summaryrefslogtreecommitdiff
path: root/src/broadcom/vulkan/v3dv_private.h
blob: b8ed1ae74e4e179a69c444e609beeb7cf1cfab15 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
/*
 * Copyright © 2019 Raspberry Pi
 *
 * based in part on anv driver which is:
 * Copyright © 2015 Intel Corporation
 *
 * based in part on radv driver which is:
 * Copyright © 2016 Red Hat.
 * Copyright © 2016 Bas Nieuwenhuizen
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the next
 * paragraph) shall be included in all copies or substantial portions of the
 * Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
 * IN THE SOFTWARE.
 */
#ifndef V3DV_PRIVATE_H
#define V3DV_PRIVATE_H

#include <stdio.h>
#include <string.h>
#include <vulkan/vulkan.h>
#include <vulkan/vk_icd.h>
#include <vk_enum_to_str.h>

#include <xf86drm.h>

#ifdef HAVE_VALGRIND
#include <valgrind/valgrind.h>
#include <valgrind/memcheck.h>
#define VG(x) x
#else
#define VG(x) ((void)0)
#endif

#include "common/v3d_device_info.h"
#include "common/v3d_limits.h"

#include "compiler/shader_enums.h"
#include "compiler/spirv/nir_spirv.h"

#include "compiler/v3d_compiler.h"

#include "vk_debug_report.h"
#include "util/set.h"
#include "util/hash_table.h"
#include "util/xmlconfig.h"

#include "v3dv_entrypoints.h"
#include "v3dv_extensions.h"
#include "v3dv_bo.h"

#include "drm-uapi/v3d_drm.h"

/* FIXME: hooks for the packet definition functions. */
static inline void
pack_emit_reloc(void *cl, const void *reloc) {}

#define __gen_user_data struct v3dv_cl
#define __gen_address_type struct v3dv_cl_reloc
#define __gen_address_offset(reloc) (((reloc)->bo ? (reloc)->bo->offset : 0) + \
                                     (reloc)->offset)
#define __gen_emit_reloc cl_pack_emit_reloc
#define __gen_unpack_address(cl, s, e) __unpack_address(cl, s, e)
#include "v3dv_cl.h"

#include "vk_alloc.h"
#include "simulator/v3d_simulator.h"


/* FIXME: pipe_box from Gallium. Needed for some v3d_tiling.c functions.
 * In the future we might want to drop that depedency, but for now it is
 * good enough.
 */
#include "util/u_box.h"
#include "wsi_common.h"

#include "broadcom/cle/v3dx_pack.h"

/* A non-fatal assert.  Useful for debugging. */
#ifdef DEBUG
#define v3dv_assert(x) ({ \
   if (unlikely(!(x))) \
      fprintf(stderr, "%s:%d ASSERT: %s", __FILE__, __LINE__, #x); \
})
#else
#define v3dv_assert(x)
#endif

#define perf_debug(...) do {                       \
   if (unlikely(V3D_DEBUG & V3D_DEBUG_PERF))       \
      fprintf(stderr, __VA_ARGS__);                \
} while (0)

#define for_each_bit(b, dword)                                               \
   for (uint32_t __dword = (dword);                                          \
        (b) = __builtin_ffs(__dword) - 1, __dword; __dword &= ~(1 << (b)))

#define typed_memcpy(dest, src, count) ({				\
			STATIC_ASSERT(sizeof(*src) == sizeof(*dest)); \
			memcpy((dest), (src), (count) * sizeof(*(src))); \
		})

#define NSEC_PER_SEC 1000000000ull

/* From vulkan spec "If the multiple viewports feature is not enabled,
 * scissorCount must be 1", ditto for viewportCount. For now we don't support
 * that feature.
 */
#define MAX_VIEWPORTS 1
#define MAX_SCISSORS  1

#define MAX_VBS 16
#define MAX_VERTEX_ATTRIBS 16

#define MAX_SETS 16

#define MAX_PUSH_CONSTANTS_SIZE 128

#define MAX_DYNAMIC_UNIFORM_BUFFERS 16
#define MAX_DYNAMIC_STORAGE_BUFFERS 8
#define MAX_DYNAMIC_BUFFERS                                                  \
   (MAX_DYNAMIC_UNIFORM_BUFFERS + MAX_DYNAMIC_STORAGE_BUFFERS)

struct v3dv_instance;

#ifdef USE_V3D_SIMULATOR
#define using_v3d_simulator true
#else
#define using_v3d_simulator false
#endif

struct v3d_simulator_file;

struct v3dv_physical_device {
   VK_LOADER_DATA _loader_data;

   struct v3dv_instance *instance;

   struct v3dv_device_extension_table supported_extensions;
   struct v3dv_physical_device_dispatch_table dispatch;

   char *name;
   int32_t render_fd;
   int32_t display_fd;
   uint8_t pipeline_cache_uuid[VK_UUID_SIZE];

   struct wsi_device wsi_device;

   VkPhysicalDeviceMemoryProperties memory;

   struct v3d_device_info devinfo;

   struct v3d_simulator_file *sim_file;

   const struct v3d_compiler *compiler;
   uint32_t next_program_id;

   struct {
      bool merge_jobs;
   } options;
};

VkResult v3dv_wsi_init(struct v3dv_physical_device *physical_device);
void v3dv_wsi_finish(struct v3dv_physical_device *physical_device);

struct v3dv_app_info {
   const char *app_name;
   uint32_t app_version;
   const char *engine_name;
   uint32_t engine_version;
   uint32_t api_version;
};

struct v3dv_instance {
   VK_LOADER_DATA _loader_data;

   VkAllocationCallbacks alloc;

   struct v3dv_app_info app_info;

   struct v3dv_instance_extension_table enabled_extensions;
   struct v3dv_instance_dispatch_table dispatch;
   struct v3dv_device_dispatch_table device_dispatch;

   int physicalDeviceCount;
   struct v3dv_physical_device physicalDevice;

   struct vk_debug_report_instance debug_report_callbacks;
};

/* Tracks wait threads spawned from a single vkQueueSubmit call */
struct v3dv_queue_submit_wait_info {
   struct list_head list_link;

   struct v3dv_device *device;

   /* List of wait threads spawned for any command buffers in a particular
    * call to vkQueueSubmit.
    */
   uint32_t wait_thread_count;
   struct {
      pthread_t thread;
      bool finished;
   } wait_threads[16];

   /* The master wait thread for the entire submit. This will wait for all
    * other threads in this submit to complete  before processing signal
    * semaphores and fences.
    */
   pthread_t master_wait_thread;

