2 * Copyright © 2016 Red Hat.
3 * Copyright © 2016 Bas Nieuwenhuizen
5 * based in part on anv driver which is:
6 * Copyright © 2015 Intel Corporation
8 * Permission is hereby granted, free of charge, to any person obtaining a
9 * copy of this software and associated documentation files (the "Software"),
10 * to deal in the Software without restriction, including without limitation
11 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
12 * and/or sell copies of the Software, and to permit persons to whom the
13 * Software is furnished to do so, subject to the following conditions:
15 * The above copyright notice and this permission notice (including the next
16 * paragraph) shall be included in all copies or substantial portions of the
19 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
20 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
21 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
22 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
23 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
24 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
32 #include "radv_debug.h"
33 #include "radv_private.h"
34 #include "radv_shader.h"
36 #include "util/disk_cache.h"
37 #include "util/strtod.h"
41 #include <amdgpu_drm.h>
42 #include "winsys/amdgpu/radv_amdgpu_winsys_public.h"
43 #include "ac_llvm_util.h"
44 #include "vk_format.h"
47 #include "util/build_id.h"
48 #include "util/debug.h"
49 #include "util/mesa-sha1.h"
50 #include "compiler/glsl_types.h"
51 #include "util/xmlpool.h"
54 radv_device_get_cache_uuid(enum radeon_family family
, void *uuid
)
57 unsigned char sha1
[20];
58 unsigned ptr_size
= sizeof(void*);
60 memset(uuid
, 0, VK_UUID_SIZE
);
61 _mesa_sha1_init(&ctx
);
63 if (!disk_cache_get_function_identifier(radv_device_get_cache_uuid
, &ctx
) ||
64 !disk_cache_get_function_identifier(LLVMInitializeAMDGPUTargetInfo
, &ctx
))
67 _mesa_sha1_update(&ctx
, &family
, sizeof(family
));
68 _mesa_sha1_update(&ctx
, &ptr_size
, sizeof(ptr_size
));
69 _mesa_sha1_final(&ctx
, sha1
);
71 memcpy(uuid
, sha1
, VK_UUID_SIZE
);
76 radv_get_driver_uuid(void *uuid
)
78 ac_compute_driver_uuid(uuid
, VK_UUID_SIZE
);
82 radv_get_device_uuid(struct radeon_info
*info
, void *uuid
)
84 ac_compute_device_uuid(info
, uuid
, VK_UUID_SIZE
);
88 radv_get_device_name(enum radeon_family family
, char *name
, size_t name_len
)
90 const char *chip_string
;
93 case CHIP_TAHITI
: chip_string
= "AMD RADV TAHITI"; break;
94 case CHIP_PITCAIRN
: chip_string
= "AMD RADV PITCAIRN"; break;
95 case CHIP_VERDE
: chip_string
= "AMD RADV CAPE VERDE"; break;
96 case CHIP_OLAND
: chip_string
= "AMD RADV OLAND"; break;
97 case CHIP_HAINAN
: chip_string
= "AMD RADV HAINAN"; break;
98 case CHIP_BONAIRE
: chip_string
= "AMD RADV BONAIRE"; break;
99 case CHIP_KAVERI
: chip_string
= "AMD RADV KAVERI"; break;
100 case CHIP_KABINI
: chip_string
= "AMD RADV KABINI"; break;
101 case CHIP_HAWAII
: chip_string
= "AMD RADV HAWAII"; break;
102 case CHIP_TONGA
: chip_string
= "AMD RADV TONGA"; break;
103 case CHIP_ICELAND
: chip_string
= "AMD RADV ICELAND"; break;
104 case CHIP_CARRIZO
: chip_string
= "AMD RADV CARRIZO"; break;
105 case CHIP_FIJI
: chip_string
= "AMD RADV FIJI"; break;
106 case CHIP_POLARIS10
: chip_string
= "AMD RADV POLARIS10"; break;
107 case CHIP_POLARIS11
: chip_string
= "AMD RADV POLARIS11"; break;
108 case CHIP_POLARIS12
: chip_string
= "AMD RADV POLARIS12"; break;
109 case CHIP_STONEY
: chip_string
= "AMD RADV STONEY"; break;
110 case CHIP_VEGAM
: chip_string
= "AMD RADV VEGA M"; break;
111 case CHIP_VEGA10
: chip_string
= "AMD RADV VEGA10"; break;
112 case CHIP_VEGA12
: chip_string
= "AMD RADV VEGA12"; break;
113 case CHIP_VEGA20
: chip_string
= "AMD RADV VEGA20"; break;
114 case CHIP_RAVEN
: chip_string
= "AMD RADV RAVEN"; break;
115 case CHIP_RAVEN2
: chip_string
= "AMD RADV RAVEN2"; break;
116 case CHIP_NAVI10
: chip_string
= "AMD RADV NAVI10"; break;
117 case CHIP_NAVI12
: chip_string
= "AMD RADV NAVI12"; break;
118 case CHIP_NAVI14
: chip_string
= "AMD RADV NAVI14"; break;
119 default: chip_string
= "AMD RADV unknown"; break;
122 snprintf(name
, name_len
, "%s (LLVM " MESA_LLVM_VERSION_STRING
")", chip_string
);
126 radv_get_visible_vram_size(struct radv_physical_device
*device
)
128 return MIN2(device
->rad_info
.vram_size
, device
->rad_info
.vram_vis_size
);
132 radv_get_vram_size(struct radv_physical_device
*device
)
134 return device
->rad_info
.vram_size
- radv_get_visible_vram_size(device
);
138 radv_physical_device_init_mem_types(struct radv_physical_device
*device
)
140 STATIC_ASSERT(RADV_MEM_HEAP_COUNT
<= VK_MAX_MEMORY_HEAPS
);
141 uint64_t visible_vram_size
= radv_get_visible_vram_size(device
);
142 uint64_t vram_size
= radv_get_vram_size(device
);
143 int vram_index
= -1, visible_vram_index
= -1, gart_index
= -1;
144 device
->memory_properties
.memoryHeapCount
= 0;
146 vram_index
= device
->memory_properties
.memoryHeapCount
++;
147 device
->memory_properties
.memoryHeaps
[vram_index
] = (VkMemoryHeap
) {
149 .flags
= VK_MEMORY_HEAP_DEVICE_LOCAL_BIT
,
152 if (visible_vram_size
) {
153 visible_vram_index
= device
->memory_properties
.memoryHeapCount
++;
154 device
->memory_properties
.memoryHeaps
[visible_vram_index
] = (VkMemoryHeap
) {
155 .size
= visible_vram_size
,
156 .flags
= VK_MEMORY_HEAP_DEVICE_LOCAL_BIT
,
159 if (device
->rad_info
.gart_size
> 0) {
160 gart_index
= device
->memory_properties
.memoryHeapCount
++;
161 device
->memory_properties
.memoryHeaps
[gart_index
] = (VkMemoryHeap
) {
162 .size
= device
->rad_info
.gart_size
,
163 .flags
= device
->rad_info
.has_dedicated_vram
? 0 : VK_MEMORY_HEAP_DEVICE_LOCAL_BIT
,
167 STATIC_ASSERT(RADV_MEM_TYPE_COUNT
<= VK_MAX_MEMORY_TYPES
);
168 unsigned type_count
= 0;
169 if (vram_index
>= 0) {
170 device
->mem_type_indices
[type_count
] = RADV_MEM_TYPE_VRAM
;
171 device
->memory_properties
.memoryTypes
[type_count
++] = (VkMemoryType
) {
172 .propertyFlags
= VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT
,
173 .heapIndex
= vram_index
,
176 if (gart_index
>= 0) {
177 device
->mem_type_indices
[type_count
] = RADV_MEM_TYPE_GTT_WRITE_COMBINE
;
178 device
->memory_properties
.memoryTypes
[type_count
++] = (VkMemoryType
) {
179 .propertyFlags
= VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT
|
180 VK_MEMORY_PROPERTY_HOST_COHERENT_BIT
|
181 (device
->rad_info
.has_dedicated_vram
? 0 : VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT
),
182 .heapIndex
= gart_index
,
185 if (visible_vram_index
>= 0) {
186 device
->mem_type_indices
[type_count
] = RADV_MEM_TYPE_VRAM_CPU_ACCESS
;
187 device
->memory_properties
.memoryTypes
[type_count
++] = (VkMemoryType
) {
188 .propertyFlags
= VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT
|
189 VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT
|
190 VK_MEMORY_PROPERTY_HOST_COHERENT_BIT
,
191 .heapIndex
= visible_vram_index
,
194 if (gart_index
>= 0) {
195 device
->mem_type_indices
[type_count
] = RADV_MEM_TYPE_GTT_CACHED
;
196 device
->memory_properties
.memoryTypes
[type_count
++] = (VkMemoryType
) {
197 .propertyFlags
= VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT
|
198 VK_MEMORY_PROPERTY_HOST_COHERENT_BIT
|
199 VK_MEMORY_PROPERTY_HOST_CACHED_BIT
|
200 (device
->rad_info
.has_dedicated_vram
? 0 : VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT
),
201 .heapIndex
= gart_index
,
204 device
->memory_properties
.memoryTypeCount
= type_count
;
208 radv_handle_env_var_force_family(struct radv_physical_device
*device
)
210 const char *family
= getenv("RADV_FORCE_FAMILY");
216 for (i
= CHIP_TAHITI
; i
< CHIP_LAST
; i
++) {
217 if (!strcmp(family
, ac_get_llvm_processor_name(i
))) {
218 /* Override family and chip_class. */
219 device
->rad_info
.family
= i
;
221 if (i
>= CHIP_NAVI10
)
222 device
->rad_info
.chip_class
= GFX10
;
223 else if (i
>= CHIP_VEGA10
)
224 device
->rad_info
.chip_class
= GFX9
;
225 else if (i
>= CHIP_TONGA
)
226 device
->rad_info
.chip_class
= GFX8
;
227 else if (i
>= CHIP_BONAIRE
)
228 device
->rad_info
.chip_class
= GFX7
;
230 device
->rad_info
.chip_class
= GFX6
;
236 fprintf(stderr
, "radv: Unknown family: %s\n", family
);
241 radv_physical_device_init(struct radv_physical_device
*device
,
242 struct radv_instance
*instance
,
243 drmDevicePtr drm_device
)
245 const char *path
= drm_device
->nodes
[DRM_NODE_RENDER
];
247 drmVersionPtr version
;
251 fd
= open(path
, O_RDWR
| O_CLOEXEC
);
253 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
254 radv_logi("Could not open device '%s'", path
);
256 return vk_error(instance
, VK_ERROR_INCOMPATIBLE_DRIVER
);
259 version
= drmGetVersion(fd
);
263 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
264 radv_logi("Could not get the kernel driver version for device '%s'", path
);
266 return vk_errorf(instance
, VK_ERROR_INCOMPATIBLE_DRIVER
,
267 "failed to get version %s: %m", path
);
270 if (strcmp(version
->name
, "amdgpu")) {
271 drmFreeVersion(version
);
274 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
275 radv_logi("Device '%s' is not using the amdgpu kernel driver.", path
);
277 return VK_ERROR_INCOMPATIBLE_DRIVER
;
279 drmFreeVersion(version
);
281 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
282 radv_logi("Found compatible device '%s'.", path
);
284 device
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
285 device
->instance
= instance
;
287 device
->ws
= radv_amdgpu_winsys_create(fd
, instance
->debug_flags
,
288 instance
->perftest_flags
);
290 result
= vk_error(instance
, VK_ERROR_INCOMPATIBLE_DRIVER
);
294 if (instance
->enabled_extensions
.KHR_display
) {
295 master_fd
= open(drm_device
->nodes
[DRM_NODE_PRIMARY
], O_RDWR
| O_CLOEXEC
);
296 if (master_fd
>= 0) {
297 uint32_t accel_working
= 0;
298 struct drm_amdgpu_info request
= {
299 .return_pointer
= (uintptr_t)&accel_working
,
300 .return_size
= sizeof(accel_working
),
301 .query
= AMDGPU_INFO_ACCEL_WORKING
304 if (drmCommandWrite(master_fd
, DRM_AMDGPU_INFO
, &request
, sizeof (struct drm_amdgpu_info
)) < 0 || !accel_working
) {
311 device
->master_fd
= master_fd
;
312 device
->local_fd
= fd
;
313 device
->ws
->query_info(device
->ws
, &device
->rad_info
);
315 radv_handle_env_var_force_family(device
);
317 radv_get_device_name(device
->rad_info
.family
, device
->name
, sizeof(device
->name
));
319 if (radv_device_get_cache_uuid(device
->rad_info
.family
, device
->cache_uuid
)) {
320 device
->ws
->destroy(device
->ws
);
321 result
= vk_errorf(instance
, VK_ERROR_INITIALIZATION_FAILED
,
322 "cannot generate UUID");
326 /* These flags affect shader compilation. */
327 uint64_t shader_env_flags
=
328 (device
->instance
->perftest_flags
& RADV_PERFTEST_SISCHED
? 0x1 : 0) |
329 (device
->instance
->debug_flags
& RADV_DEBUG_UNSAFE_MATH
? 0x2 : 0);
331 /* The gpu id is already embedded in the uuid so we just pass "radv"
332 * when creating the cache.
334 char buf
[VK_UUID_SIZE
* 2 + 1];
335 disk_cache_format_hex_id(buf
, device
->cache_uuid
, VK_UUID_SIZE
* 2);
336 device
->disk_cache
= disk_cache_create(device
->name
, buf
, shader_env_flags
);
338 if (device
->rad_info
.chip_class
< GFX8
||
339 device
->rad_info
.chip_class
> GFX9
)
340 fprintf(stderr
, "WARNING: radv is not a conformant vulkan implementation, testing use only.\n");
342 radv_get_driver_uuid(&device
->driver_uuid
);
343 radv_get_device_uuid(&device
->rad_info
, &device
->device_uuid
);
345 if (device
->rad_info
.family
== CHIP_STONEY
||
346 device
->rad_info
.chip_class
>= GFX9
) {
347 device
->has_rbplus
= true;
348 device
->rbplus_allowed
= device
->rad_info
.family
== CHIP_STONEY
||
349 device
->rad_info
.family
== CHIP_VEGA12
||
350 device
->rad_info
.family
== CHIP_RAVEN
||
351 device
->rad_info
.family
== CHIP_RAVEN2
;
354 /* The mere presence of CLEAR_STATE in the IB causes random GPU hangs
357 device
->has_clear_state
= device
->rad_info
.chip_class
>= GFX7
&&
358 device
->rad_info
.chip_class
<= GFX9
;
360 device
->cpdma_prefetch_writes_memory
= device
->rad_info
.chip_class
<= GFX8
;
362 /* Vega10/Raven need a special workaround for a hardware bug. */
363 device
->has_scissor_bug
= device
->rad_info
.family
== CHIP_VEGA10
||
364 device
->rad_info
.family
== CHIP_RAVEN
;
366 device
->has_tc_compat_zrange_bug
= device
->rad_info
.chip_class
< GFX10
;
368 /* Out-of-order primitive rasterization. */
369 device
->has_out_of_order_rast
= device
->rad_info
.chip_class
>= GFX8
&&
370 device
->rad_info
.max_se
>= 2;
371 device
->out_of_order_rast_allowed
= device
->has_out_of_order_rast
&&
372 !(device
->instance
->debug_flags
& RADV_DEBUG_NO_OUT_OF_ORDER
);
374 device
->dcc_msaa_allowed
=
375 (device
->instance
->perftest_flags
& RADV_PERFTEST_DCC_MSAA
);
377 /* TODO: Figure out how to use LOAD_CONTEXT_REG on GFX6-GFX7. */
378 device
->has_load_ctx_reg_pkt
= device
->rad_info
.chip_class
>= GFX9
||
379 (device
->rad_info
.chip_class
>= GFX8
&&
380 device
->rad_info
.me_fw_feature
>= 41);
382 device
->has_dcc_constant_encode
= device
->rad_info
.family
== CHIP_RAVEN2
||
383 device
->rad_info
.chip_class
>= GFX10
;
385 device
->use_shader_ballot
= device
->instance
->perftest_flags
& RADV_PERFTEST_SHADER_BALLOT
;
387 radv_physical_device_init_mem_types(device
);
388 radv_fill_device_extension_table(device
, &device
->supported_extensions
);
390 device
->bus_info
= *drm_device
->businfo
.pci
;
392 if ((device
->instance
->debug_flags
& RADV_DEBUG_INFO
))
393 ac_print_gpu_info(&device
->rad_info
);
395 /* The WSI is structured as a layer on top of the driver, so this has
396 * to be the last part of initialization (at least until we get other
399 result
= radv_init_wsi(device
);
400 if (result
!= VK_SUCCESS
) {
401 device
->ws
->destroy(device
->ws
);
402 vk_error(instance
, result
);
416 radv_physical_device_finish(struct radv_physical_device
*device
)
418 radv_finish_wsi(device
);
419 device
->ws
->destroy(device
->ws
);
420 disk_cache_destroy(device
->disk_cache
);
421 close(device
->local_fd
);
422 if (device
->master_fd
!= -1)
423 close(device
->master_fd
);
427 default_alloc_func(void *pUserData
, size_t size
, size_t align
,
428 VkSystemAllocationScope allocationScope
)
434 default_realloc_func(void *pUserData
, void *pOriginal
, size_t size
,
435 size_t align
, VkSystemAllocationScope allocationScope
)
437 return realloc(pOriginal
, size
);
441 default_free_func(void *pUserData
, void *pMemory
)
446 static const VkAllocationCallbacks default_alloc
= {
448 .pfnAllocation
= default_alloc_func
,
449 .pfnReallocation
= default_realloc_func
,
450 .pfnFree
= default_free_func
,
453 static const struct debug_control radv_debug_options
[] = {
454 {"nofastclears", RADV_DEBUG_NO_FAST_CLEARS
},
455 {"nodcc", RADV_DEBUG_NO_DCC
},
456 {"shaders", RADV_DEBUG_DUMP_SHADERS
},
457 {"nocache", RADV_DEBUG_NO_CACHE
},
458 {"shaderstats", RADV_DEBUG_DUMP_SHADER_STATS
},
459 {"nohiz", RADV_DEBUG_NO_HIZ
},
460 {"nocompute", RADV_DEBUG_NO_COMPUTE_QUEUE
},
461 {"unsafemath", RADV_DEBUG_UNSAFE_MATH
},
462 {"allbos", RADV_DEBUG_ALL_BOS
},
463 {"noibs", RADV_DEBUG_NO_IBS
},
464 {"spirv", RADV_DEBUG_DUMP_SPIRV
},
465 {"vmfaults", RADV_DEBUG_VM_FAULTS
},
466 {"zerovram", RADV_DEBUG_ZERO_VRAM
},
467 {"syncshaders", RADV_DEBUG_SYNC_SHADERS
},
468 {"nosisched", RADV_DEBUG_NO_SISCHED
},
469 {"preoptir", RADV_DEBUG_PREOPTIR
},
470 {"nodynamicbounds", RADV_DEBUG_NO_DYNAMIC_BOUNDS
},
471 {"nooutoforder", RADV_DEBUG_NO_OUT_OF_ORDER
},
472 {"info", RADV_DEBUG_INFO
},
473 {"errors", RADV_DEBUG_ERRORS
},
474 {"startup", RADV_DEBUG_STARTUP
},
475 {"checkir", RADV_DEBUG_CHECKIR
},
476 {"nothreadllvm", RADV_DEBUG_NOTHREADLLVM
},
477 {"nobinning", RADV_DEBUG_NOBINNING
},
478 {"noloadstoreopt", RADV_DEBUG_NO_LOAD_STORE_OPT
},
479 {"nongg", RADV_DEBUG_NO_NGG
},
484 radv_get_debug_option_name(int id
)
486 assert(id
< ARRAY_SIZE(radv_debug_options
) - 1);
487 return radv_debug_options
[id
].string
;
490 static const struct debug_control radv_perftest_options
[] = {
491 {"nobatchchain", RADV_PERFTEST_NO_BATCHCHAIN
},
492 {"sisched", RADV_PERFTEST_SISCHED
},
493 {"localbos", RADV_PERFTEST_LOCAL_BOS
},
494 {"dccmsaa", RADV_PERFTEST_DCC_MSAA
},
495 {"bolist", RADV_PERFTEST_BO_LIST
},
496 {"shader_ballot", RADV_PERFTEST_SHADER_BALLOT
},
497 {"tccompatcmask", RADV_PERFTEST_TC_COMPAT_CMASK
},
502 radv_get_perftest_option_name(int id
)
504 assert(id
< ARRAY_SIZE(radv_perftest_options
) - 1);
505 return radv_perftest_options
[id
].string
;
509 radv_handle_per_app_options(struct radv_instance
*instance
,
510 const VkApplicationInfo
*info
)
512 const char *name
= info
? info
->pApplicationName
: NULL
;
517 if (!strcmp(name
, "Talos - Linux - 32bit") ||
518 !strcmp(name
, "Talos - Linux - 64bit")) {
519 if (!(instance
->debug_flags
& RADV_DEBUG_NO_SISCHED
)) {
520 /* Force enable LLVM sisched for Talos because it looks
521 * safe and it gives few more FPS.
523 instance
->perftest_flags
|= RADV_PERFTEST_SISCHED
;
525 } else if (!strcmp(name
, "DOOM_VFR")) {
526 /* Work around a Doom VFR game bug */
527 instance
->debug_flags
|= RADV_DEBUG_NO_DYNAMIC_BOUNDS
;
528 } else if (!strcmp(name
, "MonsterHunterWorld.exe")) {
529 /* Workaround for a WaW hazard when LLVM moves/merges
530 * load/store memory operations.
