Date: Sun, 2 Jun 1996 22:56:06 +0200 (MET DST) From: Robert Eckardt <roberte@beta.mep.ruhr-uni-bochum.de> To: mbarkah@hemi.com (Ade Barkah) Cc: dennis@etinc.com, hackers@FreeBSD.ORG, freebsd-isp@FreeBSD.ORG Subject: Re: Nasty Routing BUG in 2.1R Message-ID: <199606022056.WAA03344@beta.mep.ruhr-uni-bochum.de> In-Reply-To: <199606021241.MAA00954@hemi.com> from "Ade Barkah" at Jun 2, 96 12:41:17 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> Dennis wrote:
> > route delete 200.11.1.17
> >
> > /* This causes a panic in the function rtfree(). If you dont delete
> > the route then the host address is unreachable (you get an arpresolve
> > error). */
>
> We've hit this bug (or a variation of it) three times in the past
> week (see my previous message to -hackers.)
>
> Our setup looks as follows:
>
> ---------------- --------------
> | FreeBSD 2.1R | <-- Ethernet --> | Cisco .254 |
> ---------------- --------------
> | <--- PPP connection
> ------------
> | PPP/.252 |
> ------------
>
> The .252 machine is linked via PPP to the Cisco. It also has its own
> /26 subnet. For some reason, the .252 machine (a Telebit router)
> doesn't respond to ping x.x.x.252, but it does respond to pings to
> IP addresses on its own subnet. Probably a configuration problem at
> the Telebit end.
>
> To crash the FreeBSD system, simply type 'ping x.x.x.252', which
> generates "/kernel: arpresolve: can't allocate llinfo". I think at
> this time it also generates an invalid/incomplete routing entry.
> Sometime later the system tries to purge this entry, and we get
> panic:rtfree.
Hi,
I had a very similar problem a while ago -- actually it still exists.
(running 2.0 and 2.1-RELEASE)
We have a modem connected to our server and some of us have our
'own' IP addresses for the machines at home. If you are on the
net it works fine, but when you terminate the SLIP connection and
e.g. some mail arrives you get the bad routing entry.
134.147.6.18 134.147.6.1 UGHS 0 0 ed1
% ping 134.147.6.18
PING 134.147.6.18 (134.147.6.18): 56 data bytes
2 packets transmitted, 0 packets received, 100% packet loss
134.147.6.18 134.147.6.18 UGHMS 1 2 ed1
However, I observed that the machine does not crash when I _change_
the route instead of deleting it.
% route change 134.147.6.18 134.147.6.1
change host 134.147.6.18: gateway 134.147.6.1
134.147.6.18 134.147.6.18 UGHMS 0 2 ed1
Therefore, I run every 20min the following ugly sh-script:
netstat -rn | grep UGH | while read n1 n2 nn
do
if [ $n1 = $n2 ]; then
logger -p cron.notice "Resetting route for $n1"
/sbin/route change $n1 134.147.6.1
fi
done
I see that this is just a work around, but at least the net runs.
BTW, a SysVr3.2 (ISC-2.2) machine produces also those 'wrong' route
entries. However it does not crash when it is deleted.
Ciao,
Robert
--
Robert Eckardt (
Ruhr-Universitaet Bochum, Inst.f.Theor.Physik, NB6/169 )
Universitaetsstrasse 150, D-44780 Bochum, Germany ----X---8----
Telefon: +49 234 700 3709, Telefax: +49 234 700 3712 8
E-Mail: RobertE@MEP.Ruhr-Uni-Bochum.de --------8----
URL: http://WWW.MEP.Ruhr-Uni-Bochum.de/
>>> Fuer die einen ist es bloss ein Betriebssystem, <<<
>>> fuer die anderen ist es der laengste Virus der Welt. .... Windows 95 <<<
Privat:
Steinbrink 22, D-45355 Essen, Germany -====- Telefon: +49 201 678602
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199606022056.WAA03344>