From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S268179AbUHFRGR (ORCPT ); Fri, 6 Aug 2004 13:06:17 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S268190AbUHFRDs (ORCPT ); Fri, 6 Aug 2004 13:03:48 -0400 Received: from fw.osdl.org ([65.172.181.6]:61650 "EHLO mail.osdl.org") by vger.kernel.org with ESMTP id S268200AbUHFQ6j (ORCPT ); Fri, 6 Aug 2004 12:58:39 -0400 Date: Fri, 6 Aug 2004 09:58:35 -0700 (PDT) From: Linus Torvalds To: Gene Heskett cc: Kernel Mailing List , Andrew Morton , Ingo Molnar , vda@port.imtp.ilyichevsk.odessa.ua, ak@suse.de, Chris Shoemaker , William Lee Irwin III Subject: Re: Possible dcache BUG In-Reply-To: <200408060751.07605.gene.heskett@verizon.net> Message-ID: References: <20040806073739.GA6617@elte.hu> <20040806004231.143c8bd2.akpm@osdl.org> <200408060751.07605.gene.heskett@verizon.net> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Fri, 6 Aug 2004, Gene Heskett wrote: > > Linus, Andrew, should I apply this patch too at the next remake? Might be worth it, but it's more important to see any oops at all, or lack of oopses.. > FWIW, I'm still up (20:38) this morning, and showing plenty (127+ > megs) of free memory. No crash, no odd log (other than samba > squawking about some option thats been changed & I haven't fixed the > smb.conf) so far. > > I'm beginning to like this test patch, Linus, thanks :) If the only thing you have done is add the list_del_init() debugging patch, then the only thing that has changed is really the access patterns to uncached memory. The original list_del_init() tries to only do a few single _writes_ to the dentries around it. The added debugging will do _reads_ (and thus bring it into the cache) of the dentry pointers of the dentries around it. If that change makes a real difference, I really only see two possibilities: - there really is a prefetch bug (or possibly, there's a bug in our prefetch fixup code, and the known prefetch bug just triggers the problem indirectly) - it just changes the timing enough that whatever bug you hit went away. Now, Chris Shoemaker reported dentry problems on a intel CPU and said that wli had seen something too, but I'm wondering whether Chris and wli might have been seeing the knfsd/xfs-related dentry bug that I found yesterday. So I think the prefetch theory is still alive, but we should check with Chris. Chris? Linus