Merge git://git.linux-nfs.org/pub/linux/nfs-2.6

* git://git.linux-nfs.org/pub/linux/nfs-2.6: (122 commits)
  sunrpc: drop BKL around wrap and unwrap
  NFSv4: Make sure unlock is really an unlock when cancelling a lock
  NLM: fix source address of callback to client
  SUNRPC client: add interface for binding to a local address
  SUNRPC server: record the destination address of a request
  SUNRPC: cleanup transport creation argument passing
  NFSv4: Make the NFS state model work with the nosharedcache mount option
  NFS: Error when mounting the same filesystem with different options
  NFS: Add the mount option "nosharecache"
  NFS: Add support for mounting NFSv4 file systems with string options
  NFS: Add final pieces to support in-kernel mount option parsing
  NFS: Introduce generic mount client API
  NFS: Add enums and match tables for mount option parsing
  NFS: Improve debugging output in NFS in-kernel mount client
  NFS: Clean up in-kernel NFS mount
  NFS: Remake nfsroot_mount as a permanent part of NFS client
  SUNRPC: Add a convenient default for the hostname when calling rpc_create()
  SUNRPC: Rename rpcb_getport to be consistent with new rpcb_getport_sync name
  SUNRPC: Rename rpcb_getport_external routine
  SUNRPC: Allow rpcbind requests to be interrupted by a signal.
  ...

fs/lockd/host.c

@@ -44,9 +44,8 @@ static struct nsm_handle *	nsm_find(const struct sockaddr_in *sin,
  */
 static struct nlm_host *
 nlm_lookup_host(int server, const struct sockaddr_in *sin,
-					int proto, int version,
-					const char *hostname,
-					int hostname_len)
+		int proto, int version, const char *hostname,
+		int hostname_len, const struct sockaddr_in *ssin)
 {
 	struct hlist_head *chain;
 	struct hlist_node *pos;
@@ -54,7 +53,9 @@ nlm_lookup_host(int server, const struct sockaddr_in *sin,
 	struct nsm_handle *nsm = NULL;
 	int		hash;
 
-	dprintk("lockd: nlm_lookup_host(%u.%u.%u.%u, p=%d, v=%d, my role=%s, name=%.*s)\n",
+	dprintk("lockd: nlm_lookup_host("NIPQUAD_FMT"->"NIPQUAD_FMT
+			", p=%d, v=%d, my role=%s, name=%.*s)\n",
+			NIPQUAD(ssin->sin_addr.s_addr),
 			NIPQUAD(sin->sin_addr.s_addr), proto, version,
 			server? "server" : "client",
 			hostname_len,
@@ -91,6 +92,8 @@ nlm_lookup_host(int server, const struct sockaddr_in *sin,
 			continue;
 		if (host->h_server != server)
 			continue;
+		if (!nlm_cmp_addr(&host->h_saddr, ssin))
+			continue;
 
 		/* Move to head of hash chain. */
 		hlist_del(&host->h_hash);
@@ -118,6 +121,7 @@ nlm_lookup_host(int server, const struct sockaddr_in *sin,
 	host->h_name	   = nsm->sm_name;
 	host->h_addr       = *sin;
 	host->h_addr.sin_port = 0;	/* ouch! */
+	host->h_saddr	   = *ssin;
 	host->h_version    = version;
 	host->h_proto      = proto;
 	host->h_rpcclnt    = NULL;
@@ -161,15 +165,9 @@ nlm_destroy_host(struct nlm_host *host)
 	 */
 	nsm_unmonitor(host);
 
-	if ((clnt = host->h_rpcclnt) != NULL) {
-		if (atomic_read(&clnt->cl_users)) {
-			printk(KERN_WARNING
-				"lockd: active RPC handle\n");
-			clnt->cl_dead = 1;
-		} else {
-			rpc_destroy_client(host->h_rpcclnt);
-		}
-	}
+	clnt = host->h_rpcclnt;
+	if (clnt != NULL)
+		rpc_shutdown_client(clnt);
 	kfree(host);
 }
 
@@ -180,8 +178,10 @@ struct nlm_host *
 nlmclnt_lookup_host(const struct sockaddr_in *sin, int proto, int version,
 			const char *hostname, int hostname_len)
 {
+	struct sockaddr_in ssin = {0};
+
 	return nlm_lookup_host(0, sin, proto, version,
-			       hostname, hostname_len);
+			       hostname, hostname_len, &ssin);
 }
 
 /*
@@ -191,9 +191,12 @@ struct nlm_host *
 nlmsvc_lookup_host(struct svc_rqst *rqstp,
 			const char *hostname, int hostname_len)
 {
+	struct sockaddr_in ssin = {0};
+
+	ssin.sin_addr = rqstp->rq_daddr.addr;
 	return nlm_lookup_host(1, svc_addr_in(rqstp),
 			       rqstp->rq_prot, rqstp->rq_vers,
-			       hostname, hostname_len);
+			       hostname, hostname_len, &ssin);
 }
 
 /*
@@ -204,8 +207,9 @@ nlm_bind_host(struct nlm_host *host)
 {
 	struct rpc_clnt	*clnt;
 
-	dprintk("lockd: nlm_bind_host(%08x)\n",
-			(unsigned)ntohl(host->h_addr.sin_addr.s_addr));
+	dprintk("lockd: nlm_bind_host("NIPQUAD_FMT"->"NIPQUAD_FMT")\n",
+			NIPQUAD(host->h_saddr.sin_addr),
+			NIPQUAD(host->h_addr.sin_addr));
 
 	/* Lock host handle */
 	mutex_lock(&host->h_mutex);
@@ -232,6 +236,7 @@ nlm_bind_host(struct nlm_host *host)
 			.protocol	= host->h_proto,
 			.address	= (struct sockaddr *)&host->h_addr,
 			.addrsize	= sizeof(host->h_addr),
+			.saddress	= (struct sockaddr *)&host->h_saddr,
 			.timeout	= &timeparms,
 			.servername	= host->h_name,
 			.program	= &nlm_program,

fs/lockd/mon.c

@@ -61,6 +61,7 @@ nsm_mon_unmon(struct nsm_handle *nsm, u32 proc, struct nsm_res *res)
 			status);
 	else
 		status = 0;
+	rpc_shutdown_client(clnt);
  out:
 	return status;
 }
@@ -138,7 +139,6 @@ nsm_create(void)
 		.program	= &nsm_program,
 		.version	= SM_VERSION,
 		.authflavor	= RPC_AUTH_NULL,
-		.flags		= (RPC_CLNT_CREATE_ONESHOT),
 	};
 
 	return rpc_create(&args);

fs/lockd/svc.c

@@ -123,9 +123,6 @@ lockd(struct svc_rqst *rqstp)
 	/* Process request with signals blocked, but allow SIGKILL.  */
 	allow_signal(SIGKILL);
 
-	/* kick rpciod */
-	rpciod_up();
-
 	dprintk("NFS locking service started (ver " LOCKD_VERSION ").\n");
 
 	if (!nlm_timeout)
@@ -202,9 +199,6 @@ lockd(struct svc_rqst *rqstp)
 	/* Exit the RPC thread */
 	svc_exit_thread(rqstp);
 
-	/* release rpciod */
-	rpciod_down();
-
 	/* Release module */
 	unlock_kernel();
 	module_put_and_exit(0);

fs/nfs/Makefile

@@ -6,8 +6,8 @@ obj-$(CONFIG_NFS_FS) += nfs.o
 
 nfs-y 			:= client.o dir.o file.o getroot.o inode.o super.o nfs2xdr.o \
 			   pagelist.o proc.o read.o symlink.o unlink.o \
-			   write.o namespace.o
-nfs-$(CONFIG_ROOT_NFS)	+= nfsroot.o mount_clnt.o      
+			   write.o namespace.o mount_clnt.o
+nfs-$(CONFIG_ROOT_NFS)	+= nfsroot.o
 nfs-$(CONFIG_NFS_V3)	+= nfs3proc.o nfs3xdr.o
 nfs-$(CONFIG_NFS_V3_ACL)	+= nfs3acl.o
 nfs-$(CONFIG_NFS_V4)	+= nfs4proc.o nfs4xdr.o nfs4state.o nfs4renewd.o \

fs/nfs/client.c

@@ -102,19 +102,10 @@ static struct nfs_client *nfs_alloc_client(const char *hostname,
 					   int nfsversion)
 {
 	struct nfs_client *clp;
-	int error;
 
 	if ((clp = kzalloc(sizeof(*clp), GFP_KERNEL)) == NULL)
 		goto error_0;
 
-	error = rpciod_up();
-	if (error < 0) {
-		dprintk("%s: couldn't start rpciod! Error = %d\n",
-				__FUNCTION__, error);
-		goto error_1;
-	}
-	__set_bit(NFS_CS_RPCIOD, &clp->cl_res_state);
-
 	if (nfsversion == 4) {
 		if (nfs_callback_up() < 0)
 			goto error_2;
@@ -139,8 +130,6 @@ static struct nfs_client *nfs_alloc_client(const char *hostname,
 #ifdef CONFIG_NFS_V4
 	init_rwsem(&clp->cl_sem);
 	INIT_LIST_HEAD(&clp->cl_delegations);
-	INIT_LIST_HEAD(&clp->cl_state_owners);
-	INIT_LIST_HEAD(&clp->cl_unused);
 	spin_lock_init(&clp->cl_lock);
 	INIT_DELAYED_WORK(&clp->cl_renewd, nfs4_renew_state);
 	rpc_init_wait_queue(&clp->cl_rpcwaitq, "NFS client");
@@ -154,9 +143,6 @@ error_3:
 	if (__test_and_clear_bit(NFS_CS_CALLBACK, &clp->cl_res_state))
 		nfs_callback_down();
 error_2:
-	rpciod_down();
-	__clear_bit(NFS_CS_RPCIOD, &clp->cl_res_state);
-error_1:
 	kfree(clp);
 error_0:
 	return NULL;
@@ -167,16 +153,7 @@ static void nfs4_shutdown_client(struct nfs_client *clp)
 #ifdef CONFIG_NFS_V4
 	if (__test_and_clear_bit(NFS_CS_RENEWD, &clp->cl_res_state))
 		nfs4_kill_renewd(clp);
-	while (!list_empty(&clp->cl_unused)) {
-		struct nfs4_state_owner *sp;
-
-		sp = list_entry(clp->cl_unused.next,
-				struct nfs4_state_owner,
-				so_list);
-		list_del(&sp->so_list);
-		kfree(sp);
-	}
-	BUG_ON(!list_empty(&clp->cl_state_owners));
+	BUG_ON(!RB_EMPTY_ROOT(&clp->cl_state_owners));
 	if (__test_and_clear_bit(NFS_CS_IDMAP, &clp->cl_res_state))
 		nfs_idmap_delete(clp);
 #endif
@@ -198,9 +175,6 @@ static void nfs_free_client(struct nfs_client *clp)
 	if (__test_and_clear_bit(NFS_CS_CALLBACK, &clp->cl_res_state))
 		nfs_callback_down();
 
-	if (__test_and_clear_bit(NFS_CS_RPCIOD, &clp->cl_res_state))
-		rpciod_down();
-
 	kfree(clp->cl_hostname);
 	kfree(clp);
 

fs/nfs/delegation.c

@@ -27,6 +27,13 @@ static void nfs_free_delegation(struct nfs_delegation *delegation)
 	kfree(delegation);
 }
 
+static void nfs_free_delegation_callback(struct rcu_head *head)
+{
+	struct nfs_delegation *delegation = container_of(head, struct nfs_delegation, rcu);
+
+	nfs_free_delegation(delegation);
+}
+
 static int nfs_delegation_claim_locks(struct nfs_open_context *ctx, struct nfs4_state *state)
 {
 	struct inode *inode = state->inode;
@@ -57,7 +64,7 @@ out_err:
 	return status;
 }
 
-static void nfs_delegation_claim_opens(struct inode *inode)
+static void nfs_delegation_claim_opens(struct inode *inode, const nfs4_stateid *stateid)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 	struct nfs_open_context *ctx;
@@ -72,9 +79,11 @@ again:
 			continue;
 		if (!test_bit(NFS_DELEGATED_STATE, &state->flags))
 			continue;
+		if (memcmp(state->stateid.data, stateid->data, sizeof(state->stateid.data)) != 0)
+			continue;
 		get_nfs_open_context(ctx);
 		spin_unlock(&inode->i_lock);
-		err = nfs4_open_delegation_recall(ctx->dentry, state);
+		err = nfs4_open_delegation_recall(ctx, state, stateid);
 		if (err >= 0)
 			err = nfs_delegation_claim_locks(ctx, state);
 		put_nfs_open_context(ctx);
@@ -115,10 +124,6 @@ int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct
 	struct nfs_delegation *delegation;
 	int status = 0;
 
-	/* Ensure we first revalidate the attributes and page cache! */
-	if ((nfsi->cache_validity & (NFS_INO_REVAL_PAGECACHE|NFS_INO_INVALID_ATTR)))
-		__nfs_revalidate_inode(NFS_SERVER(inode), inode);
-
 	delegation = kmalloc(sizeof(*delegation), GFP_KERNEL);
 	if (delegation == NULL)
 		return -ENOMEM;
@@ -131,10 +136,10 @@ int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct
 	delegation->inode = inode;
 
 	spin_lock(&clp->cl_lock);
-	if (nfsi->delegation == NULL) {
-		list_add(&delegation->super_list, &clp->cl_delegations);
-		nfsi->delegation = delegation;
+	if (rcu_dereference(nfsi->delegation) == NULL) {
+		list_add_rcu(&delegation->super_list, &clp->cl_delegations);
 		nfsi->delegation_state = delegation->type;
+		rcu_assign_pointer(nfsi->delegation, delegation);
 		delegation = NULL;
 	} else {
 		if (memcmp(&delegation->stateid, &nfsi->delegation->stateid,
@@ -145,6 +150,12 @@ int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct
 			status = -EIO;
 		}
 	}
+
+	/* Ensure we revalidate the attributes and page cache! */
+	spin_lock(&inode->i_lock);
+	nfsi->cache_validity |= NFS_INO_REVAL_FORCED;
+	spin_unlock(&inode->i_lock);
+
 	spin_unlock(&clp->cl_lock);
 	kfree(delegation);
 	return status;
@@ -155,7 +166,7 @@ static int nfs_do_return_delegation(struct inode *inode, struct nfs_delegation *
 	int res = 0;
 
 	res = nfs4_proc_delegreturn(inode, delegation->cred, &delegation->stateid);
-	nfs_free_delegation(delegation);
+	call_rcu(&delegation->rcu, nfs_free_delegation_callback);
 	return res;
 }
 
@@ -170,33 +181,55 @@ static void nfs_msync_inode(struct inode *inode)
 /*
  * Basic procedure for returning a delegation to the server
  */
-int __nfs_inode_return_delegation(struct inode *inode)
+static int __nfs_inode_return_delegation(struct inode *inode, struct nfs_delegation *delegation)
 {
 	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
 	struct nfs_inode *nfsi = NFS_I(inode);
-	struct nfs_delegation *delegation;
-	int res = 0;
 
 	nfs_msync_inode(inode);
 	down_read(&clp->cl_sem);
 	/* Guard against new delegated open calls */
 	down_write(&nfsi->rwsem);
-	spin_lock(&clp->cl_lock);
-	delegation = nfsi->delegation;
-	if (delegation != NULL) {
-		list_del_init(&delegation->super_list);
-		nfsi->delegation = NULL;
-		nfsi->delegation_state = 0;
-	}
-	spin_unlock(&clp->cl_lock);
-	nfs_delegation_claim_opens(inode);
+	nfs_delegation_claim_opens(inode, &delegation->stateid);
 	up_write(&nfsi->rwsem);
 	up_read(&clp->cl_sem);
 	nfs_msync_inode(inode);
 
-	if (delegation != NULL)
-		res = nfs_do_return_delegation(inode, delegation);
-	return res;
+	return nfs_do_return_delegation(inode, delegation);
+}
+
+static struct nfs_delegation *nfs_detach_delegation_locked(struct nfs_inode *nfsi, const nfs4_stateid *stateid)
+{
+	struct nfs_delegation *delegation = rcu_dereference(nfsi->delegation);
+
+	if (delegation == NULL)
+		goto nomatch;
+	if (stateid != NULL && memcmp(delegation->stateid.data, stateid->data,
+				sizeof(delegation->stateid.data)) != 0)
+		goto nomatch;
+	list_del_rcu(&delegation->super_list);
+	nfsi->delegation_state = 0;
+	rcu_assign_pointer(nfsi->delegation, NULL);
+	return delegation;
+nomatch:
+	return NULL;
+}
+
+int nfs_inode_return_delegation(struct inode *inode)
+{
+	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
+	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs_delegation *delegation;
+	int err = 0;
+
+	if (rcu_dereference(nfsi->delegation) != NULL) {
+		spin_lock(&clp->cl_lock);
+		delegation = nfs_detach_delegation_locked(nfsi, NULL);
+		spin_unlock(&clp->cl_lock);
+		if (delegation != NULL)
+			err = __nfs_inode_return_delegation(inode, delegation);
+	}
+	return err;
 }
 
 /*
@@ -211,19 +244,23 @@ void nfs_return_all_delegations(struct super_block *sb)
 	if (clp == NULL)
 		return;
 restart:
-	spin_lock(&clp->cl_lock);
-	list_for_each_entry(delegation, &clp->cl_delegations, super_list) {
+	rcu_read_lock();
+	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
 		if (delegation->inode->i_sb != sb)
 			continue;
 		inode = igrab(delegation->inode);
 		if (inode == NULL)
 			continue;
+		spin_lock(&clp->cl_lock);
+		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
 		spin_unlock(&clp->cl_lock);
-		nfs_inode_return_delegation(inode);
+		rcu_read_unlock();
+		if (delegation != NULL)
+			__nfs_inode_return_delegation(inode, delegation);
 		iput(inode);
 		goto restart;
 	}
-	spin_unlock(&clp->cl_lock);
+	rcu_read_unlock();
 }
 
 static int nfs_do_expire_all_delegations(void *ptr)
@@ -234,22 +271,26 @@ static int nfs_do_expire_all_delegations(void *ptr)
 
 	allow_signal(SIGKILL);
 restart:
-	spin_lock(&clp->cl_lock);
 	if (test_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state) != 0)
 		goto out;
 	if (test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) == 0)
 		goto out;
-	list_for_each_entry(delegation, &clp->cl_delegations, super_list) {
+	rcu_read_lock();
+	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
 		inode = igrab(delegation->inode);
 		if (inode == NULL)
 			continue;
+		spin_lock(&clp->cl_lock);
+		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
 		spin_unlock(&clp->cl_lock);
-		nfs_inode_return_delegation(inode);
+		rcu_read_unlock();
+		if (delegation)
+			__nfs_inode_return_delegation(inode, delegation);
 		iput(inode);
 		goto restart;
 	}
+	rcu_read_unlock();
 out:
-	spin_unlock(&clp->cl_lock);
 	nfs_put_client(clp);
 	module_put_and_exit(0);
 }
@@ -280,17 +321,21 @@ void nfs_handle_cb_pathdown(struct nfs_client *clp)
 	if (clp == NULL)
 		return;
 restart:
-	spin_lock(&clp->cl_lock);
-	list_for_each_entry(delegation, &clp->cl_delegations, super_list) {
+	rcu_read_lock();
+	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
 		inode = igrab(delegation->inode);
 		if (inode == NULL)
 			continue;
+		spin_lock(&clp->cl_lock);
+		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
 		spin_unlock(&clp->cl_lock);
-		nfs_inode_return_delegation(inode);
+		rcu_read_unlock();
+		if (delegation != NULL)
+			__nfs_inode_return_delegation(inode, delegation);
 		iput(inode);
 		goto restart;
 	}
-	spin_unlock(&clp->cl_lock);
+	rcu_read_unlock();
 }
 
 struct recall_threadargs {
@@ -316,21 +361,14 @@ static int recall_thread(void *data)
 	down_read(&clp->cl_sem);
 	down_write(&nfsi->rwsem);
 	spin_lock(&clp->cl_lock);
-	delegation = nfsi->delegation;
-	if (delegation != NULL && memcmp(delegation->stateid.data,
-				args->stateid->data,
-				sizeof(delegation->stateid.data)) == 0) {
-		list_del_init(&delegation->super_list);
-		nfsi->delegation = NULL;
-		nfsi->delegation_state = 0;
+	delegation = nfs_detach_delegation_locked(nfsi, args->stateid);
+	if (delegation != NULL)
 		args->result = 0;
-	} else {
-		delegation = NULL;
+	else
 		args->result = -ENOENT;
-	}
 	spin_unlock(&clp->cl_lock);
 	complete(&args->started);
-	nfs_delegation_claim_opens(inode);
+	nfs_delegation_claim_opens(inode, args->stateid);
 	up_write(&nfsi->rwsem);
 	up_read(&clp->cl_sem);
 	nfs_msync_inode(inode);
@@ -371,14 +409,14 @@ struct inode *nfs_delegation_find_inode(struct nfs_client *clp, const struct nfs
 {
 	struct nfs_delegation *delegation;
 	struct inode *res = NULL;
-	spin_lock(&clp->cl_lock);
-	list_for_each_entry(delegation, &clp->cl_delegations, super_list) {
+	rcu_read_lock();
+	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
 		if (nfs_compare_fh(fhandle, &NFS_I(delegation->inode)->fh) == 0) {
 			res = igrab(delegation->inode);
 			break;
 		}
 	}
-	spin_unlock(&clp->cl_lock);
+	rcu_read_unlock();
 	return res;
 }
 
@@ -388,10 +426,10 @@ struct inode *nfs_delegation_find_inode(struct nfs_client *clp, const struct nfs
 void nfs_delegation_mark_reclaim(struct nfs_client *clp)
 {
 	struct nfs_delegation *delegation;
-	spin_lock(&clp->cl_lock);
-	list_for_each_entry(delegation, &clp->cl_delegations, super_list)
+	rcu_read_lock();
+	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list)
 		delegation->flags |= NFS_DELEGATION_NEED_RECLAIM;
-	spin_unlock(&clp->cl_lock);
+	rcu_read_unlock();
 }
 
 /*
@@ -399,39 +437,35 @@ void nfs_delegation_mark_reclaim(struct nfs_client *clp)
  */
 void nfs_delegation_reap_unclaimed(struct nfs_client *clp)
 {
-	struct nfs_delegation *delegation, *n;
-	LIST_HEAD(head);
-	spin_lock(&clp->cl_lock);
-	list_for_each_entry_safe(delegation, n, &clp->cl_delegations, super_list) {
+	struct nfs_delegation *delegation;
+restart:
+	rcu_read_lock();
+	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
 		if ((delegation->flags & NFS_DELEGATION_NEED_RECLAIM) == 0)
 			continue;
-		list_move(&delegation->super_list, &head);
-		NFS_I(delegation->inode)->delegation = NULL;
-		NFS_I(delegation->inode)->delegation_state = 0;
-	}
-	spin_unlock(&clp->cl_lock);
-	while(!list_empty(&head)) {
-		delegation = list_entry(head.next, struct nfs_delegation, super_list);
-		list_del(&delegation->super_list);
-		nfs_free_delegation(delegation);
+		spin_lock(&clp->cl_lock);
+		delegation = nfs_detach_delegation_locked(NFS_I(delegation->inode), NULL);
+		spin_unlock(&clp->cl_lock);
+		rcu_read_unlock();
+		if (delegation != NULL)
+			call_rcu(&delegation->rcu, nfs_free_delegation_callback);
+		goto restart;
 	}
+	rcu_read_unlock();
 }
 
 int nfs4_copy_delegation_stateid(nfs4_stateid *dst, struct inode *inode)
 {
-	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
 	struct nfs_inode *nfsi = NFS_I(inode);
 	struct nfs_delegation *delegation;
-	int res = 0;
+	int ret = 0;
 
-	if (nfsi->delegation_state == 0)
-		return 0;
-	spin_lock(&clp->cl_lock);
-	delegation = nfsi->delegation;
+	rcu_read_lock();
+	delegation = rcu_dereference(nfsi->delegation);
 	if (delegation != NULL) {
 		memcpy(dst->data, delegation->stateid.data, sizeof(dst->data));
-		res = 1;
+		ret = 1;
 	}
-	spin_unlock(&clp->cl_lock);
-	return res;
+	rcu_read_unlock();
+	return ret;
 }

fs/nfs/delegation.h

@@ -22,11 +22,12 @@ struct nfs_delegation {
 	long flags;
 	loff_t maxsize;
 	__u64 change_attr;
+	struct rcu_head rcu;
 };
 
 int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res);
 void nfs_inode_reclaim_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res);
-int __nfs_inode_return_delegation(struct inode *inode);
+int nfs_inode_return_delegation(struct inode *inode);
 int nfs_async_inode_return_delegation(struct inode *inode, const nfs4_stateid *stateid);
 
 struct inode *nfs_delegation_find_inode(struct nfs_client *clp, const struct nfs_fh *fhandle);
@@ -39,27 +40,24 @@ 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 nfs4_open_delegation_recall(struct dentry *dentry, struct nfs4_state *state);
+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_copy_delegation_stateid(nfs4_stateid *dst, struct inode *inode);
 
 static inline int nfs_have_delegation(struct inode *inode, int flags)
 {
+	struct nfs_delegation *delegation;
+	int ret = 0;
+
 	flags &= FMODE_READ|FMODE_WRITE;
-	smp_rmb();
-	if ((NFS_I(inode)->delegation_state & flags) == flags)
-		return 1;
-	return 0;
+	rcu_read_lock();
+	delegation = rcu_dereference(NFS_I(inode)->delegation);
+	if (delegation != NULL && (delegation->type & flags) == flags)
+		ret = 1;
+	rcu_read_unlock();
+	return ret;
 }
 
-static inline int nfs_inode_return_delegation(struct inode *inode)
-{
-	int err = 0;
-
-	if (NFS_I(inode)->delegation != NULL)
-		err = __nfs_inode_return_delegation(inode);
-	return err;
-}
 #else
 static inline int nfs_have_delegation(struct inode *inode, int flags)
 {

fs/nfs/dir.c

@@ -897,14 +897,13 @@ int nfs_is_exclusive_create(struct inode *dir, struct nameidata *nd)
 	return (nd->intent.open.flags & O_EXCL) != 0;
 }
 
-static inline int nfs_reval_fsid(struct vfsmount *mnt, struct inode *dir,
-				 struct nfs_fh *fh, struct nfs_fattr *fattr)
+static inline int nfs_reval_fsid(struct inode *dir, const struct nfs_fattr *fattr)
 {
 	struct nfs_server *server = NFS_SERVER(dir);
 
 	if (!nfs_fsid_equal(&server->fsid, &fattr->fsid))
-		/* Revalidate fsid on root dir */
-		return __nfs_revalidate_inode(server, mnt->mnt_root->d_inode);
+		/* Revalidate fsid using the parent directory */
+		return __nfs_revalidate_inode(server, dir);
 	return 0;
 }
 
@@ -946,7 +945,7 @@ static struct dentry *nfs_lookup(struct inode *dir, struct dentry * dentry, stru
 		res = ERR_PTR(error);
 		goto out_unlock;
 	}
-	error = nfs_reval_fsid(nd->mnt, dir, &fhandle, &fattr);
+	error = nfs_reval_fsid(dir, &fattr);
 	if (error < 0) {
 		res = ERR_PTR(error);
 		goto out_unlock;
@@ -1244,7 +1243,7 @@ static int nfs_create(struct inode *dir, struct dentry *dentry, int mode,
 	attr.ia_mode = mode;
 	attr.ia_valid = ATTR_MODE;
 
-	if (nd && (nd->flags & LOOKUP_CREATE))
+	if ((nd->flags & LOOKUP_CREATE) != 0)
 		open_flags = nd->intent.open.flags;
 
 	lock_kernel();
@@ -1535,7 +1534,7 @@ static int nfs_symlink(struct inode *dir, struct dentry *dentry, const char *sym
 
 	lock_kernel();
 
-	page = alloc_page(GFP_KERNEL);
+	page = alloc_page(GFP_HIGHUSER);
 	if (!page) {
 		unlock_kernel();
 		return -ENOMEM;
@@ -1744,8 +1743,8 @@ int nfs_access_cache_shrinker(int nr_to_scan, gfp_t gfp_mask)
 	struct nfs_inode *nfsi;
 	struct nfs_access_entry *cache;
 
-	spin_lock(&nfs_access_lru_lock);
 restart:
+	spin_lock(&nfs_access_lru_lock);
 	list_for_each_entry(nfsi, &nfs_access_lru_list, access_cache_inode_lru) {
 		struct inode *inode;
 
@@ -1770,6 +1769,7 @@ remove_lru_entry:
 			clear_bit(NFS_INO_ACL_LRU_SET, &nfsi->flags);
 		}
 		spin_unlock(&inode->i_lock);
+		spin_unlock(&nfs_access_lru_lock);
 		iput(inode);
 		goto restart;
 	}

fs/nfs/direct.c

@@ -266,7 +266,7 @@ static const struct rpc_call_ops nfs_read_direct_ops = {
 static ssize_t nfs_direct_read_schedule(struct nfs_direct_req *dreq, unsigned long user_addr, size_t count, loff_t pos)
 {
 	struct nfs_open_context *ctx = dreq->ctx;
-	struct inode *inode = ctx->dentry->d_inode;
+	struct inode *inode = ctx->path.dentry->d_inode;
 	size_t rsize = NFS_SERVER(inode)->rsize;
 	unsigned int pgbase;
 	int result;
@@ -295,9 +295,14 @@ static ssize_t nfs_direct_read_schedule(struct nfs_direct_req *dreq, unsigned lo
 			break;
 		}
 		if ((unsigned)result < data->npages) {
-			nfs_direct_release_pages(data->pagevec, result);
-			nfs_readdata_release(data);
-			break;
+			bytes = result * PAGE_SIZE;
+			if (bytes <= pgbase) {
+				nfs_direct_release_pages(data->pagevec, result);
+				nfs_readdata_release(data);
+				break;
+			}
+			bytes -= pgbase;
+			data->npages = result;
 		}
 
 		get_dreq(dreq);
@@ -601,7 +606,7 @@ static const struct rpc_call_ops nfs_write_direct_ops = {
 static ssize_t nfs_direct_write_schedule(struct nfs_direct_req *dreq, unsigned long user_addr, size_t count, loff_t pos, int sync)
 {
 	struct nfs_open_context *ctx = dreq->ctx;
-	struct inode *inode = ctx->dentry->d_inode;
+	struct inode *inode = ctx->path.dentry->d_inode;
 	size_t wsize = NFS_SERVER(inode)->wsize;
 	unsigned int pgbase;
 	int result;
@@ -630,9 +635,14 @@ static ssize_t nfs_direct_write_schedule(struct nfs_direct_req *dreq, unsigned l
 			break;
 		}
 		if ((unsigned)result < data->npages) {
-			nfs_direct_release_pages(data->pagevec, result);
-			nfs_writedata_release(data);
-			break;
+			bytes = result * PAGE_SIZE;
+			if (bytes <= pgbase) {
+				nfs_direct_release_pages(data->pagevec, result);
+				nfs_writedata_release(data);
+				break;
+			}
+			bytes -= pgbase;
+			data->npages = result;
 		}
 
 		get_dreq(dreq);
@@ -763,10 +773,8 @@ ssize_t nfs_file_direct_read(struct kiocb *iocb, const struct iovec *iov,
 		(unsigned long) count, (long long) pos);
 
 	if (nr_segs != 1)
-		return -EINVAL;
-
-	if (count < 0)
 		goto out;
+
 	retval = -EFAULT;
 	if (!access_ok(VERIFY_WRITE, buf, count))
 		goto out;
@@ -814,7 +822,7 @@ out:
 ssize_t nfs_file_direct_write(struct kiocb *iocb, const struct iovec *iov,
 				unsigned long nr_segs, loff_t pos)
 {
-	ssize_t retval;
+	ssize_t retval = -EINVAL;
 	struct file *file = iocb->ki_filp;
 	struct address_space *mapping = file->f_mapping;
 	/* XXX: temporary */
@@ -827,7 +835,7 @@ ssize_t nfs_file_direct_write(struct kiocb *iocb, const struct iovec *iov,
 		(unsigned long) count, (long long) pos);
 
 	if (nr_segs != 1)
-		return -EINVAL;
+		goto out;
 
 	retval = generic_write_checks(file, &pos, &count, 0);
 	if (retval)

fs/nfs/inode.c

@@ -461,14 +461,14 @@ static struct nfs_open_context *alloc_nfs_open_context(struct vfsmount *mnt, str
 
 	ctx = kmalloc(sizeof(*ctx), GFP_KERNEL);
 	if (ctx != NULL) {
-		atomic_set(&ctx->count, 1);
-		ctx->dentry = dget(dentry);
-		ctx->vfsmnt = mntget(mnt);
+		ctx->path.dentry = dget(dentry);
+		ctx->path.mnt = mntget(mnt);
 		ctx->cred = get_rpccred(cred);
 		ctx->state = NULL;
 		ctx->lockowner = current->files;
 		ctx->error = 0;
 		ctx->dir_cookie = 0;
+		kref_init(&ctx->kref);
 	}
 	return ctx;
 }
@@ -476,27 +476,33 @@ static struct nfs_open_context *alloc_nfs_open_context(struct vfsmount *mnt, str
 struct nfs_open_context *get_nfs_open_context(struct nfs_open_context *ctx)
 {
 	if (ctx != NULL)
-		atomic_inc(&ctx->count);
+		kref_get(&ctx->kref);
 	return ctx;
 }
 
-void put_nfs_open_context(struct nfs_open_context *ctx)
+static void nfs_free_open_context(struct kref *kref)
 {
-	if (atomic_dec_and_test(&ctx->count)) {
-		if (!list_empty(&ctx->list)) {
-			struct inode *inode = ctx->dentry->d_inode;
-			spin_lock(&inode->i_lock);
-			list_del(&ctx->list);
-			spin_unlock(&inode->i_lock);
-		}
-		if (ctx->state != NULL)
-			nfs4_close_state(ctx->state, ctx->mode);
-		if (ctx->cred != NULL)
-			put_rpccred(ctx->cred);
-		dput(ctx->dentry);
-		mntput(ctx->vfsmnt);
-		kfree(ctx);
+	struct nfs_open_context *ctx = container_of(kref,
+			struct nfs_open_context, kref);
+
+	if (!list_empty(&ctx->list)) {
+		struct inode *inode = ctx->path.dentry->d_inode;
+		spin_lock(&inode->i_lock);
+		list_del(&ctx->list);
+		spin_unlock(&inode->i_lock);
 	}
+	if (ctx->state != NULL)
+		nfs4_close_state(&ctx->path, ctx->state, ctx->mode);
+	if (ctx->cred != NULL)
+		put_rpccred(ctx->cred);
+	dput(ctx->path.dentry);
+	mntput(ctx->path.mnt);
+	kfree(ctx);
+}
+
+void put_nfs_open_context(struct nfs_open_context *ctx)
+{
+	kref_put(&ctx->kref, nfs_free_open_context);
 }
 
 /*
@@ -961,8 +967,8 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 		goto out_changed;
 
 	server = NFS_SERVER(inode);
-	/* Update the fsid if and only if this is the root directory */
-	if (inode == inode->i_sb->s_root->d_inode
+	/* Update the fsid? */
+	if (S_ISDIR(inode->i_mode)
 			&& !nfs_fsid_equal(&server->fsid, &fattr->fsid))
 		server->fsid = fattr->fsid;
 
@@ -1066,8 +1072,10 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 		invalid &= ~NFS_INO_INVALID_DATA;
 	if (data_stable)
 		invalid &= ~(NFS_INO_INVALID_ATTR|NFS_INO_INVALID_ATIME|NFS_INO_REVAL_PAGECACHE);
-	if (!nfs_have_delegation(inode, FMODE_READ))
+	if (!nfs_have_delegation(inode, FMODE_READ) ||
+			(nfsi->cache_validity & NFS_INO_REVAL_FORCED))
 		nfsi->cache_validity |= invalid;
+	nfsi->cache_validity &= ~NFS_INO_REVAL_FORCED;
 
 	return 0;
  out_changed:
@@ -1103,27 +1111,10 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
  */
 void nfs4_clear_inode(struct inode *inode)
 {
-	struct nfs_inode *nfsi = NFS_I(inode);
-
 	/* If we are holding a delegation, return it! */
 	nfs_inode_return_delegation(inode);
 	/* First call standard NFS clear_inode() code */
 	nfs_clear_inode(inode);
-	/* Now clear out any remaining state */
-	while (!list_empty(&nfsi->open_states)) {
-		struct nfs4_state *state;
-		
-		state = list_entry(nfsi->open_states.next,
-				struct nfs4_state,
-				inode_states);
-		dprintk("%s(%s/%Ld): found unclaimed NFSv4 state %p\n",
-				__FUNCTION__,
-				inode->i_sb->s_id,
-				(long long)NFS_FILEID(inode),
-				state);
-		BUG_ON(atomic_read(&state->count) != 1);
-		nfs4_close_state(state, state->state);
-	}
 }
 #endif
 
@@ -1165,15 +1156,11 @@ static void init_once(void * foo, struct kmem_cache * cachep, unsigned long flag
 	struct nfs_inode *nfsi = (struct nfs_inode *) foo;
 
 	inode_init_once(&nfsi->vfs_inode);
-	spin_lock_init(&nfsi->req_lock);
-	INIT_LIST_HEAD(&nfsi->dirty);
-	INIT_LIST_HEAD(&nfsi->commit);
 	INIT_LIST_HEAD(&nfsi->open_files);
 	INIT_LIST_HEAD(&nfsi->access_cache_entry_lru);
 	INIT_LIST_HEAD(&nfsi->access_cache_inode_lru);
 	INIT_RADIX_TREE(&nfsi->nfs_page_tree, GFP_ATOMIC);
 	atomic_set(&nfsi->data_updates, 0);
-	nfsi->ndirty = 0;
 	nfsi->ncommit = 0;
 	nfsi->npages = 0;
 	nfs4_init_once(nfsi);

fs/nfs/internal.h

@@ -183,9 +183,9 @@ unsigned long nfs_block_bits(unsigned long bsize, unsigned char *nrbitsp)
 /*
  * Calculate the number of 512byte blocks used.
  */
-static inline unsigned long nfs_calc_block_size(u64 tsize)
+static inline blkcnt_t nfs_calc_block_size(u64 tsize)
 {
-	loff_t used = (tsize + 511) >> 9;
+	blkcnt_t used = (tsize + 511) >> 9;
 	return (used > ULONG_MAX) ? ULONG_MAX : used;
 }
 

fs/nfs/mount_clnt.c

@@ -1,7 +1,5 @@
 /*
- * linux/fs/nfs/mount_clnt.c
- *
- * MOUNT client to support NFSroot.
+ * In-kernel MOUNT protocol client
  *
  * Copyright (C) 1997, Olaf Kirch <okir@monad.swb.de>
  */
@@ -18,33 +16,31 @@
 #include <linux/nfs_fs.h>
 
 #ifdef RPC_DEBUG
-# define NFSDBG_FACILITY	NFSDBG_ROOT
+# define NFSDBG_FACILITY	NFSDBG_MOUNT
 #endif
 
-/*
-#define MOUNT_PROGRAM		100005
-#define MOUNT_VERSION		1
-#define MOUNT_MNT		1
-#define MOUNT_UMNT		3
- */
-
-static struct rpc_clnt *	mnt_create(char *, struct sockaddr_in *,
-								int, int);
 static struct rpc_program	mnt_program;
 
 struct mnt_fhstatus {
-	unsigned int		status;
-	struct nfs_fh *		fh;
+	u32 status;
+	struct nfs_fh *fh;
 };
 
-/*
- * Obtain an NFS file handle for the given host and path
+/**
+ * nfs_mount - Obtain an NFS file handle for the given host and path
+ * @addr: pointer to server's address
+ * @len: size of server's address
+ * @hostname: name of server host, or NULL
+ * @path: pointer to string containing export path to mount
+ * @version: mount version to use for this request
+ * @protocol: transport protocol to use for thie request
+ * @fh: pointer to location to place returned file handle
+ *
+ * Uses default timeout parameters specified by underlying transport.
  */
-int
-nfsroot_mount(struct sockaddr_in *addr, char *path, struct nfs_fh *fh,
-		int version, int protocol)
+int nfs_mount(struct sockaddr *addr, size_t len, char *hostname, char *path,
+	      int version, int protocol, struct nfs_fh *fh)
 {
-	struct rpc_clnt		*mnt_clnt;
 	struct mnt_fhstatus	result = {
 		.fh		= fh
 	};
@@ -52,16 +48,25 @@ nfsroot_mount(struct sockaddr_in *addr, char *path, struct nfs_fh *fh,
 		.rpc_argp	= path,
 		.rpc_resp	= &result,
 	};
-	char			hostname[32];
+	struct rpc_create_args args = {
+		.protocol	= protocol,
+		.address	= addr,
+		.addrsize	= len,
+		.servername	= hostname,
+		.program	= &mnt_program,
+		.version	= version,
+		.authflavor	= RPC_AUTH_UNIX,
+		.flags		= RPC_CLNT_CREATE_INTR,
+	};
+	struct rpc_clnt		*mnt_clnt;
 	int			status;
 
-	dprintk("NFS:      nfs_mount(%08x:%s)\n",
-			(unsigned)ntohl(addr->sin_addr.s_addr), path);
+	dprintk("NFS: sending MNT request for %s:%s\n",
+		(hostname ? hostname : "server"), path);
 
-	sprintf(hostname, "%u.%u.%u.%u", NIPQUAD(addr->sin_addr.s_addr));
-	mnt_clnt = mnt_create(hostname, addr, version, protocol);
+	mnt_clnt = rpc_create(&args);
 	if (IS_ERR(mnt_clnt))
-		return PTR_ERR(mnt_clnt);
+		goto out_clnt_err;
 
 	if (version == NFS_MNT3_VERSION)
 		msg.rpc_proc = &mnt_clnt->cl_procinfo[MOUNTPROC3_MNT];
@@ -69,33 +74,39 @@ nfsroot_mount(struct sockaddr_in *addr, char *path, struct nfs_fh *fh,
 		msg.rpc_proc = &mnt_clnt->cl_procinfo[MNTPROC_MNT];
 
 	status = rpc_call_sync(mnt_clnt, &msg, 0);
-	return status < 0? status : (result.status? -EACCES : 0);
-}
+	rpc_shutdown_client(mnt_clnt);
 
-static struct rpc_clnt *
-mnt_create(char *hostname, struct sockaddr_in *srvaddr, int version,
-		int protocol)
-{
-	struct rpc_create_args args = {
-		.protocol	= protocol,
-		.address	= (struct sockaddr *)srvaddr,
-		.addrsize	= sizeof(*srvaddr),
-		.servername	= hostname,
-		.program	= &mnt_program,
-		.version	= version,
-		.authflavor	= RPC_AUTH_UNIX,
-		.flags		= (RPC_CLNT_CREATE_ONESHOT |
-				   RPC_CLNT_CREATE_INTR),
-	};
+	if (status < 0)
+		goto out_call_err;
+	if (result.status != 0)
+		goto out_mnt_err;
+
+	dprintk("NFS: MNT request succeeded\n");
+	status = 0;
+
+out:
+	return status;
+
+out_clnt_err:
+	status = PTR_ERR(mnt_clnt);
+	dprintk("NFS: failed to create RPC client, status=%d\n", status);
+	goto out;
+
+out_call_err:
+	dprintk("NFS: failed to start MNT request, status=%d\n", status);
+	goto out;
 
-	return rpc_create(&args);
+out_mnt_err:
+	dprintk("NFS: MNT server returned result %d\n", result.status);
+	status = -EACCES;
+	goto out;
 }
 
 /*
  * XDR encode/decode functions for MOUNT
  */
-static int
-xdr_encode_dirpath(struct rpc_rqst *req, __be32 *p, const char *path)
+static int xdr_encode_dirpath(struct rpc_rqst *req, __be32 *p,
+			      const char *path)
 {
 	p = xdr_encode_string(p, path);
 
@@ -103,8 +114,8 @@ xdr_encode_dirpath(struct rpc_rqst *req, __be32 *p, const char *path)
 	return 0;
 }
 
-static int
-xdr_decode_fhstatus(struct rpc_rqst *req, __be32 *p, struct mnt_fhstatus *res)
+static int xdr_decode_fhstatus(struct rpc_rqst *req, __be32 *p,
+			       struct mnt_fhstatus *res)
 {
 	struct nfs_fh *fh = res->fh;
 
@@ -115,8 +126,8 @@ xdr_decode_fhstatus(struct rpc_rqst *req, __be32 *p, struct mnt_fhstatus *res)
 	return 0;
 }
 
-static int
-xdr_decode_fhstatus3(struct rpc_rqst *req, __be32 *p, struct mnt_fhstatus *res)
+static int xdr_decode_fhstatus3(struct rpc_rqst *req, __be32 *p,
+				struct mnt_fhstatus *res)
 {
 	struct nfs_fh *fh = res->fh;
 
@@ -135,53 +146,53 @@ xdr_decode_fhstatus3(struct rpc_rqst *req, __be32 *p, struct mnt_fhstatus *res)
 #define MNT_fhstatus_sz		(1 + 8)
 #define MNT_fhstatus3_sz	(1 + 16)
 
-static struct rpc_procinfo	mnt_procedures[] = {
-[MNTPROC_MNT] = {
-	  .p_proc		= MNTPROC_MNT,
-	  .p_encode		= (kxdrproc_t) xdr_encode_dirpath,	
-	  .p_decode		= (kxdrproc_t) xdr_decode_fhstatus,
-	  .p_arglen		= MNT_dirpath_sz,
-	  .p_replen		= MNT_fhstatus_sz,
-	  .p_statidx		= MNTPROC_MNT,
-	  .p_name		= "MOUNT",
+static struct rpc_procinfo mnt_procedures[] = {
+	[MNTPROC_MNT] = {
+		.p_proc		= MNTPROC_MNT,
+		.p_encode	= (kxdrproc_t) xdr_encode_dirpath,
+		.p_decode	= (kxdrproc_t) xdr_decode_fhstatus,
+		.p_arglen	= MNT_dirpath_sz,
+		.p_replen	= MNT_fhstatus_sz,
+		.p_statidx	= MNTPROC_MNT,
+		.p_name		= "MOUNT",
 	},
 };
 
 static struct rpc_procinfo mnt3_procedures[] = {
-[MOUNTPROC3_MNT] = {
-	  .p_proc		= MOUNTPROC3_MNT,
-	  .p_encode		= (kxdrproc_t) xdr_encode_dirpath,
-	  .p_decode		= (kxdrproc_t) xdr_decode_fhstatus3,
-	  .p_arglen		= MNT_dirpath_sz,
-	  .p_replen		= MNT_fhstatus3_sz,
-	  .p_statidx		= MOUNTPROC3_MNT,
-	  .p_name		= "MOUNT",
+	[MOUNTPROC3_MNT] = {
+		.p_proc		= MOUNTPROC3_MNT,
+		.p_encode	= (kxdrproc_t) xdr_encode_dirpath,
+		.p_decode	= (kxdrproc_t) xdr_decode_fhstatus3,
+		.p_arglen	= MNT_dirpath_sz,
+		.p_replen	= MNT_fhstatus3_sz,
+		.p_statidx	= MOUNTPROC3_MNT,
+		.p_name		= "MOUNT",
 	},
 };
 
 
-static struct rpc_version	mnt_version1 = {
-		.number		= 1,
-		.nrprocs 	= 2,
-		.procs 		= mnt_procedures
+static struct rpc_version mnt_version1 = {
+	.number		= 1,
+	.nrprocs	= 2,
+	.procs		= mnt_procedures,
 };
 
-static struct rpc_version       mnt_version3 = {
-		.number		= 3,
-		.nrprocs	= 2,
-		.procs		= mnt3_procedures
+static struct rpc_version mnt_version3 = {
+	.number		= 3,
+	.nrprocs	= 2,
+	.procs		= mnt3_procedures,
 };
 
