Linux Audio

Check our new training course

Embedded Linux Audio

Check our new training course
with Creative Commons CC-BY-SA
lecture materials

Bootlin logo

Elixir Cross Referencer

Loading...
   1
   2
   3
   4
   5
   6
   7
   8
   9
  10
  11
  12
  13
  14
  15
  16
  17
  18
  19
  20
  21
  22
  23
  24
  25
  26
  27
  28
  29
  30
  31
  32
  33
  34
  35
  36
  37
  38
  39
  40
  41
  42
  43
  44
  45
  46
  47
  48
  49
  50
  51
  52
  53
  54
  55
  56
  57
  58
  59
  60
  61
  62
  63
  64
  65
  66
  67
  68
  69
  70
  71
  72
  73
  74
  75
  76
  77
  78
  79
  80
  81
  82
  83
  84
  85
  86
  87
  88
  89
  90
  91
  92
  93
  94
  95
  96
  97
  98
  99
 100
 101
 102
 103
 104
 105
 106
 107
 108
 109
 110
 111
 112
 113
 114
 115
 116
 117
 118
 119
 120
 121
 122
 123
 124
 125
 126
 127
 128
 129
 130
 131
 132
 133
 134
 135
 136
 137
 138
 139
 140
 141
 142
 143
 144
 145
 146
 147
 148
 149
 150
 151
 152
 153
 154
 155
 156
 157
 158
 159
 160
 161
 162
 163
 164
 165
 166
 167
 168
 169
 170
 171
 172
 173
 174
 175
 176
 177
 178
 179
 180
 181
 182
 183
 184
 185
 186
 187
 188
 189
 190
 191
 192
 193
 194
 195
 196
 197
 198
 199
 200
 201
 202
 203
 204
 205
 206
 207
 208
 209
 210
 211
 212
 213
 214
 215
 216
 217
 218
 219
 220
 221
 222
 223
 224
 225
 226
 227
 228
 229
 230
 231
 232
 233
 234
 235
 236
 237
 238
 239
 240
 241
 242
 243
 244
 245
 246
 247
 248
 249
 250
 251
 252
 253
 254
 255
 256
 257
 258
 259
 260
 261
 262
 263
 264
 265
 266
 267
 268
 269
 270
 271
 272
 273
 274
 275
 276
 277
 278
 279
 280
 281
 282
 283
 284
 285
 286
 287
 288
 289
 290
 291
 292
 293
 294
 295
 296
 297
 298
 299
 300
 301
 302
 303
 304
 305
 306
 307
 308
 309
 310
 311
 312
 313
 314
 315
 316
 317
 318
 319
 320
 321
 322
 323
 324
 325
 326
 327
 328
 329
 330
 331
 332
 333
 334
 335
 336
 337
 338
 339
 340
 341
 342
 343
 344
 345
 346
 347
 348
 349
 350
 351
 352
 353
 354
 355
 356
 357
 358
 359
 360
 361
 362
 363
 364
 365
 366
 367
 368
 369
 370
 371
 372
 373
 374
 375
 376
 377
 378
 379
 380
 381
 382
 383
 384
 385
 386
 387
 388
 389
 390
 391
 392
 393
 394
 395
 396
 397
 398
 399
 400
 401
 402
 403
 404
 405
 406
 407
 408
 409
 410
 411
 412
 413
 414
 415
 416
 417
 418
 419
 420
 421
 422
 423
 424
 425
 426
 427
 428
 429
 430
 431
 432
 433
 434
 435
 436
 437
 438
 439
 440
 441
 442
 443
 444
 445
 446
 447
 448
 449
 450
 451
 452
 453
 454
 455
 456
 457
 458
 459
 460
 461
 462
 463
 464
 465
 466
 467
 468
 469
 470
 471
 472
 473
 474
 475
 476
 477
 478
 479
 480
 481
 482
 483
 484
 485
 486
 487
 488
 489
 490
 491
 492
 493
 494
 495
 496
 497
 498
 499
 500
 501
 502
 503
 504
 505
 506
 507
 508
 509
 510
 511
 512
 513
 514
 515
 516
 517
 518
 519
 520
 521
 522
 523
 524
 525
 526
 527
 528
 529
 530
 531
 532
 533
 534
 535
 536
 537
 538
 539
 540
 541
 542
 543
 544
 545
 546
 547
 548
 549
 550
 551
 552
 553
 554
 555
 556
 557
 558
 559
 560
 561
 562
 563
 564
 565
 566
 567
 568
 569
 570
 571
 572
 573
 574
 575
 576
 577
 578
 579
 580
 581
 582
 583
 584
 585
 586
 587
 588
 589
 590
 591
 592
 593
 594
 595
 596
 597
 598
 599
 600
 601
 602
 603
 604
 605
 606
 607
 608
 609
 610
 611
 612
 613
 614
 615
 616
 617
 618
 619
 620
 621
 622
 623
 624
 625
 626
 627
 628
 629
 630
 631
 632
 633
 634
 635
 636
 637
 638
 639
 640
 641
 642
 643
 644
 645
 646
 647
 648
 649
 650
 651
 652
 653
 654
 655
 656
 657
 658
 659
 660
 661
 662
 663
 664
 665
 666
 667
 668
 669
 670
 671
 672
 673
 674
 675
 676
 677
 678
 679
 680
 681
 682
 683
 684
 685
 686
 687
 688
 689
 690
 691
 692
 693
 694
 695
 696
 697
 698
 699
 700
 701
 702
 703
 704
 705
 706
 707
 708
 709
 710
 711
 712
 713
 714
 715
 716
 717
 718
 719
 720
 721
 722
 723
 724
 725
 726
 727
 728
 729
 730
 731
 732
 733
 734
 735
 736
 737
 738
 739
 740
 741
 742
 743
 744
 745
 746
 747
 748
 749
 750
 751
 752
 753
 754
 755
 756
 757
 758
 759
 760
 761
 762
 763
 764
 765
 766
 767
 768
 769
 770
 771
 772
 773
 774
 775
 776
 777
 778
 779
 780
 781
 782
 783
 784
 785
 786
 787
 788
 789
 790
 791
 792
 793
 794
 795
 796
 797
 798
 799
 800
 801
 802
 803
 804
 805
 806
 807
 808
 809
 810
 811
 812
 813
 814
 815
 816
 817
 818
 819
 820
 821
 822
 823
 824
 825
 826
 827
 828
 829
 830
 831
 832
 833
 834
 835
 836
 837
 838
 839
 840
 841
 842
 843
 844
 845
 846
 847
 848
 849
 850
 851
 852
 853
 854
 855
 856
 857
 858
 859
 860
 861
 862
 863
 864
 865
 866
 867
 868
 869
 870
 871
 872
 873
 874
 875
 876
 877
 878
 879
 880
 881
 882
 883
 884
 885
 886
 887
 888
 889
 890
 891
 892
 893
 894
 895
 896
 897
 898
 899
 900
 901
 902
 903
 904
 905
 906
 907
 908
 909
 910
 911
 912
 913
 914
 915
 916
 917
 918
 919
 920
 921
 922
 923
 924
 925
 926
 927
 928
 929
 930
 931
 932
 933
 934
 935
 936
 937
 938
 939
 940
 941
 942
 943
 944
 945
 946
 947
 948
 949
 950
 951
 952
 953
 954
 955
 956
 957
 958
 959
 960
 961
 962
 963
 964
 965
 966
 967
 968
 969
 970
 971
 972
 973
 974
 975
 976
 977
 978
 979
 980
 981
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
2836
2837
2838
2839
2840
2841
2842
2843
2844
2845
2846
2847
2848
2849
2850
2851
2852
2853
2854
2855
2856
2857
2858
2859
2860
2861
2862
2863
2864
2865
2866
2867
2868
2869
2870
2871
2872
2873
2874
2875
2876
2877
2878
2879
2880
2881
2882
2883
2884
2885
2886
2887
2888
2889
2890
2891
2892
2893
2894
2895
2896
2897
2898
2899
2900
2901
2902
2903
2904
2905
2906
2907
2908
2909
2910
2911
2912
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
2926
2927
2928
2929
2930
2931
2932
2933
2934
2935
2936
2937
2938
2939
2940
2941
2942
2943
2944
2945
2946
2947
2948
2949
2950
2951
2952
2953
2954
2955
2956
2957
2958
2959
2960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
2989
2990
2991
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025
3026
3027
3028
3029
3030
3031
3032
3033
3034
3035
3036
3037
3038
3039
3040
3041
3042
3043
3044
3045
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
3070
3071
3072
3073
3074
3075
3076
3077
3078
3079
3080
3081
3082
3083
3084
3085
3086
3087
3088
3089
3090
3091
3092
3093
3094
3095
3096
3097
3098
3099
3100
3101
3102
3103
3104
3105
3106
3107
3108
3109
3110
3111
3112
3113
3114
3115
3116
3117
3118
3119
3120
3121
3122
3123
3124
3125
3126
3127
3128
3129
3130
3131
3132
3133
3134
3135
3136
3137
3138
3139
3140
3141
3142
3143
3144
3145
3146
3147
3148
3149
3150
3151
3152
3153
3154
3155
3156
3157
3158
3159
3160
3161
3162
3163
3164
3165
3166
3167
3168
3169
3170
3171
3172
3173
3174
3175
3176
3177
3178
3179
3180
3181
3182
3183
3184
3185
3186
3187
3188
3189
3190
3191
3192
3193
3194
3195
3196
3197
3198
3199
3200
3201
3202
3203
3204
3205
3206
3207
3208
3209
3210
3211
3212
3213
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
3230
3231
3232
3233
3234
3235
3236
3237
3238
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
3266
3267
3268
3269
3270
3271
3272
3273
3274
3275
3276
3277
3278
3279
3280
3281
3282
3283
3284
3285
3286
3287
3288
3289
3290
3291
3292
3293
3294
3295
3296
3297
3298
3299
3300
3301
3302
3303
3304
3305
3306
3307
3308
3309
3310
3311
3312
3313
3314
3315
3316
3317
3318
3319
3320
3321
3322
3323
3324
3325
3326
3327
3328
3329
3330
3331
3332
3333
3334
3335
3336
3337
3338
3339
3340
3341
3342
3343
3344
3345
3346
3347
3348
3349
3350
3351
3352
3353
3354
3355
3356
3357
3358
3359
3360
3361
3362
3363
3364
3365
3366
3367
3368
3369
3370
3371
3372
3373
3374
3375
3376
3377
3378
3379
3380
3381
3382
3383
3384
3385
3386
3387
3388
3389
3390
3391
3392
3393
3394
3395
3396
3397
3398
3399
3400
3401
3402
3403
3404
3405
3406
3407
3408
3409
3410
3411
3412
3413
3414
3415
3416
3417
3418
3419
3420
3421
3422
3423
3424
3425
3426
3427
3428
3429
3430
3431
3432
3433
3434
3435
3436
3437
3438
3439
3440
3441
3442
3443
3444
3445
3446
3447
3448
3449
3450
3451
3452
3453
3454
3455
3456
3457
3458
3459
3460
3461
3462
3463
3464
3465
3466
3467
3468
3469
3470
3471
3472
3473
3474
3475
3476
3477
3478
3479
3480
3481
3482
3483
3484
3485
3486
3487
3488
3489
3490
3491
3492
3493
3494
3495
3496
3497
3498
3499
3500
3501
3502
3503
3504
3505
3506
3507
3508
3509
3510
3511
3512
3513
3514
3515
3516
3517
3518
3519
3520
3521
3522
3523
3524
3525
3526
3527
3528
3529
3530
3531
3532
3533
3534
3535
3536
3537
3538
3539
3540
3541
3542
3543
3544
3545
3546
3547
3548
3549
3550
3551
3552
3553
3554
3555
3556
3557
3558
3559
3560
3561
3562
3563
3564
3565
3566
3567
3568
3569
3570
3571
3572
3573
3574
3575
3576
3577
3578
3579
3580
3581
3582
3583
3584
3585
3586
3587
3588
3589
3590
3591
3592
3593
3594
3595
3596
3597
3598
3599
3600
3601
3602
3603
3604
3605
3606
3607
3608
3609
3610
3611
3612
3613
3614
3615
3616
3617
3618
3619
3620
3621
3622
3623
3624
3625
3626
3627
3628
3629
3630
3631
3632
3633
3634
3635
3636
3637
3638
3639
3640
3641
3642
3643
3644
3645
3646
3647
3648
3649
3650
3651
3652
3653
3654
3655
3656
3657
3658
3659
3660
3661
3662
3663
3664
3665
3666
3667
3668
3669
3670
3671
3672
3673
3674
3675
3676
3677
3678
3679
3680
3681
3682
3683
3684
3685
3686
3687
3688
3689
3690
3691
3692
3693
3694
3695
3696
3697
3698
3699
3700
3701
3702
3703
3704
3705
3706
3707
3708
3709
3710
3711
3712
3713
3714
3715
3716
3717
3718
3719
3720
3721
3722
3723
3724
3725
3726
3727
3728
3729
3730
3731
3732
3733
3734
3735
3736
3737
3738
3739
3740
3741
3742
3743
3744
3745
3746
3747
3748
3749
3750
3751
3752
3753
3754
3755
3756
3757
3758
3759
3760
3761
3762
3763
3764
3765
3766
3767
3768
3769
3770
3771
3772
3773
3774
3775
3776
3777
3778
3779
3780
3781
3782
3783
3784
3785
3786
3787
3788
3789
3790
3791
3792
3793
3794
3795
3796
3797
3798
3799
3800
3801
3802
3803
3804
3805
3806
3807
3808
3809
3810
3811
3812
3813
3814
3815
3816
3817
3818
3819
3820
3821
3822
3823
3824
3825
3826
3827
3828
3829
3830
3831
3832
3833
3834
3835
3836
3837
3838
3839
3840
3841
3842
3843
3844
3845
3846
3847
3848
3849
3850
3851
3852
3853
3854
3855
3856
3857
3858
3859
3860
3861
3862
3863
3864
3865
3866
3867
3868
3869
3870
3871
3872
3873
3874
3875
3876
3877
3878
3879
3880
3881
3882
3883
3884
3885
3886
3887
3888
3889
3890
3891
3892
3893
3894
3895
3896
3897
3898
3899
3900
3901
3902
3903
3904
3905
3906
3907
3908
3909
3910
3911
3912
3913
3914
3915
3916
3917
3918
3919
3920
3921
3922
3923
3924
3925
3926
3927
3928
3929
3930
3931
3932
3933
3934
3935
3936
3937
3938
3939
3940
3941
3942
3943
3944
3945
3946
3947
3948
3949
3950
3951
3952
3953
3954
3955
3956
3957
3958
3959
3960
3961
3962
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
4243
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
4435
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
4860
4861
4862
4863
4864
4865
4866
4867
4868
4869
4870
4871
4872
4873
4874
4875
4876
4877
4878
4879
4880
4881
4882
4883
4884
4885
4886
4887
4888
4889
4890
4891
4892
4893
4894
4895
4896
4897
4898
4899
4900
4901
4902
4903
4904
4905
4906
4907
4908
4909
4910
4911
4912
4913
4914
4915
4916
4917
4918
4919
4920
4921
4922
4923
4924
4925
4926
4927
4928
4929
4930
4931
4932
4933
4934
4935
4936
4937
4938
4939
4940
4941
4942
4943
4944
4945
4946
4947
4948
4949
4950
4951
4952
4953
4954
4955
4956
4957
4958
4959
4960
4961
4962
4963
4964
4965
4966
4967
4968
4969
4970
4971
4972
4973
4974
4975
4976
4977
4978
4979
4980
4981
4982
4983
4984
4985
4986
4987
4988
4989
4990
4991
4992
4993
4994
4995
4996
4997
4998
4999
5000
5001
5002
5003
5004
5005
5006
5007
5008
5009
5010
5011
5012
5013
5014
5015
5016
5017
5018
5019
5020
5021
5022
5023
5024
5025
5026
5027
5028
5029
5030
5031
5032
5033
5034
5035
5036
5037
5038
5039
5040
5041
5042
5043
5044
5045
5046
5047
5048
5049
5050
5051
5052
5053
5054
5055
5056
5057
5058
5059
5060
5061
5062
5063
5064
5065
5066
5067
5068
5069
5070
5071
5072
5073
5074
5075
5076
5077
5078
5079
5080
5081
5082
5083
5084
5085
5086
5087
5088
5089
5090
5091
5092
5093
5094
5095
5096
5097
5098
5099
5100
5101
5102
5103
5104
5105
5106
5107
5108
5109
5110
5111
5112
5113
5114
5115
5116
5117
5118
5119
5120
5121
5122
5123
5124
5125
5126
5127
5128
5129
5130
5131
5132
5133
5134
5135
5136
5137
5138
5139
5140
5141
5142
5143
5144
5145
5146
5147
5148
5149
5150
5151
5152
5153
5154
5155
5156
5157
5158
5159
5160
5161
5162
5163
5164
5165
5166
5167
5168
5169
5170
5171
5172
5173
5174
5175
5176
5177
5178
5179
5180
5181
5182
5183
5184
5185
5186
5187
5188
5189
5190
5191
5192
5193
5194
5195
5196
5197
5198
5199
5200
5201
5202
5203
5204
5205
5206
5207
5208
5209
5210
5211
5212
5213
5214
5215
5216
5217
5218
5219
5220
5221
5222
5223
5224
5225
5226
5227
5228
5229
5230
5231
5232
5233
5234
5235
5236
5237
5238
5239
5240
5241
5242
5243
5244
5245
5246
5247
5248
5249
5250
5251
5252
5253
5254
5255
5256
5257
5258
5259
5260
5261
5262
5263
5264
5265
5266
5267
5268
5269
5270
5271
5272
5273
5274
5275
5276
5277
5278
5279
5280
5281
5282
5283
5284
5285
5286
5287
5288
5289
5290
5291
5292
5293
5294
5295
5296
5297
5298
5299
5300
5301
5302
5303
5304
5305
5306
5307
5308
5309
5310
5311
5312
5313
5314
5315
5316
5317
5318
5319
5320
5321
5322
5323
5324
5325
5326
5327
5328
5329
5330
5331
5332
5333
5334
5335
5336
5337
5338
5339
5340
5341
5342
5343
5344
5345
5346
5347
5348
5349
5350
5351
5352
5353
5354
5355
5356
5357
5358
5359
5360
5361
5362
5363
5364
5365
5366
5367
5368
5369
5370
5371
5372
5373
5374
5375
5376
5377
5378
5379
5380
5381
5382
5383
5384
5385
5386
5387
5388
5389
5390
5391
5392
5393
5394
5395
5396
5397
5398
5399
5400
5401
5402
5403
5404
5405
5406
5407
5408
5409
5410
5411
5412
5413
5414
5415
5416
5417
5418
5419
5420
5421
5422
5423
5424
5425
5426
5427
5428
5429
5430
5431
5432
5433
5434
5435
5436
5437
5438
5439
5440
5441
5442
5443
5444
5445
5446
5447
5448
5449
5450
5451
5452
5453
5454
5455
5456
5457
5458
5459
5460
5461
5462
5463
5464
5465
5466
5467
5468
5469
5470
5471
5472
5473
5474
5475
5476
5477
5478
5479
5480
5481
5482
5483
5484
5485
5486
5487
5488
5489
5490
5491
5492
5493
5494
5495
5496
5497
5498
5499
5500
5501
5502
5503
5504
5505
5506
5507
5508
5509
5510
5511
5512
5513
5514
5515
5516
5517
5518
5519
5520
5521
5522
5523
5524
5525
5526
5527
5528
5529
5530
5531
5532
5533
5534
5535
5536
5537
5538
5539
5540
5541
5542
5543
5544
5545
5546
5547
5548
5549
5550
5551
5552
5553
5554
5555
5556
5557
5558
5559
5560
5561
5562
5563
5564
5565
5566
5567
5568
5569
5570
5571
5572
5573
5574
5575
5576
5577
5578
5579
5580
5581
5582
5583
5584
5585
5586
5587
5588
5589
5590
5591
5592
5593
5594
5595
5596
5597
5598
5599
5600
5601
5602
5603
5604
5605
5606
5607
5608
5609
5610
5611
5612
5613
5614
5615
5616
5617
5618
5619
5620
5621
5622
5623
5624
5625
5626
5627
5628
5629
5630
5631
5632
5633
5634
5635
5636
5637
5638
5639
5640
5641
5642
5643
5644
5645
5646
5647
5648
5649
5650
5651
5652
5653
5654
5655
5656
5657
5658
5659
5660
5661
5662
5663
5664
5665
5666
5667
5668
5669
5670
5671
5672
5673
5674
5675
5676
5677
5678
5679
5680
5681
5682
5683
5684
5685
5686
5687
5688
5689
5690
5691
5692
5693
5694
5695
5696
5697
5698
5699
5700
5701
5702
5703
5704
5705
5706
5707
5708
5709
5710
5711
5712
5713
5714
5715
5716
5717
5718
5719
5720
5721
5722
5723
5724
5725
5726
5727
5728
5729
5730
5731
5732
5733
5734
5735
5736
5737
5738
5739
5740
5741
5742
5743
5744
5745
5746
5747
5748
5749
5750
5751
5752
5753
5754
5755
5756
5757
5758
5759
5760
5761
5762
5763
5764
5765
5766
5767
5768
5769
5770
5771
5772
5773
5774
5775
5776
5777
5778
5779
5780
5781
5782
5783
5784
5785
5786
5787
5788
5789
5790
5791
5792
5793
5794
5795
5796
5797
5798
5799
5800
5801
5802
5803
5804
5805
5806
5807
5808
5809
5810
5811
5812
5813
5814
5815
5816
5817
5818
5819
5820
5821
5822
5823
5824
5825
5826
5827
5828
5829
5830
5831
5832
5833
5834
5835
5836
5837
5838
5839
5840
5841
5842
5843
5844
5845
5846
5847
5848
5849
5850
5851
5852
5853
5854
5855
5856
5857
5858
5859
5860
5861
5862
5863
5864
5865
5866
5867
5868
5869
5870
5871
5872
5873
5874
5875
5876
5877
5878
5879
5880
5881
5882
5883
5884
5885
5886
5887
5888
5889
5890
5891
5892
5893
5894
5895
5896
5897
5898
5899
5900
5901
5902
5903
5904
5905
5906
5907
5908
5909
5910
5911
5912
5913
5914
5915
5916
5917
5918
5919
5920
5921
5922
5923
5924
5925
5926
5927
5928
5929
5930
5931
5932
5933
5934
5935
5936
5937
5938
5939
5940
5941
5942
5943
5944
5945
5946
5947
5948
5949
5950
5951
5952
5953
5954
5955
5956
5957
5958
5959
5960
5961
5962
5963
5964
5965
5966
5967
5968
5969
5970
5971
5972
5973
5974
5975
5976
5977
5978
5979
5980
5981
5982
5983
5984
5985
5986
5987
5988
5989
5990
5991
5992
5993
5994
5995
5996
5997
5998
5999
6000
6001
6002
6003
6004
6005
6006
6007
6008
6009
6010
6011
6012
6013
6014
6015
6016
6017
6018
6019
6020
6021
6022
6023
6024
6025
6026
6027
6028
6029
6030
6031
6032
6033
6034
6035
6036
6037
6038
6039
6040
6041
6042
6043
6044
6045
6046
6047
6048
6049
6050
6051
6052
6053
6054
6055
6056
6057
6058
6059
6060
6061
6062
6063
6064
6065
6066
6067
6068
6069
6070
6071
6072
6073
6074
6075
6076
6077
6078
6079
6080
6081
6082
6083
6084
6085
6086
6087
6088
6089
6090
6091
6092
6093
6094
6095
6096
6097
6098
6099
6100
6101
6102
6103
6104
6105
6106
6107
6108
6109
6110
6111
6112
6113
6114
6115
6116
6117
6118
6119
6120
6121
6122
6123
6124
6125
6126
6127
6128
6129
6130
6131
6132
6133
6134
6135
6136
6137
6138
6139
6140
6141
6142
6143
6144
6145
6146
6147
6148
6149
6150
6151
6152
6153
6154
6155
6156
6157
6158
6159
6160
6161
6162
6163
6164
6165
6166
6167
6168
6169
6170
6171
6172
6173
6174
6175
6176
6177
6178
6179
6180
6181
6182
6183
6184
6185
6186
6187
6188
6189
6190
6191
6192
6193
6194
6195
6196
6197
6198
6199
6200
6201
6202
6203
6204
6205
6206
6207
6208
6209
6210
6211
6212
6213
6214
6215
6216
6217
6218
6219
6220
6221
6222
6223
6224
6225
6226
6227
6228
6229
6230
6231
6232
6233
6234
6235
6236
6237
6238
6239
6240
6241
6242
6243
6244
6245
6246
6247
6248
6249
6250
6251
6252
6253
6254
6255
6256
6257
6258
6259
6260
6261
6262
6263
6264
6265
6266
6267
6268
6269
6270
6271
6272
6273
6274
6275
6276
6277
6278
6279
6280
6281
6282
6283
6284
6285
6286
6287
6288
6289
6290
6291
6292
6293
6294
/*
 * Copyright © 2008-2015 Intel Corporation
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the next
 * paragraph) shall be included in all copies or substantial portions of the
 * Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
 * IN THE SOFTWARE.
 *
 * Authors:
 *    Eric Anholt <eric@anholt.net>
 *
 */

#include <drm/drmP.h>
#include <drm/drm_vma_manager.h>
#include <drm/i915_drm.h>
#include "i915_drv.h"
#include "i915_gem_clflush.h"
#include "i915_vgpu.h"
#include "i915_trace.h"
#include "intel_drv.h"
#include "intel_frontbuffer.h"
#include "intel_mocs.h"
#include "intel_workarounds.h"
#include "i915_gemfs.h"
#include <linux/dma-fence-array.h>
#include <linux/kthread.h>
#include <linux/reservation.h>
#include <linux/shmem_fs.h>
#include <linux/slab.h>
#include <linux/stop_machine.h>
#include <linux/swap.h>
#include <linux/pci.h>
#include <linux/dma-buf.h>

static void i915_gem_flush_free_objects(struct drm_i915_private *i915);

static bool cpu_write_needs_clflush(struct drm_i915_gem_object *obj)
{
	if (obj->cache_dirty)
		return false;

	if (!(obj->cache_coherent & I915_BO_CACHE_COHERENT_FOR_WRITE))
		return true;

	return obj->pin_global; /* currently in use by HW, keep flushed */
}

static int
insert_mappable_node(struct i915_ggtt *ggtt,
                     struct drm_mm_node *node, u32 size)
{
	memset(node, 0, sizeof(*node));
	return drm_mm_insert_node_in_range(&ggtt->vm.mm, node,
					   size, 0, I915_COLOR_UNEVICTABLE,
					   0, ggtt->mappable_end,
					   DRM_MM_INSERT_LOW);
}

static void
remove_mappable_node(struct drm_mm_node *node)
{
	drm_mm_remove_node(node);
}

/* some bookkeeping */
static void i915_gem_info_add_obj(struct drm_i915_private *dev_priv,
				  u64 size)
{
	spin_lock(&dev_priv->mm.object_stat_lock);
	dev_priv->mm.object_count++;
	dev_priv->mm.object_memory += size;
	spin_unlock(&dev_priv->mm.object_stat_lock);
}

static void i915_gem_info_remove_obj(struct drm_i915_private *dev_priv,
				     u64 size)
{
	spin_lock(&dev_priv->mm.object_stat_lock);
	dev_priv->mm.object_count--;
	dev_priv->mm.object_memory -= size;
	spin_unlock(&dev_priv->mm.object_stat_lock);
}

static int
i915_gem_wait_for_error(struct i915_gpu_error *error)
{
	int ret;

	might_sleep();

