Re: [PATCH 1/2] workqueue: rescuer_thread() processes all pwqs before exit

[Tejun Heo <tj@kernel.org>]

Hello, Lai.

On Fri, Apr 18, 2014 at 12:04:29AM +0800, Lai Jiangshan wrote:
> > I don't think this is reliable.  What if mayday requests take place
> > between wq_mayday_lock and kthread_should_stop() check?  We'll
> > probably need to run through mayday list after checking should_stop.
> 
> It is destroy_workqueue()'s responsibility to avoid this.
> destroy_workqueue() should drain all works and refuse any new work queued
> on the wq before destroy the wq.
> 
> So since there is no works, there is no new mayday request,
> and there is no mayday request take place between wq_mayday_lock
> and kthread_should_stop() check.

Hmmm?  Isn't this the same race condition that you tried to remove by
relocating the test?  It doesn't matter what destroy_workqueue() does,
the rescuer may get preempted inbetween and anything can happen
inbetween including someone maydaying and initiation of
destroy_workqueue().  Your patch doesn't change the situation at all.
It can still return with non-empty mayday list.

Thanks.

-- 
tejun