summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/etnaviv/etnaviv_gpu.c
blob: d8a9a9c68e98e8b861c821d85ae5c50db107a566 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
/*
 * Copyright (C) 2015 Etnaviv Project
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License version 2 as published by
 * the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
 * more details.
 *
 * You should have received a copy of the GNU General Public License along with
 * this program.  If not, see <http://www.gnu.org/licenses/>.
 */

#include <linux/component.h>
#include <linux/fence.h>
#include <linux/moduleparam.h>
#include <linux/of_device.h>
#include "etnaviv_dump.h"
#include "etnaviv_gpu.h"
#include "etnaviv_gem.h"
#include "etnaviv_mmu.h"
#include "etnaviv_iommu.h"
#include "etnaviv_iommu_v2.h"
#include "common.xml.h"
#include "state.xml.h"
#include "state_hi.xml.h"
#include "cmdstream.xml.h"

static const struct platform_device_id gpu_ids[] = {
	{ .name = "etnaviv-gpu,2d" },
	{ },
};

static bool etnaviv_dump_core = true;
module_param_named(dump_core, etnaviv_dump_core, bool, 0600);

/*
 * Driver functions:
 */

int etnaviv_gpu_get_param(struct etnaviv_gpu *gpu, u32 param, u64 *value)
{
	switch (param) {
	case ETNAVIV_PARAM_GPU_MODEL:
		*value = gpu->identity.model;
		break;

	case ETNAVIV_PARAM_GPU_REVISION:
		*value = gpu->identity.revision;
		break;

	case ETNAVIV_PARAM_GPU_FEATURES_0:
		*value = gpu->identity.features;
		break;

	case ETNAVIV_PARAM_GPU_FEATURES_1:
		*value = gpu->identity.minor_features0;
		break;

	case ETNAVIV_PARAM_GPU_FEATURES_2:
		*value = gpu->identity.minor_features1;
		break;

	case ETNAVIV_PARAM_GPU_FEATURES_3:
		*value = gpu->identity.minor_features2;
		break;

	case ETNAVIV_PARAM_GPU_FEATURES_4:
		*value = gpu->identity.minor_features3;
		break;

	case ETNAVIV_PARAM_GPU_FEATURES_5:
		*value = gpu->identity.minor_features4;
		break;

	case ETNAVIV_PARAM_GPU_FEATURES_6:
		*value = gpu->identity.minor_features5;
		break;

	case ETNAVIV_PARAM_GPU_STREAM_COUNT:
		*value = gpu->identity.stream_count;
		break;

	case ETNAVIV_PARAM_GPU_REGISTER_MAX:
		*value = gpu->identity.register_max;
		break;

	case ETNAVIV_PARAM_GPU_THREAD_COUNT:
		*value = gpu->identity.thread_count;
		break;

	case ETNAVIV_PARAM_GPU_VERTEX_CACHE_SIZE:
		*value = gpu->identity.vertex_cache_size;
		break;

	case ETNAVIV_PARAM_GPU_SHADER_CORE_COUNT:
		*value = gpu->identity.shader_core_count;
		break;

	case ETNAVIV_PARAM_GPU_PIXEL_PIPES:
		*value = gpu->identity.pixel_pipes;
		break;

	case ETNAVIV_PARAM_GPU_VERTEX_OUTPUT_BUFFER_SIZE:
		*value = gpu->identity.vertex_output_buffer_size;
		break;

	case ETNAVIV_PARAM_GPU_BUFFER_SIZE:
		*value = gpu->identity.buffer_size;
		break;

	case ETNAVIV_PARAM_GPU_INSTRUCTION_COUNT:
		*value = gpu->identity.instruction_count;
		break;

	case ETNAVIV_PARAM_GPU_NUM_CONSTANTS:
		*value = gpu->identity.num_constants;
		break;

	case ETNAVIV_PARAM_GPU_NUM_VARYINGS:
		*value = gpu->identity.varyings_count;
		break;

	default:
		DBG("%s: invalid param: %u", dev_name(gpu->dev), param);
		return -EINVAL;
	}

	return 0;
}


#define etnaviv_is_model_rev(gpu, mod, rev) \
	((gpu)->identity.model == chipModel_##mod && \
	 (gpu)->identity.revision == rev)
#define etnaviv_field(val, field) \
	(((val) & field##__MASK) >> field##__SHIFT)

static void etnaviv_hw_specs(struct etnaviv_gpu *gpu)
{
	if (gpu->identity.minor_features0 &
	    chipMinorFeatures0_MORE_MINOR_FEATURES) {
		u32 specs[4];
		unsigned int streams;

		specs[0] = gpu_read(gpu, VIVS_HI_CHIP_SPECS);
		specs[1] = gpu_read(gpu, VIVS_HI_CHIP_SPECS_2);
		specs[2] = gpu_read(gpu, VIVS_HI_CHIP_SPECS_3);
		specs[3] = gpu_read(gpu, VIVS_HI_CHIP_SPECS_4);

		gpu->identity.stream_count = etnaviv_field(specs[0],
					VIVS_HI_CHIP_SPECS_STREAM_COUNT);
		gpu->identity.register_max = etnaviv_field(specs[0],
					VIVS_HI_CHIP_SPECS_REGISTER_MAX);
		gpu->identity.thread_count = etnaviv_field(specs[0],
					VIVS_HI_CHIP_SPECS_THREAD_COUNT);
		gpu->identity.vertex_cache_size = etnaviv_field(specs[0],
					VIVS_HI_CHIP_SPECS_VERTEX_CACHE_SIZE);
		gpu->identity.shader_core_count = etnaviv_field(specs[0],
					VIVS_HI_CHIP_SPECS_SHADER_CORE_COUNT);
		gpu->identity.pixel_pipes = etnaviv_field(specs[0],
					VIVS_HI_CHIP_SPECS_PIXEL_PIPES);
		gpu->identity.vertex_output_buffer_size =
			etnaviv_field(specs[0],
				VIVS_HI_CHIP_SPECS_VERTEX_OUTPUT_BUFFER_SIZE);

		gpu->identity.buffer_size = etnaviv_field(specs[1],
					VIVS_HI_CHIP_SPECS_2_BUFFER_SIZE);
		gpu->identity.instruction_count = etnaviv_field(specs[1],
					VIVS_HI_CHIP_SPECS_2_INSTRUCTION_COUNT);
		gpu->identity.num_constants = etnaviv_field(specs[1],
					VIVS_HI_CHIP_SPECS_2_NUM_CONSTANTS);

		gpu->identity.varyings_count = etnaviv_field(specs[2],
					VIVS_HI_CHIP_SPECS_3_VARYINGS_COUNT);

		/* This overrides the value from older register if non-zero */
		streams = etnaviv_field(specs[3],
					VIVS_HI_CHIP_SPECS_4_STREAM_COUNT);
		if (streams)
			gpu->identity.stream_count = streams;
	}

	/* Fill in the stream count if not specified */
	if (gpu->identity.stream_count == 0) {
		if (gpu->identity.model >= 0x1000)
			gpu->identity.stream_count = 4;
		else
			gpu->identity.stream_count = 1;
	}

	/* Convert the register max value */
	if (gpu->identity.register_max)
		gpu->identity.register_max = 1 << gpu->identity.register_max;
	else if (gpu->identity.model == chipModel_GC400)
		gpu->identity.register_max = 32;
	else
		gpu->identity.register_max = 64;

