Merge git://git.linux-nfs.org/projects/trondmy/nfs-2.6

* git://git.linux-nfs.org/projects/trondmy/nfs-2.6: (70 commits)
  fs/nfs/nfs4proc.c: make nfs4_map_errors() static
  rpc: add service field to new upcall
  rpc: add target field to new upcall
  nfsd: support callbacks with gss flavors
  rpc: allow gss callbacks to client
  rpc: pass target name down to rpc level on callbacks
  nfsd: pass client principal name in rsc downcall
  rpc: implement new upcall
  rpc: store pointer to pipe inode in gss upcall message
  rpc: use count of pipe openers to wait for first open
  rpc: track number of users of the gss upcall pipe
  rpc: call release_pipe only on last close
  rpc: add an rpc_pipe_open method
  rpc: minor gss_alloc_msg cleanup
  rpc: factor out warning code from gss_pipe_destroy_msg
  rpc: remove unnecessary assignment
  NFS: remove unused status from encode routines
  NFS: increment number of operations in each encode routine
  NFS: fix comment placement in nfs4xdr.c
  NFS: fix tabs in nfs4xdr.c
  ...

fs/lockd/clntlock.c

@@ -14,6 +14,7 @@
 #include <linux/sunrpc/svc.h>
 #include <linux/lockd/lockd.h>
 #include <linux/smp_lock.h>
+#include <linux/kthread.h>
 
 #define NLMDBG_FACILITY		NLMDBG_CLIENT
 
@@ -60,7 +61,7 @@ struct nlm_host *nlmclnt_init(const struct nlmclnt_initdata *nlm_init)
 
 	host = nlmclnt_lookup_host(nlm_init->address, nlm_init->addrlen,
 				   nlm_init->protocol, nlm_version,
-				   nlm_init->hostname);
+				   nlm_init->hostname, nlm_init->noresvport);
 	if (host == NULL) {
 		lockd_down();
 		return ERR_PTR(-ENOLCK);
@@ -191,11 +192,15 @@ __be32 nlmclnt_grant(const struct sockaddr *addr, const struct nlm_lock *lock)
 void
 nlmclnt_recovery(struct nlm_host *host)
 {
+	struct task_struct *task;
+
 	if (!host->h_reclaiming++) {
 		nlm_get_host(host);
-		__module_get(THIS_MODULE);
-		if (kernel_thread(reclaimer, host, CLONE_FS | CLONE_FILES) < 0)
-			module_put(THIS_MODULE);
+		task = kthread_run(reclaimer, host, "%s-reclaim", host->h_name);
+		if (IS_ERR(task))
+			printk(KERN_ERR "lockd: unable to spawn reclaimer "
+				"thread. Locks for %s won't be reclaimed! "
+				"(%ld)\n", host->h_name, PTR_ERR(task));
 	}
 }
 
@@ -207,7 +212,6 @@ reclaimer(void *ptr)
 	struct file_lock *fl, *next;
 	u32 nsmstate;
 
-	daemonize("%s-reclaim", host->h_name);
 	allow_signal(SIGKILL);
 
 	down_write(&host->h_rwsem);
@@ -233,7 +237,12 @@ restart:
 	list_for_each_entry_safe(fl, next, &host->h_reclaim, fl_u.nfs_fl.list) {
 		list_del_init(&fl->fl_u.nfs_fl.list);
 
-		/* Why are we leaking memory here? --okir */
+		/*
+		 * sending this thread a SIGKILL will result in any unreclaimed
+		 * locks being removed from the h_granted list. This means that
+		 * the kernel will not attempt to reclaim them again if a new
+		 * reclaimer thread is spawned for this host.
+		 */
 		if (signalled())
 			continue;
 		if (nlmclnt_reclaim(host, fl) != 0)
@@ -261,5 +270,5 @@ restart:
 	nlm_release_host(host);
 	lockd_down();
 	unlock_kernel();
-	module_put_and_exit(0);
+	return 0;
 }

fs/lockd/host.c

@@ -48,6 +48,7 @@ struct nlm_lookup_host_info {
 	const size_t		hostname_len;	/* it's length */
 	const struct sockaddr	*src_sap;	/* our address (optional) */
 	const size_t		src_len;	/* it's length */
+	const int		noresvport;	/* use non-priv port */
 };
 
 /*
@@ -222,6 +223,7 @@ static struct nlm_host *nlm_lookup_host(struct nlm_lookup_host_info *ni)
 	host->h_nsmstate   = 0;			/* real NSM state */
 	host->h_nsmhandle  = nsm;
 	host->h_server	   = ni->server;
+	host->h_noresvport = ni->noresvport;
 	hlist_add_head(&host->h_hash, chain);
 	INIT_LIST_HEAD(&host->h_lockowners);
 	spin_lock_init(&host->h_lock);
@@ -272,6 +274,7 @@ nlm_destroy_host(struct nlm_host *host)
  * @protocol: transport protocol to use
  * @version: NLM protocol version
  * @hostname: '\0'-terminated hostname of server
+ * @noresvport: 1 if non-privileged port should be used
  *
  * Returns an nlm_host structure that matches the passed-in
  * [server address, transport protocol, NLM version, server hostname].
@@ -281,7 +284,9 @@ nlm_destroy_host(struct nlm_host *host)
 struct nlm_host *nlmclnt_lookup_host(const struct sockaddr *sap,
 				     const size_t salen,
 				     const unsigned short protocol,
-				     const u32 version, const char *hostname)
+				     const u32 version,
+				     const char *hostname,
+				     int noresvport)
 {
 	const struct sockaddr source = {
 		.sa_family	= AF_UNSPEC,
@@ -296,6 +301,7 @@ struct nlm_host *nlmclnt_lookup_host(const struct sockaddr *sap,
 		.hostname_len	= strlen(hostname),
 		.src_sap	= &source,
 		.src_len	= sizeof(source),
+		.noresvport	= noresvport,
 	};
 
 	dprintk("lockd: %s(host='%s', vers=%u, proto=%s)\n", __func__,
@@ -417,6 +423,8 @@ nlm_bind_host(struct nlm_host *host)
 		 */
 		if (!host->h_server)
 			args.flags |= RPC_CLNT_CREATE_HARDRTRY;
+		if (host->h_noresvport)
+			args.flags |= RPC_CLNT_CREATE_NONPRIVPORT;
 
 		clnt = rpc_create(&args);
 		if (!IS_ERR(clnt))

fs/lockd/svc.c

@@ -45,7 +45,7 @@
 static struct svc_program	nlmsvc_program;
 
 struct nlmsvc_binding *		nlmsvc_ops;
-EXPORT_SYMBOL(nlmsvc_ops);
+EXPORT_SYMBOL_GPL(nlmsvc_ops);
 
 static DEFINE_MUTEX(nlmsvc_mutex);
 static unsigned int		nlmsvc_users;
@@ -300,7 +300,7 @@ out:
 	mutex_unlock(&nlmsvc_mutex);
 	return error;
 }
-EXPORT_SYMBOL(lockd_up);
+EXPORT_SYMBOL_GPL(lockd_up);
 
 /*
  * Decrement the user count and bring down lockd if we're the last.
@@ -329,7 +329,7 @@ lockd_down(void)
 out:
 	mutex_unlock(&nlmsvc_mutex);
 }
-EXPORT_SYMBOL(lockd_down);
+EXPORT_SYMBOL_GPL(lockd_down);
 
 #ifdef CONFIG_SYSCTL
 

fs/nfs/callback.c

@@ -16,6 +16,7 @@
 #include <linux/mutex.h>
 #include <linux/freezer.h>
 #include <linux/kthread.h>
+#include <linux/sunrpc/svcauth_gss.h>
 
 #include <net/inet_sock.h>
 
@@ -182,10 +183,34 @@ void nfs_callback_down(void)
 	mutex_unlock(&nfs_callback_mutex);
 }
 
+static int check_gss_callback_principal(struct nfs_client *clp,
+					struct svc_rqst *rqstp)
+{
+	struct rpc_clnt *r = clp->cl_rpcclient;
+	char *p = svc_gss_principal(rqstp);
+
+	/*
+	 * It might just be a normal user principal, in which case
+	 * userspace won't bother to tell us the name at all.
+	 */
+	if (p == NULL)
+		return SVC_DENIED;
+
+	/* Expect a GSS_C_NT_HOSTBASED_NAME like "nfs@serverhostname" */
+
+	if (memcmp(p, "nfs@", 4) != 0)
+		return SVC_DENIED;
+	p += 4;
+	if (strcmp(p, r->cl_server) != 0)
+		return SVC_DENIED;
+	return SVC_OK;
+}
+
 static int nfs_callback_authenticate(struct svc_rqst *rqstp)
 {
 	struct nfs_client *clp;
 	RPC_IFDEBUG(char buf[RPC_MAX_ADDRBUFLEN]);
+	int ret = SVC_OK;
 
 	/* Don't talk to strangers */
 	clp = nfs_find_client(svc_addr(rqstp), 4);
@@ -194,21 +219,22 @@ static int nfs_callback_authenticate(struct svc_rqst *rqstp)
 
 	dprintk("%s: %s NFSv4 callback!\n", __func__,
 			svc_print_addr(rqstp, buf, sizeof(buf)));
-	nfs_put_client(clp);
 
 	switch (rqstp->rq_authop->flavour) {
 		case RPC_AUTH_NULL:
 			if (rqstp->rq_proc != CB_NULL)
-				return SVC_DENIED;
+				ret = SVC_DENIED;
 			break;
 		case RPC_AUTH_UNIX:
 			break;
 		case RPC_AUTH_GSS:
-			/* FIXME: RPCSEC_GSS handling? */
+			ret = check_gss_callback_principal(clp, rqstp);
+			break;
 		default:
-			return SVC_DENIED;
+			ret = SVC_DENIED;
 	}
-	return SVC_OK;
+	nfs_put_client(clp);
+	return ret;
 }
 
 /*

fs/nfs/client.c

@@ -143,7 +143,6 @@ static struct nfs_client *nfs_alloc_client(const struct nfs_client_initdata *cl_
 	clp->cl_proto = cl_init->proto;
 
 #ifdef CONFIG_NFS_V4
-	init_rwsem(&clp->cl_sem);
 	INIT_LIST_HEAD(&clp->cl_delegations);
 	spin_lock_init(&clp->cl_lock);
 	INIT_DELAYED_WORK(&clp->cl_renewd, nfs4_renew_state);
@@ -224,31 +223,54 @@ void nfs_put_client(struct nfs_client *clp)
 	}
 }
 
-static int nfs_sockaddr_match_ipaddr4(const struct sockaddr_in *sa1,
-				 const struct sockaddr_in *sa2)
+#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
+static const struct in6_addr *nfs_map_ipv4_addr(const struct sockaddr *sa, struct in6_addr *addr_mapped)
 {
-	return sa1->sin_addr.s_addr == sa2->sin_addr.s_addr;
+	switch (sa->sa_family) {
+		default:
+			return NULL;
+		case AF_INET6:
+			return &((const struct sockaddr_in6 *)sa)->sin6_addr;
+			break;
+		case AF_INET:
+			ipv6_addr_set_v4mapped(((const struct sockaddr_in *)sa)->sin_addr.s_addr,
+					addr_mapped);
+			return addr_mapped;
+	}
 }
 
-static int nfs_sockaddr_match_ipaddr6(const struct sockaddr_in6 *sa1,
-				 const struct sockaddr_in6 *sa2)
+static int nfs_sockaddr_match_ipaddr(const struct sockaddr *sa1,
+		const struct sockaddr *sa2)
+{
+	const struct in6_addr *addr1;
+	const struct in6_addr *addr2;
+	struct in6_addr addr1_mapped;
+	struct in6_addr addr2_mapped;
+
+	addr1 = nfs_map_ipv4_addr(sa1, &addr1_mapped);
+	if (likely(addr1 != NULL)) {
+		addr2 = nfs_map_ipv4_addr(sa2, &addr2_mapped);
+		if (likely(addr2 != NULL))
+			return ipv6_addr_equal(addr1, addr2);
+	}
+	return 0;
+}
+#else
+static int nfs_sockaddr_match_ipaddr4(const struct sockaddr_in *sa1,
+				 const struct sockaddr_in *sa2)
 {
-	return ipv6_addr_equal(&sa1->sin6_addr, &sa2->sin6_addr);
+	return sa1->sin_addr.s_addr == sa2->sin_addr.s_addr;
 }
 
 static int nfs_sockaddr_match_ipaddr(const struct sockaddr *sa1,
 				 const struct sockaddr *sa2)
 {
-	switch (sa1->sa_family) {
-	case AF_INET:
-		return nfs_sockaddr_match_ipaddr4((const struct sockaddr_in *)sa1,
-				(const struct sockaddr_in *)sa2);
-	case AF_INET6:
-		return nfs_sockaddr_match_ipaddr6((const struct sockaddr_in6 *)sa1,
-				(const struct sockaddr_in6 *)sa2);
-	}
-	BUG();
+	if (unlikely(sa1->sa_family != AF_INET || sa2->sa_family != AF_INET))
+		return 0;
+	return nfs_sockaddr_match_ipaddr4((const struct sockaddr_in *)sa1,
+			(const struct sockaddr_in *)sa2);
 }
+#endif
 
 /*
  * Find a client by IP address and protocol version
@@ -270,8 +292,6 @@ struct nfs_client *nfs_find_client(const struct sockaddr *addr, u32 nfsversion)
 		if (clp->rpc_ops->version != nfsversion)
 			continue;
 
-		if (addr->sa_family != clap->sa_family)
-			continue;
 		/* Match only the IP address, not the port number */
 		if (!nfs_sockaddr_match_ipaddr(addr, clap))
 			continue;
@@ -305,8 +325,6 @@ struct nfs_client *nfs_find_client_next(struct nfs_client *clp)
 		if (clp->rpc_ops->version != nfsvers)
 			continue;
 
-		if (sap->sa_family != clap->sa_family)
-			continue;
 		/* Match only the IP address, not the port number */
 		if (!nfs_sockaddr_match_ipaddr(sap, clap))
 			continue;
@@ -470,7 +488,7 @@ static void nfs_init_timeout_values(struct rpc_timeout *to, int proto,
 static int nfs_create_rpc_client(struct nfs_client *clp,
 				 const struct rpc_timeout *timeparms,
 				 rpc_authflavor_t flavor,
-				 int flags)
+				 int discrtry, int noresvport)
 {
 	struct rpc_clnt		*clnt = NULL;
 	struct rpc_create_args args = {
@@ -482,9 +500,13 @@ static int nfs_create_rpc_client(struct nfs_client *clp,
 		.program	= &nfs_program,
 		.version	= clp->rpc_ops->version,
 		.authflavor	= flavor,
-		.flags		= flags,
 	};
 
+	if (discrtry)
+		args.flags |= RPC_CLNT_CREATE_DISCRTRY;
+	if (noresvport)
+		args.flags |= RPC_CLNT_CREATE_NONPRIVPORT;
+
 	if (!IS_ERR(clp->cl_rpcclient))
 		return 0;
 
@@ -522,6 +544,8 @@ static int nfs_start_lockd(struct nfs_server *server)
 		.protocol	= server->flags & NFS_MOUNT_TCP ?
 						IPPROTO_TCP : IPPROTO_UDP,
 		.nfs_version	= clp->rpc_ops->version,
+		.noresvport	= server->flags & NFS_MOUNT_NORESVPORT ?
+					1 : 0,
 	};
 
 	if (nlm_init.nfs_version > 3)
@@ -623,7 +647,8 @@ static int nfs_init_client(struct nfs_client *clp,
 	 * Create a client RPC handle for doing FSSTAT with UNIX auth only
 	 * - RFC 2623, sec 2.3.2
 	 */
-	error = nfs_create_rpc_client(clp, timeparms, RPC_AUTH_UNIX, 0);
+	error = nfs_create_rpc_client(clp, timeparms, RPC_AUTH_UNIX,
+				      0, data->flags & NFS_MOUNT_NORESVPORT);
 	if (error < 0)
 		goto error;
 	nfs_mark_client_ready(clp, NFS_CS_READY);
@@ -965,7 +990,8 @@ error:
 static int nfs4_init_client(struct nfs_client *clp,
 		const struct rpc_timeout *timeparms,
 		const char *ip_addr,
-		rpc_authflavor_t authflavour)
+		rpc_authflavor_t authflavour,
+		int flags)
 {
 	int error;
 
@@ -979,7 +1005,7 @@ static int nfs4_init_client(struct nfs_client *clp,
 	clp->rpc_ops = &nfs_v4_clientops;
 
 	error = nfs_create_rpc_client(clp, timeparms, authflavour,
-					RPC_CLNT_CREATE_DISCRTRY);
+				      1, flags & NFS_MOUNT_NORESVPORT);
 	if (error < 0)
 		goto error;
 	memcpy(clp->cl_ipaddr, ip_addr, sizeof(clp->cl_ipaddr));
@@ -1030,7 +1056,8 @@ static int nfs4_set_client(struct nfs_server *server,
 		error = PTR_ERR(clp);
 		goto error;
 	}
-	error = nfs4_init_client(clp, timeparms, ip_addr, authflavour);
+	error = nfs4_init_client(clp, timeparms, ip_addr, authflavour,
+					server->flags);
 	if (error < 0)
 		goto error_put;
 
@@ -1059,6 +1086,10 @@ static int nfs4_init_server(struct nfs_server *server,
 	nfs_init_timeout_values(&timeparms, data->nfs_server.protocol,
 			data->timeo, data->retrans);
 
+	/* Initialise the client representation from the mount data */
+	server->flags = data->flags;
+	server->caps |= NFS_CAP_ATOMIC_OPEN;
+
 	/* Get a client record */
 	error = nfs4_set_client(server,
 			data->nfs_server.hostname,
@@ -1071,10 +1102,6 @@ static int nfs4_init_server(struct nfs_server *server,
 	if (error < 0)
 		goto error;
 
-	/* Initialise the client representation from the mount data */
-	server->flags = data->flags;
-	server->caps |= NFS_CAP_ATOMIC_OPEN;
-
 	if (data->rsize)
 		server->rsize = nfs_block_size(data->rsize, NULL);
 	if (data->wsize)
@@ -1177,6 +1204,10 @@ struct nfs_server *nfs4_create_referral_server(struct nfs_clone_mount *data,
 	parent_server = NFS_SB(data->sb);
 	parent_client = parent_server->nfs_client;
 
+	/* Initialise the client representation from the parent server */
+	nfs_server_copy_userdata(server, parent_server);
+	server->caps |= NFS_CAP_ATOMIC_OPEN;
+
 	/* Get a client representation.
 	 * Note: NFSv4 always uses TCP, */
 	error = nfs4_set_client(server, data->hostname,
@@ -1189,10 +1220,6 @@ struct nfs_server *nfs4_create_referral_server(struct nfs_clone_mount *data,
 	if (error < 0)
 		goto error;
 
-	/* Initialise the client representation from the parent server */
-	nfs_server_copy_userdata(server, parent_server);
-	server->caps |= NFS_CAP_ATOMIC_OPEN;
-
 	error = nfs_init_server_rpcclient(server, parent_server->client->cl_timeout, data->authflavor);
 	if (error < 0)
 		goto error;

fs/nfs/delegation.c

@@ -43,6 +43,27 @@ static void nfs_free_delegation(struct nfs_delegation *delegation)
 		put_rpccred(cred);
 }
 
+void nfs_mark_delegation_referenced(struct nfs_delegation *delegation)
+{
+	set_bit(NFS_DELEGATION_REFERENCED, &delegation->flags);
+}
+
+int nfs_have_delegation(struct inode *inode, fmode_t flags)
+{
+	struct nfs_delegation *delegation;
+	int ret = 0;
+
+	flags &= FMODE_READ|FMODE_WRITE;
+	rcu_read_lock();
+	delegation = rcu_dereference(NFS_I(inode)->delegation);
+	if (delegation != NULL && (delegation->type & flags) == flags) {
+		nfs_mark_delegation_referenced(delegation);
+		ret = 1;
+	}
+	rcu_read_unlock();
+	return ret;
+}
+
 static int nfs_delegation_claim_locks(struct nfs_open_context *ctx, struct nfs4_state *state)
 {
 	struct inode *inode = state->inode;
@@ -119,7 +140,7 @@ void nfs_inode_reclaim_delegation(struct inode *inode, struct rpc_cred *cred, st
 	delegation->maxsize = res->maxsize;
 	oldcred = delegation->cred;
 	delegation->cred = get_rpccred(cred);
-	delegation->flags &= ~NFS_DELEGATION_NEED_RECLAIM;
+	clear_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags);
 	NFS_I(inode)->delegation_state = delegation->type;
 	smp_wmb();
 	put_rpccred(oldcred);
@@ -134,19 +155,35 @@ static int nfs_do_return_delegation(struct inode *inode, struct nfs_delegation *
 	return res;
 }
 
