Re: [PATCH 1/1] dm-delay: fix hung task introduced by kthread mode

[Benjamin Marzinski <bmarzins@redhat.com>]

On Tue, Apr 30, 2024 at 06:12:37PM +0200, Joel Colledge wrote:
> On Tue, 30 Apr 2024 at 17:27, Christian Loehle <christian.loehle@arm.com> wrote:
> > On 30/04/2024 15:44, Joel Colledge wrote:
> > > On Tue, 30 Apr 2024 at 16:28, Christian Loehle <christian.loehle@arm.com> wrote:
> > >> Is this an issue for delay > 0 too somehow?
> > >
> > > I believe it is. If there is simply no IO to the delay device, then
> > > nothing will wake the new thread and the same issue will occur.
> >
> > Yes, but might be better to just set_current_state(TASK_INTERRUPTIBLE);
> > instead of the wakeup.
> 
> I'm afraid I don't follow what you mean.
> set_current_state(TASK_INTERRUPTIBLE) would need to be called from the
> worker, but the worker isn't running yet, so it can't do that.
> 
> When preparing this patch, I checked how common kthread_create()
> followed by wake_up_process() is. It is fairly common according to
> this very approximate metric:
> $ grep -r -I -A10 kthread_create | grep -c wake_up_process
> 49
> 
> I just looked through those matches in more detail and noticed that
> there is a kthread_run() macro which does exactly what we need. I will
> change my suggestion to use that instead.
> 
> I will wait for more comments before sending a new version of the patch.

kthread_run() looks like the right solution here. There's no need to
wait for more initialization and the kthread will put itself to sleep
momentarily.  Please send an updated patch.

-Ben

> 
> Best regards,
> Joel