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
|
/*
* Copyright 2014 Advanced Micro Devices, Inc.
*
* 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 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 COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
* OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
* ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
* OTHER DEALINGS IN THE SOFTWARE.
*/
#include <linux/bsearch.h>
#include <linux/pci.h>
#include <linux/slab.h>
#include "kfd_priv.h"
#include "kfd_device_queue_manager.h"
#include "kfd_pm4_headers_vi.h"
#include "cwsr_trap_handler.h"
#include "kfd_iommu.h"
#define MQD_SIZE_ALIGNED 768
/*
* kfd_locked is used to lock the kfd driver during suspend or reset
* once locked, kfd driver will stop any further GPU execution.
* create process (open) will return -EAGAIN.
*/
static atomic_t kfd_locked = ATOMIC_INIT(0);
#ifdef KFD_SUPPORT_IOMMU_V2
static const struct kfd_device_info kaveri_device_info = {
.asic_family = CHIP_KAVERI,
.max_pasid_bits = 16,
/* max num of queues for KV.TODO should be a dynamic value */
.max_no_of_hqd = 24,
.doorbell_size = 4,
.ih_ring_entry_size = 4 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_cik,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = false,
.needs_iommu_device = true,
.needs_pci_atomics = false,
.num_sdma_engines = 2,
};
static const struct kfd_device_info carrizo_device_info = {
.asic_family = CHIP_CARRIZO,
.max_pasid_bits = 16,
/* max num of queues for CZ.TODO should be a dynamic value */
.max_no_of_hqd = 24,
.doorbell_size = 4,
.ih_ring_entry_size = 4 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_cik,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = true,
.needs_iommu_device = true,
.needs_pci_atomics = false,
.num_sdma_engines = 2,
};
static const struct kfd_device_info raven_device_info = {
.asic_family = CHIP_RAVEN,
.max_pasid_bits = 16,
.max_no_of_hqd = 24,
.doorbell_size = 8,
.ih_ring_entry_size = 8 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_v9,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = true,
.needs_iommu_device = true,
.needs_pci_atomics = true,
.num_sdma_engines = 1,
};
#endif
static const struct kfd_device_info hawaii_device_info = {
.asic_family = CHIP_HAWAII,
.max_pasid_bits = 16,
/* max num of queues for KV.TODO should be a dynamic value */
.max_no_of_hqd = 24,
.doorbell_size = 4,
.ih_ring_entry_size = 4 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_cik,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = false,
.needs_iommu_device = false,
.needs_pci_atomics = false,
.num_sdma_engines = 2,
};
static const struct kfd_device_info tonga_device_info = {
.asic_family = CHIP_TONGA,
.max_pasid_bits = 16,
.max_no_of_hqd = 24,
.doorbell_size = 4,
.ih_ring_entry_size = 4 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_cik,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = false,
.needs_iommu_device = false,
.needs_pci_atomics = true,
.num_sdma_engines = 2,
};
static const struct kfd_device_info tonga_vf_device_info = {
.asic_family = CHIP_TONGA,
.max_pasid_bits = 16,
.max_no_of_hqd = 24,
.doorbell_size = 4,
.ih_ring_entry_size = 4 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_cik,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = false,
.needs_iommu_device = false,
.needs_pci_atomics = false,
.num_sdma_engines = 2,
};
static const struct kfd_device_info fiji_device_info = {
.asic_family = CHIP_FIJI,
.max_pasid_bits = 16,
.max_no_of_hqd = 24,
.doorbell_size = 4,
.ih_ring_entry_size = 4 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_cik,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = true,
