sched/fair: Call cpufreq update util handlers less frequently on UP

For SMP systems, update_load_avg() calls the cpufreq update util
handlers only for the top level cfs_rq (i.e. rq->cfs).

But that is not the case for UP systems. update_load_avg() calls util
handler for any cfs_rq for which it is called. This would result in way
too many calls from the scheduler to the cpufreq governors when
CONFIG_FAIR_GROUP_SCHED is enabled.

Reduce the frequency of these calls by copying the behavior from the SMP
case, i.e. Only call util handlers for the top level cfs_rq.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: linaro-kernel@lists.linaro.org
Fixes: 536bd00cdbb7 ("sched/fair: Fix !CONFIG_SMP kernel cpufreq governor breakage")
Link: http://lkml.kernel.org/r/6abf69a2107525885b616a2c1ec03d9c0946171c.1495603536.git.viresh.kumar@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index c95880e216f6900ed06e92e93ddc355d20e59092..139abf2ae2a5806d022591ec4e2ffe0468e64d51 100644 (file)
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -2790,6 +2790,29 @@ static inline void update_cfs_shares(struct sched_entity *se)
 }
 #endif /* CONFIG_FAIR_GROUP_SCHED */
 
+static inline void cfs_rq_util_change(struct cfs_rq *cfs_rq)
+{
+       if (&this_rq()->cfs == cfs_rq) {
+               /*
+                * There are a few boundary cases this might miss but it should
+                * get called often enough that that should (hopefully) not be
+                * a real problem -- added to that it only calls on the local
+                * CPU, so if we enqueue remotely we'll miss an update, but
+                * the next tick/schedule should update.
+                *
+                * It will not get called when we go idle, because the idle
+                * thread is a different class (!fair), nor will the utilization
+                * number include things like RT tasks.
+                *
+                * As is, the util number is not freq-invariant (we'd have to
+                * implement arch_scale_freq_capacity() for that).
+                *
+                * See cpu_util().
+                */
+               cpufreq_update_util(rq_of(cfs_rq), 0);
+       }
+}
+
 #ifdef CONFIG_SMP
 /*
  * Approximate:
@@ -3276,29 +3299,6 @@ static inline void set_tg_cfs_propagate(struct cfs_rq *cfs_rq) {}
 
 #endif /* CONFIG_FAIR_GROUP_SCHED */
 
-static inline void cfs_rq_util_change(struct cfs_rq *cfs_rq)
-{
-       if (&this_rq()->cfs == cfs_rq) {
-               /*
-                * There are a few boundary cases this might miss but it should
-                * get called often enough that that should (hopefully) not be
-                * a real problem -- added to that it only calls on the local
-                * CPU, so if we enqueue remotely we'll miss an update, but
-                * the next tick/schedule should update.
-                *
-                * It will not get called when we go idle, because the idle
-                * thread is a different class (!fair), nor will the utilization
-                * number include things like RT tasks.
-                *
-                * As is, the util number is not freq-invariant (we'd have to
-                * implement arch_scale_freq_capacity() for that).
-                *
-                * See cpu_util().
-                */
-               cpufreq_update_util(rq_of(cfs_rq), 0);
-       }
-}
-
 /*
  * Unsigned subtract and clamp on underflow.
  *
@@ -3544,7 +3544,7 @@ update_cfs_rq_load_avg(u64 now, struct cfs_rq *cfs_rq, bool update_freq)
 
 static inline void update_load_avg(struct sched_entity *se, int not_used1)
 {
-       cpufreq_update_util(rq_of(cfs_rq_of(se)), 0);
+       cfs_rq_util_change(cfs_rq_of(se));
 }
 
 static inline void