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"
48 #include "util/build_id.h"
49 #include "util/debug.h"
50 #include "util/mesa-sha1.h"
53 radv_device_get_cache_uuid(enum radeon_family family
, void *uuid
)
56 unsigned char sha1
[20];
57 unsigned ptr_size
= sizeof(void*);
59 memset(uuid
, 0, VK_UUID_SIZE
);
60 _mesa_sha1_init(&ctx
);
62 if (!disk_cache_get_function_identifier(radv_device_get_cache_uuid
, &ctx
) ||
63 !disk_cache_get_function_identifier(LLVMInitializeAMDGPUTargetInfo
, &ctx
))
66 _mesa_sha1_update(&ctx
, &family
, sizeof(family
));
67 _mesa_sha1_update(&ctx
, &ptr_size
, sizeof(ptr_size
));
68 _mesa_sha1_final(&ctx
, sha1
);
70 memcpy(uuid
, sha1
, VK_UUID_SIZE
);
75 radv_get_driver_uuid(void *uuid
)
77 ac_compute_driver_uuid(uuid
, VK_UUID_SIZE
);
81 radv_get_device_uuid(struct radeon_info
*info
, void *uuid
)
83 ac_compute_device_uuid(info
, uuid
, VK_UUID_SIZE
);
87 radv_get_device_name(enum radeon_family family
, char *name
, size_t name_len
)
89 const char *chip_string
;
90 char llvm_string
[32] = {};
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_MULLINS
: chip_string
= "AMD RADV MULLINS"; break;
103 case CHIP_TONGA
: chip_string
= "AMD RADV TONGA"; break;
104 case CHIP_ICELAND
: chip_string
= "AMD RADV ICELAND"; break;
105 case CHIP_CARRIZO
: chip_string
= "AMD RADV CARRIZO"; break;
106 case CHIP_FIJI
: chip_string
= "AMD RADV FIJI"; break;
107 case CHIP_POLARIS10
: chip_string
= "AMD RADV POLARIS10"; break;
108 case CHIP_POLARIS11
: chip_string
= "AMD RADV POLARIS11"; break;
109 case CHIP_POLARIS12
: chip_string
= "AMD RADV POLARIS12"; break;
110 case CHIP_STONEY
: chip_string
= "AMD RADV STONEY"; break;
111 case CHIP_VEGAM
: chip_string
= "AMD RADV VEGA M"; break;
112 case CHIP_VEGA10
: chip_string
= "AMD RADV VEGA10"; break;
113 case CHIP_VEGA12
: chip_string
= "AMD RADV VEGA12"; break;
114 case CHIP_RAVEN
: chip_string
= "AMD RADV RAVEN"; break;
115 case CHIP_RAVEN2
: chip_string
= "AMD RADV RAVEN2"; break;
116 default: chip_string
= "AMD RADV unknown"; break;
119 snprintf(llvm_string
, sizeof(llvm_string
),
120 " (LLVM %i.%i.%i)", (HAVE_LLVM
>> 8) & 0xff,
121 HAVE_LLVM
& 0xff, MESA_LLVM_VERSION_PATCH
);
122 snprintf(name
, name_len
, "%s%s", chip_string
, llvm_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_VEGA10
)
222 device
->rad_info
.chip_class
= GFX9
;
223 else if (i
>= CHIP_TONGA
)
224 device
->rad_info
.chip_class
= VI
;
225 else if (i
>= CHIP_BONAIRE
)
226 device
->rad_info
.chip_class
= CIK
;
228 device
->rad_info
.chip_class
= SI
;
234 fprintf(stderr
, "radv: Unknown family: %s\n", family
);
239 radv_physical_device_init(struct radv_physical_device
*device
,
240 struct radv_instance
*instance
,
241 drmDevicePtr drm_device
)
243 const char *path
= drm_device
->nodes
[DRM_NODE_RENDER
];
245 drmVersionPtr version
;
249 fd
= open(path
, O_RDWR
| O_CLOEXEC
);
251 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
252 radv_logi("Could not open device '%s'", path
);
254 return vk_error(instance
, VK_ERROR_INCOMPATIBLE_DRIVER
);
257 version
= drmGetVersion(fd
);
261 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
262 radv_logi("Could not get the kernel driver version for device '%s'", path
);
264 return vk_errorf(instance
, VK_ERROR_INCOMPATIBLE_DRIVER
,
265 "failed to get version %s: %m", path
);
268 if (strcmp(version
->name
, "amdgpu")) {
269 drmFreeVersion(version
);
272 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
273 radv_logi("Device '%s' is not using the amdgpu kernel driver.", path
);
275 return VK_ERROR_INCOMPATIBLE_DRIVER
;
277 drmFreeVersion(version
);
279 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
280 radv_logi("Found compatible device '%s'.", path
);
282 device
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
283 device
->instance
= instance
;
285 device
->ws
= radv_amdgpu_winsys_create(fd
, instance
->debug_flags
,
286 instance
->perftest_flags
);
288 result
= vk_error(instance
, VK_ERROR_INCOMPATIBLE_DRIVER
);
292 if (instance
->enabled_extensions
.KHR_display
) {
293 master_fd
= open(drm_device
->nodes
[DRM_NODE_PRIMARY
], O_RDWR
| O_CLOEXEC
);
294 if (master_fd
>= 0) {
295 uint32_t accel_working
= 0;
296 struct drm_amdgpu_info request
= {
297 .return_pointer
= (uintptr_t)&accel_working
,
298 .return_size
= sizeof(accel_working
),
299 .query
= AMDGPU_INFO_ACCEL_WORKING
302 if (drmCommandWrite(master_fd
, DRM_AMDGPU_INFO
, &request
, sizeof (struct drm_amdgpu_info
)) < 0 || !accel_working
) {
309 device
->master_fd
= master_fd
;
310 device
->local_fd
= fd
;
311 device
->ws
->query_info(device
->ws
, &device
->rad_info
);
313 radv_handle_env_var_force_family(device
);
315 radv_get_device_name(device
->rad_info
.family
, device
->name
, sizeof(device
->name
));
317 if (radv_device_get_cache_uuid(device
->rad_info
.family
, device
->cache_uuid
)) {
318 device
->ws
->destroy(device
->ws
);
319 result
= vk_errorf(instance
, VK_ERROR_INITIALIZATION_FAILED
,
320 "cannot generate UUID");
324 /* These flags affect shader compilation. */
325 uint64_t shader_env_flags
=
326 (device
->instance
->perftest_flags
& RADV_PERFTEST_SISCHED
? 0x1 : 0) |
327 (device
->instance
->debug_flags
& RADV_DEBUG_UNSAFE_MATH
? 0x2 : 0);
329 /* The gpu id is already embedded in the uuid so we just pass "radv"
330 * when creating the cache.
332 char buf
[VK_UUID_SIZE
* 2 + 1];
333 disk_cache_format_hex_id(buf
, device
->cache_uuid
, VK_UUID_SIZE
* 2);
334 device
->disk_cache
= disk_cache_create(device
->name
, buf
, shader_env_flags
);
336 if (device
->rad_info
.chip_class
< VI
||
337 device
->rad_info
.chip_class
> GFX9
)
338 fprintf(stderr
, "WARNING: radv is not a conformant vulkan implementation, testing use only.\n");
340 radv_get_driver_uuid(&device
->device_uuid
);
341 radv_get_device_uuid(&device
->rad_info
, &device
->device_uuid
);
343 if (device
->rad_info
.family
== CHIP_STONEY
||
344 device
->rad_info
.chip_class
>= GFX9
) {
345 device
->has_rbplus
= true;
346 device
->rbplus_allowed
= device
->rad_info
.family
== CHIP_STONEY
||
347 device
->rad_info
.family
== CHIP_VEGA12
||
348 device
->rad_info
.family
== CHIP_RAVEN
||
349 device
->rad_info
.family
== CHIP_RAVEN2
;
352 /* The mere presence of CLEAR_STATE in the IB causes random GPU hangs
355 device
->has_clear_state
= device
->rad_info
.chip_class
>= CIK
;
357 device
->cpdma_prefetch_writes_memory
= device
->rad_info
.chip_class
<= VI
;
359 /* Vega10/Raven need a special workaround for a hardware bug. */
360 device
->has_scissor_bug
= device
->rad_info
.family
== CHIP_VEGA10
||
361 device
->rad_info
.family
== CHIP_RAVEN
;
363 /* Out-of-order primitive rasterization. */
364 device
->has_out_of_order_rast
= device
->rad_info
.chip_class
>= VI
&&
365 device
->rad_info
.max_se
>= 2;
366 device
->out_of_order_rast_allowed
= device
->has_out_of_order_rast
&&
367 !(device
->instance
->debug_flags
& RADV_DEBUG_NO_OUT_OF_ORDER
);
369 device
->dcc_msaa_allowed
=
370 (device
->instance
->perftest_flags
& RADV_PERFTEST_DCC_MSAA
);
372 radv_physical_device_init_mem_types(device
);
373 radv_fill_device_extension_table(device
, &device
->supported_extensions
);
375 device
->bus_info
= *drm_device
->businfo
.pci
;
377 if ((device
->instance
->debug_flags
& RADV_DEBUG_INFO
))
378 ac_print_gpu_info(&device
->rad_info
);
380 /* The WSI is structured as a layer on top of the driver, so this has
381 * to be the last part of initialization (at least until we get other
384 result
= radv_init_wsi(device
);
385 if (result
!= VK_SUCCESS
) {
386 device
->ws
->destroy(device
->ws
);
387 vk_error(instance
, result
);
401 radv_physical_device_finish(struct radv_physical_device
*device
)
403 radv_finish_wsi(device
);
404 device
->ws
->destroy(device
->ws
);
405 disk_cache_destroy(device
->disk_cache
);
406 close(device
->local_fd
);
407 if (device
->master_fd
!= -1)
408 close(device
->master_fd
);
412 default_alloc_func(void *pUserData
, size_t size
, size_t align
,
413 VkSystemAllocationScope allocationScope
)
419 default_realloc_func(void *pUserData
, void *pOriginal
, size_t size
,
420 size_t align
, VkSystemAllocationScope allocationScope
)
422 return realloc(pOriginal
, size
);
426 default_free_func(void *pUserData
, void *pMemory
)
431 static const VkAllocationCallbacks default_alloc
= {
433 .pfnAllocation
= default_alloc_func
,
434 .pfnReallocation
= default_realloc_func
,
435 .pfnFree
= default_free_func
,
438 static const struct debug_control radv_debug_options
[] = {
439 {"nofastclears", RADV_DEBUG_NO_FAST_CLEARS
},
440 {"nodcc", RADV_DEBUG_NO_DCC
},
441 {"shaders", RADV_DEBUG_DUMP_SHADERS
},
442 {"nocache", RADV_DEBUG_NO_CACHE
},
443 {"shaderstats", RADV_DEBUG_DUMP_SHADER_STATS
},
444 {"nohiz", RADV_DEBUG_NO_HIZ
},
445 {"nocompute", RADV_DEBUG_NO_COMPUTE_QUEUE
},
446 {"unsafemath", RADV_DEBUG_UNSAFE_MATH
},
447 {"allbos", RADV_DEBUG_ALL_BOS
},
448 {"noibs", RADV_DEBUG_NO_IBS
},
449 {"spirv", RADV_DEBUG_DUMP_SPIRV
},
450 {"vmfaults", RADV_DEBUG_VM_FAULTS
},
451 {"zerovram", RADV_DEBUG_ZERO_VRAM
},
452 {"syncshaders", RADV_DEBUG_SYNC_SHADERS
},
453 {"nosisched", RADV_DEBUG_NO_SISCHED
},
454 {"preoptir", RADV_DEBUG_PREOPTIR
},
455 {"nodynamicbounds", RADV_DEBUG_NO_DYNAMIC_BOUNDS
},
456 {"nooutoforder", RADV_DEBUG_NO_OUT_OF_ORDER
},
457 {"info", RADV_DEBUG_INFO
},
458 {"errors", RADV_DEBUG_ERRORS
},
459 {"startup", RADV_DEBUG_STARTUP
},
460 {"checkir", RADV_DEBUG_CHECKIR
},
461 {"nothreadllvm", RADV_DEBUG_NOTHREADLLVM
},
462 {"nobinning", RADV_DEBUG_NOBINNING
},
467 radv_get_debug_option_name(int id
)
469 assert(id
< ARRAY_SIZE(radv_debug_options
) - 1);
470 return radv_debug_options
[id
].string
;
473 static const struct debug_control radv_perftest_options
[] = {
474 {"nobatchchain", RADV_PERFTEST_NO_BATCHCHAIN
},
475 {"sisched", RADV_PERFTEST_SISCHED
},
476 {"localbos", RADV_PERFTEST_LOCAL_BOS
},
477 {"dccmsaa", RADV_PERFTEST_DCC_MSAA
},
482 radv_get_perftest_option_name(int id
)
484 assert(id
< ARRAY_SIZE(radv_perftest_options
) - 1);
485 return radv_perftest_options
[id
].string
;
489 radv_handle_per_app_options(struct radv_instance
*instance
,
490 const VkApplicationInfo
*info
)
492 const char *name
= info
? info
->pApplicationName
: NULL
;
497 if (!strcmp(name
, "Talos - Linux - 32bit") ||
498 !strcmp(name
, "Talos - Linux - 64bit")) {
499 if (!(instance
->debug_flags
& RADV_DEBUG_NO_SISCHED
)) {
500 /* Force enable LLVM sisched for Talos because it looks
501 * safe and it gives few more FPS.
503 instance
->perftest_flags
|= RADV_PERFTEST_SISCHED
;
505 } else if (!strcmp(name
, "DOOM_VFR")) {
506 /* Work around a Doom VFR game bug */
507 instance
->debug_flags
|= RADV_DEBUG_NO_DYNAMIC_BOUNDS
;
511 static int radv_get_instance_extension_index(const char *name
)
513 for (unsigned i
= 0; i
< RADV_INSTANCE_EXTENSION_COUNT
; ++i
) {
514 if (strcmp(name
, radv_instance_extensions
[i
].extensionName
) == 0)
521 VkResult
radv_CreateInstance(
522 const VkInstanceCreateInfo
* pCreateInfo
,
523 const VkAllocationCallbacks
* pAllocator
,
524 VkInstance
* pInstance
)
526 struct radv_instance
*instance
;
529 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_INSTANCE_CREATE_INFO
);
531 uint32_t client_version
;
532 if (pCreateInfo
->pApplicationInfo
&&
533 pCreateInfo
->pApplicationInfo
->apiVersion
!= 0) {
534 client_version
= pCreateInfo
->pApplicationInfo
->apiVersion
;
536 radv_EnumerateInstanceVersion(&client_version
);
539 instance
= vk_zalloc2(&default_alloc
, pAllocator
, sizeof(*instance
), 8,
540 VK_SYSTEM_ALLOCATION_SCOPE_INSTANCE
);
542 return vk_error(NULL
, VK_ERROR_OUT_OF_HOST_MEMORY
);
544 instance
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
547 instance
->alloc
= *pAllocator
;
549 instance
->alloc
= default_alloc
;
551 instance
->apiVersion
= client_version
;
552 instance
->physicalDeviceCount
= -1;
554 instance
->debug_flags
= parse_debug_string(getenv("RADV_DEBUG"),
557 instance
->perftest_flags
= parse_debug_string(getenv("RADV_PERFTEST"),
558 radv_perftest_options
);
561 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
562 radv_logi("Created an instance");
564 for (uint32_t i
= 0; i
< pCreateInfo
->enabledExtensionCount
; i
++) {
565 const char *ext_name
= pCreateInfo
->ppEnabledExtensionNames
[i
];
566 int index
= radv_get_instance_extension_index(ext_name
);
568 if (index
< 0 || !radv_supported_instance_extensions
.extensions
[index
]) {
569 vk_free2(&default_alloc
, pAllocator
, instance
);
570 return vk_error(instance
, VK_ERROR_EXTENSION_NOT_PRESENT
);
573 instance
->enabled_extensions
.extensions
[index
] = true;
576 result
= vk_debug_report_instance_init(&instance
->debug_report_callbacks
);
577 if (result
!= VK_SUCCESS
) {
578 vk_free2(&default_alloc
, pAllocator
, instance
);
579 return vk_error(instance
, result
);
584 VG(VALGRIND_CREATE_MEMPOOL(instance
, 0, false));
586 radv_handle_per_app_options(instance
, pCreateInfo
->pApplicationInfo
);
588 *pInstance
= radv_instance_to_handle(instance
);
593 void radv_DestroyInstance(
594 VkInstance _instance
,
595 const VkAllocationCallbacks
* pAllocator
)
597 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
602 for (int i
= 0; i
< instance
->physicalDeviceCount
; ++i
) {
603 radv_physical_device_finish(instance
->physicalDevices
+ i
);
606 VG(VALGRIND_DESTROY_MEMPOOL(instance
));
610 vk_debug_report_instance_destroy(&instance
->debug_report_callbacks
);
612 vk_free(&instance
->alloc
, instance
);
616 radv_enumerate_devices(struct radv_instance
*instance
)
618 /* TODO: Check for more devices ? */
619 drmDevicePtr devices
[8];
620 VkResult result
= VK_ERROR_INCOMPATIBLE_DRIVER
;
623 instance
->physicalDeviceCount
= 0;
625 max_devices
= drmGetDevices2(0, devices
, ARRAY_SIZE(devices
));
627 if (instance
->debug_flags
& RADV_DEBUG_STARTUP
)
628 radv_logi("Found %d drm nodes", max_devices
);
631 return vk_error(instance
, VK_ERROR_INCOMPATIBLE_DRIVER
);
633 for (unsigned i
= 0; i
< (unsigned)max_devices
; i
++) {
634 if (devices
[i
]->available_nodes
& 1 << DRM_NODE_RENDER
&&
635 devices
[i
]->bustype
== DRM_BUS_PCI
&&
636 devices
[i
]->deviceinfo
.pci
->vendor_id
== ATI_VENDOR_ID
) {
638 result
= radv_physical_device_init(instance
->physicalDevices
+
639 instance
->physicalDeviceCount
,
642 if (result
== VK_SUCCESS
)
643 ++instance
->physicalDeviceCount
;
644 else if (result
!= VK_ERROR_INCOMPATIBLE_DRIVER
)
648 drmFreeDevices(devices
, max_devices
);
653 VkResult
radv_EnumeratePhysicalDevices(
654 VkInstance _instance
,
655 uint32_t* pPhysicalDeviceCount
,
656 VkPhysicalDevice
* pPhysicalDevices
)
658 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
661 if (instance
->physicalDeviceCount
< 0) {
662 result
= radv_enumerate_devices(instance
);
663 if (result
!= VK_SUCCESS
&&
664 result
!= VK_ERROR_INCOMPATIBLE_DRIVER
)
668 if (!pPhysicalDevices
) {
669 *pPhysicalDeviceCount
= instance
->physicalDeviceCount
;
671 *pPhysicalDeviceCount
= MIN2(*pPhysicalDeviceCount
, instance
->physicalDeviceCount
);
672 for (unsigned i
= 0; i
< *pPhysicalDeviceCount
; ++i
)
673 pPhysicalDevices
[i
] = radv_physical_device_to_handle(instance
->physicalDevices
+ i
);
676 return *pPhysicalDeviceCount
< instance
->physicalDeviceCount
? VK_INCOMPLETE
680 VkResult
radv_EnumeratePhysicalDeviceGroups(
681 VkInstance _instance
,
682 uint32_t* pPhysicalDeviceGroupCount
,
683 VkPhysicalDeviceGroupProperties
* pPhysicalDeviceGroupProperties
)
685 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
688 if (instance
->physicalDeviceCount
< 0) {
689 result
= radv_enumerate_devices(instance
);
690 if (result
!= VK_SUCCESS
&&
691 result
!= VK_ERROR_INCOMPATIBLE_DRIVER
)
695 if (!pPhysicalDeviceGroupProperties
) {
696 *pPhysicalDeviceGroupCount
= instance
->physicalDeviceCount
;
698 *pPhysicalDeviceGroupCount
= MIN2(*pPhysicalDeviceGroupCount
, instance
->physicalDeviceCount
);
699 for (unsigned i
= 0; i
< *pPhysicalDeviceGroupCount
; ++i
) {
700 pPhysicalDeviceGroupProperties
[i
].physicalDeviceCount
= 1;
701 pPhysicalDeviceGroupProperties
[i
].physicalDevices
[0] = radv_physical_device_to_handle(instance
->physicalDevices
+ i
);
702 pPhysicalDeviceGroupProperties
[i
].subsetAllocation
= false;
705 return *pPhysicalDeviceGroupCount
< instance
->physicalDeviceCount
? VK_INCOMPLETE
709 void radv_GetPhysicalDeviceFeatures(
710 VkPhysicalDevice physicalDevice
,
711 VkPhysicalDeviceFeatures
* pFeatures
)
713 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
714 memset(pFeatures
, 0, sizeof(*pFeatures
));
716 *pFeatures
= (VkPhysicalDeviceFeatures
) {
717 .robustBufferAccess
= true,
718 .fullDrawIndexUint32
= true,
719 .imageCubeArray
= true,
720 .independentBlend
= true,
721 .geometryShader
= true,
722 .tessellationShader
= true,
723 .sampleRateShading
= true,
724 .dualSrcBlend
= true,
726 .multiDrawIndirect
= true,
727 .drawIndirectFirstInstance
= true,
729 .depthBiasClamp
= true,
730 .fillModeNonSolid
= true,
735 .multiViewport
= true,
736 .samplerAnisotropy
= true,
737 .textureCompressionETC2
= pdevice
->rad_info
.chip_class
>= GFX9
||
738 pdevice
->rad_info
.family
== CHIP_STONEY
,
739 .textureCompressionASTC_LDR
= false,
740 .textureCompressionBC
= true,
741 .occlusionQueryPrecise
= true,
742 .pipelineStatisticsQuery
= true,
743 .vertexPipelineStoresAndAtomics
= true,
744 .fragmentStoresAndAtomics
= true,
745 .shaderTessellationAndGeometryPointSize
= true,
746 .shaderImageGatherExtended
= true,
747 .shaderStorageImageExtendedFormats
= true,
748 .shaderStorageImageMultisample
= pdevice
->rad_info
.chip_class
>= VI
,
749 .shaderUniformBufferArrayDynamicIndexing
= true,
750 .shaderSampledImageArrayDynamicIndexing
= true,
751 .shaderStorageBufferArrayDynamicIndexing
= true,
752 .shaderStorageImageArrayDynamicIndexing
= true,
753 .shaderStorageImageReadWithoutFormat
= true,
754 .shaderStorageImageWriteWithoutFormat
= true,
755 .shaderClipDistance
= true,
756 .shaderCullDistance
= true,
757 .shaderFloat64
= true,
759 .shaderInt16
= pdevice
->rad_info
.chip_class
>= GFX9
,
760 .sparseBinding
= true,
761 .variableMultisampleRate
= true,
762 .inheritedQueries
= true,
766 void radv_GetPhysicalDeviceFeatures2(
767 VkPhysicalDevice physicalDevice
,
768 VkPhysicalDeviceFeatures2
*pFeatures
)
770 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
771 vk_foreach_struct(ext
, pFeatures
->pNext
) {
772 switch (ext
->sType
) {
773 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VARIABLE_POINTER_FEATURES
: {
774 VkPhysicalDeviceVariablePointerFeatures
*features
= (void *)ext
;
775 features
->variablePointersStorageBuffer
= true;
776 features
->variablePointers
= true;
779 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MULTIVIEW_FEATURES
: {
780 VkPhysicalDeviceMultiviewFeatures
*features
= (VkPhysicalDeviceMultiviewFeatures
*)ext
;
781 features
->multiview
= true;
782 features
->multiviewGeometryShader
= true;
783 features
->multiviewTessellationShader
= true;
786 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_DRAW_PARAMETER_FEATURES
: {
787 VkPhysicalDeviceShaderDrawParameterFeatures
*features
=
788 (VkPhysicalDeviceShaderDrawParameterFeatures
*)ext
;
789 features
->shaderDrawParameters
= true;
792 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROTECTED_MEMORY_FEATURES
: {
793 VkPhysicalDeviceProtectedMemoryFeatures
*features
=
794 (VkPhysicalDeviceProtectedMemoryFeatures
*)ext
;
795 features
->protectedMemory
= false;
798 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_16BIT_STORAGE_FEATURES
: {
799 VkPhysicalDevice16BitStorageFeatures
*features
=
800 (VkPhysicalDevice16BitStorageFeatures
*)ext
;
801 bool enabled
= pdevice
->rad_info
.chip_class
>= VI
;
802 features
->storageBuffer16BitAccess
= enabled
;
803 features
->uniformAndStorageBuffer16BitAccess
= enabled
;
804 features
->storagePushConstant16
= enabled
;
805 features
->storageInputOutput16
= enabled
;
808 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SAMPLER_YCBCR_CONVERSION_FEATURES