-static struct rpc_version *	mnt_version[] = {
+static struct rpc_version *mnt_version[] = {
 	NULL,
 	&mnt_version1,
 	NULL,
 	&mnt_version3,
 };
 
-static struct rpc_stat		mnt_stats;
+static struct rpc_stat mnt_stats;
 
-static struct rpc_program	mnt_program = {
+static struct rpc_program mnt_program = {
 	.name		= "mount",
 	.number		= NFS_MNT_PROGRAM,
 	.nrvers		= ARRAY_SIZE(mnt_version),

fs/nfs/nfs2xdr.c

@@ -223,7 +223,7 @@ nfs_xdr_diropargs(struct rpc_rqst *req, __be32 *p, struct nfs_diropargs *args)
 static int
 nfs_xdr_readargs(struct rpc_rqst *req, __be32 *p, struct nfs_readargs *args)
 {
-	struct rpc_auth	*auth = req->rq_task->tk_auth;
+	struct rpc_auth	*auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	unsigned int replen;
 	u32 offset = (u32)args->offset;
 	u32 count = args->count;
@@ -380,7 +380,7 @@ static int
 nfs_xdr_readdirargs(struct rpc_rqst *req, __be32 *p, struct nfs_readdirargs *args)
 {
 	struct rpc_task	*task = req->rq_task;
-	struct rpc_auth	*auth = task->tk_auth;
+	struct rpc_auth	*auth = task->tk_msg.rpc_cred->cr_auth;
 	unsigned int replen;
 	u32 count = args->count;
 
@@ -541,7 +541,7 @@ nfs_xdr_diropres(struct rpc_rqst *req, __be32 *p, struct nfs_diropok *res)
 static int
 nfs_xdr_readlinkargs(struct rpc_rqst *req, __be32 *p, struct nfs_readlinkargs *args)
 {
-	struct rpc_auth *auth = req->rq_task->tk_auth;
+	struct rpc_auth	*auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	unsigned int replen;
 
 	p = xdr_encode_fhandle(p, args->fh);

fs/nfs/nfs3proc.c

@@ -335,9 +335,7 @@ again:
 		 * not sure this buys us anything (and I'd have
 		 * to revamp the NFSv3 XDR code) */
 		status = nfs3_proc_setattr(dentry, &fattr, sattr);
-		if (status == 0)
-			nfs_setattr_update_inode(dentry->d_inode, sattr);
-		nfs_refresh_inode(dentry->d_inode, &fattr);
+		nfs_post_op_update_inode(dentry->d_inode, &fattr);
 		dprintk("NFS reply setattr (post-create): %d\n", status);
 	}
 	if (status != 0)

fs/nfs/nfs3xdr.c

@@ -319,7 +319,7 @@ nfs3_xdr_accessargs(struct rpc_rqst *req, __be32 *p, struct nfs3_accessargs *arg
 static int
 nfs3_xdr_readargs(struct rpc_rqst *req, __be32 *p, struct nfs_readargs *args)
 {
-	struct rpc_auth	*auth = req->rq_task->tk_auth;
+	struct rpc_auth	*auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	unsigned int replen;
 	u32 count = args->count;
 
@@ -458,7 +458,7 @@ nfs3_xdr_linkargs(struct rpc_rqst *req, __be32 *p, struct nfs3_linkargs *args)
 static int
 nfs3_xdr_readdirargs(struct rpc_rqst *req, __be32 *p, struct nfs3_readdirargs *args)
 {
-	struct rpc_auth	*auth = req->rq_task->tk_auth;
+	struct rpc_auth	*auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	unsigned int replen;
 	u32 count = args->count;
 
@@ -643,7 +643,7 @@ static int
 nfs3_xdr_getaclargs(struct rpc_rqst *req, __be32 *p,
 		    struct nfs3_getaclargs *args)
 {
-	struct rpc_auth *auth = req->rq_task->tk_auth;
+	struct rpc_auth	*auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	unsigned int replen;
 
 	p = xdr_encode_fhandle(p, args->fh);
@@ -773,7 +773,7 @@ nfs3_xdr_accessres(struct rpc_rqst *req, __be32 *p, struct nfs3_accessres *res)
 static int
 nfs3_xdr_readlinkargs(struct rpc_rqst *req, __be32 *p, struct nfs3_readlinkargs *args)
 {
-	struct rpc_auth *auth = req->rq_task->tk_auth;
+	struct rpc_auth	*auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	unsigned int replen;
 
 	p = xdr_encode_fhandle(p, args->fh);

fs/nfs/nfs4_fs.h

@@ -70,19 +70,26 @@ 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
  * semantics by allowing the server to identify replayed requests.
  */
 struct nfs4_state_owner {
-	spinlock_t	     so_lock;
-	struct list_head     so_list;	 /* per-clientid list of state_owners */
+	struct nfs_unique_id so_owner_id;
 	struct nfs_client    *so_client;
-	u32                  so_id;      /* 32-bit identifier, unique */
-	atomic_t	     so_count;
+	struct nfs_server    *so_server;
+	struct rb_node	     so_client_node;
 
 	struct rpc_cred	     *so_cred;	 /* Associated cred */
+
+	spinlock_t	     so_lock;
+	atomic_t	     so_count;
 	struct list_head     so_states;
 	struct list_head     so_delegations;
 	struct nfs_seqid_counter so_seqid;
@@ -108,7 +115,7 @@ struct nfs4_lock_state {
 #define NFS_LOCK_INITIALIZED 1
 	int			ls_flags;
 	struct nfs_seqid_counter	ls_seqid;
-	u32			ls_id;
+	struct nfs_unique_id	ls_id;
 	nfs4_stateid		ls_stateid;
 	atomic_t		ls_count;
 };
@@ -116,7 +123,10 @@ struct nfs4_lock_state {
 /* bits for nfs4_state->flags */
 enum {
 	LK_STATE_IN_USE,
-	NFS_DELEGATED_STATE,
+	NFS_DELEGATED_STATE,		/* Current stateid is delegation */
+	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 */
 };
 
 struct nfs4_state {
@@ -130,11 +140,14 @@ struct nfs4_state {
 	unsigned long flags;		/* Do we hold any locks? */
 	spinlock_t state_lock;		/* Protects the lock_states list */
 
-	nfs4_stateid stateid;
+	seqlock_t seqlock;		/* Protects the stateid/open_stateid */
+	nfs4_stateid stateid;		/* Current stateid: may be delegation */
+	nfs4_stateid open_stateid;	/* OPEN stateid */
 
-	unsigned int n_rdonly;
-	unsigned int n_wronly;
-	unsigned int n_rdwr;
+	/* The following 3 fields are protected by owner->so_lock */
+	unsigned int n_rdonly;		/* Number of read-only references */
+	unsigned int n_wronly;		/* Number of write-only references */
+	unsigned int n_rdwr;		/* Number of read/write references */
 	int state;			/* State on the server (R,W, or RW) */
 	atomic_t count;
 };
@@ -165,7 +178,7 @@ extern int nfs4_proc_setclientid(struct nfs_client *, u32, unsigned short, struc
 extern int nfs4_proc_setclientid_confirm(struct nfs_client *, struct rpc_cred *);
 extern int nfs4_proc_async_renew(struct nfs_client *, struct rpc_cred *);
 extern int nfs4_proc_renew(struct nfs_client *, struct rpc_cred *);
-extern int nfs4_do_close(struct inode *inode, struct nfs4_state *state);
+extern int nfs4_do_close(struct path *path, struct nfs4_state *state);
 extern struct dentry *nfs4_atomic_open(struct inode *, struct dentry *, struct nameidata *);
 extern int nfs4_open_revalidate(struct inode *, struct dentry *, int, struct nameidata *);
 extern int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle);
@@ -189,14 +202,13 @@ extern void nfs4_renew_state(struct work_struct *);
 
 /* nfs4state.c */
 struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp);
-extern u32 nfs4_alloc_lockowner_id(struct nfs_client *);
 
 extern struct nfs4_state_owner * nfs4_get_state_owner(struct nfs_server *, struct rpc_cred *);
 extern void nfs4_put_state_owner(struct nfs4_state_owner *);
 extern void nfs4_drop_state_owner(struct nfs4_state_owner *);
 extern struct nfs4_state * nfs4_get_open_state(struct inode *, struct nfs4_state_owner *);
 extern void nfs4_put_open_state(struct nfs4_state *);
-extern void nfs4_close_state(struct nfs4_state *, mode_t);
+extern void nfs4_close_state(struct path *, struct nfs4_state *, mode_t);
 extern void nfs4_state_set_mode_locked(struct nfs4_state *, mode_t);
 extern void nfs4_schedule_state_recovery(struct nfs_client *);
 extern void nfs4_put_lock_state(struct nfs4_lock_state *lsp);
@@ -222,7 +234,7 @@ extern struct svc_version nfs4_callback_version1;
 
 #else
 
-#define nfs4_close_state(a, b) do { } while (0)
+#define nfs4_close_state(a, b, c) do { } while (0)
 
 #endif /* CONFIG_NFS_V4 */
 #endif /* __LINUX_FS_NFS_NFS4_FS.H */

fs/nfs/nfs4proc.c

@@ -65,6 +65,7 @@ static int nfs4_async_handle_error(struct rpc_task *, const struct nfs_server *)
 static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry);
 static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, struct nfs4_exception *exception);
 static int nfs4_wait_clnt_recover(struct rpc_clnt *clnt, struct nfs_client *clp);
+static int _nfs4_do_access(struct inode *inode, struct rpc_cred *cred, int openflags);
 
 /* Prevent leaks of NFSv4 errors into userland */
 int nfs4_map_errors(int err)
@@ -214,27 +215,39 @@ static void update_changeattr(struct inode *dir, struct nfs4_change_info *cinfo)
 }
 
 struct nfs4_opendata {
-	atomic_t count;
+	struct kref kref;
 	struct nfs_openargs o_arg;
 	struct nfs_openres o_res;
 	struct nfs_open_confirmargs c_arg;
 	struct nfs_open_confirmres c_res;
 	struct nfs_fattr f_attr;
 	struct nfs_fattr dir_attr;
-	struct dentry *dentry;
+	struct path path;
 	struct dentry *dir;
 	struct nfs4_state_owner *owner;
+	struct nfs4_state *state;
 	struct iattr attrs;
 	unsigned long timestamp;
+	unsigned int rpc_done : 1;
 	int rpc_status;
 	int cancelled;
 };
 
-static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
+
+static void nfs4_init_opendata_res(struct nfs4_opendata *p)
+{
+	p->o_res.f_attr = &p->f_attr;
+	p->o_res.dir_attr = &p->dir_attr;
+	p->o_res.server = p->o_arg.server;
+	nfs_fattr_init(&p->f_attr);
+	nfs_fattr_init(&p->dir_attr);
+}
+
+static struct nfs4_opendata *nfs4_opendata_alloc(struct path *path,
 		struct nfs4_state_owner *sp, int flags,
 		const struct iattr *attrs)
 {
-	struct dentry *parent = dget_parent(dentry);
+	struct dentry *parent = dget_parent(path->dentry);
 	struct inode *dir = parent->d_inode;
 	struct nfs_server *server = NFS_SERVER(dir);
 	struct nfs4_opendata *p;
@@ -245,24 +258,19 @@ static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
 	p->o_arg.seqid = nfs_alloc_seqid(&sp->so_seqid);
 	if (p->o_arg.seqid == NULL)
 		goto err_free;
-	atomic_set(&p->count, 1);
-	p->dentry = dget(dentry);
+	p->path.mnt = mntget(path->mnt);
+	p->path.dentry = dget(path->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.clientid = server->nfs_client->cl_clientid;
-	p->o_arg.id = sp->so_id;
-	p->o_arg.name = &dentry->d_name;
+	p->o_arg.id = sp->so_owner_id.id;
+	p->o_arg.name = &p->path.dentry->d_name;
 	p->o_arg.server = server;
 	p->o_arg.bitmask = server->attr_bitmask;
 	p->o_arg.claim = NFS4_OPEN_CLAIM_NULL;
-	p->o_res.f_attr = &p->f_attr;
-	p->o_res.dir_attr = &p->dir_attr;
-	p->o_res.server = server;
-	nfs_fattr_init(&p->f_attr);
-	nfs_fattr_init(&p->dir_attr);
 	if (flags & O_EXCL) {
 		u32 *s = (u32 *) p->o_arg.u.verifier.data;
 		s[0] = jiffies;
@@ -274,6 +282,8 @@ static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
 	p->c_arg.fh = &p->o_res.fh;
 	p->c_arg.stateid = &p->o_res.stateid;
 	p->c_arg.seqid = p->o_arg.seqid;
+	nfs4_init_opendata_res(p);
+	kref_init(&p->kref);
 	return p;
 err_free:
 	kfree(p);
@@ -282,27 +292,25 @@ err:
 	return NULL;
 }
 
-static void nfs4_opendata_free(struct nfs4_opendata *p)
+static void nfs4_opendata_free(struct kref *kref)
 {
-	if (p != NULL && atomic_dec_and_test(&p->count)) {
-		nfs_free_seqid(p->o_arg.seqid);
-		nfs4_put_state_owner(p->owner);
-		dput(p->dir);
-		dput(p->dentry);
-		kfree(p);
-	}
+	struct nfs4_opendata *p = container_of(kref,
+			struct nfs4_opendata, kref);
+
+	nfs_free_seqid(p->o_arg.seqid);
+	if (p->state != NULL)
+		nfs4_put_open_state(p->state);
+	nfs4_put_state_owner(p->owner);
+	dput(p->dir);
+	dput(p->path.dentry);
+	mntput(p->path.mnt);
+	kfree(p);
 }
 
-/* Helper for asynchronous RPC calls */
-static int nfs4_call_async(struct rpc_clnt *clnt,
-		const struct rpc_call_ops *tk_ops, void *calldata)
+static void nfs4_opendata_put(struct nfs4_opendata *p)
 {
-	struct rpc_task *task;
-
-	if (!(task = rpc_new_task(clnt, RPC_TASK_ASYNC, tk_ops, calldata)))
-		return -ENOMEM;
-	rpc_execute(task);
-	return 0;
+	if (p != NULL)
+		kref_put(&p->kref, nfs4_opendata_free);
 }
 
 static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
@@ -316,7 +324,34 @@ static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
 	return ret;
 }
 
-static inline void update_open_stateflags(struct nfs4_state *state, mode_t open_flags)
+static int can_open_cached(struct nfs4_state *state, int mode)
+{
+	int ret = 0;
+	switch (mode & (FMODE_READ|FMODE_WRITE|O_EXCL)) {
+		case FMODE_READ:
+			ret |= test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0;
+			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0;
+			break;
+		case FMODE_WRITE:
+			ret |= test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0;
+			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0;
+			break;
+		case FMODE_READ|FMODE_WRITE:
+			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0;
+	}
+	return ret;
+}
+
+static int can_open_delegated(struct nfs_delegation *delegation, mode_t open_flags)
+{
+	if ((delegation->type & open_flags) != open_flags)
+		return 0;
+	if (delegation->flags & NFS_DELEGATION_NEED_RECLAIM)
+		return 0;
+	return 1;
+}
+
+static void update_open_stateflags(struct nfs4_state *state, mode_t open_flags)
 {
 	switch (open_flags) {
 		case FMODE_WRITE:
@@ -328,41 +363,176 @@ static inline void update_open_stateflags(struct nfs4_state *state, mode_t open_
 		case FMODE_READ|FMODE_WRITE:
 			state->n_rdwr++;
 	}
+	nfs4_state_set_mode_locked(state, state->state | open_flags);
 }
 
-static void update_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, int open_flags)
+static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, int open_flags)
 {
-	struct inode *inode = state->inode;
+	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
+		memcpy(state->stateid.data, stateid->data, sizeof(state->stateid.data));
+	memcpy(state->open_stateid.data, stateid->data, sizeof(state->open_stateid.data));
+	switch (open_flags) {
+		case FMODE_READ:
+			set_bit(NFS_O_RDONLY_STATE, &state->flags);
+			break;
+		case FMODE_WRITE:
+			set_bit(NFS_O_WRONLY_STATE, &state->flags);
+			break;
+		case FMODE_READ|FMODE_WRITE:
+			set_bit(NFS_O_RDWR_STATE, &state->flags);
+	}
+}
+
+static void nfs_set_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, int open_flags)
+{
+	write_seqlock(&state->seqlock);
+	nfs_set_open_stateid_locked(state, stateid, open_flags);
+	write_sequnlock(&state->seqlock);
+}
 
+static void update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *deleg_stateid, int open_flags)
+{
 	open_flags &= (FMODE_READ|FMODE_WRITE);
-	/* Protect against nfs4_find_state_byowner() */
+	/*
+	 * Protect the call to nfs4_state_set_mode_locked and
+	 * serialise the stateid update
+	 */
+	write_seqlock(&state->seqlock);
+	if (deleg_stateid != NULL) {
+		memcpy(state->stateid.data, deleg_stateid->data, sizeof(state->stateid.data));
+		set_bit(NFS_DELEGATED_STATE, &state->flags);
+	}
+	if (open_stateid != NULL)
+		nfs_set_open_stateid_locked(state, open_stateid, open_flags);
+	write_sequnlock(&state->seqlock);
 	spin_lock(&state->owner->so_lock);
-	spin_lock(&inode->i_lock);
-	memcpy(&state->stateid, stateid, sizeof(state->stateid));
 	update_open_stateflags(state, open_flags);
-	nfs4_state_set_mode_locked(state, state->state | open_flags);
-	spin_unlock(&inode->i_lock);
 	spin_unlock(&state->owner->so_lock);
 }
 
+static void nfs4_return_incompatible_delegation(struct inode *inode, mode_t open_flags)
+{
+	struct nfs_delegation *delegation;
+
+	rcu_read_lock();
+	delegation = rcu_dereference(NFS_I(inode)->delegation);
+	if (delegation == NULL || (delegation->type & open_flags) == open_flags) {
+		rcu_read_unlock();
+		return;
+	}
+	rcu_read_unlock();
+	nfs_inode_return_delegation(inode);
+}
+
+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 & (FMODE_READ|FMODE_WRITE|O_EXCL);
+	nfs4_stateid stateid;
+	int ret = -EAGAIN;
+
+	rcu_read_lock();
+	delegation = rcu_dereference(nfsi->delegation);
+	for (;;) {
+		if (can_open_cached(state, open_mode)) {
+			spin_lock(&state->owner->so_lock);
+			if (can_open_cached(state, open_mode)) {
+				update_open_stateflags(state, open_mode);
+				spin_unlock(&state->owner->so_lock);
+				rcu_read_unlock();
+				goto out_return_state;
+			}
+			spin_unlock(&state->owner->so_lock);
+		}
+		if (delegation == NULL)
+			break;
+		if (!can_open_delegated(delegation, open_mode))
+			break;
+		/* Save the delegation */
+		memcpy(stateid.data, delegation->stateid.data, sizeof(stateid.data));
+		rcu_read_unlock();
+		lock_kernel();
+		ret = _nfs4_do_access(state->inode, state->owner->so_cred, open_mode);
+		unlock_kernel();
+		if (ret != 0)
+			goto out;
+		ret = -EAGAIN;
+		rcu_read_lock();
+		delegation = rcu_dereference(nfsi->delegation);
+		/* If no delegation, try a cached open */
+		if (delegation == NULL)
+			continue;
+		/* Is the delegation still valid? */
+		if (memcmp(stateid.data, delegation->stateid.data, sizeof(stateid.data)) != 0)
+			continue;
+		rcu_read_unlock();
+		update_open_stateid(state, NULL, &stateid, open_mode);
+		goto out_return_state;
+	}
+	rcu_read_unlock();
+out:
+	return ERR_PTR(ret);
+out_return_state:
+	atomic_inc(&state->count);
+	return state;
+}
+
 static struct nfs4_state *nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
 {
 	struct inode *inode;
 	struct nfs4_state *state = NULL;
+	struct nfs_delegation *delegation;
+	nfs4_stateid *deleg_stateid = NULL;
+	int ret;
 
-	if (!(data->f_attr.valid & NFS_ATTR_FATTR))
+	if (!data->rpc_done) {
+		state = nfs4_try_open_cached(data);
 		goto out;
+	}
+
+	ret = -EAGAIN;
+	if (!(data->f_attr.valid & NFS_ATTR_FATTR))
+		goto err;
 	inode = nfs_fhget(data->dir->d_sb, &data->o_res.fh, &data->f_attr);
+	ret = PTR_ERR(inode);
 	if (IS_ERR(inode))
-		goto out;
+		goto err;
+	ret = -ENOMEM;
 	state = nfs4_get_open_state(inode, data->owner);
 	if (state == NULL)
-		goto put_inode;
-	update_open_stateid(state, &data->o_res.stateid, data->o_arg.open_flags);
-put_inode:
+		goto err_put_inode;
+	if (data->o_res.delegation_type != 0) {
+		int delegation_flags = 0;
+
+		rcu_read_lock();
+		delegation = rcu_dereference(NFS_I(inode)->delegation);
+		if (delegation)
+			delegation_flags = delegation->flags;
+		rcu_read_unlock();
+		if (!(delegation_flags & NFS_DELEGATION_NEED_RECLAIM))
+			nfs_inode_set_delegation(state->inode,
+					data->owner->so_cred,
+					&data->o_res);
+		else
+			nfs_inode_reclaim_delegation(state->inode,
+					data->owner->so_cred,
+					&data->o_res);
+	}
+	rcu_read_lock();
+	delegation = rcu_dereference(NFS_I(inode)->delegation);
+	if (delegation != NULL)
+		deleg_stateid = &delegation->stateid;
+	update_open_stateid(state, &data->o_res.stateid, deleg_stateid, data->o_arg.open_flags);
+	rcu_read_unlock();
 	iput(inode);
 out:
 	return state;
+err_put_inode:
+	iput(inode);
+err:
+	return ERR_PTR(ret);
 }
 
 static struct nfs_open_context *nfs4_state_find_open_context(struct nfs4_state *state)
@@ -382,79 +552,66 @@ static struct nfs_open_context *nfs4_state_find_open_context(struct nfs4_state *
 	return ERR_PTR(-ENOENT);
 }
 
-static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, mode_t openflags, nfs4_stateid *stateid)
+static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, mode_t openflags, struct nfs4_state **res)
 {
+	struct nfs4_state *newstate;
 	int ret;
 
 	opendata->o_arg.open_flags = openflags;
+	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
+	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
+	nfs4_init_opendata_res(opendata);
 	ret = _nfs4_proc_open(opendata);
 	if (ret != 0)
 		return ret; 
-	memcpy(stateid->data, opendata->o_res.stateid.data,
-			sizeof(stateid->data));
+	newstate = nfs4_opendata_to_nfs4_state(opendata);
+	if (IS_ERR(newstate))
+		return PTR_ERR(newstate);
+	nfs4_close_state(&opendata->path, newstate, openflags);
+	*res = newstate;
 	return 0;
 }
 
 static int nfs4_open_recover(struct nfs4_opendata *opendata, struct nfs4_state *state)
 {
-	nfs4_stateid stateid;
 	struct nfs4_state *newstate;
-	int mode = 0;
-	int delegation = 0;
 	int ret;
 
 	/* memory barrier prior to reading state->n_* */
+	clear_bit(NFS_DELEGATED_STATE, &state->flags);
 	smp_rmb();
 	if (state->n_rdwr != 0) {
-		ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE, &stateid);
+		ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE, &newstate);
 		if (ret != 0)
 			return ret;
-		mode |= FMODE_READ|FMODE_WRITE;
-		if (opendata->o_res.delegation_type != 0)
-			delegation = opendata->o_res.delegation_type;
-		smp_rmb();
+		if (newstate != state)
+			return -ESTALE;
 	}
 	if (state->n_wronly != 0) {
-		ret = nfs4_open_recover_helper(opendata, FMODE_WRITE, &stateid);
+		ret = nfs4_open_recover_helper(opendata, FMODE_WRITE, &newstate);
 		if (ret != 0)
 			return ret;
-		mode |= FMODE_WRITE;
-		if (opendata->o_res.delegation_type != 0)
-			delegation = opendata->o_res.delegation_type;
-		smp_rmb();
+		if (newstate != state)
+			return -ESTALE;
 	}
 	if (state->n_rdonly != 0) {
-		ret = nfs4_open_recover_helper(opendata, FMODE_READ, &stateid);
+		ret = nfs4_open_recover_helper(opendata, FMODE_READ, &newstate);
 		if (ret != 0)
 			return ret;
-		mode |= FMODE_READ;
+		if (newstate != state)
+			return -ESTALE;
 	}
-	clear_bit(NFS_DELEGATED_STATE, &state->flags);
-	if (mode == 0)
-		return 0;
-	if (opendata->o_res.delegation_type == 0)
-		opendata->o_res.delegation_type = delegation;
-	opendata->o_arg.open_flags |= mode;
-	newstate = nfs4_opendata_to_nfs4_state(opendata);
-	if (newstate != NULL) {
-		if (opendata->o_res.delegation_type != 0) {
-			struct nfs_inode *nfsi = NFS_I(newstate->inode);
-			int delegation_flags = 0;
-			if (nfsi->delegation)
-				delegation_flags = nfsi->delegation->flags;
-			if (!(delegation_flags & NFS_DELEGATION_NEED_RECLAIM))
-				nfs_inode_set_delegation(newstate->inode,
-						opendata->owner->so_cred,
-						&opendata->o_res);
-			else
-				nfs_inode_reclaim_delegation(newstate->inode,
-						opendata->owner->so_cred,
-						&opendata->o_res);
-		}
-		nfs4_close_state(newstate, opendata->o_arg.open_flags);
+	/*
+	 * We may have performed cached opens for all three recoveries.
+	 * Check if we need to update the current stateid.
+	 */
+	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0 &&
+	    memcmp(state->stateid.data, state->open_stateid.data, sizeof(state->stateid.data)) != 0) {
+		write_seqlock(&state->seqlock);
+		if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
+			memcpy(state->stateid.data, state->open_stateid.data, sizeof(state->stateid.data));
+		write_sequnlock(&state->seqlock);
 	}
-	if (newstate != state)
-		return -ESTALE;
 	return 0;
 }
 
@@ -462,41 +619,37 @@ static int nfs4_open_recover(struct nfs4_opendata *opendata, struct nfs4_state *
  * OPEN_RECLAIM:
  * 	reclaim state on the server after a reboot.
  */
-static int _nfs4_do_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state, struct dentry *dentry)
+static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
 {
-	struct nfs_delegation *delegation = NFS_I(state->inode)->delegation;
+	struct nfs_delegation *delegation;
 	struct nfs4_opendata *opendata;
 	int delegation_type = 0;
 	int status;
 
-	if (delegation != NULL) {
-		if (!(delegation->flags & NFS_DELEGATION_NEED_RECLAIM)) {
-			memcpy(&state->stateid, &delegation->stateid,
-					sizeof(state->stateid));
-			set_bit(NFS_DELEGATED_STATE, &state->flags);
-			return 0;
-		}
-		delegation_type = delegation->type;
-	}
-	opendata = nfs4_opendata_alloc(dentry, sp, 0, NULL);
+	opendata = nfs4_opendata_alloc(&ctx->path, state->owner, 0, NULL);
 	if (opendata == NULL)
 		return -ENOMEM;
 	opendata->o_arg.claim = NFS4_OPEN_CLAIM_PREVIOUS;
 	opendata->o_arg.fh = NFS_FH(state->inode);
 	nfs_copy_fh(&opendata->o_res.fh, opendata->o_arg.fh);
+	rcu_read_lock();
+	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
+	if (delegation != NULL && (delegation->flags & NFS_DELEGATION_NEED_RECLAIM) != 0)
+		delegation_type = delegation->flags;
+	rcu_read_unlock();
 	opendata->o_arg.u.delegation_type = delegation_type;
 	status = nfs4_open_recover(opendata, state);
-	nfs4_opendata_free(opendata);
+	nfs4_opendata_put(opendata);
 	return status;
 }
 
-static int nfs4_do_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state, struct dentry *dentry)
+static int nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
 {
 	struct nfs_server *server = NFS_SERVER(state->inode);
 	struct nfs4_exception exception = { };
 	int err;
 	do {
-		err = _nfs4_do_open_reclaim(sp, state, dentry);
+		err = _nfs4_do_open_reclaim(ctx, state);
 		if (err != -NFS4ERR_DELAY)
 			break;
 		nfs4_handle_exception(server, err, &exception);
@@ -512,37 +665,35 @@ 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);
-	ret = nfs4_do_open_reclaim(sp, state, ctx->dentry);
+	ret = nfs4_do_open_reclaim(ctx, state);
 	put_nfs_open_context(ctx);
 	return ret;
 }
 
-static int _nfs4_open_delegation_recall(struct dentry *dentry, struct nfs4_state *state)
+static int _nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid)
 {
 	struct nfs4_state_owner  *sp  = state->owner;
 	struct nfs4_opendata *opendata;
 	int ret;
 
-	if (!test_bit(NFS_DELEGATED_STATE, &state->flags))
-		return 0;
-	opendata = nfs4_opendata_alloc(dentry, sp, 0, NULL);
+	opendata = nfs4_opendata_alloc(&ctx->path, sp, 0, NULL);
 	if (opendata == NULL)
 		return -ENOMEM;
 	opendata->o_arg.claim = NFS4_OPEN_CLAIM_DELEGATE_CUR;
-	memcpy(opendata->o_arg.u.delegation.data, state->stateid.data,
+	memcpy(opendata->o_arg.u.delegation.data, stateid->data,
 			sizeof(opendata->o_arg.u.delegation.data));
 	ret = nfs4_open_recover(opendata, state);
-	nfs4_opendata_free(opendata);
+	nfs4_opendata_put(opendata);
 	return ret;
 }
 
-int nfs4_open_delegation_recall(struct dentry *dentry, struct nfs4_state *state)
+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(dentry->d_inode);
+	struct nfs_server *server = NFS_SERVER(state->inode);
 	int err;
 	do {
-		err = _nfs4_open_delegation_recall(dentry, state);
+		err = _nfs4_open_delegation_recall(ctx, state, stateid);
 		switch (err) {
 			case 0:
 				return err;
@@ -582,9 +733,10 @@ static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata)
 		memcpy(data->o_res.stateid.data, data->c_res.stateid.data,
 				sizeof(data->o_res.stateid.data));
 		renew_lease(data->o_res.server, data->timestamp);
+		data->rpc_done = 1;
 	}
-	nfs_increment_open_seqid(data->rpc_status, data->c_arg.seqid);
 	nfs_confirm_seqid(&data->owner->so_seqid, data->rpc_status);
+	nfs_increment_open_seqid(data->rpc_status, data->c_arg.seqid);
 }
 
 static void nfs4_open_confirm_release(void *calldata)
@@ -596,14 +748,14 @@ static void nfs4_open_confirm_release(void *calldata)
 	if (data->cancelled == 0)
 		goto out_free;
 	/* In case of error, no cleanup! */
-	if (data->rpc_status != 0)
+	if (!data->rpc_done)
 		goto out_free;
 	nfs_confirm_seqid(&data->owner->so_seqid, 0);
 	state = nfs4_opendata_to_nfs4_state(data);
-	if (state != NULL)
-		nfs4_close_state(state, data->o_arg.open_flags);
+	if (!IS_ERR(state))
+		nfs4_close_state(&data->path, state, data->o_arg.open_flags);
 out_free:
-	nfs4_opendata_free(data);
+	nfs4_opendata_put(data);
 }
 
 static const struct rpc_call_ops nfs4_open_confirm_ops = {
@@ -621,12 +773,9 @@ static int _nfs4_proc_open_confirm(struct nfs4_opendata *data)
 	struct rpc_task *task;
 	int status;
 
-	atomic_inc(&data->count);
-	/*
-	 * If rpc_run_task() ends up calling ->rpc_release(), we
-	 * want to ensure that it takes the 'error' code path.
-	 */
-	data->rpc_status = -ENOMEM;
+	kref_get(&data->kref);
+	data->rpc_done = 0;
+	data->rpc_status = 0;
 	task = rpc_run_task(server->client, RPC_TASK_ASYNC, &nfs4_open_confirm_ops, data);
 	if (IS_ERR(task))
 		return PTR_ERR(task);
@@ -653,13 +802,35 @@ static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
 	
 	if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0)
 		return;
+	/*
+	 * Check if we still need to send an OPEN call, or if we can use
+	 * a delegation instead.
+	 */
+	if (data->state != NULL) {
+		struct nfs_delegation *delegation;
+
+		if (can_open_cached(data->state, data->o_arg.open_flags & (FMODE_READ|FMODE_WRITE|O_EXCL)))
+			goto out_no_action;
+		rcu_read_lock();
+		delegation = rcu_dereference(NFS_I(data->state->inode)->delegation);
+		if (delegation != NULL &&
+		   (delegation->flags & NFS_DELEGATION_NEED_RECLAIM) == 0) {
+			rcu_read_unlock();
+			goto out_no_action;
+		}
+		rcu_read_unlock();
+	}
 	/* Update sequence id. */
-	data->o_arg.id = sp->so_id;
+	data->o_arg.id = sp->so_owner_id.id;
 	data->o_arg.clientid = sp->so_client->cl_clientid;
 	if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS)
 		msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
 	data->timestamp = jiffies;
 	rpc_call_setup(task, &msg, 0);
+	return;
+out_no_action:
+	task->tk_action = NULL;
+
 }
 
 static void nfs4_open_done(struct rpc_task *task, void *calldata)
@@ -683,8 +854,11 @@ static void nfs4_open_done(struct rpc_task *task, void *calldata)
 				data->rpc_status = -ENOTDIR;
 		}
 		renew_lease(data->o_res.server, data->timestamp);
+		if (!(data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM))
+			nfs_confirm_seqid(&data->owner->so_seqid, 0);
 	}
 	nfs_increment_open_seqid(data->rpc_status, data->o_arg.seqid);
+	data->rpc_done = 1;
 }
 
 static void nfs4_open_release(void *calldata)
@@ -696,17 +870,17 @@ static void nfs4_open_release(void *calldata)
 	if (data->cancelled == 0)
 		goto out_free;
 	/* In case of error, no cleanup! */
-	if (data->rpc_status != 0)
+	if (data->rpc_status != 0 || !data->rpc_done)
 		goto out_free;
 	/* In case we need an open_confirm, no cleanup! */
 	if (data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM)
 		goto out_free;
 	nfs_confirm_seqid(&data->owner->so_seqid, 0);
 	state = nfs4_opendata_to_nfs4_state(data);
-	if (state != NULL)
-		nfs4_close_state(state, data->o_arg.open_flags);
+	if (!IS_ERR(state))
+		nfs4_close_state(&data->path, state, data->o_arg.open_flags);
 out_free:
-	nfs4_opendata_free(data);
+	nfs4_opendata_put(data);
 }
 
 static const struct rpc_call_ops nfs4_open_ops = {
@@ -727,12 +901,10 @@ static int _nfs4_proc_open(struct nfs4_opendata *data)
 	struct rpc_task *task;
 	int status;
 
-	atomic_inc(&data->count);
-	/*
-	 * If rpc_run_task() ends up calling ->rpc_release(), we
-	 * want to ensure that it takes the 'error' code path.
-	 */
-	data->rpc_status = -ENOMEM;
+	kref_get(&data->kref);
+	data->rpc_done = 0;
+	data->rpc_status = 0;
+	data->cancelled = 0;
 	task = rpc_run_task(server->client, RPC_TASK_ASYNC, &nfs4_open_ops, data);
 	if (IS_ERR(task))
 		return PTR_ERR(task);
@@ -743,7 +915,7 @@ static int _nfs4_proc_open(struct nfs4_opendata *data)
 	} else
 		status = data->rpc_status;
 	rpc_put_task(task);
-	if (status != 0)
+	if (status != 0 || !data->rpc_done)
 		return status;
 
 	if (o_arg->open_flags & O_CREAT) {
@@ -756,7 +928,6 @@ static int _nfs4_proc_open(struct nfs4_opendata *data)
 		if (status != 0)
 			return status;
 	}
-	nfs_confirm_seqid(&data->owner->so_seqid, 0);
 	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR))
 		return server->nfs_client->rpc_ops->getattr(server, &o_res->fh, o_res->f_attr);
 	return 0;
@@ -772,6 +943,8 @@ static int _nfs4_do_access(struct inode *inode, struct rpc_cred *cred, int openf
 		mask |= MAY_READ;
 	if (openflags & FMODE_WRITE)
 		mask |= MAY_WRITE;
+	if (openflags & FMODE_EXEC)
+		mask |= MAY_EXEC;
 	status = nfs_access_get_cached(inode, cred, &cache);
 	if (status == 0)
 		goto out;
@@ -811,43 +984,32 @@ static int nfs4_recover_expired_lease(struct nfs_server *server)
  * 	reclaim state on the server after a network partition.
  * 	Assumes caller holds the appropriate lock
  */
-static int _nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state, struct dentry *dentry)
+static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
 {
-	struct inode *inode = state->inode;
-	struct nfs_delegation *delegation = NFS_I(inode)->delegation;
 	struct nfs4_opendata *opendata;
-	int openflags = state->state & (FMODE_READ|FMODE_WRITE);
 	int ret;
 
-	if (delegation != NULL && !(delegation->flags & NFS_DELEGATION_NEED_RECLAIM)) {
-		ret = _nfs4_do_access(inode, sp->so_cred, openflags);
-		if (ret < 0)
-			return ret;
-		memcpy(&state->stateid, &delegation->stateid, sizeof(state->stateid));
-		set_bit(NFS_DELEGATED_STATE, &state->flags);
-		return 0;
-	}
-	opendata = nfs4_opendata_alloc(dentry, sp, openflags, NULL);
+	opendata = nfs4_opendata_alloc(&ctx->path, state->owner, 0, NULL);
 	if (opendata == NULL)
 		return -ENOMEM;
 	ret = nfs4_open_recover(opendata, state);
 	if (ret == -ESTALE) {
 		/* Invalidate the state owner so we don't ever use it again */
-		nfs4_drop_state_owner(sp);
-		d_drop(dentry);
+		nfs4_drop_state_owner(state->owner);
+		d_drop(ctx->path.dentry);
 	}
-	nfs4_opendata_free(opendata);
+	nfs4_opendata_put(opendata);
 	return ret;
 }
 
-static inline int nfs4_do_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state, struct dentry *dentry)
+static inline int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
 {
-	struct nfs_server *server = NFS_SERVER(dentry->d_inode);
+	struct nfs_server *server = NFS_SERVER(state->inode);
 	struct nfs4_exception exception = { };
 	int err;
 
 	do {
-		err = _nfs4_open_expired(sp, state, dentry);
+		err = _nfs4_open_expired(ctx, state);
 		if (err == -NFS4ERR_DELAY)
 			nfs4_handle_exception(server, err, &exception);
 	} while (exception.retry);
@@ -862,107 +1024,38 @@ 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);
-	ret = nfs4_do_open_expired(sp, state, ctx->dentry);
+	ret = nfs4_do_open_expired(ctx, state);
 	put_nfs_open_context(ctx);
 	return ret;
 }
 
 /*
- * Returns a referenced nfs4_state if there is an open delegation on the file
+ * on an EXCLUSIVE create, the server should send back a bitmask with FATTR4-*
+ * fields corresponding to attributes that were used to store the verifier.
+ * Make sure we clobber those fields in the later setattr call
  */
-static int _nfs4_open_delegated(struct inode *inode, int flags, struct rpc_cred *cred, struct nfs4_state **res)
-{
-	struct nfs_delegation *delegation;
-	struct nfs_server *server = NFS_SERVER(inode);
-	struct nfs_client *clp = server->nfs_client;
-	struct nfs_inode *nfsi = NFS_I(inode);
-	struct nfs4_state_owner *sp = NULL;
-	struct nfs4_state *state = NULL;
-	int open_flags = flags & (FMODE_READ|FMODE_WRITE);
-	int err;
-
-	err = -ENOMEM;
-	if (!(sp = nfs4_get_state_owner(server, cred))) {
-		dprintk("%s: nfs4_get_state_owner failed!\n", __FUNCTION__);
-		return err;
-	}
-	err = nfs4_recover_expired_lease(server);
-	if (err != 0)
-		goto out_put_state_owner;
-	/* Protect against reboot recovery - NOTE ORDER! */
-	down_read(&clp->cl_sem);
-	/* Protect against delegation recall */
-	down_read(&nfsi->rwsem);
-	delegation = NFS_I(inode)->delegation;
-	err = -ENOENT;
-	if (delegation == NULL || (delegation->type & open_flags) != open_flags)
-		goto out_err;
-	err = -ENOMEM;
-	state = nfs4_get_open_state(inode, sp);
-	if (state == NULL)
-		goto out_err;
-
-	err = -ENOENT;
-	if ((state->state & open_flags) == open_flags) {
-		spin_lock(&inode->i_lock);
-		update_open_stateflags(state, open_flags);
-		spin_unlock(&inode->i_lock);
-		goto out_ok;
-	} else if (state->state != 0)
-		goto out_put_open_state;
-
-	lock_kernel();
-	err = _nfs4_do_access(inode, cred, open_flags);
-	unlock_kernel();
-	if (err != 0)
-		goto out_put_open_state;
-	set_bit(NFS_DELEGATED_STATE, &state->flags);
-	update_open_stateid(state, &delegation->stateid, open_flags);
-out_ok:
-	nfs4_put_state_owner(sp);
-	up_read(&nfsi->rwsem);
-	up_read(&clp->cl_sem);
-	*res = state;
-	return 0;
-out_put_open_state:
-	nfs4_put_open_state(state);
-out_err:
-	up_read(&nfsi->rwsem);
-	up_read(&clp->cl_sem);
-	if (err != -EACCES)
-		nfs_inode_return_delegation(inode);
-out_put_state_owner:
-	nfs4_put_state_owner(sp);
-	return err;
-}
-
-static struct nfs4_state *nfs4_open_delegated(struct inode *inode, int flags, struct rpc_cred *cred)
+static inline void nfs4_exclusive_attrset(struct nfs4_opendata *opendata, struct iattr *sattr)
 {
-	struct nfs4_exception exception = { };
-	struct nfs4_state *res = ERR_PTR(-EIO);
-	int err;
+	if ((opendata->o_res.attrset[1] & FATTR4_WORD1_TIME_ACCESS) &&
+	    !(sattr->ia_valid & ATTR_ATIME_SET))
+		sattr->ia_valid |= ATTR_ATIME;
 
-	do {
-		err = _nfs4_open_delegated(inode, flags, cred, &res);
-		if (err == 0)
-			break;
-		res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(inode),
-					err, &exception));
-	} while (exception.retry);
-	return res;
+	if ((opendata->o_res.attrset[1] & FATTR4_WORD1_TIME_MODIFY) &&
+	    !(sattr->ia_valid & ATTR_MTIME_SET))
+		sattr->ia_valid |= ATTR_MTIME;
 }
 
 /*
  * Returns a referenced nfs4_state
  */
-static int _nfs4_do_open(struct inode *dir, struct dentry *dentry, int flags, struct iattr *sattr, struct rpc_cred *cred, struct nfs4_state **res)
+static int _nfs4_do_open(struct inode *dir, struct path *path, int flags, struct iattr *sattr, struct rpc_cred *cred, struct nfs4_state **res)
 {
 	struct nfs4_state_owner  *sp;
 	struct nfs4_state     *state = NULL;
 	struct nfs_server       *server = NFS_SERVER(dir);
 	struct nfs_client *clp = server->nfs_client;
 	struct nfs4_opendata *opendata;
-	int                     status;
+	int status;
 
 	/* Protect against reboot recovery conflicts */
 	status = -ENOMEM;
@@ -973,29 +1066,35 @@ static int _nfs4_do_open(struct inode *dir, struct dentry *dentry, int flags, st
 	status = nfs4_recover_expired_lease(server);
 	if (status != 0)
 		goto err_put_state_owner;
+	if (path->dentry->d_inode != NULL)
+		nfs4_return_incompatible_delegation(path->dentry->d_inode, flags & (FMODE_READ|FMODE_WRITE));
 	down_read(&clp->cl_sem);
 	status = -ENOMEM;
-	opendata = nfs4_opendata_alloc(dentry, sp, flags, sattr);
+	opendata = nfs4_opendata_alloc(path, sp, flags, sattr);
 	if (opendata == NULL)
 		goto err_release_rwsem;
 
+	if (path->dentry->d_inode != NULL)
+		opendata->state = nfs4_get_open_state(path->dentry->d_inode, sp);
+
 	status = _nfs4_proc_open(opendata);
 	if (status != 0)
-		goto err_opendata_free;
+		goto err_opendata_put;
+
+	if (opendata->o_arg.open_flags & O_EXCL)
+		nfs4_exclusive_attrset(opendata, sattr);
 
-	status = -ENOMEM;
 	state = nfs4_opendata_to_nfs4_state(opendata);
-	if (state == NULL)
-		goto err_opendata_free;
-	if (opendata->o_res.delegation_type != 0)
-		nfs_inode_set_delegation(state->inode, cred, &opendata->o_res);
-	nfs4_opendata_free(opendata);
+	status = PTR_ERR(state);
+	if (IS_ERR(state))
+		goto err_opendata_put;
+	nfs4_opendata_put(opendata);
 	nfs4_put_state_owner(sp);
 	up_read(&clp->cl_sem);
 	*res = state;
 	return 0;
-err_opendata_free:
-	nfs4_opendata_free(opendata);
+err_opendata_put:
+	nfs4_opendata_put(opendata);
 err_release_rwsem:
 	up_read(&clp->cl_sem);
 err_put_state_owner:
@@ -1006,14 +1105,14 @@ out_err:
 }
 
 
-static struct nfs4_state *nfs4_do_open(struct inode *dir, struct dentry *dentry, int flags, struct iattr *sattr, struct rpc_cred *cred)
+static struct nfs4_state *nfs4_do_open(struct inode *dir, struct path *path, int flags, struct iattr *sattr, struct rpc_cred *cred)
 {
 	struct nfs4_exception exception = { };
 	struct nfs4_state *res;
 	int status;
 
 	do {
-		status = _nfs4_do_open(dir, dentry, flags, sattr, cred, &res);
+		status = _nfs4_do_open(dir, path, flags, sattr, cred, &res);
 		if (status == 0)
 			break;
 		/* NOTE: BAD_SEQID means the server and client disagree about the
@@ -1028,7 +1127,9 @@ static struct nfs4_state *nfs4_do_open(struct inode *dir, struct dentry *dentry,
 		 * the user though...
 		 */
 		if (status == -NFS4ERR_BAD_SEQID) {
-			printk(KERN_WARNING "NFS: v4 server returned a bad sequence-id error!\n");
+			printk(KERN_WARNING "NFS: v4 server %s "
+					" returned a bad sequence-id error!\n",
+					NFS_SERVER(dir)->nfs_client->cl_hostname);
 			exception.retry = 1;
 			continue;
 		}
@@ -1042,6 +1143,11 @@ static struct nfs4_state *nfs4_do_open(struct inode *dir, struct dentry *dentry,
 			exception.retry = 1;
 			continue;
 		}
+		if (status == -EAGAIN) {
+			/* We must have found a delegation */
+			exception.retry = 1;
+			continue;
+		}
 		res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(dir),
 					status, &exception));
 	} while (exception.retry);
@@ -1101,6 +1207,7 @@ static int nfs4_do_setattr(struct inode *inode, struct nfs_fattr *fattr,
 }
 
 struct nfs4_closedata {
+	struct path path;
 	struct inode *inode;
 	struct nfs4_state *state;
 	struct nfs_closeargs arg;
@@ -1117,6 +1224,8 @@ static void nfs4_free_closedata(void *data)
 	nfs4_put_open_state(calldata->state);
 	nfs_free_seqid(calldata->arg.seqid);
 	nfs4_put_state_owner(sp);
+	dput(calldata->path.dentry);
+	mntput(calldata->path.mnt);
 	kfree(calldata);
 }
 
