VT8623 [Apollo CLE266]

[Kevin Brace]

Hi CK,

> Date: Tue, 27 Dec 2016 01:11:48 -0900
> From: "CK" <nibbana at gmx.us>
> To: <xorg-devel at lists.x.org>
> Subject: VT8623 [Apollo CLE266]
> Message-ID: <0MdK8t-1c40jC1FGl-00IVy6 at mail.gmx.com>
> 
> The Xorg server has been broken for 2-3 years when
> using drivers for this chipset.  I am not enough of
> an Xorg geek to know all the details, but it broke
> between FreeBSD 9.3 and later releases.  And between
> Linux Puppy Precise-5.7.1 and later releases.
> 
> And all 3 drivers were tried: the original VIA driver,
> the 0.3.3 openchrome driver, and the 0.5.0 openchrome
> driver - and these are good fanless boards that are
> still sold today (686s) and they should be supported,
> especially since they worked fine previously and the
> support was broken in some way - I still use and/or
> support over dozen of them with UNIX type systems
> that rely on Xorg.  These boards are not old
> outdated technolgy by any standard.
> 
> Please fix whatever was broken, so that boards relying
> on this chipset can function with future UNIX-type
> system releases.  Thank you.

I am the sole developer of OpenChrome graphics stack.
For OpenChrome related usage questions, please direct it to openchrome-users mailing list.

https://lists.freedesktop.org/mailman/listinfo/openchrome-users

We also have openchrome-devel mailing list, but it is primarily for discussing OpenChrome development issues.

https://lists.freedesktop.org/mailman/listinfo/openchrome-devel

I doubt airing your complaint against xorg-devel mailing list will do much good since they are not directly involved in the development.
    Regarding the OpenChrome support for CLE266 chipset, yes, I do own 2 boards with them (VIA Technologies EPIA-CL and EPIA-M), and in particular, EPIA-M gets fair amount of testing with the latest OpenChrome DDX code.
Due to VIA Technologies C3 processor's lack of PAE (Page Address Extension) support, I only test it against rather old Linux kernel and X.Org Server.
Specifically, Ubuntu 10.04 LTS since there is a mouse cursor rendering issue (mouse cursor does not get rendered on the screen) with Lubuntu 10.04.
I can attest that the latest development version OpenChrome (Version 0.5.175) should work with Ubuntu 10.04 LTS if you compile the OpenChrome code from the Git repository.
I wrote a nice instructions on how to do it exactly.

https://lists.freedesktop.org/archives/openchrome-users/2016-February/007237.html

Please follow the procedures, and you should have it running.
    The lack of PAE is the major reason why CLE266 chipset and the code does not get tested with newer Linux kernels.
I would imagine that you will have to use Gentoo / Arch Linux for CLE266 chipset at this point since they can compile the Linux kernel without PAE or CMOV instructions.
The early C3 processors (i.e., Samuel, Ezra) did not support CMOV instuctions, but later version (Nehemiah) did have it.
However, both of them do not support PAE.
Most Linux distributions require PAE support nowadays, and Gentoo / Arch Linux is pretty much the several known options left for you at this point.
    I plan to continue the support for CLE266 chipset with the next generation OpenChrome DRM with KMS (Kernel Mode Setting) support.
I got KM400 chipset to work with the next generation OpenChrome DRM in September 2016, but 2D acceleration does not work currently, so it is pretty painful to use it at this point.

https://cgit.freedesktop.org/openchrome/drm-openchrome/commit/?id=affa9b9df2045dd1275705d4897fce1ce2dd2cc6
https://cgit.freedesktop.org/openchrome/drm-openchrome/commit/?id=166edf8d0ace0681deddb5b95be05c00d718b835

That being said, due to the same PAE kernel issue, the code is untested with CLE266 chipset at this point.
There are some major differences with the display controller register addresses between CLE266 and KM400 chipsets, so OpenChrome DRM working with KM400 chipset does not necessarily guarantee that it will work right away with CLE266 chipset.
I still plan to get it working with CLE266 chipset at some point.
I do hope to get the next generation OpenChrome DRM mainlined with Linux kernel at some point, but it might take a year or more for this to happen since the code still contains too many bugs left by the previous developer.

Regards,

Kevin Brace
The OpenChrome Project maintainer / developer