Skip to contentSkip to conversation

Regression: Machine runs really slowly after commit f12a15be

BISECTED/2.6.35+ Priit Laes 3 participants 6 messages Aug 7, 2010 – Aug 9, 2010

Subsystems (3)
lore
Share & tools
lore

Priit Laes

author
#0 lore raw

Following commit makes my Dell Latitude D610 machine (x86_32) run really-really slow...

f12a15be63d1de9a35971f35f06b73088fa25c3a is the first bad commit commit f12a15be63d1de9a35971f35f06b73088fa25c3a

Author: John Stultz <[email protected]>
Date:   Tue Jul 13 17:56:27 2010 -0700

x86: Convert common clocksources to use clocksource_register_hz/khz

This converts the most common of the x86 clocksources over to use clocksource_register_hz/khz.

LKML-Reference: <[email protected]>

:040000 040000 264bcae0fcf10c54412a6c9df83b295e8c7e089b 8fc84146e6f34006cca10681afcc9869362a1cfc M arch
:040000 040000 a5f57efb367b7915d96c50fb123c053e58756423 874a8252ec8f7c9a5422e0b78834981902ad122d M drivers

#1 lore raw
Quoted 5 lines · Priit Laes

Following commit makes my Dell Latitude D610 machine (x86_32) run really-really slow...

f12a15be63d1de9a35971f35f06b73088fa25c3a is the first bad commit commit f12a15be63d1de9a35971f35f06b73088fa25c3a
Author: John Stultz <[email protected]>

Quoted 4 lines · Alexey Fisher

Hallo John,
i have regression after commit f12a15be6.
With this patch my netbook (Asus Eeepc 1005) starting really slow. It looks like printk comes 1/min.

Hi Priit, Alexey,
Thanks so much for the testing and bisecting the issue down!

Could you both please provide the output of:
cat /sys/devices/system/clocksource/clocksource0/current_clocksource cat /sys/devices/system/clocksource/clocksource0/available_clocksource

thanks
-john

#2 lore raw

Am Montag, den 09.08.2010, 12:01 -0700 schrieb john stultz:

Quoted 13 lines · Priit Laes
Quoted 5 lines

Following commit makes my Dell Latitude D610 machine (x86_32) run really-really slow...

f12a15be63d1de9a35971f35f06b73088fa25c3a is the first bad commit commit f12a15be63d1de9a35971f35f06b73088fa25c3a
Author: John Stultz <[email protected]>

Quoted 4 lines · Alexey Fisher

Hallo John,
i have regression after commit f12a15be6.
With this patch my netbook (Asus Eeepc 1005) starting really slow. It looks like printk comes 1/min.

Quoted 6 lines

Hi Priit, Alexey,
Thanks so much for the testing and bisecting the issue down!

Could you both please provide the output of:
cat /sys/devices/system/clocksource/clocksource0/current_clocksource cat /sys/devices/system/clocksource/clocksource0/available_clocksource

cat /sys/devices/system/clocksource/clocksource0/current_clocksource hpet

cat /sys/devices/system/clocksource/clocksource0/available_clocksource hpet acpi_pm

#3 lore raw
Quoted 5 lines · Alexey Fisher

cat /sys/devices/system/clocksource/clocksource0/current_clocksource hpet

cat /sys/devices/system/clocksource/clocksource0/available_clocksource hpet acpi_pm

Quoted 4 lines · Priit Laes

cat /sys/devices/system/clocksource/clocksource0/current_clocksource hpet
cat /sys/devices/system/clocksource/clocksource0/available_clocksource hpet acpi_pm

Ok. Good. Chris Wilson has created the following patch which should hopefully resolve this issue. It would be great if you could try booting with it to verify that there are no other problems lurking here.

Thanks again for the great bug reporting!

thanks
-john

From: Chris Wilson <[email protected]>
Subject: [PATCH] x86/hpet: Use the FSEC_PER_SEC constant for femto-second
periods

The current computation, introduced with f12a15be63, of FSEC_PER_SEC using the multiplication of (FSEC_PER_NSEC * NSEC_PER_SEC) is performed only with 32bit integers on small machines, resulting in an overflow and a *very* short intervals being programmed. An interrupt storm follows.