.needs_iommu_device = false,
.needs_pci_atomics = true,
.num_sdma_engines = 2,
};
static const struct kfd_device_info fiji_vf_device_info = {
.asic_family = CHIP_FIJI,
.max_pasid_bits = 16,
.max_no_of_hqd = 24,
.doorbell_size = 4,
.ih_ring_entry_size = 4 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_cik,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = true,
.needs_iommu_device = false,
.needs_pci_atomics = false,
.num_sdma_engines = 2,
};
static const struct kfd_device_info polaris10_device_info = {
.asic_family = CHIP_POLARIS10,
.max_pasid_bits = 16,
.max_no_of_hqd = 24,
.doorbell_size = 4,
.ih_ring_entry_size = 4 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_cik,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = true,
.needs_iommu_device = false,
.needs_pci_atomics = true,
.num_sdma_engines = 2,
};
static const struct kfd_device_info polaris10_vf_device_info = {
.asic_family = CHIP_POLARIS10,
.max_pasid_bits = 16,
.max_no_of_hqd = 24,
.doorbell_size = 4,
.ih_ring_entry_size = 4 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_cik,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = true,
.needs_iommu_device = false,
.needs_pci_atomics = false,
.num_sdma_engines = 2,
};
static const struct kfd_device_info polaris11_device_info = {
.asic_family = CHIP_POLARIS11,
.max_pasid_bits = 16,
.max_no_of_hqd = 24,
.doorbell_size = 4,
.ih_ring_entry_size = 4 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_cik,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = true,
.needs_iommu_device = false,
.needs_pci_atomics = true,
.num_sdma_engines = 2,
};
static const struct kfd_device_info vega10_device_info = {
.asic_family = CHIP_VEGA10,
.max_pasid_bits = 16,
.max_no_of_hqd = 24,
.doorbell_size = 8,
.ih_ring_entry_size = 8 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_v9,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = true,
.needs_iommu_device = false,
.needs_pci_atomics = false,
.num_sdma_engines = 2,
};
static const struct kfd_device_info vega10_vf_device_info = {
.asic_family = CHIP_VEGA10,
.max_pasid_bits = 16,
.max_no_of_hqd = 24,
.doorbell_size = 8,
.ih_ring_entry_size = 8 * sizeof(uint32_t),
.event_interrupt_class = &event_interrupt_class_v9,
.num_of_watch_points = 4,
.mqd_size_aligned = MQD_SIZE_ALIGNED,
.supports_cwsr = true,
.needs_iommu_device = false,
.needs_pci_atomics = false,
.num_sdma_engines = 2,
};
struct kfd_deviceid {
unsigned short did;
const struct kfd_device_info *device_info;
};
static const struct kfd_deviceid supported_devices[] = {
#ifdef KFD_SUPPORT_IOMMU_V2
{ 0x1304, &kaveri_device_info }, /* Kaveri */
{ 0x1305, &kaveri_device_info }, /* Kaveri */
{ 0x1306, &kaveri_device_info }, /* Kaveri */
{ 0x1307, &kaveri_device_info }, /* Kaveri */
{ 0x1309, &kaveri_device_info }, /* Kaveri */
{ 0x130A, &kaveri_device_info }, /* Kaveri */
{ 0x130B, &kaveri_device_info }, /* Kaveri */
{ 0x130C, &kaveri_device_info }, /* Kaveri */
{ 0x130D, &kaveri_device_info }, /* Kaveri */
{ 0x130E, &kaveri_device_info }, /* Kaveri */
{ 0x130F, &kaveri_device_info }, /* Kaveri */
{ 0x1310, &kaveri_device_info }, /* Kaveri */
{ 0x1311, &kaveri_device_info }, /* Kaveri */
{ 0x1312, &kaveri_device_info }, /* Kaveri */
{ 0x1313, &kaveri_device_info }, /* Kaveri */
{ 0x1315, &kaveri_device_info }, /* Kaveri */
{ 0x1316, &kaveri_device_info }, /* Kaveri */
{ 0x1317, &kaveri_device_info }, /* Kaveri */
{ 0x1318, &kaveri_device_info }, /* Kaveri */
{ 0x131B, &kaveri_device_info }, /* Kaveri */
{ 0x131C, &kaveri_device_info }, /* Kaveri */
{ 0x131D, &kaveri_device_info }, /* Kaveri */
{ 0x9870, &carrizo_device_info }, /* Carrizo */
