deps/v8/src/codegen/x64/macro-assembler-x64.cc - main - githubmirror/node

deps/v8/src/codegen/x64/macro-assembler-x64.cc
5 158 строк175 KB
Michaël Zasso
deps: update V8 to 14.6.202.33
24 апр 2026, 19:01
24 апр 2026, 19:01f1e0b83
// Copyright 2012 the V8 project authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

#include <climits>
#include <cstdint>

#include "src/sandbox/js-dispatch-table.h"

#if V8_TARGET_ARCH_X64

#include <optional>

#include "src/base/bits.h"
#include "src/base/division-by-constant.h"
#include "src/base/utils/random-number-generator.h"
#include "src/builtins/builtins-inl.h"
#include "src/codegen/callable.h"
#include "src/codegen/code-factory.h"
#include "src/codegen/cpu-features.h"
#include "src/codegen/external-reference-table.h"
#include "src/codegen/interface-descriptors-inl.h"
#include "src/codegen/macro-assembler.h"
#include "src/codegen/register-configuration.h"
#include "src/codegen/register.h"
#include "src/codegen/x64/assembler-x64.h"
#include "src/codegen/x64/register-x64.h"
#include "src/common/globals.h"
#include "src/debug/debug.h"
#include "src/deoptimizer/deoptimizer.h"
#include "src/execution/frames-inl.h"
#include "src/heap/mutable-page.h"
#include "src/init/bootstrapper.h"
#include "src/logging/counters.h"
#include "src/objects/instance-type-inl.h"
#include "src/objects/objects-inl.h"
#include "src/objects/smi.h"
#include "src/sandbox/external-pointer.h"
#include "src/snapshot/snapshot.h"

// Satisfy cpplint check, but don't include platform-specific header. It is
// included recursively via macro-assembler.h.
#if 0
#include "src/codegen/x64/macro-assembler-x64.h"
#endif

#define __ ACCESS_MASM(masm)

namespace v8 {
namespace internal {

Operand StackArgumentsAccessor::GetArgumentOperand(int index) const {
  DCHECK_GE(index, 0);
  // arg[0] = rsp + kPCOnStackSize;
  // arg[i] = arg[0] + i * kSystemPointerSize;
  return Operand(rsp, kPCOnStackSize + index * kSystemPointerSize);
}

void MacroAssembler::CodeEntry() {
  endbr64();
}

void MacroAssembler::ExceptionHandler() {
  CodeEntry();

  // Exception handlers are always invoked in sandboxed execution mode.
  AssertInSandboxedExecutionMode();
  // In case we're currently assembling the code of an unsandboxed builtin
  // (e.g. RunMicrotasks), we now need to exit sandboxed execution mode.
  if (sandboxing_mode() == CodeSandboxingMode::kUnsandboxed) {
    ExitSandbox();
  }
}

void MacroAssembler::Load(Register destination, ExternalReference source) {
  if (root_array_available_ && options().enable_root_relative_access) {
    intptr_t delta = RootRegisterOffsetForExternalReference(isolate(), source);
    if (is_int32(delta)) {
      movq(destination, Operand(kRootRegister, static_cast<int32_t>(delta)));
      return;
    }
  }
  // Safe code.
  if (destination == rax && !options().isolate_independent_code) {
    load_rax(source);
  } else {
    movq(destination, ExternalReferenceAsOperand(source));
  }
}

void MacroAssembler::Store(ExternalReference destination, Register source) {
  if (root_array_available_ && options().enable_root_relative_access) {
    intptr_t delta =
        RootRegisterOffsetForExternalReference(isolate(), destination);
    if (is_int32(delta)) {
      movq(Operand(kRootRegister, static_cast<int32_t>(delta)), source);
      return;
    }
  }
  // Safe code.
  if (source == rax && !options().isolate_independent_code) {
    store_rax(destination);
  } else {
    movq(ExternalReferenceAsOperand(destination), source);
  }
}

void MacroAssembler::LoadFromConstantsTable(Register destination,
                                            int constant_index) {
  DCHECK(RootsTable::IsImmortalImmovable(RootIndex::kBuiltinsConstantsTable));
  LoadRoot(destination, RootIndex::kBuiltinsConstantsTable);
  LoadTaggedField(
      destination,
      FieldOperand(destination, FixedArray::OffsetOfElementAt(constant_index)));
}

void MacroAssembler::LoadRootRegisterOffset(Register destination,
                                            intptr_t offset) {
  DCHECK(is_int32(offset));
  if (offset == 0) {
    Move(destination, kRootRegister);
  } else {
    leaq(destination, Operand(kRootRegister, static_cast<int32_t>(offset)));
  }
}

void MacroAssembler::LoadRootRelative(Register destination, int32_t offset) {
  movq(destination, Operand(kRootRegister, offset));
}

void MacroAssembler::StoreRootRelative(int32_t offset, Register value) {
  movq(Operand(kRootRegister, offset), value);
}

void MacroAssembler::LoadAddress(Register destination,
                                 ExternalReference source) {
  if (root_array_available()) {
    if (source.IsIsolateFieldId()) {
      leaq(destination,
           Operand(kRootRegister, source.offset_from_root_register()));
      return;
    }
    if (options().enable_root_relative_access) {
      intptr_t delta =
          RootRegisterOffsetForExternalReference(isolate(), source);
      if (is_int32(delta)) {
        leaq(destination, Operand(kRootRegister, static_cast<int32_t>(delta)));
        return;
      }
    } else if (options().isolate_independent_code) {
      IndirectLoadExternalReference(destination, source);
      return;
    }
  }
  Move(destination, source);
}

Operand MacroAssembler::ExternalReferenceAsOperand(ExternalReference reference,
                                                   Register scratch) {
  if (root_array_available()) {
    if (reference.IsIsolateFieldId()) {
      return Operand(kRootRegister, reference.offset_from_root_register());
    }
    if (options().enable_root_relative_access) {
      int64_t delta =
          RootRegisterOffsetForExternalReference(isolate(), reference);
      if (is_int32(delta)) {
        return Operand(kRootRegister, static_cast<int32_t>(delta));
      }
    }
    if (options().isolate_independent_code) {
      if (IsAddressableThroughRootRegister(isolate(), reference)) {
        // Some external references can be efficiently loaded as an offset from
        // kRootRegister.
        intptr_t offset =
            RootRegisterOffsetForExternalReference(isolate(), reference);
        CHECK(is_int32(offset));
        return Operand(kRootRegister, static_cast<int32_t>(offset));
      } else {
        // Otherwise, do a memory load from the external reference table.
        movq(scratch, Operand(kRootRegister,
                              RootRegisterOffsetForExternalReferenceTableEntry(
                                  isolate(), reference)));
        return Operand(scratch, 0);
      }
    }
  }
  Move(scratch, reference);
  return Operand(scratch, 0);
}

void MacroAssembler::PushAddress(ExternalReference source, Register scratch) {
  LoadAddress(scratch, source);
  Push(scratch);
}

Operand MacroAssembler::RootAsOperand(RootIndex index) {
  DCHECK(root_array_available());
  return Operand(kRootRegister, RootRegisterOffsetForRootIndex(index));
}

void MacroAssembler::LoadTaggedRoot(Register destination, RootIndex index) {
  static_assert(!CanBeImmediate(RootIndex::kUndefinedValue) ||
                std::is_same_v<Tagged_t, uint32_t>);
  if (CanBeImmediate(index)) {
    mov_tagged(destination,
               Immediate(static_cast<uint32_t>(ReadOnlyRootPtr(index))));
    return;
  }
  DCHECK(root_array_available_);
  movq(destination, RootAsOperand(index));
}

void MacroAssembler::LoadRoot(Register destination, RootIndex index) {
  if (CanBeImmediate(index)) {
    DecompressTagged(destination,
                     static_cast<uint32_t>(ReadOnlyRootPtr(index)));
    return;
  }
  DCHECK(root_array_available_);
  movq(destination, RootAsOperand(index));
}

void MacroAssembler::PushRoot(RootIndex index) {
  DCHECK(root_array_available_);
  Push(RootAsOperand(index));
}

void MacroAssembler::CompareRoot(Register with, RootIndex index,
                                 ComparisonMode mode) {
  if (mode == ComparisonMode::kFullPointer ||
      !base::IsInRange(index, RootIndex::kFirstStrongOrReadOnlyRoot,
                       RootIndex::kLastStrongOrReadOnlyRoot)) {
    // Some smi roots contain system pointer size values like stack limits.
    cmpq(with, RootAsOperand(index));
    return;
  }
  CompareTaggedRoot(with, index);
}

void MacroAssembler::CompareTaggedRoot(Register with, RootIndex index) {
  AssertSmiOrHeapObjectInMainCompressionCage(with);
  if (CanBeImmediate(index)) {
    cmp_tagged(with, Immediate(static_cast<uint32_t>(ReadOnlyRootPtr(index))));
    return;
  }
  DCHECK(root_array_available_);
  // Some smi roots contain system pointer size values like stack limits.
  DCHECK(base::IsInRange(index, RootIndex::kFirstStrongOrReadOnlyRoot,
                         RootIndex::kLastStrongOrReadOnlyRoot));
  cmp_tagged(with, RootAsOperand(index));
}

void MacroAssembler::CompareRoot(Operand with, RootIndex index) {
  if (CanBeImmediate(index)) {
    cmp_tagged(with, Immediate(static_cast<uint32_t>(ReadOnlyRootPtr(index))));
    return;
  }
  DCHECK(root_array_available_);
  DCHECK(!with.AddressUsesRegister(kScratchRegister));
  if (base::IsInRange(index, RootIndex::kFirstStrongOrReadOnlyRoot,
                      RootIndex::kLastStrongOrReadOnlyRoot)) {
    mov_tagged(kScratchRegister, RootAsOperand(index));
    cmp_tagged(with, kScratchRegister);
  } else {
    // Some smi roots contain system pointer size values like stack limits.
    movq(kScratchRegister, RootAsOperand(index));
    cmpq(with, kScratchRegister);
  }
}

void MacroAssembler::LoadCompressedMap(Register destination, Register object) {
  CHECK(COMPRESS_POINTERS_BOOL);
  mov_tagged(destination, FieldOperand(object, HeapObject::kMapOffset));
}

void MacroAssembler::LoadMap(Register destination, Register object) {
  LoadTaggedField(destination, FieldOperand(object, HeapObject::kMapOffset));
#ifdef V8_MAP_PACKING
  UnpackMapWord(destination);
#endif
}

void MacroAssembler::LoadFeedbackVector(Register dst, Register closure,
                                        Label* fbv_undef,
                                        Label::Distance distance) {
  Label done;

  // Load the feedback vector from the closure.
  TaggedRegister feedback_cell(dst);
  LoadTaggedField(feedback_cell,
                  FieldOperand(closure, JSFunction::kFeedbackCellOffset));
  LoadTaggedField(dst, FieldOperand(feedback_cell, FeedbackCell::kValueOffset));

  // Check if feedback vector is valid.
  IsObjectType(dst, FEEDBACK_VECTOR_TYPE, rcx);
  j(equal, &done, Label::kNear);

  // Not valid, load undefined.
  LoadRoot(dst, RootIndex::kUndefinedValue);
  jmp(fbv_undef, distance);

  bind(&done);
}

void MacroAssembler::LoadInterpreterDataBytecodeArray(
    Register destination, Register interpreter_data) {
  LoadProtectedPointerField(
      destination, FieldOperand(interpreter_data,
                                offsetof(InterpreterData, bytecode_array_)));
}

void MacroAssembler::LoadInterpreterDataInterpreterTrampoline(
    Register destination, Register interpreter_data) {
  LoadProtectedPointerField(
      destination,
      FieldOperand(interpreter_data,
                   offsetof(InterpreterData, interpreter_trampoline_)));
}

void MacroAssembler::LoadTaggedField(Register destination,
                                     Operand field_operand) {
  if (COMPRESS_POINTERS_BOOL) {
    DecompressTagged(destination, field_operand);
  } else {
    mov_tagged(destination, field_operand);
  }
}

void MacroAssembler::LoadTaggedField(TaggedRegister destination,
                                     Operand field_operand) {
  LoadTaggedFieldWithoutDecompressing(destination.reg(), field_operand);
}

void MacroAssembler::LoadTaggedFieldWithoutDecompressing(
    Register destination, Operand field_operand) {
  mov_tagged(destination, field_operand);
}

#ifdef V8_MAP_PACKING
void MacroAssembler::UnpackMapWord(Register r) {
  // Clear the top two bytes (which may include metadata). Must be in sync with
  // MapWord::Unpack, and vice versa.
  shlq(r, Immediate(16));
  shrq(r, Immediate(16));
  xorq(r, Immediate(Internals::kMapWordXorMask));
}
#endif

void MacroAssembler::LoadTaggedSignedField(Register destination,
                                           Operand field_operand) {
  if (COMPRESS_POINTERS_BOOL) {
    DecompressTaggedSigned(destination, field_operand);
  } else {
    mov_tagged(destination, field_operand);
  }
}

void MacroAssembler::PushTaggedField(Operand field_operand, Register scratch) {
  if (COMPRESS_POINTERS_BOOL) {
    DCHECK(!field_operand.AddressUsesRegister(scratch));
    DecompressTagged(scratch, field_operand);
    Push(scratch);
  } else {
    Push(field_operand);
  }
}

void MacroAssembler::SmiUntagField(Register dst, Operand src) {
  SmiUntag(dst, src);
}

void MacroAssembler::SmiUntagFieldUnsigned(Register dst, Operand src) {
  SmiUntagUnsigned(dst, src);
}

void MacroAssembler::StoreTaggedField(Operand dst_field_operand,
                                      Immediate value) {
  if (COMPRESS_POINTERS_BOOL) {
    movl(dst_field_operand, value);
  } else {
    movq(dst_field_operand, value);
  }
}

void MacroAssembler::StoreTaggedField(Operand dst_field_operand,
                                      Register value) {
  if (COMPRESS_POINTERS_BOOL) {
    movl(dst_field_operand, value);
  } else {
    movq(dst_field_operand, value);
  }
}

void MacroAssembler::StoreTaggedSignedField(Operand dst_field_operand,
                                            Tagged<Smi> value) {
  if (SmiValuesAre32Bits()) {
    Move(kScratchRegister, value);
    movq(dst_field_operand, kScratchRegister);
  } else {
    StoreTaggedField(dst_field_operand, Immediate(value));
  }
}

void MacroAssembler::AtomicStoreTaggedField(Operand dst_field_operand,
                                            Register value) {
  if (COMPRESS_POINTERS_BOOL) {
    movl(kScratchRegister, value);
    xchgl(kScratchRegister, dst_field_operand);
  } else {
    movq(kScratchRegister, value);
    xchgq(kScratchRegister, dst_field_operand);
  }
}

void MacroAssembler::DecompressTaggedSigned(Register destination,
                                            Operand field_operand) {
  ASM_CODE_COMMENT(this);
  movl(destination, field_operand);
}

void MacroAssembler::DecompressTagged(Register destination,
                                      Operand field_operand) {
  ASM_CODE_COMMENT(this);
  movl(destination, field_operand);
  addq(destination, kPtrComprCageBaseRegister);
}

void MacroAssembler::DecompressTagged(Register destination, Register source) {
  ASM_CODE_COMMENT(this);
  movl(destination, source);
  addq(destination, kPtrComprCageBaseRegister);
}

void MacroAssembler::DecompressTagged(Register destination,
                                      Tagged_t immediate) {
  ASM_CODE_COMMENT(this);
  leaq(destination,
       Operand(kPtrComprCageBaseRegister, static_cast<int32_t>(immediate)));
}

void MacroAssembler::DecompressProtected(Register destination,
                                         Operand field_operand) {
#if V8_ENABLE_SANDBOX
  ASM_CODE_COMMENT(this);
  movl(destination, field_operand);
  DCHECK(root_array_available_);
  orq(destination,
      Operand{kRootRegister, IsolateData::trusted_cage_base_offset()});
#else
  UNREACHABLE();
#endif  // V8_ENABLE_SANDBOX
}

void MacroAssembler::RecordWriteField(Register object, int offset,
                                      Register value, Register slot_address,
                                      SaveFPRegsMode save_fp,
                                      SmiCheck smi_check,
                                      ReadOnlyCheck ro_check,
                                      SlotDescriptor slot) {
  ASM_CODE_COMMENT(this);
  DCHECK(!AreAliased(object, value, slot_address));
  // First, check if a write barrier is even needed. The tests below
  // catch stores of Smis and read-only objects.
  Label done;

  if (ro_check == ReadOnlyCheck::kInline) {
    MaybeJumpIfReadOnlyOrSmallSmi(value, &done);
  }

  // Skip barrier if writing a smi.
  if (smi_check == SmiCheck::kInline) {
    JumpIfSmi(value, &done);
  }

  // Although the object register is tagged, the offset is relative to the start
  // of the object, so the offset must be a multiple of kTaggedSize.
  DCHECK(IsAligned(offset, kTaggedSize));

  leaq(slot_address, FieldOperand(object, offset));
  if (v8_flags.slow_debug_code) {
    ASM_CODE_COMMENT_STRING(this, "Debug check slot_address");
    Label ok;
    testb(slot_address, Immediate(kTaggedSize - 1));
    j(zero, &ok, Label::kNear);
    int3();
    bind(&ok);
  }

  RecordWrite(object, slot_address, value, save_fp, SmiCheck::kOmit,
              ReadOnlyCheck::kOmit, slot);

  bind(&done);

  // Clobber clobbered input registers when running with the debug-code flag
  // turned on to provoke errors.
  if (v8_flags.slow_debug_code) {
    ASM_CODE_COMMENT_STRING(this, "Zap scratch registers");
    Move(value, kZapValue, RelocInfo::NO_INFO);
    Move(slot_address, kZapValue, RelocInfo::NO_INFO);
  }
}

void MacroAssembler::EnterSandbox() {
#ifdef V8_ENABLE_SANDBOX_HARDWARE_SUPPORT
  pushq(rax);
  pushq(rbx);
  pushq(rcx);
  pushq(rdx);

  xorq(rcx, rcx);
  xorq(rdx, rdx);

  // TODO(350324877): it would be nicer if we could use an IsolateFieldId here.
  // However, that isn't currently possible since these routines are also used
  // in code that doesn't have the root register available. In the future, we
  // might anyway want to refactor this mechanism to instead use dedicated
  // trampoline builtins for entering/exiting sandboxed execution mode. Then we
  // could consider generating the code for that trampoline at runtime, at
  // which point we would no longer need an external reference at all.
  LoadAddress(rbx, ExternalReference::sandboxed_mode_pkey_mask_address());
  movl(rbx, Operand(rbx, 0));

  if (v8_flags.debug_code) {
    // Check that we are not in sandboxed mode.
    // Avoid calling the Abort builtin here as that would again Assert that the
    // sandboxing mode is as expected, leading to recursive aborts.
    HardAbortScope hard_abort(this);

    // If sandbox hardware support is not active, the mask will be all zeroes
    // and so this test will also pass.
    rdpkru();
    andl(rax, rbx);
    Assert(zero, AbortReason::kUnexpectedSandboxMode);
  }

