summaryrefslogtreecommitdiff
path: root/src/intel/vulkan/anv_private.h
blob: 35bc5ba0cdf38c120b2568c751e1b5c2aa37de07 (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
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
/*
 * Copyright © 2015 Intel Corporation
 *
 * 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 ANV_PRIVATE_H
#define ANV_PRIVATE_H

#include <stdlib.h>
#include <stdio.h>
#include <stdbool.h>
#include <pthread.h>
#include <assert.h>
#include <stdint.h>
#include "drm-uapi/i915_drm.h"

#ifdef HAVE_VALGRIND
#include <valgrind.h>
#include <memcheck.h>
#define VG(x) x
#ifndef NDEBUG
#define __gen_validate_value(x) VALGRIND_CHECK_MEM_IS_DEFINED(&(x), sizeof(x))
#endif
#else
#define VG(x) ((void)0)
#endif

#include "common/intel_clflush.h"
#include "common/intel_decoder.h"
#include "common/intel_gem.h"
#include "common/intel_l3_config.h"
#include "common/intel_measure.h"
#include "dev/gen_device_info.h"
#include "blorp/blorp.h"
#include "compiler/brw_compiler.h"
#include "util/bitset.h"
#include "util/bitscan.h"
#include "util/macros.h"
#include "util/hash_table.h"
#include "util/list.h"
#include "util/sparse_array.h"
#include "util/u_atomic.h"
#include "util/u_vector.h"
#include "util/u_math.h"
#include "util/vma.h"
#include "util/xmlconfig.h"
#include "vk_alloc.h"
#include "vk_debug_report.h"
#include "vk_device.h"
#include "vk_instance.h"
#include "vk_physical_device.h"
#include "vk_shader_module.h"
#include "vk_util.h"

/* Pre-declarations needed for WSI entrypoints */
struct wl_surface;
struct wl_display;
typedef struct xcb_connection_t xcb_connection_t;
typedef uint32_t xcb_visualid_t;
typedef uint32_t xcb_window_t;

struct anv_batch;
struct anv_buffer;
struct anv_buffer_view;
struct anv_image_view;
struct anv_instance;

struct intel_aux_map_context;
struct gen_perf_config;
struct gen_perf_counter_pass;
struct gen_perf_query_result;

#include <vulkan/vulkan.h>
#include <vulkan/vk_icd.h>

#include "anv_android.h"
#include "anv_entrypoints.h"
#include "isl/isl.h"

#include "dev/gen_debug.h"
#define MESA_LOG_TAG "MESA-INTEL"
#include "util/log.h"
#include "wsi_common.h"

#define NSEC_PER_SEC 1000000000ull

/* anv Virtual Memory Layout
 * =========================
 *
 * When the anv driver is determining the virtual graphics addresses of memory
 * objects itself using the softpin mechanism, the following memory ranges
 * will be used.
 *
 * Three special considerations to notice:
 *
 * (1) the dynamic state pool is located within the same 4 GiB as the low
 * heap. This is to work around a VF cache issue described in a comment in
 * anv_physical_device_init_heaps.
 *
 * (2) the binding table pool is located at lower addresses than the surface
 * state pool, within a 4 GiB range. This allows surface state base addresses
 * to cover both binding tables (16 bit offsets) and surface states (32 bit
 * offsets).
 *
 * (3) the last 4 GiB of the address space is withheld from the high
 * heap. Various hardware units will read past the end of an object for
 * various reasons. This healthy margin prevents reads from wrapping around
 * 48-bit addresses.
 */
#define GENERAL_STATE_POOL_MIN_ADDRESS     0x000000010000ULL /* 64 KiB */
#define GENERAL_STATE_POOL_MAX_ADDRESS     0x00003fffffffULL
#define LOW_HEAP_MIN_ADDRESS               0x000040000000ULL /* 1 GiB */
#define LOW_HEAP_MAX_ADDRESS               0x00007fffffffULL
#define DYNAMIC_STATE_POOL_MIN_ADDRESS     0x0000c0000000ULL /* 3 GiB */
#define DYNAMIC_STATE_POOL_MAX_ADDRESS     0x0000ffffffffULL
#define BINDING_TABLE_POOL_MIN_ADDRESS     0x000100000000ULL /* 4 GiB */
#define BINDING_TABLE_POOL_MAX_ADDRESS     0x00013fffffffULL
#define SURFACE_STATE_POOL_MIN_ADDRESS     0x000140000000ULL /* 5 GiB */
#define SURFACE_STATE_POOL_MAX_ADDRESS     0x00017fffffffULL
#define INSTRUCTION_STATE_POOL_MIN_ADDRESS 0x000180000000ULL /* 6 GiB */
#define INSTRUCTION_STATE_POOL_MAX_ADDRESS 0x0001bfffffffULL
#define CLIENT_VISIBLE_HEAP_MIN_ADDRESS    0x0001c0000000ULL /* 7 GiB */
#define CLIENT_VISIBLE_HEAP_MAX_ADDRESS    0x0002bfffffffULL
#define HIGH_HEAP_MIN_ADDRESS              0x0002c0000000ULL /* 11 GiB */

#define GENERAL_STATE_POOL_SIZE     \
   (GENERAL_STATE_POOL_MAX_ADDRESS - GENERAL_STATE_POOL_MIN_ADDRESS + 1)
#define LOW_HEAP_SIZE               \
   (LOW_HEAP_MAX_ADDRESS - LOW_HEAP_MIN_ADDRESS + 1)
#define DYNAMIC_STATE_POOL_SIZE     \
   (DYNAMIC_STATE_POOL_MAX_ADDRESS - DYNAMIC_STATE_POOL_MIN_ADDRESS + 1)
#define BINDING_TABLE_POOL_SIZE     \
   (BINDING_TABLE_POOL_MAX_ADDRESS - BINDING_TABLE_POOL_MIN_ADDRESS + 1)
#define SURFACE_STATE_POOL_SIZE     \
   (SURFACE_STATE_POOL_MAX_ADDRESS - SURFACE_STATE_POOL_MIN_ADDRESS + 1)
#define INSTRUCTION_STATE_POOL_SIZE \
   (INSTRUCTION_STATE_POOL_MAX_ADDRESS - INSTRUCTION_STATE_POOL_MIN_ADDRESS + 1)
#define CLIENT_VISIBLE_HEAP_SIZE               \
   (CLIENT_VISIBLE_HEAP_MAX_ADDRESS - CLIENT_VISIBLE_HEAP_MIN_ADDRESS + 1)

/* Allowing different clear colors requires us to perform a depth resolve at
 * the end of certain render passes. This is because while slow clears store
 * the clear color in the HiZ buffer, fast clears (without a resolve) don't.
 * See the PRMs for examples describing when additional resolves would be
 * necessary. To enable fast clears without requiring extra resolves, we set
 * the clear value to a globally-defined one. We could allow different values
 * if the user doesn't expect coherent data during or after a render passes
 * (VK_ATTACHMENT_STORE_OP_DONT_CARE), but such users (aside from the CTS)
 * don't seem to exist yet. In almost all Vulkan applications tested thus far,
 * 1.0f seems to be the only value used. The only application that doesn't set
 * this value does so through the usage of an seemingly uninitialized clear
 * value.
 */
#define ANV_HZ_FC_VAL 1.0f

#define MAX_VBS         28
#define MAX_XFB_BUFFERS  4
#define MAX_XFB_STREAMS  4
#define MAX_SETS         8
#define MAX_RTS          8
#define MAX_VIEWPORTS   16
#define MAX_SCISSORS    16
#define MAX_PUSH_CONSTANTS_SIZE 128
#define MAX_DYNAMIC_BUFFERS 16
#define MAX_IMAGES 64
#define MAX_PUSH_DESCRIPTORS 32 /* Minimum requirement */
#define MAX_INLINE_UNIFORM_BLOCK_SIZE 4096
#define MAX_INLINE_UNIFORM_BLOCK_DESCRIPTORS 32
/* We need 16 for UBO block reads to work and 32 for push UBOs. However, we
 * use 64 here to avoid cache issues. This could most likely bring it back to
 * 32 if we had different virtual addresses for the different views on a given
 * GEM object.
 */
#define ANV_UBO_ALIGNMENT 64
#define ANV_SSBO_ALIGNMENT 4
#define ANV_SSBO_BOUNDS_CHECK_ALIGNMENT 4
#define MAX_VIEWS_FOR_PRIMITIVE_REPLICATION 16
#define MAX_SAMPLE_LOCATIONS 16

/* From the Skylake PRM Vol. 7 "Binding Table Surface State Model":
 *
 *    "The surface state model is used when a Binding Table Index (specified
 *    in the message descriptor) of less than 240 is specified. In this model,
 *    the Binding Table Index is used to index into the binding table, and the
 *    binding table entry contains a pointer to the SURFACE_STATE."
 *
 * Binding table values above 240 are used for various things in the hardware
 * such as stateless, stateless with incoherent cache, SLM, and bindless.
 */
#define MAX_BINDING_TABLE_SIZE 240

/* The kernel relocation API has a limitation of a 32-bit delta value
 * applied to the address before it is written which, in spite of it being
 * unsigned, is treated as signed .  Because of the way that this maps to
 * the Vulkan API, we cannot handle an offset into a buffer that does not
 * fit into a signed 32 bits.  The only mechanism we have for dealing with
 * this at the moment is to limit all VkDeviceMemory objects to a maximum
 * of 2GB each.  The Vulkan spec allows us to do this:
 *
 *    "Some platforms may have a limit on the maximum size of a single
 *    allocation. For example, certain systems may fail to create
 *    allocations with a size greater than or equal to 4GB. Such a limit is
 *    implementation-dependent, and if such a failure occurs then the error
 *    VK_ERROR_OUT_OF_DEVICE_MEMORY should be returned."
 *
 * We don't use vk_error here because it's not an error so much as an
 * indication to the application that the allocation is too large.
 */
#define MAX_MEMORY_ALLOCATION_SIZE (1ull << 31)

#define ANV_SVGS_VB_INDEX    MAX_VBS
#define ANV_DRAWID_VB_INDEX (MAX_VBS + 1)

/* We reserve this MI ALU register for the purpose of handling predication.
 * Other code which uses the MI ALU should leave it alone.
 */
#define ANV_PREDICATE_RESULT_REG 0x2678 /* MI_ALU_REG15 */

/* We reserve this MI ALU register to pass around an offset computed from
 * VkPerformanceQuerySubmitInfoKHR::counterPassIndex VK_KHR_performance_query.
 * Other code which uses the MI ALU should leave it alone.
 */
#define ANV_PERF_QUERY_OFFSET_REG 0x2670 /* MI_ALU_REG14 */

/* For gfx12 we set the streamout buffers using 4 separate commands
 * (3DSTATE_SO_BUFFER_INDEX_*) instead of 3DSTATE_SO_BUFFER. However the layout
 * of the 3DSTATE_SO_BUFFER_INDEX_* commands is identical to that of
 * 3DSTATE_SO_BUFFER apart from the SOBufferIndex field, so for now we use the
 * 3DSTATE_SO_BUFFER command, but change the 3DCommandSubOpcode.
 * SO_BUFFER_INDEX_0_CMD is actually the 3DCommandSubOpcode for
 * 3DSTATE_SO_BUFFER_INDEX_0.
 */
#define SO_BUFFER_INDEX_0_CMD 0x60
#define anv_printflike(a, b) __attribute__((__format__(__printf__, a, b)))

static inline uint32_t
align_down_npot_u32(uint32_t v, uint32_t a)
{
   return v - (v % a);
}

static inline uint32_t
align_down_u32(uint32_t v, uint32_t a)
{
   assert(a != 0 && a == (a & -a));
   return v & ~(a - 1);
}

static inline uint32_t
align_u32(uint32_t v, uint32_t a)
{
   assert(a != 0 && a == (a & -a));
   return align_down_u32(v + a - 1, a);
}

static inline uint64_t
align_down_u64(uint64_t v, uint64_t a)
{
   assert(a != 0 && a == (a & -a));
   return v & ~(a - 1);
}

static inline uint64_t
align_u64(uint64_t v, uint64_t a)
{
   return align_down_u64(v + a - 1, a);
}

static inline int32_t
align_i32(int32_t v, int32_t a)
{
   assert(a != 0 && a == (a & -a));
   return (v + a - 1) & ~(a - 1);
}

/** Alignment must be a power of 2. */
static inline bool
anv_is_aligned(uintmax_t n, uintmax_t a)
{
   assert(a == (a & -a));
   return (n & (a - 1)) == 0;
}

static inline uint32_t
anv_minify(uint32_t n, uint32_t levels)
{
   if (unlikely(n == 0))
      return 0;
   else
      return MAX2(n >> levels, 1);
}

static inline float
anv_clamp_f(float f, float min, float max)
{
   assert(min < max);

   if (f > max)
      return max;
   else if (f < min)
      return min;
   else
      return f;
}

static inline bool
anv_clear_mask(uint32_t *inout_mask, uint32_t clear_mask)
{
   if (*inout_mask & clear_mask) {
      *inout_mask &= ~clear_mask;
      return true;
   } else {
      return false;
   }
}

static inline union isl_color_value
vk_to_isl_color(VkClearColorValue color)
{
   return (union isl_color_value) {
      .u32 = {
         color.uint32[0],
         color.uint32[1],
         color.uint32[2],
         color.uint32[3],
      },
   };
}

static inline void *anv_unpack_ptr(uintptr_t ptr, int bits, int *flags)
{
   uintptr_t mask = (1ull << bits) - 1;
   *flags = ptr & mask;
   return (void *) (ptr & ~mask);
}

static inline uintptr_t anv_pack_ptr(void *ptr, int bits, int flags)
{
   uintptr_t value = (uintptr_t) ptr;
   uintptr_t mask = (1ull << bits) - 1;
   return value | (mask & flags);
}

/* Whenever we generate an error, pass it through this function. Useful for
 * debugging, where we can break on it. Only call at error site, not when
 * propagating errors. Might be useful to plug in a stack trace here.
 */

VkResult __vk_errorv(struct anv_instance *instance,
                     const struct vk_object_base *object, VkResult error,
                     const char *file, int line, const char *format,
                     va_list args);

VkResult __vk_errorf(struct anv_instance *instance,
                     const struct vk_object_base *object, VkResult error,
                     const char *file, int line, const char *format, ...)
   anv_printflike(6, 7);

#ifdef DEBUG
#define vk_error(error) __vk_errorf(NULL, NULL, error, __FILE__, __LINE__, NULL)
#define vk_errorfi(instance, obj, error, format, ...)\
    __vk_errorf(instance, obj, error,\
                __FILE__, __LINE__, format, ## __VA_ARGS__)
#define vk_errorf(device, obj, error, format, ...)\
   vk_errorfi(anv_device_instance_or_null(device),\
              obj, error, format, ## __VA_ARGS__)
#else

static inline VkResult __dummy_vk_error(VkResult error, UNUSED const void *ignored)
{
   return error;
}

#define vk_error(error) __dummy_vk_error(error, NULL)
#define vk_errorfi(instance, obj, error, format, ...) __dummy_vk_error(error, instance)
#define vk_errorf(device, obj, error, format, ...) __dummy_vk_error(error, device)
#endif

/**
 * Warn on ignored extension structs.
 *
 * The Vulkan spec requires us to ignore unsupported or unknown structs in
 * a pNext chain.  In debug mode, emitting warnings for ignored structs may
 * help us discover structs that we should not have ignored.
 *
 *
 * From the Vulkan 1.0.38 spec:
 *
 *    Any component of the implementation (the loader, any enabled layers,
 *    and drivers) must skip over, without processing (other than reading the
 *    sType and pNext members) any chained structures with sType values not
 *    defined by extensions supported by that component.
 */
#define anv_debug_ignored_stype(sType) \
   mesa_logd("%s: ignored VkStructureType %u\n", __func__, (sType))

void __anv_perf_warn(struct anv_device *device,
                     const struct vk_object_base *object,
                     const char *file, int line, const char *format, ...)
   anv_printflike(5, 6);
void anv_loge(const char *format, ...) anv_printflike(1, 2);
void anv_loge_v(const char *format, va_list va);

/**
 * Print a FINISHME message, including its source location.
 */
#define anv_finishme(format, ...) \
   do { \
      static bool reported = false; \
      if (!reported) { \
         mesa_logw("%s:%d: FINISHME: " format, __FILE__, __LINE__, \
                    ##__VA_ARGS__); \
         reported = true; \
      } \
   } while (0)

/**
 * Print a perf warning message.  Set INTEL_DEBUG=perf to see these.
 */
#define anv_perf_warn(instance, obj, format, ...) \
   do { \
      static bool reported = false; \
      if (!reported && (INTEL_DEBUG & DEBUG_PERF)) { \
         __anv_perf_warn(instance, obj, __FILE__, __LINE__,\
                         format, ##__VA_ARGS__); \
         reported = true; \
      } \
   } while (0)

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

struct anv_bo {
   const char *name;

   uint32_t gem_handle;

   uint32_t refcount;

   /* Index into the current validation list.  This is used by the
    * validation list building alrogithm to track which buffers are already
    * in the validation list so that we can ensure uniqueness.
    */
   uint32_t index;

   /* Index for use with util_sparse_array_free_list */
   uint32_t free_index;

   /* Last known offset.  This value is provided by the kernel when we
    * execbuf and is used as the presumed offset for the next bunch of
    * relocations.
    */
   uint64_t offset;

   /** Size of the buffer not including implicit aux */
   uint64_t size;

   /* Map for internally mapped BOs.
    *
    * If ANV_BO_WRAPPER is set in flags, map points to the wrapped BO.
    */
   void *map;

   /** Size of the implicit CCS range at the end of the buffer
    *
    * On Gfx12, CCS data is always a direct 1/256 scale-down.  A single 64K
    * page of main surface data maps to a 256B chunk of CCS data and that
    * mapping is provided on TGL-LP by the AUX table which maps virtual memory
    * addresses in the main surface to virtual memory addresses for CCS data.
    *
    * Because we can't change these maps around easily and because Vulkan
    * allows two VkImages to be bound to overlapping memory regions (as long
    * as the app is careful), it's not feasible to make this mapping part of
    * the image.  (On Gfx11 and earlier, the mapping was provided via
    * RENDER_SURFACE_STATE so each image had its own main -> CCS mapping.)
    * Instead, we attach the CCS data directly to the buffer object and setup
    * the AUX table mapping at BO creation time.
    *
    * This field is for internal tracking use by the BO allocator only and
    * should not be touched by other parts of the code.  If something wants to
    * know if a BO has implicit CCS data, it should instead look at the
    * has_implicit_ccs boolean below.
    *
    * This data is not included in maps of this buffer.
    */
   uint32_t _ccs_size;

   /** Flags to pass to the kernel through drm_i915_exec_object2::flags */
   uint32_t flags;

   /** True if this BO may be shared with other processes */
   bool is_external:1;

   /** True if this BO is a wrapper
    *
    * When set to true, none of the fields in this BO are meaningful except
    * for anv_bo::is_wrapper and anv_bo::map which points to the actual BO.
    * See also anv_bo_unwrap().  Wrapper BOs are not allowed when use_softpin
    * is set in the physical device.
    */
   bool is_wrapper:1;

   /** See also ANV_BO_ALLOC_FIXED_ADDRESS */
   bool has_fixed_address:1;

   /** True if this BO wraps a host pointer */
   bool from_host_ptr:1;

   /** See also ANV_BO_ALLOC_CLIENT_VISIBLE_ADDRESS */
   bool has_client_visible_address:1;

   /** True if this BO has implicit CCS data attached to it */
   bool has_implicit_ccs:1;
};

static inline struct anv_bo *
anv_bo_ref(struct anv_bo *bo)
{
   p_atomic_inc(&bo->refcount);
   return bo;
}

static inline struct anv_bo *
anv_bo_unwrap(struct anv_bo *bo)
{
   while (bo->is_wrapper)
      bo = bo->map;
   return bo;
}

/* Represents a lock-free linked list of "free" things.  This is used by
 * both the block pool and the state pools.  Unfortunately, in order to
 * solve the ABA problem, we can't use a single uint32_t head.
 */
union anv_free_list {
   struct {
      uint32_t offset;

      /* A simple count that is incremented every time the head changes. */
      uint32_t count;
   };
   /* Make sure it's aligned to 64 bits. This will make atomic operations
    * faster on 32 bit platforms.
    */
   uint64_t u64 __attribute__ ((aligned (8)));
};

#define ANV_FREE_LIST_EMPTY ((union anv_free_list) { { UINT32_MAX, 0 } })

struct anv_block_state {
   union {
      struct {
         uint32_t next;
         uint32_t end;
      };
      /* Make sure it's aligned to 64 bits. This will make atomic operations
       * faster on 32 bit platforms.
       */
      uint64_t u64 __attribute__ ((aligned (8)));
   };
};

