-
Notifications
You must be signed in to change notification settings - Fork 393
/
OMRTreeEvaluator.cpp
6779 lines (5917 loc) · 268 KB
/
OMRTreeEvaluator.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
/*******************************************************************************
* Copyright IBM Corp. and others 2000
*
* This program and the accompanying materials are made available under
* the terms of the Eclipse Public License 2.0 which accompanies this
* distribution and is available at https://www.eclipse.org/legal/epl-2.0/
* or the Apache License, Version 2.0 which accompanies this distribution
* and is available at https://www.apache.org/licenses/LICENSE-2.0.
*
* This Source Code may also be made available under the following Secondary
* Licenses when the conditions for such availability set forth in the
* Eclipse Public License, v. 2.0 are satisfied: GNU General Public License,
* version 2 with the GNU Classpath Exception [1] and GNU General Public
* License, version 2 with the OpenJDK Assembly Exception [2].
*
* [1] https://www.gnu.org/software/classpath/license.html
* [2] https://openjdk.org/legal/assembly-exception.html
*
* SPDX-License-Identifier: EPL-2.0 OR Apache-2.0 OR GPL-2.0-only WITH Classpath-exception-2.0 OR GPL-2.0-only WITH OpenJDK-assembly-exception-1.0
*******************************************************************************/
#include <algorithm>
#include <assert.h>
#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#include "codegen/CodeGenerator.hpp"
#include "env/FrontEnd.hpp"
#include "codegen/Instruction.hpp"
#include "codegen/Linkage.hpp"
#include "codegen/Linkage_inlines.hpp"
#include "codegen/LiveRegister.hpp"
#include "codegen/Machine.hpp"
#include "codegen/MemoryReference.hpp"
#include "codegen/RealRegister.hpp"
#include "codegen/RecognizedMethods.hpp"
#include "codegen/Register.hpp"
#include "codegen/RegisterConstants.hpp"
#include "codegen/RegisterDependency.hpp"
#include "codegen/RegisterPair.hpp"
#include "codegen/RegisterRematerializationInfo.hpp"
#include "codegen/TreeEvaluator.hpp"
#include "compile/Compilation.hpp"
#include "compile/ResolvedMethod.hpp"
#include "compile/SymbolReferenceTable.hpp"
#include "control/Options.hpp"
#include "control/Options_inlines.hpp"
#include "env/CompilerEnv.hpp"
#include "env/IO.hpp"
#include "env/ObjectModel.hpp"
#include "env/TRMemory.hpp"
#include "env/jittypes.h"
#include "il/AutomaticSymbol.hpp"
#include "il/Block.hpp"
#include "il/DataTypes.hpp"
#include "il/ILOpCodes.hpp"
#include "il/ILOps.hpp"
#include "il/LabelSymbol.hpp"
#include "il/MethodSymbol.hpp"
#include "il/Node.hpp"
#include "il/Node_inlines.hpp"
#include "il/Symbol.hpp"
#include "il/SymbolReference.hpp"
#include "il/TreeTop.hpp"
#include "il/TreeTop_inlines.hpp"
#include "infra/Assert.hpp"
#include "infra/Bit.hpp"
#include "infra/List.hpp"
#include "infra/CfgEdge.hpp"
#include "ras/Debug.hpp"
#include "ras/DebugCounter.hpp"
#include "runtime/Runtime.hpp"
#include "x/codegen/HelperCallSnippet.hpp"
#include "x/codegen/OutlinedInstructions.hpp"
#include "x/codegen/RegisterRematerialization.hpp"
#include "x/codegen/X86Instruction.hpp"
#include "codegen/InstOpCode.hpp"
#include "x/codegen/BinaryCommutativeAnalyser.hpp"
#include "x/codegen/SubtractAnalyser.hpp"
#include "x/codegen/X86OpcodeTable.hpp"
class TR_OpaqueClassBlock;
class TR_OpaqueMethodBlock;
extern bool existsNextInstructionToTestFlags(TR::Instruction *startInstr,
uint8_t testMask);
TR::Register *OMR::X86::TreeEvaluator::ifxcmpoEvaluator(TR::Node *node, TR::CodeGenerator *cg)
{
TR::ILOpCodes opCode = node->getOpCodeValue();
TR_ASSERT((opCode == TR::ificmno) || (opCode == TR::ificmnno) ||
(opCode == TR::iflcmno) || (opCode == TR::iflcmnno) ||
(opCode == TR::ificmpo) || (opCode == TR::ificmpno) ||
(opCode == TR::iflcmpo) || (opCode == TR::iflcmpno), "invalid opcode");
bool nodeIs64Bit = TR::TreeEvaluator::getNodeIs64Bit(node->getFirstChild(), cg);
bool reverseBranch = (opCode == TR::ificmnno) || (opCode == TR::iflcmnno) || (opCode == TR::ificmpno) || (opCode == TR::iflcmpno);
TR::Register* rs1 = cg->evaluate(node->getFirstChild());
TR::Register* rs2 = cg->evaluate(node->getSecondChild());
if ((opCode == TR::ificmno) || (opCode == TR::ificmnno) ||
(opCode == TR::iflcmno) || (opCode == TR::iflcmnno))
{
TR::Register* tmp = cg->allocateRegister();
generateRegRegInstruction(TR::InstOpCode::MOVRegReg(nodeIs64Bit), node, tmp, rs1, cg);
generateRegRegInstruction(TR::InstOpCode::ADDRegReg(nodeIs64Bit), node, tmp, rs2, cg);
cg->stopUsingRegister(tmp);
}
else
generateRegRegInstruction(TR::InstOpCode::CMPRegReg(nodeIs64Bit), node, rs1, rs2, cg);
generateConditionalJumpInstruction(reverseBranch ? TR::InstOpCode::JNO4 : TR::InstOpCode::JO4, node, cg);
cg->decReferenceCount(node->getFirstChild());
cg->decReferenceCount(node->getSecondChild());
return NULL;
}
TR::Register *OMR::X86::TreeEvaluator::intOrLongClobberEvaluate(TR::Node *node, bool nodeIs64Bit, TR::CodeGenerator *cg)
{
// TODO:AMD64: This belongs in CodeGenerator. In fact, this whole clobberEvaluate interface needs another look.
