Re: [RFC PATCH 1/2] thermal/cpufreq_cooling: remove unused cpu_idx in get_load()

[Lukasz Luba]

On 3/23/26 10:41, Viresh Kumar wrote:
> On 23-03-26, 09:20, Lukasz Luba wrote:
> > Thanks for monitoring the development (it's always good
> > to have extra engineer opinion)!
> >
> > I've checked the commit that you referred to and the 'i++' there.
> > It's safe. That commit removed the heavy operation for only
> > tracing purpose, namely:
> > - allocate buffer for N CPUs for 'load_cpu' pointer
> > - populate CPUs' load from the idle fwk
> > - put that info into the trace
> > - free the 'load_cpu' buffer
> >
> > That has been redesigned since it was just for tracing
> > and introducing extra time spent for code run in the
> > throttling phase.
> >
> > The code in get_load() is OK with the commit that you
> > mentioned.
>
> The code
>
>          load = get_load(cpufreq_cdev, cpu, i);
>
> depends on `i` being incremented in the loop to get the correct
> `cpu_idx`. But the said commit removed it and left `i` to be set to 0
> for ever.
>
> How is that okay ? What am I missing ?

Right, there is a mix of two things.
The 'i' left but should be removed as well, since
this is !SMP code with only 1 cpu and i=0.

The whole split which has been made for getting
the load or utilization from CPU(s) needs to be
cleaned. The compiled code looks different since
it knows there is non-SMP config used.

Do you want to clean that or I should do this?