09:09fdobridge: <karolherbst🐧🦀> @airlied how good is your knowledge on ttm and nouveau's use of it?
09:09fdobridge: <karolherbst🐧🦀> we have a regression from https://gitlab.freedesktop.org/drm/nouveau/-/commit/9bff18d13473a9fdf81d5158248472a9d8ecf2bd in 6.3
09:09fdobridge: <karolherbst🐧🦀> well.. "user-after-free" regression
09:10fdobridge: <karolherbst🐧🦀> https://gitlab.freedesktop.org/drm/nouveau/-/issues/213#note_1942777
09:12fdobridge: <airlied> Let me put that on my tomorrow list, I know the code reasonably okay, it just isnt the best way to interact with ttm
09:12fdobridge: <karolherbst🐧🦀> okay, cool
09:13fdobridge: <karolherbst🐧🦀> I know almost nothing of ttm and I kinda want to get that resolved as soon as possible
21:20fdobridge: <airlied> @karolherbst so built a master kernel, just spam it with CTS to reproduce?
21:21fdobridge: <karolherbst🐧🦀> @airlied did you enable KASAN as well? But yeah.. use deqp-runner with like 8 or more threads
21:22fdobridge: <karolherbst🐧🦀> shouldn't take too long to trigger
21:24fdobridge: <airlied> yup kasan built in
21:46fdobridge: <karolherbst🐧🦀> okay. I hope you'll trigger it quickly 🙂
21:46fdobridge: <karolherbst🐧🦀> worst case, throw more threads at the problem