->pci_info
.function
;
1539 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_POINT_CLIPPING_PROPERTIES
: {
1540 VkPhysicalDevicePointClippingProperties
*properties
=
1541 (VkPhysicalDevicePointClippingProperties
*) ext
;
1542 properties
->pointClippingBehavior
= VK_POINT_CLIPPING_BEHAVIOR_USER_CLIP_PLANES_ONLY
;
1546 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROTECTED_MEMORY_PROPERTIES
: {
1547 VkPhysicalDeviceProtectedMemoryProperties
*props
=
1548 (VkPhysicalDeviceProtectedMemoryProperties
*)ext
;
1549 props
->protectedNoFault
= false;
1553 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PUSH_DESCRIPTOR_PROPERTIES_KHR
: {
1554 VkPhysicalDevicePushDescriptorPropertiesKHR
*properties
=
1555 (VkPhysicalDevicePushDescriptorPropertiesKHR
*) ext
;
1557 properties
->maxPushDescriptors
= MAX_PUSH_DESCRIPTORS
;
1561 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SAMPLER_FILTER_MINMAX_PROPERTIES_EXT
: {
1562 VkPhysicalDeviceSamplerFilterMinmaxPropertiesEXT
*properties
=
1563 (VkPhysicalDeviceSamplerFilterMinmaxPropertiesEXT
*)ext
;
1564 properties
->filterMinmaxImageComponentMapping
= pdevice
->info
.gen
>= 9;
1565 properties
->filterMinmaxSingleComponentFormats
= true;
1569 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SUBGROUP_PROPERTIES
: {
1570 VkPhysicalDeviceSubgroupProperties
*properties
= (void *)ext
;
1572 properties
->subgroupSize
= BRW_SUBGROUP_SIZE
;
1574 VkShaderStageFlags scalar_stages
= 0;
1575 for (unsigned stage
= 0; stage
< MESA_SHADER_STAGES
; stage
++) {
1576 if (pdevice
->compiler
->scalar_stage
[stage
])
1577 scalar_stages
|= mesa_to_vk_shader_stage(stage
);
1579 properties
->supportedStages
= scalar_stages
;
1581 properties
->supportedOperations
= VK_SUBGROUP_FEATURE_BASIC_BIT
|
1582 VK_SUBGROUP_FEATURE_VOTE_BIT
|
1583 VK_SUBGROUP_FEATURE_BALLOT_BIT
|
1584 VK_SUBGROUP_FEATURE_SHUFFLE_BIT
|
1585 VK_SUBGROUP_FEATURE_SHUFFLE_RELATIVE_BIT
|
1586 VK_SUBGROUP_FEATURE_QUAD_BIT
;
1587 if (pdevice
->info
.gen
>= 8) {
1588 /* TODO: There's no technical reason why these can't be made to
1589 * work on gen7 but they don't at the moment so it's best to leave
1590 * the feature disabled than enabled and broken.
1592 properties
->supportedOperations
|=
1593 VK_SUBGROUP_FEATURE_ARITHMETIC_BIT
|
1594 VK_SUBGROUP_FEATURE_CLUSTERED_BIT
;
1596 properties
->quadOperationsInAllStages
= pdevice
->info
.gen
>= 8;
1600 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SUBGROUP_SIZE_CONTROL_PROPERTIES_EXT
: {
1601 VkPhysicalDeviceSubgroupSizeControlPropertiesEXT
*props
=
1602 (VkPhysicalDeviceSubgroupSizeControlPropertiesEXT
*)ext
;
1603 STATIC_ASSERT(8 <= BRW_SUBGROUP_SIZE
&& BRW_SUBGROUP_SIZE
<= 32);
1604 props
->minSubgroupSize
= 8;
1605 props
->maxSubgroupSize
= 32;
1606 props
->maxComputeWorkgroupSubgroups
= pdevice
->info
.max_cs_threads
;
1607 props
->requiredSubgroupSizeStages
= VK_SHADER_STAGE_COMPUTE_BIT
;
1611 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TEXEL_BUFFER_ALIGNMENT_PROPERTIES_EXT
: {
1612 VkPhysicalDeviceTexelBufferAlignmentPropertiesEXT
*props
=
1613 (VkPhysicalDeviceTexelBufferAlignmentPropertiesEXT
*)ext
;
1615 /* From the SKL PRM Vol. 2d, docs for RENDER_SURFACE_STATE::Surface
1618 * "For SURFTYPE_BUFFER non-rendertarget surfaces, this field
1619 * specifies the base address of the first element of the surface,
1620 * computed in software by adding the surface base address to the
1621 * byte offset of the element in the buffer. The base address must
1622 * be aligned to element size."
1624 * The typed dataport messages require that things be texel aligned.
1625 * Otherwise, we may just load/store the wrong data or, in the worst
1626 * case, there may be hangs.
1628 props
->storageTexelBufferOffsetAlignmentBytes
= 16;
1629 props
->storageTexelBufferOffsetSingleTexelAlignment
= true;
1631 /* The sampler, however, is much more forgiving and it can handle
1632 * arbitrary byte alignment for linear and buffer surfaces. It's
1633 * hard to find a good PRM citation for this but years of empirical
1634 * experience demonstrate that this is true.
1636 props
->uniformTexelBufferOffsetAlignmentBytes
= 1;
1637 props
->uniformTexelBufferOffsetSingleTexelAlignment
= false;
1641 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TRANSFORM_FEEDBACK_PROPERTIES_EXT
: {
1642 VkPhysicalDeviceTransformFeedbackPropertiesEXT
*props
=
1643 (VkPhysicalDeviceTransformFeedbackPropertiesEXT
*)ext
;
1645 props
->maxTransformFeedbackStreams
= MAX_XFB_STREAMS
;
1646 props
->maxTransformFeedbackBuffers
= MAX_XFB_BUFFERS
;
1647 props
->maxTransformFeedbackBufferSize
= (1ull << 32);
1648 props
->maxTransformFeedbackStreamDataSize
= 128 * 4;
1649 props
->maxTransformFeedbackBufferDataSize
= 128 * 4;
1650 props
->maxTransformFeedbackBufferDataStride
= 2048;
1651 props
->transformFeedbackQueries
= true;
1652 props
->transformFeedbackStreamsLinesTriangles
= false;
1653 props
->transformFeedbackRasterizationStreamSelect
= false;
1654 props
->transformFeedbackDraw
= true;
1658 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VERTEX_ATTRIBUTE_DIVISOR_PROPERTIES_EXT
: {
1659 VkPhysicalDeviceVertexAttributeDivisorPropertiesEXT
*props
=
1660 (VkPhysicalDeviceVertexAttributeDivisorPropertiesEXT
*)ext
;
1661 /* We have to restrict this a bit for multiview */
1662 props
->maxVertexAttribDivisor
= UINT32_MAX
/ 16;
1667 anv_debug_ignored_stype(ext
->sType
);
1673 /* We support exactly one queue family. */
1674 static const VkQueueFamilyProperties
1675 anv_queue_family_properties
= {
1676 .queueFlags
= VK_QUEUE_GRAPHICS_BIT
|
1677 VK_QUEUE_COMPUTE_BIT
|
1678 VK_QUEUE_TRANSFER_BIT
,
1680 .timestampValidBits
= 36, /* XXX: Real value here */
1681 .minImageTransferGranularity
= { 1, 1, 1 },
1684 void anv_GetPhysicalDeviceQueueFamilyProperties(
1685 VkPhysicalDevice physicalDevice
,
1687 VkQueueFamilyProperties
* pQueueFamilyProperties
)
1689 VK_OUTARRAY_MAKE(out
, pQueueFamilyProperties
, pCount
);
1691 vk_outarray_append(&out
, p
) {
1692 *p
= anv_queue_family_properties
;
1696 void anv_GetPhysicalDeviceQueueFamilyProperties2(
1697 VkPhysicalDevice physicalDevice
,
1698 uint32_t* pQueueFamilyPropertyCount
,
1699 VkQueueFamilyProperties2
* pQueueFamilyProperties
)
1702 VK_OUTARRAY_MAKE(out
, pQueueFamilyProperties
, pQueueFamilyPropertyCount
);
1704 vk_outarray_append(&out
, p
) {
1705 p
->queueFamilyProperties
= anv_queue_family_properties
;
1707 vk_foreach_struct(s
, p
->pNext
) {
1708 anv_debug_ignored_stype(s
->sType
);
1713 void anv_GetPhysicalDeviceMemoryProperties(
1714 VkPhysicalDevice physicalDevice
,
1715 VkPhysicalDeviceMemoryProperties
* pMemoryProperties
)
1717 ANV_FROM_HANDLE(anv_physical_device
, physical_device
, physicalDevice
);
1719 pMemoryProperties
->memoryTypeCount
= physical_device
->memory
.type_count
;
1720 for (uint32_t i
= 0; i
< physical_device
->memory
.type_count
; i
++) {
1721 pMemoryProperties
->memoryTypes
[i
] = (VkMemoryType
) {
1722 .propertyFlags
= physical_device
->memory
.types
[i
].propertyFlags
,
1723 .heapIndex
= physical_device
->memory
.types
[i
].heapIndex
,
1727 pMemoryProperties
->memoryHeapCount
= physical_device
->memory
.heap_count
;
1728 for (uint32_t i
= 0; i
< physical_device
->memory
.heap_count
; i
++) {
1729 pMemoryProperties
->memoryHeaps
[i
] = (VkMemoryHeap
) {
1730 .size
= physical_device
->memory
.heaps
[i
].size
,
1731 .flags
= physical_device
->memory
.heaps
[i
].flags
,
1737 anv_get_memory_budget(VkPhysicalDevice physicalDevice
,
1738 VkPhysicalDeviceMemoryBudgetPropertiesEXT
*memoryBudget
)
1740 ANV_FROM_HANDLE(anv_physical_device
, device
, physicalDevice
);
1741 uint64_t sys_available
= get_available_system_memory();
1742 assert(sys_available
> 0);
1744 VkDeviceSize total_heaps_size
= 0;
1745 for (size_t i
= 0; i
< device
->memory
.heap_count
; i
++)
1746 total_heaps_size
+= device
->memory
.heaps
[i
].size
;
1748 for (size_t i
= 0; i
< device
->memory
.heap_count
; i
++) {
1749 VkDeviceSize heap_size
= device
->memory
.heaps
[i
].size
;
1750 VkDeviceSize heap_used
= device
->memory
.heaps
[i
].used
;
1751 VkDeviceSize heap_budget
;
1753 double heap_proportion
= (double) heap_size
/ total_heaps_size
;
1754 VkDeviceSize sys_available_prop
= sys_available
* heap_proportion
;
1757 * Let's not incite the app to starve the system: report at most 90% of
1758 * available system memory.
1760 uint64_t heap_available
= sys_available_prop
* 9 / 10;
1761 heap_budget
= MIN2(heap_size
, heap_used
+ heap_available
);
1764 * Round down to the nearest MB
1766 heap_budget
&= ~((1ull << 20) - 1);
1769 * The heapBudget value must be non-zero for array elements less than
1770 * VkPhysicalDeviceMemoryProperties::memoryHeapCount. The heapBudget
1771 * value must be less than or equal to VkMemoryHeap::size for each heap.
