Re: [3.2.5] NFSv3 CLOSE_WAIT hang

[dick.streefland@altium.nl (Dick Streefland)]

"Myklebust, Trond" <Trond.Myklebust@netapp.com> wrote:
| Clearly the comment is misleading and should be removed. That write lock
| 
| _is_ needed in order to throttle slots on TCP.
| 
| As far as I know, kernel 3.3 is not in stable support any more, so I
| can't help that. Can you reproduce the problem on a 3.5 kernel or
| higher?

I can easily reproduce the hang on the latest kernel (eeea3ac912)
with my script (http://permalink.gmane.org/gmane.linux.nfs/51833).
With this change, it doesn't hang:

diff --git a/net/sunrpc/xprt.c b/net/sunrpc/xprt.c
index a5a402a..6e739bf 100644
--- a/net/sunrpc/xprt.c
+++ b/net/sunrpc/xprt.c
@@ -1083,20 +1083,9 @@ void xprt_reserve(struct rpc_task *task)
 	if (task->tk_rqstp != NULL)
 		return;
 
-	/* Note: grabbing the xprt_lock_write() here is not strictly needed,
-	 * but ensures that we throttle new slot allocation if the transport
-	 * is congested (e.g. if reconnecting or if we're out of socket
-	 * write buffer space).
-	 */
-	task->tk_timeout = 0;
-	task->tk_status = -EAGAIN;
-	if (!xprt_lock_write(xprt, task))
-		return;
-
 	spin_lock(&xprt->reserve_lock);
 	xprt_alloc_slot(task);
 	spin_unlock(&xprt->reserve_lock);
-	xprt_release_write(xprt, task);
 }
 
 static inline __be32 xprt_alloc_xid(struct rpc_xprt *xprt)

-- 
Dick