Re: [PATCH v2 0/2] MR10271: Fixed thread safety issues with NtUserRegistClassExWOW and NtUserUnregisterClass.

[Yongjie Yao (＠yaoyongjie)]

On Mon Mar 23 13:43:49 2026 +0000, Rémi Bernon wrote:
> I don't think that's true? With the new user_lock scope, class creation
> and insertion in the process linked list happens atomically.
> Symmetrically, class destruction and removal from the process class list
> also happens atomically. So only scenario that
> `find_shared_session_object` can fail, is some general critical error.
> I don't think we care that much about what happens in that case, but it
> seems to me that class pointer is exclusively owned by the calling
> thread in all case until the class is inserted in the process list and
> user critical section exited, and the pointer can and should always be
> freed on error.
> Yes, you are right, with the new user_lock scope, only some critical error can cause find_shared_session_object failed, in that case, we should destroy the class, I will fixed it.

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/10271#note_133474