: {
809 VkPhysicalDeviceSamplerYcbcrConversionFeatures
*features
=
810 (VkPhysicalDeviceSamplerYcbcrConversionFeatures
*)ext
;
811 features
->samplerYcbcrConversion
= false;
814 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DESCRIPTOR_INDEXING_FEATURES_EXT
: {
815 VkPhysicalDeviceDescriptorIndexingFeaturesEXT
*features
=
816 (VkPhysicalDeviceDescriptorIndexingFeaturesEXT
*)ext
;
817 features
->shaderInputAttachmentArrayDynamicIndexing
= true;
818 features
->shaderUniformTexelBufferArrayDynamicIndexing
= true;
819 features
->shaderStorageTexelBufferArrayDynamicIndexing
= true;
820 features
->shaderUniformBufferArrayNonUniformIndexing
= false;
821 features
->shaderSampledImageArrayNonUniformIndexing
= false;
822 features
->shaderStorageBufferArrayNonUniformIndexing
= false;
823 features
->shaderStorageImageArrayNonUniformIndexing
= false;
824 features
->shaderInputAttachmentArrayNonUniformIndexing
= false;
825 features
->shaderUniformTexelBufferArrayNonUniformIndexing
= false;
826 features
->shaderStorageTexelBufferArrayNonUniformIndexing
= false;
827 features
->descriptorBindingUniformBufferUpdateAfterBind
= true;
828 features
->descriptorBindingSampledImageUpdateAfterBind
= true;
829 features
->descriptorBindingStorageImageUpdateAfterBind
= true;
830 features
->descriptorBindingStorageBufferUpdateAfterBind
= true;
831 features
->descriptorBindingUniformTexelBufferUpdateAfterBind
= true;
832 features
->descriptorBindingStorageTexelBufferUpdateAfterBind
= true;
833 features
->descriptorBindingUpdateUnusedWhilePending
= true;
834 features
->descriptorBindingPartiallyBound
= true;
835 features
->descriptorBindingVariableDescriptorCount
= true;
836 features
->runtimeDescriptorArray
= true;
839 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_CONDITIONAL_RENDERING_FEATURES_EXT
: {
840 VkPhysicalDeviceConditionalRenderingFeaturesEXT
*features
=
841 (VkPhysicalDeviceConditionalRenderingFeaturesEXT
*)ext
;
842 features
->conditionalRendering
= true;
843 features
->inheritedConditionalRendering
= false;
846 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VERTEX_ATTRIBUTE_DIVISOR_FEATURES_EXT
: {
847 VkPhysicalDeviceVertexAttributeDivisorFeaturesEXT
*features
=
848 (VkPhysicalDeviceVertexAttributeDivisorFeaturesEXT
*)ext
;
849 features
->vertexAttributeInstanceRateDivisor
= VK_TRUE
;
850 features
->vertexAttributeInstanceRateZeroDivisor
= VK_TRUE
;
853 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TRANSFORM_FEEDBACK_FEATURES_EXT
: {
854 VkPhysicalDeviceTransformFeedbackFeaturesEXT
*features
=
855 (VkPhysicalDeviceTransformFeedbackFeaturesEXT
*)ext
;
856 features
->transformFeedback
= true;
857 features
->geometryStreams
= true;
860 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SCALAR_BLOCK_LAYOUT_FEATURES_EXT
: {
861 VkPhysicalDeviceScalarBlockLayoutFeaturesEXT
*features
=
862 (VkPhysicalDeviceScalarBlockLayoutFeaturesEXT
*)ext
;
863 features
->scalarBlockLayout
= pdevice
->rad_info
.chip_class
>= CIK
;
870 return radv_GetPhysicalDeviceFeatures(physicalDevice
, &pFeatures
->features
);
873 void radv_GetPhysicalDeviceProperties(
874 VkPhysicalDevice physicalDevice
,
875 VkPhysicalDeviceProperties
* pProperties
)
877 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
878 VkSampleCountFlags sample_counts
= 0xf;
880 /* make sure that the entire descriptor set is addressable with a signed
881 * 32-bit int. So the sum of all limits scaled by descriptor size has to
882 * be at most 2 GiB. the combined image & samples object count as one of
883 * both. This limit is for the pipeline layout, not for the set layout, but
884 * there is no set limit, so we just set a pipeline limit. I don't think
885 * any app is going to hit this soon. */
886 size_t max_descriptor_set_size
= ((1ull << 31) - 16 * MAX_DYNAMIC_BUFFERS
) /
887 (32 /* uniform buffer, 32 due to potential space wasted on alignment */ +
888 32 /* storage buffer, 32 due to potential space wasted on alignment */ +
889 32 /* sampler, largest when combined with image */ +
890 64 /* sampled image */ +
891 64 /* storage image */);
893 VkPhysicalDeviceLimits limits
= {
894 .maxImageDimension1D
= (1 << 14),
895 .maxImageDimension2D
= (1 << 14),
896 .maxImageDimension3D
= (1 << 11),
897 .maxImageDimensionCube
= (1 << 14),
898 .maxImageArrayLayers
= (1 << 11),
899 .maxTexelBufferElements
= 128 * 1024 * 1024,
900 .maxUniformBufferRange
= UINT32_MAX
,
901 .maxStorageBufferRange
= UINT32_MAX
,
902 .maxPushConstantsSize
= MAX_PUSH_CONSTANTS_SIZE
,
903 .maxMemoryAllocationCount
= UINT32_MAX
,
904 .maxSamplerAllocationCount
= 64 * 1024,
905 .bufferImageGranularity
= 64, /* A cache line */
906 .sparseAddressSpaceSize
= 0xffffffffu
, /* buffer max size */
907 .maxBoundDescriptorSets
= MAX_SETS
,
908 .maxPerStageDescriptorSamplers
= max_descriptor_set_size
,
909 .maxPerStageDescriptorUniformBuffers
= max_descriptor_set_size
,
910 .maxPerStageDescriptorStorageBuffers
= max_descriptor_set_size
,
911 .maxPerStageDescriptorSampledImages
= max_descriptor_set_size
,
912 .maxPerStageDescriptorStorageImages
= max_descriptor_set_size
,
913 .maxPerStageDescriptorInputAttachments
= max_descriptor_set_size
,
914 .maxPerStageResources
= max_descriptor_set_size
,
915 .maxDescriptorSetSamplers
= max_descriptor_set_size
,
916 .maxDescriptorSetUniformBuffers
= max_descriptor_set_size
,
917 .maxDescriptorSetUniformBuffersDynamic
= MAX_DYNAMIC_UNIFORM_BUFFERS
,
918 .maxDescriptorSetStorageBuffers
= max_descriptor_set_size
,
919 .maxDescriptorSetStorageBuffersDynamic
= MAX_DYNAMIC_STORAGE_BUFFERS
,
920 .maxDescriptorSetSampledImages
= max_descriptor_set_size
,
921 .maxDescriptorSetStorageImages
= max_descriptor_set_size
,
922 .maxDescriptorSetInputAttachments
= max_descriptor_set_size
,
923 .maxVertexInputAttributes
= 32,
924 .maxVertexInputBindings
= 32,
925 .maxVertexInputAttributeOffset
= 2047,
926 .maxVertexInputBindingStride
= 2048,
927 .maxVertexOutputComponents
= 128,
928 .maxTessellationGenerationLevel
= 64,
929 .maxTessellationPatchSize
= 32,
930 .maxTessellationControlPerVertexInputComponents
= 128,
931 .maxTessellationControlPerVertexOutputComponents
= 128,
932 .maxTessellationControlPerPatchOutputComponents
= 120,
933 .maxTessellationControlTotalOutputComponents
= 4096,
934 .maxTessellationEvaluationInputComponents
= 128,
935 .maxTessellationEvaluationOutputComponents
= 128,
936 .maxGeometryShaderInvocations
= 127,
937 .maxGeometryInputComponents
= 64,
938 .maxGeometryOutputComponents
= 128,
939 .maxGeometryOutputVertices
= 256,
940 .maxGeometryTotalOutputComponents
= 1024,
941 .maxFragmentInputComponents
= 128,
942 .maxFragmentOutputAttachments
= 8,
943 .maxFragmentDualSrcAttachments
= 1,
944 .maxFragmentCombinedOutputResources
= 8,
945 .maxComputeSharedMemorySize
= 32768,
946 .maxComputeWorkGroupCount
= { 65535, 65535, 65535 },
947 .maxComputeWorkGroupInvocations
= 2048,
948 .maxComputeWorkGroupSize
= {
953 .subPixelPrecisionBits
= 8,
954 .subTexelPrecisionBits
= 8,
955 .mipmapPrecisionBits
= 8,
956 .maxDrawIndexedIndexValue
= UINT32_MAX
,
957 .maxDrawIndirectCount
= UINT32_MAX
,
958 .maxSamplerLodBias
= 16,
959 .maxSamplerAnisotropy
= 16,
960 .maxViewports
= MAX_VIEWPORTS
,
961 .maxViewportDimensions
= { (1 << 14), (1 << 14) },
962 .viewportBoundsRange
= { INT16_MIN
, INT16_MAX
},
963 .viewportSubPixelBits
= 8,
964 .minMemoryMapAlignment
= 4096, /* A page */
965 .minTexelBufferOffsetAlignment
= 1,
966 .minUniformBufferOffsetAlignment
= 4,
967 .minStorageBufferOffsetAlignment
= 4,
968 .minTexelOffset
= -32,
969 .maxTexelOffset
= 31,
970 .minTexelGatherOffset
= -32,
971 .maxTexelGatherOffset
= 31,
972 .minInterpolationOffset
= -2,
973 .maxInterpolationOffset
= 2,
974 .subPixelInterpolationOffsetBits
= 8,
975 .maxFramebufferWidth
= (1 << 14),
976 .maxFramebufferHeight
= (1 << 14),
977 .maxFramebufferLayers
= (1 << 10),
978 .framebufferColorSampleCounts
= sample_counts
,
979 .framebufferDepthSampleCounts
= sample_counts
,
980 .framebufferStencilSampleCounts
= sample_counts
,
981 .framebufferNoAttachmentsSampleCounts
= sample_counts
,
982 .maxColorAttachments
= MAX_RTS
,
983 .sampledImageColorSampleCounts
= sample_counts
,
984 .sampledImageIntegerSampleCounts
= VK_SAMPLE_COUNT_1_BIT
,
985 .sampledImageDepthSampleCounts
= sample_counts
,
986 .sampledImageStencilSampleCounts
= sample_counts
,
987 .storageImageSampleCounts
= pdevice
->rad_info
.chip_class
>= VI
? sample_counts
: VK_SAMPLE_COUNT_1_BIT
,
988 .maxSampleMaskWords
= 1,
989 .timestampComputeAndGraphics
= true,
990 .timestampPeriod
= 1000000.0 / pdevice
->rad_info
.clock_crystal_freq
,
991 .maxClipDistances
= 8,
992 .maxCullDistances
= 8,
993 .maxCombinedClipAndCullDistances
= 8,
994 .discreteQueuePriorities
= 2,
995 .pointSizeRange
= { 0.125, 255.875 },
996 .lineWidthRange
= { 0.0, 7.9921875 },
997 .pointSizeGranularity
= (1.0 / 8.0),
998 .lineWidthGranularity
= (1.0 / 128.0),
999 .strictLines
= false, /* FINISHME */
1000 .standardSampleLocations
= true,
1001 .optimalBufferCopyOffsetAlignment
= 128,
1002 .optimalBufferCopyRowPitchAlignment
= 128,
1003 .nonCoherentAtomSize
= 64,
1006 *pProperties
= (VkPhysicalDeviceProperties
) {
1007 .apiVersion
= radv_physical_device_api_version(pdevice
),
1008 .driverVersion
= vk_get_driver_version(),
1009 .vendorID
= ATI_VENDOR_ID
,
1010 .deviceID
= pdevice
->rad_info
.pci_id
,
1011 .deviceType
= pdevice
->rad_info
.has_dedicated_vram
? VK_PHYSICAL_DEVICE_TYPE_DISCRETE_GPU
: VK_PHYSICAL_DEVICE_TYPE_INTEGRATED_GPU
,
1013 .sparseProperties
= {0},
1016 strcpy(pProperties
->deviceName
, pdevice
->name
);
1017 memcpy(pProperties
->pipelineCacheUUID
, pdevice
->cache_uuid
, VK_UUID_SIZE
);
1020 void radv_GetPhysicalDeviceProperties2(
1021 VkPhysicalDevice physicalDevice
,
1022 VkPhysicalDeviceProperties2
*pProperties
)
1024 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
1025 radv_GetPhysicalDeviceProperties(physicalDevice
, &pProperties
->properties
);
1027 vk_foreach_struct(ext
, pProperties
->pNext
) {
1028 switch (ext
->sType
) {
1029 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PUSH_DESCRIPTOR_PROPERTIES_KHR
: {
1030 VkPhysicalDevicePushDescriptorPropertiesKHR
*properties
=
1031 (VkPhysicalDevicePushDescriptorPropertiesKHR
*) ext
;
1032 properties
->maxPushDescriptors
= MAX_PUSH_DESCRIPTORS
;
1035 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_ID_PROPERTIES
: {
1036 VkPhysicalDeviceIDProperties
*properties
= (VkPhysicalDeviceIDProperties
*)ext
;
1037 memcpy(properties
->driverUUID
, pdevice
->driver_uuid
, VK_UUID_SIZE
);
1038 memcpy(properties
->deviceUUID
, pdevice
->device_uuid
, VK_UUID_SIZE
);
1039 properties
->deviceLUIDValid
= false;
1042 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MULTIVIEW_PROPERTIES
: {
1043 VkPhysicalDeviceMultiviewProperties
*properties
= (VkPhysicalDeviceMultiviewProperties
*)ext
;
1044 properties
->maxMultiviewViewCount
= MAX_VIEWS
;
1045 properties
->maxMultiviewInstanceIndex
= INT_MAX
;
1048 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_POINT_CLIPPING_PROPERTIES
: {
1049 VkPhysicalDevicePointClippingProperties
*properties
=
1050 (VkPhysicalDevicePointClippingProperties
*)ext
;
1051 properties
->pointClippingBehavior
= VK_POINT_CLIPPING_BEHAVIOR_ALL_CLIP_PLANES
;
1054 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DISCARD_RECTANGLE_PROPERTIES_EXT
: {
1055 VkPhysicalDeviceDiscardRectanglePropertiesEXT
*properties
=
1056 (VkPhysicalDeviceDiscardRectanglePropertiesEXT
*)ext
;
1057 properties
->maxDiscardRectangles
= MAX_DISCARD_RECTANGLES
;
1060 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_EXTERNAL_MEMORY_HOST_PROPERTIES_EXT
: {
1061 VkPhysicalDeviceExternalMemoryHostPropertiesEXT
*properties
=
1062 (VkPhysicalDeviceExternalMemoryHostPropertiesEXT
*) ext
;
1063 properties
->minImportedHostPointerAlignment
= 4096;
1066 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SUBGROUP_PROPERTIES
: {
1067 VkPhysicalDeviceSubgroupProperties
*properties
=
1068 (VkPhysicalDeviceSubgroupProperties
*)ext
;
1069 properties
->subgroupSize
= 64;
1070 properties
->supportedStages
= VK_SHADER_STAGE_ALL
;
1071 properties
->supportedOperations
=
1072 VK_SUBGROUP_FEATURE_BASIC_BIT
|
1073 VK_SUBGROUP_FEATURE_BALLOT_BIT
|
1074 VK_SUBGROUP_FEATURE_QUAD_BIT
|
1075 VK_SUBGROUP_FEATURE_VOTE_BIT
;
1076 if (pdevice
->rad_info
.chip_class
>= VI
) {
1077 properties
->supportedOperations
|=
1078 VK_SUBGROUP_FEATURE_ARITHMETIC_BIT
|
1079 VK_SUBGROUP_FEATURE_SHUFFLE_BIT
|
1080 VK_SUBGROUP_FEATURE_SHUFFLE_RELATIVE_BIT
;
1082 properties
->quadOperationsInAllStages
= true;
1085 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MAINTENANCE_3_PROPERTIES
: {
1086 VkPhysicalDeviceMaintenance3Properties
*properties
=
1087 (VkPhysicalDeviceMaintenance3Properties
*)ext
;
1088 /* Make sure everything is addressable by a signed 32-bit int, and
1089 * our largest descriptors are 96 bytes. */
1090 properties
->maxPerSetDescriptors
= (1ull << 31) / 96;
1091 /* Our buffer size fields allow only this much */
1092 properties
->maxMemoryAllocationSize
= 0xFFFFFFFFull
;
1095 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SAMPLER_FILTER_MINMAX_PROPERTIES_EXT
: {
1096 VkPhysicalDeviceSamplerFilterMinmaxPropertiesEXT
*properties
=
1097 (VkPhysicalDeviceSamplerFilterMinmaxPropertiesEXT
*)ext
;
1098 /* GFX6-8 only support single channel min/max filter. */
1099 properties
->filterMinmaxImageComponentMapping
= pdevice
->rad_info
.chip_class
>= GFX9
;
1100 properties
->filterMinmaxSingleComponentFormats
= true;
1103 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_SHADER_CORE_PROPERTIES_AMD
: {
1104 VkPhysicalDeviceShaderCorePropertiesAMD
*properties
=
1105 (VkPhysicalDeviceShaderCorePropertiesAMD
*)ext
;
1107 /* Shader engines. */
1108 properties
->shaderEngineCount
=
1109 pdevice
->rad_info
.max_se
;
1110 properties
->shaderArraysPerEngineCount
=
1111 pdevice
->rad_info
.max_sh_per_se
;
1112 properties
->computeUnitsPerShaderArray
=
1113 pdevice
->rad_info
.num_good_cu_per_sh
;
1114 properties
->simdPerComputeUnit
= 4;
1115 properties
->wavefrontsPerSimd
=
1116 pdevice
->rad_info
.family
== CHIP_TONGA
||
1117 pdevice
->rad_info
.family
== CHIP_ICELAND
||
1118 pdevice
->rad_info
.family
== CHIP_POLARIS10
||
1119 pdevice
->rad_info
.family
== CHIP_POLARIS11
||
1120 pdevice
->rad_info
.family
== CHIP_POLARIS12
||
1121 pdevice
->rad_info
.family
== CHIP_VEGAM
? 8 : 10;
1122 properties
->wavefrontSize
= 64;
1125 properties
->sgprsPerSimd
=
1126 radv_get_num_physical_sgprs(pdevice
);
1127 properties
->minSgprAllocation
=
1128 pdevice
->rad_info
.chip_class
>= VI
? 16 : 8;
1129 properties
->maxSgprAllocation
=
1130 pdevice
->rad_info
.family
== CHIP_TONGA
||
1131 pdevice
->rad_info
.family
== CHIP_ICELAND
? 96 : 104;
1132 properties
->sgprAllocationGranularity
=
1133 pdevice
->rad_info
.chip_class
>= VI
? 16 : 8;
1136 properties
->vgprsPerSimd
= RADV_NUM_PHYSICAL_VGPRS
;
1137 properties
->minVgprAllocation
= 4;
1138 properties
->maxVgprAllocation
= 256;
1139 properties
->vgprAllocationGranularity
= 4;
1142 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VERTEX_ATTRIBUTE_DIVISOR_PROPERTIES_EXT
: {
1143 VkPhysicalDeviceVertexAttributeDivisorPropertiesEXT
*properties
=
1144 (VkPhysicalDeviceVertexAttributeDivisorPropertiesEXT
*)ext
;
1145 properties
->maxVertexAttribDivisor
= UINT32_MAX
;
1148 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DESCRIPTOR_INDEXING_PROPERTIES_EXT
: {
1149 VkPhysicalDeviceDescriptorIndexingPropertiesEXT
*properties
=
1150 (VkPhysicalDeviceDescriptorIndexingPropertiesEXT
*)ext
;
1151 properties
->maxUpdateAfterBindDescriptorsInAllPools
= UINT32_MAX
/ 64;
1152 properties
->shaderUniformBufferArrayNonUniformIndexingNative
= false;
1153 properties
->shaderSampledImageArrayNonUniformIndexingNative
= false;
1154 properties
->shaderStorageBufferArrayNonUniformIndexingNative
= false;
1155 properties
->shaderStorageImageArrayNonUniformIndexingNative
= false;
1156 properties
->shaderInputAttachmentArrayNonUniformIndexingNative
= false;
1157 properties
->robustBufferAccessUpdateAfterBind
= false;
1158 properties
->quadDivergentImplicitLod
= false;
1160 size_t max_descriptor_set_size
= ((1ull << 31) - 16 * MAX_DYNAMIC_BUFFERS
) /
1161 (32 /* uniform buffer, 32 due to potential space wasted on alignment */ +
1162 32 /* storage buffer, 32 due to potential space wasted on alignment */ +
1163 32 /* sampler, largest when combined with image */ +
1164 64 /* sampled image */ +
1165 64 /* storage image */);
1166 properties
->maxPerStageDescriptorUpdateAfterBindSamplers
= max_descriptor_set_size
;
1167 properties
->maxPerStageDescriptorUpdateAfterBindUniformBuffers
= max_descriptor_set_size
;
1168 properties
->maxPerStageDescriptorUpdateAfterBindStorageBuffers
= max_descriptor_set_size
;
1169 properties
->maxPerStageDescriptorUpdateAfterBindSampledImages
= max_descriptor_set_size
;
1170 properties
->maxPerStageDescriptorUpdateAfterBindStorageImages
= max_descriptor_set_size
;
1171 properties
->maxPerStageDescriptorUpdateAfterBindInputAttachments
= max_descriptor_set_size
;
1172 properties
->maxPerStageUpdateAfterBindResources
= max_descriptor_set_size
;
1173 properties
->maxDescriptorSetUpdateAfterBindSamplers
= max_descriptor_set_size
;
1174 properties
->maxDescriptorSetUpdateAfterBindUniformBuffers
= max_descriptor_set_size
;
1175 properties
->maxDescriptorSetUpdateAfterBindUniformBuffersDynamic
= MAX_DYNAMIC_UNIFORM_BUFFERS
;
1176 properties
->maxDescriptorSetUpdateAfterBindStorageBuffers
= max_descriptor_set_size
;
1177 properties
->maxDescriptorSetUpdateAfterBindStorageBuffersDynamic
= MAX_DYNAMIC_STORAGE_BUFFERS
;
1178 properties
->maxDescriptorSetUpdateAfterBindSampledImages
= max_descriptor_set_size
;
1179 properties
->maxDescriptorSetUpdateAfterBindStorageImages
= max_descriptor_set_size
;
1180 properties
->maxDescriptorSetUpdateAfterBindInputAttachments
= max_descriptor_set_size
;
1183 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PROTECTED_MEMORY_PROPERTIES
: {
1184 VkPhysicalDeviceProtectedMemoryProperties
*properties
=
1185 (VkPhysicalDeviceProtectedMemoryProperties
*)ext
;
1186 properties
->protectedNoFault
= false;
1189 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_CONSERVATIVE_RASTERIZATION_PROPERTIES_EXT
: {
1190 VkPhysicalDeviceConservativeRasterizationPropertiesEXT
*properties
=
1191 (VkPhysicalDeviceConservativeRasterizationPropertiesEXT
*)ext
;
1192 properties
->primitiveOverestimationSize
= 0;
1193 properties
->maxExtraPrimitiveOverestimationSize
= 0;
1194 properties
->extraPrimitiveOverestimationSizeGranularity
= 0;
1195 properties
->primitiveUnderestimation
= VK_FALSE
;
1196 properties
->conservativePointAndLineRasterization
= VK_FALSE
;
1197 properties
->degenerateTrianglesRasterized
= VK_FALSE
;
1198 properties
->degenerateLinesRasterized
= VK_FALSE
;
1199 properties
->fullyCoveredFragmentShaderInputVariable
= VK_FALSE
;
1200 properties
->conservativeRasterizationPostDepthCoverage
= VK_FALSE
;
1203 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PCI_BUS_INFO_PROPERTIES_EXT
: {
1204 VkPhysicalDevicePCIBusInfoPropertiesEXT
*properties
=
1205 (VkPhysicalDevicePCIBusInfoPropertiesEXT
*)ext
;
1206 properties
->pciDomain
= pdevice
->bus_info
.domain
;
1207 properties
->pciBus
= pdevice
->bus_info
.bus
;
1208 properties
->pciDevice
= pdevice
->bus_info
.dev
;
1209 properties
->pciFunction
= pdevice
->bus_info
.func
;
1212 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_DRIVER_PROPERTIES_KHR
: {
1213 VkPhysicalDeviceDriverPropertiesKHR
*driver_props
=
1214 (VkPhysicalDeviceDriverPropertiesKHR
*) ext
;
1216 driver_props
->driverID
= VK_DRIVER_ID_MESA_RADV_KHR
;
1217 memset(driver_props
->driverName
, 0, VK_MAX_DRIVER_NAME_SIZE_KHR
);
1218 strcpy(driver_props
->driverName
, "radv");
1220 memset(driver_props
->driverInfo
, 0, VK_MAX_DRIVER_INFO_SIZE_KHR
);
1221 snprintf(driver_props