if (nodeIs64Bit)
{
TR_ASSERT(TR::TreeEvaluator::getNodeIs64Bit(node, cg), "nodeIs64Bit must be consistent with node size");
return cg->longClobberEvaluate(node);
}
else
{
TR_ASSERT(!TR::TreeEvaluator::getNodeIs64Bit(node, cg), "nodeIs64Bit must be consistent with node size");
return cg->intClobberEvaluate(node);
}
}
TR::Instruction *OMR::X86::TreeEvaluator::compareGPMemoryToImmediate(TR::Node *node,
TR::MemoryReference *mr,
int32_t value,
TR::CodeGenerator *cg)
{
// On IA32, this is called to do half of an 8-byte compare, so even though
// the node is 64 bit, we should do a 32-bit compare
bool is64Bit = cg->comp()->target().is64Bit()? TR::TreeEvaluator::getNodeIs64Bit(node->getFirstChild(), cg) : false;
TR::InstOpCode::Mnemonic cmpOp = (value >= -128 && value <= 127) ? TR::InstOpCode::CMPMemImms(is64Bit) : TR::InstOpCode::CMPMemImm4(is64Bit);
TR::Instruction *instr = generateMemImmInstruction(cmpOp, node, mr, value, cg);
cg->setImplicitExceptionPoint(instr);
return instr;
}
void OMR::X86::TreeEvaluator::compareGPRegisterToImmediate(TR::Node *node,
TR::Register *cmpRegister,
int32_t value,
TR::CodeGenerator *cg)
{
// On IA32, this is called to do half of an 8-byte compare, so even though
// the node is 64 bit, we should do a 32-bit compare
bool is64Bit = cg->comp()->target().is64Bit()? TR::TreeEvaluator::getNodeIs64Bit(node->getFirstChild(), cg) : false;
TR::InstOpCode::Mnemonic cmpOp = (value >= -128 && value <= 127) ? TR::InstOpCode::CMPRegImms(is64Bit) : TR::InstOpCode::CMPRegImm4(is64Bit);
generateRegImmInstruction(cmpOp, node, cmpRegister, value, cg);
}
void OMR::X86::TreeEvaluator::compareGPRegisterToImmediateForEquality(TR::Node *node,
TR::Register *cmpRegister,
int32_t value,
TR::CodeGenerator *cg)
{
// On IA32, this is called to do half of an 8-byte compare, so even though
// the node is 64 bit, we should do a 32-bit compare
bool is64Bit = cg->comp()->target().is64Bit()? TR::TreeEvaluator::getNodeIs64Bit(node->getFirstChild(), cg) : false;
TR::InstOpCode::Mnemonic cmpOp = (value >= -128 && value <= 127) ? TR::InstOpCode::CMPRegImms(is64Bit) : TR::InstOpCode::CMPRegImm4(is64Bit);
if (value==0)
generateRegRegInstruction(TR::InstOpCode::TESTRegReg(is64Bit), node, cmpRegister, cmpRegister, cg);
else
generateRegImmInstruction(cmpOp, node, cmpRegister, value, cg);
}
TR::Instruction *OMR::X86::TreeEvaluator::insertLoadConstant(TR::Node *node,
TR::Register *target,
intptr_t value,
TR_RematerializableTypes type,
TR::CodeGenerator *cg,
TR::Instruction *currentInstruction)
{
TR::Compilation *comp = cg->comp();
static const TR::InstOpCode::Mnemonic ops[TR_NumRematerializableTypes+1][3] =
// load 0 load -1 load c
{ { TR::InstOpCode::UD2, TR::InstOpCode::UD2, TR::InstOpCode::UD2 }, // LEA; should not seen here
{ TR::InstOpCode::XOR4RegReg, TR::InstOpCode::OR4RegImms, TR::InstOpCode::MOV4RegImm4 }, // Byte constant
{ TR::InstOpCode::XOR4RegReg, TR::InstOpCode::OR4RegImms, TR::InstOpCode::MOV4RegImm4 }, // Short constant
{ TR::InstOpCode::XOR4RegReg, TR::InstOpCode::OR4RegImms, TR::InstOpCode::MOV4RegImm4 }, // Char constant
{ TR::InstOpCode::XOR4RegReg, TR::InstOpCode::OR4RegImms, TR::InstOpCode::MOV4RegImm4 }, // Int constant
{ TR::InstOpCode::XOR4RegReg, TR::InstOpCode::OR4RegImms, TR::InstOpCode::MOV4RegImm4 }, // 32-bit address constant
{ TR::InstOpCode::XOR4RegReg, TR::InstOpCode::OR8RegImms, TR::InstOpCode::UD2 } }; // Long address constant; MOVs handled specially
enum { XOR = 0, OR = 1, MOV = 2 };
bool is64Bit = false;
int opsRow = type;
if (cg->comp()->target().is64Bit())
{
if (type == TR_RematerializableAddress)
{
// Treat 64-bit addresses as longs
opsRow++;
is64Bit = true;
}
else
{
is64Bit = (type == TR_RematerializableLong);
}
}
else
{
TR_ASSERT(type != TR_RematerializableLong, "Longs are rematerialized as pairs of ints on IA32");
}
TR_ExternalRelocationTargetKind reloKind = TR_NoRelocation;
if (cg->profiledPointersRequireRelocation() && node && node->getOpCodeValue() == TR::aconst &&