1773 assert(0 < heap_budget
&& heap_budget
<= heap_size
);
1775 memoryBudget
->heapUsage
[i
] = heap_used
;
1776 memoryBudget
->heapBudget
[i
] = heap_budget
;
1779 /* The heapBudget and heapUsage values must be zero for array elements
1780 * greater than or equal to VkPhysicalDeviceMemoryProperties::memoryHeapCount
1782 for (uint32_t i
= device
->memory
.heap_count
; i
< VK_MAX_MEMORY_HEAPS
; i
++) {
1783 memoryBudget
->heapBudget
[i
] = 0;
1784 memoryBudget
->heapUsage
[i
] = 0;
1788 void anv_GetPhysicalDeviceMemoryProperties2(
1789 VkPhysicalDevice physicalDevice
,
1790 VkPhysicalDeviceMemoryProperties2
* pMemoryProperties
)
1792 anv_GetPhysicalDeviceMemoryProperties(physicalDevice
,
1793 &pMemoryProperties
->memoryProperties
);
1795 vk_foreach_struct(ext
, pMemoryProperties
->pNext
) {
1796 switch (ext
->sType
) {
1797 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MEMORY_BUDGET_PROPERTIES_EXT
:
1798 anv_get_memory_budget(physicalDevice
, (void*)ext
);
1801 anv_debug_ignored_stype(ext
->sType
);
1808 anv_GetDeviceGroupPeerMemoryFeatures(
1811 uint32_t localDeviceIndex
,
1812 uint32_t remoteDeviceIndex
,
1813 VkPeerMemoryFeatureFlags
* pPeerMemoryFeatures
)
1815 assert(localDeviceIndex
== 0 && remoteDeviceIndex
== 0);
1816 *pPeerMemoryFeatures
= VK_PEER_MEMORY_FEATURE_COPY_SRC_BIT
|
1817 VK_PEER_MEMORY_FEATURE_COPY_DST_BIT
|
1818 VK_PEER_MEMORY_FEATURE_GENERIC_SRC_BIT
|
1819 VK_PEER_MEMORY_FEATURE_GENERIC_DST_BIT
;
1822 PFN_vkVoidFunction
anv_GetInstanceProcAddr(
1823 VkInstance _instance
,
1826 ANV_FROM_HANDLE(anv_instance
, instance
, _instance
);
1828 /* The Vulkan 1.0 spec for vkGetInstanceProcAddr has a table of exactly
1829 * when we have to return valid function pointers, NULL, or it's left
1830 * undefined. See the table for exact details.
1835 #define LOOKUP_ANV_ENTRYPOINT(entrypoint) \
1836 if (strcmp(pName, "vk" #entrypoint) == 0) \
1837 return (PFN_vkVoidFunction)anv_##entrypoint
1839 LOOKUP_ANV_ENTRYPOINT(EnumerateInstanceExtensionProperties
);
1840 LOOKUP_ANV_ENTRYPOINT(EnumerateInstanceLayerProperties
);
1841 LOOKUP_ANV_ENTRYPOINT(EnumerateInstanceVersion
);
1842 LOOKUP_ANV_ENTRYPOINT(CreateInstance
);
1844 #undef LOOKUP_ANV_ENTRYPOINT
1846 if (instance
== NULL
)
1849 int idx
= anv_get_instance_entrypoint_index(pName
);
1851 return instance
->dispatch
.entrypoints
[idx
];
1853 idx
= anv_get_device_entrypoint_index(pName
);
1855 return instance
->device_dispatch
.entrypoints
[idx
];
1860 /* With version 1+ of the loader interface the ICD should expose
1861 * vk_icdGetInstanceProcAddr to work around certain LD_PRELOAD issues seen in apps.
1864 VKAPI_ATTR PFN_vkVoidFunction VKAPI_CALL
vk_icdGetInstanceProcAddr(
1865 VkInstance instance
,
1869 VKAPI_ATTR PFN_vkVoidFunction VKAPI_CALL
vk_icdGetInstanceProcAddr(
1870 VkInstance instance
,
1873 return anv_GetInstanceProcAddr(instance
, pName
);
1876 PFN_vkVoidFunction
anv_GetDeviceProcAddr(
1880 ANV_FROM_HANDLE(anv_device
, device
, _device
);
1882 if (!device
|| !pName
)
1885 int idx
= anv_get_device_entrypoint_index(pName
);
1889 return device
->dispatch
.entrypoints
[idx
];
1893 anv_CreateDebugReportCallbackEXT(VkInstance _instance
,
1894 const VkDebugReportCallbackCreateInfoEXT
* pCreateInfo
,
1895 const VkAllocationCallbacks
* pAllocator
,
1896 VkDebugReportCallbackEXT
* pCallback
)
1898 ANV_FROM_HANDLE(anv_instance
, instance
, _instance
);
1899 return vk_create_debug_report_callback(&instance
->debug_report_callbacks
,
1900 pCreateInfo
, pAllocator
, &instance
->alloc
,
1905 anv_DestroyDebugReportCallbackEXT(VkInstance _instance
,
1906 VkDebugReportCallbackEXT _callback
,
1907 const VkAllocationCallbacks
* pAllocator
)
1909 ANV_FROM_HANDLE(anv_instance
, instance
, _instance
);
1910 vk_destroy_debug_report_callback(&instance
->debug_report_callbacks
,
1911 _callback
, pAllocator
, &instance
->alloc
);
1915 anv_DebugReportMessageEXT(VkInstance _instance
,
1916 VkDebugReportFlagsEXT flags
,
1917 VkDebugReportObjectTypeEXT objectType
,
1920 int32_t messageCode
,
1921 const char* pLayerPrefix
,
1922 const char* pMessage
)
1924 ANV_FROM_HANDLE(anv_instance
, instance
, _instance
);
1925 vk_debug_report(&instance
->debug_report_callbacks
, flags
, objectType
,
1926 object
, location
, messageCode
, pLayerPrefix
, pMessage
);
1930 anv_queue_init(struct anv_device
*device
, struct anv_queue
*queue
)
1932 queue
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
1933 queue
->device
= device
;
1938 anv_queue_finish(struct anv_queue
*queue
)
1942 static struct anv_state
1943 anv_state_pool_emit_data(struct anv_state_pool
*pool
, size_t size
, size_t align
, const void *p
)
1945 struct anv_state state
;
1947 state
= anv_state_pool_alloc(pool
, size
, align
);
1948 memcpy(state
.map
, p
, size
);
1953 /* Haswell border color is a bit of a disaster. Float and unorm formats use a
1954 * straightforward 32-bit float color in the first 64 bytes. Instead of using
1955 * a nice float/integer union like Gen8+, Haswell specifies the integer border
1956 * color as a separate entry /after/ the float color. The layout of this entry
1957 * also depends on the format's bpp (with extra hacks for RG32), and overlaps.
1959 * Since we don't know the format/bpp, we can't make any of the border colors
1960 * containing '1' work for all formats, as it would be in the wrong place for
1961 * some of them. We opt to make 32-bit integers work as this seems like the
1962 * most common option. Fortunately, transparent black works regardless, as
1963 * all zeroes is the same in every bit-size.
1965 struct hsw_border_color
{
1969 uint32_t _pad1
[108];
1972 struct gen8_border_color
{
1977 /* Pad out to 64 bytes */
1982 anv_device_init_border_colors(struct anv_device
*device
)
1984 if (device
->info
.is_haswell
) {
1985 static const struct hsw_border_color border_colors
[] = {
1986 [VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK
] = { .float32
= { 0.0, 0.0, 0.0, 0.0 } },
1987 [VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK
] = { .float32
= { 0.0, 0.0, 0.0, 1.0 } },
1988 [VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE
] = { .float32
= { 1.0, 1.0, 1.0, 1.0 } },
1989 [VK_BORDER_COLOR_INT_TRANSPARENT_BLACK
] = { .uint32
= { 0, 0, 0, 0 } },
1990 [VK_BORDER_COLOR_INT_OPAQUE_BLACK
] = { .uint32
= { 0, 0, 0, 1 } },
1991 [VK_BORDER_COLOR_INT_OPAQUE_WHITE
] = { .uint32
= { 1, 1, 1, 1 } },
1994 device
->border_colors
=
1995 anv_state_pool_emit_data(&device
->dynamic_state_pool
,
1996 sizeof(border_colors
), 512, border_colors
);
1998 static const struct gen8_border_color border_colors
[] = {
1999 [VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK
] = { .float32
= { 0.0, 0.0, 0.0, 0.0 } },
2000 [VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK
] = { .float32
= { 0.0, 0.0, 0.0, 1.0 } },
2001 [VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE
] = { .float32
= { 1.0, 1.0, 1.0, 1.0 } },
2002 [VK_BORDER_COLOR_INT_TRANSPARENT_BLACK
] = { .uint32
= { 0, 0, 0, 0 } },
2003 [VK_BORDER_COLOR_INT_OPAQUE_BLACK
] = { .uint32
= { 0, 0, 0, 1 } },
2004 [VK_BORDER_COLOR_INT_OPAQUE_WHITE
] = { .uint32
= { 1, 1, 1, 1 } },
2007 device
->border_colors
=
2008 anv_state_pool_emit_data(&device
->dynamic_state_pool
,
2009 sizeof(border_colors
), 64, border_colors
);
2014 anv_device_init_trivial_batch(struct anv_device
*device
)
2016 anv_bo_init_new(&device
->trivial_batch_bo
, device
, 4096);
2018 if (device
->instance
->physicalDevice
.has_exec_async
)
2019 device
->trivial_batch_bo
.flags
|= EXEC_OBJECT_ASYNC
;
2021 if (device
->instance
->physicalDevice
.use_softpin
)
2022 device
->trivial_batch_bo
.flags
|= EXEC_OBJECT_PINNED
;
2024 anv_vma_alloc(device
, &device
->trivial_batch_bo
);
2026 void *map
= anv_gem_mmap(device
, device
->trivial_batch_bo
.gem_handle
,
2029 struct anv_batch batch
= {
2035 anv_batch_emit(&batch
, GEN7_MI_BATCH_BUFFER_END
, bbe
);
2036 anv_batch_emit(&batch
, GEN7_MI_NOOP
, noop
);
2038 if (!device
->info
.has_llc
)
2039 gen_clflush_range(map
, batch
.next
- map
);
2041 anv_gem_munmap(map
, device
->trivial_batch_bo
.size
);
2044 VkResult
anv_EnumerateDeviceExtensionProperties(
2045 VkPhysicalDevice physicalDevice
,
2046 const char* pLayerName
,
2047 uint32_t* pPropertyCount
,
2048 VkExtensionProperties
* pProperties
)
2050 ANV_FROM_HANDLE(anv_physical_device
, device
, physicalDevice
);
2051 VK_OUTARRAY_MAKE(out
, pProperties
, pPropertyCount
);
2053 for (int i
= 0; i
< ANV_DEVICE_EXTENSION_COUNT
; i
++) {
2054 if (device
->supported_extensions
.extensions
[i
]) {
2055 vk_outarray_append(&out
, prop
) {
2056 *prop
= anv_device_extensions
[i
];
2061 return vk_outarray_status(&out
);
2065 anv_device_init_dispatch(struct anv_device
*device
)
2067 const struct anv_device_dispatch_table
*genX_table
;
2068 switch (device
->info
.gen
) {
2070 genX_table
= &gen11_device_dispatch_table
;
2073 genX_table
= &gen10_device_dispatch_table
;
2076 genX_table
= &gen9_device_dispatch_table
;
2079 genX_table
= &gen8_device_dispatch_table
;
2082 if (device
->info
.is_haswell
)
2083 genX_table
= &gen75_device_dispatch_table
;
2085 genX_table
= &gen7_device_dispatch_table
;
2088 unreachable("unsupported gen\n");
2091 for (unsigned i
= 0; i
< ARRAY_SIZE(device
->dispatch
.entrypoints
); i
++) {
2092 /* Vulkan requires that entrypoints for extensions which have not been
2093 * enabled must not be advertised.
