Introduce leap, movzxbp, movzxwp, repmovsp and xchgp for x64 port

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/211413008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20273 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

src/x64/assembler-x64.cc

@@ -1224,17 +1224,9 @@ void Assembler::jmp(const Operand& src) {
 }
 
 
-void Assembler::lea(Register dst, const Operand& src) {
+void Assembler::emit_lea(Register dst, const Operand& src, int size) {
   EnsureSpace ensure_space(this);
-  emit_rex_64(dst, src);
-  emit(0x8D);
-  emit_operand(dst, src);
-}
-
-
-void Assembler::leal(Register dst, const Operand& src) {
-  EnsureSpace ensure_space(this);
-  emit_optional_rex_32(dst, src);
+  emit_rex(dst, src, size);
   emit(0x8D);
   emit_operand(dst, src);
 }
@@ -1473,7 +1465,7 @@ void Assembler::movsxlq(Register dst, const Operand& src) {
 }
 
 
-void Assembler::movzxbq(Register dst, const Operand& src) {
+void Assembler::emit_movzxb(Register dst, const Operand& src, int size) {
   EnsureSpace ensure_space(this);
   // 32 bit operations zero the top 32 bits of 64 bit registers.  Therefore
   // there is no need to make this a 64 bit operation.
@@ -1484,26 +1476,10 @@ void Assembler::movzxbq(Register dst, const Operand& src) {
 }
 
 
-void Assembler::movzxbl(Register dst, const Operand& src) {
-  EnsureSpace ensure_space(this);
-  emit_optional_rex_32(dst, src);
-  emit(0x0F);
-  emit(0xB6);
-  emit_operand(dst, src);
-}
-
-
-void Assembler::movzxwq(Register dst, const Operand& src) {
-  EnsureSpace ensure_space(this);
-  emit_optional_rex_32(dst, src);
-  emit(0x0F);
-  emit(0xB7);
-  emit_operand(dst, src);
-}
-
-
-void Assembler::movzxwl(Register dst, const Operand& src) {
+void Assembler::emit_movzxw(Register dst, const Operand& src, int size) {
   EnsureSpace ensure_space(this);
+  // 32 bit operations zero the top 32 bits of 64 bit registers.  Therefore
+  // there is no need to make this a 64 bit operation.
   emit_optional_rex_32(dst, src);
   emit(0x0F);
   emit(0xB7);
@@ -1511,8 +1487,10 @@ void Assembler::movzxwl(Register dst, const Operand& src) {
 }
 
 
-void Assembler::movzxwl(Register dst, Register src) {
+void Assembler::emit_movzxw(Register dst, Register src, int size) {
   EnsureSpace ensure_space(this);
+  // 32 bit operations zero the top 32 bits of 64 bit registers.  Therefore
+  // there is no need to make this a 64 bit operation.
   emit_optional_rex_32(dst, src);
   emit(0x0F);
   emit(0xB7);
@@ -1535,17 +1513,10 @@ void Assembler::repmovsw() {
 }
 
 
-void Assembler::repmovsl() {
-  EnsureSpace ensure_space(this);
-  emit(0xF3);
-  emit(0xA5);
-}
-
-
-void Assembler::repmovsq() {
+void Assembler::emit_repmovs(int size) {
   EnsureSpace ensure_space(this);
   emit(0xF3);
-  emit_rex_64();
+  emit_rex(size);
   emit(0xA5);
 }
 
@@ -1789,36 +1760,18 @@ void Assembler::shrd(Register dst, Register src) {
 }
 
 
-void Assembler::xchgq(Register dst, Register src) {
-  EnsureSpace ensure_space(this);
-  if (src.is(rax) || dst.is(rax)) {  // Single-byte encoding
-    Register other = src.is(rax) ? dst : src;
-    emit_rex_64(other);
-    emit(0x90 | other.low_bits());
-  } else if (dst.low_bits() == 4) {
-    emit_rex_64(dst, src);
-    emit(0x87);
-    emit_modrm(dst, src);
-  } else {
-    emit_rex_64(src, dst);
-    emit(0x87);
-    emit_modrm(src, dst);
-  }
-}
-
-
-void Assembler::xchgl(Register dst, Register src) {
+void Assembler::emit_xchg(Register dst, Register src, int size) {
   EnsureSpace ensure_space(this);
   if (src.is(rax) || dst.is(rax)) {  // Single-byte encoding
     Register other = src.is(rax) ? dst : src;
-    emit_optional_rex_32(other);
+    emit_rex(other, size);
     emit(0x90 | other.low_bits());
   } else if (dst.low_bits() == 4) {
-    emit_optional_rex_32(dst, src);
+    emit_rex(dst, src, size);
     emit(0x87);
     emit_modrm(dst, src);
   } else {
-    emit_optional_rex_32(src, dst);
+    emit_rex(src, dst, size);
     emit(0x87);
     emit_modrm(src, dst);
   }

src/x64/assembler-x64.h

@@ -516,11 +516,16 @@ class CpuFeatures : public AllStatic {
   V(idiv)                               \
   V(imul)                               \
   V(inc)                                \
+  V(lea)                                \
   V(mov)                                \
+  V(movzxb)                             \
+  V(movzxw)                             \
   V(neg)                                \
+  V(repmovs)                            \
   V(sbb)                                \
   V(sub)                                \
-  V(test)
+  V(test)                               \
+  V(xchg)
 
 
 class Assembler : public AssemblerBase {
@@ -773,18 +778,14 @@ class Assembler : public AssemblerBase {
   void movsxwq(Register dst, const Operand& src);
   void movsxlq(Register dst, Register src);
   void movsxlq(Register dst, const Operand& src);
-  void movzxbq(Register dst, const Operand& src);
-  void movzxbl(Register dst, const Operand& src);
-  void movzxwq(Register dst, const Operand& src);
-  void movzxwl(Register dst, const Operand& src);
-  void movzxwl(Register dst, Register src);
 
   // Repeated moves.
 
   void repmovsb();
   void repmovsw();
-  void repmovsl();
-  void repmovsq();
+  void repmovsp() { emit_repmovs(kPointerSize); }
+  void repmovsl() { emit_repmovs(kInt32Size); }
+  void repmovsq() { emit_repmovs(kInt64Size); }
 
