mirrors/cpython
1
0
Fork 0
mirror of https://github.com/python/cpython.git synced 2025-08-04 17:08:35 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 13

GH-104584: Fix ENTER_EXECUTOR (GH-106141)

Browse source 
* Check eval-breaker in ENTER_EXECUTOR.

* Make sure that frame->prev_instr is set before entering executor.
This commit is contained in:
Mark Shannon 2023-07-03 21:28:27 +01:00 • committed by GitHub
parent 7f4c8121db
commit e5862113dd
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
8 changed files with 272 additions and 269 deletions
Download patch file Download diff file Expand all files Collapse all files

2
Include/internal/pycore_ceval.h
View file

@ -152,6 +152,8 @@ extern struct _PyInterpreterFrame* _PyEval_GetFrame(void);
extern PyObject* _Py_MakeCoro(PyFunctionObject *func);
/* Handle signals, pending calls, GIL drop request
and asynchronous exception */
extern int _Py_HandlePending(PyThreadState *tstate);

19
Python/bytecodes.c
View file

@ -141,8 +141,8 @@ dummy_func(
ERROR_IF(err, error);
next_instr--;
}
else if (_Py_atomic_load_relaxed_int32(&tstate->interp->ceval.eval_breaker) && oparg < 2) {
goto handle_eval_breaker;
else if (oparg < 2) {
CHECK_EVAL_BREAKER();
}
}
@ -158,6 +158,9 @@ dummy_func(
next_instr--;
}
else {
if (oparg < 2) {
CHECK_EVAL_BREAKER();
}
_PyFrame_SetStackPointer(frame, stack_pointer);
int err = _Py_call_instrumentation(
tstate, oparg > 0, frame, next_instr-1);
@ -168,9 +171,6 @@ dummy_func(
next_instr = frame->prev_instr;
DISPATCH();
}
if (_Py_atomic_load_relaxed_int32(&tstate->interp->ceval.eval_breaker) && oparg < 2) {
goto handle_eval_breaker;
}
}
}
@ -2223,6 +2223,7 @@ dummy_func(
}
inst(JUMP_BACKWARD, (--)) {
CHECK_EVAL_BREAKER();
_Py_CODEUNIT *here = next_instr - 1;
assert(oparg <= INSTR_OFFSET());
JUMPBY(1-oparg);
@ -2240,7 +2241,6 @@ dummy_func(
goto resume_frame;
}
#endif /* ENABLE_SPECIALIZATION */
CHECK_EVAL_BREAKER();
}
pseudo(JUMP) = {
@ -2254,8 +2254,13 @@ dummy_func(
};
inst(ENTER_EXECUTOR, (--)) {
CHECK_EVAL_BREAKER();
PyCodeObject *code = _PyFrame_GetCode(frame);
_PyExecutorObject *executor = (_PyExecutorObject *)code->co_executors->executors[oparg&255];
int original_oparg = executor->vm_data.oparg | (oparg & 0xfffff00);
JUMPBY(1-original_oparg);
frame->prev_instr = next_instr - 1;
Py_INCREF(executor);
frame = executor->execute(executor, frame, stack_pointer);
if (frame == NULL) {
@ -3570,8 +3575,8 @@ dummy_func(
}
inst(INSTRUMENTED_JUMP_BACKWARD, ( -- )) {
INSTRUMENTED_JUMP(next_instr-1, next_instr+1-oparg, PY_MONITORING_EVENT_JUMP);
CHECK_EVAL_BREAKER();
INSTRUMENTED_JUMP(next_instr-1, next_instr+1-oparg, PY_MONITORING_EVENT_JUMP);
}
inst(INSTRUMENTED_POP_JUMP_IF_TRUE, ( -- )) {

