Merge tag 'nfs-for-4.9-1' of git://git.linux-nfs.org/projects/anna/linux-nfs

Pull NFS client updates from Anna Schumaker:
 "Highlights include:

  Stable bugfixes:
   - sunrpc: fix writ espace race causing stalls
   - NFS: Fix inode corruption in nfs_prime_dcache()
   - NFSv4: Don't report revoked delegations as valid in nfs_have_delegation()
   - NFSv4: nfs4_copy_delegation_stateid() must fail if the delegation is invalid
   - NFSv4: Open state recovery must account for file permission changes
   - NFSv4.2: Fix a reference leak in nfs42_proc_layoutstats_generic

  Features:
   - Add support for tracking multiple layout types with an ordered list
   - Add support for using multiple backchannel threads on the client
   - Add support for pNFS file layout session trunking
   - Delay xprtrdma use of DMA API (for device driver removal)
   - Add support for xprtrdma remote invalidation
   - Add support for larger xprtrdma inline thresholds
   - Use a scatter/gather list for sending xprtrdma RPC calls
   - Add support for the CB_NOTIFY_LOCK callback
   - Improve hashing sunrpc auth_creds by using both uid and gid

  Bugfixes:
   - Fix xprtrdma use of DMA API
   - Validate filenames before adding to the dcache
   - Fix corruption of xdr->nwords in xdr_copy_to_scratch
   - Fix setting buffer length in xdr_set_next_buffer()
   - Don't deadlock the state manager on the SEQUENCE status flags
   - Various delegation and stateid related fixes
   - Retry operations if an interrupted slot receives EREMOTEIO
   - Make nfs boot time y2038 safe"

* tag 'nfs-for-4.9-1' of git://git.linux-nfs.org/projects/anna/linux-nfs: (100 commits)
  NFSv4.2: Fix a reference leak in nfs42_proc_layoutstats_generic
  fs: nfs: Make nfs boot time y2038 safe
  sunrpc: replace generic auth_cred hash with auth-specific function
  sunrpc: add RPCSEC_GSS hash_cred() function
  sunrpc: add auth_unix hash_cred() function
  sunrpc: add generic_auth hash_cred() function
  sunrpc: add hash_cred() function to rpc_authops struct
  Retry operation on EREMOTEIO on an interrupted slot
  pNFS: Fix atime updates on pNFS clients
  sunrpc: queue work on system_power_efficient_wq
  NFSv4.1: Even if the stateid is OK, we may need to recover the open modes
  NFSv4: If recovery failed for a specific open stateid, then don't retry
  NFSv4: Fix retry issues with nfs41_test/free_stateid
  NFSv4: Open state recovery must account for file permission changes
  NFSv4: Mark the lock and open stateids as invalid after freeing them
  NFSv4: Don't test open_stateid unless it is set
  NFSv4: nfs4_do_handle_exception() handle revoke/expiry of a single stateid
  NFS: Always call nfs_inode_find_state_and_recover() when revoking a delegation
  NFSv4: Fix a race when updating an open_stateid
  NFSv4: Fix a race in nfs_inode_reclaim_delegation()
  ...

Documentation/kernel-parameters.txt

@@ -2470,6 +2470,11 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
 	nfsrootdebug	[NFS] enable nfsroot debugging messages.
 			See Documentation/filesystems/nfs/nfsroot.txt.
 
+	nfs.callback_nr_threads=
+			[NFSv4] set the total number of threads that the
+			NFS client will assign to service NFSv4 callback
+			requests.
+
 	nfs.callback_tcpport=
 			[NFS] set the TCP port on which the NFSv4 callback
 			channel should listen.
@@ -2493,6 +2498,13 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
 			of returning the full 64-bit number.
 			The default is to return 64-bit inode numbers.
 
+	nfs.max_session_cb_slots=
+			[NFSv4.1] Sets the maximum number of session
+			slots the client will assign to the callback
+			channel. This determines the maximum number of
+			callbacks the client will process in parallel for
+			a particular server.
+
 	nfs.max_session_slots=
 			[NFSv4.1] Sets the maximum number of session slots
 			the client will attempt to negotiate with the server.

fs/nfs/cache_lib.c

@@ -76,7 +76,7 @@ static void nfs_dns_cache_revisit(struct cache_deferred_req *d, int toomany)
 
 	dreq = container_of(d, struct nfs_cache_defer_req, deferred_req);
 
-	complete_all(&dreq->completion);
+	complete(&dreq->completion);
 	nfs_cache_defer_req_put(dreq);
 }
 

fs/nfs/callback.c

@@ -31,8 +31,6 @@
 struct nfs_callback_data {
 	unsigned int users;
 	struct svc_serv *serv;
-	struct svc_rqst *rqst;
-	struct task_struct *task;
 };
 
 static struct nfs_callback_data nfs_callback_info[NFS4_MAX_MINOR_VERSION + 1];
@@ -89,15 +87,6 @@ nfs4_callback_svc(void *vrqstp)
 	return 0;
 }
 
-/*
- * Prepare to bring up the NFSv4 callback service
- */
-static struct svc_rqst *
-nfs4_callback_up(struct svc_serv *serv)
-{
-	return svc_prepare_thread(serv, &serv->sv_pools[0], NUMA_NO_NODE);
-}
-
 #if defined(CONFIG_NFS_V4_1)
 /*
  * The callback service for NFSv4.1 callbacks
@@ -139,29 +128,6 @@ nfs41_callback_svc(void *vrqstp)
 	return 0;
 }
 
-/*
- * Bring up the NFSv4.1 callback service
- */
-static struct svc_rqst *
-nfs41_callback_up(struct svc_serv *serv)
-{
-	struct svc_rqst *rqstp;
-
-	INIT_LIST_HEAD(&serv->sv_cb_list);
-	spin_lock_init(&serv->sv_cb_lock);
-	init_waitqueue_head(&serv->sv_cb_waitq);
-	rqstp = svc_prepare_thread(serv, &serv->sv_pools[0], NUMA_NO_NODE);
-	dprintk("--> %s return %d\n", __func__, PTR_ERR_OR_ZERO(rqstp));
-	return rqstp;
-}
-
-static void nfs_minorversion_callback_svc_setup(struct svc_serv *serv,
-		struct svc_rqst **rqstpp, int (**callback_svc)(void *vrqstp))
-{
-	*rqstpp = nfs41_callback_up(serv);
-	*callback_svc = nfs41_callback_svc;
-}
-
 static inline void nfs_callback_bc_serv(u32 minorversion, struct rpc_xprt *xprt,
 		struct svc_serv *serv)
 {
@@ -173,13 +139,6 @@ static inline void nfs_callback_bc_serv(u32 minorversion, struct rpc_xprt *xprt,
 		xprt->bc_serv = serv;
 }
 #else
-static void nfs_minorversion_callback_svc_setup(struct svc_serv *serv,
-		struct svc_rqst **rqstpp, int (**callback_svc)(void *vrqstp))
-{
-	*rqstpp = ERR_PTR(-ENOTSUPP);
-	*callback_svc = ERR_PTR(-ENOTSUPP);
-}
-
 static inline void nfs_callback_bc_serv(u32 minorversion, struct rpc_xprt *xprt,
 		struct svc_serv *serv)
 {
@@ -189,45 +148,22 @@ static inline void nfs_callback_bc_serv(u32 minorversion, struct rpc_xprt *xprt,
 static int nfs_callback_start_svc(int minorversion, struct rpc_xprt *xprt,
 				  struct svc_serv *serv)
 {
-	struct svc_rqst *rqstp;
-	int (*callback_svc)(void *vrqstp);
-	struct nfs_callback_data *cb_info = &nfs_callback_info[minorversion];
+	int nrservs = nfs_callback_nr_threads;
 	int ret;
 
 	nfs_callback_bc_serv(minorversion, xprt, serv);
 
-	if (cb_info->task)
-		return 0;
+	if (nrservs < NFS4_MIN_NR_CALLBACK_THREADS)
+		nrservs = NFS4_MIN_NR_CALLBACK_THREADS;
 
-	switch (minorversion) {
-	case 0:
-		/* v4.0 callback setup */
-		rqstp = nfs4_callback_up(serv);
-		callback_svc = nfs4_callback_svc;
-		break;
-	default:
-		nfs_minorversion_callback_svc_setup(serv,
-				&rqstp, &callback_svc);
-	}
-
-	if (IS_ERR(rqstp))
-		return PTR_ERR(rqstp);
-
-	svc_sock_update_bufs(serv);
+	if (serv->sv_nrthreads-1 == nrservs)
+		return 0;
 
-	cb_info->serv = serv;
-	cb_info->rqst = rqstp;
-	cb_info->task = kthread_create(callback_svc, cb_info->rqst,
-				    "nfsv4.%u-svc", minorversion);
-	if (IS_ERR(cb_info->task)) {
-		ret = PTR_ERR(cb_info->task);
-		svc_exit_thread(cb_info->rqst);
-		cb_info->rqst = NULL;
-		cb_info->task = NULL;
+	ret = serv->sv_ops->svo_setup(serv, NULL, nrservs);
+	if (ret) {
+		serv->sv_ops->svo_setup(serv, NULL, 0);
 		return ret;
 	}
-	rqstp->rq_task = cb_info->task;
-	wake_up_process(cb_info->task);
 	dprintk("nfs_callback_up: service started\n");
 	return 0;
 }
@@ -281,19 +217,41 @@ err_bind:
 	return ret;
 }
 
-static struct svc_serv_ops nfs_cb_sv_ops = {
+static struct svc_serv_ops nfs40_cb_sv_ops = {
+	.svo_function		= nfs4_callback_svc,
 	.svo_enqueue_xprt	= svc_xprt_do_enqueue,
+	.svo_setup		= svc_set_num_threads,
+	.svo_module		= THIS_MODULE,
+};
+#if defined(CONFIG_NFS_V4_1)
+static struct svc_serv_ops nfs41_cb_sv_ops = {
+	.svo_function		= nfs41_callback_svc,
+	.svo_enqueue_xprt	= svc_xprt_do_enqueue,
+	.svo_setup		= svc_set_num_threads,
+	.svo_module		= THIS_MODULE,
+};
+
+struct svc_serv_ops *nfs4_cb_sv_ops[] = {
+	[0] = &nfs40_cb_sv_ops,
+	[1] = &nfs41_cb_sv_ops,
+};
+#else
+struct svc_serv_ops *nfs4_cb_sv_ops[] = {
+	[0] = &nfs40_cb_sv_ops,
+	[1] = NULL,
 };
+#endif
 
 static struct svc_serv *nfs_callback_create_svc(int minorversion)
 {
 	struct nfs_callback_data *cb_info = &nfs_callback_info[minorversion];
 	struct svc_serv *serv;
+	struct svc_serv_ops *sv_ops;
 
 	/*
 	 * Check whether we're already up and running.
 	 */
-	if (cb_info->task) {
+	if (cb_info->serv) {
 		/*
 		 * Note: increase service usage, because later in case of error
 		 * svc_destroy() will be called.
@@ -302,6 +260,17 @@ static struct svc_serv *nfs_callback_create_svc(int minorversion)
 		return cb_info->serv;
 	}
 
+	switch (minorversion) {
+	case 0:
+		sv_ops = nfs4_cb_sv_ops[0];
+		break;
+	default:
+		sv_ops = nfs4_cb_sv_ops[1];
+	}
+
+	if (sv_ops == NULL)
+		return ERR_PTR(-ENOTSUPP);
+
 	/*
 	 * Sanity check: if there's no task,
 	 * we should be the first user ...
@@ -310,11 +279,12 @@ static struct svc_serv *nfs_callback_create_svc(int minorversion)
 		printk(KERN_WARNING "nfs_callback_create_svc: no kthread, %d users??\n",
 			cb_info->users);
 
-	serv = svc_create(&nfs4_callback_program, NFS4_CALLBACK_BUFSIZE, &nfs_cb_sv_ops);
+	serv = svc_create(&nfs4_callback_program, NFS4_CALLBACK_BUFSIZE, sv_ops);
 	if (!serv) {
 		printk(KERN_ERR "nfs_callback_create_svc: create service failed\n");
 		return ERR_PTR(-ENOMEM);
 	}
+	cb_info->serv = serv;
 	/* As there is only one thread we need to over-ride the
 	 * default maximum of 80 connections
 	 */
@@ -357,6 +327,8 @@ int nfs_callback_up(u32 minorversion, struct rpc_xprt *xprt)
 	 * thread exits.
 	 */
 err_net:
+	if (!cb_info->users)
+		cb_info->serv = NULL;
 	svc_destroy(serv);
 err_create:
 	mutex_unlock(&nfs_callback_mutex);
@@ -374,18 +346,18 @@ err_start:
 void nfs_callback_down(int minorversion, struct net *net)
 {
 	struct nfs_callback_data *cb_info = &nfs_callback_info[minorversion];
+	struct svc_serv *serv;
 
 	mutex_lock(&nfs_callback_mutex);
-	nfs_callback_down_net(minorversion, cb_info->serv, net);
+	serv = cb_info->serv;
+	nfs_callback_down_net(minorversion, serv, net);
 	cb_info->users--;
-	if (cb_info->users == 0 && cb_info->task != NULL) {
-		kthread_stop(cb_info->task);
-		dprintk("nfs_callback_down: service stopped\n");
-		svc_exit_thread(cb_info->rqst);
+	if (cb_info->users == 0) {
+		svc_get(serv);
+		serv->sv_ops->svo_setup(serv, NULL, 0);
+		svc_destroy(serv);
 		dprintk("nfs_callback_down: service destroyed\n");
 		cb_info->serv = NULL;
-		cb_info->rqst = NULL;
-		cb_info->task = NULL;
 	}
 	mutex_unlock(&nfs_callback_mutex);
 }

fs/nfs/callback.h

@@ -179,6 +179,15 @@ extern __be32 nfs4_callback_devicenotify(
 	struct cb_devicenotifyargs *args,
 	void *dummy, struct cb_process_state *cps);
 
+struct cb_notify_lock_args {
+	struct nfs_fh			cbnl_fh;
+	struct nfs_lowner		cbnl_owner;
+	bool				cbnl_valid;
+};
+
+extern __be32 nfs4_callback_notify_lock(struct cb_notify_lock_args *args,
+					 void *dummy,
+					 struct cb_process_state *cps);
 #endif /* CONFIG_NFS_V4_1 */
 extern int check_gss_callback_principal(struct nfs_client *, struct svc_rqst *);
 extern __be32 nfs4_callback_getattr(struct cb_getattrargs *args,
@@ -198,6 +207,9 @@ extern void nfs_callback_down(int minorversion, struct net *net);
 #define NFS41_BC_MIN_CALLBACKS 1
 #define NFS41_BC_MAX_CALLBACKS 1
 
+#define NFS4_MIN_NR_CALLBACK_THREADS 1
+
 extern unsigned int nfs_callback_set_tcpport;
+extern unsigned short nfs_callback_nr_threads;
 
 #endif /* __LINUX_FS_NFS_CALLBACK_H */

fs/nfs/callback_proc.c

@@ -628,4 +628,20 @@ out:
 	dprintk("%s: exit with status = %d\n", __func__, ntohl(status));
 	return status;
 }
+
+__be32 nfs4_callback_notify_lock(struct cb_notify_lock_args *args, void *dummy,
+				 struct cb_process_state *cps)
+{
+	if (!cps->clp) /* set in cb_sequence */
+		return htonl(NFS4ERR_OP_NOT_IN_SESSION);
+
+	dprintk_rcu("NFS: CB_NOTIFY_LOCK request from %s\n",
+		rpc_peeraddr2str(cps->clp->cl_rpcclient, RPC_DISPLAY_ADDR));
+
+	/* Don't wake anybody if the string looked bogus */
+	if (args->cbnl_valid)
+		__wake_up(&cps->clp->cl_lock_waitq, TASK_NORMAL, 0, args);
+
+	return htonl(NFS4_OK);
+}
 #endif /* CONFIG_NFS_V4_1 */

fs/nfs/callback_xdr.c

@@ -35,6 +35,7 @@
 					 (1 + 3) * 4) // seqid, 3 slotids
 #define CB_OP_RECALLANY_RES_MAXSZ	(CB_OP_HDR_RES_MAXSZ)
 #define CB_OP_RECALLSLOT_RES_MAXSZ	(CB_OP_HDR_RES_MAXSZ)
+#define CB_OP_NOTIFY_LOCK_RES_MAXSZ	(CB_OP_HDR_RES_MAXSZ)
 #endif /* CONFIG_NFS_V4_1 */
 
 #define NFSDBG_FACILITY NFSDBG_CALLBACK
@@ -72,7 +73,7 @@ static int nfs4_encode_void(struct svc_rqst *rqstp, __be32 *p, void *dummy)
 	return xdr_ressize_check(rqstp, p);
 }
 
-static __be32 *read_buf(struct xdr_stream *xdr, int nbytes)
+static __be32 *read_buf(struct xdr_stream *xdr, size_t nbytes)
 {
 	__be32 *p;
 
@@ -534,6 +535,49 @@ static __be32 decode_recallslot_args(struct svc_rqst *rqstp,
 	return 0;
 }
 
+static __be32 decode_lockowner(struct xdr_stream *xdr, struct cb_notify_lock_args *args)
+{
+	__be32		*p;
+	unsigned int	len;
+
+	p = read_buf(xdr, 12);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_BADXDR);
+
+	p = xdr_decode_hyper(p, &args->cbnl_owner.clientid);
+	len = be32_to_cpu(*p);
+
+	p = read_buf(xdr, len);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_BADXDR);
+
+	/* Only try to decode if the length is right */
+	if (len == 20) {
+		p += 2;	/* skip "lock id:" */
+		args->cbnl_owner.s_dev = be32_to_cpu(*p++);
+		xdr_decode_hyper(p, &args->cbnl_owner.id);
+		args->cbnl_valid = true;
+	} else {
+		args->cbnl_owner.s_dev = 0;
+		args->cbnl_owner.id = 0;
+		args->cbnl_valid = false;
+	}
+	return 0;
+}
+
+static __be32 decode_notify_lock_args(struct svc_rqst *rqstp, struct xdr_stream *xdr, struct cb_notify_lock_args *args)
+{
+	__be32 status;
+
+	status = decode_fh(xdr, &args->cbnl_fh);
+	if (unlikely(status != 0))
+		goto out;
+	status = decode_lockowner(xdr, args);
+out:
+	dprintk("%s: exit with status = %d\n", __func__, ntohl(status));
+	return status;
+}
+
 #endif /* CONFIG_NFS_V4_1 */
 
 static __be32 encode_string(struct xdr_stream *xdr, unsigned int len, const char *str)
@@ -746,6 +790,7 @@ preprocess_nfs41_op(int nop, unsigned int op_nr, struct callback_op **op)
 	case OP_CB_RECALL_SLOT:
 	case OP_CB_LAYOUTRECALL:
 	case OP_CB_NOTIFY_DEVICEID:
+	case OP_CB_NOTIFY_LOCK:
 		*op = &callback_ops[op_nr];
 		break;
 
@@ -753,7 +798,6 @@ preprocess_nfs41_op(int nop, unsigned int op_nr, struct callback_op **op)
 	case OP_CB_PUSH_DELEG:
 	case OP_CB_RECALLABLE_OBJ_AVAIL:
 	case OP_CB_WANTS_CANCELLED:
-	case OP_CB_NOTIFY_LOCK:
 		return htonl(NFS4ERR_NOTSUPP);
 
 	default:
@@ -1006,6 +1050,11 @@ static struct callback_op callback_ops[] = {
 		.decode_args = (callback_decode_arg_t)decode_recallslot_args,
 		.res_maxsize = CB_OP_RECALLSLOT_RES_MAXSZ,
 	},
+	[OP_CB_NOTIFY_LOCK] = {
+		.process_op = (callback_process_op_t)nfs4_callback_notify_lock,
+		.decode_args = (callback_decode_arg_t)decode_notify_lock_args,
+		.res_maxsize = CB_OP_NOTIFY_LOCK_RES_MAXSZ,
+	},
 #endif /* CONFIG_NFS_V4_1 */
 };
 

fs/nfs/client.c

@@ -313,7 +313,10 @@ static struct nfs_client *nfs_match_client(const struct nfs_client_initdata *dat
 			continue;
 		/* Match the full socket address */
 		if (!rpc_cmp_addr_port(sap, clap))
-			continue;
+			/* Match all xprt_switch full socket addresses */
+			if (!rpc_clnt_xprt_switch_has_addr(clp->cl_rpcclient,
+							   sap))
+				continue;
 
 		atomic_inc(&clp->cl_count);
 		return clp;
@@ -785,7 +788,8 @@ int nfs_probe_fsinfo(struct nfs_server *server, struct nfs_fh *mntfh, struct nfs
 	}
 
 	fsinfo.fattr = fattr;
-	fsinfo.layouttype = 0;
+	fsinfo.nlayouttypes = 0;
+	memset(fsinfo.layouttype, 0, sizeof(fsinfo.layouttype));
 	error = clp->rpc_ops->fsinfo(server, mntfh, &fsinfo);
 	if (error < 0)
 		goto out_error;
@@ -1078,7 +1082,7 @@ void nfs_clients_init(struct net *net)
 	idr_init(&nn->cb_ident_idr);
 #endif
 	spin_lock_init(&nn->nfs_client_lock);
-	nn->boot_time = CURRENT_TIME;
+	nn->boot_time = ktime_get_real();
 }
 
 #ifdef CONFIG_PROC_FS

fs/nfs/delegation.c

@@ -41,6 +41,17 @@ void nfs_mark_delegation_referenced(struct nfs_delegation *delegation)
 	set_bit(NFS_DELEGATION_REFERENCED, &delegation->flags);
 }
 
+static bool
+nfs4_is_valid_delegation(const struct nfs_delegation *delegation,
+		fmode_t flags)
+{
+	if (delegation != NULL && (delegation->type & flags) == flags &&
+	    !test_bit(NFS_DELEGATION_REVOKED, &delegation->flags) &&
+	    !test_bit(NFS_DELEGATION_RETURNING, &delegation->flags))
+		return true;
+	return false;
+}
+
 static int
 nfs4_do_check_delegation(struct inode *inode, fmode_t flags, bool mark)
 {
@@ -50,8 +61,7 @@ nfs4_do_check_delegation(struct inode *inode, fmode_t flags, bool mark)
 	flags &= FMODE_READ|FMODE_WRITE;
 	rcu_read_lock();
 	delegation = rcu_dereference(NFS_I(inode)->delegation);
-	if (delegation != NULL && (delegation->type & flags) == flags &&
-	    !test_bit(NFS_DELEGATION_RETURNING, &delegation->flags)) {
+	if (nfs4_is_valid_delegation(delegation, flags)) {
 		if (mark)
 			nfs_mark_delegation_referenced(delegation);
 		ret = 1;
@@ -185,15 +195,13 @@ void nfs_inode_reclaim_delegation(struct inode *inode, struct rpc_cred *cred,
 			rcu_read_unlock();
 			put_rpccred(oldcred);
 			trace_nfs4_reclaim_delegation(inode, res->delegation_type);
-		} else {
-			/* We appear to have raced with a delegation return. */
-			spin_unlock(&delegation->lock);
-			rcu_read_unlock();
-			nfs_inode_set_delegation(inode, cred, res);
+			return;
 		}
-	} else {
-		rcu_read_unlock();
+		/* We appear to have raced with a delegation return. */
+		spin_unlock(&delegation->lock);
 	}
+	rcu_read_unlock();
+	nfs_inode_set_delegation(inode, cred, res);
 }
 
 static int nfs_do_return_delegation(struct inode *inode, struct nfs_delegation *delegation, int issync)
@@ -642,28 +650,49 @@ static void nfs_client_mark_return_unused_delegation_types(struct nfs_client *cl
 	rcu_read_unlock();
 }
 
-static void nfs_revoke_delegation(struct inode *inode)
+static void nfs_mark_delegation_revoked(struct nfs_server *server,
+		struct nfs_delegation *delegation)
+{
+	set_bit(NFS_DELEGATION_REVOKED, &delegation->flags);
+	delegation->stateid.type = NFS4_INVALID_STATEID_TYPE;
+	nfs_mark_return_delegation(server, delegation);
+}
+
+static bool nfs_revoke_delegation(struct inode *inode,
+		const nfs4_stateid *stateid)
 {
 	struct nfs_delegation *delegation;
+	nfs4_stateid tmp;
+	bool ret = false;
+
 	rcu_read_lock();
 	delegation = rcu_dereference(NFS_I(inode)->delegation);
-	if (delegation != NULL) {
-		set_bit(NFS_DELEGATION_REVOKED, &delegation->flags);
-		nfs_mark_return_delegation(NFS_SERVER(inode), delegation);
-	}
+	if (delegation == NULL)
+		goto out;
+	if (stateid == NULL) {
+		nfs4_stateid_copy(&tmp, &delegation->stateid);
+		stateid = &tmp;
+	} else if (!nfs4_stateid_match(stateid, &delegation->stateid))
+		goto out;
+	nfs_mark_delegation_revoked(NFS_SERVER(inode), delegation);
+	ret = true;
+out:
 	rcu_read_unlock();
+	if (ret)
+		nfs_inode_find_state_and_recover(inode, stateid);
+	return ret;
 }
 
-void nfs_remove_bad_delegation(struct inode *inode)
+void nfs_remove_bad_delegation(struct inode *inode,
+		const nfs4_stateid *stateid)
 {
 	struct nfs_delegation *delegation;
 
-	nfs_revoke_delegation(inode);
+	if (!nfs_revoke_delegation(inode, stateid))
+		return;
 	delegation = nfs_inode_detach_delegation(inode);
-	if (delegation) {
-		nfs_inode_find_state_and_recover(inode, &delegation->stateid);
+	if (delegation)
 		nfs_free_delegation(delegation);
-	}
 }
 EXPORT_SYMBOL_GPL(nfs_remove_bad_delegation);
 
@@ -786,8 +815,15 @@ static void nfs_delegation_mark_reclaim_server(struct nfs_server *server)
 {
 	struct nfs_delegation *delegation;
 
-	list_for_each_entry_rcu(delegation, &server->delegations, super_list)
+	list_for_each_entry_rcu(delegation, &server->delegations, super_list) {
+		/*
+		 * If the delegation may have been admin revoked, then we
+		 * cannot reclaim it.
+		 */
+		if (test_bit(NFS_DELEGATION_TEST_EXPIRED, &delegation->flags))
+			continue;
 		set_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags);
+	}
 }
 
 /**
@@ -851,6 +887,141 @@ restart:
 	rcu_read_unlock();
 }
 
+static inline bool nfs4_server_rebooted(const struct nfs_client *clp)
+{
+	return (clp->cl_state & (BIT(NFS4CLNT_CHECK_LEASE) |
+				BIT(NFS4CLNT_LEASE_EXPIRED) |
+				BIT(NFS4CLNT_SESSION_RESET))) != 0;
+}
+
+static void nfs_mark_test_expired_delegation(struct nfs_server *server,
+	    struct nfs_delegation *delegation)
+{
+	if (delegation->stateid.type == NFS4_INVALID_STATEID_TYPE)
+		return;
+	clear_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags);
+	set_bit(NFS_DELEGATION_TEST_EXPIRED, &delegation->flags);
+	set_bit(NFS4CLNT_DELEGATION_EXPIRED, &server->nfs_client->cl_state);
+}
+
+static void nfs_inode_mark_test_expired_delegation(struct nfs_server *server,
+		struct inode *inode)
+{
+	struct nfs_delegation *delegation;
+
+	rcu_read_lock();
+	delegation = rcu_dereference(NFS_I(inode)->delegation);
+	if (delegation)
+		nfs_mark_test_expired_delegation(server, delegation);
+	rcu_read_unlock();
+
+}
+
+static void nfs_delegation_mark_test_expired_server(struct nfs_server *server)
+{
+	struct nfs_delegation *delegation;
+
+	list_for_each_entry_rcu(delegation, &server->delegations, super_list)
+		nfs_mark_test_expired_delegation(server, delegation);
+}
+
+/**
+ * nfs_mark_test_expired_all_delegations - mark all delegations for testing
+ * @clp: nfs_client to process
+ *
+ * Iterates through all the delegations associated with this server and
+ * marks them as needing to be checked for validity.
+ */
+void nfs_mark_test_expired_all_delegations(struct nfs_client *clp)
+{
+	struct nfs_server *server;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link)
+		nfs_delegation_mark_test_expired_server(server);
+	rcu_read_unlock();
+}
+
+/**
+ * nfs_reap_expired_delegations - reap expired delegations
+ * @clp: nfs_client to process
+ *
+ * Iterates through all the delegations associated with this server and
+ * checks if they have may have been revoked. This function is usually
+ * expected to be called in cases where the server may have lost its
+ * lease.
+ */
+void nfs_reap_expired_delegations(struct nfs_client *clp)
+{
+	const struct nfs4_minor_version_ops *ops = clp->cl_mvops;
+	struct nfs_delegation *delegation;
+	struct nfs_server *server;
+	struct inode *inode;
+	struct rpc_cred *cred;
+	nfs4_stateid stateid;
+
+restart:
+	rcu_read_lock();
+	list_for_each_entry_rcu(server, &clp->cl_superblocks, client_link) {
+		list_for_each_entry_rcu(delegation, &server->delegations,
+								super_list) {
+			if (test_bit(NFS_DELEGATION_RETURNING,
+						&delegation->flags))
+				continue;
+			if (test_bit(NFS_DELEGATION_TEST_EXPIRED,
+						&delegation->flags) == 0)
+				continue;
+			if (!nfs_sb_active(server->super))
+				continue;
+			inode = nfs_delegation_grab_inode(delegation);
+			if (inode == NULL) {
+				rcu_read_unlock();
+				nfs_sb_deactive(server->super);
+				goto restart;
+			}
+			cred = get_rpccred_rcu(delegation->cred);
+			nfs4_stateid_copy(&stateid, &delegation->stateid);
+			clear_bit(NFS_DELEGATION_TEST_EXPIRED, &delegation->flags);
+			rcu_read_unlock();
+			if (cred != NULL &&
+			    ops->test_and_free_expired(server, &stateid, cred) < 0) {
+				nfs_revoke_delegation(inode, &stateid);
+				nfs_inode_find_state_and_recover(inode, &stateid);
+			}
+			put_rpccred(cred);
+			if (nfs4_server_rebooted(clp)) {
+				nfs_inode_mark_test_expired_delegation(server,inode);
+				iput(inode);
+				nfs_sb_deactive(server->super);
+				return;
+			}
+			iput(inode);
+			nfs_sb_deactive(server->super);
+			goto restart;
+		}
+	}
+	rcu_read_unlock();
+}
+
+void nfs_inode_find_delegation_state_and_recover(struct inode *inode,
+		const nfs4_stateid *stateid)
+{
+	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
+	struct nfs_delegation *delegation;
+	bool found = false;
+
+	rcu_read_lock();
+	delegation = rcu_dereference(NFS_I(inode)->delegation);
+	if (delegation &&
+	    nfs4_stateid_match_other(&delegation->stateid, stateid)) {
+		nfs_mark_test_expired_delegation(NFS_SERVER(inode), delegation);
+		found = true;
+	}
+	rcu_read_unlock();
+	if (found)
+		nfs4_schedule_state_manager(clp);
+}
+
 /**
  * nfs_delegations_present - check for existence of delegations
  * @clp: client state handle
@@ -893,7 +1064,7 @@ bool nfs4_copy_delegation_stateid(struct inode *inode, fmode_t flags,
 	flags &= FMODE_READ|FMODE_WRITE;
 	rcu_read_lock();
 	delegation = rcu_dereference(nfsi->delegation);
-	ret = (delegation != NULL && (delegation->type & flags) == flags);
+	ret = nfs4_is_valid_delegation(delegation, flags);
 	if (ret) {
 		nfs4_stateid_copy(dst, &delegation->stateid);
 		nfs_mark_delegation_referenced(delegation);

fs/nfs/delegation.h

@@ -32,6 +32,7 @@ enum {
 	NFS_DELEGATION_REFERENCED,
 	NFS_DELEGATION_RETURNING,
 	NFS_DELEGATION_REVOKED,
+	NFS_DELEGATION_TEST_EXPIRED,
 };
 
 int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res);
@@ -47,11 +48,14 @@ void nfs_expire_unused_delegation_types(struct nfs_client *clp, fmode_t flags);
 void nfs_expire_unreferenced_delegations(struct nfs_client *clp);
 int nfs_client_return_marked_delegations(struct nfs_client *clp);
 int nfs_delegations_present(struct nfs_client *clp);
-void nfs_remove_bad_delegation(struct inode *inode);
+void nfs_remove_bad_delegation(struct inode *inode, const nfs4_stateid *stateid);
 
 void nfs_delegation_mark_reclaim(struct nfs_client *clp);
 void nfs_delegation_reap_unclaimed(struct nfs_client *clp);
 