   // Instruction to load from an immediate 64-bit pointer into RAX.
   void load_rax(void* ptr, RelocInfo::Mode rmode);
@@ -796,10 +797,6 @@ class Assembler : public AssemblerBase {
   void cmovl(Condition cc, Register dst, Register src);
   void cmovl(Condition cc, Register dst, const Operand& src);
 
-  // Exchange two registers
-  void xchgq(Register dst, Register src);
-  void xchgl(Register dst, Register src);
-
   void cmpb(Register dst, Immediate src) {
     immediate_arithmetic_op_8(0x7, dst, src);
   }
@@ -886,9 +883,6 @@ class Assembler : public AssemblerBase {
   // Sign-extends eax into edx:eax.
   void cdq();
 
-  void lea(Register dst, const Operand& src);
-  void leal(Register dst, const Operand& src);
-
   // Multiply rax by src, put the result in rdx:rax.
   void mul(Register src);
 
@@ -1483,6 +1477,14 @@ class Assembler : public AssemblerBase {
   // numbers have a high bit set.
   inline void emit_optional_rex_32(const Operand& op);
 
+  void emit_rex(int size) {
+    if (size == kInt64Size) {
+      emit_rex_64();
+    } else {
+      ASSERT(size == kInt32Size);
+    }
+  }
+
   template<class P1>
   void emit_rex(P1 p1, int size) {
     if (size == kInt64Size) {
@@ -1696,15 +1698,23 @@ class Assembler : public AssemblerBase {
   void emit_inc(Register dst, int size);
   void emit_inc(const Operand& dst, int size);
 
+  void emit_lea(Register dst, const Operand& src, int size);
+
   void emit_mov(Register dst, const Operand& src, int size);
   void emit_mov(Register dst, Register src, int size);
   void emit_mov(const Operand& dst, Register src, int size);
   void emit_mov(Register dst, Immediate value, int size);
   void emit_mov(const Operand& dst, Immediate value, int size);
 
+  void emit_movzxb(Register dst, const Operand& src, int size);
+  void emit_movzxw(Register dst, const Operand& src, int size);
+  void emit_movzxw(Register dst, Register src, int size);
+
   void emit_neg(Register dst, int size);
   void emit_neg(const Operand& dst, int size);
 
+  void emit_repmovs(int size);
+
   void emit_sbb(Register dst, Register src, int size) {
     if (size == kInt64Size) {
       arithmetic_op(0x1b, dst, src);
@@ -1764,6 +1774,9 @@ class Assembler : public AssemblerBase {
   void emit_test(const Operand& op, Register reg, int size);
   void emit_test(const Operand& op, Immediate mask, int size);
 
+  // Exchange two registers
+  void emit_xchg(Register dst, Register src, int size);
+
   friend class CodePatcher;
   friend class EnsureSpace;
   friend class RegExpMacroAssemblerX64;

src/x64/builtins-x64.cc

@@ -93,13 +93,13 @@ static void GenerateTailCallToSharedCode(MacroAssembler* masm) {
           FieldOperand(rdi, JSFunction::kSharedFunctionInfoOffset));
   __ movp(kScratchRegister,
           FieldOperand(kScratchRegister, SharedFunctionInfo::kCodeOffset));
-  __ lea(kScratchRegister, FieldOperand(kScratchRegister, Code::kHeaderSize));
+  __ leap(kScratchRegister, FieldOperand(kScratchRegister, Code::kHeaderSize));
   __ jmp(kScratchRegister);
 }
 
 
 static void GenerateTailCallToReturnedCode(MacroAssembler* masm) {
-  __ lea(rax, FieldOperand(rax, Code::kHeaderSize));
+  __ leap(rax, FieldOperand(rax, Code::kHeaderSize));
   __ jmp(rax);
 }
 
@@ -213,7 +213,7 @@ static void Generate_JSConstructStubHelper(MacroAssembler* masm,
       }
 
       // Now allocate the JSObject on the heap.
-      __ movzxbq(rdi, FieldOperand(rax, Map::kInstanceSizeOffset));
+      __ movzxbp(rdi, FieldOperand(rax, Map::kInstanceSizeOffset));
       __ shl(rdi, Immediate(kPointerSizeLog2));
       if (create_memento) {
         __ addp(rdi, Immediate(AllocationMemento::kSize));
@@ -238,12 +238,12 @@ static void Generate_JSConstructStubHelper(MacroAssembler* masm,
       // rax: initial map
       // rbx: JSObject
       // rdi: start of next object (including memento if create_memento)
-      __ lea(rcx, Operand(rbx, JSObject::kHeaderSize));
+      __ leap(rcx, Operand(rbx, JSObject::kHeaderSize));
       __ LoadRoot(rdx, Heap::kUndefinedValueRootIndex);
       if (count_constructions) {
-        __ movzxbq(rsi,
+        __ movzxbp(rsi,
                    FieldOperand(rax, Map::kPreAllocatedPropertyFieldsOffset));
-        __ lea(rsi,
+        __ leap(rsi,
                Operand(rbx, rsi, times_pointer_size, JSObject::kHeaderSize));
         // rsi: offset of first field after pre-allocated fields
         if (FLAG_debug_code) {
@@ -255,7 +255,7 @@ static void Generate_JSConstructStubHelper(MacroAssembler* masm,
         __ LoadRoot(rdx, Heap::kOnePointerFillerMapRootIndex);
         __ InitializeFieldsWithFiller(rcx, rdi, rdx);
       } else if (create_memento) {
-        __ lea(rsi, Operand(rdi, -AllocationMemento::kSize));
+        __ leap(rsi, Operand(rdi, -AllocationMemento::kSize));
         __ InitializeFieldsWithFiller(rcx, rsi, rdx);
 
