summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/i915/gt/intel_gt.c
blob: 7a008e829d4d099e9372be32b37d70ad60b67384 (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
// SPDX-License-Identifier: MIT
/*
 * Copyright © 2019 Intel Corporation
 */

#include <drm/drm_managed.h>
#include <drm/intel-gtt.h>

#include "gem/i915_gem_internal.h"
#include "gem/i915_gem_lmem.h"

#include "i915_drv.h"
#include "i915_perf_oa_regs.h"
#include "i915_reg.h"
#include "intel_context.h"
#include "intel_engine_pm.h"
#include "intel_engine_regs.h"
#include "intel_ggtt_gmch.h"
#include "intel_gt.h"
#include "intel_gt_buffer_pool.h"
#include "intel_gt_clock_utils.h"
#include "intel_gt_debugfs.h"
#include "intel_gt_mcr.h"
#include "intel_gt_pm.h"
#include "intel_gt_print.h"
#include "intel_gt_regs.h"
#include "intel_gt_requests.h"
#include "intel_migrate.h"
#include "intel_mocs.h"
#include "intel_pci_config.h"
#include "intel_rc6.h"
#include "intel_renderstate.h"
#include "intel_rps.h"
#include "intel_sa_media.h"
#include "intel_gt_sysfs.h"
#include "intel_uncore.h"
#include "shmem_utils.h"

void intel_gt_common_init_early(struct intel_gt *gt)
{
	spin_lock_init(gt->irq_lock);

	INIT_LIST_HEAD(&gt->closed_vma);
	spin_lock_init(&gt->closed_lock);

	init_llist_head(&gt->watchdog.list);
	INIT_WORK(&gt->watchdog.work, intel_gt_watchdog_work);

	intel_gt_init_buffer_pool(gt);
	intel_gt_init_reset(gt);
	intel_gt_init_requests(gt);
	intel_gt_init_timelines(gt);
	mutex_init(&gt->tlb.invalidate_lock);
	seqcount_mutex_init(&gt->tlb.seqno, &gt->tlb.invalidate_lock);
	intel_gt_pm_init_early(gt);

	intel_wopcm_init_early(&gt->wopcm);
	intel_uc_init_early(&gt->uc);
	intel_rps_init_early(&gt->rps);
}

/* Preliminary initialization of Tile 0 */
int intel_root_gt_init_early(struct drm_i915_private *i915)
{
	struct intel_gt *gt = to_gt(i915);

	gt->i915 = i915;
	gt->uncore = &i915->uncore;
	gt->irq_lock = drmm_kzalloc(&i915->drm, sizeof(*gt->irq_lock), GFP_KERNEL);
	if (!gt->irq_lock)
		return -ENOMEM;

	intel_gt_common_init_early(gt);

	return 0;
}

static int intel_gt_probe_lmem(struct intel_gt *gt)
{
	struct drm_i915_private *i915 = gt->i915;
	unsigned int instance = gt->info.id;
	int id = INTEL_REGION_LMEM_0 + instance;
	struct intel_memory_region *mem;
	int err;

	mem = intel_gt_setup_lmem(gt);
	if (IS_ERR(mem)) {
		err = PTR_ERR(mem);
		if (err == -ENODEV)
			return 0;

		gt_err(gt, "Failed to setup region(%d) type=%d\n",
		       err, INTEL_MEMORY_LOCAL);
		return err;
	}

	mem->id = id;
	mem->instance = instance;

	intel_memory_region_set_name(mem, "local%u", mem->instance);

	GEM_BUG_ON(!HAS_REGION(i915, id));
	GEM_BUG_ON(i915->mm.regions[id]);
	i915->mm.regions[id] = mem;

	return 0;
}

int intel_gt_assign_ggtt(struct intel_gt *gt)
{
	/* Media GT shares primary GT's GGTT */
	if (gt->type == GT_MEDIA) {
		gt->ggtt = to_gt(gt->i915)->ggtt;
	} else {
		gt->ggtt = i915_ggtt_create(gt->i915);
		if (IS_ERR(gt->ggtt))
			return PTR_ERR(gt->ggtt);
	}

	list_add_tail(&gt->ggtt_link, &gt->ggtt->gt_list);

	return 0;
}

int intel_gt_init_mmio(struct intel_gt *gt)
{
	intel_gt_init_clock_frequency(gt);

	intel_uc_init_mmio(&gt->uc);
	intel_sseu_info_init(gt);
	intel_gt_mcr_init(gt);

