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
|
/*
* 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 "amdgpu_amdkfd.h"
#include "amd_shared.h"
#include "amdgpu.h"
#include "amdgpu_gfx.h"
#include "amdgpu_dma_buf.h"
#include <linux/module.h>
#include <linux/dma-buf.h>
#include "amdgpu_xgmi.h"
#include <uapi/linux/kfd_ioctl.h>
/* Total memory size in system memory and all GPU VRAM. Used to
* estimate worst case amount of memory to reserve for page tables
*/
uint64_t amdgpu_amdkfd_total_mem_size;
int amdgpu_amdkfd_init(void)
{
struct sysinfo si;
int ret;
si_meminfo(&si);
amdgpu_amdkfd_total_mem_size = si.totalram - si.totalhigh;
amdgpu_amdkfd_total_mem_size *= si.mem_unit;
#ifdef CONFIG_HSA_AMD
ret = kgd2kfd_init();
amdgpu_amdkfd_gpuvm_init_mem_limits();
#else
ret = -ENOENT;
#endif
return ret;
}
void amdgpu_amdkfd_fini(void)
{
kgd2kfd_exit();
}
void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
{
bool vf = amdgpu_sriov_vf(adev);
adev->kfd.dev = kgd2kfd_probe((struct kgd_dev *)adev,
adev->pdev, adev->asic_type, vf);
if (adev->kfd.dev)
amdgpu_amdkfd_total_mem_size += adev->gmc.real_vram_size;
}
/**
* amdgpu_doorbell_get_kfd_info - Report doorbell configuration required to
* setup amdkfd
*
* @adev: amdgpu_device pointer
* @aperture_base: output returning doorbell aperture base physical address
* @aperture_size: output returning doorbell aperture size in bytes
* @start_offset: output returning # of doorbell bytes reserved for amdgpu.
*
* amdgpu and amdkfd share the doorbell aperture. amdgpu sets it up,
* takes doorbells required for its own rings and reports the setup to amdkfd.
* amdgpu reserved doorbells are at the start of the doorbell aperture.
*/
static void amdgpu_doorbell_get_kfd_info(struct amdgpu_device *adev,
phys_addr_t *aperture_base,
size_t *aperture_size,
size_t *start_offset)
{
/*
* The first num_doorbells are used by amdgpu.
* amdkfd takes whatever's left in the aperture.
*/
if (adev->doorbell.size > adev->doorbell.num_doorbells * sizeof(u32)) {
*aperture_base = adev->doorbell.base;
*aperture_size = adev->doorbell.size;
*start_offset = adev->doorbell.num_doorbells * sizeof(u32);
} else {
*aperture_base = 0;
*aperture_size = 0;
*start_offset = 0;
}
}
void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
{
int i;
int last_valid_bit;
if (adev->kfd.dev) {
struct kgd2kfd_shared_resources gpu_resources = {
.compute_vmid_bitmap =
((1 << AMDGPU_NUM_VMID) - 1) -
((1 << adev->vm_manager.first_kfd_vmid) - 1),
.num_pipe_per_mec = adev->gfx.mec.num_pipe_per_mec,
.num_queue_per_pipe = adev->gfx.mec.num_queue_per_pipe,
.gpuvm_size = min(adev->vm_manager.max_pfn
<< AMDGPU_GPU_PAGE_SHIFT,
AMDGPU_GMC_HOLE_START),
.drm_render_minor = adev_to_drm(adev)->render->index,
.sdma_doorbell_idx = adev->doorbell_index.sdma_engine,
};
/* this is going to have a few of the MSBs set that we need to
* clear
*/
bitmap_complement(gpu_resources.cp_queue_bitmap,
adev->gfx.mec.queue_bitmap,
KGD_MAX_QUEUES);
/* According to linux/bitmap.h we shouldn't use bitmap_clear if
* nbits is not compile time constant
*/
last_valid_bit = 1 /* only first MEC can have compute queues */
* adev->gfx.mec.num_pipe_per_mec
* adev->gfx.mec.num_queue_per_pipe;
for (i = last_valid_bit; i < KGD_MAX_QUEUES; ++i)
clear_bit(i, gpu_resources.cp_queue_bitmap);
amdgpu_doorbell_get_kfd_info(adev,
&gpu_resources.doorbell_physical_address,
&gpu_resources.doorbell_aperture_size,
&gpu_resources.doorbell_start_offset);
/* Since SOC15, BIF starts to statically use the
* lower 12 bits of doorbell addresses for routing
* based on settings in registers like
* SDMA0_DOORBELL_RANGE etc..
