Bug 283646 – sys-kernel/gentoo-sources-2.6.30-r5 cpufreq can't set governor to ondemand

• Product: Gentoo Linux
• Status: RESOLVED INVALID

[Description g0th1c 2009-09-04 10:31:05 UTC]

Hi, I have p4_clockmod in use:

# cat /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 15
model           : 3
model name      : Intel(R) Pentium(R) 4 CPU 2.80GHz
stepping        : 4
cpu MHz         : 2799.930
cache size      : 1024 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 5
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe constant_tsc pebs bts pni dtes64 monitor ds_cpl cid xtpr
bogomips        : 5612.91
clflush size    : 64
power management:

Affected cpus:

# cat /sys/devices/system/cpu/cpu0/cpufreq/affected_cpus
0

probably all good, but:



Reproducible: Always

Steps to Reproduce:
1. # cpufreq-set -g ondemand
2. # cpufreq-info
cpufrequtils 005: cpufreq-info (C) Dominik Brodowski 2004-2006
Report errors and bugs to [e-mail], please.
analyzing CPU 0:
  driver: p4-clockmod
  CPUs which need to switch frequency at the same time: 0
  hardware limits: 350 MHz - 2.80 GHz
  available frequency steps: 350 MHz, 700 MHz, 1.05 GHz, 1.40 GHz, 1.75 GHz, 2.10 GHz, 2.45 GHz, 2.80 GHz
  available cpufreq governors: ondemand, performance
  current policy: frequency should be within 350 MHz and 2.80 GHz.
                  The governor "performance" may decide which speed to use
                  within this range.
  current CPU frequency is 2.80 GHz (asserted by call to hardware).

or in this style:

1. # cat scaling_governor
performance
2. # echo ondemand > scaling_governor    
3. # cat scaling_governor
performance

Userspace governor is unlike ondemand:

1. # cat scaling_governor
performance
2. # echo userspace > scaling_governor
3. # cat scaling_governor
userspace


Actual Results:  
Governors is work good: performance, userspace, powersave.
Governors is not work: ondemand, conservative.


Please help me.

[Comment 1 Mike Pagano 2009-09-04 20:59:37 UTC]

post your .config

[Comment 2 g0th1c 2009-09-05 16:03:34 UTC]

Created attachment 203202 [details]
my kernel config

All need governor is enabled.

[Comment 3 George Kadianakis (RETIRED) 2009-09-08 13:48:44 UTC]

I did some brief research but I ran out of time and I can't read on how cpufreq/p4-clockmod/cpufreq-governors work. I'll just spit the links here 'till I find some time to study the issue.

http://bugs.archlinux.org/task/15284
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=536183
http://kerneltrap.org/mailarchive/linux-kernel/2008/6/2/1999844
https://bugzilla.redhat.com/show_bug.cgi?id=474499

It seems that p4-clockmod and the ondemand governor are not really cooperating with each other, but I'm still not sure on the why and how.

[Comment 4 George Kadianakis (RETIRED) 2009-09-16 14:49:26 UTC]

Okie, the p4-clockmod driver does not use governors (and never did), it talks directly with the ACPI and acts accordingly (see: https://bugzilla.redhat.com/show_bug.cgi?id=474499#c2).

The illusion of a regression bug is created because in 2.6.29 p4-clockmod no longer exports a cpufreq interface to sysfs which ondemand was using all that time to be set.
The thing is that p4-clockmod never worked according to ondemand's rules whatsoever (see: http://www.codemonkey.org.uk/2009/01/18/forthcoming-p4clockmod/)

Because of the above, I'm gonna mark this bug as INVALID.

Thanks for your bug report :)

[Comment 5 George Kadianakis (RETIRED) 2009-10-04 04:04:43 UTC]

*** Bug 287463 has been marked as a duplicate of this bug. ***

[Comment 6 Robert Bradbury 2010-06-15 11:20:43 UTC]

I believe this bug should be reopened for reasons I have outlined in Bug #287463.

I may not fully understand how this works but would be inclined to disagree with Comment #4 because the acpi-cpufreq module cannot load due to errors with the acpi/processor_perflib functions due to insufficient support in the ACPI BIOS for cpu frequency scaling.  On a CPU without Intel "Enhanced" SpeedStep(R) (and the associated ACPI BIOS functionality) the correct way to control power consumption seems to be to use the ondemand or conservative governors.

[Comment 7 George Kadianakis (RETIRED) 2010-06-17 04:19:31 UTC]

(In reply to comment #6)
> I believe this bug should be reopened for reasons I have outlined in Bug
> #287463.
> 
> I may not fully understand how this works but would be inclined to disagree
> with Comment #4 because the acpi-cpufreq module cannot load due to errors with
> the acpi/processor_perflib functions due to insufficient support in the ACPI
> BIOS for cpu frequency scaling.  On a CPU without Intel "Enhanced" SpeedStep(R)
> (and the associated ACPI BIOS functionality) the correct way to control power
> consumption seems to be to use the ondemand or conservative governors.
> 

Hey Robert,

really, my knowledge in this area is not enough to constructively (dis)agree with your arguments and so I would recommend you to pass this upstream.

Do post the URL of the upstream bug report here so that we can follow this :)

Thank you!

PS1: I sent (some months ago) a mail to Dave Jones (the guy behind codemonkey.org) telling him that there are well put complaints regarding the p4-clockmod change and pointed him to bug #283646 and bug #287463, but I didn't get a reply.

PS2: I'll leave this as INVALID, 'till I see an answer from upstream if that's okay with you.