Re: [Question]: questions about 'Use sysidle detection to let the timekeeper sleep'

Re: [Question]: questions about 'Use sysidle detection to let the timekeeper sleep'

[Xiongfeng Wang]

Sorry, I forgot to Cc the maillist again. Cc it in this mail.

On 2021/11/3 10:59, Xiongfeng Wang wrote:
> Hi Frederic,
> 
> Sorry to disturb you. It's just that I came across the same problem as you
> mentioned in the below link.
> https://lwn.net/Articles/577424/
> 
> This is the problem in my situation. All the CPUs on my machine are in the same
> power domain. When nohz_full is enabled, if one of the CPUs keep receiving tick
> interrupts, all the CPUs can not be put into powerdown state, even thought all
> other CPUs are idle.
> 
> I found your first version patchset in the below link.
> https://lore.kernel.org/lkml/1387320692-28460-1-git-send-email-fweisbec@gmail.com/
> But I couldn't found other verions. Even 'full-system-idle' is reverted in the
> mailline in the following commit.
>   commit fe5ac724d81a3c7803e60c2232718f212f3f38d4
>   rcu: Remove nohz_full full-system-idle state machine
> 
> I am not sure if we have any other alternative solutions, or do we have rebased
> the patchset to mainline kernel. The first version patchset I found is based on
> v3.13.
> 
> Thanks,
> Xiongfeng
>