summaryrefslogtreecommitdiffstats
path: root/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
blob: c705c986bad04acab82656203774a17c6f7babf2 (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
/*
 * Copyright 2006 VMware, Inc.
 * All Rights Reserved.
 *
 * 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 VMWARE AND/OR ITS SUPPLIERS 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.
 */

#include <GL/gl.h>
#include <GL/internal/dri_interface.h>

#include "intel_batchbuffer.h"
#include "intel_mipmap_tree.h"
#include "intel_resolve_map.h"
#include "intel_tex.h"
#include "intel_blit.h"
#include "intel_fbo.h"

#include "brw_blorp.h"
#include "brw_context.h"
#include "brw_state.h"

#include "main/enums.h"
#include "main/fbobject.h"
#include "main/formats.h"
#include "main/glformats.h"
#include "main/texcompress_etc.h"
#include "main/teximage.h"
#include "main/streaming-load-memcpy.h"
#include "x86/common_x86_asm.h"

#define FILE_DEBUG_FLAG DEBUG_MIPTREE

static void *intel_miptree_map_raw(struct brw_context *brw,
                                   struct intel_mipmap_tree *mt);

static void intel_miptree_unmap_raw(struct intel_mipmap_tree *mt);

static bool
intel_miptree_alloc_mcs(struct brw_context *brw,
                        struct intel_mipmap_tree *mt,
                        GLuint num_samples);

/**
 * Determine which MSAA layout should be used by the MSAA surface being
 * created, based on the chip generation and the surface type.
 */
static enum intel_msaa_layout
compute_msaa_layout(struct brw_context *brw, mesa_format format,
                    bool disable_aux_buffers)
{
   /* Prior to Gen7, all MSAA surfaces used IMS layout. */
   if (brw->gen < 7)
      return INTEL_MSAA_LAYOUT_IMS;

   /* In Gen7, IMS layout is only used for depth and stencil buffers. */
   switch (_mesa_get_format_base_format(format)) {
   case GL_DEPTH_COMPONENT:
   case GL_STENCIL_INDEX:
   case GL_DEPTH_STENCIL:
      return INTEL_MSAA_LAYOUT_IMS;
   default:
      /* From the Ivy Bridge PRM, Vol4 Part1 p77 ("MCS Enable"):
       *
       *   This field must be set to 0 for all SINT MSRTs when all RT channels
       *   are not written
       *
       * In practice this means that we have to disable MCS for all signed
       * integer MSAA buffers.  The alternative, to disable MCS only when one
       * of the render target channels is disabled, is impractical because it
       * would require converting between CMS and UMS MSAA layouts on the fly,
       * which is expensive.
       */
      if (brw->gen == 7 && _mesa_get_format_datatype(format) == GL_INT) {
         return INTEL_MSAA_LAYOUT_UMS;
      } else if (disable_aux_buffers) {
         /* We can't use the CMS layout because it uses an aux buffer, the MCS
          * buffer. So fallback to UMS, which is identical to CMS without the
          * MCS. */
         return INTEL_MSAA_LAYOUT_UMS;
      } else {
         return INTEL_MSAA_LAYOUT_CMS;
      }
   }
}


/**
 * For single-sampled render targets ("non-MSRT"), the MCS buffer is a
 * scaled-down bitfield representation of the color buffer which is capable of
 * recording when blocks of the color buffer are equal to the clear value.
 * This function returns the block size that will be used by the MCS buffer
 * corresponding to a certain color miptree.
 *
 * From the Ivy Bridge PRM, Vol2 Part1 11.7 "MCS Buffer for Render Target(s)",
 * beneath the "Fast Color Clear" bullet (p327):
 *
 *     The following table describes the RT alignment
 *
 *                       Pixels  Lines
 *         TiledY RT CL
 *             bpp
 *              32          8      4
 *              64          4      4
 *             128          2      4
 *         TiledX RT CL
 *             bpp
 *              32         16      2
 *              64          8      2
 *             128          4      2
 *
 * This alignment has the following uses:
 *
 * - For figuring out the size of the MCS buffer.  Each 4k tile in the MCS
 *   buffer contains 128 blocks horizontally and 256 blocks vertically.
 *
 * - For figuring out alignment restrictions for a fast clear operation.  Fast
 *   clear operations must always clear aligned multiples of 16 blocks
 *   horizontally and 32 blocks vertically.
 *
 * - For scaling down the coordinates sent through the render pipeline during
 *   a fast clear.  X coordinates must be scaled down by 8 times the block
 *   width, and Y coordinates by 16 times the block height.
 *
 * - For scaling down the coordinates sent through the render pipeline during
 *   a "Render Target Resolve" operation.  X coordinates must be scaled down
 *   by half the block width, and Y coordinates by half the block height.
 */
void
intel_get_non_msrt_mcs_alignment(const struct intel_mipmap_tree *mt,
                                 unsigned *width_px, unsigned *height)
{
   switch (mt->tiling) {
   default:
      unreachable("Non-MSRT MCS requires X or Y tiling");
      /* In release builds, fall through */
   case I915_TILING_Y:
      *width_px = 32 / mt->cpp;
      *height = 4;
      break;
   case I915_TILING_X:
      *width_px = 64 / mt->cpp;
      *height = 2;
   }
}

bool
intel_tiling_supports_non_msrt_mcs(const struct brw_context *brw,
                                   unsigned tiling)
{
   /* From the Ivy Bridge PRM, Vol2 Part1 11.7 "MCS Buffer for Render
    * Target(s)", beneath the "Fast Color Clear" bullet (p326):
    *
    *     - Support is limited to tiled render targets.
    *
    * Gen9 changes the restriction to Y-tile only.
    */
   if (brw->gen >= 9)
      return tiling == I915_TILING_Y;
   else if (brw->gen >= 7)
      return tiling != I915_TILING_NONE;
   else
      return false;
}

/**
 * For a single-sampled render target ("non-MSRT"), determine if an MCS buffer
 * can be used. This doesn't (and should not) inspect any of the properties of
 * the miptree's BO.
 *
 * From the Ivy Bridge PRM, Vol2 Part1 11.7 "MCS Buffer for Render Target(s)",
 * beneath the "Fast Color Clear" bullet (p326):
 *
 *     - Support is for non-mip-mapped and non-array surface types only.
 *
 * And then later, on p327:
 *
 *     - MCS buffer for non-MSRT is supported only for RT formats 32bpp,
 *       64bpp, and 128bpp.
 *
 * From the Skylake documentation, it is made clear that X-tiling is no longer
 * supported:
 *
 *     - MCS and Lossless compression is supported for TiledY/TileYs/TileYf
 *     non-MSRTs only.
 */
bool
intel_miptree_supports_non_msrt_fast_clear(struct brw_context *brw,
                                           const struct intel_mipmap_tree *mt)
{
   /* MCS support does not exist prior to Gen7 */
   if (brw->gen < 7)
      return false;

   if (mt->disable_aux_buffers)
      return false;

   /* This function applies only to non-multisampled render targets. */
   if (mt->num_samples > 1)
      return false;

   /* MCS is only supported for color buffers */
   switch (_mesa_get_format_base_format(mt->format)) {
   case GL_DEPTH_COMPONENT:
   case GL_DEPTH_STENCIL:
   case GL_STENCIL_INDEX:
      return false;
   }

   if (mt->cpp != 4 && mt->cpp != 8 && mt->cpp != 16)
      return false;
   if (mt->first_level != 0 || mt->last_level != 0) {
      if (brw->gen >= 8) {
         perf_debug("Multi-LOD fast clear - giving up (%dx%dx%d).\n",
                    mt->logical_width0, mt->logical_height0, mt->last_level);
      }

      return false;
   }

   /* Check for layered surfaces. */
   if (mt->physical_depth0 != 1) {
       /* Multisample surfaces with the CMS layout are not layered surfaces,
        * yet still have physical_depth0 > 1. Assert that we don't
        * accidentally reject a multisampled surface here. We should have
        * rejected it earlier by explicitly checking the sample count.
        */
      assert(mt->num_samples <= 1);

      if (brw->gen >= 8) {
         perf_debug("Layered fast clear - giving up. (%dx%d%d)\n",
                    mt->logical_width0, mt->logical_height0,
                    mt->physical_depth0);
      }

      return false;
   }

   /* There's no point in using an MCS buffer if the surface isn't in a
    * renderable format.
    */
   if (!brw->format_supported_as_render_target[mt->format])
      return false;

   if (brw->gen >= 9) {
      mesa_format linear_format = _mesa_get_srgb_format_linear(mt->format);
      const uint32_t brw_format = brw_format_for_mesa_format(linear_format);
      return isl_format_supports_lossless_compression(brw->intelScreen->devinfo,
                                                      brw_format);
   } else
      return true;
}

/* On Gen9 support for color buffer compression was extended to single
 * sampled surfaces. This is a helper considering both auxiliary buffer
 * type and number of samples telling if the given miptree represents
 * the new single sampled case - also called lossless compression.
 */
bool
intel_miptree_is_lossless_compressed(const struct brw_context *brw,
                                     const struct intel_mipmap_tree *mt)
{
   /* Only available from Gen9 onwards. */
   if (brw->gen < 9)
      return false;

   /* Compression always requires auxiliary buffer. */
   if (!mt->mcs_mt)
      return false;

   /* Single sample compression is represented re-using msaa compression
    * layout type: "Compressed Multisampled Surfaces".
    */
   if (mt->msaa_layout != INTEL_MSAA_LAYOUT_CMS)
      return false;

   /* And finally distinguish between msaa and single sample case. */
   return mt->num_samples <= 1;
}

bool
intel_miptree_supports_lossless_compressed(struct brw_context *brw,
                                           const struct intel_mipmap_tree *mt)
{
   /* For now compression is only enabled for integer formats even though
    * there exist supported floating point formats also. This is a heuristic
    * decision based on current public benchmarks. In none of the cases these
    * formats provided any improvement but a few cases were seen to regress.
    * Hence these are left to to be enabled in the future when they are known
    * to improve things.
    */
   if (_mesa_get_format_datatype(mt->format) == GL_FLOAT)
      return false;

   /* Fast clear mechanism and lossless compression go hand in hand. */
   if (!intel_miptree_supports_non_msrt_fast_clear(brw, mt))
      return false;

   /* Fast clear can be also used to clear srgb surfaces by using equivalent
    * linear format. This trick, however, can't be extended to be used with
    * lossless compression and therefore a check is needed to see if the format
    * really is linear.
    */
   return _mesa_get_srgb_format_linear(mt->format) == mt->format;
}

/**
 * Determine depth format corresponding to a depth+stencil format,
 * for separate stencil.
 */
mesa_format
intel_depth_format_for_depthstencil_format(mesa_format format) {
   switch (format) {
   case MESA_FORMAT_Z24_UNORM_S8_UINT:
      return MESA_FORMAT_Z24_UNORM_X8_UINT;
   case MESA_FORMAT_Z32_FLOAT_S8X24_UINT:
      return MESA_FORMAT_Z_FLOAT32;
   default:
      return format;
   }
}


/**
 * @param for_bo Indicates that the caller is
 *        intel_miptree_create_for_bo(). If true, then do not create
 *        \c stencil_mt.
 */
static struct intel_mipmap_tree *
intel_miptree_create_layout(struct brw_context *brw,
                            GLenum target,
                            mesa_format format,
                            GLuint first_level,
                            GLuint last_level,
                            GLuint width0,
                            GLuint height0,
                            GLuint depth0,
                            GLuint num_samples,
                            uint32_t layout_flags)
{
   struct intel_mipmap_tree *mt = calloc(sizeof(*mt), 1);
   if (!mt)
      return NULL;

   DBG("%s target %s format %s level %d..%d slices %d <-- %p\n", __func__,
       _mesa_enum_to_string(target),
       _mesa_get_format_name(format),
       first_level, last_level, depth0, mt);

   if (target == GL_TEXTURE_1D_ARRAY) {
      /* For a 1D Array texture the OpenGL API will treat the height0
       * parameter as the number of array slices. For Intel hardware, we treat
       * the 1D array as a 2D Array with a height of 1.
       *
       * So, when we first come through this path to create a 1D Array
       * texture, height0 stores the number of slices, and depth0 is 1. In
       * this case, we want to swap height0 and depth0.
       *
       * Since some miptrees will be created based on the base miptree, we may
       * come through this path and see height0 as 1 and depth0 being the
       * number of slices. In this case we don't need to do the swap.
       */
      assert(height0 == 1 || depth0 == 1);
      if (height0 > 1) {
         depth0 = height0;
         height0 = 1;
      }
   }

   mt->target = target;
   mt->format = format;
   mt->first_level = first_level;
   mt->last_level = last_level;
   mt->logical_width0 = width0;
   mt->logical_height0 = height0;
   mt->logical_depth0 = depth0;
   mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_NO_MCS;
   mt->disable_aux_buffers = (layout_flags & MIPTREE_LAYOUT_DISABLE_AUX) != 0;
   mt->is_scanout = (layout_flags & MIPTREE_LAYOUT_FOR_SCANOUT) != 0;
   exec_list_make_empty(&mt->hiz_map);
   mt->cpp = _mesa_get_format_bytes(format);
   mt->num_samples = num_samples;
   mt->compressed = _mesa_is_format_compressed(format);
   mt->msaa_layout = INTEL_MSAA_LAYOUT_NONE;
   mt->refcount = 1;

