{{ message }}
forked from python/cpython
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathceval_macros.h
More file actions
683 lines (603 loc) · 27.3 KB
/
Copy pathceval_macros.h
File metadata and controls
683 lines (603 loc) · 27.3 KB
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
// Macros and other things needed by ceval.c, and bytecodes.c
/* Computed GOTOs, or
the-optimization-commonly-but-improperly-known-as-"threaded code"
using gcc's labels-as-values extension
(http://gcc.gnu.org/onlinedocs/gcc/Labels-as-Values.html).
The traditional bytecode evaluation loop uses a "switch" statement, which
decent compilers will optimize as a single indirect branch instruction
combined with a lookup table of jump addresses. However, since the
indirect jump instruction is shared by all opcodes, the CPU will have a
hard time making the right prediction for where to jump next (actually,
it will be always wrong except in the uncommon case of a sequence of
several identical opcodes).
"Threaded code" in contrast, uses an explicit jump table and an explicit
indirect jump instruction at the end of each opcode. Since the jump
instruction is at a different address for each opcode, the CPU will make a
separate prediction for each of these instructions, which is equivalent to
predicting the second opcode of each opcode pair. These predictions have
a much better chance to turn out valid, especially in small bytecode loops.
A mispredicted branch on a modern CPU flushes the whole pipeline and
can cost several CPU cycles (depending on the pipeline depth),
and potentially many more instructions (depending on the pipeline width).
A correctly predicted branch, however, is nearly free.
At the time of this writing, the "threaded code" version is up to 15-20%
faster than the normal "switch" version, depending on the compiler and the
CPU architecture.
NOTE: care must be taken that the compiler doesn't try to "optimize" the
indirect jumps by sharing them between all opcodes. Such optimizations
can be disabled on gcc by using the -fno-gcse flag (or possibly
-fno-crossjumping).
*/
/* Use macros rather than inline functions, to make it as clear as possible
* to the C compiler that the tracing check is a simple test then branch.
* We want to be sure that the compiler knows this before it generates
* the CFG.
*/
#ifdef WITH_DTRACE
#define OR_DTRACE_LINE | (PyDTrace_LINE_ENABLED() ? 255 : 0)
#else
#define OR_DTRACE_LINE
#endif
#ifdef HAVE_COMPUTED_GOTOS
#ifndef USE_COMPUTED_GOTOS
#define USE_COMPUTED_GOTOS 1
#endif
#else
#if defined(USE_COMPUTED_GOTOS) && USE_COMPUTED_GOTOS
#error "Computed gotos are not supported on this compiler."
#endif
#undef USE_COMPUTED_GOTOS
#define USE_COMPUTED_GOTOS 0
#endif
#ifdef Py_STATS
#define INSTRUCTION_STATS(op) \
do { \
PyStats *s = _PyStats_GET(); \
OPCODE_EXE_INC(op); \
if (s) s->opcode_stats[lastopcode].pair_count[op]++; \
lastopcode = op; \
} while (0)
#else
#define INSTRUCTION_STATS(op) ((void)0)
#endif
#ifdef Py_STATS
# define TAIL_CALL_PARAMS _PyInterpreterFrame *frame, _PyStackRef *stack_pointer, PyThreadState *tstate, _Py_CODEUNIT *next_instr, const void *instruction_funcptr_table, int oparg, int lastopcode
# define TAIL_CALL_ARGS frame, stack_pointer, tstate, next_instr, instruction_funcptr_table, oparg, lastopcode
#else
# define TAIL_CALL_PARAMS _PyInterpreterFrame *frame, _PyStackRef *stack_pointer, PyThreadState *tstate, _Py_CODEUNIT *next_instr, const void *instruction_funcptr_table, int oparg
