From f65c64f311ee2f1ddc1eb395ed8b20e6b9d14e85 Mon Sep 17 00:00:00 2001 From: Wang Yaxin Date: Fri, 20 Dec 2024 17:31:05 +0800 Subject: delayacct: add delay min to record delay peak Delay accounting can now calculate the average delay of processes, detect the overall system load, and also record the 'delay max' to identify potential abnormal delays. However, 'delay min' can help us identify another useful delay peak. By comparing the difference between 'delay max' and 'delay min', we can understand the optimization space for latency, providing a reference for the optimization of latency performance. Use case ========= bash-4.4# ./getdelays -d -t 242 print delayacct stats ON TGID 242 CPU count real total virtual total delay total delay average delay max delay min 39 156000000 156576579 2111069 0.054ms 0.212296ms 0.031307ms IO count delay total delay average delay max delay min 0 0 0.000ms 0.000000ms 0.000000ms SWAP count delay total delay average delay max delay min 0 0 0.000ms 0.000000ms 0.000000ms RECLAIM count delay total delay average delay max delay min 0 0 0.000ms 0.000000ms 0.000000ms THRASHING count delay total delay average delay max delay min 0 0 0.000ms 0.000000ms 0.000000ms COMPACT count delay total delay average delay max delay min 0 0 0.000ms 0.000000ms 0.000000ms WPCOPY count delay total delay average delay max delay min 156 11215873 0.072ms 0.207403ms 0.033913ms IRQ count delay total delay average delay max delay min 0 0 0.000ms 0.000000ms 0.000000ms Link: https://lkml.kernel.org/r/20241220173105906EOdsPhzjMLYNJJBqgz1ga@zte.com.cn Co-developed-by: Wang Yong Signed-off-by: Wang Yong Co-developed-by: xu xin Signed-off-by: xu xin Signed-off-by: Wang Yaxin Co-developed-by: Kun Jiang Signed-off-by: Kun Jiang Cc: Balbir Singh Cc: David Hildenbrand Cc: Fan Yu Cc: Peilin He Cc: tuqiang Cc: ye xingchen Cc: Yunkai Zhang Signed-off-by: Andrew Morton --- kernel/delayacct.c | 32 +++++++++++++++++++++++++------- kernel/sched/stats.h | 4 ++++ 2 files changed, 29 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/delayacct.c b/kernel/delayacct.c index 23212a0c88e4..b238eb8c6573 100644 --- a/kernel/delayacct.c +++ b/kernel/delayacct.c @@ -95,7 +95,7 @@ void __delayacct_tsk_init(struct task_struct *tsk) * Finish delay accounting for a statistic using its timestamps (@start), * accumulator (@total) and @count */ -static void delayacct_end(raw_spinlock_t *lock, u64 *start, u64 *total, u32 *count, u64 *max) +static void delayacct_end(raw_spinlock_t *lock, u64 *start, u64 *total, u32 *count, u64 *max, u64 *min) { s64 ns = local_clock() - *start; unsigned long flags; @@ -106,6 +106,8 @@ static void delayacct_end(raw_spinlock_t *lock, u64 *start, u64 *total, u32 *cou (*count)++; if (ns > *max) *max = ns; + if (*min == 0 || ns < *min) + *min = ns; raw_spin_unlock_irqrestore(lock, flags); } } @@ -125,7 +127,8 @@ void __delayacct_blkio_end(struct task_struct *p) &p->delays->blkio_start, &p->delays->blkio_delay, &p->delays->blkio_count, - &p->delays->blkio_delay_max); + &p->delays->blkio_delay_max, + &p->delays->blkio_delay_min); } int delayacct_add_tsk(struct taskstats *d, struct task_struct *tsk) @@ -157,6 +160,7 @@ int delayacct_add_tsk(struct taskstats *d, struct task_struct *tsk) d->cpu_count += t1; d->cpu_delay_max = tsk->sched_info.max_run_delay; + d->cpu_delay_min = tsk->sched_info.min_run_delay; tmp = (s64)d->cpu_delay_total + t2; d->cpu_delay_total = (tmp < (s64)d->cpu_delay_total) ? 0 : tmp; tmp = (s64)d->cpu_run_virtual_total + t3; @@ -170,24 +174,31 @@ int delayacct_add_tsk(struct taskstats *d, struct task_struct *tsk) /* zero XXX_total, non-zero XXX_count implies XXX stat overflowed */ raw_spin_lock_irqsave(&tsk->delays->lock, flags); d->blkio_delay_max = tsk->delays->blkio_delay_max; + d->blkio_delay_min = tsk->delays->blkio_delay_min; tmp = d->blkio_delay_total + tsk->delays->blkio_delay; d->blkio_delay_total = (tmp < d->blkio_delay_total) ? 0 : tmp; d->swapin_delay_max = tsk->delays->swapin_delay_max; + d->swapin_delay_min = tsk->delays->swapin_delay_min; tmp = d->swapin_delay_total + tsk->delays->swapin_delay; d->swapin_delay_total = (tmp < d->swapin_delay_total) ? 