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
|
/*==============================================================================
Copyright(c) 2019 Intel Corporation
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 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.
Description: AUX-Table management functions
(contains functions to assign memory to
AUX-Tables with valid entries,
and update their entries on request)
============================================================================*/
#include "Internal/Common/GmmLibInc.h"
#include "../TranslationTable/GmmUmdTranslationTable.h"
#if !defined(__GMM_KMD__)
//=============================================================================
//
// Function: MapNullCCS
//
// Desc: Maps given resource, with dummy null-ccs chain, on Aux Table
//
// Caller: UpdateAuxTable (map op for null-tiles)
//
// Parameters:
// UmdContext: Caller-thread specific info (regarding BB for TR-Aux udpate, cmdQ to use etc)
// BaseAdr: Start adr of main surface
// Size: Main-surface size in bytes
// PartialL1e: Aux-metadata other than AuxVA
// DoNotWait: 1 for CPU update, 0 for async(Gpu) update
//-----------------------------------------------------------------------------
GMM_STATUS GmmLib::AuxTable::MapNullCCS(GMM_UMD_SYNCCONTEXT *UmdContext, GMM_GFX_ADDRESS BaseAdr, GMM_GFX_SIZE_T Size, uint64_t PartialL1e, uint8_t DoNotWait)
{
GMM_STATUS Status = GMM_SUCCESS;
GMM_GFX_SIZE_T L1TableSize = ((GMM_GFX_SIZE_T)GMM_L1_SIZE(AUXTT, GetGmmLibContext())) * (WA16K(GetGmmLibContext()) ? GMM_KBYTE(16) : GMM_KBYTE(64)); // TGL and above : L1TableSize = 256x64K OR 16x1M
GMM_GFX_ADDRESS Addr = 0;
GMM_GFX_ADDRESS L3GfxAddress = 0;
GMM_CLIENT ClientType;
GET_GMM_CLIENT_TYPE(pClientContext, ClientType);
EnterCriticalSection(&TTLock);
DoNotWait |= (!UmdContext || !UmdContext->pCommandQueueHandle);
if(TTL3.L3Handle)
{
L3GfxAddress = TTL3.GfxAddress;
}
else
{
LeaveCriticalSection(&TTLock);
return GMM_ERROR;
}
if(!DoNotWait)
{
PageTableMgr->TTCb.pfPrologTranslationTable(
UmdContext->pCommandQueueHandle);
}
// For each L1 table
for(Addr = GFX_ALIGN_FLOOR(BaseAdr, L1TableSize); // Start at begining of L1 table
Addr < BaseAdr + Size;
Addr += L1TableSize) // Increment by 1 L1 table
{
GMM_GFX_ADDRESS L1GfxAddress, L2GfxAddress;
GMM_GFX_ADDRESS L1CPUAddress, L2CPUAddress;
GMM_GFX_ADDRESS StartAddress = 0;
GMM_GFX_ADDRESS EndAddress = 0;
GMM_GFX_ADDRESS TileAddr = 0;
GMM_GFX_SIZE_T L2eIdx = 0;
StartAddress = Addr < BaseAdr ? BaseAdr : Addr;
EndAddress = Addr + L1TableSize;
if(EndAddress > BaseAdr + Size)
{
EndAddress = BaseAdr + Size;
}
GetL1L2TableAddr(StartAddress,
&L1GfxAddress,
&L2GfxAddress);
// If tables are not there, then they are already invalidated as part of
// AUX-TT initialization or other APIs.
if(L2GfxAddress == GMM_NO_TABLE ||
L1GfxAddress == GMM_NO_TABLE)
{
//Clear Valid-bit for L3Entry or L2Entry
uint64_t Data = 0;
GMM_GFX_ADDRESS TableGfxAddress = (L2GfxAddress == GMM_NO_TABLE) ? L3GfxAddress : L2GfxAddress;
GMM_GFX_ADDRESS TableCPUAddress = (L2GfxAddress == GMM_NO_TABLE) ? TTL3.CPUAddress : pTTL2[GMM_L3_ENTRY_IDX(AUXTT, StartAddress)].GetCPUAddress();
uint32_t TableEntryIdx = (L2GfxAddress == GMM_NO_TABLE) ? static_cast<uint32_t>(GMM_L3_ENTRY_IDX(AUXTT, StartAddress)) : static_cast<uint32_t>(GMM_L2_ENTRY_IDX(AUXTT, StartAddress));
L2CPUAddress = (L2GfxAddress == GMM_NO_TABLE) ? 0 : TableCPUAddress;
if(!NullL1Table || !NullL2Table)
{
AllocateDummyTables(&NullL2Table, &NullL1Table);
if(!NullL1Table || !NullL2Table)
{
//report error
LeaveCriticalSection(&TTLock);
return GMM_OUT_OF_MEMORY;
}
else
{
//Initialize dummy table entries (one-time)
GMM_GFX_ADDRESS TableAddr = NullL2Table->GetCPUAddress();
GMM_AUXTTL2e L2e = {0};
L2e.Valid = 1;
GMM_TO_AUX_L2e_L1GFXADDR_2((NullL1Table->GetPool()->GetGfxAddress() + PAGE_SIZE * NullL1Table->GetNodeIdx()), L2e, (!WA64K(GetGmmLibContext()) && !WA16K(GetGmmLibContext()))) // populate L2e.L1GfxAddr
for(int i = 0; i < GMM_AUX_L2_SIZE; i++)
{
//initialize L2e ie clear Valid bit for all entries
((GMM_AUXTTL2e *)TableAddr)[i].Value = L2e.Value;
}
TableAddr = NullL1Table->GetCPUAddress();
GMM_AUXTTL1e L1e = {0};
L1e.Valid = 1;
if(!WA64K(GetGmmLibContext()))
{
L1e.GfxAddress = (NullCCSTile >> 12); /*********** 4kb-aligned CCS adr *****/
}
else
{
L1e.Reserved4 = (NullCCSTile >> 8); /*********** 4 lsbs of 256B-aligned CCS adr *****/
L1e.GfxAddress = (NullCCSTile >> 12); /*********** 4kb-aligned CCS adr *****/
}
for(int i = 0; i < GMM_AUX_L1_SIZE(GetGmmLibContext()); i++)
{
//initialize L1e with null ccs tile
((GMM_AUXTTL1e *)TableAddr)[i].Value = L1e.Value;