2095 if (!anv_device_entrypoint_is_enabled(i
, device
->instance
->app_info
.api_version
,
2096 &device
->instance
->enabled_extensions
,
2097 &device
->enabled_extensions
)) {
2098 device
->dispatch
.entrypoints
[i
] = NULL
;
2099 } else if (genX_table
->entrypoints
[i
]) {
2100 device
->dispatch
.entrypoints
[i
] = genX_table
->entrypoints
[i
];
2102 device
->dispatch
.entrypoints
[i
] =
2103 anv_device_dispatch_table
.entrypoints
[i
];
2109 vk_priority_to_gen(int priority
)
2112 case VK_QUEUE_GLOBAL_PRIORITY_LOW_EXT
:
2113 return GEN_CONTEXT_LOW_PRIORITY
;
2114 case VK_QUEUE_GLOBAL_PRIORITY_MEDIUM_EXT
:
2115 return GEN_CONTEXT_MEDIUM_PRIORITY
;
2116 case VK_QUEUE_GLOBAL_PRIORITY_HIGH_EXT
:
2117 return GEN_CONTEXT_HIGH_PRIORITY
;
2118 case VK_QUEUE_GLOBAL_PRIORITY_REALTIME_EXT
:
2119 return GEN_CONTEXT_REALTIME_PRIORITY
;
2121 unreachable("Invalid priority");
2126 anv_device_init_hiz_clear_value_bo(struct anv_device
*device
)
2128 anv_bo_init_new(&device
->hiz_clear_bo
, device
, 4096);
2130 if (device
->instance
->physicalDevice
.has_exec_async
)
2131 device
->hiz_clear_bo
.flags
|= EXEC_OBJECT_ASYNC
;
2133 if (device
->instance
->physicalDevice
.use_softpin
)
2134 device
->hiz_clear_bo
.flags
|= EXEC_OBJECT_PINNED
;
2136 anv_vma_alloc(device
, &device
->hiz_clear_bo
);
2138 uint32_t *map
= anv_gem_mmap(device
, device
->hiz_clear_bo
.gem_handle
,
2141 union isl_color_value hiz_clear
= { .u32
= { 0, } };
2142 hiz_clear
.f32
[0] = ANV_HZ_FC_VAL
;
2144 memcpy(map
, hiz_clear
.u32
, sizeof(hiz_clear
.u32
));
2145 anv_gem_munmap(map
, device
->hiz_clear_bo
.size
);
2149 get_bo_from_pool(struct gen_batch_decode_bo
*ret
,
2150 struct anv_block_pool
*pool
,
2153 for (uint32_t i
= 0; i
< pool
->nbos
; i
++) {
2154 uint64_t bo_address
= pool
->bos
[i
].offset
& (~0ull >> 16);
2155 uint32_t bo_size
= pool
->bos
[i
].size
;
2156 if (address
>= bo_address
&& address
< (bo_address
+ bo_size
)) {
2157 *ret
= (struct gen_batch_decode_bo
) {
2160 .map
= pool
->bos
[i
].map
,
2168 /* Finding a buffer for batch decoding */
2169 static struct gen_batch_decode_bo
2170 decode_get_bo(void *v_batch
, bool ppgtt
, uint64_t address
)
2172 struct anv_device
*device
= v_batch
;
2173 struct gen_batch_decode_bo ret_bo
= {};
2177 if (get_bo_from_pool(&ret_bo
, &device
->dynamic_state_pool
.block_pool
, address
))
2179 if (get_bo_from_pool(&ret_bo
, &device
->instruction_state_pool
.block_pool
, address
))
2181 if (get_bo_from_pool(&ret_bo
, &device
->binding_table_pool
.block_pool
, address
))
2183 if (get_bo_from_pool(&ret_bo
, &device
->surface_state_pool
.block_pool
, address
))
2186 if (!device
->cmd_buffer_being_decoded
)
2187 return (struct gen_batch_decode_bo
) { };
2189 struct anv_batch_bo
**bo
;
2191 u_vector_foreach(bo
, &device
->cmd_buffer_being_decoded
->seen_bbos
) {
2192 /* The decoder zeroes out the top 16 bits, so we need to as well */
2193 uint64_t bo_address
= (*bo
)->bo
.offset
& (~0ull >> 16);
2195 if (address
>= bo_address
&& address
< bo_address
+ (*bo
)->bo
.size
) {
2196 return (struct gen_batch_decode_bo
) {
2198 .size
= (*bo
)->bo
.size
,
2199 .map
= (*bo
)->bo
.map
,
2204 return (struct gen_batch_decode_bo
) { };
2207 VkResult
anv_CreateDevice(
2208 VkPhysicalDevice physicalDevice
,
2209 const VkDeviceCreateInfo
* pCreateInfo
,
2210 const VkAllocationCallbacks
* pAllocator
,
2213 ANV_FROM_HANDLE(anv_physical_device
, physical_device
, physicalDevice
);
2215 struct anv_device
*device
;
2217 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_DEVICE_CREATE_INFO
);
2219 struct anv_device_extension_table enabled_extensions
= { };
2220 for (uint32_t i
= 0; i
< pCreateInfo
->enabledExtensionCount
; i
++) {
2222 for (idx
= 0; idx
< ANV_DEVICE_EXTENSION_COUNT
; idx
++) {
2223 if (strcmp(pCreateInfo
->ppEnabledExtensionNames
[i
],
2224 anv_device_extensions
[idx
].extensionName
) == 0)
2228 if (idx
>= ANV_DEVICE_EXTENSION_COUNT
)
2229 return vk_error(VK_ERROR_EXTENSION_NOT_PRESENT
);
2231 if (!physical_device
->supported_extensions
.extensions
[idx
])
2232 return vk_error(VK_ERROR_EXTENSION_NOT_PRESENT
);
2234 enabled_extensions
.extensions
[idx
] = true;
2237 /* Check enabled features */
2238 if (pCreateInfo
->pEnabledFeatures
) {
2239 VkPhysicalDeviceFeatures supported_features
;
2240 anv_GetPhysicalDeviceFeatures(physicalDevice
, &supported_features
);
2241 VkBool32
*supported_feature
= (VkBool32
*)&supported_features
;
2242 VkBool32
*enabled_feature
= (VkBool32
*)pCreateInfo
->pEnabledFeatures
;
2243 unsigned num_features
= sizeof(VkPhysicalDeviceFeatures
) / sizeof(VkBool32
);
2244 for (uint32_t i
= 0; i
< num_features
; i
++) {
2245 if (enabled_feature
[i
] && !supported_feature
[i
])
2246 return vk_error(VK_ERROR_FEATURE_NOT_PRESENT
);
2250 /* Check requested queues and fail if we are requested to create any
2251 * queues with flags we don't support.
2253 assert(pCreateInfo
->queueCreateInfoCount
> 0);
2254 for (uint32_t i
= 0; i
< pCreateInfo
->queueCreateInfoCount
; i
++) {
2255 if (pCreateInfo
->pQueueCreateInfos
[i
].flags
!= 0)
2256 return vk_error(VK_ERROR_INITIALIZATION_FAILED
);
2259 /* Check if client specified queue priority. */
2260 const VkDeviceQueueGlobalPriorityCreateInfoEXT
*queue_priority
=
2261 vk_find_struct_const(pCreateInfo
->pQueueCreateInfos
[0].pNext
,
2262 DEVICE_QUEUE_GLOBAL_PRIORITY_CREATE_INFO_EXT
);
2264 VkQueueGlobalPriorityEXT priority
=
2265 queue_priority
? queue_priority
->globalPriority
:
2266 VK_QUEUE_GLOBAL_PRIORITY_MEDIUM_EXT
;
2268 device
= vk_alloc2(&physical_device
->instance
->alloc
, pAllocator
,
2270 VK_SYSTEM_ALLOCATION_SCOPE_DEVICE
);
2272 return vk_error(VK_ERROR_OUT_OF_HOST_MEMORY
);
2274 if (INTEL_DEBUG
& DEBUG_BATCH
) {
2275 const unsigned decode_flags
=
2276 GEN_BATCH_DECODE_FULL
|
2277 ((INTEL_DEBUG
& DEBUG_COLOR
) ? GEN_BATCH_DECODE_IN_COLOR
: 0) |
2278 GEN_BATCH_DECODE_OFFSETS
|
2279 GEN_BATCH_DECODE_FLOATS
;
2281 gen_batch_decode_ctx_init(&device
->decoder_ctx
,
2282 &physical_device
->info
,
2283 stderr
, decode_flags
, NULL
,
2284 decode_get_bo
, NULL
, device
);
2287 device
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
2288 device
->instance
= physical_device
->instance
;
2289 device
->chipset_id
= physical_device
->chipset_id
;
2290 device
->no_hw
= physical_device
->no_hw
;
2291 device
->_lost
= false;
2294 device
->alloc
= *pAllocator
;
2296 device
->alloc
= physical_device
->instance
->alloc
;
2298 /* XXX(chadv): Can we dup() physicalDevice->fd here? */
2299 device
->fd
= open(physical_device
->path
, O_RDWR
| O_CLOEXEC
);
2300 if (device
->fd
== -1) {
2301 result
= vk_error(VK_ERROR_INITIALIZATION_FAILED
);
2305 device
->context_id
= anv_gem_create_context(device
);
2306 if (device
->context_id
== -1) {
2307 result
= vk_error(VK_ERROR_INITIALIZATION_FAILED
);
2311 if (physical_device
->use_softpin
) {
2312 if (pthread_mutex_init(&device
->vma_mutex
, NULL
) != 0) {
2313 result
= vk_error(VK_ERROR_INITIALIZATION_FAILED
);
2317 /* keep the page with address zero out of the allocator */
2318 struct anv_memory_heap
*low_heap
=
2319 &physical_device
->memory
.heaps
[physical_device
->memory
.heap_count
- 1];
2320 util_vma_heap_init(&device
->vma_lo
, low_heap
->vma_start
, low_heap
->vma_size
);
2321 device
->vma_lo_available
= low_heap
->size
;
2323 struct anv_memory_heap
*high_heap
=
2324 &physical_device
->memory
.heaps
[0];
2325 util_vma_heap_init(&device
->vma_hi
, high_heap
->vma_start
, high_heap
->vma_size
);
2326 device
->vma_hi_available
= physical_device
->memory
.heap_count
== 1 ? 0 :
2330 list_inithead(&device
->memory_objects
);
2332 /* As per spec, the driver implementation may deny requests to acquire
2333 * a priority above the default priority (MEDIUM) if the caller does not
2334 * have sufficient privileges. In this scenario VK_ERROR_NOT_PERMITTED_EXT
2337 if (physical_device
->has_context_priority
) {
2338 int err
= anv_gem_set_context_param(device
->fd
, device
->context_id
,
2339 I915_CONTEXT_PARAM_PRIORITY
,
2340 vk_priority_to_gen(priority
));
2341 if (err
!= 0 && priority
> VK_QUEUE_GLOBAL_PRIORITY_MEDIUM_EXT
) {
2342 result
= vk_error(VK_ERROR_NOT_PERMITTED_EXT
);
2347 device
->info
= physical_device
->info
;
2348 device
->isl_dev
= physical_device
->isl_dev
;
2350 /* On Broadwell and later, we can use batch chaining to more efficiently
2351 * implement growing command buffers. Prior to Haswell, the kernel
2352 * command parser gets in the way and we have to fall back to growing
2355 device
->can_chain_batches
= device
->info
.gen
>= 8;
2357 device
->robust_buffer_access
= pCreateInfo
