From cf4ab538f1516606d3ae730dce15d6f33d96b7e1 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 8 Mar 2013 12:56:37 -0500 Subject: [PATCH 01/47] NFSv4: Fix the string length returned by the idmapper Functions like nfs_map_uid_to_name() and nfs_map_gid_to_group() are expected to return a string without any terminating NUL character. Regression introduced by commit 57e62324e469e092ecc6c94a7a86fe4bd6ac5172 (NFS: Store the legacy idmapper result in the keyring). Reported-by: Dave Chiluk Signed-off-by: Trond Myklebust Cc: Bryan Schumaker Cc: stable@vger.kernel.org [>=3.4] --- fs/nfs/idmap.c | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-) diff --git a/fs/nfs/idmap.c b/fs/nfs/idmap.c index dc0f98dfa717..c516da5873fd 100644 --- a/fs/nfs/idmap.c +++ b/fs/nfs/idmap.c @@ -726,9 +726,9 @@ static int nfs_idmap_legacy_upcall(struct key_construction *cons, return ret; } -static int nfs_idmap_instantiate(struct key *key, struct key *authkey, char *data) +static int nfs_idmap_instantiate(struct key *key, struct key *authkey, char *data, size_t datalen) { - return key_instantiate_and_link(key, data, strlen(data) + 1, + return key_instantiate_and_link(key, data, datalen, id_resolver_cache->thread_keyring, authkey); } @@ -738,6 +738,7 @@ static int nfs_idmap_read_and_verify_message(struct idmap_msg *im, struct key *key, struct key *authkey) { char id_str[NFS_UINT_MAXLEN]; + size_t len; int ret = -ENOKEY; /* ret = -ENOKEY */ @@ -747,13 +748,15 @@ static int nfs_idmap_read_and_verify_message(struct idmap_msg *im, case IDMAP_CONV_NAMETOID: if (strcmp(upcall->im_name, im->im_name) != 0) break; - sprintf(id_str, "%d", im->im_id); - ret = nfs_idmap_instantiate(key, authkey, id_str); + /* Note: here we store the NUL terminator too */ + len = sprintf(id_str, "%d", im->im_id) + 1; + ret = nfs_idmap_instantiate(key, authkey, id_str, len); break; case IDMAP_CONV_IDTONAME: if (upcall->im_id != im->im_id) break; - ret = nfs_idmap_instantiate(key, authkey, im->im_name); + len = strlen(im->im_name); + ret = nfs_idmap_instantiate(key, authkey, im->im_name, len); break; default: ret = -EINVAL; From 4376c94618c26225e69e17b7c91169c45a90b292 Mon Sep 17 00:00:00 2001 From: fanchaoting Date: Thu, 21 Mar 2013 09:15:30 +0800 Subject: [PATCH 02/47] pnfs-block: removing DM device maybe cause oops when call dev_remove when pnfs block using device mapper,if umounting later,it maybe cause oops. we apply "1 + sizeof(bl_umount_request)" memory for msg->data, the memory maybe overflow when we do "memcpy(&dataptr [sizeof(bl_msg)], &bl_umount_request, sizeof(bl_umount_request))", because the size of bl_msg is more than 1 byte. Signed-off-by: fanchaoting Cc: stable@vger.kernel.org Signed-off-by: Trond Myklebust --- fs/nfs/blocklayout/blocklayoutdm.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fs/nfs/blocklayout/blocklayoutdm.c b/fs/nfs/blocklayout/blocklayoutdm.c index 737d839bc17b..6fc7b5cae92b 100644 --- a/fs/nfs/blocklayout/blocklayoutdm.c +++ b/fs/nfs/blocklayout/blocklayoutdm.c @@ -55,7 +55,8 @@ static void dev_remove(struct net *net, dev_t dev) bl_pipe_msg.bl_wq = &nn->bl_wq; memset(msg, 0, sizeof(*msg)); - msg->data = kzalloc(1 + sizeof(bl_umount_request), GFP_NOFS); + msg->len = sizeof(bl_msg) + bl_msg.totallen; + msg->data = kzalloc(msg->len, GFP_NOFS); if (!msg->data) goto out; @@ -66,7 +67,6 @@ static void dev_remove(struct net *net, dev_t dev) memcpy(msg->data, &bl_msg, sizeof(bl_msg)); dataptr = (uint8_t *) msg->data; memcpy(&dataptr[sizeof(bl_msg)], &bl_umount_request, sizeof(bl_umount_request)); - msg->len = sizeof(bl_msg) + bl_msg.totallen; add_wait_queue(&nn->bl_wq, &wq); if (rpc_queue_upcall(nn->bl_device_pipe, msg) < 0) { From a073dbff359f4741013ae4b8395f5364c5e00b48 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 20 Mar 2013 12:34:32 -0400 Subject: [PATCH 03/47] NFSv4.1: Fix a race in pNFS layoutcommit We need to clear the NFS_LSEG_LAYOUTCOMMIT bits atomically with the NFS_INO_LAYOUTCOMMIT bit, otherwise we may end up with situations where the two are out of sync. The first half of the problem is to ensure that pnfs_layoutcommit_inode clears the NFS_LSEG_LAYOUTCOMMIT bit through pnfs_list_write_lseg. We still need to keep the reference to those segments until the RPC call is finished, so in order to make it clear _where_ those references come from, we add a helper pnfs_list_write_lseg_done() that cleans up after pnfs_list_write_lseg. Signed-off-by: Trond Myklebust Acked-by: Benny Halevy Cc: stable@vger.kernel.org --- fs/nfs/nfs4proc.c | 14 -------------- fs/nfs/pnfs.c | 19 ++++++++++++++++++- 2 files changed, 18 insertions(+), 15 deletions(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index b2671cb0f901..6ccdd4fd9b59 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -6416,22 +6416,8 @@ nfs4_layoutcommit_done(struct rpc_task *task, void *calldata) static void nfs4_layoutcommit_release(void *calldata) { struct nfs4_layoutcommit_data *data = calldata; - struct pnfs_layout_segment *lseg, *tmp; - unsigned long *bitlock = &NFS_I(data->args.inode)->flags; pnfs_cleanup_layoutcommit(data); - /* Matched by references in pnfs_set_layoutcommit */ - list_for_each_entry_safe(lseg, tmp, &data->lseg_list, pls_lc_list) { - list_del_init(&lseg->pls_lc_list); - if (test_and_clear_bit(NFS_LSEG_LAYOUTCOMMIT, - &lseg->pls_flags)) - pnfs_put_lseg(lseg); - } - - clear_bit_unlock(NFS_INO_LAYOUTCOMMITTING, bitlock); - smp_mb__after_clear_bit(); - wake_up_bit(bitlock, NFS_INO_LAYOUTCOMMITTING); - put_rpccred(data->cred); kfree(data); } diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index 48ac5aad6258..3d900916fd41 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -1746,11 +1746,27 @@ static void pnfs_list_write_lseg(struct inode *inode, struct list_head *listp) list_for_each_entry(lseg, &NFS_I(inode)->layout->plh_segs, pls_list) { if (lseg->pls_range.iomode == IOMODE_RW && - test_bit(NFS_LSEG_LAYOUTCOMMIT, &lseg->pls_flags)) + test_and_clear_bit(NFS_LSEG_LAYOUTCOMMIT, &lseg->pls_flags)) list_add(&lseg->pls_lc_list, listp); } } +static void pnfs_list_write_lseg_done(struct inode *inode, struct list_head *listp) +{ + struct pnfs_layout_segment *lseg, *tmp; + unsigned long *bitlock = &NFS_I(inode)->flags; + + /* Matched by references in pnfs_set_layoutcommit */ + list_for_each_entry_safe(lseg, tmp, listp, pls_lc_list) { + list_del_init(&lseg->pls_lc_list); + pnfs_put_lseg(lseg); + } + + clear_bit_unlock(NFS_INO_LAYOUTCOMMITTING, bitlock); + smp_mb__after_clear_bit(); + wake_up_bit(bitlock, NFS_INO_LAYOUTCOMMITTING); +} + void pnfs_set_lo_fail(struct pnfs_layout_segment *lseg) { pnfs_layout_io_set_failed(lseg->pls_layout, lseg->pls_range.iomode); @@ -1795,6 +1811,7 @@ void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data) if (nfss->pnfs_curr_ld->cleanup_layoutcommit) nfss->pnfs_curr_ld->cleanup_layoutcommit(data); + pnfs_list_write_lseg_done(data->args.inode, &data->lseg_list); } /* From 24956804349ca0eadcdde032d65e8c00b4214096 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 20 Mar 2013 13:03:00 -0400 Subject: [PATCH 04/47] NFSv4.1: Always clear the NFS_INO_LAYOUTCOMMIT in layoutreturn Note that clearing NFS_INO_LAYOUTCOMMIT is tricky, since it requires you to also clear the NFS_LSEG_LAYOUTCOMMIT bits from the layout segments. The only two sites that need to do this are the ones that call pnfs_return_layout() without first doing a layout commit. Signed-off-by: Trond Myklebust Acked-by: Benny Halevy Cc: stable@vger.kernel.org --- fs/nfs/nfs4filelayout.c | 1 - fs/nfs/pnfs.c | 35 +++++++++++++++++++++++++++-------- 2 files changed, 27 insertions(+), 9 deletions(-) diff --git a/fs/nfs/nfs4filelayout.c b/fs/nfs/nfs4filelayout.c index 49eeb044c109..4fb234d3aefb 100644 --- a/fs/nfs/nfs4filelayout.c +++ b/fs/nfs/nfs4filelayout.c @@ -129,7 +129,6 @@ static void filelayout_fenceme(struct inode *inode, struct pnfs_layout_hdr *lo) { if (!test_and_clear_bit(NFS_LAYOUT_RETURN, &lo->plh_flags)) return; - clear_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(inode)->flags); pnfs_return_layout(inode); } diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index 3d900916fd41..5044142c1216 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -417,6 +417,16 @@ should_free_lseg(struct pnfs_layout_range *lseg_range, lo_seg_intersecting(lseg_range, recall_range); } +static bool pnfs_lseg_dec_and_remove_zero(struct pnfs_layout_segment *lseg, + struct list_head *tmp_list) +{ + if (!atomic_dec_and_test(&lseg->pls_refcount)) + return false; + pnfs_layout_remove_lseg(lseg->pls_layout, lseg); + list_add(&lseg->pls_list, tmp_list); + return true; +} + /* Returns 1 if lseg is removed from list, 0 otherwise */ static int mark_lseg_invalid(struct pnfs_layout_segment *lseg, struct list_head *tmp_list) @@ -430,11 +440,8 @@ static int mark_lseg_invalid(struct pnfs_layout_segment *lseg, */ dprintk("%s: lseg %p ref %d\n", __func__, lseg, atomic_read(&lseg->pls_refcount)); - if (atomic_dec_and_test(&lseg->pls_refcount)) { - pnfs_layout_remove_lseg(lseg->pls_layout, lseg); - list_add(&lseg->pls_list, tmp_list); + if (pnfs_lseg_dec_and_remove_zero(lseg, tmp_list)) rv = 1; - } } return rv; } @@ -777,6 +784,21 @@ send_layoutget(struct pnfs_layout_hdr *lo, return lseg; } +static void pnfs_clear_layoutcommit(struct inode *inode, + struct list_head *head) +{ + struct nfs_inode *nfsi = NFS_I(inode); + struct pnfs_layout_segment *lseg, *tmp; + + if (!test_and_clear_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)) + return; + list_for_each_entry_safe(lseg, tmp, &nfsi->layout->plh_segs, pls_list) { + if (!test_and_clear_bit(NFS_LSEG_LAYOUTCOMMIT, &lseg->pls_flags)) + continue; + pnfs_lseg_dec_and_remove_zero(lseg, head); + } +} + /* * Initiates a LAYOUTRETURN(FILE), and removes the pnfs_layout_hdr * when the layout segment list is empty. @@ -808,6 +830,7 @@ _pnfs_return_layout(struct inode *ino) /* Reference matched in nfs4_layoutreturn_release */ pnfs_get_layout_hdr(lo); empty = list_empty(&lo->plh_segs); + pnfs_clear_layoutcommit(ino, &tmp_list); pnfs_mark_matching_lsegs_invalid(lo, &tmp_list, NULL); /* Don't send a LAYOUTRETURN if list was initially empty */ if (empty) { @@ -820,8 +843,6 @@ _pnfs_return_layout(struct inode *ino) spin_unlock(&ino->i_lock); pnfs_free_lseg_list(&tmp_list); - WARN_ON(test_bit(NFS_INO_LAYOUTCOMMIT, &nfsi->flags)); - lrp = kzalloc(sizeof(*lrp), GFP_KERNEL); if (unlikely(lrp == NULL)) { status = -ENOMEM; @@ -1458,7 +1479,6 @@ static void pnfs_ld_handle_write_error(struct nfs_write_data *data) dprintk("pnfs write error = %d\n", hdr->pnfs_error); if (NFS_SERVER(hdr->inode)->pnfs_curr_ld->flags & PNFS_LAYOUTRET_ON_ERROR) { - clear_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(hdr->inode)->flags); pnfs_return_layout(hdr->inode); } if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags)) @@ -1613,7 +1633,6 @@ static void pnfs_ld_handle_read_error(struct nfs_read_data *data) dprintk("pnfs read error = %d\n", hdr->pnfs_error); if (NFS_SERVER(hdr->inode)->pnfs_curr_ld->flags & PNFS_LAYOUTRET_ON_ERROR) { - clear_bit(NFS_INO_LAYOUTCOMMIT, &NFS_I(hdr->inode)->flags); pnfs_return_layout(hdr->inode); } if (!test_and_set_bit(NFS_IOHDR_REDO, &hdr->flags)) From 240286725d854331422cb15957f8d9bf2741d4e3 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 20 Mar 2013 13:23:33 -0400 Subject: [PATCH 05/47] NFSv4.1: Add a helper pnfs_commit_and_return_layout In order to be able to safely return the layout in nfs4_proc_setattr, we need to block new uses of the layout, wait for all outstanding users of the layout to complete, commit the layout and then return it. This patch adds a helper in order to do all this safely. Signed-off-by: Trond Myklebust Cc: Boaz Harrosh --- fs/nfs/nfs4proc.c | 2 +- fs/nfs/pnfs.c | 27 +++++++++++++++++++++++++++ fs/nfs/pnfs.h | 6 ++++++ 3 files changed, 34 insertions(+), 1 deletion(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 6ccdd4fd9b59..26431cf62ddb 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -2632,7 +2632,7 @@ nfs4_proc_setattr(struct dentry *dentry, struct nfs_fattr *fattr, int status; if (pnfs_ld_layoutret_on_setattr(inode)) - pnfs_return_layout(inode); + pnfs_commit_and_return_layout(inode); nfs_fattr_init(fattr); diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index 5044142c1216..4bdffe0ba025 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -866,6 +866,33 @@ _pnfs_return_layout(struct inode *ino) } EXPORT_SYMBOL_GPL(_pnfs_return_layout); +int +pnfs_commit_and_return_layout(struct inode *inode) +{ + struct pnfs_layout_hdr *lo; + int ret; + + spin_lock(&inode->i_lock); + lo = NFS_I(inode)->layout; + if (lo == NULL) { + spin_unlock(&inode->i_lock); + return 0; + } + pnfs_get_layout_hdr(lo); + /* Block new layoutgets and read/write to ds */ + lo->plh_block_lgets++; + spin_unlock(&inode->i_lock); + filemap_fdatawait(inode->i_mapping); + ret = pnfs_layoutcommit_inode(inode, true); + if (ret == 0) + ret = _pnfs_return_layout(inode); + spin_lock(&inode->i_lock); + lo->plh_block_lgets--; + spin_unlock(&inode->i_lock); + pnfs_put_layout_hdr(lo); + return ret; +} + bool pnfs_roc(struct inode *ino) { struct pnfs_layout_hdr *lo; diff --git a/fs/nfs/pnfs.h b/fs/nfs/pnfs.h index 94ba80417748..f5f8a470a647 100644 --- a/fs/nfs/pnfs.h +++ b/fs/nfs/pnfs.h @@ -219,6 +219,7 @@ void pnfs_set_layoutcommit(struct nfs_write_data *wdata); void pnfs_cleanup_layoutcommit(struct nfs4_layoutcommit_data *data); int pnfs_layoutcommit_inode(struct inode *inode, bool sync); int _pnfs_return_layout(struct inode *); +int pnfs_commit_and_return_layout(struct inode *); void pnfs_ld_write_done(struct nfs_write_data *); void pnfs_ld_read_done(struct nfs_read_data *); struct pnfs_layout_segment *pnfs_update_layout(struct inode *ino, @@ -407,6 +408,11 @@ static inline int pnfs_return_layout(struct inode *ino) return 0; } +static inline int pnfs_commit_and_return_layout(struct inode *inode) +{ + return 0; +} + static inline bool pnfs_ld_layoutret_on_setattr(struct inode *inode) { From 1166fde6a923c30f4351515b6a9a1efc513e7d00 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 25 Mar 2013 11:23:40 -0400 Subject: [PATCH 06/47] SUNRPC: Add barriers to ensure read ordering in rpc_wake_up_task_queue_locked We need to be careful when testing task->tk_waitqueue in rpc_wake_up_task_queue_locked, because it can be changed while we are holding the queue->lock. By adding appropriate memory barriers, we can ensure that it is safe to test task->tk_waitqueue for equality if the RPC_TASK_QUEUED bit is set. Signed-off-by: Trond Myklebust Cc: stable@vger.kernel.org --- net/sunrpc/sched.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) diff --git a/net/sunrpc/sched.c b/net/sunrpc/sched.c index fb20f25ddec9..f8529fc8e542 100644 --- a/net/sunrpc/sched.c +++ b/net/sunrpc/sched.c @@ -180,6 +180,8 @@ static void __rpc_add_wait_queue(struct rpc_wait_queue *queue, list_add_tail(&task->u.tk_wait.list, &queue->tasks[0]); task->tk_waitqueue = queue; queue->qlen++; + /* barrier matches the read in rpc_wake_up_task_queue_locked() */ + smp_wmb(); rpc_set_queued(task); dprintk("RPC: %5u added to queue %p \"%s\"\n", @@ -430,8 +432,11 @@ static void __rpc_do_wake_up_task(struct rpc_wait_queue *queue, struct rpc_task */ static void rpc_wake_up_task_queue_locked(struct rpc_wait_queue *queue, struct rpc_task *task) { - if (RPC_IS_QUEUED(task) && task->tk_waitqueue == queue) - __rpc_do_wake_up_task(queue, task); + if (RPC_IS_QUEUED(task)) { + smp_rmb(); + if (task->tk_waitqueue == queue) + __rpc_do_wake_up_task(queue, task); + } } /* From 3ed5e2a2c394df4e03a680842c2d07a8680f133b Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 4 Mar 2013 17:29:33 -0500 Subject: [PATCH 07/47] SUNRPC: Report network/connection errors correctly for SOFTCONN rpc tasks In the case of a SOFTCONN rpc task, we really want to ensure that it reports errors like ENETUNREACH back to the caller. Currently, only some of these errors are being reported back (connect errors are not), and they are being converted by the RPC layer into EIO. Reported-by: Jan Engelhardt Signed-off-by: Trond Myklebust --- net/sunrpc/clnt.c | 24 ++++++++++++++---------- net/sunrpc/xprtsock.c | 8 ++++---- 2 files changed, 18 insertions(+), 14 deletions(-) diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c index dcc446e7fbf6..b95a0a2d5eea 100644 --- a/net/sunrpc/clnt.c +++ b/net/sunrpc/clnt.c @@ -1644,22 +1644,26 @@ call_connect_status(struct rpc_task *task) dprint_status(task); - task->tk_status = 0; - if (status >= 0 || status == -EAGAIN) { - clnt->cl_stats->netreconn++; - task->tk_action = call_transmit; - return; - } - trace_rpc_connect_status(task, status); switch (status) { /* if soft mounted, test if we've timed out */ case -ETIMEDOUT: task->tk_action = call_timeout; - break; - default: - rpc_exit(task, -EIO); + return; + case -ECONNREFUSED: + case -ECONNRESET: + case -ENETUNREACH: + if (RPC_IS_SOFTCONN(task)) + break; + /* retry with existing socket, after a delay */ + case 0: + case -EAGAIN: + task->tk_status = 0; + clnt->cl_stats->netreconn++; + task->tk_action = call_transmit; + return; } + rpc_exit(task, status); } /* diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c index c1d8476b7692..3081620cb02c 100644 --- a/net/sunrpc/xprtsock.c +++ b/net/sunrpc/xprtsock.c @@ -2202,10 +2202,6 @@ static void xs_tcp_setup_socket(struct work_struct *work) */ xs_tcp_force_close(xprt); break; - case -ECONNREFUSED: - case -ECONNRESET: - case -ENETUNREACH: - /* retry with existing socket, after a delay */ case 0: case -EINPROGRESS: case -EALREADY: @@ -2216,6 +2212,10 @@ static void xs_tcp_setup_socket(struct work_struct *work) /* Happens, for instance, if the user specified a link * local IPv6 address without a scope-id. */ + case -ECONNREFUSED: + case -ECONNRESET: + case -ENETUNREACH: + /* retry with existing socket, after a delay */ goto out; } out_eagain: From 5d422301f97b821301efcdb6fc9d1a83a5c102d6 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Thu, 14 Mar 2013 16:57:48 -0400 Subject: [PATCH 08/47] NFSv4: Fail I/O if the state recovery fails irrevocably If state recovery fails with an ESTALE or a ENOENT, then we shouldn't keep retrying. Instead, mark the stateid as being invalid and fail the I/O with an EIO error. For other operations such as POSIX and BSD file locking, truncate etc, fail with an EBADF to indicate that this file descriptor is no longer valid. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4_fs.h | 8 +++++++- fs/nfs/nfs4filelayout.c | 12 +++++++++--- fs/nfs/nfs4proc.c | 37 +++++++++++++++++++++++++++++-------- fs/nfs/nfs4state.c | 19 ++++++++++++++----- fs/nfs/pnfs.c | 2 ++ 5 files changed, 61 insertions(+), 17 deletions(-) diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h index 944c9a5c1039..9ce90135bf22 100644 --- a/fs/nfs/nfs4_fs.h +++ b/fs/nfs/nfs4_fs.h @@ -149,6 +149,7 @@ enum { NFS_STATE_RECLAIM_REBOOT, /* OPEN stateid server rebooted */ NFS_STATE_RECLAIM_NOGRACE, /* OPEN stateid needs to recover state */ NFS_STATE_POSIX_LOCKS, /* Posix locks are supported */ + NFS_STATE_RECOVERY_FAILED, /* OPEN stateid state recovery failed */ }; struct nfs4_state { @@ -347,7 +348,7 @@ extern int nfs4_wait_clnt_recover(struct nfs_client *clp); extern int nfs4_client_recover_expired_lease(struct nfs_client *clp); extern void nfs4_schedule_state_manager(struct nfs_client *); extern void nfs4_schedule_path_down_recovery(struct nfs_client *clp); -extern void nfs4_schedule_stateid_recovery(const struct nfs_server *, struct nfs4_state *); +extern int nfs4_schedule_stateid_recovery(const struct nfs_server *, struct nfs4_state *); extern void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags); extern void nfs41_handle_server_scope(struct nfs_client *, struct nfs41_server_scope **); @@ -412,6 +413,11 @@ static inline bool nfs4_stateid_match(const nfs4_stateid *dst, const nfs4_statei return memcmp(dst, src, sizeof(*dst)) == 0; } +static inline bool nfs4_valid_open_stateid(const struct nfs4_state *state) +{ + return test_bit(NFS_STATE_RECOVERY_FAILED, &state->flags) == 0; +} + #else #define nfs4_close_state(a, b) do { } while (0) diff --git a/fs/nfs/nfs4filelayout.c b/fs/nfs/nfs4filelayout.c index 4fb234d3aefb..1ee5737211d7 100644 --- a/fs/nfs/nfs4filelayout.c +++ b/fs/nfs/nfs4filelayout.c @@ -158,11 +158,14 @@ static int filelayout_async_handle_error(struct rpc_task *task, case -NFS4ERR_OPENMODE: if (state == NULL) break; - nfs4_schedule_stateid_recovery(mds_server, state); + if (nfs4_schedule_stateid_recovery(mds_server, state) < 0) + goto out_bad_stateid; goto wait_on_recovery; case -NFS4ERR_EXPIRED: - if (state != NULL) - nfs4_schedule_stateid_recovery(mds_server, state); + if (state != NULL) { + if (nfs4_schedule_stateid_recovery(mds_server, state) < 0) + goto out_bad_stateid; + } nfs4_schedule_lease_recovery(mds_client); goto wait_on_recovery; /* DS session errors */ @@ -226,6 +229,9 @@ static int filelayout_async_handle_error(struct rpc_task *task, out: task->tk_status = 0; return -EAGAIN; +out_bad_stateid: + task->tk_status = -EIO; + return 0; wait_on_recovery: rpc_sleep_on(&mds_client->cl_rpcwaitq, task, NULL); if (test_bit(NFS4CLNT_MANAGER_RUNNING, &mds_client->cl_state) == 0) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 26431cf62ddb..c3bbb6c53d61 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -295,7 +295,9 @@ static int nfs4_handle_exception(struct nfs_server *server, int errorcode, struc } if (state == NULL) break; - nfs4_schedule_stateid_recovery(server, state); + ret = nfs4_schedule_stateid_recovery(server, state); + if (ret < 0) + break; goto wait_on_recovery; case -NFS4ERR_DELEG_REVOKED: case -NFS4ERR_ADMIN_REVOKED: @@ -303,11 +305,16 @@ static int nfs4_handle_exception(struct nfs_server *server, int errorcode, struc if (state == NULL) break; nfs_remove_bad_delegation(state->inode); - nfs4_schedule_stateid_recovery(server, state); + ret = nfs4_schedule_stateid_recovery(server, state); + if (ret < 0) + break; goto wait_on_recovery; case -NFS4ERR_EXPIRED: - if (state != NULL) - nfs4_schedule_stateid_recovery(server, state); + if (state != NULL) { + ret = nfs4_schedule_stateid_recovery(server, state); + if (ret < 0) + break; + } case -NFS4ERR_STALE_STATEID: case -NFS4ERR_STALE_CLIENTID: nfs4_schedule_lease_recovery(clp); @@ -2053,7 +2060,7 @@ static int _nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred, nfs_fattr_init(fattr); - if (state != NULL) { + if (state != NULL && nfs4_valid_open_stateid(state)) { struct nfs_lockowner lockowner = { .l_owner = current->files, .l_pid = current->tgid, @@ -2201,6 +2208,8 @@ static void nfs4_close_prepare(struct rpc_task *task, void *data) calldata->arg.fmode &= ~FMODE_WRITE; } } + if (!nfs4_valid_open_stateid(state)) + call_close = 0; spin_unlock(&state->owner->so_lock); if (!call_close) { @@ -3980,11 +3989,14 @@ nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server, case -NFS4ERR_OPENMODE: if (state == NULL) break; - nfs4_schedule_stateid_recovery(server, state); + if (nfs4_schedule_stateid_recovery(server, state) < 0) + goto stateid_invalid; goto wait_on_recovery; case -NFS4ERR_EXPIRED: - if (state != NULL) - nfs4_schedule_stateid_recovery(server, state); + if (state != NULL) { + if (nfs4_schedule_stateid_recovery(server, state) < 0) + goto stateid_invalid; + } case -NFS4ERR_STALE_STATEID: case -NFS4ERR_STALE_CLIENTID: nfs4_schedule_lease_recovery(clp); @@ -4016,6 +4028,9 @@ nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server, } task->tk_status = nfs4_map_errors(task->tk_status); return 0; +stateid_invalid: + task->tk_status = -EIO; + return 0; wait_on_recovery: rpc_sleep_on(&clp->cl_rpcwaitq, task, NULL); if (test_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) == 0) @@ -4632,12 +4647,18 @@ static void nfs4_lock_prepare(struct rpc_task *task, void *calldata) data->res.open_seqid = data->arg.open_seqid; } else data->arg.new_lock_owner = 0; + if (!nfs4_valid_open_stateid(state)) { + data->rpc_status = -EBADF; + task->tk_action = NULL; + goto out_release_open_seqid; + } data->timestamp = jiffies; if (nfs4_setup_sequence(data->server, &data->arg.seq_args, &data->res.seq_res, task) == 0) return; +out_release_open_seqid: nfs_release_seqid(data->arg.open_seqid); out_release_lock_seqid: nfs_release_seqid(data->arg.lock_seqid); diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index 6ace365c6334..fec1c5bb4863 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -699,6 +699,8 @@ __nfs4_find_state_byowner(struct inode *inode, struct nfs4_state_owner *owner) list_for_each_entry(state, &nfsi->open_states, inode_states) { if (state->owner != owner) continue; + if (!nfs4_valid_open_stateid(state)) + continue; if (atomic_inc_not_zero(&state->count)) return state; } @@ -1286,14 +1288,17 @@ static int nfs4_state_mark_reclaim_nograce(struct nfs_client *clp, struct nfs4_s return 1; } -void nfs4_schedule_stateid_recovery(const struct nfs_server *server, struct nfs4_state *state) +int nfs4_schedule_stateid_recovery(const struct nfs_server *server, struct nfs4_state *state) { struct nfs_client *clp = server->nfs_client; + if (!nfs4_valid_open_stateid(state)) + return -EBADF; nfs4_state_mark_reclaim_nograce(clp, state); dprintk("%s: scheduling stateid recovery for server %s\n", __func__, clp->cl_hostname); nfs4_schedule_state_manager(clp); + return 0; } EXPORT_SYMBOL_GPL(nfs4_schedule_stateid_recovery); @@ -1323,6 +1328,11 @@ void nfs_inode_find_state_and_recover(struct inode *inode, nfs4_schedule_state_manager(clp); } +static void nfs4_state_mark_recovery_failed(struct nfs4_state *state, int error) +{ + set_bit(NFS_STATE_RECOVERY_FAILED, &state->flags); +} + static int nfs4_reclaim_locks(struct nfs4_state *state, const struct nfs4_state_recovery_ops *ops) { @@ -1398,6 +1408,8 @@ static int nfs4_reclaim_open_state(struct nfs4_state_owner *sp, const struct nfs list_for_each_entry(state, &sp->so_states, open_states) { if (!test_and_clear_bit(ops->state_flag_bit, &state->flags)) continue; + if (!nfs4_valid_open_stateid(state)) + continue; if (state->state == 0) continue; atomic_inc(&state->count); @@ -1430,10 +1442,7 @@ static int nfs4_reclaim_open_state(struct nfs4_state_owner *sp, const struct nfs * Open state on this file cannot be recovered * All we can do is revert to using the zero stateid. */ - memset(&state->stateid, 0, - sizeof(state->stateid)); - /* Mark the file as being 'closed' */ - state->state = 0; + nfs4_state_mark_recovery_failed(state, status); break; case -NFS4ERR_ADMIN_REVOKED: case -NFS4ERR_STALE_STATEID: diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index 4bdffe0ba025..c5bd758e5637 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -718,6 +718,8 @@ pnfs_choose_layoutget_stateid(nfs4_stateid *dst, struct pnfs_layout_hdr *lo, spin_lock(&lo->plh_inode->i_lock); if (pnfs_layoutgets_blocked(lo, 1)) { status = -EAGAIN; + } else if (!nfs4_valid_open_stateid(open_state)) { + status = -EBADF; } else if (list_empty(&lo->plh_segs)) { int seq; From c58c844187df61ef7cc103d0abb5dd6198bcfcd6 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 18 Mar 2013 19:45:14 -0400 Subject: [PATCH 09/47] NFS: Don't accept more reads/writes if the open context recovery failed If the state recovery failed, we want to ensure that the application doesn't try to use the same file descriptor for more reads or writes. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4filelayout.c | 8 ++++++++ fs/nfs/nfs4state.c | 16 ++++++++++++++++ fs/nfs/pagelist.c | 2 ++ fs/nfs/read.c | 2 ++ fs/nfs/write.c | 2 ++ include/linux/nfs_fs.h | 1 + 6 files changed, 31 insertions(+) diff --git a/fs/nfs/nfs4filelayout.c b/fs/nfs/nfs4filelayout.c index 1ee5737211d7..4ba32e23eddd 100644 --- a/fs/nfs/nfs4filelayout.c +++ b/fs/nfs/nfs4filelayout.c @@ -305,6 +305,10 @@ static void filelayout_read_prepare(struct rpc_task *task, void *data) { struct nfs_read_data *rdata = data; + if (unlikely(test_bit(NFS_CONTEXT_BAD, &rdata->args.context->flags))) { + rpc_exit(task, -EIO); + return; + } if (filelayout_reset_to_mds(rdata->header->lseg)) { dprintk("%s task %u reset io to MDS\n", __func__, task->tk_pid); filelayout_reset_read(rdata); @@ -407,6 +411,10 @@ static void filelayout_write_prepare(struct rpc_task *task, void *data) { struct nfs_write_data *wdata = data; + if (unlikely(test_bit(NFS_CONTEXT_BAD, &wdata->args.context->flags))) { + rpc_exit(task, -EIO); + return; + } if (filelayout_reset_to_mds(wdata->header->lseg)) { dprintk("%s task %u reset io to MDS\n", __func__, task->tk_pid); filelayout_reset_write(wdata); diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index fec1c5bb4863..8db102c7add6 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -1328,9 +1328,25 @@ void nfs_inode_find_state_and_recover(struct inode *inode, nfs4_schedule_state_manager(clp); } +static void nfs4_state_mark_open_context_bad(struct nfs4_state *state) +{ + struct inode *inode = state->inode; + struct nfs_inode *nfsi = NFS_I(inode); + struct nfs_open_context *ctx; + + spin_lock(&inode->i_lock); + list_for_each_entry(ctx, &nfsi->open_files, list) { + if (ctx->state != state) + continue; + set_bit(NFS_CONTEXT_BAD, &ctx->flags); + } + spin_unlock(&inode->i_lock); +} + static void nfs4_state_mark_recovery_failed(struct nfs4_state *state, int error) { set_bit(NFS_STATE_RECOVERY_FAILED, &state->flags); + nfs4_state_mark_open_context_bad(state); } diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c index e56e846e9d2d..7f0933086b36 100644 --- a/fs/nfs/pagelist.c +++ b/fs/nfs/pagelist.c @@ -104,6 +104,8 @@ nfs_create_request(struct nfs_open_context *ctx, struct inode *inode, struct nfs_page *req; struct nfs_lock_context *l_ctx; + if (test_bit(NFS_CONTEXT_BAD, &ctx->flags)) + return ERR_PTR(-EBADF); /* try to allocate the request struct */ req = nfs_page_alloc(); if (req == NULL) diff --git a/fs/nfs/read.c b/fs/nfs/read.c index a5e5d9899d56..70a26c651f09 100644 --- a/fs/nfs/read.c +++ b/fs/nfs/read.c @@ -514,6 +514,8 @@ 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); } static const struct rpc_call_ops nfs_read_common_ops = { diff --git a/fs/nfs/write.c b/fs/nfs/write.c index c483cc50b82e..a2c7c28049d5 100644 --- a/fs/nfs/write.c +++ b/fs/nfs/write.c @@ -1251,6 +1251,8 @@ 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); } void nfs_commit_prepare(struct rpc_task *task, void *calldata) diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index 1cc25682b20b..f6b1956f3c86 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -77,6 +77,7 @@ struct nfs_open_context { unsigned long flags; #define NFS_CONTEXT_ERROR_WRITE (0) #define NFS_CONTEXT_RESEND_WRITES (1) +#define NFS_CONTEXT_BAD (2) int error; struct list_head list; From 8c86899f62738b8a22ca3a5f060e269b92e5545a Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 15 Mar 2013 18:11:31 -0400 Subject: [PATCH 10/47] NFS: __nfs_find_lock_context needs to check ctx->lock_context for a match too Currently, we're forcing an unnecessary duplication of the initial nfs_lock_context in calls to nfs_get_lock_context, since __nfs_find_lock_context ignores the ctx->lock_context. Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 1f941674b089..55b840f05ab2 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -565,16 +565,17 @@ static void nfs_init_lock_context(struct nfs_lock_context *l_ctx) static struct nfs_lock_context *__nfs_find_lock_context(struct nfs_open_context *ctx) { - struct nfs_lock_context *pos; + struct nfs_lock_context *head = &ctx->lock_context; + struct nfs_lock_context *pos = head; - list_for_each_entry(pos, &ctx->lock_context.list, list) { + do { if (pos->lockowner.l_owner != current->files) continue; if (pos->lockowner.l_pid != current->tgid) continue; atomic_inc(&pos->count); return pos; - } + } while ((pos = list_entry(pos->list.next, typeof(*pos), list)) != head); return NULL; } From 9b20614988199fb03580b335a28250922e902098 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 17 Mar 2013 15:52:00 -0400 Subject: [PATCH 11/47] NFSv4: The stateid must remain the same for replayed RPC calls If we replay a READ or WRITE call, we should not be changing the stateid. Currently, we may end up doing so, because the stateid is only selected at xdr encode time. This patch ensures that we select the stateid after we get an NFSv4.1 session slot, and that we keep that same stateid across retries. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4_fs.h | 4 ++++ fs/nfs/nfs4filelayout.c | 14 ++++++++++---- fs/nfs/nfs4proc.c | 27 +++++++++++++++++++++++---- fs/nfs/nfs4xdr.c | 28 ++-------------------------- include/linux/nfs_xdr.h | 2 ++ 5 files changed, 41 insertions(+), 34 deletions(-) diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h index 9ce90135bf22..8309e98c44f9 100644 --- a/fs/nfs/nfs4_fs.h +++ b/fs/nfs/nfs4_fs.h @@ -234,6 +234,10 @@ extern struct rpc_clnt *nfs4_proc_lookup_mountpoint(struct inode *, struct qstr extern int nfs4_proc_secinfo(struct inode *, const struct qstr *, struct nfs4_secinfo_flavors *); extern int nfs4_release_lockowner(struct nfs4_lock_state *); extern const struct xattr_handler *nfs4_xattr_handlers[]; +extern void nfs4_set_rw_stateid(nfs4_stateid *stateid, + const struct nfs_open_context *ctx, + const struct nfs_lock_context *l_ctx, + fmode_t fmode); #if defined(CONFIG_NFS_V4_1) static inline struct nfs4_session *nfs4_get_session(const struct nfs_server *server) diff --git a/fs/nfs/nfs4filelayout.c b/fs/nfs/nfs4filelayout.c index 4ba32e23eddd..22d10623f5ee 100644 --- a/fs/nfs/nfs4filelayout.c +++ b/fs/nfs/nfs4filelayout.c @@ -317,10 +317,13 @@ static void filelayout_read_prepare(struct rpc_task *task, void *data) } rdata->read_done_cb = filelayout_read_done_cb; - nfs41_setup_sequence(rdata->ds_clp->cl_session, + if (nfs41_setup_sequence(rdata->ds_clp->cl_session, &rdata->args.seq_args, &rdata->res.seq_res, - task); + task)) + return; + nfs4_set_rw_stateid(&rdata->args.stateid, rdata->args.context, + rdata->args.lock_context, FMODE_READ); } static void filelayout_read_call_done(struct rpc_task *task, void *data) @@ -421,10 +424,13 @@ static void filelayout_write_prepare(struct rpc_task *task, void *data) rpc_exit(task, 0); return; } - nfs41_setup_sequence(wdata->ds_clp->cl_session, + if (nfs41_setup_sequence(wdata->ds_clp->cl_session, &wdata->args.seq_args, &wdata->res.seq_res, - task); + task)) + return; + nfs4_set_rw_stateid(&wdata->args.stateid, wdata->args.context, + wdata->args.lock_context, FMODE_WRITE); } static void filelayout_write_call_done(struct rpc_task *task, void *data) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index c3bbb6c53d61..26176ce3d96a 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -3454,6 +3454,19 @@ static int nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle, return err; } +void nfs4_set_rw_stateid(nfs4_stateid *stateid, + const struct nfs_open_context *ctx, + const struct nfs_lock_context *l_ctx, + fmode_t fmode) +{ + const struct nfs_lockowner *lockowner = NULL; + + if (l_ctx != NULL) + lockowner = &l_ctx->lockowner; + nfs4_select_rw_stateid(stateid, ctx->state, fmode, lockowner); +} +EXPORT_SYMBOL_GPL(nfs4_set_rw_stateid); + void __nfs4_read_done_cb(struct nfs_read_data *data) { nfs_invalidate_atime(data->header->inode); @@ -3496,10 +3509,13 @@ static void nfs4_proc_read_setup(struct nfs_read_data *data, struct rpc_message static void nfs4_proc_read_rpc_prepare(struct rpc_task *task, struct nfs_read_data *data) { - nfs4_setup_sequence(NFS_SERVER(data->header->inode), + if (nfs4_setup_sequence(NFS_SERVER(data->header->inode), &data->args.seq_args, &data->res.seq_res, - task); + task)) + return; + nfs4_set_rw_stateid(&data->args.stateid, data->args.context, + data->args.lock_context, FMODE_READ); } static int nfs4_write_done_cb(struct rpc_task *task, struct nfs_write_data *data) @@ -3560,10 +3576,13 @@ static void nfs4_proc_write_setup(struct nfs_write_data *data, struct rpc_messag static void nfs4_proc_write_rpc_prepare(struct rpc_task *task, struct nfs_write_data *data) { - nfs4_setup_sequence(NFS_SERVER(data->header->inode), + if (nfs4_setup_sequence(NFS_SERVER(data->header->inode), &data->args.seq_args, &data->res.seq_res, - task); + task)) + return; + nfs4_set_rw_stateid(&data->args.stateid, data->args.context, + data->args.lock_context, FMODE_WRITE); } static void nfs4_proc_commit_rpc_prepare(struct rpc_task *task, struct nfs_commit_data *data) diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c index e3edda554ac7..9d328777b4c1 100644 --- a/fs/nfs/nfs4xdr.c +++ b/fs/nfs/nfs4xdr.c @@ -1506,35 +1506,12 @@ static void encode_putrootfh(struct xdr_stream *xdr, struct compound_hdr *hdr) encode_op_hdr(xdr, OP_PUTROOTFH, decode_putrootfh_maxsz, hdr); } -static void encode_open_stateid(struct xdr_stream *xdr, - const struct nfs_open_context *ctx, - const struct nfs_lock_context *l_ctx, - fmode_t fmode, - int zero_seqid) -{ - nfs4_stateid stateid; - - if (ctx->state != NULL) { - const struct nfs_lockowner *lockowner = NULL; - - if (l_ctx != NULL) - lockowner = &l_ctx->lockowner; - nfs4_select_rw_stateid(&stateid, ctx->state, - fmode, lockowner); - if (zero_seqid) - stateid.seqid = 0; - encode_nfs4_stateid(xdr, &stateid); - } else - encode_nfs4_stateid(xdr, &zero_stateid); -} - static void encode_read(struct xdr_stream *xdr, const struct nfs_readargs *args, struct compound_hdr *hdr) { __be32 *p; encode_op_hdr(xdr, OP_READ, decode_read_maxsz, hdr); - encode_open_stateid(xdr, args->context, args->lock_context, - FMODE_READ, hdr->minorversion); + encode_nfs4_stateid(xdr, &args->stateid); p = reserve_space(xdr, 12); p = xdr_encode_hyper(p, args->offset); @@ -1670,8 +1647,7 @@ static void encode_write(struct xdr_stream *xdr, const struct nfs_writeargs *arg __be32 *p; encode_op_hdr(xdr, OP_WRITE, decode_write_maxsz, hdr); - encode_open_stateid(xdr, args->context, args->lock_context, - FMODE_WRITE, hdr->minorversion); + encode_nfs4_stateid(xdr, &args->stateid); p = reserve_space(xdr, 16); p = xdr_encode_hyper(p, args->offset); diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 4b993d358dad..90a4aa190b43 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -486,6 +486,7 @@ struct nfs_readargs { struct nfs_fh * fh; struct nfs_open_context *context; struct nfs_lock_context *lock_context; + nfs4_stateid stateid; __u64 offset; __u32 count; unsigned int pgbase; @@ -507,6 +508,7 @@ struct nfs_writeargs { struct nfs_fh * fh; struct nfs_open_context *context; struct nfs_lock_context *lock_context; + nfs4_stateid stateid; __u64 offset; __u32 count; enum nfs3_stable_how stable; From 5521abfdcf4d67c3441d4414f29e1acd7cc43380 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sat, 16 Mar 2013 20:54:34 -0400 Subject: [PATCH 12/47] NFSv4: Resend the READ/WRITE RPC call if a stateid change causes an error Adds logic to ensure that if the server returns a BAD_STATEID, or other state related error, then we check if the stateid has already changed. If it has, then rather than start state recovery, we should just resend the failed RPC call with the new stateid. Allow nfs4_select_rw_stateid to notify that the stateid is unstable by having it return -EWOULDBLOCK if an RPC is underway that might change the stateid. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4_fs.h | 4 +-- fs/nfs/nfs4proc.c | 64 +++++++++++++++++++++++++++++++++++++++++++--- fs/nfs/nfs4state.c | 32 ++++++++++++++++------- 3 files changed, 86 insertions(+), 14 deletions(-) diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h index 8309e98c44f9..627a74f0e248 100644 --- a/fs/nfs/nfs4_fs.h +++ b/fs/nfs/nfs4_fs.h @@ -234,7 +234,7 @@ extern struct rpc_clnt *nfs4_proc_lookup_mountpoint(struct inode *, struct qstr extern int nfs4_proc_secinfo(struct inode *, const struct qstr *, struct nfs4_secinfo_flavors *); extern int nfs4_release_lockowner(struct nfs4_lock_state *); extern const struct xattr_handler *nfs4_xattr_handlers[]; -extern void nfs4_set_rw_stateid(nfs4_stateid *stateid, +extern int nfs4_set_rw_stateid(nfs4_stateid *stateid, const struct nfs_open_context *ctx, const struct nfs_lock_context *l_ctx, fmode_t fmode); @@ -358,7 +358,7 @@ extern void nfs41_handle_server_scope(struct nfs_client *, struct nfs41_server_scope **); extern void nfs4_put_lock_state(struct nfs4_lock_state *lsp); extern int nfs4_set_lock_state(struct nfs4_state *state, struct file_lock *fl); -extern void nfs4_select_rw_stateid(nfs4_stateid *, struct nfs4_state *, +extern int nfs4_select_rw_stateid(nfs4_stateid *, struct nfs4_state *, fmode_t, const struct nfs_lockowner *); extern struct nfs_seqid *nfs_alloc_seqid(struct nfs_seqid_counter *counter, gfp_t gfp_mask); diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 26176ce3d96a..6ad06121d88c 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -3454,7 +3454,7 @@ static int nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle, return err; } -void nfs4_set_rw_stateid(nfs4_stateid *stateid, +int nfs4_set_rw_stateid(nfs4_stateid *stateid, const struct nfs_open_context *ctx, const struct nfs_lock_context *l_ctx, fmode_t fmode) @@ -3463,10 +3463,37 @@ void nfs4_set_rw_stateid(nfs4_stateid *stateid, if (l_ctx != NULL) lockowner = &l_ctx->lockowner; - nfs4_select_rw_stateid(stateid, ctx->state, fmode, lockowner); + return nfs4_select_rw_stateid(stateid, ctx->state, fmode, lockowner); } EXPORT_SYMBOL_GPL(nfs4_set_rw_stateid); +static bool nfs4_stateid_is_current(nfs4_stateid *stateid, + const struct nfs_open_context *ctx, + const struct nfs_lock_context *l_ctx, + fmode_t fmode) +{ + nfs4_stateid current_stateid; + + if (nfs4_set_rw_stateid(¤t_stateid, ctx, l_ctx, fmode)) + return false; + return nfs4_stateid_match(stateid, ¤t_stateid); +} + +static bool nfs4_error_stateid_expired(int err) +{ + switch (err) { + case -NFS4ERR_DELEG_REVOKED: + case -NFS4ERR_ADMIN_REVOKED: + case -NFS4ERR_BAD_STATEID: + case -NFS4ERR_STALE_STATEID: + case -NFS4ERR_OLD_STATEID: + case -NFS4ERR_OPENMODE: + case -NFS4ERR_EXPIRED: + return true; + } + return false; +} + void __nfs4_read_done_cb(struct nfs_read_data *data) { nfs_invalidate_atime(data->header->inode); @@ -3487,6 +3514,20 @@ static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_read_data *data) return 0; } +static bool nfs4_read_stateid_changed(struct rpc_task *task, + struct nfs_readargs *args) +{ + + if (!nfs4_error_stateid_expired(task->tk_status) || + nfs4_stateid_is_current(&args->stateid, + args->context, + args->lock_context, + FMODE_READ)) + return false; + rpc_restart_call_prepare(task); + return true; +} + static int nfs4_read_done(struct rpc_task *task, struct nfs_read_data *data) { @@ -3494,7 +3535,8 @@ static int nfs4_read_done(struct rpc_task *task, struct nfs_read_data *data) if (!nfs4_sequence_done(task, &data->res.seq_res)) return -EAGAIN; - + if (nfs4_read_stateid_changed(task, &data->args)) + return -EAGAIN; return data->read_done_cb ? data->read_done_cb(task, data) : nfs4_read_done_cb(task, data); } @@ -3533,10 +3575,26 @@ static int nfs4_write_done_cb(struct rpc_task *task, struct nfs_write_data *data return 0; } +static bool nfs4_write_stateid_changed(struct rpc_task *task, + struct nfs_writeargs *args) +{ + + if (!nfs4_error_stateid_expired(task->tk_status) || + nfs4_stateid_is_current(&args->stateid, + args->context, + args->lock_context, + FMODE_WRITE)) + return false; + rpc_restart_call_prepare(task); + return true; +} + static int nfs4_write_done(struct rpc_task *task, struct nfs_write_data *data) { if (!nfs4_sequence_done(task, &data->res.seq_res)) return -EAGAIN; + if (nfs4_write_stateid_changed(task, &data->args)) + return -EAGAIN; return data->write_done_cb ? data->write_done_cb(task, data) : nfs4_write_done_cb(task, data); } diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index 8db102c7add6..4e95bd72f480 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -989,13 +989,14 @@ int nfs4_set_lock_state(struct nfs4_state *state, struct file_lock *fl) return 0; } -static bool nfs4_copy_lock_stateid(nfs4_stateid *dst, struct nfs4_state *state, +static int nfs4_copy_lock_stateid(nfs4_stateid *dst, + struct nfs4_state *state, const struct nfs_lockowner *lockowner) { struct nfs4_lock_state *lsp; fl_owner_t fl_owner; pid_t fl_pid; - bool ret = false; + int ret = -ENOENT; if (lockowner == NULL) @@ -1010,7 +1011,10 @@ static bool nfs4_copy_lock_stateid(nfs4_stateid *dst, struct nfs4_state *state, 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) { nfs4_stateid_copy(dst, &lsp->ls_stateid); - ret = true; + ret = 0; + smp_rmb(); + if (!list_empty(&lsp->ls_seqid.list)) + ret = -EWOULDBLOCK; } spin_unlock(&state->state_lock); nfs4_put_lock_state(lsp); @@ -1018,28 +1022,38 @@ static bool nfs4_copy_lock_stateid(nfs4_stateid *dst, struct nfs4_state *state, return ret; } -static void nfs4_copy_open_stateid(nfs4_stateid *dst, struct nfs4_state *state) +static int nfs4_copy_open_stateid(nfs4_stateid *dst, struct nfs4_state *state) { + int ret; int seq; do { seq = read_seqbegin(&state->seqlock); nfs4_stateid_copy(dst, &state->stateid); + ret = 0; + smp_rmb(); + if (!list_empty(&state->owner->so_seqid.list)) + ret = -EWOULDBLOCK; } while (read_seqretry(&state->seqlock, seq)); + return ret; } /* * Byte-range lock aware utility to initialize the stateid of read/write * requests. */ -void nfs4_select_rw_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; if (nfs4_copy_delegation_stateid(dst, state->inode, fmode)) - return; - if (nfs4_copy_lock_stateid(dst, state, lockowner)) - return; - nfs4_copy_open_stateid(dst, state); + goto out; + ret = nfs4_copy_lock_stateid(dst, state, lockowner); + if (ret != -ENOENT) + goto out; + ret = nfs4_copy_open_stateid(dst, state); +out: + return ret; } struct nfs_seqid *nfs_alloc_seqid(struct nfs_seqid_counter *counter, gfp_t gfp_mask) From 39c6daae70fe7692fc2d28738c7ebd63f0318741 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 15 Mar 2013 16:11:57 -0400 Subject: [PATCH 13/47] NFSv4: Prepare for minorversion-specific nfs_server capabilities Clean up the setting of the nfs_server->caps, by shoving it all into nfs4_server_common_setup(). Then add an 'initial capabilities' field into struct nfs4_minor_version_ops. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4_fs.h | 1 + fs/nfs/nfs4client.c | 24 +++++++++++++----------- fs/nfs/nfs4proc.c | 8 ++++++++ 3 files changed, 22 insertions(+), 11 deletions(-) diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h index 627a74f0e248..7ef19ce6d832 100644 --- a/fs/nfs/nfs4_fs.h +++ b/fs/nfs/nfs4_fs.h @@ -36,6 +36,7 @@ enum nfs4_client_state { struct nfs4_minor_version_ops { u32 minor_version; + unsigned init_caps; int (*call_sync)(struct rpc_clnt *clnt, struct nfs_server *server, diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c index ac4fc9a8fdbc..17b34b2da2df 100644 --- a/fs/nfs/nfs4client.c +++ b/fs/nfs/nfs4client.c @@ -717,6 +717,19 @@ static int nfs4_server_common_setup(struct nfs_server *server, if (error < 0) goto out; + /* Set the basic capabilities */ + server->caps |= server->nfs_client->cl_mvops->init_caps; + if (server->flags & NFS_MOUNT_NORDIRPLUS) + server->caps &= ~NFS_CAP_READDIRPLUS; + /* + * Don't use NFS uid/gid mapping if we're using AUTH_SYS or lower + * authentication. + */ + if (nfs4_disable_idmapping && + server->client->cl_auth->au_flavor == RPC_AUTH_UNIX) + server->caps |= NFS_CAP_UIDGID_NOMAP; + + /* Probe the root fh to retrieve its FSID and filehandle */ error = nfs4_get_rootfh(server, mntfh); if (error < 0) @@ -760,9 +773,6 @@ static int nfs4_init_server(struct nfs_server *server, /* Initialise the client representation from the mount data */ server->flags = data->flags; - server->caps |= NFS_CAP_ATOMIC_OPEN|NFS_CAP_CHANGE_ATTR|NFS_CAP_POSIX_LOCK; - if (!(data->flags & NFS_MOUNT_NORDIRPLUS)) - server->caps |= NFS_CAP_READDIRPLUS; server->options = data->options; /* Get a client record */ @@ -779,13 +789,6 @@ static int nfs4_init_server(struct nfs_server *server, if (error < 0) goto error; - /* - * Don't use NFS uid/gid mapping if we're using AUTH_SYS or lower - * authentication. - */ - if (nfs4_disable_idmapping && data->auth_flavors[0] == RPC_AUTH_UNIX) - server->caps |= NFS_CAP_UIDGID_NOMAP; - if (data->rsize) server->rsize = nfs_block_size(data->rsize, NULL); if (data->wsize) @@ -863,7 +866,6 @@ struct nfs_server *nfs4_create_referral_server(struct nfs_clone_mount *data, /* Initialise the client representation from the parent server */ nfs_server_copy_userdata(server, parent_server); - server->caps |= NFS_CAP_ATOMIC_OPEN|NFS_CAP_CHANGE_ATTR; /* Get a client representation. * Note: NFSv4 always uses TCP, */ diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 6ad06121d88c..22c80f9ed824 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -6823,6 +6823,10 @@ static const struct nfs4_state_maintenance_ops nfs41_state_renewal_ops = { static const struct nfs4_minor_version_ops nfs_v4_0_minor_ops = { .minor_version = 0, + .init_caps = NFS_CAP_READDIRPLUS + | NFS_CAP_ATOMIC_OPEN + | NFS_CAP_CHANGE_ATTR + | NFS_CAP_POSIX_LOCK, .call_sync = _nfs4_call_sync, .match_stateid = nfs4_match_stateid, .find_root_sec = nfs4_find_root_sec, @@ -6834,6 +6838,10 @@ static const struct nfs4_minor_version_ops nfs_v4_0_minor_ops = { #if defined(CONFIG_NFS_V4_1) static const struct nfs4_minor_version_ops nfs_v4_1_minor_ops = { .minor_version = 1, + .init_caps = NFS_CAP_READDIRPLUS + | NFS_CAP_ATOMIC_OPEN + | NFS_CAP_CHANGE_ATTR + | NFS_CAP_POSIX_LOCK, .call_sync = nfs4_call_sync_sequence, .match_stateid = nfs41_match_stateid, .find_root_sec = nfs41_find_root_sec, From 3b66486c4c7136f8d4bbe1306d581fadc6bce4c7 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 17 Mar 2013 15:31:15 -0400 Subject: [PATCH 14/47] NFSv4.1: Select the "most recent locking state" for read/write/setattr stateids Follow the practice described in section 8.2.2 of RFC5661: When sending a read/write or setattr stateid, set the seqid field to zero in order to signal that the NFS server should apply the most recent locking state. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 3 ++- fs/nfs/nfs4state.c | 2 ++ include/linux/nfs_fs_sb.h | 1 + 3 files changed, 5 insertions(+), 1 deletion(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 22c80f9ed824..625a729ebcca 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -6841,7 +6841,8 @@ static const struct nfs4_minor_version_ops nfs_v4_1_minor_ops = { .init_caps = NFS_CAP_READDIRPLUS | NFS_CAP_ATOMIC_OPEN | NFS_CAP_CHANGE_ATTR - | NFS_CAP_POSIX_LOCK, + | NFS_CAP_POSIX_LOCK + | NFS_CAP_STATEID_NFSV41, .call_sync = nfs4_call_sync_sequence, .match_stateid = nfs41_match_stateid, .find_root_sec = nfs41_find_root_sec, diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index 4e95bd72f480..685b1e953ed8 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -1053,6 +1053,8 @@ int nfs4_select_rw_stateid(nfs4_stateid *dst, struct nfs4_state *state, goto out; ret = nfs4_copy_open_stateid(dst, state); out: + if (nfs_server_capable(state->inode, NFS_CAP_STATEID_NFSV41)) + dst->seqid = 0; return ret; } diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h index 6c6ed153a9b4..74c9e52c9338 100644 --- a/include/linux/nfs_fs_sb.h +++ b/include/linux/nfs_fs_sb.h @@ -197,5 +197,6 @@ struct nfs_server { #define NFS_CAP_MTIME (1U << 13) #define NFS_CAP_POSIX_LOCK (1U << 14) #define NFS_CAP_UIDGID_NOMAP (1U << 15) +#define NFS_CAP_STATEID_NFSV41 (1U << 16) #endif From 4a1c089345a0f00ad0229bf895bf0e94d101f059 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 15 Mar 2013 14:57:33 -0400 Subject: [PATCH 15/47] NFSv4: Clean up nfs4_opendata_alloc in preparation for NFSv4.1 open modes Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 38 +++++++++++++++++++++++++++----------- 1 file changed, 27 insertions(+), 11 deletions(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 625a729ebcca..7bbb06f280fc 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -782,6 +782,7 @@ static void nfs4_init_opendata_res(struct nfs4_opendata *p) static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry, struct nfs4_state_owner *sp, fmode_t fmode, int flags, const struct iattr *attrs, + enum open_claim_type4 claim, gfp_t gfp_mask) { struct dentry *parent = dget_parent(dentry); @@ -800,7 +801,6 @@ static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry, p->dir = parent; p->owner = sp; atomic_inc(&sp->so_count); - p->o_arg.fh = NFS_FH(dir); p->o_arg.open_flags = flags; p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE); /* don't put an ACCESS op in OPEN compound if O_EXCL, because ACCESS @@ -818,7 +818,19 @@ static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry, p->o_arg.server = server; p->o_arg.bitmask = server->attr_bitmask; p->o_arg.open_bitmap = &nfs4_fattr_bitmap[0]; - p->o_arg.claim = NFS4_OPEN_CLAIM_NULL; + p->o_arg.claim = claim; + switch (claim) { + case NFS4_OPEN_CLAIM_NULL: + case NFS4_OPEN_CLAIM_DELEGATE_CUR: + case NFS4_OPEN_CLAIM_DELEGATE_PREV: + p->o_arg.fh = NFS_FH(dir); + break; + case NFS4_OPEN_CLAIM_PREVIOUS: + case NFS4_OPEN_CLAIM_FH: + case NFS4_OPEN_CLAIM_DELEG_CUR_FH: + case NFS4_OPEN_CLAIM_DELEG_PREV_FH: + p->o_arg.fh = NFS_FH(dentry->d_inode); + } if (attrs != NULL && attrs->ia_valid != 0) { __be32 verf[2]; @@ -1200,11 +1212,13 @@ static struct nfs_open_context *nfs4_state_find_open_context(struct nfs4_state * return ERR_PTR(-ENOENT); } -static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context *ctx, struct nfs4_state *state) +static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context *ctx, + struct nfs4_state *state, enum open_claim_type4 claim) { struct nfs4_opendata *opendata; - opendata = nfs4_opendata_alloc(ctx->dentry, state->owner, 0, 0, NULL, GFP_NOFS); + opendata = nfs4_opendata_alloc(ctx->dentry, state->owner, 0, 0, + NULL, claim, GFP_NOFS); if (opendata == NULL) return ERR_PTR(-ENOMEM); opendata->state = state; @@ -1290,11 +1304,10 @@ static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state fmode_t delegation_type = 0; int status; - opendata = nfs4_open_recoverdata_alloc(ctx, state); + opendata = nfs4_open_recoverdata_alloc(ctx, state, + NFS4_OPEN_CLAIM_PREVIOUS); if (IS_ERR(opendata)) return PTR_ERR(opendata); - opendata->o_arg.claim = NFS4_OPEN_CLAIM_PREVIOUS; - opendata->o_arg.fh = NFS_FH(state->inode); rcu_read_lock(); delegation = rcu_dereference(NFS_I(state->inode)->delegation); if (delegation != NULL && test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) != 0) @@ -1338,10 +1351,10 @@ static int _nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs struct nfs4_opendata *opendata; int ret; - opendata = nfs4_open_recoverdata_alloc(ctx, state); + opendata = nfs4_open_recoverdata_alloc(ctx, state, + NFS4_OPEN_CLAIM_DELEGATE_CUR); if (IS_ERR(opendata)) return PTR_ERR(opendata); - opendata->o_arg.claim = NFS4_OPEN_CLAIM_DELEGATE_CUR; nfs4_stateid_copy(&opendata->o_arg.u.delegation, stateid); ret = nfs4_open_recover(opendata, state); nfs4_opendata_put(opendata); @@ -1727,7 +1740,8 @@ static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *s struct nfs4_opendata *opendata; int ret; - opendata = nfs4_open_recoverdata_alloc(ctx, state); + opendata = nfs4_open_recoverdata_alloc(ctx, state, + NFS4_OPEN_CLAIM_NULL); if (IS_ERR(opendata)) return PTR_ERR(opendata); ret = nfs4_open_recover(opendata, state); @@ -1927,7 +1941,9 @@ static int _nfs4_do_open(struct inode *dir, if (dentry->d_inode != NULL) nfs4_return_incompatible_delegation(dentry->d_inode, fmode); status = -ENOMEM; - opendata = nfs4_opendata_alloc(dentry, sp, fmode, flags, sattr, GFP_KERNEL); + opendata = nfs4_opendata_alloc(dentry, sp, fmode, flags, sattr, + NFS4_OPEN_CLAIM_NULL, + GFP_KERNEL); if (opendata == NULL) goto err_put_state_owner; From d9fc6619ca98943d56d2931ce9e4caa4039fd0f1 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 15 Mar 2013 15:39:06 -0400 Subject: [PATCH 16/47] NFSv4.1: Add xdr support for CLAIM_FH and CLAIM_DELEG_CUR_FH opens Signed-off-by: Trond Myklebust --- fs/nfs/nfs4xdr.c | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c index 9d328777b4c1..650fa322b903 100644 --- a/fs/nfs/nfs4xdr.c +++ b/fs/nfs/nfs4xdr.c @@ -1459,6 +1459,23 @@ static inline void encode_claim_delegate_cur(struct xdr_stream *xdr, const struc encode_string(xdr, name->len, name->name); } +static inline void encode_claim_fh(struct xdr_stream *xdr) +{ + __be32 *p; + + p = reserve_space(xdr, 4); + *p = cpu_to_be32(NFS4_OPEN_CLAIM_FH); +} + +static inline void encode_claim_delegate_cur_fh(struct xdr_stream *xdr, const nfs4_stateid *stateid) +{ + __be32 *p; + + p = reserve_space(xdr, 4); + *p = cpu_to_be32(NFS4_OPEN_CLAIM_DELEG_CUR_FH); + encode_nfs4_stateid(xdr, stateid); +} + static void encode_open(struct xdr_stream *xdr, const struct nfs_openargs *arg, struct compound_hdr *hdr) { encode_op_hdr(xdr, OP_OPEN, decode_open_maxsz, hdr); @@ -1474,6 +1491,12 @@ static void encode_open(struct xdr_stream *xdr, const struct nfs_openargs *arg, case NFS4_OPEN_CLAIM_DELEGATE_CUR: encode_claim_delegate_cur(xdr, arg->name, &arg->u.delegation); break; + case NFS4_OPEN_CLAIM_FH: + encode_claim_fh(xdr); + break; + case NFS4_OPEN_CLAIM_DELEG_CUR_FH: + encode_claim_delegate_cur_fh(xdr, &arg->u.delegation); + break; default: BUG(); } From 49f9a0fafd844c32f2abada047c0b9a5ba0d6255 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 15 Mar 2013 16:44:28 -0400 Subject: [PATCH 17/47] NFSv4.1: Enable open-by-filehandle Sometimes, we actually _want_ to do open-by-filehandle, for instance when recovering opens after a network partition, or when called from nfs4_file_open. Enable that functionality using a new capability NFS_CAP_ATOMIC_OPEN_V1, and which is only enabled for NFSv4.1 servers that support it. Signed-off-by: Trond Myklebust --- fs/nfs/dir.c | 2 ++ fs/nfs/nfs4proc.c | 53 +++++++++++++++++++++++++++++++++------ include/linux/nfs_fs_sb.h | 1 + 3 files changed, 49 insertions(+), 7 deletions(-) diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c index f23f455be42b..e093e73178b7 100644 --- a/fs/nfs/dir.c +++ b/fs/nfs/dir.c @@ -1486,6 +1486,8 @@ static int nfs4_lookup_revalidate(struct dentry *dentry, unsigned int flags) goto no_open; if (d_mountpoint(dentry)) goto no_open; + if (NFS_SB(dentry->d_sb)->caps & NFS_CAP_ATOMIC_OPEN_V1) + goto no_open; inode = dentry->d_inode; parent = dget_parent(dentry); diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 7bbb06f280fc..732b76f703d6 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -767,6 +767,35 @@ struct nfs4_opendata { int cancelled; }; +static bool nfs4_clear_cap_atomic_open_v1(struct nfs_server *server, + int err, struct nfs4_exception *exception) +{ + if (err != -EINVAL) + return false; + if (!(server->caps & NFS_CAP_ATOMIC_OPEN_V1)) + return false; + server->caps &= ~NFS_CAP_ATOMIC_OPEN_V1; + exception->retry = 1; + return true; +} + +static enum open_claim_type4 +nfs4_map_atomic_open_claim(struct nfs_server *server, + enum open_claim_type4 claim) +{ + if (server->caps & NFS_CAP_ATOMIC_OPEN_V1) + return claim; + switch (claim) { + default: + return claim; + case NFS4_OPEN_CLAIM_FH: + return NFS4_OPEN_CLAIM_NULL; + case NFS4_OPEN_CLAIM_DELEG_CUR_FH: + return NFS4_OPEN_CLAIM_DELEGATE_CUR; + case NFS4_OPEN_CLAIM_DELEG_PREV_FH: + return NFS4_OPEN_CLAIM_DELEGATE_PREV; + } +} static void nfs4_init_opendata_res(struct nfs4_opendata *p) { @@ -818,8 +847,8 @@ static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry, p->o_arg.server = server; p->o_arg.bitmask = server->attr_bitmask; p->o_arg.open_bitmap = &nfs4_fattr_bitmap[0]; - p->o_arg.claim = claim; - switch (claim) { + p->o_arg.claim = nfs4_map_atomic_open_claim(server, claim); + switch (p->o_arg.claim) { case NFS4_OPEN_CLAIM_NULL: case NFS4_OPEN_CLAIM_DELEGATE_CUR: case NFS4_OPEN_CLAIM_DELEGATE_PREV: @@ -1326,6 +1355,8 @@ static int nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state int err; do { err = _nfs4_do_open_reclaim(ctx, state); + if (nfs4_clear_cap_atomic_open_v1(server, err, &exception)) + continue; if (err != -NFS4ERR_DELAY) break; nfs4_handle_exception(server, err, &exception); @@ -1741,7 +1772,7 @@ static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *s int ret; opendata = nfs4_open_recoverdata_alloc(ctx, state, - NFS4_OPEN_CLAIM_NULL); + NFS4_OPEN_CLAIM_FH); if (IS_ERR(opendata)) return PTR_ERR(opendata); ret = nfs4_open_recover(opendata, state); @@ -1759,6 +1790,8 @@ static int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4_state do { err = _nfs4_open_expired(ctx, state); + if (nfs4_clear_cap_atomic_open_v1(server, err, &exception)) + continue; switch (err) { default: goto out; @@ -1926,6 +1959,7 @@ static int _nfs4_do_open(struct inode *dir, struct nfs4_state *state = NULL; struct nfs_server *server = NFS_SERVER(dir); struct nfs4_opendata *opendata; + enum open_claim_type4 claim = NFS4_OPEN_CLAIM_NULL; int status; /* Protect against reboot recovery conflicts */ @@ -1941,9 +1975,10 @@ static int _nfs4_do_open(struct inode *dir, if (dentry->d_inode != NULL) nfs4_return_incompatible_delegation(dentry->d_inode, fmode); status = -ENOMEM; + if (dentry->d_inode) + claim = NFS4_OPEN_CLAIM_FH; opendata = nfs4_opendata_alloc(dentry, sp, fmode, flags, sattr, - NFS4_OPEN_CLAIM_NULL, - GFP_KERNEL); + claim, GFP_KERNEL); if (opendata == NULL) goto err_put_state_owner; @@ -2001,6 +2036,7 @@ static struct nfs4_state *nfs4_do_open(struct inode *dir, struct rpc_cred *cred, struct nfs4_threshold **ctx_th) { + struct nfs_server *server = NFS_SERVER(dir); struct nfs4_exception exception = { }; struct nfs4_state *res; int status; @@ -2044,7 +2080,9 @@ static struct nfs4_state *nfs4_do_open(struct inode *dir, exception.retry = 1; continue; } - res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(dir), + if (nfs4_clear_cap_atomic_open_v1(server, status, &exception)) + continue; + res = ERR_PTR(nfs4_handle_exception(server, status, &exception)); } while (exception.retry); return res; @@ -6858,7 +6896,8 @@ static const struct nfs4_minor_version_ops nfs_v4_1_minor_ops = { | NFS_CAP_ATOMIC_OPEN | NFS_CAP_CHANGE_ATTR | NFS_CAP_POSIX_LOCK - | NFS_CAP_STATEID_NFSV41, + | NFS_CAP_STATEID_NFSV41 + | NFS_CAP_ATOMIC_OPEN_V1, .call_sync = nfs4_call_sync_sequence, .match_stateid = nfs41_match_stateid, .find_root_sec = nfs41_find_root_sec, diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h index 74c9e52c9338..d8fdfdc7a8fe 100644 --- a/include/linux/nfs_fs_sb.h +++ b/include/linux/nfs_fs_sb.h @@ -198,5 +198,6 @@ struct nfs_server { #define NFS_CAP_POSIX_LOCK (1U << 14) #define NFS_CAP_UIDGID_NOMAP (1U << 15) #define NFS_CAP_STATEID_NFSV41 (1U << 16) +#define NFS_CAP_ATOMIC_OPEN_V1 (1U << 17) #endif From ccb46e2063fe0816c3f28a8d328a9bbac7e88648 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 18 Mar 2013 12:50:59 -0400 Subject: [PATCH 18/47] NFSv4.1: Use CLAIM_DELEG_CUR_FH opens when available Now that we do CLAIM_FH opens, we may run into situations where we get a delegation but don't have perfect knowledge of the file path. When returning the delegation, we might therefore not be able to us CLAIM_DELEGATE_CUR opens to convert the delegation into OPEN stateids and locks. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 732b76f703d6..06e5a72b5fa2 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -1383,7 +1383,7 @@ static int _nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs int ret; opendata = nfs4_open_recoverdata_alloc(ctx, state, - NFS4_OPEN_CLAIM_DELEGATE_CUR); + NFS4_OPEN_CLAIM_DELEG_CUR_FH); if (IS_ERR(opendata)) return PTR_ERR(opendata); nfs4_stateid_copy(&opendata->o_arg.u.delegation, stateid); From 6e3cf2415269894895074f139ef2e95ef14ddd5d Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sat, 23 Mar 2013 15:22:45 -0400 Subject: [PATCH 19/47] NFSv4: Add a mapping for NFS4ERR_FILE_OPEN in nfs4_map_errors With unlink is an asynchronous operation in the sillyrename case, it expects nfs4_async_handle_error() to map the error correctly. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 06e5a72b5fa2..3e7d42fb775c 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -107,6 +107,8 @@ static int nfs4_map_errors(int err) return -EPROTONOSUPPORT; case -NFS4ERR_ACCESS: return -EACCES; + case -NFS4ERR_FILE_OPEN: + return -EBUSY; default: dprintk("%s could not handle NFSv4 error %d\n", __func__, -err); From 91876b13b8b3cbff5cca8476d4b4eebe5f6038cc Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Thu, 28 Mar 2013 14:01:33 -0400 Subject: [PATCH 20/47] NFSv4: Fix another reboot recovery race If the open_context for the file is not yet fully initialised, then open recovery cannot succeed, and since nfs4_state_find_open_context returns an ENOENT, we end up treating the file as being irrecoverable. What we really want to do, is just defer the recovery until later. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 8 +++----- fs/nfs/nfs4state.c | 2 ++ 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 3e7d42fb775c..81343944e096 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -1373,7 +1373,7 @@ static int nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *sta ctx = nfs4_state_find_open_context(state); if (IS_ERR(ctx)) - return PTR_ERR(ctx); + return -EAGAIN; ret = nfs4_do_open_reclaim(ctx, state); put_nfs_open_context(ctx); return ret; @@ -1814,7 +1814,7 @@ static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *sta ctx = nfs4_state_find_open_context(state); if (IS_ERR(ctx)) - return PTR_ERR(ctx); + return -EAGAIN; ret = nfs4_do_open_expired(ctx, state); put_nfs_open_context(ctx); return ret; @@ -1936,10 +1936,8 @@ static int _nfs4_open_and_get_state(struct nfs4_opendata *opendata, if (ret != 0) goto out; - if (read_seqcount_retry(&sp->so_reclaim_seqcount, seq)) { + if (read_seqcount_retry(&sp->so_reclaim_seqcount, seq)) nfs4_schedule_stateid_recovery(server, state); - nfs4_wait_clnt_recover(server->nfs_client); - } *res = state; out: return ret; diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index 685b1e953ed8..b924bdd69494 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -1476,6 +1476,8 @@ static int nfs4_reclaim_open_state(struct nfs4_state_owner *sp, const struct nfs */ nfs4_state_mark_recovery_failed(state, status); break; + case -EAGAIN: + ssleep(1); case -NFS4ERR_ADMIN_REVOKED: case -NFS4ERR_STALE_STATEID: case -NFS4ERR_BAD_STATEID: From 809b426c7fc46b07065eb382c0289710c34b37e4 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 27 Mar 2013 11:54:45 -0400 Subject: [PATCH 21/47] NFSv4: Fix Oopses in the fs_locations code If the server sends us a pathname with more components than the client limit of NFS4_PATHNAME_MAXCOMPONENTS, more server entries than the client limit of NFS4_FS_LOCATION_MAXSERVERS, or sends a total number of fs_locations entries than the client limit of NFS4_FS_LOCATIONS_MAXENTRIES then we will currently Oops because the limit checks are done _after_ we've decoded the data into the arrays. Reported-by: fanchaoting Signed-off-by: Trond Myklebust --- fs/nfs/nfs4xdr.c | 43 ++++++++++++++++++++----------------------- 1 file changed, 20 insertions(+), 23 deletions(-) diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c index 650fa322b903..0b744895b9e1 100644 --- a/fs/nfs/nfs4xdr.c +++ b/fs/nfs/nfs4xdr.c @@ -3496,8 +3496,11 @@ static int decode_pathname(struct xdr_stream *xdr, struct nfs4_pathname *path) if (n == 0) goto root_path; dprintk("pathname4: "); - path->ncomponents = 0; - while (path->ncomponents < n) { + if (n > NFS4_PATHNAME_MAXCOMPONENTS) { + dprintk("cannot parse %d components in path\n", n); + goto out_eio; + } + for (path->ncomponents = 0; path->ncomponents < n; path->ncomponents++) { struct nfs4_string *component = &path->components[path->ncomponents]; status = decode_opaque_inline(xdr, &component->len, &component->data); if (unlikely(status != 0)) @@ -3506,12 +3509,6 @@ static int decode_pathname(struct xdr_stream *xdr, struct nfs4_pathname *path) pr_cont("%s%.*s ", (path->ncomponents != n ? "/ " : ""), component->len, component->data); - if (path->ncomponents < NFS4_PATHNAME_MAXCOMPONENTS) - path->ncomponents++; - else { - dprintk("cannot parse %d components in path\n", n); - goto out_eio; - } } out: return status; @@ -3556,27 +3553,23 @@ static int decode_attr_fs_locations(struct xdr_stream *xdr, uint32_t *bitmap, st n = be32_to_cpup(p); if (n <= 0) goto out_eio; - res->nlocations = 0; - while (res->nlocations < n) { + for (res->nlocations = 0; res->nlocations < n; res->nlocations++) { u32 m; - struct nfs4_fs_location *loc = &res->locations[res->nlocations]; + struct nfs4_fs_location *loc; + if (res->nlocations == NFS4_FS_LOCATIONS_MAXENTRIES) + break; + loc = &res->locations[res->nlocations]; p = xdr_inline_decode(xdr, 4); if (unlikely(!p)) goto out_overflow; m = be32_to_cpup(p); - loc->nservers = 0; dprintk("%s: servers:\n", __func__); - while (loc->nservers < m) { - struct nfs4_string *server = &loc->servers[loc->nservers]; - status = decode_opaque_inline(xdr, &server->len, &server->data); - if (unlikely(status != 0)) - goto out_eio; - dprintk("%s ", server->data); - if (loc->nservers < NFS4_FS_LOCATION_MAXSERVERS) - loc->nservers++; - else { + for (loc->nservers = 0; loc->nservers < m; loc->nservers++) { + struct nfs4_string *server; + + if (loc->nservers == NFS4_FS_LOCATION_MAXSERVERS) { unsigned int i; dprintk("%s: using first %u of %u servers " "returned for location %u\n", @@ -3590,13 +3583,17 @@ static int decode_attr_fs_locations(struct xdr_stream *xdr, uint32_t *bitmap, st if (unlikely(status != 0)) goto out_eio; } + break; } + server = &loc->servers[loc->nservers]; + status = decode_opaque_inline(xdr, &server->len, &server->data); + if (unlikely(status != 0)) + goto out_eio; + dprintk("%s ", server->data); } status = decode_pathname(xdr, &loc->rootpath); if (unlikely(status != 0)) goto out_eio; - if (res->nlocations < NFS4_FS_LOCATIONS_MAXENTRIES) - res->nlocations++; } if (res->nlocations != 0) status = NFS_ATTR_FATTR_V4_LOCATIONS; From 25d280aad82e559cc7bc762410cd35ea5b98e294 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Mon, 25 Mar 2013 07:59:57 -0400 Subject: [PATCH 22/47] nfs: allow the v4.1 callback thread to freeze The v4.1 callback thread has set_freezable() at the top, but it doesn't ever try to freeze within the loop. Have it call try_to_freeze() at the top of the loop. If a freeze event occurs, recheck kthread_should_stop() after thawing. Reported-by: Yanchuan Nian Signed-off-by: Jeff Layton Signed-off-by: Trond Myklebust --- fs/nfs/callback.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/fs/nfs/callback.c b/fs/nfs/callback.c index 5088b57b078a..cff089a412c7 100644 --- a/fs/nfs/callback.c +++ b/fs/nfs/callback.c @@ -125,6 +125,9 @@ nfs41_callback_svc(void *vrqstp) set_freezable(); while (!kthread_should_stop()) { + if (try_to_freeze()) + continue; + prepare_to_wait(&serv->sv_cb_waitq, &wq, TASK_INTERRUPTIBLE); spin_lock_bh(&serv->sv_cb_lock); if (!list_empty(&serv->sv_cb_list)) { From cb6071874703565762c190e1e7ebde762e525063 Mon Sep 17 00:00:00 2001 From: Paul Bolle Date: Sat, 9 Mar 2013 17:02:31 +0100 Subject: [PATCH 23/47] sunrpc: drop "select NETVM" The Kconfig entry for SUNRPC_SWAP selects NETVM. That select statement was added in commit a564b8f0398636ba30b07c0eaebdef7ff7837249 ("nfs: enable swap on NFS"). But there's no Kconfig symbol NETVM. It apparently was only in used in development versions of the swap over nfs functionality but never entered mainline. Anyhow, it is a nop and can safely be dropped. Signed-off-by: Paul Bolle Signed-off-by: Trond Myklebust --- net/sunrpc/Kconfig | 1 - 1 file changed, 1 deletion(-) diff --git a/net/sunrpc/Kconfig b/net/sunrpc/Kconfig index 516fe2caac2c..262caf03bd5f 100644 --- a/net/sunrpc/Kconfig +++ b/net/sunrpc/Kconfig @@ -24,7 +24,6 @@ config SUNRPC_XPRT_RDMA config SUNRPC_SWAP bool depends on SUNRPC - select NETVM config RPCSEC_GSS_KRB5 tristate "Secure RPC: Kerberos V mechanism" From dbb21c25a35a71baf413f5176f028ee11b88cfbc Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 1 Apr 2013 14:27:29 -0400 Subject: [PATCH 24/47] NFSv4: Handle NFS4ERR_DELAY and NFS4ERR_GRACE in nfs4_lock_delegation_recall A server shouldn't normally return NFS4ERR_GRACE if the client holds a delegation, since no conflicting lock reclaims can be granted, however the spec does not require the server to grant the lock in this instance. Signed-off-by: Trond Myklebust Cc: stable@vger.kernel.org --- fs/nfs/nfs4proc.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 81343944e096..dc4037c52cf7 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -5176,6 +5176,12 @@ int nfs4_lock_delegation_recall(struct nfs4_state *state, struct file_lock *fl) nfs4_schedule_stateid_recovery(server, state); err = 0; goto out; + case -NFS4ERR_DELAY: + case -NFS4ERR_GRACE: + set_bit(NFS_DELEGATED_STATE, &state->flags); + ssleep(1); + err = -EAGAIN; + goto out; case -ENOMEM: case -NFS4ERR_DENIED: /* kill_proc(fl->fl_pid, SIGLOST, 1); */ From 8b6cc4d6f841d31f72fe7478453759166d366274 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 1 Apr 2013 15:34:05 -0400 Subject: [PATCH 25/47] NFSv4: Handle NFS4ERR_DELAY and NFS4ERR_GRACE in nfs4_open_delegation_recall A server shouldn't normally return NFS4ERR_GRACE if the client holds a delegation, since no conflicting lock reclaims can be granted, however the spec does not require the server to grant the open in this instance Signed-off-by: Trond Myklebust Cc: stable@vger.kernel.org --- fs/nfs/nfs4proc.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index dc4037c52cf7..6630db1336a9 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -1432,6 +1432,12 @@ int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state case -ENOMEM: err = 0; goto out; + case -NFS4ERR_DELAY: + case -NFS4ERR_GRACE: + set_bit(NFS_DELEGATED_STATE, &state->flags); + ssleep(1); + err = -EAGAIN; + goto out; } set_bit(NFS_DELEGATED_STATE, &state->flags); err = nfs4_handle_exception(server, err, &exception); From 4a706fa09fd59b9c73b2a933953d27f3e18619c1 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 1 Apr 2013 14:47:22 -0400 Subject: [PATCH 26/47] NFSv4: Clean up nfs4_lock_delegation_recall All error cases are handled by the switch() statement, meaning that the call to nfs4_handle_exception() is unreachable. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 86 +++++++++++++++++++++-------------------------- 1 file changed, 38 insertions(+), 48 deletions(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 6630db1336a9..5f2010e0926e 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -5144,59 +5144,49 @@ nfs4_proc_lock(struct file *filp, int cmd, struct file_lock *request) int nfs4_lock_delegation_recall(struct nfs4_state *state, struct file_lock *fl) { struct nfs_server *server = NFS_SERVER(state->inode); - struct nfs4_exception exception = { }; int err; err = nfs4_set_lock_state(state, fl); if (err != 0) goto out; - do { - err = _nfs4_do_setlk(state, F_SETLK, fl, NFS_LOCK_NEW); - switch (err) { - default: - printk(KERN_ERR "NFS: %s: unhandled error " - "%d.\n", __func__, err); - case 0: - case -ESTALE: - goto out; - case -NFS4ERR_STALE_CLIENTID: - case -NFS4ERR_STALE_STATEID: - set_bit(NFS_DELEGATED_STATE, &state->flags); - case -NFS4ERR_EXPIRED: - nfs4_schedule_lease_recovery(server->nfs_client); - err = -EAGAIN; - goto out; - case -NFS4ERR_BADSESSION: - case -NFS4ERR_BADSLOT: - case -NFS4ERR_BAD_HIGH_SLOT: - case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION: - case -NFS4ERR_DEADSESSION: - set_bit(NFS_DELEGATED_STATE, &state->flags); - nfs4_schedule_session_recovery(server->nfs_client->cl_session, err); - err = -EAGAIN; - goto out; - case -NFS4ERR_DELEG_REVOKED: - case -NFS4ERR_ADMIN_REVOKED: - case -NFS4ERR_BAD_STATEID: - case -NFS4ERR_OPENMODE: - nfs4_schedule_stateid_recovery(server, state); - err = 0; - goto out; - case -NFS4ERR_DELAY: - case -NFS4ERR_GRACE: - set_bit(NFS_DELEGATED_STATE, &state->flags); - ssleep(1); - err = -EAGAIN; - goto out; - case -ENOMEM: - case -NFS4ERR_DENIED: - /* kill_proc(fl->fl_pid, SIGLOST, 1); */ - err = 0; - goto out; - } - set_bit(NFS_DELEGATED_STATE, &state->flags); - err = nfs4_handle_exception(server, err, &exception); - } while (exception.retry); + err = _nfs4_do_setlk(state, F_SETLK, fl, NFS_LOCK_NEW); + switch (err) { + default: + printk(KERN_ERR "NFS: %s: unhandled error " + "%d.\n", __func__, err); + case 0: + case -ESTALE: + goto out; + case -NFS4ERR_STALE_CLIENTID: + case -NFS4ERR_STALE_STATEID: + set_bit(NFS_DELEGATED_STATE, &state->flags); + case -NFS4ERR_EXPIRED: + nfs4_schedule_lease_recovery(server->nfs_client); + return -EAGAIN; + case -NFS4ERR_BADSESSION: + case -NFS4ERR_BADSLOT: + case -NFS4ERR_BAD_HIGH_SLOT: + case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION: + case -NFS4ERR_DEADSESSION: + set_bit(NFS_DELEGATED_STATE, &state->flags); + nfs4_schedule_session_recovery(server->nfs_client->cl_session, err); + return -EAGAIN; + case -NFS4ERR_DELEG_REVOKED: + case -NFS4ERR_ADMIN_REVOKED: + case -NFS4ERR_BAD_STATEID: + case -NFS4ERR_OPENMODE: + nfs4_schedule_stateid_recovery(server, state); + return 0; + case -NFS4ERR_DELAY: + case -NFS4ERR_GRACE: + set_bit(NFS_DELEGATED_STATE, &state->flags); + ssleep(1); + return -EAGAIN; + case -ENOMEM: + case -NFS4ERR_DENIED: + /* kill_proc(fl->fl_pid, SIGLOST, 1); */ + return 0; + } out: return err; } From be76b5b68d15861d50ccd1d3e66dac3cb19963dd Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 1 Apr 2013 15:40:44 -0400 Subject: [PATCH 27/47] NFSv4: Clean up nfs4_open_delegation_recall Make it symmetric with nfs4_lock_delegation_recall Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 96 ++++++++++++++++++++--------------------------- 1 file changed, 41 insertions(+), 55 deletions(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 5f2010e0926e..57b5fb778f16 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -1379,70 +1379,56 @@ static int nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *sta return ret; } -static int _nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid) +int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid) { + struct nfs_server *server = NFS_SERVER(state->inode); struct nfs4_opendata *opendata; - int ret; + int err; opendata = nfs4_open_recoverdata_alloc(ctx, state, NFS4_OPEN_CLAIM_DELEG_CUR_FH); if (IS_ERR(opendata)) return PTR_ERR(opendata); nfs4_stateid_copy(&opendata->o_arg.u.delegation, stateid); - ret = nfs4_open_recover(opendata, state); + err = nfs4_open_recover(opendata, state); nfs4_opendata_put(opendata); - return ret; -} - -int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid) -{ - struct nfs4_exception exception = { }; - struct nfs_server *server = NFS_SERVER(state->inode); - int err; - do { - err = _nfs4_open_delegation_recall(ctx, state, stateid); - switch (err) { - case 0: - case -ENOENT: - case -ESTALE: - goto out; - case -NFS4ERR_BADSESSION: - case -NFS4ERR_BADSLOT: - case -NFS4ERR_BAD_HIGH_SLOT: - case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION: - case -NFS4ERR_DEADSESSION: - set_bit(NFS_DELEGATED_STATE, &state->flags); - nfs4_schedule_session_recovery(server->nfs_client->cl_session, err); - err = -EAGAIN; - goto out; - case -NFS4ERR_STALE_CLIENTID: - case -NFS4ERR_STALE_STATEID: - set_bit(NFS_DELEGATED_STATE, &state->flags); - case -NFS4ERR_EXPIRED: - /* Don't recall a delegation if it was lost */ - nfs4_schedule_lease_recovery(server->nfs_client); - err = -EAGAIN; - goto out; - case -NFS4ERR_DELEG_REVOKED: - case -NFS4ERR_ADMIN_REVOKED: - case -NFS4ERR_BAD_STATEID: - nfs_inode_find_state_and_recover(state->inode, - stateid); - nfs4_schedule_stateid_recovery(server, state); - case -ENOMEM: - err = 0; - goto out; - case -NFS4ERR_DELAY: - case -NFS4ERR_GRACE: - set_bit(NFS_DELEGATED_STATE, &state->flags); - ssleep(1); - err = -EAGAIN; - goto out; - } - set_bit(NFS_DELEGATED_STATE, &state->flags); - err = nfs4_handle_exception(server, err, &exception); - } while (exception.retry); -out: + switch (err) { + default: + printk(KERN_ERR "NFS: %s: unhandled error " + "%d.\n", __func__, err); + case 0: + case -ENOENT: + case -ESTALE: + break; + case -NFS4ERR_BADSESSION: + case -NFS4ERR_BADSLOT: + case -NFS4ERR_BAD_HIGH_SLOT: + case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION: + case -NFS4ERR_DEADSESSION: + set_bit(NFS_DELEGATED_STATE, &state->flags); + nfs4_schedule_session_recovery(server->nfs_client->cl_session, err); + return -EAGAIN; + case -NFS4ERR_STALE_CLIENTID: + case -NFS4ERR_STALE_STATEID: + set_bit(NFS_DELEGATED_STATE, &state->flags); + case -NFS4ERR_EXPIRED: + /* Don't recall a delegation if it was lost */ + nfs4_schedule_lease_recovery(server->nfs_client); + return -EAGAIN; + case -NFS4ERR_DELEG_REVOKED: + case -NFS4ERR_ADMIN_REVOKED: + case -NFS4ERR_BAD_STATEID: + nfs_inode_find_state_and_recover(state->inode, + stateid); + nfs4_schedule_stateid_recovery(server, state); + case -ENOMEM: + return 0; + case -NFS4ERR_DELAY: + case -NFS4ERR_GRACE: + set_bit(NFS_DELEGATED_STATE, &state->flags); + ssleep(1); + return -EAGAIN; + } return err; } From db4f2e637f6d5dff6d07112e62605275be6050b3 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 1 Apr 2013 15:56:46 -0400 Subject: [PATCH 28/47] NFSv4: Clean up delegation recall error handling Unify the error handling in nfs4_open_delegation_recall and nfs4_lock_delegation_recall. Signed-off-by: Trond Myklebust --- fs/nfs/delegation.c | 6 ++-- fs/nfs/delegation.h | 2 +- fs/nfs/nfs4proc.c | 79 ++++++++++++++------------------------------- 3 files changed, 29 insertions(+), 58 deletions(-) diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c index 6390a4b5fee7..fc8a213497a1 100644 --- a/fs/nfs/delegation.c +++ b/fs/nfs/delegation.c @@ -64,7 +64,7 @@ int nfs4_have_delegation(struct inode *inode, fmode_t flags) return ret; } -static int nfs_delegation_claim_locks(struct nfs_open_context *ctx, struct nfs4_state *state) +static int nfs_delegation_claim_locks(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid) { struct inode *inode = state->inode; struct file_lock *fl; @@ -83,7 +83,7 @@ static int nfs_delegation_claim_locks(struct nfs_open_context *ctx, struct nfs4_ if (nfs_file_open_context(fl->fl_file) != ctx) continue; unlock_flocks(); - status = nfs4_lock_delegation_recall(state, fl); + status = nfs4_lock_delegation_recall(fl, state, stateid); if (status < 0) goto out; lock_flocks(); @@ -120,7 +120,7 @@ static int nfs_delegation_claim_opens(struct inode *inode, const nfs4_stateid *s seq = raw_seqcount_begin(&sp->so_reclaim_seqcount); err = nfs4_open_delegation_recall(ctx, state, stateid); if (!err) - err = nfs_delegation_claim_locks(ctx, state); + err = nfs_delegation_claim_locks(ctx, state, stateid); if (!err && read_seqcount_retry(&sp->so_reclaim_seqcount, seq)) err = -EAGAIN; mutex_unlock(&sp->so_delegreturn_mutex); diff --git a/fs/nfs/delegation.h b/fs/nfs/delegation.h index d54d4fca6793..88573d107502 100644 --- a/fs/nfs/delegation.h +++ b/fs/nfs/delegation.h @@ -53,7 +53,7 @@ void nfs_delegation_reap_unclaimed(struct nfs_client *clp); /* NFSv4 delegation-related procedures */ int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync); int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid); -int nfs4_lock_delegation_recall(struct nfs4_state *state, struct file_lock *fl); +int nfs4_lock_delegation_recall(struct file_lock *fl, struct nfs4_state *state, const nfs4_stateid *stateid); bool nfs4_copy_delegation_stateid(nfs4_stateid *dst, struct inode *inode, fmode_t flags); void nfs_mark_delegation_referenced(struct nfs_delegation *delegation); diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 57b5fb778f16..bdf3d07ce974 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -1379,19 +1379,8 @@ static int nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *sta return ret; } -int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid) +static int nfs4_handle_delegation_recall_error(struct nfs_server *server, struct nfs4_state *state, const nfs4_stateid *stateid, int err) { - struct nfs_server *server = NFS_SERVER(state->inode); - struct nfs4_opendata *opendata; - int err; - - opendata = nfs4_open_recoverdata_alloc(ctx, state, - NFS4_OPEN_CLAIM_DELEG_CUR_FH); - if (IS_ERR(opendata)) - return PTR_ERR(opendata); - nfs4_stateid_copy(&opendata->o_arg.u.delegation, stateid); - err = nfs4_open_recover(opendata, state); - nfs4_opendata_put(opendata); switch (err) { default: printk(KERN_ERR "NFS: %s: unhandled error " @@ -1418,20 +1407,40 @@ int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state case -NFS4ERR_DELEG_REVOKED: case -NFS4ERR_ADMIN_REVOKED: case -NFS4ERR_BAD_STATEID: + case -NFS4ERR_OPENMODE: nfs_inode_find_state_and_recover(state->inode, stateid); nfs4_schedule_stateid_recovery(server, state); - case -ENOMEM: return 0; case -NFS4ERR_DELAY: case -NFS4ERR_GRACE: set_bit(NFS_DELEGATED_STATE, &state->flags); ssleep(1); return -EAGAIN; + case -ENOMEM: + case -NFS4ERR_DENIED: + /* kill_proc(fl->fl_pid, SIGLOST, 1); */ + return 0; } return err; } +int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid) +{ + struct nfs_server *server = NFS_SERVER(state->inode); + struct nfs4_opendata *opendata; + int err; + + opendata = nfs4_open_recoverdata_alloc(ctx, state, + NFS4_OPEN_CLAIM_DELEG_CUR_FH); + if (IS_ERR(opendata)) + return PTR_ERR(opendata); + nfs4_stateid_copy(&opendata->o_arg.u.delegation, stateid); + err = nfs4_open_recover(opendata, state); + nfs4_opendata_put(opendata); + return nfs4_handle_delegation_recall_error(server, state, stateid, err); +} + static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata) { struct nfs4_opendata *data = calldata; @@ -5127,54 +5136,16 @@ nfs4_proc_lock(struct file *filp, int cmd, struct file_lock *request) return status; } -int nfs4_lock_delegation_recall(struct nfs4_state *state, struct file_lock *fl) +int nfs4_lock_delegation_recall(struct file_lock *fl, struct nfs4_state *state, const nfs4_stateid *stateid) { struct nfs_server *server = NFS_SERVER(state->inode); int err; err = nfs4_set_lock_state(state, fl); if (err != 0) - goto out; + return err; err = _nfs4_do_setlk(state, F_SETLK, fl, NFS_LOCK_NEW); - switch (err) { - default: - printk(KERN_ERR "NFS: %s: unhandled error " - "%d.\n", __func__, err); - case 0: - case -ESTALE: - goto out; - case -NFS4ERR_STALE_CLIENTID: - case -NFS4ERR_STALE_STATEID: - set_bit(NFS_DELEGATED_STATE, &state->flags); - case -NFS4ERR_EXPIRED: - nfs4_schedule_lease_recovery(server->nfs_client); - return -EAGAIN; - case -NFS4ERR_BADSESSION: - case -NFS4ERR_BADSLOT: - case -NFS4ERR_BAD_HIGH_SLOT: - case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION: - case -NFS4ERR_DEADSESSION: - set_bit(NFS_DELEGATED_STATE, &state->flags); - nfs4_schedule_session_recovery(server->nfs_client->cl_session, err); - return -EAGAIN; - case -NFS4ERR_DELEG_REVOKED: - case -NFS4ERR_ADMIN_REVOKED: - case -NFS4ERR_BAD_STATEID: - case -NFS4ERR_OPENMODE: - nfs4_schedule_stateid_recovery(server, state); - return 0; - case -NFS4ERR_DELAY: - case -NFS4ERR_GRACE: - set_bit(NFS_DELEGATED_STATE, &state->flags); - ssleep(1); - return -EAGAIN; - case -ENOMEM: - case -NFS4ERR_DENIED: - /* kill_proc(fl->fl_pid, SIGLOST, 1); */ - return 0; - } -out: - return err; + return nfs4_handle_delegation_recall_error(server, state, stateid, err); } struct nfs_release_lockowner_data { From b757144fd77cf5512f5b60179ba5ca8dcc5184b4 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 3 Apr 2013 14:33:49 -0400 Subject: [PATCH 29/47] NFSv4: Be less aggressive about returning delegations for open files Currently, if the application that holds the file open isn't doing I/O, we may end up returning the delegation. This means that we can no longer cache the file as aggressively, and often also that we multiply the state that both the server and the client needs to track. This patch adds a check for open files to the routine that scans for delegations that are unreferenced. Signed-off-by: Trond Myklebust --- fs/nfs/delegation.c | 30 +++++++++++++++++++++++++++--- fs/nfs/delegation.h | 1 + 2 files changed, 28 insertions(+), 3 deletions(-) diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c index fc8a213497a1..a377ea36381e 100644 --- a/fs/nfs/delegation.c +++ b/fs/nfs/delegation.c @@ -389,6 +389,24 @@ static int nfs_end_delegation_return(struct inode *inode, struct nfs_delegation return err; } +static bool nfs_delegation_need_return(struct nfs_delegation *delegation) +{ + bool ret = false; + + if (test_and_clear_bit(NFS_DELEGATION_RETURN, &delegation->flags)) + ret = true; + if (test_and_clear_bit(NFS_DELEGATION_RETURN_IF_CLOSED, &delegation->flags) && !ret) { + struct inode *inode; + + spin_lock(&delegation->lock); + inode = delegation->inode; + if (inode && list_empty(&NFS_I(inode)->open_files)) + ret = true; + spin_unlock(&delegation->lock); + } + return ret; +} + /** * nfs_client_return_marked_delegations - return previously marked delegations * @clp: nfs_client to process @@ -411,8 +429,7 @@ int nfs_client_return_marked_delegations(struct nfs_client *clp) list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link) { list_for_each_entry_rcu(delegation, &server->delegations, super_list) { - if (!test_and_clear_bit(NFS_DELEGATION_RETURN, - &delegation->flags)) + if (!nfs_delegation_need_return(delegation)) continue; inode = nfs_delegation_grab_inode(delegation); if (inode == NULL) @@ -471,6 +488,13 @@ int nfs4_inode_return_delegation(struct inode *inode) return err; } +static void nfs_mark_return_if_closed_delegation(struct nfs_server *server, + struct nfs_delegation *delegation) +{ + set_bit(NFS_DELEGATION_RETURN_IF_CLOSED, &delegation->flags); + set_bit(NFS4CLNT_DELEGRETURN, &server->nfs_client->cl_state); +} + static void nfs_mark_return_delegation(struct nfs_server *server, struct nfs_delegation *delegation) { @@ -574,7 +598,7 @@ static void nfs_mark_return_unreferenced_delegations(struct nfs_server *server) list_for_each_entry_rcu(delegation, &server->delegations, super_list) { if (test_and_clear_bit(NFS_DELEGATION_REFERENCED, &delegation->flags)) continue; - nfs_mark_return_delegation(server, delegation); + nfs_mark_return_if_closed_delegation(server, delegation); } } diff --git a/fs/nfs/delegation.h b/fs/nfs/delegation.h index 88573d107502..22f0a138e551 100644 --- a/fs/nfs/delegation.h +++ b/fs/nfs/delegation.h @@ -28,6 +28,7 @@ struct nfs_delegation { enum { NFS_DELEGATION_NEED_RECLAIM = 0, NFS_DELEGATION_RETURN, + NFS_DELEGATION_RETURN_IF_CLOSED, NFS_DELEGATION_REFERENCED, NFS_DELEGATION_RETURNING, }; From 5c31e2368f39aee062cb697d4b8857c64c7cef7c Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 3 Apr 2013 19:04:58 -0400 Subject: [PATCH 30/47] NFSv4: Fix nfs_server_return_all_delegations If the state manager thread is already running, we may end up racing with it in nfs_client_return_marked_delegations. Better to just allow the state manager thread to do the job. Signed-off-by: Trond Myklebust --- fs/nfs/delegation.c | 24 +++++++++++++++++------- 1 file changed, 17 insertions(+), 7 deletions(-) diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c index a377ea36381e..213f1bbeb828 100644 --- a/fs/nfs/delegation.c +++ b/fs/nfs/delegation.c @@ -502,6 +502,18 @@ static void nfs_mark_return_delegation(struct nfs_server *server, set_bit(NFS4CLNT_DELEGRETURN, &server->nfs_client->cl_state); } +static bool nfs_server_mark_return_all_delegations(struct nfs_server *server) +{ + struct nfs_delegation *delegation; + bool ret = false; + + list_for_each_entry_rcu(delegation, &server->delegations, super_list) { + nfs_mark_return_delegation(server, delegation); + ret = true; + } + return ret; +} + /** * nfs_super_return_all_delegations - return delegations for one superblock * @sb: sb to process @@ -510,21 +522,19 @@ static void nfs_mark_return_delegation(struct nfs_server *server, void nfs_server_return_all_delegations(struct nfs_server *server) { struct nfs_client *clp = server->nfs_client; - struct nfs_delegation *delegation; + bool need_wait; if (clp == NULL) return; rcu_read_lock(); - list_for_each_entry_rcu(delegation, &server->delegations, super_list) { - spin_lock(&delegation->lock); - set_bit(NFS_DELEGATION_RETURN, &delegation->flags); - spin_unlock(&delegation->lock); - } + need_wait = nfs_server_mark_return_all_delegations(server); rcu_read_unlock(); - if (nfs_client_return_marked_delegations(clp) != 0) + if (need_wait) { nfs4_schedule_state_manager(clp); + nfs4_wait_clnt_recover(clp); + } } static void nfs_mark_return_all_delegation_types(struct nfs_server *server, From b02ba0b66095d2fdcc6b74538aa5a0ae13976745 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 3 Apr 2013 19:23:58 -0400 Subject: [PATCH 31/47] NFSv4: Clean up nfs_expire_all_delegations Signed-off-by: Trond Myklebust --- fs/nfs/delegation.c | 43 +++++++++++++++++++++++++++---------------- 1 file changed, 27 insertions(+), 16 deletions(-) diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c index 213f1bbeb828..60e2e6e3e323 100644 --- a/fs/nfs/delegation.c +++ b/fs/nfs/delegation.c @@ -514,6 +514,33 @@ static bool nfs_server_mark_return_all_delegations(struct nfs_server *server) return ret; } +static void nfs_client_mark_return_all_delegations(struct nfs_client *clp) +{ + struct nfs_server *server; + + rcu_read_lock(); + list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link) + nfs_server_mark_return_all_delegations(server); + rcu_read_unlock(); +} + +static void nfs_delegation_run_state_manager(struct nfs_client *clp) +{ + if (test_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state)) + nfs4_schedule_state_manager(clp); +} + +/** + * nfs_expire_all_delegations + * @clp: client to process + * + */ +void nfs_expire_all_delegations(struct nfs_client *clp) +{ + nfs_client_mark_return_all_delegations(clp); + nfs_delegation_run_state_manager(clp); +} + /** * nfs_super_return_all_delegations - return delegations for one superblock * @sb: sb to process @@ -561,12 +588,6 @@ static void nfs_client_mark_return_all_delegation_types(struct nfs_client *clp, rcu_read_unlock(); } -static void nfs_delegation_run_state_manager(struct nfs_client *clp) -{ - if (test_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state)) - nfs4_schedule_state_manager(clp); -} - void nfs_remove_bad_delegation(struct inode *inode) { struct nfs_delegation *delegation; @@ -591,16 +612,6 @@ void nfs_expire_all_delegation_types(struct nfs_client *clp, fmode_t flags) nfs_delegation_run_state_manager(clp); } -/** - * nfs_expire_all_delegations - * @clp: client to process - * - */ -void nfs_expire_all_delegations(struct nfs_client *clp) -{ - nfs_expire_all_delegation_types(clp, FMODE_READ|FMODE_WRITE); -} - static void nfs_mark_return_unreferenced_delegations(struct nfs_server *server) { struct nfs_delegation *delegation; From 826e0013082a86fb16f2e414314a5268f744fb96 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Wed, 3 Apr 2013 19:27:52 -0400 Subject: [PATCH 32/47] NFSv4: Fix CB_RECALL_ANY to only return delegations that are not in use Signed-off-by: Trond Myklebust --- fs/nfs/callback_proc.c | 2 +- fs/nfs/delegation.c | 14 +++++++------- fs/nfs/delegation.h | 2 +- 3 files changed, 9 insertions(+), 9 deletions(-) diff --git a/fs/nfs/callback_proc.c b/fs/nfs/callback_proc.c index 2960512792c2..a13d26ede254 100644 --- a/fs/nfs/callback_proc.c +++ b/fs/nfs/callback_proc.c @@ -500,7 +500,7 @@ __be32 nfs4_callback_recallany(struct cb_recallanyargs *args, void *dummy, &args->craa_type_mask)) pnfs_recall_all_layouts(cps->clp); if (flags) - nfs_expire_all_delegation_types(cps->clp, flags); + nfs_expire_unused_delegation_types(cps->clp, flags); out: dprintk("%s: exit with status = %d\n", __func__, ntohl(status)); return status; diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c index 60e2e6e3e323..f30bd1ec63b9 100644 --- a/fs/nfs/delegation.c +++ b/fs/nfs/delegation.c @@ -564,7 +564,7 @@ void nfs_server_return_all_delegations(struct nfs_server *server) } } -static void nfs_mark_return_all_delegation_types(struct nfs_server *server, +static void nfs_mark_return_unused_delegation_types(struct nfs_server *server, fmode_t flags) { struct nfs_delegation *delegation; @@ -573,18 +573,18 @@ static void nfs_mark_return_all_delegation_types(struct nfs_server *server, if ((delegation->type == (FMODE_READ|FMODE_WRITE)) && !(flags & FMODE_WRITE)) continue; if (delegation->type & flags) - nfs_mark_return_delegation(server, delegation); + nfs_mark_return_if_closed_delegation(server, delegation); } } -static void nfs_client_mark_return_all_delegation_types(struct nfs_client *clp, +static void nfs_client_mark_return_unused_delegation_types(struct nfs_client *clp, fmode_t flags) { struct nfs_server *server; rcu_read_lock(); list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link) - nfs_mark_return_all_delegation_types(server, flags); + nfs_mark_return_unused_delegation_types(server, flags); rcu_read_unlock(); } @@ -601,14 +601,14 @@ void nfs_remove_bad_delegation(struct inode *inode) EXPORT_SYMBOL_GPL(nfs_remove_bad_delegation); /** - * nfs_expire_all_delegation_types + * nfs_expire_unused_delegation_types * @clp: client to process * @flags: delegation types to expire * */ -void nfs_expire_all_delegation_types(struct nfs_client *clp, fmode_t flags) +void nfs_expire_unused_delegation_types(struct nfs_client *clp, fmode_t flags) { - nfs_client_mark_return_all_delegation_types(clp, flags); + nfs_client_mark_return_unused_delegation_types(clp, flags); nfs_delegation_run_state_manager(clp); } diff --git a/fs/nfs/delegation.h b/fs/nfs/delegation.h index 22f0a138e551..9a79c7a99d6d 100644 --- a/fs/nfs/delegation.h +++ b/fs/nfs/delegation.h @@ -42,7 +42,7 @@ void nfs_inode_return_delegation_noreclaim(struct inode *inode); struct inode *nfs_delegation_find_inode(struct nfs_client *clp, const struct nfs_fh *fhandle); void nfs_server_return_all_delegations(struct nfs_server *); void nfs_expire_all_delegations(struct nfs_client *clp); -void nfs_expire_all_delegation_types(struct nfs_client *clp, fmode_t flags); +void nfs_expire_unused_delegation_types(struct nfs_client *clp, fmode_t flags); void nfs_expire_unreferenced_delegations(struct nfs_client *clp); int nfs_client_return_marked_delegations(struct nfs_client *clp); int nfs_delegations_present(struct nfs_client *clp); From bc7a05ca5156915a5aada26d64ee035fdd5e5d25 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 8 Apr 2013 17:50:28 -0400 Subject: [PATCH 33/47] NFSv4: Handle timeouts correctly when probing for lease validity When we send a RENEW or SEQUENCE operation in order to probe if the lease is still valid, we want it to be able to time out since the lease we are probing is likely to time out too. Currently, because we use soft mount semantics for these RPC calls, the return value is EIO, which causes the state manager to exit with an "unhandled error" message. This patch changes the call semantics, so that the RPC layer returns ETIMEDOUT instead of EIO. We then have the state manager default to a simple retry instead of exiting. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 6 +++--- fs/nfs/nfs4state.c | 4 ++++ 2 files changed, 7 insertions(+), 3 deletions(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index bdf3d07ce974..264ddb16255f 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -3797,7 +3797,7 @@ static int nfs4_proc_async_renew(struct nfs_client *clp, struct rpc_cred *cred, return -ENOMEM; data->client = clp; data->timestamp = jiffies; - return rpc_call_async(clp->cl_rpcclient, &msg, RPC_TASK_SOFT, + return rpc_call_async(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT, &nfs4_renew_ops, data); } @@ -3811,7 +3811,7 @@ static int nfs4_proc_renew(struct nfs_client *clp, struct rpc_cred *cred) unsigned long now = jiffies; int status; - status = rpc_call_sync(clp->cl_rpcclient, &msg, 0); + status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT); if (status < 0) return status; do_renew_lease(clp, now); @@ -5959,7 +5959,7 @@ static struct rpc_task *_nfs41_proc_sequence(struct nfs_client *clp, .rpc_client = clp->cl_rpcclient, .rpc_message = &msg, .callback_ops = &nfs41_sequence_ops, - .flags = RPC_TASK_ASYNC | RPC_TASK_SOFT, + .flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT, }; if (!atomic_inc_not_zero(&clp->cl_count)) diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index b924bdd69494..1eb17285c99a 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -1739,6 +1739,10 @@ static int nfs4_check_lease(struct nfs_client *clp) } status = ops->renew_lease(clp, cred); put_rpccred(cred); + if (status == -ETIMEDOUT) { + set_bit(NFS4CLNT_CHECK_LEASE, &clp->cl_state); + return 0; + } out: return nfs4_recovery_handle_error(clp, status); } From 577b42327d707fbe7166aad6902c2eeee6a65015 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 8 Apr 2013 21:38:12 -0400 Subject: [PATCH 34/47] NFS: Add functionality to allow waiting on all outstanding reads to complete This will later allow NFS locking code to wait for readahead to complete before releasing byte range locks. Signed-off-by: Trond Myklebust --- fs/nfs/inode.c | 1 + fs/nfs/internal.h | 7 ++++++ fs/nfs/pagelist.c | 51 ++++++++++++++++++++++++++++++++++++++++++ include/linux/nfs_fs.h | 7 ++++++ 4 files changed, 66 insertions(+) diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index 55b840f05ab2..c1c7a9d78722 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -561,6 +561,7 @@ static void nfs_init_lock_context(struct nfs_lock_context *l_ctx) l_ctx->lockowner.l_owner = current->files; l_ctx->lockowner.l_pid = current->tgid; INIT_LIST_HEAD(&l_ctx->list); + nfs_iocounter_init(&l_ctx->io_count); } static struct nfs_lock_context *__nfs_find_lock_context(struct nfs_open_context *ctx) diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h index 541c9ebdbc5a..91e59a39fc08 100644 --- a/fs/nfs/internal.h +++ b/fs/nfs/internal.h @@ -229,6 +229,13 @@ extern void nfs_pgheader_init(struct nfs_pageio_descriptor *desc, struct nfs_pgio_header *hdr, void (*release)(struct nfs_pgio_header *hdr)); void nfs_set_pgio_error(struct nfs_pgio_header *hdr, int error, loff_t pos); +int nfs_iocounter_wait(struct nfs_io_counter *c); + +static inline void nfs_iocounter_init(struct nfs_io_counter *c) +{ + c->flags = 0; + atomic_set(&c->io_count, 0); +} /* nfs2xdr.c */ extern struct rpc_procinfo nfs_procedures[]; diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c index 7f0933086b36..29cfb7ade121 100644 --- a/fs/nfs/pagelist.c +++ b/fs/nfs/pagelist.c @@ -84,6 +84,55 @@ nfs_page_free(struct nfs_page *p) kmem_cache_free(nfs_page_cachep, p); } +static void +nfs_iocounter_inc(struct nfs_io_counter *c) +{ + atomic_inc(&c->io_count); +} + +static void +nfs_iocounter_dec(struct nfs_io_counter *c) +{ + if (atomic_dec_and_test(&c->io_count)) { + clear_bit(NFS_IO_INPROGRESS, &c->flags); + smp_mb__after_clear_bit(); + wake_up_bit(&c->flags, NFS_IO_INPROGRESS); + } +} + +static int +__nfs_iocounter_wait(struct nfs_io_counter *c) +{ + wait_queue_head_t *wq = bit_waitqueue(&c->flags, NFS_IO_INPROGRESS); + DEFINE_WAIT_BIT(q, &c->flags, NFS_IO_INPROGRESS); + int ret = 0; + + do { + prepare_to_wait(wq, &q.wait, TASK_KILLABLE); + set_bit(NFS_IO_INPROGRESS, &c->flags); + if (atomic_read(&c->io_count) == 0) + break; + ret = nfs_wait_bit_killable(&c->flags); + } while (atomic_read(&c->io_count) != 0); + finish_wait(wq, &q.wait); + return ret; +} + +/** + * nfs_iocounter_wait - wait for i/o to complete + * @c: nfs_io_counter to use + * + * returns -ERESTARTSYS if interrupted by a fatal signal. + * Otherwise returns 0 once the io_count hits 0. + */ +int +nfs_iocounter_wait(struct nfs_io_counter *c) +{ + if (atomic_read(&c->io_count) == 0) + return 0; + return __nfs_iocounter_wait(c); +} + /** * nfs_create_request - Create an NFS read/write request. * @ctx: open context to use @@ -118,6 +167,7 @@ nfs_create_request(struct nfs_open_context *ctx, struct inode *inode, return ERR_CAST(l_ctx); } req->wb_lock_context = l_ctx; + nfs_iocounter_inc(&l_ctx->io_count); /* Initialize the request struct. Initially, we assume a * long write-back delay. This will be adjusted in @@ -177,6 +227,7 @@ static void nfs_clear_request(struct nfs_page *req) req->wb_page = NULL; } if (l_ctx != NULL) { + nfs_iocounter_dec(&l_ctx->io_count); nfs_put_lock_context(l_ctx); req->wb_lock_context = NULL; } diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index f6b1956f3c86..fc01d5cb4cf1 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -59,11 +59,18 @@ struct nfs_lockowner { pid_t l_pid; }; +#define NFS_IO_INPROGRESS 0 +struct nfs_io_counter { + unsigned long flags; + atomic_t io_count; +}; + struct nfs_lock_context { atomic_t count; struct list_head list; struct nfs_open_context *open_context; struct nfs_lockowner lockowner; + struct nfs_io_counter io_count; }; struct nfs4_state; From 7a8203d8cbab63b7c556ebd00e12685a3ed176b5 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 8 Apr 2013 21:49:53 -0400 Subject: [PATCH 35/47] NFS: Ensure that NFS file unlock waits for readahead to complete Signed-off-by: Trond Myklebust --- fs/nfs/file.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/fs/nfs/file.c b/fs/nfs/file.c index 29f4a48a0ee6..a87a44f84113 100644 --- a/fs/nfs/file.c +++ b/fs/nfs/file.c @@ -744,6 +744,7 @@ static int do_unlk(struct file *filp, int cmd, struct file_lock *fl, int is_local) { struct inode *inode = filp->f_mapping->host; + struct nfs_lock_context *l_ctx; int status; /* @@ -752,6 +753,14 @@ do_unlk(struct file *filp, int cmd, struct file_lock *fl, int is_local) */ nfs_sync_mapping(filp->f_mapping); + l_ctx = nfs_get_lock_context(nfs_file_open_context(filp)); + if (!IS_ERR(l_ctx)) { + status = nfs_iocounter_wait(&l_ctx->io_count); + nfs_put_lock_context(l_ctx); + if (status < 0) + return status; + } + /* NOTE: special case * If we're signalled while cleaning up locks on process exit, we * still need to complete the unlock. From 314d7cc05d10208d6e63b0b1185f02c73a16adfb Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Wed, 10 Apr 2013 15:36:48 -0400 Subject: [PATCH 36/47] nfs: remove unnecessary check for NULL inode->i_flock from nfs_delegation_claim_locks The second check was added in commit 65b62a29 but it will never be true. Signed-off-by: Jeff Layton Signed-off-by: Trond Myklebust --- fs/nfs/delegation.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/fs/nfs/delegation.c b/fs/nfs/delegation.c index f30bd1ec63b9..57db3244f4d9 100644 --- a/fs/nfs/delegation.c +++ b/fs/nfs/delegation.c @@ -70,11 +70,9 @@ static int nfs_delegation_claim_locks(struct nfs_open_context *ctx, struct nfs4_ struct file_lock *fl; int status = 0; - if (inode->i_flock == NULL) - return 0; - if (inode->i_flock == NULL) goto out; + /* Protect inode->i_flock using the file locks lock */ lock_flocks(); for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) { From b9536ad52152ff440231dcc2dd9cfa644f5413c9 Mon Sep 17 00:00:00 2001 From: Andy Adamson Date: Thu, 11 Apr 2013 09:28:45 -0400 Subject: [PATCH 37/47] NFSv4 release the sequence id in the return on close case Otherwise we deadlock if state recovery is initiated while we sleep. Signed-off-by: Andy Adamson Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 264ddb16255f..83197fb60d5f 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -2275,8 +2275,10 @@ static void nfs4_close_prepare(struct rpc_task *task, void *data) if (calldata->arg.fmode == 0) { task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE]; if (calldata->roc && - pnfs_roc_drain(inode, &calldata->roc_barrier, task)) + pnfs_roc_drain(inode, &calldata->roc_barrier, task)) { + nfs_release_seqid(calldata->arg.seqid); goto out_wait; + } } nfs_fattr_init(calldata->res.fattr); From b570a975ed276335dc7d148658c1f880ac0a507f Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Fri, 12 Apr 2013 15:04:51 -0400 Subject: [PATCH 38/47] NFSv4: Fix handling of revoked delegations by setattr Currently, _nfs4_do_setattr() will use the delegation stateid if no writeable open file stateid is available. If the server revokes that delegation stateid, then the call to nfs4_handle_exception() will fail to handle the error due to the lack of a struct nfs4_state, and will just convert the error into an EIO. This patch just removes the requirement that we must have a struct nfs4_state in order to invalidate the delegation and retry. Reported-by: Andy Adamson Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 83197fb60d5f..af05df3c7c79 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -304,9 +304,13 @@ static int nfs4_handle_exception(struct nfs_server *server, int errorcode, struc case -NFS4ERR_DELEG_REVOKED: case -NFS4ERR_ADMIN_REVOKED: case -NFS4ERR_BAD_STATEID: + if (inode != NULL && nfs4_have_delegation(inode, FMODE_READ)) { + nfs_remove_bad_delegation(inode); + exception->retry = 1; + break; + } if (state == NULL) break; - nfs_remove_bad_delegation(state->inode); ret = nfs4_schedule_stateid_recovery(server, state); if (ret < 0) break; From ba60eb25ff6be6f8e60488cdfd454e5c612bce60 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 14 Apr 2013 10:49:37 -0400 Subject: [PATCH 39/47] SUNRPC: Fix a livelock problem in the xprt->backlog queue This patch ensures that we throttle new RPC requests if there are requests already waiting in the xprt->backlog queue. The reason for doing this is to fix livelock issues that can occur when an existing (high priority) task is waiting in the backlog queue, gets woken up by xprt_free_slot(), but a new task then steals the slot. Signed-off-by: Trond Myklebust --- include/linux/sunrpc/xprt.h | 2 ++ net/sunrpc/clnt.c | 17 ++++++++++- net/sunrpc/xprt.c | 61 +++++++++++++++++++++++++++++++++++-- 3 files changed, 76 insertions(+), 4 deletions(-) diff --git a/include/linux/sunrpc/xprt.h b/include/linux/sunrpc/xprt.h index 30834be03011..12815f66a9c4 100644 --- a/include/linux/sunrpc/xprt.h +++ b/include/linux/sunrpc/xprt.h @@ -279,6 +279,7 @@ struct xprt_class { struct rpc_xprt *xprt_create_transport(struct xprt_create *args); void xprt_connect(struct rpc_task *task); void xprt_reserve(struct rpc_task *task); +void xprt_retry_reserve(struct rpc_task *task); int xprt_reserve_xprt(struct rpc_xprt *xprt, struct rpc_task *task); int xprt_reserve_xprt_cong(struct rpc_xprt *xprt, struct rpc_task *task); void xprt_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task); @@ -334,6 +335,7 @@ int xs_swapper(struct rpc_xprt *xprt, int enable); #define XPRT_CLOSING (6) #define XPRT_CONNECTION_ABORT (7) #define XPRT_CONNECTION_CLOSE (8) +#define XPRT_CONGESTED (9) static inline void xprt_set_connected(struct rpc_xprt *xprt) { diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c index b95a0a2d5eea..a80ee9b80dcf 100644 --- a/net/sunrpc/clnt.c +++ b/net/sunrpc/clnt.c @@ -1306,6 +1306,8 @@ call_reserve(struct rpc_task *task) xprt_reserve(task); } +static void call_retry_reserve(struct rpc_task *task); + /* * 1b. Grok the result of xprt_reserve() */ @@ -1347,7 +1349,7 @@ call_reserveresult(struct rpc_task *task) case -ENOMEM: rpc_delay(task, HZ >> 2); case -EAGAIN: /* woken up; retry */ - task->tk_action = call_reserve; + task->tk_action = call_retry_reserve; return; case -EIO: /* probably a shutdown */ break; @@ -1359,6 +1361,19 @@ call_reserveresult(struct rpc_task *task) rpc_exit(task, status); } +/* + * 1c. Retry reserving an RPC call slot + */ +static void +call_retry_reserve(struct rpc_task *task) +{ + dprint_status(task); + + task->tk_status = 0; + task->tk_action = call_reserveresult; + xprt_retry_reserve(task); +} + /* * 2. Bind and/or refresh the credentials */ diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c index b7478d5e7ffd..745fca3cfd36 100644 --- a/net/sunrpc/xprt.c +++ b/net/sunrpc/xprt.c @@ -948,6 +948,34 @@ void xprt_transmit(struct rpc_task *task) spin_unlock_bh(&xprt->transport_lock); } +static void xprt_add_backlog(struct rpc_xprt *xprt, struct rpc_task *task) +{ + set_bit(XPRT_CONGESTED, &xprt->state); + rpc_sleep_on(&xprt->backlog, task, NULL); +} + +static void xprt_wake_up_backlog(struct rpc_xprt *xprt) +{ + if (rpc_wake_up_next(&xprt->backlog) == NULL) + clear_bit(XPRT_CONGESTED, &xprt->state); +} + +static bool xprt_throttle_congested(struct rpc_xprt *xprt, struct rpc_task *task) +{ + bool ret = false; + + if (!test_bit(XPRT_CONGESTED, &xprt->state)) + goto out; + spin_lock(&xprt->reserve_lock); + if (test_bit(XPRT_CONGESTED, &xprt->state)) { + rpc_sleep_on(&xprt->backlog, task, NULL); + ret = true; + } + spin_unlock(&xprt->reserve_lock); +out: + return ret; +} + static struct rpc_rqst *xprt_dynamic_alloc_slot(struct rpc_xprt *xprt, gfp_t gfp_flags) { struct rpc_rqst *req = ERR_PTR(-EAGAIN); @@ -992,7 +1020,7 @@ void xprt_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task) task->tk_status = -ENOMEM; break; case -EAGAIN: - rpc_sleep_on(&xprt->backlog, task, NULL); + xprt_add_backlog(xprt, task); dprintk("RPC: waiting for request slot\n"); default: task->tk_status = -EAGAIN; @@ -1028,7 +1056,7 @@ static void xprt_free_slot(struct rpc_xprt *xprt, struct rpc_rqst *req) memset(req, 0, sizeof(*req)); /* mark unused */ list_add(&req->rq_list, &xprt->free); } - rpc_wake_up_next(&xprt->backlog); + xprt_wake_up_backlog(xprt); spin_unlock(&xprt->reserve_lock); } @@ -1092,13 +1120,40 @@ EXPORT_SYMBOL_GPL(xprt_free); * xprt_reserve - allocate an RPC request slot * @task: RPC task requesting a slot allocation * - * If no more slots are available, place the task on the transport's + * If the transport is marked as being congested, or if no more + * slots are available, place the task on the transport's * backlog queue. */ void xprt_reserve(struct rpc_task *task) { struct rpc_xprt *xprt; + task->tk_status = 0; + if (task->tk_rqstp != NULL) + return; + + task->tk_timeout = 0; + task->tk_status = -EAGAIN; + rcu_read_lock(); + xprt = rcu_dereference(task->tk_client->cl_xprt); + if (!xprt_throttle_congested(xprt, task)) + xprt->ops->alloc_slot(xprt, task); + rcu_read_unlock(); +} + +/** + * xprt_retry_reserve - allocate an RPC request slot + * @task: RPC task requesting a slot allocation + * + * If no more slots are available, place the task on the transport's + * backlog queue. + * Note that the only difference with xprt_reserve is that we now + * ignore the value of the XPRT_CONGESTED flag. + */ +void xprt_retry_reserve(struct rpc_task *task) +{ + struct rpc_xprt *xprt; + task->tk_status = 0; if (task->tk_rqstp != NULL) return; From b7993cebb841b0da7a33e9d5ce301a9fd3209165 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 14 Apr 2013 11:42:00 -0400 Subject: [PATCH 40/47] SUNRPC: Allow rpc_create() to request that TCP slots be unlimited This is mainly for use by NFSv4.1, where the session negotiation ultimately wants to decide how many RPC slots we can fill. Signed-off-by: Trond Myklebust --- include/linux/sunrpc/clnt.h | 1 + include/linux/sunrpc/xprt.h | 3 +++ net/sunrpc/clnt.c | 2 ++ net/sunrpc/xprtsock.c | 6 +++++- 4 files changed, 11 insertions(+), 1 deletion(-) diff --git a/include/linux/sunrpc/clnt.h b/include/linux/sunrpc/clnt.h index 2cf4ffaa3cd4..e7d492ce7c18 100644 --- a/include/linux/sunrpc/clnt.h +++ b/include/linux/sunrpc/clnt.h @@ -124,6 +124,7 @@ struct rpc_create_args { #define RPC_CLNT_CREATE_NOPING (1UL << 4) #define RPC_CLNT_CREATE_DISCRTRY (1UL << 5) #define RPC_CLNT_CREATE_QUIET (1UL << 6) +#define RPC_CLNT_CREATE_INFINITE_SLOTS (1UL << 7) struct rpc_clnt *rpc_create(struct rpc_create_args *args); struct rpc_clnt *rpc_bind_new_program(struct rpc_clnt *, diff --git a/include/linux/sunrpc/xprt.h b/include/linux/sunrpc/xprt.h index 12815f66a9c4..ff5392421cb2 100644 --- a/include/linux/sunrpc/xprt.h +++ b/include/linux/sunrpc/xprt.h @@ -255,6 +255,8 @@ static inline int bc_prealloc(struct rpc_rqst *req) } #endif /* CONFIG_SUNRPC_BACKCHANNEL */ +#define XPRT_CREATE_INFINITE_SLOTS (1U) + struct xprt_create { int ident; /* XPRT_TRANSPORT identifier */ struct net * net; @@ -263,6 +265,7 @@ struct xprt_create { size_t addrlen; const char *servername; struct svc_xprt *bc_xprt; /* NFSv4.1 backchannel */ + unsigned int flags; }; struct xprt_class { diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c index a80ee9b80dcf..651245aa829a 100644 --- a/net/sunrpc/clnt.c +++ b/net/sunrpc/clnt.c @@ -414,6 +414,8 @@ struct rpc_clnt *rpc_create(struct rpc_create_args *args) }; char servername[48]; + if (args->flags & RPC_CLNT_CREATE_INFINITE_SLOTS) + xprtargs.flags |= XPRT_CREATE_INFINITE_SLOTS; /* * If the caller chooses not to specify a hostname, whip * up a string representation of the passed-in address. diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c index 3081620cb02c..726e702b7a29 100644 --- a/net/sunrpc/xprtsock.c +++ b/net/sunrpc/xprtsock.c @@ -2762,9 +2762,13 @@ static struct rpc_xprt *xs_setup_tcp(struct xprt_create *args) struct rpc_xprt *xprt; struct sock_xprt *transport; struct rpc_xprt *ret; + unsigned int max_slot_table_size = xprt_max_tcp_slot_table_entries; + + if (args->flags & XPRT_CREATE_INFINITE_SLOTS) + max_slot_table_size = RPC_MAX_SLOT_TABLE_LIMIT; xprt = xs_setup_xprt(args, xprt_tcp_slot_table_entries, - xprt_max_tcp_slot_table_entries); + max_slot_table_size); if (IS_ERR(xprt)) return xprt; transport = container_of(xprt, struct sock_xprt, xprt); From 98f98cf571b7b1f34683455a61dae9f35e7222a1 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 14 Apr 2013 11:49:51 -0400 Subject: [PATCH 41/47] NFSv4.1: Set the RPC_CLNT_CREATE_INFINITE_SLOTS flag for NFSv4.1 transports This ensures that the RPC layer doesn't override the NFS session negotiation. Signed-off-by: Trond Myklebust --- fs/nfs/client.c | 2 ++ fs/nfs/nfs4client.c | 2 ++ include/linux/nfs_fs_sb.h | 1 + 3 files changed, 5 insertions(+) diff --git a/fs/nfs/client.c b/fs/nfs/client.c index 84d8eae203a7..c513b0cc835f 100644 --- a/fs/nfs/client.c +++ b/fs/nfs/client.c @@ -593,6 +593,8 @@ int nfs_create_rpc_client(struct nfs_client *clp, args.flags |= RPC_CLNT_CREATE_DISCRTRY; if (test_bit(NFS_CS_NORESVPORT, &clp->cl_flags)) args.flags |= RPC_CLNT_CREATE_NONPRIVPORT; + if (test_bit(NFS_CS_INFINITE_SLOTS, &clp->cl_flags)) + args.flags |= RPC_CLNT_CREATE_INFINITE_SLOTS; if (!IS_ERR(clp->cl_rpcclient)) return 0; diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c index 17b34b2da2df..f4d4d4ec6bf7 100644 --- a/fs/nfs/nfs4client.c +++ b/fs/nfs/nfs4client.c @@ -198,6 +198,8 @@ struct nfs_client *nfs4_init_client(struct nfs_client *clp, /* Check NFS protocol revision and initialize RPC op vector */ clp->rpc_ops = &nfs_v4_clientops; + if (clp->cl_minorversion != 0) + __set_bit(NFS_CS_INFINITE_SLOTS, &clp->cl_flags); __set_bit(NFS_CS_DISCRTRY, &clp->cl_flags); error = nfs_create_rpc_client(clp, timeparms, authflavour); if (error < 0) diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h index d8fdfdc7a8fe..3b7fa2abecca 100644 --- a/include/linux/nfs_fs_sb.h +++ b/include/linux/nfs_fs_sb.h @@ -40,6 +40,7 @@ struct nfs_client { #define NFS_CS_NORESVPORT 0 /* - use ephemeral src port */ #define NFS_CS_DISCRTRY 1 /* - disconnect on RPC retry */ #define NFS_CS_MIGRATION 2 /* - transparent state migr */ +#define NFS_CS_INFINITE_SLOTS 3 /* - don't limit TCP slots */ struct sockaddr_storage cl_addr; /* server identifier */ size_t cl_addrlen; char * cl_hostname; /* hostname of server */ From 549b19cc9f31e8fdda317625d564bac0052a3328 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Tue, 16 Apr 2013 18:42:34 -0400 Subject: [PATCH 42/47] NFSv4: Record the OPEN create mode used in the nfs4_opendata structure If we're doing NFSv4.1 against a server that has persistent sessions, then we should not need to call SETATTR in order to reset the file attributes immediately after doing an exclusive create. Note that since the create mode depends on the type of session that has been negotiated with the server, we should not choose the mode until after we've got a session slot. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 16 ++++++++++++++-- fs/nfs/nfs4xdr.c | 37 ++++++++++++++++--------------------- include/linux/nfs_xdr.h | 1 + 3 files changed, 31 insertions(+), 23 deletions(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index af05df3c7c79..282d9fa6994a 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -1525,6 +1525,7 @@ static void nfs4_open_prepare(struct rpc_task *task, void *calldata) { struct nfs4_opendata *data = calldata; struct nfs4_state_owner *sp = data->owner; + struct nfs_client *clp = sp->so_server->nfs_client; if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0) goto out_wait; @@ -1545,7 +1546,7 @@ static void nfs4_open_prepare(struct rpc_task *task, void *calldata) rcu_read_unlock(); } /* Update client id. */ - data->o_arg.clientid = sp->so_server->nfs_client->cl_clientid; + data->o_arg.clientid = clp->cl_clientid; if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS) { task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR]; data->o_arg.open_bitmap = &nfs4_open_noattr_bitmap[0]; @@ -1557,6 +1558,16 @@ static void nfs4_open_prepare(struct rpc_task *task, void *calldata) &data->o_res.seq_res, task) != 0) nfs_release_seqid(data->o_arg.seqid); + + /* Set the create mode (note dependency on the session type) */ + data->o_arg.createmode = NFS4_CREATE_UNCHECKED; + if (data->o_arg.open_flags & O_EXCL) { + data->o_arg.createmode = NFS4_CREATE_EXCLUSIVE; + if (nfs4_has_persistent_session(clp)) + data->o_arg.createmode = NFS4_CREATE_GUARDED; + else if (clp->cl_mvops->minor_version > 0) + data->o_arg.createmode = NFS4_CREATE_EXCLUSIVE4_1; + } return; unlock_no_action: rcu_read_unlock(); @@ -2000,7 +2011,8 @@ static int _nfs4_do_open(struct inode *dir, if (status != 0) goto err_opendata_put; - if (opendata->o_arg.open_flags & O_EXCL) { + if ((opendata->o_arg.open_flags & O_EXCL) && + (opendata->o_arg.createmode != NFS4_CREATE_GUARDED)) { nfs4_exclusive_attrset(opendata, sattr); nfs_fattr_init(opendata->o_res.f_attr); diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c index 0b744895b9e1..fef71cbec501 100644 --- a/fs/nfs/nfs4xdr.c +++ b/fs/nfs/nfs4xdr.c @@ -1366,33 +1366,28 @@ static inline void encode_openhdr(struct xdr_stream *xdr, const struct nfs_opena static inline void encode_createmode(struct xdr_stream *xdr, const struct nfs_openargs *arg) { + struct iattr dummy; __be32 *p; - struct nfs_client *clp; p = reserve_space(xdr, 4); - switch(arg->open_flags & O_EXCL) { - case 0: + switch(arg->createmode) { + case NFS4_CREATE_UNCHECKED: *p = cpu_to_be32(NFS4_CREATE_UNCHECKED); encode_attrs(xdr, arg->u.attrs, arg->server); break; - default: - clp = arg->server->nfs_client; - if (clp->cl_mvops->minor_version > 0) { - if (nfs4_has_persistent_session(clp)) { - *p = cpu_to_be32(NFS4_CREATE_GUARDED); - encode_attrs(xdr, arg->u.attrs, arg->server); - } else { - struct iattr dummy; - - *p = cpu_to_be32(NFS4_CREATE_EXCLUSIVE4_1); - encode_nfs4_verifier(xdr, &arg->u.verifier); - dummy.ia_valid = 0; - encode_attrs(xdr, &dummy, arg->server); - } - } else { - *p = cpu_to_be32(NFS4_CREATE_EXCLUSIVE); - encode_nfs4_verifier(xdr, &arg->u.verifier); - } + case NFS4_CREATE_GUARDED: + *p = cpu_to_be32(NFS4_CREATE_GUARDED); + encode_attrs(xdr, arg->u.attrs, arg->server); + break; + case NFS4_CREATE_EXCLUSIVE: + *p = cpu_to_be32(NFS4_CREATE_EXCLUSIVE); + encode_nfs4_verifier(xdr, &arg->u.verifier); + break; + case NFS4_CREATE_EXCLUSIVE4_1: + *p = cpu_to_be32(NFS4_CREATE_EXCLUSIVE4_1); + encode_nfs4_verifier(xdr, &arg->u.verifier); + dummy.ia_valid = 0; + encode_attrs(xdr, &dummy, arg->server); } } diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index 90a4aa190b43..bdc100f66dfb 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -349,6 +349,7 @@ struct nfs_openargs { const u32 * bitmask; const u32 * open_bitmap; __u32 claim; + enum createmode4 createmode; }; struct nfs_openres { From 042ad0b398ea4e937e38324dd096bdf9d2c495f7 Mon Sep 17 00:00:00 2001 From: Bryan Schumaker Date: Fri, 19 Apr 2013 16:09:37 -0400 Subject: [PATCH 43/47] nfs: Send atime and mtime as a 64bit value RFC 3530 says that the seconds value of a nfstime4 structure is a 64bit value, but we are instead sending a 32-bit 0 and then a 32bit conversion of the 64bit Linux value. This means that if we try to set atime to a value before the epoch (touch -t 196001010101) the client will only send part of the new value due to lost precision. Signed-off-by: Bryan Schumaker Signed-off-by: Trond Myklebust --- fs/nfs/nfs4xdr.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c index fef71cbec501..c2cbf0d90a31 100644 --- a/fs/nfs/nfs4xdr.c +++ b/fs/nfs/nfs4xdr.c @@ -1058,8 +1058,7 @@ static void encode_attrs(struct xdr_stream *xdr, const struct iattr *iap, const if (iap->ia_valid & ATTR_ATIME_SET) { bmval1 |= FATTR4_WORD1_TIME_ACCESS_SET; *p++ = cpu_to_be32(NFS4_SET_TO_CLIENT_TIME); - *p++ = cpu_to_be32(0); - *p++ = cpu_to_be32(iap->ia_atime.tv_sec); + p = xdr_encode_hyper(p, (s64)iap->ia_atime.tv_sec); *p++ = cpu_to_be32(iap->ia_atime.tv_nsec); } else if (iap->ia_valid & ATTR_ATIME) { @@ -1069,8 +1068,7 @@ static void encode_attrs(struct xdr_stream *xdr, const struct iattr *iap, const if (iap->ia_valid & ATTR_MTIME_SET) { bmval1 |= FATTR4_WORD1_TIME_MODIFY_SET; *p++ = cpu_to_be32(NFS4_SET_TO_CLIENT_TIME); - *p++ = cpu_to_be32(0); - *p++ = cpu_to_be32(iap->ia_mtime.tv_sec); + p = xdr_encode_hyper(p, (s64)iap->ia_mtime.tv_sec); *p++ = cpu_to_be32(iap->ia_mtime.tv_nsec); } else if (iap->ia_valid & ATTR_MTIME) { From 92b40e93849e29f9ca661de6442bb66282738bf7 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sat, 20 Apr 2013 01:25:45 -0400 Subject: [PATCH 44/47] NFSv4: Use the open stateid if the delegation has the wrong mode Fix nfs4_select_rw_stateid() so that it chooses the open stateid (or an all-zero stateid) if the delegation does not match the selected read/write mode. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4_fs.h | 1 + fs/nfs/nfs4proc.c | 1 + fs/nfs/nfs4state.c | 6 +++++- 3 files changed, 7 insertions(+), 1 deletion(-) diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h index 7ef19ce6d832..553a83cc4106 100644 --- a/fs/nfs/nfs4_fs.h +++ b/fs/nfs/nfs4_fs.h @@ -144,6 +144,7 @@ struct nfs4_lock_state { enum { LK_STATE_IN_USE, NFS_DELEGATED_STATE, /* Current stateid is delegation */ + NFS_OPEN_STATE, /* OPEN stateid is set */ NFS_O_RDONLY_STATE, /* OPEN stateid has read-only state */ NFS_O_WRONLY_STATE, /* OPEN stateid has write-only state */ NFS_O_RDWR_STATE, /* OPEN stateid has read/write state */ diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 282d9fa6994a..39849360f9f6 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -978,6 +978,7 @@ static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid * if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0) nfs4_stateid_copy(&state->stateid, stateid); nfs4_stateid_copy(&state->open_stateid, stateid); + set_bit(NFS_OPEN_STATE, &state->flags); switch (fmode) { case FMODE_READ: set_bit(NFS_O_RDONLY_STATE, &state->flags); diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c index 1eb17285c99a..b7796950eceb 100644 --- a/fs/nfs/nfs4state.c +++ b/fs/nfs/nfs4state.c @@ -1024,12 +1024,16 @@ static int nfs4_copy_lock_stateid(nfs4_stateid *dst, static int nfs4_copy_open_stateid(nfs4_stateid *dst, struct nfs4_state *state) { + const nfs4_stateid *src; int ret; int seq; do { + src = &zero_stateid; seq = read_seqbegin(&state->seqlock); - nfs4_stateid_copy(dst, &state->stateid); + if (test_bit(NFS_OPEN_STATE, &state->flags)) + src = &state->open_stateid; + nfs4_stateid_copy(dst, src); ret = 0; smp_rmb(); if (!list_empty(&state->owner->so_seqid.list)) From 8e472f33b5ba742f9bfcf16f0328f5a7842b32d1 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sat, 20 Apr 2013 01:30:53 -0400 Subject: [PATCH 45/47] NFSv4: Ensure the LOCK call cannot use the delegation stateid Defensive patch to ensure that we copy the state->open_stateid, which can never be set to the delegation stateid. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 39849360f9f6..ed01a66dc391 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -4793,7 +4793,7 @@ static void nfs4_lock_prepare(struct rpc_task *task, void *calldata) if (nfs_wait_on_sequence(data->arg.open_seqid, task) != 0) { goto out_release_lock_seqid; } - data->arg.open_stateid = &state->stateid; + data->arg.open_stateid = &state->open_stateid; data->arg.new_lock_owner = 1; data->res.open_seqid = data->arg.open_seqid; } else From 1dfd89af8697a299e7982ae740d4695ecd917eef Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Sun, 21 Apr 2013 18:01:06 -0400 Subject: [PATCH 46/47] LOCKD: Ensure that nlmclnt_block resets block->b_status after a server reboot After a server reboot, the reclaimer thread will recover all the existing locks. For locks that are blocked, however, it will change the value of block->b_status to nlm_lck_denied_grace_period in order to signal that they need to wake up and resend the original blocking lock request. Due to a bug, however, the block->b_status never gets reset after the blocked locks have been woken up, and so the process goes into an infinite loop of resends until the blocked lock is satisfied. Reported-by: Marc Eshel Signed-off-by: Trond Myklebust Cc: stable@vger.kernel.org --- fs/lockd/clntlock.c | 3 +++ fs/lockd/clntproc.c | 3 --- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/fs/lockd/clntlock.c b/fs/lockd/clntlock.c index 0796c45d0d4d..01bfe7662751 100644 --- a/fs/lockd/clntlock.c +++ b/fs/lockd/clntlock.c @@ -144,6 +144,9 @@ int nlmclnt_block(struct nlm_wait *block, struct nlm_rqst *req, long timeout) timeout); if (ret < 0) return -ERESTARTSYS; + /* Reset the lock status after a server reboot so we resend */ + if (block->b_status == nlm_lck_denied_grace_period) + block->b_status = nlm_lck_blocked; req->a_res.status = block->b_status; return 0; } diff --git a/fs/lockd/clntproc.c b/fs/lockd/clntproc.c index 7e529c3c45c0..9760ecb9b60f 100644 --- a/fs/lockd/clntproc.c +++ b/fs/lockd/clntproc.c @@ -550,9 +550,6 @@ nlmclnt_lock(struct nlm_rqst *req, struct file_lock *fl) status = nlmclnt_block(block, req, NLMCLNT_POLL_TIMEOUT); if (status < 0) break; - /* Resend the blocking lock request after a server reboot */ - if (resp->status == nlm_lck_denied_grace_period) - continue; if (resp->status != nlm_lck_blocked) break; } From fd068b200fb86e8fa52368e6f6088d2ab297a5d7 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 22 Apr 2013 11:29:51 -0400 Subject: [PATCH 47/47] NFSv4: Ensure that we clear the NFS_OPEN_STATE flag when appropriate We should always clear it before initiating file recovery. Also ensure that we clear it after a CLOSE and/or after TEST_STATEID fails. Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 18 ++++++++++++++---- 1 file changed, 14 insertions(+), 4 deletions(-) diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index ed01a66dc391..c13144911d20 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -1290,6 +1290,7 @@ static int nfs4_open_recover(struct nfs4_opendata *opendata, struct nfs4_state * /* memory barrier prior to reading state->n_* */ clear_bit(NFS_DELEGATED_STATE, &state->flags); + clear_bit(NFS_OPEN_STATE, &state->flags); smp_rmb(); if (state->n_rdwr != 0) { clear_bit(NFS_O_RDWR_STATE, &state->flags); @@ -1893,6 +1894,7 @@ static int nfs41_check_open_stateid(struct nfs4_state *state) clear_bit(NFS_O_RDONLY_STATE, &state->flags); clear_bit(NFS_O_WRONLY_STATE, &state->flags); clear_bit(NFS_O_RDWR_STATE, &state->flags); + clear_bit(NFS_OPEN_STATE, &state->flags); } return status; } @@ -2208,11 +2210,19 @@ static void nfs4_close_clear_stateid_flags(struct nfs4_state *state, fmode_t fmode) { spin_lock(&state->owner->so_lock); - if (!(fmode & FMODE_READ)) - clear_bit(NFS_O_RDONLY_STATE, &state->flags); - if (!(fmode & FMODE_WRITE)) - clear_bit(NFS_O_WRONLY_STATE, &state->flags); clear_bit(NFS_O_RDWR_STATE, &state->flags); + switch (fmode & (FMODE_READ|FMODE_WRITE)) { + case FMODE_WRITE: + clear_bit(NFS_O_RDONLY_STATE, &state->flags); + break; + case FMODE_READ: + clear_bit(NFS_O_WRONLY_STATE, &state->flags); + break; + case 0: + clear_bit(NFS_O_RDONLY_STATE, &state->flags); + clear_bit(NFS_O_WRONLY_STATE, &state->flags); + clear_bit(NFS_OPEN_STATE, &state->flags); + } spin_unlock(&state->owner->so_lock); }