[Bug 220217] deadlock on enc and pf

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=220217

--- Comment #2 from vegeta at tuxpowered.net ---
Under normal circumstances the kernel does not panic, it just goes 100% usage
on bird or on interrupt. The packet which causes it is bird connecting to BGP
router on the other side, at least on the dump I obtained from this particular
crash.

I have the screenshot of crash with debug-enabled kernel but the stack trace is
so long, that the original message is not visible anymore. Does kernel debugger
console support scrolling? Pressing scroll-lock did not have any effect.

The line 411 where the error happens is:

372 __rw_rlock(volatile uintptr_t *c, const char *file, int line)
...
411         KASSERT(rw_wowner(rw) != curthread,
412             ("rw_rlock: wlock already held for %s @ %s:%d",                 
413             rw->lock_object.lo_name, file, line));

Can I obtain information about the other process holding the lock? (struct
thread)*inp->inp_lock->rw_lock points to the same process "swi4: clock (0)". I
know the live debugger has some helpers to show locks, but I would need to kill
the machine again for that and set up serial console. I would prefer to use
existing memory dump if possible.

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