Re: [PATCH V4] sched/cpufreq: initialize iowait_boost_max and iowait_boost with cpu capacity
From: Quentin Perret
Date: Fri Feb 22 2019 - 06:00:08 EST
On Friday 22 Feb 2019 at 18:37:46 (+0800), Chunyan Zhang wrote:
> @@ -823,6 +823,8 @@ static int sugov_start(struct cpufreq_policy *policy)
> {
> struct sugov_policy *sg_policy = policy->governor_data;
> unsigned int cpu;
> + unsigned long max_cap = arch_scale_cpu_capacity(NULL, policy->cpu);
> + unsigned long min_cap = max_cap * policy->min / policy->cpuinfo.max_freq;
>
> sg_policy->freq_update_delay_ns = sg_policy->tunables->rate_limit_us * NSEC_PER_USEC;
> sg_policy->last_freq_update_time = 0;
> @@ -837,7 +839,9 @@ static int sugov_start(struct cpufreq_policy *policy)
> memset(sg_cpu, 0, sizeof(*sg_cpu));
> sg_cpu->cpu = cpu;
> sg_cpu->sg_policy = sg_policy;
> - sg_cpu->iowait_boost_max = policy->cpuinfo.max_freq;
> + sg_cpu->max = max_cap;
> + sg_cpu->min = min_cap;
> + sg_cpu->iowait_boost_max = max_cap;
Unfortunately, I don't think you can do that only here. The return value
of arch_scale_cpu_capacity() can change at run time. And it does on arm64,
see drivers/base/arch_topology.c.
> }
>
> for_each_cpu(cpu, policy->cpus) {
> --
> 2.17.1
>
Thanks,
Quentin