+static struct inode *nfs_delegation_grab_inode(struct nfs_delegation *delegation)
+{
+	struct inode *inode = NULL;
+
+	spin_lock(&delegation->lock);
+	if (delegation->inode != NULL)
+		inode = igrab(delegation->inode);
+	spin_unlock(&delegation->lock);
+	return inode;
+}
+
 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;
+	spin_lock(&delegation->lock);
 	if (stateid != NULL && memcmp(delegation->stateid.data, stateid->data,
 				sizeof(delegation->stateid.data)) != 0)
-		goto nomatch;
+		goto nomatch_unlock;
 	list_del_rcu(&delegation->super_list);
+	delegation->inode = NULL;
 	nfsi->delegation_state = 0;
 	rcu_assign_pointer(nfsi->delegation, NULL);
+	spin_unlock(&delegation->lock);
 	return delegation;
+nomatch_unlock:
+	spin_unlock(&delegation->lock);
 nomatch:
 	return NULL;
 }
@@ -172,6 +209,8 @@ int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct
 	delegation->change_attr = nfsi->change_attr;
 	delegation->cred = get_rpccred(cred);
 	delegation->inode = inode;
+	delegation->flags = 1<<NFS_DELEGATION_REFERENCED;
+	spin_lock_init(&delegation->lock);
 
 	spin_lock(&clp->cl_lock);
 	if (rcu_dereference(nfsi->delegation) != NULL) {
@@ -226,21 +265,46 @@ static void nfs_msync_inode(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);
 
 	nfs_msync_inode(inode);
-	down_read(&clp->cl_sem);
 	/* Guard against new delegated open calls */
 	down_write(&nfsi->rwsem);
 	nfs_delegation_claim_opens(inode, &delegation->stateid);
 	up_write(&nfsi->rwsem);
-	up_read(&clp->cl_sem);
 	nfs_msync_inode(inode);
 
 	return nfs_do_return_delegation(inode, delegation, 1);
 }
 
+/*
+ * Return all delegations that have been marked for return
+ */
+void nfs_client_return_marked_delegations(struct nfs_client *clp)
+{
+	struct nfs_delegation *delegation;
+	struct inode *inode;
+
+restart:
+	rcu_read_lock();
+	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
+		if (!test_and_clear_bit(NFS_DELEGATION_RETURN, &delegation->flags))
+			continue;
+		inode = nfs_delegation_grab_inode(delegation);
+		if (inode == NULL)
+			continue;
+		spin_lock(&clp->cl_lock);
+		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
+		spin_unlock(&clp->cl_lock);
+		rcu_read_unlock();
+		if (delegation != NULL)
+			__nfs_inode_return_delegation(inode, delegation);
+		iput(inode);
+		goto restart;
+	}
+	rcu_read_unlock();
+}
+
 /*
  * This function returns the delegation without reclaiming opens
  * or protecting against delegation reclaims.
@@ -279,83 +343,55 @@ int nfs_inode_return_delegation(struct inode *inode)
 	return err;
 }
 
+static void nfs_mark_return_delegation(struct nfs_client *clp, struct nfs_delegation *delegation)
+{
+	set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
+	set_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state);
+}
+
 /*
  * Return all delegations associated to a super block
  */
-void nfs_return_all_delegations(struct super_block *sb)
+void nfs_super_return_all_delegations(struct super_block *sb)
 {
 	struct nfs_client *clp = NFS_SB(sb)->nfs_client;
 	struct nfs_delegation *delegation;
-	struct inode *inode;
 
 	if (clp == NULL)
 		return;
-restart:
 	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);
-		rcu_read_unlock();
-		if (delegation != NULL)
-			__nfs_inode_return_delegation(inode, delegation);
-		iput(inode);
-		goto restart;
+		spin_lock(&delegation->lock);
+		if (delegation->inode != NULL && delegation->inode->i_sb == sb)
+			set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
+		spin_unlock(&delegation->lock);
 	}
 	rcu_read_unlock();
+	nfs_client_return_marked_delegations(clp);
 }
 
-static int nfs_do_expire_all_delegations(void *ptr)
+static void nfs_client_mark_return_all_delegations(struct nfs_client *clp)
 {
-	struct nfs_client *clp = ptr;
 	struct nfs_delegation *delegation;
-	struct inode *inode;
 
-	allow_signal(SIGKILL);
-restart:
-	if (test_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state) != 0)
-		goto out;
-	if (test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) == 0)
-		goto out;
 	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);
-		rcu_read_unlock();
-		if (delegation)
-			__nfs_inode_return_delegation(inode, delegation);
-		iput(inode);
-		goto restart;
+		set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
+		set_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state);
 	}
 	rcu_read_unlock();
-out:
-	nfs_put_client(clp);
-	module_put_and_exit(0);
+}
+
+static void nfs_delegation_run_state_manager(struct nfs_client *clp)
+{
+	if (test_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state))
+		nfs4_schedule_state_manager(clp);
 }
 
 void nfs_expire_all_delegations(struct nfs_client *clp)
 {
-	struct task_struct *task;
-
-	__module_get(THIS_MODULE);
-	atomic_inc(&clp->cl_count);
-	task = kthread_run(nfs_do_expire_all_delegations, clp,
-				"%s-delegreturn",
-				rpc_peeraddr2str(clp->cl_rpcclient,
-							RPC_DISPLAY_ADDR));
-	if (!IS_ERR(task))
-		return;
-	nfs_put_client(clp);
-	module_put(THIS_MODULE);
+	nfs_client_mark_return_all_delegations(clp);
+	nfs_delegation_run_state_manager(clp);
 }
 
 /*
@@ -363,68 +399,29 @@ void nfs_expire_all_delegations(struct nfs_client *clp)
  */
 void nfs_handle_cb_pathdown(struct nfs_client *clp)
 {
-	struct nfs_delegation *delegation;
-	struct inode *inode;
-
 	if (clp == NULL)
 		return;
-restart:
+	nfs_client_mark_return_all_delegations(clp);
+}
+
+static void nfs_client_mark_return_unreferenced_delegations(struct nfs_client *clp)
+{
+	struct nfs_delegation *delegation;
+
 	rcu_read_lock();
 	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
-		inode = igrab(delegation->inode);
-		if (inode == NULL)
+		if (test_and_clear_bit(NFS_DELEGATION_REFERENCED, &delegation->flags))
 			continue;
-		spin_lock(&clp->cl_lock);
-		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
-		spin_unlock(&clp->cl_lock);
-		rcu_read_unlock();
-		if (delegation != NULL)
-			__nfs_inode_return_delegation(inode, delegation);
-		iput(inode);
-		goto restart;
+		set_bit(NFS_DELEGATION_RETURN, &delegation->flags);
+		set_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state);
 	}
 	rcu_read_unlock();
 }
 
-struct recall_threadargs {
-	struct inode *inode;
-	struct nfs_client *clp;
-	const nfs4_stateid *stateid;
-
-	struct completion started;
-	int result;
-};
-
-static int recall_thread(void *data)
+void nfs_expire_unreferenced_delegations(struct nfs_client *clp)
 {
-	struct recall_threadargs *args = (struct recall_threadargs *)data;
-	struct inode *inode = igrab(args->inode);
-	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
-	struct nfs_inode *nfsi = NFS_I(inode);
-	struct nfs_delegation *delegation;
-
-	daemonize("nfsv4-delegreturn");
-
-	nfs_msync_inode(inode);
-	down_read(&clp->cl_sem);
-	down_write(&nfsi->rwsem);
-	spin_lock(&clp->cl_lock);
-	delegation = nfs_detach_delegation_locked(nfsi, args->stateid);
-	if (delegation != NULL)
-		args->result = 0;
-	else
-		args->result = -ENOENT;
-	spin_unlock(&clp->cl_lock);
-	complete(&args->started);
-	nfs_delegation_claim_opens(inode, args->stateid);
-	up_write(&nfsi->rwsem);
-	up_read(&clp->cl_sem);
-	nfs_msync_inode(inode);
-
-	if (delegation != NULL)
-		nfs_do_return_delegation(inode, delegation, 1);
-	iput(inode);
-	module_put_and_exit(0);
+	nfs_client_mark_return_unreferenced_delegations(clp);
+	nfs_delegation_run_state_manager(clp);
 }
 
 /*
@@ -432,22 +429,20 @@ static int recall_thread(void *data)
  */
 int nfs_async_inode_return_delegation(struct inode *inode, const nfs4_stateid *stateid)
 {
-	struct recall_threadargs data = {
-		.inode = inode,
-		.stateid = stateid,
-	};
-	int status;
+	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
+	struct nfs_delegation *delegation;
 
-	init_completion(&data.started);
-	__module_get(THIS_MODULE);
-	status = kernel_thread(recall_thread, &data, CLONE_KERNEL);
-	if (status < 0)
-		goto out_module_put;
-	wait_for_completion(&data.started);
-	return data.result;
-out_module_put:
-	module_put(THIS_MODULE);
-	return status;
+	rcu_read_lock();
+	delegation = rcu_dereference(NFS_I(inode)->delegation);
+	if (delegation == NULL || memcmp(delegation->stateid.data, stateid->data,
+				sizeof(delegation->stateid.data)) != 0) {
+		rcu_read_unlock();
+		return -ENOENT;
+	}
+	nfs_mark_return_delegation(clp, delegation);
+	rcu_read_unlock();
+	nfs_delegation_run_state_manager(clp);
+	return 0;
 }
 
 /*
@@ -459,10 +454,14 @@ struct inode *nfs_delegation_find_inode(struct nfs_client *clp, const struct nfs
 	struct inode *res = NULL;
 	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) {
+		spin_lock(&delegation->lock);
+		if (delegation->inode != NULL &&
+		    nfs_compare_fh(fhandle, &NFS_I(delegation->inode)->fh) == 0) {
 			res = igrab(delegation->inode);
-			break;
 		}
+		spin_unlock(&delegation->lock);
+		if (res != NULL)
+			break;
 	}
 	rcu_read_unlock();
 	return res;
@@ -476,7 +475,7 @@ void nfs_delegation_mark_reclaim(struct nfs_client *clp)
 	struct nfs_delegation *delegation;
 	rcu_read_lock();
 	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list)
-		delegation->flags |= NFS_DELEGATION_NEED_RECLAIM;
+		set_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags);
 	rcu_read_unlock();
 }
 
@@ -486,17 +485,22 @@ void nfs_delegation_mark_reclaim(struct nfs_client *clp)
 void nfs_delegation_reap_unclaimed(struct nfs_client *clp)
 {
 	struct nfs_delegation *delegation;
+	struct inode *inode;
 restart:
 	rcu_read_lock();
 	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
-		if ((delegation->flags & NFS_DELEGATION_NEED_RECLAIM) == 0)
+		if (test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) == 0)
+			continue;
+		inode = nfs_delegation_grab_inode(delegation);
+		if (inode == NULL)
 			continue;
 		spin_lock(&clp->cl_lock);
-		delegation = nfs_detach_delegation_locked(NFS_I(delegation->inode), NULL);
+		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
 		spin_unlock(&clp->cl_lock);
 		rcu_read_unlock();
 		if (delegation != NULL)
 			nfs_free_delegation(delegation);
+		iput(inode);
 		goto restart;
 	}
 	rcu_read_unlock();

fs/nfs/delegation.h

@@ -17,14 +17,20 @@ struct nfs_delegation {
 	struct rpc_cred *cred;
 	struct inode *inode;
 	nfs4_stateid stateid;
-	int type;
-#define NFS_DELEGATION_NEED_RECLAIM 1
-	long flags;
+	fmode_t type;
 	loff_t maxsize;
 	__u64 change_attr;
+	unsigned long flags;
+	spinlock_t lock;
 	struct rcu_head rcu;
 };
 
+enum {
+	NFS_DELEGATION_NEED_RECLAIM = 0,
+	NFS_DELEGATION_RETURN,
+	NFS_DELEGATION_REFERENCED,
+};
+
 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);
@@ -32,9 +38,11 @@ int nfs_async_inode_return_delegation(struct inode *inode, const nfs4_stateid *s
 void nfs_inode_return_delegation_noreclaim(struct inode *inode);
 
 struct inode *nfs_delegation_find_inode(struct nfs_client *clp, const struct nfs_fh *fhandle);
-void nfs_return_all_delegations(struct super_block *sb);
+void nfs_super_return_all_delegations(struct super_block *sb);
 void nfs_expire_all_delegations(struct nfs_client *clp);
+void nfs_expire_unreferenced_delegations(struct nfs_client *clp);
 void nfs_handle_cb_pathdown(struct nfs_client *clp);
+void nfs_client_return_marked_delegations(struct nfs_client *clp);
 
 void nfs_delegation_mark_reclaim(struct nfs_client *clp);
 void nfs_delegation_reap_unclaimed(struct nfs_client *clp);
@@ -45,22 +53,11 @@ int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state
 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;
-	rcu_read_lock();
-	delegation = rcu_dereference(NFS_I(inode)->delegation);
-	if (delegation != NULL && (delegation->type & flags) == flags)
-		ret = 1;
-	rcu_read_unlock();
-	return ret;
-}
+void nfs_mark_delegation_referenced(struct nfs_delegation *delegation);
+int nfs_have_delegation(struct inode *inode, fmode_t flags);
 
 #else
-static inline int nfs_have_delegation(struct inode *inode, int flags)
+static inline int nfs_have_delegation(struct inode *inode, fmode_t flags)
 {
 	return 0;
 }

fs/nfs/dir.c

@@ -799,6 +799,9 @@ static int nfs_lookup_revalidate(struct dentry * dentry, struct nameidata *nd)
 		goto out_bad;
 	}
 
+	if (nfs_have_delegation(inode, FMODE_READ))
+		goto out_set_verifier;
+
 	/* Force a full look up iff the parent directory has changed */
 	if (!nfs_is_exclusive_create(dir, nd) && nfs_check_verifier(dir, dentry)) {
 		if (nfs_lookup_verify_inode(inode, nd))
@@ -817,6 +820,7 @@ static int nfs_lookup_revalidate(struct dentry * dentry, struct nameidata *nd)
 	if ((error = nfs_refresh_inode(inode, &fattr)) != 0)
 		goto out_bad;
 
+out_set_verifier:
 	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
  out_valid:
 	dput(parent);
@@ -973,7 +977,7 @@ struct dentry_operations nfs4_dentry_operations = {
  * Use intent information to determine whether we need to substitute
  * the NFSv4-style stateful OPEN for the LOOKUP call
  */
-static int is_atomic_open(struct inode *dir, struct nameidata *nd)
+static int is_atomic_open(struct nameidata *nd)
 {
 	if (nd == NULL || nfs_lookup_check_intent(nd, LOOKUP_OPEN) == 0)
 		return 0;
@@ -996,7 +1000,7 @@ static struct dentry *nfs_atomic_lookup(struct inode *dir, struct dentry *dentry
 			dir->i_sb->s_id, dir->i_ino, dentry->d_name.name);
 
 	/* Check that we are indeed trying to open this file */
-	if (!is_atomic_open(dir, nd))
+	if (!is_atomic_open(nd))
 		goto no_open;
 
 	if (dentry->d_name.len > NFS_SERVER(dir)->namelen) {
@@ -1047,10 +1051,10 @@ static int nfs_open_revalidate(struct dentry *dentry, struct nameidata *nd)
 	struct inode *dir;
 	int openflags, ret = 0;
 
+	if (!is_atomic_open(nd))
+		goto no_open;
 	parent = dget_parent(dentry);
 	dir = parent->d_inode;
-	if (!is_atomic_open(dir, nd))
-		goto no_open;
 	/* We can't create new files in nfs_open_revalidate(), so we
 	 * optimize away revalidation of negative dentries.
 	 */
@@ -1062,11 +1066,11 @@ static int nfs_open_revalidate(struct dentry *dentry, struct nameidata *nd)
 
 	/* NFS only supports OPEN on regular files */
 	if (!S_ISREG(inode->i_mode))
-		goto no_open;
+		goto no_open_dput;
 	openflags = nd->intent.open.flags;
 	/* We cannot do exclusive creation on a positive dentry */
 	if ((openflags & (O_CREAT|O_EXCL)) == (O_CREAT|O_EXCL))
-		goto no_open;
+		goto no_open_dput;
 	/* We can't create new files, or truncate existing ones here */
 	openflags &= ~(O_CREAT|O_TRUNC);
 
@@ -1081,10 +1085,9 @@ out:
 	if (!ret)
 		d_drop(dentry);
 	return ret;
-no_open:
+no_open_dput:
 	dput(parent);
-	if (inode != NULL && nfs_have_delegation(inode, FMODE_READ))
-		return 1;
+no_open:
 	return nfs_lookup_revalidate(dentry, nd);
 }
 #endif /* CONFIG_NFSV4 */
@@ -1794,7 +1797,8 @@ static int nfs_access_get_cached(struct inode *inode, struct rpc_cred *cred, str
 	cache = nfs_access_search_rbtree(inode, cred);
 	if (cache == NULL)
 		goto out;
-	if (!time_in_range(jiffies, cache->jiffies, cache->jiffies + nfsi->attrtimeo))
+	if (!nfs_have_delegation(inode, FMODE_READ) &&
+	    !time_in_range_open(jiffies, cache->jiffies, cache->jiffies + nfsi->attrtimeo))
 		goto out_stale;
 	res->jiffies = cache->jiffies;
 	res->cred = cache->cred;

fs/nfs/inode.c

@@ -592,7 +592,7 @@ static void nfs_file_set_open_context(struct file *filp, struct nfs_open_context
 /*
  * Given an inode, search for an open context with the desired characteristics
  */
-struct nfs_open_context *nfs_find_open_context(struct inode *inode, struct rpc_cred *cred, int mode)
+struct nfs_open_context *nfs_find_open_context(struct inode *inode, struct rpc_cred *cred, fmode_t mode)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 	struct nfs_open_context *pos, *ctx = NULL;
@@ -712,14 +712,7 @@ int nfs_attribute_timeout(struct inode *inode)
 
 	if (nfs_have_delegation(inode, FMODE_READ))
 		return 0;
-	/*
-	 * Special case: if the attribute timeout is set to 0, then always
-	 * 		 treat the cache as having expired (unless holding
-	 * 		 a delegation).
-	 */
-	if (nfsi->attrtimeo == 0)
-		return 1;
-	return !time_in_range(jiffies, nfsi->read_cache_jiffies, nfsi->read_cache_jiffies + nfsi->attrtimeo);
+	return !time_in_range_open(jiffies, nfsi->read_cache_jiffies, nfsi->read_cache_jiffies + nfsi->attrtimeo);
 }
 
 /**
@@ -1182,7 +1175,7 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr)
 		nfsi->attrtimeo_timestamp = now;
 		nfsi->attr_gencount = nfs_inc_attr_generation_counter();
 	} else {
-		if (!time_in_range(now, nfsi->attrtimeo_timestamp, nfsi->attrtimeo_timestamp + nfsi->attrtimeo)) {
+		if (!time_in_range_open(now, nfsi->attrtimeo_timestamp, nfsi->attrtimeo_timestamp + nfsi->attrtimeo)) {
 			if ((nfsi->attrtimeo <<= 1) > NFS_MAXATTRTIMEO(inode))
 				nfsi->attrtimeo = NFS_MAXATTRTIMEO(inode);
 			nfsi->attrtimeo_timestamp = now;

fs/nfs/internal.h

@@ -63,6 +63,20 @@ struct nfs_parsed_mount_data {
 	struct security_mnt_opts lsm_opts;
 };
 
+/* mount_clnt.c */
+struct nfs_mount_request {
+	struct sockaddr		*sap;
+	size_t			salen;
+	char			*hostname;
+	char			*dirpath;
+	u32			version;
+	unsigned short		protocol;
+	struct nfs_fh		*fh;
+	int			noresvport;
+};
+
+extern int nfs_mount(struct nfs_mount_request *info);
+
 /* client.c */
 extern struct rpc_program nfs_program;
 

fs/nfs/mount_clnt.c

@@ -29,47 +29,43 @@ struct mnt_fhstatus {
 
 /**
  * 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
+ * @info: pointer to mount request arguments
  *
  * Uses default timeout parameters specified by underlying transport.
  */
-int nfs_mount(struct sockaddr *addr, size_t len, char *hostname, char *path,
-	      int version, int protocol, struct nfs_fh *fh)
+int nfs_mount(struct nfs_mount_request *info)
 {
 	struct mnt_fhstatus	result = {
-		.fh		= fh
+		.fh		= info->fh
 	};
 	struct rpc_message msg	= {
-		.rpc_argp	= path,
+		.rpc_argp	= info->dirpath,
 		.rpc_resp	= &result,
 	};
 	struct rpc_create_args args = {
-		.protocol	= protocol,
-		.address	= addr,
-		.addrsize	= len,
-		.servername	= hostname,
+		.protocol	= info->protocol,
+		.address	= info->sap,
+		.addrsize	= info->salen,
+		.servername	= info->hostname,
 		.program	= &mnt_program,
-		.version	= version,
+		.version	= info->version,
 		.authflavor	= RPC_AUTH_UNIX,
-		.flags		= 0,
 	};
 	struct rpc_clnt		*mnt_clnt;
 	int			status;
 
 	dprintk("NFS: sending MNT request for %s:%s\n",
-		(hostname ? hostname : "server"), path);
+		(info->hostname ? info->hostname : "server"),
+			info->dirpath);
+
+	if (info->noresvport)
+		args.flags |= RPC_CLNT_CREATE_NONPRIVPORT;
 
 	mnt_clnt = rpc_create(&args);
 	if (IS_ERR(mnt_clnt))
 		goto out_clnt_err;
 
-	if (version == NFS_MNT3_VERSION)
+	if (info->version == NFS_MNT3_VERSION)
 		msg.rpc_proc = &mnt_clnt->cl_procinfo[MOUNTPROC3_MNT];
 	else
 		msg.rpc_proc = &mnt_clnt->cl_procinfo[MNTPROC_MNT];

fs/nfs/nfs4_fs.h

@@ -38,8 +38,12 @@ struct idmap;
  ((err) != NFSERR_NOFILEHANDLE))
 
 enum nfs4_client_state {
-	NFS4CLNT_STATE_RECOVER  = 0,
+	NFS4CLNT_MANAGER_RUNNING  = 0,
+	NFS4CLNT_CHECK_LEASE,
 	NFS4CLNT_LEASE_EXPIRED,
+	NFS4CLNT_RECLAIM_REBOOT,
+	NFS4CLNT_RECLAIM_NOGRACE,
+	NFS4CLNT_DELEGRETURN,
 };
 
 /*
@@ -90,12 +94,18 @@ struct nfs4_state_owner {
 
 	spinlock_t	     so_lock;
 	atomic_t	     so_count;
+	unsigned long	     so_flags;
 	struct list_head     so_states;
 	struct list_head     so_delegations;
 	struct nfs_seqid_counter so_seqid;
 	struct rpc_sequence  so_sequence;
 };
 
+enum {
+	NFS_OWNER_RECLAIM_REBOOT,
+	NFS_OWNER_RECLAIM_NOGRACE
+};
+
 /*
  * struct nfs4_state maintains the client-side state for a given
  * (state_owner,inode) tuple (OPEN) or state_owner (LOCK).
@@ -128,6 +138,8 @@ enum {
 	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 */
+	NFS_STATE_RECLAIM_REBOOT,	/* OPEN stateid server rebooted */
+	NFS_STATE_RECLAIM_NOGRACE,	/* OPEN stateid needs to recover state */
 };
 
 struct nfs4_state {
@@ -149,7 +161,7 @@ struct nfs4_state {
 	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) */
+	fmode_t state;			/* State on the server (R,W, or RW) */
 	atomic_t count;
 };
 
