Moved around when the locking and unlocking of cuda_mem_map_mutex happens so it doesn't call a function that locks it while still holding the lock.
Fixes T85089
Details
Details
- Reviewers
Patrick Mours (pmoursnv) - Maniphest Tasks
- T84734: Denoising Data with large tiles causes instant crash and exit.
T85089: Crash when rendering with CUDA / OptiX - Commits
- rC88c43b68e6d9: Fix T85089: Crash when rendering scene that does not fit into GPU memory with…
rB4fbeb3e6be3e: Fix T85089: Crash when rendering scene that does not fit into GPU memory with…
Diff Detail
Diff Detail
- Repository
- rB Blender
Event Timeline
Comment Actions
Looks good to me. My main concern with locking directly inside generic_free would have been the double lock when called from tex_free, but you addressed by unlocking before, so everything looks functional.
I'll do some more testing to be sure and commit later. Thank you for this!