/
omptarget.cpp
1492 lines (1349 loc) · 59.2 KB
/
omptarget.cpp
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
//===------ omptarget.cpp - Target independent OpenMP target RTL -- C++ -*-===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// Implementation of the interface to be used by Clang during the codegen of a
// target region.
//
//===----------------------------------------------------------------------===//
#include "omptarget.h"
#include "device.h"
#include "private.h"
#include "rtl.h"
#include <cassert>
#include <vector>
int AsyncInfoTy::synchronize() {
int Result = OFFLOAD_SUCCESS;
if (AsyncInfo.Queue) {
// If we have a queue we need to synchronize it now.
Result = Device.synchronize(*this);
assert(AsyncInfo.Queue == nullptr &&
"The device plugin should have nulled the queue to indicate there "
"are no outstanding actions!");
}
return Result;
}
void *&AsyncInfoTy::getVoidPtrLocation() {
BufferLocations.push_back(nullptr);
return BufferLocations.back();
}
/* All begin addresses for partially mapped structs must be 8-aligned in order
* to ensure proper alignment of members. E.g.
*
* struct S {
* int a; // 4-aligned
* int b; // 4-aligned
* int *p; // 8-aligned
* } s1;
* ...
* #pragma omp target map(tofrom: s1.b, s1.p[0:N])
* {
* s1.b = 5;
* for (int i...) s1.p[i] = ...;
* }
*
* Here we are mapping s1 starting from member b, so BaseAddress=&s1=&s1.a and
* BeginAddress=&s1.b. Let's assume that the struct begins at address 0x100,
* then &s1.a=0x100, &s1.b=0x104, &s1.p=0x108. Each member obeys the alignment
* requirements for its type. Now, when we allocate memory on the device, in
* CUDA's case cuMemAlloc() returns an address which is at least 256-aligned.
* This means that the chunk of the struct on the device will start at a
* 256-aligned address, let's say 0x200. Then the address of b will be 0x200 and
* address of p will be a misaligned 0x204 (on the host there was no need to add
* padding between b and p, so p comes exactly 4 bytes after b). If the device
* kernel tries to access s1.p, a misaligned address error occurs (as reported
* by the CUDA plugin). By padding the begin address down to a multiple of 8 and
* extending the size of the allocated chuck accordingly, the chuck on the
* device will start at 0x200 with the padding (4 bytes), then &s1.b=0x204 and
* &s1.p=0x208, as they should be to satisfy the alignment requirements.
*/
static const int64_t Alignment = 8;
/// Map global data and execute pending ctors
static int InitLibrary(DeviceTy &Device) {
/*
* Map global data
*/
int32_t device_id = Device.DeviceID;
int rc = OFFLOAD_SUCCESS;
bool supportsEmptyImages = Device.RTL->supports_empty_images &&
Device.RTL->supports_empty_images() > 0;
Device.PendingGlobalsMtx.lock();
PM->TrlTblMtx.lock();
for (auto *HostEntriesBegin : PM->HostEntriesBeginRegistrationOrder) {
TranslationTable *TransTable =
&PM->HostEntriesBeginToTransTable[HostEntriesBegin];
if (TransTable->HostTable.EntriesBegin ==
TransTable->HostTable.EntriesEnd &&
!supportsEmptyImages) {
// No host entry so no need to proceed
continue;
}
if (TransTable->TargetsTable[device_id] != 0) {
// Library entries have already been processed
continue;
}
// 1) get image.
assert(TransTable->TargetsImages.size() > (size_t)device_id &&
"Not expecting a device ID outside the table's bounds!");
__tgt_device_image *img = TransTable->TargetsImages[device_id];
if (!img) {
REPORT("No image loaded for device id %d.\n", device_id);
rc = OFFLOAD_FAIL;
break;
}
// 2) load image into the target table.
__tgt_target_table *TargetTable = TransTable->TargetsTable[device_id] =
Device.load_binary(img);
// Unable to get table for this image: invalidate image and fail.
if (!TargetTable) {
REPORT("Unable to generate entries table for device id %d.\n", device_id);
TransTable->TargetsImages[device_id] = 0;
rc = OFFLOAD_FAIL;
break;
}
// Verify whether the two table sizes match.
size_t hsize =
TransTable->HostTable.EntriesEnd - TransTable->HostTable.EntriesBegin;
size_t tsize = TargetTable->EntriesEnd - TargetTable->EntriesBegin;
// Invalid image for these host entries!
if (hsize != tsize) {
REPORT("Host and Target tables mismatch for device id %d [%zx != %zx].\n",
device_id, hsize, tsize);
TransTable->TargetsImages[device_id] = 0;
TransTable->TargetsTable[device_id] = 0;
rc = OFFLOAD_FAIL;
break;
}
// process global data that needs to be mapped.
Device.DataMapMtx.lock();
__tgt_target_table *HostTable = &TransTable->HostTable;
for (__tgt_offload_entry *CurrDeviceEntry = TargetTable->EntriesBegin,
*CurrHostEntry = HostTable->EntriesBegin,
*EntryDeviceEnd = TargetTable->EntriesEnd;
CurrDeviceEntry != EntryDeviceEnd;
CurrDeviceEntry++, CurrHostEntry++) {
if (CurrDeviceEntry->size != 0) {
// has data.
assert(CurrDeviceEntry->size == CurrHostEntry->size &&
"data size mismatch");
// Fortran may use multiple weak declarations for the same symbol,
// therefore we must allow for multiple weak symbols to be loaded from
// the fat binary. Treat these mappings as any other "regular" mapping.
