-
-
Notifications
You must be signed in to change notification settings - Fork 31.7k
gh-111926: Make weakrefs thread-safe in free-threaded builds #117168
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
We need to tag weakrefs and their referents
The `is_dead` check from the default build is insufficient, since the refcount may go to zero between when we check and when we incref. More generally, `Py_NewRef` is unsafe to use if you have a borrowed refernce.
Locking is handled in the implementation
@colesbury - Would you take another look at this when you get a chance, please? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is looking good. Some comments inline.
Additionally, proxy_dealloc
should remove the redundant PyObject_GC_UnTrack
inside the if statement. It's unnecessary and the plain read of self->wr_callback
is possibly a data race.
🤖 New build scheduled with the buildbot fleet by @colesbury for commit fece88d 🤖 If you want to schedule another build, you need to add the 🔨 test-with-refleak-buildbots label again. |
We cannot assert now that we clear weakrefs without holding the lock in _PyStaticType_ClearWeakRefs
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Co-authored-by: Erlend E. Aasland <erlend.aasland@protonmail.com>
🤖 New build scheduled with the buildbot fleet by @colesbury for commit 73466bf 🤖 If you want to schedule another build, you need to add the 🔨 test-with-refleak-buildbots label again. |
…ython#117168) Most mutable data is protected by a striped lock that is keyed on the referenced object's address. The weakref's hash is protected using the weakref's per-object lock. Note that this only affects free-threaded builds. Apart from some minor refactoring, the added code is all either gated by `ifdef`s or is a no-op (e.g. `Py_BEGIN_CRITICAL_SECTION`).
I've started seeing a test failure that bisects to this PR. Please see issue #118331. |
This makes weakrefs thread-safe in free-threaded builds. Please see the comment at the beginning of
Objects/weakrefobject.c
for an explanation of the approach.Note that this only affects free-threaded builds. Apart from some minor refactoring, the added code is all either gated by
ifdef
s or is a no-op (e.g.Py_BEGIN_CRITICAL_SECTION
).