sched, timers: move calc_load() to scheduler

Dimitri Sivanich noticed that xtime_lock is held write locked across
calc_load() which iterates over all online CPUs. That can cause long
latencies for xtime_lock readers on large SMP systems. 

The load average calculation is an rough estimate anyway so there is
no real need to protect the readers vs. the update. It's not a problem
when the avenrun array is updated while a reader copies the values.

Instead of iterating over all online CPUs let the scheduler_tick code
update the number of active tasks shortly before the avenrun update
happens. The avenrun update itself is handled by the CPU which calls
do_timer().

[ Impact: reduce xtime_lock write locked section ]

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>

1 files changed, 74 insertions, 10 deletions

diff --git a/kernel/sched.c b/kernel/sched.c
index 8908d190a348..f4eb88153bd1 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -630,6 +630,10 @@ struct rq {
 	struct list_head migration_queue;
 #endif
 
+	/* calc_load related fields */
+	unsigned long calc_load_update;
+	long calc_load_active;
+
 #ifdef CONFIG_SCHED_HRTICK
 #ifdef CONFIG_SMP
 	int hrtick_csd_pending;
@@ -1728,6 +1732,8 @@ static void cfs_rq_set_shares(struct cfs_rq *cfs_rq, unsigned long shares)
 }
 #endif
 
+static void calc_load_account_active(struct rq *this_rq);
+
 #include "sched_stats.h"
 #include "sched_idletask.c"
 #include "sched_fair.c"
@@ -2856,19 +2862,57 @@ unsigned long nr_iowait(void)
 	return sum;
 }
 
-unsigned long nr_active(void)
+/* Variables and functions for calc_load */
+static atomic_long_t calc_load_tasks;
+static unsigned long calc_load_update;
+unsigned long avenrun[3];
+EXPORT_SYMBOL(avenrun);
+
+static unsigned long
+calc_load(unsigned long load, unsigned long exp, unsigned long active)
 {
-	unsigned long i, running = 0, uninterruptible = 0;
+	load *= exp;
+	load += active * (FIXED_1 - exp);
+	return load >> FSHIFT;
+}
 
-	for_each_online_cpu(i) {
-		running += cpu_rq(i)->nr_running;
-		uninterruptible += cpu_rq(i)->nr_uninterruptible;
-	}
+/*
+ * calc_load - update the avenrun load estimates 10 ticks after the
+ * CPUs have updated calc_load_tasks.
+ */
+void calc_global_load(void)
+{
+	unsigned long upd = calc_load_update + 10;
+	long active;
+
+	if (time_before(jiffies, upd))
+		return;
 
-	if (unlikely((long)uninterruptible < 0))
-		uninterruptible = 0;
+	active = atomic_long_read(&calc_load_tasks);
+	active = active > 0 ? active * FIXED_1 : 0;
 
-	return running + uninterruptible;
+	avenrun[0] = calc_load(avenrun[0], EXP_1, active);
+	avenrun[1] = calc_load(avenrun[1], EXP_5, active);
+	avenrun[2] = calc_load(avenrun[2], EXP_15, active);
+
+	calc_load_update += LOAD_FREQ;
+}
+
+/*
+ * Either called from update_cpu_load() or from a cpu going idle
+ */
+static void calc_load_account_active(struct rq *this_rq)
+{
+	long nr_active, delta;
+
+	nr_active = this_rq->nr_running;
+	nr_active += (long) this_rq->nr_uninterruptible;
+
+	if (nr_active != this_rq->calc_load_active) {
+		delta = nr_active - this_rq->calc_load_active;
+		this_rq->calc_load_active = nr_active;
+		atomic_long_add(delta, &calc_load_tasks);
+	}
 }
 
 /*
@@ -2899,6 +2943,11 @@ static void update_cpu_load(struct rq *this_rq)
 			new_load += scale-1;
 		this_rq->cpu_load[i] = (old_load*(scale-1) + new_load) >> i;
 	}
+
+	if (time_after_eq(jiffies, this_rq->calc_load_update)) {
+		this_rq->calc_load_update += LOAD_FREQ;
+		calc_load_account_active(this_rq);
+	}
 }
 
 #ifdef CONFIG_SMP
@@ -7091,6 +7140,14 @@ static void migrate_dead_tasks(unsigned int dead_cpu)
 
 	}
 }
+
+/*
+ * remove the tasks which were accounted by rq from calc_load_tasks.
+ */
+static void calc_global_load_remove(struct rq *rq)
+{
+	atomic_long_sub(rq->calc_load_active, &calc_load_tasks);
+}
 #endif /* CONFIG_HOTPLUG_CPU */
 
 #if defined(CONFIG_SCHED_DEBUG) && defined(CONFIG_SYSCTL)
@@ -7325,6 +7382,8 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu)
 		/* Update our root-domain */
 		rq = cpu_rq(cpu);
 		spin_lock_irqsave(&rq->lock, flags);
+		rq->calc_load_update = calc_load_update;
+		rq->calc_load_active = 0;
 		if (rq->rd) {
 			BUG_ON(!cpumask_test_cpu(cpu, rq->rd->span));
 
@@ -7364,7 +7423,7 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu)
 		cpuset_unlock();
 		migrate_nr_uninterruptible(rq);
 		BUG_ON(rq->nr_running != 0);
-
+		calc_global_load_remove(rq);
 		/*
 		 * No need to migrate the tasks: it was best-effort if
 		 * they didn't take sched_hotcpu_mutex. Just wake up
@@ -9059,6 +9118,8 @@ void __init sched_init(void)
 		rq = cpu_rq(i);
 		spin_lock_init(&rq->lock);
 		rq->nr_running = 0;
+		rq->calc_load_active = 0;
+		rq->calc_load_update = jiffies + LOAD_FREQ;
 		init_cfs_rq(&rq->cfs, rq);
 		init_rt_rq(&rq->rt, rq);
 #ifdef CONFIG_FAIR_GROUP_SCHED
@@ -9166,6 +9227,9 @@ void __init sched_init(void)
 	 * when this runqueue becomes "idle".
 	 */
 	init_idle(current, smp_processor_id());
+
+	calc_load_update = jiffies + LOAD_FREQ;
+
 	/*
 	 * During early bootup we pretend to be a normal task:
 	 */