	/*
	 * Only wait 10 seconds for the gpu reset to complete to avoid hanging
	 * userspace. If it takes that long something really bad is going on and
	 * we should simply try to bail out and fail as gracefully as possible.
	 */
	ret = wait_event_interruptible_timeout(error->reset_queue,
					       !i915_reset_backoff(error),
					       I915_RESET_TIMEOUT);
	if (ret == 0) {
		DRM_ERROR("Timed out waiting for the gpu reset to complete\n");
		return -EIO;
	} else if (ret < 0) {
		return ret;
	} else {
		return 0;
	}
}

int i915_mutex_lock_interruptible(struct drm_device *dev)
{
	struct drm_i915_private *dev_priv = to_i915(dev);
	int ret;

	ret = i915_gem_wait_for_error(&dev_priv->gpu_error);
	if (ret)
		return ret;

	ret = mutex_lock_interruptible(&dev->struct_mutex);
	if (ret)
		return ret;

	return 0;
}

static u32 __i915_gem_park(struct drm_i915_private *i915)
{
	GEM_TRACE("\n");

	lockdep_assert_held(&i915->drm.struct_mutex);
	GEM_BUG_ON(i915->gt.active_requests);
	GEM_BUG_ON(!list_empty(&i915->gt.active_rings));

	if (!i915->gt.awake)
		return I915_EPOCH_INVALID;

	GEM_BUG_ON(i915->gt.epoch == I915_EPOCH_INVALID);

	/*
	 * Be paranoid and flush a concurrent interrupt to make sure
	 * we don't reactivate any irq tasklets after parking.
	 *
	 * FIXME: Note that even though we have waited for execlists to be idle,
	 * there may still be an in-flight interrupt even though the CSB
	 * is now empty. synchronize_irq() makes sure that a residual interrupt
	 * is completed before we continue, but it doesn't prevent the HW from
	 * raising a spurious interrupt later. To complete the shield we should
	 * coordinate disabling the CS irq with flushing the interrupts.
	 */
	synchronize_irq(i915->drm.irq);

	intel_engines_park(i915);
	i915_timelines_park(i915);

	i915_pmu_gt_parked(i915);
	i915_vma_parked(i915);

	i915->gt.awake = false;

	if (INTEL_GEN(i915) >= 6)
		gen6_rps_idle(i915);

	intel_display_power_put(i915, POWER_DOMAIN_GT_IRQ);

	intel_runtime_pm_put(i915);

	return i915->gt.epoch;
}

void i915_gem_park(struct drm_i915_private *i915)
{
	GEM_TRACE("\n");

	lockdep_assert_held(&i915->drm.struct_mutex);
	GEM_BUG_ON(i915->gt.active_requests);

	if (!i915->gt.awake)
		return;

	/* Defer the actual call to __i915_gem_park() to prevent ping-pongs */
	mod_delayed_work(i915->wq, &i915->gt.idle_work, msecs_to_jiffies(100));
}

void i915_gem_unpark(struct drm_i915_private *i915)
{
	GEM_TRACE("\n");

	lockdep_assert_held(&i915->drm.struct_mutex);
	GEM_BUG_ON(!i915->gt.active_requests);

	if (i915->gt.awake)
		return;

	intel_runtime_pm_get_noresume(i915);

	/*
	 * It seems that the DMC likes to transition between the DC states a lot
	 * when there are no connected displays (no active power domains) during
	 * command submission.
	 *
	 * This activity has negative impact on the performance of the chip with
	 * huge latencies observed in the interrupt handler and elsewhere.
	 *
	 * Work around it by grabbing a GT IRQ power domain whilst there is any
	 * GT activity, preventing any DC state transitions.
	 */
	intel_display_power_get(i915, POWER_DOMAIN_GT_IRQ);

	i915->gt.awake = true;
	if (unlikely(++i915->gt.epoch == 0)) /* keep 0 as invalid */
		i915->gt.epoch = 1;

	intel_enable_gt_powersave(i915);
	i915_update_gfx_val(i915);
	if (INTEL_GEN(i915) >= 6)
		gen6_rps_busy(i915);
	i915_pmu_gt_unparked(i915);

	intel_engines_unpark(i915);

	i915_queue_hangcheck(i915);

	queue_delayed_work(i915->wq,
			   &i915->gt.retire_work,
			   round_jiffies_up_relative(HZ));
}

int
i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data,
			    struct drm_file *file)
{
	struct drm_i915_private *dev_priv = to_i915(dev);
	struct i915_ggtt *ggtt = &dev_priv->ggtt;
	struct drm_i915_gem_get_aperture *args = data;
	struct i915_vma *vma;
	u64 pinned;

	pinned = ggtt->vm.reserved;
	mutex_lock(&dev->struct_mutex);
	list_for_each_entry(vma, &ggtt->vm.active_list, vm_link)
		if (i915_vma_is_pinned(vma))
			pinned += vma->node.size;
	list_for_each_entry(vma, &ggtt->vm.inactive_list, vm_link)
		if (i915_vma_is_pinned(vma))
			pinned += vma->node.size;
	mutex_unlock(&dev->struct_mutex);

	args->aper_size = ggtt->vm.total;
	args->aper_available_size = args->aper_size - pinned;

	return 0;
}

static int i915_gem_object_get_pages_phys(struct drm_i915_gem_object *obj)
{
	struct address_space *mapping = obj->base.filp->f_mapping;
	drm_dma_handle_t *phys;
	struct sg_table *st;
	struct scatterlist *sg;
	char *vaddr;
	int i;
	int err;

	if (WARN_ON(i915_gem_object_needs_bit17_swizzle(obj)))
		return -EINVAL;

	/* Always aligning to the object size, allows a single allocation
	 * to handle all possible callers, and given typical object sizes,
	 * the alignment of the buddy allocation will naturally match.
	 */
	phys = drm_pci_alloc(obj->base.dev,
			     roundup_pow_of_two(obj->base.size),
			     roundup_pow_of_two(obj->base.size));
	if (!phys)
		return -ENOMEM;

	vaddr = phys->vaddr;
	for (i = 0; i < obj->base.size / PAGE_SIZE; i++) {
		struct page *page;
		char *src;

		page = shmem_read_mapping_page(mapping, i);
		if (IS_ERR(page)) {
			err = PTR_ERR(page);
			goto err_phys;
		}

		src = kmap_atomic(page);
		memcpy(vaddr, src, PAGE_SIZE);
		drm_clflush_virt_range(vaddr, PAGE_SIZE);
		kunmap_atomic(src);

		put_page(page);
		vaddr += PAGE_SIZE;
	}

	i915_gem_chipset_flush(to_i915(obj->base.dev));

	st = kmalloc(sizeof(*st), GFP_KERNEL);
	if (!st) {
		err = -ENOMEM;
		goto err_phys;
	}

	if (sg_alloc_table(st, 1, GFP_KERNEL)) {
		kfree(st);
		err = -ENOMEM;
		goto err_phys;
	}

	sg = st->sgl;
	sg->offset = 0;
	sg->length = obj->base.size;

	sg_dma_address(sg) = phys->busaddr;
	sg_dma_len(sg) = obj->base.size;

	obj->phys_handle = phys;

	__i915_gem_object_set_pages(obj, st, sg->length);

	return 0;

err_phys:
	drm_pci_free(obj->base.dev, phys);

	return err;
}

static void __start_cpu_write(struct drm_i915_gem_object *obj)
{
	obj->read_domains = I915_GEM_DOMAIN_CPU;
	obj->write_domain = I915_GEM_DOMAIN_CPU;
	if (cpu_write_needs_clflush(obj))
		obj->cache_dirty = true;
}

static void
__i915_gem_object_release_shmem(struct drm_i915_gem_object *obj,
				struct sg_table *pages,
				bool needs_clflush)
{
	GEM_BUG_ON(obj->mm.madv == __I915_MADV_PURGED);

	if (obj->mm.madv == I915_MADV_DONTNEED)
		obj->mm.dirty = false;

	if (needs_clflush &&
	    (obj->read_domains & I915_GEM_DOMAIN_CPU) == 0 &&
	    !(obj->cache_coherent & I915_BO_CACHE_COHERENT_FOR_READ))
		drm_clflush_sg(pages);

	__start_cpu_write(obj);
}

static void
i915_gem_object_put_pages_phys(struct drm_i915_gem_object *obj,
			       struct sg_table *pages)
{
	__i915_gem_object_release_shmem(obj, pages, false);

	if (obj->mm.dirty) {
		struct address_space *mapping = obj->base.filp->f_mapping;
		char *vaddr = obj->phys_handle->vaddr;
		int i;

		for (i = 0; i < obj->base.size / PAGE_SIZE; i++) {
			struct page *page;
			char *dst;

			page = shmem_read_mapping_page(mapping, i);
			if (IS_ERR(page))
				continue;

			dst = kmap_atomic(page);
			drm_clflush_virt_range(vaddr, PAGE_SIZE);
			memcpy(dst, vaddr, PAGE_SIZE);
			kunmap_atomic(dst);

			set_page_dirty(page);
			if (obj->mm.madv == I915_MADV_WILLNEED)
				mark_page_accessed(page);
			put_page(page);
			vaddr += PAGE_SIZE;
		}
		obj->mm.dirty = false;
	}

	sg_free_table(pages);
	kfree(pages);

	drm_pci_free(obj->base.dev, obj->phys_handle);
}

static void
i915_gem_object_release_phys(struct drm_i915_gem_object *obj)
{
	i915_gem_object_unpin_pages(obj);
}

static const struct drm_i915_gem_object_ops i915_gem_phys_ops = {
	.get_pages = i915_gem_object_get_pages_phys,
	.put_pages = i915_gem_object_put_pages_phys,
	.release = i915_gem_object_release_phys,
};

static const struct drm_i915_gem_object_ops i915_gem_object_ops;

int i915_gem_object_unbind(struct drm_i915_gem_object *obj)
{
	struct i915_vma *vma;
	LIST_HEAD(still_in_list);
	int ret;

	lockdep_assert_held(&obj->base.dev->struct_mutex);

	/* Closed vma are removed from the obj->vma_list - but they may
	 * still have an active binding on the object. To remove those we
	 * must wait for all rendering to complete to the object (as unbinding
	 * must anyway), and retire the requests.
	 */
	ret = i915_gem_object_set_to_cpu_domain(obj, false);
	if (ret)
		return ret;

	while ((vma = list_first_entry_or_null(&obj->vma_list,
					       struct i915_vma,
					       obj_link))) {
		list_move_tail(&vma->obj_link, &still_in_list);
		ret = i915_vma_unbind(vma);
		if (ret)
			break;
	}
	list_splice(&still_in_list, &obj->vma_list);

	return ret;
}

static long
i915_gem_object_wait_fence(struct dma_fence *fence,
			   unsigned int flags,
			   long timeout,
			   struct intel_rps_client *rps_client)
{
	struct i915_request *rq;

	BUILD_BUG_ON(I915_WAIT_INTERRUPTIBLE != 0x1);

	if (test_bit(DMA_FENCE_FLAG_SIGNALED_BIT, &fence->flags))
		return timeout;

	if (!dma_fence_is_i915(fence))
		return dma_fence_wait_timeout(fence,
					      flags & I915_WAIT_INTERRUPTIBLE,
					      timeout);

	rq = to_request(fence);
	if (i915_request_completed(rq))
		goto out;

	/*
	 * This client is about to stall waiting for the GPU. In many cases
	 * this is undesirable and limits the throughput of the system, as
	 * many clients cannot continue processing user input/output whilst
	 * blocked. RPS autotuning may take tens of milliseconds to respond
	 * to the GPU load and thus incurs additional latency for the client.
	 * We can circumvent that by promoting the GPU frequency to maximum
	 * before we wait. This makes the GPU throttle up much more quickly
	 * (good for benchmarks and user experience, e.g. window animations),
	 * but at a cost of spending more power processing the workload
	 * (bad for battery). Not all clients even want their results
	 * immediately and for them we should just let the GPU select its own
	 * frequency to maximise efficiency. To prevent a single client from
	 * forcing the clocks too high for the whole system, we only allow
	 * each client to waitboost once in a busy period.
	 */
	if (rps_client && !i915_request_started(rq)) {
		if (INTEL_GEN(rq->i915) >= 6)
			gen6_rps_boost(rq, rps_client);
	}

	timeout = i915_request_wait(rq, flags, timeout);

out:
	if (flags & I915_WAIT_LOCKED && i915_request_completed(rq))
		i915_request_retire_upto(rq);

	return timeout;
}

static long
i915_gem_object_wait_reservation(struct reservation_object *resv,
				 unsigned int flags,
				 long timeout,
				 struct intel_rps_client *rps_client)
{
	unsigned int seq = __read_seqcount_begin(&resv->seq);
	struct dma_fence *excl;
	bool prune_fences = false;

	if (flags & I915_WAIT_ALL) {
		struct dma_fence **shared;
		unsigned int count, i;
		int ret;

		ret = reservation_object_get_fences_rcu(resv,
							&excl, &count, &shared);
		if (ret)
			return ret;

		for (i = 0; i < count; i++) {
			timeout = i915_gem_object_wait_fence(shared[i],
							     flags, timeout,
							     rps_client);
			if (timeout < 0)
				break;

			dma_fence_put(shared[i]);
		}

		for (; i < count; i++)
			dma_fence_put(shared[i]);
		kfree(shared);

		/*
		 * If both shared fences and an exclusive fence exist,
		 * then by construction the shared fences must be later
		 * than the exclusive fence. If we successfully wait for
		 * all the shared fences, we know that the exclusive fence
		 * must all be signaled. If all the shared fences are
		 * signaled, we can prune the array and recover the
		 * floating references on the fences/requests.
		 */
		prune_fences = count && timeout >= 0;
	} else {
		excl = reservation_object_get_excl_rcu(resv);
	}

	if (excl && timeout >= 0)
		timeout = i915_gem_object_wait_fence(excl, flags, timeout,
						     rps_client);

	dma_fence_put(excl);

	/*
	 * Opportunistically prune the fences iff we know they have *all* been
	 * signaled and that the reservation object has not been changed (i.e.
	 * no new fences have been added).
	 */
	if (prune_fences && !__read_seqcount_retry(&resv->seq, seq)) {
		if (reservation_object_trylock(resv)) {
			if (!__read_seqcount_retry(&resv->seq, seq))
				reservation_object_add_excl_fence(resv, NULL);
			reservation_object_unlock(resv);
		}
	}

	return timeout;
}

static void __fence_set_priority(struct dma_fence *fence,
				 const struct i915_sched_attr *attr)
{
	struct i915_request *rq;
	struct intel_engine_cs *engine;

	if (dma_fence_is_signaled(fence) || !dma_fence_is_i915(fence))
		return;

	rq = to_request(fence);
	engine = rq->engine;

	local_bh_disable();
	rcu_read_lock(); /* RCU serialisation for set-wedged protection */
	if (engine->schedule)
		engine->schedule(rq, attr);
	rcu_read_unlock();
	local_bh_enable(); /* kick the tasklets if queues were reprioritised */
}

static void fence_set_priority(struct dma_fence *fence,
			       const struct i915_sched_attr *attr)
{
	/* Recurse once into a fence-array */
	if (dma_fence_is_array(fence)) {
		struct dma_fence_array *array = to_dma_fence_array(fence);
		int i;

		for (i = 0; i < array->num_fences; i++)
			__fence_set_priority(array->fences[i], attr);
	} else {
		__fence_set_priority(fence, attr);
	}
}

int
i915_gem_object_wait_priority(struct drm_i915_gem_object *obj,
			      unsigned int flags,
			      const struct i915_sched_attr *attr)
{
	struct dma_fence *excl;

	if (flags & I915_WAIT_ALL) {
		struct dma_fence **shared;
		unsigned int count, i;
		int ret;

		ret = reservation_object_get_fences_rcu(obj->resv,
							&excl, &count, &shared);
		if (ret)
			return ret;

		for (i = 0; i < count; i++) {
			fence_set_priority(shared[i], attr);
			dma_fence_put(shared[i]);
		}

		kfree(shared);
	} else {
		excl = reservation_object_get_excl_rcu(obj->resv);
	}

	if (excl) {
		fence_set_priority(excl, attr);
		dma_fence_put(excl);
	}
	return 0;
}

/**
 * Waits for rendering to the object to be completed
 * @obj: i915 gem object
 * @flags: how to wait (under a lock, for all rendering or just for writes etc)
 * @timeout: how long to wait
 * @rps_client: client (user process) to charge for any waitboosting
 */
int
i915_gem_object_wait(struct drm_i915_gem_object *obj,
		     unsigned int flags,
		     long timeout,
		     struct intel_rps_client *rps_client)
{
	might_sleep();
#if IS_ENABLED(CONFIG_LOCKDEP)
	GEM_BUG_ON(debug_locks &&
		   !!lockdep_is_held(&obj->base.dev->struct_mutex) !=
		   !!(flags & I915_WAIT_LOCKED));
#endif
	GEM_BUG_ON(timeout < 0);

	timeout = i915_gem_object_wait_reservation(obj->resv,
						   flags, timeout,
						   rps_client);
	return timeout < 0 ? timeout : 0;
}

static struct intel_rps_client *to_rps_client(struct drm_file *file)
{
	struct drm_i915_file_private *fpriv = file->driver_priv;

	return &fpriv->rps_client;
}

static int
i915_gem_phys_pwrite(struct drm_i915_gem_object *obj,
		     struct drm_i915_gem_pwrite *args,
		     struct drm_file *file)
{
	void *vaddr = obj->phys_handle->vaddr + args->offset;
	char __user *user_data = u64_to_user_ptr(args->data_ptr);

	/* We manually control the domain here and pretend that it
	 * remains coherent i.e. in the GTT domain, like shmem_pwrite.
	 */
	intel_fb_obj_invalidate(obj, ORIGIN_CPU);
	if (copy_from_user(vaddr, user_data, args->size))
		return -EFAULT;

	drm_clflush_virt_range(vaddr, args->size);
	i915_gem_chipset_flush(to_i915(obj->base.dev));

	intel_fb_obj_flush(obj, ORIGIN_CPU);
	return 0;
}

void *i915_gem_object_alloc(struct drm_i915_private *dev_priv)
{
	return kmem_cache_zalloc(dev_priv->objects, GFP_KERNEL);
}

void i915_gem_object_free(struct drm_i915_gem_object *obj)
{
	struct drm_i915_private *dev_priv = to_i915(obj->base.dev);
	kmem_cache_free(dev_priv->objects, obj);
}

static int
i915_gem_create(struct drm_file *file,
		struct drm_i915_private *dev_priv,
		uint64_t size,
		uint32_t *handle_p)
{
	struct drm_i915_gem_object *obj;
	int ret;
	u32 handle;

	size = roundup(size, PAGE_SIZE);
	if (size == 0)
		return -EINVAL;

	/* Allocate the new object */
	obj = i915_gem_object_create(dev_priv, size);
	if (IS_ERR(obj))
		return PTR_ERR(obj);

	ret = drm_gem_handle_create(file, &obj->base, &handle);
	/* drop reference from allocate - handle holds it now */
	i915_gem_object_put(obj);
	if (ret)
		return ret;

	*handle_p = handle;
	return 0;
}

int
i915_gem_dumb_create(struct drm_file *file,
		     struct drm_device *dev,
		     struct drm_mode_create_dumb *args)
{
	/* have to work out size/pitch and return them */
	args->pitch = ALIGN(args->width * DIV_ROUND_UP(args->bpp, 8), 64);
	args->size = args->pitch * args->height;
	return i915_gem_create(file, to_i915(dev),
			       args->size, &args->handle);
}

static bool gpu_write_needs_clflush(struct drm_i915_gem_object *obj)
{
	return !(obj->cache_level == I915_CACHE_NONE ||
		 obj->cache_level == I915_CACHE_WT);
}

/**
 * Creates a new mm object and returns a handle to it.
 * @dev: drm device pointer
 * @data: ioctl data blob
 * @file: drm file pointer
 */
int
i915_gem_create_ioctl(struct drm_device *dev, void *data,
		      struct drm_file *file)
{
	struct drm_i915_private *dev_priv = to_i915(dev);
	struct drm_i915_gem_create *args = data;

	i915_gem_flush_free_objects(dev_priv);

	return i915_gem_create(file, dev_priv,
			       args->size, &args->handle);
}

static inline enum fb_op_origin
fb_write_origin(struct drm_i915_gem_object *obj, unsigned int domain)
{
	return (domain == I915_GEM_DOMAIN_GTT ?
		obj->frontbuffer_ggtt_origin : ORIGIN_CPU);
}

void i915_gem_flush_ggtt_writes(struct drm_i915_private *dev_priv)
{
	/*
	 * No actual flushing is required for the GTT write domain for reads
	 * from the GTT domain. Writes to it "immediately" go to main memory
	 * as far as we know, so there's no chipset flush. It also doesn't
	 * land in the GPU render cache.
	 *
	 * However, we do have to enforce the order so that all writes through
	 * the GTT land before any writes to the device, such as updates to
	 * the GATT itself.
	 *
	 * We also have to wait a bit for the writes to land from the GTT.
	 * An uncached read (i.e. mmio) seems to be ideal for the round-trip
	 * timing. This issue has only been observed when switching quickly
	 * between GTT writes and CPU reads from inside the kernel on recent hw,
	 * and it appears to only affect discrete GTT blocks (i.e. on LLC
	 * system agents we cannot reproduce this behaviour, until Cannonlake
	 * that was!).
	 */

	wmb();

	if (INTEL_INFO(dev_priv)->has_coherent_ggtt)
		return;

	i915_gem_chipset_flush(dev_priv);

	intel_runtime_pm_get(dev_priv);
	spin_lock_irq(&dev_priv->uncore.lock);

	POSTING_READ_FW(RING_HEAD(RENDER_RING_BASE));

	spin_unlock_irq(&dev_priv->uncore.lock);
	intel_runtime_pm_put(dev_priv);
}

static void
flush_write_domain(struct drm_i915_gem_object *obj, unsigned int flush_domains)
{
	struct drm_i915_private *dev_priv = to_i915(obj->base.dev);
	struct i915_vma *vma;

	if (!(obj->write_domain & flush_domains))
		return;

	switch (obj->write_domain) {
	case I915_GEM_DOMAIN_GTT:
		i915_gem_flush_ggtt_writes(dev_priv);

		intel_fb_obj_flush(obj,
				   fb_write_origin(obj, I915_GEM_DOMAIN_GTT));

		for_each_ggtt_vma(vma, obj) {
			if (vma->iomap)
				continue;

			i915_vma_unset_ggtt_write(vma);
		}
		break;

	case I915_GEM_DOMAIN_WC:
		wmb();
		break;

	case I915_GEM_DOMAIN_CPU:
		i915_gem_clflush_object(obj, I915_CLFLUSH_SYNC);
		break;

	case I915_GEM_DOMAIN_RENDER:
		if (gpu_write_needs_clflush(obj))
			obj->cache_dirty = true;
		break;
	}

	obj->write_domain = 0;
}

static inline int
__copy_to_user_swizzled(char __user *cpu_vaddr,
			const char *gpu_vaddr, int gpu_offset,
			int length)
{
	int ret, cpu_offset = 0;

	while (length > 0) {
		int cacheline_end = ALIGN(gpu_offset + 1, 64);
		int this_length = min(cacheline_end - gpu_offset, length);
		int swizzled_gpu_offset = gpu_offset ^ 64;

		ret = __copy_to_user(cpu_vaddr + cpu_offset,
				     gpu_vaddr + swizzled_gpu_offset,
				     this_length);
		if (ret)
			return ret + length;

		cpu_offset += this_length;
		gpu_offset += this_length;
		length -= this_length;
	}

	return 0;
}

static inline int
__copy_from_user_swizzled(char *gpu_vaddr, int gpu_offset,
			  const char __user *cpu_vaddr,
			  int length)
{
	int ret, cpu_offset = 0;

	while (length > 0) {
		int cacheline_end = ALIGN(gpu_offset + 1, 64);
		int this_length = min(cacheline_end - gpu_offset, length);
		int swizzled_gpu_offset = gpu_offset ^ 64;

		ret = __copy_from_user(gpu_vaddr + swizzled_gpu_offset,
				       cpu_vaddr + cpu_offset,
				       this_length);
		if (ret)
			return ret + length;

		cpu_offset += this_length;
		gpu_offset += this_length;
		length -= this_length;
	}

	return 0;
}

/*
 * Pins the specified object's pages and synchronizes the object with
 * GPU accesses. Sets needs_clflush to non-zero if the caller should
 * flush the object from the CPU cache.
 */
int i915_gem_obj_prepare_shmem_read(struct drm_i915_gem_object *obj,
				    unsigned int *needs_clflush)
{
	int ret;

	lockdep_assert_held(&obj->base.dev->struct_mutex);

	*needs_clflush = 0;
	if (!i915_gem_object_has_struct_page(obj))
		return -ENODEV;

	ret = i915_gem_object_wait(obj,
				   I915_WAIT_INTERRUPTIBLE |
				   I915_WAIT_LOCKED,
				   MAX_SCHEDULE_TIMEOUT,
				   NULL);
	if (ret)
		return ret;

	ret = i915_gem_object_pin_pages(obj);
	if (ret)
		return ret;

	if (obj->cache_coherent & I915_BO_CACHE_COHERENT_FOR_READ ||
	    !static_cpu_has(X86_FEATURE_CLFLUSH)) {
		ret = i915_gem_object_set_to_cpu_domain(obj, false);
		if (ret)
			goto err_unpin;
		else
			goto out;
	}

	flush_write_domain(obj, ~I915_GEM_DOMAIN_CPU);

	/* If we're not in the cpu read domain, set ourself into the gtt
	 * read domain and manually flush cachelines (if required). This
	 * optimizes for the case when the gpu will dirty the data
	 * anyway again before the next pread happens.
	 */
	if (!obj->cache_dirty &&
	    !(obj->read_domains & I915_GEM_DOMAIN_CPU))
		*needs_clflush = CLFLUSH_BEFORE;

out:
	/* return with the pages pinned */
	return 0;

err_unpin:
	i915_gem_object_unpin_pages(obj);
	return ret;
}

int i915_gem_obj_prepare_shmem_write(struct drm_i915_gem_object *obj,
				     unsigned int *needs_clflush)
{
	int ret;

	lockdep_assert_held(&obj->base.dev->struct_mutex);

	*needs_clflush = 0;
	if (!i915_gem_object_has_struct_page(obj))
		return -ENODEV;

	ret = i915_gem_object_wait(obj,
				   I915_WAIT_INTERRUPTIBLE |
				   I915_WAIT_LOCKED |
				   I915_WAIT_ALL,
				   MAX_SCHEDULE_TIMEOUT,
				   NULL);
	if (ret)
		return ret;

	ret = i915_gem_object_pin_pages(obj);
	if (ret)
		return ret;

	if (obj->cache_coherent & I915_BO_CACHE_COHERENT_FOR_WRITE ||
	    !static_cpu_has(X86_FEATURE_CLFLUSH)) {
		ret = i915_gem_object_set_to_cpu_domain(obj, true);
		if (ret)
			goto err_unpin;
		else
			goto out;
	}

	flush_write_domain(obj, ~I915_GEM_DOMAIN_CPU);

	/* If we're not in the cpu write domain, set ourself into the
	 * gtt write domain and manually flush cachelines (as required).
	 * This optimizes for the case when the gpu will use the data
	 * right away and we therefore have to clflush anyway.
	 */
	if (!obj->cache_dirty) {
		*needs_clflush |= CLFLUSH_AFTER;

		/*
		 * Same trick applies to invalidate partially written
		 * cachelines read before writing.
		 */
		if (!(obj->read_domains & I915_GEM_DOMAIN_CPU))
			*needs_clflush |= CLFLUSH_BEFORE;
	}

out:
	intel_fb_obj_invalidate(obj, ORIGIN_CPU);
	obj->mm.dirty = true;
	/* return with the pages pinned */
	return 0;

err_unpin:
	i915_gem_object_unpin_pages(obj);
	return ret;
}

static void
shmem_clflush_swizzled_range(char *addr, unsigned long length,
			     bool swizzled)
{
	if (unlikely(swizzled)) {
		unsigned long start = (unsigned long) addr;
		unsigned long end = (unsigned long) addr + length;

		/* For swizzling simply ensure that we always flush both
		 * channels. Lame, but simple and it works. Swizzled
		 * pwrite/pread is far from a hotpath - current userspace
		 * doesn't use it at all. */
		start = round_down(start, 128);
		end = round_up(end, 128);

		drm_clflush_virt_range((void *)start, end - start);
	} else {
		drm_clflush_virt_range(addr, length);
	}