   /* List of semaphores (and fence) to signal after all wait threads completed
    * and all command buffer jobs in the submission have been sent to the GPU.
    */
   uint32_t signal_semaphore_count;
   VkSemaphore *signal_semaphores;
   VkFence fence;
};

struct v3dv_queue {
   VK_LOADER_DATA _loader_data;

   struct v3dv_device *device;
   VkDeviceQueueCreateFlags flags;

   /* A list of active v3dv_queue_submit_wait_info */
   struct list_head submit_wait_list;

   /* A mutex to prevent concurrent access to the list of wait threads */
   mtx_t mutex;
};

struct v3dv_meta_color_clear_pipeline {
   VkPipeline pipeline;
   VkRenderPass pass;
};

struct v3dv_meta_blit_pipeline {
   VkPipeline pipeline;
   VkRenderPass pass;
};

#define V3DV_META_BLIT_CACHE_KEY_SIZE (3 * sizeof(uint32_t))

struct v3dv_device {
   VK_LOADER_DATA _loader_data;

   VkAllocationCallbacks alloc;

   struct v3dv_instance *instance;

   struct v3dv_device_extension_table enabled_extensions;
   struct v3dv_device_dispatch_table dispatch;

   int32_t render_fd;
   int32_t display_fd;
   struct v3d_device_info devinfo;
   struct v3dv_queue queue;

   /* A sync object to track the last job submitted to the GPU. */
   uint32_t last_job_sync;

   /* A mutex to prevent concurrent access to last_job_sync from the queue */
   mtx_t mutex;

   /* Resources used for meta operations */
   struct {
      mtx_t mtx;
      struct {
         VkPipelineLayout playout;
         struct hash_table *cache; /* v3dv_meta_color_clear_pipeline */
      } color_clear;
      struct {
         VkDescriptorSetLayout dslayout;
         VkPipelineLayout playout;
         struct hash_table *cache[3]; /* v3dv_meta_blit_pipeline for 1d, 2d, 3d */
      } blit;
   } meta;
};

struct v3dv_device_memory {
   struct v3dv_bo *bo;
   const VkMemoryType *type;
};

#define V3D_OUTPUT_IMAGE_FORMAT_NO 255
#define TEXTURE_DATA_FORMAT_NO     255

struct v3dv_format {
   bool supported;

   /* One of V3D33_OUTPUT_IMAGE_FORMAT_*, or OUTPUT_IMAGE_FORMAT_NO */
   uint8_t rt_type;

   /* One of V3D33_TEXTURE_DATA_FORMAT_*. */
   uint8_t tex_type;

   /* Swizzle to apply to the RGBA shader output for storing to the tile
    * buffer, to the RGBA tile buffer to produce shader input (for
    * blending), and for turning the rgba8888 texture sampler return
    * value into shader rgba values.
    */
   uint8_t swizzle[4];

   /* Whether the return value is 16F/I/UI or 32F/I/UI. */
   uint8_t return_size;

   /* If the format supports (linear) filtering when texturing. */
   bool supports_filtering;
};

/**
 * Tiling mode enum used for v3d_resource.c, which maps directly to the Memory
 * Format field of render target and Z/Stencil config.
 */
enum v3d_tiling_mode {
   /* Untiled resources.  Not valid as texture inputs. */
   VC5_TILING_RASTER,

   /* Single line of u-tiles. */
   VC5_TILING_LINEARTILE,

   /* Departure from standard 4-UIF block column format. */
   VC5_TILING_UBLINEAR_1_COLUMN,

   /* Departure from standard 4-UIF block column format. */
   VC5_TILING_UBLINEAR_2_COLUMN,

   /* Normal tiling format: grouped in 4x4 UIFblocks, each of which is
    * split 2x2 into utiles.
    */
   VC5_TILING_UIF_NO_XOR,

   /* Normal tiling format: grouped in 4x4 UIFblocks, each of which is
    * split 2x2 into utiles.
    */
   VC5_TILING_UIF_XOR,
};

struct v3d_resource_slice {
   uint32_t offset;
   uint32_t stride;
   uint32_t padded_height;
   /* Size of a single pane of the slice.  For 3D textures, there will be
    * a number of panes equal to the minified, power-of-two-aligned
    * depth.
    */
   uint32_t size;
   uint8_t ub_pad;
   enum v3d_tiling_mode tiling;
   uint32_t padded_height_of_output_image_in_uif_blocks;
};

struct v3dv_image {
   VkImageType type;
   VkImageAspectFlags aspects;

   VkExtent3D extent;
   uint32_t levels;
   uint32_t array_size;
   uint32_t samples;
   VkImageUsageFlags usage;
   VkImageCreateFlags create_flags;
   VkImageTiling tiling;

   VkFormat vk_format;
   const struct v3dv_format *format;

   uint32_t cpp;

   uint64_t drm_format_mod;
   bool tiled;

   struct v3d_resource_slice slices[V3D_MAX_MIP_LEVELS];
   uint32_t size; /* Total size in bytes */
   uint32_t cube_map_stride;
   uint32_t alignment;

   struct v3dv_device_memory *mem;
   VkDeviceSize mem_offset;
};

VkImageViewType v3dv_image_type_to_view_type(VkImageType type);

struct v3dv_image_view {
   const struct v3dv_image *image;
   VkImageAspectFlags aspects;
   VkExtent3D extent;

   VkFormat vk_format;
   const struct v3dv_format *format;
   bool swap_rb;
   uint32_t internal_bpp;
   uint32_t internal_type;

   uint32_t base_level;
   uint32_t max_level;
   uint32_t first_layer;
   uint32_t last_layer;
   uint32_t offset;

   /* Precomputed (composed from createinfo->components and formar swizzle)
    * swizzles to pass in to the shader key.
    *
    * FIXME: this is also a candidate to be included on the descriptor info.
    */
   uint8_t swizzle[4];

   /* FIXME: here we store the packet TEXTURE_SHADER_STATE, that is referenced
    * as part of the tmu configuration, and the content is set per sampler. A
    * possible perf improvement, to avoid bo fragmentation, would be to save
    * the state as static, have the bo as part of the descriptor (booked from
    * the descriptor pools), and then copy this content to the descriptor bo
    * on UpdateDescriptor. This also makes sense because not all the images
    * are used as textures.
    */
   struct v3dv_bo *texture_shader_state;
};

uint32_t v3dv_layer_offset(const struct v3dv_image *image, uint32_t level, uint32_t layer);

struct v3dv_buffer {
   VkDeviceSize size;
   VkBufferUsageFlags usage;
   uint32_t alignment;

   struct v3dv_device_memory *mem;
   VkDeviceSize mem_offset;
};

struct v3dv_buffer_view {
   const struct v3dv_buffer *buffer;

