mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Heiko Carstens <heiko.carstens@de.ibm.com>
To: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>,
	Andrew Morton <akpm@linux-foundation.org>,
	Martin Schwidefsky <schwidefsky@de.ibm.com>,
	Mustafa Mesanovic <mustafa.mesanovic@de.ibm.com>,
	Arjan van de Ven <arjan@infradead.org>,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH/Resend] timer stats: add quick check to avoid function calls
Date: Tue, 23 Jun 2009 17:38:15 +0200	[thread overview]
Message-ID: <20090623153811.GA4641@osiris.boeblingen.de.ibm.com> (raw)
In-Reply-To: <20090623150803.GD13415@elte.hu>

On Tue, Jun 23, 2009 at 05:08:03PM +0200, Ingo Molnar wrote:
> 
> * Heiko Carstens <heiko.carstens@de.ibm.com> wrote:
> 
> > From: Heiko Carstens <heiko.carstens@de.ibm.com>
> > 
> > When the kernel is configured with CONFIG_TIMER_STATS but timer stats
> > are runtime disabled we still get calls to __timer_stats_timer_set_start_info
> > which initializes some fields in the corresponding struct timer_list.
> > 
> > So add some quick checks in the the timer stats setup functions to avoid
> > function calls to __timer_stats_timer_set_start_info when timer stats are
> > disabled.
> > 
> > In an artificial workload that does nothing but playing ping pong with a
> > single tcp packet via loopback this decreases cpu consumption by 1 - 1.5%.
> > 
> > This is part of a modified function trace output on SLES11:
> > 
> > perl-2497  [00] 28630647177732388 [+  125]: sk_reset_timer <-tcp_v4_rcv
> > perl-2497  [00] 28630647177732513 [+  125]: mod_timer <-sk_reset_timer
> > perl-2497  [00] 28630647177732638 [+  125]: __timer_stats_timer_set_start_info <-mod_timer
> > perl-2497  [00] 28630647177732763 [+  125]: __mod_timer <-mod_timer
> > perl-2497  [00] 28630647177732888 [+  125]: __timer_stats_timer_set_start_info <-__mod_timer
> > perl-2497  [00] 28630647177733013 [+   93]: lock_timer_base <-__mod_timer
> 
> makes sense.
> 
> Only one very small detail:

[..]

> > Index: linux-2.6/include/linux/timer.h
> > ===================================================================
> > --- linux-2.6.orig/include/linux/timer.h
> > +++ linux-2.6/include/linux/timer.h
> > @@ -187,6 +187,8 @@ extern unsigned long get_next_timer_inte
> >   */
> >  #ifdef CONFIG_TIMER_STATS
> >  
> > +extern int timer_stats_active;
> 
> Why is this declared twice, once in hrtimer.h and once in timer.h? 
> One should suffice, right?

Actually I only followed the example of timer_stats_update_stats()
which is declared also in both files.
But including timer.h from hrtimer.h doesn't bring up any compile
breakages on the few configs I tested.
Updated patch which avoids the double declaration but adds an
include of timer.h to hrtimer.h

Subject: [PATCH] timer stats: add quick check to avoid function calls

From: Heiko Carstens <heiko.carstens@de.ibm.com>

When the kernel is configured with CONFIG_TIMER_STATS but timer stats
are runtime disabled we still get calls to __timer_stats_timer_set_start_info
which initializes some fields in the corresponding struct timer_list.

So add some quick checks in the the timer stats setup functions to avoid
function calls to __timer_stats_timer_set_start_info when timer stats are
disabled.

In an artificial workload that does nothing but playing ping pong with a
single tcp packet via loopback this decreases cpu consumption by 1 - 1.5%.

This is part of a modified function trace output on SLES11:

perl-2497  [00] 28630647177732388 [+  125]: sk_reset_timer <-tcp_v4_rcv
perl-2497  [00] 28630647177732513 [+  125]: mod_timer <-sk_reset_timer
perl-2497  [00] 28630647177732638 [+  125]: __timer_stats_timer_set_start_info <-mod_timer
perl-2497  [00] 28630647177732763 [+  125]: __mod_timer <-mod_timer
perl-2497  [00] 28630647177732888 [+  125]: __timer_stats_timer_set_start_info <-__mod_timer
perl-2497  [00] 28630647177733013 [+   93]: lock_timer_base <-__mod_timer

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
---

 include/linux/hrtimer.h   |    5 +++++
 include/linux/timer.h     |    4 ++++
 kernel/time/timer_stats.c |   16 ++++++++--------
 kernel/timer.c            |    2 ++
 4 files changed, 19 insertions(+), 8 deletions(-)

