forked from JuliaLang/julia
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtask.c
1822 lines (1702 loc) · 66.1 KB
/
task.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
// This file is a part of Julia. License is MIT: https://julialang.org/license
/*
task.c
lightweight processes (symmetric coroutines)
*/
// need this to get the real definition of ucontext_t,
// if we're going to use the ucontext_t implementation there
//#if defined(__APPLE__) && defined(JL_HAVE_UCONTEXT)
//#pragma push_macro("_XOPEN_SOURCE")
//#define _XOPEN_SOURCE
//#include <ucontext.h>
//#pragma pop_macro("_XOPEN_SOURCE")
//#endif
// this is needed for !COPY_STACKS to work on linux
#ifdef _FORTIFY_SOURCE
// disable __longjmp_chk validation so that we can jump between stacks
// (which would normally be invalid to do with setjmp / longjmp)
#pragma push_macro("_FORTIFY_SOURCE")
#undef _FORTIFY_SOURCE
#include <setjmp.h>
#pragma pop_macro("_FORTIFY_SOURCE")
#endif
#include "platform.h"
#include <stdlib.h>
#include <string.h>
#include <signal.h>
#include <unistd.h>
#include <errno.h>
#include <inttypes.h>
#include "julia.h"
#include "julia_internal.h"
#include "threading.h"
#include "julia_assert.h"
#ifdef __cplusplus
extern "C" {
#endif
#if defined(_COMPILER_ASAN_ENABLED_)
static inline void sanitizer_start_switch_fiber(jl_ptls_t ptls, jl_task_t *from, jl_task_t *to) {
if (to->copy_stack)
__sanitizer_start_switch_fiber(&from->ctx.asan_fake_stack, (char*)ptls->stackbase-ptls->stacksize, ptls->stacksize);
else
__sanitizer_start_switch_fiber(&from->ctx.asan_fake_stack, to->stkbuf, to->bufsz);
}
static inline void sanitizer_start_switch_fiber_killed(jl_ptls_t ptls, jl_task_t *to) {
if (to->copy_stack)
__sanitizer_start_switch_fiber(NULL, (char*)ptls->stackbase-ptls->stacksize, ptls->stacksize);
else
__sanitizer_start_switch_fiber(NULL, to->stkbuf, to->bufsz);
}
static inline void sanitizer_finish_switch_fiber(jl_task_t *last, jl_task_t *current) {
__sanitizer_finish_switch_fiber(current->ctx.asan_fake_stack, NULL, NULL);
//(const void**)&last->stkbuf,
//&last->bufsz);
}
#else
static inline void sanitizer_start_switch_fiber(jl_ptls_t ptls, jl_task_t *from, jl_task_t *to) JL_NOTSAFEPOINT {}
static inline void sanitizer_start_switch_fiber_killed(jl_ptls_t ptls, jl_task_t *to) JL_NOTSAFEPOINT {}
static inline void sanitizer_finish_switch_fiber(jl_task_t *last, jl_task_t *current) JL_NOTSAFEPOINT {}
#endif
#if defined(_COMPILER_TSAN_ENABLED_)
// must defined as macros, since the function containing them must not return before the longjmp
#define tsan_destroy_ctx(_ptls, _ctx) do { \
jl_ucontext_t *_tsan_macro_ctx = (_ctx); \
if (_tsan_macro_ctx != &(_ptls)->root_task->ctx) { \
__tsan_destroy_fiber(_tsan_macro_ctx->tsan_state); \
} \
_tsan_macro_ctx->tsan_state = NULL; \
} while (0)
#define tsan_switch_to_ctx(_ctx) do { \
jl_ucontext_t *_tsan_macro_ctx = (_ctx); \
__tsan_switch_to_fiber(_tsan_macro_ctx->tsan_state, 0); \
} while (0)
#ifdef COPY_STACKS
#define tsan_destroy_copyctx(_ptls, _ctx) do { \
jl_ucontext_t *_tsan_macro_ctx = (_ctx); \
if (_tsan_macro_ctx != &(_ptls)->root_task->ctx) { \
__tsan_destroy_fiber(_tsan_macro_ctx->tsan_state); \
} \
_tsan_macro_ctx->tsan_state = NULL; \
} while (0)
#define tsan_switch_to_copyctx(_ctx) do { \
struct jl_stack_context_t *_tsan_macro_ctx = (_ctx); \
__tsan_switch_to_fiber(_tsan_macro_ctx->tsan_state, 0); \
} while (0)
#endif
#else
// just do minimal type-checking on the arguments
#define tsan_destroy_ctx(_ptls, _ctx) do { \
