Now that we're using the right variable-mode, this is the same, and
prepares us a bit more for the future where all variables will be in the
same list.

The variable here gets added to the nir_shader::system_values list
rather than the nir_shader::inputs list, so it should have the
nir_var_system_value variable mode instead of nir_var_shader_in.

Not doing this will have horrible consequences when rebasing on
upstream, because all of these variables will be stored in the same list
there. This means the variable-mode is the only detail left to let us
know what this is.

When we expose this without exposing SSBOs, gallium gets confused and
tries to use SSBOs for atomic operations, even if we explicitly suppory
zero buffers.

Connor Abbott authored 4 years ago and Erik Faye-Lund committed 4 years ago

The st now does this for us.

Part-of: <mesa/mesa!4732>

Connor Abbott authored 4 years ago and Erik Faye-Lund committed 4 years ago

This was trying to emulate the effect of mapping GL -> TGSI -> NIR,
but failed to handle VARYING_SLOT_PNTC which led to a kludgy workaround
in freedreno.

Part-of: <mesa/mesa!4732>

If we don't use the pre-draw flag here, we won't end up transitioning
these until some time after the draw, which isn't what we want.

These flags are never set, so let's simplify the logic a bit here.

This gets rid of some repeated code.

Jesse Natalie authored 4 years ago and Erik Faye-Lund committed 4 years ago

This is necessary to avoid having derefs with uniform mode, which
lower_explicit_io would turn into address computations.

Faith Ekstrand authored 4 years ago and Erik Faye-Lund committed 4 years ago

The OpenCL image_width/height/depth functions have variants which can
take an LOD parameter.  More importantly, LLVM-SPIRV-Translator always
generates OpImageQuerySizeLod even if the LOD is guaranteed to be zero.
Given that over half the hardware out there has an LOD field for image
size queries (based on a rudimentary scan through their NIR -> whatever
code), we may as well just add the source to the NIR intrinsic.  If this
is ever a problem for anyone, the lowering is pretty trivial.

I've also added asserts to everyone's drivers that should alert them if
they ever see an LOD other than zero.  This will never happen with GL or
Vulkan so there's no need for panic.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

Jesse Natalie authored 4 years ago and Erik Faye-Lund committed 4 years ago

This reverts commit d9d4a1ae.

Jesse Natalie authored 4 years ago and Erik Faye-Lund committed 4 years ago

Delete unintentionally left debugging code