->pEnabledFeatures
&&
2358 pCreateInfo
->pEnabledFeatures
->robustBufferAccess
;
2359 device
->enabled_extensions
= enabled_extensions
;
2361 anv_device_init_dispatch(device
);
2363 if (pthread_mutex_init(&device
->mutex
, NULL
) != 0) {
2364 result
= vk_error(VK_ERROR_INITIALIZATION_FAILED
);
2365 goto fail_context_id
;
2368 pthread_condattr_t condattr
;
2369 if (pthread_condattr_init(&condattr
) != 0) {
2370 result
= vk_error(VK_ERROR_INITIALIZATION_FAILED
);
2373 if (pthread_condattr_setclock(&condattr
, CLOCK_MONOTONIC
) != 0) {
2374 pthread_condattr_destroy(&condattr
);
2375 result
= vk_error(VK_ERROR_INITIALIZATION_FAILED
);
2378 if (pthread_cond_init(&device
->queue_submit
, &condattr
) != 0) {
2379 pthread_condattr_destroy(&condattr
);
2380 result
= vk_error(VK_ERROR_INITIALIZATION_FAILED
);
2383 pthread_condattr_destroy(&condattr
);
2386 (physical_device
->supports_48bit_addresses
? EXEC_OBJECT_SUPPORTS_48B_ADDRESS
: 0) |
2387 (physical_device
->has_exec_async
? EXEC_OBJECT_ASYNC
: 0) |
2388 (physical_device
->has_exec_capture
? EXEC_OBJECT_CAPTURE
: 0) |
2389 (physical_device
->use_softpin
? EXEC_OBJECT_PINNED
: 0);
2391 anv_bo_pool_init(&device
->batch_bo_pool
, device
, bo_flags
);
2393 result
= anv_bo_cache_init(&device
->bo_cache
);
2394 if (result
!= VK_SUCCESS
)
2395 goto fail_batch_bo_pool
;
2397 if (!physical_device
->use_softpin
)
2398 bo_flags
&= ~EXEC_OBJECT_SUPPORTS_48B_ADDRESS
;
2400 result
= anv_state_pool_init(&device
->dynamic_state_pool
, device
,
2401 DYNAMIC_STATE_POOL_MIN_ADDRESS
,
2404 if (result
!= VK_SUCCESS
)
2407 result
= anv_state_pool_init(&device
->instruction_state_pool
, device
,
2408 INSTRUCTION_STATE_POOL_MIN_ADDRESS
,
2411 if (result
!= VK_SUCCESS
)
2412 goto fail_dynamic_state_pool
;
2414 result
= anv_state_pool_init(&device
->surface_state_pool
, device
,
2415 SURFACE_STATE_POOL_MIN_ADDRESS
,
2418 if (result
!= VK_SUCCESS
)
2419 goto fail_instruction_state_pool
;
2421 if (physical_device
->use_softpin
) {
2422 result
= anv_state_pool_init(&device
->binding_table_pool
, device
,
2423 BINDING_TABLE_POOL_MIN_ADDRESS
,
2426 if (result
!= VK_SUCCESS
)
2427 goto fail_surface_state_pool
;
2430 result
= anv_bo_init_new(&device
->workaround_bo
, device
, 4096);
2431 if (result
!= VK_SUCCESS
)
2432 goto fail_binding_table_pool
;
2434 if (physical_device
->use_softpin
)
2435 device
->workaround_bo
.flags
|= EXEC_OBJECT_PINNED
;
2437 if (!anv_vma_alloc(device
, &device
->workaround_bo
))
2438 goto fail_workaround_bo
;
2440 anv_device_init_trivial_batch(device
);
2442 if (device
->info
.gen
>= 10)
2443 anv_device_init_hiz_clear_value_bo(device
);
2445 anv_scratch_pool_init(device
, &device
->scratch_pool
);
2447 anv_queue_init(device
, &device
->queue
);
2449 switch (device
->info
.gen
) {
2451 if (!device
->info
.is_haswell
)
2452 result
= gen7_init_device_state(device
);
2454 result
= gen75_init_device_state(device
);
2457 result
= gen8_init_device_state(device
);
2460 result
= gen9_init_device_state(device
);
2463 result
= gen10_init_device_state(device
);
2466 result
= gen11_init_device_state(device
);
2469 /* Shouldn't get here as we don't create physical devices for any other
2471 unreachable("unhandled gen");
2473 if (result
!= VK_SUCCESS
)
2474 goto fail_workaround_bo
;
2476 anv_pipeline_cache_init(&device
->default_pipeline_cache
, device
, true);
2478 anv_device_init_blorp(device
);
2480 anv_device_init_border_colors(device
);
2482 *pDevice
= anv_device_to_handle(device
);
2487 anv_queue_finish(&device
->queue
);
2488 anv_scratch_pool_finish(device
, &device
->scratch_pool
);
2489 anv_gem_munmap(device
->workaround_bo
.map
, device
->workaround_bo
.size
);
2490 anv_gem_close(device
, device
->workaround_bo
.gem_handle
);
2491 fail_binding_table_pool
:
2492 if (physical_device
->use_softpin
)
2493 anv_state_pool_finish(&device
->binding_table_pool
);
2494 fail_surface_state_pool
:
2495 anv_state_pool_finish(&device
->surface_state_pool
);
2496 fail_instruction_state_pool
:
2497 anv_state_pool_finish(&device
->instruction_state_pool
);
2498 fail_dynamic_state_pool
:
2499 anv_state_pool_finish(&device
->dynamic_state_pool
);
2501 anv_bo_cache_finish(&device
->bo_cache
);
2503 anv_bo_pool_finish(&device
->batch_bo_pool
);
2504 pthread_cond_destroy(&device
->queue_submit
);
2506 pthread_mutex_destroy(&device
->mutex
);
2508 anv_gem_destroy_context(device
, device
->context_id
);
2512 vk_free(&device
->alloc
, device
);
2517 void anv_DestroyDevice(
2519 const VkAllocationCallbacks
* pAllocator
)
2521 ANV_FROM_HANDLE(anv_device
, device
, _device
);
2522 struct anv_physical_device
*physical_device
;
2527 physical_device
= &device
->instance
->physicalDevice
;
2529 anv_device_finish_blorp(device
);
2531 anv_pipeline_cache_finish(&device
->default_pipeline_cache
);
2533 anv_queue_finish(&device
->queue
);
2535 #ifdef HAVE_VALGRIND
2536 /* We only need to free these to prevent valgrind errors. The backing
2537 * BO will go away in a couple of lines so we don't actually leak.
2539 anv_state_pool_free(&device
->dynamic_state_pool
, device
->border_colors
);
2542 anv_scratch_pool_finish(device
, &device
->scratch_pool
);
2544 anv_gem_munmap(device
->workaround_bo
.map
, device
->workaround_bo
.size
);
2545 anv_vma_free(device
, &device
->workaround_bo
);
2546 anv_gem_close(device
, device
->workaround_bo
.gem_handle
);
2548 anv_vma_free(device
, &device
->trivial_batch_bo
);
2549 anv_gem_close(device
, device
->trivial_batch_bo
.gem_handle
);
2550 if (device
->info
.gen
>= 10)
2551 anv_gem_close(device
, device
->hiz_clear_bo
.gem_handle
);
2553 if (physical_device
->use_softpin
)
2554 anv_state_pool_finish(&device
->binding_table_pool
);
2555 anv_state_pool_finish(&device
->surface_state_pool
);
2556 anv_state_pool_finish(&device
->instruction_state_pool
);
2557 anv_state_pool_finish(&device
->dynamic_state_pool
);
2559 anv_bo_cache_finish(&device
->bo_cache
);
2561 anv_bo_pool_finish(&device
->batch_bo_pool
);
2563 pthread_cond_destroy(&device
->queue_submit
);
2564 pthread_mutex_destroy(&device
->mutex
);
2566 anv_gem_destroy_context(device
, device
->context_id
);
2568 if (INTEL_DEBUG
& DEBUG_BATCH
)
2569 gen_batch_decode_ctx_finish(&device
->decoder_ctx
);
2573 vk_free(&device
->alloc
, device
);
2576 VkResult
anv_EnumerateInstanceLayerProperties(
2577 uint32_t* pPropertyCount
,
2578 VkLayerProperties
* pProperties
)
2580 if (pProperties
== NULL
) {
2581 *pPropertyCount
= 0;
2585 /* None supported at this time */
2586 return vk_error(VK_ERROR_LAYER_NOT_PRESENT
);
2589 VkResult
anv_EnumerateDeviceLayerProperties(
2590 VkPhysicalDevice physicalDevice
,
2591 uint32_t* pPropertyCount
,
2592 VkLayerProperties
* pProperties
)
2594 if (pProperties
== NULL
) {
2595 *pPropertyCount
= 0;
2599 /* None supported at this time */
2600 return vk_error(VK_ERROR_LAYER_NOT_PRESENT
);
2603 void anv_GetDeviceQueue(
2605 uint32_t queueNodeIndex
,
2606 uint32_t queueIndex
,
2609 ANV_FROM_HANDLE(anv_device
, device
, _device
);
2611 assert(queueIndex
== 0);
2613 *pQueue
= anv_queue_to_handle(&device
->queue
);
2616 void anv_GetDeviceQueue2(
2618 const VkDeviceQueueInfo2
* pQueueInfo
,
2621 ANV_FROM_HANDLE(anv_device
, device
, _device
);
2623 assert(pQueueInfo
->queueIndex
== 0);
2625 if (pQueueInfo
->flags
== device
->queue
.flags
)
2626 *pQueue
= anv_queue_to_handle(&device
->queue
);
2632 _anv_device_set_lost(struct anv_device
*device
,
2633 const char *file
, int line
,
2634 const char *msg
, ...)
2639 device
->_lost
= true;
2642 err
= __vk_errorv(device
->instance
, device
,
2643 VK_DEBUG_REPORT_OBJECT_TYPE_DEVICE_EXT
,
2644 VK_ERROR_DEVICE_LOST
, file
, line
, msg
, ap
);
2647 if (env_var_as_boolean("ANV_ABORT_ON_DEVICE_LOSS", false))
2654 anv_device_query_status(struct anv_device
*device
)
2656 /* This isn't likely as most of the callers of this function already check
2657 * for it. However, it doesn't hurt to check and it potentially lets us
2660 if (anv_device_is_lost(device
))
2661 return VK_ERROR_DEVICE_LOST
;
2663 uint32_t active
, pending
;
2664 int ret
= anv_gem_gpu_get_reset_stats(device
, &active
, &pending
);
2666 /* We don't know the real error. */
2667 return anv_device_set_lost(device
, "get_reset_stats failed: %m");
2671 return anv_device_set_lost(device
, "GPU hung on one of our command buffers");
2672 } else if (pending
) {
2673 return anv_device_set_lost(device
, "GPU hung with commands in-flight");
2680 anv_device_bo_busy(struct anv_device
*device
, struct anv_bo
*bo
)
2682 /* Note: This only returns whether or not the BO is in use by an i915 GPU.