jl_ucontext_t *_tsan_macro_ctx = (_ctx); \
(void)_tsan_macro_ctx; \
} while (0)
#define tsan_switch_to_ctx(_ctx) do { \
jl_ucontext_t *_tsan_macro_ctx = (_ctx); \
(void)_tsan_macro_ctx; \
} while (0)
#ifdef COPY_STACKS
#define tsan_destroy_copyctx(_ptls, _ctx) do { \
jl_ucontext_t *_tsan_macro_ctx = (_ctx); \
(void)_tsan_macro_ctx; \
} while (0)
#define tsan_switch_to_copyctx(_ctx) do { \
jl_ucontext_t *_tsan_macro_ctx = (_ctx); \
(void)_tsan_macro_ctx; \
} while (0)
#endif
#endif
// empirically, jl_finish_task needs about 64k stack space to infer/run
// and additionally, gc-stack reserves 64k for the guard pages
#if defined(MINSIGSTKSZ)
#define MINSTKSZ (MINSIGSTKSZ > 131072 ? MINSIGSTKSZ : 131072)
#else
#define MINSTKSZ 131072
#endif
#ifdef _COMPILER_ASAN_ENABLED_
#define ROOT_TASK_STACK_ADJUSTMENT 0
#else
#define ROOT_TASK_STACK_ADJUSTMENT 3000000
#endif
#ifdef JL_HAVE_ASYNCIFY
// Switching logic is implemented in JavaScript
#define STATIC_OR_JS JL_DLLEXPORT
#else
#define STATIC_OR_JS static
#endif
static char *jl_alloc_fiber(_jl_ucontext_t *t, size_t *ssize, jl_task_t *owner) JL_NOTSAFEPOINT;
STATIC_OR_JS void jl_set_fiber(jl_ucontext_t *t);
STATIC_OR_JS void jl_swap_fiber(jl_ucontext_t *lastt, jl_ucontext_t *t);
STATIC_OR_JS void jl_start_fiber_swap(jl_ucontext_t *savet, jl_ucontext_t *t);
STATIC_OR_JS void jl_start_fiber_set(jl_ucontext_t *t);
#ifdef ALWAYS_COPY_STACKS
# ifndef COPY_STACKS
# error "ALWAYS_COPY_STACKS requires COPY_STACKS"
# endif
static int always_copy_stacks = 1;
#else
static int always_copy_stacks = 0;
#endif
#if defined(_COMPILER_ASAN_ENABLED_)
extern void __asan_get_shadow_mapping(size_t *shadow_scale, size_t *shadow_offset);
JL_NO_ASAN void *memcpy_noasan(void *dest, const void *src, size_t n) {
char *d = (char*)dest;
const char *s = (const char *)src;
for (size_t i = 0; i < n; ++i)
d[i] = s[i];
return dest;
}
JL_NO_ASAN void *memcpy_a16_noasan(uint64_t *dest, const uint64_t *src, size_t nb) {
uint64_t *end = (uint64_t*)((char*)src + nb);
while (src < end)
*(dest++) = *(src++);
return dest;
}
/* Copy stack are allocated as regular bigval objects and do no go through free_stack,
which would otherwise unpoison it before returning to the GC pool */
static void asan_free_copy_stack(void *stkbuf, size_t bufsz) {
__asan_unpoison_stack_memory((uintptr_t)stkbuf, bufsz);
}
#else
static void asan_free_copy_stack(void *stkbuf, size_t bufsz) {}
#endif
#ifdef COPY_STACKS
static void JL_NO_ASAN JL_NO_MSAN memcpy_stack_a16(uint64_t *to, uint64_t *from, size_t nb)
{
#if defined(_COMPILER_ASAN_ENABLED_)
/* Asan keeps shadow memory for everything on the stack. However, in general,
this function may touch invalid portions of the stack, since it just moves
the stack around. To keep ASAN's stack tracking capability intact, we need
to move the shadow memory along with the stack memory itself. */
size_t shadow_offset;
size_t shadow_scale;
__asan_get_shadow_mapping(&shadow_scale, &shadow_offset);
uintptr_t from_addr = (((uintptr_t)from) >> shadow_scale) + shadow_offset;
uintptr_t to_addr = (((uintptr_t)to) >> shadow_scale) + shadow_offset;
// Make sure that the shadow scale is compatible with the alignment, so
// we can copy whole bytes.
assert(shadow_scale <= 4);
size_t shadow_nb = nb >> shadow_scale;
// Copy over the shadow memory
memcpy_noasan((char*)to_addr, (char*)from_addr, shadow_nb);
memcpy_a16_noasan(jl_assume_aligned(to, 16), jl_assume_aligned(from, 16), nb);
#elif defined(_COMPILER_MSAN_ENABLED_)