Note that we also have to specify FSEC_PER_SEC as being long long to overcome the same limitations.

---
 arch/x86/kernel/hpet.c |    4 ++--
 include/linux/time.h   |    2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)
arch/x86/kernel/hpet.c +2 -2
index 33dbcc4..351f9c0 100644
@@ -582,7 +582,7 @@ static void init_one_hpet_msi_clockevent(struct hpet_dev *hdev, int cpu)
 	 * scaled math multiplication factor for nanosecond to hpet tick
 	 * conversion.
 	 */
-	hpet_freq = 1000000000000000ULL;
+	hpet_freq = FSEC_PER_SEC;
 	do_div(hpet_freq, hpet_period);
 	evt->mult = div_sc((unsigned long) hpet_freq,
 				      NSEC_PER_SEC, evt->shift);
@@ -837,7 +837,7 @@ static int hpet_clocksource_register(void)
 	 * cyc/sec = FSEC_PER_SEC/hpet_period(fsec/cyc)
 	 * cyc/sec = (FSEC_PER_NSEC * NSEC_PER_SEC)/hpet_period
 	 */
-	hpet_freq = FSEC_PER_NSEC * NSEC_PER_SEC;
+	hpet_freq = FSEC_PER_SEC;
 	do_div(hpet_freq, hpet_period);
 	clocksource_register_hz(&clocksource_hpet, (u32)hpet_freq);
 
include/linux/time.h +1 -1
index cb34e35..1261270 100644
@@ -38,7 +38,7 @@ extern struct timezone sys_tz;
 #define NSEC_PER_MSEC	1000000L
 #define USEC_PER_SEC	1000000L
 #define NSEC_PER_SEC	1000000000L
-#define FSEC_PER_SEC	1000000000000000L
+#define FSEC_PER_SEC	1000000000000000LL
 
 #define TIME_T_MAX	(time_t)((1UL << ((sizeof(time_t) << 3) - 1)) - 1)
 
Signature · 1 line
1.7.1
#4 lore raw
Quoted 34 lines · john stultz
Quoted 5 lines · Alexey Fisher

cat /sys/devices/system/clocksource/clocksource0/current_clocksource hpet

cat /sys/devices/system/clocksource/clocksource0/available_clocksource hpet acpi_pm

Quoted 4 lines · Priit Laes

cat /sys/devices/system/clocksource/clocksource0/current_clocksource hpet
cat /sys/devices/system/clocksource/clocksource0/available_clocksource hpet acpi_pm

Ok. Good. Chris Wilson has created the following patch which should hopefully resolve this issue. It would be great if you could try booting with it to verify that there are no other problems lurking here.

Thanks again for the great bug reporting!

thanks
-john

From: Chris Wilson <[email protected]>
Subject: [PATCH] x86/hpet: Use the FSEC_PER_SEC constant for femto-second
periods

The current computation, introduced with f12a15be63, of FSEC_PER_SEC using the multiplication of (FSEC_PER_NSEC * NSEC_PER_SEC) is performed only with 32bit integers on small machines, resulting in an overflow and a *very* short intervals being programmed. An interrupt storm follows.

Quoted 7 lines

Note that we also have to specify FSEC_PER_SEC as being long long to overcome the same limitations.

---

This patch work for me. At least it boot.
Tank you.

author
#5 lore raw
In reply to john stultz

Ühel kenal päeval, E, 2010-08-09 kell 12:01, kirjutas john stultz:

Quoted 10 lines · Priit Laes
Quoted 5 lines

Following commit makes my Dell Latitude D610 machine (x86_32) run really-really slow...

f12a15be63d1de9a35971f35f06b73088fa25c3a is the first bad commit commit f12a15be63d1de9a35971f35f06b73088fa25c3a
Author: John Stultz <[email protected]>

Could you both please provide the output of:
cat /sys/devices/system/clocksource/clocksource0/current_clocksource cat /sys/devices/system/clocksource/clocksource0/available_clocksource

This is with kernel 2.6

cat /sys/devices/system/clocksource/clocksource0/current_clocksource hpet
cat /sys/devices/system/clocksource/clocksource0/available_clocksource hpet acpi_pm

Päikest,
Priit ;)