* In order to route a doorbell to CP engine, the lower
* 12 bits of its address has to be outside the range
* set for SDMA, VCN, and IH blocks.
*/
if (adev->asic_type >= CHIP_VEGA10) {
gpu_resources.non_cp_doorbells_start =
adev->doorbell_index.first_non_cp;
gpu_resources.non_cp_doorbells_end =
adev->doorbell_index.last_non_cp;
}
kgd2kfd_device_init(adev->kfd.dev, adev_to_drm(adev), &gpu_resources);
}
}
void amdgpu_amdkfd_device_fini(struct amdgpu_device *adev)
{
if (adev->kfd.dev) {
kgd2kfd_device_exit(adev->kfd.dev);
adev->kfd.dev = NULL;
}
}
void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,
const void *ih_ring_entry)
{
if (adev->kfd.dev)
kgd2kfd_interrupt(adev->kfd.dev, ih_ring_entry);
}
void amdgpu_amdkfd_suspend(struct amdgpu_device *adev, bool run_pm)
{
if (adev->kfd.dev)
kgd2kfd_suspend(adev->kfd.dev, run_pm);
}
int amdgpu_amdkfd_resume(struct amdgpu_device *adev, bool run_pm)
{
int r = 0;
if (adev->kfd.dev)
r = kgd2kfd_resume(adev->kfd.dev, run_pm);
return r;
}
int amdgpu_amdkfd_pre_reset(struct amdgpu_device *adev)
{
int r = 0;
if (adev->kfd.dev)
r = kgd2kfd_pre_reset(adev->kfd.dev);
return r;
}
int amdgpu_amdkfd_post_reset(struct amdgpu_device *adev)
{
int r = 0;
if (adev->kfd.dev)
r = kgd2kfd_post_reset(adev->kfd.dev);
return r;
}
void amdgpu_amdkfd_gpu_reset(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
if (amdgpu_device_should_recover_gpu(adev))
amdgpu_device_gpu_recover(adev, NULL);
}
int amdgpu_amdkfd_alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
void **mem_obj, uint64_t *gpu_addr,
void **cpu_ptr, bool cp_mqd_gfx9)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
struct amdgpu_bo *bo = NULL;
struct amdgpu_bo_param bp;
int r;
void *cpu_ptr_tmp = NULL;
memset(&bp, 0, sizeof(bp));
bp.size = size;
bp.byte_align = PAGE_SIZE;
bp.domain = AMDGPU_GEM_DOMAIN_GTT;
bp.flags = AMDGPU_GEM_CREATE_CPU_GTT_USWC;
bp.type = ttm_bo_type_kernel;
bp.resv = NULL;
if (cp_mqd_gfx9)
bp.flags |= AMDGPU_GEM_CREATE_CP_MQD_GFX9;
r = amdgpu_bo_create(adev, &bp, &bo);
if (r) {
dev_err(adev->dev,
"failed to allocate BO for amdkfd (%d)\n", r);
return r;
}
/* map the buffer */
r = amdgpu_bo_reserve(bo, true);
if (r) {
dev_err(adev->dev, "(%d) failed to reserve bo for amdkfd\n", r);
goto allocate_mem_reserve_bo_failed;
}
r = amdgpu_bo_pin(bo, AMDGPU_GEM_DOMAIN_GTT);
if (r) {
dev_err(adev->dev, "(%d) failed to pin bo for amdkfd\n", r);
goto allocate_mem_pin_bo_failed;
}
r = amdgpu_ttm_alloc_gart(&bo->tbo);
if (r) {
dev_err(adev->dev, "%p bind failed\n", bo);
goto allocate_mem_kmap_bo_failed;
}
r = amdgpu_bo_kmap(bo, &cpu_ptr_tmp);
if (r) {
dev_err(adev->dev,
"(%d) failed to map bo to kernel for amdkfd\n", r);
goto allocate_mem_kmap_bo_failed;
}
*mem_obj = bo;
*gpu_addr = amdgpu_bo_gpu_offset(bo);
*cpu_ptr = cpu_ptr_tmp;
amdgpu_bo_unreserve(bo);
return 0;
allocate_mem_kmap_bo_failed:
amdgpu_bo_unpin(bo);
allocate_mem_pin_bo_failed:
amdgpu_bo_unreserve(bo);
allocate_mem_reserve_bo_failed:
amdgpu_bo_unref(&bo);
return r;
}
void amdgpu_amdkfd_free_gtt_mem(struct kgd_dev *kgd, void *mem_obj)
{
struct amdgpu_bo *bo = (struct amdgpu_bo *) mem_obj;
amdgpu_bo_reserve(bo, true);
amdgpu_bo_kunmap(bo);
amdgpu_bo_unpin(bo);
amdgpu_bo_unreserve(bo);
amdgpu_bo_unref(&(bo));
}
int amdgpu_amdkfd_alloc_gws(struct kgd_dev *kgd, size_t size,
void **mem_obj)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
