summaryrefslogtreecommitdiffstats
path: root/Python/ceval_gil.c
blob: 83f4e91e54573a487568fd0de23eac79673dddac (plain)
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
1001
1002
1003
1004
1005
1006

#include "Python.h"
#include "pycore_atomic.h"        // _Py_atomic_int
#include "pycore_ceval.h"         // _PyEval_SignalReceived()
#include "pycore_pyerrors.h"      // _PyErr_Fetch()
#include "pycore_pylifecycle.h"   // _PyErr_Print()
#include "pycore_initconfig.h"    // _PyStatus_OK()
#include "pycore_interp.h"        // _Py_RunGC()
#include "pycore_pymem.h"         // _PyMem_IsPtrFreed()

/*
   Notes about the implementation:

   - The GIL is just a boolean variable (locked) whose access is protected
     by a mutex (gil_mutex), and whose changes are signalled by a condition
     variable (gil_cond). gil_mutex is taken for short periods of time,
     and therefore mostly uncontended.

   - In the GIL-holding thread, the main loop (PyEval_EvalFrameEx) must be
     able to release the GIL on demand by another thread. A volatile boolean
     variable (gil_drop_request) is used for that purpose, which is checked
     at every turn of the eval loop. That variable is set after a wait of
     `interval` microseconds on `gil_cond` has timed out.

      [Actually, another volatile boolean variable (eval_breaker) is used
       which ORs several conditions into one. Volatile booleans are
       sufficient as inter-thread signalling means since Python is run
       on cache-coherent architectures only.]

   - A thread wanting to take the GIL will first let pass a given amount of
     time (`interval` microseconds) before setting gil_drop_request. This
     encourages a defined switching period, but doesn't enforce it since
     opcodes can take an arbitrary time to execute.

     The `interval` value is available for the user to read and modify
     using the Python API `sys.{get,set}switchinterval()`.

   - When a thread releases the GIL and gil_drop_request is set, that thread
     ensures that another GIL-awaiting thread gets scheduled.
     It does so by waiting on a condition variable (switch_cond) until
     the value of last_holder is changed to something else than its
     own thread state pointer, indicating that another thread was able to
     take the GIL.

     This is meant to prohibit the latency-adverse behaviour on multi-core
     machines where one thread would speculatively release the GIL, but still
     run and end up being the first to re-acquire it, making the "timeslices"
     much longer than expected.
     (Note: this mechanism is enabled with FORCE_SWITCHING above)
*/

// GH-89279: Force inlining by using a macro.
#if defined(_MSC_VER) && SIZEOF_INT == 4
#define _Py_atomic_load_relaxed_int32(ATOMIC_VAL) (assert(sizeof((ATOMIC_VAL)->_value) == 4), *((volatile int*)&((ATOMIC_VAL)->_value)))
#else
#define _Py_atomic_load_relaxed_int32(ATOMIC_VAL) _Py_atomic_load_relaxed(ATOMIC_VAL)
#endif

/* This can set eval_breaker to 0 even though gil_drop_request became
   1.  We believe this is all right because the eval loop will release
   the GIL eventually anyway. */
static inline void
COMPUTE_EVAL_BREAKER(PyInterpreterState *interp,
                     struct _ceval_runtime_state *ceval,
                     struct _ceval_state *ceval2)
{
    _Py_atomic_store_relaxed(&ceval2->eval_breaker,
        _Py_atomic_load_relaxed_int32(&ceval2->gil_drop_request)
        | (_Py_atomic_load_relaxed_int32(&ceval->signals_pending)
           && _Py_ThreadCanHandleSignals(interp))
        | (_Py_atomic_load_relaxed_int32(&ceval2->pending.calls_to_do)
           && _Py_ThreadCanHandlePendingCalls())
        | ceval2->pending.async_exc
        | _Py_atomic_load_relaxed_int32(&ceval2->gc_scheduled));
}


static inline void
SET_GIL_DROP_REQUEST(PyInterpreterState *interp)
{
    struct _ceval_state *ceval2 = &interp->ceval;
    _Py_atomic_store_relaxed(&ceval2->gil_drop_request, 1);
    _Py_atomic_store_relaxed(&ceval2->eval_breaker, 1);
}


static inline void
RESET_GIL_DROP_REQUEST(PyInterpreterState *interp)
{
    struct _ceval_runtime_state *ceval = &interp->runtime->ceval;
    struct _ceval_state *ceval2 = &interp->ceval;
    _Py_atomic_store_relaxed(&ceval2->gil_drop_request, 0);
    COMPUTE_EVAL_BREAKER(interp, ceval, ceval2);
}