{ 0x9874, &carrizo_device_info }, /* Carrizo */
{ 0x9875, &carrizo_device_info }, /* Carrizo */
{ 0x9876, &carrizo_device_info }, /* Carrizo */
{ 0x9877, &carrizo_device_info }, /* Carrizo */
{ 0x15DD, &raven_device_info }, /* Raven */
#endif
{ 0x67A0, &hawaii_device_info }, /* Hawaii */
{ 0x67A1, &hawaii_device_info }, /* Hawaii */
{ 0x67A2, &hawaii_device_info }, /* Hawaii */
{ 0x67A8, &hawaii_device_info }, /* Hawaii */
{ 0x67A9, &hawaii_device_info }, /* Hawaii */
{ 0x67AA, &hawaii_device_info }, /* Hawaii */
{ 0x67B0, &hawaii_device_info }, /* Hawaii */
{ 0x67B1, &hawaii_device_info }, /* Hawaii */
{ 0x67B8, &hawaii_device_info }, /* Hawaii */
{ 0x67B9, &hawaii_device_info }, /* Hawaii */
{ 0x67BA, &hawaii_device_info }, /* Hawaii */
{ 0x67BE, &hawaii_device_info }, /* Hawaii */
{ 0x6920, &tonga_device_info }, /* Tonga */
{ 0x6921, &tonga_device_info }, /* Tonga */
{ 0x6928, &tonga_device_info }, /* Tonga */
{ 0x6929, &tonga_device_info }, /* Tonga */
{ 0x692B, &tonga_device_info }, /* Tonga */
{ 0x692F, &tonga_vf_device_info }, /* Tonga vf */
{ 0x6938, &tonga_device_info }, /* Tonga */
{ 0x6939, &tonga_device_info }, /* Tonga */
{ 0x7300, &fiji_device_info }, /* Fiji */
{ 0x730F, &fiji_vf_device_info }, /* Fiji vf*/
{ 0x67C0, &polaris10_device_info }, /* Polaris10 */
{ 0x67C1, &polaris10_device_info }, /* Polaris10 */
{ 0x67C2, &polaris10_device_info }, /* Polaris10 */
{ 0x67C4, &polaris10_device_info }, /* Polaris10 */
{ 0x67C7, &polaris10_device_info }, /* Polaris10 */
{ 0x67C8, &polaris10_device_info }, /* Polaris10 */
{ 0x67C9, &polaris10_device_info }, /* Polaris10 */
{ 0x67CA, &polaris10_device_info }, /* Polaris10 */
{ 0x67CC, &polaris10_device_info }, /* Polaris10 */
{ 0x67CF, &polaris10_device_info }, /* Polaris10 */
{ 0x67D0, &polaris10_vf_device_info }, /* Polaris10 vf*/
{ 0x67DF, &polaris10_device_info }, /* Polaris10 */
{ 0x67E0, &polaris11_device_info }, /* Polaris11 */
{ 0x67E1, &polaris11_device_info }, /* Polaris11 */
{ 0x67E3, &polaris11_device_info }, /* Polaris11 */
{ 0x67E7, &polaris11_device_info }, /* Polaris11 */
{ 0x67E8, &polaris11_device_info }, /* Polaris11 */
{ 0x67E9, &polaris11_device_info }, /* Polaris11 */
{ 0x67EB, &polaris11_device_info }, /* Polaris11 */
{ 0x67EF, &polaris11_device_info }, /* Polaris11 */
{ 0x67FF, &polaris11_device_info }, /* Polaris11 */
{ 0x6860, &vega10_device_info }, /* Vega10 */
{ 0x6861, &vega10_device_info }, /* Vega10 */
{ 0x6862, &vega10_device_info }, /* Vega10 */
{ 0x6863, &vega10_device_info }, /* Vega10 */
{ 0x6864, &vega10_device_info }, /* Vega10 */
{ 0x6867, &vega10_device_info }, /* Vega10 */
{ 0x6868, &vega10_device_info }, /* Vega10 */
{ 0x686C, &vega10_vf_device_info }, /* Vega10 vf*/
{ 0x687F, &vega10_device_info }, /* Vega10 */
};
static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size,
unsigned int chunk_size);
static void kfd_gtt_sa_fini(struct kfd_dev *kfd);
static int kfd_resume(struct kfd_dev *kfd);
static const struct kfd_device_info *lookup_device_info(unsigned short did)
{
size_t i;
for (i = 0; i < ARRAY_SIZE(supported_devices); i++) {
if (supported_devices[i].did == did) {
WARN_ON(!supported_devices[i].device_info);
return supported_devices[i].device_info;
}
}
dev_warn(kfd_device, "DID %04x is missing in supported_devices\n",
did);
return NULL;
}
struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd,
struct pci_dev *pdev, const struct kfd2kgd_calls *f2g)
{
struct kfd_dev *kfd;
int ret;
const struct kfd_device_info *device_info =
lookup_device_info(pdev->device);
if (!device_info) {
dev_err(kfd_device, "kgd2kfd_probe failed\n");
return NULL;
}
/* Allow BIF to recode atomics to PCIe 3.0 AtomicOps.
