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
|
/*
* Copyright 2013 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 (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*
* Authors: Marek Olšák <maraeo@gmail.com>
*
*/
#include "r600_pipe_common.h"
#include "r600_cs.h"
#include "tgsi/tgsi_parse.h"
#include "util/u_draw_quad.h"
#include "util/u_memory.h"
#include "util/u_format_s3tc.h"
#include "util/u_upload_mgr.h"
#include "vl/vl_decoder.h"
#include "vl/vl_video_buffer.h"
#include "radeon/radeon_video.h"
#include <inttypes.h>
#ifndef HAVE_LLVM
#define HAVE_LLVM 0
#endif
/*
* pipe_context
*/
void r600_draw_rectangle(struct blitter_context *blitter,
int x1, int y1, int x2, int y2, float depth,
enum blitter_attrib_type type,
const union pipe_color_union *attrib)
{
struct r600_common_context *rctx =
(struct r600_common_context*)util_blitter_get_pipe(blitter);
struct pipe_viewport_state viewport;
struct pipe_resource *buf = NULL;
unsigned offset = 0;
float *vb;
if (type == UTIL_BLITTER_ATTRIB_TEXCOORD) {
util_blitter_draw_rectangle(blitter, x1, y1, x2, y2, depth, type, attrib);
return;
}
/* Some operations (like color resolve on r6xx) don't work
* with the conventional primitive types.
* One that works is PT_RECTLIST, which we use here. */
/* setup viewport */
viewport.scale[0] = 1.0f;
viewport.scale[1] = 1.0f;
viewport.scale[2] = 1.0f;
viewport.translate[0] = 0.0f;
viewport.translate[1] = 0.0f;
viewport.translate[2] = 0.0f;
rctx->b.set_viewport_states(&rctx->b, 0, 1, &viewport);
/* Upload vertices. The hw rectangle has only 3 vertices,
* I guess the 4th one is derived from the first 3.
* The vertex specification should match u_blitter's vertex element state. */
u_upload_alloc(rctx->uploader, 0, sizeof(float) * 24, &offset, &buf, (void**)&vb);
vb[0] = x1;
vb[1] = y1;
vb[2] = depth;
vb[3] = 1;
vb[8] = x1;
vb[9] = y2;
vb[10] = depth;
vb[11] = 1;
vb[16] = x2;
vb[17] = y1;
vb[18] = depth;
vb[19] = 1;
if (attrib) {
memcpy(vb+4, attrib->f, sizeof(float)*4);
memcpy(vb+12, attrib->f, sizeof(float)*4);
memcpy(vb+20, attrib->f, sizeof(float)*4);
}
/* draw */
util_draw_vertex_buffer(&rctx->b, NULL, buf, blitter->vb_slot, offset,
R600_PRIM_RECTANGLE_LIST, 3, 2);
pipe_resource_reference(&buf, NULL);
}
void r600_need_dma_space(struct r600_common_context *ctx, unsigned num_dw)
{
/* Flush if there's not enough space. */
if ((num_dw + ctx->rings.dma.cs->cdw) > ctx->rings.dma.cs->max_dw) {
ctx->rings.dma.flush(ctx, RADEON_FLUSH_ASYNC, NULL);
assert((num_dw + ctx->rings.dma.cs->cdw) <= ctx->rings.dma.cs->max_dw);
}
}
static void r600_memory_barrier(struct pipe_context *ctx, unsigned flags)
{
}
void r600_preflush_suspend_features(struct r600_common_context *ctx)
{
/* Disable render condition. */
ctx->saved_render_cond = NULL;
ctx->saved_render_cond_cond = FALSE;
ctx->saved_render_cond_mode = 0;
if (ctx->current_render_cond) {
ctx->saved_render_cond = ctx->current_render_cond;
ctx->saved_render_cond_cond = ctx->current_render_cond_cond;
ctx->saved_render_cond_mode = ctx->current_render_cond_mode;
ctx->b.render_condition(&ctx->b, NULL, FALSE, 0);
}
/* suspend queries */
ctx->queries_suspended_for_flush = false;
if (ctx->num_cs_dw_nontimer_queries_suspend) {
r600_suspend_nontimer_queries(ctx);
r600_suspend_timer_queries(ctx);
ctx->queries_suspended_for_flush = true;
}
ctx->streamout.suspended = false;
if (ctx->streamout.begin_emitted) {
r600_emit_streamout_end(ctx);
ctx->streamout.suspended = true;
}
}
void r600_postflush_resume_features(struct r600_common_context *ctx)
{
if (ctx->streamout.suspended) {
ctx->streamout.append_bitmask = ctx->streamout.enabled_mask;
r600_streamout_buffers_dirty(ctx);
}
/* resume queries */
if (ctx->queries_suspended_for_flush) {
r600_resume_nontimer_queries(ctx);
r600_resume_timer_queries(ctx);
}
/* Re-enable render condition. */
if (ctx->saved_render_cond) {
