From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: with ECARTIS (v1.0.0; list xfs); Mon, 23 Jun 2008 11:37:50 -0700 (PDT) Received: from cuda.sgi.com (cuda1.sgi.com [192.48.168.28]) by oss.sgi.com (8.12.11.20060308/8.12.11/SuSE Linux 0.7) with ESMTP id m5NIbjBD016502 for ; Mon, 23 Jun 2008 11:37:46 -0700 Received: from gir.skynet.ie (localhost [127.0.0.1]) by cuda.sgi.com (Spam Firewall) with ESMTP id 1C467D29B5A for ; Mon, 23 Jun 2008 11:38:44 -0700 (PDT) Received: from gir.skynet.ie (gir.skynet.ie [193.1.99.77]) by cuda.sgi.com with ESMTP id u7x8dbemztD7Gjnq for ; Mon, 23 Jun 2008 11:38:44 -0700 (PDT) Date: Mon, 23 Jun 2008 19:38:41 +0100 From: Mel Gorman Subject: Re: [2.6.26-rc7] shrink_icache from pagefault locking (nee: nfsd hangs for a few sec)... Message-ID: <20080623183840.GA1824@csn.ul.ie> References: <6278d2220806220256g674304ectb945c14e7e09fede@mail.gmail.com> <6278d2220806220258p28de00c1x615ad7b2f708e3f8@mail.gmail.com> <20080622221930.GA11558@disturbed> <20080623002415.GB21597@csn.ul.ie> <20080623072227.GB11986@infradead.org> MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-15 Content-Disposition: inline In-Reply-To: <20080623072227.GB11986@infradead.org> Sender: xfs-bounce@oss.sgi.com Errors-to: xfs-bounce@oss.sgi.com List-Id: xfs To: Christoph Hellwig Cc: Daniel J Blueman , Christoph Lameter , Linus Torvalds , Alexander Beregalov , Linux Kernel , xfs@oss.sgi.com, david@fromorbit.com On (23/06/08 03:22), Christoph Hellwig didst pronounce: > On Mon, Jun 23, 2008 at 01:24:15AM +0100, Mel Gorman wrote: > > In that case, have you any theory as to why this circular dependency is > > being reported now but wasn't before 2.6.26-rc1? I'm beginning to wonder > > if the bisecting fingering the zonelist modifiation is just a > > co-incidence. > > I've seen this traces since lockdep was added when running xfsqa. > Oh right, so this isn't even 2.6.26-rc1 as such. It's an older problem that seems to be happening in more cases now, right? At this point, I believe the bisection fingering the zonelist modification was a co-incidence as reclaim behaviour at least is equivilant although catching the memory leak early was a lucky positive outcome. It's still not clear why the circular warning is happening more regularly now but it's "something else". Considering the number of changes made to NFS, XFS, reclaim and other areas since, I'm not sure how to go about finding the real underlying problem or if it can be dealt with in a trivial manner. -- Mel Gorman Part-time Phd Student Linux Technology Center University of Limerick IBM Dublin Software Lab