// Add entry to map.
if (Device.getTgtPtrBegin(CurrHostEntry->addr, CurrHostEntry->size))
continue;
DP("Add mapping from host " DPxMOD " to device " DPxMOD " with size %zu"
"\n",
DPxPTR(CurrHostEntry->addr), DPxPTR(CurrDeviceEntry->addr),
CurrDeviceEntry->size);
Device.HostDataToTargetMap.emplace(
(uintptr_t)CurrHostEntry->addr /*HstPtrBase*/,
(uintptr_t)CurrHostEntry->addr /*HstPtrBegin*/,
(uintptr_t)CurrHostEntry->addr + CurrHostEntry->size /*HstPtrEnd*/,
(uintptr_t)CurrDeviceEntry->addr /*TgtPtrBegin*/, nullptr,
true /*IsRefCountINF*/);
}
}
Device.DataMapMtx.unlock();
}
PM->TrlTblMtx.unlock();
if (rc != OFFLOAD_SUCCESS) {
Device.PendingGlobalsMtx.unlock();
return rc;
}
/*
* Run ctors for static objects
*/
if (!Device.PendingCtorsDtors.empty()) {
AsyncInfoTy AsyncInfo(Device);
// Call all ctors for all libraries registered so far
for (auto &lib : Device.PendingCtorsDtors) {
if (!lib.second.PendingCtors.empty()) {
DP("Has pending ctors... call now\n");
for (auto &entry : lib.second.PendingCtors) {
void *ctor = entry;
int rc =
target(nullptr, Device, ctor, 0, nullptr, nullptr, nullptr,
nullptr, nullptr, nullptr, 1, 1, true /*team*/, AsyncInfo);
if (rc != OFFLOAD_SUCCESS) {
REPORT("Running ctor " DPxMOD " failed.\n", DPxPTR(ctor));
Device.PendingGlobalsMtx.unlock();
return OFFLOAD_FAIL;
}
}
// Clear the list to indicate that this device has been used
lib.second.PendingCtors.clear();
DP("Done with pending ctors for lib " DPxMOD "\n", DPxPTR(lib.first));
}
}
// All constructors have been issued, wait for them now.
if (AsyncInfo.synchronize() != OFFLOAD_SUCCESS)
return OFFLOAD_FAIL;
}
Device.HasPendingGlobals = false;
Device.PendingGlobalsMtx.unlock();
return OFFLOAD_SUCCESS;
}
void handleTargetOutcome(bool Success, ident_t *Loc) {
switch (PM->TargetOffloadPolicy) {
case tgt_disabled:
if (Success) {
FATAL_MESSAGE0(1, "expected no offloading while offloading is disabled");
}
break;
case tgt_default:
FATAL_MESSAGE0(1, "default offloading policy must be switched to "
"mandatory or disabled");
break;
case tgt_mandatory:
if (!Success) {
if (getInfoLevel() & OMP_INFOTYPE_DUMP_TABLE)
for (auto &Device : PM->Devices)
dumpTargetPointerMappings(Loc, Device);
else
FAILURE_MESSAGE("Run with LIBOMPTARGET_INFO=%d to dump host-target "
"pointer mappings.\n",
OMP_INFOTYPE_DUMP_TABLE);
SourceInfo info(Loc);
if (info.isAvailible())
fprintf(stderr, "%s:%d:%d: ", info.getFilename(), info.getLine(),
info.getColumn());
else
FAILURE_MESSAGE("Source location information not present. Compile with "
"-g or -gline-tables-only.\n");
FATAL_MESSAGE0(
1, "failure of target construct while offloading is mandatory");
} else {
if (getInfoLevel() & OMP_INFOTYPE_DUMP_TABLE)
for (auto &Device : PM->Devices)
dumpTargetPointerMappings(Loc, Device);
}
break;
}
}
static void handleDefaultTargetOffload() {
PM->TargetOffloadMtx.lock();
if (PM->TargetOffloadPolicy == tgt_default) {
if (omp_get_num_devices() > 0) {
DP("Default TARGET OFFLOAD policy is now mandatory "
"(devices were found)\n");
PM->TargetOffloadPolicy = tgt_mandatory;
} else {
DP("Default TARGET OFFLOAD policy is now disabled "
"(no devices were found)\n");
PM->TargetOffloadPolicy = tgt_disabled;
}
}
PM->TargetOffloadMtx.unlock();
}
static bool isOffloadDisabled() {
if (PM->TargetOffloadPolicy == tgt_default)
handleDefaultTargetOffload();
return PM->TargetOffloadPolicy == tgt_disabled;
}
// If offload is enabled, ensure that device DeviceID has been initialized,
// global ctors have been executed, and global data has been mapped.
//
// There are three possible results:
// - Return OFFLOAD_SUCCESS if the device is ready for offload.
// - Return OFFLOAD_FAIL without reporting a runtime error if offload is
// disabled, perhaps because the initial device was specified.
// - Report a runtime error and return OFFLOAD_FAIL.
//
// If DeviceID == OFFLOAD_DEVICE_DEFAULT, set DeviceID to the default device.
