workqueue: Fix NULL pointer dereference
authorLi Bin <huawei.libin@huawei.com>
Sat, 28 Oct 2017 03:07:28 +0000 (11:07 +0800)
committerTejun Heo <tj@kernel.org>
Mon, 30 Oct 2017 14:56:01 +0000 (07:56 -0700)
commitcef572ad9bd7f85035ba8272e5352040e8be0152
treea324f1e505d3698cd0df3d22898f1c5c5e68370d
parent692b48258dda7c302e777d7d5f4217244478f1f6
workqueue: Fix NULL pointer dereference

When queue_work() is used in irq (not in task context), there is
a potential case that trigger NULL pointer dereference.
----------------------------------------------------------------
worker_thread()
|-spin_lock_irq()
|-process_one_work()
|-worker->current_pwq = pwq
|-spin_unlock_irq()
|-worker->current_func(work)
|-spin_lock_irq()
  |-worker->current_pwq = NULL
|-spin_unlock_irq()

//interrupt here
|-irq_handler
|-__queue_work()
//assuming that the wq is draining
|-is_chained_work(wq)
|-current_wq_worker()
//Here, 'current' is the interrupted worker!
|-current->current_pwq is NULL here!
|-schedule()
----------------------------------------------------------------

Avoid it by checking for task context in current_wq_worker(), and
if not in task context, we shouldn't use the 'current' to check the
condition.

Reported-by: Xiaofei Tan <tanxiaofei@huawei.com>
Signed-off-by: Li Bin <huawei.libin@huawei.com>
Reviewed-by: Lai Jiangshan <jiangshanlai@gmail.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: 8d03ecfe4718 ("workqueue: reimplement is_chained_work() using current_wq_worker()")
Cc: stable@vger.kernel.org # v3.9+
kernel/workqueue_internal.h