SUNRPC: Ensure that call_connect times out correctly
authorTrond Myklebust <trond.myklebust@primarydata.com>
Mon, 17 Mar 2014 16:51:44 +0000 (12:51 -0400)
committerTrond Myklebust <trond.myklebust@primarydata.com>
Mon, 17 Mar 2014 19:14:18 +0000 (15:14 -0400)
When the server is unavailable due to a networking error, etc, we want
the RPC client to respect the timeout delays when attempting to reconnect.

Reported-by: Neil Brown <neilb@suse.de>
Fixes: 561ec1603171 (SUNRPC: call_connect_status should recheck bind..)
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
net/sunrpc/clnt.c

index 0edada973434e4df883f63a85173f4a7a1de0885..5a1b8fa9ca132548760b3ec88fd11d8ba94b696e 100644 (file)
@@ -1798,10 +1798,6 @@ call_connect_status(struct rpc_task *task)
        trace_rpc_connect_status(task, status);
        task->tk_status = 0;
        switch (status) {
-               /* if soft mounted, test if we've timed out */
-       case -ETIMEDOUT:
-               task->tk_action = call_timeout;
-               return;
        case -ECONNREFUSED:
        case -ECONNRESET:
        case -ECONNABORTED:
@@ -1812,7 +1808,9 @@ call_connect_status(struct rpc_task *task)
                if (RPC_IS_SOFTCONN(task))
                        break;
        case -EAGAIN:
-               task->tk_action = call_bind;
+               /* Check for timeouts before looping back to call_bind */
+       case -ETIMEDOUT:
+               task->tk_action = call_timeout;
                return;
        case 0:
                clnt->cl_stats->netreconn++;