From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261801AbTJRTmt (ORCPT ); Sat, 18 Oct 2003 15:42:49 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261806AbTJRTmt (ORCPT ); Sat, 18 Oct 2003 15:42:49 -0400 Received: from dbl.q-ag.de ([80.146.160.66]:21461 "EHLO dbl.q-ag.de") by vger.kernel.org with ESMTP id S261801AbTJRTmp (ORCPT ); Sat, 18 Oct 2003 15:42:45 -0400 Message-ID: <3F919763.1000901@colorfullife.com> Date: Sat, 18 Oct 2003 21:41:23 +0200 From: Manfred Spraul User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4) Gecko/20030701 X-Accept-Language: en-us, en MIME-Version: 1.0 To: John Mock CC: linux-kernel@vger.kernel.org, Benjamin Herrenschmidt Subject: Re: PPC: slab error in cache_free_debugcheck() from sd_revalidate_disk Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org John wrote: >Hardware: PowerMac 8500 (see attached 'lspci' and 'dmesg' for details), > ADB keyboard/mouse, USB mouse, Sony 100ES monitor >Software: Debian 'Testing' branch > >Sorry, this may not be a very helpful bug report. Let me know what additional >information might be helpful and perhaps i can rig up a serial console for >this beast. > > [snip] >Oct 14 15:47:34 penngrove kernel: cddeec88: redzone 1: 0x0, redzone 2: 0x170fc2a5. >O > That looks like a bug Ben reported some time ago: The dma controller trashes the whole cacheline when it transfers data from disk to memory. I think the only solution is to disable redzoning for the kmalloc caches, or to use get_free_pages instead of kmalloc for the disk buffers,. -- Manfred