}
}
}
if(L2GfxAddress == GMM_NO_TABLE)
{
GMM_AUXTTL3e L3e = {0};
L3e.Valid = 1;
L3e.L2GfxAddr = (NullL2Table->GetPool()->GetGfxAddress() + PAGE_SIZE * NullL2Table->GetNodeIdx()) >> 15;
Data = L3e.Value;
}
else
{
GMM_AUXTTL2e L2e = {0};
L2e.Valid = 1;
GMM_TO_AUX_L2e_L1GFXADDR_2((NullL1Table->GetPool()->GetGfxAddress() + PAGE_SIZE * NullL1Table->GetNodeIdx()), L2e, (!WA64K(GetGmmLibContext()) && !WA16K(GetGmmLibContext())))
Data = L2e.Value;
}
if(DoNotWait)
{
//Sync update on CPU
((GMM_AUXTTL2e *)TableCPUAddress)[TableEntryIdx].Value = Data;
}
else
{
if(L2GfxAddress != GMM_NO_TABLE)
{
pTTL2[GMM_L3_ENTRY_IDX(AUXTT, StartAddress)].UpdatePoolFence(UmdContext, false);
}
PageTableMgr->TTCb.pfWriteL2L3Entry(UmdContext->pCommandQueueHandle,
TableGfxAddress + TableEntryIdx * GMM_AUX_L2e_SIZE,
Data);
}
continue;
}
else
{
uint32_t L3eIdx = static_cast<uint32_t>(GMM_L3_ENTRY_IDX(AUXTT, StartAddress));
L2CPUAddress = pTTL2[L3eIdx].GetCPUAddress();
L2eIdx = GMM_L2_ENTRY_IDX(AUXTT, StartAddress);
if(DoNotWait)
{
//Sync update on CPU
((GMM_AUXTTL3e *)(TTL3.CPUAddress))[L3eIdx].Valid = 1; //set Valid bit
((GMM_AUXTTL3e *)(TTL3.CPUAddress))[L3eIdx].L2GfxAddr = L2GfxAddress >> 15;
((GMM_AUXTTL2e *)L2CPUAddress)[L2eIdx].Valid = 1; //set Valid bit
GMM_TO_AUX_L2e_L1GFXADDR_2(L1GfxAddress, ((GMM_AUXTTL2e *)L2CPUAddress)[L2eIdx], (!WA64K(GetGmmLibContext()) && !WA16K(GetGmmLibContext())))
}
else
{
GMM_AUXTTL3e L3e = {0};
L3e.Valid = 1;
L3e.L2GfxAddr = L2GfxAddress >> 15;
PageTableMgr->TTCb.pfWriteL2L3Entry(
UmdContext->pCommandQueueHandle,
L3GfxAddress + (L3eIdx * GMM_AUX_L3e_SIZE),
L3e.Value);
pTTL2[L3eIdx].UpdatePoolFence(UmdContext, false);
GMM_AUXTTL2e L2e = {0};
L2e.Valid = 1;
GMM_TO_AUX_L2e_L1GFXADDR_2(L1GfxAddress, L2e, (!WA64K(GetGmmLibContext()) && !WA16K(GetGmmLibContext())))
PageTableMgr->TTCb.pfWriteL2L3Entry(
UmdContext->pCommandQueueHandle,
L2GfxAddress + (L2eIdx * GMM_AUX_L2e_SIZE),
L2e.Value);
}
}
// For each 64KB or 16KB of main surface (entry) in L1 table
for(TileAddr = StartAddress;
TileAddr < EndAddress;
TileAddr += (WA16K(GetGmmLibContext()) ? GMM_KBYTE(16) : WA64K(GetGmmLibContext()) ? GMM_KBYTE(64) : GMM_MBYTE(1)))
{
uint64_t Data = PartialL1e | NullCCSTile | __BIT(0);
GMM_GFX_SIZE_T L1eIdx = GMM_L1_ENTRY_IDX(AUXTT, TileAddr, GetGmmLibContext());
GmmLib::LastLevelTable *pL1Tbl = NULL;
pL1Tbl = pTTL2[GMM_AUX_L3_ENTRY_IDX(TileAddr)].GetL1Table(L2eIdx, NULL);
L1CPUAddress = pL1Tbl->GetCPUAddress();
if(DoNotWait)
{
//Sync update on CPU
((GMM_AUXTTL1e *)L1CPUAddress)[L1eIdx].Value = Data;
GMM_DPF(GFXDBG_NORMAL, "##### Null-Map | Table Entry: TileAddress[0x%llX] L2eIdx[%d] :: L1eIdx[%d] L1Addr[0x%llX] L1Value[00x%llX]\n", TileAddr, L2eIdx, L1eIdx, &((GMM_AUXTTL1e *)L1CPUAddress)[L1eIdx], Data);
}
else
{
pL1Tbl->UpdatePoolFence(UmdContext, false);
/* PageTableMgr->TTCb.pfWriteL1Entries(
UmdContext->pCommandQueueHandle,
2,
L1GfxAddress + (L1eIdx * GMM_AUX_L1e_SIZE),
(uint32_t*)(&Data));*/ //**********REQUIRE UMD CHANGE TO UPDATE 64-bit ENTRY - both DWORDs must be updated atomically*******/
PageTableMgr->TTCb.pfWriteL2L3Entry(
UmdContext->pCommandQueueHandle,
L1GfxAddress + (L1eIdx * GMM_AUX_L1e_SIZE),
Data);
}
if(pL1Tbl->TrackTableUsage(AUXTT, true, TileAddr, true, GetGmmLibContext()))
{ // L1 Table is not being used anymore
GMM_AUXTTL2e L2e = {0};
GmmLib::GMM_PAGETABLEPool *PoolElem = NULL;
GmmLib::LastLevelTable * pL1Tbl = NULL, *Prev = NULL;
pL1Tbl = pTTL2[GMM_L3_ENTRY_IDX(AUXTT, TileAddr)].GetL1Table(L2eIdx, &Prev);
// Map L2-entry to Null-L1Table
L2e.Valid = 1;
GMM_TO_AUX_L2e_L1GFXADDR_2((NullL1Table->GetPool()->GetGfxAddress() + PAGE_SIZE * NullL1Table->GetNodeIdx()), L2e, (!WA64K(GetGmmLibContext()) && !WA16K(GetGmmLibContext()))) // populate L2e.L1GfxAddress/Le2.Reserved2
if(DoNotWait)
{
//Sync update on CPU
((GMM_AUXTTL2e *)L2CPUAddress)[L2eIdx].Value = L2e.Value;
}
else
{
pTTL2[GMM_L3_ENTRY_IDX(AUXTT, TileAddr)].UpdatePoolFence(UmdContext, false);
PageTableMgr->TTCb.pfWriteL2L3Entry(UmdContext->pCommandQueueHandle,
L2GfxAddress + L2eIdx * GMM_AUX_L2e_SIZE,
L2e.Value);
}
//Update usage for PoolNode assigned to L1Table, and free L1Tbl
if(pL1Tbl)
{
PoolElem = pL1Tbl->GetPool();
if(PoolElem)
{
if(pL1Tbl->GetBBInfo().BBQueueHandle)
{
PoolElem->GetNodeBBInfoAtIndex(pL1Tbl->GetNodeIdx()) = pL1Tbl->GetBBInfo();
}
DEASSIGN_POOLNODE(PageTableMgr, UmdContext, PoolElem, pL1Tbl->GetNodeIdx(), AUX_L1TABLE_SIZE_IN_POOLNODES_2(GetGmmLibContext()))
}
pTTL2[GMM_L3_ENTRY_IDX(AUXTT, TileAddr)].DeleteFromList(pL1Tbl, Prev);
}
// The L1 table is unused -- meaning everything else in this table is
// already invalid. So, break early.
