[PATCH] NFSv4: Honour the 'opened' parameter in the atomic_open() filesystem method

[PATCH] NFSv4: Honour the 'opened' parameter in the atomic_open() filesystem method

[Trond Myklebust]

Determine if we've created a new file by examining the directory change
attribute and/or the O_EXCL flag.

This fixes a regression when doing a non-exclusive create of a new file.
If the FILE_CREATED flag is not set, the atomic_open() command will
perform full file access permissions checks instead of just checking
for MAY_OPEN.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/dir.c            |  2 +-
 fs/nfs/nfs4file.c       |  3 ++-
 fs/nfs/nfs4proc.c       | 26 +++++++++++++++++++-------
 include/linux/nfs_xdr.h |  3 ++-
 4 files changed, 24 insertions(+), 10 deletions(-)

diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c
index 854a8f0..02b0df7 100644
--- a/fs/nfs/dir.c
+++ b/fs/nfs/dir.c
@@ -1458,7 +1458,7 @@ int nfs_atomic_open(struct inode *dir, struct dentry *dentry,
 
 	trace_nfs_atomic_open_enter(dir, ctx, open_flags);
 	nfs_block_sillyrename(dentry->d_parent);
-	inode = NFS_PROTO(dir)->open_context(dir, ctx, open_flags, &attr);
+	inode = NFS_PROTO(dir)->open_context(dir, ctx, open_flags, &attr, opened);
 	nfs_unblock_sillyrename(dentry->d_parent);
 	if (IS_ERR(inode)) {
 		err = PTR_ERR(inode);
diff --git a/fs/nfs/nfs4file.c b/fs/nfs/nfs4file.c
index e5b804d..77efaf1 100644
--- a/fs/nfs/nfs4file.c
+++ b/fs/nfs/nfs4file.c
@@ -19,6 +19,7 @@ nfs4_file_open(struct inode *inode, struct file *filp)
 	struct inode *dir;
 	unsigned openflags = filp->f_flags;
 	struct iattr attr;
+	int opened = 0;
 	int err;
 
 	/*
@@ -55,7 +56,7 @@ nfs4_file_open(struct inode *inode, struct file *filp)
 		nfs_wb_all(inode);
 	}
 
-	inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, &attr);
+	inode = NFS_PROTO(dir)->open_context(dir, ctx, openflags, &attr, &opened);
 	if (IS_ERR(inode)) {
 		err = PTR_ERR(inode);
 		switch (err) {
diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 989bb9d..488ef9b 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -912,6 +912,7 @@ struct nfs4_opendata {
 	struct iattr attrs;
 	unsigned long timestamp;
 	unsigned int rpc_done : 1;
+	unsigned int file_created : 1;
 	unsigned int is_recover : 1;
 	int rpc_status;
 	int cancelled;
@@ -1946,8 +1947,13 @@ static int _nfs4_proc_open(struct nfs4_opendata *data)
 
 	nfs_fattr_map_and_free_names(server, &data->f_attr);
 
-	if (o_arg->open_flags & O_CREAT)
+	if (o_arg->open_flags & O_CREAT) {
 		update_changeattr(dir, &o_res->cinfo);
+		if (o_arg->open_flags & O_EXCL)
+			data->file_created = 1;
+		else if (o_res->cinfo.before != o_res->cinfo.after)
+			data->file_created = 1;
+	}
 	if ((o_res->rflags & NFS4_OPEN_RESULT_LOCKTYPE_POSIX) == 0)
 		server->caps &= ~NFS_CAP_POSIX_LOCK;
 	if(o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
@@ -2191,7 +2197,8 @@ static int _nfs4_do_open(struct inode *dir,
 			struct nfs_open_context *ctx,
 			int flags,
 			struct iattr *sattr,
-			struct nfs4_label *label)
+			struct nfs4_label *label,
+			int *opened)
 {
 	struct nfs4_state_owner  *sp;
 	struct nfs4_state     *state = NULL;
@@ -2261,6 +2268,8 @@ static int _nfs4_do_open(struct inode *dir,
 			nfs_setsecurity(state->inode, opendata->o_res.f_attr, olabel);
 		}
 	}
+	if (opendata->file_created)
+		*opened |= FILE_CREATED;
 