531 * See https://reviews.llvm.org/D61313
533 if (HAVE_LLVM
< 0x900)
534 instance
->debug_flags
|= RADV_DEBUG_NO_LOAD_STORE_OPT
;
538 static int radv_get_instance_extension_index(const char *name
)
540 for (unsigned i
= 0; i
< RADV_INSTANCE_EXTENSION_COUNT
; ++i
) {
541 if (strcmp(name
, radv_instance_extensions
[i
].extensionName
) == 0)
547 static const char radv_dri_options_xml
[] =
549 DRI_CONF_SECTION_QUALITY
550 DRI_CONF_ADAPTIVE_SYNC("true")
554 static void radv_init_dri_options(struct radv_instance
*instance
)
556 driParseOptionInfo(&instance
->available_dri_options
, radv_dri_options_xml
);
557 driParseConfigFiles(&instance
->dri_options
,
558 &instance
->available_dri_options
,
562 VkResult
radv_CreateInstance(
563 const VkInstanceCreateInfo
* pCreateInfo
,
564 const VkAllocationCallbacks
* pAllocator
,
565 VkInstance
* pInstance
)
567 struct radv_instance
*instance
;
570 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_INSTANCE_CREATE_INFO
);
572 uint32_t client_version
;
573 if (pCreateInfo
->pApplicationInfo
&&
574 pCreateInfo
->pApplicationInfo
->apiVersion
!= 0) {
575 client_version
= pCreateInfo
->pApplicationInfo
->apiVersion
;
577 client_version
= VK_API_VERSION_1_0
;
580 instance
= vk_zalloc2(&default_alloc
, pAllocator
, sizeof(*instance
), 8,
581 VK_SYSTEM_ALLOCATION_SCOPE_INSTANCE
);
583 return vk_error(NULL
, VK_ERROR_OUT_OF_HOST_MEMORY
);
585 instance
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
588 instance
->alloc
= *pAllocator
;
590 instance
->alloc
= default_alloc
;
592 instance
->apiVersion
= client_version
;
593 instance
->physicalDeviceCount
= -1;
595 instance
->debug_flags
= parse_debug_string(getenv("RADV_DEBUG"),
598 instance
->perftest_flags
= parse_debug_string(getenv("RADV_PERFTEST"),
599 radv_perftest_options
);
602 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
603 radv_logi("Created an instance");
605 for (uint32_t i
= 0; i
< pCreateInfo
->enabledExtensionCount
; i
++) {
606 const char *ext_name
= pCreateInfo
->ppEnabledExtensionNames
[i
];
607 int index
= radv_get_instance_extension_index(ext_name
);
609 if (index
< 0 || !radv_supported_instance_extensions
.extensions
[index
]) {
610 vk_free2(&default_alloc
, pAllocator
, instance
);
611 return vk_error(instance
, VK_ERROR_EXTENSION_NOT_PRESENT
);
614 instance
->enabled_extensions
.extensions
[index
] = true;
617 result
= vk_debug_report_instance_init(&instance
->debug_report_callbacks
);
618 if (result
!= VK_SUCCESS
) {
619 vk_free2(&default_alloc
, pAllocator
, instance
);
620 return vk_error(instance
, result
);
624 glsl_type_singleton_init_or_ref();
626 VG(VALGRIND_CREATE_MEMPOOL(instance
, 0, false));
628 radv_init_dri_options(instance
);
629 radv_handle_per_app_options(instance
, pCreateInfo
->pApplicationInfo
);
631 *pInstance
= radv_instance_to_handle(instance
);
636 void radv_DestroyInstance(
637 VkInstance _instance
,
638 const VkAllocationCallbacks
* pAllocator
)
640 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
645 for (int i
= 0; i
< instance
->physicalDeviceCount
; ++i
) {
646 radv_physical_device_finish(instance
->physicalDevices
+ i
);
649 VG(VALGRIND_DESTROY_MEMPOOL(instance
));
651 glsl_type_singleton_decref();
654 driDestroyOptionCache(&instance
->dri_options
);
655 driDestroyOptionInfo(&instance
->available_dri_options
);
657 vk_debug_report_instance_destroy(&instance
->debug_report_callbacks
);
659 vk_free(&instance
->alloc
, instance
);
663 radv_enumerate_devices(struct radv_instance
*instance
)
665 /* TODO: Check for more devices ? */
666 drmDevicePtr devices
[8];
667 VkResult result
= VK_ERROR_INCOMPATIBLE_DRIVER
;
670 instance
->physicalDeviceCount
= 0;
672 max_devices
= drmGetDevices2(0, devices
, ARRAY_SIZE(devices
));
674 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
675 radv_logi("Found %d drm nodes", max_devices
);
678 return vk_error(instance
, VK_ERROR_INCOMPATIBLE_DRIVER
);
680 for (unsigned i
= 0; i
< (unsigned)max_devices
; i
++) {
681 if (devices
[i
]->available_nodes
& 1 << DRM_NODE_RENDER
&&
682 devices
[i
]->bustype
== DRM_BUS_PCI
&&
683 devices
[i
]->deviceinfo
.pci
->vendor_id
== ATI_VENDOR_ID
) {
685 result
= radv_physical_device_init(instance
->physicalDevices
+
686 instance
->physicalDeviceCount
,
689 if (result
== VK_SUCCESS
)
690 ++instance
->physicalDeviceCount
;
691 else if (result
!= VK_ERROR_INCOMPATIBLE_DRIVER
)
695 drmFreeDevices(devices
, max_devices
);
700 VkResult
radv_EnumeratePhysicalDevices(
701 VkInstance _instance
,
702 uint32_t* pPhysicalDeviceCount
,
703 VkPhysicalDevice
* pPhysicalDevices
)
705 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
708 if (instance
->physicalDeviceCount
< 0) {
709 result
= radv_enumerate_devices(instance
);
710 if (result
!= VK_SUCCESS
&&
711 result
!= VK_ERROR_INCOMPATIBLE_DRIVER
)
715 if (!pPhysicalDevices
) {
716 *pPhysicalDeviceCount
= instance
->physicalDeviceCount
;
718 *pPhysicalDeviceCount
= MIN2(*pPhysicalDeviceCount
, instance
->physicalDeviceCount
);
719 for (unsigned i
= 0; i
< *pPhysicalDeviceCount
; ++i
)
720 pPhysicalDevices
[i
] = radv_physical_device_to_handle(instance
->physicalDevices
+ i
);
723 return *pPhysicalDeviceCount
< instance
->physicalDeviceCount
? VK_INCOMPLETE
727 VkResult
radv_EnumeratePhysicalDeviceGroups(
728 VkInstance _instance
,
729 uint32_t* pPhysicalDeviceGroupCount
,
730 VkPhysicalDeviceGroupProperties
* pPhysicalDeviceGroupProperties
)
732 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
735 if (instance
->physicalDeviceCount
< 0) {
736 result
= radv_enumerate_devices(instance
);
737 if (result
!= VK_SUCCESS
&&
738 result
!= VK_ERROR_INCOMPATIBLE_DRIVER
)
742 if (!pPhysicalDeviceGroupProperties
) {
743 *pPhysicalDeviceGroupCount
= instance
->physicalDeviceCount
;
745 *pPhysicalDeviceGroupCount
= MIN2(*pPhysicalDeviceGroupCount
, instance
->physicalDeviceCount
);
746 for (unsigned i
= 0; i
< *pPhysicalDeviceGroupCount
; ++i
) {
747 pPhysicalDeviceGroupProperties
[i
].physicalDeviceCount
= 1;
748 pPhysicalDeviceGroupProperties
[i
].physicalDevices
[0] = radv_physical_device_to_handle(instance
->physicalDevices
+ i
);
749 pPhysicalDeviceGroupProperties
[i
].subsetAllocation
= false;
752 return *pPhysicalDeviceGroupCount
< instance
->physicalDeviceCount
? VK_INCOMPLETE
756 void radv_GetPhysicalDeviceFeatures(
757 VkPhysicalDevice physicalDevice
,
758 VkPhysicalDeviceFeatures
* pFeatures
)
760 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
761 memset(pFeatures
, 0, sizeof(*pFeatures
));
763 *pFeatures
= (VkPhysicalDeviceFeatures
) {
764 .robustBufferAccess
= true,
765 .fullDrawIndexUint32
= true,
766 .imageCubeArray
= true,
767 .independentBlend
= true,
768 .geometryShader
= true,
769 .tessellationShader
= true,
770 .sampleRateShading
= true,
771 .dualSrcBlend
= true,
773 .multiDrawIndirect
= true,
774 .drawIndirectFirstInstance
= true,
776 .depthBiasClamp
= true,
777 .fillModeNonSolid
= true,
782 .multiViewport
= true,
783 .samplerAnisotropy
= true,
784 .textureCompressionETC2
= radv_device_supports_etc(pdevice
),
785 .textureCompressionASTC_LDR
= false,
786 .textureCompressionBC
= true,
787 .occlusionQueryPrecise
= true,
788 .pipelineStatisticsQuery
= true,
789 .vertexPipelineStoresAndAtomics
= true,
790 .fragmentStoresAndAtomics
= true,
791 .shaderTessellationAndGeometryPointSize
= true,
792 .shaderImageGatherExtended
= true,
793 .shaderStorageImageExtendedFormats
= true,
794 .shaderStorageImageMultisample
= pdevice
->rad_info
.chip_class
>= GFX8
,
795 .shaderUniformBufferArrayDynamicIndexing
= true,
796 .shaderSampledImageArrayDynamicIndexing
= true,
797 .shaderStorageBufferArrayDynamicIndexing
= true,
798 .shaderStorageImageArrayDynamicIndexing
= true,
799 .shaderStorageImageReadWithoutFormat
= true,
800 .shaderStorageImageWriteWithoutFormat
= true,
801 .shaderClipDistance
= true,
802 .shaderCullDistance
= true,
803 .shaderFloat64
= true,
805 .shaderInt16
= pdevice
->rad_info
.chip_class
>= GFX9
,
806 .sparseBinding
= true,
807 .variableMultisampleRate
= true,
808 .inheritedQueries
= true,
812 void radv_GetPhysicalDeviceFeatures2(
813 VkPhysicalDevice physicalDevice
,
814 VkPhysicalDeviceFeatures2
*pFeatures
)
816 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
817 vk_foreach_struct(ext
, pFeatures
->pNext
) {
818 switch (ext
->sType
) {
819 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VARIABLE_POINTERS_FEATURES
: {
820 VkPhysicalDeviceVariablePointersFeatures
*features
= (void *)ext
;
821 features
->variablePointersStorageBuffer
= true;
822 features
->variablePointers
= true;
825 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MULTIVIEW_FEATURES
: {
826 VkPhysicalDeviceMultiviewFeatures
*features
= (VkPhysicalDeviceMultiviewFeatures
*)ext
;
827 features
->multiview
= true;
828 features
->multiviewGeometryShader
= true;
829 features
->multiviewTessellationShader
= true;
832 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_DRAW_PARAMETERS_FEATURES
: {
833 VkPhysicalDeviceShaderDrawParametersFeatures
*features
=
834 (VkPhysicalDeviceShaderDrawParametersFeatures
*)ext
;
835 features
->shaderDrawParameters
= true;
838 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROTECTED_MEMORY_FEATURES
: {
839 VkPhysicalDeviceProtectedMemoryFeatures
*features
=
840 (VkPhysicalDeviceProtectedMemoryFeatures
*)ext
;
841 features
->protectedMemory
= false;
844 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_16BIT_STORAGE_FEATURES
: {
845 VkPhysicalDevice16BitStorageFeatures
*features
=
846 (VkPhysicalDevice16BitStorageFeatures
*)ext
;
847 bool enabled
= pdevice
->rad_info
.chip_class
>= GFX8
;
848 features
->storageBuffer16BitAccess
= enabled
;
849 features
->uniformAndStorageBuffer16BitAccess
= enabled
;
850 features
->storagePushConstant16
= enabled
;
851 features
->storageInputOutput16
= enabled
&& HAVE_LLVM
>= 0x900;
854 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SAMPLER_YCBCR_CONVERSION_FEATURES
: {
855 VkPhysicalDeviceSamplerYcbcrConversionFeatures
*features
=
856 (VkPhysicalDeviceSamplerYcbcrConversionFeatures
*)ext
;
857 features
->samplerYcbcrConversion
= true;
860 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DESCRIPTOR_INDEXING_FEATURES_EXT
: {
861 VkPhysicalDeviceDescriptorIndexingFeaturesEXT
*features
=
862 (VkPhysicalDeviceDescriptorIndexingFeaturesEXT
*)ext
;
863 features
->shaderInputAttachmentArrayDynamicIndexing
= true;
864 features
->shaderUniformTexelBufferArrayDynamicIndexing
= true;
865 features
->shaderStorageTexelBufferArrayDynamicIndexing
= true;
866 features
->shaderUniformBufferArrayNonUniformIndexing
= true;
867 features
->shaderSampledImageArrayNonUniformIndexing
= true;
868 features
->shaderStorageBufferArrayNonUniformIndexing
= true;
869 features
->shaderStorageImageArrayNonUniformIndexing
= true;
870 features
->shaderInputAttachmentArrayNonUniformIndexing
= true;
871 features
->shaderUniformTexelBufferArrayNonUniformIndexing
= true;
872 features
->shaderStorageTexelBufferArrayNonUniformIndexing
= true;
873 features
->descriptorBindingUniformBufferUpdateAfterBind
= true;
874 features
->descriptorBindingSampledImageUpdateAfterBind
= true;
875 features
->descriptorBindingStorageImageUpdateAfterBind
= true;
876 features
->descriptorBindingStorageBufferUpdateAfterBind
= true;
877 features
->descriptorBindingUniformTexelBufferUpdateAfterBind
= true;
878 features
->descriptorBindingStorageTexelBufferUpdateAfterBind
= true;
879 features
->descriptorBindingUpdateUnusedWhilePending
= true;
880 features
->descriptorBindingPartiallyBound
= true;
881 features
->descriptorBindingVariableDescriptorCount
= true;
882 features
->runtimeDescriptorArray
= true;
885 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_CONDITIONAL_RENDERING_FEATURES_EXT
: {
886 VkPhysicalDeviceConditionalRenderingFeaturesEXT
*features
=
887 (VkPhysicalDeviceConditionalRenderingFeaturesEXT
*)ext
;
888 features
->conditionalRendering
= true;
889 features
->inheritedConditionalRendering
= false;
892 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VERTEX_ATTRIBUTE_DIVISOR_FEATURES_EXT
: {
893 VkPhysicalDeviceVertexAttributeDivisorFeaturesEXT
*features
=
894 (VkPhysicalDeviceVertexAttributeDivisorFeaturesEXT
*)ext
;
895 features
->vertexAttributeInstanceRateDivisor
= VK_TRUE
;
896 features
->vertexAttributeInstanceRateZeroDivisor
= VK_TRUE
;
899 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TRANSFORM_FEEDBACK_FEATURES_EXT
: {
900 VkPhysicalDeviceTransformFeedbackFeaturesEXT
*features
=
901 (VkPhysicalDeviceTransformFeedbackFeaturesEXT
*)ext
;
902 features
->transformFeedback
= true;
903 features
->geometryStreams
= true;
906 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SCALAR_BLOCK_LAYOUT_FEATURES_EXT
: {
907 VkPhysicalDeviceScalarBlockLayoutFeaturesEXT
*features
=
908 (VkPhysicalDeviceScalarBlockLayoutFeaturesEXT
*)ext
;
909 features
->scalarBlockLayout
= pdevice
->rad_info
.chip_class
>= GFX7
;
912 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MEMORY_PRIORITY_FEATURES_EXT
: {
913 VkPhysicalDeviceMemoryPriorityFeaturesEXT
*features
=
914 (VkPhysicalDeviceMemoryPriorityFeaturesEXT
*)ext
;
915 features
->memoryPriority
= VK_TRUE
;
918 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_BUFFER_DEVICE_ADDRESS_FEATURES_EXT
: {
919 VkPhysicalDeviceBufferDeviceAddressFeaturesEXT
*features
=
920 (VkPhysicalDeviceBufferDeviceAddressFeaturesEXT
*)ext
;
921 features
->bufferDeviceAddress
= true;
922 features
->bufferDeviceAddressCaptureReplay
= false;
923 features
->bufferDeviceAddressMultiDevice
= false;
926 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DEPTH_CLIP_ENABLE_FEATURES_EXT
: {
927 VkPhysicalDeviceDepthClipEnableFeaturesEXT
*features
=
928 (VkPhysicalDeviceDepthClipEnableFeaturesEXT
*)ext
;
929 features
->depthClipEnable
= true;
932 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_HOST_QUERY_RESET_FEATURES_EXT
: {
933 VkPhysicalDeviceHostQueryResetFeaturesEXT
*features
=
934 (VkPhysicalDeviceHostQueryResetFeaturesEXT
*)ext
;
935 features
->hostQueryReset
= true;
938 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_8BIT_STORAGE_FEATURES_KHR
: {
939 VkPhysicalDevice8BitStorageFeaturesKHR
*features
=
940 (VkPhysicalDevice8BitStorageFeaturesKHR
*)ext
;
941 bool enabled
= pdevice
->rad_info
.chip_class
>= GFX8
;
942 features
->storageBuffer8BitAccess
= enabled
;
943 features
->uniformAndStorageBuffer8BitAccess
= enabled
;
944 features
->storagePushConstant8
= enabled
;
947 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_FLOAT16_INT8_FEATURES_KHR
: {
948 VkPhysicalDeviceFloat16Int8FeaturesKHR
*features
=
949 (VkPhysicalDeviceFloat16Int8FeaturesKHR
*)ext
;
950 features
->shaderFloat16
= pdevice
->rad_info
.chip_class
>= GFX8
&& HAVE_LLVM
>= 0x0800;
951 features
->shaderInt8
= true;
954 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_ATOMIC_INT64_FEATURES_KHR
: {
955 VkPhysicalDeviceShaderAtomicInt64FeaturesKHR
*features
=
956 (VkPhysicalDeviceShaderAtomicInt64FeaturesKHR
*)ext
;
957 /* TODO: Enable this once the driver supports 64-bit
958 * compare&swap atomic operations.
960 features
->shaderBufferInt64Atomics
= false;
961 features
->shaderSharedInt64Atomics
= false;
964 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_INLINE_UNIFORM_BLOCK_FEATURES_EXT
: {
965 VkPhysicalDeviceInlineUniformBlockFeaturesEXT
*features
=
966 (VkPhysicalDeviceInlineUniformBlockFeaturesEXT
*)ext
;
968 features
->inlineUniformBlock
= true;
969 features
->descriptorBindingInlineUniformBlockUpdateAfterBind
= true;
972 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_COMPUTE_SHADER_DERIVATIVES_FEATURES_NV
: {
973 VkPhysicalDeviceComputeShaderDerivativesFeaturesNV
*features
=
974 (VkPhysicalDeviceComputeShaderDerivativesFeaturesNV
*)ext
;
975 features
->computeDerivativeGroupQuads
= false;
976 features
->computeDerivativeGroupLinear
= true;
979 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_YCBCR_IMAGE_ARRAYS_FEATURES_EXT
: {
980 VkPhysicalDeviceYcbcrImageArraysFeaturesEXT
*features
=
981 (VkPhysicalDeviceYcbcrImageArraysFeaturesEXT
*)ext
;
982 features
->ycbcrImageArrays
= true;
985 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_UNIFORM_BUFFER_STANDARD_LAYOUT_FEATURES_KHR
: {
986 VkPhysicalDeviceUniformBufferStandardLayoutFeaturesKHR
*features
=
987 (VkPhysicalDeviceUniformBufferStandardLayoutFeaturesKHR
*)ext
;
988 features
->uniformBufferStandardLayout
= true;
995 return radv_GetPhysicalDeviceFeatures(physicalDevice
, &pFeatures
->features
);
998 void radv_GetPhysicalDeviceProperties(
999 VkPhysicalDevice physicalDevice
,
1000 VkPhysicalDeviceProperties
* pProperties
)
1002 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
1003 VkSampleCountFlags sample_counts
= 0xf;
1005 /* make sure that the entire descriptor set is addressable with a signed
1006 * 32-bit int. So the sum of all limits scaled by descriptor size has to
1007 * be at most 2 GiB. the combined image & samples object count as one of
1008 * both. This limit is for the pipeline layout, not for the set layout, but
1009 * there is no set limit, so we just set a pipeline limit. I don't think
1010 * any app is going to hit this soon. */
1011 size_t max_descriptor_set_size
= ((1ull << 31) - 16 * MAX_DYNAMIC_BUFFERS
) /
1012 (32 /* uniform buffer, 32 due to potential space wasted on alignment */ +
1013 32 /* storage buffer, 32 due to potential space wasted on alignment */ +
1014 32 /* sampler, largest when combined with image */ +
1015 64 /* sampled image */ +
1016 64 /* storage image */);
1018 VkPhysicalDeviceLimits limits
= {
1019 .maxImageDimension1D
= (1 << 14),
1020 .maxImageDimension2D
= (1 << 14),
1021 .maxImageDimension3D
= (1 << 11),
1022 .maxImageDimensionCube
= (1 << 14),
1023 .maxImageArrayLayers
= (1 << 11),
1024 .maxTexelBufferElements
= 128 * 1024 * 1024,
1025 .maxUniformBufferRange
= UINT32_MAX
,
1026 .maxStorageBufferRange
= UINT32_MAX
,
1027 .maxPushConstantsSize
= MAX_PUSH_CONSTANTS_SIZE
,
1028 .maxMemoryAllocationCount
= UINT32_MAX
,
1029 .maxSamplerAllocationCount
= 64 * 1024,
1030 .bufferImageGranularity
= 64, /* A cache line */
1031 .sparseAddressSpaceSize
= 0xffffffffu
, /* buffer max size */
1032 .maxBoundDescriptorSets
= MAX_SETS
,
1033 .maxPerStageDescriptorSamplers
= max_descriptor_set_size
,
1034 .maxPerStageDescriptorUniformBuffers
= max_descriptor_set_size
,
1035 .maxPerStageDescriptorStorageBuffers
= max_descriptor_set_size
,
1036 .maxPerStageDescriptorSampledImages
= max_descriptor_set_size
,
1037 .maxPerStageDescriptorStorageImages
= max_descriptor_set_size
,
1038 .maxPerStageDescriptorInputAttachments
= max_descriptor_set_size
,
1039 .maxPerStageResources
= max_descriptor_set_size
,
1040 .maxDescriptorSetSamplers
= max_descriptor_set_size
,
1041 .maxDescriptorSetUniformBuffers
= max_descriptor_set_size
,
1042 .maxDescriptorSetUniformBuffersDynamic
= MAX_DYNAMIC_UNIFORM_BUFFERS
,
1043 .maxDescriptorSetStorageBuffers
= max_descriptor_set_size
,
1044 .maxDescriptorSetStorageBuffersDynamic
= MAX_DYNAMIC_STORAGE_BUFFERS
,
1045 .maxDescriptorSetSampledImages
= max_descriptor_set_size
,
1046 .maxDescriptorSetStorageImages
= max_descriptor_set_size
,
1047 .maxDescriptorSetInputAttachments
= max_descriptor_set_size
,
1048 .maxVertexInputAttributes
= MAX_VERTEX_ATTRIBS
,
1049 .maxVertexInputBindings
= MAX_VBS
,
1050 .maxVertexInputAttributeOffset
= 2047,
1051 .maxVertexInputBindingStride
= 2048,
1052 .maxVertexOutputComponents
= 128,
1053 .maxTessellationGenerationLevel
= 64,
1054 .maxTessellationPatchSize
= 32,
1055 .maxTessellationControlPerVertexInputComponents
= 128,
1056 .maxTessellationControlPerVertexOutputComponents
= 128,
1057 .maxTessellationControlPerPatchOutputComponents
= 120,
1058 .maxTessellationControlTotalOutputComponents
= 4096,
1059 .maxTessellationEvaluationInputComponents
= 128,
1060 .maxTessellationEvaluationOutputComponents
= 128,
1061 .maxGeometryShaderInvocations
= 127,
1062 .maxGeometryInputComponents
= 64,
1063 .maxGeometryOutputComponents
= 128,
1064 .maxGeometryOutputVertices
= 256,
1065 .maxGeometryTotalOutputComponents
= 1024,
1066 .maxFragmentInputComponents
= 128,
1067 .maxFragmentOutputAttachments
= 8,
1068 .maxFragmentDualSrcAttachments
= 1,
1069 .maxFragmentCombinedOutputResources
= 8,
1070 .maxComputeSharedMemorySize
= 32768,
1071 .maxComputeWorkGroupCount
= { 65535, 65535, 65535 },
1072 .maxComputeWorkGroupInvocations
= 2048,
1073 .maxComputeWorkGroupSize
= {
1078 .subPixelPrecisionBits
= 8,
1079 .subTexelPrecisionBits
= 8,
1080 .mipmapPrecisionBits
= 8,
1081 .maxDrawIndexedIndexValue
= UINT32_MAX
,
1082 .maxDrawIndirectCount
= UINT32_MAX
,
1083 .maxSamplerLodBias
= 16,
1084 .maxSamplerAnisotropy
= 16,
1085 .maxViewports
= MAX_VIEWPORTS
,
1086 .maxViewportDimensions
= { (1 << 14), (1 << 14) },
1087 .viewportBoundsRange
= { INT16_MIN
, INT16_MAX
},
1088 .viewportSubPixelBits
= 8,
1089 .minMemoryMapAlignment
= 4096, /* A page */
1090 .minTexelBufferOffsetAlignment
= 1,
1091 .minUniformBufferOffsetAlignment
= 4,
1092 .minStorageBufferOffsetAlignment
= 4,
1093 .minTexelOffset
= -32,
1094 .maxTexelOffset
= 31,
1095 .minTexelGatherOffset
= -32,
1096 .maxTexelGatherOffset
= 31,
1097 .minInterpolationOffset
= -2,
1098 .maxInterpolationOffset
= 2,
1099 .subPixelInterpolationOffsetBits
= 8,
1100 .maxFramebufferWidth
= (1 << 14),
1101 .maxFramebufferHeight
= (1 << 14),
1102 .maxFramebufferLayers
= (1 << 10),
1103 .framebufferColorSampleCounts
= sample_counts
,
1104 .framebufferDepthSampleCounts
= sample_counts
,
1105 .framebufferStencilSampleCounts
= sample_counts
,
1106 .framebufferNoAttachmentsSampleCounts
= sample_counts
,
1107 .maxColorAttachments
= MAX_RTS
,
1108 .sampledImageColorSampleCounts
= sample_counts
,
1109 .sampledImageIntegerSampleCounts
= VK_SAMPLE_COUNT_1_BIT
,
1110 .sampledImageDepthSampleCounts
= sample_counts
,
1111 .sampledImageStencilSampleCounts
= sample_counts
,
1112 .storageImageSampleCounts
= pdevice
->rad_info
.chip_class
>= GFX8
? sample_counts
: VK_SAMPLE_COUNT_1_BIT
,
1113 .maxSampleMaskWords
= 1,
1114 .timestampComputeAndGraphics
= true,
1115 .timestampPeriod
= 1000000.0 / pdevice
->rad_info
.clock_crystal_freq
,
1116 .maxClipDistances
= 8,
1117 .maxCullDistances
= 8,
1118 .maxCombinedClipAndCullDistances
= 8,
1119 .discreteQueuePriorities
= 2,
1120 .pointSizeRange
= { 0.0, 8192.0 },
1121 .lineWidthRange
= { 0.0, 7.9921875 },
1122 .pointSizeGranularity
= (1.0 / 8.0),
1123 .lineWidthGranularity
= (1.0 / 128.0),
1124 .strictLines
= false, /* FINISHME */
1125 .standardSampleLocations
= true,
1126 .optimalBufferCopyOffsetAlignment
= 128,
1127 .optimalBufferCopyRowPitchAlignment
= 128,
1128 .nonCoherentAtomSize
= 64,
1131 *pProperties
= (VkPhysicalDeviceProperties
) {
1132 .apiVersion
= radv_physical_device_api_version(pdevice
),
1133 .driverVersion
= vk_get_driver_version(),
1134 .vendorID
= ATI_VENDOR_ID
,
1135 .deviceID
= pdevice
->rad_info
.pci_id
,
1136 .deviceType
= pdevice
->rad_info
.has_dedicated_vram
? VK_PHYSICAL_DEVICE_TYPE_DISCRETE_GPU
: VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU
,
1138 .sparseProperties
= {0},
1141 strcpy(pProperties
->deviceName
, pdevice
->name
);
1142 memcpy(pProperties
->pipelineCacheUUID
, pdevice
->cache_uuid
, VK_UUID_SIZE
);
1145 void radv_GetPhysicalDeviceProperties2(
1146 VkPhysicalDevice physicalDevice
,
1147 VkPhysicalDeviceProperties2
*pProperties
)
1149 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
1150 radv_GetPhysicalDeviceProperties(physicalDevice
, &pProperties
->properties
);
1152 vk_foreach_struct(ext
, pProperties
->pNext
) {
1153 switch (ext
->sType
) {
1154 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PUSH_DESCRIPTOR_PROPERTIES_KHR
: {
1155 VkPhysicalDevicePushDescriptorPropertiesKHR
*properties
=
1156 (VkPhysicalDevicePushDescriptorPropertiesKHR
*) ext
;
1157 properties
->maxPushDescriptors
= MAX_PUSH_DESCRIPTORS
;
1160 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_ID_PROPERTIES
: {
1161 VkPhysicalDeviceIDProperties
*properties
= (VkPhysicalDeviceIDProperties
*)ext
;
1162 memcpy(properties
->driverUUID
, pdevice
->driver_uuid
, VK_UUID_SIZE
);
1163 memcpy(properties
->deviceUUID
, pdevice
->device_uuid
, VK_UUID_SIZE
);
1164 properties
->deviceLUIDValid
= false;
1167 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MULTIVIEW_PROPERTIES
: {
1168 VkPhysicalDeviceMultiviewProperties
*properties
= (VkPhysicalDeviceMultiviewProperties
*)ext
;
1169 properties
->maxMultiviewViewCount
= MAX_VIEWS
;
1170 properties
->maxMultiviewInstanceIndex
= INT_MAX
;
1173 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_POINT_CLIPPING_PROPERTIES
: {
1174 VkPhysicalDevicePointClippingProperties
*properties
=
1175 (VkPhysicalDevicePointClippingProperties
*)ext
;
1176 properties
->pointClippingBehavior
= VK_POINT_CLIPPING_BEHAVIOR_ALL_CLIP_PLANES
;
1179 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DISCARD_RECTANGLE_PROPERTIES_EXT
: {
1180 VkPhysicalDeviceDiscardRectanglePropertiesEXT
*properties
=
1181 (VkPhysicalDeviceDiscardRectanglePropertiesEXT
*)ext
;
1182 properties
->maxDiscardRectangles
= MAX_DISCARD_RECTANGLES
;
1185 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_EXTERNAL_MEMORY_HOST_PROPERTIES_EXT
: {
1186 VkPhysicalDeviceExternalMemoryHostPropertiesEXT
*properties
=
1187 (VkPhysicalDeviceExternalMemoryHostPropertiesEXT
*) ext
;
1188 properties
->minImportedHostPointerAlignment
= 4096;
1191 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SUBGROUP_PROPERTIES
: {
1192 VkPhysicalDeviceSubgroupProperties
*properties
=
1193 (VkPhysicalDeviceSubgroupProperties
*)ext
;
1194 properties
->subgroupSize
= 64;
1195 properties
->supportedStages
= VK_SHADER_STAGE_ALL
;
1196 properties
->supportedOperations
=
1197 VK_SUBGROUP_FEATURE_BASIC_BIT
|
1198 VK_SUBGROUP_FEATURE_BALLOT_BIT
|
1199 VK_SUBGROUP_FEATURE_QUAD_BIT
|
1200 VK_SUBGROUP_FEATURE_VOTE_BIT
;
1201 if (pdevice
->rad_info
.chip_class
>= GFX8
) {
1202 properties
->supportedOperations
|=
1203 VK_SUBGROUP_FEATURE_ARITHMETIC_BIT
|
1204 VK_SUBGROUP_FEATURE_SHUFFLE_BIT
|
1205 VK_SUBGROUP_FEATURE_SHUFFLE_RELATIVE_BIT
;
1207 properties
->quadOperationsInAllStages
= true;
1210 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MAINTENANCE_3_PROPERTIES
: {
1211 VkPhysicalDeviceMaintenance3Properties
*properties
=
1212 (VkPhysicalDeviceMaintenance3Properties
*)ext