(node->isClassPointerConstant() || node->isMethodPointerConstant()))
{
if (node->isClassPointerConstant())
reloKind = TR_ClassPointer;
else if (node->isMethodPointerConstant())
reloKind = TR_MethodPointer;
else
TR_ASSERT(0, "Unexpected node, don't know how to relocate");
}
if (currentInstruction)
{
// Optimized loads inserted arbitrarily into the instruction stream must be checked
// to ensure they don't modify any eflags needed by surrounding instructions.
//
if ((value == 0 || value == -1))
{
uint8_t EFlags = TR::InstOpCode::getModifiedEFlags(ops[opsRow][((value == 0) ? XOR : OR)]);
if (existsNextInstructionToTestFlags(currentInstruction, EFlags) || cg->requiresCarry())
{
// Can't alter flags, so must use MOV. Fall through.
}
else if (value == 0)
return generateRegRegInstruction(currentInstruction, ops[opsRow][XOR], target, target, cg);
else if (value == -1)
return generateRegImmInstruction(currentInstruction, ops[opsRow][OR], target, (uint32_t)-1, cg);
}
// No luck optimizing this. Just use a MOV
//
TR::Instruction *movInstruction = NULL;
if (is64Bit)
{
if (cg->constantAddressesCanChangeSize(node) && node && node->getOpCodeValue() == TR::aconst &&
(node->isClassPointerConstant() || node->isMethodPointerConstant()))
{
movInstruction = generateRegImm64Instruction(currentInstruction, TR::InstOpCode::MOV8RegImm64, target, value, cg, reloKind);
}
else if (IS_32BIT_UNSIGNED(value))
{
// zero-extended 4-byte MOV
movInstruction = generateRegImmInstruction(currentInstruction, TR::InstOpCode::MOV4RegImm4, target, static_cast<int32_t>(value), cg, reloKind);
}
else if (IS_32BIT_SIGNED(value)) // TODO:AMD64: Is there some way we could get RIP too?
{
movInstruction = generateRegImmInstruction(currentInstruction, TR::InstOpCode::MOV8RegImm4, target, static_cast<int32_t>(value), cg, reloKind);
}
else
{
movInstruction = generateRegImm64Instruction(currentInstruction, TR::InstOpCode::MOV8RegImm64, target, value, cg, reloKind);
}
}
else
{
movInstruction = generateRegImmInstruction(currentInstruction, ops[opsRow][MOV], target, static_cast<int32_t>(value), cg, reloKind);
}
if (target && node &&
node->getOpCodeValue() == TR::aconst &&
node->isClassPointerConstant() &&
(cg->fe()->isUnloadAssumptionRequired((TR_OpaqueClassBlock *) node->getAddress(),
comp->getCurrentMethod()) ||
cg->profiledPointersRequireRelocation()))
{
comp->getStaticPICSites()->push_front(movInstruction);
}
if (target && node &&
node->getOpCodeValue() == TR::aconst &&
node->isMethodPointerConstant() &&
(cg->fe()->isUnloadAssumptionRequired(cg->fe()->createResolvedMethod(cg->trMemory(), (TR_OpaqueMethodBlock *) node->getAddress(), comp->getCurrentMethod())->classOfMethod(), comp->getCurrentMethod()) ||
cg->profiledPointersRequireRelocation()))
{
traceMsg(comp, "Adding instr %p to MethodPICSites for node %p\n", movInstruction, node);
comp->getStaticMethodPICSites()->push_front(movInstruction);
}
return movInstruction;
}
else
{
// constant loads between a compare and a branch cannot clobber the EFLAGS register
bool canClobberEFLAGS = !(cg->getCurrentEvaluationTreeTop()->getNode()->getOpCode().isIf() || cg->requiresCarry());
if (value == 0 && canClobberEFLAGS)
{
return generateRegRegInstruction(ops[opsRow][XOR], node, target, target, cg);
}
else if (value == -1 && canClobberEFLAGS)
{
return generateRegImmInstruction(ops[opsRow][OR], node, target, (uint32_t)-1, cg);
}
else
{
TR::Instruction *movInstruction = NULL;
if (is64Bit)
{
if (cg->constantAddressesCanChangeSize(node) && node && node->getOpCodeValue() == TR::aconst &&
(node->isClassPointerConstant() || node->isMethodPointerConstant()))
{
movInstruction = generateRegImm64Instruction(TR::InstOpCode::MOV8RegImm64, node, target, value, cg, reloKind);
}
else if (IS_32BIT_UNSIGNED(value))
{
// zero-extended 4-byte MOV
movInstruction = generateRegImmInstruction(TR::InstOpCode::MOV4RegImm4, node, target, static_cast<int32_t>(value), cg, reloKind);
}
else if (IS_32BIT_SIGNED(value)) // TODO:AMD64: Is there some way we could get RIP too?