	/* Convert thread count */
	if (gpu->identity.thread_count)
		gpu->identity.thread_count = 1 << gpu->identity.thread_count;
	else if (gpu->identity.model == chipModel_GC400)
		gpu->identity.thread_count = 64;
	else if (gpu->identity.model == chipModel_GC500 ||
		 gpu->identity.model == chipModel_GC530)
		gpu->identity.thread_count = 128;
	else
		gpu->identity.thread_count = 256;

	if (gpu->identity.vertex_cache_size == 0)
		gpu->identity.vertex_cache_size = 8;

	if (gpu->identity.shader_core_count == 0) {
		if (gpu->identity.model >= 0x1000)
			gpu->identity.shader_core_count = 2;
		else
			gpu->identity.shader_core_count = 1;
	}

	if (gpu->identity.pixel_pipes == 0)
		gpu->identity.pixel_pipes = 1;

	/* Convert virtex buffer size */
	if (gpu->identity.vertex_output_buffer_size) {
		gpu->identity.vertex_output_buffer_size =
			1 << gpu->identity.vertex_output_buffer_size;
	} else if (gpu->identity.model == chipModel_GC400) {
		if (gpu->identity.revision < 0x4000)
			gpu->identity.vertex_output_buffer_size = 512;
		else if (gpu->identity.revision < 0x4200)
			gpu->identity.vertex_output_buffer_size = 256;
		else
			gpu->identity.vertex_output_buffer_size = 128;
	} else {
		gpu->identity.vertex_output_buffer_size = 512;
	}

	switch (gpu->identity.instruction_count) {
	case 0:
		if (etnaviv_is_model_rev(gpu, GC2000, 0x5108) ||
		    gpu->identity.model == chipModel_GC880)
			gpu->identity.instruction_count = 512;
		else
			gpu->identity.instruction_count = 256;
		break;

	case 1:
		gpu->identity.instruction_count = 1024;
		break;

	case 2:
		gpu->identity.instruction_count = 2048;
		break;

	default:
		gpu->identity.instruction_count = 256;
		break;
	}

	if (gpu->identity.num_constants == 0)
		gpu->identity.num_constants = 168;

	if (gpu->identity.varyings_count == 0) {
		if (gpu->identity.minor_features1 & chipMinorFeatures1_HALTI0)
			gpu->identity.varyings_count = 12;
		else
			gpu->identity.varyings_count = 8;
	}

	/*
	 * For some cores, two varyings are consumed for position, so the
	 * maximum varying count needs to be reduced by one.
	 */
	if (etnaviv_is_model_rev(gpu, GC5000, 0x5434) ||
	    etnaviv_is_model_rev(gpu, GC4000, 0x5222) ||
	    etnaviv_is_model_rev(gpu, GC4000, 0x5245) ||
	    etnaviv_is_model_rev(gpu, GC4000, 0x5208) ||
	    etnaviv_is_model_rev(gpu, GC3000, 0x5435) ||
	    etnaviv_is_model_rev(gpu, GC2200, 0x5244) ||
	    etnaviv_is_model_rev(gpu, GC2100, 0x5108) ||
	    etnaviv_is_model_rev(gpu, GC2000, 0x5108) ||
	    etnaviv_is_model_rev(gpu, GC1500, 0x5246) ||
	    etnaviv_is_model_rev(gpu, GC880, 0x5107) ||
	    etnaviv_is_model_rev(gpu, GC880, 0x5106))
		gpu->identity.varyings_count -= 1;
}

static void etnaviv_hw_identify(struct etnaviv_gpu *gpu)
{
	u32 chipIdentity;

	chipIdentity = gpu_read(gpu, VIVS_HI_CHIP_IDENTITY);

	/* Special case for older graphic cores. */
	if (etnaviv_field(chipIdentity, VIVS_HI_CHIP_IDENTITY_FAMILY) == 0x01) {
		gpu->identity.model    = chipModel_GC500;
		gpu->identity.revision = etnaviv_field(chipIdentity,
					 VIVS_HI_CHIP_IDENTITY_REVISION);
	} else {

		gpu->identity.model = gpu_read(gpu, VIVS_HI_CHIP_MODEL);
		gpu->identity.revision = gpu_read(gpu, VIVS_HI_CHIP_REV);

		/*
		 * !!!! HACK ALERT !!!!
		 * Because people change device IDs without letting software
		 * know about it - here is the hack to make it all look the
		 * same.  Only for GC400 family.
		 */
		if ((gpu->identity.model & 0xff00) == 0x0400 &&
		    gpu->identity.model != chipModel_GC420) {
			gpu->identity.model = gpu->identity.model & 0x0400;
		}

		/* Another special case */
		if (etnaviv_is_model_rev(gpu, GC300, 0x2201)) {
			u32 chipDate = gpu_read(gpu, VIVS_HI_CHIP_DATE);
			u32 chipTime = gpu_read(gpu, VIVS_HI_CHIP_TIME);

			if (chipDate == 0x20080814 && chipTime == 0x12051100) {
				/*
				 * This IP has an ECO; put the correct
				 * revision in it.
				 */
				gpu->identity.revision = 0x1051;
			}
		}
	}

	dev_info(gpu->dev, "model: GC%x, revision: %x\n",
		 gpu->identity.model, gpu->identity.revision);

	gpu->identity.features = gpu_read(gpu, VIVS_HI_CHIP_FEATURE);

	/* Disable fast clear on GC700. */
	if (gpu->identity.model == chipModel_GC700)
		gpu->identity.features &= ~chipFeatures_FAST_CLEAR;

	if ((gpu->identity.model == chipModel_GC500 &&
	     gpu->identity.revision < 2) ||
	    (gpu->identity.model == chipModel_GC300 &&
	     gpu->identity.revision < 0x2000)) {

		/*
		 * GC500 rev 1.x and GC300 rev < 2.0 doesn't have these
		 * registers.
		 */
		gpu->identity.minor_features0 = 0;
		gpu->identity.minor_features1 = 0;
		gpu->identity.minor_features2 = 0;
		gpu->identity.minor_features3 = 0;
		gpu->identity.minor_features4 = 0;
		gpu->identity.minor_features5 = 0;
	} else
		gpu->identity.minor_features0 =
				gpu_read(gpu, VIVS_HI_CHIP_MINOR_FEATURE_0);

	if (gpu->identity.minor_features0 &
	    chipMinorFeatures0_MORE_MINOR_FEATURES) {
		gpu->identity.minor_features1 =
				gpu_read(gpu, VIVS_HI_CHIP_MINOR_FEATURE_1);
		gpu->identity.minor_features2 =
				gpu_read(gpu, VIVS_HI_CHIP_MINOR_FEATURE_2);
		gpu->identity.minor_features3 =
				gpu_read(gpu, VIVS_HI_CHIP_MINOR_FEATURE_3);
		gpu->identity.minor_features4 =
				gpu_read(gpu, VIVS_HI_CHIP_MINOR_FEATURE_4);
		gpu->identity.minor_features5 =
				gpu_read(gpu, VIVS_HI_CHIP_MINOR_FEATURE_5);
	}

	/* GC600 idle register reports zero bits where modules aren't present */
	if (gpu->identity.model == chipModel_GC600) {
		gpu->idle_mask = VIVS_HI_IDLE_STATE_TX |
				 VIVS_HI_IDLE_STATE_RA |
				 VIVS_HI_IDLE_STATE_SE |
				 VIVS_HI_IDLE_STATE_PA |
				 VIVS_HI_IDLE_STATE_SH |
				 VIVS_HI_IDLE_STATE_PE |
				 VIVS_HI_IDLE_STATE_DE |
				 VIVS_HI_IDLE_STATE_FE;
	} else {
		gpu->idle_mask = ~VIVS_HI_IDLE_STATE_AXI_LP;
	}

	etnaviv_hw_specs(gpu);
}

static void etnaviv_gpu_load_clock(struct etnaviv_gpu *gpu, u32 clock)
{
	gpu_write(gpu, VIVS_HI_CLOCK_CONTROL, clock |
		  VIVS_HI_CLOCK_CONTROL_FSCALE_CMD_LOAD);
	gpu_write(gpu, VIVS_HI_CLOCK_CONTROL, clock);
}

static int etnaviv_hw_reset(struct etnaviv_gpu *gpu)
{
	u32 control, idle;
	unsigned long timeout;
	bool failed = true;

