6:16 a.m.
Hello,
I got a problem with my (more or less) fresh Fedora 14 installation: The
cpufrequency stays at 1Ghz all the time, even with high cpu usage. I'm
also not able to scale the frequency manually using the applet cpu
scaling inside gnome (If I try this, nothing happens).
The interessting thing about this problem is, that it works ca. 30
minutes after (re)boot, then something crashes and the decribed problem
arises.
I attached a few infos, if you need anything else, please ask.
Thanks in advance
icehawk
7:08 a.m.
Martin Haug <martinhaug <at> piratenpartei.de> writes:
...
First, make sure your system is up-to-date.
$ lsmod |grep -i freq
...
cpufreq_ondemand 7262 2
acpi_cpufreq 6285 1
mperf 1141 1 acpi_cpufreq
...
There are two packages I can choose from:
1.
$ yum info gnome-applets
$ rpm -V gnome-applets
$ rpm -ql gnome-applets
...
/usr/bin/cpufreq-selector
...
$ /usr/bin/cpufreq-selector --help
Add to your panel: CPU Frequency Scaling Monitor.
2.
$ yum info cpufrequtils
$ rpm -V cpufrequtils
$ rpm -ql cpufrequtils
...
/usr/bin/cpufreq-info
/usr/bin/cpufreq-set
...
$ /usr/bin/cpufreq-info --help
$ /usr/bin/cpufreq-set --help
$ apropos cpufreq
cpufreq-info (1) - Utility to retrieve cpufreq kernel information
cpufreq-set (1) - A small tool which allows to modify cpufreq settings.
I think you should be set (you have to be root to set a governor or
frequencies).
JB
7:49 a.m.
JB <jb.1234abcd <at> gmail.com> writes:
...
Of course, make sure that your cpuspeed service
(this script enables/disables processor frequency scaling support, either using
the cpuspeed daemon or in-kernel frequency scaling support)
is enabled and running on your 2,3,4,5 runlevels.
You can do it manually (chkconfig --help) or via System - Administration -
Services.
JB
11:46 a.m.
On 12/27/2010 02:08 PM, JB wrote:
First, make sure your system is up-to-date.
It is.
Here the Infos you requested:
[icehawk@icehawk-laptop linuxhome]$ lsmod |grep -i freq
cpufreq_ondemand 7262 0
acpi_cpufreq 6285 0
mperf 1141 1 acpi_cpufreq
[icehawk@icehawk-laptop linuxhome]$ sudo cpufreq-info
cpufrequtils 008: cpufreq-info (C) Dominik Brodowski 2004-2009
Report errors and bugs to cpufreq(a)vger.kernel.org, please.
analyzing CPU 0:
driver: acpi-cpufreq
CPUs which run at the same hardware frequency: 0 1
CPUs which need to have their frequency coordinated by software: 0
maximum transition latency: 10.0 us.
hardware limits: 1000 MHz - 2.00 GHz
available frequency steps: 2.00 GHz, 1.67 GHz, 1.33 GHz, 1000 MHz
available cpufreq governors: ondemand, userspace, performance
current policy: *frequency should be within 1000 MHz and 1000 MHz.*
The governor "userspace" may decide which speed to use
within this range.
current CPU frequency is 1000 MHz (asserted by call to hardware).
analyzing CPU 1:
driver: acpi-cpufreq
CPUs which run at the same hardware frequency: 0 1
CPUs which need to have their frequency coordinated by software: 1
maximum transition latency: 10.0 us.
hardware limits: 1000 MHz - 2.00 GHz
available frequency steps: 2.00 GHz, 1.67 GHz, 1.33 GHz, 1000 MHz
available cpufreq governors: ondemand, userspace, performance
current policy: *frequency should be within 1000 MHz and 1000 MHz*.
The governor "userspace" may decide which speed to use
within this range.
current CPU frequency is 1000 MHz (asserted by call to hardware).
[icehawk@icehawk-laptop linuxhome]$ sudo cpufreq-set -c 1 --g userspace
-u 2000000
[icehawk@icehawk-laptop linuxhome]$ sudo cpufreq-set -c 0 --g userspace
-u 2000000
[icehawk@icehawk-laptop linuxhome]$ sudo cpufreq-info
cpufrequtils 008: cpufreq-info (C) Dominik Brodowski 2004-2009
Report errors and bugs to cpufreq(a)vger.kernel.org, please.
analyzing CPU 0:
driver: acpi-cpufreq
CPUs which run at the same hardware frequency: 0 1
CPUs which need to have their frequency coordinated by software: 0
maximum transition latency: 10.0 us.
hardware limits: 1000 MHz - 2.00 GHz
available frequency steps: 2.00 GHz, 1.67 GHz, 1.33 GHz, 1000 MHz
available cpufreq governors: ondemand, userspace, performance
current policy: *frequency should be within 1000 MHz and 1000 MHz*.