struct amdgpu_bo *bo = NULL;
struct amdgpu_bo_param bp;
int r;
memset(&bp, 0, sizeof(bp));
bp.size = size;
bp.byte_align = 1;
bp.domain = AMDGPU_GEM_DOMAIN_GWS;
bp.flags = AMDGPU_GEM_CREATE_NO_CPU_ACCESS;
bp.type = ttm_bo_type_device;
bp.resv = NULL;
r = amdgpu_bo_create(adev, &bp, &bo);
if (r) {
dev_err(adev->dev,
"failed to allocate gws BO for amdkfd (%d)\n", r);
return r;
}
*mem_obj = bo;
return 0;
}
void amdgpu_amdkfd_free_gws(struct kgd_dev *kgd, void *mem_obj)
{
struct amdgpu_bo *bo = (struct amdgpu_bo *)mem_obj;
amdgpu_bo_unref(&bo);
}
uint32_t amdgpu_amdkfd_get_fw_version(struct kgd_dev *kgd,
enum kgd_engine_type type)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
switch (type) {
case KGD_ENGINE_PFP:
return adev->gfx.pfp_fw_version;
case KGD_ENGINE_ME:
return adev->gfx.me_fw_version;
case KGD_ENGINE_CE:
return adev->gfx.ce_fw_version;
case KGD_ENGINE_MEC1:
return adev->gfx.mec_fw_version;
case KGD_ENGINE_MEC2:
return adev->gfx.mec2_fw_version;
case KGD_ENGINE_RLC:
return adev->gfx.rlc_fw_version;
case KGD_ENGINE_SDMA1:
return adev->sdma.instance[0].fw_version;
case KGD_ENGINE_SDMA2:
return adev->sdma.instance[1].fw_version;
default:
return 0;
}
return 0;
}
void amdgpu_amdkfd_get_local_mem_info(struct kgd_dev *kgd,
struct kfd_local_mem_info *mem_info)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
uint64_t address_mask = adev->dev->dma_mask ? ~*adev->dev->dma_mask :
~((1ULL << 32) - 1);
resource_size_t aper_limit = adev->gmc.aper_base + adev->gmc.aper_size;
memset(mem_info, 0, sizeof(*mem_info));
if (!(adev->gmc.aper_base & address_mask || aper_limit & address_mask)) {
mem_info->local_mem_size_public = adev->gmc.visible_vram_size;
mem_info->local_mem_size_private = adev->gmc.real_vram_size -
adev->gmc.visible_vram_size;
} else {
mem_info->local_mem_size_public = 0;
mem_info->local_mem_size_private = adev->gmc.real_vram_size;
}
mem_info->vram_width = adev->gmc.vram_width;
pr_debug("Address base: %pap limit %pap public 0x%llx private 0x%llx\n",
&adev->gmc.aper_base, &aper_limit,
mem_info->local_mem_size_public,
mem_info->local_mem_size_private);
if (amdgpu_sriov_vf(adev))
mem_info->mem_clk_max = adev->clock.default_mclk / 100;
else if (adev->pm.dpm_enabled) {
if (amdgpu_emu_mode == 1)
mem_info->mem_clk_max = 0;
else
mem_info->mem_clk_max = amdgpu_dpm_get_mclk(adev, false) / 100;
} else
mem_info->mem_clk_max = 100;
}
uint64_t amdgpu_amdkfd_get_gpu_clock_counter(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
if (adev->gfx.funcs->get_gpu_clock_counter)
return adev->gfx.funcs->get_gpu_clock_counter(adev);
return 0;
}
uint32_t amdgpu_amdkfd_get_max_engine_clock_in_mhz(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
/* the sclk is in quantas of 10kHz */
if (amdgpu_sriov_vf(adev))
return adev->clock.default_sclk / 100;
else if (adev->pm.dpm_enabled)
return amdgpu_dpm_get_sclk(adev, false) / 100;
else
return 100;
}
void amdgpu_amdkfd_get_cu_info(struct kgd_dev *kgd, struct kfd_cu_info *cu_info)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
struct amdgpu_cu_info acu_info = adev->gfx.cu_info;
memset(cu_info, 0, sizeof(*cu_info));
if (sizeof(cu_info->cu_bitmap) != sizeof(acu_info.bitmap))
return;
cu_info->cu_active_number = acu_info.number;
cu_info->cu_ao_mask = acu_info.ao_cu_mask;
memcpy(&cu_info->cu_bitmap[0], &acu_info.bitmap[0],
sizeof(acu_info.bitmap));
