On Sun, Jul 03, 2022 at 01:57:20PM +0900, Dae R. Jeong wrote: > Hello, > > We observed an issue "INFO: task hung in blk_trace_remove" during > fuzzing. > > Unfortunately, we have not found a reproducer for the crash yet. We > will inform you if we have any update on this crash. > > The crash information is attached at the end of this email. This looks like it is due to locking an on the /dev/sg node instead of the more usual (and more tested) block device node. It might make sense to run with lockdep enabled to get a better grip on potential locking issues here.