[PATCH 1/2] SUNRPC: Prevent thundering herd when the socket is not connected

[PATCH 1/2] SUNRPC: Prevent thundering herd when the socket is not connected

[Trond Myklebust]

If the socket is not connected, then we want to initiate a reconnect
rather that trying to transmit requests. If there is a large number
of requests queued and waiting for the lock in call_transmit(),
then it can take a while for one of the to loop back and retake
the lock in call_connect.

Fixes: 89f90fe1ad8b ("SUNRPC: Allow calls to xprt_transmit() to drain...")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
---
 net/sunrpc/clnt.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index 241e8423fd0c..e7ae75a045c9 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -2003,6 +2003,12 @@ call_transmit(struct rpc_task *task)
 	if (test_bit(RPC_TASK_NEED_XMIT, &task->tk_runstate)) {
 		if (!xprt_prepare_transmit(task))
 			return;
+		/* Check that the connection is OK */
+		if (!xprt_connected(task->tk_xprt) ||
+		    !xprt_bound(task->tk_xprt)) {
+			task->tk_action = call_bind;
+			return;
+		}
 		xprt_transmit(task);
 	}
 	task->tk_action = call_transmit_status;
-- 
2.20.1

[PATCH 2/2] SUNRPC: Micro-optimisation to avoid call_bind+call_commit

[Trond Myklebust]

Most tasks should not have to deal with disconnected or unbound
socket states, so let's move them out of the common state machine
path.

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

diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index e7ae75a045c9..8678a05192d3 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -1807,7 +1807,7 @@ call_encode(struct rpc_task *task)
 		xprt_request_enqueue_receive(task);
 	xprt_request_enqueue_transmit(task);
 out:
-	task->tk_action = call_bind;
+	task->tk_action = call_transmit;
 }
 
 /*
-- 
2.20.1