gh-94673: Properly Initialize and Finalize Static Builtin Types for Each Interpreter (gh-104072)

Until now, we haven't been initializing nor finalizing the per-interpreter state properly.

8 files changed, 103 insertions, 91 deletions

diff --git a/Objects/exceptions.c b/Objects/exceptions.c
index 6c9dfbd9b4..ba5ee291f0 100644
--- a/Objects/exceptions.c
+++ b/Objects/exceptions.c
@@ -3598,7 +3598,7 @@ _PyExc_InitTypes(PyInterpreterState *interp)
 {
     for (size_t i=0; i < Py_ARRAY_LENGTH(static_exceptions); i++) {
         PyTypeObject *exc = static_exceptions[i].exc;
-        if (_PyStaticType_InitBuiltin(exc) < 0) {
+        if (_PyStaticType_InitBuiltin(interp, exc) < 0) {
             return -1;
         }
     }
@@ -3609,13 +3609,9 @@ _PyExc_InitTypes(PyInterpreterState *interp)
 static void
 _PyExc_FiniTypes(PyInterpreterState *interp)
 {
-    if (!_Py_IsMainInterpreter(interp)) {
-        return;
-    }
-
     for (Py_ssize_t i=Py_ARRAY_LENGTH(static_exceptions) - 1; i >= 0; i--) {
         PyTypeObject *exc = static_exceptions[i].exc;
-        _PyStaticType_Dealloc(exc);
+        _PyStaticType_Dealloc(interp, exc);
     }
 }
 
diff --git a/Objects/floatobject.c b/Objects/floatobject.c
index a694ddcd01..d257857d9c 100644
--- a/Objects/floatobject.c
+++ b/Objects/floatobject.c
@@ -1991,8 +1991,9 @@ PyStatus
 _PyFloat_InitTypes(PyInterpreterState *interp)
 {
     /* Init float info */
-    if (_PyStructSequence_InitBuiltin(&FloatInfoType,
-                                      &floatinfo_desc) < 0) {
+    if (_PyStructSequence_InitBuiltin(interp, &FloatInfoType,
+                                      &floatinfo_desc) < 0)
+    {
         return _PyStatus_ERR("can't init float info type");
     }
 
@@ -2028,9 +2029,7 @@ _PyFloat_Fini(PyInterpreterState *interp)
 void
 _PyFloat_FiniType(PyInterpreterState *interp)
 {
-    if (_Py_IsMainInterpreter(interp)) {
-        _PyStructSequence_FiniBuiltin(&FloatInfoType);
-    }
+    _PyStructSequence_FiniBuiltin(interp, &FloatInfoType);
 }
 
 /* Print summary info about the state of the optimized allocator */
diff --git a/Objects/longobject.c b/Objects/longobject.c
index de043488d7..853e934e21 100644
--- a/Objects/longobject.c
+++ b/Objects/longobject.c
@@ -7,7 +7,6 @@
 #include "pycore_initconfig.h"    // _PyStatus_OK()
 #include "pycore_long.h"          // _Py_SmallInts
 #include "pycore_object.h"        // _PyObject_Init()
-#include "pycore_pystate.h"       // _Py_IsMainInterpreter()
 #include "pycore_runtime.h"       // _PY_NSMALLPOSINTS
 #include "pycore_structseq.h"     // _PyStructSequence_FiniBuiltin()
 
@@ -6352,7 +6351,9 @@ PyStatus
 _PyLong_InitTypes(PyInterpreterState *interp)
 {
     /* initialize int_info */
-    if (_PyStructSequence_InitBuiltin(&Int_InfoType, &int_info_desc) < 0) {
+    if (_PyStructSequence_InitBuiltin(interp, &Int_InfoType,
+                                      &int_info_desc) < 0)
+    {
         return _PyStatus_ERR("can't init int info type");
     }
 
