SUNRPC: Handle connection breakages correctly in call_status()
authorTrond Myklebust <trond.myklebust@hammerspace.com>
Fri, 16 Aug 2019 12:58:48 +0000 (08:58 -0400)
committerTrond Myklebust <trond.myklebust@hammerspace.com>
Mon, 26 Aug 2019 19:31:29 +0000 (15:31 -0400)
If the connection breaks while we're waiting for a reply from the
server, then we want to immediately try to reconnect.

Fixes: ec6017d90359 ("SUNRPC fix regression in umount of a secure mount")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
net/sunrpc/clnt.c

index 8a25440..a07b516 100644 (file)
@@ -2382,7 +2382,7 @@ call_status(struct rpc_task *task)
        case -ECONNABORTED:
        case -ENOTCONN:
                rpc_force_rebind(clnt);
-               /* fall through */
+               break;
        case -EADDRINUSE:
                rpc_delay(task, 3*HZ);
                /* fall through */