AMDGPU: Use source modifiers with f16->f32 conversions

The operand types were defined to fit the fp16_to_fp node, which
has the half as an integer type. v_cvt_f32_f16 does support
source modifiers, so change this to have an FP type and modifiers.

For targets without legal f16, this requires recognizing the
bit operations and trying to produce them.

llvm-svn: 293857

1 files changed, 2 insertions, 0 deletions

diff --git a/llvm/lib/Target/AMDGPU/MCTargetDesc/SIMCCodeEmitter.cpp b/llvm/lib/Target/AMDGPU/MCTargetDesc/SIMCCodeEmitter.cpp
index 0c5bb0648a1..bdaa5aef83d 100644
--- a/llvm/lib/Target/AMDGPU/MCTargetDesc/SIMCCodeEmitter.cpp
+++ b/llvm/lib/Target/AMDGPU/MCTargetDesc/SIMCCodeEmitter.cpp
@@ -226,6 +226,8 @@ uint32_t SIMCCodeEmitter::getLitEncoding(const MCOperand &MO,
   case 8:
     return getLit64Encoding(static_cast<uint64_t>(Imm), STI);
   case 2:
+    // FIXME Is this correct? What do inline immediates do on SI for f16 src
+    // which does not have f16 support?
     return getLit16Encoding(static_cast<uint16_t>(Imm), STI);
   default:
     llvm_unreachable("invalid operand size");