* 32 and 64-bit requests are possible and must be
* supported.
*/
ret = pci_enable_atomic_ops_to_root(pdev,
PCI_EXP_DEVCAP2_ATOMIC_COMP32 |
PCI_EXP_DEVCAP2_ATOMIC_COMP64);
if (device_info->needs_pci_atomics && ret < 0) {
dev_info(kfd_device,
"skipped device %x:%x, PCI rejects atomics\n",
pdev->vendor, pdev->device);
return NULL;
}
kfd = kzalloc(sizeof(*kfd), GFP_KERNEL);
if (!kfd)
return NULL;
kfd->kgd = kgd;
kfd->device_info = device_info;
kfd->pdev = pdev;
kfd->init_complete = false;
kfd->kfd2kgd = f2g;
mutex_init(&kfd->doorbell_mutex);
memset(&kfd->doorbell_available_index, 0,
sizeof(kfd->doorbell_available_index));
return kfd;
}
static void kfd_cwsr_init(struct kfd_dev *kfd)
{
if (cwsr_enable && kfd->device_info->supports_cwsr) {
if (kfd->device_info->asic_family < CHIP_VEGA10) {
BUILD_BUG_ON(sizeof(cwsr_trap_gfx8_hex) > PAGE_SIZE);
kfd->cwsr_isa = cwsr_trap_gfx8_hex;
kfd->cwsr_isa_size = sizeof(cwsr_trap_gfx8_hex);
} else {
BUILD_BUG_ON(sizeof(cwsr_trap_gfx9_hex) > PAGE_SIZE);
kfd->cwsr_isa = cwsr_trap_gfx9_hex;
kfd->cwsr_isa_size = sizeof(cwsr_trap_gfx9_hex);
}
kfd->cwsr_enabled = true;
}
}
bool kgd2kfd_device_init(struct kfd_dev *kfd,
const struct kgd2kfd_shared_resources *gpu_resources)
{
unsigned int size;
kfd->shared_resources = *gpu_resources;
kfd->vm_info.first_vmid_kfd = ffs(gpu_resources->compute_vmid_bitmap)-1;
kfd->vm_info.last_vmid_kfd = fls(gpu_resources->compute_vmid_bitmap)-1;
kfd->vm_info.vmid_num_kfd = kfd->vm_info.last_vmid_kfd
- kfd->vm_info.first_vmid_kfd + 1;
/* Verify module parameters regarding mapped process number*/
if ((hws_max_conc_proc < 0)
|| (hws_max_conc_proc > kfd->vm_info.vmid_num_kfd)) {
dev_err(kfd_device,
"hws_max_conc_proc %d must be between 0 and %d, use %d instead\n",
hws_max_conc_proc, kfd->vm_info.vmid_num_kfd,
kfd->vm_info.vmid_num_kfd);
kfd->max_proc_per_quantum = kfd->vm_info.vmid_num_kfd;
} else
kfd->max_proc_per_quantum = hws_max_conc_proc;
/* calculate max size of mqds needed for queues */
size = max_num_of_queues_per_device *
kfd->device_info->mqd_size_aligned;
/*
* calculate max size of runlist packet.