The governor "userspace" may decide which speed to use
within this range.
current CPU frequency is 1000 MHz (asserted by call to hardware).
analyzing CPU 1:
driver: acpi-cpufreq
CPUs which run at the same hardware frequency: 0 1
CPUs which need to have their frequency coordinated by software: 1
maximum transition latency: 10.0 us.
hardware limits: 1000 MHz - 2.00 GHz
available frequency steps: 2.00 GHz, 1.67 GHz, 1.33 GHz, 1000 MHz
available cpufreq governors: ondemand, userspace, performance
current policy: *frequency should be within 1000 MHz and 1000 MHz.*
The governor "userspace" may decide which speed to use
within this range.
current CPU frequency is 1000 MHz (asserted by call to hardware).
[icehawk@icehawk-laptop linuxhome]$ stress -t 30 -c 2 &
[1] 22884
[icehawk@icehawk-laptop linuxhome]$ stress: info: [22884] dispatching
hogs: 2 cpu, 0 io, 0 vm, 0 hdd
[icehawk@icehawk-laptop linuxhome]$ grep -i mhz /proc/cpuinfo
cpu MHz : 1000.000
cpu MHz : 1000.000
[icehawk@icehawk-laptop linuxhome]$ grep -i mhz /proc/cpuinfo
cpu MHz : 1000.000
cpu MHz : 1000.000
[icehawk@icehawk-laptop linuxhome]$ stress: info: [22884] successful run
completed in 30s
mfg
icehawk
1:40 p.m.
Martin Haug <martinhaug <at> piratenpartei.de> writes:
...
[icehawk <at> icehawk-laptop linuxhome]$ sudo cpufreq-set -c 1 --g userspace
-u 2000000
[icehawk <at> icehawk-laptop linuxhome]$ sudo cpufreq-set -c 0 --g userspace
-u 2000000
[icehawk <at> icehawk-laptop linuxhome]$ sudo cpufreq-info
...
current policy: *frequency should be within 1000 MHz and 1000 MHz*.
...
This is not right for both CPUs.
It should be in your case:
current policy: *frequency should be within 1000 MHz and 2000 MHz*.
So, I am testing and verifying it on my machine (also Intel dual core).
Please do the following:
# /etc/init.d/cpuspeed stop
# lsmod |grep -i cpu
acpi_cpufreq 6285 1
mperf 1141 1 acpi_cpufreq
Note: no cpufreq_ondemand module
# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
userspace
# cpufreq-set -c 0 -u 2GHz
# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq
2000000
# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq
1000000
# cpufreq-set -c 1 -u 2GHz
# cat /sys/devices/system/cpu/cpu1/cpufreq/scaling_max_freq
2000000
# cat /sys/devices/system/cpu/cpu1/cpufreq/scaling_min_freq
1000000
Verify:
# cpufreq-info
...
current policy: frequency should be within 1000 MHz and 2 GHz.
The governor "userspace" may decide which speed to use
within this range.
...
You should get this for both CPUs.
Now, set your desired working freqency ('cpufreq-set --help'):
# cpufreq-set -c 0 -f 1.5GHz
# cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_setspeed
1500000
# cpufreq-set -c 1 -f 1.5GHz
# cat /sys/devices/system/cpu/cpu1/cpufreq/scaling_setspeed
1500000
Verify:
# grep -i mhz /proc/cpuinfo
cpu MHz : 1500.000
cpu MHz : 1500.000
# cpufreq-info
...
current policy: frequency should be within 1000 MHz and 2 GHz.
The governor "userspace" may decide which speed to use
within this range.
current CPU frequency is 1.50 GHz (asserted by call to hardware).
...
You should get this for both CPUs.
Now do your stress test :-)
PS.
Some docs to read:
$ yum install kernel-doc
$ less /usr/share/doc/kernel-doc-2.6.35.10/Documentation/cpu-freq/user-guide.txt
...
2.1 Policy
----------
On these systems, all you can do is select the lower and upper
frequency limit as well as whether you want more aggressive
power-saving or more instantly available processing power.
2.2 Governor
------------
On all other cpufreq implementations, these boundaries still need to
be set. Then, a "governor" must be selected. Such a "governor"
decides
what speed the processor shall run within the boundaries. One such
"governor" is the "userspace" governor. This one allows the user - or
a yet-to-implement userspace program - to decide what specific speed
the processor shall run at.
3. How to change the CPU cpufreq policy and/or speed
...
If you have selected the "userspace" governor which allows you to
set the CPU operating frequency to a specific value, you can read out
the current frequency in
scaling_setspeed. By "echoing" a new frequency into this
you can change the speed of the CPU,
but only within the limits of
scaling_min_freq and scaling_max_freq.
JB
2:37 p.m.
Martin Haug <martinhaug <at> piratenpartei.de> writes:
...
I would prefer that you make these entries manually and verify each step
immediatelly (what you show here does not feel right ...).
[icehawk <at> icehawk-laptop linuxhome]$ sudo
/etc/init.d/cpuspeed stop
[icehawk <at> icehawk-laptop linuxhome]$ sudo lsmod|grep -i cpu
cpufreq_ondemand 7262 0 <====================================???????
acpi_cpufreq 6285 0
mperf 1141 1 acpi_cpufreq
[icehawk <at> icehawk-laptop linuxhome]$ cat
/sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
userspace
<========================================================== OK
The above is contradictory...
