From: Peter Zijlstra <peterz@infradead.org>
To: Venkatesh Pallipadi <venki@google.com>
Cc: Damien Wyart <damien.wyart@free.fr>,
Chase Douglas <chase.douglas@canonical.com>,
Ingo Molnar <mingo@elte.hu>,
tmhikaru@gmail.com, Thomas Gleixner <tglx@linutronix.de>,
linux-kernel@vger.kernel.org
Subject: Re: High CPU load when machine is idle (related to PROBLEM: Unusually high load average when idle in 2.6.35, 2.6.35.1 and later)
Date: Fri, 29 Oct 2010 21:42:23 +0200 [thread overview]
Message-ID: <1288381343.1988.12.camel@laptop> (raw)
In-Reply-To: <1288101958.15336.284.camel@twins>
On Tue, 2010-10-26 at 16:05 +0200, Peter Zijlstra wrote:
>
> a1 = a0 * e + a * (1 - e)
>
> a2 = a1 * e + a * (1 - e)
> = (a0 * e + a * (1 - e)) * e + a * (1 - e)
> = a0 * e^2 + a * (1 - e) * (1 + e)
>
> a3 = a2 * e + a * (1 - e)
> = (a0 * e^2 + a * (1 - e) * (1 + e)) * e + a * (1 - e)
> = a0 * e^3 + a * (1 - e) * (1 + e + e^2)
>
> an = a0 * e^n + a * (1 - e) * (1 + e + ... + e^n-1)
> = a0 * e^n + a * (1 - e) * (1 - e^n)/(1 - e)
> = a0 * e^n + a * (1 - e^n)
not tested other than compile, but how does that look...?
---
include/linux/sched.h | 2 +-
kernel/sched.c | 74 +++++++++++++++++++++++++++++++++++++++++++++---
kernel/timer.c | 2 +-
3 files changed, 71 insertions(+), 7 deletions(-)
diff --git a/include/linux/sched.h b/include/linux/sched.h
index b3d07df..f878db3 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -143,7 +143,7 @@ extern unsigned long nr_iowait_cpu(int cpu);
extern unsigned long this_cpu_load(void);
-extern void calc_global_load(void);
+extern void calc_global_load(int ticks);
extern unsigned long get_parent_ip(unsigned long addr);
diff --git a/kernel/sched.c b/kernel/sched.c
index 41f1869..125417e 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -3168,10 +3168,64 @@ calc_load(unsigned long load, unsigned long exp, unsigned long active)
}
/*
+ * fixed_power_int - compute: x^n in O(log n) time
+ * @x: base of the power
+ * frac_bits: fractional bits of @x
+ * @n: power to raise @x to.
+ */
+static unsigned long
+fixed_power_int(unsigned long x, unsigned int frac_bits, unsigned int n)
+{
+ unsigned long result = 1UL << frac_bits;
+
+ while (n) {
+ if (n & 1) {
+ result *= x;
+ result += 1UL << (frac_bits - 1);
+ result >>= frac_bits;
+ }
+ n >>= 1;
+ x *= x;
+ x >>= frac_bits;
+ }
+
+ return result;
+}
+
+/*
+ * a1 = a0 * e + a * (1 - e)
+ *
+ * a2 = a1 * e + a * (1 - e)
+ * = (a0 * e + a * (1 - e)) * e + a * (1 - e)
+ * = a0 * e^2 + a * (1 - e) * (1 + e)
+ *
+ * a3 = a2 * e + a * (1 - e)
+ * = (a0 * e^2 + a * (1 - e) * (1 + e)) * e + a * (1 - e)
+ * = a0 * e^3 + a * (1 - e) * (1 + e + e^2)
+ *
+ * an = a0 * e^n + a * (1 - e) * (1 + e + ... + e^n-1)
+ * = a0 * e^n + a * (1 - e) * (1 - e^n)/(1 - e)
+ * = a0 * e^n + a * (1 - e^n)
+ */
+static unsigned long
+calc_load_n(unsigned long load, unsigned long exp,
+ unsigned long active, unsigned int n)
+{
+ unsigned long en = fixed_power_int(exp, FSHIFT, n);
+
+ load *= en;
+ load += active * (FIXED_1 - en);
+ load += 1UL << (FSHIFT - 1);
+ load >>= FSHIFT;
+
+ return load;
+}
+
+/*
* calc_load - update the avenrun load estimates 10 ticks after the
* CPUs have updated calc_load_tasks.