{
movInstruction = generateRegImmInstruction(TR::InstOpCode::MOV8RegImm4, node, target, static_cast<int32_t>(value), cg, reloKind);
}
else
{
movInstruction = generateRegImm64Instruction(TR::InstOpCode::MOV8RegImm64, node, target, value, cg, reloKind);
}
}
else
{
movInstruction = generateRegImmInstruction(ops[opsRow][MOV], node, target, static_cast<int32_t>(value), cg, reloKind);
}
// HCR register PIC site in TR::TreeEvaluator::insertLoadConstant
TR::Symbol *symbol = NULL;
if (node && node->getOpCode().hasSymbolReference())
symbol = node->getSymbol();
bool isPICCandidate = symbol ? target && symbol->isStatic() && symbol->isClassObject() : false;
if (isPICCandidate && comp->getOption(TR_EnableHCR))
{
comp->getStaticHCRPICSites()->push_front(movInstruction);
}
if (target &&
node &&
node->getOpCodeValue() == TR::aconst &&
node->isClassPointerConstant() &&
(cg->fe()->isUnloadAssumptionRequired((TR_OpaqueClassBlock *) node->getAddress(),
comp->getCurrentMethod()) ||
cg->profiledPointersRequireRelocation()))
{
comp->getStaticPICSites()->push_front(movInstruction);
}
if (target && node &&
node->getOpCodeValue() == TR::aconst &&
node->isMethodPointerConstant() &&
(cg->fe()->isUnloadAssumptionRequired(cg->fe()->createResolvedMethod(cg->trMemory(), (TR_OpaqueMethodBlock *) node->getAddress(), comp->getCurrentMethod())->classOfMethod(), comp->getCurrentMethod()) ||
cg->profiledPointersRequireRelocation()))
{
traceMsg(comp, "Adding instr %p to MethodPICSites for node %p\n", movInstruction, node);
comp->getStaticMethodPICSites()->push_front(movInstruction);
}
return movInstruction;
}
}
}
TR::Register *OMR::X86::TreeEvaluator::loadConstant(TR::Node * node, intptr_t value, TR_RematerializableTypes type, TR::CodeGenerator *cg, TR::Register *targetRegister)
{
if (targetRegister == NULL)
{
targetRegister = cg->allocateRegister();
}
TR::Instruction *instr = TR::TreeEvaluator::insertLoadConstant(node, targetRegister, value, type, cg);
// Do not rematerialize register for class pointer or method pointer if
// it's AOT compilation because it doesn't have node info in register
// rematerialization to create relocation record for the class pointer
// or the method pointer.
if (cg->enableRematerialisation() &&
!(cg->comp()->compileRelocatableCode() && node && node->getOpCodeValue() == TR::aconst && (node->isClassPointerConstant() || node->isMethodPointerConstant())))
{
if (node && node->getOpCode().hasSymbolReference() && node->getSymbol() && node->getSymbol()->isClassObject())
(TR::Compiler->om.generateCompressedObjectHeaders() || cg->comp()->target().is32Bit()) ? type = TR_RematerializableInt : type = TR_RematerializableLong;
setDiscardableIfPossible(type, targetRegister, node, instr, value, cg);
}
return targetRegister;
}
TR::Instruction *
OMR::X86::TreeEvaluator::insertLoadMemory(
TR::Node *node,
TR::Register *target,
TR::MemoryReference *tempMR,
TR_RematerializableTypes type,
TR::CodeGenerator *cg,
TR::Instruction *currentInstruction)
{
TR::Compilation *comp = cg->comp();
static const TR::InstOpCode::Mnemonic ops[TR_NumRematerializableTypes] =
{
TR::InstOpCode::LEARegMem(), // Load Effective Address
TR::InstOpCode::L1RegMem, // Byte
TR::InstOpCode::L2RegMem, // Short
TR::InstOpCode::L2RegMem, // Char
TR::InstOpCode::L4RegMem, // Int
TR::InstOpCode::L4RegMem, // Address (64-bit addresses handled specailly below)
TR::InstOpCode::L8RegMem, // Long
};
TR::InstOpCode::Mnemonic opCode = ops[type];
if (cg->comp()->target().is64Bit())
{
if (type == TR_RematerializableAddress)
{
opCode = TR::InstOpCode::L8RegMem;
if (node && node->getOpCode().hasSymbolReference() &&
TR::Compiler->om.generateCompressedObjectHeaders())
{
if (node->getSymbol()->isClassObject() ||
(node->getSymbolReference() == comp->getSymRefTab()->findVftSymbolRef()))
opCode = TR::InstOpCode::L4RegMem;
}
}
}
else
TR_ASSERT(type != TR_RematerializableLong, "Longs are rematerialized as pairs of ints on IA32");
// If we are trying to rematerialize a byte of mem into a non-byte register,
// use movzx rather than a simple move
//
if (type == TR_RematerializableByte && target->getAssignedRealRegister())
{
TR::RealRegister::RegNum r = toRealRegister(target->getAssignedRealRegister())->getRegisterNumber();
if (r > TR::RealRegister::Last8BitGPR)
opCode = TR::InstOpCode::MOVZXReg4Mem1;
}
TR::Instruction *i;
if (currentInstruction)
{
i = generateRegMemInstruction(currentInstruction, opCode, target, tempMR, cg);
}
else
{
i = generateRegMemInstruction(opCode, node, target, tempMR, cg);
}
return i;
}
// If an unresolved data reference is followed immediately by an unconditional
// branch instruction then it is possible that the branch displacement could
// change after the unresolved data snippet has been generated (due to a forward
// reference). Insert padding such that the bytes copied from the mainline
// code do not change.
