-
Notifications
You must be signed in to change notification settings - Fork 107
/
Copy pathemulate.c
1347 lines (1214 loc) · 41.2 KB
/
emulate.c
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
/*
* rv32emu is freely redistributable under the MIT License. See the file
* "LICENSE" for information on usage and redistribution of this file.
*/
#include <assert.h>
#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#ifdef __EMSCRIPTEN__
#include <emscripten.h>
#endif
#if RV32_HAS(EXT_F)
#include <math.h>
#include "softfloat.h"
#endif /* RV32_HAS(EXT_F) */
#if RV32_HAS(GDBSTUB)
extern struct target_ops gdbstub_ops;
#endif
#include "decode.h"
#include "mpool.h"
#include "riscv.h"
#include "riscv_private.h"
#include "utils.h"
#if RV32_HAS(JIT)
#include "cache.h"
#include "jit.h"
#endif
/* Shortcuts for comparing each field of specified RISC-V instruction */
#define IF_insn(i, o) (i->opcode == rv_insn_##o)
#define IF_rd(i, r) (i->rd == rv_reg_##r)
#define IF_rs1(i, r) (i->rs1 == rv_reg_##r)
#define IF_rs2(i, r) (i->rs2 == rv_reg_##r)
#define IF_imm(i, v) (i->imm == v)
#if RV32_HAS(SYSTEM)
#if !RV32_HAS(JIT)
static bool need_clear_block_map = false;
#endif
static uint32_t reloc_enable_mmu_jalr_addr;
static bool reloc_enable_mmu = false;
bool need_retranslate = false;
bool need_handle_signal = false;
#endif
static void rv_trap_default_handler(riscv_t *rv)
{
rv->csr_mepc += rv->compressed ? 2 : 4;
rv->PC = rv->csr_mepc; /* mret */
}
#if RV32_HAS(SYSTEM)
static void __trap_handler(riscv_t *rv);
#endif /* RV32_HAS(SYSTEM) */
/* wrap load/store and insn misaligned handler
* @mask_or_pc: mask for load/store and pc for insn misaligned handler.
* @type: type of misaligned handler
* @compress: compressed instruction or not
* @IO: whether the misaligned handler is for load/store or insn.
*/
#define RV_EXC_MISALIGN_HANDLER(mask_or_pc, type, compress, IO) \
IIF(IO) \
(if (!PRIV(rv)->allow_misalign && unlikely(addr & (mask_or_pc))), \
if (unlikely(insn_is_misaligned(PC)))) \
{ \
rv->compressed = compress; \
rv->csr_cycle = cycle; \
rv->PC = PC; \
SET_CAUSE_AND_TVAL_THEN_TRAP(rv, type##_MISALIGNED, \
IIF(IO)(addr, mask_or_pc)); \
return false; \
}
/* FIXME: use more precise methods for updating time, e.g., RTC */
#if RV32_HAS(Zicsr)
static uint64_t ctr = 0;
static inline void update_time(riscv_t *rv)
{
rv->csr_time[0] = ctr & 0xFFFFFFFF;
rv->csr_time[1] = ctr >> 32;
}
/* get a pointer to a CSR */
static uint32_t *csr_get_ptr(riscv_t *rv, uint32_t csr)
{
/* csr & 0xFFF prevent sign-extension in decode stage */
switch (csr & 0xFFF) {
case CSR_MSTATUS: /* Machine Status */
return (uint32_t *) (&rv->csr_mstatus);
case CSR_MTVEC: /* Machine Trap Handler */
return (uint32_t *) (&rv->csr_mtvec);
case CSR_MISA: /* Machine ISA and Extensions */
return (uint32_t *) (&rv->csr_misa);
/* Machine Trap Handling */
case CSR_MEDELEG: /* Machine Exception Delegation Register */
return (uint32_t *) (&rv->csr_medeleg);
case CSR_MIDELEG: /* Machine Interrupt Delegation Register */
return (uint32_t *) (&rv->csr_mideleg);
case CSR_MSCRATCH: /* Machine Scratch Register */
