Re: NetBSD 2.1.1: panic during cvs update

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Mon, Jul 07, 2008 at 01:59:58PM +0300, markucz%gmail.com@localhost wrote:
> Hello,
> 
> One of my machines suddenly panic'ed during 'cvs update'. That hasn't ever
> happened before. However, it's an old machine running NetBSD 2.1.1 GENERIC
> kernel and I suspect the cause is either 1) softdeps or 2) bad blocks. I'm
> inclined to think it's 2), but not long ago someone mentioned that softdeps
> can silently corrupt a filesystem. The filesystem trashed was /usr/src so
> it's not a big deal, I just ran newfs again. I've attached a file with ddb's
> output.

What would be interesting is the dmesg up to the panic. You can get
it with 'dmesg' in ddb (for next time :), but you can also get it
from the core dump, if it got written to disk (dmesg -M /path/to/dump).

Indeed, I've seen panics in softdep caused by write error at hardware
level.

> I'd be very grateful if someone would look into it and tell me what
> the cause may be. If it happens to be 2), is there a tool to report bad
> blocks? I only found bad144 and badsect but neither of them seems to be the
> right thing to use.

maybe smartmontools (in pkgsrc) can be of some help here.

-- 
Manuel Bouyer, LIP6, Universite Paris VI.           
Manuel.Bouyer%lip6.fr@localhost
     NetBSD: 26 ans d'experience feront toujours la difference
--