sched: Add sysctl to enable power aware scheduling

Add sysctl to enable energy awareness at runtime. This is useful for
performance/power tuning/measurements and debugging. In addition this
will match up with the Documentation/scheduler/sched-hmp.txt documentation.

Change-Id: I0a9185498640d66917b38bf5d55f6c59fc60ad5c
Signed-off-by: Olav Haugan <ohaugan@codeaurora.org>
[rameezmustafa@codeaurora.org]: Port to msm-3.18]
Signed-off-by: Syed Rameez Mustafa <rameezmustafa@codeaurora.org

6 files changed, 19 insertions, 11 deletions

diff --git a/Documentation/scheduler/sched-hmp.txt b/Documentation/scheduler/sched-hmp.txt
index 54add3c4b94f..62c33b8b45fd 100644
--- a/Documentation/scheduler/sched-hmp.txt
+++ b/Documentation/scheduler/sched-hmp.txt
@@ -1086,7 +1086,7 @@ high-performance CPU).
 
 Appears at: /proc/sys/kernel/sched_enable_power_aware
 
-Default value: 1
+Default value: 0
 
 Controls whether or not per-CPU power values are used in determining
 task placement. If this is disabled, tasks are simply placed on the
diff --git a/include/linux/sched/sysctl.h b/include/linux/sched/sysctl.h
index 0ec9fc8cd361..5437c599fd54 100644
--- a/include/linux/sched/sysctl.h
+++ b/include/linux/sched/sysctl.h
@@ -49,6 +49,7 @@ extern unsigned int sysctl_sched_freq_account_wait_time;
 extern unsigned int sysctl_sched_migration_fixup;
 extern unsigned int sysctl_sched_heavy_task_pct;
 extern unsigned int sysctl_sched_min_runtime;
+extern unsigned int sysctl_sched_enable_power_aware;
 
 #if defined(CONFIG_SCHED_FREQ_INPUT) || defined(CONFIG_SCHED_HMP)
 extern unsigned int sysctl_sched_init_task_load_pct;
diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 290b8df6f6e0..332b920af31e 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -1097,7 +1097,7 @@ static int __init set_sched_enable_power_aware(char *str)
 
 	get_option(&str, &enable_power_aware);
 
-	sched_enable_power_aware = !!enable_power_aware;
+	sysctl_sched_enable_power_aware = !!enable_power_aware;
 
 	return 0;
 }
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index ec085a9e4da1..d9d4694d58da 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -2579,8 +2579,9 @@ unsigned int __read_mostly sysctl_sched_spill_nr_run = 10;
 /*
  * Control whether or not individual CPU power consumption is used to
  * guide task placement.
+ * This sysctl can be set to a default value using boot command line arguments.
  */
-unsigned int __read_mostly sched_enable_power_aware = 0;
+unsigned int __read_mostly sysctl_sched_enable_power_aware = 0;
 
 /*
  * This specifies the maximum percent power difference between 2
@@ -3028,7 +3029,7 @@ unsigned int power_cost_at_freq(int cpu, unsigned int freq)
 	struct cpu_pstate_pwr *costs;
 
 	if (!per_cpu_info || !per_cpu_info[cpu].ptable ||
-	    !sched_enable_power_aware)
+	    !sysctl_sched_enable_power_aware)
 		/* When power aware scheduling is not in use, or CPU
 		 * power data is not available, just use the CPU
 		 * capacity as a rough stand-in for real CPU power
@@ -3059,7 +3060,7 @@ static unsigned int power_cost(struct task_struct *p, int cpu)
 	unsigned int task_freq;
 	unsigned int cur_freq = cpu_rq(cpu)->cur_freq;
 
-	if (!sched_enable_power_aware)
+	if (!sysctl_sched_enable_power_aware)
 		return cpu_rq(cpu)->max_possible_capacity;
 
 	/* calculate % of max freq needed */
@@ -3690,7 +3691,7 @@ static inline int migration_needed(struct rq *rq, struct task_struct *p)
 	if (!task_will_fit(p, cpu_of(rq)))
 		return MOVE_TO_BIG_CPU;
 
-	if (sched_enable_power_aware &&
+	if (sysctl_sched_enable_power_aware &&
 	    !is_task_migration_throttled(p) &&
 	    is_cpu_throttling_imminent(cpu_of(rq)) &&
 	    lower_power_cpu_available(p, cpu_of(rq)))
@@ -3785,7 +3786,7 @@ unsigned int cpu_temp(int cpu)
 
 #else	/* CONFIG_SCHED_HMP */
 
-#define sched_enable_power_aware 0
+#define sysctl_sched_enable_power_aware 0
 
 static inline int task_will_fit(struct task_struct *p, int cpu)
 {
@@ -7827,7 +7828,7 @@ group_type group_classify(struct sched_group *group,
 	 * prioritize spreading work over power optimization.
 	 */
 	cpu = cpumask_first(sched_group_cpus(group));
-	if (sched_enable_power_aware &&
+	if (sysctl_sched_enable_power_aware &&
 	    (capacity(env->dst_rq) == group_rq_capacity(group)) &&
 	    sgs->sum_nr_running && (env->idle != CPU_NOT_IDLE) &&
 	    power_cost_at_freq(env->dst_cpu, 0) <
@@ -8935,7 +8936,7 @@ static int idle_balance(struct rq *this_rq)
 	 */
 	rcu_read_lock();
 	sd = rcu_dereference_check_sched_domain(this_rq->sd);
-	if (sd && sched_enable_power_aware) {
+	if (sd && sysctl_sched_enable_power_aware) {
 		for_each_cpu(i, sched_domain_span(sd)) {
 			if (i == this_cpu || idle_cpu(i)) {
 				cost = power_cost_at_freq(i, 0);
@@ -9429,7 +9430,7 @@ static int select_lowest_power_cpu(struct cpumask *cpus)
 	int lowest_power_cpu = -1;
 	int lowest_power = INT_MAX;
 
-	if (sched_enable_power_aware) {
+	if (sysctl_sched_enable_power_aware) {
 		for_each_cpu(i, cpus) {
 			cost = power_cost_at_freq(i, 0);
 			if (cost < lowest_power) {
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 898e24e9f63d..89b2f4dee627 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1150,7 +1150,6 @@ static inline void clear_reserved(int cpu)
 }
 
 extern unsigned int sched_enable_hmp;
-extern unsigned int sched_enable_power_aware;
 
 int mostly_idle_cpu(int cpu);
 extern void check_for_migration(struct rq *rq, struct task_struct *p);
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 69a379c4573f..5e0354d6317c 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -442,6 +442,13 @@ static struct ctl_table kern_table[] = {
 		.mode		= 0644,
 		.proc_handler	= sched_boost_handler,
 	},
+	{
+		.procname	= "sched_enable_power_aware",
+		.data		= &sysctl_sched_enable_power_aware,
+		.maxlen		= sizeof(unsigned int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec,
+	},
 #endif	/* CONFIG_SCHED_HMP */
 #ifdef CONFIG_SCHED_DEBUG
 	{