 	if (pnfs_use_threshold(ctx_th, opendata->f_attr.mdsthreshold, server))
 		*ctx_th = opendata->f_attr.mdsthreshold;
@@ -2289,7 +2298,8 @@ static struct nfs4_state *nfs4_do_open(struct inode *dir,
 					struct nfs_open_context *ctx,
 					int flags,
 					struct iattr *sattr,
-					struct nfs4_label *label)
+					struct nfs4_label *label,
+					int *opened)
 {
 	struct nfs_server *server = NFS_SERVER(dir);
 	struct nfs4_exception exception = { };
@@ -2297,7 +2307,7 @@ static struct nfs4_state *nfs4_do_open(struct inode *dir,
 	int status;
 
 	do {
-		status = _nfs4_do_open(dir, ctx, flags, sattr, label);
+		status = _nfs4_do_open(dir, ctx, flags, sattr, label, opened);
 		res = ctx->state;
 		trace_nfs4_open_file(ctx, flags, status);
 		if (status == 0)
@@ -2659,7 +2669,8 @@ out:
 }
 
 static struct inode *
-nfs4_atomic_open(struct inode *dir, struct nfs_open_context *ctx, int open_flags, struct iattr *attr)
+nfs4_atomic_open(struct inode *dir, struct nfs_open_context *ctx,
+		int open_flags, struct iattr *attr, int *opened)
 {
 	struct nfs4_state *state;
 	struct nfs4_label l = {0, 0, 0, NULL}, *label = NULL;
@@ -2667,7 +2678,7 @@ nfs4_atomic_open(struct inode *dir, struct nfs_open_context *ctx, int open_flags
 	label = nfs4_label_init_security(dir, ctx->dentry, attr, &l);
 
 	/* Protect against concurrent sillydeletes */
-	state = nfs4_do_open(dir, ctx, open_flags, attr, label);
+	state = nfs4_do_open(dir, ctx, open_flags, attr, label, opened);
 
 	nfs4_label_release_security(label);
 
@@ -3332,6 +3343,7 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 	struct nfs4_label l, *ilabel = NULL;
 	struct nfs_open_context *ctx;
 	struct nfs4_state *state;
+	int opened = 0;
 	int status = 0;
 
 	ctx = alloc_nfs_open_context(dentry, FMODE_READ);
@@ -3341,7 +3353,7 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 	ilabel = nfs4_label_init_security(dir, dentry, sattr, &l);
 
 	sattr->ia_mode &= ~current_umask();
-	state = nfs4_do_open(dir, ctx, flags, sattr, ilabel);
+	state = nfs4_do_open(dir, ctx, flags, sattr, ilabel, &opened);
 	if (IS_ERR(state)) {
 		status = PTR_ERR(state);
 		goto out;
diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h
index 01fd84b..49f52c8 100644
--- a/include/linux/nfs_xdr.h
+++ b/include/linux/nfs_xdr.h
@@ -1455,7 +1455,8 @@ struct nfs_rpc_ops {
 	struct inode * (*open_context) (struct inode *dir,
 				struct nfs_open_context *ctx,
 				int open_flags,
-				struct iattr *iattr);
+				struct iattr *iattr,
+				int *);
 	int (*have_delegation)(struct inode *, fmode_t);
 	int (*return_delegation)(struct inode *);
 	struct nfs_client *(*alloc_client) (const struct nfs_client_initdata *);
-- 
1.8.3.1

[PATCH 1/7] SUNRPC: Enable the keepalive option for TCP sockets

[Trond Myklebust]

For NFSv4 we want to avoid retransmitting RPC calls unless the TCP
connection breaks. However we still want to detect TCP connection
breakage as soon as possible. Do this by setting the keepalive option
with the idle timeout and count set to the 'timeo' and 'retrans' mount
options.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 net/sunrpc/xprtsock.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index ee03d35..208a763 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -2112,6 +2112,19 @@ static int xs_tcp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
 
 	if (!transport->inet) {
 		struct sock *sk = sock->sk;
+		unsigned int keepidle = xprt->timeout->to_initval / HZ;
+		unsigned int keepcnt = xprt->timeout->to_retries + 1;
+		unsigned int opt_on = 1;
+
+		/* TCP Keepalive options */
+		kernel_setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE,
+				(char *)&opt_on, sizeof(opt_on));
+		kernel_setsockopt(sock, SOL_TCP, TCP_KEEPIDLE,
+				(char *)&keepidle, sizeof(keepidle));
+		kernel_setsockopt(sock, SOL_TCP, TCP_KEEPINTVL,
+				(char *)&keepidle, sizeof(keepidle));
+		kernel_setsockopt(sock, SOL_TCP, TCP_KEEPCNT,
+				(char *)&keepcnt, sizeof(keepcnt));
 
 		write_lock_bh(&sk->sk_callback_lock);
 
