Add Neon VCVT instructions for f32 <-> f16 conversions.
Clang is now providing intrinsics for these and so we need to support them
in the backend.  Radar 8068427.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@121902 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/Target/ARM/ARM.td b/lib/Target/ARM/ARM.td
index 5b43716..bf4315f 100644
--- a/lib/Target/ARM/ARM.td
+++ b/lib/Target/ARM/ARM.td
@@ -103,7 +103,8 @@
                                     FeatureHasSlowFPVMLx, FeatureT2XtPk]>;
 def ProcA9      : SubtargetFeature<"a9", "ARMProcFamily", "CortexA9",
                                    "Cortex-A9 ARM processors",
-                                   [FeatureHasSlowFPVMLx, FeatureT2XtPk]>;
+                                   [FeatureHasSlowFPVMLx, FeatureT2XtPk,
+                                    FeatureFP16]>;
 
 class ProcNoItin<string Name, list<SubtargetFeature> Features>
  : Processor<Name, GenericItineraries, Features>;