Return-Path: Received: from mx1.redhat.com ([209.132.183.28]:54353 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755901Ab0IUV62 (ORCPT ); Tue, 21 Sep 2010 17:58:28 -0400 Date: Tue, 21 Sep 2010 14:58:23 -0700 From: Jeff Layton To: Trond Myklebust Cc: "J. Bruce Fields" , linux-nfs@vger.kernel.org Subject: Re: [bfields@pop.test.fieldses.org: all f08d6374 Merge branch 'nfs-for-next' into linux-next results] Message-ID: <20100921145823.2ec2a81c@corrin.poochiereds.net> In-Reply-To: <1285101981.30222.27.camel@heimdal.trondhjem.org> References: <20100918171546.GA9859@fieldses.org> <1284831680.2787.1.camel@heimdal.trondhjem.org> <20100919184549.GD32071@fieldses.org> <20100920130106.GB4580@fieldses.org> <20100920131025.GC4580@fieldses.org> <1285091815.30222.19.camel@heimdal.trondhjem.org> <20100921111657.330e7838@corrin.poochiereds.net> <20100921192005.GC4385@fieldses.org> <1285097787.30222.24.camel@heimdal.trondhjem.org> <1285099232.30222.25.camel@heimdal.trondhjem.org> <20100921202411.GA10570@fieldses.org> <1285101981.30222.27.camel@heimdal.trondhjem.org> Content-Type: text/plain; charset=US-ASCII Sender: linux-nfs-owner@vger.kernel.org List-ID: MIME-Version: 1.0 On Tue, 21 Sep 2010 16:46:21 -0400 Trond Myklebust wrote: > On Tue, 2010-09-21 at 16:24 -0400, J. Bruce Fields wrote: > > On Tue, Sep 21, 2010 at 04:00:32PM -0400, Trond Myklebust wrote: > > > On Tue, 2010-09-21 at 15:36 -0400, Trond Myklebust wrote: > > > > On Tue, 2010-09-21 at 15:20 -0400, J. Bruce Fields wrote: > > > > > All I need to do is mount nfs4.1 and run cthon -s. The last output I > > > > > see from the test is: > > > > > > > > > > check for proper open/unlink operation > > > > > nfsjunk files before unlink: > > > > > > > > > > > > > Oh... I bet I see what it is. > > > > > > > > It's that damned res->sr_slotid = NFS4_MAX_SLOT_TABLE and related > > > > initialisation junk that's biting us in the arse again... > > > > > > > > I'll fix it... > > > > > > Does this work? > > > > Yup. > > It seems to fix things for me too... Once you mentioned NFSv4.1 it was > easy to reproduce the bug. > > OK. I'll merge these into nfs-for-2.6.37... Thanks for fixing it. I should have mentioned that the 4.1 parts were written using the "cargo-cult" programming method of copying what unlink does, and that they needed careful review. Mea culpa! -- Jeff Layton