}

/* Only difference to the fast-path function is that this can handle bit17
 * and uses non-atomic copy and kmap functions. */
static int
shmem_pread_slow(struct page *page, int offset, int length,
		 char __user *user_data,
		 bool page_do_bit17_swizzling, bool needs_clflush)
{
	char *vaddr;
	int ret;

	vaddr = kmap(page);
	if (needs_clflush)
		shmem_clflush_swizzled_range(vaddr + offset, length,
					     page_do_bit17_swizzling);

	if (page_do_bit17_swizzling)
		ret = __copy_to_user_swizzled(user_data, vaddr, offset, length);
	else
		ret = __copy_to_user(user_data, vaddr + offset, length);
	kunmap(page);

	return ret ? - EFAULT : 0;
}

static int
shmem_pread(struct page *page, int offset, int length, char __user *user_data,
	    bool page_do_bit17_swizzling, bool needs_clflush)
{
	int ret;

	ret = -ENODEV;
	if (!page_do_bit17_swizzling) {
		char *vaddr = kmap_atomic(page);

		if (needs_clflush)
			drm_clflush_virt_range(vaddr + offset, length);
		ret = __copy_to_user_inatomic(user_data, vaddr + offset, length);
		kunmap_atomic(vaddr);
	}
	if (ret == 0)
		return 0;

	return shmem_pread_slow(page, offset, length, user_data,
				page_do_bit17_swizzling, needs_clflush);
}

static int
i915_gem_shmem_pread(struct drm_i915_gem_object *obj,
		     struct drm_i915_gem_pread *args)
{
	char __user *user_data;
	u64 remain;
	unsigned int obj_do_bit17_swizzling;
	unsigned int needs_clflush;
	unsigned int idx, offset;
	int ret;

	obj_do_bit17_swizzling = 0;
	if (i915_gem_object_needs_bit17_swizzle(obj))
		obj_do_bit17_swizzling = BIT(17);

	ret = mutex_lock_interruptible(&obj->base.dev->struct_mutex);
	if (ret)
		return ret;

	ret = i915_gem_obj_prepare_shmem_read(obj, &needs_clflush);
	mutex_unlock(&obj->base.dev->struct_mutex);
	if (ret)
		return ret;

	remain = args->size;
	user_data = u64_to_user_ptr(args->data_ptr);
	offset = offset_in_page(args->offset);
	for (idx = args->offset >> PAGE_SHIFT; remain; idx++) {
		struct page *page = i915_gem_object_get_page(obj, idx);
		unsigned int length = min_t(u64, remain, PAGE_SIZE - offset);

		ret = shmem_pread(page, offset, length, user_data,
				  page_to_phys(page) & obj_do_bit17_swizzling,
				  needs_clflush);
		if (ret)
			break;

		remain -= length;
		user_data += length;
		offset = 0;
	}

	i915_gem_obj_finish_shmem_access(obj);
	return ret;
}

static inline bool
gtt_user_read(struct io_mapping *mapping,
	      loff_t base, int offset,
	      char __user *user_data, int length)
{
	void __iomem *vaddr;
	unsigned long unwritten;

	/* We can use the cpu mem copy function because this is X86. */
	vaddr = io_mapping_map_atomic_wc(mapping, base);
	unwritten = __copy_to_user_inatomic(user_data,
					    (void __force *)vaddr + offset,
					    length);
	io_mapping_unmap_atomic(vaddr);
	if (unwritten) {
		vaddr = io_mapping_map_wc(mapping, base, PAGE_SIZE);
		unwritten = copy_to_user(user_data,
					 (void __force *)vaddr + offset,
					 length);
		io_mapping_unmap(vaddr);
	}
	return unwritten;
}

static int
i915_gem_gtt_pread(struct drm_i915_gem_object *obj,
		   const struct drm_i915_gem_pread *args)
{
	struct drm_i915_private *i915 = to_i915(obj->base.dev);
	struct i915_ggtt *ggtt = &i915->ggtt;
	struct drm_mm_node node;
	struct i915_vma *vma;
	void __user *user_data;
	u64 remain, offset;
	int ret;

	ret = mutex_lock_interruptible(&i915->drm.struct_mutex);
	if (ret)
		return ret;

	intel_runtime_pm_get(i915);
	vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0,
				       PIN_MAPPABLE |
				       PIN_NONFAULT |
				       PIN_NONBLOCK);
	if (!IS_ERR(vma)) {
		node.start = i915_ggtt_offset(vma);
		node.allocated = false;
		ret = i915_vma_put_fence(vma);
		if (ret) {
			i915_vma_unpin(vma);
			vma = ERR_PTR(ret);
		}
	}
	if (IS_ERR(vma)) {
		ret = insert_mappable_node(ggtt, &node, PAGE_SIZE);
		if (ret)
			goto out_unlock;
		GEM_BUG_ON(!node.allocated);
	}

	ret = i915_gem_object_set_to_gtt_domain(obj, false);
	if (ret)
		goto out_unpin;

	mutex_unlock(&i915->drm.struct_mutex);

	user_data = u64_to_user_ptr(args->data_ptr);
	remain = args->size;
	offset = args->offset;

	while (remain > 0) {
		/* Operation in this page
		 *
		 * page_base = page offset within aperture
		 * page_offset = offset within page
		 * page_length = bytes to copy for this page
		 */
		u32 page_base = node.start;
		unsigned page_offset = offset_in_page(offset);
		unsigned page_length = PAGE_SIZE - page_offset;
		page_length = remain < page_length ? remain : page_length;
		if (node.allocated) {
			wmb();
			ggtt->vm.insert_page(&ggtt->vm,
					     i915_gem_object_get_dma_address(obj, offset >> PAGE_SHIFT),
					     node.start, I915_CACHE_NONE, 0);
			wmb();
		} else {
			page_base += offset & PAGE_MASK;
		}

		if (gtt_user_read(&ggtt->iomap, page_base, page_offset,
				  user_data, page_length)) {
			ret = -EFAULT;
			break;
		}

		remain -= page_length;
		user_data += page_length;
		offset += page_length;
	}

	mutex_lock(&i915->drm.struct_mutex);
out_unpin:
	if (node.allocated) {
		wmb();
		ggtt->vm.clear_range(&ggtt->vm, node.start, node.size);
		remove_mappable_node(&node);
	} else {
		i915_vma_unpin(vma);
	}
out_unlock:
	intel_runtime_pm_put(i915);
	mutex_unlock(&i915->drm.struct_mutex);

	return ret;
}

/**
 * Reads data from the object referenced by handle.
 * @dev: drm device pointer
 * @data: ioctl data blob
 * @file: drm file pointer
 *
 * On error, the contents of *data are undefined.
 */
int
i915_gem_pread_ioctl(struct drm_device *dev, void *data,
		     struct drm_file *file)
{
	struct drm_i915_gem_pread *args = data;
	struct drm_i915_gem_object *obj;
	int ret;

	if (args->size == 0)
		return 0;

	if (!access_ok(u64_to_user_ptr(args->data_ptr),
		       args->size))
		return -EFAULT;

	obj = i915_gem_object_lookup(file, args->handle);
	if (!obj)
		return -ENOENT;

	/* Bounds check source.  */
	if (range_overflows_t(u64, args->offset, args->size, obj->base.size)) {
		ret = -EINVAL;
		goto out;
	}

	trace_i915_gem_object_pread(obj, args->offset, args->size);

	ret = i915_gem_object_wait(obj,
				   I915_WAIT_INTERRUPTIBLE,
				   MAX_SCHEDULE_TIMEOUT,
				   to_rps_client(file));
	if (ret)
		goto out;

	ret = i915_gem_object_pin_pages(obj);
	if (ret)
		goto out;

	ret = i915_gem_shmem_pread(obj, args);
	if (ret == -EFAULT || ret == -ENODEV)
		ret = i915_gem_gtt_pread(obj, args);

	i915_gem_object_unpin_pages(obj);
out:
	i915_gem_object_put(obj);
	return ret;
}

/* This is the fast write path which cannot handle
 * page faults in the source data
 */

static inline bool
ggtt_write(struct io_mapping *mapping,
	   loff_t base, int offset,
	   char __user *user_data, int length)
{
	void __iomem *vaddr;
	unsigned long unwritten;

	/* We can use the cpu mem copy function because this is X86. */
	vaddr = io_mapping_map_atomic_wc(mapping, base);
	unwritten = __copy_from_user_inatomic_nocache((void __force *)vaddr + offset,
						      user_data, length);
	io_mapping_unmap_atomic(vaddr);
	if (unwritten) {
		vaddr = io_mapping_map_wc(mapping, base, PAGE_SIZE);
		unwritten = copy_from_user((void __force *)vaddr + offset,
					   user_data, length);
		io_mapping_unmap(vaddr);
	}

	return unwritten;
}

/**
 * This is the fast pwrite path, where we copy the data directly from the
 * user into the GTT, uncached.
 * @obj: i915 GEM object
 * @args: pwrite arguments structure
 */
static int
i915_gem_gtt_pwrite_fast(struct drm_i915_gem_object *obj,
			 const struct drm_i915_gem_pwrite *args)
{
	struct drm_i915_private *i915 = to_i915(obj->base.dev);
	struct i915_ggtt *ggtt = &i915->ggtt;
	struct drm_mm_node node;
	struct i915_vma *vma;
	u64 remain, offset;
	void __user *user_data;
	int ret;

	ret = mutex_lock_interruptible(&i915->drm.struct_mutex);
	if (ret)
		return ret;

	if (i915_gem_object_has_struct_page(obj)) {
		/*
		 * Avoid waking the device up if we can fallback, as
		 * waking/resuming is very slow (worst-case 10-100 ms
		 * depending on PCI sleeps and our own resume time).
		 * This easily dwarfs any performance advantage from
		 * using the cache bypass of indirect GGTT access.
		 */
		if (!intel_runtime_pm_get_if_in_use(i915)) {
			ret = -EFAULT;
			goto out_unlock;
		}
	} else {
		/* No backing pages, no fallback, we must force GGTT access */
		intel_runtime_pm_get(i915);
	}

	vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0,
				       PIN_MAPPABLE |
				       PIN_NONFAULT |
				       PIN_NONBLOCK);
	if (!IS_ERR(vma)) {
		node.start = i915_ggtt_offset(vma);
		node.allocated = false;
		ret = i915_vma_put_fence(vma);
		if (ret) {
			i915_vma_unpin(vma);
			vma = ERR_PTR(ret);
		}
	}
	if (IS_ERR(vma)) {
		ret = insert_mappable_node(ggtt, &node, PAGE_SIZE);
		if (ret)
			goto out_rpm;
		GEM_BUG_ON(!node.allocated);
	}

	ret = i915_gem_object_set_to_gtt_domain(obj, true);
	if (ret)
		goto out_unpin;

	mutex_unlock(&i915->drm.struct_mutex);

	intel_fb_obj_invalidate(obj, ORIGIN_CPU);

	user_data = u64_to_user_ptr(args->data_ptr);
	offset = args->offset;
	remain = args->size;
	while (remain) {
		/* Operation in this page
		 *
		 * page_base = page offset within aperture
		 * page_offset = offset within page
		 * page_length = bytes to copy for this page
		 */
		u32 page_base = node.start;
		unsigned int page_offset = offset_in_page(offset);
		unsigned int page_length = PAGE_SIZE - page_offset;
		page_length = remain < page_length ? remain : page_length;
		if (node.allocated) {
			wmb(); /* flush the write before we modify the GGTT */
			ggtt->vm.insert_page(&ggtt->vm,
					     i915_gem_object_get_dma_address(obj, offset >> PAGE_SHIFT),
					     node.start, I915_CACHE_NONE, 0);
			wmb(); /* flush modifications to the GGTT (insert_page) */
		} else {
			page_base += offset & PAGE_MASK;
		}
		/* If we get a fault while copying data, then (presumably) our
		 * source page isn't available.  Return the error and we'll
		 * retry in the slow path.
		 * If the object is non-shmem backed, we retry again with the
		 * path that handles page fault.
		 */
		if (ggtt_write(&ggtt->iomap, page_base, page_offset,
			       user_data, page_length)) {
			ret = -EFAULT;
			break;
		}

		remain -= page_length;
		user_data += page_length;
		offset += page_length;
	}
	intel_fb_obj_flush(obj, ORIGIN_CPU);

	mutex_lock(&i915->drm.struct_mutex);
out_unpin:
	if (node.allocated) {
		wmb();
		ggtt->vm.clear_range(&ggtt->vm, node.start, node.size);
		remove_mappable_node(&node);
	} else {
		i915_vma_unpin(vma);
	}
out_rpm:
	intel_runtime_pm_put(i915);
out_unlock:
	mutex_unlock(&i915->drm.struct_mutex);
	return ret;
}

static int
shmem_pwrite_slow(struct page *page, int offset, int length,
		  char __user *user_data,
		  bool page_do_bit17_swizzling,
		  bool needs_clflush_before,
		  bool needs_clflush_after)
{
	char *vaddr;
	int ret;

	vaddr = kmap(page);
	if (unlikely(needs_clflush_before || page_do_bit17_swizzling))
		shmem_clflush_swizzled_range(vaddr + offset, length,
					     page_do_bit17_swizzling);
	if (page_do_bit17_swizzling)
		ret = __copy_from_user_swizzled(vaddr, offset, user_data,
						length);
	else
		ret = __copy_from_user(vaddr + offset, user_data, length);
	if (needs_clflush_after)
		shmem_clflush_swizzled_range(vaddr + offset, length,
					     page_do_bit17_swizzling);
	kunmap(page);

	return ret ? -EFAULT : 0;
}

/* Per-page copy function for the shmem pwrite fastpath.
 * Flushes invalid cachelines before writing to the target if
 * needs_clflush_before is set and flushes out any written cachelines after
 * writing if needs_clflush is set.
 */
static int
shmem_pwrite(struct page *page, int offset, int len, char __user *user_data,
	     bool page_do_bit17_swizzling,
	     bool needs_clflush_before,
	     bool needs_clflush_after)
{
	int ret;

	ret = -ENODEV;
	if (!page_do_bit17_swizzling) {
		char *vaddr = kmap_atomic(page);

		if (needs_clflush_before)
			drm_clflush_virt_range(vaddr + offset, len);
		ret = __copy_from_user_inatomic(vaddr + offset, user_data, len);
		if (needs_clflush_after)
			drm_clflush_virt_range(vaddr + offset, len);

		kunmap_atomic(vaddr);
	}
	if (ret == 0)
		return ret;

	return shmem_pwrite_slow(page, offset, len, user_data,
				 page_do_bit17_swizzling,
				 needs_clflush_before,
				 needs_clflush_after);
}

static int
i915_gem_shmem_pwrite(struct drm_i915_gem_object *obj,
		      const struct drm_i915_gem_pwrite *args)
{
	struct drm_i915_private *i915 = to_i915(obj->base.dev);
	void __user *user_data;
	u64 remain;
	unsigned int obj_do_bit17_swizzling;
	unsigned int partial_cacheline_write;
	unsigned int needs_clflush;
	unsigned int offset, idx;
	int ret;

	ret = mutex_lock_interruptible(&i915->drm.struct_mutex);
	if (ret)
		return ret;

	ret = i915_gem_obj_prepare_shmem_write(obj, &needs_clflush);
	mutex_unlock(&i915->drm.struct_mutex);
	if (ret)
		return ret;

	obj_do_bit17_swizzling = 0;
	if (i915_gem_object_needs_bit17_swizzle(obj))
		obj_do_bit17_swizzling = BIT(17);

	/* If we don't overwrite a cacheline completely we need to be
	 * careful to have up-to-date data by first clflushing. Don't
	 * overcomplicate things and flush the entire patch.
	 */
	partial_cacheline_write = 0;
	if (needs_clflush & CLFLUSH_BEFORE)
		partial_cacheline_write = boot_cpu_data.x86_clflush_size - 1;

	user_data = u64_to_user_ptr(args->data_ptr);
	remain = args->size;
	offset = offset_in_page(args->offset);
	for (idx = args->offset >> PAGE_SHIFT; remain; idx++) {
		struct page *page = i915_gem_object_get_page(obj, idx);
		unsigned int length = min_t(u64, remain, PAGE_SIZE - offset);

		ret = shmem_pwrite(page, offset, length, user_data,
				   page_to_phys(page) & obj_do_bit17_swizzling,
				   (offset | length) & partial_cacheline_write,
				   needs_clflush & CLFLUSH_AFTER);
		if (ret)
			break;

		remain -= length;
		user_data += length;
		offset = 0;
	}

	intel_fb_obj_flush(obj, ORIGIN_CPU);
	i915_gem_obj_finish_shmem_access(obj);
	return ret;
}

/**
 * Writes data to the object referenced by handle.
 * @dev: drm device
 * @data: ioctl data blob
 * @file: drm file
 *
 * On error, the contents of the buffer that were to be modified are undefined.
 */
int
i915_gem_pwrite_ioctl(struct drm_device *dev, void *data,
		      struct drm_file *file)
{
	struct drm_i915_gem_pwrite *args = data;
	struct drm_i915_gem_object *obj;
	int ret;

	if (args->size == 0)
		return 0;

	if (!access_ok(u64_to_user_ptr(args->data_ptr), args->size))
		return -EFAULT;

	obj = i915_gem_object_lookup(file, args->handle);
	if (!obj)
		return -ENOENT;

	/* Bounds check destination. */
	if (range_overflows_t(u64, args->offset, args->size, obj->base.size)) {
		ret = -EINVAL;
		goto err;
	}

	/* Writes not allowed into this read-only object */
	if (i915_gem_object_is_readonly(obj)) {
		ret = -EINVAL;
		goto err;
	}

	trace_i915_gem_object_pwrite(obj, args->offset, args->size);

	ret = -ENODEV;
	if (obj->ops->pwrite)
		ret = obj->ops->pwrite(obj, args);
	if (ret != -ENODEV)
		goto err;

	ret = i915_gem_object_wait(obj,
				   I915_WAIT_INTERRUPTIBLE |
				   I915_WAIT_ALL,
				   MAX_SCHEDULE_TIMEOUT,
				   to_rps_client(file));
	if (ret)
		goto err;

	ret = i915_gem_object_pin_pages(obj);
	if (ret)
		goto err;

	ret = -EFAULT;
	/* We can only do the GTT pwrite on untiled buffers, as otherwise
	 * it would end up going through the fenced access, and we'll get
	 * different detiling behavior between reading and writing.
	 * pread/pwrite currently are reading and writing from the CPU
	 * perspective, requiring manual detiling by the client.
	 */
	if (!i915_gem_object_has_struct_page(obj) ||
	    cpu_write_needs_clflush(obj))
		/* Note that the gtt paths might fail with non-page-backed user
		 * pointers (e.g. gtt mappings when moving data between
		 * textures). Fallback to the shmem path in that case.
		 */
		ret = i915_gem_gtt_pwrite_fast(obj, args);

	if (ret == -EFAULT || ret == -ENOSPC) {
		if (obj->phys_handle)
			ret = i915_gem_phys_pwrite(obj, args, file);
		else
			ret = i915_gem_shmem_pwrite(obj, args);
	}

	i915_gem_object_unpin_pages(obj);
err:
	i915_gem_object_put(obj);
	return ret;
}

static void i915_gem_object_bump_inactive_ggtt(struct drm_i915_gem_object *obj)
{
	struct drm_i915_private *i915;
	struct list_head *list;
	struct i915_vma *vma;

	GEM_BUG_ON(!i915_gem_object_has_pinned_pages(obj));

	for_each_ggtt_vma(vma, obj) {
		if (i915_vma_is_active(vma))
			continue;

		if (!drm_mm_node_allocated(&vma->node))
			continue;

		list_move_tail(&vma->vm_link, &vma->vm->inactive_list);
	}

	i915 = to_i915(obj->base.dev);
	spin_lock(&i915->mm.obj_lock);
	list = obj->bind_count ? &i915->mm.bound_list : &i915->mm.unbound_list;
	list_move_tail(&obj->mm.link, list);
	spin_unlock(&i915->mm.obj_lock);
}

/**
 * Called when user space prepares to use an object with the CPU, either
 * through the mmap ioctl's mapping or a GTT mapping.
 * @dev: drm device
 * @data: ioctl data blob
 * @file: drm file
 */
int
i915_gem_set_domain_ioctl(struct drm_device *dev, void *data,
			  struct drm_file *file)
{
	struct drm_i915_gem_set_domain *args = data;
	struct drm_i915_gem_object *obj;
	uint32_t read_domains = args->read_domains;
	uint32_t write_domain = args->write_domain;
	int err;

	/* Only handle setting domains to types used by the CPU. */
	if ((write_domain | read_domains) & I915_GEM_GPU_DOMAINS)
		return -EINVAL;

	/* Having something in the write domain implies it's in the read
	 * domain, and only that read domain.  Enforce that in the request.
	 */
	if (write_domain != 0 && read_domains != write_domain)
		return -EINVAL;

	obj = i915_gem_object_lookup(file, args->handle);
	if (!obj)
		return -ENOENT;

	/* Try to flush the object off the GPU without holding the lock.
	 * We will repeat the flush holding the lock in the normal manner
	 * to catch cases where we are gazumped.
	 */
	err = i915_gem_object_wait(obj,
				   I915_WAIT_INTERRUPTIBLE |
				   I915_WAIT_PRIORITY |
				   (write_domain ? I915_WAIT_ALL : 0),
				   MAX_SCHEDULE_TIMEOUT,
				   to_rps_client(file));
	if (err)
		goto out;

	/*
	 * Proxy objects do not control access to the backing storage, ergo
	 * they cannot be used as a means to manipulate the cache domain
	 * tracking for that backing storage. The proxy object is always
	 * considered to be outside of any cache domain.
	 */
	if (i915_gem_object_is_proxy(obj)) {
		err = -ENXIO;
		goto out;
	}

	/*
	 * Flush and acquire obj->pages so that we are coherent through
	 * direct access in memory with previous cached writes through
	 * shmemfs and that our cache domain tracking remains valid.
	 * For example, if the obj->filp was moved to swap without us
	 * being notified and releasing the pages, we would mistakenly
	 * continue to assume that the obj remained out of the CPU cached
	 * domain.
	 */
	err = i915_gem_object_pin_pages(obj);
	if (err)
		goto out;

	err = i915_mutex_lock_interruptible(dev);
	if (err)
		goto out_unpin;

	if (read_domains & I915_GEM_DOMAIN_WC)
		err = i915_gem_object_set_to_wc_domain(obj, write_domain);
	else if (read_domains & I915_GEM_DOMAIN_GTT)
		err = i915_gem_object_set_to_gtt_domain(obj, write_domain);
	else
		err = i915_gem_object_set_to_cpu_domain(obj, write_domain);

	/* And bump the LRU for this access */
	i915_gem_object_bump_inactive_ggtt(obj);

	mutex_unlock(&dev->struct_mutex);

	if (write_domain != 0)
		intel_fb_obj_invalidate(obj,
					fb_write_origin(obj, write_domain));

out_unpin:
	i915_gem_object_unpin_pages(obj);
out:
	i915_gem_object_put(obj);
	return err;
}

/**
 * Called when user space has done writes to this buffer
 * @dev: drm device
 * @data: ioctl data blob
 * @file: drm file
 */
int
i915_gem_sw_finish_ioctl(struct drm_device *dev, void *data,
			 struct drm_file *file)
{
	struct drm_i915_gem_sw_finish *args = data;
	struct drm_i915_gem_object *obj;

	obj = i915_gem_object_lookup(file, args->handle);
	if (!obj)
		return -ENOENT;

	/*
	 * Proxy objects are barred from CPU access, so there is no
	 * need to ban sw_finish as it is a nop.
	 */

	/* Pinned buffers may be scanout, so flush the cache */
	i915_gem_object_flush_if_display(obj);
	i915_gem_object_put(obj);

	return 0;
}

static inline bool
__vma_matches(struct vm_area_struct *vma, struct file *filp,
	      unsigned long addr, unsigned long size)
{
	if (vma->vm_file != filp)
		return false;

	return vma->vm_start == addr && (vma->vm_end - vma->vm_start) == size;
}

/**
 * i915_gem_mmap_ioctl - Maps the contents of an object, returning the address
 *			 it is mapped to.
 * @dev: drm device
 * @data: ioctl data blob
 * @file: drm file
 *
 * While the mapping holds a reference on the contents of the object, it doesn't
 * imply a ref on the object itself.
 *
 * IMPORTANT:
 *
 * DRM driver writers who look a this function as an example for how to do GEM
 * mmap support, please don't implement mmap support like here. The modern way
 * to implement DRM mmap support is with an mmap offset ioctl (like
 * i915_gem_mmap_gtt) and then using the mmap syscall on the DRM fd directly.
 * That way debug tooling like valgrind will understand what's going on, hiding
 * the mmap call in a driver private ioctl will break that. The i915 driver only
 * does cpu mmaps this way because we didn't know better.
 */
int
i915_gem_mmap_ioctl(struct drm_device *dev, void *data,
		    struct drm_file *file)
{
	struct drm_i915_gem_mmap *args = data;
	struct drm_i915_gem_object *obj;
	unsigned long addr;

	if (args->flags & ~(I915_MMAP_WC))
		return -EINVAL;

	if (args->flags & I915_MMAP_WC && !boot_cpu_has(X86_FEATURE_PAT))
		return -ENODEV;

	obj = i915_gem_object_lookup(file, args->handle);
	if (!obj)
		return -ENOENT;

	/* prime objects have no backing filp to GEM mmap
	 * pages from.
	 */
	if (!obj->base.filp) {
		i915_gem_object_put(obj);
		return -ENXIO;
	}

	addr = vm_mmap(obj->base.filp, 0, args->size,
		       PROT_READ | PROT_WRITE, MAP_SHARED,
		       args->offset);
	if (args->flags & I915_MMAP_WC) {
		struct mm_struct *mm = current->mm;
		struct vm_area_struct *vma;

		if (down_write_killable(&mm->mmap_sem)) {
			i915_gem_object_put(obj);
			return -EINTR;
		}
		vma = find_vma(mm, addr);
		if (vma && __vma_matches(vma, obj->base.filp, addr, args->size))
			vma->vm_page_prot =
				pgprot_writecombine(vm_get_page_prot(vma->vm_flags));
		else
			addr = -ENOMEM;
		up_write(&mm->mmap_sem);

		/* This may race, but that's ok, it only gets set */
		WRITE_ONCE(obj->frontbuffer_ggtt_origin, ORIGIN_CPU);
	}
	i915_gem_object_put(obj);
	if (IS_ERR((void *)addr))
		return addr;

	args->addr_ptr = (uint64_t) addr;

	return 0;
}

static unsigned int tile_row_pages(const struct drm_i915_gem_object *obj)
{
	return i915_gem_object_get_tile_row_size(obj) >> PAGE_SHIFT;
}