@@ -157,9 +169,12 @@ struct nfs4_state {
 struct nfs4_exception {
 	long timeout;
 	int retry;
+	struct nfs4_state *state;
 };
 
 struct nfs4_state_recovery_ops {
+	int owner_flag_bit;
+	int state_flag_bit;
 	int (*recover_open)(struct nfs4_state_owner *, struct nfs4_state *);
 	int (*recover_lock)(struct nfs4_state *, struct file_lock *);
 };
@@ -174,7 +189,6 @@ extern ssize_t nfs4_listxattr(struct dentry *, char *, size_t);
 
 
 /* nfs4proc.c */
-extern int nfs4_map_errors(int err);
 extern int nfs4_proc_setclientid(struct nfs_client *, u32, unsigned short, struct rpc_cred *);
 extern int nfs4_proc_setclientid_confirm(struct nfs_client *, struct rpc_cred *);
 extern int nfs4_proc_async_renew(struct nfs_client *, struct rpc_cred *);
@@ -187,7 +201,7 @@ extern int nfs4_proc_fs_locations(struct inode *dir, const struct qstr *name,
 		struct nfs4_fs_locations *fs_locations, struct page *page);
 
 extern struct nfs4_state_recovery_ops nfs4_reboot_recovery_ops;
-extern struct nfs4_state_recovery_ops nfs4_network_partition_recovery_ops;
+extern struct nfs4_state_recovery_ops nfs4_nograce_recovery_ops;
 
 extern const u32 nfs4_fattr_bitmap[2];
 extern const u32 nfs4_statfs_bitmap[2];
@@ -202,16 +216,18 @@ extern void nfs4_kill_renewd(struct nfs_client *);
 extern void nfs4_renew_state(struct work_struct *);
 
 /* nfs4state.c */
-struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp);
+struct rpc_cred *nfs4_get_renew_cred_locked(struct nfs_client *clp);
 
 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 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 path *, struct nfs4_state *, mode_t);
-extern void nfs4_close_sync(struct path *, struct nfs4_state *, mode_t);
-extern void nfs4_state_set_mode_locked(struct nfs4_state *, mode_t);
+extern void nfs4_close_state(struct path *, struct nfs4_state *, fmode_t);
+extern void nfs4_close_sync(struct path *, struct nfs4_state *, fmode_t);
+extern void nfs4_state_set_mode_locked(struct nfs4_state *, fmode_t);
 extern void nfs4_schedule_state_recovery(struct nfs_client *);
+extern void nfs4_schedule_state_manager(struct nfs_client *);
+extern int nfs4_state_mark_reclaim_nograce(struct nfs_client *clp, struct nfs4_state *state);
 extern void nfs4_put_lock_state(struct nfs4_lock_state *lsp);
 extern int nfs4_set_lock_state(struct nfs4_state *state, struct file_lock *fl);
 extern void nfs4_copy_stateid(nfs4_stateid *, struct nfs4_state *, fl_owner_t);

fs/nfs/nfs4proc.c

@@ -62,14 +62,12 @@
 struct nfs4_opendata;
 static int _nfs4_proc_open(struct nfs4_opendata *data);
 static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *);
-static int nfs4_async_handle_error(struct rpc_task *, const struct nfs_server *);
-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_async_handle_error(struct rpc_task *, const struct nfs_server *, struct nfs4_state *);
 static int _nfs4_proc_lookup(struct inode *dir, const struct qstr *name, struct nfs_fh *fhandle, struct nfs_fattr *fattr);
 static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr);
 
 /* Prevent leaks of NFSv4 errors into userland */
-int nfs4_map_errors(int err)
+static int nfs4_map_errors(int err)
 {
 	if (err < -1000) {
 		dprintk("%s could not handle NFSv4 error %d\n",
@@ -195,6 +193,83 @@ static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dent
 	kunmap_atomic(start, KM_USER0);
 }
 
+static int nfs4_wait_bit_killable(void *word)
+{
+	if (fatal_signal_pending(current))
+		return -ERESTARTSYS;
+	schedule();
+	return 0;
+}
+
+static int nfs4_wait_clnt_recover(struct nfs_client *clp)
+{
+	int res;
+
+	might_sleep();
+
+	res = wait_on_bit(&clp->cl_state, NFS4CLNT_MANAGER_RUNNING,
+			nfs4_wait_bit_killable, TASK_KILLABLE);
+	return res;
+}
+
+static int nfs4_delay(struct rpc_clnt *clnt, long *timeout)
+{
+	int res = 0;
+
+	might_sleep();
+
+	if (*timeout <= 0)
+		*timeout = NFS4_POLL_RETRY_MIN;
+	if (*timeout > NFS4_POLL_RETRY_MAX)
+		*timeout = NFS4_POLL_RETRY_MAX;
+	schedule_timeout_killable(*timeout);
+	if (fatal_signal_pending(current))
+		res = -ERESTARTSYS;
+	*timeout <<= 1;
+	return res;
+}
+
+/* This is the error handling routine for processes that are allowed
+ * to sleep.
+ */
+static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
+{
+	struct nfs_client *clp = server->nfs_client;
+	struct nfs4_state *state = exception->state;
+	int ret = errorcode;
+
+	exception->retry = 0;
+	switch(errorcode) {
+		case 0:
+			return 0;
+		case -NFS4ERR_ADMIN_REVOKED:
+		case -NFS4ERR_BAD_STATEID:
+		case -NFS4ERR_OPENMODE:
+			if (state == NULL)
+				break;
+			nfs4_state_mark_reclaim_nograce(clp, state);
+		case -NFS4ERR_STALE_CLIENTID:
+		case -NFS4ERR_STALE_STATEID:
+		case -NFS4ERR_EXPIRED:
+			nfs4_schedule_state_recovery(clp);
+			ret = nfs4_wait_clnt_recover(clp);
+			if (ret == 0)
+				exception->retry = 1;
+			break;
+		case -NFS4ERR_FILE_OPEN:
+		case -NFS4ERR_GRACE:
+		case -NFS4ERR_DELAY:
+			ret = nfs4_delay(server->client, &exception->timeout);
+			if (ret != 0)
+				break;
+		case -NFS4ERR_OLD_STATEID:
+			exception->retry = 1;
+	}
+	/* We failed to handle the error */
+	return nfs4_map_errors(ret);
+}
+
+
 static void renew_lease(const struct nfs_server *server, unsigned long timestamp)
 {
 	struct nfs_client *clp = server->nfs_client;
@@ -248,7 +323,7 @@ static void nfs4_init_opendata_res(struct nfs4_opendata *p)
 }
 
 static struct nfs4_opendata *nfs4_opendata_alloc(struct path *path,
-		struct nfs4_state_owner *sp, int flags,
+		struct nfs4_state_owner *sp, fmode_t fmode, int flags,
 		const struct iattr *attrs)
 {
 	struct dentry *parent = dget_parent(path->dentry);
@@ -268,7 +343,8 @@ static struct nfs4_opendata *nfs4_opendata_alloc(struct path *path,
 	p->owner = sp;
 	atomic_inc(&sp->so_count);
 	p->o_arg.fh = NFS_FH(dir);
-	p->o_arg.open_flags = flags,
+	p->o_arg.open_flags = flags;
+	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
 	p->o_arg.clientid = server->nfs_client->cl_clientid;
 	p->o_arg.id = sp->so_owner_id.id;
 	p->o_arg.name = &p->path.dentry->d_name;
@@ -324,10 +400,13 @@ static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
 	return ret;
 }
 
-static int can_open_cached(struct nfs4_state *state, int mode)
+static int can_open_cached(struct nfs4_state *state, fmode_t mode, int open_mode)
 {
 	int ret = 0;
-	switch (mode & (FMODE_READ|FMODE_WRITE|O_EXCL)) {
+
+	if (open_mode & O_EXCL)
+		goto out;
+	switch (mode & (FMODE_READ|FMODE_WRITE)) {
 		case FMODE_READ:
 			ret |= test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0;
 			break;
@@ -337,21 +416,23 @@ static int can_open_cached(struct nfs4_state *state, int mode)
 		case FMODE_READ|FMODE_WRITE:
 			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0;
 	}
+out:
 	return ret;
 }
 
-static int can_open_delegated(struct nfs_delegation *delegation, mode_t open_flags)
+static int can_open_delegated(struct nfs_delegation *delegation, fmode_t fmode)
 {
-	if ((delegation->type & open_flags) != open_flags)
+	if ((delegation->type & fmode) != fmode)
 		return 0;
-	if (delegation->flags & NFS_DELEGATION_NEED_RECLAIM)
+	if (test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags))
 		return 0;
+	nfs_mark_delegation_referenced(delegation);
 	return 1;
 }
 
-static void update_open_stateflags(struct nfs4_state *state, mode_t open_flags)
+static void update_open_stateflags(struct nfs4_state *state, fmode_t fmode)
 {
-	switch (open_flags) {
+	switch (fmode) {
 		case FMODE_WRITE:
 			state->n_wronly++;
 			break;
@@ -361,15 +442,15 @@ static void update_open_stateflags(struct nfs4_state *state, mode_t open_flags)
 		case FMODE_READ|FMODE_WRITE:
 			state->n_rdwr++;
 	}
-	nfs4_state_set_mode_locked(state, state->state | open_flags);
+	nfs4_state_set_mode_locked(state, state->state | fmode);
 }
 
-static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, int open_flags)
+static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
 {
 	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) {
+	switch (fmode) {
 		case FMODE_READ:
 			set_bit(NFS_O_RDONLY_STATE, &state->flags);
 			break;
@@ -381,16 +462,15 @@ static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *
 	}
 }
 
-static void nfs_set_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, int open_flags)
+static void nfs_set_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
 {
 	write_seqlock(&state->seqlock);
-	nfs_set_open_stateid_locked(state, stateid, open_flags);
+	nfs_set_open_stateid_locked(state, stateid, fmode);
 	write_sequnlock(&state->seqlock);
 }
 
-static void update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *deleg_stateid, int open_flags)
+static void __update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, const nfs4_stateid *deleg_stateid, fmode_t fmode)
 {
-	open_flags &= (FMODE_READ|FMODE_WRITE);
 	/*
 	 * Protect the call to nfs4_state_set_mode_locked and
 	 * serialise the stateid update
@@ -401,20 +481,60 @@ static void update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_sta
 		set_bit(NFS_DELEGATED_STATE, &state->flags);
 	}
 	if (open_stateid != NULL)
-		nfs_set_open_stateid_locked(state, open_stateid, open_flags);
+		nfs_set_open_stateid_locked(state, open_stateid, fmode);
 	write_sequnlock(&state->seqlock);
 	spin_lock(&state->owner->so_lock);
-	update_open_stateflags(state, open_flags);
+	update_open_stateflags(state, fmode);
 	spin_unlock(&state->owner->so_lock);
 }
 
-static void nfs4_return_incompatible_delegation(struct inode *inode, mode_t open_flags)
+static int update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *delegation, fmode_t fmode)
+{
+	struct nfs_inode *nfsi = NFS_I(state->inode);
+	struct nfs_delegation *deleg_cur;
+	int ret = 0;
+
+	fmode &= (FMODE_READ|FMODE_WRITE);
+
+	rcu_read_lock();
+	deleg_cur = rcu_dereference(nfsi->delegation);
+	if (deleg_cur == NULL)
+		goto no_delegation;
+
+	spin_lock(&deleg_cur->lock);
+	if (nfsi->delegation != deleg_cur ||
+	    (deleg_cur->type & fmode) != fmode)
+		goto no_delegation_unlock;
+
+	if (delegation == NULL)
+		delegation = &deleg_cur->stateid;
+	else if (memcmp(deleg_cur->stateid.data, delegation->data, NFS4_STATEID_SIZE) != 0)
+		goto no_delegation_unlock;
+
+	nfs_mark_delegation_referenced(deleg_cur);
+	__update_open_stateid(state, open_stateid, &deleg_cur->stateid, fmode);
+	ret = 1;
+no_delegation_unlock:
+	spin_unlock(&deleg_cur->lock);
+no_delegation:
+	rcu_read_unlock();
+
+	if (!ret && open_stateid != NULL) {
+		__update_open_stateid(state, open_stateid, NULL, fmode);
+		ret = 1;
+	}
+
+	return ret;
+}
+
+
+static void nfs4_return_incompatible_delegation(struct inode *inode, fmode_t fmode)
 {
 	struct nfs_delegation *delegation;
 
 	rcu_read_lock();
 	delegation = rcu_dereference(NFS_I(inode)->delegation);
-	if (delegation == NULL || (delegation->type & open_flags) == open_flags) {
+	if (delegation == NULL || (delegation->type & fmode) == fmode) {
 		rcu_read_unlock();
 		return;
 	}
@@ -427,27 +547,28 @@ 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);
+	int open_mode = opendata->o_arg.open_flags & O_EXCL;
+	fmode_t fmode = opendata->o_arg.fmode;
 	nfs4_stateid stateid;
 	int ret = -EAGAIN;
 
-	rcu_read_lock();
-	delegation = rcu_dereference(nfsi->delegation);
 	for (;;) {
-		if (can_open_cached(state, open_mode)) {
+		if (can_open_cached(state, fmode, open_mode)) {
 			spin_lock(&state->owner->so_lock);
-			if (can_open_cached(state, open_mode)) {
-				update_open_stateflags(state, open_mode);
+			if (can_open_cached(state, fmode, open_mode)) {
+				update_open_stateflags(state, fmode);
 				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))
+		rcu_read_lock();
+		delegation = rcu_dereference(nfsi->delegation);
+		if (delegation == NULL ||
+		    !can_open_delegated(delegation, fmode)) {
+			rcu_read_unlock();
 			break;
+		}
 		/* Save the delegation */
 		memcpy(stateid.data, delegation->stateid.data, sizeof(stateid.data));
 		rcu_read_unlock();
@@ -455,19 +576,11 @@ static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
 		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;
+
+		/* Try to update the stateid using the delegation */
+		if (update_open_stateid(state, NULL, &stateid, fmode))
+			goto out_return_state;
 	}
-	rcu_read_unlock();
 out:
 	return ERR_PTR(ret);
 out_return_state:
@@ -480,7 +593,6 @@ 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->rpc_done) {
@@ -507,7 +619,7 @@ static struct nfs4_state *nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data
 		if (delegation)
 			delegation_flags = delegation->flags;
 		rcu_read_unlock();
-		if (!(delegation_flags & NFS_DELEGATION_NEED_RECLAIM))
+		if ((delegation_flags & 1UL<<NFS_DELEGATION_NEED_RECLAIM) == 0)
 			nfs_inode_set_delegation(state->inode,
 					data->owner->so_cred,
 					&data->o_res);
@@ -516,12 +628,9 @@ static struct nfs4_state *nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data
 					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();
+
+	update_open_stateid(state, &data->o_res.stateid, NULL,
+			data->o_arg.fmode);
 	iput(inode);
 out:
 	return state;
@@ -552,7 +661,7 @@ static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context
 {
 	struct nfs4_opendata *opendata;
 
-	opendata = nfs4_opendata_alloc(&ctx->path, state->owner, 0, NULL);
+	opendata = nfs4_opendata_alloc(&ctx->path, state->owner, 0, 0, NULL);
 	if (opendata == NULL)
 		return ERR_PTR(-ENOMEM);
 	opendata->state = state;
@@ -560,12 +669,13 @@ static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context
 	return opendata;
 }
 
-static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, mode_t openflags, struct nfs4_state **res)
+static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, fmode_t fmode, struct nfs4_state **res)
 {
 	struct nfs4_state *newstate;
 	int ret;
 
-	opendata->o_arg.open_flags = openflags;
+	opendata->o_arg.open_flags = 0;
+	opendata->o_arg.fmode = fmode;
 	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
 	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
 	nfs4_init_opendata_res(opendata);
@@ -575,7 +685,7 @@ static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, mode_t openf
 	newstate = nfs4_opendata_to_nfs4_state(opendata);
 	if (IS_ERR(newstate))
 		return PTR_ERR(newstate);
-	nfs4_close_state(&opendata->path, newstate, openflags);
+	nfs4_close_state(&opendata->path, newstate, fmode);
 	*res = newstate;
 	return 0;
 }
@@ -631,7 +741,7 @@ static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state
 {
 	struct nfs_delegation *delegation;
 	struct nfs4_opendata *opendata;
-	int delegation_type = 0;
+	fmode_t delegation_type = 0;
 	int status;
 
 	opendata = nfs4_open_recoverdata_alloc(ctx, state);
@@ -641,7 +751,7 @@ static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state
 	opendata->o_arg.fh = NFS_FH(state->inode);
 	rcu_read_lock();
 	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
-	if (delegation != NULL && (delegation->flags & NFS_DELEGATION_NEED_RECLAIM) != 0)
+	if (delegation != NULL && test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) != 0)
 		delegation_type = delegation->type;
 	rcu_read_unlock();
 	opendata->o_arg.u.delegation_type = delegation_type;
@@ -744,7 +854,7 @@ static void nfs4_open_confirm_release(void *calldata)
 		goto out_free;
 	state = nfs4_opendata_to_nfs4_state(data);
 	if (!IS_ERR(state))
-		nfs4_close_state(&data->path, state, data->o_arg.open_flags);
+		nfs4_close_state(&data->path, state, data->o_arg.fmode);
 out_free:
 	nfs4_opendata_put(data);
 }
@@ -808,12 +918,12 @@ static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
 	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)))
+		if (can_open_cached(data->state, data->o_arg.fmode, data->o_arg.open_flags))
 			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) {
+		    test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) == 0) {
 			rcu_read_unlock();
 			goto out_no_action;
 		}
@@ -877,7 +987,7 @@ static void nfs4_open_release(void *calldata)
 		goto out_free;
 	state = nfs4_opendata_to_nfs4_state(data);
 	if (!IS_ERR(state))
-		nfs4_close_state(&data->path, state, data->o_arg.open_flags);
+		nfs4_close_state(&data->path, state, data->o_arg.fmode);
 out_free:
 	nfs4_opendata_put(data);
 }
@@ -955,10 +1065,11 @@ static int nfs4_recover_expired_lease(struct nfs_server *server)
 	int ret;
 
 	for (;;) {
-		ret = nfs4_wait_clnt_recover(server->client, clp);
+		ret = nfs4_wait_clnt_recover(clp);
 		if (ret != 0)
 			return ret;
-		if (!test_and_clear_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state))
+		if (!test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) &&
+		    !test_bit(NFS4CLNT_CHECK_LEASE,&clp->cl_state))
 			break;
 		nfs4_schedule_state_recovery(clp);
 	}
@@ -993,8 +1104,9 @@ static inline int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4
 
 	do {
 		err = _nfs4_open_expired(ctx, state);
-		if (err == -NFS4ERR_DELAY)
-			nfs4_handle_exception(server, err, &exception);
+		if (err != -NFS4ERR_DELAY)
+			break;
+		nfs4_handle_exception(server, err, &exception);
 	} while (exception.retry);
 	return err;
 }
@@ -1031,12 +1143,11 @@ static inline void nfs4_exclusive_attrset(struct nfs4_opendata *opendata, struct
 /*
  * Returns a referenced nfs4_state
  */
-static int _nfs4_do_open(struct inode *dir, struct path *path, int flags, struct iattr *sattr, struct rpc_cred *cred, struct nfs4_state **res)
+static int _nfs4_do_open(struct inode *dir, struct path *path, fmode_t fmode, 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;
 
@@ -1050,12 +1161,11 @@ static int _nfs4_do_open(struct inode *dir, struct path *path, int flags, struct
 	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);
+		nfs4_return_incompatible_delegation(path->dentry->d_inode, fmode);
 	status = -ENOMEM;
-	opendata = nfs4_opendata_alloc(path, sp, flags, sattr);
+	opendata = nfs4_opendata_alloc(path, sp, fmode, flags, sattr);
 	if (opendata == NULL)
-		goto err_release_rwsem;
+		goto err_put_state_owner;
 
 	if (path->dentry->d_inode != NULL)
 		opendata->state = nfs4_get_open_state(path->dentry->d_inode, sp);
@@ -1073,13 +1183,10 @@ static int _nfs4_do_open(struct inode *dir, struct path *path, int flags, struct
 		goto err_opendata_put;
 	nfs4_opendata_put(opendata);
 	nfs4_put_state_owner(sp);
-	up_read(&clp->cl_sem);
 	*res = state;
 	return 0;
 err_opendata_put:
 	nfs4_opendata_put(opendata);
-err_release_rwsem:
-	up_read(&clp->cl_sem);
 err_put_state_owner:
 	nfs4_put_state_owner(sp);
 out_err:
@@ -1088,14 +1195,14 @@ out_err:
 }
 
 
-static struct nfs4_state *nfs4_do_open(struct inode *dir, struct path *path, int flags, struct iattr *sattr, struct rpc_cred *cred)
+static struct nfs4_state *nfs4_do_open(struct inode *dir, struct path *path, fmode_t fmode, 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, path, flags, sattr, cred, &res);
+		status = _nfs4_do_open(dir, path, fmode, flags, sattr, cred, &res);
 		if (status == 0)
 			break;
 		/* NOTE: BAD_SEQID means the server and client disagree about the
@@ -1230,10 +1337,13 @@ static void nfs4_close_done(struct rpc_task *task, void *data)
 			renew_lease(server, calldata->timestamp);
 			break;
 		case -NFS4ERR_STALE_STATEID:
+		case -NFS4ERR_OLD_STATEID:
+		case -NFS4ERR_BAD_STATEID:
 		case -NFS4ERR_EXPIRED:
-			break;
+			if (calldata->arg.fmode == 0)
+				break;
 		default:
-			if (nfs4_async_handle_error(task, server) == -EAGAIN) {
+			if (nfs4_async_handle_error(task, server, state) == -EAGAIN) {
 				rpc_restart_call(task);
 				return;
 			}
@@ -1272,10 +1382,10 @@ static void nfs4_close_prepare(struct rpc_task *task, void *data)
 	nfs_fattr_init(calldata->res.fattr);
 	if (test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0) {
 		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
-		calldata->arg.open_flags = FMODE_READ;
+		calldata->arg.fmode = FMODE_READ;
 	} else if (test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0) {
 		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
-		calldata->arg.open_flags = FMODE_WRITE;
+		calldata->arg.fmode = FMODE_WRITE;
 	}
 	calldata->timestamp = jiffies;
 	rpc_call_start(task);
@@ -1328,6 +1438,7 @@ int nfs4_do_close(struct path *path, struct nfs4_state *state, int wait)
 	calldata->arg.seqid = nfs_alloc_seqid(&state->owner->so_seqid);
 	if (calldata->arg.seqid == NULL)
 		goto out_free_calldata;
+	calldata->arg.fmode = 0;
 	calldata->arg.bitmask = server->attr_bitmask;
 	calldata->res.fattr = &calldata->fattr;
 	calldata->res.seqid = calldata->arg.seqid;
@@ -1354,13 +1465,13 @@ out:
 	return status;
 }
 
