[PATCH v2 01/10] NFS: Remove unnecessary includes from linux/nfs_fs_i.h

[PATCH v2 01/10] NFS: Remove unnecessary includes from linux/nfs_fs_i.h

[Trond Myklebust]

Also from linux/nfs_xdr.h.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/pagelist.c        |    1 +
 include/linux/nfs_fs_i.h |    4 ----
 include/linux/nfs_xdr.h  |    1 -
 3 files changed, 1 insertions(+), 5 deletions(-)

diff --git a/fs/nfs/pagelist.c b/fs/nfs/pagelist.c
index 5668f7c..77a184e 100644
--- a/fs/nfs/pagelist.c
+++ b/fs/nfs/pagelist.c
@@ -13,6 +13,7 @@
 #include <linux/file.h>
 #include <linux/sched.h>
 #include <linux/sunrpc/clnt.h>
+#include <linux/nfs.h>
 #include <linux/nfs3.h>
 #include <linux/nfs4.h>
 #include <linux/nfs_page.h>
diff --git a/include/linux/nfs_fs_i.h b/include/linux/nfs_fs_i.h
index 8617302..a5c50d9 100644
--- a/include/linux/nfs_fs_i.h
+++ b/include/linux/nfs_fs_i.h
@@ -1,10 +1,6 @@
 #ifndef _NFS_FS_I
 #define _NFS_FS_I
 
-#include <asm/types.h>
-#include <linux/list.h>
-#include <linux/nfs.h>
-
 struct nlm_lockowner;
 
 /*
diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
index a764cef..f5188e1 100644
--- a/include/linux/nfs_xdr.h
+++ b/include/linux/nfs_xdr.h
@@ -2,7 +2,6 @@
 #define _LINUX_NFS_XDR_H
 
 #include <linux/nfsacl.h>
-#include <linux/nfs3.h>
 #include <linux/sunrpc/gss_api.h>
 
 /*
-- 
1.7.7.5

[PATCH v2 02/10] NFSv4: Clean up nfs4_get_state_owner

[Trond Myklebust]

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/nfs4_fs.h   |    2 +-
 fs/nfs/nfs4proc.c  |    2 +-
 fs/nfs/nfs4state.c |   24 +++++++++++-------------
 3 files changed, 13 insertions(+), 15 deletions(-)

diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index 4d7d0ae..6d439c2 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -319,7 +319,7 @@ static inline void nfs4_schedule_session_recovery(struct nfs4_session *session)
 }
 #endif /* CONFIG_NFS_V4_1 */
 
-extern struct nfs4_state_owner * nfs4_get_state_owner(struct nfs_server *, struct rpc_cred *);
+extern struct nfs4_state_owner * nfs4_get_state_owner(struct nfs_server *, struct rpc_cred *, gfp_t);
 extern void nfs4_put_state_owner(struct nfs4_state_owner *);
 extern void nfs4_purge_state_owners(struct nfs_server *);
 extern struct nfs4_state * nfs4_get_open_state(struct inode *, struct nfs4_state_owner *);
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index f0c849c..4072754 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -1754,7 +1754,7 @@ static int _nfs4_do_open(struct inode *dir, struct dentry *dentry, fmode_t fmode
 
 	/* Protect against reboot recovery conflicts */
 	status = -ENOMEM;
-	if (!(sp = nfs4_get_state_owner(server, cred))) {
+	if (!(sp = nfs4_get_state_owner(server, cred, GFP_KERNEL))) {
 		dprintk("nfs4_do_open: nfs4_get_state_owner failed!\n");
 		goto out_err;
 	}
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index a53f33b..a8a42a6 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -444,13 +444,17 @@ nfs4_remove_state_owner_locked(struct nfs4_state_owner *sp)
  *
  */
 static struct nfs4_state_owner *
-nfs4_alloc_state_owner(void)
+nfs4_alloc_state_owner(struct nfs_server *server,
+		struct rpc_cred *cred,
+		gfp_t gfp_flags)
 {
 	struct nfs4_state_owner *sp;
 
-	sp = kzalloc(sizeof(*sp),GFP_NOFS);
+	sp = kzalloc(sizeof(*sp), gfp_flags);
 	if (!sp)
 		return NULL;
+	sp->so_server = server;
+	sp->so_cred = get_rpccred(cred);
 	spin_lock_init(&sp->so_lock);
 	INIT_LIST_HEAD(&sp->so_states);
 	rpc_init_wait_queue(&sp->so_sequence.wait, "Seqid_waitqueue");
@@ -516,7 +520,8 @@ static void nfs4_gc_state_owners(struct nfs_server *server)
  * Returns a pointer to an instantiated nfs4_state_owner struct, or NULL.
  */
 struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *server,
-					      struct rpc_cred *cred)
+					      struct rpc_cred *cred,
+					      gfp_t gfp_flags)
 {
 	struct nfs_client *clp = server->nfs_client;
 	struct nfs4_state_owner *sp, *new;
@@ -526,20 +531,13 @@ struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *server,
 	spin_unlock(&clp->cl_lock);
 	if (sp != NULL)
 		goto out;
-	new = nfs4_alloc_state_owner();
+	new = nfs4_alloc_state_owner(server, cred, gfp_flags);
 	if (new == NULL)
 		goto out;
-	new->so_server = server;
-	new->so_cred = cred;
-	spin_lock(&clp->cl_lock);
 	sp = nfs4_insert_state_owner_locked(new);
 	spin_unlock(&clp->cl_lock);
-	if (sp == new)
-		get_rpccred(cred);
-	else {
-		rpc_destroy_wait_queue(&new->so_sequence.wait);
-		kfree(new);
-	}
+	if (sp != new)
+		nfs4_free_state_owner(new);
 out:
 	nfs4_gc_state_owners(server);
 	return sp;
-- 
1.7.7.5

[PATCH v2 03/10] NFSv4: Replace state_owner->so_owner_id with an ida based allocator

[Trond Myklebust]

We're unlikely to ever need more than 2^31 simultaneous open owners,
so let's replace the custom allocator with the generic ida allocator.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/client.c           |    2 ++
 fs/nfs/nfs4_fs.h          |    2 +-
 fs/nfs/nfs4proc.c         |    4 ++--
 fs/nfs/nfs4state.c        |   17 ++++++++++++-----
 include/linux/nfs_fs_sb.h |    3 ++-
 5 files changed, 19 insertions(+), 9 deletions(-)

diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index ca9a4aa..8d1739d 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -1092,6 +1092,7 @@ static struct nfs_server *nfs_alloc_server(void)
 		return NULL;
 	}
 
+	ida_init(&server->openowner_id);
 	pnfs_init_server(server);
 
 	return server;
@@ -1117,6 +1118,7 @@ void nfs_free_server(struct nfs_server *server)
 
 	nfs_put_client(server->nfs_client);
 
+	ida_destroy(&server->openowner_id);
 	nfs_free_iostats(server->io_stats);
 	bdi_destroy(&server->backing_dev_info);
 	kfree(server);
diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index 6d439c2..a754f47 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -92,7 +92,6 @@ struct nfs_unique_id {
  * semantics by allowing the server to identify replayed requests.
  */
 struct nfs4_state_owner {
-	struct nfs_unique_id so_owner_id;
 	struct nfs_server    *so_server;
 	struct list_head     so_lru;
 	unsigned long        so_expires;
@@ -106,6 +105,7 @@ struct nfs4_state_owner {
 	struct list_head     so_states;
 	struct nfs_seqid_counter so_seqid;
 	struct rpc_sequence  so_sequence;
+	int		     so_owner_id;
 };
 
 enum {
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 4072754..1ddedf0 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -815,7 +815,7 @@ static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
 	p->o_arg.open_flags = flags;
 	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
 	p->o_arg.clientid = server->nfs_client->cl_clientid;
-	p->o_arg.id = sp->so_owner_id.id;
+	p->o_arg.id = sp->so_owner_id;
 	p->o_arg.name = &dentry->d_name;
 	p->o_arg.server = server;
 	p->o_arg.bitmask = server->attr_bitmask;
@@ -1440,7 +1440,7 @@ static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
 		rcu_read_unlock();
 	}
 	/* Update sequence id. */
-	data->o_arg.id = sp->so_owner_id.id;
+	data->o_arg.id = sp->so_owner_id;
 	data->o_arg.clientid = sp->so_server->nfs_client->cl_clientid;
 	if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS) {
 		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index a8a42a6..8472707 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -405,6 +405,7 @@ nfs4_insert_state_owner_locked(struct nfs4_state_owner *new)
 	struct rb_node **p = &server->state_owners.rb_node,
 		       *parent = NULL;
 	struct nfs4_state_owner *sp;
+	int err;
 
 	while (*p != NULL) {
 		parent = *p;
@@ -421,8 +422,9 @@ nfs4_insert_state_owner_locked(struct nfs4_state_owner *new)
 			return sp;
 		}
 	}
-	nfs_alloc_unique_id_locked(&server->openowner_id,
-					&new->so_owner_id, 1, 64);
+	err = ida_get_new(&server->openowner_id, &new->so_owner_id);
+	if (err)
+		return ERR_PTR(err);
 	rb_link_node(&new->so_server_node, parent, p);
 	rb_insert_color(&new->so_server_node, &server->state_owners);
 	return new;
@@ -435,7 +437,7 @@ nfs4_remove_state_owner_locked(struct nfs4_state_owner *sp)
 
 	if (!RB_EMPTY_NODE(&sp->so_server_node))
 		rb_erase(&sp->so_server_node, &server->state_owners);
-	nfs_free_unique_id(&server->openowner_id, &sp->so_owner_id);
+	ida_remove(&server->openowner_id, sp->so_owner_id);
 }
 
