[builtins] Make Math.ceil, Math.trunc and Math.round optimizable.

src/compiler/raw-machine-assembler.h

Index: src/compiler/raw-machine-assembler.h
diff --git a/src/compiler/raw-machine-assembler.h b/src/compiler/raw-machine-assembler.h
index 7a1621920a493e1930c1198a2a0a446fcc9eeba4..ce6826da978e05d66b722770e3b8ad202c6fe36c 100644
--- a/src/compiler/raw-machine-assembler.h
+++ b/src/compiler/raw-machine-assembler.h
@@ -391,6 +391,7 @@ class RawMachineAssembler {
     return AddNode(machine()->Float32Min().op(), a, b);
   }
   Node* Float32Abs(Node* a) { return AddNode(machine()->Float32Abs(), a); }
+  Node* Float32Neg(Node* a) { return Float32Sub(Float32Constant(-0.0f), a); }
   Node* Float32Sqrt(Node* a) { return AddNode(machine()->Float32Sqrt(), a); }
   Node* Float32Equal(Node* a, Node* b) {
     return AddNode(machine()->Float32Equal(), a, b);
@@ -431,6 +432,7 @@ class RawMachineAssembler {
     return AddNode(machine()->Float64Min().op(), a, b);
   }
   Node* Float64Abs(Node* a) { return AddNode(machine()->Float64Abs(), a); }
+  Node* Float64Neg(Node* a) { return Float64Sub(Float64Constant(-0.0), a); }
   Node* Float64Sqrt(Node* a) { return AddNode(machine()->Float64Sqrt(), a); }
   Node* Float64Equal(Node* a, Node* b) {
     return AddNode(machine()->Float64Equal(), a, b);