  // PKEY permissions bits:
  //   00: kNoRestrictions
  //   01: kDisableAccess
  //   10: kDisableWrite
  rdpkru();
  // Set the bit from the pkey mask, thereby restricting access to the
  // out-of-sandbox pkey.
  orl(rax, rbx);
  wrpkru();

  popq(rdx);
  popq(rcx);
  popq(rbx);
  popq(rax);
#endif  // V8_ENABLE_SANDBOX_HARDWARE_SUPPORT
}

void MacroAssembler::ExitSandbox() {
#ifdef V8_ENABLE_SANDBOX_HARDWARE_SUPPORT
  pushq(rax);
  pushq(rbx);
  pushq(rcx);
  pushq(rdx);

  xorq(rcx, rcx);
  xorq(rdx, rdx);

  LoadAddress(rbx, ExternalReference::sandboxed_mode_pkey_mask_address());
  movl(rbx, Operand(rbx, 0));

  if (v8_flags.debug_code) {
    // Check that we are in sandboxed mode.
    // Avoid calling the Abort builtin here as that would again Assert that the
    // sandboxing mode is as expected, leading to recursive aborts.
    HardAbortScope hard_abort(this);

    // If sandbox hardware support is not active, the mask will be all zeroes
    // and so we need to handle this here.
    Label hardware_support_not_active;
    testl(rbx, rbx);
    j(zero, &hardware_support_not_active);
    rdpkru();
    andl(rax, rbx);
    Assert(not_zero, AbortReason::kUnexpectedSandboxMode);
    bind(&hardware_support_not_active);
  }

  // PKEY permissions bits:
  //   00: kNoRestrictions
  //   01: kDisableAccess
  //   10: kDisableWrite
  rdpkru();
  // Clear the bit from the pkey mask, thereby restoring full access to the
  // out-of-sandbox pkey.
  notl(rbx);
  andl(rax, rbx);
  wrpkru();

  popq(rdx);
  popq(rcx);
  popq(rbx);
  popq(rax);
#endif  // V8_ENABLE_SANDBOX_HARDWARE_SUPPORT
}

void MacroAssembler::AssertInSandboxedExecutionMode() {
#ifdef V8_ENABLE_SANDBOX_HARDWARE_SUPPORT
  if (v8_flags.debug_code) {
    // Avoid calling the Abort builtin here as that would again Assert that the
    // sandboxing mode is as expected, leading to recursive aborts.
    HardAbortScope hard_abort(this);

    pushq(rax);
    pushq(rbx);
    pushq(rcx);
    pushq(rdx);

    xorq(rcx, rcx);
    xorq(rdx, rdx);

    LoadAddress(rbx, ExternalReference::sandboxed_mode_pkey_mask_address());
    movl(rbx, Operand(rbx, 0));

    // If sandbox hardware support is not active, the mask will be all zeroes
    // and so we need to handle this here.
    Label hardware_support_not_active;
    testl(rbx, rbx);
    j(zero, &hardware_support_not_active);

    rdpkru();
    andl(rax, rbx);
    Assert(not_zero, AbortReason::kUnexpectedSandboxMode);

    bind(&hardware_support_not_active);

    popq(rdx);
    popq(rcx);
    popq(rbx);
    popq(rax);
  }
#endif  // V8_ENABLE_SANDBOX_HARDWARE_SUPPORT
}

void MacroAssembler::SwitchSandboxingModeTo(CodeSandboxingMode mode) {
  switch (mode) {
    case CodeSandboxingMode::kSandboxed:
      return EnterSandbox();
    case CodeSandboxingMode::kUnsandboxed:
      return ExitSandbox();
  }
}

CodeSandboxingMode MacroAssembler::SwitchSandboxingModeBeforeCallIfNeeded(
    CodeSandboxingMode target_sandboxing_mode) {
  CodeSandboxingMode previous_sandboxing_mode = sandboxing_mode();
  if (sandboxing_mode() != target_sandboxing_mode) {
    SwitchSandboxingModeTo(target_sandboxing_mode);
    sandboxing_mode_ = target_sandboxing_mode;
  }
  return previous_sandboxing_mode;
}

void MacroAssembler::SwitchSandboxingModeAfterCallIfNeeded(
    CodeSandboxingMode previous_sandboxing_mode) {
  if (sandboxing_mode() != previous_sandboxing_mode) {
    SwitchSandboxingModeTo(previous_sandboxing_mode);
    sandboxing_mode_ = previous_sandboxing_mode;
  }
}

void MacroAssembler::EncodeSandboxedPointer(Register value) {
  ASM_CODE_COMMENT(this);
#ifdef V8_ENABLE_SANDBOX
  subq(value, kPtrComprCageBaseRegister);
  shlq(value, Immediate(kSandboxedPointerShift));
#else
  UNREACHABLE();
#endif
}

void MacroAssembler::DecodeSandboxedPointer(Register value) {
  ASM_CODE_COMMENT(this);
#ifdef V8_ENABLE_SANDBOX
  shrq(value, Immediate(kSandboxedPointerShift));
  addq(value, kPtrComprCageBaseRegister);
#else
  UNREACHABLE();
#endif
}

void MacroAssembler::LoadSandboxedPointerField(Register destination,
                                               Operand field_operand) {
  ASM_CODE_COMMENT(this);
  movq(destination, field_operand);
  DecodeSandboxedPointer(destination);
}

void MacroAssembler::StoreSandboxedPointerField(Operand dst_field_operand,
                                                Register value) {
  ASM_CODE_COMMENT(this);
  DCHECK(!AreAliased(value, kScratchRegister));
  DCHECK(!dst_field_operand.AddressUsesRegister(kScratchRegister));
  movq(kScratchRegister, value);
  EncodeSandboxedPointer(kScratchRegister);
  movq(dst_field_operand, kScratchRegister);
}

void MacroAssembler::LoadExternalPointerField(
    Register destination, Operand field_operand,
    ExternalPointerTagRange tag_range, Register scratch,
    IsolateRootLocation isolateRootLocation) {
  DCHECK(!AreAliased(destination, scratch));
#ifdef V8_ENABLE_SANDBOX
  DCHECK(!tag_range.IsEmpty());
  DCHECK(!IsSharedExternalPointerType(tag_range));
  DCHECK(!field_operand.AddressUsesRegister(scratch));
  if (isolateRootLocation == IsolateRootLocation::kInRootRegister) {
    DCHECK(root_array_available_);
    // TODO(saelo): consider using an ExternalReference here.
    movq(scratch,
         Operand(kRootRegister,
                 IsolateData::external_pointer_table_offset() +
                     Internals::kExternalEntityTableBasePointerOffset));
  } else {
    DCHECK(isolateRootLocation == IsolateRootLocation::kInScratchRegister);
    movq(
        scratch,
        Operand(scratch, IsolateData::external_pointer_table_offset() +
                             Internals::kExternalEntityTableBasePointerOffset));
  }
  movl(destination, field_operand);
  shrq(destination, Immediate(kExternalPointerIndexShift));
  static_assert(kExternalPointerTableEntrySize == 8);
  movq(destination, Operand(scratch, destination, times_8, 0));

  // We don't expect to see empty fields here. If this is ever needed, consider
  // using an dedicated empty value entry for those tags instead (i.e. an entry
  // with the right tag and nullptr payload).
  // Although interceptor callbacks can be empty in general, once we decide
  // to generate a code loading a callback value it's guaranteed that the
  // external pointer handle is not empty.
  DCHECK(!ExternalPointerCanBeEmpty(tag_range) ||
         kAnyInterceptorInfoExternalPointerTagRange.Contains(tag_range));

  if (tag_range.Size() == 1) {
    // The common and simple case: we expect exactly one tag.
    movq(scratch, destination);
    shrq(scratch, Immediate(kExternalPointerTagShift));
    andl(scratch, Immediate(kExternalPointerShiftedTagMask));
    cmpl(scratch, Immediate(tag_range.first));
    SbxCheck(equal, AbortReason::kExternalPointerTagMismatch);
    movq(scratch, Immediate64(kExternalPointerPayloadMask));
    andq(destination, scratch);
  } else {
    // Not currently supported. Implement once needed.
    DCHECK_NE(tag_range, kAnyExternalPointerTagRange);
    UNREACHABLE();
  }
#else
  movq(destination, field_operand);
#endif  // V8_ENABLE_SANDBOX
}

void MacroAssembler::LoadTrustedPointerField(Register destination,
                                             Operand field_operand,
                                             IndirectPointerTagRange tag_range,
                                             Register scratch) {
#ifdef V8_ENABLE_SANDBOX
  LoadIndirectPointerField(destination, field_operand, tag_range, scratch);
#else
  LoadTaggedField(destination, field_operand);
#endif  // V8_ENABLE_SANDBOX
}

void MacroAssembler::LoadTrustedUnknownPointerField(
    Register destination, Operand field_operand, Register scratch,
    const std::initializer_list<
        std::tuple<InstanceType, Label*, Label::Distance>>& cases) {
  DCHECK(!AreAliased(destination, scratch));
  Label done;

#ifdef V8_ENABLE_SANDBOX
  {
    Register handle = scratch;
    movl(handle, field_operand);

    bool handles_code_case = false;
    for (auto& [type, label, distance] : cases) {
      if (type == CODE_TYPE) {
        handles_code_case = true;

        Label not_code_handle;
        testl(handle, Immediate(kCodePointerHandleMarker));
        j(zero, &not_code_handle, Label::kNear);

        ResolveCodePointerHandle(destination, handle);
        jmp(label, distance);

        bind(&not_code_handle);
        break;
      }
    }
    if (!handles_code_case) {
      testl(handle, Immediate(kCodePointerHandleMarker));
      j(not_zero, &done, Label::kNear);
    }

    ResolveTrustedPointerHandle(destination, handle, kAllTrustedPointerTags);
  }
#else
  LoadTaggedField(destination, field_operand);
#endif  // V8_ENABLE_SANDBOX

#if V8_STATIC_ROOTS_BOOL
  LoadCompressedMap(scratch, destination);
  for (auto& [type, label, distance] : cases) {
    if (V8_ENABLE_SANDBOX_BOOL && type == CODE_TYPE) {
      continue;
    }
    CompareInstanceTypeWithUniqueCompressedMap(scratch, type);
    j(equal, label, distance);
  }
#else
  LoadMap(scratch, destination);
  for (auto& [type, label, distance] : cases) {
    if (V8_ENABLE_SANDBOX_BOOL && type == CODE_TYPE) {
      continue;
    }
    CmpInstanceType(scratch, type);
    j(equal, label, distance);
  }
#endif  // V8_STATIC_ROOTS_BOOL

  bind(&done);
  xorq(destination, destination);
}

void MacroAssembler::StoreTrustedPointerField(Operand dst_field_operand,
                                              Register value) {
#ifdef V8_ENABLE_SANDBOX
  StoreIndirectPointerField(dst_field_operand, value);
#else
  StoreTaggedField(dst_field_operand, value);
#endif  // V8_ENABLE_SANDBOX
}

void MacroAssembler::LoadIndirectPointerField(Register destination,
                                              Operand field_operand,
                                              IndirectPointerTagRange tag_range,
                                              Register scratch) {
#ifdef V8_ENABLE_SANDBOX
  DCHECK(!AreAliased(destination, scratch));
  Register handle = scratch;
  movl(handle, field_operand);
  ResolveIndirectPointerHandle(destination, handle, tag_range);
#else
  UNREACHABLE();
#endif  // V8_ENABLE_SANDBOX
}

void MacroAssembler::StoreIndirectPointerField(Operand dst_field_operand,
                                               Register value) {
#ifdef V8_ENABLE_SANDBOX
  movl(kScratchRegister,
       FieldOperand(value, ExposedTrustedObject::kSelfIndirectPointerOffset));
  movl(dst_field_operand, kScratchRegister);
#else
  UNREACHABLE();
#endif  // V8_ENABLE_SANDBOX
}

#ifdef V8_ENABLE_SANDBOX
void MacroAssembler::ResolveIndirectPointerHandle(
    Register destination, Register handle, IndirectPointerTagRange tag_range) {
  // This function must not be used to resolve kAllIndirectPointerTags. Use
  // LoadTrustedUnknownPointerField for that instead.
  CHECK_NE(tag_range, kAllIndirectPointerTags);

  // The tag implies which pointer table to use.
  if (tag_range == kCodeIndirectPointerTag) {
    ResolveCodePointerHandle(destination, handle);
  } else {
    DCHECK(!tag_range.Contains(kCodeIndirectPointerTag));
    ResolveTrustedPointerHandle(destination, handle, tag_range);
  }
}

void MacroAssembler::ResolveTrustedPointerHandle(
    Register destination, Register handle, IndirectPointerTagRange tag_range) {
  DCHECK(!tag_range.Contains(kCodeIndirectPointerTag));
  DCHECK(!AreAliased(handle, destination));
  shrl(handle, Immediate(kTrustedPointerHandleShift));
  static_assert(kTrustedPointerTableEntrySize == 8);
  DCHECK(root_array_available_);
  movq(destination,
       Operand{kRootRegister, IsolateData::trusted_pointer_table_offset()});
  movq(destination, Operand{destination, handle, times_8, 0});

  if (IsFastIndirectPointerTagRange(tag_range)) {
    uint64_t mask = ComputeUntaggingMaskForFastIndirectPointerTag(tag_range);
    movq(kScratchRegister, Immediate64(mask));
    andq(destination, kScratchRegister);
  } else {
    Register scratch = kScratchRegister;
    movq(scratch, destination);
    shrq(scratch, Immediate(kTrustedPointerTableTagShift));

    Label done;
    if (tag_range.Size() == 1) {
      cmpl(scratch, Immediate(tag_range.first));
      j(equal, &done, Label::kNear);
    } else {
      subl(scratch, Immediate(tag_range.first));
      cmpl(scratch, Immediate(tag_range.last - tag_range.first));
      j(below_equal, &done, Label::kNear);
    }
    xorq(destination, destination);
    bind(&done);

    shlq(destination, Immediate(16));
    shrq(destination, Immediate(16));
  }
}

void MacroAssembler::ResolveCodePointerHandle(Register destination,
                                              Register handle) {
  DCHECK(!AreAliased(handle, destination));
  Register table = destination;
  LoadCodePointerTableBase(table);
  shrl(handle, Immediate(kCodePointerHandleShift));
  // The code pointer table entry size is 16 bytes, so we have to do an
  // explicit shift first (times_16 doesn't exist).
  shll(handle, Immediate(kCodePointerTableEntrySizeLog2));
  movq(destination,
       Operand(table, handle, times_1, kCodePointerTableEntryCodeObjectOffset));
  // The LSB is used as marking bit by the code pointer table, so here we have
  // to set it using a bitwise OR as it may or may not be set.
  orq(destination, Immediate(kHeapObjectTag));
}

void MacroAssembler::LoadCodeEntrypointViaCodePointer(Register destination,
                                                      Operand field_operand,
                                                      CodeEntrypointTag tag) {
  DCHECK(!AreAliased(destination, kScratchRegister));
  DCHECK(!field_operand.AddressUsesRegister(kScratchRegister));
  DCHECK_NE(tag, kInvalidEntrypointTag);
  LoadCodePointerTableBase(kScratchRegister);
  movl(destination, field_operand);
  shrl(destination, Immediate(kCodePointerHandleShift));
  shll(destination, Immediate(kCodePointerTableEntrySizeLog2));
  movq(destination, Operand(kScratchRegister, destination, times_1, 0));
  if (tag != 0) {
    // Can this be improved?
    movq(kScratchRegister, Immediate64(tag));
    xorq(destination, kScratchRegister);
  }
}

void MacroAssembler::LoadCodePointerTableBase(Register destination) {
#ifdef V8_COMPRESS_POINTERS_IN_MULTIPLE_CAGES
  if (!options().isolate_independent_code && isolate()) {
    // Embed the code pointer table address into the code.
    LoadAddress(destination,
                ExternalReference::code_pointer_table_base_address(isolate()));
  } else {
    // Force indirect load via root register as a workaround for
    // isolate-independent code (for example, for Wasm).
    Load(destination,
         ExternalReference::address_of_code_pointer_table_base_address());
  }
#else
  // Embed the code pointer table address into the code.
  LoadAddress(destination,
              ExternalReference::global_code_pointer_table_base_address());
#endif  // V8_COMPRESS_POINTERS_IN_MULTIPLE_CAGES
}
#endif  // V8_ENABLE_SANDBOX

void MacroAssembler::LoadEntrypointFromJSDispatchTable(
    Register destination, Register dispatch_handle) {
  DCHECK(!AreAliased(destination, dispatch_handle, kScratchRegister));
  CHECK(root_array_available());
  movq(kScratchRegister,
       ExternalReferenceAsOperand(IsolateFieldId::kJSDispatchTable));
  movq(destination, dispatch_handle);
  shrl(destination, Immediate(kJSDispatchHandleShift));
  shll(destination, Immediate(kJSDispatchTableEntrySizeLog2));
  movq(destination, Operand(kScratchRegister, destination, times_1,
                            JSDispatchEntry::kEntrypointOffset));
}

void MacroAssembler::LoadParameterCountFromJSDispatchTable(
    Register destination, Register dispatch_handle) {
  DCHECK(!AreAliased(destination, dispatch_handle, kScratchRegister));
  CHECK(root_array_available());
  movq(kScratchRegister,
       ExternalReferenceAsOperand(IsolateFieldId::kJSDispatchTable));
  movq(destination, dispatch_handle);
  shrl(destination, Immediate(kJSDispatchHandleShift));
  shll(destination, Immediate(kJSDispatchTableEntrySizeLog2));
  static_assert(JSDispatchEntry::kParameterCountMask == 0xffff);
  movzxwq(destination, Operand(kScratchRegister, destination, times_1,
                               JSDispatchEntry::kCodeObjectOffset));
}

void MacroAssembler::LoadEntrypointAndParameterCountFromJSDispatchTable(
    Register entrypoint, Register parameter_count, Register dispatch_handle) {
  DCHECK(!AreAliased(entrypoint, parameter_count, dispatch_handle,
                     kScratchRegister));
  CHECK(root_array_available());
  movq(kScratchRegister,
       ExternalReferenceAsOperand(IsolateFieldId::kJSDispatchTable));
  Register offset = parameter_count;
  movq(offset, dispatch_handle);
  shrl(offset, Immediate(kJSDispatchHandleShift));
  shll(offset, Immediate(kJSDispatchTableEntrySizeLog2));
  movq(entrypoint, Operand(kScratchRegister, offset, times_1,
                           JSDispatchEntry::kEntrypointOffset));
  static_assert(JSDispatchEntry::kParameterCountMask == 0xffff);
  movzxwq(parameter_count, Operand(kScratchRegister, offset, times_1,
                                   JSDispatchEntry::kCodeObjectOffset));
}

void MacroAssembler::LoadProtectedPointerField(Register destination,
                                               Operand field_operand) {
  DCHECK(root_array_available());
#ifdef V8_ENABLE_SANDBOX
  DecompressProtected(destination, field_operand);
#else
  LoadTaggedField(destination, field_operand);
#endif
}

void MacroAssembler::CallEphemeronKeyBarrier(Register object,
                                             Register slot_address,
                                             SaveFPRegsMode fp_mode) {
  ASM_CODE_COMMENT(this);
  DCHECK(!AreAliased(object, slot_address));
  RegList registers =
      WriteBarrierDescriptor::ComputeSavedRegisters(object, slot_address);
  PushAll(registers);

  Register object_parameter = WriteBarrierDescriptor::ObjectRegister();
  Register slot_address_parameter =
      WriteBarrierDescriptor::SlotAddressRegister();
  MovePair(slot_address_parameter, slot_address, object_parameter, object);