# warning This function is imcompletely implemented for MSAN (TODO).
memcpy((char*)jl_assume_aligned(to, 16), (char*)jl_assume_aligned(from, 16), nb);
#else
memcpy((char*)jl_assume_aligned(to, 16), (char*)jl_assume_aligned(from, 16), nb);
//uint64_t *end = (uint64_t*)((char*)from + nb);
//while (from < end)
// *(to++) = *(from++);
#endif
}
static void NOINLINE save_stack(jl_ptls_t ptls, jl_task_t *lastt, jl_task_t **pt)
{
char *frame_addr = (char*)((uintptr_t)jl_get_frame_addr() & ~15);
char *stackbase = (char*)ptls->stackbase;
assert(stackbase > frame_addr);
size_t nb = stackbase - frame_addr;
void *buf;
if (lastt->bufsz < nb) {
asan_free_copy_stack(lastt->stkbuf, lastt->bufsz);
buf = (void*)jl_gc_alloc_buf(ptls, nb);
lastt->stkbuf = buf;
lastt->bufsz = nb;
}
else {
buf = lastt->stkbuf;
}
*pt = NULL; // clear the gc-root for the target task before copying the stack for saving
lastt->copy_stack = nb;
lastt->sticky = 1;
memcpy_stack_a16((uint64_t*)buf, (uint64_t*)frame_addr, nb);
// this task's stack could have been modified after
// it was marked by an incremental collection
// move the barrier back instead of walking it again here
jl_gc_wb_back(lastt);
}
JL_NO_ASAN static void NOINLINE JL_NORETURN restore_stack(jl_task_t *t, jl_ptls_t ptls, char *p)
{
size_t nb = t->copy_stack;
char *_x = (char*)ptls->stackbase - nb;
if (!p) {
// switch to a stackframe that's beyond the bounds of the last switch
p = _x;
if ((char*)&_x > _x) {
p = (char*)alloca((char*)&_x - _x);
}
restore_stack(t, ptls, p); // pass p to ensure the compiler can't tailcall this or avoid the alloca
}
void *_y = t->stkbuf;
assert(_x != NULL && _y != NULL);
memcpy_stack_a16((uint64_t*)_x, (uint64_t*)_y, nb); // destroys all but the current stackframe
#if defined(_OS_WINDOWS_)
jl_setcontext(&t->ctx.copy_ctx);
#else
jl_longjmp(t->ctx.copy_ctx.uc_mcontext, 1);
#endif
abort(); // unreachable
}
JL_NO_ASAN static void restore_stack2(jl_task_t *t, jl_ptls_t ptls, jl_task_t *lastt)
{
assert(t->copy_stack && !lastt->copy_stack);
size_t nb = t->copy_stack;
char *_x = (char*)ptls->stackbase - nb;
void *_y = t->stkbuf;
assert(_x != NULL && _y != NULL);
memcpy_stack_a16((uint64_t*)_x, (uint64_t*)_y, nb); // destroys all but the current stackframe
#if defined(JL_HAVE_UNW_CONTEXT)
volatile int returns = 0;
int r = unw_getcontext(&lastt->ctx.ctx);
if (++returns == 2) // r is garbage after the first return
return;
if (r != 0 || returns != 1)
abort();
#elif defined(JL_HAVE_ASM) || defined(JL_HAVE_SIGALTSTACK) || defined(_OS_WINDOWS_)
if (jl_setjmp(lastt->ctx.copy_ctx.uc_mcontext, 0))
return;
#else
#error COPY_STACKS is incompatible with this platform
#endif
tsan_switch_to_copyctx(&t->ctx);
#if defined(_OS_WINDOWS_)
jl_setcontext(&t->ctx.copy_ctx);
#else
jl_longjmp(t->ctx.copy_ctx.uc_mcontext, 1);
#endif
}
#endif
/* Rooted by the base module */
static _Atomic(jl_function_t*) task_done_hook_func JL_GLOBALLY_ROOTED = NULL;
void JL_NORETURN jl_finish_task(jl_task_t *t)
{
jl_task_t *ct = jl_current_task;
JL_PROBE_RT_FINISH_TASK(ct);
JL_SIGATOMIC_BEGIN();
if (ct->metrics_enabled) {
// [task] user_time -finished-> wait_time
assert(jl_atomic_load_relaxed(&ct->first_enqueued_at) != 0);
uint64_t now = jl_hrtime();
jl_atomic_store_relaxed(&ct->finished_at, now);
jl_atomic_fetch_add_relaxed(&ct->running_time_ns, now - jl_atomic_load_relaxed(&ct->last_started_running_at));
}
if (jl_atomic_load_relaxed(&ct->_isexception))
jl_atomic_store_release(&ct->_state, JL_TASK_STATE_FAILED);
else
jl_atomic_store_release(&t->_state, JL_TASK_STATE_DONE);
if (t->copy_stack) { // early free of stkbuf
asan_free_copy_stack(t->stkbuf, t->bufsz);
t->stkbuf = NULL;
}
// ensure that state is cleared
ct->ptls->in_finalizer = 0;
ct->ptls->in_pure_callback = 0;
ct->world_age = jl_atomic_load_acquire(&jl_world_counter);
// let the runtime know this task is dead and find a new task to run
jl_function_t *done = jl_atomic_load_relaxed(&task_done_hook_func);
if (done == NULL) {
done = (jl_function_t*)jl_get_global(jl_base_module, jl_symbol("task_done_hook"));
if (done != NULL)
jl_atomic_store_release(&task_done_hook_func, done);
}
if (done != NULL) {
jl_value_t *args[2] = {done, (jl_value_t*)t};
JL_TRY {
jl_apply(args, 2);
}
JL_CATCH {
jl_no_exc_handler(jl_current_exception(), ct);
}
}
jl_gc_debug_critical_error();
abort();
}
JL_DLLEXPORT void *jl_task_stack_buffer(jl_task_t *task, size_t *size, int *ptid)
{
size_t off = 0;
#ifndef _OS_WINDOWS_
jl_ptls_t ptls0 = jl_atomic_load_relaxed(&jl_all_tls_states)[0];
if (ptls0->root_task == task) {
// See jl_init_root_task(). The root task of the main thread
// has its buffer enlarged by an artificial 3000000 bytes, but
// that means that the start of the buffer usually points to
// inaccessible memory. We need to correct for this.
