Re: [RFC] workqueue: fix detecting reentrance

[Michal Hocko <mhocko@kernel.org>]

Cc maintainers.

On Thu 12-12-19 17:15:48, Hillf Danton wrote:
> 
> If we can find a valid worker that is serving the specified work at
> the moment then the worker itself is enough to ensure reentrance and
> workqueue doesn't matter here because it is permitted for a work to
> requeue itself either on different cpu or numa node or even on another
> workqueue.

I do not follow what is the actual problem you are trying to fix here.

> Fixes: c9178087acd7 ("workqueue: perform non-reentrancy test when queueing to unbound workqueues too")
> Fixes: 18aa9effad4a ("workqueue: implement WQ_NON_REENTRANT")
> Signed-off-by: Hillf Danton <hdanton@sina.com>
> ---
> 
> --- f/kernel/workqueue.c
> +++ g/kernel/workqueue.c
> @@ -1433,7 +1433,7 @@ retry:
>  
>  		worker = find_worker_executing_work(last_pool, work);
>  
> -		if (worker && worker->current_pwq->wq == wq) {
> +		if (worker) {
>  			pwq = worker->current_pwq;
>  		} else {
>  			/* meh... not running there, queue here */
> 

-- 
Michal Hocko
SUSE Labs