On Friday, June 24, 2016 7:03:35 PM PDT Ilia Mirkin wrote:
> On Fri, Jun 24, 2016 at 6:41 PM, Kenneth Graunke <kenn...@whitecape.org> 
> wrote:
> > The only part of an ir_texture which can be an array is the
> > offsets array in textureGatherOffsets() calls.  We don't want
> > to lower those, because they're required to remain constants.
> >
> > Fixes textureGatherOffsets with Gallium drivers such as llvmpipe,
> > which commit ef78df8d3b0cf540e5f08c8c2f6caa338b64a6c7 regressed.
> 
> I believe you need to also CC this to stable, as that one was tagged
> for stable as well.

Right.  Fixed locally.

> 
> I suppose you could have a situation like
> 
> vec2 coords[] = { ... }
> 
> texture(coords[n])
> 
> in which case this won't cause the coords array to get lowered?
> 
>   -ilia

That would show up as

    (assign () (var_ref coords) (constant (array vec2 n) ...))

    (tex ... (array_ref (var_ref coords) <n>) ...)

So the constant array would still get promoted.

Attachment: signature.asc
Description: This is a digitally signed message part.

_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Reply via email to