# define TAIL_CALL_ARGS frame, stack_pointer, tstate, next_instr, instruction_funcptr_table, oparg
#endif
#if _Py_TAIL_CALL_INTERP
# if defined(__clang__) || defined(__GNUC__)
# if !_Py__has_attribute(preserve_none) || !_Py__has_attribute(musttail)
# error "This compiler does not have support for efficient tail calling."
# endif
# elif defined(_MSC_VER) && (_MSC_VER < 1950)
# error "You need at least VS 2026 / PlatformToolset v145 for tail calling."
# endif
# if defined(_MSC_VER) && !defined(__clang__)
# define Py_MUSTTAIL [[msvc::musttail]]
# define Py_PRESERVE_NONE_CC __preserve_none
# else
# define Py_MUSTTAIL __attribute__((musttail))
# define Py_PRESERVE_NONE_CC __attribute__((preserve_none))
# endif
typedef PyObject *(Py_PRESERVE_NONE_CC *py_tail_call_funcptr)(TAIL_CALL_PARAMS);
# define DISPATCH_TABLE_VAR instruction_funcptr_table
# define DISPATCH_TABLE instruction_funcptr_handler_table
# define TRACING_DISPATCH_TABLE instruction_funcptr_tracing_table
# define TARGET(op) Py_NO_INLINE PyObject *Py_PRESERVE_NONE_CC _TAIL_CALL_##op(TAIL_CALL_PARAMS)
# define DISPATCH_GOTO() \
do { \
Py_MUSTTAIL return (((py_tail_call_funcptr *)instruction_funcptr_table)[opcode])(TAIL_CALL_ARGS); \
} while (0)
# define DISPATCH_GOTO_NON_TRACING() \
do { \
Py_MUSTTAIL return (((py_tail_call_funcptr *)DISPATCH_TABLE)[opcode])(TAIL_CALL_ARGS); \
} while (0)
# define JUMP_TO_LABEL(name) \
do { \
Py_MUSTTAIL return (_TAIL_CALL_##name)(TAIL_CALL_ARGS); \
} while (0)
# ifdef Py_STATS
# define JUMP_TO_PREDICTED(name) \
do { \
Py_MUSTTAIL return (_TAIL_CALL_##name)(frame, stack_pointer, tstate, this_instr, instruction_funcptr_table, oparg, lastopcode); \
} while (0)
# else
# define JUMP_TO_PREDICTED(name) \
do { \
Py_MUSTTAIL return (_TAIL_CALL_##name)(frame, stack_pointer, tstate, this_instr, instruction_funcptr_table, oparg); \
} while (0)
# endif
# define LABEL(name) TARGET(name)
#elif USE_COMPUTED_GOTOS
# define DISPATCH_TABLE_VAR opcode_targets
# define DISPATCH_TABLE opcode_targets_table
# define TRACING_DISPATCH_TABLE opcode_tracing_targets_table
# define TARGET(op) TARGET_##op:
# define DISPATCH_GOTO() goto *opcode_targets[opcode]
# define DISPATCH_GOTO_NON_TRACING() goto *DISPATCH_TABLE[opcode];
# define JUMP_TO_LABEL(name) goto name;
# define JUMP_TO_PREDICTED(name) goto PREDICTED_##name;
# define LABEL(name) name:
#else
# define TARGET(op) case op: TARGET_##op:
# define DISPATCH_GOTO() dispatch_code = opcode | tracing_mode ; goto dispatch_opcode
# define DISPATCH_GOTO_NON_TRACING() dispatch_code = opcode; goto dispatch_opcode
# define JUMP_TO_LABEL(name) goto name;
# define JUMP_TO_PREDICTED(name) goto PREDICTED_##name;
# define LABEL(name) name:
#endif
#if (_Py_TAIL_CALL_INTERP || USE_COMPUTED_GOTOS) && _Py_TIER2
# define IS_JIT_TRACING() (DISPATCH_TABLE_VAR == TRACING_DISPATCH_TABLE)
# define ENTER_TRACING() \
DISPATCH_TABLE_VAR = TRACING_DISPATCH_TABLE;
# define LEAVE_TRACING() \
DISPATCH_TABLE_VAR = DISPATCH_TABLE;
#else
# define IS_JIT_TRACING() (tracing_mode != 0)
# define ENTER_TRACING() tracing_mode = 255
# define LEAVE_TRACING() tracing_mode = 0
#endif
#if _Py_TIER2
#define STOP_TRACING() \
do { \
if (IS_JIT_TRACING()) { \
LEAVE_TRACING(); \
_PyJit_FinalizeTracing(tstate, 0); \
} \
} while (0);
#else
#define STOP_TRACING() ((void)(0));
#endif
/* PRE_DISPATCH_GOTO() does lltrace if enabled. Normally a no-op */
#ifdef Py_DEBUG
#define PRE_DISPATCH_GOTO() if (frame->lltrace >= 5) { \
lltrace_instruction(frame, stack_pointer, next_instr, opcode, oparg); }
#else
#define PRE_DISPATCH_GOTO() ((void)0)
#endif