#define anv_block_pool_foreach_bo(bo, pool)  \
   for (struct anv_bo **_pp_bo = (pool)->bos, *bo; \
        _pp_bo != &(pool)->bos[(pool)->nbos] && (bo = *_pp_bo, true); \
        _pp_bo++)

#define ANV_MAX_BLOCK_POOL_BOS 20

struct anv_block_pool {
   const char *name;

   struct anv_device *device;
   bool use_softpin;

   /* Wrapper BO for use in relocation lists.  This BO is simply a wrapper
    * around the actual BO so that we grow the pool after the wrapper BO has
    * been put in a relocation list.  This is only used in the non-softpin
    * case.
    */
   struct anv_bo wrapper_bo;

   struct anv_bo *bos[ANV_MAX_BLOCK_POOL_BOS];
   struct anv_bo *bo;
   uint32_t nbos;

   uint64_t size;

   /* The address where the start of the pool is pinned. The various bos that
    * are created as the pool grows will have addresses in the range
    * [start_address, start_address + BLOCK_POOL_MEMFD_SIZE).
    */
   uint64_t start_address;

   /* The offset from the start of the bo to the "center" of the block
    * pool.  Pointers to allocated blocks are given by
    * bo.map + center_bo_offset + offsets.
    */
   uint32_t center_bo_offset;

   /* Current memory map of the block pool.  This pointer may or may not
    * point to the actual beginning of the block pool memory.  If
    * anv_block_pool_alloc_back has ever been called, then this pointer
    * will point to the "center" position of the buffer and all offsets
    * (negative or positive) given out by the block pool alloc functions
    * will be valid relative to this pointer.
    *
    * In particular, map == bo.map + center_offset
    *
    * DO NOT access this pointer directly. Use anv_block_pool_map() instead,
    * since it will handle the softpin case as well, where this points to NULL.
    */
   void *map;
   int fd;

   /**
    * Array of mmaps and gem handles owned by the block pool, reclaimed when
    * the block pool is destroyed.
    */
   struct u_vector mmap_cleanups;

   struct anv_block_state state;

   struct anv_block_state back_state;
};

/* Block pools are backed by a fixed-size 1GB memfd */
#define BLOCK_POOL_MEMFD_SIZE (1ul << 30)

/* The center of the block pool is also the middle of the memfd.  This may
 * change in the future if we decide differently for some reason.
 */
#define BLOCK_POOL_MEMFD_CENTER (BLOCK_POOL_MEMFD_SIZE / 2)

static inline uint32_t
anv_block_pool_size(struct anv_block_pool *pool)
{
   return pool->state.end + pool->back_state.end;
}

struct anv_state {
   int32_t offset;
   uint32_t alloc_size;
   void *map;
   uint32_t idx;
};

#define ANV_STATE_NULL ((struct anv_state) { .alloc_size = 0 })

struct anv_fixed_size_state_pool {
   union anv_free_list free_list;
   struct anv_block_state block;
};

#define ANV_MIN_STATE_SIZE_LOG2 6
#define ANV_MAX_STATE_SIZE_LOG2 21

#define ANV_STATE_BUCKETS (ANV_MAX_STATE_SIZE_LOG2 - ANV_MIN_STATE_SIZE_LOG2 + 1)

struct anv_free_entry {
   uint32_t next;
   struct anv_state state;
};

struct anv_state_table {
   struct anv_device *device;
   int fd;
   struct anv_free_entry *map;
   uint32_t size;
   struct anv_block_state state;
   struct u_vector cleanups;
};

struct anv_state_pool {
   struct anv_block_pool block_pool;

   /* Offset into the relevant state base address where the state pool starts
    * allocating memory.
    */
   int32_t start_offset;

   struct anv_state_table table;

   /* The size of blocks which will be allocated from the block pool */
   uint32_t block_size;

   /** Free list for "back" allocations */
   union anv_free_list back_alloc_free_list;

   struct anv_fixed_size_state_pool buckets[ANV_STATE_BUCKETS];
};

struct anv_state_reserved_pool {
   struct anv_state_pool *pool;
   union anv_free_list reserved_blocks;
   uint32_t count;
};

struct anv_state_stream {
   struct anv_state_pool *state_pool;

   /* The size of blocks to allocate from the state pool */
   uint32_t block_size;

   /* Current block we're allocating from */
   struct anv_state block;

   /* Offset into the current block at which to allocate the next state */
   uint32_t next;

   /* List of all blocks allocated from this pool */
   struct util_dynarray all_blocks;
};

/* The block_pool functions exported for testing only.  The block pool should
 * only be used via a state pool (see below).
 */
VkResult anv_block_pool_init(struct anv_block_pool *pool,
                             struct anv_device *device,
                             const char *name,
                             uint64_t start_address,
                             uint32_t initial_size);
void anv_block_pool_finish(struct anv_block_pool *pool);
int32_t anv_block_pool_alloc(struct anv_block_pool *pool,
                             uint32_t block_size, uint32_t *padding);
int32_t anv_block_pool_alloc_back(struct anv_block_pool *pool,
                                  uint32_t block_size);
void* anv_block_pool_map(struct anv_block_pool *pool, int32_t offset, uint32_t
size);

VkResult anv_state_pool_init(struct anv_state_pool *pool,
                             struct anv_device *device,
                             const char *name,
                             uint64_t base_address,
                             int32_t start_offset,
                             uint32_t block_size);
void anv_state_pool_finish(struct anv_state_pool *pool);
struct anv_state anv_state_pool_alloc(struct anv_state_pool *pool,
                                      uint32_t state_size, uint32_t alignment);
struct anv_state anv_state_pool_alloc_back(struct anv_state_pool *pool);
void anv_state_pool_free(struct anv_state_pool *pool, struct anv_state state);
void anv_state_stream_init(struct anv_state_stream *stream,
                           struct anv_state_pool *state_pool,
                           uint32_t block_size);
void anv_state_stream_finish(struct anv_state_stream *stream);
struct anv_state anv_state_stream_alloc(struct anv_state_stream *stream,
                                        uint32_t size, uint32_t alignment);

void anv_state_reserved_pool_init(struct anv_state_reserved_pool *pool,
                                      struct anv_state_pool *parent,
                                      uint32_t count, uint32_t size,
                                      uint32_t alignment);
void anv_state_reserved_pool_finish(struct anv_state_reserved_pool *pool);
struct anv_state anv_state_reserved_pool_alloc(struct anv_state_reserved_pool *pool);
void anv_state_reserved_pool_free(struct anv_state_reserved_pool *pool,
                                  struct anv_state state);

VkResult anv_state_table_init(struct anv_state_table *table,
                             struct anv_device *device,
                             uint32_t initial_entries);
void anv_state_table_finish(struct anv_state_table *table);
VkResult anv_state_table_add(struct anv_state_table *table, uint32_t *idx,
                             uint32_t count);
void anv_free_list_push(union anv_free_list *list,
                        struct anv_state_table *table,
                        uint32_t idx, uint32_t count);
struct anv_state* anv_free_list_pop(union anv_free_list *list,
                                    struct anv_state_table *table);


static inline struct anv_state *
anv_state_table_get(struct anv_state_table *table, uint32_t idx)
{
   return &table->map[idx].state;
}
/**
 * Implements a pool of re-usable BOs.  The interface is identical to that
 * of block_pool except that each block is its own BO.
 */
struct anv_bo_pool {
   const char *name;

   struct anv_device *device;

   struct util_sparse_array_free_list free_list[16];
};

void anv_bo_pool_init(struct anv_bo_pool *pool, struct anv_device *device,
                      const char *name);
void anv_bo_pool_finish(struct anv_bo_pool *pool);
VkResult anv_bo_pool_alloc(struct anv_bo_pool *pool, uint32_t size,
                           struct anv_bo **bo_out);
void anv_bo_pool_free(struct anv_bo_pool *pool, struct anv_bo *bo);

struct anv_scratch_pool {
   /* Indexed by Per-Thread Scratch Space number (the hardware value) and stage */
   struct anv_bo *bos[16][MESA_SHADER_STAGES];
};

void anv_scratch_pool_init(struct anv_device *device,
                           struct anv_scratch_pool *pool);
void anv_scratch_pool_finish(struct anv_device *device,
                             struct anv_scratch_pool *pool);
struct anv_bo *anv_scratch_pool_alloc(struct anv_device *device,
                                      struct anv_scratch_pool *pool,
                                      gl_shader_stage stage,
                                      unsigned per_thread_scratch);

/** Implements a BO cache that ensures a 1-1 mapping of GEM BOs to anv_bos */
struct anv_bo_cache {
   struct util_sparse_array bo_map;
   pthread_mutex_t mutex;
};

VkResult anv_bo_cache_init(struct anv_bo_cache *cache);
void anv_bo_cache_finish(struct anv_bo_cache *cache);

struct anv_queue_family {
   /* Standard bits passed on to the client */
   VkQueueFlags   queueFlags;
   uint32_t       queueCount;

   /* Driver internal information */
   enum drm_i915_gem_engine_class engine_class;
};

#define ANV_MAX_QUEUE_FAMILIES 3

struct anv_memory_type {
   /* Standard bits passed on to the client */
   VkMemoryPropertyFlags   propertyFlags;
   uint32_t                heapIndex;
};

struct anv_memory_heap {
   /* Standard bits passed on to the client */
   VkDeviceSize      size;
   VkMemoryHeapFlags flags;

   /** Driver-internal book-keeping.
    *
    * Align it to 64 bits to make atomic operations faster on 32 bit platforms.
    */
   VkDeviceSize      used __attribute__ ((aligned (8)));

   bool              is_local_mem;
};

struct anv_memregion {
   uint64_t size;
};

struct anv_physical_device {
    struct vk_physical_device                   vk;

    /* Link in anv_instance::physical_devices */
    struct list_head                            link;

    struct anv_instance *                       instance;
    bool                                        no_hw;
    char                                        path[20];
    const char *                                name;
    struct {
       uint16_t                                 domain;
       uint8_t                                  bus;
       uint8_t                                  device;
       uint8_t                                  function;
    }                                           pci_info;
    struct gen_device_info                      info;
    /** Amount of "GPU memory" we want to advertise
     *
     * Clearly, this value is bogus since Intel is a UMA architecture.  On
     * gfx7 platforms, we are limited by GTT size unless we want to implement
     * fine-grained tracking and GTT splitting.  On Broadwell and above we are
     * practically unlimited.  However, we will never report more than 3/4 of
     * the total system ram to try and avoid running out of RAM.
     */
    bool                                        supports_48bit_addresses;
    struct brw_compiler *                       compiler;
    struct isl_device                           isl_dev;
    struct gen_perf_config *                    perf;
    /*
     * Number of commands required to implement a performance query begin +
     * end.
     */
    uint32_t                                    n_perf_query_commands;
    int                                         cmd_parser_version;
    bool                                        has_softpin;
    bool                                        has_exec_async;
    bool                                        has_exec_capture;
    bool                                        has_exec_fence;
    bool                                        has_syncobj;
    bool                                        has_syncobj_wait;
    bool                                        has_syncobj_wait_available;
    bool                                        has_context_priority;
    bool                                        has_context_isolation;
    bool                                        has_thread_submit;
    bool                                        has_mem_available;
    bool                                        has_mmap_offset;
    uint64_t                                    gtt_size;

    bool                                        use_softpin;
    bool                                        always_use_bindless;
    bool                                        use_call_secondary;

    /** True if we can access buffers using A64 messages */
    bool                                        has_a64_buffer_access;
    /** True if we can use bindless access for images */
    bool                                        has_bindless_images;
    /** True if we can use bindless access for samplers */
    bool                                        has_bindless_samplers;
    /** True if we can use timeline semaphores through execbuf */
    bool                                        has_exec_timeline;

    /** True if we can read the GPU timestamp register
     *
     * When running in a virtual context, the timestamp register is unreadable
     * on Gfx12+.
     */
    bool                                        has_reg_timestamp;

    /** True if this device has implicit AUX
     *
     * If true, CCS is handled as an implicit attachment to the BO rather than
     * as an explicitly bound surface.
     */
    bool                                        has_implicit_ccs;

    bool                                        always_flush_cache;

    uint32_t                                    eu_total;
    uint32_t                                    subslice_total;

    struct {
      uint32_t                                  family_count;
      struct anv_queue_family                   families[ANV_MAX_QUEUE_FAMILIES];
    } queue;

    struct {
      uint32_t                                  type_count;
      struct anv_memory_type                    types[VK_MAX_MEMORY_TYPES];
      uint32_t                                  heap_count;
      struct anv_memory_heap                    heaps[VK_MAX_MEMORY_HEAPS];
    } memory;

    struct anv_memregion                        vram;
    struct anv_memregion                        sys;
    uint8_t                                     driver_build_sha1[20];
    uint8_t                                     pipeline_cache_uuid[VK_UUID_SIZE];
    uint8_t                                     driver_uuid[VK_UUID_SIZE];
    uint8_t                                     device_uuid[VK_UUID_SIZE];

    struct disk_cache *                         disk_cache;

    struct wsi_device                       wsi_device;
    int                                         local_fd;
    int                                         master_fd;
    struct drm_i915_query_engine_info *         engine_info;

    void (*cmd_emit_timestamp)(struct anv_batch *, struct anv_bo *, uint32_t );
    struct intel_measure_device                 measure_device;
};

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

struct anv_instance {
    struct vk_instance                          vk;

    bool                                        physical_devices_enumerated;
    struct list_head                            physical_devices;

    bool                                        pipeline_cache_enabled;

    struct driOptionCache                       dri_options;
    struct driOptionCache                       available_dri_options;
};

VkResult anv_init_wsi(struct anv_physical_device *physical_device);
void anv_finish_wsi(struct anv_physical_device *physical_device);

struct anv_queue_submit {
   struct anv_cmd_buffer **                  cmd_buffers;
   uint32_t                                  cmd_buffer_count;
   uint32_t                                  cmd_buffer_array_length;

   uint32_t                                  fence_count;
   uint32_t                                  fence_array_length;
   struct drm_i915_gem_exec_fence *          fences;
   uint64_t *                                fence_values;

   uint32_t                                  temporary_semaphore_count;
   uint32_t                                  temporary_semaphore_array_length;
   struct anv_semaphore_impl *               temporary_semaphores;

   /* Semaphores to be signaled with a SYNC_FD. */
   struct anv_semaphore **                   sync_fd_semaphores;
   uint32_t                                  sync_fd_semaphore_count;
   uint32_t                                  sync_fd_semaphore_array_length;

   /* Allocated only with non shareable timelines. */
   union {
      struct anv_timeline **                 wait_timelines;
      uint32_t *                             wait_timeline_syncobjs;
   };
   uint32_t                                  wait_timeline_count;
   uint32_t                                  wait_timeline_array_length;
   uint64_t *                                wait_timeline_values;

   struct anv_timeline **                    signal_timelines;
   uint32_t                                  signal_timeline_count;
   uint32_t                                  signal_timeline_array_length;
   uint64_t *                                signal_timeline_values;

   int                                       in_fence;
   bool                                      need_out_fence;
   int                                       out_fence;

   uint32_t                                  fence_bo_count;
   uint32_t                                  fence_bo_array_length;
   /* An array of struct anv_bo pointers with lower bit used as a flag to
    * signal we will wait on that BO (see anv_(un)pack_ptr).
    */
   uintptr_t *                               fence_bos;

   int                                       perf_query_pass;
   struct anv_query_pool *                   perf_query_pool;

   const VkAllocationCallbacks *             alloc;
   VkSystemAllocationScope                   alloc_scope;

   struct anv_bo *                           simple_bo;
   uint32_t                                  simple_bo_size;

   struct list_head                          link;
};

struct anv_queue {
   struct vk_object_base                     base;

   struct anv_device *                       device;

   VkDeviceQueueCreateFlags                  flags;
   const struct anv_queue_family *           family;

   uint32_t                                  exec_flags;

   /* Set once from the device api calls. */
   bool                                      lost_signaled;

   /* Only set once atomically by the queue */
   int                                       lost;
   int                                       error_line;
   const char *                              error_file;
   char                                      error_msg[80];

   /*
    * This mutext protects the variables below.
    */
   pthread_mutex_t                           mutex;

   pthread_t                                 thread;
   pthread_cond_t                            cond;

   /*
    * A list of struct anv_queue_submit to be submitted to i915.
    */
   struct list_head                          queued_submits;

   /* Set to true to stop the submission thread */
   bool                                      quit;
};

struct anv_pipeline_cache {
   struct vk_object_base                        base;
   struct anv_device *                          device;
   pthread_mutex_t                              mutex;

   struct hash_table *                          nir_cache;

   struct hash_table *                          cache;

   bool                                         external_sync;
};

struct nir_xfb_info;
struct anv_pipeline_bind_map;

void anv_pipeline_cache_init(struct anv_pipeline_cache *cache,
                             struct anv_device *device,
                             bool cache_enabled,
                             bool external_sync);
void anv_pipeline_cache_finish(struct anv_pipeline_cache *cache);

struct anv_shader_bin *
anv_pipeline_cache_search(struct anv_pipeline_cache *cache,
                          const void *key, uint32_t key_size);
struct anv_shader_bin *
anv_pipeline_cache_upload_kernel(struct anv_pipeline_cache *cache,
                                 gl_shader_stage stage,
                                 const void *key_data, uint32_t key_size,
                                 const void *kernel_data, uint32_t kernel_size,
                                 const struct brw_stage_prog_data *prog_data,
                                 uint32_t prog_data_size,
                                 const struct brw_compile_stats *stats,
                                 uint32_t num_stats,
                                 const struct nir_xfb_info *xfb_info,
                                 const struct anv_pipeline_bind_map *bind_map);

struct anv_shader_bin *
anv_device_search_for_kernel(struct anv_device *device,
                             struct anv_pipeline_cache *cache,
                             const void *key_data, uint32_t key_size,
                             bool *user_cache_bit);

struct anv_shader_bin *
anv_device_upload_kernel(struct anv_device *device,
                         struct anv_pipeline_cache *cache,
                         gl_shader_stage stage,
                         const void *key_data, uint32_t key_size,
                         const void *kernel_data, uint32_t kernel_size,
                         const struct brw_stage_prog_data *prog_data,
                         uint32_t prog_data_size,
                         const struct brw_compile_stats *stats,
                         uint32_t num_stats,
                         const struct nir_xfb_info *xfb_info,
                         const struct anv_pipeline_bind_map *bind_map);

struct nir_shader;
struct nir_shader_compiler_options;

struct nir_shader *
anv_device_search_for_nir(struct anv_device *device,
                          struct anv_pipeline_cache *cache,
                          const struct nir_shader_compiler_options *nir_options,
                          unsigned char sha1_key[20],
                          void *mem_ctx);

void
anv_device_upload_nir(struct anv_device *device,
                      struct anv_pipeline_cache *cache,
                      const struct nir_shader *nir,
                      unsigned char sha1_key[20]);

struct anv_address {
   struct anv_bo *bo;
   uint32_t offset;
};

struct anv_device {
    struct vk_device                            vk;

    struct anv_physical_device *                physical;
    bool                                        no_hw;
    struct gen_device_info                      info;
    struct isl_device                           isl_dev;
    int                                         context_id;
    int                                         fd;
    bool                                        can_chain_batches;
    bool                                        robust_buffer_access;
    bool                                        has_thread_submit;

    pthread_mutex_t                             vma_mutex;
    struct util_vma_heap                        vma_lo;
    struct util_vma_heap                        vma_cva;
    struct util_vma_heap                        vma_hi;

    /** List of all anv_device_memory objects */
    struct list_head                            memory_objects;

    struct anv_bo_pool                          batch_bo_pool;

    struct anv_bo_cache                         bo_cache;

    struct anv_state_pool                       general_state_pool;
    struct anv_state_pool                       dynamic_state_pool;
    struct anv_state_pool                       instruction_state_pool;
    struct anv_state_pool                       binding_table_pool;
    struct anv_state_pool                       surface_state_pool;

    struct anv_state_reserved_pool              custom_border_colors;

