linux-kernel - Re: [PATCH v2 2/2] sched/uclamp: Add uclamp_is

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: <CAKfTPtDuZSL4UdCbjkAqy2bnKo_FrqEU33EcuuPwfTS+sO8R3w@mail.gmail.com>
Date: Wed, 19 Feb 2025 17:25:12 +0100
From: Vincent Guittot <vincent.guittot@...aro.org>
To: Xuewen Yan <xuewen.yan@...soc.com>
Cc: mingo@...hat.com, peterz@...radead.org, juri.lelli@...hat.com, 
	dietmar.eggemann@....com, rostedt@...dmis.org, bsegall@...gle.com, 
	mgorman@...e.de, vschneid@...hat.com, christian.loehle@....com, 
	hongyan.xia2@....com, ke.wang@...soc.com, di.shen@...soc.com, 
	xuewen.yan94@...il.com, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 2/2] sched/uclamp: Add uclamp_is_used() check before
 enable it

On Wed, 19 Feb 2025 at 10:40, Xuewen Yan <xuewen.yan@...soc.com> wrote:
>
> Because the static_branch_enable() would get the cpus_read_lock(),
> and sometimes users may frequently set the uclamp value of tasks,
> and this operation would call the static_branch_enable()
> frequently, so add the uclamp_is_used() check to prevent calling
> the cpus_read_lock() frequently.
> And to make the code more concise, add a helper function to encapsulate
> this and use it everywhere we enable sched_uclamp_used.
>
> Signed-off-by: Xuewen Yan <xuewen.yan@...soc.com>

Apart the small comment below

Reviewed-by: Vincent Guittot <vincent.guittot@...aro.org>

> ---
> V2:
> - change some commit message;
> - Add a helper function and use it everywhere when enable uclamp (Vincent)
> ---
> ---
>  kernel/sched/core.c     |  6 +++---
>  kernel/sched/sched.h    | 14 ++++++++++++++
>  kernel/sched/syscalls.c |  2 +-
>  3 files changed, 18 insertions(+), 4 deletions(-)
>
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 38a7192bfc19..0466a2f61b99 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -1942,12 +1942,12 @@ static int sysctl_sched_uclamp_handler(const struct ctl_table *table, int write,
>         }
>
>         if (update_root_tg) {
> -               static_branch_enable(&sched_uclamp_used);
> +               sched_uclamp_enable();
>                 uclamp_update_root_tg();
>         }
>
>         if (old_min_rt != sysctl_sched_uclamp_util_min_rt_default) {
> -               static_branch_enable(&sched_uclamp_used);
> +               sched_uclamp_enable();
>                 uclamp_sync_util_min_rt_default();
>         }
>
> @@ -9295,7 +9295,7 @@ static ssize_t cpu_uclamp_write(struct kernfs_open_file *of, char *buf,
>         if (req.ret)
>                 return req.ret;
>
> -       static_branch_enable(&sched_uclamp_used);
> +       sched_uclamp_enable();
>
>         guard(mutex)(&uclamp_mutex);
>         guard(rcu)();
> diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
> index 3624fdce5536..bd22af347d0b 100644
> --- a/kernel/sched/sched.h
> +++ b/kernel/sched/sched.h
> @@ -3407,6 +3407,18 @@ static inline bool uclamp_is_used(void)
>         return static_branch_likely(&sched_uclamp_used);
>  }
>
> +/*
> + * Enabling static branches would get the cpus_read_lock(),
> + * check whether uclamp_is_used before enable it to avoid always
> + * calling cpus_read_lock(). Because we never disable this
> + * static key once enable it.

Maybe, simply:

Enabling static branches would get the cpus_read_lock(),
check uclamp_is_used before enabling it. There is no race issue
because we never disable this static key once enabled.


> + */
> +static inline void sched_uclamp_enable(void)
> +{
> +       if (!uclamp_is_used())
> +               static_branch_enable(&sched_uclamp_used);
> +}
> +
>  static inline unsigned long uclamp_rq_get(struct rq *rq,
>                                           enum uclamp_id clamp_id)
>  {
> @@ -3486,6 +3498,8 @@ static inline bool uclamp_is_used(void)
>         return false;
>  }
>
> +static inline void sched_uclamp_enable(void) {}
> +
>  static inline unsigned long
>  uclamp_rq_get(struct rq *rq, enum uclamp_id clamp_id)
>  {
> diff --git a/kernel/sched/syscalls.c b/kernel/sched/syscalls.c
> index 456d339be98f..9100a77e9d79 100644
> --- a/kernel/sched/syscalls.c
> +++ b/kernel/sched/syscalls.c
> @@ -368,7 +368,7 @@ static int uclamp_validate(struct task_struct *p,
>          * blocking operation which obviously cannot be done while holding
>          * scheduler locks.
>          */
> -       static_branch_enable(&sched_uclamp_used);
> +       sched_uclamp_enable();
>
>         return 0;
>  }
> --
> 2.25.1
>