Re: [PATCH, RFC] reimplement flush_workqueue()

[Srivatsa Vaddagiri <vatsa@in.ibm.com>]

On Thu, Jan 04, 2007 at 09:18:50AM -0800, Andrew Morton wrote:
> This?

This can still lead to the problem spotted by Oleg here:

	http://lkml.org/lkml/2006/12/30/37

and you would need a similar patch he posted there.

>  void fastcall flush_workqueue(struct workqueue_struct *wq)
>  {
> -	mutex_lock(&workqueue_mutex);
> +	preempt_disable();		/* CPU hotplug */
>  	if (is_single_threaded(wq)) {
>  		/* Always use first cpu's area. */
>  		flush_cpu_workqueue(per_cpu_ptr(wq->cpu_wq, singlethread_cpu));
> @@ -459,7 +463,7 @@ void fastcall flush_workqueue(struct wor
>  		for_each_online_cpu(cpu)
>  			flush_cpu_workqueue(per_cpu_ptr(wq->cpu_wq, cpu));
>  	}
> -	mutex_unlock(&workqueue_mutex);
> +	preempt_enable();
>  }
>  EXPORT_SYMBOL_GPL(flush_workqueue);

-- 
Regards,
vatsa