   VkFormat vk_format;
   const struct v3dv_format *format;
   uint32_t internal_bpp;
   uint32_t internal_type;

   uint32_t offset;
   uint32_t size;
   uint32_t num_elements;
};

struct v3dv_subpass_attachment {
   uint32_t attachment;
   VkImageLayout layout;
};

struct v3dv_subpass {
   uint32_t input_count;
   struct v3dv_subpass_attachment *input_attachments;

   uint32_t color_count;
   struct v3dv_subpass_attachment *color_attachments;
   struct v3dv_subpass_attachment *resolve_attachments;

   struct v3dv_subpass_attachment ds_attachment;

   bool has_srgb_rt;
};

struct v3dv_render_pass_attachment {
   VkAttachmentDescription desc;
   uint32_t first_subpass;
   uint32_t last_subpass;
};

struct v3dv_render_pass {
   uint32_t attachment_count;
   struct v3dv_render_pass_attachment *attachments;

   uint32_t subpass_count;
   struct v3dv_subpass *subpasses;

   struct v3dv_subpass_attachment *subpass_attachments;
};

void v3dv_subpass_get_granularity(struct v3dv_render_pass *pass,
                                  uint32_t subpass_idx,
                                  VkExtent2D *granularity);

struct v3dv_framebuffer {
   uint32_t width;
   uint32_t height;
   uint32_t layers;

   uint32_t attachment_count;
   uint32_t color_attachment_count;
   struct v3dv_image_view *attachments[0];
};

struct v3dv_frame_tiling {
   uint32_t width;
   uint32_t height;
   uint32_t render_target_count;
   uint32_t internal_bpp;
   uint32_t layers;
   uint32_t tile_width;
   uint32_t tile_height;
   uint32_t draw_tiles_x;
   uint32_t draw_tiles_y;
   uint32_t supertile_width;
   uint32_t supertile_height;
   uint32_t frame_width_in_supertiles;
   uint32_t frame_height_in_supertiles;
};

uint8_t v3dv_framebuffer_compute_internal_bpp(const struct v3dv_framebuffer *framebuffer,
                                              const struct v3dv_subpass *subpass);

struct v3dv_cmd_pool {
   VkAllocationCallbacks alloc;
   struct list_head cmd_buffers;
};

enum v3dv_cmd_buffer_status {
   V3DV_CMD_BUFFER_STATUS_NEW           = 0,
   V3DV_CMD_BUFFER_STATUS_INITIALIZED   = 1,
   V3DV_CMD_BUFFER_STATUS_RECORDING     = 2,
   V3DV_CMD_BUFFER_STATUS_EXECUTABLE    = 3
};

union v3dv_clear_value {
   uint32_t color[4];
   struct {
      float z;
      uint8_t s;
   };
};

struct v3dv_cmd_buffer_attachment_state {
   /* The original clear value as provided by the Vulkan API */
   VkClearValue vk_clear_value;

   /* The hardware clear value */
   union v3dv_clear_value clear_value;
};

void v3dv_get_hw_clear_color(const VkClearColorValue *color,
                             uint32_t internal_type,
                             uint32_t internal_size,
                             uint32_t *hw_color);

struct v3dv_viewport_state {
   uint32_t count;
   VkViewport viewports[MAX_VIEWPORTS];
   float translate[MAX_VIEWPORTS][3];
   float scale[MAX_VIEWPORTS][3];
};

struct v3dv_scissor_state {
   uint32_t count;
   VkRect2D scissors[MAX_SCISSORS];
};

/* Mostly a v3dv mapping of VkDynamicState, used to track which data as
 * defined as dynamic
 */
enum v3dv_dynamic_state_bits {
   V3DV_DYNAMIC_VIEWPORT                  = 1 << 0,
   V3DV_DYNAMIC_SCISSOR                   = 1 << 1,
   V3DV_DYNAMIC_STENCIL_COMPARE_MASK      = 1 << 2,
   V3DV_DYNAMIC_STENCIL_WRITE_MASK        = 1 << 3,
   V3DV_DYNAMIC_STENCIL_REFERENCE         = 1 << 4,
   V3DV_DYNAMIC_BLEND_CONSTANTS           = 1 << 5,
   V3DV_DYNAMIC_DEPTH_BIAS                = 1 << 6,
   V3DV_DYNAMIC_LINE_WIDTH                = 1 << 7,
   V3DV_DYNAMIC_ALL                       = (1 << 8) - 1,
};

/* Flags for dirty pipeline state.
 */
enum v3dv_cmd_dirty_bits {
   V3DV_CMD_DIRTY_VIEWPORT                  = 1 << 0,
   V3DV_CMD_DIRTY_SCISSOR                   = 1 << 1,
   V3DV_CMD_DIRTY_STENCIL_COMPARE_MASK      = 1 << 2,
   V3DV_CMD_DIRTY_STENCIL_WRITE_MASK        = 1 << 3,
   V3DV_CMD_DIRTY_STENCIL_REFERENCE         = 1 << 4,
   V3DV_CMD_DIRTY_PIPELINE                  = 1 << 5,
   V3DV_CMD_DIRTY_VERTEX_BUFFER             = 1 << 6,
   V3DV_CMD_DIRTY_DESCRIPTOR_SETS           = 1 << 7,
   V3DV_CMD_DIRTY_PUSH_CONSTANTS            = 1 << 8,
   V3DV_CMD_DIRTY_BLEND_CONSTANTS           = 1 << 9,
   V3DV_CMD_DIRTY_OCCLUSION_QUERY           = 1 << 10,
   V3DV_CMD_DIRTY_DEPTH_BIAS                = 1 << 11,
   V3DV_CMD_DIRTY_LINE_WIDTH                = 1 << 12,
};

struct v3dv_dynamic_state {
   /**
    * Bitmask of (1 << VK_DYNAMIC_STATE_*).
    * Defines the set of saved dynamic state.
    */
   uint32_t mask;

   struct v3dv_viewport_state viewport;

   struct v3dv_scissor_state scissor;

   struct {
      uint32_t front;
      uint32_t back;
   } stencil_compare_mask;

   struct {
      uint32_t front;
      uint32_t back;
   } stencil_write_mask;

   struct {
      uint32_t front;
      uint32_t back;
   } stencil_reference;

   float blend_constants[4];

   struct {
      float constant_factor;
      float slope_factor;
   } depth_bias;

   float line_width;
};

extern const struct v3dv_dynamic_state default_dynamic_state;

void v3dv_viewport_compute_xform(const VkViewport *viewport,
                                 float scale[3],
                                 float translate[3]);

enum v3dv_ez_state {
   VC5_EZ_UNDECIDED = 0,
   VC5_EZ_GT_GE,
   VC5_EZ_LT_LE,
   VC5_EZ_DISABLED,
};

