/
macroAssembler_aarch64.hpp
1470 lines (1202 loc) · 56.3 KB
/
macroAssembler_aarch64.hpp
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 (c) 1997, 2022, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2014, 2021, Red Hat Inc. All rights reserved.
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
*
* This code is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License version 2 only, as
* published by the Free Software Foundation.
*
* This code is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
* version 2 for more details (a copy is included in the LICENSE file that
* accompanied this code).
*
* You should have received a copy of the GNU General Public License version
* 2 along with this work; if not, write to the Free Software Foundation,
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
*
* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
* or visit www.oracle.com if you need additional information or have any
* questions.
*
*/
#ifndef CPU_AARCH64_MACROASSEMBLER_AARCH64_HPP
#define CPU_AARCH64_MACROASSEMBLER_AARCH64_HPP
#include "asm/assembler.inline.hpp"
#include "metaprogramming/enableIf.hpp"
#include "oops/compressedOops.hpp"
#include "runtime/vm_version.hpp"
#include "utilities/powerOfTwo.hpp"
// MacroAssembler extends Assembler by frequently used macros.
//
// Instructions for which a 'better' code sequence exists depending
// on arguments should also go in here.
class MacroAssembler: public Assembler {
friend class LIR_Assembler;
public:
using Assembler::mov;
using Assembler::movi;
protected:
// Support for VM calls
//
// This is the base routine called by the different versions of call_VM_leaf. The interpreter
// may customize this version by overriding it for its purposes (e.g., to save/restore
// additional registers when doing a VM call).
virtual void call_VM_leaf_base(
address entry_point, // the entry point
int number_of_arguments, // the number of arguments to pop after the call
Label *retaddr = NULL
);
virtual void call_VM_leaf_base(
address entry_point, // the entry point
int number_of_arguments, // the number of arguments to pop after the call
Label &retaddr) {
call_VM_leaf_base(entry_point, number_of_arguments, &retaddr);
}
// This is the base routine called by the different versions of call_VM. The interpreter
// may customize this version by overriding it for its purposes (e.g., to save/restore
// additional registers when doing a VM call).
//
// If no java_thread register is specified (noreg) than rthread will be used instead. call_VM_base
// returns the register which contains the thread upon return. If a thread register has been
// specified, the return value will correspond to that register. If no last_java_sp is specified
// (noreg) than rsp will be used instead.
virtual void call_VM_base( // returns the register containing the thread upon return
Register oop_result, // where an oop-result ends up if any; use noreg otherwise
Register java_thread, // the thread if computed before ; use noreg otherwise
Register last_java_sp, // to set up last_Java_frame in stubs; use noreg otherwise
address entry_point, // the entry point
int number_of_arguments, // the number of arguments (w/o thread) to pop after the call
bool check_exceptions // whether to check for pending exceptions after return
);
void call_VM_helper(Register oop_result, address entry_point, int number_of_arguments, bool check_exceptions = true);
enum KlassDecodeMode {
KlassDecodeNone,
KlassDecodeZero,
KlassDecodeXor,
KlassDecodeMovk
};
KlassDecodeMode klass_decode_mode();
private:
static KlassDecodeMode _klass_decode_mode;
public:
MacroAssembler(CodeBuffer* code) : Assembler(code) {}
// These routines should emit JVMTI PopFrame and ForceEarlyReturn handling code.
// The implementation is only non-empty for the InterpreterMacroAssembler,
// as only the interpreter handles PopFrame and ForceEarlyReturn requests.
virtual void check_and_handle_popframe(Register java_thread);
virtual void check_and_handle_earlyret(Register java_thread);
void safepoint_poll(Label& slow_path, bool at_return, bool acquire, bool in_nmethod);
// Helper functions for statistics gathering.
// Unconditional atomic increment.