return (uint32_t *) (&rv->csr_mscratch);
case CSR_MEPC: /* Machine Exception Program Counter */
return (uint32_t *) (&rv->csr_mepc);
case CSR_MCAUSE: /* Machine Exception Cause */
return (uint32_t *) (&rv->csr_mcause);
case CSR_MTVAL: /* Machine Trap Value */
return (uint32_t *) (&rv->csr_mtval);
case CSR_MIP: /* Machine Interrupt Pending */
return (uint32_t *) (&rv->csr_mip);
/* Machine Counter/Timers */
case CSR_CYCLE: /* Cycle counter for RDCYCLE instruction */
return (uint32_t *) &rv->csr_cycle;
case CSR_CYCLEH: /* Upper 32 bits of cycle */
return &((uint32_t *) &rv->csr_cycle)[1];
/* TIME/TIMEH - very roughly about 1 ms per tick */
case CSR_TIME: /* Timer for RDTIME instruction */
update_time(rv);
return &rv->csr_time[0];
case CSR_TIMEH: /* Upper 32 bits of time */
update_time(rv);
return &rv->csr_time[1];
case CSR_INSTRET: /* Number of Instructions Retired Counter */
return (uint32_t *) (&rv->csr_cycle);
#if RV32_HAS(EXT_F)
case CSR_FFLAGS:
return (uint32_t *) (&rv->csr_fcsr);
case CSR_FCSR:
return (uint32_t *) (&rv->csr_fcsr);
#endif
case CSR_SSTATUS:
return (uint32_t *) (&rv->csr_sstatus);
case CSR_SIE:
return (uint32_t *) (&rv->csr_sie);
case CSR_STVEC:
return (uint32_t *) (&rv->csr_stvec);
case CSR_SCOUNTEREN:
return (uint32_t *) (&rv->csr_scounteren);
case CSR_SSCRATCH:
return (uint32_t *) (&rv->csr_sscratch);
case CSR_SEPC:
return (uint32_t *) (&rv->csr_sepc);
case CSR_SCAUSE:
return (uint32_t *) (&rv->csr_scause);
case CSR_STVAL:
return (uint32_t *) (&rv->csr_stval);
case CSR_SIP:
return (uint32_t *) (&rv->csr_sip);
case CSR_SATP:
return (uint32_t *) (&rv->csr_satp);
default:
return NULL;
}
}
/* CSRRW (Atomic Read/Write CSR) instruction atomically swaps values in the
* CSRs and integer registers. CSRRW reads the old value of the CSR,
* zero-extends the value to XLEN bits, and then writes it to register rd.
* The initial value in rs1 is written to the CSR.
* If rd == x0, then the instruction shall not read the CSR and shall not cause
* any of the side effects that might occur on a CSR read.
*/
static uint32_t csr_csrrw(riscv_t *rv, uint32_t csr, uint32_t val)
{
uint32_t *c = csr_get_ptr(rv, csr);
if (!c)
return 0;
uint32_t out = *c;
#if RV32_HAS(EXT_F)
if (csr == CSR_FFLAGS)
out &= FFLAG_MASK;
#endif
*c = val;
#if !RV32_HAS(JIT) && RV32_HAS(SYSTEM)
/*
* guestOS's process might have same VA, so block map cannot be reused
*
* Instead of calling block_map_clear() directly here,
* a flag is set to indicate that the block map should be cleared,
* and the clearing occurs after the corresponding 'code' of RVOP
* has executed. This prevents the 'code' of RVOP from potentially
* accessing a NULL ir.
*/
if (c == &rv->csr_satp)
need_clear_block_map = true;
#endif
return out;
}
/* perform csrrs (atomic read and set) */
static uint32_t csr_csrrs(riscv_t *rv, uint32_t csr, uint32_t val)
{
uint32_t *c = csr_get_ptr(rv, csr);
if (!c)
return 0;
uint32_t out = *c;
#if RV32_HAS(EXT_F)
if (csr == CSR_FFLAGS)
out &= FFLAG_MASK;
#endif
*c |= val;
return out;
}
/* perform csrrc (atomic read and clear)
* Read old value of CSR, zero-extend to XLEN bits, write to rd.