//
// The padding is always inserted for certain classes of unresolved references
// because it is difficult to accurately predict when the following instruction
// will actually be one with a forward reference. This could potentially be
// solved cleaner by running relocation processing after snippet generation.
//
// Also if the unresolved data reference is immediately followed by another patchable
// instruction (such as a virtual guard nop) with a different alignment boundary size
// then there could be an overlap between the two codes being patching, unless
// further padding is inserted.
//
void OMR::X86::TreeEvaluator::padUnresolvedDataReferences(
TR::Node *node,
TR::SymbolReference &symRef,
TR::CodeGenerator *cg)
{
TR_ASSERT(symRef.isUnresolved(), "expecting an unresolved symbol");
TR::Compilation *comp = cg->comp();
uint8_t padBytes = 0;
if (cg->comp()->target().is32Bit())
padBytes = 2; // needs at least 2 bytes as we are patching 8 bytes at a time
else
{
TR::Symbol *symbol = symRef.getSymbol();
if (symbol && symbol->isShadow())
{
padBytes = 2;
}
}
if (padBytes > 0)
{
TR::Instruction *pi = generatePaddingInstruction(padBytes, node, cg);
if (comp->getOption(TR_TraceCG))
traceMsg(comp, "adding %d pad bytes following unresolved data instruction %p\n", padBytes, pi->getPrev());
}
}
TR::Register *
OMR::X86::TreeEvaluator::loadMemory(
TR::Node *node,
TR::MemoryReference *sourceMR,
TR_RematerializableTypes type,
bool markImplicitExceptionPoint,
TR::CodeGenerator *cg)
{
TR::Register *targetRegister = cg->allocateRegister();
TR::Instruction *instr;
instr = TR::TreeEvaluator::insertLoadMemory(node, targetRegister, sourceMR, type, cg);
TR::SymbolReference& symRef = sourceMR->getSymbolReference();
if (symRef.isUnresolved())
{
TR::TreeEvaluator::padUnresolvedDataReferences(node, symRef, cg);
}
if (cg->enableRematerialisation())
{
if (node && node->getOpCode().hasSymbolReference() && node->getSymbol() && node->getSymbol()->isClassObject())
(TR::Compiler->om.generateCompressedObjectHeaders() || cg->comp()->target().is32Bit()) ? type = TR_RematerializableInt : type = TR_RematerializableLong;
setDiscardableIfPossible(type, targetRegister, node, instr, sourceMR, cg);
}
if (markImplicitExceptionPoint)
cg->setImplicitExceptionPoint(instr);
return targetRegister;
}
void OMR::X86::TreeEvaluator::removeLiveDiscardableStatics(TR::CodeGenerator *cg)
{
TR_ASSERT(cg->supportsStaticMemoryRematerialization(),
"should not be called without rematerialisation of statics\n");
TR::Compilation *comp = cg->comp();
for(auto iterator = cg->getLiveDiscardableRegisters().begin(); iterator != cg->getLiveDiscardableRegisters().end(); )
{
if ((*iterator)->getRematerializationInfo()->isRematerializableFromMemory() &&
(*iterator)->getRematerializationInfo()->getSymbolReference()->getSymbol()->isStatic())
{
TR::Register* regCursor = *iterator;
iterator = cg->getLiveDiscardableRegisters().erase(iterator);
regCursor->resetIsDiscardable();
if (debug("dumpRemat"))
{
diagnostic("\n---> Deleting static discardable candidate %s at %s instruction " POINTER_PRINTF_FORMAT,
regCursor->getRegisterName(comp),
cg->getAppendInstruction()->getOpCode().getOpCodeName(cg),
cg->getAppendInstruction());
}
}
else
++iterator;
}
}
TR::Register *OMR::X86::TreeEvaluator::performIload(TR::Node *node, TR::MemoryReference *sourceMR, TR::CodeGenerator *cg)
{
TR::Register *reg = TR::TreeEvaluator::loadMemory(node, sourceMR, TR_RematerializableInt, node->getOpCode().isIndirect(), cg);
node->setRegister(reg);
return reg;
}
// also handles iaload
TR::Register *OMR::X86::TreeEvaluator::aloadEvaluator(TR::Node *node, TR::CodeGenerator *cg)
{
TR::MemoryReference *sourceMR = generateX86MemoryReference(node, cg);
TR::Register *reg = TR::TreeEvaluator::loadMemory(node, sourceMR, TR_RematerializableAddress, node->getOpCode().isIndirect(), cg);
reg->setMemRef(sourceMR);
TR::Compilation *comp = cg->comp();
if (!node->getSymbolReference()->isUnresolved() &&
(node->getSymbolReference()->getSymbol()->getKind() == TR::Symbol::IsShadow) &&
(node->getSymbolReference()->getCPIndex() >= 0) &&
(comp->getMethodHotness()>=scorching))