;
1213 /* Make sure everything is addressable by a signed 32-bit int, and
1214 * our largest descriptors are 96 bytes. */
1215 properties
->maxPerSetDescriptors
= (1ull << 31) / 96;
1216 /* Our buffer size fields allow only this much */
1217 properties
->maxMemoryAllocationSize
= 0xFFFFFFFFull
;
1220 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SAMPLER_FILTER_MINMAX_PROPERTIES_EXT
: {
1221 VkPhysicalDeviceSamplerFilterMinmaxPropertiesEXT
*properties
=
1222 (VkPhysicalDeviceSamplerFilterMinmaxPropertiesEXT
*)ext
;
1223 /* GFX6-8 only support single channel min/max filter. */
1224 properties
->filterMinmaxImageComponentMapping
= pdevice
->rad_info
.chip_class
>= GFX9
;
1225 properties
->filterMinmaxSingleComponentFormats
= true;
1228 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_CORE_PROPERTIES_AMD
: {
1229 VkPhysicalDeviceShaderCorePropertiesAMD
*properties
=
1230 (VkPhysicalDeviceShaderCorePropertiesAMD
*)ext
;
1232 /* Shader engines. */
1233 properties
->shaderEngineCount
=
1234 pdevice
->rad_info
.max_se
;
1235 properties
->shaderArraysPerEngineCount
=
1236 pdevice
->rad_info
.max_sh_per_se
;
1237 properties
->computeUnitsPerShaderArray
=
1238 pdevice
->rad_info
.num_good_cu_per_sh
;
1239 properties
->simdPerComputeUnit
= 4;
1240 properties
->wavefrontsPerSimd
=
1241 pdevice
->rad_info
.family
== CHIP_TONGA
||
1242 pdevice
->rad_info
.family
== CHIP_ICELAND
||
1243 pdevice
->rad_info
.family
== CHIP_POLARIS10
||
1244 pdevice
->rad_info
.family
== CHIP_POLARIS11
||
1245 pdevice
->rad_info
.family
== CHIP_POLARIS12
||
1246 pdevice
->rad_info
.family
== CHIP_VEGAM
? 8 : 10;
1247 properties
->wavefrontSize
= 64;
1250 properties
->sgprsPerSimd
=
1251 ac_get_num_physical_sgprs(pdevice
->rad_info
.chip_class
);
1252 properties
->minSgprAllocation
=
1253 pdevice
->rad_info
.chip_class
>= GFX8
? 16 : 8;
1254 properties
->maxSgprAllocation
=
1255 pdevice
->rad_info
.family
== CHIP_TONGA
||
1256 pdevice
->rad_info
.family
== CHIP_ICELAND
? 96 : 104;
1257 properties
->sgprAllocationGranularity
=
1258 pdevice
->rad_info
.chip_class
>= GFX8
? 16 : 8;
1261 properties
->vgprsPerSimd
= RADV_NUM_PHYSICAL_VGPRS
;
1262 properties
->minVgprAllocation
= 4;
1263 properties
->maxVgprAllocation
= 256;
1264 properties
->vgprAllocationGranularity
= 4;
1267 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VERTEX_ATTRIBUTE_DIVISOR_PROPERTIES_EXT
: {
1268 VkPhysicalDeviceVertexAttributeDivisorPropertiesEXT
*properties
=
1269 (VkPhysicalDeviceVertexAttributeDivisorPropertiesEXT
*)ext
;
1270 properties
->maxVertexAttribDivisor
= UINT32_MAX
;
1273 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DESCRIPTOR_INDEXING_PROPERTIES_EXT
: {
1274 VkPhysicalDeviceDescriptorIndexingPropertiesEXT
*properties
=
1275 (VkPhysicalDeviceDescriptorIndexingPropertiesEXT
*)ext
;
1276 properties
->maxUpdateAfterBindDescriptorsInAllPools
= UINT32_MAX
/ 64;
1277 properties
->shaderUniformBufferArrayNonUniformIndexingNative
= false;
1278 properties
->shaderSampledImageArrayNonUniformIndexingNative
= false;
1279 properties
->shaderStorageBufferArrayNonUniformIndexingNative
= false;
1280 properties
->shaderStorageImageArrayNonUniformIndexingNative
= false;
1281 properties
->shaderInputAttachmentArrayNonUniformIndexingNative
= false;
1282 properties
->robustBufferAccessUpdateAfterBind
= false;
1283 properties
->quadDivergentImplicitLod
= false;
1285 size_t max_descriptor_set_size
= ((1ull << 31) - 16 * MAX_DYNAMIC_BUFFERS
-
1286 MAX_INLINE_UNIFORM_BLOCK_SIZE
* MAX_INLINE_UNIFORM_BLOCK_COUNT
) /
1287 (32 /* uniform buffer, 32 due to potential space wasted on alignment */ +
1288 32 /* storage buffer, 32 due to potential space wasted on alignment */ +
1289 32 /* sampler, largest when combined with image */ +
1290 64 /* sampled image */ +
1291 64 /* storage image */);
1292 properties
->maxPerStageDescriptorUpdateAfterBindSamplers
= max_descriptor_set_size
;
1293 properties
->maxPerStageDescriptorUpdateAfterBindUniformBuffers
= max_descriptor_set_size
;
1294 properties
->maxPerStageDescriptorUpdateAfterBindStorageBuffers
= max_descriptor_set_size
;
1295 properties
->maxPerStageDescriptorUpdateAfterBindSampledImages
= max_descriptor_set_size
;
1296 properties
->maxPerStageDescriptorUpdateAfterBindStorageImages
= max_descriptor_set_size
;
1297 properties
->maxPerStageDescriptorUpdateAfterBindInputAttachments
= max_descriptor_set_size
;
1298 properties
->maxPerStageUpdateAfterBindResources
= max_descriptor_set_size
;
1299 properties
->maxDescriptorSetUpdateAfterBindSamplers
= max_descriptor_set_size
;
1300 properties
->maxDescriptorSetUpdateAfterBindUniformBuffers
= max_descriptor_set_size
;
1301 properties
->maxDescriptorSetUpdateAfterBindUniformBuffersDynamic
= MAX_DYNAMIC_UNIFORM_BUFFERS
;
1302 properties
->maxDescriptorSetUpdateAfterBindStorageBuffers
= max_descriptor_set_size
;
1303 properties
->maxDescriptorSetUpdateAfterBindStorageBuffersDynamic
= MAX_DYNAMIC_STORAGE_BUFFERS
;
1304 properties
->maxDescriptorSetUpdateAfterBindSampledImages
= max_descriptor_set_size
;
1305 properties
->maxDescriptorSetUpdateAfterBindStorageImages
= max_descriptor_set_size
;
1306 properties
->maxDescriptorSetUpdateAfterBindInputAttachments
= max_descriptor_set_size
;
1309 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROTECTED_MEMORY_PROPERTIES
: {
1310 VkPhysicalDeviceProtectedMemoryProperties
*properties
=
1311 (VkPhysicalDeviceProtectedMemoryProperties
*)ext
;
1312 properties
->protectedNoFault
= false;
1315 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_CONSERVATIVE_RASTERIZATION_PROPERTIES_EXT
: {
1316 VkPhysicalDeviceConservativeRasterizationPropertiesEXT
*properties
=
1317 (VkPhysicalDeviceConservativeRasterizationPropertiesEXT
*)ext
;
1318 properties
->primitiveOverestimationSize
= 0;
1319 properties
->maxExtraPrimitiveOverestimationSize
= 0;
1320 properties
->extraPrimitiveOverestimationSizeGranularity
= 0;
1321 properties
->primitiveUnderestimation
= VK_FALSE
;
1322 properties
->conservativePointAndLineRasterization
= VK_FALSE
;
1323 properties
->degenerateTrianglesRasterized
= VK_FALSE
;
1324 properties
->degenerateLinesRasterized
= VK_FALSE
;
1325 properties
->fullyCoveredFragmentShaderInputVariable
= VK_FALSE
;
1326 properties
->conservativeRasterizationPostDepthCoverage
= VK_FALSE
;
1329 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PCI_BUS_INFO_PROPERTIES_EXT
: {
1330 VkPhysicalDevicePCIBusInfoPropertiesEXT
*properties
=
1331 (VkPhysicalDevicePCIBusInfoPropertiesEXT
*)ext
;
1332 properties
->pciDomain
= pdevice
->bus_info
.domain
;
1333 properties
->pciBus
= pdevice
->bus_info
.bus
;
1334 properties
->pciDevice
= pdevice
->bus_info
.dev
;
1335 properties
->pciFunction
= pdevice
->bus_info
.func
;
1338 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DRIVER_PROPERTIES_KHR
: {
1339 VkPhysicalDeviceDriverPropertiesKHR
*driver_props
=
1340 (VkPhysicalDeviceDriverPropertiesKHR
*) ext
;
1342 driver_props
->driverID
= VK_DRIVER_ID_MESA_RADV_KHR
;
1343 snprintf(driver_props
->driverName
, VK_MAX_DRIVER_NAME_SIZE_KHR
, "radv");
1344 snprintf(driver_props
->driverInfo
, VK_MAX_DRIVER_INFO_SIZE_KHR
,
1345 "Mesa " PACKAGE_VERSION MESA_GIT_SHA1
1346 " (LLVM " MESA_LLVM_VERSION_STRING
")");
1348 driver_props
->conformanceVersion
= (VkConformanceVersionKHR
) {
1356 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TRANSFORM_FEEDBACK_PROPERTIES_EXT
: {
1357 VkPhysicalDeviceTransformFeedbackPropertiesEXT
*properties
=
1358 (VkPhysicalDeviceTransformFeedbackPropertiesEXT
*)ext
;
1359 properties
->maxTransformFeedbackStreams
= MAX_SO_STREAMS
;
1360 properties
->maxTransformFeedbackBuffers
= MAX_SO_BUFFERS
;
1361 properties
->maxTransformFeedbackBufferSize
= UINT32_MAX
;
1362 properties
->maxTransformFeedbackStreamDataSize
= 512;
1363 properties
->maxTransformFeedbackBufferDataSize
= UINT32_MAX
;
1364 properties
->maxTransformFeedbackBufferDataStride
= 512;
1365 properties
->transformFeedbackQueries
= true;
1366 properties
->transformFeedbackStreamsLinesTriangles
= true;
1367 properties
->transformFeedbackRasterizationStreamSelect
= false;
1368 properties
->transformFeedbackDraw
= true;
1371 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_INLINE_UNIFORM_BLOCK_PROPERTIES_EXT
: {
1372 VkPhysicalDeviceInlineUniformBlockPropertiesEXT
*props
=
1373 (VkPhysicalDeviceInlineUniformBlockPropertiesEXT
*)ext
;
1375 props
->maxInlineUniformBlockSize
= MAX_INLINE_UNIFORM_BLOCK_SIZE
;
1376 props
->maxPerStageDescriptorInlineUniformBlocks
= MAX_INLINE_UNIFORM_BLOCK_SIZE
* MAX_SETS
;
1377 props
->maxPerStageDescriptorUpdateAfterBindInlineUniformBlocks
= MAX_INLINE_UNIFORM_BLOCK_SIZE
* MAX_SETS
;
1378 props
->maxDescriptorSetInlineUniformBlocks
= MAX_INLINE_UNIFORM_BLOCK_COUNT
;
1379 props
->maxDescriptorSetUpdateAfterBindInlineUniformBlocks
= MAX_INLINE_UNIFORM_BLOCK_COUNT
;
1382 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SAMPLE_LOCATIONS_PROPERTIES_EXT
: {
1383 VkPhysicalDeviceSampleLocationsPropertiesEXT
*properties
=
1384 (VkPhysicalDeviceSampleLocationsPropertiesEXT
*)ext
;
1385 properties
->sampleLocationSampleCounts
= VK_SAMPLE_COUNT_2_BIT
|
1386 VK_SAMPLE_COUNT_4_BIT
|
1387 VK_SAMPLE_COUNT_8_BIT
;
1388 properties
->maxSampleLocationGridSize
= (VkExtent2D
){ 2 , 2 };
1389 properties
->sampleLocationCoordinateRange
[0] = 0.0f
;
1390 properties
->sampleLocationCoordinateRange
[1] = 0.9375f
;
1391 properties
->sampleLocationSubPixelBits
= 4;
1392 properties
->variableSampleLocations
= VK_FALSE
;
1395 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DEPTH_STENCIL_RESOLVE_PROPERTIES_KHR
: {
1396 VkPhysicalDeviceDepthStencilResolvePropertiesKHR
*properties
=
1397 (VkPhysicalDeviceDepthStencilResolvePropertiesKHR
*)ext
;
1399 /* We support all of the depth resolve modes */
1400 properties
->supportedDepthResolveModes
=
1401 VK_RESOLVE_MODE_SAMPLE_ZERO_BIT_KHR
|
1402 VK_RESOLVE_MODE_AVERAGE_BIT_KHR
|
1403 VK_RESOLVE_MODE_MIN_BIT_KHR
|
1404 VK_RESOLVE_MODE_MAX_BIT_KHR
;
1406 /* Average doesn't make sense for stencil so we don't support that */
1407 properties
->supportedStencilResolveModes
=
1408 VK_RESOLVE_MODE_SAMPLE_ZERO_BIT_KHR
|
1409 VK_RESOLVE_MODE_MIN_BIT_KHR
|
1410 VK_RESOLVE_MODE_MAX_BIT_KHR
;
1412 properties
->independentResolveNone
= VK_TRUE
;
1413 properties
->independentResolve
= VK_TRUE
;
1422 static void radv_get_physical_device_queue_family_properties(
1423 struct radv_physical_device
* pdevice
,
1425 VkQueueFamilyProperties
** pQueueFamilyProperties
)
1427 int num_queue_families
= 1;
1429 if (pdevice
->rad_info
.num_compute_rings
> 0 &&
1430 !(pdevice
->instance
->debug_flags
& RADV_DEBUG_NO_COMPUTE_QUEUE
))
1431 num_queue_families
++;
1433 if (pQueueFamilyProperties
== NULL
) {
1434 *pCount
= num_queue_families
;
1443 *pQueueFamilyProperties
[idx
] = (VkQueueFamilyProperties
) {
1444 .queueFlags
= VK_QUEUE_GRAPHICS_BIT
|
1445 VK_QUEUE_COMPUTE_BIT
|
1446 VK_QUEUE_TRANSFER_BIT
|
1447 VK_QUEUE_SPARSE_BINDING_BIT
,
1449 .timestampValidBits
= 64,
1450 .minImageTransferGranularity
= (VkExtent3D
) { 1, 1, 1 },
1455 if (pdevice
->rad_info
.num_compute_rings
> 0 &&
1456 !(pdevice
->instance
->debug_flags
& RADV_DEBUG_NO_COMPUTE_QUEUE
)) {
1457 if (*pCount
> idx
) {
1458 *pQueueFamilyProperties
[idx
] = (VkQueueFamilyProperties
) {
1459 .queueFlags
= VK_QUEUE_COMPUTE_BIT
|
1460 VK_QUEUE_TRANSFER_BIT
|
1461 VK_QUEUE_SPARSE_BINDING_BIT
,
1462 .queueCount
= pdevice
->rad_info
.num_compute_rings
,
1463 .timestampValidBits
= 64,
1464 .minImageTransferGranularity
= (VkExtent3D
) { 1, 1, 1 },
1472 void radv_GetPhysicalDeviceQueueFamilyProperties(
1473 VkPhysicalDevice physicalDevice
,
1475 VkQueueFamilyProperties
* pQueueFamilyProperties
)
1477 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
1478 if (!pQueueFamilyProperties
) {
1479 radv_get_physical_device_queue_family_properties(pdevice
, pCount
, NULL
);
1482 VkQueueFamilyProperties
*properties
[] = {
1483 pQueueFamilyProperties
+ 0,
1484 pQueueFamilyProperties
+ 1,
1485 pQueueFamilyProperties
+ 2,
1487 radv_get_physical_device_queue_family_properties(pdevice
, pCount
, properties
);
1488 assert(*pCount
<= 3);
1491 void radv_GetPhysicalDeviceQueueFamilyProperties2(
1492 VkPhysicalDevice physicalDevice
,
1494 VkQueueFamilyProperties2
*pQueueFamilyProperties
)
1496 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
1497 if (!pQueueFamilyProperties
) {
1498 radv_get_physical_device_queue_family_properties(pdevice
, pCount
, NULL
);
1501 VkQueueFamilyProperties
*properties
[] = {
1502 &pQueueFamilyProperties
[0].queueFamilyProperties
,
1503 &pQueueFamilyProperties
[1].queueFamilyProperties
,
1504 &pQueueFamilyProperties
[2].queueFamilyProperties
,
1506 radv_get_physical_device_queue_family_properties(pdevice
, pCount
, properties
);
1507 assert(*pCount
<= 3);
1510 void radv_GetPhysicalDeviceMemoryProperties(
1511 VkPhysicalDevice physicalDevice
,
1512 VkPhysicalDeviceMemoryProperties
*pMemoryProperties
)
1514 RADV_FROM_HANDLE(radv_physical_device
, physical_device
, physicalDevice
);
1516 *pMemoryProperties
= physical_device
->memory_properties
;
1520 radv_get_memory_budget_properties(VkPhysicalDevice physicalDevice
,
1521 VkPhysicalDeviceMemoryBudgetPropertiesEXT
*memoryBudget
)
1523 RADV_FROM_HANDLE(radv_physical_device
, device
, physicalDevice
);
1524 VkPhysicalDeviceMemoryProperties
*memory_properties
= &device
->memory_properties
;
1525 uint64_t visible_vram_size
= radv_get_visible_vram_size(device
);
1526 uint64_t vram_size
= radv_get_vram_size(device
);
1527 uint64_t gtt_size
= device
->rad_info
.gart_size
;
1528 uint64_t heap_budget
, heap_usage
;
1530 /* For all memory heaps, the computation of budget is as follow:
1531 * heap_budget = heap_size - global_heap_usage + app_heap_usage
1533 * The Vulkan spec 1.1.97 says that the budget should include any
1534 * currently allocated device memory.
1536 * Note that the application heap usages are not really accurate (eg.
1537 * in presence of shared buffers).
1539 for (int i
= 0; i
< device
->memory_properties
.memoryTypeCount
; i
++) {
1540 uint32_t heap_index
= device
->memory_properties
.memoryTypes
[i
].heapIndex
;
1542 switch (device
->mem_type_indices
[i
]) {
1543 case RADV_MEM_TYPE_VRAM
:
1544 heap_usage
= device
->ws
->query_value(device
->ws
,
1545 RADEON_ALLOCATED_VRAM
);
1547 heap_budget
= vram_size
-
1548 device
->ws
->query_value(device
->ws
, RADEON_VRAM_USAGE
) +
1551 memoryBudget
->heapBudget
[heap_index
] = heap_budget
;
1552 memoryBudget
->heapUsage
[heap_index
] = heap_usage
;
1554 case RADV_MEM_TYPE_VRAM_CPU_ACCESS
:
1555 heap_usage
= device
->ws
->query_value(device
->ws
,
1556 RADEON_ALLOCATED_VRAM_VIS
);
1558 heap_budget
= visible_vram_size
-
1559 device
->ws
->query_value(device
->ws
, RADEON_VRAM_VIS_USAGE
) +
1562 memoryBudget
->heapBudget
[heap_index
] = heap_budget
;
1563 memoryBudget
->heapUsage
[heap_index
] = heap_usage
;
1565 case RADV_MEM_TYPE_GTT_WRITE_COMBINE
:
1566 heap_usage
= device
->ws
->query_value(device
->ws
,
1567 RADEON_ALLOCATED_GTT
);
1569 heap_budget
= gtt_size
-
1570 device
->ws
->query_value(device
->ws
, RADEON_GTT_USAGE
) +
1573 memoryBudget
->heapBudget
[heap_index
] = heap_budget
;
1574 memoryBudget
->heapUsage
[heap_index
] = heap_usage
;
1581 /* The heapBudget and heapUsage values must be zero for array elements
1582 * greater than or equal to
1583 * VkPhysicalDeviceMemoryProperties::memoryHeapCount.
1585 for (uint32_t i
= memory_properties
->memoryHeapCount
; i
< VK_MAX_MEMORY_HEAPS
; i
++) {
1586 memoryBudget
->heapBudget
[i
] = 0;
1587 memoryBudget
->heapUsage
[i
] = 0;
1591 void radv_GetPhysicalDeviceMemoryProperties2(
1592 VkPhysicalDevice physicalDevice
,
1593 VkPhysicalDeviceMemoryProperties2
*pMemoryProperties
)
1595 radv_GetPhysicalDeviceMemoryProperties(physicalDevice
,
1596 &pMemoryProperties
->memoryProperties
);
1598 VkPhysicalDeviceMemoryBudgetPropertiesEXT
*memory_budget
=
1599 vk_find_struct(pMemoryProperties
->pNext
,
1600 PHYSICAL_DEVICE_MEMORY_BUDGET_PROPERTIES_EXT
);
1602 radv_get_memory_budget_properties(physicalDevice
, memory_budget
);
1605 VkResult
radv_GetMemoryHostPointerPropertiesEXT(
1607 VkExternalMemoryHandleTypeFlagBits handleType
,
1608 const void *pHostPointer
,
1609 VkMemoryHostPointerPropertiesEXT
*pMemoryHostPointerProperties
)
1611 RADV_FROM_HANDLE(radv_device
, device
, _device
);
1615 case VK_EXTERNAL_MEMORY_HANDLE_TYPE_HOST_ALLOCATION_BIT_EXT
: {
1616 const struct radv_physical_device
*physical_device
= device
->physical_device
;
1617 uint32_t memoryTypeBits
= 0;
1618 for (int i
= 0; i
< physical_device
->memory_properties
.memoryTypeCount
; i
++) {
1619 if (physical_device
->mem_type_indices
[i
] == RADV_MEM_TYPE_GTT_CACHED
) {
1620 memoryTypeBits
= (1 << i
);
1624 pMemoryHostPointerProperties
->memoryTypeBits
= memoryTypeBits
;
1628 return VK_ERROR_INVALID_EXTERNAL_HANDLE
;
1632 static enum radeon_ctx_priority
1633 radv_get_queue_global_priority(const VkDeviceQueueGlobalPriorityCreateInfoEXT
*pObj
)
1635 /* Default to MEDIUM when a specific global priority isn't requested */
1637 return RADEON_CTX_PRIORITY_MEDIUM
;
1639 switch(pObj
->globalPriority
) {
1640 case VK_QUEUE_GLOBAL_PRIORITY_REALTIME_EXT
:
1641 return RADEON_CTX_PRIORITY_REALTIME
;
1642 case VK_QUEUE_GLOBAL_PRIORITY_HIGH_EXT
:
1643 return RADEON_CTX_PRIORITY_HIGH
;
1644 case VK_QUEUE_GLOBAL_PRIORITY_MEDIUM_EXT
:
1645 return RADEON_CTX_PRIORITY_MEDIUM
;
1646 case VK_QUEUE_GLOBAL_PRIORITY_LOW_EXT
:
1647 return RADEON_CTX_PRIORITY_LOW
;
1649 unreachable("Illegal global priority value");
1650 return RADEON_CTX_PRIORITY_INVALID
;
1655 radv_queue_init(struct radv_device
*device
, struct radv_queue
*queue
,
1656 uint32_t queue_family_index
, int idx
,
1657 VkDeviceQueueCreateFlags flags
,
1658 const VkDeviceQueueGlobalPriorityCreateInfoEXT
*global_priority
)
1660 queue
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
1661 queue
->device
= device
;
1662 queue
->queue_family_index
= queue_family_index
;
1663 queue
->queue_idx
= idx
;
1664 queue
->priority
= radv_get_queue_global_priority(global_priority
);
1665 queue
->flags
= flags
;
1667 queue
->hw_ctx
= device
->ws
->ctx_create(device
->ws
, queue
->priority
);
1669 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
1675 radv_queue_finish(struct radv_queue
*queue
)
1678 queue
->device
->ws
->ctx_destroy(queue
->hw_ctx
);
1680 if (queue
->initial_full_flush_preamble_cs
)
1681 queue
->device
->ws
->cs_destroy(queue
->initial_full_flush_preamble_cs
);
1682 if (queue
->initial_preamble_cs
)
1683 queue
->device
->ws
->cs_destroy(queue
->initial_preamble_cs
);
1684 if (queue
->continue_preamble_cs
)
1685 queue
->device
->ws
->cs_destroy(queue
->continue_preamble_cs
);
1686 if (queue
->descriptor_bo
)
1687 queue
->device
->ws
->buffer_destroy(queue
->descriptor_bo
);
1688 if (queue
->scratch_bo
)
1689 queue
->device
->ws
->buffer_destroy(queue
->scratch_bo
);
1690 if (queue
->esgs_ring_bo
)
1691 queue
->device
->ws
->buffer_destroy(queue
->esgs_ring_bo
);
1692 if (queue
->gsvs_ring_bo
)
1693 queue
->device
->ws
->buffer_destroy(queue
->gsvs_ring_bo
);
1694 if (queue
->tess_rings_bo
)
1695 queue
->device
->ws
->buffer_destroy(queue
->tess_rings_bo
);
1696 if (queue
->compute_scratch_bo
)
1697 queue
->device
->ws
->buffer_destroy(queue
->compute_scratch_bo
);
1701 radv_bo_list_init(struct radv_bo_list
*bo_list
)
1703 pthread_mutex_init(&bo_list
->mutex
, NULL
);
1704 bo_list
->list
.count
= bo_list
->capacity
= 0;
1705 bo_list
->list
.bos
= NULL
;
1709 radv_bo_list_finish(struct radv_bo_list
*bo_list
)
1711 free(bo_list
->list
.bos
);
1712 pthread_mutex_destroy(&bo_list
->mutex
);
1715 static VkResult
radv_bo_list_add(struct radv_device
*device
,
1716 struct radeon_winsys_bo
*bo
)
1718 struct radv_bo_list
*bo_list
= &device
->bo_list
;
1723 if (unlikely(!device
->use_global_bo_list
))
1726 pthread_mutex_lock(&bo_list
->mutex
);
1727 if (bo_list
->list
.count
== bo_list
->capacity
) {
1728 unsigned capacity
= MAX2(4, bo_list
->capacity
* 2);
1729 void *data
= realloc(bo_list
->list
.bos
, capacity
* sizeof(struct radeon_winsys_bo
*));
1732 pthread_mutex_unlock(&bo_list
->mutex
);
1733 return VK_ERROR_OUT_OF_HOST_MEMORY
;
1736 bo_list
->list
.bos
= (struct radeon_winsys_bo
**)data
;
1737 bo_list
->capacity
= capacity
;
1740 bo_list
->list
.bos
[bo_list
->list
.count
++] = bo
;
1741 pthread_mutex_unlock(&bo_list
->mutex
);
1745 static void radv_bo_list_remove(struct radv_device
*device
,
1746 struct radeon_winsys_bo
*bo
)
1748 struct radv_bo_list
*bo_list
= &device
->bo_list
;
1753 if (unlikely(!device
->use_global_bo_list
))
1756 pthread_mutex_lock(&bo_list
->mutex
);
1757 for(unsigned i
= 0; i
< bo_list
->list
.count
; ++i
) {
1758 if (bo_list
->list
.bos
[i
] == bo
) {
1759 bo_list
->list
.bos
[i
] = bo_list
->list
.bos
[bo_list
->list
.count
- 1];
1760 --bo_list
->list
.count
;
1764 pthread_mutex_unlock(&bo_list
->mutex
);
1768 radv_device_init_gs_info(struct radv_device
*device
)
1770 device
->gs_table_depth
= ac_get_gs_table_depth(device
->physical_device
->rad_info
.chip_class
,
1771 device
->physical_device
->rad_info
.family
);
1774 static int radv_get_device_extension_index(const char *name
)
1776 for (unsigned i
= 0; i
< RADV_DEVICE_EXTENSION_COUNT
; ++i
) {
1777 if (strcmp(name
, radv_device_extensions
[i
].extensionName
) == 0)
1784 radv_get_int_debug_option(const char *name
, int default_value
)
1791 result
= default_value
;
1795 result
= strtol(str
, &endptr
, 0);
1796 if (str
== endptr
) {
1797 /* No digits founs. */
1798 result
= default_value
;
1805 VkResult
radv_CreateDevice(
1806 VkPhysicalDevice physicalDevice
,
1807 const VkDeviceCreateInfo
* pCreateInfo
,
1808 const VkAllocationCallbacks
* pAllocator
,
1811 RADV_FROM_HANDLE(radv_physical_device
, physical_device
, physicalDevice
);
1813 struct radv_device
*device
;
1815 bool keep_shader_info
= false;
1817 /* Check enabled features */
1818 if (pCreateInfo
->pEnabledFeatures
) {
1819 VkPhysicalDeviceFeatures supported_features
;
1820 radv_GetPhysicalDeviceFeatures(physicalDevice
, &supported_features
);
1821 VkBool32
*supported_feature
= (VkBool32
*)&supported_features
;
1822 VkBool32
*enabled_feature
= (VkBool32
*)pCreateInfo
->pEnabledFeatures
;
1823 unsigned num_features
= sizeof(VkPhysicalDeviceFeatures
) / sizeof(VkBool32
);
1824 for (uint32_t i
= 0; i
< num_features
; i
++) {
1825 if (enabled_feature
[i
] && !supported_feature
[i
])
1826 return vk_error(physical_device
->instance
, VK_ERROR_FEATURE_NOT_PRESENT
);
1830 device
= vk_zalloc2(&physical_device
->instance
->alloc
, pAllocator
,
1832 VK_SYSTEM_ALLOCATION_SCOPE_DEVICE
);
1834 return vk_error(physical_device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
1836 device
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
1837 device
->instance
= physical_device
->instance
;
1838 device
->physical_device
= physical_device
;
1840 device
->ws
= physical_device
->ws
;
1842 device
->alloc
= *pAllocator
;
1844 device
->alloc
= physical_device
->instance
->alloc
;
1846 for (uint32_t i
= 0; i
< pCreateInfo
->enabledExtensionCount
; i
++) {
1847 const char *ext_name
= pCreateInfo
->ppEnabledExtensionNames
[i
];
1848 int index
= radv_get_device_extension_index(ext_name
);
1849 if (index
< 0 || !physical_device
->supported_extensions
.extensions
[index
]) {
1850 vk_free(&device
->alloc
, device
);
1851 return vk_error(physical_device
->instance
, VK_ERROR_EXTENSION_NOT_PRESENT
);
1854 device
->enabled_extensions
.extensions
[index
] = true;
1857 keep_shader_info
= device
->enabled_extensions
.AMD_shader_info
;
1859 /* With update after bind we can't attach bo's to the command buffer
1860 * from the descriptor set anymore, so we have to use a global BO list.
1862 device
->use_global_bo_list
=
1863 (device
->instance
->perftest_flags
& RADV_PERFTEST_BO_LIST
) ||
1864 device
->enabled_extensions
.EXT_descriptor_indexing
||
1865 device
->enabled_extensions
.EXT_buffer_device_address
;
1867 mtx_init(&device
->shader_slab_mutex
, mtx_plain
);
1868 list_inithead(&device
->shader_slabs
);
1870 radv_bo_list_init(&device
->bo_list
);
1872 for (unsigned i
= 0; i
< pCreateInfo
->queueCreateInfoCount
; i
++) {
1873 const VkDeviceQueueCreateInfo
*queue_create
= &pCreateInfo
->pQueueCreateInfos
[i
];
1874 uint32_t qfi
= queue_create
->queueFamilyIndex
;
1875 const VkDeviceQueueGlobalPriorityCreateInfoEXT
*global_priority
=
1876 vk_find_struct_const(queue_create
->pNext
, DEVICE_QUEUE_GLOBAL_PRIORITY_CREATE_INFO_EXT
);
1878 assert(!global_priority
|| device
->physical_device
->rad_info
.has_ctx_priority
);
1880 device
->queues
[qfi
] = vk_alloc(&device
->alloc
,
1881 queue_create
->queueCount
* sizeof(struct radv_queue
), 8, VK_SYSTEM_ALLOCATION_SCOPE_DEVICE
);
1882 if (!device
->queues
[qfi
]) {
1883 result
= VK_ERROR_OUT_OF_HOST_MEMORY
;
1887 memset(device
->queues
[qfi
], 0, queue_create
->queueCount
* sizeof(struct radv_queue
));
1889 device
->queue_count
[qfi
] = queue_create
->queueCount
;
1891 for (unsigned q
= 0; q
< queue_create
->queueCount
; q
++) {
1892 result
= radv_queue_init(device
, &device
->queues
[qfi
][q
],
1893 qfi
, q
, queue_create
->flags
,
1895 if (result
!= VK_SUCCESS
)
1900 /* TODO: Enable binning for GFX10. */
1901 device
->pbb_allowed
= device
->physical_device
->rad_info
.chip_class
== GFX9
&&
1902 !(device
->instance
->debug_flags
& RADV_DEBUG_NOBINNING
);
1904 /* Disabled and not implemented for now. */
1905 device
->dfsm_allowed
= device
->pbb_allowed
&&
1906 (device
->physical_device
->rad_info
.family
== CHIP_RAVEN
||
1907 device
->physical_device
->rad_info
.family
== CHIP_RAVEN2
);
1910 device
->always_use_syncobj
= device
->physical_device
->rad_info
.has_syncobj_wait_for_submit
;
1913 /* The maximum number of scratch waves. Scratch space isn't divided
1914 * evenly between CUs. The number is only a function of the number of CUs.
1915 * We can decrease the constant to decrease the scratch buffer size.
1917 * sctx->scratch_waves must be >= the maximum possible size of
1918 * 1 threadgroup, so that the hw doesn't hang from being unable
1921 * The recommended value is 4 per CU at most. Higher numbers don't
1922 * bring much benefit, but they still occupy chip resources (think
1923 * async compute). I've seen ~2% performance difference between 4 and 32.
