From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Fri, 3 May 2002 15:21:48 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Fri, 3 May 2002 15:21:47 -0400 Received: from pc132.utati.net ([216.143.22.132]:33182 "HELO merlin.webofficenow.com") by vger.kernel.org with SMTP id ; Fri, 3 May 2002 15:21:45 -0400 Content-Type: text/plain; charset=US-ASCII From: Rob Landley To: Sebastian Droege Subject: Re: sis900 ethernet driver/IP stack getting REALLY confused... Date: Fri, 3 May 2002 09:23:16 -0400 X-Mailer: KMail [version 1.3.1] Cc: linux-kernel@vger.kernel.org In-Reply-To: <20020503183407.AA9FC644@merlin.webofficenow.com> <20020503202352.04858a22.sebastian.droege@gmx.de> MIME-Version: 1.0 Content-Transfer-Encoding: 7BIT Message-Id: <20020503194504.B353D644@merlin.webofficenow.com> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Friday 03 May 2002 02:23 pm, Sebastian Droege wrote: > Hi, > I have a very similar problem... when I ping the box with the sis900 > ethernet adapter everything wents fine except ~3 % of the pings. This pings > need 10 times longer to get back to me than the other (eg. 72 ms <-> 1004 > ms) There is no other network activity on the network > And sometimes I get corrupted echo replies when there is traffic over the > sis900... > > I hope this helps somehow... > > Bye Yeah, that's what I'm seeing here too. I'm 99% certain it's not a hardware problem, because: A) It occurs on multiple motherboards (built in sis900). B) A soft reboot makes the problem go away. C) Unplugging the cat 5 cable and plugging it back in makes the problem start manifesting, until the next reboot. (On the systems I've seen it on anyway, I haven't tested otherwise healthy boxes to see if I can make them go nuts by temporarily unplugging their ethernet cables, but if anybody thinks it would help...) D) The problem itself is packets getting delayed ~10 seconds. (I think the ping corruption warning is just ping getting confused by the delay and profound out-of-sequenceness, but in any case 10 second old packets aren't interesting so if they went bad in storage it's a side issue...) That smells like a driver problem to me, but I can't honestly say I have too much of a clue in this area. I'm working with Advanced Clue Substitute at best... Rob