ocfs2/dlm: clear migration_pending when migration target goes down

xuejiufei · torvalds · commit cc28d6d80f6a · 2015-12-29T17:45:49.000-08:00
We have found a BUG on res-&gt;migration_pending when migrating lock
resources.  The situation is as follows.

dlm_mark_lockres_migration
  res-&gt;migration_pending = 1;
  __dlm_lockres_reserve_ast
  dlm_lockres_release_ast returns with res-&gt;migration_pending remains
      because other threads reserve asts
  wait dlm_migration_can_proceed returns 1
  &gt;&gt;&gt;&gt;&gt;&gt;&gt; o2hb found that target goes down and remove target
          from domain_map
  dlm_migration_can_proceed returns 1
  dlm_mark_lockres_migrating returns -ESHOTDOWN with
      res-&gt;migration_pending still remains.

When reentering dlm_mark_lockres_migrating(), it will trigger the BUG_ON
with res-&gt;migration_pending.  So clear migration_pending when target is
down.

Signed-off-by: Jiufei Xue &lt;xuejiufei@huawei.com&gt;
Reviewed-by: Joseph Qi &lt;joseph.qi@huawei.com&gt;
Cc: Mark Fasheh &lt;mfasheh@suse.de&gt;
Cc: Joel Becker &lt;jlbec@evilplan.org&gt;
Cc: Junxiao Bi &lt;junxiao.bi@oracle.com&gt;
Signed-off-by: Andrew Morton &lt;akpm@linux-foundation.org&gt;
Signed-off-by: Linus Torvalds &lt;torvalds@linux-foundation.org&gt;
diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c
@@ -2843,6 +2843,8 @@ static int dlm_mark_lockres_migrating(struct dlm_ctxt *dlm,
 	res->state &= ~DLM_LOCK_RES_BLOCK_DIRTY;
 	if (!ret)
 		BUG_ON(!(res->state & DLM_LOCK_RES_MIGRATING));
+	else
+		res->migration_pending = 0;
 	spin_unlock(&res->spinlock);
 
 	/*
