Page MenuHomeFreeBSD

Code must not unlock a mutex while owning the thread lock.
ClosedPublic

Authored by kib on Jan 13 2020, 10:24 AM.
Tags
None
Referenced Files
F157985844: D23150.id66682.diff
Wed, May 27, 4:38 AM
F157980100: D23150.id.diff
Wed, May 27, 2:59 AM
Unknown Object (File)
Mon, May 25, 12:23 AM
Unknown Object (File)
Thu, May 21, 7:55 PM
Unknown Object (File)
Wed, May 20, 9:38 PM
Unknown Object (File)
Wed, May 6, 2:12 AM
Unknown Object (File)
Mon, May 4, 4:09 PM
Unknown Object (File)
Thu, Apr 30, 6:50 PM

Details

Summary

Sponsored by: Mellanox Technologies

Diff Detail

Repository
rS FreeBSD src repository - subversion
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

sys/compat/linuxkpi/common/src/linux_rcu.c has same bug. Please also fix there.

Should thread_lock() assert Giant not owned?

Should thread_lock() assert Giant not owned?

I do not think so. It might make sense to make a witness assert that no sleepable mutexes are owned, but really the check for recursion in the thread_lock_flags() catches the problematic cases.

Also change linux RCU code.

This revision is now accepted and ready to land.Jan 13 2020, 10:56 AM