static inline void
SIGNAL_PENDING_CALLS(PyInterpreterState *interp)
{
    struct _ceval_runtime_state *ceval = &interp->runtime->ceval;
    struct _ceval_state *ceval2 = &interp->ceval;
    _Py_atomic_store_relaxed(&ceval2->pending.calls_to_do, 1);
    COMPUTE_EVAL_BREAKER(interp, ceval, ceval2);
}


static inline void
UNSIGNAL_PENDING_CALLS(PyInterpreterState *interp)
{
    struct _ceval_runtime_state *ceval = &interp->runtime->ceval;
    struct _ceval_state *ceval2 = &interp->ceval;
    _Py_atomic_store_relaxed(&ceval2->pending.calls_to_do, 0);
    COMPUTE_EVAL_BREAKER(interp, ceval, ceval2);
}


static inline void
SIGNAL_PENDING_SIGNALS(PyInterpreterState *interp, int force)
{
    struct _ceval_runtime_state *ceval = &interp->runtime->ceval;
    struct _ceval_state *ceval2 = &interp->ceval;
    _Py_atomic_store_relaxed(&ceval->signals_pending, 1);
    if (force) {
        _Py_atomic_store_relaxed(&ceval2->eval_breaker, 1);
    }
    else {
        /* eval_breaker is not set to 1 if thread_can_handle_signals() is false */
        COMPUTE_EVAL_BREAKER(interp, ceval, ceval2);
    }
}


static inline void
UNSIGNAL_PENDING_SIGNALS(PyInterpreterState *interp)
{
    struct _ceval_runtime_state *ceval = &interp->runtime->ceval;
    struct _ceval_state *ceval2 = &interp->ceval;
    _Py_atomic_store_relaxed(&ceval->signals_pending, 0);
    COMPUTE_EVAL_BREAKER(interp, ceval, ceval2);
}


static inline void
SIGNAL_ASYNC_EXC(PyInterpreterState *interp)
{
    struct _ceval_state *ceval2 = &interp->ceval;
    ceval2->pending.async_exc = 1;
    _Py_atomic_store_relaxed(&ceval2->eval_breaker, 1);
}


static inline void
UNSIGNAL_ASYNC_EXC(PyInterpreterState *interp)
{
    struct _ceval_runtime_state *ceval = &interp->runtime->ceval;
    struct _ceval_state *ceval2 = &interp->ceval;
    ceval2->pending.async_exc = 0;
    COMPUTE_EVAL_BREAKER(interp, ceval, ceval2);
}

#ifndef NDEBUG
/* Ensure that tstate is valid */
static int
is_tstate_valid(PyThreadState *tstate)
{
    assert(!_PyMem_IsPtrFreed(tstate));
    assert(!_PyMem_IsPtrFreed(tstate->interp));
    return 1;
}
#endif

/*
 * Implementation of the Global Interpreter Lock (GIL).
 */

#include <stdlib.h>
#include <errno.h>

#include "pycore_atomic.h"


#include "condvar.h"

#define MUTEX_INIT(mut) \
    if (PyMUTEX_INIT(&(mut))) { \
        Py_FatalError("PyMUTEX_INIT(" #mut ") failed"); };
#define MUTEX_FINI(mut) \
    if (PyMUTEX_FINI(&(mut))) { \
        Py_FatalError("PyMUTEX_FINI(" #mut ") failed"); };
#define MUTEX_LOCK(mut) \
    if (PyMUTEX_LOCK(&(mut))) { \
        Py_FatalError("PyMUTEX_LOCK(" #mut ") failed"); };
#define MUTEX_UNLOCK(mut) \
    if (PyMUTEX_UNLOCK(&(mut))) { \
        Py_FatalError("PyMUTEX_UNLOCK(" #mut ") failed"); };