-static int nfs4_intent_set_file(struct nameidata *nd, struct path *path, struct nfs4_state *state)
+static int nfs4_intent_set_file(struct nameidata *nd, struct path *path, struct nfs4_state *state, fmode_t fmode)
 {
 	struct file *filp;
 	int ret;
 
 	/* If the open_intent is for execute, we have an extra check to make */
-	if (nd->intent.open.flags & FMODE_EXEC) {
+	if (fmode & FMODE_EXEC) {
 		ret = nfs_may_open(state->inode,
 				state->owner->so_cred,
 				nd->intent.open.flags);
@@ -1376,7 +1487,7 @@ static int nfs4_intent_set_file(struct nameidata *nd, struct path *path, struct
 	}
 	ret = PTR_ERR(filp);
 out_close:
-	nfs4_close_sync(path, state, nd->intent.open.flags);
+	nfs4_close_sync(path, state, fmode & (FMODE_READ|FMODE_WRITE));
 	return ret;
 }
 
@@ -1392,6 +1503,7 @@ nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
 	struct rpc_cred *cred;
 	struct nfs4_state *state;
 	struct dentry *res;
+	fmode_t fmode = nd->intent.open.flags & (FMODE_READ | FMODE_WRITE | FMODE_EXEC);
 
 	if (nd->flags & LOOKUP_CREATE) {
 		attr.ia_mode = nd->intent.open.create_mode;
@@ -1409,7 +1521,7 @@ nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
 	parent = dentry->d_parent;
 	/* Protect against concurrent sillydeletes */
 	nfs_block_sillyrename(parent);
-	state = nfs4_do_open(dir, &path, nd->intent.open.flags, &attr, cred);
+	state = nfs4_do_open(dir, &path, fmode, nd->intent.open.flags, &attr, cred);
 	put_rpccred(cred);
 	if (IS_ERR(state)) {
 		if (PTR_ERR(state) == -ENOENT) {
@@ -1424,7 +1536,7 @@ nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
 		path.dentry = res;
 	nfs_set_verifier(path.dentry, nfs_save_change_attribute(dir));
 	nfs_unblock_sillyrename(parent);
-	nfs4_intent_set_file(nd, &path, state);
+	nfs4_intent_set_file(nd, &path, state, fmode);
 	return res;
 }
 
@@ -1437,11 +1549,12 @@ nfs4_open_revalidate(struct inode *dir, struct dentry *dentry, int openflags, st
 	};
 	struct rpc_cred *cred;
 	struct nfs4_state *state;
+	fmode_t fmode = openflags & (FMODE_READ | FMODE_WRITE);
 
 	cred = rpc_lookup_cred();
 	if (IS_ERR(cred))
 		return PTR_ERR(cred);
-	state = nfs4_do_open(dir, &path, openflags, NULL, cred);
+	state = nfs4_do_open(dir, &path, fmode, openflags, NULL, cred);
 	put_rpccred(cred);
 	if (IS_ERR(state)) {
 		switch (PTR_ERR(state)) {
@@ -1458,10 +1571,10 @@ nfs4_open_revalidate(struct inode *dir, struct dentry *dentry, int openflags, st
 	}
 	if (state->inode == dentry->d_inode) {
 		nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
-		nfs4_intent_set_file(nd, &path, state);
+		nfs4_intent_set_file(nd, &path, state, fmode);
 		return 1;
 	}
-	nfs4_close_sync(&path, state, openflags);
+	nfs4_close_sync(&path, state, fmode);
 out_drop:
 	d_drop(dentry);
 	return 0;
@@ -1887,6 +2000,7 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 	};
 	struct nfs4_state *state;
 	struct rpc_cred *cred;
+	fmode_t fmode = flags & (FMODE_READ | FMODE_WRITE);
 	int status = 0;
 
 	cred = rpc_lookup_cred();
@@ -1894,7 +2008,7 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 		status = PTR_ERR(cred);
 		goto out;
 	}
-	state = nfs4_do_open(dir, &path, flags, sattr, cred);
+	state = nfs4_do_open(dir, &path, fmode, flags, sattr, cred);
 	d_drop(dentry);
 	if (IS_ERR(state)) {
 		status = PTR_ERR(state);
@@ -1910,9 +2024,9 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 		nfs_post_op_update_inode(state->inode, &fattr);
 	}
 	if (status == 0 && (nd->flags & LOOKUP_OPEN) != 0)
-		status = nfs4_intent_set_file(nd, &path, state);
+		status = nfs4_intent_set_file(nd, &path, state, fmode);
 	else
-		nfs4_close_sync(&path, state, flags);
+		nfs4_close_sync(&path, state, fmode);
 out_putcred:
 	put_rpccred(cred);
 out:
@@ -1974,7 +2088,7 @@ static int nfs4_proc_unlink_done(struct rpc_task *task, struct inode *dir)
 {
 	struct nfs_removeres *res = task->tk_msg.rpc_resp;
 
-	if (nfs4_async_handle_error(task, res->server) == -EAGAIN)
+	if (nfs4_async_handle_error(task, res->server, NULL) == -EAGAIN)
 		return 0;
 	update_changeattr(dir, &res->cinfo);
 	nfs_post_op_update_inode(dir, &res->dir_attr);
@@ -2402,7 +2516,7 @@ static int nfs4_read_done(struct rpc_task *task, struct nfs_read_data *data)
 {
 	struct nfs_server *server = NFS_SERVER(data->inode);
 
-	if (nfs4_async_handle_error(task, server) == -EAGAIN) {
+	if (nfs4_async_handle_error(task, server, data->args.context->state) == -EAGAIN) {
 		rpc_restart_call(task);
 		return -EAGAIN;
 	}
@@ -2423,7 +2537,7 @@ static int nfs4_write_done(struct rpc_task *task, struct nfs_write_data *data)
 {
 	struct inode *inode = data->inode;
 	
-	if (nfs4_async_handle_error(task, NFS_SERVER(inode)) == -EAGAIN) {
+	if (nfs4_async_handle_error(task, NFS_SERVER(inode), data->args.context->state) == -EAGAIN) {
 		rpc_restart_call(task);
 		return -EAGAIN;
 	}
@@ -2449,7 +2563,7 @@ static int nfs4_commit_done(struct rpc_task *task, struct nfs_write_data *data)
 {
 	struct inode *inode = data->inode;
 	
-	if (nfs4_async_handle_error(task, NFS_SERVER(inode)) == -EAGAIN) {
+	if (nfs4_async_handle_error(task, NFS_SERVER(inode), NULL) == -EAGAIN) {
 		rpc_restart_call(task);
 		return -EAGAIN;
 	}
@@ -2742,19 +2856,25 @@ static int nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen
 }
 
 static int
-nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server)
+nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server, struct nfs4_state *state)
 {
 	struct nfs_client *clp = server->nfs_client;
 
 	if (!clp || task->tk_status >= 0)
 		return 0;
 	switch(task->tk_status) {
+		case -NFS4ERR_ADMIN_REVOKED:
+		case -NFS4ERR_BAD_STATEID:
+		case -NFS4ERR_OPENMODE:
+			if (state == NULL)
+				break;
+			nfs4_state_mark_reclaim_nograce(clp, state);
 		case -NFS4ERR_STALE_CLIENTID:
 		case -NFS4ERR_STALE_STATEID:
 		case -NFS4ERR_EXPIRED:
 			rpc_sleep_on(&clp->cl_rpcwaitq, task, NULL);
 			nfs4_schedule_state_recovery(clp);
-			if (test_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state) == 0)
+			if (test_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) == 0)
 				rpc_wake_up_queued_task(&clp->cl_rpcwaitq, task);
 			task->tk_status = 0;
 			return -EAGAIN;
@@ -2772,79 +2892,6 @@ nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server)
 	return 0;
 }
 
-static int nfs4_wait_bit_killable(void *word)
-{
-	if (fatal_signal_pending(current))
-		return -ERESTARTSYS;
-	schedule();
-	return 0;
-}
-
-static int nfs4_wait_clnt_recover(struct rpc_clnt *clnt, struct nfs_client *clp)
-{
-	int res;
-
-	might_sleep();
-
-	rwsem_acquire(&clp->cl_sem.dep_map, 0, 0, _RET_IP_);
-
-	res = wait_on_bit(&clp->cl_state, NFS4CLNT_STATE_RECOVER,
-			nfs4_wait_bit_killable, TASK_KILLABLE);
-
-	rwsem_release(&clp->cl_sem.dep_map, 1, _RET_IP_);
-	return res;
-}
-
-static int nfs4_delay(struct rpc_clnt *clnt, long *timeout)
-{
-	int res = 0;
-
-	might_sleep();
-
-	if (*timeout <= 0)
-		*timeout = NFS4_POLL_RETRY_MIN;
-	if (*timeout > NFS4_POLL_RETRY_MAX)
-		*timeout = NFS4_POLL_RETRY_MAX;
-	schedule_timeout_killable(*timeout);
-	if (fatal_signal_pending(current))
-		res = -ERESTARTSYS;
-	*timeout <<= 1;
-	return res;
-}
-
-/* This is the error handling routine for processes that are allowed
- * to sleep.
- */
-static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
-{
-	struct nfs_client *clp = server->nfs_client;
-	int ret = errorcode;
-
-	exception->retry = 0;
-	switch(errorcode) {
-		case 0:
-			return 0;
-		case -NFS4ERR_STALE_CLIENTID:
-		case -NFS4ERR_STALE_STATEID:
-		case -NFS4ERR_EXPIRED:
-			nfs4_schedule_state_recovery(clp);
-			ret = nfs4_wait_clnt_recover(server->client, clp);
-			if (ret == 0)
-				exception->retry = 1;
-			break;
-		case -NFS4ERR_FILE_OPEN:
-		case -NFS4ERR_GRACE:
-		case -NFS4ERR_DELAY:
-			ret = nfs4_delay(server->client, &exception->timeout);
-			if (ret != 0)
-				break;
-		case -NFS4ERR_OLD_STATEID:
-			exception->retry = 1;
-	}
-	/* We failed to handle the error */
-	return nfs4_map_errors(ret);
-}
-
 int nfs4_proc_setclientid(struct nfs_client *clp, u32 program, unsigned short port, struct rpc_cred *cred)
 {
 	nfs4_verifier sc_verifier;
@@ -2916,7 +2963,6 @@ static int _nfs4_proc_setclientid_confirm(struct nfs_client *clp, struct rpc_cre
 		spin_lock(&clp->cl_lock);
 		clp->cl_lease_time = fsinfo.lease_time * HZ;
 		clp->cl_last_renewal = now;
-		clear_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
 		spin_unlock(&clp->cl_lock);
 	}
 	return status;
@@ -3074,7 +3120,6 @@ static int _nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock
 	struct nfs4_lock_state *lsp;
 	int status;
 
-	down_read(&clp->cl_sem);
 	arg.lock_owner.clientid = clp->cl_clientid;
 	status = nfs4_set_lock_state(state, request);
 	if (status != 0)
@@ -3091,7 +3136,6 @@ static int _nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock
 	}
 	request->fl_ops->fl_release_private(request);
 out:
-	up_read(&clp->cl_sem);
 	return status;
 }
 
@@ -3181,11 +3225,13 @@ static void nfs4_locku_done(struct rpc_task *task, void *data)
 					sizeof(calldata->lsp->ls_stateid.data));
 			renew_lease(calldata->server, calldata->timestamp);
 			break;
+		case -NFS4ERR_BAD_STATEID:
+		case -NFS4ERR_OLD_STATEID:
 		case -NFS4ERR_STALE_STATEID:
 		case -NFS4ERR_EXPIRED:
 			break;
 		default:
-			if (nfs4_async_handle_error(task, calldata->server) == -EAGAIN)
+			if (nfs4_async_handle_error(task, calldata->server, NULL) == -EAGAIN)
 				rpc_restart_call(task);
 	}
 }
@@ -3248,6 +3294,7 @@ static struct rpc_task *nfs4_do_unlck(struct file_lock *fl,
 
 static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
 {
+	struct nfs_inode *nfsi = NFS_I(state->inode);
 	struct nfs_seqid *seqid;
 	struct nfs4_lock_state *lsp;
 	struct rpc_task *task;
@@ -3257,8 +3304,12 @@ static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *
 	status = nfs4_set_lock_state(state, request);
 	/* Unlock _before_ we do the RPC call */
 	request->fl_flags |= FL_EXISTS;
-	if (do_vfs_lock(request->fl_file, request) == -ENOENT)
+	down_read(&nfsi->rwsem);
+	if (do_vfs_lock(request->fl_file, request) == -ENOENT) {
+		up_read(&nfsi->rwsem);
 		goto out;
+	}
+	up_read(&nfsi->rwsem);
 	if (status != 0)
 		goto out;
 	/* Is this a delegated lock? */
@@ -3484,7 +3535,7 @@ static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request
 
 static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 {
-	struct nfs_client *clp = state->owner->so_client;
+	struct nfs_inode *nfsi = NFS_I(state->inode);
 	unsigned char fl_flags = request->fl_flags;
 	int status;
 
@@ -3496,19 +3547,13 @@ static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock
 	status = do_vfs_lock(request->fl_file, request);
 	if (status < 0)
 		goto out;
-	down_read(&clp->cl_sem);
+	down_read(&nfsi->rwsem);
 	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
-		struct nfs_inode *nfsi = NFS_I(state->inode);
 		/* Yes: cache locks! */
-		down_read(&nfsi->rwsem);
 		/* ...but avoid races with delegation recall... */
-		if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
-			request->fl_flags = fl_flags & ~FL_SLEEP;
-			status = do_vfs_lock(request->fl_file, request);
-			up_read(&nfsi->rwsem);
-			goto out_unlock;
-		}
-		up_read(&nfsi->rwsem);
+		request->fl_flags = fl_flags & ~FL_SLEEP;
+		status = do_vfs_lock(request->fl_file, request);
+		goto out_unlock;
 	}
 	status = _nfs4_do_setlk(state, cmd, request, 0);
 	if (status != 0)
@@ -3518,7 +3563,7 @@ static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock
 	if (do_vfs_lock(request->fl_file, request) < 0)
 		printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n", __func__);
 out_unlock:
-	up_read(&clp->cl_sem);
+	up_read(&nfsi->rwsem);
 out:
 	request->fl_flags = fl_flags;
 	return status;
@@ -3664,11 +3709,15 @@ int nfs4_proc_fs_locations(struct inode *dir, const struct qstr *name,
 }
 
 struct nfs4_state_recovery_ops nfs4_reboot_recovery_ops = {
+	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
+	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
 	.recover_open	= nfs4_open_reclaim,
 	.recover_lock	= nfs4_lock_reclaim,
 };
 
-struct nfs4_state_recovery_ops nfs4_network_partition_recovery_ops = {
+struct nfs4_state_recovery_ops nfs4_nograce_recovery_ops = {
+	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
+	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
 	.recover_open	= nfs4_open_expired,
 	.recover_lock	= nfs4_lock_expired,
 };

fs/nfs/nfs4renewd.c

@@ -65,7 +65,6 @@ nfs4_renew_state(struct work_struct *work)
 	long lease, timeout;
 	unsigned long last, now;
 
-	down_read(&clp->cl_sem);
 	dprintk("%s: start\n", __func__);
 	/* Are there any active superblocks? */
 	if (list_empty(&clp->cl_superblocks))
@@ -77,17 +76,19 @@ nfs4_renew_state(struct work_struct *work)
 	timeout = (2 * lease) / 3 + (long)last - (long)now;
 	/* Are we close to a lease timeout? */
 	if (time_after(now, last + lease/3)) {
-		cred = nfs4_get_renew_cred(clp);
+		cred = nfs4_get_renew_cred_locked(clp);
+		spin_unlock(&clp->cl_lock);
 		if (cred == NULL) {
-			set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
-			spin_unlock(&clp->cl_lock);
+			if (list_empty(&clp->cl_delegations)) {
+				set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
+				goto out;
+			}
 			nfs_expire_all_delegations(clp);
-			goto out;
+		} else {
+			/* Queue an asynchronous RENEW. */
+			nfs4_proc_async_renew(clp, cred);
+			put_rpccred(cred);
 		}
-		spin_unlock(&clp->cl_lock);
-		/* Queue an asynchronous RENEW. */
-		nfs4_proc_async_renew(clp, cred);
-		put_rpccred(cred);
 		timeout = (2 * lease) / 3;
 		spin_lock(&clp->cl_lock);
 	} else
@@ -100,12 +101,11 @@ nfs4_renew_state(struct work_struct *work)
 	cancel_delayed_work(&clp->cl_renewd);
 	schedule_delayed_work(&clp->cl_renewd, timeout);
 	spin_unlock(&clp->cl_lock);
+	nfs_expire_unreferenced_delegations(clp);
 out:
-	up_read(&clp->cl_sem);
 	dprintk("%s: done\n", __func__);
 }
 
-/* Must be called with clp->cl_sem locked for writes */
 void
 nfs4_schedule_state_renewal(struct nfs_client *clp)
 {

fs/nfs/nfs4state.c

@@ -71,14 +71,12 @@ static int nfs4_init_client(struct nfs_client *clp, struct rpc_cred *cred)
 	return status;
 }
 
-static struct rpc_cred *nfs4_get_machine_cred(struct nfs_client *clp)
+static struct rpc_cred *nfs4_get_machine_cred_locked(struct nfs_client *clp)
 {
 	struct rpc_cred *cred = NULL;
 
-	spin_lock(&clp->cl_lock);
 	if (clp->cl_machine_cred != NULL)
 		cred = get_rpccred(clp->cl_machine_cred);
-	spin_unlock(&clp->cl_lock);
 	return cred;
 }
 
@@ -94,7 +92,7 @@ static void nfs4_clear_machine_cred(struct nfs_client *clp)
 		put_rpccred(cred);
 }
 
-struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp)
+struct rpc_cred *nfs4_get_renew_cred_locked(struct nfs_client *clp)
 {
 	struct nfs4_state_owner *sp;
 	struct rb_node *pos;
@@ -110,13 +108,24 @@ struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp)
 	return cred;
 }
 
+static struct rpc_cred *nfs4_get_renew_cred(struct nfs_client *clp)
+{
+	struct rpc_cred *cred;
+
+	spin_lock(&clp->cl_lock);
+	cred = nfs4_get_renew_cred_locked(clp);
+	spin_unlock(&clp->cl_lock);
+	return cred;
+}
+
 static struct rpc_cred *nfs4_get_setclientid_cred(struct nfs_client *clp)
 {
 	struct nfs4_state_owner *sp;
 	struct rb_node *pos;
 	struct rpc_cred *cred;
 
-	cred = nfs4_get_machine_cred(clp);
+	spin_lock(&clp->cl_lock);
+	cred = nfs4_get_machine_cred_locked(clp);
 	if (cred != NULL)
 		goto out;
 	pos = rb_first(&clp->cl_state_owners);
@@ -125,6 +134,7 @@ static struct rpc_cred *nfs4_get_setclientid_cred(struct nfs_client *clp)
 		cred = get_rpccred(sp->so_cred);
 	}
 out:
+	spin_unlock(&clp->cl_lock);
 	return cred;
 }
 
@@ -295,10 +305,6 @@ nfs4_drop_state_owner(struct nfs4_state_owner *sp)
 	}
 }
 
-/*
- * Note: must be called with clp->cl_sem held in order to prevent races
- *       with reboot recovery!
- */
 struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *server, struct rpc_cred *cred)
 {
 	struct nfs_client *clp = server->nfs_client;
@@ -327,10 +333,6 @@ struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *server, struct
 	return sp;
 }
 
-/*
- * Must be called with clp->cl_sem held in order to avoid races
- * with state recovery...
- */
 void nfs4_put_state_owner(struct nfs4_state_owner *sp)
 {
 	struct nfs_client *clp = sp->so_client;
@@ -361,18 +363,18 @@ nfs4_alloc_open_state(void)
 }
 
 void
-nfs4_state_set_mode_locked(struct nfs4_state *state, mode_t mode)
+nfs4_state_set_mode_locked(struct nfs4_state *state, fmode_t fmode)
 {
-	if (state->state == mode)
+	if (state->state == fmode)
 		return;
 	/* NB! List reordering - see the reclaim code for why.  */
-	if ((mode & FMODE_WRITE) != (state->state & FMODE_WRITE)) {
-		if (mode & FMODE_WRITE)
+	if ((fmode & FMODE_WRITE) != (state->state & FMODE_WRITE)) {
+		if (fmode & FMODE_WRITE)
 			list_move(&state->open_states, &state->owner->so_states);
 		else
 			list_move_tail(&state->open_states, &state->owner->so_states);
 	}
-	state->state = mode;
+	state->state = fmode;
 }
 
 static struct nfs4_state *
@@ -432,10 +434,6 @@ out:
 	return state;
 }
 
-/*
- * Beware! Caller must be holding exactly one
- * reference to clp->cl_sem!
- */
 void nfs4_put_open_state(struct nfs4_state *state)
 {
 	struct inode *inode = state->inode;
@@ -456,16 +454,16 @@ void nfs4_put_open_state(struct nfs4_state *state)
 /*
  * Close the current file.
  */
-static void __nfs4_close(struct path *path, struct nfs4_state *state, mode_t mode, int wait)
+static void __nfs4_close(struct path *path, struct nfs4_state *state, fmode_t fmode, int wait)
 {
 	struct nfs4_state_owner *owner = state->owner;
 	int call_close = 0;
-	int newstate;
+	fmode_t newstate;
 
 	atomic_inc(&owner->so_count);
 	/* Protect against nfs4_find_state() */
 	spin_lock(&owner->so_lock);
-	switch (mode & (FMODE_READ | FMODE_WRITE)) {
+	switch (fmode & (FMODE_READ | FMODE_WRITE)) {
 		case FMODE_READ:
 			state->n_rdonly--;
 			break;
@@ -500,14 +498,14 @@ static void __nfs4_close(struct path *path, struct nfs4_state *state, mode_t mod
 		nfs4_do_close(path, state, wait);
 }
 
