SUNRPC: RPC transport queue must be low latency

rpciod can easily get congested due to the long list of queued rpc_tasks.
Having the receive queue wait in turn for those tasks to complete can
therefore be a bottleneck.

Address the problem by separating the workqueues into:
- rpciod: manages rpc_tasks
- xprtiod: manages transport related work.

Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>

1 files changed, 20 insertions, 4 deletions

diff --git a/net/sunrpc/sched.c b/net/sunrpc/sched.c
index fcfd48d263f6..a9f786247ffb 100644
--- a/net/sunrpc/sched.c
+++ b/net/sunrpc/sched.c
@@ -54,7 +54,8 @@ static struct rpc_wait_queue delay_queue;
 /*
  * rpciod-related stuff
  */
-struct workqueue_struct *rpciod_workqueue;
+struct workqueue_struct *rpciod_workqueue __read_mostly;
+struct workqueue_struct *xprtiod_workqueue __read_mostly;
 
 /*
  * Disable the timer for a given RPC task. Should be called with
@@ -1071,10 +1072,22 @@ static int rpciod_start(void)
 	 * Create the rpciod thread and wait for it to start.
 	 */
 	dprintk("RPC:       creating workqueue rpciod\n");
-	/* Note: highpri because network receive is latency sensitive */
-	wq = alloc_workqueue("rpciod", WQ_MEM_RECLAIM | WQ_HIGHPRI, 0);
+	wq = alloc_workqueue("rpciod", WQ_MEM_RECLAIM, 0);
+	if (!wq)
+		goto out_failed;
 	rpciod_workqueue = wq;
-	return rpciod_workqueue != NULL;
+	/* Note: highpri because network receive is latency sensitive */
+	wq = alloc_workqueue("xprtiod", WQ_MEM_RECLAIM | WQ_HIGHPRI, 0);
+	if (!wq)
+		goto free_rpciod;
+	xprtiod_workqueue = wq;
+	return 1;
+free_rpciod:
+	wq = rpciod_workqueue;
+	rpciod_workqueue = NULL;
+	destroy_workqueue(wq);
+out_failed:
+	return 0;
 }
 
 static void rpciod_stop(void)
@@ -1088,6 +1101,9 @@ static void rpciod_stop(void)
 	wq = rpciod_workqueue;
 	rpciod_workqueue = NULL;
 	destroy_workqueue(wq);
+	wq = xprtiod_workqueue;
+	xprtiod_workqueue = NULL;
+	destroy_workqueue(wq);
 }
 
 void