pnfs: retry after a bad stateid error from layoutget
authorChristoph Hellwig <hch@lst.de>
Thu, 21 Aug 2014 16:09:21 +0000 (11:09 -0500)
committerTrond Myklebust <trond.myklebust@primarydata.com>
Wed, 10 Sep 2014 19:47:01 +0000 (12:47 -0700)
Currently we fall through to nfs4_async_handle_error when we get
a bad stateid error back from layoutget.  nfs4_async_handle_error
with a NULL state argument will never retry the operations but return
the error to higher layer, causing an avoiable fallback to MDS I/O.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
fs/nfs/nfs4proc.c

index 18eb31c..47fa67a 100644 (file)
@@ -7577,11 +7577,16 @@ static void nfs4_layoutget_done(struct rpc_task *task, void *calldata)
                } else {
                        LIST_HEAD(head);
 
+                       /*
+                        * Mark the bad layout state as invalid, then retry
+                        * with the current stateid.
+                        */
                        pnfs_mark_matching_lsegs_invalid(lo, &head, NULL);
                        spin_unlock(&inode->i_lock);
-                       /* Mark the bad layout state as invalid, then
-                        * retry using the open stateid. */
                        pnfs_free_lseg_list(&head);
+       
+                       task->tk_status = 0;
+                       rpc_restart_call_prepare(task);
                }
        }
        if (nfs4_async_handle_error(task, server, state) == -EAGAIN)