   if (num_samples > 1) {
      /* Adjust width/height/depth for MSAA */
      mt->msaa_layout = compute_msaa_layout(brw, format,
                                            mt->disable_aux_buffers);
      if (mt->msaa_layout == INTEL_MSAA_LAYOUT_IMS) {
         /* From the Ivybridge PRM, Volume 1, Part 1, page 108:
          * "If the surface is multisampled and it is a depth or stencil
          *  surface or Multisampled Surface StorageFormat in SURFACE_STATE is
          *  MSFMT_DEPTH_STENCIL, WL and HL must be adjusted as follows before
          *  proceeding:
          *
          *  +----------------------------------------------------------------+
          *  | Num Multisamples |        W_l =         |        H_l =         |
          *  +----------------------------------------------------------------+
          *  |         2        | ceiling(W_l / 2) * 4 | H_l (no adjustment)  |
          *  |         4        | ceiling(W_l / 2) * 4 | ceiling(H_l / 2) * 4 |
          *  |         8        | ceiling(W_l / 2) * 8 | ceiling(H_l / 2) * 4 |
          *  |        16        | ceiling(W_l / 2) * 8 | ceiling(H_l / 2) * 8 |
          *  +----------------------------------------------------------------+
          * "
          *
          * Note that MSFMT_DEPTH_STENCIL just means the IMS (interleaved)
          * format rather than UMS/CMS (array slices).  The Sandybridge PRM,
          * Volume 1, Part 1, Page 111 has the same formula for 4x MSAA.
          *
          * Another more complicated explanation for these adjustments comes
          * from the Sandybridge PRM, volume 4, part 1, page 31:
          *
          *     "Any of the other messages (sample*, LOD, load4) used with a
          *      (4x) multisampled surface will in-effect sample a surface with
          *      double the height and width as that indicated in the surface
          *      state. Each pixel position on the original-sized surface is
          *      replaced with a 2x2 of samples with the following arrangement:
          *
          *         sample 0 sample 2
          *         sample 1 sample 3"
          *
          * Thus, when sampling from a multisampled texture, it behaves as
          * though the layout in memory for (x,y,sample) is:
          *
          *      (0,0,0) (0,0,2)   (1,0,0) (1,0,2)
          *      (0,0,1) (0,0,3)   (1,0,1) (1,0,3)
          *
          *      (0,1,0) (0,1,2)   (1,1,0) (1,1,2)
          *      (0,1,1) (0,1,3)   (1,1,1) (1,1,3)
          *
          * However, the actual layout of multisampled data in memory is:
          *
          *      (0,0,0) (1,0,0)   (0,0,1) (1,0,1)
          *      (0,1,0) (1,1,0)   (0,1,1) (1,1,1)
          *
          *      (0,0,2) (1,0,2)   (0,0,3) (1,0,3)
          *      (0,1,2) (1,1,2)   (0,1,3) (1,1,3)
          *
          * This pattern repeats for each 2x2 pixel block.
          *
          * As a result, when calculating the size of our 4-sample buffer for
          * an odd width or height, we have to align before scaling up because
          * sample 3 is in that bottom right 2x2 block.
          */
         switch (num_samples) {
         case 2:
            assert(brw->gen >= 8);
            width0 = ALIGN(width0, 2) * 2;
            height0 = ALIGN(height0, 2);
            break;
         case 4:
            width0 = ALIGN(width0, 2) * 2;
            height0 = ALIGN(height0, 2) * 2;
            break;
         case 8:
            width0 = ALIGN(width0, 2) * 4;
            height0 = ALIGN(height0, 2) * 2;
            break;
         case 16:
            width0 = ALIGN(width0, 2) * 4;
            height0 = ALIGN(height0, 2) * 4;
            break;
         default:
            /* num_samples should already have been quantized to 0, 1, 2, 4, 8
             * or 16.
             */
            unreachable("not reached");
         }
      } else {
         /* Non-interleaved */
         depth0 *= num_samples;
      }
   }

   /* Set array_layout to ALL_SLICES_AT_EACH_LOD when array_spacing_lod0 can
    * be used. array_spacing_lod0 is only used for non-IMS MSAA surfaces on
    * Gen 7 and 8. On Gen 8 and 9 this layout is not available but it is still
    * used on Gen8 to make it pick a qpitch value which doesn't include space
    * for the mipmaps. On Gen9 this is not necessary because it will
    * automatically pick a packed qpitch value whenever mt->first_level ==
    * mt->last_level.
    * TODO: can we use it elsewhere?
    * TODO: also disable this on Gen8 and pick the qpitch value like Gen9
    */
   if (brw->gen >= 9) {
      mt->array_layout = ALL_LOD_IN_EACH_SLICE;
   } else {
      switch (mt->msaa_layout) {
      case INTEL_MSAA_LAYOUT_NONE:
      case INTEL_MSAA_LAYOUT_IMS:
         mt->array_layout = ALL_LOD_IN_EACH_SLICE;
         break;
      case INTEL_MSAA_LAYOUT_UMS:
      case INTEL_MSAA_LAYOUT_CMS:
         mt->array_layout = ALL_SLICES_AT_EACH_LOD;
         break;
      }
   }

   if (target == GL_TEXTURE_CUBE_MAP) {
      assert(depth0 == 1);
      depth0 = 6;
   }

   mt->physical_width0 = width0;
   mt->physical_height0 = height0;
   mt->physical_depth0 = depth0;

   if (!(layout_flags & MIPTREE_LAYOUT_FOR_BO) &&
       _mesa_get_format_base_format(format) == GL_DEPTH_STENCIL &&
       (brw->must_use_separate_stencil ||
	(brw->has_separate_stencil &&
         intel_miptree_wants_hiz_buffer(brw, mt)))) {
      uint32_t stencil_flags = MIPTREE_LAYOUT_ACCELERATED_UPLOAD;
      if (brw->gen == 6) {
         stencil_flags |= MIPTREE_LAYOUT_FORCE_ALL_SLICE_AT_LOD |
                          MIPTREE_LAYOUT_TILING_ANY;
      }

      mt->stencil_mt = intel_miptree_create(brw,
                                            mt->target,
                                            MESA_FORMAT_S_UINT8,
                                            mt->first_level,
                                            mt->last_level,
                                            mt->logical_width0,
                                            mt->logical_height0,
                                            mt->logical_depth0,
                                            num_samples,
                                            stencil_flags);

      if (!mt->stencil_mt) {
	 intel_miptree_release(&mt);
	 return NULL;
      }

      /* Fix up the Z miptree format for how we're splitting out separate
       * stencil.  Gen7 expects there to be no stencil bits in its depth buffer.
       */
      mt->format = intel_depth_format_for_depthstencil_format(mt->format);
      mt->cpp = 4;

      if (format == mt->format) {
         _mesa_problem(NULL, "Unknown format %s in separate stencil mt\n",
                       _mesa_get_format_name(mt->format));
      }
   }

   if (layout_flags & MIPTREE_LAYOUT_FORCE_ALL_SLICE_AT_LOD)
      mt->array_layout = ALL_SLICES_AT_EACH_LOD;

   /*
    * Obey HALIGN_16 constraints for Gen8 and Gen9 buffers which are
    * multisampled or have an AUX buffer attached to it.
    *
    * GEN  |    MSRT        | AUX_CCS_* or AUX_MCS
    *  -------------------------------------------
    *  9   |  HALIGN_16     |    HALIGN_16
    *  8   |  HALIGN_ANY    |    HALIGN_16
    *  7   |      ?         |        ?
    *  6   |      ?         |        ?
    */
   if (intel_miptree_supports_non_msrt_fast_clear(brw, mt)) {
      if (brw->gen >= 9 || (brw->gen == 8 && num_samples <= 1))
         layout_flags |= MIPTREE_LAYOUT_FORCE_HALIGN16;
   } else if (brw->gen >= 9 && num_samples > 1) {
      layout_flags |= MIPTREE_LAYOUT_FORCE_HALIGN16;
   } else {
      const UNUSED bool is_lossless_compressed_aux =
         brw->gen >= 9 && num_samples == 1 &&
         mt->format == MESA_FORMAT_R_UINT32;

      /* For now, nothing else has this requirement */
      assert(is_lossless_compressed_aux ||
             (layout_flags & MIPTREE_LAYOUT_FORCE_HALIGN16) == 0);
   }

   brw_miptree_layout(brw, mt, layout_flags);

   if (mt->disable_aux_buffers)
      assert(mt->msaa_layout != INTEL_MSAA_LAYOUT_CMS);

   return mt;
}


/**
 * Choose an appropriate uncompressed format for a requested
 * compressed format, if unsupported.
 */
mesa_format
intel_lower_compressed_format(struct brw_context *brw, mesa_format format)
{
   /* No need to lower ETC formats on these platforms,
    * they are supported natively.
    */
   if (brw->gen >= 8 || brw->is_baytrail)
      return format;

   switch (format) {
   case MESA_FORMAT_ETC1_RGB8:
      return MESA_FORMAT_R8G8B8X8_UNORM;
   case MESA_FORMAT_ETC2_RGB8:
      return MESA_FORMAT_R8G8B8X8_UNORM;
   case MESA_FORMAT_ETC2_SRGB8:
   case MESA_FORMAT_ETC2_SRGB8_ALPHA8_EAC:
   case MESA_FORMAT_ETC2_SRGB8_PUNCHTHROUGH_ALPHA1:
      return MESA_FORMAT_B8G8R8A8_SRGB;
   case MESA_FORMAT_ETC2_RGBA8_EAC:
   case MESA_FORMAT_ETC2_RGB8_PUNCHTHROUGH_ALPHA1:
      return MESA_FORMAT_R8G8B8A8_UNORM;
   case MESA_FORMAT_ETC2_R11_EAC:
      return MESA_FORMAT_R_UNORM16;
   case MESA_FORMAT_ETC2_SIGNED_R11_EAC:
      return MESA_FORMAT_R_SNORM16;
   case MESA_FORMAT_ETC2_RG11_EAC:
      return MESA_FORMAT_R16G16_UNORM;
   case MESA_FORMAT_ETC2_SIGNED_RG11_EAC:
      return MESA_FORMAT_R16G16_SNORM;
   default:
      /* Non ETC1 / ETC2 format */
      return format;
   }
}

/* This function computes Yf/Ys tiled bo size, alignment and pitch. */
static unsigned long
intel_get_yf_ys_bo_size(struct intel_mipmap_tree *mt, unsigned *alignment,
                        unsigned long *pitch)
{
   uint32_t tile_width, tile_height;
   unsigned long stride, size, aligned_y;

   assert(mt->tr_mode != INTEL_MIPTREE_TRMODE_NONE);
   intel_get_tile_dims(mt->tiling, mt->tr_mode, mt->cpp,
                       &tile_width, &tile_height);

   aligned_y = ALIGN(mt->total_height, tile_height);
   stride = mt->total_width * mt->cpp;
   stride = ALIGN(stride, tile_width);
   size = stride * aligned_y;

   if (mt->tr_mode == INTEL_MIPTREE_TRMODE_YF) {
      assert(size % 4096 == 0);
      *alignment = 4096;
   } else {
      assert(size % (64 * 1024) == 0);
      *alignment = 64 * 1024;
   }
   *pitch = stride;
   return size;
}

static struct intel_mipmap_tree *
miptree_create(struct brw_context *brw,
               GLenum target,
               mesa_format format,
               GLuint first_level,
               GLuint last_level,
               GLuint width0,
               GLuint height0,
               GLuint depth0,
               GLuint num_samples,
               uint32_t layout_flags)
{
   struct intel_mipmap_tree *mt;
   mesa_format tex_format = format;
   mesa_format etc_format = MESA_FORMAT_NONE;
   uint32_t alloc_flags = 0;

   format = intel_lower_compressed_format(brw, format);

   etc_format = (format != tex_format) ? tex_format : MESA_FORMAT_NONE;

   assert((layout_flags & MIPTREE_LAYOUT_DISABLE_AUX) == 0);
   assert((layout_flags & MIPTREE_LAYOUT_FOR_BO) == 0);
   mt = intel_miptree_create_layout(brw, target, format,
                                    first_level, last_level, width0,
                                    height0, depth0, num_samples,
                                    layout_flags);
   /*
    * pitch == 0 || height == 0  indicates the null texture
    */
   if (!mt || !mt->total_width || !mt->total_height) {
      intel_miptree_release(&mt);
      return NULL;
   }

   if (mt->tiling == (I915_TILING_Y | I915_TILING_X))
      mt->tiling = I915_TILING_Y;

   if (layout_flags & MIPTREE_LAYOUT_ACCELERATED_UPLOAD)
      alloc_flags |= BO_ALLOC_FOR_RENDER;

   unsigned long pitch;
   mt->etc_format = etc_format;

   if (mt->tr_mode != INTEL_MIPTREE_TRMODE_NONE) {
      unsigned alignment = 0;
      unsigned long size;
      size = intel_get_yf_ys_bo_size(mt, &alignment, &pitch);
      assert(size);
      mt->bo = drm_intel_bo_alloc_for_render(brw->bufmgr, "miptree",
                                             size, alignment);
   } else {
      if (format == MESA_FORMAT_S_UINT8) {
         /* Align to size of W tile, 64x64. */
         mt->bo = drm_intel_bo_alloc_tiled(brw->bufmgr, "miptree",
                                           ALIGN(mt->total_width, 64),
                                           ALIGN(mt->total_height, 64),
                                           mt->cpp, &mt->tiling, &pitch,
                                           alloc_flags);
      } else {
         mt->bo = drm_intel_bo_alloc_tiled(brw->bufmgr, "miptree",
                                           mt->total_width, mt->total_height,
                                           mt->cpp, &mt->tiling, &pitch,
                                           alloc_flags);
      }
   }

   mt->pitch = pitch;

   return mt;
}

struct intel_mipmap_tree *
intel_miptree_create(struct brw_context *brw,
                     GLenum target,
                     mesa_format format,
                     GLuint first_level,
                     GLuint last_level,
                     GLuint width0,
                     GLuint height0,
                     GLuint depth0,
                     GLuint num_samples,
                     uint32_t layout_flags)
{
   struct intel_mipmap_tree *mt = miptree_create(
                                     brw, target, format,
                                     first_level, last_level,
                                     width0, height0, depth0, num_samples,
                                     layout_flags);

   /* If the BO is too large to fit in the aperture, we need to use the
    * BLT engine to support it.  Prior to Sandybridge, the BLT paths can't
    * handle Y-tiling, so we need to fall back to X.
    */
   if (brw->gen < 6 && mt->bo->size >= brw->max_gtt_map_object_size &&
       mt->tiling == I915_TILING_Y) {
      unsigned long pitch = mt->pitch;
      const uint32_t alloc_flags =
         (layout_flags & MIPTREE_LAYOUT_ACCELERATED_UPLOAD) ?
         BO_ALLOC_FOR_RENDER : 0;
      perf_debug("%dx%d miptree larger than aperture; falling back to X-tiled\n",
                 mt->total_width, mt->total_height);

      mt->tiling = I915_TILING_X;
      drm_intel_bo_unreference(mt->bo);
      mt->bo = drm_intel_bo_alloc_tiled(brw->bufmgr, "miptree",
                                  mt->total_width, mt->total_height, mt->cpp,
                                  &mt->tiling, &pitch, alloc_flags);
      mt->pitch = pitch;
   }

   mt->offset = 0;

   if (!mt->bo) {
       intel_miptree_release(&mt);
       return NULL;
   }


   if (mt->msaa_layout == INTEL_MSAA_LAYOUT_CMS) {
      assert(mt->num_samples > 1);
      if (!intel_miptree_alloc_mcs(brw, mt, num_samples)) {
         intel_miptree_release(&mt);
         return NULL;
      }
   }