	return intel_engines_init_mmio(gt);
}

static void init_unused_ring(struct intel_gt *gt, u32 base)
{
	struct intel_uncore *uncore = gt->uncore;

	intel_uncore_write(uncore, RING_CTL(base), 0);
	intel_uncore_write(uncore, RING_HEAD(base), 0);
	intel_uncore_write(uncore, RING_TAIL(base), 0);
	intel_uncore_write(uncore, RING_START(base), 0);
}

static void init_unused_rings(struct intel_gt *gt)
{
	struct drm_i915_private *i915 = gt->i915;

	if (IS_I830(i915)) {
		init_unused_ring(gt, PRB1_BASE);
		init_unused_ring(gt, SRB0_BASE);
		init_unused_ring(gt, SRB1_BASE);
		init_unused_ring(gt, SRB2_BASE);
		init_unused_ring(gt, SRB3_BASE);
	} else if (GRAPHICS_VER(i915) == 2) {
		init_unused_ring(gt, SRB0_BASE);
		init_unused_ring(gt, SRB1_BASE);
	} else if (GRAPHICS_VER(i915) == 3) {
		init_unused_ring(gt, PRB1_BASE);
		init_unused_ring(gt, PRB2_BASE);
	}
}

int intel_gt_init_hw(struct intel_gt *gt)
{
	struct drm_i915_private *i915 = gt->i915;
	struct intel_uncore *uncore = gt->uncore;
	int ret;

	gt->last_init_time = ktime_get();

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

	if (HAS_EDRAM(i915) && GRAPHICS_VER(i915) < 9)
		intel_uncore_rmw(uncore, HSW_IDICR, 0, IDIHASHMSK(0xf));

	if (IS_HASWELL(i915))
		intel_uncore_write(uncore,
				   HSW_MI_PREDICATE_RESULT_2,
				   IS_HSW_GT3(i915) ?
				   LOWER_SLICE_ENABLED : LOWER_SLICE_DISABLED);

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

	intel_gt_init_swizzling(gt);

	/*
	 * 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(gt);

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

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

	intel_mocs_init(gt);

out:
	intel_uncore_forcewake_put(uncore, FORCEWAKE_ALL);
	return ret;
}

static void gen6_clear_engine_error_register(struct intel_engine_cs *engine)
{
	GEN6_RING_FAULT_REG_RMW(engine, RING_FAULT_VALID, 0);
	GEN6_RING_FAULT_REG_POSTING_READ(engine);
}

i915_reg_t intel_gt_perf_limit_reasons_reg(struct intel_gt *gt)
{
	/* GT0_PERF_LIMIT_REASONS is available only for Gen11+ */
	if (GRAPHICS_VER(gt->i915) < 11)
		return INVALID_MMIO_REG;

	return gt->type == GT_MEDIA ?
		MTL_MEDIA_PERF_LIMIT_REASONS : GT0_PERF_LIMIT_REASONS;
}

void
intel_gt_clear_error_registers(struct intel_gt *gt,
			       intel_engine_mask_t engine_mask)
{
	struct drm_i915_private *i915 = gt->i915;
	struct intel_uncore *uncore = gt->uncore;
	u32 eir;

	if (GRAPHICS_VER(i915) != 2)
		intel_uncore_write(uncore, PGTBL_ER, 0);

	if (GRAPHICS_VER(i915) < 4)
		intel_uncore_write(uncore, IPEIR(RENDER_RING_BASE), 0);
	else
		intel_uncore_write(uncore, IPEIR_I965, 0);

	intel_uncore_write(uncore, EIR, 0);
	eir = intel_uncore_read(uncore, EIR);
	if (eir) {
		/*
		 * some errors might have become stuck,
		 * mask them.
		 */
		gt_dbg(gt, "EIR stuck: 0x%08x, masking\n", eir);
		intel_uncore_rmw(uncore, EMR, 0, eir);
		intel_uncore_write(uncore, GEN2_IIR,
				   I915_MASTER_ERROR_INTERRUPT);
	}

	if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 50)) {
		intel_gt_mcr_multicast_rmw(gt, XEHP_RING_FAULT_REG,
					   RING_FAULT_VALID, 0);
		intel_gt_mcr_read_any(gt, XEHP_RING_FAULT_REG);
	} else if (GRAPHICS_VER(i915) >= 12) {
		intel_uncore_rmw(uncore, GEN12_RING_FAULT_REG, RING_FAULT_VALID, 0);
		intel_uncore_posting_read(uncore, GEN12_RING_FAULT_REG);
	} else if (GRAPHICS_VER(i915) >= 8) {
		intel_uncore_rmw(uncore, GEN8_RING_FAULT_REG, RING_FAULT_VALID, 0);
		intel_uncore_posting_read(uncore, GEN8_RING_FAULT_REG);
	} else if (GRAPHICS_VER(i915) >= 6) {
		struct intel_engine_cs *engine;
		enum intel_engine_id id;

