netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Kok, Auke" <auke-jan.h.kok@intel.com>
To: speedy <speedy@3d-io.com>
Cc: netdev@vger.kernel.org
Subject: Re: [BUG REPORT, 2.6.22] e1000: detected tx unit hang
Date: Mon, 21 Apr 2008 15:06:31 -0700	[thread overview]
Message-ID: <480D0FE7.4010609@intel.com> (raw)
In-Reply-To: <304359202.20080421235657@3d-io.com>

speedy wrote:
> Hello Auke,
> 
> Monday, April 21, 2008, 11:44:59 PM, you wrote:
> 
> KA> [dropped lkml from the Cc]
> 
> KA> speedy wrote:
>>> Hello Linux crew,
>>>
>>>         I've just switched the
>>>
>>>         Ethernet controller: Intel Corporation 82545GM Gigabit Ethernet Controller (rev 01)
>>>
>>>         netword card to an NForce 2 based motherboard and after a day
>>>         of work it got stuck with "detected tx unit hang" messages
>>>         showing in the console.
>>>
>>>         The card worked flawlessly under load in a different computer
>>>         for two years now, under the same/similar Ubuntu operating system.
>>>
>>>         /var/log/messages:  http://87.230.23.147/messages.txt
>>>         /proc/interrupts: http://87.230.23.147/proc_interrupts.txt
>>>         lspci -vv: http://87.230.23.147/lspcivv.txt
>>>
>>>
>>>         If more info is needed, let me know.
> 
> 
> KA> basically it's inserted into a new motherboard?
> 
> Yup.
> 
> I've changed the PCI slot in which the card is inserted (just out of
> hunch) and rebooted the server. I'll let you know if the problem
> happens again.
> 
> KA> what was the old motherboard?
> 
> QDI Legend KinetiZ 7B
> 
> http://www.qdigrp.com/qdisite/eng/products/K7B.htm
> 
> (had uptimes of 200+ days :)
> 
> KA> can you check the BIOS and disable things like "PCI Write combining" or
> KA> "Writeback" or any option looking similar to that?
> 
> I'm curious to see how often does the problem happen. I'll try such
> measures if it reproduces itself.
> 
> KA> It appears you hit an issue that is exposed by these adapters on some AMD/NVIDIA
> KA> chipset-based motherboards. This issue is known and we are investigating this and
> KA> have been for a long time. The root cause is still yet unknown however.
> 
> Does it also happen with newer AMD/NVIDIA motherboards? :(

yes, that's what the reports are.

it appears to be related to a bridge chip which is common on both older and newer
motherboards.

> KA> For some people disabling TSO helps to relieve the situation. You could give that
> KA> a try.
> 
> TSO? What is that and how to disable it? :)

TCP Segmentation offload - the hardware will split up the payload into MTU-size
fragments itself instead of doing it in the kernel.

ethtool -K ethX tso off


Auke

      reply	other threads:[~2008-04-21 22:07 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-04-21 20:52 [BUG REPORT, 2.6.22] e1000: detected tx unit hang speedy
2008-04-21 21:44 ` Kok, Auke
2008-04-21 21:56   ` Re[2]: " speedy
2008-04-21 22:06     ` Kok, Auke [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=480D0FE7.4010609@intel.com \
    --to=auke-jan.h.kok@intel.com \
    --cc=netdev@vger.kernel.org \
    --cc=speedy@3d-io.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).