From patchwork Wed Sep 4 07:04:49 2013 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: NeilBrown X-Patchwork-Id: 2853543 Return-Path: X-Original-To: patchwork-linux-nfs@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork1.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.19.201]) by patchwork1.web.kernel.org (Postfix) with ESMTP id 8EA469F4D4 for ; Wed, 4 Sep 2013 07:05:18 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id 164502021E for ; Wed, 4 Sep 2013 07:05:17 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 21137201EF for ; Wed, 4 Sep 2013 07:05:15 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1762036Ab3IDHFN (ORCPT ); Wed, 4 Sep 2013 03:05:13 -0400 Received: from cantor2.suse.de ([195.135.220.15]:40998 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755775Ab3IDHFM (ORCPT ); Wed, 4 Sep 2013 03:05:12 -0400 Received: from relay1.suse.de (unknown [195.135.220.254]) by mx2.suse.de (Postfix) with ESMTP id A29D8A51BB; Wed, 4 Sep 2013 09:05:11 +0200 (CEST) Date: Wed, 4 Sep 2013 17:04:49 +1000 From: NeilBrown To: "Myklebust, Trond" Cc: NFS Subject: [PATCH - alt-2] NFSv4: Don't try to recover NFSv4 locks when they are lost. Message-ID: <20130904170449.7d2662cd@notabene.brown> 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 When an NFSv4 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 for now this patch just causes writes to fail. For NFSv4 (unlike v2/v3) there is a strong link between the lock and the write request so we can fairly easily fail any IO of the lock is gone. While some applications might not expect this, it is still safer than allowing the write to succeed. Because this is a fairly big change in behaviour a module parameter, "recover_locks", is introduced 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. Signed-off-by: NeilBrown --- This alternative uses a module parameter which defaults to the current, incorrect, behaviour. I suspect we don't want that one.. NeilBrown diff --git a/fs/nfs/nfs3proc.c b/fs/nfs/nfs3proc.c index f5c84c3..de0229b 100644 --- a/fs/nfs/nfs3proc.c +++ b/fs/nfs/nfs3proc.c @@ -826,9 +826,10 @@ static void nfs3_proc_read_setup(struct nfs_read_data *data, struct rpc_message msg->rpc_proc = &nfs3_procedures[NFS3PROC_READ]; } -static void nfs3_proc_read_rpc_prepare(struct rpc_task *task, struct nfs_read_data *data) +static int nfs3_proc_read_rpc_prepare(struct rpc_task *task, struct nfs_read_data *data) { rpc_call_start(task); + return 0; } static int nfs3_write_done(struct rpc_task *task, struct nfs_write_data *data) @@ -847,9 +848,10 @@ static void nfs3_proc_write_setup(struct nfs_write_data *data, struct rpc_messag msg->rpc_proc = &nfs3_procedures[NFS3PROC_WRITE]; } -static void nfs3_proc_write_rpc_prepare(struct rpc_task *task, struct nfs_write_data *data) +static int nfs3_proc_write_rpc_prepare(struct rpc_task *task, struct nfs_write_data *data) { rpc_call_start(task); + return 0; } static void nfs3_proc_commit_rpc_prepare(struct rpc_task *task, struct nfs_commit_data *data) diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h index ee81e35..a468b345 100644 --- a/fs/nfs/nfs4_fs.h +++ b/fs/nfs/nfs4_fs.h @@ -135,6 +135,7 @@ struct nfs4_lock_state { struct list_head ls_locks; /* Other lock stateids */ struct nfs4_state * ls_state; /* Pointer to open state */ #define NFS_LOCK_INITIALIZED 0 +#define NFS_LOCK_LOST 1 unsigned long ls_flags; struct nfs_seqid_counter ls_seqid; nfs4_stateid ls_stateid; diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 108a774..6ef4016 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -3905,15 +3905,19 @@ static void nfs4_proc_read_setup(struct nfs_read_data *data, struct rpc_message nfs41_init_sequence(&data->args.seq_args, &data->res.seq_res, 0); } -static void nfs4_proc_read_rpc_prepare(struct rpc_task *task, struct nfs_read_data *data) +static int nfs4_proc_read_rpc_prepare(struct rpc_task *task, struct nfs_read_data *data) { if (nfs4_setup_sequence(NFS_SERVER(data->header->inode), &data->args.seq_args, &data->res.seq_res, task)) - return; - nfs4_set_rw_stateid(&data->args.stateid, data->args.context, - data->args.lock_context, FMODE_READ); + return 0; + if (nfs4_set_rw_stateid(&data->args.stateid, data->args.context, + data->args.lock_context, FMODE_READ) == -EIO) + return -EIO; + if (unlikely(test_bit(NFS_CONTEXT_BAD, &data->args.context->flags))) + return -EIO; + return 0; } static int nfs4_write_done_cb(struct rpc_task *task, struct nfs_write_data *data) @@ -3988,15 +3992,19 @@ static void nfs4_proc_write_setup(struct nfs_write_data *data, struct rpc_messag nfs41_init_sequence(&data->args.seq_args, &data->res.seq_res, 1); } -static void nfs4_proc_write_rpc_prepare(struct rpc_task *task, struct nfs_write_data *data) +static int nfs4_proc_write_rpc_prepare(struct rpc_task *task, struct nfs_write_data *data) { if (nfs4_setup_sequence(NFS_SERVER(data->header->inode), &data->args.seq_args, &data->res.seq_res, task)) - return; - nfs4_set_rw_stateid(&data->args.stateid, data->args.context, - data->args.lock_context, FMODE_WRITE); + return 0; + if (nfs4_set_rw_stateid(&data->args.stateid, data->args.context, + data->args.lock_context, FMODE_WRITE) == -EIO) + return -EIO; + if (unlikely(test_bit(NFS_CONTEXT_BAD, &data->args.context->flags))) + return -EIO; + return 0; } static void nfs4_proc_commit_rpc_prepare(struct rpc_task *task, struct nfs_commit_data *data) @@ -5378,6 +5386,11 @@ static int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request return err; } +bool recover_locks = true; +module_param(recover_locks, bool, 0644); +MODULE_PARM_DESC(recovery_locks, + "If the server reports that a lock might be lost, " + "try to recovery it risking corruption."); static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request) { struct nfs_server *server = NFS_SERVER(state->inode); @@ -5389,6 +5402,10 @@ static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request err = nfs4_set_lock_state(state, request); if (err != 0) return err; + if (!recover_locks) { + set_bit(NFS_LOCK_LOST, &request->fl_u.nfs4_fl.owner->ls_flags); + return 0; + } do { if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0) return 0; diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index e22862f..1e26e8c 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -998,7 +998,9 @@ static int nfs4_copy_lock_stateid(nfs4_stateid *dst, fl_pid = lockowner->l_pid; spin_lock(&state->state_lock); lsp = __nfs4_find_lock_state(state, fl_owner, fl_pid, NFS4_ANY_LOCK_TYPE); - if (lsp != NULL && test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) != 0) { + if (lsp && test_bit(NFS_LOCK_LOST, &lsp->ls_flags)) + ret = -EIO; + else if (lsp != NULL && test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) != 0) { nfs4_stateid_copy(dst, &lsp->ls_stateid); ret = 0; smp_rmb(); @@ -1038,11 +1040,17 @@ 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) + /* A lost lock - don't even consider delegations */ + goto out; if (nfs4_copy_delegation_stateid(dst, state->inode, fmode)) goto out; - ret = nfs4_copy_lock_stateid(dst, state, lockowner); if (ret != -ENOENT) + /* nfs4_copy_delegation_stateid() didn't over-write + * dst, so it still has the lock stateid which we now + * choose to use. + */ goto out; ret = nfs4_copy_open_stateid(dst, state); out: diff --git a/fs/nfs/proc.c b/fs/nfs/proc.c index c041c41..a8f57c7 100644 --- a/fs/nfs/proc.c +++ b/fs/nfs/proc.c @@ -623,9 +623,10 @@ static void nfs_proc_read_setup(struct nfs_read_data *data, struct rpc_message * msg->rpc_proc = &nfs_procedures[NFSPROC_READ]; } -static void nfs_proc_read_rpc_prepare(struct rpc_task *task, struct nfs_read_data *data) +static int nfs_proc_read_rpc_prepare(struct rpc_task *task, struct nfs_read_data *data) { rpc_call_start(task); + return 0; } static int nfs_write_done(struct rpc_task *task, struct nfs_write_data *data) @@ -644,9 +645,10 @@ static void nfs_proc_write_setup(struct nfs_write_data *data, struct rpc_message msg->rpc_proc = &nfs_procedures[NFSPROC_WRITE]; } -static void nfs_proc_write_rpc_prepare(struct rpc_task *task, struct nfs_write_data *data) +static int nfs_proc_write_rpc_prepare(struct rpc_task *task, struct nfs_write_data *data) { rpc_call_start(task); + return 0; } static void nfs_proc_commit_rpc_prepare(struct rpc_task *task, struct nfs_commit_data *data) diff --git a/fs/nfs/read.c b/fs/nfs/read.c index 70a26c6..31db5c3 100644 --- a/fs/nfs/read.c +++ b/fs/nfs/read.c @@ -513,9 +513,10 @@ static void nfs_readpage_release_common(void *calldata) void nfs_read_prepare(struct rpc_task *task, void *calldata) { struct nfs_read_data *data = calldata; - NFS_PROTO(data->header->inode)->read_rpc_prepare(task, data); - if (unlikely(test_bit(NFS_CONTEXT_BAD, &data->args.context->flags))) - rpc_exit(task, -EIO); + int err; + err = NFS_PROTO(data->header->inode)->read_rpc_prepare(task, data); + if (err) + rpc_exit(task, err); } static const struct rpc_call_ops nfs_read_common_ops = { diff --git a/fs/nfs/write.c b/fs/nfs/write.c index f1bdb72..7816801 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -1265,9 +1265,10 @@ EXPORT_SYMBOL_GPL(nfs_pageio_reset_write_mds); void nfs_write_prepare(struct rpc_task *task, void *calldata) { struct nfs_write_data *data = calldata; - NFS_PROTO(data->header->inode)->write_rpc_prepare(task, data); - if (unlikely(test_bit(NFS_CONTEXT_BAD, &data->args.context->flags))) - rpc_exit(task, -EIO); + int err; + err = NFS_PROTO(data->header->inode)->write_rpc_prepare(task, data); + if (err) + rpc_exit(task, err); } void nfs_commit_prepare(struct rpc_task *task, void *calldata) diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 8651574..c71e12b 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1419,12 +1419,12 @@ struct nfs_rpc_ops { void (*read_setup) (struct nfs_read_data *, struct rpc_message *); void (*read_pageio_init)(struct nfs_pageio_descriptor *, struct inode *, const struct nfs_pgio_completion_ops *); - void (*read_rpc_prepare)(struct rpc_task *, struct nfs_read_data *); + int (*read_rpc_prepare)(struct rpc_task *, struct nfs_read_data *); int (*read_done) (struct rpc_task *, struct nfs_read_data *); void (*write_setup) (struct nfs_write_data *, struct rpc_message *); void (*write_pageio_init)(struct nfs_pageio_descriptor *, struct inode *, int, const struct nfs_pgio_completion_ops *); - void (*write_rpc_prepare)(struct rpc_task *, struct nfs_write_data *); + int (*write_rpc_prepare)(struct rpc_task *, struct nfs_write_data *); int (*write_done) (struct rpc_task *, struct nfs_write_data *); void (*commit_setup) (struct nfs_commit_data *, struct rpc_message *); void (*commit_rpc_prepare)(struct rpc_task *, struct nfs_commit_data *);