		for_each_engine_masked(engine, gt, engine_mask, id)
			gen6_clear_engine_error_register(engine);
	}
}

static void gen6_check_faults(struct intel_gt *gt)
{
	struct intel_engine_cs *engine;
	enum intel_engine_id id;
	u32 fault;

	for_each_engine(engine, gt, id) {
		fault = GEN6_RING_FAULT_REG_READ(engine);
		if (fault & RING_FAULT_VALID) {
			gt_dbg(gt, "Unexpected fault\n"
			       "\tAddr: 0x%08lx\n"
			       "\tAddress space: %s\n"
			       "\tSource ID: %d\n"
			       "\tType: %d\n",
			       fault & PAGE_MASK,
			       fault & RING_FAULT_GTTSEL_MASK ?
			       "GGTT" : "PPGTT",
			       RING_FAULT_SRCID(fault),
			       RING_FAULT_FAULT_TYPE(fault));
		}
	}
}

static void xehp_check_faults(struct intel_gt *gt)
{
	u32 fault;

	/*
	 * Although the fault register now lives in an MCR register range,
	 * the GAM registers are special and we only truly need to read
	 * the "primary" GAM instance rather than handling each instance
	 * individually.  intel_gt_mcr_read_any() will automatically steer
	 * toward the primary instance.
	 */
	fault = intel_gt_mcr_read_any(gt, XEHP_RING_FAULT_REG);
	if (fault & RING_FAULT_VALID) {
		u32 fault_data0, fault_data1;
		u64 fault_addr;

		fault_data0 = intel_gt_mcr_read_any(gt, XEHP_FAULT_TLB_DATA0);
		fault_data1 = intel_gt_mcr_read_any(gt, XEHP_FAULT_TLB_DATA1);

		fault_addr = ((u64)(fault_data1 & FAULT_VA_HIGH_BITS) << 44) |
			     ((u64)fault_data0 << 12);

		gt_dbg(gt, "Unexpected fault\n"
		       "\tAddr: 0x%08x_%08x\n"
		       "\tAddress space: %s\n"
		       "\tEngine ID: %d\n"
		       "\tSource ID: %d\n"
		       "\tType: %d\n",
		       upper_32_bits(fault_addr), lower_32_bits(fault_addr),
		       fault_data1 & FAULT_GTT_SEL ? "GGTT" : "PPGTT",
		       GEN8_RING_FAULT_ENGINE_ID(fault),
		       RING_FAULT_SRCID(fault),
		       RING_FAULT_FAULT_TYPE(fault));
	}
}

static void gen8_check_faults(struct intel_gt *gt)
{
	struct intel_uncore *uncore = gt->uncore;
	i915_reg_t fault_reg, fault_data0_reg, fault_data1_reg;
	u32 fault;

	if (GRAPHICS_VER(gt->i915) >= 12) {
		fault_reg = GEN12_RING_FAULT_REG;
		fault_data0_reg = GEN12_FAULT_TLB_DATA0;
		fault_data1_reg = GEN12_FAULT_TLB_DATA1;
	} else {
		fault_reg = GEN8_RING_FAULT_REG;
		fault_data0_reg = GEN8_FAULT_TLB_DATA0;
		fault_data1_reg = GEN8_FAULT_TLB_DATA1;
	}

	fault = intel_uncore_read(uncore, fault_reg);
	if (fault & RING_FAULT_VALID) {
		u32 fault_data0, fault_data1;
		u64 fault_addr;

		fault_data0 = intel_uncore_read(uncore, fault_data0_reg);
		fault_data1 = intel_uncore_read(uncore, fault_data1_reg);

		fault_addr = ((u64)(fault_data1 & FAULT_VA_HIGH_BITS) << 44) |
			     ((u64)fault_data0 << 12);

		gt_dbg(gt, "Unexpected fault\n"
		       "\tAddr: 0x%08x_%08x\n"
		       "\tAddress space: %s\n"
		       "\tEngine ID: %d\n"
		       "\tSource ID: %d\n"
		       "\tType: %d\n",
		       upper_32_bits(fault_addr), lower_32_bits(fault_addr),
		       fault_data1 & FAULT_GTT_SEL ? "GGTT" : "PPGTT",
		       GEN8_RING_FAULT_ENGINE_ID(fault),
		       RING_FAULT_SRCID(fault),
		       RING_FAULT_FAULT_TYPE(fault));
	}
}

void intel_gt_check_and_clear_faults(struct intel_gt *gt)
{
	struct drm_i915_private *i915 = gt->i915;