off = ROOT_TASK_STACK_ADJUSTMENT;
}
#endif
jl_ptls_t ptls2 = task->ptls;
*ptid = -1;
if (ptls2) {
*ptid = jl_atomic_load_relaxed(&task->tid);
#ifdef COPY_STACKS
if (task->copy_stack) {
*size = ptls2->stacksize;
return (char *)ptls2->stackbase - *size;
}
#endif
}
*size = task->bufsz - off;
return (void *)((char *)task->stkbuf + off);
}
JL_DLLEXPORT void jl_active_task_stack(jl_task_t *task,
char **active_start, char **active_end,
char **total_start, char **total_end)
{
if (!task->started) {
*total_start = *active_start = 0;
*total_end = *active_end = 0;
return;
}
jl_ptls_t ptls2 = task->ptls;
if (task->copy_stack && ptls2) {
*total_start = *active_start = (char*)ptls2->stackbase - ptls2->stacksize;
*total_end = *active_end = (char*)ptls2->stackbase;
}
else if (task->stkbuf) {
*total_start = *active_start = (char*)task->stkbuf;
#ifndef _OS_WINDOWS_
jl_ptls_t ptls0 = jl_atomic_load_relaxed(&jl_all_tls_states)[0];
if (ptls0->root_task == task) {
// See jl_init_root_task(). The root task of the main thread
// has its buffer enlarged by an artificial 3000000 bytes, but
// that means that the start of the buffer usually points to
// inaccessible memory. We need to correct for this.
*active_start += ROOT_TASK_STACK_ADJUSTMENT;
*total_start += ROOT_TASK_STACK_ADJUSTMENT;
}
#endif
*total_end = *active_end = (char*)task->stkbuf + task->bufsz;
#ifdef COPY_STACKS
// save_stack stores the stack of an inactive task in stkbuf, and the
// actual number of used bytes in copy_stack.
if (task->copy_stack > 1)
*active_end = (char*)task->stkbuf + task->copy_stack;
#endif
}
else {
// no stack allocated yet
*total_start = *active_start = 0;
*total_end = *active_end = 0;
return;
}
if (task == jl_current_task) {
// scan up to current `sp` for current thread and task
*active_start = (char*)jl_get_frame_addr();
}
}
// Marked noinline so we can consistently skip the associated frame.
// `skip` is number of additional frames to skip.
NOINLINE static void record_backtrace(jl_ptls_t ptls, int skip) JL_NOTSAFEPOINT
{
// storing bt_size in ptls ensures roots in bt_data will be found
ptls->bt_size = rec_backtrace(ptls->bt_data, JL_MAX_BT_SIZE, skip + 1);
}
JL_DLLEXPORT void jl_set_next_task(jl_task_t *task) JL_NOTSAFEPOINT
{
jl_current_task->ptls->next_task = task;
}
JL_DLLEXPORT jl_task_t *jl_get_next_task(void) JL_NOTSAFEPOINT
{
jl_task_t *ct = jl_current_task;
if (ct->ptls->next_task)
return ct->ptls->next_task;
return ct;
}
#ifdef _COMPILER_TSAN_ENABLED_
const char tsan_state_corruption[] = "TSAN state corrupted. Exiting HARD!\n";
#endif
JL_NO_ASAN static void ctx_switch(jl_task_t *lastt)
{
jl_ptls_t ptls = lastt->ptls;
jl_task_t **pt = &ptls->next_task;
jl_task_t *t = *pt;
assert(t != lastt);
// none of these locks should be held across a task switch
assert(ptls->locks.len == 0);
#ifdef _COMPILER_TSAN_ENABLED_
if (lastt->ctx.tsan_state != __tsan_get_current_fiber()) {
// Something went really wrong - don't even assume that we can
// use assert/abort which involve lots of signal handling that
// looks at the tsan state.
write(STDERR_FILENO, tsan_state_corruption, sizeof(tsan_state_corruption) - 1);
_exit(1);
}
#endif
int killed = jl_atomic_load_relaxed(&lastt->_state) != JL_TASK_STATE_RUNNABLE;
if (!t->started && !t->copy_stack) {
// may need to allocate the stack
if (t->stkbuf == NULL) {
t->stkbuf = jl_alloc_fiber(&t->ctx.ctx, &t->bufsz, t);
if (t->stkbuf == NULL) {
#ifdef COPY_STACKS
// fall back to stack copying if mmap fails
t->copy_stack = 1;
t->sticky = 1;
t->bufsz = 0;
if (always_copy_stacks)
memcpy(&t->ctx.copy_ctx, &ptls->copy_stack_ctx, sizeof(t->ctx.copy_ctx));
else
memcpy(&t->ctx.ctx, &ptls->base_ctx, sizeof(t->ctx.ctx));
#else
jl_throw(jl_memory_exception);
#endif
}
}
}
if (killed) {
*pt = NULL; // can't fail after here: clear the gc-root for the target task now
lastt->gcstack = NULL;
lastt->eh = NULL;
if (!lastt->copy_stack && lastt->stkbuf) {
// early free of stkbuf back to the pool
jl_release_task_stack(ptls, lastt);
}
}
else {
#ifdef COPY_STACKS
if (lastt->copy_stack) { // save the old copy-stack
save_stack(ptls, lastt, pt); // allocates (gc-safepoint, and can also fail)
if (jl_setjmp(lastt->ctx.copy_ctx.uc_mcontext, 0)) {
sanitizer_finish_switch_fiber(ptls->previous_task, jl_atomic_load_relaxed(&ptls->current_task));
// TODO: mutex unlock the thread we just switched from
return;
}
}
else
#endif
*pt = NULL; // can't fail after here: clear the gc-root for the target task now
}
// set up global state for new task and clear global state for old task
t->ptls = ptls;
jl_atomic_store_relaxed(&ptls->current_task, t);
JL_GC_PROMISE_ROOTED(t);
jl_signal_fence();
jl_set_pgcstack(&t->gcstack);
jl_signal_fence();
lastt->ptls = NULL;
#ifdef MIGRATE_TASKS
ptls->previous_task = lastt;
#endif
if (t->started) {
#ifdef COPY_STACKS
if (t->copy_stack) {
if (lastt->copy_stack) {
// Switching from copystack to copystack. Clear any shadow stack
// memory above the saved shadow stack.
