On Sun, 01 Jun 2025 21:22:59 +0200,
"Ted Unangst" <tedu@tedunangst.com> wrote:
> 
> This introduces a helper function for hw.perfpolicy that allows more
> machine specific policies to be set.
> 
> On my laptop, there's also a fan policy. This affects performance, but
> it's not really related to the existing hw.setperf mechanism. For 
> example, a "silent" fan setting and "high" CPU setting is reasonable.
> Exactly what I want, in fact.
> 
> Fortunately, perfpolicy takes a string, which makes it flexible.
> This uses two new optional functions to parse and append the policy.
> 
> hw.perfpolicy=silent,auto
>

It looks like as a very good direction to support things like Intel Hardware
P-state via dedicated driver.

I like it!

> @@ -671,18 +673,22 @@ sysctl_hwperfpolicy(void *oldp, size_t *
>  	if (!cpu_setperf)
>  		return EOPNOTSUPP;
>  
> +	policy[0] = 0;
> +	if (cpu_getpolicy)
> +		cpu_getpolicy(policy, sizeof(policy));
> +
>  	switch (current_perfpolicy()) {
>  	case PERFPOL_MANUAL:
> -		strlcpy(policy, "manual", sizeof(policy));
> +		strlcat(policy, "manual", sizeof(policy));
>  		break;
>  	case PERFPOL_AUTO:
> -		strlcpy(policy, "auto", sizeof(policy));
> +		strlcat(policy, "auto", sizeof(policy));
>  		break;
>  	case PERFPOL_HIGH:
> -		strlcpy(policy, "high", sizeof(policy));
> +		strlcat(policy, "high", sizeof(policy));
>  		break;
>  	default:
> -		strlcpy(policy, "unknown", sizeof(policy));
> +		strlcat(policy, "unknown", sizeof(policy));
>  		break;
>  	}
>

I haven't tried it but after reading the diff I think that after

sysctl hw.perfpolicy=silent,auto

you'll have result of sysctl hw.perfpolicy as:
 - auto on AC
 - unknown on battery

-- 
wbr, Kirill