nir: Fix nir_lower_alu_to_scalar's instr filtering.
authorEric Anholt <eric@anholt.net>
Tue, 16 Jul 2019 19:38:10 +0000 (12:38 -0700)
committerEric Anholt <eric@anholt.net>
Wed, 17 Jul 2019 17:30:43 +0000 (10:30 -0700)
commit28a808a11b0dac153115776b4cf1c32c73a891a4
treef03e21be79c0f2d0ce8795e7f096d004d43581bc
parenta301250ece2e3d656008941308db83145b2579ca
nir: Fix nir_lower_alu_to_scalar's instr filtering.

It was checking if the dest or src[0] SSA values were vectors, rather than
whether the ALU op was using the source as a vector resulting in a
nir_fdot4 making it through to vc4 and v3d:

vec1 32 ssa_6 = fdot4 ssa_4.xxxx, ssa_5

Fixes: c1cffa4249ca ("nir/alu_to_scalar: Use the new NIR lowering framework")
v2: Use Jason's recommendation to look at input_sizes.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
src/compiler/nir/nir_lower_alu_to_scalar.c