->driverInfo
, VK_MAX_DRIVER_INFO_SIZE_KHR
,
1222 "Mesa " PACKAGE_VERSION MESA_GIT_SHA1
1224 (HAVE_LLVM
>> 8) & 0xff, HAVE_LLVM
& 0xff,
1225 MESA_LLVM_VERSION_PATCH
);
1227 driver_props
->conformanceVersion
= (VkConformanceVersionKHR
) {
1235 case VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TRANSFORM_FEEDBACK_PROPERTIES_EXT
: {
1236 VkPhysicalDeviceTransformFeedbackPropertiesEXT
*properties
=
1237 (VkPhysicalDeviceTransformFeedbackPropertiesEXT
*)ext
;
1238 properties
->maxTransformFeedbackStreams
= MAX_SO_STREAMS
;
1239 properties
->maxTransformFeedbackBuffers
= MAX_SO_BUFFERS
;
1240 properties
->maxTransformFeedbackBufferSize
= UINT32_MAX
;
1241 properties
->maxTransformFeedbackStreamDataSize
= 512;
1242 properties
->maxTransformFeedbackBufferDataSize
= UINT32_MAX
;
1243 properties
->maxTransformFeedbackBufferDataStride
= 512;
1244 properties
->transformFeedbackQueries
= true;
1245 properties
->transformFeedbackStreamsLinesTriangles
= false;
1246 properties
->transformFeedbackRasterizationStreamSelect
= false;
1247 properties
->transformFeedbackDraw
= true;
1256 static void radv_get_physical_device_queue_family_properties(
1257 struct radv_physical_device
* pdevice
,
1259 VkQueueFamilyProperties
** pQueueFamilyProperties
)
1261 int num_queue_families
= 1;
1263 if (pdevice
->rad_info
.num_compute_rings
> 0 &&
1264 !(pdevice
->instance
->debug_flags
& RADV_DEBUG_NO_COMPUTE_QUEUE
))
1265 num_queue_families
++;
1267 if (pQueueFamilyProperties
== NULL
) {
1268 *pCount
= num_queue_families
;
1277 *pQueueFamilyProperties
[idx
] = (VkQueueFamilyProperties
) {
1278 .queueFlags
= VK_QUEUE_GRAPHICS_BIT
|
1279 VK_QUEUE_COMPUTE_BIT
|
1280 VK_QUEUE_TRANSFER_BIT
|
1281 VK_QUEUE_SPARSE_BINDING_BIT
,
1283 .timestampValidBits
= 64,
1284 .minImageTransferGranularity
= (VkExtent3D
) { 1, 1, 1 },
1289 if (pdevice
->rad_info
.num_compute_rings
> 0 &&
1290 !(pdevice
->instance
->debug_flags
& RADV_DEBUG_NO_COMPUTE_QUEUE
)) {
1291 if (*pCount
> idx
) {
1292 *pQueueFamilyProperties
[idx
] = (VkQueueFamilyProperties
) {
1293 .queueFlags
= VK_QUEUE_COMPUTE_BIT
|
1294 VK_QUEUE_TRANSFER_BIT
|
1295 VK_QUEUE_SPARSE_BINDING_BIT
,
1296 .queueCount
= pdevice
->rad_info
.num_compute_rings
,
1297 .timestampValidBits
= 64,
1298 .minImageTransferGranularity
= (VkExtent3D
) { 1, 1, 1 },
1306 void radv_GetPhysicalDeviceQueueFamilyProperties(
1307 VkPhysicalDevice physicalDevice
,
1309 VkQueueFamilyProperties
* pQueueFamilyProperties
)
1311 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
1312 if (!pQueueFamilyProperties
) {
1313 radv_get_physical_device_queue_family_properties(pdevice
, pCount
, NULL
);
1316 VkQueueFamilyProperties
*properties
[] = {
1317 pQueueFamilyProperties
+ 0,
1318 pQueueFamilyProperties
+ 1,
1319 pQueueFamilyProperties
+ 2,
1321 radv_get_physical_device_queue_family_properties(pdevice
, pCount
, properties
);
1322 assert(*pCount
<= 3);
1325 void radv_GetPhysicalDeviceQueueFamilyProperties2(
1326 VkPhysicalDevice physicalDevice
,
1328 VkQueueFamilyProperties2
*pQueueFamilyProperties
)
1330 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
1331 if (!pQueueFamilyProperties
) {
1332 radv_get_physical_device_queue_family_properties(pdevice
, pCount
, NULL
);
1335 VkQueueFamilyProperties
*properties
[] = {
1336 &pQueueFamilyProperties
[0].queueFamilyProperties
,
1337 &pQueueFamilyProperties
[1].queueFamilyProperties
,
1338 &pQueueFamilyProperties
[2].queueFamilyProperties
,
1340 radv_get_physical_device_queue_family_properties(pdevice
, pCount
, properties
);
1341 assert(*pCount
<= 3);
1344 void radv_GetPhysicalDeviceMemoryProperties(
1345 VkPhysicalDevice physicalDevice
,
1346 VkPhysicalDeviceMemoryProperties
*pMemoryProperties
)
1348 RADV_FROM_HANDLE(radv_physical_device
, physical_device
, physicalDevice
);
1350 *pMemoryProperties
= physical_device
->memory_properties
;
1353 void radv_GetPhysicalDeviceMemoryProperties2(
1354 VkPhysicalDevice physicalDevice
,
1355 VkPhysicalDeviceMemoryProperties2
*pMemoryProperties
)
1357 radv_GetPhysicalDeviceMemoryProperties(physicalDevice
,
1358 &pMemoryProperties
->memoryProperties
);
1361 VkResult
radv_GetMemoryHostPointerPropertiesEXT(
1363 VkExternalMemoryHandleTypeFlagBits handleType
,
1364 const void *pHostPointer
,
1365 VkMemoryHostPointerPropertiesEXT
*pMemoryHostPointerProperties
)
1367 RADV_FROM_HANDLE(radv_device
, device
, _device
);
1371 case VK_EXTERNAL_MEMORY_HANDLE_TYPE_HOST_ALLOCATION_BIT_EXT
: {
1372 const struct radv_physical_device
*physical_device
= device
->physical_device
;
1373 uint32_t memoryTypeBits
= 0;
1374 for (int i
= 0; i
< physical_device
->memory_properties
.memoryTypeCount
; i
++) {
1375 if (physical_device
->mem_type_indices
[i
] == RADV_MEM_TYPE_GTT_CACHED
) {
1376 memoryTypeBits
= (1 << i
);
1380 pMemoryHostPointerProperties
->memoryTypeBits
= memoryTypeBits
;
1384 return VK_ERROR_INVALID_EXTERNAL_HANDLE
;
1388 static enum radeon_ctx_priority
1389 radv_get_queue_global_priority(const VkDeviceQueueGlobalPriorityCreateInfoEXT
*pObj
)
1391 /* Default to MEDIUM when a specific global priority isn't requested */
1393 return RADEON_CTX_PRIORITY_MEDIUM
;
1395 switch(pObj
->globalPriority
) {
1396 case VK_QUEUE_GLOBAL_PRIORITY_REALTIME_EXT
:
1397 return RADEON_CTX_PRIORITY_REALTIME
;
1398 case VK_QUEUE_GLOBAL_PRIORITY_HIGH_EXT
:
1399 return RADEON_CTX_PRIORITY_HIGH
;
1400 case VK_QUEUE_GLOBAL_PRIORITY_MEDIUM_EXT
:
1401 return RADEON_CTX_PRIORITY_MEDIUM
;
1402 case VK_QUEUE_GLOBAL_PRIORITY_LOW_EXT
:
1403 return RADEON_CTX_PRIORITY_LOW
;
1405 unreachable("Illegal global priority value");
1406 return RADEON_CTX_PRIORITY_INVALID
;
1411 radv_queue_init(struct radv_device
*device
, struct radv_queue
*queue
,
1412 uint32_t queue_family_index
, int idx
,
1413 VkDeviceQueueCreateFlags flags
,
1414 const VkDeviceQueueGlobalPriorityCreateInfoEXT
*global_priority
)
1416 queue
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
1417 queue
->device
= device
;
1418 queue
->queue_family_index
= queue_family_index
;
1419 queue
->queue_idx
= idx
;
1420 queue
->priority
= radv_get_queue_global_priority(global_priority
);
1421 queue
->flags
= flags
;
1423 queue
->hw_ctx
= device
->ws
->ctx_create(device
->ws
, queue
->priority
);
1425 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
1431 radv_queue_finish(struct radv_queue
*queue
)
1434 queue
->device
->ws
->ctx_destroy(queue
->hw_ctx
);
1436 if (queue
->initial_full_flush_preamble_cs
)
1437 queue
->device
->ws
->cs_destroy(queue
->initial_full_flush_preamble_cs
);
1438 if (queue
->initial_preamble_cs
)
1439 queue
->device
->ws
->cs_destroy(queue
->initial_preamble_cs
);
1440 if (queue
->continue_preamble_cs
)
1441 queue
->device
->ws
->cs_destroy(queue
->continue_preamble_cs
);
1442 if (queue
->descriptor_bo
)
1443 queue
->device
->ws
->buffer_destroy(queue
->descriptor_bo
);
1444 if (queue
->scratch_bo
)
1445 queue
->device
->ws
->buffer_destroy(queue
->scratch_bo
);
1446 if (queue
->esgs_ring_bo
)
1447 queue
->device
->ws
->buffer_destroy(queue
->esgs_ring_bo
);
1448 if (queue
->gsvs_ring_bo
)
1449 queue
->device
->ws
->buffer_destroy(queue
->gsvs_ring_bo
);
1450 if (queue
->tess_rings_bo
)
1451 queue
->device
->ws
->buffer_destroy(queue
->tess_rings_bo
);
1452 if (queue
->compute_scratch_bo
)
1453 queue
->device
->ws
->buffer_destroy(queue
->compute_scratch_bo
);
1457 radv_bo_list_init(struct radv_bo_list
*bo_list
)
1459 pthread_mutex_init(&bo_list
->mutex
, NULL
);
1460 bo_list
->list
.count
= bo_list
->capacity
= 0;
1461 bo_list
->list
.bos
= NULL
;
1465 radv_bo_list_finish(struct radv_bo_list
*bo_list
)
1467 free(bo_list
->list
.bos
);
1468 pthread_mutex_destroy(&bo_list
->mutex
);
1471 static VkResult
radv_bo_list_add(struct radv_device
*device
,
1472 struct radeon_winsys_bo
*bo
)
1474 struct radv_bo_list
*bo_list
= &device
->bo_list
;
1476 if (unlikely(!device
->use_global_bo_list
))
1479 pthread_mutex_lock(&bo_list
->mutex
);
1480 if (bo_list
->list
.count
== bo_list
->capacity
) {
1481 unsigned capacity
= MAX2(4, bo_list
->capacity
* 2);
1482 void *data
= realloc(bo_list
->list
.bos
, capacity
* sizeof(struct radeon_winsys_bo
*));
1485 pthread_mutex_unlock(&bo_list
->mutex
);
1486 return VK_ERROR_OUT_OF_HOST_MEMORY
;
1489 bo_list
->list
.bos
= (struct radeon_winsys_bo
**)data
;
1490 bo_list
->capacity
= capacity
;
1493 bo_list
->list
.bos
[bo_list
->list
.count
++] = bo
;
1494 pthread_mutex_unlock(&bo_list
->mutex
);
1498 static void radv_bo_list_remove(struct radv_device
*device
,
1499 struct radeon_winsys_bo
*bo
)
1501 struct radv_bo_list
*bo_list
= &device
->bo_list
;
1503 if (unlikely(!device
->use_global_bo_list
))
1506 pthread_mutex_lock(&bo_list
->mutex
);
1507 for(unsigned i
= 0; i
< bo_list
->list
.count
; ++i
) {
1508 if (bo_list
->list
.bos
[i
] == bo
) {
1509 bo_list
->list
.bos
[i
] = bo_list
->list
.bos
[bo_list
->list
.count
- 1];
1510 --bo_list
->list
.count
;
1514 pthread_mutex_unlock(&bo_list
->mutex
);
1518 radv_device_init_gs_info(struct radv_device
*device
)
1520 device
->gs_table_depth
= ac_get_gs_table_depth(device
->physical_device
->rad_info
.chip_class
,
1521 device
->physical_device
->rad_info
.family
);
1524 static int radv_get_device_extension_index(const char *name
)
1526 for (unsigned i
= 0; i
< RADV_DEVICE_EXTENSION_COUNT
; ++i
) {
1527 if (strcmp(name
, radv_device_extensions
[i
].extensionName
) == 0)
1534 radv_get_int_debug_option(const char *name
, int default_value
)
1541 result
= default_value
;
1545 result
= strtol(str
, &endptr
, 0);
1546 if (str
== endptr
) {
1547 /* No digits founs. */
1548 result
= default_value
;
1555 VkResult
radv_CreateDevice(
1556 VkPhysicalDevice physicalDevice
,
1557 const VkDeviceCreateInfo
* pCreateInfo
,
1558 const VkAllocationCallbacks
* pAllocator
,
1561 RADV_FROM_HANDLE(radv_physical_device
, physical_device
, physicalDevice
);
1563 struct radv_device
*device
;
1565 bool keep_shader_info
= false;
1567 /* Check enabled features */
1568 if (pCreateInfo
->pEnabledFeatures
) {
1569 VkPhysicalDeviceFeatures supported_features
;
1570 radv_GetPhysicalDeviceFeatures(physicalDevice
, &supported_features
);
1571 VkBool32
*supported_feature
= (VkBool32
*)&supported_features
;
1572 VkBool32
*enabled_feature
= (VkBool32
*)pCreateInfo
->pEnabledFeatures
;
1573 unsigned num_features
= sizeof(VkPhysicalDeviceFeatures
) / sizeof(VkBool32
);
1574 for (uint32_t i
= 0; i
< num_features
; i
++) {
1575 if (enabled_feature
[i
] && !supported_feature
[i
])
1576 return vk_error(physical_device
->instance
, VK_ERROR_FEATURE_NOT_PRESENT
);
1580 device
= vk_zalloc2(&physical_device
->instance
->alloc
, pAllocator
,
1582 VK_SYSTEM_ALLOCATION_SCOPE_DEVICE
);
1584 return vk_error(physical_device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
1586 device
->_loader_data
.loaderMagic
= ICD_LOADER_MAGIC
;
1587 device
->instance
= physical_device
->instance
;
1588 device
->physical_device
= physical_device
;
1590 device
->ws
= physical_device
->ws
;
1592 device
->alloc
= *pAllocator
;
1594 device
->alloc
= physical_device
->instance
->alloc
;
1596 for (uint32_t i
= 0; i
< pCreateInfo
->enabledExtensionCount
; i
++) {
1597 const char *ext_name
= pCreateInfo
->ppEnabledExtensionNames
[i
];
1598 int index
= radv_get_device_extension_index(ext_name
);
1599 if (index
< 0 || !physical_device
->supported_extensions
.extensions
[index
]) {
1600 vk_free(&device
->alloc
, device
);
1601 return vk_error(physical_device
->instance
, VK_ERROR_EXTENSION_NOT_PRESENT
);
1604 device
->enabled_extensions
.extensions
[index
] = true;
1607 keep_shader_info
= device
->enabled_extensions
.AMD_shader_info
;
1609 /* With update after bind we can't attach bo's to the command buffer
1610 * from the descriptor set anymore, so we have to use a global BO list.
1612 device
->use_global_bo_list
=
1613 device
->enabled_extensions
.EXT_descriptor_indexing
;
1615 mtx_init(&device
->shader_slab_mutex
, mtx_plain
);
1616 list_inithead(&device
->shader_slabs
);
1618 radv_bo_list_init(&device
->bo_list
);
1620 for (unsigned i
= 0; i
< pCreateInfo
->queueCreateInfoCount
; i
++) {
1621 const VkDeviceQueueCreateInfo
*queue_create
= &pCreateInfo
->pQueueCreateInfos
[i
];
1622 uint32_t qfi
= queue_create
->queueFamilyIndex
;
1623 const VkDeviceQueueGlobalPriorityCreateInfoEXT
*global_priority
=
1624 vk_find_struct_const(queue_create
->pNext
, DEVICE_QUEUE_GLOBAL_PRIORITY_CREATE_INFO_EXT
);
1626 assert(!global_priority
|| device
->physical_device
->rad_info
.has_ctx_priority
);
1628 device
->queues
[qfi
] = vk_alloc(&device
->alloc
,
1629 queue_create
->queueCount
* sizeof(struct radv_queue
), 8, VK_SYSTEM_ALLOCATION_SCOPE_DEVICE
);
1630 if (!device
->queues
[qfi
]) {
1631 result
= VK_ERROR_OUT_OF_HOST_MEMORY
;
1635 memset(device
->queues
[qfi
], 0, queue_create
->queueCount
* sizeof(struct radv_queue
));
1637 device
->queue_count
[qfi
] = queue_create
->queueCount
;
1639 for (unsigned q
= 0; q
< queue_create
->queueCount
; q
++) {
1640 result
= radv_queue_init(device
, &device
->queues
[qfi
][q
],
1641 qfi
, q
, queue_create
->flags
,
1643 if (result
!= VK_SUCCESS
)
1648 device
->pbb_allowed
= device
->physical_device
->rad_info
.chip_class
>= GFX9
&&
1649 !(device
->instance
->debug_flags
& RADV_DEBUG_NOBINNING
);
1651 /* Disabled and not implemented for now. */
1652 device
->dfsm_allowed
= device
->pbb_allowed
&&
1653 (device
->physical_device
->rad_info
.family
== CHIP_RAVEN
||
1654 device
->physical_device
->rad_info
.family
== CHIP_RAVEN2
);
1657 device
->always_use_syncobj
= device
->physical_device
->rad_info
.has_syncobj_wait_for_submit
;
1660 /* The maximum number of scratch waves. Scratch space isn't divided
1661 * evenly between CUs. The number is only a function of the number of CUs.
1662 * We can decrease the constant to decrease the scratch buffer size.
1664 * sctx->scratch_waves must be >= the maximum possible size of
1665 * 1 threadgroup, so that the hw doesn't hang from being unable
1668 * The recommended value is 4 per CU at most. Higher numbers don't
1669 * bring much benefit, but they still occupy chip resources (think
1670 * async compute). I've seen ~2% performance difference between 4 and 32.
1672 uint32_t max_threads_per_block
= 2048;
1673 device
->scratch_waves
= MAX2(32 * physical_device
->rad_info
.num_good_compute_units
,
1674 max_threads_per_block
/ 64);
1676 device
->dispatch_initiator
= S_00B800_COMPUTE_SHADER_EN(1);
1678 if (device
->physical_device
->rad_info
.chip_class
>= CIK
) {
1679 /* If the KMD allows it (there is a KMD hw register for it),
1680 * allow launching waves out-of-order.
1682 device
->dispatch_initiator
|= S_00B800_ORDER_MODE(1);
1685 radv_device_init_gs_info(device
);
1687 device
->tess_offchip_block_dw_size
=
1688 device
->physical_device
->rad_info
.family
== CHIP_HAWAII
? 4096 : 8192;
1689 device
->has_distributed_tess
=
1690 device
->physical_device
->rad_info
.chip_class
>= VI
&&
1691 device
->physical_device
->rad_info
.max_se
>= 2;
1693 if (getenv("RADV_TRACE_FILE")) {
1694 const char *filename
= getenv("RADV_TRACE_FILE");
1696 keep_shader_info
= true;
1698 if (!radv_init_trace(device
))
1701 fprintf(stderr
, "*****************************************************************************\n");
1702 fprintf(stderr
, "* WARNING: RADV_TRACE_FILE is costly and should only be used for debugging! *\n");
1703 fprintf(stderr
, "*****************************************************************************\n");
1705 fprintf(stderr
, "Trace file will be dumped to %s\n", filename
);
1706 radv_dump_enabled_options(device
, stderr
);
1709 device
->keep_shader_info
= keep_shader_info
;
1711 result
= radv_device_init_meta(device
);
1712 if (result
!= VK_SUCCESS
)
1715 radv_device_init_msaa(device
);
1717 for (int family
= 0; family
< RADV_MAX_QUEUE_FAMILIES
; ++family
) {
1718 device
->empty_cs
[family
] = device
->ws
->cs_create(device
->ws
, family
);
1720 case RADV_QUEUE_GENERAL
:
1721 radeon_emit(device
->empty_cs
[family
], PKT3(PKT3_CONTEXT_CONTROL
, 1, 0));
1722 radeon_emit(device
->empty_cs
[family
], CONTEXT_CONTROL_LOAD_ENABLE(1));
1723 radeon_emit(device
->empty_cs
[family
], CONTEXT_CONTROL_SHADOW_ENABLE(1));
1725 case RADV_QUEUE_COMPUTE
:
1726 radeon_emit(device
->empty_cs
[family
], PKT3(PKT3_NOP
, 0, 0));
1727 radeon_emit(device
->empty_cs
[family
], 0);
1730 device
->ws
->cs_finalize(device
->empty_cs
[family
]);
1733 if (device
->physical_device
->rad_info
.chip_class
>= CIK
)
1734 cik_create_gfx_config(device
);
1736 VkPipelineCacheCreateInfo ci
;
1737 ci
.sType
= VK_STRUCTURE_TYPE_PIPELINE_CACHE_CREATE_INFO
;
1740 ci
.pInitialData
= NULL
;
1741 ci
.initialDataSize
= 0;
1743 result
= radv_CreatePipelineCache(radv_device_to_handle(device
),
1745 if (result
!= VK_SUCCESS
)
1748 device
->mem_cache
= radv_pipeline_cache_from_handle(pc
);
1750 device
->force_aniso
=
1751 MIN2(16, radv_get_int_debug_option("RADV_TEX_ANISO", -1));
1752 if (device
->force_aniso
>= 0) {
1753 fprintf(stderr
, "radv: Forcing anisotropy filter to %ix\n",
1754 1 << util_logbase2(device
->force_aniso
));
1757 *pDevice
= radv_device_to_handle(device
);
1761 radv_device_finish_meta(device
);
1763 radv_bo_list_finish(&device
->bo_list
);
1765 if (device
->trace_bo
)
1766 device
->ws
->buffer_destroy(device
->trace_bo
);
1768 if (device
->gfx_init
)
1769 device
->ws
->buffer_destroy(device
->gfx_init
);
1771 for (unsigned i
= 0; i
< RADV_MAX_QUEUE_FAMILIES
; i
++) {
1772 for (unsigned q
= 0; q
< device
->queue_count
[i
]; q
++)
1773 radv_queue_finish(&device
->queues
[i
][q
]);
1774 if (device
->queue_count
[i
])
1775 vk_free(&device
->alloc
, device
->queues
[i
]);
1778 vk_free(&device
->alloc
, device
);
1782 void radv_DestroyDevice(
1784 const VkAllocationCallbacks
* pAllocator
)
1786 RADV_FROM_HANDLE(radv_device
, device
, _device
);
1791 if (device
->trace_bo
)
1792 device
->ws
->buffer_destroy(device
->trace_bo
);
1794 if (device
->gfx_init
)
1795 device
->ws
->buffer_destroy(device
->gfx_init
);
1797 for (unsigned i
= 0; i
< RADV_MAX_QUEUE_FAMILIES
; i
++) {
1798 for (unsigned q
= 0; q
< device
->queue_count
[i
]; q
++)
1799 radv_queue_finish(&device
->queues
[i
][q
]);
1800 if (device
->queue_count
[i
])
1801 vk_free(&device
->alloc
, device
->queues
[i
]);
1802 if (device
->empty_cs
[i
])
1803 device
->ws
->cs_destroy(device
->empty_cs
[i
]);
1805 radv_device_finish_meta(device
);
1807 VkPipelineCache pc
= radv_pipeline_cache_to_handle(device
->mem_cache
);
1808 radv_DestroyPipelineCache(radv_device_to_handle(device
), pc
, NULL
);
1810 radv_destroy_shader_slabs(device
);
1812 radv_bo_list_finish(&device
->bo_list
);
1813 vk_free(&device
->alloc
, device
);
1816 VkResult
radv_EnumerateInstanceLayerProperties(
1817 uint32_t* pPropertyCount
,
1818 VkLayerProperties
* pProperties
)
1820 if (pProperties
== NULL
) {
1821 *pPropertyCount
= 0;
1825 /* None supported at this time */
1826 return vk_error(NULL
, VK_ERROR_LAYER_NOT_PRESENT
);
1829 VkResult
radv_EnumerateDeviceLayerProperties(
1830 VkPhysicalDevice physicalDevice
,
1831 uint32_t* pPropertyCount
,
1832 VkLayerProperties
* pProperties
)
1834 if (pProperties
== NULL
) {
1835 *pPropertyCount
= 0;
1839 /* None supported at this time */
1840 return vk_error(NULL
, VK_ERROR_LAYER_NOT_PRESENT
);
1843 void radv_GetDeviceQueue2(
1845 const VkDeviceQueueInfo2
* pQueueInfo
,
1848 RADV_FROM_HANDLE(radv_device
, device
, _device
);
1849 struct radv_queue
*queue
;
1851 queue
= &device
->queues
[pQueueInfo
->queueFamilyIndex
][pQueueInfo
->queueIndex
];
1852 if (pQueueInfo
->flags
!= queue
->flags
) {
1853 /* From the Vulkan 1.1.70 spec:
1855 * "The queue returned by vkGetDeviceQueue2 must have the same
1856 * flags value from this structure as that used at device
1857 * creation time in a VkDeviceQueueCreateInfo instance. If no
1858 * matching flags were specified at device creation time then
1859 * pQueue will return VK_NULL_HANDLE."
