gh-132775: Add _PyCode_VerifyStateless()

Include/internal/pycore_code.h

@@ -616,6 +616,46 @@ PyAPI_FUNC(int) _PyCode_SetUnboundVarCounts(
         PyObject *globalsns,
         PyObject *builtinsns);
 
+
+/* "Stateless" code is a function or code object which does not rely on
+ * external state or internal state.  It may rely on arguments and
+ * builtins, but not globals or a closure.  Thus it does not need to
+ * rely __globals__ or __closure__, and a stateless function
+ * is equivalent to its code object.
+ *
+ * Stateless code also does not keep any persistent state
+ * of its own, so it can't have any executors, monitoring,
+ * instrumentation, or "extras" (i.e. co_extra).
+ *
+ * Stateless code may create nested functions, including closures.
+ * However, nested functions must themselves be stateless, except they
+ * *can* close on the enclosing locals.
+ *
+ * Stateless code may return any value, including nested functions and closures.
+ *
+ * Stateless code that takes no arguments and doesn't return anything
+ * may be treated like a script.
+ *
+ * Note that we can consider stateless code to be
+ * "portable" if it does not return any object that holds a reference
+ * to any of the code's locals.  Thus generators and coroutines are not
+ * portable.  Likewise a function that returns a closure isn't portable.
+ * The concept of portability is useful in cases where the code is run
+ * in a different execution context than where the return value
+ * will be used. */
+
+PyAPI_FUNC(const char *) _PyCode_CheckNoInternalState(PyCodeObject *);
+PyAPI_FUNC(const char *) _PyCode_CheckNoExternalState(
+        PyCodeObject *,
+        _PyCode_var_counts_t *);
+PyAPI_FUNC(int) _PyCode_VerifyStateless(
+        PyThreadState *,
+        PyCodeObject *,
+        PyObject *globalnames,
+        PyObject *globalsns,
+        PyObject *builtinsns);
+
+PyAPI_FUNC(const char *) _PyCode_CheckPureFunction(PyCodeObject *);
 PyAPI_FUNC(int) _PyCode_ReturnsOnlyNone(PyCodeObject *);
 
 

Include/internal/pycore_function.h

@@ -35,6 +35,13 @@ PyFunctionObject *_PyFunction_LookupByVersion(uint32_t version, PyObject **p_cod
 extern PyObject *_Py_set_function_type_params(
     PyThreadState* unused, PyObject *func, PyObject *type_params);
 
+
+/* See pycore_code.h for explanation about what "stateless" means. */
+
+PyAPI_FUNC(int)
+_PyFunction_VerifyStateless(PyThreadState *, PyObject *);
+
+
 #ifdef __cplusplus
 }
 #endif

Lib/test/_code_definitions.py

@@ -12,6 +12,40 @@ def spam_minimal():
     return
 
 
+def spam_with_builtins():
+    x = 42
+    values = (42,)
+    checks = tuple(callable(v) for v in values)
+    res = callable(values), tuple(values), list(values), checks
+    print(res)
+
+
+def spam_with_globals_and_builtins():
+    func1 = spam
+    func2 = spam_minimal
+    funcs = (func1, func2)
+    checks = tuple(callable(f) for f in funcs)
+    res = callable(funcs), tuple(funcs), list(funcs), checks
+    print(res)
+
+
+def spam_returns_arg(x):
+    return x
+
+
+def spam_with_inner_not_closure():
+    def eggs():
+        pass
+    eggs()
+
+
+def spam_with_inner_closure():
+    x = 42
+    def eggs():
+        print(x)
+    eggs()
+
+
 def spam_full(a, b, /, c, d:int=1, *args, e, f:object=None, **kwargs) -> tuple:
     # arg defaults, kwarg defaults
     # annotations
@@ -98,6 +132,11 @@ def ham_C_closure(z):
 TOP_FUNCTIONS = [
     # shallow
     spam_minimal,
+    spam_with_builtins,
+    spam_with_globals_and_builtins,
+    spam_returns_arg,
+    spam_with_inner_not_closure,
+    spam_with_inner_closure,
     spam_full,
     spam,
     # outer func
@@ -127,6 +166,30 @@ def ham_C_closure(z):
     *NESTED_FUNCTIONS,
 ]
 