1925 uint32_t max_threads_per_block
= 2048;
1926 device
->scratch_waves
= MAX2(32 * physical_device
->rad_info
.num_good_compute_units
,
1927 max_threads_per_block
/ 64);
1929 device
->dispatch_initiator
= S_00B800_COMPUTE_SHADER_EN(1);
1931 if (device
->physical_device
->rad_info
.chip_class
>= GFX7
) {
1932 /* If the KMD allows it (there is a KMD hw register for it),
1933 * allow launching waves out-of-order.
1935 device
->dispatch_initiator
|= S_00B800_ORDER_MODE(1);
1938 radv_device_init_gs_info(device
);
1940 device
->tess_offchip_block_dw_size
=
1941 device
->physical_device
->rad_info
.family
== CHIP_HAWAII
? 4096 : 8192;
1942 device
->has_distributed_tess
=
1943 device
->physical_device
->rad_info
.chip_class
>= GFX8
&&
1944 device
->physical_device
->rad_info
.max_se
>= 2;
1946 if (getenv("RADV_TRACE_FILE")) {
1947 const char *filename
= getenv("RADV_TRACE_FILE");
1949 keep_shader_info
= true;
1951 if (!radv_init_trace(device
))
1954 fprintf(stderr
, "*****************************************************************************\n");
1955 fprintf(stderr
, "* WARNING: RADV_TRACE_FILE is costly and should only be used for debugging! *\n");
1956 fprintf(stderr
, "*****************************************************************************\n");
1958 fprintf(stderr
, "Trace file will be dumped to %s\n", filename
);
1959 radv_dump_enabled_options(device
, stderr
);
1962 device
->keep_shader_info
= keep_shader_info
;
1964 result
= radv_device_init_meta(device
);
1965 if (result
!= VK_SUCCESS
)
1968 radv_device_init_msaa(device
);
1970 for (int family
= 0; family
< RADV_MAX_QUEUE_FAMILIES
; ++family
) {
1971 device
->empty_cs
[family
] = device
->ws
->cs_create(device
->ws
, family
);
1973 case RADV_QUEUE_GENERAL
:
1974 radeon_emit(device
->empty_cs
[family
], PKT3(PKT3_CONTEXT_CONTROL
, 1, 0));
1975 radeon_emit(device
->empty_cs
[family
], CONTEXT_CONTROL_LOAD_ENABLE(1));
1976 radeon_emit(device
->empty_cs
[family
], CONTEXT_CONTROL_SHADOW_ENABLE(1));
1978 case RADV_QUEUE_COMPUTE
:
1979 radeon_emit(device
->empty_cs
[family
], PKT3(PKT3_NOP
, 0, 0));
1980 radeon_emit(device
->empty_cs
[family
], 0);
1983 device
->ws
->cs_finalize(device
->empty_cs
[family
]);
1986 if (device
->physical_device
->rad_info
.chip_class
>= GFX7
)
1987 cik_create_gfx_config(device
);
1989 VkPipelineCacheCreateInfo ci
;
1990 ci
.sType
= VK_STRUCTURE_TYPE_PIPELINE_CACHE_CREATE_INFO
;
1993 ci
.pInitialData
= NULL
;
1994 ci
.initialDataSize
= 0;
1996 result
= radv_CreatePipelineCache(radv_device_to_handle(device
),
1998 if (result
!= VK_SUCCESS
)
2001 device
->mem_cache
= radv_pipeline_cache_from_handle(pc
);
2003 device
->force_aniso
=
2004 MIN2(16, radv_get_int_debug_option("RADV_TEX_ANISO", -1));
2005 if (device
->force_aniso
>= 0) {
2006 fprintf(stderr
, "radv: Forcing anisotropy filter to %ix\n",
2007 1 << util_logbase2(device
->force_aniso
));
2010 *pDevice
= radv_device_to_handle(device
);
2014 radv_device_finish_meta(device
);
2016 radv_bo_list_finish(&device
->bo_list
);
2018 if (device
->trace_bo
)
2019 device
->ws
->buffer_destroy(device
->trace_bo
);
2021 if (device
->gfx_init
)
2022 device
->ws
->buffer_destroy(device
->gfx_init
);
2024 for (unsigned i
= 0; i
< RADV_MAX_QUEUE_FAMILIES
; i
++) {
2025 for (unsigned q
= 0; q
< device
->queue_count
[i
]; q
++)
2026 radv_queue_finish(&device
->queues
[i
][q
]);
2027 if (device
->queue_count
[i
])
2028 vk_free(&device
->alloc
, device
->queues
[i
]);
2031 vk_free(&device
->alloc
, device
);
2035 void radv_DestroyDevice(
2037 const VkAllocationCallbacks
* pAllocator
)
2039 RADV_FROM_HANDLE(radv_device
, device
, _device
);
2044 if (device
->trace_bo
)
2045 device
->ws
->buffer_destroy(device
->trace_bo
);
2047 if (device
->gfx_init
)
2048 device
->ws
->buffer_destroy(device
->gfx_init
);
2050 for (unsigned i
= 0; i
< RADV_MAX_QUEUE_FAMILIES
; i
++) {
2051 for (unsigned q
= 0; q
< device
->queue_count
[i
]; q
++)
2052 radv_queue_finish(&device
->queues
[i
][q
]);
2053 if (device
->queue_count
[i
])
2054 vk_free(&device
->alloc
, device
->queues
[i
]);
2055 if (device
->empty_cs
[i
])
2056 device
->ws
->cs_destroy(device
->empty_cs
[i
]);
2058 radv_device_finish_meta(device
);
2060 VkPipelineCache pc
= radv_pipeline_cache_to_handle(device
->mem_cache
);
2061 radv_DestroyPipelineCache(radv_device_to_handle(device
), pc
, NULL
);
2063 radv_destroy_shader_slabs(device
);
2065 radv_bo_list_finish(&device
->bo_list
);
2066 vk_free(&device
->alloc
, device
);
2069 VkResult
radv_EnumerateInstanceLayerProperties(
2070 uint32_t* pPropertyCount
,
2071 VkLayerProperties
* pProperties
)
2073 if (pProperties
== NULL
) {
2074 *pPropertyCount
= 0;
2078 /* None supported at this time */
2079 return vk_error(NULL
, VK_ERROR_LAYER_NOT_PRESENT
);
2082 VkResult
radv_EnumerateDeviceLayerProperties(
2083 VkPhysicalDevice physicalDevice
,
2084 uint32_t* pPropertyCount
,
2085 VkLayerProperties
* pProperties
)
2087 if (pProperties
== NULL
) {
2088 *pPropertyCount
= 0;
2092 /* None supported at this time */
2093 return vk_error(NULL
, VK_ERROR_LAYER_NOT_PRESENT
);
2096 void radv_GetDeviceQueue2(
2098 const VkDeviceQueueInfo2
* pQueueInfo
,
2101 RADV_FROM_HANDLE(radv_device
, device
, _device
);
2102 struct radv_queue
*queue
;
2104 queue
= &device
->queues
[pQueueInfo
->queueFamilyIndex
][pQueueInfo
->queueIndex
];
2105 if (pQueueInfo
->flags
!= queue
->flags
) {
2106 /* From the Vulkan 1.1.70 spec:
2108 * "The queue returned by vkGetDeviceQueue2 must have the same
2109 * flags value from this structure as that used at device
2110 * creation time in a VkDeviceQueueCreateInfo instance. If no
2111 * matching flags were specified at device creation time then
2112 * pQueue will return VK_NULL_HANDLE."
2114 *pQueue
= VK_NULL_HANDLE
;
2118 *pQueue
= radv_queue_to_handle(queue
);
2121 void radv_GetDeviceQueue(
2123 uint32_t queueFamilyIndex
,
2124 uint32_t queueIndex
,
2127 const VkDeviceQueueInfo2 info
= (VkDeviceQueueInfo2
) {
2128 .sType
= VK_STRUCTURE_TYPE_DEVICE_QUEUE_INFO_2
,
2129 .queueFamilyIndex
= queueFamilyIndex
,
2130 .queueIndex
= queueIndex
2133 radv_GetDeviceQueue2(_device
, &info
, pQueue
);
2137 fill_geom_tess_rings(struct radv_queue
*queue
,
2139 bool add_sample_positions
,
2140 uint32_t esgs_ring_size
,
2141 struct radeon_winsys_bo
*esgs_ring_bo
,
2142 uint32_t gsvs_ring_size
,
2143 struct radeon_winsys_bo
*gsvs_ring_bo
,
2144 uint32_t tess_factor_ring_size
,
2145 uint32_t tess_offchip_ring_offset
,
2146 uint32_t tess_offchip_ring_size
,
2147 struct radeon_winsys_bo
*tess_rings_bo
)
2149 uint32_t *desc
= &map
[4];
2152 uint64_t esgs_va
= radv_buffer_get_va(esgs_ring_bo
);
2154 /* stride 0, num records - size, add tid, swizzle, elsize4,
2157 desc
[1] = S_008F04_BASE_ADDRESS_HI(esgs_va
>> 32) |
2158 S_008F04_STRIDE(0) |
2159 S_008F04_SWIZZLE_ENABLE(true);
2160 desc
[2] = esgs_ring_size
;
2161 desc
[3] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
2162 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
2163 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
2164 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
) |
2165 S_008F0C_ELEMENT_SIZE(1) |
2166 S_008F0C_INDEX_STRIDE(3) |
2167 S_008F0C_ADD_TID_ENABLE(true);
2169 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
2170 desc
[3] |= S_008F0C_FORMAT(V_008F0C_IMG_FORMAT_32_FLOAT
) |
2171 S_008F0C_OOB_SELECT(2) |
2172 S_008F0C_RESOURCE_LEVEL(1);
2174 desc
[3] |= S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
2175 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
);
2178 /* GS entry for ES->GS ring */
2179 /* stride 0, num records - size, elsize0,
2182 desc
[5] = S_008F04_BASE_ADDRESS_HI(esgs_va
>> 32)|
2183 S_008F04_STRIDE(0) |
2184 S_008F04_SWIZZLE_ENABLE(false);
2185 desc
[6] = esgs_ring_size
;
2186 desc
[7] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
2187 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
2188 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
2189 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
) |
2190 S_008F0C_ELEMENT_SIZE(0) |
2191 S_008F0C_INDEX_STRIDE(0) |
2192 S_008F0C_ADD_TID_ENABLE(false);
2194 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
2195 desc
[7] |= S_008F0C_FORMAT(V_008F0C_IMG_FORMAT_32_FLOAT
) |
2196 S_008F0C_OOB_SELECT(2) |
2197 S_008F0C_RESOURCE_LEVEL(1);
2199 desc
[7] |= S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
2200 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
);
2207 uint64_t gsvs_va
= radv_buffer_get_va(gsvs_ring_bo
);
2209 /* VS entry for GS->VS ring */
2210 /* stride 0, num records - size, elsize0,
2213 desc
[1] = S_008F04_BASE_ADDRESS_HI(gsvs_va
>> 32)|
2214 S_008F04_STRIDE(0) |
2215 S_008F04_SWIZZLE_ENABLE(false);
2216 desc
[2] = gsvs_ring_size
;
2217 desc
[3] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
2218 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
2219 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
2220 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
) |
2221 S_008F0C_ELEMENT_SIZE(0) |
2222 S_008F0C_INDEX_STRIDE(0) |
2223 S_008F0C_ADD_TID_ENABLE(false);
2225 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
2226 desc
[3] |= S_008F0C_FORMAT(V_008F0C_IMG_FORMAT_32_FLOAT
) |
2227 S_008F0C_OOB_SELECT(2) |
2228 S_008F0C_RESOURCE_LEVEL(1);
2230 desc
[3] |= S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
2231 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
);
2234 /* stride gsvs_itemsize, num records 64
2235 elsize 4, index stride 16 */
2236 /* shader will patch stride and desc[2] */
2238 desc
[5] = S_008F04_BASE_ADDRESS_HI(gsvs_va
>> 32)|
2239 S_008F04_STRIDE(0) |
2240 S_008F04_SWIZZLE_ENABLE(true);
2242 desc
[7] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
2243 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
2244 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
2245 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
) |
2246 S_008F0C_ELEMENT_SIZE(1) |
2247 S_008F0C_INDEX_STRIDE(1) |
2248 S_008F0C_ADD_TID_ENABLE(true);
2250 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
2251 desc
[7] |= S_008F0C_FORMAT(V_008F0C_IMG_FORMAT_32_FLOAT
) |
2252 S_008F0C_OOB_SELECT(2) |
2253 S_008F0C_RESOURCE_LEVEL(1);
2255 desc
[7] |= S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
2256 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
);
2263 if (tess_rings_bo
) {
2264 uint64_t tess_va
= radv_buffer_get_va(tess_rings_bo
);
2265 uint64_t tess_offchip_va
= tess_va
+ tess_offchip_ring_offset
;
2268 desc
[1] = S_008F04_BASE_ADDRESS_HI(tess_va
>> 32) |
2269 S_008F04_STRIDE(0) |
2270 S_008F04_SWIZZLE_ENABLE(false);
2271 desc
[2] = tess_factor_ring_size
;
2272 desc
[3] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
2273 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
2274 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
2275 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
);
2277 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
2278 desc
[3] |= S_008F0C_FORMAT(V_008F0C_IMG_FORMAT_32_FLOAT
) |
2279 S_008F0C_OOB_SELECT(3) |
2280 S_008F0C_RESOURCE_LEVEL(1);
2282 desc
[3] |= S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
2283 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
);
2286 desc
[4] = tess_offchip_va
;
2287 desc
[5] = S_008F04_BASE_ADDRESS_HI(tess_offchip_va
>> 32) |
2288 S_008F04_STRIDE(0) |
2289 S_008F04_SWIZZLE_ENABLE(false);
2290 desc
[6] = tess_offchip_ring_size
;
2291 desc
[7] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
2292 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
2293 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
2294 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
);
2296 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
2297 desc
[7] |= S_008F0C_FORMAT(V_008F0C_IMG_FORMAT_32_FLOAT
) |
2298 S_008F0C_OOB_SELECT(3) |
2299 S_008F0C_RESOURCE_LEVEL(1);
2301 desc
[7] |= S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
2302 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
);
2308 if (add_sample_positions
) {
2309 /* add sample positions after all rings */
2310 memcpy(desc
, queue
->device
->sample_locations_1x
, 8);
2312 memcpy(desc
, queue
->device
->sample_locations_2x
, 16);
2314 memcpy(desc
, queue
->device
->sample_locations_4x
, 32);
2316 memcpy(desc
, queue
->device
->sample_locations_8x
, 64);
2321 radv_get_hs_offchip_param(struct radv_device
*device
, uint32_t *max_offchip_buffers_p
)
2323 bool double_offchip_buffers
= device
->physical_device
->rad_info
.chip_class
>= GFX7
&&
2324 device
->physical_device
->rad_info
.family
!= CHIP_CARRIZO
&&
2325 device
->physical_device
->rad_info
.family
!= CHIP_STONEY
;
2326 unsigned max_offchip_buffers_per_se
= double_offchip_buffers
? 128 : 64;
2327 unsigned max_offchip_buffers
;
2328 unsigned offchip_granularity
;
2329 unsigned hs_offchip_param
;
2333 * This must be one less than the maximum number due to a hw limitation.
2334 * Various hardware bugs need thGFX7
2337 * Vega10 should limit max_offchip_buffers to 508 (4 * 127).
2338 * Gfx7 should limit max_offchip_buffers to 508
2339 * Gfx6 should limit max_offchip_buffers to 126 (2 * 63)
2341 * Follow AMDVLK here.
2343 if (device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
2344 max_offchip_buffers_per_se
= 256;
2345 } else if (device
->physical_device
->rad_info
.family
== CHIP_VEGA10
||
2346 device
->physical_device
->rad_info
.chip_class
== GFX7
||
2347 device
->physical_device
->rad_info
.chip_class
== GFX6
)
2348 --max_offchip_buffers_per_se
;
2350 max_offchip_buffers
= max_offchip_buffers_per_se
*
2351 device
->physical_device
->rad_info
.max_se
;
2353 /* Hawaii has a bug with offchip buffers > 256 that can be worked
2354 * around by setting 4K granularity.
2356 if (device
->tess_offchip_block_dw_size
== 4096) {
2357 assert(device
->physical_device
->rad_info
.family
== CHIP_HAWAII
);
2358 offchip_granularity
= V_03093C_X_4K_DWORDS
;
2360 assert(device
->tess_offchip_block_dw_size
== 8192);
2361 offchip_granularity
= V_03093C_X_8K_DWORDS
;
2364 switch (device
->physical_device
->rad_info
.chip_class
) {
2366 max_offchip_buffers
= MIN2(max_offchip_buffers
, 126);
2371 max_offchip_buffers
= MIN2(max_offchip_buffers
, 508);
2379 *max_offchip_buffers_p
= max_offchip_buffers
;
2380 if (device
->physical_device
->rad_info
.chip_class
>= GFX7
) {
2381 if (device
->physical_device
->rad_info
.chip_class
>= GFX8
)
2382 --max_offchip_buffers
;
2384 S_03093C_OFFCHIP_BUFFERING(max_offchip_buffers
) |
2385 S_03093C_OFFCHIP_GRANULARITY(offchip_granularity
);
2388 S_0089B0_OFFCHIP_BUFFERING(max_offchip_buffers
);
2390 return hs_offchip_param
;
2394 radv_emit_gs_ring_sizes(struct radv_queue
*queue
, struct radeon_cmdbuf
*cs
,
2395 struct radeon_winsys_bo
*esgs_ring_bo
,
2396 uint32_t esgs_ring_size
,
2397 struct radeon_winsys_bo
*gsvs_ring_bo
,
2398 uint32_t gsvs_ring_size
)
2400 if (!esgs_ring_bo
&& !gsvs_ring_bo
)
2404 radv_cs_add_buffer(queue
->device
->ws
, cs
, esgs_ring_bo
);
2407 radv_cs_add_buffer(queue
->device
->ws
, cs
, gsvs_ring_bo
);
2409 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX7
) {
2410 radeon_set_uconfig_reg_seq(cs
, R_030900_VGT_ESGS_RING_SIZE
, 2);
2411 radeon_emit(cs
, esgs_ring_size
>> 8);
2412 radeon_emit(cs
, gsvs_ring_size
>> 8);
2414 radeon_set_config_reg_seq(cs
, R_0088C8_VGT_ESGS_RING_SIZE
, 2);
2415 radeon_emit(cs
, esgs_ring_size
>> 8);
2416 radeon_emit(cs
, gsvs_ring_size
>> 8);
2421 radv_emit_tess_factor_ring(struct radv_queue
*queue
, struct radeon_cmdbuf
*cs
,
2422 unsigned hs_offchip_param
, unsigned tf_ring_size
,
2423 struct radeon_winsys_bo
*tess_rings_bo
)
2430 tf_va
= radv_buffer_get_va(tess_rings_bo
);
2432 radv_cs_add_buffer(queue
->device
->ws
, cs
, tess_rings_bo
);
2434 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX7
) {
2435 radeon_set_uconfig_reg(cs
, R_030938_VGT_TF_RING_SIZE
,
2436 S_030938_SIZE(tf_ring_size
/ 4));
2437 radeon_set_uconfig_reg(cs
, R_030940_VGT_TF_MEMORY_BASE
,
2440 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
2441 radeon_set_uconfig_reg(cs
, R_030984_VGT_TF_MEMORY_BASE_HI_UMD
,
2442 S_030984_BASE_HI(tf_va
>> 40));
2443 } else if (queue
->device
->physical_device
->rad_info
.chip_class
== GFX9
) {
2444 radeon_set_uconfig_reg(cs
, R_030944_VGT_TF_MEMORY_BASE_HI
,
2445 S_030944_BASE_HI(tf_va
>> 40));
2447 radeon_set_uconfig_reg(cs
, R_03093C_VGT_HS_OFFCHIP_PARAM
,
2450 radeon_set_config_reg(cs
, R_008988_VGT_TF_RING_SIZE
,
2451 S_008988_SIZE(tf_ring_size
/ 4));
2452 radeon_set_config_reg(cs
, R_0089B8_VGT_TF_MEMORY_BASE
,
2454 radeon_set_config_reg(cs
, R_0089B0_VGT_HS_OFFCHIP_PARAM
,
2460 radv_emit_compute_scratch(struct radv_queue
*queue
, struct radeon_cmdbuf
*cs
,
2461 struct radeon_winsys_bo
*compute_scratch_bo
)
2463 uint64_t scratch_va
;
2465 if (!compute_scratch_bo
)
2468 scratch_va
= radv_buffer_get_va(compute_scratch_bo
);
2470 radv_cs_add_buffer(queue
->device
->ws
, cs
, compute_scratch_bo
);
2472 radeon_set_sh_reg_seq(cs
, R_00B900_COMPUTE_USER_DATA_0
, 2);
2473 radeon_emit(cs
, scratch_va
);
2474 radeon_emit(cs
, S_008F04_BASE_ADDRESS_HI(scratch_va
>> 32) |
2475 S_008F04_SWIZZLE_ENABLE(1));
2479 radv_emit_global_shader_pointers(struct radv_queue
*queue
,
2480 struct radeon_cmdbuf
*cs
,
2481 struct radeon_winsys_bo
*descriptor_bo
)
2488 va
= radv_buffer_get_va(descriptor_bo
);
2490 radv_cs_add_buffer(queue
->device
->ws
, cs
, descriptor_bo
);
2492 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
2493 uint32_t regs
[] = {R_00B030_SPI_SHADER_USER_DATA_PS_0
,
2494 R_00B130_SPI_SHADER_USER_DATA_VS_0
,
2495 R_00B208_SPI_SHADER_USER_DATA_ADDR_LO_GS
,
2496 R_00B408_SPI_SHADER_USER_DATA_ADDR_LO_HS
};
2498 for (int i
= 0; i
< ARRAY_SIZE(regs
); ++i
) {
2499 radv_emit_shader_pointer(queue
->device
, cs
, regs
[i
],
2502 } else if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX9
) {
2503 uint32_t regs
[] = {R_00B030_SPI_SHADER_USER_DATA_PS_0
,
2504 R_00B130_SPI_SHADER_USER_DATA_VS_0
,
2505 R_00B208_SPI_SHADER_USER_DATA_ADDR_LO_GS
,
2506 R_00B408_SPI_SHADER_USER_DATA_ADDR_LO_HS
};
2508 for (int i
= 0; i
< ARRAY_SIZE(regs
); ++i
) {
2509 radv_emit_shader_pointer(queue
->device
, cs
, regs
[i
],
2513 uint32_t regs
[] = {R_00B030_SPI_SHADER_USER_DATA_PS_0
,
2514 R_00B130_SPI_SHADER_USER_DATA_VS_0
,
2515 R_00B230_SPI_SHADER_USER_DATA_GS_0
,
2516 R_00B330_SPI_SHADER_USER_DATA_ES_0
,
2517 R_00B430_SPI_SHADER_USER_DATA_HS_0
,
2518 R_00B530_SPI_SHADER_USER_DATA_LS_0
};
2520 for (int i
= 0; i
< ARRAY_SIZE(regs
); ++i
) {
2521 radv_emit_shader_pointer(queue
->device
, cs
, regs
[i
],
2528 radv_init_graphics_state(struct radeon_cmdbuf
*cs
, struct radv_queue
*queue
)
2530 struct radv_device
*device
= queue
->device
;
2532 if (device
->gfx_init
) {
2533 uint64_t va
= radv_buffer_get_va(device
->gfx_init
);
2535 radeon_emit(cs
, PKT3(PKT3_INDIRECT_BUFFER_CIK
, 2, 0));
2536 radeon_emit(cs
, va
);
2537 radeon_emit(cs
, va
>> 32);
2538 radeon_emit(cs
, device
->gfx_init_size_dw
& 0xffff);
2540 radv_cs_add_buffer(device
->ws
, cs
, device
->gfx_init
);
2542 struct radv_physical_device
*physical_device
= device
->physical_device
;
2543 si_emit_graphics(physical_device
, cs
);
2548 radv_init_compute_state(struct radeon_cmdbuf
*cs
, struct radv_queue
*queue
)
2550 struct radv_physical_device
*physical_device
= queue
->device
->physical_device
;
2551 si_emit_compute(physical_device
, cs
);
2555 radv_get_preamble_cs(struct radv_queue
*queue
,
2556 uint32_t scratch_size
,
2557 uint32_t compute_scratch_size
,
2558 uint32_t esgs_ring_size
,
2559 uint32_t gsvs_ring_size
,
2560 bool needs_tess_rings
,
2561 bool needs_sample_positions
,
2562 struct radeon_cmdbuf
**initial_full_flush_preamble_cs
,
2563 struct radeon_cmdbuf
**initial_preamble_cs
,
2564 struct radeon_cmdbuf
**continue_preamble_cs
)
2566 struct radeon_winsys_bo
*scratch_bo
= NULL
;
2567 struct radeon_winsys_bo
*descriptor_bo
= NULL
;
2568 struct radeon_winsys_bo
*compute_scratch_bo
= NULL
;
2569 struct radeon_winsys_bo
*esgs_ring_bo
= NULL
;
2570 struct radeon_winsys_bo
*gsvs_ring_bo
