Update retroactive comments from GH-117741 (segfault in `FutureIter_dealloc`) (GH-121638)

Address comments
This commit is contained in:
Savannah Ostrowski 2024-07-12 01:34:30 -07:00 committed by GitHub
parent e8c91d90ba
commit 65fededf9c
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 10 additions and 13 deletions

View File

@ -354,16 +354,6 @@ asend().throw()
..
.. date: 2024-04-13-18-59-25
.. gh-issue: 115874
.. nonce: c3xG-E
.. section: Core and Builtins
Fixed a possible segfault during garbage collection of
``_asyncio.FutureIter`` objects
..
.. date: 2024-04-13-16-55-53
.. gh-issue: 117536
.. nonce: xkVbfv
@ -883,6 +873,16 @@ Alex Waygood.
..
.. date: 2024-04-13-18-59-25
.. gh-issue: 115874
.. nonce: c3xG-E
.. section: Library
Fixed a possible segfault during garbage collection of
``_asyncio.FutureIter`` objects. Patch by Savannah Ostrowski.
..
.. date: 2024-04-13-01-45-15
.. gh-issue: 115060
.. nonce: IxoM03

View File

@ -1650,7 +1650,6 @@ FutureIter_dealloc(futureiterobject *it)
{
PyTypeObject *tp = Py_TYPE(it);
// FutureIter is a heap type so any subclass must also be a heap type.
assert(_PyType_HasFeature(tp, Py_TPFLAGS_HEAPTYPE));
PyObject *module = ((PyHeapTypeObject*)tp)->ht_module;
@ -1661,8 +1660,6 @@ FutureIter_dealloc(futureiterobject *it)
// GH-115874: We can't use PyType_GetModuleByDef here as the type might have
// already been cleared, which is also why we must check if ht_module != NULL.
// Due to this restriction, subclasses that belong to a different module
// will not be able to use the free list.
if (module && _PyModule_GetDef(module) == &_asynciomodule) {
state = get_asyncio_state(module);
}