    /** BO used for various workarounds
     *
     * There are a number of workarounds on our hardware which require writing
     * data somewhere and it doesn't really matter where.  For that, we use
     * this BO and just write to the first dword or so.
     *
     * We also need to be able to handle NULL buffers bound as pushed UBOs.
     * For that, we use the high bytes (>= 1024) of the workaround BO.
     */
    struct anv_bo *                             workaround_bo;
    struct anv_address                          workaround_address;

    struct anv_bo *                             trivial_batch_bo;
    struct anv_bo *                             hiz_clear_bo;
    struct anv_state                            null_surface_state;

    struct anv_pipeline_cache                   default_pipeline_cache;
    struct blorp_context                        blorp;

    struct anv_state                            border_colors;

    struct anv_state                            slice_hash;

    uint32_t                                    queue_count;
    struct anv_queue  *                         queues;

    struct anv_scratch_pool                     scratch_pool;

    pthread_mutex_t                             mutex;
    pthread_cond_t                              queue_submit;
    int                                         _lost;
    int                                         lost_reported;

    struct intel_batch_decode_ctx               decoder_ctx;
    /*
     * When decoding a anv_cmd_buffer, we might need to search for BOs through
     * the cmd_buffer's list.
     */
    struct anv_cmd_buffer                      *cmd_buffer_being_decoded;

    int                                         perf_fd; /* -1 if no opened */
    uint64_t                                    perf_metric; /* 0 if unset */

    struct intel_aux_map_context                *aux_map_ctx;

    const struct intel_l3_config                *l3_config;

    struct gen_debug_block_frame                *debug_frame_desc;
};

static inline struct anv_instance *
anv_device_instance_or_null(const struct anv_device *device)
{
   return device ? device->physical->instance : NULL;
}

static inline struct anv_state_pool *
anv_binding_table_pool(struct anv_device *device)
{
   if (device->physical->use_softpin)
      return &device->binding_table_pool;
   else
      return &device->surface_state_pool;
}

static inline struct anv_state
anv_binding_table_pool_alloc(struct anv_device *device) {
   if (device->physical->use_softpin)
      return anv_state_pool_alloc(&device->binding_table_pool,
                                  device->binding_table_pool.block_size, 0);
   else
      return anv_state_pool_alloc_back(&device->surface_state_pool);
}

static inline void
anv_binding_table_pool_free(struct anv_device *device, struct anv_state state) {
   anv_state_pool_free(anv_binding_table_pool(device), state);
}

static inline uint32_t
anv_mocs(const struct anv_device *device,
         const struct anv_bo *bo,
         isl_surf_usage_flags_t usage)
{
   return isl_mocs(&device->isl_dev, usage, bo && bo->is_external);
}

void anv_device_init_blorp(struct anv_device *device);
void anv_device_finish_blorp(struct anv_device *device);

void _anv_device_report_lost(struct anv_device *device);
VkResult _anv_device_set_lost(struct anv_device *device,
                              const char *file, int line,
                              const char *msg, ...)
   anv_printflike(4, 5);
VkResult _anv_queue_set_lost(struct anv_queue *queue,
                              const char *file, int line,
                              const char *msg, ...)
   anv_printflike(4, 5);
#define anv_device_set_lost(dev, ...) \
   _anv_device_set_lost(dev, __FILE__, __LINE__, __VA_ARGS__)
#define anv_queue_set_lost(queue, ...) \
   (queue)->device->has_thread_submit ? \
   _anv_queue_set_lost(queue, __FILE__, __LINE__, __VA_ARGS__) : \
   _anv_device_set_lost(queue->device, __FILE__, __LINE__, __VA_ARGS__)

static inline bool
anv_device_is_lost(struct anv_device *device)
{
   int lost = p_atomic_read(&device->_lost);
   if (unlikely(lost && !device->lost_reported))
      _anv_device_report_lost(device);
   return lost;
}

VkResult anv_device_query_status(struct anv_device *device);


enum anv_bo_alloc_flags {
   /** Specifies that the BO must have a 32-bit address
    *
    * This is the opposite of EXEC_OBJECT_SUPPORTS_48B_ADDRESS.
    */
   ANV_BO_ALLOC_32BIT_ADDRESS =  (1 << 0),

   /** Specifies that the BO may be shared externally */
   ANV_BO_ALLOC_EXTERNAL =       (1 << 1),

   /** Specifies that the BO should be mapped */
   ANV_BO_ALLOC_MAPPED =         (1 << 2),

   /** Specifies that the BO should be snooped so we get coherency */
   ANV_BO_ALLOC_SNOOPED =        (1 << 3),

   /** Specifies that the BO should be captured in error states */
   ANV_BO_ALLOC_CAPTURE =        (1 << 4),

   /** Specifies that the BO will have an address assigned by the caller
    *
    * Such BOs do not exist in any VMA heap.
    */
   ANV_BO_ALLOC_FIXED_ADDRESS = (1 << 5),

   /** Enables implicit synchronization on the BO
    *
    * This is the opposite of EXEC_OBJECT_ASYNC.
    */
   ANV_BO_ALLOC_IMPLICIT_SYNC =  (1 << 6),

   /** Enables implicit synchronization on the BO
    *
    * This is equivalent to EXEC_OBJECT_WRITE.
    */
   ANV_BO_ALLOC_IMPLICIT_WRITE = (1 << 7),

   /** Has an address which is visible to the client */
   ANV_BO_ALLOC_CLIENT_VISIBLE_ADDRESS = (1 << 8),

   /** This buffer has implicit CCS data attached to it */
   ANV_BO_ALLOC_IMPLICIT_CCS = (1 << 9),
};

VkResult anv_device_alloc_bo(struct anv_device *device,
                             const char *name, uint64_t size,
                             enum anv_bo_alloc_flags alloc_flags,
                             uint64_t explicit_address,
                             struct anv_bo **bo);
VkResult anv_device_import_bo_from_host_ptr(struct anv_device *device,
                                            void *host_ptr, uint32_t size,
                                            enum anv_bo_alloc_flags alloc_flags,
                                            uint64_t client_address,
                                            struct anv_bo **bo_out);
VkResult anv_device_import_bo(struct anv_device *device, int fd,
                              enum anv_bo_alloc_flags alloc_flags,
                              uint64_t client_address,
                              struct anv_bo **bo);
VkResult anv_device_export_bo(struct anv_device *device,
                              struct anv_bo *bo, int *fd_out);
void anv_device_release_bo(struct anv_device *device,
                           struct anv_bo *bo);

static inline struct anv_bo *
anv_device_lookup_bo(struct anv_device *device, uint32_t gem_handle)
{
   return util_sparse_array_get(&device->bo_cache.bo_map, gem_handle);
}

VkResult anv_device_bo_busy(struct anv_device *device, struct anv_bo *bo);
VkResult anv_device_wait(struct anv_device *device, struct anv_bo *bo,
                         int64_t timeout);

VkResult anv_queue_init(struct anv_device *device, struct anv_queue *queue,
                        uint32_t exec_flags,
                        const VkDeviceQueueCreateInfo *pCreateInfo);
void anv_queue_finish(struct anv_queue *queue);

VkResult anv_queue_execbuf_locked(struct anv_queue *queue, struct anv_queue_submit *submit);
VkResult anv_queue_submit_simple_batch(struct anv_queue *queue,
                                       struct anv_batch *batch);

uint64_t anv_gettime_ns(void);
uint64_t anv_get_absolute_timeout(uint64_t timeout);

void* anv_gem_mmap(struct anv_device *device,
                   uint32_t gem_handle, uint64_t offset, uint64_t size, uint32_t flags);
void anv_gem_munmap(struct anv_device *device, void *p, uint64_t size);
uint32_t anv_gem_create(struct anv_device *device, uint64_t size);
void anv_gem_close(struct anv_device *device, uint32_t gem_handle);
uint32_t anv_gem_userptr(struct anv_device *device, void *mem, size_t size);
int anv_gem_busy(struct anv_device *device, uint32_t gem_handle);
int anv_gem_wait(struct anv_device *device, uint32_t gem_handle, int64_t *timeout_ns);
int anv_gem_execbuffer(struct anv_device *device,
                       struct drm_i915_gem_execbuffer2 *execbuf);
int anv_gem_set_tiling(struct anv_device *device, uint32_t gem_handle,
                       uint32_t stride, uint32_t tiling);
int anv_gem_create_context(struct anv_device *device);
int anv_gem_create_context_engines(struct anv_device *device,
                                   const struct drm_i915_query_engine_info *info,
                                   int num_engines,
                                   uint16_t *engine_classes);
bool anv_gem_has_context_priority(int fd);
int anv_gem_destroy_context(struct anv_device *device, int context);
int anv_gem_set_context_param(int fd, int context, uint32_t param,
                              uint64_t value);
int anv_gem_get_context_param(int fd, int context, uint32_t param,
                              uint64_t *value);
int anv_gem_get_param(int fd, uint32_t param);
uint64_t anv_gem_get_drm_cap(int fd, uint32_t capability);
int anv_gem_get_tiling(struct anv_device *device, uint32_t gem_handle);
bool anv_gem_get_bit6_swizzle(int fd, uint32_t tiling);
int anv_gem_context_get_reset_stats(int fd, int context,
                                    uint32_t *active, uint32_t *pending);
int anv_gem_handle_to_fd(struct anv_device *device, uint32_t gem_handle);
int anv_gem_reg_read(int fd, uint32_t offset, uint64_t *result);
uint32_t anv_gem_fd_to_handle(struct anv_device *device, int fd);
int anv_gem_set_caching(struct anv_device *device, uint32_t gem_handle, uint32_t caching);
int anv_gem_set_domain(struct anv_device *device, uint32_t gem_handle,
                       uint32_t read_domains, uint32_t write_domain);
int anv_gem_sync_file_merge(struct anv_device *device, int fd1, int fd2);
uint32_t anv_gem_syncobj_create(struct anv_device *device, uint32_t flags);
void anv_gem_syncobj_destroy(struct anv_device *device, uint32_t handle);
int anv_gem_syncobj_handle_to_fd(struct anv_device *device, uint32_t handle);
uint32_t anv_gem_syncobj_fd_to_handle(struct anv_device *device, int fd);
int anv_gem_syncobj_export_sync_file(struct anv_device *device,
                                     uint32_t handle);
int anv_gem_syncobj_import_sync_file(struct anv_device *device,
                                     uint32_t handle, int fd);
void anv_gem_syncobj_reset(struct anv_device *device, uint32_t handle);
bool anv_gem_supports_syncobj_wait(int fd);
int anv_gem_syncobj_wait(struct anv_device *device,
                         const uint32_t *handles, uint32_t num_handles,
                         int64_t abs_timeout_ns, bool wait_all);
int anv_gem_syncobj_timeline_wait(struct anv_device *device,
                                  const uint32_t *handles, const uint64_t *points,
                                  uint32_t num_items, int64_t abs_timeout_ns,
                                  bool wait_all, bool wait_materialize);
int anv_gem_syncobj_timeline_signal(struct anv_device *device,
                                    const uint32_t *handles, const uint64_t *points,
                                    uint32_t num_items);
int anv_gem_syncobj_timeline_query(struct anv_device *device,
                                   const uint32_t *handles, uint64_t *points,
                                   uint32_t num_items);
int anv_i915_query(int fd, uint64_t query_id, void *buffer,
                   int32_t *buffer_len);
struct drm_i915_query_engine_info *anv_gem_get_engine_info(int fd);
int anv_gem_count_engines(const struct drm_i915_query_engine_info *info,
                          uint16_t engine_class);

uint64_t anv_vma_alloc(struct anv_device *device,
                       uint64_t size, uint64_t align,
                       enum anv_bo_alloc_flags alloc_flags,
                       uint64_t client_address);
void anv_vma_free(struct anv_device *device,
                  uint64_t address, uint64_t size);

struct anv_reloc_list {
   uint32_t                                     num_relocs;
   uint32_t                                     array_length;
   struct drm_i915_gem_relocation_entry *       relocs;
   struct anv_bo **                             reloc_bos;
   uint32_t                                     dep_words;
   BITSET_WORD *                                deps;
};

VkResult anv_reloc_list_init(struct anv_reloc_list *list,
                             const VkAllocationCallbacks *alloc);
void anv_reloc_list_finish(struct anv_reloc_list *list,
                           const VkAllocationCallbacks *alloc);

VkResult anv_reloc_list_add(struct anv_reloc_list *list,
                            const VkAllocationCallbacks *alloc,
                            uint32_t offset, struct anv_bo *target_bo,
                            uint32_t delta, uint64_t *address_u64_out);

struct anv_batch_bo {
   /* Link in the anv_cmd_buffer.owned_batch_bos list */
   struct list_head                             link;

   struct anv_bo *                              bo;

   /* Bytes actually consumed in this batch BO */
   uint32_t                                     length;

   /* When this batch BO is used as part of a primary batch buffer, this
    * tracked whether it is chained to another primary batch buffer.
    *
    * If this is the case, the relocation list's last entry points the
    * location of the MI_BATCH_BUFFER_START chaining to the next batch.
    */
   bool                                         chained;

   struct anv_reloc_list                        relocs;
};

struct anv_batch {
   const VkAllocationCallbacks *                alloc;

   struct anv_address                           start_addr;

   void *                                       start;
   void *                                       end;
   void *                                       next;

   struct anv_reloc_list *                      relocs;

   /* This callback is called (with the associated user data) in the event
    * that the batch runs out of space.
    */
   VkResult (*extend_cb)(struct anv_batch *, void *);
   void *                                       user_data;

   /**
    * Current error status of the command buffer. Used to track inconsistent
    * or incomplete command buffer states that are the consequence of run-time
    * errors such as out of memory scenarios. We want to track this in the
    * batch because the command buffer object is not visible to some parts
    * of the driver.
    */
   VkResult                                     status;
};

void *anv_batch_emit_dwords(struct anv_batch *batch, int num_dwords);
void anv_batch_emit_batch(struct anv_batch *batch, struct anv_batch *other);
uint64_t anv_batch_emit_reloc(struct anv_batch *batch,
                              void *location, struct anv_bo *bo, uint32_t offset);
struct anv_address anv_batch_address(struct anv_batch *batch, void *batch_location);

static inline void
anv_batch_set_storage(struct anv_batch *batch, struct anv_address addr,
                      void *map, size_t size)
{
   batch->start_addr = addr;
   batch->next = batch->start = map;
   batch->end = map + size;
}

static inline VkResult
anv_batch_set_error(struct anv_batch *batch, VkResult error)
{
   assert(error != VK_SUCCESS);
   if (batch->status == VK_SUCCESS)
      batch->status = error;
   return batch->status;
}

static inline bool
anv_batch_has_error(struct anv_batch *batch)
{
   return batch->status != VK_SUCCESS;
}

#define ANV_NULL_ADDRESS ((struct anv_address) { NULL, 0 })

static inline bool
anv_address_is_null(struct anv_address addr)
{
   return addr.bo == NULL && addr.offset == 0;
}

static inline uint64_t
anv_address_physical(struct anv_address addr)
{
   if (addr.bo && (addr.bo->flags & EXEC_OBJECT_PINNED))
      return intel_canonical_address(addr.bo->offset + addr.offset);
   else
      return intel_canonical_address(addr.offset);
}

static inline struct anv_address
anv_address_add(struct anv_address addr, uint64_t offset)
{
   addr.offset += offset;
   return addr;
}

static inline void
write_reloc(const struct anv_device *device, void *p, uint64_t v, bool flush)
{
   unsigned reloc_size = 0;
   if (device->info.ver >= 8) {
      reloc_size = sizeof(uint64_t);
      *(uint64_t *)p = intel_canonical_address(v);
   } else {
      reloc_size = sizeof(uint32_t);
      *(uint32_t *)p = v;
   }

   if (flush && !device->info.has_llc)
      intel_flush_range(p, reloc_size);
}

static inline uint64_t
_anv_combine_address(struct anv_batch *batch, void *location,
                     const struct anv_address address, uint32_t delta)
{
   if (address.bo == NULL) {
      return address.offset + delta;
   } else {
      assert(batch->start <= location && location < batch->end);

      return anv_batch_emit_reloc(batch, location, address.bo, address.offset + delta);
   }
}

#define __gen_address_type struct anv_address
#define __gen_user_data struct anv_batch
#define __gen_combine_address _anv_combine_address

/* Wrapper macros needed to work around preprocessor argument issues.  In
 * particular, arguments don't get pre-evaluated if they are concatenated.
 * This means that, if you pass GENX(3DSTATE_PS) into the emit macro, the
 * GENX macro won't get evaluated if the emit macro contains "cmd ## foo".
 * We can work around this easily enough with these helpers.
 */
#define __anv_cmd_length(cmd) cmd ## _length
#define __anv_cmd_length_bias(cmd) cmd ## _length_bias
#define __anv_cmd_header(cmd) cmd ## _header
#define __anv_cmd_pack(cmd) cmd ## _pack
#define __anv_reg_num(reg) reg ## _num

#define anv_pack_struct(dst, struc, ...) do {                              \
      struct struc __template = {                                          \
         __VA_ARGS__                                                       \
      };                                                                   \
      __anv_cmd_pack(struc)(NULL, dst, &__template);                       \
      VG(VALGRIND_CHECK_MEM_IS_DEFINED(dst, __anv_cmd_length(struc) * 4)); \
   } while (0)

#define anv_batch_emitn(batch, n, cmd, ...) ({             \
      void *__dst = anv_batch_emit_dwords(batch, n);       \
      if (__dst) {                                         \
         struct cmd __template = {                         \
            __anv_cmd_header(cmd),                         \
           .DWordLength = n - __anv_cmd_length_bias(cmd),  \
            __VA_ARGS__                                    \
         };                                                \
         __anv_cmd_pack(cmd)(batch, __dst, &__template);   \
      }                                                    \
      __dst;                                               \
   })

#define anv_batch_emit_merge(batch, dwords0, dwords1)                   \
   do {                                                                 \
      uint32_t *dw;                                                     \
                                                                        \
      STATIC_ASSERT(ARRAY_SIZE(dwords0) == ARRAY_SIZE(dwords1));        \
      dw = anv_batch_emit_dwords((batch), ARRAY_SIZE(dwords0));         \
      if (!dw)                                                          \
         break;                                                         \
      for (uint32_t i = 0; i < ARRAY_SIZE(dwords0); i++)                \
         dw[i] = (dwords0)[i] | (dwords1)[i];                           \
      VG(VALGRIND_CHECK_MEM_IS_DEFINED(dw, ARRAY_SIZE(dwords0) * 4));\
   } while (0)

#define anv_batch_emit(batch, cmd, name)                            \
   for (struct cmd name = { __anv_cmd_header(cmd) },                    \
        *_dst = anv_batch_emit_dwords(batch, __anv_cmd_length(cmd));    \
        __builtin_expect(_dst != NULL, 1);                              \
        ({ __anv_cmd_pack(cmd)(batch, _dst, &name);                     \
           VG(VALGRIND_CHECK_MEM_IS_DEFINED(_dst, __anv_cmd_length(cmd) * 4)); \
           _dst = NULL;                                                 \
         }))

#define anv_batch_write_reg(batch, reg, name)                           \
   for (struct reg name = {}, *_cont = (struct reg *)1; _cont != NULL;  \
        ({                                                              \
            uint32_t _dw[__anv_cmd_length(reg)];                        \
            __anv_cmd_pack(reg)(NULL, _dw, &name);                      \
            for (unsigned i = 0; i < __anv_cmd_length(reg); i++) {      \
               anv_batch_emit(batch, GENX(MI_LOAD_REGISTER_IMM), lri) { \
                  lri.RegisterOffset   = __anv_reg_num(reg);            \
                  lri.DataDWord        = _dw[i];                        \
               }                                                        \
            }                                                           \
           _cont = NULL;                                                \
         }))

/* #define __gen_get_batch_dwords anv_batch_emit_dwords */
/* #define __gen_get_batch_address anv_batch_address */
/* #define __gen_address_value anv_address_physical */
/* #define __gen_address_offset anv_address_add */

struct anv_device_memory {
   struct vk_object_base                        base;

   struct list_head                             link;

   struct anv_bo *                              bo;
   const struct anv_memory_type *               type;
   VkDeviceSize                                 map_size;
   void *                                       map;

   /* If set, we are holding reference to AHardwareBuffer
    * which we must release when memory is freed.
    */
   struct AHardwareBuffer *                     ahw;

   /* If set, this memory comes from a host pointer. */
   void *                                       host_ptr;
};

/**
 * Header for Vertex URB Entry (VUE)
 */
struct anv_vue_header {
   uint32_t Reserved;
   uint32_t RTAIndex; /* RenderTargetArrayIndex */
   uint32_t ViewportIndex;
   float PointWidth;
};

/** Struct representing a sampled image descriptor
 *
 * This descriptor layout is used for sampled images, bare sampler, and
 * combined image/sampler descriptors.
 */
struct anv_sampled_image_descriptor {
   /** Bindless image handle
    *
    * This is expected to already be shifted such that the 20-bit
    * SURFACE_STATE table index is in the top 20 bits.
    */
   uint32_t image;

   /** Bindless sampler handle
    *
    * This is assumed to be a 32B-aligned SAMPLER_STATE pointer relative
    * to the dynamic state base address.
    */
   uint32_t sampler;
};

struct anv_texture_swizzle_descriptor {
   /** Texture swizzle
    *
    * See also nir_intrinsic_channel_select_intel
    */
   uint8_t swizzle[4];

   /** Unused padding to ensure the struct is a multiple of 64 bits */
   uint32_t _pad;
};