  CallBuiltin(Builtins::EphemeronKeyBarrier(fp_mode));
  PopAll(registers);
}

void MacroAssembler::CallIndirectPointerBarrier(Register object,
                                                Register slot_address,
                                                SaveFPRegsMode fp_mode,
                                                IndirectPointerTag tag) {
  ASM_CODE_COMMENT(this);
  DCHECK(!AreAliased(object, slot_address));
  // TODO(saelo) if necessary, we could introduce a "SaveRegisters version of
  // this function and make this code not save clobbered registers. It's
  // probably not currently worth the effort though since stores to indirect
  // pointer fields are fairly rare.
  RegList registers =
      IndirectPointerWriteBarrierDescriptor::ComputeSavedRegisters(
          object, slot_address);
  PushAll(registers);

  Register object_parameter =
      IndirectPointerWriteBarrierDescriptor::ObjectRegister();
  Register slot_address_parameter =
      IndirectPointerWriteBarrierDescriptor::SlotAddressRegister();
  MovePair(slot_address_parameter, slot_address, object_parameter, object);

  Register tag_parameter =
      IndirectPointerWriteBarrierDescriptor::IndirectPointerTagRegister();
  Move(tag_parameter, tag);

  CallBuiltin(Builtins::IndirectPointerBarrier(fp_mode));
  PopAll(registers);
}

void MacroAssembler::CallRecordWriteStubSaveRegisters(Register object,
                                                      Register slot_address,
                                                      SaveFPRegsMode fp_mode,
                                                      StubCallMode mode) {
  ASM_CODE_COMMENT(this);
  DCHECK(!AreAliased(object, slot_address));
  RegList registers =
      WriteBarrierDescriptor::ComputeSavedRegisters(object, slot_address);
  PushAll(registers);
  Register object_parameter = WriteBarrierDescriptor::ObjectRegister();
  Register slot_address_parameter =
      WriteBarrierDescriptor::SlotAddressRegister();
  MovePair(object_parameter, object, slot_address_parameter, slot_address);

  CallRecordWriteStub(object_parameter, slot_address_parameter, fp_mode, mode);
  PopAll(registers);
}

void MacroAssembler::CallRecordWriteStub(Register object, Register slot_address,
                                         SaveFPRegsMode fp_mode,
                                         StubCallMode mode) {
  ASM_CODE_COMMENT(this);
  // Use CallRecordWriteStubSaveRegisters if the object and slot registers
  // need to be caller saved.
  DCHECK_EQ(WriteBarrierDescriptor::ObjectRegister(), object);
  DCHECK_EQ(WriteBarrierDescriptor::SlotAddressRegister(), slot_address);
#if V8_ENABLE_WEBASSEMBLY
  if (mode == StubCallMode::kCallWasmRuntimeStub) {
    // Use {near_call} for direct Wasm call within a module.
    Builtin wasm_target = wasm::WasmCode::GetRecordWriteBuiltin(fp_mode);
    // TODO(429142815): replace with a DCHECK_EQ(sandboxing_mode(),
    // target_sandboxing_mode()) once write barrier builtins can run in
    // sandboxed execution mode.
    CodeSandboxingMode previous_mode = SwitchSandboxingModeBeforeCallIfNeeded(
        Builtins::SandboxingModeOf(wasm_target));
    near_call(static_cast<intptr_t>(wasm_target), RelocInfo::WASM_STUB_CALL);
    SwitchSandboxingModeAfterCallIfNeeded(previous_mode);
#else
  if (false) {
#endif
  } else {
    CallBuiltin(Builtins::RecordWrite(fp_mode));
  }
}

void MacroAssembler::CallVerifySkippedWriteBarrierStubSaveRegisters(
    Register object, Register value, SaveFPRegsMode fp_mode) {
  ASM_CODE_COMMENT(this);
  PushCallerSaved(fp_mode);
  CallVerifySkippedWriteBarrierStub(object, value);
  PopCallerSaved(fp_mode);
}

void MacroAssembler::CallVerifySkippedWriteBarrierStub(Register object,
                                                       Register value) {
  ASM_CODE_COMMENT(this);
  MovePair(kCArgRegs[0], object, kCArgRegs[1], value);
  PrepareCallCFunction(2);
  CallCFunction(ExternalReference::verify_skipped_write_barrier(), 2);
}

void MacroAssembler::CallVerifySkippedIndirectWriteBarrierStubSaveRegisters(
    Register object, Register value, SaveFPRegsMode fp_mode) {
  ASM_CODE_COMMENT(this);
  PushCallerSaved(fp_mode);
  CallVerifySkippedIndirectWriteBarrierStub(object, value);
  PopCallerSaved(fp_mode);
}

void MacroAssembler::CallVerifySkippedIndirectWriteBarrierStub(Register object,
                                                               Register value) {
  ASM_CODE_COMMENT(this);
  MovePair(kCArgRegs[0], object, kCArgRegs[1], value);
  PrepareCallCFunction(2);
  CallCFunction(ExternalReference::verify_skipped_indirect_write_barrier(), 2);
}

#ifdef V8_IS_TSAN
void MacroAssembler::CallTSANStoreStub(Register address, Register value,
                                       SaveFPRegsMode fp_mode, int size,
                                       StubCallMode mode,
                                       std::memory_order order) {
  ASM_CODE_COMMENT(this);
  DCHECK(!AreAliased(address, value));
  TSANStoreDescriptor descriptor;
  RegList registers = descriptor.allocatable_registers();

  PushAll(registers);

  Register address_parameter(
      descriptor.GetRegisterParameter(TSANStoreDescriptor::kAddress));
  Register value_parameter(
      descriptor.GetRegisterParameter(TSANStoreDescriptor::kValue));

  // Prepare argument registers for calling GetTSANStoreStub.
  MovePair(address_parameter, address, value_parameter, value);

#if V8_ENABLE_WEBASSEMBLY
  if (mode != StubCallMode::kCallWasmRuntimeStub) {
    // JS functions and Wasm wrappers.
    CallBuiltin(CodeFactory::GetTSANStoreStub(fp_mode, size, order));
  } else {
    // Wasm functions should call builtins through their far jump table.
    auto wasm_target = static_cast<intptr_t>(
        wasm::WasmCode::GetTSANStoreBuiltin(fp_mode, size, order));
    near_call(wasm_target, RelocInfo::WASM_STUB_CALL);
  }
#else
  CallBuiltin(CodeFactory::GetTSANStoreStub(fp_mode, size, order));
#endif  // V8_ENABLE_WEBASSEMBLY

  PopAll(registers);
}

void MacroAssembler::CallTSANRelaxedLoadStub(Register address,
                                             SaveFPRegsMode fp_mode, int size,
                                             StubCallMode mode) {
  TSANLoadDescriptor descriptor;
  RegList registers = descriptor.allocatable_registers();

  PushAll(registers);

  Register address_parameter(
      descriptor.GetRegisterParameter(TSANLoadDescriptor::kAddress));

  // Prepare argument registers for calling TSANRelaxedLoad.
  Move(address_parameter, address);

#if V8_ENABLE_WEBASSEMBLY
  if (mode != StubCallMode::kCallWasmRuntimeStub) {
    // JS functions and Wasm wrappers.
    CallBuiltin(CodeFactory::GetTSANRelaxedLoadStub(fp_mode, size));
  } else {
    // Wasm functions should call builtins through their far jump table.
    auto wasm_target = static_cast<intptr_t>(
        wasm::WasmCode::GetTSANRelaxedLoadBuiltin(fp_mode, size));
    near_call(wasm_target, RelocInfo::WASM_STUB_CALL);
  }
#else
  CallBuiltin(CodeFactory::GetTSANRelaxedLoadStub(fp_mode, size));
#endif  // V8_ENABLE_WEBASSEMBLY

  PopAll(registers);
}
#endif  // V8_IS_TSAN

void MacroAssembler::MaybeJumpIfReadOnlyOrSmallSmi(Register value,
                                                   Label* dest) {
#if V8_STATIC_ROOTS_BOOL && CONTIGUOUS_COMPRESSED_READ_ONLY_SPACE_BOOL
  // Quick check for Read-only and small Smi values.
  // This optimization requires contiguous compressed RO space to ensure RO
  // space is at the beginning of the cage; otherwise, objects from other spaces
  // could alias with low addresses.
  constexpr int kLastStaticRootPage =
      RoundUp<kRegularPageSize>(StaticReadOnlyRoot::kLastAllocatedRoot);
  static_assert(kLastStaticRootPage <= kContiguousReadOnlyReservationSize);
  JumpIfUnsignedLessThan(value, kContiguousReadOnlyReservationSize, dest);
#endif  // V8_STATIC_ROOTS_BOOL && CONTIGUOUS_COMPRESSED_READ_ONLY_SPACE_BOOL
}

void MacroAssembler::RecordWrite(Register object, Register slot_address,
                                 Register value, SaveFPRegsMode fp_mode,
                                 SmiCheck smi_check, ReadOnlyCheck ro_check,
                                 SlotDescriptor slot) {
  ASM_CODE_COMMENT(this);
  DCHECK(!AreAliased(object, slot_address, value));
  AssertNotSmi(object);

  if (v8_flags.disable_write_barriers) {
    return;
  }

  if (v8_flags.slow_debug_code) {
    ASM_CODE_COMMENT_STRING(this, "Debug check slot_address");
    Label ok;
    if (slot.contains_indirect_pointer()) {
      Push(object);  // Use object register as scratch
      Register scratch = object;
      Push(slot_address);  // Use slot address register to load the value into
      Register value_in_slot = slot_address;
      LoadIndirectPointerField(value_in_slot, Operand(slot_address, 0),
                               slot.indirect_pointer_tag(), scratch);
      cmp_tagged(value, value_in_slot);
      // These pops don't affect the flag registers, so we can do them before
      // the conditional jump below.
      Pop(slot_address);
      Pop(object);
    } else {
      cmp_tagged(value, Operand(slot_address, 0));
    }
    j(equal, &ok, Label::kNear);
    int3();
    bind(&ok);
  }

  // First, check if a write barrier is even needed. The tests below
  // catch stores of smis and read-only objects, as well as stores into the
  // young generation.
  Label done;

  if (ro_check == ReadOnlyCheck::kInline) {
    MaybeJumpIfReadOnlyOrSmallSmi(value, &done);
  }

  if (smi_check == SmiCheck::kInline) {
    // Skip barrier if writing a smi.
    JumpIfSmi(value, &done);
  }

  if (slot.contains_indirect_pointer()) {
    // The indirect pointer write barrier is only enabled during marking.
    JumpIfNotMarking(&done);
  } else {
#if V8_ENABLE_STICKY_MARK_BITS_BOOL
    DCHECK(!AreAliased(kScratchRegister, object, slot_address, value));
    Label stub_call;

    JumpIfMarking(&stub_call);

    // Save the slot_address in the xmm scratch register.
    movq(kScratchDoubleReg, slot_address);
    Register scratch0 = slot_address;
    CheckMarkBit(object, kScratchRegister, scratch0, carry, &done);
#if CONTIGUOUS_COMPRESSED_READ_ONLY_SPACE_BOOL
    JumpIfUnsignedLessThan(value, kContiguousReadOnlyReservationSize, &done);
#else   // !CONTIGUOUS_COMPRESSED_READ_ONLY_SPACE_BOOL
    CheckPageFlag(value, kScratchRegister, MemoryChunk::kIsInReadOnlyHeapMask,
                  not_zero, &done, Label::kFar);
#endif  // !CONTIGUOUS_COMPRESSED_READ_ONLY_SPACE_BOOL
    CheckMarkBit(value, kScratchRegister, scratch0, carry, &done);
    movq(slot_address, kScratchDoubleReg);
    bind(&stub_call);
#else   // !V8_ENABLE_STICKY_MARK_BITS_BOOL
    CheckPageFlag(value,
                  value,  // Used as scratch.
                  MemoryChunk::kPointersToHereAreInterestingMask, zero, &done,
                  Label::kFar);

    CheckPageFlag(object,
                  value,  // Used as scratch.
                  MemoryChunk::kPointersFromHereAreInterestingMask, zero, &done,
                  Label::kFar);
#endif  // !V8_ENABLE_STICKY_MARK_BITS_BOOL
  }

  if (slot.contains_direct_pointer()) {
    CallRecordWriteStub(object, slot_address, fp_mode,
                        StubCallMode::kCallBuiltinPointer);
  } else {
    DCHECK(slot.contains_indirect_pointer());
    CallIndirectPointerBarrier(object, slot_address, fp_mode,
                               slot.indirect_pointer_tag());
  }

  bind(&done);

  // Clobber clobbered registers when running with the debug-code flag
  // turned on to provoke errors.
  if (v8_flags.slow_debug_code) {
    ASM_CODE_COMMENT_STRING(this, "Zap scratch registers");
    Move(slot_address, kZapValue, RelocInfo::NO_INFO);
    Move(value, kZapValue, RelocInfo::NO_INFO);
  }
}

void MacroAssembler::Check(Condition cc, AbortReason reason) {
  Label L;
  j(cc, &L, Label::kNear);
  Abort(reason);
  // Control will not return here.
  bind(&L);
}

void MacroAssembler::SbxCheck(Condition cc, AbortReason reason) {
  Check(cc, reason);
}

void MacroAssembler::CheckStackAlignment() {
  int frame_alignment = base::OS::ActivationFrameAlignment();
  int frame_alignment_mask = frame_alignment - 1;
  if (frame_alignment > kSystemPointerSize) {
    ASM_CODE_COMMENT(this);
    DCHECK(base::bits::IsPowerOfTwo(frame_alignment));
    Label alignment_as_expected;
    testq(rsp, Immediate(frame_alignment_mask));
    j(zero, &alignment_as_expected, Label::kNear);
    // Abort if stack is not aligned.
    int3();
    bind(&alignment_as_expected);
  }
}

void MacroAssembler::AlignStackPointer() {
  const int kFrameAlignment = base::OS::ActivationFrameAlignment();
  if (kFrameAlignment > 0) {
    DCHECK(base::bits::IsPowerOfTwo(kFrameAlignment));
    DCHECK(is_int8(kFrameAlignment));
    andq(rsp, Immediate(-kFrameAlignment));
  }
}

void MacroAssembler::Abort(AbortReason reason) {
  ASM_CODE_COMMENT(this);
  if (v8_flags.code_comments) {
    RecordComment("Abort message:", SourceLocation{});
    RecordComment(GetAbortReason(reason), SourceLocation{});
  }

  // Without debug code, save the code size and just trap.
  if (!v8_flags.debug_code || v8_flags.trap_on_abort) {
    int3();
    return;
  }

  if (should_abort_hard()) {
    // We don't care if we constructed a frame. Just pretend we did.
    FrameScope assume_frame(this, StackFrame::NO_FRAME_TYPE);
    Move(kCArgRegs[0], static_cast<int>(reason));
    PrepareCallCFunction(1);
    LoadAddress(rax, ExternalReference::abort_with_reason());
    call(rax);
    return;
  }

  Move(rdx, Smi::FromInt(static_cast<int>(reason)));

  {
    // We don't actually want to generate a pile of code for this, so just
    // claim there is a stack frame, without generating one.
    FrameScope scope(this, StackFrame::NO_FRAME_TYPE);
    if (root_array_available()) {
      // Generate an indirect call via builtins entry table here in order to
      // ensure that the interpreter_entry_return_pc_offset is the same for
      // InterpreterEntryTrampoline and InterpreterEntryTrampolineForProfiling
      // when v8_flags.debug_code is enabled.
      Call(EntryFromBuiltinAsOperand(Builtin::kAbort));
    } else {
      CallBuiltin(Builtin::kAbort);
    }
  }

  // Control will not return here.
  int3();
}

void MacroAssembler::CallRuntime(const Runtime::Function* f,
                                 int num_arguments) {
  ASM_CODE_COMMENT(this);
  // If the expected number of arguments of the runtime function is
  // constant, we check that the actual number of arguments match the
  // expectation.
  CHECK(f->nargs < 0 || f->nargs == num_arguments);

  // TODO(1236192): Most runtime routines don't need the number of
  // arguments passed in because it is constant. At some point we
  // should remove this need and make the runtime routine entry code
  // smarter.
  Move(rax, num_arguments);
  LoadAddress(rbx, ExternalReference::Create(f));

  bool switch_to_central = options().is_wasm;
  CallBuiltin(Builtins::RuntimeCEntry(f->result_size, switch_to_central));
}

void MacroAssembler::TailCallRuntime(Runtime::FunctionId fid) {
  // ----------- S t a t e -------------
  //  -- rsp[0]                 : return address
  //  -- rsp[8]                 : argument num_arguments - 1
  //  ...
  //  -- rsp[8 * num_arguments] : argument 0 (receiver)
  //
  //  For runtime functions with variable arguments:
  //  -- rax                    : number of  arguments
  // -----------------------------------
  ASM_CODE_COMMENT(this);
  const Runtime::Function* function = Runtime::FunctionForId(fid);
  DCHECK_EQ(1, function->result_size);
  if (function->nargs >= 0) {
    Move(rax, function->nargs);
  }
  JumpToExternalReference(ExternalReference::Create(fid));
}

void MacroAssembler::JumpToExternalReference(const ExternalReference& ext,
                                             bool builtin_exit_frame) {
  ASM_CODE_COMMENT(this);
  // Set the entry point and jump to the C entry runtime stub.
  LoadAddress(rbx, ext);
  TailCallBuiltin(Builtins::CEntry(1, ArgvMode::kStack, builtin_exit_frame));
}

#ifdef V8_ENABLE_DEBUG_CODE
void MacroAssembler::AssertFeedbackCell(Register object, Register scratch) {
  if (v8_flags.debug_code) {
    IsObjectType(object, FEEDBACK_CELL_TYPE, scratch);
    Assert(equal, AbortReason::kExpectedFeedbackCell);
  }
}
void MacroAssembler::AssertFeedbackVector(Register object, Register scratch) {
  if (v8_flags.debug_code) {
    IsObjectType(object, FEEDBACK_VECTOR_TYPE, scratch);
    Assert(equal, AbortReason::kExpectedFeedbackVector);
  }
}
#endif  // V8_ENABLE_DEBUG_CODE

void MacroAssembler::GenerateTailCallToReturnedCode(
    Runtime::FunctionId function_id, JumpMode jump_mode) {
  // ----------- S t a t e -------------
  //  -- rax : actual argument count (preserved for callee)
  //  -- rdx : new target (preserved for callee)
  //  -- rdi : target function (preserved for callee)
  //  -- r15 : dispatch handle (preserved for callee)
  // -----------------------------------
  ASM_CODE_COMMENT(this);
  {
    FrameScope scope(this, StackFrame::INTERNAL);
    // Push a copy of the target function, the new target, the actual argument
    // count, and the dispatch handle.
    Push(kJavaScriptCallTargetRegister);
    Push(kJavaScriptCallNewTargetRegister);
    SmiTag(kJavaScriptCallArgCountRegister);
    Push(kJavaScriptCallArgCountRegister);
#ifdef V8_JS_LINKAGE_INCLUDES_DISPATCH_HANDLE
    // No need to SmiTag since dispatch handles always look like Smis.
    static_assert(kJSDispatchHandleShift > 0);
    AssertSmi(kJavaScriptCallDispatchHandleRegister);
    Push(kJavaScriptCallDispatchHandleRegister);
#endif
    // Function is also the parameter to the runtime call.
    Push(kJavaScriptCallTargetRegister);

    CallRuntime(function_id, 1);