	/* TODO
	 *
	 * - clock gating
	 * - puls eater
	 * - what about VG?
	 */

	/* We hope that the GPU resets in under one second */
	timeout = jiffies + msecs_to_jiffies(1000);

	while (time_is_after_jiffies(timeout)) {
		control = VIVS_HI_CLOCK_CONTROL_DISABLE_DEBUG_REGISTERS |
			  VIVS_HI_CLOCK_CONTROL_FSCALE_VAL(0x40);

		/* enable clock */
		etnaviv_gpu_load_clock(gpu, control);

		/* Wait for stable clock.  Vivante's code waited for 1ms */
		usleep_range(1000, 10000);

		/* isolate the GPU. */
		control |= VIVS_HI_CLOCK_CONTROL_ISOLATE_GPU;
		gpu_write(gpu, VIVS_HI_CLOCK_CONTROL, control);

		/* set soft reset. */
		control |= VIVS_HI_CLOCK_CONTROL_SOFT_RESET;
		gpu_write(gpu, VIVS_HI_CLOCK_CONTROL, control);

		/* wait for reset. */
		msleep(1);

		/* reset soft reset bit. */
		control &= ~VIVS_HI_CLOCK_CONTROL_SOFT_RESET;
		gpu_write(gpu, VIVS_HI_CLOCK_CONTROL, control);

		/* reset GPU isolation. */
		control &= ~VIVS_HI_CLOCK_CONTROL_ISOLATE_GPU;
		gpu_write(gpu, VIVS_HI_CLOCK_CONTROL, control);

		/* read idle register. */
		idle = gpu_read(gpu, VIVS_HI_IDLE_STATE);

		/* try reseting again if FE it not idle */
		if ((idle & VIVS_HI_IDLE_STATE_FE) == 0) {
			dev_dbg(gpu->dev, "FE is not idle\n");
			continue;
		}

		/* read reset register. */
		control = gpu_read(gpu, VIVS_HI_CLOCK_CONTROL);

		/* is the GPU idle? */
		if (((control & VIVS_HI_CLOCK_CONTROL_IDLE_3D) == 0) ||
		    ((control & VIVS_HI_CLOCK_CONTROL_IDLE_2D) == 0)) {
			dev_dbg(gpu->dev, "GPU is not idle\n");
			continue;
		}

		failed = false;
		break;
	}

	if (failed) {
		idle = gpu_read(gpu, VIVS_HI_IDLE_STATE);
		control = gpu_read(gpu, VIVS_HI_CLOCK_CONTROL);

		dev_err(gpu->dev, "GPU failed to reset: FE %sidle, 3D %sidle, 2D %sidle\n",
			idle & VIVS_HI_IDLE_STATE_FE ? "" : "not ",
			control & VIVS_HI_CLOCK_CONTROL_IDLE_3D ? "" : "not ",
			control & VIVS_HI_CLOCK_CONTROL_IDLE_2D ? "" : "not ");

		return -EBUSY;
	}

	/* We rely on the GPU running, so program the clock */
	control = VIVS_HI_CLOCK_CONTROL_DISABLE_DEBUG_REGISTERS |
		  VIVS_HI_CLOCK_CONTROL_FSCALE_VAL(0x40);

	/* enable clock */
	etnaviv_gpu_load_clock(gpu, control);

	return 0;
}

static void etnaviv_gpu_hw_init(struct etnaviv_gpu *gpu)
{
	u16 prefetch;

	if ((etnaviv_is_model_rev(gpu, GC320, 0x5007) ||
	     etnaviv_is_model_rev(gpu, GC320, 0x5220)) &&
	    gpu_read(gpu, VIVS_HI_CHIP_TIME) != 0x2062400) {
		u32 mc_memory_debug;

		mc_memory_debug = gpu_read(gpu, VIVS_MC_DEBUG_MEMORY) & ~0xff;

		if (gpu->identity.revision == 0x5007)
			mc_memory_debug |= 0x0c;
		else
			mc_memory_debug |= 0x08;

		gpu_write(gpu, VIVS_MC_DEBUG_MEMORY, mc_memory_debug);
	}

	/*
	 * Update GPU AXI cache atttribute to "cacheable, no allocate".
	 * This is necessary to prevent the iMX6 SoC locking up.
	 */
	gpu_write(gpu, VIVS_HI_AXI_CONFIG,
		  VIVS_HI_AXI_CONFIG_AWCACHE(2) |
		  VIVS_HI_AXI_CONFIG_ARCACHE(2));

	/* GC2000 rev 5108 needs a special bus config */
	if (etnaviv_is_model_rev(gpu, GC2000, 0x5108)) {
		u32 bus_config = gpu_read(gpu, VIVS_MC_BUS_CONFIG);
		bus_config &= ~(VIVS_MC_BUS_CONFIG_FE_BUS_CONFIG__MASK |
				VIVS_MC_BUS_CONFIG_TX_BUS_CONFIG__MASK);
		bus_config |= VIVS_MC_BUS_CONFIG_FE_BUS_CONFIG(1) |
			      VIVS_MC_BUS_CONFIG_TX_BUS_CONFIG(0);
		gpu_write(gpu, VIVS_MC_BUS_CONFIG, bus_config);
	}

	/* set base addresses */
	gpu_write(gpu, VIVS_MC_MEMORY_BASE_ADDR_RA, gpu->memory_base);
	gpu_write(gpu, VIVS_MC_MEMORY_BASE_ADDR_FE, gpu->memory_base);
	gpu_write(gpu, VIVS_MC_MEMORY_BASE_ADDR_TX, gpu->memory_base);
	gpu_write(gpu, VIVS_MC_MEMORY_BASE_ADDR_PEZ, gpu->memory_base);
	gpu_write(gpu, VIVS_MC_MEMORY_BASE_ADDR_PE, gpu->memory_base);

	/* setup the MMU page table pointers */
	etnaviv_iommu_domain_restore(gpu, gpu->mmu->domain);

	/* Start command processor */
	prefetch = etnaviv_buffer_init(gpu);

	gpu_write(gpu, VIVS_HI_INTR_ENBL, ~0U);
	gpu_write(gpu, VIVS_FE_COMMAND_ADDRESS,
		  gpu->buffer->paddr - gpu->memory_base);
	gpu_write(gpu, VIVS_FE_COMMAND_CONTROL,
		  VIVS_FE_COMMAND_CONTROL_ENABLE |
		  VIVS_FE_COMMAND_CONTROL_PREFETCH(prefetch));
}

int etnaviv_gpu_init(struct etnaviv_gpu *gpu)
{
	int ret, i;
	struct iommu_domain *iommu;
	enum etnaviv_iommu_version version;
	bool mmuv2;

	ret = pm_runtime_get_sync(gpu->dev);
	if (ret < 0)
		return ret;

	etnaviv_hw_identify(gpu);

	if (gpu->identity.model == 0) {
		dev_err(gpu->dev, "Unknown GPU model\n");
		ret = -ENXIO;
		goto fail;
	}