1861 *pQueue
= VK_NULL_HANDLE
;
1865 *pQueue
= radv_queue_to_handle(queue
);
1868 void radv_GetDeviceQueue(
1870 uint32_t queueFamilyIndex
,
1871 uint32_t queueIndex
,
1874 const VkDeviceQueueInfo2 info
= (VkDeviceQueueInfo2
) {
1875 .sType
= VK_STRUCTURE_TYPE_DEVICE_QUEUE_INFO_2
,
1876 .queueFamilyIndex
= queueFamilyIndex
,
1877 .queueIndex
= queueIndex
1880 radv_GetDeviceQueue2(_device
, &info
, pQueue
);
1884 fill_geom_tess_rings(struct radv_queue
*queue
,
1886 bool add_sample_positions
,
1887 uint32_t esgs_ring_size
,
1888 struct radeon_winsys_bo
*esgs_ring_bo
,
1889 uint32_t gsvs_ring_size
,
1890 struct radeon_winsys_bo
*gsvs_ring_bo
,
1891 uint32_t tess_factor_ring_size
,
1892 uint32_t tess_offchip_ring_offset
,
1893 uint32_t tess_offchip_ring_size
,
1894 struct radeon_winsys_bo
*tess_rings_bo
)
1896 uint64_t esgs_va
= 0, gsvs_va
= 0;
1897 uint64_t tess_va
= 0, tess_offchip_va
= 0;
1898 uint32_t *desc
= &map
[4];
1901 esgs_va
= radv_buffer_get_va(esgs_ring_bo
);
1903 gsvs_va
= radv_buffer_get_va(gsvs_ring_bo
);
1904 if (tess_rings_bo
) {
1905 tess_va
= radv_buffer_get_va(tess_rings_bo
);
1906 tess_offchip_va
= tess_va
+ tess_offchip_ring_offset
;
1909 /* stride 0, num records - size, add tid, swizzle, elsize4,
1912 desc
[1] = S_008F04_BASE_ADDRESS_HI(esgs_va
>> 32) |
1913 S_008F04_STRIDE(0) |
1914 S_008F04_SWIZZLE_ENABLE(true);
1915 desc
[2] = esgs_ring_size
;
1916 desc
[3] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
1917 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
1918 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
1919 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
) |
1920 S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
1921 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
) |
1922 S_008F0C_ELEMENT_SIZE(1) |
1923 S_008F0C_INDEX_STRIDE(3) |
1924 S_008F0C_ADD_TID_ENABLE(true);
1927 /* GS entry for ES->GS ring */
1928 /* stride 0, num records - size, elsize0,
1931 desc
[1] = S_008F04_BASE_ADDRESS_HI(esgs_va
>> 32)|
1932 S_008F04_STRIDE(0) |
1933 S_008F04_SWIZZLE_ENABLE(false);
1934 desc
[2] = esgs_ring_size
;
1935 desc
[3] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
1936 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
1937 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
1938 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
) |
1939 S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
1940 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
) |
1941 S_008F0C_ELEMENT_SIZE(0) |
1942 S_008F0C_INDEX_STRIDE(0) |
1943 S_008F0C_ADD_TID_ENABLE(false);
1946 /* VS entry for GS->VS ring */
1947 /* stride 0, num records - size, elsize0,
1950 desc
[1] = S_008F04_BASE_ADDRESS_HI(gsvs_va
>> 32)|
1951 S_008F04_STRIDE(0) |
1952 S_008F04_SWIZZLE_ENABLE(false);
1953 desc
[2] = gsvs_ring_size
;
1954 desc
[3] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
1955 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
1956 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
1957 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
) |
1958 S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
1959 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
) |
1960 S_008F0C_ELEMENT_SIZE(0) |
1961 S_008F0C_INDEX_STRIDE(0) |
1962 S_008F0C_ADD_TID_ENABLE(false);
1965 /* stride gsvs_itemsize, num records 64
1966 elsize 4, index stride 16 */
1967 /* shader will patch stride and desc[2] */
1969 desc
[1] = S_008F04_BASE_ADDRESS_HI(gsvs_va
>> 32)|
1970 S_008F04_STRIDE(0) |
1971 S_008F04_SWIZZLE_ENABLE(true);
1973 desc
[3] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
1974 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
1975 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
1976 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
) |
1977 S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
1978 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
) |
1979 S_008F0C_ELEMENT_SIZE(1) |
1980 S_008F0C_INDEX_STRIDE(1) |
1981 S_008F0C_ADD_TID_ENABLE(true);
1985 desc
[1] = S_008F04_BASE_ADDRESS_HI(tess_va
>> 32) |
1986 S_008F04_STRIDE(0) |
1987 S_008F04_SWIZZLE_ENABLE(false);
1988 desc
[2] = tess_factor_ring_size
;
1989 desc
[3] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
1990 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
1991 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
1992 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
) |
1993 S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
1994 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
) |
1995 S_008F0C_ELEMENT_SIZE(0) |
1996 S_008F0C_INDEX_STRIDE(0) |
1997 S_008F0C_ADD_TID_ENABLE(false);
2000 desc
[0] = tess_offchip_va
;
2001 desc
[1] = S_008F04_BASE_ADDRESS_HI(tess_offchip_va
>> 32) |
2002 S_008F04_STRIDE(0) |
2003 S_008F04_SWIZZLE_ENABLE(false);
2004 desc
[2] = tess_offchip_ring_size
;
2005 desc
[3] = S_008F0C_DST_SEL_X(V_008F0C_SQ_SEL_X
) |
2006 S_008F0C_DST_SEL_Y(V_008F0C_SQ_SEL_Y
) |
2007 S_008F0C_DST_SEL_Z(V_008F0C_SQ_SEL_Z
) |
2008 S_008F0C_DST_SEL_W(V_008F0C_SQ_SEL_W
) |
2009 S_008F0C_NUM_FORMAT(V_008F0C_BUF_NUM_FORMAT_FLOAT
) |
2010 S_008F0C_DATA_FORMAT(V_008F0C_BUF_DATA_FORMAT_32
) |
2011 S_008F0C_ELEMENT_SIZE(0) |
2012 S_008F0C_INDEX_STRIDE(0) |
2013 S_008F0C_ADD_TID_ENABLE(false);
2016 /* add sample positions after all rings */
2017 memcpy(desc
, queue
->device
->sample_locations_1x
, 8);
2019 memcpy(desc
, queue
->device
->sample_locations_2x
, 16);
2021 memcpy(desc
, queue
->device
->sample_locations_4x
, 32);
2023 memcpy(desc
, queue
->device
->sample_locations_8x
, 64);
2025 memcpy(desc
, queue
->device
->sample_locations_16x
, 128);
2029 radv_get_hs_offchip_param(struct radv_device
*device
, uint32_t *max_offchip_buffers_p
)
2031 bool double_offchip_buffers
= device
->physical_device
->rad_info
.chip_class
>= CIK
&&
2032 device
->physical_device
->rad_info
.family
!= CHIP_CARRIZO
&&
2033 device
->physical_device
->rad_info
.family
!= CHIP_STONEY
;
2034 unsigned max_offchip_buffers_per_se
= double_offchip_buffers
? 128 : 64;
2035 unsigned max_offchip_buffers
;
2036 unsigned offchip_granularity
;
2037 unsigned hs_offchip_param
;
2041 * This must be one less than the maximum number due to a hw limitation.
2042 * Various hardware bugs in SI, CIK, and GFX9 need this.
2045 * Vega10 should limit max_offchip_buffers to 508 (4 * 127).
2046 * Gfx7 should limit max_offchip_buffers to 508
2047 * Gfx6 should limit max_offchip_buffers to 126 (2 * 63)
2049 * Follow AMDVLK here.
2051 if (device
->physical_device
->rad_info
.family
== CHIP_VEGA10
||
2052 device
->physical_device
->rad_info
.chip_class
== CIK
||
2053 device
->physical_device
->rad_info
.chip_class
== SI
)
2054 --max_offchip_buffers_per_se
;
2056 max_offchip_buffers
= max_offchip_buffers_per_se
*
2057 device
->physical_device
->rad_info
.max_se
;
2059 /* Hawaii has a bug with offchip buffers > 256 that can be worked
2060 * around by setting 4K granularity.
2062 if (device
->tess_offchip_block_dw_size
== 4096) {
2063 assert(device
->physical_device
->rad_info
.family
== CHIP_HAWAII
);
2064 offchip_granularity
= V_03093C_X_4K_DWORDS
;
2066 assert(device
->tess_offchip_block_dw_size
== 8192);
2067 offchip_granularity
= V_03093C_X_8K_DWORDS
;
2070 switch (device
->physical_device
->rad_info
.chip_class
) {
2072 max_offchip_buffers
= MIN2(max_offchip_buffers
, 126);
2078 max_offchip_buffers
= MIN2(max_offchip_buffers
, 508);
2082 *max_offchip_buffers_p
= max_offchip_buffers
;
2083 if (device
->physical_device
->rad_info
.chip_class
>= CIK
) {
2084 if (device
->physical_device
->rad_info
.chip_class
>= VI
)
2085 --max_offchip_buffers
;
2087 S_03093C_OFFCHIP_BUFFERING(max_offchip_buffers
) |
2088 S_03093C_OFFCHIP_GRANULARITY(offchip_granularity
);
2091 S_0089B0_OFFCHIP_BUFFERING(max_offchip_buffers
);
2093 return hs_offchip_param
;
2097 radv_emit_gs_ring_sizes(struct radv_queue
*queue
, struct radeon_cmdbuf
*cs
,
2098 struct radeon_winsys_bo
*esgs_ring_bo
,
2099 uint32_t esgs_ring_size
,
2100 struct radeon_winsys_bo
*gsvs_ring_bo
,
2101 uint32_t gsvs_ring_size
)
2103 if (!esgs_ring_bo
&& !gsvs_ring_bo
)
2107 radv_cs_add_buffer(queue
->device
->ws
, cs
, esgs_ring_bo
);
2110 radv_cs_add_buffer(queue
->device
->ws
, cs
, gsvs_ring_bo
);
2112 if (queue
->device
->physical_device
->rad_info
.chip_class
>= CIK
) {
2113 radeon_set_uconfig_reg_seq(cs
, R_030900_VGT_ESGS_RING_SIZE
, 2);
2114 radeon_emit(cs
, esgs_ring_size
>> 8);
2115 radeon_emit(cs
, gsvs_ring_size
>> 8);
2117 radeon_set_config_reg_seq(cs
, R_0088C8_VGT_ESGS_RING_SIZE
, 2);
2118 radeon_emit(cs
, esgs_ring_size
>> 8);
2119 radeon_emit(cs
, gsvs_ring_size
>> 8);
2124 radv_emit_tess_factor_ring(struct radv_queue
*queue
, struct radeon_cmdbuf
*cs
,
2125 unsigned hs_offchip_param
, unsigned tf_ring_size
,
2126 struct radeon_winsys_bo
*tess_rings_bo
)
2133 tf_va
= radv_buffer_get_va(tess_rings_bo
);
2135 radv_cs_add_buffer(queue
->device
->ws
, cs
, tess_rings_bo
);
2137 if (queue
->device
->physical_device
->rad_info
.chip_class
>= CIK
) {
2138 radeon_set_uconfig_reg(cs
, R_030938_VGT_TF_RING_SIZE
,
2139 S_030938_SIZE(tf_ring_size
/ 4));
2140 radeon_set_uconfig_reg(cs
, R_030940_VGT_TF_MEMORY_BASE
,
2142 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX9
) {
2143 radeon_set_uconfig_reg(cs
, R_030944_VGT_TF_MEMORY_BASE_HI
,
2144 S_030944_BASE_HI(tf_va
>> 40));
2146 radeon_set_uconfig_reg(cs
, R_03093C_VGT_HS_OFFCHIP_PARAM
,
2149 radeon_set_config_reg(cs
, R_008988_VGT_TF_RING_SIZE
,
2150 S_008988_SIZE(tf_ring_size
/ 4));
2151 radeon_set_config_reg(cs
, R_0089B8_VGT_TF_MEMORY_BASE
,
2153 radeon_set_config_reg(cs
, R_0089B0_VGT_HS_OFFCHIP_PARAM
,
2159 radv_emit_compute_scratch(struct radv_queue
*queue
, struct radeon_cmdbuf
*cs
,
2160 struct radeon_winsys_bo
*compute_scratch_bo
)
2162 uint64_t scratch_va
;
2164 if (!compute_scratch_bo
)
2167 scratch_va
= radv_buffer_get_va(compute_scratch_bo
);
2169 radv_cs_add_buffer(queue
->device
->ws
, cs
, compute_scratch_bo
);
2171 radeon_set_sh_reg_seq(cs
, R_00B900_COMPUTE_USER_DATA_0
, 2);
2172 radeon_emit(cs
, scratch_va
);
2173 radeon_emit(cs
, S_008F04_BASE_ADDRESS_HI(scratch_va
>> 32) |
2174 S_008F04_SWIZZLE_ENABLE(1));
2178 radv_emit_global_shader_pointers(struct radv_queue
*queue
,
2179 struct radeon_cmdbuf
*cs
,
2180 struct radeon_winsys_bo
*descriptor_bo
)
2187 va
= radv_buffer_get_va(descriptor_bo
);
2189 radv_cs_add_buffer(queue
->device
->ws
, cs
, descriptor_bo
);
2191 if (queue
->device
->physical_device
->rad_info
.chip_class
>= GFX9
) {
2192 uint32_t regs
[] = {R_00B030_SPI_SHADER_USER_DATA_PS_0
,
2193 R_00B130_SPI_SHADER_USER_DATA_VS_0
,
2194 R_00B208_SPI_SHADER_USER_DATA_ADDR_LO_GS
,
2195 R_00B408_SPI_SHADER_USER_DATA_ADDR_LO_HS
};
2197 for (int i
= 0; i
< ARRAY_SIZE(regs
); ++i
) {
2198 radv_emit_shader_pointer(queue
->device
, cs
, regs
[i
],
2202 uint32_t regs
[] = {R_00B030_SPI_SHADER_USER_DATA_PS_0
,
2203 R_00B130_SPI_SHADER_USER_DATA_VS_0
,
2204 R_00B230_SPI_SHADER_USER_DATA_GS_0
,
2205 R_00B330_SPI_SHADER_USER_DATA_ES_0
,
2206 R_00B430_SPI_SHADER_USER_DATA_HS_0
,
2207 R_00B530_SPI_SHADER_USER_DATA_LS_0
};
2209 for (int i
= 0; i
< ARRAY_SIZE(regs
); ++i
) {
2210 radv_emit_shader_pointer(queue
->device
, cs
, regs
[i
],
2217 radv_init_graphics_state(struct radeon_cmdbuf
*cs
, struct radv_queue
*queue
)
2219 struct radv_device
*device
= queue
->device
;
2221 if (device
->gfx_init
) {
2222 uint64_t va
= radv_buffer_get_va(device
->gfx_init
);
2224 radeon_emit(cs
, PKT3(PKT3_INDIRECT_BUFFER_CIK
, 2, 0));
2225 radeon_emit(cs
, va
);
2226 radeon_emit(cs
, va
>> 32);
2227 radeon_emit(cs
, device
->gfx_init_size_dw
& 0xffff);
2229 radv_cs_add_buffer(device
->ws
, cs
, device
->gfx_init
);
2231 struct radv_physical_device
*physical_device
= device
->physical_device
;
2232 si_emit_graphics(physical_device
, cs
);
2237 radv_init_compute_state(struct radeon_cmdbuf
*cs
, struct radv_queue
*queue
)
2239 struct radv_physical_device
*physical_device
= queue
->device
->physical_device
;
2240 si_emit_compute(physical_device
, cs
);
2244 radv_get_preamble_cs(struct radv_queue
*queue
,
2245 uint32_t scratch_size
,
2246 uint32_t compute_scratch_size
,
2247 uint32_t esgs_ring_size
,
2248 uint32_t gsvs_ring_size
,
2249 bool needs_tess_rings
,
2250 bool needs_sample_positions
,
2251 struct radeon_cmdbuf
**initial_full_flush_preamble_cs
,
2252 struct radeon_cmdbuf
**initial_preamble_cs
,
2253 struct radeon_cmdbuf
**continue_preamble_cs
)
2255 struct radeon_winsys_bo
*scratch_bo
= NULL
;
2256 struct radeon_winsys_bo
*descriptor_bo
= NULL
;
2257 struct radeon_winsys_bo
*compute_scratch_bo
= NULL
;
2258 struct radeon_winsys_bo
*esgs_ring_bo
= NULL
;
2259 struct radeon_winsys_bo
*gsvs_ring_bo
= NULL
;
2260 struct radeon_winsys_bo
*tess_rings_bo
= NULL
;
2261 struct radeon_cmdbuf
*dest_cs
[3] = {0};
2262 bool add_tess_rings
= false, add_sample_positions
= false;
2263 unsigned tess_factor_ring_size
= 0, tess_offchip_ring_size
= 0;
2264 unsigned max_offchip_buffers
;
2265 unsigned hs_offchip_param
= 0;
2266 unsigned tess_offchip_ring_offset
;
2267 uint32_t ring_bo_flags
= RADEON_FLAG_NO_CPU_ACCESS
| RADEON_FLAG_NO_INTERPROCESS_SHARING
;
2268 if (!queue
->has_tess_rings
) {
2269 if (needs_tess_rings
)
2270 add_tess_rings
= true;
2272 if (!queue
->has_sample_positions
) {
2273 if (needs_sample_positions
)
2274 add_sample_positions
= true;
2276 tess_factor_ring_size
= 32768 * queue
->device
->physical_device
->rad_info
.max_se
;
2277 hs_offchip_param
= radv_get_hs_offchip_param(queue
->device
,
2278 &max_offchip_buffers
);
2279 tess_offchip_ring_offset
= align(tess_factor_ring_size
, 64 * 1024);
2280 tess_offchip_ring_size
= max_offchip_buffers
*
2281 queue
->device
->tess_offchip_block_dw_size
* 4;
2283 if (scratch_size
<= queue
->scratch_size
&&
2284 compute_scratch_size
<= queue
->compute_scratch_size
&&
2285 esgs_ring_size
<= queue
->esgs_ring_size
&&
2286 gsvs_ring_size
<= queue
->gsvs_ring_size
&&
2287 !add_tess_rings
&& !add_sample_positions
&&
2288 queue
->initial_preamble_cs
) {
2289 *initial_full_flush_preamble_cs
= queue
->initial_full_flush_preamble_cs
;
2290 *initial_preamble_cs
= queue
->initial_preamble_cs
;
2291 *continue_preamble_cs
= queue
->continue_preamble_cs
;
2292 if (!scratch_size
&& !compute_scratch_size
&& !esgs_ring_size
&& !gsvs_ring_size
)
2293 *continue_preamble_cs
= NULL
;
2297 if (scratch_size
> queue
->scratch_size
) {
2298 scratch_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2306 scratch_bo
= queue
->scratch_bo
;
2308 if (compute_scratch_size
> queue
->compute_scratch_size
) {
2309 compute_scratch_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2310 compute_scratch_size
,
2314 if (!compute_scratch_bo
)
2318 compute_scratch_bo
= queue
->compute_scratch_bo
;
2320 if (esgs_ring_size
> queue
->esgs_ring_size
) {
2321 esgs_ring_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2329 esgs_ring_bo
= queue
->esgs_ring_bo
;
2330 esgs_ring_size
= queue
->esgs_ring_size
;
2333 if (gsvs_ring_size
> queue
->gsvs_ring_size
) {
2334 gsvs_ring_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2342 gsvs_ring_bo
= queue
->gsvs_ring_bo
;
2343 gsvs_ring_size
= queue
->gsvs_ring_size
;
2346 if (add_tess_rings
) {
2347 tess_rings_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2348 tess_offchip_ring_offset
+ tess_offchip_ring_size
,
2355 tess_rings_bo
= queue
->tess_rings_bo
;
2358 if (scratch_bo
!= queue
->scratch_bo
||
2359 esgs_ring_bo
!= queue
->esgs_ring_bo
||
2360 gsvs_ring_bo
!= queue
->gsvs_ring_bo
||
2361 tess_rings_bo
!= queue
->tess_rings_bo
||
2362 add_sample_positions
) {
2364 if (gsvs_ring_bo
|| esgs_ring_bo
||
2365 tess_rings_bo
|| add_sample_positions
) {
2366 size
= 112; /* 2 dword + 2 padding + 4 dword * 6 */
2367 if (add_sample_positions
)
2368 size
+= 256; /* 32+16+8+4+2+1 samples * 4 * 2 = 248 bytes. */
2370 else if (scratch_bo
)
2371 size
= 8; /* 2 dword */
2373 descriptor_bo
= queue
->device
->ws
->buffer_create(queue
->device
->ws
,
2377 RADEON_FLAG_CPU_ACCESS
|
2378 RADEON_FLAG_NO_INTERPROCESS_SHARING
|
2379 RADEON_FLAG_READ_ONLY
);
2383 descriptor_bo
= queue
->descriptor_bo
;
2385 for(int i
= 0; i
< 3; ++i
) {
2386 struct radeon_cmdbuf
*cs
= NULL
;
2387 cs
= queue
->device
->ws
->cs_create(queue
->device
->ws
,
2388 queue
->queue_family_index
? RING_COMPUTE
: RING_GFX
);
2395 radv_cs_add_buffer(queue
->device
->ws
, cs
, scratch_bo
);
2397 /* Emit initial configuration. */
2398 switch (queue
->queue_family_index
) {
2399 case RADV_QUEUE_GENERAL
:
2400 radv_init_graphics_state(cs
, queue
);
2402 case RADV_QUEUE_COMPUTE
:
2403 radv_init_compute_state(cs
, queue
);
2405 case RADV_QUEUE_TRANSFER
:
2409 if (descriptor_bo
!= queue
->descriptor_bo
) {
2410 uint32_t *map
= (uint32_t*)queue
->device
->ws
->buffer_map(descriptor_bo
);
2413 uint64_t scratch_va
= radv_buffer_get_va(scratch_bo
);
2414 uint32_t rsrc1
= S_008F04_BASE_ADDRESS_HI(scratch_va
>> 32) |
2415 S_008F04_SWIZZLE_ENABLE(1);
2416 map
[0] = scratch_va
;
2420 if (esgs_ring_bo
|| gsvs_ring_bo
|| tess_rings_bo
||
2421 add_sample_positions
)
2422 fill_geom_tess_rings(queue
, map
, add_sample_positions
,
2423 esgs_ring_size
, esgs_ring_bo
,
2424 gsvs_ring_size
, gsvs_ring_bo
,
2425 tess_factor_ring_size
,
2426 tess_offchip_ring_offset
,
2427 tess_offchip_ring_size
,
2430 queue
->device
->ws
->buffer_unmap(descriptor_bo
);
2433 if (esgs_ring_bo
|| gsvs_ring_bo
|| tess_rings_bo
) {
2434 radeon_emit(cs
, PKT3(PKT3_EVENT_WRITE
, 0, 0));
2435 radeon_emit(cs
, EVENT_TYPE(V_028A90_VS_PARTIAL_FLUSH
) | EVENT_INDEX(4));
2436 radeon_emit(cs
, PKT3(PKT3_EVENT_WRITE
, 0, 0));
2437 radeon_emit(cs
, EVENT_TYPE(V_028A90_VGT_FLUSH
) | EVENT_INDEX(0));
2440 radv_emit_gs_ring_sizes(queue
, cs
, esgs_ring_bo
, esgs_ring_size
,
2441 gsvs_ring_bo
, gsvs_ring_size
);
2442 radv_emit_tess_factor_ring(queue
, cs
, hs_offchip_param
,
2443 tess_factor_ring_size
, tess_rings_bo
);
2444 radv_emit_global_shader_pointers(queue
, cs
, descriptor_bo
);
2445 radv_emit_compute_scratch(queue
, cs
, compute_scratch_bo
);
2448 si_cs_emit_cache_flush(cs
,
2449 queue
->device
->physical_device
->rad_info
.chip_class
,
2451 queue
->queue_family_index
== RING_COMPUTE
&&
2452 queue
->device
->physical_device
->rad_info
.chip_class
>= CIK
,
2453 (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
)) |
2454 RADV_CMD_FLAG_INV_ICACHE
|
2455 RADV_CMD_FLAG_INV_SMEM_L1
|
2456 RADV_CMD_FLAG_INV_VMEM_L1
|
2457 RADV_CMD_FLAG_INV_GLOBAL_L2
|
2458 RADV_CMD_FLAG_START_PIPELINE_STATS
, 0);
2459 } else if (i
== 1) {
2460 si_cs_emit_cache_flush(cs
,
2461 queue
->device
->physical_device
->rad_info
.chip_class
,
2463 queue
->queue_family_index
== RING_COMPUTE
&&
2464 queue
->device
->physical_device
->rad_info
.chip_class
>= CIK
,
2465 RADV_CMD_FLAG_INV_ICACHE
|
2466 RADV_CMD_FLAG_INV_SMEM_L1
|
2467 RADV_CMD_FLAG_INV_VMEM_L1
|
2468 RADV_CMD_FLAG_INV_GLOBAL_L2
|
2469 RADV_CMD_FLAG_START_PIPELINE_STATS
, 0);
2472 if (!queue
->device
->ws
->cs_finalize(cs
))
2476 if (queue
->initial_full_flush_preamble_cs
)
2477 queue
->device
->ws
->cs_destroy(queue
->initial_full_flush_preamble_cs
);
2479 if (queue
->initial_preamble_cs
)
2480 queue
->device
->ws
->cs_destroy(queue
->initial_preamble_cs
);
2482 if (queue
->continue_preamble_cs
)
2483 queue
->device
->ws
->cs_destroy(queue
->continue_preamble_cs
);
2485 queue
->initial_full_flush_preamble_cs
= dest_cs
[0];
2486 queue
->initial_preamble_cs
= dest_cs
[1];
2487 queue
->continue_preamble_cs
= dest_cs
[2];
2489 if (scratch_bo
!= queue
->scratch_bo
) {
2490 if (queue
->scratch_bo
)
2491 queue
->device
->ws
->buffer_destroy(queue
->scratch_bo
);
2492 queue
->scratch_bo
= scratch_bo
;
2493 queue
->scratch_size
= scratch_size
;
2496 if (compute_scratch_bo
!= queue
->compute_scratch_bo
) {
2497 if (queue
->compute_scratch_bo
)
2498 queue
->device
->ws
->buffer_destroy(queue
