From patchwork Wed Sep 4 00:49:46 2013 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: NeilBrown X-Patchwork-Id: 2853459 Return-Path: X-Original-To: patchwork-linux-nfs@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork2.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.19.201]) by patchwork2.web.kernel.org (Postfix) with ESMTP id 5A580C0AB5 for ; Wed, 4 Sep 2013 00:50:03 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id 0A6BF20451 for ; Wed, 4 Sep 2013 00:50:02 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id BA40420444 for ; Wed, 4 Sep 2013 00:50:00 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1761549Ab3IDAt7 (ORCPT ); Tue, 3 Sep 2013 20:49:59 -0400 Received: from cantor2.suse.de ([195.135.220.15]:34217 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1761538Ab3IDAt7 (ORCPT ); Tue, 3 Sep 2013 20:49:59 -0400 Received: from relay2.suse.de (unknown [195.135.220.254]) by mx2.suse.de (Postfix) with ESMTP id B496EA52D7; Wed, 4 Sep 2013 02:49:57 +0200 (CEST) Date: Wed, 4 Sep 2013 10:49:46 +1000 From: NeilBrown To: "Myklebust, Trond" Cc: NFS , "Schumaker, Bryan" Subject: Re: [PATCH/RFC] Don't try to recover NFS locks when they are lost. Message-ID: <20130904104946.1be45e52@notabene.brown> In-Reply-To: <1378233802.6410.34.camel@leira.trondhjem.org> References: <20130815123604.47755509@notabene.brown> <1378233802.6410.34.camel@leira.trondhjem.org> X-Mailer: Claws Mail 3.9.0 (GTK+ 2.24.18; x86_64-suse-linux-gnu) Mime-Version: 1.0 Sender: linux-nfs-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-nfs@vger.kernel.org X-Spam-Status: No, score=-9.3 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_HI, RP_MATCHES_RCVD, T_TVD_MIME_EPI, UNPARSEABLE_RELAY autolearn=unavailable version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on mail.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP On Tue, 3 Sep 2013 18:43:23 +0000 "Myklebust, Trond" wrote: > On Thu, 2013-08-15 at 12:36 +1000, NeilBrown wrote: > > > > When an NFS (V4 specifically) client loses contact with the server it can > > lose any locks that it holds. > > Currently when it reconnects to the server it simply tries to reclaim > > those locks. This might succeed even though some other client has held and > > released a lock in the mean time. So the first client might think the file > > is unchanged, but it isn't. This isn't good. > > > > If, when recovery happens, the locks cannot be claimed because some other > > client still holds the lock, then we get a message in the kernel logs, but > > the client can still write. So two clients can both think they have a lock > > and can both write at the same time. This is equally not good. > > > > There was a patch a while ago > > http://comments.gmane.org/gmane.linux.nfs/41917 > > > > which tried to address some of this, but it didn't seem to go anywhere. > > That patch would also send a signal to the process. That might be useful > > but I'm really just interested in failing the writes. > > For NFSv4 (unlike v2/v3) there is a strong link between the lock and the > > write request so we can fairly easily fail an IO of the lock is gone. > > > > The patch below attempts to do this. Does it make sense? > > Because this is a fairly big change I introduces a module parameter > > "recover_locks" which defaults to true (the current behaviour) but can be set > > to "false" to tell the client not to try to recover things that were lost. > > > > Comments? > > I think this patch is close to being usable. A couple of questions, > though: > > 1. What happens if another process' open() causes us to receive a > delegation after NFS_LOCK_LOST has been set on our lock stateid, > but before we call nfs4_set_rw_stateid()? Good point. I think we need to check for NFS_LOCK_LOST before checking for a delegation. Does the incremental patch below look OK? It takes a spinlock in the case where we have a delegation and hold some locks which it didn't have to take before. Is that a concern? > 2. Shouldn't we clear NFS_LOCK_LOST at some point? It looks to me > as if a process which sees the EIO, and decides to recover by > calling close(), reopen()ing the file and then locking it again, > might find NFS_LOCK_LOST still being set. NFS_LOCK_LOST is per nfs4_lock_state which should be freed by nfs4_fl_release_lock(). So when the files is closed, the locks a dropped, and the structure holding the NFS_LOCK_LOST flag will go away. Or did I miss something? Thanks, NeilBrown diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index 4d103ff..bb1fd5d 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -1040,10 +1040,11 @@ static int nfs4_copy_open_stateid(nfs4_stateid *dst, struct nfs4_state *state) int nfs4_select_rw_stateid(nfs4_stateid *dst, struct nfs4_state *state, fmode_t fmode, const struct nfs_lockowner *lockowner) { - int ret = 0; + int ret = nfs4_copy_lock_stateid(dst, state, lockowner); + if (ret == -EIO) + goto out; if (nfs4_copy_delegation_stateid(dst, state->inode, fmode)) goto out; - ret = nfs4_copy_lock_stateid(dst, state, lockowner); if (ret != -ENOENT) goto out; ret = nfs4_copy_open_stateid(dst, state);