enum v3dv_job_type {
   V3DV_JOB_TYPE_GPU_CL = 0,
   V3DV_JOB_TYPE_GPU_TFU,
   V3DV_JOB_TYPE_GPU_CSD,
   V3DV_JOB_TYPE_CPU_RESET_QUERIES,
   V3DV_JOB_TYPE_CPU_END_QUERY,
   V3DV_JOB_TYPE_CPU_COPY_QUERY_RESULTS,
   V3DV_JOB_TYPE_CPU_SET_EVENT,
   V3DV_JOB_TYPE_CPU_WAIT_EVENTS,
};

struct v3dv_reset_query_cpu_job_info {
   struct v3dv_query_pool *pool;
   uint32_t first;
   uint32_t count;
};

struct v3dv_end_query_cpu_job_info {
   struct v3dv_query_pool *pool;
   uint32_t query;
};

struct v3dv_copy_query_results_cpu_job_info {
   struct v3dv_query_pool *pool;
   uint32_t first;
   uint32_t count;
   struct v3dv_buffer *dst;
   uint32_t offset;
   uint32_t stride;
   VkQueryResultFlags flags;
};

struct v3dv_event_set_cpu_job_info {
   struct v3dv_event *event;
   int state;
};

struct v3dv_event_wait_cpu_job_info {
   /* List of events to wait on */
   uint32_t event_count;
   struct v3dv_event **events;

   /* Whether any postponed jobs after the wait should wait on semaphores */
   bool sem_wait;
};

struct v3dv_job {
   struct list_head list_link;

   /* We only create job clones when executing secondary command buffers into
    * primaries. These clones don't make deep copies of the original object
    * so we want to flag them to avoid freeing resources they don't own.
    */
   bool is_clone;

   enum v3dv_job_type type;

   struct v3dv_device *device;

   struct v3dv_cmd_buffer *cmd_buffer;

   struct v3dv_cl bcl;
   struct v3dv_cl rcl;
   struct v3dv_cl indirect;

   /* Set of all BOs referenced by the job. This will be used for making
    * the list of BOs that the kernel will need to have paged in to
    * execute our job.
    */
   struct set *bos;
   uint32_t bo_count;

   struct v3dv_bo *tile_alloc;
   struct v3dv_bo *tile_state;

   bool tmu_dirty_rcl;

   uint32_t first_subpass;

   /* When the current subpass is split into multiple jobs, this flag is set
    * to true for any jobs after the first in the same subpass.
    */
   bool is_subpass_continue;

   /* If this job is the last job emitted for a subpass. */
   bool is_subpass_finish;

   struct v3dv_frame_tiling frame_tiling;

   enum v3dv_ez_state ez_state;
   enum v3dv_ez_state first_ez_state;

   /* Number of draw calls recorded into the job */
   uint32_t draw_count;

   /* A flag indicating whether we want to flush every draw separately. This
    * can be used for debugging, or for cases where special circumstances
    * require this behavior.
    */
   bool always_flush;

   /* Job specs for CPU jobs */
   union {
      struct v3dv_reset_query_cpu_job_info        query_reset;
      struct v3dv_end_query_cpu_job_info          query_end;
      struct v3dv_copy_query_results_cpu_job_info query_copy_results;
      struct v3dv_event_set_cpu_job_info          event_set;
      struct v3dv_event_wait_cpu_job_info         event_wait;
   } cpu;

   /* Job spects for TFU jobs */
   struct drm_v3d_submit_tfu tfu;
};

void v3dv_job_init(struct v3dv_job *job,
                   enum v3dv_job_type type,
                   struct v3dv_device *device,
                   struct v3dv_cmd_buffer *cmd_buffer,
                   int32_t subpass_idx);
void v3dv_job_destroy(struct v3dv_job *job);
void v3dv_job_add_bo(struct v3dv_job *job, struct v3dv_bo *bo);
void v3dv_job_emit_binning_flush(struct v3dv_job *job);
void v3dv_job_start_frame(struct v3dv_job *job,
                          uint32_t width,
                          uint32_t height,
                          uint32_t layers,
                          uint32_t render_target_count,
                          uint8_t max_internal_bpp);

struct v3dv_vertex_binding {
   struct v3dv_buffer *buffer;
   VkDeviceSize offset;
};

struct v3dv_descriptor_state {
   struct v3dv_descriptor_set *descriptor_sets[MAX_SETS];
   uint32_t valid;
   uint32_t dynamic_offsets[MAX_DYNAMIC_BUFFERS];
};

struct v3dv_cmd_buffer_state {
   struct v3dv_render_pass *pass;
   struct v3dv_framebuffer *framebuffer;
   VkRect2D render_area;

   /* Current job being recorded */
   struct v3dv_job *job;

   uint32_t subpass_idx;

   struct v3dv_pipeline *pipeline;
   struct v3dv_descriptor_state descriptor_state;

   struct v3dv_dynamic_state dynamic;
   uint32_t dirty;

   /* Current clip window. We use this to check whether we have an active
    * scissor, since in that case we can't use TLB clears and need to fallback
    * to drawing rects.
    */
   VkRect2D clip_window;

   /* Whether our render area is aligned to tile boundaries. If this is false
    * then we have tiles that are only partially covered by the render area,
    * and therefore, we need to be careful with our loads and stores so we don't
    * modify pixels for the tile area that is not covered by the render area.
    * This means, for example, that we can't use the TLB to clear, since that
    * always clears full tiles.
    */
   bool tile_aligned_render_area;

   uint32_t attachment_count;
   struct v3dv_cmd_buffer_attachment_state *attachments;

   struct v3dv_vertex_binding vertex_bindings[MAX_VBS];

   uint8_t index_size;

   /* Used to flag OOM conditions during command buffer recording */
   bool oom;

   /* Command buffer state saved during a meta operation */
   struct {
      uint32_t subpass_idx;
      VkRenderPass pass;
      VkPipeline pipeline;
      VkFramebuffer framebuffer;

      uint32_t attachment_alloc_count;
      uint32_t attachment_count;
      struct v3dv_cmd_buffer_attachment_state *attachments;

      bool tile_aligned_render_area;
      VkRect2D render_area;

      struct v3dv_dynamic_state dynamic;

      struct v3dv_descriptor_state descriptor_state;

      uint32_t push_constants[MAX_PUSH_CONSTANTS_SIZE / 4];
   } meta;

   /* Command buffer state for queries */
   struct {
      /* A list of vkCmdQueryEnd commands recorded in the command buffer during
       * a render pass. We queue these here and then schedule the corresponding
       * CPU jobs for them at the time we finish the GPU job in which they have
       * been recorded.
       */
      struct {
         uint32_t used_count;
         uint32_t alloc_count;
         struct v3dv_end_query_cpu_job_info *states;
      } end;