    // Restore target function, new target, actual argument count, and dispatch
    // handle.
#ifdef V8_JS_LINKAGE_INCLUDES_DISPATCH_HANDLE
    Pop(kJavaScriptCallDispatchHandleRegister);
#endif
    Pop(kJavaScriptCallArgCountRegister);
    SmiUntagUnsigned(kJavaScriptCallArgCountRegister);
    Pop(kJavaScriptCallNewTargetRegister);
    Pop(kJavaScriptCallTargetRegister);
  }

  static_assert(kJavaScriptCallCodeStartRegister == rcx, "ABI mismatch");
#ifndef V8_JS_LINKAGE_INCLUDES_DISPATCH_HANDLE
  movl(kJavaScriptCallDispatchHandleRegister,
       FieldOperand(kJavaScriptCallTargetRegister,
                    JSFunction::kDispatchHandleOffset));
#endif
  LoadEntrypointFromJSDispatchTable(rcx, kJavaScriptCallDispatchHandleRegister);
  DCHECK_EQ(jump_mode, JumpMode::kJump);
  jmp(rcx);
}


int MacroAssembler::RequiredStackSizeForCallerSaved(SaveFPRegsMode fp_mode,
                                                    Register exclusion) const {
  int bytes = 0;
  RegList saved_regs = kCallerSaved - exclusion;
  bytes += kSystemPointerSize * saved_regs.Count();

  // R12 to r15 are callee save on all platforms.
  if (fp_mode == SaveFPRegsMode::kSave) {
    bytes += kStackSavedSavedFPSize * kAllocatableDoubleRegisters.Count();
  }

  return bytes;
}

int MacroAssembler::PushCallerSaved(SaveFPRegsMode fp_mode,
                                    Register exclusion) {
  ASM_CODE_COMMENT(this);
  int bytes = 0;
  bytes += PushAll(kCallerSaved - exclusion);
  if (fp_mode == SaveFPRegsMode::kSave) {
    bytes += PushAll(kAllocatableDoubleRegisters);
  }

  return bytes;
}

int MacroAssembler::PopCallerSaved(SaveFPRegsMode fp_mode, Register exclusion) {
  ASM_CODE_COMMENT(this);
  int bytes = 0;
  if (fp_mode == SaveFPRegsMode::kSave) {
    bytes += PopAll(kAllocatableDoubleRegisters);
  }
  bytes += PopAll(kCallerSaved - exclusion);

  return bytes;
}

int MacroAssembler::PushAll(RegList registers) {
  int bytes = 0;
  for (Register reg : registers) {
    pushq(reg);
    bytes += kSystemPointerSize;
  }
  return bytes;
}

int MacroAssembler::PopAll(RegList registers) {
  int bytes = 0;
  for (Register reg : base::Reversed(registers)) {
    popq(reg);
    bytes += kSystemPointerSize;
  }
  return bytes;
}

int MacroAssembler::PushAll(DoubleRegList registers, int stack_slot_size) {
  if (registers.is_empty()) return 0;
  const int delta = stack_slot_size * registers.Count();
  AllocateStackSpace(delta);
  int slot = 0;
  for (XMMRegister reg : registers) {
    if (stack_slot_size == kDoubleSize) {
      Movsd(Operand(rsp, slot), reg);
    } else {
      DCHECK_EQ(stack_slot_size, 2 * kDoubleSize);
      Movdqu(Operand(rsp, slot), reg);
    }
    slot += stack_slot_size;
  }
  DCHECK_EQ(slot, delta);
  return delta;
}

int MacroAssembler::PopAll(DoubleRegList registers, int stack_slot_size) {
  if (registers.is_empty()) return 0;
  int slot = 0;
  for (XMMRegister reg : registers) {
    if (stack_slot_size == kDoubleSize) {
      Movsd(reg, Operand(rsp, slot));
    } else {
      DCHECK_EQ(stack_slot_size, 2 * kDoubleSize);
      Movdqu(reg, Operand(rsp, slot));
    }
    slot += stack_slot_size;
  }
  DCHECK_EQ(slot, stack_slot_size * registers.Count());
  addq(rsp, Immediate(slot));
  return slot;
}

void MacroAssembler::Movq(XMMRegister dst, Register src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope avx_scope(this, AVX);
    vmovq(dst, src);
  } else {
    movq(dst, src);
  }
}

void MacroAssembler::Movq(Register dst, XMMRegister src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope avx_scope(this, AVX);
    vmovq(dst, src);
  } else {
    movq(dst, src);
  }
}

void MacroAssembler::Pextrq(Register dst, XMMRegister src, int8_t imm8) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope avx_scope(this, AVX);
    vpextrq(dst, src, imm8);
  } else {
    CpuFeatureScope sse_scope(this, SSE4_1);
    pextrq(dst, src, imm8);
  }
}

void MacroAssembler::Cvtss2sd(XMMRegister dst, XMMRegister src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtss2sd(dst, src, src);
  } else {
    cvtss2sd(dst, src);
  }
}

void MacroAssembler::Cvtss2sd(XMMRegister dst, Operand src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtss2sd(dst, dst, src);
  } else {
    cvtss2sd(dst, src);
  }
}

void MacroAssembler::Cvtsd2ss(XMMRegister dst, XMMRegister src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtsd2ss(dst, src, src);
  } else {
    cvtsd2ss(dst, src);
  }
}

void MacroAssembler::Cvtsd2ss(XMMRegister dst, Operand src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtsd2ss(dst, dst, src);
  } else {
    cvtsd2ss(dst, src);
  }
}

void MacroAssembler::Cvtlsi2sd(XMMRegister dst, Register src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtlsi2sd(dst, kScratchDoubleReg, src);
  } else {
    xorpd(dst, dst);
    cvtlsi2sd(dst, src);
  }
}

void MacroAssembler::Cvtlsi2sd(XMMRegister dst, Operand src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtlsi2sd(dst, kScratchDoubleReg, src);
  } else {
    xorpd(dst, dst);
    cvtlsi2sd(dst, src);
  }
}

void MacroAssembler::Cvtlsi2ss(XMMRegister dst, Register src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtlsi2ss(dst, kScratchDoubleReg, src);
  } else {
    xorps(dst, dst);
    cvtlsi2ss(dst, src);
  }
}

void MacroAssembler::Cvtlsi2ss(XMMRegister dst, Operand src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtlsi2ss(dst, kScratchDoubleReg, src);
  } else {
    xorps(dst, dst);
    cvtlsi2ss(dst, src);
  }
}

void MacroAssembler::Cvtqsi2ss(XMMRegister dst, Register src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtqsi2ss(dst, kScratchDoubleReg, src);
  } else {
    xorps(dst, dst);
    cvtqsi2ss(dst, src);
  }
}

void MacroAssembler::Cvtqsi2ss(XMMRegister dst, Operand src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtqsi2ss(dst, kScratchDoubleReg, src);
  } else {
    xorps(dst, dst);
    cvtqsi2ss(dst, src);
  }
}

void MacroAssembler::Cvtqsi2sd(XMMRegister dst, Register src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtqsi2sd(dst, kScratchDoubleReg, src);
  } else {
    xorpd(dst, dst);
    cvtqsi2sd(dst, src);
  }
}

void MacroAssembler::Cvtqsi2sd(XMMRegister dst, Operand src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvtqsi2sd(dst, kScratchDoubleReg, src);
  } else {
    xorpd(dst, dst);
    cvtqsi2sd(dst, src);
  }
}

void MacroAssembler::Cvtlui2ss(XMMRegister dst, Register src) {
  // Zero-extend the 32 bit value to 64 bit.
  movl(kScratchRegister, src);
  Cvtqsi2ss(dst, kScratchRegister);
}

void MacroAssembler::Cvtlui2ss(XMMRegister dst, Operand src) {
  // Zero-extend the 32 bit value to 64 bit.
  movl(kScratchRegister, src);
  Cvtqsi2ss(dst, kScratchRegister);
}

void MacroAssembler::Cvtlui2sd(XMMRegister dst, Register src) {
  // Zero-extend the 32 bit value to 64 bit.
  movl(kScratchRegister, src);
  Cvtqsi2sd(dst, kScratchRegister);
}

void MacroAssembler::Cvtlui2sd(XMMRegister dst, Operand src) {
  // Zero-extend the 32 bit value to 64 bit.
  movl(kScratchRegister, src);
  Cvtqsi2sd(dst, kScratchRegister);
}

void MacroAssembler::Cvtqui2ss(XMMRegister dst, Register src) {
  Label done;
  Cvtqsi2ss(dst, src);
  testq(src, src);
  j(positive, &done, Label::kNear);

  // Compute {src/2 | (src&1)} (retain the LSB to avoid rounding errors).
  if (src != kScratchRegister) movq(kScratchRegister, src);
  shrq(kScratchRegister, Immediate(1));
  // The LSB is shifted into CF. If it is set, set the LSB in {tmp}.
  Label msb_not_set;
  j(not_carry, &msb_not_set, Label::kNear);
  orq(kScratchRegister, Immediate(1));
  bind(&msb_not_set);
  Cvtqsi2ss(dst, kScratchRegister);
  Addss(dst, dst);
  bind(&done);
}

void MacroAssembler::Cvtqui2ss(XMMRegister dst, Operand src) {
  movq(kScratchRegister, src);
  Cvtqui2ss(dst, kScratchRegister);
}

void MacroAssembler::Cvtqui2sd(XMMRegister dst, Register src) {
  Label done;
  Cvtqsi2sd(dst, src);
  testq(src, src);
  j(positive, &done, Label::kNear);

  // Compute {src/2 | (src&1)} (retain the LSB to avoid rounding errors).
  if (src != kScratchRegister) movq(kScratchRegister, src);
  shrq(kScratchRegister, Immediate(1));
  // The LSB is shifted into CF. If it is set, set the LSB in {tmp}.
  Label msb_not_set;
  j(not_carry, &msb_not_set, Label::kNear);
  orq(kScratchRegister, Immediate(1));
  bind(&msb_not_set);
  Cvtqsi2sd(dst, kScratchRegister);
  Addsd(dst, dst);
  bind(&done);
}

void MacroAssembler::Cvtqui2sd(XMMRegister dst, Operand src) {
  movq(kScratchRegister, src);
  Cvtqui2sd(dst, kScratchRegister);
}

void MacroAssembler::Cvttss2si(Register dst, XMMRegister src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvttss2si(dst, src);
  } else {
    cvttss2si(dst, src);
  }
}

void MacroAssembler::Cvttss2si(Register dst, Operand src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvttss2si(dst, src);
  } else {
    cvttss2si(dst, src);
  }
}

void MacroAssembler::Cvttsd2si(Register dst, XMMRegister src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvttsd2si(dst, src);
  } else {
    cvttsd2si(dst, src);
  }
}

void MacroAssembler::Cvttsd2si(Register dst, Operand src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvttsd2si(dst, src);
  } else {
    cvttsd2si(dst, src);
  }
}

void MacroAssembler::Cvttss2siq(Register dst, XMMRegister src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvttss2siq(dst, src);
  } else {
    cvttss2siq(dst, src);
  }
}

void MacroAssembler::Cvttss2siq(Register dst, Operand src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvttss2siq(dst, src);
  } else {
    cvttss2siq(dst, src);
  }
}

void MacroAssembler::Cvttsd2siq(Register dst, XMMRegister src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvttsd2siq(dst, src);
  } else {
    cvttsd2siq(dst, src);
  }
}

void MacroAssembler::Cvttsd2siq(Register dst, Operand src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope scope(this, AVX);
    vcvttsd2siq(dst, src);
  } else {
    cvttsd2siq(dst, src);
  }
}

void MacroAssembler::Cvtph2pd(XMMRegister dst, XMMRegister src) {
  ASM_CODE_COMMENT(this);
  CpuFeatureScope f16c_scope(this, F16C);
  CpuFeatureScope avx_scope(this, AVX);

  vcvtph2ps(dst, src);
  Cvtss2sd(dst, dst);
}

void MacroAssembler::Cvtpd2ph(XMMRegister dst, XMMRegister src, Register tmp) {
  ASM_CODE_COMMENT(this);
  CpuFeatureScope f16c_scope(this, F16C);
  CpuFeatureScope avx_scope(this, AVX);
  Register tmp2 = kScratchRegister;
  DCHECK_NE(tmp, tmp2);
  DCHECK_NE(dst, src);

  // Conversion algo from
  // https://github.com/tc39/proposal-float16array/issues/12#issuecomment-2256642971
  Label f32tof16;
  // Convert Float64 -> Float32.
  Cvtsd2ss(dst, src);
  vmovd(tmp, dst);
  // Mask off sign bit.
  andl(tmp, Immediate(kFP32WithoutSignMask));
  // Underflow to zero.
  cmpl(tmp, Immediate(kFP32MinFP16ZeroRepresentable));
  j(below, &f32tof16);
  // Overflow to infinity.
  cmpl(tmp, Immediate(kFP32MaxFP16Representable));
  j(above_equal, &f32tof16);
  // Detection of subnormal numbers.
  cmpl(tmp, Immediate(kFP32SubnormalThresholdOfFP16));
  setcc(above_equal, tmp2);
  movzxbl(tmp2, tmp2);
  // Compute 0x1000 for normal and 0x0000 for denormal numbers.
  shll(tmp2, Immediate(12));
  // Look at the last thirteen bits of the mantissa which will be shifted out
  // when converting from float32 to float16. (The round and sticky bits.)
  // Normal numbers: If the round bit is set and sticky bits are zero, then
  // adjust the float32 mantissa.
  // Denormal numbers: If all bits are zero, then adjust the mantissa.
  andl(tmp, Immediate(0x1fff));
  // Check round and sticky bits.
  cmpl(tmp, tmp2);
  j(not_equal, &f32tof16);

  // Adjust mantissa by -1/0/+1.
  Move(kScratchDoubleReg, static_cast<uint32_t>(1));
  psignd(kScratchDoubleReg, src);
  paddd(dst, kScratchDoubleReg);

  bind(&f32tof16);
  // Convert Float32 -> Float16.
  vcvtps2ph(dst, dst, 4);
}

namespace {
template <typename OperandOrXMMRegister, bool is_double>
void ConvertFloatToUint64(MacroAssembler* masm, Register dst,
                          OperandOrXMMRegister src, Label* fail) {
  Label success;
  // There does not exist a native float-to-uint instruction, so we have to use
  // a float-to-int, and postprocess the result.
  if (is_double) {
    masm->Cvttsd2siq(dst, src);
  } else {
    masm->Cvttss2siq(dst, src);
  }
  // If the result of the conversion is positive, we are already done.
  masm->testq(dst, dst);
  masm->j(positive, &success);
  // The result of the first conversion was negative, which means that the
  // input value was not within the positive int64 range. We subtract 2^63
  // and convert it again to see if it is within the uint64 range.
  if (is_double) {
    masm->Move(kScratchDoubleReg, -9223372036854775808.0);
    masm->Addsd(kScratchDoubleReg, src);
    masm->Cvttsd2siq(dst, kScratchDoubleReg);
  } else {
    masm->Move(kScratchDoubleReg, -9223372036854775808.0f);
    masm->Addss(kScratchDoubleReg, src);
    masm->Cvttss2siq(dst, kScratchDoubleReg);
  }
  masm->testq(dst, dst);
  // The only possible negative value here is 0x8000000000000000, which is
  // used on x64 to indicate an integer overflow.
  masm->j(negative, fail ? fail : &success);
  // The input value is within uint64 range and the second conversion worked
  // successfully, but we still have to undo the subtraction we did
  // earlier.
  masm->Move(kScratchRegister, 0x8000000000000000);
  masm->orq(dst, kScratchRegister);
  masm->bind(&success);
}

template <typename OperandOrXMMRegister, bool is_double>
void ConvertFloatToUint32(MacroAssembler* masm, Register dst,
                          OperandOrXMMRegister src, Label* fail) {
  Label success;
  // There does not exist a native float-to-uint instruction, so we have to use
  // a float-to-int, and postprocess the result.
  if (is_double) {
    masm->Cvttsd2si(dst, src);
  } else {
    masm->Cvttss2si(dst, src);
  }
  // If the result of the conversion is positive, we are already done.
  masm->testl(dst, dst);
  masm->j(positive, &success);
  // The result of the first conversion was negative, which means that the
  // input value was not within the positive int32 range. We subtract 2^31
  // and convert it again to see if it is within the uint32 range.
  if (is_double) {
    masm->Move(kScratchDoubleReg, -2147483648.0);
    masm->Addsd(kScratchDoubleReg, src);
    masm->Cvttsd2si(dst, kScratchDoubleReg);
  } else {
    masm->Move(kScratchDoubleReg, -2147483648.0f);
    masm->Addss(kScratchDoubleReg, src);
    masm->Cvttss2si(dst, kScratchDoubleReg);
  }
  masm->testl(dst, dst);
  // The only possible negative value here is 0x80000000, which is
  // used on x64 to indicate an integer overflow.
  masm->j(negative, fail ? fail : &success);
  // The input value is within uint32 range and the second conversion worked
  // successfully, but we still have to undo the subtraction we did
  // earlier.
  masm->Move(kScratchRegister, 0x80000000);
  masm->orl(dst, kScratchRegister);
  masm->bind(&success);
}
}  // namespace

void MacroAssembler::Cvttsd2uiq(Register dst, Operand src, Label* fail) {
  ConvertFloatToUint64<Operand, true>(this, dst, src, fail);
}

void MacroAssembler::Cvttsd2uiq(Register dst, XMMRegister src, Label* fail) {
  ConvertFloatToUint64<XMMRegister, true>(this, dst, src, fail);
}

void MacroAssembler::Cvttsd2ui(Register dst, Operand src, Label* fail) {
  ConvertFloatToUint32<Operand, true>(this, dst, src, fail);
}

void MacroAssembler::Cvttsd2ui(Register dst, XMMRegister src, Label* fail) {
  ConvertFloatToUint32<XMMRegister, true>(this, dst, src, fail);
}

void MacroAssembler::Cvttss2uiq(Register dst, Operand src, Label* fail) {
  ConvertFloatToUint64<Operand, false>(this, dst, src, fail);
}

void MacroAssembler::Cvttss2uiq(Register dst, XMMRegister src, Label* fail) {
  ConvertFloatToUint64<XMMRegister, false>(this, dst, src, fail);
}

void MacroAssembler::Cvttss2ui(Register dst, Operand src, Label* fail) {
  ConvertFloatToUint32<Operand, false>(this, dst, src, fail);
}

void MacroAssembler::Cvttss2ui(Register dst, XMMRegister src, Label* fail) {
  ConvertFloatToUint32<XMMRegister, false>(this, dst, src, fail);
}

void MacroAssembler::Cmpeqss(XMMRegister dst, XMMRegister src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope avx_scope(this, AVX);
    vcmpeqss(dst, src);
  } else {
    cmpeqss(dst, src);
  }
}

void MacroAssembler::Cmpeqsd(XMMRegister dst, XMMRegister src) {
  if (CpuFeatures::IsSupported(AVX)) {
    CpuFeatureScope avx_scope(this, AVX);
    vcmpeqsd(dst, src);
  } else {
    cmpeqsd(dst, src);
  }
}

void MacroAssembler::S256Not(YMMRegister dst, YMMRegister src,
                             YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  CpuFeatureScope avx2_scope(this, AVX2);
  if (dst == src) {
    vpcmpeqd(scratch, scratch, scratch);
    vpxor(dst, dst, scratch);
  } else {
    vpcmpeqd(dst, dst, dst);
    vpxor(dst, dst, src);
  }
}

void MacroAssembler::S256Select(YMMRegister dst, YMMRegister mask,
                                YMMRegister src1, YMMRegister src2,
                                YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  CpuFeatureScope avx2_scope(this, AVX2);
  // v256.select = v256.or(v256.and(v1, c), v256.andnot(v2, c)).
  // pandn(x, y) = !x & y, so we have to flip the mask and input.
  vpandn(scratch, mask, src2);
  vpand(dst, src1, mask);
  vpor(dst, dst, scratch);
}

// ----------------------------------------------------------------------------
// Smi tagging, untagging and tag detection.