70
Python/ceval.c
View file

@ -763,68 +763,6 @@ resume_frame:
DISPATCH();
handle_eval_breaker:
/* Do periodic things, like check for signals and async I/0.
* We need to do reasonably frequently, but not too frequently.
* All loops should include a check of the eval breaker.
* We also check on return from any builtin function.
*
* ## More Details ###
*
* The eval loop (this function) normally executes the instructions
* of a code object sequentially. However, the runtime supports a
* number of out-of-band execution scenarios that may pause that
* sequential execution long enough to do that out-of-band work
* in the current thread using the current PyThreadState.
*
* The scenarios include:
*
* - cyclic garbage collection
* - GIL drop requests
* - "async" exceptions
* - "pending calls" (some only in the main thread)
* - signal handling (only in the main thread)
*
* When the need for one of the above is detected, the eval loop
* pauses long enough to handle the detected case. Then, if doing
* so didn't trigger an exception, the eval loop resumes executing
* the sequential instructions.
*
* To make this work, the eval loop periodically checks if any
* of the above needs to happen. The individual checks can be
* expensive if computed each time, so a while back we switched
* to using pre-computed, per-interpreter variables for the checks,
* and later consolidated that to a single "eval breaker" variable
* (now a PyInterpreterState field).
*
* For the longest time, the eval breaker check would happen
* frequently, every 5 or so times through the loop, regardless
* of what instruction ran last or what would run next. Then, in
* early 2021 (gh-18334, commit 4958f5d), we switched to checking
* the eval breaker less frequently, by hard-coding the check to
* specific places in the eval loop (e.g. certain instructions).
* The intent then was to check after returning from calls
* and on the back edges of loops.
*
* In addition to being more efficient, that approach keeps
* the eval loop from running arbitrary code between instructions
* that don't handle that well. (See gh-74174.)
*
* Currently, the eval breaker check happens here at the
* "handle_eval_breaker" label. Some instructions come here
* explicitly (goto) and some indirectly. Notably, the check
* happens on back edges in the control flow graph, which
* pretty much applies to all loops and most calls.
* (See bytecodes.c for exact information.)
*
* One consequence of this approach is that it might not be obvious
* how to force any specific thread to pick up the eval breaker,
* or for any specific thread to not pick it up. Mostly this
* involves judicious uses of locks and careful ordering of code,
* while avoiding code that might trigger the eval breaker
* until so desired.
*/
if (_Py_HandlePending(tstate) != 0) {
goto error;
}
@ -2796,13 +2734,7 @@ _PyUopExecute(_PyExecutorObject *executor, _PyInterpreterFrame *frame, PyObject
PyThreadState *tstate = _PyThreadState_GET();
_PyUOpExecutorObject *self = (_PyUOpExecutorObject *)executor;
// Equivalent to CHECK_EVAL_BREAKER()
_Py_CHECK_EMSCRIPTEN_SIGNALS_PERIODICALLY();
if (_Py_atomic_load_relaxed_int32(&tstate->interp->ceval.eval_breaker)) {
if (_Py_HandlePending(tstate) != 0) {
goto error;
}
}
CHECK_EVAL_BREAKER();
OBJECT_STAT_INC(optimization_traces_executed);
_Py_CODEUNIT *ip_offset = (_Py_CODEUNIT *)_PyFrame_GetCode(frame)->co_code_adaptive;