* Read value from rs1, use as bit mask to clear bits in CSR.
*/
static uint32_t csr_csrrc(riscv_t *rv, uint32_t csr, uint32_t val)
{
uint32_t *c = csr_get_ptr(rv, csr);
if (!c)
return 0;
uint32_t out = *c;
#if RV32_HAS(EXT_F)
if (csr == CSR_FFLAGS)
out &= FFLAG_MASK;
#endif
*c &= ~val;
return out;
}
#endif
#if RV32_HAS(GDBSTUB)
void rv_debug(riscv_t *rv)
{
if (!gdbstub_init(&rv->gdbstub, &gdbstub_ops,
(arch_info_t){
.reg_num = 33,
.reg_byte = 4,
.target_desc = TARGET_RV32,
},
GDBSTUB_COMM)) {
return;
}
rv->debug_mode = true;
rv->breakpoint_map = breakpoint_map_new();
rv->is_interrupted = false;
if (!gdbstub_run(&rv->gdbstub, (void *) rv))
return;
breakpoint_map_destroy(rv->breakpoint_map);
gdbstub_close(&rv->gdbstub);
}
#endif /* RV32_HAS(GDBSTUB) */
#if !RV32_HAS(JIT)
/* hash function for the block map */
HASH_FUNC_IMPL(map_hash, BLOCK_MAP_CAPACITY_BITS, 1 << BLOCK_MAP_CAPACITY_BITS)
#endif
/* allocate a basic block */
static block_t *block_alloc(riscv_t *rv)
{
block_t *block = mpool_alloc(rv->block_mp);
assert(block);
block->n_insn = 0;
#if RV32_HAS(JIT)
block->translatable = true;
block->hot = false;
block->hot2 = false;
block->has_loops = false;
block->n_invoke = 0;
INIT_LIST_HEAD(&block->list);
#if RV32_HAS(T2C)
block->compiled = false;
#endif
#endif
return block;
}
#if !RV32_HAS(JIT)
/* insert a block into block map */
static void block_insert(block_map_t *map, const block_t *block)
{
assert(map && block);
const uint32_t mask = map->block_capacity - 1;
uint32_t index = map_hash(block->pc_start);
/* insert into the block map */
for (;; index++) {
if (!map->map[index & mask]) {
map->map[index & mask] = (block_t *) block;
break;
}
}
map->size++;
}
/* try to locate an already translated block in the block map */
static block_t *block_find(const block_map_t *map, const uint32_t addr)
{
assert(map);
uint32_t index = map_hash(addr);
const uint32_t mask = map->block_capacity - 1;
/* find block in block map */
for (;; index++) {
block_t *block = map->map[index & mask];
if (!block)
return NULL;
if (block->pc_start == addr)
return block;
}
return NULL;
}
#endif
#if !RV32_HAS(EXT_C)
FORCE_INLINE bool insn_is_misaligned(uint32_t pc)
{
return pc & 0x3;
}
#endif
/* instruction length information for each RISC-V instruction */
enum {
#define _(inst, can_branch, insn_len, translatable, reg_mask) \
__rv_insn_##inst##_len = insn_len,
RV_INSN_LIST
#undef _
};
/* can-branch information for each RISC-V instruction */
enum {
#define _(inst, can_branch, insn_len, translatable, reg_mask) \
__rv_insn_##inst##_canbranch = can_branch,
RV_INSN_LIST
#undef _
};
#if RV32_HAS(GDBSTUB)
#define RVOP_NO_NEXT(ir) \
(!ir->next | rv->debug_mode IIF(RV32_HAS(SYSTEM))(| rv->is_trapped, ))
#else
#define RVOP_NO_NEXT(ir) (!ir->next IIF(RV32_HAS(SYSTEM))(| rv->is_trapped, ))
#endif
/* record whether the branch is taken or not during emulation */
static bool is_branch_taken = false;
/* record the program counter of the previous block */