void atomic_incw(Register counter_addr, Register tmp, Register tmp2);
void atomic_incw(Address counter_addr, Register tmp1, Register tmp2, Register tmp3) {
lea(tmp1, counter_addr);
atomic_incw(tmp1, tmp2, tmp3);
}
// Load Effective Address
void lea(Register r, const Address &a) {
InstructionMark im(this);
code_section()->relocate(inst_mark(), a.rspec());
a.lea(this, r);
}
/* Sometimes we get misaligned loads and stores, usually from Unsafe
accesses, and these can exceed the offset range. */
Address legitimize_address(const Address &a, int size, Register scratch) {
if (a.getMode() == Address::base_plus_offset) {
if (! Address::offset_ok_for_immed(a.offset(), exact_log2(size))) {
block_comment("legitimize_address {");
lea(scratch, a);
block_comment("} legitimize_address");
return Address(scratch);
}
}
return a;
}
void addmw(Address a, Register incr, Register scratch) {
ldrw(scratch, a);
addw(scratch, scratch, incr);
strw(scratch, a);
}
// Add constant to memory word
void addmw(Address a, int imm, Register scratch) {
ldrw(scratch, a);
if (imm > 0)
addw(scratch, scratch, (unsigned)imm);
else
subw(scratch, scratch, (unsigned)-imm);
strw(scratch, a);
}
void bind(Label& L) {
Assembler::bind(L);
code()->clear_last_insn();
}
void membar(Membar_mask_bits order_constraint);
using Assembler::ldr;
using Assembler::str;
using Assembler::ldrw;
using Assembler::strw;
void ldr(Register Rx, const Address &adr);
void ldrw(Register Rw, const Address &adr);
void str(Register Rx, const Address &adr);
void strw(Register Rx, const Address &adr);
// Frame creation and destruction shared between JITs.
void build_frame(int framesize);
void remove_frame(int framesize);
virtual void _call_Unimplemented(address call_site) {
mov(rscratch2, call_site);
}
// Microsoft's MSVC team thinks that the __FUNCSIG__ is approximately (sympathy for calling conventions) equivalent to __PRETTY_FUNCTION__
// Also, from Clang patch: "It is very similar to GCC's PRETTY_FUNCTION, except it prints the calling convention."
// https://reviews.llvm.org/D3311
#ifdef _WIN64
#define call_Unimplemented() _call_Unimplemented((address)__FUNCSIG__)
#else
#define call_Unimplemented() _call_Unimplemented((address)__PRETTY_FUNCTION__)
#endif
// aliases defined in AARCH64 spec
template<class T>
inline void cmpw(Register Rd, T imm) { subsw(zr, Rd, imm); }
inline void cmp(Register Rd, unsigned char imm8) { subs(zr, Rd, imm8); }
inline void cmp(Register Rd, unsigned imm) = delete;
inline void cmnw(Register Rd, unsigned imm) { addsw(zr, Rd, imm); }
inline void cmn(Register Rd, unsigned imm) { adds(zr, Rd, imm); }
void cset(Register Rd, Assembler::Condition cond) {
csinc(Rd, zr, zr, ~cond);
}
void csetw(Register Rd, Assembler::Condition cond) {
csincw(Rd, zr, zr, ~cond);
}
void cneg(Register Rd, Register Rn, Assembler::Condition cond) {
csneg(Rd, Rn, Rn, ~cond);
}
void cnegw(Register Rd, Register Rn, Assembler::Condition cond) {
csnegw(Rd, Rn, Rn, ~cond);
}
inline void movw(Register Rd, Register Rn) {
if (Rd == sp || Rn == sp) {
addw(Rd, Rn, 0U);
} else {
orrw(Rd, zr, Rn);
}
}
inline void mov(Register Rd, Register Rn) {
assert(Rd != r31_sp && Rn != r31_sp, "should be");
if (Rd == Rn) {
} else if (Rd == sp || Rn == sp) {
add(Rd, Rn, 0U);
} else {
orr(Rd, zr, Rn);
}
}
inline void moviw(Register Rd, unsigned imm) { orrw(Rd, zr, imm); }
inline void movi(Register Rd, unsigned imm) { orr(Rd, zr, imm); }
inline void tstw(Register Rd, Register Rn) { andsw(zr, Rd, Rn); }
