Re: [PATCH 0/9] perf: Avoid explicit cpumask var allocation from stack

[Mark Rutland]

On Wed, Apr 03, 2024 at 06:41:23PM +0800, Dawei Li wrote:
> On Tue, Apr 02, 2024 at 03:41:51PM +0100, Mark Rutland wrote:
> > Looking at this case, the only reason we need the mask is because it made the
> > logic a little easier to write. All we really want is to choose some CPU in the
> > intersection of two masks ignoring a specific CPU, and there was no helper
> > function to do that.
> > 
> > We can add a new helper to do that for us, which would avoid redundant work to
> > manipulate the entire mask, and it would make the existing code simpler.  I had
> > a series a few years back to add cpumask_any_and_but():
> > 
> >   https://lore.kernel.org/lkml/1486381132-5610-1-git-send-email-mark.rutland@xxxxxxx/
> 
> Sounds a perfect idea!
> 
> Actually I am re-implementing new series on top of your seven-years-late-yet-still-helpful
> patch, with minor update on it:
> 
> diff --git a/include/linux/cpumask.h b/include/linux/cpumask.h
> index 1c29947db848..121f3ac757ff 100644
> --- a/include/linux/cpumask.h
> +++ b/include/linux/cpumask.h
> @@ -388,6 +388,29 @@ unsigned int cpumask_any_but(const struct cpumask *mask, unsigned int cpu)
>         return i;
>  }
> 
> +/**
> + * cpumask_any_and_but - pick a "random" cpu from *mask1 & *mask2, but not this one.
> + * @mask1: the first input cpumask
> + * @mask2: the second input cpumask
> + * @cpu: the cpu to ignore
> + *
> + * Returns >= nr_cpu_ids if no cpus set.
> + */
> +static inline
> +unsigned int cpumask_any_and_but(const struct cpumask *mask1,
> +                                const struct cpumask *mask2,
> +                                unsigned int cpu)
> +{
> +       unsigned int i;
> +
> +       cpumask_check(cpu);
> +       i = cpumask_first_and(mask1, mask2);
> +       if (i != cpu)
> +               return i;
> +
> +       return cpumask_next_and(cpu, mask1, mask2);
> +}
> +
>  /**
>   * cpumask_nth - get the Nth cpu in a cpumask
>   * @srcp: the cpumask pointer
> 
> Change from your original version:
> 1 Moved to cpumask.h, just like other helpers.
> 2 Return value converted to unsigned int.
> 3 Remove EXPORT_SYMBOL, for obvious reason.

That's exactly how I rebased it locally, so that looks good to me!

> I will respin V2 as a whole as soon as possible.

Great!

Thanks,
Mark.