[Mesa-dev] [PATCH 8/8] st/mesa: add double input support including lowering (v3)

[Ilia Mirkin]

On Mon, May 4, 2015 at 7:33 PM, Dave Airlie <airlied at gmail.com> wrote:
> On 30 April 2015 at 13:23, Ilia Mirkin <imirkin at alum.mit.edu> wrote:
>> On Wed, Apr 29, 2015 at 9:14 PM, Dave Airlie <airlied at gmail.com> wrote:
>>> From: Dave Airlie <airlied at redhat.com>
>>>
>>> This takes a different approach to previously, we cannot index into the
>>> inputMapping with anything but the mesa attribute index, so we can't use
>>> the just add one to index trick, we need more info to add one to it
>>> after we've mapped the input.
>>
>> Almost certainly a failing on my part, but the above makes little sense to me.
>
> Clarifying this is also the answer to all the questions below as well I think,
>
> Due to the double taking one attribute location, but using two hw slots, I can't
> just use attr + 1, VERT_ATTRIB_GENERIC0 could be a dvec4 attribute,
> and VERT_ATTRIB_GENERIC1 could also be a different dvec4 attribute.
>
> So the mapping from TGSI input where we want
> GENERIC0 to map to input 0,1
> and
> GENERIC1 to map to input 2,3
>
> So input 1, can map to GENERIC0, not GENERIC0+1
>
> We also don't want to process each mesa attribute twice, so when we find
> a mesa attribute that is a double one, we do the double processing at
> that point,
>
> so we have the placeholder, which means this TGSI input is just the second
> part of a mesa attribute you already processed, no need to process it again,
> so we just return a NULL client array for it so we skip it.
>
> Not sure if this helps make more or less sense :-)

Yes, thanks. This all looks good to me... would still like a comment
about double_reg2 (at the definition site in the code) as well as a
newline after get_client_array. With those minor bits fixed,

Reviewed-by: Ilia Mirkin <imirkin at alum.mit.edu>