         // Fill in memento fields if necessary.
@@ -286,12 +286,12 @@ static void Generate_JSConstructStubHelper(MacroAssembler* masm,
       // rbx: JSObject
       // rdi: start of next object
       // Calculate total properties described map.
-      __ movzxbq(rdx, FieldOperand(rax, Map::kUnusedPropertyFieldsOffset));
-      __ movzxbq(rcx,
+      __ movzxbp(rdx, FieldOperand(rax, Map::kUnusedPropertyFieldsOffset));
+      __ movzxbp(rcx,
                  FieldOperand(rax, Map::kPreAllocatedPropertyFieldsOffset));
       __ addp(rdx, rcx);
       // Calculate unused properties past the end of the in-object properties.
-      __ movzxbq(rcx, FieldOperand(rax, Map::kInObjectPropertiesOffset));
+      __ movzxbp(rcx, FieldOperand(rax, Map::kInObjectPropertiesOffset));
       __ subp(rdx, rcx);
       // Done if no extra properties are to be allocated.
       __ j(zero, &allocated);
@@ -328,7 +328,7 @@ static void Generate_JSConstructStubHelper(MacroAssembler* masm,
       // rdx: number of elements
       { Label loop, entry;
         __ LoadRoot(rdx, Heap::kUndefinedValueRootIndex);
-        __ lea(rcx, Operand(rdi, FixedArray::kHeaderSize));
+        __ leap(rcx, Operand(rdi, FixedArray::kHeaderSize));
         __ jmp(&entry);
         __ bind(&loop);
         __ movp(Operand(rcx, 0), rdx);
@@ -417,7 +417,7 @@ static void Generate_JSConstructStubHelper(MacroAssembler* masm,
     __ Push(rbx);
 
     // Set up pointer to last argument.
-    __ lea(rbx, Operand(rbp, StandardFrameConstants::kCallerSPOffset));
+    __ leap(rbx, Operand(rbp, StandardFrameConstants::kCallerSPOffset));
 
     // Copy arguments and receiver to the expression stack.
     Label loop, entry;
@@ -476,7 +476,7 @@ static void Generate_JSConstructStubHelper(MacroAssembler* masm,
   // Remove caller arguments from the stack and return.
   __ PopReturnAddressTo(rcx);
   SmiIndex index = masm->SmiToIndex(rbx, rbx, kPointerSizeLog2);
-  __ lea(rsp, Operand(rsp, index.reg, index.scale, 1 * kPointerSize));
+  __ leap(rsp, Operand(rsp, index.reg, index.scale, 1 * kPointerSize));
   __ PushReturnAddressFrom(rcx);
   Counters* counters = masm->isolate()->counters();
   __ IncrementCounter(counters->constructed_objects(), 1);
@@ -1222,7 +1222,7 @@ void Builtins::Generate_StringConstructCode(MacroAssembler* masm) {
   __ j(zero, &no_arguments);
   __ movp(rbx, args.GetArgumentOperand(1));
   __ PopReturnAddressTo(rcx);
-  __ lea(rsp, Operand(rsp, rax, times_pointer_size, kPointerSize));
+  __ leap(rsp, Operand(rsp, rax, times_pointer_size, kPointerSize));
   __ PushReturnAddressFrom(rcx);
   __ movp(rax, rbx);
 
@@ -1306,7 +1306,7 @@ void Builtins::Generate_StringConstructCode(MacroAssembler* masm) {
   __ bind(&no_arguments);
   __ LoadRoot(rbx, Heap::kempty_stringRootIndex);
   __ PopReturnAddressTo(rcx);
-  __ lea(rsp, Operand(rsp, kPointerSize));
+  __ leap(rsp, Operand(rsp, kPointerSize));
   __ PushReturnAddressFrom(rcx);
   __ jmp(&argument_is_string);
 
@@ -1352,7 +1352,7 @@ static void LeaveArgumentsAdaptorFrame(MacroAssembler* masm) {
   // Remove caller arguments from the stack.
   __ PopReturnAddressTo(rcx);
   SmiIndex index = masm->SmiToIndex(rbx, rbx, kPointerSizeLog2);
-  __ lea(rsp, Operand(rsp, index.reg, index.scale, 1 * kPointerSize));
+  __ leap(rsp, Operand(rsp, index.reg, index.scale, 1 * kPointerSize));
   __ PushReturnAddressFrom(rcx);
 }
 
@@ -1381,7 +1381,7 @@ void Builtins::Generate_ArgumentsAdaptorTrampoline(MacroAssembler* masm) {
 
     // Copy receiver and all expected arguments.
     const int offset = StandardFrameConstants::kCallerSPOffset;
-    __ lea(rax, Operand(rbp, rax, times_pointer_size, offset));
+    __ leap(rax, Operand(rbp, rax, times_pointer_size, offset));
     __ Set(r8, -1);  // account for receiver
 
     Label copy;
@@ -1400,7 +1400,7 @@ void Builtins::Generate_ArgumentsAdaptorTrampoline(MacroAssembler* masm) {
 
     // Copy receiver and all actual arguments.
     const int offset = StandardFrameConstants::kCallerSPOffset;
-    __ lea(rdi, Operand(rbp, rax, times_pointer_size, offset));
+    __ leap(rdi, Operand(rbp, rax, times_pointer_size, offset));
     __ Set(r8, -1);  // account for receiver
 
     Label copy;
@@ -1469,7 +1469,7 @@ void Builtins::Generate_OnStackReplacement(MacroAssembler* masm) {
       DeoptimizationInputData::kOsrPcOffsetIndex) - kHeapObjectTag));
 
   // Compute the target address = code_obj + header_size + osr_offset
-  __ lea(rax, Operand(rax, rbx, times_1, Code::kHeaderSize - kHeapObjectTag));
+  __ leap(rax, Operand(rax, rbx, times_1, Code::kHeaderSize - kHeapObjectTag));
 
