Re: [PATCH 0/2] workqueue lockdep limitations/bugs

Hello, Johannes.

On Tue, Aug 21, 2018 at 02:03:15PM +0200, Johannes Berg wrote:

3)
The third problem is more complicated, but it means that we likely miss a
lot of potential locking problems with ordered workqueues, and I think we
should really try to fix this somehow.

Let's say we again have an ordered workqueue, and the following:

work1_function
{
	mutex_lock(&mutex);
}

Regular mutexes complain when the locker isn't the unlocker already.
Do we really care about this case?

Thanks.

-- 
tejun