From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jesper Krogh Subject: Re: niu driver - Transmit timed out - 2.6.29 Date: Sat, 28 Mar 2009 08:25:41 +0100 Message-ID: <49CDD0F5.1060802@krogh.cc> References: <49CB78A4.3020406@krogh.cc> <49CD2996.60502@krogh.cc> <49CD7266.1070002@sun.com> <49CDBE42.8030708@krogh.cc> <49CDC136.60709@sun.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: "netdev@vger.kernel.org" To: Matheos Worku Return-path: Received: from 2605ds1-ynoe.1.fullrate.dk ([90.184.12.24]:34726 "EHLO shrek.krogh.cc" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752327AbZC1HZw (ORCPT ); Sat, 28 Mar 2009 03:25:52 -0400 In-Reply-To: <49CDC136.60709@sun.com> Sender: netdev-owner@vger.kernel.org List-ID: Matheos Worku wrote: > Jesper Krogh wrote: >> Matheos Worku wrote: >>>> This is probably the interesting part: >>>> Mar 26 13:25:49 hest kernel: [25335.587191] niu 0000:84:00.0: niu: >>>> eth4: bits (40000000) of register RXDMA_CFIG1 would not clear, >>>> val[c0000000] >>> Jesper, >>> >>> One of the RX ring DMAs is failing to reset. I guess whatever is >>> hanging the TX side is affecting the RX side as well. Can you do >>> lspci on the function and its siblings? >> >> Like this(please guide me if that wasn't the correct lspci output): > > Jesper, > > I was wondering if you can get the register dump just after the NIC hangs. > > lspci -vvv -xxx -s 84:0 I will try to do that, but it involves more or less "putting a known bad driver" in production. And wait X days. (where X usually is less than 7 and more than 2). So if there is more debugging code that would be helpful to have in the driver/kernel then it would be preferrable to get it in at the same time, in order to reduce the amount of trial-and-error cycles. -- Jesper