+STATELESS_FUNCTIONS = [
+    spam,
+    spam_minimal,
+    spam_with_builtins,
+    spam_returns_arg,
+    spam_with_inner_not_closure,
+    spam_with_inner_closure,
+    spam_N,
+    spam_C,
+    spam_NN,
+    spam_NC,
+    spam_CN,
+    spam_CC,
+    eggs_nested,
+    eggs_nested_N,
+    ham_nested,
+    ham_C_nested
+]
+STATELESS_CODE = [
+    *STATELESS_FUNCTIONS,
+    spam_with_globals_and_builtins,
+    spam_full,
+]
+
 
 # generators
 

Lib/test/test_code.py

@@ -220,6 +220,7 @@
     import _testinternalcapi
 except ModuleNotFoundError:
     _testinternalcapi = None
+import test._code_definitions as defs
 
 COPY_FREE_VARS = opmap['COPY_FREE_VARS']
 
@@ -671,9 +672,31 @@ def test_local_kinds(self):
         VARARGS = CO_FAST_LOCAL | CO_FAST_ARG_VAR | CO_FAST_ARG_POS
         VARKWARGS = CO_FAST_LOCAL | CO_FAST_ARG_VAR | CO_FAST_ARG_KW
 
-        import test._code_definitions as defs
         funcs = {
             defs.spam_minimal: {},
+            defs.spam_with_builtins: {
+                'x': CO_FAST_LOCAL,
+                'values': CO_FAST_LOCAL,
+                'checks': CO_FAST_LOCAL,
+                'res': CO_FAST_LOCAL,
+            },
+            defs.spam_with_globals_and_builtins: {
+                'func1': CO_FAST_LOCAL,
+                'func2': CO_FAST_LOCAL,
+                'funcs': CO_FAST_LOCAL,
+                'checks': CO_FAST_LOCAL,
+                'res': CO_FAST_LOCAL,
+            },
+            defs.spam_returns_arg: {
+                'x': POSORKW,
+            },
+            defs.spam_with_inner_not_closure: {
+                'eggs': CO_FAST_LOCAL,
+            },
+            defs.spam_with_inner_closure: {
+                'x': CO_FAST_CELL,
+                'eggs': CO_FAST_LOCAL,
+            },
             defs.spam_full: {
                 'a': POSONLY,
                 'b': POSONLY,
@@ -859,9 +882,26 @@ def new_var_counts(*,
                 },
             }
 
-        import test._code_definitions as defs
         funcs = {
             defs.spam_minimal: new_var_counts(),
+            defs.spam_with_builtins: new_var_counts(
+                purelocals=4,
+                globalvars=4,
+            ),
+            defs.spam_with_globals_and_builtins: new_var_counts(
+                purelocals=5,
+                globalvars=6,
+            ),
+            defs.spam_returns_arg: new_var_counts(
+                posorkw=1,
+            ),
+            defs.spam_with_inner_not_closure: new_var_counts(
+                purelocals=1,
+            ),
+            defs.spam_with_inner_closure: new_var_counts(
+                othercells=1,
+                purelocals=1,
+            ),
             defs.spam_full: new_var_counts(
                 posonly=2,
                 posorkw=2,
@@ -958,55 +998,70 @@ def new_var_counts(*,
                 counts = _testinternalcapi.get_code_var_counts(func.__code__)
                 self.assertEqual(counts, expected)
 
-        def func_with_globals_and_builtins():
-            mod1 = _testinternalcapi
-            mod2 = dis
-            mods = (mod1, mod2)
-            checks = tuple(callable(m) for m in mods)
-            return callable(mod2), tuple(mods), list(mods), checks
-
-        func = func_with_globals_and_builtins
+        func = defs.spam_with_globals_and_builtins
         with self.subTest(f'{func} code'):
             expected = new_var_counts(
-                purelocals=4,
-                globalvars=5,
+                purelocals=5,
+                globalvars=6,
             )
             counts = _testinternalcapi.get_code_var_counts(func.__code__)
             self.assertEqual(counts, expected)
 
         with self.subTest(f'{func} with own globals and builtins'):
             expected = new_var_counts(
-                purelocals=4,
-                globalvars=(2, 3),
+                purelocals=5,
+                globalvars=(2, 4),
             )
             counts = _testinternalcapi.get_code_var_counts(func)
             self.assertEqual(counts, expected)
 
         with self.subTest(f'{func} without globals'):
             expected = new_var_counts(
-                purelocals=4,
-                globalvars=(0, 3, 2),
+                purelocals=5,
+                globalvars=(0, 4, 2),
             )
             counts = _testinternalcapi.get_code_var_counts(func, globalsns={})
             self.assertEqual(counts, expected)
 