inline void tst(Register Rd, Register Rn) { ands(zr, Rd, Rn); }
inline void tstw(Register Rd, uint64_t imm) { andsw(zr, Rd, imm); }
inline void tst(Register Rd, uint64_t imm) { ands(zr, Rd, imm); }
inline void bfiw(Register Rd, Register Rn, unsigned lsb, unsigned width) {
bfmw(Rd, Rn, ((32 - lsb) & 31), (width - 1));
}
inline void bfi(Register Rd, Register Rn, unsigned lsb, unsigned width) {
bfm(Rd, Rn, ((64 - lsb) & 63), (width - 1));
}
inline void bfxilw(Register Rd, Register Rn, unsigned lsb, unsigned width) {
bfmw(Rd, Rn, lsb, (lsb + width - 1));
}
inline void bfxil(Register Rd, Register Rn, unsigned lsb, unsigned width) {
bfm(Rd, Rn, lsb , (lsb + width - 1));
}
inline void sbfizw(Register Rd, Register Rn, unsigned lsb, unsigned width) {
sbfmw(Rd, Rn, ((32 - lsb) & 31), (width - 1));
}
inline void sbfiz(Register Rd, Register Rn, unsigned lsb, unsigned width) {
sbfm(Rd, Rn, ((64 - lsb) & 63), (width - 1));
}
inline void sbfxw(Register Rd, Register Rn, unsigned lsb, unsigned width) {
sbfmw(Rd, Rn, lsb, (lsb + width - 1));
}
inline void sbfx(Register Rd, Register Rn, unsigned lsb, unsigned width) {
sbfm(Rd, Rn, lsb , (lsb + width - 1));
}
inline void ubfizw(Register Rd, Register Rn, unsigned lsb, unsigned width) {
ubfmw(Rd, Rn, ((32 - lsb) & 31), (width - 1));
}
inline void ubfiz(Register Rd, Register Rn, unsigned lsb, unsigned width) {
ubfm(Rd, Rn, ((64 - lsb) & 63), (width - 1));
}
inline void ubfxw(Register Rd, Register Rn, unsigned lsb, unsigned width) {
ubfmw(Rd, Rn, lsb, (lsb + width - 1));
}
inline void ubfx(Register Rd, Register Rn, unsigned lsb, unsigned width) {
ubfm(Rd, Rn, lsb , (lsb + width - 1));
}
inline void asrw(Register Rd, Register Rn, unsigned imm) {
sbfmw(Rd, Rn, imm, 31);
}
inline void asr(Register Rd, Register Rn, unsigned imm) {
sbfm(Rd, Rn, imm, 63);
}
inline void lslw(Register Rd, Register Rn, unsigned imm) {
ubfmw(Rd, Rn, ((32 - imm) & 31), (31 - imm));
}
inline void lsl(Register Rd, Register Rn, unsigned imm) {
ubfm(Rd, Rn, ((64 - imm) & 63), (63 - imm));
}
inline void lsrw(Register Rd, Register Rn, unsigned imm) {
ubfmw(Rd, Rn, imm, 31);
}
inline void lsr(Register Rd, Register Rn, unsigned imm) {
ubfm(Rd, Rn, imm, 63);
}
inline void rorw(Register Rd, Register Rn, unsigned imm) {
extrw(Rd, Rn, Rn, imm);
}
inline void ror(Register Rd, Register Rn, unsigned imm) {
extr(Rd, Rn, Rn, imm);
}
inline void sxtbw(Register Rd, Register Rn) {
sbfmw(Rd, Rn, 0, 7);
}
inline void sxthw(Register Rd, Register Rn) {
sbfmw(Rd, Rn, 0, 15);
}
inline void sxtb(Register Rd, Register Rn) {
sbfm(Rd, Rn, 0, 7);
}
inline void sxth(Register Rd, Register Rn) {
sbfm(Rd, Rn, 0, 15);
}
inline void sxtw(Register Rd, Register Rn) {
sbfm(Rd, Rn, 0, 31);
}
inline void uxtbw(Register Rd, Register Rn) {
ubfmw(Rd, Rn, 0, 7);
}
inline void uxthw(Register Rd, Register Rn) {
ubfmw(Rd, Rn, 0, 15);
}
inline void uxtb(Register Rd, Register Rn) {
ubfm(Rd, Rn, 0, 7);
}
inline void uxth(Register Rd, Register Rn) {
ubfm(Rd, Rn, 0, 15);
}
inline void uxtw(Register Rd, Register Rn) {
ubfm(Rd, Rn, 0, 31);
}
inline void cmnw(Register Rn, Register Rm) {
addsw(zr, Rn, Rm);
}
inline void cmn(Register Rn, Register Rm) {
adds(zr, Rn, Rm);
}
inline void cmpw(Register Rn, Register Rm) {
subsw(zr, Rn, Rm);
}
inline void cmp(Register Rn, Register Rm) {
subs(zr, Rn, Rm);
}
inline void negw(Register Rd, Register Rn) {
subw(Rd, zr, Rn);
}
inline void neg(Register Rd, Register Rn) {
sub(Rd, zr, Rn);
}
inline void negsw(Register Rd, Register Rn) {
