[Pixman] Moving xf86-video-omapfb's YUV format conversion code into pixman

[Matt Turner]

On Wed, May 30, 2012 at 1:22 AM, Kalle Vahlman <kalle.vahlman at gmail.com> wrote:
> 2012/5/30 Matt Turner <mattst88 at gmail.com>
>>
>> (Resending to a different email address for Tuomas, which hopefully works.)
>>
>> Hi Kalle, Tuomas, Ian,
>>
>> I recently began looking at adding some YUV format code to pixman. The
>> X server has a rather awful xf86XVCopyYUV12ToPacked function which
>> would be great to replace with pixman, which already has the
>> infrastructure to support multiple vectorized implementations.
>>
>> Siarhei pointed me to xf86-video-omapfb's image-format-conversions.c
>> which already has some similar code. Would you be interested in
>> porting this code to pixman? It seems better suited in pixman where it
>> could be potentially useful for a number of other software projects.
>
> The code should be license-compatible at least, but I'm unsure how
> compatible in terms of passed parameters, which might impact the
> implementation. I think I checked pixman way back when while looking
> for YUV conversion implementations, but at that time the support
> wasn't really there (I think it only converted to/from RGB formats).
>
> It would be very nice to have proper YUV conversion support in pixman
> and if the x-v-o implementation would work for that, that's cool. And
> vice versa, when pixman has YUV support then we should port x-v-o to
> that.

Yeah, certainly it wouldn't just be a straight copy and paste. There
would be a small amount of infrastructure work.

pixman actually has at least some yuy2 and yv12 support already.

Thanks,
Matt