   /* If this miptree is capable of supporting fast color clears, set
    * fast_clear_state appropriately to ensure that fast clears will occur.
    * Allocation of the MCS miptree will be deferred until the first fast
    * clear actually occurs or when compressed single sampled buffer is
    * written by the GPU for the first time.
    */
   if (intel_tiling_supports_non_msrt_mcs(brw, mt->tiling) &&
       intel_miptree_supports_non_msrt_fast_clear(brw, mt)) {
      mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_RESOLVED;
      assert(brw->gen < 8 || mt->halign == 16 || num_samples <= 1);
   }

   return mt;
}

struct intel_mipmap_tree *
intel_miptree_create_for_bo(struct brw_context *brw,
                            drm_intel_bo *bo,
                            mesa_format format,
                            uint32_t offset,
                            uint32_t width,
                            uint32_t height,
                            uint32_t depth,
                            int pitch,
                            uint32_t layout_flags)
{
   struct intel_mipmap_tree *mt;
   uint32_t tiling, swizzle;
   GLenum target;

   drm_intel_bo_get_tiling(bo, &tiling, &swizzle);

   /* Nothing will be able to use this miptree with the BO if the offset isn't
    * aligned.
    */
   if (tiling != I915_TILING_NONE)
      assert(offset % 4096 == 0);

   /* miptrees can't handle negative pitch.  If you need flipping of images,
    * that's outside of the scope of the mt.
    */
   assert(pitch >= 0);

   target = depth > 1 ? GL_TEXTURE_2D_ARRAY : GL_TEXTURE_2D;

   /* The BO already has a tiling format and we shouldn't confuse the lower
    * layers by making it try to find a tiling format again.
    */
   assert((layout_flags & MIPTREE_LAYOUT_TILING_ANY) == 0);
   assert((layout_flags & MIPTREE_LAYOUT_TILING_NONE) == 0);

   layout_flags |= MIPTREE_LAYOUT_FOR_BO;
   mt = intel_miptree_create_layout(brw, target, format,
                                    0, 0,
                                    width, height, depth, 0,
                                    layout_flags);
   if (!mt)
      return NULL;

   drm_intel_bo_reference(bo);
   mt->bo = bo;
   mt->pitch = pitch;
   mt->offset = offset;
   mt->tiling = tiling;

   return mt;
}

/**
 * For a singlesample renderbuffer, this simply wraps the given BO with a
 * miptree.
 *
 * For a multisample renderbuffer, this wraps the window system's
 * (singlesample) BO with a singlesample miptree attached to the
 * intel_renderbuffer, then creates a multisample miptree attached to irb->mt
 * that will contain the actual rendering (which is lazily resolved to
 * irb->singlesample_mt).
 */
void
intel_update_winsys_renderbuffer_miptree(struct brw_context *intel,
                                         struct intel_renderbuffer *irb,
                                         drm_intel_bo *bo,
                                         uint32_t width, uint32_t height,
                                         uint32_t pitch)
{
   struct intel_mipmap_tree *singlesample_mt = NULL;
   struct intel_mipmap_tree *multisample_mt = NULL;
   struct gl_renderbuffer *rb = &irb->Base.Base;
   mesa_format format = rb->Format;
   int num_samples = rb->NumSamples;

   /* Only the front and back buffers, which are color buffers, are allocated
    * through the image loader.
    */
   assert(_mesa_get_format_base_format(format) == GL_RGB ||
          _mesa_get_format_base_format(format) == GL_RGBA);

   singlesample_mt = intel_miptree_create_for_bo(intel,
                                                 bo,
                                                 format,
                                                 0,
                                                 width,
                                                 height,
                                                 1,
                                                 pitch,
                                                 MIPTREE_LAYOUT_FOR_SCANOUT);
   if (!singlesample_mt)
      goto fail;

   /* If this miptree is capable of supporting fast color clears, set
    * mcs_state appropriately to ensure that fast clears will occur.
    * Allocation of the MCS miptree will be deferred until the first fast
    * clear actually occurs.
    */
   if (intel_tiling_supports_non_msrt_mcs(intel, singlesample_mt->tiling) &&
       intel_miptree_supports_non_msrt_fast_clear(intel, singlesample_mt)) {
      singlesample_mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_RESOLVED;
   }

   if (num_samples == 0) {
      intel_miptree_release(&irb->mt);
      irb->mt = singlesample_mt;

      assert(!irb->singlesample_mt);
   } else {
      intel_miptree_release(&irb->singlesample_mt);
      irb->singlesample_mt = singlesample_mt;

      if (!irb->mt ||
          irb->mt->logical_width0 != width ||
          irb->mt->logical_height0 != height) {
         multisample_mt = intel_miptree_create_for_renderbuffer(intel,
                                                                format,
                                                                width,
                                                                height,
                                                                num_samples);
         if (!multisample_mt)
            goto fail;

         irb->need_downsample = false;
         intel_miptree_release(&irb->mt);
         irb->mt = multisample_mt;
      }
   }
   return;

fail:
   intel_miptree_release(&irb->singlesample_mt);
   intel_miptree_release(&irb->mt);
   return;
}

struct intel_mipmap_tree*
intel_miptree_create_for_renderbuffer(struct brw_context *brw,
                                      mesa_format format,
                                      uint32_t width,
                                      uint32_t height,
                                      uint32_t num_samples)
{
   struct intel_mipmap_tree *mt;
   uint32_t depth = 1;
   bool ok;
   GLenum target = num_samples > 1 ? GL_TEXTURE_2D_MULTISAMPLE : GL_TEXTURE_2D;
   const uint32_t layout_flags = MIPTREE_LAYOUT_ACCELERATED_UPLOAD |
                                 MIPTREE_LAYOUT_TILING_ANY |
                                 MIPTREE_LAYOUT_FOR_SCANOUT;

   mt = intel_miptree_create(brw, target, format, 0, 0,
                             width, height, depth, num_samples,
                             layout_flags);
   if (!mt)
      goto fail;

   if (intel_miptree_wants_hiz_buffer(brw, mt)) {
      ok = intel_miptree_alloc_hiz(brw, mt);
      if (!ok)
         goto fail;
   }

   return mt;

fail:
   intel_miptree_release(&mt);
   return NULL;
}

void
intel_miptree_reference(struct intel_mipmap_tree **dst,
                        struct intel_mipmap_tree *src)
{
   if (*dst == src)
      return;

   intel_miptree_release(dst);

   if (src) {
      src->refcount++;
      DBG("%s %p refcount now %d\n", __func__, src, src->refcount);
   }

   *dst = src;
}


void
intel_miptree_release(struct intel_mipmap_tree **mt)
{
   if (!*mt)
      return;

   DBG("%s %p refcount will be %d\n", __func__, *mt, (*mt)->refcount - 1);
   if (--(*mt)->refcount <= 0) {
      GLuint i;

      DBG("%s deleting %p\n", __func__, *mt);

      drm_intel_bo_unreference((*mt)->bo);
      intel_miptree_release(&(*mt)->stencil_mt);
      if ((*mt)->hiz_buf) {
         if ((*mt)->hiz_buf->mt)
            intel_miptree_release(&(*mt)->hiz_buf->mt);
         else
            drm_intel_bo_unreference((*mt)->hiz_buf->bo);
         free((*mt)->hiz_buf);
      }
      intel_miptree_release(&(*mt)->mcs_mt);
      intel_resolve_map_clear(&(*mt)->hiz_map);

      intel_miptree_release(&(*mt)->plane[0]);
      intel_miptree_release(&(*mt)->plane[1]);

      for (i = 0; i < MAX_TEXTURE_LEVELS; i++) {
	 free((*mt)->level[i].slice);
      }

      free(*mt);
   }
   *mt = NULL;
}


void
intel_get_image_dims(struct gl_texture_image *image,
                     int *width, int *height, int *depth)
{
   switch (image->TexObject->Target) {
   case GL_TEXTURE_1D_ARRAY:
      /* For a 1D Array texture the OpenGL API will treat the image height as
       * the number of array slices. For Intel hardware, we treat the 1D array
       * as a 2D Array with a height of 1. So, here we want to swap image
       * height and depth.
       */
      *width = image->Width;
      *height = 1;
      *depth = image->Height;
      break;
   default:
      *width = image->Width;
      *height = image->Height;
      *depth = image->Depth;
      break;
   }
}

/**
 * Can the image be pulled into a unified mipmap tree?  This mirrors
 * the completeness test in a lot of ways.
 *
 * Not sure whether I want to pass gl_texture_image here.
 */
bool
intel_miptree_match_image(struct intel_mipmap_tree *mt,
                          struct gl_texture_image *image)
{
   struct intel_texture_image *intelImage = intel_texture_image(image);
   GLuint level = intelImage->base.Base.Level;
   int width, height, depth;

   /* glTexImage* choose the texture object based on the target passed in, and
    * objects can't change targets over their lifetimes, so this should be
    * true.
    */
   assert(image->TexObject->Target == mt->target);

   mesa_format mt_format = mt->format;
   if (mt->format == MESA_FORMAT_Z24_UNORM_X8_UINT && mt->stencil_mt)
      mt_format = MESA_FORMAT_Z24_UNORM_S8_UINT;
   if (mt->format == MESA_FORMAT_Z_FLOAT32 && mt->stencil_mt)
      mt_format = MESA_FORMAT_Z32_FLOAT_S8X24_UINT;
   if (mt->etc_format != MESA_FORMAT_NONE)
      mt_format = mt->etc_format;

   if (image->TexFormat != mt_format)
      return false;

   intel_get_image_dims(image, &width, &height, &depth);

   if (mt->target == GL_TEXTURE_CUBE_MAP)
      depth = 6;

   int level_depth = mt->level[level].depth;
   if (mt->num_samples > 1) {
      switch (mt->msaa_layout) {
      case INTEL_MSAA_LAYOUT_NONE:
      case INTEL_MSAA_LAYOUT_IMS:
         break;
      case INTEL_MSAA_LAYOUT_UMS:
      case INTEL_MSAA_LAYOUT_CMS:
         level_depth /= mt->num_samples;
         break;
      }
   }

   /* Test image dimensions against the base level image adjusted for
    * minification.  This will also catch images not present in the
    * tree, changed targets, etc.
    */
   if (width != minify(mt->logical_width0, level - mt->first_level) ||
       height != minify(mt->logical_height0, level - mt->first_level) ||
       depth != level_depth) {
      return false;
   }

   if (image->NumSamples != mt->num_samples)
      return false;

   return true;
}


void
intel_miptree_set_level_info(struct intel_mipmap_tree *mt,
			     GLuint level,
			     GLuint x, GLuint y, GLuint d)
{
   mt->level[level].depth = d;
   mt->level[level].level_x = x;
   mt->level[level].level_y = y;

   DBG("%s level %d, depth %d, offset %d,%d\n", __func__,
       level, d, x, y);

   assert(mt->level[level].slice == NULL);

   mt->level[level].slice = calloc(d, sizeof(*mt->level[0].slice));
   mt->level[level].slice[0].x_offset = mt->level[level].level_x;
   mt->level[level].slice[0].y_offset = mt->level[level].level_y;
}


void
intel_miptree_set_image_offset(struct intel_mipmap_tree *mt,
			       GLuint level, GLuint img,
			       GLuint x, GLuint y)
{
   if (img == 0 && level == 0)
      assert(x == 0 && y == 0);

   assert(img < mt->level[level].depth);

   mt->level[level].slice[img].x_offset = mt->level[level].level_x + x;
   mt->level[level].slice[img].y_offset = mt->level[level].level_y + y;

   DBG("%s level %d img %d pos %d,%d\n",
       __func__, level, img,
       mt->level[level].slice[img].x_offset,
       mt->level[level].slice[img].y_offset);
}

void
intel_miptree_get_image_offset(const struct intel_mipmap_tree *mt,
			       GLuint level, GLuint slice,
			       GLuint *x, GLuint *y)
{
   assert(slice < mt->level[level].depth);

   *x = mt->level[level].slice[slice].x_offset;
   *y = mt->level[level].slice[slice].y_offset;
}


/**
 * This function computes the tile_w (in bytes) and tile_h (in rows) of
 * different tiling patterns. If the BO is untiled, tile_w is set to cpp
 * and tile_h is set to 1.
 */
void
intel_get_tile_dims(uint32_t tiling, uint32_t tr_mode, uint32_t cpp,
                    uint32_t *tile_w, uint32_t *tile_h)
{
   if (tr_mode == INTEL_MIPTREE_TRMODE_NONE) {
      switch (tiling) {
      case I915_TILING_X:
         *tile_w = 512;
         *tile_h = 8;
         break;
      case I915_TILING_Y:
         *tile_w = 128;
         *tile_h = 32;
         break;
      case I915_TILING_NONE:
         *tile_w = cpp;
         *tile_h = 1;
         break;
      default:
         unreachable("not reached");
      }
   } else {
      uint32_t aspect_ratio = 1;
      assert(_mesa_is_pow_two(cpp));

      switch (cpp) {
      case 1:
         *tile_h = 64;
         break;
      case 2:
      case 4:
         *tile_h = 32;
         break;
      case 8:
      case 16:
         *tile_h = 16;
         break;
      default:
         unreachable("not reached");
      }

      if (cpp == 2 || cpp == 8)
         aspect_ratio = 2;

      if (tr_mode == INTEL_MIPTREE_TRMODE_YS)
         *tile_h *= 4;

      *tile_w = *tile_h * aspect_ratio * cpp;
   }
}


/**
 * This function computes masks that may be used to select the bits of the X
 * and Y coordinates that indicate the offset within a tile.  If the BO is
 * untiled, the masks are set to 0.
 */
void
intel_get_tile_masks(uint32_t tiling, uint32_t tr_mode, uint32_t cpp,
                     bool map_stencil_as_y_tiled,
                     uint32_t *mask_x, uint32_t *mask_y)
{
   uint32_t tile_w_bytes, tile_h;
   if (map_stencil_as_y_tiled)
      tiling = I915_TILING_Y;

   intel_get_tile_dims(tiling, tr_mode, cpp, &tile_w_bytes, &tile_h);

   *mask_x = tile_w_bytes / cpp - 1;
   *mask_y = tile_h - 1;
}

/**
 * Compute the offset (in bytes) from the start of the BO to the given x
 * and y coordinate.  For tiled BOs, caller must ensure that x and y are
 * multiples of the tile size.
 */
uint32_t
intel_miptree_get_aligned_offset(const struct intel_mipmap_tree *mt,
                                 uint32_t x, uint32_t y,
                                 bool map_stencil_as_y_tiled)
{
   int cpp = mt->cpp;
   uint32_t pitch = mt->pitch;
   uint32_t tiling = mt->tiling;

   if (map_stencil_as_y_tiled) {
      tiling = I915_TILING_Y;

      /* When mapping a W-tiled stencil buffer as Y-tiled, each 64-high W-tile
       * gets transformed into a 32-high Y-tile.  Accordingly, the pitch of
       * the resulting surface is twice the pitch of the original miptree,
       * since each row in the Y-tiled view corresponds to two rows in the
       * actual W-tiled surface.  So we need to correct the pitch before
       * computing the offsets.
       */
      pitch *= 2;
   }

   switch (tiling) {
   default:
      unreachable("not reached");
   case I915_TILING_NONE:
      return y * pitch + x * cpp;
   case I915_TILING_X:
      assert((x % (512 / cpp)) == 0);
      assert((y % 8) == 0);
      return y * pitch + x / (512 / cpp) * 4096;
   case I915_TILING_Y:
      assert((x % (128 / cpp)) == 0);
      assert((y % 32) == 0);
      return y * pitch + x / (128 / cpp) * 4096;
   }
}