   // Overwrite the return address on the stack.
   __ movq(StackOperandForReturnAddress(0), rax);

src/x64/codegen-x64.cc

@@ -300,7 +300,7 @@ void ElementsTransitionGenerator::GenerateSmiToDouble(
 
   // Allocate new backing store.
   __ bind(&new_backing_store);
-  __ lea(rdi, Operand(r9, times_8, FixedArray::kHeaderSize));
+  __ leap(rdi, Operand(r9, times_8, FixedArray::kHeaderSize));
   __ Allocate(rdi, r14, r11, r15, fail, TAG_OBJECT);
   // Set backing store's map
   __ LoadRoot(rdi, Heap::kFixedDoubleArrayMapRootIndex);
@@ -387,7 +387,7 @@ void ElementsTransitionGenerator::GenerateDoubleToObject(
   __ SmiToInteger32(r9, FieldOperand(r8, FixedDoubleArray::kLengthOffset));
   // r8 : source FixedDoubleArray
   // r9 : number of elements
-  __ lea(rdi, Operand(r9, times_pointer_size, FixedArray::kHeaderSize));
+  __ leap(rdi, Operand(r9, times_pointer_size, FixedArray::kHeaderSize));
   __ Allocate(rdi, r11, r14, r15, &gc_required, TAG_OBJECT);
   // r11: destination FixedArray
   __ LoadRoot(rdi, Heap::kFixedArrayMapRootIndex);
@@ -606,7 +606,7 @@ void MathExpGenerator::EmitMathExp(MacroAssembler* masm,
   __ movq(temp2, double_scratch);
   __ subsd(double_scratch, result);
   __ movsd(result, Operand(kScratchRegister, 6 * kDoubleSize));
-  __ lea(temp1, Operand(temp2, 0x1ff800));
+  __ leaq(temp1, Operand(temp2, 0x1ff800));
   __ and_(temp2, Immediate(0x7ff));
   __ shr(temp1, Immediate(11));
   __ mulsd(double_scratch, Operand(kScratchRegister, 5 * kDoubleSize));

src/x64/debug-x64.cc

@@ -156,7 +156,7 @@ static void Generate_DebugBreakCallHelper(MacroAssembler* masm,
     // Read current padding counter and skip corresponding number of words.
     __ Pop(kScratchRegister);
     __ SmiToInteger32(kScratchRegister, kScratchRegister);
-    __ lea(rsp, Operand(rsp, kScratchRegister, times_pointer_size, 0));
+    __ leap(rsp, Operand(rsp, kScratchRegister, times_pointer_size, 0));
 
     // Get rid of the internal frame.
   }
@@ -327,7 +327,7 @@ void Debug::GenerateFrameDropperLiveEdit(MacroAssembler* masm) {
   __ movp(Operand(rax, 0), Immediate(0));
 
   // We do not know our frame height, but set rsp based on rbp.
-  __ lea(rsp, Operand(rbp, -1 * kPointerSize));
+  __ leap(rsp, Operand(rbp, -1 * kPointerSize));
 
   __ Pop(rdi);  // Function.
   __ popq(rbp);
@@ -338,7 +338,7 @@ void Debug::GenerateFrameDropperLiveEdit(MacroAssembler* masm) {
   // Get function code.
   __ movp(rdx, FieldOperand(rdi, JSFunction::kSharedFunctionInfoOffset));
   __ movp(rdx, FieldOperand(rdx, SharedFunctionInfo::kCodeOffset));
-  __ lea(rdx, FieldOperand(rdx, Code::kHeaderSize));
+  __ leap(rdx, FieldOperand(rdx, Code::kHeaderSize));
 
   // Re-run JSFunction, rdi is function, rsi is context.
   __ jmp(rdx);

src/x64/deoptimizer-x64.cc

@@ -196,7 +196,7 @@ void Deoptimizer::EntryGenerator::Generate() {
   // Get the address of the location in the code object
   // and compute the fp-to-sp delta in register arg5.
   __ movp(arg_reg_4, Operand(rsp, kSavedRegistersAreaSize + 1 * kRegisterSize));
-  __ lea(arg5, Operand(rsp, kSavedRegistersAreaSize + 1 * kRegisterSize +
+  __ leap(arg5, Operand(rsp, kSavedRegistersAreaSize + 1 * kRegisterSize +
                             kPCOnStackSize));
 
   __ subp(arg5, rbp);
@@ -251,7 +251,7 @@ void Deoptimizer::EntryGenerator::Generate() {
   // Unwind the stack down to - but not including - the unwinding
   // limit and copy the contents of the activation frame to the input
   // frame description.
-  __ lea(rdx, Operand(rbx, FrameDescription::frame_content_offset()));
+  __ leap(rdx, Operand(rbx, FrameDescription::frame_content_offset()));
   Label pop_loop_header;
   __ jmp(&pop_loop_header);
   Label pop_loop;
@@ -281,7 +281,7 @@ void Deoptimizer::EntryGenerator::Generate() {
   // last FrameDescription**.
   __ movl(rdx, Operand(rax, Deoptimizer::output_count_offset()));
   __ movp(rax, Operand(rax, Deoptimizer::output_offset()));
-  __ lea(rdx, Operand(rax, rdx, times_pointer_size, 0));
+  __ leap(rdx, Operand(rax, rdx, times_pointer_size, 0));
   __ jmp(&outer_loop_header);
   __ bind(&outer_push_loop);
   // Inner loop state: rbx = current FrameDescription*, rcx = loop index.

src/x64/full-codegen-x64.cc

@@ -271,7 +271,7 @@ void FullCodeGenerator::Generate() {
     // The receiver is just before the parameters on the caller's stack.
     int num_parameters = info->scope()->num_parameters();
     int offset = num_parameters * kPointerSize;
-    __ lea(rdx,
+    __ leap(rdx,
            Operand(rbp, StandardFrameConstants::kCallerSPOffset + offset));
     __ Push(rdx);
     __ Push(Smi::FromInt(num_parameters));
@@ -2012,7 +2012,7 @@ void FullCodeGenerator::VisitYield(Yield* expr) {
       __ movp(rcx, rsi);
       __ RecordWriteField(rax, JSGeneratorObject::kContextOffset, rcx, rdx,
                           kDontSaveFPRegs);
-      __ lea(rbx, Operand(rbp, StandardFrameConstants::kExpressionsOffset));
+      __ leap(rbx, Operand(rbp, StandardFrameConstants::kExpressionsOffset));
       __ cmpp(rsp, rbx);
       __ j(equal, &post_runtime);
       __ Push(rax);  // generator object
@@ -2934,7 +2934,7 @@ void FullCodeGenerator::EmitIsObject(CallRuntime* expr) {
   __ testb(FieldOperand(rbx, Map::kBitFieldOffset),
            Immediate(1 << Map::kIsUndetectable));
   __ j(not_zero, if_false);
-  __ movzxbq(rbx, FieldOperand(rbx, Map::kInstanceTypeOffset));
+  __ movzxbp(rbx, FieldOperand(rbx, Map::kInstanceTypeOffset));
   __ cmpp(rbx, Immediate(FIRST_NONCALLABLE_SPEC_OBJECT_TYPE));
   __ j(below, if_false);
   __ cmpp(rbx, Immediate(LAST_NONCALLABLE_SPEC_OBJECT_TYPE));
@@ -3036,7 +3036,7 @@ void FullCodeGenerator::EmitIsStringWrapperSafeForDefaultValueOf(
   // Calculate the end of the descriptor array.
   __ imulp(rcx, rcx, Immediate(DescriptorArray::kDescriptorSize));
   SmiIndex index = masm_->SmiToIndex(rdx, rcx, kPointerSizeLog2);
-  __ lea(rcx,
+  __ leap(rcx,
          Operand(
              r8, index.reg, index.scale, DescriptorArray::kFirstOffset));
   // Calculate location of the first key name.
@@ -4004,7 +4004,7 @@ void FullCodeGenerator::EmitFastAsciiArrayJoin(CallRuntime* expr) {
   __ AllocateAsciiString(result_pos, string_length, scratch,
                          index, string, &bailout);
   __ movp(result_operand, result_pos);
-  __ lea(result_pos, FieldOperand(result_pos, SeqOneByteString::kHeaderSize));
+  __ leap(result_pos, FieldOperand(result_pos, SeqOneByteString::kHeaderSize));
 