// This step might be skipped if offload is disabled.
int checkDeviceAndCtors(int64_t &DeviceID, ident_t *Loc) {
if (isOffloadDisabled()) {
DP("Offload is disabled\n");
return OFFLOAD_FAIL;
}
if (DeviceID == OFFLOAD_DEVICE_DEFAULT) {
DeviceID = omp_get_default_device();
DP("Use default device id %" PRId64 "\n", DeviceID);
}
// Proposed behavior for OpenMP 5.2 in OpenMP spec github issue 2669.
if (omp_get_num_devices() == 0) {
DP("omp_get_num_devices() == 0 but offload is manadatory\n");
handleTargetOutcome(false, Loc);
return OFFLOAD_FAIL;
}
if (DeviceID == omp_get_initial_device()) {
DP("Device is host (%" PRId64 "), returning as if offload is disabled\n",
DeviceID);
return OFFLOAD_FAIL;
}
// Is device ready?
if (!device_is_ready(DeviceID)) {
REPORT("Device %" PRId64 " is not ready.\n", DeviceID);
handleTargetOutcome(false, Loc);
return OFFLOAD_FAIL;
}
// Get device info.
DeviceTy &Device = PM->Devices[DeviceID];
// Check whether global data has been mapped for this device
Device.PendingGlobalsMtx.lock();
bool hasPendingGlobals = Device.HasPendingGlobals;
Device.PendingGlobalsMtx.unlock();
if (hasPendingGlobals && InitLibrary(Device) != OFFLOAD_SUCCESS) {
REPORT("Failed to init globals on device %" PRId64 "\n", DeviceID);
handleTargetOutcome(false, Loc);
return OFFLOAD_FAIL;
}
return OFFLOAD_SUCCESS;
}
static int32_t getParentIndex(int64_t type) {
return ((type & OMP_TGT_MAPTYPE_MEMBER_OF) >> 48) - 1;
}
void *targetAllocExplicit(size_t size, int device_num, int kind,
const char *name) {
TIMESCOPE();
DP("Call to %s for device %d requesting %zu bytes\n", name, device_num, size);
if (size <= 0) {
DP("Call to %s with non-positive length\n", name);
return NULL;
}
void *rc = NULL;
if (device_num == omp_get_initial_device()) {
rc = malloc(size);
DP("%s returns host ptr " DPxMOD "\n", name, DPxPTR(rc));
return rc;
}
if (!device_is_ready(device_num)) {
DP("%s returns NULL ptr\n", name);
return NULL;
}
DeviceTy &Device = PM->Devices[device_num];
rc = Device.allocData(size, nullptr, kind);
DP("%s returns device ptr " DPxMOD "\n", name, DPxPTR(rc));
return rc;
}
/// Call the user-defined mapper function followed by the appropriate
// targetData* function (targetData{Begin,End,Update}).
int targetDataMapper(ident_t *loc, DeviceTy &Device, void *arg_base, void *arg,
int64_t arg_size, int64_t arg_type,
map_var_info_t arg_names, void *arg_mapper,
AsyncInfoTy &AsyncInfo,
TargetDataFuncPtrTy target_data_function) {
TIMESCOPE_WITH_IDENT(loc);
DP("Calling the mapper function " DPxMOD "\n", DPxPTR(arg_mapper));
// The mapper function fills up Components.
MapperComponentsTy MapperComponents;
MapperFuncPtrTy MapperFuncPtr = (MapperFuncPtrTy)(arg_mapper);
(*MapperFuncPtr)((void *)&MapperComponents, arg_base, arg, arg_size, arg_type,
arg_names);
// Construct new arrays for args_base, args, arg_sizes and arg_types
// using the information in MapperComponents and call the corresponding
// targetData* function using these new arrays.
std::vector<void *> MapperArgsBase(MapperComponents.Components.size());
std::vector<void *> MapperArgs(MapperComponents.Components.size());
std::vector<int64_t> MapperArgSizes(MapperComponents.Components.size());
std::vector<int64_t> MapperArgTypes(MapperComponents.Components.size());
std::vector<void *> MapperArgNames(MapperComponents.Components.size());
for (unsigned I = 0, E = MapperComponents.Components.size(); I < E; ++I) {
auto &C = MapperComponents.Components[I];
MapperArgsBase[I] = C.Base;
MapperArgs[I] = C.Begin;
MapperArgSizes[I] = C.Size;
MapperArgTypes[I] = C.Type;
MapperArgNames[I] = C.Name;
}
int rc = target_data_function(loc, Device, MapperComponents.Components.size(),
MapperArgsBase.data(), MapperArgs.data(),
MapperArgSizes.data(), MapperArgTypes.data(),
MapperArgNames.data(), /*arg_mappers*/ nullptr,
AsyncInfo, /*FromMapper=*/true);
return rc;
}
/// Internal function to do the mapping and transfer the data to the device
int targetDataBegin(ident_t *loc, DeviceTy &Device, int32_t arg_num,
void **args_base, void **args, int64_t *arg_sizes,
int64_t *arg_types, map_var_info_t *arg_names,
void **arg_mappers, AsyncInfoTy &AsyncInfo,
bool FromMapper) {
// process each input.
for (int32_t i = 0; i < arg_num; ++i) {
// Ignore private variables and arrays - there is no mapping for them.
if ((arg_types[i] & OMP_TGT_MAPTYPE_LITERAL) ||
(arg_types[i] & OMP_TGT_MAPTYPE_PRIVATE))
continue;
if (arg_mappers && arg_mappers[i]) {
// Instead of executing the regular path of targetDataBegin, call the
// targetDataMapper variant which will call targetDataBegin again
// with new arguments.