	/* From GEN8 onwards we only have one 'All Engine Fault Register' */
	if (GRAPHICS_VER_FULL(i915) >= IP_VER(12, 50))
		xehp_check_faults(gt);
	else if (GRAPHICS_VER(i915) >= 8)
		gen8_check_faults(gt);
	else if (GRAPHICS_VER(i915) >= 6)
		gen6_check_faults(gt);
	else
		return;

	intel_gt_clear_error_registers(gt, ALL_ENGINES);
}

void intel_gt_flush_ggtt_writes(struct intel_gt *gt)
{
	struct intel_uncore *uncore = gt->uncore;
	intel_wakeref_t wakeref;

	/*
	 * 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(gt->i915)->has_coherent_ggtt)
		return;

	intel_gt_chipset_flush(gt);

	with_intel_runtime_pm_if_in_use(uncore->rpm, wakeref) {
		unsigned long flags;

		spin_lock_irqsave(&uncore->lock, flags);
		intel_uncore_posting_read_fw(uncore,
					     RING_HEAD(RENDER_RING_BASE));
		spin_unlock_irqrestore(&uncore->lock, flags);
	}
}

void intel_gt_chipset_flush(struct intel_gt *gt)
{
	wmb();
	if (GRAPHICS_VER(gt->i915) < 6)
		intel_ggtt_gmch_flush();
}

void intel_gt_driver_register(struct intel_gt *gt)
{
	intel_gsc_init(&gt->gsc, gt->i915);

	intel_rps_driver_register(&gt->rps);

	intel_gt_debugfs_register(gt);
	intel_gt_sysfs_register(gt);
}

static int intel_gt_init_scratch(struct intel_gt *gt, unsigned int size)
{
	struct drm_i915_private *i915 = gt->i915;
	struct drm_i915_gem_object *obj;
	struct i915_vma *vma;
	int ret;

	obj = i915_gem_object_create_lmem(i915, size,
					  I915_BO_ALLOC_VOLATILE |
					  I915_BO_ALLOC_GPU_ONLY);
	if (IS_ERR(obj))
		obj = i915_gem_object_create_stolen(i915, size);
	if (IS_ERR(obj))
		obj = i915_gem_object_create_internal(i915, size);
	if (IS_ERR(obj)) {
		gt_err(gt, "Failed to allocate scratch page\n");
		return PTR_ERR(obj);
	}

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

	ret = i915_ggtt_pin(vma, NULL, 0, PIN_HIGH);
	if (ret)
		goto err_unref;

	gt->scratch = i915_vma_make_unshrinkable(vma);

	return 0;

err_unref:
	i915_gem_object_put(obj);
	return ret;
}

static void intel_gt_fini_scratch(struct intel_gt *gt)
{
	i915_vma_unpin_and_release(&gt->scratch, 0);
}

static struct i915_address_space *kernel_vm(struct intel_gt *gt)
{
	if (INTEL_PPGTT(gt->i915) > INTEL_PPGTT_ALIASING)
		return &i915_ppgtt_create(gt, I915_BO_ALLOC_PM_EARLY)->vm;
	else
		return i915_vm_get(&gt->ggtt->vm);
}

static int __engines_record_defaults(struct intel_gt *gt)
{
	struct i915_request *requests[I915_NUM_ENGINES] = {};
	struct intel_engine_cs *engine;
	enum intel_engine_id id;
	int err = 0;

	/*
	 * 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.
	 */

	for_each_engine(engine, gt, id) {
		struct intel_renderstate so;
		struct intel_context *ce;
		struct i915_request *rq;

		/* We must be able to switch to something! */
		GEM_BUG_ON(!engine->kernel_context);

		ce = intel_context_create(engine);
		if (IS_ERR(ce)) {
			err = PTR_ERR(ce);
			goto out;
		}

		err = intel_renderstate_init(&so, ce);
		if (err)
			goto err;

		rq = i915_request_create(ce);
		if (IS_ERR(rq)) {
			err = PTR_ERR(rq);
			goto err_fini;
		}

		err = intel_engine_emit_ctx_wa(rq);
		if (err)
			goto err_rq;

		err = intel_renderstate_emit(&so, rq);
		if (err)
			goto err_rq;

err_rq:
		requests[id] = i915_request_get(rq);
		i915_request_add(rq);
err_fini:
		intel_renderstate_fini(&so, ce);
err:
		if (err) {
			intel_context_put(ce);
			goto out;
		}
	}