break;
}
}
}
if(!DoNotWait)
{
PageTableMgr->TTCb.pfEpilogTranslationTable(
UmdContext->pCommandQueueHandle,
1); // ForceFlush
}
LeaveCriticalSection(&TTLock);
return Status;
}
//=============================================================================
//
// Function: InvalidateTable (InvalidateMappings)
//
// Desc: Unmaps given resource from Aux Table; and marks affected entries as invalid
//
// Caller: UpdateAuxTable (unmap op)
//
// Parameters:
// UmdContext: Caller-thread specific info (regarding BB for Aux udpate, cmdQ to use etc)
// BaseAdr: Start adr of main surface
// Size: Main-surface size in bytes? (or take GmmResInfo?)
// DoNotWait: 1 for CPU update, 0 for async(Gpu) update
//-----------------------------------------------------------------------------
GMM_STATUS GmmLib::AuxTable::InvalidateTable(GMM_UMD_SYNCCONTEXT *UmdContext, GMM_GFX_ADDRESS BaseAdr, GMM_GFX_SIZE_T Size, uint8_t DoNotWait)
{
GMM_STATUS Status = GMM_SUCCESS;
GMM_GFX_SIZE_T L1TableSize = ((GMM_GFX_SIZE_T)GMM_L1_SIZE(AUXTT, GetGmmLibContext())) * (WA16K(GetGmmLibContext()) ? GMM_KBYTE(16) : GMM_KBYTE(64)); //Each AuxTable entry maps 16K main-surface
GMM_GFX_ADDRESS Addr = 0;
GMM_GFX_ADDRESS L3GfxAddress = 0;
uint8_t isTRVA = 0;
GMM_CLIENT ClientType;
GET_GMM_CLIENT_TYPE(pClientContext, ClientType);
//NullCCSTile isn't initialized, disable TRVA path
isTRVA = (NullCCSTile ? isTRVA : 0);
EnterCriticalSection(&TTLock);
DoNotWait |= (!UmdContext || !UmdContext->pCommandQueueHandle);
if(TTL3.L3Handle)
{
L3GfxAddress = TTL3.GfxAddress;
}
else
{
LeaveCriticalSection(&TTLock);
return GMM_ERROR;
}
if(!DoNotWait)
{
PageTableMgr->TTCb.pfPrologTranslationTable(
UmdContext->pCommandQueueHandle);
}
// For each L1 table
for(Addr = GFX_ALIGN_FLOOR(BaseAdr, L1TableSize); // Start at begining of L1 table
Addr < BaseAdr + Size;
Addr += L1TableSize) // Increment by 1 L1 table
{
GMM_GFX_ADDRESS L1GfxAddress, L2GfxAddress;
GMM_GFX_ADDRESS L1CPUAddress, L2CPUAddress;
GMM_GFX_ADDRESS StartAddress = 0;
GMM_GFX_ADDRESS EndAddress = 0;
GMM_GFX_ADDRESS TileAddr = 0;
GMM_GFX_SIZE_T L2eIdx = 0;
StartAddress = Addr < BaseAdr ? BaseAdr : Addr;
EndAddress = Addr + L1TableSize;
if(EndAddress > BaseAdr + Size)
{
EndAddress = BaseAdr + Size;
}
GetL1L2TableAddr(StartAddress,
&L1GfxAddress,
&L2GfxAddress);
// If tables are not there, then they are already invalidated as part of
// AUX-TT initialization or other APIs.
if(L2GfxAddress == GMM_NO_TABLE ||
L1GfxAddress == GMM_NO_TABLE)
{
//Clear Valid-bit for L3Entry or L2Entry
GMM_AUXTTL2e L2e = {0}; //AUXTT L3e is identical to L2e, reuse.
