[Bug 283747] kernel panic after telegraf service restart

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=283747

--- Comment #14 from Gleb Smirnoff <glebius@FreeBSD.org> ---
Thanks, Eero! This gives a bit more of insight.  So we see that the actual
panic happens in kern_mutex.c:592:

                owner = lv_mtx_owner(v);
                if (TD_IS_RUNNING(owner)) {

The TD_IS_RUNNING() dereferences td->td_state, which is exactly at 0x458 offset
inside struct thread.  So, lv_mtx_owner() gave us NULL on the previous line.
But a few lines above we see that the lock value is not MTX_UNOWNED.  This
means that the value has one of the flags set: (MTX_RECURSED | MTX_CONTESTED |
MTX_DESTROYED).  Apparently it is MTX_DESTROYED, cause other flags are interal
to the mutex subsystem and having them on would mean there is a bug in mutex
itself and we would see all different kinds of a panics.

So, we got valid inpcb pointer with valid cred pointer, but the cred itself has
already went through crfree_final().  We got a step closer to understanding the
problem.  Myself or Mark might come up with an idea, but getting more
information would be very helpful.  If you find a way to reproduce that, please
let us know.

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