#define COND_INIT(cond) \
    if (PyCOND_INIT(&(cond))) { \
        Py_FatalError("PyCOND_INIT(" #cond ") failed"); };
#define COND_FINI(cond) \
    if (PyCOND_FINI(&(cond))) { \
        Py_FatalError("PyCOND_FINI(" #cond ") failed"); };
#define COND_SIGNAL(cond) \
    if (PyCOND_SIGNAL(&(cond))) { \
        Py_FatalError("PyCOND_SIGNAL(" #cond ") failed"); };
#define COND_WAIT(cond, mut) \
    if (PyCOND_WAIT(&(cond), &(mut))) { \
        Py_FatalError("PyCOND_WAIT(" #cond ") failed"); };
#define COND_TIMED_WAIT(cond, mut, microseconds, timeout_result) \
    { \
        int r = PyCOND_TIMEDWAIT(&(cond), &(mut), (microseconds)); \
        if (r < 0) \
            Py_FatalError("PyCOND_WAIT(" #cond ") failed"); \
        if (r) /* 1 == timeout, 2 == impl. can't say, so assume timeout */ \
            timeout_result = 1; \
        else \
            timeout_result = 0; \
    } \


#define DEFAULT_INTERVAL 5000

static void _gil_initialize(struct _gil_runtime_state *gil)
{
    _Py_atomic_int uninitialized = {-1};
    gil->locked = uninitialized;
    gil->interval = DEFAULT_INTERVAL;
}

static int gil_created(struct _gil_runtime_state *gil)
{
    return (_Py_atomic_load_explicit(&gil->locked, _Py_memory_order_acquire) >= 0);
}

static void create_gil(struct _gil_runtime_state *gil)
{
    MUTEX_INIT(gil->mutex);
#ifdef FORCE_SWITCHING
    MUTEX_INIT(gil->switch_mutex);
#endif
    COND_INIT(gil->cond);
#ifdef FORCE_SWITCHING
    COND_INIT(gil->switch_cond);
#endif
    _Py_atomic_store_relaxed(&gil->last_holder, 0);
    _Py_ANNOTATE_RWLOCK_CREATE(&gil->locked);
    _Py_atomic_store_explicit(&gil->locked, 0, _Py_memory_order_release);
}

static void destroy_gil(struct _gil_runtime_state *gil)
{
    /* some pthread-like implementations tie the mutex to the cond
     * and must have the cond destroyed first.
     */
    COND_FINI(gil->cond);
    MUTEX_FINI(gil->mutex);
#ifdef FORCE_SWITCHING
    COND_FINI(gil->switch_cond);
    MUTEX_FINI(gil->switch_mutex);
#endif
    _Py_atomic_store_explicit(&gil->locked, -1,
                              _Py_memory_order_release);
    _Py_ANNOTATE_RWLOCK_DESTROY(&gil->locked);
}

#ifdef HAVE_FORK
static void recreate_gil(struct _gil_runtime_state *gil)
{
    _Py_ANNOTATE_RWLOCK_DESTROY(&gil->locked);
    /* XXX should we destroy the old OS resources here? */
    create_gil(gil);
}
#endif

static void
drop_gil(struct _ceval_runtime_state *ceval, struct _ceval_state *ceval2,
         PyThreadState *tstate)
{
    struct _gil_runtime_state *gil = &ceval->gil;
    if (!_Py_atomic_load_relaxed(&gil->locked)) {
        Py_FatalError("drop_gil: GIL is not locked");
    }

    /* tstate is allowed to be NULL (early interpreter init) */
    if (tstate != NULL) {
        /* Sub-interpreter support: threads might have been switched
           under our feet using PyThreadState_Swap(). Fix the GIL last
           holder variable so that our heuristics work. */
        _Py_atomic_store_relaxed(&gil->last_holder, (uintptr_t)tstate);
    }

    MUTEX_LOCK(gil->mutex);
    _Py_ANNOTATE_RWLOCK_RELEASED(&gil->locked, /*is_write=*/1);
    _Py_atomic_store_relaxed(&gil->locked, 0);
    COND_SIGNAL(gil->cond);
    MUTEX_UNLOCK(gil->mutex);

#ifdef FORCE_SWITCHING
    if (_Py_atomic_load_relaxed(&ceval2->gil_drop_request) && tstate != NULL) {
        MUTEX_LOCK(gil->switch_mutex);
        /* Not switched yet => wait */
        if (((PyThreadState*)_Py_atomic_load_relaxed(&gil->last_holder)) == tstate)
        {
            assert(is_tstate_valid(tstate));
            RESET_GIL_DROP_REQUEST(tstate->interp);
            /* NOTE: if COND_WAIT does not atomically start waiting when
               releasing the mutex, another thread can run through, take
               the GIL and drop it again, and reset the condition
               before we even had a chance to wait for it. */
            COND_WAIT(gil->switch_cond, gil->switch_mutex);
        }
        MUTEX_UNLOCK(gil->switch_mutex);
    }
#endif
}


/* Check if a Python thread must exit immediately, rather than taking the GIL
   if Py_Finalize() has been called.