/** Struct representing a storage image descriptor */
struct anv_storage_image_descriptor {
   /** Bindless image handles
    *
    * These are expected to already be shifted such that the 20-bit
    * SURFACE_STATE table index is in the top 20 bits.
    */
   uint32_t read_write;
   uint32_t write_only;
};

/** Struct representing a address/range descriptor
 *
 * The fields of this struct correspond directly to the data layout of
 * nir_address_format_64bit_bounded_global addresses.  The last field is the
 * offset in the NIR address so it must be zero so that when you load the
 * descriptor you get a pointer to the start of the range.
 */
struct anv_address_range_descriptor {
   uint64_t address;
   uint32_t range;
   uint32_t zero;
};

enum anv_descriptor_data {
   /** The descriptor contains a BTI reference to a surface state */
   ANV_DESCRIPTOR_SURFACE_STATE  = (1 << 0),
   /** The descriptor contains a BTI reference to a sampler state */
   ANV_DESCRIPTOR_SAMPLER_STATE  = (1 << 1),
   /** The descriptor contains an actual buffer view */
   ANV_DESCRIPTOR_BUFFER_VIEW    = (1 << 2),
   /** The descriptor contains auxiliary image layout data */
   ANV_DESCRIPTOR_IMAGE_PARAM    = (1 << 3),
   /** The descriptor contains auxiliary image layout data */
   ANV_DESCRIPTOR_INLINE_UNIFORM = (1 << 4),
   /** anv_address_range_descriptor with a buffer address and range */
   ANV_DESCRIPTOR_ADDRESS_RANGE  = (1 << 5),
   /** Bindless surface handle */
   ANV_DESCRIPTOR_SAMPLED_IMAGE  = (1 << 6),
   /** Storage image handles */
   ANV_DESCRIPTOR_STORAGE_IMAGE  = (1 << 7),
   /** Storage image handles */
   ANV_DESCRIPTOR_TEXTURE_SWIZZLE  = (1 << 8),
};

struct anv_descriptor_set_binding_layout {
   /* The type of the descriptors in this binding */
   VkDescriptorType type;

   /* Flags provided when this binding was created */
   VkDescriptorBindingFlagsEXT flags;

   /* Bitfield representing the type of data this descriptor contains */
   enum anv_descriptor_data data;

   /* Maximum number of YCbCr texture/sampler planes */
   uint8_t max_plane_count;

   /* Number of array elements in this binding (or size in bytes for inline
    * uniform data)
    */
   uint16_t array_size;

   /* Index into the flattend descriptor set */
   uint16_t descriptor_index;

   /* Index into the dynamic state array for a dynamic buffer */
   int16_t dynamic_offset_index;

   /* Index into the descriptor set buffer views */
   int16_t buffer_view_index;

   /* Offset into the descriptor buffer where this descriptor lives */
   uint32_t descriptor_offset;

   /* Immutable samplers (or NULL if no immutable samplers) */
   struct anv_sampler **immutable_samplers;
};

unsigned anv_descriptor_size(const struct anv_descriptor_set_binding_layout *layout);

unsigned anv_descriptor_type_size(const struct anv_physical_device *pdevice,
                                  VkDescriptorType type);

bool anv_descriptor_supports_bindless(const struct anv_physical_device *pdevice,
                                      const struct anv_descriptor_set_binding_layout *binding,
                                      bool sampler);

bool anv_descriptor_requires_bindless(const struct anv_physical_device *pdevice,
                                      const struct anv_descriptor_set_binding_layout *binding,
                                      bool sampler);

struct anv_descriptor_set_layout {
   struct vk_object_base base;

   /* Descriptor set layouts can be destroyed at almost any time */
   uint32_t ref_cnt;

   /* Number of bindings in this descriptor set */
   uint16_t binding_count;

   /* Total number of descriptors */
   uint16_t descriptor_count;

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

   /* Number of buffer views in this descriptor set */
   uint16_t buffer_view_count;

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

   /* For each dynamic buffer, which VkShaderStageFlagBits stages are using
    * this buffer
    */
   VkShaderStageFlags dynamic_offset_stages[MAX_DYNAMIC_BUFFERS];

   /* Size of the descriptor buffer for this descriptor set */
   uint32_t descriptor_buffer_size;

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

void anv_descriptor_set_layout_destroy(struct anv_device *device,
                                       struct anv_descriptor_set_layout *layout);

static inline void
anv_descriptor_set_layout_ref(struct anv_descriptor_set_layout *layout)
{
   assert(layout && layout->ref_cnt >= 1);
   p_atomic_inc(&layout->ref_cnt);
}

static inline void
anv_descriptor_set_layout_unref(struct anv_device *device,
                                struct anv_descriptor_set_layout *layout)
{
   assert(layout && layout->ref_cnt >= 1);
   if (p_atomic_dec_zero(&layout->ref_cnt))
      anv_descriptor_set_layout_destroy(device, layout);
}

struct anv_descriptor {
   VkDescriptorType type;

   union {
      struct {
         VkImageLayout layout;
         struct anv_image_view *image_view;
         struct anv_sampler *sampler;
      };

      struct {
         struct anv_buffer *buffer;
         uint64_t offset;
         uint64_t range;
      };

      struct anv_buffer_view *buffer_view;
   };
};

struct anv_descriptor_set {
   struct vk_object_base base;

   struct anv_descriptor_pool *pool;
   struct anv_descriptor_set_layout *layout;

   /* Amount of space occupied in the the pool by this descriptor set. It can
    * be larger than the size of the descriptor set.
    */
   uint32_t size;

   /* State relative to anv_descriptor_pool::bo */
   struct anv_state desc_mem;
   /* Surface state for the descriptor buffer */
   struct anv_state desc_surface_state;

   uint32_t buffer_view_count;
   struct anv_buffer_view *buffer_views;

   /* Link to descriptor pool's desc_sets list . */
   struct list_head pool_link;

   uint32_t descriptor_count;
   struct anv_descriptor descriptors[0];
};

struct anv_buffer_view {
   struct vk_object_base base;

   enum isl_format format; /**< VkBufferViewCreateInfo::format */
   uint64_t range; /**< VkBufferViewCreateInfo::range */

   struct anv_address address;

   struct anv_state surface_state;
   struct anv_state storage_surface_state;
   struct anv_state writeonly_storage_surface_state;

   struct brw_image_param storage_image_param;
};

struct anv_push_descriptor_set {
   struct anv_descriptor_set set;

   /* Put this field right behind anv_descriptor_set so it fills up the
    * descriptors[0] field. */
   struct anv_descriptor descriptors[MAX_PUSH_DESCRIPTORS];

   /** True if the descriptor set buffer has been referenced by a draw or
    * dispatch command.
    */
   bool set_used_on_gpu;

   struct anv_buffer_view buffer_views[MAX_PUSH_DESCRIPTORS];
};

struct anv_descriptor_pool {
   struct vk_object_base base;

   uint32_t size;
   uint32_t next;
   uint32_t free_list;

   struct anv_bo *bo;
   struct util_vma_heap bo_heap;

   struct anv_state_stream surface_state_stream;
   void *surface_state_free_list;

   struct list_head desc_sets;

   char data[0];
};

enum anv_descriptor_template_entry_type {
   ANV_DESCRIPTOR_TEMPLATE_ENTRY_TYPE_IMAGE,
   ANV_DESCRIPTOR_TEMPLATE_ENTRY_TYPE_BUFFER,
   ANV_DESCRIPTOR_TEMPLATE_ENTRY_TYPE_BUFFER_VIEW
};

struct anv_descriptor_template_entry {
   /* The type of descriptor in this entry */
   VkDescriptorType type;

   /* Binding in the descriptor set */
   uint32_t binding;

   /* Offset at which to write into the descriptor set binding */
   uint32_t array_element;

   /* Number of elements to write into the descriptor set binding */
   uint32_t array_count;

   /* Offset into the user provided data */
   size_t offset;

   /* Stride between elements into the user provided data */
   size_t stride;
};

struct anv_descriptor_update_template {
    struct vk_object_base base;

    VkPipelineBindPoint bind_point;

   /* The descriptor set this template corresponds to. This value is only
    * valid if the template was created with the templateType
    * VK_DESCRIPTOR_UPDATE_TEMPLATE_TYPE_DESCRIPTOR_SET.
    */
   uint8_t set;

   /* Number of entries in this template */
   uint32_t entry_count;

   /* Entries of the template */
   struct anv_descriptor_template_entry entries[0];
};

size_t
anv_descriptor_set_layout_size(const struct anv_descriptor_set_layout *layout,
                               uint32_t var_desc_count);

uint32_t
anv_descriptor_set_layout_descriptor_buffer_size(const struct anv_descriptor_set_layout *set_layout,
                                                 uint32_t var_desc_count);

void
anv_descriptor_set_write_image_view(struct anv_device *device,
                                    struct anv_descriptor_set *set,
                                    const VkDescriptorImageInfo * const info,
                                    VkDescriptorType type,
                                    uint32_t binding,
                                    uint32_t element);

void
anv_descriptor_set_write_buffer_view(struct anv_device *device,
                                     struct anv_descriptor_set *set,
                                     VkDescriptorType type,
                                     struct anv_buffer_view *buffer_view,
                                     uint32_t binding,
                                     uint32_t element);

void
anv_descriptor_set_write_buffer(struct anv_device *device,
                                struct anv_descriptor_set *set,
                                struct anv_state_stream *alloc_stream,
                                VkDescriptorType type,
                                struct anv_buffer *buffer,
                                uint32_t binding,
                                uint32_t element,
                                VkDeviceSize offset,
                                VkDeviceSize range);
void
anv_descriptor_set_write_inline_uniform_data(struct anv_device *device,
                                             struct anv_descriptor_set *set,
                                             uint32_t binding,
                                             const void *data,
                                             size_t offset,
                                             size_t size);

void
anv_descriptor_set_write_template(struct anv_device *device,
                                  struct anv_descriptor_set *set,
                                  struct anv_state_stream *alloc_stream,
                                  const struct anv_descriptor_update_template *template,
                                  const void *data);

VkResult
anv_descriptor_set_create(struct anv_device *device,
                          struct anv_descriptor_pool *pool,
                          struct anv_descriptor_set_layout *layout,
                          uint32_t var_desc_count,
                          struct anv_descriptor_set **out_set);

void
anv_descriptor_set_destroy(struct anv_device *device,
                           struct anv_descriptor_pool *pool,
                           struct anv_descriptor_set *set);

#define ANV_DESCRIPTOR_SET_NULL             (UINT8_MAX - 5)
#define ANV_DESCRIPTOR_SET_PUSH_CONSTANTS   (UINT8_MAX - 4)
#define ANV_DESCRIPTOR_SET_DESCRIPTORS      (UINT8_MAX - 3)
#define ANV_DESCRIPTOR_SET_NUM_WORK_GROUPS  (UINT8_MAX - 2)
#define ANV_DESCRIPTOR_SET_SHADER_CONSTANTS (UINT8_MAX - 1)
#define ANV_DESCRIPTOR_SET_COLOR_ATTACHMENTS UINT8_MAX

struct anv_pipeline_binding {
   /** Index in the descriptor set
    *
    * This is a flattened index; the descriptor set layout is already taken
    * into account.
    */
   uint32_t index;

   /** The descriptor set this surface corresponds to.
    *
    * The special ANV_DESCRIPTOR_SET_* values above indicates that this
    * binding is not a normal descriptor set but something else.
    */
   uint8_t set;

   union {
      /** Plane in the binding index for images */
      uint8_t plane;

      /** Input attachment index (relative to the subpass) */
      uint8_t input_attachment_index;

      /** Dynamic offset index (for dynamic UBOs and SSBOs) */
      uint8_t dynamic_offset_index;
   };

   /** For a storage image, whether it is write-only */
   uint8_t write_only;

   /** Pad to 64 bits so that there are no holes and we can safely memcmp
    * assuming POD zero-initialization.
    */
   uint8_t pad;
};

struct anv_push_range {
   /** Index in the descriptor set */
   uint32_t index;

   /** Descriptor set index */
   uint8_t set;

   /** Dynamic offset index (for dynamic UBOs) */
   uint8_t dynamic_offset_index;

   /** Start offset in units of 32B */
   uint8_t start;

   /** Range in units of 32B */
   uint8_t length;
};

struct anv_pipeline_layout {
   struct vk_object_base base;

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

   uint32_t num_sets;

   unsigned char sha1[20];
};

struct anv_buffer {
   struct vk_object_base                        base;

   struct anv_device *                          device;
   VkDeviceSize                                 size;

   VkBufferCreateFlags                          create_flags;
   VkBufferUsageFlags                           usage;

   /* Set when bound */
   struct anv_address                           address;
};

static inline uint64_t
anv_buffer_get_range(struct anv_buffer *buffer, uint64_t offset, uint64_t range)
{
   assert(offset <= buffer->size);
   if (range == VK_WHOLE_SIZE) {
      return buffer->size - offset;
   } else {
      assert(range + offset >= range);
      assert(range + offset <= buffer->size);
      return range;
   }
}

enum anv_cmd_dirty_bits {
   ANV_CMD_DIRTY_DYNAMIC_VIEWPORT                    = 1 << 0, /* VK_DYNAMIC_STATE_VIEWPORT */
   ANV_CMD_DIRTY_DYNAMIC_SCISSOR                     = 1 << 1, /* VK_DYNAMIC_STATE_SCISSOR */
   ANV_CMD_DIRTY_DYNAMIC_LINE_WIDTH                  = 1 << 2, /* VK_DYNAMIC_STATE_LINE_WIDTH */
   ANV_CMD_DIRTY_DYNAMIC_DEPTH_BIAS                  = 1 << 3, /* VK_DYNAMIC_STATE_DEPTH_BIAS */
   ANV_CMD_DIRTY_DYNAMIC_BLEND_CONSTANTS             = 1 << 4, /* VK_DYNAMIC_STATE_BLEND_CONSTANTS */
   ANV_CMD_DIRTY_DYNAMIC_DEPTH_BOUNDS                = 1 << 5, /* VK_DYNAMIC_STATE_DEPTH_BOUNDS */
   ANV_CMD_DIRTY_DYNAMIC_STENCIL_COMPARE_MASK        = 1 << 6, /* VK_DYNAMIC_STATE_STENCIL_COMPARE_MASK */
   ANV_CMD_DIRTY_DYNAMIC_STENCIL_WRITE_MASK          = 1 << 7, /* VK_DYNAMIC_STATE_STENCIL_WRITE_MASK */
   ANV_CMD_DIRTY_DYNAMIC_STENCIL_REFERENCE           = 1 << 8, /* VK_DYNAMIC_STATE_STENCIL_REFERENCE */
   ANV_CMD_DIRTY_PIPELINE                            = 1 << 9,
   ANV_CMD_DIRTY_INDEX_BUFFER                        = 1 << 10,
   ANV_CMD_DIRTY_RENDER_TARGETS                      = 1 << 11,
   ANV_CMD_DIRTY_XFB_ENABLE                          = 1 << 12,
   ANV_CMD_DIRTY_DYNAMIC_LINE_STIPPLE                = 1 << 13, /* VK_DYNAMIC_STATE_LINE_STIPPLE_EXT */
   ANV_CMD_DIRTY_DYNAMIC_CULL_MODE                   = 1 << 14, /* VK_DYNAMIC_STATE_CULL_MODE_EXT */
   ANV_CMD_DIRTY_DYNAMIC_FRONT_FACE                  = 1 << 15, /* VK_DYNAMIC_STATE_FRONT_FACE_EXT */
   ANV_CMD_DIRTY_DYNAMIC_PRIMITIVE_TOPOLOGY          = 1 << 16, /* VK_DYNAMIC_STATE_PRIMITIVE_TOPOLOGY_EXT */
   ANV_CMD_DIRTY_DYNAMIC_VERTEX_INPUT_BINDING_STRIDE = 1 << 17, /* VK_DYNAMIC_STATE_VERTEX_INPUT_BINDING_STRIDE_EXT */
   ANV_CMD_DIRTY_DYNAMIC_DEPTH_TEST_ENABLE           = 1 << 18, /* VK_DYNAMIC_STATE_DEPTH_TEST_ENABLE_EXT */
   ANV_CMD_DIRTY_DYNAMIC_DEPTH_WRITE_ENABLE          = 1 << 19, /* VK_DYNAMIC_STATE_DEPTH_WRITE_ENABLE_EXT */
   ANV_CMD_DIRTY_DYNAMIC_DEPTH_COMPARE_OP            = 1 << 20, /* VK_DYNAMIC_STATE_DEPTH_COMPARE_OP_EXT */
   ANV_CMD_DIRTY_DYNAMIC_DEPTH_BOUNDS_TEST_ENABLE    = 1 << 21, /* VK_DYNAMIC_STATE_DEPTH_BOUNDS_TEST_ENABLE_EXT */
   ANV_CMD_DIRTY_DYNAMIC_STENCIL_TEST_ENABLE         = 1 << 22, /* VK_DYNAMIC_STATE_STENCIL_TEST_ENABLE_EXT */
   ANV_CMD_DIRTY_DYNAMIC_STENCIL_OP                  = 1 << 23, /* VK_DYNAMIC_STATE_STENCIL_OP_EXT */
   ANV_CMD_DIRTY_DYNAMIC_SAMPLE_LOCATIONS            = 1 << 24, /* VK_DYNAMIC_STATE_SAMPLE_LOCATIONS_EXT */
};
typedef uint32_t anv_cmd_dirty_mask_t;

#define ANV_CMD_DIRTY_DYNAMIC_ALL                       \
   (ANV_CMD_DIRTY_DYNAMIC_VIEWPORT |                    \
    ANV_CMD_DIRTY_DYNAMIC_SCISSOR |                     \
    ANV_CMD_DIRTY_DYNAMIC_LINE_WIDTH |                  \
    ANV_CMD_DIRTY_DYNAMIC_DEPTH_BIAS |                  \
    ANV_CMD_DIRTY_DYNAMIC_BLEND_CONSTANTS |             \
    ANV_CMD_DIRTY_DYNAMIC_DEPTH_BOUNDS |                \
    ANV_CMD_DIRTY_DYNAMIC_STENCIL_COMPARE_MASK |        \
    ANV_CMD_DIRTY_DYNAMIC_STENCIL_WRITE_MASK |          \
    ANV_CMD_DIRTY_DYNAMIC_STENCIL_REFERENCE |           \
    ANV_CMD_DIRTY_DYNAMIC_LINE_STIPPLE |                \
    ANV_CMD_DIRTY_DYNAMIC_CULL_MODE |                   \
    ANV_CMD_DIRTY_DYNAMIC_FRONT_FACE |                  \
    ANV_CMD_DIRTY_DYNAMIC_PRIMITIVE_TOPOLOGY |          \
    ANV_CMD_DIRTY_DYNAMIC_VERTEX_INPUT_BINDING_STRIDE | \
    ANV_CMD_DIRTY_DYNAMIC_DEPTH_TEST_ENABLE |           \
    ANV_CMD_DIRTY_DYNAMIC_DEPTH_WRITE_ENABLE |          \
    ANV_CMD_DIRTY_DYNAMIC_DEPTH_COMPARE_OP |            \
    ANV_CMD_DIRTY_DYNAMIC_DEPTH_BOUNDS_TEST_ENABLE |    \
    ANV_CMD_DIRTY_DYNAMIC_STENCIL_TEST_ENABLE |         \
    ANV_CMD_DIRTY_DYNAMIC_STENCIL_OP |                  \
    ANV_CMD_DIRTY_DYNAMIC_SAMPLE_LOCATIONS)

