From owner-freebsd-stable@FreeBSD.ORG Sat Jan 7 08:20:36 2006 Return-Path: X-Original-To: freebsd-stable@FreeBSD.org Delivered-To: freebsd-stable@FreeBSD.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 64B7C16A420 for ; Sat, 7 Jan 2006 08:20:36 +0000 (GMT) (envelope-from sirmoo@cowbert.2y.net) Received: from cowbert.2y.net (d46h180.public.uconn.edu [137.99.46.180]) by mx1.FreeBSD.org (Postfix) with SMTP id C9C1F43D45 for ; Sat, 7 Jan 2006 08:20:33 +0000 (GMT) (envelope-from sirmoo@cowbert.2y.net) Received: (qmail 97232 invoked by uid 1001); 7 Jan 2006 08:20:32 -0000 Date: Sat, 7 Jan 2006 03:20:32 -0500 From: "Peter C. Lai" To: PeterJeremy@optushome.com.au Message-ID: <20060107082032.GU326@cowbert.2y.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.6i Cc: freebsd-stable@FreeBSD.org Subject: Re: NIC card problems X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 07 Jan 2006 08:20:36 -0000 Peter Jeremy wrote: >Real DEC Tulip cards do this when running Tru64 as well. My guess is that >it's a bug in the NIC. (And it looks like AMDtek have copied it). Peter, Warner, Stefan, et al.: I just found this thread on the mailing list, and am responding to it, a year later :) I also believe the problem is a bug in the NIC as well, since the ADMTek 985 appears to not listen to the "automagic buffer underrun recovery" command. Silby added some patches to mbuf allocation in 2003 after stress testing dc(4), which improves the situation somewhat (ability to sustain the traffic longer) but doesn't solve it. While my system doesn't reboot (panic), it will often hang as a result of this. What happens then is that when the interface tries to transmit, a "No buffer space available" error occurs. If one can access the console, it can be rescued by bringing the interface down and then up again using ifconfig(8). This will reset the card and presumably flush the buffers. I wonder if any work has been done on the driver in -CURRENT (and I am too lazy to look), but in the next few weeks the machine is getting overhauled from 4.11 to 6 (reformat/reinstall) so we shall see if it does anything. -- Peter C. Lai Dept. of Neurobiology Yale University School of Medicine http://cowbert.2y.net/