[Openchrome-users] VX900 issues on VIA VE900 mainboard

[Mike Williams]

On 12/31/12 18:57, James Simmons wrote:
>
>> Hi,
>>
>> I have been setting up OpenBSD 5.2 on a VE900 mainboard and having fun getting
>> the openchrome driver to work with the VX900 chipset.
>
> Wow you are first BSD user that I know of :-)

There are a few of us, mostly using older VIA boards though.

>> The problem is break up on the display.  Hard to describe - it's like
>> horizontal sequences of pixels are not initially being drawn.  Moving the
>> mouse slowly results in the missing pixels being painted, but never all of
>> them.  A bit more experimentation shows that the first two starts of the X
>> server after a boot are perfectly ok but then any subsequent start of the X
>> server results in corruption from then on with no obvious way to stop it.
>>
>> Interestingly, it doesn't affect the X server root window, just the windows on
>> top of it.  The problem  can in the extreme lead to corrupted text in plain
>> console mode after exiting the X server.  I tried using XAA and EXA with no
>> compositing, no difference.
>>
>> Some choice bits from the attached Xorg log file (after display starts
>> suffering from corruption using all default options) and comparison with other
>> log files from experimenting:
>
> Looked at the logs and it appears you are running a old X driver. Can you
> try build from the code at
>
> http://cgit.freedesktop.org/openchrome/xf86-video-openchrome

Easier said than done. OpenBSD run their own packaged X distribution - 
Xenocara, being X plus patches.  This runs down into the drivers as 
well.  You can find the state of the openchrome driver here - 
http://www.openbsd.org/cgi-bin/cvsweb/xenocara/driver/xf86-video-openchrome/ 
  I'll chase the OpenBSD maintainers.

>>
>> 1. Unknown card ids:
>> [   124.916] (EE) CHROME(0): Unknown Card-Ids (7122|1019|3126), Chipset:
>> VX900; please report to openchrome-users at openchrome.org
>>
>> 2. A time out on one occasion - in the attached one:
>> [   190.474] viaWaitVideoCommandFire: Timeout.
>
> Strange that is running.
>
>> 3. Frame buffer start address "wanders" from start to start - I would have
>> thought that would be fixed, perhaps not. Interesting?
>
> Nope. Its a chunk taken out of the VRAM.
>
>> 4. The driver mode and default mode lines can vary.  Sometimes they include a
>> "(scaled from ...)" statement where the frequency various between 0.0 and some
>> varying large number.
>>
>> This is OBSD 5.2 release i386, Xserver 1.12.2, and openchrome 0.2.906 with all
>> default options and the BIOS auto 256MB video RAM (4GiB RAM installed).
>>
>> Happy to try things out.  HTH - TTFN
>
> Thanks for trying out this on BSD. Please note the driver is not
> accelerated for this platform. I also know the BSD is working on a port
> of KMS so it should be possible to support acceleration in the future.
>
>
>


-- 
Mike