->compute_scratch_bo
);
2499 queue
->compute_scratch_bo
= compute_scratch_bo
;
2500 queue
->compute_scratch_size
= compute_scratch_size
;
2503 if (esgs_ring_bo
!= queue
->esgs_ring_bo
) {
2504 if (queue
->esgs_ring_bo
)
2505 queue
->device
->ws
->buffer_destroy(queue
->esgs_ring_bo
);
2506 queue
->esgs_ring_bo
= esgs_ring_bo
;
2507 queue
->esgs_ring_size
= esgs_ring_size
;
2510 if (gsvs_ring_bo
!= queue
->gsvs_ring_bo
) {
2511 if (queue
->gsvs_ring_bo
)
2512 queue
->device
->ws
->buffer_destroy(queue
->gsvs_ring_bo
);
2513 queue
->gsvs_ring_bo
= gsvs_ring_bo
;
2514 queue
->gsvs_ring_size
= gsvs_ring_size
;
2517 if (tess_rings_bo
!= queue
->tess_rings_bo
) {
2518 queue
->tess_rings_bo
= tess_rings_bo
;
2519 queue
->has_tess_rings
= true;
2522 if (descriptor_bo
!= queue
->descriptor_bo
) {
2523 if (queue
->descriptor_bo
)
2524 queue
->device
->ws
->buffer_destroy(queue
->descriptor_bo
);
2526 queue
->descriptor_bo
= descriptor_bo
;
2529 if (add_sample_positions
)
2530 queue
->has_sample_positions
= true;
2532 *initial_full_flush_preamble_cs
= queue
->initial_full_flush_preamble_cs
;
2533 *initial_preamble_cs
= queue
->initial_preamble_cs
;
2534 *continue_preamble_cs
= queue
->continue_preamble_cs
;
2535 if (!scratch_size
&& !compute_scratch_size
&& !esgs_ring_size
&& !gsvs_ring_size
)
2536 *continue_preamble_cs
= NULL
;
2539 for (int i
= 0; i
< ARRAY_SIZE(dest_cs
); ++i
)
2541 queue
->device
->ws
->cs_destroy(dest_cs
[i
]);
2542 if (descriptor_bo
&& descriptor_bo
!= queue
->descriptor_bo
)
2543 queue
->device
->ws
->buffer_destroy(descriptor_bo
);
2544 if (scratch_bo
&& scratch_bo
!= queue
->scratch_bo
)
2545 queue
->device
->ws
->buffer_destroy(scratch_bo
);
2546 if (compute_scratch_bo
&& compute_scratch_bo
!= queue
->compute_scratch_bo
)
2547 queue
->device
->ws
->buffer_destroy(compute_scratch_bo
);
2548 if (esgs_ring_bo
&& esgs_ring_bo
!= queue
->esgs_ring_bo
)
2549 queue
->device
->ws
->buffer_destroy(esgs_ring_bo
);
2550 if (gsvs_ring_bo
&& gsvs_ring_bo
!= queue
->gsvs_ring_bo
)
2551 queue
->device
->ws
->buffer_destroy(gsvs_ring_bo
);
2552 if (tess_rings_bo
&& tess_rings_bo
!= queue
->tess_rings_bo
)
2553 queue
->device
->ws
->buffer_destroy(tess_rings_bo
);
2554 return vk_error(queue
->device
->instance
, VK_ERROR_OUT_OF_DEVICE_MEMORY
);
2557 static VkResult
radv_alloc_sem_counts(struct radv_instance
*instance
,
2558 struct radv_winsys_sem_counts
*counts
,
2560 const VkSemaphore
*sems
,
2564 int syncobj_idx
= 0, sem_idx
= 0;
2566 if (num_sems
== 0 && _fence
== VK_NULL_HANDLE
)
2569 for (uint32_t i
= 0; i
< num_sems
; i
++) {
2570 RADV_FROM_HANDLE(radv_semaphore
, sem
, sems
[i
]);
2572 if (sem
->temp_syncobj
|| sem
->syncobj
)
2573 counts
->syncobj_count
++;
2575 counts
->sem_count
++;
2578 if (_fence
!= VK_NULL_HANDLE
) {
2579 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
2580 if (fence
->temp_syncobj
|| fence
->syncobj
)
2581 counts
->syncobj_count
++;
2584 if (counts
->syncobj_count
) {
2585 counts
->syncobj
= (uint32_t *)malloc(sizeof(uint32_t) * counts
->syncobj_count
);
2586 if (!counts
->syncobj
)
2587 return vk_error(instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
2590 if (counts
->sem_count
) {
2591 counts
->sem
= (struct radeon_winsys_sem
**)malloc(sizeof(struct radeon_winsys_sem
*) * counts
->sem_count
);
2593 free(counts
->syncobj
);
2594 return vk_error(instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
2598 for (uint32_t i
= 0; i
< num_sems
; i
++) {
2599 RADV_FROM_HANDLE(radv_semaphore
, sem
, sems
[i
]);
2601 if (sem
->temp_syncobj
) {
2602 counts
->syncobj
[syncobj_idx
++] = sem
->temp_syncobj
;
2604 else if (sem
->syncobj
)
2605 counts
->syncobj
[syncobj_idx
++] = sem
->syncobj
;
2608 counts
->sem
[sem_idx
++] = sem
->sem
;
2612 if (_fence
!= VK_NULL_HANDLE
) {
2613 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
2614 if (fence
->temp_syncobj
)
2615 counts
->syncobj
[syncobj_idx
++] = fence
->temp_syncobj
;
2616 else if (fence
->syncobj
)
2617 counts
->syncobj
[syncobj_idx
++] = fence
->syncobj
;
2624 radv_free_sem_info(struct radv_winsys_sem_info
*sem_info
)
2626 free(sem_info
->wait
.syncobj
);
2627 free(sem_info
->wait
.sem
);
2628 free(sem_info
->signal
.syncobj
);
2629 free(sem_info
->signal
.sem
);
2633 static void radv_free_temp_syncobjs(struct radv_device
*device
,
2635 const VkSemaphore
*sems
)
2637 for (uint32_t i
= 0; i
< num_sems
; i
++) {
2638 RADV_FROM_HANDLE(radv_semaphore
, sem
, sems
[i
]);
2640 if (sem
->temp_syncobj
) {
2641 device
->ws
->destroy_syncobj(device
->ws
, sem
->temp_syncobj
);
2642 sem
->temp_syncobj
= 0;
2648 radv_alloc_sem_info(struct radv_instance
*instance
,
2649 struct radv_winsys_sem_info
*sem_info
,
2651 const VkSemaphore
*wait_sems
,
2652 int num_signal_sems
,
2653 const VkSemaphore
*signal_sems
,
2657 memset(sem_info
, 0, sizeof(*sem_info
));
2659 ret
= radv_alloc_sem_counts(instance
, &sem_info
->wait
, num_wait_sems
, wait_sems
, VK_NULL_HANDLE
, true);
2662 ret
= radv_alloc_sem_counts(instance
, &sem_info
->signal
, num_signal_sems
, signal_sems
, fence
, false);
2664 radv_free_sem_info(sem_info
);
2666 /* caller can override these */
2667 sem_info
->cs_emit_wait
= true;
2668 sem_info
->cs_emit_signal
= true;
2672 /* Signals fence as soon as all the work currently put on queue is done. */
2673 static VkResult
radv_signal_fence(struct radv_queue
*queue
,
2674 struct radv_fence
*fence
)
2678 struct radv_winsys_sem_info sem_info
;
2680 result
= radv_alloc_sem_info(queue
->device
->instance
, &sem_info
, 0, NULL
, 0, NULL
,
2681 radv_fence_to_handle(fence
));
2682 if (result
!= VK_SUCCESS
)
2685 ret
= queue
->device
->ws
->cs_submit(queue
->hw_ctx
, queue
->queue_idx
,
2686 &queue
->device
->empty_cs
[queue
->queue_family_index
],
2687 1, NULL
, NULL
, &sem_info
, NULL
,
2688 false, fence
->fence
);
2689 radv_free_sem_info(&sem_info
);
2692 return vk_error(queue
->device
->instance
, VK_ERROR_DEVICE_LOST
);
2697 VkResult
radv_QueueSubmit(
2699 uint32_t submitCount
,
2700 const VkSubmitInfo
* pSubmits
,
2703 RADV_FROM_HANDLE(radv_queue
, queue
, _queue
);
2704 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
2705 struct radeon_winsys_fence
*base_fence
= fence
? fence
->fence
: NULL
;
2706 struct radeon_winsys_ctx
*ctx
= queue
->hw_ctx
;
2708 uint32_t max_cs_submission
= queue
->device
->trace_bo
? 1 : UINT32_MAX
;
2709 uint32_t scratch_size
= 0;
2710 uint32_t compute_scratch_size
= 0;
2711 uint32_t esgs_ring_size
= 0, gsvs_ring_size
= 0;
2712 struct radeon_cmdbuf
*initial_preamble_cs
= NULL
, *initial_flush_preamble_cs
= NULL
, *continue_preamble_cs
= NULL
;
2714 bool fence_emitted
= false;
2715 bool tess_rings_needed
= false;
2716 bool sample_positions_needed
= false;
2718 /* Do this first so failing to allocate scratch buffers can't result in
2719 * partially executed submissions. */
2720 for (uint32_t i
= 0; i
< submitCount
; i
++) {
2721 for (uint32_t j
= 0; j
< pSubmits
[i
].commandBufferCount
; j
++) {
2722 RADV_FROM_HANDLE(radv_cmd_buffer
, cmd_buffer
,
2723 pSubmits
[i
].pCommandBuffers
[j
]);
2725 scratch_size
= MAX2(scratch_size
, cmd_buffer
->scratch_size_needed
);
2726 compute_scratch_size
= MAX2(compute_scratch_size
,
2727 cmd_buffer
->compute_scratch_size_needed
);
2728 esgs_ring_size
= MAX2(esgs_ring_size
, cmd_buffer
->esgs_ring_size_needed
);
2729 gsvs_ring_size
= MAX2(gsvs_ring_size
, cmd_buffer
->gsvs_ring_size_needed
);
2730 tess_rings_needed
|= cmd_buffer
->tess_rings_needed
;
2731 sample_positions_needed
|= cmd_buffer
->sample_positions_needed
;
2735 result
= radv_get_preamble_cs(queue
, scratch_size
, compute_scratch_size
,
2736 esgs_ring_size
, gsvs_ring_size
, tess_rings_needed
,
2737 sample_positions_needed
, &initial_flush_preamble_cs
,
2738 &initial_preamble_cs
, &continue_preamble_cs
);
2739 if (result
!= VK_SUCCESS
)
2742 for (uint32_t i
= 0; i
< submitCount
; i
++) {
2743 struct radeon_cmdbuf
**cs_array
;
2744 bool do_flush
= !i
|| pSubmits
[i
].pWaitDstStageMask
;
2745 bool can_patch
= true;
2747 struct radv_winsys_sem_info sem_info
;
2749 result
= radv_alloc_sem_info(queue
->device
->instance
,
2751 pSubmits
[i
].waitSemaphoreCount
,
2752 pSubmits
[i
].pWaitSemaphores
,
2753 pSubmits
[i
].signalSemaphoreCount
,
2754 pSubmits
[i
].pSignalSemaphores
,
2756 if (result
!= VK_SUCCESS
)
2759 if (!pSubmits
[i
].commandBufferCount
) {
2760 if (pSubmits
[i
].waitSemaphoreCount
|| pSubmits
[i
].signalSemaphoreCount
) {
2761 ret
= queue
->device
->ws
->cs_submit(ctx
, queue
->queue_idx
,
2762 &queue
->device
->empty_cs
[queue
->queue_family_index
],
2767 radv_loge("failed to submit CS %d\n", i
);
2770 fence_emitted
= true;
2772 radv_free_sem_info(&sem_info
);
2776 cs_array
= malloc(sizeof(struct radeon_cmdbuf
*) *
2777 (pSubmits
[i
].commandBufferCount
));
2779 for (uint32_t j
= 0; j
< pSubmits
[i
].commandBufferCount
; j
++) {
2780 RADV_FROM_HANDLE(radv_cmd_buffer
, cmd_buffer
,
2781 pSubmits
[i
].pCommandBuffers
[j
]);
2782 assert(cmd_buffer
->level
== VK_COMMAND_BUFFER_LEVEL_PRIMARY
);
2784 cs_array
[j
] = cmd_buffer
->cs
;
2785 if ((cmd_buffer
->usage_flags
& VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT
))
2788 cmd_buffer
->status
= RADV_CMD_BUFFER_STATUS_PENDING
;
2791 for (uint32_t j
= 0; j
< pSubmits
[i
].commandBufferCount
; j
+= advance
) {
2792 struct radeon_cmdbuf
*initial_preamble
= (do_flush
&& !j
) ? initial_flush_preamble_cs
: initial_preamble_cs
;
2793 const struct radv_winsys_bo_list
*bo_list
= NULL
;
2795 advance
= MIN2(max_cs_submission
,
2796 pSubmits
[i
].commandBufferCount
- j
);
2798 if (queue
->device
->trace_bo
)
2799 *queue
->device
->trace_id_ptr
= 0;
2801 sem_info
.cs_emit_wait
= j
== 0;
2802 sem_info
.cs_emit_signal
= j
+ advance
== pSubmits
[i
].commandBufferCount
;
2804 if (unlikely(queue
->device
->use_global_bo_list
)) {
2805 pthread_mutex_lock(&queue
->device
->bo_list
.mutex
);
2806 bo_list
= &queue
->device
->bo_list
.list
;
2809 ret
= queue
->device
->ws
->cs_submit(ctx
, queue
->queue_idx
, cs_array
+ j
,
2810 advance
, initial_preamble
, continue_preamble_cs
,
2812 can_patch
, base_fence
);
2814 if (unlikely(queue
->device
->use_global_bo_list
))
2815 pthread_mutex_unlock(&queue
->device
->bo_list
.mutex
);
2818 radv_loge("failed to submit CS %d\n", i
);
2821 fence_emitted
= true;
2822 if (queue
->device
->trace_bo
) {
2823 radv_check_gpu_hangs(queue
, cs_array
[j
]);
2827 radv_free_temp_syncobjs(queue
->device
,
2828 pSubmits
[i
].waitSemaphoreCount
,
2829 pSubmits
[i
].pWaitSemaphores
);
2830 radv_free_sem_info(&sem_info
);
2835 if (!fence_emitted
) {
2836 result
= radv_signal_fence(queue
, fence
);
2837 if (result
!= VK_SUCCESS
)
2840 fence
->submitted
= true;
2846 VkResult
radv_QueueWaitIdle(
2849 RADV_FROM_HANDLE(radv_queue
, queue
, _queue
);
2851 queue
->device
->ws
->ctx_wait_idle(queue
->hw_ctx
,
2852 radv_queue_family_to_ring(queue
->queue_family_index
),
2857 VkResult
radv_DeviceWaitIdle(
2860 RADV_FROM_HANDLE(radv_device
, device
, _device
);
2862 for (unsigned i
= 0; i
< RADV_MAX_QUEUE_FAMILIES
; i
++) {
2863 for (unsigned q
= 0; q
< device
->queue_count
[i
]; q
++) {
2864 radv_QueueWaitIdle(radv_queue_to_handle(&device
->queues
[i
][q
]));
2870 VkResult
radv_EnumerateInstanceExtensionProperties(
2871 const char* pLayerName
,
2872 uint32_t* pPropertyCount
,
2873 VkExtensionProperties
* pProperties
)
2875 VK_OUTARRAY_MAKE(out
, pProperties
, pPropertyCount
);
2877 for (int i
= 0; i
< RADV_INSTANCE_EXTENSION_COUNT
; i
++) {
2878 if (radv_supported_instance_extensions
.extensions
[i
]) {
2879 vk_outarray_append(&out
, prop
) {
2880 *prop
= radv_instance_extensions
[i
];
2885 return vk_outarray_status(&out
);
2888 VkResult
radv_EnumerateDeviceExtensionProperties(
2889 VkPhysicalDevice physicalDevice
,
2890 const char* pLayerName
,
2891 uint32_t* pPropertyCount
,
2892 VkExtensionProperties
* pProperties
)
2894 RADV_FROM_HANDLE(radv_physical_device
, device
, physicalDevice
);
2895 VK_OUTARRAY_MAKE(out
, pProperties
, pPropertyCount
);
2897 for (int i
= 0; i
< RADV_DEVICE_EXTENSION_COUNT
; i
++) {
2898 if (device
->supported_extensions
.extensions
[i
]) {
2899 vk_outarray_append(&out
, prop
) {
2900 *prop
= radv_device_extensions
[i
];
2905 return vk_outarray_status(&out
);
2908 PFN_vkVoidFunction
radv_GetInstanceProcAddr(
2909 VkInstance _instance
,
2912 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
2914 return radv_lookup_entrypoint_checked(pName
,
2915 instance
? instance
->apiVersion
: 0,
2916 instance
? &instance
->enabled_extensions
: NULL
,
2920 /* The loader wants us to expose a second GetInstanceProcAddr function
2921 * to work around certain LD_PRELOAD issues seen in apps.
2924 VKAPI_ATTR PFN_vkVoidFunction VKAPI_CALL
vk_icdGetInstanceProcAddr(
2925 VkInstance instance
,
2929 VKAPI_ATTR PFN_vkVoidFunction VKAPI_CALL
vk_icdGetInstanceProcAddr(
2930 VkInstance instance
,
2933 return radv_GetInstanceProcAddr(instance
, pName
);
2936 PFN_vkVoidFunction
radv_GetDeviceProcAddr(
2940 RADV_FROM_HANDLE(radv_device
, device
, _device
);
2942 return radv_lookup_entrypoint_checked(pName
,
2943 device
->instance
->apiVersion
,
2944 &device
->instance
->enabled_extensions
,
2945 &device
->enabled_extensions
);
2948 bool radv_get_memory_fd(struct radv_device
*device
,
2949 struct radv_device_memory
*memory
,
2952 struct radeon_bo_metadata metadata
;
2954 if (memory
->image
) {
2955 radv_init_metadata(device
, memory
->image
, &metadata
);
2956 device
->ws
->buffer_set_metadata(memory
->bo
, &metadata
);
2959 return device
->ws
->buffer_get_fd(device
->ws
, memory
->bo
,
2963 static VkResult
radv_alloc_memory(struct radv_device
*device
,
2964 const VkMemoryAllocateInfo
* pAllocateInfo
,
2965 const VkAllocationCallbacks
* pAllocator
,
2966 VkDeviceMemory
* pMem
)
2968 struct radv_device_memory
*mem
;
2970 enum radeon_bo_domain domain
;
2972 enum radv_mem_type mem_type_index
= device
->physical_device
->mem_type_indices
[pAllocateInfo
->memoryTypeIndex
];
2974 assert(pAllocateInfo
->sType
== VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO
);
2976 if (pAllocateInfo
->allocationSize
== 0) {
2977 /* Apparently, this is allowed */
2978 *pMem
= VK_NULL_HANDLE
;
2982 const VkImportMemoryFdInfoKHR
*import_info
=
2983 vk_find_struct_const(pAllocateInfo
->pNext
, IMPORT_MEMORY_FD_INFO_KHR
);
2984 const VkMemoryDedicatedAllocateInfo
*dedicate_info
=
2985 vk_find_struct_const(pAllocateInfo
->pNext
, MEMORY_DEDICATED_ALLOCATE_INFO
);
2986 const VkExportMemoryAllocateInfo
*export_info
=
2987 vk_find_struct_const(pAllocateInfo
->pNext
, EXPORT_MEMORY_ALLOCATE_INFO
);
2988 const VkImportMemoryHostPointerInfoEXT
*host_ptr_info
=
2989 vk_find_struct_const(pAllocateInfo
->pNext
, IMPORT_MEMORY_HOST_POINTER_INFO_EXT
);
2991 const struct wsi_memory_allocate_info
*wsi_info
=
2992 vk_find_struct_const(pAllocateInfo
->pNext
, WSI_MEMORY_ALLOCATE_INFO_MESA
);
2994 mem
= vk_alloc2(&device
->alloc
, pAllocator
, sizeof(*mem
), 8,
2995 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
2997 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
2999 if (wsi_info
&& wsi_info
->implicit_sync
)
3000 flags
|= RADEON_FLAG_IMPLICIT_SYNC
;
3002 if (dedicate_info
) {
3003 mem
->image
= radv_image_from_handle(dedicate_info
->image
);
3004 mem
->buffer
= radv_buffer_from_handle(dedicate_info
->buffer
);
3010 mem
->user_ptr
= NULL
;
3013 assert(import_info
->handleType
==
3014 VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD_BIT
||
3015 import_info
->handleType
==
3016 VK_EXTERNAL_MEMORY_HANDLE_TYPE_DMA_BUF_BIT_EXT
);
3017 mem
->bo
= device
->ws
->buffer_from_fd(device
->ws
, import_info
->fd
,
3020 result
= VK_ERROR_INVALID_EXTERNAL_HANDLE
;
3023 close(import_info
->fd
);
3025 } else if (host_ptr_info
) {
3026 assert(host_ptr_info
->handleType
== VK_EXTERNAL_MEMORY_HANDLE_TYPE_HOST_ALLOCATION_BIT_EXT
);
3027 assert(mem_type_index
== RADV_MEM_TYPE_GTT_CACHED
);
3028 mem
->bo
= device
->ws
->buffer_from_ptr(device
->ws
, host_ptr_info
->pHostPointer
,
3029 pAllocateInfo
->allocationSize
);
3031 result
= VK_ERROR_INVALID_EXTERNAL_HANDLE
;
3034 mem
->user_ptr
= host_ptr_info
->pHostPointer
;
3037 uint64_t alloc_size
= align_u64(pAllocateInfo
->allocationSize
, 4096);
3038 if (mem_type_index
== RADV_MEM_TYPE_GTT_WRITE_COMBINE
||
3039 mem_type_index
== RADV_MEM_TYPE_GTT_CACHED
)
3040 domain
= RADEON_DOMAIN_GTT
;
3042 domain
= RADEON_DOMAIN_VRAM
;
3044 if (mem_type_index
== RADV_MEM_TYPE_VRAM
)
3045 flags
|= RADEON_FLAG_NO_CPU_ACCESS
;
3047 flags
|= RADEON_FLAG_CPU_ACCESS
;
3049 if (mem_type_index
== RADV_MEM_TYPE_GTT_WRITE_COMBINE
)
3050 flags
|= RADEON_FLAG_GTT_WC
;
3052 if (!dedicate_info
&& !import_info
&& (!export_info
|| !export_info
->handleTypes
))
3053 flags
|= RADEON_FLAG_NO_INTERPROCESS_SHARING
;
3055 mem
->bo
= device
->ws
->buffer_create(device
->ws
, alloc_size
, device
->physical_device
->rad_info
.max_alignment
,
3059 result
= VK_ERROR_OUT_OF_DEVICE_MEMORY
;
3062 mem
->type_index
= mem_type_index
;
3065 result
= radv_bo_list_add(device
, mem
->bo
);
3066 if (result
!= VK_SUCCESS
)
3069 *pMem
= radv_device_memory_to_handle(mem
);
3074 device
->ws
->buffer_destroy(mem
->bo
);
3076 vk_free2(&device
->alloc
, pAllocator
, mem
);
3081 VkResult
radv_AllocateMemory(
3083 const VkMemoryAllocateInfo
* pAllocateInfo
,
3084 const VkAllocationCallbacks
* pAllocator
,
3085 VkDeviceMemory
* pMem
)
3087 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3088 return radv_alloc_memory(device
, pAllocateInfo
, pAllocator
, pMem
);
3091 void radv_FreeMemory(
3093 VkDeviceMemory _mem
,
3094 const VkAllocationCallbacks
* pAllocator
)
3096 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3097 RADV_FROM_HANDLE(radv_device_memory
, mem
, _mem
);
3102 radv_bo_list_remove(device
, mem
->bo
);
3103 device
->ws
->buffer_destroy(mem
->bo
);
3106 vk_free2(&device
->alloc
, pAllocator
, mem
);
3109 VkResult
radv_MapMemory(
3111 VkDeviceMemory _memory
,
3112 VkDeviceSize offset
,
3114 VkMemoryMapFlags flags
,
3117 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3118 RADV_FROM_HANDLE(radv_device_memory
, mem
, _memory
);
3126 *ppData
= mem
->user_ptr
;
3128 *ppData
= device
->ws
->buffer_map(mem
->bo
);
3135 return vk_error(device
->instance
, VK_ERROR_MEMORY_MAP_FAILED
);
3138 void radv_UnmapMemory(
3140 VkDeviceMemory _memory
)
3142 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3143 RADV_FROM_HANDLE(radv_device_memory
, mem
, _memory
);
3148 if (mem
->user_ptr
== NULL
)
3149 device
->ws
->buffer_unmap(mem
->bo
);
3152 VkResult
radv_FlushMappedMemoryRanges(
3154 uint32_t memoryRangeCount
,
3155 const VkMappedMemoryRange
* pMemoryRanges
)
3160 VkResult
radv_InvalidateMappedMemoryRanges(
3162 uint32_t memoryRangeCount
,
3163 const VkMappedMemoryRange
* pMemoryRanges
)
3168 void radv_GetBufferMemoryRequirements(
3171 VkMemoryRequirements
* pMemoryRequirements
)
3173 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3174 RADV_FROM_HANDLE(radv_buffer
, buffer
, _buffer
);
3176 pMemoryRequirements
->memoryTypeBits
= (1u << device
->physical_device
->memory_properties
.memoryTypeCount
) - 1;
3178 if (buffer
->flags
& VK_BUFFER_CREATE_SPARSE_BINDING_BIT
)
3179 pMemoryRequirements
->alignment
= 4096;
3181 pMemoryRequirements
->alignment
= 16;
3183 pMemoryRequirements
->size
= align64(buffer
->size
, pMemoryRequirements
->alignment
);
3186 void radv_GetBufferMemoryRequirements2(
3188 const VkBufferMemoryRequirementsInfo2
*pInfo
,
3189 VkMemoryRequirements2
*pMemoryRequirements
)
3191 radv_GetBufferMemoryRequirements(device
, pInfo
->buffer
,
3192 &pMemoryRequirements
->memoryRequirements
);
3193 RADV_FROM_HANDLE(radv_buffer
, buffer
, pInfo
->buffer
);
3194 vk_foreach_struct(ext
, pMemoryRequirements
->pNext
) {
3195 switch (ext
->sType
) {
3196 case VK_STRUCTURE_TYPE_MEMORY_DEDICATED_REQUIREMENTS
: {
3197 VkMemoryDedicatedRequirements