subsw(Rd, zr, Rn);
}
inline void negs(Register Rd, Register Rn) {
subs(Rd, zr, Rn);
}
inline void cmnw(Register Rn, Register Rm, enum shift_kind kind, unsigned shift = 0) {
addsw(zr, Rn, Rm, kind, shift);
}
inline void cmn(Register Rn, Register Rm, enum shift_kind kind, unsigned shift = 0) {
adds(zr, Rn, Rm, kind, shift);
}
inline void cmpw(Register Rn, Register Rm, enum shift_kind kind, unsigned shift = 0) {
subsw(zr, Rn, Rm, kind, shift);
}
inline void cmp(Register Rn, Register Rm, enum shift_kind kind, unsigned shift = 0) {
subs(zr, Rn, Rm, kind, shift);
}
inline void negw(Register Rd, Register Rn, enum shift_kind kind, unsigned shift = 0) {
subw(Rd, zr, Rn, kind, shift);
}
inline void neg(Register Rd, Register Rn, enum shift_kind kind, unsigned shift = 0) {
sub(Rd, zr, Rn, kind, shift);
}
inline void negsw(Register Rd, Register Rn, enum shift_kind kind, unsigned shift = 0) {
subsw(Rd, zr, Rn, kind, shift);
}
inline void negs(Register Rd, Register Rn, enum shift_kind kind, unsigned shift = 0) {
subs(Rd, zr, Rn, kind, shift);
}
inline void mnegw(Register Rd, Register Rn, Register Rm) {
msubw(Rd, Rn, Rm, zr);
}
inline void mneg(Register Rd, Register Rn, Register Rm) {
msub(Rd, Rn, Rm, zr);
}
inline void mulw(Register Rd, Register Rn, Register Rm) {
maddw(Rd, Rn, Rm, zr);
}
inline void mul(Register Rd, Register Rn, Register Rm) {
madd(Rd, Rn, Rm, zr);
}
inline void smnegl(Register Rd, Register Rn, Register Rm) {
smsubl(Rd, Rn, Rm, zr);
}
inline void smull(Register Rd, Register Rn, Register Rm) {
smaddl(Rd, Rn, Rm, zr);
}
inline void umnegl(Register Rd, Register Rn, Register Rm) {
umsubl(Rd, Rn, Rm, zr);
}
inline void umull(Register Rd, Register Rn, Register Rm) {
umaddl(Rd, Rn, Rm, zr);
}
#define WRAP(INSN) \
void INSN(Register Rd, Register Rn, Register Rm, Register Ra) { \
if ((VM_Version::features() & VM_Version::CPU_A53MAC) && Ra != zr) \
nop(); \
Assembler::INSN(Rd, Rn, Rm, Ra); \
}
WRAP(madd) WRAP(msub) WRAP(maddw) WRAP(msubw)
WRAP(smaddl) WRAP(smsubl) WRAP(umaddl) WRAP(umsubl)
#undef WRAP
// macro assembly operations needed for aarch64
// first two private routines for loading 32 bit or 64 bit constants
private:
void mov_immediate64(Register dst, uint64_t imm64);
void mov_immediate32(Register dst, uint32_t imm32);
int push(unsigned int bitset, Register stack);
int pop(unsigned int bitset, Register stack);
int push_fp(unsigned int bitset, Register stack);
int pop_fp(unsigned int bitset, Register stack);
int push_p(unsigned int bitset, Register stack);
int pop_p(unsigned int bitset, Register stack);
void mov(Register dst, Address a);
public:
void push(RegSet regs, Register stack) { if (regs.bits()) push(regs.bits(), stack); }
void pop(RegSet regs, Register stack) { if (regs.bits()) pop(regs.bits(), stack); }
void push_fp(FloatRegSet regs, Register stack) { if (regs.bits()) push_fp(regs.bits(), stack); }
void pop_fp(FloatRegSet regs, Register stack) { if (regs.bits()) pop_fp(regs.bits(), stack); }
static RegSet call_clobbered_gp_registers();
void push_p(PRegSet regs, Register stack) { if (regs.bits()) push_p(regs.bits(), stack); }
void pop_p(PRegSet regs, Register stack) { if (regs.bits()) pop_p(regs.bits(), stack); }
// Push and pop everything that might be clobbered by a native
// runtime call except rscratch1 and rscratch2. (They are always
// scratch, so we don't have to protect them.) Only save the lower
// 64 bits of each vector register. Additonal registers can be excluded
// in a passed RegSet.
void push_call_clobbered_registers_except(RegSet exclude);
void pop_call_clobbered_registers_except(RegSet exclude);
void push_call_clobbered_registers() {
push_call_clobbered_registers_except(RegSet());
}
void pop_call_clobbered_registers() {
pop_call_clobbered_registers_except(RegSet());
}
// now mov instructions for loading absolute addresses and 32 or
// 64 bit integers
inline void mov(Register dst, address addr) { mov_immediate64(dst, (uint64_t)addr); }
template<typename T, ENABLE_IF(std::is_integral<T>::value)>
inline void mov(Register dst, T o) { mov_immediate64(dst, (uint64_t)o); }
inline void movw(Register dst, uint32_t imm32) { mov_immediate32(dst, imm32); }
void mov(Register dst, RegisterOrConstant src) {
if (src.is_register())
mov(dst, src.as_register());
else
mov(dst, src.as_constant());
}
void movptr(Register r, uintptr_t imm64);
void mov(FloatRegister Vd, SIMD_Arrangement T, uint32_t imm32);
void mov(FloatRegister Vd, SIMD_Arrangement T, FloatRegister Vn) {
orr(Vd, T, Vn, Vn);
}
public:
// Generalized Test Bit And Branch, including a "far" variety which
// spans more than 32KiB.
void tbr(Condition cond, Register Rt, int bitpos, Label &dest, bool isfar = false) {
assert(cond == EQ || cond == NE, "must be");
if (isfar)
cond = ~cond;
void (Assembler::* branch)(Register Rt, int bitpos, Label &L);
if (cond == Assembler::EQ)
branch = &Assembler::tbz;
else
branch = &Assembler::tbnz;
if (isfar) {
Label L;
(this->*branch)(Rt, bitpos, L);
b(dest);
bind(L);
} else {
(this->*branch)(Rt, bitpos, dest);
}
}
// macro instructions for accessing and updating floating point
// status register
//
// FPSR : op1 == 011
// CRn == 0100
// CRm == 0100
// op2 == 001
inline void get_fpsr(Register reg)
{
mrs(0b11, 0b0100, 0b0100, 0b001, reg);
}
inline void set_fpsr(Register reg)
{
msr(0b011, 0b0100, 0b0100, 0b001, reg);
}
inline void clear_fpsr()
{
msr(0b011, 0b0100, 0b0100, 0b001, zr);
}
// DCZID_EL0: op1 == 011
// CRn == 0000
// CRm == 0000
// op2 == 111
inline void get_dczid_el0(Register reg)
{
mrs(0b011, 0b0000, 0b0000, 0b111, reg);
}
// CTR_EL0: op1 == 011
// CRn == 0000
// CRm == 0000
// op2 == 001
inline void get_ctr_el0(Register reg)
{
mrs(0b011, 0b0000, 0b0000, 0b001, reg);
}
// idiv variant which deals with MINLONG as dividend and -1 as divisor
int corrected_idivl(Register result, Register ra, Register rb,
bool want_remainder, Register tmp = rscratch1);
int corrected_idivq(Register result, Register ra, Register rb,
bool want_remainder, Register tmp = rscratch1);
// Support for NULL-checks
//
// Generates code that causes a NULL OS exception if the content of reg is NULL.
// If the accessed location is M[reg + offset] and the offset is known, provide the
// offset. No explicit code generation is needed if the offset is within a certain
// range (0 <= offset <= page_size).
virtual void null_check(Register reg, int offset = -1);
static bool needs_explicit_null_check(intptr_t offset);
static bool uses_implicit_null_check(void* address);
static address target_addr_for_insn(address insn_addr, unsigned insn);
static address target_addr_for_insn_or_null(address insn_addr, unsigned insn);
static address target_addr_for_insn(address insn_addr) {
unsigned insn = *(unsigned*)insn_addr;
return target_addr_for_insn(insn_addr, insn);
}
static address target_addr_for_insn_or_null(address insn_addr) {
unsigned insn = *(unsigned*)insn_addr;
return target_addr_for_insn_or_null(insn_addr, insn);
}
// Required platform-specific helpers for Label::patch_instructions.
// They _shadow_ the declarations in AbstractAssembler, which are undefined.