 /*
@@ -534,8 +536,13 @@ struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *server,
 	new = nfs4_alloc_state_owner(server, cred, gfp_flags);
 	if (new == NULL)
 		goto out;
-	sp = nfs4_insert_state_owner_locked(new);
-	spin_unlock(&clp->cl_lock);
+	do {
+		if (ida_pre_get(&server->openowner_id, gfp_flags) == 0)
+			break;
+		spin_lock(&clp->cl_lock);
+		sp = nfs4_insert_state_owner_locked(new);
+		spin_unlock(&clp->cl_lock);
+	} while (sp == ERR_PTR(-EAGAIN));
 	if (sp != new)
 		nfs4_free_state_owner(new);
 out:
diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
index e0863d3..645537e 100644
--- a/include/linux/nfs_fs_sb.h
+++ b/include/linux/nfs_fs_sb.h
@@ -3,6 +3,7 @@
 
 #include <linux/list.h>
 #include <linux/backing-dev.h>
+#include <linux/idr.h>
 #include <linux/wait.h>
 #include <linux/nfs_xdr.h>
 #include <linux/sunrpc/xprt.h>
@@ -151,9 +152,9 @@ struct nfs_server {
 
 	/* the following fields are protected by nfs_client->cl_lock */
 	struct rb_root		state_owners;
-	struct rb_root		openowner_id;
 	struct rb_root		lockowner_id;
 #endif
+	struct ida		openowner_id;
 	struct list_head	state_owners_lru;
 	struct list_head	layouts;
 	struct list_head	delegations;
-- 
1.7.7.5

[PATCH v2 04/10] NFSv4: Replace lock_owner->ld_id with an ida based allocator

[Trond Myklebust]

Again, We're unlikely to ever need more than 2^31 simultaneous lock
owners, so let's replace the custom allocator.

Now that there are no more users, we can also get rid of the custom
allocator code.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/client.c           |    2 +
 fs/nfs/nfs4_fs.h          |    7 +----
 fs/nfs/nfs4proc.c         |    6 ++--
 fs/nfs/nfs4state.c        |   74 +++++----------------------------------------
 include/linux/nfs_fs_sb.h |    2 +-
 5 files changed, 15 insertions(+), 76 deletions(-)

diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index 8d1739d..df60d99 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -1093,6 +1093,7 @@ static struct nfs_server *nfs_alloc_server(void)
 	}
 
 	ida_init(&server->openowner_id);
+	ida_init(&server->lockowner_id);
 	pnfs_init_server(server);
 
 	return server;
@@ -1118,6 +1119,7 @@ void nfs_free_server(struct nfs_server *server)
 
 	nfs_put_client(server->nfs_client);
 
+	ida_destroy(&server->lockowner_id);
 	ida_destroy(&server->openowner_id);
 	nfs_free_iostats(server->io_stats);
 	bdi_destroy(&server->backing_dev_info);
diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index a754f47..2d1f5a5 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -81,11 +81,6 @@ static inline void nfs_confirm_seqid(struct nfs_seqid_counter *seqid, int status
 		seqid->flags |= NFS_SEQID_CONFIRMED;
 }
 
-struct nfs_unique_id {
-	struct rb_node rb_node;
-	__u64 id;
-};
-
 /*
  * NFS4 state_owners and lock_owners are simply labels for ordered
  * sequences of RPC calls. Their sole purpose is to provide once-only
@@ -145,9 +140,9 @@ struct nfs4_lock_state {
 	struct nfs4_state *	ls_state;	/* Pointer to open state */
 #define NFS_LOCK_INITIALIZED 1
 	int			ls_flags;
+	int			ls_id;
 	struct nfs_seqid_counter	ls_seqid;
 	struct rpc_sequence	ls_sequence;
-	struct nfs_unique_id	ls_id;
 	nfs4_stateid		ls_stateid;
 	atomic_t		ls_count;
 	struct nfs4_lock_owner	ls_owner;
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 1ddedf0..7ccc12d 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -4016,7 +4016,7 @@ static int _nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock
 	if (status != 0)
 		goto out;
 	lsp = request->fl_u.nfs4_fl.owner;
-	arg.lock_owner.id = lsp->ls_id.id;
+	arg.lock_owner.id = lsp->ls_id;
 	arg.lock_owner.s_dev = server->s_dev;
 	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
 	switch (status) {
@@ -4261,7 +4261,7 @@ static struct nfs4_lockdata *nfs4_alloc_lockdata(struct file_lock *fl,
 		goto out_free_seqid;
 	p->arg.lock_stateid = &lsp->ls_stateid;
 	p->arg.lock_owner.clientid = server->nfs_client->cl_clientid;
-	p->arg.lock_owner.id = lsp->ls_id.id;
+	p->arg.lock_owner.id = lsp->ls_id;
 	p->arg.lock_owner.s_dev = server->s_dev;
 	p->res.lock_seqid = p->arg.lock_seqid;
 	p->lsp = lsp;
@@ -4678,7 +4678,7 @@ void nfs4_release_lockowner(const struct nfs4_lock_state *lsp)
 	if (!args)
 		return;
 	args->lock_owner.clientid = server->nfs_client->cl_clientid;
-	args->lock_owner.id = lsp->ls_id.id;
+	args->lock_owner.id = lsp->ls_id;
 	args->lock_owner.s_dev = server->s_dev;
 	msg.rpc_argp = args;
 	rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, args);
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index 8472707..5abf236 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -317,62 +317,6 @@ out:
 	return cred;
 }
 
-static void nfs_alloc_unique_id_locked(struct rb_root *root,
-				       struct nfs_unique_id *new,
-				       __u64 minval, int maxbits)
-{
-	struct rb_node **p, *parent;
-	struct nfs_unique_id *pos;
-	__u64 mask = ~0ULL;
-
-	if (maxbits < 64)
-		mask = (1ULL << maxbits) - 1ULL;
-
-	/* Ensure distribution is more or less flat */
-	get_random_bytes(&new->id, sizeof(new->id));
-	new->id &= mask;
-	if (new->id < minval)
-		new->id += minval;
-retry:
-	p = &root->rb_node;
-	parent = NULL;
-
-	while (*p != NULL) {
-		parent = *p;
-		pos = rb_entry(parent, struct nfs_unique_id, rb_node);
-
-		if (new->id < pos->id)
-			p = &(*p)->rb_left;
-		else if (new->id > pos->id)
-			p = &(*p)->rb_right;
-		else
-			goto id_exists;
-	}
-	rb_link_node(&new->rb_node, parent, p);
-	rb_insert_color(&new->rb_node, root);
-	return;
-id_exists:
-	for (;;) {
-		new->id++;
-		if (new->id < minval || (new->id & mask) != new->id) {
-			new->id = minval;
-			break;
-		}
-		parent = rb_next(parent);
-		if (parent == NULL)
-			break;
-		pos = rb_entry(parent, struct nfs_unique_id, rb_node);
-		if (new->id < pos->id)
-			break;
-	}
-	goto retry;
-}
-
-static void nfs_free_unique_id(struct rb_root *root, struct nfs_unique_id *id)
-{
-	rb_erase(&id->rb_node, root);
-}
-
 static struct nfs4_state_owner *
 nfs4_find_state_owner_locked(struct nfs_server *server, struct rpc_cred *cred)
 {
@@ -800,7 +744,6 @@ static struct nfs4_lock_state *nfs4_alloc_lock_state(struct nfs4_state *state, f
 {
 	struct nfs4_lock_state *lsp;
 	struct nfs_server *server = state->owner->so_server;
-	struct nfs_client *clp = server->nfs_client;
 
 	lsp = kzalloc(sizeof(*lsp), GFP_NOFS);
 	if (lsp == NULL)
@@ -820,24 +763,23 @@ static struct nfs4_lock_state *nfs4_alloc_lock_state(struct nfs4_state *state, f
 		lsp->ls_owner.lo_u.posix_owner = fl_owner;
 		break;
 	default:
-		kfree(lsp);
-		return NULL;
+		goto out_free;
 	}
-	spin_lock(&clp->cl_lock);
-	nfs_alloc_unique_id_locked(&server->lockowner_id, &lsp->ls_id, 1, 64);
-	spin_unlock(&clp->cl_lock);
+	lsp->ls_id = ida_simple_get(&server->lockowner_id, 0, 0, GFP_NOFS);
+	if (lsp->ls_id < 0)
+		goto out_free;
 	INIT_LIST_HEAD(&lsp->ls_locks);
 	return lsp;
+out_free:
+	kfree(lsp);
+	return NULL;
 }
 
 static void nfs4_free_lock_state(struct nfs4_lock_state *lsp)
 {
 	struct nfs_server *server = lsp->ls_state->owner->so_server;
-	struct nfs_client *clp = server->nfs_client;
 