ctx->b.render_condition(&ctx->b, ctx->saved_render_cond,
ctx->saved_render_cond_cond,
ctx->saved_render_cond_mode);
}
}
static void r600_flush_from_st(struct pipe_context *ctx,
struct pipe_fence_handle **fence,
unsigned flags)
{
struct r600_common_context *rctx = (struct r600_common_context *)ctx;
unsigned rflags = 0;
if (flags & PIPE_FLUSH_END_OF_FRAME)
rflags |= RADEON_FLUSH_END_OF_FRAME;
if (rctx->rings.dma.cs) {
rctx->rings.dma.flush(rctx, rflags, NULL);
}
rctx->rings.gfx.flush(rctx, rflags, fence);
}
static void r600_flush_dma_ring(void *ctx, unsigned flags,
struct pipe_fence_handle **fence)
{
struct r600_common_context *rctx = (struct r600_common_context *)ctx;
struct radeon_winsys_cs *cs = rctx->rings.dma.cs;
if (!cs->cdw) {
return;
}
rctx->rings.dma.flushing = true;
rctx->ws->cs_flush(cs, flags, fence, 0);
rctx->rings.dma.flushing = false;
}
static enum pipe_reset_status r600_get_reset_status(struct pipe_context *ctx)
{
struct r600_common_context *rctx = (struct r600_common_context *)ctx;
unsigned latest = rctx->ws->query_value(rctx->ws,
RADEON_GPU_RESET_COUNTER);
if (rctx->gpu_reset_counter == latest)
return PIPE_NO_RESET;
rctx->gpu_reset_counter = latest;
return PIPE_UNKNOWN_CONTEXT_RESET;
}
bool r600_common_context_init(struct r600_common_context *rctx,
struct r600_common_screen *rscreen)
{
util_slab_create(&rctx->pool_transfers,
sizeof(struct r600_transfer), 64,
UTIL_SLAB_SINGLETHREADED);
rctx->screen = rscreen;
rctx->ws = rscreen->ws;
rctx->family = rscreen->family;
rctx->chip_class = rscreen->chip_class;
if (rscreen->family == CHIP_HAWAII)
rctx->max_db = 16;
else if (rscreen->chip_class >= EVERGREEN)
rctx->max_db = 8;
else
rctx->max_db = 4;
rctx->b.transfer_map = u_transfer_map_vtbl;
rctx->b.transfer_flush_region = u_default_transfer_flush_region;
rctx->b.transfer_unmap = u_transfer_unmap_vtbl;
rctx->b.transfer_inline_write = u_default_transfer_inline_write;
rctx->b.memory_barrier = r600_memory_barrier;
rctx->b.flush = r600_flush_from_st;
if (rscreen->info.drm_major == 2 && rscreen->info.drm_minor >= 43) {
rctx->b.get_device_reset_status = r600_get_reset_status;
rctx->gpu_reset_counter =
rctx->ws->query_value(rctx->ws,
RADEON_GPU_RESET_COUNTER);
}
LIST_INITHEAD(&rctx->texture_buffers);
r600_init_context_texture_functions(rctx);
r600_streamout_init(rctx);
r600_query_init(rctx);
cayman_init_msaa(&rctx->b);
rctx->allocator_so_filled_size = u_suballocator_create(&rctx->b, 4096, 4,
0, PIPE_USAGE_DEFAULT, TRUE);
if (!rctx->allocator_so_filled_size)
return false;
rctx->uploader = u_upload_create(&rctx->b, 1024 * 1024, 256,
PIPE_BIND_INDEX_BUFFER |
PIPE_BIND_CONSTANT_BUFFER);
if (!rctx->uploader)
return false;
rctx->ctx = rctx->ws->ctx_create(rctx->ws);
if (!rctx->ctx)
return false;
if (rscreen->info.r600_has_dma && !(rscreen->debug_flags & DBG_NO_ASYNC_DMA)) {
rctx->rings.dma.cs = rctx->ws->cs_create(rctx->ctx, RING_DMA,
r600_flush_dma_ring,
rctx, NULL);
rctx->rings.dma.flush = r600_flush_dma_ring;
}
return true;
}
void r600_common_context_cleanup(struct r600_common_context *rctx)
{
if (rctx->rings.gfx.cs)
rctx->ws->cs_destroy(rctx->rings.gfx.cs);
if (rctx->rings.dma.cs)
rctx->ws->cs_destroy(rctx->rings.dma.cs);
if (rctx->ctx)
rctx->ws->ctx_destroy(rctx->ctx);
if (rctx->uploader) {
u_upload_destroy(rctx->uploader);
}
util_slab_destroy(&rctx->pool_transfers);
if (rctx->allocator_so_filled_size) {
u_suballocator_destroy(rctx->allocator_so_filled_size);
}
}
void r600_context_add_resource_size(struct pipe_context *ctx, struct pipe_resource *r)
{
struct r600_common_context *rctx = (struct r600_common_context *)ctx;
struct r600_resource *rr = (struct r600_resource *)r;
if (r == NULL) {
return;
}
/*
* The idea is to compute a gross estimate of memory requirement of
* each draw call. After each draw call, memory will be precisely
* accounted. So the uncertainty is only on the current draw call.