GMM_GFX_ADDRESS TableGfxAddress = (L2GfxAddress == GMM_NO_TABLE) ? L3GfxAddress : L2GfxAddress;
GMM_GFX_ADDRESS TableCPUAddress = (L2GfxAddress == GMM_NO_TABLE) ? TTL3.CPUAddress : pTTL2[GMM_L3_ENTRY_IDX(AUXTT, StartAddress)].GetCPUAddress();
uint32_t TableEntryIdx = (L2GfxAddress == GMM_NO_TABLE) ? static_cast<uint32_t>(GMM_L3_ENTRY_IDX(AUXTT, StartAddress)) : static_cast<uint32_t>(GMM_L2_ENTRY_IDX(AUXTT, StartAddress));
L2CPUAddress = (L2GfxAddress == GMM_NO_TABLE) ? 0 : TableCPUAddress;
if(isTRVA && NullL2Table && NullL1Table)
{
//invalidate if request spans entire stretch ie TileAdr aligns L1TableSize*GMM_L2_SIZE
uint64_t Data = 0;
if(L2GfxAddress == GMM_NO_TABLE)
{
GMM_AUXTTL3e L3e = {0};
L3e.Valid = 1;
L3e.L2GfxAddr = (NullL2Table->GetPool()->GetGfxAddress() + PAGE_SIZE * NullL2Table->GetNodeIdx()) >> 15;
Data = L3e.Value;
}
else
{
GMM_AUXTTL2e L2e = {0};
L2e.Valid = 1;
GMM_TO_AUX_L2e_L1GFXADDR_2((NullL1Table->GetPool()->GetGfxAddress() + PAGE_SIZE * NullL1Table->GetNodeIdx()), L2e, (!WA64K(GetGmmLibContext()) && !WA16K(GetGmmLibContext())))
Data = L2e.Value;
}
L2e.Value = Data;
}
else
{
L2e.Valid = 0;
}
if(DoNotWait)
{
//Sync update on CPU
((GMM_AUXTTL2e *)TableCPUAddress)[TableEntryIdx].Value = L2e.Value;
}
else
{
if(L2GfxAddress != GMM_NO_TABLE)
{
pTTL2[GMM_L3_ENTRY_IDX(AUXTT, StartAddress)].UpdatePoolFence(UmdContext, false);
}
PageTableMgr->TTCb.pfWriteL2L3Entry(UmdContext->pCommandQueueHandle,
TableGfxAddress + TableEntryIdx * GMM_AUX_L2e_SIZE,
L2e.Value);
}
continue;
}
else
{
uint32_t L3eIdx = static_cast<uint32_t>(GMM_L3_ENTRY_IDX(AUXTT, StartAddress));
L2CPUAddress = pTTL2[L3eIdx].GetCPUAddress();
L2eIdx = GMM_L2_ENTRY_IDX(AUXTT, StartAddress);
if(DoNotWait)
{
//Sync update on CPU
((GMM_AUXTTL3e *)(TTL3.CPUAddress))[L3eIdx].Valid = 1; //set Valid bit
((GMM_AUXTTL3e *)(TTL3.CPUAddress))[L3eIdx].L2GfxAddr = L2GfxAddress >> 15;
((GMM_AUXTTL2e *)L2CPUAddress)[L2eIdx].Valid = 1; //set Valid bit
GMM_TO_AUX_L2e_L1GFXADDR_2(L1GfxAddress, ((GMM_AUXTTL2e *)L2CPUAddress)[L2eIdx], (!WA64K(GetGmmLibContext()) && !WA16K(GetGmmLibContext())))
}
else
{
GMM_AUXTTL3e L3e = {0};
L3e.Valid = 1;
L3e.L2GfxAddr = L2GfxAddress >> 15;
PageTableMgr->TTCb.pfWriteL2L3Entry(
UmdContext->pCommandQueueHandle,
L3GfxAddress + (L3eIdx * GMM_AUX_L3e_SIZE),
L3e.Value);
pTTL2[L3eIdx].UpdatePoolFence(UmdContext, false);
GMM_AUXTTL2e L2e = {0};
L2e.Valid = 1;
GMM_TO_AUX_L2e_L1GFXADDR_2(L1GfxAddress, L2e, (!WA64K(GetGmmLibContext()) && !WA16K(GetGmmLibContext())))
PageTableMgr->TTCb.pfWriteL2L3Entry(
UmdContext->pCommandQueueHandle,
L2GfxAddress + (L2eIdx * GMM_AUX_L2e_SIZE),
L2e.Value);
}
}
// For each 64KB or 16KB or 1MB of main surface (entry) in L1 table
for(TileAddr = StartAddress;
TileAddr < EndAddress;
TileAddr += (WA16K(GetGmmLibContext()) ? GMM_KBYTE(16) : WA64K(GetGmmLibContext()) ? GMM_KBYTE(64) : GMM_MBYTE(1)))
{
//Invalidation of requested range irrespective of TRVA
uint64_t Data = GMM_INVALID_AUX_ENTRY;
GMM_GFX_SIZE_T L1eIdx = GMM_L1_ENTRY_IDX(AUXTT, TileAddr, GetGmmLibContext());
GmmLib::LastLevelTable *pL1Tbl = NULL;
pL1Tbl = pTTL2[GMM_AUX_L3_ENTRY_IDX(TileAddr)].GetL1Table(L2eIdx, NULL);
L1CPUAddress = pL1Tbl->GetCPUAddress();
if(DoNotWait)
{
//Sync update on CPU
((GMM_AUXTTL1e *)L1CPUAddress)[L1eIdx].Value = Data;
GMM_DPF(GFXDBG_NORMAL, "~~UnMap | Table Entry: L2addressBase[0x%llX] :: L2Valid[%d] :: L2eidx[%d] L1addressBase[0x%llX] :: L1eidx[%d] L1Valid[%d] :: DerivedCCS[0x%llX] ", (GMM_AUXTTL2e *)L2CPUAddress, ((GMM_AUXTTL2e *)L2CPUAddress)[L2eIdx].Valid, L2eIdx, GMM_L1TABLE_ADDR_FROM_AUX_L2e_L1GFXADDR(((GMM_AUXTTL2e *)L2CPUAddress)[L2eIdx], true), L1eIdx, ((GMM_AUXTTL1e *)L1CPUAddress)[L1eIdx].Valid, (((GMM_AUXTTL1e *)L1CPUAddress)[L1eIdx].GfxAddress << 12));
}
else
{
pL1Tbl->UpdatePoolFence(UmdContext, false);
/* PageTableMgr->TTCb.pfWriteL1Entries(
UmdContext->pCommandQueueHandle,
2,
(uint32_t*)(&Data));*/
//**********REQUIRE UMD CHANGE TO UPDATE 64-bit ENTRY - both DWORDs must be updated atomically*******/