      /* This is not NULL if we have an active query, that is, we have called
       * vkCmdBeginQuery but not vkCmdEndQuery.
       */
      struct v3dv_bo *active_query;
   } query;
};

struct v3dv_descriptor {
   VkDescriptorType type;

   union {
      struct {
         struct v3dv_image_view *image_view;
         struct v3dv_sampler *sampler;
      };

      struct {
         struct v3dv_buffer *buffer;
         uint32_t offset;
      };
   };
};

/* Aux struct as it is really common to have a pair bo/address. Called
 * resource because it is really likely that we would need something like that
 * if we work on reuse the same bo at different points (like the shader
 * assembly).
 */
struct v3dv_resource {
   struct v3dv_bo *bo;
   uint32_t offset;
};

struct v3dv_query {
   bool maybe_available;
   struct v3dv_bo *bo;
};

struct v3dv_query_pool {
   uint32_t query_count;
   struct v3dv_query *queries;
};

VkResult v3dv_get_query_pool_results_cpu(struct v3dv_device *device,
                                         struct v3dv_query_pool *pool,
                                         uint32_t first,
                                         uint32_t count,
                                         void *data,
                                         VkDeviceSize stride,
                                         VkQueryResultFlags flags);

typedef void (*v3dv_cmd_buffer_private_obj_destroy_cb)(VkDevice device,
                                                       uint64_t pobj,
                                                       VkAllocationCallbacks *alloc);
struct v3dv_cmd_buffer_private_obj {
   struct list_head list_link;
   uint64_t obj;
   v3dv_cmd_buffer_private_obj_destroy_cb destroy_cb;
};

struct v3dv_cmd_buffer {
   VK_LOADER_DATA _loader_data;

   struct v3dv_device *device;

   struct v3dv_cmd_pool *pool;
   struct list_head pool_link;

   /* Used at submit time to link command buffers in the submission that have
    * spawned wait threads, so we can then wait on all of them to complete
    * before we process any signal sempahores or fences.
    */
   struct list_head list_link;

   VkCommandBufferUsageFlags usage_flags;
   VkCommandBufferLevel level;

   enum v3dv_cmd_buffer_status status;

   struct v3dv_cmd_buffer_state state;

   uint32_t push_constants_data[MAX_PUSH_CONSTANTS_SIZE / 4];
   struct v3dv_resource push_constants_resource;

   /* Collection of Vulkan objects created internally by the driver (typically
    * during recording of meta operations) that are part of the command buffer
    * and should be destroyed with it.
    */
   struct list_head private_objs; /* v3dv_cmd_buffer_private_obj */

   /* Per-command buffer resources for meta operations. */
   struct {
      struct {
         VkDescriptorPool dspool;
      } blit;
   } meta;

   /* List of jobs in the command buffer. For primary command buffers it
    * represents the jobs we want to submit to the GPU. For secondary command
    * buffers it represents jobs that will be merged into a primary command
    * buffer via vkCmdExecuteCommands.
    */
   struct list_head jobs;
};

struct v3dv_job *v3dv_cmd_buffer_start_job(struct v3dv_cmd_buffer *cmd_buffer,
                                           int32_t subpass_idx);
void v3dv_cmd_buffer_finish_job(struct v3dv_cmd_buffer *cmd_buffer);

struct v3dv_job *v3dv_cmd_buffer_subpass_start(struct v3dv_cmd_buffer *cmd_buffer,
                                               uint32_t subpass_idx);
struct v3dv_job *v3dv_cmd_buffer_subpass_resume(struct v3dv_cmd_buffer *cmd_buffer,
                                                uint32_t subpass_idx);

void v3dv_cmd_buffer_subpass_finish(struct v3dv_cmd_buffer *cmd_buffer);

void v3dv_cmd_buffer_meta_state_push(struct v3dv_cmd_buffer *cmd_buffer,
                                     bool push_descriptor_state);
void v3dv_cmd_buffer_meta_state_pop(struct v3dv_cmd_buffer *cmd_buffer,
                                    uint32_t dirty_dynamic_state);

void v3dv_render_pass_setup_render_target(struct v3dv_cmd_buffer *cmd_buffer,
                                          int rt,
                                          uint32_t *rt_bpp,
                                          uint32_t *rt_type,
                                          uint32_t *rt_clamp);

void v3dv_cmd_buffer_reset_queries(struct v3dv_cmd_buffer *cmd_buffer,
                                   struct v3dv_query_pool *pool,
                                   uint32_t first,
                                   uint32_t count);

void v3dv_cmd_buffer_begin_query(struct v3dv_cmd_buffer *cmd_buffer,
                                 struct v3dv_query_pool *pool,
                                 uint32_t query,
                                 VkQueryControlFlags flags);

void v3dv_cmd_buffer_end_query(struct v3dv_cmd_buffer *cmd_buffer,
                               struct v3dv_query_pool *pool,
                               uint32_t query);

void v3dv_cmd_buffer_copy_query_results(struct v3dv_cmd_buffer *cmd_buffer,
                                        struct v3dv_query_pool *pool,
                                        uint32_t first,
                                        uint32_t count,
                                        struct v3dv_buffer *dst,
                                        uint32_t offset,
                                        uint32_t stride,
                                        VkQueryResultFlags flags);

void v3dv_cmd_buffer_add_tfu_job(struct v3dv_cmd_buffer *cmd_buffer,
                                 struct drm_v3d_submit_tfu *tfu);

void v3dv_cmd_buffer_add_private_obj(struct v3dv_cmd_buffer *cmd_buffer,
                                     uint64_t obj,
                                     v3dv_cmd_buffer_private_obj_destroy_cb destroy_cb);

struct v3dv_semaphore {
   /* A syncobject handle associated with this semaphore */
   uint32_t sync;

   /* The file handle of a fence that we imported into our syncobject */
   int32_t fd;
};

struct v3dv_fence {
   /* A syncobject handle associated with this fence */
   uint32_t sync;

   /* The file handle of a fence that we imported into our syncobject */
   int32_t fd;
};

struct v3dv_event {
   int state;
};

struct v3dv_shader_module {
   /* A NIR shader. We create NIR modules for shaders that are generated
    * internally by the driver.
    */
   struct nir_shader *nir;

   /* A SPIR-V shader */
   unsigned char sha1[20];
   uint32_t size;
   char data[0];
};

/* FIXME: the same function at anv, radv and tu, perhaps create common
 * place?
 */
static inline gl_shader_stage
vk_to_mesa_shader_stage(VkShaderStageFlagBits vk_stage)
{
   assert(__builtin_popcount(vk_stage) == 1);
   return ffs(vk_stage) - 1;
}

struct v3dv_shader_variant {
   union {
      struct v3d_prog_data *base;
      struct v3d_vs_prog_data *vs;
      struct v3d_fs_prog_data *fs;
   } prog_data;

