Re: [External] Re: [PATCH] memcg: Don't wait writeback completion when release memcg.

[Julian Sun <sunjunchao@bytedance.com>]

Hi,

On Tue, Aug 26, 2025 at 2:53 AM Tejun Heo <tj@kernel.org> wrote:
>
> Hello,
>
> On Tue, Aug 26, 2025 at 01:45:44AM +0800, Julian Sun wrote:
> ...
> > Sorry for having misunderstood what you meant before. I’m afraid that
> > init_wait_func() cannot work the same way. Because calling
> > init_wait_func() presupposes that we are preparing to wait for an
> > event(like wb_wait_completion()), but waiting for such an event might
> > lead to a hung task.
> >
> > Please correct me if I'm wrong.
>
> Using init_wait_func() does not require someone waiting for it. AFAICS, you
> should be able to do the same thing that you did - allocating the done
> entries individually and freeing them when the done count reaches zero
> without anyone waiting for it. waitq doesn't really make many assumptions
> about how it's used - when you call wake_up() on it, it just walks the
> queued entries and invoke the callbacks there.

Ah, yeah, this sounds great, many thanks for your clarification.  I’ll
send v2 of the patch after testing.
>
> Thanks.
>
> --
> tejun


Thanks,
-- 
Julian Sun <sunjunchao@bytedance.com>