From: Pierre Gondois <pierre.gondois@arm.com>
To: Xueqin Luo <luoxueqin@kylinos.cn>, lkp@intel.com
Cc: linux-kernel@vger.kernel.org, linux-pm@vger.kernel.org,
llvm@lists.linux.dev, oe-kbuild-all@lists.linux.dev,
rafael@kernel.org, sumitg@nvidia.com, viresh.kumar@linaro.org,
zhanjie9@hisilicon.com, zhenglifeng1@huawei.com
Subject: Re: [PATCH v3] cpufreq: cppc: Add update_limits support for Highest Performance changes
Date: Mon, 11 May 2026 18:16:03 +0200 [thread overview]
Message-ID: <b63b1409-3c01-418b-be27-f7e273b2e73a@arm.com> (raw)
In-Reply-To: <20260511075647.206642-1-luoxueqin@kylinos.cn>
Hello Xueqin,
On 5/11/26 09:56, Xueqin Luo wrote:
> ACPI CPPC specification requires OSPM to re-evaluate the Highest
> Performance register when Notify(0x85) is received for a processor
> device.
>
> Implement cppc_cpufreq_update_limits() to refresh the cached
> highest_perf capability through cppc_get_highest_perf() and update
> policy->cpuinfo.max_freq accordingly.
>
> When autonomous selection mode is enabled, reprogram the runtime
> MIN_PERF/MAX_PERF envelope against the updated Highest Performance
> capability through cppc_cpufreq_set_autonomous_perf().
>
> --
> change v2->v3:
> - Fix build issues reported by kernel test robot
> - Fold cppc_cpufreq_set_autonomous_perf() refactoring into
> this patch to keep it self-contained
>
> Reported-by: kernel test robot <lkp@intel.com>
> Closes: https://lore.kernel.org/oe-kbuild-all/202605101404.fqz0MXIe-lkp@intel.com/
> Signed-off-by: Xueqin Luo <luoxueqin@kylinos.cn>
> ---
> drivers/cpufreq/cppc_cpufreq.c | 89 +++++++++++++++++++++++++++++-----
> 1 file changed, 76 insertions(+), 13 deletions(-)
>
> diff --git a/drivers/cpufreq/cppc_cpufreq.c b/drivers/cpufreq/cppc_cpufreq.c
> index 7e7f9dfb7a24..9184d5570de8 100644
> --- a/drivers/cpufreq/cppc_cpufreq.c
> +++ b/drivers/cpufreq/cppc_cpufreq.c
> @@ -843,6 +843,80 @@ static int cppc_cpufreq_set_boost(struct cpufreq_policy *policy, int state)
> return 0;
> }
>
> +/**
> + * cppc_cpufreq_set_autonomous_perf - Configure performance bounds for
> + * autonomous mode
> + * @policy: cpufreq policy structure
> + *
> + * When enabling autonomous selection, program MIN_PERF and MAX_PERF from
> + * current policy limits so that the platform uses the correct performance
> + * bounds immediately.
> + *
> + * Return: 0 on success, negative error code on failure.
> + */
> +static int cppc_cpufreq_set_autonomous_perf(struct cpufreq_policy *policy)
> +{
> + struct cppc_cpudata *cpu_data = policy->driver_data;
> + u32 old_min_perf = cpu_data->perf_ctrls.min_perf;
> + u32 old_max_perf = cpu_data->perf_ctrls.max_perf;
> + int ret;
> +
> + cppc_cpufreq_update_perf_limits(cpu_data, policy);
> +
> + ret = cppc_set_perf(policy->cpu, &cpu_data->perf_ctrls);
> + if (ret) {
> + cpu_data->perf_ctrls.min_perf = old_min_perf;
> + cpu_data->perf_ctrls.max_perf = old_max_perf;
> + return ret;
> + }
> +
> + return 0;
> +}
> +
> +static void cppc_cpufreq_update_limits(struct cpufreq_policy *policy)
> +{
> + struct cppc_cpudata *cpu_data = policy->driver_data;
> + u64 prev_highest_perf;
> + u64 highest_perf;
> + u64 nominal_perf;
> + int ret;
> +
> + guard(cpufreq_policy_write)(policy);
> +
> + prev_highest_perf = cpu_data->perf_caps.highest_perf;
> +
> + ret = cppc_get_highest_perf(policy->cpu, &highest_perf);
> + if (ret)
> + return;
> +
> + if (highest_perf == prev_highest_perf)
> + return;
> +
> + cpu_data->perf_caps.highest_perf = highest_perf;
> + nominal_perf = min_t(u64, highest_perf,
> + cpu_data->perf_caps.nominal_perf);
> +
I sent a patch to try to remove setting policy->max in drivers
in cpufreq drivers in general [1]. IMO policy->max should be
updated through a callback handler, maybe:
refresh_frequency_limits()
\-cpufreq_set_policy()
[1] https://lore.kernel.org/all/20260511135538.522653-4-pierre.gondois@arm.com/
> + policy->max = cppc_perf_to_khz(&cpu_data->perf_caps,
> + policy->boost_enabled ?
