Re: [PATCH 1/2] sched: Use do_div() for 64 bit division at power utilization calculation (putil)

[Lukasz Majewski <l.majewski@samsung.com>]

Hi Alex,

> On 05/23/2013 04:34 PM, Lukasz Majewski wrote:
> > Now explicit casting is done when power usage variable (putil) is
> > calculated
> > 
> > Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
> > Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
> > ---
> > This patch was developed on top of the following Alex's repository:
> > https://github.com/alexshi/power-scheduling/commits/power-scheduling
> > ---
> >  kernel/sched/fair.c |    6 ++++--
> >  1 file changed, 4 insertions(+), 2 deletions(-)
> > 
> 
> 
> Thanks for catch this issue. seems use div_u64 is better, and there
> are 2 same bugs. so, could I rewrite the patch like following?

Yes, no problem. 

> ---
> 
> From 9f72c25607351981898d99822f5a66e0ca67a3da Mon Sep 17 00:00:00 2001
> From: Alex Shi <alex.shi@intel.com>
> Date: Wed, 29 May 2013 11:09:39 +0800
> Subject: [PATCH 1/2] sched: fix cast on power utilization calculation
> and use div_u64
> 
> Now explicit casting is done when power usage variable (putil) is
> calculated.
> div_u64 is optimized on u32.
> 
> Signed-off-by: Lukasz Majewski <l.majewski@samsung.com>
> Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
> Signed-off-by: Alex Shi <alex.shi@intel.com>
> ---
>  kernel/sched/fair.c | 14 ++++++++------
>  1 file changed, 8 insertions(+), 6 deletions(-)
> 
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index 09ae48a..3a4917c 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -1504,8 +1504,8 @@ static inline void
> update_rq_runnable_avg(struct rq *rq, int runnable)
> __update_tg_runnable_avg(&rq->avg, &rq->cfs); 
>  	period = rq->avg.runnable_avg_period ?
> rq->avg.runnable_avg_period : 1;
> -	rq->util = (u64)(rq->avg.runnable_avg_sum <<
> SCHED_POWER_SHIFT)
> -				/ period;
> +	rq->util = div_u64(((u64)rq->avg.runnable_avg_sum <<
> SCHED_POWER_SHIFT),
> +				period);
>  }
>  
>  /* Add the load generated by se into cfs_rq's child load-average */
> @@ -3407,8 +3407,8 @@ static int is_sd_full(struct sched_domain *sd,
>  		/* p maybe a new forked task */
>  		putil = FULL_UTIL;
>  	else
> -		putil = (u64)(p->se.avg.runnable_avg_sum <<
> SCHED_POWER_SHIFT)
> -				/ (p->se.avg.runnable_avg_period +
> 1);
> +		putil = div_u64(((u64)p->se.avg.runnable_avg_sum <<
> SCHED_POWER_SHIFT),
> +				p->se.avg.runnable_avg_period + 1);
>  
>  	/* Try to collect the domain's utilization */
>  	group = sd->groups;
> @@ -3463,9 +3463,11 @@ find_leader_cpu(struct sched_group *group,
> struct task_struct *p, int this_cpu, int vacancy, min_vacancy =
> INT_MAX; int leader_cpu = -1;
>  	int i;
> +
>  	/* percentage of the task's util */
> -	unsigned putil = (u64)(p->se.avg.runnable_avg_sum <<
> SCHED_POWER_SHIFT)
> -				/ (p->se.avg.runnable_avg_period +
> 1);
> +	unsigned putil;
> +	putil = div_u64(((u64)p->se.avg.runnable_avg_sum <<
> SCHED_POWER_SHIFT),
> +				p->se.avg.runnable_avg_period + 1);
>  
>  	/* bias toward local cpu */
>  	if (cpumask_test_cpu(this_cpu, tsk_cpus_allowed(p)) &&



-- 
Best regards,

Lukasz Majewski

Samsung R&D Poland (SRPOL) | Linux Platform Group