Re: [PATCH v7 4/4] workqueue: Unbind kworkers before sending them to exit()

[Valentin Schneider]

On 11/01/23 06:55, Tejun Heo wrote:
> On Wed, Jan 11, 2023 at 12:49:49PM +0000, Valentin Schneider wrote:
>> While we're here, for my own education I was trying to figure out in what
>> scenarios we can hit this manager-already-active condition. When sending
>> out v6 I had convinced myself it could happen during failed
>> initialization of a new unbound pool, but having another look at it now I'm
>> not so sure anymore.
>> 
>> The only scenario I can think of now is around maybe_create_worker()'s
>> release of pool->lock, as that implies another worker can drain the
>> pool->worklist and thus let pool->refcnt reach 0 while another worker is
>> being the pool manager. Am I looking at the right thing?
>
> To be frank, I'm not sure and can't remember why the code is like that off
> the top of my head. It could well be that I was just habitually thinking
> that MANAGER can be contended while in practice the scenario can never
> happen in this particular case. I'll need to look harder at it but maybe we
> can leave that to another day?
>

For sure :-)

> Thanks.
>
> -- 
> tejun