Skip to content

bpo-46532: Reduce number of memory writes to update call_shape.kwnames. #31231

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Feb 10, 2022
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 15 additions & 7 deletions Python/ceval.c
Original file line number Diff line number Diff line change
Expand Up @@ -1615,7 +1615,9 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr

CFrame cframe;
CallShape call_shape;
call_shape.kwnames = NULL; // Borrowed reference
call_shape.kwnames = NULL; // Borrowed reference. Reset by CALL instructions.
/* The following three values are always set by the PRECALL instructions.
They are set here to keep the compiler happy. */
call_shape.postcall_shrink = 0;
call_shape.total_args = 0;
call_shape.callable = NULL; // Strong reference
Expand Down Expand Up @@ -4446,7 +4448,7 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr
call_shape.postcall_shrink = 1;

call_shape.total_args = oparg;
call_shape.kwnames = NULL;
assert(call_shape.kwnames == NULL);
#ifdef Py_STATS
extern int _PySpecialization_ClassifyCallable(PyObject *);
_py_stats.opcode_stats[PRECALL_FUNCTION].specialization.failure++;
Expand Down Expand Up @@ -4490,12 +4492,11 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr
call_shape.postcall_shrink = 2-is_method;

call_shape.total_args = nargs;
call_shape.kwnames = NULL;
assert(call_shape.kwnames == NULL);
DISPATCH();
}

TARGET(KW_NAMES) {
assert(call_shape.kwnames == NULL);
assert(oparg < PyTuple_GET_SIZE(consts));
call_shape.kwnames = GETITEM(consts, oparg);
DISPATCH();
Expand Down Expand Up @@ -4531,6 +4532,7 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr
tstate, (PyFunctionObject *)function, locals,
stack_pointer, positional_args, call_shape.kwnames
);
call_shape.kwnames = NULL;
STACK_SHRINK(call_shape.postcall_shrink);
// The frame has stolen all the arguments from the stack,
// so there is no need to clean them up.
Expand All @@ -4556,6 +4558,7 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr
positional_args | PY_VECTORCALL_ARGUMENTS_OFFSET,
call_shape.kwnames);
}
call_shape.kwnames = NULL;
assert((res != NULL) ^ (_PyErr_Occurred(tstate) != NULL));
Py_DECREF(function);
/* Clear the stack */
Expand Down Expand Up @@ -4597,6 +4600,7 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr
}

TARGET(CALL_PY_EXACT_ARGS) {
assert(call_shape.kwnames == NULL);
SpecializedCacheEntry *caches = GET_CACHE();
int argcount = call_shape.total_args;
DEOPT_IF(!PyFunction_Check(call_shape.callable), CALL);
Expand Down Expand Up @@ -4625,6 +4629,7 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr
}

TARGET(CALL_PY_WITH_DEFAULTS) {
assert(call_shape.kwnames == NULL);
SpecializedCacheEntry *caches = GET_CACHE();
int argcount = call_shape.total_args;
DEOPT_IF(!PyFunction_Check(call_shape.callable), CALL);
Expand Down Expand Up @@ -4661,9 +4666,9 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr
}

TARGET(CALL_NO_KW_TYPE_1) {
assert(call_shape.kwnames == NULL);
assert(cframe.use_tracing == 0);
DEOPT_IF(call_shape.total_args != 1, CALL);
assert(call_shape.kwnames == NULL);
PyObject *obj = TOP();
PyObject *callable = SECOND();
DEOPT_IF(callable != (PyObject *)&PyType_Type, CALL);
Expand All @@ -4676,13 +4681,13 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr
}

TARGET(CALL_NO_KW_STR_1) {
assert(call_shape.kwnames == NULL);
assert(cframe.use_tracing == 0);
DEOPT_IF(!PyType_Check(call_shape.callable), CALL);
PyTypeObject *tp = (PyTypeObject *)call_shape.callable;
DEOPT_IF(call_shape.total_args != 1, CALL);
DEOPT_IF(tp != &PyUnicode_Type, CALL);
STAT_INC(CALL, hit);
assert(call_shape.kwnames == NULL);
PyObject *arg = TOP();
PyObject *res = PyObject_Str(arg);
Py_DECREF(arg);
Expand All @@ -4696,12 +4701,12 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr
}

TARGET(CALL_NO_KW_TUPLE_1) {
assert(call_shape.kwnames == NULL);
DEOPT_IF(!PyType_Check(call_shape.callable), CALL);
PyTypeObject *tp = (PyTypeObject *)call_shape.callable;
DEOPT_IF(call_shape.total_args != 1, CALL);
DEOPT_IF(tp != &PyTuple_Type, CALL);
STAT_INC(CALL, hit);
assert(call_shape.kwnames == NULL);
PyObject *arg = TOP();
PyObject *res = PySequence_Tuple(arg);
Py_DECREF(arg);
Expand All @@ -4724,6 +4729,7 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr
int nargs = call_shape.total_args - kwnames_len;
STACK_SHRINK(call_shape.total_args);
PyObject *res = tp->tp_vectorcall((PyObject *)tp, stack_pointer, nargs, call_shape.kwnames);
call_shape.kwnames = NULL;
/* Free the arguments. */
for (int i = 0; i < call_shape.total_args; i++) {
Py_DECREF(stack_pointer[i]);
Expand Down Expand Up @@ -4833,6 +4839,7 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, InterpreterFrame *frame, int thr
call_shape.kwnames
);
assert((res != NULL) ^ (_PyErr_Occurred(tstate) != NULL));
call_shape.kwnames = NULL;

/* Free the arguments. */
for (int i = 0; i < call_shape.total_args; i++) {
Expand Down Expand Up @@ -5398,6 +5405,7 @@ MISS_WITH_OPARG_COUNTER(STORE_SUBSCR)
}

error:
call_shape.kwnames = NULL;
/* Double-check exception status. */
#ifdef NDEBUG
if (!_PyErr_Occurred(tstate)) {
Expand Down