DP("Calling targetDataMapper for the %dth argument\n", i);
map_var_info_t arg_name = (!arg_names) ? nullptr : arg_names[i];
int rc = targetDataMapper(loc, Device, args_base[i], args[i],
arg_sizes[i], arg_types[i], arg_name,
arg_mappers[i], AsyncInfo, targetDataBegin);
if (rc != OFFLOAD_SUCCESS) {
REPORT("Call to targetDataBegin via targetDataMapper for custom mapper"
" failed.\n");
return OFFLOAD_FAIL;
}
// Skip the rest of this function, continue to the next argument.
continue;
}
void *HstPtrBegin = args[i];
void *HstPtrBase = args_base[i];
int64_t data_size = arg_sizes[i];
map_var_info_t HstPtrName = (!arg_names) ? nullptr : arg_names[i];
// Adjust for proper alignment if this is a combined entry (for structs).
// Look at the next argument - if that is MEMBER_OF this one, then this one
// is a combined entry.
int64_t padding = 0;
const int next_i = i + 1;
if (getParentIndex(arg_types[i]) < 0 && next_i < arg_num &&
getParentIndex(arg_types[next_i]) == i) {
padding = (int64_t)HstPtrBegin % Alignment;
if (padding) {
DP("Using a padding of %" PRId64 " bytes for begin address " DPxMOD
"\n",
padding, DPxPTR(HstPtrBegin));
HstPtrBegin = (char *)HstPtrBegin - padding;
data_size += padding;
}
}
// Address of pointer on the host and device, respectively.
void *Pointer_HstPtrBegin, *PointerTgtPtrBegin;
TargetPointerResultTy Pointer_TPR;
bool IsHostPtr = false;
bool IsImplicit = arg_types[i] & OMP_TGT_MAPTYPE_IMPLICIT;
// Force the creation of a device side copy of the data when:
// a close map modifier was associated with a map that contained a to.
bool HasCloseModifier = arg_types[i] & OMP_TGT_MAPTYPE_CLOSE;
bool HasPresentModifier = arg_types[i] & OMP_TGT_MAPTYPE_PRESENT;
// UpdateRef is based on MEMBER_OF instead of TARGET_PARAM because if we
// have reached this point via __tgt_target_data_begin and not __tgt_target
// then no argument is marked as TARGET_PARAM ("omp target data map" is not
// associated with a target region, so there are no target parameters). This
// may be considered a hack, we could revise the scheme in the future.
bool UpdateRef =
!(arg_types[i] & OMP_TGT_MAPTYPE_MEMBER_OF) && !(FromMapper && i == 0);
if (arg_types[i] & OMP_TGT_MAPTYPE_PTR_AND_OBJ) {
DP("Has a pointer entry: \n");
// Base is address of pointer.
//
// Usually, the pointer is already allocated by this time. For example:
//
// #pragma omp target map(s.p[0:N])
//
// The map entry for s comes first, and the PTR_AND_OBJ entry comes
// afterward, so the pointer is already allocated by the time the
// PTR_AND_OBJ entry is handled below, and PointerTgtPtrBegin is thus
// non-null. However, "declare target link" can produce a PTR_AND_OBJ
// entry for a global that might not already be allocated by the time the
// PTR_AND_OBJ entry is handled below, and so the allocation might fail
// when HasPresentModifier.
Pointer_TPR = Device.getTargetPointer(
HstPtrBase, HstPtrBase, sizeof(void *), nullptr,
MoveDataStateTy::NONE, IsImplicit, UpdateRef, HasCloseModifier,
HasPresentModifier, AsyncInfo);
PointerTgtPtrBegin = Pointer_TPR.TargetPointer;
IsHostPtr = Pointer_TPR.Flags.IsHostPointer;
if (!PointerTgtPtrBegin) {
REPORT("Call to getOrAllocTgtPtr returned null pointer (%s).\n",
HasPresentModifier ? "'present' map type modifier"
: "device failure or illegal mapping");
return OFFLOAD_FAIL;
}
DP("There are %zu bytes allocated at target address " DPxMOD " - is%s new"
"\n",
sizeof(void *), DPxPTR(PointerTgtPtrBegin),
(Pointer_TPR.Flags.IsNewEntry ? "" : " not"));
Pointer_HstPtrBegin = HstPtrBase;
// modify current entry.
HstPtrBase = *(void **)HstPtrBase;
// No need to update pointee ref count for the first element of the
// subelement that comes from mapper.
UpdateRef =
(!FromMapper || i != 0); // subsequently update ref count of pointee
}
MoveDataStateTy MoveData = MoveDataStateTy::NONE;
const bool UseUSM = PM->RTLs.RequiresFlags & OMP_REQ_UNIFIED_SHARED_MEMORY;
const bool HasFlagTo = arg_types[i] & OMP_TGT_MAPTYPE_TO;
const bool HasFlagAlways = arg_types[i] & OMP_TGT_MAPTYPE_ALWAYS;
if (HasFlagTo && (!UseUSM || HasCloseModifier))
MoveData = HasFlagAlways ? MoveDataStateTy::REQUIRED
: MoveDataStateTy::UNKNOWN;
auto TPR = Device.getTargetPointer(
HstPtrBegin, HstPtrBase, data_size, HstPtrName, MoveData, IsImplicit,
UpdateRef, HasCloseModifier, HasPresentModifier, AsyncInfo);
void *TgtPtrBegin = TPR.TargetPointer;
IsHostPtr = TPR.Flags.IsHostPointer;
// If data_size==0, then the argument could be a zero-length pointer to
// NULL, so getOrAlloc() returning NULL is not an error.
