Re: [PATCH v2] sched/uclamp: Align uclamp and util_est and call before freq update

[Kuyo Chang <kuyo.chang@mediatek.com>]

On Tue, 2025-03-25 at 09:47 +0800, Xuewen Yan wrote:
> When task's uclamp is set, we hope that the CPU frequency
> can increase as quickly as possible when the task is enqueued.
> Because the cpu frequency updating happens during the enqueue_task(),
> so the rq's uclamp needs to be updated before the task is enqueued,
> just like util_est.
> So, aline the uclamp and util_est and call before freq update.
> 
> For sched-delayed tasks, the rq uclamp/util_est should only be
> updated
> when they are enqueued upon being awakened.
> So simply the logic of util_est's enqueue/dequeue check.
> 
> Signed-off-by: Xuewen Yan <xuewen.yan@unisoc.com>
> ---
> v2:
> - simply the util-est's en/dequeue check;
> ---
> Previous discussion:
> https://lore.kernel.org/all/CAB8ipk8pEvOtCm-d0o1rsekwxPWUHk9iBGtt9TLTWW-iWTQKiA@mail.gmail.com/
> https://lore.kernel.org/all/84441660bef0a5e67fd09dc3787178d0276dad31.1740664400.git.hongyan.xia2@arm.com/T/#u
> https://lore.kernel.org/all/CAB8ipk9LpbiUDnbcV6+59+Sa=Ai7tFzO===mpLD3obNdV4=J-A@mail.gmail.com/T/#u
> https://lore.kernel.org/all/aa8baf67-a8ec-4ad8-a6a8-afdcd7036771@arm.com/
> ---
>  kernel/sched/core.c | 17 ++++++++++-------
>  kernel/sched/fair.c |  4 ++--
>  2 files changed, 12 insertions(+), 9 deletions(-)
> 
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 042351c7afce..72fbe2031e54 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -1747,7 +1747,7 @@ static inline void uclamp_rq_dec_id(struct rq
> *rq, struct task_struct *p,
>  	}
>  }
>  
> -static inline void uclamp_rq_inc(struct rq *rq, struct task_struct
> *p)
> +static inline void uclamp_rq_inc(struct rq *rq, struct task_struct
> *p, int flags)
>  {
>  	enum uclamp_id clamp_id;
>  
> @@ -1763,7 +1763,8 @@ static inline void uclamp_rq_inc(struct rq *rq,
> struct task_struct *p)
>  	if (unlikely(!p->sched_class->uclamp_enabled))
>  		return;
>  
> -	if (p->se.sched_delayed)
> +	/* Only inc the delayed task which being woken up. */
> +	if (p->se.sched_delayed && !(flags & ENQUEUE_DELAYED))
>  		return;
>  
>  	for_each_clamp_id(clamp_id)
> @@ -2031,7 +2032,7 @@ static void __init init_uclamp(void)
>  }
>  
>  #else /* !CONFIG_UCLAMP_TASK */
> -static inline void uclamp_rq_inc(struct rq *rq, struct task_struct
> *p) { }
> +static inline void uclamp_rq_inc(struct rq *rq, struct task_struct
> *p, int flags) { }
>  static inline void uclamp_rq_dec(struct rq *rq, struct task_struct
> *p) { }
>  static inline void uclamp_fork(struct task_struct *p) { }
>  static inline void uclamp_post_fork(struct task_struct *p) { }
> @@ -2067,12 +2068,14 @@ void enqueue_task(struct rq *rq, struct
> task_struct *p, int flags)
>  	if (!(flags & ENQUEUE_NOCLOCK))
>  		update_rq_clock(rq);
>  
> -	p->sched_class->enqueue_task(rq, p, flags);
>  	/*
> -	 * Must be after ->enqueue_task() because ENQUEUE_DELAYED
> can clear
> -	 * ->sched_delayed.
> +	 * Can be before ->enqueue_task() because uclamp considers
> the
> +	 * ENQUEUE_DELAYED task before its ->sched_delayed gets
> cleared
> +	 * in ->enqueue_task().
>  	 */
> -	uclamp_rq_inc(rq, p);
> +	uclamp_rq_inc(rq, p, flags);
> +
> +	p->sched_class->enqueue_task(rq, p, flags);
>  
>  	psi_enqueue(p, flags);
>  
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index c798d2795243..c92fee07fb7b 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -6930,7 +6930,7 @@ enqueue_task_fair(struct rq *rq, struct
> task_struct *p, int flags)
>  	 * Let's add the task's estimated utilization to the
> cfs_rq's
>  	 * estimated utilization, before we update schedutil.
>  	 */
> -	if (!(p->se.sched_delayed && (task_on_rq_migrating(p) ||
> (flags & ENQUEUE_RESTORE))))
> +	if (!p->se.sched_delayed || (flags & ENQUEUE_DELAYED))
>  		util_est_enqueue(&rq->cfs, p);
>  
>  	if (flags & ENQUEUE_DELAYED) {
> @@ -7168,7 +7168,7 @@ static int dequeue_entities(struct rq *rq,
> struct sched_entity *se, int flags)
>   */
>  static bool dequeue_task_fair(struct rq *rq, struct task_struct *p,
> int flags)
>  {
> -	if (!(p->se.sched_delayed && (task_on_rq_migrating(p) ||
> (flags & DEQUEUE_SAVE))))
> +	if (!p->se.sched_delayed)
>  		util_est_dequeue(&rq->cfs, p);
>  
>  	util_est_update(&rq->cfs, p, flags & DEQUEUE_SLEEP);


The HW composer detects the frame status and uses uclamp to adjust the
CPU frequency for specific tasks.
After applying this patch, there is a significant improvement in the
performance of the dispatch system.
==================================================================
Test          : FHD(1920x1080) video playback 30fps in 60hz frame rate.
Units         : jank(frame drop)
Interpretation: lower is better
Statistic     : AMean in one minutes
==================================================================
asis        with patch
6-janks     0-janks