NFS: Don't hold the group lock when calling nfs_release_request()

That can deadlock if this is the last reference since
nfs_page_group_destroy() calls nfs_page_group_sync_on_bit().
Note that even if the page was removed from the subpage list,
the req->wb_head could still be pointing to the old head.

Signed-off-by: Trond Myklebust <[email protected]>

Author: trondmypd

fs/nfs/write.c

@@ -532,9 +532,9 @@ nfs_lock_and_join_requests(struct page *page)
 		} else if (WARN_ON_ONCE(subreq->wb_offset < head->wb_offset ||
 			    ((subreq->wb_offset + subreq->wb_bytes) >
 			     (head->wb_offset + total_bytes)))) {
+			nfs_page_group_unlock(head);
 			nfs_unroll_locks(inode, head, subreq);
 			nfs_unlock_and_release_request(subreq);
-			nfs_page_group_unlock(head);
 			nfs_unlock_and_release_request(head);
 			return ERR_PTR(-EIO);
 		}