/**
 * Rendering with tiled buffers requires that the base address of the buffer
 * be aligned to a page boundary.  For renderbuffers, and sometimes with
 * textures, we may want the surface to point at a texture image level that
 * isn't at a page boundary.
 *
 * This function returns an appropriately-aligned base offset
 * according to the tiling restrictions, plus any required x/y offset
 * from there.
 */
uint32_t
intel_miptree_get_tile_offsets(const struct intel_mipmap_tree *mt,
                               GLuint level, GLuint slice,
                               uint32_t *tile_x,
                               uint32_t *tile_y)
{
   uint32_t x, y;
   uint32_t mask_x, mask_y;

   intel_get_tile_masks(mt->tiling, mt->tr_mode, mt->cpp, false, &mask_x, &mask_y);
   intel_miptree_get_image_offset(mt, level, slice, &x, &y);

   *tile_x = x & mask_x;
   *tile_y = y & mask_y;

   return intel_miptree_get_aligned_offset(mt, x & ~mask_x, y & ~mask_y, false);
}

static void
intel_miptree_copy_slice_sw(struct brw_context *brw,
                            struct intel_mipmap_tree *dst_mt,
                            struct intel_mipmap_tree *src_mt,
                            int level,
                            int slice,
                            int width,
                            int height)
{
   void *src, *dst;
   ptrdiff_t src_stride, dst_stride;
   int cpp = dst_mt->cpp;

   intel_miptree_map(brw, src_mt,
                     level, slice,
                     0, 0,
                     width, height,
                     GL_MAP_READ_BIT | BRW_MAP_DIRECT_BIT,
                     &src, &src_stride);

   intel_miptree_map(brw, dst_mt,
                     level, slice,
                     0, 0,
                     width, height,
                     GL_MAP_WRITE_BIT | GL_MAP_INVALIDATE_RANGE_BIT |
                     BRW_MAP_DIRECT_BIT,
                     &dst, &dst_stride);

   DBG("sw blit %s mt %p %p/%"PRIdPTR" -> %s mt %p %p/%"PRIdPTR" (%dx%d)\n",
       _mesa_get_format_name(src_mt->format),
       src_mt, src, src_stride,
       _mesa_get_format_name(dst_mt->format),
       dst_mt, dst, dst_stride,
       width, height);

   int row_size = cpp * width;
   if (src_stride == row_size &&
       dst_stride == row_size) {
      memcpy(dst, src, row_size * height);
   } else {
      for (int i = 0; i < height; i++) {
         memcpy(dst, src, row_size);
         dst += dst_stride;
         src += src_stride;
      }
   }

   intel_miptree_unmap(brw, dst_mt, level, slice);
   intel_miptree_unmap(brw, src_mt, level, slice);

   /* Don't forget to copy the stencil data over, too.  We could have skipped
    * passing BRW_MAP_DIRECT_BIT, but that would have meant intel_miptree_map
    * shuffling the two data sources in/out of temporary storage instead of
    * the direct mapping we get this way.
    */
   if (dst_mt->stencil_mt) {
      assert(src_mt->stencil_mt);
      intel_miptree_copy_slice_sw(brw, dst_mt->stencil_mt, src_mt->stencil_mt,
                                  level, slice, width, height);
   }
}

static void
intel_miptree_copy_slice(struct brw_context *brw,
			 struct intel_mipmap_tree *dst_mt,
			 struct intel_mipmap_tree *src_mt,
			 int level,
			 int face,
			 int depth)

{
   mesa_format format = src_mt->format;
   uint32_t width = minify(src_mt->physical_width0, level - src_mt->first_level);
   uint32_t height = minify(src_mt->physical_height0, level - src_mt->first_level);
   int slice;

   if (face > 0)
      slice = face;
   else
      slice = depth;

   assert(depth < src_mt->level[level].depth);
   assert(src_mt->format == dst_mt->format);

   if (dst_mt->compressed) {
      unsigned int i, j;
      _mesa_get_format_block_size(dst_mt->format, &i, &j);
      height = ALIGN_NPOT(height, j) / j;
      width = ALIGN_NPOT(width, i) / i;
   }

   /* If it's a packed depth/stencil buffer with separate stencil, the blit
    * below won't apply since we can't do the depth's Y tiling or the
    * stencil's W tiling in the blitter.
    */
   if (src_mt->stencil_mt) {
      intel_miptree_copy_slice_sw(brw,
                                  dst_mt, src_mt,
                                  level, slice,
                                  width, height);
      return;
   }

   uint32_t dst_x, dst_y, src_x, src_y;
   intel_miptree_get_image_offset(dst_mt, level, slice, &dst_x, &dst_y);
   intel_miptree_get_image_offset(src_mt, level, slice, &src_x, &src_y);

   DBG("validate blit mt %s %p %d,%d/%d -> mt %s %p %d,%d/%d (%dx%d)\n",
       _mesa_get_format_name(src_mt->format),
       src_mt, src_x, src_y, src_mt->pitch,
       _mesa_get_format_name(dst_mt->format),
       dst_mt, dst_x, dst_y, dst_mt->pitch,
       width, height);

   if (!intel_miptree_blit(brw,
                           src_mt, level, slice, 0, 0, false,
                           dst_mt, level, slice, 0, 0, false,
                           width, height, GL_COPY)) {
      perf_debug("miptree validate blit for %s failed\n",
                 _mesa_get_format_name(format));

      intel_miptree_copy_slice_sw(brw, dst_mt, src_mt, level, slice,
                                  width, height);
   }
}

/**
 * Copies the image's current data to the given miptree, and associates that
 * miptree with the image.
 *
 * If \c invalidate is true, then the actual image data does not need to be
 * copied, but the image still needs to be associated to the new miptree (this
 * is set to true if we're about to clear the image).
 */
void
intel_miptree_copy_teximage(struct brw_context *brw,
			    struct intel_texture_image *intelImage,
			    struct intel_mipmap_tree *dst_mt,
                            bool invalidate)
{
   struct intel_mipmap_tree *src_mt = intelImage->mt;
   struct intel_texture_object *intel_obj =
      intel_texture_object(intelImage->base.Base.TexObject);
   int level = intelImage->base.Base.Level;
   int face = intelImage->base.Base.Face;

   GLuint depth;
   if (intel_obj->base.Target == GL_TEXTURE_1D_ARRAY)
      depth = intelImage->base.Base.Height;
   else
      depth = intelImage->base.Base.Depth;

   if (!invalidate) {
      for (int slice = 0; slice < depth; slice++) {
         intel_miptree_copy_slice(brw, dst_mt, src_mt, level, face, slice);
      }
   }

   intel_miptree_reference(&intelImage->mt, dst_mt);
   intel_obj->needs_validate = true;
}

static void
intel_miptree_init_mcs(struct brw_context *brw,
                       struct intel_mipmap_tree *mt,
                       int init_value)
{
   /* From the Ivy Bridge PRM, Vol 2 Part 1 p326:
    *
    *     When MCS buffer is enabled and bound to MSRT, it is required that it
    *     is cleared prior to any rendering.
    *
    * Since we don't use the MCS buffer for any purpose other than rendering,
    * it makes sense to just clear it immediately upon allocation.
    *
    * Note: the clear value for MCS buffers is all 1's, so we memset to 0xff.
    */
   void *data = intel_miptree_map_raw(brw, mt->mcs_mt);
   memset(data, init_value, mt->mcs_mt->total_height * mt->mcs_mt->pitch);
   intel_miptree_unmap_raw(mt->mcs_mt);
   mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_CLEAR;
}

static bool
intel_miptree_alloc_mcs(struct brw_context *brw,
                        struct intel_mipmap_tree *mt,
                        GLuint num_samples)
{
   assert(brw->gen >= 7); /* MCS only used on Gen7+ */
   assert(mt->mcs_mt == NULL);
   assert(!mt->disable_aux_buffers);

   /* Choose the correct format for the MCS buffer.  All that really matters
    * is that we allocate the right buffer size, since we'll always be
    * accessing this miptree using MCS-specific hardware mechanisms, which
    * infer the correct format based on num_samples.
    */
   mesa_format format;
   switch (num_samples) {
   case 2:
   case 4:
      /* 8 bits/pixel are required for MCS data when using 4x MSAA (2 bits for
       * each sample).
       */
      format = MESA_FORMAT_R_UNORM8;
      break;
   case 8:
      /* 32 bits/pixel are required for MCS data when using 8x MSAA (3 bits
       * for each sample, plus 8 padding bits).
       */
      format = MESA_FORMAT_R_UINT32;
      break;
   case 16:
      /* 64 bits/pixel are required for MCS data when using 16x MSAA (4 bits
       * for each sample).
       */
      format = MESA_FORMAT_RG_UINT32;
      break;
   default:
      unreachable("Unrecognized sample count in intel_miptree_alloc_mcs");
   };

   /* From the Ivy Bridge PRM, Vol4 Part1 p76, "MCS Base Address":
    *
    *     "The MCS surface must be stored as Tile Y."
    */
   const uint32_t mcs_flags = MIPTREE_LAYOUT_ACCELERATED_UPLOAD |
                              MIPTREE_LAYOUT_TILING_Y;
   mt->mcs_mt = miptree_create(brw,
                               mt->target,
                               format,
                               mt->first_level,
                               mt->last_level,
                               mt->logical_width0,
                               mt->logical_height0,
                               mt->logical_depth0,
                               0 /* num_samples */,
                               mcs_flags);

   intel_miptree_init_mcs(brw, mt, 0xFF);

   return mt->mcs_mt;
}


bool
intel_miptree_alloc_non_msrt_mcs(struct brw_context *brw,
                                 struct intel_mipmap_tree *mt)
{
   assert(mt->mcs_mt == NULL);
   assert(!mt->disable_aux_buffers);

   /* The format of the MCS buffer is opaque to the driver; all that matters
    * is that we get its size and pitch right.  We'll pretend that the format
    * is R32.  Since an MCS tile covers 128 blocks horizontally, and a Y-tiled
    * R32 buffer is 32 pixels across, we'll need to scale the width down by
    * the block width and then a further factor of 4.  Since an MCS tile
    * covers 256 blocks vertically, and a Y-tiled R32 buffer is 32 rows high,
    * we'll need to scale the height down by the block height and then a
    * further factor of 8.
    */
   const mesa_format format = MESA_FORMAT_R_UINT32;
   unsigned block_width_px;
   unsigned block_height;
   intel_get_non_msrt_mcs_alignment(mt, &block_width_px, &block_height);
   unsigned width_divisor = block_width_px * 4;
   unsigned height_divisor = block_height * 8;

   /* The Skylake MCS is twice as tall as the Broadwell MCS.
    *
    * In pre-Skylake, each bit in the MCS contained the state of 2 cachelines
    * in the main surface. In Skylake, it's two bits.  The extra bit
    * doubles the MCS height, not width, because in Skylake the MCS is always
    * Y-tiled.
    */
   if (brw->gen >= 9)
      height_divisor /= 2;

   unsigned mcs_width =
      ALIGN(mt->logical_width0, width_divisor) / width_divisor;
   unsigned mcs_height =
      ALIGN(mt->logical_height0, height_divisor) / height_divisor;
   assert(mt->logical_depth0 == 1);
   uint32_t layout_flags = MIPTREE_LAYOUT_TILING_Y;

   if (brw->gen >= 8) {
      layout_flags |= MIPTREE_LAYOUT_FORCE_HALIGN16;
   }

   /* On Gen9+ clients are not currently capable of consuming compressed
    * single-sampled buffers. Disabling compression allows us to skip
    * resolves.
    */
   const bool lossless_compression_disabled = INTEL_DEBUG & DEBUG_NO_RBC;
   const bool is_lossless_compressed =
      unlikely(!lossless_compression_disabled) &&
      brw->gen >= 9 && !mt->is_scanout &&
      intel_miptree_supports_lossless_compressed(brw, mt);

   /* In case of compression mcs buffer needs to be initialised requiring the
    * buffer to be immediately mapped to cpu space for writing. Therefore do
    * not use the gpu access flag which can cause an unnecessary delay if the
    * backing pages happened to be just used by the GPU.
    */
   if (!is_lossless_compressed)
      layout_flags |= MIPTREE_LAYOUT_ACCELERATED_UPLOAD;

   mt->mcs_mt = miptree_create(brw,
                               mt->target,
                               format,
                               mt->first_level,
                               mt->last_level,
                               mcs_width,
                               mcs_height,
                               mt->logical_depth0,
                               0 /* num_samples */,
                               layout_flags);

   /* From Gen9 onwards single-sampled (non-msrt) auxiliary buffers are
    * used for lossless compression which requires similar initialisation
    * as multi-sample compression.
    */
   if (is_lossless_compressed) {
      /* Hardware sets the auxiliary buffer to all zeroes when it does full
       * resolve. Initialize it accordingly in case the first renderer is
       * cpu (or other none compression aware party).
       *
       * This is also explicitly stated in the spec (MCS Buffer for Render
       * Target(s)):
       *   "If Software wants to enable Color Compression without Fast clear,
       *    Software needs to initialize MCS with zeros."
       */
      intel_miptree_init_mcs(brw, mt, 0);
      mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_RESOLVED;
      mt->msaa_layout = INTEL_MSAA_LAYOUT_CMS;
   }

   return mt->mcs_mt;
}

void
intel_miptree_prepare_mcs(struct brw_context *brw,
                          struct intel_mipmap_tree *mt)
{
   if (mt->mcs_mt)
      return;

   if (brw->gen < 9)
      return;

   /* Single sample compression is represented re-using msaa compression
    * layout type: "Compressed Multisampled Surfaces".
    */
   if (mt->msaa_layout != INTEL_MSAA_LAYOUT_CMS || mt->num_samples > 1)
      return;

   /* Clients are not currently capable of consuming compressed
    * single-sampled buffers.
    */
   if (mt->is_scanout)
      return;

   assert(intel_tiling_supports_non_msrt_mcs(brw, mt->tiling) ||
          intel_miptree_supports_lossless_compressed(brw, mt));

   /* Consider if lossless compression is supported but the needed
    * auxiliary buffer doesn't exist yet.
    *
    * Failing to allocate the auxiliary buffer means running out of
    * memory. The pointer to the aux miptree is left NULL which should
    * signal non-compressed behavior.
    */
   if (!intel_miptree_alloc_non_msrt_mcs(brw, mt)) {
      _mesa_warning(NULL,
                    "Failed to allocated aux buffer for lossless"
                    " compressed %p %u:%u %s\n",
                    mt, mt->logical_width0, mt->logical_height0,
                    _mesa_get_format_name(mt->format));
   }
}

