On Tue, Jan 27, 2026 at 06:02:25PM -0800, Samuel Wu wrote:
> Thanks for the suggestion. I tried a few different setups, and now I'm
> fairly confident e5bf5ee26663 ("functionfs: fix the open/removal
> races") is the culprit. I did have to revert 6ca67378d0e7 ("convert
> functionfs") and c7747fafaba0 ("functionfs: switch to
> simple_remove_by_name()") to successfully build, but reverting only
> those two in isolation did not fix the issue.
Very interesting... Does 1544775687f0 (parent of e5bf5ee26663)
demonstrate that behaviour?