61
Python/ceval_gil.c
View file

@ -1052,8 +1052,65 @@ _PyEval_FiniState(struct _ceval_state *ceval)
}
}
/* Handle signals, pending calls, GIL drop request
and asynchronous exception */
/* Do periodic things, like check for signals and async I/0.
* We need to do reasonably frequently, but not too frequently.
* All loops should include a check of the eval breaker.
* We also check on return from any builtin function.
*
* ## More Details ###
*
* The eval loop (this function) normally executes the instructions
* of a code object sequentially. However, the runtime supports a
* number of out-of-band execution scenarios that may pause that
* sequential execution long enough to do that out-of-band work
* in the current thread using the current PyThreadState.
*
* The scenarios include:
*
* - cyclic garbage collection
* - GIL drop requests
* - "async" exceptions
* - "pending calls" (some only in the main thread)
* - signal handling (only in the main thread)
*
* When the need for one of the above is detected, the eval loop
* pauses long enough to handle the detected case. Then, if doing
* so didn't trigger an exception, the eval loop resumes executing
* the sequential instructions.
*
* To make this work, the eval loop periodically checks if any
* of the above needs to happen. The individual checks can be
* expensive if computed each time, so a while back we switched
* to using pre-computed, per-interpreter variables for the checks,
* and later consolidated that to a single "eval breaker" variable
* (now a PyInterpreterState field).
*
* For the longest time, the eval breaker check would happen
* frequently, every 5 or so times through the loop, regardless
* of what instruction ran last or what would run next. Then, in
* early 2021 (gh-18334, commit 4958f5d), we switched to checking
* the eval breaker less frequently, by hard-coding the check to
* specific places in the eval loop (e.g. certain instructions).
* The intent then was to check after returning from calls
* and on the back edges of loops.
*
* In addition to being more efficient, that approach keeps
* the eval loop from running arbitrary code between instructions
* that don't handle that well. (See gh-74174.)
*
* Currently, the eval breaker check happens on back edges in
* the control flow graph, which pretty much applies to all loops,
* and most calls.
* (See bytecodes.c for exact information.)
*
* One consequence of this approach is that it might not be obvious
* how to force any specific thread to pick up the eval breaker,
* or for any specific thread to not pick it up. Mostly this
* involves judicious uses of locks and careful ordering of code,
* while avoiding code that might trigger the eval breaker
* until so desired.
*/
int
_Py_HandlePending(PyThreadState *tstate)
{

4
Python/ceval_macros.h
View file

@ -117,7 +117,9 @@
#define CHECK_EVAL_BREAKER() \
_Py_CHECK_EMSCRIPTEN_SIGNALS_PERIODICALLY(); \
if (_Py_atomic_load_relaxed_int32(&tstate->interp->ceval.eval_breaker)) { \
goto handle_eval_breaker; \
if (_Py_HandlePending(tstate) != 0) { \
goto error; \
} \
}

