From mboxrd@z Thu Jan 1 00:00:00 1970 From: Auke Kok Subject: Re: [PATCH 1/3] e1000: fix shared interrupt warning message Date: Sat, 17 Feb 2007 12:47:57 -0800 Message-ID: <45D769FD.4030508@intel.com> References: <20070216223913.26963.76602.stgit@gitlost.site> <20070216223926.26963.52634.stgit@gitlost.site> <45D767C8.3040508@garzik.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, "Brandeburg, Jesse" , "Kok, Auke" , "Ronciak, John" To: Jeff Garzik Return-path: Received: from mga01.intel.com ([192.55.52.88]:2547 "EHLO mga01.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932444AbXBQUsA (ORCPT ); Sat, 17 Feb 2007 15:48:00 -0500 In-Reply-To: <45D767C8.3040508@garzik.org> Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Jeff Garzik wrote: > Kok, Auke wrote: >> Signed-off-by: Jesse Brandeburg >> Signed-off-by: Auke Kok >> Signed-off-by: Andrew Morton >> --- >> >> drivers/net/e1000/e1000_main.c | 12 ++++++------ >> 1 files changed, 6 insertions(+), 6 deletions(-) > > applied 1-3, though patch #1 had a noticable lack of description. you > should describe your change, not just the condition that prompted it. > operation reordering is something that people shouldn't have to /try/ to > understand. Things should just be obvious, and if not, document why. For the record then: It is suspected that workarounds in the _up() routine of e1000 can cause the receive unit to be enabled before we're all done initializing the adapter data. An interrupt arriving before we're all done setting up obviously causes problems. To fix, we postpone enabling interrupts completely until the software intialization is finished. Thanks, Auke