Register MacroAssembler::GetSmiConstant(Tagged<Smi> source) {
  Move(kScratchRegister, source);
  return kScratchRegister;
}

void MacroAssembler::Cmp(Register dst, int32_t src) {
  if (src == 0) {
    testl(dst, dst);
  } else {
    cmpl(dst, Immediate(src));
  }
}

void MacroAssembler::I64x4Mul(YMMRegister dst, YMMRegister lhs, YMMRegister rhs,
                              YMMRegister tmp1, YMMRegister tmp2) {
  ASM_CODE_COMMENT(this);
  DCHECK(!AreAliased(dst, tmp1, tmp2));
  DCHECK(!AreAliased(lhs, tmp1, tmp2));
  DCHECK(!AreAliased(rhs, tmp1, tmp2));
  DCHECK(CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx_scope(this, AVX2);
  // 1. Multiply high dword of each qword of left with right.
  vpsrlq(tmp1, lhs, uint8_t{32});
  vpmuludq(tmp1, tmp1, rhs);
  // 2. Multiply high dword of each qword of right with left.
  vpsrlq(tmp2, rhs, uint8_t{32});
  vpmuludq(tmp2, tmp2, lhs);
  // 3. Add 1 and 2, then shift left by 32 (this is the high dword of result).
  vpaddq(tmp2, tmp2, tmp1);
  vpsllq(tmp2, tmp2, uint8_t{32});
  // 4. Multiply low dwords (this is the low dword of result).
  vpmuludq(dst, lhs, rhs);
  // 5. Add 3 and 4.
  vpaddq(dst, dst, tmp2);
}

#define DEFINE_ISPLAT(name, suffix, instr_mov)                               \
  void MacroAssembler::name(YMMRegister dst, Register src) {                 \
    ASM_CODE_COMMENT(this);                                                  \
    DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2)); \
    CpuFeatureScope avx_scope(this, AVX);                                    \
    CpuFeatureScope avx2_scope(this, AVX2);                                  \
    instr_mov(dst, src);                                                     \
    vpbroadcast##suffix(dst, dst);                                           \
  }                                                                          \
                                                                             \
  void MacroAssembler::name(YMMRegister dst, Operand src) {                  \
    ASM_CODE_COMMENT(this);                                                  \
    DCHECK(CpuFeatures::IsSupported(AVX2));                                  \
    CpuFeatureScope avx2_scope(this, AVX2);                                  \
    vpbroadcast##suffix(dst, src);                                           \
  }

MACRO_ASM_X64_ISPLAT_LIST(DEFINE_ISPLAT)

#undef DEFINE_ISPLAT

void MacroAssembler::F64x4Splat(YMMRegister dst, XMMRegister src) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx2_scope(this, AVX2);
  vbroadcastsd(dst, src);
}

void MacroAssembler::F32x8Splat(YMMRegister dst, XMMRegister src) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx2_scope(this, AVX2);
  vbroadcastss(dst, src);
}

void MacroAssembler::F64x4Min(YMMRegister dst, YMMRegister lhs, YMMRegister rhs,
                              YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx_scope(this, AVX);
  CpuFeatureScope avx2_scope(this, AVX2);
  vminpd(scratch, lhs, rhs);
  vminpd(dst, rhs, lhs);
  vorpd(scratch, scratch, dst);
  vcmpunordpd(dst, dst, scratch);
  vorpd(scratch, scratch, dst);
  vpsrlq(dst, dst, uint8_t{13});
  vandnpd(dst, dst, scratch);
}

void MacroAssembler::F64x4Max(YMMRegister dst, YMMRegister lhs, YMMRegister rhs,
                              YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx_scope(this, AVX);
  CpuFeatureScope avx2_scope(this, AVX2);
  vmaxpd(scratch, lhs, rhs);
  vmaxpd(dst, rhs, lhs);
  vxorpd(dst, dst, scratch);
  vorpd(scratch, scratch, dst);
  vsubpd(scratch, scratch, dst);
  vcmpunordpd(dst, dst, scratch);
  vpsrlq(dst, dst, uint8_t{13});
  vandnpd(dst, dst, scratch);
}

void MacroAssembler::F32x8Min(YMMRegister dst, YMMRegister lhs, YMMRegister rhs,
                              YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx_scope(this, AVX);
  CpuFeatureScope avx2_scope(this, AVX2);
  vminps(scratch, lhs, rhs);
  vminps(dst, rhs, lhs);
  vorps(scratch, scratch, dst);
  vcmpunordps(dst, dst, scratch);
  vorps(scratch, scratch, dst);
  vpsrld(dst, dst, uint8_t{10});
  vandnps(dst, dst, scratch);
}

void MacroAssembler::F32x8Max(YMMRegister dst, YMMRegister lhs, YMMRegister rhs,
                              YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx_scope(this, AVX);
  CpuFeatureScope avx2_scope(this, AVX2);
  vmaxps(scratch, lhs, rhs);
  vmaxps(dst, rhs, lhs);
  vxorps(dst, dst, scratch);
  vorps(scratch, scratch, dst);
  vsubps(scratch, scratch, dst);
  vcmpunordps(dst, dst, scratch);
  vpsrld(dst, dst, uint8_t{10});
  vandnps(dst, dst, scratch);
}

void MacroAssembler::F16x8Min(YMMRegister dst, XMMRegister lhs, XMMRegister rhs,
                              YMMRegister scratch, YMMRegister scratch2) {
  ASM_CODE_COMMENT(this);
  CpuFeatureScope f16c_scope(this, F16C);
  CpuFeatureScope avx_scope(this, AVX);
  CpuFeatureScope avx2_scope(this, AVX2);
  vcvtph2ps(scratch, lhs);
  vcvtph2ps(scratch2, rhs);
  // The minps instruction doesn't propagate NaNs and +0's in its first
  // operand. Perform minps in both orders, merge the results, and adjust.
  vminps(dst, scratch, scratch2);
  vminps(scratch, scratch2, scratch);
  // Propagate -0's and NaNs, which may be non-canonical.
  vorps(scratch, scratch, dst);
  // Canonicalize NaNs by quieting and clearing the payload.
  vcmpunordps(dst, dst, scratch);
  vorps(scratch, scratch, dst);
  vpsrld(dst, dst, uint8_t{10});
  vandnps(dst, dst, scratch);
  vcvtps2ph(dst, dst, 0);
}

void MacroAssembler::F16x8Max(YMMRegister dst, XMMRegister lhs, XMMRegister rhs,
                              YMMRegister scratch, YMMRegister scratch2) {
  ASM_CODE_COMMENT(this);
  CpuFeatureScope f16c_scope(this, F16C);
  CpuFeatureScope avx_scope(this, AVX);
  CpuFeatureScope avx2_scope(this, AVX2);
  vcvtph2ps(scratch, lhs);
  vcvtph2ps(scratch2, rhs);
  // The maxps instruction doesn't propagate NaNs and +0's in its first
  // operand. Perform maxps in both orders, merge the results, and adjust.
  vmaxps(dst, scratch, scratch2);
  vmaxps(scratch, scratch2, scratch);
  // Find discrepancies.
  vxorps(dst, dst, scratch);
  // Propagate NaNs, which may be non-canonical.
  vorps(scratch, scratch, dst);
  // Propagate sign discrepancy and (subtle) quiet NaNs.
  vsubps(scratch, scratch, dst);
  // Canonicalize NaNs by clearing the payload. Sign is non-deterministic.
  vcmpunordps(dst, dst, scratch);
  vpsrld(dst, dst, uint8_t{10});
  vandnps(dst, dst, scratch);
  vcvtps2ph(dst, dst, 0);
}

// 1. Zero extend 4 packed 32-bit integers in src1 to 4 packed 64-bit integers
// in scratch
// 2. Zero extend 4 packed 32-bit integers in src2 to 4 packed 64-bit integers
// in dst
// 3. Multiply packed doubleword integers in scratch with dst, the extended zero
// are ignored
void MacroAssembler::I64x4ExtMul(YMMRegister dst, XMMRegister src1,
                                 XMMRegister src2, YMMRegister scratch,
                                 bool is_signed) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx_scope(this, AVX2);
  vpmovzxdq(scratch, src1);
  vpmovzxdq(dst, src2);
  if (is_signed) {
    vpmuldq(dst, scratch, dst);
  } else {
    vpmuludq(dst, scratch, dst);
  }
}

// 1. Extend 8 packed 16-bit integers in src1 to 8 packed 32-bit integers in
// scratch
// 2. Extend 8 packed 16-bit integers in src2 to 8 packed 32-bit integers in dst
// 3. Multiply the packed doubleword integers in scratch and dst and store the
// low 32 bits of each product in dst.
void MacroAssembler::I32x8ExtMul(YMMRegister dst, XMMRegister src1,
                                 XMMRegister src2, YMMRegister scratch,
                                 bool is_signed) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx_scope(this, AVX2);
  is_signed ? vpmovsxwd(scratch, src1) : vpmovzxwd(scratch, src1);
  is_signed ? vpmovsxwd(dst, src2) : vpmovzxwd(dst, src2);
  vpmulld(dst, dst, scratch);
}

void MacroAssembler::I16x16ExtMul(YMMRegister dst, XMMRegister src1,
                                  XMMRegister src2, YMMRegister scratch,
                                  bool is_signed) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx_scope(this, AVX2);
  is_signed ? vpmovsxbw(scratch, src1) : vpmovzxbw(scratch, src1);
  is_signed ? vpmovsxbw(dst, src2) : vpmovzxbw(dst, src2);
  vpmullw(dst, dst, scratch);
}

void MacroAssembler::I32x8ExtAddPairwiseI16x16S(YMMRegister dst,
                                                YMMRegister src,
                                                YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx2_scope(this, AVX2);
  Move(scratch, uint32_t{1});
  vpbroadcastw(scratch, scratch);
  // vpmaddwd multiplies signed words in src and op, producing
  // signed doublewords, then adds pairwise.
  // src = |l0|l1|...|l14|l15|
  // dst = |l0*1+l1*1|l2*1+l3*1|...|l14*1+l15*1|
  vpmaddwd(dst, src, scratch);
}

void MacroAssembler::I32x8ExtAddPairwiseI16x16U(YMMRegister dst,
                                                YMMRegister src,
                                                YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx2_scope(this, AVX2);
  // src = |l0|l1|...l14|l15|
  // scratch = |0|l0|0|l2|...|0|l14|
  vpsrld(scratch, src, 16);
  // dst = |0|l1|0|l3|...|0|l15|
  vpblendw(dst, src, scratch, 0xAA);
  vpaddd(dst, dst, scratch);
}

void MacroAssembler::I16x16ExtAddPairwiseI8x32S(YMMRegister dst,
                                                YMMRegister src,
                                                YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx2_scope(this, AVX2);
  Move(scratch, uint32_t{1});
  vpbroadcastb(scratch, scratch);
  // pmaddubsw treats the first operand as unsigned, so scratch here should
  // be first operand
  // src = |l0|l1|...|l34|l35|
  // dst = |l0*1+l1*1|l2*1+l3*1|...|l34*1+l35*1|
  vpmaddubsw(dst, scratch, src);
}

void MacroAssembler::I16x16ExtAddPairwiseI8x32U(YMMRegister dst,
                                                YMMRegister src,
                                                YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx2_scope(this, AVX2);
  Move(scratch, uint32_t{1});
  vpbroadcastb(scratch, scratch);
  vpmaddubsw(dst, src, scratch);
}

void MacroAssembler::I32x8SConvertF32x8(YMMRegister dst, YMMRegister src,
                                        YMMRegister tmp, Register scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx_scope(this, AVX);
  CpuFeatureScope avx2_scope(this, AVX2);
  Operand int32_overflow_as_float = ExternalReferenceAsOperand(
      ExternalReference::address_of_wasm_i32x8_int32_overflow_as_float(),
      scratch);
  // This algorithm works by:
  // 1. lanes with NaNs are zero-ed
  // 2. lanes ge than 2147483648.0f (MAX_INT32+1) set to 0xffff'ffff
  // 3. cvttps2dq sets all out of range lanes to 0x8000'0000
  //   a. correct for underflows (< MIN_INT32)
  //   b. wrong for overflow, and we know which lanes overflow from 2.
  // 4. adjust for 3b by xor-ing 2 and 3
  //   a. 0x8000'0000 xor 0xffff'ffff = 0x7fff'ffff (MAX_INT32)
  vcmpeqps(tmp, src, src);
  vandps(dst, src, tmp);
  vcmpgeps(tmp, src, int32_overflow_as_float);
  vcvttps2dq(dst, dst);
  vpxor(dst, dst, tmp);
}

void MacroAssembler::I16x8SConvertF16x8(YMMRegister dst, XMMRegister src,
                                        YMMRegister tmp, Register scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2) &&
         CpuFeatures::IsSupported(F16C));

  CpuFeatureScope f16c_scope(this, F16C);
  CpuFeatureScope avx_scope(this, AVX);
  CpuFeatureScope avx2_scope(this, AVX2);

  Operand op = ExternalReferenceAsOperand(
      ExternalReference::address_of_wasm_i32x8_int32_overflow_as_float(),
      scratch);
  // Convert source f16 to f32.
  vcvtph2ps(dst, src);
  // Compare it to itself, NaNs are turn to 0s because don't equal to itself.
  vcmpeqps(tmp, dst, dst);
  // Reset NaNs.
  vandps(dst, dst, tmp);
  // Detect positive Infinity as an overflow above MAX_INT32.
  vcmpgeps(tmp, dst, op);
  // Convert f32 to i32.
  vcvttps2dq(dst, dst);
  // cvttps2dq sets all out of range lanes to 0x8000'0000,
  // but as soon as source values are result of conversion from f16,
  // and so less than MAX_INT32, only +Infinity is an issue.
  // Convert all infinities to MAX_INT32 and let vpackssdw
  // clamp it to MAX_INT16 later.
  // 0x8000'0000 xor 0xffff'ffff(from 2 steps before) = 0x7fff'ffff (MAX_INT32)
  vpxor(dst, dst, tmp);
  // We now have 8 i32 values. Using one character per 16 bits:
  // dst: [AABBCCDDEEFFGGHH]
  // Create a copy of the upper four values in the lower half of {tmp}
  // (so the upper half of the immediate doesn't matter):
  vpermq(tmp, dst, 0x4E);  // 0b01001110
  // tmp: [EEFFGGHHAABBCCDD]
  // Now pack them together as i16s. Note that {vpackssdw} interleaves
  // 128-bit chunks from each input, and takes care of saturating each
  // value to kMinInt16 and kMaxInt16. We will then ignore the upper half
  // of {dst}.
  vpackssdw(dst, dst, tmp);
  // dst: [EFGHABCDABCDEFGH]
  //       <--><--><--><-->
  //         ↑   ↑   ↑   └── from lower half of {dst}
  //         │   │   └────── from lower half of {tmp}
  //         │   └────────── from upper half of {dst} (ignored)
  //         └────────────── from upper half of {tmp} (ignored)
}

void MacroAssembler::I16x8TruncF16x8U(YMMRegister dst, XMMRegister src,
                                      YMMRegister tmp) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2) &&
         CpuFeatures::IsSupported(F16C));

  CpuFeatureScope f16c_scope(this, F16C);
  CpuFeatureScope avx_scope(this, AVX);
  CpuFeatureScope avx2_scope(this, AVX2);

  Operand op = ExternalReferenceAsOperand(
      ExternalReference::address_of_wasm_i32x8_int32_overflow_as_float(),
      kScratchRegister);
  vcvtph2ps(dst, src);
  // NAN->0, negative->0.
  vpxor(tmp, tmp, tmp);
  vmaxps(dst, dst, tmp);
  // Detect positive Infinity as an overflow above MAX_INT32.
  vcmpgeps(tmp, dst, op);
  // Convert to int.
  vcvttps2dq(dst, dst);
  // cvttps2dq sets all out of range lanes to 0x8000'0000,
  // but as soon as source values are result of conversion from f16,
  // and so less than MAX_INT32, only +Infinity is an issue.
  // Convert all infinities to MAX_INT32 and let vpackusdw
  // clamp it to MAX_INT16 later.
  // 0x8000'0000 xor 0xffff'ffff(from 2 steps before) = 0x7fff'ffff (MAX_INT32)
  vpxor(dst, dst, tmp);
  // Move high part to a spare register.
  // See detailed comment in {I16x8SConvertF16x8} for how this works.
  vpermq(tmp, dst, 0x4E);  // 0b01001110
  vpackusdw(dst, dst, tmp);
}

void MacroAssembler::F16x8Qfma(YMMRegister dst, XMMRegister src1,
                               XMMRegister src2, XMMRegister src3,
                               YMMRegister tmp, YMMRegister tmp2) {
  CpuFeatureScope fma3_scope(this, FMA3);
  CpuFeatureScope f16c_scope(this, F16C);

  if (dst.code() == src2.code()) {
    vcvtph2ps(dst, dst);
    vcvtph2ps(tmp, src1);
    vcvtph2ps(tmp2, src3);
    vfmadd213ps(dst, tmp, tmp2);
  } else if (dst.code() == src3.code()) {
    vcvtph2ps(dst, dst);
    vcvtph2ps(tmp, src2);
    vcvtph2ps(tmp2, src1);
    vfmadd231ps(dst, tmp, tmp2);
  } else {
    vcvtph2ps(dst, src1);
    vcvtph2ps(tmp, src2);
    vcvtph2ps(tmp2, src3);
    vfmadd213ps(dst, tmp, tmp2);
  }
  vcvtps2ph(dst, dst, 0);
}

void MacroAssembler::F16x8Qfms(YMMRegister dst, XMMRegister src1,
                               XMMRegister src2, XMMRegister src3,
                               YMMRegister tmp, YMMRegister tmp2) {
  CpuFeatureScope fma3_scope(this, FMA3);
  CpuFeatureScope f16c_scope(this, F16C);

  if (dst.code() == src2.code()) {
    vcvtph2ps(dst, dst);
    vcvtph2ps(tmp, src1);
    vcvtph2ps(tmp2, src3);
    vfnmadd213ps(dst, tmp, tmp2);
  } else if (dst.code() == src3.code()) {
    vcvtph2ps(dst, dst);
    vcvtph2ps(tmp, src2);
    vcvtph2ps(tmp2, src1);
    vfnmadd231ps(dst, tmp, tmp2);
  } else {
    vcvtph2ps(dst, src1);
    vcvtph2ps(tmp, src2);
    vcvtph2ps(tmp2, src3);
    vfnmadd213ps(dst, tmp, tmp2);
  }
  vcvtps2ph(dst, dst, 0);
}

void MacroAssembler::F32x8Qfma(YMMRegister dst, YMMRegister src1,
                               YMMRegister src2, YMMRegister src3,
                               YMMRegister tmp) {
  QFMA(ps);
}

void MacroAssembler::F32x8Qfms(YMMRegister dst, YMMRegister src1,
                               YMMRegister src2, YMMRegister src3,
                               YMMRegister tmp) {
  QFMS(ps);
}

void MacroAssembler::F64x4Qfma(YMMRegister dst, YMMRegister src1,
                               YMMRegister src2, YMMRegister src3,
                               YMMRegister tmp) {
  QFMA(pd);
}

void MacroAssembler::F64x4Qfms(YMMRegister dst, YMMRegister src1,
                               YMMRegister src2, YMMRegister src3,
                               YMMRegister tmp) {
  QFMS(pd);
}

void MacroAssembler::I32x8DotI8x32I7x32AddS(YMMRegister dst, YMMRegister src1,
                                            YMMRegister src2, YMMRegister src3,
                                            YMMRegister scratch,
                                            YMMRegister splat_reg) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2));
  // It's guaranteed in instruction selector
  DCHECK_EQ(dst, src3);
  if (CpuFeatures::IsSupported(AVX_VNNI_INT8)) {
    CpuFeatureScope avx_vnni_int8_scope(this, AVX_VNNI_INT8);
    vpdpbssd(dst, src2, src1);
    return;
  } else if (CpuFeatures::IsSupported(AVX_VNNI)) {
    CpuFeatureScope avx_scope(this, AVX_VNNI);
    vpdpbusd(dst, src2, src1);
    return;
  }