/**
 * Helper for intel_miptree_alloc_hiz() that sets
 * \c mt->level[level].has_hiz. Return true if and only if
 * \c has_hiz was set.
 */
static bool
intel_miptree_level_enable_hiz(struct brw_context *brw,
                               struct intel_mipmap_tree *mt,
                               uint32_t level)
{
   assert(mt->hiz_buf);

   if (brw->gen >= 8 || brw->is_haswell) {
      uint32_t width = minify(mt->physical_width0, level);
      uint32_t height = minify(mt->physical_height0, level);

      /* Disable HiZ for LOD > 0 unless the width is 8 aligned
       * and the height is 4 aligned. This allows our HiZ support
       * to fulfill Haswell restrictions for HiZ ops. For LOD == 0,
       * we can grow the width & height to allow the HiZ op to
       * force the proper size alignments.
       */
      if (level > 0 && ((width & 7) || (height & 3))) {
         DBG("mt %p level %d: HiZ DISABLED\n", mt, level);
         return false;
      }
   }

   DBG("mt %p level %d: HiZ enabled\n", mt, level);
   mt->level[level].has_hiz = true;
   return true;
}


/**
 * Helper for intel_miptree_alloc_hiz() that determines the required hiz
 * buffer dimensions and allocates a bo for the hiz buffer.
 */
static struct intel_miptree_aux_buffer *
intel_gen7_hiz_buf_create(struct brw_context *brw,
                          struct intel_mipmap_tree *mt)
{
   unsigned z_width = mt->logical_width0;
   unsigned z_height = mt->logical_height0;
   const unsigned z_depth = MAX2(mt->logical_depth0, 1);
   unsigned hz_width, hz_height;
   struct intel_miptree_aux_buffer *buf = calloc(sizeof(*buf), 1);

   if (!buf)
      return NULL;

   /* Gen7 PRM Volume 2, Part 1, 11.5.3 "Hierarchical Depth Buffer" documents
    * adjustments required for Z_Height and Z_Width based on multisampling.
    */
   switch (mt->num_samples) {
   case 0:
   case 1:
      break;
   case 2:
   case 4:
      z_width *= 2;
      z_height *= 2;
      break;
   case 8:
      z_width *= 4;
      z_height *= 2;
      break;
   default:
      unreachable("unsupported sample count");
   }

   const unsigned vertical_align = 8; /* 'j' in the docs */
   const unsigned H0 = z_height;
   const unsigned h0 = ALIGN(H0, vertical_align);
   const unsigned h1 = ALIGN(minify(H0, 1), vertical_align);
   const unsigned Z0 = z_depth;

   /* HZ_Width (bytes) = ceiling(Z_Width / 16) * 16 */
   hz_width = ALIGN(z_width, 16);

   if (mt->target == GL_TEXTURE_3D) {
      unsigned H_i = H0;
      unsigned Z_i = Z0;
      hz_height = 0;
      for (unsigned level = mt->first_level; level <= mt->last_level; ++level) {
         unsigned h_i = ALIGN(H_i, vertical_align);
         /* sum(i=0 to m; h_i * max(1, floor(Z_Depth/2**i))) */
         hz_height += h_i * Z_i;
         H_i = minify(H_i, 1);
         Z_i = minify(Z_i, 1);
      }
      /* HZ_Height =
       *    (1/2) * sum(i=0 to m; h_i * max(1, floor(Z_Depth/2**i)))
       */
      hz_height = DIV_ROUND_UP(hz_height, 2);
   } else {
      const unsigned hz_qpitch = h0 + h1 + (12 * vertical_align);
      if (mt->target == GL_TEXTURE_CUBE_MAP_ARRAY ||
          mt->target == GL_TEXTURE_CUBE_MAP) {
         /* HZ_Height (rows) = Ceiling ( ( Q_pitch * Z_depth * 6/2) /8 ) * 8 */
         hz_height = DIV_ROUND_UP(hz_qpitch * Z0 * 6, 2 * 8) * 8;
      } else {
         /* HZ_Height (rows) = Ceiling ( ( Q_pitch * Z_depth/2) /8 ) * 8 */
         hz_height = DIV_ROUND_UP(hz_qpitch * Z0, 2 * 8) * 8;
      }
   }

   unsigned long pitch;
   uint32_t tiling = I915_TILING_Y;
   buf->bo = drm_intel_bo_alloc_tiled(brw->bufmgr, "hiz",
                                      hz_width, hz_height, 1,
                                      &tiling, &pitch,
                                      BO_ALLOC_FOR_RENDER);
   if (!buf->bo) {
      free(buf);
      return NULL;
   } else if (tiling != I915_TILING_Y) {
      drm_intel_bo_unreference(buf->bo);
      free(buf);
      return NULL;
   }

   buf->pitch = pitch;

   return buf;
}


/**
 * Helper for intel_miptree_alloc_hiz() that determines the required hiz
 * buffer dimensions and allocates a bo for the hiz buffer.
 */
static struct intel_miptree_aux_buffer *
intel_gen8_hiz_buf_create(struct brw_context *brw,
                          struct intel_mipmap_tree *mt)
{
   unsigned z_width = mt->logical_width0;
   unsigned z_height = mt->logical_height0;
   const unsigned z_depth = MAX2(mt->logical_depth0, 1);
   unsigned hz_width, hz_height;
   struct intel_miptree_aux_buffer *buf = calloc(sizeof(*buf), 1);

   if (!buf)
      return NULL;

   /* Gen7 PRM Volume 2, Part 1, 11.5.3 "Hierarchical Depth Buffer" documents
    * adjustments required for Z_Height and Z_Width based on multisampling.
    */
   if (brw->gen < 9) {
      switch (mt->num_samples) {
      case 0:
      case 1:
         break;
      case 2:
      case 4:
         z_width *= 2;
         z_height *= 2;
         break;
      case 8:
         z_width *= 4;
         z_height *= 2;
         break;
      default:
         unreachable("unsupported sample count");
      }
   }

   const unsigned vertical_align = 8; /* 'j' in the docs */
   const unsigned H0 = z_height;
   const unsigned h0 = ALIGN(H0, vertical_align);
   const unsigned h1 = ALIGN(minify(H0, 1), vertical_align);
   const unsigned Z0 = z_depth;

   /* HZ_Width (bytes) = ceiling(Z_Width / 16) * 16 */
   hz_width = ALIGN(z_width, 16);

   unsigned H_i = H0;
   unsigned Z_i = Z0;
   unsigned sum_h_i = 0;
   unsigned hz_height_3d_sum = 0;
   for (unsigned level = mt->first_level; level <= mt->last_level; ++level) {
      unsigned i = level - mt->first_level;
      unsigned h_i = ALIGN(H_i, vertical_align);
      /* sum(i=2 to m; h_i) */
      if (i >= 2) {
         sum_h_i += h_i;
      }
      /* sum(i=0 to m; h_i * max(1, floor(Z_Depth/2**i))) */
      hz_height_3d_sum += h_i * Z_i;
      H_i = minify(H_i, 1);
      Z_i = minify(Z_i, 1);
   }
   /* HZ_QPitch = h0 + max(h1, sum(i=2 to m; h_i)) */
   buf->qpitch = h0 + MAX2(h1, sum_h_i);

   if (mt->target == GL_TEXTURE_3D) {
      /* (1/2) * sum(i=0 to m; h_i * max(1, floor(Z_Depth/2**i))) */
      hz_height = DIV_ROUND_UP(hz_height_3d_sum, 2);
   } else {
      /* HZ_Height (rows) = ceiling( (HZ_QPitch/2)/8) *8 * Z_Depth */
      hz_height = DIV_ROUND_UP(buf->qpitch, 2 * 8) * 8 * Z0;
      if (mt->target == GL_TEXTURE_CUBE_MAP_ARRAY ||
          mt->target == GL_TEXTURE_CUBE_MAP) {
         /* HZ_Height (rows) = ceiling( (HZ_QPitch/2)/8) *8 * 6 * Z_Depth
          *
          * We can can just take our hz_height calculation from above, and
          * multiply by 6 for the cube map and cube map array types.
          */
         hz_height *= 6;
      }
   }

   unsigned long pitch;
   uint32_t tiling = I915_TILING_Y;
   buf->bo = drm_intel_bo_alloc_tiled(brw->bufmgr, "hiz",
                                      hz_width, hz_height, 1,
                                      &tiling, &pitch,
                                      BO_ALLOC_FOR_RENDER);
   if (!buf->bo) {
      free(buf);
      return NULL;
   } else if (tiling != I915_TILING_Y) {
      drm_intel_bo_unreference(buf->bo);
      free(buf);
      return NULL;
   }

   buf->pitch = pitch;

   return buf;
}


static struct intel_miptree_aux_buffer *
intel_hiz_miptree_buf_create(struct brw_context *brw,
                             struct intel_mipmap_tree *mt)
{
   struct intel_miptree_aux_buffer *buf = calloc(sizeof(*buf), 1);
   uint32_t layout_flags = MIPTREE_LAYOUT_ACCELERATED_UPLOAD;

   if (brw->gen == 6)
      layout_flags |= MIPTREE_LAYOUT_FORCE_ALL_SLICE_AT_LOD;

   if (!buf)
      return NULL;

   layout_flags |= MIPTREE_LAYOUT_TILING_ANY;
   buf->mt = intel_miptree_create(brw,
                                  mt->target,
                                  mt->format,
                                  mt->first_level,
                                  mt->last_level,
                                  mt->logical_width0,
                                  mt->logical_height0,
                                  mt->logical_depth0,
                                  mt->num_samples,
                                  layout_flags);
   if (!buf->mt) {
      free(buf);
      return NULL;
   }

   buf->bo = buf->mt->bo;
   buf->pitch = buf->mt->pitch;
   buf->qpitch = buf->mt->qpitch;

   return buf;
}

bool
intel_miptree_wants_hiz_buffer(struct brw_context *brw,
                               struct intel_mipmap_tree *mt)
{
   if (!brw->has_hiz)
      return false;

   if (mt->hiz_buf != NULL)
      return false;

   if (mt->disable_aux_buffers)
      return false;

   switch (mt->format) {
   case MESA_FORMAT_Z_FLOAT32:
   case MESA_FORMAT_Z32_FLOAT_S8X24_UINT:
   case MESA_FORMAT_Z24_UNORM_X8_UINT:
   case MESA_FORMAT_Z24_UNORM_S8_UINT:
   case MESA_FORMAT_Z_UNORM16:
      return true;
   default:
      return false;
   }
}

bool
intel_miptree_alloc_hiz(struct brw_context *brw,
			struct intel_mipmap_tree *mt)
{
   assert(mt->hiz_buf == NULL);
   assert(!mt->disable_aux_buffers);

   if (brw->gen == 7) {
      mt->hiz_buf = intel_gen7_hiz_buf_create(brw, mt);
   } else if (brw->gen >= 8) {
      mt->hiz_buf = intel_gen8_hiz_buf_create(brw, mt);
   } else {
      mt->hiz_buf = intel_hiz_miptree_buf_create(brw, mt);
   }

   if (!mt->hiz_buf)
      return false;

   /* Mark that all slices need a HiZ resolve. */
   for (unsigned level = mt->first_level; level <= mt->last_level; ++level) {
      if (!intel_miptree_level_enable_hiz(brw, mt, level))
         continue;

      for (unsigned layer = 0; layer < mt->level[level].depth; ++layer) {
         struct intel_resolve_map *m = malloc(sizeof(struct intel_resolve_map));
         exec_node_init(&m->link);
         m->level = level;
         m->layer = layer;
         m->need = GEN6_HIZ_OP_HIZ_RESOLVE;

         exec_list_push_tail(&mt->hiz_map, &m->link);
      }
   }

   return true;
}

/**
 * Does the miptree slice have hiz enabled?
 */
bool
intel_miptree_level_has_hiz(struct intel_mipmap_tree *mt, uint32_t level)
{
   intel_miptree_check_level_layer(mt, level, 0);
   return mt->level[level].has_hiz;
}

void
intel_miptree_slice_set_needs_hiz_resolve(struct intel_mipmap_tree *mt,
					  uint32_t level,
					  uint32_t layer)
{
   if (!intel_miptree_level_has_hiz(mt, level))
      return;

   intel_resolve_map_set(&mt->hiz_map,
			 level, layer, GEN6_HIZ_OP_HIZ_RESOLVE);
}


void
intel_miptree_slice_set_needs_depth_resolve(struct intel_mipmap_tree *mt,
                                            uint32_t level,
                                            uint32_t layer)
{
   if (!intel_miptree_level_has_hiz(mt, level))
      return;

   intel_resolve_map_set(&mt->hiz_map,
			 level, layer, GEN6_HIZ_OP_DEPTH_RESOLVE);
}

void
intel_miptree_set_all_slices_need_depth_resolve(struct intel_mipmap_tree *mt,
                                                uint32_t level)
{
   uint32_t layer;
   uint32_t end_layer = mt->level[level].depth;

   for (layer = 0; layer < end_layer; layer++) {
      intel_miptree_slice_set_needs_depth_resolve(mt, level, layer);
   }
}

static bool
intel_miptree_slice_resolve(struct brw_context *brw,
			    struct intel_mipmap_tree *mt,
			    uint32_t level,
			    uint32_t layer,
			    enum gen6_hiz_op need)
{
   intel_miptree_check_level_layer(mt, level, layer);

   struct intel_resolve_map *item =
	 intel_resolve_map_get(&mt->hiz_map, level, layer);

   if (!item || item->need != need)
      return false;

   intel_hiz_exec(brw, mt, level, layer, need);
   intel_resolve_map_remove(item);
   return true;
}

bool
intel_miptree_slice_resolve_hiz(struct brw_context *brw,
				struct intel_mipmap_tree *mt,
				uint32_t level,
				uint32_t layer)
{
   return intel_miptree_slice_resolve(brw, mt, level, layer,
				      GEN6_HIZ_OP_HIZ_RESOLVE);
}

bool
intel_miptree_slice_resolve_depth(struct brw_context *brw,
				  struct intel_mipmap_tree *mt,
				  uint32_t level,
				  uint32_t layer)
{
   return intel_miptree_slice_resolve(brw, mt, level, layer,
				      GEN6_HIZ_OP_DEPTH_RESOLVE);
}

static bool
intel_miptree_all_slices_resolve(struct brw_context *brw,
				 struct intel_mipmap_tree *mt,
				 enum gen6_hiz_op need)
{
   bool did_resolve = false;

   foreach_list_typed_safe(struct intel_resolve_map, map, link, &mt->hiz_map) {
      if (map->need != need)
	 continue;

      intel_hiz_exec(brw, mt, map->level, map->layer, need);
      intel_resolve_map_remove(map);
      did_resolve = true;
   }

   return did_resolve;
}

bool
intel_miptree_all_slices_resolve_hiz(struct brw_context *brw,
				     struct intel_mipmap_tree *mt)
{
   return intel_miptree_all_slices_resolve(brw, mt,
					   GEN6_HIZ_OP_HIZ_RESOLVE);
}

bool
intel_miptree_all_slices_resolve_depth(struct brw_context *brw,
				       struct intel_mipmap_tree *mt)
{
   return intel_miptree_all_slices_resolve(brw, mt,
					   GEN6_HIZ_OP_DEPTH_RESOLVE);
}


void
intel_miptree_resolve_color(struct brw_context *brw,
                            struct intel_mipmap_tree *mt,
                            int flags)
{
   /* From gen9 onwards there is new compression scheme for single sampled
    * surfaces called "lossless compressed". These don't need to be always
    * resolved.
    */
   if ((flags & INTEL_MIPTREE_IGNORE_CCS_E) &&
       intel_miptree_is_lossless_compressed(brw, mt))
      return;

   switch (mt->fast_clear_state) {
   case INTEL_FAST_CLEAR_STATE_NO_MCS:
   case INTEL_FAST_CLEAR_STATE_RESOLVED:
      /* No resolve needed */
      break;
   case INTEL_FAST_CLEAR_STATE_UNRESOLVED:
   case INTEL_FAST_CLEAR_STATE_CLEAR:
      /* Fast color clear resolves only make sense for non-MSAA buffers. */
      if (mt->msaa_layout == INTEL_MSAA_LAYOUT_NONE ||
          intel_miptree_is_lossless_compressed(brw, mt)) {
         brw_blorp_resolve_color(brw, mt);
      }
      break;
   }
}


