cpufreq regression, unable to set lower frequency

!MAILaRCHIVE_VOTE_RePLACE
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
To: <linux-kernel@...>
Date: Thursday, August 21, 2008 - 1:51 pm

Hi LKML,

I've upgraded my kernel few days earlier and suddenly my cpufreqd
stopperd working. My configuration is this:
AC adapter connected:
  set scaling_max_freq and scaling_min_freq to maximal available
frequency
AC adapter disconnected:
  set scaling_max_freq to maximal available, scaling_min_freq to minimal
available.

When I connect AC adapter, everything works fine but when I disconnect
my adapter, everything remains on maximal frequency and in logs I can
see:

cpufreqd: cpufreqd_set_profile: Couldn't set profile "Powersave Low" set
for cpu0

After some tests I've traced problem to=20

http://git.kernel.org/?p=3Dlinux/kernel/git/stable/linux-2.6.25.y.git;a=3Dc=
ommitdiff;h=3D53391fa20cab6df6b476a5a0ad6be653c9de0c46

What do you think? Is this condition correct?

# cd /sys/devices/system/cpu/cpu0/cpufreq/

when AC adapter is connected, something like this should happen:
# cat scaling_available_frequencies=20
2167000 2000000 1833000 1667000 1500000 1333000 1000000=20
# echo 2167000 > scaling_max_freq=20
# echo 2167000 > scaling_min_freq=20

when disconnected, something like this should:
# echo 1000000 > scaling_max_freq=20
-su: echo: write error: Invalid argument
# echo 1000000 > scaling_min_freq

after min is set, max can be lowered to. Before this patch, lowering max
below min lead to lowering of min to the same value.

I think that this whole condition must not be here at all. After
removing this condition the kernel behaves like this:


# cat scaling_min_freq scaling_max_freq=20
2167000
2167000
# echo 1000000 > scaling_max_freq
# cat scaling_min_freq scaling_max_freq
1000000
1000000
# echo 2167000 > scaling_min_freq
# cat scaling_min_freq scaling_max_freq
1000000
1000000
# echo 2167000 > scaling_max_freq
# echo 2167000 > scaling_min_freq
# cat scaling_min_freq scaling_max_freq
2167000
2167000


Which I think is not perfect but is better then current behavior. What
do you think?

--=20
Martin Filip <bugtraq@smoula.net>
jabber: nexus@smoula.net
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]

Messages in current thread:
cpufreq regression, unable to set lower frequency, Martin Filip, (Thu Aug 21, 1:51 pm)
Re: cpufreq regression, unable to set lower frequency, Mattia Dongili, (Thu Aug 21, 6:55 pm)
Re: cpufreq regression, unable to set lower frequency, Andrew Morton, (Fri Aug 22, 4:12 am)