-void nfs4_close_state(struct path *path, struct nfs4_state *state, mode_t mode)
+void nfs4_close_state(struct path *path, struct nfs4_state *state, fmode_t fmode)
 {
-	__nfs4_close(path, state, mode, 0);
+	__nfs4_close(path, state, fmode, 0);
 }
 
-void nfs4_close_sync(struct path *path, struct nfs4_state *state, mode_t mode)
+void nfs4_close_sync(struct path *path, struct nfs4_state *state, fmode_t fmode)
 {
-	__nfs4_close(path, state, mode, 1);
+	__nfs4_close(path, state, fmode, 1);
 }
 
 /*
@@ -568,7 +566,6 @@ static void nfs4_free_lock_state(struct nfs4_lock_state *lsp)
  * Return a compatible lock_state. If no initialized lock_state structure
  * exists, return an uninitialized one.
  *
- * The caller must be holding clp->cl_sem
  */
 static struct nfs4_lock_state *nfs4_get_lock_state(struct nfs4_state *state, fl_owner_t owner)
 {
@@ -770,32 +767,34 @@ unlock:
 	return status;
 }
 
-static int reclaimer(void *);
+static int nfs4_run_state_manager(void *);
 
-static inline void nfs4_clear_recover_bit(struct nfs_client *clp)
+static void nfs4_clear_state_manager_bit(struct nfs_client *clp)
 {
 	smp_mb__before_clear_bit();
-	clear_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state);
+	clear_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state);
 	smp_mb__after_clear_bit();
-	wake_up_bit(&clp->cl_state, NFS4CLNT_STATE_RECOVER);
+	wake_up_bit(&clp->cl_state, NFS4CLNT_MANAGER_RUNNING);
 	rpc_wake_up(&clp->cl_rpcwaitq);
 }
 
 /*
- * State recovery routine
+ * Schedule the nfs_client asynchronous state management routine
  */
-static void nfs4_recover_state(struct nfs_client *clp)
+void nfs4_schedule_state_manager(struct nfs_client *clp)
 {
 	struct task_struct *task;
 
+	if (test_and_set_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) != 0)
+		return;
 	__module_get(THIS_MODULE);
 	atomic_inc(&clp->cl_count);
-	task = kthread_run(reclaimer, clp, "%s-reclaim",
+	task = kthread_run(nfs4_run_state_manager, clp, "%s-manager",
 				rpc_peeraddr2str(clp->cl_rpcclient,
 							RPC_DISPLAY_ADDR));
 	if (!IS_ERR(task))
 		return;
-	nfs4_clear_recover_bit(clp);
+	nfs4_clear_state_manager_bit(clp);
 	nfs_put_client(clp);
 	module_put(THIS_MODULE);
 }
@@ -807,16 +806,42 @@ void nfs4_schedule_state_recovery(struct nfs_client *clp)
 {
 	if (!clp)
 		return;
-	if (test_and_set_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state) == 0)
-		nfs4_recover_state(clp);
+	if (!test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state))
+		set_bit(NFS4CLNT_CHECK_LEASE, &clp->cl_state);
+	nfs4_schedule_state_manager(clp);
 }
 
-static int nfs4_reclaim_locks(struct nfs4_state_recovery_ops *ops, struct nfs4_state *state)
+static int nfs4_state_mark_reclaim_reboot(struct nfs_client *clp, struct nfs4_state *state)
+{
+
+	set_bit(NFS_STATE_RECLAIM_REBOOT, &state->flags);
+	/* Don't recover state that expired before the reboot */
+	if (test_bit(NFS_STATE_RECLAIM_NOGRACE, &state->flags)) {
+		clear_bit(NFS_STATE_RECLAIM_REBOOT, &state->flags);
+		return 0;
+	}
+	set_bit(NFS_OWNER_RECLAIM_REBOOT, &state->owner->so_flags);
+	set_bit(NFS4CLNT_RECLAIM_REBOOT, &clp->cl_state);
+	return 1;
+}
+
+int nfs4_state_mark_reclaim_nograce(struct nfs_client *clp, struct nfs4_state *state)
+{
+	set_bit(NFS_STATE_RECLAIM_NOGRACE, &state->flags);
+	clear_bit(NFS_STATE_RECLAIM_REBOOT, &state->flags);
+	set_bit(NFS_OWNER_RECLAIM_NOGRACE, &state->owner->so_flags);
+	set_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state);
+	return 1;
+}
+
+static int nfs4_reclaim_locks(struct nfs4_state *state, const struct nfs4_state_recovery_ops *ops)
 {
 	struct inode *inode = state->inode;
+	struct nfs_inode *nfsi = NFS_I(inode);
 	struct file_lock *fl;
 	int status = 0;
 
+	down_write(&nfsi->rwsem);
 	for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) {
 		if (!(fl->fl_flags & (FL_POSIX|FL_FLOCK)))
 			continue;
@@ -839,12 +864,14 @@ static int nfs4_reclaim_locks(struct nfs4_state_recovery_ops *ops, struct nfs4_s
 				goto out_err;
 		}
 	}
+	up_write(&nfsi->rwsem);
 	return 0;
 out_err:
+	up_write(&nfsi->rwsem);
 	return status;
 }
 
-static int nfs4_reclaim_open_state(struct nfs4_state_recovery_ops *ops, struct nfs4_state_owner *sp)
+static int nfs4_reclaim_open_state(struct nfs4_state_owner *sp, const struct nfs4_state_recovery_ops *ops)
 {
 	struct nfs4_state *state;
 	struct nfs4_lock_state *lock;
@@ -858,28 +885,34 @@ static int nfs4_reclaim_open_state(struct nfs4_state_recovery_ops *ops, struct n
 	 * recovering after a network partition or a reboot from a
 	 * server that doesn't support a grace period.
 	 */
+restart:
+	spin_lock(&sp->so_lock);
 	list_for_each_entry(state, &sp->so_states, open_states) {
+		if (!test_and_clear_bit(ops->state_flag_bit, &state->flags))
+			continue;
 		if (state->state == 0)
 			continue;
+		atomic_inc(&state->count);
+		spin_unlock(&sp->so_lock);
 		status = ops->recover_open(sp, state);
 		if (status >= 0) {
-			status = nfs4_reclaim_locks(ops, state);
-			if (status < 0)
-				goto out_err;
-			list_for_each_entry(lock, &state->lock_states, ls_locks) {
-				if (!(lock->ls_flags & NFS_LOCK_INITIALIZED))
-					printk("%s: Lock reclaim failed!\n",
+			status = nfs4_reclaim_locks(state, ops);
+			if (status >= 0) {
+				list_for_each_entry(lock, &state->lock_states, ls_locks) {
+					if (!(lock->ls_flags & NFS_LOCK_INITIALIZED))
+						printk("%s: Lock reclaim failed!\n",
 							__func__);
+				}
+				nfs4_put_open_state(state);
+				goto restart;
 			}
-			continue;
 		}
 		switch (status) {
 			default:
 				printk(KERN_ERR "%s: unhandled error %d. Zeroing state\n",
 						__func__, status);
 			case -ENOENT:
-			case -NFS4ERR_RECLAIM_BAD:
-			case -NFS4ERR_RECLAIM_CONFLICT:
+			case -ESTALE:
 				/*
 				 * Open state on this file cannot be recovered
 				 * All we can do is revert to using the zero stateid.
@@ -889,84 +922,176 @@ static int nfs4_reclaim_open_state(struct nfs4_state_recovery_ops *ops, struct n
 				/* Mark the file as being 'closed' */
 				state->state = 0;
 				break;
+			case -NFS4ERR_RECLAIM_BAD:
+			case -NFS4ERR_RECLAIM_CONFLICT:
+				nfs4_state_mark_reclaim_nograce(sp->so_client, state);
+				break;
 			case -NFS4ERR_EXPIRED:
 			case -NFS4ERR_NO_GRACE:
+				nfs4_state_mark_reclaim_nograce(sp->so_client, state);
 			case -NFS4ERR_STALE_CLIENTID:
 				goto out_err;
 		}
+		nfs4_put_open_state(state);
+		goto restart;
 	}
+	spin_unlock(&sp->so_lock);
 	return 0;
 out_err:
+	nfs4_put_open_state(state);
 	return status;
 }
 
-static void nfs4_state_mark_reclaim(struct nfs_client *clp)
+static void nfs4_clear_open_state(struct nfs4_state *state)
+{
+	struct nfs4_lock_state *lock;
+
+	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.flags = 0;
+		lock->ls_flags &= ~NFS_LOCK_INITIALIZED;
+	}
+}
+
+static void nfs4_state_mark_reclaim_helper(struct nfs_client *clp, int (*mark_reclaim)(struct nfs_client *clp, struct nfs4_state *state))
 {
 	struct nfs4_state_owner *sp;
 	struct rb_node *pos;
 	struct nfs4_state *state;
-	struct nfs4_lock_state *lock;
 
 	/* Reset all sequence ids to zero */
 	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;
-				lock->ls_flags &= ~NFS_LOCK_INITIALIZED;
-			}
+			if (mark_reclaim(clp, state))
+				nfs4_clear_open_state(state);
 		}
 		spin_unlock(&sp->so_lock);
 	}
 }
 
-static int reclaimer(void *ptr)
+static void nfs4_state_start_reclaim_reboot(struct nfs_client *clp)
+{
+	/* Mark all delegations for reclaim */
+	nfs_delegation_mark_reclaim(clp);
+	nfs4_state_mark_reclaim_helper(clp, nfs4_state_mark_reclaim_reboot);
+}
+
+static void nfs4_state_end_reclaim_reboot(struct nfs_client *clp)
 {
-	struct nfs_client *clp = ptr;
 	struct nfs4_state_owner *sp;
 	struct rb_node *pos;
-	struct nfs4_state_recovery_ops *ops;
-	struct rpc_cred *cred;
+	struct nfs4_state *state;
+
+	if (!test_and_clear_bit(NFS4CLNT_RECLAIM_REBOOT, &clp->cl_state))
+		return;
+
+	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);
+		spin_lock(&sp->so_lock);
+		list_for_each_entry(state, &sp->so_states, open_states) {
+			if (!test_and_clear_bit(NFS_STATE_RECLAIM_REBOOT, &state->flags))
+				continue;
+			nfs4_state_mark_reclaim_nograce(clp, state);
+		}
+		spin_unlock(&sp->so_lock);
+	}
+
+	nfs_delegation_reap_unclaimed(clp);
+}
+
+static void nfs_delegation_clear_all(struct nfs_client *clp)
+{
+	nfs_delegation_mark_reclaim(clp);
+	nfs_delegation_reap_unclaimed(clp);
+}
+
+static void nfs4_state_start_reclaim_nograce(struct nfs_client *clp)
+{
+	nfs_delegation_clear_all(clp);
+	nfs4_state_mark_reclaim_helper(clp, nfs4_state_mark_reclaim_nograce);
+}
+
+static void nfs4_state_end_reclaim_nograce(struct nfs_client *clp)
+{
+	clear_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state);
+}
+
+static void nfs4_recovery_handle_error(struct nfs_client *clp, int error)
+{
+	switch (error) {
+		case -NFS4ERR_CB_PATH_DOWN:
+			nfs_handle_cb_pathdown(clp);
+			break;
+		case -NFS4ERR_STALE_CLIENTID:
+		case -NFS4ERR_LEASE_MOVED:
+			set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
+			nfs4_state_start_reclaim_reboot(clp);
+			break;
+		case -NFS4ERR_EXPIRED:
+			set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
+			nfs4_state_start_reclaim_nograce(clp);
+	}
+}
+
+static int nfs4_do_reclaim(struct nfs_client *clp, const struct nfs4_state_recovery_ops *ops)
+{
+	struct rb_node *pos;
 	int status = 0;
 
-	allow_signal(SIGKILL);
+restart:
+	spin_lock(&clp->cl_lock);
+	for (pos = rb_first(&clp->cl_state_owners); pos != NULL; pos = rb_next(pos)) {
+		struct nfs4_state_owner *sp = rb_entry(pos, struct nfs4_state_owner, so_client_node);
+		if (!test_and_clear_bit(ops->owner_flag_bit, &sp->so_flags))
+			continue;
+		atomic_inc(&sp->so_count);
+		spin_unlock(&clp->cl_lock);
+		status = nfs4_reclaim_open_state(sp, ops);
+		if (status < 0) {
+			set_bit(ops->owner_flag_bit, &sp->so_flags);
+			nfs4_put_state_owner(sp);
+			nfs4_recovery_handle_error(clp, status);
+			return status;
+		}
+		nfs4_put_state_owner(sp);
+		goto restart;
+	}
+	spin_unlock(&clp->cl_lock);
+	return status;
+}
 
-	/* Ensure exclusive access to NFSv4 state */
-	down_write(&clp->cl_sem);
-	/* Are there any NFS mounts out there? */
-	if (list_empty(&clp->cl_superblocks))
-		goto out;
-restart_loop:
-	ops = &nfs4_network_partition_recovery_ops;
-	/* Are there any open files on this volume? */
+static int nfs4_check_lease(struct nfs_client *clp)
+{
+	struct rpc_cred *cred;
+	int status = -NFS4ERR_EXPIRED;
+
+	/* Is the client already known to have an expired lease? */
+	if (test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state))
+		return 0;
 	cred = nfs4_get_renew_cred(clp);
-	if (cred != NULL) {
-		/* Yes there are: try to renew the old lease */
-		status = nfs4_proc_renew(clp, cred);
-		put_rpccred(cred);
-		switch (status) {
-			case 0:
-			case -NFS4ERR_CB_PATH_DOWN:
-				goto out;
-			case -NFS4ERR_STALE_CLIENTID:
-			case -NFS4ERR_LEASE_MOVED:
-				ops = &nfs4_reboot_recovery_ops;
-		}
-	} else {
-		/* "reboot" to ensure we clear all state on the server */
-		clp->cl_boot_time = CURRENT_TIME;
+	if (cred == NULL) {
+		cred = nfs4_get_setclientid_cred(clp);
+		if (cred == NULL)
+			goto out;
 	}
-	/* We're going to have to re-establish a clientid */
-	nfs4_state_mark_reclaim(clp);
-	status = -ENOENT;
+	status = nfs4_proc_renew(clp, cred);
+	put_rpccred(cred);
+out:
+	nfs4_recovery_handle_error(clp, status);
+	return status;
+}
+
+static int nfs4_reclaim_lease(struct nfs_client *clp)
+{
+	struct rpc_cred *cred;
+	int status = -ENOENT;
+
 	cred = nfs4_get_setclientid_cred(clp);
 	if (cred != NULL) {
 		status = nfs4_init_client(clp, cred);
@@ -974,42 +1099,90 @@ restart_loop:
 		/* Handle case where the user hasn't set up machine creds */
 		if (status == -EACCES && cred == clp->cl_machine_cred) {
 			nfs4_clear_machine_cred(clp);
-			goto restart_loop;
+			status = -EAGAIN;
 		}
 	}
-	if (status)
-		goto out_error;
-	/* Mark all delegations for reclaim */
-	nfs_delegation_mark_reclaim(clp);
-	/* Note: list is protected by exclusive lock on cl->cl_sem */
-	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) {
-				ops = &nfs4_network_partition_recovery_ops;
-				status = nfs4_reclaim_open_state(ops, sp);
+	return status;
+}
+
+static void nfs4_state_manager(struct nfs_client *clp)
+{
+	int status = 0;
+
+	/* Ensure exclusive access to NFSv4 state */
+	for(;;) {
+		if (test_and_clear_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state)) {
+			/* We're going to have to re-establish a clientid */
+			status = nfs4_reclaim_lease(clp);
+			if (status) {
+				set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
+				if (status == -EAGAIN)
+					continue;
+				goto out_error;
 			}
+			clear_bit(NFS4CLNT_CHECK_LEASE, &clp->cl_state);
+		}
+
+		if (test_and_clear_bit(NFS4CLNT_CHECK_LEASE, &clp->cl_state)) {
+			status = nfs4_check_lease(clp);
+			if (status != 0)
+				continue;
+		}
+
+		/* First recover reboot state... */
+		if (test_and_clear_bit(NFS4CLNT_RECLAIM_REBOOT, &clp->cl_state)) {
+			status = nfs4_do_reclaim(clp, &nfs4_reboot_recovery_ops);
 			if (status == -NFS4ERR_STALE_CLIENTID)
-				goto restart_loop;
-			if (status == -NFS4ERR_EXPIRED)
-				goto restart_loop;
+				continue;
+			nfs4_state_end_reclaim_reboot(clp);
+			continue;
+		}
+
+		/* Now recover expired state... */
+		if (test_and_clear_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state)) {
+			status = nfs4_do_reclaim(clp, &nfs4_nograce_recovery_ops);
+			if (status < 0) {
+				set_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state);
+				if (status == -NFS4ERR_STALE_CLIENTID)
+					continue;
+				if (status == -NFS4ERR_EXPIRED)
+					continue;
+				goto out_error;
+			} else
+				nfs4_state_end_reclaim_nograce(clp);
+			continue;
 		}
+
+		if (test_and_clear_bit(NFS4CLNT_DELEGRETURN, &clp->cl_state)) {
+			nfs_client_return_marked_delegations(clp);
+			continue;
+		}
+
+		nfs4_clear_state_manager_bit(clp);
+		/* Did we race with an attempt to give us more work? */
+		if (clp->cl_state == 0)
+			break;
+		if (test_and_set_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) != 0)
+			break;
 	}
-	nfs_delegation_reap_unclaimed(clp);
-out:
-	up_write(&clp->cl_sem);
-	if (status == -NFS4ERR_CB_PATH_DOWN)
-		nfs_handle_cb_pathdown(clp);
-	nfs4_clear_recover_bit(clp);
+	return;
+out_error:
+	printk(KERN_WARNING "Error: state manager failed on NFSv4 server %s"
+			" with error %d\n", clp->cl_hostname, -status);
+	if (test_bit(NFS4CLNT_RECLAIM_REBOOT, &clp->cl_state))
+		nfs4_state_end_reclaim_reboot(clp);
+	nfs4_clear_state_manager_bit(clp);
+}
+
+static int nfs4_run_state_manager(void *ptr)
+{
+	struct nfs_client *clp = ptr;
+
+	allow_signal(SIGKILL);
+	nfs4_state_manager(clp);
 	nfs_put_client(clp);
 	module_put_and_exit(0);
 	return 0;
-out_error:
-	printk(KERN_WARNING "Error: state recovery failed on NFSv4 server %s"
-			" with error %d\n", clp->cl_hostname, -status);
-	set_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state);
-	goto out;
 }
 
 /*

fs/nfs/nfsroot.c

@@ -86,6 +86,8 @@
 #include <net/ipconfig.h>
 #include <linux/parser.h>
 
+#include "internal.h"
+
 /* Define this to allow debugging output */
 #undef NFSROOT_DEBUG
 #define NFSDBG_FACILITY NFSDBG_ROOT
@@ -100,7 +102,7 @@ static char nfs_root_name[256] __initdata = "";
 static __be32 servaddr __initdata = 0;
 
 /* Name of directory to mount */
-static char nfs_path[NFS_MAXPATHLEN] __initdata = { 0, };
+static char nfs_export_path[NFS_MAXPATHLEN] __initdata = { 0, };
 
 /* NFS-related data */
 static struct nfs_mount_data nfs_data __initdata = { 0, };/* NFS mount info */
@@ -312,7 +314,7 @@ static int __init root_nfs_name(char *name)
 		printk(KERN_ERR "Root-NFS: Pathname for remote directory too long.\n");
 		return -1;
 	}
-	sprintf(nfs_path, buf, cp);
+	sprintf(nfs_export_path, buf, cp);
 
 	return 1;
 }
@@ -340,7 +342,7 @@ static int __init root_nfs_addr(void)
 static void __init root_nfs_print(void)
 {
 	printk(KERN_NOTICE "Root-NFS: Mounting %s on server %s as root\n",
-		nfs_path, nfs_data.hostname);
+		nfs_export_path, nfs_data.hostname);
 	printk(KERN_NOTICE "Root-NFS:     rsize = %d, wsize = %d, timeo = %d, retrans = %d\n",
 		nfs_data.rsize, nfs_data.wsize, nfs_data.timeo, nfs_data.retrans);
 	printk(KERN_NOTICE "Root-NFS:     acreg (min,max) = (%d,%d), acdir (min,max) = (%d,%d)\n",
@@ -485,18 +487,23 @@ static int __init root_nfs_get_handle(void)
 {
 	struct nfs_fh fh;
 	struct sockaddr_in sin;
+	struct nfs_mount_request request = {
+		.sap		= (struct sockaddr *)&sin,
+		.salen		= sizeof(sin),
+		.dirpath	= nfs_export_path,
+		.version	= (nfs_data.flags & NFS_MOUNT_VER3) ?
+					NFS_MNT3_VERSION : NFS_MNT_VERSION,
+		.protocol	= (nfs_data.flags & NFS_MOUNT_TCP) ?
+					XPRT_TRANSPORT_TCP : XPRT_TRANSPORT_UDP,
+		.fh		= &fh,
+	};
 	int status;
-	int protocol = (nfs_data.flags & NFS_MOUNT_TCP) ?
-					XPRT_TRANSPORT_TCP : XPRT_TRANSPORT_UDP;
-	int version = (nfs_data.flags & NFS_MOUNT_VER3) ?
-					NFS_MNT3_VERSION : NFS_MNT_VERSION;
 
 	set_sockaddr(&sin, servaddr, htons(mount_port));
-	status = nfs_mount((struct sockaddr *) &sin, sizeof(sin), NULL,
-			   nfs_path, version, protocol, &fh);
+	status = nfs_mount(&request);
 	if (status < 0)
 		printk(KERN_ERR "Root-NFS: Server returned error %d "
-				"while mounting %s\n", status, nfs_path);
+				"while mounting %s\n", status, nfs_export_path);
 	else {
 		nfs_data.root.size = fh.size;
 		memcpy(nfs_data.root.data, fh.data, fh.size);

fs/nfs/read.c

@@ -533,12 +533,6 @@ readpage_async_filler(void *data, struct page *page)
 	unsigned int len;
 	int error;
 