@@ -1134,8 +1243,7 @@ static void nfs4_close_done(struct rpc_task *task, void *data)
 	nfs_increment_open_seqid(task->tk_status, calldata->arg.seqid);
 	switch (task->tk_status) {
 		case 0:
-			memcpy(&state->stateid, &calldata->res.stateid,
-					sizeof(state->stateid));
+			nfs_set_open_stateid(state, &calldata->res.stateid, calldata->arg.open_flags);
 			renew_lease(server, calldata->timestamp);
 			break;
 		case -NFS4ERR_STALE_STATEID:
@@ -1160,26 +1268,30 @@ static void nfs4_close_prepare(struct rpc_task *task, void *data)
 		.rpc_resp = &calldata->res,
 		.rpc_cred = state->owner->so_cred,
 	};
-	int mode = 0, old_mode;
+	int clear_rd, clear_wr, clear_rdwr;
+	int mode;
 
 	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
 		return;
-	/* Recalculate the new open mode in case someone reopened the file
-	 * while we were waiting in line to be scheduled.
-	 */
+
+	mode = FMODE_READ|FMODE_WRITE;
+	clear_rd = clear_wr = clear_rdwr = 0;
 	spin_lock(&state->owner->so_lock);
-	spin_lock(&calldata->inode->i_lock);
-	mode = old_mode = state->state;
+	/* Calculate the change in open mode */
 	if (state->n_rdwr == 0) {
-		if (state->n_rdonly == 0)
+		if (state->n_rdonly == 0) {
 			mode &= ~FMODE_READ;
-		if (state->n_wronly == 0)
+			clear_rd |= test_and_clear_bit(NFS_O_RDONLY_STATE, &state->flags);
+			clear_rdwr |= test_and_clear_bit(NFS_O_RDWR_STATE, &state->flags);
+		}
+		if (state->n_wronly == 0) {
 			mode &= ~FMODE_WRITE;
+			clear_wr |= test_and_clear_bit(NFS_O_WRONLY_STATE, &state->flags);
+			clear_rdwr |= test_and_clear_bit(NFS_O_RDWR_STATE, &state->flags);
+		}
 	}
-	nfs4_state_set_mode_locked(state, mode);
-	spin_unlock(&calldata->inode->i_lock);
 	spin_unlock(&state->owner->so_lock);
-	if (mode == old_mode || test_bit(NFS_DELEGATED_STATE, &state->flags)) {
+	if (!clear_rd && !clear_wr && !clear_rdwr) {
 		/* Note: exit _without_ calling nfs4_close_done */
 		task->tk_action = NULL;
 		return;
@@ -1209,19 +1321,21 @@ static const struct rpc_call_ops nfs4_close_ops = {
  *
  * NOTE: Caller must be holding the sp->so_owner semaphore!
  */
-int nfs4_do_close(struct inode *inode, struct nfs4_state *state) 
+int nfs4_do_close(struct path *path, struct nfs4_state *state)
 {
-	struct nfs_server *server = NFS_SERVER(inode);
+	struct nfs_server *server = NFS_SERVER(state->inode);
 	struct nfs4_closedata *calldata;
+	struct nfs4_state_owner *sp = state->owner;
+	struct rpc_task *task;
 	int status = -ENOMEM;
 
 	calldata = kmalloc(sizeof(*calldata), GFP_KERNEL);
 	if (calldata == NULL)
 		goto out;
-	calldata->inode = inode;
+	calldata->inode = state->inode;
 	calldata->state = state;
-	calldata->arg.fh = NFS_FH(inode);
-	calldata->arg.stateid = &state->stateid;
+	calldata->arg.fh = NFS_FH(state->inode);
+	calldata->arg.stateid = &state->open_stateid;
 	/* Serialization for the sequence id */
 	calldata->arg.seqid = nfs_alloc_seqid(&state->owner->so_seqid);
 	if (calldata->arg.seqid == NULL)
@@ -1229,36 +1343,55 @@ int nfs4_do_close(struct inode *inode, struct nfs4_state *state)
 	calldata->arg.bitmask = server->attr_bitmask;
 	calldata->res.fattr = &calldata->fattr;
 	calldata->res.server = server;
+	calldata->path.mnt = mntget(path->mnt);
+	calldata->path.dentry = dget(path->dentry);
 
-	status = nfs4_call_async(server->client, &nfs4_close_ops, calldata);
-	if (status == 0)
-		goto out;
-
-	nfs_free_seqid(calldata->arg.seqid);
+	task = rpc_run_task(server->client, RPC_TASK_ASYNC, &nfs4_close_ops, calldata);
+	if (IS_ERR(task))
+		return PTR_ERR(task);
+	rpc_put_task(task);
+	return 0;
 out_free_calldata:
 	kfree(calldata);
 out:
+	nfs4_put_open_state(state);
+	nfs4_put_state_owner(sp);
 	return status;
 }
 
-static int nfs4_intent_set_file(struct nameidata *nd, struct dentry *dentry, struct nfs4_state *state)
+static int nfs4_intent_set_file(struct nameidata *nd, struct path *path, struct nfs4_state *state)
 {
 	struct file *filp;
+	int ret;
 
-	filp = lookup_instantiate_filp(nd, dentry, NULL);
+	/* If the open_intent is for execute, we have an extra check to make */
+	if (nd->intent.open.flags & FMODE_EXEC) {
+		ret = _nfs4_do_access(state->inode,
+				state->owner->so_cred,
+				nd->intent.open.flags);
+		if (ret < 0)
+			goto out_close;
+	}
+	filp = lookup_instantiate_filp(nd, path->dentry, NULL);
 	if (!IS_ERR(filp)) {
 		struct nfs_open_context *ctx;
 		ctx = (struct nfs_open_context *)filp->private_data;
 		ctx->state = state;
 		return 0;
 	}
-	nfs4_close_state(state, nd->intent.open.flags);
-	return PTR_ERR(filp);
+	ret = PTR_ERR(filp);
+out_close:
+	nfs4_close_state(path, state, nd->intent.open.flags);
+	return ret;
 }
 
 struct dentry *
 nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
 {
+	struct path path = {
+		.mnt = nd->mnt,
+		.dentry = dentry,
+	};
 	struct iattr attr;
 	struct rpc_cred *cred;
 	struct nfs4_state *state;
@@ -1277,7 +1410,7 @@ nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
 	cred = rpcauth_lookupcred(NFS_CLIENT(dir)->cl_auth, 0);
 	if (IS_ERR(cred))
 		return (struct dentry *)cred;
-	state = nfs4_do_open(dir, dentry, nd->intent.open.flags, &attr, cred);
+	state = nfs4_do_open(dir, &path, nd->intent.open.flags, &attr, cred);
 	put_rpccred(cred);
 	if (IS_ERR(state)) {
 		if (PTR_ERR(state) == -ENOENT)
@@ -1287,22 +1420,24 @@ nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
 	res = d_add_unique(dentry, igrab(state->inode));
 	if (res != NULL)
 		dentry = res;
-	nfs4_intent_set_file(nd, dentry, state);
+	nfs4_intent_set_file(nd, &path, state);
 	return res;
 }
 
 int
 nfs4_open_revalidate(struct inode *dir, struct dentry *dentry, int openflags, struct nameidata *nd)
 {
+	struct path path = {
+		.mnt = nd->mnt,
+		.dentry = dentry,
+	};
 	struct rpc_cred *cred;
 	struct nfs4_state *state;
 
 	cred = rpcauth_lookupcred(NFS_CLIENT(dir)->cl_auth, 0);
 	if (IS_ERR(cred))
 		return PTR_ERR(cred);
-	state = nfs4_open_delegated(dentry->d_inode, openflags, cred);
-	if (IS_ERR(state))
-		state = nfs4_do_open(dir, dentry, openflags, NULL, cred);
+	state = nfs4_do_open(dir, &path, openflags, NULL, cred);
 	put_rpccred(cred);
 	if (IS_ERR(state)) {
 		switch (PTR_ERR(state)) {
@@ -1318,10 +1453,10 @@ nfs4_open_revalidate(struct inode *dir, struct dentry *dentry, int openflags, st
 		}
 	}
 	if (state->inode == dentry->d_inode) {
-		nfs4_intent_set_file(nd, dentry, state);
+		nfs4_intent_set_file(nd, &path, state);
 		return 1;
 	}
-	nfs4_close_state(state, openflags);
+	nfs4_close_state(&path, state, openflags);
 out_drop:
 	d_drop(dentry);
 	return 0;
@@ -1559,8 +1694,6 @@ static int _nfs4_proc_lookupfh(struct nfs_server *server, struct nfs_fh *dirfh,
 	dprintk("NFS call  lookupfh %s\n", name->name);
 	status = rpc_call_sync(server->client, &msg, 0);
 	dprintk("NFS reply lookupfh: %d\n", status);
-	if (status == -NFS4ERR_MOVED)
-		status = -EREMOTE;
 	return status;
 }
 
@@ -1571,10 +1704,13 @@ static int nfs4_proc_lookupfh(struct nfs_server *server, struct nfs_fh *dirfh,
 	struct nfs4_exception exception = { };
 	int err;
 	do {
-		err = nfs4_handle_exception(server,
-				_nfs4_proc_lookupfh(server, dirfh, name,
-						    fhandle, fattr),
-				&exception);
+		err = _nfs4_proc_lookupfh(server, dirfh, name, fhandle, fattr);
+		/* FIXME: !!!! */
+		if (err == -NFS4ERR_MOVED) {
+			err = -EREMOTE;
+			break;
+		}
+		err = nfs4_handle_exception(server, err, &exception);
 	} while (exception.retry);
 	return err;
 }
@@ -1582,28 +1718,10 @@ static int nfs4_proc_lookupfh(struct nfs_server *server, struct nfs_fh *dirfh,
 static int _nfs4_proc_lookup(struct inode *dir, struct qstr *name,
 		struct nfs_fh *fhandle, struct nfs_fattr *fattr)
 {
-	int		       status;
-	struct nfs_server *server = NFS_SERVER(dir);
-	struct nfs4_lookup_arg args = {
-		.bitmask = server->attr_bitmask,
-		.dir_fh = NFS_FH(dir),
-		.name = name,
-	};
-	struct nfs4_lookup_res res = {
-		.server = server,
-		.fattr = fattr,
-		.fh = fhandle,
-	};
-	struct rpc_message msg = {
-		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP],
-		.rpc_argp = &args,
-		.rpc_resp = &res,
-	};
-	
-	nfs_fattr_init(fattr);
+	int status;
 	
 	dprintk("NFS call  lookup %s\n", name->name);
-	status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
+	status = _nfs4_proc_lookupfh(NFS_SERVER(dir), NFS_FH(dir), name, fhandle, fattr);
 	if (status == -NFS4ERR_MOVED)
 		status = nfs4_get_referral(dir, name, fattr, fhandle);
 	dprintk("NFS reply lookup: %d\n", status);
@@ -1752,6 +1870,10 @@ static int
 nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
                  int flags, struct nameidata *nd)
 {
+	struct path path = {
+		.mnt = nd->mnt,
+		.dentry = dentry,
+	};
 	struct nfs4_state *state;
 	struct rpc_cred *cred;
 	int status = 0;
@@ -1761,7 +1883,7 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 		status = PTR_ERR(cred);
 		goto out;
 	}
-	state = nfs4_do_open(dir, dentry, flags, sattr, cred);
+	state = nfs4_do_open(dir, &path, flags, sattr, cred);
 	put_rpccred(cred);
 	if (IS_ERR(state)) {
 		status = PTR_ERR(state);
@@ -1773,11 +1895,12 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 		status = nfs4_do_setattr(state->inode, &fattr, sattr, state);
 		if (status == 0)
 			nfs_setattr_update_inode(state->inode, sattr);
+		nfs_post_op_update_inode(state->inode, &fattr);
 	}
-	if (status == 0 && nd != NULL && (nd->flags & LOOKUP_OPEN))
-		status = nfs4_intent_set_file(nd, dentry, state);
+	if (status == 0 && (nd->flags & LOOKUP_OPEN) != 0)
+		status = nfs4_intent_set_file(nd, &path, state);
 	else
-		nfs4_close_state(state, flags);
+		nfs4_close_state(&path, state, flags);
 out:
 	return status;
 }
@@ -3008,7 +3131,7 @@ 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_id.id;
 	status = rpc_call_sync(server->client, &msg, 0);
 	switch (status) {
 		case 0:
@@ -3152,6 +3275,11 @@ static struct rpc_task *nfs4_do_unlck(struct file_lock *fl,
 {
 	struct nfs4_unlockdata *data;
 
+	/* Ensure this is an unlock - when canceling a lock, the
+	 * canceled lock is passed in, and it won't be an unlock.
+	 */
+	fl->fl_type = F_UNLCK;
+
 	data = nfs4_alloc_unlockdata(fl, ctx, lsp, seqid);
 	if (data == NULL) {
 		nfs_free_seqid(seqid);
@@ -3222,7 +3350,7 @@ static struct nfs4_lockdata *nfs4_alloc_lockdata(struct file_lock *fl,
 		goto out_free;
 	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_id.id;
 	p->lsp = lsp;
 	atomic_inc(&lsp->ls_count);
 	p->ctx = get_nfs_open_context(ctx);
@@ -3285,7 +3413,7 @@ static void nfs4_lock_done(struct rpc_task *task, void *calldata)
 		memcpy(data->lsp->ls_stateid.data, data->res.stateid.data,
 					sizeof(data->lsp->ls_stateid.data));
 		data->lsp->ls_flags |= NFS_LOCK_INITIALIZED;
-		renew_lease(NFS_SERVER(data->ctx->dentry->d_inode), data->timestamp);
+		renew_lease(NFS_SERVER(data->ctx->path.dentry->d_inode), data->timestamp);
 	}
 	nfs_increment_lock_seqid(data->rpc_status, data->arg.lock_seqid);
 out:

fs/nfs/nfs4state.c

@@ -38,12 +38,14 @@
  * subsequent patch.
  */
 
+#include <linux/kernel.h>
 #include <linux/slab.h>
 #include <linux/smp_lock.h>
 #include <linux/nfs_fs.h>
 #include <linux/nfs_idmap.h>
 #include <linux/kthread.h>
 #include <linux/module.h>
+#include <linux/random.h>
 #include <linux/workqueue.h>
 #include <linux/bitops.h>
 
@@ -69,33 +71,14 @@ static int nfs4_init_client(struct nfs_client *clp, struct rpc_cred *cred)
 	return status;
 }
 
-u32
-nfs4_alloc_lockowner_id(struct nfs_client *clp)
-{
-	return clp->cl_lockowner_id ++;
-}
-
-static struct nfs4_state_owner *
-nfs4_client_grab_unused(struct nfs_client *clp, struct rpc_cred *cred)
-{
-	struct nfs4_state_owner *sp = NULL;
-
-	if (!list_empty(&clp->cl_unused)) {
-		sp = list_entry(clp->cl_unused.next, struct nfs4_state_owner, so_list);
-		atomic_inc(&sp->so_count);
-		sp->so_cred = cred;
-		list_move(&sp->so_list, &clp->cl_state_owners);
-		clp->cl_nunused--;
-	}
-	return sp;
-}
-
 struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp)
 {
 	struct nfs4_state_owner *sp;
+	struct rb_node *pos;
 	struct rpc_cred *cred = NULL;
 
-	list_for_each_entry(sp, &clp->cl_state_owners, so_list) {
+	for (pos = rb_first(&clp->cl_state_owners); pos != NULL; pos = rb_next(pos)) {
+		sp = rb_entry(pos, struct nfs4_state_owner, so_client_node);
 		if (list_empty(&sp->so_states))
 			continue;
 		cred = get_rpccred(sp->so_cred);
@@ -107,32 +90,146 @@ struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp)
 static struct rpc_cred *nfs4_get_setclientid_cred(struct nfs_client *clp)
 {
 	struct nfs4_state_owner *sp;
+	struct rb_node *pos;
 
-	if (!list_empty(&clp->cl_state_owners)) {
-		sp = list_entry(clp->cl_state_owners.next,
-				struct nfs4_state_owner, so_list);
+	pos = rb_first(&clp->cl_state_owners);
+	if (pos != NULL) {
+		sp = rb_entry(pos, struct nfs4_state_owner, so_client_node);
 		return get_rpccred(sp->so_cred);
 	}
 	return NULL;
 }
 
+static void nfs_alloc_unique_id(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(struct nfs_client *clp, struct rpc_cred *cred)
+nfs4_find_state_owner(struct nfs_server *server, struct rpc_cred *cred)
 {
+	struct nfs_client *clp = server->nfs_client;
+	struct rb_node **p = &clp->cl_state_owners.rb_node,
+		       *parent = NULL;
 	struct nfs4_state_owner *sp, *res = NULL;
 
-	list_for_each_entry(sp, &clp->cl_state_owners, so_list) {
-		if (sp->so_cred != cred)
+	while (*p != NULL) {
+		parent = *p;
+		sp = rb_entry(parent, struct nfs4_state_owner, so_client_node);
+
+		if (server < sp->so_server) {
+			p = &parent->rb_left;
 			continue;
-		atomic_inc(&sp->so_count);
-		/* Move to the head of the list */
-		list_move(&sp->so_list, &clp->cl_state_owners);
-		res = sp;
-		break;
+		}
+		if (server > sp->so_server) {
+			p = &parent->rb_right;
+			continue;
+		}
+		if (cred < sp->so_cred)
+			p = &parent->rb_left;
+		else if (cred > sp->so_cred)
+			p = &parent->rb_right;
+		else {
+			atomic_inc(&sp->so_count);
+			res = sp;
+			break;
+		}
 	}
 	return res;
 }
 
+static struct nfs4_state_owner *
+nfs4_insert_state_owner(struct nfs_client *clp, struct nfs4_state_owner *new)
+{
+	struct rb_node **p = &clp->cl_state_owners.rb_node,
+		       *parent = NULL;
+	struct nfs4_state_owner *sp;
+
+	while (*p != NULL) {
+		parent = *p;
+		sp = rb_entry(parent, struct nfs4_state_owner, so_client_node);
+
+		if (new->so_server < sp->so_server) {
+			p = &parent->rb_left;
+			continue;
+		}
+		if (new->so_server > sp->so_server) {
+			p = &parent->rb_right;
+			continue;
+		}
+		if (new->so_cred < sp->so_cred)
+			p = &parent->rb_left;
+		else if (new->so_cred > sp->so_cred)
+			p = &parent->rb_right;
+		else {
+			atomic_inc(&sp->so_count);
+			return sp;
+		}
+	}
+	nfs_alloc_unique_id(&clp->cl_openowner_id, &new->so_owner_id, 1, 64);
+	rb_link_node(&new->so_client_node, parent, p);
+	rb_insert_color(&new->so_client_node, &clp->cl_state_owners);
+	return new;
+}
+
+static void
+nfs4_remove_state_owner(struct nfs_client *clp, struct nfs4_state_owner *sp)
+{
+	if (!RB_EMPTY_NODE(&sp->so_client_node))
+		rb_erase(&sp->so_client_node, &clp->cl_state_owners);
+	nfs_free_unique_id(&clp->cl_openowner_id, &sp->so_owner_id);
+}
+
 /*
  * nfs4_alloc_state_owner(): this is called on the OPEN or CREATE path to
  * create a new state_owner.
@@ -160,10 +257,14 @@ nfs4_alloc_state_owner(void)
 void
 nfs4_drop_state_owner(struct nfs4_state_owner *sp)
 {
-	struct nfs_client *clp = sp->so_client;
-	spin_lock(&clp->cl_lock);
-	list_del_init(&sp->so_list);
-	spin_unlock(&clp->cl_lock);
+	if (!RB_EMPTY_NODE(&sp->so_client_node)) {
+		struct nfs_client *clp = sp->so_client;
+
+		spin_lock(&clp->cl_lock);
+		rb_erase(&sp->so_client_node, &clp->cl_state_owners);
+		RB_CLEAR_NODE(&sp->so_client_node);
+		spin_unlock(&clp->cl_lock);
+	}
 }
 
 /*
@@ -175,26 +276,25 @@ struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *server, struct
 	struct nfs_client *clp = server->nfs_client;
 	struct nfs4_state_owner *sp, *new;
 
-	get_rpccred(cred);
-	new = nfs4_alloc_state_owner();
 	spin_lock(&clp->cl_lock);
-	sp = nfs4_find_state_owner(clp, cred);
-	if (sp == NULL)
-		sp = nfs4_client_grab_unused(clp, cred);
-	if (sp == NULL && new != NULL) {
-		list_add(&new->so_list, &clp->cl_state_owners);
-		new->so_client = clp;
-		new->so_id = nfs4_alloc_lockowner_id(clp);
-		new->so_cred = cred;
-		sp = new;
-		new = NULL;
-	}
+	sp = nfs4_find_state_owner(server, cred);
 	spin_unlock(&clp->cl_lock);
-	kfree(new);
 	if (sp != NULL)
 		return sp;
-	put_rpccred(cred);
-	return NULL;
+	new = nfs4_alloc_state_owner();
+	if (new == NULL)
+		return NULL;
+	new->so_client = clp;
+	new->so_server = server;
+	new->so_cred = cred;
+	spin_lock(&clp->cl_lock);
+	sp = nfs4_insert_state_owner(clp, new);
+	spin_unlock(&clp->cl_lock);
+	if (sp == new)
+		get_rpccred(cred);
+	else
+		kfree(new);
+	return sp;
 }
 
 /*
@@ -208,18 +308,7 @@ void nfs4_put_state_owner(struct nfs4_state_owner *sp)
 
 	if (!atomic_dec_and_lock(&sp->so_count, &clp->cl_lock))
 		return;
-	if (clp->cl_nunused >= OPENOWNER_POOL_SIZE)
-		goto out_free;
-	if (list_empty(&sp->so_list))
-		goto out_free;
-	list_move(&sp->so_list, &clp->cl_unused);
-	clp->cl_nunused++;
-	spin_unlock(&clp->cl_lock);
-	put_rpccred(cred);
-	cred = NULL;
-	return;
-out_free:
-	list_del(&sp->so_list);
+	nfs4_remove_state_owner(clp, sp);
 	spin_unlock(&clp->cl_lock);
 	put_rpccred(cred);
 	kfree(sp);
@@ -236,6 +325,7 @@ nfs4_alloc_open_state(void)
 	atomic_set(&state->count, 1);
 	INIT_LIST_HEAD(&state->lock_states);
 	spin_lock_init(&state->state_lock);
+	seqlock_init(&state->seqlock);
 	return state;
 }
 
@@ -263,13 +353,10 @@ __nfs4_find_state_byowner(struct inode *inode, struct nfs4_state_owner *owner)
 	struct nfs4_state *state;
 
 	list_for_each_entry(state, &nfsi->open_states, inode_states) {
-		/* Is this in the process of being freed? */
-		if (state->state == 0)
+		if (state->owner != owner)
 			continue;
-		if (state->owner == owner) {
-			atomic_inc(&state->count);
+		if (atomic_inc_not_zero(&state->count))
 			return state;
-		}
 	}
 	return NULL;
 }
@@ -341,16 +428,15 @@ void nfs4_put_open_state(struct nfs4_state *state)
 /*
  * Close the current file.
  */
-void nfs4_close_state(struct nfs4_state *state, mode_t mode)
+void nfs4_close_state(struct path *path, struct nfs4_state *state, mode_t mode)
 {
-	struct inode *inode = state->inode;
 	struct nfs4_state_owner *owner = state->owner;
-	int oldstate, newstate = 0;
+	int call_close = 0;
+	int newstate;
 
 	atomic_inc(&owner->so_count);
 	/* Protect against nfs4_find_state() */
 	spin_lock(&owner->so_lock);
-	spin_lock(&inode->i_lock);
 	switch (mode & (FMODE_READ | FMODE_WRITE)) {
 		case FMODE_READ:
 			state->n_rdonly--;
@@ -361,24 +447,29 @@ void nfs4_close_state(struct nfs4_state *state, mode_t mode)
 		case FMODE_READ|FMODE_WRITE:
 			state->n_rdwr--;
 	}
-	oldstate = newstate = state->state;
+	newstate = FMODE_READ|FMODE_WRITE;
 	if (state->n_rdwr == 0) {
-		if (state->n_rdonly == 0)
+		if (state->n_rdonly == 0) {
 			newstate &= ~FMODE_READ;
-		if (state->n_wronly == 0)
+			call_close |= test_bit(NFS_O_RDONLY_STATE, &state->flags);
+			call_close |= test_bit(NFS_O_RDWR_STATE, &state->flags);
+		}
+		if (state->n_wronly == 0) {
 			newstate &= ~FMODE_WRITE;
+			call_close |= test_bit(NFS_O_WRONLY_STATE, &state->flags);
+			call_close |= test_bit(NFS_O_RDWR_STATE, &state->flags);
+		}
+		if (newstate == 0)
+			clear_bit(NFS_DELEGATED_STATE, &state->flags);
 	}
-	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
-		nfs4_state_set_mode_locked(state, newstate);
-		oldstate = newstate;
-	}
-	spin_unlock(&inode->i_lock);
+	nfs4_state_set_mode_locked(state, newstate);
 	spin_unlock(&owner->so_lock);
 
-	if (oldstate != newstate && nfs4_do_close(inode, state) == 0)
-		return;
-	nfs4_put_open_state(state);
-	nfs4_put_state_owner(owner);
+	if (!call_close) {
+		nfs4_put_open_state(state);
+		nfs4_put_state_owner(owner);
+	} else
+		nfs4_do_close(path, state);
 }
 
 /*
@@ -415,12 +506,22 @@ static struct nfs4_lock_state *nfs4_alloc_lock_state(struct nfs4_state *state, f
 	atomic_set(&lsp->ls_count, 1);
 	lsp->ls_owner = fl_owner;
 	spin_lock(&clp->cl_lock);
-	lsp->ls_id = nfs4_alloc_lockowner_id(clp);
+	nfs_alloc_unique_id(&clp->cl_lockowner_id, &lsp->ls_id, 1, 64);
 	spin_unlock(&clp->cl_lock);
 	INIT_LIST_HEAD(&lsp->ls_locks);
 	return lsp;
 }
 
+static void nfs4_free_lock_state(struct nfs4_lock_state *lsp)
+{
+	struct nfs_client *clp = lsp->ls_state->owner->so_client;
+
+	spin_lock(&clp->cl_lock);
+	nfs_free_unique_id(&clp->cl_lockowner_id, &lsp->ls_id);
+	spin_unlock(&clp->cl_lock);
+	kfree(lsp);
+}
+
 /*
  * Return a compatible lock_state. If no initialized lock_state structure
  * exists, return an uninitialized one.
@@ -450,7 +551,8 @@ static struct nfs4_lock_state *nfs4_get_lock_state(struct nfs4_state *state, fl_
 			return NULL;
 	}
 	spin_unlock(&state->state_lock);
-	kfree(new);
+	if (new != NULL)
+		nfs4_free_lock_state(new);
 	return lsp;
 }
 
@@ -471,7 +573,7 @@ void nfs4_put_lock_state(struct nfs4_lock_state *lsp)
 	if (list_empty(&state->lock_states))
 		clear_bit(LK_STATE_IN_USE, &state->flags);
 	spin_unlock(&state->state_lock);
-	kfree(lsp);
+	nfs4_free_lock_state(lsp);
 }
 
 static void nfs4_fl_copy_lock(struct file_lock *dst, struct file_lock *src)
@@ -513,8 +615,12 @@ int nfs4_set_lock_state(struct nfs4_state *state, struct file_lock *fl)
 void nfs4_copy_stateid(nfs4_stateid *dst, struct nfs4_state *state, fl_owner_t fl_owner)
 {
 	struct nfs4_lock_state *lsp;
+	int seq;
 
-	memcpy(dst, &state->stateid, sizeof(*dst));
+	do {
+		seq = read_seqbegin(&state->seqlock);
+		memcpy(dst, &state->stateid, sizeof(*dst));
+	} while (read_seqretry(&state->seqlock, seq));
 	if (test_bit(LK_STATE_IN_USE, &state->flags) == 0)
 		return;
 
@@ -557,12 +663,18 @@ void nfs_free_seqid(struct nfs_seqid *seqid)
  * failed with a seqid incrementing error -
  * see comments nfs_fs.h:seqid_mutating_error()
  */
-static inline void nfs_increment_seqid(int status, struct nfs_seqid *seqid)
+static void nfs_increment_seqid(int status, struct nfs_seqid *seqid)
 {
 	switch (status) {
 		case 0:
 			break;
 		case -NFS4ERR_BAD_SEQID:
+			if (seqid->sequence->flags & NFS_SEQID_CONFIRMED)
+				return;
+			printk(KERN_WARNING "NFS: v4 server returned a bad"
+					"sequence-id error on an"
+					"unconfirmed sequence %p!\n",
+					seqid->sequence);
 		case -NFS4ERR_STALE_CLIENTID:
 		case -NFS4ERR_STALE_STATEID:
 		case -NFS4ERR_BAD_STATEID:
@@ -586,7 +698,7 @@ void nfs_increment_open_seqid(int status, struct nfs_seqid *seqid)
 				struct nfs4_state_owner, so_seqid);
 		nfs4_drop_state_owner(sp);
 	}
-	return nfs_increment_seqid(status, seqid);
+	nfs_increment_seqid(status, seqid);
 }
 
 /*
@@ -596,7 +708,7 @@ void nfs_increment_open_seqid(int status, struct nfs_seqid *seqid)
  */
 void nfs_increment_lock_seqid(int status, struct nfs_seqid *seqid)
 {
-	return nfs_increment_seqid(status, seqid);
+	nfs_increment_seqid(status, seqid);
 }
 
 int nfs_wait_on_sequence(struct nfs_seqid *seqid, struct rpc_task *task)
@@ -748,15 +860,21 @@ out_err:
 static void nfs4_state_mark_reclaim(struct nfs_client *clp)
 {
 	struct nfs4_state_owner *sp;
+	struct rb_node *pos;
 	struct nfs4_state *state;
 	struct nfs4_lock_state *lock;
 
 	/* Reset all sequence ids to zero */
-	list_for_each_entry(sp, &clp->cl_state_owners, so_list) {
+	for (pos = rb_first(&clp->cl_state_owners); pos != NULL; pos = rb_next(pos)) {
+		sp = rb_entry(pos, struct nfs4_state_owner, so_client_node);
 		sp->so_seqid.counter = 0;
 		sp->so_seqid.flags = 0;
 		spin_lock(&sp->so_lock);
 		list_for_each_entry(state, &sp->so_states, open_states) {
+			clear_bit(NFS_DELEGATED_STATE, &state->flags);
+			clear_bit(NFS_O_RDONLY_STATE, &state->flags);
+			clear_bit(NFS_O_WRONLY_STATE, &state->flags);
+			clear_bit(NFS_O_RDWR_STATE, &state->flags);
 			list_for_each_entry(lock, &state->lock_states, ls_locks) {
 				lock->ls_seqid.counter = 0;
 				lock->ls_seqid.flags = 0;
@@ -771,6 +889,7 @@ static int reclaimer(void *ptr)
 {
 	struct nfs_client *clp = ptr;
 	struct nfs4_state_owner *sp;
+	struct rb_node *pos;
 	struct nfs4_state_recovery_ops *ops;
 	struct rpc_cred *cred;
 	int status = 0;
@@ -816,7 +935,8 @@ restart_loop:
 	/* Mark all delegations for reclaim */
 	nfs_delegation_mark_reclaim(clp);
 	/* Note: list is protected by exclusive lock on cl->cl_sem */
-	list_for_each_entry(sp, &clp->cl_state_owners, so_list) {
+	for (pos = rb_first(&clp->cl_state_owners); pos != NULL; pos = rb_next(pos)) {
+		sp = rb_entry(pos, struct nfs4_state_owner, so_client_node);
 		status = nfs4_reclaim_open_state(ops, sp);
 		if (status < 0) {
 			if (status == -NFS4ERR_NO_GRACE) {

fs/nfs/nfs4xdr.c

@@ -68,9 +68,10 @@ static int nfs4_stat_to_errno(int);
 #endif
 
 /* lock,open owner id: 
- * we currently use size 1 (u32) out of (NFS4_OPAQUE_LIMIT  >> 2)
+ * we currently use size 2 (u64) out of (NFS4_OPAQUE_LIMIT  >> 2)
  */
-#define owner_id_maxsz          (1 + 1)
+#define open_owner_id_maxsz	(1 + 4)
+#define lock_owner_id_maxsz	(1 + 4)
 #define compound_encode_hdr_maxsz	(3 + (NFS4_MAXTAGLEN >> 2))
 #define compound_decode_hdr_maxsz	(3 + (NFS4_MAXTAGLEN >> 2))
 #define op_encode_hdr_maxsz	(1)
@@ -87,9 +88,11 @@ static int nfs4_stat_to_errno(int);
 #define encode_getattr_maxsz    (op_encode_hdr_maxsz + nfs4_fattr_bitmap_maxsz)
 #define nfs4_name_maxsz		(1 + ((3 + NFS4_MAXNAMLEN) >> 2))
 #define nfs4_path_maxsz		(1 + ((3 + NFS4_MAXPATHLEN) >> 2))
+#define nfs4_owner_maxsz	(1 + XDR_QUADLEN(IDMAP_NAMESZ))
+#define nfs4_group_maxsz	(1 + XDR_QUADLEN(IDMAP_NAMESZ))
 /* This is based on getfattr, which uses the most attributes: */
 #define nfs4_fattr_value_maxsz	(1 + (1 + 2 + 2 + 4 + 2 + 1 + 1 + 2 + 2 + \
-				3 + 3 + 3 + 2 * nfs4_name_maxsz))
+				3 + 3 + 3 + nfs4_owner_maxsz + nfs4_group_maxsz))
 #define nfs4_fattr_maxsz	(nfs4_fattr_bitmap_maxsz + \
 				nfs4_fattr_value_maxsz)
 #define decode_getattr_maxsz    (op_decode_hdr_maxsz + nfs4_fattr_maxsz)
@@ -116,8 +119,27 @@ static int nfs4_stat_to_errno(int);
 				3 + (NFS4_VERIFIER_SIZE >> 2))
 #define decode_setclientid_confirm_maxsz \
 				(op_decode_hdr_maxsz)
-#define encode_lookup_maxsz	(op_encode_hdr_maxsz + \
-				1 + ((3 + NFS4_FHSIZE) >> 2))
+#define encode_lookup_maxsz	(op_encode_hdr_maxsz + nfs4_name_maxsz)
+#define decode_lookup_maxsz	(op_decode_hdr_maxsz)
+#define encode_share_access_maxsz \
+				(2)
+#define encode_createmode_maxsz	(1 + nfs4_fattr_maxsz)
+#define encode_opentype_maxsz	(1 + encode_createmode_maxsz)
+#define encode_claim_null_maxsz	(1 + nfs4_name_maxsz)
+#define encode_open_maxsz	(op_encode_hdr_maxsz + \
+				2 + encode_share_access_maxsz + 2 + \
+				open_owner_id_maxsz + \
+				encode_opentype_maxsz + \
+				encode_claim_null_maxsz)
+#define decode_ace_maxsz	(3 + nfs4_owner_maxsz)
+#define decode_delegation_maxsz	(1 + XDR_QUADLEN(NFS4_STATEID_SIZE) + 1 + \
+				decode_ace_maxsz)
+#define decode_change_info_maxsz	(5)
+#define decode_open_maxsz	(op_decode_hdr_maxsz + \
+				XDR_QUADLEN(NFS4_STATEID_SIZE) + \
+				decode_change_info_maxsz + 1 + \
+				nfs4_fattr_bitmap_maxsz + \
+				decode_delegation_maxsz)
 #define encode_remove_maxsz	(op_encode_hdr_maxsz + \
 				nfs4_name_maxsz)
 #define encode_rename_maxsz	(op_encode_hdr_maxsz + \
@@ -134,9 +156,15 @@ static int nfs4_stat_to_errno(int);
 #define encode_create_maxsz	(op_encode_hdr_maxsz + \
 				2 + nfs4_name_maxsz + \
 				nfs4_fattr_maxsz)
-#define decode_create_maxsz	(op_decode_hdr_maxsz + 8)
+#define decode_create_maxsz	(op_decode_hdr_maxsz + \
+				decode_change_info_maxsz + \
+				nfs4_fattr_bitmap_maxsz)
 #define encode_delegreturn_maxsz (op_encode_hdr_maxsz + 4)
 #define decode_delegreturn_maxsz (op_decode_hdr_maxsz)
+#define encode_fs_locations_maxsz \
+				(encode_getattr_maxsz)
+#define decode_fs_locations_maxsz \
+				(0)
 #define NFS4_enc_compound_sz	(1024)  /* XXX: large enough? */
 #define NFS4_dec_compound_sz	(1024)  /* XXX: large enough? */
 #define NFS4_enc_read_sz	(compound_encode_hdr_maxsz + \
@@ -174,16 +202,21 @@ static int nfs4_stat_to_errno(int);
 				op_decode_hdr_maxsz + 2 + \
 				decode_getattr_maxsz)
 #define NFS4_enc_open_sz        (compound_encode_hdr_maxsz + \
-                                encode_putfh_maxsz + \
-                                op_encode_hdr_maxsz + \
-                                13 + 3 + 2 + 64 + \
-                                encode_getattr_maxsz + \
-                                encode_getfh_maxsz)
+				encode_putfh_maxsz + \
+				encode_savefh_maxsz + \
+				encode_open_maxsz + \
+				encode_getfh_maxsz + \
+				encode_getattr_maxsz + \
+				encode_restorefh_maxsz + \
+				encode_getattr_maxsz)
 #define NFS4_dec_open_sz        (compound_decode_hdr_maxsz + \
-                                decode_putfh_maxsz + \
-                                op_decode_hdr_maxsz + 4 + 5 + 2 + 3 + \
-                                decode_getattr_maxsz + \
-                                decode_getfh_maxsz)
+				decode_putfh_maxsz + \
+				decode_savefh_maxsz + \
+				decode_open_maxsz + \
+				decode_getfh_maxsz + \
+				decode_getattr_maxsz + \
+				decode_restorefh_maxsz + \
+				decode_getattr_maxsz)
 #define NFS4_enc_open_confirm_sz      \
                                 (compound_encode_hdr_maxsz + \
                                 encode_putfh_maxsz + \
@@ -193,12 +226,12 @@ static int nfs4_stat_to_errno(int);
                                         op_decode_hdr_maxsz + 4)
 #define NFS4_enc_open_noattr_sz	(compound_encode_hdr_maxsz + \
 					encode_putfh_maxsz + \
-					op_encode_hdr_maxsz + \
-					11)
+					encode_open_maxsz + \
+					encode_getattr_maxsz)
 #define NFS4_dec_open_noattr_sz	(compound_decode_hdr_maxsz + \
 					decode_putfh_maxsz + \
-					op_decode_hdr_maxsz + \
-					4 + 5 + 2 + 3)
+					decode_open_maxsz + \
+					decode_getattr_maxsz)
 #define NFS4_enc_open_downgrade_sz \
 				(compound_encode_hdr_maxsz + \
                                 encode_putfh_maxsz + \
@@ -256,19 +289,19 @@ static int nfs4_stat_to_errno(int);
 				op_encode_hdr_maxsz + \
 				1 + 1 + 2 + 2 + \
 				1 + 4 + 1 + 2 + \
-				owner_id_maxsz)
+				lock_owner_id_maxsz)
 #define NFS4_dec_lock_sz        (compound_decode_hdr_maxsz + \
 				decode_putfh_maxsz + \
 				decode_getattr_maxsz + \
 				op_decode_hdr_maxsz + \
 				2 + 2 + 1 + 2 + \
-				owner_id_maxsz)
+				lock_owner_id_maxsz)
 #define NFS4_enc_lockt_sz       (compound_encode_hdr_maxsz + \
 				encode_putfh_maxsz + \
 				encode_getattr_maxsz + \
 				op_encode_hdr_maxsz + \
 				1 + 2 + 2 + 2 + \
-				owner_id_maxsz)
+				lock_owner_id_maxsz)
 #define NFS4_dec_lockt_sz       (NFS4_dec_lock_sz)
 #define NFS4_enc_locku_sz       (compound_encode_hdr_maxsz + \
 				encode_putfh_maxsz + \
@@ -298,7 +331,7 @@ static int nfs4_stat_to_errno(int);
 				encode_getfh_maxsz)
 #define NFS4_dec_lookup_sz	(compound_decode_hdr_maxsz + \
 				decode_putfh_maxsz + \
-				op_decode_hdr_maxsz + \
+				decode_lookup_maxsz + \
 				decode_getattr_maxsz + \
 				decode_getfh_maxsz)
 #define NFS4_enc_lookup_root_sz (compound_encode_hdr_maxsz + \
@@ -417,12 +450,13 @@ static int nfs4_stat_to_errno(int);
 #define NFS4_enc_fs_locations_sz \
 				(compound_encode_hdr_maxsz + \
 				 encode_putfh_maxsz + \
-				 encode_getattr_maxsz)
+				 encode_lookup_maxsz + \
+				 encode_fs_locations_maxsz)
 #define NFS4_dec_fs_locations_sz \
 				(compound_decode_hdr_maxsz + \
 				 decode_putfh_maxsz + \
-				 op_decode_hdr_maxsz + \
-				 nfs4_fattr_bitmap_maxsz)
+				 decode_lookup_maxsz + \
+				 decode_fs_locations_maxsz)
 
 static struct {
 	unsigned int	mode;
@@ -793,13 +827,14 @@ static int encode_lock(struct xdr_stream *xdr, const struct nfs_lock_args *args)
 	WRITE64(nfs4_lock_length(args->fl));
 	WRITE32(args->new_lock_owner);
 	if (args->new_lock_owner){
-		RESERVE_SPACE(4+NFS4_STATEID_SIZE+20);
+		RESERVE_SPACE(4+NFS4_STATEID_SIZE+32);
 		WRITE32(args->open_seqid->sequence->counter);
 		WRITEMEM(args->open_stateid->data, NFS4_STATEID_SIZE);
 		WRITE32(args->lock_seqid->sequence->counter);
 		WRITE64(args->lock_owner.clientid);
-		WRITE32(4);
-		WRITE32(args->lock_owner.id);
+		WRITE32(16);
+		WRITEMEM("lock id:", 8);
+		WRITE64(args->lock_owner.id);
 	}
 	else {
 		RESERVE_SPACE(NFS4_STATEID_SIZE+4);
@@ -814,14 +849,15 @@ static int encode_lockt(struct xdr_stream *xdr, const struct nfs_lockt_args *arg
 {
 	__be32 *p;
 
-	RESERVE_SPACE(40);
+	RESERVE_SPACE(52);
 	WRITE32(OP_LOCKT);
 	WRITE32(nfs4_lock_type(args->fl, 0));
 	WRITE64(args->fl->fl_start);
 	WRITE64(nfs4_lock_length(args->fl));
 	WRITE64(args->lock_owner.clientid);
-	WRITE32(4);
-	WRITE32(args->lock_owner.id);
+	WRITE32(16);
+	WRITEMEM("lock id:", 8);
+	WRITE64(args->lock_owner.id);
 
 	return 0;
 }
@@ -886,10 +922,11 @@ static inline void encode_openhdr(struct xdr_stream *xdr, const struct nfs_opena
 	WRITE32(OP_OPEN);
 	WRITE32(arg->seqid->sequence->counter);
 	encode_share_access(xdr, arg->open_flags);
-	RESERVE_SPACE(16);
+	RESERVE_SPACE(28);
 	WRITE64(arg->clientid);
-	WRITE32(4);
-	WRITE32(arg->id);
+	WRITE32(16);
+	WRITEMEM("open id:", 8);
+	WRITE64(arg->id);
 }
 
 static inline void encode_createmode(struct xdr_stream *xdr, const struct nfs_openargs *arg)
@@ -1071,7 +1108,7 @@ static int encode_read(struct xdr_stream *xdr, const struct nfs_readargs *args)
 
 static int encode_readdir(struct xdr_stream *xdr, const struct nfs4_readdir_arg *readdir, struct rpc_rqst *req)
 {
-	struct rpc_auth *auth = req->rq_task->tk_auth;
+	struct rpc_auth *auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	uint32_t attrs[2] = {
 		FATTR4_WORD0_RDATTR_ERROR|FATTR4_WORD0_FILEID,
 		FATTR4_WORD1_MOUNTED_ON_FILEID,
@@ -1117,7 +1154,7 @@ static int encode_readdir(struct xdr_stream *xdr, const struct nfs4_readdir_arg
 
 static int encode_readlink(struct xdr_stream *xdr, const struct nfs4_readlink *readlink, struct rpc_rqst *req)
 {
-	struct rpc_auth *auth = req->rq_task->tk_auth;
+	struct rpc_auth *auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	unsigned int replen;
 	__be32 *p;
 
@@ -1735,7 +1772,7 @@ out:
  */
 static int nfs4_xdr_enc_read(struct rpc_rqst *req, __be32 *p, struct nfs_readargs *args)
 {
-	struct rpc_auth	*auth = req->rq_task->tk_auth;
+	struct rpc_auth *auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
 		.nops = 2,
@@ -1795,7 +1832,7 @@ nfs4_xdr_enc_getacl(struct rpc_rqst *req, __be32 *p,
 		struct nfs_getaclargs *args)
 {
 	struct xdr_stream xdr;
-	struct rpc_auth *auth = req->rq_task->tk_auth;
+	struct rpc_auth *auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	struct compound_hdr hdr = {
 		.nops   = 2,
 	};
@@ -2030,7 +2067,7 @@ static int nfs4_xdr_enc_fs_locations(struct rpc_rqst *req, __be32 *p, struct nfs
 	struct compound_hdr hdr = {
 		.nops = 3,
 	};
-	struct rpc_auth *auth = req->rq_task->tk_auth;
+	struct rpc_auth *auth = req->rq_task->tk_msg.rpc_cred->cr_auth;
 	int replen;
 	int status;
 
@@ -3269,7 +3306,7 @@ static int decode_delegation(struct xdr_stream *xdr, struct nfs_openres *res)
 static int decode_open(struct xdr_stream *xdr, struct nfs_openres *res)
 {
         __be32 *p;
-        uint32_t bmlen;
+	uint32_t savewords, bmlen, i;
         int status;
 
         status = decode_op_hdr(xdr, OP_OPEN);
@@ -3287,7 +3324,12 @@ static int decode_open(struct xdr_stream *xdr, struct nfs_openres *res)
                 goto xdr_error;
 
         READ_BUF(bmlen << 2);
-        p += bmlen;
+	savewords = min_t(uint32_t, bmlen, NFS4_BITMAP_SIZE);
+	for (i = 0; i < savewords; ++i)
+		READ32(res->attrset[i]);
+	for (; i < NFS4_BITMAP_SIZE; i++)
+		res->attrset[i] = 0;
+
 	return decode_delegation(xdr, res);
 xdr_error:
 	dprintk("%s: Bitmap too large! Length = %u\n", __FUNCTION__, bmlen);

fs/nfs/nfsroot.c

@@ -428,7 +428,7 @@ static int __init root_nfs_getport(int program, int version, int proto)
 	printk(KERN_NOTICE "Looking up port of RPC %d/%d on %u.%u.%u.%u\n",
 		program, version, NIPQUAD(servaddr));
 	set_sockaddr(&sin, servaddr, 0);
-	return rpcb_getport_external(&sin, program, version, proto);
+	return rpcb_getport_sync(&sin, program, version, proto);
 }
 
 
@@ -496,7 +496,8 @@ static int __init root_nfs_get_handle(void)
 					NFS_MNT3_VERSION : NFS_MNT_VERSION;
 
 	set_sockaddr(&sin, servaddr, htons(mount_port));
-	status = nfsroot_mount(&sin, nfs_path, &fh, version, protocol);
+	status = nfs_mount((struct sockaddr *) &sin, sizeof(sin), NULL,
+			   nfs_path, version, protocol, &fh);
 	if (status < 0)
 		printk(KERN_ERR "Root-NFS: Server returned error %d "
 				"while mounting %s\n", status, nfs_path);

fs/nfs/pagelist.c

@@ -85,9 +85,8 @@ nfs_create_request(struct nfs_open_context *ctx, struct inode *inode,
 	req->wb_offset  = offset;
 	req->wb_pgbase	= offset;
 	req->wb_bytes   = count;
-	atomic_set(&req->wb_count, 1);
 	req->wb_context = get_nfs_open_context(ctx);
-
+	kref_init(&req->wb_kref);
 	return req;
 }
 