-	spin_lock(&clp->cl_lock);
-	nfs_free_unique_id(&server->lockowner_id, &lsp->ls_id);
-	spin_unlock(&clp->cl_lock);
+	ida_simple_remove(&server->lockowner_id, lsp->ls_id);
 	rpc_destroy_wait_queue(&lsp->ls_sequence.wait);
 	kfree(lsp);
 }
diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
index 645537e..c234693 100644
--- a/include/linux/nfs_fs_sb.h
+++ b/include/linux/nfs_fs_sb.h
@@ -152,9 +152,9 @@ struct nfs_server {
 
 	/* the following fields are protected by nfs_client->cl_lock */
 	struct rb_root		state_owners;
-	struct rb_root		lockowner_id;
 #endif
 	struct ida		openowner_id;
+	struct ida		lockowner_id;
 	struct list_head	state_owners_lru;
 	struct list_head	layouts;
 	struct list_head	delegations;
-- 
1.7.7.5

[PATCH v2 05/10] NFSv41: Add a new helper nfs4_init_sequence()

[Trond Myklebust]

Clean up

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/nfs4_fs.h        |    6 ++--
 fs/nfs/nfs4filelayout.c |    4 +-
 fs/nfs/nfs4proc.c       |   86 +++++++++++++++++++++++++++++------------------
 fs/nfs/read.c           |    2 +-
 fs/nfs/unlink.c         |    4 +-
 fs/nfs/write.c          |    2 +-
 6 files changed, 62 insertions(+), 42 deletions(-)

diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index 2d1f5a5..583ded0 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -230,10 +230,10 @@ static inline struct nfs4_session *nfs4_get_session(const struct nfs_server *ser
 
 extern int nfs4_setup_sequence(const struct nfs_server *server,
 		struct nfs4_sequence_args *args, struct nfs4_sequence_res *res,
-		int cache_reply, struct rpc_task *task);
+		struct rpc_task *task);
 extern int nfs41_setup_sequence(struct nfs4_session *session,
 		struct nfs4_sequence_args *args, struct nfs4_sequence_res *res,
-		int cache_reply, struct rpc_task *task);
+		struct rpc_task *task);
 extern void nfs4_destroy_session(struct nfs4_session *session);
 extern struct nfs4_session *nfs4_alloc_session(struct nfs_client *clp);
 extern int nfs4_proc_create_session(struct nfs_client *);
@@ -264,7 +264,7 @@ static inline struct nfs4_session *nfs4_get_session(const struct nfs_server *ser
 
 static inline int nfs4_setup_sequence(const struct nfs_server *server,
 		struct nfs4_sequence_args *args, struct nfs4_sequence_res *res,
-		int cache_reply, struct rpc_task *task)
+		struct rpc_task *task)
 {
 	return 0;
 }
diff --git a/fs/nfs/nfs4filelayout.c b/fs/nfs/nfs4filelayout.c
index 71ec086..b4f8f96 100644
--- a/fs/nfs/nfs4filelayout.c
+++ b/fs/nfs/nfs4filelayout.c
@@ -173,7 +173,7 @@ static void filelayout_read_prepare(struct rpc_task *task, void *data)
 
 	if (nfs41_setup_sequence(rdata->ds_clp->cl_session,
 				&rdata->args.seq_args, &rdata->res.seq_res,
-				0, task))
+				task))
 		return;
 
 	rpc_call_start(task);
@@ -254,7 +254,7 @@ static void filelayout_write_prepare(struct rpc_task *task, void *data)
 
 	if (nfs41_setup_sequence(wdata->ds_clp->cl_session,
 				&wdata->args.seq_args, &wdata->res.seq_res,
-				0, task))
+				task))
 		return;
 
 	rpc_call_start(task);
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 7ccc12d..3233ee0 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -534,10 +534,19 @@ out:
 	return ret_id;
 }
 
+static void nfs41_init_sequence(struct nfs4_sequence_args *args,
+		struct nfs4_sequence_res *res, int cache_reply)
+{
+	args->sa_session = NULL;
+	args->sa_cache_this = 0;
+	if (cache_reply)
+		args->sa_cache_this = 1;
+	res->sr_slot = NULL;
+}
+
 int nfs41_setup_sequence(struct nfs4_session *session,
 				struct nfs4_sequence_args *args,
 				struct nfs4_sequence_res *res,
-				int cache_reply,
 				struct rpc_task *task)
 {
 	struct nfs4_slot *slot;
@@ -582,7 +591,6 @@ int nfs41_setup_sequence(struct nfs4_session *session,
 	slot = tbl->slots + slotid;
 	args->sa_session = session;
 	args->sa_slotid = slotid;
-	args->sa_cache_this = cache_reply;
 
 	dprintk("<-- %s slotid=%d seqid=%d\n", __func__, slotid, slot->seq_nr);
 
@@ -602,7 +610,6 @@ EXPORT_SYMBOL_GPL(nfs41_setup_sequence);
 int nfs4_setup_sequence(const struct nfs_server *server,
 			struct nfs4_sequence_args *args,
 			struct nfs4_sequence_res *res,
-			int cache_reply,
 			struct rpc_task *task)
 {
 	struct nfs4_session *session = nfs4_get_session(server);
@@ -618,8 +625,7 @@ int nfs4_setup_sequence(const struct nfs_server *server,
 		__func__, session->clp, session, res->sr_slot ?
 			res->sr_slot - session->fc_slot_table.slots : -1);
 
-	ret = nfs41_setup_sequence(session, args, res, cache_reply,
-				   task);
+	ret = nfs41_setup_sequence(session, args, res, task);
 out:
 	dprintk("<-- %s status=%d\n", __func__, ret);
 	return ret;
@@ -629,7 +635,6 @@ struct nfs41_call_sync_data {
 	const struct nfs_server *seq_server;
 	struct nfs4_sequence_args *seq_args;
 	struct nfs4_sequence_res *seq_res;
-	int cache_reply;
 };
 
 static void nfs41_call_sync_prepare(struct rpc_task *task, void *calldata)
@@ -639,7 +644,7 @@ static void nfs41_call_sync_prepare(struct rpc_task *task, void *calldata)
 	dprintk("--> %s data->seq_server %p\n", __func__, data->seq_server);
 
 	if (nfs4_setup_sequence(data->seq_server, data->seq_args,
-				data->seq_res, data->cache_reply, task))
+				data->seq_res, task))
 		return;
 	rpc_call_start(task);
 }
@@ -672,7 +677,6 @@ static int nfs4_call_sync_sequence(struct rpc_clnt *clnt,
 				   struct rpc_message *msg,
 				   struct nfs4_sequence_args *args,
 				   struct nfs4_sequence_res *res,
-				   int cache_reply,
 				   int privileged)
 {
 	int ret;
@@ -681,7 +685,6 @@ static int nfs4_call_sync_sequence(struct rpc_clnt *clnt,
 		.seq_server = server,
 		.seq_args = args,
 		.seq_res = res,
-		.cache_reply = cache_reply,
 	};
 	struct rpc_task_setup task_setup = {
 		.rpc_client = clnt,
@@ -690,7 +693,6 @@ static int nfs4_call_sync_sequence(struct rpc_clnt *clnt,
 		.callback_data = &data
 	};
 
-	res->sr_slot = NULL;
 	if (privileged)
 		task_setup.callback_ops = &nfs41_call_priv_sync_ops;
 	task = rpc_run_task(&task_setup);
@@ -710,10 +712,17 @@ int _nfs4_call_sync_session(struct rpc_clnt *clnt,
 			    struct nfs4_sequence_res *res,
 			    int cache_reply)
 {
-	return nfs4_call_sync_sequence(clnt, server, msg, args, res, cache_reply, 0);
+	nfs41_init_sequence(args, res, cache_reply);
+	return nfs4_call_sync_sequence(clnt, server, msg, args, res, 0);
 }
 
 #else
+static inline
+void nfs41_init_sequence(struct nfs4_sequence_args *args,
+		struct nfs4_sequence_res *res, int cache_reply)
+{
+}
+
 static int nfs4_sequence_done(struct rpc_task *task,
 			       struct nfs4_sequence_res *res)
 {
@@ -728,7 +737,7 @@ int _nfs4_call_sync(struct rpc_clnt *clnt,
 		    struct nfs4_sequence_res *res,
 		    int cache_reply)
 {
-	args->sa_session = res->sr_session = NULL;
+	nfs41_init_sequence(args, res, cache_reply);
 	return rpc_call_sync(clnt, msg, 0);
 }
 
