linux-kernel - Re: [PATCH] cpufreq: cppc: Use lowest_nonlinear

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <770cdb73-9538-45cc-824a-d5939834b680@huawei.com>
Date: Thu, 5 Feb 2026 19:57:35 +0800
From: "zhangpengjie (A)" <zhangpengjie2@...wei.com>
To: <rafael@...nel.org>, <viresh.kumar@...aro.org>
CC: <linux-pm@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
	<zhanjie9@...ilicon.com>, <zhenglifeng1@...wei.com>, <lihuisong@...wei.com>,
	<yubowen8@...wei.com>, <linhongye@...artners.com>, <linuxarm@...wei.com>,
	<jonathan.cameron@...wei.com>, <wangzhi12@...wei.com>
Subject: Re: [PATCH] cpufreq: cppc: Use lowest_nonlinear_perf as a soft
 minimum limit

Gentle Ping.

Thanks!

On 1/16/2026 5:45 PM, Pengjie Zhang wrote:
> The ACPI 6.5 specification describes "Lowest Performance" as the absolute
> lowest performance level of the platform. It notes:
>
> "Selecting a performance level lower than the lowest nonlinear performance
> level may actually cause an efficiency penalty, but should reduce the
> instantaneous power consumption of the processor. In traditional terms,
> this represents the T-state range of performance levels."
>
> Currently, the cpufreq core initializes the minimum QoS request to 0.
> While this allows userspace to modify limits, it causes the CPU frequency
> to drop below lowest_nonlinear_perf by default. This hurts performance with
> negligible power savings.
>
> To fix this, implement a "soft" limit in cppc_verify_policy(). If the
> policy minimum is still the default (0), it is clamped to
> lowest_nonlinear_perf. However, if userspace has explicitly set a lower
> limit, that configuration takes precedence.
>
> Signed-off-by: Pengjie Zhang <zhangpengjie2@...wei.com>
> ---
>   drivers/cpufreq/cppc_cpufreq.c | 18 ++++++++++++++++--
>   1 file changed, 16 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/cpufreq/cppc_cpufreq.c b/drivers/cpufreq/cppc_cpufreq.c
> index 9eac77c4f294..59aa1721438b 100644
> --- a/drivers/cpufreq/cppc_cpufreq.c
> +++ b/drivers/cpufreq/cppc_cpufreq.c
> @@ -301,9 +301,23 @@ static unsigned int cppc_cpufreq_fast_switch(struct cpufreq_policy *policy,
>   	return target_freq;
>   }
>   
> -static int cppc_verify_policy(struct cpufreq_policy_data *policy)
> +static int cppc_verify_policy(struct cpufreq_policy_data *policy_data)
>   {
> -	cpufreq_verify_within_cpu_limits(policy);
> +	if (policy_data->min == FREQ_QOS_MIN_DEFAULT_VALUE) {
> +		struct cpufreq_policy *policy __free(put_cpufreq_policy) =
> +					      cpufreq_cpu_get(policy_data->cpu);
> +		struct cppc_cpudata *cpu_data;
> +
> +		if (!policy)
> +			return -EINVAL;
> +
> +		cpu_data = policy->driver_data;
> +		policy_data->min = cppc_perf_to_khz(&cpu_data->perf_caps,
> +			cpu_data->perf_caps.lowest_nonlinear_perf);
> +	}
> +
> +	cpufreq_verify_within_cpu_limits(policy_data);
> +
>   	return 0;
>   }
>