linux-kernel - Re: [PATCH v2] cpufreq: Support per-policy performance boost

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] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAJZ5v0gNZ_Uw4fLH00tzPtvY3GWu2qBAvQ=y2Bjwreqa=3kuuQ@mail.gmail.com>
Date:   Wed, 23 Aug 2023 21:52:43 +0200
From:   "Rafael J. Wysocki" <rafael@...nel.org>
To:     Jie Zhan <zhanjie9@...ilicon.com>, viresh.kumar@...aro.org
Cc:     linux-pm@...r.kernel.org, linux-kernel@...r.kernel.org,
        linuxarm@...wei.com, xuwei5@...wei.com, wanghuiqiang@...wei.com,
        jonathan.cameron@...wei.com, wangxiongfeng2@...wei.com
Subject: Re: [PATCH v2] cpufreq: Support per-policy performance boost

On Tue, Aug 22, 2023 at 2:51 PM Jie Zhan <zhanjie9@...ilicon.com> wrote:
>
> The boost control currently applies to the whole system.  However, users
> may prefer to boost a subset of cores in order to provide prioritized
> performance to workloads running on the boosted cores.
>
> Enable per-policy boost by adding a 'boost' sysfs interface under each
> policy path.  This can be found at:
>
>         /sys/devices/system/cpu/cpufreq/policy<*>/boost
>
> Same to the global boost switch, writing 1/0 to the per-policy 'boost'
> enables/disables boost on a cpufreq policy respectively.
>
> The user view of global and per-policy boost controls should be:
>
> 1. Enabling global boost initially enables boost on all policies, and
> per-policy boost can then be enabled or disabled individually, given that
> the platform does support so.
>
> 2. Disabling global boost makes the per-policy boost interface illegal.
>
> Signed-off-by: Jie Zhan <zhanjie9@...ilicon.com>
> Reviewed-by: Wei Xu <xuwei5@...ilicon.com>

Viresh, you had comments on the previous version of this.  Does it
look better now?

> ---
> A possible question could be: why not just limiting 'scaling_max_freq'?
> Well, the fundamental difference is that per-policy boost could be more
> user-friendly.  When global boost is enabled, it is not straightforward
> to figure out the base frequency for setting 'scaling_max_freq' to a
> non-boost value. Also, this is supposed to take effect on the physical
> upper frequency limit, reflected through 'cpuinfo_max_freq'.
>
> v1->v2:
> - Rename the interface from 'local_boost' to 'boost'.
> - Illegalize writing 0 to per-policy even if global boost is off.
> - Show the per-policy 'boost' file only when ->set_boost() is available.
>
> v1: https://lore.kernel.org/linux-pm/20230724075827.4160512-1-zhanjie9@hisilicon.com/
>
>  drivers/cpufreq/cpufreq.c | 43 +++++++++++++++++++++++++++++++++++++++
>  include/linux/cpufreq.h   |  3 +++
>  2 files changed, 46 insertions(+)
>
> diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
> index 50bbc969ffe5..81e21fa070e1 100644
> --- a/drivers/cpufreq/cpufreq.c
> +++ b/drivers/cpufreq/cpufreq.c
> @@ -86,6 +86,7 @@ static void cpufreq_governor_limits(struct cpufreq_policy *policy);
>  static int cpufreq_set_policy(struct cpufreq_policy *policy,
>                               struct cpufreq_governor *new_gov,
>                               unsigned int new_pol);
> +static bool cpufreq_boost_supported(void);
>
>  /*
>   * Two notifier lists: the "policy" list is involved in the
> @@ -621,6 +622,40 @@ static ssize_t store_boost(struct kobject *kobj, struct kobj_attribute *attr,
>  }
>  define_one_global_rw(boost);
>
> +static ssize_t show_local_boost(struct cpufreq_policy *policy, char *buf)
> +{
> +       return sysfs_emit(buf, "%d\n", policy->boost_enabled);
> +}
> +
> +static ssize_t store_local_boost(struct cpufreq_policy *policy,
> +                                const char *buf, size_t count)
> +{
> +       int ret, enable;
> +
> +       ret = kstrtoint(buf, 10, &enable);
> +       if (ret || enable < 0 || enable > 1)
> +               return -EINVAL;
> +
> +       if (!cpufreq_driver->boost_enabled)
> +               return -EINVAL;
> +
> +       if (policy->boost_enabled == enable)
> +               return count;
> +
> +       cpus_read_lock();
> +       ret = cpufreq_driver->set_boost(policy, enable);
> +       cpus_read_unlock();
> +
> +       if (ret)
> +               return ret;
> +
> +       policy->boost_enabled = enable;
> +
> +       return count;
> +}
> +
> +static struct freq_attr local_boost = __ATTR(boost, 0644, show_local_boost, store_local_boost);
> +
>  static struct cpufreq_governor *find_governor(const char *str_governor)
>  {
>         struct cpufreq_governor *t;
> @@ -1055,6 +1090,12 @@ static int cpufreq_add_dev_interface(struct cpufreq_policy *policy)
>                         return ret;
>         }
>
> +       if (cpufreq_boost_supported()) {
> +               ret = sysfs_create_file(&policy->kobj, &local_boost.attr);
> +               if (ret)
> +                       return ret;
> +       }
> +
>         return 0;
>  }
>
> @@ -2716,6 +2757,8 @@ int cpufreq_boost_trigger_state(int state)
>                 ret = cpufreq_driver->set_boost(policy, state);
>                 if (ret)
>                         goto err_reset_state;
> +
> +               policy->boost_enabled = state;
>         }
>         cpus_read_unlock();
>
> diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h
> index 172ff51c1b2a..fa02f2fa88c4 100644
> --- a/include/linux/cpufreq.h
> +++ b/include/linux/cpufreq.h
> @@ -140,6 +140,9 @@ struct cpufreq_policy {
>          */
>         bool                    dvfs_possible_from_any_cpu;
>
> +       /* Per policy boost enabled flag. */
> +       bool                    boost_enabled;
> +
>          /* Cached frequency lookup from cpufreq_driver_resolve_freq. */
>         unsigned int cached_target_freq;
>         unsigned int cached_resolved_idx;
> --
> 2.30.0
>