static int pd_patch_instruction_size(address branch, address target);
static void pd_patch_instruction(address branch, address target, const char* file = NULL, int line = 0) {
pd_patch_instruction_size(branch, target);
}
static address pd_call_destination(address branch) {
return target_addr_for_insn(branch);
}
#ifndef PRODUCT
static void pd_print_patched_instruction(address branch);
#endif
static int patch_oop(address insn_addr, address o);
static int patch_narrow_klass(address insn_addr, narrowKlass n);
address emit_trampoline_stub(int insts_call_instruction_offset, address target);
void emit_static_call_stub();
// The following 4 methods return the offset of the appropriate move instruction
// Support for fast byte/short loading with zero extension (depending on particular CPU)
int load_unsigned_byte(Register dst, Address src);
int load_unsigned_short(Register dst, Address src);
// Support for fast byte/short loading with sign extension (depending on particular CPU)
int load_signed_byte(Register dst, Address src);
int load_signed_short(Register dst, Address src);
int load_signed_byte32(Register dst, Address src);
int load_signed_short32(Register dst, Address src);
// Support for sign-extension (hi:lo = extend_sign(lo))
void extend_sign(Register hi, Register lo);
// Load and store values by size and signed-ness
void load_sized_value(Register dst, Address src, size_t size_in_bytes, bool is_signed, Register dst2 = noreg);
void store_sized_value(Address dst, Register src, size_t size_in_bytes, Register src2 = noreg);
// Support for inc/dec with optimal instruction selection depending on value
// x86_64 aliases an unqualified register/address increment and
// decrement to call incrementq and decrementq but also supports
// explicitly sized calls to incrementq/decrementq or
// incrementl/decrementl
// for aarch64 the proper convention would be to use
// increment/decrement for 64 bit operatons and
// incrementw/decrementw for 32 bit operations. so when porting
// x86_64 code we can leave calls to increment/decrement as is,
// replace incrementq/decrementq with increment/decrement and
// replace incrementl/decrementl with incrementw/decrementw.
// n.b. increment/decrement calls with an Address destination will
// need to use a scratch register to load the value to be
// incremented. increment/decrement calls which add or subtract a
// constant value greater than 2^12 will need to use a 2nd scratch
// register to hold the constant. so, a register increment/decrement
// may trash rscratch2 and an address increment/decrement trash
// rscratch and rscratch2
void decrementw(Address dst, int value = 1);
void decrementw(Register reg, int value = 1);
void decrement(Register reg, int value = 1);
void decrement(Address dst, int value = 1);
void incrementw(Address dst, int value = 1);
void incrementw(Register reg, int value = 1);
void increment(Register reg, int value = 1);
void increment(Address dst, int value = 1);
// Alignment
void align(int modulus);
// Stack frame creation/removal
void enter(bool strip_ret_addr = false);
void leave();
// ROP Protection
void protect_return_address();
void protect_return_address(Register return_reg, Register temp_reg);
void authenticate_return_address(Register return_reg = lr);
void authenticate_return_address(Register return_reg, Register temp_reg);
void strip_return_address();
void check_return_address(Register return_reg=lr) PRODUCT_RETURN;
// Support for getting the JavaThread pointer (i.e.; a reference to thread-local information)
// The pointer will be loaded into the thread register.
void get_thread(Register thread);
// Support for VM calls
//
// It is imperative that all calls into the VM are handled via the call_VM macros.
// They make sure that the stack linkage is setup correctly. call_VM's correspond
// to ENTRY/ENTRY_X entry points while call_VM_leaf's correspond to LEAF entry points.
void call_VM(Register oop_result,
address entry_point,
bool check_exceptions = true);
void call_VM(Register oop_result,
address entry_point,
Register arg_1,
bool check_exceptions = true);
void call_VM(Register oop_result,
address entry_point,
Register arg_1, Register arg_2,
bool check_exceptions = true);
void call_VM(Register oop_result,
address entry_point,
Register arg_1, Register arg_2, Register arg_3,
bool check_exceptions = true);
// Overloadings with last_Java_sp
void call_VM(Register oop_result,
Register last_java_sp,