/**
 * i915_gem_mmap_gtt_version - report the current feature set for GTT mmaps
 *
 * A history of the GTT mmap interface:
 *
 * 0 - Everything had to fit into the GTT. Both parties of a memcpy had to
 *     aligned and suitable for fencing, and still fit into the available
 *     mappable space left by the pinned display objects. A classic problem
 *     we called the page-fault-of-doom where we would ping-pong between
 *     two objects that could not fit inside the GTT and so the memcpy
 *     would page one object in at the expense of the other between every
 *     single byte.
 *
 * 1 - Objects can be any size, and have any compatible fencing (X Y, or none
 *     as set via i915_gem_set_tiling() [DRM_I915_GEM_SET_TILING]). If the
 *     object is too large for the available space (or simply too large
 *     for the mappable aperture!), a view is created instead and faulted
 *     into userspace. (This view is aligned and sized appropriately for
 *     fenced access.)
 *
 * 2 - Recognise WC as a separate cache domain so that we can flush the
 *     delayed writes via GTT before performing direct access via WC.
 *
 * Restrictions:
 *
 *  * snoopable objects cannot be accessed via the GTT. It can cause machine
 *    hangs on some architectures, corruption on others. An attempt to service
 *    a GTT page fault from a snoopable object will generate a SIGBUS.
 *
 *  * the object must be able to fit into RAM (physical memory, though no
 *    limited to the mappable aperture).
 *
 *
 * Caveats:
 *
 *  * a new GTT page fault will synchronize rendering from the GPU and flush
 *    all data to system memory. Subsequent access will not be synchronized.
 *
 *  * all mappings are revoked on runtime device suspend.
 *
 *  * there are only 8, 16 or 32 fence registers to share between all users
 *    (older machines require fence register for display and blitter access
 *    as well). Contention of the fence registers will cause the previous users
 *    to be unmapped and any new access will generate new page faults.
 *
 *  * running out of memory while servicing a fault may generate a SIGBUS,
 *    rather than the expected SIGSEGV.
 */
int i915_gem_mmap_gtt_version(void)
{
	return 2;
}

static inline struct i915_ggtt_view
compute_partial_view(const struct drm_i915_gem_object *obj,
		     pgoff_t page_offset,
		     unsigned int chunk)
{
	struct i915_ggtt_view view;

	if (i915_gem_object_is_tiled(obj))
		chunk = roundup(chunk, tile_row_pages(obj));

	view.type = I915_GGTT_VIEW_PARTIAL;
	view.partial.offset = rounddown(page_offset, chunk);
	view.partial.size =
		min_t(unsigned int, chunk,
		      (obj->base.size >> PAGE_SHIFT) - view.partial.offset);

	/* If the partial covers the entire object, just create a normal VMA. */
	if (chunk >= obj->base.size >> PAGE_SHIFT)
		view.type = I915_GGTT_VIEW_NORMAL;

	return view;
}

/**
 * i915_gem_fault - fault a page into the GTT
 * @vmf: fault info
 *
 * The fault handler is set up by drm_gem_mmap() when a object is GTT mapped
 * from userspace.  The fault handler takes care of binding the object to
 * the GTT (if needed), allocating and programming a fence register (again,
 * only if needed based on whether the old reg is still valid or the object
 * is tiled) and inserting a new PTE into the faulting process.
 *
 * Note that the faulting process may involve evicting existing objects
 * from the GTT and/or fence registers to make room.  So performance may
 * suffer if the GTT working set is large or there are few fence registers
 * left.
 *
 * The current feature set supported by i915_gem_fault() and thus GTT mmaps
 * is exposed via I915_PARAM_MMAP_GTT_VERSION (see i915_gem_mmap_gtt_version).
 */
vm_fault_t i915_gem_fault(struct vm_fault *vmf)
{
#define MIN_CHUNK_PAGES (SZ_1M >> PAGE_SHIFT)
	struct vm_area_struct *area = vmf->vma;
	struct drm_i915_gem_object *obj = to_intel_bo(area->vm_private_data);
	struct drm_device *dev = obj->base.dev;
	struct drm_i915_private *dev_priv = to_i915(dev);
	struct i915_ggtt *ggtt = &dev_priv->ggtt;
	bool write = area->vm_flags & VM_WRITE;
	struct i915_vma *vma;
	pgoff_t page_offset;
	int ret;

	/* Sanity check that we allow writing into this object */
	if (i915_gem_object_is_readonly(obj) && write)
		return VM_FAULT_SIGBUS;

	/* We don't use vmf->pgoff since that has the fake offset */
	page_offset = (vmf->address - area->vm_start) >> PAGE_SHIFT;

	trace_i915_gem_object_fault(obj, page_offset, true, write);

	/* Try to flush the object off the GPU first without holding the lock.
	 * Upon acquiring the lock, we will perform our sanity checks and then
	 * repeat the flush holding the lock in the normal manner to catch cases
	 * where we are gazumped.
	 */
	ret = i915_gem_object_wait(obj,
				   I915_WAIT_INTERRUPTIBLE,
				   MAX_SCHEDULE_TIMEOUT,
				   NULL);
	if (ret)
		goto err;

	ret = i915_gem_object_pin_pages(obj);
	if (ret)
		goto err;

	intel_runtime_pm_get(dev_priv);

	ret = i915_mutex_lock_interruptible(dev);
	if (ret)
		goto err_rpm;

	/* Access to snoopable pages through the GTT is incoherent. */
	if (obj->cache_level != I915_CACHE_NONE && !HAS_LLC(dev_priv)) {
		ret = -EFAULT;
		goto err_unlock;
	}


	/* Now pin it into the GTT as needed */
	vma = i915_gem_object_ggtt_pin(obj, NULL, 0, 0,
				       PIN_MAPPABLE |
				       PIN_NONBLOCK |
				       PIN_NONFAULT);
	if (IS_ERR(vma)) {
		/* Use a partial view if it is bigger than available space */
		struct i915_ggtt_view view =
			compute_partial_view(obj, page_offset, MIN_CHUNK_PAGES);
		unsigned int flags;

		flags = PIN_MAPPABLE;
		if (view.type == I915_GGTT_VIEW_NORMAL)
			flags |= PIN_NONBLOCK; /* avoid warnings for pinned */

		/*
		 * Userspace is now writing through an untracked VMA, abandon
		 * all hope that the hardware is able to track future writes.
		 */
		obj->frontbuffer_ggtt_origin = ORIGIN_CPU;

		vma = i915_gem_object_ggtt_pin(obj, &view, 0, 0, flags);
		if (IS_ERR(vma) && !view.type) {
			flags = PIN_MAPPABLE;
			view.type = I915_GGTT_VIEW_PARTIAL;
			vma = i915_gem_object_ggtt_pin(obj, &view, 0, 0, flags);
		}
	}
	if (IS_ERR(vma)) {
		ret = PTR_ERR(vma);
		goto err_unlock;
	}

	ret = i915_gem_object_set_to_gtt_domain(obj, write);
	if (ret)
		goto err_unpin;

	ret = i915_vma_pin_fence(vma);
	if (ret)
		goto err_unpin;

	/* Finally, remap it using the new GTT offset */
	ret = remap_io_mapping(area,
			       area->vm_start + (vma->ggtt_view.partial.offset << PAGE_SHIFT),
			       (ggtt->gmadr.start + vma->node.start) >> PAGE_SHIFT,
			       min_t(u64, vma->size, area->vm_end - area->vm_start),
			       &ggtt->iomap);
	if (ret)
		goto err_fence;

	/* Mark as being mmapped into userspace for later revocation */
	assert_rpm_wakelock_held(dev_priv);
	if (!i915_vma_set_userfault(vma) && !obj->userfault_count++)
		list_add(&obj->userfault_link, &dev_priv->mm.userfault_list);
	GEM_BUG_ON(!obj->userfault_count);

	i915_vma_set_ggtt_write(vma);

err_fence:
	i915_vma_unpin_fence(vma);
err_unpin:
	__i915_vma_unpin(vma);
err_unlock:
	mutex_unlock(&dev->struct_mutex);
err_rpm:
	intel_runtime_pm_put(dev_priv);
	i915_gem_object_unpin_pages(obj);
err:
	switch (ret) {
	case -EIO:
		/*
		 * We eat errors when the gpu is terminally wedged to avoid
		 * userspace unduly crashing (gl has no provisions for mmaps to
		 * fail). But any other -EIO isn't ours (e.g. swap in failure)
		 * and so needs to be reported.
		 */
		if (!i915_terminally_wedged(&dev_priv->gpu_error))
			return VM_FAULT_SIGBUS;
		/* else: fall through */
	case -EAGAIN:
		/*
		 * EAGAIN means the gpu is hung and we'll wait for the error
		 * handler to reset everything when re-faulting in
		 * i915_mutex_lock_interruptible.
		 */
	case 0:
	case -ERESTARTSYS:
	case -EINTR:
	case -EBUSY:
		/*
		 * EBUSY is ok: this just means that another thread
		 * already did the job.
		 */
		return VM_FAULT_NOPAGE;
	case -ENOMEM:
		return VM_FAULT_OOM;
	case -ENOSPC:
	case -EFAULT:
		return VM_FAULT_SIGBUS;
	default:
		WARN_ONCE(ret, "unhandled error in i915_gem_fault: %i\n", ret);
		return VM_FAULT_SIGBUS;
	}
}

static void __i915_gem_object_release_mmap(struct drm_i915_gem_object *obj)
{
	struct i915_vma *vma;

	GEM_BUG_ON(!obj->userfault_count);

	obj->userfault_count = 0;
	list_del(&obj->userfault_link);
	drm_vma_node_unmap(&obj->base.vma_node,
			   obj->base.dev->anon_inode->i_mapping);

	for_each_ggtt_vma(vma, obj)
		i915_vma_unset_userfault(vma);
}

/**
 * i915_gem_release_mmap - remove physical page mappings
 * @obj: obj in question
 *
 * Preserve the reservation of the mmapping with the DRM core code, but
 * relinquish ownership of the pages back to the system.
 *
 * It is vital that we remove the page mapping if we have mapped a tiled
 * object through the GTT and then lose the fence register due to
 * resource pressure. Similarly if the object has been moved out of the
 * aperture, than pages mapped into userspace must be revoked. Removing the
 * mapping will then trigger a page fault on the next user access, allowing
 * fixup by i915_gem_fault().
 */
void
i915_gem_release_mmap(struct drm_i915_gem_object *obj)
{
	struct drm_i915_private *i915 = to_i915(obj->base.dev);

	/* Serialisation between user GTT access and our code depends upon
	 * revoking the CPU's PTE whilst the mutex is held. The next user
	 * pagefault then has to wait until we release the mutex.
	 *
	 * Note that RPM complicates somewhat by adding an additional
	 * requirement that operations to the GGTT be made holding the RPM
	 * wakeref.
	 */
	lockdep_assert_held(&i915->drm.struct_mutex);
	intel_runtime_pm_get(i915);

	if (!obj->userfault_count)
		goto out;

	__i915_gem_object_release_mmap(obj);

	/* Ensure that the CPU's PTE are revoked and there are not outstanding
	 * memory transactions from userspace before we return. The TLB
	 * flushing implied above by changing the PTE above *should* be
	 * sufficient, an extra barrier here just provides us with a bit
	 * of paranoid documentation about our requirement to serialise
	 * memory writes before touching registers / GSM.
	 */
	wmb();

out:
	intel_runtime_pm_put(i915);
}

void i915_gem_runtime_suspend(struct drm_i915_private *dev_priv)
{
	struct drm_i915_gem_object *obj, *on;
	int i;

	/*
	 * Only called during RPM suspend. All users of the userfault_list
	 * must be holding an RPM wakeref to ensure that this can not
	 * run concurrently with themselves (and use the struct_mutex for
	 * protection between themselves).
	 */

	list_for_each_entry_safe(obj, on,
				 &dev_priv->mm.userfault_list, userfault_link)
		__i915_gem_object_release_mmap(obj);

	/* The fence will be lost when the device powers down. If any were
	 * in use by hardware (i.e. they are pinned), we should not be powering
	 * down! All other fences will be reacquired by the user upon waking.
	 */
	for (i = 0; i < dev_priv->num_fence_regs; i++) {
		struct drm_i915_fence_reg *reg = &dev_priv->fence_regs[i];

		/* Ideally we want to assert that the fence register is not
		 * live at this point (i.e. that no piece of code will be
		 * trying to write through fence + GTT, as that both violates
		 * our tracking of activity and associated locking/barriers,
		 * but also is illegal given that the hw is powered down).
		 *
		 * Previously we used reg->pin_count as a "liveness" indicator.
		 * That is not sufficient, and we need a more fine-grained
		 * tool if we want to have a sanity check here.
		 */

		if (!reg->vma)
			continue;

		GEM_BUG_ON(i915_vma_has_userfault(reg->vma));
		reg->dirty = true;
	}
}

static int i915_gem_object_create_mmap_offset(struct drm_i915_gem_object *obj)
{
	struct drm_i915_private *dev_priv = to_i915(obj->base.dev);
	int err;

	err = drm_gem_create_mmap_offset(&obj->base);
	if (likely(!err))
		return 0;

	/* Attempt to reap some mmap space from dead objects */
	do {
		err = i915_gem_wait_for_idle(dev_priv,
					     I915_WAIT_INTERRUPTIBLE,
					     MAX_SCHEDULE_TIMEOUT);
		if (err)
			break;

		i915_gem_drain_freed_objects(dev_priv);
		err = drm_gem_create_mmap_offset(&obj->base);
		if (!err)
			break;

	} while (flush_delayed_work(&dev_priv->gt.retire_work));

	return err;
}

static void i915_gem_object_free_mmap_offset(struct drm_i915_gem_object *obj)
{
	drm_gem_free_mmap_offset(&obj->base);
}

int
i915_gem_mmap_gtt(struct drm_file *file,
		  struct drm_device *dev,
		  uint32_t handle,
		  uint64_t *offset)
{
	struct drm_i915_gem_object *obj;
	int ret;

	obj = i915_gem_object_lookup(file, handle);
	if (!obj)
		return -ENOENT;

	ret = i915_gem_object_create_mmap_offset(obj);
	if (ret == 0)
		*offset = drm_vma_node_offset_addr(&obj->base.vma_node);

	i915_gem_object_put(obj);
	return ret;
}

/**
 * i915_gem_mmap_gtt_ioctl - prepare an object for GTT mmap'ing
 * @dev: DRM device
 * @data: GTT mapping ioctl data
 * @file: GEM object info
 *
 * Simply returns the fake offset to userspace so it can mmap it.
 * The mmap call will end up in drm_gem_mmap(), which will set things
 * up so we can get faults in the handler above.
 *
 * The fault handler will take care of binding the object into the GTT
 * (since it may have been evicted to make room for something), allocating
 * a fence register, and mapping the appropriate aperture address into
 * userspace.
 */
int
i915_gem_mmap_gtt_ioctl(struct drm_device *dev, void *data,
			struct drm_file *file)
{
	struct drm_i915_gem_mmap_gtt *args = data;

	return i915_gem_mmap_gtt(file, dev, args->handle, &args->offset);
}

/* Immediately discard the backing storage */
static void
i915_gem_object_truncate(struct drm_i915_gem_object *obj)
{
	i915_gem_object_free_mmap_offset(obj);

	if (obj->base.filp == NULL)
		return;

	/* Our goal here is to return as much of the memory as
	 * is possible back to the system as we are called from OOM.
	 * To do this we must instruct the shmfs to drop all of its
	 * backing pages, *now*.
	 */
	shmem_truncate_range(file_inode(obj->base.filp), 0, (loff_t)-1);
	obj->mm.madv = __I915_MADV_PURGED;
	obj->mm.pages = ERR_PTR(-EFAULT);
}

/* Try to discard unwanted pages */
void __i915_gem_object_invalidate(struct drm_i915_gem_object *obj)
{
	struct address_space *mapping;

	lockdep_assert_held(&obj->mm.lock);
	GEM_BUG_ON(i915_gem_object_has_pages(obj));

	switch (obj->mm.madv) {
	case I915_MADV_DONTNEED:
		i915_gem_object_truncate(obj);
	case __I915_MADV_PURGED:
		return;
	}

	if (obj->base.filp == NULL)
		return;

	mapping = obj->base.filp->f_mapping,
	invalidate_mapping_pages(mapping, 0, (loff_t)-1);
}

/*
 * Move pages to appropriate lru and release the pagevec, decrementing the
 * ref count of those pages.
 */
static void check_release_pagevec(struct pagevec *pvec)
{
	check_move_unevictable_pages(pvec);
	__pagevec_release(pvec);
	cond_resched();
}

static void
i915_gem_object_put_pages_gtt(struct drm_i915_gem_object *obj,
			      struct sg_table *pages)
{
	struct sgt_iter sgt_iter;
	struct pagevec pvec;
	struct page *page;

	__i915_gem_object_release_shmem(obj, pages, true);

	i915_gem_gtt_finish_pages(obj, pages);

	if (i915_gem_object_needs_bit17_swizzle(obj))
		i915_gem_object_save_bit_17_swizzle(obj, pages);

	mapping_clear_unevictable(file_inode(obj->base.filp)->i_mapping);

	pagevec_init(&pvec);
	for_each_sgt_page(page, sgt_iter, pages) {
		if (obj->mm.dirty)
			set_page_dirty(page);

		if (obj->mm.madv == I915_MADV_WILLNEED)
			mark_page_accessed(page);

		if (!pagevec_add(&pvec, page))
			check_release_pagevec(&pvec);
	}
	if (pagevec_count(&pvec))
		check_release_pagevec(&pvec);
	obj->mm.dirty = false;

	sg_free_table(pages);
	kfree(pages);
}

static void __i915_gem_object_reset_page_iter(struct drm_i915_gem_object *obj)
{
	struct radix_tree_iter iter;
	void __rcu **slot;

	rcu_read_lock();
	radix_tree_for_each_slot(slot, &obj->mm.get_page.radix, &iter, 0)
		radix_tree_delete(&obj->mm.get_page.radix, iter.index);
	rcu_read_unlock();
}

static struct sg_table *
__i915_gem_object_unset_pages(struct drm_i915_gem_object *obj)
{
	struct drm_i915_private *i915 = to_i915(obj->base.dev);
	struct sg_table *pages;

	pages = fetch_and_zero(&obj->mm.pages);
	if (!pages)
		return NULL;

	spin_lock(&i915->mm.obj_lock);
	list_del(&obj->mm.link);
	spin_unlock(&i915->mm.obj_lock);

	if (obj->mm.mapping) {
		void *ptr;

		ptr = page_mask_bits(obj->mm.mapping);
		if (is_vmalloc_addr(ptr))
			vunmap(ptr);
		else
			kunmap(kmap_to_page(ptr));

		obj->mm.mapping = NULL;
	}

	__i915_gem_object_reset_page_iter(obj);
	obj->mm.page_sizes.phys = obj->mm.page_sizes.sg = 0;

	return pages;
}

void __i915_gem_object_put_pages(struct drm_i915_gem_object *obj,
				 enum i915_mm_subclass subclass)
{
	struct sg_table *pages;

	if (i915_gem_object_has_pinned_pages(obj))
		return;

	GEM_BUG_ON(obj->bind_count);
	if (!i915_gem_object_has_pages(obj))
		return;

	/* May be called by shrinker from within get_pages() (on another bo) */
	mutex_lock_nested(&obj->mm.lock, subclass);
	if (unlikely(atomic_read(&obj->mm.pages_pin_count)))
		goto unlock;

	/*
	 * ->put_pages might need to allocate memory for the bit17 swizzle
	 * array, hence protect them from being reaped by removing them from gtt
	 * lists early.
	 */
	pages = __i915_gem_object_unset_pages(obj);
	if (!IS_ERR(pages))
		obj->ops->put_pages(obj, pages);

unlock:
	mutex_unlock(&obj->mm.lock);
}

bool i915_sg_trim(struct sg_table *orig_st)
{
	struct sg_table new_st;
	struct scatterlist *sg, *new_sg;
	unsigned int i;

	if (orig_st->nents == orig_st->orig_nents)
		return false;

	if (sg_alloc_table(&new_st, orig_st->nents, GFP_KERNEL | __GFP_NOWARN))
		return false;

	new_sg = new_st.sgl;
	for_each_sg(orig_st->sgl, sg, orig_st->nents, i) {
		sg_set_page(new_sg, sg_page(sg), sg->length, 0);
		sg_dma_address(new_sg) = sg_dma_address(sg);
		sg_dma_len(new_sg) = sg_dma_len(sg);

		new_sg = sg_next(new_sg);
	}
	GEM_BUG_ON(new_sg); /* Should walk exactly nents and hit the end */

	sg_free_table(orig_st);

	*orig_st = new_st;
	return true;
}

static int i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
{
	struct drm_i915_private *dev_priv = to_i915(obj->base.dev);
	const unsigned long page_count = obj->base.size / PAGE_SIZE;
	unsigned long i;
	struct address_space *mapping;
	struct sg_table *st;
	struct scatterlist *sg;
	struct sgt_iter sgt_iter;
	struct page *page;
	unsigned long last_pfn = 0;	/* suppress gcc warning */
	unsigned int max_segment = i915_sg_segment_size();
	unsigned int sg_page_sizes;
	struct pagevec pvec;
	gfp_t noreclaim;
	int ret;

	/*
	 * Assert that the object is not currently in any GPU domain. As it
	 * wasn't in the GTT, there shouldn't be any way it could have been in
	 * a GPU cache
	 */
	GEM_BUG_ON(obj->read_domains & I915_GEM_GPU_DOMAINS);
	GEM_BUG_ON(obj->write_domain & I915_GEM_GPU_DOMAINS);

	/*
	 * If there's no chance of allocating enough pages for the whole
	 * object, bail early.
	 */
	if (page_count > totalram_pages())
		return -ENOMEM;

	st = kmalloc(sizeof(*st), GFP_KERNEL);
	if (st == NULL)
		return -ENOMEM;

rebuild_st:
	if (sg_alloc_table(st, page_count, GFP_KERNEL)) {
		kfree(st);
		return -ENOMEM;
	}

	/*
	 * Get the list of pages out of our struct file.  They'll be pinned
	 * at this point until we release them.
	 *
	 * Fail silently without starting the shrinker
	 */
	mapping = obj->base.filp->f_mapping;
	mapping_set_unevictable(mapping);
	noreclaim = mapping_gfp_constraint(mapping, ~__GFP_RECLAIM);
	noreclaim |= __GFP_NORETRY | __GFP_NOWARN;

	sg = st->sgl;
	st->nents = 0;
	sg_page_sizes = 0;
	for (i = 0; i < page_count; i++) {
		const unsigned int shrink[] = {
			I915_SHRINK_BOUND | I915_SHRINK_UNBOUND | I915_SHRINK_PURGEABLE,
			0,
		}, *s = shrink;
		gfp_t gfp = noreclaim;

		do {
			cond_resched();
			page = shmem_read_mapping_page_gfp(mapping, i, gfp);
			if (likely(!IS_ERR(page)))
				break;

			if (!*s) {
				ret = PTR_ERR(page);
				goto err_sg;
			}

			i915_gem_shrink(dev_priv, 2 * page_count, NULL, *s++);

			/*
			 * We've tried hard to allocate the memory by reaping
			 * our own buffer, now let the real VM do its job and
			 * go down in flames if truly OOM.
			 *
			 * However, since graphics tend to be disposable,
			 * defer the oom here by reporting the ENOMEM back
			 * to userspace.
			 */
			if (!*s) {
				/* reclaim and warn, but no oom */
				gfp = mapping_gfp_mask(mapping);

				/*
				 * Our bo are always dirty and so we require
				 * kswapd to reclaim our pages (direct reclaim
				 * does not effectively begin pageout of our
				 * buffers on its own). However, direct reclaim
				 * only waits for kswapd when under allocation
				 * congestion. So as a result __GFP_RECLAIM is
				 * unreliable and fails to actually reclaim our
				 * dirty pages -- unless you try over and over
				 * again with !__GFP_NORETRY. However, we still
				 * want to fail this allocation rather than
				 * trigger the out-of-memory killer and for
				 * this we want __GFP_RETRY_MAYFAIL.
				 */
				gfp |= __GFP_RETRY_MAYFAIL;
			}
		} while (1);

		if (!i ||
		    sg->length >= max_segment ||
		    page_to_pfn(page) != last_pfn + 1) {
			if (i) {
				sg_page_sizes |= sg->length;
				sg = sg_next(sg);
			}
			st->nents++;
			sg_set_page(sg, page, PAGE_SIZE, 0);
		} else {
			sg->length += PAGE_SIZE;
		}
		last_pfn = page_to_pfn(page);

		/* Check that the i965g/gm workaround works. */
		WARN_ON((gfp & __GFP_DMA32) && (last_pfn >= 0x00100000UL));
	}
	if (sg) { /* loop terminated early; short sg table */
		sg_page_sizes |= sg->length;
		sg_mark_end(sg);
	}

	/* Trim unused sg entries to avoid wasting memory. */
	i915_sg_trim(st);

	ret = i915_gem_gtt_prepare_pages(obj, st);
	if (ret) {
		/*
		 * DMA remapping failed? One possible cause is that
		 * it could not reserve enough large entries, asking
		 * for PAGE_SIZE chunks instead may be helpful.
		 */
		if (max_segment > PAGE_SIZE) {
			for_each_sgt_page(page, sgt_iter, st)
				put_page(page);
			sg_free_table(st);

			max_segment = PAGE_SIZE;
			goto rebuild_st;
		} else {
			dev_warn(&dev_priv->drm.pdev->dev,
				 "Failed to DMA remap %lu pages\n",
				 page_count);
			goto err_pages;
		}
	}

	if (i915_gem_object_needs_bit17_swizzle(obj))
		i915_gem_object_do_bit_17_swizzle(obj, st);

	__i915_gem_object_set_pages(obj, st, sg_page_sizes);

	return 0;

err_sg:
	sg_mark_end(sg);
err_pages:
	mapping_clear_unevictable(mapping);
	pagevec_init(&pvec);
	for_each_sgt_page(page, sgt_iter, st) {
		if (!pagevec_add(&pvec, page))
			check_release_pagevec(&pvec);
	}
	if (pagevec_count(&pvec))
		check_release_pagevec(&pvec);
	sg_free_table(st);
	kfree(st);

	/*
	 * shmemfs first checks if there is enough memory to allocate the page
	 * and reports ENOSPC should there be insufficient, along with the usual
	 * ENOMEM for a genuine allocation failure.
	 *
	 * We use ENOSPC in our driver to mean that we have run out of aperture
	 * space and so want to translate the error from shmemfs back to our
	 * usual understanding of ENOMEM.
	 */
	if (ret == -ENOSPC)
		ret = -ENOMEM;

	return ret;
}

void __i915_gem_object_set_pages(struct drm_i915_gem_object *obj,
				 struct sg_table *pages,
				 unsigned int sg_page_sizes)
{
	struct drm_i915_private *i915 = to_i915(obj->base.dev);
	unsigned long supported = INTEL_INFO(i915)->page_sizes;
	int i;

	lockdep_assert_held(&obj->mm.lock);

	obj->mm.get_page.sg_pos = pages->sgl;
	obj->mm.get_page.sg_idx = 0;

	obj->mm.pages = pages;

	if (i915_gem_object_is_tiled(obj) &&
	    i915->quirks & QUIRK_PIN_SWIZZLED_PAGES) {
		GEM_BUG_ON(obj->mm.quirked);
		__i915_gem_object_pin_pages(obj);
		obj->mm.quirked = true;
	}

	GEM_BUG_ON(!sg_page_sizes);
	obj->mm.page_sizes.phys = sg_page_sizes;

	/*
	 * Calculate the supported page-sizes which fit into the given
	 * sg_page_sizes. This will give us the page-sizes which we may be able
	 * to use opportunistically when later inserting into the GTT. For
	 * example if phys=2G, then in theory we should be able to use 1G, 2M,
	 * 64K or 4K pages, although in practice this will depend on a number of
	 * other factors.
	 */
	obj->mm.page_sizes.sg = 0;
	for_each_set_bit(i, &supported, ilog2(I915_GTT_MAX_PAGE_SIZE) + 1) {
		if (obj->mm.page_sizes.phys & ~0u << i)
			obj->mm.page_sizes.sg |= BIT(i);
	}
	GEM_BUG_ON(!HAS_PAGE_SIZES(i915, obj->mm.page_sizes.sg));

	spin_lock(&i915->mm.obj_lock);
	list_add(&obj->mm.link, &i915->mm.unbound_list);
	spin_unlock(&i915->mm.obj_lock);
}

static int ____i915_gem_object_get_pages(struct drm_i915_gem_object *obj)
{
	int err;

	if (unlikely(obj->mm.madv != I915_MADV_WILLNEED)) {
		DRM_DEBUG("Attempting to obtain a purgeable object\n");
		return -EFAULT;
	}

	err = obj->ops->get_pages(obj);
	GEM_BUG_ON(!err && !i915_gem_object_has_pages(obj));

	return err;
}

/* Ensure that the associated pages are gathered from the backing storage
 * and pinned into our object. i915_gem_object_pin_pages() may be called
 * multiple times before they are released by a single call to
 * i915_gem_object_unpin_pages() - once the pages are no longer referenced
 * either as a result of memory pressure (reaping pages under the shrinker)
 * or as the object is itself released.
 */
int __i915_gem_object_get_pages(struct drm_i915_gem_object *obj)
{
	int err;

	err = mutex_lock_interruptible(&obj->mm.lock);
	if (err)
		return err;

	if (unlikely(!i915_gem_object_has_pages(obj))) {
		GEM_BUG_ON(i915_gem_object_has_pinned_pages(obj));

		err = ____i915_gem_object_get_pages(obj);
		if (err)
			goto unlock;

		smp_mb__before_atomic();
	}
	atomic_inc(&obj->mm.pages_pin_count);

unlock:
	mutex_unlock(&obj->mm.lock);
	return err;
}

/* The 'mapping' part of i915_gem_object_pin_map() below */
static void *i915_gem_object_map(const struct drm_i915_gem_object *obj,
				 enum i915_map_type type)
{
	unsigned long n_pages = obj->base.size >> PAGE_SHIFT;
	struct sg_table *sgt = obj->mm.pages;
	struct sgt_iter sgt_iter;
	struct page *page;
	struct page *stack_pages[32];
	struct page **pages = stack_pages;
	unsigned long i = 0;
	pgprot_t pgprot;
	void *addr;

