ceph: force updating the msg pointer in non-split case

commit 4cafd0400bcb6187c0d4ab4d4b0229a89ac4f8c2 upstream.

When the MClientSnap reqeust's op is not CEPH_SNAP_OP_SPLIT the
request may still contain a list of 'split_realms', and we need
to skip it anyway. Or it will be parsed as a corrupt snaptrace.

Cc: stable@vger.kernel.org
Link: https://tracker.ceph.com/issues/61200
Reported-by: Frank Schilder <frans@dtu.dk>
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

diff --git a/fs/ceph/snap.c b/fs/ceph/snap.c
index 97ce1bd13bad1e47de5fbe8e35c889838cc0f969..96c79d5e9f7e0068a701a9dd06eff29ff5c4b713 100644 (file)
--- a/fs/ceph/snap.c
+++ b/fs/ceph/snap.c
@@ -1005,6 +1005,19 @@ void ceph_handle_snap(struct ceph_mds_client *mdsc,
                                continue;
                        adjust_snap_realm_parent(mdsc, child, realm->ino);
                }
+       } else {
+               /*
+                * In the non-split case both 'num_split_inos' and
+                * 'num_split_realms' should be 0, making this a no-op.
+                * However the MDS happens to populate 'split_realms' list
+                * in one of the UPDATE op cases by mistake.
+                *
+                * Skip both lists just in case to ensure that 'p' is
+                * positioned at the start of realm info, as expected by
+                * ceph_update_snap_trace().
+                */
+               p += sizeof(u64) * num_split_inos;
+               p += sizeof(u64) * num_split_realms;
        }
 
        /*