* In practice this gave very good estimate (+/- 10% of the target
* memory limit).
*/
if (rr->domains & RADEON_DOMAIN_GTT) {
rctx->gtt += rr->buf->size;
}
if (rr->domains & RADEON_DOMAIN_VRAM) {
rctx->vram += rr->buf->size;
}
}
/*
* pipe_screen
*/
static const struct debug_named_value common_debug_options[] = {
/* logging */
{ "tex", DBG_TEX, "Print texture info" },
{ "texmip", DBG_TEXMIP, "Print texture info (mipmapped only)" },
{ "compute", DBG_COMPUTE, "Print compute info" },
{ "vm", DBG_VM, "Print virtual addresses when creating resources" },
{ "trace_cs", DBG_TRACE_CS, "Trace cs and write rlockup_<csid>.c file with faulty cs" },
{ "info", DBG_INFO, "Print driver information" },
/* shaders */
{ "fs", DBG_FS, "Print fetch shaders" },
{ "vs", DBG_VS, "Print vertex shaders" },
{ "gs", DBG_GS, "Print geometry shaders" },
{ "ps", DBG_PS, "Print pixel shaders" },
{ "cs", DBG_CS, "Print compute shaders" },
{ "tcs", DBG_TCS, "Print tessellation control shaders" },
{ "tes", DBG_TES, "Print tessellation evaluation shaders" },
{ "noir", DBG_NO_IR, "Don't print the LLVM IR"},
{ "notgsi", DBG_NO_TGSI, "Don't print the TGSI"},
{ "noasm", DBG_NO_ASM, "Don't print disassembled shaders"},
/* features */
{ "nodma", DBG_NO_ASYNC_DMA, "Disable asynchronous DMA" },
{ "nohyperz", DBG_NO_HYPERZ, "Disable Hyper-Z" },
/* GL uses the word INVALIDATE, gallium uses the word DISCARD */
{ "noinvalrange", DBG_NO_DISCARD_RANGE, "Disable handling of INVALIDATE_RANGE map flags" },
{ "no2d", DBG_NO_2D_TILING, "Disable 2D tiling" },
{ "notiling", DBG_NO_TILING, "Disable tiling" },
{ "switch_on_eop", DBG_SWITCH_ON_EOP, "Program WD/IA to switch on end-of-packet." },
{ "forcedma", DBG_FORCE_DMA, "Use asynchronous DMA for all operations when possible." },
{ "precompile", DBG_PRECOMPILE, "Compile one shader variant at shader creation." },
{ "nowc", DBG_NO_WC, "Disable GTT write combining" },
DEBUG_NAMED_VALUE_END /* must be last */
};
static const char* r600_get_vendor(struct pipe_screen* pscreen)
{
return "X.Org";
}
static const char* r600_get_device_vendor(struct pipe_screen* pscreen)
{
return "AMD";
}
static const char* r600_get_chip_name(struct r600_common_screen *rscreen)
{
switch (rscreen->info.family) {
case CHIP_R600: return "AMD R600";
case CHIP_RV610: return "AMD RV610";
case CHIP_RV630: return "AMD RV630";
case CHIP_RV670: return "AMD RV670";
case CHIP_RV620: return "AMD RV620";
case CHIP_RV635: return "AMD RV635";
case CHIP_RS780: return "AMD RS780";
case CHIP_RS880: return "AMD RS880";
case CHIP_RV770: return "AMD RV770";
case CHIP_RV730: return "AMD RV730";
case CHIP_RV710: return "AMD RV710";
case CHIP_RV740: return "AMD RV740";
case CHIP_CEDAR: return "AMD CEDAR";
case CHIP_REDWOOD: return "AMD REDWOOD";
case CHIP_JUNIPER: return "AMD JUNIPER";
case CHIP_CYPRESS: return "AMD CYPRESS";
case CHIP_HEMLOCK: return "AMD HEMLOCK";
case CHIP_PALM: return "AMD PALM";
case CHIP_SUMO: return "AMD SUMO";
case CHIP_SUMO2: return "AMD SUMO2";
case CHIP_BARTS: return "AMD BARTS";
case CHIP_TURKS: return "AMD TURKS";
case CHIP_CAICOS: return "AMD CAICOS";
case CHIP_CAYMAN: return "AMD CAYMAN";
case CHIP_ARUBA: return "AMD ARUBA";
case CHIP_TAHITI: return "AMD TAHITI";
case CHIP_PITCAIRN: return "AMD PITCAIRN";
case CHIP_VERDE: return "AMD CAPE VERDE";
case CHIP_OLAND: return "AMD OLAND";
case CHIP_HAINAN: return "AMD HAINAN";
case CHIP_BONAIRE: return "AMD BONAIRE";
case CHIP_KAVERI: return "AMD KAVERI";
case CHIP_KABINI: return "AMD KABINI";
case CHIP_HAWAII: return "AMD HAWAII";
case CHIP_MULLINS: return "AMD MULLINS";