uintptr_t stacktop = (uintptr_t)ptls->stackbase - t->copy_stack;
uintptr_t stackbottom = ((uintptr_t)jl_get_frame_addr() & ~15);
if (stackbottom < stacktop)
asan_unpoison_stack_memory(stackbottom, stacktop-stackbottom);
}
if (!killed && !lastt->copy_stack) {
sanitizer_start_switch_fiber(ptls, lastt, t);
restore_stack2(t, ptls, lastt);
} else {
tsan_switch_to_copyctx(&t->ctx);
if (killed) {
sanitizer_start_switch_fiber_killed(ptls, t);
tsan_destroy_copyctx(ptls, &lastt->ctx);
} else {
sanitizer_start_switch_fiber(ptls, lastt, t);
}
if (lastt->copy_stack) {
restore_stack(t, ptls, NULL); // (doesn't return)
}
else {
restore_stack(t, ptls, (char*)1); // (doesn't return)
}
}
}
else
#endif
{
if (lastt->copy_stack) {
// Switching away from a copystack to a non-copystack. Clear
// the whole shadow stack now, because otherwise we won't know
// how much stack memory to clear the next time we switch to
// a copystack.
uintptr_t stacktop = (uintptr_t)ptls->stackbase;
uintptr_t stackbottom = ((uintptr_t)jl_get_frame_addr() & ~15);
// We're not restoring the stack, but we still need to unpoison the
// stack, so it starts with a pristine stack.
asan_unpoison_stack_memory(stackbottom, stacktop-stackbottom);
}
if (killed) {
sanitizer_start_switch_fiber_killed(ptls, t);
tsan_switch_to_ctx(&t->ctx);
tsan_destroy_ctx(ptls, &lastt->ctx);
jl_set_fiber(&t->ctx); // (doesn't return)
abort(); // unreachable
}
else {
sanitizer_start_switch_fiber(ptls, lastt, t);
if (lastt->copy_stack) {
// Resume at the jl_setjmp earlier in this function,
// don't do a full task swap
tsan_switch_to_ctx(&t->ctx);
jl_set_fiber(&t->ctx); // (doesn't return)
}
else {
jl_swap_fiber(&lastt->ctx, &t->ctx);
}
}
}
}
else {
if (lastt->copy_stack) {
uintptr_t stacktop = (uintptr_t)ptls->stackbase;
uintptr_t stackbottom = ((uintptr_t)jl_get_frame_addr() & ~15);
// We're not restoring the stack, but we still need to unpoison the
// stack, so it starts with a pristine stack.
asan_unpoison_stack_memory(stackbottom, stacktop-stackbottom);
}
if (t->copy_stack && always_copy_stacks) {
tsan_switch_to_ctx(&t->ctx);
if (killed) {
sanitizer_start_switch_fiber_killed(ptls, t);
tsan_destroy_ctx(ptls, &lastt->ctx);
} else {
sanitizer_start_switch_fiber(ptls, lastt, t);
}
#ifdef COPY_STACKS
#if defined(_OS_WINDOWS_)
jl_setcontext(&t->ctx.copy_ctx);
#else
jl_longjmp(t->ctx.copy_ctx.uc_mcontext, 1);
#endif
#endif
abort(); // unreachable
}
else {
if (killed) {
sanitizer_start_switch_fiber_killed(ptls, t);
tsan_switch_to_ctx(&t->ctx);
tsan_destroy_ctx(ptls, &lastt->ctx);
jl_start_fiber_set(&t->ctx); // (doesn't return)
abort();
}
sanitizer_start_switch_fiber(ptls, lastt, t);
if (lastt->copy_stack) {
// Resume at the jl_setjmp earlier in this function
tsan_switch_to_ctx(&t->ctx);
jl_start_fiber_set(&t->ctx); // (doesn't return)
abort();
}
else {
jl_start_fiber_swap(&lastt->ctx, &t->ctx);
}
}
}
sanitizer_finish_switch_fiber(ptls->previous_task, jl_atomic_load_relaxed(&ptls->current_task));
}
JL_DLLEXPORT void jl_switch(void) JL_NOTSAFEPOINT_LEAVE JL_NOTSAFEPOINT_ENTER
{
jl_task_t *ct = jl_current_task;
jl_ptls_t ptls = ct->ptls;
jl_task_t *t = ptls->next_task;
if (t == ct) {
return;
}
int8_t gc_state = jl_gc_unsafe_enter(ptls);
if (t->started && t->stkbuf == NULL)
jl_error("attempt to switch to exited task");
if (ptls->in_finalizer)
jl_error("task switch not allowed from inside gc finalizer");
if (ptls->in_pure_callback)
jl_error("task switch not allowed from inside staged nor pure functions");
if (!jl_set_task_tid(t, jl_atomic_load_relaxed(&ct->tid))) // manually yielding to a task
jl_error("cannot switch to task running on another thread");
JL_PROBE_RT_PAUSE_TASK(ct);
// Store old values on the stack and reset
sig_atomic_t defer_signal = ptls->defer_signal;
int finalizers_inhibited = ptls->finalizers_inhibited;
ptls->finalizers_inhibited = 0;
jl_timing_block_t *blk = jl_timing_block_task_exit(ct, ptls);
ctx_switch(ct);
#ifdef MIGRATE_TASKS
ptls = ct->ptls;
t = ptls->previous_task;
ptls->previous_task = NULL;