/**
 * Make it possible to share the BO backing the given miptree with another
 * process or another miptree.
 *
 * Fast color clears are unsafe with shared buffers, so we need to resolve and
 * then discard the MCS buffer, if present.  We also set the fast_clear_state
 * to INTEL_FAST_CLEAR_STATE_NO_MCS to ensure that no MCS buffer gets
 * allocated in the future.
 */
void
intel_miptree_make_shareable(struct brw_context *brw,
                             struct intel_mipmap_tree *mt)
{
   /* MCS buffers are also used for multisample buffers, but we can't resolve
    * away a multisample MCS buffer because it's an integral part of how the
    * pixel data is stored.  Fortunately this code path should never be
    * reached for multisample buffers.
    */
   assert(mt->msaa_layout == INTEL_MSAA_LAYOUT_NONE);

   if (mt->mcs_mt) {
      intel_miptree_resolve_color(brw, mt, 0);
      intel_miptree_release(&mt->mcs_mt);
      mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_NO_MCS;
   }
}


/**
 * \brief Get pointer offset into stencil buffer.
 *
 * The stencil buffer is W tiled. Since the GTT is incapable of W fencing, we
 * must decode the tile's layout in software.
 *
 * See
 *   - PRM, 2011 Sandy Bridge, Volume 1, Part 2, Section 4.5.2.1 W-Major Tile
 *     Format.
 *   - PRM, 2011 Sandy Bridge, Volume 1, Part 2, Section 4.5.3 Tiling Algorithm
 *
 * Even though the returned offset is always positive, the return type is
 * signed due to
 *    commit e8b1c6d6f55f5be3bef25084fdd8b6127517e137
 *    mesa: Fix return type of  _mesa_get_format_bytes() (#37351)
 */
static intptr_t
intel_offset_S8(uint32_t stride, uint32_t x, uint32_t y, bool swizzled)
{
   uint32_t tile_size = 4096;
   uint32_t tile_width = 64;
   uint32_t tile_height = 64;
   uint32_t row_size = 64 * stride;

   uint32_t tile_x = x / tile_width;
   uint32_t tile_y = y / tile_height;

   /* The byte's address relative to the tile's base addres. */
   uint32_t byte_x = x % tile_width;
   uint32_t byte_y = y % tile_height;

   uintptr_t u = tile_y * row_size
               + tile_x * tile_size
               + 512 * (byte_x / 8)
               +  64 * (byte_y / 8)
               +  32 * ((byte_y / 4) % 2)
               +  16 * ((byte_x / 4) % 2)
               +   8 * ((byte_y / 2) % 2)
               +   4 * ((byte_x / 2) % 2)
               +   2 * (byte_y % 2)
               +   1 * (byte_x % 2);

   if (swizzled) {
      /* adjust for bit6 swizzling */
      if (((byte_x / 8) % 2) == 1) {
	 if (((byte_y / 8) % 2) == 0) {
	    u += 64;
	 } else {
	    u -= 64;
	 }
      }
   }

   return u;
}

void
intel_miptree_updownsample(struct brw_context *brw,
                           struct intel_mipmap_tree *src,
                           struct intel_mipmap_tree *dst)
{
   brw_blorp_blit_miptrees(brw,
                           src, 0 /* level */, 0 /* layer */,
                           src->format, SWIZZLE_XYZW,
                           dst, 0 /* level */, 0 /* layer */, dst->format,
                           0, 0,
                           src->logical_width0, src->logical_height0,
                           0, 0,
                           dst->logical_width0, dst->logical_height0,
                           GL_NEAREST, false, false /*mirror x, y*/,
                           false, false);

   if (src->stencil_mt) {
      brw_blorp_blit_miptrees(brw,
                              src->stencil_mt, 0 /* level */, 0 /* layer */,
                              src->stencil_mt->format, SWIZZLE_XYZW,
                              dst->stencil_mt, 0 /* level */, 0 /* layer */,
                              dst->stencil_mt->format,
                              0, 0,
                              src->logical_width0, src->logical_height0,
                              0, 0,
                              dst->logical_width0, dst->logical_height0,
                              GL_NEAREST, false, false /*mirror x, y*/,
                              false, false /* decode/encode srgb */);
   }
}

static void *
intel_miptree_map_raw(struct brw_context *brw, struct intel_mipmap_tree *mt)
{
   /* CPU accesses to color buffers don't understand fast color clears, so
    * resolve any pending fast color clears before we map.
    */
   intel_miptree_resolve_color(brw, mt, 0);

   drm_intel_bo *bo = mt->bo;

   if (drm_intel_bo_references(brw->batch.bo, bo))
      intel_batchbuffer_flush(brw);

   if (mt->tiling != I915_TILING_NONE)
      brw_bo_map_gtt(brw, bo, "miptree");
   else
      brw_bo_map(brw, bo, true, "miptree");

   return bo->virtual;
}

static void
intel_miptree_unmap_raw(struct intel_mipmap_tree *mt)
{
   drm_intel_bo_unmap(mt->bo);
}

static void
intel_miptree_map_gtt(struct brw_context *brw,
		      struct intel_mipmap_tree *mt,
		      struct intel_miptree_map *map,
		      unsigned int level, unsigned int slice)
{
   unsigned int bw, bh;
   void *base;
   unsigned int image_x, image_y;
   intptr_t x = map->x;
   intptr_t y = map->y;

   /* For compressed formats, the stride is the number of bytes per
    * row of blocks.  intel_miptree_get_image_offset() already does
    * the divide.
    */
   _mesa_get_format_block_size(mt->format, &bw, &bh);
   assert(y % bh == 0);
   assert(x % bw == 0);
   y /= bh;
   x /= bw;

   base = intel_miptree_map_raw(brw, mt) + mt->offset;

   if (base == NULL)
      map->ptr = NULL;
   else {
      /* Note that in the case of cube maps, the caller must have passed the
       * slice number referencing the face.
      */
      intel_miptree_get_image_offset(mt, level, slice, &image_x, &image_y);
      x += image_x;
      y += image_y;

      map->stride = mt->pitch;
      map->ptr = base + y * map->stride + x * mt->cpp;
   }

   DBG("%s: %d,%d %dx%d from mt %p (%s) "
       "%"PRIiPTR",%"PRIiPTR" = %p/%d\n", __func__,
       map->x, map->y, map->w, map->h,
       mt, _mesa_get_format_name(mt->format),
       x, y, map->ptr, map->stride);
}

static void
intel_miptree_unmap_gtt(struct intel_mipmap_tree *mt)
{
   intel_miptree_unmap_raw(mt);
}

static void
intel_miptree_map_blit(struct brw_context *brw,
		       struct intel_mipmap_tree *mt,
		       struct intel_miptree_map *map,
		       unsigned int level, unsigned int slice)
{
   map->linear_mt = intel_miptree_create(brw, GL_TEXTURE_2D, mt->format,
                                         /* first_level */ 0,
                                         /* last_level */ 0,
                                         map->w, map->h, 1,
                                         /* samples */ 0,
                                         MIPTREE_LAYOUT_TILING_NONE);

   if (!map->linear_mt) {
      fprintf(stderr, "Failed to allocate blit temporary\n");
      goto fail;
   }
   map->stride = map->linear_mt->pitch;

   /* One of either READ_BIT or WRITE_BIT or both is set.  READ_BIT implies no
    * INVALIDATE_RANGE_BIT.  WRITE_BIT needs the original values read in unless
    * invalidate is set, since we'll be writing the whole rectangle from our
    * temporary buffer back out.
    */
   if (!(map->mode & GL_MAP_INVALIDATE_RANGE_BIT)) {
      if (!intel_miptree_blit(brw,
                              mt, level, slice,
                              map->x, map->y, false,
                              map->linear_mt, 0, 0,
                              0, 0, false,
                              map->w, map->h, GL_COPY)) {
         fprintf(stderr, "Failed to blit\n");
         goto fail;
      }
   }

   map->ptr = intel_miptree_map_raw(brw, map->linear_mt);

   DBG("%s: %d,%d %dx%d from mt %p (%s) %d,%d = %p/%d\n", __func__,
       map->x, map->y, map->w, map->h,
       mt, _mesa_get_format_name(mt->format),
       level, slice, map->ptr, map->stride);

   return;

fail:
   intel_miptree_release(&map->linear_mt);
   map->ptr = NULL;
   map->stride = 0;
}

static void
intel_miptree_unmap_blit(struct brw_context *brw,
			 struct intel_mipmap_tree *mt,
			 struct intel_miptree_map *map,
			 unsigned int level,
			 unsigned int slice)
{
   struct gl_context *ctx = &brw->ctx;

   intel_miptree_unmap_raw(map->linear_mt);

   if (map->mode & GL_MAP_WRITE_BIT) {
      bool ok = intel_miptree_blit(brw,
                                   map->linear_mt, 0, 0,
                                   0, 0, false,
                                   mt, level, slice,
                                   map->x, map->y, false,
                                   map->w, map->h, GL_COPY);
      WARN_ONCE(!ok, "Failed to blit from linear temporary mapping");
   }

   intel_miptree_release(&map->linear_mt);
}

/**
 * "Map" a buffer by copying it to an untiled temporary using MOVNTDQA.
 */
#if defined(USE_SSE41)
static void
intel_miptree_map_movntdqa(struct brw_context *brw,
                           struct intel_mipmap_tree *mt,
                           struct intel_miptree_map *map,
                           unsigned int level, unsigned int slice)
{
   assert(map->mode & GL_MAP_READ_BIT);
   assert(!(map->mode & GL_MAP_WRITE_BIT));

   DBG("%s: %d,%d %dx%d from mt %p (%s) %d,%d = %p/%d\n", __func__,
       map->x, map->y, map->w, map->h,
       mt, _mesa_get_format_name(mt->format),
       level, slice, map->ptr, map->stride);

   /* Map the original image */
   uint32_t image_x;
   uint32_t image_y;
   intel_miptree_get_image_offset(mt, level, slice, &image_x, &image_y);
   image_x += map->x;
   image_y += map->y;

   void *src = intel_miptree_map_raw(brw, mt);
   if (!src)
      return;
   src += image_y * mt->pitch;
   src += image_x * mt->cpp;

   /* Due to the pixel offsets for the particular image being mapped, our
    * src pointer may not be 16-byte aligned.  However, if the pitch is
    * divisible by 16, then the amount by which it's misaligned will remain
    * consistent from row to row.
    */
   assert((mt->pitch % 16) == 0);
   const int misalignment = ((uintptr_t) src) & 15;

   /* Create an untiled temporary buffer for the mapping. */
   const unsigned width_bytes = _mesa_format_row_stride(mt->format, map->w);

   map->stride = ALIGN(misalignment + width_bytes, 16);

   map->buffer = _mesa_align_malloc(map->stride * map->h, 16);
   /* Offset the destination so it has the same misalignment as src. */
   map->ptr = map->buffer + misalignment;

   assert((((uintptr_t) map->ptr) & 15) == misalignment);

   for (uint32_t y = 0; y < map->h; y++) {
      void *dst_ptr = map->ptr + y * map->stride;
      void *src_ptr = src + y * mt->pitch;

      _mesa_streaming_load_memcpy(dst_ptr, src_ptr, width_bytes);
   }

   intel_miptree_unmap_raw(mt);
}

static void
intel_miptree_unmap_movntdqa(struct brw_context *brw,
                             struct intel_mipmap_tree *mt,
                             struct intel_miptree_map *map,
                             unsigned int level,
                             unsigned int slice)
{
   _mesa_align_free(map->buffer);
   map->buffer = NULL;
   map->ptr = NULL;
}
#endif

static void
intel_miptree_map_s8(struct brw_context *brw,
		     struct intel_mipmap_tree *mt,
		     struct intel_miptree_map *map,
		     unsigned int level, unsigned int slice)
{
   map->stride = map->w;
   map->buffer = map->ptr = malloc(map->stride * map->h);
   if (!map->buffer)
      return;

   /* One of either READ_BIT or WRITE_BIT or both is set.  READ_BIT implies no
    * INVALIDATE_RANGE_BIT.  WRITE_BIT needs the original values read in unless
    * invalidate is set, since we'll be writing the whole rectangle from our
    * temporary buffer back out.
    */
   if (!(map->mode & GL_MAP_INVALIDATE_RANGE_BIT)) {
      uint8_t *untiled_s8_map = map->ptr;
      uint8_t *tiled_s8_map = intel_miptree_map_raw(brw, mt);
      unsigned int image_x, image_y;

      intel_miptree_get_image_offset(mt, level, slice, &image_x, &image_y);

      for (uint32_t y = 0; y < map->h; y++) {
	 for (uint32_t x = 0; x < map->w; x++) {
	    ptrdiff_t offset = intel_offset_S8(mt->pitch,
	                                       x + image_x + map->x,
	                                       y + image_y + map->y,
					       brw->has_swizzling);
	    untiled_s8_map[y * map->w + x] = tiled_s8_map[offset];
	 }
      }

      intel_miptree_unmap_raw(mt);