PageTableMgr->TTCb.pfWriteL2L3Entry(
UmdContext->pCommandQueueHandle,
L1GfxAddress + (L1eIdx * GMM_AUX_L1e_SIZE),
Data);
}
if(pL1Tbl->TrackTableUsage(AUXTT, true, TileAddr, true, GetGmmLibContext()))
{ // L1 Table is not being used anymore
GMM_AUXTTL2e L2e = {0};
GmmLib::GMM_PAGETABLEPool *PoolElem = NULL;
GmmLib::LastLevelTable * pL1Tbl = NULL, *Prev = NULL;
pL1Tbl = pTTL2[GMM_L3_ENTRY_IDX(AUXTT, TileAddr)].GetL1Table(L2eIdx, &Prev);
if(isTRVA && NullL1Table &&
((TileAddr > GFX_ALIGN_FLOOR(BaseAdr, L1TableSize) && TileAddr < GFX_ALIGN_NP2(BaseAdr, L1TableSize)) ||
(TileAddr > GFX_ALIGN_FLOOR(BaseAdr + Size, L1TableSize) && TileAddr < GFX_ALIGN_NP2(BaseAdr + Size, L1TableSize))))
{
//Invalidation affects entries out of requested range, null-map for TR
L2e.Valid = 1;
GMM_TO_AUX_L2e_L1GFXADDR_2((NullL1Table->GetPool()->GetGfxAddress() + PAGE_SIZE * NullL1Table->GetNodeIdx()), L2e, (!WA64K(GetGmmLibContext()) && !WA16K(GetGmmLibContext())))
}
else
{
// Clear valid bit of L2 entry
L2e.Valid = 0;
((GMM_AUXTTL2e *)L2CPUAddress)[L2eIdx].Valid = 0;
}
if(DoNotWait)
{
//Sync update on CPU
((GMM_AUXTTL2e *)L2CPUAddress)[L2eIdx].Value = L2e.Value;
}
else
{
pTTL2[GMM_L3_ENTRY_IDX(AUXTT, TileAddr)].UpdatePoolFence(UmdContext, false);
PageTableMgr->TTCb.pfWriteL2L3Entry(UmdContext->pCommandQueueHandle,
L2GfxAddress + L2eIdx * GMM_AUX_L2e_SIZE,
L2e.Value);
}
//Update usage for PoolNode assigned to L1Table, and free L1Tbl
if(pL1Tbl)
{
PoolElem = pL1Tbl->GetPool();
if(PoolElem)
{
if(pL1Tbl->GetBBInfo().BBQueueHandle)
{
PoolElem->GetNodeBBInfoAtIndex(pL1Tbl->GetNodeIdx()) = pL1Tbl->GetBBInfo();
}
DEASSIGN_POOLNODE(PageTableMgr, UmdContext, PoolElem, pL1Tbl->GetNodeIdx(), AUX_L1TABLE_SIZE_IN_POOLNODES_2(GetGmmLibContext()))
}
pTTL2[GMM_L3_ENTRY_IDX(AUXTT, TileAddr)].DeleteFromList(pL1Tbl, Prev);
}
// The L1 table is unused -- meaning everything else in this table is
// already invalid. So, break early.
break;
}
}
}
if(!DoNotWait)
{
PageTableMgr->TTCb.pfEpilogTranslationTable(
UmdContext->pCommandQueueHandle,
1); // ForceFlush
}
LeaveCriticalSection(&TTLock);
return Status;
}
//=============================================================================
//
// Function: MapValidEntry
//
// Desc: Maps given main-surface, on Aux-Table, to get the exact CCS cacheline tied to
// different 4x4K pages of main-surface
//
// Caller: UpdateAuxTable (map op)
//
// Parameters:
// UmdContext: ptr to thread-data
// BaseAdr: Start adr of main-surface
// BaseSize: main-surface Size in bytes
// BaseResInfo: main surface ResInfo
// AuxVA: Start adr of Aux-surface
// AuxResInfo: Aux surface ResInfo
// PartialData: Aux L1 partial data (ie w/o address)
// DoNotWait: true for CPU update, false for async(Gpu) update
//-----------------------------------------------------------------------------
GMM_STATUS GmmLib::AuxTable::MapValidEntry(GMM_UMD_SYNCCONTEXT *UmdContext, GMM_GFX_ADDRESS BaseAdr, GMM_GFX_SIZE_T BaseSize,
GMM_RESOURCE_INFO *BaseResInfo, GMM_GFX_ADDRESS AuxVA, GMM_RESOURCE_INFO *AuxResInfo, uint64_t PartialData, uint8_t DoNotWait)
{
GMM_STATUS Status = GMM_SUCCESS;
GMM_GFX_ADDRESS Addr = 0, L3TableAdr = GMM_NO_TABLE;
GMM_GFX_SIZE_T L1TableSize = GMM_AUX_L1_SIZE(GetGmmLibContext()) * (WA16K(GetGmmLibContext()) ? GMM_KBYTE(16) : GMM_KBYTE(64)); // L1TableSize maps to 16MB address space for TGL and above: 256x64k | 16x1MB
GMM_GFX_SIZE_T CCS$Adr = AuxVA;
uint8_t isTRVA =0 ;
GMM_CLIENT ClientType;
GET_GMM_CLIENT_TYPE(pClientContext, ClientType);
//NullCCSTile isn't initialized, disable TRVA path
isTRVA = (NullCCSTile ? isTRVA : 0);
EnterCriticalSection(&TTLock);
if(!TTL3.L3Handle || (!DoNotWait && !UmdContext))
{
Status = GMM_ERROR;
}
else
{
L3TableAdr = TTL3.GfxAddress;
if(!DoNotWait)
{
PageTableMgr->TTCb.pfPrologTranslationTable(UmdContext->pCommandQueueHandle);
}
// GMM_DPF(GFXDBG_CRITICAL, "Mapping surface: GPUVA=0x%016llX Size=0x%08X Aux_GPUVA=0x%016llX\n", BaseAdr, BaseSize, AuxVA);
