ocfs2: delay migration when the lockres is in migration state

We trigger a bug in __dlm_lockres_reserve_ast() when we parallel umount 4
nodes.  The situation is as follows:

1) Node A migrate all lockres it owned(eg.  lockres A) to other nodes
   say node B when it umounts.

2) Receiving MIG_LOCKRES message from A, Node B masters the lockres A
   with DLM_LOCK_RES_MIGRATING state set.

3) Then we umount ocfs2 on node B.  It also should migrate lockres A to
   another node, say node C.  But now, DLM_LOCK_RES_MIGRATING state of
   lockers A is not cleared.  Node B triggered the BUG on lockres with
   state DLM_LOCK_RES_MIGRATING.

Signed-off-by: Xuejiufei <xuejiufei@huawei.com>
Signed-off-by: Joel Becker <jlbec@evilplan.org>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Tariq Saeed <tariq.x.saeed@oracle.com>
Cc: Srinivas Eeda <srinivas.eeda@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
index cf0f103..5a0ef78 100644 (file)
--- a/fs/ocfs2/dlm/dlmmaster.c
+++ b/fs/ocfs2/dlm/dlmmaster.c
@@ -2354,6 +2354,10 @@ static int dlm_is_lockres_migrateable(struct dlm_ctxt *dlm,
 
        assert_spin_locked(&res->spinlock);
 
+       /* delay migration when the lockres is in MIGRATING state */
+       if (res->state & DLM_LOCK_RES_MIGRATING)
+               return 0;
+
        if (res->owner != dlm->node_num)
                return 0;