= NULL
;
2571 struct radeon_winsys_bo
*tess_rings_bo
= NULL
;
2572 struct radeon_cmdbuf
*dest_cs
[3] = {0};
2573 bool add_tess_rings
= false, add_sample_positions
= false;
2574 unsigned tess_factor_ring_size
= 0, tess_offchip_ring_size
= 0;
2575 unsigned max_offchip_buffers
;
2576 unsigned hs_offchip_param
= 0;
2577 unsigned tess_offchip_ring_offset
;
2578 uint32_t ring_bo_flags
= RADEON_FLAG_NO_CPU_ACCESS
| RADEON_FLAG_NO_INTERPROCESS_SHARING
;
2579 if (!queue
->has_tess_rings
) {
2580 if (needs_tess_rings
)
2581 add_tess_rings
= true;
2583 if (!queue
->has_sample_positions
) {
2584 if (needs_sample_positions
)
2585 add_sample_positions
= true;
2587 tess_factor_ring_size
= 32768 * queue
->device
->physical_device
->rad_info
.max_se
;
2588 hs_offchip_param
= radv_get_hs_offchip_param(queue
->device
,
2589 &max_offchip_buffers
);
2590 tess_offchip_ring_offset
= align(tess_factor_ring_size
, 64 * 1024);
2591 tess_offchip_ring_size
= max_offchip_buffers
*
2592 queue
->device
->tess_offchip_block_dw_size
* 4;
2594 if (scratch_size
<= queue
->scratch_size
&&
2595 compute_scratch_size
<= queue
->compute_scratch_size
&&
2596 esgs_ring_size
<= queue
->esgs_ring_size
&&
2597 gsvs_ring_size
<= queue
->gsvs_ring_size
&&
2598 !add_tess_rings
&& !add_sample_positions
&&
2599 queue
->initial_preamble_cs
) {
2600 *initial_full_flush_preamble_cs
= queue
->initial_full_flush_preamble_cs
;
2601 *initial_preamble_cs
= queue
->initial_preamble_cs
;
2602 *continue_preamble_cs
= queue
->continue_preamble_cs
;
2603 if (!scratch_size
&& !compute_scratch_size
&& !esgs_ring_size
&& !gsvs_ring_size
)
2604 *continue_preamble_cs
= NULL
;
2608 if (scratch_size
> queue
->scratch_size
) {
2609 scratch_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2614 RADV_BO_PRIORITY_SCRATCH
);
2618 scratch_bo
= queue
->scratch_bo
;
2620 if (compute_scratch_size
> queue
->compute_scratch_size
) {
2621 compute_scratch_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2622 compute_scratch_size
,
2626 RADV_BO_PRIORITY_SCRATCH
);
2627 if (!compute_scratch_bo
)
2631 compute_scratch_bo
= queue
->compute_scratch_bo
;
2633 if (esgs_ring_size
> queue
->esgs_ring_size
) {
2634 esgs_ring_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2639 RADV_BO_PRIORITY_SCRATCH
);
2643 esgs_ring_bo
= queue
->esgs_ring_bo
;
2644 esgs_ring_size
= queue
->esgs_ring_size
;
2647 if (gsvs_ring_size
> queue
->gsvs_ring_size
) {
2648 gsvs_ring_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2653 RADV_BO_PRIORITY_SCRATCH
);
2657 gsvs_ring_bo
= queue
->gsvs_ring_bo
;
2658 gsvs_ring_size
= queue
->gsvs_ring_size
;
2661 if (add_tess_rings
) {
2662 tess_rings_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2663 tess_offchip_ring_offset
+ tess_offchip_ring_size
,
2667 RADV_BO_PRIORITY_SCRATCH
);
2671 tess_rings_bo
= queue
->tess_rings_bo
;
2674 if (scratch_bo
!= queue
->scratch_bo
||
2675 esgs_ring_bo
!= queue
->esgs_ring_bo
||
2676 gsvs_ring_bo
!= queue
->gsvs_ring_bo
||
2677 tess_rings_bo
!= queue
->tess_rings_bo
||
2678 add_sample_positions
) {
2680 if (gsvs_ring_bo
|| esgs_ring_bo
||
2681 tess_rings_bo
|| add_sample_positions
) {
2682 size
= 112; /* 2 dword + 2 padding + 4 dword * 6 */
2683 if (add_sample_positions
)
2684 size
+= 128; /* 64+32+16+8 = 120 bytes */
2686 else if (scratch_bo
)
2687 size
= 8; /* 2 dword */
2689 descriptor_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2693 RADEON_FLAG_CPU_ACCESS
|
2694 RADEON_FLAG_NO_INTERPROCESS_SHARING
|
2695 RADEON_FLAG_READ_ONLY
,
2696 RADV_BO_PRIORITY_DESCRIPTOR
);
2700 descriptor_bo
= queue
->descriptor_bo
;
2702 if (descriptor_bo
!= queue
->descriptor_bo
) {
2703 uint32_t *map
= (uint32_t*)queue
->device
->ws
->buffer_map(descriptor_bo
);
2706 uint64_t scratch_va
= radv_buffer_get_va(scratch_bo
);
2707 uint32_t rsrc1
= S_008F04_BASE_ADDRESS_HI(scratch_va
>> 32) |
2708 S_008F04_SWIZZLE_ENABLE(1);
2709 map
[0] = scratch_va
;
2713 if (esgs_ring_bo
|| gsvs_ring_bo
|| tess_rings_bo
|| add_sample_positions
)
2714 fill_geom_tess_rings(queue
, map
, add_sample_positions
,
2715 esgs_ring_size
, esgs_ring_bo
,
2716 gsvs_ring_size
, gsvs_ring_bo
,
2717 tess_factor_ring_size
,
2718 tess_offchip_ring_offset
,
2719 tess_offchip_ring_size
,
2722 queue
->device
->ws
->buffer_unmap(descriptor_bo
);
2725 for(int i
= 0; i
< 3; ++i
) {
2726 struct radeon_cmdbuf
*cs
= NULL
;
2727 cs
= queue
->device
->ws
->cs_create(queue
->device
->ws
,
2728 queue
->queue_family_index
? RING_COMPUTE
: RING_GFX
);
2735 radv_cs_add_buffer(queue
->device
->ws
, cs
, scratch_bo
);
2737 /* Emit initial configuration. */
2738 switch (queue
->queue_family_index
) {
2739 case RADV_QUEUE_GENERAL
:
2740 radv_init_graphics_state(cs
, queue
);
2742 case RADV_QUEUE_COMPUTE
:
2743 radv_init_compute_state(cs
, queue
);
2745 case RADV_QUEUE_TRANSFER
:
2749 if (esgs_ring_bo
|| gsvs_ring_bo
|| tess_rings_bo
) {
2750 radeon_emit(cs
, PKT3(PKT3_EVENT_WRITE
, 0, 0));
2751 radeon_emit(cs
, EVENT_TYPE(V_028A90_VS_PARTIAL_FLUSH
) | EVENT_INDEX(4));
2753 radeon_emit(cs
, PKT3(PKT3_EVENT_WRITE
, 0, 0));
2754 radeon_emit(cs
, EVENT_TYPE(V_028A90_VGT_FLUSH
) | EVENT_INDEX(0));
2757 radv_emit_gs_ring_sizes(queue
, cs
, esgs_ring_bo
, esgs_ring_size
,
2758 gsvs_ring_bo
, gsvs_ring_size
);
2759 radv_emit_tess_factor_ring(queue
, cs
, hs_offchip_param
,
2760 tess_factor_ring_size
, tess_rings_bo
);
2761 radv_emit_global_shader_pointers(queue
, cs
, descriptor_bo
);
2762 radv_emit_compute_scratch(queue
, cs
, compute_scratch_bo
);
2765 si_cs_emit_cache_flush(cs
,
2766 queue
->device
->physical_device
->rad_info
.chip_class
,
2768 queue
->queue_family_index
== RING_COMPUTE
&&
2769 queue
->device
->physical_device
->rad_info
.chip_class
>= GFX7
,
2770 (queue
->queue_family_index
== RADV_QUEUE_COMPUTE
? RADV_CMD_FLAG_CS_PARTIAL_FLUSH
: (RADV_CMD_FLAG_CS_PARTIAL_FLUSH
| RADV_CMD_FLAG_PS_PARTIAL_FLUSH
)) |
2771 RADV_CMD_FLAG_INV_ICACHE
|
2772 RADV_CMD_FLAG_INV_SCACHE
|
2773 RADV_CMD_FLAG_INV_VCACHE
|
2774 RADV_CMD_FLAG_INV_L2
|
2775 RADV_CMD_FLAG_START_PIPELINE_STATS
, 0);
2776 } else if (i
== 1) {
2777 si_cs_emit_cache_flush(cs
,
2778 queue
->device
->physical_device
->rad_info
.chip_class
,
2780 queue
->queue_family_index
== RING_COMPUTE
&&
2781 queue
->device
->physical_device
->rad_info
.chip_class
>= GFX7
,
2782 RADV_CMD_FLAG_INV_ICACHE
|
2783 RADV_CMD_FLAG_INV_SCACHE
|
2784 RADV_CMD_FLAG_INV_VCACHE
|
2785 RADV_CMD_FLAG_INV_L2
|
2786 RADV_CMD_FLAG_START_PIPELINE_STATS
, 0);
2789 if (!queue
->device
->ws
->cs_finalize(cs
))
2793 if (queue
->initial_full_flush_preamble_cs
)
2794 queue
->device
->ws
->cs_destroy(queue
->initial_full_flush_preamble_cs
);
2796 if (queue
->initial_preamble_cs
)
2797 queue
->device
->ws
->cs_destroy(queue
->initial_preamble_cs
);
2799 if (queue
->continue_preamble_cs
)
2800 queue
->device
->ws
->cs_destroy(queue
->continue_preamble_cs
);
2802 queue
->initial_full_flush_preamble_cs
= dest_cs
[0];
2803 queue
->initial_preamble_cs
= dest_cs
[1];
2804 queue
->continue_preamble_cs
= dest_cs
[2];
2806 if (scratch_bo
!= queue
->scratch_bo
) {
2807 if (queue
->scratch_bo
)
2808 queue
->device
->ws
->buffer_destroy(queue
->scratch_bo
);
2809 queue
->scratch_bo
= scratch_bo
;
2810 queue
->scratch_size
= scratch_size
;
2813 if (compute_scratch_bo
!= queue
->compute_scratch_bo
) {
2814 if (queue
->compute_scratch_bo
)
2815 queue
->device
->ws
->buffer_destroy(queue
->compute_scratch_bo
);
2816 queue
->compute_scratch_bo
= compute_scratch_bo
;
2817 queue
->compute_scratch_size
= compute_scratch_size
;
2820 if (esgs_ring_bo
!= queue
->esgs_ring_bo
) {
2821 if (queue
->esgs_ring_bo
)
2822 queue
->device
->ws
->buffer_destroy(queue
->esgs_ring_bo
);
2823 queue
->esgs_ring_bo
= esgs_ring_bo
;
2824 queue
->esgs_ring_size
= esgs_ring_size
;
2827 if (gsvs_ring_bo
!= queue
->gsvs_ring_bo
) {
2828 if (queue
->gsvs_ring_bo
)
2829 queue
->device
->ws
->buffer_destroy(queue
->gsvs_ring_bo
);
2830 queue
->gsvs_ring_bo
= gsvs_ring_bo
;
2831 queue
->gsvs_ring_size
= gsvs_ring_size
;
2834 if (tess_rings_bo
!= queue
->tess_rings_bo
) {
2835 queue
->tess_rings_bo
= tess_rings_bo
;
2836 queue
->has_tess_rings
= true;
2839 if (descriptor_bo
!= queue
->descriptor_bo
) {
2840 if (queue
->descriptor_bo
)
2841 queue
->device
->ws
->buffer_destroy(queue
->descriptor_bo
);
2843 queue
->descriptor_bo
= descriptor_bo
;
2846 if (add_sample_positions
)
2847 queue
->has_sample_positions
= true;
2849 *initial_full_flush_preamble_cs
= queue
->initial_full_flush_preamble_cs
;
2850 *initial_preamble_cs
= queue
->initial_preamble_cs
;
2851 *continue_preamble_cs
= queue
->continue_preamble_cs
;
2852 if (!scratch_size
&& !compute_scratch_size
&& !esgs_ring_size
&& !gsvs_ring_size
)
2853 *continue_preamble_cs
= NULL
;
2856 for (int i
= 0; i
< ARRAY_SIZE(dest_cs
); ++i
)
2858 queue
->device
->ws
->cs_destroy(dest_cs
[i
]);
2859 if (descriptor_bo
&& descriptor_bo
!= queue
->descriptor_bo
)
2860 queue
->device
->ws
->buffer_destroy(descriptor_bo
);
2861 if (scratch_bo
&& scratch_bo
!= queue
->scratch_bo
)
2862 queue
->device
->ws
->buffer_destroy(scratch_bo
);
2863 if (compute_scratch_bo
&& compute_scratch_bo
!= queue
->compute_scratch_bo
)
2864 queue
->device
->ws
->buffer_destroy(compute_scratch_bo
);
2865 if (esgs_ring_bo
&& esgs_ring_bo
!= queue
->esgs_ring_bo
)
2866 queue
->device
->ws
->buffer_destroy(esgs_ring_bo
);
2867 if (gsvs_ring_bo
&& gsvs_ring_bo
!= queue
->gsvs_ring_bo
)
2868 queue
->device
->ws
->buffer_destroy(gsvs_ring_bo
);
2869 if (tess_rings_bo
&& tess_rings_bo
!= queue
->tess_rings_bo
)
2870 queue
->device
->ws
->buffer_destroy(tess_rings_bo
);
2871 return vk_error(queue
->device
->instance
, VK_ERROR_OUT_OF_DEVICE_MEMORY
);
2874 static VkResult
radv_alloc_sem_counts(struct radv_instance
*instance
,
2875 struct radv_winsys_sem_counts
*counts
,
2877 const VkSemaphore
*sems
,
2881 int syncobj_idx
= 0, sem_idx
= 0;
2883 if (num_sems
== 0 && _fence
== VK_NULL_HANDLE
)
2886 for (uint32_t i
= 0; i
< num_sems
; i
++) {
2887 RADV_FROM_HANDLE(radv_semaphore
, sem
, sems
[i
]);
2889 if (sem
->temp_syncobj
|| sem
->syncobj
)
2890 counts
->syncobj_count
++;
2892 counts
->sem_count
++;
2895 if (_fence
!= VK_NULL_HANDLE
) {
2896 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
2897 if (fence
->temp_syncobj
|| fence
->syncobj
)
2898 counts
->syncobj_count
++;
2901 if (counts
->syncobj_count
) {
2902 counts
->syncobj
= (uint32_t *)malloc(sizeof(uint32_t) * counts
->syncobj_count
);
2903 if (!counts
->syncobj
)
2904 return vk_error(instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
2907 if (counts
->sem_count
) {
2908 counts
->sem
= (struct radeon_winsys_sem
**)malloc(sizeof(struct radeon_winsys_sem
*) * counts
->sem_count
);
2910 free(counts
->syncobj
);
2911 return vk_error(instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
2915 for (uint32_t i
= 0; i
< num_sems
; i
++) {
2916 RADV_FROM_HANDLE(radv_semaphore
, sem
, sems
[i
]);
2918 if (sem
->temp_syncobj
) {
2919 counts
->syncobj
[syncobj_idx
++] = sem
->temp_syncobj
;
2921 else if (sem
->syncobj
)
2922 counts
->syncobj
[syncobj_idx
++] = sem
->syncobj
;
2925 counts
->sem
[sem_idx
++] = sem
->sem
;
2929 if (_fence
!= VK_NULL_HANDLE
) {
2930 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
2931 if (fence
->temp_syncobj
)
2932 counts
->syncobj
[syncobj_idx
++] = fence
->temp_syncobj
;
2933 else if (fence
->syncobj
)
2934 counts
->syncobj
[syncobj_idx
++] = fence
->syncobj
;
2941 radv_free_sem_info(struct radv_winsys_sem_info
*sem_info
)
2943 free(sem_info
->wait
.syncobj
);
2944 free(sem_info
->wait
.sem
);
2945 free(sem_info
->signal
.syncobj
);
2946 free(sem_info
->signal
.sem
);
2950 static void radv_free_temp_syncobjs(struct radv_device
*device
,
2952 const VkSemaphore
*sems
)
2954 for (uint32_t i
= 0; i
< num_sems
; i
++) {
2955 RADV_FROM_HANDLE(radv_semaphore
, sem
, sems
[i
]);
2957 if (sem
->temp_syncobj
) {
2958 device
->ws
->destroy_syncobj(device
->ws
, sem
->temp_syncobj
);
2959 sem
->temp_syncobj
= 0;
2965 radv_alloc_sem_info(struct radv_instance
*instance
,
2966 struct radv_winsys_sem_info
*sem_info
,
2968 const VkSemaphore
*wait_sems
,
2969 int num_signal_sems
,
2970 const VkSemaphore
*signal_sems
,
2974 memset(sem_info
, 0, sizeof(*sem_info
));
2976 ret
= radv_alloc_sem_counts(instance
, &sem_info
->wait
, num_wait_sems
, wait_sems
, VK_NULL_HANDLE
, true);
2979 ret
= radv_alloc_sem_counts(instance
, &sem_info
->signal
, num_signal_sems
, signal_sems
, fence
, false);
2981 radv_free_sem_info(sem_info
);
2983 /* caller can override these */
2984 sem_info
->cs_emit_wait
= true;
2985 sem_info
->cs_emit_signal
= true;
2989 /* Signals fence as soon as all the work currently put on queue is done. */
2990 static VkResult
radv_signal_fence(struct radv_queue
*queue
,
2991 struct radv_fence
*fence
)
2995 struct radv_winsys_sem_info sem_info
;
2997 result
= radv_alloc_sem_info(queue
->device
->instance
, &sem_info
, 0, NULL
, 0, NULL
,
2998 radv_fence_to_handle(fence
));
2999 if (result
!= VK_SUCCESS
)
3002 ret
= queue
->device
->ws
->cs_submit(queue
->hw_ctx
, queue
->queue_idx
,
3003 &queue
->device
->empty_cs
[queue
->queue_family_index
],
3004 1, NULL
, NULL
, &sem_info
, NULL
,
3005 false, fence
->fence
);
3006 radv_free_sem_info(&sem_info
);
3009 return vk_error(queue
->device
->instance
, VK_ERROR_DEVICE_LOST
);
3014 VkResult
radv_QueueSubmit(
3016 uint32_t submitCount
,
3017 const VkSubmitInfo
* pSubmits
,
3020 RADV_FROM_HANDLE(radv_queue
, queue
, _queue
);
3021 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
3022 struct radeon_winsys_fence
*base_fence
= fence
? fence
->fence
: NULL
;
3023 struct radeon_winsys_ctx
*ctx
= queue
->hw_ctx
;
3025 uint32_t max_cs_submission
= queue
->device
->trace_bo
? 1 : RADV_MAX_IBS_PER_SUBMIT
;
3026 uint32_t scratch_size
= 0;
3027 uint32_t compute_scratch_size
= 0;
3028 uint32_t esgs_ring_size
= 0, gsvs_ring_size
= 0;
3029 struct radeon_cmdbuf
*initial_preamble_cs
= NULL
, *initial_flush_preamble_cs
= NULL
, *continue_preamble_cs
= NULL
;
3031 bool fence_emitted
= false;
3032 bool tess_rings_needed
= false;
3033 bool sample_positions_needed
= false;
3035 /* Do this first so failing to allocate scratch buffers can't result in
3036 * partially executed submissions. */
3037 for (uint32_t i
= 0; i
< submitCount
; i
++) {
3038 for (uint32_t j
= 0; j
< pSubmits
[i
].commandBufferCount
; j
++) {
3039 RADV_FROM_HANDLE(radv_cmd_buffer
, cmd_buffer
,
3040 pSubmits
[i
].pCommandBuffers
[j
]);
3042 scratch_size
= MAX2(scratch_size
, cmd_buffer
->scratch_size_needed
);
3043 compute_scratch_size
= MAX2(compute_scratch_size
,
3044 cmd_buffer
->compute_scratch_size_needed
);
3045 esgs_ring_size
= MAX2(esgs_ring_size
, cmd_buffer
->esgs_ring_size_needed
);
3046 gsvs_ring_size
= MAX2(gsvs_ring_size
, cmd_buffer
->gsvs_ring_size_needed
);
3047 tess_rings_needed
|= cmd_buffer
->tess_rings_needed
;
3048 sample_positions_needed
|= cmd_buffer
->sample_positions_needed
;
3052 result
= radv_get_preamble_cs(queue
, scratch_size
, compute_scratch_size
,
3053 esgs_ring_size
, gsvs_ring_size
, tess_rings_needed
,
3054 sample_positions_needed
, &initial_flush_preamble_cs
,
3055 &initial_preamble_cs
, &continue_preamble_cs
);
3056 if (result
!= VK_SUCCESS
)
3059 for (uint32_t i
= 0; i
< submitCount
; i
++) {
3060 struct radeon_cmdbuf
**cs_array
;
3061 bool do_flush
= !i
|| pSubmits
[i
].pWaitDstStageMask
;
3062 bool can_patch
= true;
3064 struct radv_winsys_sem_info sem_info
;
3066 result
= radv_alloc_sem_info(queue
->device
->instance
,
3068 pSubmits
[i
].waitSemaphoreCount
,
3069 pSubmits
[i
].pWaitSemaphores
,
3070 pSubmits
[i
].signalSemaphoreCount
,
3071 pSubmits
[i
].pSignalSemaphores
,
3073 if (result
!= VK_SUCCESS
)
3076 if (!pSubmits
[i
].commandBufferCount
) {
3077 if (pSubmits
[i
].waitSemaphoreCount
|| pSubmits
[i
].signalSemaphoreCount
) {
3078 ret
= queue
->device
->ws
->cs_submit(ctx
, queue
->queue_idx
,
3079 &queue
->device
->empty_cs
[queue
->queue_family_index
],
3084 radv_loge("failed to submit CS %d\n", i
);
3087 fence_emitted
= true;
3089 radv_free_sem_info(&sem_info
);
3093 cs_array
= malloc(sizeof(struct radeon_cmdbuf
*) *
3094 (pSubmits
[i
].commandBufferCount
));
3096 for (uint32_t j
= 0; j
< pSubmits
[i
].commandBufferCount
; j
++) {
3097 RADV_FROM_HANDLE(radv_cmd_buffer
, cmd_buffer
,
3098 pSubmits
[i
].pCommandBuffers
[j
]);
3099 assert(cmd_buffer
->level
== VK_COMMAND_BUFFER_LEVEL_PRIMARY
);
3101 cs_array
[j
] = cmd_buffer
->cs
;
3102 if ((cmd_buffer
->usage_flags
& VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT
))
3105 cmd_buffer
->status
= RADV_CMD_BUFFER_STATUS_PENDING
;
3108 for (uint32_t j
= 0; j
< pSubmits
[i
].commandBufferCount
; j
+= advance
) {
3109 struct radeon_cmdbuf
*initial_preamble
= (do_flush
&& !j
) ? initial_flush_preamble_cs
: initial_preamble_cs
;
3110 const struct radv_winsys_bo_list
*bo_list
= NULL
;
3112 advance
= MIN2(max_cs_submission
,
3113 pSubmits
[i
].commandBufferCount
- j
);
3115 if (queue
->device
->trace_bo
)
3116 *queue
->device
->trace_id_ptr
= 0;
3118 sem_info
.cs_emit_wait
= j
== 0;
3119 sem_info
.cs_emit_signal
= j
+ advance
== pSubmits
[i
].commandBufferCount
;
3121 if (unlikely(queue
->device
->use_global_bo_list
)) {
3122 pthread_mutex_lock(&queue
->device
->bo_list
.mutex
);
3123 bo_list
= &queue
->device
->bo_list
.list
;
3126 ret
= queue
->device
->ws
->cs_submit(ctx
, queue
->queue_idx
, cs_array
+ j
,
3127 advance
, initial_preamble
, continue_preamble_cs
,
3129 can_patch
, base_fence
);
3131 if (unlikely(queue
->device
->use_global_bo_list
))
3132 pthread_mutex_unlock(&queue
->device
->bo_list
.mutex
);
3135 radv_loge("failed to submit CS %d\n", i
);
3138 fence_emitted
= true;
3139 if (queue
->device
->trace_bo
) {
3140 radv_check_gpu_hangs(queue
, cs_array
[j
]);
3144 radv_free_temp_syncobjs(queue
->device
,
3145 pSubmits
[i
].waitSemaphoreCount
,
3146 pSubmits
[i
].pWaitSemaphores
);
3147 radv_free_sem_info(&sem_info
);
3152 if (!fence_emitted
) {
3153 result
= radv_signal_fence(queue
, fence
);
3154 if (result
!= VK_SUCCESS
)
3162 VkResult
radv_QueueWaitIdle(
3165 RADV_FROM_HANDLE(radv_queue
, queue
, _queue
);
3167 queue
->device
->ws
->ctx_wait_idle(queue
->hw_ctx
,
3168 radv_queue_family_to_ring(queue
->queue_family_index
),
3173 VkResult
radv_DeviceWaitIdle(
3176 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3178 for (unsigned i
= 0; i
< RADV_MAX_QUEUE_FAMILIES
; i
++) {
3179 for (unsigned q
= 0; q
< device
->queue_count
[i
]; q
++) {
3180 radv_QueueWaitIdle(radv_queue_to_handle(&device
->queues
[i
][q
]));
3186 VkResult
radv_EnumerateInstanceExtensionProperties(
3187 const char* pLayerName
,
3188 uint32_t* pPropertyCount
,
3189 VkExtensionProperties
* pProperties
)
3191 VK_OUTARRAY_MAKE(out
, pProperties
, pPropertyCount
);
3193 for (int i
= 0; i
< RADV_INSTANCE_EXTENSION_COUNT
; i
++) {
3194 if (radv_supported_instance_extensions
.extensions
[i
]) {
3195 vk_outarray_append(&out
, prop
) {
3196 *prop
= radv_instance_extensions
[i
];
3201 return vk_outarray_status(&out
);
3204 VkResult
radv_EnumerateDeviceExtensionProperties(
3205 VkPhysicalDevice physicalDevice
,
3206 const char* pLayerName
,
3207 uint32_t* pPropertyCount
,
3208 VkExtensionProperties
* pProperties
)
3210 RADV_FROM_HANDLE(radv_physical_device
, device
, physicalDevice
);
3211 VK_OUTARRAY_MAKE(out
, pProperties
, pPropertyCount
);
3213 for (int i
= 0; i
< RADV_DEVICE_EXTENSION_COUNT
; i
++) {
3214 if (device
->supported_extensions
.extensions
[i
]) {
3215 vk_outarray_append(&out
, prop
) {
3216 *prop
= radv_device_extensions
[i
];
3221 return vk_outarray_status(&out
);
3224 PFN_vkVoidFunction
radv_GetInstanceProcAddr(
3225 VkInstance _instance
,
3228 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
3230 return radv_lookup_entrypoint_checked(pName
,
3231 instance
? instance
->apiVersion
: 0,
3232 instance
? &instance
->enabled_extensions
: NULL
,
3236 /* The loader wants us to expose a second GetInstanceProcAddr function
3237 * to work around certain LD_PRELOAD issues seen in apps.
3240 VKAPI_ATTR PFN_vkVoidFunction VKAPI_CALL
vk_icdGetInstanceProcAddr(
3241 VkInstance instance
,
3245 VKAPI_ATTR PFN_vkVoidFunction VKAPI_CALL
vk_icdGetInstanceProcAddr(
3246 VkInstance instance
,
3249 return radv_GetInstanceProcAddr(instance
, pName
);
3253 VKAPI_ATTR PFN_vkVoidFunction VKAPI_CALL
vk_icdGetPhysicalDeviceProcAddr(
3254 VkInstance _instance
,
3258 VKAPI_ATTR PFN_vkVoidFunction VKAPI_CALL
vk_icdGetPhysicalDeviceProcAddr(
3259 VkInstance _instance
,
3262 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
3264 return radv_lookup_physical_device_entrypoint_checked(pName
,
3265 instance
? instance
->apiVersion
: 0,
3266 instance
? &instance
->enabled_extensions
: NULL
);
3269 PFN_vkVoidFunction
radv_GetDeviceProcAddr(
3273 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3275 return radv_lookup_entrypoint_checked(pName
,
3276 device
->instance
->apiVersion
,
3277 &device
->instance
->enabled_extensions
,
3278 &device
->enabled_extensions
);
3281 bool radv_get_memory_fd(struct radv_device
*device
,
3282 struct radv_device_memory
*memory
,
3285 struct radeon_bo_metadata metadata
;
3287 if (memory
->image
) {
3288 radv_init_metadata(device
, memory
->image
, &metadata
);
3289 device
->ws
->buffer_set_metadata(memory
->bo
, &metadata
);
3292 return device
->ws
->buffer_get_fd(device
->ws
, memory
->bo
,
3296 static VkResult
radv_alloc_memory(struct radv_device
*device
,
3297 const VkMemoryAllocateInfo
* pAllocateInfo
,
3298 const VkAllocationCallbacks
* pAllocator
,
3299 VkDeviceMemory
* pMem
)
3301 struct radv_device_memory
*mem
;
3303 enum radeon_bo_domain domain
;
3305 enum radv_mem_type mem_type_index
= device
->physical_device
->mem_type_indices
[pAllocateInfo
->memoryTypeIndex
];
3307 assert(pAllocateInfo
->sType
== VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO
);
3309 if (pAllocateInfo
->allocationSize
== 0) {
3310 /* Apparently, this is allowed */
3311 *pMem
= VK_NULL_HANDLE
;
3315 const VkImportMemoryFdInfoKHR
*import_info
=
3316 vk_find_struct_const(pAllocateInfo
->pNext
, IMPORT_MEMORY_FD_INFO_KHR
);
3317 const VkMemoryDedicatedAllocateInfo
*dedicate_info
=
3318 vk_find_struct_const(pAllocateInfo
->pNext
, MEMORY_DEDICATED_ALLOCATE_INFO
);
3319 const VkExportMemoryAllocateInfo
*export_info
=
3320 vk_find_struct_const(pAllocateInfo
->pNext
, EXPORT_MEMORY_ALLOCATE_INFO
);
3321 const VkImportMemoryHostPointerInfoEXT
*host_ptr_info
=
3322 vk_find_struct_const(pAllocateInfo
->pNext
, IMPORT_MEMORY_HOST_POINTER_INFO_EXT
);
3324 const struct wsi_memory_allocate_info
*wsi_info
=
3325 vk_find_struct_const(pAllocateInfo
->pNext
, WSI_MEMORY_ALLOCATE_INFO_MESA
);
3327 mem
= vk_alloc2(&device
->alloc
, pAllocator
, sizeof(*mem
), 8,
3328 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
3330 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3332 if (wsi_info
&& wsi_info
->implicit_sync
)
3333 flags
|= RADEON_FLAG_IMPLICIT_SYNC
;
3335 if (dedicate_info
) {
3336 mem
->image
= radv_image_from_handle(dedicate_info
->image
);
3337 mem
->buffer
= radv_buffer_from_handle(dedicate_info
->buffer
);
3343 float priority_float
= 0.5;
3344 const struct VkMemoryPriorityAllocateInfoEXT
*priority_ext
=
3345 vk_find_struct_const(pAllocateInfo
->pNext
,
3346 MEMORY_PRIORITY_ALLOCATE_INFO_EXT
);
3348 priority_float
= priority_ext
->priority
;
3350 unsigned priority
= MIN2(RADV_BO_PRIORITY_APPLICATION_MAX
- 1,
3351 (int)(priority_float
* RADV_BO_PRIORITY_APPLICATION_MAX
));
3353 mem
->user_ptr
= NULL
;
3356 assert(import_info
->handleType
==
3357 VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD_BIT
||
3358 import_info
->handleType
==
3359 VK_EXTERNAL_MEMORY_HANDLE_TYPE_DMA_BUF_BIT_EXT
);
3360 mem
->bo
= device
->ws
->buffer_from_fd(device
->ws
, import_info
->fd
,
3361 priority
, NULL
, NULL
);
3363 result
= VK_ERROR_INVALID_EXTERNAL_HANDLE
;
3366 close(import_info
->fd
);
3368 } else if (host_ptr_info
) {
3369 assert(host_ptr_info
->handleType
== VK_EXTERNAL_MEMORY_HANDLE_TYPE_HOST_ALLOCATION_BIT_EXT
);
3370 assert(mem_type_index
== RADV_MEM_TYPE_GTT_CACHED
);
3371 mem
->bo
= device
->ws
->buffer_from_ptr(device
->ws
, host_ptr_info
->pHostPointer
,
3372 pAllocateInfo
->allocationSize
,
3375 result
= VK_ERROR_INVALID_EXTERNAL_HANDLE
;
3378 mem
->user_ptr
= host_ptr_info
->pHostPointer
;
3381 uint64_t alloc_size
= align_u64(pAllocateInfo
->allocationSize
, 4096);
3382 if (mem_type_index
== RADV_MEM_TYPE_GTT_WRITE_COMBINE
||
3383 mem_type_index
== RADV_MEM_TYPE_GTT_CACHED
)
3384 domain
= RADEON_DOMAIN_GTT
;
3386 domain
= RADEON_DOMAIN_VRAM
;
3388 if (mem_type_index
== RADV_MEM_TYPE_VRAM
)
3389 flags
|= RADEON_FLAG_NO_CPU_ACCESS
;
3391 flags
|= RADEON_FLAG_CPU_ACCESS
;
3393 if (mem_type_index
== RADV_MEM_TYPE_GTT_WRITE_COMBINE
)
3394 flags
|= RADEON_FLAG_GTT_WC
;
3396 if (!dedicate_info
&& !import_info
&& (!export_info
|| !export_info
->handleTypes
)) {
3397 flags
|= RADEON_FLAG_NO_INTERPROCESS_SHARING
;
3398 if (device
->use_global_bo_list
) {
3399 flags
|= RADEON_FLAG_PREFER_LOCAL_BO
;
3403 mem
->bo
= device
->ws
->buffer_create(device
->ws
, alloc_size
, device
->physical_device
->rad_info
.max_alignment
,
3404 domain
, flags
, priority
);
3407 result
= VK_ERROR_OUT_OF_DEVICE_MEMORY
;
3410 mem
->type_index
= mem_type_index
;
3413 result
= radv_bo_list_add(device
, mem
->bo
);
3414 if (result
!= VK_SUCCESS
)
3417 *pMem
= radv_device_memory_to_handle(mem
);
3422 device
->ws
->buffer_destroy(mem
->bo
);
3424 vk_free2(&device
->alloc
, pAllocator
, mem
);
3429 VkResult
radv_AllocateMemory(
3431 const VkMemoryAllocateInfo
* pAllocateInfo
,
3432 const VkAllocationCallbacks
* pAllocator
,
3433 VkDeviceMemory
* pMem
)
3435 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3436 return radv_alloc_memory(device
, pAllocateInfo
, pAllocator
, pMem
);
3439 void radv_FreeMemory(
3441 VkDeviceMemory _mem
,
3442 const VkAllocationCallbacks
* pAllocator
)
3444 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3445 RADV_FROM_HANDLE(radv_device_memory
, mem
, _mem
);
3450 radv_bo_list_remove(device
, mem
->bo
);
3451 device
->ws
->buffer_destroy(mem
->bo
);
3454 vk_free2(&device
->alloc
, pAllocator
, mem
);
3457 VkResult
radv_MapMemory(
3459 VkDeviceMemory _memory
,
3460 VkDeviceSize offset
,
3462 VkMemoryMapFlags flags
,
3465 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3466 RADV_FROM_HANDLE(radv_device_memory
, mem
, _memory
);
3474 *ppData
= mem
->user_ptr
;
3476 *ppData
= device
->ws
->buffer_map(mem
->bo
);
3483 return vk_error(device
->instance
, VK_ERROR_MEMORY_MAP_FAILED
);
3486 void radv_UnmapMemory(
3488 VkDeviceMemory _memory
)
3490 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3491 RADV_FROM_HANDLE(radv_device_memory
, mem
, _memory
);
3496 if (mem
->user_ptr
== NULL
)
3497 device
->ws
->buffer_unmap(mem
->bo
);
3500 VkResult
radv_FlushMappedMemoryRanges(
3502 uint32_t memoryRangeCount
,
3503 const VkMappedMemoryRange
* pMemoryRanges
)
3508 VkResult
radv_InvalidateMappedMemoryRanges(
3510 uint32_t memoryRangeCount
,
3511 const VkMappedMemoryRange
* pMemoryRanges
)
3516 void radv_GetBufferMemoryRequirements(
3519 VkMemoryRequirements
* pMemoryRequirements
)
3521 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3522 RADV_FROM_HANDLE(radv_buffer
, buffer
, _buffer
);
3524 pMemoryRequirements
->memoryTypeBits
= (1u << device
->physical_device
->memory_properties
.memoryTypeCount
) - 1;
3526 if (buffer
->flags
& VK_BUFFER_CREATE_SPARSE_BINDING_BIT
)
3527 pMemoryRequirements
->alignment
= 4096;
3529 pMemoryRequirements
->alignment
= 16;
3531 pMemoryRequirements
->size
= align64(buffer
->size
, pMemoryRequirements
->alignment
);
3534 void radv_GetBufferMemoryRequirements2(
3536 const VkBufferMemoryRequirementsInfo2
*pInfo
,
3537 VkMemoryRequirements2
*pMemoryRequirements
)
3539 radv_GetBufferMemoryRequirements(device
, pInfo
->buffer
,
3540 &pMemoryRequirements
->memoryRequirements
);
3541 RADV_FROM_HANDLE(radv_buffer
, buffer
, pInfo
->buffer
);
3542 vk_foreach_struct(ext
, pMemoryRequirements
->pNext
) {
3543 switch (ext
->sType
) {
3544 case VK_STRUCTURE_TYPE_MEMORY_DEDICATED_REQUIREMENTS
: {
3545 VkMemoryDedicatedRequirements
*req
=
3546 (VkMemoryDedicatedRequirements
*) ext
;
3547 req
->requiresDedicatedAllocation
= buffer
->shareable
;
3548 req
->prefersDedicatedAllocation
= req
->requiresDedicatedAllocation
;
3557 void radv_GetImageMemoryRequirements(
3560 VkMemoryRequirements
* pMemoryRequirements
)
3562 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3563 RADV_FROM_HANDLE(radv_image
, image
, _image
);
3565 pMemoryRequirements
->memoryTypeBits
= (1u << device
->physical_device
->memory_properties
.memoryTypeCount
) - 1;
3567 pMemoryRequirements
->size
= image
->size
;
3568 pMemoryRequirements
->alignment
= image
->alignment
;
3571 void radv_GetImageMemoryRequirements2(
3573 const VkImageMemoryRequirementsInfo2
*pInfo
,
3574 VkMemoryRequirements2
*pMemoryRequirements
)
3576 radv_GetImageMemoryRequirements(device
, pInfo
->image
,
3577 &pMemoryRequirements
->memoryRequirements
);
3579 RADV_FROM_HANDLE(radv_image
, image
, pInfo
->image
);
3581 vk_foreach_struct(ext
, pMemoryRequirements
->pNext
) {
3582 switch (ext
->sType
) {
3583 case VK_STRUCTURE_TYPE_MEMORY_DEDICATED_REQUIREMENTS
: {
3584 VkMemoryDedicatedRequirements
*req
=
3585 (VkMemoryDedicatedRequirements
*) ext
;
3586 req
->requiresDedicatedAllocation
= image
->shareable
;
3587 req
->prefersDedicatedAllocation
= req
->requiresDedicatedAllocation
;
3596 void radv_GetImageSparseMemoryRequirements(
3599 uint32_t* pSparseMemoryRequirementCount
,
3600 VkSparseImageMemoryRequirements
* pSparseMemoryRequirements
)
3605 void radv_GetImageSparseMemoryRequirements2(
3607 const VkImageSparseMemoryRequirementsInfo2
*pInfo
,
3608 uint32_t* pSparseMemoryRequirementCount
,
3609 VkSparseImageMemoryRequirements2
*pSparseMemoryRequirements
)
3614 void radv_GetDeviceMemoryCommitment(
3616 VkDeviceMemory memory
,
3617 VkDeviceSize
* pCommittedMemoryInBytes
)
3619 *pCommittedMemoryInBytes
= 0;
3622 VkResult
radv_BindBufferMemory2(VkDevice device
,
3623 uint32_t bindInfoCount
,
3624 const VkBindBufferMemoryInfo
*pBindInfos
)
3626 for (uint32_t i
= 0; i
< bindInfoCount
; ++i
) {
3627 RADV_FROM_HANDLE(radv_device_memory
, mem
, pBindInfos
[i
].memory
);
3628 RADV_FROM_HANDLE(radv_buffer
, buffer
, pBindInfos
[i
].buffer
);
3631 buffer
->bo
= mem
->bo
;
3632 buffer
->offset
= pBindInfos
[i
].memoryOffset
;
3640 VkResult
radv_BindBufferMemory(
3643 VkDeviceMemory memory
,
3644 VkDeviceSize memoryOffset
)
3646 const VkBindBufferMemoryInfo info
= {
3647 .sType
= VK_STRUCTURE_TYPE_BIND_BUFFER_MEMORY_INFO
,
3650 .memoryOffset
= memoryOffset
3653 return radv_BindBufferMemory2(device
, 1, &info
);
3656 VkResult
radv_BindImageMemory2(VkDevice device
,
3657 uint32_t bindInfoCount
,
3658 const VkBindImageMemoryInfo
*pBindInfos
)
3660 for (uint32_t i
= 0; i
< bindInfoCount
; ++i
) {
3661 RADV_FROM_HANDLE(radv_device_memory
, mem
, pBindInfos
[i
].memory
);
3662 RADV_FROM_HANDLE(radv_image
, image
, pBindInfos
[i
].image
);
3665 image
->bo
= mem
->bo
;
3666 image
->offset
= pBindInfos
[i
].memoryOffset
;
3676 VkResult
radv_BindImageMemory(
3679 VkDeviceMemory memory
,
3680 VkDeviceSize memoryOffset
)
3682 const VkBindImageMemoryInfo info
= {
3683 .sType
= VK_STRUCTURE_TYPE_BIND_BUFFER_MEMORY_INFO
,
3686 .memoryOffset
= memoryOffset
3689 return radv_BindImageMemory2(device
, 1, &info
);
3694 radv_sparse_buffer_bind_memory(struct radv_device
*device
,
3695 const VkSparseBufferMemoryBindInfo
*bind
)
3697 RADV_FROM_HANDLE(radv_buffer
, buffer
, bind
->buffer
);
3699 for (uint32_t i
= 0; i
< bind
->bindCount
; ++i
) {
3700 struct radv_device_memory
*mem
= NULL
;
3702 if (bind
->pBinds
[i
].memory
!= VK_NULL_HANDLE
)
3703 mem
= radv_device_memory_from_handle(bind
->pBinds
[i
].memory
);
3705 device
->ws
->buffer_virtual_bind(buffer
->bo
,
3706 bind
->pBinds
[i
].resourceOffset
,
3707 bind
->pBinds
[i
].size
,
3708 mem
? mem
->bo
: NULL
,
3709 bind
->pBinds
[i
].memoryOffset
);
3714 radv_sparse_image_opaque_bind_memory(struct radv_device
*device
,
3715 const VkSparseImageOpaqueMemoryBindInfo
*bind
)
3717 RADV_FROM_HANDLE(radv_image
, image
, bind
->image
);
3719 for (uint32_t i
= 0; i
< bind
->bindCount
; ++i
) {
3720 struct radv_device_memory
*mem
= NULL
;
3722 if (bind
->pBinds
[i
].memory
!= VK_NULL_HANDLE
)
3723 mem
= radv_device_memory_from_handle(bind
->pBinds
[i
].memory
);
3725 device
->ws
->buffer_virtual_bind(image
->bo
,
3726 bind
->pBinds
[i
].resourceOffset
,
3727 bind
->pBinds
[i
].size
,
3728 mem
? mem
->bo
: NULL
,
3729 bind
->pBinds
[i
].memoryOffset
);
3733 VkResult
radv_QueueBindSparse(
3735 uint32_t bindInfoCount
,
3736 const VkBindSparseInfo
* pBindInfo
,
3739 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
3740 RADV_FROM_HANDLE(radv_queue
, queue
, _queue
);
3741 struct radeon_winsys_fence
*base_fence
= fence
? fence
->fence
: NULL
;
3742 bool fence_emitted
= false;
3746 for (uint32_t i
= 0; i
< bindInfoCount
; ++i
) {
3747 struct radv_winsys_sem_info sem_info
;
3748 for (uint32_t j
= 0; j
< pBindInfo
[i
].bufferBindCount
; ++j
) {
3749 radv_sparse_buffer_bind_memory(queue
->device
,
3750 pBindInfo
[i
].pBufferBinds
+ j
);
3753 for (uint32_t j
= 0; j
< pBindInfo
[i
].imageOpaqueBindCount
; ++j
) {
3754 radv_sparse_image_opaque_bind_memory(queue
->device
,
3755 pBindInfo
[i
].pImageOpaqueBinds
+ j
);
3759 result
= radv_alloc_sem_info(queue
->device
->instance
,
3761 pBindInfo
[i
].waitSemaphoreCount
,
3762 pBindInfo
[i
].pWaitSemaphores
,
3763 pBindInfo
[i
].signalSemaphoreCount
,
3764 pBindInfo
[i
].pSignalSemaphores
,
3766 if (result
!= VK_SUCCESS
)
3769 if (pBindInfo
[i
].waitSemaphoreCount
|| pBindInfo
[i
].signalSemaphoreCount
) {
3770 ret
= queue
->device
->ws
->cs_submit(queue
->hw_ctx
, queue
->queue_idx
,
3771 &queue
->device
->empty_cs
[queue
->queue_family_index
],
3776 radv_loge("failed to submit CS %d\n", i
);
3780 fence_emitted
= true;
3783 radv_free_sem_info(&sem_info
);
3788 if (!fence_emitted
) {
3789 result
= radv_signal_fence(queue
, fence
);
3790 if (result
!= VK_SUCCESS
)
3798 VkResult
radv_CreateFence(
3800 const VkFenceCreateInfo
* pCreateInfo
,
3801 const VkAllocationCallbacks
* pAllocator
,
3804 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3805 const VkExportFenceCreateInfo
*export
=
3806 vk_find_struct_const(pCreateInfo
->pNext
, EXPORT_FENCE_CREATE_INFO
);
3807 VkExternalFenceHandleTypeFlags handleTypes
=
3808 export
? export
->handleTypes
: 0;
3810 struct radv_fence
*fence
= vk_alloc2(&device
->alloc
, pAllocator
,
3812 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
3815 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3817 fence
->fence_wsi
= NULL
;
3818 fence
->temp_syncobj
= 0;
3819 if (device
->always_use_syncobj
|| handleTypes
) {
3820 int ret
= device
->ws
->create_syncobj(device
->ws
, &fence
->syncobj
);
3822 vk_free2(&device
->alloc
, pAllocator
, fence
);
3823 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3825 if (pCreateInfo
->flags
& VK_FENCE_CREATE_SIGNALED_BIT
) {
3826 device
->ws
->signal_syncobj(device
->ws
, fence
->syncobj
);
3828 fence
->fence
= NULL
;
3830 fence
->fence
= device
->ws
->create_fence();
3831 if (!fence
->fence
) {
3832 vk_free2(&device
->alloc
, pAllocator
, fence
);
3833 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3836 if (pCreateInfo
->flags
& VK_FENCE_CREATE_SIGNALED_BIT
)
3837 device
->ws
->signal_fence(fence
->fence
);
3840 *pFence
= radv_fence_to_handle(fence
);
3845 void radv_DestroyFence(
3848 const VkAllocationCallbacks
* pAllocator
)
3850 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3851 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
3856 if (fence
->temp_syncobj
)
3857 device
->ws
->destroy_syncobj(device
->ws
, fence
->temp_syncobj
);
3859 device
->ws
->destroy_syncobj(device
->ws
, fence
->syncobj
);
3861 device
->ws
->destroy_fence(fence
->fence
);
3862 if (fence
->fence_wsi
)
3863 fence
->fence_wsi
->destroy(fence
->fence_wsi
);
3864 vk_free2(&device
->alloc
, pAllocator
, fence
);
3868 uint64_t radv_get_current_time(void)
3871 clock_gettime(CLOCK_MONOTONIC
, &tv
);
3872 return tv
.tv_nsec
+ tv
.tv_sec
*1000000000ull;
3875 static uint64_t radv_get_absolute_timeout(uint64_t timeout
)
3877 uint64_t current_time
= radv_get_current_time();
3879 timeout
= MIN2(UINT64_MAX
- current_time
, timeout
);
3881 return current_time
+ timeout
;
3885 static bool radv_all_fences_plain_and_submitted(struct radv_device
*device
,
3886 uint32_t fenceCount
, const VkFence
*pFences
)
3888 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3889 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
3890 if (fence
->fence
== NULL
|| fence
->syncobj
||
3891 fence
->temp_syncobj
|| fence
->fence_wsi
||
3892 (!device
->ws
->is_fence_waitable(fence
->fence
)))
3898 static bool radv_all_fences_syncobj(uint32_t fenceCount
, const VkFence
*pFences
)
3900 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3901 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
3902 if (fence
->syncobj
== 0 && fence
->temp_syncobj
== 0)
3908 VkResult
radv_WaitForFences(
3910 uint32_t fenceCount
,
3911 const VkFence
* pFences
,
3915 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3916 timeout
= radv_get_absolute_timeout(timeout
);
3918 if (device
->always_use_syncobj
&&
3919 radv_all_fences_syncobj(fenceCount
, pFences
))
3921 uint32_t *handles
= malloc(sizeof(uint32_t) * fenceCount
);
3923 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3925 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3926 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
3927 handles
[i
] = fence
->temp_syncobj
? fence
->temp_syncobj
: fence
->syncobj
;
3930 bool success
= device
->ws
->wait_syncobj(device
->ws
, handles
, fenceCount
, waitAll
, timeout
);
3933 return success
? VK_SUCCESS
: VK_TIMEOUT
;
3936 if (!waitAll
&& fenceCount
> 1) {
3937 /* Not doing this by default for waitAll, due to needing to allocate twice. */
3938 if (device
->physical_device
->rad_info
.drm_minor
>= 10 && radv_all_fences_plain_and_submitted(device
, fenceCount
, pFences
)) {
3939 uint32_t wait_count
= 0;
3940 struct radeon_winsys_fence
**fences
= malloc(sizeof(struct radeon_winsys_fence
*) * fenceCount
);
3942 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3944 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3945 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
3947 if (device
->ws
->fence_wait(device
->ws
, fence
->fence
, false, 0)) {
3952 fences
[wait_count
++] = fence
->fence
;
3955 bool success
= device
->ws
->fences_wait(device
->ws
, fences
, wait_count
,
3956 waitAll
, timeout
- radv_get_current_time());
3959 return success
? VK_SUCCESS
: VK_TIMEOUT
;
3962 while(radv_get_current_time() <= timeout
) {
3963 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3964 if (radv_GetFenceStatus(_device
, pFences
[i
]) == VK_SUCCESS
)
3971 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3972 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
3973 bool expired
= false;
3975 if (fence
->temp_syncobj
) {
3976 if (!device
->ws
->wait_syncobj(device
->ws
, &fence
->temp_syncobj
, 1, true, timeout
))
3981 if (fence
->syncobj
) {
3982 if (!device
->ws
->wait_syncobj(device
->ws
, &fence
->syncobj
, 1, true, timeout
))
3988 if (!device
->ws
->is_fence_waitable(fence
->fence
)) {
3989 while(!device
->ws
->is_fence_waitable(fence
->fence
) &&
3990 radv_get_current_time() <= timeout
)
3994 expired
= device
->ws
->fence_wait(device
->ws
,
4001 if (fence
->fence_wsi
) {
4002 VkResult result
= fence
->fence_wsi
->wait(fence
->fence_wsi
, timeout
);
4003 if (result
!= VK_SUCCESS
)
4011 VkResult
radv_ResetFences(VkDevice _device
,
4012 uint32_t fenceCount
,
4013 const VkFence
*pFences
)
4015 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4017 for (unsigned i
= 0; i
< fenceCount
; ++i
) {
4018 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
4020 device
->ws
->reset_fence(fence
->fence
);
4022 /* Per spec, we first restore the permanent payload, and then reset, so
4023 * having a temp syncobj should not skip resetting the permanent syncobj. */
4024 if (fence
->temp_syncobj
) {
4025 device
->ws
->destroy_syncobj(device
->ws
, fence
->temp_syncobj
);
4026 fence
->temp_syncobj
= 0;
4029 if (fence
->syncobj
) {
4030 device
->ws
->reset_syncobj(device
->ws
, fence
->syncobj
);
4037 VkResult
radv_GetFenceStatus(VkDevice _device
, VkFence _fence
)
4039 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4040 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
4042 if (fence
->temp_syncobj
) {
4043 bool success
= device
->ws
->wait_syncobj(device
->ws
, &fence
->temp_syncobj
, 1, true, 0);
4044 return success
? VK_SUCCESS
: VK_NOT_READY
;
4047 if (fence
->syncobj
) {
4048 bool success
= device
->ws
->wait_syncobj(device
->ws
, &fence
->syncobj
, 1, true, 0);
4049 return success
? VK_SUCCESS
: VK_NOT_READY
;
4053 if (!device
->ws
->fence_wait(device
->ws
, fence
->fence
, false, 0))
4054 return VK_NOT_READY
;
4056 if (fence
->fence_wsi
) {
4057 VkResult result
= fence
->fence_wsi
->wait(fence
->fence_wsi
, 0);
4059 if (result
!= VK_SUCCESS
) {
4060 if (result
== VK_TIMEOUT
)
4061 return VK_NOT_READY
;
4069 // Queue semaphore functions
4071 VkResult
radv_CreateSemaphore(
4073 const VkSemaphoreCreateInfo
* pCreateInfo
,
4074 const VkAllocationCallbacks
* pAllocator
,
4075 VkSemaphore
* pSemaphore
)
4077 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4078 const VkExportSemaphoreCreateInfo
*export
=
4079 vk_find_struct_const(pCreateInfo
->pNext
, EXPORT_SEMAPHORE_CREATE_INFO
);
4080 VkExternalSemaphoreHandleTypeFlags handleTypes
=
4081 export
? export
->handleTypes
: 0;
4083 struct radv_semaphore
*sem
= vk_alloc2(&device
->alloc
, pAllocator
,
4085 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
4087 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
4089 sem
->temp_syncobj
= 0;
4090 /* create a syncobject if we are going to export this semaphore */
4091 if (device
->always_use_syncobj
|| handleTypes
) {
4092 assert (device
->physical_device
->rad_info
.has_syncobj
);
4093 int ret
= device
->ws
->create_syncobj(device
->ws
, &sem
->syncobj
);
4095 vk_free2(&device
->alloc
, pAllocator
, sem
);
4096 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
4100 sem
->sem
= device
->ws
->create_sem(device
->ws
);
4102 vk_free2(&device
->alloc
, pAllocator
, sem
);
4103 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
4108 *pSemaphore
= radv_semaphore_to_handle(sem
);
4112 void radv_DestroySemaphore(
4114 VkSemaphore _semaphore
,
4115 const VkAllocationCallbacks
* pAllocator
)
4117 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4118 RADV_FROM_HANDLE(radv_semaphore
, sem
, _semaphore
);
4123 device
->ws
->destroy_syncobj(device
->ws
, sem
->syncobj
);
4125 device
->ws
->destroy_sem(sem
->sem
);
4126 vk_free2(&device
->alloc
, pAllocator
, sem
);
4129 VkResult
radv_CreateEvent(
4131 const VkEventCreateInfo
* pCreateInfo
,
4132 const VkAllocationCallbacks
* pAllocator
,
4135 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4136 struct radv_event
*event
= vk_alloc2(&device
->alloc
, pAllocator
,
4138 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
4141 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
4143 event
->bo
= device
->ws
->buffer_create(device
->ws
, 8, 8,
4145 RADEON_FLAG_VA_UNCACHED
| RADEON_FLAG_CPU_ACCESS
| RADEON_FLAG_NO_INTERPROCESS_SHARING
,
4146 RADV_BO_PRIORITY_FENCE
);
4148 vk_free2(&device
->alloc
, pAllocator
, event
);
4149 return vk_error(device
->instance
, VK_ERROR_OUT_OF_DEVICE_MEMORY
);
4152 event
->map
= (uint64_t*)device
->ws
->buffer_map(event
->bo
);
4154 *pEvent
= radv_event_to_handle(event
);
4159 void radv_DestroyEvent(
4162 const VkAllocationCallbacks
* pAllocator
)
4164 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4165 RADV_FROM_HANDLE(radv_event
, event
, _event
);
4169 device
->ws
->buffer_destroy(event
->bo
);
4170 vk_free2(&device
->alloc
, pAllocator
, event
);
4173 VkResult
radv_GetEventStatus(
4177 RADV_FROM_HANDLE(radv_event
, event
, _event
);
4179 if (*event
->map
== 1)
4180 return VK_EVENT_SET
;
4181 return VK_EVENT_RESET
;
4184 VkResult
radv_SetEvent(
4188 RADV_FROM_HANDLE(radv_event
, event
, _event
);
4194 VkResult
radv_ResetEvent(
4198 RADV_FROM_HANDLE(radv_event
, event
, _event
);
4204 VkResult
radv_CreateBuffer(
4206 const VkBufferCreateInfo
* pCreateInfo
,
4207 const VkAllocationCallbacks
* pAllocator
,
4210 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4211 struct radv_buffer
*buffer
;
4213 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO
);
4215 buffer
= vk_alloc2(&device
->alloc
, pAllocator
, sizeof(*buffer
), 8,
4216 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
4218 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
4220 buffer
->size
= pCreateInfo
->size
;
4221 buffer
->usage
= pCreateInfo
->usage
;
4224 buffer
->flags
= pCreateInfo
->flags
;
4226 buffer
->shareable
= vk_find_struct_const(pCreateInfo
->pNext
,
4227 EXTERNAL_MEMORY_BUFFER_CREATE_INFO
) != NULL
;
4229 if (pCreateInfo
->flags
& VK_BUFFER_CREATE_SPARSE_BINDING_BIT
) {
4230 buffer
->bo
= device
->ws
->buffer_create(device
->ws
,
4231 align64(buffer
->size
, 4096),
4232 4096, 0, RADEON_FLAG_VIRTUAL
,
4233 RADV_BO_PRIORITY_VIRTUAL
);
4235 vk_free2(&device
->alloc
, pAllocator
, buffer
);
4236 return vk_error(device
->instance
, VK_ERROR_OUT_OF_DEVICE_MEMORY
);
4240 *pBuffer
= radv_buffer_to_handle(buffer
);
4245 void radv_DestroyBuffer(
4248 const VkAllocationCallbacks
* pAllocator
)
4250 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4251 RADV_FROM_HANDLE(radv_buffer
, buffer
, _buffer
);
4256 if (buffer
->flags
& VK_BUFFER_CREATE_SPARSE_BINDING_BIT
)
4257 device
->ws
->buffer_destroy(buffer
->bo
);
4259 vk_free2(&device
->alloc
, pAllocator
, buffer
);
4262 VkDeviceAddress
radv_GetBufferDeviceAddressEXT(
4264 const VkBufferDeviceAddressInfoEXT
* pInfo
)
4266 RADV_FROM_HANDLE(radv_buffer
, buffer
, pInfo
->buffer
);
4267 return radv_buffer_get_va(buffer
->bo
) + buffer
->offset
;
4271 static inline unsigned
4272 si_tile_mode_index(const struct radv_image_plane
*plane
, unsigned level
, bool stencil
)
4275 return plane
->surface
.u
.legacy
.stencil_tiling_index
[level
];
4277 return plane
->surface
.u
.legacy
.tiling_index
[level
];
4280 static uint32_t radv_surface_max_layer_count(struct radv_image_view
*iview
)
4282 return iview
->type
== VK_IMAGE_VIEW_TYPE_3D
? iview
->extent
.depth
: (iview
->base_layer
+ iview
->layer_count
);
4286 radv_init_dcc_control_reg(struct radv_device
*device
,
4287 struct radv_image_view
*iview
)
4289 unsigned max_uncompressed_block_size
= V_028C78_MAX_BLOCK_SIZE_256B
;
4290 unsigned min_compressed_block_size
= V_028C78_MIN_BLOCK_SIZE_32B
;
4291 unsigned max_compressed_block_size
;
4292 unsigned independent_128b_blocks
;
4293 unsigned independent_64b_blocks
;
4295 if (!radv_dcc_enabled(iview
->image
, iview
->base_mip
))
4298 if (!device
->physical_device
->rad_info
.has_dedicated_vram
) {
4299 /* amdvlk: [min-compressed-block-size] should be set to 32 for
4300 * dGPU and 64 for APU because all of our APUs to date use
4301 * DIMMs which have a request granularity size of 64B while all
4302 * other chips have a 32B request size.
4304 min_compressed_block_size
= V_028C78_MIN_BLOCK_SIZE_64B
;
4307 if (device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
4308 max_compressed_block_size
= V_028C78_MAX_BLOCK_SIZE_128B
;
4309 independent_64b_blocks
= 0;
4310 independent_128b_blocks
= 1;
4312 independent_128b_blocks
= 0;
4314 if (iview
->image
->info
.samples
> 1) {
4315 if (iview
->image
->planes
[0].surface
.bpe
== 1)
4316 max_uncompressed_block_size
= V_028C78_MAX_BLOCK_SIZE_64B
;
4317 else if (iview
->image
->planes
[0].surface
.bpe
== 2)
4318 max_uncompressed_block_size
= V_028C78_MAX_BLOCK_SIZE_128B
;
4321 if (iview
->image
->usage
& (VK_IMAGE_USAGE_SAMPLED_BIT
|
4322 VK_IMAGE_USAGE_TRANSFER_SRC_BIT
|
4323 VK_IMAGE_USAGE_INPUT_ATTACHMENT_BIT
)) {
4324 /* If this DCC image is potentially going to be used in texture
4325 * fetches, we need some special settings.
