[linux-next,v2] sched/psi: Avoid update triggers and rtpoll_total when it is unnecessary

Message ID 202310071421311735972@zte.com.cn
State New
Headers
Series [linux-next,v2] sched/psi: Avoid update triggers and rtpoll_total when it is unnecessary |

Commit Message

Yang Yang Oct. 7, 2023, 6:21 a.m. UTC
  From: Yang Yang <yang.yang29@zte.com.cn>

When psimon wakes up and there are no state changes for rtpoll_states,
it's unnecessary to update triggers and rtpoll_total because the pressures
being monitored by user have not changed. This will help to slightly reduce
unnecessary computations of psi.

There are also some minor related optimizations, please see below.

The parameter update_total in update_triggers() is useless now. Since if
changed_states & group->rtpoll_states is true, new_stall in update_triggers()
will be true, then update_total should also be true. We have no need for
update_total to help judgment whether to update rtpoll_total, so delete
update_total.

Update_triggers() always return now + group->rtpoll_min_period, and the
return value is only used by psi_rtpoll_work(), so change update_triggers()
to a void function, let group->rtpoll_next_update = now +
group->rtpoll_min_period directly.

Signed-off-by: Yang Yang <yang.yang29@zte.com.cn>
Suggested-by: Suren Baghdasaryan <surenb@google.com>
---
 kernel/sched/psi.c | 23 ++++++-----------------
 1 file changed, 6 insertions(+), 17 deletions(-)
  

Comments

Ingo Molnar Oct. 9, 2023, 10:49 a.m. UTC | #1
* yang.yang29@zte.com.cn <yang.yang29@zte.com.cn> wrote:

> From: Yang Yang <yang.yang29@zte.com.cn>
> 
> When psimon wakes up and there are no state changes for rtpoll_states,
> it's unnecessary to update triggers and rtpoll_total because the pressures
> being monitored by user have not changed. This will help to slightly reduce
> unnecessary computations of psi.
> 
> There are also some minor related optimizations, please see below.
> 
> The parameter update_total in update_triggers() is useless now. Since if
> changed_states & group->rtpoll_states is true, new_stall in update_triggers()
> will be true, then update_total should also be true. We have no need for
> update_total to help judgment whether to update rtpoll_total, so delete
> update_total.
> 
> Update_triggers() always return now + group->rtpoll_min_period, and the
> return value is only used by psi_rtpoll_work(), so change update_triggers()
> to a void function, let group->rtpoll_next_update = now +
> group->rtpoll_min_period directly.

Yeah, so please split this up into 3 patches: one change per patch, even
if each patch is relatively small.

Thanks,

	Ingo
  

Patch

diff --git a/kernel/sched/psi.c b/kernel/sched/psi.c
index 1d0f634725a6..44a78774ae87 100644
--- a/kernel/sched/psi.c
+++ b/kernel/sched/psi.c
@@ -434,14 +434,13 @@  static u64 window_update(struct psi_window *win, u64 now, u64 value)
 	return growth;
 }

-static u64 update_triggers(struct psi_group *group, u64 now, bool *update_total,
+static void update_triggers(struct psi_group *group, u64 now,
 						   enum psi_aggregators aggregator)
 {
 	struct psi_trigger *t;
 	u64 *total = group->total[aggregator];
 	struct list_head *triggers;
 	u64 *aggregator_total;
-	*update_total = false;

 	if (aggregator == PSI_AVGS) {
 		triggers = &group->avg_triggers;
@@ -471,14 +470,6 @@  static u64 update_triggers(struct psi_group *group, u64 now, bool *update_total,
 		 * events without dropping any).
 		 */
 		if (new_stall) {
-			/*
-			 * Multiple triggers might be looking at the same state,
-			 * remember to update group->polling_total[] once we've
-			 * been through all of them. Also remember to extend the
-			 * polling time if we see new stall activity.
-			 */
-			*update_total = true;
-
 			/* Calculate growth since last update */
 			growth = window_update(&t->win, now, total[t->state]);
 			if (!t->pending_event) {
@@ -503,8 +494,6 @@  static u64 update_triggers(struct psi_group *group, u64 now, bool *update_total,
 		/* Reset threshold breach flag once event got generated */
 		t->pending_event = false;
 	}
-
-	return now + group->rtpoll_min_period;
 }

 static u64 update_averages(struct psi_group *group, u64 now)
@@ -565,7 +554,6 @@  static void psi_avgs_work(struct work_struct *work)
 	struct delayed_work *dwork;
 	struct psi_group *group;
 	u32 changed_states;
-	bool update_total;
 	u64 now;

 	dwork = to_delayed_work(work);
@@ -584,7 +572,7 @@  static void psi_avgs_work(struct work_struct *work)
 	 * go - see calc_avgs() and missed_periods.
 	 */
 	if (now >= group->avg_next_update) {
-		update_triggers(group, now, &update_total, PSI_AVGS);
+		update_triggers(group, now, PSI_AVGS);
 		group->avg_next_update = update_averages(group, now);
 	}

@@ -640,7 +628,6 @@  static void psi_rtpoll_work(struct psi_group *group)
 {
 	bool force_reschedule = false;
 	u32 changed_states;
-	bool update_total;
 	u64 now;

 	mutex_lock(&group->rtpoll_trigger_lock);
@@ -706,10 +693,12 @@  static void psi_rtpoll_work(struct psi_group *group)
 	}

 	if (now >= group->rtpoll_next_update) {
-		group->rtpoll_next_update = update_triggers(group, now, &update_total, PSI_POLL);
-		if (update_total)
+		if (changed_states & group->rtpoll_states) {
+			update_triggers(group, now, PSI_POLL);
 			memcpy(group->rtpoll_total, group->total[PSI_POLL],
 				   sizeof(group->rtpoll_total));
+		}
+		group->rtpoll_next_update = now + group->rtpoll_min_period;
 	}

 	psi_schedule_rtpoll_work(group,