2683 * Other usages of the BO (such as on different hardware) will not be
2684 * flagged as "busy" by this ioctl. Use with care.
2686 int ret
= anv_gem_busy(device
, bo
->gem_handle
);
2688 return VK_NOT_READY
;
2689 } else if (ret
== -1) {
2690 /* We don't know the real error. */
2691 return anv_device_set_lost(device
, "gem wait failed: %m");
2694 /* Query for device status after the busy call. If the BO we're checking
2695 * got caught in a GPU hang we don't want to return VK_SUCCESS to the
2696 * client because it clearly doesn't have valid data. Yes, this most
2697 * likely means an ioctl, but we just did an ioctl to query the busy status
2698 * so it's no great loss.
2700 return anv_device_query_status(device
);
2704 anv_device_wait(struct anv_device
*device
, struct anv_bo
*bo
,
2707 int ret
= anv_gem_wait(device
, bo
->gem_handle
, &timeout
);
2708 if (ret
== -1 && errno
== ETIME
) {
2710 } else if (ret
== -1) {
2711 /* We don't know the real error. */
2712 return anv_device_set_lost(device
, "gem wait failed: %m");
2715 /* Query for device status after the wait. If the BO we're waiting on got
2716 * caught in a GPU hang we don't want to return VK_SUCCESS to the client
2717 * because it clearly doesn't have valid data. Yes, this most likely means
2718 * an ioctl, but we just did an ioctl to wait so it's no great loss.
2720 return anv_device_query_status(device
);
2723 VkResult
anv_DeviceWaitIdle(
2726 ANV_FROM_HANDLE(anv_device
, device
, _device
);
2727 if (anv_device_is_lost(device
))
2728 return VK_ERROR_DEVICE_LOST
;
2730 struct anv_batch batch
;
2733 batch
.start
= batch
.next
= cmds
;
2734 batch
.end
= (void *) cmds
+ sizeof(cmds
);
2736 anv_batch_emit(&batch
, GEN7_MI_BATCH_BUFFER_END
, bbe
);
2737 anv_batch_emit(&batch
, GEN7_MI_NOOP
, noop
);
2739 return anv_device_submit_simple_batch(device
, &batch
);
2743 anv_vma_alloc(struct anv_device
*device
, struct anv_bo
*bo
)
2745 if (!(bo
->flags
& EXEC_OBJECT_PINNED
))
2748 pthread_mutex_lock(&device
->vma_mutex
);
2752 if (bo
->flags
& EXEC_OBJECT_SUPPORTS_48B_ADDRESS
&&
2753 device
->vma_hi_available
>= bo
->size
) {
2754 uint64_t addr
= util_vma_heap_alloc(&device
->vma_hi
, bo
->size
, 4096);
2756 bo
->offset
= gen_canonical_address(addr
);
2757 assert(addr
== gen_48b_address(bo
->offset
));
2758 device
->vma_hi_available
-= bo
->size
;
2762 if (bo
->offset
== 0 && device
->vma_lo_available
>= bo
->size
) {
2763 uint64_t addr
= util_vma_heap_alloc(&device
->vma_lo
, bo
->size
, 4096);
2765 bo
->offset
= gen_canonical_address(addr
);
2766 assert(addr
== gen_48b_address(bo
->offset
));
2767 device
->vma_lo_available
-= bo
->size
;
2771 pthread_mutex_unlock(&device
->vma_mutex
);
2773 return bo
->offset
!= 0;
2777 anv_vma_free(struct anv_device
*device
, struct anv_bo
*bo
)
2779 if (!(bo
->flags
& EXEC_OBJECT_PINNED
))
2782 const uint64_t addr_48b
= gen_48b_address(bo
->offset
);
2784 pthread_mutex_lock(&device
->vma_mutex
);
2786 if (addr_48b
>= LOW_HEAP_MIN_ADDRESS
&&
2787 addr_48b
<= LOW_HEAP_MAX_ADDRESS
) {
2788 util_vma_heap_free(&device
->vma_lo
, addr_48b
, bo
->size
);
2789 device
->vma_lo_available
+= bo
->size
;
2791 MAYBE_UNUSED
const struct anv_physical_device
*physical_device
=
2792 &device
->instance
->physicalDevice
;
2793 assert(addr_48b
>= physical_device
->memory
.heaps
[0].vma_start
&&
2794 addr_48b
< (physical_device
->memory
.heaps
[0].vma_start
+
2795 physical_device
->memory
.heaps
[0].vma_size
));
2796 util_vma_heap_free(&device
->vma_hi
, addr_48b
, bo
->size
);
2797 device
->vma_hi_available
+= bo
->size
;
2800 pthread_mutex_unlock(&device
->vma_mutex
);
2806 anv_bo_init_new(struct anv_bo
*bo
, struct anv_device
*device
, uint64_t size
)
2808 uint32_t gem_handle
= anv_gem_create(device
, size
);
2810 return vk_error(VK_ERROR_OUT_OF_DEVICE_MEMORY
);
2812 anv_bo_init(bo
, gem_handle
, size
);
2817 VkResult
anv_AllocateMemory(
2819 const VkMemoryAllocateInfo
* pAllocateInfo
,
2820 const VkAllocationCallbacks
* pAllocator
,
2821 VkDeviceMemory
* pMem
)
2823 ANV_FROM_HANDLE(anv_device
, device
, _device
);
2824 struct anv_physical_device
*pdevice
= &device
->instance
->physicalDevice
;
2825 struct anv_device_memory
*mem
;
2826 VkResult result
= VK_SUCCESS
;
2828 assert(pAllocateInfo
->sType
== VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO
);
2830 /* The Vulkan 1.0.33 spec says "allocationSize must be greater than 0". */
2831 assert(pAllocateInfo
->allocationSize
> 0);
2833 if (pAllocateInfo
->allocationSize
> MAX_MEMORY_ALLOCATION_SIZE
)
2834 return VK_ERROR_OUT_OF_DEVICE_MEMORY
;
2836 /* FINISHME: Fail if allocation request exceeds heap size. */
2838 mem
= vk_alloc2(&device
->alloc
, pAllocator
, sizeof(*mem
), 8,
2839 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
2841 return vk_error(VK_ERROR_OUT_OF_HOST_MEMORY
);
2843 assert(pAllocateInfo
->memoryTypeIndex
< pdevice
->memory
.type_count
);
2844 mem
->type
= &pdevice
->memory
.types
[pAllocateInfo
->memoryTypeIndex
];
2848 mem
->host_ptr
= NULL
;
2850 uint64_t bo_flags
= 0;
2852 assert(mem
->type
->heapIndex
< pdevice
->memory
.heap_count
);
2853 if (pdevice
->memory
.heaps
[mem
->type
->heapIndex
].supports_48bit_addresses
)
2854 bo_flags
|= EXEC_OBJECT_SUPPORTS_48B_ADDRESS
;
2856 const struct wsi_memory_allocate_info
*wsi_info
=
2857 vk_find_struct_const(pAllocateInfo
->pNext
, WSI_MEMORY_ALLOCATE_INFO_MESA
);
2858 if (wsi_info
&& wsi_info
->implicit_sync
) {
2859 /* We need to set the WRITE flag on window system buffers so that GEM
2860 * will know we're writing to them and synchronize uses on other rings
2861 * (eg if the display server uses the blitter ring).
2863 bo_flags
|= EXEC_OBJECT_WRITE
;
2864 } else if (pdevice
->has_exec_async
) {
2865 bo_flags
|= EXEC_OBJECT_ASYNC
;
2868 if (pdevice
->use_softpin
)
2869 bo_flags
|= EXEC_OBJECT_PINNED
;
2871 const VkExportMemoryAllocateInfo
*export_info
=
2872 vk_find_struct_const(pAllocateInfo
->pNext
, EXPORT_MEMORY_ALLOCATE_INFO
);
2874 /* Check if we need to support Android HW buffer export. If so,
2875 * create AHardwareBuffer and import memory from it.
2877 bool android_export
= false;
2878 if (export_info
&& export_info
->handleTypes
&
2879 VK_EXTERNAL_MEMORY_HANDLE_TYPE_ANDROID_HARDWARE_BUFFER_BIT_ANDROID
)
2880 android_export
= true;
2882 /* Android memory import. */
2883 const struct VkImportAndroidHardwareBufferInfoANDROID
*ahw_import_info
=
2884 vk_find_struct_const(pAllocateInfo
->pNext
,
2885 IMPORT_ANDROID_HARDWARE_BUFFER_INFO_ANDROID
);
2887 if (ahw_import_info
) {
2888 result
= anv_import_ahw_memory(_device
, mem
, ahw_import_info
);
2889 if (result
!= VK_SUCCESS
)
2893 } else if (android_export
) {
2894 result
= anv_create_ahw_memory(_device
, mem
, pAllocateInfo
);
2895 if (result
!= VK_SUCCESS
)
2898 const struct VkImportAndroidHardwareBufferInfoANDROID import_info
= {
2901 result
= anv_import_ahw_memory(_device
, mem
, &import_info
);
2902 if (result
!= VK_SUCCESS
)
2908 const VkImportMemoryFdInfoKHR
*fd_info
=
2909 vk_find_struct_const(pAllocateInfo
->pNext
, IMPORT_MEMORY_FD_INFO_KHR
);
2911 /* The Vulkan spec permits handleType to be 0, in which case the struct is
2914 if (fd_info
&& fd_info
->handleType
) {
2915 /* At the moment, we support only the below handle types. */
2916 assert(fd_info
->handleType
==
2917 VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD_BIT
||
2918 fd_info
->handleType
==
2919 VK_EXTERNAL_MEMORY_HANDLE_TYPE_DMA_BUF_BIT_EXT
);
2921 result
= anv_bo_cache_import(device
, &device
->bo_cache
, fd_info
->fd
,
2922 bo_flags
| ANV_BO_EXTERNAL
, &mem
->bo
);
2923 if (result
!= VK_SUCCESS
)
2926 VkDeviceSize aligned_alloc_size
=
2927 align_u64(pAllocateInfo
->allocationSize
, 4096);
2929 /* For security purposes, we reject importing the bo if it's smaller
2930 * than the requested allocation size. This prevents a malicious client
2931 * from passing a buffer to a trusted client, lying about the size, and
2932 * telling the trusted client to try and texture from an image that goes
2933 * out-of-bounds. This sort of thing could lead to GPU hangs or worse
2934 * in the trusted client. The trusted client can protect itself against
2935 * this sort of attack but only if it can trust the buffer size.
2937 if (mem
->bo
->size
< aligned_alloc_size
) {
2938 result
= vk_errorf(device
->instance
, device
,
2939 VK_ERROR_INVALID_EXTERNAL_HANDLE
,
2940 "aligned allocationSize too large for "
2941 "VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD_BIT: "
2942 "%"PRIu64
"B > %"PRIu64
"B",
2943 aligned_alloc_size
, mem
->bo
->size
);
2944 anv_bo_cache_release(device
, &device
->bo_cache
, mem
->bo
);
2948 /* From the Vulkan spec:
2950 * "Importing memory from a file descriptor transfers ownership of
2951 * the file descriptor from the application to the Vulkan
2952 * implementation. The application must not perform any operations on
2953 * the file descriptor after a successful import."
2955 * If the import fails, we leave the file descriptor open.
