<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Mon, May 7, 2018 at 9:37 AM, Kenneth Graunke <span dir="ltr"><<a href="mailto:kenneth@whitecape.org" target="_blank">kenneth@whitecape.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="HOEnZb"><div class="h5">On Monday, May 7, 2018 8:53:34 AM PDT Jason Ekstrand wrote:<br>
> On Mon, May 7, 2018 at 8:02 AM, Alejandro Piñeiro <<a href="mailto:apinheiro@igalia.com">apinheiro@igalia.com</a>><br>
> wrote:<br>
> <br>
> > Hi Jason,<br>
> ><br>
> > as part of the ARB_gl_spirv work, we are doing the linking based on the<br>
> > nir shader that comes from spirv_to_nir. On some cases,<br>
> > nir_lower_wpos_ytransform introduces a new uniform,<br>
> > "gl_FbWposYTransform". So as the uniform is there when we start to do<br>
> > the linking, it is treated as any other uniform, being added to the<br>
> > UniformStorage, accessible through the OpenGL API and so on. Something<br>
> > that I understand that we don't want. On GLSL this is not a problem<br>
> > because the linking is done based on MESA/IR, then the IR is converted<br>
> > to NIR, and the lowering adds the uniform based on the NIR shader,<br>
> > really after the linking.<br>
> ><br>
> > The first solution I thought was just mark the uniform with a special<br>
> > location [1]. It would need to be a negative value, not -1 that is<br>
> > already reserved for not-assigned locations (we still have some cases<br>
> > for those, even if in general ARB_gl_spirv makes explicit location<br>
> > mandatory). So it would be a new negative value, like lets say -5. So<br>
> > then on the linking, if the uniform location has this value, the uniform<br>
> > is skipped.<br>
> ><br>
> > After checking that this solution worked (although I only tested with<br>
> > some tests), I started to search for an alternative, as this solution<br>
> > seemed too hacky.<br>
> <br>
> <br>
> It does seem a bit hacky, yes.<br>
> <br>
> <br>
> > But this lowering needs to be run before nir_lower_io,<br>
> > and nir_lower_io is part of the uniform lowering. So I didn't see really<br>
> > reasonable to do a nir_lower_io without nir_lower_wpos_ytransform, do<br>
> > the linking, and then execute again nir_lower_io/nir_lower_wpos_<wbr>transform.<br>
> ><br>
> <br>
> This is all a bit sticky. One observation is that the newly added uniform<br>
> makes usage of the token system. Are tokens no longer used with SPIR-V? I<br>
> guess they are kind-of an old mechanism to handle the crufty ARB program<br>
> parameters. However, you could have -5 mean "It uses a token"<br>
> <br>
> Another option would be to use a system value instead of a uniform. We can<br>
> then convert it to a uniform in the back-end. We have a few places where<br>
> we do things like this for compute shaders.<br>
> <br>
> I don't have a lot of good ideas or strong opinions, I'm afraid. :-( Maybe<br>
> Ken has some?<br>
> <br>
> --Jason<br>
<br>
</div></div>In GLSL IR, we use var->how_declared == ir_var_hidden to hide internally<br>
created variables from API introspection. Why not just port something<br>
similar over to the newer NIR code?<br></blockquote><div><br></div><div>I wasn't aware of that. If we're going to do linking in NIR, something like that is probably needed.<br><br></div><div>--Jason <br></div></div><br></div></div>