SUNRPC: Move xpt_mutex into socket xpo_sendto methods

It appears that the RPC/RDMA transport does not need serialization
of calls to its xpo_sendto method. Move the mutex into the socket
methods that still need that serialization.

Tail latencies are unambiguously better with this patch applied.
fio randrw 8KB 70/30 on NFSv3, smaller numbers are better:

    clat percentiles (usec):

With xpt_mutex:
r    | 99.99th=[ 8848]
w    | 99.99th=[ 9634]

Without xpt_mutex:
r    | 99.99th=[ 8586]
w    | 99.99th=[ 8979]

Serializing the construction of RPC/RDMA transport headers is not
really necessary at this point, because the Linux NFS server
implementation never changes its credit grant on a connection. If
that should change, then svc_rdma_sendto will need to serialize
access to the transport's credit grant fields.

Reported-by: kbuild test robot <lkp@intel.com>
[ cel: fix uninitialized variable warning ]
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>

6 files changed, 64 insertions, 36 deletions

diff --git a/include/linux/sunrpc/svc_xprt.h b/include/linux/sunrpc/svc_xprt.h
index 9e1e046de176..aca35ab5cff2 100644
--- a/include/linux/sunrpc/svc_xprt.h
+++ b/include/linux/sunrpc/svc_xprt.h
@@ -117,6 +117,12 @@ static inline int register_xpt_user(struct svc_xprt *xpt, struct svc_xpt_user *u
 	return 0;
 }
 
+static inline bool svc_xprt_is_dead(const struct svc_xprt *xprt)
+{
+	return (test_bit(XPT_DEAD, &xprt->xpt_flags) != 0) ||
+		(test_bit(XPT_CLOSE, &xprt->xpt_flags) != 0);
+}
+
 int	svc_reg_xprt_class(struct svc_xprt_class *);
 void	svc_unreg_xprt_class(struct svc_xprt_class *);
 void	svc_xprt_init(struct net *, struct svc_xprt_class *, struct svc_xprt *,
diff --git a/net/sunrpc/svc_xprt.c b/net/sunrpc/svc_xprt.c
index 2284ff038dad..07cdbf7d5764 100644
--- a/net/sunrpc/svc_xprt.c
+++ b/net/sunrpc/svc_xprt.c
@@ -914,16 +914,10 @@ int svc_send(struct svc_rqst *rqstp)
 		xb->page_len +
 		xb->tail[0].iov_len;
 	trace_svc_sendto(xb);
-
-	/* Grab mutex to serialize outgoing data. */
-	mutex_lock(&xprt->xpt_mutex);
 	trace_svc_stats_latency(rqstp);
-	if (test_bit(XPT_DEAD, &xprt->xpt_flags)
-			|| test_bit(XPT_CLOSE, &xprt->xpt_flags))
-		len = -ENOTCONN;
-	else
-		len = xprt->xpt_ops->xpo_sendto(rqstp);
-	mutex_unlock(&xprt->xpt_mutex);
+
+	len = xprt->xpt_ops->xpo_sendto(rqstp);
+
 	trace_svc_send(rqstp, len);
 	svc_xprt_release(rqstp);
 
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 023514e392b3..3e7b6445e317 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -506,6 +506,9 @@ out_free:
  * svc_udp_sendto - Send out a reply on a UDP socket
  * @rqstp: completed svc_rqst
  *
+ * xpt_mutex ensures @rqstp's whole message is written to the socket
+ * without interruption.
+ *
  * Returns the number of bytes sent, or a negative errno.
  */
 static int svc_udp_sendto(struct svc_rqst *rqstp)
@@ -531,6 +534,11 @@ static int svc_udp_sendto(struct svc_rqst *rqstp)
 
 	svc_set_cmsg_data(rqstp, cmh);
 
+	mutex_lock(&xprt->xpt_mutex);
+
+	if (svc_xprt_is_dead(xprt))
+		goto out_notconn;
+
 	err = xprt_sock_sendmsg(svsk->sk_sock, &msg, xdr, 0, 0, &sent);
 	xdr_free_bvec(xdr);
 	if (err == -ECONNREFUSED) {
@@ -538,9 +546,15 @@ static int svc_udp_sendto(struct svc_rqst *rqstp)
 		err = xprt_sock_sendmsg(svsk->sk_sock, &msg, xdr, 0, 0, &sent);
 		xdr_free_bvec(xdr);
 	}
+
+	mutex_unlock(&xprt->xpt_mutex);
 	if (err < 0)
 		return err;
 	return sent;
+
+out_notconn:
+	mutex_unlock(&xprt->xpt_mutex);
+	return -ENOTCONN;
 }
 
 static int svc_udp_has_wspace(struct svc_xprt *xprt)
@@ -1063,6 +1077,9 @@ err_noclose:
  * svc_tcp_sendto - Send out a reply on a TCP socket
  * @rqstp: completed svc_rqst
  *
+ * xpt_mutex ensures @rqstp's whole message is written to the socket
+ * without interruption.
+ *
  * Returns the number of bytes sent, or a negative errno.
  */
 static int svc_tcp_sendto(struct svc_rqst *rqstp)
@@ -1080,12 +1097,19 @@ static int svc_tcp_sendto(struct svc_rqst *rqstp)
 
 	svc_release_skb(rqstp);
 