   /* FIXME: using one bo per shader. Eventually we would be interested on
    * reusing the same bo for all the shaders, like a bo per v3dv_pipeline for
    * shaders.
    */
   struct v3dv_bo *assembly_bo;
};

/*
 * Per-stage info for each stage, useful so shader_module_compile_to_nir and
 * other methods doesn't have so many parameters.
 *
 * FIXME: for the case of the coordinate shader and the vertex shader, module,
 * entrypoint, spec_info and nir are the same. There are also info only
 * relevant to some stages. But seemed too much a hassle to create a new
 * struct only to handle that. Revisit if such kind of info starts to grow.
 */
struct v3dv_pipeline_stage {
   struct v3dv_pipeline *pipeline;

   gl_shader_stage stage;
   /* FIXME: is_coord only make sense if stage == MESA_SHADER_VERTEX. Perhaps
    * a stage base/vs/fs as keys and prog_data?
    */
   bool is_coord;

   const struct v3dv_shader_module *module;
   const char *entrypoint;
   const VkSpecializationInfo *spec_info;

   nir_shader *nir;

   /** A name for this program, so you can track it in shader-db output. */
   uint32_t program_id;
   /** How many variants of this program were compiled, for shader-db. */
   uint32_t compiled_variant_count;

   /* The following are the default v3d_key populated using
    * VkCreateGraphicsPipelineCreateInfo. Variants will be created tweaking
    * them, so we don't need to maintain a copy of that create info struct
    * around
    */
   union {
      struct v3d_key base;
      struct v3d_vs_key vs;
      struct v3d_fs_key fs;
   } key;

   /* Cache with all the shader variant.
    */
   struct hash_table *cache;

   struct v3dv_shader_variant *current_variant;

   /* FIXME: only make sense on vs, so perhaps a v3dv key like radv? or a kind
    * of pipe_draw_info
    */
   enum pipe_prim_type topology;
};

/* FIXME: although the full vpm_config is not required at this point, as we
 * don't plan to initially support GS, it is more readable and serves as a
 * placeholder, to have the struct and fill it with default values.
 */
struct vpm_config {
   uint32_t As;
   uint32_t Vc;
   uint32_t Gs;
   uint32_t Gd;
   uint32_t Gv;
   uint32_t Ve;
   uint32_t gs_width;
};

struct v3dv_descriptor_pool_entry
{
   struct v3dv_descriptor_set *set;
};

struct v3dv_descriptor_pool {
   uint8_t *host_memory_base;
   uint8_t *host_memory_ptr;
   uint8_t *host_memory_end;

   uint32_t entry_count;
   uint32_t max_entry_count;
   struct v3dv_descriptor_pool_entry entries[0];
};

struct v3dv_descriptor_set {
   struct v3dv_descriptor_pool *pool;

   const struct v3dv_descriptor_set_layout *layout;

   /* The descriptors below can be indexed (set/binding) using the set_layout
    */
   struct v3dv_descriptor descriptors[0];
};

struct v3dv_descriptor_set_binding_layout {
   VkDescriptorType type;

   /* Number of array elements in this binding */
   uint32_t array_size;

   uint32_t descriptor_index;

   uint32_t dynamic_offset_count;
   uint32_t dynamic_offset_index;

   /* Offset in the v3dv_descriptor_set_layout of the immutable samplers, or 0
    * if there are no immutable samplers.
    */
   uint32_t immutable_samplers_offset;
};

struct v3dv_descriptor_set_layout {
   VkDescriptorSetLayoutCreateFlags flags;

   /* Number of bindings in this descriptor set */
   uint32_t binding_count;

   /* Total size of the descriptor set with room for all array entries */
   uint32_t size;

   /* Shader stages affected by this descriptor set */
   uint16_t shader_stages;

   /* Number of descriptors in this descriptor set */
   uint32_t descriptor_count;

   /* Number of dynamic offsets used by this descriptor set */
   uint16_t dynamic_offset_count;

   /* Bindings in this descriptor set */
   struct v3dv_descriptor_set_binding_layout binding[0];
};

struct v3dv_pipeline_layout {
   struct {
      struct v3dv_descriptor_set_layout *layout;
      uint32_t dynamic_offset_start;
   } set[MAX_SETS];

   uint32_t num_sets;
   uint32_t dynamic_offset_count;

   uint32_t push_constant_size;
};

struct v3dv_descriptor_map {
   /* TODO: avoid fixed size array/justify the size */
   unsigned num_desc; /* Number of descriptors  */
   int set[64];
   int binding[64];
   int array_index[64];
   int array_size[64];
};

struct v3dv_sampler {
   bool compare_enable;

   /* FIXME: here we store the packet SAMPLER_STATE, that is referenced as part
    * of the tmu configuration, and the content is set per sampler. A possible
    * perf improvement, to avoid bo fragmentation, would be to save the state
    * as static, have the bo as part of the descriptor (booked from the
    * descriptor pools), and then copy this content to the descriptor bo on
    * UpdateDescriptor
    */
   struct v3dv_bo *state;
};

#define V3DV_NO_SAMPLER_IDX 666

/*
 * Following two methods are using on the combined to/from texture/sampler
 * indices maps at v3dv_pipeline.
 */
static inline uint32_t
v3dv_pipeline_combined_index_key_create(uint32_t texture_index,
                                        uint32_t sampler_index)
{
   return texture_index << 24 | sampler_index;
}

static inline void
v3dv_pipeline_combined_index_key_unpack(uint32_t combined_index_key,
                                        uint32_t *texture_index,
                                        uint32_t *sampler_index)
{
   uint32_t texture = combined_index_key >> 24;
   uint32_t sampler = combined_index_key & 0xffffff;

   if (texture_index)
      *texture_index = texture;

   if (sampler_index)
      *sampler_index = sampler;
}

struct v3dv_pipeline {
   struct v3dv_device *device;

   VkShaderStageFlags active_stages;

   struct v3dv_render_pass *pass;
   struct v3dv_subpass *subpass;

   /* Note: We can't use just a MESA_SHADER_STAGES array as we need to track
    * too the coordinate shader
    */
   struct v3dv_pipeline_stage *vs;
   struct v3dv_pipeline_stage *vs_bin;
   struct v3dv_pipeline_stage *fs;

   struct v3dv_dynamic_state dynamic_state;

   struct v3dv_pipeline_layout *layout;

   enum v3dv_ez_state ez_state;

   bool primitive_restart;

   /* Accessed by binding. So vb[binding]->stride is the stride of the vertex
    * array with such binding
    */
   struct v3dv_pipeline_vertex_binding {
      uint32_t stride;
      uint32_t instance_divisor;
   } vb[MAX_VBS];
   uint32_t vb_count;

