Re: Observation on NOHZ_FULL

[Uladzislau Rezki <urezki@gmail.com>]

On Thu, Feb 15, 2024 at 08:51:54AM +0100, Andrea Righi wrote:
> On Tue, Feb 06, 2024 at 10:44:33AM -0800, Paul E. McKenney wrote:
> ...
> > > 
> > >  - with lazy RCU enabled hrtimer_interrupt() takes like 2x more to
> > >    return, respect to the other cases (is this expected?)
> > 
> > This last is surprising at first glance, but I could be missing
> > something.  Joel, Uladzislau, thoughts?
> 
> On this specific point, I repeated the tests with the latest 6.8
> kernel and now I don't notice this difference anymore.
> 
> hrtimer_interrupt() time is always in the range [4K, 16K) ns, both with
> lazy RCU off and lazy RCU on.
> 
> I don't really know why I was getting different results, probably some
> noise triggered by something running in user-space...?
> 
> In any case I will update my old post with these new results and I
> wouldn't consider this an issue for now.
> 
Thank you for your confirmation, so our results are aligned in terms of
i also do not see the difference and issues with hrtimer_interrupt()
duration time.

Thanks!

--
Uladzislau Rezki