	/* A single page can always be kmapped */
	if (n_pages == 1 && type == I915_MAP_WB)
		return kmap(sg_page(sgt->sgl));

	if (n_pages > ARRAY_SIZE(stack_pages)) {
		/* Too big for stack -- allocate temporary array instead */
		pages = kvmalloc_array(n_pages, sizeof(*pages), GFP_KERNEL);
		if (!pages)
			return NULL;
	}

	for_each_sgt_page(page, sgt_iter, sgt)
		pages[i++] = page;

	/* Check that we have the expected number of pages */
	GEM_BUG_ON(i != n_pages);

	switch (type) {
	default:
		MISSING_CASE(type);
		/* fallthrough to use PAGE_KERNEL anyway */
	case I915_MAP_WB:
		pgprot = PAGE_KERNEL;
		break;
	case I915_MAP_WC:
		pgprot = pgprot_writecombine(PAGE_KERNEL_IO);
		break;
	}
	addr = vmap(pages, n_pages, 0, pgprot);

	if (pages != stack_pages)
		kvfree(pages);

	return addr;
}

/* get, pin, and map the pages of the object into kernel space */
void *i915_gem_object_pin_map(struct drm_i915_gem_object *obj,
			      enum i915_map_type type)
{
	enum i915_map_type has_type;
	bool pinned;
	void *ptr;
	int ret;

	if (unlikely(!i915_gem_object_has_struct_page(obj)))
		return ERR_PTR(-ENXIO);

	ret = mutex_lock_interruptible(&obj->mm.lock);
	if (ret)
		return ERR_PTR(ret);

	pinned = !(type & I915_MAP_OVERRIDE);
	type &= ~I915_MAP_OVERRIDE;

	if (!atomic_inc_not_zero(&obj->mm.pages_pin_count)) {
		if (unlikely(!i915_gem_object_has_pages(obj))) {
			GEM_BUG_ON(i915_gem_object_has_pinned_pages(obj));

			ret = ____i915_gem_object_get_pages(obj);
			if (ret)
				goto err_unlock;

			smp_mb__before_atomic();
		}
		atomic_inc(&obj->mm.pages_pin_count);
		pinned = false;
	}
	GEM_BUG_ON(!i915_gem_object_has_pages(obj));

	ptr = page_unpack_bits(obj->mm.mapping, &has_type);
	if (ptr && has_type != type) {
		if (pinned) {
			ret = -EBUSY;
			goto err_unpin;
		}

		if (is_vmalloc_addr(ptr))
			vunmap(ptr);
		else
			kunmap(kmap_to_page(ptr));

		ptr = obj->mm.mapping = NULL;
	}

	if (!ptr) {
		ptr = i915_gem_object_map(obj, type);
		if (!ptr) {
			ret = -ENOMEM;
			goto err_unpin;
		}

		obj->mm.mapping = page_pack_bits(ptr, type);
	}

out_unlock:
	mutex_unlock(&obj->mm.lock);
	return ptr;

err_unpin:
	atomic_dec(&obj->mm.pages_pin_count);
err_unlock:
	ptr = ERR_PTR(ret);
	goto out_unlock;
}

static int
i915_gem_object_pwrite_gtt(struct drm_i915_gem_object *obj,
			   const struct drm_i915_gem_pwrite *arg)
{
	struct address_space *mapping = obj->base.filp->f_mapping;
	char __user *user_data = u64_to_user_ptr(arg->data_ptr);
	u64 remain, offset;
	unsigned int pg;

	/* Before we instantiate/pin the backing store for our use, we
	 * can prepopulate the shmemfs filp efficiently using a write into
	 * the pagecache. We avoid the penalty of instantiating all the
	 * pages, important if the user is just writing to a few and never
	 * uses the object on the GPU, and using a direct write into shmemfs
	 * allows it to avoid the cost of retrieving a page (either swapin
	 * or clearing-before-use) before it is overwritten.
	 */
	if (i915_gem_object_has_pages(obj))
		return -ENODEV;

	if (obj->mm.madv != I915_MADV_WILLNEED)
		return -EFAULT;

	/* Before the pages are instantiated the object is treated as being
	 * in the CPU domain. The pages will be clflushed as required before
	 * use, and we can freely write into the pages directly. If userspace
	 * races pwrite with any other operation; corruption will ensue -
	 * that is userspace's prerogative!
	 */

	remain = arg->size;
	offset = arg->offset;
	pg = offset_in_page(offset);

	do {
		unsigned int len, unwritten;
		struct page *page;
		void *data, *vaddr;
		int err;

		len = PAGE_SIZE - pg;
		if (len > remain)
			len = remain;

		err = pagecache_write_begin(obj->base.filp, mapping,
					    offset, len, 0,
					    &page, &data);
		if (err < 0)
			return err;

		vaddr = kmap(page);
		unwritten = copy_from_user(vaddr + pg, user_data, len);
		kunmap(page);

		err = pagecache_write_end(obj->base.filp, mapping,
					  offset, len, len - unwritten,
					  page, data);
		if (err < 0)
			return err;

		if (unwritten)
			return -EFAULT;

		remain -= len;
		user_data += len;
		offset += len;
		pg = 0;
	} while (remain);

	return 0;
}

static void i915_gem_client_mark_guilty(struct drm_i915_file_private *file_priv,
					const struct i915_gem_context *ctx)
{
	unsigned int score;
	unsigned long prev_hang;

	if (i915_gem_context_is_banned(ctx))
		score = I915_CLIENT_SCORE_CONTEXT_BAN;
	else
		score = 0;

	prev_hang = xchg(&file_priv->hang_timestamp, jiffies);
	if (time_before(jiffies, prev_hang + I915_CLIENT_FAST_HANG_JIFFIES))
		score += I915_CLIENT_SCORE_HANG_FAST;

	if (score) {
		atomic_add(score, &file_priv->ban_score);

		DRM_DEBUG_DRIVER("client %s: gained %u ban score, now %u\n",
				 ctx->name, score,
				 atomic_read(&file_priv->ban_score));
	}
}

static void i915_gem_context_mark_guilty(struct i915_gem_context *ctx)
{
	unsigned int score;
	bool banned, bannable;

	atomic_inc(&ctx->guilty_count);

	bannable = i915_gem_context_is_bannable(ctx);
	score = atomic_add_return(CONTEXT_SCORE_GUILTY, &ctx->ban_score);
	banned = score >= CONTEXT_SCORE_BAN_THRESHOLD;

	/* Cool contexts don't accumulate client ban score */
	if (!bannable)
		return;

	if (banned) {
		DRM_DEBUG_DRIVER("context %s: guilty %d, score %u, banned\n",
				 ctx->name, atomic_read(&ctx->guilty_count),
				 score);
		i915_gem_context_set_banned(ctx);
	}

	if (!IS_ERR_OR_NULL(ctx->file_priv))
		i915_gem_client_mark_guilty(ctx->file_priv, ctx);
}

static void i915_gem_context_mark_innocent(struct i915_gem_context *ctx)
{
	atomic_inc(&ctx->active_count);
}

struct i915_request *
i915_gem_find_active_request(struct intel_engine_cs *engine)
{
	struct i915_request *request, *active = NULL;
	unsigned long flags;

	/*
	 * We are called by the error capture, reset and to dump engine
	 * state at random points in time. In particular, note that neither is
	 * crucially ordered with an interrupt. After a hang, the GPU is dead
	 * and we assume that no more writes can happen (we waited long enough
	 * for all writes that were in transaction to be flushed) - adding an
	 * extra delay for a recent interrupt is pointless. Hence, we do
	 * not need an engine->irq_seqno_barrier() before the seqno reads.
	 * At all other times, we must assume the GPU is still running, but
	 * we only care about the snapshot of this moment.
	 */
	spin_lock_irqsave(&engine->timeline.lock, flags);
	list_for_each_entry(request, &engine->timeline.requests, link) {
		if (__i915_request_completed(request, request->global_seqno))
			continue;

		active = request;
		break;
	}
	spin_unlock_irqrestore(&engine->timeline.lock, flags);

	return active;
}

/*
 * Ensure irq handler finishes, and not run again.
 * Also return the active request so that we only search for it once.
 */
struct i915_request *
i915_gem_reset_prepare_engine(struct intel_engine_cs *engine)
{
	struct i915_request *request;

	/*
	 * During the reset sequence, we must prevent the engine from
	 * entering RC6. As the context state is undefined until we restart
	 * the engine, if it does enter RC6 during the reset, the state
	 * written to the powercontext is undefined and so we may lose
	 * GPU state upon resume, i.e. fail to restart after a reset.
	 */
	intel_uncore_forcewake_get(engine->i915, FORCEWAKE_ALL);

	request = engine->reset.prepare(engine);
	if (request && request->fence.error == -EIO)
		request = ERR_PTR(-EIO); /* Previous reset failed! */

	return request;
}

int i915_gem_reset_prepare(struct drm_i915_private *dev_priv)
{
	struct intel_engine_cs *engine;
	struct i915_request *request;
	enum intel_engine_id id;
	int err = 0;

	for_each_engine(engine, dev_priv, id) {
		request = i915_gem_reset_prepare_engine(engine);
		if (IS_ERR(request)) {
			err = PTR_ERR(request);
			continue;
		}

		engine->hangcheck.active_request = request;
	}

	i915_gem_revoke_fences(dev_priv);
	intel_uc_sanitize(dev_priv);

	return err;
}

static void engine_skip_context(struct i915_request *request)
{
	struct intel_engine_cs *engine = request->engine;
	struct i915_gem_context *hung_ctx = request->gem_context;
	struct i915_timeline *timeline = request->timeline;
	unsigned long flags;

	GEM_BUG_ON(timeline == &engine->timeline);

	spin_lock_irqsave(&engine->timeline.lock, flags);
	spin_lock(&timeline->lock);

	list_for_each_entry_continue(request, &engine->timeline.requests, link)
		if (request->gem_context == hung_ctx)
			i915_request_skip(request, -EIO);

	list_for_each_entry(request, &timeline->requests, link)
		i915_request_skip(request, -EIO);

	spin_unlock(&timeline->lock);
	spin_unlock_irqrestore(&engine->timeline.lock, flags);
}

/* Returns the request if it was guilty of the hang */
static struct i915_request *
i915_gem_reset_request(struct intel_engine_cs *engine,
		       struct i915_request *request,
		       bool stalled)
{
	/* The guilty request will get skipped on a hung engine.
	 *
	 * Users of client default contexts do not rely on logical
	 * state preserved between batches so it is safe to execute
	 * queued requests following the hang. Non default contexts
	 * rely on preserved state, so skipping a batch loses the
	 * evolution of the state and it needs to be considered corrupted.
	 * Executing more queued batches on top of corrupted state is
	 * risky. But we take the risk by trying to advance through
	 * the queued requests in order to make the client behaviour
	 * more predictable around resets, by not throwing away random
	 * amount of batches it has prepared for execution. Sophisticated
	 * clients can use gem_reset_stats_ioctl and dma fence status
	 * (exported via sync_file info ioctl on explicit fences) to observe
	 * when it loses the context state and should rebuild accordingly.
	 *
	 * The context ban, and ultimately the client ban, mechanism are safety
	 * valves if client submission ends up resulting in nothing more than
	 * subsequent hangs.
	 */

	if (i915_request_completed(request)) {
		GEM_TRACE("%s pardoned global=%d (fence %llx:%d), current %d\n",
			  engine->name, request->global_seqno,
			  request->fence.context, request->fence.seqno,
			  intel_engine_get_seqno(engine));
		stalled = false;
	}

	if (stalled) {
		i915_gem_context_mark_guilty(request->gem_context);
		i915_request_skip(request, -EIO);

		/* If this context is now banned, skip all pending requests. */
		if (i915_gem_context_is_banned(request->gem_context))
			engine_skip_context(request);
	} else {
		/*
		 * Since this is not the hung engine, it may have advanced
		 * since the hang declaration. Double check by refinding
		 * the active request at the time of the reset.
		 */
		request = i915_gem_find_active_request(engine);
		if (request) {
			unsigned long flags;

			i915_gem_context_mark_innocent(request->gem_context);
			dma_fence_set_error(&request->fence, -EAGAIN);

			/* Rewind the engine to replay the incomplete rq */
			spin_lock_irqsave(&engine->timeline.lock, flags);
			request = list_prev_entry(request, link);
			if (&request->link == &engine->timeline.requests)
				request = NULL;
			spin_unlock_irqrestore(&engine->timeline.lock, flags);
		}
	}

	return request;
}

void i915_gem_reset_engine(struct intel_engine_cs *engine,
			   struct i915_request *request,
			   bool stalled)
{
	/*
	 * Make sure this write is visible before we re-enable the interrupt
	 * handlers on another CPU, as tasklet_enable() resolves to just
	 * a compiler barrier which is insufficient for our purpose here.
	 */
	smp_store_mb(engine->irq_posted, 0);

	if (request)
		request = i915_gem_reset_request(engine, request, stalled);

	/* Setup the CS to resume from the breadcrumb of the hung request */
	engine->reset.reset(engine, request);
}

void i915_gem_reset(struct drm_i915_private *dev_priv,
		    unsigned int stalled_mask)
{
	struct intel_engine_cs *engine;
	enum intel_engine_id id;

	lockdep_assert_held(&dev_priv->drm.struct_mutex);

	i915_retire_requests(dev_priv);

	for_each_engine(engine, dev_priv, id) {
		struct intel_context *ce;

		i915_gem_reset_engine(engine,
				      engine->hangcheck.active_request,
				      stalled_mask & ENGINE_MASK(id));
		ce = fetch_and_zero(&engine->last_retired_context);
		if (ce)
			intel_context_unpin(ce);

		/*
		 * Ostensibily, we always want a context loaded for powersaving,
		 * so if the engine is idle after the reset, send a request
		 * to load our scratch kernel_context.
		 *
		 * More mysteriously, if we leave the engine idle after a reset,
		 * the next userspace batch may hang, with what appears to be
		 * an incoherent read by the CS (presumably stale TLB). An
		 * empty request appears sufficient to paper over the glitch.
		 */
		if (intel_engine_is_idle(engine)) {
			struct i915_request *rq;

			rq = i915_request_alloc(engine,
						dev_priv->kernel_context);
			if (!IS_ERR(rq))
				i915_request_add(rq);
		}
	}

	i915_gem_restore_fences(dev_priv);
}

void i915_gem_reset_finish_engine(struct intel_engine_cs *engine)
{
	engine->reset.finish(engine);

	intel_uncore_forcewake_put(engine->i915, FORCEWAKE_ALL);
}

void i915_gem_reset_finish(struct drm_i915_private *dev_priv)
{
	struct intel_engine_cs *engine;
	enum intel_engine_id id;

	lockdep_assert_held(&dev_priv->drm.struct_mutex);

	for_each_engine(engine, dev_priv, id) {
		engine->hangcheck.active_request = NULL;
		i915_gem_reset_finish_engine(engine);
	}
}

static void nop_submit_request(struct i915_request *request)
{
	unsigned long flags;

	GEM_TRACE("%s fence %llx:%d -> -EIO\n",
		  request->engine->name,
		  request->fence.context, request->fence.seqno);
	dma_fence_set_error(&request->fence, -EIO);

	spin_lock_irqsave(&request->engine->timeline.lock, flags);
	__i915_request_submit(request);
	intel_engine_init_global_seqno(request->engine, request->global_seqno);
	spin_unlock_irqrestore(&request->engine->timeline.lock, flags);
}

void i915_gem_set_wedged(struct drm_i915_private *i915)
{
	struct intel_engine_cs *engine;
	enum intel_engine_id id;

	GEM_TRACE("start\n");

	if (GEM_SHOW_DEBUG()) {
		struct drm_printer p = drm_debug_printer(__func__);

		for_each_engine(engine, i915, id)
			intel_engine_dump(engine, &p, "%s\n", engine->name);
	}

	if (test_and_set_bit(I915_WEDGED, &i915->gpu_error.flags))
		goto out;

	/*
	 * First, stop submission to hw, but do not yet complete requests by
	 * rolling the global seqno forward (since this would complete requests
	 * for which we haven't set the fence error to EIO yet).
	 */
	for_each_engine(engine, i915, id)
		i915_gem_reset_prepare_engine(engine);

	/* Even if the GPU reset fails, it should still stop the engines */
	if (INTEL_GEN(i915) >= 5)
		intel_gpu_reset(i915, ALL_ENGINES);

	for_each_engine(engine, i915, id) {
		engine->submit_request = nop_submit_request;
		engine->schedule = NULL;
	}
	i915->caps.scheduler = 0;

	/*
	 * Make sure no request can slip through without getting completed by
	 * either this call here to intel_engine_init_global_seqno, or the one
	 * in nop_submit_request.
	 */
	synchronize_rcu();

	/* Mark all executing requests as skipped */
	for_each_engine(engine, i915, id)
		engine->cancel_requests(engine);

	for_each_engine(engine, i915, id) {
		i915_gem_reset_finish_engine(engine);
		intel_engine_wakeup(engine);
	}

out:
	GEM_TRACE("end\n");

	wake_up_all(&i915->gpu_error.reset_queue);
}

bool i915_gem_unset_wedged(struct drm_i915_private *i915)
{
	struct i915_timeline *tl;

	lockdep_assert_held(&i915->drm.struct_mutex);
	if (!test_bit(I915_WEDGED, &i915->gpu_error.flags))
		return true;

	GEM_TRACE("start\n");

	/*
	 * Before unwedging, make sure that all pending operations
	 * are flushed and errored out - we may have requests waiting upon
	 * third party fences. We marked all inflight requests as EIO, and
	 * every execbuf since returned EIO, for consistency we want all
	 * the currently pending requests to also be marked as EIO, which
	 * is done inside our nop_submit_request - and so we must wait.
	 *
	 * No more can be submitted until we reset the wedged bit.
	 */
	list_for_each_entry(tl, &i915->gt.timelines, link) {
		struct i915_request *rq;

		rq = i915_gem_active_peek(&tl->last_request,
					  &i915->drm.struct_mutex);
		if (!rq)
			continue;

		/*
		 * We can't use our normal waiter as we want to
		 * avoid recursively trying to handle the current
		 * reset. The basic dma_fence_default_wait() installs
		 * a callback for dma_fence_signal(), which is
		 * triggered by our nop handler (indirectly, the
		 * callback enables the signaler thread which is
		 * woken by the nop_submit_request() advancing the seqno
		 * and when the seqno passes the fence, the signaler
		 * then signals the fence waking us up).
		 */
		if (dma_fence_default_wait(&rq->fence, true,
					   MAX_SCHEDULE_TIMEOUT) < 0)
			return false;
	}
	i915_retire_requests(i915);
	GEM_BUG_ON(i915->gt.active_requests);

	if (!intel_gpu_reset(i915, ALL_ENGINES))
		intel_engines_sanitize(i915);

	/*
	 * Undo nop_submit_request. We prevent all new i915 requests from
	 * being queued (by disallowing execbuf whilst wedged) so having
	 * waited for all active requests above, we know the system is idle
	 * and do not have to worry about a thread being inside
	 * engine->submit_request() as we swap over. So unlike installing
	 * the nop_submit_request on reset, we can do this from normal
	 * context and do not require stop_machine().
	 */
	intel_engines_reset_default_submission(i915);
	i915_gem_contexts_lost(i915);

	GEM_TRACE("end\n");

	smp_mb__before_atomic(); /* complete takeover before enabling execbuf */
	clear_bit(I915_WEDGED, &i915->gpu_error.flags);

	return true;
}

static void
i915_gem_retire_work_handler(struct work_struct *work)
{
	struct drm_i915_private *dev_priv =
		container_of(work, typeof(*dev_priv), gt.retire_work.work);
	struct drm_device *dev = &dev_priv->drm;

	/* Come back later if the device is busy... */
	if (mutex_trylock(&dev->struct_mutex)) {
		i915_retire_requests(dev_priv);
		mutex_unlock(&dev->struct_mutex);
	}

	/*
	 * Keep the retire handler running until we are finally idle.
	 * We do not need to do this test under locking as in the worst-case
	 * we queue the retire worker once too often.
	 */
	if (READ_ONCE(dev_priv->gt.awake))
		queue_delayed_work(dev_priv->wq,
				   &dev_priv->gt.retire_work,
				   round_jiffies_up_relative(HZ));
}

static void shrink_caches(struct drm_i915_private *i915)
{
	/*
	 * kmem_cache_shrink() discards empty slabs and reorders partially
	 * filled slabs to prioritise allocating from the mostly full slabs,
	 * with the aim of reducing fragmentation.
	 */
	kmem_cache_shrink(i915->priorities);
	kmem_cache_shrink(i915->dependencies);
	kmem_cache_shrink(i915->requests);
	kmem_cache_shrink(i915->luts);
	kmem_cache_shrink(i915->vmas);
	kmem_cache_shrink(i915->objects);
}

struct sleep_rcu_work {
	union {
		struct rcu_head rcu;
		struct work_struct work;
	};
	struct drm_i915_private *i915;
	unsigned int epoch;
};

static inline bool
same_epoch(struct drm_i915_private *i915, unsigned int epoch)
{
	/*
	 * There is a small chance that the epoch wrapped since we started
	 * sleeping. If we assume that epoch is at least a u32, then it will
	 * take at least 2^32 * 100ms for it to wrap, or about 326 years.
	 */
	return epoch == READ_ONCE(i915->gt.epoch);
}

static void __sleep_work(struct work_struct *work)
{
	struct sleep_rcu_work *s = container_of(work, typeof(*s), work);
	struct drm_i915_private *i915 = s->i915;
	unsigned int epoch = s->epoch;

	kfree(s);
	if (same_epoch(i915, epoch))
		shrink_caches(i915);
}

static void __sleep_rcu(struct rcu_head *rcu)
{
	struct sleep_rcu_work *s = container_of(rcu, typeof(*s), rcu);
	struct drm_i915_private *i915 = s->i915;

	destroy_rcu_head(&s->rcu);

	if (same_epoch(i915, s->epoch)) {
		INIT_WORK(&s->work, __sleep_work);
		queue_work(i915->wq, &s->work);
	} else {
		kfree(s);
	}
}

static inline bool
new_requests_since_last_retire(const struct drm_i915_private *i915)
{
	return (READ_ONCE(i915->gt.active_requests) ||
		work_pending(&i915->gt.idle_work.work));
}

static void assert_kernel_context_is_current(struct drm_i915_private *i915)
{
	struct intel_engine_cs *engine;
	enum intel_engine_id id;

	if (i915_terminally_wedged(&i915->gpu_error))
		return;

	GEM_BUG_ON(i915->gt.active_requests);
	for_each_engine(engine, i915, id) {
		GEM_BUG_ON(__i915_gem_active_peek(&engine->timeline.last_request));
		GEM_BUG_ON(engine->last_retired_context !=
			   to_intel_context(i915->kernel_context, engine));
	}
}

static void
i915_gem_idle_work_handler(struct work_struct *work)
{
	struct drm_i915_private *dev_priv =
		container_of(work, typeof(*dev_priv), gt.idle_work.work);
	unsigned int epoch = I915_EPOCH_INVALID;
	bool rearm_hangcheck;

	if (!READ_ONCE(dev_priv->gt.awake))
		return;

	if (READ_ONCE(dev_priv->gt.active_requests))
		return;

	/*
	 * Flush out the last user context, leaving only the pinned
	 * kernel context resident. When we are idling on the kernel_context,
	 * no more new requests (with a context switch) are emitted and we
	 * can finally rest. A consequence is that the idle work handler is
	 * always called at least twice before idling (and if the system is
	 * idle that implies a round trip through the retire worker).
	 */
	mutex_lock(&dev_priv->drm.struct_mutex);
	i915_gem_switch_to_kernel_context(dev_priv);
	mutex_unlock(&dev_priv->drm.struct_mutex);

	GEM_TRACE("active_requests=%d (after switch-to-kernel-context)\n",
		  READ_ONCE(dev_priv->gt.active_requests));

	/*
	 * Wait for last execlists context complete, but bail out in case a
	 * new request is submitted. As we don't trust the hardware, we
	 * continue on if the wait times out. This is necessary to allow
	 * the machine to suspend even if the hardware dies, and we will
	 * try to recover in resume (after depriving the hardware of power,
	 * it may be in a better mmod).
	 */
	__wait_for(if (new_requests_since_last_retire(dev_priv)) return,
		   intel_engines_are_idle(dev_priv),
		   I915_IDLE_ENGINES_TIMEOUT * 1000,
		   10, 500);

	rearm_hangcheck =
		cancel_delayed_work_sync(&dev_priv->gpu_error.hangcheck_work);

	if (!mutex_trylock(&dev_priv->drm.struct_mutex)) {
		/* Currently busy, come back later */
		mod_delayed_work(dev_priv->wq,
				 &dev_priv->gt.idle_work,
				 msecs_to_jiffies(50));
		goto out_rearm;
	}

	/*
	 * New request retired after this work handler started, extend active
	 * period until next instance of the work.
	 */
	if (new_requests_since_last_retire(dev_priv))
		goto out_unlock;

	epoch = __i915_gem_park(dev_priv);

	assert_kernel_context_is_current(dev_priv);

	rearm_hangcheck = false;
out_unlock:
	mutex_unlock(&dev_priv->drm.struct_mutex);

out_rearm:
	if (rearm_hangcheck) {
		GEM_BUG_ON(!dev_priv->gt.awake);
		i915_queue_hangcheck(dev_priv);
	}

	/*
	 * When we are idle, it is an opportune time to reap our caches.
	 * However, we have many objects that utilise RCU and the ordered
	 * i915->wq that this work is executing on. To try and flush any
	 * pending frees now we are idle, we first wait for an RCU grace
	 * period, and then queue a task (that will run last on the wq) to
	 * shrink and re-optimize the caches.
	 */
	if (same_epoch(dev_priv, epoch)) {
		struct sleep_rcu_work *s = kmalloc(sizeof(*s), GFP_KERNEL);
		if (s) {
			init_rcu_head(&s->rcu);
			s->i915 = dev_priv;
			s->epoch = epoch;
			call_rcu(&s->rcu, __sleep_rcu);
		}
	}
}

void i915_gem_close_object(struct drm_gem_object *gem, struct drm_file *file)
{
	struct drm_i915_private *i915 = to_i915(gem->dev);
	struct drm_i915_gem_object *obj = to_intel_bo(gem);
	struct drm_i915_file_private *fpriv = file->driver_priv;
	struct i915_lut_handle *lut, *ln;

	mutex_lock(&i915->drm.struct_mutex);

	list_for_each_entry_safe(lut, ln, &obj->lut_list, obj_link) {
		struct i915_gem_context *ctx = lut->ctx;
		struct i915_vma *vma;

		GEM_BUG_ON(ctx->file_priv == ERR_PTR(-EBADF));
		if (ctx->file_priv != fpriv)
			continue;

		vma = radix_tree_delete(&ctx->handles_vma, lut->handle);
		GEM_BUG_ON(vma->obj != obj);

