[Openchrome-devel] openchrome 0.6.0 regressions on VX900 laptop

[Xavier Bachelot]

On 21/03/2017 05:22, Kevin Brace wrote:
> Hi Xavier,
> 
>> Sent: Tuesday, March 21, 2017 at 12:57 AM
>> From: "Xavier Bachelot" <xavier at bachelot.org>
>> To: "Kevin Brace" <kevinbrace at gmx.com>
>> Cc: openchrome-devel at lists.freedesktop.org
>> Subject: Re: [Openchrome-devel] openchrome 0.6.0 regressions on VX900 laptop
>>
>> On 20/03/2017 22:38, Xavier Bachelot wrote:
>>> Hi Kevin,
>>>
>>> . . .
>>>
>>> This last paragraph got me thinking, and I just did an interesting
>>> experiment.
>>> I have booted the laptop with a 1280x1024 screen hooked up to the VGA port.
>>> This results in VGA just complaining about unsupported video mode upon
>>> starting X, but the more interesting part is the FP is displaying a
>>> 1280x1024 picture, which is not distorted.
>>> There is garbage on the right, after the 1280 horizontal boundary, which
>>> is very probably a distorted re-print of the beginning of left of the
>>> screen, and the part below the 768 vertical boundary is indeed not
>>> displayed. The cursor cannot go after the 1280 horizontal boundary, and
>>> gets vertically duplicated when going below the 768 vertical boundary.
>>>
>>> I've tried to use xrandr, but that doesn't help.
>>> X log, regs dump and xrandr output attached.
>>>
>>> Does that ring any bell ?
>>>
>>> I'm now going to compare this regs dump with the ones I took before, and
>>> then try to poke at registers without the VGA port attached.
>>>
>>> Regards,
>>> Xavier
>>>
>> I thought I was running with patch v3 applied, but this is actually
>> vanilla 0.6.0 with only the 1366x768 panel fix added, sorry for the bad
>> info.
>>
>> Anyway, it seems initializing IGA1 for VGA, together with IGA2 for FP,
>> helps getting a proper display on the FP.
>>
>> Regards,
>> Xavier
>>
> 
> I just committed the previously mentioned changes, but are you now saying that the FP is now properly working?

FP does work with vanilla 0.6.0 + the panel fix for 1366x768, if and
only if VGA is connected. VGA on the other hand doesn't work in this setup.
Sorry again for sending wrong information in the first place, and being
inaccurate.

> The code I pushed is now at Version 0.6.100.
> It has the updated mode setting code for VX855 and VX900 chipsets, and it also has the VIA panel ID 10 (0xA) fix as well.
> The code I pushed does not have the software based FP power on / off code.
> If the FP is now working properly, I am happy to hear that, but if necessary, I can also push the code tweaks to use software based FP power on / off code for VX855 and VX900 chipsets as well.

I will retry with git master, but it didn't work properly with version 2
of your patch, even with VGA connected. In this case, FP is showing the
distorted picture, and VGA is reported as using an incorrect resolution,
so there is still something to fix.

That may be related to panel scaling, which you said is only available
on IGA2, and is reported as enabled by via_reg_dump.

If you look at the regs dumps for 0.6 + panel fix with FP + VGA, you'll
see the vertical resolution for IGA1/VGA is the one from the FP (768),
not the one from the external screen (1024). 1280x768 is most probably
not a supported resolution, which would explain.

If you have the time, please send a patch adding the use of software
power sequence, I'll check if it helps or not.

> There weren't too many laptops with VX855 and VX900 chipsets produced (i.e., VIA's small niche of low cost / power market was being eroded by Intel Atom processors around Year 2009 to 2011), but VIA did have several in house mini-ITX mainboards with a LVDS FP connector, so I will like to make sure that FP works.
> If the iDOT Computers L740 laptop's FP is now properly working, then I can work on merging your code fixes you have committed to "fix_warning" branch.
> I personally have never done code merging from one branch to another, so if you want me to do it, let me know the git command line example for doing it.
> Otherwise, I can manually commit them myself with you being the author of all of the commits.
> Personally, I am fine with all of the 6 commits you have made so far in your "fix_warning" branch.
> 
I'll push them myself after adding your reviewed-by.
I will not push the last one which is adding -Wno-unused-functions.
Instead, I'll send a proper patch removing the unused-functions warning
for you to review. I don't want to risk removing functions that may
still have some use later, without you first double-checking that's it's
ok, or even hiding an otherwise useful warning.

Regards,
Xavier