[CodeGen] Don't assume fp_to_fp16 produces i16 when legalizing it.
Since r230276, we support an improved legalization for f64->f16,
which goes through a temporary f32, improving codegen when
f32->f16 is legal but not f64->f16. This requires unsafe-fp-math.
However, that legalization assumed that the second step, producing
a pseudo-softened f16, had type i16. That's not true on targets
with illegal i16, such as ARM.
Use the initial f64->f16 result type instead.
llvm-svn: 257794
diff --git a/llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp b/llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
index 5d572c4..10d06a7 100644
--- a/llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
@@ -3416,7 +3416,7 @@
SDValue FloatVal = DAG.getNode(ISD::FP_ROUND, dl, MVT::f32, Op,
DAG.getIntPtrConstant(0, dl));
Results.push_back(
- DAG.getNode(ISD::FP_TO_FP16, dl, MVT::i16, FloatVal));
+ DAG.getNode(ISD::FP_TO_FP16, dl, Node->getValueType(0), FloatVal));
}
}
break;