From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753926AbXLAMrV (ORCPT ); Sat, 1 Dec 2007 07:47:21 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1752967AbXLAMrK (ORCPT ); Sat, 1 Dec 2007 07:47:10 -0500 Received: from smtp-vbr10.xs4all.nl ([194.109.24.30]:4337 "EHLO smtp-vbr10.xs4all.nl" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752602AbXLAMrJ (ORCPT ); Sat, 1 Dec 2007 07:47:09 -0500 Message-ID: <475157CA.2040206@xs4all.nl> Date: Sat, 01 Dec 2007 13:47:06 +0100 From: Udo van den Heuvel User-Agent: Thunderbird 2.0.0.9 (Windows/20071031) MIME-Version: 1.0 To: Cyrill Gorcunov CC: LKML Subject: Re: possible BUG: unable to handle kernel NULL pointer dereference at virtual address 00000000 References: <47481590.8090009@xs4all.nl> <200711241951.37817.rjw@sisk.pl> <474870F5.4050504@xs4all.nl> <474C3659.600@xs4all.nl> <20071127164929.GA7279@cvg> <474C56C8.80206@xs4all.nl> <20071127175503.GC7279@cvg> <47513EE5.30703@xs4all.nl> <20071201121220.GA28875@cvg> In-Reply-To: <20071201121220.GA28875@cvg> X-Enigmail-Version: 0.95.5 OpenPGP: id=8300CC02 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Cyrill Gorcunov wrote: > [Udo van den Heuvel - Sat, Dec 01, 2007 at 12:00:53PM +0100] > | I only had teh bug once, that was the reason for my post. > | Since then the system has been busy, with intervals, doing kernel > | compiles and no repeat of the BUG yet... > | It's a VIA EN12000 with 1GB of RAM. How could I increase the chances of > | hitting it? > > if my guessing is right - the only chance to hit it faster - is to increase > files' activity i.e. compiling the kernel involves a lot of file being readed > thru dcache system. First we have to locate the bug on a pure kernel > and 'case it's a rare thing to happen... well I don't really know how to > get it up to known point. So i think you could play with a pure kernel > by compiling it or run several copies of 'grep' thru kernel searching > the same pattern like "grep -r -n for ./*" in kernel sources. But Udo, > I'm not a kernel specialist so my suggestions could be not really usefull ;) Ok, maybe someone else can improve on this: I now have 12 screen'ed sessions running in /usr/src/linux with: while true; do grep -r -n for ./*; done Every now and then I can check dmesg. Would that be enough? Udo