if (!TgtPtrBegin && (data_size || HasPresentModifier)) {
REPORT("Call to getOrAllocTgtPtr returned null pointer (%s).\n",
HasPresentModifier ? "'present' map type modifier"
: "device failure or illegal mapping");
return OFFLOAD_FAIL;
}
DP("There are %" PRId64 " bytes allocated at target address " DPxMOD
" - is%s new\n",
data_size, DPxPTR(TgtPtrBegin), (TPR.Flags.IsNewEntry ? "" : " not"));
if (arg_types[i] & OMP_TGT_MAPTYPE_RETURN_PARAM) {
uintptr_t Delta = (uintptr_t)HstPtrBegin - (uintptr_t)HstPtrBase;
void *TgtPtrBase = (void *)((uintptr_t)TgtPtrBegin - Delta);
DP("Returning device pointer " DPxMOD "\n", DPxPTR(TgtPtrBase));
args_base[i] = TgtPtrBase;
}
if (arg_types[i] & OMP_TGT_MAPTYPE_PTR_AND_OBJ && !IsHostPtr) {
// Check whether we need to update the pointer on the device
bool UpdateDevPtr = false;
uint64_t Delta = (uint64_t)HstPtrBegin - (uint64_t)HstPtrBase;
void *ExpectedTgtPtrBase = (void *)((uint64_t)TgtPtrBegin - Delta);
Device.ShadowMtx.lock();
auto Entry = Device.ShadowPtrMap.find(Pointer_HstPtrBegin);
// If this pointer is not in the map we need to insert it. If the map
// contains a stale entry, we need to update it (e.g. if the pointee was
// deallocated and later on is reallocated at another device address). The
// latter scenario is the subject of LIT test env/base_ptr_ref_count.c. An
// entry is removed from ShadowPtrMap only when the PTR of a PTR_AND_OBJ
// pair is deallocated, not when the OBJ is deallocated. In
// env/base_ptr_ref_count.c the PTR is a global "declare target" pointer,
// so it stays in the map for the lifetime of the application. When the
// OBJ is deallocated and later on allocated again (at a different device
// address), ShadowPtrMap still contains an entry for Pointer_HstPtrBegin
// which is stale, pointing to the old ExpectedTgtPtrBase of the OBJ.
if (Entry == Device.ShadowPtrMap.end() ||
Entry->second.TgtPtrVal != ExpectedTgtPtrBase) {
// create or update shadow pointers for this entry
Device.ShadowPtrMap[Pointer_HstPtrBegin] = {
HstPtrBase, PointerTgtPtrBegin, ExpectedTgtPtrBase};
UpdateDevPtr = true;
}
if (UpdateDevPtr) {
Pointer_TPR.MapTableEntry->lock();
Device.ShadowMtx.unlock();
DP("Update pointer (" DPxMOD ") -> [" DPxMOD "]\n",
DPxPTR(PointerTgtPtrBegin), DPxPTR(TgtPtrBegin));
void *&TgtPtrBase = AsyncInfo.getVoidPtrLocation();
TgtPtrBase = ExpectedTgtPtrBase;
int rt = Device.submitData(PointerTgtPtrBegin, &TgtPtrBase,
sizeof(void *), AsyncInfo);
Pointer_TPR.MapTableEntry->unlock();
if (rt != OFFLOAD_SUCCESS) {
REPORT("Copying data to device failed.\n");
return OFFLOAD_FAIL;
}
} else
Device.ShadowMtx.unlock();
}
}
return OFFLOAD_SUCCESS;
}
namespace {
/// This structure contains information to deallocate a target pointer, aka.
/// used to call the function \p DeviceTy::deallocTgtPtr.
struct DeallocTgtPtrInfo {
/// Host pointer used to look up into the map table
void *HstPtrBegin;
/// Size of the data
int64_t DataSize;
/// Whether it has \p close modifier
bool HasCloseModifier;
DeallocTgtPtrInfo(void *HstPtr, int64_t Size, bool HasCloseModifier)
: HstPtrBegin(HstPtr), DataSize(Size),
HasCloseModifier(HasCloseModifier) {}
};
} // namespace
/// Internal function to undo the mapping and retrieve the data from the device.
int targetDataEnd(ident_t *loc, DeviceTy &Device, int32_t ArgNum,
void **ArgBases, void **Args, int64_t *ArgSizes,
int64_t *ArgTypes, map_var_info_t *ArgNames,
void **ArgMappers, AsyncInfoTy &AsyncInfo, bool FromMapper) {
int Ret;
std::vector<DeallocTgtPtrInfo> DeallocTgtPtrs;
void *FromMapperBase = nullptr;
// process each input.
for (int32_t I = ArgNum - 1; I >= 0; --I) {
// Ignore private variables and arrays - there is no mapping for them.
// Also, ignore the use_device_ptr directive, it has no effect here.
if ((ArgTypes[I] & OMP_TGT_MAPTYPE_LITERAL) ||
(ArgTypes[I] & OMP_TGT_MAPTYPE_PRIVATE))
continue;
if (ArgMappers && ArgMappers[I]) {
// Instead of executing the regular path of targetDataEnd, call the
// targetDataMapper variant which will call targetDataEnd again
// with new arguments.
DP("Calling targetDataMapper for the %dth argument\n", I);
map_var_info_t ArgName = (!ArgNames) ? nullptr : ArgNames[I];
Ret = targetDataMapper(loc, Device, ArgBases[I], Args[I], ArgSizes[I],
ArgTypes[I], ArgName, ArgMappers[I], AsyncInfo,
targetDataEnd);
if (Ret != OFFLOAD_SUCCESS) {
REPORT("Call to targetDataEnd via targetDataMapper for custom mapper"
" failed.\n");
return OFFLOAD_FAIL;
}
// Skip the rest of this function, continue to the next argument.
continue;
}
void *HstPtrBegin = Args[I];
int64_t DataSize = ArgSizes[I];
// Adjust for proper alignment if this is a combined entry (for structs).