  DCHECK_NE(scratch, splat_reg);
  CpuFeatureScope avx_scope(this, AVX);
  CpuFeatureScope avx2_scope(this, AVX2);
  // splat_reg = i16x16.splat(1)
  vpcmpeqd(splat_reg, splat_reg, splat_reg);
  vpsrlw(splat_reg, splat_reg, uint8_t{15});
  vpmaddubsw(scratch, src2, src1);
  vpmaddwd(scratch, splat_reg, scratch);
  vpaddd(dst, src3, scratch);
}

void MacroAssembler::I32x8TruncF32x8U(YMMRegister dst, YMMRegister src,
                                      YMMRegister scratch1,
                                      YMMRegister scratch2) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx_scope(this, AVX);
  CpuFeatureScope avx2_scope(this, AVX2);

  // NAN->0, negative->0.
  vpxor(scratch1, scratch1, scratch1);
  vmaxps(dst, src, scratch1);
  // scratch1: float representation of max_signed.
  vpcmpeqd(scratch1, scratch1, scratch1);
  vpsrld(scratch1, scratch1, uint8_t{1});  // 0x7fffffff
  vcvtdq2ps(scratch1, scratch1);           // 0x4f000000
  // scratch2: convert (src-max_signed).
  // Set positive overflow lanes to 0x7FFFFFFF.
  // Set negative lanes to 0.
  vsubps(scratch2, dst, scratch1);

  vcmpleps(scratch1, scratch1, scratch2);
  vcvttps2dq(scratch2, scratch2);
  vpxor(scratch2, scratch2, scratch1);
  vpxor(scratch1, scratch1, scratch1);
  vpmaxsd(scratch2, scratch2, scratch1);
  // Convert to int. Overflow lanes above max_signed will be 0x80000000.
  vcvttps2dq(dst, dst);
  // Add (src-max_signed) for overflow lanes.
  vpaddd(dst, dst, scratch2);
}

void MacroAssembler::Negpd(YMMRegister dst, YMMRegister src,
                           YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx2_scope(this, AVX2);
  if (dst == src) {
    vpcmpeqq(scratch, scratch, scratch);
    vpsllq(scratch, scratch, uint8_t{63});
    vpxor(dst, dst, scratch);
  } else {
    vpcmpeqq(dst, dst, dst);
    vpsllq(dst, dst, uint8_t{63});
    vpxor(dst, dst, src);
  }
}

void MacroAssembler::Negps(YMMRegister dst, YMMRegister src,
                           YMMRegister scratch) {
  ASM_CODE_COMMENT(this);
  DCHECK(CpuFeatures::IsSupported(AVX) && CpuFeatures::IsSupported(AVX2));
  CpuFeatureScope avx2_scope(this, AVX2);
  if (dst == src) {
    vpcmpeqd(scratch, scratch, scratch);
    vpslld(scratch, scratch, uint8_t{31});
    vpxor(dst, dst, scratch);
  } else {
    vpcmpeqd(dst, dst, dst);
    vpslld(dst, dst, uint8_t{31});
    vpxor(dst, dst, src);
  }
}

void MacroAssembler::SmiTag(Register reg) {
  static_assert(kSmiTag == 0);
  DCHECK(SmiValuesAre32Bits() || SmiValuesAre31Bits());
  if (COMPRESS_POINTERS_BOOL) {
    DCHECK_EQ(kSmiShift, 1);
    addl(reg, reg);
  } else {
    shlq(reg, Immediate(kSmiShift));
  }
#ifdef ENABLE_SLOW_DCHECKS
  ClobberDecompressedSmiBits(reg);
#endif
}

void MacroAssembler::SmiTag(Register dst, Register src) {
  if (dst != src) {
    if (COMPRESS_POINTERS_BOOL) {
      movl(dst, src);
    } else {
      movq(dst, src);
    }
  }
  SmiTag(dst);
}

void MacroAssembler::SmiUntag(Register reg) {
  static_assert(kSmiTag == 0);
  DCHECK(SmiValuesAre32Bits() || SmiValuesAre31Bits());
  // TODO(v8:7703): Is there a way to avoid this sign extension when pointer
  // compression is enabled?
  if (COMPRESS_POINTERS_BOOL) {
    sarl(reg, Immediate(kSmiShift));
    movsxlq(reg, reg);
  } else {
    sarq(reg, Immediate(kSmiShift));
  }
}

void MacroAssembler::SmiUntagUnsigned(Register reg) {
  static_assert(kSmiTag == 0);
  DCHECK(SmiValuesAre32Bits() || SmiValuesAre31Bits());
  if (COMPRESS_POINTERS_BOOL) {
    AssertSignBitOfSmiIsZero(reg);
    shrl(reg, Immediate(kSmiShift));
  } else {
    shrq(reg, Immediate(kSmiShift));
  }
}

void MacroAssembler::SmiUntag(Register dst, Register src) {
  DCHECK(dst != src);
  if (COMPRESS_POINTERS_BOOL) {
    movsxlq(dst, src);
  } else {
    movq(dst, src);
  }
  // TODO(v8:7703): Call SmiUntag(reg) if we can find a way to avoid the extra
  // mov when pointer compression is enabled.
  static_assert(kSmiTag == 0);
  DCHECK(SmiValuesAre32Bits() || SmiValuesAre31Bits());
  sarq(dst, Immediate(kSmiShift));
}

void MacroAssembler::SmiUntag(Register dst, Operand src) {
  if (SmiValuesAre32Bits()) {
    // Sign extend to 64-bit.
    movsxlq(dst, Operand(src, kSmiShift / kBitsPerByte));
  } else {
    DCHECK(SmiValuesAre31Bits());
    if (COMPRESS_POINTERS_BOOL) {
      movsxlq(dst, src);
    } else {
      movq(dst, src);
    }
    sarq(dst, Immediate(kSmiShift));
  }
}

void MacroAssembler::SmiUntagUnsigned(Register dst, Operand src) {
  if (SmiValuesAre32Bits()) {
    // Zero extend to 64-bit.
    movl(dst, Operand(src, kSmiShift / kBitsPerByte));
  } else {
    DCHECK(SmiValuesAre31Bits());
    if (COMPRESS_POINTERS_BOOL) {
      movl(dst, src);
      AssertSignBitOfSmiIsZero(dst);
      shrl(dst, Immediate(kSmiShift));
    } else {
      movq(dst, src);
      shrq(dst, Immediate(kSmiShift));
    }
  }
}

void MacroAssembler::SmiToInt32(Register reg) {
  AssertSmi(reg);
  static_assert(kSmiTag == 0);
  DCHECK(SmiValuesAre32Bits() || SmiValuesAre31Bits());
  if (COMPRESS_POINTERS_BOOL) {
    sarl(reg, Immediate(kSmiShift));
  } else {
    shrq(reg, Immediate(kSmiShift));
  }
}

void MacroAssembler::SmiToInt32(Register dst, Register src) {
  if (dst != src) {
    mov_tagged(dst, src);
  }
  SmiToInt32(dst);
}

void MacroAssembler::SmiCompare(Register smi1, Register smi2) {
  AssertSmi(smi1);
  AssertSmi(smi2);
  cmp_tagged(smi1, smi2);
}

void MacroAssembler::SmiCompare(Register dst, Tagged<Smi> src) {
  AssertSmi(dst);
  Cmp(dst, src);
}

void MacroAssembler::Cmp(Register dst, Tagged<Smi> src) {
  if (src.value() == 0) {
    test_tagged(dst, dst);
  } else if (COMPRESS_POINTERS_BOOL) {
    cmp_tagged(dst, Immediate(src));
  } else {
    DCHECK_NE(dst, kScratchRegister);
    Register constant_reg = GetSmiConstant(src);
    cmp_tagged(dst, constant_reg);
  }
}

void MacroAssembler::SmiCompare(Register dst, Operand src) {
  AssertSmi(dst);
  AssertSmi(src);
  cmp_tagged(dst, src);
}

void MacroAssembler::SmiCompare(Operand dst, Register src) {
  AssertSmi(dst);
  AssertSmi(src);
  cmp_tagged(dst, src);
}

void MacroAssembler::SmiCompare(Operand dst, Tagged<Smi> src) {
  AssertSmi(dst);
  if (SmiValuesAre32Bits()) {
    cmpl(Operand(dst, kSmiShift / kBitsPerByte), Immediate(src.value()));
  } else {
    DCHECK(SmiValuesAre31Bits());
    cmpl(dst, Immediate(src));
  }
}

void MacroAssembler::Cmp(Operand dst, Tagged<Smi> src) {
  // The Operand cannot use the smi register.
  Register smi_reg = GetSmiConstant(src);
  DCHECK(!dst.AddressUsesRegister(smi_reg));
  cmp_tagged(dst, smi_reg);
}

void MacroAssembler::ClobberDecompressedSmiBits(Register src) {
#ifdef V8_COMPRESS_POINTERS
  ASM_CODE_COMMENT(this);
  static constexpr unsigned int clobber_mask = 0x515151;
  static constexpr int rot_to_unused =
      64 - kSmiShiftSize - kSmiTagSize - kSmiValueSize;
  rolq(src, Immediate(rot_to_unused));
  xorq(src, Immediate(clobber_mask));
  rorq(src, Immediate(rot_to_unused));
#endif
}

Condition MacroAssembler::CheckSmi(Register src) {
  static_assert(kSmiTag == 0);
  testb(src, Immediate(kSmiTagMask));
  return zero;
}

Condition MacroAssembler::CheckSmi(Operand src) {
  static_assert(kSmiTag == 0);
  testb(src, Immediate(kSmiTagMask));
  return zero;
}

void MacroAssembler::JumpIfSmi(Register src, Label* on_smi,
                               Label::Distance near_jump) {
  Condition smi = CheckSmi(src);
  j(smi, on_smi, near_jump);
}

void MacroAssembler::JumpIfNotSmi(Register src, Label* on_not_smi,
                                  Label::Distance near_jump) {
  Condition smi = CheckSmi(src);
  j(NegateCondition(smi), on_not_smi, near_jump);
}

void MacroAssembler::JumpIfNotSmi(Operand src, Label* on_not_smi,
                                  Label::Distance near_jump) {
  Condition smi = CheckSmi(src);
  j(NegateCondition(smi), on_not_smi, near_jump);
}

void MacroAssembler::SmiAddConstant(Operand dst, Tagged<Smi> constant) {
  if (constant.value() != 0) {
    if (SmiValuesAre32Bits()) {
      addl(Operand(dst, kSmiShift / kBitsPerByte), Immediate(constant.value()));
    } else {
      DCHECK(SmiValuesAre31Bits());
      if (kTaggedSize == kInt64Size) {
        // Sign-extend value after addition
        movl(kScratchRegister, dst);
        addl(kScratchRegister, Immediate(constant));
        movsxlq(kScratchRegister, kScratchRegister);
        movq(dst, kScratchRegister);
      } else {
        DCHECK_EQ(kTaggedSize, kInt32Size);
        addl(dst, Immediate(constant));
      }
    }
  }
}

SmiIndex MacroAssembler::SmiToIndex(Register dst, Register src, int shift) {
  if (SmiValuesAre32Bits()) {
    DCHECK(is_uint6(shift));
    // There is a possible optimization if shift is in the range 60-63, but that
    // will (and must) never happen.
    if (dst != src) {
      movq(dst, src);
    }
    if (shift < kSmiShift) {
      sarq(dst, Immediate(kSmiShift - shift));
    } else {
      shlq(dst, Immediate(shift - kSmiShift));
    }
    return SmiIndex(dst, times_1);
  } else {
    DCHECK(SmiValuesAre31Bits());
    // We have to sign extend the index register to 64-bit as the SMI might
    // be negative.
    movsxlq(dst, src);
    if (shift < kSmiShift) {
      sarq(dst, Immediate(kSmiShift - shift));
    } else if (shift != kSmiShift) {
      if (shift - kSmiShift <= static_cast<int>(times_8)) {
        return SmiIndex(dst, static_cast<ScaleFactor>(shift - kSmiShift));
      }
      shlq(dst, Immediate(shift - kSmiShift));
    }
    return SmiIndex(dst, times_1);
  }
}

void MacroAssembler::Switch(Register scratch, Register reg, int case_value_base,
                            Label** labels, int num_labels) {
  Register table = scratch;
  Label fallthrough, jump_table;
  if (case_value_base != 0) {
    subq(reg, Immediate(case_value_base));
  }
  cmpq(reg, Immediate(num_labels));
  j(above_equal, &fallthrough);
  leaq(table, MemOperand(&jump_table));
#ifdef V8_ENABLE_CET_IBT
  // Add the notrack prefix to disable landing pad enforcement.
  jmp(MemOperand(table, reg, times_8, 0), /*notrack=*/true);
#else
  jmp(MemOperand(table, reg, times_8, 0));
#endif
  // Emit the jump table inline, under the assumption that it's not too big.
  Align(kSystemPointerSize);
  bind(&jump_table);
  for (int i = 0; i < num_labels; ++i) {
    dq(labels[i]);
  }
  bind(&fallthrough);
}

void MacroAssembler::Push(Tagged<Smi> source) {
  intptr_t smi = static_cast<intptr_t>(source.ptr());
  if (is_int32(smi)) {
    Push(Immediate(static_cast<int32_t>(smi)));
    return;
  }
  int first_byte_set = base::bits::CountTrailingZeros64(smi) / 8;
  int last_byte_set = (63 - base::bits::CountLeadingZeros64(smi)) / 8;
  if (first_byte_set == last_byte_set) {
    // This sequence has only 7 bytes, compared to the 12 bytes below.
    Push(Immediate(0));
    movb(Operand(rsp, first_byte_set),
         Immediate(static_cast<int8_t>(smi >> (8 * first_byte_set))));
    return;
  }
  Register constant = GetSmiConstant(source);
  Push(constant);
}

// ----------------------------------------------------------------------------

void MacroAssembler::Move(Register dst, Tagged<Smi> source) {
  static_assert(kSmiTag == 0);
  int value = source.value();
  if (value == 0) {
    xorl(dst, dst);
  } else if (SmiValuesAre32Bits()) {
    Move(dst, source.ptr(), RelocInfo::NO_INFO);
  } else {
    uint32_t uvalue = static_cast<uint32_t>(source.ptr());
    Move(dst, uvalue);
  }
}

void MacroAssembler::Move(Operand dst, intptr_t x) {
  if (is_int32(x)) {
    movq(dst, Immediate(static_cast<int32_t>(x)));
  } else {
    Move(kScratchRegister, x);
    movq(dst, kScratchRegister);
  }
}

void MacroAssembler::Move(Register dst, ExternalReference ext) {
  // TODO(jgruber,v8:8887): Also consider a root-relative load when generating
  // non-isolate-independent code. In many cases it might be cheaper than
  // embedding the relocatable value.
  if (root_array_available()) {
    if (ext.IsIsolateFieldId()) {
      leaq(dst, Operand(kRootRegister, ext.offset_from_root_register()));
      return;
    } else if (options().isolate_independent_code) {
      IndirectLoadExternalReference(dst, ext);
      return;
    }
  }
  // External references should not get created with IDs if
  // `!root_array_available()`.
  CHECK(!ext.IsIsolateFieldId());
  movq(dst, Immediate64(ext.address(), RelocInfo::EXTERNAL_REFERENCE));
}

void MacroAssembler::Move(Register dst, Register src) {
  if (dst != src) {
    movq(dst, src);
  }
}

void MacroAssembler::Move(Register dst, Operand src) { movq(dst, src); }
void MacroAssembler::Move(Register dst, Immediate src) {
  if (src.rmode() == RelocInfo::Mode::NO_INFO) {
    Move(dst, src.value());
  } else {
    movl(dst, src);
  }
}

void MacroAssembler::Move(XMMRegister dst, XMMRegister src) {
  if (dst != src) {
    Movaps(dst, src);
  }
}

void MacroAssembler::MovePair(Register dst0, Register src0, Register dst1,
                              Register src1) {
  if (dst0 != src1) {
    // Normal case: Writing to dst0 does not destroy src1.
    Move(dst0, src0);
    Move(dst1, src1);
  } else if (dst1 != src0) {
    // Only dst0 and src1 are the same register,
    // but writing to dst1 does not destroy src0.
    Move(dst1, src1);
    Move(dst0, src0);
  } else {
    // dst0 == src1, and dst1 == src0, a swap is required:
    // dst0 \/ src0
    // dst1 /\ src1
    xchgq(dst0, dst1);
  }
}

void MacroAssembler::MoveNumber(Register dst, double value) {
  int32_t smi;
  if (DoubleToSmiInteger(value, &smi)) {
    Move(dst, Smi::FromInt(smi));
  } else {
    movq_heap_number(dst, value);
  }
}

void MacroAssembler::Move(XMMRegister dst, uint32_t src) {
  if (src == 0) {
    Xorps(dst, dst);
  } else {
    unsigned nlz = base::bits::CountLeadingZeros(src);
    unsigned ntz = base::bits::CountTrailingZeros(src);
    unsigned pop = base::bits::CountPopulation(src);
    DCHECK_NE(0u, pop);
    if (pop + ntz + nlz == 32) {
      Pcmpeqd(dst, dst);
      if (ntz) Pslld(dst, static_cast<uint8_t>(ntz + nlz));
      if (nlz) Psrld(dst, static_cast<uint8_t>(nlz));
    } else {
      movl(kScratchRegister, Immediate(src));
      Movd(dst, kScratchRegister);
    }
  }
}

void MacroAssembler::Move(XMMRegister dst, uint64_t src) {
  if (src == 0) {
    Xorpd(dst, dst);
  } else {
    unsigned nlz = base::bits::CountLeadingZeros(src);
    unsigned ntz = base::bits::CountTrailingZeros(src);
    unsigned pop = base::bits::CountPopulation(src);
    DCHECK_NE(0u, pop);
    if (pop + ntz + nlz == 64) {
      Pcmpeqd(dst, dst);
      if (ntz) Psllq(dst, static_cast<uint8_t>(ntz + nlz));
      if (nlz) Psrlq(dst, static_cast<uint8_t>(nlz));
    } else {
      uint32_t lower = static_cast<uint32_t>(src);
      uint32_t upper = static_cast<uint32_t>(src >> 32);
      if (upper == 0) {
        Move(dst, lower);
      } else {
        movq(kScratchRegister, src);
        Movq(dst, kScratchRegister);
      }
    }
  }
}

void MacroAssembler::Move(XMMRegister dst, uint64_t high, uint64_t low) {
  if (high == low) {
    Move(dst, low);
    Punpcklqdq(dst, dst);
    return;
  }