	/* Exclude VG cores with FE2.0 */
	if (gpu->identity.features & chipFeatures_PIPE_VG &&
	    gpu->identity.features & chipFeatures_FE20) {
		dev_info(gpu->dev, "Ignoring GPU with VG and FE2.0\n");
		ret = -ENXIO;
		goto fail;
	}

	/*
	 * Set the GPU linear window to be at the end of the DMA window, where
	 * the CMA area is likely to reside. This ensures that we are able to
	 * map the command buffers while having the linear window overlap as
	 * much RAM as possible, so we can optimize mappings for other buffers.
	 *
	 * For 3D cores only do this if MC2.0 is present, as with MC1.0 it leads
	 * to different views of the memory on the individual engines.
	 */
	if (!(gpu->identity.features & chipFeatures_PIPE_3D) ||
	    (gpu->identity.minor_features0 & chipMinorFeatures0_MC20)) {
		u32 dma_mask = (u32)dma_get_required_mask(gpu->dev);
		if (dma_mask < PHYS_OFFSET + SZ_2G)
			gpu->memory_base = PHYS_OFFSET;
		else
			gpu->memory_base = dma_mask - SZ_2G + 1;
	}

	ret = etnaviv_hw_reset(gpu);
	if (ret)
		goto fail;

	/* Setup IOMMU.. eventually we will (I think) do this once per context
	 * and have separate page tables per context.  For now, to keep things
	 * simple and to get something working, just use a single address space:
	 */
	mmuv2 = gpu->identity.minor_features1 & chipMinorFeatures1_MMU_VERSION;
	dev_dbg(gpu->dev, "mmuv2: %d\n", mmuv2);

	if (!mmuv2) {
		iommu = etnaviv_iommu_domain_alloc(gpu);
		version = ETNAVIV_IOMMU_V1;
	} else {
		iommu = etnaviv_iommu_v2_domain_alloc(gpu);
		version = ETNAVIV_IOMMU_V2;
	}

	if (!iommu) {
		ret = -ENOMEM;
		goto fail;
	}

	gpu->mmu = etnaviv_iommu_new(gpu, iommu, version);
	if (!gpu->mmu) {
		iommu_domain_free(iommu);
		ret = -ENOMEM;
		goto fail;
	}

	/* Create buffer: */
	gpu->buffer = etnaviv_gpu_cmdbuf_new(gpu, PAGE_SIZE, 0);
	if (!gpu->buffer) {
		ret = -ENOMEM;
		dev_err(gpu->dev, "could not create command buffer\n");
		goto destroy_iommu;
	}
	if (gpu->buffer->paddr - gpu->memory_base > 0x80000000) {
		ret = -EINVAL;
		dev_err(gpu->dev,
			"command buffer outside valid memory window\n");
		goto free_buffer;
	}

	/* Setup event management */
	spin_lock_init(&gpu->event_spinlock);
	init_completion(&gpu->event_free);
	for (i = 0; i < ARRAY_SIZE(gpu->event); i++) {
		gpu->event[i].used = false;
		complete(&gpu->event_free);
	}

	/* Now program the hardware */
	mutex_lock(&gpu->lock);
	etnaviv_gpu_hw_init(gpu);
	gpu->exec_state = -1;
	mutex_unlock(&gpu->lock);

	pm_runtime_mark_last_busy(gpu->dev);
	pm_runtime_put_autosuspend(gpu->dev);

	return 0;

free_buffer:
	etnaviv_gpu_cmdbuf_free(gpu->buffer);
	gpu->buffer = NULL;
destroy_iommu:
	etnaviv_iommu_destroy(gpu->mmu);
	gpu->mmu = NULL;
fail:
	pm_runtime_mark_last_busy(gpu->dev);
	pm_runtime_put_autosuspend(gpu->dev);

	return ret;
}

#ifdef CONFIG_DEBUG_FS
struct dma_debug {
	u32 address[2];
	u32 state[2];
};

static void verify_dma(struct etnaviv_gpu *gpu, struct dma_debug *debug)
{
	u32 i;

	debug->address[0] = gpu_read(gpu, VIVS_FE_DMA_ADDRESS);
	debug->state[0]   = gpu_read(gpu, VIVS_FE_DMA_DEBUG_STATE);

	for (i = 0; i < 500; i++) {
		debug->address[1] = gpu_read(gpu, VIVS_FE_DMA_ADDRESS);
		debug->state[1]   = gpu_read(gpu, VIVS_FE_DMA_DEBUG_STATE);

		if (debug->address[0] != debug->address[1])
			break;

		if (debug->state[0] != debug->state[1])
			break;
	}
}

int etnaviv_gpu_debugfs(struct etnaviv_gpu *gpu, struct seq_file *m)
{
	struct dma_debug debug;
	u32 dma_lo, dma_hi, axi, idle;
	int ret;

	seq_printf(m, "%s Status:\n", dev_name(gpu->dev));

	ret = pm_runtime_get_sync(gpu->dev);
	if (ret < 0)
		return ret;

	dma_lo = gpu_read(gpu, VIVS_FE_DMA_LOW);
	dma_hi = gpu_read(gpu, VIVS_FE_DMA_HIGH);
	axi = gpu_read(gpu, VIVS_HI_AXI_STATUS);
	idle = gpu_read(gpu, VIVS_HI_IDLE_STATE);

	verify_dma(gpu, &debug);

	seq_puts(m, "\tfeatures\n");
	seq_printf(m, "\t minor_features0: 0x%08x\n",
		   gpu->identity.minor_features0);
	seq_printf(m, "\t minor_features1: 0x%08x\n",
		   gpu->identity.minor_features1);
	seq_printf(m, "\t minor_features2: 0x%08x\n",
		   gpu->identity.minor_features2);
	seq_printf(m, "\t minor_features3: 0x%08x\n",
		   gpu->identity.minor_features3);
	seq_printf(m, "\t minor_features4: 0x%08x\n",
		   gpu->identity.minor_features4);
	seq_printf(m, "\t minor_features5: 0x%08x\n",
		   gpu->identity.minor_features5);

	seq_puts(m, "\tspecs\n");
	seq_printf(m, "\t stream_count:  %d\n",
			gpu->identity.stream_count);
	seq_printf(m, "\t register_max: %d\n",
			gpu->identity.register_max);
	seq_printf(m, "\t thread_count: %d\n",
			gpu->identity.thread_count);
	seq_printf(m, "\t vertex_cache_size: %d\n",
			gpu->identity.vertex_cache_size);
	seq_printf(m, "\t shader_core_count: %d\n",
			gpu->identity.shader_core_count);
	seq_printf(m, "\t pixel_pipes: %d\n",
			gpu->identity.pixel_pipes);
	seq_printf(m, "\t vertex_output_buffer_size: %d\n",
			gpu->identity.vertex_output_buffer_size);
	seq_printf(m, "\t buffer_size: %d\n",
			gpu->identity.buffer_size);
	seq_printf(m, "\t instruction_count: %d\n",
			gpu->identity.instruction_count);
	seq_printf(m, "\t num_constants: %d\n",
			gpu->identity.num_constants);
	seq_printf(m, "\t varyings_count: %d\n",
			gpu->identity.varyings_count);