-	error = nfs_wb_page(inode, page);
-	if (error)
-		goto out_unlock;
-	if (PageUptodate(page))
-		goto out_unlock;
-
 	len = nfs_page_length(page);
 	if (len == 0)
 		return nfs_return_empty_page(page);

fs/nfs/super.c

@@ -75,6 +75,7 @@ enum {
 	Opt_acl, Opt_noacl,
 	Opt_rdirplus, Opt_nordirplus,
 	Opt_sharecache, Opt_nosharecache,
+	Opt_resvport, Opt_noresvport,
 
 	/* Mount options that take integer arguments */
 	Opt_port,
@@ -129,6 +130,8 @@ static const match_table_t nfs_mount_option_tokens = {
 	{ Opt_nordirplus, "nordirplus" },
 	{ Opt_sharecache, "sharecache" },
 	{ Opt_nosharecache, "nosharecache" },
+	{ Opt_resvport, "resvport" },
+	{ Opt_noresvport, "noresvport" },
 
 	{ Opt_port, "port=%u" },
 	{ Opt_rsize, "rsize=%u" },
@@ -512,7 +515,8 @@ 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", ""},
+		{ NFS_MOUNT_UNSHARED, ",nosharecache", "" },
+		{ NFS_MOUNT_NORESVPORT, ",noresvport", "" },
 		{ 0, NULL, NULL }
 	};
 	const struct proc_nfs_info *nfs_infop;
@@ -1033,6 +1037,12 @@ static int nfs_parse_mount_options(char *raw,
 		case Opt_nosharecache:
 			mnt->flags |= NFS_MOUNT_UNSHARED;
 			break;
+		case Opt_resvport:
+			mnt->flags &= ~NFS_MOUNT_NORESVPORT;
+			break;
+		case Opt_noresvport:
+			mnt->flags |= NFS_MOUNT_NORESVPORT;
+			break;
 
 		/*
 		 * options that take numeric values
@@ -1327,8 +1337,14 @@ out_security_failure:
 static int nfs_try_mount(struct nfs_parsed_mount_data *args,
 			 struct nfs_fh *root_fh)
 {
-	struct sockaddr *sap = (struct sockaddr *)&args->mount_server.address;
-	char *hostname;
+	struct nfs_mount_request request = {
+		.sap		= (struct sockaddr *)
+						&args->mount_server.address,
+		.dirpath	= args->nfs_server.export_path,
+		.protocol	= args->mount_server.protocol,
+		.fh		= root_fh,
+		.noresvport	= args->flags & NFS_MOUNT_NORESVPORT,
+	};
 	int status;
 
 	if (args->mount_server.version == 0) {
@@ -1337,42 +1353,38 @@ static int nfs_try_mount(struct nfs_parsed_mount_data *args,
 		else
 			args->mount_server.version = NFS_MNT_VERSION;
 	}
+	request.version = args->mount_server.version;
 
 	if (args->mount_server.hostname)
-		hostname = args->mount_server.hostname;
+		request.hostname = args->mount_server.hostname;
 	else
-		hostname = args->nfs_server.hostname;
+		request.hostname = args->nfs_server.hostname;
 
 	/*
 	 * Construct the mount server's address.
 	 */
 	if (args->mount_server.address.ss_family == AF_UNSPEC) {
-		memcpy(sap, &args->nfs_server.address,
+		memcpy(request.sap, &args->nfs_server.address,
 		       args->nfs_server.addrlen);
 		args->mount_server.addrlen = args->nfs_server.addrlen;
 	}
+	request.salen = args->mount_server.addrlen;
 
 	/*
 	 * autobind will be used if mount_server.port == 0
 	 */
-	nfs_set_port(sap, args->mount_server.port);
+	nfs_set_port(request.sap, args->mount_server.port);
 
 	/*
 	 * Now ask the mount server to map our export path
 	 * to a file handle.
 	 */
-	status = nfs_mount(sap,
-			   args->mount_server.addrlen,
-			   hostname,
-			   args->nfs_server.export_path,
-			   args->mount_server.version,
-			   args->mount_server.protocol,
-			   root_fh);
+	status = nfs_mount(&request);
 	if (status == 0)
 		return 0;
 
 	dfprintk(MOUNT, "NFS: unable to mount server %s, error %d\n",
-			hostname, status);
+			request.hostname, status);
 	return status;
 }
 
@@ -2419,7 +2431,7 @@ static void nfs4_kill_super(struct super_block *sb)
 {
 	struct nfs_server *server = NFS_SB(sb);
 
-	nfs_return_all_delegations(sb);
+	nfs_super_return_all_delegations(sb);
 	kill_anon_super(sb);
 
 	nfs4_renewd_prepare_shutdown(server);

fs/nfs_common/nfsacl.c

@@ -29,8 +29,8 @@
 
 MODULE_LICENSE("GPL");
 
-EXPORT_SYMBOL(nfsacl_encode);
-EXPORT_SYMBOL(nfsacl_decode);
+EXPORT_SYMBOL_GPL(nfsacl_encode);
+EXPORT_SYMBOL_GPL(nfsacl_decode);
 
 struct nfsacl_encode_desc {
 	struct xdr_array2_desc desc;

fs/nfsd/nfs4callback.c

@@ -358,6 +358,7 @@ static struct rpc_program cb_program = {
 		.nrvers		= ARRAY_SIZE(nfs_cb_version),
 		.version	= nfs_cb_version,
 		.stats		= &cb_stats,
+		.pipe_dir_name  = "/nfsd4_cb",
 };
 
 /* Reference counting, callback cleanup, etc., all look racy as heck.
@@ -382,8 +383,9 @@ static int do_probe_callback(void *data)
 		.program	= &cb_program,
 		.prognumber	= cb->cb_prog,
 		.version	= nfs_cb_version[1]->number,
-		.authflavor	= RPC_AUTH_UNIX, /* XXX: need AUTH_GSS... */
+		.authflavor	= clp->cl_flavor,
 		.flags		= (RPC_CLNT_CREATE_NOPING | RPC_CLNT_CREATE_QUIET),
+		.client_name    = clp->cl_principal,
 	};
 	struct rpc_message msg = {
 		.rpc_proc       = &nfs4_cb_procedures[NFSPROC4_CLNT_CB_NULL],
@@ -392,6 +394,11 @@ static int do_probe_callback(void *data)
 	struct rpc_clnt *client;
 	int status;
 
+	if (!clp->cl_principal && (clp->cl_flavor >= RPC_AUTH_GSS_KRB5)) {
+		status = nfserr_cb_path_down;
+		goto out_err;
+	}
+
 	/* Initialize address */
 	memset(&addr, 0, sizeof(addr));
 	addr.sin_family = AF_INET;

fs/nfsd/nfs4state.c

@@ -54,6 +54,7 @@
 #include <linux/mutex.h>
 #include <linux/lockd/bind.h>
 #include <linux/module.h>
+#include <linux/sunrpc/svcauth_gss.h>
 
 #define NFSDDBG_FACILITY                NFSDDBG_PROC
 
@@ -377,6 +378,7 @@ free_client(struct nfs4_client *clp)
 	shutdown_callback_client(clp);
 	if (clp->cl_cred.cr_group_info)
 		put_group_info(clp->cl_cred.cr_group_info);
+	kfree(clp->cl_principal);
 	kfree(clp->cl_name.data);
 	kfree(clp);
 }
@@ -696,6 +698,7 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	unsigned int 		strhashval;
 	struct nfs4_client	*conf, *unconf, *new;
 	__be32 			status;
+	char			*princ;
 	char                    dname[HEXDIR_LEN];
 	
 	if (!check_name(clname))
@@ -783,6 +786,15 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	}
 	copy_verf(new, &clverifier);
 	new->cl_addr = sin->sin_addr.s_addr;
+	new->cl_flavor = rqstp->rq_flavor;
+	princ = svc_gss_principal(rqstp);
+	if (princ) {
+		new->cl_principal = kstrdup(princ, GFP_KERNEL);
+		if (new->cl_principal == NULL) {
+			free_client(new);
+			goto out;
+		}
+	}
 	copy_cred(&new->cl_cred, &rqstp->rq_cred);
 	gen_confirm(new);
 	gen_callback(new, setclid);

include/linux/jiffies.h

@@ -115,10 +115,20 @@ static inline u64 get_jiffies_64(void)
 	 ((long)(a) - (long)(b) >= 0))
 #define time_before_eq(a,b)	time_after_eq(b,a)
 
+/*
+ * Calculate whether a is in the range of [b, c].
+ */
 #define time_in_range(a,b,c) \
 	(time_after_eq(a,b) && \
 	 time_before_eq(a,c))
 
+/*
+ * Calculate whether a is in the range of [b, c).
+ */
+#define time_in_range_open(a,b,c) \
+	(time_after_eq(a,b) && \
+	 time_before(a,c))
+
 /* Same as above, but does so with platform independent 64bit types.
  * These must be used when utilizing jiffies_64 (i.e. return value of
  * get_jiffies_64() */

include/linux/lockd/bind.h

@@ -41,6 +41,7 @@ struct nlmclnt_initdata {
 	size_t			addrlen;
 	unsigned short		protocol;
 	u32			nfs_version;
+	int			noresvport;
 };
 
 /*

include/linux/lockd/lockd.h

@@ -49,6 +49,7 @@ struct nlm_host {
 	unsigned short		h_proto;	/* transport proto */
 	unsigned short		h_reclaiming : 1,
 				h_server     : 1, /* server side, not client side */
+				h_noresvport : 1,
 				h_inuse      : 1;
 	wait_queue_head_t	h_gracewait;	/* wait while reclaiming */
 	struct rw_semaphore	h_rwsem;	/* Reboot recovery lock */
@@ -220,7 +221,8 @@ struct nlm_host  *nlmclnt_lookup_host(const struct sockaddr *sap,
 					const size_t salen,
 					const unsigned short protocol,
 					const u32 version,
-					const char *hostname);
+					const char *hostname,
+					int noresvport);
 struct nlm_host  *nlmsvc_lookup_host(const struct svc_rqst *rqstp,
 					const char *hostname,
 					const size_t hostname_len);

include/linux/nfs_fs.h

@@ -83,7 +83,7 @@ struct nfs_open_context {
 	struct rpc_cred *cred;
 	struct nfs4_state *state;
 	fl_owner_t lockowner;
-	int mode;
+	fmode_t mode;
 
 	unsigned long flags;
 #define NFS_CONTEXT_ERROR_WRITE		(0)
@@ -130,7 +130,10 @@ struct nfs_inode {
 	 *
 	 * We need to revalidate the cached attrs for this inode if
 	 *
-	 *	jiffies - read_cache_jiffies > attrtimeo
+	 *	jiffies - read_cache_jiffies >= attrtimeo
+	 *
+	 * Please note the comparison is greater than or equal
+	 * so that zero timeout values can be specified.
 	 */
 	unsigned long		read_cache_jiffies;
 	unsigned long		attrtimeo;
@@ -180,7 +183,7 @@ struct nfs_inode {
         /* NFSv4 state */
 	struct list_head	open_states;
 	struct nfs_delegation	*delegation;
-	int			 delegation_state;
+	fmode_t			 delegation_state;
 	struct rw_semaphore	rwsem;
 #endif /* CONFIG_NFS_V4*/
 	struct inode		vfs_inode;
@@ -342,7 +345,7 @@ extern int nfs_setattr(struct dentry *, struct iattr *);
 extern void nfs_setattr_update_inode(struct inode *inode, struct iattr *attr);
 extern struct nfs_open_context *get_nfs_open_context(struct nfs_open_context *ctx);
 extern void put_nfs_open_context(struct nfs_open_context *ctx);
-extern struct nfs_open_context *nfs_find_open_context(struct inode *inode, struct rpc_cred *cred, int mode);
+extern struct nfs_open_context *nfs_find_open_context(struct inode *inode, struct rpc_cred *cred, fmode_t mode);
 extern u64 nfs_compat_user_ino64(u64 fileid);
 extern void nfs_fattr_init(struct nfs_fattr *fattr);
 
@@ -532,12 +535,6 @@ static inline void nfs3_forget_cached_acls(struct inode *inode)
 }
 #endif /* CONFIG_NFS_V3_ACL */
 
-/*
- * linux/fs/mount_clnt.c
- */
-extern int  nfs_mount(struct sockaddr *, size_t, char *, char *,
-		      int, int, struct nfs_fh *);
-
 /*
  * inline functions
  */

include/linux/nfs_fs_sb.h

@@ -42,12 +42,6 @@ struct nfs_client {
 	struct rb_root		cl_openowner_id;
 	struct rb_root		cl_lockowner_id;
 
-	/*
-	 * The following rwsem ensures exclusive access to the server
-	 * while we recover the state following a lease expiration.
-	 */
-	struct rw_semaphore	cl_sem;
-
 	struct list_head	cl_delegations;
 	struct rb_root		cl_state_owners;
 	spinlock_t		cl_lock;

include/linux/nfs_mount.h

@@ -45,7 +45,7 @@ struct nfs_mount_data {
 	char		context[NFS_MAX_CONTEXT_LEN + 1];	/* 6 */
 };
 
-/* bits in the flags field */
+/* bits in the flags field visible to user space */
 
 #define NFS_MOUNT_SOFT		0x0001	/* 1 */
 #define NFS_MOUNT_INTR		0x0002	/* 1 */ /* now unused, but ABI */
@@ -68,5 +68,6 @@ struct nfs_mount_data {
 /* The following are for internal use only */
 #define NFS_MOUNT_LOOKUP_CACHE_NONEG	0x10000
 #define NFS_MOUNT_LOOKUP_CACHE_NONE	0x20000
+#define NFS_MOUNT_NORESVPORT		0x40000
 
 #endif

include/linux/nfs_xdr.h

@@ -120,13 +120,14 @@ struct nfs_openargs {
 	const struct nfs_fh *	fh;
 	struct nfs_seqid *	seqid;
 	int			open_flags;
+	fmode_t			fmode;
 	__u64                   clientid;
 	__u64                   id;
 	union {
 		struct iattr *  attrs;    /* UNCHECKED, GUARDED */
 		nfs4_verifier   verifier; /* EXCLUSIVE */
 		nfs4_stateid	delegation;		/* CLAIM_DELEGATE_CUR */
-		int		delegation_type;	/* CLAIM_PREVIOUS */
+		fmode_t		delegation_type;	/* CLAIM_PREVIOUS */
 	} u;
 	const struct qstr *	name;
 	const struct nfs_server *server;	 /* Needed for ID mapping */
@@ -143,7 +144,7 @@ struct nfs_openres {
 	struct nfs_fattr *      dir_attr;
 	struct nfs_seqid *	seqid;
 	const struct nfs_server *server;
-	int			delegation_type;
+	fmode_t			delegation_type;
 	nfs4_stateid		delegation;
 	__u32			do_recall;
 	__u64			maxsize;
@@ -171,7 +172,7 @@ struct nfs_closeargs {
 	struct nfs_fh *         fh;
 	nfs4_stateid *		stateid;
 	struct nfs_seqid *	seqid;
-	int			open_flags;
+	fmode_t			fmode;
 	const u32 *		bitmask;
 };
 

include/linux/nfsd/state.h

@@ -124,6 +124,8 @@ struct nfs4_client {
 	nfs4_verifier		cl_verifier; 	/* generated by client */
 	time_t                  cl_time;        /* time of last lease renewal */
 	__be32			cl_addr; 	/* client ipaddress */
+	u32			cl_flavor;	/* setclientid pseudoflavor */
+	char			*cl_principal;	/* setclientid principal name */
 	struct svc_cred		cl_cred; 	/* setclientid principal */
 	clientid_t		cl_clientid;	/* generated by server */
 	nfs4_verifier		cl_confirm;	/* generated by server */

include/linux/sunrpc/clnt.h

@@ -58,6 +58,7 @@ struct rpc_clnt {
 	struct rpc_timeout	cl_timeout_default;
 	struct rpc_program *	cl_program;
 	char			cl_inline_name[32];
+	char			*cl_principal;	/* target to authenticate to */
 };
 
 /*
@@ -108,6 +109,7 @@ struct rpc_create_args {
 	u32			version;
 	rpc_authflavor_t	authflavor;
 	unsigned long		flags;
+	char			*client_name;
 };
 
 /* Values for "flags" field */

include/linux/sunrpc/rpc_pipe_fs.h

@@ -15,6 +15,7 @@ struct rpc_pipe_ops {
 	ssize_t (*upcall)(struct file *, struct rpc_pipe_msg *, char __user *, size_t);
 	ssize_t (*downcall)(struct file *, const char __user *, size_t);
 	void (*release_pipe)(struct inode *);
+	int (*open_pipe)(struct inode *);
 	void (*destroy_msg)(struct rpc_pipe_msg *);
 };
 

include/linux/sunrpc/svcauth_gss.h

@@ -20,6 +20,7 @@ int gss_svc_init(void);
 void gss_svc_shutdown(void);
 int svcauth_gss_register_pseudoflavor(u32 pseudoflavor, char * name);
 u32 svcauth_gss_flavor(struct auth_domain *dom);
+char *svc_gss_principal(struct svc_rqst *);
 
 #endif /* __KERNEL__ */
 #endif /* _LINUX_SUNRPC_SVCAUTH_GSS_H */

include/linux/sunrpc/xdr.h

@@ -36,21 +36,6 @@ struct xdr_netobj {
  */
 typedef int	(*kxdrproc_t)(void *rqstp, __be32 *data, void *obj);
 
-/*
- * We're still requiring the BKL in the xdr code until it's been
- * more carefully audited, at which point this wrapper will become
- * unnecessary.
- */
-static inline int rpc_call_xdrproc(kxdrproc_t xdrproc, void *rqstp, __be32 *data, void *obj)
-{
-	int ret;
-
-	lock_kernel();
-	ret = xdrproc(rqstp, data, obj);
-	unlock_kernel();
-	return ret;
-}
-
 /*
  * Basic structure for transmission/reception of a client XDR message.
  * Features a header (for a linear buffer containing RPC headers

include/linux/sunrpc/xprt.h

@@ -76,8 +76,7 @@ struct rpc_rqst {
 	struct list_head	rq_list;
 
 	__u32 *			rq_buffer;	/* XDR encode buffer */
-	size_t			rq_bufsize,
-				rq_callsize,
+	size_t			rq_callsize,
 				rq_rcvsize;
 
 	struct xdr_buf		rq_private_buf;		/* The receive buffer

net/sunrpc/auth.c

@@ -234,7 +234,7 @@ rpcauth_prune_expired(struct list_head *free, int nr_to_scan)
 	list_for_each_entry_safe(cred, next, &cred_unused, cr_lru) {
 
 		/* Enforce a 60 second garbage collection moratorium */
-		if (time_in_range(cred->cr_expire, expired, jiffies) &&
+		if (time_in_range_open(cred->cr_expire, expired, jiffies) &&
 		    test_bit(RPCAUTH_CRED_HASHED, &cred->cr_flags) != 0)
 			continue;
 
@@ -515,7 +515,7 @@ rpcauth_wrap_req(struct rpc_task *task, kxdrproc_t encode, void *rqstp,
 	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 rpc_call_xdrproc(encode, rqstp, data, obj);
+	return encode(rqstp, data, obj);
 }
 
 int
@@ -530,7 +530,7 @@ 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 rpc_call_xdrproc(decode, rqstp, data, obj);
+	return decode(rqstp, data, obj);
 }
 
 int

net/sunrpc/auth_gss/auth_gss.c

@@ -72,11 +72,25 @@ struct gss_auth {
 	struct gss_api_mech *mech;
 	enum rpc_gss_svc service;
 	struct rpc_clnt *client;
-	struct dentry *dentry;
+	/*
+	 * There are two upcall pipes; dentry[1], named "gssd", is used
+	 * for the new text-based upcall; dentry[0] is named after the
+	 * mechanism (for example, "krb5") and exists for
+	 * backwards-compatibility with older gssd's.
+	 */
+	struct dentry *dentry[2];
 };
 
+/* pipe_version >= 0 if and only if someone has a pipe open. */
+static int pipe_version = -1;
+static atomic_t pipe_users = ATOMIC_INIT(0);
+static DEFINE_SPINLOCK(pipe_version_lock);
+static struct rpc_wait_queue pipe_version_rpc_waitqueue;
+static DECLARE_WAIT_QUEUE_HEAD(pipe_version_waitqueue);
+
 static void gss_free_ctx(struct gss_cl_ctx *);
-static struct rpc_pipe_ops gss_upcall_ops;
+static struct rpc_pipe_ops gss_upcall_ops_v0;
+static struct rpc_pipe_ops gss_upcall_ops_v1;
 
 static inline struct gss_cl_ctx *
 gss_get_ctx(struct gss_cl_ctx *ctx)
@@ -220,6 +234,7 @@ err:
 	return p;
 }
 
+#define UPCALL_BUF_LEN 128
 
 struct gss_upcall_msg {
 	atomic_t count;
@@ -227,16 +242,41 @@ struct gss_upcall_msg {
 	struct rpc_pipe_msg msg;
 	struct list_head list;
 	struct gss_auth *auth;
+	struct rpc_inode *inode;
 	struct rpc_wait_queue rpc_waitqueue;
 	wait_queue_head_t waitqueue;
 	struct gss_cl_ctx *ctx;
+	char databuf[UPCALL_BUF_LEN];
 };
 