> + highest_perf : nominal_perf);
> +
There might be other things to check:
- Is the new highest_perf higher/equal than nominal_perf ?
Depending on the answer the boost capabilities/status will have to
be updated.
- IIUC it will be possible to update the nominal_perf in the next spec,
so this makes things a bit more complicated aswell.
https://lore.kernel.org/all/20260430142430.755437-3-sumitg@nvidia.com/
- Do we need to rebuild the whole topology ? Cf. highest_perf
is used as the CPU capacity in topology_init_cpu_capacity_cppc().
(maybe off topic)
- What is the relation between highest_perf and the (not yet introduced [1])
highest_freq value ? Since the highest_perf can evolve over time there
is no clear correlation between them IIUC.
https://mantis.uefi.org/mantis/view.php?id=2570
Some of the questions are more spec. related,
but this patch might be the right place to discuss them.
> + policy->cpuinfo.max_freq = policy->max;
> +
> + /*
> + * Autonomous selection mode uses MIN/MAX performance as runtime
> + * hardware control bounds.
> + *
> + * Re-program them when highest_perf changes.
> + */
> + if (cpu_data->perf_ctrls.auto_sel)
> + cppc_cpufreq_set_autonomous_perf(policy);
> +
> + pr_debug("CPU%d: highest_perf updated %llu -> %llu\n",
> + policy->cpu,
> + prev_highest_perf,
> + highest_perf);
> +}
> +
> static ssize_t show_freqdomain_cpus(struct cpufreq_policy *policy, char *buf)
> {
> struct cppc_cpudata *cpu_data = policy->driver_data;
> @@ -885,20 +959,8 @@ static ssize_t store_auto_select(struct cpufreq_policy *policy,
> cpu_data->perf_ctrls.auto_sel = val;
>
> if (val) {
> - u32 old_min_perf = cpu_data->perf_ctrls.min_perf;
> - u32 old_max_perf = cpu_data->perf_ctrls.max_perf;
> -
> - /*
> - * When enabling autonomous selection, program MIN_PERF and
> - * MAX_PERF from current policy limits so that the platform
> - * uses the correct performance bounds immediately.
> - */
> - cppc_cpufreq_update_perf_limits(cpu_data, policy);
> -
> - ret = cppc_set_perf(policy->cpu, &cpu_data->perf_ctrls);
> + ret = cppc_cpufreq_set_autonomous_perf(policy);
> if (ret) {
> - cpu_data->perf_ctrls.min_perf = old_min_perf;
> - cpu_data->perf_ctrls.max_perf = old_max_perf;
> cppc_set_auto_sel(policy->cpu, false);
> cpu_data->perf_ctrls.auto_sel = false;
> return ret;
> @@ -1009,6 +1071,7 @@ static struct cpufreq_driver cppc_cpufreq_driver = {
> .init = cppc_cpufreq_cpu_init,
> .exit = cppc_cpufreq_cpu_exit,
> .set_boost = cppc_cpufreq_set_boost,
> + .update_limits = cppc_cpufreq_update_limits,
> .attr = cppc_cpufreq_attr,
> .name = "cppc_cpufreq",
> };
next prev parent reply other threads:[~2026-05-11 16:17 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <20260508111932.185886-3-luoxueqin@kylinos.cn>
2026-05-10 6:14 ` [PATCH] cpufreq: cppc: Add update_limits support for Highest Performance changes kernel test robot
2026-05-11 7:56 ` [PATCH v3] " Xueqin Luo
2026-05-11 16:16 ` Pierre Gondois [this message]
2026-05-11 16:43 ` Rafael J. Wysocki
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=b63b1409-3c01-418b-be27-f7e273b2e73a@arm.com \
--to=pierre.gondois@arm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-pm@vger.kernel.org \
--cc=lkp@intel.com \
--cc=llvm@lists.linux.dev \
--cc=luoxueqin@kylinos.cn \
--cc=oe-kbuild-all@lists.linux.dev \
--cc=rafael@kernel.org \
--cc=sumitg@nvidia.com \
--cc=viresh.kumar@linaro.org \
--cc=zhanjie9@hisilicon.com \
--cc=zhenglifeng1@huawei.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox