Acked-by: Sunil Mushran <<a href="mailto:sunil.mushran@gmail.com">sunil.mushran@gmail.com</a>><div><br><br><div class="gmail_quote">On Mon, Aug 13, 2012 at 7:06 PM, Xue jiufei <span dir="ltr"><<a href="mailto:xuejiufei@huawei.com" target="_blank">xuejiufei@huawei.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> We trigger a bug in __dlm_lockres_reserve_ast() when we parallel umount 4 nodes. The situation is as follows:<br>
1) Node A migrate all lockres it owned(eg. lockres A) to other nodes say node B when it umounts.<br>
2) Receiving MIG_LOCKRES message from A, Node B masters the lockres A with DLM_LOCK_RES_MIGRATING state set.<br>
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<br>
of lockers A is not cleared. Node B triggered the BUG on lockres with state DLM_LOCK_RES_MIGRATING.<br>
<br>
Signed-off-by: Xuejiufei <<a href="mailto:xuejiufei@huawei.com">xuejiufei@huawei.com</a>><br>
---<br>
fs/ocfs2/dlm/dlmmaster.c | 4 ++++<br>
1 file changed, 4 insertions(+)<br>
<br>
diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c<br>
index 005261c..6c35406 100644<br>
--- a/fs/ocfs2/dlm/dlmmaster.c<br>
+++ b/fs/ocfs2/dlm/dlmmaster.c<br>
@@ -2357,6 +2357,10 @@ static int dlm_is_lockres_migrateable(struct dlm_ctxt *dlm,<br>
<br>
assert_spin_locked(&res->spinlock);<br>
<br>
+ /* delay migration when the lockres is in MIGRATING state */<br>
+ if (res->state & DLM_LOCK_RES_MIGRATING)<br>
+ return 0;<br>
+<br>
if (res->owner != dlm->node_num)<br>
return 0;<br>
<br>
-- 1.7.9.7<br>
<br>
_______________________________________________<br>
Ocfs2-devel mailing list<br>
<a href="mailto:Ocfs2-devel@oss.oracle.com">Ocfs2-devel@oss.oracle.com</a><br>
<a href="https://oss.oracle.com/mailman/listinfo/ocfs2-devel" target="_blank">https://oss.oracle.com/mailman/listinfo/ocfs2-devel</a><br>
</blockquote></div><br></div>