{
int32_t len;
const char *fieldName = node->getSymbolReference()->getOwningMethod(comp)->fieldSignatureChars(
node->getSymbolReference()->getCPIndex(), len);
if (fieldName && strstr(fieldName, "Ljava/lang/String;"))
{
generateMemInstruction(TR::InstOpCode::PREFETCHT0, node, generateX86MemoryReference(reg, 0, cg), cg);
}
}
#ifdef J9_PROJECT_SPECIFIC
if (node->getSymbolReference() == comp->getSymRefTab()->findVftSymbolRef())
TR::TreeEvaluator::generateVFTMaskInstruction(node, reg, cg);
#endif
if (node->getSymbolReference()->getSymbol()->isNotCollected())
{
if (node->getSymbolReference()->getSymbol()->isInternalPointer())
{
reg->setContainsInternalPointer();
reg->setPinningArrayPointer(node->getSymbolReference()->getSymbol()->castToInternalPointerAutoSymbol()->getPinningArrayPointer());
}
}
else
{
if (node->getSymbolReference()->getSymbol()->isInternalPointer())
{
reg->setContainsInternalPointer();
reg->setPinningArrayPointer(node->getSymbolReference()->getSymbol()->castToInternalPointerAutoSymbol()->getPinningArrayPointer());
}
else
{
reg->setContainsCollectedReference();
}
}
node->setRegister(reg);
sourceMR->decNodeReferenceCounts(cg);
return reg;
}
// generate the null test instructions if required
//
bool OMR::X86::TreeEvaluator::genNullTestSequence(TR::Node *node,
TR::Register *opReg,
TR::Register *targetReg,
TR::CodeGenerator *cg)
{
TR::Compilation *comp = cg->comp();
if (comp->useCompressedPointers() &&
node->containsCompressionSequence())
{
TR::Node *n = node;
bool isNonZero = false;
if (n->isNonZero())
isNonZero = true;
if (n->getOpCodeValue() == TR::ladd)
{
if (n->getFirstChild()->isNonZero())
isNonZero = true;
if (n->getFirstChild()->getOpCodeValue() == TR::iu2l ||
n->getFirstChild()->getOpCode().isShift())
{
if (n->getFirstChild()->getFirstChild()->isNonZero())
isNonZero = true;
}
}
if (!isNonZero)
{
if (opReg != targetReg)
generateRegRegInstruction(TR::InstOpCode::MOVRegReg(), node, targetReg, opReg, cg);
TR::Register *tempReg = cg->allocateRegister();
// addition of the negative number should result in 0
//
generateRegImm64Instruction(TR::InstOpCode::MOV8RegImm64, node, tempReg, 0, cg);
if (n->getFirstChild()->getOpCode().isShift() && n->getFirstChild()->getFirstChild()->getRegister())
{
TR::Register * r = n->getFirstChild()->getFirstChild()->getRegister();
generateRegRegInstruction(TR::InstOpCode::TESTRegReg(), node, r, r, cg);
}
else
{
generateRegRegInstruction(TR::InstOpCode::TESTRegReg(), node, opReg, opReg, cg);
}
generateRegRegInstruction(TR::InstOpCode::CMOVERegReg(), node, targetReg, tempReg, cg);
cg->stopUsingRegister(tempReg);
return true;
}
}
return false;
}
// also handles iiload
TR::Register *OMR::X86::TreeEvaluator::iloadEvaluator(TR::Node *node, TR::CodeGenerator *cg)
{
TR::MemoryReference *sourceMR = generateX86MemoryReference(node, cg);
TR::Register *reg = TR::TreeEvaluator::performIload(node, sourceMR, cg);
reg->setMemRef(sourceMR);
sourceMR->decNodeReferenceCounts(cg);
TR::Compilation *comp = cg->comp();
if (comp->useCompressedPointers() &&
(node->getOpCode().hasSymbolReference() &&
node->getSymbolReference()->getSymbol()->getDataType() == TR::Address))
{
if (!node->getSymbolReference()->isUnresolved() &&
(node->getSymbolReference()->getSymbol()->getKind() == TR::Symbol::IsShadow) &&
(node->getSymbolReference()->getCPIndex() >= 0) &&
(comp->getMethodHotness()>=scorching))
{
int32_t len;
const char *fieldName = node->getSymbolReference()->getOwningMethod(comp)->fieldSignatureChars(
node->getSymbolReference()->getCPIndex(), len);
if (fieldName && strstr(fieldName, "Ljava/lang/String;"))
{
generateMemInstruction(TR::InstOpCode::PREFETCHT0, node, generateX86MemoryReference(reg, 0, cg), cg);
}
}
}
return reg;
}
// also handles ibload
TR::Register *OMR::X86::TreeEvaluator::bloadEvaluator(TR::Node *node, TR::CodeGenerator *cg)
{
TR::MemoryReference *sourceMR = generateX86MemoryReference(node, cg);
TR::Register *reg = TR::TreeEvaluator::loadMemory(node, sourceMR, TR_RematerializableByte, node->getOpCode().isIndirect(), cg);
reg->setMemRef(sourceMR);
node->setRegister(reg);
if (cg->enableRegisterInterferences())