  Move(dst, low);
  movq(kScratchRegister, high);
  Pinsrq(dst, dst, kScratchRegister, uint8_t{1});
}

// ----------------------------------------------------------------------------

void MacroAssembler::Cmp(Register dst, Handle<Object> source) {
  if (IsSmi(*source)) {
    Cmp(dst, Cast<Smi>(*source));
  } else if (root_array_available_ && options().isolate_independent_code) {
    // TODO(jgruber,v8:8887): Also consider a root-relative load when generating
    // non-isolate-independent code. In many cases it might be cheaper than
    // embedding the relocatable value.
    // TODO(v8:9706): Fix-it! This load will always uncompress the value
    // even when we are loading a compressed embedded object.
    IndirectLoadConstant(kScratchRegister, Cast<HeapObject>(source));
    cmp_tagged(dst, kScratchRegister);
  } else if (COMPRESS_POINTERS_BOOL) {
    EmbeddedObjectIndex index = AddEmbeddedObject(Cast<HeapObject>(source));
    DCHECK(is_uint32(index));
    cmpl(dst, Immediate(static_cast<int>(index),
                        RelocInfo::COMPRESSED_EMBEDDED_OBJECT));
  } else {
    movq(kScratchRegister,
         Immediate64(source.address(), RelocInfo::FULL_EMBEDDED_OBJECT));
    cmpq(dst, kScratchRegister);
  }
}

void MacroAssembler::Cmp(Operand dst, Handle<Object> source) {
  if (IsSmi(*source)) {
    Cmp(dst, Cast<Smi>(*source));
  } else if (root_array_available_ && options().isolate_independent_code) {
    // TODO(jgruber,v8:8887): Also consider a root-relative load when generating
    // non-isolate-independent code. In many cases it might be cheaper than
    // embedding the relocatable value.
    // TODO(v8:9706): Fix-it! This load will always uncompress the value
    // even when we are loading a compressed embedded object.
    IndirectLoadConstant(kScratchRegister, Cast<HeapObject>(source));
    cmp_tagged(dst, kScratchRegister);
  } else if (COMPRESS_POINTERS_BOOL) {
    EmbeddedObjectIndex index = AddEmbeddedObject(Cast<HeapObject>(source));
    DCHECK(is_uint32(index));
    cmpl(dst, Immediate(static_cast<int>(index),
                        RelocInfo::COMPRESSED_EMBEDDED_OBJECT));
  } else {
    Move(kScratchRegister, Cast<HeapObject>(source),
         RelocInfo::FULL_EMBEDDED_OBJECT);
    cmp_tagged(dst, kScratchRegister);
  }
}

void MacroAssembler::CompareRange(Register value, unsigned lower_limit,
                                  unsigned higher_limit) {
  ASM_CODE_COMMENT(this);
  DCHECK_LT(lower_limit, higher_limit);
  if (lower_limit != 0) {
    leal(kScratchRegister, Operand(value, 0u - lower_limit));
    cmpl(kScratchRegister, Immediate(higher_limit - lower_limit));
  } else {
    cmpl(value, Immediate(higher_limit));
  }
}

void MacroAssembler::JumpIfIsInRange(Register value, unsigned lower_limit,
                                     unsigned higher_limit, Label* on_in_range,
                                     Label::Distance near_jump) {
  CompareRange(value, lower_limit, higher_limit);
  j(below_equal, on_in_range, near_jump);
}

void MacroAssembler::Push(Handle<HeapObject> source) {
  Move(kScratchRegister, source);
  Push(kScratchRegister);
}

void MacroAssembler::PushArray(Register array, Register size, Register scratch,
                               PushArrayOrder order) {
  DCHECK(!AreAliased(array, size, scratch));
  Register counter = scratch;
  Label loop, entry;
  if (order == PushArrayOrder::kReverse) {
    Move(counter, 0);
    jmp(&entry);
    bind(&loop);
    Push(Operand(array, counter, times_system_pointer_size, 0));
    incq(counter);
    bind(&entry);
    cmpq(counter, size);
    j(less, &loop, Label::kNear);
  } else {
    movq(counter, size);
    jmp(&entry);
    bind(&loop);
    Push(Operand(array, counter, times_system_pointer_size, 0));
    bind(&entry);
    decq(counter);
    j(greater_equal, &loop, Label::kNear);
  }
}

void MacroAssembler::Move(Register result, Handle<HeapObject> object,
                          RelocInfo::Mode rmode) {
  // TODO(jgruber,v8:8887): Also consider a root-relative load when generating
  // non-isolate-independent code. In many cases it might be cheaper than
  // embedding the relocatable value.
  if (root_array_available_ && options().isolate_independent_code) {
    // TODO(v8:9706): Fix-it! This load will always uncompress the value
    // even when we are loading a compressed embedded object.
    IndirectLoadConstant(result, object);
  } else if (RelocInfo::IsCompressedEmbeddedObject(rmode)) {
    EmbeddedObjectIndex index = AddEmbeddedObject(object);
    DCHECK(is_uint32(index));
    movl(result, Immediate(static_cast<int>(index), rmode));
  } else {
    DCHECK(RelocInfo::IsFullEmbeddedObject(rmode));
    movq(result, Immediate64(object.address(), rmode));
  }
}

void MacroAssembler::Move(Operand dst, Handle<HeapObject> object,
                          RelocInfo::Mode rmode) {
  Move(kScratchRegister, object, rmode);
  movq(dst, kScratchRegister);
}

void MacroAssembler::Drop(int stack_elements) {
  if (stack_elements > 0) {
    addq(rsp, Immediate(stack_elements * kSystemPointerSize));
  }
}

void MacroAssembler::DropUnderReturnAddress(int stack_elements,
                                            Register scratch) {
  DCHECK_GT(stack_elements, 0);
  if (stack_elements == 1) {
    popq(MemOperand(rsp, 0));
    return;
  }

  PopReturnAddressTo(scratch);
  Drop(stack_elements);
  PushReturnAddressFrom(scratch);
}

void MacroAssembler::DropArguments(Register count) {
  leaq(rsp, Operand(rsp, count, times_system_pointer_size, 0));
}

void MacroAssembler::DropArguments(Register count, Register scratch) {
  DCHECK(!AreAliased(count, scratch));
  PopReturnAddressTo(scratch);
  DropArguments(count);
  PushReturnAddressFrom(scratch);
}

void MacroAssembler::DropArgumentsAndPushNewReceiver(Register argc,
                                                     Register receiver,
                                                     Register scratch) {
  DCHECK(!AreAliased(argc, receiver, scratch));
  PopReturnAddressTo(scratch);
  DropArguments(argc);
  Push(receiver);
  PushReturnAddressFrom(scratch);
}

void MacroAssembler::DropArgumentsAndPushNewReceiver(Register argc,
                                                     Operand receiver,
                                                     Register scratch) {
  DCHECK(!AreAliased(argc, scratch));
  DCHECK(!receiver.AddressUsesRegister(scratch));
  PopReturnAddressTo(scratch);
  DropArguments(argc);
  Push(receiver);
  PushReturnAddressFrom(scratch);
}

void MacroAssembler::Push(Register src) { pushq(src); }

void MacroAssembler::Push(Operand src) { pushq(src); }

void MacroAssembler::PushQuad(Operand src) { pushq(src); }

void MacroAssembler::Push(Immediate value) { pushq(value); }

void MacroAssembler::PushImm32(int32_t imm32) { pushq_imm32(imm32); }

void MacroAssembler::Pop(Register dst) { popq(dst); }

void MacroAssembler::Pop(Operand dst) { popq(dst); }

void MacroAssembler::PopQuad(Operand dst) { popq(dst); }

void MacroAssembler::Jump(const ExternalReference& reference) {
  DCHECK(root_array_available());
  jmp(Operand(kRootRegister, RootRegisterOffsetForExternalReferenceTableEntry(
                                 isolate(), reference)));
}

void MacroAssembler::Jump(Operand op) { jmp(op); }

void MacroAssembler::Jump(Operand op, Condition cc) {
  Label skip;
  j(NegateCondition(cc), &skip, Label::kNear);
  Jump(op);
  bind(&skip);
}

void MacroAssembler::Jump(Address destination, RelocInfo::Mode rmode) {
  Move(kScratchRegister, destination, rmode);
  jmp(kScratchRegister);
}

void MacroAssembler::Jump(Address destination, RelocInfo::Mode rmode,
                          Condition cc) {
  Label skip;
  j(NegateCondition(cc), &skip, Label::kNear);
  Jump(destination, rmode);
  bind(&skip);
}

void MacroAssembler::Jump(Handle<Code> code_object, RelocInfo::Mode rmode) {
  DCHECK_EQ(sandboxing_mode(), code_object->sandboxing_mode());
  DCHECK_IMPLIES(options().isolate_independent_code,
                 Builtins::IsIsolateIndependentBuiltin(*code_object));
  Builtin builtin = Builtin::kNoBuiltinId;
  if (isolate()->builtins()->IsBuiltinHandle(code_object, &builtin)) {
    TailCallBuiltin(builtin);
    return;
  }
  DCHECK(RelocInfo::IsCodeTarget(rmode));
  jmp(code_object, rmode);
}

void MacroAssembler::Jump(Handle<Code> code_object, RelocInfo::Mode rmode,
                          Condition cc) {
  DCHECK_EQ(sandboxing_mode(), code_object->sandboxing_mode());
  DCHECK_IMPLIES(options().isolate_independent_code,
                 Builtins::IsIsolateIndependentBuiltin(*code_object));
  Builtin builtin = Builtin::kNoBuiltinId;
  if (isolate()->builtins()->IsBuiltinHandle(code_object, &builtin)) {
    TailCallBuiltin(builtin, cc);
    return;
  }
  DCHECK(RelocInfo::IsCodeTarget(rmode));
  j(cc, code_object, rmode);
}

void MacroAssembler::Call(ExternalReference ext) {
  // TODO(350324877): can we DCHECK that the sandboxing mode is correct here?
  LoadAddress(kScratchRegister, ext);
  call(kScratchRegister);
}

void MacroAssembler::Call(Operand op) {
  // TODO(350324877): can we DCHECK that the sandboxing mode is correct here?
  if (!CpuFeatures::IsSupported(INTEL_ATOM)) {
    call(op);
  } else {
    movq(kScratchRegister, op);
    call(kScratchRegister);
  }
}

void MacroAssembler::Call(Address destination, RelocInfo::Mode rmode) {
  Move(kScratchRegister, destination, rmode);
  call(kScratchRegister);
}

void MacroAssembler::Call(Handle<Code> code_object, RelocInfo::Mode rmode) {
  DCHECK_IMPLIES(options().isolate_independent_code,
                 Builtins::IsIsolateIndependentBuiltin(*code_object));
  Builtin builtin = Builtin::kNoBuiltinId;
  if (isolate()->builtins()->IsBuiltinHandle(code_object, &builtin)) {
    CallBuiltin(builtin);
    return;
  }
  DCHECK_EQ(sandboxing_mode(), code_object->sandboxing_mode());
  DCHECK(RelocInfo::IsCodeTarget(rmode));
  call(code_object, rmode);
}

Operand MacroAssembler::EntryFromBuiltinAsOperand(Builtin builtin) {
  DCHECK(root_array_available());
  return Operand(kRootRegister, IsolateData::BuiltinEntrySlotOffset(builtin));
}

Operand MacroAssembler::EntryFromBuiltinIndexAsOperand(Register builtin_index) {
  if (SmiValuesAre32Bits()) {
    // The builtin_index register contains the builtin index as a Smi.
    Move(kScratchRegister, builtin_index);  // Callee checks for equality.
    SmiUntagUnsigned(kScratchRegister);
    return Operand(kRootRegister, kScratchRegister, times_system_pointer_size,
                   IsolateData::builtin_entry_table_offset());
  } else {
    DCHECK(SmiValuesAre31Bits());

    // The builtin_index register contains the builtin index as a Smi.
    // Untagging is folded into the indexing operand below (we use
    // times_half_system_pointer_size since smis are already shifted by one).
    return Operand(kRootRegister, builtin_index, times_half_system_pointer_size,
                   IsolateData::builtin_entry_table_offset());
  }
}

void MacroAssembler::CallBuiltinByIndex(Register builtin_index) {
  // TODO(350324877): can we DCHECK that the sandboxing mode is correct here?
  Call(EntryFromBuiltinIndexAsOperand(builtin_index));
}

void MacroAssembler::CallBuiltin(Builtin builtin) {
  ASM_CODE_COMMENT_STRING(this, CommentForOffHeapTrampoline("call", builtin));

  // Check if this builtin call transitions out of sandboxed execution mode.
  CodeSandboxingMode previous_mode = SwitchSandboxingModeBeforeCallIfNeeded(
      Builtins::SandboxingModeOf(builtin));

  switch (options().builtin_call_jump_mode) {
    case BuiltinCallJumpMode::kAbsolute:
      Call(BuiltinEntry(builtin), RelocInfo::OFF_HEAP_TARGET);
      break;
    case BuiltinCallJumpMode::kPCRelative:
      near_call(static_cast<intptr_t>(builtin), RelocInfo::NEAR_BUILTIN_ENTRY);
      break;
    case BuiltinCallJumpMode::kIndirect:
      Call(EntryFromBuiltinAsOperand(builtin));
      break;
    case BuiltinCallJumpMode::kForMksnapshot: {
      Handle<Code> code = isolate()->builtins()->code_handle(builtin);
      call(code, RelocInfo::CODE_TARGET);
      break;
    }
  }

  SwitchSandboxingModeAfterCallIfNeeded(previous_mode);
}

void MacroAssembler::TailCallBuiltin(Builtin builtin) {
  // We cannot (currently) switch the sandboxing mode on tail calls.
  DCHECK_EQ(sandboxing_mode(), Builtins::SandboxingModeOf(builtin));

  ASM_CODE_COMMENT_STRING(this,
                          CommentForOffHeapTrampoline("tail call", builtin));
  switch (options().builtin_call_jump_mode) {
    case BuiltinCallJumpMode::kAbsolute:
      Jump(BuiltinEntry(builtin), RelocInfo::OFF_HEAP_TARGET);
      break;
    case BuiltinCallJumpMode::kPCRelative:
      near_jmp(static_cast<intptr_t>(builtin), RelocInfo::NEAR_BUILTIN_ENTRY);
      break;
    case BuiltinCallJumpMode::kIndirect:
      Jump(EntryFromBuiltinAsOperand(builtin));
      break;
    case BuiltinCallJumpMode::kForMksnapshot: {
      Handle<Code> code = isolate()->builtins()->code_handle(builtin);
      jmp(code, RelocInfo::CODE_TARGET);
      break;
    }
  }
}

void MacroAssembler::TailCallBuiltin(Builtin builtin, Condition cc) {
  // We cannot (currently) switch the sandboxing mode on tail calls.
  DCHECK_EQ(sandboxing_mode(), Builtins::SandboxingModeOf(builtin));

  ASM_CODE_COMMENT_STRING(this,
                          CommentForOffHeapTrampoline("tail call", builtin));
  switch (options().builtin_call_jump_mode) {
    case BuiltinCallJumpMode::kAbsolute:
      Jump(BuiltinEntry(builtin), RelocInfo::OFF_HEAP_TARGET, cc);
      break;
    case BuiltinCallJumpMode::kPCRelative:
      near_j(cc, static_cast<intptr_t>(builtin), RelocInfo::NEAR_BUILTIN_ENTRY);
      break;
    case BuiltinCallJumpMode::kIndirect:
      Jump(EntryFromBuiltinAsOperand(builtin), cc);
      break;
    case BuiltinCallJumpMode::kForMksnapshot: {
      Handle<Code> code = isolate()->builtins()->code_handle(builtin);
      j(cc, code, RelocInfo::CODE_TARGET);
      break;
    }
  }
}

void MacroAssembler::LoadCodeInstructionStart(Register destination,
                                              Register code_object,
                                              CodeEntrypointTag tag) {
  ASM_CODE_COMMENT(this);
#ifdef V8_ENABLE_SANDBOX
  LoadCodeEntrypointViaCodePointer(
      destination, FieldOperand(code_object, Code::kSelfIndirectPointerOffset),
      tag);
#else
  movq(destination, FieldOperand(code_object, Code::kInstructionStartOffset));
#endif
}

void MacroAssembler::CallCodeObject(Register code_object,
                                    CodeEntrypointTag tag) {
  LoadCodeInstructionStart(code_object, code_object, tag);
  call(code_object);
}

void MacroAssembler::JumpCodeObject(Register code_object, CodeEntrypointTag tag,
                                    JumpMode jump_mode) {
  // TODO(saelo): can we avoid using this for JavaScript functions
  // (kJSEntrypointTag) and instead use a variant that ensures that the caller
  // and callee agree on the signature (i.e. parameter count)?
  LoadCodeInstructionStart(code_object, code_object, tag);
  switch (jump_mode) {
    case JumpMode::kJump:
      jmp(code_object);
      return;
    case JumpMode::kPushAndReturn:
      pushq(code_object);
      Ret();
      return;
  }
}

void MacroAssembler::CallJSFunction(Register function_object,
                                    uint16_t argument_count) {
  static_assert(kJavaScriptCallCodeStartRegister == rcx, "ABI mismatch");
  static_assert(kJavaScriptCallDispatchHandleRegister == r15, "ABI mismatch");
  movl(r15, FieldOperand(function_object, JSFunction::kDispatchHandleOffset));
  LoadEntrypointAndParameterCountFromJSDispatchTable(rcx, rbx, r15);
  // Force a safe crash if the parameter count doesn't match.
  // TODO(412398354): to avoid this runtime check, we should switch all
  // remaining users to call the function via its dispatch handle instead. See
  // CallJSDispatchEntry below and crbug.com/412398354 for more details.
  cmpl(rbx, Immediate(argument_count));
  SbxCheck(less_equal, AbortReason::kJSSignatureMismatch);
  call(rcx);
}

void MacroAssembler::CallJSDispatchEntry(JSDispatchHandle dispatch_handle,
                                         uint16_t argument_count) {
  static_assert(kJavaScriptCallCodeStartRegister == rcx, "ABI mismatch");
  static_assert(kJavaScriptCallDispatchHandleRegister == r15, "ABI mismatch");
  movl(kJavaScriptCallDispatchHandleRegister,
       Immediate(dispatch_handle.value(), RelocInfo::JS_DISPATCH_HANDLE));
  LoadEntrypointFromJSDispatchTable(rcx, kJavaScriptCallDispatchHandleRegister);
  CHECK_EQ(argument_count,
           isolate()->js_dispatch_table().GetParameterCount(dispatch_handle));
  call(rcx);
}

void MacroAssembler::JumpJSFunction(Register function_object,
                                    JumpMode jump_mode) {
  CHECK(!V8_ENABLE_SANDBOX_BOOL);
  // This implementation is not currently used because callers usually need
  // to load both entry point and parameter count and then do something with
  // the latter before the actual call.
  UNREACHABLE();
}

#ifdef V8_ENABLE_WEBASSEMBLY

void MacroAssembler::CallWasmCodePointer(Register target,
                                         uint64_t signature_hash,
                                         CallJumpMode call_jump_mode) {
  ASM_CODE_COMMENT(this);
  Move(kScratchRegister, ExternalReference::wasm_code_pointer_table());

