Re: [RFC PATCH 0/1] sched/pelt: Change PELT halflife at runtime

[Peter Zijlstra <peterz@infradead.org>]

On Thu, Sep 29, 2022 at 03:41:47PM +0100, Kajetan Puchalski wrote:

> Based on all the tests we've seen, jankbench or otherwise, the
> improvement can mainly be attributed to the faster ramp up of frequency
> caused by the shorter PELT window while using schedutil.

Would something terrible like the below help some?

If not, I suppose it could be modified to take the current state as
history. But basically it runs a faster pelt sum along side the regular
signal just for ramping up the frequency.

diff --git a/kernel/sched/features.h b/kernel/sched/features.h
index ee7f23c76bd3..9ba07a1d19f6 100644
--- a/kernel/sched/features.h
+++ b/kernel/sched/features.h
@@ -96,6 +96,7 @@ SCHED_FEAT(WA_BIAS, true)
  */
 SCHED_FEAT(UTIL_EST, true)
 SCHED_FEAT(UTIL_EST_FASTUP, true)
+SCHED_FEAT(UTIL_EST_FASTER, true)
 
 SCHED_FEAT(LATENCY_WARN, false)
 
diff --git a/kernel/sched/pelt.c b/kernel/sched/pelt.c
index 0f310768260c..13cd9e27ce3e 100644
--- a/kernel/sched/pelt.c
+++ b/kernel/sched/pelt.c
@@ -148,6 +148,22 @@ accumulate_sum(u64 delta, struct sched_avg *sa,
 	return periods;
 }
 
+/*
+ * Compute a pelt util_avg assuming no history and @delta runtime.
+ */
+unsigned long faster_est_approx(u64 delta)
+{
+	unsigned long contrib = (unsigned long)delta; /* p == 0 -> delta < 1024 */
+	u64 periods = delta / 1024;
+
+	if (periods) {
+		delta %= 1024;
+		contrib = __accumulate_pelt_segments(periods, 1024, delta);
+	}
+
+	return (contrib << SCHED_CAPACITY_SHIFT) / PELT_MIN_DIVIDER;
+}
+
 /*
  * We can represent the historical contribution to runnable average as the
  * coefficients of a geometric series.  To do this we sub-divide our runnable
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index a4a20046e586..99827d5dda27 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -2922,6 +2922,8 @@ static inline unsigned long cpu_util_dl(struct rq *rq)
 	return READ_ONCE(rq->avg_dl.util_avg);
 }
 
+extern unsigned long faster_est_approx(u64 runtime);
+
 /**
  * cpu_util_cfs() - Estimates the amount of CPU capacity used by CFS tasks.
  * @cpu: the CPU to get the utilization for.
@@ -2956,13 +2958,26 @@ static inline unsigned long cpu_util_dl(struct rq *rq)
  */
 static inline unsigned long cpu_util_cfs(int cpu)
 {
+	struct rq *rq = cpu_rq(cpu);
 	struct cfs_rq *cfs_rq;
 	unsigned long util;
 
-	cfs_rq = &cpu_rq(cpu)->cfs;
+	cfs_rq = &rq->cfs;
 	util = READ_ONCE(cfs_rq->avg.util_avg);
 
 	if (sched_feat(UTIL_EST)) {
+		if (sched_feat(UTIL_EST_FASTER)) {
+			struct task_struct *curr;
+
+			rcu_read_lock();
+			curr = rcu_dereference(rq->curr);
+			if (likely(curr->sched_class == &fair_sched_class)) {
+				u64 runtime = curr->se.sum_exec_runtime - curr->se.exec_start;
+				util = max_t(unsigned long, util,
+					     faster_est_approx(runtime * 2));
+			}
+			rcu_read_unlock();
+		}
 		util = max_t(unsigned long, util,
 			     READ_ONCE(cfs_rq->avg.util_est.enqueued));
 	}