Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1030366AbVI3QpL (ORCPT ); Fri, 30 Sep 2005 12:45:11 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1030368AbVI3QpL (ORCPT ); Fri, 30 Sep 2005 12:45:11 -0400 Received: from mx.pathscale.com ([64.160.42.68]:32988 "EHLO mx.pathscale.com") by vger.kernel.org with ESMTP id S1030366AbVI3QpK (ORCPT ); Fri, 30 Sep 2005 12:45:10 -0400 Subject: Re: 2.6.14-rc1-git-now still dying in mm/slab - this time line 1849 From: "Bryan O'Sullivan" To: Christoph Lameter Cc: Ravikiran G Thirumalai , Andrew Morton , vandrove@vc.cvut.cz, alokk@calsoftinc.com, linux-kernel@vger.kernel.org, manfred@colorfullife.com, shai@scalex86.org, ananth@in.ibm.com, ak@suse.de In-Reply-To: References: <20050919112912.18daf2eb.akpm@osdl.org> <20050919122847.4322df95.akpm@osdl.org> <20050919221614.6c01c2d1.akpm@osdl.org> <43301578.8040305@vc.cvut.cz> <20050928210245.GA3760@localhost.localdomain> <433C1999.2060201@vc.cvut.cz> <20050930054556.GA3599@localhost.localdomain> <20050929230540.6a8651fa.akpm@osdl.org> <20050930062853.GB3599@localhost.localdomain> <1128093382.10913.92.camel@serpentine.pathscale.com> Content-Type: text/plain Date: Fri, 30 Sep 2005 09:45:04 -0700 Message-Id: <1128098704.10913.100.camel@serpentine.pathscale.com> Mime-Version: 1.0 X-Mailer: Evolution 2.2.3 (2.2.3-2.fc4) Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 848 Lines: 28 On Fri, 2005-09-30 at 08:57 -0700, Christoph Lameter wrote: > On Fri, 30 Sep 2005, Bryan O'Sullivan wrote: > > > Kiran, your patch works for me, too. I can boot 2.6.14-rc2 with your > > patch, but not without it. > > The patch is not in rc2-mm2 right? Correct. > I can now reproduce it on a AMD64 > single processor with numa emulation (numa=fake=2). That's helpful for reproducing the problem. Thanks. > So all x86_64 NUMA > systems will throw these same stacktraces for rc2-mm2? I've only tried with HDAMA motherboards, but based on your report and Petr's, it seems somewhat likely.