@@ -109,30 +108,31 @@ void nfs_unlock_request(struct nfs_page *req)
 }
 
 /**
- * nfs_set_page_writeback_locked - Lock a request for writeback
+ * nfs_set_page_tag_locked - Tag a request as locked
  * @req:
  */
-int nfs_set_page_writeback_locked(struct nfs_page *req)
+static int nfs_set_page_tag_locked(struct nfs_page *req)
 {
-	struct nfs_inode *nfsi = NFS_I(req->wb_context->dentry->d_inode);
+	struct nfs_inode *nfsi = NFS_I(req->wb_context->path.dentry->d_inode);
 
 	if (!nfs_lock_request(req))
 		return 0;
-	radix_tree_tag_set(&nfsi->nfs_page_tree, req->wb_index, NFS_PAGE_TAG_WRITEBACK);
+	radix_tree_tag_set(&nfsi->nfs_page_tree, req->wb_index, NFS_PAGE_TAG_LOCKED);
 	return 1;
 }
 
 /**
- * nfs_clear_page_writeback - Unlock request and wake up sleepers
+ * nfs_clear_page_tag_locked - Clear request tag and wake up sleepers
  */
-void nfs_clear_page_writeback(struct nfs_page *req)
+void nfs_clear_page_tag_locked(struct nfs_page *req)
 {
-	struct nfs_inode *nfsi = NFS_I(req->wb_context->dentry->d_inode);
+	struct inode *inode = req->wb_context->path.dentry->d_inode;
+	struct nfs_inode *nfsi = NFS_I(inode);
 
 	if (req->wb_page != NULL) {
-		spin_lock(&nfsi->req_lock);
-		radix_tree_tag_clear(&nfsi->nfs_page_tree, req->wb_index, NFS_PAGE_TAG_WRITEBACK);
-		spin_unlock(&nfsi->req_lock);
+		spin_lock(&inode->i_lock);
+		radix_tree_tag_clear(&nfsi->nfs_page_tree, req->wb_index, NFS_PAGE_TAG_LOCKED);
+		spin_unlock(&inode->i_lock);
 	}
 	nfs_unlock_request(req);
 }
@@ -160,11 +160,9 @@ void nfs_clear_request(struct nfs_page *req)
  *
  * Note: Should never be called with the spinlock held!
  */
-void
-nfs_release_request(struct nfs_page *req)
+static void nfs_free_request(struct kref *kref)
 {
-	if (!atomic_dec_and_test(&req->wb_count))
-		return;
+	struct nfs_page *req = container_of(kref, struct nfs_page, wb_kref);
 
 	/* Release struct file or cached credential */
 	nfs_clear_request(req);
@@ -172,6 +170,11 @@ nfs_release_request(struct nfs_page *req)
 	nfs_page_free(req);
 }
 
+void nfs_release_request(struct nfs_page *req)
+{
+	kref_put(&req->wb_kref, nfs_free_request);
+}
+
 static int nfs_wait_bit_interruptible(void *word)
 {
 	int ret = 0;
@@ -193,7 +196,7 @@ static int nfs_wait_bit_interruptible(void *word)
 int
 nfs_wait_on_request(struct nfs_page *req)
 {
-        struct rpc_clnt	*clnt = NFS_CLIENT(req->wb_context->dentry->d_inode);
+	struct rpc_clnt *clnt = NFS_CLIENT(req->wb_context->path.dentry->d_inode);
 	sigset_t oldmask;
 	int ret = 0;
 
@@ -379,20 +382,20 @@ void nfs_pageio_cond_complete(struct nfs_pageio_descriptor *desc, pgoff_t index)
 /**
  * nfs_scan_list - Scan a list for matching requests
  * @nfsi: NFS inode
- * @head: One of the NFS inode request lists
  * @dst: Destination list
  * @idx_start: lower bound of page->index to scan
  * @npages: idx_start + npages sets the upper bound to scan.
+ * @tag: tag to scan for
  *
  * Moves elements from one of the inode request lists.
  * If the number of requests is set to 0, the entire address_space
  * starting at index idx_start, is scanned.
  * The requests are *not* checked to ensure that they form a contiguous set.
- * You must be holding the inode's req_lock when calling this function
+ * You must be holding the inode's i_lock when calling this function
  */
-int nfs_scan_list(struct nfs_inode *nfsi, struct list_head *head,
+int nfs_scan_list(struct nfs_inode *nfsi,
 		struct list_head *dst, pgoff_t idx_start,
-		unsigned int npages)
+		unsigned int npages, int tag)
 {
 	struct nfs_page *pgvec[NFS_SCAN_MAXENTRIES];
 	struct nfs_page *req;
@@ -407,9 +410,9 @@ int nfs_scan_list(struct nfs_inode *nfsi, struct list_head *head,
 		idx_end = idx_start + npages - 1;
 
 	for (;;) {
-		found = radix_tree_gang_lookup(&nfsi->nfs_page_tree,
+		found = radix_tree_gang_lookup_tag(&nfsi->nfs_page_tree,
 				(void **)&pgvec[0], idx_start,
-				NFS_SCAN_MAXENTRIES);
+				NFS_SCAN_MAXENTRIES, tag);
 		if (found <= 0)
 			break;
 		for (i = 0; i < found; i++) {
@@ -417,15 +420,18 @@ int nfs_scan_list(struct nfs_inode *nfsi, struct list_head *head,
 			if (req->wb_index > idx_end)
 				goto out;
 			idx_start = req->wb_index + 1;
-			if (req->wb_list_head != head)
-				continue;
-			if (nfs_set_page_writeback_locked(req)) {
+			if (nfs_set_page_tag_locked(req)) {
 				nfs_list_remove_request(req);
+				radix_tree_tag_clear(&nfsi->nfs_page_tree,
+						req->wb_index, tag);
 				nfs_list_add_request(req, dst);
 				res++;
+				if (res == INT_MAX)
+					goto out;
 			}
 		}
-
+		/* for latency reduction */
+		cond_resched_lock(&nfsi->vfs_inode.i_lock);
 	}
 out:
 	return res;

fs/nfs/read.c

@@ -145,8 +145,8 @@ static void nfs_readpage_release(struct nfs_page *req)
 	unlock_page(req->wb_page);
 
 	dprintk("NFS: read done (%s/%Ld %d@%Ld)\n",
-			req->wb_context->dentry->d_inode->i_sb->s_id,
-			(long long)NFS_FILEID(req->wb_context->dentry->d_inode),
+			req->wb_context->path.dentry->d_inode->i_sb->s_id,
+			(long long)NFS_FILEID(req->wb_context->path.dentry->d_inode),
 			req->wb_bytes,
 			(long long)req_offset(req));
 	nfs_clear_request(req);
@@ -164,7 +164,7 @@ static void nfs_read_rpcsetup(struct nfs_page *req, struct nfs_read_data *data,
 	int flags;
 
 	data->req	  = req;
-	data->inode	  = inode = req->wb_context->dentry->d_inode;
+	data->inode	  = inode = req->wb_context->path.dentry->d_inode;
 	data->cred	  = req->wb_context->cred;
 
 	data->args.fh     = NFS_FH(inode);
@@ -483,17 +483,19 @@ int nfs_readpage(struct file *file, struct page *page)
 	 */
 	error = nfs_wb_page(inode, page);
 	if (error)
-		goto out_error;
+		goto out_unlock;
+	if (PageUptodate(page))
+		goto out_unlock;
 
 	error = -ESTALE;
 	if (NFS_STALE(inode))
-		goto out_error;
+		goto out_unlock;
 
 	if (file == NULL) {
 		error = -EBADF;
 		ctx = nfs_find_open_context(inode, NULL, FMODE_READ);
 		if (ctx == NULL)
-			goto out_error;
+			goto out_unlock;
 	} else
 		ctx = get_nfs_open_context((struct nfs_open_context *)
 				file->private_data);
@@ -502,8 +504,7 @@ int nfs_readpage(struct file *file, struct page *page)
 
 	put_nfs_open_context(ctx);
 	return error;
-
-out_error:
+out_unlock:
 	unlock_page(page);
 	return error;
 }
@@ -520,21 +521,32 @@ readpage_async_filler(void *data, struct page *page)
 	struct inode *inode = page->mapping->host;
 	struct nfs_page *new;
 	unsigned int len;
+	int error;
+
+	error = nfs_wb_page(inode, page);
+	if (error)
+		goto out_unlock;
+	if (PageUptodate(page))
+		goto out_unlock;
 
-	nfs_wb_page(inode, page);
 	len = nfs_page_length(page);
 	if (len == 0)
 		return nfs_return_empty_page(page);
+
 	new = nfs_create_request(desc->ctx, inode, page, 0, len);
-	if (IS_ERR(new)) {
-			SetPageError(page);
-			unlock_page(page);
-			return PTR_ERR(new);
-	}
+	if (IS_ERR(new))
+		goto out_error;
+
 	if (len < PAGE_CACHE_SIZE)
 		zero_user_page(page, len, PAGE_CACHE_SIZE - len, KM_USER0);
 	nfs_pageio_add_request(desc->pgio, new);
 	return 0;
+out_error:
+	error = PTR_ERR(new);
+	SetPageError(page);
+out_unlock:
+	unlock_page(page);
+	return error;
 }
 
 int nfs_readpages(struct file *filp, struct address_space *mapping,

fs/nfs/super.c

@@ -45,6 +45,7 @@
 #include <linux/inet.h>
 #include <linux/nfs_xdr.h>
 #include <linux/magic.h>
+#include <linux/parser.h>
 
 #include <asm/system.h>
 #include <asm/uaccess.h>
@@ -57,6 +58,167 @@
 
 #define NFSDBG_FACILITY		NFSDBG_VFS
 
+
+struct nfs_parsed_mount_data {
+	int			flags;
+	int			rsize, wsize;
+	int			timeo, retrans;
+	int			acregmin, acregmax,
+				acdirmin, acdirmax;
+	int			namlen;
+	unsigned int		bsize;
+	unsigned int		auth_flavor_len;
+	rpc_authflavor_t	auth_flavors[1];
+	char			*client_address;
+
+	struct {
+		struct sockaddr_in	address;
+		unsigned int		program;
+		unsigned int		version;
+		unsigned short		port;
+		int			protocol;
+	} mount_server;
+
+	struct {
+		struct sockaddr_in	address;
+		char			*hostname;
+		char			*export_path;
+		unsigned int		program;
+		int			protocol;
+	} nfs_server;
+};
+
+enum {
+	/* Mount options that take no arguments */
+	Opt_soft, Opt_hard,
+	Opt_intr, Opt_nointr,
+	Opt_posix, Opt_noposix,
+	Opt_cto, Opt_nocto,
+	Opt_ac, Opt_noac,
+	Opt_lock, Opt_nolock,
+	Opt_v2, Opt_v3,
+	Opt_udp, Opt_tcp,
+	Opt_acl, Opt_noacl,
+	Opt_rdirplus, Opt_nordirplus,
+	Opt_sharecache, Opt_nosharecache,
+
+	/* Mount options that take integer arguments */
+	Opt_port,
+	Opt_rsize, Opt_wsize, Opt_bsize,
+	Opt_timeo, Opt_retrans,
+	Opt_acregmin, Opt_acregmax,
+	Opt_acdirmin, Opt_acdirmax,
+	Opt_actimeo,
+	Opt_namelen,
+	Opt_mountport,
+	Opt_mountprog, Opt_mountvers,
+	Opt_nfsprog, Opt_nfsvers,
+
+	/* Mount options that take string arguments */
+	Opt_sec, Opt_proto, Opt_mountproto,
+	Opt_addr, Opt_mounthost, Opt_clientaddr,
+
+	/* Mount options that are ignored */
+	Opt_userspace, Opt_deprecated,
+
+	Opt_err
+};
+
+static match_table_t nfs_mount_option_tokens = {
+	{ Opt_userspace, "bg" },
+	{ Opt_userspace, "fg" },
+	{ Opt_soft, "soft" },
+	{ Opt_hard, "hard" },
+	{ Opt_intr, "intr" },
+	{ Opt_nointr, "nointr" },
+	{ Opt_posix, "posix" },
+	{ Opt_noposix, "noposix" },
+	{ Opt_cto, "cto" },
+	{ Opt_nocto, "nocto" },
+	{ Opt_ac, "ac" },
+	{ Opt_noac, "noac" },
+	{ Opt_lock, "lock" },
+	{ Opt_nolock, "nolock" },
+	{ Opt_v2, "v2" },
+	{ Opt_v3, "v3" },
+	{ Opt_udp, "udp" },
+	{ Opt_tcp, "tcp" },
+	{ Opt_acl, "acl" },
+	{ Opt_noacl, "noacl" },
+	{ Opt_rdirplus, "rdirplus" },
+	{ Opt_nordirplus, "nordirplus" },
+	{ Opt_sharecache, "sharecache" },
+	{ Opt_nosharecache, "nosharecache" },
+
+	{ Opt_port, "port=%u" },
+	{ Opt_rsize, "rsize=%u" },
+	{ Opt_wsize, "wsize=%u" },
+	{ Opt_bsize, "bsize=%u" },
+	{ Opt_timeo, "timeo=%u" },
+	{ Opt_retrans, "retrans=%u" },
+	{ Opt_acregmin, "acregmin=%u" },
+	{ Opt_acregmax, "acregmax=%u" },
+	{ Opt_acdirmin, "acdirmin=%u" },
+	{ Opt_acdirmax, "acdirmax=%u" },
+	{ Opt_actimeo, "actimeo=%u" },
+	{ Opt_userspace, "retry=%u" },
+	{ Opt_namelen, "namlen=%u" },
+	{ Opt_mountport, "mountport=%u" },
+	{ Opt_mountprog, "mountprog=%u" },
+	{ Opt_mountvers, "mountvers=%u" },
+	{ Opt_nfsprog, "nfsprog=%u" },
+	{ Opt_nfsvers, "nfsvers=%u" },
+	{ Opt_nfsvers, "vers=%u" },
+
+	{ Opt_sec, "sec=%s" },
+	{ Opt_proto, "proto=%s" },
+	{ Opt_mountproto, "mountproto=%s" },
+	{ Opt_addr, "addr=%s" },
+	{ Opt_clientaddr, "clientaddr=%s" },
+	{ Opt_mounthost, "mounthost=%s" },
+
+	{ Opt_err, NULL }
+};
+
+enum {
+	Opt_xprt_udp, Opt_xprt_tcp,
+
+	Opt_xprt_err
+};
+
+static match_table_t nfs_xprt_protocol_tokens = {
+	{ Opt_xprt_udp, "udp" },
+	{ Opt_xprt_tcp, "tcp" },
+
+	{ Opt_xprt_err, NULL }
+};
+
+enum {
+	Opt_sec_none, Opt_sec_sys,
+	Opt_sec_krb5, Opt_sec_krb5i, Opt_sec_krb5p,
+	Opt_sec_lkey, Opt_sec_lkeyi, Opt_sec_lkeyp,
+	Opt_sec_spkm, Opt_sec_spkmi, Opt_sec_spkmp,
+
+	Opt_sec_err
+};
+
+static match_table_t nfs_secflavor_tokens = {
+	{ Opt_sec_none, "none" },
+	{ Opt_sec_none, "null" },
+	{ Opt_sec_sys, "sys" },
+
+	{ Opt_sec_krb5, "krb5" },
+	{ Opt_sec_krb5i, "krb5i" },
+	{ Opt_sec_krb5p, "krb5p" },
+
+	{ Opt_sec_lkey, "lkey" },
+	{ Opt_sec_lkeyi, "lkeyi" },
+	{ Opt_sec_lkeyp, "lkeyp" },
+
+	{ Opt_sec_err, NULL }
+};
+
+
 static void nfs_umount_begin(struct vfsmount *, int);
 static int  nfs_statfs(struct dentry *, struct kstatfs *);
 static int  nfs_show_options(struct seq_file *, struct vfsmount *);
@@ -263,11 +425,11 @@ static const char *nfs_pseudoflavour_to_name(rpc_authflavor_t flavour)
 		{ RPC_AUTH_GSS_SPKM, "spkm" },
 		{ RPC_AUTH_GSS_SPKMI, "spkmi" },
 		{ RPC_AUTH_GSS_SPKMP, "spkmp" },
-		{ -1, "unknown" }
+		{ UINT_MAX, "unknown" }
 	};
 	int i;
 
-	for (i=0; sec_flavours[i].flavour != -1; i++) {
+	for (i = 0; sec_flavours[i].flavour != UINT_MAX; i++) {
 		if (sec_flavours[i].flavour == flavour)
 			break;
 	}
@@ -291,6 +453,7 @@ static void nfs_show_mount_options(struct seq_file *m, struct nfs_server *nfss,
 		{ NFS_MOUNT_NONLM, ",nolock", "" },
 		{ NFS_MOUNT_NOACL, ",noacl", "" },
 		{ NFS_MOUNT_NORDIRPLUS, ",nordirplus", "" },
+		{ NFS_MOUNT_UNSHARED, ",nosharecache", ""},
 		{ 0, NULL, NULL }
 	};
 	const struct proc_nfs_info *nfs_infop;
@@ -430,87 +593,641 @@ static int nfs_show_stats(struct seq_file *m, struct vfsmount *mnt)
  */
 static void nfs_umount_begin(struct vfsmount *vfsmnt, int flags)
 {
+	struct nfs_server *server = NFS_SB(vfsmnt->mnt_sb);
+	struct rpc_clnt *rpc;
+
 	shrink_submounts(vfsmnt, &nfs_automount_list);
+
+	if (!(flags & MNT_FORCE))
+		return;
+	/* -EIO all pending I/O */
+	rpc = server->client_acl;
+	if (!IS_ERR(rpc))
+		rpc_killall_tasks(rpc);
+	rpc = server->client;
+	if (!IS_ERR(rpc))
+		rpc_killall_tasks(rpc);
 }
 
 /*
- * Validate the NFS2/NFS3 mount data
- * - fills in the mount root filehandle
+ * Sanity-check a server address provided by the mount command
  */
-static int nfs_validate_mount_data(struct nfs_mount_data *data,
-				   struct nfs_fh *mntfh)
+static int nfs_verify_server_address(struct sockaddr *addr)
 {
-	if (data == NULL) {
-		dprintk("%s: missing data argument\n", __FUNCTION__);
-		return -EINVAL;
+	switch (addr->sa_family) {
+	case AF_INET: {
+		struct sockaddr_in *sa = (struct sockaddr_in *) addr;
+		if (sa->sin_addr.s_addr != INADDR_ANY)
+			return 1;
+		break;
+	}
 	}
 
-	if (data->version <= 0 || data->version > NFS_MOUNT_VERSION) {
-		dprintk("%s: bad mount version\n", __FUNCTION__);
-		return -EINVAL;
+	return 0;
+}
+
+/*
+ * Error-check and convert a string of mount options from user space into
+ * a data structure
+ */
+static int nfs_parse_mount_options(char *raw,
+				   struct nfs_parsed_mount_data *mnt)
+{
+	char *p, *string;
+
+	if (!raw) {
+		dfprintk(MOUNT, "NFS: mount options string was NULL.\n");
+		return 1;
 	}
+	dfprintk(MOUNT, "NFS: nfs mount opts='%s'\n", raw);
 
-	switch (data->version) {
-		case 1:
-			data->namlen = 0;
-		case 2:
-			data->bsize  = 0;
-		case 3:
-			if (data->flags & NFS_MOUNT_VER3) {
-				dprintk("%s: mount structure version %d does not support NFSv3\n",
-						__FUNCTION__,
-						data->version);
-				return -EINVAL;
+	while ((p = strsep(&raw, ",")) != NULL) {
+		substring_t args[MAX_OPT_ARGS];
+		int option, token;
+
+		if (!*p)
+			continue;
+
+		dfprintk(MOUNT, "NFS:   parsing nfs mount option '%s'\n", p);
+
+		token = match_token(p, nfs_mount_option_tokens, args);
+		switch (token) {
+		case Opt_soft:
+			mnt->flags |= NFS_MOUNT_SOFT;
+			break;
+		case Opt_hard:
+			mnt->flags &= ~NFS_MOUNT_SOFT;
+			break;
+		case Opt_intr:
+			mnt->flags |= NFS_MOUNT_INTR;
+			break;
+		case Opt_nointr:
+			mnt->flags &= ~NFS_MOUNT_INTR;
+			break;
+		case Opt_posix:
+			mnt->flags |= NFS_MOUNT_POSIX;
+			break;
+		case Opt_noposix:
+			mnt->flags &= ~NFS_MOUNT_POSIX;
+			break;
+		case Opt_cto:
+			mnt->flags &= ~NFS_MOUNT_NOCTO;
+			break;
+		case Opt_nocto:
+			mnt->flags |= NFS_MOUNT_NOCTO;
+			break;
+		case Opt_ac:
+			mnt->flags &= ~NFS_MOUNT_NOAC;
+			break;
+		case Opt_noac:
+			mnt->flags |= NFS_MOUNT_NOAC;
+			break;
+		case Opt_lock:
+			mnt->flags &= ~NFS_MOUNT_NONLM;
+			break;
+		case Opt_nolock:
+			mnt->flags |= NFS_MOUNT_NONLM;
+			break;
+		case Opt_v2:
+			mnt->flags &= ~NFS_MOUNT_VER3;
+			break;
+		case Opt_v3:
+			mnt->flags |= NFS_MOUNT_VER3;
+			break;
+		case Opt_udp:
+			mnt->flags &= ~NFS_MOUNT_TCP;
+			mnt->nfs_server.protocol = IPPROTO_UDP;
+			mnt->timeo = 7;
+			mnt->retrans = 5;
+			break;
+		case Opt_tcp:
+			mnt->flags |= NFS_MOUNT_TCP;
+			mnt->nfs_server.protocol = IPPROTO_TCP;
+			mnt->timeo = 600;
+			mnt->retrans = 2;
+			break;
+		case Opt_acl:
+			mnt->flags &= ~NFS_MOUNT_NOACL;
+			break;
+		case Opt_noacl:
+			mnt->flags |= NFS_MOUNT_NOACL;
+			break;
+		case Opt_rdirplus:
+			mnt->flags &= ~NFS_MOUNT_NORDIRPLUS;
+			break;
+		case Opt_nordirplus:
+			mnt->flags |= NFS_MOUNT_NORDIRPLUS;
+			break;
+		case Opt_sharecache:
+			mnt->flags &= ~NFS_MOUNT_UNSHARED;
+			break;
+		case Opt_nosharecache:
+			mnt->flags |= NFS_MOUNT_UNSHARED;
+			break;
+
+		case Opt_port:
+			if (match_int(args, &option))
+				return 0;
+			if (option < 0 || option > 65535)
+				return 0;
+			mnt->nfs_server.address.sin_port = htonl(option);
+			break;
+		case Opt_rsize:
+			if (match_int(args, &mnt->rsize))
+				return 0;
+			break;
+		case Opt_wsize:
+			if (match_int(args, &mnt->wsize))
+				return 0;
+			break;
+		case Opt_bsize:
+			if (match_int(args, &option))
+				return 0;
+			if (option < 0)
+				return 0;
+			mnt->bsize = option;
+			break;
+		case Opt_timeo:
+			if (match_int(args, &mnt->timeo))
+				return 0;
+			break;
+		case Opt_retrans:
+			if (match_int(args, &mnt->retrans))
+				return 0;
+			break;
+		case Opt_acregmin:
+			if (match_int(args, &mnt->acregmin))
+				return 0;
+			break;
+		case Opt_acregmax:
+			if (match_int(args, &mnt->acregmax))
+				return 0;
+			break;
+		case Opt_acdirmin:
+			if (match_int(args, &mnt->acdirmin))
+				return 0;
+			break;
+		case Opt_acdirmax:
+			if (match_int(args, &mnt->acdirmax))
+				return 0;
+			break;
+		case Opt_actimeo:
+			if (match_int(args, &option))
+				return 0;
+			if (option < 0)
+				return 0;
+			mnt->acregmin =
+			mnt->acregmax =
+			mnt->acdirmin =
+			mnt->acdirmax = option;
+			break;
+		case Opt_namelen:
+			if (match_int(args, &mnt->namlen))
+				return 0;
+			break;
+		case Opt_mountport:
+			if (match_int(args, &option))
+				return 0;
+			if (option < 0 || option > 65535)
+				return 0;
+			mnt->mount_server.port = option;
+			break;
+		case Opt_mountprog:
+			if (match_int(args, &option))
+				return 0;
+			if (option < 0)
+				return 0;
+			mnt->mount_server.program = option;
+			break;
+		case Opt_mountvers:
+			if (match_int(args, &option))
+				return 0;
+			if (option < 0)
+				return 0;
+			mnt->mount_server.version = option;
+			break;
+		case Opt_nfsprog:
+			if (match_int(args, &option))
+				return 0;
+			if (option < 0)
+				return 0;
+			mnt->nfs_server.program = option;
+			break;
+		case Opt_nfsvers:
+			if (match_int(args, &option))
+				return 0;
+			switch (option) {
+			case 2:
+				mnt->flags &= ~NFS_MOUNT_VER3;
+				break;
+			case 3:
+				mnt->flags |= NFS_MOUNT_VER3;
+				break;
+			default:
+				goto out_unrec_vers;
 			}
-			data->root.size = NFS2_FHSIZE;
-			memcpy(data->root.data, data->old_root.data, NFS2_FHSIZE);
-		case 4:
-			if (data->flags & NFS_MOUNT_SECFLAVOUR) {
-				dprintk("%s: mount structure version %d does not support strong security\n",
-						__FUNCTION__,
-						data->version);
-				return -EINVAL;
+			break;
+
+		case Opt_sec:
+			string = match_strdup(args);
+			if (string == NULL)
+				goto out_nomem;
+			token = match_token(string, nfs_secflavor_tokens, args);
+			kfree(string);
+
+			/*
+			 * The flags setting is for v2/v3.  The flavor_len
+			 * setting is for v4.  v2/v3 also need to know the
+			 * difference between NULL and UNIX.
+			 */
+			switch (token) {
+			case Opt_sec_none:
+				mnt->flags &= ~NFS_MOUNT_SECFLAVOUR;
+				mnt->auth_flavor_len = 0;
+				mnt->auth_flavors[0] = RPC_AUTH_NULL;
+				break;
+			case Opt_sec_sys:
+				mnt->flags &= ~NFS_MOUNT_SECFLAVOUR;
+				mnt->auth_flavor_len = 0;
+				mnt->auth_flavors[0] = RPC_AUTH_UNIX;
+				break;
+			case Opt_sec_krb5:
+				mnt->flags |= NFS_MOUNT_SECFLAVOUR;
+				mnt->auth_flavor_len = 1;
+				mnt->auth_flavors[0] = RPC_AUTH_GSS_KRB5;
+				break;
+			case Opt_sec_krb5i:
+				mnt->flags |= NFS_MOUNT_SECFLAVOUR;
+				mnt->auth_flavor_len = 1;
+				mnt->auth_flavors[0] = RPC_AUTH_GSS_KRB5I;
+				break;
+			case Opt_sec_krb5p:
+				mnt->flags |= NFS_MOUNT_SECFLAVOUR;
+				mnt->auth_flavor_len = 1;
+				mnt->auth_flavors[0] = RPC_AUTH_GSS_KRB5P;
+				break;
+			case Opt_sec_lkey:
+				mnt->flags |= NFS_MOUNT_SECFLAVOUR;
+				mnt->auth_flavor_len = 1;
+				mnt->auth_flavors[0] = RPC_AUTH_GSS_LKEY;
+				break;
+			case Opt_sec_lkeyi:
+				mnt->flags |= NFS_MOUNT_SECFLAVOUR;
+				mnt->auth_flavor_len = 1;
+				mnt->auth_flavors[0] = RPC_AUTH_GSS_LKEYI;
+				break;
+			case Opt_sec_lkeyp:
+				mnt->flags |= NFS_MOUNT_SECFLAVOUR;
+				mnt->auth_flavor_len = 1;
+				mnt->auth_flavors[0] = RPC_AUTH_GSS_LKEYP;
+				break;
+			case Opt_sec_spkm:
+				mnt->flags |= NFS_MOUNT_SECFLAVOUR;
+				mnt->auth_flavor_len = 1;
+				mnt->auth_flavors[0] = RPC_AUTH_GSS_SPKM;
+				break;
+			case Opt_sec_spkmi:
+				mnt->flags |= NFS_MOUNT_SECFLAVOUR;
+				mnt->auth_flavor_len = 1;
+				mnt->auth_flavors[0] = RPC_AUTH_GSS_SPKMI;
+				break;
+			case Opt_sec_spkmp:
+				mnt->flags |= NFS_MOUNT_SECFLAVOUR;
+				mnt->auth_flavor_len = 1;
+				mnt->auth_flavors[0] = RPC_AUTH_GSS_SPKMP;
+				break;
+			default:
+				goto out_unrec_sec;
 			}
-		case 5:
-			memset(data->context, 0, sizeof(data->context));
-	}
+			break;
+		case Opt_proto:
+			string = match_strdup(args);
+			if (string == NULL)
+				goto out_nomem;
+			token = match_token(string,
+					    nfs_xprt_protocol_tokens, args);
+			kfree(string);
+
+			switch (token) {
+			case Opt_udp:
+				mnt->flags &= ~NFS_MOUNT_TCP;
+				mnt->nfs_server.protocol = IPPROTO_UDP;
+				mnt->timeo = 7;
+				mnt->retrans = 5;
+				break;
+			case Opt_tcp:
+				mnt->flags |= NFS_MOUNT_TCP;
+				mnt->nfs_server.protocol = IPPROTO_TCP;
+				mnt->timeo = 600;
+				mnt->retrans = 2;
+				break;
+			default:
+				goto out_unrec_xprt;
+			}
+			break;
+		case Opt_mountproto:
+			string = match_strdup(args);
+			if (string == NULL)
+				goto out_nomem;
+			token = match_token(string,
+					    nfs_xprt_protocol_tokens, args);
+			kfree(string);
+
+			switch (token) {
+			case Opt_udp:
+				mnt->mount_server.protocol = IPPROTO_UDP;
+				break;
+			case Opt_tcp:
+				mnt->mount_server.protocol = IPPROTO_TCP;
+				break;
+			default:
+				goto out_unrec_xprt;
+			}
+			break;
+		case Opt_addr:
+			string = match_strdup(args);
+			if (string == NULL)
+				goto out_nomem;
+			mnt->nfs_server.address.sin_family = AF_INET;
+			mnt->nfs_server.address.sin_addr.s_addr =
+							in_aton(string);
+			kfree(string);
+			break;
+		case Opt_clientaddr:
+			string = match_strdup(args);
+			if (string == NULL)
+				goto out_nomem;
+			mnt->client_address = string;
+			break;
+		case Opt_mounthost:
+			string = match_strdup(args);
+			if (string == NULL)
+				goto out_nomem;
+			mnt->mount_server.address.sin_family = AF_INET;
+			mnt->mount_server.address.sin_addr.s_addr =
+							in_aton(string);
+			kfree(string);
+			break;
 
-	/* Set the pseudoflavor */
-	if (!(data->flags & NFS_MOUNT_SECFLAVOUR))
-		data->pseudoflavor = RPC_AUTH_UNIX;
+		case Opt_userspace:
+		case Opt_deprecated:
+			break;
 
-#ifndef CONFIG_NFS_V3
-	/* If NFSv3 is not compiled in, return -EPROTONOSUPPORT */
-	if (data->flags & NFS_MOUNT_VER3) {
-		dprintk("%s: NFSv3 not compiled into kernel\n", __FUNCTION__);
-		return -EPROTONOSUPPORT;
+		default:
+			goto out_unknown;
+		}
 	}
-#endif /* CONFIG_NFS_V3 */
 
-	/* We now require that the mount process passes the remote address */
-	if (data->addr.sin_addr.s_addr == INADDR_ANY) {
-		dprintk("%s: mount program didn't pass remote address!\n",
-			__FUNCTION__);
-		return -EINVAL;
+	return 1;
+
+out_nomem:
+	printk(KERN_INFO "NFS: not enough memory to parse option\n");
+	return 0;
+
+out_unrec_vers:
+	printk(KERN_INFO "NFS: unrecognized NFS version number\n");
+	return 0;
+
+out_unrec_xprt:
+	printk(KERN_INFO "NFS: unrecognized transport protocol\n");
+	return 0;
+
+out_unrec_sec:
+	printk(KERN_INFO "NFS: unrecognized security flavor\n");
+	return 0;
+
+out_unknown:
+	printk(KERN_INFO "NFS: unknown mount option: %s\n", p);
+	return 0;
+}
+
+/*
+ * Use the remote server's MOUNT service to request the NFS file handle
+ * corresponding to the provided path.
+ */
+static int nfs_try_mount(struct nfs_parsed_mount_data *args,
+			 struct nfs_fh *root_fh)
+{
+	struct sockaddr_in sin;
+	int status;
+
+	if (args->mount_server.version == 0) {
+		if (args->flags & NFS_MOUNT_VER3)
+			args->mount_server.version = NFS_MNT3_VERSION;
+		else
+			args->mount_server.version = NFS_MNT_VERSION;
 	}
 
-	/* Prepare the root filehandle */
-	if (data->flags & NFS_MOUNT_VER3)
-		mntfh->size = data->root.size;
+	/*
+	 * Construct the mount server's address.
+	 */
+	if (args->mount_server.address.sin_addr.s_addr != INADDR_ANY)
+		sin = args->mount_server.address;
 	else
-		mntfh->size = NFS2_FHSIZE;
+		sin = args->nfs_server.address;
+	if (args->mount_server.port == 0) {
+		status = rpcb_getport_sync(&sin,
+					   args->mount_server.program,
+					   args->mount_server.version,
+					   args->mount_server.protocol);
+		if (status < 0)
+			goto out_err;
+		sin.sin_port = htons(status);
+	} else
+		sin.sin_port = htons(args->mount_server.port);
+
+	/*
+	 * Now ask the mount server to map our export path
+	 * to a file handle.
+	 */
+	status = nfs_mount((struct sockaddr *) &sin,
+			   sizeof(sin),
+			   args->nfs_server.hostname,
+			   args->nfs_server.export_path,
+			   args->mount_server.version,
+			   args->mount_server.protocol,
+			   root_fh);
+	if (status < 0)
+		goto out_err;
+
+	return status;
 
-	if (mntfh->size > sizeof(mntfh->data)) {
-		dprintk("%s: invalid root filehandle\n", __FUNCTION__);
-		return -EINVAL;
+out_err:
+	dfprintk(MOUNT, "NFS: unable to contact server on host "
+		 NIPQUAD_FMT "\n", NIPQUAD(sin.sin_addr.s_addr));
+	return status;
+}
+
+/*
+ * Validate the NFS2/NFS3 mount data
+ * - fills in the mount root filehandle
+ *
+ * For option strings, user space handles the following behaviors:
+ *
+ * + DNS: mapping server host name to IP address ("addr=" option)
+ *
+ * + failure mode: how to behave if a mount request can't be handled
+ *   immediately ("fg/bg" option)
+ *
+ * + retry: how often to retry a mount request ("retry=" option)
+ *
+ * + breaking back: trying proto=udp after proto=tcp, v2 after v3,
+ *   mountproto=tcp after mountproto=udp, and so on
+ *
+ * XXX: as far as I can tell, changing the NFS program number is not
+ *      supported in the NFS client.
+ */
+static int nfs_validate_mount_data(struct nfs_mount_data **options,
+				   struct nfs_fh *mntfh,
+				   const char *dev_name)
+{
+	struct nfs_mount_data *data = *options;
+
+	if (data == NULL)
+		goto out_no_data;
+
+	switch (data->version) {
+	case 1:
+		data->namlen = 0;
+	case 2:
+		data->bsize = 0;
+	case 3:
+		if (data->flags & NFS_MOUNT_VER3)
+			goto out_no_v3;
+		data->root.size = NFS2_FHSIZE;
+		memcpy(data->root.data, data->old_root.data, NFS2_FHSIZE);
+	case 4:
+		if (data->flags & NFS_MOUNT_SECFLAVOUR)
+			goto out_no_sec;
+	case 5:
+		memset(data->context, 0, sizeof(data->context));
+	case 6:
+		if (data->flags & NFS_MOUNT_VER3)
+			mntfh->size = data->root.size;
+		else
+			mntfh->size = NFS2_FHSIZE;
+
+		if (mntfh->size > sizeof(mntfh->data))
+			goto out_invalid_fh;
+
+		memcpy(mntfh->data, data->root.data, mntfh->size);
+		if (mntfh->size < sizeof(mntfh->data))
+			memset(mntfh->data + mntfh->size, 0,
+			       sizeof(mntfh->data) - mntfh->size);
+		break;
+	default: {
+		unsigned int len;
+		char *c;
+		int status;
+		struct nfs_parsed_mount_data args = {
+			.flags		= (NFS_MOUNT_VER3 | NFS_MOUNT_TCP),
+			.rsize		= NFS_MAX_FILE_IO_SIZE,
+			.wsize		= NFS_MAX_FILE_IO_SIZE,
+			.timeo		= 600,
+			.retrans	= 2,
+			.acregmin	= 3,
+			.acregmax	= 60,
+			.acdirmin	= 30,
+			.acdirmax	= 60,
+			.mount_server.protocol = IPPROTO_UDP,
+			.mount_server.program = NFS_MNT_PROGRAM,
+			.nfs_server.protocol = IPPROTO_TCP,
+			.nfs_server.program = NFS_PROGRAM,
+		};
+
+		if (nfs_parse_mount_options((char *) *options, &args) == 0)
+			return -EINVAL;
+
+		data = kzalloc(sizeof(*data), GFP_KERNEL);
+		if (data == NULL)
+			return -ENOMEM;
+
+		/*
+		 * NB: after this point, caller will free "data"
+		 * if we return an error
+		 */
+		*options = data;
+
+		c = strchr(dev_name, ':');
+		if (c == NULL)
+			return -EINVAL;
+		len = c - dev_name - 1;
+		if (len > sizeof(data->hostname))
+			return -EINVAL;
+		strncpy(data->hostname, dev_name, len);
+		args.nfs_server.hostname = data->hostname;
+
+		c++;
+		if (strlen(c) > NFS_MAXPATHLEN)
+			return -EINVAL;
+		args.nfs_server.export_path = c;
+
+		status = nfs_try_mount(&args, mntfh);
+		if (status)
+			return -EINVAL;
+
+		/*
+		 * Translate to nfs_mount_data, which nfs_fill_super
+		 * can deal with.
+		 */
+		data->version		= 6;
+		data->flags		= args.flags;
+		data->rsize		= args.rsize;
+		data->wsize		= args.wsize;
+		data->timeo		= args.timeo;
+		data->retrans		= args.retrans;
+		data->acregmin		= args.acregmin;
+		data->acregmax		= args.acregmax;
+		data->acdirmin		= args.acdirmin;
+		data->acdirmax		= args.acdirmax;
+		data->addr		= args.nfs_server.address;
+		data->namlen		= args.namlen;
+		data->bsize		= args.bsize;
+		data->pseudoflavor	= args.auth_flavors[0];
+
+		break;
+		}
 	}
 
-	memcpy(mntfh->data, data->root.data, mntfh->size);
-	if (mntfh->size < sizeof(mntfh->data))
-		memset(mntfh->data + mntfh->size, 0,
-		       sizeof(mntfh->data) - mntfh->size);
+	if (!(data->flags & NFS_MOUNT_SECFLAVOUR))
+		data->pseudoflavor = RPC_AUTH_UNIX;
+
+#ifndef CONFIG_NFS_V3
+	if (data->flags & NFS_MOUNT_VER3)
+		goto out_v3_not_compiled;
+#endif /* !CONFIG_NFS_V3 */
+
+	if (!nfs_verify_server_address((struct sockaddr *) &data->addr))
+		goto out_no_address;
 
 	return 0;
+
+out_no_data:
+	dfprintk(MOUNT, "NFS: mount program didn't pass any mount data\n");
+	return -EINVAL;
+
+out_no_v3:
+	dfprintk(MOUNT, "NFS: nfs_mount_data version %d does not support v3\n",
+		 data->version);
+	return -EINVAL;
+
+out_no_sec:
+	dfprintk(MOUNT, "NFS: nfs_mount_data version supports only AUTH_SYS\n");
+	return -EINVAL;
+
+#ifndef CONFIG_NFS_V3
+out_v3_not_compiled:
+	dfprintk(MOUNT, "NFS: NFSv3 is not compiled into kernel\n");
+	return -EPROTONOSUPPORT;
+#endif /* !CONFIG_NFS_V3 */
+
+out_no_address:
+	dfprintk(MOUNT, "NFS: mount program didn't pass remote address\n");
+	return -EINVAL;
+
+out_invalid_fh:
+	dfprintk(MOUNT, "NFS: invalid root filehandle\n");
+	return -EINVAL;
 }
 
 /*
@@ -600,13 +1317,51 @@ static int nfs_compare_super(struct super_block *sb, void *data)
 {
 	struct nfs_server *server = data, *old = NFS_SB(sb);
 
-	if (old->nfs_client != server->nfs_client)
+	if (memcmp(&old->nfs_client->cl_addr,
+				&server->nfs_client->cl_addr,
+				sizeof(old->nfs_client->cl_addr)) != 0)
+		return 0;
+	/* Note: NFS_MOUNT_UNSHARED == NFS4_MOUNT_UNSHARED */
+	if (old->flags & NFS_MOUNT_UNSHARED)
 		return 0;
 	if (memcmp(&old->fsid, &server->fsid, sizeof(old->fsid)) != 0)
 		return 0;
 	return 1;
 }
 
+#define NFS_MS_MASK (MS_RDONLY|MS_NOSUID|MS_NODEV|MS_NOEXEC|MS_SYNCHRONOUS)
+
+static int nfs_compare_mount_options(const struct super_block *s, const struct nfs_server *b, int flags)
+{
+	const struct nfs_server *a = s->s_fs_info;
+	const struct rpc_clnt *clnt_a = a->client;
+	const struct rpc_clnt *clnt_b = b->client;
+
+	if ((s->s_flags & NFS_MS_MASK) != (flags & NFS_MS_MASK))
+		goto Ebusy;
+	if (a->nfs_client != b->nfs_client)
+		goto Ebusy;
+	if (a->flags != b->flags)
+		goto Ebusy;
+	if (a->wsize != b->wsize)
+		goto Ebusy;
+	if (a->rsize != b->rsize)
+		goto Ebusy;
+	if (a->acregmin != b->acregmin)
+		goto Ebusy;
+	if (a->acregmax != b->acregmax)
+		goto Ebusy;
+	if (a->acdirmin != b->acdirmin)
+		goto Ebusy;
+	if (a->acdirmax != b->acdirmax)
+		goto Ebusy;
+	if (clnt_a->cl_auth->au_flavor != clnt_b->cl_auth->au_flavor)
+		goto Ebusy;
+	return 0;
+Ebusy:
+	return -EBUSY;
+}
+
 static int nfs_get_sb(struct file_system_type *fs_type,
 	int flags, const char *dev_name, void *raw_data, struct vfsmount *mnt)
 {
@@ -615,30 +1370,37 @@ static int nfs_get_sb(struct file_system_type *fs_type,
 	struct nfs_fh mntfh;
 	struct nfs_mount_data *data = raw_data;
 	struct dentry *mntroot;
+	int (*compare_super)(struct super_block *, void *) = nfs_compare_super;
 	int error;
 
 	/* Validate the mount data */
-	error = nfs_validate_mount_data(data, &mntfh);
+	error = nfs_validate_mount_data(&data, &mntfh, dev_name);
 	if (error < 0)
-		return error;
+		goto out;
 
 	/* Get a volume representation */
 	server = nfs_create_server(data, &mntfh);
 	if (IS_ERR(server)) {
 		error = PTR_ERR(server);
-		goto out_err_noserver;
+		goto out;
 	}
 
+	if (server->flags & NFS_MOUNT_UNSHARED)
+		compare_super = NULL;
+
 	/* Get a superblock - note that we may end up sharing one that already exists */
-	s = sget(fs_type, nfs_compare_super, nfs_set_super, server);
+	s = sget(fs_type, compare_super, nfs_set_super, server);
 	if (IS_ERR(s)) {
 		error = PTR_ERR(s);
 		goto out_err_nosb;
 	}
 
 	if (s->s_fs_info != server) {
+		error = nfs_compare_mount_options(s, server, flags);
 		nfs_free_server(server);
 		server = NULL;
+		if (error < 0)
+			goto error_splat_super;
 	}
 
 	if (!s->s_root) {
@@ -656,17 +1418,21 @@ static int nfs_get_sb(struct file_system_type *fs_type,
 	s->s_flags |= MS_ACTIVE;
 	mnt->mnt_sb = s;
 	mnt->mnt_root = mntroot;
-	return 0;
+	error = 0;
+
+out:
+	if (data != raw_data)
+		kfree(data);
+	return error;
 
 out_err_nosb:
 	nfs_free_server(server);
-out_err_noserver:
-	return error;
+	goto out;
 
 error_splat_super:
 	up_write(&s->s_umount);
 	deactivate_super(s);
-	return error;
+	goto out;
 }
 
 /*
@@ -691,6 +1457,7 @@ static int nfs_xdev_get_sb(struct file_system_type *fs_type, int flags,
 	struct super_block *s;
 	struct nfs_server *server;
 	struct dentry *mntroot;
+	int (*compare_super)(struct super_block *, void *) = nfs_compare_super;
 	int error;
 
 	dprintk("--> nfs_xdev_get_sb()\n");
@@ -702,16 +1469,22 @@ static int nfs_xdev_get_sb(struct file_system_type *fs_type, int flags,
 		goto out_err_noserver;
 	}
 
+	if (server->flags & NFS_MOUNT_UNSHARED)
+		compare_super = NULL;
+
 	/* Get a superblock - note that we may end up sharing one that already exists */
-	s = sget(&nfs_fs_type, nfs_compare_super, nfs_set_super, server);
+	s = sget(&nfs_fs_type, compare_super, nfs_set_super, server);
 	if (IS_ERR(s)) {
 		error = PTR_ERR(s);
 		goto out_err_nosb;
 	}
 
 	if (s->s_fs_info != server) {
+		error = nfs_compare_mount_options(s, server, flags);
 		nfs_free_server(server);
 		server = NULL;
+		if (error < 0)
+			goto error_splat_super;
 	}
 
 	if (!s->s_root) {
@@ -772,25 +1545,164 @@ static void nfs4_fill_super(struct super_block *sb)
 	nfs_initialise_sb(sb);
 }
 