static uint32_t last_pc = 0;
#if RV32_HAS(JIT)
static set_t pc_set;
static bool has_loops = false;
#endif
#if RV32_HAS(SYSTEM) && !RV32_HAS(ELF_LOADER)
extern void emu_update_uart_interrupts(riscv_t *rv);
static uint32_t peripheral_update_ctr = 64;
#endif
/* Interpreter-based execution path */
#define RVOP(inst, code, asm) \
static bool do_##inst(riscv_t *rv, const rv_insn_t *ir, uint64_t cycle, \
uint32_t PC) \
{ \
IIF(RV32_HAS(SYSTEM))(ctr++;, ) cycle++; \
code; \
IIF(RV32_HAS(SYSTEM)) \
( \
if (need_handle_signal) { \
need_handle_signal = false; \
return true; \
}, ) nextop : PC += __rv_insn_##inst##_len; \
IIF(RV32_HAS(SYSTEM)) \
(IIF(RV32_HAS(JIT))( \
, if (unlikely(need_clear_block_map)) { \
block_map_clear(rv); \
need_clear_block_map = false; \
rv->csr_cycle = cycle; \
rv->PC = PC; \
return false; \
}), ); \
if (unlikely(RVOP_NO_NEXT(ir))) \
goto end_op; \
const rv_insn_t *next = ir->next; \
MUST_TAIL return next->impl(rv, next, cycle, PC); \
end_op: \
rv->csr_cycle = cycle; \
rv->PC = PC; \
return true; \
}
#include "rv32_template.c"
#undef RVOP
/* multiple LUI */
static bool do_fuse1(riscv_t *rv,
const rv_insn_t *ir,
uint64_t cycle,
uint32_t PC)
{
cycle += ir->imm2;
opcode_fuse_t *fuse = ir->fuse;
for (int i = 0; i < ir->imm2; i++)
rv->X[fuse[i].rd] = fuse[i].imm;
PC += ir->imm2 * 4;
if (unlikely(RVOP_NO_NEXT(ir))) {
rv->csr_cycle = cycle;
rv->PC = PC;
return true;
}
const rv_insn_t *next = ir->next;
MUST_TAIL return next->impl(rv, next, cycle, PC);
}
/* LUI + ADD */
static bool do_fuse2(riscv_t *rv,
const rv_insn_t *ir,
uint64_t cycle,
uint32_t PC)
{
cycle += 2;
rv->X[ir->rd] = ir->imm;
rv->X[ir->rs2] = rv->X[ir->rd] + rv->X[ir->rs1];
PC += 8;
if (unlikely(RVOP_NO_NEXT(ir))) {
rv->csr_cycle = cycle;
rv->PC = PC;
return true;
}
const rv_insn_t *next = ir->next;
MUST_TAIL return next->impl(rv, next, cycle, PC);
}
/* multiple SW */
static bool do_fuse3(riscv_t *rv,
const rv_insn_t *ir,
uint64_t cycle,
uint32_t PC)
{
cycle += ir->imm2;
opcode_fuse_t *fuse = ir->fuse;
/* The memory addresses of the sw instructions are contiguous, thus only
* the first SW instruction needs to be checked to determine if its memory
* address is misaligned or if the memory chunk does not exist.
*/
for (int i = 0; i < ir->imm2; i++) {
uint32_t addr = rv->X[fuse[i].rs1] + fuse[i].imm;
RV_EXC_MISALIGN_HANDLER(3, STORE, false, 1);
rv->io.mem_write_w(rv, addr, rv->X[fuse[i].rs2]);
}
PC += ir->imm2 * 4;
if (unlikely(RVOP_NO_NEXT(ir))) {
rv->csr_cycle = cycle;
rv->PC = PC;
return true;
}
const rv_insn_t *next = ir->next;
MUST_TAIL return next->impl(rv, next, cycle, PC);
}
/* multiple LW */
static bool do_fuse4(riscv_t *rv,
const rv_insn_t *ir,
uint64_t cycle,
uint32_t PC)
{
cycle += ir->imm2;
opcode_fuse_t *fuse = ir->fuse;
/* The memory addresses of the lw instructions are contiguous, therefore
* only the first LW instruction needs to be checked to determine if its
* memory address is misaligned or if the memory chunk does not exist.
