Re: [PATCH] workqueue: not allow recursion run_workqueue

[Oleg Nesterov <oleg@redhat.com>]

On 01/21, Peter Zijlstra wrote:
>
> On Wed, 2009-01-21 at 17:42 +0800, Lai Jiangshan wrote:
> > 1) lockdep will complain when recursion run_workqueue
> > 2) works is not run orderly when recursion run_workqueue
> >
> > 3) BUG!
> >    We use recursion run_workqueue to hidden deadlock when
> >    keventd trying to flush its own queue.
> >
> >    It's bug. When flush_workqueue()(nested in a work callback)returns,
> >    the workqueue is not really flushed, the sequence statement of
> >    this work callback will do some thing bad.
> >
> >    So we should not allow workqueue trying to flush its own queue.
>
> The patch looks good, but I'm utterly failing to comprehend this
> changelog. What exactly can go wrong (other than the obvious too deep
> nest and the fact that lockdep will complain)?

I am confused too.


But the change itself looks good to me, I am only worried if we still
have the callers of flush() from within work->func().

> +	WARN_ON(cwq->thread == current);

probably BUG_ON() is better, we are going to deadlock in this case.

Oleg.