#ifdef Py_DEBUG
#define LLTRACE_RESUME_FRAME() \
do { \
_PyFrame_SetStackPointer(frame, stack_pointer); \
int lltrace = maybe_lltrace_resume_frame(frame, GLOBALS()); \
stack_pointer = _PyFrame_GetStackPointer(frame); \
frame->lltrace = lltrace; \
} while (0)
#else
#define LLTRACE_RESUME_FRAME() ((void)0)
#endif
#ifdef Py_GIL_DISABLED
#define QSBR_QUIESCENT_STATE(tstate) _Py_qsbr_quiescent_state(((_PyThreadStateImpl *)tstate)->qsbr)
#else
#define QSBR_QUIESCENT_STATE(tstate)
#endif
/* Do interpreter dispatch accounting for tracing and instrumentation */
#define DISPATCH() \
{ \
_PyFrame_StackAssertInvalid(frame); \
NEXTOPARG(); \
PRE_DISPATCH_GOTO(); \
DISPATCH_GOTO(); \
}
#define DISPATCH_NON_TRACING() \
{ \
_PyFrame_StackAssertInvalid(frame); \
NEXTOPARG(); \
PRE_DISPATCH_GOTO(); \
DISPATCH_GOTO_NON_TRACING(); \
}
#define DISPATCH_SAME_OPARG() \
{ \
opcode = next_instr->op.code; \
PRE_DISPATCH_GOTO(); \
DISPATCH_GOTO_NON_TRACING(); \
}
#define DISPATCH_INLINED(NEW_FRAME) \
do { \
assert(!IS_PEP523_HOOKED(tstate)); \
_PyFrame_SetStackPointer(frame, stack_pointer); \
_PyFrame_StackPointerValidate(frame); \
assert((NEW_FRAME)->previous == frame); \
frame = tstate->current_frame = (NEW_FRAME); \
CALL_STAT_INC(inlined_py_calls); \
JUMP_TO_LABEL(start_frame); \
} while (0)
/* Tuple access macros */
#ifndef Py_DEBUG
#define GETITEM(v, i) PyTuple_GET_ITEM((v), (i))
#else
static inline PyObject *
GETITEM(PyObject *v, Py_ssize_t i) {
assert(PyTuple_Check(v));
assert(i >= 0);
assert(i < PyTuple_GET_SIZE(v));
return PyTuple_GET_ITEM(v, i);
}
#endif
/* Code access macros */
/* The integer overflow is checked by an assertion below. */
#define INSTR_OFFSET() ((int)(next_instr - _PyFrame_GetBytecode(frame)))
#define NEXTOPARG() do { \
_Py_CODEUNIT word = {.cache = FT_ATOMIC_LOAD_UINT16_RELAXED(*(uint16_t*)next_instr)}; \
opcode = word.op.code; \
oparg = word.op.arg; \
} while (0)
/* JUMPBY makes the generator identify the instruction as a jump. SKIP_OVER is
* for advancing to the next instruction, taking into account cache entries
* and skipped instructions.
*/
#define JUMPBY(x) (next_instr += (x))
#define SKIP_OVER(x) (next_instr += (x))
#define STACK_LEVEL() ((int)(stack_pointer - _PyFrame_Stackbase(frame)))
#define STACK_SIZE() (_PyFrame_GetCode(frame)->co_stacksize)
#define WITHIN_STACK_BOUNDS() \
(frame->owner == FRAME_OWNED_BY_INTERPRETER || (STACK_LEVEL() >= 0 && STACK_LEVEL() <= STACK_SIZE()))
#if defined(Py_DEBUG) && !defined(_Py_JIT)
// This allows temporary stack "overflows", provided it's all in the cache at any point of time.
#define ASSERT_WITHIN_STACK_BOUNDS_IGNORING_CACHE(F, L) \
assert(frame->owner == FRAME_OWNED_BY_INTERPRETER || (STACK_LEVEL() >= 0 && (STACK_LEVEL()) <= STACK_SIZE()))
#else
#define ASSERT_WITHIN_STACK_BOUNDS_IGNORING_CACHE ASSERT_WITHIN_STACK_BOUNDS
#endif
/* Data access macros */
#define FRAME_CO_CONSTS (_PyFrame_GetCode(frame)->co_consts)
#define FRAME_CO_NAMES (_PyFrame_GetCode(frame)->co_names)
/* Local variable macros */
#define LOCALS_ARRAY (frame->localsplus)
#define GETLOCAL(i) (frame->localsplus[i])
#ifdef Py_STATS
#define UPDATE_MISS_STATS(INSTNAME) \
do { \
STAT_INC(opcode, miss); \
STAT_INC((INSTNAME), miss); \
/* The counter is always the first cache entry: */ \
if (ADAPTIVE_COUNTER_TRIGGERS(next_instr->cache)) { \
STAT_INC((INSTNAME), deopt); \
} \
} while (0)
#else
#define UPDATE_MISS_STATS(INSTNAME) ((void)0)
#endif
// Try to lock an object in the free threading build, if it's not already
// locked. Use with a DEOPT_IF() to deopt if the object is already locked.
