From: Martin Schwidefsky <schwidefsky@de.ibm.com>
To: linux-arch@vger.kernel.org
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>,
Paul Mackerras <paulus@samba.org>,
Benjamin Herrenschmidt <benh@kernel.crashing.org>,
Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>,
Tony Luck <tony.luck@intel.com>,
Jeremy Fitzhardinge <jeremy@xensource.com>,
Chris Wright <chrisw@sous-sol.org>,
Michael Neuling <mikey@neuling.org>,
Martin Schwidefsky <schwidefsky@de.ibm.com>
Subject: [patch 3/4] improve precision of idle accounting.
Date: Wed, 08 Oct 2008 18:20:01 +0200 [thread overview]
Message-ID: <20081008162144.772227050@de.ibm.com> (raw)
In-Reply-To: 20081008161958.767142939@de.ibm.com
[-- Attachment #1: 204-cputime-idle-stck.diff --]
[-- Type: text/plain, Size: 5683 bytes --]
From: Martin Schwidefsky <schwidefsky@de.ibm.com>
Increase the precision of the idle time calculation that is exported
to user space via /sys/devices/system/cpu/cpu<x>/idle_time_us
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/include/asm/cpu.h | 3 -
arch/s390/kernel/process.c | 67 +++++++++++++++++++++++++++++---------------
arch/s390/kernel/smp.c | 25 ++++++++--------
3 files changed, 59 insertions(+), 36 deletions(-)
Index: linux-idle/arch/s390/include/asm/cpu.h
===================================================================
--- linux-idle.orig/arch/s390/include/asm/cpu.h
+++ linux-idle/arch/s390/include/asm/cpu.h
@@ -14,7 +14,6 @@
struct s390_idle_data {
spinlock_t lock;
- unsigned int in_idle;
unsigned long long idle_count;
unsigned long long idle_enter;
unsigned long long idle_time;
@@ -26,7 +25,7 @@ void s390_idle_leave(void);
static inline void s390_idle_check(void)
{
- if ((&__get_cpu_var(s390_idle))->in_idle)
+ if ((&__get_cpu_var(s390_idle))->idle_enter != 0ULL)
s390_idle_leave();
}
Index: linux-idle/arch/s390/kernel/process.c
===================================================================
--- linux-idle.orig/arch/s390/kernel/process.c
+++ linux-idle/arch/s390/kernel/process.c
@@ -38,6 +38,7 @@
#include <linux/utsname.h>
#include <linux/tick.h>
#include <linux/elfcore.h>
+#include <linux/kernel_stat.h>
#include <asm/uaccess.h>
#include <asm/pgtable.h>
#include <asm/system.h>
@@ -79,30 +80,19 @@ DEFINE_PER_CPU(struct s390_idle_data, s3
.lock = __SPIN_LOCK_UNLOCKED(s390_idle.lock)
};
-static int s390_idle_enter(void)
+void s390_idle_leave(void)
{
struct s390_idle_data *idle;
+ unsigned long long idle_time;
idle = &__get_cpu_var(s390_idle);
+ idle_time = S390_lowcore.int_clock - idle->idle_enter;
spin_lock(&idle->lock);
+ idle->idle_time += idle_time;
+ idle->idle_enter = 0ULL;
idle->idle_count++;
- idle->in_idle = 1;
- idle->idle_enter = get_clock();
spin_unlock(&idle->lock);
- vtime_stop_cpu_timer();
- return NOTIFY_OK;
-}
-
-void s390_idle_leave(void)
-{
- struct s390_idle_data *idle;
-
vtime_start_cpu_timer();
- idle = &__get_cpu_var(s390_idle);
- spin_lock(&idle->lock);
- idle->idle_time += get_clock() - idle->idle_enter;
- idle->in_idle = 0;
- spin_unlock(&idle->lock);
}
extern void s390_handle_mcck(void);
@@ -111,16 +101,16 @@ extern void s390_handle_mcck(void);
*/
static void default_idle(void)
{
+ struct s390_idle_data *idle = &__get_cpu_var(s390_idle);
+ unsigned long addr;
+ psw_t psw;
+
/* CPU is going idle. */
local_irq_disable();
if (need_resched()) {
local_irq_enable();
return;
}
- if (s390_idle_enter() == NOTIFY_BAD) {
- local_irq_enable();
- return;
- }
#ifdef CONFIG_HOTPLUG_CPU
if (cpu_is_offline(smp_processor_id())) {
preempt_enable_no_resched();
@@ -136,9 +126,42 @@ static void default_idle(void)
return;
}
trace_hardirqs_on();
+ vtime_stop_cpu_timer();
+
+ /*
+ * The inline assembly is equivalent to
+ * idle->idle_enter = get_clock();
+ * __load_psw_mask(psw_kernel_bits | PSW_MASK_WAIT |
+ * PSW_MASK_IO | PSW_MASK_EXT);
+ * The difference is that the inline assembly makes sure that
+ * the stck instruction is right before the lpsw instruction.
