ocfs2/dlm: clear refmap bit of recovery lock while doing local recovery cleanup
authorxuejiufei <[email protected]>
Fri, 5 Feb 2016 23:36:47 +0000 (15:36 -0800)
committerLinus Torvalds <[email protected]>
Sat, 6 Feb 2016 02:10:40 +0000 (18:10 -0800)
When recovery master down, dlm_do_local_recovery_cleanup() only remove
the $RECOVERY lock owned by dead node, but do not clear the refmap bit.
Which will make umount thread falling in dead loop migrating $RECOVERY
to the dead node.

Signed-off-by: xuejiufei <[email protected]>
Reviewed-by: Joseph Qi <[email protected]>
Cc: Mark Fasheh <[email protected]>
Cc: Joel Becker <[email protected]>
Cc: Junxiao Bi <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>
fs/ocfs2/dlm/dlmrecovery.c

index c5bdf02c213bd7ad5c086331a1ea8194bde40578..b94a425f0175fa84818d78742659c42293d96dba 100644 (file)
@@ -2367,6 +2367,8 @@ static void dlm_do_local_recovery_cleanup(struct dlm_ctxt *dlm, u8 dead_node)
                                                break;
                                        }
                                }
+                               dlm_lockres_clear_refmap_bit(dlm, res,
+                                               dead_node);
                                spin_unlock(&res->spinlock);
                                continue;
                        }