PublicArchive/cpython
2
0
Fork 0
mirror of https://github.com/python/cpython.git synced 2026-05-06 04:37:33 -04:00
Code Issues Packages Projects Releases Wiki Activity

gh-148211: decompose _SHUFFLE_3_LOAD_CONST_INLINE_BORROW in JIT (GH-148816)

Browse Source
This commit is contained in:
Neko Asakura
2026-04-28 20:48:23 +08:00
committed by GitHub
parent be968c7210
commit 2b6a13710f
7 changed files with 1029 additions and 1167 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Include/internal/pycore_uop_ids.h
Generated
+1018 -1023
View File
File diff suppressed because it is too large Load Diff
Include/internal/pycore_uop_metadata.h
Generated
-21
View File
@@ -397,7 +397,6 @@ const uint32_t _PyUop_Flags[MAX_UOP_ID+1] = {
[_CHECK_VALIDITY] = HAS_DEOPT_FLAG,
[_LOAD_CONST_INLINE] = HAS_PURE_FLAG,
[_LOAD_CONST_INLINE_BORROW] = HAS_PURE_FLAG,
[_SHUFFLE_3_LOAD_CONST_INLINE_BORROW] = 0,
[_START_EXECUTOR] = HAS_DEOPT_FLAG,
[_MAKE_WARM] = 0,
[_FATAL_ERROR] = 0,
@@ -3700,15 +3699,6 @@ const _PyUopCachingInfo _PyUop_Caching[MAX_UOP_ID+1] = {
{ -1, -1, -1 },
},
},
[_SHUFFLE_3_LOAD_CONST_INLINE_BORROW] = {
.best = { 0, 1, 2, 3 },
.entries = {
{ 3, 0, _SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r03 },
{ 3, 1, _SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r13 },
{ 3, 2, _SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r23 },
{ 3, 3, _SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r33 },
},
},
[_START_EXECUTOR] = {
.best = { 0, 0, 0, 0 },
.entries = {
@@ -4705,10 +4695,6 @@ const uint16_t _PyUop_Uncached[MAX_UOP_REGS_ID+1] = {
[_LOAD_CONST_INLINE_BORROW_r01] = _LOAD_CONST_INLINE_BORROW,
[_LOAD_CONST_INLINE_BORROW_r12] = _LOAD_CONST_INLINE_BORROW,
[_LOAD_CONST_INLINE_BORROW_r23] = _LOAD_CONST_INLINE_BORROW,
[_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r03] = _SHUFFLE_3_LOAD_CONST_INLINE_BORROW,
[_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r13] = _SHUFFLE_3_LOAD_CONST_INLINE_BORROW,
[_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r23] = _SHUFFLE_3_LOAD_CONST_INLINE_BORROW,
[_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r33] = _SHUFFLE_3_LOAD_CONST_INLINE_BORROW,
[_START_EXECUTOR_r00] = _START_EXECUTOR,
[_MAKE_WARM_r00] = _MAKE_WARM,
[_MAKE_WARM_r11] = _MAKE_WARM,
@@ -5933,11 +5919,6 @@ const char *const _PyOpcode_uop_name[MAX_UOP_REGS_ID+1] = {
[_SET_IP_r33] = "_SET_IP_r33",
[_SET_UPDATE] = "_SET_UPDATE",
[_SET_UPDATE_r11] = "_SET_UPDATE_r11",
[_SHUFFLE_3_LOAD_CONST_INLINE_BORROW] = "_SHUFFLE_3_LOAD_CONST_INLINE_BORROW",
[_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r03] = "_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r03",
[_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r13] = "_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r13",
[_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r23] = "_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r23",
[_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r33] = "_SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r33",
[_SPILL_OR_RELOAD] = "_SPILL_OR_RELOAD",
[_SPILL_OR_RELOAD_r01] = "_SPILL_OR_RELOAD_r01",
[_SPILL_OR_RELOAD_r02] = "_SPILL_OR_RELOAD_r02",
@@ -6827,8 +6808,6 @@ int _PyUop_num_popped(int opcode, int oparg)
return 0;
case _LOAD_CONST_INLINE_BORROW:
return 0;
case _SHUFFLE_3_LOAD_CONST_INLINE_BORROW:
return 3;
case _START_EXECUTOR:
return 0;
case _MAKE_WARM:
Lib/test/test_capi/test_opt.py
+1 -1
View File
@@ -2963,7 +2963,7 @@ class TestUopsOptimization(unittest.TestCase):
self.assertIsNotNone(ex)
uops = get_opnames(ex)
self.assertNotIn("_CALL_LEN", uops)
self.assertEqual(count_ops(ex, "_SHUFFLE_3_LOAD_CONST_INLINE_BORROW"), 4)
self.assertGreaterEqual(count_ops(ex, "_LOAD_CONST_INLINE_BORROW"), 8)
def test_call_len_known_length_small_int(self):
