workqueue: fix bugs in wq_update_unbound_numa() failure path
authorDaeseok Youn <daeseok.youn@gmail.com>
Wed, 16 Apr 2014 05:32:29 +0000 (14:32 +0900)
committerJiri Slaby <jslaby@suse.cz>
Mon, 9 Jun 2014 13:53:35 +0000 (15:53 +0200)
commit3b3b224978f8426000d0c26ecdde037356f3d873
treed72e5db462cd7836874afad51a4854a651c06825
parent94b7b52bd91e0bb9ed304d7f5de3711defdc42c6
workqueue: fix bugs in wq_update_unbound_numa() failure path

commit 77f300b198f93328c26191b52655ce1b62e202cf upstream.

wq_update_unbound_numa() failure path has the following two bugs.

- alloc_unbound_pwq() is called without holding wq->mutex; however, if
  the allocation fails, it jumps to out_unlock which tries to unlock
  wq->mutex.

- The function should switch to dfl_pwq on failure but didn't do so
  after alloc_unbound_pwq() failure.

Fix it by regrabbing wq->mutex and jumping to use_dfl_pwq on
alloc_unbound_pwq() failure.

Signed-off-by: Daeseok Youn <daeseok.youn@gmail.com>
Acked-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: 4c16bd327c74 ("workqueue: implement NUMA affinity for unbound workqueues")
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
kernel/workqueue.c