   __ movp(string, separator_operand);
   __ SmiCompare(FieldOperand(string, SeqOneByteString::kLengthOffset),
@@ -4032,7 +4032,7 @@ void FullCodeGenerator::EmitFastAsciiArrayJoin(CallRuntime* expr) {
                                FixedArray::kHeaderSize));
   __ SmiToInteger32(string_length,
                     FieldOperand(string, String::kLengthOffset));
-  __ lea(string,
+  __ leap(string,
          FieldOperand(string, SeqOneByteString::kHeaderSize));
   __ CopyBytes(result_pos, string, string_length);
   __ incl(index);
@@ -4076,7 +4076,7 @@ void FullCodeGenerator::EmitFastAsciiArrayJoin(CallRuntime* expr) {
                                FixedArray::kHeaderSize));
   __ SmiToInteger32(string_length,
                     FieldOperand(string, String::kLengthOffset));
-  __ lea(string,
+  __ leap(string,
          FieldOperand(string, SeqOneByteString::kHeaderSize));
   __ CopyBytes(result_pos, string, string_length);
   __ incl(index);
@@ -4092,7 +4092,7 @@ void FullCodeGenerator::EmitFastAsciiArrayJoin(CallRuntime* expr) {
   // count from -array_length to zero, so we don't need to maintain
   // a loop limit.
   __ movl(index, array_length_operand);
-  __ lea(elements, FieldOperand(elements, index, times_pointer_size,
+  __ leap(elements, FieldOperand(elements, index, times_pointer_size,
                                 FixedArray::kHeaderSize));
   __ negq(index);
 
@@ -4101,7 +4101,7 @@ void FullCodeGenerator::EmitFastAsciiArrayJoin(CallRuntime* expr) {
   __ movp(string, separator_operand);
   __ SmiToInteger32(scratch,
                     FieldOperand(string, String::kLengthOffset));
-  __ lea(string,
+  __ leap(string,
          FieldOperand(string, SeqOneByteString::kHeaderSize));
   __ movp(separator_operand, string);
 
@@ -4127,7 +4127,7 @@ void FullCodeGenerator::EmitFastAsciiArrayJoin(CallRuntime* expr) {
   __ movp(string, Operand(elements, index, times_pointer_size, 0));
   __ SmiToInteger32(string_length,
                     FieldOperand(string, String::kLengthOffset));
-  __ lea(string,
+  __ leap(string,
          FieldOperand(string, SeqOneByteString::kHeaderSize));
   __ CopyBytes(result_pos, string, string_length);
   __ incq(index);

src/x64/ic-x64.cc

@@ -212,7 +212,7 @@ static void GenerateDictionaryStore(MacroAssembler* masm,
 
   // Store the value at the masked, scaled index.
   const int kValueOffset = kElementsStartOffset + kPointerSize;
-  __ lea(scratch1, Operand(elements,
+  __ leap(scratch1, Operand(elements,
                            scratch1,
                            times_pointer_size,
                            kValueOffset - kHeapObjectTag));
@@ -467,7 +467,7 @@ void KeyedLoadIC::GenerateGeneric(MacroAssembler* masm) {
     }
     __ LoadAddress(kScratchRegister, cache_field_offsets);
     __ movl(rdi, Operand(kScratchRegister, rcx, times_4, 0));
-    __ movzxbq(rcx, FieldOperand(rbx, Map::kInObjectPropertiesOffset));
+    __ movzxbp(rcx, FieldOperand(rbx, Map::kInObjectPropertiesOffset));
     __ subp(rdi, rcx);
     __ j(above_equal, &property_array_property);
     if (i != 0) {
@@ -477,7 +477,7 @@ void KeyedLoadIC::GenerateGeneric(MacroAssembler* masm) {
 
   // Load in-object property.
   __ bind(&load_in_object_property);
-  __ movzxbq(rcx, FieldOperand(rbx, Map::kInstanceSizeOffset));
+  __ movzxbp(rcx, FieldOperand(rbx, Map::kInstanceSizeOffset));
   __ addp(rcx, rdi);
   __ movp(rax, FieldOperand(rdx, rcx, times_pointer_size, 0));
   __ IncrementCounter(counters->keyed_load_generic_lookup_cache(), 1);
@@ -945,7 +945,7 @@ void KeyedStoreIC::GenerateSloppyArguments(MacroAssembler* masm) {
   Operand mapped_location = GenerateMappedArgumentsLookup(
       masm, rdx, rcx, rbx, rdi, r8, &notin, &slow);
   __ movp(mapped_location, rax);
-  __ lea(r9, mapped_location);
+  __ leap(r9, mapped_location);
   __ movp(r8, rax);
   __ RecordWrite(rbx,
                  r9,
@@ -959,7 +959,7 @@ void KeyedStoreIC::GenerateSloppyArguments(MacroAssembler* masm) {
   Operand unmapped_location =
       GenerateUnmappedArgumentsLookup(masm, rcx, rbx, rdi, &slow);
   __ movp(unmapped_location, rax);
-  __ lea(r9, unmapped_location);
+  __ leap(r9, unmapped_location);
   __ movp(r8, rax);
   __ RecordWrite(rbx,
                  r9,

