libceph: unregister request in __map_request failed and nofail == false
authormajianpeng <majianpeng@gmail.com>
Tue, 16 Jul 2013 07:45:48 +0000 (15:45 +0800)
committerSage Weil <sage@inktank.com>
Sat, 10 Aug 2013 00:52:06 +0000 (17:52 -0700)
For nofail == false request, if __map_request failed, the caller does
cleanup work, like releasing the relative pages.  It doesn't make any sense
to retry this request.

CC: stable@vger.kernel.org
Signed-off-by: Jianpeng Ma <majianpeng@gmail.com>
Reviewed-by: Sage Weil <sage@inktank.com>
net/ceph/osd_client.c

index dd47889..dbc0a73 100644 (file)
@@ -2129,6 +2129,8 @@ int ceph_osdc_start_request(struct ceph_osd_client *osdc,
                        dout("osdc_start_request failed map, "
                                " will retry %lld\n", req->r_tid);
                        rc = 0;
+               } else {
+                       __unregister_request(osdc, req);
                }
                goto out_unlock;
        }