On Fri, Nov 11, 2016 at 1:39 PM, Ilia Mirkin <imir...@alum.mit.edu> wrote:
> On Fri, Nov 11, 2016 at 10:23 AM, Jason Ekstrand <ja...@jlekstrand.net>
> wrote:
> > On Nov 11, 2016 6:16 AM, "Ilia Mirkin" <imir...@alum.mit.edu> wrote:
> >>
> >> I haven't checked spirv for the specific naming, but note that prim ID
> can
> >> be an output in gs.
> >
> > You're right. I'm not 100% site how we handle that today. We probably
> just
> > pass it through as a varying. It looks like the code here needs to be a
> bit
> > more complex.
>
> Based on some quick spec and vtn reading, you can just do
>
> if (*mode == nir_var_shader_out) {
> *location = VARYING_SLOT_PRIMITIVE_ID;
> } else {
> set_mode_system_value(mode);
> *location = SYSTEM_VALUE_PRIMITIVE_ID;
> }
>
That's exactly the code I was intending to write!
> Note that this will cause gl_PrimitiveID input to become a sysval in
> *every* stage (i.e. TES, TCS, GS gl_PrimitiveIDIn, FS).
>
Yup.
> Cheers,
>
> -ilia
>
> >
> >>
> >> On Nov 11, 2016 1:34 AM, "Jason Ekstrand" <ja...@jlekstrand.net> wrote:
> >>>
> >>> Before, we were treating it as an output which is just completely
> bogus.
> >>> We could go for an input but a system value makes more sense.
> >>>
> >>> Cc: "13.0" <mesa-sta...@lists.freedesktop.org>
> >>> ---
> >>> src/compiler/spirv/vtn_variables.c | 4 ++--
> >>> 1 file changed, 2 insertions(+), 2 deletions(-)
> >>>
> >>> diff --git a/src/compiler/spirv/vtn_variables.c
> >>> b/src/compiler/spirv/vtn_variables.c
> >>> index bbcca14..a4d5367 100644
> >>> --- a/src/compiler/spirv/vtn_variables.c
> >>> +++ b/src/compiler/spirv/vtn_variables.c
> >>> @@ -806,8 +806,8 @@ vtn_get_builtin_location(struct vtn_builder *b,
> >>> set_mode_system_value(mode);
> >>> break;
> >>> case SpvBuiltInPrimitiveId:
> >>> - *location = VARYING_SLOT_PRIMITIVE_ID;
> >>> - *mode = nir_var_shader_out;
> >>> + *location = SYSTEM_VALUE_PRIMITIVE_ID;
> >>> + set_mode_system_value(mode);
> >>> break;
> >>> case SpvBuiltInInvocationId:
> >>> *location = SYSTEM_VALUE_INVOCATION_ID;
> >>> --
> >>> 2.5.0.400.gff86faf
> >>>
> >>> _______________________________________________
> >>> mesa-dev mailing list
> >>> mesa-dev@lists.freedesktop.org
> >>> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
>
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
