lkml.org 
[lkml]   [2012]   [Jun]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH 2/2] sched: Try fix sampling time line for idle
Date
In our mind per-cpu sampling for cpu idle and non-idle is equal. But
actually may not. For non-idle cpu sampling, it's right the load when
sampling. But for idle, cause of nohz, the sampling will be delayed to
nohz exit(less than 1 tick after nohz exit). Nohz exit is always caused
by processes woken up--non-idle model. It's not fair here. Idle
sampling will be turned to non-idle sampling. And cause loadavg being
higher than normal.

time-expected-sampling
| time-do-sampling
| |
V V
-|-------------------------|--
start_nohz stop_nohz

CC: Peter Zijlstra <peterz@infradead.org>
CC: Doug Smythies <dsmythies@telus.net>
CC: Ingo Molnar <mingo@redhat.com>
CC: Tao Ma <tm@tao.ma>
CC: Sha Zhengju <handai.szj@gmail.com>
Reported-by: Sha Zhengju <handai.szj@gmail.com>
Signed-off-by: Charles Wang <muming.wq@taobao.com>

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 691e7ec..2983838 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -2166,6 +2166,7 @@ unsigned long this_cpu_load(void)
static atomic_long_t calc_load_tasks;
static unsigned long calc_load_update;
static unsigned long calc_mask_update;
+static unsigned long calc_load_time_start = LOAD_FREQ;
unsigned long avenrun[3];
EXPORT_SYMBOL(avenrun);

@@ -2260,6 +2261,7 @@ void calc_load_account_idle(struct rq *this_rq)
long delta;

delta = calc_load_fold_active(this_rq);
+ this_rq->last_idle_enter = jiffies;
if (delta) {
atomic_long_add(delta, &calc_load_tasks_idle);
/*
@@ -2499,6 +2501,8 @@ void prepare_calc_load(void)
static void calc_load_account_active(struct rq *this_rq)
{
long delta;
+ unsigned long delta_time;
+ long last_idle_time_elapse;

if (time_before(jiffies, this_rq->calc_load_update))
return;
@@ -2508,15 +2512,24 @@ static void calc_load_account_active(struct rq *this_rq)
atomic_long_xchg(&calc_unmask_cpu_load_idle, 0);
delta = atomic_long_xchg(&calc_load_tasks_idle, 0);
atomic_long_add(delta, &calc_load_tasks);
+ calc_load_time_start = jiffies;
}
/* mark this cpu as load calculated */
set_cpu_load_update_mask();

+ last_idle_time_elapse = this_rq->last_idle_enter - calc_load_time_start;
+ delta_time = jiffies - this_rq->calc_load_update;
+ if (last_idle_time_elapse > 0)
+ goto out;
+ if ((last_idle_time_elapse > -1) && (delta_time >= 1))
+ goto out;
+
delta = calc_load_fold_active(this_rq);
delta += calc_load_fold_unmask_idle();
if (delta)
atomic_long_add(delta, &calc_load_tasks);

+out:
this_rq->calc_load_update += LOAD_FREQ;
}

diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 4134d37..a356588 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -438,6 +438,7 @@ struct rq {

/* calc_load related fields */
unsigned long calc_load_update;
+ unsigned long last_idle_enter;
long calc_load_active;

#ifdef CONFIG_SCHED_HRTICK
--
1.7.9.5


\
 
 \ /
  Last update: 2012-06-18 10:01    [W:0.032 / U:3.392 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site