+void nfs_mark_test_expired_all_delegations(struct nfs_client *clp);
+void nfs_reap_expired_delegations(struct nfs_client *clp);
+
 /* NFSv4 delegation-related procedures */
 int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync);
 int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid, fmode_t type);
@@ -62,6 +66,8 @@ void nfs_mark_delegation_referenced(struct nfs_delegation *delegation);
 int nfs4_have_delegation(struct inode *inode, fmode_t flags);
 int nfs4_check_delegation(struct inode *inode, fmode_t flags);
 bool nfs4_delegation_flush_on_close(const struct inode *inode);
+void nfs_inode_find_delegation_state_and_recover(struct inode *inode,
+		const nfs4_stateid *stateid);
 
 #endif
 

fs/nfs/dir.c

@@ -435,11 +435,11 @@ int nfs_same_file(struct dentry *dentry, struct nfs_entry *entry)
 		return 0;
 
 	nfsi = NFS_I(inode);
-	if (entry->fattr->fileid == nfsi->fileid)
-		return 1;
-	if (nfs_compare_fh(entry->fh, &nfsi->fh) == 0)
-		return 1;
-	return 0;
+	if (entry->fattr->fileid != nfsi->fileid)
+		return 0;
+	if (entry->fh->size && nfs_compare_fh(entry->fh, &nfsi->fh) != 0)
+		return 0;
+	return 1;
 }
 
 static
@@ -496,6 +496,14 @@ void nfs_prime_dcache(struct dentry *parent, struct nfs_entry *entry)
 		return;
 	if (!(entry->fattr->valid & NFS_ATTR_FATTR_FSID))
 		return;
+	if (filename.len == 0)
+		return;
+	/* Validate that the name doesn't contain any illegal '\0' */
+	if (strnlen(filename.name, filename.len) != filename.len)
+		return;
+	/* ...or '/' */
+	if (strnchr(filename.name, filename.len, '/'))
+		return;
 	if (filename.name[0] == '.') {
 		if (filename.len == 1)
 			return;
@@ -517,6 +525,8 @@ again:
 					&entry->fattr->fsid))
 			goto out;
 		if (nfs_same_file(dentry, entry)) {
+			if (!entry->fh->size)
+				goto out;
 			nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
 			status = nfs_refresh_inode(d_inode(dentry), entry->fattr);
 			if (!status)
@@ -529,6 +539,10 @@ again:
 			goto again;
 		}
 	}
+	if (!entry->fh->size) {
+		d_lookup_done(dentry);
+		goto out;
+	}
 
 	inode = nfs_fhget(dentry->d_sb, entry->fh, entry->fattr, entry->label);
 	alias = d_splice_alias(inode, dentry);

fs/nfs/direct.c

@@ -387,7 +387,7 @@ static void nfs_direct_complete(struct nfs_direct_req *dreq)
 		dreq->iocb->ki_complete(dreq->iocb, res, 0);
 	}
 
-	complete_all(&dreq->completion);
+	complete(&dreq->completion);
 
 	nfs_direct_req_release(dreq);
 }

fs/nfs/file.c

@@ -520,7 +520,9 @@ const struct address_space_operations nfs_file_aops = {
 	.invalidatepage = nfs_invalidate_page,
 	.releasepage = nfs_release_page,
 	.direct_IO = nfs_direct_IO,
+#ifdef CONFIG_MIGRATION
 	.migratepage = nfs_migrate_page,
+#endif
 	.launder_page = nfs_launder_page,
 	.is_dirty_writeback = nfs_check_dirty_writeback,
 	.error_remove_page = generic_error_remove_page,
@@ -685,11 +687,6 @@ out_noconflict:
 	goto out;
 }
 
-static int do_vfs_lock(struct file *file, struct file_lock *fl)
-{
-	return locks_lock_file_wait(file, fl);
-}
-
 static int
 do_unlk(struct file *filp, int cmd, struct file_lock *fl, int is_local)
 {
@@ -722,7 +719,7 @@ do_unlk(struct file *filp, int cmd, struct file_lock *fl, int is_local)
 	if (!is_local)
 		status = NFS_PROTO(inode)->lock(filp, cmd, fl);
 	else
-		status = do_vfs_lock(filp, fl);
+		status = locks_lock_file_wait(filp, fl);
 	return status;
 }
 
@@ -747,7 +744,7 @@ do_setlk(struct file *filp, int cmd, struct file_lock *fl, int is_local)
 	if (!is_local)
 		status = NFS_PROTO(inode)->lock(filp, cmd, fl);
 	else
-		status = do_vfs_lock(filp, fl);
+		status = locks_lock_file_wait(filp, fl);
 	if (status < 0)
 		goto out;
 

fs/nfs/flexfilelayout/flexfilelayout.c

@@ -1080,7 +1080,7 @@ static int ff_layout_async_handle_error_v4(struct rpc_task *task,
 	case -NFS4ERR_BAD_STATEID:
 		if (state == NULL)
 			break;
-		nfs_remove_bad_delegation(state->inode);
+		nfs_remove_bad_delegation(state->inode, NULL);
 	case -NFS4ERR_OPENMODE:
 		if (state == NULL)
 			break;

fs/nfs/internal.h

@@ -534,12 +534,9 @@ void nfs_clear_pnfs_ds_commit_verifiers(struct pnfs_ds_commit_info *cinfo)
 }
 #endif
 
-
 #ifdef CONFIG_MIGRATION
 extern int nfs_migrate_page(struct address_space *,
 		struct page *, struct page *, enum migrate_mode);
-#else
-#define nfs_migrate_page NULL
 #endif
 
 static inline int
@@ -562,7 +559,6 @@ void nfs_init_cinfo_from_dreq(struct nfs_commit_info *cinfo,
 extern ssize_t nfs_dreq_bytes_left(struct nfs_direct_req *dreq);
 
 /* nfs4proc.c */
-extern void __nfs4_read_done_cb(struct nfs_pgio_header *);
 extern struct nfs_client *nfs4_init_client(struct nfs_client *clp,
 			    const struct nfs_client_initdata *);
 extern int nfs40_walk_client_list(struct nfs_client *clp,
@@ -571,6 +567,9 @@ extern int nfs40_walk_client_list(struct nfs_client *clp,
 extern int nfs41_walk_client_list(struct nfs_client *clp,
 				struct nfs_client **result,
 				struct rpc_cred *cred);
+extern int nfs4_test_session_trunk(struct rpc_clnt *,
+				struct rpc_xprt *,
+				void *);
 
 static inline struct inode *nfs_igrab_and_active(struct inode *inode)
 {

fs/nfs/netns.h

@@ -29,7 +29,7 @@ struct nfs_net {
 	int cb_users[NFS4_MAX_MINOR_VERSION + 1];
 #endif
 	spinlock_t nfs_client_lock;
-	struct timespec boot_time;
+	ktime_t boot_time;
 #ifdef CONFIG_PROC_FS
 	struct proc_dir_entry *proc_nfsfs;
 #endif

fs/nfs/nfs42proc.c

@@ -443,6 +443,7 @@ int nfs42_proc_layoutstats_generic(struct nfs_server *server,
 	task = rpc_run_task(&task_setup);
 	if (IS_ERR(task))
 		return PTR_ERR(task);
+	rpc_put_task(task);
 	return 0;
 }
 

fs/nfs/nfs4_fs.h

@@ -39,6 +39,7 @@ enum nfs4_client_state {
 	NFS4CLNT_BIND_CONN_TO_SESSION,
 	NFS4CLNT_MOVED,
 	NFS4CLNT_LEASE_MOVED,
+	NFS4CLNT_DELEGATION_EXPIRED,
 };
 
 #define NFS4_RENEW_TIMEOUT		0x01
@@ -57,8 +58,11 @@ struct nfs4_minor_version_ops {
 			struct nfs_fsinfo *);
 	void	(*free_lock_state)(struct nfs_server *,
 			struct nfs4_lock_state *);
+	int	(*test_and_free_expired)(struct nfs_server *,
+			nfs4_stateid *, struct rpc_cred *);
 	struct nfs_seqid *
 		(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
+	int	(*session_trunk)(struct rpc_clnt *, struct rpc_xprt *, void *);
 	const struct rpc_call_ops *call_sync_ops;
 	const struct nfs4_state_recovery_ops *reboot_recovery_ops;
 	const struct nfs4_state_recovery_ops *nograce_recovery_ops;
@@ -156,6 +160,7 @@ enum {
 	NFS_STATE_RECLAIM_NOGRACE,	/* OPEN stateid needs to recover state */
 	NFS_STATE_POSIX_LOCKS,		/* Posix locks are supported */
 	NFS_STATE_RECOVERY_FAILED,	/* OPEN stateid state recovery failed */
+	NFS_STATE_MAY_NOTIFY_LOCK,	/* server may CB_NOTIFY_LOCK */
 };
 
 struct nfs4_state {
@@ -203,6 +208,11 @@ struct nfs4_state_recovery_ops {
 		struct rpc_cred *);
 };
 
+struct nfs4_add_xprt_data {
+	struct nfs_client	*clp;
+	struct rpc_cred		*cred;
+};
+
 struct nfs4_state_maintenance_ops {
 	int (*sched_state_renewal)(struct nfs_client *, struct rpc_cred *, unsigned);
 	struct rpc_cred * (*get_state_renewal_cred_locked)(struct nfs_client *);
@@ -278,6 +288,8 @@ extern int nfs4_proc_get_lease_time(struct nfs_client *clp,
 		struct nfs_fsinfo *fsinfo);
 extern int nfs4_proc_layoutcommit(struct nfs4_layoutcommit_data *data,
 				  bool sync);
+extern int nfs4_detect_session_trunking(struct nfs_client *clp,
+		struct nfs41_exchange_id_res *res, struct rpc_xprt *xprt);
 
 static inline bool
 is_ds_only_client(struct nfs_client *clp)
@@ -439,7 +451,7 @@ extern void nfs4_schedule_path_down_recovery(struct nfs_client *clp);
 extern int nfs4_schedule_stateid_recovery(const struct nfs_server *, struct nfs4_state *);
 extern int nfs4_schedule_migration_recovery(const struct nfs_server *);
 extern void nfs4_schedule_lease_moved_recovery(struct nfs_client *);
-extern void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags);
+extern void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags, bool);
 extern void nfs41_handle_server_scope(struct nfs_client *,
 				      struct nfs41_server_scope **);
 extern void nfs4_put_lock_state(struct nfs4_lock_state *lsp);
@@ -471,6 +483,7 @@ extern struct nfs_subversion nfs_v4;
 struct dentry *nfs4_try_mount(int, const char *, struct nfs_mount_info *, struct nfs_subversion *);
 extern bool nfs4_disable_idmapping;
 extern unsigned short max_session_slots;
+extern unsigned short max_session_cb_slots;
 extern unsigned short send_implementation_id;
 extern bool recover_lost_locks;
 

fs/nfs/nfs4client.c

@@ -199,6 +199,9 @@ struct nfs_client *nfs4_alloc_client(const struct nfs_client_initdata *cl_init)
 	clp->cl_minorversion = cl_init->minorversion;
 	clp->cl_mvops = nfs_v4_minor_ops[cl_init->minorversion];
 	clp->cl_mig_gen = 1;
+#if IS_ENABLED(CONFIG_NFS_V4_1)
+	init_waitqueue_head(&clp->cl_lock_waitq);
+#endif
 	return clp;
 
 error:
@@ -562,15 +565,15 @@ out:
 /*
  * Returns true if the client IDs match
  */
-static bool nfs4_match_clientids(struct nfs_client *a, struct nfs_client *b)
+static bool nfs4_match_clientids(u64 a, u64 b)
 {
-	if (a->cl_clientid != b->cl_clientid) {
+	if (a != b) {
 		dprintk("NFS: --> %s client ID %llx does not match %llx\n",
-			__func__, a->cl_clientid, b->cl_clientid);
+			__func__, a, b);
 		return false;
 	}
 	dprintk("NFS: --> %s client ID %llx matches %llx\n",
-		__func__, a->cl_clientid, b->cl_clientid);
+		__func__, a, b);
 	return true;
 }
 
@@ -578,17 +581,15 @@ static bool nfs4_match_clientids(struct nfs_client *a, struct nfs_client *b)
  * Returns true if the server major ids match
  */
 static bool
-nfs4_check_clientid_trunking(struct nfs_client *a, struct nfs_client *b)
+nfs4_check_serverowner_major_id(struct nfs41_server_owner *o1,
+				struct nfs41_server_owner *o2)
 {
-	struct nfs41_server_owner *o1 = a->cl_serverowner;
-	struct nfs41_server_owner *o2 = b->cl_serverowner;
-
 	if (o1->major_id_sz != o2->major_id_sz)
 		goto out_major_mismatch;
 	if (memcmp(o1->major_id, o2->major_id, o1->major_id_sz) != 0)
 		goto out_major_mismatch;
 
-	dprintk("NFS: --> %s server owners match\n", __func__);
+	dprintk("NFS: --> %s server owner major IDs match\n", __func__);
 	return true;
 
 out_major_mismatch:
@@ -597,6 +598,100 @@ out_major_mismatch:
 	return false;
 }
 
+/*
+ * Returns true if server minor ids match
+ */
+static bool
+nfs4_check_serverowner_minor_id(struct nfs41_server_owner *o1,
+				struct nfs41_server_owner *o2)
+{
+	/* Check eir_server_owner so_minor_id */
+	if (o1->minor_id != o2->minor_id)
+		goto out_minor_mismatch;
+
+	dprintk("NFS: --> %s server owner minor IDs match\n", __func__);
+	return true;
+
+out_minor_mismatch:
+	dprintk("NFS: --> %s server owner minor IDs do not match\n", __func__);
+	return false;
+}
+
+/*
+ * Returns true if the server scopes match
+ */
+static bool
+nfs4_check_server_scope(struct nfs41_server_scope *s1,
+			struct nfs41_server_scope *s2)
+{
+	if (s1->server_scope_sz != s2->server_scope_sz)
+		goto out_scope_mismatch;
+	if (memcmp(s1->server_scope, s2->server_scope,
+		   s1->server_scope_sz) != 0)
+		goto out_scope_mismatch;
+
+	dprintk("NFS: --> %s server scopes match\n", __func__);
+	return true;
+
+out_scope_mismatch:
+	dprintk("NFS: --> %s server scopes do not match\n",
+		__func__);
+	return false;
+}
+
+/**
+ * nfs4_detect_session_trunking - Checks for session trunking.
+ *
+ * Called after a successful EXCHANGE_ID on a multi-addr connection.
+ * Upon success, add the transport.
+ *
+ * @clp:    original mount nfs_client
+ * @res:    result structure from an exchange_id using the original mount
+ *          nfs_client with a new multi_addr transport
+ *
+ * Returns zero on success, otherwise -EINVAL
+ *
+ * Note: since the exchange_id for the new multi_addr transport uses the
+ * same nfs_client from the original mount, the cl_owner_id is reused,
+ * so eir_clientowner is the same.
+ */
+int nfs4_detect_session_trunking(struct nfs_client *clp,
+				 struct nfs41_exchange_id_res *res,
+				 struct rpc_xprt *xprt)
+{
+	/* Check eir_clientid */
+	if (!nfs4_match_clientids(clp->cl_clientid, res->clientid))
+		goto out_err;
+
+	/* Check eir_server_owner so_major_id */
+	if (!nfs4_check_serverowner_major_id(clp->cl_serverowner,
+					     res->server_owner))
+		goto out_err;
+
+	/* Check eir_server_owner so_minor_id */
+	if (!nfs4_check_serverowner_minor_id(clp->cl_serverowner,
+					     res->server_owner))
+		goto out_err;
+
+	/* Check eir_server_scope */
+	if (!nfs4_check_server_scope(clp->cl_serverscope, res->server_scope))
+		goto out_err;
+
+	/* Session trunking passed, add the xprt */
+	rpc_clnt_xprt_switch_add_xprt(clp->cl_rpcclient, xprt);
+
+	pr_info("NFS:  %s: Session trunking succeeded for %s\n",
+		clp->cl_hostname,
+		xprt->address_strings[RPC_DISPLAY_ADDR]);
+
+	return 0;
+out_err:
+	pr_info("NFS:  %s: Session trunking failed for %s\n", clp->cl_hostname,
+		xprt->address_strings[RPC_DISPLAY_ADDR]);
+
+	return -EINVAL;
+}
+
 /**
  * nfs41_walk_client_list - Find nfs_client that matches a client/server owner
  *
@@ -650,7 +745,7 @@ int nfs41_walk_client_list(struct nfs_client *new,
 		if (pos->cl_cons_state != NFS_CS_READY)
 			continue;
 
-		if (!nfs4_match_clientids(pos, new))
+		if (!nfs4_match_clientids(pos->cl_clientid, new->cl_clientid))
 			continue;
 
 		/*
@@ -658,7 +753,8 @@ int nfs41_walk_client_list(struct nfs_client *new,
 		 * client id trunking. In either case, we want to fall back
 		 * to using the existing nfs_client.
 		 */
-		if (!nfs4_check_clientid_trunking(pos, new))
+		if (!nfs4_check_serverowner_major_id(pos->cl_serverowner,
+						     new->cl_serverowner))
 			continue;
 
 		/* Unlike NFSv4.0, we know that NFSv4.1 always uses the

fs/nfs/nfs4proc.c

@@ -99,8 +99,8 @@ static int nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
 #ifdef CONFIG_NFS_V4_1
 static int nfs41_test_stateid(struct nfs_server *, nfs4_stateid *,
 		struct rpc_cred *);
-static int nfs41_free_stateid(struct nfs_server *, nfs4_stateid *,
-		struct rpc_cred *);
+static int nfs41_free_stateid(struct nfs_server *, const nfs4_stateid *,
+		struct rpc_cred *, bool);
 #endif
 
 #ifdef CONFIG_NFS_V4_SECURITY_LABEL
@@ -328,6 +328,33 @@ static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dent
 	kunmap_atomic(start);
 }
 
+static void nfs4_test_and_free_stateid(struct nfs_server *server,
+		nfs4_stateid *stateid,
+		struct rpc_cred *cred)
+{
+	const struct nfs4_minor_version_ops *ops = server->nfs_client->cl_mvops;
+
+	ops->test_and_free_expired(server, stateid, cred);
+}
+
+static void __nfs4_free_revoked_stateid(struct nfs_server *server,
+		nfs4_stateid *stateid,
+		struct rpc_cred *cred)
+{
+	stateid->type = NFS4_REVOKED_STATEID_TYPE;
+	nfs4_test_and_free_stateid(server, stateid, cred);
+}
+
+static void nfs4_free_revoked_stateid(struct nfs_server *server,
+		const nfs4_stateid *stateid,
+		struct rpc_cred *cred)
+{
+	nfs4_stateid tmp;
+
+	nfs4_stateid_copy(&tmp, stateid);
+	__nfs4_free_revoked_stateid(server, &tmp, cred);
+}
+
 static long nfs4_update_delay(long *timeout)
 {
 	long ret;
@@ -370,13 +397,23 @@ static int nfs4_do_handle_exception(struct nfs_server *server,
 	exception->delay = 0;
 	exception->recovering = 0;
 	exception->retry = 0;
+
+	if (stateid == NULL && state != NULL)
+		stateid = &state->stateid;
+
 	switch(errorcode) {
 		case 0:
 			return 0;
-		case -NFS4ERR_OPENMODE:
 		case -NFS4ERR_DELEG_REVOKED:
 		case -NFS4ERR_ADMIN_REVOKED:
+		case -NFS4ERR_EXPIRED:
 		case -NFS4ERR_BAD_STATEID:
+			if (inode != NULL && stateid != NULL) {
+				nfs_inode_find_state_and_recover(inode,
+						stateid);
+				goto wait_on_recovery;
+			}
+		case -NFS4ERR_OPENMODE:
 			if (inode) {
 				int err;
 
@@ -395,12 +432,6 @@ static int nfs4_do_handle_exception(struct nfs_server *server,
 			if (ret < 0)
 				break;
 			goto wait_on_recovery;
-		case -NFS4ERR_EXPIRED:
-			if (state != NULL) {
-				ret = nfs4_schedule_stateid_recovery(server, state);
-				if (ret < 0)
-					break;
-			}
 		case -NFS4ERR_STALE_STATEID:
 		case -NFS4ERR_STALE_CLIENTID:
 			nfs4_schedule_lease_recovery(clp);
@@ -616,6 +647,7 @@ int nfs40_setup_sequence(struct nfs4_slot_table *tbl,
 	}
 	spin_unlock(&tbl->slot_tbl_lock);
 
+	slot->privileged = args->sa_privileged ? 1 : 0;
 	args->sa_slot = slot;
 	res->sr_slot = slot;
 
@@ -723,12 +755,20 @@ static int nfs41_sequence_process(struct rpc_task *task,
 	/* Check the SEQUENCE operation status */
 	switch (res->sr_status) {
 	case 0:
+		/* If previous op on slot was interrupted and we reused
+		 * the seq# and got a reply from the cache, then retry
+		 */
+		if (task->tk_status == -EREMOTEIO && interrupted) {
+			++slot->seq_nr;
+			goto retry_nowait;
+		}
 		/* Update the slot's sequence and clientid lease timer */
 		slot->seq_done = 1;
 		clp = session->clp;
 		do_renew_lease(clp, res->sr_timestamp);
 		/* Check sequence flags */
-		nfs41_handle_sequence_flag_errors(clp, res->sr_status_flags);
+		nfs41_handle_sequence_flag_errors(clp, res->sr_status_flags,
+				!!slot->privileged);
 		nfs41_update_target_slotid(slot->table, slot, res);
 		break;
 	case 1:
@@ -875,6 +915,7 @@ int nfs41_setup_sequence(struct nfs4_session *session,
 	}
 	spin_unlock(&tbl->slot_tbl_lock);
 
+	slot->privileged = args->sa_privileged ? 1 : 0;
 	args->sa_slot = slot;
 
 	dprintk("<-- %s slotid=%u seqid=%u\n", __func__,
@@ -1353,6 +1394,19 @@ static void update_open_stateflags(struct nfs4_state *state, fmode_t fmode)
 	nfs4_state_set_mode_locked(state, state->state | fmode);
 }
 
+#ifdef CONFIG_NFS_V4_1
+static bool nfs_open_stateid_recover_openmode(struct nfs4_state *state)
+{
+	if (state->n_rdonly && !test_bit(NFS_O_RDONLY_STATE, &state->flags))
+		return true;
+	if (state->n_wronly && !test_bit(NFS_O_WRONLY_STATE, &state->flags))
+		return true;
+	if (state->n_rdwr && !test_bit(NFS_O_RDWR_STATE, &state->flags))
+		return true;
+	return false;
+}
+#endif /* CONFIG_NFS_V4_1 */
+
 static void nfs_test_and_clear_all_open_stateid(struct nfs4_state *state)
 {
 	struct nfs_client *clp = state->owner->so_server->nfs_client;
@@ -1369,11 +1423,12 @@ static void nfs_test_and_clear_all_open_stateid(struct nfs4_state *state)
 }
 
 static bool nfs_need_update_open_stateid(struct nfs4_state *state,
-		nfs4_stateid *stateid)
+		const nfs4_stateid *stateid, nfs4_stateid *freeme)
 {
 	if (test_and_set_bit(NFS_OPEN_STATE, &state->flags) == 0)
 		return true;
 	if (!nfs4_stateid_match_other(stateid, &state->open_stateid)) {
+		nfs4_stateid_copy(freeme, &state->open_stateid);
 		nfs_test_and_clear_all_open_stateid(state);
 		return true;
 	}
@@ -1437,7 +1492,9 @@ static void nfs_clear_open_stateid(struct nfs4_state *state,
 		nfs4_schedule_state_manager(state->owner->so_server->nfs_client);
 }
 
-static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
+static void nfs_set_open_stateid_locked(struct nfs4_state *state,
+		const nfs4_stateid *stateid, fmode_t fmode,
+		nfs4_stateid *freeme)
 {
 	switch (fmode) {
 		case FMODE_READ:
@@ -1449,14 +1506,18 @@ static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *
 		case FMODE_READ|FMODE_WRITE:
 			set_bit(NFS_O_RDWR_STATE, &state->flags);
 	}
-	if (!nfs_need_update_open_stateid(state, stateid))
+	if (!nfs_need_update_open_stateid(state, stateid, freeme))
 		return;
 	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
 		nfs4_stateid_copy(&state->stateid, stateid);
 	nfs4_stateid_copy(&state->open_stateid, stateid);
 }
 
-static void __update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, const nfs4_stateid *deleg_stateid, fmode_t fmode)
+static void __update_open_stateid(struct nfs4_state *state,
+		const nfs4_stateid *open_stateid,
+		const nfs4_stateid *deleg_stateid,
+		fmode_t fmode,
+		nfs4_stateid *freeme)
 {
 	/*
 	 * Protect the call to nfs4_state_set_mode_locked and
@@ -1469,16 +1530,22 @@ static void __update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_s
 		set_bit(NFS_DELEGATED_STATE, &state->flags);
 	}
 	if (open_stateid != NULL)
-		nfs_set_open_stateid_locked(state, open_stateid, fmode);
+		nfs_set_open_stateid_locked(state, open_stateid, fmode, freeme);
 	write_sequnlock(&state->seqlock);
 	update_open_stateflags(state, fmode);
 	spin_unlock(&state->owner->so_lock);
 }
 
-static int update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *delegation, fmode_t fmode)
+static int update_open_stateid(struct nfs4_state *state,
+		const nfs4_stateid *open_stateid,
+		const nfs4_stateid *delegation,
+		fmode_t fmode)
 {
+	struct nfs_server *server = NFS_SERVER(state->inode);
+	struct nfs_client *clp = server->nfs_client;
 	struct nfs_inode *nfsi = NFS_I(state->inode);
 	struct nfs_delegation *deleg_cur;
+	nfs4_stateid freeme = {0};
 	int ret = 0;
 
 	fmode &= (FMODE_READ|FMODE_WRITE);
@@ -1500,7 +1567,8 @@ static int update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stat
 		goto no_delegation_unlock;
 
 	nfs_mark_delegation_referenced(deleg_cur);
-	__update_open_stateid(state, open_stateid, &deleg_cur->stateid, fmode);
+	__update_open_stateid(state, open_stateid, &deleg_cur->stateid,
+			fmode, &freeme);
 	ret = 1;
 no_delegation_unlock:
 	spin_unlock(&deleg_cur->lock);
@@ -1508,11 +1576,14 @@ no_delegation:
 	rcu_read_unlock();
 
 	if (!ret && open_stateid != NULL) {
-		__update_open_stateid(state, open_stateid, NULL, fmode);
+		__update_open_stateid(state, open_stateid, NULL, fmode, &freeme);
 		ret = 1;
 	}
 	if (test_bit(NFS_STATE_RECLAIM_NOGRACE, &state->flags))
-		nfs4_schedule_state_manager(state->owner->so_server->nfs_client);
+		nfs4_schedule_state_manager(clp);
+	if (freeme.type != 0)
+		nfs4_test_and_free_stateid(server, &freeme,
+				state->owner->so_cred);
 
 	return ret;
 }
@@ -1889,7 +1960,6 @@ static int nfs4_handle_delegation_recall_error(struct nfs_server *server, struct
 		case -NFS4ERR_STALE_CLIENTID:
 		case -NFS4ERR_STALE_STATEID:
 			set_bit(NFS_DELEGATED_STATE, &state->flags);
-		case -NFS4ERR_EXPIRED:
 			/* Don't recall a delegation if it was lost */
 			nfs4_schedule_lease_recovery(server->nfs_client);
 			return -EAGAIN;
@@ -1901,6 +1971,7 @@ static int nfs4_handle_delegation_recall_error(struct nfs_server *server, struct
 			return -EAGAIN;
 		case -NFS4ERR_DELEG_REVOKED:
 		case -NFS4ERR_ADMIN_REVOKED:
+		case -NFS4ERR_EXPIRED:
 		case -NFS4ERR_BAD_STATEID:
 		case -NFS4ERR_OPENMODE:
 			nfs_inode_find_state_and_recover(state->inode,
@@ -2382,9 +2453,10 @@ static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *sta
 	return ret;
 }
 
-static void nfs_finish_clear_delegation_stateid(struct nfs4_state *state)
+static void nfs_finish_clear_delegation_stateid(struct nfs4_state *state,
+		const nfs4_stateid *stateid)
 {
-	nfs_remove_bad_delegation(state->inode);
+	nfs_remove_bad_delegation(state->inode, stateid);
 	write_seqlock(&state->seqlock);
 	nfs4_stateid_copy(&state->stateid, &state->open_stateid);
 	write_sequnlock(&state->seqlock);
@@ -2394,7 +2466,7 @@ static void nfs_finish_clear_delegation_stateid(struct nfs4_state *state)
 static void nfs40_clear_delegation_stateid(struct nfs4_state *state)
 {
 	if (rcu_access_pointer(NFS_I(state->inode)->delegation) != NULL)
-		nfs_finish_clear_delegation_stateid(state);
+		nfs_finish_clear_delegation_stateid(state, NULL);
 }
 
 static int nfs40_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
@@ -2404,7 +2476,45 @@ static int nfs40_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *st
 	return nfs4_open_expired(sp, state);
 }
 
+static int nfs40_test_and_free_expired_stateid(struct nfs_server *server,
+		nfs4_stateid *stateid,
+		struct rpc_cred *cred)
+{
+	return -NFS4ERR_BAD_STATEID;
+}
+
 #if defined(CONFIG_NFS_V4_1)
+static int nfs41_test_and_free_expired_stateid(struct nfs_server *server,
+		nfs4_stateid *stateid,
+		struct rpc_cred *cred)
+{
+	int status;
+
+	switch (stateid->type) {
+	default:
+		break;
+	case NFS4_INVALID_STATEID_TYPE:
+	case NFS4_SPECIAL_STATEID_TYPE:
+		return -NFS4ERR_BAD_STATEID;
+	case NFS4_REVOKED_STATEID_TYPE:
+		goto out_free;
+	}
+
+	status = nfs41_test_stateid(server, stateid, cred);
+	switch (status) {
+	case -NFS4ERR_EXPIRED:
+	case -NFS4ERR_ADMIN_REVOKED:
+	case -NFS4ERR_DELEG_REVOKED:
+		break;
+	default:
+		return status;
+	}
+out_free:
+	/* Ack the revoked state to the server */
+	nfs41_free_stateid(server, stateid, cred, true);
+	return -NFS4ERR_EXPIRED;
+}
+
 static void nfs41_check_delegation_stateid(struct nfs4_state *state)
 {
 	struct nfs_server *server = NFS_SERVER(state->inode);
@@ -2422,22 +2532,67 @@ static void nfs41_check_delegation_stateid(struct nfs4_state *state)
 	}
 
 	nfs4_stateid_copy(&stateid, &delegation->stateid);
+	if (test_bit(NFS_DELEGATION_REVOKED, &delegation->flags)) {
+		rcu_read_unlock();
+		nfs_finish_clear_delegation_stateid(state, &stateid);
+		return;
+	}
+
+	if (!test_and_clear_bit(NFS_DELEGATION_TEST_EXPIRED, &delegation->flags)) {
+		rcu_read_unlock();
+		return;
+	}
+
 	cred = get_rpccred(delegation->cred);
 	rcu_read_unlock();
-	status = nfs41_test_stateid(server, &stateid, cred);
+	status = nfs41_test_and_free_expired_stateid(server, &stateid, cred);
 	trace_nfs4_test_delegation_stateid(state, NULL, status);
-
-	if (status != NFS_OK) {
-		/* Free the stateid unless the server explicitly
-		 * informs us the stateid is unrecognized. */
-		if (status != -NFS4ERR_BAD_STATEID)
-			nfs41_free_stateid(server, &stateid, cred);
-		nfs_finish_clear_delegation_stateid(state);
-	}
+	if (status == -NFS4ERR_EXPIRED || status == -NFS4ERR_BAD_STATEID)
+		nfs_finish_clear_delegation_stateid(state, &stateid);
 
 	put_rpccred(cred);
 }
 
+/**
+ * nfs41_check_expired_locks - possibly free a lock stateid
+ *
+ * @state: NFSv4 state for an inode
+ *
+ * Returns NFS_OK if recovery for this stateid is now finished.
+ * Otherwise a negative NFS4ERR value is returned.
+ */
+static int nfs41_check_expired_locks(struct nfs4_state *state)
+{
+	int status, ret = NFS_OK;
+	struct nfs4_lock_state *lsp;
+	struct nfs_server *server = NFS_SERVER(state->inode);
+
+	if (!test_bit(LK_STATE_IN_USE, &state->flags))
+		goto out;
+	list_for_each_entry(lsp, &state->lock_states, ls_locks) {
+		if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags)) {
+			struct rpc_cred *cred = lsp->ls_state->owner->so_cred;
+
+			status = nfs41_test_and_free_expired_stateid(server,
+					&lsp->ls_stateid,
+					cred);
+			trace_nfs4_test_lock_stateid(state, lsp, status);
+			if (status == -NFS4ERR_EXPIRED ||
+			    status == -NFS4ERR_BAD_STATEID) {
+				clear_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags);
+				lsp->ls_stateid.type = NFS4_INVALID_STATEID_TYPE;
+				if (!recover_lost_locks)
+					set_bit(NFS_LOCK_LOST, &lsp->ls_flags);
+			} else if (status != NFS_OK) {
+				ret = status;
+				break;
+			}
+		}
+	};
+out:
+	return ret;
+}
+
 /**
  * nfs41_check_open_stateid - possibly free an open stateid
  *
@@ -2453,26 +2608,28 @@ static int nfs41_check_open_stateid(struct nfs4_state *state)
 	struct rpc_cred *cred = state->owner->so_cred;
 	int status;
 
-	/* If a state reset has been done, test_stateid is unneeded */
-	if ((test_bit(NFS_O_RDONLY_STATE, &state->flags) == 0) &&
-	    (test_bit(NFS_O_WRONLY_STATE, &state->flags) == 0) &&
-	    (test_bit(NFS_O_RDWR_STATE, &state->flags) == 0))
+	if (test_bit(NFS_OPEN_STATE, &state->flags) == 0) {
+		if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)  {
+			if (nfs4_have_delegation(state->inode, state->state))
+				return NFS_OK;
+			return -NFS4ERR_OPENMODE;
+		}
 		return -NFS4ERR_BAD_STATEID;
-
-	status = nfs41_test_stateid(server, stateid, cred);
+	}
+	status = nfs41_test_and_free_expired_stateid(server, stateid, cred);
 	trace_nfs4_test_open_stateid(state, NULL, status);
-	if (status != NFS_OK) {
-		/* Free the stateid unless the server explicitly
-		 * informs us the stateid is unrecognized. */
-		if (status != -NFS4ERR_BAD_STATEID)
-			nfs41_free_stateid(server, stateid, cred);
-
+	if (status == -NFS4ERR_EXPIRED || status == -NFS4ERR_BAD_STATEID) {
 		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
 		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
 		clear_bit(NFS_O_RDWR_STATE, &state->flags);
 		clear_bit(NFS_OPEN_STATE, &state->flags);
+		stateid->type = NFS4_INVALID_STATEID_TYPE;
 	}
-	return status;
+	if (status != NFS_OK)
+		return status;
+	if (nfs_open_stateid_recover_openmode(state))
+		return -NFS4ERR_OPENMODE;
+	return NFS_OK;
 }
 
 static int nfs41_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
@@ -2480,6 +2637,9 @@ static int nfs41_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *st
 	int status;
 
 	nfs41_check_delegation_stateid(state);
+	status = nfs41_check_expired_locks(state);
+	if (status != NFS_OK)
+		return status;
 	status = nfs41_check_open_stateid(state);
 	if (status != NFS_OK)
 		status = nfs4_open_expired(sp, state);
@@ -2537,6 +2697,8 @@ static int _nfs4_open_and_get_state(struct nfs4_opendata *opendata,
 		goto out;
 	if (server->caps & NFS_CAP_POSIX_LOCK)
 		set_bit(NFS_STATE_POSIX_LOCKS, &state->flags);
+	if (opendata->o_res.rflags & NFS4_OPEN_RESULT_MAY_NOTIFY_LOCK)
+		set_bit(NFS_STATE_MAY_NOTIFY_LOCK, &state->flags);
 
 	dentry = opendata->dentry;
 	if (d_really_is_negative(dentry)) {
@@ -2899,9 +3061,12 @@ static void nfs4_close_done(struct rpc_task *task, void *data)
 			break;
 		case -NFS4ERR_ADMIN_REVOKED:
 		case -NFS4ERR_STALE_STATEID:
+		case -NFS4ERR_EXPIRED:
+			nfs4_free_revoked_stateid(server,
+					&calldata->arg.stateid,
+					task->tk_msg.rpc_cred);
 		case -NFS4ERR_OLD_STATEID:
 		case -NFS4ERR_BAD_STATEID:
-		case -NFS4ERR_EXPIRED:
 			if (!nfs4_stateid_match(&calldata->arg.stateid,
 						&state->open_stateid)) {
 				rpc_restart_call_prepare(task);
@@ -4312,7 +4477,7 @@ static int nfs4_proc_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, s
 	if (error == 0) {
 		/* block layout checks this! */
 		server->pnfs_blksize = fsinfo->blksize;
-		set_pnfs_layoutdriver(server, fhandle, fsinfo->layouttype);
+		set_pnfs_layoutdriver(server, fhandle, fsinfo);
 	}
 
 	return error;
@@ -4399,24 +4564,25 @@ static bool nfs4_error_stateid_expired(int err)
 	return false;
 }
 
-void __nfs4_read_done_cb(struct nfs_pgio_header *hdr)
-{
-	nfs_invalidate_atime(hdr->inode);
-}
-
 static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_pgio_header *hdr)
 {
 	struct nfs_server *server = NFS_SERVER(hdr->inode);
 
 	trace_nfs4_read(hdr, task->tk_status);
-	if (nfs4_async_handle_error(task, server,
-				    hdr->args.context->state,
-				    NULL) == -EAGAIN) {
-		rpc_restart_call_prepare(task);
-		return -EAGAIN;
+	if (task->tk_status < 0) {
+		struct nfs4_exception exception = {
+			.inode = hdr->inode,
+			.state = hdr->args.context->state,
+			.stateid = &hdr->args.stateid,
+		};
+		task->tk_status = nfs4_async_handle_exception(task,
+				server, task->tk_status, &exception);
+		if (exception.retry) {
+			rpc_restart_call_prepare(task);
+			return -EAGAIN;
+		}
 	}
 
-	__nfs4_read_done_cb(hdr);
 	if (task->tk_status > 0)
 		renew_lease(server, hdr->timestamp);
 	return 0;
@@ -4445,6 +4611,8 @@ static int nfs4_read_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
 		return -EAGAIN;
 	if (nfs4_read_stateid_changed(task, &hdr->args))
 		return -EAGAIN;
+	if (task->tk_status > 0)
+		nfs_invalidate_atime(hdr->inode);
 	return hdr->pgio_done_cb ? hdr->pgio_done_cb(task, hdr) :
 				    nfs4_read_done_cb(task, hdr);
 }
@@ -4482,11 +4650,19 @@ static int nfs4_write_done_cb(struct rpc_task *task,
 	struct inode *inode = hdr->inode;
 
 	trace_nfs4_write(hdr, task->tk_status);
-	if (nfs4_async_handle_error(task, NFS_SERVER(inode),
-				    hdr->args.context->state,
-				    NULL) == -EAGAIN) {
-		rpc_restart_call_prepare(task);
-		return -EAGAIN;
+	if (task->tk_status < 0) {
+		struct nfs4_exception exception = {
+			.inode = hdr->inode,
+			.state = hdr->args.context->state,
+			.stateid = &hdr->args.stateid,
+		};
+		task->tk_status = nfs4_async_handle_exception(task,
+				NFS_SERVER(inode), task->tk_status,
+				&exception);
+		if (exception.retry) {
+			rpc_restart_call_prepare(task);
+			return -EAGAIN;
+		}
 	}
 	if (task->tk_status >= 0) {
 		renew_lease(NFS_SERVER(inode), hdr->timestamp);
@@ -5123,12 +5299,14 @@ static void nfs4_init_boot_verifier(const struct nfs_client *clp,
 	if (test_bit(NFS4CLNT_PURGE_STATE, &clp->cl_state)) {
 		/* An impossible timestamp guarantees this value
 		 * will never match a generated boot time. */
-		verf[0] = 0;
-		verf[1] = cpu_to_be32(NSEC_PER_SEC + 1);
+		verf[0] = cpu_to_be32(U32_MAX);
+		verf[1] = cpu_to_be32(U32_MAX);
 	} else {
 		struct nfs_net *nn = net_generic(clp->cl_net, nfs_net_id);
-		verf[0] = cpu_to_be32(nn->boot_time.tv_sec);
-		verf[1] = cpu_to_be32(nn->boot_time.tv_nsec);
+		u64 ns = ktime_to_ns(nn->boot_time);
+
+		verf[0] = cpu_to_be32(ns >> 32);
+		verf[1] = cpu_to_be32(ns);
 	}
 	memcpy(bootverf->data, verf, sizeof(bootverf->data));
 }
@@ -5393,10 +5571,13 @@ static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
 		renew_lease(data->res.server, data->timestamp);
 	case -NFS4ERR_ADMIN_REVOKED:
 	case -NFS4ERR_DELEG_REVOKED:
+	case -NFS4ERR_EXPIRED:
+		nfs4_free_revoked_stateid(data->res.server,
+				data->args.stateid,
+				task->tk_msg.rpc_cred);
 	case -NFS4ERR_BAD_STATEID:
 	case -NFS4ERR_OLD_STATEID:
 	case -NFS4ERR_STALE_STATEID:
-	case -NFS4ERR_EXPIRED:
 		task->tk_status = 0;
 		if (data->roc)
 			pnfs_roc_set_barrier(data->inode, data->roc_barrier);
@@ -5528,22 +5709,6 @@ int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4
 	return err;
 }
 
-#define NFS4_LOCK_MINTIMEOUT (1 * HZ)
-#define NFS4_LOCK_MAXTIMEOUT (30 * HZ)
-
-/* 
- * sleep, with exponential backoff, and retry the LOCK operation. 
- */
-static unsigned long
-nfs4_set_lock_task_retry(unsigned long timeout)
-{
-	freezable_schedule_timeout_killable_unsafe(timeout);
-	timeout <<= 1;
-	if (timeout > NFS4_LOCK_MAXTIMEOUT)
-		return NFS4_LOCK_MAXTIMEOUT;
-	return timeout;
-}
-
 static int _nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 {
 	struct inode *inode = state->inode;
@@ -5600,11 +5765,6 @@ static int nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *
 	return err;
 }
 
-static int do_vfs_lock(struct inode *inode, struct file_lock *fl)
-{
-	return locks_lock_inode_wait(inode, fl);
-}
-
 struct nfs4_unlockdata {
 	struct nfs_locku_args arg;
 	struct nfs_locku_res res;
@@ -5657,14 +5817,18 @@ static void nfs4_locku_done(struct rpc_task *task, void *data)
 	switch (task->tk_status) {
 		case 0:
 			renew_lease(calldata->server, calldata->timestamp);
-			do_vfs_lock(calldata->lsp->ls_state->inode, &calldata->fl);
+			locks_lock_inode_wait(calldata->lsp->ls_state->inode, &calldata->fl);
 			if (nfs4_update_lock_stateid(calldata->lsp,
 					&calldata->res.stateid))
 				break;
+		case -NFS4ERR_ADMIN_REVOKED:
+		case -NFS4ERR_EXPIRED:
+			nfs4_free_revoked_stateid(calldata->server,
+					&calldata->arg.stateid,
+					task->tk_msg.rpc_cred);
 		case -NFS4ERR_BAD_STATEID:
 		case -NFS4ERR_OLD_STATEID:
 		case -NFS4ERR_STALE_STATEID:
-		case -NFS4ERR_EXPIRED:
 			if (!nfs4_stateid_match(&calldata->arg.stateid,
 						&calldata->lsp->ls_stateid))
 				rpc_restart_call_prepare(task);
@@ -5765,7 +5929,7 @@ static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *
 	mutex_lock(&sp->so_delegreturn_mutex);
 	/* Exclude nfs4_reclaim_open_stateid() - note nesting! */
 	down_read(&nfsi->rwsem);
-	if (do_vfs_lock(inode, request) == -ENOENT) {
+	if (locks_lock_inode_wait(inode, request) == -ENOENT) {
 		up_read(&nfsi->rwsem);
 		mutex_unlock(&sp->so_delegreturn_mutex);
 		goto out;
@@ -5906,7 +6070,7 @@ static void nfs4_lock_done(struct rpc_task *task, void *calldata)
 				data->timestamp);
 		if (data->arg.new_lock) {
 			data->fl.fl_flags &= ~(FL_SLEEP | FL_ACCESS);
-			if (do_vfs_lock(lsp->ls_state->inode, &data->fl) < 0) {
+			if (locks_lock_inode_wait(lsp->ls_state->inode, &data->fl) < 0) {
 				rpc_restart_call_prepare(task);
 				break;
 			}
@@ -5965,6 +6129,7 @@ static void nfs4_handle_setlk_error(struct nfs_server *server, struct nfs4_lock_
 {
 	switch (error) {
 	case -NFS4ERR_ADMIN_REVOKED:
+	case -NFS4ERR_EXPIRED:
 	case -NFS4ERR_BAD_STATEID:
 		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
 		if (new_lock_owner != 0 ||
@@ -5973,7 +6138,6 @@ static void nfs4_handle_setlk_error(struct nfs_server *server, struct nfs4_lock_
 		break;
 	case -NFS4ERR_STALE_STATEID:
 		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
-	case -NFS4ERR_EXPIRED:
 		nfs4_schedule_lease_recovery(server->nfs_client);
 	};
 }
@@ -6083,52 +6247,19 @@ out:
 }
 
 #if defined(CONFIG_NFS_V4_1)
-/**
- * nfs41_check_expired_locks - possibly free a lock stateid
- *
- * @state: NFSv4 state for an inode
- *
- * Returns NFS_OK if recovery for this stateid is now finished.
- * Otherwise a negative NFS4ERR value is returned.
- */
-static int nfs41_check_expired_locks(struct nfs4_state *state)
-{
-	int status, ret = -NFS4ERR_BAD_STATEID;
-	struct nfs4_lock_state *lsp;
-	struct nfs_server *server = NFS_SERVER(state->inode);
-
-	list_for_each_entry(lsp, &state->lock_states, ls_locks) {
-		if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags)) {
-			struct rpc_cred *cred = lsp->ls_state->owner->so_cred;
-
-			status = nfs41_test_stateid(server,
-					&lsp->ls_stateid,
-					cred);
-			trace_nfs4_test_lock_stateid(state, lsp, status);
-			if (status != NFS_OK) {
-				/* Free the stateid unless the server
-				 * informs us the stateid is unrecognized. */
-				if (status != -NFS4ERR_BAD_STATEID)
-					nfs41_free_stateid(server,
-							&lsp->ls_stateid,
-							cred);
-				clear_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags);
-				ret = status;
-			}
-		}
-	};
-
-	return ret;
-}
-
 static int nfs41_lock_expired(struct nfs4_state *state, struct file_lock *request)
 {
-	int status = NFS_OK;
+	struct nfs4_lock_state *lsp;
+	int status;
 
-	if (test_bit(LK_STATE_IN_USE, &state->flags))
-		status = nfs41_check_expired_locks(state);
-	if (status != NFS_OK)
-		status = nfs4_lock_expired(state, request);
+	status = nfs4_set_lock_state(state, request);
+	if (status != 0)
+		return status;
+	lsp = request->fl_u.nfs4_fl.owner;
+	if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) ||
+	    test_bit(NFS_LOCK_LOST, &lsp->ls_flags))
+		return 0;
+	status = nfs4_lock_expired(state, request);
 	return status;
 }
 #endif
@@ -6138,17 +6269,10 @@ static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock
 	struct nfs_inode *nfsi = NFS_I(state->inode);
 	struct nfs4_state_owner *sp = state->owner;
 	unsigned char fl_flags = request->fl_flags;
-	int status = -ENOLCK;
+	int status;
 
-	if ((fl_flags & FL_POSIX) &&
-			!test_bit(NFS_STATE_POSIX_LOCKS, &state->flags))
-		goto out;
-	/* Is this a delegated open? */
-	status = nfs4_set_lock_state(state, request);
-	if (status != 0)
-		goto out;
 	request->fl_flags |= FL_ACCESS;
-	status = do_vfs_lock(state->inode, request);
+	status = locks_lock_inode_wait(state->inode, request);
 	if (status < 0)
 		goto out;
 	mutex_lock(&sp->so_delegreturn_mutex);
@@ -6157,7 +6281,7 @@ static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock
 		/* Yes: cache locks! */
 		/* ...but avoid races with delegation recall... */
 		request->fl_flags = fl_flags & ~FL_SLEEP;
-		status = do_vfs_lock(state->inode, request);
+		status = locks_lock_inode_wait(state->inode, request);
 		up_read(&nfsi->rwsem);
 		mutex_unlock(&sp->so_delegreturn_mutex);
 		goto out;
@@ -6188,12 +6312,124 @@ static int nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *
 	return err;
 }
 
+#define NFS4_LOCK_MINTIMEOUT (1 * HZ)
+#define NFS4_LOCK_MAXTIMEOUT (30 * HZ)
+
+static int
+nfs4_retry_setlk_simple(struct nfs4_state *state, int cmd,
+			struct file_lock *request)
+{
+	int		status = -ERESTARTSYS;
+	unsigned long	timeout = NFS4_LOCK_MINTIMEOUT;
+
+	while(!signalled()) {
+		status = nfs4_proc_setlk(state, cmd, request);
+		if ((status != -EAGAIN) || IS_SETLK(cmd))
+			break;
+		freezable_schedule_timeout_interruptible(timeout);
+		timeout *= 2;
+		timeout = min_t(unsigned long, NFS4_LOCK_MAXTIMEOUT, timeout);
+		status = -ERESTARTSYS;
+	}
+	return status;
+}
+
+#ifdef CONFIG_NFS_V4_1
+struct nfs4_lock_waiter {
+	struct task_struct	*task;
+	struct inode		*inode;
+	struct nfs_lowner	*owner;
+	bool			notified;
+};
+
+static int
+nfs4_wake_lock_waiter(wait_queue_t *wait, unsigned int mode, int flags, void *key)
+{
+	int ret;
+	struct cb_notify_lock_args *cbnl = key;
+	struct nfs4_lock_waiter	*waiter	= wait->private;
+	struct nfs_lowner	*lowner = &cbnl->cbnl_owner,
+				*wowner = waiter->owner;
+
+	/* Only wake if the callback was for the same owner */
+	if (lowner->clientid != wowner->clientid ||
+	    lowner->id != wowner->id		 ||
+	    lowner->s_dev != wowner->s_dev)
+		return 0;
+
+	/* Make sure it's for the right inode */
+	if (nfs_compare_fh(NFS_FH(waiter->inode), &cbnl->cbnl_fh))
+		return 0;
+
+	waiter->notified = true;
+
+	/* override "private" so we can use default_wake_function */
+	wait->private = waiter->task;
+	ret = autoremove_wake_function(wait, mode, flags, key);
+	wait->private = waiter;
+	return ret;
+}
+
+static int
+nfs4_retry_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
+{
+	int status = -ERESTARTSYS;
+	unsigned long flags;
+	struct nfs4_lock_state *lsp = request->fl_u.nfs4_fl.owner;
+	struct nfs_server *server = NFS_SERVER(state->inode);
+	struct nfs_client *clp = server->nfs_client;
+	wait_queue_head_t *q = &clp->cl_lock_waitq;
+	struct nfs_lowner owner = { .clientid = clp->cl_clientid,
+				    .id = lsp->ls_seqid.owner_id,
+				    .s_dev = server->s_dev };
+	struct nfs4_lock_waiter waiter = { .task  = current,
+					   .inode = state->inode,
+					   .owner = &owner,
+					   .notified = false };
+	wait_queue_t wait;
+
+	/* Don't bother with waitqueue if we don't expect a callback */
+	if (!test_bit(NFS_STATE_MAY_NOTIFY_LOCK, &state->flags))
+		return nfs4_retry_setlk_simple(state, cmd, request);
+
+	init_wait(&wait);
+	wait.private = &waiter;
+	wait.func = nfs4_wake_lock_waiter;
+	add_wait_queue(q, &wait);
+
+	while(!signalled()) {
+		status = nfs4_proc_setlk(state, cmd, request);
+		if ((status != -EAGAIN) || IS_SETLK(cmd))
+			break;
+
+		status = -ERESTARTSYS;
+		spin_lock_irqsave(&q->lock, flags);
+		if (waiter.notified) {
+			spin_unlock_irqrestore(&q->lock, flags);
+			continue;
+		}
+		set_current_state(TASK_INTERRUPTIBLE);
+		spin_unlock_irqrestore(&q->lock, flags);
+
+		freezable_schedule_timeout_interruptible(NFS4_LOCK_MAXTIMEOUT);
+	}
+
+	finish_wait(q, &wait);
+	return status;
+}
+#else /* !CONFIG_NFS_V4_1 */
+static inline int
+nfs4_retry_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
+{
+	return nfs4_retry_setlk_simple(state, cmd, request);
+}
+#endif
+
 static int
 nfs4_proc_lock(struct file *filp, int cmd, struct file_lock *request)
 {
 	struct nfs_open_context *ctx;
 	struct nfs4_state *state;
-	unsigned long timeout = NFS4_LOCK_MINTIMEOUT;
 	int status;
 
 	/* verify open state */
@@ -6220,6 +6456,11 @@ nfs4_proc_lock(struct file *filp, int cmd, struct file_lock *request)
 
 	if (state == NULL)
 		return -ENOLCK;
+
+	if ((request->fl_flags & FL_POSIX) &&
+	    !test_bit(NFS_STATE_POSIX_LOCKS, &state->flags))
+		return -ENOLCK;
+
 	/*
 	 * Don't rely on the VFS having checked the file open mode,
 	 * since it won't do this for flock() locks.
@@ -6234,16 +6475,11 @@ nfs4_proc_lock(struct file *filp, int cmd, struct file_lock *request)
 			return -EBADF;
 	}
 
-	do {
-		status = nfs4_proc_setlk(state, cmd, request);
-		if ((status != -EAGAIN) || IS_SETLK(cmd))
-			break;
-		timeout = nfs4_set_lock_task_retry(timeout);
-		status = -ERESTARTSYS;
-		if (signalled())
-			break;
-	} while(status < 0);
-	return status;
+	status = nfs4_set_lock_state(state, request);
+	if (status != 0)
+		return status;
+
+	return nfs4_retry_setlk(state, cmd, request);
 }
 
 int nfs4_lock_delegation_recall(struct file_lock *fl, struct nfs4_state *state, const nfs4_stateid *stateid)
@@ -7104,75 +7340,161 @@ static int nfs4_sp4_select_mode(struct nfs_client *clp,
 	return 0;
 }
 