* There can be only 2 packets at once
*/
size += (KFD_MAX_NUM_OF_PROCESSES * sizeof(struct pm4_mes_map_process) +
max_num_of_queues_per_device * sizeof(struct pm4_mes_map_queues)
+ sizeof(struct pm4_mes_runlist)) * 2;
/* Add size of HIQ & DIQ */
size += KFD_KERNEL_QUEUE_SIZE * 2;
/* add another 512KB for all other allocations on gart (HPD, fences) */
size += 512 * 1024;
if (kfd->kfd2kgd->init_gtt_mem_allocation(
kfd->kgd, size, &kfd->gtt_mem,
&kfd->gtt_start_gpu_addr, &kfd->gtt_start_cpu_ptr,
false)) {
dev_err(kfd_device, "Could not allocate %d bytes\n", size);
goto out;
}
dev_info(kfd_device, "Allocated %d bytes on gart\n", size);
/* Initialize GTT sa with 512 byte chunk size */
if (kfd_gtt_sa_init(kfd, size, 512) != 0) {
dev_err(kfd_device, "Error initializing gtt sub-allocator\n");
goto kfd_gtt_sa_init_error;
}
if (kfd_doorbell_init(kfd)) {
dev_err(kfd_device,
"Error initializing doorbell aperture\n");
goto kfd_doorbell_error;
}
if (kfd->kfd2kgd->get_hive_id)
kfd->hive_id = kfd->kfd2kgd->get_hive_id(kfd->kgd);
if (kfd_topology_add_device(kfd)) {
dev_err(kfd_device, "Error adding device to topology\n");
goto kfd_topology_add_device_error;
}
if (kfd_interrupt_init(kfd)) {
dev_err(kfd_device, "Error initializing interrupts\n");
goto kfd_interrupt_error;
}
kfd->dqm = device_queue_manager_init(kfd);
if (!kfd->dqm) {
dev_err(kfd_device, "Error initializing queue manager\n");
goto device_queue_manager_error;
}
if (kfd_iommu_device_init(kfd)) {
dev_err(kfd_device, "Error initializing iommuv2\n");
goto device_iommu_error;
}
kfd_cwsr_init(kfd);
if (kfd_resume(kfd))
goto kfd_resume_error;
kfd->dbgmgr = NULL;
kfd->init_complete = true;
dev_info(kfd_device, "added device %x:%x\n", kfd->pdev->vendor,
kfd->pdev->device);
pr_debug("Starting kfd with the following scheduling policy %d\n",
kfd->dqm->sched_policy);
goto out;
kfd_resume_error:
device_iommu_error:
device_queue_manager_uninit(kfd->dqm);
device_queue_manager_error:
kfd_interrupt_exit(kfd);
kfd_interrupt_error:
kfd_topology_remove_device(kfd);
kfd_topology_add_device_error:
kfd_doorbell_fini(kfd);
kfd_doorbell_error:
kfd_gtt_sa_fini(kfd);
kfd_gtt_sa_init_error:
kfd->kfd2kgd->free_gtt_mem(kfd->kgd, kfd->gtt_mem);
dev_err(kfd_device,
"device %x:%x NOT added due to errors\n",
kfd->pdev->vendor, kfd->pdev->device);
out:
return kfd->init_complete;
}
void kgd2kfd_device_exit(struct kfd_dev *kfd)
{
if (kfd->init_complete) {
kgd2kfd_suspend(kfd);
device_queue_manager_uninit(kfd->dqm);
kfd_interrupt_exit(kfd);
kfd_topology_remove_device(kfd);
kfd_doorbell_fini(kfd);
kfd_gtt_sa_fini(kfd);
kfd->kfd2kgd->free_gtt_mem(kfd->kgd, kfd->gtt_mem);
}
kfree(kfd);
}
int kgd2kfd_pre_reset(struct kfd_dev *kfd)
{
if (!kfd->init_complete)
return 0;
kgd2kfd_suspend(kfd);
/* hold dqm->lock to prevent further execution*/
dqm_lock(kfd->dqm);
kfd_signal_reset_event(kfd);
return 0;
}
/*
* Fix me. KFD won't be able to resume existing process for now.