2961 const VkImportMemoryHostPointerInfoEXT
*host_ptr_info
=
2962 vk_find_struct_const(pAllocateInfo
->pNext
,
2963 IMPORT_MEMORY_HOST_POINTER_INFO_EXT
);
2964 if (host_ptr_info
&& host_ptr_info
->handleType
) {
2965 if (host_ptr_info
->handleType
==
2966 VK_EXTERNAL_MEMORY_HANDLE_TYPE_HOST_MAPPED_FOREIGN_MEMORY_BIT_EXT
) {
2967 result
= vk_error(VK_ERROR_INVALID_EXTERNAL_HANDLE
);
2971 assert(host_ptr_info
->handleType
==
2972 VK_EXTERNAL_MEMORY_HANDLE_TYPE_HOST_ALLOCATION_BIT_EXT
);
2974 result
= anv_bo_cache_import_host_ptr(
2975 device
, &device
->bo_cache
, host_ptr_info
->pHostPointer
,
2976 pAllocateInfo
->allocationSize
, bo_flags
, &mem
->bo
);
2978 if (result
!= VK_SUCCESS
)
2981 mem
->host_ptr
= host_ptr_info
->pHostPointer
;
2985 /* Regular allocate (not importing memory). */
2987 if (export_info
&& export_info
->handleTypes
)
2988 bo_flags
|= ANV_BO_EXTERNAL
;
2990 result
= anv_bo_cache_alloc(device
, &device
->bo_cache
,
2991 pAllocateInfo
->allocationSize
, bo_flags
,
2993 if (result
!= VK_SUCCESS
)
2996 const VkMemoryDedicatedAllocateInfo
*dedicated_info
=
2997 vk_find_struct_const(pAllocateInfo
->pNext
, MEMORY_DEDICATED_ALLOCATE_INFO
);
2998 if (dedicated_info
&& dedicated_info
->image
!= VK_NULL_HANDLE
) {
2999 ANV_FROM_HANDLE(anv_image
, image
, dedicated_info
->image
);
3001 /* Some legacy (non-modifiers) consumers need the tiling to be set on
3002 * the BO. In this case, we have a dedicated allocation.
3004 if (image
->needs_set_tiling
) {
3005 const uint32_t i915_tiling
=
3006 isl_tiling_to_i915_tiling(image
->planes
[0].surface
.isl
.tiling
);
3007 int ret
= anv_gem_set_tiling(device
, mem
->bo
->gem_handle
,
3008 image
->planes
[0].surface
.isl
.row_pitch_B
,
3011 anv_bo_cache_release(device
, &device
->bo_cache
, mem
->bo
);
3012 return vk_errorf(device
->instance
, NULL
,
3013 VK_ERROR_OUT_OF_DEVICE_MEMORY
,
3014 "failed to set BO tiling: %m");
3020 pthread_mutex_lock(&device
->mutex
);
3021 list_addtail(&mem
->link
, &device
->memory_objects
);
3022 pthread_mutex_unlock(&device
->mutex
);
3024 *pMem
= anv_device_memory_to_handle(mem
);
3026 p_atomic_add(&pdevice
->memory
.heaps
[mem
->type
->heapIndex
].used
,
3032 vk_free2(&device
->alloc
, pAllocator
, mem
);
3037 VkResult
anv_GetMemoryFdKHR(
3039 const VkMemoryGetFdInfoKHR
* pGetFdInfo
,
3042 ANV_FROM_HANDLE(anv_device
, dev
, device_h
);
3043 ANV_FROM_HANDLE(anv_device_memory
, mem
, pGetFdInfo
->memory
);
3045 assert(pGetFdInfo
->sType
== VK_STRUCTURE_TYPE_MEMORY_GET_FD_INFO_KHR
);
3047 assert(pGetFdInfo
->handleType
== VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD_BIT
||
3048 pGetFdInfo
->handleType
== VK_EXTERNAL_MEMORY_HANDLE_TYPE_DMA_BUF_BIT_EXT
);
3050 return anv_bo_cache_export(dev
, &dev
->bo_cache
, mem
->bo
, pFd
);
3053 VkResult
anv_GetMemoryFdPropertiesKHR(
3055 VkExternalMemoryHandleTypeFlagBits handleType
,
3057 VkMemoryFdPropertiesKHR
* pMemoryFdProperties
)
3059 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3060 struct anv_physical_device
*pdevice
= &device
->instance
->physicalDevice
;
3062 switch (handleType
) {
3063 case VK_EXTERNAL_MEMORY_HANDLE_TYPE_DMA_BUF_BIT_EXT
:
3064 /* dma-buf can be imported as any memory type */
3065 pMemoryFdProperties
->memoryTypeBits
=
3066 (1 << pdevice
->memory
.type_count
) - 1;
3070 /* The valid usage section for this function says:
3072 * "handleType must not be one of the handle types defined as
3075 * So opaque handle types fall into the default "unsupported" case.
3077 return vk_error(VK_ERROR_INVALID_EXTERNAL_HANDLE
);
3081 VkResult
anv_GetMemoryHostPointerPropertiesEXT(
3083 VkExternalMemoryHandleTypeFlagBits handleType
,
3084 const void* pHostPointer
,
3085 VkMemoryHostPointerPropertiesEXT
* pMemoryHostPointerProperties
)
3087 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3089 assert(pMemoryHostPointerProperties
->sType
==
3090 VK_STRUCTURE_TYPE_MEMORY_HOST_POINTER_PROPERTIES_EXT
);
3092 switch (handleType
) {
3093 case VK_EXTERNAL_MEMORY_HANDLE_TYPE_HOST_ALLOCATION_BIT_EXT
: {
3094 struct anv_physical_device
*pdevice
= &device
->instance
->physicalDevice
;
3096 /* Host memory can be imported as any memory type. */
3097 pMemoryHostPointerProperties
->memoryTypeBits
=
3098 (1ull << pdevice
->memory
.type_count
) - 1;
3103 return VK_ERROR_INVALID_EXTERNAL_HANDLE
;
3107 void anv_FreeMemory(
3109 VkDeviceMemory _mem
,
3110 const VkAllocationCallbacks
* pAllocator
)
3112 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3113 ANV_FROM_HANDLE(anv_device_memory
, mem
, _mem
);
3114 struct anv_physical_device
*pdevice
= &device
->instance
->physicalDevice
;
3119 pthread_mutex_lock(&device
->mutex
);
3120 list_del(&mem
->link
);
3121 pthread_mutex_unlock(&device
->mutex
);
3124 anv_UnmapMemory(_device
, _mem
);
3126 p_atomic_add(&pdevice
->memory
.heaps
[mem
->type
->heapIndex
].used
,
3129 anv_bo_cache_release(device
, &device
->bo_cache
, mem
->bo
);
3131 #if defined(ANDROID) && ANDROID_API_LEVEL >= 26
3133 AHardwareBuffer_release(mem
->ahw
);
3136 vk_free2(&device
->alloc
, pAllocator
, mem
);
3139 VkResult
anv_MapMemory(
3141 VkDeviceMemory _memory
,
3142 VkDeviceSize offset
,
3144 VkMemoryMapFlags flags
,
3147 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3148 ANV_FROM_HANDLE(anv_device_memory
, mem
, _memory
);
3155 if (mem
->host_ptr
) {
3156 *ppData
= mem
->host_ptr
+ offset
;
3160 if (size
== VK_WHOLE_SIZE
)
3161 size
= mem
->bo
->size
- offset
;
3163 /* From the Vulkan spec version 1.0.32 docs for MapMemory:
3165 * * If size is not equal to VK_WHOLE_SIZE, size must be greater than 0
3166 * assert(size != 0);
3167 * * If size is not equal to VK_WHOLE_SIZE, size must be less than or
3168 * equal to the size of the memory minus offset
3171 assert(offset
+ size
<= mem
->bo
->size
);
3173 /* FIXME: Is this supposed to be thread safe? Since vkUnmapMemory() only
3174 * takes a VkDeviceMemory pointer, it seems like only one map of the memory
3175 * at a time is valid. We could just mmap up front and return an offset
3176 * pointer here, but that may exhaust virtual memory on 32 bit
3179 uint32_t gem_flags
= 0;
3181 if (!device
->info
.has_llc
&&
3182 (mem
->type
->propertyFlags
& VK_MEMORY_PROPERTY_HOST_COHERENT_BIT
))
3183 gem_flags
|= I915_MMAP_WC
;
3185 /* GEM will fail to map if the offset isn't 4k-aligned. Round down. */
3186 uint64_t map_offset
= offset
& ~4095ull;
3187 assert(offset
>= map_offset
);
3188 uint64_t map_size
= (offset
+ size
) - map_offset
;
3190 /* Let's map whole pages */
3191 map_size
= align_u64(map_size
, 4096);
3193 void *map
= anv_gem_mmap(device
, mem
->bo
->gem_handle
,
3194 map_offset
, map_size
, gem_flags
);
3195 if (map
== MAP_FAILED
)
3196 return vk_error(VK_ERROR_MEMORY_MAP_FAILED
);
3199 mem
->map_size
= map_size
;
3201 *ppData
= mem
->map
+ (offset
- map_offset
);
3206 void anv_UnmapMemory(
3208 VkDeviceMemory _memory
)
3210 ANV_FROM_HANDLE(anv_device_memory
, mem
, _memory
);
3212 if (mem
== NULL
|| mem
->host_ptr
)
3215 anv_gem_munmap(mem
->map
, mem
->map_size
);
3222 clflush_mapped_ranges(struct anv_device
*device
,
3224 const VkMappedMemoryRange
*ranges
)
3226 for (uint32_t i
= 0; i
< count
; i
++) {
3227 ANV_FROM_HANDLE(anv_device_memory
, mem
, ranges
[i
].memory
);
3228 if (ranges
[i
].offset
>= mem
->map_size
)
3231 gen_clflush_range(mem
->map
+ ranges
[i
].offset
,
3232 MIN2(ranges
[i
].size
, mem
->map_size
- ranges
[i
].offset
));
3236 VkResult
anv_FlushMappedMemoryRanges(
3238 uint32_t memoryRangeCount
,
3239 const VkMappedMemoryRange
* pMemoryRanges
)
3241 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3243 if (device
->info
.has_llc
)
3246 /* Make sure the writes we're flushing have landed. */
3247 __builtin_ia32_mfence();
3249 clflush_mapped_ranges(device
, memoryRangeCount
, pMemoryRanges
);
3254 VkResult
anv_InvalidateMappedMemoryRanges(
3256 uint32_t memoryRangeCount
,
3257 const VkMappedMemoryRange
* pMemoryRanges
)
3259 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3261 if (device
->info
.has_llc
)
3264 clflush_mapped_ranges(device
, memoryRangeCount
, pMemoryRanges
);
3266 /* Make sure no reads get moved up above the invalidate. */
3267 __builtin_ia32_mfence();
3272 void anv_GetBufferMemoryRequirements(
3275 VkMemoryRequirements
* pMemoryRequirements
)
3277 ANV_FROM_HANDLE(anv_buffer
, buffer
, _buffer
);
3278 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3279 struct anv_physical_device
*pdevice
= &device
->instance
->physicalDevice
;
3281 /* The Vulkan spec (git aaed022) says:
3283 * memoryTypeBits is a bitfield and contains one bit set for every
3284 * supported memory type for the resource. The bit `1<<i` is set if and
3285 * only if the memory type `i` in the VkPhysicalDeviceMemoryProperties
3286 * structure for the physical device is supported.