What you should get is:
# /etc/init.d/cpuspeed stop
# /etc/init.d/cpuspeed status
cpuspeed is stopped
# lsmod |grep -i cpu
acpi_cpufreq 6285 1
mperf 1141 1 acpi_cpufreq
This is the package:
$ yum list installed cpuspeed
cpuspeed.i686 1:1.5-14.fc14 @anaconda-InstallationRepo-201010031452.i386
$ rpm -ql cpuspeed
/etc/rc.d/init.d/cpuspeed
/etc/sysconfig/cpuspeed
/usr/sbin/cpuspeed
/usr/share/doc/cpuspeed-1.5
/usr/share/doc/cpuspeed-1.5/CHANGES
/usr/share/doc/cpuspeed-1.5/CONTRIBUTORS
/usr/share/doc/cpuspeed-1.5/EXAMPLES
/usr/share/doc/cpuspeed-1.5/FEATURES
/usr/share/doc/cpuspeed-1.5/README
/usr/share/doc/cpuspeed-1.5/USAGE
/usr/share/doc/cpuspeed-1.5/license.txt
/usr/share/man/man8/cpuspeed.8.gz
So,
$ rpm -V cpuspeed
Take a look at
/etc/sysconfig/cpuspeed
which should not have been modified (normally) - it is a config file.
If you are in doubt, reinstall this package (nor problem...):
# yum reinstall cpuspeed
and start all over as I showed above.
After that things should just work ...
> [icehawk <at> icehawk-laptop linuxhome]$ sudo cpufreq-set -c 0 -u 2GHz
...
There is always a possibility that some CPUs do not work (such note is in cpu
freq docs), but it is not obvious why in this case.
JB
4:46 p.m.
JB <jb.1234abcd <at> gmail.com> writes:
...
There is a similar case on fedora.testers mailing list
http://article.gmane.org/gmane.linux.redhat.fedora.testers/86733
It may be a case of outdated or misconfigured BIOS, hardware related.
Take a look at it.
JB
1:54 a.m.
On 28.12.2010, JB wrote:
It may be a case of outdated or misconfigured BIOS, hardware related.
He might be affected by this phenomenon:
https://bugzilla.kernel.org/show_bug.cgi?id=19702
4:56 a.m.
On 12/27/2010 09:37 PM, JB wrote:
Martin Haug <martinhaug <at> piratenpartei.de> writes:
> ...
I would prefer that you make these entries manually and verify each step
immediatelly (what you show here does not feel right ...).
I do that, the line
"script has startet/ended, file is $whatever" comes
from the script-command, which starts a shell and records everything
that is printed on the screen. I do not start a shellscript for the
commands you give me.
So maybe we are near the root of the problem? :-)
Take a look at
/etc/sysconfig/cpuspeed
which should not have been modified (normally) - it is a config file.
It wasn't
modified, but I tried to set the min_speed and max_speed
variables manually, unfortunately that doesn't help. :-(
If you are in doubt, reinstall this package (nor problem...):
# yum reinstall cpuspeed
and start all over as I showed above.
Did that, results are attached.
There is always a possibility that some CPUs do not work (such note
is in cpu
freq docs), but it is not obvious why in this case.
I may forgot to mention it, but
this isn't my first install of Fedora 14
on this machine: I had Fedora 14 before and it worked pretty well, then
I accidently deleted the root-partition and after reinstallation I got
the discussed Problem. So CPU works, BIOS stayed the same, too. (But I
will try to do a BIOS-Update)
mfg
icehawk
7:08 a.m.
Martin Haug <martinhaug <at> piratenpartei.de> writes:
...
I may forgot to mention it, but this isn't my first install of Fedora 14
on this machine: I had Fedora 14 before and it worked pretty well, then
I accidently deleted the root-partition and after reinstallation I got
the discussed Problem. So CPU works, BIOS stayed the same, too. (But I
will try to do a BIOS-Update)
...
In connection with the reinstallation, if you maintained some dirs like /home
on separate partitions and they were reused, it is possible that something
got screwed up and saved on them during the crash (root / deletion).
Besides that, after reading that Bugzilla discussion (post by Heinz), where
some guy (at the end of comments as I looed at them yesterday) fixed his
machine just by changing his BIOS freq to performance instead of on-demand,
and looking at my own BIOS's Power settings, with all those options having to
do with SpeedStep (freq), thermal control, etc, I think you should try that as
well.
Besides, settings BIOS may unlock some ACPI data read by kernel that will
allow you to make the desired changes (governor, freq values) thru cpufreq
entries later on.
Another possibility is, after you update the BIOS, to reinstall F14 fresh, with
reformatting partitions, and not inheriting any separate dirs/partitions if
applicable.
Of course, make sure you do not have any problems with harddisk - after a crash
things happen.
Before reformatting partition, select disk health check in anaconda.
Install smartmontools package and run a full test.
One more - if you have space, install another distro and see if things work
there.
JB
4859
days inactive
4862
days old
10 comments
3 participants
participants (3)
-
Heinz Diehl -
JB -
Martin Haug