Re: port-amd64/39283: Kernel crash on Dell Poweredge 2950

[fredrik%netbsd.se@localhost]

The following reply was made to PR port-amd64/39283; it has been noted by GNATS.

From: fredrik%netbsd.se@localhost
To: "David Holland" <dholland-bugs%netbsd.org@localhost>
Cc: fredrik%netbsd.se@localhost,
 gnats-bugs%netbsd.org@localhost,
 port-amd64-maintainer%netbsd.org@localhost,
 gnats-admin%netbsd.org@localhost,
 netbsd-bugs%netbsd.org@localhost
Subject: Re: port-amd64/39283: Kernel crash on Dell Poweredge 2950
Date: Mon, 1 Sep 2008 08:01:01 +0200 (CEST)

 > On Mon, Aug 25, 2008 at 08:05:33AM +0200, fredrik%netbsd.se@localhost wrote:
 >  > >  My guess is a bad callout, but that doesn't narrow it down very
 > much.
 >  >
 >  > The machine has been running stable (with -1 in boot.cfg) for the last
 > 21
 >  > days, with a lot of network, I/O and CPU load. What can we do to
 > collect
 >  > more data to help the troubleshooting?
 >
 > Probably the best thing to do is build a kernel with DIAGNOSTIC; and
 > on the conjecture that it may be a bad callout, add this patch, which
 > should cause it to panic in a recognizable way instead of crashing if
 > that's the problem.
 >
 > Or it may panic somewhere else, if you weren't previously running a
 > DIAGNOSTIC kernel.
 >
 > Then, wait for it to crash. :-/
 >
 > Index: kern_timeout.c
 > ===================================================================
 > RCS file: /cvsroot/src/sys/kern/kern_timeout.c,v
 > retrieving revision 1.41
 > diff -u -p -r1.41 kern_timeout.c
 > --- kern_timeout.c   2 Jul 2008 14:47:34 -0000       1.41
 > +++ kern_timeout.c   31 Aug 2008 22:26:51 -0000
 > @@ -722,6 +722,7 @@ callout_softclock(void *v)
 >              cc->cc_active = c;
 >
 >              mutex_spin_exit(&cc->cc_lock);
 > +            KASSERT(func != NULL);
 >              if (!mpsafe) {
 >                      KERNEL_LOCK(1, NULL);
 >                      (*func)(arg);
 >
 >
 > --
 > David A. Holland
 > dholland%netbsd.org@localhost
 >
 
 We are already running with DEBUG and DIAGNOSTIC, hopefully it will be a
 little more verbose after your patch.
 
 The current options:
 options DEBUG
 options DIAGNOSTIC
 options LOCKDEBUG
 makeoptions DEBUG="-g"
 
 Regards
 Fredrik