LOR in pf on 6.2

[Max Laier]

Am Di, 29.01.2008, 19:37, schrieb Bill Marquette:
> Been having some kernel locks on some machines at work, not sure if
> this LOR is related (and I see an XXX LOR comment in the code too, so
> I'm guessing it's been seen before) although I have certainly had some
> of the machines lock during bootup right around the place that this
> LOR prints out.
>
> Jan 29 12:11:46 1st 0xc09e4420 pf task mtx (pf task mtx) @
> /usr/src/sys/contrib/pf/net/pf.c:6386
> Jan 29 12:11:46 2nd 0xc0a5142c udp (udp) @
> /usr/src/sys/contrib/pf/net/pf.c:2744
>
> I've got a couple machines that reliably lock w/in 24 hours or so of
> uptime, I can easily test any source patches on them.  All machines
> are built off of FreeBSD 6.2-RELEASE-p10 sources.  Thanks

>From the pf.conf(5) in RELENG_6_2:

BUGS
  Due to a lock order reversal (LOR) with the socket layer, the use of the
  group and user filter parameter in conjuction with a Giant-free netstack
  can result in a deadlock.  If you have to use group or user you must set
  debug.mpsafenet to ``0'' from the loader(8), for the moment.  This work-
  around will still produce the LOR, but Giant will protect from the dead-
  lock.

A better fix is in RELENG_7 ... backporting won't make much sense.

-- 
/"\  Best regards,                      | mlaier at freebsd.org
\ /  Max Laier                          | ICQ #67774661
 X   http://pf4freebsd.love2party.net/  | mlaier at EFnet
/ \  ASCII Ribbon Campaign              | Against HTML Mail and News