From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: [Bug #11382] e1000e: 2.6.27-rc1 corrupts EEPROM/NVM Date: Tue, 23 Sep 2008 14:05:19 -0700 (PDT) Message-ID: <20080923.140519.268233735.davem@davemloft.net> References: <21d7e9970809221826i76081719pdd749237580fc68e@mail.gmail.com> <20080922.185902.80812984.davem@davemloft.net> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: kernel-testers-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: Content-Type: Text/Plain; charset="us-ascii" To: jkosina-AlSwsSmVLrQ@public.gmane.org Cc: david.vrabel-kQvG35nSl+M@public.gmane.org, airlied-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org, rjw-KKrjLPT3xs0@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, kernel-testers-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, chrisl-pghWNbHTmq7QT0dZR+AlfA@public.gmane.org From: Jiri Kosina Date: Tue, 23 Sep 2008 16:29:16 +0200 (CEST) > On Mon, 22 Sep 2008, David Miller wrote: > > > So I went through the changes from 2.6.27-rc5 until the SHA1 > > ID ec0c15afb41fd9ad45b53468b60db50170e22346 and there were > > definitely no E1000 or E1000E changes during that time. > > Some recent comments on [1] seem to indicate that this is somehow coupled > into prior problems/panics with Intel graphics. My current suspicion in all of this is either the GEM kernel patches or recent X server. However, the eeprom/nvram programming sequence seems non-trivial on the e1000e. You have to execute a set of precise register writes and register polls to successfully write things out to the nvram. This makes something like a random scribble out to MMIO space less likely to cause this problem. Is there some linear mapping of the nvram that could be written to on these cards? From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755687AbYIWVFs (ORCPT ); Tue, 23 Sep 2008 17:05:48 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752976AbYIWVFd (ORCPT ); Tue, 23 Sep 2008 17:05:33 -0400 Received: from 74-93-104-97-Washington.hfc.comcastbusiness.net ([74.93.104.97]:47731 "EHLO sunset.davemloft.net" rhost-flags-OK-FAIL-OK-OK) by vger.kernel.org with ESMTP id S1755448AbYIWVFb (ORCPT ); Tue, 23 Sep 2008 17:05:31 -0400 Date: Tue, 23 Sep 2008 14:05:19 -0700 (PDT) Message-Id: <20080923.140519.268233735.davem@davemloft.net> To: jkosina@suse.cz Cc: david.vrabel@csr.com, airlied@gmail.com, rjw@sisk.pl, linux-kernel@vger.kernel.org, kernel-testers@vger.kernel.org, chrisl@vmware.com Subject: Re: [Bug #11382] e1000e: 2.6.27-rc1 corrupts EEPROM/NVM From: David Miller In-Reply-To: References: <21d7e9970809221826i76081719pdd749237580fc68e@mail.gmail.com> <20080922.185902.80812984.davem@davemloft.net> X-Mailer: Mew version 6.1 on Emacs 22.1 / Mule 5.0 (SAKAKI) Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org From: Jiri Kosina Date: Tue, 23 Sep 2008 16:29:16 +0200 (CEST) > On Mon, 22 Sep 2008, David Miller wrote: > > > So I went through the changes from 2.6.27-rc5 until the SHA1 > > ID ec0c15afb41fd9ad45b53468b60db50170e22346 and there were > > definitely no E1000 or E1000E changes during that time. > > Some recent comments on [1] seem to indicate that this is somehow coupled > into prior problems/panics with Intel graphics. My current suspicion in all of this is either the GEM kernel patches or recent X server. However, the eeprom/nvram programming sequence seems non-trivial on the e1000e. You have to execute a set of precise register writes and register polls to successfully write things out to the nvram. This makes something like a random scribble out to MMIO space less likely to cause this problem. Is there some linear mapping of the nvram that could be written to on these cards?