	seq_printf(m, "\taxi: 0x%08x\n", axi);
	seq_printf(m, "\tidle: 0x%08x\n", idle);
	idle |= ~gpu->idle_mask & ~VIVS_HI_IDLE_STATE_AXI_LP;
	if ((idle & VIVS_HI_IDLE_STATE_FE) == 0)
		seq_puts(m, "\t FE is not idle\n");
	if ((idle & VIVS_HI_IDLE_STATE_DE) == 0)
		seq_puts(m, "\t DE is not idle\n");
	if ((idle & VIVS_HI_IDLE_STATE_PE) == 0)
		seq_puts(m, "\t PE is not idle\n");
	if ((idle & VIVS_HI_IDLE_STATE_SH) == 0)
		seq_puts(m, "\t SH is not idle\n");
	if ((idle & VIVS_HI_IDLE_STATE_PA) == 0)
		seq_puts(m, "\t PA is not idle\n");
	if ((idle & VIVS_HI_IDLE_STATE_SE) == 0)
		seq_puts(m, "\t SE is not idle\n");
	if ((idle & VIVS_HI_IDLE_STATE_RA) == 0)
		seq_puts(m, "\t RA is not idle\n");
	if ((idle & VIVS_HI_IDLE_STATE_TX) == 0)
		seq_puts(m, "\t TX is not idle\n");
	if ((idle & VIVS_HI_IDLE_STATE_VG) == 0)
		seq_puts(m, "\t VG is not idle\n");
	if ((idle & VIVS_HI_IDLE_STATE_IM) == 0)
		seq_puts(m, "\t IM is not idle\n");
	if ((idle & VIVS_HI_IDLE_STATE_FP) == 0)
		seq_puts(m, "\t FP is not idle\n");
	if ((idle & VIVS_HI_IDLE_STATE_TS) == 0)
		seq_puts(m, "\t TS is not idle\n");
	if (idle & VIVS_HI_IDLE_STATE_AXI_LP)
		seq_puts(m, "\t AXI low power mode\n");

	if (gpu->identity.features & chipFeatures_DEBUG_MODE) {
		u32 read0 = gpu_read(gpu, VIVS_MC_DEBUG_READ0);
		u32 read1 = gpu_read(gpu, VIVS_MC_DEBUG_READ1);
		u32 write = gpu_read(gpu, VIVS_MC_DEBUG_WRITE);

		seq_puts(m, "\tMC\n");
		seq_printf(m, "\t read0: 0x%08x\n", read0);
		seq_printf(m, "\t read1: 0x%08x\n", read1);
		seq_printf(m, "\t write: 0x%08x\n", write);
	}

	seq_puts(m, "\tDMA ");

	if (debug.address[0] == debug.address[1] &&
	    debug.state[0] == debug.state[1]) {
		seq_puts(m, "seems to be stuck\n");
	} else if (debug.address[0] == debug.address[1]) {
		seq_puts(m, "address is constant\n");
	} else {
		seq_puts(m, "is running\n");
	}

	seq_printf(m, "\t address 0: 0x%08x\n", debug.address[0]);
	seq_printf(m, "\t address 1: 0x%08x\n", debug.address[1]);
	seq_printf(m, "\t state 0: 0x%08x\n", debug.state[0]);
	seq_printf(m, "\t state 1: 0x%08x\n", debug.state[1]);
	seq_printf(m, "\t last fetch 64 bit word: 0x%08x 0x%08x\n",
		   dma_lo, dma_hi);

	ret = 0;

	pm_runtime_mark_last_busy(gpu->dev);
	pm_runtime_put_autosuspend(gpu->dev);

	return ret;
}
#endif

/*
 * Power Management:
 */
static int enable_clk(struct etnaviv_gpu *gpu)
{
	if (gpu->clk_core)
		clk_prepare_enable(gpu->clk_core);
	if (gpu->clk_shader)
		clk_prepare_enable(gpu->clk_shader);

	return 0;
}

static int disable_clk(struct etnaviv_gpu *gpu)
{
	if (gpu->clk_core)
		clk_disable_unprepare(gpu->clk_core);
	if (gpu->clk_shader)
		clk_disable_unprepare(gpu->clk_shader);

	return 0;
}

static int enable_axi(struct etnaviv_gpu *gpu)
{
	if (gpu->clk_bus)
		clk_prepare_enable(gpu->clk_bus);

	return 0;
}

static int disable_axi(struct etnaviv_gpu *gpu)
{
	if (gpu->clk_bus)
		clk_disable_unprepare(gpu->clk_bus);

	return 0;
}

/*
 * Hangcheck detection for locked gpu:
 */
static void recover_worker(struct work_struct *work)
{
	struct etnaviv_gpu *gpu = container_of(work, struct etnaviv_gpu,
					       recover_work);
	unsigned long flags;
	unsigned int i;

	dev_err(gpu->dev, "hangcheck recover!\n");

	if (pm_runtime_get_sync(gpu->dev) < 0)
		return;

	mutex_lock(&gpu->lock);

	/* Only catch the first event, or when manually re-armed */
	if (etnaviv_dump_core) {
		etnaviv_core_dump(gpu);
		etnaviv_dump_core = false;
	}

	etnaviv_hw_reset(gpu);

	/* complete all events, the GPU won't do it after the reset */
	spin_lock_irqsave(&gpu->event_spinlock, flags);
	for (i = 0; i < ARRAY_SIZE(gpu->event); i++) {
		if (!gpu->event[i].used)
			continue;
		fence_signal(gpu->event[i].fence);
		gpu->event[i].fence = NULL;
		gpu->event[i].used = false;
		complete(&gpu->event_free);
	}
	spin_unlock_irqrestore(&gpu->event_spinlock, flags);
	gpu->completed_fence = gpu->active_fence;

	etnaviv_gpu_hw_init(gpu);
	gpu->switch_context = true;
	gpu->exec_state = -1;

	mutex_unlock(&gpu->lock);
	pm_runtime_mark_last_busy(gpu->dev);
	pm_runtime_put_autosuspend(gpu->dev);

	/* Retire the buffer objects in a work */
	etnaviv_queue_work(gpu->drm, &gpu->retire_work);
}

static void hangcheck_timer_reset(struct etnaviv_gpu *gpu)
{
	DBG("%s", dev_name(gpu->dev));
	mod_timer(&gpu->hangcheck_timer,
		  round_jiffies_up(jiffies + DRM_ETNAVIV_HANGCHECK_JIFFIES));
}

static void hangcheck_handler(unsigned long data)
{
	struct etnaviv_gpu *gpu = (struct etnaviv_gpu *)data;
	u32 fence = gpu->completed_fence;
	bool progress = false;

	if (fence != gpu->hangcheck_fence) {
		gpu->hangcheck_fence = fence;
		progress = true;
	}

	if (!progress) {
		u32 dma_addr = gpu_read(gpu, VIVS_FE_DMA_ADDRESS);
		int change = dma_addr - gpu->hangcheck_dma_addr;

		if (change < 0 || change > 16) {
			gpu->hangcheck_dma_addr = dma_addr;
			progress = true;
		}
	}

	if (!progress && fence_after(gpu->active_fence, fence)) {
		dev_err(gpu->dev, "hangcheck detected gpu lockup!\n");
		dev_err(gpu->dev, "     completed fence: %u\n", fence);
		dev_err(gpu->dev, "     active fence: %u\n",
			gpu->active_fence);
		etnaviv_queue_work(gpu->drm, &gpu->recover_work);
	}

	/* if still more pending work, reset the hangcheck timer: */
	if (fence_after(gpu->active_fence, gpu->hangcheck_fence))
		hangcheck_timer_reset(gpu);
}

static void hangcheck_disable(struct etnaviv_gpu *gpu)
{
	del_timer_sync(&gpu->hangcheck_timer);
	cancel_work_sync(&gpu->recover_work);
}