// Look at the next argument - if that is MEMBER_OF this one, then this one
// is a combined entry.
const int NextI = I + 1;
if (getParentIndex(ArgTypes[I]) < 0 && NextI < ArgNum &&
getParentIndex(ArgTypes[NextI]) == I) {
int64_t Padding = (int64_t)HstPtrBegin % Alignment;
if (Padding) {
DP("Using a Padding of %" PRId64 " bytes for begin address " DPxMOD
"\n",
Padding, DPxPTR(HstPtrBegin));
HstPtrBegin = (char *)HstPtrBegin - Padding;
DataSize += Padding;
}
}
bool IsLast, IsHostPtr;
bool IsImplicit = ArgTypes[I] & OMP_TGT_MAPTYPE_IMPLICIT;
bool UpdateRef = (!(ArgTypes[I] & OMP_TGT_MAPTYPE_MEMBER_OF) ||
(ArgTypes[I] & OMP_TGT_MAPTYPE_PTR_AND_OBJ)) &&
!(FromMapper && I == 0);
bool ForceDelete = ArgTypes[I] & OMP_TGT_MAPTYPE_DELETE;
bool HasCloseModifier = ArgTypes[I] & OMP_TGT_MAPTYPE_CLOSE;
bool HasPresentModifier = ArgTypes[I] & OMP_TGT_MAPTYPE_PRESENT;
// If PTR_AND_OBJ, HstPtrBegin is address of pointee
void *TgtPtrBegin =
Device.getTgtPtrBegin(HstPtrBegin, DataSize, IsLast, UpdateRef,
IsHostPtr, !IsImplicit, ForceDelete);
if (!TgtPtrBegin && (DataSize || HasPresentModifier)) {
DP("Mapping does not exist (%s)\n",
(HasPresentModifier ? "'present' map type modifier" : "ignored"));
if (HasPresentModifier) {
// OpenMP 5.1, sec. 2.21.7.1 "map Clause", p. 350 L10-13:
// "If a map clause appears on a target, target data, target enter data
// or target exit data construct with a present map-type-modifier then
// on entry to the region if the corresponding list item does not appear
// in the device data environment then an error occurs and the program
// terminates."
//
// This should be an error upon entering an "omp target exit data". It
// should not be an error upon exiting an "omp target data" or "omp
// target". For "omp target data", Clang thus doesn't include present
// modifiers for end calls. For "omp target", we have not found a valid
// OpenMP program for which the error matters: it appears that, if a
// program can guarantee that data is present at the beginning of an
// "omp target" region so that there's no error there, that data is also
// guaranteed to be present at the end.
MESSAGE("device mapping required by 'present' map type modifier does "
"not exist for host address " DPxMOD " (%" PRId64 " bytes)",
DPxPTR(HstPtrBegin), DataSize);
return OFFLOAD_FAIL;
}
} else {
DP("There are %" PRId64 " bytes allocated at target address " DPxMOD
" - is%s last\n",
DataSize, DPxPTR(TgtPtrBegin), (IsLast ? "" : " not"));
}
// OpenMP 5.1, sec. 2.21.7.1 "map Clause", p. 351 L14-16:
// "If the map clause appears on a target, target data, or target exit data
// construct and a corresponding list item of the original list item is not
// present in the device data environment on exit from the region then the
// list item is ignored."
if (!TgtPtrBegin)
continue;
bool DelEntry = IsLast;
// If the last element from the mapper (for end transfer args comes in
// reverse order), do not remove the partial entry, the parent struct still
// exists.
if ((ArgTypes[I] & OMP_TGT_MAPTYPE_MEMBER_OF) &&
!(ArgTypes[I] & OMP_TGT_MAPTYPE_PTR_AND_OBJ)) {
DelEntry = false; // protect parent struct from being deallocated
}
if ((ArgTypes[I] & OMP_TGT_MAPTYPE_FROM) || DelEntry) {
// Move data back to the host
if (ArgTypes[I] & OMP_TGT_MAPTYPE_FROM) {
bool Always = ArgTypes[I] & OMP_TGT_MAPTYPE_ALWAYS;
bool CopyMember = false;
if (!(PM->RTLs.RequiresFlags & OMP_REQ_UNIFIED_SHARED_MEMORY) ||
HasCloseModifier) {
if (IsLast)
CopyMember = true;
}
if ((DelEntry || Always || CopyMember) &&
!(PM->RTLs.RequiresFlags & OMP_REQ_UNIFIED_SHARED_MEMORY &&
TgtPtrBegin == HstPtrBegin)) {
DP("Moving %" PRId64 " bytes (tgt:" DPxMOD ") -> (hst:" DPxMOD ")\n",
DataSize, DPxPTR(TgtPtrBegin), DPxPTR(HstPtrBegin));
Ret = Device.retrieveData(HstPtrBegin, TgtPtrBegin, DataSize,
AsyncInfo);
if (Ret != OFFLOAD_SUCCESS) {
REPORT("Copying data from device failed.\n");
return OFFLOAD_FAIL;
}
}
}
if (DelEntry && FromMapper && I == 0) {
DelEntry = false;
FromMapperBase = HstPtrBegin;
}
// If we copied back to the host a struct/array containing pointers, we
// need to restore the original host pointer values from their shadow
// copies. If the struct is going to be deallocated, remove any remaining
// shadow pointer entries for this struct.