3288 uint32_t memory_types
= 0;
3289 for (uint32_t i
= 0; i
< pdevice
->memory
.type_count
; i
++) {
3290 uint32_t valid_usage
= pdevice
->memory
.types
[i
].valid_buffer_usage
;
3291 if ((valid_usage
& buffer
->usage
) == buffer
->usage
)
3292 memory_types
|= (1u << i
);
3295 /* Base alignment requirement of a cache line */
3296 uint32_t alignment
= 16;
3298 /* We need an alignment of 32 for pushing UBOs */
3299 if (buffer
->usage
& VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT
)
3300 alignment
= MAX2(alignment
, 32);
3302 pMemoryRequirements
->size
= buffer
->size
;
3303 pMemoryRequirements
->alignment
= alignment
;
3305 /* Storage and Uniform buffers should have their size aligned to
3306 * 32-bits to avoid boundary checks when last DWord is not complete.
3307 * This would ensure that not internal padding would be needed for
3310 if (device
->robust_buffer_access
&&
3311 (buffer
->usage
& VK_BUFFER_USAGE_UNIFORM_BUFFER_BIT
||
3312 buffer
->usage
& VK_BUFFER_USAGE_STORAGE_BUFFER_BIT
))
3313 pMemoryRequirements
->size
= align_u64(buffer
->size
, 4);
3315 pMemoryRequirements
->memoryTypeBits
= memory_types
;
3318 void anv_GetBufferMemoryRequirements2(
3320 const VkBufferMemoryRequirementsInfo2
* pInfo
,
3321 VkMemoryRequirements2
* pMemoryRequirements
)
3323 anv_GetBufferMemoryRequirements(_device
, pInfo
->buffer
,
3324 &pMemoryRequirements
->memoryRequirements
);
3326 vk_foreach_struct(ext
, pMemoryRequirements
->pNext
) {
3327 switch (ext
->sType
) {
3328 case VK_STRUCTURE_TYPE_MEMORY_DEDICATED_REQUIREMENTS
: {
3329 VkMemoryDedicatedRequirements
*requirements
= (void *)ext
;
3330 requirements
->prefersDedicatedAllocation
= false;
3331 requirements
->requiresDedicatedAllocation
= false;
3336 anv_debug_ignored_stype(ext
->sType
);
3342 void anv_GetImageMemoryRequirements(
3345 VkMemoryRequirements
* pMemoryRequirements
)
3347 ANV_FROM_HANDLE(anv_image
, image
, _image
);
3348 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3349 struct anv_physical_device
*pdevice
= &device
->instance
->physicalDevice
;
3351 /* The Vulkan spec (git aaed022) says:
3353 * memoryTypeBits is a bitfield and contains one bit set for every
3354 * supported memory type for the resource. The bit `1<<i` is set if and
3355 * only if the memory type `i` in the VkPhysicalDeviceMemoryProperties
3356 * structure for the physical device is supported.
3358 * All types are currently supported for images.
3360 uint32_t memory_types
= (1ull << pdevice
->memory
.type_count
) - 1;
3362 /* We must have image allocated or imported at this point. According to the
3363 * specification, external images must have been bound to memory before
3364 * calling GetImageMemoryRequirements.
3366 assert(image
->size
> 0);
3368 pMemoryRequirements
->size
= image
->size
;
3369 pMemoryRequirements
->alignment
= image
->alignment
;
3370 pMemoryRequirements
->memoryTypeBits
= memory_types
;
3373 void anv_GetImageMemoryRequirements2(
3375 const VkImageMemoryRequirementsInfo2
* pInfo
,
3376 VkMemoryRequirements2
* pMemoryRequirements
)
3378 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3379 ANV_FROM_HANDLE(anv_image
, image
, pInfo
->image
);
3381 anv_GetImageMemoryRequirements(_device
, pInfo
->image
,
3382 &pMemoryRequirements
->memoryRequirements
);
3384 vk_foreach_struct_const(ext
, pInfo
->pNext
) {
3385 switch (ext
->sType
) {
3386 case VK_STRUCTURE_TYPE_IMAGE_PLANE_MEMORY_REQUIREMENTS_INFO
: {
3387 struct anv_physical_device
*pdevice
= &device
->instance
->physicalDevice
;
3388 const VkImagePlaneMemoryRequirementsInfo
*plane_reqs
=
3389 (const VkImagePlaneMemoryRequirementsInfo
*) ext
;
3390 uint32_t plane
= anv_image_aspect_to_plane(image
->aspects
,
3391 plane_reqs
->planeAspect
);
3393 assert(image
->planes
[plane
].offset
== 0);
3395 /* The Vulkan spec (git aaed022) says:
3397 * memoryTypeBits is a bitfield and contains one bit set for every
3398 * supported memory type for the resource. The bit `1<<i` is set
3399 * if and only if the memory type `i` in the
3400 * VkPhysicalDeviceMemoryProperties structure for the physical
3401 * device is supported.
3403 * All types are currently supported for images.
3405 pMemoryRequirements
->memoryRequirements
.memoryTypeBits
=
3406 (1ull << pdevice
->memory
.type_count
) - 1;
3408 /* We must have image allocated or imported at this point. According to the
3409 * specification, external images must have been bound to memory before
3410 * calling GetImageMemoryRequirements.
3412 assert(image
->planes
[plane
].size
> 0);
3414 pMemoryRequirements
->memoryRequirements
.size
= image
->planes
[plane
].size
;
3415 pMemoryRequirements
->memoryRequirements
.alignment
=
3416 image
->planes
[plane
].alignment
;
3421 anv_debug_ignored_stype(ext
->sType
);
3426 vk_foreach_struct(ext
, pMemoryRequirements
->pNext
) {
3427 switch (ext
->sType
) {
3428 case VK_STRUCTURE_TYPE_MEMORY_DEDICATED_REQUIREMENTS
: {
3429 VkMemoryDedicatedRequirements
*requirements
= (void *)ext
;
3430 if (image
->needs_set_tiling
|| image
->external_format
) {
3431 /* If we need to set the tiling for external consumers, we need a
3432 * dedicated allocation.
3434 * See also anv_AllocateMemory.
3436 requirements
->prefersDedicatedAllocation
= true;
3437 requirements
->requiresDedicatedAllocation
= true;
3439 requirements
->prefersDedicatedAllocation
= false;
3440 requirements
->requiresDedicatedAllocation
= false;
3446 anv_debug_ignored_stype(ext
->sType
);
3452 void anv_GetImageSparseMemoryRequirements(
3455 uint32_t* pSparseMemoryRequirementCount
,
3456 VkSparseImageMemoryRequirements
* pSparseMemoryRequirements
)
3458 *pSparseMemoryRequirementCount
= 0;
3461 void anv_GetImageSparseMemoryRequirements2(
3463 const VkImageSparseMemoryRequirementsInfo2
* pInfo
,
3464 uint32_t* pSparseMemoryRequirementCount
,
3465 VkSparseImageMemoryRequirements2
* pSparseMemoryRequirements
)
3467 *pSparseMemoryRequirementCount
= 0;
3470 void anv_GetDeviceMemoryCommitment(
3472 VkDeviceMemory memory
,
3473 VkDeviceSize
* pCommittedMemoryInBytes
)
3475 *pCommittedMemoryInBytes
= 0;
3479 anv_bind_buffer_memory(const VkBindBufferMemoryInfo
*pBindInfo
)
3481 ANV_FROM_HANDLE(anv_device_memory
, mem
, pBindInfo
->memory
);
3482 ANV_FROM_HANDLE(anv_buffer
, buffer
, pBindInfo
->buffer
);
3484 assert(pBindInfo
->sType
== VK_STRUCTURE_TYPE_BIND_BUFFER_MEMORY_INFO
);
3487 assert((buffer
->usage
& mem
->type
->valid_buffer_usage
) == buffer
->usage
);
3488 buffer
->address
= (struct anv_address
) {
3490 .offset
= pBindInfo
->memoryOffset
,
3493 buffer
->address
= ANV_NULL_ADDRESS
;
3497 VkResult
anv_BindBufferMemory(
3500 VkDeviceMemory memory
,
3501 VkDeviceSize memoryOffset
)
3503 anv_bind_buffer_memory(
3504 &(VkBindBufferMemoryInfo
) {
3505 .sType
= VK_STRUCTURE_TYPE_BIND_BUFFER_MEMORY_INFO
,
3508 .memoryOffset
= memoryOffset
,
3514 VkResult
anv_BindBufferMemory2(
3516 uint32_t bindInfoCount
,
3517 const VkBindBufferMemoryInfo
* pBindInfos
)
3519 for (uint32_t i
= 0; i
< bindInfoCount
; i
++)
3520 anv_bind_buffer_memory(&pBindInfos
[i
]);
3525 VkResult
anv_QueueBindSparse(
3527 uint32_t bindInfoCount
,
3528 const VkBindSparseInfo
* pBindInfo
,
3531 ANV_FROM_HANDLE(anv_queue
, queue
, _queue
);
3532 if (anv_device_is_lost(queue
->device
))
3533 return VK_ERROR_DEVICE_LOST
;
3535 return vk_error(VK_ERROR_FEATURE_NOT_PRESENT
);
3540 VkResult
anv_CreateEvent(
3542 const VkEventCreateInfo
* pCreateInfo
,
3543 const VkAllocationCallbacks
* pAllocator
,
3546 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3547 struct anv_state state
;
3548 struct anv_event
*event
;
3550 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_EVENT_CREATE_INFO
);
3552 state
= anv_state_pool_alloc(&device
->dynamic_state_pool
,
3555 event
->state
= state
;
3556 event
->semaphore
= VK_EVENT_RESET
;
3558 if (!device
->info
.has_llc
) {
3559 /* Make sure the writes we're flushing have landed. */
3560 __builtin_ia32_mfence();
3561 __builtin_ia32_clflush(event
);
3564 *pEvent
= anv_event_to_handle(event
);
3569 void anv_DestroyEvent(
3572 const VkAllocationCallbacks
* pAllocator
)
3574 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3575 ANV_FROM_HANDLE(anv_event
, event
, _event
);
3580 anv_state_pool_free(&device
->dynamic_state_pool
, event
->state
);
3583 VkResult
anv_GetEventStatus(
3587 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3588 ANV_FROM_HANDLE(anv_event
, event
, _event
);
3590 if (anv_device_is_lost(device
))
3591 return VK_ERROR_DEVICE_LOST
;
3593 if (!device
->info
.has_llc
) {
3594 /* Invalidate read cache before reading event written by GPU. */
3595 __builtin_ia32_clflush(event
);
3596 __builtin_ia32_mfence();
3600 return event
->semaphore
;
3603 VkResult
anv_SetEvent(
3607 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3608 ANV_FROM_HANDLE(anv_event
, event
, _event
);
3610 event
->semaphore
= VK_EVENT_SET
;
3612 if (!device
->info
.has_llc
) {
3613 /* Make sure the writes we're flushing have landed. */
3614 __builtin_ia32_mfence();
3615 __builtin_ia32_clflush(event
);
3621 VkResult
anv_ResetEvent(
3625 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3626 ANV_FROM_HANDLE(anv_event
, event
, _event
);
3628 event
->semaphore
= VK_EVENT_RESET
;
3630 if (!device
->info
.has_llc
) {
3631 /* Make sure the writes we're flushing have landed. */
3632 __builtin_ia32_mfence();
3633 __builtin_ia32_clflush(event
);
3641 VkResult
anv_CreateBuffer(
3643 const VkBufferCreateInfo
* pCreateInfo
,
3644 const VkAllocationCallbacks
* pAllocator
,
3647 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3648 struct anv_buffer
*buffer
;
3650 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO
);
3652 buffer
= vk_alloc2(&device
->alloc
, pAllocator
, sizeof(*buffer
), 8,
3653 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
3655 return vk_error(VK_ERROR_OUT_OF_HOST_MEMORY
);
3657 buffer
->size
= pCreateInfo
->size
;
3658 buffer
->usage
= pCreateInfo
->usage
;
3659 buffer
->address
= ANV_NULL_ADDRESS
;
3661 *pBuffer
= anv_buffer_to_handle(buffer
);
3666 void anv_DestroyBuffer(
3669 const VkAllocationCallbacks
* pAllocator
)
3671 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3672 ANV_FROM_HANDLE(anv_buffer
, buffer
, _buffer
);
3677 vk_free2(&device
->alloc
, pAllocator
, buffer
);
3680 VkDeviceAddress
anv_GetBufferDeviceAddressEXT(
3682 const VkBufferDeviceAddressInfoEXT
* pInfo
)
3684 ANV_FROM_HANDLE(anv_buffer
, buffer
, pInfo
->buffer
);
3686 assert(buffer
->address
.bo
->flags
& EXEC_OBJECT_PINNED
);
3688 return anv_address_physical(buffer
->address
);
3692 anv_fill_buffer_surface_state(struct anv_device
*device
, struct anv_state state
,
3693 enum isl_format format
,
3694 struct anv_address address
,
3695 uint32_t range
, uint32_t stride
)
3697 isl_buffer_fill_state(&device
->isl_dev
, state
.map
,
3698 .address
= anv_address_physical(address
),
3699 .mocs
= device
->default_mocs
,
3702 .swizzle
= ISL_SWIZZLE_IDENTITY
,
3703 .stride_B
= stride
);
3706 void anv_DestroySampler(
3709 const VkAllocationCallbacks
* pAllocator
)
3711 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3712 ANV_FROM_HANDLE(anv_sampler
, sampler
, _sampler
);
3717 if (sampler
->bindless_state
.map
) {
3718 anv_state_pool_free(&device
->dynamic_state_pool
,
3719 sampler
->bindless_state
);
3722 vk_free2(&device
->alloc
, pAllocator
, sampler
);
3725 VkResult
anv_CreateFramebuffer(
3727 const VkFramebufferCreateInfo
* pCreateInfo
,
3728 const VkAllocationCallbacks
* pAllocator
,
3729 VkFramebuffer
* pFramebuffer
)
3731 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3732 struct anv_framebuffer
*framebuffer
;
3734 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_FRAMEBUFFER_CREATE_INFO
);
3736 size_t size
= sizeof(*framebuffer
);
3738 /* VK_KHR_imageless_framebuffer extension says:
3740 * If flags includes VK_FRAMEBUFFER_CREATE_IMAGELESS_BIT_KHR,
3741 * parameter pAttachments is ignored.