cg->getLiveRegisters(TR_GPR)->setByteRegisterAssociation(reg);
sourceMR->decNodeReferenceCounts(cg);
return reg;
}
// also handles isload
TR::Register *OMR::X86::TreeEvaluator::sloadEvaluator(TR::Node *node, TR::CodeGenerator *cg)
{
TR::MemoryReference *sourceMR = generateX86MemoryReference(node, cg);
TR::Register *reg = TR::TreeEvaluator::loadMemory(node, sourceMR, TR_RematerializableShort, node->getOpCode().isIndirect(), cg);
reg->setMemRef(sourceMR);
node->setRegister(reg);
sourceMR->decNodeReferenceCounts(cg);
return reg;
}
TR::Register *
OMR::X86::TreeEvaluator::fwrtbarEvaluator(TR::Node *node, TR::CodeGenerator *cg)
{
// The wrtbar IL op represents a store with side effects.
// Currently we don't use the side effect node. So just evaluate it and decrement the reference count.
TR::Node *sideEffectNode = node->getSecondChild();
cg->evaluate(sideEffectNode);
cg->decReferenceCount(sideEffectNode);
// Delegate the evaluation of the remaining children and the store operation to the storeEvaluator.
return TR::TreeEvaluator::floatingPointStoreEvaluator(node, cg);
}
TR::Register *
OMR::X86::TreeEvaluator::fwrtbariEvaluator(TR::Node *node, TR::CodeGenerator *cg)
{
// The wrtbar IL op represents a store with side effects.
// Currently we don't use the side effect node. So just evaluate it and decrement the reference count.
TR::Node *sideEffectNode = node->getThirdChild();
cg->evaluate(sideEffectNode);
cg->decReferenceCount(sideEffectNode);
// Delegate the evaluation of the remaining children and the store operation to the storeEvaluator.
return TR::TreeEvaluator::floatingPointStoreEvaluator(node, cg);
}
// iiload handled by iloadEvaluator
// ilload handled by lloadEvaluator
// ialoadEvaluator handled by iloadEvaluator
// ibloadEvaluator handled by bloadEvaluator
// isloadEvaluator handled by sloadEvaluator
// also used for iistore, astore and iastore
TR::Register *OMR::X86::TreeEvaluator::integerStoreEvaluator(TR::Node *node, TR::CodeGenerator *cg)
{
TR::Node *valueChild = NULL;
TR::Register *valueReg = NULL;
bool genLockedOpOutOfline = true;
TR::Compilation *comp = cg->comp();
bool usingCompressedPointers = false;
TR::Node *origRef = NULL; // input to compression sequence (when there is one)
bool isCompressedNullStore = false;
bool isCompressedWithoutShift = false;
node->getFirstChild()->oneParentSupportsLazyClobber(comp);
if (node->getOpCode().isIndirect())
{
valueChild = node->getSecondChild();
valueChild->oneParentSupportsLazyClobber(comp);
if (comp->useCompressedPointers() &&
(node->getSymbolReference()->getSymbol()->getDataType() == TR::Address))
{
if (valueChild->getOpCodeValue() == TR::l2i)
{
TR::Node *inner = valueChild->getChild(0);
bool haveShift = inner->getOpCode().isRightShift();
if (haveShift)
inner = inner->getChild(0);
if (inner->getOpCodeValue() == TR::a2l)
{
usingCompressedPointers = true;
origRef = inner->getChild(0);
isCompressedNullStore =
valueChild->isZero() || inner->isZero() || origRef->isNull();
// Ignore the shift (if any) if the value is null anyway.
isCompressedWithoutShift = !haveShift || isCompressedNullStore;
}
}
}
}
else
valueChild = node->getFirstChild();
int32_t size = node->getOpCode().getSize();
TR::MemoryReference *tempMR = NULL;
TR::Instruction *instr = NULL;
TR::InstOpCode::Mnemonic opCode;
TR::Node *originalValueChild = valueChild;
bool childIsConstant = false;
int64_t childConstValue = TR::getMinSigned<TR::Int64>();
if (valueChild->getOpCode().isLoadConst() && valueChild->getRegister() == NULL)
{
childIsConstant = true;
childConstValue = valueChild->getConstValue();
}
else if (isCompressedNullStore && origRef->getRegister() == NULL)
{
childIsConstant = true;
childConstValue = 0;
}
if (childIsConstant && (size <= 4 || IS_32BIT_SIGNED(childConstValue)))
{
cg->recursivelyDecReferenceCount(valueChild); // skip evaluation of entire subtree
TR::LabelSymbol * dstStored = NULL;
TR::RegisterDependencyConditions *deps = NULL;
// Note that we can truncate to 32-bit here for all sizes, since only the
// low order "size" bytes of the int will be used by the instruction,
// and longs only get here if the constant fits in 32 bits.
