[Intel-gfx] S6010 - brightness adjustment not available

[Thomas Richter]

Hi Ville

>> just testing your alm_fixes11 branch. So far, everything works fine,
>> including suspend2ram, for the first time! Yippiee!
>>
>> However, there is one thing that bothers me, namely that the brightness
>> adjustment is no longer working. Specifically, fujitsu_laptop fails with:
>>
>> Fujitsu laptop FUJ02B1 ACPI brightness driver: probe of FUJ02B1:00
>> failed with error -22
>
> My impression was that those were happening since forever. But I can't
> be sure.

Actually, it did work with the alm-fixes5 patch, based on 3.15.0 rc7+ IIRC.

>> The source of fujitsu_laptop does not seem to have changed, thus I
>> believe it must be somehow related to the modifications you made. Any
>> idea where to look?
>
> Hmm. My impression was that the backlight control is handled by some
> hardware/firmware magic and OS can't touch it. IIRC that's how it seemed
> to work on my 855 machine which is a slighly newer Fujitsu-Siemens
> Lifebook model. At least on that machine backlight buttons worked
> a while ago.

Oh, sorry, I should have clarified. Yes, the *buttons* work, those are 
handled by the firmware (via the Bios service management interrupt, I 
suppose), but usually the fujitsu-laptop module *also* creates a device 
which provides access to the brightness control such that tools like the 
Gnome brightness applet can pick the information up and allows 
user-control of the brightness. The corresponding device does no longer 
exist on 3.16.0. Not yet fully clear whether that's due to the patches 
you made or something else, but I was running out of time yesterday to 
start a bisection, and I'm currently on the road and cannot test 
(greetings from Newark airport, BTW.)

> I can re-test the s6010 backlight next week and see if it works for me.
>
> I guess one idea might be the "init important ns2501 registers" patch.

No, I don't think so. You had that in in alm_fixes5 already, and there 
it worked. So the DVO is innocent, at least this time. As said, I 
haven't had the chance to have a more detailed look, but it seems that 
the fujitsu laptop extras go through some scary ACPI access path to 
control the backlight, and somehow this seems to be broken now, for a 
reason I haven't tried to understand.

> Maybe the values I used weren't entirely correct for this machine and
> we need to use the exact same values as the BIOS. All the other register
> values I used were definitely the same that hte BIOS set up, but I have
> nagging feeling there was some slight difference with these few
> registers.

Probably try something else first if you get the chance. My next 
possibility to work on this is next Saturday when I'm back from San 
Diego (this is where I'm going right now) and then take out your patches 
one by one.

Greetings,
	Thomas