uintptr_t LB = (uintptr_t)HstPtrBegin;
uintptr_t UB = (uintptr_t)HstPtrBegin + DataSize;
Device.ShadowMtx.lock();
for (ShadowPtrListTy::iterator Itr = Device.ShadowPtrMap.begin();
Itr != Device.ShadowPtrMap.end();) {
void **ShadowHstPtrAddr = (void **)Itr->first;
// An STL map is sorted on its keys; use this property
// to quickly determine when to break out of the loop.
if ((uintptr_t)ShadowHstPtrAddr < LB) {
++Itr;
continue;
}
if ((uintptr_t)ShadowHstPtrAddr >= UB)
break;
// If we copied the struct to the host, we need to restore the pointer.
if (ArgTypes[I] & OMP_TGT_MAPTYPE_FROM) {
DP("Restoring original host pointer value " DPxMOD " for host "
"pointer " DPxMOD "\n",
DPxPTR(Itr->second.HstPtrVal), DPxPTR(ShadowHstPtrAddr));
*ShadowHstPtrAddr = Itr->second.HstPtrVal;
}
// If the struct is to be deallocated, remove the shadow entry.
if (DelEntry) {
DP("Removing shadow pointer " DPxMOD "\n", DPxPTR(ShadowHstPtrAddr));
Itr = Device.ShadowPtrMap.erase(Itr);
} else {
++Itr;
}
}
Device.ShadowMtx.unlock();
// Add pointer to the buffer for later deallocation
if (DelEntry)
DeallocTgtPtrs.emplace_back(HstPtrBegin, DataSize, HasCloseModifier);
}
}
// TODO: We should not synchronize here but pass the AsyncInfo object to the
// allocate/deallocate device APIs.
//
// We need to synchronize before deallocating data.
Ret = AsyncInfo.synchronize();
if (Ret != OFFLOAD_SUCCESS)
return OFFLOAD_FAIL;
// Deallocate target pointer
for (DeallocTgtPtrInfo &Info : DeallocTgtPtrs) {
if (FromMapperBase && FromMapperBase == Info.HstPtrBegin)
continue;
Ret = Device.deallocTgtPtr(Info.HstPtrBegin, Info.DataSize,
Info.HasCloseModifier);
if (Ret != OFFLOAD_SUCCESS) {
REPORT("Deallocating data from device failed.\n");
return OFFLOAD_FAIL;
}
}
return OFFLOAD_SUCCESS;
}
static int targetDataContiguous(ident_t *loc, DeviceTy &Device, void *ArgsBase,
void *HstPtrBegin, int64_t ArgSize,
int64_t ArgType, AsyncInfoTy &AsyncInfo) {
TIMESCOPE_WITH_IDENT(loc);
bool IsLast, IsHostPtr;
void *TgtPtrBegin = Device.getTgtPtrBegin(HstPtrBegin, ArgSize, IsLast, false,
IsHostPtr, /*MustContain=*/true);
if (!TgtPtrBegin) {
DP("hst data:" DPxMOD " not found, becomes a noop\n", DPxPTR(HstPtrBegin));
if (ArgType & OMP_TGT_MAPTYPE_PRESENT) {
MESSAGE("device mapping required by 'present' motion modifier does not "
"exist for host address " DPxMOD " (%" PRId64 " bytes)",
DPxPTR(HstPtrBegin), ArgSize);
return OFFLOAD_FAIL;
}
return OFFLOAD_SUCCESS;
}
if (PM->RTLs.RequiresFlags & OMP_REQ_UNIFIED_SHARED_MEMORY &&
TgtPtrBegin == HstPtrBegin) {
DP("hst data:" DPxMOD " unified and shared, becomes a noop\n",
DPxPTR(HstPtrBegin));
return OFFLOAD_SUCCESS;
}
if (ArgType & OMP_TGT_MAPTYPE_FROM) {
DP("Moving %" PRId64 " bytes (tgt:" DPxMOD ") -> (hst:" DPxMOD ")\n",
ArgSize, DPxPTR(TgtPtrBegin), DPxPTR(HstPtrBegin));
int Ret = Device.retrieveData(HstPtrBegin, TgtPtrBegin, ArgSize, AsyncInfo);
if (Ret != OFFLOAD_SUCCESS) {
REPORT("Copying data from device failed.\n");
return OFFLOAD_FAIL;
}
uintptr_t LB = (uintptr_t)HstPtrBegin;
uintptr_t UB = (uintptr_t)HstPtrBegin + ArgSize;
Device.ShadowMtx.lock();
for (ShadowPtrListTy::iterator IT = Device.ShadowPtrMap.begin();
IT != Device.ShadowPtrMap.end(); ++IT) {
void **ShadowHstPtrAddr = (void **)IT->first;
if ((uintptr_t)ShadowHstPtrAddr < LB)
continue;
if ((uintptr_t)ShadowHstPtrAddr >= UB)
break;
DP("Restoring original host pointer value " DPxMOD
" for host pointer " DPxMOD "\n",
DPxPTR(IT->second.HstPtrVal), DPxPTR(ShadowHstPtrAddr));
*ShadowHstPtrAddr = IT->second.HstPtrVal;
}
Device.ShadowMtx.unlock();
}
if (ArgType & OMP_TGT_MAPTYPE_TO) {
DP("Moving %" PRId64 " bytes (hst:" DPxMOD ") -> (tgt:" DPxMOD ")\n",
ArgSize, DPxPTR(HstPtrBegin), DPxPTR(TgtPtrBegin));
int Ret = Device.submitData(TgtPtrBegin, HstPtrBegin, ArgSize, AsyncInfo);