   /* Note that a lot of info from VkVertexInputAttributeDescription is
    * already prepacked, so storing here only those that need recheck later
    *
    * Note that they are not indexed by the location or nir driver location,
    * as we are defining here only the inputs that the shader are really
    * using.
    */
   struct v3dv_pipeline_vertex_attrib {
      uint32_t binding;
      uint32_t offset;
      /* We store driver_location instead of location because most v3d structs
       * are indexed by driver_location
       */
      uint32_t driver_location;
      VkFormat vk_format;
   } va[MAX_VERTEX_ATTRIBS];
   uint32_t va_count;

   struct v3dv_descriptor_map ubo_map;
   struct v3dv_descriptor_map ssbo_map;

   struct v3dv_descriptor_map sampler_map;
   struct v3dv_descriptor_map texture_map;

   /*
    * Vulkan has separate texture and sampler objects. Previous sampler and
    * texture map uses a sampler and texture index respectively, that can be
    * different. But OpenGL combine both (or in other words, they are the
    * same). The v3d compiler and all the nir lowerings that they use were
    * written under that assumption. In order to not update all those, we
    * combine the indexes, and we use the following maps to get one or the
    * other. In general the driver side uses the tex/sampler indexes to gather
    * resources, and the compiler side uses the combined index (so the v3d key
    * texture info will be indexed using the combined index).
    */
   struct hash_table *combined_index_map;
   uint32_t combined_index_to_key_map[32];
   uint32_t next_combined_index;

   /* FIXME: this bo is another candidate to data to be uploaded using a
    * resource manager, instead of a individual bo
    */
   struct v3dv_bo *default_attribute_values;

   struct vpm_config vpm_cfg;
   struct vpm_config vpm_cfg_bin;

   /* If the pipeline should emit any of the stencil configuration packets */
   bool emit_stencil_cfg[2];

   /* If the pipeline is using push constants */
   bool use_push_constants;

   /* Blend state */
   struct {
      /* Per-RT bit mask with blend enables */
      uint8_t enables;
      /* Per-RT prepacked blend config packets */
      uint8_t cfg[V3D_MAX_DRAW_BUFFERS][cl_packet_length(BLEND_CFG)];
      /* Flag indicating whether the blend factors in use require
       * color constants.
       */
      bool needs_color_constants;
      /* Mask with enabled color channels for each RT (4 bits per RT) */
      uint32_t color_write_masks;
   } blend;

   /* Depth bias */
   struct {
      bool enabled;
      bool is_z16;
   } depth_bias;

   /* Packets prepacked during pipeline creation
    */
   uint8_t cfg_bits[cl_packet_length(CFG_BITS)];
   uint8_t shader_state_record[cl_packet_length(GL_SHADER_STATE_RECORD)];
   uint8_t vcm_cache_size[cl_packet_length(VCM_CACHE_SIZE)];
   uint8_t vertex_attrs[cl_packet_length(GL_SHADER_STATE_ATTRIBUTE_RECORD) *
                        MAX_VERTEX_ATTRIBS];
   uint8_t stencil_cfg[2][cl_packet_length(STENCIL_CFG)];
};

const nir_shader_compiler_options *v3dv_pipeline_get_nir_options(void);

static inline uint32_t
v3dv_zs_buffer_from_aspect_bits(VkImageAspectFlags aspects)
{
   const VkImageAspectFlags zs_aspects =
      VK_IMAGE_ASPECT_DEPTH_BIT | VK_IMAGE_ASPECT_STENCIL_BIT;
   const VkImageAspectFlags filtered_aspects = aspects & zs_aspects;

   if (filtered_aspects == zs_aspects)
      return ZSTENCIL;
   else if (filtered_aspects == VK_IMAGE_ASPECT_DEPTH_BIT)
      return Z;
   else if (filtered_aspects == VK_IMAGE_ASPECT_STENCIL_BIT)
      return STENCIL;
   else
      return NONE;
}

static inline uint32_t
v3dv_zs_buffer_from_vk_format(VkFormat format)
{
   switch (format) {
   case VK_FORMAT_D16_UNORM_S8_UINT:
   case VK_FORMAT_D24_UNORM_S8_UINT:
   case VK_FORMAT_D32_SFLOAT_S8_UINT:
      return ZSTENCIL;
   case VK_FORMAT_D16_UNORM:
   case VK_FORMAT_D32_SFLOAT:
   case VK_FORMAT_X8_D24_UNORM_PACK32:
      return Z;
   case VK_FORMAT_S8_UINT:
      return STENCIL;
   default:
      return NONE;
   }
}

uint32_t v3dv_physical_device_api_version(struct v3dv_physical_device *dev);

int v3dv_get_instance_entrypoint_index(const char *name);
int v3dv_get_device_entrypoint_index(const char *name);
int v3dv_get_physical_device_entrypoint_index(const char *name);

const char *v3dv_get_instance_entry_name(int index);
const char *v3dv_get_physical_device_entry_name(int index);
const char *v3dv_get_device_entry_name(int index);

bool
v3dv_instance_entrypoint_is_enabled(int index, uint32_t core_version,
                                    const struct v3dv_instance_extension_table *instance);
bool
v3dv_physical_device_entrypoint_is_enabled(int index, uint32_t core_version,
                                           const struct v3dv_instance_extension_table *instance);
bool
v3dv_device_entrypoint_is_enabled(int index, uint32_t core_version,
                                  const struct v3dv_instance_extension_table *instance,
                                  const struct v3dv_device_extension_table *device);

void *v3dv_lookup_entrypoint(const struct v3d_device_info *devinfo,
                             const char *name);

#define v3dv_printflike(a, b) __attribute__((__format__(__printf__, a, b)))

VkResult __vk_errorf(struct v3dv_instance *instance, VkResult error,
                     const char *file, int line,
                     const char *format, ...);

#define vk_error(instance, error) __vk_errorf(instance, error, __FILE__, __LINE__, NULL);
#define vk_errorf(instance, error, format, ...) __vk_errorf(instance, error, __FILE__, __LINE__, format, ## __VA_ARGS__);

void v3dv_loge(const char *format, ...) v3dv_printflike(1, 2);
void v3dv_loge_v(const char *format, va_list va);