-- 
1.8.3.1

[PATCH 2/7] SUNRPC: Only update the TCP connect cookie on a successful connect

[Trond Myklebust]

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 net/sunrpc/xprtsock.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 208a763..9928ba1 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -1511,6 +1511,7 @@ static void xs_tcp_state_change(struct sock *sk)
 			transport->tcp_copied = 0;
 			transport->tcp_flags =
 				TCP_RCV_COPY_FRAGHDR | TCP_RCV_COPY_XID;
+			xprt->connect_cookie++;
 
 			xprt_wake_pending_tasks(xprt, -EAGAIN);
 		}
@@ -2164,7 +2165,6 @@ static int xs_tcp_finish_connecting(struct rpc_xprt *xprt, struct socket *sock)
 	case 0:
 	case -EINPROGRESS:
 		/* SYN_SENT! */
-		xprt->connect_cookie++;
 		if (xprt->reestablish_timeout < XS_TCP_INIT_REEST_TO)
 			xprt->reestablish_timeout = XS_TCP_INIT_REEST_TO;
 	}
-- 
1.8.3.1

[PATCH 3/7] SUNRPC: Don't set the request connect_cookie until a successful transmit

[Trond Myklebust]

We're using the request connect_cookie to track whether or not a
request was successfully transmitted on the current transport
connection or not. For that reason we should ensure that it is
only set after we've successfully transmitted the request.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 net/sunrpc/xprt.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index 095363e..e9ee7bf 100644
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -912,7 +912,6 @@ void xprt_transmit(struct rpc_task *task)
 	} else if (!req->rq_bytes_sent)
 		return;
 
-	req->rq_connect_cookie = xprt->connect_cookie;
 	req->rq_xtime = ktime_get();
 	status = xprt->ops->send_request(task);
 	if (status != 0) {
@@ -938,12 +937,14 @@ void xprt_transmit(struct rpc_task *task)
 	/* Don't race with disconnect */
 	if (!xprt_connected(xprt))
 		task->tk_status = -ENOTCONN;
-	else if (!req->rq_reply_bytes_recvd && rpc_reply_expected(task)) {
+	else {
 		/*
 		 * Sleep on the pending queue since
 		 * we're expecting a reply.
 		 */
-		rpc_sleep_on(&xprt->pending, task, xprt_timer);
+		if (!req->rq_reply_bytes_recvd && rpc_reply_expected(task))
+			rpc_sleep_on(&xprt->pending, task, xprt_timer);
+		req->rq_connect_cookie = xprt->connect_cookie;
 	}
 	spin_unlock_bh(&xprt->transport_lock);
 }
@@ -1186,6 +1187,7 @@ static void xprt_request_init(struct rpc_task *task, struct rpc_xprt *xprt)
 	req->rq_xprt    = xprt;
 	req->rq_buffer  = NULL;
 	req->rq_xid     = xprt_alloc_xid(xprt);
+	req->rq_connect_cookie = xprt->connect_cookie - 1;
 	req->rq_release_snd_buf = NULL;
 	xprt_reset_majortimeo(req);
 	dprintk("RPC: %5u reserved req %p xid %08x\n", task->tk_pid,
-- 
1.8.3.1

[PATCH 4/7] SUNRPC: Clean up - convert xprt_prepare_transmit to return a bool

[Trond Myklebust]

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 include/linux/sunrpc/xprt.h |  2 +-
 net/sunrpc/clnt.c           |  6 ++----
 net/sunrpc/xprt.c           | 15 +++++++++------
 3 files changed, 12 insertions(+), 11 deletions(-)

diff --git a/include/linux/sunrpc/xprt.h b/include/linux/sunrpc/xprt.h
index cec7b9b..8097b9d 100644
--- a/include/linux/sunrpc/xprt.h
+++ b/include/linux/sunrpc/xprt.h
@@ -288,7 +288,7 @@ int			xprt_reserve_xprt(struct rpc_xprt *xprt, struct rpc_task *task);
 int			xprt_reserve_xprt_cong(struct rpc_xprt *xprt, struct rpc_task *task);
 void			xprt_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task);
 void			xprt_lock_and_alloc_slot(struct rpc_xprt *xprt, struct rpc_task *task);
-int			xprt_prepare_transmit(struct rpc_task *task);
+bool			xprt_prepare_transmit(struct rpc_task *task);
 void			xprt_transmit(struct rpc_task *task);
 void			xprt_end_transmit(struct rpc_task *task);
 int			xprt_adjust_timeout(struct rpc_rqst *req);
diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index 7747960..fa50e42 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -1722,8 +1722,7 @@ call_transmit(struct rpc_task *task)
 	task->tk_action = call_status;
 	if (task->tk_status < 0)
 		return;
-	task->tk_status = xprt_prepare_transmit(task);
-	if (task->tk_status != 0)
+	if (!xprt_prepare_transmit(task))
 		return;
 	task->tk_action = call_transmit_status;
 	/* Encode here so that rpcsec_gss can use correct sequence number. */
@@ -1811,8 +1810,7 @@ call_bc_transmit(struct rpc_task *task)
 {
 	struct rpc_rqst *req = task->tk_rqstp;
 
-	task->tk_status = xprt_prepare_transmit(task);
-	if (task->tk_status == -EAGAIN) {
+	if (!xprt_prepare_transmit(task)) {
 		/*
 		 * Could not reserve the transport. Try again after the
 		 * transport is released.
diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index e9ee7bf..c94f4a2 100644
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -854,24 +854,27 @@ static inline int xprt_has_timer(struct rpc_xprt *xprt)
  * @task: RPC task about to send a request
  *
  */
-int xprt_prepare_transmit(struct rpc_task *task)
+bool xprt_prepare_transmit(struct rpc_task *task)
 {
 	struct rpc_rqst	*req = task->tk_rqstp;
 	struct rpc_xprt	*xprt = req->rq_xprt;
-	int err = 0;
+	bool ret = false;
 
 	dprintk("RPC: %5u xprt_prepare_transmit\n", task->tk_pid);
 
 	spin_lock_bh(&xprt->transport_lock);
 	if (req->rq_reply_bytes_recvd && !req->rq_bytes_sent) {
-		err = req->rq_reply_bytes_recvd;
+		task->tk_status = req->rq_reply_bytes_recvd;
 		goto out_unlock;
 	}
-	if (!xprt->ops->reserve_xprt(xprt, task))
-		err = -EAGAIN;
+	if (!xprt->ops->reserve_xprt(xprt, task)) {
+		task->tk_status = -EAGAIN;
+		goto out_unlock;
+	}
+	ret = true;
 out_unlock:
 	spin_unlock_bh(&xprt->transport_lock);
-	return err;
+	return ret;
 }
 
 void xprt_end_transmit(struct rpc_task *task)
-- 
1.8.3.1

[PATCH 5/7] SUNRPC: Add RPC task and client level options to disable the resend timeout

[Trond Myklebust]

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 include/linux/sunrpc/clnt.h  | 2 ++
 include/linux/sunrpc/sched.h | 1 +
 net/sunrpc/clnt.c            | 5 ++++-
 net/sunrpc/xprt.c            | 7 +++++++
 4 files changed, 14 insertions(+), 1 deletion(-)

diff --git a/include/linux/sunrpc/clnt.h b/include/linux/sunrpc/clnt.h
index 6740801..943ee89 100644
--- a/include/linux/sunrpc/clnt.h
+++ b/include/linux/sunrpc/clnt.h
@@ -49,6 +49,7 @@ struct rpc_clnt {
 
 	unsigned int		cl_softrtry : 1,/* soft timeouts */
 				cl_discrtry : 1,/* disconnect before retry */
+				cl_noretranstimeo: 1,/* No retransmit timeouts */
 				cl_autobind : 1,/* use getport() */
 				cl_chatty   : 1;/* be verbose */
 
@@ -126,6 +127,7 @@ struct rpc_create_args {
 #define RPC_CLNT_CREATE_QUIET		(1UL << 6)
 #define RPC_CLNT_CREATE_INFINITE_SLOTS	(1UL << 7)
 #define RPC_CLNT_CREATE_NO_IDLE_TIMEOUT	(1UL << 8)
+#define RPC_CLNT_CREATE_NO_RETRANS_TIMEOUT	(1UL << 9)
 
 struct rpc_clnt *rpc_create(struct rpc_create_args *args);
 struct rpc_clnt	*rpc_bind_new_program(struct rpc_clnt *,
diff --git a/include/linux/sunrpc/sched.h b/include/linux/sunrpc/sched.h
index 096ee58..3a847de 100644
--- a/include/linux/sunrpc/sched.h
+++ b/include/linux/sunrpc/sched.h
@@ -122,6 +122,7 @@ struct rpc_task_setup {
 #define RPC_TASK_SENT		0x0800		/* message was sent */
 #define RPC_TASK_TIMEOUT	0x1000		/* fail with ETIMEDOUT on timeout */
 #define RPC_TASK_NOCONNECT	0x2000		/* return ENOTCONN if not connected */
+#define RPC_TASK_NO_RETRANS_TIMEOUT	0x4000		/* wait forever for a reply */
 
 #define RPC_IS_ASYNC(t)		((t)->tk_flags & RPC_TASK_ASYNC)
 #define RPC_IS_SWAPPER(t)	((t)->tk_flags & RPC_TASK_SWAPPER)
diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index fa50e42..b585250 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -772,6 +772,8 @@ void rpc_task_set_client(struct rpc_task *task, struct rpc_clnt *clnt)
 		atomic_inc(&clnt->cl_count);
 		if (clnt->cl_softrtry)
 			task->tk_flags |= RPC_TASK_SOFT;
+		if (clnt->cl_noretranstimeo)
+			task->tk_flags |= RPC_TASK_NO_RETRANS_TIMEOUT;
 		if (sk_memalloc_socks()) {
 			struct rpc_xprt *xprt;
 
@@ -1898,7 +1900,8 @@ call_status(struct rpc_task *task)
 		rpc_delay(task, 3*HZ);
 	case -ETIMEDOUT:
 		task->tk_action = call_timeout;
-		if (task->tk_client->cl_discrtry)
+		if (!(task->tk_flags & RPC_TASK_NO_RETRANS_TIMEOUT)
+		    && task->tk_client->cl_discrtry)
 			xprt_conditional_disconnect(req->rq_xprt,
 					req->rq_connect_cookie);
 		break;
diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index c94f4a2..d992d7f 100644
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -867,6 +867,13 @@ bool xprt_prepare_transmit(struct rpc_task *task)
 		task->tk_status = req->rq_reply_bytes_recvd;
 		goto out_unlock;
 	}
+	if ((task->tk_flags & RPC_TASK_NO_RETRANS_TIMEOUT)
+	    && xprt_connected(xprt)
+	    && req->rq_connect_cookie == xprt->connect_cookie) {
+		xprt->ops->set_retrans_timeout(task);
+		rpc_sleep_on(&xprt->pending, task, xprt_timer);
+		goto out_unlock;
+	}
 	if (!xprt->ops->reserve_xprt(xprt, task)) {
 		task->tk_status = -EAGAIN;
 		goto out_unlock;
-- 
1.8.3.1

[PATCH 6/7] NFSv4: Ensure that we disable the resend timeout for NFSv4

[Trond Myklebust]

The spec states that the client should not resend requests because
the server will disconnect if it needs to drop an RPC request.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 fs/nfs/client.c           | 2 ++
 fs/nfs/nfs4client.c       | 1 +
 include/linux/nfs_fs_sb.h | 1 +
 3 files changed, 4 insertions(+)

diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index 2dceee4..af03258 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -590,6 +590,8 @@ int nfs_create_rpc_client(struct nfs_client *clp,
 