address entry_point,
int number_of_arguments = 0,
bool check_exceptions = true);
void call_VM(Register oop_result,
Register last_java_sp,
address entry_point,
Register arg_1, bool
check_exceptions = true);
void call_VM(Register oop_result,
Register last_java_sp,
address entry_point,
Register arg_1, Register arg_2,
bool check_exceptions = true);
void call_VM(Register oop_result,
Register last_java_sp,
address entry_point,
Register arg_1, Register arg_2, Register arg_3,
bool check_exceptions = true);
void get_vm_result (Register oop_result, Register thread);
void get_vm_result_2(Register metadata_result, Register thread);
// These always tightly bind to MacroAssembler::call_VM_base
// bypassing the virtual implementation
void super_call_VM(Register oop_result, Register last_java_sp, address entry_point, int number_of_arguments = 0, bool check_exceptions = true);
void super_call_VM(Register oop_result, Register last_java_sp, address entry_point, Register arg_1, bool check_exceptions = true);
void super_call_VM(Register oop_result, Register last_java_sp, address entry_point, Register arg_1, Register arg_2, bool check_exceptions = true);
void super_call_VM(Register oop_result, Register last_java_sp, address entry_point, Register arg_1, Register arg_2, Register arg_3, bool check_exceptions = true);
void super_call_VM(Register oop_result, Register last_java_sp, address entry_point, Register arg_1, Register arg_2, Register arg_3, Register arg_4, bool check_exceptions = true);
void call_VM_leaf(address entry_point,
int number_of_arguments = 0);
void call_VM_leaf(address entry_point,
Register arg_1);
void call_VM_leaf(address entry_point,
Register arg_1, Register arg_2);
void call_VM_leaf(address entry_point,
Register arg_1, Register arg_2, Register arg_3);
// These always tightly bind to MacroAssembler::call_VM_leaf_base
// bypassing the virtual implementation
void super_call_VM_leaf(address entry_point);
void super_call_VM_leaf(address entry_point, Register arg_1);
void super_call_VM_leaf(address entry_point, Register arg_1, Register arg_2);
void super_call_VM_leaf(address entry_point, Register arg_1, Register arg_2, Register arg_3);
void super_call_VM_leaf(address entry_point, Register arg_1, Register arg_2, Register arg_3, Register arg_4);
// last Java Frame (fills frame anchor)
void set_last_Java_frame(Register last_java_sp,
Register last_java_fp,
address last_java_pc,
Register scratch);
void set_last_Java_frame(Register last_java_sp,
Register last_java_fp,
Label &last_java_pc,
Register scratch);
void set_last_Java_frame(Register last_java_sp,
Register last_java_fp,
Register last_java_pc,
Register scratch);
void reset_last_Java_frame(Register thread);
// thread in the default location (rthread)
void reset_last_Java_frame(bool clear_fp);
// Stores
void store_check(Register obj); // store check for obj - register is destroyed afterwards
void store_check(Register obj, Address dst); // same as above, dst is exact store location (reg. is destroyed)
void resolve_jobject(Register value, Register thread, Register tmp);
// C 'boolean' to Java boolean: x == 0 ? 0 : 1
void c2bool(Register x);
void load_method_holder_cld(Register rresult, Register rmethod);
void load_method_holder(Register holder, Register method);
// oop manipulations
void load_klass(Register dst, Register src);
void store_klass(Register dst, Register src);
void cmp_klass(Register oop, Register trial_klass, Register tmp);
void resolve_weak_handle(Register result, Register tmp);
void resolve_oop_handle(Register result, Register tmp = r5);
void load_mirror(Register dst, Register method, Register tmp = r5);
void access_load_at(BasicType type, DecoratorSet decorators, Register dst, Address src,
Register tmp1, Register tmp_thread);
void access_store_at(BasicType type, DecoratorSet decorators, Address dst, Register src,
Register tmp1, Register tmp_thread);
void load_heap_oop(Register dst, Address src, Register tmp1 = noreg,
Register thread_tmp = noreg, DecoratorSet decorators = 0);
void load_heap_oop_not_null(Register dst, Address src, Register tmp1 = noreg,
Register thread_tmp = noreg, DecoratorSet decorators = 0);
void store_heap_oop(Address dst, Register src, Register tmp1 = noreg,
Register tmp_thread = noreg, DecoratorSet decorators = 0);
// currently unimplemented
// Used for storing NULL. All other oop constants should be
// stored using routines that take a jobject.
void store_heap_oop_null(Address dst);
void store_klass_gap(Register dst, Register src);
// This dummy is to prevent a call to store_heap_oop from
// converting a zero (like NULL) into a Register by giving
// the compiler two choices it can't resolve
void store_heap_oop(Address dst, void* dummy);
void encode_heap_oop(Register d, Register s);
void encode_heap_oop(Register r) { encode_heap_oop(r, r); }
void decode_heap_oop(Register d, Register s);
void decode_heap_oop(Register r) { decode_heap_oop(r, r); }
void encode_heap_oop_not_null(Register r);
void decode_heap_oop_not_null(Register r);
void encode_heap_oop_not_null(Register dst, Register src);
void decode_heap_oop_not_null(Register dst, Register src);
void set_narrow_oop(Register dst, jobject obj);
void encode_klass_not_null(Register r);
void decode_klass_not_null(Register r);
void encode_klass_not_null(Register dst, Register src);
void decode_klass_not_null(Register dst, Register src);
void set_narrow_klass(Register dst, Klass* k);
// if heap base register is used - reinit it with the correct value
void reinit_heapbase();
DEBUG_ONLY(void verify_heapbase(const char* msg);)
void push_CPU_state(bool save_vectors = false, bool use_sve = false,
int sve_vector_size_in_bytes = 0, int total_predicate_in_bytes = 0);
void pop_CPU_state(bool restore_vectors = false, bool use_sve = false,
int sve_vector_size_in_bytes = 0, int total_predicate_in_bytes = 0);
// Round up to a power of two
void round_to(Register reg, int modulus);
// allocation
void eden_allocate(
Register obj, // result: pointer to object after successful allocation
Register var_size_in_bytes, // object size in bytes if unknown at compile time; invalid otherwise
int con_size_in_bytes, // object size in bytes if known at compile time
Register t1, // temp register
Label& slow_case // continuation point if fast allocation fails
);
void tlab_allocate(
Register obj, // result: pointer to object after successful allocation
Register var_size_in_bytes, // object size in bytes if unknown at compile time; invalid otherwise
int con_size_in_bytes, // object size in bytes if known at compile time
Register t1, // temp register
Register t2, // temp register
Label& slow_case // continuation point if fast allocation fails
);
void verify_tlab();
// interface method calling
void lookup_interface_method(Register recv_klass,
Register intf_klass,
RegisterOrConstant itable_index,
Register method_result,
Register scan_temp,
Label& no_such_interface,
bool return_method = true);
// virtual method calling
// n.b. x86 allows RegisterOrConstant for vtable_index
void lookup_virtual_method(Register recv_klass,
RegisterOrConstant vtable_index,
Register method_result);
// Test sub_klass against super_klass, with fast and slow paths.
// The fast path produces a tri-state answer: yes / no / maybe-slow.
// One of the three labels can be NULL, meaning take the fall-through.
// If super_check_offset is -1, the value is loaded up from super_klass.
// No registers are killed, except temp_reg.
void check_klass_subtype_fast_path(Register sub_klass,
Register super_klass,
Register temp_reg,
Label* L_success,
Label* L_failure,
Label* L_slow_path,
RegisterOrConstant super_check_offset = RegisterOrConstant(-1));
// The rest of the type check; must be wired to a corresponding fast path.
// It does not repeat the fast path logic, so don't use it standalone.
// The temp_reg and temp2_reg can be noreg, if no temps are available.
// Updates the sub's secondary super cache as necessary.
// If set_cond_codes, condition codes will be Z on success, NZ on failure.
void check_klass_subtype_slow_path(Register sub_klass,
Register super_klass,
Register temp_reg,
Register temp2_reg,
Label* L_success,
Label* L_failure,
bool set_cond_codes = false);
// Simplified, combined version, good for typical uses.
// Falls through on failure.
void check_klass_subtype(Register sub_klass,
Register super_klass,
Register temp_reg,
Label& L_success);
void clinit_barrier(Register klass,
Register thread,
Label* L_fast_path = NULL,
Label* L_slow_path = NULL);
Address argument_address(RegisterOrConstant arg_slot, int extra_slot_offset = 0);
void verify_sve_vector_length();
void reinitialize_ptrue() {
if (UseSVE > 0) {
sve_ptrue(ptrue, B);
}
}
void verify_ptrue();
// Debugging
// only if +VerifyOops
void verify_oop(Register reg, const char* s = "broken oop");
void verify_oop_addr(Address addr, const char * s = "broken oop addr");
// TODO: verify method and klass metadata (compare against vptr?)
void _verify_method_ptr(Register reg, const char * msg, const char * file, int line) {}
void _verify_klass_ptr(Register reg, const char * msg, const char * file, int line){}
#define verify_method_ptr(reg) _verify_method_ptr(reg, "broken method " #reg, __FILE__, __LINE__)
#define verify_klass_ptr(reg) _verify_klass_ptr(reg, "broken klass " #reg, __FILE__, __LINE__)
// only if +VerifyFPU
void verify_FPU(int stack_depth, const char* s = "illegal FPU state");
// prints msg, dumps registers and stops execution
void stop(const char* msg);
static void debug64(char* msg, int64_t pc, int64_t regs[]);
void untested() { stop("untested"); }
void unimplemented(const char* what = "");
void should_not_reach_here() { stop("should not reach here"); }
// Stack overflow checking
void bang_stack_with_offset(int offset) {
// stack grows down, caller passes positive offset
assert(offset > 0, "must bang with negative offset");
sub(rscratch2, sp, offset);
str(zr, Address(rscratch2));
}
// Writes to stack successive pages until offset reached to check for
// stack overflow + shadow pages. Also, clobbers tmp
void bang_stack_size(Register size, Register tmp);