@@ -6363,9 +6364,5 @@ _PyLong_InitTypes(PyInterpreterState *interp)
 void
 _PyLong_FiniTypes(PyInterpreterState *interp)
 {
-    if (!_Py_IsMainInterpreter(interp)) {
-        return;
-    }
-
-    _PyStructSequence_FiniBuiltin(&Int_InfoType);
+    _PyStructSequence_FiniBuiltin(interp, &Int_InfoType);
 }
diff --git a/Objects/object.c b/Objects/object.c
index 4ce10cf119..ee8690101d 100644
--- a/Objects/object.c
+++ b/Objects/object.c
@@ -2105,7 +2105,7 @@ _PyTypes_InitTypes(PyInterpreterState *interp)
     // All other static types (unless initialized elsewhere)
     for (size_t i=0; i < Py_ARRAY_LENGTH(static_types); i++) {
         PyTypeObject *type = static_types[i];
-        if (_PyStaticType_InitBuiltin(type) < 0) {
+        if (_PyStaticType_InitBuiltin(interp, type) < 0) {
             return _PyStatus_ERR("Can't initialize builtin type");
         }
         if (type == &PyType_Type) {
@@ -2128,15 +2128,11 @@ _PyTypes_InitTypes(PyInterpreterState *interp)
 void
 _PyTypes_FiniTypes(PyInterpreterState *interp)
 {
-    if (!_Py_IsMainInterpreter(interp)) {
-        return;
-    }
-
     // Deallocate types in the reverse order to deallocate subclasses before
     // their base classes.
     for (Py_ssize_t i=Py_ARRAY_LENGTH(static_types)-1; i>=0; i--) {
         PyTypeObject *type = static_types[i];
-        _PyStaticType_Dealloc(type);
+        _PyStaticType_Dealloc(interp, type);
     }
 }
 
diff --git a/Objects/structseq.c b/Objects/structseq.c
index d8f55dc1ea..ea476bf7a6 100644
--- a/Objects/structseq.c
+++ b/Objects/structseq.c
@@ -502,7 +502,8 @@ initialize_static_type(PyTypeObject *type, PyStructSequence_Desc *desc,
 }
 
 int
-_PyStructSequence_InitBuiltinWithFlags(PyTypeObject *type,
+_PyStructSequence_InitBuiltinWithFlags(PyInterpreterState *interp,
+                                       PyTypeObject *type,
                                        PyStructSequence_Desc *desc,
                                        unsigned long tp_flags)
 {
@@ -536,7 +537,7 @@ _PyStructSequence_InitBuiltinWithFlags(PyTypeObject *type,
     }
 #endif
 
-    if (_PyStaticType_InitBuiltin(type) < 0) {
+    if (_PyStaticType_InitBuiltin(interp, type) < 0) {
         PyErr_Format(PyExc_RuntimeError,
                      "Can't initialize builtin type %s",
                      desc->name);
@@ -606,7 +607,7 @@ PyStructSequence_InitType(PyTypeObject *type, PyStructSequence_Desc *desc)
    initialized via _PyStructSequence_InitBuiltinWithFlags(). */
 
 void
-_PyStructSequence_FiniBuiltin(PyTypeObject *type)
+_PyStructSequence_FiniBuiltin(PyInterpreterState *interp, PyTypeObject *type)
 {
     // Ensure that the type is initialized
     assert(type->tp_name != NULL);
@@ -620,13 +621,15 @@ _PyStructSequence_FiniBuiltin(PyTypeObject *type)
         return;
     }
 
-    _PyStaticType_Dealloc(type);
+    _PyStaticType_Dealloc(interp, type);
 
-    // Undo _PyStructSequence_InitBuiltinWithFlags().
-    type->tp_name = NULL;
-    PyMem_Free(type->tp_members);
-    type->tp_members = NULL;
-    type->tp_base = NULL;
+    if (_Py_IsMainInterpreter(interp)) {
+        // Undo _PyStructSequence_InitBuiltinWithFlags().
+        type->tp_name = NULL;
+        PyMem_Free(type->tp_members);
+        type->tp_members = NULL;
+        type->tp_base = NULL;
+    }
 }
 
 
diff --git a/Objects/typeobject.c b/Objects/typeobject.c
index 060d14e254..2ed806fb01 100644
--- a/Objects/typeobject.c
+++ b/Objects/typeobject.c
@@ -69,13 +69,11 @@ static inline PyTypeObject * subclass_from_ref(PyObject *ref);
 
