afs: add afs_wq and use it instead of the system workqueue

flush_scheduled_work() is going away.  afs needs to make sure all the
works it has queued have finished before being unloaded and there can
be arbitrary number of pending works.  Add afs_wq and use it as the
flush domain instead of the system workqueue.

Also, convert cancel_delayed_work() + flush_scheduled_work() to
cancel_delayed_work_sync() in afs_mntpt_kill_timer().

Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Cc: linux-afs@lists.infradead.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
diff --git a/fs/afs/rxrpc.c b/fs/afs/rxrpc.c
index 654d8fd..e45a323 100644
--- a/fs/afs/rxrpc.c
+++ b/fs/afs/rxrpc.c
@@ -410,7 +410,7 @@
 	if (!call) {
 		/* its an incoming call for our callback service */
 		skb_queue_tail(&afs_incoming_calls, skb);
-		schedule_work(&afs_collect_incoming_call_work);
+		queue_work(afs_wq, &afs_collect_incoming_call_work);
 	} else {
 		/* route the messages directly to the appropriate call */
 		skb_queue_tail(&call->rx_queue, skb);