assert(t != ct);
assert(jl_atomic_load_relaxed(&t->tid) == ptls->tid);
if (!t->sticky && !t->copy_stack)
jl_atomic_store_release(&t->tid, -1);
#else
assert(ptls == ct->ptls);
#endif
// Pop old values back off the stack
assert(ct == jl_current_task &&
0 != ct->ptls &&
0 == ptls->finalizers_inhibited);
ptls->finalizers_inhibited = finalizers_inhibited;
jl_timing_block_task_enter(ct, ptls, blk); (void)blk;
sig_atomic_t other_defer_signal = ptls->defer_signal;
ptls->defer_signal = defer_signal;
if (other_defer_signal && !defer_signal)
jl_sigint_safepoint(ptls);
JL_PROBE_RT_RUN_TASK(ct);
jl_gc_unsafe_leave(ptls, gc_state);
}
JL_DLLEXPORT void jl_switchto(jl_task_t **pt) JL_NOTSAFEPOINT_ENTER // n.b. this does not actually enter a safepoint
{
jl_set_next_task(*pt);
jl_switch();
}
JL_DLLEXPORT JL_NORETURN void jl_no_exc_handler(jl_value_t *e, jl_task_t *ct)
{
// NULL exception objects are used when rethrowing. we don't have a handler to process
// the exception stack, so at least report the exception at the top of the stack.
if (!e)
e = jl_current_exception();
jl_printf((JL_STREAM*)STDERR_FILENO, "fatal: error thrown and no exception handler available.\n");
jl_static_show((JL_STREAM*)STDERR_FILENO, e);
jl_printf((JL_STREAM*)STDERR_FILENO, "\n");
jlbacktrace(); // written to STDERR_FILENO
if (ct == NULL)
jl_raise(6);
jl_exit(1);
}
/* throw_internal - yield to exception handler */
#ifdef ENABLE_TIMINGS
#define pop_timings_stack() \
jl_timing_block_t *cur_block = ptls->timing_stack; \
while (cur_block && eh->timing_stack != cur_block) { \
cur_block = jl_timing_block_pop(cur_block); \
} \
assert(cur_block == eh->timing_stack);
#else
#define pop_timings_stack() /* Nothing */
#endif
#define throw_internal_body(altstack) \
assert(!jl_get_safe_restore()); \
jl_ptls_t ptls = ct->ptls; \
ptls->io_wait = 0; \
jl_gc_unsafe_enter(ptls); \
if (exception) { \
/* The temporary ptls->bt_data is rooted by special purpose code in the\
GC. This exists only for the purpose of preserving bt_data until we \
set ptls->bt_size=0 below. */ \
jl_push_excstack(&ct->excstack, exception, \
ptls->bt_data, ptls->bt_size); \
ptls->bt_size = 0; \
} \
assert(ct->excstack && ct->excstack->top); \
jl_handler_t *eh = ct->eh; \
if (eh != NULL) { \
if (altstack) ptls->sig_exception = NULL; \
pop_timings_stack() \
asan_unpoison_task_stack(ct, &eh->eh_ctx); \
jl_longjmp(eh->eh_ctx, 1); \
} \
else { \
jl_no_exc_handler(exception, ct); \
} \
assert(0);
static void JL_NORETURN throw_internal(jl_task_t *ct, jl_value_t *exception JL_MAYBE_UNROOTED)
{
CFI_NORETURN
JL_GC_PUSH1(&exception);
throw_internal_body(0);
jl_unreachable();
}
/* On the signal stack, we don't want to create any asan frames, but we do on the
normal, stack, so we split this function in two, depending on which context
we're calling it in. This also lets us avoid making a GC frame on the altstack,
which might end up getting corrupted if we recur here through another signal. */
JL_NO_ASAN static void JL_NORETURN throw_internal_altstack(jl_task_t *ct, jl_value_t *exception)
{
CFI_NORETURN
throw_internal_body(1);
jl_unreachable();
}
// record backtrace and raise an error
JL_DLLEXPORT void jl_throw(jl_value_t *e JL_MAYBE_UNROOTED)
{
assert(e != NULL);
jl_jmp_buf *safe_restore = jl_get_safe_restore();
jl_task_t *ct = jl_get_current_task();
if (safe_restore) {
asan_unpoison_task_stack(ct, safe_restore);
jl_longjmp(*safe_restore, 1);
}
if (ct == NULL) // During startup, or on other threads
jl_no_exc_handler(e, ct);
record_backtrace(ct->ptls, 1);
throw_internal(ct, e);
}
// rethrow with current excstack state
JL_DLLEXPORT void jl_rethrow(void)
{
jl_task_t *ct = jl_current_task;
jl_excstack_t *excstack = ct->excstack;
if (!excstack || excstack->top == 0)
jl_error("rethrow() not allowed outside a catch block");
throw_internal(ct, NULL);
}
// Special case throw for errors detected inside signal handlers. This is not
// (cannot be) called directly in the signal handler itself, but is returned to
// after the signal handler exits.