	/* Flush the default context image to memory, and enable powersaving. */
	if (intel_gt_wait_for_idle(gt, I915_GEM_IDLE_TIMEOUT) == -ETIME) {
		err = -EIO;
		goto out;
	}

	for (id = 0; id < ARRAY_SIZE(requests); id++) {
		struct i915_request *rq;
		struct file *state;

		rq = requests[id];
		if (!rq)
			continue;

		if (rq->fence.error) {
			err = -EIO;
			goto out;
		}

		GEM_BUG_ON(!test_bit(CONTEXT_ALLOC_BIT, &rq->context->flags));
		if (!rq->context->state)
			continue;

		/* Keep a copy of the state's backing pages; free the obj */
		state = shmem_create_from_object(rq->context->state->obj);
		if (IS_ERR(state)) {
			err = PTR_ERR(state);
			goto out;
		}
		rq->engine->default_state = state;
	}

out:
	/*
	 * If we have to abandon now, we expect the engines to be idle
	 * and ready to be torn-down. The quickest way we can accomplish
	 * this is by declaring ourselves wedged.
	 */
	if (err)
		intel_gt_set_wedged(gt);

	for (id = 0; id < ARRAY_SIZE(requests); id++) {
		struct intel_context *ce;
		struct i915_request *rq;

		rq = requests[id];
		if (!rq)
			continue;

		ce = rq->context;
		i915_request_put(rq);
		intel_context_put(ce);
	}
	return err;
}

static int __engines_verify_workarounds(struct intel_gt *gt)
{
	struct intel_engine_cs *engine;
	enum intel_engine_id id;
	int err = 0;

	if (!IS_ENABLED(CONFIG_DRM_I915_DEBUG_GEM))
		return 0;

	for_each_engine(engine, gt, id) {
		if (intel_engine_verify_workarounds(engine, "load"))
			err = -EIO;
	}

	/* Flush and restore the kernel context for safety */
	if (intel_gt_wait_for_idle(gt, I915_GEM_IDLE_TIMEOUT) == -ETIME)
		err = -EIO;

	return err;
}

static void __intel_gt_disable(struct intel_gt *gt)
{
	intel_gt_set_wedged_on_fini(gt);

	intel_gt_suspend_prepare(gt);
	intel_gt_suspend_late(gt);

	GEM_BUG_ON(intel_gt_pm_is_awake(gt));
}

int intel_gt_wait_for_idle(struct intel_gt *gt, long timeout)
{
	long remaining_timeout;

	/* If the device is asleep, we have no requests outstanding */
	if (!intel_gt_pm_is_awake(gt))
		return 0;

	while ((timeout = intel_gt_retire_requests_timeout(gt, timeout,
							   &remaining_timeout)) > 0) {
		cond_resched();
		if (signal_pending(current))
			return -EINTR;
	}

	if (timeout)
		return timeout;

	if (remaining_timeout < 0)
		remaining_timeout = 0;

	return intel_uc_wait_for_idle(&gt->uc, remaining_timeout);
}

int intel_gt_init(struct intel_gt *gt)
{
	int err;

	err = i915_inject_probe_error(gt->i915, -ENODEV);
	if (err)
		return err;

	intel_gt_init_workarounds(gt);

	/*
	 * 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.
	 */
	intel_uncore_forcewake_get(gt->uncore, FORCEWAKE_ALL);

	err = intel_gt_init_scratch(gt,
				    GRAPHICS_VER(gt->i915) == 2 ? SZ_256K : SZ_4K);
	if (err)
		goto out_fw;

	intel_gt_pm_init(gt);

	gt->vm = kernel_vm(gt);
	if (!gt->vm) {
		err = -ENOMEM;
		goto err_pm;
	}

	intel_set_mocs_index(gt);

	err = intel_engines_init(gt);
	if (err)
		goto err_engines;

	err = intel_uc_init(&gt->uc);
	if (err)
		goto err_engines;

	err = intel_gt_resume(gt);
	if (err)
		goto err_uc_init;