*req
=
3198 (VkMemoryDedicatedRequirements
*) ext
;
3199 req
->requiresDedicatedAllocation
= buffer
->shareable
;
3200 req
->prefersDedicatedAllocation
= req
->requiresDedicatedAllocation
;
3209 void radv_GetImageMemoryRequirements(
3212 VkMemoryRequirements
* pMemoryRequirements
)
3214 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3215 RADV_FROM_HANDLE(radv_image
, image
, _image
);
3217 pMemoryRequirements
->memoryTypeBits
= (1u << device
->physical_device
->memory_properties
.memoryTypeCount
) - 1;
3219 pMemoryRequirements
->size
= image
->size
;
3220 pMemoryRequirements
->alignment
= image
->alignment
;
3223 void radv_GetImageMemoryRequirements2(
3225 const VkImageMemoryRequirementsInfo2
*pInfo
,
3226 VkMemoryRequirements2
*pMemoryRequirements
)
3228 radv_GetImageMemoryRequirements(device
, pInfo
->image
,
3229 &pMemoryRequirements
->memoryRequirements
);
3231 RADV_FROM_HANDLE(radv_image
, image
, pInfo
->image
);
3233 vk_foreach_struct(ext
, pMemoryRequirements
->pNext
) {
3234 switch (ext
->sType
) {
3235 case VK_STRUCTURE_TYPE_MEMORY_DEDICATED_REQUIREMENTS
: {
3236 VkMemoryDedicatedRequirements
*req
=
3237 (VkMemoryDedicatedRequirements
*) ext
;
3238 req
->requiresDedicatedAllocation
= image
->shareable
;
3239 req
->prefersDedicatedAllocation
= req
->requiresDedicatedAllocation
;
3248 void radv_GetImageSparseMemoryRequirements(
3251 uint32_t* pSparseMemoryRequirementCount
,
3252 VkSparseImageMemoryRequirements
* pSparseMemoryRequirements
)
3257 void radv_GetImageSparseMemoryRequirements2(
3259 const VkImageSparseMemoryRequirementsInfo2
*pInfo
,
3260 uint32_t* pSparseMemoryRequirementCount
,
3261 VkSparseImageMemoryRequirements2
*pSparseMemoryRequirements
)
3266 void radv_GetDeviceMemoryCommitment(
3268 VkDeviceMemory memory
,
3269 VkDeviceSize
* pCommittedMemoryInBytes
)
3271 *pCommittedMemoryInBytes
= 0;
3274 VkResult
radv_BindBufferMemory2(VkDevice device
,
3275 uint32_t bindInfoCount
,
3276 const VkBindBufferMemoryInfo
*pBindInfos
)
3278 for (uint32_t i
= 0; i
< bindInfoCount
; ++i
) {
3279 RADV_FROM_HANDLE(radv_device_memory
, mem
, pBindInfos
[i
].memory
);
3280 RADV_FROM_HANDLE(radv_buffer
, buffer
, pBindInfos
[i
].buffer
);
3283 buffer
->bo
= mem
->bo
;
3284 buffer
->offset
= pBindInfos
[i
].memoryOffset
;
3292 VkResult
radv_BindBufferMemory(
3295 VkDeviceMemory memory
,
3296 VkDeviceSize memoryOffset
)
3298 const VkBindBufferMemoryInfo info
= {
3299 .sType
= VK_STRUCTURE_TYPE_BIND_BUFFER_MEMORY_INFO
,
3302 .memoryOffset
= memoryOffset
3305 return radv_BindBufferMemory2(device
, 1, &info
);
3308 VkResult
radv_BindImageMemory2(VkDevice device
,
3309 uint32_t bindInfoCount
,
3310 const VkBindImageMemoryInfo
*pBindInfos
)
3312 for (uint32_t i
= 0; i
< bindInfoCount
; ++i
) {
3313 RADV_FROM_HANDLE(radv_device_memory
, mem
, pBindInfos
[i
].memory
);
3314 RADV_FROM_HANDLE(radv_image
, image
, pBindInfos
[i
].image
);
3317 image
->bo
= mem
->bo
;
3318 image
->offset
= pBindInfos
[i
].memoryOffset
;
3328 VkResult
radv_BindImageMemory(
3331 VkDeviceMemory memory
,
3332 VkDeviceSize memoryOffset
)
3334 const VkBindImageMemoryInfo info
= {
3335 .sType
= VK_STRUCTURE_TYPE_BIND_BUFFER_MEMORY_INFO
,
3338 .memoryOffset
= memoryOffset
3341 return radv_BindImageMemory2(device
, 1, &info
);
3346 radv_sparse_buffer_bind_memory(struct radv_device
*device
,
3347 const VkSparseBufferMemoryBindInfo
*bind
)
3349 RADV_FROM_HANDLE(radv_buffer
, buffer
, bind
->buffer
);
3351 for (uint32_t i
= 0; i
< bind
->bindCount
; ++i
) {
3352 struct radv_device_memory
*mem
= NULL
;
3354 if (bind
->pBinds
[i
].memory
!= VK_NULL_HANDLE
)
3355 mem
= radv_device_memory_from_handle(bind
->pBinds
[i
].memory
);
3357 device
->ws
->buffer_virtual_bind(buffer
->bo
,
3358 bind
->pBinds
[i
].resourceOffset
,
3359 bind
->pBinds
[i
].size
,
3360 mem
? mem
->bo
: NULL
,
3361 bind
->pBinds
[i
].memoryOffset
);
3366 radv_sparse_image_opaque_bind_memory(struct radv_device
*device
,
3367 const VkSparseImageOpaqueMemoryBindInfo
*bind
)
3369 RADV_FROM_HANDLE(radv_image
, image
, bind
->image
);
3371 for (uint32_t i
= 0; i
< bind
->bindCount
; ++i
) {
3372 struct radv_device_memory
*mem
= NULL
;
3374 if (bind
->pBinds
[i
].memory
!= VK_NULL_HANDLE
)
3375 mem
= radv_device_memory_from_handle(bind
->pBinds
[i
].memory
);
3377 device
->ws
->buffer_virtual_bind(image
->bo
,
3378 bind
->pBinds
[i
].resourceOffset
,
3379 bind
->pBinds
[i
].size
,
3380 mem
? mem
->bo
: NULL
,
3381 bind
->pBinds
[i
].memoryOffset
);
3385 VkResult
radv_QueueBindSparse(
3387 uint32_t bindInfoCount
,
3388 const VkBindSparseInfo
* pBindInfo
,
3391 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
3392 RADV_FROM_HANDLE(radv_queue
, queue
, _queue
);
3393 struct radeon_winsys_fence
*base_fence
= fence
? fence
->fence
: NULL
;
3394 bool fence_emitted
= false;
3398 for (uint32_t i
= 0; i
< bindInfoCount
; ++i
) {
3399 struct radv_winsys_sem_info sem_info
;
3400 for (uint32_t j
= 0; j
< pBindInfo
[i
].bufferBindCount
; ++j
) {
3401 radv_sparse_buffer_bind_memory(queue
->device
,
3402 pBindInfo
[i
].pBufferBinds
+ j
);
3405 for (uint32_t j
= 0; j
< pBindInfo
[i
].imageOpaqueBindCount
; ++j
) {
3406 radv_sparse_image_opaque_bind_memory(queue
->device
,
3407 pBindInfo
[i
].pImageOpaqueBinds
+ j
);
3411 result
= radv_alloc_sem_info(queue
->device
->instance
,
3413 pBindInfo
[i
].waitSemaphoreCount
,
3414 pBindInfo
[i
].pWaitSemaphores
,
3415 pBindInfo
[i
].signalSemaphoreCount
,
3416 pBindInfo
[i
].pSignalSemaphores
,
3418 if (result
!= VK_SUCCESS
)
3421 if (pBindInfo
[i
].waitSemaphoreCount
|| pBindInfo
[i
].signalSemaphoreCount
) {
3422 ret
= queue
->device
->ws
->cs_submit(queue
->hw_ctx
, queue
->queue_idx
,
3423 &queue
->device
->empty_cs
[queue
->queue_family_index
],
3428 radv_loge("failed to submit CS %d\n", i
);
3432 fence_emitted
= true;
3434 fence
->submitted
= true;
3437 radv_free_sem_info(&sem_info
);
3442 if (!fence_emitted
) {
3443 result
= radv_signal_fence(queue
, fence
);
3444 if (result
!= VK_SUCCESS
)
3447 fence
->submitted
= true;
3453 VkResult
radv_CreateFence(
3455 const VkFenceCreateInfo
* pCreateInfo
,
3456 const VkAllocationCallbacks
* pAllocator
,
3459 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3460 const VkExportFenceCreateInfo
*export
=
3461 vk_find_struct_const(pCreateInfo
->pNext
, EXPORT_FENCE_CREATE_INFO
);
3462 VkExternalFenceHandleTypeFlags handleTypes
=
3463 export
? export
->handleTypes
: 0;
3465 struct radv_fence
*fence
= vk_alloc2(&device
->alloc
, pAllocator
,
3467 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
3470 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3472 fence
->fence_wsi
= NULL
;
3473 fence
->submitted
= false;
3474 fence
->signalled
= !!(pCreateInfo
->flags
& VK_FENCE_CREATE_SIGNALED_BIT
);
3475 fence
->temp_syncobj
= 0;
3476 if (device
->always_use_syncobj
|| handleTypes
) {
3477 int ret
= device
->ws
->create_syncobj(device
->ws
, &fence
->syncobj
);
3479 vk_free2(&device
->alloc
, pAllocator
, fence
);
3480 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3482 if (pCreateInfo
->flags
& VK_FENCE_CREATE_SIGNALED_BIT
) {
3483 device
->ws
->signal_syncobj(device
->ws
, fence
->syncobj
);
3485 fence
->fence
= NULL
;
3487 fence
->fence
= device
->ws
->create_fence();
3488 if (!fence
->fence
) {
3489 vk_free2(&device
->alloc
, pAllocator
, fence
);
3490 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3495 *pFence
= radv_fence_to_handle(fence
);
3500 void radv_DestroyFence(
3503 const VkAllocationCallbacks
* pAllocator
)
3505 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3506 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
3511 if (fence
->temp_syncobj
)
3512 device
->ws
->destroy_syncobj(device
->ws
, fence
->temp_syncobj
);
3514 device
->ws
->destroy_syncobj(device
->ws
, fence
->syncobj
);
3516 device
->ws
->destroy_fence(fence
->fence
);
3517 if (fence
->fence_wsi
)
3518 fence
->fence_wsi
->destroy(fence
->fence_wsi
);
3519 vk_free2(&device
->alloc
, pAllocator
, fence
);
3523 static uint64_t radv_get_current_time()
3526 clock_gettime(CLOCK_MONOTONIC
, &tv
);
3527 return tv
.tv_nsec
+ tv
.tv_sec
*1000000000ull;
3530 static uint64_t radv_get_absolute_timeout(uint64_t timeout
)
3532 uint64_t current_time
= radv_get_current_time();
3534 timeout
= MIN2(UINT64_MAX
- current_time
, timeout
);
3536 return current_time
+ timeout
;
3540 static bool radv_all_fences_plain_and_submitted(uint32_t fenceCount
, const VkFence
*pFences
)
3542 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3543 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
3544 if (fence
->fence
== NULL
|| fence
->syncobj
||
3545 fence
->temp_syncobj
||
3546 (!fence
->signalled
&& !fence
->submitted
))
3552 static bool radv_all_fences_syncobj(uint32_t fenceCount
, const VkFence
*pFences
)
3554 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3555 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
3556 if (fence
->syncobj
== 0 && fence
->temp_syncobj
== 0)
3562 VkResult
radv_WaitForFences(
3564 uint32_t fenceCount
,
3565 const VkFence
* pFences
,
3569 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3570 timeout
= radv_get_absolute_timeout(timeout
);
3572 if (device
->always_use_syncobj
&&
3573 radv_all_fences_syncobj(fenceCount
, pFences
))
3575 uint32_t *handles
= malloc(sizeof(uint32_t) * fenceCount
);
3577 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3579 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3580 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
3581 handles
[i
] = fence
->temp_syncobj
? fence
->temp_syncobj
: fence
->syncobj
;
3584 bool success
= device
->ws
->wait_syncobj(device
->ws
, handles
, fenceCount
, waitAll
, timeout
);
3587 return success
? VK_SUCCESS
: VK_TIMEOUT
;
3590 if (!waitAll
&& fenceCount
> 1) {
3591 /* Not doing this by default for waitAll, due to needing to allocate twice. */
3592 if (device
->physical_device
->rad_info
.drm_minor
>= 10 && radv_all_fences_plain_and_submitted(fenceCount
, pFences
)) {
3593 uint32_t wait_count
= 0;
3594 struct radeon_winsys_fence
**fences
= malloc(sizeof(struct radeon_winsys_fence
*) * fenceCount
);
3596 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3598 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3599 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
3601 if (fence
->signalled
) {
3606 fences
[wait_count
++] = fence
->fence
;
3609 bool success
= device
->ws
->fences_wait(device
->ws
, fences
, wait_count
,
3610 waitAll
, timeout
- radv_get_current_time());
3613 return success
? VK_SUCCESS
: VK_TIMEOUT
;
3616 while(radv_get_current_time() <= timeout
) {
3617 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3618 if (radv_GetFenceStatus(_device
, pFences
[i
]) == VK_SUCCESS
)
3625 for (uint32_t i
= 0; i
< fenceCount
; ++i
) {
3626 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
3627 bool expired
= false;
3629 if (fence
->temp_syncobj
) {
3630 if (!device
->ws
->wait_syncobj(device
->ws
, &fence
->temp_syncobj
, 1, true, timeout
))
3635 if (fence
->syncobj
) {
3636 if (!device
->ws
->wait_syncobj(device
->ws
, &fence
->syncobj
, 1, true, timeout
))
3641 if (fence
->signalled
)
3645 if (!fence
->submitted
) {
3646 while(radv_get_current_time() <= timeout
&&
3650 if (!fence
->submitted
)
3653 /* Recheck as it may have been set by
3654 * submitting operations. */
3656 if (fence
->signalled
)
3660 expired
= device
->ws
->fence_wait(device
->ws
,
3667 if (fence
->fence_wsi
) {
3668 VkResult result
= fence
->fence_wsi
->wait(fence
->fence_wsi
, timeout
);
3669 if (result
!= VK_SUCCESS
)
3673 fence
->signalled
= true;
3679 VkResult
radv_ResetFences(VkDevice _device
,
3680 uint32_t fenceCount
,
3681 const VkFence
*pFences
)
3683 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3685 for (unsigned i
= 0; i
< fenceCount
; ++i
) {
3686 RADV_FROM_HANDLE(radv_fence
, fence
, pFences
[i
]);
3687 fence
->submitted
= fence
->signalled
= false;
3689 /* Per spec, we first restore the permanent payload, and then reset, so
3690 * having a temp syncobj should not skip resetting the permanent syncobj. */
3691 if (fence
->temp_syncobj
) {
3692 device
->ws
->destroy_syncobj(device
->ws
, fence
->temp_syncobj
);
3693 fence
->temp_syncobj
= 0;
3696 if (fence
->syncobj
) {
3697 device
->ws
->reset_syncobj(device
->ws
, fence
->syncobj
);
3704 VkResult
radv_GetFenceStatus(VkDevice _device
, VkFence _fence
)
3706 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3707 RADV_FROM_HANDLE(radv_fence
, fence
, _fence
);
3709 if (fence
->temp_syncobj
) {
3710 bool success
= device
->ws
->wait_syncobj(device
->ws
, &fence
->temp_syncobj
, 1, true, 0);
3711 return success
? VK_SUCCESS
: VK_NOT_READY
;
3714 if (fence
->syncobj
) {
3715 bool success
= device
->ws
->wait_syncobj(device
->ws
, &fence
->syncobj
, 1, true, 0);
3716 return success
? VK_SUCCESS
: VK_NOT_READY
;
3719 if (fence
->signalled
)
3721 if (!fence
->submitted
)
3722 return VK_NOT_READY
;
3724 if (!device
->ws
->fence_wait(device
->ws
, fence
->fence
, false, 0))
3725 return VK_NOT_READY
;
3727 if (fence
->fence_wsi
) {
3728 VkResult result
= fence
->fence_wsi
->wait(fence
->fence_wsi
, 0);
3730 if (result
!= VK_SUCCESS
) {
3731 if (result
== VK_TIMEOUT
)
3732 return VK_NOT_READY
;
3740 // Queue semaphore functions
3742 VkResult
radv_CreateSemaphore(
3744 const VkSemaphoreCreateInfo
* pCreateInfo
,
3745 const VkAllocationCallbacks
* pAllocator
,
3746 VkSemaphore
* pSemaphore
)
3748 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3749 const VkExportSemaphoreCreateInfo
*export
=
3750 vk_find_struct_const(pCreateInfo
->pNext
, EXPORT_SEMAPHORE_CREATE_INFO
);
3751 VkExternalSemaphoreHandleTypeFlags handleTypes
=
3752 export
? export
->handleTypes
: 0;
3754 struct radv_semaphore
*sem
= vk_alloc2(&device
->alloc
, pAllocator
,
3756 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
3758 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3760 sem
->temp_syncobj
= 0;
3761 /* create a syncobject if we are going to export this semaphore */
3762 if (device
->always_use_syncobj
|| handleTypes
) {
3763 assert (device
->physical_device
->rad_info
.has_syncobj
);
3764 int ret
= device
->ws
->create_syncobj(device
->ws
, &sem
->syncobj
);
3766 vk_free2(&device
->alloc
, pAllocator
, sem
);
3767 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3771 sem
->sem
= device
->ws
->create_sem(device
->ws
);
3773 vk_free2(&device
->alloc
, pAllocator
, sem
);
3774 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3779 *pSemaphore
= radv_semaphore_to_handle(sem
);
3783 void radv_DestroySemaphore(
3785 VkSemaphore _semaphore
,
3786 const VkAllocationCallbacks
* pAllocator
)
3788 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3789 RADV_FROM_HANDLE(radv_semaphore
, sem
, _semaphore
);
3794 device
->ws
->destroy_syncobj(device
->ws
, sem
->syncobj
);
3796 device
->ws
->destroy_sem(sem
->sem
);
3797 vk_free2(&device
->alloc
, pAllocator
, sem
);
3800 VkResult
radv_CreateEvent(
3802 const VkEventCreateInfo
* pCreateInfo
,
3803 const VkAllocationCallbacks
* pAllocator
,
3806 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3807 struct radv_event
*event
= vk_alloc2(&device
->alloc
, pAllocator
,
3809 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
3812 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3814 event
->bo
= device
->ws
->buffer_create(device
->ws
, 8, 8,
3816 RADEON_FLAG_VA_UNCACHED
| RADEON_FLAG_CPU_ACCESS
| RADEON_FLAG_NO_INTERPROCESS_SHARING
);
3818 vk_free2(&device
->alloc
, pAllocator
, event
);
3819 return vk_error(device
->instance
, VK_ERROR_OUT_OF_DEVICE_MEMORY
);
3822 event
->map
= (uint64_t*)device
->ws
->buffer_map(event
->bo
);
3824 *pEvent
= radv_event_to_handle(event
);
3829 void radv_DestroyEvent(
3832 const VkAllocationCallbacks
* pAllocator
)
3834 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3835 RADV_FROM_HANDLE(radv_event
, event
, _event
);
3839 device
->ws
->buffer_destroy(event
->bo
);
3840 vk_free2(&device
->alloc
, pAllocator
, event
);
3843 VkResult
radv_GetEventStatus(
3847 RADV_FROM_HANDLE(radv_event
, event
, _event
);
3849 if (*event
->map
== 1)
3850 return VK_EVENT_SET
;
3851 return VK_EVENT_RESET
;
3854 VkResult
radv_SetEvent(
3858 RADV_FROM_HANDLE(radv_event
, event
, _event
);
3864 VkResult
radv_ResetEvent(
3868 RADV_FROM_HANDLE(radv_event
, event
, _event
);
3874 VkResult
radv_CreateBuffer(
3876 const VkBufferCreateInfo
* pCreateInfo
,
3877 const VkAllocationCallbacks
* pAllocator
,
3880 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3881 struct radv_buffer
*buffer
;
3883 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO
);
3885 buffer
= vk_alloc2(&device
->alloc
, pAllocator
, sizeof(*buffer
), 8,
3886 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
3888 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
3890 buffer
->size
= pCreateInfo
->size
;
3891 buffer
->usage
= pCreateInfo
->usage
;
3894 buffer
->flags
= pCreateInfo
->flags
;
3896 buffer
->shareable
= vk_find_struct_const(pCreateInfo
->pNext
,
3897 EXTERNAL_MEMORY_BUFFER_CREATE_INFO
) != NULL
;
3899 if (pCreateInfo
->flags
& VK_BUFFER_CREATE_SPARSE_BINDING_BIT
) {
3900 buffer
->bo
= device
->ws
->buffer_create(device
->ws
,
3901 align64(buffer
->size
, 4096),
3902 4096, 0, RADEON_FLAG_VIRTUAL
);
3904 vk_free2(&device
->alloc
, pAllocator
, buffer
);
3905 return vk_error(device
->instance
, VK_ERROR_OUT_OF_DEVICE_MEMORY
);
3909 *pBuffer
= radv_buffer_to_handle(buffer
);
3914 void radv_DestroyBuffer(
3917 const VkAllocationCallbacks
* pAllocator
)
3919 RADV_FROM_HANDLE(radv_device
, device
, _device
);
3920 RADV_FROM_HANDLE(radv_buffer
, buffer
, _buffer
);
3925 if (buffer
->flags
& VK_BUFFER_CREATE_SPARSE_BINDING_BIT
)
3926 device
->ws
->buffer_destroy(buffer
->bo
);
3928 vk_free2(&device
->alloc
, pAllocator
, buffer
);
3931 static inline unsigned
3932 si_tile_mode_index(const struct radv_image
*image
, unsigned level
, bool stencil
)
3935 return image
->surface
.u
.legacy
.stencil_tiling_index
[level
];
3937 return image
->surface
.u
.legacy
.tiling_index
[level
];
3940 static uint32_t radv_surface_max_layer_count(struct radv_image_view
*iview
)
3942 return iview
->type
== VK_IMAGE_VIEW_TYPE_3D
? iview
->extent
.depth
: (iview
->base_layer
+ iview
->layer_count
);
3946 radv_init_dcc_control_reg(struct radv_device
*device
,
3947 struct radv_image_view
*iview
)
3949 unsigned max_uncompressed_block_size
= V_028C78_MAX_BLOCK_SIZE_256B
;
3950 unsigned min_compressed_block_size
= V_028C78_MIN_BLOCK_SIZE_32B
;
3951 unsigned max_compressed_block_size
;
3952 unsigned independent_64b_blocks
;
3954 if (!radv_image_has_dcc(iview
->image
))
3957 if (iview
->image
->info
.samples
> 1) {
3958 if (iview
->image
->surface
.bpe
== 1)
3959 max_uncompressed_block_size
= V_028C78_MAX_BLOCK_SIZE_64B
;
3960 else if (iview
->image
->surface
.bpe
== 2)
3961 max_uncompressed_block_size
= V_028C78_MAX_BLOCK_SIZE_128B
;
3964 if (!device
->physical_device
->rad_info
.has_dedicated_vram
) {
3965 /* amdvlk: [min-compressed-block-size] should be set to 32 for
3966 * dGPU and 64 for APU because all of our APUs to date use
3967 * DIMMs which have a request granularity size of 64B while all
3968 * other chips have a 32B request size.
3970 min_compressed_block_size
= V_028C78_MIN_BLOCK_SIZE_64B
;
3973 if (iview
->image
->usage
& (VK_IMAGE_USAGE_SAMPLED_BIT
|
3974 VK_IMAGE_USAGE_TRANSFER_SRC_BIT
|
3975 VK_IMAGE_USAGE_INPUT_ATTACHMENT_BIT
)) {
3976 /* If this DCC image is potentially going to be used in texture
3977 * fetches, we need some special settings.
3979 independent_64b_blocks
= 1;
3980 max_compressed_block_size
= V_028C78_MAX_BLOCK_SIZE_64B
;
3982 /* MAX_UNCOMPRESSED_BLOCK_SIZE must be >=
3983 * MAX_COMPRESSED_BLOCK_SIZE. Set MAX_COMPRESSED_BLOCK_SIZE as
3984 * big as possible for better compression state.
