optimize _CALL_BUILTIN_FAST_WITH_KEYWORDS

python · Fidget-Spinner · Apr 11, 2026 · Apr 11, 2026 · Apr 11, 2026 · Apr 11, 2026
commit 60aaed74368f7735c989d6f650416ec4fce08419
diff --git a/Include/internal/pycore_ceval.h b/Include/internal/pycore_ceval.h
@@ -439,7 +439,7 @@ _Py_BuiltinCallFast_StackRef(
     int total_args);
 
 PyAPI_FUNC(PyObject *)
-_Py_BuiltinCallFastWithKeywords_StackRefSteal(
+_Py_BuiltinCallFastWithKeywords_StackRef(
     _PyStackRef callable,
     _PyStackRef *arguments,
     int total_args);

diff --git a/Include/internal/pycore_opcode_metadata.h b/Include/internal/pycore_opcode_metadata.h
diff --git a/Include/internal/pycore_uop_ids.h b/Include/internal/pycore_uop_ids.h
diff --git a/Include/internal/pycore_uop_metadata.h b/Include/internal/pycore_uop_metadata.h
diff --git a/Lib/test/test_capi/test_opt.py b/Lib/test/test_capi/test_opt.py
@@ -2746,7 +2746,12 @@ def test_call_builtin_fast_with_keywords(self):
         def testfunc(n):
             x = 0
             for _ in range(n):
+                # _CALL_BUILTIN_FAST_WITH_KEYWORDS: 1 _POP_TOP_NOP, 1 POP_TOP
+                # _LIST_EXTEND: 1 _POP_TOP_NOP,
                 y = sorted([3, 1, 2])
+
+                # _BINARY_OP_SUBSCR_LIST_INT: 2 _POP_TOP_NOP
+                # _BINARY_OP_ADD_INT: 1 _POP_TOP_NOP
                 x += y[0]
             return x
 
@@ -2756,6 +2761,8 @@ def testfunc(n):
         uops = get_opnames(ex)
         self.assertIn("_CALL_BUILTIN_FAST_WITH_KEYWORDS", uops)
         self.assertNotIn("_GUARD_CALLABLE_BUILTIN_FAST_WITH_KEYWORDS", uops)
+        self.assertGreaterEqual(count_ops(ex, "_POP_TOP_NOP"), 5)
+        self.assertGreaterEqual(count_ops(ex, "_POP_TOP"), 3)
 
     def test_call_method_descriptor_o(self):
         def testfunc(n):

diff --git a/Modules/_testinternalcapi/test_cases.c.h b/Modules/_testinternalcapi/test_cases.c.h
@@ -4689,7 +4689,7 @@ dummy_func(
             EXIT_IF(PyCFunction_GET_FLAGS(callable_o) != (METH_FASTCALL | METH_KEYWORDS));
         }
 
-        op(_CALL_BUILTIN_FAST_WITH_KEYWORDS, (callable, self_or_null, args[oparg] -- res)) {
+        op(_CALL_BUILTIN_FAST_WITH_KEYWORDS, (callable, self_or_null, args[oparg] -- callable, self_or_null, args[oparg])) {
             /* Builtin METH_FASTCALL | METH_KEYWORDS functions */
             int total_args = oparg;
             _PyStackRef *arguments = args;
@@ -4698,12 +4698,13 @@ dummy_func(
                 total_args++;
             }
             STAT_INC(CALL, hit);
-            PyObject *res_o = _Py_BuiltinCallFastWithKeywords_StackRefSteal(callable, arguments, total_args);
-            DEAD(args);
-            DEAD(self_or_null);
-            DEAD(callable);
-            ERROR_IF(res_o == NULL);
-            res = PyStackRef_FromPyObjectSteal(res_o);
+            PyObject *res_o = _Py_BuiltinCallFastWithKeywords_StackRef(callable, arguments, total_args);
+            if (res_o == NULL) {
+                ERROR_NO_POP();
+            }
+            _PyStackRef temp = callable;
+            callable = PyStackRef_FromPyObjectSteal(res_o);
+            PyStackRef_CLOSE(temp);
         }
 
         macro(CALL_BUILTIN_FAST_WITH_KEYWORDS) =
@@ -4712,6 +4713,8 @@ dummy_func(
             unused/2 +
             _GUARD_CALLABLE_BUILTIN_FAST_WITH_KEYWORDS +
             _CALL_BUILTIN_FAST_WITH_KEYWORDS +
+            _POP_TOP_OPARG +
+            POP_TOP +
             _CHECK_PERIODIC_AT_END;
 
         macro(CALL_LEN) =

@@ -832,32 +832,22 @@ _Py_BuiltinCallFast_StackRef(
 }
 
 PyObject *
-_Py_BuiltinCallFastWithKeywords_StackRefSteal(
+_Py_BuiltinCallFastWithKeywords_StackRef(
     _PyStackRef callable,
     _PyStackRef *arguments,
     int total_args)
 {
     PyObject *res;
     STACKREFS_TO_PYOBJECTS(arguments, total_args, args_o);
     if (CONVERSION_FAILED(args_o)) {
-        res = NULL;
-        goto cleanup;
+        return NULL;
     }
     PyObject *callable_o = PyStackRef_AsPyObjectBorrow(callable);
     PyCFunctionFastWithKeywords cfunc =
         _PyCFunctionFastWithKeywords_CAST(PyCFunction_GET_FUNCTION(callable_o));
     res = cfunc(PyCFunction_GET_SELF(callable_o), args_o, total_args, NULL);
     STACKREFS_TO_PYOBJECTS_CLEANUP(args_o);
     assert((res != NULL) ^ (PyErr_Occurred() != NULL));
-cleanup:
-    // arguments is a pointer into the GC visible stack,
-    // so we must NULL out values as we clear them.
-    for (int i = total_args-1; i >= 0; i--) {
-        _PyStackRef tmp = arguments[i];
-        arguments[i] = PyStackRef_NULL;
-        PyStackRef_CLOSE(tmp);
-    }
-    PyStackRef_CLOSE(callable);
     return res;
 }
 

diff --git a/Python/executor_cases.c.h b/Python/executor_cases.c.h
diff --git a/Python/generated_cases.c.h b/Python/generated_cases.c.h
@@ -1325,6 +1325,10 @@ dummy_func(void) {
         }
     }
 
+    op(_CALL_BUILTIN_FAST_WITH_KEYWORDS, (callable, self_or_null, args[oparg] -- callable, self_or_null, args[oparg])) {
+        callable = sym_new_not_null(ctx);
+    }
+
     op(_CALL_BUILTIN_O, (callable, self_or_null, args[oparg] -- res, c, s)) {
         res = sym_new_not_null(ctx);
         c = callable;

diff --git a/Python/optimizer_cases.c.h b/Python/optimizer_cases.c.h