44
Python/executor_cases.c.h generated
View file

@ -1629,7 +1629,7 @@
case GET_LEN: {
PyObject *obj = stack_pointer[-1];
PyObject *len_o;
#line 2304 "Python/bytecodes.c"
#line 2309 "Python/bytecodes.c"
// PUSH(len(TOS))
Py_ssize_t len_i = PyObject_Length(obj);
if (len_i < 0) goto error;
@ -1646,7 +1646,7 @@
PyObject *type = stack_pointer[-2];
PyObject *subject = stack_pointer[-3];
PyObject *attrs;
#line 2312 "Python/bytecodes.c"
#line 2317 "Python/bytecodes.c"
// Pop TOS and TOS1. Set TOS to a tuple of attributes on success, or
// None on failure.
assert(PyTuple_CheckExact(names));
@ -1655,7 +1655,7 @@
Py_DECREF(subject);
Py_DECREF(type);
Py_DECREF(names);
#line 2317 "Python/bytecodes.c"
#line 2322 "Python/bytecodes.c"
if (attrs) {
assert(PyTuple_CheckExact(attrs)); // Success!
}
@ -1672,7 +1672,7 @@
case MATCH_MAPPING: {
PyObject *subject = stack_pointer[-1];
PyObject *res;
#line 2327 "Python/bytecodes.c"
#line 2332 "Python/bytecodes.c"
int match = Py_TYPE(subject)->tp_flags & Py_TPFLAGS_MAPPING;
res = match ? Py_True : Py_False;
#line 1679 "Python/executor_cases.c.h"
@ -1684,7 +1684,7 @@
case MATCH_SEQUENCE: {
PyObject *subject = stack_pointer[-1];
PyObject *res;
#line 2332 "Python/bytecodes.c"
#line 2337 "Python/bytecodes.c"
int match = Py_TYPE(subject)->tp_flags & Py_TPFLAGS_SEQUENCE;
res = match ? Py_True : Py_False;
#line 1691 "Python/executor_cases.c.h"
@ -1697,7 +1697,7 @@
PyObject *keys = stack_pointer[-1];
PyObject *subject = stack_pointer[-2];
PyObject *values_or_none;
#line 2337 "Python/bytecodes.c"
#line 2342 "Python/bytecodes.c"
// On successful match, PUSH(values). Otherwise, PUSH(None).
values_or_none = match_keys(tstate, subject, keys);
if (values_or_none == NULL) goto error;
@ -1710,12 +1710,12 @@
case GET_ITER: {
PyObject *iterable = stack_pointer[-1];
PyObject *iter;
#line 2343 "Python/bytecodes.c"
#line 2348 "Python/bytecodes.c"
/* before: [obj]; after [getiter(obj)] */
iter = PyObject_GetIter(iterable);
#line 1717 "Python/executor_cases.c.h"
Py_DECREF(iterable);
#line 2346 "Python/bytecodes.c"
#line 2351 "Python/bytecodes.c"
if (iter == NULL) goto pop_1_error;
#line 1721 "Python/executor_cases.c.h"
stack_pointer[-1] = iter;
@ -1725,7 +1725,7 @@
case GET_YIELD_FROM_ITER: {
PyObject *iterable = stack_pointer[-1];
PyObject *iter;
#line 2350 "Python/bytecodes.c"
#line 2355 "Python/bytecodes.c"
/* before: [obj]; after [getiter(obj)] */
if (PyCoro_CheckExact(iterable)) {
/* `iterable` is a coroutine */
@ -1750,7 +1750,7 @@
}
#line 1752 "Python/executor_cases.c.h"
Py_DECREF(iterable);
#line 2373 "Python/bytecodes.c"
#line 2378 "Python/bytecodes.c"
}
#line 1756 "Python/executor_cases.c.h"
stack_pointer[-1] = iter;
@ -1762,7 +1762,7 @@
PyObject *lasti = stack_pointer[-3];
PyObject *exit_func = stack_pointer[-4];
PyObject *res;
#line 2605 "Python/bytecodes.c"
#line 2610 "Python/bytecodes.c"
/* At the top of the stack are 4 values:
- val: TOP = exc_info()
- unused: SECOND = previous exception
@ -1792,7 +1792,7 @@
case PUSH_EXC_INFO: {
PyObject *new_exc = stack_pointer[-1];
PyObject *prev_exc;
#line 2644 "Python/bytecodes.c"
#line 2649 "Python/bytecodes.c"
_PyErr_StackItem *exc_info = tstate->exc_info;
if (exc_info->exc_value != NULL) {
prev_exc = exc_info->exc_value;
@ -1811,7 +1811,7 @@
case EXIT_INIT_CHECK: {
PyObject *should_be_none = stack_pointer[-1];
#line 3013 "Python/bytecodes.c"
#line 3018 "Python/bytecodes.c"
assert(STACK_LEVEL() == 2);