   When this function is called by a daemon thread after Py_Finalize() has been
   called, the GIL does no longer exist.

   tstate must be non-NULL. */
static inline int
tstate_must_exit(PyThreadState *tstate)
{
    /* bpo-39877: Access _PyRuntime directly rather than using
       tstate->interp->runtime to support calls from Python daemon threads.
       After Py_Finalize() has been called, tstate can be a dangling pointer:
       point to PyThreadState freed memory. */
    PyThreadState *finalizing = _PyRuntimeState_GetFinalizing(&_PyRuntime);
    return (finalizing != NULL && finalizing != tstate);
}


/* Take the GIL.

   The function saves errno at entry and restores its value at exit.

   tstate must be non-NULL. */
static void
take_gil(PyThreadState *tstate)
{
    int err = errno;

    assert(tstate != NULL);

    if (tstate_must_exit(tstate)) {
        /* bpo-39877: If Py_Finalize() has been called and tstate is not the
           thread which called Py_Finalize(), exit immediately the thread.

           This code path can be reached by a daemon thread after Py_Finalize()
           completes. In this case, tstate is a dangling pointer: points to
           PyThreadState freed memory. */
        PyThread_exit_thread();
    }

    assert(is_tstate_valid(tstate));
    PyInterpreterState *interp = tstate->interp;
    struct _ceval_runtime_state *ceval = &interp->runtime->ceval;
    struct _ceval_state *ceval2 = &interp->ceval;
    struct _gil_runtime_state *gil = &ceval->gil;

    /* Check that _PyEval_InitThreads() was called to create the lock */
    assert(gil_created(gil));

    MUTEX_LOCK(gil->mutex);

    if (!_Py_atomic_load_relaxed(&gil->locked)) {
        goto _ready;
    }

    int drop_requested = 0;
    while (_Py_atomic_load_relaxed(&gil->locked)) {
        unsigned long saved_switchnum = gil->switch_number;

        unsigned long interval = (gil->interval >= 1 ? gil->interval : 1);
        int timed_out = 0;
        COND_TIMED_WAIT(gil->cond, gil->mutex, interval, timed_out);

        /* If we timed out and no switch occurred in the meantime, it is time
           to ask the GIL-holding thread to drop it. */
        if (timed_out &&
            _Py_atomic_load_relaxed(&gil->locked) &&
            gil->switch_number == saved_switchnum)
        {
            if (tstate_must_exit(tstate)) {
                MUTEX_UNLOCK(gil->mutex);
                // gh-96387: If the loop requested a drop request in a previous
                // iteration, reset the request. Otherwise, drop_gil() can
                // block forever waiting for the thread which exited. Drop
                // requests made by other threads are also reset: these threads
                // may have to request again a drop request (iterate one more
                // time).
                if (drop_requested) {
                    RESET_GIL_DROP_REQUEST(interp);
                }
                PyThread_exit_thread();
            }
            assert(is_tstate_valid(tstate));

            SET_GIL_DROP_REQUEST(interp);
            drop_requested = 1;
        }
    }

_ready:
#ifdef FORCE_SWITCHING
    /* This mutex must be taken before modifying gil->last_holder:
       see drop_gil(). */
    MUTEX_LOCK(gil->switch_mutex);
#endif
    /* We now hold the GIL */
    _Py_atomic_store_relaxed(&gil->locked, 1);
    _Py_ANNOTATE_RWLOCK_ACQUIRED(&gil->locked, /*is_write=*/1);

    if (tstate != (PyThreadState*)_Py_atomic_load_relaxed(&gil->last_holder)) {
        _Py_atomic_store_relaxed(&gil->last_holder, (uintptr_t)tstate);
        ++gil->switch_number;
    }

#ifdef FORCE_SWITCHING
    COND_SIGNAL(gil->switch_cond);
    MUTEX_UNLOCK(gil->switch_mutex);
#endif

    if (tstate_must_exit(tstate)) {
        /* bpo-36475: If Py_Finalize() has been called and tstate is not
           the thread which called Py_Finalize(), exit immediately the
           thread.