	err = intel_gt_init_hwconfig(gt);
	if (err)
		gt_err(gt, "Failed to retrieve hwconfig table: %pe\n", ERR_PTR(err));

	err = __engines_record_defaults(gt);
	if (err)
		goto err_gt;

	err = __engines_verify_workarounds(gt);
	if (err)
		goto err_gt;

	err = i915_inject_probe_error(gt->i915, -EIO);
	if (err)
		goto err_gt;

	intel_uc_init_late(&gt->uc);

	intel_migrate_init(&gt->migrate, gt);

	goto out_fw;
err_gt:
	__intel_gt_disable(gt);
	intel_uc_fini_hw(&gt->uc);
err_uc_init:
	intel_uc_fini(&gt->uc);
err_engines:
	intel_engines_release(gt);
	i915_vm_put(fetch_and_zero(&gt->vm));
err_pm:
	intel_gt_pm_fini(gt);
	intel_gt_fini_scratch(gt);
out_fw:
	if (err)
		intel_gt_set_wedged_on_init(gt);
	intel_uncore_forcewake_put(gt->uncore, FORCEWAKE_ALL);
	return err;
}

void intel_gt_driver_remove(struct intel_gt *gt)
{
	__intel_gt_disable(gt);

	intel_migrate_fini(&gt->migrate);
	intel_uc_driver_remove(&gt->uc);

	intel_engines_release(gt);

	intel_gt_flush_buffer_pool(gt);
}

void intel_gt_driver_unregister(struct intel_gt *gt)
{
	intel_wakeref_t wakeref;

	intel_gt_sysfs_unregister(gt);
	intel_rps_driver_unregister(&gt->rps);
	intel_gsc_fini(&gt->gsc);

	/*
	 * If we unload the driver and wedge before the GSC worker is complete,
	 * the worker will hit an error on its submission to the GSC engine and
	 * then exit. This is hard to hit for a user, but it is reproducible
	 * with skipping selftests. The error is handled gracefully by the
	 * worker, so there are no functional issues, but we still end up with
	 * an error message in dmesg, which is something we want to avoid as
	 * this is a supported scenario. We could modify the worker to better
	 * handle a wedging occurring during its execution, but that gets
	 * complicated for a couple of reasons:
	 * - We do want the error on runtime wedging, because there are
	 *   implications for subsystems outside of GT (i.e., PXP, HDCP), it's
	 *   only the error on driver unload that we want to silence.
	 * - The worker is responsible for multiple submissions (GSC FW load,
	 *   HuC auth, SW proxy), so all of those will have to be adapted to
	 *   handle the wedged_on_fini scenario.
	 * Therefore, it's much simpler to just wait for the worker to be done
	 * before wedging on driver removal, also considering that the worker
	 * will likely already be idle in the great majority of non-selftest
	 * scenarios.
	 */
	intel_gsc_uc_flush_work(&gt->uc.gsc);

	/*
	 * Upon unregistering the device to prevent any new users, cancel
	 * all in-flight requests so that we can quickly unbind the active
	 * resources.
	 */
	intel_gt_set_wedged_on_fini(gt);

	/* Scrub all HW state upon release */
	with_intel_runtime_pm(gt->uncore->rpm, wakeref)
		__intel_gt_reset(gt, ALL_ENGINES);
}

void intel_gt_driver_release(struct intel_gt *gt)
{
	struct i915_address_space *vm;

	vm = fetch_and_zero(&gt->vm);
	if (vm) /* FIXME being called twice on error paths :( */
		i915_vm_put(vm);

	intel_wa_list_free(&gt->wa_list);
	intel_gt_pm_fini(gt);
	intel_gt_fini_scratch(gt);
	intel_gt_fini_buffer_pool(gt);
	intel_gt_fini_hwconfig(gt);
}

void intel_gt_driver_late_release_all(struct drm_i915_private *i915)
{
	struct intel_gt *gt;
	unsigned int id;