/* fence object management */
struct etnaviv_fence {
	struct etnaviv_gpu *gpu;
	struct fence base;
};

static inline struct etnaviv_fence *to_etnaviv_fence(struct fence *fence)
{
	return container_of(fence, struct etnaviv_fence, base);
}

static const char *etnaviv_fence_get_driver_name(struct fence *fence)
{
	return "etnaviv";
}

static const char *etnaviv_fence_get_timeline_name(struct fence *fence)
{
	struct etnaviv_fence *f = to_etnaviv_fence(fence);

	return dev_name(f->gpu->dev);
}

static bool etnaviv_fence_enable_signaling(struct fence *fence)
{
	return true;
}

static bool etnaviv_fence_signaled(struct fence *fence)
{
	struct etnaviv_fence *f = to_etnaviv_fence(fence);

	return fence_completed(f->gpu, f->base.seqno);
}

static void etnaviv_fence_release(struct fence *fence)
{
	struct etnaviv_fence *f = to_etnaviv_fence(fence);

	kfree_rcu(f, base.rcu);
}

static const struct fence_ops etnaviv_fence_ops = {
	.get_driver_name = etnaviv_fence_get_driver_name,
	.get_timeline_name = etnaviv_fence_get_timeline_name,
	.enable_signaling = etnaviv_fence_enable_signaling,
	.signaled = etnaviv_fence_signaled,
	.wait = fence_default_wait,
	.release = etnaviv_fence_release,
};

static struct fence *etnaviv_gpu_fence_alloc(struct etnaviv_gpu *gpu)
{
	struct etnaviv_fence *f;

	f = kzalloc(sizeof(*f), GFP_KERNEL);
	if (!f)
		return NULL;

	f->gpu = gpu;

	fence_init(&f->base, &etnaviv_fence_ops, &gpu->fence_spinlock,
		   gpu->fence_context, ++gpu->next_fence);

	return &f->base;
}

int etnaviv_gpu_fence_sync_obj(struct etnaviv_gem_object *etnaviv_obj,
	unsigned int context, bool exclusive)
{
	struct reservation_object *robj = etnaviv_obj->resv;
	struct reservation_object_list *fobj;
	struct fence *fence;
	int i, ret;

	if (!exclusive) {
		ret = reservation_object_reserve_shared(robj);
		if (ret)
			return ret;
	}

	/*
	 * If we have any shared fences, then the exclusive fence
	 * should be ignored as it will already have been signalled.
	 */
	fobj = reservation_object_get_list(robj);
	if (!fobj || fobj->shared_count == 0) {
		/* Wait on any existing exclusive fence which isn't our own */
		fence = reservation_object_get_excl(robj);
		if (fence && fence->context != context) {
			ret = fence_wait(fence, true);
			if (ret)
				return ret;
		}
	}

	if (!exclusive || !fobj)
		return 0;

	for (i = 0; i < fobj->shared_count; i++) {
		fence = rcu_dereference_protected(fobj->shared[i],
						reservation_object_held(robj));
		if (fence->context != context) {
			ret = fence_wait(fence, true);
			if (ret)
				return ret;
		}
	}

	return 0;
}

/*
 * event management:
 */

static unsigned int event_alloc(struct etnaviv_gpu *gpu)
{
	unsigned long ret, flags;
	unsigned int i, event = ~0U;

	ret = wait_for_completion_timeout(&gpu->event_free,
					  msecs_to_jiffies(10 * 10000));
	if (!ret)
		dev_err(gpu->dev, "wait_for_completion_timeout failed");

	spin_lock_irqsave(&gpu->event_spinlock, flags);

	/* find first free event */
	for (i = 0; i < ARRAY_SIZE(gpu->event); i++) {
		if (gpu->event[i].used == false) {
			gpu->event[i].used = true;
			event = i;
			break;
		}
	}

	spin_unlock_irqrestore(&gpu->event_spinlock, flags);

	return event;
}

static void event_free(struct etnaviv_gpu *gpu, unsigned int event)
{
	unsigned long flags;

	spin_lock_irqsave(&gpu->event_spinlock, flags);

	if (gpu->event[event].used == false) {
		dev_warn(gpu->dev, "event %u is already marked as free",
			 event);
		spin_unlock_irqrestore(&gpu->event_spinlock, flags);
	} else {
		gpu->event[event].used = false;
		spin_unlock_irqrestore(&gpu->event_spinlock, flags);

		complete(&gpu->event_free);
	}
}

/*
 * Cmdstream submission/retirement:
 */

struct etnaviv_cmdbuf *etnaviv_gpu_cmdbuf_new(struct etnaviv_gpu *gpu, u32 size,
	size_t nr_bos)
{
	struct etnaviv_cmdbuf *cmdbuf;
	size_t sz = size_vstruct(nr_bos, sizeof(cmdbuf->bo_map[0]),
				 sizeof(*cmdbuf));

	cmdbuf = kzalloc(sz, GFP_KERNEL);
	if (!cmdbuf)
		return NULL;

	cmdbuf->vaddr = dma_alloc_wc(gpu->dev, size, &cmdbuf->paddr,
				     GFP_KERNEL);
	if (!cmdbuf->vaddr) {
		kfree(cmdbuf);
		return NULL;
	}

	cmdbuf->gpu = gpu;
	cmdbuf->size = size;

	return cmdbuf;
}

void etnaviv_gpu_cmdbuf_free(struct etnaviv_cmdbuf *cmdbuf)
{
	dma_free_wc(cmdbuf->gpu->dev, cmdbuf->size, cmdbuf->vaddr,
		    cmdbuf->paddr);
	kfree(cmdbuf);
}

static void retire_worker(struct work_struct *work)
{
	struct etnaviv_gpu *gpu = container_of(work, struct etnaviv_gpu,
					       retire_work);
	u32 fence = gpu->completed_fence;
	struct etnaviv_cmdbuf *cmdbuf, *tmp;
	unsigned int i;

	mutex_lock(&gpu->lock);
	list_for_each_entry_safe(cmdbuf, tmp, &gpu->active_cmd_list, node) {
		if (!fence_is_signaled(cmdbuf->fence))
			break;

		list_del(&cmdbuf->node);
		fence_put(cmdbuf->fence);

		for (i = 0; i < cmdbuf->nr_bos; i++) {
			struct etnaviv_vram_mapping *mapping = cmdbuf->bo_map[i];
			struct etnaviv_gem_object *etnaviv_obj = mapping->object;

			atomic_dec(&etnaviv_obj->gpu_active);
			/* drop the refcount taken in etnaviv_gpu_submit */
			etnaviv_gem_mapping_unreference(mapping);
		}

		etnaviv_gpu_cmdbuf_free(cmdbuf);
		/*
		 * We need to balance the runtime PM count caused by
		 * each submission.  Upon submission, we increment
		 * the runtime PM counter, and allocate one event.
		 * So here, we put the runtime PM count for each
		 * completed event.
		 */
		pm_runtime_put_autosuspend(gpu->dev);
	}

	gpu->retired_fence = fence;

	mutex_unlock(&gpu->lock);

	wake_up_all(&gpu->fence_event);
}

int etnaviv_gpu_wait_fence_interruptible(struct etnaviv_gpu *gpu,
	u32 fence, struct timespec *timeout)
{
	int ret;

	if (fence_after(fence, gpu->next_fence)) {
		DRM_ERROR("waiting on invalid fence: %u (of %u)\n",
				fence, gpu->next_fence);
		return -EINVAL;
	}

	if (!timeout) {
		/* No timeout was requested: just test for completion */
		ret = fence_completed(gpu, fence) ? 0 : -EBUSY;
	} else {
		unsigned long remaining = etnaviv_timeout_to_jiffies(timeout);