# Make sure that len(t) is optimized for a tuple of length 5.
Python/bytecodes.c
-7
View File
@@ -6107,13 +6107,6 @@ dummy_func(
value = PyStackRef_FromPyObjectBorrow(ptr);
}
tier2 op(_SHUFFLE_3_LOAD_CONST_INLINE_BORROW, (ptr/4, callable, null, arg -- res, a, c)) {
res = PyStackRef_FromPyObjectBorrow(ptr);
a = arg;
c = callable;
INPUTS_DEAD();
}
tier2 op(_START_EXECUTOR, (executor/4 --)) {
#ifndef _Py_JIT
assert(current_executor == (_PyExecutorObject*)executor);
Python/executor_cases.c.h
Generated
-100
View File
@@ -22337,106 +22337,6 @@
break;
}
case _SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r03: {
CHECK_CURRENT_CACHED_VALUES(0);
assert(WITHIN_STACK_BOUNDS_IGNORING_CACHE());
_PyStackRef arg;
_PyStackRef callable;
_PyStackRef res;
_PyStackRef a;
_PyStackRef c;
arg = stack_pointer[-1];
callable = stack_pointer[-3];
PyObject *ptr = (PyObject *)CURRENT_OPERAND0_64();
res = PyStackRef_FromPyObjectBorrow(ptr);
a = arg;
c = callable;
_tos_cache2 = c;
_tos_cache1 = a;
_tos_cache0 = res;
SET_CURRENT_CACHED_VALUES(3);
stack_pointer += -3;
ASSERT_WITHIN_STACK_BOUNDS(__FILE__, __LINE__);
assert(WITHIN_STACK_BOUNDS_IGNORING_CACHE());
break;
}
case _SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r13: {
CHECK_CURRENT_CACHED_VALUES(1);
assert(WITHIN_STACK_BOUNDS_IGNORING_CACHE());
_PyStackRef arg;
_PyStackRef callable;
_PyStackRef res;
_PyStackRef a;
_PyStackRef c;
_PyStackRef _stack_item_0 = _tos_cache0;
arg = _stack_item_0;
callable = stack_pointer[-2];
PyObject *ptr = (PyObject *)CURRENT_OPERAND0_64();
res = PyStackRef_FromPyObjectBorrow(ptr);
a = arg;
c = callable;
_tos_cache2 = c;
_tos_cache1 = a;
_tos_cache0 = res;
SET_CURRENT_CACHED_VALUES(3);
stack_pointer += -2;
ASSERT_WITHIN_STACK_BOUNDS(__FILE__, __LINE__);
assert(WITHIN_STACK_BOUNDS_IGNORING_CACHE());
break;
}
case _SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r23: {
CHECK_CURRENT_CACHED_VALUES(2);
assert(WITHIN_STACK_BOUNDS_IGNORING_CACHE());
_PyStackRef arg;
_PyStackRef callable;
_PyStackRef res;
_PyStackRef a;
_PyStackRef c;
_PyStackRef _stack_item_0 = _tos_cache0;
_PyStackRef _stack_item_1 = _tos_cache1;
arg = _stack_item_1;
callable = stack_pointer[-1];
PyObject *ptr = (PyObject *)CURRENT_OPERAND0_64();
res = PyStackRef_FromPyObjectBorrow(ptr);
a = arg;
c = callable;
_tos_cache2 = c;