if (Ret != OFFLOAD_SUCCESS) {
REPORT("Copying data to device failed.\n");
return OFFLOAD_FAIL;
}
uintptr_t LB = (uintptr_t)HstPtrBegin;
uintptr_t UB = (uintptr_t)HstPtrBegin + ArgSize;
Device.ShadowMtx.lock();
for (ShadowPtrListTy::iterator IT = Device.ShadowPtrMap.begin();
IT != Device.ShadowPtrMap.end(); ++IT) {
void **ShadowHstPtrAddr = (void **)IT->first;
if ((uintptr_t)ShadowHstPtrAddr < LB)
continue;
if ((uintptr_t)ShadowHstPtrAddr >= UB)
break;
DP("Restoring original target pointer value " DPxMOD " for target "
"pointer " DPxMOD "\n",
DPxPTR(IT->second.TgtPtrVal), DPxPTR(IT->second.TgtPtrAddr));
Ret = Device.submitData(IT->second.TgtPtrAddr, &IT->second.TgtPtrVal,
sizeof(void *), AsyncInfo);
if (Ret != OFFLOAD_SUCCESS) {
REPORT("Copying data to device failed.\n");
Device.ShadowMtx.unlock();
return OFFLOAD_FAIL;
}
}
Device.ShadowMtx.unlock();
}
return OFFLOAD_SUCCESS;
}
static int targetDataNonContiguous(ident_t *loc, DeviceTy &Device,
void *ArgsBase,
__tgt_target_non_contig *NonContig,
uint64_t Size, int64_t ArgType,
int CurrentDim, int DimSize, uint64_t Offset,
AsyncInfoTy &AsyncInfo) {
TIMESCOPE_WITH_IDENT(loc);
int Ret = OFFLOAD_SUCCESS;
if (CurrentDim < DimSize) {
for (unsigned int I = 0; I < NonContig[CurrentDim].Count; ++I) {
uint64_t CurOffset =
(NonContig[CurrentDim].Offset + I) * NonContig[CurrentDim].Stride;
// we only need to transfer the first element for the last dimension
// since we've already got a contiguous piece.
if (CurrentDim != DimSize - 1 || I == 0) {
Ret = targetDataNonContiguous(loc, Device, ArgsBase, NonContig, Size,
ArgType, CurrentDim + 1, DimSize,
Offset + CurOffset, AsyncInfo);
// Stop the whole process if any contiguous piece returns anything
// other than OFFLOAD_SUCCESS.
if (Ret != OFFLOAD_SUCCESS)
return Ret;
}
}
} else {
char *Ptr = (char *)ArgsBase + Offset;
DP("Transfer of non-contiguous : host ptr " DPxMOD " offset %" PRIu64
" len %" PRIu64 "\n",
DPxPTR(Ptr), Offset, Size);
Ret = targetDataContiguous(loc, Device, ArgsBase, Ptr, Size, ArgType,
AsyncInfo);
}
return Ret;
}
static int getNonContigMergedDimension(__tgt_target_non_contig *NonContig,
int32_t DimSize) {
int RemovedDim = 0;
for (int I = DimSize - 1; I > 0; --I) {
if (NonContig[I].Count * NonContig[I].Stride == NonContig[I - 1].Stride)
RemovedDim++;
}
return RemovedDim;
}
/// Internal function to pass data to/from the target.
int targetDataUpdate(ident_t *loc, DeviceTy &Device, int32_t ArgNum,
void **ArgsBase, void **Args, int64_t *ArgSizes,
int64_t *ArgTypes, map_var_info_t *ArgNames,
void **ArgMappers, AsyncInfoTy &AsyncInfo, bool) {
// process each input.
for (int32_t I = 0; I < ArgNum; ++I) {
if ((ArgTypes[I] & OMP_TGT_MAPTYPE_LITERAL) ||
(ArgTypes[I] & OMP_TGT_MAPTYPE_PRIVATE))
continue;
if (ArgMappers && ArgMappers[I]) {
// Instead of executing the regular path of targetDataUpdate, call the
// targetDataMapper variant which will call targetDataUpdate again
// with new arguments.
DP("Calling targetDataMapper for the %dth argument\n", I);
map_var_info_t ArgName = (!ArgNames) ? nullptr : ArgNames[I];
int Ret = targetDataMapper(loc, Device, ArgsBase[I], Args[I], ArgSizes[I],
ArgTypes[I], ArgName, ArgMappers[I], AsyncInfo,
targetDataUpdate);
if (Ret != OFFLOAD_SUCCESS) {
REPORT("Call to targetDataUpdate via targetDataMapper for custom mapper"
" failed.\n");
return OFFLOAD_FAIL;
}
// Skip the rest of this function, continue to the next argument.
continue;
}
int Ret = OFFLOAD_SUCCESS;
if (ArgTypes[I] & OMP_TGT_MAPTYPE_NON_CONTIG) {
__tgt_target_non_contig *NonContig = (__tgt_target_non_contig *)Args[I];
int32_t DimSize = ArgSizes[I];
uint64_t Size =
NonContig[DimSize - 1].Count * NonContig[DimSize - 1].Stride;
int32_t MergedDim = getNonContigMergedDimension(NonContig, DimSize);