JL_DLLEXPORT JL_NO_ASAN void JL_NORETURN jl_sig_throw(void)
{
CFI_NORETURN
jl_jmp_buf *safe_restore = jl_get_safe_restore();
jl_task_t *ct = jl_current_task;
if (safe_restore) {
asan_unpoison_task_stack(ct, safe_restore);
jl_longjmp(*safe_restore, 1);
}
jl_ptls_t ptls = ct->ptls;
jl_value_t *e = ptls->sig_exception;
JL_GC_PROMISE_ROOTED(e);
throw_internal_altstack(ct, e);
}
JL_DLLEXPORT void jl_rethrow_other(jl_value_t *e JL_MAYBE_UNROOTED)
{
// TODO: Should uses of `rethrow(exc)` be replaced with a normal throw, now
// that exception stacks allow root cause analysis?
jl_task_t *ct = jl_current_task;
jl_excstack_t *excstack = ct->excstack;
if (!excstack || excstack->top == 0)
jl_error("rethrow(exc) not allowed outside a catch block");
// overwrite exception on top of stack. see jl_excstack_exception
jl_excstack_raw(excstack)[excstack->top-1].jlvalue = e;
JL_GC_PROMISE_ROOTED(e);
throw_internal(ct, NULL);
}
/* This is xoshiro256++ 1.0, used for tasklocal random number generation in Julia.
This implementation is intended for embedders and internal use by the runtime, and is
based on the reference implementation at https://prng.di.unimi.it
Credits go to David Blackman and Sebastiano Vigna for coming up with this PRNG.
They described xoshiro256++ in "Scrambled Linear Pseudorandom Number Generators",
ACM Trans. Math. Softw., 2021.
There is a pure Julia implementation in stdlib that tends to be faster when used from
within Julia, due to inlining and more aggressive architecture-specific optimizations.
*/
uint64_t jl_genrandom(uint64_t rngState[4]) JL_NOTSAFEPOINT
{
uint64_t s0 = rngState[0];
uint64_t s1 = rngState[1];
uint64_t s2 = rngState[2];
uint64_t s3 = rngState[3];
uint64_t t = s1 << 17;
uint64_t tmp = s0 + s3;
uint64_t res = ((tmp << 23) | (tmp >> 41)) + s0;
s2 ^= s0;
s3 ^= s1;
s1 ^= s2;
s0 ^= s3;
s2 ^= t;
s3 = (s3 << 45) | (s3 >> 19);
rngState[0] = s0;
rngState[1] = s1;
rngState[2] = s2;
rngState[3] = s3;
return res;
}
/*
The jl_rng_split function forks a task's RNG state in a way that is essentially
guaranteed to avoid collisions between the RNG streams of all tasks. The main
RNG is the xoshiro256++ RNG whose state is stored in rngState[0..3]. There is
also a small internal RNG used for task forking stored in rngState[4]. This
state is used to iterate a LCG (linear congruential generator), which is then
put through four different variations of the strongest PCG output function,
referred to as PCG-RXS-M-XS-64 [1]. This output function is invertible: it maps
a 64-bit state to 64-bit output; which is one of the reasons it's not
recommended for general purpose RNGs unless space is at a premium, but in our
usage invertibility is actually a benefit, as is explained below.
The goal of jl_rng_split is to perturb the state of each child task's RNG in
such a way each that for an entire tree of tasks spawned starting with a given
state in a root task, no two tasks have the same RNG state. Moreover, we want to
do this in a way that is deterministic and repeatable based on (1) the root
task's seed, (2) how many random numbers are generated, and (3) the task tree
structure. The RNG state of a parent task is allowed to affect the initial RNG
state of a child task, but the mere fact that a child was spawned should not
alter the RNG output of the parent. This second requirement rules out using the
main RNG to seed children -- some separate state must be maintained and changed
upon forking a child task while leaving the main RNG state unchanged.
The basic approach is that used by the DotMix [2] and SplitMix [3] RNG systems:
each task is uniquely identified by a sequence of "pedigree" numbers, indicating
where in the task tree it was spawned. This vector of pedigree coordinates is
then reduced to a single value by computing a dot product with a common vector
of random weights. The DotMix paper provides a proof that this dot product hash
value (referred to as a "compression function") is collision resistant in the
sense the the pairwise collision probability of two distinct tasks is 1/N where
N is the number of possible weight values. Both DotMix and SplitMix use a prime
value of N because the proof requires that the difference between two distinct
pedigree coordinates must be invertible, which is guaranteed by N being prime.
We take a different approach: we instead limit pedigree coordinates to being
binary instead -- when a task spawns a child, both tasks share the same pedigree
prefix, with the parent appending a zero and the child appending a one. This way
a binary pedigree vector uniquely identifies each task. Moreover, since the
coordinates are binary, the difference between coordinates is always one which
is its own inverse regardless of whether N is prime or not. This allows us to
compute the dot product modulo 2^64 using native machine arithmetic, which is
considerably more efficient and simpler to implement than arithmetic in a prime
modulus. It also means that when accumulating the dot product incrementally, as
described in SplitMix, we don't need to multiply weights by anything, we simply
add the random weight for the current task tree depth to the parent's dot
product to derive the child's dot product.