4327 independent_64b_blocks
= 1;
4328 max_compressed_block_size
= V_028C78_MAX_BLOCK_SIZE_64B
;
4330 /* MAX_UNCOMPRESSED_BLOCK_SIZE must be >=
4331 * MAX_COMPRESSED_BLOCK_SIZE. Set MAX_COMPRESSED_BLOCK_SIZE as
4332 * big as possible for better compression state.
4334 independent_64b_blocks
= 0;
4335 max_compressed_block_size
= max_uncompressed_block_size
;
4339 return S_028C78_MAX_UNCOMPRESSED_BLOCK_SIZE(max_uncompressed_block_size
) |
4340 S_028C78_MAX_COMPRESSED_BLOCK_SIZE(max_compressed_block_size
) |
4341 S_028C78_MIN_COMPRESSED_BLOCK_SIZE(min_compressed_block_size
) |
4342 S_028C78_INDEPENDENT_64B_BLOCKS(independent_64b_blocks
) |
4343 S_028C78_INDEPENDENT_128B_BLOCKS(independent_128b_blocks
);
4347 radv_initialise_color_surface(struct radv_device
*device
,
4348 struct radv_color_buffer_info
*cb
,
4349 struct radv_image_view
*iview
)
4351 const struct vk_format_description
*desc
;
4352 unsigned ntype
, format
, swap
, endian
;
4353 unsigned blend_clamp
= 0, blend_bypass
= 0;
4355 const struct radv_image_plane
*plane
= &iview
->image
->planes
[iview
->plane_id
];
4356 const struct radeon_surf
*surf
= &plane
->surface
;
4358 desc
= vk_format_description(iview
->vk_format
);
4360 memset(cb
, 0, sizeof(*cb
));
4362 /* Intensity is implemented as Red, so treat it that way. */
4363 cb
->cb_color_attrib
= S_028C74_FORCE_DST_ALPHA_1(desc
->swizzle
[3] == VK_SWIZZLE_1
);
4365 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
+ plane
->offset
;
4367 cb
->cb_color_base
= va
>> 8;
4369 if (device
->physical_device
->rad_info
.chip_class
>= GFX9
) {
4370 struct gfx9_surf_meta_flags meta
;
4371 if (iview
->image
->dcc_offset
)
4372 meta
= surf
->u
.gfx9
.dcc
;
4374 meta
= surf
->u
.gfx9
.cmask
;
4376 if (device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
4377 cb
->cb_color_attrib3
|= S_028EE0_COLOR_SW_MODE(surf
->u
.gfx9
.surf
.swizzle_mode
) |
4378 S_028EE0_FMASK_SW_MODE(surf
->u
.gfx9
.fmask
.swizzle_mode
) |
4379 S_028EE0_CMASK_PIPE_ALIGNED(surf
->u
.gfx9
.cmask
.pipe_aligned
) |
4380 S_028EE0_DCC_PIPE_ALIGNED(surf
->u
.gfx9
.dcc
.pipe_aligned
);
4382 cb
->cb_color_attrib
|= S_028C74_COLOR_SW_MODE(surf
->u
.gfx9
.surf
.swizzle_mode
) |
4383 S_028C74_FMASK_SW_MODE(surf
->u
.gfx9
.fmask
.swizzle_mode
) |
4384 S_028C74_RB_ALIGNED(meta
.rb_aligned
) |
4385 S_028C74_PIPE_ALIGNED(meta
.pipe_aligned
);
4386 cb
->cb_mrt_epitch
= S_0287A0_EPITCH(surf
->u
.gfx9
.surf
.epitch
);
4389 cb
->cb_color_base
+= surf
->u
.gfx9
.surf_offset
>> 8;
4390 cb
->cb_color_base
|= surf
->tile_swizzle
;
4392 const struct legacy_surf_level
*level_info
= &surf
->u
.legacy
.level
[iview
->base_mip
];
4393 unsigned pitch_tile_max
, slice_tile_max
, tile_mode_index
;
4395 cb
->cb_color_base
+= level_info
->offset
>> 8;
4396 if (level_info
->mode
== RADEON_SURF_MODE_2D
)
4397 cb
->cb_color_base
|= surf
->tile_swizzle
;
4399 pitch_tile_max
= level_info
->nblk_x
/ 8 - 1;
4400 slice_tile_max
= (level_info
->nblk_x
* level_info
->nblk_y
) / 64 - 1;
4401 tile_mode_index
= si_tile_mode_index(plane
, iview
->base_mip
, false);
4403 cb
->cb_color_pitch
= S_028C64_TILE_MAX(pitch_tile_max
);
4404 cb
->cb_color_slice
= S_028C68_TILE_MAX(slice_tile_max
);
4405 cb
->cb_color_cmask_slice
= iview
->image
->cmask
.slice_tile_max
;
4407 cb
->cb_color_attrib
|= S_028C74_TILE_MODE_INDEX(tile_mode_index
);
4409 if (radv_image_has_fmask(iview
->image
)) {
4410 if (device
->physical_device
->rad_info
.chip_class
>= GFX7
)
4411 cb
->cb_color_pitch
|= S_028C64_FMASK_TILE_MAX(iview
->image
->fmask
.pitch_in_pixels
/ 8 - 1);
4412 cb
->cb_color_attrib
|= S_028C74_FMASK_TILE_MODE_INDEX(iview
->image
->fmask
.tile_mode_index
);
4413 cb
->cb_color_fmask_slice
= S_028C88_TILE_MAX(iview
->image
->fmask
.slice_tile_max
);
4415 /* This must be set for fast clear to work without FMASK. */
4416 if (device
->physical_device
->rad_info
.chip_class
>= GFX7
)
4417 cb
->cb_color_pitch
|= S_028C64_FMASK_TILE_MAX(pitch_tile_max
);
4418 cb
->cb_color_attrib
|= S_028C74_FMASK_TILE_MODE_INDEX(tile_mode_index
);
4419 cb
->cb_color_fmask_slice
= S_028C88_TILE_MAX(slice_tile_max
);
4423 /* CMASK variables */
4424 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
;
4425 va
+= iview
->image
->cmask
.offset
;
4426 cb
->cb_color_cmask
= va
>> 8;
4428 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
;
4429 va
+= iview
->image
->dcc_offset
;
4431 if (radv_dcc_enabled(iview
->image
, iview
->base_mip
) &&
4432 device
->physical_device
->rad_info
.chip_class
<= GFX8
)
4433 va
+= plane
->surface
.u
.legacy
.level
[iview
->base_mip
].dcc_offset
;
4435 unsigned dcc_tile_swizzle
= surf
->tile_swizzle
;
4436 dcc_tile_swizzle
&= (surf
->dcc_alignment
- 1) >> 8;
4438 cb
->cb_dcc_base
= va
>> 8;
4439 cb
->cb_dcc_base
|= dcc_tile_swizzle
;
4441 /* GFX10 field has the same base shift as the GFX6 field. */
4442 uint32_t max_slice
= radv_surface_max_layer_count(iview
) - 1;
4443 cb
->cb_color_view
= S_028C6C_SLICE_START(iview
->base_layer
) |
4444 S_028C6C_SLICE_MAX_GFX10(max_slice
);
4446 if (iview
->image
->info
.samples
> 1) {
4447 unsigned log_samples
= util_logbase2(iview
->image
->info
.samples
);
4449 cb
->cb_color_attrib
|= S_028C74_NUM_SAMPLES(log_samples
) |
4450 S_028C74_NUM_FRAGMENTS(log_samples
);
4453 if (radv_image_has_fmask(iview
->image
)) {
4454 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
+ iview
->image
->fmask
.offset
;
4455 cb
->cb_color_fmask
= va
>> 8;
4456 cb
->cb_color_fmask
|= iview
->image
->fmask
.tile_swizzle
;
4458 cb
->cb_color_fmask
= cb
->cb_color_base
;
4461 ntype
= radv_translate_color_numformat(iview
->vk_format
,
4463 vk_format_get_first_non_void_channel(iview
->vk_format
));
4464 format
= radv_translate_colorformat(iview
->vk_format
);
4465 if (format
== V_028C70_COLOR_INVALID
|| ntype
== ~0u)
4466 radv_finishme("Illegal color\n");
4467 swap
= radv_translate_colorswap(iview
->vk_format
, FALSE
);
4468 endian
= radv_colorformat_endian_swap(format
);
4470 /* blend clamp should be set for all NORM/SRGB types */
4471 if (ntype
== V_028C70_NUMBER_UNORM
||
4472 ntype
== V_028C70_NUMBER_SNORM
||
4473 ntype
== V_028C70_NUMBER_SRGB
)
4476 /* set blend bypass according to docs if SINT/UINT or
4477 8/24 COLOR variants */
4478 if (ntype
== V_028C70_NUMBER_UINT
|| ntype
== V_028C70_NUMBER_SINT
||
4479 format
== V_028C70_COLOR_8_24
|| format
== V_028C70_COLOR_24_8
||
4480 format
== V_028C70_COLOR_X24_8_32_FLOAT
) {
4485 if ((ntype
== V_028C70_NUMBER_UINT
|| ntype
== V_028C70_NUMBER_SINT
) &&
4486 (format
== V_028C70_COLOR_8
||
4487 format
== V_028C70_COLOR_8_8
||
4488 format
== V_028C70_COLOR_8_8_8_8
))
4489 ->color_is_int8
= true;
4491 cb
->cb_color_info
= S_028C70_FORMAT(format
) |
4492 S_028C70_COMP_SWAP(swap
) |
4493 S_028C70_BLEND_CLAMP(blend_clamp
) |
4494 S_028C70_BLEND_BYPASS(blend_bypass
) |
4495 S_028C70_SIMPLE_FLOAT(1) |
4496 S_028C70_ROUND_MODE(ntype
!= V_028C70_NUMBER_UNORM
&&
4497 ntype
!= V_028C70_NUMBER_SNORM
&&
4498 ntype
!= V_028C70_NUMBER_SRGB
&&
4499 format
!= V_028C70_COLOR_8_24
&&
4500 format
!= V_028C70_COLOR_24_8
) |
4501 S_028C70_NUMBER_TYPE(ntype
) |
4502 S_028C70_ENDIAN(endian
);
4503 if (radv_image_has_fmask(iview
->image
)) {
4504 cb
->cb_color_info
|= S_028C70_COMPRESSION(1);
4505 if (device
->physical_device
->rad_info
.chip_class
== GFX6
) {
4506 unsigned fmask_bankh
= util_logbase2(iview
->image
->fmask
.bank_height
);
4507 cb
->cb_color_attrib
|= S_028C74_FMASK_BANK_HEIGHT(fmask_bankh
);
4510 if (radv_image_is_tc_compat_cmask(iview
->image
)) {
4511 /* Allow the texture block to read FMASK directly
4512 * without decompressing it. This bit must be cleared
4513 * when performing FMASK_DECOMPRESS or DCC_COMPRESS,
4514 * otherwise the operation doesn't happen.
4516 cb
->cb_color_info
|= S_028C70_FMASK_COMPRESS_1FRAG_ONLY(1);
4518 /* Set CMASK into a tiling format that allows the
4519 * texture block to read it.
4521 cb
->cb_color_info
|= S_028C70_CMASK_ADDR_TYPE(2);
4525 if (radv_image_has_cmask(iview
->image
) &&
4526 !(device
->instance
->debug_flags
& RADV_DEBUG_NO_FAST_CLEARS
))
4527 cb
->cb_color_info
|= S_028C70_FAST_CLEAR(1);
4529 if (radv_dcc_enabled(iview
->image
, iview
->base_mip
))
4530 cb
->cb_color_info
|= S_028C70_DCC_ENABLE(1);
4532 cb
->cb_dcc_control
= radv_init_dcc_control_reg(device
, iview
);
4534 /* This must be set for fast clear to work without FMASK. */
4535 if (!radv_image_has_fmask(iview
->image
) &&
4536 device
->physical_device
->rad_info
.chip_class
== GFX6
) {
4537 unsigned bankh
= util_logbase2(surf
->u
.legacy
.bankh
);
4538 cb
->cb_color_attrib
|= S_028C74_FMASK_BANK_HEIGHT(bankh
);
4541 if (device
->physical_device
->rad_info
.chip_class
>= GFX9
) {
4542 const struct vk_format_description
*format_desc
= vk_format_description(iview
->image
->vk_format
);
4544 unsigned mip0_depth
= iview
->image
->type
== VK_IMAGE_TYPE_3D
?
4545 (iview
->extent
.depth
- 1) : (iview
->image
->info
.array_size
- 1);
4546 unsigned width
= iview
->extent
.width
/ (iview
->plane_id
? format_desc
->width_divisor
: 1);
4547 unsigned height
= iview
->extent
.height
/ (iview
->plane_id
? format_desc
->height_divisor
: 1);
4549 if (device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
4550 cb
->cb_color_view
|= S_028C6C_MIP_LEVEL_GFX10(iview
->base_mip
);
4552 cb
->cb_color_attrib3
|= S_028EE0_MIP0_DEPTH(mip0_depth
) |
4553 S_028EE0_RESOURCE_TYPE(surf
->u
.gfx9
.resource_type
) |
4554 S_028EE0_RESOURCE_LEVEL(1);
4556 cb
->cb_color_view
|= S_028C6C_MIP_LEVEL_GFX9(iview
->base_mip
);
4557 cb
->cb_color_attrib
|= S_028C74_MIP0_DEPTH(mip0_depth
) |
4558 S_028C74_RESOURCE_TYPE(surf
->u
.gfx9
.resource_type
);
4561 cb
->cb_color_attrib2
= S_028C68_MIP0_WIDTH(width
- 1) |
4562 S_028C68_MIP0_HEIGHT(height
- 1) |
4563 S_028C68_MAX_MIP(iview
->image
->info
.levels
- 1);
4568 radv_calc_decompress_on_z_planes(struct radv_device
*device
,
4569 struct radv_image_view
*iview
)
4571 unsigned max_zplanes
= 0;
4573 assert(radv_image_is_tc_compat_htile(iview
->image
));
4575 if (device
->physical_device
->rad_info
.chip_class
>= GFX9
) {
4576 /* Default value for 32-bit depth surfaces. */
4579 if (iview
->vk_format
== VK_FORMAT_D16_UNORM
&&
4580 iview
->image
->info
.samples
> 1)
4583 max_zplanes
= max_zplanes
+ 1;
4585 if (iview
->vk_format
== VK_FORMAT_D16_UNORM
) {
4586 /* Do not enable Z plane compression for 16-bit depth
4587 * surfaces because isn't supported on GFX8. Only
4588 * 32-bit depth surfaces are supported by the hardware.
4589 * This allows to maintain shader compatibility and to
4590 * reduce the number of depth decompressions.
4594 if (iview
->image
->info
.samples
<= 1)
4596 else if (iview
->image
->info
.samples
<= 4)
4607 radv_initialise_ds_surface(struct radv_device
*device
,
4608 struct radv_ds_buffer_info
*ds
,
4609 struct radv_image_view
*iview
)
4611 unsigned level
= iview
->base_mip
;
4612 unsigned format
, stencil_format
;
4613 uint64_t va
, s_offs
, z_offs
;
4614 bool stencil_only
= false;
4615 const struct radv_image_plane
*plane
= &iview
->image
->planes
[0];
4616 const struct radeon_surf
*surf
= &plane
->surface
;
4618 assert(vk_format_get_plane_count(iview
->image
->vk_format
) == 1);
4620 memset(ds
, 0, sizeof(*ds
));
4621 switch (iview
->image
->vk_format
) {
4622 case VK_FORMAT_D24_UNORM_S8_UINT
:
4623 case VK_FORMAT_X8_D24_UNORM_PACK32
:
4624 ds
->pa_su_poly_offset_db_fmt_cntl
= S_028B78_POLY_OFFSET_NEG_NUM_DB_BITS(-24);
4625 ds
->offset_scale
= 2.0f
;
4627 case VK_FORMAT_D16_UNORM
:
4628 case VK_FORMAT_D16_UNORM_S8_UINT
:
4629 ds
->pa_su_poly_offset_db_fmt_cntl
= S_028B78_POLY_OFFSET_NEG_NUM_DB_BITS(-16);
4630 ds
->offset_scale
= 4.0f
;
4632 case VK_FORMAT_D32_SFLOAT
:
4633 case VK_FORMAT_D32_SFLOAT_S8_UINT
:
4634 ds
->pa_su_poly_offset_db_fmt_cntl
= S_028B78_POLY_OFFSET_NEG_NUM_DB_BITS(-23) |
4635 S_028B78_POLY_OFFSET_DB_IS_FLOAT_FMT(1);
4636 ds
->offset_scale
= 1.0f
;
4638 case VK_FORMAT_S8_UINT
:
4639 stencil_only
= true;
4645 format
= radv_translate_dbformat(iview
->image
->vk_format
);
4646 stencil_format
= surf
->has_stencil
?
4647 V_028044_STENCIL_8
: V_028044_STENCIL_INVALID
;
4649 uint32_t max_slice
= radv_surface_max_layer_count(iview
) - 1;
4650 ds
->db_depth_view
= S_028008_SLICE_START(iview
->base_layer
) |
4651 S_028008_SLICE_MAX(max_slice
);
4652 if (device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
4653 ds
->db_depth_view
|= S_028008_SLICE_START_HI(iview
->base_layer
>> 11) |
4654 S_028008_SLICE_MAX_HI(max_slice
>> 11);
4657 ds
->db_htile_data_base
= 0;
4658 ds
->db_htile_surface
= 0;
4660 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
;
4661 s_offs
= z_offs
= va
;
4663 if (device
->physical_device
->rad_info
.chip_class
>= GFX9
) {
4664 assert(surf
->u
.gfx9
.surf_offset
== 0);
4665 s_offs
+= surf
->u
.gfx9
.stencil_offset
;
4667 ds
->db_z_info
= S_028038_FORMAT(format
) |
4668 S_028038_NUM_SAMPLES(util_logbase2(iview
->image
->info
.samples
)) |
4669 S_028038_SW_MODE(surf
->u
.gfx9
.surf
.swizzle_mode
) |
4670 S_028038_MAXMIP(iview
->image
->info
.levels
- 1) |
4671 S_028038_ZRANGE_PRECISION(1);
4672 ds
->db_stencil_info
= S_02803C_FORMAT(stencil_format
) |
4673 S_02803C_SW_MODE(surf
->u
.gfx9
.stencil
.swizzle_mode
);
4675 if (device
->physical_device
->rad_info
.chip_class
== GFX9
) {
4676 ds
->db_z_info2
= S_028068_EPITCH(surf
->u
.gfx9
.surf
.epitch
);
4677 ds
->db_stencil_info2
= S_02806C_EPITCH(surf
->u
.gfx9
.stencil
.epitch
);
4680 ds
->db_depth_view
|= S_028008_MIPID(level
);
4681 ds
->db_depth_size
= S_02801C_X_MAX(iview
->image
->info
.width
- 1) |
4682 S_02801C_Y_MAX(iview
->image
->info
.height
- 1);
4684 if (radv_htile_enabled(iview
->image
, level
)) {
4685 ds
->db_z_info
|= S_028038_TILE_SURFACE_ENABLE(1);
4687 if (radv_image_is_tc_compat_htile(iview
->image
)) {
4688 unsigned max_zplanes
=
4689 radv_calc_decompress_on_z_planes(device
, iview
);
4691 ds
->db_z_info
|= S_028038_DECOMPRESS_ON_N_ZPLANES(max_zplanes
);
4693 if (device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
4694 ds
->db_z_info
|= S_028040_ITERATE_FLUSH(1);
4695 ds
->db_stencil_info
|= S_028044_ITERATE_FLUSH(1);
4697 ds
->db_z_info
|= S_028038_ITERATE_FLUSH(1);
4698 ds
->db_stencil_info
|= S_02803C_ITERATE_FLUSH(1);
4702 if (!surf
->has_stencil
)
4703 /* Use all of the htile_buffer for depth if there's no stencil. */
4704 ds
->db_stencil_info
|= S_02803C_TILE_STENCIL_DISABLE(1);
4705 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
+
4706 iview
->image
->htile_offset
;
4707 ds
->db_htile_data_base
= va
>> 8;
4708 ds
->db_htile_surface
= S_028ABC_FULL_CACHE(1) |
4709 S_028ABC_PIPE_ALIGNED(surf
->u
.gfx9
.htile
.pipe_aligned
);
4711 if (device
->physical_device
->rad_info
.chip_class
== GFX9
) {
4712 ds
->db_htile_surface
|= S_028ABC_RB_ALIGNED(surf
->u
.gfx9
.htile
.rb_aligned
);
4716 const struct legacy_surf_level
*level_info
= &surf
->u
.legacy
.level
[level
];
4719 level_info
= &surf
->u
.legacy
.stencil_level
[level
];
4721 z_offs
+= surf
->u
.legacy
.level
[level
].offset
;
4722 s_offs
+= surf
->u
.legacy
.stencil_level
[level
].offset
;
4724 ds
->db_depth_info
= S_02803C_ADDR5_SWIZZLE_MASK(!radv_image_is_tc_compat_htile(iview
->image
));
4725 ds
->db_z_info
= S_028040_FORMAT(format
) | S_028040_ZRANGE_PRECISION(1);
4726 ds
->db_stencil_info
= S_028044_FORMAT(stencil_format
);
4728 if (iview
->image
->info
.samples
> 1)
4729 ds
->db_z_info
|= S_028040_NUM_SAMPLES(util_logbase2(iview
->image
->info
.samples
));
4731 if (device
->physical_device
->rad_info
.chip_class
>= GFX7
) {
4732 struct radeon_info
*info
= &device
->physical_device
->rad_info
;
4733 unsigned tiling_index
= surf
->u
.legacy
.tiling_index
[level
];
4734 unsigned stencil_index
= surf
->u
.legacy
.stencil_tiling_index
[level
];
4735 unsigned macro_index
= surf
->u
.legacy
.macro_tile_index
;
4736 unsigned tile_mode
= info
->si_tile_mode_array
[tiling_index
];
4737 unsigned stencil_tile_mode
= info
->si_tile_mode_array
[stencil_index
];
4738 unsigned macro_mode
= info
->cik_macrotile_mode_array
[macro_index
];
4741 tile_mode
= stencil_tile_mode
;
4743 ds
->db_depth_info
|=
4744 S_02803C_ARRAY_MODE(G_009910_ARRAY_MODE(tile_mode
)) |
4745 S_02803C_PIPE_CONFIG(G_009910_PIPE_CONFIG(tile_mode
)) |
4746 S_02803C_BANK_WIDTH(G_009990_BANK_WIDTH(macro_mode
)) |
4747 S_02803C_BANK_HEIGHT(G_009990_BANK_HEIGHT(macro_mode
)) |
4748 S_02803C_MACRO_TILE_ASPECT(G_009990_MACRO_TILE_ASPECT(macro_mode
)) |
4749 S_02803C_NUM_BANKS(G_009990_NUM_BANKS(macro_mode
));
4750 ds
->db_z_info
|= S_028040_TILE_SPLIT(G_009910_TILE_SPLIT(tile_mode
));
4751 ds
->db_stencil_info
|= S_028044_TILE_SPLIT(G_009910_TILE_SPLIT(stencil_tile_mode
));
4753 unsigned tile_mode_index
= si_tile_mode_index(&iview
->image
->planes
[0], level
, false);
4754 ds
->db_z_info
|= S_028040_TILE_MODE_INDEX(tile_mode_index
);
4755 tile_mode_index
= si_tile_mode_index(&iview
->image
->planes
[0], level
, true);
4756 ds
->db_stencil_info
|= S_028044_TILE_MODE_INDEX(tile_mode_index
);
4758 ds
->db_z_info
|= S_028040_TILE_MODE_INDEX(tile_mode_index
);
4761 ds
->db_depth_size
= S_028058_PITCH_TILE_MAX((level_info
->nblk_x
/ 8) - 1) |
4762 S_028058_HEIGHT_TILE_MAX((level_info
->nblk_y
/ 8) - 1);
4763 ds
->db_depth_slice
= S_02805C_SLICE_TILE_MAX((level_info
->nblk_x
* level_info
->nblk_y
) / 64 - 1);
4765 if (radv_htile_enabled(iview
->image
, level
)) {
4766 ds
->db_z_info
|= S_028040_TILE_SURFACE_ENABLE(1);
4768 if (!surf
->has_stencil
&&
4769 !radv_image_is_tc_compat_htile(iview
->image
))
4770 /* Use all of the htile_buffer for depth if there's no stencil. */
4771 ds
->db_stencil_info
|= S_028044_TILE_STENCIL_DISABLE(1);
4773 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
+
4774 iview
->image
->htile_offset
;
4775 ds
->db_htile_data_base
= va
>> 8;
4776 ds
->db_htile_surface
= S_028ABC_FULL_CACHE(1);
4778 if (radv_image_is_tc_compat_htile(iview
->image
)) {
4779 unsigned max_zplanes
=
4780 radv_calc_decompress_on_z_planes(device
, iview
);
4782 ds
->db_htile_surface
|= S_028ABC_TC_COMPATIBLE(1);
4783 ds
->db_z_info
|= S_028040_DECOMPRESS_ON_N_ZPLANES(max_zplanes
);
4788 ds
->db_z_read_base
= ds
->db_z_write_base
= z_offs
>> 8;
4789 ds
->db_stencil_read_base
= ds
->db_stencil_write_base
= s_offs
>> 8;
4792 VkResult
radv_CreateFramebuffer(
4794 const VkFramebufferCreateInfo
* pCreateInfo
,
4795 const VkAllocationCallbacks
* pAllocator
,
4796 VkFramebuffer
* pFramebuffer
)
4798 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4799 struct radv_framebuffer
*framebuffer
;
4801 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_FRAMEBUFFER_CREATE_INFO
);
4803 size_t size
= sizeof(*framebuffer
) +
4804 sizeof(struct radv_attachment_info
) * pCreateInfo
->attachmentCount
;
4805 framebuffer
= vk_alloc2(&device
->alloc
, pAllocator
, size
, 8,
4806 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
4807 if (framebuffer
== NULL
)
4808 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
4810 framebuffer
->attachment_count
= pCreateInfo
->attachmentCount
;
4811 framebuffer
->width
= pCreateInfo
->width
;
4812 framebuffer
->height
= pCreateInfo
->height
;
4813 framebuffer
->layers
= pCreateInfo
->layers
;
4814 for (uint32_t i
= 0; i
< pCreateInfo
->attachmentCount
; i
++) {
4815 VkImageView _iview
= pCreateInfo
->pAttachments
[i
];
4816 struct radv_image_view
*iview
= radv_image_view_from_handle(_iview
);
4817 framebuffer
->attachments
[i
].attachment
= iview
;
4818 if (iview
->aspect_mask
& (VK_IMAGE_ASPECT_DEPTH_BIT
| VK_IMAGE_ASPECT_STENCIL_BIT
)) {
4819 radv_initialise_ds_surface(device
, &framebuffer
->attachments
[i
].ds
, iview
);
4821 radv_initialise_color_surface(device
, &framebuffer
->attachments
[i
].cb
, iview
);
4823 framebuffer
->width
= MIN2(framebuffer
->width
, iview
->extent
.width
);
4824 framebuffer
->height
= MIN2(framebuffer
->height
, iview
->extent
.height
);
4825 framebuffer
->layers
= MIN2(framebuffer
->layers
, radv_surface_max_layer_count(iview
));
4828 *pFramebuffer
= radv_framebuffer_to_handle(framebuffer
);
4832 void radv_DestroyFramebuffer(
4835 const VkAllocationCallbacks
* pAllocator
)
4837 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4838 RADV_FROM_HANDLE(radv_framebuffer
, fb
, _fb
);
4842 vk_free2(&device
->alloc
, pAllocator
, fb
);
4845 static unsigned radv_tex_wrap(VkSamplerAddressMode address_mode
)
4847 switch (address_mode
) {
4848 case VK_SAMPLER_ADDRESS_MODE_REPEAT
:
4849 return V_008F30_SQ_TEX_WRAP
;
4850 case VK_SAMPLER_ADDRESS_MODE_MIRRORED_REPEAT
:
4851 return V_008F30_SQ_TEX_MIRROR
;
4852 case VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE
:
4853 return V_008F30_SQ_TEX_CLAMP_LAST_TEXEL
;
4854 case VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER
:
4855 return V_008F30_SQ_TEX_CLAMP_BORDER
;
4856 case VK_SAMPLER_ADDRESS_MODE_MIRROR_CLAMP_TO_EDGE
:
4857 return V_008F30_SQ_TEX_MIRROR_ONCE_LAST_TEXEL
;
4859 unreachable("illegal tex wrap mode");
4865 radv_tex_compare(VkCompareOp op
)
4868 case VK_COMPARE_OP_NEVER
:
4869 return V_008F30_SQ_TEX_DEPTH_COMPARE_NEVER
;
4870 case VK_COMPARE_OP_LESS
:
4871 return V_008F30_SQ_TEX_DEPTH_COMPARE_LESS
;
4872 case VK_COMPARE_OP_EQUAL
:
4873 return V_008F30_SQ_TEX_DEPTH_COMPARE_EQUAL
;
4874 case VK_COMPARE_OP_LESS_OR_EQUAL
:
4875 return V_008F30_SQ_TEX_DEPTH_COMPARE_LESSEQUAL
;
4876 case VK_COMPARE_OP_GREATER
:
4877 return V_008F30_SQ_TEX_DEPTH_COMPARE_GREATER
;
4878 case VK_COMPARE_OP_NOT_EQUAL
:
4879 return V_008F30_SQ_TEX_DEPTH_COMPARE_NOTEQUAL
;
4880 case VK_COMPARE_OP_GREATER_OR_EQUAL
:
4881 return V_008F30_SQ_TEX_DEPTH_COMPARE_GREATEREQUAL
;
4882 case VK_COMPARE_OP_ALWAYS
:
4883 return V_008F30_SQ_TEX_DEPTH_COMPARE_ALWAYS
;
4885 unreachable("illegal compare mode");
4891 radv_tex_filter(VkFilter filter
, unsigned max_ansio
)
4894 case VK_FILTER_NEAREST
:
4895 return (max_ansio
> 1 ? V_008F38_SQ_TEX_XY_FILTER_ANISO_POINT
:
4896 V_008F38_SQ_TEX_XY_FILTER_POINT
