Revert "Introduce MoveDouble to the X64 MacroAssembler" (r17383)

The direction is to use "movl, movq, movp" instead of introducing new
Macro assembler instructions.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/101083002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@19210 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

src/x64/codegen-x64.cc

@@ -422,7 +422,7 @@ void ElementsTransitionGenerator::GenerateDoubleToObject(
   // Non-hole double, copy value into a heap number.
   __ AllocateHeapNumber(rax, r15, &gc_required);
   // rax: new heap number
-  __ MoveDouble(FieldOperand(rax, HeapNumber::kValueOffset), r14);
+  __ movq(FieldOperand(rax, HeapNumber::kValueOffset), r14);
   __ movp(FieldOperand(r11,
                        r9,
                        times_pointer_size,

src/x64/lithium-codegen-x64.cc

@@ -3417,10 +3417,10 @@ void LCodeGen::DoDeferredMathAbsTaggedHeapNumber(LMathAbs* instr) {
   __ LoadFromSafepointRegisterSlot(input_reg, input_reg);
 
   __ bind(&allocated);
-  __ MoveDouble(tmp2, FieldOperand(input_reg, HeapNumber::kValueOffset));
+  __ movq(tmp2, FieldOperand(input_reg, HeapNumber::kValueOffset));
   __ shl(tmp2, Immediate(1));
   __ shr(tmp2, Immediate(1));
-  __ MoveDouble(FieldOperand(tmp, HeapNumber::kValueOffset), tmp2);
+  __ movq(FieldOperand(tmp, HeapNumber::kValueOffset), tmp2);
   __ StoreToSafepointRegisterSlot(input_reg, tmp);
 
   __ bind(&done);

src/x64/macro-assembler-x64.h

@@ -841,9 +841,6 @@ class MacroAssembler: public Assembler {
   void Pop(Register dst) { pop(dst); }
   void PushReturnAddressFrom(Register src) { push(src); }
   void PopReturnAddressTo(Register dst) { pop(dst); }
-  void MoveDouble(Register dst, const Operand& src) { movq(dst, src); }
-  void MoveDouble(const Operand& dst, Register src) { movq(dst, src); }
-
   void Move(Register dst, ExternalReference ext) {
     movp(dst, reinterpret_cast<Address>(ext.address()),
          RelocInfo::EXTERNAL_REFERENCE);