[Bug 198014] Signals can lead to an inconsistency in PI mutex ownership

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=198014

--- Comment #9 from Konstantin Belousov <kib at FreeBSD.org> ---
I agree with all three patches, with one detail.

In the first patch, which does umtx_pi_disown() for empty queue, shouldn't the
case of pi->pi_owner != curthread be handled the same as it is done in the
non-empty queue case ?  The 'might' condition from the the comment really means
that the umutex and the kernel state are un-synchronized, and the 'userland
messed the mutex' situation is applicable in the same way there.

-- 
You are receiving this mail because:
You are the assignee for the bug.