*/
for (int i = 0; i < ir->imm2; i++) {
uint32_t addr = rv->X[fuse[i].rs1] + fuse[i].imm;
RV_EXC_MISALIGN_HANDLER(3, LOAD, false, 1);
rv->X[fuse[i].rd] = rv->io.mem_read_w(rv, addr);
}
PC += ir->imm2 * 4;
if (unlikely(RVOP_NO_NEXT(ir))) {
rv->csr_cycle = cycle;
rv->PC = PC;
return true;
}
const rv_insn_t *next = ir->next;
MUST_TAIL return next->impl(rv, next, cycle, PC);
}
/* multiple shift immediate */
static bool do_fuse5(riscv_t *rv,
const rv_insn_t *ir,
uint64_t cycle,
uint32_t PC)
{
cycle += ir->imm2;
opcode_fuse_t *fuse = ir->fuse;
for (int i = 0; i < ir->imm2; i++)
shift_func(rv, (const rv_insn_t *) (&fuse[i]));
PC += ir->imm2 * 4;
if (unlikely(RVOP_NO_NEXT(ir))) {
rv->csr_cycle = cycle;
rv->PC = PC;
return true;
}
const rv_insn_t *next = ir->next;
MUST_TAIL return next->impl(rv, next, cycle, PC);
}
/* clang-format off */
static const void *dispatch_table[] = {
/* RV32 instructions */
#define _(inst, can_branch, insn_len, translatable, reg_mask) [rv_insn_##inst] = do_##inst,
RV_INSN_LIST
#undef _
/* Macro operation fusion instructions */
#define _(inst) [rv_insn_##inst] = do_##inst,
FUSE_INSN_LIST
#undef _
};
/* clang-format on */
FORCE_INLINE bool insn_is_branch(uint8_t opcode)
{
switch (opcode) {
#define _(inst, can_branch, insn_len, translatable, reg_mask) \
IIF(can_branch) \
(case rv_insn_##inst:, )
RV_INSN_LIST
#undef _
return true;
}
return false;
}
#if RV32_HAS(JIT)
FORCE_INLINE bool insn_is_translatable(uint8_t opcode)
{
switch (opcode) {
#define _(inst, can_branch, insn_len, translatable, reg_mask) \
IIF(translatable) \
(case rv_insn_##inst:, )
RV_INSN_LIST
#undef _
return true;
}
return false;
}
#endif
#if RV32_HAS(BLOCK_CHAINING)
FORCE_INLINE bool insn_is_unconditional_branch(uint8_t opcode)
{
switch (opcode) {
case rv_insn_ecall:
case rv_insn_ebreak:
case rv_insn_jal:
case rv_insn_jalr:
case rv_insn_mret:
#if RV32_HAS(Zicsr)
case rv_insn_csrrw:
#endif
#if RV32_HAS(SYSTEM)
case rv_insn_sret:
#endif
#if RV32_HAS(EXT_C)
case rv_insn_cj:
case rv_insn_cjalr:
case rv_insn_cjal:
case rv_insn_cjr:
case rv_insn_cebreak:
#endif
return true;
default:
return false;
}
}
FORCE_INLINE bool insn_is_direct_branch(uint8_t opcode)
{
switch (opcode) {
case rv_insn_jal:
#if RV32_HAS(EXT_C)
case rv_insn_cjal:
case rv_insn_cj:
#endif
return true;
default:
return false;
}
}
#endif
FORCE_INLINE bool insn_is_indirect_branch(uint8_t opcode)
{
switch (opcode) {
case rv_insn_jalr:
#if RV32_HAS(EXT_C)
case rv_insn_cjalr:
case rv_insn_cjr:
#endif
return true;
default:
return false;
}
}