           This code path can be reached by a daemon thread which was waiting
           in take_gil() while the main thread called
           wait_for_thread_shutdown() from Py_Finalize(). */
        MUTEX_UNLOCK(gil->mutex);
        drop_gil(ceval, ceval2, tstate);
        PyThread_exit_thread();
    }
    assert(is_tstate_valid(tstate));

    if (_Py_atomic_load_relaxed(&ceval2->gil_drop_request)) {
        RESET_GIL_DROP_REQUEST(interp);
    }
    else {
        /* bpo-40010: eval_breaker should be recomputed to be set to 1 if there
           is a pending signal: signal received by another thread which cannot
           handle signals.

           Note: RESET_GIL_DROP_REQUEST() calls COMPUTE_EVAL_BREAKER(). */
        COMPUTE_EVAL_BREAKER(interp, ceval, ceval2);
    }

    /* Don't access tstate if the thread must exit */
    if (tstate->async_exc != NULL) {
        _PyEval_SignalAsyncExc(tstate->interp);
    }

    MUTEX_UNLOCK(gil->mutex);

    errno = err;
}

void _PyEval_SetSwitchInterval(unsigned long microseconds)
{
    struct _gil_runtime_state *gil = &_PyRuntime.ceval.gil;
    gil->interval = microseconds;
}

unsigned long _PyEval_GetSwitchInterval()
{
    struct _gil_runtime_state *gil = &_PyRuntime.ceval.gil;
    return gil->interval;
}


int
_PyEval_ThreadsInitialized(_PyRuntimeState *runtime)
{
    return gil_created(&runtime->ceval.gil);
}

int
PyEval_ThreadsInitialized(void)
{
    _PyRuntimeState *runtime = &_PyRuntime;
    return _PyEval_ThreadsInitialized(runtime);
}

PyStatus
_PyEval_InitGIL(PyThreadState *tstate)
{
    if (!_Py_IsMainInterpreter(tstate->interp)) {
        /* Currently, the GIL is shared by all interpreters,
           and only the main interpreter is responsible to create
           and destroy it. */
        return _PyStatus_OK();
    }

    struct _gil_runtime_state *gil = &tstate->interp->runtime->ceval.gil;
    assert(!gil_created(gil));

    PyThread_init_thread();
    create_gil(gil);

    take_gil(tstate);

    assert(gil_created(gil));
    return _PyStatus_OK();
}

void
_PyEval_FiniGIL(PyInterpreterState *interp)
{
    if (!_Py_IsMainInterpreter(interp)) {
        /* Currently, the GIL is shared by all interpreters,
           and only the main interpreter is responsible to create
           and destroy it. */
        return;
    }

    struct _gil_runtime_state *gil = &interp->runtime->ceval.gil;
    if (!gil_created(gil)) {
        /* First Py_InitializeFromConfig() call: the GIL doesn't exist
           yet: do nothing. */
        return;
    }

    destroy_gil(gil);
    assert(!gil_created(gil));
}

void
PyEval_InitThreads(void)
{
    /* Do nothing: kept for backward compatibility */
}

void
_PyEval_Fini(void)
{
#ifdef Py_STATS
    _Py_PrintSpecializationStats(1);
#endif
}
void
PyEval_AcquireLock(void)
{
    _PyRuntimeState *runtime = &_PyRuntime;
    PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
    _Py_EnsureTstateNotNULL(tstate);

    take_gil(tstate);
}

void
PyEval_ReleaseLock(void)
{
    _PyRuntimeState *runtime = &_PyRuntime;
    PyThreadState *tstate = _PyRuntimeState_GetThreadState(runtime);
    /* This function must succeed when the current thread state is NULL.
       We therefore avoid PyThreadState_Get() which dumps a fatal error
       in debug mode. */
    struct _ceval_runtime_state *ceval = &runtime->ceval;
    struct _ceval_state *ceval2 = &tstate->interp->ceval;
    drop_gil(ceval, ceval2, tstate);
}

void
_PyEval_ReleaseLock(PyThreadState *tstate)
{
    struct _ceval_runtime_state *ceval = &tstate->interp->runtime->ceval;
    struct _ceval_state *ceval2 = &tstate->interp->ceval;
    drop_gil(ceval, ceval2, tstate);
}

void
PyEval_AcquireThread(PyThreadState *tstate)
{
    _Py_EnsureTstateNotNULL(tstate);

    take_gil(tstate);

    struct _gilstate_runtime_state *gilstate = &tstate->interp->runtime->gilstate;
    if (_PyThreadState_Swap(gilstate, tstate) != NULL) {
        Py_FatalError("non-NULL old thread state");
    }
}

void
PyEval_ReleaseThread(PyThreadState *tstate)
{
    assert(is_tstate_valid(tstate));