+struct nfs41_exchange_id_data {
+	struct nfs41_exchange_id_res res;
+	struct nfs41_exchange_id_args args;
+	struct rpc_xprt *xprt;
+	int rpc_status;
+};
+
+static void nfs4_exchange_id_done(struct rpc_task *task, void *data)
+{
+	struct nfs41_exchange_id_data *cdata =
+					(struct nfs41_exchange_id_data *)data;
+	struct nfs_client *clp = cdata->args.client;
+	int status = task->tk_status;
+
+	trace_nfs4_exchange_id(clp, status);
+
+	if (status == 0)
+		status = nfs4_check_cl_exchange_flags(cdata->res.flags);
+
+	if (cdata->xprt && status == 0) {
+		status = nfs4_detect_session_trunking(clp, &cdata->res,
+						      cdata->xprt);
+		goto out;
+	}
+
+	if (status  == 0)
+		status = nfs4_sp4_select_mode(clp, &cdata->res.state_protect);
+
+	if (status == 0) {
+		clp->cl_clientid = cdata->res.clientid;
+		clp->cl_exchange_flags = cdata->res.flags;
+		/* Client ID is not confirmed */
+		if (!(cdata->res.flags & EXCHGID4_FLAG_CONFIRMED_R)) {
+			clear_bit(NFS4_SESSION_ESTABLISHED,
+			&clp->cl_session->session_state);
+			clp->cl_seqid = cdata->res.seqid;
+		}
+
+		kfree(clp->cl_serverowner);
+		clp->cl_serverowner = cdata->res.server_owner;
+		cdata->res.server_owner = NULL;
+
+		/* use the most recent implementation id */
+		kfree(clp->cl_implid);
+		clp->cl_implid = cdata->res.impl_id;
+		cdata->res.impl_id = NULL;
+
+		if (clp->cl_serverscope != NULL &&
+		    !nfs41_same_server_scope(clp->cl_serverscope,
+					cdata->res.server_scope)) {
+			dprintk("%s: server_scope mismatch detected\n",
+				__func__);
+			set_bit(NFS4CLNT_SERVER_SCOPE_MISMATCH, &clp->cl_state);
+			kfree(clp->cl_serverscope);
+			clp->cl_serverscope = NULL;
+		}
+
+		if (clp->cl_serverscope == NULL) {
+			clp->cl_serverscope = cdata->res.server_scope;
+			cdata->res.server_scope = NULL;
+		}
+		/* Save the EXCHANGE_ID verifier session trunk tests */
+		memcpy(clp->cl_confirm.data, cdata->args.verifier->data,
+		       sizeof(clp->cl_confirm.data));
+	}
+out:
+	cdata->rpc_status = status;
+	return;
+}
+
+static void nfs4_exchange_id_release(void *data)
+{
+	struct nfs41_exchange_id_data *cdata =
+					(struct nfs41_exchange_id_data *)data;
+
+	nfs_put_client(cdata->args.client);
+	if (cdata->xprt) {
+		xprt_put(cdata->xprt);
+		rpc_clnt_xprt_switch_put(cdata->args.client->cl_rpcclient);
+	}
+	kfree(cdata->res.impl_id);
+	kfree(cdata->res.server_scope);
+	kfree(cdata->res.server_owner);
+	kfree(cdata);
+}
+
+static const struct rpc_call_ops nfs4_exchange_id_call_ops = {
+	.rpc_call_done = nfs4_exchange_id_done,
+	.rpc_release = nfs4_exchange_id_release,
+};
+
 /*
  * _nfs4_proc_exchange_id()
  *
  * Wrapper for EXCHANGE_ID operation.
  */
 static int _nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred,
-	u32 sp4_how)
+			u32 sp4_how, struct rpc_xprt *xprt)
 {
 	nfs4_verifier verifier;
-	struct nfs41_exchange_id_args args = {
-		.verifier = &verifier,
-		.client = clp,
-#ifdef CONFIG_NFS_V4_1_MIGRATION
-		.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER |
-			 EXCHGID4_FLAG_BIND_PRINC_STATEID |
-			 EXCHGID4_FLAG_SUPP_MOVED_MIGR,
-#else
-		.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER |
-			 EXCHGID4_FLAG_BIND_PRINC_STATEID,
-#endif
-	};
-	struct nfs41_exchange_id_res res = {
-		0
-	};
-	int status;
 	struct rpc_message msg = {
 		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_EXCHANGE_ID],
-		.rpc_argp = &args,
-		.rpc_resp = &res,
 		.rpc_cred = cred,
 	};
+	struct rpc_task_setup task_setup_data = {
+		.rpc_client = clp->cl_rpcclient,
+		.callback_ops = &nfs4_exchange_id_call_ops,
+		.rpc_message = &msg,
+		.flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT,
+	};
+	struct nfs41_exchange_id_data *calldata;
+	struct rpc_task *task;
+	int status = -EIO;
+
+	if (!atomic_inc_not_zero(&clp->cl_count))
+		goto out;
+
+	status = -ENOMEM;
+	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
+	if (!calldata)
+		goto out;
 
-	nfs4_init_boot_verifier(clp, &verifier);
+	if (!xprt)
+		nfs4_init_boot_verifier(clp, &verifier);
 
 	status = nfs4_init_uniform_client_string(clp);
 	if (status)
-		goto out;
+		goto out_calldata;
 
 	dprintk("NFS call  exchange_id auth=%s, '%s'\n",
 		clp->cl_rpcclient->cl_auth->au_ops->au_name,
 		clp->cl_owner_id);
 
-	res.server_owner = kzalloc(sizeof(struct nfs41_server_owner),
-					GFP_NOFS);
-	if (unlikely(res.server_owner == NULL)) {
-		status = -ENOMEM;
-		goto out;
-	}
+	calldata->res.server_owner = kzalloc(sizeof(struct nfs41_server_owner),
+						GFP_NOFS);
+	status = -ENOMEM;
+	if (unlikely(calldata->res.server_owner == NULL))
+		goto out_calldata;
 
-	res.server_scope = kzalloc(sizeof(struct nfs41_server_scope),
+	calldata->res.server_scope = kzalloc(sizeof(struct nfs41_server_scope),
 					GFP_NOFS);
-	if (unlikely(res.server_scope == NULL)) {
-		status = -ENOMEM;
+	if (unlikely(calldata->res.server_scope == NULL))
 		goto out_server_owner;
-	}
 
-	res.impl_id = kzalloc(sizeof(struct nfs41_impl_id), GFP_NOFS);
-	if (unlikely(res.impl_id == NULL)) {
-		status = -ENOMEM;
+	calldata->res.impl_id = kzalloc(sizeof(struct nfs41_impl_id), GFP_NOFS);
+	if (unlikely(calldata->res.impl_id == NULL))
 		goto out_server_scope;
-	}
 
 	switch (sp4_how) {
 	case SP4_NONE:
-		args.state_protect.how = SP4_NONE;
+		calldata->args.state_protect.how = SP4_NONE;
 		break;
 
 	case SP4_MACH_CRED:
-		args.state_protect = nfs4_sp4_mach_cred_request;
+		calldata->args.state_protect = nfs4_sp4_mach_cred_request;
 		break;
 
 	default:
@@ -7181,56 +7503,42 @@ static int _nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred,
 		status = -EINVAL;
 		goto out_impl_id;
 	}
+	if (xprt) {
+		calldata->xprt = xprt;
+		task_setup_data.rpc_xprt = xprt;
+		task_setup_data.flags =
+				RPC_TASK_SOFT|RPC_TASK_SOFTCONN|RPC_TASK_ASYNC;
+		calldata->args.verifier = &clp->cl_confirm;
+	} else {
+		calldata->args.verifier = &verifier;
+	}
+	calldata->args.client = clp;
+#ifdef CONFIG_NFS_V4_1_MIGRATION
+	calldata->args.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER |
+	EXCHGID4_FLAG_BIND_PRINC_STATEID |
+	EXCHGID4_FLAG_SUPP_MOVED_MIGR,
+#else
+	calldata->args.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER |
+	EXCHGID4_FLAG_BIND_PRINC_STATEID,
+#endif
+	msg.rpc_argp = &calldata->args;
+	msg.rpc_resp = &calldata->res;
+	task_setup_data.callback_data = calldata;
 
-	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
-	trace_nfs4_exchange_id(clp, status);
-	if (status == 0)
-		status = nfs4_check_cl_exchange_flags(res.flags);
-
-	if (status == 0)
-		status = nfs4_sp4_select_mode(clp, &res.state_protect);
-
-	if (status == 0) {
-		clp->cl_clientid = res.clientid;
-		clp->cl_exchange_flags = res.flags;
-		/* Client ID is not confirmed */
-		if (!(res.flags & EXCHGID4_FLAG_CONFIRMED_R)) {
-			clear_bit(NFS4_SESSION_ESTABLISHED,
-					&clp->cl_session->session_state);
-			clp->cl_seqid = res.seqid;
-		}
-
-		kfree(clp->cl_serverowner);
-		clp->cl_serverowner = res.server_owner;
-		res.server_owner = NULL;
-
-		/* use the most recent implementation id */
-		kfree(clp->cl_implid);
-		clp->cl_implid = res.impl_id;
-		res.impl_id = NULL;
-
-		if (clp->cl_serverscope != NULL &&
-		    !nfs41_same_server_scope(clp->cl_serverscope,
-					     res.server_scope)) {
-			dprintk("%s: server_scope mismatch detected\n",
-				__func__);
-			set_bit(NFS4CLNT_SERVER_SCOPE_MISMATCH, &clp->cl_state);
-			kfree(clp->cl_serverscope);
-			clp->cl_serverscope = NULL;
-		}
-
-		if (clp->cl_serverscope == NULL) {
-			clp->cl_serverscope = res.server_scope;
-			res.server_scope = NULL;
-		}
+	task = rpc_run_task(&task_setup_data);
+	if (IS_ERR(task)) {
+	status = PTR_ERR(task);
+		goto out_impl_id;
 	}
 
-out_impl_id:
-	kfree(res.impl_id);
-out_server_scope:
-	kfree(res.server_scope);
-out_server_owner:
-	kfree(res.server_owner);
+	if (!xprt) {
+		status = rpc_wait_for_completion_task(task);
+		if (!status)
+			status = calldata->rpc_status;
+	} else	/* session trunking test */
+		status = calldata->rpc_status;
+
+	rpc_put_task(task);
 out:
 	if (clp->cl_implid != NULL)
 		dprintk("NFS reply exchange_id: Server Implementation ID: "
@@ -7240,6 +7548,16 @@ out:
 			clp->cl_implid->date.nseconds);
 	dprintk("NFS reply exchange_id: %d\n", status);
 	return status;
+
+out_impl_id:
+	kfree(calldata->res.impl_id);
+out_server_scope:
+	kfree(calldata->res.server_scope);
+out_server_owner:
+	kfree(calldata->res.server_owner);
+out_calldata:
+	kfree(calldata);
+	goto out;
 }
 
 /*
@@ -7262,14 +7580,45 @@ int nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred)
 	/* try SP4_MACH_CRED if krb5i/p	*/
 	if (authflavor == RPC_AUTH_GSS_KRB5I ||
 	    authflavor == RPC_AUTH_GSS_KRB5P) {
-		status = _nfs4_proc_exchange_id(clp, cred, SP4_MACH_CRED);
+		status = _nfs4_proc_exchange_id(clp, cred, SP4_MACH_CRED, NULL);
 		if (!status)
 			return 0;
 	}
 
 	/* try SP4_NONE */
-	return _nfs4_proc_exchange_id(clp, cred, SP4_NONE);
+	return _nfs4_proc_exchange_id(clp, cred, SP4_NONE, NULL);
+}
+
+/**
+ * nfs4_test_session_trunk
+ *
+ * This is an add_xprt_test() test function called from
+ * rpc_clnt_setup_test_and_add_xprt.
+ *
+ * The rpc_xprt_switch is referrenced by rpc_clnt_setup_test_and_add_xprt
+ * and is dereferrenced in nfs4_exchange_id_release
+ *
+ * Upon success, add the new transport to the rpc_clnt
+ *
+ * @clnt: struct rpc_clnt to get new transport
+ * @xprt: the rpc_xprt to test
+ * @data: call data for _nfs4_proc_exchange_id.
+ */
+int nfs4_test_session_trunk(struct rpc_clnt *clnt, struct rpc_xprt *xprt,
+			    void *data)
+{
+	struct nfs4_add_xprt_data *adata = (struct nfs4_add_xprt_data *)data;
+	u32 sp4_how;
+
+	dprintk("--> %s try %s\n", __func__,
+		xprt->address_strings[RPC_DISPLAY_ADDR]);
+
+	sp4_how = (adata->clp->cl_sp4_flags == 0 ? SP4_NONE : SP4_MACH_CRED);
+
+	/* Test connection for session trunking. Async exchange_id call */
+	return  _nfs4_proc_exchange_id(adata->clp, adata->cred, sp4_how, xprt);
 }
+EXPORT_SYMBOL_GPL(nfs4_test_session_trunk);
 
 static int _nfs4_proc_destroy_clientid(struct nfs_client *clp,
 		struct rpc_cred *cred)
@@ -7463,7 +7812,7 @@ static void nfs4_init_channel_attrs(struct nfs41_create_session_args *args,
 	args->bc_attrs.max_resp_sz = max_bc_payload;
 	args->bc_attrs.max_resp_sz_cached = 0;
 	args->bc_attrs.max_ops = NFS4_MAX_BACK_CHANNEL_OPS;
-	args->bc_attrs.max_reqs = NFS41_BC_MAX_CALLBACKS;
+	args->bc_attrs.max_reqs = min_t(unsigned short, max_session_cb_slots, 1);
 
 	dprintk("%s: Back Channel : max_rqst_sz=%u max_resp_sz=%u "
 		"max_resp_sz_cached=%u max_ops=%u max_reqs=%u\n",
@@ -7510,10 +7859,9 @@ static int nfs4_verify_back_channel_attrs(struct nfs41_create_session_args *args
 		return -EINVAL;
 	if (rcvd->max_resp_sz_cached > sent->max_resp_sz_cached)
 		return -EINVAL;
-	/* These would render the backchannel useless: */
-	if (rcvd->max_ops != sent->max_ops)
+	if (rcvd->max_ops > sent->max_ops)
 		return -EINVAL;
-	if (rcvd->max_reqs != sent->max_reqs)
+	if (rcvd->max_reqs > sent->max_reqs)
 		return -EINVAL;
 out:
 	return 0;
@@ -7982,6 +8330,8 @@ nfs4_layoutget_handle_exception(struct rpc_task *task,
 	case -NFS4ERR_RECALLCONFLICT:
 		status = -ERECALLCONFLICT;
 		break;
+	case -NFS4ERR_DELEG_REVOKED:
+	case -NFS4ERR_ADMIN_REVOKED:
 	case -NFS4ERR_EXPIRED:
 	case -NFS4ERR_BAD_STATEID:
 		exception->timeout = 0;
@@ -7993,6 +8343,7 @@ nfs4_layoutget_handle_exception(struct rpc_task *task,
 					&lgp->args.ctx->state->stateid)) {
 			spin_unlock(&inode->i_lock);
 			exception->state = lgp->args.ctx->state;
+			exception->stateid = &lgp->args.stateid;
 			break;
 		}
 
@@ -8591,6 +8942,24 @@ static int _nfs41_test_stateid(struct nfs_server *server,
 	return -res.status;
 }
 
+static void nfs4_handle_delay_or_session_error(struct nfs_server *server,
+		int err, struct nfs4_exception *exception)
+{
+	exception->retry = 0;
+	switch(err) {
+	case -NFS4ERR_DELAY:
+	case -NFS4ERR_RETRY_UNCACHED_REP:
+		nfs4_handle_exception(server, err, exception);
+		break;
+	case -NFS4ERR_BADSESSION:
+	case -NFS4ERR_BADSLOT:
+	case -NFS4ERR_BAD_HIGH_SLOT:
+	case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
+	case -NFS4ERR_DEADSESSION:
+		nfs4_do_handle_exception(server, err, exception);
+	}
+}
+
 /**
  * nfs41_test_stateid - perform a TEST_STATEID operation
  *
@@ -8610,9 +8979,7 @@ static int nfs41_test_stateid(struct nfs_server *server,
 	int err;
 	do {
 		err = _nfs41_test_stateid(server, stateid, cred);
-		if (err != -NFS4ERR_DELAY)
-			break;
-		nfs4_handle_exception(server, err, &exception);
+		nfs4_handle_delay_or_session_error(server, err, &exception);
 	} while (exception.retry);
 	return err;
 }
@@ -8657,7 +9024,7 @@ static const struct rpc_call_ops nfs41_free_stateid_ops = {
 };
 
 static struct rpc_task *_nfs41_free_stateid(struct nfs_server *server,
-		nfs4_stateid *stateid,
+		const nfs4_stateid *stateid,
 		struct rpc_cred *cred,
 		bool privileged)
 {
@@ -8687,7 +9054,7 @@ static struct rpc_task *_nfs41_free_stateid(struct nfs_server *server,
 
 	msg.rpc_argp = &data->args;
 	msg.rpc_resp = &data->res;
-	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
+	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
 	if (privileged)
 		nfs4_set_sequence_privileged(&data->args.seq_args);
 
@@ -8700,38 +9067,31 @@ static struct rpc_task *_nfs41_free_stateid(struct nfs_server *server,
  * @server: server / transport on which to perform the operation
  * @stateid: state ID to release
  * @cred: credential
+ * @is_recovery: set to true if this call needs to be privileged
  *
- * Returns NFS_OK if the server freed "stateid".  Otherwise a
- * negative NFS4ERR value is returned.
+ * Note: this function is always asynchronous.
  */
 static int nfs41_free_stateid(struct nfs_server *server,
-		nfs4_stateid *stateid,
-		struct rpc_cred *cred)
+		const nfs4_stateid *stateid,
+		struct rpc_cred *cred,
+		bool is_recovery)
 {
 	struct rpc_task *task;
-	int ret;
 
-	task = _nfs41_free_stateid(server, stateid, cred, true);
+	task = _nfs41_free_stateid(server, stateid, cred, is_recovery);
 	if (IS_ERR(task))
 		return PTR_ERR(task);
-	ret = rpc_wait_for_completion_task(task);
-	if (!ret)
-		ret = task->tk_status;
 	rpc_put_task(task);
-	return ret;
+	return 0;
 }
 
 static void
 nfs41_free_lock_state(struct nfs_server *server, struct nfs4_lock_state *lsp)
 {
-	struct rpc_task *task;
 	struct rpc_cred *cred = lsp->ls_state->owner->so_cred;
 
-	task = _nfs41_free_stateid(server, &lsp->ls_stateid, cred, false);
+	nfs41_free_stateid(server, &lsp->ls_stateid, cred, false);
 	nfs4_free_lock_state(server, lsp);
-	if (IS_ERR(task))
-		return;
-	rpc_put_task(task);
 }
 
 static bool nfs41_match_stateid(const nfs4_stateid *s1,
@@ -8835,6 +9195,7 @@ static const struct nfs4_minor_version_ops nfs_v4_0_minor_ops = {
 	.match_stateid = nfs4_match_stateid,
 	.find_root_sec = nfs4_find_root_sec,
 	.free_lock_state = nfs4_release_lockowner,
+	.test_and_free_expired = nfs40_test_and_free_expired_stateid,
 	.alloc_seqid = nfs_alloc_seqid,
 	.call_sync_ops = &nfs40_call_sync_ops,
 	.reboot_recovery_ops = &nfs40_reboot_recovery_ops,
@@ -8862,7 +9223,9 @@ static const struct nfs4_minor_version_ops nfs_v4_1_minor_ops = {
 	.match_stateid = nfs41_match_stateid,
 	.find_root_sec = nfs41_find_root_sec,
 	.free_lock_state = nfs41_free_lock_state,
+	.test_and_free_expired = nfs41_test_and_free_expired_stateid,
 	.alloc_seqid = nfs_alloc_no_seqid,
+	.session_trunk = nfs4_test_session_trunk,
 	.call_sync_ops = &nfs41_call_sync_ops,
 	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
 	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
@@ -8891,7 +9254,9 @@ static const struct nfs4_minor_version_ops nfs_v4_2_minor_ops = {
 	.find_root_sec = nfs41_find_root_sec,
 	.free_lock_state = nfs41_free_lock_state,
 	.call_sync_ops = &nfs41_call_sync_ops,
+	.test_and_free_expired = nfs41_test_and_free_expired_stateid,
 	.alloc_seqid = nfs_alloc_no_seqid,
+	.session_trunk = nfs4_test_session_trunk,
 	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
 	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
 	.state_renewal_ops = &nfs41_state_renewal_ops,

fs/nfs/nfs4session.h

@@ -9,6 +9,7 @@
 
 /* maximum number of slots to use */
 #define NFS4_DEF_SLOT_TABLE_SIZE (64U)
+#define NFS4_DEF_CB_SLOT_TABLE_SIZE (1U)
 #define NFS4_MAX_SLOT_TABLE (1024U)
 #define NFS4_NO_SLOT ((u32)-1)
 
@@ -22,6 +23,7 @@ struct nfs4_slot {
 	u32			slot_nr;
 	u32		 	seq_nr;
 	unsigned int		interrupted : 1,
+				privileged : 1,
 				seq_done : 1;
 };
 

fs/nfs/nfs4state.c

@@ -991,6 +991,8 @@ int nfs4_select_rw_stateid(struct nfs4_state *state,
 {
 	int ret;
 
+	if (!nfs4_valid_open_stateid(state))
+		return -EIO;
 	if (cred != NULL)
 		*cred = NULL;
 	ret = nfs4_copy_lock_stateid(dst, state, lockowner);
@@ -1303,6 +1305,8 @@ void nfs4_schedule_path_down_recovery(struct nfs_client *clp)
 static int nfs4_state_mark_reclaim_reboot(struct nfs_client *clp, struct nfs4_state *state)
 {
 
+	if (!nfs4_valid_open_stateid(state))
+		return 0;
 	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)) {
@@ -1316,6 +1320,8 @@ static int nfs4_state_mark_reclaim_reboot(struct nfs_client *clp, struct nfs4_st
 
 int nfs4_state_mark_reclaim_nograce(struct nfs_client *clp, struct nfs4_state *state)
 {
+	if (!nfs4_valid_open_stateid(state))
+		return 0;
 	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);
@@ -1327,9 +1333,8 @@ int nfs4_schedule_stateid_recovery(const struct nfs_server *server, struct nfs4_
 {
 	struct nfs_client *clp = server->nfs_client;
 
-	if (!nfs4_valid_open_stateid(state))
+	if (!nfs4_state_mark_reclaim_nograce(clp, state))
 		return -EBADF;
-	nfs4_state_mark_reclaim_nograce(clp, state);
 	dprintk("%s: scheduling stateid recovery for server %s\n", __func__,
 			clp->cl_hostname);
 	nfs4_schedule_state_manager(clp);
@@ -1337,6 +1342,35 @@ int nfs4_schedule_stateid_recovery(const struct nfs_server *server, struct nfs4_
 }
 EXPORT_SYMBOL_GPL(nfs4_schedule_stateid_recovery);
 
+static struct nfs4_lock_state *
+nfs_state_find_lock_state_by_stateid(struct nfs4_state *state,
+		const nfs4_stateid *stateid)
+{
+	struct nfs4_lock_state *pos;
+
+	list_for_each_entry(pos, &state->lock_states, ls_locks) {
+		if (!test_bit(NFS_LOCK_INITIALIZED, &pos->ls_flags))
+			continue;
+		if (nfs4_stateid_match_other(&pos->ls_stateid, stateid))
+			return pos;
+	}
+	return NULL;
+}
+
+static bool nfs_state_lock_state_matches_stateid(struct nfs4_state *state,
+		const nfs4_stateid *stateid)
+{
+	bool found = false;
+
+	if (test_bit(LK_STATE_IN_USE, &state->flags)) {
+		spin_lock(&state->state_lock);
+		if (nfs_state_find_lock_state_by_stateid(state, stateid))
+			found = true;
+		spin_unlock(&state->state_lock);
+	}
+	return found;
+}
+
 void nfs_inode_find_state_and_recover(struct inode *inode,
 		const nfs4_stateid *stateid)
 {
@@ -1351,14 +1385,18 @@ void nfs_inode_find_state_and_recover(struct inode *inode,
 		state = ctx->state;
 		if (state == NULL)
 			continue;
-		if (!test_bit(NFS_DELEGATED_STATE, &state->flags))
+		if (nfs4_stateid_match_other(&state->stateid, stateid) &&
+		    nfs4_state_mark_reclaim_nograce(clp, state)) {
+			found = true;
 			continue;
-		if (!nfs4_stateid_match(&state->stateid, stateid))
-			continue;
-		nfs4_state_mark_reclaim_nograce(clp, state);
-		found = true;
+		}
+		if (nfs_state_lock_state_matches_stateid(state, stateid) &&
+		    nfs4_state_mark_reclaim_nograce(clp, state))
+			found = true;
 	}
 	spin_unlock(&inode->i_lock);
+
+	nfs_inode_find_delegation_state_and_recover(inode, stateid);
 	if (found)
 		nfs4_schedule_state_manager(clp);
 }
@@ -1498,6 +1536,9 @@ restart:
 					__func__, status);
 			case -ENOENT:
 			case -ENOMEM:
+			case -EACCES:
+			case -EROFS:
+			case -EIO:
 			case -ESTALE:
 				/* Open state on this file cannot be recovered */
 				nfs4_state_mark_recovery_failed(state, status);
@@ -1656,15 +1697,9 @@ static void nfs4_state_end_reclaim_reboot(struct nfs_client *clp)
 	put_rpccred(cred);
 }
 
-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);
+	nfs_mark_test_expired_all_delegations(clp);
 	nfs4_state_mark_reclaim_helper(clp, nfs4_state_mark_reclaim_nograce);
 }
 
@@ -2195,7 +2230,7 @@ static void nfs41_handle_all_state_revoked(struct nfs_client *clp)
 
 static void nfs41_handle_some_state_revoked(struct nfs_client *clp)
 {
-	nfs4_state_mark_reclaim_helper(clp, nfs4_state_mark_reclaim_nograce);
+	nfs4_state_start_reclaim_nograce(clp);
 	nfs4_schedule_state_manager(clp);
 
 	dprintk("%s: state revoked on server %s\n", __func__, clp->cl_hostname);
@@ -2227,13 +2262,22 @@ static void nfs41_handle_cb_path_down(struct nfs_client *clp)
 		nfs4_schedule_state_manager(clp);
 }
 
-void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags)
+void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags,
+		bool recovery)
 {
 	if (!flags)
 		return;
 
 	dprintk("%s: \"%s\" (client ID %llx) flags=0x%08x\n",
 		__func__, clp->cl_hostname, clp->cl_clientid, flags);
+	/*
+	 * If we're called from the state manager thread, then assume we're
+	 * already handling the RECLAIM_NEEDED and/or STATE_REVOKED.
+	 * Those flags are expected to remain set until we're done
+	 * recovering (see RFC5661, section 18.46.3).
+	 */
+	if (recovery)
+		goto out_recovery;
 
 	if (flags & SEQ4_STATUS_RESTART_RECLAIM_NEEDED)
 		nfs41_handle_server_reboot(clp);
@@ -2246,6 +2290,7 @@ void nfs41_handle_sequence_flag_errors(struct nfs_client *clp, u32 flags)
 		nfs4_schedule_lease_moved_recovery(clp);
 	if (flags & SEQ4_STATUS_RECALLABLE_STATE_REVOKED)
 		nfs41_handle_recallable_state_revoked(clp);
+out_recovery:
 	if (flags & SEQ4_STATUS_BACKCHANNEL_FAULT)
 		nfs41_handle_backchannel_fault(clp);
 	else if (flags & (SEQ4_STATUS_CB_PATH_DOWN |
@@ -2410,6 +2455,13 @@ static void nfs4_state_manager(struct nfs_client *clp)
 			nfs4_state_end_reclaim_reboot(clp);
 		}
 
+		/* Detect expired delegations... */
+		if (test_and_clear_bit(NFS4CLNT_DELEGATION_EXPIRED, &clp->cl_state)) {
+			section = "detect expired delegations";
+			nfs_reap_expired_delegations(clp);
+			continue;
+		}
+
 		/* Now recover expired state... */
 		if (test_and_clear_bit(NFS4CLNT_RECLAIM_NOGRACE, &clp->cl_state)) {
 			section = "reclaim nograce";

fs/nfs/nfs4xdr.c

@@ -1850,7 +1850,7 @@ static void encode_create_session(struct xdr_stream *xdr,
 	*p++ = cpu_to_be32(RPC_AUTH_UNIX);			/* auth_sys */
 
 	/* authsys_parms rfc1831 */
-	*p++ = cpu_to_be32(nn->boot_time.tv_nsec);	/* stamp */
+	*p++ = cpu_to_be32(ktime_to_ns(nn->boot_time));	/* stamp */
 	p = xdr_encode_array(p, clnt->cl_nodename, clnt->cl_nodelen);
 	*p++ = cpu_to_be32(0);				/* UID */
 	*p++ = cpu_to_be32(0);				/* GID */
@@ -4725,34 +4725,37 @@ static int decode_getfattr(struct xdr_stream *xdr, struct nfs_fattr *fattr,
 }
 
 /*
- * Decode potentially multiple layout types. Currently we only support
- * one layout driver per file system.
+ * Decode potentially multiple layout types.
  */
-static int decode_first_pnfs_layout_type(struct xdr_stream *xdr,
-					 uint32_t *layouttype)
+static int decode_pnfs_layout_types(struct xdr_stream *xdr,
+				    struct nfs_fsinfo *fsinfo)
 {
 	__be32 *p;
-	int num;
+	uint32_t i;
 
 	p = xdr_inline_decode(xdr, 4);
 	if (unlikely(!p))
 		goto out_overflow;
-	num = be32_to_cpup(p);
+	fsinfo->nlayouttypes = be32_to_cpup(p);
 
 	/* pNFS is not supported by the underlying file system */
-	if (num == 0) {
-		*layouttype = 0;
+	if (fsinfo->nlayouttypes == 0)
 		return 0;
-	}
-	if (num > 1)
-		printk(KERN_INFO "NFS: %s: Warning: Multiple pNFS layout "
-			"drivers per filesystem not supported\n", __func__);
 
