[turbofan] Also optimize unsigned division by constant.

src/compiler/arm/instruction-selector-arm.cc

Index: src/compiler/arm/instruction-selector-arm.cc
diff --git a/src/compiler/arm/instruction-selector-arm.cc b/src/compiler/arm/instruction-selector-arm.cc
index 3f3aff4541b0380d06ac2b76a09b4044a106a26d..a071bbccf8422ff371018a68085d46fea377505d 100644
--- a/src/compiler/arm/instruction-selector-arm.cc
+++ b/src/compiler/arm/instruction-selector-arm.cc
@@ -86,6 +86,7 @@ class ArmOperandGenerator : public OperandGenerator {
       case kArmMls:
       case kArmSmmul:
       case kArmSmmla:
+      case kArmUmull:
       case kArmSdiv:
       case kArmUdiv:
       case kArmBfc:
@@ -658,6 +659,15 @@ void InstructionSelector::VisitInt32MulHigh(Node* node) {
 }
 
 
+void InstructionSelector::VisitUint32MulHigh(Node* node) {
+  ArmOperandGenerator g(this);
+  InstructionOperand* outputs[] = {g.TempRegister(), g.DefineAsRegister(node)};
+  InstructionOperand* inputs[] = {g.UseRegister(node->InputAt(0)),
+                                  g.UseRegister(node->InputAt(1))};
+  Emit(kArmUmull, arraysize(outputs), outputs, arraysize(inputs), inputs);
+}
+
+
 static void EmitDiv(InstructionSelector* selector, ArchOpcode div_opcode,
                     ArchOpcode f64i32_opcode, ArchOpcode i32f64_opcode,
                     InstructionOperand* result_operand,