[mypyc] Fix segfault when top level raises exception (#10586)

If an error occurs while running the initialization code, set the
CPyModule_<name>_internal module pointer to NULL so future attempts to
import that same module don't mistakenly think that the module is
already initialized due to the fact that that module pointer is not
NULL.

Clearing that module pointer on error allows us to keep initializing it
at the beginning of the function before running any top level code,
which is necessary to prevent RecursionErrors when dealing with circular
imports.

Author: pranavrajpal

mypyc/codegen/emitmodule.py

@@ -888,15 +888,15 @@ def generate_module_def(self, emitter: Emitter, module_name: str, module: Module
 
         emitter.emit_lines('{} = PyModule_Create(&{}module);'.format(module_static, module_prefix),
                            'if (unlikely({} == NULL))'.format(module_static),
-                           '    return NULL;')
+                           '    goto fail;')
         emitter.emit_line(
             'PyObject *modname = PyObject_GetAttrString((PyObject *){}, "__name__");'.format(
                 module_static))
 
         module_globals = emitter.static_name('globals', module_name)
         emitter.emit_lines('{} = PyModule_GetDict({});'.format(module_globals, module_static),
                            'if (unlikely({} == NULL))'.format(module_globals),
-                           '    return NULL;')
+                           '    goto fail;')
 
         # HACK: Manually instantiate generated classes here
         for cl in module.classes:
@@ -907,16 +907,19 @@ def generate_module_def(self, emitter: Emitter, module_name: str, module: Module
                     '(PyObject *){t}_template, NULL, modname);'
                     .format(t=type_struct))
                 emitter.emit_lines('if (unlikely(!{}))'.format(type_struct),
-                                   '    return NULL;')
+                                   '    goto fail;')
 
         emitter.emit_lines('if (CPyGlobalsInit() < 0)',
-                           '    return NULL;')
+                           '    goto fail;')
 
         self.generate_top_level_call(module, emitter)
 
         emitter.emit_lines('Py_DECREF(modname);')
 
         emitter.emit_line('return {};'.format(module_static))
+        emitter.emit_lines('fail:',
+                           '{} = NULL;'.format(module_static),
+                           'return NULL;')
         emitter.emit_line('}')
 
     def generate_top_level_call(self, module: ModuleIR, emitter: Emitter) -> None:
@@ -927,7 +930,7 @@ def generate_top_level_call(self, module: ModuleIR, emitter: Emitter) -> None:
                 emitter.emit_lines(
                     'char result = {}();'.format(emitter.native_function_name(fn.decl)),
                     'if (result == 2)',
-                    '    return NULL;',
+                    '    goto fail;',
                 )
                 break
 

mypyc/test-data/run-misc.test

@@ -1037,3 +1037,27 @@ C = sys.platform == 'x' and (lambda x: y + x)
 assert not A
 assert not B
 assert not C
+
+[case testDoesntSegfaultWhenTopLevelFails]
+# make the initial import fail
+assert False
+
+class C:
+    def __init__(self):
+        self.x = 1
+        self.y = 2
+def test() -> None:
+    a = C()
+[file driver.py]
+# load native, cause PyInit to be run, create the module but don't finish initializing the globals
+try:
+    import native
+except:
+    pass
+try:
+    # try accessing those globals that were never properly initialized
+    import native
+    native.test()
+# should fail with AssertionError due to `assert False` in other function
+except AssertionError:
+    pass