Re: [PATCH v2 7/8] sched/schedutil: Add a new tunable to dictate response time

[Qais Yousef]

On 12/08/23 00:23, Qais Yousef wrote:

> +static inline u64 sugov_calc_freq_response_ms(struct sugov_policy *sg_policy)
> +{
> +	int cpu = cpumask_first(sg_policy->policy->cpus);
> +	unsigned long cap = arch_scale_cpu_capacity(cpu);
> +	unsigned int max_freq, sec_max_freq;
> +
> +	max_freq = sg_policy->policy->cpuinfo.max_freq;
> +	sec_max_freq = __resolve_freq(sg_policy->policy,
> +				      max_freq - 1,
> +				      CPUFREQ_RELATION_H);
> +
> +	/*
> +	 * We will request max_freq as soon as util crosses the capacity at
> +	 * second highest frequency. So effectively our response time is the
> +	 * util at which we cross the cap@2nd_highest_freq.
> +	 */
> +	cap = sec_max_freq * cap / max_freq;
> +
> +	return approximate_runtime(cap + 1);

After Linus problem (and more testing) I realize this is not correct. This
value is correct for the biggest core only, for smaller cores I must stretch
time with capacity. I have similar invariance issues that I need to address in
this series and uclamp max aggregation series.