*/
-void calc_global_load(void)
+void calc_global_load(int ticks)
{
unsigned long upd = calc_load_update + 10;
long active;
@@ -3182,11 +3236,21 @@ void calc_global_load(void)
active = atomic_long_read(&calc_load_tasks);
active = active > 0 ? active * FIXED_1 : 0;
- 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);
+ if (ticks > LOAD_FREQ) {
+ int n = (jiffies - upd) / LOAD_FREQ;
+
+ avenrun[0] = calc_load_n(avenrun[0], EXP_1, active, n);
+ avenrun[1] = calc_load_n(avenrun[1], EXP_5, active, n);
+ avenrun[2] = calc_load_n(avenrun[2], EXP_15, active, n);
- calc_load_update += LOAD_FREQ;
+ calc_load_update += n * LOAD_FREQ;
+ } else {
+ 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;
+ }
}
/*
diff --git a/kernel/timer.c b/kernel/timer.c
index d6ccb90..9f82b2a 100644
--- a/kernel/timer.c
+++ b/kernel/timer.c
@@ -1297,7 +1297,7 @@ void do_timer(unsigned long ticks)
{
jiffies_64 += ticks;
update_wall_time();
- calc_global_load();
+ calc_global_load(ticks);
}
#ifdef __ARCH_WANT_SYS_ALARM
next prev parent reply other threads:[~2010-10-29 19:44 UTC|newest]
Thread overview: 55+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-09-29 7:01 High CPU load when machine is idle Damien Wyart
2010-10-14 14:58 ` High CPU load when machine is idle (related to PROBLEM: Unusually high load average when idle in 2.6.35, 2.6.35.1 and later) Damien Wyart
2010-10-14 15:29 ` Chase Douglas
2010-10-14 15:56 ` Damien Wyart
2010-10-15 11:08 ` Peter Zijlstra
2010-10-18 12:32 ` Peter Zijlstra
2010-10-20 13:27 ` Damien Wyart
2010-10-20 13:30 ` Peter Zijlstra
2010-10-20 13:43 ` Peter Zijlstra
2010-10-20 14:14 ` Peter Zijlstra
2010-10-20 14:25 ` Peter Zijlstra
2010-10-20 17:26 ` Peter Zijlstra
2010-10-20 20:24 ` Damien Wyart
2010-10-21 1:48 ` tmhikaru
2010-10-21 1:53 ` tmhikaru
2010-10-21 8:22 ` Ingo Molnar
2010-10-21 8:57 ` tmhikaru
2010-10-21 18:36 ` tmhikaru
2010-10-22 1:37 ` tmhikaru
2010-10-21 12:09 ` Peter Zijlstra
2010-10-21 17:18 ` Venkatesh Pallipadi
2010-10-22 21:03 ` Venkatesh Pallipadi
2010-10-22 23:03 ` High CPU load when machine is idle (related to PROBLEM: Unusually high load average when idle in 2.6.35, 2.6.35.1 and later) Venkatesh Pallipadi
2010-10-23 2:13 ` tmhikaru
2010-10-25 10:12 ` Peter Zijlstra
2010-10-25 16:29 ` Venkatesh Pallipadi
2010-10-26 12:44 ` Peter Zijlstra
2010-10-26 14:05 ` Peter Zijlstra
2010-10-29 19:42 ` Peter Zijlstra [this message]
2010-11-09 18:55 ` Kyle McMartin
2010-11-09 19:02 ` Peter Zijlstra
2010-11-10 2:37 ` tmhikaru
2010-11-10 12:01 ` Peter Zijlstra
2010-11-10 3:45 ` Kyle McMartin
2010-11-10 12:00 ` Peter Zijlstra
2010-11-14 5:14 ` tmhikaru
2010-11-25 13:31 ` Damien Wyart
2010-11-25 14:03 ` Peter Zijlstra
2010-11-27 20:15 ` Peter Zijlstra
2010-11-28 4:26 ` Kyle McMartin
2010-11-28 11:40 ` Damien Wyart
2010-11-28 18:07 ` Valdis.Kletnieks
2010-11-29 11:38 ` Peter Zijlstra
2010-11-29 19:40 ` tmhikaru
2010-11-29 23:01 ` Peter Zijlstra
2010-11-30 14:59 ` Peter Zijlstra
2010-11-30 15:39 ` Kyle McMartin
2010-11-30 20:04 ` Kyle McMartin
2010-11-30 16:53 ` Damien Wyart
2010-11-30 17:29 ` Peter Zijlstra
2010-12-01 21:27 ` tmhikaru
2010-12-02 10:16 ` tmhikaru
2010-12-08 20:40 ` [tip:sched/urgent] sched: Cure more NO_HZ load average woes tip-bot for Peter Zijlstra
2010-11-30 20:01 ` High CPU load when machine is idle (related to PROBLEM: Unusually high load average when idle in 2.6.35, 2.6.35.1 and later) tmhikaru
2010-11-30 16:49 ` Damien Wyart
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1288381343.1988.12.camel@laptop \
--to=peterz@infradead.org \
--cc=chase.douglas@canonical.com \
--cc=damien.wyart@free.fr \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=tglx@linutronix.de \
--cc=tmhikaru@gmail.com \
--cc=venki@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.