for(Addr = GFX_ALIGN_FLOOR(BaseAdr, L1TableSize); Addr < BaseAdr + BaseSize; Addr += L1TableSize)
{
GMM_GFX_ADDRESS StartAdr, EndAdr, TileAdr;
GMM_GFX_ADDRESS L1TableAdr = GMM_NO_TABLE, L2TableAdr = GMM_NO_TABLE;
GMM_GFX_ADDRESS L1TableCPUAdr = GMM_NO_TABLE, L2TableCPUAdr = GMM_NO_TABLE;
GMM_GFX_SIZE_T L2eIdx = 0;
GMM_GFX_SIZE_T L3eIdx = 0;
bool AllocateL1 = false, AllocateL2 = false;
EndAdr = Addr + L1TableSize;
EndAdr = EndAdr > BaseAdr + BaseSize ? BaseAdr + BaseSize : EndAdr;
StartAdr = Addr < BaseAdr ? BaseAdr : Addr;
L2eIdx = GMM_L2_ENTRY_IDX(AUXTT, StartAdr);
L3eIdx = GMM_L3_ENTRY_IDX(AUXTT, StartAdr);
//Allocate L2/L1 Table -- get L2 Table Adr for <StartAdr,EndAdr>
GetL1L2TableAddr(Addr, &L1TableAdr, &L2TableAdr);
if(L2TableAdr == GMM_NO_TABLE || L1TableAdr == GMM_NO_TABLE)
{
AllocateL1 = GMM_NO_TABLE == L1TableAdr;
AllocateL2 = GMM_NO_TABLE == L2TableAdr;
AllocateL1L2Table(Addr, &L1TableAdr, &L2TableAdr);
if(L2TableAdr == GMM_NO_TABLE || L1TableAdr == GMM_NO_TABLE)
{
LeaveCriticalSection(&TTLock);
return GMM_OUT_OF_MEMORY;
}
if(AllocateL2)
{
uint32_t i = 0;
GMM_AUXTTL2e InvalidEntry;
InvalidEntry.Value = 0;
if(isTRVA && NullL1Table)
{
InvalidEntry.Valid = 1;
GMM_TO_AUX_L2e_L1GFXADDR_2((NullL1Table->GetPool()->GetGfxAddress() + PAGE_SIZE * NullL1Table->GetNodeIdx()), InvalidEntry, (!WA16K(GetGmmLibContext()) && !WA64K(GetGmmLibContext())))
}
if(DoNotWait)
{
L2TableCPUAdr = pTTL2[L3eIdx].GetCPUAddress();
((GMM_AUXTTL3e *)(TTL3.CPUAddress))[L3eIdx].Value = 0;
((GMM_AUXTTL3e *)(TTL3.CPUAddress))[L3eIdx].L2GfxAddr = L2TableAdr >> 15;
((GMM_AUXTTL3e *)(TTL3.CPUAddress))[L3eIdx].Valid = 1;
for(i = 0; i < GMM_AUX_L2_SIZE; i++)
{
//initialize L2e ie clear Valid bit for all entries
((GMM_AUXTTL2e *)L2TableCPUAdr)[i].Value = InvalidEntry.Value;
}
}
else
{
GMM_AUXTTL3e L3e = {0};
L3e.Valid = 1;
L3e.L2GfxAddr = L2TableAdr >> 15;
PageTableMgr->TTCb.pfWriteL2L3Entry(UmdContext->pCommandQueueHandle,
L3TableAdr + L3eIdx * GMM_AUX_L3e_SIZE,
L3e.Value);
//initialize L2e ie clear valid bit for all entries
for(i = 0; i < GMM_AUX_L2_SIZE; i++)
{
PageTableMgr->TTCb.pfWriteL2L3Entry(UmdContext->pCommandQueueHandle,
L2TableAdr + i * GMM_AUX_L2e_SIZE,
InvalidEntry.Value);
}
}
}
if(AllocateL1)
{
uint64_t InvalidEntry = (!isTRVA) ? GMM_INVALID_AUX_ENTRY : (NullCCSTile | __BIT(0));
uint32_t i = 0;
if(DoNotWait)
{
GmmLib::LastLevelTable *pL1Tbl = NULL;
pL1Tbl = pTTL2[L3eIdx].GetL1Table(L2eIdx, NULL);
L2TableCPUAdr = pTTL2[L3eIdx].GetCPUAddress();
L1TableCPUAdr = pL1Tbl->GetCPUAddress();
//Sync update on CPU
((GMM_AUXTTL2e *)L2TableCPUAdr)[L2eIdx].Value = 0;
GMM_TO_AUX_L2e_L1GFXADDR_2(L1TableAdr, ((GMM_AUXTTL2e *)L2TableCPUAdr)[L2eIdx], (!WA64K(GetGmmLibContext()) && !WA16K(GetGmmLibContext()))) // populate L2e.L1GfxAddr
((GMM_AUXTTL2e *)L2TableCPUAdr)[L2eIdx]
.Valid = 1;
for(i = 0; i < (uint32_t)GMM_AUX_L1_SIZE(GetGmmLibContext()); i++)
{
//initialize L1e ie mark all entries with Null tile value
((GMM_AUXTTL1e *)L1TableCPUAdr)[i].Value = InvalidEntry;
}
}
else
{
GMM_AUXTTL2e L2e = {0};
L2e.Valid = 1;
GMM_TO_AUX_L2e_L1GFXADDR_2(L1TableAdr, L2e, (!WA64K(GetGmmLibContext()) && !WA16K(GetGmmLibContext())))
pTTL2[L3eIdx]
.UpdatePoolFence(UmdContext, false);
PageTableMgr->TTCb.pfWriteL2L3Entry(UmdContext->pCommandQueueHandle,
L2TableAdr + L2eIdx * GMM_AUX_L2e_SIZE,
L2e.Value);
//initialize all L1e with invalid entries
for(i = 0; i < (uint32_t)GMM_AUX_L1_SIZE(GetGmmLibContext()); i++)
{
PageTableMgr->TTCb.pfWriteL2L3Entry(UmdContext->pCommandQueueHandle,
L1TableAdr + i * sizeof(uint64_t),
InvalidEntry);
}
}
}
}
//GMM_DPF(GFXDBG_NORMAL, "Mapping surface: GPUVA=0x%016llx Size=0x%08x Aux_GPUVA=0x%016llx", StartAdr, BaseSize, CCS$Adr);
for(TileAdr = StartAdr; TileAdr < EndAdr; TileAdr += (WA16K(GetGmmLibContext()) ? GMM_KBYTE(16) : WA64K(GetGmmLibContext()) ? GMM_KBYTE(64) : GMM_MBYTE(1)),
CCS$Adr += (pClientContext->GetLibContext()->GetSkuTable().FtrLinearCCS ?