src/x64/lithium-codegen-x64.cc

@@ -356,7 +356,7 @@ bool LCodeGen::GenerateDeferredCode() {
         __ pushq(rbp);  // Caller's frame pointer.
         __ Push(Operand(rbp, StandardFrameConstants::kContextOffset));
         __ Push(Smi::FromInt(StackFrame::STUB));
-        __ lea(rbp, Operand(rsp, 2 * kPointerSize));
+        __ leap(rbp, Operand(rsp, 2 * kPointerSize));
         Comment(";;; Deferred code");
       }
       code->Generate();
@@ -1687,7 +1687,7 @@ void LCodeGen::DoSeqStringGetChar(LSeqStringGetChar* instr) {
   if (FLAG_debug_code) {
     __ Push(string);
     __ movp(string, FieldOperand(string, HeapObject::kMapOffset));
-    __ movzxbq(string, FieldOperand(string, Map::kInstanceTypeOffset));
+    __ movzxbp(string, FieldOperand(string, Map::kInstanceTypeOffset));
 
     __ andb(string, Immediate(kStringRepresentationMask | kStringEncodingMask));
     static const uint32_t one_byte_seq_type = kSeqStringTag | kOneByteStringTag;
@@ -1755,7 +1755,7 @@ void LCodeGen::DoAddI(LAddI* instr) {
     if (right->IsConstantOperand()) {
       int32_t offset = ToInteger32(LConstantOperand::cast(right));
       if (is_p) {
-        __ lea(ToRegister(instr->result()),
+        __ leap(ToRegister(instr->result()),
                 MemOperand(ToRegister(left), offset));
       } else {
         __ leal(ToRegister(instr->result()),
@@ -1764,7 +1764,7 @@ void LCodeGen::DoAddI(LAddI* instr) {
     } else {
       Operand address(ToRegister(left), ToRegister(right), times_1, 0);
       if (is_p) {
-        __ lea(ToRegister(instr->result()), address);
+        __ leap(ToRegister(instr->result()), address);
       } else {
         __ leal(ToRegister(instr->result()), address);
       }
@@ -2978,7 +2978,7 @@ void LCodeGen::DoLoadKeyedExternalArray(LLoadKeyed* instr) {
       case EXTERNAL_UINT8_CLAMPED_ELEMENTS:
       case UINT8_ELEMENTS:
       case UINT8_CLAMPED_ELEMENTS:
-        __ movzxbq(result, operand);
+        __ movzxbp(result, operand);
         break;
       case EXTERNAL_INT16_ELEMENTS:
       case INT16_ELEMENTS:
@@ -2986,7 +2986,7 @@ void LCodeGen::DoLoadKeyedExternalArray(LLoadKeyed* instr) {
         break;
       case EXTERNAL_UINT16_ELEMENTS:
       case UINT16_ELEMENTS:
-        __ movzxwq(result, operand);
+        __ movzxwp(result, operand);
         break;
       case EXTERNAL_INT32_ELEMENTS:
       case INT32_ELEMENTS:
@@ -3172,7 +3172,7 @@ void LCodeGen::DoArgumentsElements(LArgumentsElements* instr) {
   Register result = ToRegister(instr->result());
 
   if (instr->hydrogen()->from_inlined()) {
-    __ lea(result, Operand(rsp, -kFPOnStackSize + -kPCOnStackSize));
+    __ leap(result, Operand(rsp, -kFPOnStackSize + -kPCOnStackSize));
   } else {
     // Check for arguments adapter frame.
     Label done, adapted;
@@ -3907,7 +3907,7 @@ void LCodeGen::DoCallRuntime(LCallRuntime* instr) {
 void LCodeGen::DoStoreCodeEntry(LStoreCodeEntry* instr) {
   Register function = ToRegister(instr->function());
   Register code_object = ToRegister(instr->code_object());
-  __ lea(code_object, FieldOperand(code_object, Code::kHeaderSize));
+  __ leap(code_object, FieldOperand(code_object, Code::kHeaderSize));
   __ movp(FieldOperand(function, JSFunction::kCodeEntryOffset), code_object);
 }
 
@@ -3917,10 +3917,10 @@ void LCodeGen::DoInnerAllocatedObject(LInnerAllocatedObject* instr) {
   Register base = ToRegister(instr->base_object());
   if (instr->offset()->IsConstantOperand()) {
     LConstantOperand* offset = LConstantOperand::cast(instr->offset());
-    __ lea(result, Operand(base, ToInteger32(offset)));
+    __ leap(result, Operand(base, ToInteger32(offset)));
   } else {
     Register offset = ToRegister(instr->offset());
-    __ lea(result, Operand(base, offset, times_1, 0));
+    __ leap(result, Operand(base, offset, times_1, 0));
   }
 }
 
@@ -4322,7 +4322,7 @@ void LCodeGen::DoStoreKeyedFixedArray(LStoreKeyed* instr) {
             ? OMIT_SMI_CHECK : INLINE_SMI_CHECK;
     // Compute address of modified element and store it into key register.
     Register key_reg(ToRegister(key));
-    __ lea(key_reg, operand);
+    __ leap(key_reg, operand);
     __ RecordWrite(elements,
                    key_reg,
                    value,

src/x64/regexp-macro-assembler-x64.cc

@@ -203,7 +203,7 @@ void RegExpMacroAssemblerX64::CheckAtStart(Label* on_at_start) {
   __ cmpl(Operand(rbp, kStartIndex), Immediate(0));
   BranchOrBacktrack(not_equal, &not_at_start);
   // If we did, are we still at the start of the input?
-  __ lea(rax, Operand(rsi, rdi, times_1, 0));
+  __ leap(rax, Operand(rsi, rdi, times_1, 0));
   __ cmpp(rax, Operand(rbp, kInputStart));
   BranchOrBacktrack(equal, on_at_start);
   __ bind(&not_at_start);
@@ -215,7 +215,7 @@ void RegExpMacroAssemblerX64::CheckNotAtStart(Label* on_not_at_start) {
   __ cmpl(Operand(rbp, kStartIndex), Immediate(0));
   BranchOrBacktrack(not_equal, on_not_at_start);
   // If we did, are we still at the start of the input?
-  __ lea(rax, Operand(rsi, rdi, times_1, 0));
+  __ leap(rax, Operand(rsi, rdi, times_1, 0));
   __ cmpp(rax, Operand(rbp, kInputStart));
   BranchOrBacktrack(not_equal, on_not_at_start);
 }
@@ -273,8 +273,8 @@ void RegExpMacroAssemblerX64::CheckNotBackReferenceIgnoreCase(
       on_no_match = &backtrack_label_;
     }
 