case CHIP_TONGA: return "AMD TONGA";
case CHIP_ICELAND: return "AMD ICELAND";
case CHIP_CARRIZO: return "AMD CARRIZO";
default: return "AMD unknown";
}
}
static const char* r600_get_name(struct pipe_screen* pscreen)
{
struct r600_common_screen *rscreen = (struct r600_common_screen*)pscreen;
return rscreen->renderer_string;
}
static float r600_get_paramf(struct pipe_screen* pscreen,
enum pipe_capf param)
{
struct r600_common_screen *rscreen = (struct r600_common_screen *)pscreen;
switch (param) {
case PIPE_CAPF_MAX_LINE_WIDTH:
case PIPE_CAPF_MAX_LINE_WIDTH_AA:
case PIPE_CAPF_MAX_POINT_WIDTH:
case PIPE_CAPF_MAX_POINT_WIDTH_AA:
if (rscreen->family >= CHIP_CEDAR)
return 16384.0f;
else
return 8192.0f;
case PIPE_CAPF_MAX_TEXTURE_ANISOTROPY:
return 16.0f;
case PIPE_CAPF_MAX_TEXTURE_LOD_BIAS:
return 16.0f;
case PIPE_CAPF_GUARD_BAND_LEFT:
case PIPE_CAPF_GUARD_BAND_TOP:
case PIPE_CAPF_GUARD_BAND_RIGHT:
case PIPE_CAPF_GUARD_BAND_BOTTOM:
return 0.0f;
}
return 0.0f;
}
static int r600_get_video_param(struct pipe_screen *screen,
enum pipe_video_profile profile,
enum pipe_video_entrypoint entrypoint,
enum pipe_video_cap param)
{
switch (param) {
case PIPE_VIDEO_CAP_SUPPORTED:
return vl_profile_supported(screen, profile, entrypoint);
case PIPE_VIDEO_CAP_NPOT_TEXTURES:
return 1;
case PIPE_VIDEO_CAP_MAX_WIDTH:
case PIPE_VIDEO_CAP_MAX_HEIGHT:
return vl_video_buffer_max_size(screen);
case PIPE_VIDEO_CAP_PREFERED_FORMAT:
return PIPE_FORMAT_NV12;
case PIPE_VIDEO_CAP_PREFERS_INTERLACED:
return false;
case PIPE_VIDEO_CAP_SUPPORTS_INTERLACED:
return false;
case PIPE_VIDEO_CAP_SUPPORTS_PROGRESSIVE:
return true;
case PIPE_VIDEO_CAP_MAX_LEVEL:
return vl_level_supported(screen, profile);
default:
return 0;
}
}
const char *r600_get_llvm_processor_name(enum radeon_family family)
{
switch (family) {
case CHIP_R600:
case CHIP_RV630:
case CHIP_RV635:
case CHIP_RV670:
return "r600";
case CHIP_RV610:
case CHIP_RV620:
case CHIP_RS780:
case CHIP_RS880:
return "rs880";
case CHIP_RV710:
return "rv710";
case CHIP_RV730:
return "rv730";
case CHIP_RV740:
case CHIP_RV770:
return "rv770";
case CHIP_PALM:
case CHIP_CEDAR:
return "cedar";
case CHIP_SUMO:
case CHIP_SUMO2:
return "sumo";
case CHIP_REDWOOD:
return "redwood";
case CHIP_JUNIPER:
return "juniper";
case CHIP_HEMLOCK:
case CHIP_CYPRESS:
return "cypress";
case CHIP_BARTS:
return "barts";
case CHIP_TURKS:
return "turks";
case CHIP_CAICOS:
return "caicos";
case CHIP_CAYMAN:
case CHIP_ARUBA:
return "cayman";
case CHIP_TAHITI: return "tahiti";
case CHIP_PITCAIRN: return "pitcairn";
case CHIP_VERDE: return "verde";
case CHIP_OLAND: return "oland";
case CHIP_HAINAN: return "hainan";
case CHIP_BONAIRE: return "bonaire";
case CHIP_KABINI: return "kabini";
case CHIP_KAVERI: return "kaveri";
case CHIP_HAWAII: return "hawaii";
case CHIP_MULLINS:
#if HAVE_LLVM >= 0x0305
return "mullins";
#else
return "kabini";
#endif
case CHIP_TONGA: return "tonga";
case CHIP_ICELAND: return "iceland";
case CHIP_CARRIZO: return "carrizo";
default: return "";
}
}
static int r600_get_compute_param(struct pipe_screen *screen,
enum pipe_compute_cap param,
void *ret)
{
struct r600_common_screen *rscreen = (struct r600_common_screen *)screen;
//TODO: select these params by asic
switch (param) {
case PIPE_COMPUTE_CAP_IR_TARGET: {
const char *gpu;
const char *triple;
if (rscreen->family <= CHIP_ARUBA || HAVE_LLVM < 0x0306) {
triple = "r600--";
} else {
triple = "amdgcn--";
}
switch(rscreen->family) {
/* Clang < 3.6 is missing Hainan in its list of
* GPUs, so we need to use the name of a similar GPU.