 	/* Decode and set first layout type, move xdr->p past unused types */
-	p = xdr_inline_decode(xdr, num * 4);
+	p = xdr_inline_decode(xdr, fsinfo->nlayouttypes * 4);
 	if (unlikely(!p))
 		goto out_overflow;
-	*layouttype = be32_to_cpup(p);
+
+	/* If we get too many, then just cap it at the max */
+	if (fsinfo->nlayouttypes > NFS_MAX_LAYOUT_TYPES) {
+		printk(KERN_INFO "NFS: %s: Warning: Too many (%u) pNFS layout types\n",
+			__func__, fsinfo->nlayouttypes);
+		fsinfo->nlayouttypes = NFS_MAX_LAYOUT_TYPES;
+	}
+
+	for(i = 0; i < fsinfo->nlayouttypes; ++i)
+		fsinfo->layouttype[i] = be32_to_cpup(p++);
 	return 0;
 out_overflow:
 	print_overflow_msg(__func__, xdr);
@@ -4764,7 +4767,7 @@ out_overflow:
  * Note we must ensure that layouttype is set in any non-error case.
  */
 static int decode_attr_pnfstype(struct xdr_stream *xdr, uint32_t *bitmap,
-				uint32_t *layouttype)
+				struct nfs_fsinfo *fsinfo)
 {
 	int status = 0;
 
@@ -4772,10 +4775,9 @@ static int decode_attr_pnfstype(struct xdr_stream *xdr, uint32_t *bitmap,
 	if (unlikely(bitmap[1] & (FATTR4_WORD1_FS_LAYOUT_TYPES - 1U)))
 		return -EIO;
 	if (bitmap[1] & FATTR4_WORD1_FS_LAYOUT_TYPES) {
-		status = decode_first_pnfs_layout_type(xdr, layouttype);
+		status = decode_pnfs_layout_types(xdr, fsinfo);
 		bitmap[1] &= ~FATTR4_WORD1_FS_LAYOUT_TYPES;
-	} else
-		*layouttype = 0;
+	}
 	return status;
 }
 
@@ -4856,7 +4858,7 @@ static int decode_fsinfo(struct xdr_stream *xdr, struct nfs_fsinfo *fsinfo)
 	status = decode_attr_time_delta(xdr, bitmap, &fsinfo->time_delta);
 	if (status != 0)
 		goto xdr_error;
-	status = decode_attr_pnfstype(xdr, bitmap, &fsinfo->layouttype);
+	status = decode_attr_pnfstype(xdr, bitmap, fsinfo);
 	if (status != 0)
 		goto xdr_error;
 

fs/nfs/pnfs.c

@@ -30,6 +30,7 @@
 #include <linux/nfs_fs.h>
 #include <linux/nfs_page.h>
 #include <linux/module.h>
+#include <linux/sort.h>
 #include "internal.h"
 #include "pnfs.h"
 #include "iostat.h"
@@ -98,36 +99,80 @@ unset_pnfs_layoutdriver(struct nfs_server *nfss)
 	nfss->pnfs_curr_ld = NULL;
 }
 
+/*
+ * When the server sends a list of layout types, we choose one in the order
+ * given in the list below.
+ *
+ * FIXME: should this list be configurable in some fashion? module param?
+ * 	  mount option? something else?
+ */
+static const u32 ld_prefs[] = {
+	LAYOUT_SCSI,
+	LAYOUT_BLOCK_VOLUME,
+	LAYOUT_OSD2_OBJECTS,
+	LAYOUT_FLEX_FILES,
+	LAYOUT_NFSV4_1_FILES,
+	0
+};
+
+static int
+ld_cmp(const void *e1, const void *e2)
+{
+	u32 ld1 = *((u32 *)e1);
+	u32 ld2 = *((u32 *)e2);
+	int i;
+
+	for (i = 0; ld_prefs[i] != 0; i++) {
+		if (ld1 == ld_prefs[i])
+			return -1;
+
+		if (ld2 == ld_prefs[i])
+			return 1;
+	}
+	return 0;
+}
+
 /*
  * Try to set the server's pnfs module to the pnfs layout type specified by id.
  * Currently only one pNFS layout driver per filesystem is supported.
  *
- * @id layout type. Zero (illegal layout type) indicates pNFS not in use.
+ * @ids array of layout types supported by MDS.
  */
 void
 set_pnfs_layoutdriver(struct nfs_server *server, const struct nfs_fh *mntfh,
-		      u32 id)
+		      struct nfs_fsinfo *fsinfo)
 {
 	struct pnfs_layoutdriver_type *ld_type = NULL;
+	u32 id;
+	int i;
 
-	if (id == 0)
-		goto out_no_driver;
 	if (!(server->nfs_client->cl_exchange_flags &
 		 (EXCHGID4_FLAG_USE_NON_PNFS | EXCHGID4_FLAG_USE_PNFS_MDS))) {
-		printk(KERN_ERR "NFS: %s: id %u cl_exchange_flags 0x%x\n",
-			__func__, id, server->nfs_client->cl_exchange_flags);
+		printk(KERN_ERR "NFS: %s: cl_exchange_flags 0x%x\n",
+			__func__, server->nfs_client->cl_exchange_flags);
 		goto out_no_driver;
 	}
-	ld_type = find_pnfs_driver(id);
-	if (!ld_type) {
-		request_module("%s-%u", LAYOUT_NFSV4_1_MODULE_PREFIX, id);
+
+	sort(fsinfo->layouttype, fsinfo->nlayouttypes,
+		sizeof(*fsinfo->layouttype), ld_cmp, NULL);
+
+	for (i = 0; i < fsinfo->nlayouttypes; i++) {
+		id = fsinfo->layouttype[i];
 		ld_type = find_pnfs_driver(id);
 		if (!ld_type) {
-			dprintk("%s: No pNFS module found for %u.\n",
-				__func__, id);
-			goto out_no_driver;
+			request_module("%s-%u", LAYOUT_NFSV4_1_MODULE_PREFIX,
+					id);
+			ld_type = find_pnfs_driver(id);
 		}
+		if (ld_type)
+			break;
+	}
+
+	if (!ld_type) {
+		dprintk("%s: No pNFS module found!\n", __func__);
+		goto out_no_driver;
 	}
+
 	server->pnfs_curr_ld = ld_type;
 	if (ld_type->set_layoutdriver
 	    && ld_type->set_layoutdriver(server, mntfh)) {
@@ -2185,10 +2230,8 @@ static void pnfs_ld_handle_read_error(struct nfs_pgio_header *hdr)
  */
 void pnfs_ld_read_done(struct nfs_pgio_header *hdr)
 {
-	if (likely(!hdr->pnfs_error)) {
-		__nfs4_read_done_cb(hdr);
+	if (likely(!hdr->pnfs_error))
 		hdr->mds_ops->rpc_call_done(&hdr->task, hdr);
-	}
 	trace_nfs4_pnfs_read(hdr, hdr->pnfs_error);
 	if (unlikely(hdr->pnfs_error))
 		pnfs_ld_handle_read_error(hdr);

fs/nfs/pnfs.h

@@ -236,7 +236,7 @@ void pnfs_get_layout_hdr(struct pnfs_layout_hdr *lo);
 void pnfs_put_lseg(struct pnfs_layout_segment *lseg);
 void pnfs_put_lseg_locked(struct pnfs_layout_segment *lseg);
 
-void set_pnfs_layoutdriver(struct nfs_server *, const struct nfs_fh *, u32);
+void set_pnfs_layoutdriver(struct nfs_server *, const struct nfs_fh *, struct nfs_fsinfo *);
 void unset_pnfs_layoutdriver(struct nfs_server *);
 void pnfs_generic_pg_init_read(struct nfs_pageio_descriptor *, struct nfs_page *);
 int pnfs_generic_pg_readpages(struct nfs_pageio_descriptor *desc);
@@ -657,7 +657,8 @@ pnfs_wait_on_layoutreturn(struct inode *ino, struct rpc_task *task)
 }
 
 static inline void set_pnfs_layoutdriver(struct nfs_server *s,
-					 const struct nfs_fh *mntfh, u32 id)
+					 const struct nfs_fh *mntfh,
+					 struct nfs_fsinfo *fsinfo)
 {
 }
 

fs/nfs/pnfs_nfs.c

@@ -690,13 +690,50 @@ static int _nfs4_pnfs_v4_ds_connect(struct nfs_server *mds_srv,
 		dprintk("%s: DS %s: trying address %s\n",
 			__func__, ds->ds_remotestr, da->da_remotestr);
 
-		clp = nfs4_set_ds_client(mds_srv,
-					(struct sockaddr *)&da->da_addr,
-					da->da_addrlen, IPPROTO_TCP,
-					timeo, retrans, minor_version,
-					au_flavor);
-		if (!IS_ERR(clp))
-			break;
+		if (!IS_ERR(clp) && clp->cl_mvops->session_trunk) {
+			struct xprt_create xprt_args = {
+				.ident = XPRT_TRANSPORT_TCP,
+				.net = clp->cl_net,
+				.dstaddr = (struct sockaddr *)&da->da_addr,
+				.addrlen = da->da_addrlen,
+				.servername = clp->cl_hostname,
+			};
+			struct nfs4_add_xprt_data xprtdata = {
+				.clp = clp,
+				.cred = nfs4_get_clid_cred(clp),
+			};
+			struct rpc_add_xprt_test rpcdata = {
+				.add_xprt_test = clp->cl_mvops->session_trunk,
+				.data = &xprtdata,
+			};
+
+			/**
+			* Test this address for session trunking and
+			* add as an alias
+			*/
+			rpc_clnt_add_xprt(clp->cl_rpcclient, &xprt_args,
+					  rpc_clnt_setup_test_and_add_xprt,
+					  &rpcdata);
+			if (xprtdata.cred)
+				put_rpccred(xprtdata.cred);
+		} else {
+			clp = nfs4_set_ds_client(mds_srv,
+						(struct sockaddr *)&da->da_addr,
+						da->da_addrlen, IPPROTO_TCP,
+						timeo, retrans, minor_version,
+						au_flavor);
+			if (IS_ERR(clp))
+				continue;
+
+			status = nfs4_init_ds_session(clp,
+					mds_srv->nfs_client->cl_lease_time);
+			if (status) {
+				nfs_put_client(clp);
+				clp = ERR_PTR(-EIO);
+				continue;
+			}
+
+		}
 	}
 
 	if (IS_ERR(clp)) {
@@ -704,18 +741,11 @@ static int _nfs4_pnfs_v4_ds_connect(struct nfs_server *mds_srv,
 		goto out;
 	}
 
-	status = nfs4_init_ds_session(clp, mds_srv->nfs_client->cl_lease_time);
-	if (status)
-		goto out_put;
-
 	smp_wmb();
 	ds->ds_clp = clp;
 	dprintk("%s [new] addr: %s\n", __func__, ds->ds_remotestr);
 out:
 	return status;
-out_put:
-	nfs_put_client(clp);
-	goto out;
 }
 
 /*

fs/nfs/super.c

@@ -2848,19 +2848,23 @@ out_invalid_transport_udp:
  * NFS client for backwards compatibility
  */
 unsigned int nfs_callback_set_tcpport;
+unsigned short nfs_callback_nr_threads;
 /* Default cache timeout is 10 minutes */
 unsigned int nfs_idmap_cache_timeout = 600;
 /* Turn off NFSv4 uid/gid mapping when using AUTH_SYS */
 bool nfs4_disable_idmapping = true;
 unsigned short max_session_slots = NFS4_DEF_SLOT_TABLE_SIZE;
+unsigned short max_session_cb_slots = NFS4_DEF_CB_SLOT_TABLE_SIZE;
 unsigned short send_implementation_id = 1;
 char nfs4_client_id_uniquifier[NFS4_CLIENT_ID_UNIQ_LEN] = "";
 bool recover_lost_locks = false;
 
+EXPORT_SYMBOL_GPL(nfs_callback_nr_threads);
 EXPORT_SYMBOL_GPL(nfs_callback_set_tcpport);
 EXPORT_SYMBOL_GPL(nfs_idmap_cache_timeout);
 EXPORT_SYMBOL_GPL(nfs4_disable_idmapping);
 EXPORT_SYMBOL_GPL(max_session_slots);
+EXPORT_SYMBOL_GPL(max_session_cb_slots);
 EXPORT_SYMBOL_GPL(send_implementation_id);
 EXPORT_SYMBOL_GPL(nfs4_client_id_uniquifier);
 EXPORT_SYMBOL_GPL(recover_lost_locks);
@@ -2887,6 +2891,9 @@ static const struct kernel_param_ops param_ops_portnr = {
 #define param_check_portnr(name, p) __param_check(name, p, unsigned int);
 
 module_param_named(callback_tcpport, nfs_callback_set_tcpport, portnr, 0644);
+module_param_named(callback_nr_threads, nfs_callback_nr_threads, ushort, 0644);
+MODULE_PARM_DESC(callback_nr_threads, "Number of threads that will be "
+		"assigned to the NFSv4 callback channels.");
 module_param(nfs_idmap_cache_timeout, int, 0644);
 module_param(nfs4_disable_idmapping, bool, 0644);
 module_param_string(nfs4_unique_id, nfs4_client_id_uniquifier,
@@ -2896,6 +2903,9 @@ MODULE_PARM_DESC(nfs4_disable_idmapping,
 module_param(max_session_slots, ushort, 0644);
 MODULE_PARM_DESC(max_session_slots, "Maximum number of outstanding NFSv4.1 "
 		"requests the client will negotiate");
+module_param(max_session_cb_slots, ushort, 0644);
+MODULE_PARM_DESC(max_session_slots, "Maximum number of parallel NFSv4.1 "
+		"callbacks the client will process for a given server");
 module_param(send_implementation_id, ushort, 0644);
 MODULE_PARM_DESC(send_implementation_id,
 		"Send implementation ID with NFSv4.1 exchange_id");

include/linux/nfs4.h

@@ -67,6 +67,7 @@ struct nfs4_stateid_struct {
 		NFS4_DELEGATION_STATEID_TYPE,
 		NFS4_LAYOUT_STATEID_TYPE,
 		NFS4_PNFS_DS_STATEID_TYPE,
+		NFS4_REVOKED_STATEID_TYPE,
 	} type;
 };
 

include/linux/nfs_fs_sb.h

@@ -103,6 +103,9 @@ struct nfs_client {
 #define NFS_SP4_MACH_CRED_WRITE    5	/* WRITE */
 #define NFS_SP4_MACH_CRED_COMMIT   6	/* COMMIT */
 #define NFS_SP4_MACH_CRED_PNFS_CLEANUP  7 /* LAYOUTRETURN */
+#if IS_ENABLED(CONFIG_NFS_V4_1)
+	wait_queue_head_t	cl_lock_waitq;
+#endif /* CONFIG_NFS_V4_1 */
 #endif /* CONFIG_NFS_V4 */
 
 	/* Our own IP address, as a null-terminated string.

include/linux/nfs_xdr.h

@@ -124,6 +124,11 @@ struct nfs_fattr {
 		| NFS_ATTR_FATTR_SPACE_USED \
 		| NFS_ATTR_FATTR_V4_SECURITY_LABEL)
 
+/*
+ * Maximal number of supported layout drivers.
+ */
+#define NFS_MAX_LAYOUT_TYPES 8
+
 /*
  * Info on the file system
  */
@@ -139,7 +144,8 @@ struct nfs_fsinfo {
 	__u64			maxfilesize;
 	struct timespec		time_delta; /* server time granularity */
 	__u32			lease_time; /* in seconds */
-	__u32			layouttype; /* supported pnfs layout driver */
+	__u32			nlayouttypes; /* number of layouttypes */
+	__u32			layouttype[NFS_MAX_LAYOUT_TYPES]; /* supported pnfs layout driver */
 	__u32			blksize; /* preferred pnfs io block size */
 	__u32			clone_blksize; /* granularity of a CLONE operation */
 };

include/linux/sunrpc/auth.h

@@ -131,6 +131,7 @@ struct rpc_authops {
 	struct rpc_auth *	(*create)(struct rpc_auth_create_args *, struct rpc_clnt *);
 	void			(*destroy)(struct rpc_auth *);
 
+	int			(*hash_cred)(struct auth_cred *, unsigned int);
 	struct rpc_cred *	(*lookup_cred)(struct rpc_auth *, struct auth_cred *, int);
 	struct rpc_cred *	(*crcreate)(struct rpc_auth*, struct auth_cred *, int, gfp_t);
 	int			(*list_pseudoflavors)(rpc_authflavor_t *, int);

include/linux/sunrpc/clnt.h

@@ -125,6 +125,13 @@ struct rpc_create_args {
 	struct svc_xprt		*bc_xprt;	/* NFSv4.1 backchannel */
 };
 
+struct rpc_add_xprt_test {
+	int (*add_xprt_test)(struct rpc_clnt *,
+		struct rpc_xprt *,
+		void *calldata);
+	void *data;
+};
+
 /* Values for "flags" field */
 #define RPC_CLNT_CREATE_HARDRTRY	(1UL << 0)
 #define RPC_CLNT_CREATE_AUTOBIND	(1UL << 2)
@@ -198,6 +205,16 @@ int		rpc_clnt_add_xprt(struct rpc_clnt *, struct xprt_create *,
 void		rpc_cap_max_reconnect_timeout(struct rpc_clnt *clnt,
 			unsigned long timeo);
 
+int		rpc_clnt_setup_test_and_add_xprt(struct rpc_clnt *,
+			struct rpc_xprt_switch *,
+			struct rpc_xprt *,
+			void *);
+
 const char *rpc_proc_name(const struct rpc_task *task);
+
+void rpc_clnt_xprt_switch_put(struct rpc_clnt *);
+void rpc_clnt_xprt_switch_add_xprt(struct rpc_clnt *, struct rpc_xprt *);
+bool rpc_clnt_xprt_switch_has_addr(struct rpc_clnt *clnt,
+			const struct sockaddr *sap);
 #endif /* __KERNEL__ */
 #endif /* _LINUX_SUNRPC_CLNT_H */

include/linux/sunrpc/rpc_rdma.h

@@ -46,6 +46,10 @@
 #define RPCRDMA_VERSION		1
 #define rpcrdma_version		cpu_to_be32(RPCRDMA_VERSION)
 
+enum {
+	RPCRDMA_V1_DEF_INLINE_SIZE	= 1024,
+};
+
 struct rpcrdma_segment {
 	__be32 rs_handle;	/* Registered memory handle */
 	__be32 rs_length;	/* Length of the chunk in bytes */

include/linux/sunrpc/sched.h

@@ -239,8 +239,8 @@ struct rpc_task *rpc_wake_up_first(struct rpc_wait_queue *,
 					void *);
 void		rpc_wake_up_status(struct rpc_wait_queue *, int);
 void		rpc_delay(struct rpc_task *, unsigned long);
-void *		rpc_malloc(struct rpc_task *, size_t);
-void		rpc_free(void *);
+int		rpc_malloc(struct rpc_task *);
+void		rpc_free(struct rpc_task *);
 int		rpciod_up(void);
 void		rpciod_down(void);
 int		__rpc_wait_for_completion_task(struct rpc_task *task, wait_bit_action_f *);

include/linux/sunrpc/xdr.h

@@ -67,6 +67,18 @@ struct xdr_buf {
 			len;		/* Length of XDR encoded message */
 };
 
+static inline void
+xdr_buf_init(struct xdr_buf *buf, void *start, size_t len)
+{
+	buf->head[0].iov_base = start;
+	buf->head[0].iov_len = len;
+	buf->tail[0].iov_len = 0;
+	buf->page_len = 0;
+	buf->flags = 0;
+	buf->len = 0;
+	buf->buflen = len;
+}
+
 /*
  * pre-xdr'ed macros.
  */

include/linux/sunrpc/xprt.h

@@ -83,9 +83,11 @@ struct rpc_rqst {
 	void (*rq_release_snd_buf)(struct rpc_rqst *); /* release rq_enc_pages */
 	struct list_head	rq_list;
 
-	__u32 *			rq_buffer;	/* XDR encode buffer */
-	size_t			rq_callsize,
-				rq_rcvsize;
+	void			*rq_xprtdata;	/* Per-xprt private data */
+	void			*rq_buffer;	/* Call XDR encode buffer */
+	size_t			rq_callsize;
+	void			*rq_rbuffer;	/* Reply XDR decode buffer */
+	size_t			rq_rcvsize;
 	size_t			rq_xmit_bytes_sent;	/* total bytes sent */
 	size_t			rq_reply_bytes_recvd;	/* total reply bytes */
 							/* received */
@@ -127,8 +129,8 @@ struct rpc_xprt_ops {
 	void		(*rpcbind)(struct rpc_task *task);
 	void		(*set_port)(struct rpc_xprt *xprt, unsigned short port);
 	void		(*connect)(struct rpc_xprt *xprt, struct rpc_task *task);
-	void *		(*buf_alloc)(struct rpc_task *task, size_t size);
-	void		(*buf_free)(void *buffer);
+	int		(*buf_alloc)(struct rpc_task *task);
+	void		(*buf_free)(struct rpc_task *task);
 	int		(*send_request)(struct rpc_task *task);
 	void		(*set_retrans_timeout)(struct rpc_task *task);
 	void		(*timer)(struct rpc_xprt *xprt, struct rpc_task *task);

include/linux/sunrpc/xprtmultipath.h

@@ -66,4 +66,6 @@ extern struct rpc_xprt *xprt_iter_xprt(struct rpc_xprt_iter *xpi);
 extern struct rpc_xprt *xprt_iter_get_xprt(struct rpc_xprt_iter *xpi);
 extern struct rpc_xprt *xprt_iter_get_next(struct rpc_xprt_iter *xpi);
 
+extern bool rpc_xprt_switch_has_addr(struct rpc_xprt_switch *xps,
+		const struct sockaddr *sap);
 #endif

include/linux/sunrpc/xprtrdma.h

@@ -53,8 +53,8 @@
 #define RPCRDMA_MAX_SLOT_TABLE	(256U)
 
 #define RPCRDMA_MIN_INLINE  (1024)	/* min inline thresh */
-#define RPCRDMA_DEF_INLINE  (1024)	/* default inline thresh */
-#define RPCRDMA_MAX_INLINE  (3068)	/* max inline thresh */
+#define RPCRDMA_DEF_INLINE  (4096)	/* default inline thresh */
+#define RPCRDMA_MAX_INLINE  (65536)	/* max inline thresh */
 
 /* Memory registration strategies, by number.
  * This is part of a kernel / user space API. Do not remove. */

net/sunrpc/auth.c

@@ -551,7 +551,7 @@ rpcauth_lookup_credcache(struct rpc_auth *auth, struct auth_cred * acred,
 			*entry, *new;
 	unsigned int nr;
 
-	nr = hash_long(from_kuid(&init_user_ns, acred->uid), cache->hashbits);
+	nr = auth->au_ops->hash_cred(acred, cache->hashbits);
 
 	rcu_read_lock();
 	hlist_for_each_entry_rcu(entry, &cache->hashtable[nr], cr_hash) {

net/sunrpc/auth_generic.c

@@ -78,6 +78,14 @@ static struct rpc_cred *generic_bind_cred(struct rpc_task *task,
 	return auth->au_ops->lookup_cred(auth, acred, lookupflags);
 }
 
+static int
+generic_hash_cred(struct auth_cred *acred, unsigned int hashbits)
+{
+	return hash_64(from_kgid(&init_user_ns, acred->gid) |
+		((u64)from_kuid(&init_user_ns, acred->uid) <<
+			(sizeof(gid_t) * 8)), hashbits);
+}
+
 /*
  * Lookup generic creds for current process
  */
@@ -258,6 +266,7 @@ generic_key_timeout(struct rpc_auth *auth, struct rpc_cred *cred)
 static const struct rpc_authops generic_auth_ops = {
 	.owner = THIS_MODULE,
 	.au_name = "Generic",
+	.hash_cred = generic_hash_cred,
 	.lookup_cred = generic_lookup_cred,
 	.crcreate = generic_create_cred,
 	.key_timeout = generic_key_timeout,

net/sunrpc/auth_gss/auth_gss.c

@@ -1298,6 +1298,12 @@ gss_destroy_cred(struct rpc_cred *cred)
 	gss_destroy_nullcred(cred);
 }
 
+static int
+gss_hash_cred(struct auth_cred *acred, unsigned int hashbits)
+{
+	return hash_64(from_kuid(&init_user_ns, acred->uid), hashbits);
+}
+
 /*
  * Lookup RPCSEC_GSS cred for the current process
  */
@@ -1982,6 +1988,7 @@ static const struct rpc_authops authgss_ops = {
 	.au_name	= "RPCSEC_GSS",
 	.create		= gss_create,
 	.destroy	= gss_destroy,
+	.hash_cred	= gss_hash_cred,
 	.lookup_cred	= gss_lookup_cred,
 	.crcreate	= gss_create_cred,
 	.list_pseudoflavors = gss_mech_list_pseudoflavors,

net/sunrpc/auth_unix.c

@@ -46,6 +46,14 @@ unx_destroy(struct rpc_auth *auth)
 	rpcauth_clear_credcache(auth->au_credcache);
 }
 
+static int
+unx_hash_cred(struct auth_cred *acred, unsigned int hashbits)
+{
+	return hash_64(from_kgid(&init_user_ns, acred->gid) |
+		((u64)from_kuid(&init_user_ns, acred->uid) <<
+			(sizeof(gid_t) * 8)), hashbits);
+}
+
 /*
  * Lookup AUTH_UNIX creds for current process
  */
@@ -220,6 +228,7 @@ const struct rpc_authops authunix_ops = {
 	.au_name	= "UNIX",
 	.create		= unx_create,
 	.destroy	= unx_destroy,
+	.hash_cred	= unx_hash_cred,
 	.lookup_cred	= unx_lookup_cred,
 	.crcreate	= unx_create_cred,
 };

net/sunrpc/backchannel_rqst.c

@@ -76,13 +76,7 @@ static int xprt_alloc_xdr_buf(struct xdr_buf *buf, gfp_t gfp_flags)
 	page = alloc_page(gfp_flags);
 	if (page == NULL)
 		return -ENOMEM;
-	buf->head[0].iov_base = page_address(page);
-	buf->head[0].iov_len = PAGE_SIZE;
-	buf->tail[0].iov_base = NULL;
-	buf->tail[0].iov_len = 0;
-	buf->page_len = 0;
-	buf->len = 0;
-	buf->buflen = PAGE_SIZE;
+	xdr_buf_init(buf, page_address(page), PAGE_SIZE);
 	return 0;
 }
 

net/sunrpc/cache.c

@@ -353,7 +353,7 @@ void sunrpc_init_cache_detail(struct cache_detail *cd)
 	spin_unlock(&cache_list_lock);
 
 	/* start the cleaning process */
-	schedule_delayed_work(&cache_cleaner, 0);
+	queue_delayed_work(system_power_efficient_wq, &cache_cleaner, 0);
 }
 EXPORT_SYMBOL_GPL(sunrpc_init_cache_detail);
 
@@ -476,7 +476,8 @@ static void do_cache_clean(struct work_struct *work)
 		delay = 0;
 
 	if (delay)
-		schedule_delayed_work(&cache_cleaner, delay);
+		queue_delayed_work(system_power_efficient_wq,
+				   &cache_cleaner, delay);
 }
 
 

net/sunrpc/clnt.c

@@ -184,7 +184,6 @@ static int __rpc_clnt_handle_event(struct rpc_clnt *clnt, unsigned long event,
 				   struct super_block *sb)
 {
 	struct dentry *dentry;
-	int err = 0;
 
 	switch (event) {
 	case RPC_PIPEFS_MOUNT:
@@ -201,7 +200,7 @@ static int __rpc_clnt_handle_event(struct rpc_clnt *clnt, unsigned long event,
 		printk(KERN_ERR "%s: unknown event: %ld\n", __func__, event);
 		return -ENOTSUPP;
 	}
-	return err;
+	return 0;
 }
 
 static int __rpc_pipefs_event(struct rpc_clnt *clnt, unsigned long event,
@@ -988,7 +987,6 @@ void rpc_task_set_client(struct rpc_task *task, struct rpc_clnt *clnt)
 {
 
 	if (clnt != NULL) {
-		rpc_task_release_client(task);
 		if (task->tk_xprt == NULL)
 			task->tk_xprt = xprt_iter_get_next(&clnt->cl_xpi);
 		task->tk_client = clnt;
@@ -1693,6 +1691,7 @@ call_allocate(struct rpc_task *task)
 	struct rpc_rqst *req = task->tk_rqstp;
 	struct rpc_xprt *xprt = req->rq_xprt;
 	struct rpc_procinfo *proc = task->tk_msg.rpc_proc;
+	int status;
 
 	dprint_status(task);
 
@@ -1718,11 +1717,14 @@ call_allocate(struct rpc_task *task)
 	req->rq_rcvsize = RPC_REPHDRSIZE + slack + proc->p_replen;
 	req->rq_rcvsize <<= 2;
 
-	req->rq_buffer = xprt->ops->buf_alloc(task,
-					req->rq_callsize + req->rq_rcvsize);
-	if (req->rq_buffer != NULL)
-		return;
+	status = xprt->ops->buf_alloc(task);
 	xprt_inject_disconnect(xprt);
+	if (status == 0)
+		return;
+	if (status != -ENOMEM) {
+		rpc_exit(task, status);
+		return;
+	}
 
 	dprintk("RPC: %5u rpc_buffer allocation failed\n", task->tk_pid);
 
@@ -1748,18 +1750,6 @@ rpc_task_force_reencode(struct rpc_task *task)
 	task->tk_rqstp->rq_bytes_sent = 0;
 }
 
-static inline void
-rpc_xdr_buf_init(struct xdr_buf *buf, void *start, size_t len)
-{
-	buf->head[0].iov_base = start;
-	buf->head[0].iov_len = len;
-	buf->tail[0].iov_len = 0;
-	buf->page_len = 0;
-	buf->flags = 0;
-	buf->len = 0;
-	buf->buflen = len;
-}
-
 /*
  * 3.	Encode arguments of an RPC call
  */
@@ -1772,12 +1762,12 @@ rpc_xdr_encode(struct rpc_task *task)
 
 	dprint_status(task);
 