(WA16K(pClientContext->GetLibContext()) ? GMM_BYTES(64) : WA64K(pClientContext->GetLibContext()) ? GMM_BYTES(256) : GMM_KBYTE(4)) :
0))
{
GMM_GFX_SIZE_T L1eIdx = GMM_L1_ENTRY_IDX(AUXTT, TileAdr, GetGmmLibContext());
GMM_AUXTTL1e L1e = {0};
L1e.Value = PartialData;
L1e.Valid = 1;
if(L1eIdx == 15 || L1eIdx == 14)
{
GMM_DPF(GFXDBG_NORMAL, "\n****** switching over L1*******\n");
}
CCS$Adr = (pClientContext->GetLibContext()->GetSkuTable().FtrLinearCCS ? CCS$Adr :
__GetCCSCacheline(BaseResInfo, BaseAdr, AuxResInfo, AuxVA, TileAdr - BaseAdr));
if(WA16K(GetGmmLibContext()))
{
L1e.Reserved2 = CCS$Adr >> 6; /*********** 2 lsbs of 64B-aligned CCS adr *****/
L1e.Reserved4 = CCS$Adr >> 8; /*********** 256B-aligned CCS adr *****/
L1e.GfxAddress = CCS$Adr >> 12; /*********** 4KB-aligned CCS adr *****/
}
else if(WA64K(GetGmmLibContext()))
{
__GMM_ASSERT((CCS$Adr & 0xFF) == 0x0);
__GMM_ASSERT(GFX_IS_ALIGNED(CCS$Adr, GMM_BYTES(256)));
__GMM_ASSERT(GFX_IS_ALIGNED(TileAdr, GMM_KBYTE(64)));
L1e.Reserved4 = CCS$Adr >> 8; /*********** 4 lsbs of 256B-aligned CCS adr *****/
L1e.GfxAddress = CCS$Adr >> 12; /*********** 4KB-aligned CCS adr *****/
}
else // 1MB aligned address
{
__GMM_ASSERT((CCS$Adr & 0xFF) == 0x0);
__GMM_ASSERT(GFX_IS_ALIGNED(CCS$Adr, GMM_KBYTE(4)));
__GMM_ASSERT(GFX_IS_ALIGNED(TileAdr, GMM_MBYTE(1)));
L1e.GfxAddress = CCS$Adr >> 12; /*********** 4KB-aligned CCS adr *****/
}
GMM_DPF(GFXDBG_NORMAL, "--------------------------------MAP AuxTT Map Address: TileAddr[0x%llX], Size[0x%x], CCSAddr[0x%llX], L3eIdx[%d], L2eIdx[%d], L1eIdx[%d], \n L1CCSAddres[0x%llX] \n", TileAdr, BaseSize, CCS$Adr, L3eIdx, L2eIdx, L1eIdx, L1e.GfxAddress);
GmmLib::LastLevelTable *pL1Tbl = NULL;
pL1Tbl = pTTL2[L3eIdx].GetL1Table(L2eIdx, NULL);
L1TableCPUAdr = pL1Tbl->GetCPUAddress();
if(DoNotWait)
{
//Sync update on CPU
((GMM_AUXTTL1e *)L1TableCPUAdr)[L1eIdx].Value = L1e.Value;
}
else
{
pL1Tbl->UpdatePoolFence(UmdContext, false);
PageTableMgr->TTCb.pfWriteL2L3Entry(UmdContext->pCommandQueueHandle,
L1TableAdr + L1eIdx * GMM_AUX_L1e_SIZE,
L1e.Value);
}
GMM_DPF(GFXDBG_NORMAL, "Map | L3 Table Entry: L3AddressBase[0x%llX] :: L3.L2GfxAddr[0x%llX] :: L3Valid[0x%llX] \n", (GMM_AUXTTL3e *)(TTL3.CPUAddress), ((GMM_AUXTTL3e *)(TTL3.CPUAddress))[L3eIdx].L2GfxAddr, ((GMM_AUXTTL3e *)(TTL3.CPUAddress))[L3eIdx].Valid);
GMM_DPF(GFXDBG_NORMAL, "Map | L2 Table Entry: L2addressBase[0x%llX] :: L2.L1GfxAddr[0x%llX] :: L2Valid[0x%llX] \n", ((GMM_AUXTTL2e *)pTTL2[L3eIdx].GetCPUAddress()), ((GMM_AUXTTL2e *)pTTL2[L3eIdx].GetCPUAddress())[L2eIdx].L1GfxAddr, ((GMM_AUXTTL2e *)pTTL2[L3eIdx].GetCPUAddress())[L2eIdx].Valid);
GMM_DPF(GFXDBG_NORMAL, "Map | L1 Table Entry: L1addressBase[0x%llX] :: L1.CCSAddr[0x%llX] :: L1ValueReserved4[0x%llX] ::L1ValueReserved2[0x%llX] :: L1Valid[0x%llX] :: DerivedCCS[0x%llX] \n\n", ((GMM_AUXTTL1e *)L1TableCPUAdr), ((GMM_AUXTTL1e *)L1TableCPUAdr)[L1eIdx].GfxAddress, ((GMM_AUXTTL1e *)L1TableCPUAdr)[L1eIdx].Reserved4, ((GMM_AUXTTL1e *)L1TableCPUAdr)[L1eIdx].Reserved2, ((GMM_AUXTTL1e *)L1TableCPUAdr)[L1eIdx].Valid, (((GMM_AUXTTL1e *)L1TableCPUAdr)[L1eIdx].GfxAddress << 12));
GMM_DPF(GFXDBG_NORMAL, "**Map | Table Entry: L2addressBase[0x%llX] :: L2Valid[%d] :: L2eidx[%d] L1addressBase[0x%llX] :: L1eidx[%d] L1Valid[0x%llX] :: DerivedCCS[0x%llX]", ((GMM_AUXTTL2e *)pTTL2[L3eIdx].GetCPUAddress()), ((GMM_AUXTTL2e *)pTTL2[L3eIdx].GetCPUAddress())[L2eIdx].Valid, L2eIdx, GMM_L1TABLE_ADDR_FROM_AUX_L2e_L1GFXADDR(((GMM_AUXTTL2e *)pTTL2[L3eIdx].GetCPUAddress())[L2eIdx], true), L1eIdx, ((GMM_AUXTTL1e *)L1TableCPUAdr)[L1eIdx].Valid, (((GMM_AUXTTL1e *)L1TableCPUAdr)[L1eIdx].GfxAddress << 12));
// Since we are mapping a non-null entry, no need to check whether
// L1 table is unused.