// These are no-ops in the default GIL build. The general pattern is:
//
// DEOPT_IF(!LOCK_OBJECT(op));
// if (/* condition fails */) {
// UNLOCK_OBJECT(op);
// DEOPT_IF(true);
// }
// ...
// UNLOCK_OBJECT(op);
//
// NOTE: The object must be unlocked on every exit code path and you should
// avoid any potentially escaping calls (like PyStackRef_CLOSE) while the
// object is locked.
#ifdef Py_GIL_DISABLED
# define LOCK_OBJECT(op) PyMutex_LockFast(&(_PyObject_CAST(op))->ob_mutex)
# define UNLOCK_OBJECT(op) PyMutex_Unlock(&(_PyObject_CAST(op))->ob_mutex)
#else
# define LOCK_OBJECT(op) (1)
# define UNLOCK_OBJECT(op) ((void)0)
#endif
#define GLOBALS() frame->f_globals
#define BUILTINS() frame->f_builtins
#define LOCALS() frame->f_locals
#define CONSTS() _PyFrame_GetCode(frame)->co_consts
#define NAMES() _PyFrame_GetCode(frame)->co_names
#if defined(WITH_DTRACE) && !defined(Py_BUILD_CORE_MODULE)
static void dtrace_function_entry(_PyInterpreterFrame *);
static void dtrace_function_return(_PyInterpreterFrame *);
#define DTRACE_FUNCTION_ENTRY() \
if (PyDTrace_FUNCTION_ENTRY_ENABLED()) { \
dtrace_function_entry(frame); \
}
#define DTRACE_FUNCTION_RETURN() \
if (PyDTrace_FUNCTION_RETURN_ENABLED()) { \
dtrace_function_return(frame); \
}
#else
#define DTRACE_FUNCTION_ENTRY() ((void)0)
#define DTRACE_FUNCTION_RETURN() ((void)0)
#endif
/* This takes a uint16_t instead of a _Py_BackoffCounter,
* because it is used directly on the cache entry in generated code,
* which is always an integral type. */
// Force re-specialization when tracing a side exit to get good side exits.
#define ADAPTIVE_COUNTER_TRIGGERS(COUNTER) \
backoff_counter_triggers(forge_backoff_counter((COUNTER)))
#ifdef Py_GIL_DISABLED
/* Counters are unreachable when thread-local bytecode is disabled,
* so there is no need to update them. */
#define ADVANCE_ADAPTIVE_COUNTER(COUNTER) \
do { \
_Py_BackoffCounter cnt = (COUNTER); \
if (!backoff_counter_is_unreachable(cnt)) { \
(COUNTER) = advance_backoff_counter(cnt); \
} \
} while (0);
#define PAUSE_ADAPTIVE_COUNTER(COUNTER) \
do { \
_Py_BackoffCounter cnt = (COUNTER); \
if (!backoff_counter_is_unreachable(cnt)) { \
(COUNTER) = pause_backoff_counter(cnt); \
} \
} while (0);
#else
#define ADVANCE_ADAPTIVE_COUNTER(COUNTER) \
do { \
(COUNTER) = advance_backoff_counter((COUNTER)); \
} while (0);
#define PAUSE_ADAPTIVE_COUNTER(COUNTER) \
do { \
(COUNTER) = pause_backoff_counter((COUNTER)); \
} while (0);
#endif
#ifdef ENABLE_SPECIALIZATION
/* Multiple threads may execute these concurrently if thread-local bytecode is
* disabled and they all execute the main copy of the bytecode. Specialization
* is disabled in that case so the value is unused, but the RMW cycle should be
* free of data races.