		ret = wait_event_interruptible_timeout(gpu->fence_event,
						fence_completed(gpu, fence),
						remaining);
		if (ret == 0) {
			DBG("timeout waiting for fence: %u (retired: %u completed: %u)",
				fence, gpu->retired_fence,
				gpu->completed_fence);
			ret = -ETIMEDOUT;
		} else if (ret != -ERESTARTSYS) {
			ret = 0;
		}
	}

	return ret;
}

/*
 * Wait for an object to become inactive.  This, on it's own, is not race
 * free: the object is moved by the retire worker off the active list, and
 * then the iova is put.  Moreover, the object could be re-submitted just
 * after we notice that it's become inactive.
 *
 * Although the retirement happens under the gpu lock, we don't want to hold
 * that lock in this function while waiting.
 */
int etnaviv_gpu_wait_obj_inactive(struct etnaviv_gpu *gpu,
	struct etnaviv_gem_object *etnaviv_obj, struct timespec *timeout)
{
	unsigned long remaining;
	long ret;

	if (!timeout)
		return !is_active(etnaviv_obj) ? 0 : -EBUSY;

	remaining = etnaviv_timeout_to_jiffies(timeout);

	ret = wait_event_interruptible_timeout(gpu->fence_event,
					       !is_active(etnaviv_obj),
					       remaining);
	if (ret > 0) {
		struct etnaviv_drm_private *priv = gpu->drm->dev_private;

		/* Synchronise with the retire worker */
		flush_workqueue(priv->wq);
		return 0;
	} else if (ret == -ERESTARTSYS) {
		return -ERESTARTSYS;
	} else {
		return -ETIMEDOUT;
	}
}

int etnaviv_gpu_pm_get_sync(struct etnaviv_gpu *gpu)
{
	return pm_runtime_get_sync(gpu->dev);
}

void etnaviv_gpu_pm_put(struct etnaviv_gpu *gpu)
{
	pm_runtime_mark_last_busy(gpu->dev);
	pm_runtime_put_autosuspend(gpu->dev);
}

/* add bo's to gpu's ring, and kick gpu: */
int etnaviv_gpu_submit(struct etnaviv_gpu *gpu,
	struct etnaviv_gem_submit *submit, struct etnaviv_cmdbuf *cmdbuf)
{
	struct fence *fence;
	unsigned int event, i;
	int ret;

	ret = etnaviv_gpu_pm_get_sync(gpu);
	if (ret < 0)
		return ret;

	mutex_lock(&gpu->lock);

	/*
	 * TODO
	 *
	 * - flush
	 * - data endian
	 * - prefetch
	 *
	 */

	event = event_alloc(gpu);
	if (unlikely(event == ~0U)) {
		DRM_ERROR("no free event\n");
		ret = -EBUSY;
		goto out_unlock;
	}

	fence = etnaviv_gpu_fence_alloc(gpu);
	if (!fence) {
		event_free(gpu, event);
		ret = -ENOMEM;
		goto out_unlock;
	}

	gpu->event[event].fence = fence;
	submit->fence = fence->seqno;
	gpu->active_fence = submit->fence;

	if (gpu->lastctx != cmdbuf->ctx) {
		gpu->mmu->need_flush = true;
		gpu->switch_context = true;
		gpu->lastctx = cmdbuf->ctx;
	}

	etnaviv_buffer_queue(gpu, event, cmdbuf);

	cmdbuf->fence = fence;
	list_add_tail(&cmdbuf->node, &gpu->active_cmd_list);

	/* We're committed to adding this command buffer, hold a PM reference */
	pm_runtime_get_noresume(gpu->dev);

	for (i = 0; i < submit->nr_bos; i++) {
		struct etnaviv_gem_object *etnaviv_obj = submit->bos[i].obj;

		/* Each cmdbuf takes a refcount on the mapping */
		etnaviv_gem_mapping_reference(submit->bos[i].mapping);
		cmdbuf->bo_map[i] = submit->bos[i].mapping;
		atomic_inc(&etnaviv_obj->gpu_active);

		if (submit->bos[i].flags & ETNA_SUBMIT_BO_WRITE)
			reservation_object_add_excl_fence(etnaviv_obj->resv,
							  fence);
		else
			reservation_object_add_shared_fence(etnaviv_obj->resv,
							    fence);
	}
	cmdbuf->nr_bos = submit->nr_bos;
	hangcheck_timer_reset(gpu);
	ret = 0;

out_unlock:
	mutex_unlock(&gpu->lock);

	etnaviv_gpu_pm_put(gpu);

	return ret;
}

/*
 * Init/Cleanup:
 */
static irqreturn_t irq_handler(int irq, void *data)
{
	struct etnaviv_gpu *gpu = data;
	irqreturn_t ret = IRQ_NONE;

	u32 intr = gpu_read(gpu, VIVS_HI_INTR_ACKNOWLEDGE);

	if (intr != 0) {
		int event;

		pm_runtime_mark_last_busy(gpu->dev);

		dev_dbg(gpu->dev, "intr 0x%08x\n", intr);

		if (intr & VIVS_HI_INTR_ACKNOWLEDGE_AXI_BUS_ERROR) {
			dev_err(gpu->dev, "AXI bus error\n");
			intr &= ~VIVS_HI_INTR_ACKNOWLEDGE_AXI_BUS_ERROR;
		}

		while ((event = ffs(intr)) != 0) {
			struct fence *fence;

			event -= 1;

			intr &= ~(1 << event);

			dev_dbg(gpu->dev, "event %u\n", event);

			fence = gpu->event[event].fence;
			gpu->event[event].fence = NULL;
			fence_signal(fence);

			/*
			 * Events can be processed out of order.  Eg,
			 * - allocate and queue event 0
			 * - allocate event 1
			 * - event 0 completes, we process it
			 * - allocate and queue event 0
			 * - event 1 and event 0 complete
			 * we can end up processing event 0 first, then 1.
			 */
			if (fence_after(fence->seqno, gpu->completed_fence))
				gpu->completed_fence = fence->seqno;

			event_free(gpu, event);
		}

		/* Retire the buffer objects in a work */
		etnaviv_queue_work(gpu->drm, &gpu->retire_work);

		ret = IRQ_HANDLED;
	}

	return ret;
}

static int etnaviv_gpu_clk_enable(struct etnaviv_gpu *gpu)
{
	int ret;

	ret = enable_clk(gpu);
	if (ret)
		return ret;

	ret = enable_axi(gpu);
	if (ret) {
		disable_clk(gpu);
		return ret;
	}

	return 0;
}

static int etnaviv_gpu_clk_disable(struct etnaviv_gpu *gpu)
{
	int ret;

	ret = disable_axi(gpu);
	if (ret)
		return ret;

	ret = disable_clk(gpu);
	if (ret)
		return ret;

	return 0;
}

static int etnaviv_gpu_hw_suspend(struct etnaviv_gpu *gpu)
{
	if (gpu->buffer) {
		unsigned long timeout;

		/* Replace the last WAIT with END */
		etnaviv_buffer_end(gpu);