         with self.subTest(f'{func} without both'):
             expected = new_var_counts(
-                purelocals=4,
-                globalvars=5,
+                purelocals=5,
+                globalvars=6,
             )
             counts = _testinternalcapi.get_code_var_counts(func, globalsns={},
                   builtinsns={})
             self.assertEqual(counts, expected)
 
         with self.subTest(f'{func} without builtins'):
             expected = new_var_counts(
-                purelocals=4,
-                globalvars=(2, 0, 3),
+                purelocals=5,
+                globalvars=(2, 0, 4),
             )
             counts = _testinternalcapi.get_code_var_counts(func, builtinsns={})
             self.assertEqual(counts, expected)
 
+    @unittest.skipIf(_testinternalcapi is None, "missing _testinternalcapi")
+    def test_stateless(self):
+        self.maxDiff = None
+
+        for func in defs.STATELESS_CODE:
+            with self.subTest((func, '(code)')):
+                _testinternalcapi.verify_stateless_code(func.__code__)
+        for func in defs.STATELESS_FUNCTIONS:
+            with self.subTest((func, '(func)')):
+                _testinternalcapi.verify_stateless_code(func)
+
+        for func in defs.FUNCTIONS:
+            if func not in defs.STATELESS_CODE:
+                with self.subTest((func, '(code)')):
+                    with self.assertRaises(Exception):
+                        _testinternalcapi.verify_stateless_code(func.__code__)
+
+            if func not in defs.STATELESS_FUNCTIONS:
+                with self.subTest((func, '(func)')):
+                    with self.assertRaises(Exception):
+                        _testinternalcapi.verify_stateless_code(func)
+
 
 def isinterned(s):
     return s is sys.intern(('_' + s + '_')[1:-1])

Modules/_testinternalcapi.c

@@ -1165,6 +1165,48 @@ get_code_var_counts(PyObject *self, PyObject *_args, PyObject *_kwargs)
     return NULL;
 }
 
+static PyObject *
+verify_stateless_code(PyObject *self, PyObject *args, PyObject *kwargs)
+{
+    PyThreadState *tstate = _PyThreadState_GET();
+    PyObject *codearg;
+    PyObject *globalnames = NULL;
+    PyObject *globalsns = NULL;
+    PyObject *builtinsns = NULL;
+    static char *kwlist[] = {"code", "globalnames",
+                             "globalsns", "builtinsns", NULL};
+    if (!PyArg_ParseTupleAndKeywords(args, kwargs,
+                    "O|O!O!O!:get_code_var_counts", kwlist,
+                    &codearg, &PySet_Type, &globalnames,
+                    &PyDict_Type, &globalsns, &PyDict_Type, &builtinsns))
+    {
+        return NULL;
+    }
+    if (PyFunction_Check(codearg)) {
+        if (globalsns == NULL) {
+            globalsns = PyFunction_GET_GLOBALS(codearg);
+        }
+        if (builtinsns == NULL) {
+            builtinsns = PyFunction_GET_BUILTINS(codearg);
+        }
+        codearg = PyFunction_GET_CODE(codearg);
+    }
+    else if (!PyCode_Check(codearg)) {
+        PyErr_SetString(PyExc_TypeError,
+                        "argument must be a code object or a function");
+        return NULL;
+    }
+    PyCodeObject *code = (PyCodeObject *)codearg;
+
+    if (_PyCode_VerifyStateless(
+                tstate, code, globalnames, globalsns, builtinsns) < 0)
+    {
+        return NULL;
+    }
+    Py_RETURN_NONE;
+}
+
+
 static PyObject *
 jit_enabled(PyObject *self, PyObject *arg)
 {
@@ -2288,6 +2330,8 @@ static PyMethodDef module_functions[] = {
     {"get_co_localskinds", get_co_localskinds, METH_O, NULL},
     {"get_code_var_counts", _PyCFunction_CAST(get_code_var_counts),
      METH_VARARGS | METH_KEYWORDS, NULL},
+    {"verify_stateless_code", _PyCFunction_CAST(verify_stateless_code),
+     METH_VARARGS | METH_KEYWORDS, NULL},
     {"jit_enabled", jit_enabled,  METH_NOARGS, NULL},
 #ifdef _Py_TIER2
     {"add_executor_dependency", add_executor_dependency, METH_VARARGS, NULL},