From: Chris Redpath <chris.redpath@arm.com>
Date: Thu, 8 Aug 2013 15:31:07 +0000 (+0100)
Subject: sched: track per-rq 'last migration time'
X-Git-Tag: firefly_0821_release~3680^2~16^2^2~48
X-Git-Url: http://plrg.eecs.uci.edu/git/?a=commitdiff_plain;h=add684211e0ff4a08f419f6547fc311a72c35391;p=firefly-linux-kernel-4.4.55.git

sched: track per-rq 'last migration time'

Track when migrations were performed to runqueues.

Use this to decide between runqueues as migration targets when run
queues in an hmp domain have equal load.

Intention is to spread migration load amongst CPUs more fairly.

When all CPUs in an hmp domain are fully loaded, the existing code
always selects the last CPU as a migration target - this is unfair
and little better than doing no selection.

Signed-off-by: Chris Redpath <chris.redpath@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
Signed-off-by: Jon Medhurst <tixy@linaro.org>
---

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index b801eb0330e1..62302a372796 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -3764,17 +3764,21 @@ static inline unsigned int hmp_select_slower_cpu(struct task_struct *tsk,
 static inline void hmp_next_up_delay(struct sched_entity *se, int cpu)
 {
 	struct cfs_rq *cfs_rq = &cpu_rq(cpu)->cfs;
-
-	se->avg.hmp_last_up_migration = cfs_rq_clock_task(cfs_rq);
+	u64 now = cfs_rq_clock_task(cfs_rq);
+	se->avg.hmp_last_up_migration = now;
 	se->avg.hmp_last_down_migration = 0;
+	cpu_rq(cpu)->avg.hmp_last_up_migration = now;
+	cpu_rq(cpu)->avg.hmp_last_down_migration = 0;
 }
 
 static inline void hmp_next_down_delay(struct sched_entity *se, int cpu)
 {
 	struct cfs_rq *cfs_rq = &cpu_rq(cpu)->cfs;
-
-	se->avg.hmp_last_down_migration = cfs_rq_clock_task(cfs_rq);
+	u64 now = cfs_rq_clock_task(cfs_rq);
+	se->avg.hmp_last_down_migration = now;
 	se->avg.hmp_last_up_migration = 0;
+	cpu_rq(cpu)->avg.hmp_last_down_migration = now;
+	cpu_rq(cpu)->avg.hmp_last_up_migration = 0;
 }
 
 #ifdef CONFIG_HMP_VARIABLE_SCALE
@@ -3946,15 +3950,37 @@ static inline unsigned int hmp_domain_min_load(struct hmp_domain *hmpd,
 {
 	int cpu;
 	int min_cpu_runnable_temp = NR_CPUS;
+	u64 min_target_last_migration = ULLONG_MAX;
+	u64 curr_last_migration;
 	unsigned long min_runnable_load = INT_MAX;
-	unsigned long contrib;
+	unsigned long scaled_min_runnable_load = INT_MAX;
+	unsigned long contrib, scaled_contrib;
+	struct sched_avg *avg;
 
 	for_each_cpu_mask(cpu, hmpd->cpus) {
+		avg = &cpu_rq(cpu)->avg;
+		/* used for both up and down migration */
+		curr_last_migration = avg->hmp_last_up_migration ?
+			avg->hmp_last_up_migration : avg->hmp_last_down_migration;
+
 		/* don't use the divisor in the loop, just at the end */
-		contrib = cpu_rq(cpu)->avg.runnable_avg_sum * scale_load_down(1024);
-		if (contrib < min_runnable_load) {
+		contrib = avg->runnable_avg_sum * scale_load_down(1024);
+		scaled_contrib = contrib >> 22;
+
+		if ((contrib < min_runnable_load) ||
+			(scaled_contrib == scaled_min_runnable_load &&
+			 curr_last_migration < min_target_last_migration)) {
+			/*
+			 * if the load is the same target the CPU with
+			 * the longest time since a migration.
+			 * This is to spread migration load between
+			 * members of a domain more evenly when the
+			 * domain is fully loaded
+			 */
 			min_runnable_load = contrib;
+			scaled_min_runnable_load = scaled_contrib;
 			min_cpu_runnable_temp = cpu;
+			min_target_last_migration = curr_last_migration;
 		}
 	}