@@ -1449,7 +1458,7 @@ static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
 	data->timestamp = jiffies;
 	if (nfs4_setup_sequence(data->o_arg.server,
 				&data->o_arg.seq_args,
-				&data->o_res.seq_res, 1, task))
+				&data->o_res.seq_res, task))
 		return;
 	rpc_call_start(task);
 	return;
@@ -1551,6 +1560,7 @@ static int nfs4_run_open_task(struct nfs4_opendata *data, int isrecover)
 	};
 	int status;
 
+	nfs41_init_sequence(&o_arg->seq_args, &o_res->seq_res, 1);
 	kref_get(&data->kref);
 	data->rpc_done = 0;
 	data->rpc_status = 0;
@@ -2029,8 +2039,9 @@ static void nfs4_close_prepare(struct rpc_task *task, void *data)
 	nfs_fattr_init(calldata->res.fattr);
 	calldata->timestamp = jiffies;
 	if (nfs4_setup_sequence(NFS_SERVER(calldata->inode),
-				&calldata->arg.seq_args, &calldata->res.seq_res,
-				1, task))
+				&calldata->arg.seq_args,
+				&calldata->res.seq_res,
+				task))
 		return;
 	rpc_call_start(task);
 }
@@ -2074,6 +2085,7 @@ int nfs4_do_close(struct nfs4_state *state, gfp_t gfp_mask, int wait, bool roc)
 	calldata = kzalloc(sizeof(*calldata), gfp_mask);
 	if (calldata == NULL)
 		goto out;
+	nfs41_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 1);
 	calldata->inode = state->inode;
 	calldata->state = state;
 	calldata->arg.fh = NFS_FH(state->inode);
@@ -2712,8 +2724,8 @@ static void nfs4_proc_unlink_setup(struct rpc_message *msg, struct inode *dir)
 
 	args->bitmask = server->cache_consistency_bitmask;
 	res->server = server;
-	res->seq_res.sr_slot = NULL;
 	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE];
+	nfs41_init_sequence(&args->seq_args, &res->seq_res, 1);
 }
 
 static int nfs4_proc_unlink_done(struct rpc_task *task, struct inode *dir)
@@ -2738,6 +2750,7 @@ static void nfs4_proc_rename_setup(struct rpc_message *msg, struct inode *dir)
 	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RENAME];
 	arg->bitmask = server->attr_bitmask;
 	res->server = server;
+	nfs41_init_sequence(&arg->seq_args, &res->seq_res, 1);
 }
 
 static int nfs4_proc_rename_done(struct rpc_task *task, struct inode *old_dir,
@@ -3232,6 +3245,7 @@ static void nfs4_proc_read_setup(struct nfs_read_data *data, struct rpc_message
 	data->timestamp   = jiffies;
 	data->read_done_cb = nfs4_read_done_cb;
 	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ];
+	nfs41_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
 }
 
 /* Reset the the nfs_read_data to send the read to the MDS. */
@@ -3305,6 +3319,7 @@ static void nfs4_proc_write_setup(struct nfs_write_data *data, struct rpc_messag
 	data->timestamp   = jiffies;
 
 	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE];
+	nfs41_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
 }
 
 static int nfs4_commit_done_cb(struct rpc_task *task, struct nfs_write_data *data)
@@ -3891,7 +3906,7 @@ static void nfs4_delegreturn_prepare(struct rpc_task *task, void *data)
 
 	if (nfs4_setup_sequence(d_data->res.server,
 				&d_data->args.seq_args,
-				&d_data->res.seq_res, 1, task))
+				&d_data->res.seq_res, task))
 		return;
 	rpc_call_start(task);
 }
@@ -3925,6 +3940,7 @@ static int _nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, co
 	data = kzalloc(sizeof(*data), GFP_NOFS);
 	if (data == NULL)
 		return -ENOMEM;
+	nfs41_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
 	data->args.fhandle = &data->fh;
 	data->args.stateid = &data->stateid;
 	data->args.bitmask = server->attr_bitmask;
@@ -4142,7 +4158,7 @@ static void nfs4_locku_prepare(struct rpc_task *task, void *data)
 	calldata->timestamp = jiffies;
 	if (nfs4_setup_sequence(calldata->server,
 				&calldata->arg.seq_args,
-				&calldata->res.seq_res, 1, task))
+				&calldata->res.seq_res, task))
 		return;
 	rpc_call_start(task);
 }
@@ -4182,6 +4198,7 @@ static struct rpc_task *nfs4_do_unlck(struct file_lock *fl,
 		return ERR_PTR(-ENOMEM);
 	}
 
+	nfs41_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1);
 	msg.rpc_argp = &data->arg;
 	msg.rpc_resp = &data->res;
 	task_setup_data.callback_data = data;
@@ -4297,7 +4314,7 @@ static void nfs4_lock_prepare(struct rpc_task *task, void *calldata)
 	data->timestamp = jiffies;
 	if (nfs4_setup_sequence(data->server,
 				&data->arg.seq_args,
-				&data->res.seq_res, 1, task))
+				&data->res.seq_res, task))
 		return;
 	rpc_call_start(task);
 	dprintk("%s: done!, ret = %d\n", __func__, data->rpc_status);
@@ -4415,6 +4432,7 @@ static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *f
 			data->arg.reclaim = NFS_LOCK_RECLAIM;
 		task_setup_data.callback_ops = &nfs4_recover_lock_ops;
 	}
+	nfs41_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1);
 	msg.rpc_argp = &data->arg;
 	msg.rpc_resp = &data->res;
 	task_setup_data.callback_data = data;
@@ -4930,7 +4948,7 @@ static void nfs4_get_lease_time_prepare(struct rpc_task *task,
 	   since we're invoked within one */
 	ret = nfs41_setup_sequence(data->clp->cl_session,
 				   &data->args->la_seq_args,
-				   &data->res->lr_seq_res, 0, task);
+				   &data->res->lr_seq_res, task);
 
 	BUG_ON(ret == -EAGAIN);
 	rpc_call_start(task);
@@ -4994,6 +5012,7 @@ int nfs4_proc_get_lease_time(struct nfs_client *clp, struct nfs_fsinfo *fsinfo)
 	};
 	int status;
 
+	nfs41_init_sequence(&args.la_seq_args, &res.lr_seq_res, 0);
 	dprintk("--> %s\n", __func__);
 	task = rpc_run_task(&task_setup);
 
@@ -5467,7 +5486,7 @@ static void nfs41_sequence_prepare(struct rpc_task *task, void *data)
 	args = task->tk_msg.rpc_argp;
 	res = task->tk_msg.rpc_resp;
 
-	if (nfs41_setup_sequence(clp->cl_session, args, res, 0, task))
+	if (nfs41_setup_sequence(clp->cl_session, args, res, task))
 		return;
 	rpc_call_start(task);
 }
@@ -5499,6 +5518,7 @@ static struct rpc_task *_nfs41_proc_sequence(struct nfs_client *clp, struct rpc_
 		nfs_put_client(clp);
 		return ERR_PTR(-ENOMEM);
 	}
+	nfs41_init_sequence(&calldata->args, &calldata->res, 0);
 	msg.rpc_argp = &calldata->args;
 	msg.rpc_resp = &calldata->res;
 	calldata->clp = clp;
@@ -5560,7 +5580,7 @@ static void nfs4_reclaim_complete_prepare(struct rpc_task *task, void *data)
 	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
 	if (nfs41_setup_sequence(calldata->clp->cl_session,
 				&calldata->arg.seq_args,
-				&calldata->res.seq_res, 0, task))
+				&calldata->res.seq_res, task))
 		return;
 
 	rpc_call_start(task);
@@ -5639,6 +5659,7 @@ static int nfs41_proc_reclaim_complete(struct nfs_client *clp)
 	calldata->clp = clp;
 	calldata->arg.one_fs = 0;
 
+	nfs41_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 0);
 	msg.rpc_argp = &calldata->arg;
 	msg.rpc_resp = &calldata->res;
 	task_setup_data.callback_data = calldata;
@@ -5670,7 +5691,7 @@ nfs4_layoutget_prepare(struct rpc_task *task, void *calldata)
 	 * to be no way to prevent it completely.
 	 */
 	if (nfs4_setup_sequence(server, &lgp->args.seq_args,
-				&lgp->res.seq_res, 0, task))
+				&lgp->res.seq_res, task))
 		return;
 	if (pnfs_choose_layoutget_stateid(&lgp->args.stateid,
 					  NFS_I(lgp->args.inode)->layout,
@@ -5745,6 +5766,7 @@ int nfs4_proc_layoutget(struct nfs4_layoutget *lgp)
 
 	lgp->res.layoutp = &lgp->args.layout;
 	lgp->res.seq_res.sr_slot = NULL;
+	nfs41_init_sequence(&lgp->args.seq_args, &lgp->res.seq_res, 0);
 	task = rpc_run_task(&task_setup_data);
 	if (IS_ERR(task))
 		return PTR_ERR(task);
@@ -5765,7 +5787,7 @@ nfs4_layoutreturn_prepare(struct rpc_task *task, void *calldata)
 
 	dprintk("--> %s\n", __func__);
 	if (nfs41_setup_sequence(lrp->clp->cl_session, &lrp->args.seq_args,
-				&lrp->res.seq_res, 0, task))
+				&lrp->res.seq_res, task))
 		return;
 	rpc_call_start(task);
 }
@@ -5831,6 +5853,7 @@ int nfs4_proc_layoutreturn(struct nfs4_layoutreturn *lrp)
 	int status;
 
 	dprintk("--> %s\n", __func__);
+	nfs41_init_sequence(&lrp->args.seq_args, &lrp->res.seq_res, 1);
 	task = rpc_run_task(&task_setup_data);
 	if (IS_ERR(task))
 		return PTR_ERR(task);
@@ -5931,7 +5954,7 @@ static void nfs4_layoutcommit_prepare(struct rpc_task *task, void *calldata)
 	struct nfs_server *server = NFS_SERVER(data->args.inode);
 
 	if (nfs4_setup_sequence(server, &data->args.seq_args,
-				&data->res.seq_res, 1, task))
+				&data->res.seq_res, task))
 		return;
 	rpc_call_start(task);
 }
@@ -6018,6 +6041,7 @@ nfs4_proc_layoutcommit(struct nfs4_layoutcommit_data *data, bool sync)
 		data->args.lastbytewritten,
 		data->args.inode->i_ino);
 
+	nfs41_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
 	task = rpc_run_task(&task_setup_data);
 	if (IS_ERR(task))
 		return PTR_ERR(task);
@@ -6113,7 +6137,6 @@ out:
 }
 static int _nfs41_test_stateid(struct nfs_server *server, struct nfs4_state *state)
 {
-	int status;
 	struct nfs41_test_stateid_args args = {
 		.stateid = &state->stateid,
 	};
@@ -6123,9 +6146,8 @@ static int _nfs41_test_stateid(struct nfs_server *server, struct nfs4_state *sta
 		.rpc_argp = &args,
 		.rpc_resp = &res,
 	};
-	args.seq_args.sa_session = res.seq_res.sr_session = NULL;
-	status = nfs4_call_sync_sequence(server->client, server, &msg, &args.seq_args, &res.seq_res, 0, 1);
-	return status;
+	nfs41_init_sequence(&args.seq_args, &res.seq_res, 0);
+	return nfs4_call_sync_sequence(server->client, server, &msg, &args.seq_args, &res.seq_res, 1);
 }
 
 static int nfs41_test_stateid(struct nfs_server *server, struct nfs4_state *state)
@@ -6142,7 +6164,6 @@ static int nfs41_test_stateid(struct nfs_server *server, struct nfs4_state *stat
 
 static int _nfs4_free_stateid(struct nfs_server *server, struct nfs4_state *state)
 {
-	int status;
 	struct nfs41_free_stateid_args args = {
 		.stateid = &state->stateid,
 	};
@@ -6153,9 +6174,8 @@ static int _nfs4_free_stateid(struct nfs_server *server, struct nfs4_state *stat
 		.rpc_resp = &res,
 	};
 
-	args.seq_args.sa_session = res.seq_res.sr_session = NULL;
-	status = nfs4_call_sync_sequence(server->client, server, &msg, &args.seq_args, &res.seq_res, 0, 1);
-	return status;
+	nfs41_init_sequence(&args.seq_args, &res.seq_res, 0);
+	return nfs4_call_sync_sequence(server->client, server, &msg, &args.seq_args, &res.seq_res, 1);
 }
 
 static int nfs41_free_stateid(struct nfs_server *server, struct nfs4_state *state)
diff --git a/fs/nfs/read.c b/fs/nfs/read.c
index cfa175c..3c2540d 100644
--- a/fs/nfs/read.c
+++ b/fs/nfs/read.c
@@ -472,7 +472,7 @@ void nfs_read_prepare(struct rpc_task *task, void *calldata)
 
 	if (nfs4_setup_sequence(NFS_SERVER(data->inode),
 				&data->args.seq_args, &data->res.seq_res,
-				0, task))
+				task))
 		return;
 	rpc_call_start(task);
 }
diff --git a/fs/nfs/unlink.c b/fs/nfs/unlink.c
index 4f9319a..490613b 100644
--- a/fs/nfs/unlink.c
+++ b/fs/nfs/unlink.c
@@ -114,7 +114,7 @@ void nfs_unlink_prepare(struct rpc_task *task, void *calldata)
 	struct nfs_server *server = NFS_SERVER(data->dir);
 
 	if (nfs4_setup_sequence(server, &data->args.seq_args,
-				&data->res.seq_res, 1, task))
+				&data->res.seq_res, task))
 		return;
 	rpc_call_start(task);
 }
@@ -410,7 +410,7 @@ static void nfs_rename_prepare(struct rpc_task *task, void *calldata)
 	struct nfs_server *server = NFS_SERVER(data->old_dir);
 
 	if (nfs4_setup_sequence(server, &data->args.seq_args,
-				&data->res.seq_res, 1, task))
+				&data->res.seq_res, task))
 		return;
 	rpc_call_start(task);
 }
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index 834f0fe..0b1831d 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -1135,7 +1135,7 @@ void nfs_write_prepare(struct rpc_task *task, void *calldata)
 
 	if (nfs4_setup_sequence(NFS_SERVER(data->inode),
 				&data->args.seq_args,
-				&data->res.seq_res, 1, task))
+				&data->res.seq_res, task))
 		return;
 	rpc_call_start(task);
 }
-- 
1.7.7.5

[PATCH v2 06/10] NFSv4: Move contents of struct rpc_sequence into struct nfs_seqid_counter

[Trond Myklebust]

Clean up.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/nfs4_fs.h   |   16 +++-------------
 fs/nfs/nfs4state.c |   36 +++++++++++++++++++++++-------------
 2 files changed, 26 insertions(+), 26 deletions(-)

diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index 583ded0..cef9d5d 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -53,21 +53,13 @@ struct nfs4_minor_version_ops {
 	const struct nfs4_state_maintenance_ops *state_renewal_ops;
 };
 
-/*
- * struct rpc_sequence ensures that RPC calls are sent in the exact
- * order that they appear on the list.
- */
-struct rpc_sequence {
-	struct rpc_wait_queue	wait;	/* RPC call delay queue */
-	spinlock_t lock;		/* Protects the list */
-	struct list_head list;		/* Defines sequence of RPC calls */
-};
-
 #define NFS_SEQID_CONFIRMED 1
 struct nfs_seqid_counter {
-	struct rpc_sequence *sequence;
 	int flags;
 	u32 counter;
+	spinlock_t lock;		/* Protects the list */
+	struct list_head list;		/* Defines sequence of RPC calls */
+	struct rpc_wait_queue	wait;	/* RPC call delay queue */
 };
 
 struct nfs_seqid {
@@ -99,7 +91,6 @@ struct nfs4_state_owner {
 	unsigned long	     so_flags;
 	struct list_head     so_states;
 	struct nfs_seqid_counter so_seqid;
-	struct rpc_sequence  so_sequence;
 	int		     so_owner_id;
 };
 
@@ -142,7 +133,6 @@ struct nfs4_lock_state {
 	int			ls_flags;
 	int			ls_id;
 	struct nfs_seqid_counter	ls_seqid;
-	struct rpc_sequence	ls_sequence;
 	nfs4_stateid		ls_stateid;
 	atomic_t		ls_count;
 	struct nfs4_lock_owner	ls_owner;
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index 5abf236..cf7bc39 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -384,6 +384,22 @@ nfs4_remove_state_owner_locked(struct nfs4_state_owner *sp)
 	ida_remove(&server->openowner_id, sp->so_owner_id);
 }
 