static inline enum anv_cmd_dirty_bits
anv_cmd_dirty_bit_for_vk_dynamic_state(VkDynamicState vk_state)
{
   switch (vk_state) {
   case VK_DYNAMIC_STATE_VIEWPORT:
   case VK_DYNAMIC_STATE_VIEWPORT_WITH_COUNT_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_VIEWPORT;
   case VK_DYNAMIC_STATE_SCISSOR:
   case VK_DYNAMIC_STATE_SCISSOR_WITH_COUNT_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_SCISSOR;
   case VK_DYNAMIC_STATE_LINE_WIDTH:
      return ANV_CMD_DIRTY_DYNAMIC_LINE_WIDTH;
   case VK_DYNAMIC_STATE_DEPTH_BIAS:
      return ANV_CMD_DIRTY_DYNAMIC_DEPTH_BIAS;
   case VK_DYNAMIC_STATE_BLEND_CONSTANTS:
      return ANV_CMD_DIRTY_DYNAMIC_BLEND_CONSTANTS;
   case VK_DYNAMIC_STATE_DEPTH_BOUNDS:
      return ANV_CMD_DIRTY_DYNAMIC_DEPTH_BOUNDS;
   case VK_DYNAMIC_STATE_STENCIL_COMPARE_MASK:
      return ANV_CMD_DIRTY_DYNAMIC_STENCIL_COMPARE_MASK;
   case VK_DYNAMIC_STATE_STENCIL_WRITE_MASK:
      return ANV_CMD_DIRTY_DYNAMIC_STENCIL_WRITE_MASK;
   case VK_DYNAMIC_STATE_STENCIL_REFERENCE:
      return ANV_CMD_DIRTY_DYNAMIC_STENCIL_REFERENCE;
   case VK_DYNAMIC_STATE_LINE_STIPPLE_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_LINE_STIPPLE;
   case VK_DYNAMIC_STATE_CULL_MODE_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_CULL_MODE;
   case VK_DYNAMIC_STATE_FRONT_FACE_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_FRONT_FACE;
   case VK_DYNAMIC_STATE_PRIMITIVE_TOPOLOGY_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_PRIMITIVE_TOPOLOGY;
   case VK_DYNAMIC_STATE_VERTEX_INPUT_BINDING_STRIDE_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_VERTEX_INPUT_BINDING_STRIDE;
   case VK_DYNAMIC_STATE_DEPTH_TEST_ENABLE_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_DEPTH_TEST_ENABLE;
   case VK_DYNAMIC_STATE_DEPTH_WRITE_ENABLE_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_DEPTH_WRITE_ENABLE;
   case VK_DYNAMIC_STATE_DEPTH_COMPARE_OP_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_DEPTH_COMPARE_OP;
   case VK_DYNAMIC_STATE_DEPTH_BOUNDS_TEST_ENABLE_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_DEPTH_BOUNDS_TEST_ENABLE;
   case VK_DYNAMIC_STATE_STENCIL_TEST_ENABLE_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_STENCIL_TEST_ENABLE;
   case VK_DYNAMIC_STATE_STENCIL_OP_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_STENCIL_OP;
   case VK_DYNAMIC_STATE_SAMPLE_LOCATIONS_EXT:
      return ANV_CMD_DIRTY_DYNAMIC_SAMPLE_LOCATIONS;
   default:
      assert(!"Unsupported dynamic state");
      return 0;
   }
}


enum anv_pipe_bits {
   ANV_PIPE_DEPTH_CACHE_FLUSH_BIT            = (1 << 0),
   ANV_PIPE_STALL_AT_SCOREBOARD_BIT          = (1 << 1),
   ANV_PIPE_STATE_CACHE_INVALIDATE_BIT       = (1 << 2),
   ANV_PIPE_CONSTANT_CACHE_INVALIDATE_BIT    = (1 << 3),
   ANV_PIPE_VF_CACHE_INVALIDATE_BIT          = (1 << 4),
   ANV_PIPE_DATA_CACHE_FLUSH_BIT             = (1 << 5),
   ANV_PIPE_TILE_CACHE_FLUSH_BIT             = (1 << 6),
   ANV_PIPE_TEXTURE_CACHE_INVALIDATE_BIT     = (1 << 10),
   ANV_PIPE_INSTRUCTION_CACHE_INVALIDATE_BIT = (1 << 11),
   ANV_PIPE_RENDER_TARGET_CACHE_FLUSH_BIT    = (1 << 12),
   ANV_PIPE_DEPTH_STALL_BIT                  = (1 << 13),
   ANV_PIPE_CS_STALL_BIT                     = (1 << 20),
   ANV_PIPE_END_OF_PIPE_SYNC_BIT             = (1 << 21),

   /* This bit does not exist directly in PIPE_CONTROL.  Instead it means that
    * a flush has happened but not a CS stall.  The next time we do any sort
    * of invalidation we need to insert a CS stall at that time.  Otherwise,
    * we would have to CS stall on every flush which could be bad.
    */
   ANV_PIPE_NEEDS_END_OF_PIPE_SYNC_BIT       = (1 << 22),

   /* This bit does not exist directly in PIPE_CONTROL. It means that render
    * target operations related to transfer commands with VkBuffer as
    * destination are ongoing. Some operations like copies on the command
    * streamer might need to be aware of this to trigger the appropriate stall
    * before they can proceed with the copy.
    */
   ANV_PIPE_RENDER_TARGET_BUFFER_WRITES      = (1 << 23),

   /* This bit does not exist directly in PIPE_CONTROL. It means that Gfx12
    * AUX-TT data has changed and we need to invalidate AUX-TT data.  This is
    * done by writing the AUX-TT register.
    */
   ANV_PIPE_AUX_TABLE_INVALIDATE_BIT         = (1 << 24),

   /* This bit does not exist directly in PIPE_CONTROL. It means that a
    * PIPE_CONTROL with a post-sync operation will follow. This is used to
    * implement a workaround for Gfx9.
    */
   ANV_PIPE_POST_SYNC_BIT                    = (1 << 25),
};

#define ANV_PIPE_FLUSH_BITS ( \
   ANV_PIPE_DEPTH_CACHE_FLUSH_BIT | \
   ANV_PIPE_DATA_CACHE_FLUSH_BIT | \
   ANV_PIPE_RENDER_TARGET_CACHE_FLUSH_BIT | \
   ANV_PIPE_TILE_CACHE_FLUSH_BIT)

#define ANV_PIPE_STALL_BITS ( \
   ANV_PIPE_STALL_AT_SCOREBOARD_BIT | \
   ANV_PIPE_DEPTH_STALL_BIT | \
   ANV_PIPE_CS_STALL_BIT)

#define ANV_PIPE_INVALIDATE_BITS ( \
   ANV_PIPE_STATE_CACHE_INVALIDATE_BIT | \
   ANV_PIPE_CONSTANT_CACHE_INVALIDATE_BIT | \
   ANV_PIPE_VF_CACHE_INVALIDATE_BIT | \
   ANV_PIPE_DATA_CACHE_FLUSH_BIT | \
   ANV_PIPE_TEXTURE_CACHE_INVALIDATE_BIT | \
   ANV_PIPE_INSTRUCTION_CACHE_INVALIDATE_BIT | \
   ANV_PIPE_AUX_TABLE_INVALIDATE_BIT)

static inline enum anv_pipe_bits
anv_pipe_flush_bits_for_access_flags(struct anv_device *device,
                                     VkAccessFlags flags)
{
   enum anv_pipe_bits pipe_bits = 0;

   u_foreach_bit(b, flags) {
      switch ((VkAccessFlagBits)(1 << b)) {
      case VK_ACCESS_SHADER_WRITE_BIT:
         /* We're transitioning a buffer that was previously used as write
          * destination through the data port. To make its content available
          * to future operations, flush the data cache.
          */
         pipe_bits |= ANV_PIPE_DATA_CACHE_FLUSH_BIT;
         break;
      case VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT:
         /* We're transitioning a buffer that was previously used as render
          * target. To make its content available to future operations, flush
          * the render target cache.
          */
         pipe_bits |= ANV_PIPE_RENDER_TARGET_CACHE_FLUSH_BIT;
         break;
      case VK_ACCESS_DEPTH_STENCIL_ATTACHMENT_WRITE_BIT:
         /* We're transitioning a buffer that was previously used as depth
          * buffer. To make its content available to future operations, flush
          * the depth cache.
          */
         pipe_bits |= ANV_PIPE_DEPTH_CACHE_FLUSH_BIT;
         break;
      case VK_ACCESS_TRANSFER_WRITE_BIT:
         /* We're transitioning a buffer that was previously used as a
          * transfer write destination. Generic write operations include color
          * & depth operations as well as buffer operations like :
          *     - vkCmdClearColorImage()
          *     - vkCmdClearDepthStencilImage()
          *     - vkCmdBlitImage()
          *     - vkCmdCopy*(), vkCmdUpdate*(), vkCmdFill*()
          *
          * Most of these operations are implemented using Blorp which writes
          * through the render target, so flush that cache to make it visible
          * to future operations. And for depth related operations we also
          * need to flush the depth cache.
          */
         pipe_bits |= ANV_PIPE_RENDER_TARGET_CACHE_FLUSH_BIT;
         pipe_bits |= ANV_PIPE_DEPTH_CACHE_FLUSH_BIT;
         break;
      case VK_ACCESS_MEMORY_WRITE_BIT:
         /* We're transitioning a buffer for generic write operations. Flush
          * all the caches.
          */
         pipe_bits |= ANV_PIPE_FLUSH_BITS;
         break;
      default:
         break; /* Nothing to do */
      }
   }

   return pipe_bits;
}

static inline enum anv_pipe_bits
anv_pipe_invalidate_bits_for_access_flags(struct anv_device *device,
                                          VkAccessFlags flags)
{
   enum anv_pipe_bits pipe_bits = 0;

   u_foreach_bit(b, flags) {
      switch ((VkAccessFlagBits)(1 << b)) {
      case VK_ACCESS_INDIRECT_COMMAND_READ_BIT:
         /* Indirect draw commands take a buffer as input that we're going to
          * read from the command streamer to load some of the HW registers
          * (see genX_cmd_buffer.c:load_indirect_parameters). This requires a
          * command streamer stall so that all the cache flushes have
          * completed before the command streamer loads from memory.
          */
         pipe_bits |=  ANV_PIPE_CS_STALL_BIT;
         /* Indirect draw commands also set gl_BaseVertex & gl_BaseIndex
          * through a vertex buffer, so invalidate that cache.
          */
         pipe_bits |= ANV_PIPE_VF_CACHE_INVALIDATE_BIT;
         /* For CmdDipatchIndirect, we also load gl_NumWorkGroups through a
          * UBO from the buffer, so we need to invalidate constant cache.
          */
         pipe_bits |= ANV_PIPE_CONSTANT_CACHE_INVALIDATE_BIT;
         break;
      case VK_ACCESS_INDEX_READ_BIT:
      case VK_ACCESS_VERTEX_ATTRIBUTE_READ_BIT:
         /* We transitioning a buffer to be used for as input for vkCmdDraw*
          * commands, so we invalidate the VF cache to make sure there is no
          * stale data when we start rendering.
          */
         pipe_bits |= ANV_PIPE_VF_CACHE_INVALIDATE_BIT;
         break;
      case VK_ACCESS_UNIFORM_READ_BIT:
         /* We transitioning a buffer to be used as uniform data. Because
          * uniform is accessed through the data port & sampler, we need to
          * invalidate the texture cache (sampler) & constant cache (data
          * port) to avoid stale data.
          */
         pipe_bits |= ANV_PIPE_CONSTANT_CACHE_INVALIDATE_BIT;
         if (device->physical->compiler->indirect_ubos_use_sampler)
            pipe_bits |= ANV_PIPE_TEXTURE_CACHE_INVALIDATE_BIT;
         else
            pipe_bits |= ANV_PIPE_DATA_CACHE_FLUSH_BIT;
         break;
      case VK_ACCESS_SHADER_READ_BIT:
      case VK_ACCESS_INPUT_ATTACHMENT_READ_BIT:
      case VK_ACCESS_TRANSFER_READ_BIT:
         /* Transitioning a buffer to be read through the sampler, so
          * invalidate the texture cache, we don't want any stale data.
          */
         pipe_bits |= ANV_PIPE_TEXTURE_CACHE_INVALIDATE_BIT;
         break;
      case VK_ACCESS_MEMORY_READ_BIT:
         /* Transitioning a buffer for generic read, invalidate all the
          * caches.
          */
         pipe_bits |= ANV_PIPE_INVALIDATE_BITS;
         break;
      case VK_ACCESS_MEMORY_WRITE_BIT:
         /* Generic write, make sure all previously written things land in
          * memory.
          */
         pipe_bits |= ANV_PIPE_FLUSH_BITS;
         break;
      case VK_ACCESS_CONDITIONAL_RENDERING_READ_BIT_EXT:
         /* Transitioning a buffer for conditional rendering. We'll load the
          * content of this buffer into HW registers using the command
          * streamer, so we need to stall the command streamer to make sure
          * any in-flight flush operations have completed.
          */
         pipe_bits |= ANV_PIPE_CS_STALL_BIT;
         break;
      default:
         break; /* Nothing to do */
      }
   }

   return pipe_bits;
}

#define VK_IMAGE_ASPECT_ANY_COLOR_BIT_ANV (         \
   VK_IMAGE_ASPECT_COLOR_BIT | \
   VK_IMAGE_ASPECT_PLANE_0_BIT | \
   VK_IMAGE_ASPECT_PLANE_1_BIT | \
   VK_IMAGE_ASPECT_PLANE_2_BIT)
#define VK_IMAGE_ASPECT_PLANES_BITS_ANV ( \
   VK_IMAGE_ASPECT_PLANE_0_BIT | \
   VK_IMAGE_ASPECT_PLANE_1_BIT | \
   VK_IMAGE_ASPECT_PLANE_2_BIT)

struct anv_vertex_binding {
   struct anv_buffer *                          buffer;
   VkDeviceSize                                 offset;
   VkDeviceSize                                 stride;
   VkDeviceSize                                 size;
};

struct anv_xfb_binding {
   struct anv_buffer *                          buffer;
   VkDeviceSize                                 offset;
   VkDeviceSize                                 size;
};

struct anv_push_constants {
   /** Push constant data provided by the client through vkPushConstants */
   uint8_t client_data[MAX_PUSH_CONSTANTS_SIZE];

   /** Dynamic offsets for dynamic UBOs and SSBOs */
   uint32_t dynamic_offsets[MAX_DYNAMIC_BUFFERS];

   /* Robust access pushed registers. */
   uint64_t push_reg_mask[MESA_SHADER_STAGES];

   /** Pad out to a multiple of 32 bytes */
   uint32_t pad[2];

   struct {
      /** Base workgroup ID
       *
       * Used for vkCmdDispatchBase.
       */
      uint32_t base_work_group_id[3];

      /** Subgroup ID
       *
       * This is never set by software but is implicitly filled out when
       * uploading the push constants for compute shaders.
       */
      uint32_t subgroup_id;
   } cs;
};

struct anv_dynamic_state {
   struct {
      uint32_t                                  count;
      VkViewport                                viewports[MAX_VIEWPORTS];
   } viewport;

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

   float                                        line_width;

   struct {
      float                                     bias;
      float                                     clamp;
      float                                     slope;
   } depth_bias;

   float                                        blend_constants[4];

   struct {
      float                                     min;
      float                                     max;
   } depth_bounds;

   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;

   struct {
      struct {
         VkStencilOp fail_op;
         VkStencilOp pass_op;
         VkStencilOp depth_fail_op;
         VkCompareOp compare_op;
      } front;
      struct {
         VkStencilOp fail_op;
         VkStencilOp pass_op;
         VkStencilOp depth_fail_op;
         VkCompareOp compare_op;
      } back;
   } stencil_op;

   struct {
      uint32_t                                  factor;
      uint16_t                                  pattern;
   } line_stipple;

   struct {
      uint32_t                                  samples;
      VkSampleLocationEXT                       locations[MAX_SAMPLE_LOCATIONS];
   } sample_locations;

   VkCullModeFlags                              cull_mode;
   VkFrontFace                                  front_face;
   VkPrimitiveTopology                          primitive_topology;
   bool                                         depth_test_enable;
   bool                                         depth_write_enable;
   VkCompareOp                                  depth_compare_op;
   bool                                         depth_bounds_test_enable;
   bool                                         stencil_test_enable;
   bool                                         dyn_vbo_stride;
   bool                                         dyn_vbo_size;
};

extern const struct anv_dynamic_state default_dynamic_state;

uint32_t anv_dynamic_state_copy(struct anv_dynamic_state *dest,
                                const struct anv_dynamic_state *src,
                                uint32_t copy_mask);

struct anv_surface_state {
   struct anv_state state;
   /** Address of the surface referred to by this state
    *
    * This address is relative to the start of the BO.
    */
   struct anv_address address;
   /* Address of the aux surface, if any
    *
    * This field is ANV_NULL_ADDRESS if and only if no aux surface exists.
    *
    * With the exception of gfx8, the bottom 12 bits of this address' offset
    * include extra aux information.
    */
   struct anv_address aux_address;
   /* Address of the clear color, if any
    *
    * This address is relative to the start of the BO.
    */
   struct anv_address clear_address;
};

/**
 * Attachment state when recording a renderpass instance.
 *
 * The clear value is valid only if there exists a pending clear.
 */
struct anv_attachment_state {
   enum isl_aux_usage                           aux_usage;
   struct anv_surface_state                     color;
   struct anv_surface_state                     input;

   VkImageLayout                                current_layout;
   VkImageLayout                                current_stencil_layout;
   VkImageAspectFlags                           pending_clear_aspects;
   VkImageAspectFlags                           pending_load_aspects;
   bool                                         fast_clear;
   VkClearValue                                 clear_value;

   /* When multiview is active, attachments with a renderpass clear
    * operation have their respective layers cleared on the first
    * subpass that uses them, and only in that subpass. We keep track
    * of this using a bitfield to indicate which layers of an attachment
    * have not been cleared yet when multiview is active.
    */
   uint32_t                                     pending_clear_views;
   struct anv_image_view *                      image_view;
};

/** State tracking for vertex buffer flushes
 *
 * On Gfx8-9, the VF cache only considers the bottom 32 bits of memory
 * addresses.  If you happen to have two vertex buffers which get placed
 * exactly 4 GiB apart and use them in back-to-back draw calls, you can get
 * collisions.  In order to solve this problem, we track vertex address ranges
 * which are live in the cache and invalidate the cache if one ever exceeds 32
 * bits.
 */
struct anv_vb_cache_range {
   /* Virtual address at which the live vertex buffer cache range starts for
    * this vertex buffer index.
    */
   uint64_t start;

   /* Virtual address of the byte after where vertex buffer cache range ends.
    * This is exclusive such that end - start is the size of the range.
    */
   uint64_t end;
};

/** State tracking for particular pipeline bind point
 *
 * This struct is the base struct for anv_cmd_graphics_state and
 * anv_cmd_compute_state.  These are used to track state which is bound to a
 * particular type of pipeline.  Generic state that applies per-stage such as
 * binding table offsets and push constants is tracked generically with a
 * per-stage array in anv_cmd_state.
 */
struct anv_cmd_pipeline_state {
   struct anv_descriptor_set *descriptors[MAX_SETS];
   struct anv_push_descriptor_set *push_descriptors[MAX_SETS];

   struct anv_push_constants push_constants;

   /* Push constant state allocated when flushing push constants. */
   struct anv_state          push_constants_state;
};

/** State tracking for graphics pipeline
 *
 * This has anv_cmd_pipeline_state as a base struct to track things which get
 * bound to a graphics pipeline.  Along with general pipeline bind point state
 * which is in the anv_cmd_pipeline_state base struct, it also contains other
 * state which is graphics-specific.
 */
struct anv_cmd_graphics_state {
   struct anv_cmd_pipeline_state base;

   struct anv_graphics_pipeline *pipeline;

   anv_cmd_dirty_mask_t dirty;
   uint32_t vb_dirty;

   struct anv_vb_cache_range ib_bound_range;
   struct anv_vb_cache_range ib_dirty_range;
   struct anv_vb_cache_range vb_bound_ranges[33];
   struct anv_vb_cache_range vb_dirty_ranges[33];

   VkShaderStageFlags push_constant_stages;

   struct anv_dynamic_state dynamic;

   uint32_t primitive_topology;

   struct {
      struct anv_buffer *index_buffer;
      uint32_t index_type; /**< 3DSTATE_INDEX_BUFFER.IndexFormat */
      uint32_t index_offset;
   } gfx7;
};

/** State tracking for compute pipeline
 *
 * This has anv_cmd_pipeline_state as a base struct to track things which get
 * bound to a compute pipeline.  Along with general pipeline bind point state
 * which is in the anv_cmd_pipeline_state base struct, it also contains other
 * state which is compute-specific.
 */
struct anv_cmd_compute_state {
   struct anv_cmd_pipeline_state base;

   struct anv_compute_pipeline *pipeline;

   bool pipeline_dirty;

   struct anv_state push_data;

   struct anv_address num_workgroups;
};

/** State required while building cmd buffer */
struct anv_cmd_state {
   /* PIPELINE_SELECT.PipelineSelection */
   uint32_t                                     current_pipeline;
   const struct intel_l3_config *               current_l3_config;
   uint32_t                                     last_aux_map_state;

   struct anv_cmd_graphics_state                gfx;
   struct anv_cmd_compute_state                 compute;

   enum anv_pipe_bits                           pending_pipe_bits;
   VkShaderStageFlags                           descriptors_dirty;
   VkShaderStageFlags                           push_constants_dirty;

   struct anv_framebuffer *                     framebuffer;
   struct anv_render_pass *                     pass;
   struct anv_subpass *                         subpass;
   VkRect2D                                     render_area;
   uint32_t                                     restart_index;
   struct anv_vertex_binding                    vertex_bindings[MAX_VBS];
   bool                                         xfb_enabled;
   struct anv_xfb_binding                       xfb_bindings[MAX_XFB_BUFFERS];
   struct anv_state                             binding_tables[MESA_SHADER_STAGES];
   struct anv_state                             samplers[MESA_SHADER_STAGES];

   unsigned char                                sampler_sha1s[MESA_SHADER_STAGES][20];
   unsigned char                                surface_sha1s[MESA_SHADER_STAGES][20];
   unsigned char                                push_sha1s[MESA_SHADER_STAGES][20];