*/
#if HAVE_LLVM < 0x0306
case CHIP_HAINAN:
gpu = "oland";
break;
#endif
default:
gpu = r600_get_llvm_processor_name(rscreen->family);
break;
}
if (ret) {
sprintf(ret, "%s-%s", gpu, triple);
}
/* +2 for dash and terminating NIL byte */
return (strlen(triple) + strlen(gpu) + 2) * sizeof(char);
}
case PIPE_COMPUTE_CAP_GRID_DIMENSION:
if (ret) {
uint64_t *grid_dimension = ret;
grid_dimension[0] = 3;
}
return 1 * sizeof(uint64_t);
case PIPE_COMPUTE_CAP_MAX_GRID_SIZE:
if (ret) {
uint64_t *grid_size = ret;
grid_size[0] = 65535;
grid_size[1] = 65535;
grid_size[2] = 1;
}
return 3 * sizeof(uint64_t) ;
case PIPE_COMPUTE_CAP_MAX_BLOCK_SIZE:
if (ret) {
uint64_t *block_size = ret;
block_size[0] = 256;
block_size[1] = 256;
block_size[2] = 256;
}
return 3 * sizeof(uint64_t);
case PIPE_COMPUTE_CAP_MAX_THREADS_PER_BLOCK:
if (ret) {
uint64_t *max_threads_per_block = ret;
*max_threads_per_block = 256;
}
return sizeof(uint64_t);
case PIPE_COMPUTE_CAP_MAX_GLOBAL_SIZE:
if (ret) {
uint64_t *max_global_size = ret;
uint64_t max_mem_alloc_size;
r600_get_compute_param(screen,
PIPE_COMPUTE_CAP_MAX_MEM_ALLOC_SIZE,
&max_mem_alloc_size);
/* In OpenCL, the MAX_MEM_ALLOC_SIZE must be at least
* 1/4 of the MAX_GLOBAL_SIZE. Since the
* MAX_MEM_ALLOC_SIZE is fixed for older kernels,
* make sure we never report more than
* 4 * MAX_MEM_ALLOC_SIZE.
*/
*max_global_size = MIN2(4 * max_mem_alloc_size,
rscreen->info.gart_size +
rscreen->info.vram_size);
}
return sizeof(uint64_t);
case PIPE_COMPUTE_CAP_MAX_LOCAL_SIZE:
if (ret) {
uint64_t *max_local_size = ret;
/* Value reported by the closed source driver. */
*max_local_size = 32768;
}
return sizeof(uint64_t);
case PIPE_COMPUTE_CAP_MAX_INPUT_SIZE:
if (ret) {
uint64_t *max_input_size = ret;
/* Value reported by the closed source driver. */
*max_input_size = 1024;
}
return sizeof(uint64_t);
case PIPE_COMPUTE_CAP_MAX_MEM_ALLOC_SIZE:
if (ret) {
uint64_t *max_mem_alloc_size = ret;
/* XXX: The limit in older kernels is 256 MB. We
* should add a query here for newer kernels.