+static void
+nfs4_init_seqid_counter(struct nfs_seqid_counter *sc)
+{
+	sc->flags = 0;
+	sc->counter = 0;
+	spin_lock_init(&sc->lock);
+	INIT_LIST_HEAD(&sc->list);
+	rpc_init_wait_queue(&sc->wait, "Seqid_waitqueue");
+}
+
+static void
+nfs4_destroy_seqid_counter(struct nfs_seqid_counter *sc)
+{
+	rpc_destroy_wait_queue(&sc->wait);
+}
+
 /*
  * nfs4_alloc_state_owner(): this is called on the OPEN or CREATE path to
  * create a new state_owner.
@@ -403,10 +419,7 @@ nfs4_alloc_state_owner(struct nfs_server *server,
 	sp->so_cred = get_rpccred(cred);
 	spin_lock_init(&sp->so_lock);
 	INIT_LIST_HEAD(&sp->so_states);
-	rpc_init_wait_queue(&sp->so_sequence.wait, "Seqid_waitqueue");
-	sp->so_seqid.sequence = &sp->so_sequence;
-	spin_lock_init(&sp->so_sequence.lock);
-	INIT_LIST_HEAD(&sp->so_sequence.list);
+	nfs4_init_seqid_counter(&sp->so_seqid);
 	atomic_set(&sp->so_count, 1);
 	INIT_LIST_HEAD(&sp->so_lru);
 	return sp;
@@ -428,7 +441,7 @@ nfs4_drop_state_owner(struct nfs4_state_owner *sp)
 
 static void nfs4_free_state_owner(struct nfs4_state_owner *sp)
 {
-	rpc_destroy_wait_queue(&sp->so_sequence.wait);
+	nfs4_destroy_seqid_counter(&sp->so_seqid);
 	put_rpccred(sp->so_cred);
 	kfree(sp);
 }
@@ -748,10 +761,7 @@ static struct nfs4_lock_state *nfs4_alloc_lock_state(struct nfs4_state *state, f
 	lsp = kzalloc(sizeof(*lsp), GFP_NOFS);
 	if (lsp == NULL)
 		return NULL;
-	rpc_init_wait_queue(&lsp->ls_sequence.wait, "lock_seqid_waitqueue");
-	spin_lock_init(&lsp->ls_sequence.lock);
-	INIT_LIST_HEAD(&lsp->ls_sequence.list);
-	lsp->ls_seqid.sequence = &lsp->ls_sequence;
+	nfs4_init_seqid_counter(&lsp->ls_seqid);
 	atomic_set(&lsp->ls_count, 1);
 	lsp->ls_state = state;
 	lsp->ls_owner.lo_type = type;
@@ -780,7 +790,7 @@ static void nfs4_free_lock_state(struct nfs4_lock_state *lsp)
 	struct nfs_server *server = lsp->ls_state->owner->so_server;
 
 	ida_simple_remove(&server->lockowner_id, lsp->ls_id);
-	rpc_destroy_wait_queue(&lsp->ls_sequence.wait);
+	nfs4_destroy_seqid_counter(&lsp->ls_seqid);
 	kfree(lsp);
 }
 
@@ -914,7 +924,7 @@ struct nfs_seqid *nfs_alloc_seqid(struct nfs_seqid_counter *counter, gfp_t gfp_m
 void nfs_release_seqid(struct nfs_seqid *seqid)
 {
 	if (!list_empty(&seqid->list)) {
-		struct rpc_sequence *sequence = seqid->sequence->sequence;
+		struct nfs_seqid_counter *sequence = seqid->sequence;
 
 		spin_lock(&sequence->lock);
 		list_del_init(&seqid->list);
@@ -936,7 +946,7 @@ void nfs_free_seqid(struct nfs_seqid *seqid)
  */
 static void nfs_increment_seqid(int status, struct nfs_seqid *seqid)
 {
-	BUG_ON(list_first_entry(&seqid->sequence->sequence->list, struct nfs_seqid, list) != seqid);
+	BUG_ON(list_first_entry(&seqid->sequence->list, struct nfs_seqid, list) != seqid);
 	switch (status) {
 		case 0:
 			break;
@@ -987,7 +997,7 @@ void nfs_increment_lock_seqid(int status, struct nfs_seqid *seqid)
 
 int nfs_wait_on_sequence(struct nfs_seqid *seqid, struct rpc_task *task)
 {
-	struct rpc_sequence *sequence = seqid->sequence->sequence;
+	struct nfs_seqid_counter *sequence = seqid->sequence;
 	int status = 0;
 
 	spin_lock(&sequence->lock);
-- 
1.7.7.5

[PATCH v2 07/10] NFS: Move struct nfs_unique_id into struct nfs_seqid_counter

[Trond Myklebust]

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/nfs4_fs.h   |    8 ++++++--
 fs/nfs/nfs4proc.c  |   10 +++++-----
 fs/nfs/nfs4state.c |   10 +++++-----
 3 files changed, 16 insertions(+), 12 deletions(-)

diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index cef9d5d..b7e5a1d 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -53,8 +53,14 @@ struct nfs4_minor_version_ops {
 	const struct nfs4_state_maintenance_ops *state_renewal_ops;
 };
 
+struct nfs_unique_id {
+	struct rb_node rb_node;
+	__u64 id;
+};
+
 #define NFS_SEQID_CONFIRMED 1
 struct nfs_seqid_counter {
+	int owner_id;
 	int flags;
 	u32 counter;
 	spinlock_t lock;		/* Protects the list */
@@ -91,7 +97,6 @@ struct nfs4_state_owner {
 	unsigned long	     so_flags;
 	struct list_head     so_states;
 	struct nfs_seqid_counter so_seqid;
-	int		     so_owner_id;
 };
 
 enum {
@@ -131,7 +136,6 @@ struct nfs4_lock_state {
 	struct nfs4_state *	ls_state;	/* Pointer to open state */
 #define NFS_LOCK_INITIALIZED 1
 	int			ls_flags;
-	int			ls_id;
 	struct nfs_seqid_counter	ls_seqid;
 	nfs4_stateid		ls_stateid;
 	atomic_t		ls_count;
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 3233ee0..7d431d7 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -824,7 +824,7 @@ static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
 	p->o_arg.open_flags = flags;
 	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
 	p->o_arg.clientid = server->nfs_client->cl_clientid;
-	p->o_arg.id = sp->so_owner_id;
+	p->o_arg.id = sp->so_seqid.owner_id;
 	p->o_arg.name = &dentry->d_name;
 	p->o_arg.server = server;
 	p->o_arg.bitmask = server->attr_bitmask;
@@ -1449,7 +1449,7 @@ static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
 		rcu_read_unlock();
 	}
 	/* Update sequence id. */
-	data->o_arg.id = sp->so_owner_id;
+	data->o_arg.id = sp->so_seqid.owner_id;
 	data->o_arg.clientid = sp->so_server->nfs_client->cl_clientid;
 	if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS) {
 		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
@@ -4032,7 +4032,7 @@ static int _nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock
 	if (status != 0)
 		goto out;
 	lsp = request->fl_u.nfs4_fl.owner;
-	arg.lock_owner.id = lsp->ls_id;
+	arg.lock_owner.id = lsp->ls_seqid.owner_id;
 	arg.lock_owner.s_dev = server->s_dev;
 	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
 	switch (status) {
@@ -4278,7 +4278,7 @@ static struct nfs4_lockdata *nfs4_alloc_lockdata(struct file_lock *fl,
 		goto out_free_seqid;
 	p->arg.lock_stateid = &lsp->ls_stateid;
 	p->arg.lock_owner.clientid = server->nfs_client->cl_clientid;
-	p->arg.lock_owner.id = lsp->ls_id;
+	p->arg.lock_owner.id = lsp->ls_seqid.owner_id;
 	p->arg.lock_owner.s_dev = server->s_dev;
 	p->res.lock_seqid = p->arg.lock_seqid;
 	p->lsp = lsp;
@@ -4696,7 +4696,7 @@ void nfs4_release_lockowner(const struct nfs4_lock_state *lsp)
 	if (!args)
 		return;
 	args->lock_owner.clientid = server->nfs_client->cl_clientid;
-	args->lock_owner.id = lsp->ls_id;
+	args->lock_owner.id = lsp->ls_seqid.owner_id;
 	args->lock_owner.s_dev = server->s_dev;
 	msg.rpc_argp = args;
 	rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, args);
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index cf7bc39..a42e60d 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -366,7 +366,7 @@ nfs4_insert_state_owner_locked(struct nfs4_state_owner *new)
 			return sp;
 		}
 	}
-	err = ida_get_new(&server->openowner_id, &new->so_owner_id);
+	err = ida_get_new(&server->openowner_id, &new->so_seqid.owner_id);
 	if (err)
 		return ERR_PTR(err);
 	rb_link_node(&new->so_server_node, parent, p);
@@ -381,7 +381,7 @@ nfs4_remove_state_owner_locked(struct nfs4_state_owner *sp)
 
 	if (!RB_EMPTY_NODE(&sp->so_server_node))
 		rb_erase(&sp->so_server_node, &server->state_owners);
-	ida_remove(&server->openowner_id, sp->so_owner_id);
+	ida_remove(&server->openowner_id, sp->so_seqid.owner_id);
 }
 
 static void
@@ -775,8 +775,8 @@ static struct nfs4_lock_state *nfs4_alloc_lock_state(struct nfs4_state *state, f
 	default:
 		goto out_free;
 	}
-	lsp->ls_id = ida_simple_get(&server->lockowner_id, 0, 0, GFP_NOFS);
-	if (lsp->ls_id < 0)
+	lsp->ls_seqid.owner_id = ida_simple_get(&server->lockowner_id, 0, 0, GFP_NOFS);
+	if (lsp->ls_seqid.owner_id < 0)
 		goto out_free;
 	INIT_LIST_HEAD(&lsp->ls_locks);
 	return lsp;
@@ -789,7 +789,7 @@ static void nfs4_free_lock_state(struct nfs4_lock_state *lsp)
 {
 	struct nfs_server *server = lsp->ls_state->owner->so_server;
 
