From owner-freebsd-current  Fri Oct  3 02:20:35 1997
Return-Path: <owner-freebsd-current>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id CAA02806
          for current-outgoing; Fri, 3 Oct 1997 02:20:35 -0700 (PDT)
Received: from oxmail4.ox.ac.uk (oxmail4.ox.ac.uk [163.1.2.33])
          by hub.freebsd.org (8.8.7/8.8.7) with SMTP id CAA02801
          for <freebsd-current@freebsd.org>; Fri, 3 Oct 1997 02:20:28 -0700 (PDT)
Received: from njl2.materials.ox.ac.uk by oxmail4 with SMTP (PP);
          Fri, 3 Oct 1997 10:19:41 +0100
Received: by njl2.materials.ox.ac.uk (950413.SGI.8.6.12/940406.SGI)	for freebsd-current@FreeBSD.ORG 
          id KAA12629; Fri, 3 Oct 1997 10:19:27 +0100
Date: Fri, 3 Oct 1997 10:19:27 +0100
From: neil.long@materials.oxford.ac.uk (Neil J Long)
Message-Id: <9710031019.ZM12627@njl2.materials.ox.ac.uk>
X-Mailer: Z-Mail-SGI (3.2S.3 08feb96 MediaMail)
To: freebsd-current@freebsd.org
Subject: Network related lockups
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Sender: owner-freebsd-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

Hello

I have a 486/33 running RELENG_2_2 and continue to experience network related
lockups whose nature has changed with the recent kernel network mods.

The machine is currently on a thin-wire segment and runs arpwatch. There is
very high activity on this segment at night while the amanda server (SGI
Challenge S) backups up a lot of machines but the total data is around 2G over
a couple of hours. The PC also gets backed up during this spell and the only
solution I have found so far is to use at to killall arpwatch and re-start it
later and reduce the number of amanda 'dumpers' per client to 1 from 2.

I used to get messages related to routed (some may remember my previous
postings) but the current changes give

/kernel: ed0: NIC memory corrupt - invalid packet length 2800
/kernel: ed0: NIC memory corrupt - invalid packet length 2776

This is using a NE2000 clone, I will switch back again to a ep0 3Com card and
see what happens tonight.

It seems that arpwatch is just too much for the network card to cope with when
the network is heavily utilised and I will switch to using another SG box to
monitor for users hell-bent on misconfiguring their W95 boxes.

Just for info - I never seem to be able to get any more debug info to go on _ I
only got the above error by unloading the lkm screen saver otherwise the lockup
seems total needing a power cycle.

Cheers
Neil