* We will keep all existing process in a evicted state and
* wait the process to be terminated.
*/
int kgd2kfd_post_reset(struct kfd_dev *kfd)
{
int ret, count;
if (!kfd->init_complete)
return 0;
dqm_unlock(kfd->dqm);
ret = kfd_resume(kfd);
if (ret)
return ret;
count = atomic_dec_return(&kfd_locked);
WARN_ONCE(count != 0, "KFD reset ref. error");
return 0;
}
bool kfd_is_locked(void)
{
return (atomic_read(&kfd_locked) > 0);
}
void kgd2kfd_suspend(struct kfd_dev *kfd)
{
if (!kfd->init_complete)
return;
/* For first KFD device suspend all the KFD processes */
if (atomic_inc_return(&kfd_locked) == 1)
kfd_suspend_all_processes();
kfd->dqm->ops.stop(kfd->dqm);
kfd_iommu_suspend(kfd);
}
int kgd2kfd_resume(struct kfd_dev *kfd)
{
int ret, count;
if (!kfd->init_complete)
return 0;
ret = kfd_resume(kfd);
if (ret)
return ret;
count = atomic_dec_return(&kfd_locked);
WARN_ONCE(count < 0, "KFD suspend / resume ref. error");
if (count == 0)
ret = kfd_resume_all_processes();
return ret;
}
static int kfd_resume(struct kfd_dev *kfd)
{
int err = 0;
err = kfd_iommu_resume(kfd);
if (err) {
dev_err(kfd_device,
"Failed to resume IOMMU for device %x:%x\n",
kfd->pdev->vendor, kfd->pdev->device);
return err;
}
err = kfd->dqm->ops.start(kfd->dqm);
if (err) {
dev_err(kfd_device,
"Error starting queue manager for device %x:%x\n",
kfd->pdev->vendor, kfd->pdev->device);
goto dqm_start_error;
}
return err;
dqm_start_error:
kfd_iommu_suspend(kfd);
return err;
}
/* This is called directly from KGD at ISR. */
void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry)
{
uint32_t patched_ihre[KFD_MAX_RING_ENTRY_SIZE];
bool is_patched = false;
if (!kfd->init_complete)
return;
if (kfd->device_info->ih_ring_entry_size > sizeof(patched_ihre)) {
dev_err_once(kfd_device, "Ring entry too small\n");
return;
}
spin_lock(&kfd->interrupt_lock);
if (kfd->interrupts_active
&& interrupt_is_wanted(kfd, ih_ring_entry,
patched_ihre, &is_patched)
&& enqueue_ih_ring_entry(kfd,
is_patched ? patched_ihre : ih_ring_entry))
queue_work(kfd->ih_wq, &kfd->interrupt_work);
spin_unlock(&kfd->interrupt_lock);
}
int kgd2kfd_quiesce_mm(struct mm_struct *mm)
{
struct kfd_process *p;
int r;
/* Because we are called from arbitrary context (workqueue) as opposed
* to process context, kfd_process could attempt to exit while we are
* running so the lookup function increments the process ref count.
*/
p = kfd_lookup_process_by_mm(mm);
if (!p)
return -ESRCH;
r = kfd_process_evict_queues(p);
kfd_unref_process(p);
return r;
}
int kgd2kfd_resume_mm(struct mm_struct *mm)
{
struct kfd_process *p;
int r;
/* Because we are called from arbitrary context (workqueue) as opposed
* to process context, kfd_process could attempt to exit while we are
* running so the lookup function increments the process ref count.
*/
p = kfd_lookup_process_by_mm(mm);
if (!p)
return -ESRCH;
r = kfd_process_restore_queues(p);
kfd_unref_process(p);
return r;
}
/** kgd2kfd_schedule_evict_and_restore_process - Schedules work queue that will
* prepare for safe eviction of KFD BOs that belong to the specified
* process.