-	ida_simple_remove(&server->lockowner_id, lsp->ls_id);
+	ida_simple_remove(&server->lockowner_id, lsp->ls_seqid.owner_id);
 	nfs4_destroy_seqid_counter(&lsp->ls_seqid);
 	kfree(lsp);
 }
-- 
1.7.7.5

[PATCH v2 08/10] NFS: Optimise away unnecessary setattrs for open(O_TRUNC);

[Trond Myklebust]

Currently, we will correctly optimise away a truncate that doesn't
change the file size. However, in the case of open(O_TRUNC), we
also want to optimise away the time changes.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/dir.c      |   25 +++++++++++++++++++------
 fs/nfs/inode.c    |    4 ++--
 fs/nfs/nfs4proc.c |   10 +++++++---
 3 files changed, 28 insertions(+), 11 deletions(-)

diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index fd9a872..bb132a8 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -1429,6 +1429,7 @@ static struct dentry *nfs_atomic_lookup(struct inode *dir, struct dentry *dentry
 	}
 
 	open_flags = nd->intent.open.flags;
+	attr.ia_valid = 0;
 
 	ctx = create_nfs_open_context(dentry, open_flags);
 	res = ERR_CAST(ctx);
@@ -1437,11 +1438,14 @@ static struct dentry *nfs_atomic_lookup(struct inode *dir, struct dentry *dentry
 
 	if (nd->flags & LOOKUP_CREATE) {
 		attr.ia_mode = nd->intent.open.create_mode;
-		attr.ia_valid = ATTR_MODE;
+		attr.ia_valid |= ATTR_MODE;
 		attr.ia_mode &= ~current_umask();
-	} else {
+	} else
 		open_flags &= ~(O_EXCL | O_CREAT);
-		attr.ia_valid = 0;
+
+	if (open_flags & O_TRUNC) {
+		attr.ia_valid |= ATTR_SIZE;
+		attr.ia_size = 0;
 	}
 
 	/* Open the file on the server */
@@ -1495,6 +1499,7 @@ static int nfs_open_revalidate(struct dentry *dentry, struct nameidata *nd)
 	struct inode *inode;
 	struct inode *dir;
 	struct nfs_open_context *ctx;
+	struct iattr attr;
 	int openflags, ret = 0;
 
 	if (nd->flags & LOOKUP_RCU)
@@ -1523,19 +1528,27 @@ static int nfs_open_revalidate(struct dentry *dentry, struct nameidata *nd)
 	/* We cannot do exclusive creation on a positive dentry */
 	if ((openflags & (O_CREAT|O_EXCL)) == (O_CREAT|O_EXCL))
 		goto no_open_dput;
-	/* We can't create new files, or truncate existing ones here */
-	openflags &= ~(O_CREAT|O_EXCL|O_TRUNC);
+	/* We can't create new files here */
+	openflags &= ~(O_CREAT|O_EXCL);
 
 	ctx = create_nfs_open_context(dentry, openflags);
 	ret = PTR_ERR(ctx);
 	if (IS_ERR(ctx))
 		goto out;
+
+	attr.ia_valid = 0;
+	if (openflags & O_TRUNC) {
+		attr.ia_valid |= ATTR_SIZE;
+		attr.ia_size = 0;
+		nfs_wb_all(inode);
+	}
+
 	/*
 	 * Note: we're not holding inode->i_mutex and so may be racing with
 	 * operations that change the directory. We therefore save the
 	 * change attribute *before* we do the RPC call.
 	 */
-	inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, NULL);
+	inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, &attr);
 	if (IS_ERR(inode)) {
 		ret = PTR_ERR(inode);
 		switch (ret) {
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 577ad5a..65486e6 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -402,7 +402,7 @@ out_no_inode:
 	goto out;
 }
 
-#define NFS_VALID_ATTRS (ATTR_MODE|ATTR_UID|ATTR_GID|ATTR_SIZE|ATTR_ATIME|ATTR_ATIME_SET|ATTR_MTIME|ATTR_MTIME_SET|ATTR_FILE)
+#define NFS_VALID_ATTRS (ATTR_MODE|ATTR_UID|ATTR_GID|ATTR_SIZE|ATTR_ATIME|ATTR_ATIME_SET|ATTR_MTIME|ATTR_MTIME_SET|ATTR_FILE|ATTR_OPEN)
 
 int
 nfs_setattr(struct dentry *dentry, struct iattr *attr)
@@ -424,7 +424,7 @@ nfs_setattr(struct dentry *dentry, struct iattr *attr)
 
 	/* Optimization: if the end result is no change, don't RPC */
 	attr->ia_valid &= NFS_VALID_ATTRS;
-	if ((attr->ia_valid & ~ATTR_FILE) == 0)
+	if ((attr->ia_valid & ~(ATTR_FILE|ATTR_OPEN)) == 0)
 		return 0;
 
 	/* Write all dirty data */
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 7d431d7..f2a4d32 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -830,7 +830,7 @@ static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
 	p->o_arg.bitmask = server->attr_bitmask;
 	p->o_arg.dir_bitmask = server->cache_consistency_bitmask;
 	p->o_arg.claim = NFS4_OPEN_CLAIM_NULL;
-	if (flags & O_CREAT) {
+	if (attrs != NULL && attrs->ia_valid != 0) {
 		u32 *s;
 
 		p->o_arg.u.attrs = &p->attrs;
@@ -887,7 +887,7 @@ static int can_open_cached(struct nfs4_state *state, fmode_t mode, int open_mode
 {
 	int ret = 0;
 
-	if (open_mode & O_EXCL)
+	if (open_mode & (O_EXCL|O_TRUNC))
 		goto out;
 	switch (mode & (FMODE_READ|FMODE_WRITE)) {
 		case FMODE_READ:
@@ -1035,7 +1035,7 @@ static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
 	struct nfs4_state *state = opendata->state;
 	struct nfs_inode *nfsi = NFS_I(state->inode);
 	struct nfs_delegation *delegation;
-	int open_mode = opendata->o_arg.open_flags & O_EXCL;
+	int open_mode = opendata->o_arg.open_flags & (O_EXCL|O_TRUNC);
 	fmode_t fmode = opendata->o_arg.fmode;
 	nfs4_stateid stateid;
 	int ret = -EAGAIN;
@@ -2432,6 +2432,10 @@ nfs4_proc_setattr(struct dentry *dentry, struct nfs_fattr *fattr,
 		}
 	}
 
+	/* Deal with open(O_TRUNC) */
+	if (sattr->ia_valid & ATTR_OPEN)
+		sattr->ia_valid &= ~(ATTR_MTIME|ATTR_CTIME|ATTR_OPEN);
+
 	status = nfs4_do_setattr(inode, cred, fattr, sattr, state);
 	if (status == 0)
 		nfs_setattr_update_inode(inode, sattr);
-- 
1.7.7.5

[PATCH v2 09/10] NFS: Ensure that mmapped pages remain stable during writeback

[Trond Myklebust]

Ensure that nfs_vm_page_mkwrite() waits for the page writeback to
complete before the application is allowed to modify page
contents.
The main reason for wanting to do this in NFS is to ensure that the
server doesn't get confused if we have to resend the RPC request
due to a dropped/missed reply.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/file.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index c43a452..4fdaaa6 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -530,6 +530,8 @@ static int nfs_vm_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
 	if (mapping != dentry->d_inode->i_mapping)
 		goto out_unlock;
 
+	wait_on_page_writeback(page);
+
 	pagelen = nfs_page_length(page);
 	if (pagelen == 0)
 		goto out_unlock;
-- 
1.7.7.5

[PATCH v2 10/10] SUNRPC: Fix potential races in xprt_lock_write_next()

[Trond Myklebust]

We have to ensure that the wake up from the waitqueue and the assignment
of xprt->snd_task are atomic. We can do this by assigning the snd_task
while under the waitqueue spinlock.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/nfs4_fs.h             |    1 +
 fs/nfs/nfs4proc.c            |   13 ++++++----
 fs/nfs/nfs4state.c           |   17 ++++++-------
 include/linux/sunrpc/sched.h |    3 ++
 net/sunrpc/sched.c           |   42 ++++++++++++++++++++++++++++-------
 net/sunrpc/xprt.c            |   49 ++++++++++++++++++++++-------------------
 6 files changed, 79 insertions(+), 46 deletions(-)

diff --git a/fs/nfs/nfs4_fs.h b/fs/nfs/nfs4_fs.h
index b7e5a1d..fbb1bc0 100644
--- a/fs/nfs/nfs4_fs.h
+++ b/fs/nfs/nfs4_fs.h
@@ -222,6 +222,7 @@ static inline struct nfs4_session *nfs4_get_session(const struct nfs_server *ser
 	return server->nfs_client->cl_session;
 }
 