*/
#define RECORD_BRANCH_TAKEN(bitset, flag) \
FT_ATOMIC_STORE_UINT16_RELAXED( \
bitset, (FT_ATOMIC_LOAD_UINT16_RELAXED(bitset) << 1) | (flag))
#else
#define RECORD_BRANCH_TAKEN(bitset, flag)
#endif
#define UNBOUNDLOCAL_ERROR_MSG \
"cannot access local variable '%s' where it is not associated with a value"
#define UNBOUNDFREE_ERROR_MSG \
"cannot access free variable '%s' where it is not associated with a value" \
" in enclosing scope"
#define NAME_ERROR_MSG "name '%.200s' is not defined"
// If a trace function sets a new f_lineno and
// *then* raises, we use the destination when searching
// for an exception handler, displaying the traceback, and so on
#define INSTRUMENTED_JUMP(src, dest, event) \
do { \
_Py_CODEUNIT *_dest = (dest); \
if (tstate->tracing) {\
next_instr = _dest; \
} else { \
_PyFrame_SetStackPointer(frame, stack_pointer); \
next_instr = _Py_call_instrumentation_jump(this_instr, tstate, event, frame, src, _dest); \
stack_pointer = _PyFrame_GetStackPointer(frame); \
if (next_instr == NULL) { \
next_instr = _dest + 1; \
JUMP_TO_LABEL(error); \
} \
} \
} while (0);
static inline int _Py_EnterRecursivePy(PyThreadState *tstate) {
return (tstate->py_recursion_remaining-- <= 0) &&
_Py_CheckRecursiveCallPy(tstate);
}
static inline void _Py_LeaveRecursiveCallPy(PyThreadState *tstate) {
tstate->py_recursion_remaining++;
}
/* Implementation of "macros" that modify the instruction pointer,
* stack pointer, or frame pointer.
* These need to treated differently by tier 1 and 2.
* The Tier 1 version is here; Tier 2 is inlined in ceval.c. */
#define LOAD_IP(OFFSET) do { \
next_instr = frame->instr_ptr + (OFFSET); \
} while (0)
/* There's no STORE_IP(), it's inlined by the code generator. */
#define LOAD_SP() \
stack_pointer = _PyFrame_GetStackPointer(frame)
#define SAVE_SP() \
_PyFrame_SetStackPointer(frame, stack_pointer)
/* Tier-switching macros. */
#define TIER1_TO_TIER2(EXECUTOR) \
do { \
OPT_STAT_INC(traces_executed); \
next_instr = _Py_jit_entry((EXECUTOR), frame, stack_pointer, tstate); \
frame = tstate->current_frame; \
stack_pointer = _PyFrame_GetStackPointer(frame); \
int keep_tracing_bit = (uintptr_t)next_instr & 1; \
next_instr = (_Py_CODEUNIT *)(((uintptr_t)next_instr) & (~1)); \
if (next_instr == NULL) { \
/* gh-140104: The exception handler expects frame->instr_ptr
to after this_instr, not this_instr! */ \
next_instr = frame->instr_ptr + 1; \
JUMP_TO_LABEL(error); \
} \
if (keep_tracing_bit) { \
assert(uop_buffer_length(&((_PyThreadStateImpl *)tstate)->jit_tracer_state->code_buffer)); \
ENTER_TRACING(); \
DISPATCH_NON_TRACING(); \
} \
DISPATCH(); \
} while (0)
#define TIER2_TO_TIER2(EXECUTOR) \
do { \
OPT_STAT_INC(traces_executed); \
current_executor = (EXECUTOR); \
goto tier2_start; \
} while (0)
#define GOTO_TIER_ONE_SETUP \
tstate->current_executor = NULL; \
OPT_HIST(trace_uop_execution_counter, trace_run_length_hist); \
_PyFrame_SetStackPointer(frame, stack_pointer);
#define GOTO_TIER_ONE(TARGET) \
do \
{ \
GOTO_TIER_ONE_SETUP \
return (_Py_CODEUNIT *)(TARGET); \
} while (0)
#define GOTO_TIER_ONE_CONTINUE_TRACING(TARGET) \
do \
{ \
GOTO_TIER_ONE_SETUP \
return (_Py_CODEUNIT *)(((uintptr_t)(TARGET))| 1); \
} while (0)
#define CURRENT_OPARG() (next_uop[-1].oparg)
#define CURRENT_OPERAND0_64() (next_uop[-1].operand0)