+	mutex_lock(&xprt->xpt_mutex);
+	if (svc_xprt_is_dead(xprt))
+		goto out_notconn;
 	err = xprt_sock_sendmsg(svsk->sk_sock, &msg, xdr, 0, marker, &sent);
 	xdr_free_bvec(xdr);
 	if (err < 0 || sent != (xdr->len + sizeof(marker)))
 		goto out_close;
+	mutex_unlock(&xprt->xpt_mutex);
 	return sent;
 
+out_notconn:
+	mutex_unlock(&xprt->xpt_mutex);
+	return -ENOTCONN;
 out_close:
 	pr_notice("rpc-srv/tcp: %s: %s %d when sending %d bytes - shutting down socket\n",
 		  xprt->xpt_server->sv_name,
@@ -1093,6 +1117,7 @@ out_close:
 		  (err < 0) ? err : sent, xdr->len);
 	set_bit(XPT_CLOSE, &xprt->xpt_flags);
 	svc_xprt_enqueue(xprt);
+	mutex_unlock(&xprt->xpt_mutex);
 	return -EAGAIN;
 }
 
diff --git a/net/sunrpc/xprtrdma/svc_rdma_backchannel.c b/net/sunrpc/xprtrdma/svc_rdma_backchannel.c
index af7eb8d202ae..d9aab4504f2c 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_backchannel.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_backchannel.c
@@ -210,34 +210,31 @@ drop_connection:
 	return -ENOTCONN;
 }
 
-/* Send an RPC call on the passive end of a transport
- * connection.
+/**
+ * xprt_rdma_bc_send_request - Send a reverse-direction Call
+ * @rqst: rpc_rqst containing Call message to be sent
+ *
+ * Return values:
+ *   %0 if the message was sent successfully
+ *   %ENOTCONN if the message was not sent
  */
-static int
-xprt_rdma_bc_send_request(struct rpc_rqst *rqst)
+static int xprt_rdma_bc_send_request(struct rpc_rqst *rqst)
 {
 	struct svc_xprt *sxprt = rqst->rq_xprt->bc_xprt;
-	struct svcxprt_rdma *rdma;
+	struct svcxprt_rdma *rdma =
+		container_of(sxprt, struct svcxprt_rdma, sc_xprt);
 	int ret;
 
 	dprintk("svcrdma: sending bc call with xid: %08x\n",
 		be32_to_cpu(rqst->rq_xid));
 
-	mutex_lock(&sxprt->xpt_mutex);
-
-	ret = -ENOTCONN;
-	rdma = container_of(sxprt, struct svcxprt_rdma, sc_xprt);
-	if (!test_bit(XPT_DEAD, &sxprt->xpt_flags)) {
-		ret = rpcrdma_bc_send_request(rdma, rqst);
-		if (ret == -ENOTCONN)
-			svc_close_xprt(sxprt);
-	}
+	if (test_bit(XPT_DEAD, &sxprt->xpt_flags))
+		return -ENOTCONN;
 
-	mutex_unlock(&sxprt->xpt_mutex);
-
-	if (ret < 0)
-		return ret;
-	return 0;
+	ret = rpcrdma_bc_send_request(rdma, rqst);
+	if (ret == -ENOTCONN)
+		svc_close_xprt(sxprt);
+	return ret;
 }
 
 static void
diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
index b6c8643867f2..38e7c3c8c4a9 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
@@ -868,12 +868,10 @@ int svc_rdma_sendto(struct svc_rqst *rqstp)
 	__be32 *p;
 	int ret;
 
-	/* Create the RDMA response header. xprt->xpt_mutex,
-	 * acquired in svc_send(), serializes RPC replies. The
-	 * code path below that inserts the credit grant value
-	 * into each transport header runs only inside this
-	 * critical section.
-	 */
+	ret = -ENOTCONN;
+	if (svc_xprt_is_dead(xprt))
+		goto err0;
+
 	ret = -ENOMEM;
 	sctxt = svc_rdma_send_ctxt_get(rdma);
 	if (!sctxt)
diff --git a/net/sunrpc/xprtsock.c b/net/sunrpc/xprtsock.c
index 845d0be805ec..839c49330785 100644
--- a/net/sunrpc/xprtsock.c
+++ b/net/sunrpc/xprtsock.c
@@ -2548,8 +2548,16 @@ static int bc_sendto(struct rpc_rqst *req)
 	return sent;
 }
 
-/*
- * The send routine. Borrows from svc_send
+/**
+ * bc_send_request - Send a backchannel Call on a TCP socket
+ * @req: rpc_rqst containing Call message to be sent
+ *
+ * xpt_mutex ensures @rqstp's whole message is written to the socket
+ * without interruption.
+ *
+ * Return values:
+ *   %0 if the message was sent successfully
+ *   %ENOTCONN if the message was not sent
  */
 static int bc_send_request(struct rpc_rqst *req)
 {