[Debian-eeepc-devel] eeepc-acpi-scripts bits for testing

Darren Salt linux at youmustbejoking.demon.co.uk
Wed May 27 19:10:32 UTC 2009

I demand that Santi Béjar may or may not have written...

> 2009/5/27 Santi Béjar <santi at agolina.net>:
>> 2009/5/25 Darren Salt <linux at youmustbejoking.demon.co.uk>:
>>> We have two branches in eeepc-acpi-scripts.git which need a few quick
>>> tests.
>> [...]
>>> "shengine": only for those who are running Linux 2.6.30-rc7 (or have
>>> applied the ACPI patch which is merged into -rc7). You'll need to edit
>>> /etc/default/eeepc-acpi-scripts to assign "handle_shengine" to a key; I
>>> suggest the "task manager" key (Fn-F6 or Fn-F9).

> So now I've tested it, and it works somehow. The problem is that I cannot
> set cpufv=2, it is stuck at cpufv=1 so it doesn't cycle between the three
> (or two) values. (I can change it from 0 to 1 with the key).

Hmm. That sounds like the BIOS and/or hardware only supports two states; what
hardware? (I'm guessing one of the Celeron-based ones, since all three states
are useable on my 901.)

(Incidentally, your build failure doesn't happen here.)

>> But I have one question, can this be done automatic like the ondemand
>> cpufreq?

> Now that I've tested I know why, it's really slow.

It can, but I'd tie it to the current power source: if mains power is
removed, slow down. (I have code which does this, but it requires editing of
a conffile to alter the settings.)

> Another question, is there a way to know how much power the eeepc is using,
> the ACPI one returned by powertop is bogus. I've read that it is because of
> wrong units, but is there another way? Thanks.

I'm only aware of the ACPI method, and "wrong units" doesn't make it fake,
just wrong.

| Darren Salt      | linux at youmustbejoking | nr. Ashington, | Doon
| Debian GNU/Linux | or ds    ,demon,co,uk    | Northumberland | Army
| + Output less CO2 => avoid boiling weather.     TIME IS RUNNING OUT *FAST*.

How you look depends on where you go.

More information about the Debian-eeepc-devel mailing list