#define CURRENT_OPERAND1_64() (next_uop[-1].operand1)
#define CURRENT_OPERAND0_32() (next_uop[-1].operand0)
#define CURRENT_OPERAND1_32() (next_uop[-1].operand1)
#define CURRENT_OPERAND0_16() (next_uop[-1].operand0)
#define CURRENT_OPERAND1_16() (next_uop[-1].operand1)
#define CURRENT_TARGET() (next_uop[-1].target)
#define JUMP_TO_JUMP_TARGET() goto jump_to_jump_target
#define JUMP_TO_ERROR() goto jump_to_error_target
/* Stackref macros */
/* How much scratch space to give stackref to PyObject* conversion. */
#define MAX_STACKREF_SCRATCH 10
#define STACKREFS_TO_PYOBJECTS(ARGS, ARG_COUNT, NAME) \
/* +1 because vectorcall might use -1 to write self */ \
PyObject *NAME##_temp[MAX_STACKREF_SCRATCH+1]; \
PyObject **NAME = _PyObjectArray_FromStackRefArray(ARGS, ARG_COUNT, NAME##_temp);
#define STACKREFS_TO_PYOBJECTS_CLEANUP(NAME) \
/* +1 because we +1 previously */ \
_PyObjectArray_Free(NAME - 1, NAME##_temp);
#define CONVERSION_FAILED(NAME) ((NAME) == NULL)
#if defined(Py_DEBUG) && !defined(_Py_JIT)
#define SET_CURRENT_CACHED_VALUES(N) current_cached_values = (N)
#define CHECK_CURRENT_CACHED_VALUES(N) assert(current_cached_values == (N))
#else
#define SET_CURRENT_CACHED_VALUES(N) ((void)0)
#define CHECK_CURRENT_CACHED_VALUES(N) ((void)0)
#endif
#define IS_PEP523_HOOKED(tstate) (tstate->interp->eval_frame != NULL)
static inline int
check_periodics(PyThreadState *tstate) {
_Py_CHECK_EMSCRIPTEN_SIGNALS_PERIODICALLY();
QSBR_QUIESCENT_STATE(tstate);
if (_Py_atomic_load_uintptr_relaxed(&tstate->eval_breaker) & _PY_EVAL_EVENTS_MASK) {
return _Py_HandlePending(tstate);
}
return 0;
}
static inline int
check_periodics_at_end(PyThreadState *tstate, _PyInterpreterFrame *frame) {
_Py_CHECK_EMSCRIPTEN_SIGNALS_PERIODICALLY();
QSBR_QUIESCENT_STATE(tstate);
if (_Py_atomic_load_uintptr_relaxed(&tstate->eval_breaker) & _PY_EVAL_EVENTS_MASK) {
// Do not handle pending interrupts if the previous instruction was LOAD_SPECIAL
// This may also not handle interrupts if a cache looks like LOAD_SPECIAL,
// but this is benign as we won't skip periodic checks indefinitely.
if (frame->instr_ptr[-1].op.code == LOAD_SPECIAL) {
return 0;
}
return _Py_HandlePending(tstate);
}
return 0;
}
// Mark the generator as executing. Returns true if the state was changed,
// false if it was already executing or finished.
static inline bool
gen_try_set_executing(PyGenObject *gen)
{
#ifdef Py_GIL_DISABLED
if (!_PyObject_IsUniquelyReferenced((PyObject *)gen)) {
int8_t frame_state = _Py_atomic_load_int8_relaxed(&gen->gi_frame_state);
while (frame_state < FRAME_SUSPENDED_YIELD_FROM_LOCKED) {
if (_Py_atomic_compare_exchange_int8(&gen->gi_frame_state,
&frame_state,
FRAME_EXECUTING)) {
return true;
}
}
// NB: We return false for FRAME_SUSPENDED_YIELD_FROM_LOCKED as well.
// That case is rare enough that we can just handle it in the deopt.
return false;
}
#endif
// Use faster non-atomic modifications in the GIL-enabled build and when
// the object is uniquely referenced in the free-threaded build.
if (gen->gi_frame_state < FRAME_EXECUTING) {
assert(gen->gi_frame_state != FRAME_SUSPENDED_YIELD_FROM_LOCKED);
gen->gi_frame_state = FRAME_EXECUTING;
return true;
}
return false;
}
// Macro for inplace float binary ops (tier 2 only).
// Mutates the uniquely-referenced TARGET operand in place.
// TARGET must be either left or right.