static void block_translate(riscv_t *rv, block_t *block)
{
retranslate:
block->pc_start = block->pc_end = rv->PC;
rv_insn_t *prev_ir = NULL;
rv_insn_t *ir = mpool_calloc(rv->block_ir_mp);
block->ir_head = ir;
/* translate the basic block */
while (true) {
if (prev_ir)
prev_ir->next = ir;
/* fetch the next instruction */
uint32_t insn = rv->io.mem_ifetch(rv, block->pc_end);
#if RV32_HAS(SYSTEM)
if (!insn && need_retranslate) {
memset(block, 0, sizeof(block_t));
need_retranslate = false;
goto retranslate;
}
#endif
assert(insn);
/* decode the instruction */
if (!rv_decode(ir, insn)) {
rv->compressed = is_compressed(insn);
SET_CAUSE_AND_TVAL_THEN_TRAP(rv, ILLEGAL_INSN, insn);
break;
}
ir->impl = dispatch_table[ir->opcode];
ir->pc = block->pc_end; /* compute the end of pc */
block->pc_end += is_compressed(insn) ? 2 : 4;
block->n_insn++;
prev_ir = ir;
#if RV32_HAS(JIT)
if (!insn_is_translatable(ir->opcode))
block->translatable = false;
#endif
/* stop on branch */
if (insn_is_branch(ir->opcode)) {
if (insn_is_indirect_branch(ir->opcode)) {
ir->branch_table = calloc(1, sizeof(branch_history_table_t));
assert(ir->branch_table);
memset(ir->branch_table->PC, -1,
sizeof(uint32_t) * HISTORY_SIZE);
}
break;
}
ir = mpool_calloc(rv->block_ir_mp);
}
assert(prev_ir);
block->ir_tail = prev_ir;
block->ir_tail->next = NULL;
}
#if RV32_HAS(MOP_FUSION)
#define COMBINE_MEM_OPS(RW) \
next_ir = ir->next; \
count = 1; \
while (1) { \
if (next_ir->opcode != IIF(RW)(rv_insn_lw, rv_insn_sw)) \
break; \
count++; \
if (!next_ir->next) \
break; \
next_ir = next_ir->next; \
} \
if (count > 1) { \
ir->opcode = IIF(RW)(rv_insn_fuse4, rv_insn_fuse3); \
ir->fuse = malloc(count * sizeof(opcode_fuse_t)); \
assert(ir->fuse); \
ir->imm2 = count; \
memcpy(ir->fuse, ir, sizeof(opcode_fuse_t)); \
ir->impl = dispatch_table[ir->opcode]; \
next_ir = ir->next; \
for (int j = 1; j < count; j++, next_ir = next_ir->next) \
memcpy(ir->fuse + j, next_ir, sizeof(opcode_fuse_t)); \
remove_next_nth_ir(rv, ir, block, count - 1); \
}
static inline void remove_next_nth_ir(const riscv_t *rv,
rv_insn_t *ir,
block_t *block,
uint8_t n)
{
for (uint8_t i = 0; i < n; i++) {
rv_insn_t *next = ir->next;
ir->next = ir->next->next;
mpool_free(rv->block_ir_mp, next);
}
if (!ir->next)
block->ir_tail = ir;
block->n_insn -= n;
}
/* Check if instructions in a block match a specific pattern. If they do,
* rewrite them as fused instructions.
*
* Strategies are being devised to increase the number of instructions that
* match the pattern, including possible instruction reordering.