_tos_cache1 = a;
_tos_cache0 = res;
SET_CURRENT_CACHED_VALUES(3);
stack_pointer += -1;
ASSERT_WITHIN_STACK_BOUNDS(__FILE__, __LINE__);
assert(WITHIN_STACK_BOUNDS_IGNORING_CACHE());
break;
}
case _SHUFFLE_3_LOAD_CONST_INLINE_BORROW_r33: {
CHECK_CURRENT_CACHED_VALUES(3);
assert(WITHIN_STACK_BOUNDS_IGNORING_CACHE());
_PyStackRef arg;
_PyStackRef callable;
_PyStackRef res;
_PyStackRef a;
_PyStackRef c;
_PyStackRef _stack_item_0 = _tos_cache0;
_PyStackRef _stack_item_1 = _tos_cache1;
_PyStackRef _stack_item_2 = _tos_cache2;
arg = _stack_item_2;
callable = _stack_item_0;
PyObject *ptr = (PyObject *)CURRENT_OPERAND0_64();
res = PyStackRef_FromPyObjectBorrow(ptr);
a = arg;
c = callable;
_tos_cache2 = c;
_tos_cache1 = a;
_tos_cache0 = res;
SET_CURRENT_CACHED_VALUES(3);
assert(WITHIN_STACK_BOUNDS_IGNORING_CACHE());
break;
}
case _START_EXECUTOR_r00: {
CHECK_CURRENT_CACHED_VALUES(0);
assert(WITHIN_STACK_BOUNDS_IGNORING_CACHE());
Python/optimizer_bytecodes.c
+4 -1
View File
@@ -2330,7 +2330,10 @@ dummy_func(void) {
goto error;
}
if (_Py_IsImmortal(temp)) {
ADD_OP(_SHUFFLE_3_LOAD_CONST_INLINE_BORROW, 0, (uintptr_t)temp);
ADD_OP(_SWAP, 2, 0);
optimize_pop_top(ctx, this_instr, null);
ADD_OP(_LOAD_CONST_INLINE_BORROW, 0, (uintptr_t)temp);
ADD_OP(_SWAP, 3, 0);
}
res = sym_new_const(ctx, temp);
Py_DECREF(temp);
Python/optimizer_cases.c.h
Generated
+6 -14
View File
@@ -4514,11 +4514,13 @@
case _CALL_LEN: {
JitOptRef arg;
JitOptRef null;
JitOptRef callable;
JitOptRef res;
JitOptRef a;
JitOptRef c;
arg = stack_pointer[-1];
null = stack_pointer[-2];
callable = stack_pointer[-3];
res = sym_new_type(ctx, &PyLong_Type);
Py_ssize_t length = sym_tuple_length(arg);
@@ -4544,7 +4546,10 @@
goto error;
}
if (_Py_IsImmortal(temp)) {
ADD_OP(_SHUFFLE_3_LOAD_CONST_INLINE_BORROW, 0, (uintptr_t)temp);
ADD_OP(_SWAP, 2, 0);
optimize_pop_top(ctx, this_instr, null);
ADD_OP(_LOAD_CONST_INLINE_BORROW, 0, (uintptr_t)temp);
ADD_OP(_SWAP, 3, 0);
}
res = sym_new_const(ctx, temp);
CHECK_STACK_BOUNDS(-2);
@@ -5497,19 +5502,6 @@
break;
}
case _SHUFFLE_3_LOAD_CONST_INLINE_BORROW: {
JitOptRef res;
JitOptRef a;
JitOptRef c;
res = sym_new_not_null(ctx);
a = sym_new_not_null(ctx);
c = sym_new_not_null(ctx);
stack_pointer[-3] = res;
stack_pointer[-2] = a;
stack_pointer[-1] = c;
break;
}
case _START_EXECUTOR: {
break;
}