#define FLOAT_INPLACE_OP(left, right, TARGET, OP) \
do { \
PyObject *left_o = PyStackRef_AsPyObjectBorrow(left); \
PyObject *right_o = PyStackRef_AsPyObjectBorrow(right); \
assert(PyFloat_CheckExact(left_o)); \
assert(PyFloat_CheckExact(right_o)); \
assert(_PyObject_IsUniquelyReferenced( \
PyStackRef_AsPyObjectBorrow(TARGET))); \
STAT_INC(BINARY_OP, hit); \
double _dres = \
((PyFloatObject *)left_o)->ob_fval \
OP ((PyFloatObject *)right_o)->ob_fval; \
((PyFloatObject *)PyStackRef_AsPyObjectBorrow(TARGET)) \
->ob_fval = _dres; \
} while (0)
// Inplace float true division. Sets _divop_err to 1 on zero division.
// Caller must check _divop_err and call ERROR_NO_POP() if set.
#define FLOAT_INPLACE_DIVOP(left, right, TARGET) \
int _divop_err = 0; \
do { \
PyObject *left_o = PyStackRef_AsPyObjectBorrow(left); \
PyObject *right_o = PyStackRef_AsPyObjectBorrow(right); \
assert(PyFloat_CheckExact(left_o)); \
assert(PyFloat_CheckExact(right_o)); \
assert(_PyObject_IsUniquelyReferenced( \
PyStackRef_AsPyObjectBorrow(TARGET))); \
STAT_INC(BINARY_OP, hit); \
double _divisor = ((PyFloatObject *)right_o)->ob_fval; \
if (_divisor == 0.0) { \
PyErr_SetString(PyExc_ZeroDivisionError, \
"float division by zero"); \
_divop_err = 1; \
break; \
} \
double _dres = ((PyFloatObject *)left_o)->ob_fval / _divisor; \
((PyFloatObject *)PyStackRef_AsPyObjectBorrow(TARGET)) \
->ob_fval = _dres; \
} while (0)
// Inplace compact int operation. TARGET is expected to be uniquely
// referenced at the optimizer level, but at runtime it may be a
// cached small int singleton. We check _Py_IsImmortal on TARGET
// to decide whether inplace mutation is safe.
//
// After the macro, _int_inplace_res holds the result (may be NULL
// on allocation failure). On success, TARGET was mutated in place
// and _int_inplace_res is a DUP'd reference to it. On fallback
// (small int target, small int result, or overflow), _int_inplace_res
// is from FUNC (_PyCompactLong_Add etc.).
// FUNC is the fallback function (_PyCompactLong_Add etc.)
#define INT_INPLACE_OP(left, right, TARGET, OP, FUNC) \
_PyStackRef _int_inplace_res = PyStackRef_NULL; \
do { \
PyObject *target_o = PyStackRef_AsPyObjectBorrow(TARGET); \
if (_Py_IsImmortal(target_o)) { \
break; \
} \
assert(_PyObject_IsUniquelyReferenced(target_o)); \
Py_ssize_t left_val = _PyLong_CompactValue( \
(PyLongObject *)PyStackRef_AsPyObjectBorrow(left)); \
Py_ssize_t right_val = _PyLong_CompactValue( \
(PyLongObject *)PyStackRef_AsPyObjectBorrow(right)); \
Py_ssize_t result = left_val OP right_val; \
if (!_PY_IS_SMALL_INT(result) \
&& ((twodigits)((stwodigits)result) + PyLong_MASK \
< (twodigits)PyLong_MASK + PyLong_BASE)) \
{ \
_PyLong_SetSignAndDigitCount( \
(PyLongObject *)target_o, result < 0 ? -1 : 1, 1); \
((PyLongObject *)target_o)->long_value.ob_digit[0] = \
(digit)(result < 0 ? -result : result); \
_int_inplace_res = PyStackRef_DUP(TARGET); \
break; \
} \
} while (0); \
if (PyStackRef_IsNull(_int_inplace_res)) { \
_int_inplace_res = FUNC( \
(PyLongObject *)PyStackRef_AsPyObjectBorrow(left), \
(PyLongObject *)PyStackRef_AsPyObjectBorrow(right)); \
}
#define CALL_TP_ITERITEM_NO_ESCAPE(ITER, INDEX) \
Py_TYPE(ITER)->_tp_iteritem((ITER), (INDEX))
You can’t perform that action at this time.