Skip to content
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

bpo-46841: Use inline caching for UNPACK_SEQUENCE #31591

Merged
merged 2 commits into from
Feb 28, 2022
Merged
Show file tree
Hide file tree
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
9 changes: 8 additions & 1 deletion Include/internal/pycore_code.h
Original file line number Diff line number Diff line change
Expand Up @@ -68,9 +68,16 @@ typedef struct {
_Py_CODEUNIT counter;
} _PyBinaryOpCache;

typedef struct {
_Py_CODEUNIT counter;
} _PyUnpackSequenceCache;

#define INLINE_CACHE_ENTRIES_BINARY_OP \
(sizeof(_PyBinaryOpCache) / sizeof(_Py_CODEUNIT))

#define INLINE_CACHE_ENTRIES_UNPACK_SEQUENCE \
(sizeof(_PyUnpackSequenceCache) / sizeof(_Py_CODEUNIT))

/* Maximum size of code to quicken, in code units. */
#define MAX_SIZE_TO_QUICKEN 5000

Expand Down Expand Up @@ -312,7 +319,7 @@ extern void _Py_Specialize_BinaryOp(PyObject *lhs, PyObject *rhs, _Py_CODEUNIT *
int oparg);
extern void _Py_Specialize_CompareOp(PyObject *lhs, PyObject *rhs, _Py_CODEUNIT *instr, SpecializedCacheEntry *cache);
extern void _Py_Specialize_UnpackSequence(PyObject *seq, _Py_CODEUNIT *instr,
SpecializedCacheEntry *cache);
int oparg);

/* Deallocator function for static codeobjects used in deepfreeze.py */
extern void _PyStaticCode_Dealloc(PyCodeObject *co);
Expand Down
1 change: 1 addition & 0 deletions Include/opcode.h

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

3 changes: 2 additions & 1 deletion Lib/importlib/_bootstrap_external.py
Original file line number Diff line number Diff line change
Expand Up @@ -388,6 +388,7 @@ def _write_atomic(path, data, mode=0o666):
# Python 3.11a5 3479 (Add PUSH_NULL opcode)
# Python 3.11a5 3480 (New CALL opcodes, second iteration)
# Python 3.11a5 3481 (Use inline CACHE instructions)
# Python 3.11a5 3482 (Use inline caching for UNPACK_SEQUENCE)

# Python 3.12 will start with magic number 3500

Expand All @@ -402,7 +403,7 @@ def _write_atomic(path, data, mode=0o666):
# Whenever MAGIC_NUMBER is changed, the ranges in the magic_values array
# in PC/launcher.c must also be updated.

MAGIC_NUMBER = (3481).to_bytes(2, 'little') + b'\r\n'
MAGIC_NUMBER = (3482).to_bytes(2, 'little') + b'\r\n'
_RAW_MAGIC_NUMBER = int.from_bytes(MAGIC_NUMBER, 'little') # For import.c

_PYCACHE = '__pycache__'
Expand Down
2 changes: 1 addition & 1 deletion Lib/opcode.py
Original file line number Diff line number Diff line change
Expand Up @@ -109,7 +109,7 @@ def jabs_op(name, op, entries=0):

name_op('STORE_NAME', 90) # Index in name list
name_op('DELETE_NAME', 91) # ""
def_op('UNPACK_SEQUENCE', 92) # Number of tuple items
def_op('UNPACK_SEQUENCE', 92, 1) # Number of tuple items
jrel_op('FOR_ITER', 93)
def_op('UNPACK_EX', 94)
name_op('STORE_ATTR', 95) # Index in name list
Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
Use inline caching for :opcode:`UNPACK_SEQUENCE`.
29 changes: 15 additions & 14 deletions Python/ceval.c
Original file line number Diff line number Diff line change
Expand Up @@ -2758,22 +2758,22 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, _PyInterpreterFrame *frame, int
}
STACK_GROW(oparg);
Py_DECREF(seq);
JUMPBY(INLINE_CACHE_ENTRIES_UNPACK_SEQUENCE);
DISPATCH();
}

TARGET(UNPACK_SEQUENCE_ADAPTIVE) {
assert(cframe.use_tracing == 0);
SpecializedCacheEntry *cache = GET_CACHE();
if (cache->adaptive.counter == 0) {
_PyUnpackSequenceCache *cache = (_PyUnpackSequenceCache *)next_instr;
if (cache->counter == 0) {
PyObject *seq = TOP();
next_instr--;
_Py_Specialize_UnpackSequence(seq, next_instr, cache);
_Py_Specialize_UnpackSequence(seq, next_instr, oparg);
DISPATCH();
}
else {
STAT_INC(UNPACK_SEQUENCE, deferred);
cache->adaptive.counter--;
oparg = cache->adaptive.original_oparg;
cache->counter--;
JUMP_TO_INSTRUCTION(UNPACK_SEQUENCE);
}
}
Expand All @@ -2786,36 +2786,37 @@ _PyEval_EvalFrameDefault(PyThreadState *tstate, _PyInterpreterFrame *frame, int
SET_TOP(Py_NewRef(PyTuple_GET_ITEM(seq, 1)));
PUSH(Py_NewRef(PyTuple_GET_ITEM(seq, 0)));
Py_DECREF(seq);
JUMPBY(INLINE_CACHE_ENTRIES_UNPACK_SEQUENCE);
NOTRACE_DISPATCH();
}