3743 if (!(pCreateInfo
->flags
& VK_FRAMEBUFFER_CREATE_IMAGELESS_BIT_KHR
)) {
3744 size
+= sizeof(struct anv_image_view
*) * pCreateInfo
->attachmentCount
;
3745 framebuffer
= vk_alloc2(&device
->alloc
, pAllocator
, size
, 8,
3746 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
3747 if (framebuffer
== NULL
)
3748 return vk_error(VK_ERROR_OUT_OF_HOST_MEMORY
);
3750 for (uint32_t i
= 0; i
< pCreateInfo
->attachmentCount
; i
++) {
3751 ANV_FROM_HANDLE(anv_image_view
, iview
, pCreateInfo
->pAttachments
[i
]);
3752 framebuffer
->attachments
[i
] = iview
;
3754 framebuffer
->attachment_count
= pCreateInfo
->attachmentCount
;
3756 assert(device
->enabled_extensions
.KHR_imageless_framebuffer
);
3757 framebuffer
= vk_alloc2(&device
->alloc
, pAllocator
, size
, 8,
3758 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
3759 if (framebuffer
== NULL
)
3760 return vk_error(VK_ERROR_OUT_OF_HOST_MEMORY
);
3762 framebuffer
->attachment_count
= 0;
3765 framebuffer
->width
= pCreateInfo
->width
;
3766 framebuffer
->height
= pCreateInfo
->height
;
3767 framebuffer
->layers
= pCreateInfo
->layers
;
3769 *pFramebuffer
= anv_framebuffer_to_handle(framebuffer
);
3774 void anv_DestroyFramebuffer(
3777 const VkAllocationCallbacks
* pAllocator
)
3779 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3780 ANV_FROM_HANDLE(anv_framebuffer
, fb
, _fb
);
3785 vk_free2(&device
->alloc
, pAllocator
, fb
);
3788 static const VkTimeDomainEXT anv_time_domains
[] = {
3789 VK_TIME_DOMAIN_DEVICE_EXT
,
3790 VK_TIME_DOMAIN_CLOCK_MONOTONIC_EXT
,
3791 VK_TIME_DOMAIN_CLOCK_MONOTONIC_RAW_EXT
,
3794 VkResult
anv_GetPhysicalDeviceCalibrateableTimeDomainsEXT(
3795 VkPhysicalDevice physicalDevice
,
3796 uint32_t *pTimeDomainCount
,
3797 VkTimeDomainEXT
*pTimeDomains
)
3800 VK_OUTARRAY_MAKE(out
, pTimeDomains
, pTimeDomainCount
);
3802 for (d
= 0; d
< ARRAY_SIZE(anv_time_domains
); d
++) {
3803 vk_outarray_append(&out
, i
) {
3804 *i
= anv_time_domains
[d
];
3808 return vk_outarray_status(&out
);
3812 anv_clock_gettime(clockid_t clock_id
)
3814 struct timespec current
;
3817 ret
= clock_gettime(clock_id
, ¤t
);
3818 if (ret
< 0 && clock_id
== CLOCK_MONOTONIC_RAW
)
3819 ret
= clock_gettime(CLOCK_MONOTONIC
, ¤t
);
3823 return (uint64_t) current
.tv_sec
* 1000000000ULL + current
.tv_nsec
;
3826 #define TIMESTAMP 0x2358
3828 VkResult
anv_GetCalibratedTimestampsEXT(
3830 uint32_t timestampCount
,
3831 const VkCalibratedTimestampInfoEXT
*pTimestampInfos
,
3832 uint64_t *pTimestamps
,
3833 uint64_t *pMaxDeviation
)
3835 ANV_FROM_HANDLE(anv_device
, device
, _device
);
3836 uint64_t timestamp_frequency
= device
->info
.timestamp_frequency
;
3839 uint64_t begin
, end
;
3840 uint64_t max_clock_period
= 0;
3842 begin
= anv_clock_gettime(CLOCK_MONOTONIC_RAW
);
3844 for (d
= 0; d
< timestampCount
; d
++) {
3845 switch (pTimestampInfos
[d
].timeDomain
) {
3846 case VK_TIME_DOMAIN_DEVICE_EXT
:
3847 ret
= anv_gem_reg_read(device
, TIMESTAMP
| 1,
3851 return anv_device_set_lost(device
, "Failed to read the TIMESTAMP "
3854 uint64_t device_period
= DIV_ROUND_UP(1000000000, timestamp_frequency
);
3855 max_clock_period
= MAX2(max_clock_period
, device_period
);
3857 case VK_TIME_DOMAIN_CLOCK_MONOTONIC_EXT
:
3858 pTimestamps
[d
] = anv_clock_gettime(CLOCK_MONOTONIC
);
3859 max_clock_period
= MAX2(max_clock_period
, 1);
3862 case VK_TIME_DOMAIN_CLOCK_MONOTONIC_RAW_EXT
:
3863 pTimestamps
[d
] = begin
;
3871 end
= anv_clock_gettime(CLOCK_MONOTONIC_RAW
);
3874 * The maximum deviation is the sum of the interval over which we
3875 * perform the sampling and the maximum period of any sampled
3876 * clock. That's because the maximum skew between any two sampled
3877 * clock edges is when the sampled clock with the largest period is
3878 * sampled at the end of that period but right at the beginning of the
3879 * sampling interval and some other clock is sampled right at the
3880 * begining of its sampling period and right at the end of the
3881 * sampling interval. Let's assume the GPU has the longest clock
3882 * period and that the application is sampling GPU and monotonic:
3885 * w x y z 0 1 2 3 4 5 6 7 8 9 a b c d e f
3886 * Raw -_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-
3890 * GPU -----_____-----_____-----_____-----_____
3893 * x y z 0 1 2 3 4 5 6 7 8 9 a b c
3894 * Monotonic -_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-
3896 * Interval <----------------->
3897 * Deviation <-------------------------->
3901 * m = read(monotonic) 2
3904 * We round the sample interval up by one tick to cover sampling error
3905 * in the interval clock
3908 uint64_t sample_interval
= end
- begin
+ 1;
3910 *pMaxDeviation
= sample_interval
+ max_clock_period
;
3915 /* vk_icd.h does not declare this function, so we declare it here to
3916 * suppress Wmissing-prototypes.
3918 PUBLIC VKAPI_ATTR VkResult VKAPI_CALL
3919 vk_icdNegotiateLoaderICDInterfaceVersion(uint32_t* pSupportedVersion
);
3921 PUBLIC VKAPI_ATTR VkResult VKAPI_CALL
3922 vk_icdNegotiateLoaderICDInterfaceVersion(uint32_t* pSupportedVersion
)
3924 /* For the full details on loader interface versioning, see
3925 * <https://github.com/KhronosGroup/Vulkan-LoaderAndValidationLayers/blob/master/loader/LoaderAndLayerInterface.md>.
3926 * What follows is a condensed summary, to help you navigate the large and
3927 * confusing official doc.
3929 * - Loader interface v0 is incompatible with later versions. We don't
3932 * - In loader interface v1:
3933 * - The first ICD entrypoint called by the loader is
3934 * vk_icdGetInstanceProcAddr(). The ICD must statically expose this
3936 * - The ICD must statically expose no other Vulkan symbol unless it is
3937 * linked with -Bsymbolic.
3938 * - Each dispatchable Vulkan handle created by the ICD must be
3939 * a pointer to a struct whose first member is VK_LOADER_DATA. The
3940 * ICD must initialize VK_LOADER_DATA.loadMagic to ICD_LOADER_MAGIC.
3941 * - The loader implements vkCreate{PLATFORM}SurfaceKHR() and
3942 * vkDestroySurfaceKHR(). The ICD must be capable of working with
3943 * such loader-managed surfaces.
3945 * - Loader interface v2 differs from v1 in:
3946 * - The first ICD entrypoint called by the loader is
3947 * vk_icdNegotiateLoaderICDInterfaceVersion(). The ICD must
3948 * statically expose this entrypoint.
3950 * - Loader interface v3 differs from v2 in:
3951 * - The ICD must implement vkCreate{PLATFORM}SurfaceKHR(),
3952 * vkDestroySurfaceKHR(), and other API which uses VKSurfaceKHR,
3953 * because the loader no longer does so.
3955 *pSupportedVersion
= MIN2(*pSupportedVersion
, 3u);