		/* We allow the process to have multiple handles to the same
		 * vma, in the same fd namespace, by virtue of flink/open.
		 */
		GEM_BUG_ON(!vma->open_count);
		if (!--vma->open_count && !i915_vma_is_ggtt(vma))
			i915_vma_close(vma);

		list_del(&lut->obj_link);
		list_del(&lut->ctx_link);

		kmem_cache_free(i915->luts, lut);
		__i915_gem_object_release_unless_active(obj);
	}

	mutex_unlock(&i915->drm.struct_mutex);
}

static unsigned long to_wait_timeout(s64 timeout_ns)
{
	if (timeout_ns < 0)
		return MAX_SCHEDULE_TIMEOUT;

	if (timeout_ns == 0)
		return 0;

	return nsecs_to_jiffies_timeout(timeout_ns);
}

/**
 * i915_gem_wait_ioctl - implements DRM_IOCTL_I915_GEM_WAIT
 * @dev: drm device pointer
 * @data: ioctl data blob
 * @file: drm file pointer
 *
 * Returns 0 if successful, else an error is returned with the remaining time in
 * the timeout parameter.
 *  -ETIME: object is still busy after timeout
 *  -ERESTARTSYS: signal interrupted the wait
 *  -ENONENT: object doesn't exist
 * Also possible, but rare:
 *  -EAGAIN: incomplete, restart syscall
 *  -ENOMEM: damn
 *  -ENODEV: Internal IRQ fail
 *  -E?: The add request failed
 *
 * The wait ioctl with a timeout of 0 reimplements the busy ioctl. With any
 * non-zero timeout parameter the wait ioctl will wait for the given number of
 * nanoseconds on an object becoming unbusy. Since the wait itself does so
 * without holding struct_mutex the object may become re-busied before this
 * function completes. A similar but shorter * race condition exists in the busy
 * ioctl
 */
int
i915_gem_wait_ioctl(struct drm_device *dev, void *data, struct drm_file *file)
{
	struct drm_i915_gem_wait *args = data;
	struct drm_i915_gem_object *obj;
	ktime_t start;
	long ret;

	if (args->flags != 0)
		return -EINVAL;

	obj = i915_gem_object_lookup(file, args->bo_handle);
	if (!obj)
		return -ENOENT;

	start = ktime_get();

	ret = i915_gem_object_wait(obj,
				   I915_WAIT_INTERRUPTIBLE |
				   I915_WAIT_PRIORITY |
				   I915_WAIT_ALL,
				   to_wait_timeout(args->timeout_ns),
				   to_rps_client(file));

	if (args->timeout_ns > 0) {
		args->timeout_ns -= ktime_to_ns(ktime_sub(ktime_get(), start));
		if (args->timeout_ns < 0)
			args->timeout_ns = 0;

		/*
		 * Apparently ktime isn't accurate enough and occasionally has a
		 * bit of mismatch in the jiffies<->nsecs<->ktime loop. So patch
		 * things up to make the test happy. We allow up to 1 jiffy.
		 *
		 * This is a regression from the timespec->ktime conversion.
		 */
		if (ret == -ETIME && !nsecs_to_jiffies(args->timeout_ns))
			args->timeout_ns = 0;

		/* Asked to wait beyond the jiffie/scheduler precision? */
		if (ret == -ETIME && args->timeout_ns)
			ret = -EAGAIN;
	}

	i915_gem_object_put(obj);
	return ret;
}

static long wait_for_timeline(struct i915_timeline *tl,
			      unsigned int flags, long timeout)
{
	struct i915_request *rq;

	rq = i915_gem_active_get_unlocked(&tl->last_request);
	if (!rq)
		return timeout;

	/*
	 * "Race-to-idle".
	 *
	 * Switching to the kernel context is often used a synchronous
	 * step prior to idling, e.g. in suspend for flushing all
	 * current operations to memory before sleeping. These we
	 * want to complete as quickly as possible to avoid prolonged
	 * stalls, so allow the gpu to boost to maximum clocks.
	 */
	if (flags & I915_WAIT_FOR_IDLE_BOOST)
		gen6_rps_boost(rq, NULL);

	timeout = i915_request_wait(rq, flags, timeout);
	i915_request_put(rq);

	return timeout;
}

static int wait_for_engines(struct drm_i915_private *i915)
{
	if (wait_for(intel_engines_are_idle(i915), I915_IDLE_ENGINES_TIMEOUT)) {
		dev_err(i915->drm.dev,
			"Failed to idle engines, declaring wedged!\n");
		GEM_TRACE_DUMP();
		i915_gem_set_wedged(i915);
		return -EIO;
	}

	return 0;
}

int i915_gem_wait_for_idle(struct drm_i915_private *i915,
			   unsigned int flags, long timeout)
{
	GEM_TRACE("flags=%x (%s), timeout=%ld%s\n",
		  flags, flags & I915_WAIT_LOCKED ? "locked" : "unlocked",
		  timeout, timeout == MAX_SCHEDULE_TIMEOUT ? " (forever)" : "");

	/* If the device is asleep, we have no requests outstanding */
	if (!READ_ONCE(i915->gt.awake))
		return 0;

	if (flags & I915_WAIT_LOCKED) {
		struct i915_timeline *tl;
		int err;

		lockdep_assert_held(&i915->drm.struct_mutex);

		list_for_each_entry(tl, &i915->gt.timelines, link) {
			timeout = wait_for_timeline(tl, flags, timeout);
			if (timeout < 0)
				return timeout;
		}
		if (GEM_SHOW_DEBUG() && !timeout) {
			/* Presume that timeout was non-zero to begin with! */
			dev_warn(&i915->drm.pdev->dev,
				 "Missed idle-completion interrupt!\n");
			GEM_TRACE_DUMP();
		}

		err = wait_for_engines(i915);
		if (err)
			return err;

		i915_retire_requests(i915);
		GEM_BUG_ON(i915->gt.active_requests);
	} else {
		struct intel_engine_cs *engine;
		enum intel_engine_id id;

		for_each_engine(engine, i915, id) {
			struct i915_timeline *tl = &engine->timeline;

			timeout = wait_for_timeline(tl, flags, timeout);
			if (timeout < 0)
				return timeout;
		}
	}

	return 0;
}

static void __i915_gem_object_flush_for_display(struct drm_i915_gem_object *obj)
{
	/*
	 * We manually flush the CPU domain so that we can override and
	 * force the flush for the display, and perform it asyncrhonously.
	 */
	flush_write_domain(obj, ~I915_GEM_DOMAIN_CPU);
	if (obj->cache_dirty)
		i915_gem_clflush_object(obj, I915_CLFLUSH_FORCE);
	obj->write_domain = 0;
}

void i915_gem_object_flush_if_display(struct drm_i915_gem_object *obj)
{
	if (!READ_ONCE(obj->pin_global))
		return;

	mutex_lock(&obj->base.dev->struct_mutex);
	__i915_gem_object_flush_for_display(obj);
	mutex_unlock(&obj->base.dev->struct_mutex);
}

/**
 * Moves a single object to the WC read, and possibly write domain.
 * @obj: object to act on
 * @write: ask for write access or read only
 *
 * This function returns when the move is complete, including waiting on
 * flushes to occur.
 */
int
i915_gem_object_set_to_wc_domain(struct drm_i915_gem_object *obj, bool write)
{
	int ret;

	lockdep_assert_held(&obj->base.dev->struct_mutex);

	ret = i915_gem_object_wait(obj,
				   I915_WAIT_INTERRUPTIBLE |
				   I915_WAIT_LOCKED |
				   (write ? I915_WAIT_ALL : 0),
				   MAX_SCHEDULE_TIMEOUT,
				   NULL);
	if (ret)
		return ret;

	if (obj->write_domain == I915_GEM_DOMAIN_WC)
		return 0;

	/* Flush and acquire obj->pages so that we are coherent through
	 * direct access in memory with previous cached writes through
	 * shmemfs and that our cache domain tracking remains valid.
	 * For example, if the obj->filp was moved to swap without us
	 * being notified and releasing the pages, we would mistakenly
	 * continue to assume that the obj remained out of the CPU cached
	 * domain.
	 */
	ret = i915_gem_object_pin_pages(obj);
	if (ret)
		return ret;

	flush_write_domain(obj, ~I915_GEM_DOMAIN_WC);

	/* Serialise direct access to this object with the barriers for
	 * coherent writes from the GPU, by effectively invalidating the
	 * WC domain upon first access.
	 */
	if ((obj->read_domains & I915_GEM_DOMAIN_WC) == 0)
		mb();

	/* It should now be out of any other write domains, and we can update
	 * the domain values for our changes.
	 */
	GEM_BUG_ON((obj->write_domain & ~I915_GEM_DOMAIN_WC) != 0);
	obj->read_domains |= I915_GEM_DOMAIN_WC;
	if (write) {
		obj->read_domains = I915_GEM_DOMAIN_WC;
		obj->write_domain = I915_GEM_DOMAIN_WC;
		obj->mm.dirty = true;
	}

	i915_gem_object_unpin_pages(obj);
	return 0;
}

/**
 * Moves a single object to the GTT read, and possibly write domain.
 * @obj: object to act on
 * @write: ask for write access or read only
 *
 * This function returns when the move is complete, including waiting on
 * flushes to occur.
 */
int
i915_gem_object_set_to_gtt_domain(struct drm_i915_gem_object *obj, bool write)
{
	int ret;

	lockdep_assert_held(&obj->base.dev->struct_mutex);

	ret = i915_gem_object_wait(obj,
				   I915_WAIT_INTERRUPTIBLE |
				   I915_WAIT_LOCKED |
				   (write ? I915_WAIT_ALL : 0),
				   MAX_SCHEDULE_TIMEOUT,
				   NULL);
	if (ret)
		return ret;

	if (obj->write_domain == I915_GEM_DOMAIN_GTT)
		return 0;

	/* Flush and acquire obj->pages so that we are coherent through
	 * direct access in memory with previous cached writes through
	 * shmemfs and that our cache domain tracking remains valid.
	 * For example, if the obj->filp was moved to swap without us
	 * being notified and releasing the pages, we would mistakenly
	 * continue to assume that the obj remained out of the CPU cached
	 * domain.
	 */
	ret = i915_gem_object_pin_pages(obj);
	if (ret)
		return ret;

	flush_write_domain(obj, ~I915_GEM_DOMAIN_GTT);

	/* Serialise direct access to this object with the barriers for
	 * coherent writes from the GPU, by effectively invalidating the
	 * GTT domain upon first access.
	 */
	if ((obj->read_domains & I915_GEM_DOMAIN_GTT) == 0)
		mb();

	/* It should now be out of any other write domains, and we can update
	 * the domain values for our changes.
	 */
	GEM_BUG_ON((obj->write_domain & ~I915_GEM_DOMAIN_GTT) != 0);
	obj->read_domains |= I915_GEM_DOMAIN_GTT;
	if (write) {
		obj->read_domains = I915_GEM_DOMAIN_GTT;
		obj->write_domain = I915_GEM_DOMAIN_GTT;
		obj->mm.dirty = true;
	}

	i915_gem_object_unpin_pages(obj);
	return 0;
}

/**
 * Changes the cache-level of an object across all VMA.
 * @obj: object to act on
 * @cache_level: new cache level to set for the object
 *
 * After this function returns, the object will be in the new cache-level
 * across all GTT and the contents of the backing storage will be coherent,
 * with respect to the new cache-level. In order to keep the backing storage
 * coherent for all users, we only allow a single cache level to be set
 * globally on the object and prevent it from being changed whilst the
 * hardware is reading from the object. That is if the object is currently
 * on the scanout it will be set to uncached (or equivalent display
 * cache coherency) and all non-MOCS GPU access will also be uncached so
 * that all direct access to the scanout remains coherent.
 */
int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj,
				    enum i915_cache_level cache_level)
{
	struct i915_vma *vma;
	int ret;

	lockdep_assert_held(&obj->base.dev->struct_mutex);

	if (obj->cache_level == cache_level)
		return 0;

	/* Inspect the list of currently bound VMA and unbind any that would
	 * be invalid given the new cache-level. This is principally to
	 * catch the issue of the CS prefetch crossing page boundaries and
	 * reading an invalid PTE on older architectures.
	 */
restart:
	list_for_each_entry(vma, &obj->vma_list, obj_link) {
		if (!drm_mm_node_allocated(&vma->node))
			continue;

		if (i915_vma_is_pinned(vma)) {
			DRM_DEBUG("can not change the cache level of pinned objects\n");
			return -EBUSY;
		}

		if (!i915_vma_is_closed(vma) &&
		    i915_gem_valid_gtt_space(vma, cache_level))
			continue;

		ret = i915_vma_unbind(vma);
		if (ret)
			return ret;

		/* As unbinding may affect other elements in the
		 * obj->vma_list (due to side-effects from retiring
		 * an active vma), play safe and restart the iterator.
		 */
		goto restart;
	}

	/* We can reuse the existing drm_mm nodes but need to change the
	 * cache-level on the PTE. We could simply unbind them all and
	 * rebind with the correct cache-level on next use. However since
	 * we already have a valid slot, dma mapping, pages etc, we may as
	 * rewrite the PTE in the belief that doing so tramples upon less
	 * state and so involves less work.
	 */
	if (obj->bind_count) {
		/* Before we change the PTE, the GPU must not be accessing it.
		 * If we wait upon the object, we know that all the bound
		 * VMA are no longer active.
		 */
		ret = i915_gem_object_wait(obj,
					   I915_WAIT_INTERRUPTIBLE |
					   I915_WAIT_LOCKED |
					   I915_WAIT_ALL,
					   MAX_SCHEDULE_TIMEOUT,
					   NULL);
		if (ret)
			return ret;

		if (!HAS_LLC(to_i915(obj->base.dev)) &&
		    cache_level != I915_CACHE_NONE) {
			/* Access to snoopable pages through the GTT is
			 * incoherent and on some machines causes a hard
			 * lockup. Relinquish the CPU mmaping to force
			 * userspace to refault in the pages and we can
			 * then double check if the GTT mapping is still
			 * valid for that pointer access.
			 */
			i915_gem_release_mmap(obj);

			/* As we no longer need a fence for GTT access,
			 * we can relinquish it now (and so prevent having
			 * to steal a fence from someone else on the next
			 * fence request). Note GPU activity would have
			 * dropped the fence as all snoopable access is
			 * supposed to be linear.
			 */
			for_each_ggtt_vma(vma, obj) {
				ret = i915_vma_put_fence(vma);
				if (ret)
					return ret;
			}
		} else {
			/* We either have incoherent backing store and
			 * so no GTT access or the architecture is fully
			 * coherent. In such cases, existing GTT mmaps
			 * ignore the cache bit in the PTE and we can
			 * rewrite it without confusing the GPU or having
			 * to force userspace to fault back in its mmaps.
			 */
		}

		list_for_each_entry(vma, &obj->vma_list, obj_link) {
			if (!drm_mm_node_allocated(&vma->node))
				continue;

			ret = i915_vma_bind(vma, cache_level, PIN_UPDATE);
			if (ret)
				return ret;
		}
	}

	list_for_each_entry(vma, &obj->vma_list, obj_link)
		vma->node.color = cache_level;
	i915_gem_object_set_cache_coherency(obj, cache_level);
	obj->cache_dirty = true; /* Always invalidate stale cachelines */

	return 0;
}

int i915_gem_get_caching_ioctl(struct drm_device *dev, void *data,
			       struct drm_file *file)
{
	struct drm_i915_gem_caching *args = data;
	struct drm_i915_gem_object *obj;
	int err = 0;

	rcu_read_lock();
	obj = i915_gem_object_lookup_rcu(file, args->handle);
	if (!obj) {
		err = -ENOENT;
		goto out;
	}

	switch (obj->cache_level) {
	case I915_CACHE_LLC:
	case I915_CACHE_L3_LLC:
		args->caching = I915_CACHING_CACHED;
		break;

	case I915_CACHE_WT:
		args->caching = I915_CACHING_DISPLAY;
		break;

	default:
		args->caching = I915_CACHING_NONE;
		break;
	}
out:
	rcu_read_unlock();
	return err;
}

int i915_gem_set_caching_ioctl(struct drm_device *dev, void *data,
			       struct drm_file *file)
{
	struct drm_i915_private *i915 = to_i915(dev);
	struct drm_i915_gem_caching *args = data;
	struct drm_i915_gem_object *obj;
	enum i915_cache_level level;
	int ret = 0;

	switch (args->caching) {
	case I915_CACHING_NONE:
		level = I915_CACHE_NONE;
		break;
	case I915_CACHING_CACHED:
		/*
		 * Due to a HW issue on BXT A stepping, GPU stores via a
		 * snooped mapping may leave stale data in a corresponding CPU
		 * cacheline, whereas normally such cachelines would get
		 * invalidated.
		 */
		if (!HAS_LLC(i915) && !HAS_SNOOP(i915))
			return -ENODEV;

		level = I915_CACHE_LLC;
		break;
	case I915_CACHING_DISPLAY:
		level = HAS_WT(i915) ? I915_CACHE_WT : I915_CACHE_NONE;
		break;
	default:
		return -EINVAL;
	}

	obj = i915_gem_object_lookup(file, args->handle);
	if (!obj)
		return -ENOENT;

	/*
	 * The caching mode of proxy object is handled by its generator, and
	 * not allowed to be changed by userspace.
	 */
	if (i915_gem_object_is_proxy(obj)) {
		ret = -ENXIO;
		goto out;
	}

	if (obj->cache_level == level)
		goto out;

	ret = i915_gem_object_wait(obj,
				   I915_WAIT_INTERRUPTIBLE,
				   MAX_SCHEDULE_TIMEOUT,
				   to_rps_client(file));
	if (ret)
		goto out;

	ret = i915_mutex_lock_interruptible(dev);
	if (ret)
		goto out;

	ret = i915_gem_object_set_cache_level(obj, level);
	mutex_unlock(&dev->struct_mutex);

out:
	i915_gem_object_put(obj);
	return ret;
}

/*
 * Prepare buffer for display plane (scanout, cursors, etc). Can be called from
 * an uninterruptible phase (modesetting) and allows any flushes to be pipelined
 * (for pageflips). We only flush the caches while preparing the buffer for
 * display, the callers are responsible for frontbuffer flush.
 */
struct i915_vma *
i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj,
				     u32 alignment,
				     const struct i915_ggtt_view *view,
				     unsigned int flags)
{
	struct i915_vma *vma;
	int ret;

	lockdep_assert_held(&obj->base.dev->struct_mutex);

	/* Mark the global pin early so that we account for the
	 * display coherency whilst setting up the cache domains.
	 */
	obj->pin_global++;

	/* The display engine is not coherent with the LLC cache on gen6.  As
	 * a result, we make sure that the pinning that is about to occur is
	 * done with uncached PTEs. This is lowest common denominator for all
	 * chipsets.
	 *
	 * However for gen6+, we could do better by using the GFDT bit instead
	 * of uncaching, which would allow us to flush all the LLC-cached data
	 * with that bit in the PTE to main memory with just one PIPE_CONTROL.
	 */
	ret = i915_gem_object_set_cache_level(obj,
					      HAS_WT(to_i915(obj->base.dev)) ?
					      I915_CACHE_WT : I915_CACHE_NONE);
	if (ret) {
		vma = ERR_PTR(ret);
		goto err_unpin_global;
	}

	/* As the user may map the buffer once pinned in the display plane
	 * (e.g. libkms for the bootup splash), we have to ensure that we
	 * always use map_and_fenceable for all scanout buffers. However,
	 * it may simply be too big to fit into mappable, in which case
	 * put it anyway and hope that userspace can cope (but always first
	 * try to preserve the existing ABI).
	 */
	vma = ERR_PTR(-ENOSPC);
	if ((flags & PIN_MAPPABLE) == 0 &&
	    (!view || view->type == I915_GGTT_VIEW_NORMAL))
		vma = i915_gem_object_ggtt_pin(obj, view, 0, alignment,
					       flags |
					       PIN_MAPPABLE |
					       PIN_NONBLOCK);
	if (IS_ERR(vma))
		vma = i915_gem_object_ggtt_pin(obj, view, 0, alignment, flags);
	if (IS_ERR(vma))
		goto err_unpin_global;

	vma->display_alignment = max_t(u64, vma->display_alignment, alignment);

	__i915_gem_object_flush_for_display(obj);

	/* It should now be out of any other write domains, and we can update
	 * the domain values for our changes.
	 */
	obj->read_domains |= I915_GEM_DOMAIN_GTT;

	return vma;

err_unpin_global:
	obj->pin_global--;
	return vma;
}

void
i915_gem_object_unpin_from_display_plane(struct i915_vma *vma)
{
	lockdep_assert_held(&vma->vm->i915->drm.struct_mutex);

	if (WARN_ON(vma->obj->pin_global == 0))
		return;

	if (--vma->obj->pin_global == 0)
		vma->display_alignment = I915_GTT_MIN_ALIGNMENT;

	/* Bump the LRU to try and avoid premature eviction whilst flipping  */
	i915_gem_object_bump_inactive_ggtt(vma->obj);

	i915_vma_unpin(vma);
}

/**
 * Moves a single object to the CPU read, and possibly write domain.
 * @obj: object to act on
 * @write: requesting write or read-only access
 *
 * This function returns when the move is complete, including waiting on
 * flushes to occur.
 */
int
i915_gem_object_set_to_cpu_domain(struct drm_i915_gem_object *obj, bool write)
{
	int ret;

	lockdep_assert_held(&obj->base.dev->struct_mutex);

	ret = i915_gem_object_wait(obj,
				   I915_WAIT_INTERRUPTIBLE |
				   I915_WAIT_LOCKED |
				   (write ? I915_WAIT_ALL : 0),
				   MAX_SCHEDULE_TIMEOUT,
				   NULL);
	if (ret)
		return ret;

	flush_write_domain(obj, ~I915_GEM_DOMAIN_CPU);

	/* Flush the CPU cache if it's still invalid. */
	if ((obj->read_domains & I915_GEM_DOMAIN_CPU) == 0) {
		i915_gem_clflush_object(obj, I915_CLFLUSH_SYNC);
		obj->read_domains |= I915_GEM_DOMAIN_CPU;
	}

	/* It should now be out of any other write domains, and we can update
	 * the domain values for our changes.
	 */
	GEM_BUG_ON(obj->write_domain & ~I915_GEM_DOMAIN_CPU);

	/* If we're writing through the CPU, then the GPU read domains will
	 * need to be invalidated at next use.
	 */
	if (write)
		__start_cpu_write(obj);

	return 0;
}

/* Throttle our rendering by waiting until the ring has completed our requests
 * emitted over 20 msec ago.
 *
 * Note that if we were to use the current jiffies each time around the loop,
 * we wouldn't escape the function with any frames outstanding if the time to
 * render a frame was over 20ms.
 *
 * This should get us reasonable parallelism between CPU and GPU but also
 * relatively low latency when blocking on a particular request to finish.
 */
static int
i915_gem_ring_throttle(struct drm_device *dev, struct drm_file *file)
{
	struct drm_i915_private *dev_priv = to_i915(dev);
	struct drm_i915_file_private *file_priv = file->driver_priv;
	unsigned long recent_enough = jiffies - DRM_I915_THROTTLE_JIFFIES;
	struct i915_request *request, *target = NULL;
	long ret;

	/* ABI: return -EIO if already wedged */
	if (i915_terminally_wedged(&dev_priv->gpu_error))
		return -EIO;

	spin_lock(&file_priv->mm.lock);
	list_for_each_entry(request, &file_priv->mm.request_list, client_link) {
		if (time_after_eq(request->emitted_jiffies, recent_enough))
			break;

		if (target) {
			list_del(&target->client_link);
			target->file_priv = NULL;
		}

		target = request;
	}
	if (target)
		i915_request_get(target);
	spin_unlock(&file_priv->mm.lock);

	if (target == NULL)
		return 0;

	ret = i915_request_wait(target,
				I915_WAIT_INTERRUPTIBLE,
				MAX_SCHEDULE_TIMEOUT);
	i915_request_put(target);

	return ret < 0 ? ret : 0;
}

struct i915_vma *
i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj,
			 const struct i915_ggtt_view *view,
			 u64 size,
			 u64 alignment,
			 u64 flags)
{
	struct drm_i915_private *dev_priv = to_i915(obj->base.dev);
	struct i915_address_space *vm = &dev_priv->ggtt.vm;
	struct i915_vma *vma;
	int ret;

	lockdep_assert_held(&obj->base.dev->struct_mutex);

	if (flags & PIN_MAPPABLE &&
	    (!view || view->type == I915_GGTT_VIEW_NORMAL)) {
		/* If the required space is larger than the available
		 * aperture, we will not able to find a slot for the
		 * object and unbinding the object now will be in
		 * vain. Worse, doing so may cause us to ping-pong
		 * the object in and out of the Global GTT and
		 * waste a lot of cycles under the mutex.
		 */
		if (obj->base.size > dev_priv->ggtt.mappable_end)
			return ERR_PTR(-E2BIG);

		/* If NONBLOCK is set the caller is optimistically
		 * trying to cache the full object within the mappable
		 * aperture, and *must* have a fallback in place for
		 * situations where we cannot bind the object. We
		 * can be a little more lax here and use the fallback
		 * more often to avoid costly migrations of ourselves
		 * and other objects within the aperture.
		 *
		 * Half-the-aperture is used as a simple heuristic.
		 * More interesting would to do search for a free
		 * block prior to making the commitment to unbind.
		 * That caters for the self-harm case, and with a
		 * little more heuristics (e.g. NOFAULT, NOEVICT)
		 * we could try to minimise harm to others.
		 */
		if (flags & PIN_NONBLOCK &&
		    obj->base.size > dev_priv->ggtt.mappable_end / 2)
			return ERR_PTR(-ENOSPC);
	}

	vma = i915_vma_instance(obj, vm, view);
	if (unlikely(IS_ERR(vma)))
		return vma;

	if (i915_vma_misplaced(vma, size, alignment, flags)) {
		if (flags & PIN_NONBLOCK) {
			if (i915_vma_is_pinned(vma) || i915_vma_is_active(vma))
				return ERR_PTR(-ENOSPC);

			if (flags & PIN_MAPPABLE &&
			    vma->fence_size > dev_priv->ggtt.mappable_end / 2)
				return ERR_PTR(-ENOSPC);
		}

		WARN(i915_vma_is_pinned(vma),
		     "bo is already pinned in ggtt with incorrect alignment:"
		     " offset=%08x, req.alignment=%llx,"
		     " req.map_and_fenceable=%d, vma->map_and_fenceable=%d\n",
		     i915_ggtt_offset(vma), alignment,
		     !!(flags & PIN_MAPPABLE),
		     i915_vma_is_map_and_fenceable(vma));
		ret = i915_vma_unbind(vma);
		if (ret)
			return ERR_PTR(ret);
	}

	ret = i915_vma_pin(vma, size, alignment, flags | PIN_GLOBAL);
	if (ret)
		return ERR_PTR(ret);

	return vma;
}

static __always_inline unsigned int __busy_read_flag(unsigned int id)
{
	/* Note that we could alias engines in the execbuf API, but
	 * that would be very unwise as it prevents userspace from
	 * fine control over engine selection. Ahem.
	 *
	 * This should be something like EXEC_MAX_ENGINE instead of
	 * I915_NUM_ENGINES.
	 */
	BUILD_BUG_ON(I915_NUM_ENGINES > 16);
	return 0x10000 << id;
}

static __always_inline unsigned int __busy_write_id(unsigned int id)
{
	/* The uABI guarantees an active writer is also amongst the read
	 * engines. This would be true if we accessed the activity tracking
	 * under the lock, but as we perform the lookup of the object and
	 * its activity locklessly we can not guarantee that the last_write
	 * being active implies that we have set the same engine flag from
	 * last_read - hence we always set both read and write busy for
	 * last_write.
	 */
	return id | __busy_read_flag(id);
}

static __always_inline unsigned int
__busy_set_if_active(const struct dma_fence *fence,
		     unsigned int (*flag)(unsigned int id))
{
	struct i915_request *rq;

	/* We have to check the current hw status of the fence as the uABI
	 * guarantees forward progress. We could rely on the idle worker
	 * to eventually flush us, but to minimise latency just ask the
	 * hardware.
	 *
	 * Note we only report on the status of native fences.
	 */
	if (!dma_fence_is_i915(fence))
		return 0;