*
* @mm: mm_struct that identifies the specified KFD process
* @fence: eviction fence attached to KFD process BOs
*
*/
int kgd2kfd_schedule_evict_and_restore_process(struct mm_struct *mm,
struct dma_fence *fence)
{
struct kfd_process *p;
unsigned long active_time;
unsigned long delay_jiffies = msecs_to_jiffies(PROCESS_ACTIVE_TIME_MS);
if (!fence)
return -EINVAL;
if (dma_fence_is_signaled(fence))
return 0;
p = kfd_lookup_process_by_mm(mm);
if (!p)
return -ENODEV;
if (fence->seqno == p->last_eviction_seqno)
goto out;
p->last_eviction_seqno = fence->seqno;
/* Avoid KFD process starvation. Wait for at least
* PROCESS_ACTIVE_TIME_MS before evicting the process again
*/
active_time = get_jiffies_64() - p->last_restore_timestamp;
if (delay_jiffies > active_time)
delay_jiffies -= active_time;
else
delay_jiffies = 0;
/* During process initialization eviction_work.dwork is initialized
* to kfd_evict_bo_worker
*/
schedule_delayed_work(&p->eviction_work, delay_jiffies);
out:
kfd_unref_process(p);
return 0;
}
static int kfd_gtt_sa_init(struct kfd_dev *kfd, unsigned int buf_size,
unsigned int chunk_size)
{
unsigned int num_of_longs;
if (WARN_ON(buf_size < chunk_size))
return -EINVAL;
if (WARN_ON(buf_size == 0))
return -EINVAL;
if (WARN_ON(chunk_size == 0))
return -EINVAL;
kfd->gtt_sa_chunk_size = chunk_size;
kfd->gtt_sa_num_of_chunks = buf_size / chunk_size;
num_of_longs = (kfd->gtt_sa_num_of_chunks + BITS_PER_LONG - 1) /
BITS_PER_LONG;
kfd->gtt_sa_bitmap = kcalloc(num_of_longs, sizeof(long), GFP_KERNEL);
if (!kfd->gtt_sa_bitmap)
return -ENOMEM;
pr_debug("gtt_sa_num_of_chunks = %d, gtt_sa_bitmap = %p\n",
kfd->gtt_sa_num_of_chunks, kfd->gtt_sa_bitmap);
mutex_init(&kfd->gtt_sa_lock);
return 0;
}
static void kfd_gtt_sa_fini(struct kfd_dev *kfd)
{
mutex_destroy(&kfd->gtt_sa_lock);
kfree(kfd->gtt_sa_bitmap);
}
static inline uint64_t kfd_gtt_sa_calc_gpu_addr(uint64_t start_addr,
unsigned int bit_num,
unsigned int chunk_size)
{
return start_addr + bit_num * chunk_size;
}
static inline uint32_t *kfd_gtt_sa_calc_cpu_addr(void *start_addr,
unsigned int bit_num,
unsigned int chunk_size)
{
return (uint32_t *) ((uint64_t) start_addr + bit_num * chunk_size);
}
int kfd_gtt_sa_allocate(struct kfd_dev *kfd, unsigned int size,
struct kfd_mem_obj **mem_obj)
{
unsigned int found, start_search, cur_size;
if (size == 0)
return -EINVAL;
if (size > kfd->gtt_sa_num_of_chunks * kfd->gtt_sa_chunk_size)
return -ENOMEM;
*mem_obj = kzalloc(sizeof(struct kfd_mem_obj), GFP_KERNEL);
if (!(*mem_obj))
return -ENOMEM;
pr_debug("Allocated mem_obj = %p for size = %d\n", *mem_obj, size);
start_search = 0;
mutex_lock(&kfd->gtt_sa_lock);
kfd_gtt_restart_search:
/* Find the first chunk that is free */
found = find_next_zero_bit(kfd->gtt_sa_bitmap,
kfd->gtt_sa_num_of_chunks,
start_search);
pr_debug("Found = %d\n", found);
