Re: Recent kernel crashes at shutdown

[christos%astron.com@localhost (Christos Zoulas)]

In article <51195988.1000504%anduin.org.uk@localhost>,
Dave Tyson  <dtyson%anduin.org.uk@localhost> wrote:
>I having been running 6.99.16 on a i386 desktop system using the releng
>snapshot from 28th January. Its been completely stable and I have been
>using it with XFCE4 with no problems until today. I added a bunch of
>extra packages and started them from rc.conf. On shutting down the
>system it paniced with an assertion failure:
>
> Feb 11 19:58:11 darkstar savecore: reboot after panic: panic: kernel
>diagnostic
>assertion "lwpcnt >= 0" failed: file
>"/home/builds/ab/HEAD/src/sys/kern/kern_uid
>info.c", line 217
>
>$NetBSD: kern_uidinfo.c,v 1.7 2012/06/09 02:55:32 christos Exp $
>
>
>This happened just after swap and the shutdown hooks had been run.
>Sometimes the system hangs, but mostly it takes a dump and reboots. A
>bit of investigation showed that if HAL is running then the system will
>crash, but if it isn't then shutdown proceeds OK.
>
>I tried with the latest releng kernel from 10th Feb and that exhibits
>the same issue.
>
>kernel which first showed the problem:
>
>NetBSD 6.99.16 (GENERIC) #0: Mon Jan 28 16:05: 30 UTC 2013
>builds%b8.netbsd.org@localhost:/home/builds/ab/HEAD/i386/201301281320Z-obj/home/builds/ab/HEAD/src/sys/arch/i386/compile/GENERIC
>
>I have core dumps from the crash on both the Jan 28/Feb 10 kernels if
>needed.
>
>Worth a PR ?

I think there is one. Someone else is seeing the same problem with hald,
and I've given him http://www.netbsd.org/~christos/lwp-hald.diff, but we
could not figure out why on his machine hald ended up running with the
wrong uid! I could not duplicate the issue on mine.

christos