[Mesa-dev] [PATCH 0/6] Fix Various Compilation Issues With Bindless

[Marek Olšák]

On Sat, Jun 30, 2018 at 10:56 PM, Karol Herbst <kherbst at redhat.com> wrote:
> On Mon, Jun 11, 2018 at 5:10 PM, Marek Olšák <maraeo at gmail.com> wrote:
>> The series is OK with me, even though radeonsi can't support the new
>> opcodes.
>>
>
> How would you handle the case when a local variable might get a
> bindless or a non bindless sampler value assigned? Meaning, how can a
> compliant implementation of bindless_texture then look like on
> radeonsi?

radeonsi is designed such that nonbindless and bindless arrays of
binding slots are separate. You either get a nonbindless index within
nonbindless per-shader slots, or you get a bindless handle that is an
index into bindless slots. We could merge those arrays of binding
slots into one, though at the cost of added CPU overhead. So we are
not going to do that. This is one of the weird things that get added
into the spec even though they are unnecessary.

Marek