pL1Tbl->TrackTableUsage(AUXTT, true, TileAdr, false, GetGmmLibContext());
}
}
if(!DoNotWait)
{
PageTableMgr->TTCb.pfEpilogTranslationTable(
UmdContext->pCommandQueueHandle,
1);
}
}
LeaveCriticalSection(&TTLock);
return Status;
}
GMM_AUXTTL1e GmmLib::AuxTable::CreateAuxL1Data(GMM_RESOURCE_INFO *BaseResInfo)
{
GMM_RESOURCE_FORMAT Format;
Format = BaseResInfo->GetResourceFormat();
Format = ((Format > GMM_FORMAT_INVALID) && (Format < GMM_RESOURCE_FORMATS)) ? Format : GMM_FORMAT_INVALID;
GMM_FORMAT_ENTRY FormatInfo = pClientContext->GetLibContext()->GetPlatformInfo().FormatTable[Format];
GMM_AUXTTL1e L1ePartial = {0};
#define GMM_REGISTRY_UMD_PATH "SOFTWARE\\Intel\\IGFX\\GMM\\"
#define GMM_E2EC_OVERRIDEDEPTH16BPPTO12 "ForceYUV16To12BPP"
L1ePartial.Mode = BaseResInfo->GetResFlags().Info.RenderCompressed ? 0x1 : 0x0; //MC on VCS supports all compression modes,
//MC on Render pipe only 128B compr (until B-step)
//Recognize which .MC surfaces needs Render pipe access
if(pClientContext->GetLibContext()->GetWaTable().WaLimit128BMediaCompr)
{
L1ePartial.Mode = 0x1; //Limit media compression to 128B (same as RC) on gen12LP A0
}
//L1ePartial.Lossy = 0; // when to set it
L1ePartial.TileMode = BaseResInfo->GetResFlags().Info.TiledYs ? 0 : 1;
L1ePartial.Format = FormatInfo.CompressionFormat.AuxL1eFormat;
L1ePartial.LumaChroma = GmmIsPlanar(Format);
if(pClientContext->GetLibContext()->GetWaTable().WaUntypedBufferCompression && BaseResInfo->GetResourceType() == RESOURCE_BUFFER)
{
//Gen12LP WA to support untyped raw buffer compression on HDC ie MLC(machine-learning compression)
L1ePartial.TileMode = 0;
L1ePartial.Depth = 0x6;
L1ePartial.Format = GMM_E2ECOMP_FORMAT_RGBAFLOAT16;
}
__GMM_ASSERT(L1ePartial.Format > GMM_E2ECOMP_MIN_FORMAT && //Are we going to reuse 0x00 for uncompressed indication? CCS contains that info, but only known by HW
L1ePartial.Format <= GMM_E2ECOMP_MAX_FORMAT); //Could SW use it as surface-wide uncompressed state indicator? If so, remove teh assert (Need to make sure, all format encodings are correct)
if(BaseResInfo->GetResFlags().Info.RenderCompressed)
{
if(BaseResInfo->GetResourceType() != RESOURCE_BUFFER)
{
switch(FormatInfo.Element.BitsPer)
{
case 8:
L1ePartial.Depth = 0x4;
break;
case 16:
L1ePartial.Depth = 0x0;
break;
case 32:
L1ePartial.Depth = 0x5;
break;
case 64:
L1ePartial.Depth = 0x6;
break;
case 128:
L1ePartial.Depth = 0x7;
break;
default:
L1ePartial.Depth = 0x3;
}
}
}
else
{
switch(Format)
{
case GMM_FORMAT_P012:
case GMM_FORMAT_Y412:
case GMM_FORMAT_Y212: //which format encoding for Y212, Y412, P012?
L1ePartial.Depth = 0x2;
break;
case GMM_FORMAT_P010:
//case GMM_FORMAT_Y410:
case GMM_FORMAT_Y210: //which format encoding for Y210?
L1ePartial.Depth = 0x1;
break;
case GMM_FORMAT_P016: //per HAS, separate encoding than P010, but a comment says to use P010 in AuxTable?
case GMM_FORMAT_Y416:
case GMM_FORMAT_Y216:
L1ePartial.Depth = 0x0;
break;
default:
L1ePartial.Depth = 0x3; //For MC, bpp got from format encoding
}
if(L1ePartial.Format == GMM_E2ECOMP_FORMAT_R10G10B10A2_UNORM)
{
L1ePartial.Format = GMM_E2ECOMP_FORMAT_RGB10b;
}
}
return L1ePartial;
}
GMM_GFX_ADDRESS GMM_INLINE GmmLib::AuxTable::__GetCCSCacheline(GMM_RESOURCE_INFO *BaseResInfo, GMM_GFX_ADDRESS BaseAdr,
GMM_RESOURCE_INFO *AuxResInfo, GMM_GFX_ADDRESS AuxVA, GMM_GFX_SIZE_T AdrOffset)
{
GMM_GFX_ADDRESS CCSChunkAdr = 0xFFFFFFF0;
uint32_t x = 0, y = 0;
uint32_t i = 0, j = 0;
uint32_t CCSXTile = 0, CCSYTile = 0;
GMM_UNREFERENCED_PARAMETER(BaseAdr);
bool BaseIsYF = BaseResInfo->GetResFlags().Info.TiledYf ? true : false;
uint32_t BasePitchInTiles = BaseResInfo->GetRenderPitchTiles();
//Find YF/YS TileId <x,y> for given main surface 16K-chunk
//and CCS$Id <i,j> corresponding to main's <x,y>
AdrOffset >>= 14; //AdrOffset must be 16K-aligned chunk, since mapping unit is 4 YF pages
if(BaseIsYF)
{
uint32_t PitchIn4YF = BasePitchInTiles / 4; //Base Pitch is physically padded to 4x1 YF width
if (PitchIn4YF != 0)
{
i = static_cast<uint32_t>(AdrOffset % PitchIn4YF);
j = static_cast<uint32_t>(AdrOffset / PitchIn4YF);
}
else
{
__GMM_ASSERT(PitchIn4YF != 0);
return 0;
}
}
else if(BasePitchInTiles != 0) //TileYs
{
x = static_cast<uint32_t>(AdrOffset >> 2); //YS-tile count
y = x / BasePitchInTiles; //YS- tile id <x,y>
x = x % BasePitchInTiles;
i = 2 * x;
j = 2 * y;
switch(AdrOffset % 4) //YS : XYXY [XYXY YF] ie 2x2 16K-units in Y-major
{
case 0:
break;
case 1:
j++;
break;
case 2:
i++;
break;
case 3:
i++;
j++;
break;
}
}
//Compute CCS$ address for <i,j>
CCSXTile = (i >= 8) ? i / 8 : 0; //8x8 CLs make one CCS Tile; get TileOffset
CCSYTile = (j >= 8) ? j / 8 : 0;
i %= 8;
j %= 8;
uint32_t AuxPitchInTiles = AuxResInfo ? AuxResInfo->GetRenderPitchTiles() : BaseResInfo->GetRenderAuxPitchTiles();
CCSChunkAdr = AuxVA + ((CCSXTile + CCSYTile * AuxPitchInTiles) * GMM_KBYTE(4)) + (8 * GMM_BYTES(64) * i) + (GMM_BYTES(64) * j);
return CCSChunkAdr;
}
#endif /*!__GMM_KMD__*/
|