cu_info->num_shader_engines = adev->gfx.config.max_shader_engines;
cu_info->num_shader_arrays_per_engine = adev->gfx.config.max_sh_per_se;
cu_info->num_cu_per_sh = adev->gfx.config.max_cu_per_sh;
cu_info->simd_per_cu = acu_info.simd_per_cu;
cu_info->max_waves_per_simd = acu_info.max_waves_per_simd;
cu_info->wave_front_size = acu_info.wave_front_size;
cu_info->max_scratch_slots_per_cu = acu_info.max_scratch_slots_per_cu;
cu_info->lds_size = acu_info.lds_size;
}
int amdgpu_amdkfd_get_dmabuf_info(struct kgd_dev *kgd, int dma_buf_fd,
struct kgd_dev **dma_buf_kgd,
uint64_t *bo_size, void *metadata_buffer,
size_t buffer_size, uint32_t *metadata_size,
uint32_t *flags)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
struct dma_buf *dma_buf;
struct drm_gem_object *obj;
struct amdgpu_bo *bo;
uint64_t metadata_flags;
int r = -EINVAL;
dma_buf = dma_buf_get(dma_buf_fd);
if (IS_ERR(dma_buf))
return PTR_ERR(dma_buf);
if (dma_buf->ops != &amdgpu_dmabuf_ops)
/* Can't handle non-graphics buffers */
goto out_put;
obj = dma_buf->priv;
if (obj->dev->driver != adev_to_drm(adev)->driver)
/* Can't handle buffers from different drivers */
goto out_put;
adev = drm_to_adev(obj->dev);
bo = gem_to_amdgpu_bo(obj);
if (!(bo->preferred_domains & (AMDGPU_GEM_DOMAIN_VRAM |
AMDGPU_GEM_DOMAIN_GTT)))
/* Only VRAM and GTT BOs are supported */
goto out_put;
r = 0;
if (dma_buf_kgd)
*dma_buf_kgd = (struct kgd_dev *)adev;
if (bo_size)
*bo_size = amdgpu_bo_size(bo);
if (metadata_size)
*metadata_size = bo->metadata_size;
if (metadata_buffer)
r = amdgpu_bo_get_metadata(bo, metadata_buffer, buffer_size,
metadata_size, &metadata_flags);
if (flags) {
*flags = (bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM) ?
KFD_IOC_ALLOC_MEM_FLAGS_VRAM
: KFD_IOC_ALLOC_MEM_FLAGS_GTT;
if (bo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)
*flags |= KFD_IOC_ALLOC_MEM_FLAGS_PUBLIC;
}
out_put:
dma_buf_put(dma_buf);
return r;
}
uint64_t amdgpu_amdkfd_get_vram_usage(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
return amdgpu_vram_mgr_usage(&adev->mman.bdev.man[TTM_PL_VRAM]);
}
uint64_t amdgpu_amdkfd_get_hive_id(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
return adev->gmc.xgmi.hive_id;
}
uint64_t amdgpu_amdkfd_get_unique_id(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
return adev->unique_id;
}
uint8_t amdgpu_amdkfd_get_xgmi_hops_count(struct kgd_dev *dst, struct kgd_dev *src)
{
struct amdgpu_device *peer_adev = (struct amdgpu_device *)src;
struct amdgpu_device *adev = (struct amdgpu_device *)dst;
int ret = amdgpu_xgmi_get_hops_count(adev, peer_adev);
if (ret < 0) {
DRM_ERROR("amdgpu: failed to get xgmi hops count between node %d and %d. ret = %d\n",
adev->gmc.xgmi.physical_node_id,
peer_adev->gmc.xgmi.physical_node_id, ret);
ret = 0;
}
return (uint8_t)ret;
}
uint64_t amdgpu_amdkfd_get_mmio_remap_phys_addr(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
return adev->rmmio_remap.bus_addr;
}
uint32_t amdgpu_amdkfd_get_num_gws(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
return adev->gds.gws_size;
}
uint32_t amdgpu_amdkfd_get_asic_rev_id(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
return adev->rev_id;
}
int amdgpu_amdkfd_submit_ib(struct kgd_dev *kgd, enum kgd_engine_type engine,
uint32_t vmid, uint64_t gpu_addr,
uint32_t *ib_cmd, uint32_t ib_len)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
