[turbofan] Add operators for float32 support.

src/conversions-inl.h

Index: src/conversions-inl.h
diff --git a/src/conversions-inl.h b/src/conversions-inl.h
index bac15768270f4fe0808c31332e261dac81312dce..3b816019c1e79e54c972a347ba04fcaed3599551 100644
--- a/src/conversions-inl.h
+++ b/src/conversions-inl.h
@@ -67,6 +67,13 @@ inline unsigned int FastD2UI(double x) {
 }
 
 
+inline float DoubleToFloat32(double x) {

[titzer, 2014/09/22 11:24:54]

I think you will need to assign the result to a volatile float local variable,
because of C++.

Ugh.

[Benedikt Meurer, 2014/09/22 11:37:53]

Done.

+  // TODO(yanggou): This static_cast is implementation-defined behaviour in C++,
+  // so we may need to do the conversion manually instead to match the spec.
+  return static_cast<float>(x);
+}
+
+
 inline double DoubleToInteger(double x) {
   if (std::isnan(x)) return 0;
   if (!std::isfinite(x) || x == 0) return x;