 /* helpers for for static builtin types */
 
-#ifndef NDEBUG
 static inline int
 static_builtin_index_is_set(PyTypeObject *self)
 {
     return self->tp_subclasses != NULL;
 }
-#endif
 
 static inline size_t
 static_builtin_index_get(PyTypeObject *self)
@@ -107,43 +105,46 @@ static_builtin_state_get(PyInterpreterState *interp, PyTypeObject *self)
 
 /* For static types we store some state in an array on each interpreter. */
 static_builtin_state *
-_PyStaticType_GetState(PyTypeObject *self)
+_PyStaticType_GetState(PyInterpreterState *interp, PyTypeObject *self)
 {
     assert(self->tp_flags & _Py_TPFLAGS_STATIC_BUILTIN);
-    PyInterpreterState *interp = _PyInterpreterState_GET();
     return static_builtin_state_get(interp, self);
 }
 
+/* Set the type's per-interpreter state. */
 static void
-static_builtin_state_init(PyTypeObject *self)
+static_builtin_state_init(PyInterpreterState *interp, PyTypeObject *self)
 {
-    /* Set the type's per-interpreter state. */
-    PyInterpreterState *interp = _PyInterpreterState_GET();
+    if (!static_builtin_index_is_set(self)) {
+        static_builtin_index_set(self, interp->types.num_builtins_initialized);
+    }
+    static_builtin_state *state = static_builtin_state_get(interp, self);
 
     /* It should only be called once for each builtin type. */
-    assert(!static_builtin_index_is_set(self));
-
-    static_builtin_index_set(self, interp->types.num_builtins_initialized);
-    interp->types.num_builtins_initialized++;
-
-    static_builtin_state *state = static_builtin_state_get(interp, self);
+    assert(state->type == NULL);
     state->type = self;
+
     /* state->tp_subclasses is left NULL until init_subclasses() sets it. */
     /* state->tp_weaklist is left NULL until insert_head() or insert_after()
        (in weakrefobject.c) sets it. */
+
+    interp->types.num_builtins_initialized++;
 }
 
+/* Reset the type's per-interpreter state.
+   This basically undoes what static_builtin_state_init() did. */
 static void
-static_builtin_state_clear(PyTypeObject *self)
+static_builtin_state_clear(PyInterpreterState *interp, PyTypeObject *self)
 {
-    /* Reset the type's per-interpreter state.
-       This basically undoes what static_builtin_state_init() did. */
-    PyInterpreterState *interp = _PyInterpreterState_GET();
-
     static_builtin_state *state = static_builtin_state_get(interp, self);
+
+    assert(state->type != NULL);
     state->type = NULL;
     assert(state->tp_weaklist == NULL);  // It was already cleared out.
-    static_builtin_index_clear(self);
+
+    if (_Py_IsMainInterpreter(interp)) {
+        static_builtin_index_clear(self);
+    }
 
     assert(interp->types.num_builtins_initialized > 0);
     interp->types.num_builtins_initialized--;
@@ -4491,33 +4492,37 @@ clear_static_tp_subclasses(PyTypeObject *type)
     clear_subclasses(type);
 }
 
+static void
+clear_static_type_objects(PyInterpreterState *interp, PyTypeObject *type)
+{
+    if (_Py_IsMainInterpreter(interp)) {
+        Py_CLEAR(type->tp_dict);
+        Py_CLEAR(type->tp_bases);
+        Py_CLEAR(type->tp_mro);
+        Py_CLEAR(type->tp_cache);
+    }
+    clear_static_tp_subclasses(type);
+}
+
 void
-_PyStaticType_Dealloc(PyTypeObject *type)
+_PyStaticType_Dealloc(PyInterpreterState *interp, PyTypeObject *type)
 {
-    assert(!(type->tp_flags & Py_TPFLAGS_HEAPTYPE));
+    assert(type->tp_flags & _Py_TPFLAGS_STATIC_BUILTIN);
+    assert(_Py_IsImmortal((PyObject *)type));
 
     type_dealloc_common(type);
 
-    Py_CLEAR(type->tp_dict);
-    Py_CLEAR(type->tp_bases);
-    Py_CLEAR(type->tp_mro);
-    Py_CLEAR(type->tp_cache);
-    clear_static_tp_subclasses(type);
+    clear_static_type_objects(interp, type);
 