+ * This is done to increase the precision.
+ */
+
/* Wait for external, I/O or machine check interrupt. */
- __load_psw_mask(psw_kernel_bits | PSW_MASK_WAIT |
- PSW_MASK_IO | PSW_MASK_EXT);
+ psw.mask = psw_kernel_bits|PSW_MASK_WAIT|PSW_MASK_IO|PSW_MASK_EXT;
+#ifndef __s390x__
+ asm volatile(
+ " basr %0,0\n"
+ "0: ahi %0,1f-0b\n"
+ " st %0,4(%2)\n"
+ " stck 0(%3)\n"
+ " lpsw 0(%2)\n"
+ "1:"
+ : "=&d" (addr), "=m" (idle->idle_enter)
+ : "a" (&psw), "a" (&idle->idle_enter), "m" (psw)
+ : "memory", "cc");
+#else /* __s390x__ */
+ asm volatile(
+ " larl %0,1f\n"
+ " stg %0,8(%2)\n"
+ " stck 0(%3)\n"
+ " lpswe 0(%2)\n"
+ "1:"
+ : "=&d" (addr), "=m" (idle->idle_enter)
+ : "a" (&psw), "a" (&idle->idle_enter), "m" (psw)
+ : "memory", "cc");
+#endif /* __s390x__ */
}
void cpu_idle(void)
Index: linux-idle/arch/s390/kernel/smp.c
===================================================================
--- linux-idle.orig/arch/s390/kernel/smp.c
+++ linux-idle/arch/s390/kernel/smp.c
@@ -992,9 +992,11 @@ static ssize_t show_idle_count(struct sy
unsigned long long idle_count;
idle = &per_cpu(s390_idle, dev->id);
- spin_lock_irq(&idle->lock);
+ spin_lock(&idle->lock);
idle_count = idle->idle_count;
- spin_unlock_irq(&idle->lock);
+ if (idle->idle_enter)
+ idle_count++;
+ spin_unlock(&idle->lock);
return sprintf(buf, "%llu\n", idle_count);
}
static SYSDEV_ATTR(idle_count, 0444, show_idle_count, NULL);
@@ -1003,18 +1005,17 @@ static ssize_t show_idle_time(struct sys
struct sysdev_attribute *attr, char *buf)
{
struct s390_idle_data *idle;
- unsigned long long new_time;
+ unsigned long long now, idle_time, idle_enter;
idle = &per_cpu(s390_idle, dev->id);
- spin_lock_irq(&idle->lock);
- if (idle->in_idle) {
- new_time = get_clock();
- idle->idle_time += new_time - idle->idle_enter;
- idle->idle_enter = new_time;
- }
- new_time = idle->idle_time;
- spin_unlock_irq(&idle->lock);
- return sprintf(buf, "%llu\n", new_time >> 12);
+ spin_lock(&idle->lock);
+ now = get_clock();
+ idle_time = idle->idle_time;
+ idle_enter = idle->idle_enter;
+ if (idle_enter != 0ULL && idle_enter < now)
+ idle_time += now - idle_enter;
+ spin_unlock(&idle->lock);
+ return sprintf(buf, "%llu\n", idle_time >> 12);
}
static SYSDEV_ATTR(idle_time_us, 0444, show_idle_time, NULL);
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
next prev parent reply other threads:[~2008-10-08 16:29 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-10-08 16:19 [patch 0/4] [RFC] true vs. system idle cputime Martin Schwidefsky
2008-10-08 16:19 ` [patch 1/4] fix scaled & unscaled cputime accounting Martin Schwidefsky
2008-10-16 4:31 ` Paul Mackerras
2008-10-08 16:20 ` [patch 2/4] idle " Martin Schwidefsky
2008-10-16 4:59 ` Paul Mackerras
2008-10-16 6:42 ` Martin Schwidefsky
2008-10-16 9:08 ` Martin Schwidefsky
2008-10-08 16:20 ` Martin Schwidefsky [this message]
2008-10-08 16:20 ` [patch 4/4] improve " Martin Schwidefsky
2008-10-08 21:22 ` [patch 0/4] [RFC] true vs. system idle cputime Luck, Tony
2008-10-09 8:03 ` Martin Schwidefsky
2008-10-15 14:01 ` Martin Schwidefsky
2008-10-15 20:56 ` Benjamin Herrenschmidt
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=20081008162144.772227050@de.ibm.com \
--to=schwidefsky@de.ibm.com \
--cc=benh@kernel.crashing.org \
--cc=chrisw@sous-sol.org \
--cc=heiko.carstens@de.ibm.com \
--cc=jeremy@xensource.com \
--cc=linux-arch@vger.kernel.org \
--cc=mikey@neuling.org \
--cc=paulus@samba.org \
--cc=seto.hidetoshi@jp.fujitsu.com \
--cc=tony.luck@intel.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).