-    __ lea(r9, Operand(rsi, rdx, times_1, 0));
-    __ lea(r11, Operand(rsi, rdi, times_1, 0));
+    __ leap(r9, Operand(rsi, rdx, times_1, 0));
+    __ leap(r11, Operand(rsi, rdi, times_1, 0));
     __ addp(rbx, r9);  // End of capture
     // ---------------------
     // r11 - current input character address
@@ -337,18 +337,18 @@ void RegExpMacroAssemblerX64::CheckNotBackReferenceIgnoreCase(
     //   Isolate* isolate
 #ifdef _WIN64
     // Compute and set byte_offset1 (start of capture).
-    __ lea(rcx, Operand(rsi, rdx, times_1, 0));
+    __ leap(rcx, Operand(rsi, rdx, times_1, 0));
     // Set byte_offset2.
-    __ lea(rdx, Operand(rsi, rdi, times_1, 0));
+    __ leap(rdx, Operand(rsi, rdi, times_1, 0));
     // Set byte_length.
     __ movp(r8, rbx);
     // Isolate.
     __ LoadAddress(r9, ExternalReference::isolate_address(isolate()));
 #else  // AMD64 calling convention
     // Compute byte_offset2 (current position = rsi+rdi).
-    __ lea(rax, Operand(rsi, rdi, times_1, 0));
+    __ leap(rax, Operand(rsi, rdi, times_1, 0));
     // Compute and set byte_offset1 (start of capture).
-    __ lea(rdi, Operand(rsi, rdx, times_1, 0));
+    __ leap(rdi, Operand(rsi, rdx, times_1, 0));
     // Set byte_offset2.
     __ movp(rsi, rax);
     // Set byte_length.
@@ -412,9 +412,9 @@ void RegExpMacroAssemblerX64::CheckNotBackReference(
   BranchOrBacktrack(greater, on_no_match);
 
   // Compute pointers to match string and capture string
-  __ lea(rbx, Operand(rsi, rdi, times_1, 0));  // Start of match.
+  __ leap(rbx, Operand(rsi, rdi, times_1, 0));  // Start of match.
   __ addp(rdx, rsi);  // Start of capture.
-  __ lea(r9, Operand(rdx, rax, times_1, 0));  // End of capture
+  __ leap(r9, Operand(rdx, rax, times_1, 0));  // End of capture
 
