Skip to contentSkip to conversation

sched: Feature to disable APERF/MPERF cpu_power

tip:sched/core tip-bot for Peter Zijlstra 1 participant 1 message Sep 16, 2009

lore
Share & tools
lore
#0 lore raw
Commit-ID:  8e6598af3f35629c37249a610cf13e73f70db279
Gitweb:     http://git.kernel.org/tip/8e6598af3f35629c37249a610cf13e73f70db279
Author:     Peter Zijlstra <[email protected]>
AuthorDate: Thu, 3 Sep 2009 13:20:03 +0200
Committer:  Ingo Molnar <[email protected]>
CommitDate: Tue, 15 Sep 2009 16:51:28 +0200

sched: Feature to disable APERF/MPERF cpu_power

I suspect a feed-back loop between cpuidle and the aperf/mperf cpu_power bits, where when we have idle C-states lower the ratio, which leads to lower cpu_power and then less load, which generates more idle time, etc..

Put in a knob to disable it.

LKML-Reference: <new-submission>

---
 kernel/sched.c          |   12 ++++++++++--
 kernel/sched_features.h |    5 +++++
 2 files changed, 15 insertions(+), 2 deletions(-)
kernel/sched.c +10 -2
index c210321..e8e603b 100644
@@ -3602,11 +3602,19 @@ static void update_cpu_power(struct sched_domain *sd, int cpu)
 	unsigned long power = SCHED_LOAD_SCALE;
 	struct sched_group *sdg = sd->groups;
 
-	power *= arch_scale_freq_power(sd, cpu);
+	if (sched_feat(ARCH_POWER))
+		power *= arch_scale_freq_power(sd, cpu);
+	else
+		power *= default_scale_freq_power(sd, cpu);
+
 	power >>= SCHED_LOAD_SHIFT;
 
 	if ((sd->flags & SD_SHARE_CPUPOWER) && weight > 1) {
-		power *= arch_scale_smt_power(sd, cpu);
+		if (sched_feat(ARCH_POWER))
+			power *= arch_scale_smt_power(sd, cpu);
+		else
+			power *= default_scale_smt_power(sd, cpu);
+
 		power >>= SCHED_LOAD_SHIFT;
 	}
 
kernel/sched_features.h +5 -0
index e98c2e8..294e10e 100644
@@ -82,6 +82,11 @@ SCHED_FEAT(LAST_BUDDY, 1)
  */
 SCHED_FEAT(CACHE_HOT_BUDDY, 1)
 
+/*
+ * Use arch dependent cpu power functions
+ */
+SCHED_FEAT(ARCH_POWER, 0)
+
 SCHED_FEAT(HRTICK, 0)
 SCHED_FEAT(DOUBLE_TICK, 0)
 SCHED_FEAT(LB_BIAS, 1)