-static void *nfs_copy_user_string(char *dst, struct nfs_string *src, int maxlen)
+/*
+ * Validate NFSv4 mount options
+ */
+static int nfs4_validate_mount_data(struct nfs4_mount_data **options,
+				    const char *dev_name,
+				    struct sockaddr_in *addr,
+				    rpc_authflavor_t *authflavour,
+				    char **hostname,
+				    char **mntpath,
+				    char **ip_addr)
 {
-	void *p = NULL;
-
-	if (!src->len)
-		return ERR_PTR(-EINVAL);
-	if (src->len < maxlen)
-		maxlen = src->len;
-	if (dst == NULL) {
-		p = dst = kmalloc(maxlen + 1, GFP_KERNEL);
-		if (p == NULL)
-			return ERR_PTR(-ENOMEM);
-	}
-	if (copy_from_user(dst, src->data, maxlen)) {
-		kfree(p);
-		return ERR_PTR(-EFAULT);
+	struct nfs4_mount_data *data = *options;
+	char *c;
+
+	if (data == NULL)
+		goto out_no_data;
+
+	switch (data->version) {
+	case 1:
+		if (data->host_addrlen != sizeof(*addr))
+			goto out_no_address;
+		if (copy_from_user(addr, data->host_addr, sizeof(*addr)))
+			return -EFAULT;
+		if (addr->sin_port == 0)
+			addr->sin_port = htons(NFS_PORT);
+		if (!nfs_verify_server_address((struct sockaddr *) addr))
+			goto out_no_address;
+
+		switch (data->auth_flavourlen) {
+		case 0:
+			*authflavour = RPC_AUTH_UNIX;
+			break;
+		case 1:
+			if (copy_from_user(authflavour, data->auth_flavours,
+					   sizeof(*authflavour)))
+				return -EFAULT;
+			break;
+		default:
+			goto out_inval_auth;
+		}
+
+		c = strndup_user(data->hostname.data, NFS4_MAXNAMLEN);
+		if (IS_ERR(c))
+			return PTR_ERR(c);
+		*hostname = c;
+
+		c = strndup_user(data->mnt_path.data, NFS4_MAXPATHLEN);
+		if (IS_ERR(c))
+			return PTR_ERR(c);
+		*mntpath = c;
+		dfprintk(MOUNT, "NFS: MNTPATH: '%s'\n", *mntpath);
+
+		c = strndup_user(data->client_addr.data, 16);
+		if (IS_ERR(c))
+			return PTR_ERR(c);
+		*ip_addr = c;
+
+		break;
+	default: {
+		unsigned int len;
+		struct nfs_parsed_mount_data args = {
+			.rsize		= NFS_MAX_FILE_IO_SIZE,
+			.wsize		= NFS_MAX_FILE_IO_SIZE,
+			.timeo		= 600,
+			.retrans	= 2,
+			.acregmin	= 3,
+			.acregmax	= 60,
+			.acdirmin	= 30,
+			.acdirmax	= 60,
+			.nfs_server.protocol = IPPROTO_TCP,
+		};
+
+		if (nfs_parse_mount_options((char *) *options, &args) == 0)
+			return -EINVAL;
+
+		if (!nfs_verify_server_address((struct sockaddr *)
+						&args.nfs_server.address))
+			return -EINVAL;
+		*addr = args.nfs_server.address;
+
+		switch (args.auth_flavor_len) {
+		case 0:
+			*authflavour = RPC_AUTH_UNIX;
+			break;
+		case 1:
+			*authflavour = (rpc_authflavor_t) args.auth_flavors[0];
+			break;
+		default:
+			goto out_inval_auth;
+		}
+
+		/*
+		 * Translate to nfs4_mount_data, which nfs4_fill_super
+		 * can deal with.
+		 */
+		data = kzalloc(sizeof(*data), GFP_KERNEL);
+		if (data == NULL)
+			return -ENOMEM;
+		*options = data;
+
+		data->version	= 1;
+		data->flags	= args.flags & NFS4_MOUNT_FLAGMASK;
+		data->rsize	= args.rsize;
+		data->wsize	= args.wsize;
+		data->timeo	= args.timeo;
+		data->retrans	= args.retrans;
+		data->acregmin	= args.acregmin;
+		data->acregmax	= args.acregmax;
+		data->acdirmin	= args.acdirmin;
+		data->acdirmax	= args.acdirmax;
+		data->proto	= args.nfs_server.protocol;
+
+		/*
+		 * Split "dev_name" into "hostname:mntpath".
+		 */
+		c = strchr(dev_name, ':');
+		if (c == NULL)
+			return -EINVAL;
+		/* while calculating len, pretend ':' is '\0' */
+		len = c - dev_name;
+		if (len > NFS4_MAXNAMLEN)
+			return -EINVAL;
+		*hostname = kzalloc(len, GFP_KERNEL);
+		if (*hostname == NULL)
+			return -ENOMEM;
+		strncpy(*hostname, dev_name, len - 1);
+
+		c++;			/* step over the ':' */
+		len = strlen(c);
+		if (len > NFS4_MAXPATHLEN)
+			return -EINVAL;
+		*mntpath = kzalloc(len + 1, GFP_KERNEL);
+		if (*mntpath == NULL)
+			return -ENOMEM;
+		strncpy(*mntpath, c, len);
+
+		dprintk("MNTPATH: %s\n", *mntpath);
+
+		*ip_addr = args.client_address;
+
+		break;
+		}
 	}
-	dst[maxlen] = '\0';
-	return dst;
+
+	return 0;
+
+out_no_data:
+	dfprintk(MOUNT, "NFS4: mount program didn't pass any mount data\n");
+	return -EINVAL;
+
+out_inval_auth:
+	dfprintk(MOUNT, "NFS4: Invalid number of RPC auth flavours %d\n",
+		 data->auth_flavourlen);
+	return -EINVAL;
+
+out_no_address:
+	dfprintk(MOUNT, "NFS4: mount program didn't pass remote address\n");
+	return -EINVAL;
 }
 
 /*
@@ -806,81 +1718,29 @@ static int nfs4_get_sb(struct file_system_type *fs_type,
 	rpc_authflavor_t authflavour;
 	struct nfs_fh mntfh;
 	struct dentry *mntroot;
-	char *mntpath = NULL, *hostname = NULL, ip_addr[16];
-	void *p;
+	char *mntpath = NULL, *hostname = NULL, *ip_addr = NULL;
+	int (*compare_super)(struct super_block *, void *) = nfs_compare_super;
 	int error;
 
-	if (data == NULL) {
-		dprintk("%s: missing data argument\n", __FUNCTION__);
-		return -EINVAL;
-	}
-	if (data->version <= 0 || data->version > NFS4_MOUNT_VERSION) {
-		dprintk("%s: bad mount version\n", __FUNCTION__);
-		return -EINVAL;
-	}
-
-	/* We now require that the mount process passes the remote address */
-	if (data->host_addrlen != sizeof(addr))
-		return -EINVAL;
-
-	if (copy_from_user(&addr, data->host_addr, sizeof(addr)))
-		return -EFAULT;
-
-	if (addr.sin_family != AF_INET ||
-	    addr.sin_addr.s_addr == INADDR_ANY
-	    ) {
-		dprintk("%s: mount program didn't pass remote IP address!\n",
-				__FUNCTION__);
-		return -EINVAL;
-	}
-	/* RFC3530: The default port for NFS is 2049 */
-	if (addr.sin_port == 0)
-		addr.sin_port = htons(NFS_PORT);
-
-	/* Grab the authentication type */
-	authflavour = RPC_AUTH_UNIX;
-	if (data->auth_flavourlen != 0) {
-		if (data->auth_flavourlen != 1) {
-			dprintk("%s: Invalid number of RPC auth flavours %d.\n",
-					__FUNCTION__, data->auth_flavourlen);
-			error = -EINVAL;
-			goto out_err_noserver;
-		}
-
-		if (copy_from_user(&authflavour, data->auth_flavours,
-				   sizeof(authflavour))) {
-			error = -EFAULT;
-			goto out_err_noserver;
-		}
-	}
-
-	p = nfs_copy_user_string(NULL, &data->hostname, 256);
-	if (IS_ERR(p))
-		goto out_err;
-	hostname = p;
-
-	p = nfs_copy_user_string(NULL, &data->mnt_path, 1024);
-	if (IS_ERR(p))
-		goto out_err;
-	mntpath = p;
-
-	dprintk("MNTPATH: %s\n", mntpath);
-
-	p = nfs_copy_user_string(ip_addr, &data->client_addr,
-				 sizeof(ip_addr) - 1);
-	if (IS_ERR(p))
-		goto out_err;
+	/* Validate the mount data */
+	error = nfs4_validate_mount_data(&data, dev_name, &addr, &authflavour,
+					 &hostname, &mntpath, &ip_addr);
+	if (error < 0)
+		goto out;
 
 	/* Get a volume representation */
 	server = nfs4_create_server(data, hostname, &addr, mntpath, ip_addr,
 				    authflavour, &mntfh);
 	if (IS_ERR(server)) {
 		error = PTR_ERR(server);
-		goto out_err_noserver;
+		goto out;
 	}
 
+	if (server->flags & NFS4_MOUNT_UNSHARED)
+		compare_super = NULL;
+
 	/* Get a superblock - note that we may end up sharing one that already exists */
-	s = sget(fs_type, nfs_compare_super, nfs_set_super, server);
+	s = sget(fs_type, compare_super, nfs_set_super, server);
 	if (IS_ERR(s)) {
 		error = PTR_ERR(s);
 		goto out_free;
@@ -906,25 +1766,22 @@ static int nfs4_get_sb(struct file_system_type *fs_type,
 	s->s_flags |= MS_ACTIVE;
 	mnt->mnt_sb = s;
 	mnt->mnt_root = mntroot;
+	error = 0;
+
+out:
+	kfree(ip_addr);
 	kfree(mntpath);
 	kfree(hostname);
-	return 0;
-
-out_err:
-	error = PTR_ERR(p);
-	goto out_err_noserver;
+	return error;
 
 out_free:
 	nfs_free_server(server);
-out_err_noserver:
-	kfree(mntpath);
-	kfree(hostname);
-	return error;
+	goto out;
 
 error_splat_super:
 	up_write(&s->s_umount);
 	deactivate_super(s);
-	goto out_err_noserver;
+	goto out;
 }
 
 static void nfs4_kill_super(struct super_block *sb)
@@ -949,6 +1806,7 @@ static int nfs4_xdev_get_sb(struct file_system_type *fs_type, int flags,
 	struct super_block *s;
 	struct nfs_server *server;
 	struct dentry *mntroot;
+	int (*compare_super)(struct super_block *, void *) = nfs_compare_super;
 	int error;
 
 	dprintk("--> nfs4_xdev_get_sb()\n");
@@ -960,8 +1818,11 @@ static int nfs4_xdev_get_sb(struct file_system_type *fs_type, int flags,
 		goto out_err_noserver;
 	}
 
+	if (server->flags & NFS4_MOUNT_UNSHARED)
+		compare_super = NULL;
+
 	/* Get a superblock - note that we may end up sharing one that already exists */
-	s = sget(&nfs_fs_type, nfs_compare_super, nfs_set_super, server);
+	s = sget(&nfs_fs_type, compare_super, nfs_set_super, server);
 	if (IS_ERR(s)) {
 		error = PTR_ERR(s);
 		goto out_err_nosb;
@@ -1016,6 +1877,7 @@ static int nfs4_referral_get_sb(struct file_system_type *fs_type, int flags,
 	struct nfs_server *server;
 	struct dentry *mntroot;
 	struct nfs_fh mntfh;
+	int (*compare_super)(struct super_block *, void *) = nfs_compare_super;
 	int error;
 
 	dprintk("--> nfs4_referral_get_sb()\n");
@@ -1027,8 +1889,11 @@ static int nfs4_referral_get_sb(struct file_system_type *fs_type, int flags,
 		goto out_err_noserver;
 	}
 
+	if (server->flags & NFS4_MOUNT_UNSHARED)
+		compare_super = NULL;
+
 	/* Get a superblock - note that we may end up sharing one that already exists */
-	s = sget(&nfs_fs_type, nfs_compare_super, nfs_set_super, server);
+	s = sget(&nfs_fs_type, compare_super, nfs_set_super, server);
 	if (IS_ERR(s)) {
 		error = PTR_ERR(s);
 		goto out_err_nosb;

fs/nfs/write.c

@@ -117,19 +117,19 @@ static struct nfs_page *nfs_page_find_request_locked(struct page *page)
 	if (PagePrivate(page)) {
 		req = (struct nfs_page *)page_private(page);
 		if (req != NULL)
-			atomic_inc(&req->wb_count);
+			kref_get(&req->wb_kref);
 	}
 	return req;
 }
 
 static struct nfs_page *nfs_page_find_request(struct page *page)
 {
+	struct inode *inode = page->mapping->host;
 	struct nfs_page *req = NULL;
-	spinlock_t *req_lock = &NFS_I(page->mapping->host)->req_lock;
 
-	spin_lock(req_lock);
+	spin_lock(&inode->i_lock);
 	req = nfs_page_find_request_locked(page);
-	spin_unlock(req_lock);
+	spin_unlock(&inode->i_lock);
 	return req;
 }
 
@@ -191,8 +191,6 @@ static int nfs_writepage_setup(struct nfs_open_context *ctx, struct page *page,
 	}
 	/* Update file length */
 	nfs_grow_file(page, offset, count);
-	/* Set the PG_uptodate flag? */
-	nfs_mark_uptodate(page, offset, count);
 	nfs_unlock_request(req);
 	return 0;
 }
@@ -253,16 +251,16 @@ static void nfs_end_page_writeback(struct page *page)
 static int nfs_page_async_flush(struct nfs_pageio_descriptor *pgio,
 				struct page *page)
 {
+	struct inode *inode = page->mapping->host;
+	struct nfs_inode *nfsi = NFS_I(inode);
 	struct nfs_page *req;
-	struct nfs_inode *nfsi = NFS_I(page->mapping->host);
-	spinlock_t *req_lock = &nfsi->req_lock;
 	int ret;
 
-	spin_lock(req_lock);
+	spin_lock(&inode->i_lock);
 	for(;;) {
 		req = nfs_page_find_request_locked(page);
 		if (req == NULL) {
-			spin_unlock(req_lock);
+			spin_unlock(&inode->i_lock);
 			return 1;
 		}
 		if (nfs_lock_request_dontget(req))
@@ -272,28 +270,28 @@ static int nfs_page_async_flush(struct nfs_pageio_descriptor *pgio,
 		 *	 succeed provided that someone hasn't already marked the
 		 *	 request as dirty (in which case we don't care).
 		 */
-		spin_unlock(req_lock);
+		spin_unlock(&inode->i_lock);
 		ret = nfs_wait_on_request(req);
 		nfs_release_request(req);
 		if (ret != 0)
 			return ret;
-		spin_lock(req_lock);
+		spin_lock(&inode->i_lock);
 	}
 	if (test_bit(PG_NEED_COMMIT, &req->wb_flags)) {
 		/* This request is marked for commit */
-		spin_unlock(req_lock);
+		spin_unlock(&inode->i_lock);
 		nfs_unlock_request(req);
 		nfs_pageio_complete(pgio);
 		return 1;
 	}
 	if (nfs_set_page_writeback(page) != 0) {
-		spin_unlock(req_lock);
+		spin_unlock(&inode->i_lock);
 		BUG();
 	}
 	radix_tree_tag_set(&nfsi->nfs_page_tree, req->wb_index,
-			NFS_PAGE_TAG_WRITEBACK);
+			NFS_PAGE_TAG_LOCKED);
 	ret = test_bit(PG_NEED_FLUSH, &req->wb_flags);
-	spin_unlock(req_lock);
+	spin_unlock(&inode->i_lock);
 	nfs_pageio_add_request(pgio, req);
 	return ret;
 }
@@ -400,7 +398,7 @@ static int nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
 	if (PageDirty(req->wb_page))
 		set_bit(PG_NEED_FLUSH, &req->wb_flags);
 	nfsi->npages++;
-	atomic_inc(&req->wb_count);
+	kref_get(&req->wb_kref);
 	return 0;
 }
 
@@ -409,12 +407,12 @@ static int nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
  */
 static void nfs_inode_remove_request(struct nfs_page *req)
 {
-	struct inode *inode = req->wb_context->dentry->d_inode;
+	struct inode *inode = req->wb_context->path.dentry->d_inode;
 	struct nfs_inode *nfsi = NFS_I(inode);
 
 	BUG_ON (!NFS_WBACK_BUSY(req));
 
-	spin_lock(&nfsi->req_lock);
+	spin_lock(&inode->i_lock);
 	set_page_private(req->wb_page, 0);
 	ClearPagePrivate(req->wb_page);
 	radix_tree_delete(&nfsi->nfs_page_tree, req->wb_index);
@@ -422,11 +420,11 @@ static void nfs_inode_remove_request(struct nfs_page *req)
 		__set_page_dirty_nobuffers(req->wb_page);
 	nfsi->npages--;
 	if (!nfsi->npages) {
-		spin_unlock(&nfsi->req_lock);
+		spin_unlock(&inode->i_lock);
 		nfs_end_data_update(inode);
 		iput(inode);
 	} else
-		spin_unlock(&nfsi->req_lock);
+		spin_unlock(&inode->i_lock);
 	nfs_clear_request(req);
 	nfs_release_request(req);
 }
@@ -457,14 +455,16 @@ nfs_dirty_request(struct nfs_page *req)
 static void
 nfs_mark_request_commit(struct nfs_page *req)
 {
-	struct inode *inode = req->wb_context->dentry->d_inode;
+	struct inode *inode = req->wb_context->path.dentry->d_inode;
 	struct nfs_inode *nfsi = NFS_I(inode);
 
-	spin_lock(&nfsi->req_lock);
-	nfs_list_add_request(req, &nfsi->commit);
+	spin_lock(&inode->i_lock);
 	nfsi->ncommit++;
 	set_bit(PG_NEED_COMMIT, &(req)->wb_flags);
-	spin_unlock(&nfsi->req_lock);
+	radix_tree_tag_set(&nfsi->nfs_page_tree,
+			req->wb_index,
+			NFS_PAGE_TAG_COMMIT);
+	spin_unlock(&inode->i_lock);
 	inc_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
 	__mark_inode_dirty(inode, I_DIRTY_DATASYNC);
 }
@@ -526,18 +526,18 @@ static int nfs_wait_on_requests_locked(struct inode *inode, pgoff_t idx_start, u
 		idx_end = idx_start + npages - 1;
 
 	next = idx_start;
-	while (radix_tree_gang_lookup_tag(&nfsi->nfs_page_tree, (void **)&req, next, 1, NFS_PAGE_TAG_WRITEBACK)) {
+	while (radix_tree_gang_lookup_tag(&nfsi->nfs_page_tree, (void **)&req, next, 1, NFS_PAGE_TAG_LOCKED)) {
 		if (req->wb_index > idx_end)
 			break;
 
 		next = req->wb_index + 1;
 		BUG_ON(!NFS_WBACK_BUSY(req));
 
-		atomic_inc(&req->wb_count);
-		spin_unlock(&nfsi->req_lock);
+		kref_get(&req->wb_kref);
+		spin_unlock(&inode->i_lock);
 		error = nfs_wait_on_request(req);
 		nfs_release_request(req);
-		spin_lock(&nfsi->req_lock);
+		spin_lock(&inode->i_lock);
 		if (error < 0)
 			return error;
 		res++;
@@ -577,10 +577,9 @@ nfs_scan_commit(struct inode *inode, struct list_head *dst, pgoff_t idx_start, u
 	int res = 0;
 
 	if (nfsi->ncommit != 0) {
-		res = nfs_scan_list(nfsi, &nfsi->commit, dst, idx_start, npages);
+		res = nfs_scan_list(nfsi, dst, idx_start, npages,
+				NFS_PAGE_TAG_COMMIT);
 		nfsi->ncommit -= res;
-		if ((nfsi->ncommit == 0) != list_empty(&nfsi->commit))
-			printk(KERN_ERR "NFS: desynchronized value of nfs_i.ncommit.\n");
 	}
 	return res;
 }
@@ -603,7 +602,6 @@ static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx,
 {
 	struct address_space *mapping = page->mapping;
 	struct inode *inode = mapping->host;
-	struct nfs_inode *nfsi = NFS_I(inode);
 	struct nfs_page		*req, *new = NULL;
 	pgoff_t		rqend, end;
 
@@ -613,13 +611,13 @@ static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx,
 		/* Loop over all inode entries and see if we find
 		 * A request for the page we wish to update
 		 */
-		spin_lock(&nfsi->req_lock);
+		spin_lock(&inode->i_lock);
 		req = nfs_page_find_request_locked(page);
 		if (req) {
 			if (!nfs_lock_request_dontget(req)) {
 				int error;
 
-				spin_unlock(&nfsi->req_lock);
+				spin_unlock(&inode->i_lock);
 				error = nfs_wait_on_request(req);
 				nfs_release_request(req);
 				if (error < 0) {
@@ -629,7 +627,7 @@ static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx,
 				}
 				continue;
 			}
-			spin_unlock(&nfsi->req_lock);
+			spin_unlock(&inode->i_lock);
 			if (new)
 				nfs_release_request(new);
 			break;
@@ -640,14 +638,14 @@ static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx,
 			nfs_lock_request_dontget(new);
 			error = nfs_inode_add_request(inode, new);
 			if (error) {
-				spin_unlock(&nfsi->req_lock);
+				spin_unlock(&inode->i_lock);
 				nfs_unlock_request(new);
 				return ERR_PTR(error);
 			}
-			spin_unlock(&nfsi->req_lock);
+			spin_unlock(&inode->i_lock);
 			return new;
 		}
-		spin_unlock(&nfsi->req_lock);
+		spin_unlock(&inode->i_lock);
 
 		new = nfs_create_request(ctx, inode, page, offset, bytes);
 		if (IS_ERR(new))
@@ -751,12 +749,17 @@ int nfs_updatepage(struct file *file, struct page *page,
 static void nfs_writepage_release(struct nfs_page *req)
 {
 
-	if (PageError(req->wb_page) || !nfs_reschedule_unstable_write(req)) {
+	if (PageError(req->wb_page)) {
+		nfs_end_page_writeback(req->wb_page);
+		nfs_inode_remove_request(req);
+	} else if (!nfs_reschedule_unstable_write(req)) {
+		/* Set the PG_uptodate flag */
+		nfs_mark_uptodate(req->wb_page, req->wb_pgbase, req->wb_bytes);
 		nfs_end_page_writeback(req->wb_page);
 		nfs_inode_remove_request(req);
 	} else
 		nfs_end_page_writeback(req->wb_page);
-	nfs_clear_page_writeback(req);
+	nfs_clear_page_tag_locked(req);
 }
 
 static inline int flush_task_priority(int how)
@@ -786,7 +789,7 @@ static void nfs_write_rpcsetup(struct nfs_page *req,
 	 * NB: take care not to mess about with data->commit et al. */
 
 	data->req = req;
-	data->inode = inode = req->wb_context->dentry->d_inode;
+	data->inode = inode = req->wb_context->path.dentry->d_inode;
 	data->cred = req->wb_context->cred;
 
 	data->args.fh     = NFS_FH(inode);
@@ -885,7 +888,7 @@ out_bad:
 	}
 	nfs_redirty_request(req);
 	nfs_end_page_writeback(req->wb_page);
-	nfs_clear_page_writeback(req);
+	nfs_clear_page_tag_locked(req);
 	return -ENOMEM;
 }
 
@@ -928,7 +931,7 @@ static int nfs_flush_one(struct inode *inode, struct list_head *head, unsigned i
 		nfs_list_remove_request(req);
 		nfs_redirty_request(req);
 		nfs_end_page_writeback(req->wb_page);
-		nfs_clear_page_writeback(req);
+		nfs_clear_page_tag_locked(req);
 	}
 	return -ENOMEM;
 }
@@ -954,8 +957,8 @@ static void nfs_writeback_done_partial(struct rpc_task *task, void *calldata)
 	struct page		*page = req->wb_page;
 
 	dprintk("NFS: write (%s/%Ld %d@%Ld)",
-		req->wb_context->dentry->d_inode->i_sb->s_id,
-		(long long)NFS_FILEID(req->wb_context->dentry->d_inode),
+		req->wb_context->path.dentry->d_inode->i_sb->s_id,
+		(long long)NFS_FILEID(req->wb_context->path.dentry->d_inode),
 		req->wb_bytes,
 		(long long)req_offset(req));
 
@@ -970,9 +973,9 @@ static void nfs_writeback_done_partial(struct rpc_task *task, void *calldata)
 	}
 
 	if (nfs_write_need_commit(data)) {
-		spinlock_t *req_lock = &NFS_I(page->mapping->host)->req_lock;
+		struct inode *inode = page->mapping->host;
 
-		spin_lock(req_lock);
+		spin_lock(&inode->i_lock);
 		if (test_bit(PG_NEED_RESCHED, &req->wb_flags)) {
 			/* Do nothing we need to resend the writes */
 		} else if (!test_and_set_bit(PG_NEED_COMMIT, &req->wb_flags)) {
@@ -983,7 +986,7 @@ static void nfs_writeback_done_partial(struct rpc_task *task, void *calldata)
 			clear_bit(PG_NEED_COMMIT, &req->wb_flags);
 			dprintk(" server reboot detected\n");
 		}
-		spin_unlock(req_lock);
+		spin_unlock(&inode->i_lock);
 	} else
 		dprintk(" OK\n");
 
@@ -1020,8 +1023,8 @@ static void nfs_writeback_done_full(struct rpc_task *task, void *calldata)
 		page = req->wb_page;
 
 		dprintk("NFS: write (%s/%Ld %d@%Ld)",
-			req->wb_context->dentry->d_inode->i_sb->s_id,
-			(long long)NFS_FILEID(req->wb_context->dentry->d_inode),
+			req->wb_context->path.dentry->d_inode->i_sb->s_id,
+			(long long)NFS_FILEID(req->wb_context->path.dentry->d_inode),
 			req->wb_bytes,
 			(long long)req_offset(req));
 
@@ -1039,12 +1042,14 @@ static void nfs_writeback_done_full(struct rpc_task *task, void *calldata)
 			dprintk(" marked for commit\n");
 			goto next;
 		}
+		/* Set the PG_uptodate flag? */
+		nfs_mark_uptodate(page, req->wb_pgbase, req->wb_bytes);
 		dprintk(" OK\n");
 remove_request:
 		nfs_end_page_writeback(page);
 		nfs_inode_remove_request(req);
 	next:
-		nfs_clear_page_writeback(req);
+		nfs_clear_page_tag_locked(req);
 	}
 }
 
@@ -1157,7 +1162,7 @@ static void nfs_commit_rpcsetup(struct list_head *head,
 
 	list_splice_init(head, &data->pages);
 	first = nfs_list_entry(data->pages.next);
-	inode = first->wb_context->dentry->d_inode;
+	inode = first->wb_context->path.dentry->d_inode;
 
 	data->inode	  = inode;
 	data->cred	  = first->wb_context->cred;
@@ -1207,7 +1212,7 @@ nfs_commit_list(struct inode *inode, struct list_head *head, int how)
 		nfs_list_remove_request(req);
 		nfs_mark_request_commit(req);
 		dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
-		nfs_clear_page_writeback(req);
+		nfs_clear_page_tag_locked(req);
 	}
 	return -ENOMEM;
 }
@@ -1234,8 +1239,8 @@ static void nfs_commit_done(struct rpc_task *task, void *calldata)
 		dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
 
 		dprintk("NFS: commit (%s/%Ld %d@%Ld)",
-			req->wb_context->dentry->d_inode->i_sb->s_id,
-			(long long)NFS_FILEID(req->wb_context->dentry->d_inode),
+			req->wb_context->path.dentry->d_inode->i_sb->s_id,
+			(long long)NFS_FILEID(req->wb_context->path.dentry->d_inode),
 			req->wb_bytes,
 			(long long)req_offset(req));
 		if (task->tk_status < 0) {
@@ -1249,6 +1254,9 @@ static void nfs_commit_done(struct rpc_task *task, void *calldata)
 		 * returned by the server against all stored verfs. */
 		if (!memcmp(req->wb_verf.verifier, data->verf.verifier, sizeof(data->verf.verifier))) {
 			/* We have a match */
+			/* Set the PG_uptodate flag */
+			nfs_mark_uptodate(req->wb_page, req->wb_pgbase,
+					req->wb_bytes);
 			nfs_inode_remove_request(req);
 			dprintk(" OK\n");
 			goto next;
@@ -1257,7 +1265,7 @@ static void nfs_commit_done(struct rpc_task *task, void *calldata)
 		dprintk(" mismatch\n");
 		nfs_redirty_request(req);
 	next:
-		nfs_clear_page_writeback(req);
+		nfs_clear_page_tag_locked(req);
 	}
 }
 
@@ -1268,13 +1276,12 @@ static const struct rpc_call_ops nfs_commit_ops = {
 
 int nfs_commit_inode(struct inode *inode, int how)
 {
-	struct nfs_inode *nfsi = NFS_I(inode);
 	LIST_HEAD(head);
 	int res;
 
-	spin_lock(&nfsi->req_lock);
+	spin_lock(&inode->i_lock);
 	res = nfs_scan_commit(inode, &head, 0, 0);
-	spin_unlock(&nfsi->req_lock);
+	spin_unlock(&inode->i_lock);
 	if (res) {
 		int error = nfs_commit_list(inode, &head, how);
 		if (error < 0)
@@ -1292,7 +1299,6 @@ static inline int nfs_commit_list(struct inode *inode, struct list_head *head, i
 long nfs_sync_mapping_wait(struct address_space *mapping, struct writeback_control *wbc, int how)
 {
 	struct inode *inode = mapping->host;
-	struct nfs_inode *nfsi = NFS_I(inode);
 	pgoff_t idx_start, idx_end;
 	unsigned int npages = 0;
 	LIST_HEAD(head);
@@ -1314,7 +1320,7 @@ long nfs_sync_mapping_wait(struct address_space *mapping, struct writeback_contr
 		}
 	}
 	how &= ~FLUSH_NOCOMMIT;
-	spin_lock(&nfsi->req_lock);
+	spin_lock(&inode->i_lock);
 	do {
 		ret = nfs_wait_on_requests_locked(inode, idx_start, npages);
 		if (ret != 0)
@@ -1325,18 +1331,19 @@ long nfs_sync_mapping_wait(struct address_space *mapping, struct writeback_contr
 		if (pages == 0)
 			break;
 		if (how & FLUSH_INVALIDATE) {
-			spin_unlock(&nfsi->req_lock);
+			spin_unlock(&inode->i_lock);
 			nfs_cancel_commit_list(&head);
 			ret = pages;
-			spin_lock(&nfsi->req_lock);
+			spin_lock(&inode->i_lock);
 			continue;
 		}
 		pages += nfs_scan_commit(inode, &head, 0, 0);
-		spin_unlock(&nfsi->req_lock);
+		spin_unlock(&inode->i_lock);
 		ret = nfs_commit_list(inode, &head, how);
-		spin_lock(&nfsi->req_lock);
+		spin_lock(&inode->i_lock);
+
 	} while (ret >= 0);
-	spin_unlock(&nfsi->req_lock);
+	spin_unlock(&inode->i_lock);
 	return ret;
 }
 
@@ -1430,7 +1437,6 @@ int nfs_set_page_dirty(struct page *page)
 {
 	struct address_space *mapping = page->mapping;
 	struct inode *inode;
-	spinlock_t *req_lock;
 	struct nfs_page *req;
 	int ret;
 
@@ -1439,18 +1445,17 @@ int nfs_set_page_dirty(struct page *page)
 	inode = mapping->host;
 	if (!inode)
 		goto out_raced;
-	req_lock = &NFS_I(inode)->req_lock;
-	spin_lock(req_lock);
+	spin_lock(&inode->i_lock);
 	req = nfs_page_find_request_locked(page);
 	if (req != NULL) {
 		/* Mark any existing write requests for flushing */
 		ret = !test_and_set_bit(PG_NEED_FLUSH, &req->wb_flags);
-		spin_unlock(req_lock);
+		spin_unlock(&inode->i_lock);
 		nfs_release_request(req);
 		return ret;
 	}
 	ret = __set_page_dirty_nobuffers(page);
-	spin_unlock(req_lock);
+	spin_unlock(&inode->i_lock);
 	return ret;
 out_raced:
 	return !TestSetPageDirty(page);

fs/nfsd/nfs4callback.c

@@ -394,7 +394,6 @@ nfsd4_probe_callback(struct nfs4_client *clp)
 		.rpc_proc       = &nfs4_cb_procedures[NFSPROC4_CLNT_CB_NULL],
 		.rpc_argp       = clp,
 	};
-	char clientname[16];
 	int status;
 
 	if (atomic_read(&cb->cb_set))
@@ -417,11 +416,6 @@ nfsd4_probe_callback(struct nfs4_client *clp)
 	memset(program->stats, 0, sizeof(cb->cb_stat));
 	program->stats->program = program;
 
-	/* Just here to make some printk's more useful: */
-	snprintf(clientname, sizeof(clientname),
-		"%u.%u.%u.%u", NIPQUAD(addr.sin_addr));
-	args.servername = clientname;
-
 	/* Create RPC client */
 	cb->cb_client = rpc_create(&args);
 	if (IS_ERR(cb->cb_client)) {
@@ -429,29 +423,23 @@ nfsd4_probe_callback(struct nfs4_client *clp)
 		goto out_err;
 	}
 
-	/* Kick rpciod, put the call on the wire. */
-	if (rpciod_up() != 0)
-		goto out_clnt;
-
 	/* the task holds a reference to the nfs4_client struct */
 	atomic_inc(&clp->cl_count);
 
 	msg.rpc_cred = nfsd4_lookupcred(clp,0);
 	if (IS_ERR(msg.rpc_cred))
-		goto out_rpciod;
+		goto out_release_clp;
 	status = rpc_call_async(cb->cb_client, &msg, RPC_TASK_ASYNC, &nfs4_cb_null_ops, NULL);
 	put_rpccred(msg.rpc_cred);
 
 	if (status != 0) {
 		dprintk("NFSD: asynchronous NFSPROC4_CB_NULL failed!\n");
-		goto out_rpciod;
+		goto out_release_clp;
 	}
 	return;
 
-out_rpciod:
+out_release_clp:
 	atomic_dec(&clp->cl_count);
-	rpciod_down();
-out_clnt:
 	rpc_shutdown_client(cb->cb_client);
 out_err:
 	cb->cb_client = NULL;

fs/nfsd/nfs4state.c

@@ -378,7 +378,6 @@ shutdown_callback_client(struct nfs4_client *clp)
 	if (clnt) {
 		clp->cl_callback.cb_client = NULL;
 		rpc_shutdown_client(clnt);
-		rpciod_down();
 	}
 }
 

include/linux/lockd/lockd.h

@@ -39,6 +39,7 @@
 struct nlm_host {
 	struct hlist_node	h_hash;		/* doubly linked list */
 	struct sockaddr_in	h_addr;		/* peer address */
+	struct sockaddr_in	h_saddr;	/* our address (optional) */
 	struct rpc_clnt	*	h_rpcclnt;	/* RPC client to talk to peer */
 	char *			h_name;		/* remote hostname */
 	u32			h_version;	/* interface version */

include/linux/nfs4.h

@@ -15,6 +15,7 @@
 
 #include <linux/types.h>
 
+#define NFS4_BITMAP_SIZE	2
 #define NFS4_VERIFIER_SIZE	8
 #define NFS4_STATEID_SIZE	16
 #define NFS4_FHSIZE		128

include/linux/nfs4_mount.h

@@ -65,6 +65,7 @@ struct nfs4_mount_data {
 #define NFS4_MOUNT_NOCTO	0x0010	/* 1 */
 #define NFS4_MOUNT_NOAC		0x0020	/* 1 */
 #define NFS4_MOUNT_STRICTLOCK	0x1000	/* 1 */
-#define NFS4_MOUNT_FLAGMASK	0xFFFF
+#define NFS4_MOUNT_UNSHARED	0x8000	/* 1 */
+#define NFS4_MOUNT_FLAGMASK	0x9033
 
 #endif

include/linux/nfs_fs.h

@@ -30,7 +30,9 @@
 #ifdef __KERNEL__
 
 #include <linux/in.h>
+#include <linux/kref.h>
 #include <linux/mm.h>
+#include <linux/namei.h>
 #include <linux/pagemap.h>
 #include <linux/rbtree.h>
 #include <linux/rwsem.h>
@@ -69,9 +71,8 @@ struct nfs_access_entry {
 
 struct nfs4_state;
 struct nfs_open_context {
-	atomic_t count;
-	struct vfsmount *vfsmnt;
-	struct dentry *dentry;
+	struct kref kref;
+	struct path path;
 	struct rpc_cred *cred;
 	struct nfs4_state *state;
 	fl_owner_t lockowner;
@@ -155,13 +156,9 @@ struct nfs_inode {
 	/*
 	 * This is the list of dirty unwritten pages.
 	 */
-	spinlock_t		req_lock;
-	struct list_head	dirty;
-	struct list_head	commit;
 	struct radix_tree_root	nfs_page_tree;
 
-	unsigned int		ndirty,
-				ncommit,
+	unsigned long		ncommit,
 				npages;
 
 	/* Open contexts for shared mmap writes */
@@ -187,6 +184,7 @@ struct nfs_inode {
 #define NFS_INO_INVALID_ACCESS	0x0008		/* cached access cred invalid */
 #define NFS_INO_INVALID_ACL	0x0010		/* cached acls are invalid */
 #define NFS_INO_REVAL_PAGECACHE	0x0020		/* must revalidate pagecache */
+#define NFS_INO_REVAL_FORCED	0x0040		/* force revalidation ignoring a delegation */
 
 /*
  * Bit offsets in flags field
@@ -496,21 +494,18 @@ static inline void nfs3_forget_cached_acls(struct inode *inode)
 
 /*
  * linux/fs/mount_clnt.c
- * (Used only by nfsroot module)
  */
-extern int  nfsroot_mount(struct sockaddr_in *, char *, struct nfs_fh *,
-		int, int);
+extern int  nfs_mount(struct sockaddr *, size_t, char *, char *,
+		      int, int, struct nfs_fh *);
 
 /*
  * inline functions
  */
 
-static inline loff_t
-nfs_size_to_loff_t(__u64 size)
+static inline loff_t nfs_size_to_loff_t(__u64 size)
 {
-	loff_t maxsz = (((loff_t) ULONG_MAX) << PAGE_CACHE_SHIFT) + PAGE_CACHE_SIZE - 1;
-	if (size > maxsz)
-		return maxsz;
+	if (size > (__u64) OFFSET_MAX - 1)
+		return OFFSET_MAX - 1;
 	return (loff_t) size;
 }
 
@@ -557,6 +552,7 @@ extern void * nfs_root_data(void);
 #define NFSDBG_ROOT		0x0080
 #define NFSDBG_CALLBACK		0x0100
 #define NFSDBG_CLIENT		0x0200
+#define NFSDBG_MOUNT		0x0400
 #define NFSDBG_ALL		0xFFFF
 
 #ifdef __KERNEL__

include/linux/nfs_fs_sb.h

@@ -16,7 +16,6 @@ struct nfs_client {
 #define NFS_CS_INITING		1		/* busy initialising */
 	int			cl_nfsversion;	/* NFS protocol version */
 	unsigned long		cl_res_state;	/* NFS resources state */
-#define NFS_CS_RPCIOD		0		/* - rpciod started */
 #define NFS_CS_CALLBACK		1		/* - callback started */
 #define NFS_CS_IDMAP		2		/* - idmap started */
 #define NFS_CS_RENEWD		3		/* - renewd started */
@@ -35,7 +34,8 @@ struct nfs_client {
 	nfs4_verifier		cl_confirm;
 	unsigned long		cl_state;
 
-	u32			cl_lockowner_id;
+	struct rb_root		cl_openowner_id;
+	struct rb_root		cl_lockowner_id;
 
 	/*
 	 * The following rwsem ensures exclusive access to the server
@@ -44,9 +44,7 @@ struct nfs_client {
 	struct rw_semaphore	cl_sem;
 
 	struct list_head	cl_delegations;
-	struct list_head	cl_state_owners;
-	struct list_head	cl_unused;
-	int			cl_nunused;
+	struct rb_root		cl_state_owners;
 	spinlock_t		cl_lock;
 
 	unsigned long		cl_lease_time;

include/linux/nfs_mount.h

@@ -37,7 +37,7 @@ struct nfs_mount_data {
 	int		acdirmin;		/* 1 */
 	int		acdirmax;		/* 1 */
 	struct sockaddr_in addr;		/* 1 */
-	char		hostname[256];		/* 1 */
+	char		hostname[NFS_MAXNAMLEN + 1];		/* 1 */
 	int		namlen;			/* 2 */
 	unsigned int	bsize;			/* 3 */
 	struct nfs3_fh	root;			/* 4 */
@@ -62,6 +62,7 @@ struct nfs_mount_data {
 #define NFS_MOUNT_STRICTLOCK	0x1000	/* reserved for NFSv4 */
 #define NFS_MOUNT_SECFLAVOUR	0x2000	/* 5 */
 #define NFS_MOUNT_NORDIRPLUS	0x4000	/* 5 */
+#define NFS_MOUNT_UNSHARED	0x8000	/* 5 */
 #define NFS_MOUNT_FLAGMASK	0xFFFF
 
 #endif

include/linux/nfs_page.h

@@ -16,12 +16,13 @@
 #include <linux/sunrpc/auth.h>
 #include <linux/nfs_xdr.h>
 
-#include <asm/atomic.h>
+#include <linux/kref.h>
 
 /*
  * Valid flags for the radix tree
  */
-#define NFS_PAGE_TAG_WRITEBACK	0
+#define NFS_PAGE_TAG_LOCKED	0
+#define NFS_PAGE_TAG_COMMIT	1
 
 /*
  * Valid flags for a dirty buffer
@@ -33,8 +34,7 @@
 
 struct nfs_inode;
 struct nfs_page {
-	struct list_head	wb_list,	/* Defines state of page: */
-				*wb_list_head;	/*      read/write/commit */
+	struct list_head	wb_list;	/* Defines state of page: */
 	struct page		*wb_page;	/* page to read in/write out */
 	struct nfs_open_context	*wb_context;	/* File state context info */
 	atomic_t		wb_complete;	/* i/os we're waiting for */
@@ -42,7 +42,7 @@ struct nfs_page {
 	unsigned int		wb_offset,	/* Offset & ~PAGE_CACHE_MASK */
 				wb_pgbase,	/* Start of page data */
 				wb_bytes;	/* Length of request */
-	atomic_t		wb_count;	/* reference count */
+	struct kref		wb_kref;	/* reference count */
 	unsigned long		wb_flags;
 	struct nfs_writeverf	wb_verf;	/* Commit cookie */
 };
@@ -71,8 +71,8 @@ extern	void nfs_clear_request(struct nfs_page *req);
 extern	void nfs_release_request(struct nfs_page *req);
 
 
-extern	int nfs_scan_list(struct nfs_inode *nfsi, struct list_head *head, struct list_head *dst,
-			  pgoff_t idx_start, unsigned int npages);
+extern	int nfs_scan_list(struct nfs_inode *nfsi, struct list_head *dst,
+			  pgoff_t idx_start, unsigned int npages, int tag);
 extern	void nfs_pageio_init(struct nfs_pageio_descriptor *desc,
 			     struct inode *inode,
 			     int (*doio)(struct inode *, struct list_head *, unsigned int, size_t, int),
@@ -84,12 +84,11 @@ extern	void nfs_pageio_complete(struct nfs_pageio_descriptor *desc);
 extern	void nfs_pageio_cond_complete(struct nfs_pageio_descriptor *, pgoff_t);
 extern  int nfs_wait_on_request(struct nfs_page *);
 extern	void nfs_unlock_request(struct nfs_page *req);
-extern  int nfs_set_page_writeback_locked(struct nfs_page *req);
-extern  void nfs_clear_page_writeback(struct nfs_page *req);
+extern  void nfs_clear_page_tag_locked(struct nfs_page *req);
 
 
 /*
- * Lock the page of an asynchronous request without incrementing the wb_count
+ * Lock the page of an asynchronous request without getting a new reference
  */
 static inline int
 nfs_lock_request_dontget(struct nfs_page *req)
@@ -98,14 +97,14 @@ nfs_lock_request_dontget(struct nfs_page *req)
 }
 
 /*
- * Lock the page of an asynchronous request
+ * Lock the page of an asynchronous request and take a reference
  */
 static inline int
 nfs_lock_request(struct nfs_page *req)
 {
 	if (test_and_set_bit(PG_BUSY, &req->wb_flags))
 		return 0;
-	atomic_inc(&req->wb_count);
+	kref_get(&req->wb_kref);
 	return 1;
 }
 
@@ -118,7 +117,6 @@ static inline void
 nfs_list_add_request(struct nfs_page *req, struct list_head *head)
 {
 	list_add_tail(&req->wb_list, head);
-	req->wb_list_head = head;
 }
 
 
@@ -132,7 +130,6 @@ nfs_list_remove_request(struct nfs_page *req)
 	if (list_empty(&req->wb_list))
 		return;
 	list_del_init(&req->wb_list);
-	req->wb_list_head = NULL;
 }
 
 static inline struct nfs_page *

include/linux/nfs_xdr.h

@@ -119,7 +119,7 @@ struct nfs_openargs {
 	struct nfs_seqid *	seqid;
 	int			open_flags;
 	__u64                   clientid;
-	__u32                   id;
+	__u64                   id;
 	union {
 		struct iattr *  attrs;    /* UNCHECKED, GUARDED */
 		nfs4_verifier   verifier; /* EXCLUSIVE */
@@ -144,6 +144,7 @@ struct nfs_openres {
 	nfs4_stateid		delegation;
 	__u32			do_recall;
 	__u64			maxsize;
+	__u32			attrset[NFS4_BITMAP_SIZE];
 };
 
 /*
@@ -180,7 +181,7 @@ struct nfs_closeres {
  *   */
 struct nfs_lowner {
 	__u64			clientid;
-	u32			id;
+	__u64			id;
 };
 
 struct nfs_lock_args {

include/linux/sunrpc/auth.h

@@ -16,6 +16,7 @@
 #include <linux/sunrpc/xdr.h>
 
 #include <asm/atomic.h>
+#include <linux/rcupdate.h>
 
 /* size of the nodename buffer */
 #define UNX_MAXNODENAME	32
@@ -30,22 +31,28 @@ struct auth_cred {
 /*
  * Client user credentials
  */
+struct rpc_auth;
+struct rpc_credops;
 struct rpc_cred {
 	struct hlist_node	cr_hash;	/* hash chain */
-	struct rpc_credops *	cr_ops;
-	unsigned long		cr_expire;	/* when to gc */
-	atomic_t		cr_count;	/* ref count */
-	unsigned short		cr_flags;	/* various flags */
+	struct list_head	cr_lru;		/* lru garbage collection */
+	struct rcu_head		cr_rcu;
+	struct rpc_auth *	cr_auth;
+	const struct rpc_credops *cr_ops;
 #ifdef RPC_DEBUG
 	unsigned long		cr_magic;	/* 0x0f4aa4f0 */
 #endif
+	unsigned long		cr_expire;	/* when to gc */
+	unsigned long		cr_flags;	/* various flags */
+	atomic_t		cr_count;	/* ref count */
 
 	uid_t			cr_uid;
 
 	/* per-flavor data */
 };
-#define RPCAUTH_CRED_NEW	0x0001
-#define RPCAUTH_CRED_UPTODATE	0x0002
+#define RPCAUTH_CRED_NEW	0
+#define RPCAUTH_CRED_UPTODATE	1
+#define RPCAUTH_CRED_HASHED	2
 
 #define RPCAUTH_CRED_MAGIC	0x0f4aa4f0
 
@@ -56,10 +63,10 @@ struct rpc_cred {
 #define RPC_CREDCACHE_MASK	(RPC_CREDCACHE_NR - 1)
 struct rpc_cred_cache {
 	struct hlist_head	hashtable[RPC_CREDCACHE_NR];
-	unsigned long		nextgc;		/* next garbage collection */
-	unsigned long		expire;		/* cache expiry interval */
+	spinlock_t		lock;
 };
 
+struct rpc_authops;
 struct rpc_auth {
 	unsigned int		au_cslack;	/* call cred size estimate */
 				/* guess at number of u32's auth adds before
@@ -69,7 +76,7 @@ struct rpc_auth {
 	unsigned int		au_verfsize;
 
 	unsigned int		au_flags;	/* various flags */
-	struct rpc_authops *	au_ops;		/* operations */
+	const struct rpc_authops *au_ops;		/* operations */
 	rpc_authflavor_t	au_flavor;	/* pseudoflavor (note may
 						 * differ from the flavor in
 						 * au_ops->au_flavor in gss
@@ -115,17 +122,19 @@ struct rpc_credops {
 						void *, __be32 *, void *);
 };
 
-extern struct rpc_authops	authunix_ops;
-extern struct rpc_authops	authnull_ops;
-#ifdef CONFIG_SUNRPC_SECURE
-extern struct rpc_authops	authdes_ops;
-#endif
+extern const struct rpc_authops	authunix_ops;
+extern const struct rpc_authops	authnull_ops;
+
+void __init		rpc_init_authunix(void);
+void __init		rpcauth_init_module(void);
+void __exit		rpcauth_remove_module(void);
 
