Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 11 Sep 2000 12:19:25 -0700 (PDT)
From:      David Kirchner <dpk@parodius.com>
To:        Alfred Perlstein <bright@wintelcom.net>
Cc:        freebsd-questions@FreeBSD.ORG
Subject:   Re: Strange procfs bug(?) 3.2-RELEASE
Message-ID:  <Pine.BSF.4.21.0009111218440.80992-100000@pentarou.parodius.com>
In-Reply-To: <20000911120640.J12231@fw.wintelcom.net>

next in thread | previous in thread | raw e-mail | index | archive | help

Is this related? :

"Don't call calcru() on a swapped-out process. calcru() access p_stats,
which is in U-area."

That's from CVSweb at file procfs_status.c, revision 1.14

On Mon, 11 Sep 2000, Alfred Perlstein wrote:

> * David Kirchner <dpk@parodius.com> [000911 11:49] wrote:
> > 
> > I've been trying to track down a procfs bug which has semi-recently popped
> > up. I'm not 100% certain it is a bug in procfs, but it seems to be the
> > most likely candidate.
> > 
> > We first noticed the bug when we used the 'killall' command to kill off
> > processes. Whenever we'd run it, the machine would panic. We assumed it
> > was because we might have been killing something that was in a weird
> > state. But later on when I was trying to figure out something else on the
> > same box, I was running 'more /proc/*/status' and it panic'd. Just now, I
> > ran a little script:
> > 
> > cd /proc
> > for i in *
> > do
> >         ps -uxp $i >> /root/killallfindbug.log 2>&1
> >         sync;sync
> >         cat /proc/$i/status >> /root/killallfindbug.log 2>&1
> >         sync;sync
> > done
> > 
> > which also caused a panic (double panic, this time). The last line in the
> > log file was:
> > 
> > www   1756  0.0  0.8  5900 4364  ??  S    11:14AM   0:00.22
> > /usr/local/bin/http
> > 
> > (That would have been /usr/local/bin/httpd -f
> > /usr/local/www/server/conf/httpd.conf if I added -w to the ps flags)
> > 
> > This is on a 3.2-RELEASE machine. We haven't rebuilt the kernels on these
> > boxes in a long time, and they haven't been modified (been watching with
> > tripwire) to our knowledge anyhow.
> 
> Yes, there's a problem with procfs in 3.2 when you stress it heavily
> like that, my suggestion is to get 4-stable running, or at least try
> to get 3.5.1 up and running, but I'm unsure if it's fixed in 3.5.1.
> 
> -Alfred
> 



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0009111218440.80992-100000>