/* If there wasn't any free chunk, bail out */
if (found == kfd->gtt_sa_num_of_chunks)
goto kfd_gtt_no_free_chunk;
/* Update fields of mem_obj */
(*mem_obj)->range_start = found;
(*mem_obj)->range_end = found;
(*mem_obj)->gpu_addr = kfd_gtt_sa_calc_gpu_addr(
kfd->gtt_start_gpu_addr,
found,
kfd->gtt_sa_chunk_size);
(*mem_obj)->cpu_ptr = kfd_gtt_sa_calc_cpu_addr(
kfd->gtt_start_cpu_ptr,
found,
kfd->gtt_sa_chunk_size);
pr_debug("gpu_addr = %p, cpu_addr = %p\n",
(uint64_t *) (*mem_obj)->gpu_addr, (*mem_obj)->cpu_ptr);
/* If we need only one chunk, mark it as allocated and get out */
if (size <= kfd->gtt_sa_chunk_size) {
pr_debug("Single bit\n");
set_bit(found, kfd->gtt_sa_bitmap);
goto kfd_gtt_out;
}
/* Otherwise, try to see if we have enough contiguous chunks */
cur_size = size - kfd->gtt_sa_chunk_size;
do {
(*mem_obj)->range_end =
find_next_zero_bit(kfd->gtt_sa_bitmap,
kfd->gtt_sa_num_of_chunks, ++found);
/*
* If next free chunk is not contiguous than we need to
* restart our search from the last free chunk we found (which
* wasn't contiguous to the previous ones
*/
if ((*mem_obj)->range_end != found) {
start_search = found;
goto kfd_gtt_restart_search;
}
/*
* If we reached end of buffer, bail out with error
*/
if (found == kfd->gtt_sa_num_of_chunks)
goto kfd_gtt_no_free_chunk;
/* Check if we don't need another chunk */
if (cur_size <= kfd->gtt_sa_chunk_size)
cur_size = 0;
else
cur_size -= kfd->gtt_sa_chunk_size;
} while (cur_size > 0);
pr_debug("range_start = %d, range_end = %d\n",
(*mem_obj)->range_start, (*mem_obj)->range_end);
/* Mark the chunks as allocated */
for (found = (*mem_obj)->range_start;
found <= (*mem_obj)->range_end;
found++)
set_bit(found, kfd->gtt_sa_bitmap);
kfd_gtt_out:
mutex_unlock(&kfd->gtt_sa_lock);
return 0;
kfd_gtt_no_free_chunk:
pr_debug("Allocation failed with mem_obj = %p\n", mem_obj);
mutex_unlock(&kfd->gtt_sa_lock);
kfree(mem_obj);
return -ENOMEM;
}
int kfd_gtt_sa_free(struct kfd_dev *kfd, struct kfd_mem_obj *mem_obj)
{
unsigned int bit;
/* Act like kfree when trying to free a NULL object */
if (!mem_obj)
return 0;
pr_debug("Free mem_obj = %p, range_start = %d, range_end = %d\n",
mem_obj, mem_obj->range_start, mem_obj->range_end);
mutex_lock(&kfd->gtt_sa_lock);
/* Mark the chunks as free */
for (bit = mem_obj->range_start;
bit <= mem_obj->range_end;
bit++)
clear_bit(bit, kfd->gtt_sa_bitmap);
mutex_unlock(&kfd->gtt_sa_lock);
kfree(mem_obj);
return 0;
}
#if defined(CONFIG_DEBUG_FS)
/* This function will send a package to HIQ to hang the HWS
* which will trigger a GPU reset and bring the HWS back to normal state
*/
int kfd_debugfs_hang_hws(struct kfd_dev *dev)
{
int r = 0;
if (dev->dqm->sched_policy != KFD_SCHED_POLICY_HWS) {
pr_err("HWS is not enabled");
return -EINVAL;
}
r = pm_debugfs_hang_hws(&dev->dqm->packets);
if (!r)
r = dqm_debugfs_execute_queues(dev->dqm);
return r;
}
#endif
|