*/
*max_mem_alloc_size = 256 * 1024 * 1024;
}
return sizeof(uint64_t);
case PIPE_COMPUTE_CAP_MAX_CLOCK_FREQUENCY:
if (ret) {
uint32_t *max_clock_frequency = ret;
*max_clock_frequency = rscreen->info.max_sclk;
}
return sizeof(uint32_t);
case PIPE_COMPUTE_CAP_MAX_COMPUTE_UNITS:
if (ret) {
uint32_t *max_compute_units = ret;
*max_compute_units = rscreen->info.max_compute_units;
}
return sizeof(uint32_t);
case PIPE_COMPUTE_CAP_IMAGES_SUPPORTED:
if (ret) {
uint32_t *images_supported = ret;
*images_supported = 0;
}
return sizeof(uint32_t);
case PIPE_COMPUTE_CAP_MAX_PRIVATE_SIZE:
break; /* unused */
case PIPE_COMPUTE_CAP_SUBGROUP_SIZE:
if (ret) {
uint32_t *subgroup_size = ret;
*subgroup_size = r600_wavefront_size(rscreen->family);
}
return sizeof(uint32_t);
}
fprintf(stderr, "unknown PIPE_COMPUTE_CAP %d\n", param);
return 0;
}
static uint64_t r600_get_timestamp(struct pipe_screen *screen)
{
struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
return 1000000 * rscreen->ws->query_value(rscreen->ws, RADEON_TIMESTAMP) /
rscreen->info.r600_clock_crystal_freq;
}
static int r600_get_driver_query_info(struct pipe_screen *screen,
unsigned index,
struct pipe_driver_query_info *info)
{
struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
struct pipe_driver_query_info list[] = {
{"num-compilations", R600_QUERY_NUM_COMPILATIONS, {0}, PIPE_DRIVER_QUERY_TYPE_UINT64,
PIPE_DRIVER_QUERY_RESULT_TYPE_CUMULATIVE},
{"num-shaders-created", R600_QUERY_NUM_SHADERS_CREATED, {0}, PIPE_DRIVER_QUERY_TYPE_UINT64,
PIPE_DRIVER_QUERY_RESULT_TYPE_CUMULATIVE},
{"draw-calls", R600_QUERY_DRAW_CALLS, {0}},
{"requested-VRAM", R600_QUERY_REQUESTED_VRAM, {rscreen->info.vram_size}, PIPE_DRIVER_QUERY_TYPE_BYTES},
{"requested-GTT", R600_QUERY_REQUESTED_GTT, {rscreen->info.gart_size}, PIPE_DRIVER_QUERY_TYPE_BYTES},
{"buffer-wait-time", R600_QUERY_BUFFER_WAIT_TIME, {0}, PIPE_DRIVER_QUERY_TYPE_MICROSECONDS,
PIPE_DRIVER_QUERY_RESULT_TYPE_CUMULATIVE},
{"num-cs-flushes", R600_QUERY_NUM_CS_FLUSHES, {0}},
{"num-bytes-moved", R600_QUERY_NUM_BYTES_MOVED, {0}, PIPE_DRIVER_QUERY_TYPE_BYTES,
PIPE_DRIVER_QUERY_RESULT_TYPE_CUMULATIVE},
{"VRAM-usage", R600_QUERY_VRAM_USAGE, {rscreen->info.vram_size}, PIPE_DRIVER_QUERY_TYPE_BYTES},
{"GTT-usage", R600_QUERY_GTT_USAGE, {rscreen->info.gart_size}, PIPE_DRIVER_QUERY_TYPE_BYTES},
{"GPU-load", R600_QUERY_GPU_LOAD, {100}},
{"temperature", R600_QUERY_GPU_TEMPERATURE, {100}},
{"shader-clock", R600_QUERY_CURRENT_GPU_SCLK, {0}, PIPE_DRIVER_QUERY_TYPE_HZ},
{"memory-clock", R600_QUERY_CURRENT_GPU_MCLK, {0}, PIPE_DRIVER_QUERY_TYPE_HZ},
};
unsigned num_queries;
if (rscreen->info.drm_major == 2 && rscreen->info.drm_minor >= 42)
num_queries = Elements(list);
else if (rscreen->info.drm_major == 3)
num_queries = Elements(list) - 3;
else
num_queries = Elements(list) - 4;
if (!info)
return num_queries;
if (index >= num_queries)
return 0;
*info = list[index];
return 1;
}
static void r600_fence_reference(struct pipe_screen *screen,
struct pipe_fence_handle **ptr,
struct pipe_fence_handle *fence)
{
struct radeon_winsys *rws = ((struct r600_common_screen*)screen)->ws;
rws->fence_reference(ptr, fence);
}
static boolean r600_fence_finish(struct pipe_screen *screen,
struct pipe_fence_handle *fence,
uint64_t timeout)
{
struct radeon_winsys *rws = ((struct r600_common_screen*)screen)->ws;
return rws->fence_wait(rws, fence, timeout);
}
static bool r600_interpret_tiling(struct r600_common_screen *rscreen,
uint32_t tiling_config)
{
switch ((tiling_config & 0xe) >> 1) {
case 0:
rscreen->tiling_info.num_channels = 1;
break;
case 1:
rscreen->tiling_info.num_channels = 2;
break;
case 2:
rscreen->tiling_info.num_channels = 4;
break;
case 3:
rscreen->tiling_info.num_channels = 8;
break;
default:
return false;
}
switch ((tiling_config & 0x30) >> 4) {
case 0:
rscreen->tiling_info.num_banks = 4;
break;
case 1:
rscreen->tiling_info.num_banks = 8;
break;
default:
return false;
}
switch ((tiling_config & 0xc0) >> 6) {
case 0:
rscreen->tiling_info.group_bytes = 256;
break;
case 1:
rscreen->tiling_info.group_bytes = 512;
break;
default:
return false;
}
return true;
}
static bool evergreen_interpret_tiling(struct r600_common_screen *rscreen,
uint32_t tiling_config)
{
switch (tiling_config & 0xf) {
case 0:
rscreen->tiling_info.num_channels = 1;
break;
case 1:
rscreen->tiling_info.num_channels = 2;
break;
case 2:
rscreen->tiling_info.num_channels = 4;
break;
case 3:
rscreen->tiling_info.num_channels = 8;
break;
default:
return false;
}
switch ((tiling_config & 0xf0) >> 4) {
case 0:
rscreen->tiling_info.num_banks = 4;
break;
case 1:
rscreen->tiling_info.num_banks = 8;
break;
case 2:
rscreen->tiling_info.num_banks = 16;
break;
default:
return false;
}
switch ((tiling_config & 0xf00) >> 8) {
case 0:
rscreen->tiling_info.group_bytes = 256;
break;
case 1:
rscreen->tiling_info.group_bytes = 512;
break;
default:
return false;
}
return true;
}
static bool r600_init_tiling(struct r600_common_screen *rscreen)
{
uint32_t tiling_config = rscreen->info.r600_tiling_config;
/* set default group bytes, overridden by tiling info ioctl */
if (rscreen->chip_class <= R700) {
rscreen->tiling_info.group_bytes = 256;
} else {
rscreen->tiling_info.group_bytes = 512;
}
if (!tiling_config)
return true;
if (rscreen->chip_class <= R700) {
return r600_interpret_tiling(rscreen, tiling_config);
} else {
return evergreen_interpret_tiling(rscreen, tiling_config);
}
}
struct pipe_resource *r600_resource_create_common(struct pipe_screen *screen,
const struct pipe_resource *templ)
{
if (templ->target == PIPE_BUFFER) {
return r600_buffer_create(screen, templ, 4096);
} else {
return r600_texture_create(screen, templ);
}
}
bool r600_common_screen_init(struct r600_common_screen *rscreen,
struct radeon_winsys *ws)
{
char llvm_string[32] = {};
ws->query_info(ws, &rscreen->info);
#if HAVE_LLVM
snprintf(llvm_string, sizeof(llvm_string),
", LLVM %i.%i.%i", (HAVE_LLVM >> 8) & 0xff,
HAVE_LLVM & 0xff, MESA_LLVM_VERSION_PATCH);
#endif
snprintf(rscreen->renderer_string, sizeof(rscreen->renderer_string),
"%s (DRM %i.%i.%i%s)",
r600_get_chip_name(rscreen), rscreen->info.drm_major,
rscreen->info.drm_minor, rscreen->info.drm_patchlevel,
llvm_string);
rscreen->b.get_name = r600_get_name;
rscreen->b.get_vendor = r600_get_vendor;
rscreen->b.get_device_vendor = r600_get_device_vendor;
rscreen->b.get_compute_param = r600_get_compute_param;
rscreen->b.get_paramf = r600_get_paramf;
rscreen->b.get_driver_query_info = r600_get_driver_query_info;
rscreen->b.get_timestamp = r600_get_timestamp;
rscreen->b.fence_finish = r600_fence_finish;
rscreen->b.fence_reference = r600_fence_reference;
rscreen->b.resource_destroy = u_resource_destroy_vtbl;
rscreen->b.resource_from_user_memory = r600_buffer_from_user_memory;
if (rscreen->info.has_uvd) {
rscreen->b.get_video_param = rvid_get_video_param;
rscreen->b.is_video_format_supported = rvid_is_format_supported;
} else {
rscreen->b.get_video_param = r600_get_video_param;
rscreen->b.is_video_format_supported = vl_video_buffer_is_format_supported;
}
r600_init_screen_texture_functions(rscreen);
rscreen->ws = ws;
rscreen->family = rscreen->info.family;
rscreen->chip_class = rscreen->info.chip_class;
rscreen->debug_flags = debug_get_flags_option("R600_DEBUG", common_debug_options, 0);
if (!r600_init_tiling(rscreen)) {
return false;
}
util_format_s3tc_init();
pipe_mutex_init(rscreen->aux_context_lock);
