[Mesa-dev] i965 Gen4/5: Garbage rendering when gl_BackColor is written but gl_FrontColor is not

[Paul Berry]

On 9 June 2013 03:30, Chris Forbes <chrisf at ijw.co.nz> wrote:

> I noticed that if you write gl_BackColor but not gl_FrontColor, the
> value that lands in gl_Color in the fragment shader is junk; same for
> the corresponding secondary color builtins. This causes us to fail 12
> of the interpolation piglits.
>
> There's a series from July of last year which was supposed to fix this:
>
> http://lists.freedesktop.org/archives/mesa-dev/2012-July/024117.html
>
> ... but it looks like it went nowhere.
>
> What's the current thinking on this? I'm happy to do the work if
> there's some consensus.
>

BTW did you notice that Olivier also sent out a second version of the
series?  It's here:

http://lists.freedesktop.org/archives/mesa-dev/2012-July/024333.html

I'd love to see this get picked up again.  Reading through the comments on
version 2 of the series, it looks like a decision got made on how to handle
all the outstanding issues except for one: how do we handle vertex shaders
that write to gl_BackColor but not gl_FrontColor?  (See the discussion
here: http://lists.freedesktop.org/archives/mesa-dev/2012-July/024153.html).
Unfortunately I don't remember the issues involved well enough to have a
coherent opinion.  I would appreciate hearing what you think.

Once that issue gets resolved, it looks like it's a fairly simple matter of
taking Olivier's second series, making the fixes suggested by Eric in his
review comments, and turning it into a v3 series.  I'd say go for it.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20130609/579cba25/attachment.html>