[PATCH] RPC: proper soft timeout behavior for rpcbind

 Implement a best practice:  for soft mounts, an rpcbind timeout should
 cause an RPC request to fail.

 This also provides an FSM hook for retrying an rpcbind with a different
 rpcbind protocol version.  We'll use this later to try multiple rpcbind
 protocol versions when binding.  To enable this, expose the RPC error
 code returned during a portmap request to the FSM so it can make some
 decision about how to report, retry, or fail the request.

 Test-plan:
 Hundreds of passes with connectathon NFSv3 locking suite, on the client
 and server.

 Version: Thu, 11 Aug 2005 16:01:53 -0400

 Signed-off-by: Chuck Lever <cel@netapp.com>
 Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

1 files changed, 77 insertions, 20 deletions

diff --git a/net/sunrpc/clnt.c b/net/sunrpc/clnt.c
index f17e6153b688..2d3cf0a52d82 100644
--- a/net/sunrpc/clnt.c
+++ b/net/sunrpc/clnt.c
@@ -53,6 +53,7 @@ static void	call_allocate(struct rpc_task *task);
 static void     call_encode(struct rpc_task *task);
 static void     call_decode(struct rpc_task *task);
 static void     call_bind(struct rpc_task *task);
+static void     call_bind_status(struct rpc_task *task);
 static void     call_transmit(struct rpc_task *task);
 static void     call_status(struct rpc_task *task);
 static void     call_refresh(struct rpc_task *task);
@@ -734,43 +735,94 @@ static void
 call_bind(struct rpc_task *task)
 {
         struct rpc_clnt *clnt = task->tk_client;
-        struct rpc_xprt *xprt = clnt->cl_xprt;
-
-        dprintk("RPC: %4d call_bind xprt %p %s connected\n", task->tk_pid,
-                        xprt, (xprt_connected(xprt) ? "is" : "is not"));
 
-        task->tk_action = (xprt_connected(xprt)) ? call_transmit : call_connect;
+        dprintk("RPC: %4d call_bind (status %d)\n",
+                                task->tk_pid, task->tk_status);
 
+        task->tk_action = call_connect;
         if (!clnt->cl_port) {
-                task->tk_action = call_connect;
+                task->tk_action = call_bind_status;
                 task->tk_timeout = RPC_CONNECT_TIMEOUT;
                 rpc_getport(task, clnt);
         }
 }
 
 /*
- * 4a.  Connect to the RPC server (TCP case)
+ * 4a.  Sort out bind result
+ */
+static void
+call_bind_status(struct rpc_task *task)
+{
+        int status = -EACCES;
+
+        if (task->tk_status >= 0) {
+                dprintk("RPC: %4d call_bind_status (status %d)\n",
+                                        task->tk_pid, task->tk_status);
+                task->tk_status = 0;
+                task->tk_action = call_connect;
+                return;
+        }
+
+        switch (task->tk_status) {
+        case -EACCES:
+                dprintk("RPC: %4d remote rpcbind: RPC program/version unavailable\n",
+                                task->tk_pid);
+                break;
+        case -ETIMEDOUT:
+                dprintk("RPC: %4d rpcbind request timed out\n",
+                                task->tk_pid);
+                if (RPC_IS_SOFT(task)) {
+                        status = -EIO;
+                        break;
+                }
+                goto retry_bind;
+        case -EPFNOSUPPORT:
+                dprintk("RPC: %4d remote rpcbind service unavailable\n",
+                                task->tk_pid);
+                break;
+        case -EPROTONOSUPPORT:
+                dprintk("RPC: %4d remote rpcbind version 2 unavailable\n",
+                                task->tk_pid);
+                break;
+        default:
+                dprintk("RPC: %4d unrecognized rpcbind error (%d)\n",
+                                task->tk_pid, -task->tk_status);
+                status = -EIO;
+                break;
+        }
+
+        rpc_exit(task, status);
+        return;
+
+retry_bind:
+        task->tk_status = 0;
+        task->tk_action = call_bind;
+        return;
+}
+
+/*
+ * 4b.  Connect to the RPC server
  */
 static void
 call_connect(struct rpc_task *task)
 {
-        struct rpc_clnt *clnt = task->tk_client;
+        struct rpc_xprt *xprt = task->tk_xprt;
 
-        dprintk("RPC: %4d call_connect status %d\n",
-                                task->tk_pid, task->tk_status);
+        dprintk("RPC: %4d call_connect xprt %p %s connected\n",
+                        task->tk_pid, xprt,
+                        (xprt_connected(xprt) ? "is" : "is not"));
 
-        if (xprt_connected(clnt->cl_xprt)) {
-                task->tk_action = call_transmit;
-                return;
+        task->tk_action = call_transmit;
+        if (!xprt_connected(xprt)) {
+                task->tk_action = call_connect_status;
+                if (task->tk_status < 0)
+                        return;
+                xprt_connect(task);
         }
-        task->tk_action = call_connect_status;
-        if (task->tk_status < 0)
-                return;
-        xprt_connect(task);
 }
 
 /*
- * 4b. Sort out connect result
+ * 4c.  Sort out connect result
  */
 static void
 call_connect_status(struct rpc_task *task)
@@ -778,6 +830,9 @@ call_connect_status(struct rpc_task *task)
         struct rpc_clnt *clnt = task->tk_client;
         int status = task->tk_status;
 
+        dprintk("RPC: %5u call_connect_status (status %d)\n", 
+                                task->tk_pid, task->tk_status);
+
         task->tk_status = 0;
         if (status >= 0) {
                 clnt->cl_stats->netreconn++;
@@ -785,17 +840,19 @@ call_connect_status(struct rpc_task *task)
                 return;
         }
 
-        /* Something failed: we may have to rebind */
+        /* Something failed: remote service port may have changed */
         if (clnt->cl_autobind)
                 clnt->cl_port = 0;
+
         switch (status) {
         case -ENOTCONN:
         case -ETIMEDOUT:
         case -EAGAIN:
-                task->tk_action = (clnt->cl_port == 0) ? call_bind : call_connect;
+                task->tk_action = call_bind;
                 break;
         default:
                 rpc_exit(task, -EIO);
+                break;
         }
 }