pipe_mutex_init(rscreen->gpu_load_mutex);
if (((rscreen->info.drm_major == 2 && rscreen->info.drm_minor >= 28) ||
rscreen->info.drm_major == 3) &&
(rscreen->debug_flags & DBG_TRACE_CS)) {
rscreen->trace_bo = (struct r600_resource*)pipe_buffer_create(&rscreen->b,
PIPE_BIND_CUSTOM,
PIPE_USAGE_STAGING,
4096);
if (rscreen->trace_bo) {
rscreen->trace_ptr = rscreen->ws->buffer_map(rscreen->trace_bo->cs_buf, NULL,
PIPE_TRANSFER_UNSYNCHRONIZED);
}
}
if (rscreen->debug_flags & DBG_INFO) {
printf("pci_id = 0x%x\n", rscreen->info.pci_id);
printf("family = %i\n", rscreen->info.family);
printf("chip_class = %i\n", rscreen->info.chip_class);
printf("gart_size = %i MB\n", (int)(rscreen->info.gart_size >> 20));
printf("vram_size = %i MB\n", (int)(rscreen->info.vram_size >> 20));
printf("max_sclk = %i\n", rscreen->info.max_sclk);
printf("max_compute_units = %i\n", rscreen->info.max_compute_units);
printf("max_se = %i\n", rscreen->info.max_se);
printf("max_sh_per_se = %i\n", rscreen->info.max_sh_per_se);
printf("drm = %i.%i.%i\n", rscreen->info.drm_major,
rscreen->info.drm_minor, rscreen->info.drm_patchlevel);
printf("has_uvd = %i\n", rscreen->info.has_uvd);
printf("vce_fw_version = %i\n", rscreen->info.vce_fw_version);
printf("r600_num_backends = %i\n", rscreen->info.r600_num_backends);
printf("r600_clock_crystal_freq = %i\n", rscreen->info.r600_clock_crystal_freq);
printf("r600_tiling_config = 0x%x\n", rscreen->info.r600_tiling_config);
printf("r600_num_tile_pipes = %i\n", rscreen->info.r600_num_tile_pipes);
printf("r600_max_pipes = %i\n", rscreen->info.r600_max_pipes);
printf("r600_virtual_address = %i\n", rscreen->info.r600_virtual_address);
printf("r600_has_dma = %i\n", rscreen->info.r600_has_dma);
printf("r600_backend_map = %i\n", rscreen->info.r600_backend_map);
printf("r600_backend_map_valid = %i\n", rscreen->info.r600_backend_map_valid);
printf("si_tile_mode_array_valid = %i\n", rscreen->info.si_tile_mode_array_valid);
printf("cik_macrotile_mode_array_valid = %i\n", rscreen->info.cik_macrotile_mode_array_valid);
}
return true;
}
void r600_destroy_common_screen(struct r600_common_screen *rscreen)
{
r600_gpu_load_kill_thread(rscreen);
pipe_mutex_destroy(rscreen->gpu_load_mutex);
pipe_mutex_destroy(rscreen->aux_context_lock);
rscreen->aux_context->destroy(rscreen->aux_context);
if (rscreen->trace_bo)
pipe_resource_reference((struct pipe_resource**)&rscreen->trace_bo, NULL);
rscreen->ws->destroy(rscreen->ws);
FREE(rscreen);
}
bool r600_can_dump_shader(struct r600_common_screen *rscreen,
const struct tgsi_token *tokens)
{
/* Compute shader don't have tgsi_tokens */
if (!tokens)
return (rscreen->debug_flags & DBG_CS) != 0;
switch (tgsi_get_processor_type(tokens)) {
case TGSI_PROCESSOR_VERTEX:
return (rscreen->debug_flags & DBG_VS) != 0;
case TGSI_PROCESSOR_TESS_CTRL:
return (rscreen->debug_flags & DBG_TCS) != 0;
case TGSI_PROCESSOR_TESS_EVAL:
return (rscreen->debug_flags & DBG_TES) != 0;
case TGSI_PROCESSOR_GEOMETRY:
return (rscreen->debug_flags & DBG_GS) != 0;
case TGSI_PROCESSOR_FRAGMENT:
return (rscreen->debug_flags & DBG_PS) != 0;
case TGSI_PROCESSOR_COMPUTE:
return (rscreen->debug_flags & DBG_CS) != 0;
default:
return false;
}
}
void r600_screen_clear_buffer(struct r600_common_screen *rscreen, struct pipe_resource *dst,
unsigned offset, unsigned size, unsigned value,
bool is_framebuffer)
{
struct r600_common_context *rctx = (struct r600_common_context*)rscreen->aux_context;
pipe_mutex_lock(rscreen->aux_context_lock);
rctx->clear_buffer(&rctx->b, dst, offset, size, value, is_framebuffer);
rscreen->aux_context->flush(rscreen->aux_context, NULL, 0);
pipe_mutex_unlock(rscreen->aux_context_lock);
}
|