*/
static void match_pattern(riscv_t *rv, block_t *block)
{
uint32_t i;
rv_insn_t *ir;
for (i = 0, ir = block->ir_head; i < block->n_insn - 1;
i++, ir = ir->next) {
assert(ir);
rv_insn_t *next_ir = NULL;
int32_t count = 0;
switch (ir->opcode) {
case rv_insn_lui:
next_ir = ir->next;
switch (next_ir->opcode) {
case rv_insn_add:
if (ir->rd == next_ir->rs2 || ir->rd == next_ir->rs1) {
ir->opcode = rv_insn_fuse2;
ir->rs2 = next_ir->rd;
if (ir->rd == next_ir->rs2)
ir->rs1 = next_ir->rs1;
else
ir->rs1 = next_ir->rs2;
ir->impl = dispatch_table[ir->opcode];
remove_next_nth_ir(rv, ir, block, 1);
}
break;
case rv_insn_lui:
count = 1;
while (1) {
if (!IF_insn(next_ir, lui))
break;
count++;
if (!next_ir->next)
break;
next_ir = next_ir->next;
}
if (count > 1) {
ir->opcode = rv_insn_fuse1;
ir->fuse = malloc(count * sizeof(opcode_fuse_t));
assert(ir->fuse);
ir->imm2 = count;
memcpy(ir->fuse, ir, sizeof(opcode_fuse_t));
ir->impl = dispatch_table[ir->opcode];
next_ir = ir->next;
for (int j = 1; j < count; j++, next_ir = next_ir->next)
memcpy(ir->fuse + j, next_ir, sizeof(opcode_fuse_t));
remove_next_nth_ir(rv, ir, block, count - 1);
}
break;
}
break;
/* If the memory addresses of a sequence of store or load instructions
* are contiguous, combine these instructions.
*/
case rv_insn_sw:
COMBINE_MEM_OPS(0);
break;
case rv_insn_lw:
COMBINE_MEM_OPS(1);
break;
/* TODO: mixture of SW and LW */
/* TODO: reorder insturction to match pattern */
case rv_insn_slli:
case rv_insn_srli:
case rv_insn_srai:
count = 1;
next_ir = ir->next;
while (1) {
if (!IF_insn(next_ir, slli) && !IF_insn(next_ir, srli) &&
!IF_insn(next_ir, srai))
break;
count++;
if (!next_ir->next)
break;
next_ir = next_ir->next;
}
if (count > 1) {
ir->fuse = malloc(count * sizeof(opcode_fuse_t));
assert(ir->fuse);
memcpy(ir->fuse, ir, sizeof(opcode_fuse_t));
ir->opcode = rv_insn_fuse5;
ir->imm2 = count;
ir->impl = dispatch_table[ir->opcode];
next_ir = ir->next;
for (int j = 1; j < count; j++, next_ir = next_ir->next)
memcpy(ir->fuse + j, next_ir, sizeof(opcode_fuse_t));
remove_next_nth_ir(rv, ir, block, count - 1);
}
break;
}
}
}
#endif
typedef struct {
bool is_constant[N_RV_REGS];
uint32_t const_val[N_RV_REGS];
} constopt_info_t;
#define CONSTOPT(inst, code) \
static void constopt_##inst(rv_insn_t *ir UNUSED, \
constopt_info_t *info UNUSED) \
{ \
code; \
}
#include "rv32_constopt.c"
static const void *constopt_table[] = {
#define _(inst, can_branch, insn_len, translatable, reg_mask) \
[rv_insn_##inst] = constopt_##inst,
RV_INSN_LIST
#undef _
};
#undef CONSTOPT
typedef void (*constopt_func_t)(rv_insn_t *, constopt_info_t *);
static void optimize_constant(riscv_t *rv UNUSED, block_t *block)
{
constopt_info_t info = {.is_constant[0] = true};
assert(rv->X[0] == 0);
uint32_t i;
rv_insn_t *ir;
for (i = 0, ir = block->ir_head; i < block->n_insn; i++, ir = ir->next)
((constopt_func_t) constopt_table[ir->opcode])(ir, &info);
}
static block_t *prev = NULL;
static block_t *block_find_or_translate(riscv_t *rv)
{
#if !RV32_HAS(JIT)
block_map_t *map = &rv->block_map;
/* lookup the next block in the block map */
block_t *next_blk = block_find(map, rv->PC);
#else
/* lookup the next block in the block cache */
/*
* The function "cache_get()" gets the cached block by the given "key (PC)".
