Re: [PATCH] cpufreq: CPPC: Program policy limits during initialization

From: Rafael J. Wysocki (Intel)

Date: Fri Nov 13 2026 - 11:03:00 EST


On Thu, Nov 12, 2026 at 12:32 PM Christian Loehle
<christian.loehle@xxxxxxx> wrote:
>
> cppc_cpufreq_cpu_init() keeps valid firmware MIN/MAX limits and relies on
> the governor to update them. If its first target equals policy->cur, the
> core can skip the driver callback, allowing autonomous selection below
> scaling_min_freq.
>
> Program MIN from Lowest Nonlinear and MAX from Nominal or Highest according
> to the boost state before starting the governor. Set Desired to MAX and
> publish policy->cur only after the write succeeds. Leave policy->max unset
> so the initial maximum QoS request does not restrict later boost use.
>
> Reported-by: Sashiko <sashiko-bot@xxxxxxxxxx>
> Link: https://sashiko.dev/#/patchset/20261106193304.3347374-1-sumitg@xxxxxxxxxx
> Fixes: ea3db45ae476 ("cpufreq: cppc: Update MIN_PERF/MAX_PERF in target callbacks")
> Link: https://lore.kernel.org/all/CAJZ5v0jQVRQhJzLoAy3btKovHGYkqGKJAP_5VfENM_WAREQnyQ@xxxxxxxxxxxxxx/
> Signed-off-by: Christian Loehle <christian.loehle@xxxxxxx>
> ---
> drivers/cpufreq/cppc_cpufreq.c | 34 +++++++++++++---------------------
> 1 file changed, 13 insertions(+), 21 deletions(-)
>
> diff --git a/drivers/cpufreq/cppc_cpufreq.c b/drivers/cpufreq/cppc_cpufreq.c
> index a62589c0674f..92a5bb28571c 100644
> --- a/drivers/cpufreq/cppc_cpufreq.c
> +++ b/drivers/cpufreq/cppc_cpufreq.c
> @@ -1018,29 +1018,20 @@ static int cppc_cpufreq_cpu_init(struct cpufreq_policy *policy)
> policy->boost_supported = true;
>
> /*
> - * The request starts from the live limits, which may be unusable.
> - * Firmware may leave them zeroed. An interrupted earlier update may
> - * have left MIN above MAX. cppc_set_perf() would reject either, so
> - * request the full range instead.
> - */
> - if (!cpu_data->perf_ctrls.max_perf ||
> - cpu_data->perf_ctrls.min_perf > cpu_data->perf_ctrls.max_perf) {
> - cpu_data->perf_ctrls.min_perf = caps->lowest_perf;
> - cpu_data->perf_ctrls.min_perf_valid = true;
> - cpu_data->perf_ctrls.max_perf = caps->highest_perf;
> - }
> -
> - /*
> - * Keep the initial request within controls left by an earlier driver
> - * instance or firmware. The governor will update the full tuple later.
> + * Program the initial policy bounds before starting the governor. Its
> + * first target request may be skipped if it already matches policy->cur.
> + * cppc_set_perf() orders the writes against the live firmware limits.
> + *
> + * New policies start with boost disabled, so this matches the initial
> + * cpuinfo.max_freq. The core applies global boost after governor startup
> + * and handles any resulting limit change through QoS.
> */
> - desired_perf = caps->highest_perf;
> - if (cpu_data->perf_ctrls.max_perf)
> - desired_perf = min(desired_perf, cpu_data->perf_ctrls.max_perf);
> - if (cpu_data->perf_ctrls.min_perf_valid)
> - desired_perf = max(desired_perf, cpu_data->perf_ctrls.min_perf);
> + cpu_data->perf_ctrls.min_perf = caps->lowest_nonlinear_perf;
> + cpu_data->perf_ctrls.min_perf_valid = true;
> + cpu_data->perf_ctrls.max_perf = policy->boost_enabled ?
> + caps->highest_perf : caps->nominal_perf;
>
> - policy->cur = cppc_perf_to_khz(caps, desired_perf);
> + desired_perf = cpu_data->perf_ctrls.max_perf;
> cpu_data->perf_ctrls.desired_perf = desired_perf;
>
> cppc_cpufreq_policy_state(policy)->suspend_regs_handled = false;
> @@ -1057,6 +1048,7 @@ static int cppc_cpufreq_cpu_init(struct cpufreq_policy *policy)
> goto out;
> }
> cpu_data->perf_limits_valid = true;
> + policy->cur = cppc_perf_to_khz(caps, desired_perf);
>
> cppc_cpufreq_cpu_fie_init(policy);
> return 0;
> --

Applied as 7.5 material, thanks!