	/* opencode to_request() in order to avoid const warnings */
	rq = container_of(fence, struct i915_request, fence);
	if (i915_request_completed(rq))
		return 0;

	return flag(rq->engine->uabi_id);
}

static __always_inline unsigned int
busy_check_reader(const struct dma_fence *fence)
{
	return __busy_set_if_active(fence, __busy_read_flag);
}

static __always_inline unsigned int
busy_check_writer(const struct dma_fence *fence)
{
	if (!fence)
		return 0;

	return __busy_set_if_active(fence, __busy_write_id);
}

int
i915_gem_busy_ioctl(struct drm_device *dev, void *data,
		    struct drm_file *file)
{
	struct drm_i915_gem_busy *args = data;
	struct drm_i915_gem_object *obj;
	struct reservation_object_list *list;
	unsigned int seq;
	int err;

	err = -ENOENT;
	rcu_read_lock();
	obj = i915_gem_object_lookup_rcu(file, args->handle);
	if (!obj)
		goto out;

	/* A discrepancy here is that we do not report the status of
	 * non-i915 fences, i.e. even though we may report the object as idle,
	 * a call to set-domain may still stall waiting for foreign rendering.
	 * This also means that wait-ioctl may report an object as busy,
	 * where busy-ioctl considers it idle.
	 *
	 * We trade the ability to warn of foreign fences to report on which
	 * i915 engines are active for the object.
	 *
	 * Alternatively, we can trade that extra information on read/write
	 * activity with
	 *	args->busy =
	 *		!reservation_object_test_signaled_rcu(obj->resv, true);
	 * to report the overall busyness. This is what the wait-ioctl does.
	 *
	 */
retry:
	seq = raw_read_seqcount(&obj->resv->seq);

	/* Translate the exclusive fence to the READ *and* WRITE engine */
	args->busy = busy_check_writer(rcu_dereference(obj->resv->fence_excl));

	/* Translate shared fences to READ set of engines */
	list = rcu_dereference(obj->resv->fence);
	if (list) {
		unsigned int shared_count = list->shared_count, i;

		for (i = 0; i < shared_count; ++i) {
			struct dma_fence *fence =
				rcu_dereference(list->shared[i]);

			args->busy |= busy_check_reader(fence);
		}
	}

	if (args->busy && read_seqcount_retry(&obj->resv->seq, seq))
		goto retry;

	err = 0;
out:
	rcu_read_unlock();
	return err;
}

int
i915_gem_throttle_ioctl(struct drm_device *dev, void *data,
			struct drm_file *file_priv)
{
	return i915_gem_ring_throttle(dev, file_priv);
}

int
i915_gem_madvise_ioctl(struct drm_device *dev, void *data,
		       struct drm_file *file_priv)
{
	struct drm_i915_private *dev_priv = to_i915(dev);
	struct drm_i915_gem_madvise *args = data;
	struct drm_i915_gem_object *obj;
	int err;

	switch (args->madv) {
	case I915_MADV_DONTNEED:
	case I915_MADV_WILLNEED:
	    break;
	default:
	    return -EINVAL;
	}

	obj = i915_gem_object_lookup(file_priv, args->handle);
	if (!obj)
		return -ENOENT;

	err = mutex_lock_interruptible(&obj->mm.lock);
	if (err)
		goto out;

	if (i915_gem_object_has_pages(obj) &&
	    i915_gem_object_is_tiled(obj) &&
	    dev_priv->quirks & QUIRK_PIN_SWIZZLED_PAGES) {
		if (obj->mm.madv == I915_MADV_WILLNEED) {
			GEM_BUG_ON(!obj->mm.quirked);
			__i915_gem_object_unpin_pages(obj);
			obj->mm.quirked = false;
		}
		if (args->madv == I915_MADV_WILLNEED) {
			GEM_BUG_ON(obj->mm.quirked);
			__i915_gem_object_pin_pages(obj);
			obj->mm.quirked = true;
		}
	}

	if (obj->mm.madv != __I915_MADV_PURGED)
		obj->mm.madv = args->madv;

	/* if the object is no longer attached, discard its backing storage */
	if (obj->mm.madv == I915_MADV_DONTNEED &&
	    !i915_gem_object_has_pages(obj))
		i915_gem_object_truncate(obj);

	args->retained = obj->mm.madv != __I915_MADV_PURGED;
	mutex_unlock(&obj->mm.lock);

out:
	i915_gem_object_put(obj);
	return err;
}

static void
frontbuffer_retire(struct i915_gem_active *active, struct i915_request *request)
{
	struct drm_i915_gem_object *obj =
		container_of(active, typeof(*obj), frontbuffer_write);

	intel_fb_obj_flush(obj, ORIGIN_CS);
}

void i915_gem_object_init(struct drm_i915_gem_object *obj,
			  const struct drm_i915_gem_object_ops *ops)
{
	mutex_init(&obj->mm.lock);

	INIT_LIST_HEAD(&obj->vma_list);
	INIT_LIST_HEAD(&obj->lut_list);
	INIT_LIST_HEAD(&obj->batch_pool_link);

	init_rcu_head(&obj->rcu);

	obj->ops = ops;

	reservation_object_init(&obj->__builtin_resv);
	obj->resv = &obj->__builtin_resv;

	obj->frontbuffer_ggtt_origin = ORIGIN_GTT;
	init_request_active(&obj->frontbuffer_write, frontbuffer_retire);

	obj->mm.madv = I915_MADV_WILLNEED;
	INIT_RADIX_TREE(&obj->mm.get_page.radix, GFP_KERNEL | __GFP_NOWARN);
	mutex_init(&obj->mm.get_page.lock);

	i915_gem_info_add_obj(to_i915(obj->base.dev), obj->base.size);
}

static const struct drm_i915_gem_object_ops i915_gem_object_ops = {
	.flags = I915_GEM_OBJECT_HAS_STRUCT_PAGE |
		 I915_GEM_OBJECT_IS_SHRINKABLE,

	.get_pages = i915_gem_object_get_pages_gtt,
	.put_pages = i915_gem_object_put_pages_gtt,

	.pwrite = i915_gem_object_pwrite_gtt,
};

static int i915_gem_object_create_shmem(struct drm_device *dev,
					struct drm_gem_object *obj,
					size_t size)
{
	struct drm_i915_private *i915 = to_i915(dev);
	unsigned long flags = VM_NORESERVE;
	struct file *filp;

	drm_gem_private_object_init(dev, obj, size);

	if (i915->mm.gemfs)
		filp = shmem_file_setup_with_mnt(i915->mm.gemfs, "i915", size,
						 flags);
	else
		filp = shmem_file_setup("i915", size, flags);

	if (IS_ERR(filp))
		return PTR_ERR(filp);

	obj->filp = filp;

	return 0;
}

struct drm_i915_gem_object *
i915_gem_object_create(struct drm_i915_private *dev_priv, u64 size)
{
	struct drm_i915_gem_object *obj;
	struct address_space *mapping;
	unsigned int cache_level;
	gfp_t mask;
	int ret;

	/* There is a prevalence of the assumption that we fit the object's
	 * page count inside a 32bit _signed_ variable. Let's document this and
	 * catch if we ever need to fix it. In the meantime, if you do spot
	 * such a local variable, please consider fixing!
	 */
	if (size >> PAGE_SHIFT > INT_MAX)
		return ERR_PTR(-E2BIG);

	if (overflows_type(size, obj->base.size))
		return ERR_PTR(-E2BIG);

	obj = i915_gem_object_alloc(dev_priv);
	if (obj == NULL)
		return ERR_PTR(-ENOMEM);

	ret = i915_gem_object_create_shmem(&dev_priv->drm, &obj->base, size);
	if (ret)
		goto fail;

	mask = GFP_HIGHUSER | __GFP_RECLAIMABLE;
	if (IS_I965GM(dev_priv) || IS_I965G(dev_priv)) {
		/* 965gm cannot relocate objects above 4GiB. */
		mask &= ~__GFP_HIGHMEM;
		mask |= __GFP_DMA32;
	}

	mapping = obj->base.filp->f_mapping;
	mapping_set_gfp_mask(mapping, mask);
	GEM_BUG_ON(!(mapping_gfp_mask(mapping) & __GFP_RECLAIM));

	i915_gem_object_init(obj, &i915_gem_object_ops);

	obj->write_domain = I915_GEM_DOMAIN_CPU;
	obj->read_domains = I915_GEM_DOMAIN_CPU;

	if (HAS_LLC(dev_priv))
		/* On some devices, we can have the GPU use the LLC (the CPU
		 * cache) for about a 10% performance improvement
		 * compared to uncached.  Graphics requests other than
		 * display scanout are coherent with the CPU in
		 * accessing this cache.  This means in this mode we
		 * don't need to clflush on the CPU side, and on the
		 * GPU side we only need to flush internal caches to
		 * get data visible to the CPU.
		 *
		 * However, we maintain the display planes as UC, and so
		 * need to rebind when first used as such.
		 */
		cache_level = I915_CACHE_LLC;
	else
		cache_level = I915_CACHE_NONE;

	i915_gem_object_set_cache_coherency(obj, cache_level);

	trace_i915_gem_object_create(obj);

	return obj;

fail:
	i915_gem_object_free(obj);
	return ERR_PTR(ret);
}

static bool discard_backing_storage(struct drm_i915_gem_object *obj)
{
	/* If we are the last user of the backing storage (be it shmemfs
	 * pages or stolen etc), we know that the pages are going to be
	 * immediately released. In this case, we can then skip copying
	 * back the contents from the GPU.
	 */

	if (obj->mm.madv != I915_MADV_WILLNEED)
		return false;

	if (obj->base.filp == NULL)
		return true;

	/* At first glance, this looks racy, but then again so would be
	 * userspace racing mmap against close. However, the first external
	 * reference to the filp can only be obtained through the
	 * i915_gem_mmap_ioctl() which safeguards us against the user
	 * acquiring such a reference whilst we are in the middle of
	 * freeing the object.
	 */
	return atomic_long_read(&obj->base.filp->f_count) == 1;
}

static void __i915_gem_free_objects(struct drm_i915_private *i915,
				    struct llist_node *freed)
{
	struct drm_i915_gem_object *obj, *on;

	intel_runtime_pm_get(i915);
	llist_for_each_entry_safe(obj, on, freed, freed) {
		struct i915_vma *vma, *vn;

		trace_i915_gem_object_destroy(obj);

		mutex_lock(&i915->drm.struct_mutex);

		GEM_BUG_ON(i915_gem_object_is_active(obj));
		list_for_each_entry_safe(vma, vn,
					 &obj->vma_list, obj_link) {
			GEM_BUG_ON(i915_vma_is_active(vma));
			vma->flags &= ~I915_VMA_PIN_MASK;
			i915_vma_destroy(vma);
		}
		GEM_BUG_ON(!list_empty(&obj->vma_list));
		GEM_BUG_ON(!RB_EMPTY_ROOT(&obj->vma_tree));

		/* This serializes freeing with the shrinker. Since the free
		 * is delayed, first by RCU then by the workqueue, we want the
		 * shrinker to be able to free pages of unreferenced objects,
		 * or else we may oom whilst there are plenty of deferred
		 * freed objects.
		 */
		if (i915_gem_object_has_pages(obj)) {
			spin_lock(&i915->mm.obj_lock);
			list_del_init(&obj->mm.link);
			spin_unlock(&i915->mm.obj_lock);
		}

		mutex_unlock(&i915->drm.struct_mutex);

		GEM_BUG_ON(obj->bind_count);
		GEM_BUG_ON(obj->userfault_count);
		GEM_BUG_ON(atomic_read(&obj->frontbuffer_bits));
		GEM_BUG_ON(!list_empty(&obj->lut_list));

		if (obj->ops->release)
			obj->ops->release(obj);

		if (WARN_ON(i915_gem_object_has_pinned_pages(obj)))
			atomic_set(&obj->mm.pages_pin_count, 0);
		__i915_gem_object_put_pages(obj, I915_MM_NORMAL);
		GEM_BUG_ON(i915_gem_object_has_pages(obj));

		if (obj->base.import_attach)
			drm_prime_gem_destroy(&obj->base, NULL);

		reservation_object_fini(&obj->__builtin_resv);
		drm_gem_object_release(&obj->base);
		i915_gem_info_remove_obj(i915, obj->base.size);

		kfree(obj->bit_17);
		i915_gem_object_free(obj);

		GEM_BUG_ON(!atomic_read(&i915->mm.free_count));
		atomic_dec(&i915->mm.free_count);

		if (on)
			cond_resched();
	}
	intel_runtime_pm_put(i915);
}

static void i915_gem_flush_free_objects(struct drm_i915_private *i915)
{
	struct llist_node *freed;

	/* Free the oldest, most stale object to keep the free_list short */
	freed = NULL;
	if (!llist_empty(&i915->mm.free_list)) { /* quick test for hotpath */
		/* Only one consumer of llist_del_first() allowed */
		spin_lock(&i915->mm.free_lock);
		freed = llist_del_first(&i915->mm.free_list);
		spin_unlock(&i915->mm.free_lock);
	}
	if (unlikely(freed)) {
		freed->next = NULL;
		__i915_gem_free_objects(i915, freed);
	}
}

static void __i915_gem_free_work(struct work_struct *work)
{
	struct drm_i915_private *i915 =
		container_of(work, struct drm_i915_private, mm.free_work);
	struct llist_node *freed;

	/*
	 * All file-owned VMA should have been released by this point through
	 * i915_gem_close_object(), or earlier by i915_gem_context_close().
	 * However, the object may also be bound into the global GTT (e.g.
	 * older GPUs without per-process support, or for direct access through
	 * the GTT either for the user or for scanout). Those VMA still need to
	 * unbound now.
	 */

	spin_lock(&i915->mm.free_lock);
	while ((freed = llist_del_all(&i915->mm.free_list))) {
		spin_unlock(&i915->mm.free_lock);

		__i915_gem_free_objects(i915, freed);
		if (need_resched())
			return;

		spin_lock(&i915->mm.free_lock);
	}
	spin_unlock(&i915->mm.free_lock);
}

static void __i915_gem_free_object_rcu(struct rcu_head *head)
{
	struct drm_i915_gem_object *obj =
		container_of(head, typeof(*obj), rcu);
	struct drm_i915_private *i915 = to_i915(obj->base.dev);

	/*
	 * We reuse obj->rcu for the freed list, so we had better not treat
	 * it like a rcu_head from this point forwards. And we expect all
	 * objects to be freed via this path.
	 */
	destroy_rcu_head(&obj->rcu);

	/*
	 * Since we require blocking on struct_mutex to unbind the freed
	 * object from the GPU before releasing resources back to the
	 * system, we can not do that directly from the RCU callback (which may
	 * be a softirq context), but must instead then defer that work onto a
	 * kthread. We use the RCU callback rather than move the freed object
	 * directly onto the work queue so that we can mix between using the
	 * worker and performing frees directly from subsequent allocations for
	 * crude but effective memory throttling.
	 */
	if (llist_add(&obj->freed, &i915->mm.free_list))
		queue_work(i915->wq, &i915->mm.free_work);
}

void i915_gem_free_object(struct drm_gem_object *gem_obj)
{
	struct drm_i915_gem_object *obj = to_intel_bo(gem_obj);

	if (obj->mm.quirked)
		__i915_gem_object_unpin_pages(obj);

	if (discard_backing_storage(obj))
		obj->mm.madv = I915_MADV_DONTNEED;

	/*
	 * Before we free the object, make sure any pure RCU-only
	 * read-side critical sections are complete, e.g.
	 * i915_gem_busy_ioctl(). For the corresponding synchronized
	 * lookup see i915_gem_object_lookup_rcu().
	 */
	atomic_inc(&to_i915(obj->base.dev)->mm.free_count);
	call_rcu(&obj->rcu, __i915_gem_free_object_rcu);
}

void __i915_gem_object_release_unless_active(struct drm_i915_gem_object *obj)
{
	lockdep_assert_held(&obj->base.dev->struct_mutex);

	if (!i915_gem_object_has_active_reference(obj) &&
	    i915_gem_object_is_active(obj))
		i915_gem_object_set_active_reference(obj);
	else
		i915_gem_object_put(obj);
}

void i915_gem_sanitize(struct drm_i915_private *i915)
{
	int err;

	GEM_TRACE("\n");

	mutex_lock(&i915->drm.struct_mutex);

	intel_runtime_pm_get(i915);
	intel_uncore_forcewake_get(i915, FORCEWAKE_ALL);

	/*
	 * As we have just resumed the machine and woken the device up from
	 * deep PCI sleep (presumably D3_cold), assume the HW has been reset
	 * back to defaults, recovering from whatever wedged state we left it
	 * in and so worth trying to use the device once more.
	 */
	if (i915_terminally_wedged(&i915->gpu_error))
		i915_gem_unset_wedged(i915);

	/*
	 * If we inherit context state from the BIOS or earlier occupants
	 * of the GPU, the GPU may be in an inconsistent state when we
	 * try to take over. The only way to remove the earlier state
	 * is by resetting. However, resetting on earlier gen is tricky as
	 * it may impact the display and we are uncertain about the stability
	 * of the reset, so this could be applied to even earlier gen.
	 */
	err = -ENODEV;
	if (INTEL_GEN(i915) >= 5 && intel_has_gpu_reset(i915))
		err = WARN_ON(intel_gpu_reset(i915, ALL_ENGINES));
	if (!err)
		intel_engines_sanitize(i915);

	intel_uncore_forcewake_put(i915, FORCEWAKE_ALL);
	intel_runtime_pm_put(i915);

	i915_gem_contexts_lost(i915);
	mutex_unlock(&i915->drm.struct_mutex);
}

int i915_gem_suspend(struct drm_i915_private *i915)
{
	int ret;

	GEM_TRACE("\n");

	intel_runtime_pm_get(i915);
	intel_suspend_gt_powersave(i915);

	mutex_lock(&i915->drm.struct_mutex);

	/*
	 * We have to flush all the executing contexts to main memory so
	 * that they can saved in the hibernation image. To ensure the last
	 * context image is coherent, we have to switch away from it. That
	 * leaves the i915->kernel_context still active when
	 * we actually suspend, and its image in memory may not match the GPU
	 * state. Fortunately, the kernel_context is disposable and we do
	 * not rely on its state.
	 */
	if (!i915_terminally_wedged(&i915->gpu_error)) {
		ret = i915_gem_switch_to_kernel_context(i915);
		if (ret)
			goto err_unlock;

		ret = i915_gem_wait_for_idle(i915,
					     I915_WAIT_INTERRUPTIBLE |
					     I915_WAIT_LOCKED |
					     I915_WAIT_FOR_IDLE_BOOST,
					     MAX_SCHEDULE_TIMEOUT);
		if (ret && ret != -EIO)
			goto err_unlock;

		assert_kernel_context_is_current(i915);
	}
	i915_retire_requests(i915); /* ensure we flush after wedging */

	mutex_unlock(&i915->drm.struct_mutex);

	intel_uc_suspend(i915);

	cancel_delayed_work_sync(&i915->gpu_error.hangcheck_work);
	cancel_delayed_work_sync(&i915->gt.retire_work);

	/*
	 * As the idle_work is rearming if it detects a race, play safe and
	 * repeat the flush until it is definitely idle.
	 */
	drain_delayed_work(&i915->gt.idle_work);

	/*
	 * Assert that we successfully flushed all the work and
	 * reset the GPU back to its idle, low power state.
	 */
	WARN_ON(i915->gt.awake);
	if (WARN_ON(!intel_engines_are_idle(i915)))
		i915_gem_set_wedged(i915); /* no hope, discard everything */

	intel_runtime_pm_put(i915);
	return 0;

err_unlock:
	mutex_unlock(&i915->drm.struct_mutex);
	intel_runtime_pm_put(i915);
	return ret;
}

void i915_gem_suspend_late(struct drm_i915_private *i915)
{
	struct drm_i915_gem_object *obj;
	struct list_head *phases[] = {
		&i915->mm.unbound_list,
		&i915->mm.bound_list,
		NULL
	}, **phase;

	/*
	 * Neither the BIOS, ourselves or any other kernel
	 * expects the system to be in execlists mode on startup,
	 * so we need to reset the GPU back to legacy mode. And the only
	 * known way to disable logical contexts is through a GPU reset.
	 *
	 * So in order to leave the system in a known default configuration,
	 * always reset the GPU upon unload and suspend. Afterwards we then
	 * clean up the GEM state tracking, flushing off the requests and
	 * leaving the system in a known idle state.
	 *
	 * Note that is of the upmost importance that the GPU is idle and
	 * all stray writes are flushed *before* we dismantle the backing
	 * storage for the pinned objects.
	 *
	 * However, since we are uncertain that resetting the GPU on older
	 * machines is a good idea, we don't - just in case it leaves the
	 * machine in an unusable condition.
	 */

	mutex_lock(&i915->drm.struct_mutex);
	for (phase = phases; *phase; phase++) {
		list_for_each_entry(obj, *phase, mm.link)
			WARN_ON(i915_gem_object_set_to_gtt_domain(obj, false));
	}
	mutex_unlock(&i915->drm.struct_mutex);

	intel_uc_sanitize(i915);
	i915_gem_sanitize(i915);
}

void i915_gem_resume(struct drm_i915_private *i915)
{
	GEM_TRACE("\n");

	WARN_ON(i915->gt.awake);

	mutex_lock(&i915->drm.struct_mutex);
	intel_uncore_forcewake_get(i915, FORCEWAKE_ALL);

	i915_gem_restore_gtt_mappings(i915);
	i915_gem_restore_fences(i915);

	/*
	 * As we didn't flush the kernel context before suspend, we cannot
	 * guarantee that the context image is complete. So let's just reset
	 * it and start again.
	 */
	i915->gt.resume(i915);

	if (i915_gem_init_hw(i915))
		goto err_wedged;

	intel_uc_resume(i915);

	/* Always reload a context for powersaving. */
	if (i915_gem_switch_to_kernel_context(i915))
		goto err_wedged;

out_unlock:
	intel_uncore_forcewake_put(i915, FORCEWAKE_ALL);
	mutex_unlock(&i915->drm.struct_mutex);
	return;

err_wedged:
	if (!i915_terminally_wedged(&i915->gpu_error)) {
		DRM_ERROR("failed to re-initialize GPU, declaring wedged!\n");
		i915_gem_set_wedged(i915);
	}
	goto out_unlock;
}

void i915_gem_init_swizzling(struct drm_i915_private *dev_priv)
{
	if (INTEL_GEN(dev_priv) < 5 ||
	    dev_priv->mm.bit_6_swizzle_x == I915_BIT_6_SWIZZLE_NONE)
		return;

	I915_WRITE(DISP_ARB_CTL, I915_READ(DISP_ARB_CTL) |
				 DISP_TILE_SURFACE_SWIZZLING);

	if (IS_GEN5(dev_priv))
		return;

	I915_WRITE(TILECTL, I915_READ(TILECTL) | TILECTL_SWZCTL);
	if (IS_GEN6(dev_priv))
		I915_WRITE(ARB_MODE, _MASKED_BIT_ENABLE(ARB_MODE_SWIZZLE_SNB));
	else if (IS_GEN7(dev_priv))
		I915_WRITE(ARB_MODE, _MASKED_BIT_ENABLE(ARB_MODE_SWIZZLE_IVB));
	else if (IS_GEN8(dev_priv))
		I915_WRITE(GAMTARBMODE, _MASKED_BIT_ENABLE(ARB_MODE_SWIZZLE_BDW));
	else
		BUG();
}

static void init_unused_ring(struct drm_i915_private *dev_priv, u32 base)
{
	I915_WRITE(RING_CTL(base), 0);
	I915_WRITE(RING_HEAD(base), 0);
	I915_WRITE(RING_TAIL(base), 0);
	I915_WRITE(RING_START(base), 0);
}

static void init_unused_rings(struct drm_i915_private *dev_priv)
{
	if (IS_I830(dev_priv)) {
		init_unused_ring(dev_priv, PRB1_BASE);
		init_unused_ring(dev_priv, SRB0_BASE);
		init_unused_ring(dev_priv, SRB1_BASE);
		init_unused_ring(dev_priv, SRB2_BASE);
		init_unused_ring(dev_priv, SRB3_BASE);
	} else if (IS_GEN2(dev_priv)) {
		init_unused_ring(dev_priv, SRB0_BASE);
		init_unused_ring(dev_priv, SRB1_BASE);
	} else if (IS_GEN3(dev_priv)) {
		init_unused_ring(dev_priv, PRB1_BASE);
		init_unused_ring(dev_priv, PRB2_BASE);
	}
}

static int __i915_gem_restart_engines(void *data)
{
	struct drm_i915_private *i915 = data;
	struct intel_engine_cs *engine;
	enum intel_engine_id id;
	int err;

	for_each_engine(engine, i915, id) {
		err = engine->init_hw(engine);
		if (err) {
			DRM_ERROR("Failed to restart %s (%d)\n",
				  engine->name, err);
			return err;
		}
	}

	return 0;
}

int i915_gem_init_hw(struct drm_i915_private *dev_priv)
{
	int ret;

	dev_priv->gt.last_init_time = ktime_get();

	/* Double layer security blanket, see i915_gem_init() */
	intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL);

	if (HAS_EDRAM(dev_priv) && INTEL_GEN(dev_priv) < 9)
		I915_WRITE(HSW_IDICR, I915_READ(HSW_IDICR) | IDIHASHMSK(0xf));

	if (IS_HASWELL(dev_priv))
		I915_WRITE(MI_PREDICATE_RESULT_2, IS_HSW_GT3(dev_priv) ?
			   LOWER_SLICE_ENABLED : LOWER_SLICE_DISABLED);

	/* Apply the GT workarounds... */
	intel_gt_apply_workarounds(dev_priv);
	/* ...and determine whether they are sticking. */
	intel_gt_verify_workarounds(dev_priv, "init");

	i915_gem_init_swizzling(dev_priv);

	/*
	 * At least 830 can leave some of the unused rings
	 * "active" (ie. head != tail) after resume which
	 * will prevent c3 entry. Makes sure all unused rings
	 * are totally idle.
	 */
	init_unused_rings(dev_priv);

	BUG_ON(!dev_priv->kernel_context);
	if (i915_terminally_wedged(&dev_priv->gpu_error)) {
		ret = -EIO;
		goto out;
	}

	ret = i915_ppgtt_init_hw(dev_priv);
	if (ret) {
		DRM_ERROR("Enabling PPGTT failed (%d)\n", ret);
		goto out;
	}

	ret = intel_wopcm_init_hw(&dev_priv->wopcm);
	if (ret) {
		DRM_ERROR("Enabling WOPCM failed (%d)\n", ret);
		goto out;
	}

	/* We can't enable contexts until all firmware is loaded */
	ret = intel_uc_init_hw(dev_priv);
	if (ret) {
		DRM_ERROR("Enabling uc failed (%d)\n", ret);
		goto out;
	}

	intel_mocs_init_l3cc_table(dev_priv);

	/* Only when the HW is re-initialised, can we replay the requests */
	ret = __i915_gem_restart_engines(dev_priv);
	if (ret)
		goto cleanup_uc;

	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);

	return 0;

cleanup_uc:
	intel_uc_fini_hw(dev_priv);
out:
	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);

	return ret;
}

static int __intel_engines_record_defaults(struct drm_i915_private *i915)
{
	struct i915_gem_context *ctx;
	struct intel_engine_cs *engine;
	enum intel_engine_id id;
	int err;

	/*
	 * As we reset the gpu during very early sanitisation, the current
	 * register state on the GPU should reflect its defaults values.
	 * We load a context onto the hw (with restore-inhibit), then switch
	 * over to a second context to save that default register state. We
	 * can then prime every new context with that state so they all start
	 * from the same default HW values.
	 */

	ctx = i915_gem_context_create_kernel(i915, 0);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);

	for_each_engine(engine, i915, id) {
		struct i915_request *rq;

		rq = i915_request_alloc(engine, ctx);
		if (IS_ERR(rq)) {
			err = PTR_ERR(rq);
			goto out_ctx;
		}

		err = 0;
		if (engine->init_context)
			err = engine->init_context(rq);

		i915_request_add(rq);
		if (err)
			goto err_active;
	}

	err = i915_gem_switch_to_kernel_context(i915);
	if (err)
		goto err_active;

	if (i915_gem_wait_for_idle(i915, I915_WAIT_LOCKED, HZ / 5)) {
		i915_gem_set_wedged(i915);
		err = -EIO; /* Caller will declare us wedged */
		goto err_active;
	}

	assert_kernel_context_is_current(i915);