	/* We need to wait for inflight RCU frees to release their grip */
	rcu_barrier();

	for_each_gt(gt, i915, id) {
		intel_uc_driver_late_release(&gt->uc);
		intel_gt_fini_requests(gt);
		intel_gt_fini_reset(gt);
		intel_gt_fini_timelines(gt);
		mutex_destroy(&gt->tlb.invalidate_lock);
		intel_engines_free(gt);
	}
}

static int intel_gt_tile_setup(struct intel_gt *gt, phys_addr_t phys_addr)
{
	int ret;

	if (!gt_is_root(gt)) {
		struct intel_uncore *uncore;
		spinlock_t *irq_lock;

		uncore = drmm_kzalloc(&gt->i915->drm, sizeof(*uncore), GFP_KERNEL);
		if (!uncore)
			return -ENOMEM;

		irq_lock = drmm_kzalloc(&gt->i915->drm, sizeof(*irq_lock), GFP_KERNEL);
		if (!irq_lock)
			return -ENOMEM;

		gt->uncore = uncore;
		gt->irq_lock = irq_lock;

		intel_gt_common_init_early(gt);
	}

	intel_uncore_init_early(gt->uncore, gt);

	ret = intel_uncore_setup_mmio(gt->uncore, phys_addr);
	if (ret)
		return ret;

	gt->phys_addr = phys_addr;

	return 0;
}

int intel_gt_probe_all(struct drm_i915_private *i915)
{
	struct pci_dev *pdev = to_pci_dev(i915->drm.dev);
	struct intel_gt *gt = &i915->gt0;
	const struct intel_gt_definition *gtdef;
	phys_addr_t phys_addr;
	unsigned int mmio_bar;
	unsigned int i;
	int ret;

	mmio_bar = intel_mmio_bar(GRAPHICS_VER(i915));
	phys_addr = pci_resource_start(pdev, mmio_bar);

	/*
	 * We always have at least one primary GT on any device
	 * and it has been already initialized early during probe
	 * in i915_driver_probe()
	 */
	gt->i915 = i915;
	gt->name = "Primary GT";
	gt->info.engine_mask = RUNTIME_INFO(i915)->platform_engine_mask;

	gt_dbg(gt, "Setting up %s\n", gt->name);
	ret = intel_gt_tile_setup(gt, phys_addr);
	if (ret)
		return ret;

	i915->gt[0] = gt;

	if (!HAS_EXTRA_GT_LIST(i915))
		return 0;

	for (i = 1, gtdef = &INTEL_INFO(i915)->extra_gt_list[i - 1];
	     gtdef->name != NULL;
	     i++, gtdef = &INTEL_INFO(i915)->extra_gt_list[i - 1]) {
		gt = drmm_kzalloc(&i915->drm, sizeof(*gt), GFP_KERNEL);
		if (!gt) {
			ret = -ENOMEM;
			goto err;
		}

		gt->i915 = i915;
		gt->name = gtdef->name;
		gt->type = gtdef->type;
		gt->info.engine_mask = gtdef->engine_mask;
		gt->info.id = i;

		gt_dbg(gt, "Setting up %s\n", gt->name);
		if (GEM_WARN_ON(range_overflows_t(resource_size_t,
						  gtdef->mapping_base,
						  SZ_16M,
						  pci_resource_len(pdev, mmio_bar)))) {
			ret = -ENODEV;
			goto err;
		}

		switch (gtdef->type) {
		case GT_TILE:
			ret = intel_gt_tile_setup(gt, phys_addr + gtdef->mapping_base);
			break;

		case GT_MEDIA:
			ret = intel_sa_mediagt_setup(gt, phys_addr + gtdef->mapping_base,
						     gtdef->gsi_offset);
			break;

		case GT_PRIMARY:
			/* Primary GT should not appear in extra GT list */
		default:
			MISSING_CASE(gtdef->type);
			ret = -ENODEV;
		}

		if (ret)
			goto err;

		i915->gt[i] = gt;
	}

	return 0;

err:
	i915_probe_error(i915, "Failed to initialize %s! (%d)\n", gtdef->name, ret);
	intel_gt_release_all(i915);

	return ret;
}

int intel_gt_tiles_init(struct drm_i915_private *i915)
{
	struct intel_gt *gt;
	unsigned int id;
	int ret;

	for_each_gt(gt, i915, id) {
		ret = intel_gt_probe_lmem(gt);
		if (ret)
			return ret;
	}

	return 0;
}

void intel_gt_release_all(struct drm_i915_private *i915)
{
	struct intel_gt *gt;
	unsigned int id;

	for_each_gt(gt, i915, id)
		i915->gt[id] = NULL;
}

void intel_gt_info_print(const struct intel_gt_info *info,
			 struct drm_printer *p)
{
	drm_printf(p, "available engines: %x\n", info->engine_mask);

	intel_sseu_dump(&info->sseu, p);
}

/*
 * HW architecture suggest typical invalidation time at 40us,
 * with pessimistic cases up to 100us and a recommendation to
 * cap at 1ms. We go a bit higher just in case.
 */
#define TLB_INVAL_TIMEOUT_US 100
#define TLB_INVAL_TIMEOUT_MS 4