struct amdgpu_job *job;
struct amdgpu_ib *ib;
struct amdgpu_ring *ring;
struct dma_fence *f = NULL;
int ret;
switch (engine) {
case KGD_ENGINE_MEC1:
ring = &adev->gfx.compute_ring[0];
break;
case KGD_ENGINE_SDMA1:
ring = &adev->sdma.instance[0].ring;
break;
case KGD_ENGINE_SDMA2:
ring = &adev->sdma.instance[1].ring;
break;
default:
pr_err("Invalid engine in IB submission: %d\n", engine);
ret = -EINVAL;
goto err;
}
ret = amdgpu_job_alloc(adev, 1, &job, NULL);
if (ret)
goto err;
ib = &job->ibs[0];
memset(ib, 0, sizeof(struct amdgpu_ib));
ib->gpu_addr = gpu_addr;
ib->ptr = ib_cmd;
ib->length_dw = ib_len;
/* This works for NO_HWS. TODO: need to handle without knowing VMID */
job->vmid = vmid;
ret = amdgpu_ib_schedule(ring, 1, ib, job, &f);
if (ret) {
DRM_ERROR("amdgpu: failed to schedule IB.\n");
goto err_ib_sched;
}
ret = dma_fence_wait(f, false);
err_ib_sched:
dma_fence_put(f);
amdgpu_job_free(job);
err:
return ret;
}
void amdgpu_amdkfd_set_compute_idle(struct kgd_dev *kgd, bool idle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
amdgpu_dpm_switch_power_profile(adev,
PP_SMC_POWER_PROFILE_COMPUTE,
!idle);
}
bool amdgpu_amdkfd_is_kfd_vmid(struct amdgpu_device *adev, u32 vmid)
{
if (adev->kfd.dev)
return vmid >= adev->vm_manager.first_kfd_vmid;
return false;
}
int amdgpu_amdkfd_flush_gpu_tlb_vmid(struct kgd_dev *kgd, uint16_t vmid)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
if (adev->family == AMDGPU_FAMILY_AI) {
int i;
for (i = 0; i < adev->num_vmhubs; i++)
amdgpu_gmc_flush_gpu_tlb(adev, vmid, i, 0);
} else {
amdgpu_gmc_flush_gpu_tlb(adev, vmid, AMDGPU_GFXHUB_0, 0);
}
return 0;
}
int amdgpu_amdkfd_flush_gpu_tlb_pasid(struct kgd_dev *kgd, uint16_t pasid)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
const uint32_t flush_type = 0;
bool all_hub = false;
if (adev->family == AMDGPU_FAMILY_AI)
all_hub = true;
return amdgpu_gmc_flush_gpu_tlb_pasid(adev, pasid, flush_type, all_hub);
}
bool amdgpu_amdkfd_have_atomics_support(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
return adev->have_atomics_support;
}
#ifndef CONFIG_HSA_AMD
bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm)
{
return false;
}
void amdgpu_amdkfd_unreserve_memory_limit(struct amdgpu_bo *bo)
{
}
int amdgpu_amdkfd_remove_fence_on_pt_pd_bos(struct amdgpu_bo *bo)
{
return 0;
}
void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev,
struct amdgpu_vm *vm)
{
}
struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f)
{
return NULL;
}
int amdgpu_amdkfd_evict_userptr(struct kgd_mem *mem, struct mm_struct *mm)
{
return 0;
}
struct kfd_dev *kgd2kfd_probe(struct kgd_dev *kgd, struct pci_dev *pdev,
unsigned int asic_type, bool vf)
{
return NULL;
}
bool kgd2kfd_device_init(struct kfd_dev *kfd,
struct drm_device *ddev,
const struct kgd2kfd_shared_resources *gpu_resources)
{
return false;
}
void kgd2kfd_device_exit(struct kfd_dev *kfd)
{
}
void kgd2kfd_exit(void)
{
}
void kgd2kfd_suspend(struct kfd_dev *kfd, bool run_pm)
{
}
int kgd2kfd_resume(struct kfd_dev *kfd, bool run_pm)
{
return 0;
}
int kgd2kfd_pre_reset(struct kfd_dev *kfd)
{
return 0;
}
int kgd2kfd_post_reset(struct kfd_dev *kfd)
{
return 0;
}
void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry)
{
}
void kgd2kfd_set_sram_ecc_flag(struct kfd_dev *kfd)
{
}
void kgd2kfd_smi_event_throttle(struct kfd_dev *kfd, uint32_t throttle_bitmask)
{
}
#endif
|