	/*
	 * Immediately park the GPU so that we enable powersaving and
	 * treat it as idle. The next time we issue a request, we will
	 * unpark and start using the engine->pinned_default_state, otherwise
	 * it is in limbo and an early reset may fail.
	 */
	__i915_gem_park(i915);

	for_each_engine(engine, i915, id) {
		struct i915_vma *state;
		void *vaddr;

		GEM_BUG_ON(to_intel_context(ctx, engine)->pin_count);

		state = to_intel_context(ctx, engine)->state;
		if (!state)
			continue;

		/*
		 * As we will hold a reference to the logical state, it will
		 * not be torn down with the context, and importantly the
		 * object will hold onto its vma (making it possible for a
		 * stray GTT write to corrupt our defaults). Unmap the vma
		 * from the GTT to prevent such accidents and reclaim the
		 * space.
		 */
		err = i915_vma_unbind(state);
		if (err)
			goto err_active;

		err = i915_gem_object_set_to_cpu_domain(state->obj, false);
		if (err)
			goto err_active;

		engine->default_state = i915_gem_object_get(state->obj);

		/* Check we can acquire the image of the context state */
		vaddr = i915_gem_object_pin_map(engine->default_state,
						I915_MAP_FORCE_WB);
		if (IS_ERR(vaddr)) {
			err = PTR_ERR(vaddr);
			goto err_active;
		}

		i915_gem_object_unpin_map(engine->default_state);
	}

	if (IS_ENABLED(CONFIG_DRM_I915_DEBUG_GEM)) {
		unsigned int found = intel_engines_has_context_isolation(i915);

		/*
		 * Make sure that classes with multiple engine instances all
		 * share the same basic configuration.
		 */
		for_each_engine(engine, i915, id) {
			unsigned int bit = BIT(engine->uabi_class);
			unsigned int expected = engine->default_state ? bit : 0;

			if ((found & bit) != expected) {
				DRM_ERROR("mismatching default context state for class %d on engine %s\n",
					  engine->uabi_class, engine->name);
			}
		}
	}

out_ctx:
	i915_gem_context_set_closed(ctx);
	i915_gem_context_put(ctx);
	return err;

err_active:
	/*
	 * If we have to abandon now, we expect the engines to be idle
	 * and ready to be torn-down. First try to flush any remaining
	 * request, ensure we are pointing at the kernel context and
	 * then remove it.
	 */
	if (WARN_ON(i915_gem_switch_to_kernel_context(i915)))
		goto out_ctx;

	if (WARN_ON(i915_gem_wait_for_idle(i915,
					   I915_WAIT_LOCKED,
					   MAX_SCHEDULE_TIMEOUT)))
		goto out_ctx;

	i915_gem_contexts_lost(i915);
	goto out_ctx;
}

static int
i915_gem_init_scratch(struct drm_i915_private *i915, unsigned int size)
{
	struct drm_i915_gem_object *obj;
	struct i915_vma *vma;
	int ret;

	obj = i915_gem_object_create_stolen(i915, size);
	if (!obj)
		obj = i915_gem_object_create_internal(i915, size);
	if (IS_ERR(obj)) {
		DRM_ERROR("Failed to allocate scratch page\n");
		return PTR_ERR(obj);
	}

	vma = i915_vma_instance(obj, &i915->ggtt.vm, NULL);
	if (IS_ERR(vma)) {
		ret = PTR_ERR(vma);
		goto err_unref;
	}

	ret = i915_vma_pin(vma, 0, 0, PIN_GLOBAL | PIN_HIGH);
	if (ret)
		goto err_unref;

	i915->gt.scratch = vma;
	return 0;

err_unref:
	i915_gem_object_put(obj);
	return ret;
}

static void i915_gem_fini_scratch(struct drm_i915_private *i915)
{
	i915_vma_unpin_and_release(&i915->gt.scratch, 0);
}

int i915_gem_init(struct drm_i915_private *dev_priv)
{
	int ret;

	/* We need to fallback to 4K pages if host doesn't support huge gtt. */
	if (intel_vgpu_active(dev_priv) && !intel_vgpu_has_huge_gtt(dev_priv))
		mkwrite_device_info(dev_priv)->page_sizes =
			I915_GTT_PAGE_SIZE_4K;

	dev_priv->mm.unordered_timeline = dma_fence_context_alloc(1);

	if (HAS_LOGICAL_RING_CONTEXTS(dev_priv)) {
		dev_priv->gt.resume = intel_lr_context_resume;
		dev_priv->gt.cleanup_engine = intel_logical_ring_cleanup;
	} else {
		dev_priv->gt.resume = intel_legacy_submission_resume;
		dev_priv->gt.cleanup_engine = intel_engine_cleanup;
	}

	ret = i915_gem_init_userptr(dev_priv);
	if (ret)
		return ret;

	ret = intel_uc_init_misc(dev_priv);
	if (ret)
		return ret;

	ret = intel_wopcm_init(&dev_priv->wopcm);
	if (ret)
		goto err_uc_misc;

	/* This is just a security blanket to placate dragons.
	 * On some systems, we very sporadically observe that the first TLBs
	 * used by the CS may be stale, despite us poking the TLB reset. If
	 * we hold the forcewake during initialisation these problems
	 * just magically go away.
	 */
	mutex_lock(&dev_priv->drm.struct_mutex);
	intel_uncore_forcewake_get(dev_priv, FORCEWAKE_ALL);

	ret = i915_gem_init_ggtt(dev_priv);
	if (ret) {
		GEM_BUG_ON(ret == -EIO);
		goto err_unlock;
	}

	ret = i915_gem_init_scratch(dev_priv,
				    IS_GEN2(dev_priv) ? SZ_256K : PAGE_SIZE);
	if (ret) {
		GEM_BUG_ON(ret == -EIO);
		goto err_ggtt;
	}

	ret = i915_gem_contexts_init(dev_priv);
	if (ret) {
		GEM_BUG_ON(ret == -EIO);
		goto err_scratch;
	}

	ret = intel_engines_init(dev_priv);
	if (ret) {
		GEM_BUG_ON(ret == -EIO);
		goto err_context;
	}

	intel_init_gt_powersave(dev_priv);

	ret = intel_uc_init(dev_priv);
	if (ret)
		goto err_pm;

	ret = i915_gem_init_hw(dev_priv);
	if (ret)
		goto err_uc_init;

	/*
	 * Despite its name intel_init_clock_gating applies both display
	 * clock gating workarounds; GT mmio workarounds and the occasional
	 * GT power context workaround. Worse, sometimes it includes a context
	 * register workaround which we need to apply before we record the
	 * default HW state for all contexts.
	 *
	 * FIXME: break up the workarounds and apply them at the right time!
	 */
	intel_init_clock_gating(dev_priv);

	ret = __intel_engines_record_defaults(dev_priv);
	if (ret)
		goto err_init_hw;

	if (i915_inject_load_failure()) {
		ret = -ENODEV;
		goto err_init_hw;
	}

	if (i915_inject_load_failure()) {
		ret = -EIO;
		goto err_init_hw;
	}

	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
	mutex_unlock(&dev_priv->drm.struct_mutex);

	return 0;

	/*
	 * Unwinding is complicated by that we want to handle -EIO to mean
	 * disable GPU submission but keep KMS alive. We want to mark the
	 * HW as irrevisibly wedged, but keep enough state around that the
	 * driver doesn't explode during runtime.
	 */
err_init_hw:
	mutex_unlock(&dev_priv->drm.struct_mutex);

	WARN_ON(i915_gem_suspend(dev_priv));
	i915_gem_suspend_late(dev_priv);

	i915_gem_drain_workqueue(dev_priv);

	mutex_lock(&dev_priv->drm.struct_mutex);
	intel_uc_fini_hw(dev_priv);
err_uc_init:
	intel_uc_fini(dev_priv);
err_pm:
	if (ret != -EIO) {
		intel_cleanup_gt_powersave(dev_priv);
		i915_gem_cleanup_engines(dev_priv);
	}
err_context:
	if (ret != -EIO)
		i915_gem_contexts_fini(dev_priv);
err_scratch:
	i915_gem_fini_scratch(dev_priv);
err_ggtt:
err_unlock:
	intel_uncore_forcewake_put(dev_priv, FORCEWAKE_ALL);
	mutex_unlock(&dev_priv->drm.struct_mutex);

err_uc_misc:
	intel_uc_fini_misc(dev_priv);

	if (ret != -EIO)
		i915_gem_cleanup_userptr(dev_priv);

	if (ret == -EIO) {
		mutex_lock(&dev_priv->drm.struct_mutex);

		/*
		 * Allow engine initialisation to fail by marking the GPU as
		 * wedged. But we only want to do this where the GPU is angry,
		 * for all other failure, such as an allocation failure, bail.
		 */
		if (!i915_terminally_wedged(&dev_priv->gpu_error)) {
			i915_load_error(dev_priv,
					"Failed to initialize GPU, declaring it wedged!\n");
			i915_gem_set_wedged(dev_priv);
		}

		/* Minimal basic recovery for KMS */
		ret = i915_ggtt_enable_hw(dev_priv);
		i915_gem_restore_gtt_mappings(dev_priv);
		i915_gem_restore_fences(dev_priv);
		intel_init_clock_gating(dev_priv);

		mutex_unlock(&dev_priv->drm.struct_mutex);
	}

	i915_gem_drain_freed_objects(dev_priv);
	return ret;
}

void i915_gem_fini(struct drm_i915_private *dev_priv)
{
	i915_gem_suspend_late(dev_priv);
	intel_disable_gt_powersave(dev_priv);

	/* Flush any outstanding unpin_work. */
	i915_gem_drain_workqueue(dev_priv);

	mutex_lock(&dev_priv->drm.struct_mutex);
	intel_uc_fini_hw(dev_priv);
	intel_uc_fini(dev_priv);
	i915_gem_cleanup_engines(dev_priv);
	i915_gem_contexts_fini(dev_priv);
	i915_gem_fini_scratch(dev_priv);
	mutex_unlock(&dev_priv->drm.struct_mutex);

	intel_wa_list_free(&dev_priv->gt_wa_list);

	intel_cleanup_gt_powersave(dev_priv);

	intel_uc_fini_misc(dev_priv);
	i915_gem_cleanup_userptr(dev_priv);

	i915_gem_drain_freed_objects(dev_priv);

	WARN_ON(!list_empty(&dev_priv->contexts.list));
}

void i915_gem_init_mmio(struct drm_i915_private *i915)
{
	i915_gem_sanitize(i915);
}

void
i915_gem_cleanup_engines(struct drm_i915_private *dev_priv)
{
	struct intel_engine_cs *engine;
	enum intel_engine_id id;

	for_each_engine(engine, dev_priv, id)
		dev_priv->gt.cleanup_engine(engine);
}

void
i915_gem_load_init_fences(struct drm_i915_private *dev_priv)
{
	int i;

	if (INTEL_GEN(dev_priv) >= 7 && !IS_VALLEYVIEW(dev_priv) &&
	    !IS_CHERRYVIEW(dev_priv))
		dev_priv->num_fence_regs = 32;
	else if (INTEL_GEN(dev_priv) >= 4 ||
		 IS_I945G(dev_priv) || IS_I945GM(dev_priv) ||
		 IS_G33(dev_priv) || IS_PINEVIEW(dev_priv))
		dev_priv->num_fence_regs = 16;
	else
		dev_priv->num_fence_regs = 8;

	if (intel_vgpu_active(dev_priv))
		dev_priv->num_fence_regs =
				I915_READ(vgtif_reg(avail_rs.fence_num));

	/* Initialize fence registers to zero */
	for (i = 0; i < dev_priv->num_fence_regs; i++) {
		struct drm_i915_fence_reg *fence = &dev_priv->fence_regs[i];

		fence->i915 = dev_priv;
		fence->id = i;
		list_add_tail(&fence->link, &dev_priv->mm.fence_list);
	}
	i915_gem_restore_fences(dev_priv);

	i915_gem_detect_bit_6_swizzle(dev_priv);
}

static void i915_gem_init__mm(struct drm_i915_private *i915)
{
	spin_lock_init(&i915->mm.object_stat_lock);
	spin_lock_init(&i915->mm.obj_lock);
	spin_lock_init(&i915->mm.free_lock);

	init_llist_head(&i915->mm.free_list);

	INIT_LIST_HEAD(&i915->mm.unbound_list);
	INIT_LIST_HEAD(&i915->mm.bound_list);
	INIT_LIST_HEAD(&i915->mm.fence_list);
	INIT_LIST_HEAD(&i915->mm.userfault_list);

	INIT_WORK(&i915->mm.free_work, __i915_gem_free_work);
}

int i915_gem_init_early(struct drm_i915_private *dev_priv)
{
	int err = -ENOMEM;

	dev_priv->objects = KMEM_CACHE(drm_i915_gem_object, SLAB_HWCACHE_ALIGN);
	if (!dev_priv->objects)
		goto err_out;

	dev_priv->vmas = KMEM_CACHE(i915_vma, SLAB_HWCACHE_ALIGN);
	if (!dev_priv->vmas)
		goto err_objects;

	dev_priv->luts = KMEM_CACHE(i915_lut_handle, 0);
	if (!dev_priv->luts)
		goto err_vmas;

	dev_priv->requests = KMEM_CACHE(i915_request,
					SLAB_HWCACHE_ALIGN |
					SLAB_RECLAIM_ACCOUNT |
					SLAB_TYPESAFE_BY_RCU);
	if (!dev_priv->requests)
		goto err_luts;

	dev_priv->dependencies = KMEM_CACHE(i915_dependency,
					    SLAB_HWCACHE_ALIGN |
					    SLAB_RECLAIM_ACCOUNT);
	if (!dev_priv->dependencies)
		goto err_requests;

	dev_priv->priorities = KMEM_CACHE(i915_priolist, SLAB_HWCACHE_ALIGN);
	if (!dev_priv->priorities)
		goto err_dependencies;

	INIT_LIST_HEAD(&dev_priv->gt.timelines);
	INIT_LIST_HEAD(&dev_priv->gt.active_rings);
	INIT_LIST_HEAD(&dev_priv->gt.closed_vma);

	i915_gem_init__mm(dev_priv);

	INIT_DELAYED_WORK(&dev_priv->gt.retire_work,
			  i915_gem_retire_work_handler);
	INIT_DELAYED_WORK(&dev_priv->gt.idle_work,
			  i915_gem_idle_work_handler);
	init_waitqueue_head(&dev_priv->gpu_error.wait_queue);
	init_waitqueue_head(&dev_priv->gpu_error.reset_queue);

	atomic_set(&dev_priv->mm.bsd_engine_dispatch_index, 0);

	spin_lock_init(&dev_priv->fb_tracking.lock);

	err = i915_gemfs_init(dev_priv);
	if (err)
		DRM_NOTE("Unable to create a private tmpfs mount, hugepage support will be disabled(%d).\n", err);

	return 0;

err_dependencies:
	kmem_cache_destroy(dev_priv->dependencies);
err_requests:
	kmem_cache_destroy(dev_priv->requests);
err_luts:
	kmem_cache_destroy(dev_priv->luts);
err_vmas:
	kmem_cache_destroy(dev_priv->vmas);
err_objects:
	kmem_cache_destroy(dev_priv->objects);
err_out:
	return err;
}

void i915_gem_cleanup_early(struct drm_i915_private *dev_priv)
{
	i915_gem_drain_freed_objects(dev_priv);
	GEM_BUG_ON(!llist_empty(&dev_priv->mm.free_list));
	GEM_BUG_ON(atomic_read(&dev_priv->mm.free_count));
	WARN_ON(dev_priv->mm.object_count);
	WARN_ON(!list_empty(&dev_priv->gt.timelines));

	kmem_cache_destroy(dev_priv->priorities);
	kmem_cache_destroy(dev_priv->dependencies);
	kmem_cache_destroy(dev_priv->requests);
	kmem_cache_destroy(dev_priv->luts);
	kmem_cache_destroy(dev_priv->vmas);
	kmem_cache_destroy(dev_priv->objects);

	/* And ensure that our DESTROY_BY_RCU slabs are truly destroyed */
	rcu_barrier();

	i915_gemfs_fini(dev_priv);
}

int i915_gem_freeze(struct drm_i915_private *dev_priv)
{
	/* Discard all purgeable objects, let userspace recover those as
	 * required after resuming.
	 */
	i915_gem_shrink_all(dev_priv);

	return 0;
}

int i915_gem_freeze_late(struct drm_i915_private *i915)
{
	struct drm_i915_gem_object *obj;
	struct list_head *phases[] = {
		&i915->mm.unbound_list,
		&i915->mm.bound_list,
		NULL
	}, **phase;

	/*
	 * Called just before we write the hibernation image.
	 *
	 * We need to update the domain tracking to reflect that the CPU
	 * will be accessing all the pages to create and restore from the
	 * hibernation, and so upon restoration those pages will be in the
	 * CPU domain.
	 *
	 * To make sure the hibernation image contains the latest state,
	 * we update that state just before writing out the image.
	 *
	 * To try and reduce the hibernation image, we manually shrink
	 * the objects as well, see i915_gem_freeze()
	 */

	i915_gem_shrink(i915, -1UL, NULL, I915_SHRINK_UNBOUND);
	i915_gem_drain_freed_objects(i915);

	mutex_lock(&i915->drm.struct_mutex);
	for (phase = phases; *phase; phase++) {
		list_for_each_entry(obj, *phase, mm.link)
			WARN_ON(i915_gem_object_set_to_cpu_domain(obj, true));
	}
	mutex_unlock(&i915->drm.struct_mutex);

	return 0;
}

void i915_gem_release(struct drm_device *dev, struct drm_file *file)
{
	struct drm_i915_file_private *file_priv = file->driver_priv;
	struct i915_request *request;

	/* Clean up our request list when the client is going away, so that
	 * later retire_requests won't dereference our soon-to-be-gone
	 * file_priv.
	 */
	spin_lock(&file_priv->mm.lock);
	list_for_each_entry(request, &file_priv->mm.request_list, client_link)
		request->file_priv = NULL;
	spin_unlock(&file_priv->mm.lock);
}

int i915_gem_open(struct drm_i915_private *i915, struct drm_file *file)
{
	struct drm_i915_file_private *file_priv;
	int ret;

	DRM_DEBUG("\n");

	file_priv = kzalloc(sizeof(*file_priv), GFP_KERNEL);
	if (!file_priv)
		return -ENOMEM;

	file->driver_priv = file_priv;
	file_priv->dev_priv = i915;
	file_priv->file = file;

	spin_lock_init(&file_priv->mm.lock);
	INIT_LIST_HEAD(&file_priv->mm.request_list);

	file_priv->bsd_engine = -1;
	file_priv->hang_timestamp = jiffies;

	ret = i915_gem_context_open(i915, file);
	if (ret)
		kfree(file_priv);

	return ret;
}

/**
 * i915_gem_track_fb - update frontbuffer tracking
 * @old: current GEM buffer for the frontbuffer slots
 * @new: new GEM buffer for the frontbuffer slots
 * @frontbuffer_bits: bitmask of frontbuffer slots
 *
 * This updates the frontbuffer tracking bits @frontbuffer_bits by clearing them
 * from @old and setting them in @new. Both @old and @new can be NULL.
 */
void i915_gem_track_fb(struct drm_i915_gem_object *old,
		       struct drm_i915_gem_object *new,
		       unsigned frontbuffer_bits)
{
	/* Control of individual bits within the mask are guarded by
	 * the owning plane->mutex, i.e. we can never see concurrent
	 * manipulation of individual bits. But since the bitfield as a whole
	 * is updated using RMW, we need to use atomics in order to update
	 * the bits.
	 */
	BUILD_BUG_ON(INTEL_FRONTBUFFER_BITS_PER_PIPE * I915_MAX_PIPES >
		     BITS_PER_TYPE(atomic_t));

	if (old) {
		WARN_ON(!(atomic_read(&old->frontbuffer_bits) & frontbuffer_bits));
		atomic_andnot(frontbuffer_bits, &old->frontbuffer_bits);
	}

	if (new) {
		WARN_ON(atomic_read(&new->frontbuffer_bits) & frontbuffer_bits);
		atomic_or(frontbuffer_bits, &new->frontbuffer_bits);
	}
}

/* Allocate a new GEM object and fill it with the supplied data */
struct drm_i915_gem_object *
i915_gem_object_create_from_data(struct drm_i915_private *dev_priv,
			         const void *data, size_t size)
{
	struct drm_i915_gem_object *obj;
	struct file *file;
	size_t offset;
	int err;

	obj = i915_gem_object_create(dev_priv, round_up(size, PAGE_SIZE));
	if (IS_ERR(obj))
		return obj;

	GEM_BUG_ON(obj->write_domain != I915_GEM_DOMAIN_CPU);

	file = obj->base.filp;
	offset = 0;
	do {
		unsigned int len = min_t(typeof(size), size, PAGE_SIZE);
		struct page *page;
		void *pgdata, *vaddr;

		err = pagecache_write_begin(file, file->f_mapping,
					    offset, len, 0,
					    &page, &pgdata);
		if (err < 0)
			goto fail;

		vaddr = kmap(page);
		memcpy(vaddr, data, len);
		kunmap(page);

		err = pagecache_write_end(file, file->f_mapping,
					  offset, len, len,
					  page, pgdata);
		if (err < 0)
			goto fail;

		size -= len;
		data += len;
		offset += len;
	} while (size);

	return obj;

fail:
	i915_gem_object_put(obj);
	return ERR_PTR(err);
}

struct scatterlist *
i915_gem_object_get_sg(struct drm_i915_gem_object *obj,
		       unsigned int n,
		       unsigned int *offset)
{
	struct i915_gem_object_page_iter *iter = &obj->mm.get_page;
	struct scatterlist *sg;
	unsigned int idx, count;

	might_sleep();
	GEM_BUG_ON(n >= obj->base.size >> PAGE_SHIFT);
	GEM_BUG_ON(!i915_gem_object_has_pinned_pages(obj));

	/* As we iterate forward through the sg, we record each entry in a
	 * radixtree for quick repeated (backwards) lookups. If we have seen
	 * this index previously, we will have an entry for it.
	 *
	 * Initial lookup is O(N), but this is amortized to O(1) for
	 * sequential page access (where each new request is consecutive
	 * to the previous one). Repeated lookups are O(lg(obj->base.size)),
	 * i.e. O(1) with a large constant!
	 */
	if (n < READ_ONCE(iter->sg_idx))
		goto lookup;

	mutex_lock(&iter->lock);

	/* We prefer to reuse the last sg so that repeated lookup of this
	 * (or the subsequent) sg are fast - comparing against the last
	 * sg is faster than going through the radixtree.
	 */

	sg = iter->sg_pos;
	idx = iter->sg_idx;
	count = __sg_page_count(sg);

	while (idx + count <= n) {
		void *entry;
		unsigned long i;
		int ret;

		/* If we cannot allocate and insert this entry, or the
		 * individual pages from this range, cancel updating the
		 * sg_idx so that on this lookup we are forced to linearly
		 * scan onwards, but on future lookups we will try the
		 * insertion again (in which case we need to be careful of
		 * the error return reporting that we have already inserted
		 * this index).
		 */
		ret = radix_tree_insert(&iter->radix, idx, sg);
		if (ret && ret != -EEXIST)
			goto scan;

		entry = xa_mk_value(idx);
		for (i = 1; i < count; i++) {
			ret = radix_tree_insert(&iter->radix, idx + i, entry);
			if (ret && ret != -EEXIST)
				goto scan;
		}

		idx += count;
		sg = ____sg_next(sg);
		count = __sg_page_count(sg);
	}

scan:
	iter->sg_pos = sg;
	iter->sg_idx = idx;

	mutex_unlock(&iter->lock);

	if (unlikely(n < idx)) /* insertion completed by another thread */
		goto lookup;

	/* In case we failed to insert the entry into the radixtree, we need
	 * to look beyond the current sg.
	 */
	while (idx + count <= n) {
		idx += count;
		sg = ____sg_next(sg);
		count = __sg_page_count(sg);
	}

	*offset = n - idx;
	return sg;

lookup:
	rcu_read_lock();

	sg = radix_tree_lookup(&iter->radix, n);
	GEM_BUG_ON(!sg);

	/* If this index is in the middle of multi-page sg entry,
	 * the radix tree will contain a value entry that points
	 * to the start of that range. We will return the pointer to
	 * the base page and the offset of this page within the
	 * sg entry's range.
	 */
	*offset = 0;
	if (unlikely(xa_is_value(sg))) {
		unsigned long base = xa_to_value(sg);

		sg = radix_tree_lookup(&iter->radix, base);
		GEM_BUG_ON(!sg);

		*offset = n - base;
	}

	rcu_read_unlock();

	return sg;
}

struct page *
i915_gem_object_get_page(struct drm_i915_gem_object *obj, unsigned int n)
{
	struct scatterlist *sg;
	unsigned int offset;

	GEM_BUG_ON(!i915_gem_object_has_struct_page(obj));

	sg = i915_gem_object_get_sg(obj, n, &offset);
	return nth_page(sg_page(sg), offset);
}

/* Like i915_gem_object_get_page(), but mark the returned page dirty */
struct page *
i915_gem_object_get_dirty_page(struct drm_i915_gem_object *obj,
			       unsigned int n)
{
	struct page *page;

	page = i915_gem_object_get_page(obj, n);
	if (!obj->mm.dirty)
		set_page_dirty(page);

	return page;
}

dma_addr_t
i915_gem_object_get_dma_address(struct drm_i915_gem_object *obj,
				unsigned long n)
{
	struct scatterlist *sg;
	unsigned int offset;

	sg = i915_gem_object_get_sg(obj, n, &offset);
	return sg_dma_address(sg) + (offset << PAGE_SHIFT);
}

int i915_gem_object_attach_phys(struct drm_i915_gem_object *obj, int align)
{
	struct sg_table *pages;
	int err;

	if (align > obj->base.size)
		return -EINVAL;

	if (obj->ops == &i915_gem_phys_ops)
		return 0;

	if (obj->ops != &i915_gem_object_ops)
		return -EINVAL;

	err = i915_gem_object_unbind(obj);
	if (err)
		return err;

	mutex_lock(&obj->mm.lock);

	if (obj->mm.madv != I915_MADV_WILLNEED) {
		err = -EFAULT;
		goto err_unlock;
	}

	if (obj->mm.quirked) {
		err = -EFAULT;
		goto err_unlock;
	}

	if (obj->mm.mapping) {
		err = -EBUSY;
		goto err_unlock;
	}

	pages = __i915_gem_object_unset_pages(obj);

	obj->ops = &i915_gem_phys_ops;

	err = ____i915_gem_object_get_pages(obj);
	if (err)
		goto err_xfer;

	/* Perma-pin (until release) the physical set of pages */
	__i915_gem_object_pin_pages(obj);

	if (!IS_ERR_OR_NULL(pages))
		i915_gem_object_ops.put_pages(obj, pages);
	mutex_unlock(&obj->mm.lock);
	return 0;

err_xfer:
	obj->ops = &i915_gem_object_ops;
	if (!IS_ERR_OR_NULL(pages)) {
		unsigned int sg_page_sizes = i915_sg_page_sizes(pages->sgl);

		__i915_gem_object_set_pages(obj, pages, sg_page_sizes);
	}
err_unlock:
	mutex_unlock(&obj->mm.lock);
	return err;
}

#if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
#include "selftests/scatterlist.c"
#include "selftests/mock_gem_device.c"
#include "selftests/huge_gem_object.c"
#include "selftests/huge_pages.c"
#include "selftests/i915_gem_object.c"
#include "selftests/i915_gem_coherency.c"
#include "selftests/i915_gem.c"
#endif