/*
 * On Xe_HP the TLB invalidation registers are located at the same MMIO offsets
 * but are now considered MCR registers.  Since they exist within a GAM range,
 * the primary instance of the register rolls up the status from each unit.
 */
static int wait_for_invalidate(struct intel_engine_cs *engine)
{
	if (engine->tlb_inv.mcr)
		return intel_gt_mcr_wait_for_reg(engine->gt,
						 engine->tlb_inv.reg.mcr_reg,
						 engine->tlb_inv.done,
						 0,
						 TLB_INVAL_TIMEOUT_US,
						 TLB_INVAL_TIMEOUT_MS);
	else
		return __intel_wait_for_register_fw(engine->gt->uncore,
						    engine->tlb_inv.reg.reg,
						    engine->tlb_inv.done,
						    0,
						    TLB_INVAL_TIMEOUT_US,
						    TLB_INVAL_TIMEOUT_MS,
						    NULL);
}

static void mmio_invalidate_full(struct intel_gt *gt)
{
	struct drm_i915_private *i915 = gt->i915;
	struct intel_uncore *uncore = gt->uncore;
	struct intel_engine_cs *engine;
	intel_engine_mask_t awake, tmp;
	enum intel_engine_id id;
	unsigned long flags;

	if (GRAPHICS_VER(i915) < 8)
		return;

	intel_uncore_forcewake_get(uncore, FORCEWAKE_ALL);

	intel_gt_mcr_lock(gt, &flags);
	spin_lock(&uncore->lock); /* serialise invalidate with GT reset */

	awake = 0;
	for_each_engine(engine, gt, id) {
		if (!intel_engine_pm_is_awake(engine))
			continue;

		if (engine->tlb_inv.mcr)
			intel_gt_mcr_multicast_write_fw(gt,
							engine->tlb_inv.reg.mcr_reg,
							engine->tlb_inv.request);
		else
			intel_uncore_write_fw(uncore,
					      engine->tlb_inv.reg.reg,
					      engine->tlb_inv.request);

		awake |= engine->mask;
	}

	GT_TRACE(gt, "invalidated engines %08x\n", awake);

	/* Wa_2207587034:tgl,dg1,rkl,adl-s,adl-p */
	if (awake &&
	    (IS_TIGERLAKE(i915) ||
	     IS_DG1(i915) ||
	     IS_ROCKETLAKE(i915) ||
	     IS_ALDERLAKE_S(i915) ||
	     IS_ALDERLAKE_P(i915)))
		intel_uncore_write_fw(uncore, GEN12_OA_TLB_INV_CR, 1);

	spin_unlock(&uncore->lock);
	intel_gt_mcr_unlock(gt, flags);

	for_each_engine_masked(engine, gt, awake, tmp) {
		if (wait_for_invalidate(engine))
			gt_err_ratelimited(gt,
					   "%s TLB invalidation did not complete in %ums!\n",
					   engine->name, TLB_INVAL_TIMEOUT_MS);
	}

	/*
	 * Use delayed put since a) we mostly expect a flurry of TLB
	 * invalidations so it is good to avoid paying the forcewake cost and
	 * b) it works around a bug in Icelake which cannot cope with too rapid
	 * transitions.
	 */
	intel_uncore_forcewake_put_delayed(uncore, FORCEWAKE_ALL);
}

static bool tlb_seqno_passed(const struct intel_gt *gt, u32 seqno)
{
	u32 cur = intel_gt_tlb_seqno(gt);

	/* Only skip if a *full* TLB invalidate barrier has passed */
	return (s32)(cur - ALIGN(seqno, 2)) > 0;
}

void intel_gt_invalidate_tlb(struct intel_gt *gt, u32 seqno)
{
	intel_wakeref_t wakeref;

	if (I915_SELFTEST_ONLY(gt->awake == -ENODEV))
		return;

	if (intel_gt_is_wedged(gt))
		return;

	if (tlb_seqno_passed(gt, seqno))
		return;

	with_intel_gt_pm_if_awake(gt, wakeref) {
		mutex_lock(&gt->tlb.invalidate_lock);
		if (tlb_seqno_passed(gt, seqno))
			goto unlock;

		mmio_invalidate_full(gt);

		write_seqcount_invalidate(&gt->tlb.seqno);
unlock:
		mutex_unlock(&gt->tlb.invalidate_lock);
	}
}

#if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
#include "selftest_tlb.c"
#endif