 	if (test_bit(NFS_CS_DISCRTRY, &clp->cl_flags))
 		args.flags |= RPC_CLNT_CREATE_DISCRTRY;
+	if (test_bit(NFS_CS_NO_RETRANS_TIMEOUT, &clp->cl_flags))
+		args.flags |= RPC_CLNT_CREATE_NO_RETRANS_TIMEOUT;
 	if (test_bit(NFS_CS_NORESVPORT, &clp->cl_flags))
 		args.flags |= RPC_CLNT_CREATE_NONPRIVPORT;
 	if (test_bit(NFS_CS_INFINITE_SLOTS, &clp->cl_flags))
diff --git a/fs/nfs/nfs4client.c b/fs/nfs/nfs4client.c
index a860ab5..511cdce 100644
--- a/fs/nfs/nfs4client.c
+++ b/fs/nfs/nfs4client.c
@@ -368,6 +368,7 @@ struct nfs_client *nfs4_init_client(struct nfs_client *clp,
 	if (clp->cl_minorversion != 0)
 		__set_bit(NFS_CS_INFINITE_SLOTS, &clp->cl_flags);
 	__set_bit(NFS_CS_DISCRTRY, &clp->cl_flags);
+	__set_bit(NFS_CS_NO_RETRANS_TIMEOUT, &clp->cl_flags);
 	error = nfs_create_rpc_client(clp, timeparms, RPC_AUTH_GSS_KRB5I);
 	if (error == -EINVAL)
 		error = nfs_create_rpc_client(clp, timeparms, RPC_AUTH_UNIX);
diff --git a/include/linux/nfs_fs_sb.h b/include/linux/nfs_fs_sb.h
index b8cedce..f9c0a6c 100644
--- a/include/linux/nfs_fs_sb.h
+++ b/include/linux/nfs_fs_sb.h
@@ -41,6 +41,7 @@ struct nfs_client {
 #define NFS_CS_DISCRTRY		1		/* - disconnect on RPC retry */
 #define NFS_CS_MIGRATION	2		/* - transparent state migr */
 #define NFS_CS_INFINITE_SLOTS	3		/* - don't limit TCP slots */
+#define NFS_CS_NO_RETRANS_TIMEOUT	4	/* - Disable retransmit timeouts */
 	struct sockaddr_storage	cl_addr;	/* server identifier */
 	size_t			cl_addrlen;
 	char *			cl_hostname;	/* hostname of server */
-- 
1.8.3.1

[PATCH 7/7] SUNRPC: Fix RPC call retransmission statistics

[Trond Myklebust]

A retransmit should be when you successfully transmit an RPC call to
the server a second time.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 net/sunrpc/clnt.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index b585250..46d061f 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -1719,6 +1719,8 @@ call_connect_status(struct rpc_task *task)
 static void
 call_transmit(struct rpc_task *task)
 {
+	bool is_retrans = (task->tk_flags & RPC_TASK_SENT) != 0;
+
 	dprint_status(task);
 
 	task->tk_action = call_status;
@@ -1743,6 +1745,8 @@ call_transmit(struct rpc_task *task)
 	xprt_transmit(task);
 	if (task->tk_status < 0)
 		return;
+	if (is_retrans)
+		task->tk_client->cl_stats->rpcretrans++;
 	/*
 	 * On success, ensure that we call xprt_end_transmit() before sleeping
 	 * in order to allow access to the socket to other RPC requests.
@@ -1983,7 +1987,6 @@ call_timeout(struct rpc_task *task)
 	rpcauth_invalcred(task);
 
 retry:
-	clnt->cl_stats->rpcretrans++;
 	task->tk_action = call_bind;
 	task->tk_status = 0;
 }
@@ -2026,7 +2029,6 @@ call_decode(struct rpc_task *task)
 	if (req->rq_rcv_buf.len < 12) {
 		if (!RPC_IS_SOFT(task)) {
 			task->tk_action = call_bind;
-			clnt->cl_stats->rpcretrans++;
 			goto out_retry;
 		}
 		dprintk("RPC:       %s: too small RPC reply size (%d bytes)\n",
-- 
1.8.3.1

[PATCH] SUNRPC: call_connect_status should recheck bind and connect status on error

[Trond Myklebust]

Currently, we go directly to call_transmit which sends us to call_status
on error. If we know that the connect attempt failed, we should rather
just jump straight back to call_bind and/or call_transmit

Also ensure that EAGAIN checks for a timeout event.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
---
 net/sunrpc/clnt.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index 46d061f..6ebf74c 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -1692,20 +1692,23 @@ call_connect_status(struct rpc_task *task)
 	dprint_status(task);
 
 	trace_rpc_connect_status(task, status);
+	task->tk_status = 0;
 	switch (status) {
 		/* if soft mounted, test if we've timed out */
+	case -EAGAIN:
 	case -ETIMEDOUT:
 		task->tk_action = call_timeout;
 		return;
 	case -ECONNREFUSED:
 	case -ECONNRESET:
 	case -ENETUNREACH:
+		/* retry with existing socket, after a delay */
+		rpc_delay(task, 3*HZ);
 		if (RPC_IS_SOFTCONN(task))
 			break;
-		/* retry with existing socket, after a delay */
+		task->tk_action = call_bind;
+		return;
 	case 0:
-	case -EAGAIN:
-		task->tk_status = 0;
 		clnt->cl_stats->netreconn++;
 		task->tk_action = call_transmit;
 		return;
-- 
1.8.3.1