if (should_be_none != Py_None) {
PyErr_Format(PyExc_TypeError,
@ -1827,7 +1827,7 @@
case MAKE_FUNCTION: {
PyObject *codeobj = stack_pointer[-1];
PyObject *func;
#line 3427 "Python/bytecodes.c"
#line 3432 "Python/bytecodes.c"
PyFunctionObject *func_obj = (PyFunctionObject *)
PyFunction_New(codeobj, GLOBALS());
@ -1847,7 +1847,7 @@
case SET_FUNCTION_ATTRIBUTE: {
PyObject *func = stack_pointer[-1];
PyObject *attr = stack_pointer[-2];
#line 3441 "Python/bytecodes.c"
#line 3446 "Python/bytecodes.c"
assert(PyFunction_Check(func));
PyFunctionObject *func_obj = (PyFunctionObject *)func;
switch(oparg) {
@ -1883,13 +1883,13 @@
PyObject *stop = stack_pointer[-(1 + ((oparg == 3) ? 1 : 0))];
PyObject *start = stack_pointer[-(2 + ((oparg == 3) ? 1 : 0))];
PyObject *slice;
#line 3491 "Python/bytecodes.c"
#line 3496 "Python/bytecodes.c"
slice = PySlice_New(start, stop, step);
#line 1889 "Python/executor_cases.c.h"
Py_DECREF(start);
Py_DECREF(stop);
Py_XDECREF(step);
#line 3493 "Python/bytecodes.c"
#line 3498 "Python/bytecodes.c"
if (slice == NULL) { STACK_SHRINK(((oparg == 3) ? 1 : 0)); goto pop_2_error; }
#line 1895 "Python/executor_cases.c.h"
STACK_SHRINK(((oparg == 3) ? 1 : 0));
@ -1901,7 +1901,7 @@
case CONVERT_VALUE: {
PyObject *value = stack_pointer[-1];
PyObject *result;
#line 3497 "Python/bytecodes.c"
#line 3502 "Python/bytecodes.c"
convertion_func_ptr conv_fn;
assert(oparg >= FVC_STR && oparg <= FVC_ASCII);
conv_fn = CONVERSION_FUNCTIONS[oparg];
@ -1916,7 +1916,7 @@
case FORMAT_SIMPLE: {
PyObject *value = stack_pointer[-1];
PyObject *res;
#line 3506 "Python/bytecodes.c"
#line 3511 "Python/bytecodes.c"
/* If value is a unicode object, then we know the result
* of format(value) is value itself. */
if (!PyUnicode_CheckExact(value)) {
@ -1936,7 +1936,7 @@
PyObject *fmt_spec = stack_pointer[-1];
PyObject *value = stack_pointer[-2];
PyObject *res;
#line 3519 "Python/bytecodes.c"
#line 3524 "Python/bytecodes.c"
res = PyObject_Format(value, fmt_spec);
Py_DECREF(value);
Py_DECREF(fmt_spec);
@ -1950,7 +1950,7 @@
case COPY: {
PyObject *bottom = stack_pointer[-(1 + (oparg-1))];
PyObject *top;
#line 3526 "Python/bytecodes.c"
#line 3531 "Python/bytecodes.c"
assert(oparg > 0);
top = Py_NewRef(bottom);
#line 1957 "Python/executor_cases.c.h"
@ -1962,7 +1962,7 @@
case SWAP: {
PyObject *top = stack_pointer[-1];
PyObject *bottom = stack_pointer[-(2 + (oparg-2))];
#line 3551 "Python/bytecodes.c"
#line 3556 "Python/bytecodes.c"
assert(oparg >= 2);
#line 1968 "Python/executor_cases.c.h"
stack_pointer[-1] = bottom;

339
Python/generated_cases.c.h generated
View file

File diff suppressed because it is too large Load diff

2
Python/opcode_metadata.h generated
View file

@ -1087,7 +1087,7 @@ const struct opcode_metadata _PyOpcode_opcode_metadata[512] = {
[JUMP_BACKWARD] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_JUMP_FLAG },
[JUMP] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_JUMP_FLAG },
[JUMP_NO_INTERRUPT] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_JUMP_FLAG },
[ENTER_EXECUTOR] = { true, INSTR_FMT_IB, HAS_ARG_FLAG },
[ENTER_EXECUTOR] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_JUMP_FLAG },
[POP_JUMP_IF_FALSE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_JUMP_FLAG },
[POP_JUMP_IF_TRUE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_JUMP_FLAG },
[POP_JUMP_IF_NOT_NONE] = { true, INSTR_FMT_IB, HAS_ARG_FLAG | HAS_JUMP_FLAG },