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>

diff --git a/include/linux/sched.h b/include/linux/sched.h
index b4c38bc8049cbbea17e0ca4f929f35df9cddbe1f..6eb4892efe453c9b97260239b291132edb3a8596 100644 (file)
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -135,8 +135,8 @@ DECLARE_PER_CPU(unsigned long, process_counts);
 extern int nr_processes(void);
 extern unsigned long nr_running(void);
 extern unsigned long nr_uninterruptible(void);
-extern unsigned long nr_active(void);
 extern unsigned long nr_iowait(void);
+extern void calc_global_load(void);
 
 extern unsigned long get_parent_ip(unsigned long addr);
 

diff --git a/kernel/sched.c b/kernel/sched.c
index 8908d190a348ec8afde622bfd8b3b2a042d69664..f4eb88153bd185cdf54e68643319c058461ce59b 100644 (file)
--- 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:
         */

diff --git a/kernel/sched_idletask.c b/kernel/sched_idletask.c
index 8a21a2e28c13ac2e4620e980888702e22a4f2387..499672c10cbd615141a362cafe804711bbcf1ff3 100644 (file)
--- a/kernel/sched_idletask.c
+++ b/kernel/sched_idletask.c
@@ -22,7 +22,8 @@ static void check_preempt_curr_idle(struct rq *rq, struct task_struct *p, int sy
 static struct task_struct *pick_next_task_idle(struct rq *rq)
 {
        schedstat_inc(rq, sched_goidle);
-
+       /* adjust the active tasks as we might go into a long sleep */
+       calc_load_account_active(rq);
        return rq->idle;
 }
 

diff --git a/kernel/time/timekeeping.c b/kernel/time/timekeeping.c
index 687dff49f6e7da5ec92199f6bbae08de57e5ba43..52a8bf8931f3903128af16a13d32c5d4b8b376aa 100644 (file)
--- a/kernel/time/timekeeping.c
+++ b/kernel/time/timekeeping.c
@@ -22,7 +22,7 @@
 
 /*
  * This read-write spinlock protects us from races in SMP while
- * playing with xtime and avenrun.
+ * playing with xtime.
  */
 __cacheline_aligned_in_smp DEFINE_SEQLOCK(xtime_lock);
 

diff --git a/kernel/timer.c b/kernel/timer.c
index cffffad01c31cf1670490f2860f0e5de0f25d301..6a21d7af9620910168de16f1bd9f21ba3cb7c5b7 100644 (file)
--- a/kernel/timer.c
+++ b/kernel/timer.c
@@ -1122,47 +1122,6 @@ void update_process_times(int user_tick)
        run_posix_cpu_timers(p);
 }
 
-/*
- * Nr of active tasks - counted in fixed-point numbers
- */
-static unsigned long count_active_tasks(void)
-{
-       return nr_active() * FIXED_1;
-}
-
-/*
- * Hmm.. Changed this, as the GNU make sources (load.c) seems to
- * imply that avenrun[] is the standard name for this kind of thing.
- * Nothing else seems to be standardized: the fractional size etc
- * all seem to differ on different machines.
- *
- * Requires xtime_lock to access.
- */
-unsigned long avenrun[3];
-
-EXPORT_SYMBOL(avenrun);
-
-/*
- * calc_load - given tick count, update the avenrun load estimates.
- * This is called while holding a write_lock on xtime_lock.
- */
-static inline void calc_load(unsigned long ticks)
-{
-       unsigned long active_tasks; /* fixed-point */
-       static int count = LOAD_FREQ;
-
-       count -= ticks;
-       if (unlikely(count < 0)) {
-               active_tasks = count_active_tasks();
-               do {
-                       CALC_LOAD(avenrun[0], EXP_1, active_tasks);
-                       CALC_LOAD(avenrun[1], EXP_5, active_tasks);
-                       CALC_LOAD(avenrun[2], EXP_15, active_tasks);
-                       count += LOAD_FREQ;
-               } while (count < 0);
-       }
-}
-
 /*
  * This function runs timers and the timer-tq in bottom half context.
  */
@@ -1186,16 +1145,6 @@ void run_local_timers(void)
        softlockup_tick();
 }
 
-/*
- * Called by the timer interrupt. xtime_lock must already be taken
- * by the timer IRQ!
- */
-static inline void update_times(unsigned long ticks)
-{
-       update_wall_time();
-       calc_load(ticks);
-}
-
 /*
  * The 64-bit jiffies value is not atomic - you MUST NOT read it
  * without sampling the sequence number in xtime_lock.
@@ -1205,7 +1154,8 @@ static inline void update_times(unsigned long ticks)
 void do_timer(unsigned long ticks)
 {
        jiffies_64 += ticks;
-       update_times(ticks);
+       update_wall_time();
+       calc_global_load();
 }
 
 #ifdef __ARCH_WANT_SYS_ALARM