* In system simulation, the returned block might be dropped because it is
* not the one from the current process (by checking SATP CSR register).
*/
block_t *next_blk = (block_t *) cache_get(rv->block_cache, rv->PC, true);
#endif
if (next_blk)
return next_blk;
#if !RV32_HAS(JIT)
/* clear block list if it is going to be filled */
if (map->size * 1.25 > map->block_capacity) {
block_map_clear(rv);
prev = NULL;
}
#endif
/* allocate a new block */
next_blk = block_alloc(rv);
block_translate(rv, next_blk);
optimize_constant(rv, next_blk);
#if RV32_HAS(MOP_FUSION)
/* macro operation fusion */
match_pattern(rv, next_blk);
#endif
#if !RV32_HAS(JIT)
/* insert the block into block map */
block_insert(&rv->block_map, next_blk);
#else
list_add(&next_blk->list, &rv->block_list);
#if RV32_HAS(T2C)
pthread_mutex_lock(&rv->cache_lock);
#endif
/* insert the block into block cache */
block_t *replaced_blk = cache_put(rv->block_cache, rv->PC, next_blk);
if (!replaced_blk) {
#if RV32_HAS(T2C)
pthread_mutex_unlock(&rv->cache_lock);
#endif
return next_blk;
}
list_del_init(&replaced_blk->list);
if (prev == replaced_blk)
prev = NULL;
/* remove the connection from parents */
rv_insn_t *replaced_blk_entry = replaced_blk->ir_head;
/* TODO: record parents of each block to avoid traversing all blocks */
block_t *entry;
list_for_each_entry (entry, &rv->block_list, list) {
rv_insn_t *taken = entry->ir_tail->branch_taken,
*untaken = entry->ir_tail->branch_untaken;
if (taken == replaced_blk_entry) {
entry->ir_tail->branch_taken = NULL;
}
if (untaken == replaced_blk_entry) {
entry->ir_tail->branch_untaken = NULL;
}
}
/* free IRs in replaced block */
for (rv_insn_t *ir = replaced_blk->ir_head, *next_ir; ir != NULL;
ir = next_ir) {
next_ir = ir->next;
if (ir->fuse)
free(ir->fuse);
mpool_free(rv->block_ir_mp, ir);
}
mpool_free(rv->block_mp, replaced_blk);
#if RV32_HAS(T2C)
pthread_mutex_unlock(&rv->cache_lock);
#endif
#endif
assert(next_blk);
return next_blk;
}
#if RV32_HAS(JIT)
static bool runtime_profiler(riscv_t *rv, block_t *block)
{
/* Based on our observations, a significant number of true hotspots are
* characterized by high usage frequency and including loop. Consequently,
* we posit that our profiler could effectively identify hotspots using
* three key indicators.
*/
uint32_t freq = cache_freq(rv->block_cache, block->pc_start);
/* To profile a block after chaining, it must first be executed. */
if (unlikely(freq >= 2 && block->has_loops))
return true;
/* using frequency exceeds predetermined threshold */
if (unlikely(freq >= THRESHOLD))
return true;
return false;
}
#endif
#if RV32_HAS(SYSTEM) && !RV32_HAS(ELF_LOADER)
static bool rv_has_plic_trap(riscv_t *rv)
{
return ((rv->csr_sstatus & SSTATUS_SIE || !rv->priv_mode) &&
(rv->csr_sip & rv->csr_sie));
}
static void rv_check_interrupt(riscv_t *rv)
{
vm_attr_t *attr = PRIV(rv);
if (peripheral_update_ctr-- == 0) {
peripheral_update_ctr = 64;
u8250_check_ready(PRIV(rv)->uart);
if (PRIV(rv)->uart->in_ready)
emu_update_uart_interrupts(rv);
}
if (ctr > attr->timer)
rv->csr_sip |= RV_INT_STI;
else
rv->csr_sip &= ~RV_INT_STI;
if (rv_has_plic_trap(rv)) {