//
int32_t konst = (int32_t)childConstValue;
tempMR = generateX86MemoryReference(node, cg);
if (size == 1)
opCode = TR::InstOpCode::S1MemImm1;
else if (size == 2)
opCode = TR::InstOpCode::S2MemImm2;
else if (size == 4)
opCode = TR::InstOpCode::S4MemImm4;
else
opCode = TR::InstOpCode::S8MemImm4;
instr = generateMemImmInstruction(opCode, node, tempMR, konst, cg);
}
else
{
// See if this can be a direct memory update operation.
// If so, mark the value child so that when it is evaluated it can generate
// the direct memory update.
//
if (!usingCompressedPointers && cg->isMemoryUpdate(node))
{
if (valueChild->getFirstChild()->getReferenceCount() == 1)
{
// Memory update always a win if the result value is not needed again
//
valueChild->setDirectMemoryUpdate(true);
}
else
{
int32_t numRegs = cg->getLiveRegisters(TR_GPR)->getNumberOfLiveRegisters();
if (numRegs >= TR::RealRegister::LastAssignableGPR - 2) // -1 for VM thread reg, -1 fudge
valueChild->setDirectMemoryUpdate(true);
}
}
TR::Register *translatedReg = valueReg;
// try to avoid unnecessary sign-extension (or reg-reg moves in the case
// of a compression sequence)
if (valueChild->getRegister() == 0 &&
valueChild->getReferenceCount() == 1 &&
(valueChild->getOpCodeValue() == TR::l2i ||
valueChild->getOpCodeValue() == TR::l2s ||
valueChild->getOpCodeValue() == TR::l2b))
{
valueChild = isCompressedWithoutShift ? origRef : valueChild->getChild(0);
// Skipping evaluation of originalValueChild, but not valueChild.
cg->incReferenceCount(valueChild);
cg->recursivelyDecReferenceCount(originalValueChild);
if (cg->comp()->target().is64Bit())
translatedReg = cg->evaluate(valueChild);
else
translatedReg = cg->evaluate(valueChild)->getLowOrder();
}
else
{
translatedReg = cg->evaluate(valueChild);
}
valueReg = translatedReg;
// If the evaluation of the child resulted in a NULL register value, the
// store has already been done by the child
//
if (valueReg)
{
if (size == 1)
opCode = TR::InstOpCode::S1MemReg;
else if (size == 2)
opCode = TR::InstOpCode::S2MemReg;
else if (size == 4)
opCode = TR::InstOpCode::S4MemReg;
else
opCode = TR::InstOpCode::S8MemReg;
if (TR::Compiler->om.generateCompressedObjectHeaders() &&
(node->getSymbol()->isClassObject() ||
(node->getSymbolReference() == comp->getSymRefTab()->findVftSymbolRef())))
opCode = TR::InstOpCode::S4MemReg;
tempMR = generateX86MemoryReference(node, cg);
// in comp->useCompressedPointers we should write 4 bytes
// since the iastore has been changed to an iistore, size will be 4
//
instr = generateMemRegInstruction(opCode, node, tempMR, valueReg, cg);
TR::SymbolReference& symRef = tempMR->getSymbolReference();
if (symRef.isUnresolved())
{
TR::TreeEvaluator::padUnresolvedDataReferences(node, symRef, cg);
}
// Make the register being stored rematerializable from the destination of
// the store, if possible. Only do this if the register is not already
// discardable, otherwise the original rematerialization information will
// be lost.
//
if (cg->enableRematerialisation() && !valueReg->getRematerializationInfo())
{
// WARNING: if the store truncates/modifies the data associated with
// a virtual register then there will not be enough information to
// rematerialise that virtual register
//
if (originalValueChild == valueChild)
{
TR_RematerializableTypes type;
// This is why we should use TR::DataType in place of TR_RematerializableTypes...
//
switch (node->getDataType())
{
case TR::Int8:
type = TR_RematerializableByte;
break;
case TR::Int16:
type = TR_RematerializableShort;
break;
case TR::Int32:
type = TR_RematerializableInt;
break;
case TR::Int64:
type = TR_RematerializableLong;
break;
case TR::Address:
{
if (node && node->getOpCode().hasSymbolReference() && node->getSymbol() && node->getSymbol()->isClassObject())
(TR::Compiler->om.generateCompressedObjectHeaders() || cg->comp()->target().is32Bit()) ? type = TR_RematerializableInt : type = TR_RematerializableLong;
else
type = TR_RematerializableAddress;
break;
}
default:
TR_ASSERT(0, "Unsupported dataType for integerStoreEvaluator");
break;
}
setDiscardableIfPossible(type, valueReg, node, instr, tempMR, cg);
}
}
}