ocfs2/dlm: ignore cleaning the migration mle that is inuse
authorxuejiufei <xuejiufei@huawei.com>
Thu, 14 Jan 2016 23:17:38 +0000 (15:17 -0800)
committerBen Hutchings <ben@decadent.org.uk>
Sat, 13 Feb 2016 10:34:10 +0000 (10:34 +0000)
commitb99688d443e58ba1f1206039de3316bc7d169235
tree255baf25924169c9f4a4bf06427c986c213e4432
parentda7b239bbdcc174c94cc6d0a4b55f9d421508d0c
ocfs2/dlm: ignore cleaning the migration mle that is inuse

commit bef5502de074b6f6fa647b94b73155d675694420 upstream.

We have found that migration source will trigger a BUG that the refcount
of mle is already zero before put when the target is down during
migration.  The situation is as follows:

dlm_migrate_lockres
  dlm_add_migration_mle
  dlm_mark_lockres_migrating
  dlm_get_mle_inuse
  <<<<<< Now the refcount of the mle is 2.
  dlm_send_one_lockres and wait for the target to become the
  new master.
  <<<<<< o2hb detect the target down and clean the migration
  mle. Now the refcount is 1.

dlm_migrate_lockres woken, and put the mle twice when found the target
goes down which trigger the BUG with the following message:

  "ERROR: bad mle: ".

Signed-off-by: Jiufei Xue <xuejiufei@huawei.com>
Reviewed-by: Joseph Qi <joseph.qi@huawei.com>
Cc: Mark Fasheh <mfasheh@suse.de>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
fs/ocfs2/dlm/dlmmaster.c