+static int get_pipe_version(void)
+{
+	int ret;
+
+	spin_lock(&pipe_version_lock);
+	if (pipe_version >= 0) {
+		atomic_inc(&pipe_users);
+		ret = pipe_version;
+	} else
+		ret = -EAGAIN;
+	spin_unlock(&pipe_version_lock);
+	return ret;
+}
+
+static void put_pipe_version(void)
+{
+	if (atomic_dec_and_lock(&pipe_users, &pipe_version_lock)) {
+		pipe_version = -1;
+		spin_unlock(&pipe_version_lock);
+	}
+}
+
 static void
 gss_release_msg(struct gss_upcall_msg *gss_msg)
 {
 	if (!atomic_dec_and_test(&gss_msg->count))
 		return;
+	put_pipe_version();
 	BUG_ON(!list_empty(&gss_msg->list));
 	if (gss_msg->ctx != NULL)
 		gss_put_ctx(gss_msg->ctx);
@@ -266,8 +306,8 @@ __gss_find_upcall(struct rpc_inode *rpci, 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 rpc_inode *rpci = gss_msg->inode;
+	struct inode *inode = &rpci->vfs_inode;
 	struct gss_upcall_msg *old;
 
 	spin_lock(&inode->i_lock);
@@ -293,8 +333,7 @@ __gss_unhash_msg(struct gss_upcall_msg *gss_msg)
 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;
+	struct inode *inode = &gss_msg->inode->vfs_inode;
 
 	if (list_empty(&gss_msg->list))
 		return;
@@ -310,7 +349,7 @@ 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;
+	struct inode *inode = &gss_msg->inode->vfs_inode;
 
 	spin_lock(&inode->i_lock);
 	if (gss_msg->ctx)
@@ -323,22 +362,75 @@ gss_upcall_callback(struct rpc_task *task)
 	gss_release_msg(gss_msg);
 }
 
+static void gss_encode_v0_msg(struct gss_upcall_msg *gss_msg)
+{
+	gss_msg->msg.data = &gss_msg->uid;
+	gss_msg->msg.len = sizeof(gss_msg->uid);
+}
+
+static void gss_encode_v1_msg(struct gss_upcall_msg *gss_msg,
+				struct rpc_clnt *clnt, int machine_cred)
+{
+	char *p = gss_msg->databuf;
+	int len = 0;
+
+	gss_msg->msg.len = sprintf(gss_msg->databuf, "mech=%s uid=%d ",
+				   gss_msg->auth->mech->gm_name,
+				   gss_msg->uid);
+	p += gss_msg->msg.len;
+	if (clnt->cl_principal) {
+		len = sprintf(p, "target=%s ", clnt->cl_principal);
+		p += len;
+		gss_msg->msg.len += len;
+	}
+	if (machine_cred) {
+		len = sprintf(p, "service=* ");
+		p += len;
+		gss_msg->msg.len += len;
+	} else if (!strcmp(clnt->cl_program->name, "nfs4_cb")) {
+		len = sprintf(p, "service=nfs ");
+		p += len;
+		gss_msg->msg.len += len;
+	}
+	len = sprintf(p, "\n");
+	gss_msg->msg.len += len;
+
+	gss_msg->msg.data = gss_msg->databuf;
+	BUG_ON(gss_msg->msg.len > UPCALL_BUF_LEN);
+}
+
+static void gss_encode_msg(struct gss_upcall_msg *gss_msg,
+				struct rpc_clnt *clnt, int machine_cred)
+{
+	if (pipe_version == 0)
+		gss_encode_v0_msg(gss_msg);
+	else /* pipe_version == 1 */
+		gss_encode_v1_msg(gss_msg, clnt, machine_cred);
+}
+
 static inline struct gss_upcall_msg *
-gss_alloc_msg(struct gss_auth *gss_auth, uid_t uid)
+gss_alloc_msg(struct gss_auth *gss_auth, uid_t uid, struct rpc_clnt *clnt,
+		int machine_cred)
 {
 	struct gss_upcall_msg *gss_msg;
+	int vers;
 
 	gss_msg = kzalloc(sizeof(*gss_msg), GFP_NOFS);
-	if (gss_msg != NULL) {
-		INIT_LIST_HEAD(&gss_msg->list);
-		rpc_init_wait_queue(&gss_msg->rpc_waitqueue, "RPCSEC_GSS upcall waitq");
-		init_waitqueue_head(&gss_msg->waitqueue);
-		atomic_set(&gss_msg->count, 1);
-		gss_msg->msg.data = &gss_msg->uid;
-		gss_msg->msg.len = sizeof(gss_msg->uid);
-		gss_msg->uid = uid;
-		gss_msg->auth = gss_auth;
+	if (gss_msg == NULL)
+		return ERR_PTR(-ENOMEM);
+	vers = get_pipe_version();
+	if (vers < 0) {
+		kfree(gss_msg);
+		return ERR_PTR(vers);
 	}
+	gss_msg->inode = RPC_I(gss_auth->dentry[vers]->d_inode);
+	INIT_LIST_HEAD(&gss_msg->list);
+	rpc_init_wait_queue(&gss_msg->rpc_waitqueue, "RPCSEC_GSS upcall waitq");
+	init_waitqueue_head(&gss_msg->waitqueue);
+	atomic_set(&gss_msg->count, 1);
+	gss_msg->uid = uid;
+	gss_msg->auth = gss_auth;
+	gss_encode_msg(gss_msg, clnt, machine_cred);
 	return gss_msg;
 }
 
@@ -350,16 +442,13 @@ gss_setup_upcall(struct rpc_clnt *clnt, struct gss_auth *gss_auth, struct rpc_cr
 	struct gss_upcall_msg *gss_new, *gss_msg;
 	uid_t uid = cred->cr_uid;
 
-	/* Special case: rpc.gssd assumes that uid == 0 implies machine creds */
-	if (gss_cred->gc_machine_cred != 0)
-		uid = 0;
-
-	gss_new = gss_alloc_msg(gss_auth, uid);
-	if (gss_new == NULL)
-		return ERR_PTR(-ENOMEM);
+	gss_new = gss_alloc_msg(gss_auth, uid, clnt, gss_cred->gc_machine_cred);
+	if (IS_ERR(gss_new))
+		return gss_new;
 	gss_msg = gss_add_msg(gss_auth, gss_new);
 	if (gss_msg == gss_new) {
-		int res = rpc_queue_upcall(gss_auth->dentry->d_inode, &gss_new->msg);
+		struct inode *inode = &gss_new->inode->vfs_inode;
+		int res = rpc_queue_upcall(inode, &gss_new->msg);
 		if (res) {
 			gss_unhash_msg(gss_new);
 			gss_msg = ERR_PTR(res);
@@ -369,6 +458,18 @@ gss_setup_upcall(struct rpc_clnt *clnt, struct gss_auth *gss_auth, struct rpc_cr
 	return gss_msg;
 }
 
+static void warn_gssd(void)
+{
+	static unsigned long ratelimit;
+	unsigned long now = jiffies;
+
+	if (time_after(now, ratelimit)) {
+		printk(KERN_WARNING "RPC: AUTH_GSS upcall timed out.\n"
+				"Please check user daemon is running.\n");
+		ratelimit = now + 15*HZ;
+	}
+}
+
 static inline int
 gss_refresh_upcall(struct rpc_task *task)
 {
@@ -378,16 +479,25 @@ gss_refresh_upcall(struct rpc_task *task)
 	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;
+	struct inode *inode;
 	int err = 0;
 
 	dprintk("RPC: %5u gss_refresh_upcall for uid %u\n", task->tk_pid,
 								cred->cr_uid);
 	gss_msg = gss_setup_upcall(task->tk_client, gss_auth, cred);
+	if (IS_ERR(gss_msg) == -EAGAIN) {
+		/* XXX: warning on the first, under the assumption we
+		 * shouldn't normally hit this case on a refresh. */
+		warn_gssd();
+		task->tk_timeout = 15*HZ;
+		rpc_sleep_on(&pipe_version_rpc_waitqueue, task, NULL);
+		return 0;
+	}
 	if (IS_ERR(gss_msg)) {
 		err = PTR_ERR(gss_msg);
 		goto out;
 	}
+	inode = &gss_msg->inode->vfs_inode;
 	spin_lock(&inode->i_lock);
 	if (gss_cred->gc_upcall != NULL)
 		rpc_sleep_on(&gss_cred->gc_upcall->rpc_waitqueue, task, NULL);
@@ -414,18 +524,29 @@ 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 inode *inode;
 	struct rpc_cred *cred = &gss_cred->gc_base;
 	struct gss_upcall_msg *gss_msg;
 	DEFINE_WAIT(wait);
 	int err = 0;
 
 	dprintk("RPC:       gss_upcall for uid %u\n", cred->cr_uid);
+retry:
 	gss_msg = gss_setup_upcall(gss_auth->client, gss_auth, cred);
+	if (PTR_ERR(gss_msg) == -EAGAIN) {
+		err = wait_event_interruptible_timeout(pipe_version_waitqueue,
+				pipe_version >= 0, 15*HZ);
+		if (err)
+			goto out;
+		if (pipe_version < 0)
+			warn_gssd();
+		goto retry;
+	}
 	if (IS_ERR(gss_msg)) {
 		err = PTR_ERR(gss_msg);
 		goto out;
 	}
+	inode = &gss_msg->inode->vfs_inode;
 	for (;;) {
 		prepare_to_wait(&gss_msg->waitqueue, &wait, TASK_INTERRUPTIBLE);
 		spin_lock(&inode->i_lock);
@@ -543,6 +664,38 @@ out:
 	return err;
 }
 
+static int gss_pipe_open(struct inode *inode, int new_version)
+{
+	int ret = 0;
+
+	spin_lock(&pipe_version_lock);
+	if (pipe_version < 0) {
+		/* First open of any gss pipe determines the version: */
+		pipe_version = new_version;
+		rpc_wake_up(&pipe_version_rpc_waitqueue);
+		wake_up(&pipe_version_waitqueue);
+	} else if (pipe_version != new_version) {
+		/* Trying to open a pipe of a different version */
+		ret = -EBUSY;
+		goto out;
+	}
+	atomic_inc(&pipe_users);
+out:
+	spin_unlock(&pipe_version_lock);
+	return ret;
+
+}
+
+static int gss_pipe_open_v0(struct inode *inode)
+{
+	return gss_pipe_open(inode, 0);
+}
+
+static int gss_pipe_open_v1(struct inode *inode)
+{
+	return gss_pipe_open(inode, 1);
+}
+
 static void
 gss_pipe_release(struct inode *inode)
 {
@@ -562,27 +715,22 @@ gss_pipe_release(struct inode *inode)
 		spin_lock(&inode->i_lock);
 	}
 	spin_unlock(&inode->i_lock);
+
+	put_pipe_version();
 }
 
 static void
 gss_pipe_destroy_msg(struct rpc_pipe_msg *msg)
 {
 	struct gss_upcall_msg *gss_msg = container_of(msg, struct gss_upcall_msg, msg);
-	static unsigned long ratelimit;
 
 	if (msg->errno < 0) {
 		dprintk("RPC:       gss_pipe_destroy_msg releasing msg %p\n",
 				gss_msg);
 		atomic_inc(&gss_msg->count);
 		gss_unhash_msg(gss_msg);
-		if (msg->errno == -ETIMEDOUT) {
-			unsigned long now = jiffies;
-			if (time_after(now, ratelimit)) {
-				printk(KERN_WARNING "RPC: AUTH_GSS upcall timed out.\n"
-						    "Please check user daemon is running!\n");
-				ratelimit = now + 15*HZ;
-			}
-		}
+		if (msg->errno == -ETIMEDOUT)
+			warn_gssd();
 		gss_release_msg(gss_msg);
 	}
 }
@@ -623,20 +771,38 @@ gss_create(struct rpc_clnt *clnt, rpc_authflavor_t flavor)
 	atomic_set(&auth->au_count, 1);
 	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);
-	if (IS_ERR(gss_auth->dentry)) {
-		err = PTR_ERR(gss_auth->dentry);
+	/*
+	 * Note: if we created the old pipe first, then someone who
+	 * examined the directory at the right moment might conclude
+	 * that we supported only the old pipe.  So we instead create
+	 * the new pipe first.
+	 */
+	gss_auth->dentry[1] = rpc_mkpipe(clnt->cl_dentry,
+					 "gssd",
+					 clnt, &gss_upcall_ops_v1,
+					 RPC_PIPE_WAIT_FOR_OPEN);
+	if (IS_ERR(gss_auth->dentry[1])) {
+		err = PTR_ERR(gss_auth->dentry[1]);
 		goto err_put_mech;
 	}
 
+	gss_auth->dentry[0] = rpc_mkpipe(clnt->cl_dentry,
+					 gss_auth->mech->gm_name,
+					 clnt, &gss_upcall_ops_v0,
+					 RPC_PIPE_WAIT_FOR_OPEN);
+	if (IS_ERR(gss_auth->dentry[0])) {
+		err = PTR_ERR(gss_auth->dentry[0]);
+		goto err_unlink_pipe_1;
+	}
 	err = rpcauth_init_credcache(auth);
 	if (err)
-		goto err_unlink_pipe;
+		goto err_unlink_pipe_0;
 
 	return auth;
-err_unlink_pipe:
-	rpc_unlink(gss_auth->dentry);
+err_unlink_pipe_0:
+	rpc_unlink(gss_auth->dentry[0]);
+err_unlink_pipe_1:
+	rpc_unlink(gss_auth->dentry[1]);
 err_put_mech:
 	gss_mech_put(gss_auth->mech);
 err_free:
@@ -649,8 +815,8 @@ out_dec:
 static void
 gss_free(struct gss_auth *gss_auth)
 {
-	rpc_unlink(gss_auth->dentry);
-	gss_auth->dentry = NULL;
+	rpc_unlink(gss_auth->dentry[1]);
+	rpc_unlink(gss_auth->dentry[0]);
 	gss_mech_put(gss_auth->mech);
 
 	kfree(gss_auth);
@@ -693,7 +859,7 @@ gss_destroying_context(struct rpc_cred *cred)
 	struct rpc_task *task;
 
 	if (gss_cred->gc_ctx == NULL ||
-	    test_and_clear_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) == 0)
+	    test_bit(RPCAUTH_CRED_UPTODATE, &cred->cr_flags) == 0)
 		return 0;
 
 	gss_cred->gc_ctx->gc_proc = RPC_GSS_PROC_DESTROY;
@@ -757,14 +923,12 @@ gss_free_cred_callback(struct rcu_head *head)
 }
 
 static void
-gss_destroy_cred(struct rpc_cred *cred)
+gss_destroy_nullcred(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)
@@ -772,6 +936,15 @@ gss_destroy_cred(struct rpc_cred *cred)
 	kref_put(&gss_auth->kref, gss_free_callback);
 }
 
+static void
+gss_destroy_cred(struct rpc_cred *cred)
+{
+
+	if (gss_destroying_context(cred))
+		return;
+	gss_destroy_nullcred(cred);
+}
+
 /*
  * Lookup RPCSEC_GSS cred for the current process
  */
@@ -1017,7 +1190,7 @@ 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);
 
-	status = rpc_call_xdrproc(encode, rqstp, p, obj);
+	status = encode(rqstp, p, obj);
 	if (status)
 		return status;
 
@@ -1111,7 +1284,7 @@ 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);
 
-	status = rpc_call_xdrproc(encode, rqstp, p, obj);
+	status = encode(rqstp, p, obj);
 	if (status)
 		return status;
 
@@ -1170,12 +1343,12 @@ 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.
 		 */
-		status = rpc_call_xdrproc(encode, rqstp, p, obj);
+		status = encode(rqstp, p, obj);
 		goto out;
 	}
 	switch (gss_cred->gc_service) {
 		case RPC_GSS_SVC_NONE:
-			status = rpc_call_xdrproc(encode, rqstp, p, obj);
+			status = encode(rqstp, p, obj);
 			break;
 		case RPC_GSS_SVC_INTEGRITY:
 			status = gss_wrap_req_integ(cred, ctx, encode,
@@ -1291,7 +1464,7 @@ gss_unwrap_resp(struct rpc_task *task,
 	cred->cr_auth->au_rslack = cred->cr_auth->au_verfsize + (p - savedp)
 						+ (savedlen - head->iov_len);
 out_decode:
-	status = rpc_call_xdrproc(decode, rqstp, p, obj);
+	status = decode(rqstp, p, obj);
 out:
 	gss_put_ctx(ctx);
 	dprintk("RPC: %5u gss_unwrap_resp returning %d\n", task->tk_pid,
@@ -1324,7 +1497,7 @@ static const struct rpc_credops gss_credops = {
 
 static const struct rpc_credops gss_nullops = {
 	.cr_name	= "AUTH_GSS",
-	.crdestroy	= gss_destroy_cred,
+	.crdestroy	= gss_destroy_nullcred,
 	.crbind		= rpcauth_generic_bind_cred,
 	.crmatch	= gss_match,
 	.crmarshal	= gss_marshal,
@@ -1334,10 +1507,19 @@ static const struct rpc_credops gss_nullops = {
 	.crunwrap_resp	= gss_unwrap_resp,
 };
 
-static struct rpc_pipe_ops gss_upcall_ops = {
+static struct rpc_pipe_ops gss_upcall_ops_v0 = {
+	.upcall		= gss_pipe_upcall,
+	.downcall	= gss_pipe_downcall,
+	.destroy_msg	= gss_pipe_destroy_msg,
+	.open_pipe	= gss_pipe_open_v0,
+	.release_pipe	= gss_pipe_release,
+};
+
+static struct rpc_pipe_ops gss_upcall_ops_v1 = {
 	.upcall		= gss_pipe_upcall,
 	.downcall	= gss_pipe_downcall,
 	.destroy_msg	= gss_pipe_destroy_msg,
+	.open_pipe	= gss_pipe_open_v1,
 	.release_pipe	= gss_pipe_release,
 };
 
@@ -1354,6 +1536,7 @@ static int __init init_rpcsec_gss(void)
 	err = gss_svc_init();
 	if (err)
 		goto out_unregister;
+	rpc_init_wait_queue(&pipe_version_rpc_waitqueue, "gss pipe version");
 	return 0;
 out_unregister:
 	rpcauth_unregister(&authgss_ops);

net/sunrpc/auth_gss/gss_generic_token.c

@@ -152,7 +152,7 @@ g_token_size(struct xdr_netobj *mech, unsigned int body_size)
 	return(1 + der_length_size(body_size) + body_size);
 }
 
-EXPORT_SYMBOL(g_token_size);
+EXPORT_SYMBOL_GPL(g_token_size);
 
 /* fills in a buffer with the token header.  The buffer is assumed to
    be the right size.  buf is advanced past the token header */
@@ -167,7 +167,7 @@ g_make_token_header(struct xdr_netobj *mech, int body_size, unsigned char **buf)
 	TWRITE_STR(*buf, mech->data, ((int) mech->len));
 }
 
-EXPORT_SYMBOL(g_make_token_header);
+EXPORT_SYMBOL_GPL(g_make_token_header);
 
 /*
  * Given a buffer containing a token, reads and verifies the token,
@@ -231,5 +231,5 @@ g_verify_token_header(struct xdr_netobj *mech, int *body_size,
 	return(ret);
 }
 
-EXPORT_SYMBOL(g_verify_token_header);
+EXPORT_SYMBOL_GPL(g_verify_token_header);
 

net/sunrpc/auth_gss/gss_mech_switch.c

@@ -117,7 +117,7 @@ gss_mech_register(struct gss_api_mech *gm)
 	return 0;
 }
 
-EXPORT_SYMBOL(gss_mech_register);
+EXPORT_SYMBOL_GPL(gss_mech_register);
 
 void
 gss_mech_unregister(struct gss_api_mech *gm)
@@ -129,7 +129,7 @@ gss_mech_unregister(struct gss_api_mech *gm)
 	gss_mech_free(gm);
 }
 
-EXPORT_SYMBOL(gss_mech_unregister);
+EXPORT_SYMBOL_GPL(gss_mech_unregister);
 
 struct gss_api_mech *
 gss_mech_get(struct gss_api_mech *gm)
@@ -138,7 +138,7 @@ gss_mech_get(struct gss_api_mech *gm)
 	return gm;
 }
 
-EXPORT_SYMBOL(gss_mech_get);
+EXPORT_SYMBOL_GPL(gss_mech_get);
 
 struct gss_api_mech *
 gss_mech_get_by_name(const char *name)
@@ -158,7 +158,7 @@ gss_mech_get_by_name(const char *name)
 
 }
 
-EXPORT_SYMBOL(gss_mech_get_by_name);
+EXPORT_SYMBOL_GPL(gss_mech_get_by_name);
 
 static inline int
 mech_supports_pseudoflavor(struct gss_api_mech *gm, u32 pseudoflavor)
@@ -191,7 +191,7 @@ gss_mech_get_by_pseudoflavor(u32 pseudoflavor)
 	return gm;
 }
 
-EXPORT_SYMBOL(gss_mech_get_by_pseudoflavor);
+EXPORT_SYMBOL_GPL(gss_mech_get_by_pseudoflavor);
 
 u32
 gss_svc_to_pseudoflavor(struct gss_api_mech *gm, u32 service)
@@ -205,7 +205,7 @@ gss_svc_to_pseudoflavor(struct gss_api_mech *gm, u32 service)
 	}
 	return RPC_AUTH_MAXFLAVOR; /* illegal value */
 }
-EXPORT_SYMBOL(gss_svc_to_pseudoflavor);
+EXPORT_SYMBOL_GPL(gss_svc_to_pseudoflavor);
 
 u32
 gss_pseudoflavor_to_service(struct gss_api_mech *gm, u32 pseudoflavor)
@@ -219,7 +219,7 @@ gss_pseudoflavor_to_service(struct gss_api_mech *gm, u32 pseudoflavor)
 	return 0;
 }
 
