sched/fair: Fix load_above_capacity fixed point arithmetic width

Since commit:

  2159197d66 ("sched/core: Enable increased load resolution on 64-bit kernels")

we now have two different fixed point units for load.

load_above_capacity has to have 10 bits fixed point unit like PELT,
whereas NICE_0_LOAD has 20 bit fixed point unit on 64-bit kernels.

Fix this by scaling down NICE_0_LOAD when multiplying
load_above_capacity with it.

Signed-off-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Vincent Guittot <vincent.guittot@linaro.org>
Acked-by: Morten Rasmussen <morten.rasmussen@arm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Yuyang Du <yuyang.du@intel.com>
Link: http://lkml.kernel.org/r/1470824847-5316-1-git-send-email-dietmar.eggemann@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This commit is contained in:
Dietmar Eggemann 2016-08-10 11:27:27 +01:00 committed by Ingo Molnar
parent d8206bb3ff
commit 2665621506
1 changed files with 1 additions and 1 deletions

View File

@ -7193,7 +7193,7 @@ static inline void calculate_imbalance(struct lb_env *env, struct sd_lb_stats *s
load_above_capacity = busiest->sum_nr_running * SCHED_CAPACITY_SCALE; load_above_capacity = busiest->sum_nr_running * SCHED_CAPACITY_SCALE;
if (load_above_capacity > busiest->group_capacity) { if (load_above_capacity > busiest->group_capacity) {
load_above_capacity -= busiest->group_capacity; load_above_capacity -= busiest->group_capacity;
load_above_capacity *= NICE_0_LOAD; load_above_capacity *= scale_load_down(NICE_0_LOAD);
load_above_capacity /= busiest->group_capacity; load_above_capacity /= busiest->group_capacity;
} else } else
load_above_capacity = ~0UL; load_above_capacity = ~0UL;