-    // PyObject_ClearWeakRefs() raises an exception if Py_REFCNT() != 0
-    if (Py_REFCNT(type) == 0) {
-        PyObject_ClearWeakRefs((PyObject *)type);
+    if (_Py_IsMainInterpreter(interp)) {
+        type->tp_flags &= ~Py_TPFLAGS_READY;
+        type->tp_flags &= ~Py_TPFLAGS_VALID_VERSION_TAG;
+        type->tp_version_tag = 0;
     }
 
-    type->tp_flags &= ~Py_TPFLAGS_READY;
-    type->tp_flags &= ~Py_TPFLAGS_VALID_VERSION_TAG;
-    type->tp_version_tag = 0;
-
-    if (type->tp_flags & _Py_TPFLAGS_STATIC_BUILTIN) {
-        _PyStaticType_ClearWeakRefs(type);
-        static_builtin_state_clear(type);
-        /* We leave _Py_TPFLAGS_STATIC_BUILTIN set on tp_flags. */
-    }
+    _PyStaticType_ClearWeakRefs(interp, type);
+    static_builtin_state_clear(interp, type);
+    /* We leave _Py_TPFLAGS_STATIC_BUILTIN set on tp_flags. */
 }
 
 
@@ -4564,7 +4569,8 @@ static PyObject *
 lookup_subclasses(PyTypeObject *self)
 {
     if (self->tp_flags & _Py_TPFLAGS_STATIC_BUILTIN) {
-        static_builtin_state *state = _PyStaticType_GetState(self);
+        PyInterpreterState *interp = _PyInterpreterState_GET();
+        static_builtin_state *state = _PyStaticType_GetState(interp, self);
         assert(state != NULL);
         return state->tp_subclasses;
     }
@@ -4574,8 +4580,9 @@ lookup_subclasses(PyTypeObject *self)
 int
 _PyType_HasSubclasses(PyTypeObject *self)
 {
+    PyInterpreterState *interp = _PyInterpreterState_GET();
     if (self->tp_flags & _Py_TPFLAGS_STATIC_BUILTIN &&
-            _PyStaticType_GetState(self) == NULL) {
+            _PyStaticType_GetState(interp, self) == NULL) {
         return 0;
     }
     if (lookup_subclasses(self) == NULL) {
@@ -6938,7 +6945,8 @@ type_ready_post_checks(PyTypeObject *type)
     else if (type->tp_dictoffset < (Py_ssize_t)sizeof(PyObject)) {
         if (type->tp_dictoffset + type->tp_basicsize <= 0) {
             PyErr_Format(PyExc_SystemError,
-                         "type %s has a tp_dictoffset that is too small");
+                         "type %s has a tp_dictoffset that is too small",
+                         type->tp_name);
         }
     }
     return 0;
@@ -7029,17 +7037,32 @@ PyType_Ready(PyTypeObject *type)
 }
 
 int
-_PyStaticType_InitBuiltin(PyTypeObject *self)
+_PyStaticType_InitBuiltin(PyInterpreterState *interp, PyTypeObject *self)
 {
     assert(_Py_IsImmortal((PyObject *)self));
     assert(!(self->tp_flags & Py_TPFLAGS_HEAPTYPE));
+    assert(!(self->tp_flags & Py_TPFLAGS_MANAGED_DICT));
+    assert(!(self->tp_flags & Py_TPFLAGS_MANAGED_WEAKREF));
 
+#ifndef NDEBUG
+    int ismain = _Py_IsMainInterpreter(interp);
+#endif
     if (self->tp_flags & Py_TPFLAGS_READY) {
+        assert(!ismain);
         assert(self->tp_flags & _Py_TPFLAGS_STATIC_BUILTIN);
+        assert(self->tp_flags & Py_TPFLAGS_VALID_VERSION_TAG);
+
+        static_builtin_state_init(interp, self);
+
+        /* Per-interpreter tp_subclasses is done lazily.
+           Otherwise we would initialize it here. */
+
         assert(_PyType_CheckConsistency(self));
         return 0;
     }
 
+    assert(ismain);
+
     self->tp_flags |= _Py_TPFLAGS_STATIC_BUILTIN;
     self->tp_flags |= Py_TPFLAGS_IMMUTABLETYPE;
 