-	rpc_xdr_buf_init(&req->rq_snd_buf,
-			 req->rq_buffer,
-			 req->rq_callsize);
-	rpc_xdr_buf_init(&req->rq_rcv_buf,
-			 (char *)req->rq_buffer + req->rq_callsize,
-			 req->rq_rcvsize);
+	xdr_buf_init(&req->rq_snd_buf,
+		     req->rq_buffer,
+		     req->rq_callsize);
+	xdr_buf_init(&req->rq_rcv_buf,
+		     req->rq_rbuffer,
+		     req->rq_rcvsize);
 
 	p = rpc_encode_header(task);
 	if (p == NULL) {
@@ -2615,6 +2605,70 @@ int rpc_clnt_test_and_add_xprt(struct rpc_clnt *clnt,
 }
 EXPORT_SYMBOL_GPL(rpc_clnt_test_and_add_xprt);
 
+/**
+ * rpc_clnt_setup_test_and_add_xprt()
+ *
+ * This is an rpc_clnt_add_xprt setup() function which returns 1 so:
+ *   1) caller of the test function must dereference the rpc_xprt_switch
+ *   and the rpc_xprt.
+ *   2) test function must call rpc_xprt_switch_add_xprt, usually in
+ *   the rpc_call_done routine.
+ *
+ * Upon success (return of 1), the test function adds the new
+ * transport to the rpc_clnt xprt switch
+ *
+ * @clnt: struct rpc_clnt to get the new transport
+ * @xps:  the rpc_xprt_switch to hold the new transport
+ * @xprt: the rpc_xprt to test
+ * @data: a struct rpc_add_xprt_test pointer that holds the test function
+ *        and test function call data
+ */
+int rpc_clnt_setup_test_and_add_xprt(struct rpc_clnt *clnt,
+				     struct rpc_xprt_switch *xps,
+				     struct rpc_xprt *xprt,
+				     void *data)
+{
+	struct rpc_cred *cred;
+	struct rpc_task *task;
+	struct rpc_add_xprt_test *xtest = (struct rpc_add_xprt_test *)data;
+	int status = -EADDRINUSE;
+
+	xprt = xprt_get(xprt);
+	xprt_switch_get(xps);
+
+	if (rpc_xprt_switch_has_addr(xps, (struct sockaddr *)&xprt->addr))
+		goto out_err;
+
+	/* Test the connection */
+	cred = authnull_ops.lookup_cred(NULL, NULL, 0);
+	task = rpc_call_null_helper(clnt, xprt, cred,
+				    RPC_TASK_SOFT | RPC_TASK_SOFTCONN,
+				    NULL, NULL);
+	put_rpccred(cred);
+	if (IS_ERR(task)) {
+		status = PTR_ERR(task);
+		goto out_err;
+	}
+	status = task->tk_status;
+	rpc_put_task(task);
+
+	if (status < 0)
+		goto out_err;
+
+	/* rpc_xprt_switch and rpc_xprt are deferrenced by add_xprt_test() */
+	xtest->add_xprt_test(clnt, xprt, xtest->data);
+
+	/* so that rpc_clnt_add_xprt does not call rpc_xprt_switch_add_xprt */
+	return 1;
+out_err:
+	xprt_put(xprt);
+	xprt_switch_put(xps);
+	pr_info("RPC:   rpc_clnt_test_xprt failed: %d addr %s not added\n",
+		status, xprt->address_strings[RPC_DISPLAY_ADDR]);
+	return status;
+}
+EXPORT_SYMBOL_GPL(rpc_clnt_setup_test_and_add_xprt);
+
 /**
  * rpc_clnt_add_xprt - Add a new transport to a rpc_clnt
  * @clnt: pointer to struct rpc_clnt
@@ -2697,6 +2751,34 @@ rpc_cap_max_reconnect_timeout(struct rpc_clnt *clnt, unsigned long timeo)
 }
 EXPORT_SYMBOL_GPL(rpc_cap_max_reconnect_timeout);
 
+void rpc_clnt_xprt_switch_put(struct rpc_clnt *clnt)
+{
+	xprt_switch_put(rcu_dereference(clnt->cl_xpi.xpi_xpswitch));
+}
+EXPORT_SYMBOL_GPL(rpc_clnt_xprt_switch_put);
+
+void rpc_clnt_xprt_switch_add_xprt(struct rpc_clnt *clnt, struct rpc_xprt *xprt)
+{
+	rpc_xprt_switch_add_xprt(rcu_dereference(clnt->cl_xpi.xpi_xpswitch),
+				 xprt);
+}
+EXPORT_SYMBOL_GPL(rpc_clnt_xprt_switch_add_xprt);
+
+bool rpc_clnt_xprt_switch_has_addr(struct rpc_clnt *clnt,
+				   const struct sockaddr *sap)
+{
+	struct rpc_xprt_switch *xps;
+	bool ret;
+
+	xps = rcu_dereference(clnt->cl_xpi.xpi_xpswitch);
+
+	rcu_read_lock();
+	ret = rpc_xprt_switch_has_addr(xps, sap);
+	rcu_read_unlock();
+	return ret;
+}
+EXPORT_SYMBOL_GPL(rpc_clnt_xprt_switch_has_addr);
+
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
 static void rpc_show_header(void)
 {

net/sunrpc/sched.c

@@ -849,14 +849,17 @@ static void rpc_async_schedule(struct work_struct *work)
 }
 
 /**
- * rpc_malloc - allocate an RPC buffer
- * @task: RPC task that will use this buffer
- * @size: requested byte size
+ * rpc_malloc - allocate RPC buffer resources
+ * @task: RPC task
+ *
+ * A single memory region is allocated, which is split between the
+ * RPC call and RPC reply that this task is being used for. When
+ * this RPC is retired, the memory is released by calling rpc_free.
  *
  * To prevent rpciod from hanging, this allocator never sleeps,
- * returning NULL and suppressing warning if the request cannot be serviced
- * immediately.
- * The caller can arrange to sleep in a way that is safe for rpciod.
+ * returning -ENOMEM and suppressing warning if the request cannot
+ * be serviced immediately. The caller can arrange to sleep in a
+ * way that is safe for rpciod.
  *
  * Most requests are 'small' (under 2KiB) and can be serviced from a
  * mempool, ensuring that NFS reads and writes can always proceed,
@@ -865,8 +868,10 @@ static void rpc_async_schedule(struct work_struct *work)
  * In order to avoid memory starvation triggering more writebacks of
  * NFS requests, we avoid using GFP_KERNEL.
  */
-void *rpc_malloc(struct rpc_task *task, size_t size)
+int rpc_malloc(struct rpc_task *task)
 {
+	struct rpc_rqst *rqst = task->tk_rqstp;
+	size_t size = rqst->rq_callsize + rqst->rq_rcvsize;
 	struct rpc_buffer *buf;
 	gfp_t gfp = GFP_NOIO | __GFP_NOWARN;
 
@@ -880,28 +885,28 @@ void *rpc_malloc(struct rpc_task *task, size_t size)
 		buf = kmalloc(size, gfp);
 
 	if (!buf)
-		return NULL;
+		return -ENOMEM;
 
 	buf->len = size;
 	dprintk("RPC: %5u allocated buffer of size %zu at %p\n",
 			task->tk_pid, size, buf);
-	return &buf->data;
+	rqst->rq_buffer = buf->data;
+	rqst->rq_rbuffer = (char *)rqst->rq_buffer + rqst->rq_callsize;
+	return 0;
 }
 EXPORT_SYMBOL_GPL(rpc_malloc);
 
 /**
- * rpc_free - free buffer allocated via rpc_malloc
- * @buffer: buffer to free
+ * rpc_free - free RPC buffer resources allocated via rpc_malloc
+ * @task: RPC task
  *
  */
-void rpc_free(void *buffer)
+void rpc_free(struct rpc_task *task)
 {
+	void *buffer = task->tk_rqstp->rq_buffer;
 	size_t size;
 	struct rpc_buffer *buf;
 
-	if (!buffer)
-		return;
-
 	buf = container_of(buffer, struct rpc_buffer, data);
 	size = buf->len;
 

net/sunrpc/svc.c

@@ -401,6 +401,21 @@ int svc_bind(struct svc_serv *serv, struct net *net)
 }
 EXPORT_SYMBOL_GPL(svc_bind);
 
+#if defined(CONFIG_SUNRPC_BACKCHANNEL)
+static void
+__svc_init_bc(struct svc_serv *serv)
+{
+	INIT_LIST_HEAD(&serv->sv_cb_list);
+	spin_lock_init(&serv->sv_cb_lock);
+	init_waitqueue_head(&serv->sv_cb_waitq);
+}
+#else
+static void
+__svc_init_bc(struct svc_serv *serv)
+{
+}
+#endif
+
 /*
  * Create an RPC service
  */
@@ -443,6 +458,8 @@ __svc_create(struct svc_program *prog, unsigned int bufsize, int npools,
 	init_timer(&serv->sv_temptimer);
 	spin_lock_init(&serv->sv_lock);
 
+	__svc_init_bc(serv);
+
 	serv->sv_nrpools = npools;
 	serv->sv_pools =
 		kcalloc(serv->sv_nrpools, sizeof(struct svc_pool),

net/sunrpc/xdr.c

@@ -767,7 +767,7 @@ static void xdr_set_next_page(struct xdr_stream *xdr)
 	newbase -= xdr->buf->page_base;
 
 	if (xdr_set_page_base(xdr, newbase, PAGE_SIZE) < 0)
-		xdr_set_iov(xdr, xdr->buf->tail, xdr->buf->len);
+		xdr_set_iov(xdr, xdr->buf->tail, xdr->nwords << 2);
 }
 
 static bool xdr_set_next_buffer(struct xdr_stream *xdr)
@@ -776,7 +776,7 @@ static bool xdr_set_next_buffer(struct xdr_stream *xdr)
 		xdr_set_next_page(xdr);
 	else if (xdr->iov == xdr->buf->head) {
 		if (xdr_set_page_base(xdr, 0, PAGE_SIZE) < 0)
-			xdr_set_iov(xdr, xdr->buf->tail, xdr->buf->len);
+			xdr_set_iov(xdr, xdr->buf->tail, xdr->nwords << 2);
 	}
 	return xdr->p != xdr->end;
 }
@@ -859,12 +859,15 @@ EXPORT_SYMBOL_GPL(xdr_set_scratch_buffer);
 static __be32 *xdr_copy_to_scratch(struct xdr_stream *xdr, size_t nbytes)
 {
 	__be32 *p;
-	void *cpdest = xdr->scratch.iov_base;
+	char *cpdest = xdr->scratch.iov_base;
 	size_t cplen = (char *)xdr->end - (char *)xdr->p;
 
 	if (nbytes > xdr->scratch.iov_len)
 		return NULL;
-	memcpy(cpdest, xdr->p, cplen);
+	p = __xdr_inline_decode(xdr, cplen);
+	if (p == NULL)
+		return NULL;
+	memcpy(cpdest, p, cplen);
 	cpdest += cplen;
 	nbytes -= cplen;
 	if (!xdr_set_next_buffer(xdr))

net/sunrpc/xprt.c

@@ -1295,7 +1295,7 @@ void xprt_release(struct rpc_task *task)
 	xprt_schedule_autodisconnect(xprt);
 	spin_unlock_bh(&xprt->transport_lock);
 	if (req->rq_buffer)
-		xprt->ops->buf_free(req->rq_buffer);
+		xprt->ops->buf_free(task);
 	xprt_inject_disconnect(xprt);
 	if (req->rq_cred != NULL)
 		put_rpccred(req->rq_cred);

net/sunrpc/xprtmultipath.c

@@ -15,6 +15,7 @@
 #include <asm/cmpxchg.h>
 #include <linux/spinlock.h>
 #include <linux/sunrpc/xprt.h>
+#include <linux/sunrpc/addr.h>
 #include <linux/sunrpc/xprtmultipath.h>
 
 typedef struct rpc_xprt *(*xprt_switch_find_xprt_t)(struct list_head *head,
@@ -49,7 +50,8 @@ void rpc_xprt_switch_add_xprt(struct rpc_xprt_switch *xps,
 	if (xprt == NULL)
 		return;
 	spin_lock(&xps->xps_lock);
-	if (xps->xps_net == xprt->xprt_net || xps->xps_net == NULL)
+	if ((xps->xps_net == xprt->xprt_net || xps->xps_net == NULL) &&
+	    !rpc_xprt_switch_has_addr(xps, (struct sockaddr *)&xprt->addr))
 		xprt_switch_add_xprt_locked(xps, xprt);
 	spin_unlock(&xps->xps_lock);
 }
@@ -232,6 +234,26 @@ struct rpc_xprt *xprt_iter_current_entry(struct rpc_xprt_iter *xpi)
 	return xprt_switch_find_current_entry(head, xpi->xpi_cursor);
 }
 
+bool rpc_xprt_switch_has_addr(struct rpc_xprt_switch *xps,
+			      const struct sockaddr *sap)
+{
+	struct list_head *head;
+	struct rpc_xprt *pos;
+
+	if (xps == NULL || sap == NULL)
+		return false;
+
+	head = &xps->xps_xprt_list;
+	list_for_each_entry_rcu(pos, head, xprt_switch) {
+		if (rpc_cmp_addr_port(sap, (struct sockaddr *)&pos->addr)) {
+			pr_info("RPC:   addr %s already in xprt switch\n",
+				pos->address_strings[RPC_DISPLAY_ADDR]);
+			return true;
+		}
+	}
+	return false;
+}
+
 static
 struct rpc_xprt *xprt_switch_find_next_entry(struct list_head *head,
 		const struct rpc_xprt *cur)

net/sunrpc/xprtrdma/backchannel.c

@@ -27,7 +27,7 @@ static void rpcrdma_bc_free_rqst(struct rpcrdma_xprt *r_xprt,
 	list_del(&req->rl_all);
 	spin_unlock(&buf->rb_reqslock);
 
-	rpcrdma_destroy_req(&r_xprt->rx_ia, req);
+	rpcrdma_destroy_req(req);
 
 	kfree(rqst);
 }
@@ -35,10 +35,8 @@ static void rpcrdma_bc_free_rqst(struct rpcrdma_xprt *r_xprt,
 static int rpcrdma_bc_setup_rqst(struct rpcrdma_xprt *r_xprt,
 				 struct rpc_rqst *rqst)
 {
-	struct rpcrdma_ia *ia = &r_xprt->rx_ia;
 	struct rpcrdma_regbuf *rb;
 	struct rpcrdma_req *req;
-	struct xdr_buf *buf;
 	size_t size;
 
 	req = rpcrdma_create_req(r_xprt);
@@ -46,30 +44,19 @@ static int rpcrdma_bc_setup_rqst(struct rpcrdma_xprt *r_xprt,
 		return PTR_ERR(req);
 	req->rl_backchannel = true;
 
-	size = RPCRDMA_INLINE_WRITE_THRESHOLD(rqst);
-	rb = rpcrdma_alloc_regbuf(ia, size, GFP_KERNEL);
+	rb = rpcrdma_alloc_regbuf(RPCRDMA_HDRBUF_SIZE,
+				  DMA_TO_DEVICE, GFP_KERNEL);
 	if (IS_ERR(rb))
 		goto out_fail;
 	req->rl_rdmabuf = rb;
 
-	size += RPCRDMA_INLINE_READ_THRESHOLD(rqst);
-	rb = rpcrdma_alloc_regbuf(ia, size, GFP_KERNEL);
+	size = r_xprt->rx_data.inline_rsize;
+	rb = rpcrdma_alloc_regbuf(size, DMA_TO_DEVICE, GFP_KERNEL);
 	if (IS_ERR(rb))
 		goto out_fail;
-	rb->rg_owner = req;
 	req->rl_sendbuf = rb;
-	/* so that rpcr_to_rdmar works when receiving a request */
-	rqst->rq_buffer = (void *)req->rl_sendbuf->rg_base;
-
-	buf = &rqst->rq_snd_buf;
-	buf->head[0].iov_base = rqst->rq_buffer;
-	buf->head[0].iov_len = 0;
-	buf->tail[0].iov_base = NULL;
-	buf->tail[0].iov_len = 0;
-	buf->page_len = 0;
-	buf->len = 0;
-	buf->buflen = size;
-
+	xdr_buf_init(&rqst->rq_snd_buf, rb->rg_base, size);
+	rpcrdma_set_xprtdata(rqst, req);
 	return 0;
 
 out_fail:
@@ -219,7 +206,6 @@ int rpcrdma_bc_marshal_reply(struct rpc_rqst *rqst)
 	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
 	struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
 	struct rpcrdma_msg *headerp;
-	size_t rpclen;
 
 	headerp = rdmab_to_msg(req->rl_rdmabuf);
 	headerp->rm_xid = rqst->rq_xid;
@@ -231,26 +217,9 @@ int rpcrdma_bc_marshal_reply(struct rpc_rqst *rqst)
 	headerp->rm_body.rm_chunks[1] = xdr_zero;
 	headerp->rm_body.rm_chunks[2] = xdr_zero;
 
-	rpclen = rqst->rq_svec[0].iov_len;
-
-#ifdef RPCRDMA_BACKCHANNEL_DEBUG
-	pr_info("RPC:       %s: rpclen %zd headerp 0x%p lkey 0x%x\n",
-		__func__, rpclen, headerp, rdmab_lkey(req->rl_rdmabuf));
-	pr_info("RPC:       %s: RPC/RDMA: %*ph\n",
-		__func__, (int)RPCRDMA_HDRLEN_MIN, headerp);
-	pr_info("RPC:       %s:      RPC: %*ph\n",
-		__func__, (int)rpclen, rqst->rq_svec[0].iov_base);
-#endif
-
-	req->rl_send_iov[0].addr = rdmab_addr(req->rl_rdmabuf);
-	req->rl_send_iov[0].length = RPCRDMA_HDRLEN_MIN;
-	req->rl_send_iov[0].lkey = rdmab_lkey(req->rl_rdmabuf);
-
-	req->rl_send_iov[1].addr = rdmab_addr(req->rl_sendbuf);
-	req->rl_send_iov[1].length = rpclen;
-	req->rl_send_iov[1].lkey = rdmab_lkey(req->rl_sendbuf);
-
-	req->rl_niovs = 2;
+	if (!rpcrdma_prepare_send_sges(&r_xprt->rx_ia, req, RPCRDMA_HDRLEN_MIN,
+				       &rqst->rq_snd_buf, rpcrdma_noch))
+		return -EIO;
 	return 0;
 }
 
@@ -402,7 +371,7 @@ out_overflow:
 out_short:
 	pr_warn("RPC/RDMA short backward direction call\n");
 
-	if (rpcrdma_ep_post_recv(&r_xprt->rx_ia, &r_xprt->rx_ep, rep))
+	if (rpcrdma_ep_post_recv(&r_xprt->rx_ia, rep))
 		xprt_disconnect_done(xprt);
 	else
 		pr_warn("RPC:       %s: reposting rep %p\n",

net/sunrpc/xprtrdma/fmr_ops.c

@@ -160,9 +160,8 @@ static int
 fmr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep,
 	    struct rpcrdma_create_data_internal *cdata)
 {
-	rpcrdma_set_max_header_sizes(ia, cdata, max_t(unsigned int, 1,
-						      RPCRDMA_MAX_DATA_SEGS /
-						      RPCRDMA_MAX_FMR_SGES));
+	ia->ri_max_segs = max_t(unsigned int, 1, RPCRDMA_MAX_DATA_SEGS /
+				RPCRDMA_MAX_FMR_SGES);
 	return 0;
 }
 
@@ -274,6 +273,7 @@ fmr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
 	 */
 	list_for_each_entry(mw, &req->rl_registered, mw_list)
 		list_add_tail(&mw->fmr.fm_mr->list, &unmap_list);
+	r_xprt->rx_stats.local_inv_needed++;
 	rc = ib_unmap_fmr(&unmap_list);
 	if (rc)
 		goto out_reset;
@@ -331,4 +331,5 @@ const struct rpcrdma_memreg_ops rpcrdma_fmr_memreg_ops = {
 	.ro_init_mr			= fmr_op_init_mr,
 	.ro_release_mr			= fmr_op_release_mr,
 	.ro_displayname			= "fmr",
+	.ro_send_w_inv_ok		= 0,
 };

net/sunrpc/xprtrdma/frwr_ops.c

@@ -67,6 +67,8 @@
  * pending send queue WRs before the transport is reconnected.
  */
 
+#include <linux/sunrpc/rpc_rdma.h>
+
 #include "xprt_rdma.h"
 
 #if IS_ENABLED(CONFIG_SUNRPC_DEBUG)
@@ -161,7 +163,7 @@ __frwr_reset_mr(struct rpcrdma_ia *ia, struct rpcrdma_mw *r)
 		return PTR_ERR(f->fr_mr);
 	}
 
-	dprintk("RPC:       %s: recovered FRMR %p\n", __func__, r);
+	dprintk("RPC:       %s: recovered FRMR %p\n", __func__, f);
 	f->fr_state = FRMR_IS_INVALID;
 	return 0;
 }
@@ -242,9 +244,8 @@ frwr_op_open(struct rpcrdma_ia *ia, struct rpcrdma_ep *ep,
 					       depth;
 	}
 
-	rpcrdma_set_max_header_sizes(ia, cdata, max_t(unsigned int, 1,
-						      RPCRDMA_MAX_DATA_SEGS /
-						      ia->ri_max_frmr_depth));
+	ia->ri_max_segs = max_t(unsigned int, 1, RPCRDMA_MAX_DATA_SEGS /
+				ia->ri_max_frmr_depth);
 	return 0;
 }
 
@@ -329,7 +330,7 @@ frwr_wc_localinv_wake(struct ib_cq *cq, struct ib_wc *wc)
 	frmr = container_of(cqe, struct rpcrdma_frmr, fr_cqe);
 	if (wc->status != IB_WC_SUCCESS)
 		__frwr_sendcompletion_flush(wc, frmr, "localinv");
-	complete_all(&frmr->fr_linv_done);
+	complete(&frmr->fr_linv_done);
 }
 
 /* Post a REG_MR Work Request to register a memory region
@@ -396,7 +397,7 @@ frwr_op_map(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr_seg *seg,
 		goto out_mapmr_err;
 
 	dprintk("RPC:       %s: Using frmr %p to map %u segments (%u bytes)\n",
-		__func__, mw, mw->mw_nents, mr->length);
+		__func__, frmr, mw->mw_nents, mr->length);
 
 	key = (u8)(mr->rkey & 0x000000FF);
 	ib_update_fast_reg_key(mr, ++key);
@@ -449,6 +450,8 @@ __frwr_prepare_linv_wr(struct rpcrdma_mw *mw)
 	struct rpcrdma_frmr *f = &mw->frmr;
 	struct ib_send_wr *invalidate_wr;
 
+	dprintk("RPC:       %s: invalidating frmr %p\n", __func__, f);
+
 	f->fr_state = FRMR_IS_INVALID;
 	invalidate_wr = &f->fr_invwr;
 
@@ -472,6 +475,7 @@ static void
 frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
 {
 	struct ib_send_wr *invalidate_wrs, *pos, *prev, *bad_wr;
+	struct rpcrdma_rep *rep = req->rl_reply;
 	struct rpcrdma_ia *ia = &r_xprt->rx_ia;
 	struct rpcrdma_mw *mw, *tmp;
 	struct rpcrdma_frmr *f;
@@ -487,6 +491,12 @@ frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
 	f = NULL;
 	invalidate_wrs = pos = prev = NULL;
 	list_for_each_entry(mw, &req->rl_registered, mw_list) {
+		if ((rep->rr_wc_flags & IB_WC_WITH_INVALIDATE) &&
+		    (mw->mw_handle == rep->rr_inv_rkey)) {
+			mw->frmr.fr_state = FRMR_IS_INVALID;
+			continue;
+		}
+
 		pos = __frwr_prepare_linv_wr(mw);
 
 		if (!invalidate_wrs)
@@ -496,6 +506,8 @@ frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
 		prev = pos;
 		f = &mw->frmr;
 	}
+	if (!f)
+		goto unmap;
 
 	/* Strong send queue ordering guarantees that when the
 	 * last WR in the chain completes, all WRs in the chain
@@ -510,6 +522,7 @@ frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
 	 * replaces the QP. The RPC reply handler won't call us
 	 * unless ri_id->qp is a valid pointer.
 	 */
+	r_xprt->rx_stats.local_inv_needed++;
 	rc = ib_post_send(ia->ri_id->qp, invalidate_wrs, &bad_wr);
 	if (rc)
 		goto reset_mrs;
@@ -521,6 +534,8 @@ frwr_op_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)
 	 */
 unmap:
 	list_for_each_entry_safe(mw, tmp, &req->rl_registered, mw_list) {
+		dprintk("RPC:       %s: unmapping frmr %p\n",
+			__func__, &mw->frmr);
 		list_del_init(&mw->mw_list);
 		ib_dma_unmap_sg(ia->ri_device,
 				mw->mw_sg, mw->mw_nents, mw->mw_dir);
@@ -576,4 +591,5 @@ const struct rpcrdma_memreg_ops rpcrdma_frwr_memreg_ops = {
 	.ro_init_mr			= frwr_op_init_mr,
 	.ro_release_mr			= frwr_op_release_mr,
 	.ro_displayname			= "frwr",
+	.ro_send_w_inv_ok		= RPCRDMA_CMP_F_SND_W_INV_OK,
 };

net/sunrpc/xprtrdma/rpc_rdma.c

@@ -53,14 +53,6 @@
 # define RPCDBG_FACILITY	RPCDBG_TRANS
 #endif
 
-enum rpcrdma_chunktype {
-	rpcrdma_noch = 0,
-	rpcrdma_readch,
-	rpcrdma_areadch,
-	rpcrdma_writech,
-	rpcrdma_replych
-};
-
 static const char transfertypes[][12] = {
 	"inline",	/* no chunks */
 	"read list",	/* some argument via rdma read */
@@ -118,10 +110,12 @@ static unsigned int rpcrdma_max_reply_header_size(unsigned int maxsegs)
 	return size;
 }
 
-void rpcrdma_set_max_header_sizes(struct rpcrdma_ia *ia,
-				  struct rpcrdma_create_data_internal *cdata,
-				  unsigned int maxsegs)
+void rpcrdma_set_max_header_sizes(struct rpcrdma_xprt *r_xprt)
 {
+	struct rpcrdma_create_data_internal *cdata = &r_xprt->rx_data;
+	struct rpcrdma_ia *ia = &r_xprt->rx_ia;
+	unsigned int maxsegs = ia->ri_max_segs;
+
 	ia->ri_max_inline_write = cdata->inline_wsize -
 				  rpcrdma_max_call_header_size(maxsegs);
 	ia->ri_max_inline_read = cdata->inline_rsize -
@@ -155,42 +149,6 @@ static bool rpcrdma_results_inline(struct rpcrdma_xprt *r_xprt,
 	return rqst->rq_rcv_buf.buflen <= ia->ri_max_inline_read;
 }
 
-static int
-rpcrdma_tail_pullup(struct xdr_buf *buf)
-{
-	size_t tlen = buf->tail[0].iov_len;
-	size_t skip = tlen & 3;
-
-	/* Do not include the tail if it is only an XDR pad */
-	if (tlen < 4)
-		return 0;
-
-	/* xdr_write_pages() adds a pad at the beginning of the tail
-	 * if the content in "buf->pages" is unaligned. Force the
-	 * tail's actual content to land at the next XDR position
-	 * after the head instead.
-	 */
-	if (skip) {
-		unsigned char *src, *dst;
-		unsigned int count;
-
-		src = buf->tail[0].iov_base;
-		dst = buf->head[0].iov_base;
-		dst += buf->head[0].iov_len;
-
-		src += skip;
-		tlen -= skip;
-
-		dprintk("RPC:       %s: skip=%zu, memmove(%p, %p, %zu)\n",
-			__func__, skip, dst, src, tlen);
-
-		for (count = tlen; count; count--)
-			*dst++ = *src++;
-	}
-
-	return tlen;
-}
-
 /* Split "vec" on page boundaries into segments. FMR registers pages,
  * not a byte range. Other modes coalesce these segments into a single
  * MR when they can.
@@ -229,7 +187,8 @@ rpcrdma_convert_kvec(struct kvec *vec, struct rpcrdma_mr_seg *seg, int n)
 
 static int
 rpcrdma_convert_iovs(struct xdr_buf *xdrbuf, unsigned int pos,
-	enum rpcrdma_chunktype type, struct rpcrdma_mr_seg *seg)
+	enum rpcrdma_chunktype type, struct rpcrdma_mr_seg *seg,
+	bool reminv_expected)
 {
 	int len, n, p, page_base;
 	struct page **ppages;
@@ -271,6 +230,13 @@ rpcrdma_convert_iovs(struct xdr_buf *xdrbuf, unsigned int pos,
 	if (type == rpcrdma_readch)
 		return n;
 
+	/* When encoding the Write list, some servers need to see an extra
+	 * segment for odd-length Write chunks. The upper layer provides
+	 * space in the tail iovec for this purpose.
+	 */
+	if (type == rpcrdma_writech && reminv_expected)
+		return n;
+
 	if (xdrbuf->tail[0].iov_len) {
 		/* the rpcrdma protocol allows us to omit any trailing
 		 * xdr pad bytes, saving the server an RDMA operation. */
@@ -327,7 +293,7 @@ rpcrdma_encode_read_list(struct rpcrdma_xprt *r_xprt,
 	if (rtype == rpcrdma_areadch)
 		pos = 0;
 	seg = req->rl_segments;
-	nsegs = rpcrdma_convert_iovs(&rqst->rq_snd_buf, pos, rtype, seg);
+	nsegs = rpcrdma_convert_iovs(&rqst->rq_snd_buf, pos, rtype, seg, false);
 	if (nsegs < 0)
 		return ERR_PTR(nsegs);
 
