Re: [PATCH v2] tg: show the sum wait time of an task group

[çè]

Hi, folks

On 2018/7/4 äå11:27, çè wrote:
> Although we can rely on cpuacct to present the cpu usage of task
> group, it is hard to tell how intense the competition is between
> these groups on cpu resources.
>
> Monitoring the wait time of each process or sched_debug could cost
> too much, and there is no good way to accurately represent the
> conflict with these info, we need the wait time on group dimension.
>
> Thus we introduced group's wait_sum represent the conflict between
> task groups, which is simply sum the wait time of group's cfs_rq.
>
> The 'cpu.stat' is modified to show the statistic, like:
>
>  Â nr_periods 0
>  Â nr_throttled 0
>  Â throttled_time 0
>  Â wait_sum 2035098795584
>
> Now we can monitor the changing on wait_sum to tell how suffering
> a task group is in the fight of cpu resources.
>
> For example:
>  Â (wait_sum - last_wait_sum) * 100 / (nr_cpu * period_ns) == X%
>
> means the task group paid X percentage of period on waiting
> for the cpu.

Any comments please?

Regards,
Michael Wang


> Signed-off-by: Michael Wang <yun.wang@xxxxxxxxxxxxxxxxx>
> ---
>
> Since v1:
>  Â Use schedstat_val to avoid compile error
>  Â Check and skip root_task_group
>
>  Âkernel/sched/core.c | 8 ++++++++
>  Â1 file changed, 8 insertions(+)
>
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 78d8fac..80ab995 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -6781,6 +6781,8 @@ static int __cfs_schedulable(struct task_group *tg, u64 period, u64 quota)
>
>  Âstatic int cpu_cfs_stat_show(struct seq_file *sf, void *v)
>  Â{
> +ÂÂÂ int i;
> +ÂÂÂ u64 ws = 0;
>  ÂÂÂÂ struct task_group *tg = css_tg(seq_css(sf));
>  ÂÂÂÂ struct cfs_bandwidth *cfs_b = &tg->cfs_bandwidth;
>
> @@ -6788,6 +6790,12 @@ static int cpu_cfs_stat_show(struct seq_file *sf, void *v)
>  ÂÂÂÂ seq_printf(sf, "nr_throttled %d\n", cfs_b->nr_throttled);
>  ÂÂÂÂ seq_printf(sf, "throttled_time %llu\n", cfs_b->throttled_time);
>
> +ÂÂÂ if (schedstat_enabled() && tg != &root_task_group) {
> +ÂÂÂÂÂÂÂ for_each_possible_cpu(i)
> +ÂÂÂÂÂÂÂÂÂÂÂ ws += schedstat_val(tg->se[i]->statistics.wait_sum);
> +ÂÂÂÂÂÂÂ seq_printf(sf, "wait_sum %llu\n", ws);
> +ÂÂÂ }
> +
>  ÂÂÂÂ return 0;
>  Â}
>  Â#endif /* CONFIG_CFS_BANDWIDTH */