    _PyRuntimeState *runtime = tstate->interp->runtime;
    PyThreadState *new_tstate = _PyThreadState_Swap(&runtime->gilstate, NULL);
    if (new_tstate != tstate) {
        Py_FatalError("wrong thread state");
    }
    struct _ceval_runtime_state *ceval = &runtime->ceval;
    struct _ceval_state *ceval2 = &tstate->interp->ceval;
    drop_gil(ceval, ceval2, tstate);
}

#ifdef HAVE_FORK
/* This function is called from PyOS_AfterFork_Child to destroy all threads
   which are not running in the child process, and clear internal locks
   which might be held by those threads. */
PyStatus
_PyEval_ReInitThreads(PyThreadState *tstate)
{
    _PyRuntimeState *runtime = tstate->interp->runtime;

    struct _gil_runtime_state *gil = &runtime->ceval.gil;
    if (!gil_created(gil)) {
        return _PyStatus_OK();
    }
    recreate_gil(gil);

    take_gil(tstate);

    struct _pending_calls *pending = &tstate->interp->ceval.pending;
    if (_PyThread_at_fork_reinit(&pending->lock) < 0) {
        return _PyStatus_ERR("Can't reinitialize pending calls lock");
    }

    /* Destroy all threads except the current one */
    _PyThreadState_DeleteExcept(runtime, tstate);
    return _PyStatus_OK();
}
#endif

/* This function is used to signal that async exceptions are waiting to be
   raised. */

void
_PyEval_SignalAsyncExc(PyInterpreterState *interp)
{
    SIGNAL_ASYNC_EXC(interp);
}

PyThreadState *
PyEval_SaveThread(void)
{
    _PyRuntimeState *runtime = &_PyRuntime;
    PyThreadState *tstate = _PyThreadState_Swap(&runtime->gilstate, NULL);
    _Py_EnsureTstateNotNULL(tstate);

    struct _ceval_runtime_state *ceval = &runtime->ceval;
    struct _ceval_state *ceval2 = &tstate->interp->ceval;
    assert(gil_created(&ceval->gil));
    drop_gil(ceval, ceval2, tstate);
    return tstate;
}

void
PyEval_RestoreThread(PyThreadState *tstate)
{
    _Py_EnsureTstateNotNULL(tstate);

    take_gil(tstate);

    struct _gilstate_runtime_state *gilstate = &tstate->interp->runtime->gilstate;
    _PyThreadState_Swap(gilstate, tstate);
}


/* Mechanism whereby asynchronously executing callbacks (e.g. UNIX
   signal handlers or Mac I/O completion routines) can schedule calls
   to a function to be called synchronously.
   The synchronous function is called with one void* argument.
   It should return 0 for success or -1 for failure -- failure should
   be accompanied by an exception.

   If registry succeeds, the registry function returns 0; if it fails
   (e.g. due to too many pending calls) it returns -1 (without setting
   an exception condition).

   Note that because registry may occur from within signal handlers,
   or other asynchronous events, calling malloc() is unsafe!

   Any thread can schedule pending calls, but only the main thread
   will execute them.
   There is no facility to schedule calls to a particular thread, but
   that should be easy to change, should that ever be required.  In
   that case, the static variables here should go into the python
   threadstate.
*/

void
_PyEval_SignalReceived(PyInterpreterState *interp)
{
#ifdef MS_WINDOWS
    // bpo-42296: On Windows, _PyEval_SignalReceived() is called from a signal
    // handler which can run in a thread different than the Python thread, in
    // which case _Py_ThreadCanHandleSignals() is wrong. Ignore
    // _Py_ThreadCanHandleSignals() and always set eval_breaker to 1.
    //
    // The next eval_frame_handle_pending() call will call
    // _Py_ThreadCanHandleSignals() to recompute eval_breaker.
    int force = 1;
#else
    int force = 0;
#endif
    /* bpo-30703: Function called when the C signal handler of Python gets a
       signal. We cannot queue a callback using _PyEval_AddPendingCall() since
       that function is not async-signal-safe. */
    SIGNAL_PENDING_SIGNALS(interp, force);
}

/* Push one item onto the queue while holding the lock. */
static int
_push_pending_call(struct _pending_calls *pending,
                   int (*func)(void *), void *arg)
{
    int i = pending->last;
    int j = (i + 1) % NPENDINGCALLS;
    if (j == pending->first) {
        return -1; /* Queue full */
    }
    pending->calls[i].func = func;
    pending->calls[i].arg = arg;
    pending->last = j;
    return 0;
}

/* Pop one item off the queue while holding the lock. */
static void
_pop_pending_call(struct _pending_calls *pending,
                  int (**func)(void *), void **arg)
{
    int i = pending->first;
    if (i == pending->last) {
        return; /* Queue empty */
    }

    *func = pending->calls[i].func;
    *arg = pending->calls[i].arg;
    pending->first = (i + 1) % NPENDINGCALLS;
}

/* This implementation is thread-safe.  It allows
   scheduling to be made from any thread, and even from an executing
   callback.
 */

int
_PyEval_AddPendingCall(PyInterpreterState *interp,
                       int (*func)(void *), void *arg)
{
    struct _pending_calls *pending = &interp->ceval.pending;
    /* Ensure that _PyEval_InitState() was called
       and that _PyEval_FiniState() is not called yet. */
    assert(pending->lock != NULL);

    PyThread_acquire_lock(pending->lock, WAIT_LOCK);
    int result = _push_pending_call(pending, func, arg);
    PyThread_release_lock(pending->lock);

    /* signal main loop */
    SIGNAL_PENDING_CALLS(interp);
    return result;
}

int
Py_AddPendingCall(int (*func)(void *), void *arg)
{
    /* Best-effort to support subinterpreters and calls with the GIL released.

       First attempt _PyThreadState_GET() since it supports subinterpreters.

       If the GIL is released, _PyThreadState_GET() returns NULL . In this
       case, use PyGILState_GetThisThreadState() which works even if the GIL
       is released.

       Sadly, PyGILState_GetThisThreadState() doesn't support subinterpreters:
       see bpo-10915 and bpo-15751.

       Py_AddPendingCall() doesn't require the caller to hold the GIL. */
    PyThreadState *tstate = _PyThreadState_GET();
    if (tstate == NULL) {
        tstate = PyGILState_GetThisThreadState();
    }

    PyInterpreterState *interp;
    if (tstate != NULL) {
        interp = tstate->interp;
    }
    else {
        /* Last resort: use the main interpreter */
        interp = _PyInterpreterState_Main();
    }
    return _PyEval_AddPendingCall(interp, func, arg);
}

static int
handle_signals(PyThreadState *tstate)
{
    assert(is_tstate_valid(tstate));
    if (!_Py_ThreadCanHandleSignals(tstate->interp)) {
        return 0;
    }

    UNSIGNAL_PENDING_SIGNALS(tstate->interp);
    if (_PyErr_CheckSignalsTstate(tstate) < 0) {
        /* On failure, re-schedule a call to handle_signals(). */
        SIGNAL_PENDING_SIGNALS(tstate->interp, 0);
        return -1;
    }
    return 0;
}

static int
make_pending_calls(PyInterpreterState *interp)
{
    /* only execute pending calls on main thread */
    if (!_Py_ThreadCanHandlePendingCalls()) {
        return 0;
    }

    /* don't perform recursive pending calls */
    if (interp->ceval.pending.busy) {
        return 0;
    }
    interp->ceval.pending.busy = 1;

    /* unsignal before starting to call callbacks, so that any callback
       added in-between re-signals */
    UNSIGNAL_PENDING_CALLS(interp);
    int res = 0;

    /* perform a bounded number of calls, in case of recursion */
    struct _pending_calls *pending = &interp->ceval.pending;
    for (int i=0; i<NPENDINGCALLS; i++) {
        int (*func)(void *) = NULL;
        void *arg = NULL;

        /* pop one item off the queue while holding the lock */
        PyThread_acquire_lock(pending->lock, WAIT_LOCK);
        _pop_pending_call(pending, &func, &arg);
        PyThread_release_lock(pending->lock);

        /* having released the lock, perform the callback */
        if (func == NULL) {
            break;
        }
        res = func(arg);
        if (res) {
            goto error;
        }
    }

    interp->ceval.pending.busy = 0;
    return res;

error:
    interp->ceval.pending.busy = 0;
    SIGNAL_PENDING_CALLS(interp);
    return res;
}

void
_Py_FinishPendingCalls(PyThreadState *tstate)
{
    assert(PyGILState_Check());
    assert(is_tstate_valid(tstate));

    struct _pending_calls *pending = &tstate->interp->ceval.pending;

    if (!_Py_atomic_load_relaxed_int32(&(pending->calls_to_do))) {
        return;
    }

    if (make_pending_calls(tstate->interp) < 0) {
        PyObject *exc, *val, *tb;
        _PyErr_Fetch(tstate, &exc, &val, &tb);
        PyErr_BadInternalCall();
        _PyErr_ChainExceptions(exc, val, tb);
        _PyErr_Print(tstate);
    }
}

/* Py_MakePendingCalls() is a simple wrapper for the sake
   of backward-compatibility. */
int
Py_MakePendingCalls(void)
{
    assert(PyGILState_Check());

    PyThreadState *tstate = _PyThreadState_GET();
    assert(is_tstate_valid(tstate));

    /* Python signal handler doesn't really queue a callback: it only signals
       that a signal was received, see _PyEval_SignalReceived(). */
    int res = handle_signals(tstate);
    if (res != 0) {
        return res;
    }

    res = make_pending_calls(tstate->interp);
    if (res != 0) {
        return res;
    }

    return 0;
}

/* The interpreter's recursion limit */

void
_PyEval_InitRuntimeState(struct _ceval_runtime_state *ceval)
{
    _gil_initialize(&ceval->gil);
}

void
_PyEval_InitState(struct _ceval_state *ceval, PyThread_type_lock pending_lock)
{
    struct _pending_calls *pending = &ceval->pending;
    assert(pending->lock == NULL);

    pending->lock = pending_lock;
}

void
_PyEval_FiniState(struct _ceval_state *ceval)
{
    struct _pending_calls *pending = &ceval->pending;
    if (pending->lock != NULL) {
        PyThread_free_lock(pending->lock);
        pending->lock = NULL;
    }
}

/* Handle signals, pending calls, GIL drop request
   and asynchronous exception */
int
_Py_HandlePending(PyThreadState *tstate)
{
    _PyRuntimeState * const runtime = &_PyRuntime;
    struct _ceval_runtime_state *ceval = &runtime->ceval;
    struct _ceval_state *interp_ceval_state = &tstate->interp->ceval;

    /* Pending signals */
    if (_Py_atomic_load_relaxed_int32(&ceval->signals_pending)) {
        if (handle_signals(tstate) != 0) {
            return -1;
        }
    }

    /* Pending calls */
    if (_Py_atomic_load_relaxed_int32(&interp_ceval_state->pending.calls_to_do)) {
        if (make_pending_calls(tstate->interp) != 0) {
            return -1;
        }
    }

    /* GC scheduled to run */
    if (_Py_atomic_load_relaxed_int32(&interp_ceval_state->gc_scheduled)) {
        _Py_atomic_store_relaxed(&interp_ceval_state->gc_scheduled, 0);
        COMPUTE_EVAL_BREAKER(tstate->interp, ceval, interp_ceval_state);
        _Py_RunGC(tstate);
    }

    /* GIL drop request */
    if (_Py_atomic_load_relaxed_int32(&interp_ceval_state->gil_drop_request)) {
        /* Give another thread a chance */
        if (_PyThreadState_Swap(&runtime->gilstate, NULL) != tstate) {
            Py_FatalError("tstate mix-up");
        }
        drop_gil(ceval, interp_ceval_state, tstate);

        /* Other threads may run now */

        take_gil(tstate);

        if (_PyThreadState_Swap(&runtime->gilstate, tstate) != NULL) {
            Py_FatalError("orphan tstate");
        }
    }

    /* Check for asynchronous exception. */
    if (tstate->async_exc != NULL) {
        PyObject *exc = tstate->async_exc;
        tstate->async_exc = NULL;
        UNSIGNAL_ASYNC_EXC(tstate->interp);
        _PyErr_SetNone(tstate, exc);
        Py_DECREF(exc);
        return -1;
    }


    // It is possible that some of the conditions that trigger the eval breaker
    // are called in a different thread than the Python thread. An example of
    // this is bpo-42296: On Windows, _PyEval_SignalReceived() can be called in
    // a different thread than the Python thread, in which case
    // _Py_ThreadCanHandleSignals() is wrong. Recompute eval_breaker in the
    // current Python thread with the correct _Py_ThreadCanHandleSignals()
    // value. It prevents to interrupt the eval loop at every instruction if
    // the current Python thread cannot handle signals (if
    // _Py_ThreadCanHandleSignals() is false).
    COMPUTE_EVAL_BREAKER(tstate->interp, ceval, interp_ceval_state);

    return 0;
}