TARGET(UNPACK_SEQUENCE_TUPLE) {
PyObject *seq = TOP();
int len = GET_CACHE()->adaptive.original_oparg;
DEOPT_IF(!PyTuple_CheckExact(seq), UNPACK_SEQUENCE);
DEOPT_IF(PyTuple_GET_SIZE(seq) != len, UNPACK_SEQUENCE);
DEOPT_IF(PyTuple_GET_SIZE(seq) != oparg, UNPACK_SEQUENCE);
STAT_INC(UNPACK_SEQUENCE, hit);
STACK_SHRINK(1);
PyObject **items = _PyTuple_ITEMS(seq);
while (len--) {
PUSH(Py_NewRef(items[len]));
while (oparg--) {
PUSH(Py_NewRef(items[oparg]));
}
Py_DECREF(seq);
JUMPBY(INLINE_CACHE_ENTRIES_UNPACK_SEQUENCE);
NOTRACE_DISPATCH();
}

TARGET(UNPACK_SEQUENCE_LIST) {
PyObject *seq = TOP();
int len = GET_CACHE()->adaptive.original_oparg;
DEOPT_IF(!PyList_CheckExact(seq), UNPACK_SEQUENCE);
DEOPT_IF(PyList_GET_SIZE(seq) != len, UNPACK_SEQUENCE);
DEOPT_IF(PyList_GET_SIZE(seq) != oparg, UNPACK_SEQUENCE);
STAT_INC(UNPACK_SEQUENCE, hit);
STACK_SHRINK(1);
PyObject **items = _PyList_ITEMS(seq);
while (len--) {
PUSH(Py_NewRef(items[len]));
while (oparg--) {
PUSH(Py_NewRef(items[oparg]));
}
Py_DECREF(seq);
JUMPBY(INLINE_CACHE_ENTRIES_UNPACK_SEQUENCE);
NOTRACE_DISPATCH();
}

Expand Down Expand Up @@ -5600,7 +5601,7 @@ MISS_WITH_CACHE(CALL)
MISS_WITH_INLINE_CACHE(BINARY_OP)
MISS_WITH_CACHE(COMPARE_OP)
MISS_WITH_CACHE(BINARY_SUBSCR)
MISS_WITH_CACHE(UNPACK_SEQUENCE)
MISS_WITH_INLINE_CACHE(UNPACK_SEQUENCE)
MISS_WITH_OPARG_COUNTER(STORE_SUBSCR)

LOAD_ATTR_INSTANCE_VALUE_miss:
Expand Down
23 changes: 11 additions & 12 deletions Python/specialize.c
Original file line number Diff line number Diff line change
Expand Up @@ -67,7 +67,6 @@ static uint8_t cache_requirements[256] = {
[PRECALL] = 2, /* _PyAdaptiveEntry and _PyObjectCache/_PyCallCache */
[STORE_ATTR] = 1, // _PyAdaptiveEntry
[COMPARE_OP] = 1, /* _PyAdaptiveEntry */
[UNPACK_SEQUENCE] = 1, // _PyAdaptiveEntry
};

Py_ssize_t _Py_QuickenedCount = 0;
Expand Down Expand Up @@ -2133,39 +2132,39 @@ unpack_sequence_fail_kind(PyObject *seq)
#endif

void
_Py_Specialize_UnpackSequence(PyObject *seq, _Py_CODEUNIT *instr,
SpecializedCacheEntry *cache)
_Py_Specialize_UnpackSequence(PyObject *seq, _Py_CODEUNIT *instr, int oparg)
{
_PyAdaptiveEntry *adaptive = &cache->adaptive;
assert(_PyOpcode_InlineCacheEntries[UNPACK_SEQUENCE] ==
INLINE_CACHE_ENTRIES_UNPACK_SEQUENCE);
_PyUnpackSequenceCache *cache = (_PyUnpackSequenceCache *)(instr + 1);
if (PyTuple_CheckExact(seq)) {
if (PyTuple_GET_SIZE(seq) != adaptive->original_oparg) {
if (PyTuple_GET_SIZE(seq) != oparg) {
SPECIALIZATION_FAIL(UNPACK_SEQUENCE, SPEC_FAIL_EXPECTED_ERROR);
goto failure;
}
if (PyTuple_GET_SIZE(seq) == 2) {
*instr = _Py_MAKECODEUNIT(UNPACK_SEQUENCE_TWO_TUPLE,
_Py_OPARG(*instr));
*instr = _Py_MAKECODEUNIT(UNPACK_SEQUENCE_TWO_TUPLE, oparg);
goto success;
}
*instr = _Py_MAKECODEUNIT(UNPACK_SEQUENCE_TUPLE, _Py_OPARG(*instr));
*instr = _Py_MAKECODEUNIT(UNPACK_SEQUENCE_TUPLE, oparg);
goto success;
}
if (PyList_CheckExact(seq)) {
if (PyList_GET_SIZE(seq) != adaptive->original_oparg) {
if (PyList_GET_SIZE(seq) != oparg) {
SPECIALIZATION_FAIL(UNPACK_SEQUENCE, SPEC_FAIL_EXPECTED_ERROR);
goto failure;
}
*instr = _Py_MAKECODEUNIT(UNPACK_SEQUENCE_LIST, _Py_OPARG(*instr));
*instr = _Py_MAKECODEUNIT(UNPACK_SEQUENCE_LIST, oparg);
goto success;
}
SPECIALIZATION_FAIL(UNPACK_SEQUENCE, unpack_sequence_fail_kind(seq));
failure:
STAT_INC(UNPACK_SEQUENCE, failure);
cache_backoff(adaptive);
cache->counter = ADAPTIVE_CACHE_BACKOFF;
return;
success:
STAT_INC(UNPACK_SEQUENCE, success);
adaptive->counter = initial_counter_value();
cache->counter = initial_counter_value();
}

#ifdef Py_STATS
Expand Down