@@ -391,7 +357,8 @@ rpcrdma_encode_write_list(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req,
 	seg = req->rl_segments;
 	nsegs = rpcrdma_convert_iovs(&rqst->rq_rcv_buf,
 				     rqst->rq_rcv_buf.head[0].iov_len,
-				     wtype, seg);
+				     wtype, seg,
+				     r_xprt->rx_ia.ri_reminv_expected);
 	if (nsegs < 0)
 		return ERR_PTR(nsegs);
 
@@ -456,7 +423,8 @@ rpcrdma_encode_reply_chunk(struct rpcrdma_xprt *r_xprt,
 	}
 
 	seg = req->rl_segments;
-	nsegs = rpcrdma_convert_iovs(&rqst->rq_rcv_buf, 0, wtype, seg);
+	nsegs = rpcrdma_convert_iovs(&rqst->rq_rcv_buf, 0, wtype, seg,
+				     r_xprt->rx_ia.ri_reminv_expected);
 	if (nsegs < 0)
 		return ERR_PTR(nsegs);
 
@@ -491,74 +459,184 @@ rpcrdma_encode_reply_chunk(struct rpcrdma_xprt *r_xprt,
 	return iptr;
 }
 
-/*
- * Copy write data inline.
- * This function is used for "small" requests. Data which is passed
- * to RPC via iovecs (or page list) is copied directly into the
- * pre-registered memory buffer for this request. For small amounts
- * of data, this is efficient. The cutoff value is tunable.
+/* Prepare the RPC-over-RDMA header SGE.
  */
-static void rpcrdma_inline_pullup(struct rpc_rqst *rqst)
+static bool
+rpcrdma_prepare_hdr_sge(struct rpcrdma_ia *ia, struct rpcrdma_req *req,
+			u32 len)
 {
-	int i, npages, curlen;
-	int copy_len;
-	unsigned char *srcp, *destp;
-	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt);
-	int page_base;
-	struct page **ppages;
+	struct rpcrdma_regbuf *rb = req->rl_rdmabuf;
+	struct ib_sge *sge = &req->rl_send_sge[0];
+
+	if (unlikely(!rpcrdma_regbuf_is_mapped(rb))) {
+		if (!__rpcrdma_dma_map_regbuf(ia, rb))
+			return false;
+		sge->addr = rdmab_addr(rb);
+		sge->lkey = rdmab_lkey(rb);
+	}
+	sge->length = len;
+
+	ib_dma_sync_single_for_device(ia->ri_device, sge->addr,
+				      sge->length, DMA_TO_DEVICE);
+	req->rl_send_wr.num_sge++;
+	return true;
+}
 
-	destp = rqst->rq_svec[0].iov_base;
-	curlen = rqst->rq_svec[0].iov_len;
-	destp += curlen;
+/* Prepare the Send SGEs. The head and tail iovec, and each entry
+ * in the page list, gets its own SGE.
+ */
+static bool
+rpcrdma_prepare_msg_sges(struct rpcrdma_ia *ia, struct rpcrdma_req *req,
+			 struct xdr_buf *xdr, enum rpcrdma_chunktype rtype)
+{
+	unsigned int sge_no, page_base, len, remaining;
+	struct rpcrdma_regbuf *rb = req->rl_sendbuf;
+	struct ib_device *device = ia->ri_device;
+	struct ib_sge *sge = req->rl_send_sge;
+	u32 lkey = ia->ri_pd->local_dma_lkey;
+	struct page *page, **ppages;
+
+	/* The head iovec is straightforward, as it is already
+	 * DMA-mapped. Sync the content that has changed.
+	 */
+	if (!rpcrdma_dma_map_regbuf(ia, rb))
+		return false;
+	sge_no = 1;
+	sge[sge_no].addr = rdmab_addr(rb);
+	sge[sge_no].length = xdr->head[0].iov_len;
+	sge[sge_no].lkey = rdmab_lkey(rb);
+	ib_dma_sync_single_for_device(device, sge[sge_no].addr,
+				      sge[sge_no].length, DMA_TO_DEVICE);
+
+	/* If there is a Read chunk, the page list is being handled
+	 * via explicit RDMA, and thus is skipped here. However, the
+	 * tail iovec may include an XDR pad for the page list, as
+	 * well as additional content, and may not reside in the
+	 * same page as the head iovec.
+	 */
+	if (rtype == rpcrdma_readch) {
+		len = xdr->tail[0].iov_len;
 
-	dprintk("RPC:       %s: destp 0x%p len %d hdrlen %d\n",
-		__func__, destp, rqst->rq_slen, curlen);
+		/* Do not include the tail if it is only an XDR pad */
+		if (len < 4)
+			goto out;
 
-	copy_len = rqst->rq_snd_buf.page_len;
+		page = virt_to_page(xdr->tail[0].iov_base);
+		page_base = (unsigned long)xdr->tail[0].iov_base & ~PAGE_MASK;
 
-	if (rqst->rq_snd_buf.tail[0].iov_len) {
-		curlen = rqst->rq_snd_buf.tail[0].iov_len;
-		if (destp + copy_len != rqst->rq_snd_buf.tail[0].iov_base) {
-			memmove(destp + copy_len,
-				rqst->rq_snd_buf.tail[0].iov_base, curlen);
-			r_xprt->rx_stats.pullup_copy_count += curlen;
+		/* If the content in the page list is an odd length,
+		 * xdr_write_pages() has added a pad at the beginning
+		 * of the tail iovec. Force the tail's non-pad content
+		 * to land at the next XDR position in the Send message.
+		 */
+		page_base += len & 3;
+		len -= len & 3;
+		goto map_tail;
+	}
+
+	/* If there is a page list present, temporarily DMA map
+	 * and prepare an SGE for each page to be sent.
+	 */
+	if (xdr->page_len) {
+		ppages = xdr->pages + (xdr->page_base >> PAGE_SHIFT);
+		page_base = xdr->page_base & ~PAGE_MASK;
+		remaining = xdr->page_len;
+		while (remaining) {
+			sge_no++;
+			if (sge_no > RPCRDMA_MAX_SEND_SGES - 2)
+				goto out_mapping_overflow;
+
+			len = min_t(u32, PAGE_SIZE - page_base, remaining);
+			sge[sge_no].addr = ib_dma_map_page(device, *ppages,
+							   page_base, len,
+							   DMA_TO_DEVICE);
+			if (ib_dma_mapping_error(device, sge[sge_no].addr))
+				goto out_mapping_err;
+			sge[sge_no].length = len;
+			sge[sge_no].lkey = lkey;
+
+			req->rl_mapped_sges++;
+			ppages++;
+			remaining -= len;
+			page_base = 0;
 		}
-		dprintk("RPC:       %s: tail destp 0x%p len %d\n",
-			__func__, destp + copy_len, curlen);
-		rqst->rq_svec[0].iov_len += curlen;
 	}
-	r_xprt->rx_stats.pullup_copy_count += copy_len;
 
-	page_base = rqst->rq_snd_buf.page_base;
-	ppages = rqst->rq_snd_buf.pages + (page_base >> PAGE_SHIFT);
-	page_base &= ~PAGE_MASK;
-	npages = PAGE_ALIGN(page_base+copy_len) >> PAGE_SHIFT;
-	for (i = 0; copy_len && i < npages; i++) {
-		curlen = PAGE_SIZE - page_base;
-		if (curlen > copy_len)
-			curlen = copy_len;
-		dprintk("RPC:       %s: page %d destp 0x%p len %d curlen %d\n",
-			__func__, i, destp, copy_len, curlen);
-		srcp = kmap_atomic(ppages[i]);
-		memcpy(destp, srcp+page_base, curlen);
-		kunmap_atomic(srcp);
-		rqst->rq_svec[0].iov_len += curlen;
-		destp += curlen;
-		copy_len -= curlen;
-		page_base = 0;
+	/* The tail iovec is not always constructed in the same
+	 * page where the head iovec resides (see, for example,
+	 * gss_wrap_req_priv). To neatly accommodate that case,
+	 * DMA map it separately.
+	 */
+	if (xdr->tail[0].iov_len) {
+		page = virt_to_page(xdr->tail[0].iov_base);
+		page_base = (unsigned long)xdr->tail[0].iov_base & ~PAGE_MASK;
+		len = xdr->tail[0].iov_len;
+
+map_tail:
+		sge_no++;
+		sge[sge_no].addr = ib_dma_map_page(device, page,
+						   page_base, len,
+						   DMA_TO_DEVICE);
+		if (ib_dma_mapping_error(device, sge[sge_no].addr))
+			goto out_mapping_err;
+		sge[sge_no].length = len;
+		sge[sge_no].lkey = lkey;
+		req->rl_mapped_sges++;
 	}
-	/* header now contains entire send message */
+
+out:
+	req->rl_send_wr.num_sge = sge_no + 1;
+	return true;
+
+out_mapping_overflow:
+	pr_err("rpcrdma: too many Send SGEs (%u)\n", sge_no);
+	return false;
+
+out_mapping_err:
+	pr_err("rpcrdma: Send mapping error\n");
+	return false;
+}
+
+bool
+rpcrdma_prepare_send_sges(struct rpcrdma_ia *ia, struct rpcrdma_req *req,
+			  u32 hdrlen, struct xdr_buf *xdr,
+			  enum rpcrdma_chunktype rtype)
+{
+	req->rl_send_wr.num_sge = 0;
+	req->rl_mapped_sges = 0;
+
+	if (!rpcrdma_prepare_hdr_sge(ia, req, hdrlen))
+		goto out_map;
+
+	if (rtype != rpcrdma_areadch)
+		if (!rpcrdma_prepare_msg_sges(ia, req, xdr, rtype))
+			goto out_map;
+
+	return true;
+
+out_map:
+	pr_err("rpcrdma: failed to DMA map a Send buffer\n");
+	return false;
+}
+
+void
+rpcrdma_unmap_sges(struct rpcrdma_ia *ia, struct rpcrdma_req *req)
+{
+	struct ib_device *device = ia->ri_device;
+	struct ib_sge *sge;
+	int count;
+
+	sge = &req->rl_send_sge[2];
+	for (count = req->rl_mapped_sges; count--; sge++)
+		ib_dma_unmap_page(device, sge->addr, sge->length,
+				  DMA_TO_DEVICE);
+	req->rl_mapped_sges = 0;
 }
 
 /*
  * Marshal a request: the primary job of this routine is to choose
  * the transfer modes. See comments below.
  *
- * Prepares up to two IOVs per Call message:
- *
- *  [0] -- RPC RDMA header
- *  [1] -- the RPC header/data
- *
  * Returns zero on success, otherwise a negative errno.
  */
 
@@ -626,12 +704,11 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst)
 	 */
 	if (rpcrdma_args_inline(r_xprt, rqst)) {
 		rtype = rpcrdma_noch;
-		rpcrdma_inline_pullup(rqst);
-		rpclen = rqst->rq_svec[0].iov_len;
+		rpclen = rqst->rq_snd_buf.len;
 	} else if (ddp_allowed && rqst->rq_snd_buf.flags & XDRBUF_WRITE) {
 		rtype = rpcrdma_readch;
-		rpclen = rqst->rq_svec[0].iov_len;
-		rpclen += rpcrdma_tail_pullup(&rqst->rq_snd_buf);
+		rpclen = rqst->rq_snd_buf.head[0].iov_len +
+			 rqst->rq_snd_buf.tail[0].iov_len;
 	} else {
 		r_xprt->rx_stats.nomsg_call_count++;
 		headerp->rm_type = htonl(RDMA_NOMSG);
@@ -673,34 +750,18 @@ rpcrdma_marshal_req(struct rpc_rqst *rqst)
 		goto out_unmap;
 	hdrlen = (unsigned char *)iptr - (unsigned char *)headerp;
 
-	if (hdrlen + rpclen > RPCRDMA_INLINE_WRITE_THRESHOLD(rqst))
-		goto out_overflow;
-
 	dprintk("RPC: %5u %s: %s/%s: hdrlen %zd rpclen %zd\n",
 		rqst->rq_task->tk_pid, __func__,
 		transfertypes[rtype], transfertypes[wtype],
 		hdrlen, rpclen);
 
-	req->rl_send_iov[0].addr = rdmab_addr(req->rl_rdmabuf);
-	req->rl_send_iov[0].length = hdrlen;
-	req->rl_send_iov[0].lkey = rdmab_lkey(req->rl_rdmabuf);
-
-	req->rl_niovs = 1;
-	if (rtype == rpcrdma_areadch)
-		return 0;
-
-	req->rl_send_iov[1].addr = rdmab_addr(req->rl_sendbuf);
-	req->rl_send_iov[1].length = rpclen;
-	req->rl_send_iov[1].lkey = rdmab_lkey(req->rl_sendbuf);
-
-	req->rl_niovs = 2;
+	if (!rpcrdma_prepare_send_sges(&r_xprt->rx_ia, req, hdrlen,
+				       &rqst->rq_snd_buf, rtype)) {
+		iptr = ERR_PTR(-EIO);
+		goto out_unmap;
+	}
 	return 0;
 
-out_overflow:
-	pr_err("rpcrdma: send overflow: hdrlen %zd rpclen %zu %s/%s\n",
-		hdrlen, rpclen, transfertypes[rtype], transfertypes[wtype]);
-	iptr = ERR_PTR(-EIO);
-
 out_unmap:
 	r_xprt->rx_ia.ri_ops->ro_unmap_safe(r_xprt, req, false);
 	return PTR_ERR(iptr);
@@ -916,8 +977,10 @@ rpcrdma_conn_func(struct rpcrdma_ep *ep)
  * allowed to timeout, to discover the errors at that time.
  */
 void
-rpcrdma_reply_handler(struct rpcrdma_rep *rep)
+rpcrdma_reply_handler(struct work_struct *work)
 {
+	struct rpcrdma_rep *rep =
+			container_of(work, struct rpcrdma_rep, rr_work);
 	struct rpcrdma_msg *headerp;
 	struct rpcrdma_req *req;
 	struct rpc_rqst *rqst;
@@ -1132,6 +1195,6 @@ out_duplicate:
 
 repost:
 	r_xprt->rx_stats.bad_reply_count++;
-	if (rpcrdma_ep_post_recv(&r_xprt->rx_ia, &r_xprt->rx_ep, rep))
+	if (rpcrdma_ep_post_recv(&r_xprt->rx_ia, rep))
 		rpcrdma_recv_buffer_put(rep);
 }

net/sunrpc/xprtrdma/svc_rdma_backchannel.c

@@ -159,33 +159,34 @@ out_unmap:
 /* Server-side transport endpoint wants a whole page for its send
  * buffer. The client RPC code constructs the RPC header in this
  * buffer before it invokes ->send_request.
- *
- * Returns NULL if there was a temporary allocation failure.
  */
-static void *
-xprt_rdma_bc_allocate(struct rpc_task *task, size_t size)
+static int
+xprt_rdma_bc_allocate(struct rpc_task *task)
 {
 	struct rpc_rqst *rqst = task->tk_rqstp;
 	struct svc_xprt *sxprt = rqst->rq_xprt->bc_xprt;
+	size_t size = rqst->rq_callsize;
 	struct svcxprt_rdma *rdma;
 	struct page *page;
 
 	rdma = container_of(sxprt, struct svcxprt_rdma, sc_xprt);
 
-	/* Prevent an infinite loop: try to make this case work */
-	if (size > PAGE_SIZE)
+	if (size > PAGE_SIZE) {
 		WARN_ONCE(1, "svcrdma: large bc buffer request (size %zu)\n",
 			  size);
+		return -EINVAL;
+	}
 
 	page = alloc_page(RPCRDMA_DEF_GFP);
 	if (!page)
-		return NULL;
+		return -ENOMEM;
 
-	return page_address(page);
+	rqst->rq_buffer = page_address(page);
+	return 0;
 }
 
 static void
-xprt_rdma_bc_free(void *buffer)
+xprt_rdma_bc_free(struct rpc_task *task)
 {
 	/* No-op: ctxt and page have already been freed. */
 }

net/sunrpc/xprtrdma/transport.c

@@ -97,7 +97,7 @@ static struct ctl_table xr_tunables_table[] = {
 		.data		= &xprt_rdma_max_inline_read,
 		.maxlen		= sizeof(unsigned int),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_dointvec_minmax,
 		.extra1		= &min_inline_size,
 		.extra2		= &max_inline_size,
 	},
@@ -106,7 +106,7 @@ static struct ctl_table xr_tunables_table[] = {
 		.data		= &xprt_rdma_max_inline_write,
 		.maxlen		= sizeof(unsigned int),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_dointvec_minmax,
 		.extra1		= &min_inline_size,
 		.extra2		= &max_inline_size,
 	},
@@ -477,115 +477,152 @@ xprt_rdma_connect(struct rpc_xprt *xprt, struct rpc_task *task)
 	}
 }
 
-/*
- * The RDMA allocate/free functions need the task structure as a place
- * to hide the struct rpcrdma_req, which is necessary for the actual send/recv
- * sequence.
+/* Allocate a fixed-size buffer in which to construct and send the
+ * RPC-over-RDMA header for this request.
+ */
+static bool
+rpcrdma_get_rdmabuf(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req,
+		    gfp_t flags)
+{
+	size_t size = RPCRDMA_HDRBUF_SIZE;
+	struct rpcrdma_regbuf *rb;
+
+	if (req->rl_rdmabuf)
+		return true;
+
+	rb = rpcrdma_alloc_regbuf(size, DMA_TO_DEVICE, flags);
+	if (IS_ERR(rb))
+		return false;
+
+	r_xprt->rx_stats.hardway_register_count += size;
+	req->rl_rdmabuf = rb;
+	return true;
+}
+
+static bool
+rpcrdma_get_sendbuf(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req,
+		    size_t size, gfp_t flags)
+{
+	struct rpcrdma_regbuf *rb;
+
+	if (req->rl_sendbuf && rdmab_length(req->rl_sendbuf) >= size)
+		return true;
+
+	rb = rpcrdma_alloc_regbuf(size, DMA_TO_DEVICE, flags);
+	if (IS_ERR(rb))
+		return false;
+
+	rpcrdma_free_regbuf(req->rl_sendbuf);
+	r_xprt->rx_stats.hardway_register_count += size;
+	req->rl_sendbuf = rb;
+	return true;
+}
+
+/* The rq_rcv_buf is used only if a Reply chunk is necessary.
+ * The decision to use a Reply chunk is made later in
+ * rpcrdma_marshal_req. This buffer is registered at that time.
  *
- * The RPC layer allocates both send and receive buffers in the same call
- * (rq_send_buf and rq_rcv_buf are both part of a single contiguous buffer).
- * We may register rq_rcv_buf when using reply chunks.
+ * Otherwise, the associated RPC Reply arrives in a separate
+ * Receive buffer, arbitrarily chosen by the HCA. The buffer
+ * allocated here for the RPC Reply is not utilized in that
+ * case. See rpcrdma_inline_fixup.
+ *
+ * A regbuf is used here to remember the buffer size.
  */
-static void *
-xprt_rdma_allocate(struct rpc_task *task, size_t size)
+static bool
+rpcrdma_get_recvbuf(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req,
+		    size_t size, gfp_t flags)
 {
-	struct rpc_xprt *xprt = task->tk_rqstp->rq_xprt;
-	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(xprt);
 	struct rpcrdma_regbuf *rb;
+
+	if (req->rl_recvbuf && rdmab_length(req->rl_recvbuf) >= size)
+		return true;
+
+	rb = rpcrdma_alloc_regbuf(size, DMA_NONE, flags);
+	if (IS_ERR(rb))
+		return false;
+
+	rpcrdma_free_regbuf(req->rl_recvbuf);
+	r_xprt->rx_stats.hardway_register_count += size;
+	req->rl_recvbuf = rb;
+	return true;
+}
+
+/**
+ * xprt_rdma_allocate - allocate transport resources for an RPC
+ * @task: RPC task
+ *
+ * Return values:
+ *        0:	Success; rq_buffer points to RPC buffer to use
+ *   ENOMEM:	Out of memory, call again later
+ *      EIO:	A permanent error occurred, do not retry
+ *
+ * The RDMA allocate/free functions need the task structure as a place
+ * to hide the struct rpcrdma_req, which is necessary for the actual
+ * send/recv sequence.
+ *
+ * xprt_rdma_allocate provides buffers that are already mapped for
+ * DMA, and a local DMA lkey is provided for each.
+ */
+static int
+xprt_rdma_allocate(struct rpc_task *task)
+{
+	struct rpc_rqst *rqst = task->tk_rqstp;
+	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt);
 	struct rpcrdma_req *req;
-	size_t min_size;
 	gfp_t flags;
 
 	req = rpcrdma_buffer_get(&r_xprt->rx_buf);
 	if (req == NULL)
-		return NULL;
+		return -ENOMEM;
 
 	flags = RPCRDMA_DEF_GFP;
 	if (RPC_IS_SWAPPER(task))
 		flags = __GFP_MEMALLOC | GFP_NOWAIT | __GFP_NOWARN;
 
-	if (req->rl_rdmabuf == NULL)
-		goto out_rdmabuf;
-	if (req->rl_sendbuf == NULL)
-		goto out_sendbuf;
-	if (size > req->rl_sendbuf->rg_size)
-		goto out_sendbuf;
-
-out:
-	dprintk("RPC:       %s: size %zd, request 0x%p\n", __func__, size, req);
-	req->rl_connect_cookie = 0;	/* our reserved value */
-	req->rl_task = task;
-	return req->rl_sendbuf->rg_base;
-
-out_rdmabuf:
-	min_size = RPCRDMA_INLINE_WRITE_THRESHOLD(task->tk_rqstp);
-	rb = rpcrdma_alloc_regbuf(&r_xprt->rx_ia, min_size, flags);
-	if (IS_ERR(rb))
+	if (!rpcrdma_get_rdmabuf(r_xprt, req, flags))
 		goto out_fail;
-	req->rl_rdmabuf = rb;
-
-out_sendbuf:
-	/* XDR encoding and RPC/RDMA marshaling of this request has not
-	 * yet occurred. Thus a lower bound is needed to prevent buffer
-	 * overrun during marshaling.
-	 *
-	 * RPC/RDMA marshaling may choose to send payload bearing ops
-	 * inline, if the result is smaller than the inline threshold.
-	 * The value of the "size" argument accounts for header
-	 * requirements but not for the payload in these cases.
-	 *
-	 * Likewise, allocate enough space to receive a reply up to the
-	 * size of the inline threshold.
-	 *
-	 * It's unlikely that both the send header and the received
-	 * reply will be large, but slush is provided here to allow
-	 * flexibility when marshaling.
-	 */
-	min_size = RPCRDMA_INLINE_READ_THRESHOLD(task->tk_rqstp);
-	min_size += RPCRDMA_INLINE_WRITE_THRESHOLD(task->tk_rqstp);
-	if (size < min_size)
-		size = min_size;
-
-	rb = rpcrdma_alloc_regbuf(&r_xprt->rx_ia, size, flags);
-	if (IS_ERR(rb))
+	if (!rpcrdma_get_sendbuf(r_xprt, req, rqst->rq_callsize, flags))
+		goto out_fail;
+	if (!rpcrdma_get_recvbuf(r_xprt, req, rqst->rq_rcvsize, flags))
 		goto out_fail;
-	rb->rg_owner = req;
 
-	r_xprt->rx_stats.hardway_register_count += size;
-	rpcrdma_free_regbuf(&r_xprt->rx_ia, req->rl_sendbuf);
-	req->rl_sendbuf = rb;
-	goto out;
+	dprintk("RPC: %5u %s: send size = %zd, recv size = %zd, req = %p\n",
+		task->tk_pid, __func__, rqst->rq_callsize,
+		rqst->rq_rcvsize, req);
+
+	req->rl_connect_cookie = 0;	/* our reserved value */
+	rpcrdma_set_xprtdata(rqst, req);
+	rqst->rq_buffer = req->rl_sendbuf->rg_base;
+	rqst->rq_rbuffer = req->rl_recvbuf->rg_base;
+	return 0;
 
 out_fail:
 	rpcrdma_buffer_put(req);
-	return NULL;
+	return -ENOMEM;
 }
 
-/*
- * This function returns all RDMA resources to the pool.
+/**
+ * xprt_rdma_free - release resources allocated by xprt_rdma_allocate
+ * @task: RPC task
+ *
+ * Caller guarantees rqst->rq_buffer is non-NULL.
  */
 static void
-xprt_rdma_free(void *buffer)
+xprt_rdma_free(struct rpc_task *task)
 {
-	struct rpcrdma_req *req;
-	struct rpcrdma_xprt *r_xprt;
-	struct rpcrdma_regbuf *rb;
-
-	if (buffer == NULL)
-		return;
+	struct rpc_rqst *rqst = task->tk_rqstp;
+	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt);
+	struct rpcrdma_req *req = rpcr_to_rdmar(rqst);
+	struct rpcrdma_ia *ia = &r_xprt->rx_ia;
 
-	rb = container_of(buffer, struct rpcrdma_regbuf, rg_base[0]);
-	req = rb->rg_owner;
 	if (req->rl_backchannel)
 		return;
 
-	r_xprt = container_of(req->rl_buffer, struct rpcrdma_xprt, rx_buf);
-
 	dprintk("RPC:       %s: called on 0x%p\n", __func__, req->rl_reply);
 
-	r_xprt->rx_ia.ri_ops->ro_unmap_safe(r_xprt, req,
-					    !RPC_IS_ASYNC(req->rl_task));
-
+	ia->ri_ops->ro_unmap_safe(r_xprt, req, !RPC_IS_ASYNC(task));
+	rpcrdma_unmap_sges(ia, req);
 	rpcrdma_buffer_put(req);
 }
 
@@ -685,10 +722,11 @@ void xprt_rdma_print_stats(struct rpc_xprt *xprt, struct seq_file *seq)
 		   r_xprt->rx_stats.failed_marshal_count,
 		   r_xprt->rx_stats.bad_reply_count,
 		   r_xprt->rx_stats.nomsg_call_count);
-	seq_printf(seq, "%lu %lu %lu\n",
+	seq_printf(seq, "%lu %lu %lu %lu\n",
 		   r_xprt->rx_stats.mrs_recovered,
 		   r_xprt->rx_stats.mrs_orphaned,
-		   r_xprt->rx_stats.mrs_allocated);
+		   r_xprt->rx_stats.mrs_allocated,
+		   r_xprt->rx_stats.local_inv_needed);
 }
 
 static int

net/sunrpc/xprtrdma/verbs.c

@@ -129,15 +129,6 @@ rpcrdma_wc_send(struct ib_cq *cq, struct ib_wc *wc)
 		       wc->status, wc->vendor_err);
 }
 
-static void
-rpcrdma_receive_worker(struct work_struct *work)
-{
-	struct rpcrdma_rep *rep =
-			container_of(work, struct rpcrdma_rep, rr_work);
-
-	rpcrdma_reply_handler(rep);
-}
-
 /* Perform basic sanity checking to avoid using garbage
  * to update the credit grant value.
  */
@@ -161,13 +152,13 @@ rpcrdma_update_granted_credits(struct rpcrdma_rep *rep)
 }
 
 /**
- * rpcrdma_receive_wc - Invoked by RDMA provider for each polled Receive WC
+ * rpcrdma_wc_receive - Invoked by RDMA provider for each polled Receive WC
  * @cq:	completion queue (ignored)
  * @wc:	completed WR
  *
  */
 static void
-rpcrdma_receive_wc(struct ib_cq *cq, struct ib_wc *wc)
+rpcrdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc)
 {
 	struct ib_cqe *cqe = wc->wr_cqe;
 	struct rpcrdma_rep *rep = container_of(cqe, struct rpcrdma_rep,
@@ -185,6 +176,9 @@ rpcrdma_receive_wc(struct ib_cq *cq, struct ib_wc *wc)
 		__func__, rep, wc->byte_len);
 
 	rep->rr_len = wc->byte_len;
+	rep->rr_wc_flags = wc->wc_flags;
+	rep->rr_inv_rkey = wc->ex.invalidate_rkey;
+
 	ib_dma_sync_single_for_cpu(rep->rr_device,
 				   rdmab_addr(rep->rr_rdmabuf),
 				   rep->rr_len, DMA_FROM_DEVICE);
@@ -204,6 +198,36 @@ out_fail:
 	goto out_schedule;
 }
 
+static void
+rpcrdma_update_connect_private(struct rpcrdma_xprt *r_xprt,
+			       struct rdma_conn_param *param)
+{
+	struct rpcrdma_create_data_internal *cdata = &r_xprt->rx_data;
+	const struct rpcrdma_connect_private *pmsg = param->private_data;
+	unsigned int rsize, wsize;
+
+	/* Default settings for RPC-over-RDMA Version One */
+	r_xprt->rx_ia.ri_reminv_expected = false;
+	rsize = RPCRDMA_V1_DEF_INLINE_SIZE;
+	wsize = RPCRDMA_V1_DEF_INLINE_SIZE;
+
+	if (pmsg &&
+	    pmsg->cp_magic == rpcrdma_cmp_magic &&
+	    pmsg->cp_version == RPCRDMA_CMP_VERSION) {
+		r_xprt->rx_ia.ri_reminv_expected = true;
+		rsize = rpcrdma_decode_buffer_size(pmsg->cp_send_size);
+		wsize = rpcrdma_decode_buffer_size(pmsg->cp_recv_size);
+	}
+
+	if (rsize < cdata->inline_rsize)
+		cdata->inline_rsize = rsize;
+	if (wsize < cdata->inline_wsize)
+		cdata->inline_wsize = wsize;
+	pr_info("rpcrdma: max send %u, max recv %u\n",
+		cdata->inline_wsize, cdata->inline_rsize);
+	rpcrdma_set_max_header_sizes(r_xprt);
+}
+
 static int
 rpcrdma_conn_upcall(struct rdma_cm_id *id, struct rdma_cm_event *event)
 {
@@ -244,6 +268,7 @@ rpcrdma_conn_upcall(struct rdma_cm_id *id, struct rdma_cm_event *event)
 			" (%d initiator)\n",
 			__func__, attr->max_dest_rd_atomic,
 			attr->max_rd_atomic);
+		rpcrdma_update_connect_private(xprt, &event->param.conn);
 		goto connected;
 	case RDMA_CM_EVENT_CONNECT_ERROR:
 		connstate = -ENOTCONN;
@@ -454,11 +479,12 @@ int
 rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia,
 				struct rpcrdma_create_data_internal *cdata)
 {
+	struct rpcrdma_connect_private *pmsg = &ep->rep_cm_private;
 	struct ib_cq *sendcq, *recvcq;
 	unsigned int max_qp_wr;
 	int rc;
 