+extern bool nfs4_set_task_privileged(struct rpc_task *task, void *dummy);
 extern int nfs4_setup_sequence(const struct nfs_server *server,
 		struct nfs4_sequence_args *args, struct nfs4_sequence_res *res,
 		struct rpc_task *task);
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index f2a4d32..734bd5a 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -385,17 +385,20 @@ nfs4_free_slot(struct nfs4_slot_table *tbl, u8 free_slotid)
 		free_slotid, tbl->highest_used_slotid);
 }
 
+bool nfs4_set_task_privileged(struct rpc_task *task, void *dummy)
+{
+	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
+	return true;
+}
+
 /*
  * Signal state manager thread if session fore channel is drained
  */
 static void nfs4_check_drain_fc_complete(struct nfs4_session *ses)
 {
-	struct rpc_task *task;
-
 	if (!test_bit(NFS4_SESSION_DRAINING, &ses->session_state)) {
-		task = rpc_wake_up_next(&ses->fc_slot_table.slot_tbl_waitq);
-		if (task)
-			rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
+		rpc_wake_up_first(&ses->fc_slot_table.slot_tbl_waitq,
+				nfs4_set_task_privileged, NULL);
 		return;
 	}
 
diff --git a/fs/nfs/nfs4state.c b/fs/nfs/nfs4state.c
index a42e60d..f0e9881 100644
--- a/fs/nfs/nfs4state.c
+++ b/fs/nfs/nfs4state.c
@@ -190,23 +190,22 @@ static int nfs41_setup_state_renewal(struct nfs_client *clp)
 static void nfs4_end_drain_session(struct nfs_client *clp)
 {
 	struct nfs4_session *ses = clp->cl_session;
+	struct nfs4_slot_table *tbl;
 	int max_slots;
 
 	if (ses == NULL)
 		return;
+	tbl = &ses->fc_slot_table;
 	if (test_and_clear_bit(NFS4_SESSION_DRAINING, &ses->session_state)) {
-		spin_lock(&ses->fc_slot_table.slot_tbl_lock);
-		max_slots = ses->fc_slot_table.max_slots;
+		spin_lock(&tbl->slot_tbl_lock);
+		max_slots = tbl->max_slots;
 		while (max_slots--) {
-			struct rpc_task *task;
-
-			task = rpc_wake_up_next(&ses->fc_slot_table.
-						slot_tbl_waitq);
-			if (!task)
+			if (rpc_wake_up_first(&tbl->slot_tbl_waitq,
+						nfs4_set_task_privileged,
+						NULL) == NULL)
 				break;
-			rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
 		}
-		spin_unlock(&ses->fc_slot_table.slot_tbl_lock);
+		spin_unlock(&tbl->slot_tbl_lock);
 	}
 }
 
diff --git a/include/linux/sunrpc/sched.h b/include/linux/sunrpc/sched.h
index b16243a..bd337f9 100644
--- a/include/linux/sunrpc/sched.h
+++ b/include/linux/sunrpc/sched.h
@@ -235,6 +235,9 @@ void		rpc_wake_up_queued_task(struct rpc_wait_queue *,
 					struct rpc_task *);
 void		rpc_wake_up(struct rpc_wait_queue *);
 struct rpc_task *rpc_wake_up_next(struct rpc_wait_queue *);
+struct rpc_task *rpc_wake_up_first(struct rpc_wait_queue *,
+					bool (*)(struct rpc_task *, void *),
+					void *);
 void		rpc_wake_up_status(struct rpc_wait_queue *, int);
 int		rpc_queue_empty(struct rpc_wait_queue *);
 void		rpc_delay(struct rpc_task *, unsigned long);
diff --git a/net/sunrpc/sched.c b/net/sunrpc/sched.c
index 3341d89..7c8142e 100644
--- a/net/sunrpc/sched.c
+++ b/net/sunrpc/sched.c
@@ -422,7 +422,7 @@ EXPORT_SYMBOL_GPL(rpc_wake_up_queued_task);
 /*
  * Wake up the next task on a priority queue.
  */
-static struct rpc_task * __rpc_wake_up_next_priority(struct rpc_wait_queue *queue)
+static struct rpc_task *__rpc_find_next_queued_priority(struct rpc_wait_queue *queue)
 {
 	struct list_head *q;
 	struct rpc_task *task;
@@ -467,30 +467,54 @@ new_queue:
 new_owner:
 	rpc_set_waitqueue_owner(queue, task->tk_owner);
 out:
-	rpc_wake_up_task_queue_locked(queue, task);
 	return task;
 }
 
+static struct rpc_task *__rpc_find_next_queued(struct rpc_wait_queue *queue)
+{
+	if (RPC_IS_PRIORITY(queue))
+		return __rpc_find_next_queued_priority(queue);
+	if (!list_empty(&queue->tasks[0]))
+		return list_first_entry(&queue->tasks[0], struct rpc_task, u.tk_wait.list);
+	return NULL;
+}
+
 /*
- * Wake up the next task on the wait queue.
+ * Wake up the first task on the wait queue.
  */
-struct rpc_task * rpc_wake_up_next(struct rpc_wait_queue *queue)
+struct rpc_task *rpc_wake_up_first(struct rpc_wait_queue *queue,
+		bool (*func)(struct rpc_task *, void *), void *data)
 {
 	struct rpc_task	*task = NULL;
 
-	dprintk("RPC:       wake_up_next(%p \"%s\")\n",
+	dprintk("RPC:       wake_up_first(%p \"%s\")\n",
 			queue, rpc_qname(queue));
 	spin_lock_bh(&queue->lock);
-	if (RPC_IS_PRIORITY(queue))
-		task = __rpc_wake_up_next_priority(queue);
-	else {
-		task_for_first(task, &queue->tasks[0])
+	task = __rpc_find_next_queued(queue);
+	if (task != NULL) {
+		if (func(task, data))
 			rpc_wake_up_task_queue_locked(queue, task);
+		else
+			task = NULL;
 	}
 	spin_unlock_bh(&queue->lock);
 
 	return task;
 }
+EXPORT_SYMBOL_GPL(rpc_wake_up_first);
+
+static bool rpc_wake_up_next_func(struct rpc_task *task, void *data)
+{
+	return true;
+}
+
+/*
+ * Wake up the next task on the wait queue.
+*/
+struct rpc_task * rpc_wake_up_next(struct rpc_wait_queue *queue)
+{
+	return rpc_wake_up_first(queue, rpc_wake_up_next_func, NULL);
+}
 EXPORT_SYMBOL_GPL(rpc_wake_up_next);
 
 /**
diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index c64c0ef..839f6ef 100644
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -292,54 +292,57 @@ static inline int xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task)
 	return retval;
 }
 
-static void __xprt_lock_write_next(struct rpc_xprt *xprt)
+static bool __xprt_lock_write_func(struct rpc_task *task, void *data)
 {
-	struct rpc_task *task;
+	struct rpc_xprt *xprt = data;
 	struct rpc_rqst *req;
 
-	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
-		return;
-
-	task = rpc_wake_up_next(&xprt->sending);
-	if (task == NULL)
-		goto out_unlock;
-
 	req = task->tk_rqstp;
 	xprt->snd_task = task;
 	if (req) {
 		req->rq_bytes_sent = 0;
 		req->rq_ntrans++;
 	}
-	return;
+	return true;
+}
 
-out_unlock:
+static void __xprt_lock_write_next(struct rpc_xprt *xprt)
+{
+	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
+		return;
+
+	if (rpc_wake_up_first(&xprt->sending, __xprt_lock_write_func, xprt))
+		return;
 	xprt_clear_locked(xprt);
 }
 
-static void __xprt_lock_write_next_cong(struct rpc_xprt *xprt)
+static bool __xprt_lock_write_cong_func(struct rpc_task *task, void *data)
 {
-	struct rpc_task *task;
+	struct rpc_xprt *xprt = data;
 	struct rpc_rqst *req;
 
-	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
-		return;
-	if (RPCXPRT_CONGESTED(xprt))
-		goto out_unlock;
-	task = rpc_wake_up_next(&xprt->sending);
-	if (task == NULL)
-		goto out_unlock;
-
 	req = task->tk_rqstp;
 	if (req == NULL) {
 		xprt->snd_task = task;
-		return;
+		return true;
 	}
 	if (__xprt_get_cong(xprt, task)) {
 		xprt->snd_task = task;
 		req->rq_bytes_sent = 0;
 		req->rq_ntrans++;
-		return;
+		return true;
 	}
+	return false;
+}
+
+static void __xprt_lock_write_next_cong(struct rpc_xprt *xprt)
+{
+	if (test_and_set_bit(XPRT_LOCKED, &xprt->state))
+		return;
+	if (RPCXPRT_CONGESTED(xprt))
+		goto out_unlock;
+	if (rpc_wake_up_first(&xprt->sending, __xprt_lock_write_cong_func, xprt))
+		return;
 out_unlock:
 	xprt_clear_locked(xprt);
 }
-- 
1.7.7.5