Hack because this only works for colour preloads (no depth/stencil preloads, nor
general blits) and because it'll break Bifrost pretty hard.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

With a Valhall driver stack written, all that's left is probing on the new
hardware!

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Layout is changed since Bifrost, the information stored is basically the
same.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Required for proper texturing.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Valhall-style IDVS uses a distinct job type which has to be handled separately.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

It's implicit.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

I don't know the encoding and right now don't care. Fixes
dEQP-GLES2.functional.shaders.indexing.varying_array.vec4_static_write_dynamic_read on Valhall.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Doesn't work yet. Kick the can down the road; I'd like to get textures
and FBOs working at all before worrying about compressing them.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Replace LOAD.ubo with LD_BUFFER since the .ubo segment doesn't exist on Valhall.
We could do this with a lowering pass instead but this is probably fine.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

It is unclear if FP32 point sizes are supported on Valhall -- I can't get the
DDK to use them at any rate. Always lower them to FP16 and store them as FP16
for hardware use.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

To match driver behaviour.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Is this right on bifrost..?

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

We need a slightly different idiom on Valhall, so let's first split the
helpers for encapsulation.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

We use bi_dontcare() to specify any encoding where we don't care about
the value, with a preference for power-efficient encodings. On Bifrost,
a (possibly nonexistant) FAU read is the best encoding. On Valhall, that
encoding doesn't exist so just use a zero. That should be good enough in
practice.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Needed for good code gen of varying shaders with multiple varyings.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Needed to handle LEA_VARY which writes a register pair.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Varying stores was changed in Valhall. Rather than using attribute descriptors
like on Bifrost and Midgard, on Valhall we store to memory directly with
hardware-allocated buffers. This requires a new implementation of store_output,
with special provisions for writing gl_PointSize from a position shader.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

There are 4 distinct cases of fsqrt:

1. FP16 on Bifrost

Here we may use lower to `x * rsqrt(x)` with a .left modifier on the
FMA_RSCALE.v2f16 used to carry out the multiplication, ensuring correct
handling of NaN and Inf.

2. FP32 on G71

Missing FRSQ.f32 instruction, do something simple since we don't even
probe the driver on G71...

3. FP32 on G72 and newer

We can do the same lowering as FP16 in theory. However, this may have
precision issues. The DDK uses extra FREXPM/FREXPE instructions in a
.sqrt mode for a range reduction. It's unknown if this is necessary for
OpenGL (ES), Vulkan, OpenCL, or some combination thereof.

4. FP16 on Valhall

We want to use the same strategy as on Bifrost, but Valhall removed the
FMA_RSCALE.v2f16 instruction. Instead we use an ordinary FMA.v2f16 for
the multiply and check the special case explicitly with CSEL.v2f32 ...
I'm not sure if this is right for infinity but the DDK does it so ¯\_(ツ)_/¯

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Hack because this bloats bi_instr and fails to model multiple staging
registers.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Fewer arguments compared to Bifrost.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>

Still need to r/e the op.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>