Index: linux-2.6/include/linux/hrtimer.h
===================================================================
--- linux-2.6.orig/include/linux/hrtimer.h
+++ linux-2.6/include/linux/hrtimer.h
@@ -21,6 +21,7 @@
 #include <linux/list.h>
 #include <linux/wait.h>
 #include <linux/percpu.h>
+#include <linux/timer.h>
 
 
 struct hrtimer_clock_base;
@@ -447,6 +448,8 @@ extern void timer_stats_update_stats(voi
 
 static inline void timer_stats_account_hrtimer(struct hrtimer *timer)
 {
+	if (likely(!timer->start_pid))
+		return;
 	timer_stats_update_stats(timer, timer->start_pid, timer->start_site,
 				 timer->function, timer->start_comm, 0);
 }
@@ -456,6 +459,8 @@ extern void __timer_stats_hrtimer_set_st
 
 static inline void timer_stats_hrtimer_set_start_info(struct hrtimer *timer)
 {
+	if (likely(!timer_stats_active))
+		return;
 	__timer_stats_hrtimer_set_start_info(timer, __builtin_return_address(0));
 }
 
Index: linux-2.6/include/linux/timer.h
===================================================================
--- linux-2.6.orig/include/linux/timer.h
+++ linux-2.6/include/linux/timer.h
@@ -190,6 +190,8 @@ extern unsigned long get_next_timer_inte
  */
 #ifdef CONFIG_TIMER_STATS
 
+extern int timer_stats_active;
+
 #define TIMER_STATS_FLAG_DEFERRABLE	0x1
 
 extern void init_timer_stats(void);
@@ -203,6 +205,8 @@ extern void __timer_stats_timer_set_star
 
 static inline void timer_stats_timer_set_start_info(struct timer_list *timer)
 {
+	if (likely(!timer_stats_active))
+		return;
 	__timer_stats_timer_set_start_info(timer, __builtin_return_address(0));
 }
 
Index: linux-2.6/kernel/timer.c
===================================================================
--- linux-2.6.orig/kernel/timer.c
+++ linux-2.6/kernel/timer.c
@@ -380,6 +380,8 @@ static void timer_stats_account_timer(st
 {
 	unsigned int flag = 0;
 
+	if (likely(!timer->start_site))
+		return;
 	if (unlikely(tbase_get_deferrable(timer->base)))
 		flag |= TIMER_STATS_FLAG_DEFERRABLE;
 
Index: linux-2.6/kernel/time/timer_stats.c
===================================================================
--- linux-2.6.orig/kernel/time/timer_stats.c
+++ linux-2.6/kernel/time/timer_stats.c
@@ -96,7 +96,7 @@ static DEFINE_MUTEX(show_mutex);
 /*
  * Collection status, active/inactive:
  */
-static int __read_mostly active;
+int __read_mostly timer_stats_active;
 
 /*
  * Beginning/end timestamps of measurement:
@@ -242,7 +242,7 @@ void timer_stats_update_stats(void *time
 	struct entry *entry, input;
 	unsigned long flags;
 
-	if (likely(!active))
+	if (likely(!timer_stats_active))
 		return;
 
 	lock = &per_cpu(lookup_lock, raw_smp_processor_id());
@@ -254,7 +254,7 @@ void timer_stats_update_stats(void *time
 	input.timer_flag = timer_flag;
 
 	spin_lock_irqsave(lock, flags);
-	if (!active)
+	if (!timer_stats_active)
 		goto out_unlock;
 
 	entry = tstat_lookup(&input, comm);
@@ -290,7 +290,7 @@ static int tstats_show(struct seq_file *
 	/*
 	 * If still active then calculate up to now:
 	 */
-	if (active)
+	if (timer_stats_active)
 		time_stop = ktime_get();
 
 	time = ktime_sub(time_stop, time_start);
@@ -368,18 +368,18 @@ static ssize_t tstats_write(struct file 
 	mutex_lock(&show_mutex);
 	switch (ctl[0]) {
 	case '0':
-		if (active) {
-			active = 0;
+		if (timer_stats_active) {
+			timer_stats_active = 0;
 			time_stop = ktime_get();
 			sync_access();
 		}
 		break;
 	case '1':
-		if (!active) {
+		if (!timer_stats_active) {
 			reset_entries();
 			time_start = ktime_get();
 			smp_mb();
-			active = 1;
+			timer_stats_active = 1;
 		}
 		break;
 	default:

  reply	other threads:[~2009-06-23 15:38 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2009-06-23 14:12 Heiko Carstens
2009-06-23 15:08 ` Ingo Molnar
2009-06-23 15:38   ` Heiko Carstens [this message]
2009-06-24  9:47     ` [tip:timers/urgent] timer stats: Optimize by adding " tip-bot for Heiko Carstens

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=20090623153811.GA4641@osiris.boeblingen.de.ibm.com \
    --to=heiko.carstens@de.ibm.com \
    --cc=akpm@linux-foundation.org \
    --cc=arjan@infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=mustafa.mesanovic@de.ibm.com \
    --cc=schwidefsky@de.ibm.com \
    --cc=tglx@linutronix.de \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

Powered by JetHome