-	if (ia->ri_device->attrs.max_sge < RPCRDMA_MAX_IOVS) {
+	if (ia->ri_device->attrs.max_sge < RPCRDMA_MAX_SEND_SGES) {
 		dprintk("RPC:       %s: insufficient sge's available\n",
 			__func__);
 		return -ENOMEM;
@@ -487,7 +513,7 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia,
 	ep->rep_attr.cap.max_recv_wr = cdata->max_requests;
 	ep->rep_attr.cap.max_recv_wr += RPCRDMA_BACKWARD_WRS;
 	ep->rep_attr.cap.max_recv_wr += 1;	/* drain cqe */
-	ep->rep_attr.cap.max_send_sge = RPCRDMA_MAX_IOVS;
+	ep->rep_attr.cap.max_send_sge = RPCRDMA_MAX_SEND_SGES;
 	ep->rep_attr.cap.max_recv_sge = 1;
 	ep->rep_attr.cap.max_inline_data = 0;
 	ep->rep_attr.sq_sig_type = IB_SIGNAL_REQ_WR;
@@ -536,9 +562,14 @@ rpcrdma_ep_create(struct rpcrdma_ep *ep, struct rpcrdma_ia *ia,
 	/* Initialize cma parameters */
 	memset(&ep->rep_remote_cma, 0, sizeof(ep->rep_remote_cma));
 
-	/* RPC/RDMA does not use private data */
-	ep->rep_remote_cma.private_data = NULL;
-	ep->rep_remote_cma.private_data_len = 0;
+	/* Prepare RDMA-CM private message */
+	pmsg->cp_magic = rpcrdma_cmp_magic;
+	pmsg->cp_version = RPCRDMA_CMP_VERSION;
+	pmsg->cp_flags |= ia->ri_ops->ro_send_w_inv_ok;
+	pmsg->cp_send_size = rpcrdma_encode_buffer_size(cdata->inline_wsize);
+	pmsg->cp_recv_size = rpcrdma_encode_buffer_size(cdata->inline_rsize);
+	ep->rep_remote_cma.private_data = pmsg;
+	ep->rep_remote_cma.private_data_len = sizeof(*pmsg);
 
 	/* Client offers RDMA Read but does not initiate */
 	ep->rep_remote_cma.initiator_depth = 0;
@@ -849,6 +880,10 @@ rpcrdma_create_req(struct rpcrdma_xprt *r_xprt)
 	req->rl_cqe.done = rpcrdma_wc_send;
 	req->rl_buffer = &r_xprt->rx_buf;
 	INIT_LIST_HEAD(&req->rl_registered);
+	req->rl_send_wr.next = NULL;
+	req->rl_send_wr.wr_cqe = &req->rl_cqe;
+	req->rl_send_wr.sg_list = req->rl_send_sge;
+	req->rl_send_wr.opcode = IB_WR_SEND;
 	return req;
 }
 
@@ -865,17 +900,21 @@ rpcrdma_create_rep(struct rpcrdma_xprt *r_xprt)
 	if (rep == NULL)
 		goto out;
 
-	rep->rr_rdmabuf = rpcrdma_alloc_regbuf(ia, cdata->inline_rsize,
-					       GFP_KERNEL);
+	rep->rr_rdmabuf = rpcrdma_alloc_regbuf(cdata->inline_rsize,
+					       DMA_FROM_DEVICE, GFP_KERNEL);
 	if (IS_ERR(rep->rr_rdmabuf)) {
 		rc = PTR_ERR(rep->rr_rdmabuf);
 		goto out_free;
 	}
 
 	rep->rr_device = ia->ri_device;
-	rep->rr_cqe.done = rpcrdma_receive_wc;
+	rep->rr_cqe.done = rpcrdma_wc_receive;
 	rep->rr_rxprt = r_xprt;
-	INIT_WORK(&rep->rr_work, rpcrdma_receive_worker);
+	INIT_WORK(&rep->rr_work, rpcrdma_reply_handler);
+	rep->rr_recv_wr.next = NULL;
+	rep->rr_recv_wr.wr_cqe = &rep->rr_cqe;
+	rep->rr_recv_wr.sg_list = &rep->rr_rdmabuf->rg_iov;
+	rep->rr_recv_wr.num_sge = 1;
 	return rep;
 
 out_free:
@@ -966,17 +1005,18 @@ rpcrdma_buffer_get_rep_locked(struct rpcrdma_buffer *buf)
 }
 
 static void
-rpcrdma_destroy_rep(struct rpcrdma_ia *ia, struct rpcrdma_rep *rep)
+rpcrdma_destroy_rep(struct rpcrdma_rep *rep)
 {
-	rpcrdma_free_regbuf(ia, rep->rr_rdmabuf);
+	rpcrdma_free_regbuf(rep->rr_rdmabuf);
 	kfree(rep);
 }
 
 void
-rpcrdma_destroy_req(struct rpcrdma_ia *ia, struct rpcrdma_req *req)
+rpcrdma_destroy_req(struct rpcrdma_req *req)
 {
-	rpcrdma_free_regbuf(ia, req->rl_sendbuf);
-	rpcrdma_free_regbuf(ia, req->rl_rdmabuf);
+	rpcrdma_free_regbuf(req->rl_recvbuf);
+	rpcrdma_free_regbuf(req->rl_sendbuf);
+	rpcrdma_free_regbuf(req->rl_rdmabuf);
 	kfree(req);
 }
 
@@ -1009,15 +1049,13 @@ rpcrdma_destroy_mrs(struct rpcrdma_buffer *buf)
 void
 rpcrdma_buffer_destroy(struct rpcrdma_buffer *buf)
 {
-	struct rpcrdma_ia *ia = rdmab_to_ia(buf);
-
 	cancel_delayed_work_sync(&buf->rb_recovery_worker);
 
 	while (!list_empty(&buf->rb_recv_bufs)) {
 		struct rpcrdma_rep *rep;
 
 		rep = rpcrdma_buffer_get_rep_locked(buf);
-		rpcrdma_destroy_rep(ia, rep);
+		rpcrdma_destroy_rep(rep);
 	}
 	buf->rb_send_count = 0;
 
@@ -1030,7 +1068,7 @@ rpcrdma_buffer_destroy(struct rpcrdma_buffer *buf)
 		list_del(&req->rl_all);
 
 		spin_unlock(&buf->rb_reqslock);
-		rpcrdma_destroy_req(ia, req);
+		rpcrdma_destroy_req(req);
 		spin_lock(&buf->rb_reqslock);
 	}
 	spin_unlock(&buf->rb_reqslock);
@@ -1129,7 +1167,7 @@ rpcrdma_buffer_put(struct rpcrdma_req *req)
 	struct rpcrdma_buffer *buffers = req->rl_buffer;
 	struct rpcrdma_rep *rep = req->rl_reply;
 
-	req->rl_niovs = 0;
+	req->rl_send_wr.num_sge = 0;
 	req->rl_reply = NULL;
 
 	spin_lock(&buffers->rb_lock);
@@ -1171,70 +1209,81 @@ rpcrdma_recv_buffer_put(struct rpcrdma_rep *rep)
 	spin_unlock(&buffers->rb_lock);
 }
 
-/*
- * Wrappers for internal-use kmalloc memory registration, used by buffer code.
- */
-
 /**
- * rpcrdma_alloc_regbuf - kmalloc and register memory for SEND/RECV buffers
- * @ia: controlling rpcrdma_ia
+ * rpcrdma_alloc_regbuf - allocate and DMA-map memory for SEND/RECV buffers
  * @size: size of buffer to be allocated, in bytes
+ * @direction: direction of data movement
  * @flags: GFP flags
  *
- * Returns pointer to private header of an area of internally
- * registered memory, or an ERR_PTR. The registered buffer follows
- * the end of the private header.
+ * Returns an ERR_PTR, or a pointer to a regbuf, a buffer that
+ * can be persistently DMA-mapped for I/O.
  *
  * xprtrdma uses a regbuf for posting an outgoing RDMA SEND, or for
- * receiving the payload of RDMA RECV operations. regbufs are not
- * used for RDMA READ/WRITE operations, thus are registered only for
- * LOCAL access.
+ * receiving the payload of RDMA RECV operations. During Long Calls
+ * or Replies they may be registered externally via ro_map.
  */
 struct rpcrdma_regbuf *
-rpcrdma_alloc_regbuf(struct rpcrdma_ia *ia, size_t size, gfp_t flags)
+rpcrdma_alloc_regbuf(size_t size, enum dma_data_direction direction,
+		     gfp_t flags)
 {
 	struct rpcrdma_regbuf *rb;
-	struct ib_sge *iov;
 
 	rb = kmalloc(sizeof(*rb) + size, flags);
 	if (rb == NULL)
-		goto out;
+		return ERR_PTR(-ENOMEM);
 
-	iov = &rb->rg_iov;
-	iov->addr = ib_dma_map_single(ia->ri_device,
-				      (void *)rb->rg_base, size,
-				      DMA_BIDIRECTIONAL);
-	if (ib_dma_mapping_error(ia->ri_device, iov->addr))
-		goto out_free;
+	rb->rg_device = NULL;
+	rb->rg_direction = direction;
+	rb->rg_iov.length = size;
 
-	iov->length = size;
-	iov->lkey = ia->ri_pd->local_dma_lkey;
-	rb->rg_size = size;
-	rb->rg_owner = NULL;
 	return rb;
+}
 
-out_free:
-	kfree(rb);
-out:
-	return ERR_PTR(-ENOMEM);
+/**
+ * __rpcrdma_map_regbuf - DMA-map a regbuf
+ * @ia: controlling rpcrdma_ia
+ * @rb: regbuf to be mapped
+ */
+bool
+__rpcrdma_dma_map_regbuf(struct rpcrdma_ia *ia, struct rpcrdma_regbuf *rb)
+{
+	if (rb->rg_direction == DMA_NONE)
+		return false;
+
+	rb->rg_iov.addr = ib_dma_map_single(ia->ri_device,
+					    (void *)rb->rg_base,
+					    rdmab_length(rb),
+					    rb->rg_direction);
+	if (ib_dma_mapping_error(ia->ri_device, rdmab_addr(rb)))
+		return false;
+
+	rb->rg_device = ia->ri_device;
+	rb->rg_iov.lkey = ia->ri_pd->local_dma_lkey;
+	return true;
+}
+
+static void
+rpcrdma_dma_unmap_regbuf(struct rpcrdma_regbuf *rb)
+{
+	if (!rpcrdma_regbuf_is_mapped(rb))
+		return;
+
+	ib_dma_unmap_single(rb->rg_device, rdmab_addr(rb),
+			    rdmab_length(rb), rb->rg_direction);
+	rb->rg_device = NULL;
 }
 
 /**
  * rpcrdma_free_regbuf - deregister and free registered buffer
- * @ia: controlling rpcrdma_ia
  * @rb: regbuf to be deregistered and freed
  */
 void
-rpcrdma_free_regbuf(struct rpcrdma_ia *ia, struct rpcrdma_regbuf *rb)
+rpcrdma_free_regbuf(struct rpcrdma_regbuf *rb)
 {
-	struct ib_sge *iov;
-
 	if (!rb)
 		return;
 
-	iov = &rb->rg_iov;
-	ib_dma_unmap_single(ia->ri_device,
-			    iov->addr, iov->length, DMA_BIDIRECTIONAL);
+	rpcrdma_dma_unmap_regbuf(rb);
 	kfree(rb);
 }
 
@@ -1248,39 +1297,28 @@ rpcrdma_ep_post(struct rpcrdma_ia *ia,
 		struct rpcrdma_ep *ep,
 		struct rpcrdma_req *req)
 {
-	struct ib_device *device = ia->ri_device;
-	struct ib_send_wr send_wr, *send_wr_fail;
-	struct rpcrdma_rep *rep = req->rl_reply;
-	struct ib_sge *iov = req->rl_send_iov;
-	int i, rc;
+	struct ib_send_wr *send_wr = &req->rl_send_wr;
+	struct ib_send_wr *send_wr_fail;
+	int rc;
 
-	if (rep) {
-		rc = rpcrdma_ep_post_recv(ia, ep, rep);
+	if (req->rl_reply) {
+		rc = rpcrdma_ep_post_recv(ia, req->rl_reply);
 		if (rc)
 			return rc;
 		req->rl_reply = NULL;
 	}
 
-	send_wr.next = NULL;
-	send_wr.wr_cqe = &req->rl_cqe;
-	send_wr.sg_list = iov;
-	send_wr.num_sge = req->rl_niovs;
-	send_wr.opcode = IB_WR_SEND;
-
-	for (i = 0; i < send_wr.num_sge; i++)
-		ib_dma_sync_single_for_device(device, iov[i].addr,
-					      iov[i].length, DMA_TO_DEVICE);
 	dprintk("RPC:       %s: posting %d s/g entries\n",
-		__func__, send_wr.num_sge);
+		__func__, send_wr->num_sge);
 
 	if (DECR_CQCOUNT(ep) > 0)
-		send_wr.send_flags = 0;
+		send_wr->send_flags = 0;
 	else { /* Provider must take a send completion every now and then */
 		INIT_CQCOUNT(ep);
-		send_wr.send_flags = IB_SEND_SIGNALED;
+		send_wr->send_flags = IB_SEND_SIGNALED;
 	}
 
-	rc = ib_post_send(ia->ri_id->qp, &send_wr, &send_wr_fail);
+	rc = ib_post_send(ia->ri_id->qp, send_wr, &send_wr_fail);
 	if (rc)
 		goto out_postsend_err;
 	return 0;
@@ -1290,32 +1328,24 @@ out_postsend_err:
 	return -ENOTCONN;
 }
 
-/*
- * (Re)post a receive buffer.
- */
 int
 rpcrdma_ep_post_recv(struct rpcrdma_ia *ia,
-		     struct rpcrdma_ep *ep,
 		     struct rpcrdma_rep *rep)
 {
-	struct ib_recv_wr recv_wr, *recv_wr_fail;
+	struct ib_recv_wr *recv_wr_fail;
 	int rc;
 
-	recv_wr.next = NULL;
-	recv_wr.wr_cqe = &rep->rr_cqe;
-	recv_wr.sg_list = &rep->rr_rdmabuf->rg_iov;
-	recv_wr.num_sge = 1;
-
-	ib_dma_sync_single_for_cpu(ia->ri_device,
-				   rdmab_addr(rep->rr_rdmabuf),
-				   rdmab_length(rep->rr_rdmabuf),
-				   DMA_BIDIRECTIONAL);
-
-	rc = ib_post_recv(ia->ri_id->qp, &recv_wr, &recv_wr_fail);
+	if (!rpcrdma_dma_map_regbuf(ia, rep->rr_rdmabuf))
+		goto out_map;
+	rc = ib_post_recv(ia->ri_id->qp, &rep->rr_recv_wr, &recv_wr_fail);
 	if (rc)
 		goto out_postrecv;
 	return 0;
 
+out_map:
+	pr_err("rpcrdma: failed to DMA map the Receive buffer\n");
+	return -EIO;
+
 out_postrecv:
 	pr_err("rpcrdma: ib_post_recv returned %i\n", rc);
 	return -ENOTCONN;
@@ -1333,7 +1363,6 @@ rpcrdma_ep_post_extra_recv(struct rpcrdma_xprt *r_xprt, unsigned int count)
 {
 	struct rpcrdma_buffer *buffers = &r_xprt->rx_buf;
 	struct rpcrdma_ia *ia = &r_xprt->rx_ia;
-	struct rpcrdma_ep *ep = &r_xprt->rx_ep;
 	struct rpcrdma_rep *rep;
 	int rc;
 
@@ -1344,7 +1373,7 @@ rpcrdma_ep_post_extra_recv(struct rpcrdma_xprt *r_xprt, unsigned int count)
 		rep = rpcrdma_buffer_get_rep_locked(buffers);
 		spin_unlock(&buffers->rb_lock);
 
-		rc = rpcrdma_ep_post_recv(ia, ep, rep);
+		rc = rpcrdma_ep_post_recv(ia, rep);
 		if (rc)
 			goto out_rc;
 	}

net/sunrpc/xprtrdma/xprt_rdma.h

@@ -70,9 +70,11 @@ struct rpcrdma_ia {
 	struct ib_pd		*ri_pd;
 	struct completion	ri_done;
 	int			ri_async_rc;
+	unsigned int		ri_max_segs;
 	unsigned int		ri_max_frmr_depth;
 	unsigned int		ri_max_inline_write;
 	unsigned int		ri_max_inline_read;
+	bool			ri_reminv_expected;
 	struct ib_qp_attr	ri_qp_attr;
 	struct ib_qp_init_attr	ri_qp_init_attr;
 };
@@ -87,6 +89,7 @@ struct rpcrdma_ep {
 	int			rep_connected;
 	struct ib_qp_init_attr	rep_attr;
 	wait_queue_head_t 	rep_connect_wait;
+	struct rpcrdma_connect_private	rep_cm_private;
 	struct rdma_conn_param	rep_remote_cma;
 	struct sockaddr_storage	rep_remote_addr;
 	struct delayed_work	rep_connect_worker;
@@ -112,9 +115,9 @@ struct rpcrdma_ep {
  */
 
 struct rpcrdma_regbuf {
-	size_t			rg_size;
-	struct rpcrdma_req	*rg_owner;
 	struct ib_sge		rg_iov;
+	struct ib_device	*rg_device;
+	enum dma_data_direction	rg_direction;
 	__be32			rg_base[0] __attribute__ ((aligned(256)));
 };
 
@@ -162,7 +165,10 @@ rdmab_to_msg(struct rpcrdma_regbuf *rb)
  * The smallest inline threshold is 1024 bytes, ensuring that
  * at least 750 bytes are available for RPC messages.
  */
-#define RPCRDMA_MAX_HDR_SEGS	(8)
+enum {
+	RPCRDMA_MAX_HDR_SEGS = 8,
+	RPCRDMA_HDRBUF_SIZE = 256,
+};
 
 /*
  * struct rpcrdma_rep -- this structure encapsulates state required to recv
@@ -182,10 +188,13 @@ rdmab_to_msg(struct rpcrdma_regbuf *rb)
 struct rpcrdma_rep {
 	struct ib_cqe		rr_cqe;
 	unsigned int		rr_len;
+	int			rr_wc_flags;
+	u32			rr_inv_rkey;
 	struct ib_device	*rr_device;
 	struct rpcrdma_xprt	*rr_rxprt;
 	struct work_struct	rr_work;
 	struct list_head	rr_list;
+	struct ib_recv_wr	rr_recv_wr;
 	struct rpcrdma_regbuf	*rr_rdmabuf;
 };
 
@@ -276,19 +285,30 @@ struct rpcrdma_mr_seg {		/* chunk descriptors */
 	char		*mr_offset;	/* kva if no page, else offset */
 };
 
-#define RPCRDMA_MAX_IOVS	(2)
+/* Reserve enough Send SGEs to send a maximum size inline request:
+ * - RPC-over-RDMA header
+ * - xdr_buf head iovec
+ * - RPCRDMA_MAX_INLINE bytes, possibly unaligned, in pages
+ * - xdr_buf tail iovec
+ */
+enum {
+	RPCRDMA_MAX_SEND_PAGES = PAGE_SIZE + RPCRDMA_MAX_INLINE - 1,
+	RPCRDMA_MAX_PAGE_SGES = (RPCRDMA_MAX_SEND_PAGES >> PAGE_SHIFT) + 1,
+	RPCRDMA_MAX_SEND_SGES = 1 + 1 + RPCRDMA_MAX_PAGE_SGES + 1,
+};
 
 struct rpcrdma_buffer;
 struct rpcrdma_req {
 	struct list_head	rl_free;
-	unsigned int		rl_niovs;
+	unsigned int		rl_mapped_sges;
 	unsigned int		rl_connect_cookie;
-	struct rpc_task		*rl_task;
 	struct rpcrdma_buffer	*rl_buffer;
-	struct rpcrdma_rep	*rl_reply;/* holder for reply buffer */
-	struct ib_sge		rl_send_iov[RPCRDMA_MAX_IOVS];
-	struct rpcrdma_regbuf	*rl_rdmabuf;
-	struct rpcrdma_regbuf	*rl_sendbuf;
+	struct rpcrdma_rep	*rl_reply;
+	struct ib_send_wr	rl_send_wr;
+	struct ib_sge		rl_send_sge[RPCRDMA_MAX_SEND_SGES];
+	struct rpcrdma_regbuf	*rl_rdmabuf;	/* xprt header */
+	struct rpcrdma_regbuf	*rl_sendbuf;	/* rq_snd_buf */
+	struct rpcrdma_regbuf	*rl_recvbuf;	/* rq_rcv_buf */
 
 	struct ib_cqe		rl_cqe;
 	struct list_head	rl_all;
@@ -298,14 +318,16 @@ struct rpcrdma_req {
 	struct rpcrdma_mr_seg	rl_segments[RPCRDMA_MAX_SEGS];
 };
 
+static inline void
+rpcrdma_set_xprtdata(struct rpc_rqst *rqst, struct rpcrdma_req *req)
+{
+	rqst->rq_xprtdata = req;
+}
+
 static inline struct rpcrdma_req *
 rpcr_to_rdmar(struct rpc_rqst *rqst)
 {
-	void *buffer = rqst->rq_buffer;
-	struct rpcrdma_regbuf *rb;
-
-	rb = container_of(buffer, struct rpcrdma_regbuf, rg_base);
-	return rb->rg_owner;
+	return rqst->rq_xprtdata;
 }
 
 /*
@@ -356,15 +378,6 @@ struct rpcrdma_create_data_internal {
 	unsigned int	padding;	/* non-rdma write header padding */
 };
 
-#define RPCRDMA_INLINE_READ_THRESHOLD(rq) \
-	(rpcx_to_rdmad(rq->rq_xprt).inline_rsize)
-
-#define RPCRDMA_INLINE_WRITE_THRESHOLD(rq)\
-	(rpcx_to_rdmad(rq->rq_xprt).inline_wsize)
-
-#define RPCRDMA_INLINE_PAD_VALUE(rq)\
-	rpcx_to_rdmad(rq->rq_xprt).padding
-
 /*
  * Statistics for RPCRDMA
  */
@@ -386,6 +399,7 @@ struct rpcrdma_stats {
 	unsigned long		mrs_recovered;
 	unsigned long		mrs_orphaned;
 	unsigned long		mrs_allocated;
+	unsigned long		local_inv_needed;
 };
 
 /*
@@ -409,6 +423,7 @@ struct rpcrdma_memreg_ops {
 				      struct rpcrdma_mw *);
 	void		(*ro_release_mr)(struct rpcrdma_mw *);
 	const char	*ro_displayname;
+	const int	ro_send_w_inv_ok;
 };
 
 extern const struct rpcrdma_memreg_ops rpcrdma_fmr_memreg_ops;
@@ -461,15 +476,14 @@ void rpcrdma_ep_disconnect(struct rpcrdma_ep *, struct rpcrdma_ia *);
 
 int rpcrdma_ep_post(struct rpcrdma_ia *, struct rpcrdma_ep *,
 				struct rpcrdma_req *);
-int rpcrdma_ep_post_recv(struct rpcrdma_ia *, struct rpcrdma_ep *,
-				struct rpcrdma_rep *);
+int rpcrdma_ep_post_recv(struct rpcrdma_ia *, struct rpcrdma_rep *);
 
 /*
  * Buffer calls - xprtrdma/verbs.c
  */
 struct rpcrdma_req *rpcrdma_create_req(struct rpcrdma_xprt *);
 struct rpcrdma_rep *rpcrdma_create_rep(struct rpcrdma_xprt *);
-void rpcrdma_destroy_req(struct rpcrdma_ia *, struct rpcrdma_req *);
+void rpcrdma_destroy_req(struct rpcrdma_req *);
 int rpcrdma_buffer_create(struct rpcrdma_xprt *);
 void rpcrdma_buffer_destroy(struct rpcrdma_buffer *);
 
@@ -482,10 +496,24 @@ void rpcrdma_recv_buffer_put(struct rpcrdma_rep *);
 
 void rpcrdma_defer_mr_recovery(struct rpcrdma_mw *);
 
-struct rpcrdma_regbuf *rpcrdma_alloc_regbuf(struct rpcrdma_ia *,
-					    size_t, gfp_t);
-void rpcrdma_free_regbuf(struct rpcrdma_ia *,
-			 struct rpcrdma_regbuf *);
+struct rpcrdma_regbuf *rpcrdma_alloc_regbuf(size_t, enum dma_data_direction,
+					    gfp_t);
+bool __rpcrdma_dma_map_regbuf(struct rpcrdma_ia *, struct rpcrdma_regbuf *);
+void rpcrdma_free_regbuf(struct rpcrdma_regbuf *);
+
+static inline bool
+rpcrdma_regbuf_is_mapped(struct rpcrdma_regbuf *rb)
+{
+	return rb->rg_device != NULL;
+}
+
+static inline bool
+rpcrdma_dma_map_regbuf(struct rpcrdma_ia *ia, struct rpcrdma_regbuf *rb)
+{
+	if (likely(rpcrdma_regbuf_is_mapped(rb)))
+		return true;
+	return __rpcrdma_dma_map_regbuf(ia, rb);
+}
 
 int rpcrdma_ep_post_extra_recv(struct rpcrdma_xprt *, unsigned int);
 
@@ -507,15 +535,25 @@ rpcrdma_data_dir(bool writing)
  */
 void rpcrdma_connect_worker(struct work_struct *);
 void rpcrdma_conn_func(struct rpcrdma_ep *);
-void rpcrdma_reply_handler(struct rpcrdma_rep *);
+void rpcrdma_reply_handler(struct work_struct *);
 
 /*
  * RPC/RDMA protocol calls - xprtrdma/rpc_rdma.c
  */
+
+enum rpcrdma_chunktype {
+	rpcrdma_noch = 0,
+	rpcrdma_readch,
+	rpcrdma_areadch,
+	rpcrdma_writech,
+	rpcrdma_replych
+};
+
+bool rpcrdma_prepare_send_sges(struct rpcrdma_ia *, struct rpcrdma_req *,
+			       u32, struct xdr_buf *, enum rpcrdma_chunktype);
+void rpcrdma_unmap_sges(struct rpcrdma_ia *, struct rpcrdma_req *);
 int rpcrdma_marshal_req(struct rpc_rqst *);
-void rpcrdma_set_max_header_sizes(struct rpcrdma_ia *,
-				  struct rpcrdma_create_data_internal *,
-				  unsigned int);
+void rpcrdma_set_max_header_sizes(struct rpcrdma_xprt *);
 
 /* RPC/RDMA module init - xprtrdma/transport.c
  */

net/sunrpc/xprtsock.c

@@ -473,7 +473,16 @@ static int xs_nospace(struct rpc_task *task)
 	spin_unlock_bh(&xprt->transport_lock);
 
 	/* Race breaker in case memory is freed before above code is called */
-	sk->sk_write_space(sk);
+	if (ret == -EAGAIN) {
+		struct socket_wq *wq;
+
+		rcu_read_lock();
+		wq = rcu_dereference(sk->sk_wq);
+		set_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags);
+		rcu_read_unlock();
+
+		sk->sk_write_space(sk);
+	}
 	return ret;
 }
 
@@ -2533,35 +2542,38 @@ static void xs_tcp_print_stats(struct rpc_xprt *xprt, struct seq_file *seq)
  * we allocate pages instead doing a kmalloc like rpc_malloc is because we want
  * to use the server side send routines.
  */
-static void *bc_malloc(struct rpc_task *task, size_t size)
+static int bc_malloc(struct rpc_task *task)
 {
+	struct rpc_rqst *rqst = task->tk_rqstp;
+	size_t size = rqst->rq_callsize;
 	struct page *page;
 	struct rpc_buffer *buf;
 
-	WARN_ON_ONCE(size > PAGE_SIZE - sizeof(struct rpc_buffer));
-	if (size > PAGE_SIZE - sizeof(struct rpc_buffer))
-		return NULL;
+	if (size > PAGE_SIZE - sizeof(struct rpc_buffer)) {
+		WARN_ONCE(1, "xprtsock: large bc buffer request (size %zu)\n",
+			  size);
+		return -EINVAL;
+	}
 
 	page = alloc_page(GFP_KERNEL);
 	if (!page)
-		return NULL;
+		return -ENOMEM;
 
 	buf = page_address(page);
 	buf->len = PAGE_SIZE;
 
-	return buf->data;
+	rqst->rq_buffer = buf->data;
+	return 0;
 }
 
 /*
  * Free the space allocated in the bc_alloc routine
  */
-static void bc_free(void *buffer)
+static void bc_free(struct rpc_task *task)
 {
+	void *buffer = task->tk_rqstp->rq_buffer;
 	struct rpc_buffer *buf;
 
-	if (!buffer)
-		return;
-
 	buf = container_of(buffer, struct rpc_buffer, data);
 	free_page((unsigned long)buf);
 }