Comment 48 for bug 333386

gidantribal wrote:
> I don't know what you mean with 'elusive' problem. To replicate it you
> just need an Acer Aspire 6920G or 6930G (my friend has the same problem)
> and a FRESH installation of Jaunty 9.04 x64.

This is the first "problem". I do not have one. I only got an Acer Aspire One
for which the maintainer of acer-wmi claims has a completely non-working wmi
interface. So it is not much help in problem determination.
Sure, you posted anything that was asked. Still it is sometimes hard to get the
picture together in the head.

> As said before, in kernel 2.6.27-9 everything worked fine,
> applet+brightness commands. At least I remember I had to switch to
> 2.6.27-4 in order to change brightness. Now I don't have this kernel
> image anymore and have to use Vista. Anyway if you really need more
> feedback from the previous kernel images I'll install them and see
> (problem is that i have to disable Nvidia drivers and so on)...

I can understand this is annoying and also painful. Unfortunately, with
interruptions and distractions, the picture only builds slowly and then later
the are questions coming up. Like: was the sysfs interface working whith
2.6.27-9 or was something else in control. Wouldn't a live-cd run work? This
should be 2.6.27-7. That would safe you the hassle of installing/reconfiguring.

> How is it possible ACPI-WMI are installed (directories do exist) and at
> the same time there's no trace of effective brightness change? Maybe
> diff between ACPI in 2.6.27-9 and in 2.6.28-11 should be done?

Depending on which of the interfaces are used (another thing that just appeared
which might be good to know "cat /sys/devices/platform/acer-wmi/interface") the
driver seems to write to the embedded controller or uses the acpi wmi
interface. wmi.c did not change much ec.c did while it is more likely that the
wmi interface is used. acpi in general changed a lot. As the acer-wmi driver
did not change much (only on places that relate to wireless and 3g), that would
require something in the interface itself to be changed. But that does not seem
to have happened. Also this should then have more impact on other drivers as well.

The driver also seems to have a debugfs interface, though the AAO does not
anything there (maybe because it is broken). After "mounr -tdebugfs none
/sys/kernel/debug" you would find an acer-wmi/devices there.

--

When all other means of communication fail, try words!