   /**
    * Whether or not the gfx8 PMA fix is enabled.  We ensure that, at the top
    * of any command buffer it is disabled by disabling it in EndCommandBuffer
    * and before invoking the secondary in ExecuteCommands.
    */
   bool                                         pma_fix_enabled;

   /**
    * Whether or not we know for certain that HiZ is enabled for the current
    * subpass.  If, for whatever reason, we are unsure as to whether HiZ is
    * enabled or not, this will be false.
    */
   bool                                         hiz_enabled;

   bool                                         conditional_render_enabled;

   /**
    * Last rendering scale argument provided to
    * genX(cmd_buffer_emit_hashing_mode)().
    */
   unsigned                                     current_hash_scale;

   /**
    * Array length is anv_cmd_state::pass::attachment_count. Array content is
    * valid only when recording a render pass instance.
    */
   struct anv_attachment_state *                attachments;

   /**
    * Surface states for color render targets.  These are stored in a single
    * flat array.  For depth-stencil attachments, the surface state is simply
    * left blank.
    */
   struct anv_state                             attachment_states;

   /**
    * A null surface state of the right size to match the framebuffer.  This
    * is one of the states in attachment_states.
    */
   struct anv_state                             null_surface_state;
};

struct anv_cmd_pool {
   struct vk_object_base                        base;
   VkAllocationCallbacks                        alloc;
   struct list_head                             cmd_buffers;

   VkCommandPoolCreateFlags                     flags;
};

#define ANV_CMD_BUFFER_BATCH_SIZE 8192

enum anv_cmd_buffer_exec_mode {
   ANV_CMD_BUFFER_EXEC_MODE_PRIMARY,
   ANV_CMD_BUFFER_EXEC_MODE_EMIT,
   ANV_CMD_BUFFER_EXEC_MODE_GROW_AND_EMIT,
   ANV_CMD_BUFFER_EXEC_MODE_CHAIN,
   ANV_CMD_BUFFER_EXEC_MODE_COPY_AND_CHAIN,
   ANV_CMD_BUFFER_EXEC_MODE_CALL_AND_RETURN,
};

struct anv_measure_batch;

struct anv_cmd_buffer {
   struct vk_object_base                        base;

   struct anv_device *                          device;

   struct anv_cmd_pool *                        pool;
   struct list_head                             pool_link;

   struct anv_batch                             batch;

   /* Pointer to the location in the batch where MI_BATCH_BUFFER_END was
    * recorded upon calling vkEndCommandBuffer(). This is useful if we need to
    * rewrite the end to chain multiple batch together at vkQueueSubmit().
    */
   void *                                       batch_end;

   /* Fields required for the actual chain of anv_batch_bo's.
    *
    * These fields are initialized by anv_cmd_buffer_init_batch_bo_chain().
    */
   struct list_head                             batch_bos;
   enum anv_cmd_buffer_exec_mode                exec_mode;

   /* A vector of anv_batch_bo pointers for every batch or surface buffer
    * referenced by this command buffer
    *
    * initialized by anv_cmd_buffer_init_batch_bo_chain()
    */
   struct u_vector                            seen_bbos;

   /* A vector of int32_t's for every block of binding tables.
    *
    * initialized by anv_cmd_buffer_init_batch_bo_chain()
    */
   struct u_vector                              bt_block_states;
   struct anv_state                             bt_next;

   struct anv_reloc_list                        surface_relocs;
   /** Last seen surface state block pool center bo offset */
   uint32_t                                     last_ss_pool_center;

   /* Serial for tracking buffer completion */
   uint32_t                                     serial;

   /* Stream objects for storing temporary data */
   struct anv_state_stream                      surface_state_stream;
   struct anv_state_stream                      dynamic_state_stream;
   struct anv_state_stream                      general_state_stream;

   VkCommandBufferUsageFlags                    usage_flags;
   VkCommandBufferLevel                         level;

   struct anv_query_pool                       *perf_query_pool;

   struct anv_cmd_state                         state;

   struct anv_address                           return_addr;

   /* Set by SetPerformanceMarkerINTEL, written into queries by CmdBeginQuery */
   uint64_t                                     intel_perf_marker;

   struct anv_measure_batch *measure;

   /**
    * KHR_performance_query requires self modifying command buffers and this
    * array has the location of modifying commands to the query begin and end
    * instructions storing performance counters. The array length is
    * anv_physical_device::n_perf_query_commands.
    */
   struct mi_address_token                  *self_mod_locations;

   /**
    * Index tracking which of the self_mod_locations items have already been
    * used.
    */
   uint32_t                                      perf_reloc_idx;
};

/* Determine whether we can chain a given cmd_buffer to another one. We need
 * softpin and we also need to make sure that we can edit the end of the batch
 * to point to next one, which requires the command buffer to not be used
 * simultaneously.
 */
static inline bool
anv_cmd_buffer_is_chainable(struct anv_cmd_buffer *cmd_buffer)
{
   return cmd_buffer->device->physical->use_softpin &&
      !(cmd_buffer->usage_flags & VK_COMMAND_BUFFER_USAGE_SIMULTANEOUS_USE_BIT);
}

VkResult anv_cmd_buffer_init_batch_bo_chain(struct anv_cmd_buffer *cmd_buffer);
void anv_cmd_buffer_fini_batch_bo_chain(struct anv_cmd_buffer *cmd_buffer);
void anv_cmd_buffer_reset_batch_bo_chain(struct anv_cmd_buffer *cmd_buffer);
void anv_cmd_buffer_end_batch_buffer(struct anv_cmd_buffer *cmd_buffer);
void anv_cmd_buffer_add_secondary(struct anv_cmd_buffer *primary,
                                  struct anv_cmd_buffer *secondary);
void anv_cmd_buffer_prepare_execbuf(struct anv_cmd_buffer *cmd_buffer);
VkResult anv_cmd_buffer_execbuf(struct anv_queue *queue,
                                struct anv_cmd_buffer *cmd_buffer,
                                const VkSemaphore *in_semaphores,
                                const uint64_t *in_wait_values,
                                uint32_t num_in_semaphores,
                                const VkSemaphore *out_semaphores,
                                const uint64_t *out_signal_values,
                                uint32_t num_out_semaphores,
                                VkFence fence,
                                int perf_query_pass);

VkResult anv_cmd_buffer_reset(struct anv_cmd_buffer *cmd_buffer);

struct anv_state anv_cmd_buffer_emit_dynamic(struct anv_cmd_buffer *cmd_buffer,
                                             const void *data, uint32_t size, uint32_t alignment);
struct anv_state anv_cmd_buffer_merge_dynamic(struct anv_cmd_buffer *cmd_buffer,
                                              uint32_t *a, uint32_t *b,
                                              uint32_t dwords, uint32_t alignment);

struct anv_address
anv_cmd_buffer_surface_base_address(struct anv_cmd_buffer *cmd_buffer);
struct anv_state
anv_cmd_buffer_alloc_binding_table(struct anv_cmd_buffer *cmd_buffer,
                                   uint32_t entries, uint32_t *state_offset);
struct anv_state
anv_cmd_buffer_alloc_surface_state(struct anv_cmd_buffer *cmd_buffer);
struct anv_state
anv_cmd_buffer_alloc_dynamic_state(struct anv_cmd_buffer *cmd_buffer,
                                   uint32_t size, uint32_t alignment);

VkResult
anv_cmd_buffer_new_binding_table_block(struct anv_cmd_buffer *cmd_buffer);

void gfx8_cmd_buffer_emit_viewport(struct anv_cmd_buffer *cmd_buffer);
void gfx8_cmd_buffer_emit_depth_viewport(struct anv_cmd_buffer *cmd_buffer,
                                         bool depth_clamp_enable);
void gfx7_cmd_buffer_emit_scissor(struct anv_cmd_buffer *cmd_buffer);

void anv_cmd_buffer_setup_attachments(struct anv_cmd_buffer *cmd_buffer,
                                      struct anv_render_pass *pass,
                                      struct anv_framebuffer *framebuffer,
                                      const VkClearValue *clear_values);

void anv_cmd_buffer_emit_state_base_address(struct anv_cmd_buffer *cmd_buffer);

struct anv_state
anv_cmd_buffer_gfx_push_constants(struct anv_cmd_buffer *cmd_buffer);
struct anv_state
anv_cmd_buffer_cs_push_constants(struct anv_cmd_buffer *cmd_buffer);

const struct anv_image_view *
anv_cmd_buffer_get_depth_stencil_view(const struct anv_cmd_buffer *cmd_buffer);

VkResult
anv_cmd_buffer_alloc_blorp_binding_table(struct anv_cmd_buffer *cmd_buffer,
                                         uint32_t num_entries,
                                         uint32_t *state_offset,
                                         struct anv_state *bt_state);

void anv_cmd_buffer_dump(struct anv_cmd_buffer *cmd_buffer);

void anv_cmd_emit_conditional_render_predicate(struct anv_cmd_buffer *cmd_buffer);

enum anv_fence_type {
   ANV_FENCE_TYPE_NONE = 0,
   ANV_FENCE_TYPE_BO,
   ANV_FENCE_TYPE_WSI_BO,
   ANV_FENCE_TYPE_SYNCOBJ,
   ANV_FENCE_TYPE_WSI,
};

enum anv_bo_fence_state {
   /** Indicates that this is a new (or newly reset fence) */
   ANV_BO_FENCE_STATE_RESET,

   /** Indicates that this fence has been submitted to the GPU but is still
    * (as far as we know) in use by the GPU.
    */
   ANV_BO_FENCE_STATE_SUBMITTED,

   ANV_BO_FENCE_STATE_SIGNALED,
};

struct anv_fence_impl {
   enum anv_fence_type type;

   union {
      /** Fence implementation for BO fences
       *
       * These fences use a BO and a set of CPU-tracked state flags.  The BO
       * is added to the object list of the last execbuf call in a QueueSubmit
       * and is marked EXEC_WRITE.  The state flags track when the BO has been
       * submitted to the kernel.  We need to do this because Vulkan lets you
       * wait on a fence that has not yet been submitted and I915_GEM_BUSY
       * will say it's idle in this case.
       */
      struct {
         struct anv_bo *bo;
         enum anv_bo_fence_state state;
      } bo;

      /** DRM syncobj handle for syncobj-based fences */
      uint32_t syncobj;

      /** WSI fence */
      struct wsi_fence *fence_wsi;
   };
};

struct anv_fence {
   struct vk_object_base base;

   /* Permanent fence state.  Every fence has some form of permanent state
    * (type != ANV_SEMAPHORE_TYPE_NONE).  This may be a BO to fence on (for
    * cross-process fences) or it could just be a dummy for use internally.
    */
   struct anv_fence_impl permanent;

   /* Temporary fence state.  A fence *may* have temporary state.  That state
    * is added to the fence by an import operation and is reset back to
    * ANV_SEMAPHORE_TYPE_NONE when the fence is reset.  A fence with temporary
    * state cannot be signaled because the fence must already be signaled
    * before the temporary state can be exported from the fence in the other
    * process and imported here.
    */
   struct anv_fence_impl temporary;
};

void anv_fence_reset_temporary(struct anv_device *device,
                               struct anv_fence *fence);

struct anv_event {
   struct vk_object_base                        base;
   uint64_t                                     semaphore;
   struct anv_state                             state;
};

enum anv_semaphore_type {
   ANV_SEMAPHORE_TYPE_NONE = 0,
   ANV_SEMAPHORE_TYPE_DUMMY,
   ANV_SEMAPHORE_TYPE_BO,
   ANV_SEMAPHORE_TYPE_WSI_BO,
   ANV_SEMAPHORE_TYPE_SYNC_FILE,
   ANV_SEMAPHORE_TYPE_DRM_SYNCOBJ,
   ANV_SEMAPHORE_TYPE_TIMELINE,
   ANV_SEMAPHORE_TYPE_DRM_SYNCOBJ_TIMELINE,
};

struct anv_timeline_point {
   struct list_head link;

   uint64_t serial;

   /* Number of waiter on this point, when > 0 the point should not be garbage
    * collected.
    */
   int waiting;

   /* BO used for synchronization. */
   struct anv_bo *bo;
};

struct anv_timeline {
   pthread_mutex_t mutex;
   pthread_cond_t  cond;

   uint64_t highest_past;
   uint64_t highest_pending;

   struct list_head points;
   struct list_head free_points;
};

struct anv_semaphore_impl {
   enum anv_semaphore_type type;

   union {
      /* A BO representing this semaphore when type == ANV_SEMAPHORE_TYPE_BO
       * or type == ANV_SEMAPHORE_TYPE_WSI_BO.  This BO will be added to the
       * object list on any execbuf2 calls for which this semaphore is used as
       * a wait or signal fence.  When used as a signal fence or when type ==
       * ANV_SEMAPHORE_TYPE_WSI_BO, the EXEC_OBJECT_WRITE flag will be set.
       */
      struct anv_bo *bo;

      /* The sync file descriptor when type == ANV_SEMAPHORE_TYPE_SYNC_FILE.
       * If the semaphore is in the unsignaled state due to either just being
       * created or because it has been used for a wait, fd will be -1.
       */
      int fd;

      /* Sync object handle when type == ANV_SEMAPHORE_TYPE_DRM_SYNCOBJ.
       * Unlike GEM BOs, DRM sync objects aren't deduplicated by the kernel on
       * import so we don't need to bother with a userspace cache.
       */
      uint32_t syncobj;

      /* Non shareable timeline semaphore
       *
       * Used when kernel don't have support for timeline semaphores.
       */
      struct anv_timeline timeline;
   };
};

struct anv_semaphore {
   struct vk_object_base base;

   uint32_t refcount;

   /* Permanent semaphore state.  Every semaphore has some form of permanent
    * state (type != ANV_SEMAPHORE_TYPE_NONE).  This may be a BO to fence on
    * (for cross-process semaphores0 or it could just be a dummy for use
    * internally.
    */
   struct anv_semaphore_impl permanent;

   /* Temporary semaphore state.  A semaphore *may* have temporary state.
    * That state is added to the semaphore by an import operation and is reset
    * back to ANV_SEMAPHORE_TYPE_NONE when the semaphore is waited on.  A
    * semaphore with temporary state cannot be signaled because the semaphore
    * must already be signaled before the temporary state can be exported from
    * the semaphore in the other process and imported here.
    */
   struct anv_semaphore_impl temporary;
};

void anv_semaphore_reset_temporary(struct anv_device *device,
                                   struct anv_semaphore *semaphore);

#define ANV_STAGE_MASK ((1 << MESA_SHADER_STAGES) - 1)

#define anv_foreach_stage(stage, stage_bits)                         \
   for (gl_shader_stage stage,                                       \
        __tmp = (gl_shader_stage)((stage_bits) & ANV_STAGE_MASK);    \
        stage = __builtin_ffs(__tmp) - 1, __tmp;                     \
        __tmp &= ~(1 << (stage)))

enum anv_shader_reloc {
   ANV_SHADER_RELOC_CONST_DATA_ADDR_LOW,
   ANV_SHADER_RELOC_CONST_DATA_ADDR_HIGH,
};

struct anv_pipeline_bind_map {
   unsigned char                                surface_sha1[20];
   unsigned char                                sampler_sha1[20];
   unsigned char                                push_sha1[20];

   uint32_t surface_count;
   uint32_t sampler_count;

   struct anv_pipeline_binding *                surface_to_descriptor;
   struct anv_pipeline_binding *                sampler_to_descriptor;

   struct anv_push_range                        push_ranges[4];
};

struct anv_shader_bin_key {
   uint32_t size;
   uint8_t data[0];
};

struct anv_shader_bin {
   uint32_t ref_cnt;

   gl_shader_stage stage;

   const struct anv_shader_bin_key *key;

   struct anv_state kernel;
   uint32_t kernel_size;

   const struct brw_stage_prog_data *prog_data;
   uint32_t prog_data_size;

   struct brw_compile_stats stats[3];
   uint32_t num_stats;

   struct nir_xfb_info *xfb_info;

   struct anv_pipeline_bind_map bind_map;
};

struct anv_shader_bin *
anv_shader_bin_create(struct anv_device *device,
                      gl_shader_stage stage,
                      const void *key, uint32_t key_size,
                      const void *kernel, uint32_t kernel_size,
                      const struct brw_stage_prog_data *prog_data,
                      uint32_t prog_data_size,
                      const struct brw_compile_stats *stats, uint32_t num_stats,
                      const struct nir_xfb_info *xfb_info,
                      const struct anv_pipeline_bind_map *bind_map);

void
anv_shader_bin_destroy(struct anv_device *device, struct anv_shader_bin *shader);

static inline void
anv_shader_bin_ref(struct anv_shader_bin *shader)
{
   assert(shader && shader->ref_cnt >= 1);
   p_atomic_inc(&shader->ref_cnt);
}

static inline void
anv_shader_bin_unref(struct anv_device *device, struct anv_shader_bin *shader)
{
   assert(shader && shader->ref_cnt >= 1);
   if (p_atomic_dec_zero(&shader->ref_cnt))
      anv_shader_bin_destroy(device, shader);
}

struct anv_pipeline_executable {
   gl_shader_stage stage;

   struct brw_compile_stats stats;

   char *nir;
   char *disasm;
};

enum anv_pipeline_type {
   ANV_PIPELINE_GRAPHICS,
   ANV_PIPELINE_COMPUTE,
};

struct anv_pipeline {
   struct vk_object_base                        base;

   struct anv_device *                          device;

   struct anv_batch                             batch;
   struct anv_reloc_list                        batch_relocs;

   void *                                       mem_ctx;

   enum anv_pipeline_type                       type;
   VkPipelineCreateFlags                        flags;

   struct util_dynarray                         executables;

   const struct intel_l3_config *               l3_config;
};

struct anv_graphics_pipeline {
   struct anv_pipeline                          base;

   uint32_t                                     batch_data[512];

   anv_cmd_dirty_mask_t                         dynamic_state_mask;
   struct anv_dynamic_state                     dynamic_state;

   uint32_t                                     topology;

   struct anv_subpass *                         subpass;

   struct anv_shader_bin *                      shaders[MESA_SHADER_STAGES];

   VkShaderStageFlags                           active_stages;

   bool                                         primitive_restart;
   bool                                         writes_depth;
   bool                                         depth_test_enable;
   bool                                         writes_stencil;
   bool                                         stencil_test_enable;
   bool                                         depth_clamp_enable;
   bool                                         depth_clip_enable;
   bool                                         sample_shading_enable;
   bool                                         kill_pixel;
   bool                                         depth_bounds_test_enable;

   /* When primitive replication is used, subpass->view_mask will describe what
    * views to replicate.
    */
   bool                                         use_primitive_replication;

   struct anv_state                             blend_state;

   uint32_t                                     vb_used;
   struct anv_pipeline_vertex_binding {
      uint32_t                                  stride;
      bool                                      instanced;
      uint32_t                                  instance_divisor;
   } vb[MAX_VBS];

   struct {
      uint32_t                                  sf[7];
      uint32_t                                  depth_stencil_state[3];
      uint32_t                                  clip[4];
      uint32_t                                  xfb_bo_pitch[4];
   } gfx7;

   struct {
      uint32_t                                  sf[4];
      uint32_t                                  raster[5];
      uint32_t                                  wm_depth_stencil[3];
   } gfx8;

   struct {
      uint32_t                                  wm_depth_stencil[4];
   } gfx9;
};

struct anv_compute_pipeline {
   struct anv_pipeline                          base;

   struct anv_shader_bin *                      cs;
   uint32_t                                     cs_right_mask;
   uint32_t                                     batch_data[9];
   uint32_t                                     interface_descriptor_data[8];
};

#define ANV_DECL_PIPELINE_DOWNCAST(pipe_type, pipe_enum)             \
   static inline struct anv_##pipe_type##_pipeline *                 \
   anv_pipeline_to_##pipe_type(struct anv_pipeline *pipeline)      \
   {                                                                 \
      assert(pipeline->type == pipe_enum);                           \
      return (struct anv_##pipe_type##_pipeline *) pipeline;         \
   }

ANV_DECL_PIPELINE_DOWNCAST(graphics, ANV_PIPELINE_GRAPHICS)
ANV_DECL_PIPELINE_DOWNCAST(compute, ANV_PIPELINE_COMPUTE)

static inline bool
anv_pipeline_has_stage(const struct anv_graphics_pipeline *pipeline,
                       gl_shader_stage stage)
{
   return (pipeline->active_stages & mesa_to_vk_shader_stage(stage)) != 0;
}

