From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1764894AbXJSRgJ (ORCPT ); Fri, 19 Oct 2007 13:36:09 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1755932AbXJSRfx (ORCPT ); Fri, 19 Oct 2007 13:35:53 -0400 Received: from rtr.ca ([76.10.145.34]:1962 "EHLO mail.rtr.ca" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S935424AbXJSRfw (ORCPT ); Fri, 19 Oct 2007 13:35:52 -0400 Message-ID: <4718EAF5.6000700@rtr.ca> Date: Fri, 19 Oct 2007 13:35:49 -0400 From: Mark Lord User-Agent: Thunderbird 2.0.0.6 (X11/20070728) MIME-Version: 1.0 To: Ray Lee Cc: Giangiacomo Mariotti , linux-kernel@vger.kernel.org Subject: Re: PROBLEM: 2.6.23.1 Freezes on GB data transfers References: <12bfabe40710170420m6882ea8eg3b8e6e7db72a0cdd@mail.gmail.com> <2c0942db0710190858n647192d6t270e43b2a3075320@mail.gmail.com> <4718DB8F.3070903@rtr.ca> <2c0942db0710191018n894a773mc2b0400067c0fdf@mail.gmail.com> In-Reply-To: <2c0942db0710191018n894a773mc2b0400067c0fdf@mail.gmail.com> 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 Ray Lee wrote: > On 10/19/07, Mark Lord wrote: > >> I believe this is now the forth report of blinking-leds lockup with 2.6.23.1. >> >> I wonder what's causing it? >> My system has this as well, totally random, once every day or so. >> New behaviour since 2.6.23-rc9 (I posted previously about this). > > Boy, there's just not a lot between 2.6.23-rc9 and 2.6.23 proper. > There are two things that kinda pop out, but this is at best a WAG. Yeah, I looked at the changes and it's all very innocent looking. More likely in my case, I suppose, is that the bug was there earlier and then some timing changed somewhere and now it triggers more often. One supporting evidence of that, is that with the powertop patches applied, it crashes much less often. And all that those do is adjust timings in various places. > First, are you on x86-32 and happen to have CONFIG_HIGHPTE set? Core2Duo, 2GB, x86-32, CONFIG_HIGHPTE is not set. > Second is another memory thing, where in filemap_fault we now do a > page_cache_release where we didn't before, but that appears to only be > in a case where we send a sigbus, so I wouldn't expect that to be > hitting. It seems to happen most often here on resume from suspend (RAM), and when hotplugging hardware. But it's infrequent enough that this may not be a reliable clue. Cheers