);
4897 case VK_FILTER_LINEAR
:
4898 return (max_ansio
> 1 ? V_008F38_SQ_TEX_XY_FILTER_ANISO_BILINEAR
:
4899 V_008F38_SQ_TEX_XY_FILTER_BILINEAR
);
4900 case VK_FILTER_CUBIC_IMG
:
4902 fprintf(stderr
, "illegal texture filter");
4908 radv_tex_mipfilter(VkSamplerMipmapMode mode
)
4911 case VK_SAMPLER_MIPMAP_MODE_NEAREST
:
4912 return V_008F38_SQ_TEX_Z_FILTER_POINT
;
4913 case VK_SAMPLER_MIPMAP_MODE_LINEAR
:
4914 return V_008F38_SQ_TEX_Z_FILTER_LINEAR
;
4916 return V_008F38_SQ_TEX_Z_FILTER_NONE
;
4921 radv_tex_bordercolor(VkBorderColor bcolor
)
4924 case VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK
:
4925 case VK_BORDER_COLOR_INT_TRANSPARENT_BLACK
:
4926 return V_008F3C_SQ_TEX_BORDER_COLOR_TRANS_BLACK
;
4927 case VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK
:
4928 case VK_BORDER_COLOR_INT_OPAQUE_BLACK
:
4929 return V_008F3C_SQ_TEX_BORDER_COLOR_OPAQUE_BLACK
;
4930 case VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE
:
4931 case VK_BORDER_COLOR_INT_OPAQUE_WHITE
:
4932 return V_008F3C_SQ_TEX_BORDER_COLOR_OPAQUE_WHITE
;
4940 radv_tex_aniso_filter(unsigned filter
)
4954 radv_tex_filter_mode(VkSamplerReductionModeEXT mode
)
4957 case VK_SAMPLER_REDUCTION_MODE_WEIGHTED_AVERAGE_EXT
:
4958 return V_008F30_SQ_IMG_FILTER_MODE_BLEND
;
4959 case VK_SAMPLER_REDUCTION_MODE_MIN_EXT
:
4960 return V_008F30_SQ_IMG_FILTER_MODE_MIN
;
4961 case VK_SAMPLER_REDUCTION_MODE_MAX_EXT
:
4962 return V_008F30_SQ_IMG_FILTER_MODE_MAX
;
4970 radv_get_max_anisotropy(struct radv_device
*device
,
4971 const VkSamplerCreateInfo
*pCreateInfo
)
4973 if (device
->force_aniso
>= 0)
4974 return device
->force_aniso
;
4976 if (pCreateInfo
->anisotropyEnable
&&
4977 pCreateInfo
->maxAnisotropy
> 1.0f
)
4978 return (uint32_t)pCreateInfo
->maxAnisotropy
;
4984 radv_init_sampler(struct radv_device
*device
,
4985 struct radv_sampler
*sampler
,
4986 const VkSamplerCreateInfo
*pCreateInfo
)
4988 uint32_t max_aniso
= radv_get_max_anisotropy(device
, pCreateInfo
);
4989 uint32_t max_aniso_ratio
= radv_tex_aniso_filter(max_aniso
);
4990 bool compat_mode
= device
->physical_device
->rad_info
.chip_class
== GFX8
||
4991 device
->physical_device
->rad_info
.chip_class
== GFX9
;
4992 unsigned filter_mode
= V_008F30_SQ_IMG_FILTER_MODE_BLEND
;
4994 const struct VkSamplerReductionModeCreateInfoEXT
*sampler_reduction
=
4995 vk_find_struct_const(pCreateInfo
->pNext
,
4996 SAMPLER_REDUCTION_MODE_CREATE_INFO_EXT
);
4997 if (sampler_reduction
)
4998 filter_mode
= radv_tex_filter_mode(sampler_reduction
->reductionMode
);
5000 sampler
->state
[0] = (S_008F30_CLAMP_X(radv_tex_wrap(pCreateInfo
->addressModeU
)) |
5001 S_008F30_CLAMP_Y(radv_tex_wrap(pCreateInfo
->addressModeV
)) |
5002 S_008F30_CLAMP_Z(radv_tex_wrap(pCreateInfo
->addressModeW
)) |
5003 S_008F30_MAX_ANISO_RATIO(max_aniso_ratio
) |
5004 S_008F30_DEPTH_COMPARE_FUNC(radv_tex_compare(pCreateInfo
->compareOp
)) |
5005 S_008F30_FORCE_UNNORMALIZED(pCreateInfo
->unnormalizedCoordinates
? 1 : 0) |
5006 S_008F30_ANISO_THRESHOLD(max_aniso_ratio
>> 1) |
5007 S_008F30_ANISO_BIAS(max_aniso_ratio
) |
5008 S_008F30_DISABLE_CUBE_WRAP(0) |
5009 S_008F30_COMPAT_MODE(compat_mode
) |
5010 S_008F30_FILTER_MODE(filter_mode
));
5011 sampler
->state
[1] = (S_008F34_MIN_LOD(S_FIXED(CLAMP(pCreateInfo
->minLod
, 0, 15), 8)) |
5012 S_008F34_MAX_LOD(S_FIXED(CLAMP(pCreateInfo
->maxLod
, 0, 15), 8)) |
5013 S_008F34_PERF_MIP(max_aniso_ratio
? max_aniso_ratio
+ 6 : 0));
5014 sampler
->state
[2] = (S_008F38_LOD_BIAS(S_FIXED(CLAMP(pCreateInfo
->mipLodBias
, -16, 16), 8)) |
5015 S_008F38_XY_MAG_FILTER(radv_tex_filter(pCreateInfo
->magFilter
, max_aniso
)) |
5016 S_008F38_XY_MIN_FILTER(radv_tex_filter(pCreateInfo
->minFilter
, max_aniso
)) |
5017 S_008F38_MIP_FILTER(radv_tex_mipfilter(pCreateInfo
->mipmapMode
)) |
5018 S_008F38_MIP_POINT_PRECLAMP(0));
5019 sampler
->state
[3] = (S_008F3C_BORDER_COLOR_PTR(0) |
5020 S_008F3C_BORDER_COLOR_TYPE(radv_tex_bordercolor(pCreateInfo
->borderColor
)));
5022 if (device
->physical_device
->rad_info
.chip_class
>= GFX10
) {
5023 sampler
->state
[2] |= S_008F38_ANISO_OVERRIDE_GFX10(1);
5025 sampler
->state
[2] |=
5026 S_008F38_DISABLE_LSB_CEIL(device
->physical_device
->rad_info
.chip_class
<= GFX8
) |
5027 S_008F38_FILTER_PREC_FIX(1) |
5028 S_008F38_ANISO_OVERRIDE_GFX6(device
->physical_device
->rad_info
.chip_class
>= GFX8
);
5032 VkResult
radv_CreateSampler(
5034 const VkSamplerCreateInfo
* pCreateInfo
,
5035 const VkAllocationCallbacks
* pAllocator
,
5036 VkSampler
* pSampler
)
5038 RADV_FROM_HANDLE(radv_device
, device
, _device
);
5039 struct radv_sampler
*sampler
;
5041 const struct VkSamplerYcbcrConversionInfo
*ycbcr_conversion
=
5042 vk_find_struct_const(pCreateInfo
->pNext
,
5043 SAMPLER_YCBCR_CONVERSION_INFO
);
5045 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO
);
5047 sampler
= vk_alloc2(&device
->alloc
, pAllocator
, sizeof(*sampler
), 8,
5048 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
5050 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
5052 radv_init_sampler(device
, sampler
, pCreateInfo
);
5054 sampler
->ycbcr_sampler
= ycbcr_conversion
? radv_sampler_ycbcr_conversion_from_handle(ycbcr_conversion
->conversion
): NULL
;
5055 *pSampler
= radv_sampler_to_handle(sampler
);
5060 void radv_DestroySampler(
5063 const VkAllocationCallbacks
* pAllocator
)
5065 RADV_FROM_HANDLE(radv_device
, device
, _device
);
5066 RADV_FROM_HANDLE(radv_sampler
, sampler
, _sampler
);
5070 vk_free2(&device
->alloc
, pAllocator
, sampler
);
5073 /* vk_icd.h does not declare this function, so we declare it here to
5074 * suppress Wmissing-prototypes.
5076 PUBLIC VKAPI_ATTR VkResult VKAPI_CALL
5077 vk_icdNegotiateLoaderICDInterfaceVersion(uint32_t *pSupportedVersion
);
5079 PUBLIC VKAPI_ATTR VkResult VKAPI_CALL
5080 vk_icdNegotiateLoaderICDInterfaceVersion(uint32_t *pSupportedVersion
)
5082 /* For the full details on loader interface versioning, see
5083 * <https://github.com/KhronosGroup/Vulkan-LoaderAndValidationLayers/blob/master/loader/LoaderAndLayerInterface.md>.
5084 * What follows is a condensed summary, to help you navigate the large and
5085 * confusing official doc.
5087 * - Loader interface v0 is incompatible with later versions. We don't
5090 * - In loader interface v1:
5091 * - The first ICD entrypoint called by the loader is
5092 * vk_icdGetInstanceProcAddr(). The ICD must statically expose this
5094 * - The ICD must statically expose no other Vulkan symbol unless it is
5095 * linked with -Bsymbolic.
5096 * - Each dispatchable Vulkan handle created by the ICD must be
5097 * a pointer to a struct whose first member is VK_LOADER_DATA. The
5098 * ICD must initialize VK_LOADER_DATA.loadMagic to ICD_LOADER_MAGIC.
5099 * - The loader implements vkCreate{PLATFORM}SurfaceKHR() and
5100 * vkDestroySurfaceKHR(). The ICD must be capable of working with
5101 * such loader-managed surfaces.
5103 * - Loader interface v2 differs from v1 in:
5104 * - The first ICD entrypoint called by the loader is
5105 * vk_icdNegotiateLoaderICDInterfaceVersion(). The ICD must
5106 * statically expose this entrypoint.
5108 * - Loader interface v3 differs from v2 in:
5109 * - The ICD must implement vkCreate{PLATFORM}SurfaceKHR(),
5110 * vkDestroySurfaceKHR(), and other API which uses VKSurfaceKHR,
5111 * because the loader no longer does so.
5113 *pSupportedVersion
= MIN2(*pSupportedVersion
, 4u);
5117 VkResult
radv_GetMemoryFdKHR(VkDevice _device
,
5118 const VkMemoryGetFdInfoKHR
*pGetFdInfo
,
5121 RADV_FROM_HANDLE(radv_device
, device
, _device
);
5122 RADV_FROM_HANDLE(radv_device_memory
, memory
, pGetFdInfo
->memory
);
5124 assert(pGetFdInfo
->sType
== VK_STRUCTURE_TYPE_MEMORY_GET_FD_INFO_KHR
);
5126 /* At the moment, we support only the below handle types. */
5127 assert(pGetFdInfo
->handleType
==
5128 VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD_BIT
||
5129 pGetFdInfo
->handleType
==
5130 VK_EXTERNAL_MEMORY_HANDLE_TYPE_DMA_BUF_BIT_EXT
);
5132 bool ret
= radv_get_memory_fd(device
, memory
, pFD
);
5134 return vk_error(device
->instance
, VK_ERROR_OUT_OF_DEVICE_MEMORY
);
5138 VkResult
radv_GetMemoryFdPropertiesKHR(VkDevice _device
,
5139 VkExternalMemoryHandleTypeFlagBits handleType
,
5141 VkMemoryFdPropertiesKHR
*pMemoryFdProperties
)
5143 RADV_FROM_HANDLE(radv_device
, device
, _device
);
5145 switch (handleType
) {
5146 case VK_EXTERNAL_MEMORY_HANDLE_TYPE_DMA_BUF_BIT_EXT
:
5147 pMemoryFdProperties
->memoryTypeBits
= (1 << RADV_MEM_TYPE_COUNT
) - 1;
5151 /* The valid usage section for this function says:
5153 * "handleType must not be one of the handle types defined as
5156 * So opaque handle types fall into the default "unsupported" case.
5158 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
5162 static VkResult
radv_import_opaque_fd(struct radv_device
*device
,
5166 uint32_t syncobj_handle
= 0;
5167 int ret
= device
->ws
->import_syncobj(device
->ws
, fd
, &syncobj_handle
);
5169 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
5172 device
->ws
->destroy_syncobj(device
->ws
, *syncobj
);
5174 *syncobj
= syncobj_handle
;
5180 static VkResult
radv_import_sync_fd(struct radv_device
*device
,
5184 /* If we create a syncobj we do it locally so that if we have an error, we don't
5185 * leave a syncobj in an undetermined state in the fence. */
5186 uint32_t syncobj_handle
= *syncobj
;
5187 if (!syncobj_handle
) {
5188 int ret
= device
->ws
->create_syncobj(device
->ws
, &syncobj_handle
);
5190 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
5195 device
->ws
->signal_syncobj(device
->ws
, syncobj_handle
);
5197 int ret
= device
->ws
->import_syncobj_from_sync_file(device
->ws
, syncobj_handle
, fd
);
5199 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
5202 *syncobj
= syncobj_handle
;
5209 VkResult
radv_ImportSemaphoreFdKHR(VkDevice _device
,
5210 const VkImportSemaphoreFdInfoKHR
*pImportSemaphoreFdInfo
)
5212 RADV_FROM_HANDLE(radv_device
, device
, _device
);
5213 RADV_FROM_HANDLE(radv_semaphore
, sem
, pImportSemaphoreFdInfo
->semaphore
);
5214 uint32_t *syncobj_dst
= NULL
;
5216 if (pImportSemaphoreFdInfo
->flags
& VK_SEMAPHORE_IMPORT_TEMPORARY_BIT
) {
5217 syncobj_dst
= &sem
->temp_syncobj
;
5219 syncobj_dst
= &sem
->syncobj
;
5222 switch(pImportSemaphoreFdInfo
->handleType
) {
5223 case VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
:
5224 return radv_import_opaque_fd(device
, pImportSemaphoreFdInfo
->fd
, syncobj_dst
);
5225 case VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_SYNC_FD_BIT
:
5226 return radv_import_sync_fd(device
, pImportSemaphoreFdInfo
->fd
, syncobj_dst
);
5228 unreachable("Unhandled semaphore handle type");
5232 VkResult
radv_GetSemaphoreFdKHR(VkDevice _device
,
5233 const VkSemaphoreGetFdInfoKHR
*pGetFdInfo
,
5236 RADV_FROM_HANDLE(radv_device
, device
, _device
);
5237 RADV_FROM_HANDLE(radv_semaphore
, sem
, pGetFdInfo
->semaphore
);
5239 uint32_t syncobj_handle
;
5241 if (sem
->temp_syncobj
)
5242 syncobj_handle
= sem
->temp_syncobj
;
5244 syncobj_handle
= sem
->syncobj
;
5246 switch(pGetFdInfo
->handleType
) {
5247 case VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
:
5248 ret
= device
->ws
->export_syncobj(device
->ws
, syncobj_handle
, pFd
);
5250 case VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_SYNC_FD_BIT
:
5251 ret
= device
->ws
->export_syncobj_to_sync_file(device
->ws
, syncobj_handle
, pFd
);
5253 if (sem
->temp_syncobj
) {
5254 close (sem
->temp_syncobj
);
5255 sem
->temp_syncobj
= 0;
5257 device
->ws
->reset_syncobj(device
->ws
, syncobj_handle
);
5262 unreachable("Unhandled semaphore handle type");
5266 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
5270 void radv_GetPhysicalDeviceExternalSemaphoreProperties(
5271 VkPhysicalDevice physicalDevice
,
5272 const VkPhysicalDeviceExternalSemaphoreInfo
*pExternalSemaphoreInfo
,
5273 VkExternalSemaphoreProperties
*pExternalSemaphoreProperties
)
5275 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
5277 /* Require has_syncobj_wait_for_submit for the syncobj signal ioctl introduced at virtually the same time */
5278 if (pdevice
->rad_info
.has_syncobj_wait_for_submit
&&
5279 (pExternalSemaphoreInfo
->handleType
== VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
||
5280 pExternalSemaphoreInfo
->handleType
== VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_SYNC_FD_BIT
)) {
5281 pExternalSemaphoreProperties
->exportFromImportedHandleTypes
= VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
| VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_SYNC_FD_BIT
;
5282 pExternalSemaphoreProperties
->compatibleHandleTypes
= VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
| VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_SYNC_FD_BIT
;
5283 pExternalSemaphoreProperties
->externalSemaphoreFeatures
= VK_EXTERNAL_SEMAPHORE_FEATURE_EXPORTABLE_BIT
|
5284 VK_EXTERNAL_SEMAPHORE_FEATURE_IMPORTABLE_BIT
;
5285 } else if (pExternalSemaphoreInfo
->handleType
== VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
) {
5286 pExternalSemaphoreProperties
->exportFromImportedHandleTypes
= VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
;
5287 pExternalSemaphoreProperties
->compatibleHandleTypes
= VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
;
5288 pExternalSemaphoreProperties
->externalSemaphoreFeatures
= VK_EXTERNAL_SEMAPHORE_FEATURE_EXPORTABLE_BIT
|
5289 VK_EXTERNAL_SEMAPHORE_FEATURE_IMPORTABLE_BIT
;
5291 pExternalSemaphoreProperties
->exportFromImportedHandleTypes
= 0;
5292 pExternalSemaphoreProperties
->compatibleHandleTypes
= 0;
5293 pExternalSemaphoreProperties
->externalSemaphoreFeatures
= 0;
5297 VkResult
radv_ImportFenceFdKHR(VkDevice _device
,
5298 const VkImportFenceFdInfoKHR
*pImportFenceFdInfo
)
5300 RADV_FROM_HANDLE(radv_device
, device
, _device
);
5301 RADV_FROM_HANDLE(radv_fence
, fence
, pImportFenceFdInfo
->fence
);
5302 uint32_t *syncobj_dst
= NULL
;
5305 if (pImportFenceFdInfo
->flags
& VK_FENCE_IMPORT_TEMPORARY_BIT
) {
5306 syncobj_dst
= &fence
->temp_syncobj
;
5308 syncobj_dst
= &fence
->syncobj
;
5311 switch(pImportFenceFdInfo
->handleType
) {
5312 case VK_EXTERNAL_FENCE_HANDLE_TYPE_OPAQUE_FD_BIT
:
5313 return radv_import_opaque_fd(device
, pImportFenceFdInfo
->fd
, syncobj_dst
);
5314 case VK_EXTERNAL_FENCE_HANDLE_TYPE_SYNC_FD_BIT
:
5315 return radv_import_sync_fd(device
, pImportFenceFdInfo
->fd
, syncobj_dst
);
5317 unreachable("Unhandled fence handle type");
5321 VkResult
radv_GetFenceFdKHR(VkDevice _device
,
5322 const VkFenceGetFdInfoKHR
*pGetFdInfo
,
5325 RADV_FROM_HANDLE(radv_device
, device
, _device
);
5326 RADV_FROM_HANDLE(radv_fence
, fence
, pGetFdInfo
->fence
);
5328 uint32_t syncobj_handle
;
5330 if (fence
->temp_syncobj
)
5331 syncobj_handle
= fence
->temp_syncobj
;
5333 syncobj_handle
= fence
->syncobj
;
5335 switch(pGetFdInfo
->handleType
) {
5336 case VK_EXTERNAL_FENCE_HANDLE_TYPE_OPAQUE_FD_BIT
:
5337 ret
= device
->ws
->export_syncobj(device
->ws
, syncobj_handle
, pFd
);
5339 case VK_EXTERNAL_FENCE_HANDLE_TYPE_SYNC_FD_BIT
:
5340 ret
= device
->ws
->export_syncobj_to_sync_file(device
->ws
, syncobj_handle
, pFd
);
5342 if (fence
->temp_syncobj
) {
5343 close (fence
->temp_syncobj
);
5344 fence
->temp_syncobj
= 0;
5346 device
->ws
->reset_syncobj(device
->ws
, syncobj_handle
);
5351 unreachable("Unhandled fence handle type");
5355 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
5359 void radv_GetPhysicalDeviceExternalFenceProperties(
5360 VkPhysicalDevice physicalDevice
,
5361 const VkPhysicalDeviceExternalFenceInfo
*pExternalFenceInfo
,
5362 VkExternalFenceProperties
*pExternalFenceProperties
)
5364 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
5366 if (pdevice
->rad_info
.has_syncobj_wait_for_submit
&&
5367 (pExternalFenceInfo
->handleType
== VK_EXTERNAL_FENCE_HANDLE_TYPE_OPAQUE_FD_BIT
||
5368 pExternalFenceInfo
->handleType
== VK_EXTERNAL_FENCE_HANDLE_TYPE_SYNC_FD_BIT
)) {
5369 pExternalFenceProperties
->exportFromImportedHandleTypes
= VK_EXTERNAL_FENCE_HANDLE_TYPE_OPAQUE_FD_BIT
| VK_EXTERNAL_FENCE_HANDLE_TYPE_SYNC_FD_BIT
;
5370 pExternalFenceProperties
->compatibleHandleTypes
= VK_EXTERNAL_FENCE_HANDLE_TYPE_OPAQUE_FD_BIT
| VK_EXTERNAL_FENCE_HANDLE_TYPE_SYNC_FD_BIT
;
5371 pExternalFenceProperties
->externalFenceFeatures
= VK_EXTERNAL_FENCE_FEATURE_EXPORTABLE_BIT
|
5372 VK_EXTERNAL_SEMAPHORE_FEATURE_IMPORTABLE_BIT
;
5374 pExternalFenceProperties
->exportFromImportedHandleTypes
= 0;
5375 pExternalFenceProperties
->compatibleHandleTypes
= 0;
5376 pExternalFenceProperties
->externalFenceFeatures
= 0;
5381 radv_CreateDebugReportCallbackEXT(VkInstance _instance
,
5382 const VkDebugReportCallbackCreateInfoEXT
* pCreateInfo
,
5383 const VkAllocationCallbacks
* pAllocator
,
5384 VkDebugReportCallbackEXT
* pCallback
)
5386 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
5387 return vk_create_debug_report_callback(&instance
->debug_report_callbacks
,
5388 pCreateInfo
, pAllocator
, &instance
->alloc
,
5393 radv_DestroyDebugReportCallbackEXT(VkInstance _instance
,
5394 VkDebugReportCallbackEXT _callback
,
5395 const VkAllocationCallbacks
* pAllocator
)
5397 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
5398 vk_destroy_debug_report_callback(&instance
->debug_report_callbacks
,
5399 _callback
, pAllocator
, &instance
->alloc
);
5403 radv_DebugReportMessageEXT(VkInstance _instance
,
5404 VkDebugReportFlagsEXT flags
,
5405 VkDebugReportObjectTypeEXT objectType
,
5408 int32_t messageCode
,
5409 const char* pLayerPrefix
,
5410 const char* pMessage
)
5412 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
5413 vk_debug_report(&instance
->debug_report_callbacks
, flags
, objectType
,
5414 object
, location
, messageCode
, pLayerPrefix
, pMessage
);
5418 radv_GetDeviceGroupPeerMemoryFeatures(
5421 uint32_t localDeviceIndex
,
5422 uint32_t remoteDeviceIndex
,
5423 VkPeerMemoryFeatureFlags
* pPeerMemoryFeatures
)
5425 assert(localDeviceIndex
== remoteDeviceIndex
);
5427 *pPeerMemoryFeatures
= VK_PEER_MEMORY_FEATURE_COPY_SRC_BIT
|
5428 VK_PEER_MEMORY_FEATURE_COPY_DST_BIT
|
5429 VK_PEER_MEMORY_FEATURE_GENERIC_SRC_BIT
|
5430 VK_PEER_MEMORY_FEATURE_GENERIC_DST_BIT
;
5433 static const VkTimeDomainEXT radv_time_domains
[] = {
5434 VK_TIME_DOMAIN_DEVICE_EXT
,
5435 VK_TIME_DOMAIN_CLOCK_MONOTONIC_EXT
,
5436 VK_TIME_DOMAIN_CLOCK_MONOTONIC_RAW_EXT
,
5439 VkResult
radv_GetPhysicalDeviceCalibrateableTimeDomainsEXT(
5440 VkPhysicalDevice physicalDevice
,
5441 uint32_t *pTimeDomainCount
,
5442 VkTimeDomainEXT
*pTimeDomains
)
5445 VK_OUTARRAY_MAKE(out
, pTimeDomains
, pTimeDomainCount
);
5447 for (d
= 0; d
< ARRAY_SIZE(radv_time_domains
); d
++) {
5448 vk_outarray_append(&out
, i
) {
5449 *i
= radv_time_domains
[d
];
5453 return vk_outarray_status(&out
);
5457 radv_clock_gettime(clockid_t clock_id
)
5459 struct timespec current
;
5462 ret
= clock_gettime(clock_id
, ¤t
);
5463 if (ret
< 0 && clock_id
== CLOCK_MONOTONIC_RAW
)
5464 ret
= clock_gettime(CLOCK_MONOTONIC
, ¤t
);
5468 return (uint64_t) current
.tv_sec
* 1000000000ULL + current
.tv_nsec
;
5471 VkResult
radv_GetCalibratedTimestampsEXT(
5473 uint32_t timestampCount
,
5474 const VkCalibratedTimestampInfoEXT
*pTimestampInfos
,
5475 uint64_t *pTimestamps
,
5476 uint64_t *pMaxDeviation
)
5478 RADV_FROM_HANDLE(radv_device
, device
, _device
);
5479 uint32_t clock_crystal_freq
= device
->physical_device
->rad_info
.clock_crystal_freq
;
5481 uint64_t begin
, end
;
5482 uint64_t max_clock_period
= 0;
5484 begin
= radv_clock_gettime(CLOCK_MONOTONIC_RAW
);
5486 for (d
= 0; d
< timestampCount
; d
++) {
5487 switch (pTimestampInfos
[d
].timeDomain
) {
5488 case VK_TIME_DOMAIN_DEVICE_EXT
:
5489 pTimestamps
[d
] = device
->ws
->query_value(device
->ws
,
5491 uint64_t device_period
= DIV_ROUND_UP(1000000, clock_crystal_freq
);
5492 max_clock_period
= MAX2(max_clock_period
, device_period
);
5494 case VK_TIME_DOMAIN_CLOCK_MONOTONIC_EXT
:
5495 pTimestamps
[d
] = radv_clock_gettime(CLOCK_MONOTONIC
);
5496 max_clock_period
= MAX2(max_clock_period
, 1);
5499 case VK_TIME_DOMAIN_CLOCK_MONOTONIC_RAW_EXT
:
5500 pTimestamps
[d
] = begin
;
5508 end
= radv_clock_gettime(CLOCK_MONOTONIC_RAW
);
5511 * The maximum deviation is the sum of the interval over which we
5512 * perform the sampling and the maximum period of any sampled
5513 * clock. That's because the maximum skew between any two sampled
5514 * clock edges is when the sampled clock with the largest period is
5515 * sampled at the end of that period but right at the beginning of the
5516 * sampling interval and some other clock is sampled right at the
5517 * begining of its sampling period and right at the end of the
5518 * sampling interval. Let's assume the GPU has the longest clock
5519 * period and that the application is sampling GPU and monotonic:
5522 * w x y z 0 1 2 3 4 5 6 7 8 9 a b c d e f
5523 * Raw -_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-
5527 * GPU -----_____-----_____-----_____-----_____
5530 * x y z 0 1 2 3 4 5 6 7 8 9 a b c
5531 * Monotonic -_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-
5533 * Interval <----------------->
5534 * Deviation <-------------------------->
5538 * m = read(monotonic) 2
5541 * We round the sample interval up by one tick to cover sampling error
5542 * in the interval clock
5545 uint64_t sample_interval
= end
- begin
+ 1;
5547 *pMaxDeviation
= sample_interval
+ max_clock_period
;
5552 void radv_GetPhysicalDeviceMultisamplePropertiesEXT(
5553 VkPhysicalDevice physicalDevice
,
5554 VkSampleCountFlagBits samples
,
5555 VkMultisamplePropertiesEXT
* pMultisampleProperties
)
5557 if (samples
& (VK_SAMPLE_COUNT_2_BIT
|
5558 VK_SAMPLE_COUNT_4_BIT
|
5559 VK_SAMPLE_COUNT_8_BIT
)) {
5560 pMultisampleProperties
->maxSampleLocationGridSize
= (VkExtent2D
){ 2, 2 };
5562 pMultisampleProperties
->maxSampleLocationGridSize
= (VkExtent2D
){ 0, 0 };