3986 independent_64b_blocks
= 0;
3987 max_compressed_block_size
= max_uncompressed_block_size
;
3990 return S_028C78_MAX_UNCOMPRESSED_BLOCK_SIZE(max_uncompressed_block_size
) |
3991 S_028C78_MAX_COMPRESSED_BLOCK_SIZE(max_compressed_block_size
) |
3992 S_028C78_MIN_COMPRESSED_BLOCK_SIZE(min_compressed_block_size
) |
3993 S_028C78_INDEPENDENT_64B_BLOCKS(independent_64b_blocks
);
3997 radv_initialise_color_surface(struct radv_device
*device
,
3998 struct radv_color_buffer_info
*cb
,
3999 struct radv_image_view
*iview
)
4001 const struct vk_format_description
*desc
;
4002 unsigned ntype
, format
, swap
, endian
;
4003 unsigned blend_clamp
= 0, blend_bypass
= 0;
4005 const struct radeon_surf
*surf
= &iview
->image
->surface
;
4007 desc
= vk_format_description(iview
->vk_format
);
4009 memset(cb
, 0, sizeof(*cb
));
4011 /* Intensity is implemented as Red, so treat it that way. */
4012 cb
->cb_color_attrib
= S_028C74_FORCE_DST_ALPHA_1(desc
->swizzle
[3] == VK_SWIZZLE_1
);
4014 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
;
4016 cb
->cb_color_base
= va
>> 8;
4018 if (device
->physical_device
->rad_info
.chip_class
>= GFX9
) {
4019 struct gfx9_surf_meta_flags meta
;
4020 if (iview
->image
->dcc_offset
)
4021 meta
= iview
->image
->surface
.u
.gfx9
.dcc
;
4023 meta
= iview
->image
->surface
.u
.gfx9
.cmask
;
4025 cb
->cb_color_attrib
|= S_028C74_COLOR_SW_MODE(iview
->image
->surface
.u
.gfx9
.surf
.swizzle_mode
) |
4026 S_028C74_FMASK_SW_MODE(iview
->image
->surface
.u
.gfx9
.fmask
.swizzle_mode
) |
4027 S_028C74_RB_ALIGNED(meta
.rb_aligned
) |
4028 S_028C74_PIPE_ALIGNED(meta
.pipe_aligned
);
4030 cb
->cb_color_base
+= iview
->image
->surface
.u
.gfx9
.surf_offset
>> 8;
4031 cb
->cb_color_base
|= iview
->image
->surface
.tile_swizzle
;
4033 const struct legacy_surf_level
*level_info
= &surf
->u
.legacy
.level
[iview
->base_mip
];
4034 unsigned pitch_tile_max
, slice_tile_max
, tile_mode_index
;
4036 cb
->cb_color_base
+= level_info
->offset
>> 8;
4037 if (level_info
->mode
== RADEON_SURF_MODE_2D
)
4038 cb
->cb_color_base
|= iview
->image
->surface
.tile_swizzle
;
4040 pitch_tile_max
= level_info
->nblk_x
/ 8 - 1;
4041 slice_tile_max
= (level_info
->nblk_x
* level_info
->nblk_y
) / 64 - 1;
4042 tile_mode_index
= si_tile_mode_index(iview
->image
, iview
->base_mip
, false);
4044 cb
->cb_color_pitch
= S_028C64_TILE_MAX(pitch_tile_max
);
4045 cb
->cb_color_slice
= S_028C68_TILE_MAX(slice_tile_max
);
4046 cb
->cb_color_cmask_slice
= iview
->image
->cmask
.slice_tile_max
;
4048 cb
->cb_color_attrib
|= S_028C74_TILE_MODE_INDEX(tile_mode_index
);
4050 if (radv_image_has_fmask(iview
->image
)) {
4051 if (device
->physical_device
->rad_info
.chip_class
>= CIK
)
4052 cb
->cb_color_pitch
|= S_028C64_FMASK_TILE_MAX(iview
->image
->fmask
.pitch_in_pixels
/ 8 - 1);
4053 cb
->cb_color_attrib
|= S_028C74_FMASK_TILE_MODE_INDEX(iview
->image
->fmask
.tile_mode_index
);
4054 cb
->cb_color_fmask_slice
= S_028C88_TILE_MAX(iview
->image
->fmask
.slice_tile_max
);
4056 /* This must be set for fast clear to work without FMASK. */
4057 if (device
->physical_device
->rad_info
.chip_class
>= CIK
)
4058 cb
->cb_color_pitch
|= S_028C64_FMASK_TILE_MAX(pitch_tile_max
);
4059 cb
->cb_color_attrib
|= S_028C74_FMASK_TILE_MODE_INDEX(tile_mode_index
);
4060 cb
->cb_color_fmask_slice
= S_028C88_TILE_MAX(slice_tile_max
);
4064 /* CMASK variables */
4065 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
;
4066 va
+= iview
->image
->cmask
.offset
;
4067 cb
->cb_color_cmask
= va
>> 8;
4069 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
;
4070 va
+= iview
->image
->dcc_offset
;
4071 cb
->cb_dcc_base
= va
>> 8;
4072 cb
->cb_dcc_base
|= iview
->image
->surface
.tile_swizzle
;
4074 uint32_t max_slice
= radv_surface_max_layer_count(iview
) - 1;
4075 cb
->cb_color_view
= S_028C6C_SLICE_START(iview
->base_layer
) |
4076 S_028C6C_SLICE_MAX(max_slice
);
4078 if (iview
->image
->info
.samples
> 1) {
4079 unsigned log_samples
= util_logbase2(iview
->image
->info
.samples
);
4081 cb
->cb_color_attrib
|= S_028C74_NUM_SAMPLES(log_samples
) |
4082 S_028C74_NUM_FRAGMENTS(log_samples
);
4085 if (radv_image_has_fmask(iview
->image
)) {
4086 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
+ iview
->image
->fmask
.offset
;
4087 cb
->cb_color_fmask
= va
>> 8;
4088 cb
->cb_color_fmask
|= iview
->image
->fmask
.tile_swizzle
;
4090 cb
->cb_color_fmask
= cb
->cb_color_base
;
4093 ntype
= radv_translate_color_numformat(iview
->vk_format
,
4095 vk_format_get_first_non_void_channel(iview
->vk_format
));
4096 format
= radv_translate_colorformat(iview
->vk_format
);
4097 if (format
== V_028C70_COLOR_INVALID
|| ntype
== ~0u)
4098 radv_finishme("Illegal color\n");
4099 swap
= radv_translate_colorswap(iview
->vk_format
, FALSE
);
4100 endian
= radv_colorformat_endian_swap(format
);
4102 /* blend clamp should be set for all NORM/SRGB types */
4103 if (ntype
== V_028C70_NUMBER_UNORM
||
4104 ntype
== V_028C70_NUMBER_SNORM
||
4105 ntype
== V_028C70_NUMBER_SRGB
)
4108 /* set blend bypass according to docs if SINT/UINT or
4109 8/24 COLOR variants */
4110 if (ntype
== V_028C70_NUMBER_UINT
|| ntype
== V_028C70_NUMBER_SINT
||
4111 format
== V_028C70_COLOR_8_24
|| format
== V_028C70_COLOR_24_8
||
4112 format
== V_028C70_COLOR_X24_8_32_FLOAT
) {
4117 if ((ntype
== V_028C70_NUMBER_UINT
|| ntype
== V_028C70_NUMBER_SINT
) &&
4118 (format
== V_028C70_COLOR_8
||
4119 format
== V_028C70_COLOR_8_8
||
4120 format
== V_028C70_COLOR_8_8_8_8
))
4121 ->color_is_int8
= true;
4123 cb
->cb_color_info
= S_028C70_FORMAT(format
) |
4124 S_028C70_COMP_SWAP(swap
) |
4125 S_028C70_BLEND_CLAMP(blend_clamp
) |
4126 S_028C70_BLEND_BYPASS(blend_bypass
) |
4127 S_028C70_SIMPLE_FLOAT(1) |
4128 S_028C70_ROUND_MODE(ntype
!= V_028C70_NUMBER_UNORM
&&
4129 ntype
!= V_028C70_NUMBER_SNORM
&&
4130 ntype
!= V_028C70_NUMBER_SRGB
&&
4131 format
!= V_028C70_COLOR_8_24
&&
4132 format
!= V_028C70_COLOR_24_8
) |
4133 S_028C70_NUMBER_TYPE(ntype
) |
4134 S_028C70_ENDIAN(endian
);
4135 if (radv_image_has_fmask(iview
->image
)) {
4136 cb
->cb_color_info
|= S_028C70_COMPRESSION(1);
4137 if (device
->physical_device
->rad_info
.chip_class
== SI
) {
4138 unsigned fmask_bankh
= util_logbase2(iview
->image
->fmask
.bank_height
);
4139 cb
->cb_color_attrib
|= S_028C74_FMASK_BANK_HEIGHT(fmask_bankh
);
4143 if (radv_image_has_cmask(iview
->image
) &&
4144 !(device
->instance
->debug_flags
& RADV_DEBUG_NO_FAST_CLEARS
))
4145 cb
->cb_color_info
|= S_028C70_FAST_CLEAR(1);
4147 if (radv_dcc_enabled(iview
->image
, iview
->base_mip
))
4148 cb
->cb_color_info
|= S_028C70_DCC_ENABLE(1);
4150 cb
->cb_dcc_control
= radv_init_dcc_control_reg(device
, iview
);
4152 /* This must be set for fast clear to work without FMASK. */
4153 if (!radv_image_has_fmask(iview
->image
) &&
4154 device
->physical_device
->rad_info
.chip_class
== SI
) {
4155 unsigned bankh
= util_logbase2(iview
->image
->surface
.u
.legacy
.bankh
);
4156 cb
->cb_color_attrib
|= S_028C74_FMASK_BANK_HEIGHT(bankh
);
4159 if (device
->physical_device
->rad_info
.chip_class
>= GFX9
) {
4160 unsigned mip0_depth
= iview
->image
->type
== VK_IMAGE_TYPE_3D
?
4161 (iview
->extent
.depth
- 1) : (iview
->image
->info
.array_size
- 1);
4163 cb
->cb_color_view
|= S_028C6C_MIP_LEVEL(iview
->base_mip
);
4164 cb
->cb_color_attrib
|= S_028C74_MIP0_DEPTH(mip0_depth
) |
4165 S_028C74_RESOURCE_TYPE(iview
->image
->surface
.u
.gfx9
.resource_type
);
4166 cb
->cb_color_attrib2
= S_028C68_MIP0_WIDTH(iview
->extent
.width
- 1) |
4167 S_028C68_MIP0_HEIGHT(iview
->extent
.height
- 1) |
4168 S_028C68_MAX_MIP(iview
->image
->info
.levels
- 1);
4173 radv_calc_decompress_on_z_planes(struct radv_device
*device
,
4174 struct radv_image_view
*iview
)
4176 unsigned max_zplanes
= 0;
4178 assert(radv_image_is_tc_compat_htile(iview
->image
));
4180 if (device
->physical_device
->rad_info
.chip_class
>= GFX9
) {
4181 /* Default value for 32-bit depth surfaces. */
4184 if (iview
->vk_format
== VK_FORMAT_D16_UNORM
&&
4185 iview
->image
->info
.samples
> 1)
4188 max_zplanes
= max_zplanes
+ 1;
4190 if (iview
->vk_format
== VK_FORMAT_D16_UNORM
) {
4191 /* Do not enable Z plane compression for 16-bit depth
4192 * surfaces because isn't supported on GFX8. Only
4193 * 32-bit depth surfaces are supported by the hardware.
4194 * This allows to maintain shader compatibility and to
4195 * reduce the number of depth decompressions.
4199 if (iview
->image
->info
.samples
<= 1)
4201 else if (iview
->image
->info
.samples
<= 4)
4212 radv_initialise_ds_surface(struct radv_device
*device
,
4213 struct radv_ds_buffer_info
*ds
,
4214 struct radv_image_view
*iview
)
4216 unsigned level
= iview
->base_mip
;
4217 unsigned format
, stencil_format
;
4218 uint64_t va
, s_offs
, z_offs
;
4219 bool stencil_only
= false;
4220 memset(ds
, 0, sizeof(*ds
));
4221 switch (iview
->image
->vk_format
) {
4222 case VK_FORMAT_D24_UNORM_S8_UINT
:
4223 case VK_FORMAT_X8_D24_UNORM_PACK32
:
4224 ds
->pa_su_poly_offset_db_fmt_cntl
= S_028B78_POLY_OFFSET_NEG_NUM_DB_BITS(-24);
4225 ds
->offset_scale
= 2.0f
;
4227 case VK_FORMAT_D16_UNORM
:
4228 case VK_FORMAT_D16_UNORM_S8_UINT
:
4229 ds
->pa_su_poly_offset_db_fmt_cntl
= S_028B78_POLY_OFFSET_NEG_NUM_DB_BITS(-16);
4230 ds
->offset_scale
= 4.0f
;
4232 case VK_FORMAT_D32_SFLOAT
:
4233 case VK_FORMAT_D32_SFLOAT_S8_UINT
:
4234 ds
->pa_su_poly_offset_db_fmt_cntl
= S_028B78_POLY_OFFSET_NEG_NUM_DB_BITS(-23) |
4235 S_028B78_POLY_OFFSET_DB_IS_FLOAT_FMT(1);
4236 ds
->offset_scale
= 1.0f
;
4238 case VK_FORMAT_S8_UINT
:
4239 stencil_only
= true;
4245 format
= radv_translate_dbformat(iview
->image
->vk_format
);
4246 stencil_format
= iview
->image
->surface
.has_stencil
?
4247 V_028044_STENCIL_8
: V_028044_STENCIL_INVALID
;
4249 uint32_t max_slice
= radv_surface_max_layer_count(iview
) - 1;
4250 ds
->db_depth_view
= S_028008_SLICE_START(iview
->base_layer
) |
4251 S_028008_SLICE_MAX(max_slice
);
4253 ds
->db_htile_data_base
= 0;
4254 ds
->db_htile_surface
= 0;
4256 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
;
4257 s_offs
= z_offs
= va
;
4259 if (device
->physical_device
->rad_info
.chip_class
>= GFX9
) {
4260 assert(iview
->image
->surface
.u
.gfx9
.surf_offset
== 0);
4261 s_offs
+= iview
->image
->surface
.u
.gfx9
.stencil_offset
;
4263 ds
->db_z_info
= S_028038_FORMAT(format
) |
4264 S_028038_NUM_SAMPLES(util_logbase2(iview
->image
->info
.samples
)) |
4265 S_028038_SW_MODE(iview
->image
->surface
.u
.gfx9
.surf
.swizzle_mode
) |
4266 S_028038_MAXMIP(iview
->image
->info
.levels
- 1) |
4267 S_028038_ZRANGE_PRECISION(1);
4268 ds
->db_stencil_info
= S_02803C_FORMAT(stencil_format
) |
4269 S_02803C_SW_MODE(iview
->image
->surface
.u
.gfx9
.stencil
.swizzle_mode
);
4271 ds
->db_z_info2
= S_028068_EPITCH(iview
->image
->surface
.u
.gfx9
.surf
.epitch
);
4272 ds
->db_stencil_info2
= S_02806C_EPITCH(iview
->image
->surface
.u
.gfx9
.stencil
.epitch
);
4273 ds
->db_depth_view
|= S_028008_MIPID(level
);
4275 ds
->db_depth_size
= S_02801C_X_MAX(iview
->image
->info
.width
- 1) |
4276 S_02801C_Y_MAX(iview
->image
->info
.height
- 1);
4278 if (radv_htile_enabled(iview
->image
, level
)) {
4279 ds
->db_z_info
|= S_028038_TILE_SURFACE_ENABLE(1);
4281 if (radv_image_is_tc_compat_htile(iview
->image
)) {
4282 unsigned max_zplanes
=
4283 radv_calc_decompress_on_z_planes(device
, iview
);
4285 ds
->db_z_info
|= S_028038_DECOMPRESS_ON_N_ZPLANES(max_zplanes
) |
4286 S_028038_ITERATE_FLUSH(1);
4287 ds
->db_stencil_info
|= S_02803C_ITERATE_FLUSH(1);
4290 if (!iview
->image
->surface
.has_stencil
)
4291 /* Use all of the htile_buffer for depth if there's no stencil. */
4292 ds
->db_stencil_info
|= S_02803C_TILE_STENCIL_DISABLE(1);
4293 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
+
4294 iview
->image
->htile_offset
;
4295 ds
->db_htile_data_base
= va
>> 8;
4296 ds
->db_htile_surface
= S_028ABC_FULL_CACHE(1) |
4297 S_028ABC_PIPE_ALIGNED(iview
->image
->surface
.u
.gfx9
.htile
.pipe_aligned
) |
4298 S_028ABC_RB_ALIGNED(iview
->image
->surface
.u
.gfx9
.htile
.rb_aligned
);
4301 const struct legacy_surf_level
*level_info
= &iview
->image
->surface
.u
.legacy
.level
[level
];
4304 level_info
= &iview
->image
->surface
.u
.legacy
.stencil_level
[level
];
4306 z_offs
+= iview
->image
->surface
.u
.legacy
.level
[level
].offset
;
4307 s_offs
+= iview
->image
->surface
.u
.legacy
.stencil_level
[level
].offset
;
4309 ds
->db_depth_info
= S_02803C_ADDR5_SWIZZLE_MASK(!radv_image_is_tc_compat_htile(iview
->image
));
4310 ds
->db_z_info
= S_028040_FORMAT(format
) | S_028040_ZRANGE_PRECISION(1);
4311 ds
->db_stencil_info
= S_028044_FORMAT(stencil_format
);
4313 if (iview
->image
->info
.samples
> 1)
4314 ds
->db_z_info
|= S_028040_NUM_SAMPLES(util_logbase2(iview
->image
->info
.samples
));
4316 if (device
->physical_device
->rad_info
.chip_class
>= CIK
) {
4317 struct radeon_info
*info
= &device
->physical_device
->rad_info
;
4318 unsigned tiling_index
= iview
->image
->surface
.u
.legacy
.tiling_index
[level
];
4319 unsigned stencil_index
= iview
->image
->surface
.u
.legacy
.stencil_tiling_index
[level
];
4320 unsigned macro_index
= iview
->image
->surface
.u
.legacy
.macro_tile_index
;
4321 unsigned tile_mode
= info
->si_tile_mode_array
[tiling_index
];
4322 unsigned stencil_tile_mode
= info
->si_tile_mode_array
[stencil_index
];
4323 unsigned macro_mode
= info
->cik_macrotile_mode_array
[macro_index
];
4326 tile_mode
= stencil_tile_mode
;
4328 ds
->db_depth_info
|=
4329 S_02803C_ARRAY_MODE(G_009910_ARRAY_MODE(tile_mode
)) |
4330 S_02803C_PIPE_CONFIG(G_009910_PIPE_CONFIG(tile_mode
)) |
4331 S_02803C_BANK_WIDTH(G_009990_BANK_WIDTH(macro_mode
)) |
4332 S_02803C_BANK_HEIGHT(G_009990_BANK_HEIGHT(macro_mode
)) |
4333 S_02803C_MACRO_TILE_ASPECT(G_009990_MACRO_TILE_ASPECT(macro_mode
)) |
4334 S_02803C_NUM_BANKS(G_009990_NUM_BANKS(macro_mode
));
4335 ds
->db_z_info
|= S_028040_TILE_SPLIT(G_009910_TILE_SPLIT(tile_mode
));
4336 ds
->db_stencil_info
|= S_028044_TILE_SPLIT(G_009910_TILE_SPLIT(stencil_tile_mode
));
4338 unsigned tile_mode_index
= si_tile_mode_index(iview
->image
, level
, false);
4339 ds
->db_z_info
|= S_028040_TILE_MODE_INDEX(tile_mode_index
);
4340 tile_mode_index
= si_tile_mode_index(iview
->image
, level
, true);
4341 ds
->db_stencil_info
|= S_028044_TILE_MODE_INDEX(tile_mode_index
);
4343 ds
->db_z_info
|= S_028040_TILE_MODE_INDEX(tile_mode_index
);
4346 ds
->db_depth_size
= S_028058_PITCH_TILE_MAX((level_info
->nblk_x
/ 8) - 1) |
4347 S_028058_HEIGHT_TILE_MAX((level_info
->nblk_y
/ 8) - 1);
4348 ds
->db_depth_slice
= S_02805C_SLICE_TILE_MAX((level_info
->nblk_x
* level_info
->nblk_y
) / 64 - 1);
4350 if (radv_htile_enabled(iview
->image
, level
)) {
4351 ds
->db_z_info
|= S_028040_TILE_SURFACE_ENABLE(1);
4353 if (!iview
->image
->surface
.has_stencil
&&
4354 !radv_image_is_tc_compat_htile(iview
->image
))
4355 /* Use all of the htile_buffer for depth if there's no stencil. */
4356 ds
->db_stencil_info
|= S_028044_TILE_STENCIL_DISABLE(1);
4358 va
= radv_buffer_get_va(iview
->bo
) + iview
->image
->offset
+
4359 iview
->image
->htile_offset
;
4360 ds
->db_htile_data_base
= va
>> 8;
4361 ds
->db_htile_surface
= S_028ABC_FULL_CACHE(1);
4363 if (radv_image_is_tc_compat_htile(iview
->image
)) {
4364 unsigned max_zplanes
=
4365 radv_calc_decompress_on_z_planes(device
, iview
);
4367 ds
->db_htile_surface
|= S_028ABC_TC_COMPATIBLE(1);
4368 ds
->db_z_info
|= S_028040_DECOMPRESS_ON_N_ZPLANES(max_zplanes
);
4373 ds
->db_z_read_base
= ds
->db_z_write_base
= z_offs
>> 8;
4374 ds
->db_stencil_read_base
= ds
->db_stencil_write_base
= s_offs
>> 8;
4377 VkResult
radv_CreateFramebuffer(
4379 const VkFramebufferCreateInfo
* pCreateInfo
,
4380 const VkAllocationCallbacks
* pAllocator
,
4381 VkFramebuffer
* pFramebuffer
)
4383 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4384 struct radv_framebuffer
*framebuffer
;
4386 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_FRAMEBUFFER_CREATE_INFO
);
4388 size_t size
= sizeof(*framebuffer
) +
4389 sizeof(struct radv_attachment_info
) * pCreateInfo
->attachmentCount
;
4390 framebuffer
= vk_alloc2(&device
->alloc
, pAllocator
, size
, 8,
4391 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
4392 if (framebuffer
== NULL
)
4393 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
4395 framebuffer
->attachment_count
= pCreateInfo
->attachmentCount
;
4396 framebuffer
->width
= pCreateInfo
->width
;
4397 framebuffer
->height
= pCreateInfo
->height
;
4398 framebuffer
->layers
= pCreateInfo
->layers
;
4399 for (uint32_t i
= 0; i
< pCreateInfo
->attachmentCount
; i
++) {
4400 VkImageView _iview
= pCreateInfo
->pAttachments
[i
];
4401 struct radv_image_view
*iview
= radv_image_view_from_handle(_iview
);
4402 framebuffer
->attachments
[i
].attachment
= iview
;
4403 if (iview
->aspect_mask
& VK_IMAGE_ASPECT_COLOR_BIT
) {
4404 radv_initialise_color_surface(device
, &framebuffer
->attachments
[i
].cb
, iview
);
4405 } else if (iview
->aspect_mask
& (VK_IMAGE_ASPECT_DEPTH_BIT
| VK_IMAGE_ASPECT_STENCIL_BIT
)) {
4406 radv_initialise_ds_surface(device
, &framebuffer
->attachments
[i
].ds
, iview
);
4408 framebuffer
->width
= MIN2(framebuffer
->width
, iview
->extent
.width
);
4409 framebuffer
->height
= MIN2(framebuffer
->height
, iview
->extent
.height
);
4410 framebuffer
->layers
= MIN2(framebuffer
->layers
, radv_surface_max_layer_count(iview
));
4413 *pFramebuffer
= radv_framebuffer_to_handle(framebuffer
);
4417 void radv_DestroyFramebuffer(
4420 const VkAllocationCallbacks
* pAllocator
)
4422 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4423 RADV_FROM_HANDLE(radv_framebuffer
, fb
, _fb
);
4427 vk_free2(&device
->alloc
, pAllocator
, fb
);
4430 static unsigned radv_tex_wrap(VkSamplerAddressMode address_mode
)
4432 switch (address_mode
) {
4433 case VK_SAMPLER_ADDRESS_MODE_REPEAT
:
4434 return V_008F30_SQ_TEX_WRAP
;
4435 case VK_SAMPLER_ADDRESS_MODE_MIRRORED_REPEAT
:
4436 return V_008F30_SQ_TEX_MIRROR
;
4437 case VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE
:
4438 return V_008F30_SQ_TEX_CLAMP_LAST_TEXEL
;
4439 case VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER
:
4440 return V_008F30_SQ_TEX_CLAMP_BORDER
;
4441 case VK_SAMPLER_ADDRESS_MODE_MIRROR_CLAMP_TO_EDGE
:
4442 return V_008F30_SQ_TEX_MIRROR_ONCE_LAST_TEXEL
;
4444 unreachable("illegal tex wrap mode");
4450 radv_tex_compare(VkCompareOp op
)
4453 case VK_COMPARE_OP_NEVER
:
4454 return V_008F30_SQ_TEX_DEPTH_COMPARE_NEVER
;
4455 case VK_COMPARE_OP_LESS
:
4456 return V_008F30_SQ_TEX_DEPTH_COMPARE_LESS
;
4457 case VK_COMPARE_OP_EQUAL
:
4458 return V_008F30_SQ_TEX_DEPTH_COMPARE_EQUAL
;
4459 case VK_COMPARE_OP_LESS_OR_EQUAL
:
4460 return V_008F30_SQ_TEX_DEPTH_COMPARE_LESSEQUAL
;
4461 case VK_COMPARE_OP_GREATER
:
4462 return V_008F30_SQ_TEX_DEPTH_COMPARE_GREATER
;
4463 case VK_COMPARE_OP_NOT_EQUAL
:
4464 return V_008F30_SQ_TEX_DEPTH_COMPARE_NOTEQUAL
;
4465 case VK_COMPARE_OP_GREATER_OR_EQUAL
:
4466 return V_008F30_SQ_TEX_DEPTH_COMPARE_GREATEREQUAL
;
4467 case VK_COMPARE_OP_ALWAYS
:
4468 return V_008F30_SQ_TEX_DEPTH_COMPARE_ALWAYS
;
4470 unreachable("illegal compare mode");
4476 radv_tex_filter(VkFilter filter
, unsigned max_ansio
)
4479 case VK_FILTER_NEAREST
:
4480 return (max_ansio
> 1 ? V_008F38_SQ_TEX_XY_FILTER_ANISO_POINT
:
4481 V_008F38_SQ_TEX_XY_FILTER_POINT
);
4482 case VK_FILTER_LINEAR
:
4483 return (max_ansio
> 1 ? V_008F38_SQ_TEX_XY_FILTER_ANISO_BILINEAR
:
4484 V_008F38_SQ_TEX_XY_FILTER_BILINEAR
);
4485 case VK_FILTER_CUBIC_IMG
:
4487 fprintf(stderr
, "illegal texture filter");
4493 radv_tex_mipfilter(VkSamplerMipmapMode mode
)
4496 case VK_SAMPLER_MIPMAP_MODE_NEAREST
:
4497 return V_008F38_SQ_TEX_Z_FILTER_POINT
;
4498 case VK_SAMPLER_MIPMAP_MODE_LINEAR
:
4499 return V_008F38_SQ_TEX_Z_FILTER_LINEAR
;
4501 return V_008F38_SQ_TEX_Z_FILTER_NONE
;
4506 radv_tex_bordercolor(VkBorderColor bcolor
)
4509 case VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK
:
4510 case VK_BORDER_COLOR_INT_TRANSPARENT_BLACK
:
4511 return V_008F3C_SQ_TEX_BORDER_COLOR_TRANS_BLACK
;
4512 case VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK
:
4513 case VK_BORDER_COLOR_INT_OPAQUE_BLACK
:
4514 return V_008F3C_SQ_TEX_BORDER_COLOR_OPAQUE_BLACK
;
4515 case VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE
:
4516 case VK_BORDER_COLOR_INT_OPAQUE_WHITE
:
4517 return V_008F3C_SQ_TEX_BORDER_COLOR_OPAQUE_WHITE
;
4525 radv_tex_aniso_filter(unsigned filter
)
4539 radv_tex_filter_mode(VkSamplerReductionModeEXT mode
)
4542 case VK_SAMPLER_REDUCTION_MODE_WEIGHTED_AVERAGE_EXT
:
4543 return V_008F30_SQ_IMG_FILTER_MODE_BLEND
;
4544 case VK_SAMPLER_REDUCTION_MODE_MIN_EXT
:
4545 return V_008F30_SQ_IMG_FILTER_MODE_MIN
;
4546 case VK_SAMPLER_REDUCTION_MODE_MAX_EXT
:
4547 return V_008F30_SQ_IMG_FILTER_MODE_MAX
;
4555 radv_get_max_anisotropy(struct radv_device
*device
,
4556 const VkSamplerCreateInfo
*pCreateInfo
)
4558 if (device
->force_aniso
>= 0)
4559 return device
->force_aniso
;
4561 if (pCreateInfo
->anisotropyEnable
&&
4562 pCreateInfo
->maxAnisotropy
> 1.0f
)
4563 return (uint32_t)pCreateInfo
->maxAnisotropy
;
4569 radv_init_sampler(struct radv_device
*device
,
4570 struct radv_sampler
*sampler
,
4571 const VkSamplerCreateInfo
*pCreateInfo
)
4573 uint32_t max_aniso
= radv_get_max_anisotropy(device
, pCreateInfo
);
4574 uint32_t max_aniso_ratio
= radv_tex_aniso_filter(max_aniso
);
4575 bool is_vi
= (device
->physical_device
->rad_info
.chip_class
>= VI
);
4576 unsigned filter_mode
= V_008F30_SQ_IMG_FILTER_MODE_BLEND
;
4578 const struct VkSamplerReductionModeCreateInfoEXT
*sampler_reduction
=
4579 vk_find_struct_const(pCreateInfo
->pNext
,
4580 SAMPLER_REDUCTION_MODE_CREATE_INFO_EXT
);
4581 if (sampler_reduction
)
4582 filter_mode
= radv_tex_filter_mode(sampler_reduction
->reductionMode
);
4584 sampler
->state
[0] = (S_008F30_CLAMP_X(radv_tex_wrap(pCreateInfo
->addressModeU
)) |
4585 S_008F30_CLAMP_Y(radv_tex_wrap(pCreateInfo
->addressModeV
)) |
4586 S_008F30_CLAMP_Z(radv_tex_wrap(pCreateInfo
->addressModeW
)) |
4587 S_008F30_MAX_ANISO_RATIO(max_aniso_ratio
) |
4588 S_008F30_DEPTH_COMPARE_FUNC(radv_tex_compare(pCreateInfo
->compareOp
)) |
4589 S_008F30_FORCE_UNNORMALIZED(pCreateInfo
->unnormalizedCoordinates
? 1 : 0) |
4590 S_008F30_ANISO_THRESHOLD(max_aniso_ratio
>> 1) |
4591 S_008F30_ANISO_BIAS(max_aniso_ratio
) |
4592 S_008F30_DISABLE_CUBE_WRAP(0) |
4593 S_008F30_COMPAT_MODE(is_vi
) |
4594 S_008F30_FILTER_MODE(filter_mode
));
4595 sampler
->state
[1] = (S_008F34_MIN_LOD(S_FIXED(CLAMP(pCreateInfo
->minLod
, 0, 15), 8)) |
4596 S_008F34_MAX_LOD(S_FIXED(CLAMP(pCreateInfo
->maxLod
, 0, 15), 8)) |
4597 S_008F34_PERF_MIP(max_aniso_ratio
? max_aniso_ratio
+ 6 : 0));
4598 sampler
->state
[2] = (S_008F38_LOD_BIAS(S_FIXED(CLAMP(pCreateInfo
->mipLodBias
, -16, 16), 8)) |
4599 S_008F38_XY_MAG_FILTER(radv_tex_filter(pCreateInfo
->magFilter
, max_aniso
)) |
4600 S_008F38_XY_MIN_FILTER(radv_tex_filter(pCreateInfo
->minFilter
, max_aniso
)) |
4601 S_008F38_MIP_FILTER(radv_tex_mipfilter(pCreateInfo
->mipmapMode
)) |
4602 S_008F38_MIP_POINT_PRECLAMP(0) |
4603 S_008F38_DISABLE_LSB_CEIL(device
->physical_device
->rad_info
.chip_class
<= VI
) |
4604 S_008F38_FILTER_PREC_FIX(1) |
4605 S_008F38_ANISO_OVERRIDE(is_vi
));
4606 sampler
->state
[3] = (S_008F3C_BORDER_COLOR_PTR(0) |
4607 S_008F3C_BORDER_COLOR_TYPE(radv_tex_bordercolor(pCreateInfo
->borderColor
)));
4610 VkResult
radv_CreateSampler(
4612 const VkSamplerCreateInfo
* pCreateInfo
,
4613 const VkAllocationCallbacks
* pAllocator
,
4614 VkSampler
* pSampler
)
4616 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4617 struct radv_sampler
*sampler
;
4619 assert(pCreateInfo
->sType
== VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO
);
4621 sampler
= vk_alloc2(&device
->alloc
, pAllocator
, sizeof(*sampler
), 8,
4622 VK_SYSTEM_ALLOCATION_SCOPE_OBJECT
);
4624 return vk_error(device
->instance
, VK_ERROR_OUT_OF_HOST_MEMORY
);
4626 radv_init_sampler(device
, sampler
, pCreateInfo
);
4627 *pSampler
= radv_sampler_to_handle(sampler
);
4632 void radv_DestroySampler(
4635 const VkAllocationCallbacks
* pAllocator
)
4637 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4638 RADV_FROM_HANDLE(radv_sampler
, sampler
, _sampler
);
4642 vk_free2(&device
->alloc
, pAllocator
, sampler
);
4645 /* vk_icd.h does not declare this function, so we declare it here to
4646 * suppress Wmissing-prototypes.