      DBG("%s: %d,%d %dx%d from mt %p %d,%d = %p/%d\n", __func__,
	  map->x, map->y, map->w, map->h,
	  mt, map->x + image_x, map->y + image_y, map->ptr, map->stride);
   } else {
      DBG("%s: %d,%d %dx%d from mt %p = %p/%d\n", __func__,
	  map->x, map->y, map->w, map->h,
	  mt, map->ptr, map->stride);
   }
}

static void
intel_miptree_unmap_s8(struct brw_context *brw,
		       struct intel_mipmap_tree *mt,
		       struct intel_miptree_map *map,
		       unsigned int level,
		       unsigned int slice)
{
   if (map->mode & GL_MAP_WRITE_BIT) {
      unsigned int image_x, image_y;
      uint8_t *untiled_s8_map = map->ptr;
      uint8_t *tiled_s8_map = intel_miptree_map_raw(brw, mt);

      intel_miptree_get_image_offset(mt, level, slice, &image_x, &image_y);

      for (uint32_t y = 0; y < map->h; y++) {
	 for (uint32_t x = 0; x < map->w; x++) {
	    ptrdiff_t offset = intel_offset_S8(mt->pitch,
	                                       image_x + x + map->x,
	                                       image_y + y + map->y,
					       brw->has_swizzling);
	    tiled_s8_map[offset] = untiled_s8_map[y * map->w + x];
	 }
      }

      intel_miptree_unmap_raw(mt);
   }

   free(map->buffer);
}

static void
intel_miptree_map_etc(struct brw_context *brw,
                      struct intel_mipmap_tree *mt,
                      struct intel_miptree_map *map,
                      unsigned int level,
                      unsigned int slice)
{
   assert(mt->etc_format != MESA_FORMAT_NONE);
   if (mt->etc_format == MESA_FORMAT_ETC1_RGB8) {
      assert(mt->format == MESA_FORMAT_R8G8B8X8_UNORM);
   }

   assert(map->mode & GL_MAP_WRITE_BIT);
   assert(map->mode & GL_MAP_INVALIDATE_RANGE_BIT);

   map->stride = _mesa_format_row_stride(mt->etc_format, map->w);
   map->buffer = malloc(_mesa_format_image_size(mt->etc_format,
                                                map->w, map->h, 1));
   map->ptr = map->buffer;
}

static void
intel_miptree_unmap_etc(struct brw_context *brw,
                        struct intel_mipmap_tree *mt,
                        struct intel_miptree_map *map,
                        unsigned int level,
                        unsigned int slice)
{
   uint32_t image_x;
   uint32_t image_y;
   intel_miptree_get_image_offset(mt, level, slice, &image_x, &image_y);

   image_x += map->x;
   image_y += map->y;

   uint8_t *dst = intel_miptree_map_raw(brw, mt)
                + image_y * mt->pitch
                + image_x * mt->cpp;

   if (mt->etc_format == MESA_FORMAT_ETC1_RGB8)
      _mesa_etc1_unpack_rgba8888(dst, mt->pitch,
                                 map->ptr, map->stride,
                                 map->w, map->h);
   else
      _mesa_unpack_etc2_format(dst, mt->pitch,
                               map->ptr, map->stride,
                               map->w, map->h, mt->etc_format);

   intel_miptree_unmap_raw(mt);
   free(map->buffer);
}

/**
 * Mapping function for packed depth/stencil miptrees backed by real separate
 * miptrees for depth and stencil.
 *
 * On gen7, and to support HiZ pre-gen7, we have to have the stencil buffer
 * separate from the depth buffer.  Yet at the GL API level, we have to expose
 * packed depth/stencil textures and FBO attachments, and Mesa core expects to
 * be able to map that memory for texture storage and glReadPixels-type
 * operations.  We give Mesa core that access by mallocing a temporary and
 * copying the data between the actual backing store and the temporary.
 */
static void
intel_miptree_map_depthstencil(struct brw_context *brw,
			       struct intel_mipmap_tree *mt,
			       struct intel_miptree_map *map,
			       unsigned int level, unsigned int slice)
{
   struct intel_mipmap_tree *z_mt = mt;
   struct intel_mipmap_tree *s_mt = mt->stencil_mt;
   bool map_z32f_x24s8 = mt->format == MESA_FORMAT_Z_FLOAT32;
   int packed_bpp = map_z32f_x24s8 ? 8 : 4;

   map->stride = map->w * packed_bpp;
   map->buffer = map->ptr = malloc(map->stride * map->h);
   if (!map->buffer)
      return;

   /* One of either READ_BIT or WRITE_BIT or both is set.  READ_BIT implies no
    * INVALIDATE_RANGE_BIT.  WRITE_BIT needs the original values read in unless
    * invalidate is set, since we'll be writing the whole rectangle from our
    * temporary buffer back out.
    */
   if (!(map->mode & GL_MAP_INVALIDATE_RANGE_BIT)) {
      uint32_t *packed_map = map->ptr;
      uint8_t *s_map = intel_miptree_map_raw(brw, s_mt);
      uint32_t *z_map = intel_miptree_map_raw(brw, z_mt);
      unsigned int s_image_x, s_image_y;
      unsigned int z_image_x, z_image_y;

      intel_miptree_get_image_offset(s_mt, level, slice,
				     &s_image_x, &s_image_y);
      intel_miptree_get_image_offset(z_mt, level, slice,
				     &z_image_x, &z_image_y);

      for (uint32_t y = 0; y < map->h; y++) {
	 for (uint32_t x = 0; x < map->w; x++) {
	    int map_x = map->x + x, map_y = map->y + y;
	    ptrdiff_t s_offset = intel_offset_S8(s_mt->pitch,
						 map_x + s_image_x,
						 map_y + s_image_y,
						 brw->has_swizzling);
	    ptrdiff_t z_offset = ((map_y + z_image_y) *
                                  (z_mt->pitch / 4) +
				  (map_x + z_image_x));
	    uint8_t s = s_map[s_offset];
	    uint32_t z = z_map[z_offset];

	    if (map_z32f_x24s8) {
	       packed_map[(y * map->w + x) * 2 + 0] = z;
	       packed_map[(y * map->w + x) * 2 + 1] = s;
	    } else {
	       packed_map[y * map->w + x] = (s << 24) | (z & 0x00ffffff);
	    }
	 }
      }

      intel_miptree_unmap_raw(s_mt);
      intel_miptree_unmap_raw(z_mt);

      DBG("%s: %d,%d %dx%d from z mt %p %d,%d, s mt %p %d,%d = %p/%d\n",
	  __func__,
	  map->x, map->y, map->w, map->h,
	  z_mt, map->x + z_image_x, map->y + z_image_y,
	  s_mt, map->x + s_image_x, map->y + s_image_y,
	  map->ptr, map->stride);
   } else {
      DBG("%s: %d,%d %dx%d from mt %p = %p/%d\n", __func__,
	  map->x, map->y, map->w, map->h,
	  mt, map->ptr, map->stride);
   }
}

static void
intel_miptree_unmap_depthstencil(struct brw_context *brw,
				 struct intel_mipmap_tree *mt,
				 struct intel_miptree_map *map,
				 unsigned int level,
				 unsigned int slice)
{
   struct intel_mipmap_tree *z_mt = mt;
   struct intel_mipmap_tree *s_mt = mt->stencil_mt;
   bool map_z32f_x24s8 = mt->format == MESA_FORMAT_Z_FLOAT32;

   if (map->mode & GL_MAP_WRITE_BIT) {
      uint32_t *packed_map = map->ptr;
      uint8_t *s_map = intel_miptree_map_raw(brw, s_mt);
      uint32_t *z_map = intel_miptree_map_raw(brw, z_mt);
      unsigned int s_image_x, s_image_y;
      unsigned int z_image_x, z_image_y;

      intel_miptree_get_image_offset(s_mt, level, slice,
				     &s_image_x, &s_image_y);
      intel_miptree_get_image_offset(z_mt, level, slice,
				     &z_image_x, &z_image_y);

      for (uint32_t y = 0; y < map->h; y++) {
	 for (uint32_t x = 0; x < map->w; x++) {
	    ptrdiff_t s_offset = intel_offset_S8(s_mt->pitch,
						 x + s_image_x + map->x,
						 y + s_image_y + map->y,
						 brw->has_swizzling);
	    ptrdiff_t z_offset = ((y + z_image_y + map->y) *
                                  (z_mt->pitch / 4) +
				  (x + z_image_x + map->x));

	    if (map_z32f_x24s8) {
	       z_map[z_offset] = packed_map[(y * map->w + x) * 2 + 0];
	       s_map[s_offset] = packed_map[(y * map->w + x) * 2 + 1];
	    } else {
	       uint32_t packed = packed_map[y * map->w + x];
	       s_map[s_offset] = packed >> 24;
	       z_map[z_offset] = packed;
	    }
	 }
      }

      intel_miptree_unmap_raw(s_mt);
      intel_miptree_unmap_raw(z_mt);

      DBG("%s: %d,%d %dx%d from z mt %p (%s) %d,%d, s mt %p %d,%d = %p/%d\n",
	  __func__,
	  map->x, map->y, map->w, map->h,
	  z_mt, _mesa_get_format_name(z_mt->format),
	  map->x + z_image_x, map->y + z_image_y,
	  s_mt, map->x + s_image_x, map->y + s_image_y,
	  map->ptr, map->stride);
   }

   free(map->buffer);
}

/**
 * Create and attach a map to the miptree at (level, slice). Return the
 * attached map.
 */
static struct intel_miptree_map*
intel_miptree_attach_map(struct intel_mipmap_tree *mt,
                         unsigned int level,
                         unsigned int slice,
                         unsigned int x,
                         unsigned int y,
                         unsigned int w,
                         unsigned int h,
                         GLbitfield mode)
{
   struct intel_miptree_map *map = calloc(1, sizeof(*map));

   if (!map)
      return NULL;

   assert(mt->level[level].slice[slice].map == NULL);
   mt->level[level].slice[slice].map = map;

   map->mode = mode;
   map->x = x;
   map->y = y;
   map->w = w;
   map->h = h;

   return map;
}

/**
 * Release the map at (level, slice).
 */
static void
intel_miptree_release_map(struct intel_mipmap_tree *mt,
                         unsigned int level,
                         unsigned int slice)
{
   struct intel_miptree_map **map;

   map = &mt->level[level].slice[slice].map;
   free(*map);
   *map = NULL;
}

static bool
can_blit_slice(struct intel_mipmap_tree *mt,
               unsigned int level, unsigned int slice)
{
   uint32_t image_x;
   uint32_t image_y;
   intel_miptree_get_image_offset(mt, level, slice, &image_x, &image_y);
   if (image_x >= 32768 || image_y >= 32768)
      return false;

   /* See intel_miptree_blit() for details on the 32k pitch limit. */
   if (mt->pitch >= 32768)
      return false;

   return true;
}

static bool
use_intel_mipree_map_blit(struct brw_context *brw,
                          struct intel_mipmap_tree *mt,
                          GLbitfield mode,
                          unsigned int level,
                          unsigned int slice)
{
   if (brw->has_llc &&
      /* It's probably not worth swapping to the blit ring because of
       * all the overhead involved. But, we must use blitter for the
       * surfaces with INTEL_MIPTREE_TRMODE_{YF,YS}.
       */
       (!(mode & GL_MAP_WRITE_BIT) ||
        mt->tr_mode != INTEL_MIPTREE_TRMODE_NONE) &&
       !mt->compressed &&
       (mt->tiling == I915_TILING_X ||
        /* Prior to Sandybridge, the blitter can't handle Y tiling */
        (brw->gen >= 6 && mt->tiling == I915_TILING_Y) ||
        /* Fast copy blit on skl+ supports all tiling formats. */
        brw->gen >= 9) &&
       can_blit_slice(mt, level, slice))
      return true;

   if (mt->tiling != I915_TILING_NONE &&
       mt->bo->size >= brw->max_gtt_map_object_size) {
      assert(can_blit_slice(mt, level, slice));
      return true;
   }

   return false;
}

/**
 * Parameter \a out_stride has type ptrdiff_t not because the buffer stride may
 * exceed 32 bits but to diminish the likelihood subtle bugs in pointer
 * arithmetic overflow.
 *
 * If you call this function and use \a out_stride, then you're doing pointer
 * arithmetic on \a out_ptr. The type of \a out_stride doesn't prevent all
 * bugs.  The caller must still take care to avoid 32-bit overflow errors in
 * all arithmetic expressions that contain buffer offsets and pixel sizes,
 * which usually have type uint32_t or GLuint.
 */
void
intel_miptree_map(struct brw_context *brw,
                  struct intel_mipmap_tree *mt,
                  unsigned int level,
                  unsigned int slice,
                  unsigned int x,
                  unsigned int y,
                  unsigned int w,
                  unsigned int h,
                  GLbitfield mode,
                  void **out_ptr,
                  ptrdiff_t *out_stride)
{
   struct intel_miptree_map *map;

   assert(mt->num_samples <= 1);

   map = intel_miptree_attach_map(mt, level, slice, x, y, w, h, mode);
   if (!map){
      *out_ptr = NULL;
      *out_stride = 0;
      return;
   }

   intel_miptree_slice_resolve_depth(brw, mt, level, slice);
   if (map->mode & GL_MAP_WRITE_BIT) {
      intel_miptree_slice_set_needs_hiz_resolve(mt, level, slice);
   }

   if (mt->format == MESA_FORMAT_S_UINT8) {
      intel_miptree_map_s8(brw, mt, map, level, slice);
   } else if (mt->etc_format != MESA_FORMAT_NONE &&
              !(mode & BRW_MAP_DIRECT_BIT)) {
      intel_miptree_map_etc(brw, mt, map, level, slice);
   } else if (mt->stencil_mt && !(mode & BRW_MAP_DIRECT_BIT)) {
      intel_miptree_map_depthstencil(brw, mt, map, level, slice);
   } else if (use_intel_mipree_map_blit(brw, mt, mode, level, slice)) {
      intel_miptree_map_blit(brw, mt, map, level, slice);
#if defined(USE_SSE41)
   } else if (!(mode & GL_MAP_WRITE_BIT) &&
              !mt->compressed && cpu_has_sse4_1 &&
              (mt->pitch % 16 == 0)) {
      intel_miptree_map_movntdqa(brw, mt, map, level, slice);
#endif
   } else {
      /* intel_miptree_map_gtt() doesn't support surfaces with Yf/Ys tiling. */
      assert(mt->tr_mode == INTEL_MIPTREE_TRMODE_NONE);
      intel_miptree_map_gtt(brw, mt, map, level, slice);
   }