		/*
		 * We know that only the FE is busy here, this should
		 * happen quickly (as the WAIT is only 200 cycles).  If
		 * we fail, just warn and continue.
		 */
		timeout = jiffies + msecs_to_jiffies(100);
		do {
			u32 idle = gpu_read(gpu, VIVS_HI_IDLE_STATE);

			if ((idle & gpu->idle_mask) == gpu->idle_mask)
				break;

			if (time_is_before_jiffies(timeout)) {
				dev_warn(gpu->dev,
					 "timed out waiting for idle: idle=0x%x\n",
					 idle);
				break;
			}

			udelay(5);
		} while (1);
	}

	return etnaviv_gpu_clk_disable(gpu);
}

#ifdef CONFIG_PM
static int etnaviv_gpu_hw_resume(struct etnaviv_gpu *gpu)
{
	u32 clock;
	int ret;

	ret = mutex_lock_killable(&gpu->lock);
	if (ret)
		return ret;

	clock = VIVS_HI_CLOCK_CONTROL_DISABLE_DEBUG_REGISTERS |
		VIVS_HI_CLOCK_CONTROL_FSCALE_VAL(0x40);

	etnaviv_gpu_load_clock(gpu, clock);
	etnaviv_gpu_hw_init(gpu);

	gpu->switch_context = true;
	gpu->exec_state = -1;

	mutex_unlock(&gpu->lock);

	return 0;
}
#endif

static int etnaviv_gpu_bind(struct device *dev, struct device *master,
	void *data)
{
	struct drm_device *drm = data;
	struct etnaviv_drm_private *priv = drm->dev_private;
	struct etnaviv_gpu *gpu = dev_get_drvdata(dev);
	int ret;

#ifdef CONFIG_PM
	ret = pm_runtime_get_sync(gpu->dev);
#else
	ret = etnaviv_gpu_clk_enable(gpu);
#endif
	if (ret < 0)
		return ret;

	gpu->drm = drm;
	gpu->fence_context = fence_context_alloc(1);
	spin_lock_init(&gpu->fence_spinlock);

	INIT_LIST_HEAD(&gpu->active_cmd_list);
	INIT_WORK(&gpu->retire_work, retire_worker);
	INIT_WORK(&gpu->recover_work, recover_worker);
	init_waitqueue_head(&gpu->fence_event);

	setup_timer(&gpu->hangcheck_timer, hangcheck_handler,
			(unsigned long)gpu);

	priv->gpu[priv->num_gpus++] = gpu;

	pm_runtime_mark_last_busy(gpu->dev);
	pm_runtime_put_autosuspend(gpu->dev);

	return 0;
}

static void etnaviv_gpu_unbind(struct device *dev, struct device *master,
	void *data)
{
	struct etnaviv_gpu *gpu = dev_get_drvdata(dev);

	DBG("%s", dev_name(gpu->dev));

	hangcheck_disable(gpu);

#ifdef CONFIG_PM
	pm_runtime_get_sync(gpu->dev);
	pm_runtime_put_sync_suspend(gpu->dev);
#else
	etnaviv_gpu_hw_suspend(gpu);
#endif

	if (gpu->buffer) {
		etnaviv_gpu_cmdbuf_free(gpu->buffer);
		gpu->buffer = NULL;
	}

	if (gpu->mmu) {
		etnaviv_iommu_destroy(gpu->mmu);
		gpu->mmu = NULL;
	}

	gpu->drm = NULL;
}

static const struct component_ops gpu_ops = {
	.bind = etnaviv_gpu_bind,
	.unbind = etnaviv_gpu_unbind,
};

static const struct of_device_id etnaviv_gpu_match[] = {
	{
		.compatible = "vivante,gc"
	},
	{ /* sentinel */ }
};

static int etnaviv_gpu_platform_probe(struct platform_device *pdev)
{
	struct device *dev = &pdev->dev;
	struct etnaviv_gpu *gpu;
	int err = 0;

	gpu = devm_kzalloc(dev, sizeof(*gpu), GFP_KERNEL);
	if (!gpu)
		return -ENOMEM;

	gpu->dev = &pdev->dev;
	mutex_init(&gpu->lock);

	/* Map registers: */
	gpu->mmio = etnaviv_ioremap(pdev, NULL, dev_name(gpu->dev));
	if (IS_ERR(gpu->mmio))
		return PTR_ERR(gpu->mmio);

	/* Get Interrupt: */
	gpu->irq = platform_get_irq(pdev, 0);
	if (gpu->irq < 0) {
		err = gpu->irq;
		dev_err(dev, "failed to get irq: %d\n", err);
		goto fail;
	}

	err = devm_request_irq(&pdev->dev, gpu->irq, irq_handler, 0,
			       dev_name(gpu->dev), gpu);
	if (err) {
		dev_err(dev, "failed to request IRQ%u: %d\n", gpu->irq, err);
		goto fail;
	}

	/* Get Clocks: */
	gpu->clk_bus = devm_clk_get(&pdev->dev, "bus");
	DBG("clk_bus: %p", gpu->clk_bus);
	if (IS_ERR(gpu->clk_bus))
		gpu->clk_bus = NULL;

	gpu->clk_core = devm_clk_get(&pdev->dev, "core");
	DBG("clk_core: %p", gpu->clk_core);
	if (IS_ERR(gpu->clk_core))
		gpu->clk_core = NULL;

	gpu->clk_shader = devm_clk_get(&pdev->dev, "shader");
	DBG("clk_shader: %p", gpu->clk_shader);
	if (IS_ERR(gpu->clk_shader))
		gpu->clk_shader = NULL;

	/* TODO: figure out max mapped size */
	dev_set_drvdata(dev, gpu);

	/*
	 * We treat the device as initially suspended.  The runtime PM
	 * autosuspend delay is rather arbitary: no measurements have
	 * yet been performed to determine an appropriate value.
	 */
	pm_runtime_use_autosuspend(gpu->dev);
	pm_runtime_set_autosuspend_delay(gpu->dev, 200);
	pm_runtime_enable(gpu->dev);

	err = component_add(&pdev->dev, &gpu_ops);
	if (err < 0) {
		dev_err(&pdev->dev, "failed to register component: %d\n", err);
		goto fail;
	}

	return 0;

fail:
	return err;
}

static int etnaviv_gpu_platform_remove(struct platform_device *pdev)
{
	component_del(&pdev->dev, &gpu_ops);
	pm_runtime_disable(&pdev->dev);
	return 0;
}

#ifdef CONFIG_PM
static int etnaviv_gpu_rpm_suspend(struct device *dev)
{
	struct etnaviv_gpu *gpu = dev_get_drvdata(dev);
	u32 idle, mask;

	/* If we have outstanding fences, we're not idle */
	if (gpu->completed_fence != gpu->active_fence)
		return -EBUSY;

	/* Check whether the hardware (except FE) is idle */
	mask = gpu->idle_mask & ~VIVS_HI_IDLE_STATE_FE;
	idle = gpu_read(gpu, VIVS_HI_IDLE_STATE) & mask;
	if (idle != mask)
		return -EBUSY;

	return etnaviv_gpu_hw_suspend(gpu);
}

static int etnaviv_gpu_rpm_resume(struct device *dev)
{
	struct etnaviv_gpu *gpu = dev_get_drvdata(dev);
	int ret;

	ret = etnaviv_gpu_clk_enable(gpu);
	if (ret)
		return ret;

	/* Re-initialise the basic hardware state */
	if (gpu->drm && gpu->buffer) {
		ret = etnaviv_gpu_hw_resume(gpu);
		if (ret) {
			etnaviv_gpu_clk_disable(gpu);
			return ret;
		}
	}

	return 0;
}
#endif

static const struct dev_pm_ops etnaviv_gpu_pm_ops = {
	SET_RUNTIME_PM_OPS(etnaviv_gpu_rpm_suspend, etnaviv_gpu_rpm_resume,
			   NULL)
};

struct platform_driver etnaviv_gpu_driver = {
	.driver = {
		.name = "etnaviv-gpu",
		.owner = THIS_MODULE,
		.pm = &etnaviv_gpu_pm_ops,
		.of_match_table = etnaviv_gpu_match,
	},
	.probe = etnaviv_gpu_platform_probe,
	.remove = etnaviv_gpu_platform_remove,
	.id_table = gpu_ids,
};