#define ANV_DECL_GET_GRAPHICS_PROG_DATA_FUNC(prefix, stage)             \
static inline const struct brw_##prefix##_prog_data *                   \
get_##prefix##_prog_data(const struct anv_graphics_pipeline *pipeline)  \
{                                                                       \
   if (anv_pipeline_has_stage(pipeline, stage)) {                       \
      return (const struct brw_##prefix##_prog_data *)                  \
             pipeline->shaders[stage]->prog_data;                       \
   } else {                                                             \
      return NULL;                                                      \
   }                                                                    \
}

ANV_DECL_GET_GRAPHICS_PROG_DATA_FUNC(vs, MESA_SHADER_VERTEX)
ANV_DECL_GET_GRAPHICS_PROG_DATA_FUNC(tcs, MESA_SHADER_TESS_CTRL)
ANV_DECL_GET_GRAPHICS_PROG_DATA_FUNC(tes, MESA_SHADER_TESS_EVAL)
ANV_DECL_GET_GRAPHICS_PROG_DATA_FUNC(gs, MESA_SHADER_GEOMETRY)
ANV_DECL_GET_GRAPHICS_PROG_DATA_FUNC(wm, MESA_SHADER_FRAGMENT)

static inline const struct brw_cs_prog_data *
get_cs_prog_data(const struct anv_compute_pipeline *pipeline)
{
   assert(pipeline->cs);
   return (const struct brw_cs_prog_data *) pipeline->cs->prog_data;
}

static inline const struct brw_vue_prog_data *
anv_pipeline_get_last_vue_prog_data(const struct anv_graphics_pipeline *pipeline)
{
   if (anv_pipeline_has_stage(pipeline, MESA_SHADER_GEOMETRY))
      return &get_gs_prog_data(pipeline)->base;
   else if (anv_pipeline_has_stage(pipeline, MESA_SHADER_TESS_EVAL))
      return &get_tes_prog_data(pipeline)->base;
   else
      return &get_vs_prog_data(pipeline)->base;
}

VkResult
anv_pipeline_init(struct anv_pipeline *pipeline,
                  struct anv_device *device,
                  enum anv_pipeline_type type,
                  VkPipelineCreateFlags flags,
                  const VkAllocationCallbacks *pAllocator);

void
anv_pipeline_finish(struct anv_pipeline *pipeline,
                    struct anv_device *device,
                    const VkAllocationCallbacks *pAllocator);

VkResult
anv_graphics_pipeline_init(struct anv_graphics_pipeline *pipeline, struct anv_device *device,
                           struct anv_pipeline_cache *cache,
                           const VkGraphicsPipelineCreateInfo *pCreateInfo,
                           const VkAllocationCallbacks *alloc);

VkResult
anv_pipeline_compile_cs(struct anv_compute_pipeline *pipeline,
                        struct anv_pipeline_cache *cache,
                        const VkComputePipelineCreateInfo *info,
                        const struct vk_shader_module *module,
                        const char *entrypoint,
                        const VkSpecializationInfo *spec_info);

struct anv_cs_parameters {
   uint32_t group_size;
   uint32_t simd_size;
   uint32_t threads;
};

struct anv_cs_parameters
anv_cs_parameters(const struct anv_compute_pipeline *pipeline);

struct anv_format_plane {
   enum isl_format isl_format:16;
   struct isl_swizzle swizzle;

   /* Whether this plane contains chroma channels */
   bool has_chroma;

   /* For downscaling of YUV planes */
   uint8_t denominator_scales[2];

   /* How to map sampled ycbcr planes to a single 4 component element. */
   struct isl_swizzle ycbcr_swizzle;

   /* What aspect is associated to this plane */
   VkImageAspectFlags aspect;
};


struct anv_format {
   struct anv_format_plane planes[3];
   VkFormat vk_format;
   uint8_t n_planes;
   bool can_ycbcr;
};

/**
 * Return the aspect's _format_ plane, not its _memory_ plane (using the
 * vocabulary of VK_EXT_image_drm_format_modifier). As a consequence, \a
 * aspect_mask may contain VK_IMAGE_ASPECT_PLANE_*, but must not contain
 * VK_IMAGE_ASPECT_MEMORY_PLANE_* .
 */
static inline uint32_t
anv_image_aspect_to_plane(VkImageAspectFlags image_aspects,
                          VkImageAspectFlags aspect_mask)
{
   switch (aspect_mask) {
   case VK_IMAGE_ASPECT_COLOR_BIT:
   case VK_IMAGE_ASPECT_DEPTH_BIT:
   case VK_IMAGE_ASPECT_PLANE_0_BIT:
      return 0;
   case VK_IMAGE_ASPECT_STENCIL_BIT:
      if ((image_aspects & VK_IMAGE_ASPECT_DEPTH_BIT) == 0)
         return 0;
      /* Fall-through */
   case VK_IMAGE_ASPECT_PLANE_1_BIT:
      return 1;
   case VK_IMAGE_ASPECT_PLANE_2_BIT:
      return 2;
   default:
      /* Purposefully assert with depth/stencil aspects. */
      unreachable("invalid image aspect");
   }
}

static inline VkImageAspectFlags
anv_plane_to_aspect(VkImageAspectFlags image_aspects,
                    uint32_t plane)
{
   if (image_aspects & VK_IMAGE_ASPECT_ANY_COLOR_BIT_ANV) {
      if (util_bitcount(image_aspects) > 1)
         return VK_IMAGE_ASPECT_PLANE_0_BIT << plane;
      return VK_IMAGE_ASPECT_COLOR_BIT;
   }
   if (image_aspects & VK_IMAGE_ASPECT_DEPTH_BIT)
      return VK_IMAGE_ASPECT_DEPTH_BIT << plane;
   assert(image_aspects == VK_IMAGE_ASPECT_STENCIL_BIT);
   return VK_IMAGE_ASPECT_STENCIL_BIT;
}

#define anv_foreach_image_aspect_bit(b, image, aspects) \
   u_foreach_bit(b, anv_image_expand_aspects(image, aspects))

const struct anv_format *
anv_get_format(VkFormat format);

static inline uint32_t
anv_get_format_planes(VkFormat vk_format)
{
   const struct anv_format *format = anv_get_format(vk_format);

   return format != NULL ? format->n_planes : 0;
}

struct anv_format_plane
anv_get_format_plane(const struct gen_device_info *devinfo, VkFormat vk_format,
                     VkImageAspectFlagBits aspect, VkImageTiling tiling);

static inline enum isl_format
anv_get_isl_format(const struct gen_device_info *devinfo, VkFormat vk_format,
                   VkImageAspectFlags aspect, VkImageTiling tiling)
{
   return anv_get_format_plane(devinfo, vk_format, aspect, tiling).isl_format;
}

bool anv_formats_ccs_e_compatible(const struct gen_device_info *devinfo,
                                  VkImageCreateFlags create_flags,
                                  VkFormat vk_format,
                                  VkImageTiling vk_tiling,
                                  const VkImageFormatListCreateInfoKHR *fmt_list);

extern VkFormat
vk_format_from_android(unsigned android_format, unsigned android_usage);

static inline struct isl_swizzle
anv_swizzle_for_render(struct isl_swizzle swizzle)
{
   /* Sometimes the swizzle will have alpha map to one.  We do this to fake
    * RGB as RGBA for texturing
    */
   assert(swizzle.a == ISL_CHANNEL_SELECT_ONE ||
          swizzle.a == ISL_CHANNEL_SELECT_ALPHA);

   /* But it doesn't matter what we render to that channel */
   swizzle.a = ISL_CHANNEL_SELECT_ALPHA;

   return swizzle;
}

void
anv_pipeline_setup_l3_config(struct anv_pipeline *pipeline, bool needs_slm);

/**
 * Describes how each part of anv_image will be bound to memory.
 */
struct anv_image_memory_range {
   /**
    * Disjoint bindings into which each portion of the image will be bound.
    *
    * Binding images to memory can be complicated and invold binding different
    * portions of the image to different memory objects or regions.  For most
    * images, everything lives in the MAIN binding and gets bound by
    * vkBindImageMemory.  For disjoint multi-planar images, each plane has
    * a unique, disjoint binding and gets bound by vkBindImageMemory2 with
    * VkBindImagePlaneMemoryInfo.  There may also exist bits of memory which are
    * implicit or driver-managed and live in special-case bindings.
    */
   enum anv_image_memory_binding {
      /**
       * Used if and only if image is not multi-planar disjoint. Bound by
       * vkBindImageMemory2 without VkBindImagePlaneMemoryInfo.
       */
      ANV_IMAGE_MEMORY_BINDING_MAIN,

      /**
       * Used if and only if image is multi-planar disjoint.  Bound by
       * vkBindImageMemory2 with VkBindImagePlaneMemoryInfo.
       */
      ANV_IMAGE_MEMORY_BINDING_PLANE_0,
      ANV_IMAGE_MEMORY_BINDING_PLANE_1,
      ANV_IMAGE_MEMORY_BINDING_PLANE_2,

      /**
       * Driver-private bo. In special cases we may store the aux surface and/or
       * aux state in this binding.
       */
      ANV_IMAGE_MEMORY_BINDING_PRIVATE,

      /** Sentinel */
      ANV_IMAGE_MEMORY_BINDING_END,
   } binding;

   /**
    * Offset is relative to the start of the binding created by
    * vkBindImageMemory, not to the start of the bo.
    */
   uint64_t offset;

   uint64_t size;
   uint32_t alignment;
};

/**
 * Subsurface of an anv_image.
 */
struct anv_surface {
   struct isl_surf isl;
   struct anv_image_memory_range memory_range;
};

static inline bool MUST_CHECK
anv_surface_is_valid(const struct anv_surface *surface)
{
   return surface->isl.size_B > 0 && surface->memory_range.size > 0;
}

struct anv_image {
   struct vk_object_base base;

   VkImageType type; /**< VkImageCreateInfo::imageType */
   /* The original VkFormat provided by the client.  This may not match any
    * of the actual surface formats.
    */
   VkFormat vk_format;
   const struct anv_format *format;

   VkImageAspectFlags aspects;
   VkExtent3D extent;
   uint32_t levels;
   uint32_t array_size;
   uint32_t samples; /**< VkImageCreateInfo::samples */
   uint32_t n_planes;
   VkImageUsageFlags usage; /**< VkImageCreateInfo::usage. */
   VkImageUsageFlags stencil_usage;
   VkImageCreateFlags create_flags; /* Flags used when creating image. */
   VkImageTiling tiling; /** VkImageCreateInfo::tiling */

   /** True if this is needs to be bound to an appropriately tiled BO.
    *
    * When not using modifiers, consumers such as X11, Wayland, and KMS need
    * the tiling passed via I915_GEM_SET_TILING.  When exporting these buffers
    * we require a dedicated allocation so that we can know to allocate a
    * tiled buffer.
    */
   bool needs_set_tiling;

   /**
    * Must be DRM_FORMAT_MOD_INVALID unless tiling is
    * VK_IMAGE_TILING_DRM_FORMAT_MODIFIER_EXT.
    */
   uint64_t drm_format_mod;

   /**
    * Image has multi-planar format and was created with
    * VK_IMAGE_CREATE_DISJOINT_BIT.
    */
   bool disjoint;

   /* Image was created with external format. */
   bool external_format;

   /**
    * Image was imported from gralloc with VkNativeBufferANDROID. The gralloc bo
    * must be released when the image is destroyed.
    */
   bool from_gralloc;

   /**
    * The memory bindings created by vkCreateImage and vkBindImageMemory.
    *
    * For details on the image's memory layout, see check_memory_bindings().
    *
    * vkCreateImage constructs the `memory_range` for each
    * anv_image_memory_binding.  After vkCreateImage, each binding is valid if
    * and only if `memory_range::size > 0`.
    *
    * vkBindImageMemory binds each valid `memory_range` to an `address`.
    * Usually, the app will provide the address via the parameters of
    * vkBindImageMemory.  However, special-case bindings may be bound to
    * driver-private memory.
    */
   struct anv_image_binding {
      struct anv_image_memory_range memory_range;
      struct anv_address address;
   } bindings[ANV_IMAGE_MEMORY_BINDING_END];

   /**
    * Image subsurfaces
    *
    * For each foo, anv_image::planes[x].surface is valid if and only if
    * anv_image::aspects has a x aspect. Refer to anv_image_aspect_to_plane()
    * to figure the number associated with a given aspect.
    *
    * The hardware requires that the depth buffer and stencil buffer be
    * separate surfaces.  From Vulkan's perspective, though, depth and stencil
    * reside in the same VkImage.  To satisfy both the hardware and Vulkan, we
    * allocate the depth and stencil buffers as separate surfaces in the same
    * bo.
    */
   struct anv_image_plane {
      struct anv_surface primary_surface;

      /**
       * A surface which shadows the main surface and may have different
       * tiling. This is used for sampling using a tiling that isn't supported
       * for other operations.
       */
      struct anv_surface shadow_surface;

      /**
       * The base aux usage for this image.  For color images, this can be
       * either CCS_E or CCS_D depending on whether or not we can reliably
       * leave CCS on all the time.
       */
      enum isl_aux_usage aux_usage;

      struct anv_surface aux_surface;

      /** Location of the fast clear state.  */
      struct anv_image_memory_range fast_clear_memory_range;
   } planes[3];
};

/* The ordering of this enum is important */
enum anv_fast_clear_type {
   /** Image does not have/support any fast-clear blocks */
   ANV_FAST_CLEAR_NONE = 0,
   /** Image has/supports fast-clear but only to the default value */
   ANV_FAST_CLEAR_DEFAULT_VALUE = 1,
   /** Image has/supports fast-clear with an arbitrary fast-clear value */
   ANV_FAST_CLEAR_ANY = 2,
};

/* Returns the number of auxiliary buffer levels attached to an image. */
static inline uint8_t
anv_image_aux_levels(const struct anv_image * const image,
                     VkImageAspectFlagBits aspect)
{
   uint32_t plane = anv_image_aspect_to_plane(image->aspects, aspect);
   if (image->planes[plane].aux_usage == ISL_AUX_USAGE_NONE)
      return 0;

   return image->levels;
}

/* Returns the number of auxiliary buffer layers attached to an image. */
static inline uint32_t
anv_image_aux_layers(const struct anv_image * const image,
                     VkImageAspectFlagBits aspect,
                     const uint8_t miplevel)
{
   assert(image);

   /* The miplevel must exist in the main buffer. */
   assert(miplevel < image->levels);

   if (miplevel >= anv_image_aux_levels(image, aspect)) {
      /* There are no layers with auxiliary data because the miplevel has no
       * auxiliary data.
       */
      return 0;
   }

   return MAX2(image->array_size, image->extent.depth >> miplevel);
}

static inline struct anv_address MUST_CHECK
anv_image_address(const struct anv_image *image,
                  const struct anv_image_memory_range *mem_range)
{
   const struct anv_image_binding *binding = &image->bindings[mem_range->binding];
   assert(binding->memory_range.offset == 0);

   if (mem_range->size == 0)
      return ANV_NULL_ADDRESS;

   return anv_address_add(binding->address, mem_range->offset);
}

static inline struct anv_address
anv_image_get_clear_color_addr(UNUSED const struct anv_device *device,
                               const struct anv_image *image,
                               VkImageAspectFlagBits aspect)
{
   assert(image->aspects & VK_IMAGE_ASPECT_ANY_COLOR_BIT_ANV);

   uint32_t plane = anv_image_aspect_to_plane(image->aspects, aspect);
   const struct anv_image_memory_range *mem_range =
      &image->planes[plane].fast_clear_memory_range;

   return anv_image_address(image, mem_range);
}

static inline struct anv_address
anv_image_get_fast_clear_type_addr(const struct anv_device *device,
                                   const struct anv_image *image,
                                   VkImageAspectFlagBits aspect)
{
   struct anv_address addr =
      anv_image_get_clear_color_addr(device, image, aspect);

   const unsigned clear_color_state_size = device->info.ver >= 10 ?
      device->isl_dev.ss.clear_color_state_size :
      device->isl_dev.ss.clear_value_size;
   return anv_address_add(addr, clear_color_state_size);
}

static inline struct anv_address
anv_image_get_compression_state_addr(const struct anv_device *device,
                                     const struct anv_image *image,
                                     VkImageAspectFlagBits aspect,
                                     uint32_t level, uint32_t array_layer)
{
   assert(level < anv_image_aux_levels(image, aspect));
   assert(array_layer < anv_image_aux_layers(image, aspect, level));
   UNUSED uint32_t plane = anv_image_aspect_to_plane(image->aspects, aspect);
   assert(image->planes[plane].aux_usage == ISL_AUX_USAGE_CCS_E);

   /* Relative to start of the plane's fast clear memory range */
   uint32_t offset;

   offset = 4; /* Go past the fast clear type */

   if (image->type == VK_IMAGE_TYPE_3D) {
      for (uint32_t l = 0; l < level; l++)
         offset += anv_minify(image->extent.depth, l) * 4;
   } else {
      offset += level * image->array_size * 4;
   }

   offset += array_layer * 4;

   assert(offset < image->planes[plane].fast_clear_memory_range.size);

   return anv_address_add(
      anv_image_get_fast_clear_type_addr(device, image, aspect),
      offset);
}

/* Returns true if a HiZ-enabled depth buffer can be sampled from. */
static inline bool
anv_can_sample_with_hiz(const struct gen_device_info * const devinfo,
                        const struct anv_image *image)
{
   if (!(image->aspects & VK_IMAGE_ASPECT_DEPTH_BIT))
      return false;

   /* For Gfx8-11, there are some restrictions around sampling from HiZ.
    * The Skylake PRM docs for RENDER_SURFACE_STATE::AuxiliarySurfaceMode
    * say:
    *
    *    "If this field is set to AUX_HIZ, Number of Multisamples must
    *    be MULTISAMPLECOUNT_1, and Surface Type cannot be SURFTYPE_3D."
    */
   if (image->type == VK_IMAGE_TYPE_3D)
      return false;

   /* Allow this feature on BDW even though it is disabled in the BDW devinfo
    * struct. There's documentation which suggests that this feature actually
    * reduces performance on BDW, but it has only been observed to help so
    * far. Sampling fast-cleared blocks on BDW must also be handled with care
    * (see depth_stencil_attachment_compute_aux_usage() for more info).
    */
   if (devinfo->ver != 8 && !devinfo->has_sample_with_hiz)
      return false;

   return image->samples == 1;
}

static inline bool
anv_image_plane_uses_aux_map(const struct anv_device *device,
                             const struct anv_image *image,
                             uint32_t plane)
{
   return device->info.has_aux_map &&
      isl_aux_usage_has_ccs(image->planes[plane].aux_usage);
}

void
anv_cmd_buffer_mark_image_written(struct anv_cmd_buffer *cmd_buffer,
                                  const struct anv_image *image,
                                  VkImageAspectFlagBits aspect,
                                  enum isl_aux_usage aux_usage,
                                  uint32_t level,
                                  uint32_t base_layer,
                                  uint32_t layer_count);

void
anv_image_clear_color(struct anv_cmd_buffer *cmd_buffer,
                      const struct anv_image *image,
                      VkImageAspectFlagBits aspect,
                      enum isl_aux_usage aux_usage,
                      enum isl_format format, struct isl_swizzle swizzle,
                      uint32_t level, uint32_t base_layer, uint32_t layer_count,
                      VkRect2D area, union isl_color_value clear_color);
void
anv_image_clear_depth_stencil(struct anv_cmd_buffer *cmd_buffer,
                              const struct anv_image *image,
                              VkImageAspectFlags aspects,
                              enum isl_aux_usage depth_aux_usage,
                              uint32_t level,
                              uint32_t base_layer, uint32_t layer_count,
                              VkRect2D area,
                              float depth_value, uint8_t stencil_value);
void
anv_image_msaa_resolve(struct anv_cmd_buffer *cmd_buffer,
                       const struct anv_image *src_image,
                       enum isl_aux_usage src_aux_usage,
                       uint32_t src_level, uint32_t src_base_layer,
                       const struct anv_image *dst_image,
                       enum isl_aux_usage dst_aux_usage,
                       uint32_t dst_level, uint32_t dst_base_layer,
                       VkImageAspectFlagBits aspect,
                       uint32_t src_x, uint32_t src_y,
                       uint32_t dst_x, uint32_t dst_y,
                       uint32_t width, uint32_t height,
                       uint32_t layer_count,
                       enum blorp_filter filter);
void
anv_image_hiz_op(struct anv_cmd_buffer *cmd_buffer,
                 const struct anv_image *image,
                 VkImageAspectFlagBits aspect, uint32_t level,
                 uint32_t base_layer, uint32_t layer_count,
                 enum isl_aux_op hiz_op);
void
anv_image_hiz_clear(struct anv_cmd_buffer *cmd_buffer,
                    const struct anv_image *image,
                    VkImageAspectFlags aspects,
                    uint32_t level,
                    uint32_t base_layer, uint32_t layer_count,
                    VkRect2D area, uint8_t stencil_value);
void
anv_image_mcs_op(struct anv_cmd_buffer *cmd_buffer,
                 const struct anv_image *image,
                 enum isl_format format, struct isl_swizzle swizzle,
                 VkImageAspectFlagBits aspect,
                 uint32_t base_layer, uint32_t layer_count,
                 enum isl_aux_op mcs_op, union isl_color_value *clear_value,
                 bool predicate);
void
anv_image_ccs_op(struct anv_cmd_buffer *cmd_buffer,
                 const struct anv_image *image,
                 enum isl_format format, struct isl_swizzle swizzle,
                 VkImageAspectFlagBits aspect, uint32_t level,
                 uint32_t base_layer, uint32_t layer_count,
                 enum isl_aux_op ccs_op, union isl_color_value *clear_value,
                 bool predicate);

void
anv_image_copy_to_shadow(struct anv_cmd_buffer *cmd_buffer,
                         const struct anv_image *image,
                         VkImageAspectFlagBits aspect,
                         uint32_t base_level, uint32_t level_count,
                         uint32_t base_layer, uint32_t layer_count);

enum isl_aux_state ATTRIBUTE_PURE
anv_layout_to_aux_state(const struct gen_device_info * const devinfo,
                        const struct anv_image *image,
                        const VkImageAspectFlagBits aspect,
                        const VkImageLayout layout);

enum isl_aux_usage ATTRIBUTE_PURE
anv_layout_to_aux_usage(const struct gen_device_info * const devinfo,
                        const struct anv_image *image,
                        const VkImageAspectFlagBits aspect,
                        const VkImageUsageFlagBits usage,
                        const VkImageLayout layout);

enum anv_fast_clear_type ATTRIBUTE_PURE
anv_layout_to_fast_clear_type(const struct gen_device_info * const devinfo,
                              const struct anv_image * const image,
                              const VkImageAspectFlagBits aspect,
                              const VkImageLayout layout);

