cpudyn sets wrong scaling_min_freq if "-minf" option is given on some CPU's

[hm]

the core of the problem is that the array of numbers scaling_available_frequencies does not have to be sorted in increasing order by the kernel. But cpudyn in the current ubuntu package supposed that it is.

This bug was already solved by the author. Ubuntu just needs to update the package to the next version from the author's web page http://mnm.uib.es/~gallir/cpudyn/

[Achim Bohnet (allee)]

On Wednesday 09 March 2005 11:23, Hermann Schwarting wrote:
> Package: cpudyn
> Version: 1.0-2
> Severity: normal

Strange! After a fresh boot:

allee(0) ~ $ cat /proc/cpuinfo | grep MH
cpu MHz : 598.259
allee(0) ~ $ /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq
bash: /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq: Permission denied
allee(126) ~ $ /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq
allee(1) ~ $
allee(1) ~ $
allee(1) ~ $
allee(1) ~ $ cat /proc/cpuinfo | grep MH
cpu MHz : 1794.779
allee(0) ~ $ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq
600000
allee(0) ~ $ cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_frequencies
1800000 1800000 1600000 1400000 1200000 1000000 800000 600000
allee(0) ~ $ cat /proc/cpuinfo | grep MH
cpu MHz : 1794.779

min_freq was right at the start. Even with available_frequencies in reverse
order. But ...

... for whatever reason the cpufreq now sticks on 1.8 GHz. (top shows 0.16
average load and 92%idle) There's nothing in kernlog.

It looks like trying to execute /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq
(forgot the cat) as a normal user. Pushed freq to stay at 1.8GHz. Of course this can
be triggered by a background system daemon too ;)

Debian sarge with debian kernel from sid 2.6.10-1-686 (2.6.10-4) on Dell D600
laptop.

allee(0) ~ $ cat /proc/cpuinfo
processor : 0
vendor_id : GenuineIntel
cpu family : 6
model : 13
model name : Intel(R) Pentium(R) M processor 1.80GHz
stepping : 6
cpu MHz : 1794.779
cache size : 2048 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr mce cx8 sep mtrr pge mca cmov pat clflush dts acpi mmx fxsr sse sse2 ss tm pbe est tm2
bogomips : 3555.32

--
  To me vi is Zen. To use vi is to practice zen. Every command is
  a koan. Profound to the user, unintelligible to the uninitiated.
  You discover truth everytime you use it.
                                      -- <email address hidden>

[Achim Bohnet (allee)]

I have to correct myself. Login out of my KDE session and
cpufreq in back to 600 MHz. So it's not cpudyn fault.

FWIW: in KDE: top shows load 0.2, cpu is 92% idle so it's not clear
why cpufreq turn to max freq with default cpudyn settings:

 $ ps -ef | grep cpudy[n]
 root 5328 1 0 13:48 ? 00:00:00 /usr/sbin/cpudynd -i 1 -p 0.5 0.9 -l 7

Sorry incomplete infos.

Achim

--
  To me vi is Zen. To use vi is to practice zen. Every command is
  a koan. Profound to the user, unintelligible to the uninitiated.
  You discover truth everytime you use it.
                                      -- <email address hidden>

[Ricardo Galli (gallir)]

On Wednesday 09 March 2005 13:39, Achim Bohnet shaped the electrons to
shout:
> On Wednesday 09 March 2005 11:23, Hermann Schwarting wrote:
> > Package: cpudyn
> > Version: 1.0-2
> > Severity: normal
>
> Strange! After a fresh boot:

Should be fixed in the last version (1.0.1):
http://mnm.uib.es/gallir/cpudyn/

Thanks for reporting it.

--
  ricardo galli GPG id C8114D34
  http://mnm.uib.es/gallir/
  We all live in a #FFFF00 submarine

[hm (colimit)]

the core of the problem is that the array of numbers scaling_available_frequencies does not have to be sorted in increasing order by the kernel. But cpudyn in the current ubuntu package supposed that it is.

This bug was already solved by the author. Ubuntu just needs to update the package to the next version from the author's web page http://mnm.uib.es/~gallir/cpudyn/

[Daniel Robitaille (robitaille)]

According to a comment in the Debian bug report, this is fixed in version 1.0.1 (currently not available in either Ubuntu or Debian)

[hm (colimit)]

Well, I hope you read my initial post. I already said that it was fixed by
the author, but this fix is not reflected in the Ubuntu distribution. You
didn't have to dig into Debian bug reports, just read my post.

The update is not yet in Debian. So what? Ubuntu declares itself as "we
adopt new releases of the packages much faster than Debian". So, initialy I
said, and I can repeat it again: "please use the latest version of cpudyn.
It has the bugs fixed". Sounds simple, and people don't have to debug cpudyn
themselves and figure out that the latest version fixes the bug.

On 4/19/06, Daniel Robitaille <email address hidden> wrote:
>
> According to a comment in the Debian bug report, this is fixed in
> version 1.0.1 (currently not available in either Ubuntu or Debian)
>
>
> ** Bug watch added: Debian Bug tracker #298691
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=298691
>
> ** Also affects: cpudyn (Debian) via
> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=298691
> Severity: Unknown
> Priority: Unknown
> Status: Unknown
>
> --
> cpudyn sets wrong scaling_min_freq if "-minf" option is given on some
> CPU's
> https://launchpad.net/bugs/5385
>

[Carthik Sharma (carthik)]

Confirmed, bug present in Dapper.

Note to reporter: It is normal procedure to see if the packages have been updated in Debian, and then synch with the debian package. Thanks for reporting the bug.

[dino99 (9d9)]

That package is no more an Ubuntu one.