0 : tmp; d->freepages_delay_max = tsk->delays->freepages_delay_max; + d->freepages_delay_min = tsk->delays->freepages_delay_min; tmp = d->freepages_delay_total + tsk->delays->freepages_delay; d->freepages_delay_total = (tmp < d->freepages_delay_total) ? 0 : tmp; d->thrashing_delay_max = tsk->delays->thrashing_delay_max; + d->thrashing_delay_min = tsk->delays->thrashing_delay_min; tmp = d->thrashing_delay_total + tsk->delays->thrashing_delay; d->thrashing_delay_total = (tmp < d->thrashing_delay_total) ? 0 : tmp; d->compact_delay_max = tsk->delays->compact_delay_max; + d->compact_delay_min = tsk->delays->compact_delay_min; tmp = d->compact_delay_total + tsk->delays->compact_delay; d->compact_delay_total = (tmp < d->compact_delay_total) ? 0 : tmp; d->wpcopy_delay_max = tsk->delays->wpcopy_delay_max; + d->wpcopy_delay_min = tsk->delays->wpcopy_delay_min; tmp = d->wpcopy_delay_total + tsk->delays->wpcopy_delay; d->wpcopy_delay_total = (tmp < d->wpcopy_delay_total) ? 0 : tmp; d->irq_delay_max = tsk->delays->irq_delay_max; + d->irq_delay_min = tsk->delays->irq_delay_min; tmp = d->irq_delay_total + tsk->delays->irq_delay; d->irq_delay_total = (tmp < d->irq_delay_total) ? 0 : tmp; d->blkio_count += tsk->delays->blkio_count; @@ -224,7 +235,8 @@ void __delayacct_freepages_end(void) ¤t->delays->freepages_start, ¤t->delays->freepages_delay, ¤t->delays->freepages_count, - ¤t->delays->freepages_delay_max); + ¤t->delays->freepages_delay_max, + ¤t->delays->freepages_delay_min); } void __delayacct_thrashing_start(bool *in_thrashing) @@ -247,7 +259,8 @@ void __delayacct_thrashing_end(bool *in_thrashing) ¤t->delays->thrashing_start, ¤t->delays->thrashing_delay, ¤t->delays->thrashing_count, - ¤t->delays->thrashing_delay_max); + ¤t->delays->thrashing_delay_max, + ¤t->delays->thrashing_delay_min); } void __delayacct_swapin_start(void) @@ -261,7 +274,8 @@ void __delayacct_swapin_end(void) ¤t->delays->swapin_start, ¤t->delays->swapin_delay, ¤t->delays->swapin_count, - ¤t->delays->swapin_delay_max); + ¤t->delays->swapin_delay_max, + ¤t->delays->swapin_delay_min); } void __delayacct_compact_start(void) @@ -275,7 +289,8 @@ void __delayacct_compact_end(void) ¤t->delays->compact_start, ¤t->delays->compact_delay, ¤t->delays->compact_count, - ¤t->delays->compact_delay_max); + ¤t->delays->compact_delay_max, + ¤t->delays->compact_delay_min); } void __delayacct_wpcopy_start(void) @@ -289,7 +304,8 @@ void __delayacct_wpcopy_end(void) ¤t->delays->wpcopy_start, ¤t->delays->wpcopy_delay, ¤t->delays->wpcopy_count, - ¤t->delays->wpcopy_delay_max); + ¤t->delays->wpcopy_delay_max, + ¤t->delays->wpcopy_delay_min); } void __delayacct_irq(struct task_struct *task, u32 delta) @@ -301,6 +317,8 @@ void __delayacct_irq(struct task_struct *task, u32 delta) task->delays->irq_count++; if (delta > task->delays->irq_delay_max) task->delays->irq_delay_max = delta; + if (delta && (!task->delays->irq_delay_min || delta < task->delays->irq_delay_min)) + task->delays->irq_delay_min = delta; raw_spin_unlock_irqrestore(&task->delays->lock, flags); } diff --git a/kernel/sched/stats.h b/kernel/sched/stats.h index ed72435aef51..693537b908a1 100644 --- a/kernel/sched/stats.h +++ b/kernel/sched/stats.h @@ -246,6 +246,8 @@ static inline void sched_info_dequeue(struct rq *rq, struct task_struct *t) t->sched_info.run_delay += delta; if (delta > t->sched_info.max_run_delay) t->sched_info.max_run_delay = delta; + if (delta && (!t->sched_info.min_run_delay || delta < t->sched_info.min_run_delay)) + t->sched_info.min_run_delay = delta; rq_sched_info_dequeue(rq, delta); } @@ -269,6 +271,8 @@ static void sched_info_arrive(struct rq *rq, struct task_struct *t) t->sched_info.pcount++; if (delta > t->sched_info.max_run_delay) t->sched_info.max_run_delay = delta; + if (delta && (!t->sched_info.min_run_delay || delta < t->sched_info.min_run_delay)) + t->sched_info.min_run_delay = delta; rq_sched_info_arrive(rq, delta); } -- cgit v1.2.3