-EXPORT_SYMBOL(gss_pseudoflavor_to_service);
+EXPORT_SYMBOL_GPL(gss_pseudoflavor_to_service);
 
 char *
 gss_service_to_auth_domain_name(struct gss_api_mech *gm, u32 service)
@@ -233,7 +233,7 @@ gss_service_to_auth_domain_name(struct gss_api_mech *gm, u32 service)
 	return NULL;
 }
 
-EXPORT_SYMBOL(gss_service_to_auth_domain_name);
+EXPORT_SYMBOL_GPL(gss_service_to_auth_domain_name);
 
 void
 gss_mech_put(struct gss_api_mech * gm)
@@ -242,7 +242,7 @@ gss_mech_put(struct gss_api_mech * gm)
 		module_put(gm->gm_owner);
 }
 
-EXPORT_SYMBOL(gss_mech_put);
+EXPORT_SYMBOL_GPL(gss_mech_put);
 
 /* The mech could probably be determined from the token instead, but it's just
  * as easy for now to pass it in. */

net/sunrpc/auth_gss/svcauth_gss.c

@@ -332,6 +332,7 @@ struct rsc {
 	struct svc_cred		cred;
 	struct gss_svc_seq_data	seqdata;
 	struct gss_ctx		*mechctx;
+	char			*client_name;
 };
 
 static struct cache_head *rsc_table[RSC_HASHMAX];
@@ -346,6 +347,7 @@ static void rsc_free(struct rsc *rsci)
 		gss_delete_sec_context(&rsci->mechctx);
 	if (rsci->cred.cr_group_info)
 		put_group_info(rsci->cred.cr_group_info);
+	kfree(rsci->client_name);
 }
 
 static void rsc_put(struct kref *ref)
@@ -383,6 +385,7 @@ rsc_init(struct cache_head *cnew, struct cache_head *ctmp)
 	tmp->handle.data = NULL;
 	new->mechctx = NULL;
 	new->cred.cr_group_info = NULL;
+	new->client_name = NULL;
 }
 
 static void
@@ -397,6 +400,8 @@ update_rsc(struct cache_head *cnew, struct cache_head *ctmp)
 	spin_lock_init(&new->seqdata.sd_lock);
 	new->cred = tmp->cred;
 	tmp->cred.cr_group_info = NULL;
+	new->client_name = tmp->client_name;
+	tmp->client_name = NULL;
 }
 
 static struct cache_head *
@@ -486,6 +491,15 @@ static int rsc_parse(struct cache_detail *cd,
 		status = gss_import_sec_context(buf, len, gm, &rsci.mechctx);
 		if (status)
 			goto out;
+
+		/* get client name */
+		len = qword_get(&mesg, buf, mlen);
+		if (len > 0) {
+			rsci.client_name = kstrdup(buf, GFP_KERNEL);
+			if (!rsci.client_name)
+				goto out;
+		}
+
 	}
 	rsci.h.expiry_time = expiry;
 	rscp = rsc_update(&rsci, rscp);
@@ -746,7 +760,7 @@ u32 svcauth_gss_flavor(struct auth_domain *dom)
 	return gd->pseudoflavor;
 }
 
-EXPORT_SYMBOL(svcauth_gss_flavor);
+EXPORT_SYMBOL_GPL(svcauth_gss_flavor);
 
 int
 svcauth_gss_register_pseudoflavor(u32 pseudoflavor, char * name)
@@ -780,7 +794,7 @@ out:
 	return stat;
 }
 
-EXPORT_SYMBOL(svcauth_gss_register_pseudoflavor);
+EXPORT_SYMBOL_GPL(svcauth_gss_register_pseudoflavor);
 
 static inline int
 read_u32_from_xdr_buf(struct xdr_buf *buf, int base, u32 *obj)
@@ -913,6 +927,16 @@ struct gss_svc_data {
 	struct rsc			*rsci;
 };
 
+char *svc_gss_principal(struct svc_rqst *rqstp)
+{
+	struct gss_svc_data *gd = (struct gss_svc_data *)rqstp->rq_auth_data;
+
+	if (gd && gd->rsci)
+		return gd->rsci->client_name;
+	return NULL;
+}
+EXPORT_SYMBOL_GPL(svc_gss_principal);
+
 static int
 svcauth_gss_set_client(struct svc_rqst *rqstp)
 {

net/sunrpc/clnt.c

@@ -197,6 +197,12 @@ static struct rpc_clnt * rpc_new_client(const struct rpc_create_args *args, stru
 
 	clnt->cl_rtt = &clnt->cl_rtt_default;
 	rpc_init_rtt(&clnt->cl_rtt_default, clnt->cl_timeout->to_initval);
+	clnt->cl_principal = NULL;
+	if (args->client_name) {
+		clnt->cl_principal = kstrdup(args->client_name, GFP_KERNEL);
+		if (!clnt->cl_principal)
+			goto out_no_principal;
+	}
 
 	kref_init(&clnt->cl_kref);
 
@@ -226,6 +232,8 @@ out_no_auth:
 		rpc_put_mount();
 	}
 out_no_path:
+	kfree(clnt->cl_principal);
+out_no_principal:
 	rpc_free_iostats(clnt->cl_metrics);
 out_no_stats:
 	if (clnt->cl_server != clnt->cl_inline_name)
@@ -354,6 +362,11 @@ rpc_clone_client(struct rpc_clnt *clnt)
 	new->cl_metrics = rpc_alloc_iostats(clnt);
 	if (new->cl_metrics == NULL)
 		goto out_no_stats;
+	if (clnt->cl_principal) {
+		new->cl_principal = kstrdup(clnt->cl_principal, GFP_KERNEL);
+		if (new->cl_principal == NULL)
+			goto out_no_principal;
+	}
 	kref_init(&new->cl_kref);
 	err = rpc_setup_pipedir(new, clnt->cl_program->pipe_dir_name);
 	if (err != 0)
@@ -366,6 +379,8 @@ rpc_clone_client(struct rpc_clnt *clnt)
 	rpciod_up();
 	return new;
 out_no_path:
+	kfree(new->cl_principal);
+out_no_principal:
 	rpc_free_iostats(new->cl_metrics);
 out_no_stats:
 	kfree(new);
@@ -417,6 +432,7 @@ rpc_free_client(struct kref *kref)
 out_free:
 	rpc_unregister_client(clnt);
 	rpc_free_iostats(clnt->cl_metrics);
+	kfree(clnt->cl_principal);
 	clnt->cl_metrics = NULL;
 	xprt_put(clnt->cl_xprt);
 	rpciod_down();

net/sunrpc/rpc_pipe.c

@@ -113,7 +113,7 @@ out:
 	wake_up(&rpci->waitq);
 	return res;
 }
-EXPORT_SYMBOL(rpc_queue_upcall);
+EXPORT_SYMBOL_GPL(rpc_queue_upcall);
 
 static inline void
 rpc_inode_setowner(struct inode *inode, void *private)
@@ -126,13 +126,14 @@ rpc_close_pipes(struct inode *inode)
 {
 	struct rpc_inode *rpci = RPC_I(inode);
 	struct rpc_pipe_ops *ops;
+	int need_release;
 
 	mutex_lock(&inode->i_mutex);
 	ops = rpci->ops;
 	if (ops != NULL) {
 		LIST_HEAD(free_list);
-
 		spin_lock(&inode->i_lock);
+		need_release = rpci->nreaders != 0 || rpci->nwriters != 0;
 		rpci->nreaders = 0;
 		list_splice_init(&rpci->in_upcall, &free_list);
 		list_splice_init(&rpci->pipe, &free_list);
@@ -141,7 +142,7 @@ rpc_close_pipes(struct inode *inode)
 		spin_unlock(&inode->i_lock);
 		rpc_purge_list(rpci, &free_list, ops->destroy_msg, -EPIPE);
 		rpci->nwriters = 0;
-		if (ops->release_pipe)
+		if (need_release && ops->release_pipe)
 			ops->release_pipe(inode);
 		cancel_delayed_work_sync(&rpci->queue_timeout);
 	}
@@ -169,16 +170,24 @@ static int
 rpc_pipe_open(struct inode *inode, struct file *filp)
 {
 	struct rpc_inode *rpci = RPC_I(inode);
+	int first_open;
 	int res = -ENXIO;
 
 	mutex_lock(&inode->i_mutex);
-	if (rpci->ops != NULL) {
-		if (filp->f_mode & FMODE_READ)
-			rpci->nreaders ++;
-		if (filp->f_mode & FMODE_WRITE)
-			rpci->nwriters ++;
-		res = 0;
+	if (rpci->ops == NULL)
+		goto out;
+	first_open = rpci->nreaders == 0 && rpci->nwriters == 0;
+	if (first_open && rpci->ops->open_pipe) {
+		res = rpci->ops->open_pipe(inode);
+		if (res)
+			goto out;
 	}
+	if (filp->f_mode & FMODE_READ)
+		rpci->nreaders++;
+	if (filp->f_mode & FMODE_WRITE)
+		rpci->nwriters++;
+	res = 0;
+out:
 	mutex_unlock(&inode->i_mutex);
 	return res;
 }
@@ -188,6 +197,7 @@ rpc_pipe_release(struct inode *inode, struct file *filp)
 {
 	struct rpc_inode *rpci = RPC_I(inode);
 	struct rpc_pipe_msg *msg;
+	int last_close;
 
 	mutex_lock(&inode->i_mutex);
 	if (rpci->ops == NULL)
@@ -214,7 +224,8 @@ rpc_pipe_release(struct inode *inode, struct file *filp)
 					rpci->ops->destroy_msg, -EAGAIN);
 		}
 	}
-	if (rpci->ops->release_pipe)
+	last_close = rpci->nwriters == 0 && rpci->nreaders == 0;
+	if (last_close && rpci->ops->release_pipe)
 		rpci->ops->release_pipe(inode);
 out:
 	mutex_unlock(&inode->i_mutex);
@@ -396,6 +407,7 @@ enum {
 	RPCAUTH_nfs,
 	RPCAUTH_portmap,
 	RPCAUTH_statd,
+	RPCAUTH_nfsd4_cb,
 	RPCAUTH_RootEOF
 };
 
@@ -429,6 +441,10 @@ static struct rpc_filelist files[] = {
 		.name = "statd",
 		.mode = S_IFDIR | S_IRUGO | S_IXUGO,
 	},
+	[RPCAUTH_nfsd4_cb] = {
+		.name = "nfsd4_cb",
+		.mode = S_IFDIR | S_IRUGO | S_IXUGO,
+	},
 };
 
 enum {
@@ -748,7 +764,7 @@ rpc_rmdir(struct dentry *dentry)
  * @name: name of pipe
  * @private: private data to associate with the pipe, for the caller's use
  * @ops: operations defining the behavior of the pipe: upcall, downcall,
- *	release_pipe, and destroy_msg.
+ *	release_pipe, open_pipe, and destroy_msg.
  * @flags: rpc_inode flags
  *
  * Data is made available for userspace to read by calls to
@@ -808,7 +824,7 @@ err_dput:
 			-ENOMEM);
 	goto out;
 }
-EXPORT_SYMBOL(rpc_mkpipe);
+EXPORT_SYMBOL_GPL(rpc_mkpipe);
 
 /**
  * rpc_unlink - remove a pipe
@@ -839,7 +855,7 @@ rpc_unlink(struct dentry *dentry)
 	dput(parent);
 	return error;
 }
-EXPORT_SYMBOL(rpc_unlink);
+EXPORT_SYMBOL_GPL(rpc_unlink);
 
 /*
  * populate the filesystem

net/sunrpc/xdr.c

@@ -28,7 +28,7 @@ xdr_encode_netobj(__be32 *p, const struct xdr_netobj *obj)
 	memcpy(p, obj->data, obj->len);
 	return p + XDR_QUADLEN(obj->len);
 }
-EXPORT_SYMBOL(xdr_encode_netobj);
+EXPORT_SYMBOL_GPL(xdr_encode_netobj);
 
 __be32 *
 xdr_decode_netobj(__be32 *p, struct xdr_netobj *obj)
@@ -41,7 +41,7 @@ xdr_decode_netobj(__be32 *p, struct xdr_netobj *obj)
 	obj->data = (u8 *) p;
 	return p + XDR_QUADLEN(len);
 }
-EXPORT_SYMBOL(xdr_decode_netobj);
+EXPORT_SYMBOL_GPL(xdr_decode_netobj);
 
 /**
  * xdr_encode_opaque_fixed - Encode fixed length opaque data
@@ -71,7 +71,7 @@ __be32 *xdr_encode_opaque_fixed(__be32 *p, const void *ptr, unsigned int nbytes)
 	}
 	return p;
 }
-EXPORT_SYMBOL(xdr_encode_opaque_fixed);
+EXPORT_SYMBOL_GPL(xdr_encode_opaque_fixed);
 
 /**
  * xdr_encode_opaque - Encode variable length opaque data
@@ -86,14 +86,14 @@ __be32 *xdr_encode_opaque(__be32 *p, const void *ptr, unsigned int nbytes)
 	*p++ = htonl(nbytes);
 	return xdr_encode_opaque_fixed(p, ptr, nbytes);
 }
-EXPORT_SYMBOL(xdr_encode_opaque);
+EXPORT_SYMBOL_GPL(xdr_encode_opaque);
 
 __be32 *
 xdr_encode_string(__be32 *p, const char *string)
 {
 	return xdr_encode_array(p, string, strlen(string));
 }
-EXPORT_SYMBOL(xdr_encode_string);
+EXPORT_SYMBOL_GPL(xdr_encode_string);
 
 __be32 *
 xdr_decode_string_inplace(__be32 *p, char **sp,
@@ -108,7 +108,7 @@ xdr_decode_string_inplace(__be32 *p, char **sp,
 	*sp = (char *) p;
 	return p + XDR_QUADLEN(len);
 }
-EXPORT_SYMBOL(xdr_decode_string_inplace);
+EXPORT_SYMBOL_GPL(xdr_decode_string_inplace);
 
 void
 xdr_encode_pages(struct xdr_buf *xdr, struct page **pages, unsigned int base,
@@ -136,7 +136,7 @@ xdr_encode_pages(struct xdr_buf *xdr, struct page **pages, unsigned int base,
 	xdr->buflen += len;
 	xdr->len += len;
 }
-EXPORT_SYMBOL(xdr_encode_pages);
+EXPORT_SYMBOL_GPL(xdr_encode_pages);
 
 void
 xdr_inline_pages(struct xdr_buf *xdr, unsigned int offset,
@@ -158,7 +158,7 @@ xdr_inline_pages(struct xdr_buf *xdr, unsigned int offset,
 
 	xdr->buflen += len;
 }
-EXPORT_SYMBOL(xdr_inline_pages);
+EXPORT_SYMBOL_GPL(xdr_inline_pages);
 
 /*
  * Helper routines for doing 'memmove' like operations on a struct xdr_buf
@@ -428,7 +428,7 @@ xdr_shift_buf(struct xdr_buf *buf, size_t len)
 {
 	xdr_shrink_bufhead(buf, len);
 }
-EXPORT_SYMBOL(xdr_shift_buf);
+EXPORT_SYMBOL_GPL(xdr_shift_buf);
 
 /**
  * xdr_init_encode - Initialize a struct xdr_stream for sending data.
@@ -465,7 +465,7 @@ void xdr_init_encode(struct xdr_stream *xdr, struct xdr_buf *buf, __be32 *p)
 		iov->iov_len += len;
 	}
 }
-EXPORT_SYMBOL(xdr_init_encode);
+EXPORT_SYMBOL_GPL(xdr_init_encode);
 
 /**
  * xdr_reserve_space - Reserve buffer space for sending
@@ -492,7 +492,7 @@ __be32 * xdr_reserve_space(struct xdr_stream *xdr, size_t nbytes)
 	xdr->buf->len += nbytes;
 	return p;
 }
-EXPORT_SYMBOL(xdr_reserve_space);
+EXPORT_SYMBOL_GPL(xdr_reserve_space);
 
 /**
  * xdr_write_pages - Insert a list of pages into an XDR buffer for sending
@@ -527,7 +527,7 @@ void xdr_write_pages(struct xdr_stream *xdr, struct page **pages, unsigned int b
 	buf->buflen += len;
 	buf->len += len;
 }
-EXPORT_SYMBOL(xdr_write_pages);
+EXPORT_SYMBOL_GPL(xdr_write_pages);
 
 /**
  * xdr_init_decode - Initialize an xdr_stream for decoding data.
@@ -547,7 +547,7 @@ void xdr_init_decode(struct xdr_stream *xdr, struct xdr_buf *buf, __be32 *p)
 	xdr->p = p;
 	xdr->end = (__be32 *)((char *)iov->iov_base + len);
 }
-EXPORT_SYMBOL(xdr_init_decode);
+EXPORT_SYMBOL_GPL(xdr_init_decode);
 
 /**
  * xdr_inline_decode - Retrieve non-page XDR data to decode
@@ -569,7 +569,7 @@ __be32 * xdr_inline_decode(struct xdr_stream *xdr, size_t nbytes)
 	xdr->p = q;
 	return p;
 }
-EXPORT_SYMBOL(xdr_inline_decode);
+EXPORT_SYMBOL_GPL(xdr_inline_decode);
 
 /**
  * xdr_read_pages - Ensure page-based XDR data to decode is aligned at current pointer position
@@ -613,7 +613,7 @@ void xdr_read_pages(struct xdr_stream *xdr, unsigned int len)
 	xdr->p = (__be32 *)((char *)iov->iov_base + padding);
 	xdr->end = (__be32 *)((char *)iov->iov_base + end);
 }
-EXPORT_SYMBOL(xdr_read_pages);
+EXPORT_SYMBOL_GPL(xdr_read_pages);
 
 /**
  * xdr_enter_page - decode data from the XDR page
@@ -638,7 +638,7 @@ void xdr_enter_page(struct xdr_stream *xdr, unsigned int len)
 	xdr->p = (__be32 *)(kaddr + xdr->buf->page_base);
 	xdr->end = (__be32 *)((char *)xdr->p + len);
 }
-EXPORT_SYMBOL(xdr_enter_page);
+EXPORT_SYMBOL_GPL(xdr_enter_page);
 
 static struct kvec empty_iov = {.iov_base = NULL, .iov_len = 0};
 
@@ -650,7 +650,7 @@ xdr_buf_from_iov(struct kvec *iov, struct xdr_buf *buf)
 	buf->page_len = 0;
 	buf->buflen = buf->len = iov->iov_len;
 }
-EXPORT_SYMBOL(xdr_buf_from_iov);
+EXPORT_SYMBOL_GPL(xdr_buf_from_iov);
 
 /* Sets subbuf to the portion of buf of length len beginning base bytes
  * from the start of buf. Returns -1 if base of length are out of bounds. */
@@ -699,7 +699,7 @@ xdr_buf_subsegment(struct xdr_buf *buf, struct xdr_buf *subbuf,
 		return -1;
 	return 0;
 }
-EXPORT_SYMBOL(xdr_buf_subsegment);
+EXPORT_SYMBOL_GPL(xdr_buf_subsegment);
 
 static void __read_bytes_from_xdr_buf(struct xdr_buf *subbuf, void *obj, unsigned int len)
 {
@@ -730,7 +730,7 @@ int read_bytes_from_xdr_buf(struct xdr_buf *buf, unsigned int base, void *obj, u
 	__read_bytes_from_xdr_buf(&subbuf, obj, len);
 	return 0;
 }
-EXPORT_SYMBOL(read_bytes_from_xdr_buf);
+EXPORT_SYMBOL_GPL(read_bytes_from_xdr_buf);
 
 static void __write_bytes_to_xdr_buf(struct xdr_buf *subbuf, void *obj, unsigned int len)
 {
@@ -774,7 +774,7 @@ xdr_decode_word(struct xdr_buf *buf, unsigned int base, u32 *obj)
 	*obj = ntohl(raw);
 	return 0;
 }
-EXPORT_SYMBOL(xdr_decode_word);
+EXPORT_SYMBOL_GPL(xdr_decode_word);
 
 int
 xdr_encode_word(struct xdr_buf *buf, unsigned int base, u32 obj)
@@ -783,7 +783,7 @@ xdr_encode_word(struct xdr_buf *buf, unsigned int base, u32 obj)
 
 	return write_bytes_to_xdr_buf(buf, base, &raw, sizeof(obj));
 }
-EXPORT_SYMBOL(xdr_encode_word);
+EXPORT_SYMBOL_GPL(xdr_encode_word);
 
 /* If the netobj starting offset bytes from the start of xdr_buf is contained
  * entirely in the head or the tail, set object to point to it; otherwise
@@ -821,7 +821,7 @@ int xdr_buf_read_netobj(struct xdr_buf *buf, struct xdr_netobj *obj, unsigned in
 	__read_bytes_from_xdr_buf(&subbuf, obj->data, obj->len);
 	return 0;
 }
-EXPORT_SYMBOL(xdr_buf_read_netobj);
+EXPORT_SYMBOL_GPL(xdr_buf_read_netobj);
 
 /* Returns 0 on success, or else a negative error code. */
 static int
@@ -1027,7 +1027,7 @@ xdr_decode_array2(struct xdr_buf *buf, unsigned int base,
 
 	return xdr_xcode_array2(buf, base, desc, 0);
 }
-EXPORT_SYMBOL(xdr_decode_array2);
+EXPORT_SYMBOL_GPL(xdr_decode_array2);
 
 int
 xdr_encode_array2(struct xdr_buf *buf, unsigned int base,
@@ -1039,7 +1039,7 @@ xdr_encode_array2(struct xdr_buf *buf, unsigned int base,
 
 	return xdr_xcode_array2(buf, base, desc, 1);
 }
-EXPORT_SYMBOL(xdr_encode_array2);
+EXPORT_SYMBOL_GPL(xdr_encode_array2);
 
 int
 xdr_process_buf(struct xdr_buf *buf, unsigned int offset, unsigned int len,
@@ -1106,5 +1106,5 @@ xdr_process_buf(struct xdr_buf *buf, unsigned int offset, unsigned int len,
 out:
 	return ret;
 }
-EXPORT_SYMBOL(xdr_process_buf);
+EXPORT_SYMBOL_GPL(xdr_process_buf);