sched: add a packing level knob

The knob is used to set an average load threshold that will be used to trig
the inclusion/removal of CPUs in the packing effort list.

Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>

3 files changed, 52 insertions, 0 deletions

diff --git a/include/linux/sched/sysctl.h b/include/linux/sched/sysctl.h
index bf8086b2506e..f41afa52d2df 100644
--- a/include/linux/sched/sysctl.h
+++ b/include/linux/sched/sysctl.h
@@ -44,6 +44,14 @@ enum sched_tunable_scaling {
 };
 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
 
+#ifdef CONFIG_SCHED_PACKING_TASKS
+extern int  __read_mostly sysctl_sched_packing_level;
+
+int sched_proc_update_packing(struct ctl_table *table, int write,
+		void __user *buffer, size_t *lenp,
+		loff_t *ppos);
+#endif
+
 extern unsigned int sysctl_numa_balancing_scan_delay;
 extern unsigned int sysctl_numa_balancing_scan_period_min;
 extern unsigned int sysctl_numa_balancing_scan_period_max;
@@ -61,6 +69,7 @@ extern unsigned int sysctl_sched_shares_window;
 int sched_proc_update_handler(struct ctl_table *table, int write,
 		void __user *buffer, size_t *length,
 		loff_t *ppos);
+
 #endif
 #ifdef CONFIG_SCHED_DEBUG
 static inline unsigned int get_sysctl_timer_migration(void)
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 7149f38f37fc..556898049ec4 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -186,6 +186,32 @@ void sched_init_granularity(void)
  */
 DEFINE_PER_CPU(int, sd_pack_buddy);
 
+/*
+ * The packing level of the scheduler
+ *
+ * This level define the activity % above which we should add another CPU to
+ * participate to the packing effort of the tasks
+ */
+#define DEFAULT_PACKING_LEVEL 80
+int __read_mostly sysctl_sched_packing_level = DEFAULT_PACKING_LEVEL;
+
+unsigned int sd_pack_threshold = (100 * 1024) / DEFAULT_PACKING_LEVEL;
+
+
+int sched_proc_update_packing(struct ctl_table *table, int write,
+		void __user *buffer, size_t *lenp,
+		loff_t *ppos)
+{
+	int ret = proc_dointvec_minmax(table, write, buffer, lenp, ppos);
+	if (ret || !write)
+		return ret;
+
+	if (sysctl_sched_packing_level)
+		sd_pack_threshold = (100 * 1024) / sysctl_sched_packing_level;
+
+	return 0;
+}
+
 static inline bool is_packing_cpu(int cpu)
 {
 	int my_buddy = per_cpu(sd_pack_buddy, cpu);
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index b2f06f3c6a3f..77383fc5032a 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -255,11 +255,17 @@ static struct ctl_table sysctl_base_table[] = {
 	{ }
 };
 
+#ifdef CONFIG_SCHED_PACKING_TASKS
+static int min_sched_packing_level;
+static int max_sched_packing_level = 100;
+#endif /* CONFIG_SMP */
+
 #ifdef CONFIG_SCHED_DEBUG
 static int min_sched_granularity_ns = 100000;		/* 100 usecs */
 static int max_sched_granularity_ns = NSEC_PER_SEC;	/* 1 second */
 static int min_wakeup_granularity_ns;			/* 0 usecs */
 static int max_wakeup_granularity_ns = NSEC_PER_SEC;	/* 1 second */
+
 #ifdef CONFIG_SMP
 static int min_sched_tunable_scaling = SCHED_TUNABLESCALING_NONE;
 static int max_sched_tunable_scaling = SCHED_TUNABLESCALING_END-1;
@@ -279,6 +285,17 @@ static struct ctl_table kern_table[] = {
 		.mode		= 0644,
 		.proc_handler	= proc_dointvec,
 	},
+#ifdef CONFIG_SCHED_PACKING_TASKS
+	{
+		.procname	= "sched_packing_level",
+		.data		= &sysctl_sched_packing_level,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= sched_proc_update_packing,
+		.extra1		= &min_sched_packing_level,
+		.extra2		= &max_sched_packing_level,
+	},
+#endif
 #ifdef CONFIG_SCHED_DEBUG
 	{
 		.procname	= "sched_min_granularity_ns",