-int			rpcauth_register(struct rpc_authops *);
-int			rpcauth_unregister(struct rpc_authops *);
+int			rpcauth_register(const struct rpc_authops *);
+int			rpcauth_unregister(const struct rpc_authops *);
 struct rpc_auth *	rpcauth_create(rpc_authflavor_t, struct rpc_clnt *);
-void			rpcauth_destroy(struct rpc_auth *);
+void			rpcauth_release(struct rpc_auth *);
 struct rpc_cred *	rpcauth_lookup_credcache(struct rpc_auth *, struct auth_cred *, int);
+void			rpcauth_init_cred(struct rpc_cred *, const struct auth_cred *, struct rpc_auth *, const struct rpc_credops *);
 struct rpc_cred *	rpcauth_lookupcred(struct rpc_auth *, int);
 struct rpc_cred *	rpcauth_bindcred(struct rpc_task *);
 void			rpcauth_holdcred(struct rpc_task *);
@@ -138,8 +147,9 @@ int			rpcauth_unwrap_resp(struct rpc_task *task, kxdrproc_t decode, void *rqstp,
 int			rpcauth_refreshcred(struct rpc_task *);
 void			rpcauth_invalcred(struct rpc_task *);
 int			rpcauth_uptodatecred(struct rpc_task *);
-int			rpcauth_init_credcache(struct rpc_auth *, unsigned long);
-void			rpcauth_free_credcache(struct rpc_auth *);
+int			rpcauth_init_credcache(struct rpc_auth *);
+void			rpcauth_destroy_credcache(struct rpc_auth *);
+void			rpcauth_clear_credcache(struct rpc_cred_cache *);
 
 static inline
 struct rpc_cred *	get_rpccred(struct rpc_cred *cred)

include/linux/sunrpc/auth_gss.h

@@ -75,6 +75,7 @@ struct gss_cl_ctx {
 	struct xdr_netobj	gc_wire_ctx;
 	u32			gc_win;
 	unsigned long		gc_expiry;
+	struct rcu_head		gc_rcu;
 };
 
 struct gss_upcall_msg;
@@ -85,11 +86,6 @@ struct gss_cred {
 	struct gss_upcall_msg	*gc_upcall;
 };
 
-#define gc_uid			gc_base.cr_uid
-#define gc_count		gc_base.cr_count
-#define gc_flags		gc_base.cr_flags
-#define gc_expire		gc_base.cr_expire
-
 #endif /* __KERNEL__ */
 #endif /* _LINUX_SUNRPC_AUTH_GSS_H */
 

include/linux/sunrpc/clnt.h

@@ -24,8 +24,10 @@ struct rpc_inode;
  * The high-level client handle
  */
 struct rpc_clnt {
-	atomic_t		cl_count;	/* Number of clones */
-	atomic_t		cl_users;	/* number of references */
+	struct kref		cl_kref;	/* Number of references */
+	struct list_head	cl_clients;	/* Global list of clients */
+	struct list_head	cl_tasks;	/* List of tasks */
+	spinlock_t		cl_lock;	/* spinlock */
 	struct rpc_xprt *	cl_xprt;	/* transport */
 	struct rpc_procinfo *	cl_procinfo;	/* procedure info */
 	u32			cl_prog,	/* RPC program number */
@@ -41,9 +43,7 @@ struct rpc_clnt {
 	unsigned int		cl_softrtry : 1,/* soft timeouts */
 				cl_intr     : 1,/* interruptible */
 				cl_discrtry : 1,/* disconnect before retry */
-				cl_autobind : 1,/* use getport() */
-				cl_oneshot  : 1,/* dispose after use */
-				cl_dead     : 1;/* abandoned */
+				cl_autobind : 1;/* use getport() */
 
 	struct rpc_rtt *	cl_rtt;		/* RTO estimator data */
 
@@ -98,6 +98,7 @@ struct rpc_create_args {
 	int			protocol;
 	struct sockaddr		*address;
 	size_t			addrsize;
+	struct sockaddr		*saddress;
 	struct rpc_timeout	*timeout;
 	char			*servername;
 	struct rpc_program	*program;
@@ -110,20 +111,20 @@ struct rpc_create_args {
 #define RPC_CLNT_CREATE_HARDRTRY	(1UL << 0)
 #define RPC_CLNT_CREATE_INTR		(1UL << 1)
 #define RPC_CLNT_CREATE_AUTOBIND	(1UL << 2)
-#define RPC_CLNT_CREATE_ONESHOT		(1UL << 3)
-#define RPC_CLNT_CREATE_NONPRIVPORT	(1UL << 4)
-#define RPC_CLNT_CREATE_NOPING		(1UL << 5)
-#define RPC_CLNT_CREATE_DISCRTRY	(1UL << 6)
+#define RPC_CLNT_CREATE_NONPRIVPORT	(1UL << 3)
+#define RPC_CLNT_CREATE_NOPING		(1UL << 4)
+#define RPC_CLNT_CREATE_DISCRTRY	(1UL << 5)
 
 struct rpc_clnt *rpc_create(struct rpc_create_args *args);
 struct rpc_clnt	*rpc_bind_new_program(struct rpc_clnt *,
 				struct rpc_program *, int);
 struct rpc_clnt *rpc_clone_client(struct rpc_clnt *);
-int		rpc_shutdown_client(struct rpc_clnt *);
-int		rpc_destroy_client(struct rpc_clnt *);
+void		rpc_shutdown_client(struct rpc_clnt *);
 void		rpc_release_client(struct rpc_clnt *);
+
 int		rpcb_register(u32, u32, int, unsigned short, int *);
-void		rpcb_getport(struct rpc_task *);
+int		rpcb_getport_sync(struct sockaddr_in *, __u32, __u32, int);
+void		rpcb_getport_async(struct rpc_task *);
 
 void		rpc_call_setup(struct rpc_task *, struct rpc_message *, int);
 
@@ -132,20 +133,16 @@ int		rpc_call_async(struct rpc_clnt *clnt, struct rpc_message *msg,
 			       void *calldata);
 int		rpc_call_sync(struct rpc_clnt *clnt, struct rpc_message *msg,
 			      int flags);
+struct rpc_task *rpc_call_null(struct rpc_clnt *clnt, struct rpc_cred *cred,
+			       int flags);
 void		rpc_restart_call(struct rpc_task *);
 void		rpc_clnt_sigmask(struct rpc_clnt *clnt, sigset_t *oldset);
 void		rpc_clnt_sigunmask(struct rpc_clnt *clnt, sigset_t *oldset);
 void		rpc_setbufsize(struct rpc_clnt *, unsigned int, unsigned int);
 size_t		rpc_max_payload(struct rpc_clnt *);
 void		rpc_force_rebind(struct rpc_clnt *);
-int		rpc_ping(struct rpc_clnt *clnt, int flags);
 size_t		rpc_peeraddr(struct rpc_clnt *, struct sockaddr *, size_t);
 char *		rpc_peeraddr2str(struct rpc_clnt *, enum rpc_display_format_t);
 
-/*
- * Helper function for NFSroot support
- */
-int		rpcb_getport_external(struct sockaddr_in *, __u32, __u32, int);
-
 #endif /* __KERNEL__ */
 #endif /* _LINUX_SUNRPC_CLNT_H */

include/linux/sunrpc/gss_api.h

@@ -77,7 +77,7 @@ struct gss_api_mech {
 	struct module		*gm_owner;
 	struct xdr_netobj	gm_oid;
 	char			*gm_name;
-	struct gss_api_ops	*gm_ops;
+	const struct gss_api_ops *gm_ops;
 	/* pseudoflavors supported by this mechanism: */
 	int			gm_pf_num;
 	struct pf_desc *	gm_pfs;

include/linux/sunrpc/rpc_pipe_fs.h

@@ -23,9 +23,11 @@ struct rpc_inode {
 	void *private;
 	struct list_head pipe;
 	struct list_head in_upcall;
+	struct list_head in_downcall;
 	int pipelen;
 	int nreaders;
 	int nwriters;
+	int nkern_readwriters;
 	wait_queue_head_t waitq;
 #define RPC_PIPE_WAIT_FOR_OPEN	1
 	int flags;

include/linux/sunrpc/sched.h

@@ -98,7 +98,6 @@ struct rpc_task {
 	unsigned short		tk_pid;		/* debugging aid */
 #endif
 };
-#define tk_auth			tk_client->cl_auth
 #define tk_xprt			tk_client->cl_xprt
 
 /* support walking a list of tasks on a wait queue */
@@ -110,11 +109,6 @@ struct rpc_task {
 	if (!list_empty(head) &&  \
 	    ((task=list_entry((head)->next, struct rpc_task, u.tk_wait.list)),1))
 
-/* .. and walking list of all tasks */
-#define	alltask_for_each(task, pos, head) \
-	list_for_each(pos, head) \
-		if ((task=list_entry(pos, struct rpc_task, tk_task)),1)
-
 typedef void			(*rpc_action)(struct rpc_task *);
 
 struct rpc_call_ops {

include/linux/sunrpc/svcsock.h

@@ -59,6 +59,7 @@ struct svc_sock {
 	/* cache of various info for TCP sockets */
 	void			*sk_info_authunix;
 
+	struct sockaddr_storage	sk_local;	/* local address */
 	struct sockaddr_storage	sk_remote;	/* remote peer's address */
 	int			sk_remotelen;	/* length of address */
 };

include/linux/sunrpc/xprt.h

@@ -17,6 +17,8 @@
 #include <linux/sunrpc/xdr.h>
 #include <linux/sunrpc/msg_prot.h>
 
+#ifdef __KERNEL__
+
 extern unsigned int xprt_udp_slot_table_entries;
 extern unsigned int xprt_tcp_slot_table_entries;
 
@@ -194,7 +196,13 @@ struct rpc_xprt {
 	char *			address_strings[RPC_DISPLAY_MAX];
 };
 
-#ifdef __KERNEL__
+struct rpc_xprtsock_create {
+	int			proto;		/* IPPROTO_UDP or IPPROTO_TCP */
+	struct sockaddr *	srcaddr;	/* optional local address */
+	struct sockaddr *	dstaddr;	/* remote peer address */
+	size_t			addrlen;
+	struct rpc_timeout *	timeout;	/* optional timeout parameters */
+};
 
 /*
  * Transport operations used by ULPs
@@ -204,7 +212,7 @@ void			xprt_set_timeout(struct rpc_timeout *to, unsigned int retr, unsigned long
 /*
  * Generic internal transport functions
  */
-struct rpc_xprt *	xprt_create_transport(int proto, struct sockaddr *addr, size_t size, struct rpc_timeout *toparms);
+struct rpc_xprt *	xprt_create_transport(struct rpc_xprtsock_create *args);
 void			xprt_connect(struct rpc_task *task);
 void			xprt_reserve(struct rpc_task *task);
 int			xprt_reserve_xprt(struct rpc_task *task);
@@ -242,8 +250,8 @@ void			xprt_disconnect(struct rpc_xprt *xprt);
 /*
  * Socket transport setup operations
  */
-struct rpc_xprt *	xs_setup_udp(struct sockaddr *addr, size_t addrlen, struct rpc_timeout *to);
-struct rpc_xprt *	xs_setup_tcp(struct sockaddr *addr, size_t addrlen, struct rpc_timeout *to);
+struct rpc_xprt *	xs_setup_udp(struct rpc_xprtsock_create *args);
+struct rpc_xprt *	xs_setup_tcp(struct rpc_xprtsock_create *args);
 int			init_socket_xprt(void);
 void			cleanup_socket_xprt(void);
 

net/sunrpc/auth.c

@@ -13,17 +13,22 @@
 #include <linux/errno.h>
 #include <linux/sunrpc/clnt.h>
 #include <linux/spinlock.h>
+#include <linux/smp_lock.h>
 
 #ifdef RPC_DEBUG
 # define RPCDBG_FACILITY	RPCDBG_AUTH
 #endif
 
-static struct rpc_authops *	auth_flavors[RPC_AUTH_MAXFLAVOR] = {
+static DEFINE_SPINLOCK(rpc_authflavor_lock);
+static const struct rpc_authops *auth_flavors[RPC_AUTH_MAXFLAVOR] = {
 	&authnull_ops,		/* AUTH_NULL */
 	&authunix_ops,		/* AUTH_UNIX */
 	NULL,			/* others can be loadable modules */
 };
 
+static LIST_HEAD(cred_unused);
+static unsigned long number_cred_unused;
+
 static u32
 pseudoflavor_to_flavor(u32 flavor) {
 	if (flavor >= RPC_AUTH_MAXFLAVOR)
@@ -32,55 +37,67 @@ pseudoflavor_to_flavor(u32 flavor) {
 }
 
 int
-rpcauth_register(struct rpc_authops *ops)
+rpcauth_register(const struct rpc_authops *ops)
 {
 	rpc_authflavor_t flavor;
+	int ret = -EPERM;
 
 	if ((flavor = ops->au_flavor) >= RPC_AUTH_MAXFLAVOR)
 		return -EINVAL;
-	if (auth_flavors[flavor] != NULL)
-		return -EPERM;		/* what else? */
-	auth_flavors[flavor] = ops;
-	return 0;
+	spin_lock(&rpc_authflavor_lock);
+	if (auth_flavors[flavor] == NULL) {
+		auth_flavors[flavor] = ops;
+		ret = 0;
+	}
+	spin_unlock(&rpc_authflavor_lock);
+	return ret;
 }
 
 int
-rpcauth_unregister(struct rpc_authops *ops)
+rpcauth_unregister(const struct rpc_authops *ops)
 {
 	rpc_authflavor_t flavor;
+	int ret = -EPERM;
 
 	if ((flavor = ops->au_flavor) >= RPC_AUTH_MAXFLAVOR)
 		return -EINVAL;
-	if (auth_flavors[flavor] != ops)
-		return -EPERM;		/* what else? */
-	auth_flavors[flavor] = NULL;
-	return 0;
+	spin_lock(&rpc_authflavor_lock);
+	if (auth_flavors[flavor] == ops) {
+		auth_flavors[flavor] = NULL;
+		ret = 0;
+	}
+	spin_unlock(&rpc_authflavor_lock);
+	return ret;
 }
 
 struct rpc_auth *
 rpcauth_create(rpc_authflavor_t pseudoflavor, struct rpc_clnt *clnt)
 {
 	struct rpc_auth		*auth;
-	struct rpc_authops	*ops;
+	const struct rpc_authops *ops;
 	u32			flavor = pseudoflavor_to_flavor(pseudoflavor);
 
 	auth = ERR_PTR(-EINVAL);
 	if (flavor >= RPC_AUTH_MAXFLAVOR)
 		goto out;
 
-	/* FIXME - auth_flavors[] really needs an rw lock,
-	 * and module refcounting. */
 #ifdef CONFIG_KMOD
 	if ((ops = auth_flavors[flavor]) == NULL)
 		request_module("rpc-auth-%u", flavor);
 #endif
-	if ((ops = auth_flavors[flavor]) == NULL)
+	spin_lock(&rpc_authflavor_lock);
+	ops = auth_flavors[flavor];
+	if (ops == NULL || !try_module_get(ops->owner)) {
+		spin_unlock(&rpc_authflavor_lock);
 		goto out;
+	}
+	spin_unlock(&rpc_authflavor_lock);
 	auth = ops->create(clnt, pseudoflavor);
+	module_put(ops->owner);
 	if (IS_ERR(auth))
 		return auth;
 	if (clnt->cl_auth)
-		rpcauth_destroy(clnt->cl_auth);
+		rpcauth_release(clnt->cl_auth);
 	clnt->cl_auth = auth;
 
 out:
@@ -88,7 +105,7 @@ out:
 }
 
 void
-rpcauth_destroy(struct rpc_auth *auth)
+rpcauth_release(struct rpc_auth *auth)
 {
 	if (!atomic_dec_and_test(&auth->au_count))
 		return;
@@ -97,11 +114,31 @@ rpcauth_destroy(struct rpc_auth *auth)
 
 static DEFINE_SPINLOCK(rpc_credcache_lock);
 
+static void
+rpcauth_unhash_cred_locked(struct rpc_cred *cred)
+{
+	hlist_del_rcu(&cred->cr_hash);
+	smp_mb__before_clear_bit();
+	clear_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags);
+}
+
+static void
+rpcauth_unhash_cred(struct rpc_cred *cred)
+{
+	spinlock_t *cache_lock;
+
+	cache_lock = &cred->cr_auth->au_credcache->lock;
+	spin_lock(cache_lock);
+	if (atomic_read(&cred->cr_count) == 0)
+		rpcauth_unhash_cred_locked(cred);
+	spin_unlock(cache_lock);
+}
+
 /*
  * Initialize RPC credential cache
  */
 int
-rpcauth_init_credcache(struct rpc_auth *auth, unsigned long expire)
+rpcauth_init_credcache(struct rpc_auth *auth)
 {
 	struct rpc_cred_cache *new;
 	int i;
@@ -111,8 +148,7 @@ rpcauth_init_credcache(struct rpc_auth *auth, unsigned long expire)
 		return -ENOMEM;
 	for (i = 0; i < RPC_CREDCACHE_NR; i++)
 		INIT_HLIST_HEAD(&new->hashtable[i]);
-	new->expire = expire;
-	new->nextgc = jiffies + (expire >> 1);
+	spin_lock_init(&new->lock);
 	auth->au_credcache = new;
 	return 0;
 }
@@ -121,13 +157,13 @@ rpcauth_init_credcache(struct rpc_auth *auth, unsigned long expire)
  * Destroy a list of credentials
  */
 static inline
-void rpcauth_destroy_credlist(struct hlist_head *head)
+void rpcauth_destroy_credlist(struct list_head *head)
 {
 	struct rpc_cred *cred;
 
-	while (!hlist_empty(head)) {
-		cred = hlist_entry(head->first, struct rpc_cred, cr_hash);
-		hlist_del_init(&cred->cr_hash);
+	while (!list_empty(head)) {
+		cred = list_entry(head->next, struct rpc_cred, cr_lru);
+		list_del_init(&cred->cr_lru);
 		put_rpccred(cred);
 	}
 }
@@ -137,58 +173,95 @@ void rpcauth_destroy_credlist(struct hlist_head *head)
  * that are not referenced.
  */
 void
-rpcauth_free_credcache(struct rpc_auth *auth)
+rpcauth_clear_credcache(struct rpc_cred_cache *cache)
 {
-	struct rpc_cred_cache *cache = auth->au_credcache;
-	HLIST_HEAD(free);
-	struct hlist_node *pos, *next;
+	LIST_HEAD(free);
+	struct hlist_head *head;
 	struct rpc_cred	*cred;
 	int		i;
 
 	spin_lock(&rpc_credcache_lock);
+	spin_lock(&cache->lock);
 	for (i = 0; i < RPC_CREDCACHE_NR; i++) {
-		hlist_for_each_safe(pos, next, &cache->hashtable[i]) {
-			cred = hlist_entry(pos, struct rpc_cred, cr_hash);
-			__hlist_del(&cred->cr_hash);
-			hlist_add_head(&cred->cr_hash, &free);
+		head = &cache->hashtable[i];
+		while (!hlist_empty(head)) {
+			cred = hlist_entry(head->first, struct rpc_cred, cr_hash);
+			get_rpccred(cred);
+			if (!list_empty(&cred->cr_lru)) {
+				list_del(&cred->cr_lru);
+				number_cred_unused--;
+			}
+			list_add_tail(&cred->cr_lru, &free);
+			rpcauth_unhash_cred_locked(cred);
 		}
 	}
+	spin_unlock(&cache->lock);
 	spin_unlock(&rpc_credcache_lock);
 	rpcauth_destroy_credlist(&free);
 }
 
-static void
-rpcauth_prune_expired(struct rpc_auth *auth, struct rpc_cred *cred, struct hlist_head *free)
+/*
+ * Destroy the RPC credential cache
+ */
+void
+rpcauth_destroy_credcache(struct rpc_auth *auth)
 {
-	if (atomic_read(&cred->cr_count) != 1)
-	       return;
-	if (time_after(jiffies, cred->cr_expire + auth->au_credcache->expire))
-		cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE;
-	if (!(cred->cr_flags & RPCAUTH_CRED_UPTODATE)) {
-		__hlist_del(&cred->cr_hash);
-		hlist_add_head(&cred->cr_hash, free);
+	struct rpc_cred_cache *cache = auth->au_credcache;
+
+	if (cache) {
+		auth->au_credcache = NULL;
+		rpcauth_clear_credcache(cache);
+		kfree(cache);
 	}
 }
 
 /*
  * Remove stale credentials. Avoid sleeping inside the loop.
  */
-static void
-rpcauth_gc_credcache(struct rpc_auth *auth, struct hlist_head *free)
+static int
+rpcauth_prune_expired(struct list_head *free, int nr_to_scan)
 {
-	struct rpc_cred_cache *cache = auth->au_credcache;
-	struct hlist_node *pos, *next;
-	struct rpc_cred	*cred;
-	int		i;
+	spinlock_t *cache_lock;
+	struct rpc_cred *cred;
 
-	dprintk("RPC:       gc'ing RPC credentials for auth %p\n", auth);
-	for (i = 0; i < RPC_CREDCACHE_NR; i++) {
-		hlist_for_each_safe(pos, next, &cache->hashtable[i]) {
-			cred = hlist_entry(pos, struct rpc_cred, cr_hash);
-			rpcauth_prune_expired(auth, cred, free);
+	while (!list_empty(&cred_unused)) {
+		cred = list_entry(cred_unused.next, struct rpc_cred, cr_lru);
+		list_del_init(&cred->cr_lru);
+		number_cred_unused--;
+		if (atomic_read(&cred->cr_count) != 0)
+			continue;
+		cache_lock = &cred->cr_auth->au_credcache->lock;
+		spin_lock(cache_lock);
+		if (atomic_read(&cred->cr_count) == 0) {
+			get_rpccred(cred);
+			list_add_tail(&cred->cr_lru, free);
+			rpcauth_unhash_cred_locked(cred);
+			nr_to_scan--;
 		}
+		spin_unlock(cache_lock);
+		if (nr_to_scan == 0)
+			break;
 	}
-	cache->nextgc = jiffies + cache->expire;
+	return nr_to_scan;
+}
+
+/*
+ * Run memory cache shrinker.
+ */
+static int
+rpcauth_cache_shrinker(int nr_to_scan, gfp_t gfp_mask)
+{
+	LIST_HEAD(free);
+	int res;
+
+	if (list_empty(&cred_unused))
+		return 0;
+	spin_lock(&rpc_credcache_lock);
+	nr_to_scan = rpcauth_prune_expired(&free, nr_to_scan);
+	res = (number_cred_unused / 100) * sysctl_vfs_cache_pressure;
+	spin_unlock(&rpc_credcache_lock);
+	rpcauth_destroy_credlist(&free);
+	return res;
 }
 
 /*
@@ -198,53 +271,56 @@ struct rpc_cred *
 rpcauth_lookup_credcache(struct rpc_auth *auth, struct auth_cred * acred,
 		int flags)
 {
+	LIST_HEAD(free);
 	struct rpc_cred_cache *cache = auth->au_credcache;
-	HLIST_HEAD(free);
-	struct hlist_node *pos, *next;
-	struct rpc_cred	*new = NULL,
-			*cred = NULL;
+	struct hlist_node *pos;
+	struct rpc_cred	*cred = NULL,
+			*entry, *new;
 	int		nr = 0;
 
 	if (!(flags & RPCAUTH_LOOKUP_ROOTCREDS))
 		nr = acred->uid & RPC_CREDCACHE_MASK;
-retry:
-	spin_lock(&rpc_credcache_lock);
-	if (time_before(cache->nextgc, jiffies))
-		rpcauth_gc_credcache(auth, &free);
-	hlist_for_each_safe(pos, next, &cache->hashtable[nr]) {
-		struct rpc_cred *entry;
-		entry = hlist_entry(pos, struct rpc_cred, cr_hash);
-		if (entry->cr_ops->crmatch(acred, entry, flags)) {
-			hlist_del(&entry->cr_hash);
-			cred = entry;
-			break;
+
+	rcu_read_lock();
+	hlist_for_each_entry_rcu(entry, pos, &cache->hashtable[nr], cr_hash) {
+		if (!entry->cr_ops->crmatch(acred, entry, flags))
+			continue;
+		spin_lock(&cache->lock);
+		if (test_bit(RPCAUTH_CRED_HASHED, &entry->cr_flags) == 0) {
+			spin_unlock(&cache->lock);
+			continue;
 		}
-		rpcauth_prune_expired(auth, entry, &free);
+		cred = get_rpccred(entry);
+		spin_unlock(&cache->lock);
+		break;
 	}
-	if (new) {
-		if (cred)
-			hlist_add_head(&new->cr_hash, &free);
-		else
-			cred = new;
-	}
-	if (cred) {
-		hlist_add_head(&cred->cr_hash, &cache->hashtable[nr]);
-		get_rpccred(cred);
-	}
-	spin_unlock(&rpc_credcache_lock);
+	rcu_read_unlock();
 
-	rpcauth_destroy_credlist(&free);
+	if (cred != NULL)
+		goto found;
 
-	if (!cred) {
-		new = auth->au_ops->crcreate(auth, acred, flags);
-		if (!IS_ERR(new)) {
-#ifdef RPC_DEBUG
-			new->cr_magic = RPCAUTH_CRED_MAGIC;
-#endif
-			goto retry;
-		} else
-			cred = new;
-	} else if ((cred->cr_flags & RPCAUTH_CRED_NEW)
+	new = auth->au_ops->crcreate(auth, acred, flags);
+	if (IS_ERR(new)) {
+		cred = new;
+		goto out;
+	}
+
+	spin_lock(&cache->lock);
+	hlist_for_each_entry(entry, pos, &cache->hashtable[nr], cr_hash) {
+		if (!entry->cr_ops->crmatch(acred, entry, flags))
+			continue;
+		cred = get_rpccred(entry);
+		break;
+	}
+	if (cred == NULL) {
+		cred = new;
+		set_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags);
+		hlist_add_head_rcu(&cred->cr_hash, &cache->hashtable[nr]);
+	} else
+		list_add_tail(&new->cr_lru, &free);
+	spin_unlock(&cache->lock);
+found:
+	if (test_bit(RPCAUTH_CRED_NEW, &cred->cr_flags)
 			&& cred->cr_ops->cr_init != NULL
 			&& !(flags & RPCAUTH_LOOKUP_NEW)) {
 		int res = cred->cr_ops->cr_init(auth, cred);
@@ -253,8 +329,9 @@ retry:
 			cred = ERR_PTR(res);
 		}
 	}
-
-	return (struct rpc_cred *) cred;
+	rpcauth_destroy_credlist(&free);
+out:
+	return cred;
 }
 
 struct rpc_cred *
@@ -275,10 +352,27 @@ rpcauth_lookupcred(struct rpc_auth *auth, int flags)
 	return ret;
 }
 
+void
+rpcauth_init_cred(struct rpc_cred *cred, const struct auth_cred *acred,
+		  struct rpc_auth *auth, const struct rpc_credops *ops)
+{
+	INIT_HLIST_NODE(&cred->cr_hash);
+	INIT_LIST_HEAD(&cred->cr_lru);
+	atomic_set(&cred->cr_count, 1);
+	cred->cr_auth = auth;
+	cred->cr_ops = ops;
+	cred->cr_expire = jiffies;
+#ifdef RPC_DEBUG
+	cred->cr_magic = RPCAUTH_CRED_MAGIC;
+#endif
+	cred->cr_uid = acred->uid;
+}
+EXPORT_SYMBOL(rpcauth_init_cred);
+
 struct rpc_cred *
 rpcauth_bindcred(struct rpc_task *task)
 {
-	struct rpc_auth *auth = task->tk_auth;
+	struct rpc_auth *auth = task->tk_client->cl_auth;
 	struct auth_cred acred = {
 		.uid = current->fsuid,
 		.gid = current->fsgid,
@@ -288,7 +382,7 @@ rpcauth_bindcred(struct rpc_task *task)
 	int flags = 0;
 
 	dprintk("RPC: %5u looking up %s cred\n",
-		task->tk_pid, task->tk_auth->au_ops->au_name);
+		task->tk_pid, task->tk_client->cl_auth->au_ops->au_name);
 	get_group_info(acred.group_info);
 	if (task->tk_flags & RPC_TASK_ROOTCREDS)
 		flags |= RPCAUTH_LOOKUP_ROOTCREDS;
@@ -304,19 +398,42 @@ rpcauth_bindcred(struct rpc_task *task)
 void
 rpcauth_holdcred(struct rpc_task *task)
 {
-	dprintk("RPC: %5u holding %s cred %p\n",
-		task->tk_pid, task->tk_auth->au_ops->au_name,
-		task->tk_msg.rpc_cred);
-	if (task->tk_msg.rpc_cred)
-		get_rpccred(task->tk_msg.rpc_cred);
+	struct rpc_cred *cred = task->tk_msg.rpc_cred;
+	if (cred != NULL) {
+		get_rpccred(cred);
+		dprintk("RPC: %5u holding %s cred %p\n", task->tk_pid,
+				cred->cr_auth->au_ops->au_name, cred);
+	}
 }
 
 void
 put_rpccred(struct rpc_cred *cred)
 {
-	cred->cr_expire = jiffies;
+	/* Fast path for unhashed credentials */
+	if (test_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags) != 0)
+		goto need_lock;
+
 	if (!atomic_dec_and_test(&cred->cr_count))
 		return;
+	goto out_destroy;
+need_lock:
+	if (!atomic_dec_and_lock(&cred->cr_count, &rpc_credcache_lock))
+		return;
+	if (!list_empty(&cred->cr_lru)) {
+		number_cred_unused--;
+		list_del_init(&cred->cr_lru);
+	}
+	if (test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) == 0)
+		rpcauth_unhash_cred(cred);
+	else if (test_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags) != 0) {
+		cred->cr_expire = jiffies;
+		list_add_tail(&cred->cr_lru, &cred_unused);
+		number_cred_unused++;
+		spin_unlock(&rpc_credcache_lock);
+		return;
+	}
+	spin_unlock(&rpc_credcache_lock);
+out_destroy:
 	cred->cr_ops->crdestroy(cred);
 }
 
@@ -326,7 +443,7 @@ rpcauth_unbindcred(struct rpc_task *task)
 	struct rpc_cred	*cred = task->tk_msg.rpc_cred;
 
 	dprintk("RPC: %5u releasing %s cred %p\n",
-		task->tk_pid, task->tk_auth->au_ops->au_name, cred);
+		task->tk_pid, cred->cr_auth->au_ops->au_name, cred);
 
 	put_rpccred(cred);
 	task->tk_msg.rpc_cred = NULL;
@@ -338,7 +455,7 @@ rpcauth_marshcred(struct rpc_task *task, __be32 *p)
 	struct rpc_cred	*cred = task->tk_msg.rpc_cred;
 
 	dprintk("RPC: %5u marshaling %s cred %p\n",
-		task->tk_pid, task->tk_auth->au_ops->au_name, cred);
+		task->tk_pid, cred->cr_auth->au_ops->au_name, cred);
 
 	return cred->cr_ops->crmarshal(task, p);
 }
@@ -349,7 +466,7 @@ rpcauth_checkverf(struct rpc_task *task, __be32 *p)
 	struct rpc_cred	*cred = task->tk_msg.rpc_cred;
 
 	dprintk("RPC: %5u validating %s cred %p\n",
-		task->tk_pid, task->tk_auth->au_ops->au_name, cred);
+		task->tk_pid, cred->cr_auth->au_ops->au_name, cred);
 
 	return cred->cr_ops->crvalidate(task, p);
 }
@@ -359,13 +476,17 @@ rpcauth_wrap_req(struct rpc_task *task, kxdrproc_t encode, void *rqstp,
 		__be32 *data, void *obj)
 {
 	struct rpc_cred *cred = task->tk_msg.rpc_cred;
+	int ret;
 
 	dprintk("RPC: %5u using %s cred %p to wrap rpc data\n",
 			task->tk_pid, cred->cr_ops->cr_name, cred);
 	if (cred->cr_ops->crwrap_req)
 		return cred->cr_ops->crwrap_req(task, encode, rqstp, data, obj);
 	/* By default, we encode the arguments normally. */
-	return encode(rqstp, data, obj);
+	lock_kernel();
+	ret = encode(rqstp, data, obj);
+	unlock_kernel();
+	return ret;
 }
 
 int
@@ -373,6 +494,7 @@ rpcauth_unwrap_resp(struct rpc_task *task, kxdrproc_t decode, void *rqstp,
 		__be32 *data, void *obj)
 {
 	struct rpc_cred *cred = task->tk_msg.rpc_cred;
+	int ret;
 
 	dprintk("RPC: %5u using %s cred %p to unwrap rpc data\n",
 			task->tk_pid, cred->cr_ops->cr_name, cred);
@@ -380,7 +502,10 @@ rpcauth_unwrap_resp(struct rpc_task *task, kxdrproc_t decode, void *rqstp,
 		return cred->cr_ops->crunwrap_resp(task, decode, rqstp,
 						   data, obj);
 	/* By default, we decode the arguments normally. */
-	return decode(rqstp, data, obj);
+	lock_kernel();
+	ret = decode(rqstp, data, obj);
+	unlock_kernel();
+	return ret;
 }
 
 int
@@ -390,7 +515,7 @@ rpcauth_refreshcred(struct rpc_task *task)
 	int err;
 
 	dprintk("RPC: %5u refreshing %s cred %p\n",
-		task->tk_pid, task->tk_auth->au_ops->au_name, cred);
+		task->tk_pid, cred->cr_auth->au_ops->au_name, cred);
 
 	err = cred->cr_ops->crrefresh(task);
 	if (err < 0)
@@ -401,17 +526,34 @@ rpcauth_refreshcred(struct rpc_task *task)
 void
 rpcauth_invalcred(struct rpc_task *task)
 {
+	struct rpc_cred *cred = task->tk_msg.rpc_cred;
+
 	dprintk("RPC: %5u invalidating %s cred %p\n",
-		task->tk_pid, task->tk_auth->au_ops->au_name, task->tk_msg.rpc_cred);
-	spin_lock(&rpc_credcache_lock);
-	if (task->tk_msg.rpc_cred)
-		task->tk_msg.rpc_cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE;
-	spin_unlock(&rpc_credcache_lock);
+		task->tk_pid, cred->cr_auth->au_ops->au_name, cred);
+	if (cred)
+		clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags);
 }
 
 int
 rpcauth_uptodatecred(struct rpc_task *task)
 {
-	return !(task->tk_msg.rpc_cred) ||
-		(task->tk_msg.rpc_cred->cr_flags & RPCAUTH_CRED_UPTODATE);
+	struct rpc_cred *cred = task->tk_msg.rpc_cred;
+
+	return cred == NULL ||
+		test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) != 0;
+}
+
+
+static struct shrinker *rpc_cred_shrinker;
+
+void __init rpcauth_init_module(void)
+{
+	rpc_init_authunix();
+	rpc_cred_shrinker = set_shrinker(DEFAULT_SEEKS, rpcauth_cache_shrinker);
+}
+
+void __exit rpcauth_remove_module(void)
+{
+	if (rpc_cred_shrinker != NULL)
+		remove_shrinker(rpc_cred_shrinker);
 }

net/sunrpc/auth_gss/auth_gss.c

@@ -54,9 +54,10 @@
 #include <linux/sunrpc/gss_api.h>
 #include <asm/uaccess.h>
 
-static struct rpc_authops authgss_ops;
+static const struct rpc_authops authgss_ops;
 
-static struct rpc_credops gss_credops;
+static const struct rpc_credops gss_credops;
+static const struct rpc_credops gss_nullops;
 
 #ifdef RPC_DEBUG
 # define RPCDBG_FACILITY	RPCDBG_AUTH
@@ -64,7 +65,6 @@ static struct rpc_credops gss_credops;
 
 #define NFS_NGROUPS	16
 
-#define GSS_CRED_EXPIRE		(60 * HZ)	/* XXX: reasonable? */
 #define GSS_CRED_SLACK		1024		/* XXX: unused */
 /* length of a krb5 verifier (48), plus data added before arguments when
  * using integrity (two 4-byte integers): */
@@ -79,19 +79,16 @@ static struct rpc_credops gss_credops;
 /* dump the buffer in `emacs-hexl' style */
 #define isprint(c)      ((c > 0x1f) && (c < 0x7f))
 
-static DEFINE_RWLOCK(gss_ctx_lock);
-
 struct gss_auth {
+	struct kref kref;
 	struct rpc_auth rpc_auth;
 	struct gss_api_mech *mech;
 	enum rpc_gss_svc service;
-	struct list_head upcalls;
 	struct rpc_clnt *client;
 	struct dentry *dentry;
-	spinlock_t lock;
 };
 
-static void gss_destroy_ctx(struct gss_cl_ctx *);
+static void gss_free_ctx(struct gss_cl_ctx *);
 static struct rpc_pipe_ops gss_upcall_ops;
 
 static inline struct gss_cl_ctx *
@@ -105,20 +102,24 @@ static inline void
 gss_put_ctx(struct gss_cl_ctx *ctx)
 {
 	if (atomic_dec_and_test(&ctx->count))
-		gss_destroy_ctx(ctx);
+		gss_free_ctx(ctx);
 }
 
+/* gss_cred_set_ctx:
+ * called by gss_upcall_callback and gss_create_upcall in order
+ * to set the gss context. The actual exchange of an old context
+ * and a new one is protected by the inode->i_lock.
+ */
 static void
 gss_cred_set_ctx(struct rpc_cred *cred, struct gss_cl_ctx *ctx)
 {
 	struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base);
 	struct gss_cl_ctx *old;
-	write_lock(&gss_ctx_lock);
+
 	old = gss_cred->gc_ctx;
-	gss_cred->gc_ctx = ctx;
-	cred->cr_flags |= RPCAUTH_CRED_UPTODATE;
-	cred->cr_flags &= ~RPCAUTH_CRED_NEW;
-	write_unlock(&gss_ctx_lock);
+	rcu_assign_pointer(gss_cred->gc_ctx, ctx);
+	set_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags);
+	clear_bit(RPCAUTH_CRED_NEW, &cred->cr_flags);
 	if (old)
 		gss_put_ctx(old);
 }
@@ -129,10 +130,10 @@ gss_cred_is_uptodate_ctx(struct rpc_cred *cred)
 	struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base);
 	int res = 0;
 
-	read_lock(&gss_ctx_lock);
-	if ((cred->cr_flags & RPCAUTH_CRED_UPTODATE) && gss_cred->gc_ctx)
+	rcu_read_lock();
+	if (test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) && gss_cred->gc_ctx)
 		res = 1;
-	read_unlock(&gss_ctx_lock);
+	rcu_read_unlock();
 	return res;
 }
 
@@ -171,10 +172,10 @@ gss_cred_get_ctx(struct rpc_cred *cred)
 	struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base);
 	struct gss_cl_ctx *ctx = NULL;
 
-	read_lock(&gss_ctx_lock);
+	rcu_read_lock();
 	if (gss_cred->gc_ctx)
 		ctx = gss_get_ctx(gss_cred->gc_ctx);
-	read_unlock(&gss_ctx_lock);
+	rcu_read_unlock();
 	return ctx;
 }
 
@@ -269,10 +270,10 @@ gss_release_msg(struct gss_upcall_msg *gss_msg)
 }
 
 static struct gss_upcall_msg *
-__gss_find_upcall(struct gss_auth *gss_auth, uid_t uid)
+__gss_find_upcall(struct rpc_inode *rpci, uid_t uid)
 {
 	struct gss_upcall_msg *pos;
-	list_for_each_entry(pos, &gss_auth->upcalls, list) {
+	list_for_each_entry(pos, &rpci->in_downcall, list) {
 		if (pos->uid != uid)
 			continue;
 		atomic_inc(&pos->count);
@@ -290,24 +291,24 @@ __gss_find_upcall(struct gss_auth *gss_auth, uid_t uid)
 static inline struct gss_upcall_msg *
 gss_add_msg(struct gss_auth *gss_auth, struct gss_upcall_msg *gss_msg)
 {
+	struct inode *inode = gss_auth->dentry->d_inode;
+	struct rpc_inode *rpci = RPC_I(inode);
 	struct gss_upcall_msg *old;
 
-	spin_lock(&gss_auth->lock);
-	old = __gss_find_upcall(gss_auth, gss_msg->uid);
+	spin_lock(&inode->i_lock);
+	old = __gss_find_upcall(rpci, gss_msg->uid);
 	if (old == NULL) {
 		atomic_inc(&gss_msg->count);
-		list_add(&gss_msg->list, &gss_auth->upcalls);
+		list_add(&gss_msg->list, &rpci->in_downcall);
 	} else
 		gss_msg = old;
-	spin_unlock(&gss_auth->lock);
+	spin_unlock(&inode->i_lock);
 	return gss_msg;
 }
 
 static void
 __gss_unhash_msg(struct gss_upcall_msg *gss_msg)
 {
-	if (list_empty(&gss_msg->list))
-		return;
 	list_del_init(&gss_msg->list);
 	rpc_wake_up_status(&gss_msg->rpc_waitqueue, gss_msg->msg.errno);
 	wake_up_all(&gss_msg->waitqueue);
@@ -318,10 +319,14 @@ static void
 gss_unhash_msg(struct gss_upcall_msg *gss_msg)
 {
 	struct gss_auth *gss_auth = gss_msg->auth;
+	struct inode *inode = gss_auth->dentry->d_inode;
 
-	spin_lock(&gss_auth->lock);
-	__gss_unhash_msg(gss_msg);
-	spin_unlock(&gss_auth->lock);
+	if (list_empty(&gss_msg->list))
+		return;
+	spin_lock(&inode->i_lock);
+	if (!list_empty(&gss_msg->list))
+		__gss_unhash_msg(gss_msg);
+	spin_unlock(&inode->i_lock);
 }
 
 static void
@@ -330,16 +335,16 @@ gss_upcall_callback(struct rpc_task *task)
 	struct gss_cred *gss_cred = container_of(task->tk_msg.rpc_cred,
 			struct gss_cred, gc_base);
 	struct gss_upcall_msg *gss_msg = gss_cred->gc_upcall;
+	struct inode *inode = gss_msg->auth->dentry->d_inode;
 
-	BUG_ON(gss_msg == NULL);
+	spin_lock(&inode->i_lock);
 	if (gss_msg->ctx)
 		gss_cred_set_ctx(task->tk_msg.rpc_cred, gss_get_ctx(gss_msg->ctx));
 	else
 		task->tk_status = gss_msg->msg.errno;
-	spin_lock(&gss_msg->auth->lock);
 	gss_cred->gc_upcall = NULL;
 	rpc_wake_up_status(&gss_msg->rpc_waitqueue, gss_msg->msg.errno);
-	spin_unlock(&gss_msg->auth->lock);
+	spin_unlock(&inode->i_lock);
 	gss_release_msg(gss_msg);
 }
 
@@ -386,11 +391,12 @@ static inline int
 gss_refresh_upcall(struct rpc_task *task)
 {
 	struct rpc_cred *cred = task->tk_msg.rpc_cred;
-	struct gss_auth *gss_auth = container_of(task->tk_client->cl_auth,
+	struct gss_auth *gss_auth = container_of(cred->cr_auth,
 			struct gss_auth, rpc_auth);
 	struct gss_cred *gss_cred = container_of(cred,
 			struct gss_cred, gc_base);
 	struct gss_upcall_msg *gss_msg;
+	struct inode *inode = gss_auth->dentry->d_inode;
 	int err = 0;
 
 	dprintk("RPC: %5u gss_refresh_upcall for uid %u\n", task->tk_pid,
@@ -400,7 +406,7 @@ gss_refresh_upcall(struct rpc_task *task)
 		err = PTR_ERR(gss_msg);
 		goto out;
 	}
-	spin_lock(&gss_auth->lock);
+	spin_lock(&inode->i_lock);
 	if (gss_cred->gc_upcall != NULL)
 		rpc_sleep_on(&gss_cred->gc_upcall->rpc_waitqueue, task, NULL, NULL);
 	else if (gss_msg->ctx == NULL && gss_msg->msg.errno >= 0) {
@@ -411,7 +417,7 @@ gss_refresh_upcall(struct rpc_task *task)
 		rpc_sleep_on(&gss_msg->rpc_waitqueue, task, gss_upcall_callback, NULL);
 	} else
 		err = gss_msg->msg.errno;
-	spin_unlock(&gss_auth->lock);
+	spin_unlock(&inode->i_lock);
 	gss_release_msg(gss_msg);
 out:
 	dprintk("RPC: %5u gss_refresh_upcall for uid %u result %d\n",
@@ -422,6 +428,7 @@ out:
 static inline int
 gss_create_upcall(struct gss_auth *gss_auth, struct gss_cred *gss_cred)
 {
+	struct inode *inode = gss_auth->dentry->d_inode;
 	struct rpc_cred *cred = &gss_cred->gc_base;
 	struct gss_upcall_msg *gss_msg;
 	DEFINE_WAIT(wait);
@@ -435,12 +442,11 @@ gss_create_upcall(struct gss_auth *gss_auth, struct gss_cred *gss_cred)
 	}
 	for (;;) {
 		prepare_to_wait(&gss_msg->waitqueue, &wait, TASK_INTERRUPTIBLE);
-		spin_lock(&gss_auth->lock);
+		spin_lock(&inode->i_lock);
 		if (gss_msg->ctx != NULL || gss_msg->msg.errno < 0) {
-			spin_unlock(&gss_auth->lock);
 			break;
 		}
-		spin_unlock(&gss_auth->lock);
+		spin_unlock(&inode->i_lock);
 		if (signalled()) {
 			err = -ERESTARTSYS;
 			goto out_intr;
@@ -451,6 +457,7 @@ gss_create_upcall(struct gss_auth *gss_auth, struct gss_cred *gss_cred)
 		gss_cred_set_ctx(cred, gss_get_ctx(gss_msg->ctx));
 	else
 		err = gss_msg->msg.errno;
+	spin_unlock(&inode->i_lock);
 out_intr:
 	finish_wait(&gss_msg->waitqueue, &wait);
 	gss_release_msg(gss_msg);
@@ -489,12 +496,11 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
 	const void *p, *end;
 	void *buf;
 	struct rpc_clnt *clnt;
-	struct gss_auth *gss_auth;
-	struct rpc_cred *cred;
 	struct gss_upcall_msg *gss_msg;
+	struct inode *inode = filp->f_path.dentry->d_inode;
 	struct gss_cl_ctx *ctx;
 	uid_t uid;
-	int err = -EFBIG;
+	ssize_t err = -EFBIG;
 
 	if (mlen > MSG_BUF_MAXSIZE)
 		goto out;
@@ -503,7 +509,7 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
 	if (!buf)
 		goto out;
 
-	clnt = RPC_I(filp->f_path.dentry->d_inode)->private;
+	clnt = RPC_I(inode)->private;
 	err = -EFAULT;
 	if (copy_from_user(buf, src, mlen))
 		goto err;
@@ -519,43 +525,38 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)
 	ctx = gss_alloc_context();
 	if (ctx == NULL)
 		goto err;
-	err = 0;
-	gss_auth = container_of(clnt->cl_auth, struct gss_auth, rpc_auth);
-	p = gss_fill_context(p, end, ctx, gss_auth->mech);
+
+	err = -ENOENT;
+	/* Find a matching upcall */
+	spin_lock(&inode->i_lock);
+	gss_msg = __gss_find_upcall(RPC_I(inode), uid);
+	if (gss_msg == NULL) {
+		spin_unlock(&inode->i_lock);
+		goto err_put_ctx;
+	}
+	list_del_init(&gss_msg->list);
+	spin_unlock(&inode->i_lock);
+
+	p = gss_fill_context(p, end, ctx, gss_msg->auth->mech);
 	if (IS_ERR(p)) {
 		err = PTR_ERR(p);
-		if (err != -EACCES)
-			goto err_put_ctx;
-	}
-	spin_lock(&gss_auth->lock);
-	gss_msg = __gss_find_upcall(gss_auth, uid);
-	if (gss_msg) {
-		if (err == 0 && gss_msg->ctx == NULL)
-			gss_msg->ctx = gss_get_ctx(ctx);
-		gss_msg->msg.errno = err;
-		__gss_unhash_msg(gss_msg);
-		spin_unlock(&gss_auth->lock);
-		gss_release_msg(gss_msg);
-	} else {
-		struct auth_cred acred = { .uid = uid };
-		spin_unlock(&gss_auth->lock);
-		cred = rpcauth_lookup_credcache(clnt->cl_auth, &acred, RPCAUTH_LOOKUP_NEW);
-		if (IS_ERR(cred)) {
-			err = PTR_ERR(cred);
-			goto err_put_ctx;
-		}
-		gss_cred_set_ctx(cred, gss_get_ctx(ctx));
+		gss_msg->msg.errno = (err == -EACCES) ? -EACCES : -EAGAIN;
+		goto err_release_msg;
 	}
-	gss_put_ctx(ctx);
-	kfree(buf);
-	dprintk("RPC:       gss_pipe_downcall returning length %Zu\n", mlen);
-	return mlen;
+	gss_msg->ctx = gss_get_ctx(ctx);
+	err = mlen;
+
+err_release_msg:
+	spin_lock(&inode->i_lock);
+	__gss_unhash_msg(gss_msg);
+	spin_unlock(&inode->i_lock);
+	gss_release_msg(gss_msg);
 err_put_ctx:
 	gss_put_ctx(ctx);
 err:
 	kfree(buf);
 out:
-	dprintk("RPC:       gss_pipe_downcall returning %d\n", err);
+	dprintk("RPC:       gss_pipe_downcall returning %Zd\n", err);
 	return err;
 }
 
@@ -563,27 +564,21 @@ static void
 gss_pipe_release(struct inode *inode)
 {
 	struct rpc_inode *rpci = RPC_I(inode);
-	struct rpc_clnt *clnt;
-	struct rpc_auth *auth;
-	struct gss_auth *gss_auth;
+	struct gss_upcall_msg *gss_msg;
 
-	clnt = rpci->private;
-	auth = clnt->cl_auth;
-	gss_auth = container_of(auth, struct gss_auth, rpc_auth);
-	spin_lock(&gss_auth->lock);
-	while (!list_empty(&gss_auth->upcalls)) {
-		struct gss_upcall_msg *gss_msg;
+	spin_lock(&inode->i_lock);
+	while (!list_empty(&rpci->in_downcall)) {
 
-		gss_msg = list_entry(gss_auth->upcalls.next,
+		gss_msg = list_entry(rpci->in_downcall.next,
 				struct gss_upcall_msg, list);
 		gss_msg->msg.errno = -EPIPE;
 		atomic_inc(&gss_msg->count);
 		__gss_unhash_msg(gss_msg);
-		spin_unlock(&gss_auth->lock);
+		spin_unlock(&inode->i_lock);
 		gss_release_msg(gss_msg);
-		spin_lock(&gss_auth->lock);
+		spin_lock(&inode->i_lock);
 	}
-	spin_unlock(&gss_auth->lock);
+	spin_unlock(&inode->i_lock);
 }
 
 static void
@@ -637,18 +632,13 @@ gss_create(struct rpc_clnt *clnt, rpc_authflavor_t flavor)
 	gss_auth->service = gss_pseudoflavor_to_service(gss_auth->mech, flavor);
 	if (gss_auth->service == 0)
 		goto err_put_mech;
-	INIT_LIST_HEAD(&gss_auth->upcalls);
-	spin_lock_init(&gss_auth->lock);
 	auth = &gss_auth->rpc_auth;
 	auth->au_cslack = GSS_CRED_SLACK >> 2;
 	auth->au_rslack = GSS_VERF_SLACK >> 2;
 	auth->au_ops = &authgss_ops;
 	auth->au_flavor = flavor;
 	atomic_set(&auth->au_count, 1);
-
-	err = rpcauth_init_credcache(auth, GSS_CRED_EXPIRE);
-	if (err)
-		goto err_put_mech;
+	kref_init(&gss_auth->kref);
 
 	gss_auth->dentry = rpc_mkpipe(clnt->cl_dentry, gss_auth->mech->gm_name,
 			clnt, &gss_upcall_ops, RPC_PIPE_WAIT_FOR_OPEN);
@@ -657,7 +647,13 @@ gss_create(struct rpc_clnt *clnt, rpc_authflavor_t flavor)
 		goto err_put_mech;
 	}
 
+	err = rpcauth_init_credcache(auth);
+	if (err)
+		goto err_unlink_pipe;
+
 	return auth;
+err_unlink_pipe:
+	rpc_unlink(gss_auth->dentry);
 err_put_mech:
 	gss_mech_put(gss_auth->mech);
 err_free:
@@ -667,6 +663,25 @@ out_dec:
 	return ERR_PTR(err);
 }
 
+static void
+gss_free(struct gss_auth *gss_auth)
+{
+	rpc_unlink(gss_auth->dentry);
+	gss_auth->dentry = NULL;
+	gss_mech_put(gss_auth->mech);
+
+	kfree(gss_auth);
+	module_put(THIS_MODULE);
+}
+
+static void
+gss_free_callback(struct kref *kref)
+{
+	struct gss_auth *gss_auth = container_of(kref, struct gss_auth, kref);
+
+	gss_free(gss_auth);
+}
+
 static void
 gss_destroy(struct rpc_auth *auth)
 {
@@ -675,23 +690,51 @@ gss_destroy(struct rpc_auth *auth)
 	dprintk("RPC:       destroying GSS authenticator %p flavor %d\n",
 			auth, auth->au_flavor);
 
+	rpcauth_destroy_credcache(auth);
+
 	gss_auth = container_of(auth, struct gss_auth, rpc_auth);
-	rpc_unlink(gss_auth->dentry);
-	gss_auth->dentry = NULL;
-	gss_mech_put(gss_auth->mech);
+	kref_put(&gss_auth->kref, gss_free_callback);
+}
 
-	rpcauth_free_credcache(auth);
-	kfree(gss_auth);
-	module_put(THIS_MODULE);
+/*
+ * gss_destroying_context will cause the RPCSEC_GSS to send a NULL RPC call
+ * to the server with the GSS control procedure field set to
+ * RPC_GSS_PROC_DESTROY. This should normally cause the server to release
+ * all RPCSEC_GSS state associated with that context.
+ */
+static int
+gss_destroying_context(struct rpc_cred *cred)
+{
+	struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base);
+	struct gss_auth *gss_auth = container_of(cred->cr_auth, struct gss_auth, rpc_auth);
+	struct rpc_task *task;
+
+	if (gss_cred->gc_ctx == NULL ||
+			gss_cred->gc_ctx->gc_proc == RPC_GSS_PROC_DESTROY)
+		return 0;
+
+	gss_cred->gc_ctx->gc_proc = RPC_GSS_PROC_DESTROY;
+	cred->cr_ops = &gss_nullops;
+
+	/* Take a reference to ensure the cred will be destroyed either
+	 * by the RPC call or by the put_rpccred() below */
+	get_rpccred(cred);
+
+	task = rpc_call_null(gss_auth->client, cred, RPC_TASK_ASYNC);
+	if (!IS_ERR(task))
+		rpc_put_task(task);
+
+	put_rpccred(cred);
+	return 1;
 }
 
-/* gss_destroy_cred (and gss_destroy_ctx) are used to clean up after failure
+/* gss_destroy_cred (and gss_free_ctx) are used to clean up after failure
  * to create a new cred or context, so they check that things have been
  * allocated before freeing them. */
 static void
-gss_destroy_ctx(struct gss_cl_ctx *ctx)
+gss_do_free_ctx(struct gss_cl_ctx *ctx)
 {
-	dprintk("RPC:       gss_destroy_ctx\n");
+	dprintk("RPC:       gss_free_ctx\n");
 
 	if (ctx->gc_gss_ctx)
 		gss_delete_sec_context(&ctx->gc_gss_ctx);
@@ -701,15 +744,46 @@ gss_destroy_ctx(struct gss_cl_ctx *ctx)
 }
 
 static void
-gss_destroy_cred(struct rpc_cred *rc)
+gss_free_ctx_callback(struct rcu_head *head)
 {
-	struct gss_cred *cred = container_of(rc, struct gss_cred, gc_base);
+	struct gss_cl_ctx *ctx = container_of(head, struct gss_cl_ctx, gc_rcu);
+	gss_do_free_ctx(ctx);
+}
 
-	dprintk("RPC:       gss_destroy_cred \n");
+static void
+gss_free_ctx(struct gss_cl_ctx *ctx)
+{
+	call_rcu(&ctx->gc_rcu, gss_free_ctx_callback);
+}
 
-	if (cred->gc_ctx)
-		gss_put_ctx(cred->gc_ctx);
-	kfree(cred);
+static void
+gss_free_cred(struct gss_cred *gss_cred)
+{
+	dprintk("RPC:       gss_free_cred %p\n", gss_cred);
+	kfree(gss_cred);
+}
+
+static void
+gss_free_cred_callback(struct rcu_head *head)
+{
+	struct gss_cred *gss_cred = container_of(head, struct gss_cred, gc_base.cr_rcu);
+	gss_free_cred(gss_cred);
+}
+
+static void
+gss_destroy_cred(struct rpc_cred *cred)
+{
+	struct gss_cred *gss_cred = container_of(cred, struct gss_cred, gc_base);
+	struct gss_auth *gss_auth = container_of(cred->cr_auth, struct gss_auth, rpc_auth);
+	struct gss_cl_ctx *ctx = gss_cred->gc_ctx;
+
+	if (gss_destroying_context(cred))
+		return;
+	rcu_assign_pointer(gss_cred->gc_ctx, NULL);
+	call_rcu(&cred->cr_rcu, gss_free_cred_callback);
+	if (ctx)
+		gss_put_ctx(ctx);
+	kref_put(&gss_auth->kref, gss_free_callback);
 }
 
 /*
@@ -734,16 +808,14 @@ gss_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags)
 	if (!(cred = kzalloc(sizeof(*cred), GFP_KERNEL)))
 		goto out_err;
 
-	atomic_set(&cred->gc_count, 1);
-	cred->gc_uid = acred->uid;
+	rpcauth_init_cred(&cred->gc_base, acred, auth, &gss_credops);
 	/*
 	 * Note: in order to force a call to call_refresh(), we deliberately
 	 * fail to flag the credential as RPCAUTH_CRED_UPTODATE.
 	 */
-	cred->gc_flags = 0;
-	cred->gc_base.cr_ops = &gss_credops;
-	cred->gc_base.cr_flags = RPCAUTH_CRED_NEW;
+	cred->gc_base.cr_flags = 1UL << RPCAUTH_CRED_NEW;
 	cred->gc_service = gss_auth->service;
+	kref_get(&gss_auth->kref);
 	return &cred->gc_base;
 
 out_err:
@@ -774,7 +846,7 @@ gss_match(struct auth_cred *acred, struct rpc_cred *rc, int flags)
 	 * we don't really care if the credential has expired or not,
 	 * since the caller should be prepared to reinitialise it.
 	 */
-	if ((flags & RPCAUTH_LOOKUP_NEW) && (rc->cr_flags & RPCAUTH_CRED_NEW))
+	if ((flags & RPCAUTH_LOOKUP_NEW) && test_bit(RPCAUTH_CRED_NEW, &rc->cr_flags))
 		goto out;
 	/* Don't match with creds that have expired. */
 	if (gss_cred->gc_ctx && time_after(jiffies, gss_cred->gc_ctx->gc_expiry))
@@ -830,7 +902,7 @@ gss_marshal(struct rpc_task *task, __be32 *p)
 	mic.data = (u8 *)(p + 1);
 	maj_stat = gss_get_mic(ctx->gc_gss_ctx, &verf_buf, &mic);
 	if (maj_stat == GSS_S_CONTEXT_EXPIRED) {
-		cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE;
+		clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags);
 	} else if (maj_stat != 0) {
 		printk("gss_marshal: gss_get_mic FAILED (%d)\n", maj_stat);
 		goto out_put_ctx;
@@ -855,6 +927,13 @@ gss_refresh(struct rpc_task *task)
 	return 0;
 }
 
+/* Dummy refresh routine: used only when destroying the context */
+static int
+gss_refresh_null(struct rpc_task *task)
+{
+	return -EACCES;
+}
+
 static __be32 *
 gss_validate(struct rpc_task *task, __be32 *p)
 {
@@ -883,12 +962,15 @@ gss_validate(struct rpc_task *task, __be32 *p)
 
 	maj_stat = gss_verify_mic(ctx->gc_gss_ctx, &verf_buf, &mic);
 	if (maj_stat == GSS_S_CONTEXT_EXPIRED)
-		cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE;
-	if (maj_stat)
+		clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags);
+	if (maj_stat) {
+		dprintk("RPC: %5u gss_validate: gss_verify_mic returned"
+				"error 0x%08x\n", task->tk_pid, maj_stat);
 		goto out_bad;
+	}
 	/* We leave it to unwrap to calculate au_rslack. For now we just
 	 * calculate the length of the verifier: */
-	task->tk_auth->au_verfsize = XDR_QUADLEN(len) + 2;
+	cred->cr_auth->au_verfsize = XDR_QUADLEN(len) + 2;
 	gss_put_ctx(ctx);
 	dprintk("RPC: %5u gss_validate: gss_verify_mic succeeded.\n",
 			task->tk_pid);
@@ -917,7 +999,9 @@ gss_wrap_req_integ(struct rpc_cred *cred, struct gss_cl_ctx *ctx,
 	offset = (u8 *)p - (u8 *)snd_buf->head[0].iov_base;
 	*p++ = htonl(rqstp->rq_seqno);
 
+	lock_kernel();
 	status = encode(rqstp, p, obj);
+	unlock_kernel();
 	if (status)
 		return status;
 
@@ -937,7 +1021,7 @@ gss_wrap_req_integ(struct rpc_cred *cred, struct gss_cl_ctx *ctx,
 	maj_stat = gss_get_mic(ctx->gc_gss_ctx, &integ_buf, &mic);
 	status = -EIO; /* XXX? */
 	if (maj_stat == GSS_S_CONTEXT_EXPIRED)
-		cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE;
+		clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags);
 	else if (maj_stat)
 		return status;
 	q = xdr_encode_opaque(p, NULL, mic.len);
@@ -1011,7 +1095,9 @@ gss_wrap_req_priv(struct rpc_cred *cred, struct gss_cl_ctx *ctx,
 	offset = (u8 *)p - (u8 *)snd_buf->head[0].iov_base;
 	*p++ = htonl(rqstp->rq_seqno);
 
+	lock_kernel();
 	status = encode(rqstp, p, obj);
+	unlock_kernel();
 	if (status)
 		return status;
 
@@ -1036,7 +1122,7 @@ gss_wrap_req_priv(struct rpc_cred *cred, struct gss_cl_ctx *ctx,
 	/* We're assuming that when GSS_S_CONTEXT_EXPIRED, the encryption was
 	 * done anyway, so it's safe to put the request on the wire: */
 	if (maj_stat == GSS_S_CONTEXT_EXPIRED)
-		cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE;
+		clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags);
 	else if (maj_stat)
 		return status;
 
@@ -1070,12 +1156,16 @@ gss_wrap_req(struct rpc_task *task,
 		/* The spec seems a little ambiguous here, but I think that not
 		 * wrapping context destruction requests makes the most sense.
 		 */
+		lock_kernel();
 		status = encode(rqstp, p, obj);
+		unlock_kernel();
 		goto out;
 	}
 	switch (gss_cred->gc_service) {
 		case RPC_GSS_SVC_NONE:
+			lock_kernel();
 			status = encode(rqstp, p, obj);
+			unlock_kernel();
 			break;
 		case RPC_GSS_SVC_INTEGRITY:
 			status = gss_wrap_req_integ(cred, ctx, encode,
@@ -1123,7 +1213,7 @@ gss_unwrap_resp_integ(struct rpc_cred *cred, struct gss_cl_ctx *ctx,
 
 	maj_stat = gss_verify_mic(ctx->gc_gss_ctx, &integ_buf, &mic);
 	if (maj_stat == GSS_S_CONTEXT_EXPIRED)
-		cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE;
+		clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags);
 	if (maj_stat != GSS_S_COMPLETE)
 		return status;
 	return 0;
@@ -1148,7 +1238,7 @@ gss_unwrap_resp_priv(struct rpc_cred *cred, struct gss_cl_ctx *ctx,
 
 	maj_stat = gss_unwrap(ctx->gc_gss_ctx, offset, rcv_buf);
 	if (maj_stat == GSS_S_CONTEXT_EXPIRED)
-		cred->cr_flags &= ~RPCAUTH_CRED_UPTODATE;
+		clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags);
 	if (maj_stat != GSS_S_COMPLETE)
 		return status;
 	if (ntohl(*(*p)++) != rqstp->rq_seqno)
@@ -1188,10 +1278,12 @@ gss_unwrap_resp(struct rpc_task *task,
 			break;
 	}
 	/* take into account extra slack for integrity and privacy cases: */
-	task->tk_auth->au_rslack = task->tk_auth->au_verfsize + (p - savedp)
+	cred->cr_auth->au_rslack = cred->cr_auth->au_verfsize + (p - savedp)
 						+ (savedlen - head->iov_len);
 out_decode:
+	lock_kernel();
 	status = decode(rqstp, p, obj);
+	unlock_kernel();
 out:
 	gss_put_ctx(ctx);
 	dprintk("RPC: %5u gss_unwrap_resp returning %d\n", task->tk_pid,
@@ -1199,7 +1291,7 @@ out:
 	return status;
 }
 
-static struct rpc_authops authgss_ops = {
+static const struct rpc_authops authgss_ops = {
 	.owner		= THIS_MODULE,
 	.au_flavor	= RPC_AUTH_GSS,
 #ifdef RPC_DEBUG
@@ -1211,7 +1303,7 @@ static struct rpc_authops authgss_ops = {
 	.crcreate	= gss_create_cred
 };
 
-static struct rpc_credops gss_credops = {
+static const struct rpc_credops gss_credops = {
 	.cr_name	= "AUTH_GSS",
 	.crdestroy	= gss_destroy_cred,
 	.cr_init	= gss_cred_init,
@@ -1223,6 +1315,17 @@ static struct rpc_credops gss_credops = {
 	.crunwrap_resp	= gss_unwrap_resp,
 };
 
+static const struct rpc_credops gss_nullops = {
+	.cr_name	= "AUTH_GSS",
+	.crdestroy	= gss_destroy_cred,
+	.crmatch	= gss_match,
+	.crmarshal	= gss_marshal,
+	.crrefresh	= gss_refresh_null,
+	.crvalidate	= gss_validate,
+	.crwrap_req	= gss_wrap_req,
+	.crunwrap_resp	= gss_unwrap_resp,
+};
+
 static struct rpc_pipe_ops gss_upcall_ops = {
 	.upcall		= gss_pipe_upcall,
 	.downcall	= gss_pipe_downcall,

net/sunrpc/auth_gss/gss_krb5_mech.c

@@ -201,7 +201,7 @@ gss_delete_sec_context_kerberos(void *internal_ctx) {
 	kfree(kctx);
 }
 
-static struct gss_api_ops gss_kerberos_ops = {
+static const struct gss_api_ops gss_kerberos_ops = {
 	.gss_import_sec_context	= gss_import_sec_context_kerberos,
 	.gss_get_mic		= gss_get_mic_kerberos,
 	.gss_verify_mic		= gss_verify_mic_kerberos,

net/sunrpc/auth_gss/gss_spkm3_mech.c

@@ -202,7 +202,7 @@ gss_get_mic_spkm3(struct gss_ctx	*ctx,
 	return err;
 }
 
-static struct gss_api_ops gss_spkm3_ops = {
+static const struct gss_api_ops gss_spkm3_ops = {
 	.gss_import_sec_context	= gss_import_sec_context_spkm3,
 	.gss_get_mic		= gss_get_mic_spkm3,
 	.gss_verify_mic		= gss_verify_mic_spkm3,

net/sunrpc/auth_null.c

@@ -76,7 +76,7 @@ nul_marshal(struct rpc_task *task, __be32 *p)
 static int
 nul_refresh(struct rpc_task *task)
 {
-	task->tk_msg.rpc_cred->cr_flags |= RPCAUTH_CRED_UPTODATE;
+	set_bit(RPCAUTH_CRED_UPTODATE, &task->tk_msg.rpc_cred->cr_flags);
 	return 0;
 }
 
@@ -101,7 +101,7 @@ nul_validate(struct rpc_task *task, __be32 *p)
 	return p;
 }
 
-struct rpc_authops authnull_ops = {
+const struct rpc_authops authnull_ops = {
 	.owner		= THIS_MODULE,
 	.au_flavor	= RPC_AUTH_NULL,
 #ifdef RPC_DEBUG
@@ -122,7 +122,7 @@ struct rpc_auth null_auth = {
 };
 
 static
-struct rpc_credops	null_credops = {
+const struct rpc_credops null_credops = {
 	.cr_name	= "AUTH_NULL",
 	.crdestroy	= nul_destroy_cred,
 	.crmatch	= nul_match,
@@ -133,9 +133,11 @@ struct rpc_credops	null_credops = {
 
 static
 struct rpc_cred null_cred = {
+	.cr_lru		= LIST_HEAD_INIT(null_cred.cr_lru),
+	.cr_auth	= &null_auth,
 	.cr_ops		= &null_credops,
 	.cr_count	= ATOMIC_INIT(1),
-	.cr_flags	= RPCAUTH_CRED_UPTODATE,
+	.cr_flags	= 1UL << RPCAUTH_CRED_UPTODATE,
 #ifdef RPC_DEBUG
 	.cr_magic	= RPCAUTH_CRED_MAGIC,
 #endif

net/sunrpc/auth_unix.c

@@ -20,11 +20,6 @@ struct unx_cred {
 	gid_t			uc_gids[NFS_NGROUPS];
 };
 #define uc_uid			uc_base.cr_uid
-#define uc_count		uc_base.cr_count
-#define uc_flags		uc_base.cr_flags
-#define uc_expire		uc_base.cr_expire
-
-#define UNX_CRED_EXPIRE		(60 * HZ)
 
 #define UNX_WRITESLACK		(21 + (UNX_MAXNODENAME >> 2))
 
@@ -34,15 +29,14 @@ struct unx_cred {
 
 static struct rpc_auth		unix_auth;
 static struct rpc_cred_cache	unix_cred_cache;
-static struct rpc_credops	unix_credops;
+static const struct rpc_credops	unix_credops;
 
 static struct rpc_auth *
 unx_create(struct rpc_clnt *clnt, rpc_authflavor_t flavor)
 {
 	dprintk("RPC:       creating UNIX authenticator for client %p\n",
 			clnt);
-	if (atomic_inc_return(&unix_auth.au_count) == 0)
-		unix_cred_cache.nextgc = jiffies + (unix_cred_cache.expire >> 1);
+	atomic_inc(&unix_auth.au_count);
 	return &unix_auth;
 }
 
@@ -50,7 +44,7 @@ static void
 unx_destroy(struct rpc_auth *auth)
 {
 	dprintk("RPC:       destroying UNIX authenticator %p\n", auth);
-	rpcauth_free_credcache(auth);
+	rpcauth_clear_credcache(auth->au_credcache);
 }
 
 /*
@@ -74,8 +68,8 @@ unx_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags)
 	if (!(cred = kmalloc(sizeof(*cred), GFP_KERNEL)))
 		return ERR_PTR(-ENOMEM);
 
-	atomic_set(&cred->uc_count, 1);
-	cred->uc_flags = RPCAUTH_CRED_UPTODATE;
+	rpcauth_init_cred(&cred->uc_base, acred, auth, &unix_credops);
+	cred->uc_base.cr_flags = 1UL << RPCAUTH_CRED_UPTODATE;
 	if (flags & RPCAUTH_LOOKUP_ROOTCREDS) {
 		cred->uc_uid = 0;
 		cred->uc_gid = 0;
@@ -85,22 +79,34 @@ unx_create_cred(struct rpc_auth *auth, struct auth_cred *acred, int flags)
 		if (groups > NFS_NGROUPS)
 			groups = NFS_NGROUPS;
 
-		cred->uc_uid = acred->uid;
 		cred->uc_gid = acred->gid;
 		for (i = 0; i < groups; i++)
 			cred->uc_gids[i] = GROUP_AT(acred->group_info, i);
 		if (i < NFS_NGROUPS)
 		  cred->uc_gids[i] = NOGROUP;
 	}
-	cred->uc_base.cr_ops = &unix_credops;
 
-	return (struct rpc_cred *) cred;
+	return &cred->uc_base;
+}
+
+static void
+unx_free_cred(struct unx_cred *unx_cred)
+{
+	dprintk("RPC:       unx_free_cred %p\n", unx_cred);
+	kfree(unx_cred);
+}
+
+static void
+unx_free_cred_callback(struct rcu_head *head)
+{
+	struct unx_cred *unx_cred = container_of(head, struct unx_cred, uc_base.cr_rcu);
+	unx_free_cred(unx_cred);
 }
 
 static void
 unx_destroy_cred(struct rpc_cred *cred)
 {
-	kfree(cred);
+	call_rcu(&cred->cr_rcu, unx_free_cred_callback);
 }
 
 /*
@@ -111,7 +117,7 @@ unx_destroy_cred(struct rpc_cred *cred)
 static int
 unx_match(struct auth_cred *acred, struct rpc_cred *rcred, int flags)
 {
-	struct unx_cred	*cred = (struct unx_cred *) rcred;
+	struct unx_cred	*cred = container_of(rcred, struct unx_cred, uc_base);
 	int		i;
 
 	if (!(flags & RPCAUTH_LOOKUP_ROOTCREDS)) {
@@ -142,7 +148,7 @@ static __be32 *
 unx_marshal(struct rpc_task *task, __be32 *p)
 {
 	struct rpc_clnt	*clnt = task->tk_client;
-	struct unx_cred	*cred = (struct unx_cred *) task->tk_msg.rpc_cred;
+	struct unx_cred	*cred = container_of(task->tk_msg.rpc_cred, struct unx_cred, uc_base);
 	__be32		*base, *hold;
 	int		i;
 
@@ -175,7 +181,7 @@ unx_marshal(struct rpc_task *task, __be32 *p)
 static int
 unx_refresh(struct rpc_task *task)
 {
-	task->tk_msg.rpc_cred->cr_flags |= RPCAUTH_CRED_UPTODATE;
+	set_bit(RPCAUTH_CRED_UPTODATE, &task->tk_msg.rpc_cred->cr_flags);
 	return 0;
 }
 
@@ -198,13 +204,18 @@ unx_validate(struct rpc_task *task, __be32 *p)
 		printk("RPC: giant verf size: %u\n", size);
 		return NULL;
 	}
-	task->tk_auth->au_rslack = (size >> 2) + 2;
+	task->tk_msg.rpc_cred->cr_auth->au_rslack = (size >> 2) + 2;
 	p += (size >> 2);
 
 	return p;
 }
 
-struct rpc_authops	authunix_ops = {
+void __init rpc_init_authunix(void)
+{
+	spin_lock_init(&unix_cred_cache.lock);
+}
+
+const struct rpc_authops authunix_ops = {
 	.owner		= THIS_MODULE,
 	.au_flavor	= RPC_AUTH_UNIX,
 #ifdef RPC_DEBUG
@@ -218,7 +229,6 @@ struct rpc_authops	authunix_ops = {
 
 static
 struct rpc_cred_cache	unix_cred_cache = {
-	.expire		= UNX_CRED_EXPIRE,
 };
 
 static
@@ -232,7 +242,7 @@ struct rpc_auth		unix_auth = {
 };
 
 static
-struct rpc_credops	unix_credops = {
+const struct rpc_credops unix_credops = {
 	.cr_name	= "AUTH_UNIX",
 	.crdestroy	= unx_destroy_cred,
 	.crmatch	= unx_match,

net/sunrpc/clnt.c

@@ -44,6 +44,12 @@
 	dprintk("RPC: %5u %s (status %d)\n", t->tk_pid,		\
 			__FUNCTION__, t->tk_status)
 
+/*
+ * All RPC clients are linked into this list
+ */
+static LIST_HEAD(all_clients);
+static DEFINE_SPINLOCK(rpc_client_lock);
+
 static DECLARE_WAIT_QUEUE_HEAD(destroy_wait);
 
 
@@ -66,6 +72,21 @@ static void	call_connect_status(struct rpc_task *task);
 static __be32 *	call_header(struct rpc_task *task);
 static __be32 *	call_verify(struct rpc_task *task);
 
+static int	rpc_ping(struct rpc_clnt *clnt, int flags);
+
+static void rpc_register_client(struct rpc_clnt *clnt)
+{
+	spin_lock(&rpc_client_lock);
+	list_add(&clnt->cl_clients, &all_clients);
+	spin_unlock(&rpc_client_lock);
+}
+
+static void rpc_unregister_client(struct rpc_clnt *clnt)
+{
+	spin_lock(&rpc_client_lock);
+	list_del(&clnt->cl_clients);
+	spin_unlock(&rpc_client_lock);
+}
 
 static int
 rpc_setup_pipedir(struct rpc_clnt *clnt, char *dir_name)
@@ -111,6 +132,9 @@ static struct rpc_clnt * rpc_new_client(struct rpc_xprt *xprt, char *servname, s
 	dprintk("RPC:       creating %s client for %s (xprt %p)\n",
 			program->name, servname, xprt);
 
+	err = rpciod_up();
+	if (err)
+		goto out_no_rpciod;
 	err = -EINVAL;
 	if (!xprt)
 		goto out_no_xprt;
@@ -121,8 +145,6 @@ static struct rpc_clnt * rpc_new_client(struct rpc_xprt *xprt, char *servname, s
 	clnt = kzalloc(sizeof(*clnt), GFP_KERNEL);
 	if (!clnt)
 		goto out_err;
-	atomic_set(&clnt->cl_users, 0);
-	atomic_set(&clnt->cl_count, 1);
 	clnt->cl_parent = clnt;
 
 	clnt->cl_server = clnt->cl_inline_name;
@@ -148,6 +170,8 @@ static struct rpc_clnt * rpc_new_client(struct rpc_xprt *xprt, char *servname, s
 	if (clnt->cl_metrics == NULL)
 		goto out_no_stats;
 	clnt->cl_program  = program;
+	INIT_LIST_HEAD(&clnt->cl_tasks);
+	spin_lock_init(&clnt->cl_lock);
 
 	if (!xprt_bound(clnt->cl_xprt))
 		clnt->cl_autobind = 1;
@@ -155,6 +179,8 @@ static struct rpc_clnt * rpc_new_client(struct rpc_xprt *xprt, char *servname, s
 	clnt->cl_rtt = &clnt->cl_rtt_default;
 	rpc_init_rtt(&clnt->cl_rtt_default, xprt->timeout.to_initval);
 
+	kref_init(&clnt->cl_kref);
+
 	err = rpc_setup_pipedir(clnt, program->pipe_dir_name);
 	if (err < 0)
 		goto out_no_path;
@@ -172,6 +198,7 @@ static struct rpc_clnt * rpc_new_client(struct rpc_xprt *xprt, char *servname, s
 	if (clnt->cl_nodelen > UNX_MAXNODENAME)
 		clnt->cl_nodelen = UNX_MAXNODENAME;
 	memcpy(clnt->cl_nodename, utsname()->nodename, clnt->cl_nodelen);
+	rpc_register_client(clnt);
 	return clnt;
 
 out_no_auth:
@@ -188,6 +215,8 @@ out_no_stats:
 out_err:
 	xprt_put(xprt);
 out_no_xprt:
+	rpciod_down();
+out_no_rpciod:
 	return ERR_PTR(err);
 }
 
@@ -205,12 +234,31 @@ struct rpc_clnt *rpc_create(struct rpc_create_args *args)
 {
 	struct rpc_xprt *xprt;
 	struct rpc_clnt *clnt;
+	struct rpc_xprtsock_create xprtargs = {
+		.proto = args->protocol,
+		.srcaddr = args->saddress,
+		.dstaddr = args->address,
+		.addrlen = args->addrsize,
+		.timeout = args->timeout
+	};
+	char servername[20];
 
-	xprt = xprt_create_transport(args->protocol, args->address,
-					args->addrsize, args->timeout);
+	xprt = xprt_create_transport(&xprtargs);
 	if (IS_ERR(xprt))
 		return (struct rpc_clnt *)xprt;
 
+	/*
+	 * If the caller chooses not to specify a hostname, whip
+	 * up a string representation of the passed-in address.
+	 */
+	if (args->servername == NULL) {
+		struct sockaddr_in *addr =
+					(struct sockaddr_in *) &args->address;
+		snprintf(servername, sizeof(servername), NIPQUAD_FMT,
+			NIPQUAD(addr->sin_addr.s_addr));
+		args->servername = servername;
+	}
+
 	/*
 	 * By default, kernel RPC client connects from a reserved port.
 	 * CAP_NET_BIND_SERVICE will not be set for unprivileged requesters,
@@ -245,8 +293,6 @@ struct rpc_clnt *rpc_create(struct rpc_create_args *args)
 		clnt->cl_intr = 1;
 	if (args->flags & RPC_CLNT_CREATE_AUTOBIND)
 		clnt->cl_autobind = 1;
-	if (args->flags & RPC_CLNT_CREATE_ONESHOT)
-		clnt->cl_oneshot = 1;
 	if (args->flags & RPC_CLNT_CREATE_DISCRTRY)
 		clnt->cl_discrtry = 1;
 
@@ -268,24 +314,25 @@ rpc_clone_client(struct rpc_clnt *clnt)
 	new = kmemdup(clnt, sizeof(*new), GFP_KERNEL);
 	if (!new)
 		goto out_no_clnt;
-	atomic_set(&new->cl_count, 1);
-	atomic_set(&new->cl_users, 0);
+	new->cl_parent = clnt;
+	/* Turn off autobind on clones */
+	new->cl_autobind = 0;
+	INIT_LIST_HEAD(&new->cl_tasks);
+	spin_lock_init(&new->cl_lock);
+	rpc_init_rtt(&new->cl_rtt_default, clnt->cl_xprt->timeout.to_initval);
 	new->cl_metrics = rpc_alloc_iostats(clnt);
 	if (new->cl_metrics == NULL)
 		goto out_no_stats;
+	kref_init(&new->cl_kref);
 	err = rpc_setup_pipedir(new, clnt->cl_program->pipe_dir_name);
 	if (err != 0)
 		goto out_no_path;
-	new->cl_parent = clnt;
-	atomic_inc(&clnt->cl_count);
-	new->cl_xprt = xprt_get(clnt->cl_xprt);
-	/* Turn off autobind on clones */
-	new->cl_autobind = 0;
-	new->cl_oneshot = 0;
-	new->cl_dead = 0;
-	rpc_init_rtt(&new->cl_rtt_default, clnt->cl_xprt->timeout.to_initval);
 	if (new->cl_auth)
 		atomic_inc(&new->cl_auth->au_count);
+	xprt_get(clnt->cl_xprt);
+	kref_get(&clnt->cl_kref);
+	rpc_register_client(new);
+	rpciod_up();
 	return new;
 out_no_path:
 	rpc_free_iostats(new->cl_metrics);
@@ -298,86 +345,86 @@ out_no_clnt:
 
 /*
  * Properly shut down an RPC client, terminating all outstanding
- * requests. Note that we must be certain that cl_oneshot and
- * cl_dead are cleared, or else the client would be destroyed
- * when the last task releases it.
+ * requests.
  */
-int
-rpc_shutdown_client(struct rpc_clnt *clnt)
+void rpc_shutdown_client(struct rpc_clnt *clnt)
 {
-	dprintk("RPC:       shutting down %s client for %s, tasks=%d\n",
-			clnt->cl_protname, clnt->cl_server,
-			atomic_read(&clnt->cl_users));
-
-	while (atomic_read(&clnt->cl_users) > 0) {
-		/* Don't let rpc_release_client destroy us */
-		clnt->cl_oneshot = 0;
-		clnt->cl_dead = 0;
+	dprintk("RPC:       shutting down %s client for %s\n",
+			clnt->cl_protname, clnt->cl_server);
+
+	while (!list_empty(&clnt->cl_tasks)) {
 		rpc_killall_tasks(clnt);
 		wait_event_timeout(destroy_wait,
-			!atomic_read(&clnt->cl_users), 1*HZ);
-	}
-
-	if (atomic_read(&clnt->cl_users) < 0) {
-		printk(KERN_ERR "RPC: rpc_shutdown_client clnt %p tasks=%d\n",
-				clnt, atomic_read(&clnt->cl_users));
-#ifdef RPC_DEBUG
-		rpc_show_tasks();
-#endif
-		BUG();
+			list_empty(&clnt->cl_tasks), 1*HZ);
 	}
 
-	return rpc_destroy_client(clnt);
+	rpc_release_client(clnt);
 }
 
 /*
- * Delete an RPC client
+ * Free an RPC client
  */
-int
-rpc_destroy_client(struct rpc_clnt *clnt)
+static void
+rpc_free_client(struct kref *kref)
 {
-	if (!atomic_dec_and_test(&clnt->cl_count))
-		return 1;
-	BUG_ON(atomic_read(&clnt->cl_users) != 0);
+	struct rpc_clnt *clnt = container_of(kref, struct rpc_clnt, cl_kref);
 
 	dprintk("RPC:       destroying %s client for %s\n",
 			clnt->cl_protname, clnt->cl_server);
-	if (clnt->cl_auth) {
-		rpcauth_destroy(clnt->cl_auth);
-		clnt->cl_auth = NULL;
-	}
 	if (!IS_ERR(clnt->cl_dentry)) {
 		rpc_rmdir(clnt->cl_dentry);
 		rpc_put_mount();
 	}
 	if (clnt->cl_parent != clnt) {
-		rpc_destroy_client(clnt->cl_parent);
+		rpc_release_client(clnt->cl_parent);
 		goto out_free;
 	}
 	if (clnt->cl_server != clnt->cl_inline_name)
 		kfree(clnt->cl_server);
 out_free:
+	rpc_unregister_client(clnt);
 	rpc_free_iostats(clnt->cl_metrics);
 	clnt->cl_metrics = NULL;
 	xprt_put(clnt->cl_xprt);
+	rpciod_down();
 	kfree(clnt);
-	return 0;
 }
 
 /*
- * Release an RPC client
+ * Free an RPC client
+ */
+static void
+rpc_free_auth(struct kref *kref)
+{
+	struct rpc_clnt *clnt = container_of(kref, struct rpc_clnt, cl_kref);
+
+	if (clnt->cl_auth == NULL) {
+		rpc_free_client(kref);
+		return;
+	}
+
+	/*
+	 * Note: RPCSEC_GSS may need to send NULL RPC calls in order to
+	 *       release remaining GSS contexts. This mechanism ensures
+	 *       that it can do so safely.
+	 */
+	kref_init(kref);
+	rpcauth_release(clnt->cl_auth);
+	clnt->cl_auth = NULL;
+	kref_put(kref, rpc_free_client);
+}
+
+/*
+ * Release reference to the RPC client
  */
 void
 rpc_release_client(struct rpc_clnt *clnt)
 {
-	dprintk("RPC:       rpc_release_client(%p, %d)\n",
-			clnt, atomic_read(&clnt->cl_users));
+	dprintk("RPC:       rpc_release_client(%p)\n", clnt);
 
-	if (!atomic_dec_and_test(&clnt->cl_users))
-		return;
-	wake_up(&destroy_wait);
-	if (clnt->cl_oneshot || clnt->cl_dead)
-		rpc_destroy_client(clnt);
+	if (list_empty(&clnt->cl_tasks))
+		wake_up(&destroy_wait);
+	kref_put(&clnt->cl_kref, rpc_free_auth);
 }
 
 /**
@@ -468,82 +515,96 @@ void rpc_clnt_sigunmask(struct rpc_clnt *clnt, sigset_t *oldset)
 	rpc_restore_sigmask(oldset);
 }
 
-/*
- * New rpc_call implementation
+static
+struct rpc_task *rpc_do_run_task(struct rpc_clnt *clnt,
+		struct rpc_message *msg,
+		int flags,
+		const struct rpc_call_ops *ops,
+		void *data)
+{
+	struct rpc_task *task, *ret;
+	sigset_t oldset;
+
+	task = rpc_new_task(clnt, flags, ops, data);
+	if (task == NULL) {
+		rpc_release_calldata(ops, data);
+		return ERR_PTR(-ENOMEM);
+	}
+
+	/* Mask signals on synchronous RPC calls and RPCSEC_GSS upcalls */
+	rpc_task_sigmask(task, &oldset);
+	if (msg != NULL) {
+		rpc_call_setup(task, msg, 0);
+		if (task->tk_status != 0) {
+			ret = ERR_PTR(task->tk_status);
+			rpc_put_task(task);
+			goto out;
+		}
+	}
+	atomic_inc(&task->tk_count);
+	rpc_execute(task);
+	ret = task;
+out:
+	rpc_restore_sigmask(&oldset);
+	return ret;
+}
+
+/**
+ * rpc_call_sync - Perform a synchronous RPC call
+ * @clnt: pointer to RPC client
+ * @msg: RPC call parameters
+ * @flags: RPC call flags
  */
 int rpc_call_sync(struct rpc_clnt *clnt, struct rpc_message *msg, int flags)
 {
 	struct rpc_task	*task;
-	sigset_t	oldset;
-	int		status;
-
-	/* If this client is slain all further I/O fails */
-	if (clnt->cl_dead)
-		return -EIO;
+	int status;
 
 	BUG_ON(flags & RPC_TASK_ASYNC);
 
-	task = rpc_new_task(clnt, flags, &rpc_default_ops, NULL);
-	if (task == NULL)
-		return -ENOMEM;
-
-	/* Mask signals on RPC calls _and_ GSS_AUTH upcalls */
-	rpc_task_sigmask(task, &oldset);
-
-	/* Set up the call info struct and execute the task */
-	rpc_call_setup(task, msg, 0);
-	if (task->tk_status == 0) {
-		atomic_inc(&task->tk_count);
-		rpc_execute(task);
-	}
+	task = rpc_do_run_task(clnt, msg, flags, &rpc_default_ops, NULL);
+	if (IS_ERR(task))
+		return PTR_ERR(task);
 	status = task->tk_status;
 	rpc_put_task(task);
-	rpc_restore_sigmask(&oldset);
 	return status;
 }
 
-/*
- * New rpc_call implementation
+/**
+ * rpc_call_async - Perform an asynchronous RPC call
+ * @clnt: pointer to RPC client
+ * @msg: RPC call parameters
+ * @flags: RPC call flags
+ * @ops: RPC call ops
+ * @data: user call data
  */
 int
 rpc_call_async(struct rpc_clnt *clnt, struct rpc_message *msg, int flags,
 	       const struct rpc_call_ops *tk_ops, void *data)
 {
 	struct rpc_task	*task;
-	sigset_t	oldset;
-	int		status;
-
-	/* If this client is slain all further I/O fails */
-	status = -EIO;
-	if (clnt->cl_dead)
-		goto out_release;
-
-	flags |= RPC_TASK_ASYNC;
-
-	/* Create/initialize a new RPC task */
-	status = -ENOMEM;
-	if (!(task = rpc_new_task(clnt, flags, tk_ops, data)))
-		goto out_release;
-
-	/* Mask signals on GSS_AUTH upcalls */
-	rpc_task_sigmask(task, &oldset);
 
-	rpc_call_setup(task, msg, 0);
-
-	/* Set up the call info struct and execute the task */
-	status = task->tk_status;
-	if (status == 0)
-		rpc_execute(task);
-	else
-		rpc_put_task(task);
-
-	rpc_restore_sigmask(&oldset);
-	return status;
-out_release:
-	rpc_release_calldata(tk_ops, data);
-	return status;
+	task = rpc_do_run_task(clnt, msg, flags|RPC_TASK_ASYNC, tk_ops, data);
+	if (IS_ERR(task))
+		return PTR_ERR(task);
+	rpc_put_task(task);
+	return 0;
 }
 
+/**
+ * rpc_run_task - Allocate a new RPC task, then run rpc_execute against it
+ * @clnt: pointer to RPC client
+ * @flags: RPC flags
+ * @ops: RPC call ops
+ * @data: user call data
+ */
+struct rpc_task *rpc_run_task(struct rpc_clnt *clnt, int flags,
+					const struct rpc_call_ops *tk_ops,
+					void *data)
+{
+	return rpc_do_run_task(clnt, NULL, flags, tk_ops, data);
+}
+EXPORT_SYMBOL(rpc_run_task);
 
 void
 rpc_call_setup(struct rpc_task *task, struct rpc_message *msg, int flags)
@@ -745,7 +806,7 @@ call_reserveresult(struct rpc_task *task)
 static void
 call_allocate(struct rpc_task *task)
 {
-	unsigned int slack = task->tk_auth->au_cslack;
+	unsigned int slack = task->tk_msg.rpc_cred->cr_auth->au_cslack;
 	struct rpc_rqst *req = task->tk_rqstp;
 	struct rpc_xprt *xprt = task->tk_xprt;
 	struct rpc_procinfo *proc = task->tk_msg.rpc_proc;
@@ -843,10 +904,8 @@ call_encode(struct rpc_task *task)
 	if (encode == NULL)
 		return;
 
-	lock_kernel();
 	task->tk_status = rpcauth_wrap_req(task, encode, req, p,
 			task->tk_msg.rpc_argp);
-	unlock_kernel();
 	if (task->tk_status == -ENOMEM) {
 		/* XXX: Is this sane? */
 		rpc_delay(task, 3*HZ);
@@ -1177,10 +1236,8 @@ call_decode(struct rpc_task *task)
 	task->tk_action = rpc_exit_task;
 
 	if (decode) {
-		lock_kernel();
 		task->tk_status = rpcauth_unwrap_resp(task, decode, req, p,
 						      task->tk_msg.rpc_resp);
-		unlock_kernel();
 	}
 	dprintk("RPC: %5u call_decode result %d\n", task->tk_pid,
 			task->tk_status);
@@ -1273,9 +1330,9 @@ call_verify(struct rpc_task *task)
 		 * - if it isn't pointer subtraction in the NFS client may give
 		 *   undefined results
 		 */
-		printk(KERN_WARNING
-		       "call_verify: XDR representation not a multiple of"
-		       " 4 bytes: 0x%x\n", task->tk_rqstp->rq_rcv_buf.len);
+		dprintk("RPC: %5u %s: XDR representation not a multiple of"
+		       " 4 bytes: 0x%x\n", task->tk_pid, __FUNCTION__,
+		       task->tk_rqstp->rq_rcv_buf.len);
 		goto out_eio;
 	}
 	if ((len -= 3) < 0)
@@ -1283,7 +1340,8 @@ call_verify(struct rpc_task *task)
 	p += 1;	/* skip XID */
 
 	if ((n = ntohl(*p++)) != RPC_REPLY) {
-		printk(KERN_WARNING "call_verify: not an RPC reply: %x\n", n);
+		dprintk("RPC: %5u %s: not an RPC reply: %x\n",
+				task->tk_pid, __FUNCTION__, n);
 		goto out_garbage;
 	}
 	if ((n = ntohl(*p++)) != RPC_MSG_ACCEPTED) {
@@ -1334,7 +1392,8 @@ call_verify(struct rpc_task *task)
 			       "authentication.\n", task->tk_client->cl_server);
 			break;
 		default:
-			printk(KERN_WARNING "call_verify: unknown auth error: %x\n", n);
+			dprintk("RPC: %5u %s: unknown auth error: %x\n",
+					task->tk_pid, __FUNCTION__, n);
 			error = -EIO;
 		}
 		dprintk("RPC: %5u %s: call rejected %d\n",
@@ -1342,7 +1401,8 @@ call_verify(struct rpc_task *task)
 		goto out_err;
 	}
 	if (!(p = rpcauth_checkverf(task, p))) {
-		printk(KERN_WARNING "call_verify: auth check failed\n");
+		dprintk("RPC: %5u %s: auth check failed\n",
+				task->tk_pid, __FUNCTION__);
 		goto out_garbage;		/* bad verifier, retry */
 	}
 	len = p - (__be32 *)iov->iov_base - 1;
@@ -1381,7 +1441,8 @@ call_verify(struct rpc_task *task)
 				task->tk_pid, __FUNCTION__);
 		break;			/* retry */
 	default:
-		printk(KERN_WARNING "call_verify: server accept status: %x\n", n);
+		dprintk("RPC: %5u %s: server accept status: %x\n",
+				task->tk_pid, __FUNCTION__, n);
 		/* Also retry */
 	}
 