   // -----------------------
   // rbx - current capture character address.
@@ -489,7 +489,7 @@ void RegExpMacroAssemblerX64::CheckNotCharacterAfterMinusAnd(
     uc16 mask,
     Label* on_not_equal) {
   ASSERT(minus < String::kMaxUtf16CodeUnit);
-  __ lea(rax, Operand(current_character(), -minus));
+  __ leap(rax, Operand(current_character(), -minus));
   __ and_(rax, Immediate(mask));
   __ cmpl(rax, Immediate(c));
   BranchOrBacktrack(not_equal, on_not_equal);
@@ -536,7 +536,7 @@ bool RegExpMacroAssemblerX64::CheckSpecialCharacterClass(uc16 type,
                                                          Label* on_no_match) {
   // Range checks (c in min..max) are generally implemented by an unsigned
   // (c - min) <= (max - min) check, using the sequence:
-  //   lea(rax, Operand(current_character(), -min)) or sub(rax, Immediate(min))
+  //   leap(rax, Operand(current_character(), -min)) or sub(rax, Immediate(min))
   //   cmp(rax, Immediate(max - min))
   switch (type) {
   case 's':
@@ -547,7 +547,7 @@ bool RegExpMacroAssemblerX64::CheckSpecialCharacterClass(uc16 type,
       __ cmpl(current_character(), Immediate(' '));
       __ j(equal, &success, Label::kNear);
       // Check range 0x09..0x0d
-      __ lea(rax, Operand(current_character(), -'\t'));
+      __ leap(rax, Operand(current_character(), -'\t'));
       __ cmpl(rax, Immediate('\r' - '\t'));
       __ j(below_equal, &success, Label::kNear);
       // \u00a0 (NBSP).
@@ -562,13 +562,13 @@ bool RegExpMacroAssemblerX64::CheckSpecialCharacterClass(uc16 type,
     return false;
   case 'd':
     // Match ASCII digits ('0'..'9')
-    __ lea(rax, Operand(current_character(), -'0'));
+    __ leap(rax, Operand(current_character(), -'0'));
     __ cmpl(rax, Immediate('9' - '0'));
     BranchOrBacktrack(above, on_no_match);
     return true;
   case 'D':
     // Match non ASCII-digits
-    __ lea(rax, Operand(current_character(), -'0'));
+    __ leap(rax, Operand(current_character(), -'0'));
     __ cmpl(rax, Immediate('9' - '0'));
     BranchOrBacktrack(below_equal, on_no_match);
     return true;
@@ -753,9 +753,9 @@ Handle<HeapObject> RegExpMacroAssemblerX64::GetCode(Handle<String> source) {
   __ movp(rbx, Operand(rbp, kStartIndex));
   __ negq(rbx);
   if (mode_ == UC16) {
-    __ lea(rax, Operand(rdi, rbx, times_2, -char_size()));
+    __ leap(rax, Operand(rdi, rbx, times_2, -char_size()));
   } else {
-    __ lea(rax, Operand(rdi, rbx, times_1, -char_size()));
+    __ leap(rax, Operand(rdi, rbx, times_1, -char_size()));
   }
   // Store this value in a local variable, for use when clearing
   // position registers.
@@ -826,7 +826,7 @@ Handle<HeapObject> RegExpMacroAssemblerX64::GetCode(Handle<String> source) {
       __ movp(rcx, Operand(rbp, kInputEnd));
       __ subp(rcx, Operand(rbp, kInputStart));
       if (mode_ == UC16) {
-        __ lea(rcx, Operand(rcx, rdx, times_2, 0));
+        __ leap(rcx, Operand(rcx, rdx, times_2, 0));
       } else {
         __ addp(rcx, rdx);
       }
@@ -896,7 +896,7 @@ Handle<HeapObject> RegExpMacroAssemblerX64::GetCode(Handle<String> source) {
   __ bind(&return_rax);
 #ifdef _WIN64
   // Restore callee save registers.
-  __ lea(rsp, Operand(rbp, kLastCalleeSaveRegister));
+  __ leap(rsp, Operand(rbp, kLastCalleeSaveRegister));
   __ popq(rbx);
   __ popq(rdi);
   __ popq(rsi);
@@ -960,12 +960,12 @@ Handle<HeapObject> RegExpMacroAssemblerX64::GetCode(Handle<String> source) {
 #ifdef _WIN64
     // Microsoft passes parameters in rcx, rdx, r8.
     // First argument, backtrack stackpointer, is already in rcx.
-    __ lea(rdx, Operand(rbp, kStackHighEnd));  // Second argument
+    __ leap(rdx, Operand(rbp, kStackHighEnd));  // Second argument
     __ LoadAddress(r8, ExternalReference::isolate_address(isolate()));
 #else
     // AMD64 ABI passes parameters in rdi, rsi, rdx.
     __ movp(rdi, backtrack_stackpointer());   // First argument.
-    __ lea(rsi, Operand(rbp, kStackHighEnd));  // Second argument.
+    __ leap(rsi, Operand(rbp, kStackHighEnd));  // Second argument.
     __ LoadAddress(rdx, ExternalReference::isolate_address(isolate()));
 #endif
     ExternalReference grow_stack =
@@ -1125,7 +1125,7 @@ void RegExpMacroAssemblerX64::WriteCurrentPositionToRegister(int reg,
   if (cp_offset == 0) {
     __ movp(register_location(reg), rdi);
   } else {
-    __ lea(rax, Operand(rdi, cp_offset * char_size()));
+    __ leap(rax, Operand(rdi, cp_offset * char_size()));
     __ movp(register_location(reg), rax);
   }
 }
@@ -1161,7 +1161,7 @@ void RegExpMacroAssemblerX64::CallCheckStackGuardState() {
   __ movp(r8, rbp);
   // First argument: Next address on the stack (will be address of
   // return address).
-  __ lea(rcx, Operand(rsp, -kPointerSize));
+  __ leap(rcx, Operand(rsp, -kPointerSize));
 #else
   // Third argument: RegExp code frame pointer.
   __ movp(rdx, rbp);
@@ -1169,7 +1169,7 @@ void RegExpMacroAssemblerX64::CallCheckStackGuardState() {
   __ movp(rsi, code_object_pointer());
   // First argument: Next address on the stack (will be address of
   // return address).
-  __ lea(rdi, Operand(rsp, -kPointerSize));
+  __ leap(rdi, Operand(rsp, -kRegisterSize));
 #endif
   ExternalReference stack_check =
       ExternalReference::re_check_stack_guard_state(isolate());

src/x64/stub-cache-x64.cc

@@ -64,7 +64,7 @@ static void ProbeTable(Isolate* isolate,
   Label miss;
 
   // Multiply by 3 because there are 3 fields per entry (name, code, map).
-  __ lea(offset, Operand(offset, offset, times_2, 0));
+  __ leap(offset, Operand(offset, offset, times_2, 0));
 
   __ LoadAddress(kScratchRegister, key_offset);
 

test/cctest/test-code-stubs-x64.cc

@@ -70,7 +70,7 @@ ConvertDToIFunc MakeConvertDToIFuncTrampoline(Isolate* isolate,
     // registers.
     int double_argument_slot =
         (Register::NumAllocatableRegisters() - 1) * kPointerSize + kDoubleSize;
-    __ lea(source_reg, MemOperand(rsp, -double_argument_slot - offset));
+    __ leaq(source_reg, MemOperand(rsp, -double_argument_slot - offset));
   }
 
   // Save registers make sure they don't get clobbered.

test/cctest/test-disasm-x64.cc

@@ -173,7 +173,7 @@ TEST(DisasmX64) {
   // TODO(mstarzinger): The following is protected.
   // __ jmp(Operand(rbx, rcx, times_4, 10000));
 
-  __ lea(rdx, Operand(rbx, rcx, times_4, 10000));
+  __ leaq(rdx, Operand(rbx, rcx, times_4, 10000));
   __ or_(rdx, Immediate(12345));
   __ or_(rdx, Operand(rbx, rcx, times_4, 10000));
 

test/cctest/test-macro-assembler-x64.cc

@@ -2343,9 +2343,9 @@ TEST(OperandOffset) {
   // r15 = rsp[3]
   // rbx = rsp[5]
   // r13 = rsp[7]
-  __ lea(r14, Operand(rsp, 3 * kPointerSize));
-  __ lea(r13, Operand(rbp, -3 * kPointerSize));
-  __ lea(rbx, Operand(rbp, -5 * kPointerSize));
+  __ leaq(r14, Operand(rsp, 3 * kPointerSize));
+  __ leaq(r13, Operand(rbp, -3 * kPointerSize));
+  __ leaq(rbx, Operand(rbp, -5 * kPointerSize));
   __ movl(rcx, Immediate(2));
   __ Move(r8, reinterpret_cast<Address>(&data[128]), RelocInfo::NONE64);
   __ movl(rax, Immediate(1));
@@ -2643,7 +2643,7 @@ TEST(OperandOffset) {
 
   __ movl(rax, Immediate(0));
   __ bind(&exit);
-  __ lea(rsp, Operand(rbp, kPointerSize));
+  __ leaq(rsp, Operand(rbp, kPointerSize));
   __ popq(rbp);
   __ popq(rbx);
   __ popq(r14);