#define v3dv_debug_ignored_stype(sType) \
   v3dv_loge("%s: ignored VkStructureType %u:%s\n", __func__, (sType), vk_StructureType_to_str(sType))

const struct v3dv_format *v3dv_get_format(VkFormat);
const uint8_t *v3dv_get_format_swizzle(VkFormat f);
void v3dv_get_internal_type_bpp_for_output_format(uint32_t format, uint32_t *type, uint32_t *bpp);
uint8_t v3dv_get_tex_return_size(const struct v3dv_format *vf, bool compare_enable);
bool v3dv_tfu_supports_tex_format(const struct v3d_device_info *devinfo,
                                  uint32_t tex_format);



uint32_t v3d_utile_width(int cpp);
uint32_t v3d_utile_height(int cpp);

void v3d_load_tiled_image(void *dst, uint32_t dst_stride,
                          void *src, uint32_t src_stride,
                          enum v3d_tiling_mode tiling_format,
                          int cpp, uint32_t image_h,
                          const struct pipe_box *box);

void v3d_store_tiled_image(void *dst, uint32_t dst_stride,
                           void *src, uint32_t src_stride,
                           enum v3d_tiling_mode tiling_format,
                           int cpp, uint32_t image_h,
                           const struct pipe_box *box);

struct v3dv_cl_reloc v3dv_write_uniforms(struct v3dv_cmd_buffer *cmd_buffer,
                                         struct v3dv_pipeline_stage *p_stage);

struct v3dv_shader_variant *
v3dv_get_shader_variant(struct v3dv_pipeline_stage *p_stage,
                        struct v3d_key *key,
                        size_t key_size,
                        const VkAllocationCallbacks *pAllocator,
                        VkResult *out_vk_result);

struct v3dv_descriptor *
v3dv_descriptor_map_get_descriptor(struct v3dv_descriptor_state *descriptor_state,
                                   struct v3dv_descriptor_map *map,
                                   struct v3dv_pipeline_layout *pipeline_layout,
                                   uint32_t index,
                                   uint32_t *dynamic_offset);

const struct v3dv_sampler *
v3dv_descriptor_map_get_sampler(struct v3dv_descriptor_state *descriptor_state,
                                struct v3dv_descriptor_map *map,
                                struct v3dv_pipeline_layout *pipeline_layout,
                                uint32_t index);

struct v3dv_image_view *
v3dv_descriptor_map_get_image_view(struct v3dv_descriptor_state *descriptor_state,
                                   struct v3dv_descriptor_map *map,
                                   struct v3dv_pipeline_layout *pipeline_layout,
                                   uint32_t index);

static inline const struct v3dv_sampler *
v3dv_immutable_samplers(const struct v3dv_descriptor_set_layout *set,
                        const struct v3dv_descriptor_set_binding_layout *binding)
{
   assert(binding->immutable_samplers_offset);
   return (const struct v3dv_sampler *) ((const char *) set + binding->immutable_samplers_offset);
}

#define V3DV_DEFINE_HANDLE_CASTS(__v3dv_type, __VkType)   \
                                                        \
   static inline struct __v3dv_type *                    \
   __v3dv_type ## _from_handle(__VkType _handle)         \
   {                                                    \
      return (struct __v3dv_type *) _handle;             \
   }                                                    \
                                                        \
   static inline __VkType                               \
   __v3dv_type ## _to_handle(struct __v3dv_type *_obj)    \
   {                                                    \
      return (__VkType) _obj;                           \
   }

#define V3DV_DEFINE_NONDISP_HANDLE_CASTS(__v3dv_type, __VkType)              \
                                                                           \
   static inline struct __v3dv_type *                                       \
   __v3dv_type ## _from_handle(__VkType _handle)                            \
   {                                                                       \
      return (struct __v3dv_type *)(uintptr_t) _handle;                     \
   }                                                                       \
                                                                           \
   static inline __VkType                                                  \
   __v3dv_type ## _to_handle(struct __v3dv_type *_obj)                       \
   {                                                                       \
      return (__VkType)(uintptr_t) _obj;                                   \
   }

#define V3DV_FROM_HANDLE(__v3dv_type, __name, __handle)			\
   struct __v3dv_type *__name = __v3dv_type ## _from_handle(__handle)

V3DV_DEFINE_HANDLE_CASTS(v3dv_cmd_buffer, VkCommandBuffer)
V3DV_DEFINE_HANDLE_CASTS(v3dv_device, VkDevice)
V3DV_DEFINE_HANDLE_CASTS(v3dv_instance, VkInstance)
V3DV_DEFINE_HANDLE_CASTS(v3dv_physical_device, VkPhysicalDevice)
V3DV_DEFINE_HANDLE_CASTS(v3dv_queue, VkQueue)

V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_cmd_pool, VkCommandPool)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_buffer, VkBuffer)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_buffer_view, VkBufferView)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_device_memory, VkDeviceMemory)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_descriptor_pool, VkDescriptorPool)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_descriptor_set, VkDescriptorSet)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_descriptor_set_layout, VkDescriptorSetLayout)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_event, VkEvent)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_fence, VkFence)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_framebuffer, VkFramebuffer)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_image, VkImage)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_image_view, VkImageView)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_pipeline, VkPipeline)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_pipeline_layout, VkPipelineLayout)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_query_pool, VkQueryPool)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_render_pass, VkRenderPass)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_sampler, VkSampler)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_semaphore, VkSemaphore)
V3DV_DEFINE_NONDISP_HANDLE_CASTS(v3dv_shader_module, VkShaderModule)

/* This is defined as a macro so that it works for both
 * VkImageSubresourceRange and VkImageSubresourceLayers
 */
#define v3dv_layer_count(_image, _range) \
   ((_range)->layerCount == VK_REMAINING_ARRAY_LAYERS ? \
    (_image)->array_size - (_range)->baseArrayLayer : (_range)->layerCount)

#define v3dv_level_count(_image, _range) \
   ((_range)->levelCount == VK_REMAINING_MIP_LEVELS ? \
    (_image)->levels - (_range)->baseMipLevel : (_range)->levelCount)

static inline int
v3dv_ioctl(int fd, unsigned long request, void *arg)
{
   if (using_v3d_simulator)
      return v3d_simulator_ioctl(fd, request, arg);
   else
      return drmIoctl(fd, request, arg);
}

/* Flags OOM conditions in command buffer state.
 *
 * Note: notice that no-op jobs don't have a command buffer reference.
 */
static inline void
v3dv_flag_oom(struct v3dv_cmd_buffer *cmd_buffer, struct v3dv_job *job)
{
   if (cmd_buffer) {
      cmd_buffer->state.oom = true;
   } else {
      assert(job);
      if (job->cmd_buffer)
         job->cmd_buffer->state.oom = true;
   }
}

#define v3dv_return_if_oom(_cmd_buffer, _job) do {                  \
   const struct v3dv_cmd_buffer *__cmd_buffer = _cmd_buffer;        \
   if (__cmd_buffer && __cmd_buffer->state.oom)                     \
      return;                                                       \
   const struct v3dv_job *__job = _job;                             \
   if (__job && __job->cmd_buffer && __job->cmd_buffer->state.oom)  \
      return;                                                       \
} while(0)                                                          \

#endif /* V3DV_PRIVATE_H */