We use the LCG in rngState[4] to derive generate pseudorandom weights for the
dot product. Each time a child is forked, we update the LCG in both parent and
child tasks. In the parent, that's all we have to do -- the main RNG state
remains unchanged (recall that spawning a child should *not* affect subsequence
RNG draws in the parent). The next time the parent forks a child, the dot
product weight used will be different, corresponding to being a level deeper in
the binary task tree. In the child, we use the LCG state to generate four
pseudorandom 64-bit weights (more below) and add each weight to one of the
xoshiro256 state registers, rngState[0..3]. If we assume the main RNG remains
unused in all tasks, then each register rngState[0..3] accumulates a different
Dot/SplitMix dot product hash as additional child tasks are spawned. Each one is
collision resistant with a pairwise collision chance of only 1/2^64. Assuming
that the four pseudorandom 64-bit weight streams are sufficiently independent,
the pairwise collision probability for distinct tasks is 1/2^256. If we somehow
managed to spawn a trillion tasks, the probability of a collision would be on
the order of 1/10^54. Practically impossible. Put another way, this is the same
as the probability of two SHA256 hash values accidentally colliding, which we
generally consider so unlikely as not to be worth worrying about.
What about the random "junk" that's in the xoshiro256 state registers from
normal use of the RNG? For a tree of tasks spawned with no intervening samples
taken from the main RNG, all tasks start with the same junk which doesn't affect
the chance of collision. The Dot/SplitMix papers even suggest adding a random
base value to the dot product, so we can consider whatever happens to be in the
xoshiro256 registers to be that. What if the main RNG gets used between task
forks? In that case, the initial state registers will be different. The DotMix
collision resistance proof doesn't apply without modification, but we can
generalize the setup by adding a different base constant to each compression
function and observe that we still have a 1/N chance of the weight value
matching that exact difference. This proves collision resistance even between
tasks whose dot product hashes are computed with arbitrary offsets. We can
conclude that this scheme provides collision resistance even in the face of
different starting states of the main RNG. Does this seem too good to be true?
Perhaps another way of thinking about it will help. Suppose we seeded each task
completely randomly. Then there would also be a 1/2^256 chance of collision,
just as the DotMix proof gives. Essentially what the proof is telling us is that
if the weights are chosen uniformly and uncorrelated with the rest of the
compression function, then the dot product construction is a good enough way to
pseudorandomly seed each task. From that perspective, it's easier to believe
that adding an arbitrary constant to each seed doesn't worsen its randomness.
This leaves us with the question of how to generate four pseudorandom weights to
add to the rngState[0..3] registers at each depth of the task tree. The scheme
used here is that a single 64-bit LCG state is iterated in both parent and child
at each task fork, and four different variations of the PCG-RXS-M-XS-64 output
function are applied to that state to generate four different pseudorandom
weights. Another obvious way to generate four weights would be to iterate the
LCG four times per task split. There are two main reasons we've chosen to use
four output variants instead:
1. Advancing four times per fork reduces the set of possible weights that each
register can be perturbed by from 2^64 to 2^60. Since collision resistance is
proportional to the number of possible weight values, that would reduce
collision resistance.
2. It's easier to compute four PCG output variants in parallel. Iterating the
LCG is inherently sequential. Each PCG variant can be computed independently
from the LCG state. All four can even be computed at once with SIMD vector
instructions, but the compiler doesn't currently choose to do that.
A key question is whether the approach of using four variations of PCG-RXS-M-XS
is sufficiently random both within and between streams to provide the collision
resistance we expect. We obviously can't test that with 256 bits, but we have
tested it with a reduced state analogue using four PCG-RXS-M-XS-8 output
variations applied to a common 8-bit LCG. Test results do indicate sufficient
independence: a single register has collisions at 2^5 while four registers only
start having collisions at 2^20, which is actually better scaling of collision
resistance than we expect in theory. In theory, with one byte of resistance we
have a 50% chance of some collision at 20, which matches, but four bytes gives a
50% chance of collision at 2^17 and our (reduced size analogue) construction is
still collision free at 2^19. This may be due to the next observation, which guarantees collision avoidance for certain shapes of task trees as a result of using an
invertible RNG to generate weights.
In the specific case where a parent task spawns a sequence of child tasks with
no intervening usage of its main RNG, the parent and child tasks are actually
_guaranteed_ to have different RNG states. This is true because the four PCG
streams each produce every possible 2^64 bit output exactly once in the full
2^64 period of the LCG generator. This is considered a weakness of PCG-RXS-M-XS
when used as a general purpose RNG, but is quite beneficial in this application.
Since each of up to 2^64 children will be perturbed by different weights, they
cannot have hash collisions. What about parent colliding with child? That can
only happen if all four main RNG registers are perturbed by exactly zero. This
seems unlikely, but could it occur? Consider this part of each output function:
p ^= p >> ((p >> 59) + 5);
p *= m[i];
p ^= p >> 43
It's easy to check that this maps zero to zero. An unchanged parent RNG can only
happen if all four `p` values are zero at the end of this, which implies that
they were all zero at the beginning. However, that is impossible since the four
`p` values differ from `x` by different additive constants, so they cannot all
be zero. Stated more generally, this non-collision property: assuming the main
RNG isn't used between task forks, sibling and parent tasks cannot have RNG
collisions. If the task tree structure is more deeply nested or if there are