[Openchrome-devel] openchrome 0.6.0 regressions on VX900 laptop

[Kevin Brace]

Hi Xavier,

I prefer if you can file this as a bug over at https://bugs.freedesktop.org, but if we are able to fix this soon, you may not have to.
The register dump was helpful, and I think the cause of the FP not working is due to LVDS1 (The VIA datasheet calls it LVDS Channel 0 sometimes, but I will call it LVDS Channel 1. Accordingly, LVDS2 will be called LVDS Channel 2.) sending IGA1 display stream, not IGA2.
When the computer boots, it probably sets LVDS1 display stream source to IGA1, and the register dump indicates this.
If you manually switch the LVDS1 to IGA2, that should solve the bug.

$ cd tools
$ sudo ./via_regs_dump -r 3d5.99

The screen according to your Version 0.6 + your panel index fix patch returned 0x01.
Now, write 0x11 to 3X5.99

$ sudo ./via_regs_dump -w 3d5.99 11

Bit 4 of 3X5.99 determines the display source.
0 for IGA1 and 1 for IGA2.
I hope this experiment will get the screen back.
It appears that the LCDCK PLL is set correctly, so hopefully this fix will work.
    The mode setting code is still imperfect, and this is due to myself having only up to VX700 chipset laptop.
I do not own laptops with VX800, VX855, and VX900 chipsets.
I think there are very few VX855 and VX900 laptops in existence.
I have only found one VX900 laptop called N17WA sold by a Chinese vendor for the Chinese market.

https://upload.wikimedia.org/wikipedia/commons/thumb/d/d4/VIA_QuadCore_%40_Computex_2011-N17WA_Netbook_%285810895993%29.jpg/1280px-VIA_QuadCore_%40_Computex_2011-N17WA_Netbook_%285810895993%29.jpg

Assuming the test fixes the problem, I will rewrite the mode setting code to permanently fix the flaw.

Regards,

Kevin Brace
The OpenChrome Project maintainer / developer


> Sent: Tuesday, March 14, 2017 at 12:52 AM
> From: "Xavier Bachelot" <xavier at bachelot.org>
> To: "Kevin Brace" <kevinbrace at gmx.com>, openchrome-devel at lists.freedesktop.org
> Subject: Re: [Openchrome-devel] openchrome 0.6.0 regressions on VX900 laptop
>
> Hi Kevin,
> 
> Agreed, better a real fix than a revert, but that's what I came up with
> after bisecting the issue. Oh, and no need for VGA, ssh is much easier.
> 
> So here are 2 registers dumps, one with plain 0.5.126, another with
> 0.6.0 plus only the panel resolution table fix in order to reduce the
> differences.
> 
> Regards,
> Xavier
> 
> 
> 
>