   *out_ptr = map->ptr;
   *out_stride = map->stride;

   if (map->ptr == NULL)
      intel_miptree_release_map(mt, level, slice);
}

void
intel_miptree_unmap(struct brw_context *brw,
                    struct intel_mipmap_tree *mt,
                    unsigned int level,
                    unsigned int slice)
{
   struct intel_miptree_map *map = mt->level[level].slice[slice].map;

   assert(mt->num_samples <= 1);

   if (!map)
      return;

   DBG("%s: mt %p (%s) level %d slice %d\n", __func__,
       mt, _mesa_get_format_name(mt->format), level, slice);

   if (mt->format == MESA_FORMAT_S_UINT8) {
      intel_miptree_unmap_s8(brw, mt, map, level, slice);
   } else if (mt->etc_format != MESA_FORMAT_NONE &&
              !(map->mode & BRW_MAP_DIRECT_BIT)) {
      intel_miptree_unmap_etc(brw, mt, map, level, slice);
   } else if (mt->stencil_mt && !(map->mode & BRW_MAP_DIRECT_BIT)) {
      intel_miptree_unmap_depthstencil(brw, mt, map, level, slice);
   } else if (map->linear_mt) {
      intel_miptree_unmap_blit(brw, mt, map, level, slice);
#if defined(USE_SSE41)
   } else if (map->buffer && cpu_has_sse4_1) {
      intel_miptree_unmap_movntdqa(brw, mt, map, level, slice);
#endif
   } else {
      intel_miptree_unmap_gtt(mt);
   }

   intel_miptree_release_map(mt, level, slice);
}

void
intel_miptree_get_isl_surf(struct brw_context *brw,
                           const struct intel_mipmap_tree *mt,
                           struct isl_surf *surf)
{
   switch (mt->target) {
   case GL_TEXTURE_1D:
   case GL_TEXTURE_1D_ARRAY: {
      surf->dim = ISL_SURF_DIM_1D;
      if (brw->gen >= 9 && mt->tiling == I915_TILING_NONE)
         surf->dim_layout = ISL_DIM_LAYOUT_GEN9_1D;
      else
         surf->dim_layout = ISL_DIM_LAYOUT_GEN4_2D;
      break;
   }
   case GL_TEXTURE_2D:
   case GL_TEXTURE_2D_ARRAY:
   case GL_TEXTURE_RECTANGLE:
   case GL_TEXTURE_CUBE_MAP:
   case GL_TEXTURE_CUBE_MAP_ARRAY:
   case GL_TEXTURE_2D_MULTISAMPLE:
   case GL_TEXTURE_2D_MULTISAMPLE_ARRAY:
   case GL_TEXTURE_EXTERNAL_OES:
      surf->dim = ISL_SURF_DIM_2D;
      surf->dim_layout = ISL_DIM_LAYOUT_GEN4_2D;
      break;
   case GL_TEXTURE_3D:
      surf->dim = ISL_SURF_DIM_3D;
      if (brw->gen >= 9)
         surf->dim_layout = ISL_DIM_LAYOUT_GEN4_2D;
      else
         surf->dim_layout = ISL_DIM_LAYOUT_GEN4_3D;
      break;
   default:
      unreachable("Invalid texture target");
   }

   if (mt->num_samples > 1) {
      switch (mt->msaa_layout) {
      case INTEL_MSAA_LAYOUT_IMS:
         surf->msaa_layout = ISL_MSAA_LAYOUT_INTERLEAVED;
         break;
      case INTEL_MSAA_LAYOUT_UMS:
      case INTEL_MSAA_LAYOUT_CMS:
         surf->msaa_layout = ISL_MSAA_LAYOUT_ARRAY;
         break;
      default:
         unreachable("Invalid MSAA layout");
      }
   } else {
      surf->msaa_layout = ISL_MSAA_LAYOUT_NONE;
   }

   if (mt->format == MESA_FORMAT_S_UINT8) {
      surf->tiling = ISL_TILING_W;
      /* The ISL definition of row_pitch matches the surface state pitch field
       * a bit better than intel_mipmap_tree.  In particular, ISL incorporates
       * the factor of 2 for W-tiling in row_pitch.
       */
      surf->row_pitch = 2 * mt->pitch;
   } else {
      switch (mt->tiling) {
      case I915_TILING_NONE:
         surf->tiling = ISL_TILING_LINEAR;
         break;
      case I915_TILING_X:
         surf->tiling = ISL_TILING_X;
         break;
      case I915_TILING_Y:
         switch (mt->tr_mode) {
         case INTEL_MIPTREE_TRMODE_NONE:
            surf->tiling = ISL_TILING_Y0;
            break;
         case INTEL_MIPTREE_TRMODE_YF:
            surf->tiling = ISL_TILING_Yf;
            break;
         case INTEL_MIPTREE_TRMODE_YS:
            surf->tiling = ISL_TILING_Ys;
            break;
         }
         break;
      default:
         unreachable("Invalid tiling mode");
      }

      surf->row_pitch = mt->pitch;
   }

   surf->format = translate_tex_format(brw, mt->format, false);

   if (brw->gen >= 9) {
      if (surf->dim == ISL_SURF_DIM_1D && surf->tiling == ISL_TILING_LINEAR) {
         /* For gen9 1-D surfaces, intel_mipmap_tree has a bogus alignment. */
         surf->image_alignment_el = isl_extent3d(64, 1, 1);
      } else {
         /* On gen9+, intel_mipmap_tree stores the horizontal and vertical
          * alignment in terms of surface elements like we want.
          */
         surf->image_alignment_el = isl_extent3d(mt->halign, mt->valign, 1);
      }
   } else {
      /* On earlier gens it's stored in pixels. */
      unsigned bw, bh;
      _mesa_get_format_block_size(mt->format, &bw, &bh);
      surf->image_alignment_el =
         isl_extent3d(mt->halign / bw, mt->valign / bh, 1);
   }

   surf->logical_level0_px.width = mt->logical_width0;
   surf->logical_level0_px.height = mt->logical_height0;
   if (surf->dim == ISL_SURF_DIM_3D) {
      surf->logical_level0_px.depth = mt->logical_depth0;
      surf->logical_level0_px.array_len = 1;
   } else if (mt->target == GL_TEXTURE_CUBE_MAP ||
              mt->target == GL_TEXTURE_CUBE_MAP_ARRAY) {
      /* For cube maps, mt->logical_depth0 is in number of cubes */
      surf->logical_level0_px.depth = 1;
      surf->logical_level0_px.array_len = mt->logical_depth0 * 6;
   } else {
      surf->logical_level0_px.depth = 1;
      surf->logical_level0_px.array_len = mt->logical_depth0;
   }

   surf->phys_level0_sa.width = mt->physical_width0;
   surf->phys_level0_sa.height = mt->physical_height0;
   if (surf->dim == ISL_SURF_DIM_3D) {
      surf->phys_level0_sa.depth = mt->physical_depth0;
      surf->phys_level0_sa.array_len = 1;
   } else {
      surf->phys_level0_sa.depth = 1;
      surf->phys_level0_sa.array_len = mt->physical_depth0;
   }

   surf->levels = mt->last_level + 1;
   surf->samples = MAX2(mt->num_samples, 1);

   surf->size = 0; /* TODO */
   surf->alignment = 0; /* TODO */

   switch (surf->dim_layout) {
   case ISL_DIM_LAYOUT_GEN4_2D:
   case ISL_DIM_LAYOUT_GEN4_3D:
      if (brw->gen >= 9) {
         surf->array_pitch_el_rows = mt->qpitch;
      } else {
         unsigned bw, bh;
         _mesa_get_format_block_size(mt->format, &bw, &bh);
         assert(mt->qpitch % bh == 0);
         surf->array_pitch_el_rows = mt->qpitch / bh;
      }
      break;
   case ISL_DIM_LAYOUT_GEN9_1D:
      surf->array_pitch_el_rows = 1;
      break;
   }

   switch (mt->array_layout) {
   case ALL_LOD_IN_EACH_SLICE:
      surf->array_pitch_span = ISL_ARRAY_PITCH_SPAN_FULL;
      break;
   case ALL_SLICES_AT_EACH_LOD:
      surf->array_pitch_span = ISL_ARRAY_PITCH_SPAN_COMPACT;
      break;
   default:
      unreachable("Invalid array layout");
   }

   surf->usage = 0; /* TODO */
}

/* WARNING: THE SURFACE CREATED BY THIS FUNCTION IS NOT COMPLETE AND CANNOT BE
 * USED FOR ANY REAL CALCULATIONS.  THE ONLY VALID USE OF SUCH A SURFACE IS TO
 * PASS IT INTO isl_surf_fill_state.
 */
void
intel_miptree_get_aux_isl_surf(struct brw_context *brw,
                               const struct intel_mipmap_tree *mt,
                               struct isl_surf *surf,
                               enum isl_aux_usage *usage)
{
   /* Much is the same as the regular surface */
   intel_miptree_get_isl_surf(brw, mt->mcs_mt, surf);

   /* Figure out the layout */
   if (_mesa_get_format_base_format(mt->format) == GL_DEPTH_COMPONENT) {
      *usage = ISL_AUX_USAGE_HIZ;
   } else if (mt->num_samples > 1) {
      assert(mt->msaa_layout == INTEL_MSAA_LAYOUT_CMS);
      *usage = ISL_AUX_USAGE_MCS;
   } else if (intel_miptree_is_lossless_compressed(brw, mt)) {
      assert(brw->gen >= 9);
      *usage = ISL_AUX_USAGE_CCS_E;
   } else if (mt->fast_clear_state != INTEL_FAST_CLEAR_STATE_NO_MCS) {
      *usage = ISL_AUX_USAGE_CCS_D;
   } else {
      unreachable("Invalid MCS miptree");
   }

   /* Figure out the format and tiling of the auxiliary surface */
   switch (*usage) {
   case ISL_AUX_USAGE_NONE:
      unreachable("Invalid MCS miptree");

   case ISL_AUX_USAGE_HIZ:
      surf->format = ISL_FORMAT_HIZ;
      surf->tiling = ISL_TILING_HIZ;
      surf->usage = ISL_SURF_USAGE_HIZ_BIT;
      break;

   case ISL_AUX_USAGE_MCS:
      /*
       * From the SKL PRM:
       *    "When Auxiliary Surface Mode is set to AUX_CCS_D or AUX_CCS_E,
       *    HALIGN 16 must be used."
       */
      if (brw->gen >= 9)
         assert(mt->halign == 16);

      surf->usage = ISL_SURF_USAGE_MCS_BIT;

      switch (mt->num_samples) {
      case 2:  surf->format = ISL_FORMAT_MCS_2X;   break;
      case 4:  surf->format = ISL_FORMAT_MCS_4X;   break;
      case 8:  surf->format = ISL_FORMAT_MCS_8X;   break;
      case 16: surf->format = ISL_FORMAT_MCS_16X;  break;
      default:
         unreachable("Invalid number of samples");
      }
      break;

   case ISL_AUX_USAGE_CCS_D:
   case ISL_AUX_USAGE_CCS_E:
      /*
       * From the BDW PRM, Volume 2d, page 260 (RENDER_SURFACE_STATE):
       *
       *    "When MCS is enabled for non-MSRT, HALIGN_16 must be used"
       *
       * From the hardware spec for GEN9:
       *
       *    "When Auxiliary Surface Mode is set to AUX_CCS_D or AUX_CCS_E,
       *    HALIGN 16 must be used."
       */
      assert(mt->num_samples <= 1);
      if (brw->gen >= 8)
         assert(mt->halign == 16);

      surf->tiling = ISL_TILING_CCS;
      surf->usage = ISL_SURF_USAGE_CCS_BIT;

      if (brw->gen >= 9) {
         assert(mt->tiling == I915_TILING_Y);
         switch (_mesa_get_format_bytes(mt->format)) {
         case 4:  surf->format = ISL_FORMAT_GEN9_CCS_32BPP;   break;
         case 8:  surf->format = ISL_FORMAT_GEN9_CCS_64BPP;   break;
         case 16: surf->format = ISL_FORMAT_GEN9_CCS_128BPP;  break;
         default:
            unreachable("Invalid format size for color compression");
         }
      } else if (mt->tiling == I915_TILING_Y) {
         switch (_mesa_get_format_bytes(mt->format)) {
         case 4:  surf->format = ISL_FORMAT_GEN7_CCS_32BPP_Y;    break;
         case 8:  surf->format = ISL_FORMAT_GEN7_CCS_64BPP_Y;    break;
         case 16: surf->format = ISL_FORMAT_GEN7_CCS_128BPP_Y;   break;
         default:
            unreachable("Invalid format size for color compression");
         }
      } else {
         assert(mt->tiling == I915_TILING_X);
         switch (_mesa_get_format_bytes(mt->format)) {
         case 4:  surf->format = ISL_FORMAT_GEN7_CCS_32BPP_X;    break;
         case 8:  surf->format = ISL_FORMAT_GEN7_CCS_64BPP_X;    break;
         case 16: surf->format = ISL_FORMAT_GEN7_CCS_128BPP_X;   break;
         default:
            unreachable("Invalid format size for color compression");
         }
      }
      break;
   }

   /* Auxiliary surfaces in ISL have compressed formats and array_pitch_el_rows
    * is in elements.  This doesn't match intel_mipmap_tree::qpitch which is
    * in elements of the primary color surface so we have to divide by the
    * compression block height.
    */
   surf->array_pitch_el_rows = mt->qpitch / isl_format_get_layout(surf->format)->bh;
}

union isl_color_value
intel_miptree_get_isl_clear_color(struct brw_context *brw,
                                  const struct intel_mipmap_tree *mt)
{
   union isl_color_value clear_color;

   if (brw->gen >= 9) {
      clear_color.i32[0] = mt->gen9_fast_clear_color.i[0];
      clear_color.i32[1] = mt->gen9_fast_clear_color.i[1];
      clear_color.i32[2] = mt->gen9_fast_clear_color.i[2];
      clear_color.i32[3] = mt->gen9_fast_clear_color.i[3];
   } else if (_mesa_is_format_integer(mt->format)) {
      clear_color.i32[0] = (mt->fast_clear_color_value & (1u << 31)) != 0;
      clear_color.i32[1] = (mt->fast_clear_color_value & (1u << 30)) != 0;
      clear_color.i32[2] = (mt->fast_clear_color_value & (1u << 29)) != 0;
      clear_color.i32[3] = (mt->fast_clear_color_value & (1u << 28)) != 0;
   } else {
      clear_color.f32[0] = (mt->fast_clear_color_value & (1u << 31)) != 0;
      clear_color.f32[1] = (mt->fast_clear_color_value & (1u << 30)) != 0;
      clear_color.f32[2] = (mt->fast_clear_color_value & (1u << 29)) != 0;
      clear_color.f32[3] = (mt->fast_clear_color_value & (1u << 28)) != 0;
   }

   return clear_color;
}