SuSE 8.2 Throttling on Lid Close (long)
John Andersen <[email protected]>
| Newsgroups | gmane.linux.hardware.sony |
|---|---|
| Organization | NORCOM |
| Message-ID | <[email protected]> |
The latest SuSE 8.2 (along with KDE 3.1.2) has most of the acpi stuff
that affects hardware quite nicely handled (Even Klaptop works - although
its still a far cry from "akpi").
HOWEVER...
SuSE's acpi event handeling has one annoying little problem on
my GRX-570 that I discovered quite by accident.
If you close the lid, or (like I did) casually stick a pen into the
right hand lid latch hole (next to the green power-on LED) the
acpid will send a lid-close event, and SuSE will throttle
the machine WAY DOWN, and set Performance state
to the lowest to conserve battery.
Sounds good so far...
BUT re-open the lid (or extract the stupid pen from the
lid-latch-hole) and it DOES NOT DE-THROTTLE.
How to test if this affects you:
Set up an xterm and type
cat /proc/acpi/processor/CPU0/performance
You should see something like this
state count: 2
active state: P0
states:
*P0: 1600 MHz, 30000 mW, 250 uS
P1: 1200 MHz, 20200 mW, 250 uS
Note that the Asterisk indicates the current state.
Then type cat /proc/acpi/processor/CPU0/throttling
and you should see state T0: with the asterisk.
Now reach for your pen, or close the lid briefly and
re-open.
Again type cat /proc/acpi/processor/CPU0/performance
and perhaps something other than P0 will have the asterisk,
and cat /proc/acpi/processor/CPU0/throttling may
show the highest state of throttling (T7: in my case)
and the machine will be DOG SLOW and it will
stay that way.
Hint: To restore useability, as root type
echo 0 >cat /proc/acpi/processor/CPU0/throttling
echo 0 >cat /proc/acpi/processor/CPU0/performance
Note:
This behavior (throttle/dethrottle) is specified
in /etc/sysconfig/powermanagement (text file).
This file is managed by Yast2 / System / Editor for
Sysconfig Files under the heading of
System/Powermanagement/ACPI/EventHandeling
I wondered if the VAIO did not send the Lid-Open event
OR if that the dethrottle command does not actually work.
So, using the debug setting (same file or Yast location)
I ran some tests with debug set on.
Tailing /var/log/messages indicates that the Lid Open event
is triggered when the lid is closed, and the Lid Close event is
triggered when the lid is opened. EXACTLY BACKWARDS!!!
I don't know if the VAIO actually sends these events
backward or if something in the SuSE acpid mis-reports
what is sent.
In any event, the logs reveal that both the throttle and
dethrottle commands do work, but they are triggered in
the wrong places. Simply reversing them in the
/etc/sysconfig/powermanagement file is a sutible
workaround untill the real fix can be found.
Either that, or I have to learn to keep my pen
away from the computer.
--
John Andersen - NORCOM
http://www.norcomsoftware.com/