18 Apr
2025
18 Apr
'25
4:47 p.m.
On Fri Apr 18 16:47:26 2025 +0000, Etaash Mathamsetty wrote:
Doesn't it resolve the previous race condition, since the accum value will get reset on every init? It makes the race less likely, but it can still happen if we get a destruction and recreation of relative motion while an event handler is running:
1. Thread A: Entering relative motion handler 2. Thread B: Destroys relative motion 3. Thread B: Creates new relative motion (zeroes accum_x/y) 4. Thread A: Relative motion handler (from now destroyed relative motion) continues, updates accum_x/y. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/7806#note_101412