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

static inline uint32_t
anv_get_levelCount(const struct anv_image *image,
                   const VkImageSubresourceRange *range)
{
   return range->levelCount == VK_REMAINING_MIP_LEVELS ?
          image->levels - range->baseMipLevel : range->levelCount;
}

static inline VkImageAspectFlags
anv_image_expand_aspects(const struct anv_image *image,
                         VkImageAspectFlags aspects)
{
   /* If the underlying image has color plane aspects and
    * VK_IMAGE_ASPECT_COLOR_BIT has been requested, then return the aspects of
    * the underlying image. */
   if ((image->aspects & VK_IMAGE_ASPECT_PLANES_BITS_ANV) != 0 &&
       aspects == VK_IMAGE_ASPECT_COLOR_BIT)
      return image->aspects;

   return aspects;
}

static inline bool
anv_image_aspects_compatible(VkImageAspectFlags aspects1,
                             VkImageAspectFlags aspects2)
{
   if (aspects1 == aspects2)
      return true;

   /* Only 1 color aspects are compatibles. */
   if ((aspects1 & VK_IMAGE_ASPECT_ANY_COLOR_BIT_ANV) != 0 &&
       (aspects2 & VK_IMAGE_ASPECT_ANY_COLOR_BIT_ANV) != 0 &&
       util_bitcount(aspects1) == util_bitcount(aspects2))
      return true;

   return false;
}

struct anv_image_view {
   struct vk_object_base base;

   const struct anv_image *image; /**< VkImageViewCreateInfo::image */

   VkImageAspectFlags aspect_mask;
   VkFormat vk_format;
   VkExtent3D extent; /**< Extent of VkImageViewCreateInfo::baseMipLevel. */

   unsigned n_planes;
   struct {
      uint32_t image_plane;

      struct isl_view isl;

      /**
       * RENDER_SURFACE_STATE when using image as a sampler surface with an
       * image layout of SHADER_READ_ONLY_OPTIMAL or
       * DEPTH_STENCIL_READ_ONLY_OPTIMAL.
       */
      struct anv_surface_state optimal_sampler_surface_state;

      /**
       * RENDER_SURFACE_STATE when using image as a sampler surface with an
       * image layout of GENERAL.
       */
      struct anv_surface_state general_sampler_surface_state;

      /**
       * RENDER_SURFACE_STATE when using image as a storage image. Separate
       * states for write-only and readable, using the real format for
       * write-only and the lowered format for readable.
       */
      struct anv_surface_state storage_surface_state;
      struct anv_surface_state writeonly_storage_surface_state;

      struct brw_image_param storage_image_param;
   } planes[3];
};

enum anv_image_view_state_flags {
   ANV_IMAGE_VIEW_STATE_STORAGE_WRITE_ONLY   = (1 << 0),
   ANV_IMAGE_VIEW_STATE_TEXTURE_OPTIMAL      = (1 << 1),
};

void anv_image_fill_surface_state(struct anv_device *device,
                                  const struct anv_image *image,
                                  VkImageAspectFlagBits aspect,
                                  const struct isl_view *view,
                                  isl_surf_usage_flags_t view_usage,
                                  enum isl_aux_usage aux_usage,
                                  const union isl_color_value *clear_color,
                                  enum anv_image_view_state_flags flags,
                                  struct anv_surface_state *state_inout,
                                  struct brw_image_param *image_param_out);

struct anv_image_create_info {
   const VkImageCreateInfo *vk_info;

   /** An opt-in bitmask which filters an ISL-mapping of the Vulkan tiling. */
   isl_tiling_flags_t isl_tiling_flags;

   /** These flags will be added to any derived from VkImageCreateInfo. */
   isl_surf_usage_flags_t isl_extra_usage_flags;

   bool external_format;
};

VkResult anv_image_create(VkDevice _device,
                          const struct anv_image_create_info *info,
                          const VkAllocationCallbacks* alloc,
                          VkImage *pImage);

enum isl_format
anv_isl_format_for_descriptor_type(const struct anv_device *device,
                                   VkDescriptorType type);

static inline VkExtent3D
anv_sanitize_image_extent(const VkImageType imageType,
                          const VkExtent3D imageExtent)
{
   switch (imageType) {
   case VK_IMAGE_TYPE_1D:
      return (VkExtent3D) { imageExtent.width, 1, 1 };
   case VK_IMAGE_TYPE_2D:
      return (VkExtent3D) { imageExtent.width, imageExtent.height, 1 };
   case VK_IMAGE_TYPE_3D:
      return imageExtent;
   default:
      unreachable("invalid image type");
   }
}

static inline VkOffset3D
anv_sanitize_image_offset(const VkImageType imageType,
                          const VkOffset3D imageOffset)
{
   switch (imageType) {
   case VK_IMAGE_TYPE_1D:
      return (VkOffset3D) { imageOffset.x, 0, 0 };
   case VK_IMAGE_TYPE_2D:
      return (VkOffset3D) { imageOffset.x, imageOffset.y, 0 };
   case VK_IMAGE_TYPE_3D:
      return imageOffset;
   default:
      unreachable("invalid image type");
   }
}

VkFormatFeatureFlags
anv_get_image_format_features(const struct gen_device_info *devinfo,
                              VkFormat vk_format,
                              const struct anv_format *anv_format,
                              VkImageTiling vk_tiling,
                              const struct isl_drm_modifier_info *isl_mod_info);

void anv_fill_buffer_surface_state(struct anv_device *device,
                                   struct anv_state state,
                                   enum isl_format format,
                                   isl_surf_usage_flags_t usage,
                                   struct anv_address address,
                                   uint32_t range, uint32_t stride);

static inline void
anv_clear_color_from_att_state(union isl_color_value *clear_color,
                               const struct anv_attachment_state *att_state,
                               const struct anv_image_view *iview)
{
   const struct isl_format_layout *view_fmtl =
      isl_format_get_layout(iview->planes[0].isl.format);

#define COPY_CLEAR_COLOR_CHANNEL(c, i) \
   if (view_fmtl->channels.c.bits) \
      clear_color->u32[i] = att_state->clear_value.color.uint32[i]

   COPY_CLEAR_COLOR_CHANNEL(r, 0);
   COPY_CLEAR_COLOR_CHANNEL(g, 1);
   COPY_CLEAR_COLOR_CHANNEL(b, 2);
   COPY_CLEAR_COLOR_CHANNEL(a, 3);

#undef COPY_CLEAR_COLOR_CHANNEL
}


/* Haswell border color is a bit of a disaster.  Float and unorm formats use a
 * straightforward 32-bit float color in the first 64 bytes.  Instead of using
 * a nice float/integer union like Gfx8+, Haswell specifies the integer border
 * color as a separate entry /after/ the float color.  The layout of this entry
 * also depends on the format's bpp (with extra hacks for RG32), and overlaps.
 *
 * Since we don't know the format/bpp, we can't make any of the border colors
 * containing '1' work for all formats, as it would be in the wrong place for
 * some of them.  We opt to make 32-bit integers work as this seems like the
 * most common option.  Fortunately, transparent black works regardless, as
 * all zeroes is the same in every bit-size.
 */
struct hsw_border_color {
   float float32[4];
   uint32_t _pad0[12];
   uint32_t uint32[4];
   uint32_t _pad1[108];
};

struct gfx8_border_color {
   union {
      float float32[4];
      uint32_t uint32[4];
   };
   /* Pad out to 64 bytes */
   uint32_t _pad[12];
};

struct anv_ycbcr_conversion {
   struct vk_object_base base;

   const struct anv_format *        format;
   VkSamplerYcbcrModelConversion    ycbcr_model;
   VkSamplerYcbcrRange              ycbcr_range;
   VkComponentSwizzle               mapping[4];
   VkChromaLocation                 chroma_offsets[2];
   VkFilter                         chroma_filter;
   bool                             chroma_reconstruction;
};

struct anv_sampler {
   struct vk_object_base        base;

   uint32_t                     state[3][4];
   uint32_t                     n_planes;
   struct anv_ycbcr_conversion *conversion;

   /* Blob of sampler state data which is guaranteed to be 32-byte aligned
    * and with a 32-byte stride for use as bindless samplers.
    */
   struct anv_state             bindless_state;

   struct anv_state             custom_border_color;
};

struct anv_framebuffer {
   struct vk_object_base                        base;

   uint32_t                                     width;
   uint32_t                                     height;
   uint32_t                                     layers;

   uint32_t                                     attachment_count;
   struct anv_image_view *                      attachments[0];
};

struct anv_subpass_attachment {
   VkImageUsageFlagBits usage;
   uint32_t attachment;
   VkImageLayout layout;

   /* Used only with attachment containing stencil data. */
   VkImageLayout stencil_layout;
};

struct anv_subpass {
   uint32_t                                     attachment_count;

   /**
    * A pointer to all attachment references used in this subpass.
    * Only valid if ::attachment_count > 0.
    */
   struct anv_subpass_attachment *              attachments;
   uint32_t                                     input_count;
   struct anv_subpass_attachment *              input_attachments;
   uint32_t                                     color_count;
   struct anv_subpass_attachment *              color_attachments;
   struct anv_subpass_attachment *              resolve_attachments;

   struct anv_subpass_attachment *              depth_stencil_attachment;
   struct anv_subpass_attachment *              ds_resolve_attachment;
   VkResolveModeFlagBitsKHR                     depth_resolve_mode;
   VkResolveModeFlagBitsKHR                     stencil_resolve_mode;

   uint32_t                                     view_mask;

   /** Subpass has a depth/stencil self-dependency */
   bool                                         has_ds_self_dep;

   /** Subpass has at least one color resolve attachment */
   bool                                         has_color_resolve;
};

static inline unsigned
anv_subpass_view_count(const struct anv_subpass *subpass)
{
   return MAX2(1, util_bitcount(subpass->view_mask));
}

struct anv_render_pass_attachment {
   /* TODO: Consider using VkAttachmentDescription instead of storing each of
    * its members individually.
    */
   VkFormat                                     format;
   uint32_t                                     samples;
   VkImageUsageFlags                            usage;
   VkAttachmentLoadOp                           load_op;
   VkAttachmentStoreOp                          store_op;
   VkAttachmentLoadOp                           stencil_load_op;
   VkImageLayout                                initial_layout;
   VkImageLayout                                final_layout;
   VkImageLayout                                first_subpass_layout;

   VkImageLayout                                stencil_initial_layout;
   VkImageLayout                                stencil_final_layout;

   /* The subpass id in which the attachment will be used last. */
   uint32_t                                     last_subpass_idx;
};

struct anv_render_pass {
   struct vk_object_base                        base;

   uint32_t                                     attachment_count;
   uint32_t                                     subpass_count;
   /* An array of subpass_count+1 flushes, one per subpass boundary */
   enum anv_pipe_bits *                         subpass_flushes;
   struct anv_render_pass_attachment *          attachments;
   struct anv_subpass                           subpasses[0];
};

#define ANV_PIPELINE_STATISTICS_MASK 0x000007ff

struct anv_query_pool {
   struct vk_object_base                        base;

   VkQueryType                                  type;
   VkQueryPipelineStatisticFlags                pipeline_statistics;
   /** Stride between slots, in bytes */
   uint32_t                                     stride;
   /** Number of slots in this query pool */
   uint32_t                                     slots;
   struct anv_bo *                              bo;

   /* KHR perf queries : */
   uint32_t                                     pass_size;
   uint32_t                                     data_offset;
   uint32_t                                     snapshot_size;
   uint32_t                                     n_counters;
   struct gen_perf_counter_pass                *counter_pass;
   uint32_t                                     n_passes;
   struct gen_perf_query_info                 **pass_query;
};

static inline uint32_t khr_perf_query_preamble_offset(const struct anv_query_pool *pool,
                                                      uint32_t pass)
{
   return pool->pass_size * pass + 8;
}

int anv_get_instance_entrypoint_index(const char *name);
int anv_get_device_entrypoint_index(const char *name);
int anv_get_physical_device_entrypoint_index(const char *name);

const char *anv_get_instance_entry_name(int index);
const char *anv_get_physical_device_entry_name(int index);
const char *anv_get_device_entry_name(int index);

bool
anv_instance_entrypoint_is_enabled(int index, uint32_t core_version,
                                   const struct vk_instance_extension_table *instance);
bool
anv_physical_device_entrypoint_is_enabled(int index, uint32_t core_version,
                                          const struct vk_instance_extension_table *instance);
bool
anv_device_entrypoint_is_enabled(int index, uint32_t core_version,
                                 const struct vk_instance_extension_table *instance,
                                 const struct vk_device_extension_table *device);

const struct vk_device_dispatch_table *
anv_get_device_dispatch_table(const struct gen_device_info *devinfo);

static inline uint32_t
anv_get_subpass_id(const struct anv_cmd_state * const cmd_state)
{
   /* This function must be called from within a subpass. */
   assert(cmd_state->pass && cmd_state->subpass);

   const uint32_t subpass_id = cmd_state->subpass - cmd_state->pass->subpasses;

   /* The id of this subpass shouldn't exceed the number of subpasses in this
    * render pass minus 1.
    */
   assert(subpass_id < cmd_state->pass->subpass_count);
   return subpass_id;
}

struct anv_performance_configuration_intel {
   struct vk_object_base      base;

   struct gen_perf_registers *register_config;

   uint64_t                   config_id;
};

void anv_physical_device_init_perf(struct anv_physical_device *device, int fd);
void anv_device_perf_init(struct anv_device *device);
void anv_perf_write_pass_results(struct gen_perf_config *perf,
                                 struct anv_query_pool *pool, uint32_t pass,
                                 const struct gen_perf_query_result *accumulated_results,
                                 union VkPerformanceCounterResultKHR *results);

#define ANV_FROM_HANDLE(__anv_type, __name, __handle) \
   VK_FROM_HANDLE(__anv_type, __name, __handle)

VK_DEFINE_HANDLE_CASTS(anv_cmd_buffer, base, VkCommandBuffer,
                       VK_OBJECT_TYPE_COMMAND_BUFFER)
VK_DEFINE_HANDLE_CASTS(anv_device, vk.base, VkDevice, VK_OBJECT_TYPE_DEVICE)
VK_DEFINE_HANDLE_CASTS(anv_instance, vk.base, VkInstance, VK_OBJECT_TYPE_INSTANCE)
VK_DEFINE_HANDLE_CASTS(anv_physical_device, vk.base, VkPhysicalDevice,
                       VK_OBJECT_TYPE_PHYSICAL_DEVICE)
VK_DEFINE_HANDLE_CASTS(anv_queue, base, VkQueue, VK_OBJECT_TYPE_QUEUE)

VK_DEFINE_NONDISP_HANDLE_CASTS(anv_cmd_pool, base, VkCommandPool,
                               VK_OBJECT_TYPE_COMMAND_POOL)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_buffer, base, VkBuffer,
                               VK_OBJECT_TYPE_BUFFER)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_buffer_view, base, VkBufferView,
                               VK_OBJECT_TYPE_BUFFER_VIEW)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_descriptor_pool, base, VkDescriptorPool,
                               VK_OBJECT_TYPE_DESCRIPTOR_POOL)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_descriptor_set, base, VkDescriptorSet,
                               VK_OBJECT_TYPE_DESCRIPTOR_SET)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_descriptor_set_layout, base,
                               VkDescriptorSetLayout,
                               VK_OBJECT_TYPE_DESCRIPTOR_SET_LAYOUT)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_descriptor_update_template, base,
                               VkDescriptorUpdateTemplate,
                               VK_OBJECT_TYPE_DESCRIPTOR_UPDATE_TEMPLATE)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_device_memory, base, VkDeviceMemory,
                               VK_OBJECT_TYPE_DEVICE_MEMORY)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_fence, base, VkFence, VK_OBJECT_TYPE_FENCE)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_event, base, VkEvent, VK_OBJECT_TYPE_EVENT)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_framebuffer, base, VkFramebuffer,
                               VK_OBJECT_TYPE_FRAMEBUFFER)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_image, base, VkImage, VK_OBJECT_TYPE_IMAGE)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_image_view, base, VkImageView,
                               VK_OBJECT_TYPE_IMAGE_VIEW);
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_pipeline_cache, base, VkPipelineCache,
                               VK_OBJECT_TYPE_PIPELINE_CACHE)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_pipeline, base, VkPipeline,
                               VK_OBJECT_TYPE_PIPELINE)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_pipeline_layout, base, VkPipelineLayout,
                               VK_OBJECT_TYPE_PIPELINE_LAYOUT)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_query_pool, base, VkQueryPool,
                               VK_OBJECT_TYPE_QUERY_POOL)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_render_pass, base, VkRenderPass,
                               VK_OBJECT_TYPE_RENDER_PASS)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_sampler, base, VkSampler,
                               VK_OBJECT_TYPE_SAMPLER)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_semaphore, base, VkSemaphore,
                               VK_OBJECT_TYPE_SEMAPHORE)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_ycbcr_conversion, base,
                               VkSamplerYcbcrConversion,
                               VK_OBJECT_TYPE_SAMPLER_YCBCR_CONVERSION)
VK_DEFINE_NONDISP_HANDLE_CASTS(anv_performance_configuration_intel, base,
                               VkPerformanceConfigurationINTEL,
                               VK_OBJECT_TYPE_PERFORMANCE_CONFIGURATION_INTEL)

#define anv_genX(devinfo, thing) ({             \
   __typeof(&gfx9_##thing) genX_thing;          \
   switch ((devinfo)->verx10) {                 \
   case 70:                                     \
      genX_thing = &gfx7_##thing;               \
      break;                                    \
   case 75:                                     \
      genX_thing = &gfx75_##thing;              \
      break;                                    \
   case 80:                                     \
      genX_thing = &gfx8_##thing;               \
      break;                                    \
   case 90:                                     \
      genX_thing = &gfx9_##thing;               \
      break;                                    \
   case 110:                                    \
      genX_thing = &gfx11_##thing;              \
      break;                                    \
   case 120:                                    \
      genX_thing = &gfx12_##thing;              \
      break;                                    \
   case 125:                                    \
      genX_thing = &gfx125_##thing;             \
      break;                                    \
   default:                                     \
      unreachable("Unknown hardware generation"); \
   }                                            \
   genX_thing;                                  \
})

/* Gen-specific function declarations */
#ifdef genX
#  include "anv_genX.h"
#else
#  define genX(x) gfx7_##x
#  include "anv_genX.h"
#  undef genX
#  define genX(x) gfx75_##x
#  include "anv_genX.h"
#  undef genX
#  define genX(x) gfx8_##x
#  include "anv_genX.h"
#  undef genX
#  define genX(x) gfx9_##x
#  include "anv_genX.h"
#  undef genX
#  define genX(x) gfx11_##x
#  include "anv_genX.h"
#  undef genX
#  define genX(x) gfx12_##x
#  include "anv_genX.h"
#  undef genX
#  define genX(x) gfx125_##x
#  include "anv_genX.h"
#  undef genX
#endif

#endif /* ANV_PRIVATE_H */