4648 PUBLIC VKAPI_ATTR VkResult VKAPI_CALL
4649 vk_icdNegotiateLoaderICDInterfaceVersion(uint32_t *pSupportedVersion
);
4651 PUBLIC VKAPI_ATTR VkResult VKAPI_CALL
4652 vk_icdNegotiateLoaderICDInterfaceVersion(uint32_t *pSupportedVersion
)
4654 /* For the full details on loader interface versioning, see
4655 * <https://github.com/KhronosGroup/Vulkan-LoaderAndValidationLayers/blob/master/loader/LoaderAndLayerInterface.md>.
4656 * What follows is a condensed summary, to help you navigate the large and
4657 * confusing official doc.
4659 * - Loader interface v0 is incompatible with later versions. We don't
4662 * - In loader interface v1:
4663 * - The first ICD entrypoint called by the loader is
4664 * vk_icdGetInstanceProcAddr(). The ICD must statically expose this
4666 * - The ICD must statically expose no other Vulkan symbol unless it is
4667 * linked with -Bsymbolic.
4668 * - Each dispatchable Vulkan handle created by the ICD must be
4669 * a pointer to a struct whose first member is VK_LOADER_DATA. The
4670 * ICD must initialize VK_LOADER_DATA.loadMagic to ICD_LOADER_MAGIC.
4671 * - The loader implements vkCreate{PLATFORM}SurfaceKHR() and
4672 * vkDestroySurfaceKHR(). The ICD must be capable of working with
4673 * such loader-managed surfaces.
4675 * - Loader interface v2 differs from v1 in:
4676 * - The first ICD entrypoint called by the loader is
4677 * vk_icdNegotiateLoaderICDInterfaceVersion(). The ICD must
4678 * statically expose this entrypoint.
4680 * - Loader interface v3 differs from v2 in:
4681 * - The ICD must implement vkCreate{PLATFORM}SurfaceKHR(),
4682 * vkDestroySurfaceKHR(), and other API which uses VKSurfaceKHR,
4683 * because the loader no longer does so.
4685 *pSupportedVersion
= MIN2(*pSupportedVersion
, 3u);
4689 VkResult
radv_GetMemoryFdKHR(VkDevice _device
,
4690 const VkMemoryGetFdInfoKHR
*pGetFdInfo
,
4693 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4694 RADV_FROM_HANDLE(radv_device_memory
, memory
, pGetFdInfo
->memory
);
4696 assert(pGetFdInfo
->sType
== VK_STRUCTURE_TYPE_MEMORY_GET_FD_INFO_KHR
);
4698 /* At the moment, we support only the below handle types. */
4699 assert(pGetFdInfo
->handleType
==
4700 VK_EXTERNAL_MEMORY_HANDLE_TYPE_OPAQUE_FD_BIT
||
4701 pGetFdInfo
->handleType
==
4702 VK_EXTERNAL_MEMORY_HANDLE_TYPE_DMA_BUF_BIT_EXT
);
4704 bool ret
= radv_get_memory_fd(device
, memory
, pFD
);
4706 return vk_error(device
->instance
, VK_ERROR_OUT_OF_DEVICE_MEMORY
);
4710 VkResult
radv_GetMemoryFdPropertiesKHR(VkDevice _device
,
4711 VkExternalMemoryHandleTypeFlagBits handleType
,
4713 VkMemoryFdPropertiesKHR
*pMemoryFdProperties
)
4715 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4717 switch (handleType
) {
4718 case VK_EXTERNAL_MEMORY_HANDLE_TYPE_DMA_BUF_BIT_EXT
:
4719 pMemoryFdProperties
->memoryTypeBits
= (1 << RADV_MEM_TYPE_COUNT
) - 1;
4723 /* The valid usage section for this function says:
4725 * "handleType must not be one of the handle types defined as
4728 * So opaque handle types fall into the default "unsupported" case.
4730 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
4734 static VkResult
radv_import_opaque_fd(struct radv_device
*device
,
4738 uint32_t syncobj_handle
= 0;
4739 int ret
= device
->ws
->import_syncobj(device
->ws
, fd
, &syncobj_handle
);
4741 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
4744 device
->ws
->destroy_syncobj(device
->ws
, *syncobj
);
4746 *syncobj
= syncobj_handle
;
4752 static VkResult
radv_import_sync_fd(struct radv_device
*device
,
4756 /* If we create a syncobj we do it locally so that if we have an error, we don't
4757 * leave a syncobj in an undetermined state in the fence. */
4758 uint32_t syncobj_handle
= *syncobj
;
4759 if (!syncobj_handle
) {
4760 int ret
= device
->ws
->create_syncobj(device
->ws
, &syncobj_handle
);
4762 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
4767 device
->ws
->signal_syncobj(device
->ws
, syncobj_handle
);
4769 int ret
= device
->ws
->import_syncobj_from_sync_file(device
->ws
, syncobj_handle
, fd
);
4771 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
4774 *syncobj
= syncobj_handle
;
4781 VkResult
radv_ImportSemaphoreFdKHR(VkDevice _device
,
4782 const VkImportSemaphoreFdInfoKHR
*pImportSemaphoreFdInfo
)
4784 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4785 RADV_FROM_HANDLE(radv_semaphore
, sem
, pImportSemaphoreFdInfo
->semaphore
);
4786 uint32_t *syncobj_dst
= NULL
;
4788 if (pImportSemaphoreFdInfo
->flags
& VK_SEMAPHORE_IMPORT_TEMPORARY_BIT
) {
4789 syncobj_dst
= &sem
->temp_syncobj
;
4791 syncobj_dst
= &sem
->syncobj
;
4794 switch(pImportSemaphoreFdInfo
->handleType
) {
4795 case VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
:
4796 return radv_import_opaque_fd(device
, pImportSemaphoreFdInfo
->fd
, syncobj_dst
);
4797 case VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_SYNC_FD_BIT
:
4798 return radv_import_sync_fd(device
, pImportSemaphoreFdInfo
->fd
, syncobj_dst
);
4800 unreachable("Unhandled semaphore handle type");
4804 VkResult
radv_GetSemaphoreFdKHR(VkDevice _device
,
4805 const VkSemaphoreGetFdInfoKHR
*pGetFdInfo
,
4808 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4809 RADV_FROM_HANDLE(radv_semaphore
, sem
, pGetFdInfo
->semaphore
);
4811 uint32_t syncobj_handle
;
4813 if (sem
->temp_syncobj
)
4814 syncobj_handle
= sem
->temp_syncobj
;
4816 syncobj_handle
= sem
->syncobj
;
4818 switch(pGetFdInfo
->handleType
) {
4819 case VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
:
4820 ret
= device
->ws
->export_syncobj(device
->ws
, syncobj_handle
, pFd
);
4822 case VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_SYNC_FD_BIT
:
4823 ret
= device
->ws
->export_syncobj_to_sync_file(device
->ws
, syncobj_handle
, pFd
);
4825 if (sem
->temp_syncobj
) {
4826 close (sem
->temp_syncobj
);
4827 sem
->temp_syncobj
= 0;
4829 device
->ws
->reset_syncobj(device
->ws
, syncobj_handle
);
4834 unreachable("Unhandled semaphore handle type");
4838 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
4842 void radv_GetPhysicalDeviceExternalSemaphoreProperties(
4843 VkPhysicalDevice physicalDevice
,
4844 const VkPhysicalDeviceExternalSemaphoreInfo
*pExternalSemaphoreInfo
,
4845 VkExternalSemaphoreProperties
*pExternalSemaphoreProperties
)
4847 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
4849 /* Require has_syncobj_wait_for_submit for the syncobj signal ioctl introduced at virtually the same time */
4850 if (pdevice
->rad_info
.has_syncobj_wait_for_submit
&&
4851 (pExternalSemaphoreInfo
->handleType
== VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
||
4852 pExternalSemaphoreInfo
->handleType
== VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_SYNC_FD_BIT
)) {
4853 pExternalSemaphoreProperties
->exportFromImportedHandleTypes
= VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
| VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_SYNC_FD_BIT
;
4854 pExternalSemaphoreProperties
->compatibleHandleTypes
= VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
| VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_SYNC_FD_BIT
;
4855 pExternalSemaphoreProperties
->externalSemaphoreFeatures
= VK_EXTERNAL_SEMAPHORE_FEATURE_EXPORTABLE_BIT
|
4856 VK_EXTERNAL_SEMAPHORE_FEATURE_IMPORTABLE_BIT
;
4857 } else if (pExternalSemaphoreInfo
->handleType
== VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
) {
4858 pExternalSemaphoreProperties
->exportFromImportedHandleTypes
= VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
;
4859 pExternalSemaphoreProperties
->compatibleHandleTypes
= VK_EXTERNAL_SEMAPHORE_HANDLE_TYPE_OPAQUE_FD_BIT
;
4860 pExternalSemaphoreProperties
->externalSemaphoreFeatures
= VK_EXTERNAL_SEMAPHORE_FEATURE_EXPORTABLE_BIT
|
4861 VK_EXTERNAL_SEMAPHORE_FEATURE_IMPORTABLE_BIT
;
4863 pExternalSemaphoreProperties
->exportFromImportedHandleTypes
= 0;
4864 pExternalSemaphoreProperties
->compatibleHandleTypes
= 0;
4865 pExternalSemaphoreProperties
->externalSemaphoreFeatures
= 0;
4869 VkResult
radv_ImportFenceFdKHR(VkDevice _device
,
4870 const VkImportFenceFdInfoKHR
*pImportFenceFdInfo
)
4872 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4873 RADV_FROM_HANDLE(radv_fence
, fence
, pImportFenceFdInfo
->fence
);
4874 uint32_t *syncobj_dst
= NULL
;
4877 if (pImportFenceFdInfo
->flags
& VK_FENCE_IMPORT_TEMPORARY_BIT
) {
4878 syncobj_dst
= &fence
->temp_syncobj
;
4880 syncobj_dst
= &fence
->syncobj
;
4883 switch(pImportFenceFdInfo
->handleType
) {
4884 case VK_EXTERNAL_FENCE_HANDLE_TYPE_OPAQUE_FD_BIT
:
4885 return radv_import_opaque_fd(device
, pImportFenceFdInfo
->fd
, syncobj_dst
);
4886 case VK_EXTERNAL_FENCE_HANDLE_TYPE_SYNC_FD_BIT
:
4887 return radv_import_sync_fd(device
, pImportFenceFdInfo
->fd
, syncobj_dst
);
4889 unreachable("Unhandled fence handle type");
4893 VkResult
radv_GetFenceFdKHR(VkDevice _device
,
4894 const VkFenceGetFdInfoKHR
*pGetFdInfo
,
4897 RADV_FROM_HANDLE(radv_device
, device
, _device
);
4898 RADV_FROM_HANDLE(radv_fence
, fence
, pGetFdInfo
->fence
);
4900 uint32_t syncobj_handle
;
4902 if (fence
->temp_syncobj
)
4903 syncobj_handle
= fence
->temp_syncobj
;
4905 syncobj_handle
= fence
->syncobj
;
4907 switch(pGetFdInfo
->handleType
) {
4908 case VK_EXTERNAL_FENCE_HANDLE_TYPE_OPAQUE_FD_BIT
:
4909 ret
= device
->ws
->export_syncobj(device
->ws
, syncobj_handle
, pFd
);
4911 case VK_EXTERNAL_FENCE_HANDLE_TYPE_SYNC_FD_BIT
:
4912 ret
= device
->ws
->export_syncobj_to_sync_file(device
->ws
, syncobj_handle
, pFd
);
4914 if (fence
->temp_syncobj
) {
4915 close (fence
->temp_syncobj
);
4916 fence
->temp_syncobj
= 0;
4918 device
->ws
->reset_syncobj(device
->ws
, syncobj_handle
);
4923 unreachable("Unhandled fence handle type");
4927 return vk_error(device
->instance
, VK_ERROR_INVALID_EXTERNAL_HANDLE
);
4931 void radv_GetPhysicalDeviceExternalFenceProperties(
4932 VkPhysicalDevice physicalDevice
,
4933 const VkPhysicalDeviceExternalFenceInfo
*pExternalFenceInfo
,
4934 VkExternalFenceProperties
*pExternalFenceProperties
)
4936 RADV_FROM_HANDLE(radv_physical_device
, pdevice
, physicalDevice
);
4938 if (pdevice
->rad_info
.has_syncobj_wait_for_submit
&&
4939 (pExternalFenceInfo
->handleType
== VK_EXTERNAL_FENCE_HANDLE_TYPE_OPAQUE_FD_BIT
||
4940 pExternalFenceInfo
->handleType
== VK_EXTERNAL_FENCE_HANDLE_TYPE_SYNC_FD_BIT
)) {
4941 pExternalFenceProperties
->exportFromImportedHandleTypes
= VK_EXTERNAL_FENCE_HANDLE_TYPE_OPAQUE_FD_BIT
| VK_EXTERNAL_FENCE_HANDLE_TYPE_SYNC_FD_BIT
;
4942 pExternalFenceProperties
->compatibleHandleTypes
= VK_EXTERNAL_FENCE_HANDLE_TYPE_OPAQUE_FD_BIT
| VK_EXTERNAL_FENCE_HANDLE_TYPE_SYNC_FD_BIT
;
4943 pExternalFenceProperties
->externalFenceFeatures
= VK_EXTERNAL_FENCE_FEATURE_EXPORTABLE_BIT
|
4944 VK_EXTERNAL_SEMAPHORE_FEATURE_IMPORTABLE_BIT
;
4946 pExternalFenceProperties
->exportFromImportedHandleTypes
= 0;
4947 pExternalFenceProperties
->compatibleHandleTypes
= 0;
4948 pExternalFenceProperties
->externalFenceFeatures
= 0;
4953 radv_CreateDebugReportCallbackEXT(VkInstance _instance
,
4954 const VkDebugReportCallbackCreateInfoEXT
* pCreateInfo
,
4955 const VkAllocationCallbacks
* pAllocator
,
4956 VkDebugReportCallbackEXT
* pCallback
)
4958 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
4959 return vk_create_debug_report_callback(&instance
->debug_report_callbacks
,
4960 pCreateInfo
, pAllocator
, &instance
->alloc
,
4965 radv_DestroyDebugReportCallbackEXT(VkInstance _instance
,
4966 VkDebugReportCallbackEXT _callback
,
4967 const VkAllocationCallbacks
* pAllocator
)
4969 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
4970 vk_destroy_debug_report_callback(&instance
->debug_report_callbacks
,
4971 _callback
, pAllocator
, &instance
->alloc
);
4975 radv_DebugReportMessageEXT(VkInstance _instance
,
4976 VkDebugReportFlagsEXT flags
,
4977 VkDebugReportObjectTypeEXT objectType
,
4980 int32_t messageCode
,
4981 const char* pLayerPrefix
,
4982 const char* pMessage
)
4984 RADV_FROM_HANDLE(radv_instance
, instance
, _instance
);
4985 vk_debug_report(&instance
->debug_report_callbacks
, flags
, objectType
,
4986 object
, location
, messageCode
, pLayerPrefix
, pMessage
);
4990 radv_GetDeviceGroupPeerMemoryFeatures(
4993 uint32_t localDeviceIndex
,
4994 uint32_t remoteDeviceIndex
,
4995 VkPeerMemoryFeatureFlags
* pPeerMemoryFeatures
)
4997 assert(localDeviceIndex
== remoteDeviceIndex
);
4999 *pPeerMemoryFeatures
= VK_PEER_MEMORY_FEATURE_COPY_SRC_BIT
|
5000 VK_PEER_MEMORY_FEATURE_COPY_DST_BIT
|
5001 VK_PEER_MEMORY_FEATURE_GENERIC_SRC_BIT
|
5002 VK_PEER_MEMORY_FEATURE_GENERIC_DST_BIT
;
5005 static const VkTimeDomainEXT radv_time_domains
[] = {
5006 VK_TIME_DOMAIN_DEVICE_EXT
,
5007 VK_TIME_DOMAIN_CLOCK_MONOTONIC_EXT
,
5008 VK_TIME_DOMAIN_CLOCK_MONOTONIC_RAW_EXT
,
5011 VkResult
radv_GetPhysicalDeviceCalibrateableTimeDomainsEXT(
5012 VkPhysicalDevice physicalDevice
,
5013 uint32_t *pTimeDomainCount
,
5014 VkTimeDomainEXT
*pTimeDomains
)
5017 VK_OUTARRAY_MAKE(out
, pTimeDomains
, pTimeDomainCount
);
5019 for (d
= 0; d
< ARRAY_SIZE(radv_time_domains
); d
++) {
5020 vk_outarray_append(&out
, i
) {
5021 *i
= radv_time_domains
[d
];
5025 return vk_outarray_status(&out
);
5029 radv_clock_gettime(clockid_t clock_id
)
5031 struct timespec current
;
5034 ret
= clock_gettime(clock_id
, ¤t
);
5035 if (ret
< 0 && clock_id
== CLOCK_MONOTONIC_RAW
)
5036 ret
= clock_gettime(CLOCK_MONOTONIC
, ¤t
);
5040 return (uint64_t) current
.tv_sec
* 1000000000ULL + current
.tv_nsec
;
5043 VkResult
radv_GetCalibratedTimestampsEXT(
5045 uint32_t timestampCount
,
5046 const VkCalibratedTimestampInfoEXT
*pTimestampInfos
,
5047 uint64_t *pTimestamps
,
5048 uint64_t *pMaxDeviation
)
5050 RADV_FROM_HANDLE(radv_device
, device
, _device
);
5051 uint32_t clock_crystal_freq
= device
->physical_device
->rad_info
.clock_crystal_freq
;
5053 uint64_t begin
, end
;
5054 uint64_t max_clock_period
= 0;
5056 begin
= radv_clock_gettime(CLOCK_MONOTONIC_RAW
);
5058 for (d
= 0; d
< timestampCount
; d
++) {
5059 switch (pTimestampInfos
[d
].timeDomain
) {
5060 case VK_TIME_DOMAIN_DEVICE_EXT
:
5061 pTimestamps
[d
] = device
->ws
->query_value(device
->ws
,
5063 uint64_t device_period
= DIV_ROUND_UP(1000000, clock_crystal_freq
);
5064 max_clock_period
= MAX2(max_clock_period
, device_period
);
5066 case VK_TIME_DOMAIN_CLOCK_MONOTONIC_EXT
:
5067 pTimestamps
[d
] = radv_clock_gettime(CLOCK_MONOTONIC
);
5068 max_clock_period
= MAX2(max_clock_period
, 1);
5071 case VK_TIME_DOMAIN_CLOCK_MONOTONIC_RAW_EXT
:
5072 pTimestamps
[d
] = begin
;
5080 end
= radv_clock_gettime(CLOCK_MONOTONIC_RAW
);
5083 * The maximum deviation is the sum of the interval over which we
5084 * perform the sampling and the maximum period of any sampled
5085 * clock. That's because the maximum skew between any two sampled
5086 * clock edges is when the sampled clock with the largest period is
5087 * sampled at the end of that period but right at the beginning of the
5088 * sampling interval and some other clock is sampled right at the
5089 * begining of its sampling period and right at the end of the
5090 * sampling interval. Let's assume the GPU has the longest clock
5091 * period and that the application is sampling GPU and monotonic:
5094 * w x y z 0 1 2 3 4 5 6 7 8 9 a b c d e f
5095 * Raw -_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-
5099 * GPU -----_____-----_____-----_____-----_____
5102 * x y z 0 1 2 3 4 5 6 7 8 9 a b c
5103 * Monotonic -_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-
5105 * Interval <----------------->
5106 * Deviation <-------------------------->
5110 * m = read(monotonic) 2
5113 * We round the sample interval up by one tick to cover sampling error
5114 * in the interval clock
5117 uint64_t sample_interval
= end
- begin
+ 1;
5119 *pMaxDeviation
= sample_interval
+ max_clock_period
;