@@ -7047,11 +7070,11 @@ _PyStaticType_InitBuiltin(PyTypeObject *self)
     self->tp_version_tag = NEXT_GLOBAL_VERSION_TAG++;
     self->tp_flags |= Py_TPFLAGS_VALID_VERSION_TAG;
 
-    static_builtin_state_init(self);
+    static_builtin_state_init(interp, self);
 
     int res = type_ready(self);
     if (res < 0) {
-        static_builtin_state_clear(self);
+        static_builtin_state_clear(interp, self);
     }
     return res;
 }
@@ -7065,7 +7088,8 @@ init_subclasses(PyTypeObject *self)
         return NULL;
     }
     if (self->tp_flags & _Py_TPFLAGS_STATIC_BUILTIN) {
-        static_builtin_state *state = _PyStaticType_GetState(self);
+        PyInterpreterState *interp = _PyInterpreterState_GET();
+        static_builtin_state *state = _PyStaticType_GetState(interp, self);
         state->tp_subclasses = subclasses;
         return subclasses;
     }
@@ -7080,7 +7104,8 @@ clear_subclasses(PyTypeObject *self)
        callers also test if tp_subclasses is NULL to check if a static type
        has no subclass. */
     if (self->tp_flags & _Py_TPFLAGS_STATIC_BUILTIN) {
-        static_builtin_state *state = _PyStaticType_GetState(self);
+        PyInterpreterState *interp = _PyInterpreterState_GET();
+        static_builtin_state *state = _PyStaticType_GetState(interp, self);
         Py_CLEAR(state->tp_subclasses);
         return;
     }
diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index 7537c12e92..6ae68cc20f 100644
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -14573,13 +14573,13 @@ _PyUnicode_InitGlobalObjects(PyInterpreterState *interp)
 PyStatus
 _PyUnicode_InitTypes(PyInterpreterState *interp)
 {
-    if (_PyStaticType_InitBuiltin(&EncodingMapType) < 0) {
+    if (_PyStaticType_InitBuiltin(interp, &EncodingMapType) < 0) {
         goto error;
     }
-    if (_PyStaticType_InitBuiltin(&PyFieldNameIter_Type) < 0) {
+    if (_PyStaticType_InitBuiltin(interp, &PyFieldNameIter_Type) < 0) {
         goto error;
     }
-    if (_PyStaticType_InitBuiltin(&PyFormatterIter_Type) < 0) {
+    if (_PyStaticType_InitBuiltin(interp, &PyFormatterIter_Type) < 0) {
         goto error;
     }
     return _PyStatus_OK();
@@ -15158,13 +15158,9 @@ unicode_is_finalizing(void)
 void
 _PyUnicode_FiniTypes(PyInterpreterState *interp)
 {
-    if (!_Py_IsMainInterpreter(interp)) {
-        return;
-    }
-
-    _PyStaticType_Dealloc(&EncodingMapType);
-    _PyStaticType_Dealloc(&PyFieldNameIter_Type);
-    _PyStaticType_Dealloc(&PyFormatterIter_Type);
+    _PyStaticType_Dealloc(interp, &EncodingMapType);
+    _PyStaticType_Dealloc(interp, &PyFieldNameIter_Type);
+    _PyStaticType_Dealloc(interp, &PyFormatterIter_Type);
 }
 
 
diff --git a/Objects/weakrefobject.c b/Objects/weakrefobject.c
index c1afe63ecf..aee79fc141 100644
--- a/Objects/weakrefobject.c
+++ b/Objects/weakrefobject.c
@@ -1017,9 +1017,9 @@ PyObject_ClearWeakRefs(PyObject *object)
  * or anything else.
  */
 void
-_PyStaticType_ClearWeakRefs(PyTypeObject *type)
+_PyStaticType_ClearWeakRefs(PyInterpreterState *interp, PyTypeObject *type)
 {
-    static_builtin_state *state = _PyStaticType_GetState(type);
+    static_builtin_state *state = _PyStaticType_GetState(interp, type);
     PyObject **list = _PyStaticType_GET_WEAKREFS_LISTPTR(state);
     while (*list != NULL) {
         /* Note that clear_weakref() pops the first ref off the type's