@@ -1395,14 +1456,16 @@ out_garbage:
 out_retry:
 		return ERR_PTR(-EAGAIN);
 	}
-	printk(KERN_WARNING "RPC %s: retry failed, exit EIO\n", __FUNCTION__);
 out_eio:
 	error = -EIO;
 out_err:
 	rpc_exit(task, error);
+	dprintk("RPC: %5u %s: call failed with error %d\n", task->tk_pid,
+			__FUNCTION__, error);
 	return ERR_PTR(error);
 out_overflow:
-	printk(KERN_WARNING "RPC %s: server reply was truncated.\n", __FUNCTION__);
+	dprintk("RPC: %5u %s: server reply was truncated.\n", task->tk_pid,
+			__FUNCTION__);
 	goto out_garbage;
 }
 
@@ -1421,7 +1484,7 @@ static struct rpc_procinfo rpcproc_null = {
 	.p_decode = rpcproc_decode_null,
 };
 
-int rpc_ping(struct rpc_clnt *clnt, int flags)
+static int rpc_ping(struct rpc_clnt *clnt, int flags)
 {
 	struct rpc_message msg = {
 		.rpc_proc = &rpcproc_null,
@@ -1432,3 +1495,51 @@ int rpc_ping(struct rpc_clnt *clnt, int flags)
 	put_rpccred(msg.rpc_cred);
 	return err;
 }
+
+struct rpc_task *rpc_call_null(struct rpc_clnt *clnt, struct rpc_cred *cred, int flags)
+{
+	struct rpc_message msg = {
+		.rpc_proc = &rpcproc_null,
+		.rpc_cred = cred,
+	};
+	return rpc_do_run_task(clnt, &msg, flags, &rpc_default_ops, NULL);
+}
+EXPORT_SYMBOL(rpc_call_null);
+
+#ifdef RPC_DEBUG
+void rpc_show_tasks(void)
+{
+	struct rpc_clnt *clnt;
+	struct rpc_task *t;
+
+	spin_lock(&rpc_client_lock);
+	if (list_empty(&all_clients))
+		goto out;
+	printk("-pid- proc flgs status -client- -prog- --rqstp- -timeout "
+		"-rpcwait -action- ---ops--\n");
+	list_for_each_entry(clnt, &all_clients, cl_clients) {
+		if (list_empty(&clnt->cl_tasks))
+			continue;
+		spin_lock(&clnt->cl_lock);
+		list_for_each_entry(t, &clnt->cl_tasks, tk_task) {
+			const char *rpc_waitq = "none";
+
+			if (RPC_IS_QUEUED(t))
+				rpc_waitq = rpc_qname(t->u.tk_wait.rpc_waitq);
+
+			printk("%5u %04d %04x %6d %8p %6d %8p %8ld %8s %8p %8p\n",
+				t->tk_pid,
+				(t->tk_msg.rpc_proc ? t->tk_msg.rpc_proc->p_proc : -1),
+				t->tk_flags, t->tk_status,
+				t->tk_client,
+				(t->tk_client ? t->tk_client->cl_prog : 0),
+				t->tk_rqstp, t->tk_timeout,
+				rpc_waitq,
+				t->tk_action, t->tk_ops);
+		}
+		spin_unlock(&clnt->cl_lock);
+	}
+out:
+	spin_unlock(&rpc_client_lock);
+}
+#endif

net/sunrpc/rpc_pipe.c

@@ -344,7 +344,7 @@ rpc_info_open(struct inode *inode, struct file *file)
 		mutex_lock(&inode->i_mutex);
 		clnt = RPC_I(inode)->private;
 		if (clnt) {
-			atomic_inc(&clnt->cl_users);
+			kref_get(&clnt->cl_kref);
 			m->private = clnt;
 		} else {
 			single_release(inode, file);
@@ -448,6 +448,15 @@ void rpc_put_mount(void)
 	simple_release_fs(&rpc_mount, &rpc_mount_count);
 }
 
+static int rpc_delete_dentry(struct dentry *dentry)
+{
+	return 1;
+}
+
+static struct dentry_operations rpc_dentry_operations = {
+	.d_delete = rpc_delete_dentry,
+};
+
 static int
 rpc_lookup_parent(char *path, struct nameidata *nd)
 {
@@ -506,7 +515,7 @@ rpc_get_inode(struct super_block *sb, int mode)
  * FIXME: This probably has races.
  */
 static void
-rpc_depopulate(struct dentry *parent)
+rpc_depopulate(struct dentry *parent, int start, int eof)
 {
 	struct inode *dir = parent->d_inode;
 	struct list_head *pos, *next;
@@ -518,6 +527,10 @@ repeat:
 	spin_lock(&dcache_lock);
 	list_for_each_safe(pos, next, &parent->d_subdirs) {
 		dentry = list_entry(pos, struct dentry, d_u.d_child);
+		if (!dentry->d_inode ||
+				dentry->d_inode->i_ino < start ||
+				dentry->d_inode->i_ino >= eof)
+			continue;
 		spin_lock(&dentry->d_lock);
 		if (!d_unhashed(dentry)) {
 			dget_locked(dentry);
@@ -533,11 +546,11 @@ repeat:
 	if (n) {
 		do {
 			dentry = dvec[--n];
-			if (dentry->d_inode) {
-				rpc_close_pipes(dentry->d_inode);
+			if (S_ISREG(dentry->d_inode->i_mode))
 				simple_unlink(dir, dentry);
-			}
-			inode_dir_notify(dir, DN_DELETE);
+			else if (S_ISDIR(dentry->d_inode->i_mode))
+				simple_rmdir(dir, dentry);
+			d_delete(dentry);
 			dput(dentry);
 		} while (n);
 		goto repeat;
@@ -560,6 +573,7 @@ rpc_populate(struct dentry *parent,
 		dentry = d_alloc_name(parent, files[i].name);
 		if (!dentry)
 			goto out_bad;
+		dentry->d_op = &rpc_dentry_operations;
 		mode = files[i].mode;
 		inode = rpc_get_inode(dir->i_sb, mode);
 		if (!inode) {
@@ -607,21 +621,14 @@ static int
 __rpc_rmdir(struct inode *dir, struct dentry *dentry)
 {
 	int error;
-
-	shrink_dcache_parent(dentry);
-	if (d_unhashed(dentry))
-		return 0;
-	if ((error = simple_rmdir(dir, dentry)) != 0)
-		return error;
-	if (!error) {
-		inode_dir_notify(dir, DN_DELETE);
-		d_drop(dentry);
-	}
-	return 0;
+	error = simple_rmdir(dir, dentry);
+	if (!error)
+		d_delete(dentry);
+	return error;
 }
 
 static struct dentry *
-rpc_lookup_create(struct dentry *parent, const char *name, int len)
+rpc_lookup_create(struct dentry *parent, const char *name, int len, int exclusive)
 {
 	struct inode *dir = parent->d_inode;
 	struct dentry *dentry;
@@ -630,7 +637,9 @@ rpc_lookup_create(struct dentry *parent, const char *name, int len)
 	dentry = lookup_one_len(name, parent, len);
 	if (IS_ERR(dentry))
 		goto out_err;
-	if (dentry->d_inode) {
+	if (!dentry->d_inode)
+		dentry->d_op = &rpc_dentry_operations;
+	else if (exclusive) {
 		dput(dentry);
 		dentry = ERR_PTR(-EEXIST);
 		goto out_err;
@@ -649,7 +658,7 @@ rpc_lookup_negative(char *path, struct nameidata *nd)
 
 	if ((error = rpc_lookup_parent(path, nd)) != 0)
 		return ERR_PTR(error);
-	dentry = rpc_lookup_create(nd->dentry, nd->last.name, nd->last.len);
+	dentry = rpc_lookup_create(nd->dentry, nd->last.name, nd->last.len, 1);
 	if (IS_ERR(dentry))
 		rpc_release_path(nd);
 	return dentry;
@@ -681,7 +690,7 @@ out:
 	rpc_release_path(&nd);
 	return dentry;
 err_depopulate:
-	rpc_depopulate(dentry);
+	rpc_depopulate(dentry, RPCAUTH_info, RPCAUTH_EOF);
 	__rpc_rmdir(dir, dentry);
 err_dput:
 	dput(dentry);
@@ -701,7 +710,7 @@ rpc_rmdir(struct dentry *dentry)
 	parent = dget_parent(dentry);
 	dir = parent->d_inode;
 	mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
-	rpc_depopulate(dentry);
+	rpc_depopulate(dentry, RPCAUTH_info, RPCAUTH_EOF);
 	error = __rpc_rmdir(dir, dentry);
 	dput(dentry);
 	mutex_unlock(&dir->i_mutex);
@@ -716,10 +725,21 @@ rpc_mkpipe(struct dentry *parent, const char *name, void *private, struct rpc_pi
 	struct inode *dir, *inode;
 	struct rpc_inode *rpci;
 
-	dentry = rpc_lookup_create(parent, name, strlen(name));
+	dentry = rpc_lookup_create(parent, name, strlen(name), 0);
 	if (IS_ERR(dentry))
 		return dentry;
 	dir = parent->d_inode;
+	if (dentry->d_inode) {
+		rpci = RPC_I(dentry->d_inode);
+		if (rpci->private != private ||
+				rpci->ops != ops ||
+				rpci->flags != flags) {
+			dput (dentry);
+			dentry = ERR_PTR(-EBUSY);
+		}
+		rpci->nkern_readwriters++;
+		goto out;
+	}
 	inode = rpc_get_inode(dir->i_sb, S_IFIFO | S_IRUSR | S_IWUSR);
 	if (!inode)
 		goto err_dput;
@@ -730,6 +750,7 @@ rpc_mkpipe(struct dentry *parent, const char *name, void *private, struct rpc_pi
 	rpci->private = private;
 	rpci->flags = flags;
 	rpci->ops = ops;
+	rpci->nkern_readwriters = 1;
 	inode_dir_notify(dir, DN_CREATE);
 	dget(dentry);
 out:
@@ -754,13 +775,11 @@ rpc_unlink(struct dentry *dentry)
 	parent = dget_parent(dentry);
 	dir = parent->d_inode;
 	mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
-	if (!d_unhashed(dentry)) {
-		d_drop(dentry);
-		if (dentry->d_inode) {
-			rpc_close_pipes(dentry->d_inode);
-			error = simple_unlink(dir, dentry);
-		}
-		inode_dir_notify(dir, DN_DELETE);
+	if (--RPC_I(dentry->d_inode)->nkern_readwriters == 0) {
+		rpc_close_pipes(dentry->d_inode);
+		error = simple_unlink(dir, dentry);
+		if (!error)
+			d_delete(dentry);
 	}
 	dput(dentry);
 	mutex_unlock(&dir->i_mutex);
@@ -833,6 +852,7 @@ init_once(void * foo, struct kmem_cache * cachep, unsigned long flags)
 	rpci->nreaders = 0;
 	rpci->nwriters = 0;
 	INIT_LIST_HEAD(&rpci->in_upcall);
+	INIT_LIST_HEAD(&rpci->in_downcall);
 	INIT_LIST_HEAD(&rpci->pipe);
 	rpci->pipelen = 0;
 	init_waitqueue_head(&rpci->waitq);

net/sunrpc/rpcb_clnt.c

@@ -12,6 +12,8 @@
  *  Copyright (C) 1996, Olaf Kirch <okir@monad.swb.de>
  */
 
+#include <linux/module.h>
+
 #include <linux/types.h>
 #include <linux/socket.h>
 #include <linux/kernel.h>
@@ -184,8 +186,8 @@ static struct rpc_clnt *rpcb_create(char *hostname, struct sockaddr *srvaddr,
 		.program	= &rpcb_program,
 		.version	= version,
 		.authflavor	= RPC_AUTH_UNIX,
-		.flags		= (RPC_CLNT_CREATE_ONESHOT |
-				   RPC_CLNT_CREATE_NOPING),
+		.flags		= (RPC_CLNT_CREATE_NOPING |
+				   RPC_CLNT_CREATE_INTR),
 	};
 
 	((struct sockaddr_in *)srvaddr)->sin_port = htons(RPCBIND_PORT);
@@ -238,6 +240,7 @@ int rpcb_register(u32 prog, u32 vers, int prot, unsigned short port, int *okay)
 
 	error = rpc_call_sync(rpcb_clnt, &msg, 0);
 
+	rpc_shutdown_client(rpcb_clnt);
 	if (error < 0)
 		printk(KERN_WARNING "RPC: failed to contact local rpcbind "
 				"server (errno %d).\n", -error);
@@ -246,21 +249,20 @@ int rpcb_register(u32 prog, u32 vers, int prot, unsigned short port, int *okay)
 	return error;
 }
 
-#ifdef CONFIG_ROOT_NFS
 /**
- * rpcb_getport_external - obtain the port for an RPC service on a given host
+ * rpcb_getport_sync - obtain the port for an RPC service on a given host
  * @sin: address of remote peer
  * @prog: RPC program number to bind
  * @vers: RPC version number to bind
  * @prot: transport protocol to use to make this request
  *
  * Called from outside the RPC client in a synchronous task context.
+ * Uses default timeout parameters specified by underlying transport.
  *
- * For now, this supports only version 2 queries, but is used only by
- * mount_clnt for NFS_ROOT.
+ * XXX: Needs to support IPv6, and rpcbind versions 3 and 4
  */
-int rpcb_getport_external(struct sockaddr_in *sin, __u32 prog,
-				__u32 vers, int prot)
+int rpcb_getport_sync(struct sockaddr_in *sin, __u32 prog,
+		      __u32 vers, int prot)
 {
 	struct rpcbind_args map = {
 		.r_prog		= prog,
@@ -277,15 +279,16 @@ int rpcb_getport_external(struct sockaddr_in *sin, __u32 prog,
 	char hostname[40];
 	int status;
 
-	dprintk("RPC:       rpcb_getport_external(%u.%u.%u.%u, %u, %u, %d)\n",
-			NIPQUAD(sin->sin_addr.s_addr), prog, vers, prot);
+	dprintk("RPC:       %s(" NIPQUAD_FMT ", %u, %u, %d)\n",
+		__FUNCTION__, NIPQUAD(sin->sin_addr.s_addr), prog, vers, prot);
 
-	sprintf(hostname, "%u.%u.%u.%u", NIPQUAD(sin->sin_addr.s_addr));
+	sprintf(hostname, NIPQUAD_FMT, NIPQUAD(sin->sin_addr.s_addr));
 	rpcb_clnt = rpcb_create(hostname, (struct sockaddr *)sin, prot, 2, 0);
 	if (IS_ERR(rpcb_clnt))
 		return PTR_ERR(rpcb_clnt);
 
 	status = rpc_call_sync(rpcb_clnt, &msg, 0);
+	rpc_shutdown_client(rpcb_clnt);
 
 	if (status >= 0) {
 		if (map.r_port != 0)
@@ -294,16 +297,16 @@ int rpcb_getport_external(struct sockaddr_in *sin, __u32 prog,
 	}
 	return status;
 }
-#endif
+EXPORT_SYMBOL_GPL(rpcb_getport_sync);
 
 /**
- * rpcb_getport - obtain the port for a given RPC service on a given host
+ * rpcb_getport_async - obtain the port for a given RPC service on a given host
  * @task: task that is waiting for portmapper request
  *
  * This one can be called for an ongoing RPC request, and can be used in
  * an async (rpciod) context.
  */
-void rpcb_getport(struct rpc_task *task)
+void rpcb_getport_async(struct rpc_task *task)
 {
 	struct rpc_clnt *clnt = task->tk_client;
 	int bind_version;
@@ -314,17 +317,17 @@ void rpcb_getport(struct rpc_task *task)
 	struct sockaddr addr;
 	int status;
 
-	dprintk("RPC: %5u rpcb_getport(%s, %u, %u, %d)\n",
-			task->tk_pid, clnt->cl_server,
-			clnt->cl_prog, clnt->cl_vers, xprt->prot);
+	dprintk("RPC: %5u %s(%s, %u, %u, %d)\n",
+		task->tk_pid, __FUNCTION__,
+		clnt->cl_server, clnt->cl_prog, clnt->cl_vers, xprt->prot);
 
 	/* Autobind on cloned rpc clients is discouraged */
 	BUG_ON(clnt->cl_parent != clnt);
 
 	if (xprt_test_and_set_binding(xprt)) {
 		status = -EACCES;		/* tell caller to check again */
-		dprintk("RPC: %5u rpcb_getport waiting for another binder\n",
-				task->tk_pid);
+		dprintk("RPC: %5u %s: waiting for another binder\n",
+			task->tk_pid, __FUNCTION__);
 		goto bailout_nowake;
 	}
 
@@ -335,27 +338,28 @@ void rpcb_getport(struct rpc_task *task)
 	/* Someone else may have bound if we slept */
 	if (xprt_bound(xprt)) {
 		status = 0;
-		dprintk("RPC: %5u rpcb_getport already bound\n", task->tk_pid);
+		dprintk("RPC: %5u %s: already bound\n",
+			task->tk_pid, __FUNCTION__);
 		goto bailout_nofree;
 	}
 
 	if (rpcb_next_version[xprt->bind_index].rpc_proc == NULL) {
 		xprt->bind_index = 0;
 		status = -EACCES;	/* tell caller to try again later */
-		dprintk("RPC: %5u rpcb_getport no more getport versions "
-				"available\n", task->tk_pid);
+		dprintk("RPC: %5u %s: no more getport versions available\n",
+			task->tk_pid, __FUNCTION__);
 		goto bailout_nofree;
 	}
 	bind_version = rpcb_next_version[xprt->bind_index].rpc_vers;
 
-	dprintk("RPC: %5u rpcb_getport trying rpcbind version %u\n",
-			task->tk_pid, bind_version);
+	dprintk("RPC: %5u %s: trying rpcbind version %u\n",
+		task->tk_pid, __FUNCTION__, bind_version);
 
 	map = kzalloc(sizeof(struct rpcbind_args), GFP_ATOMIC);
 	if (!map) {
 		status = -ENOMEM;
-		dprintk("RPC: %5u rpcb_getport no memory available\n",
-				task->tk_pid);
+		dprintk("RPC: %5u %s: no memory available\n",
+			task->tk_pid, __FUNCTION__);
 		goto bailout_nofree;
 	}
 	map->r_prog = clnt->cl_prog;
@@ -373,16 +377,17 @@ void rpcb_getport(struct rpc_task *task)
 	rpcb_clnt = rpcb_create(clnt->cl_server, &addr, xprt->prot, bind_version, 0);
 	if (IS_ERR(rpcb_clnt)) {
 		status = PTR_ERR(rpcb_clnt);
-		dprintk("RPC: %5u rpcb_getport rpcb_create failed, error %ld\n",
-				task->tk_pid, PTR_ERR(rpcb_clnt));
+		dprintk("RPC: %5u %s: rpcb_create failed, error %ld\n",
+			task->tk_pid, __FUNCTION__, PTR_ERR(rpcb_clnt));
 		goto bailout;
 	}
 
 	child = rpc_run_task(rpcb_clnt, RPC_TASK_ASYNC, &rpcb_getport_ops, map);
+	rpc_release_client(rpcb_clnt);
 	if (IS_ERR(child)) {
 		status = -EIO;
-		dprintk("RPC: %5u rpcb_getport rpc_run_task failed\n",
-				task->tk_pid);
+		dprintk("RPC: %5u %s: rpc_run_task failed\n",
+			task->tk_pid, __FUNCTION__);
 		goto bailout_nofree;
 	}
 	rpc_put_task(child);

net/sunrpc/sched.c

@@ -25,7 +25,6 @@
 #ifdef RPC_DEBUG
 #define RPCDBG_FACILITY		RPCDBG_SCHED
 #define RPC_TASK_MAGIC_ID	0xf00baa
-static int			rpc_task_id;
 #endif
 
 /*
@@ -40,7 +39,6 @@ static mempool_t	*rpc_task_mempool __read_mostly;
 static mempool_t	*rpc_buffer_mempool __read_mostly;
 
 static void			__rpc_default_timer(struct rpc_task *task);
-static void			rpciod_killall(void);
 static void			rpc_async_schedule(struct work_struct *);
 static void			 rpc_release_task(struct rpc_task *task);
 
@@ -49,23 +47,13 @@ static void			 rpc_release_task(struct rpc_task *task);
  */
 static RPC_WAITQ(delay_queue, "delayq");
 
-/*
- * All RPC tasks are linked into this list
- */
-static LIST_HEAD(all_tasks);
-
 /*
  * rpciod-related stuff
  */
 static DEFINE_MUTEX(rpciod_mutex);
-static unsigned int		rpciod_users;
+static atomic_t rpciod_users = ATOMIC_INIT(0);
 struct workqueue_struct *rpciod_workqueue;
 
-/*
- * Spinlock for other critical sections of code.
- */
-static DEFINE_SPINLOCK(rpc_sched_lock);
-
 /*
  * Disable the timer for a given RPC task. Should be called with
  * queue->lock and bh_disabled in order to avoid races within
@@ -267,18 +255,33 @@ static int rpc_wait_bit_interruptible(void *word)
 	return 0;
 }
 
+#ifdef RPC_DEBUG
+static void rpc_task_set_debuginfo(struct rpc_task *task)
+{
+	static atomic_t rpc_pid;
+
+	task->tk_magic = RPC_TASK_MAGIC_ID;
+	task->tk_pid = atomic_inc_return(&rpc_pid);
+}
+#else
+static inline void rpc_task_set_debuginfo(struct rpc_task *task)
+{
+}
+#endif
+
 static void rpc_set_active(struct rpc_task *task)
 {
+	struct rpc_clnt *clnt;
 	if (test_and_set_bit(RPC_TASK_ACTIVE, &task->tk_runstate) != 0)
 		return;
-	spin_lock(&rpc_sched_lock);
-#ifdef RPC_DEBUG
-	task->tk_magic = RPC_TASK_MAGIC_ID;
-	task->tk_pid = rpc_task_id++;
-#endif
+	rpc_task_set_debuginfo(task);
 	/* Add to global list of all tasks */
-	list_add_tail(&task->tk_task, &all_tasks);
-	spin_unlock(&rpc_sched_lock);
+	clnt = task->tk_client;
+	if (clnt != NULL) {
+		spin_lock(&clnt->cl_lock);
+		list_add_tail(&task->tk_task, &clnt->cl_tasks);
+		spin_unlock(&clnt->cl_lock);
+	}
 }
 
 /*
@@ -818,6 +821,7 @@ void rpc_init_task(struct rpc_task *task, struct rpc_clnt *clnt, int flags, cons
 	if (tk_ops->rpc_call_prepare != NULL)
 		task->tk_action = rpc_prepare_task;
 	task->tk_calldata = calldata;
+	INIT_LIST_HEAD(&task->tk_task);
 
 	/* Initialize retry counters */
 	task->tk_garb_retry = 2;
@@ -830,7 +834,7 @@ void rpc_init_task(struct rpc_task *task, struct rpc_clnt *clnt, int flags, cons
 	task->tk_workqueue = rpciod_workqueue;
 
 	if (clnt) {
-		atomic_inc(&clnt->cl_users);
+		kref_get(&clnt->cl_kref);
 		if (clnt->cl_softrtry)
 			task->tk_flags |= RPC_TASK_SOFT;
 		if (!clnt->cl_intr)
@@ -860,9 +864,7 @@ static void rpc_free_task(struct rcu_head *rcu)
 }
 
 /*
- * Create a new task for the specified client.  We have to
- * clean up after an allocation failure, as the client may
- * have specified "oneshot".
+ * Create a new task for the specified client.
  */
 struct rpc_task *rpc_new_task(struct rpc_clnt *clnt, int flags, const struct rpc_call_ops *tk_ops, void *calldata)
 {
@@ -870,7 +872,7 @@ struct rpc_task *rpc_new_task(struct rpc_clnt *clnt, int flags, const struct rpc
 
 	task = rpc_alloc_task();
 	if (!task)
-		goto cleanup;
+		goto out;
 
 	rpc_init_task(task, clnt, flags, tk_ops, calldata);
 
@@ -878,16 +880,6 @@ struct rpc_task *rpc_new_task(struct rpc_clnt *clnt, int flags, const struct rpc
 	task->tk_flags |= RPC_TASK_DYNAMIC;
 out:
 	return task;
-
-cleanup:
-	/* Check whether to release the client */
-	if (clnt) {
-		printk("rpc_new_task: failed, users=%d, oneshot=%d\n",
-			atomic_read(&clnt->cl_users), clnt->cl_oneshot);
-		atomic_inc(&clnt->cl_users); /* pretend we were used ... */
-		rpc_release_client(clnt);
-	}
-	goto out;
 }
 
 
@@ -920,11 +912,13 @@ static void rpc_release_task(struct rpc_task *task)
 #endif
 	dprintk("RPC: %5u release task\n", task->tk_pid);
 
-	/* Remove from global task list */
-	spin_lock(&rpc_sched_lock);
-	list_del(&task->tk_task);
-	spin_unlock(&rpc_sched_lock);
-
+	if (!list_empty(&task->tk_task)) {
+		struct rpc_clnt *clnt = task->tk_client;
+		/* Remove from client task list */
+		spin_lock(&clnt->cl_lock);
+		list_del(&task->tk_task);
+		spin_unlock(&clnt->cl_lock);
+	}
 	BUG_ON (RPC_IS_QUEUED(task));
 
 	/* Synchronously delete any running timer */
@@ -939,29 +933,6 @@ static void rpc_release_task(struct rpc_task *task)
 	rpc_put_task(task);
 }
 
-/**
- * rpc_run_task - Allocate a new RPC task, then run rpc_execute against it
- * @clnt: pointer to RPC client
- * @flags: RPC flags
- * @ops: RPC call ops
- * @data: user call data
- */
-struct rpc_task *rpc_run_task(struct rpc_clnt *clnt, int flags,
-					const struct rpc_call_ops *ops,
-					void *data)
-{
-	struct rpc_task *task;
-	task = rpc_new_task(clnt, flags, ops, data);
-	if (task == NULL) {
-		rpc_release_calldata(ops, data);
-		return ERR_PTR(-ENOMEM);
-	}
-	atomic_inc(&task->tk_count);
-	rpc_execute(task);
-	return task;
-}
-EXPORT_SYMBOL(rpc_run_task);
-
 /*
  * Kill all tasks for the given client.
  * XXX: kill their descendants as well?
@@ -969,44 +940,25 @@ EXPORT_SYMBOL(rpc_run_task);
 void rpc_killall_tasks(struct rpc_clnt *clnt)
 {
 	struct rpc_task	*rovr;
-	struct list_head *le;
 
-	dprintk("RPC:       killing all tasks for client %p\n", clnt);
 
+	if (list_empty(&clnt->cl_tasks))
+		return;
+	dprintk("RPC:       killing all tasks for client %p\n", clnt);
 	/*
 	 * Spin lock all_tasks to prevent changes...
 	 */
-	spin_lock(&rpc_sched_lock);
-	alltask_for_each(rovr, le, &all_tasks) {
+	spin_lock(&clnt->cl_lock);
+	list_for_each_entry(rovr, &clnt->cl_tasks, tk_task) {
 		if (! RPC_IS_ACTIVATED(rovr))
 			continue;
-		if (!clnt || rovr->tk_client == clnt) {
+		if (!(rovr->tk_flags & RPC_TASK_KILLED)) {
 			rovr->tk_flags |= RPC_TASK_KILLED;
 			rpc_exit(rovr, -EIO);
 			rpc_wake_up_task(rovr);
 		}
 	}
-	spin_unlock(&rpc_sched_lock);
-}
-
-static void rpciod_killall(void)
-{
-	unsigned long flags;
-
-	while (!list_empty(&all_tasks)) {
-		clear_thread_flag(TIF_SIGPENDING);
-		rpc_killall_tasks(NULL);
-		flush_workqueue(rpciod_workqueue);
-		if (!list_empty(&all_tasks)) {
-			dprintk("RPC:       rpciod_killall: waiting for tasks "
-					"to exit\n");
-			yield();
-		}
-	}
-
-	spin_lock_irqsave(&current->sighand->siglock, flags);
-	recalc_sigpending();
-	spin_unlock_irqrestore(&current->sighand->siglock, flags);
+	spin_unlock(&clnt->cl_lock);
 }
 
 /*
@@ -1018,28 +970,27 @@ rpciod_up(void)
 	struct workqueue_struct *wq;
 	int error = 0;
 
+	if (atomic_inc_not_zero(&rpciod_users))
+		return 0;
+
 	mutex_lock(&rpciod_mutex);
-	dprintk("RPC:       rpciod_up: users %u\n", rpciod_users);
-	rpciod_users++;
-	if (rpciod_workqueue)
-		goto out;
-	/*
-	 * If there's no pid, we should be the first user.
-	 */
-	if (rpciod_users > 1)
-		printk(KERN_WARNING "rpciod_up: no workqueue, %u users??\n", rpciod_users);
+
+	/* Guard against races with rpciod_down() */
+	if (rpciod_workqueue != NULL)
+		goto out_ok;
 	/*
 	 * Create the rpciod thread and wait for it to start.
 	 */
+	dprintk("RPC:       creating workqueue rpciod\n");
 	error = -ENOMEM;
 	wq = create_workqueue("rpciod");
-	if (wq == NULL) {
-		printk(KERN_WARNING "rpciod_up: create workqueue failed, error=%d\n", error);
-		rpciod_users--;
+	if (wq == NULL)
 		goto out;
-	}
+
 	rpciod_workqueue = wq;
 	error = 0;
+out_ok:
+	atomic_inc(&rpciod_users);
 out:
 	mutex_unlock(&rpciod_mutex);
 	return error;
@@ -1048,59 +999,19 @@ out:
 void
 rpciod_down(void)
 {
+	if (!atomic_dec_and_test(&rpciod_users))
+		return;
+
 	mutex_lock(&rpciod_mutex);
-	dprintk("RPC:       rpciod_down sema %u\n", rpciod_users);
-	if (rpciod_users) {
-		if (--rpciod_users)
-			goto out;
-	} else
-		printk(KERN_WARNING "rpciod_down: no users??\n");
+	dprintk("RPC:       destroying workqueue rpciod\n");
 
-	if (!rpciod_workqueue) {
-		dprintk("RPC:       rpciod_down: Nothing to do!\n");
-		goto out;
+	if (atomic_read(&rpciod_users) == 0 && rpciod_workqueue != NULL) {
+		destroy_workqueue(rpciod_workqueue);
+		rpciod_workqueue = NULL;
 	}
-	rpciod_killall();
-
-	destroy_workqueue(rpciod_workqueue);
-	rpciod_workqueue = NULL;
- out:
 	mutex_unlock(&rpciod_mutex);
 }
 
-#ifdef RPC_DEBUG
-void rpc_show_tasks(void)
-{
-	struct list_head *le;
-	struct rpc_task *t;
-
-	spin_lock(&rpc_sched_lock);
-	if (list_empty(&all_tasks)) {
-		spin_unlock(&rpc_sched_lock);
-		return;
-	}
-	printk("-pid- proc flgs status -client- -prog- --rqstp- -timeout "
-		"-rpcwait -action- ---ops--\n");
-	alltask_for_each(t, le, &all_tasks) {
-		const char *rpc_waitq = "none";
-
-		if (RPC_IS_QUEUED(t))
-			rpc_waitq = rpc_qname(t->u.tk_wait.rpc_waitq);
-
-		printk("%5u %04d %04x %6d %8p %6d %8p %8ld %8s %8p %8p\n",
-			t->tk_pid,
-			(t->tk_msg.rpc_proc ? t->tk_msg.rpc_proc->p_proc : -1),
-			t->tk_flags, t->tk_status,
-			t->tk_client,
-			(t->tk_client ? t->tk_client->cl_prog : 0),
-			t->tk_rqstp, t->tk_timeout,
-			rpc_waitq,
-			t->tk_action, t->tk_ops);
-	}
-	spin_unlock(&rpc_sched_lock);
-}
-#endif
-
 void
 rpc_destroy_mempool(void)
 {

net/sunrpc/sunrpc_syms.c

@@ -28,15 +28,11 @@ EXPORT_SYMBOL(rpc_init_task);
 EXPORT_SYMBOL(rpc_sleep_on);
 EXPORT_SYMBOL(rpc_wake_up_next);
 EXPORT_SYMBOL(rpc_wake_up_task);
-EXPORT_SYMBOL(rpciod_down);
-EXPORT_SYMBOL(rpciod_up);
-EXPORT_SYMBOL(rpc_new_task);
 EXPORT_SYMBOL(rpc_wake_up_status);
 
 /* RPC client functions */
 EXPORT_SYMBOL(rpc_clone_client);
 EXPORT_SYMBOL(rpc_bind_new_program);
-EXPORT_SYMBOL(rpc_destroy_client);
 EXPORT_SYMBOL(rpc_shutdown_client);
 EXPORT_SYMBOL(rpc_killall_tasks);
 EXPORT_SYMBOL(rpc_call_sync);
@@ -61,7 +57,7 @@ EXPORT_SYMBOL(rpcauth_unregister);
 EXPORT_SYMBOL(rpcauth_create);
 EXPORT_SYMBOL(rpcauth_lookupcred);
 EXPORT_SYMBOL(rpcauth_lookup_credcache);
-EXPORT_SYMBOL(rpcauth_free_credcache);
+EXPORT_SYMBOL(rpcauth_destroy_credcache);
 EXPORT_SYMBOL(rpcauth_init_credcache);
 EXPORT_SYMBOL(put_rpccred);
 
@@ -156,6 +152,7 @@ init_sunrpc(void)
 	cache_register(&ip_map_cache);
 	cache_register(&unix_gid_cache);
 	init_socket_xprt();
+	rpcauth_init_module();
 out:
 	return err;
 }
@@ -163,6 +160,7 @@ out:
 static void __exit
 cleanup_sunrpc(void)
 {
+	rpcauth_remove_module();
 	cleanup_socket_xprt();
 	unregister_rpc_pipefs();
 	rpc_destroy_mempool();

net/sunrpc/svcsock.c

@@ -644,6 +644,7 @@ svc_recvfrom(struct svc_rqst *rqstp, struct kvec *iov, int nr, int buflen)
 	struct msghdr msg = {
 		.msg_flags	= MSG_DONTWAIT,
 	};
+	struct sockaddr *sin;
 	int len;
 
 	len = kernel_recvmsg(svsk->sk_sock, &msg, iov, nr, buflen,
@@ -654,6 +655,19 @@ svc_recvfrom(struct svc_rqst *rqstp, struct kvec *iov, int nr, int buflen)
 	memcpy(&rqstp->rq_addr, &svsk->sk_remote, svsk->sk_remotelen);
 	rqstp->rq_addrlen = svsk->sk_remotelen;
 
+	/* Destination address in request is needed for binding the
+	 * source address in RPC callbacks later.
+	 */
+	sin = (struct sockaddr *)&svsk->sk_local;
+	switch (sin->sa_family) {
+	case AF_INET:
+		rqstp->rq_daddr.addr = ((struct sockaddr_in *)sin)->sin_addr;
+		break;
+	case AF_INET6:
+		rqstp->rq_daddr.addr6 = ((struct sockaddr_in6 *)sin)->sin6_addr;
+		break;
+	}
+
 	dprintk("svc: socket %p recvfrom(%p, %Zu) = %d\n",
 		svsk, iov[0].iov_base, iov[0].iov_len, len);
 
@@ -1064,6 +1078,12 @@ svc_tcp_accept(struct svc_sock *svsk)
 		goto failed;
 	memcpy(&newsvsk->sk_remote, sin, slen);
 	newsvsk->sk_remotelen = slen;
+	err = kernel_getsockname(newsock, sin, &slen);
+	if (unlikely(err < 0)) {
+		dprintk("svc_tcp_accept: kernel_getsockname error %d\n", -err);
+		slen = offsetof(struct sockaddr, sa_data);
+	}
+	memcpy(&newsvsk->sk_local, sin, slen);
 
 	svc_sock_received(newsvsk);
 

net/sunrpc/xprt.c

@@ -127,7 +127,7 @@ static void xprt_clear_locked(struct rpc_xprt *xprt)
 		clear_bit(XPRT_LOCKED, &xprt->state);
 		smp_mb__after_clear_bit();
 	} else
-		schedule_work(&xprt->task_cleanup);
+		queue_work(rpciod_workqueue, &xprt->task_cleanup);
 }
 
 /*
@@ -515,7 +515,7 @@ xprt_init_autodisconnect(unsigned long data)
 	if (xprt_connecting(xprt))
 		xprt_release_write(xprt, NULL);
 	else
-		schedule_work(&xprt->task_cleanup);
+		queue_work(rpciod_workqueue, &xprt->task_cleanup);
 	return;
 out_abort:
 	spin_unlock(&xprt->transport_lock);
@@ -886,27 +886,24 @@ void xprt_set_timeout(struct rpc_timeout *to, unsigned int retr, unsigned long i
 
 /**
  * xprt_create_transport - create an RPC transport
- * @proto: requested transport protocol
- * @ap: remote peer address
- * @size: length of address
- * @to: timeout parameters
+ * @args: rpc transport creation arguments
  *
  */
-struct rpc_xprt *xprt_create_transport(int proto, struct sockaddr *ap, size_t size, struct rpc_timeout *to)
+struct rpc_xprt *xprt_create_transport(struct rpc_xprtsock_create *args)
 {
 	struct rpc_xprt	*xprt;
 	struct rpc_rqst	*req;
 
-	switch (proto) {
+	switch (args->proto) {
 	case IPPROTO_UDP:
-		xprt = xs_setup_udp(ap, size, to);
+		xprt = xs_setup_udp(args);
 		break;
 	case IPPROTO_TCP:
-		xprt = xs_setup_tcp(ap, size, to);
+		xprt = xs_setup_tcp(args);
 		break;
 	default:
 		printk(KERN_ERR "RPC: unrecognized transport protocol: %d\n",
-				proto);
+				args->proto);
 		return ERR_PTR(-EIO);
 	}
 	if (IS_ERR(xprt)) {

net/sunrpc/xprtsock.c

@@ -235,6 +235,7 @@ struct sock_xprt {
 	 * Connection of transports
 	 */
 	struct delayed_work	connect_worker;
+	struct sockaddr_storage	addr;
 	unsigned short		port;
 
 	/*
@@ -653,8 +654,7 @@ static void xs_destroy(struct rpc_xprt *xprt)
 
 	dprintk("RPC:       xs_destroy xprt %p\n", xprt);
 
-	cancel_delayed_work(&transport->connect_worker);
-	flush_scheduled_work();
+	cancel_rearming_delayed_work(&transport->connect_worker);
 
 	xprt_disconnect(xprt);
 	xs_close(xprt);
@@ -1001,7 +1001,7 @@ static void xs_tcp_state_change(struct sock *sk)
 		/* Try to schedule an autoclose RPC calls */
 		set_bit(XPRT_CLOSE_WAIT, &xprt->state);
 		if (test_and_set_bit(XPRT_LOCKED, &xprt->state) == 0)
-			schedule_work(&xprt->task_cleanup);
+			queue_work(rpciod_workqueue, &xprt->task_cleanup);
 	default:
 		xprt_disconnect(xprt);
 	}
@@ -1146,31 +1146,36 @@ static void xs_set_port(struct rpc_xprt *xprt, unsigned short port)
 	sap->sin_port = htons(port);
 }
 
-static int xs_bindresvport(struct sock_xprt *transport, struct socket *sock)
+static int xs_bind(struct sock_xprt *transport, struct socket *sock)
 {
 	struct sockaddr_in myaddr = {
 		.sin_family = AF_INET,
 	};
+	struct sockaddr_in *sa;
 	int err;
 	unsigned short port = transport->port;
 
+	if (!transport->xprt.resvport)
+		port = 0;
+	sa = (struct sockaddr_in *)&transport->addr;
+	myaddr.sin_addr = sa->sin_addr;
 	do {
 		myaddr.sin_port = htons(port);
 		err = kernel_bind(sock, (struct sockaddr *) &myaddr,
 						sizeof(myaddr));
+		if (!transport->xprt.resvport)
+			break;
 		if (err == 0) {
 			transport->port = port;
-			dprintk("RPC:       xs_bindresvport bound to port %u\n",
-					port);
-			return 0;
+			break;
 		}
 		if (port <= xprt_min_resvport)
 			port = xprt_max_resvport;
 		else
 			port--;
 	} while (err == -EADDRINUSE && port != transport->port);
-
-	dprintk("RPC:       can't bind to reserved port (%d).\n", -err);
+	dprintk("RPC:       xs_bind "NIPQUAD_FMT":%u: %s (%d)\n",
+		NIPQUAD(myaddr.sin_addr), port, err ? "failed" : "ok", err);
 	return err;
 }
 
@@ -1229,7 +1234,7 @@ static void xs_udp_connect_worker(struct work_struct *work)
 	}
 	xs_reclassify_socket(sock);
 
-	if (xprt->resvport && xs_bindresvport(transport, sock) < 0) {
+	if (xs_bind(transport, sock)) {
 		sock_release(sock);
 		goto out;
 	}
@@ -1316,7 +1321,7 @@ static void xs_tcp_connect_worker(struct work_struct *work)
 		}
 		xs_reclassify_socket(sock);
 
-		if (xprt->resvport && xs_bindresvport(transport, sock) < 0) {
+		if (xs_bind(transport, sock)) {
 			sock_release(sock);
 			goto out;
 		}
@@ -1410,18 +1415,16 @@ static void xs_connect(struct rpc_task *task)
 		dprintk("RPC:       xs_connect delayed xprt %p for %lu "
 				"seconds\n",
 				xprt, xprt->reestablish_timeout / HZ);
-		schedule_delayed_work(&transport->connect_worker,
-					xprt->reestablish_timeout);
+		queue_delayed_work(rpciod_workqueue,
+				   &transport->connect_worker,
+				   xprt->reestablish_timeout);
 		xprt->reestablish_timeout <<= 1;
 		if (xprt->reestablish_timeout > XS_TCP_MAX_REEST_TO)
 			xprt->reestablish_timeout = XS_TCP_MAX_REEST_TO;
 	} else {
 		dprintk("RPC:       xs_connect scheduled xprt %p\n", xprt);
-		schedule_delayed_work(&transport->connect_worker, 0);
-
-		/* flush_scheduled_work can sleep... */
-		if (!RPC_IS_ASYNC(task))
-			flush_scheduled_work();
+		queue_delayed_work(rpciod_workqueue,
+				   &transport->connect_worker, 0);
 	}
 }
 
@@ -1476,7 +1479,7 @@ static struct rpc_xprt_ops xs_udp_ops = {
 	.set_buffer_size	= xs_udp_set_buffer_size,
 	.reserve_xprt		= xprt_reserve_xprt_cong,
 	.release_xprt		= xprt_release_xprt_cong,
-	.rpcbind		= rpcb_getport,
+	.rpcbind		= rpcb_getport_async,
 	.set_port		= xs_set_port,
 	.connect		= xs_connect,
 	.buf_alloc		= rpc_malloc,
@@ -1493,7 +1496,7 @@ static struct rpc_xprt_ops xs_udp_ops = {
 static struct rpc_xprt_ops xs_tcp_ops = {
 	.reserve_xprt		= xprt_reserve_xprt,
 	.release_xprt		= xs_tcp_release_xprt,
-	.rpcbind		= rpcb_getport,
+	.rpcbind		= rpcb_getport_async,
 	.set_port		= xs_set_port,
 	.connect		= xs_connect,
 	.buf_alloc		= rpc_malloc,
@@ -1505,12 +1508,12 @@ static struct rpc_xprt_ops xs_tcp_ops = {
 	.print_stats		= xs_tcp_print_stats,
 };
 
-static struct rpc_xprt *xs_setup_xprt(struct sockaddr *addr, size_t addrlen, unsigned int slot_table_size)
+static struct rpc_xprt *xs_setup_xprt(struct rpc_xprtsock_create *args, unsigned int slot_table_size)
 {
 	struct rpc_xprt *xprt;
 	struct sock_xprt *new;
 
-	if (addrlen > sizeof(xprt->addr)) {
+	if (args->addrlen > sizeof(xprt->addr)) {
 		dprintk("RPC:       xs_setup_xprt: address too large\n");
 		return ERR_PTR(-EBADF);
 	}
@@ -1532,8 +1535,10 @@ static struct rpc_xprt *xs_setup_xprt(struct sockaddr *addr, size_t addrlen, uns
 		return ERR_PTR(-ENOMEM);
 	}
 
-	memcpy(&xprt->addr, addr, addrlen);
-	xprt->addrlen = addrlen;
+	memcpy(&xprt->addr, args->dstaddr, args->addrlen);
+	xprt->addrlen = args->addrlen;
+	if (args->srcaddr)
+		memcpy(&new->addr, args->srcaddr, args->addrlen);
 	new->port = xs_get_random_port();
 
 	return xprt;
@@ -1541,22 +1546,20 @@ static struct rpc_xprt *xs_setup_xprt(struct sockaddr *addr, size_t addrlen, uns
 
 /**
  * xs_setup_udp - Set up transport to use a UDP socket
- * @addr: address of remote server
- * @addrlen: length of address in bytes
- * @to:   timeout parameters
+ * @args: rpc transport creation arguments
  *
  */
-struct rpc_xprt *xs_setup_udp(struct sockaddr *addr, size_t addrlen, struct rpc_timeout *to)
+struct rpc_xprt *xs_setup_udp(struct rpc_xprtsock_create *args)
 {
 	struct rpc_xprt *xprt;
 	struct sock_xprt *transport;
 
-	xprt = xs_setup_xprt(addr, addrlen, xprt_udp_slot_table_entries);
+	xprt = xs_setup_xprt(args, xprt_udp_slot_table_entries);
 	if (IS_ERR(xprt))
 		return xprt;
 	transport = container_of(xprt, struct sock_xprt, xprt);
 
-	if (ntohs(((struct sockaddr_in *)addr)->sin_port) != 0)
+	if (ntohs(((struct sockaddr_in *)args->dstaddr)->sin_port) != 0)
 		xprt_set_bound(xprt);
 
 	xprt->prot = IPPROTO_UDP;
@@ -1572,8 +1575,8 @@ struct rpc_xprt *xs_setup_udp(struct sockaddr *addr, size_t addrlen, struct rpc_
 
 	xprt->ops = &xs_udp_ops;
 
-	if (to)
-		xprt->timeout = *to;
+	if (args->timeout)
+		xprt->timeout = *args->timeout;
 	else
 		xprt_set_timeout(&xprt->timeout, 5, 5 * HZ);
 
@@ -1586,22 +1589,20 @@ struct rpc_xprt *xs_setup_udp(struct sockaddr *addr, size_t addrlen, struct rpc_
 
 /**
  * xs_setup_tcp - Set up transport to use a TCP socket
- * @addr: address of remote server
- * @addrlen: length of address in bytes
- * @to: timeout parameters
+ * @args: rpc transport creation arguments
  *
  */
-struct rpc_xprt *xs_setup_tcp(struct sockaddr *addr, size_t addrlen, struct rpc_timeout *to)
+struct rpc_xprt *xs_setup_tcp(struct rpc_xprtsock_create *args)
 {
 	struct rpc_xprt *xprt;
 	struct sock_xprt *transport;
 
-	xprt = xs_setup_xprt(addr, addrlen, xprt_tcp_slot_table_entries);
+	xprt = xs_setup_xprt(args, xprt_tcp_slot_table_entries);
 	if (IS_ERR(xprt))
 		return xprt;
 	transport = container_of(xprt, struct sock_xprt, xprt);
 
-	if (ntohs(((struct sockaddr_in *)addr)->sin_port) != 0)
+	if (ntohs(((struct sockaddr_in *)args->dstaddr)->sin_port) != 0)
 		xprt_set_bound(xprt);
 
 	xprt->prot = IPPROTO_TCP;
@@ -1616,8 +1617,8 @@ struct rpc_xprt *xs_setup_tcp(struct sockaddr *addr, size_t addrlen, struct rpc_
 
 	xprt->ops = &xs_tcp_ops;
 
-	if (to)
-		xprt->timeout = *to;
+	if (args->timeout)
+		xprt->timeout = *args->timeout;
 	else
 		xprt_set_timeout(&xprt->timeout, 2, 60 * HZ);