[Intel-gfx] Where is SDVO mode setting code in kernel module sources?

[4ernov]

2011/10/19 Adam Jackson <ajax at redhat.com>:
> On Wed, 2011-10-19 at 18:43 +0400, 4ernov wrote:
>
>> And is it guessed somewhere in case of VBT missing?
>
> Well, not yet, no.  Like I said, there's no code at all yet for tweaking
> SDVO LVDS.

OK, it's clear for me now. But is there any right place where I can
hardcode this settings to test my idea (the idea is that low
brightness is because of 18 bit LVDS is on output but the monitor is
working in 24 bit LVDS mode)?

>
>> In my case there's
>> no OpRegion support (i945GME chipset don't support it as mainboard
>> manufacturer informed)
>
> This is, at best, stretching the truth.

Yes, I also in doubt and even found some mentions that it actually
supports, but the manufacturer insists that BIOS of 945 knows nothing
about OpRegion..

>
>> thus there're only "*ERROR* VBT signature
>> missing" and "failed to find VBIOS tables" messages printed and I
>> think the module doesn't use VBT after this.
>
> I suppose in this case you'd have to hope that the SDVO init code at
> POST time set up the device correctly, so that if you called
> SDVO_CMD_GET_LVDS_PANEL_INFORMATION it'd tell you the truth.  But that
> assumes that the SDVO init code does that, and it might not, I don't
> actually know; Intel doesn't release the docs for that publicly (or even
> under NDA unless you're an OEM).

Yes, it seems to init SDVO somehow, but perhaps in not right manner as
there's a problem with brightness I mentioned. Thank you for
SDVO_CMD_GET_LVDS_PANEL_INFORMATION reference, will try it.

> That or we'd need to add a way of passing in a VBT from the kernel
> command line, like ACPI DSDT override does.

I think it could be nice workaround for such cases.

>
> - ajax
>