#ifdef V8_ENABLE_SANDBOX
  // Execute a left shift followed by right shift to achieve two things:
  // - Only keep `kNumRelevantBits` bits (to avoid OOB access to the table),
  // - shift by `kLeftShift` to translate from index to offset into the table.
  static constexpr int kLeftShift =
      base::bits::WhichPowerOfTwo(sizeof(wasm::WasmCodePointerTableEntry));
  static constexpr int kNumRelevantBits =
      base::bits::WhichPowerOfTwo(WasmCodePointer::kIndexSpaceSize);
  static constexpr int kNumClearedHighBits = 32 - kNumRelevantBits;
  shll(target, Immediate(kNumClearedHighBits));
  shrl(target, Immediate(kNumClearedHighBits - kLeftShift));

  // Add `target` and `kScratchRegister` early to free `kScratchRegister` again.
  addq(target, kScratchRegister);

  Operand signature_hash_op{target,
                            wasm::WasmCodePointerTable::kOffsetOfSignatureHash};
  if (is_int32(signature_hash)) {
    // cmpq sign-extends the 32-bit immediate.
    cmpq(signature_hash_op, Immediate(static_cast<int32_t>(signature_hash)));
  } else {
    Move(kScratchRegister, signature_hash);
    cmpq(kScratchRegister, signature_hash_op);
  }
  Label fail, ok;
  j(Condition::kNotEqual, &fail, Label::Distance::kNear);
  jmp(&ok, Label::Distance::kNear);

  bind(&fail);
  Abort(AbortReason::kWasmSignatureMismatch);

  bind(&ok);
  Operand target_op{target, 0};
#else
  static_assert(sizeof(wasm::WasmCodePointerTableEntry) == 8);
  Operand target_op{kScratchRegister, target, ScaleFactor::times_8, 0};
#endif

  if (call_jump_mode == CallJumpMode::kTailCall) {
    jmp(target_op);
  } else {
    call(target_op);
  }
}

void MacroAssembler::CallWasmCodePointerNoSignatureCheck(Register target) {
  Move(kScratchRegister, ExternalReference::wasm_code_pointer_table());

#ifdef V8_ENABLE_SANDBOX
  // Execute a left shift followed by right shift to achieve two things:
  // - Only keep `kNumRelevantBits` bits (to avoid OOB access to the table),
  // - shift by `kLeftShift` to translate from index to offset into the table.
  static constexpr int kLeftShift =
      base::bits::WhichPowerOfTwo(sizeof(wasm::WasmCodePointerTableEntry));
  static constexpr int kNumRelevantBits =
      base::bits::WhichPowerOfTwo(WasmCodePointer::kIndexSpaceSize);
  static constexpr int kNumClearedHighBits = 32 - kNumRelevantBits;
  static_assert(kNumClearedHighBits == 9);
  shll(target, Immediate(kNumClearedHighBits));
  shrl(target, Immediate(kNumClearedHighBits - kLeftShift));

  call(Operand(kScratchRegister, target, ScaleFactor::times_1, 0));
#else
  static_assert(sizeof(wasm::WasmCodePointerTableEntry) == 8);
  call(Operand(kScratchRegister, target, ScaleFactor::times_8, 0));
#endif
}

void MacroAssembler::LoadWasmCodePointer(Register dst, Operand src) {
  static_assert(sizeof(WasmCodePointer) == 4);
  movl(dst, src);
}

#endif

void MacroAssembler::PextrdPreSse41(Register dst, XMMRegister src,
                                    uint8_t imm8) {
  if (imm8 == 0) {
    Movd(dst, src);
    return;
  }
  DCHECK_EQ(1, imm8);
  movq(dst, src);
  shrq(dst, Immediate(32));
}

namespace {
template <typename Op>
void PinsrdPreSse41Helper(MacroAssembler* masm, XMMRegister dst, Op src,
                          uint8_t imm8, uint32_t* load_pc_offset) {
  masm->Movd(kScratchDoubleReg, src);
  if (load_pc_offset) *load_pc_offset = masm->pc_offset();
  if (imm8 == 1) {
    masm->punpckldq(dst, kScratchDoubleReg);
  } else {
    DCHECK_EQ(0, imm8);
    masm->Movss(dst, kScratchDoubleReg);
  }
}
}  // namespace

void MacroAssembler::PinsrdPreSse41(XMMRegister dst, Register src, uint8_t imm8,
                                    uint32_t* load_pc_offset) {
  PinsrdPreSse41Helper(this, dst, src, imm8, load_pc_offset);
}

void MacroAssembler::PinsrdPreSse41(XMMRegister dst, Operand src, uint8_t imm8,
                                    uint32_t* load_pc_offset) {
  PinsrdPreSse41Helper(this, dst, src, imm8, load_pc_offset);
}

void MacroAssembler::Pinsrq(XMMRegister dst, XMMRegister src1, Register src2,
                            uint8_t imm8, uint32_t* load_pc_offset) {
  PinsrHelper(this, &Assembler::vpinsrq, &Assembler::pinsrq, dst, src1, src2,
              imm8, load_pc_offset, {SSE4_1});
}

void MacroAssembler::Pinsrq(XMMRegister dst, XMMRegister src1, Operand src2,
                            uint8_t imm8, uint32_t* load_pc_offset) {
  PinsrHelper(this, &Assembler::vpinsrq, &Assembler::pinsrq, dst, src1, src2,
              imm8, load_pc_offset, {SSE4_1});
}

void MacroAssembler::Lzcntl(Register dst, Register src) {
  if (CpuFeatures::IsSupported(LZCNT)) {
    CpuFeatureScope scope(this, LZCNT);
    lzcntl(dst, src);
    return;
  }
  Label not_zero_src;
  bsrl(dst, src);
  j(not_zero, &not_zero_src, Label::kNear);
  Move(dst, 63);  // 63^31 == 32
  bind(&not_zero_src);
  xorl(dst, Immediate(31));  // for x in [0..31], 31^x == 31 - x
}

void MacroAssembler::Lzcntl(Register dst, Operand src) {
  if (CpuFeatures::IsSupported(LZCNT)) {
    CpuFeatureScope scope(this, LZCNT);
    lzcntl(dst, src);
    return;
  }
  Label not_zero_src;
  bsrl(dst, src);
  j(not_zero, &not_zero_src, Label::kNear);
  Move(dst, 63);  // 63^31 == 32
  bind(&not_zero_src);
  xorl(dst, Immediate(31));  // for x in [0..31], 31^x == 31 - x
}

void MacroAssembler::Lzcntq(Register dst, Register src) {
  if (CpuFeatures::IsSupported(LZCNT)) {
    CpuFeatureScope scope(this, LZCNT);
    lzcntq(dst, src);
    return;
  }
  Label not_zero_src;
  bsrq(dst, src);
  j(not_zero, &not_zero_src, Label::kNear);
  Move(dst, 127);  // 127^63 == 64
  bind(&not_zero_src);
  xorl(dst, Immediate(63));  // for x in [0..63], 63^x == 63 - x
}

void MacroAssembler::Lzcntq(Register dst, Operand src) {
  if (CpuFeatures::IsSupported(LZCNT)) {
    CpuFeatureScope scope(this, LZCNT);
    lzcntq(dst, src);
    return;
  }
  Label not_zero_src;
  bsrq(dst, src);
  j(not_zero, &not_zero_src, Label::kNear);
  Move(dst, 127);  // 127^63 == 64
  bind(&not_zero_src);
  xorl(dst, Immediate(63));  // for x in [0..63], 63^x == 63 - x
}

void MacroAssembler::Tzcntq(Register dst, Register src) {
  if (CpuFeatures::IsSupported(BMI1)) {
    CpuFeatureScope scope(this, BMI1);
    tzcntq(dst, src);
    return;
  }
  Label not_zero_src;
  bsfq(dst, src);
  j(not_zero, &not_zero_src, Label::kNear);
  // Define the result of tzcnt(0) separately, because bsf(0) is undefined.
  Move(dst, 64);
  bind(&not_zero_src);
}

void MacroAssembler::Tzcntq(Register dst, Operand src) {
  if (CpuFeatures::IsSupported(BMI1)) {
    CpuFeatureScope scope(this, BMI1);
    tzcntq(dst, src);
    return;
  }
  Label not_zero_src;
  bsfq(dst, src);
  j(not_zero, &not_zero_src, Label::kNear);
  // Define the result of tzcnt(0) separately, because bsf(0) is undefined.
  Move(dst, 64);
  bind(&not_zero_src);
}

void MacroAssembler::Tzcntl(Register dst, Register src) {
  if (CpuFeatures::IsSupported(BMI1)) {
    CpuFeatureScope scope(this, BMI1);
    tzcntl(dst, src);
    return;
  }
  Label not_zero_src;
  bsfl(dst, src);
  j(not_zero, &not_zero_src, Label::kNear);
  Move(dst, 32);  // The result of tzcnt is 32 if src = 0.
  bind(&not_zero_src);
}

void MacroAssembler::Tzcntl(Register dst, Operand src) {
  if (CpuFeatures::IsSupported(BMI1)) {
    CpuFeatureScope scope(this, BMI1);
    tzcntl(dst, src);
    return;
  }
  Label not_zero_src;
  bsfl(dst, src);
  j(not_zero, &not_zero_src, Label::kNear);
  Move(dst, 32);  // The result of tzcnt is 32 if src = 0.
  bind(&not_zero_src);
}

void MacroAssembler::Popcntl(Register dst, Register src) {
  if (CpuFeatures::IsSupported(POPCNT)) {
    CpuFeatureScope scope(this, POPCNT);
    popcntl(dst, src);
    return;
  }
  UNREACHABLE();
}

void MacroAssembler::Popcntl(Register dst, Operand src) {
  if (CpuFeatures::IsSupported(POPCNT)) {
    CpuFeatureScope scope(this, POPCNT);
    popcntl(dst, src);
    return;
  }
  UNREACHABLE();
}

void MacroAssembler::Popcntq(Register dst, Register src) {
  if (CpuFeatures::IsSupported(POPCNT)) {
    CpuFeatureScope scope(this, POPCNT);
    popcntq(dst, src);
    return;
  }
  UNREACHABLE();
}

void MacroAssembler::Popcntq(Register dst, Operand src) {
  if (CpuFeatures::IsSupported(POPCNT)) {
    CpuFeatureScope scope(this, POPCNT);
    popcntq(dst, src);
    return;
  }
  UNREACHABLE();
}

void MacroAssembler::PushStackHandler() {
  // Adjust this code if not the case.
  static_assert(StackHandlerConstants::kSize == 2 * kSystemPointerSize);
  static_assert(StackHandlerConstants::kNextOffset == 0);

  Push(Immediate(0));  // Padding.

  // Link the current handler as the next handler.
  Push(AsMemOperand(IsolateFieldId::kHandler));

  // Set this new handler as the current one.
  movq(AsMemOperand(IsolateFieldId::kHandler), rsp);
}

void MacroAssembler::PopStackHandler() {
  static_assert(StackHandlerConstants::kNextOffset == 0);
  Pop(AsMemOperand(IsolateFieldId::kHandler));
  addq(rsp, Immediate(StackHandlerConstants::kSize - kSystemPointerSize));
}

void MacroAssembler::Ret() { ret(0); }

void MacroAssembler::Ret(int bytes_dropped, Register scratch) {
  if (is_uint16(bytes_dropped)) {
    ret(bytes_dropped);
  } else {
    PopReturnAddressTo(scratch);
    addq(rsp, Immediate(bytes_dropped));
    // Push and ret (instead of jmp) to keep the RSB and the CET shadow stack
    // balanced.
    PushReturnAddressFrom(scratch);
    ret(0);
  }
}

void MacroAssembler::IncsspqIfSupported(Register number_of_words,
                                        Register scratch) {
  // Optimized code can validate at runtime whether the cpu supports the
  // incsspq instruction, so it shouldn't use this method.
  CHECK(isolate()->IsGeneratingEmbeddedBuiltins());
  DCHECK_NE(number_of_words, scratch);
  Label not_supported;
  ExternalReference supports_cetss =
      ExternalReference::supports_cetss_address();
  Operand supports_cetss_operand =
      ExternalReferenceAsOperand(supports_cetss, scratch);
  cmpb(supports_cetss_operand, Immediate(0));
  j(equal, &not_supported, Label::kNear);
  incsspq(number_of_words);
  bind(&not_supported);
}

#if V8_STATIC_ROOTS_BOOL
void MacroAssembler::CompareInstanceTypeWithUniqueCompressedMap(
    Register map, InstanceType type) {
  std::optional<RootIndex> expected =
      InstanceTypeChecker::UniqueMapOfInstanceType(type);
  CHECK(expected);
  Tagged_t expected_ptr = ReadOnlyRootPtr(*expected);
  cmp_tagged(map, Immediate(expected_ptr));
}

void MacroAssembler::IsObjectTypeFast(Register object, InstanceType type,
                                      Register compressed_map_scratch) {
  ASM_CODE_COMMENT(this);
  CHECK(InstanceTypeChecker::UniqueMapOfInstanceType(type));
  LoadCompressedMap(compressed_map_scratch, object);
  CompareInstanceTypeWithUniqueCompressedMap(compressed_map_scratch, type);
}
#endif  // V8_STATIC_ROOTS_BOOL

void MacroAssembler::IsObjectType(Register heap_object, InstanceType type,
                                  Register map) {
#if V8_STATIC_ROOTS_BOOL
  if (InstanceTypeChecker::UniqueMapOfInstanceType(type)) {
    LoadCompressedMap(map, heap_object);
    CompareInstanceTypeWithUniqueCompressedMap(map, type);
    return;
  }
#endif  // V8_STATIC_ROOTS_BOOL
  CmpObjectType(heap_object, type, map);
}

void MacroAssembler::IsObjectTypeInRange(Register heap_object,
                                         InstanceType lower_limit,
                                         InstanceType higher_limit,
                                         Register scratch) {
  DCHECK_LT(lower_limit, higher_limit);
#if V8_STATIC_ROOTS_BOOL
  if (auto range = InstanceTypeChecker::UniqueMapRangeOfInstanceTypeRange(
          lower_limit, higher_limit)) {
    LoadCompressedMap(scratch, heap_object);
    CompareRange(scratch, range->first, range->second);
    return;
  }
#endif  // V8_STATIC_ROOTS_BOOL
  LoadMap(scratch, heap_object);
  CmpInstanceTypeRange(scratch, scratch, lower_limit, higher_limit);
}

void MacroAssembler::JumpIfJSAnyIsNotPrimitive(Register heap_object,
                                               Register scratch, Label* target,
                                               Label::Distance distance,
                                               Condition cc) {
  CHECK(cc == Condition::kUnsignedLessThan ||
        cc == Condition::kUnsignedGreaterThanEqual);
  if (V8_STATIC_ROOTS_BOOL) {
#ifdef DEBUG
    Label ok;
    LoadMap(scratch, heap_object);
    CmpInstanceTypeRange(scratch, scratch, FIRST_JS_RECEIVER_TYPE,
                         LAST_JS_RECEIVER_TYPE);
    j(Condition::kUnsignedLessThanEqual, &ok, Label::Distance::kNear);
    LoadMap(scratch, heap_object);
    CmpInstanceTypeRange(scratch, scratch, FIRST_PRIMITIVE_HEAP_OBJECT_TYPE,
                         LAST_PRIMITIVE_HEAP_OBJECT_TYPE);
    j(Condition::kUnsignedLessThanEqual, &ok, Label::Distance::kNear);
    Abort(AbortReason::kInvalidReceiver);
    bind(&ok);
#endif  // DEBUG

    // All primitive object's maps are allocated at the start of the read only
    // heap. Thus JS_RECEIVER's must have maps with larger (compressed)
    // addresses.
    LoadCompressedMap(scratch, heap_object);
    cmp_tagged(scratch, Immediate(InstanceTypeChecker::kNonJsReceiverMapLimit));
  } else {
    static_assert(LAST_JS_RECEIVER_TYPE == LAST_TYPE);
    CmpObjectType(heap_object, FIRST_JS_RECEIVER_TYPE, scratch);
  }
  j(cc, target, distance);
}

void MacroAssembler::CmpObjectType(Register heap_object, InstanceType type,
                                   Register map) {
  LoadMap(map, heap_object);
  CmpInstanceType(map, type);
}

void MacroAssembler::CmpInstanceType(Register map, InstanceType type) {
  cmpw(FieldOperand(map, Map::kInstanceTypeOffset), Immediate(type));
}

void MacroAssembler::CmpInstanceTypeRange(Register map,
                                          Register instance_type_out,
                                          InstanceType lower_limit,
                                          InstanceType higher_limit) {
  DCHECK_LT(lower_limit, higher_limit);
  movzxwl(instance_type_out, FieldOperand(map, Map::kInstanceTypeOffset));
  CompareRange(instance_type_out, lower_limit, higher_limit);
}

void MacroAssembler::TestCodeIsMarkedForDeoptimization(Register code) {
  const int kByteWithDeoptBitOffset = 0 * kByteSize;
  const int kByteWithDeoptBitOffsetInBits = kByteWithDeoptBitOffset * 8;
  static_assert(V8_TARGET_LITTLE_ENDIAN == 1);
  static_assert(FIELD_SIZE(Code::kFlagsOffset) * kBitsPerByte == 32);
  static_assert(Code::kMarkedForDeoptimizationBit >
                kByteWithDeoptBitOffsetInBits);
  testb(FieldOperand(code, Code::kFlagsOffset + kByteWithDeoptBitOffset),
        Immediate(1 << (Code::kMarkedForDeoptimizationBit -
                        kByteWithDeoptBitOffsetInBits)));
}

void MacroAssembler::TestCodeIsTurbofanned(Register code) {
  testl(FieldOperand(code, Code::kFlagsOffset),
        Immediate(1 << Code::kIsTurbofannedBit));
}

Immediate MacroAssembler::ClearedValue() const {
  return Immediate(static_cast<int32_t>(i::kClearedWeakValue.ptr()));
}

#ifdef V8_ENABLE_DEBUG_CODE
void MacroAssembler::AssertNotSmi(Register object) {
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  Condition is_smi = CheckSmi(object);
  Check(NegateCondition(is_smi), AbortReason::kOperandIsASmi);
}

void MacroAssembler::AssertSmi(Register object) {
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  Condition is_smi = CheckSmi(object);
  Check(is_smi, AbortReason::kOperandIsNotASmi);
#ifdef ENABLE_SLOW_DCHECKS
  ClobberDecompressedSmiBits(object);
#endif
}

void MacroAssembler::AssertSmi(Operand object) {
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  Condition is_smi = CheckSmi(object);
  Check(is_smi, AbortReason::kOperandIsNotASmi);
}

void MacroAssembler::AssertZeroExtended(Register int32_register) {
  if (!v8_flags.slow_debug_code) return;
  ASM_CODE_COMMENT(this);
  DCHECK_NE(int32_register, kScratchRegister);
  movl(kScratchRegister, Immediate(kMaxUInt32));  // zero-extended
  cmpq(int32_register, kScratchRegister);
  Check(below_equal, AbortReason::k32BitValueInRegisterIsNotZeroExtended);
}

void MacroAssembler::AssertSignBitOfSmiIsZero(Register smi_register) {
  if (!v8_flags.slow_debug_code) return;
  ASM_CODE_COMMENT(this);
  DCHECK(COMPRESS_POINTERS_BOOL);
  constexpr int kSmiShiftBits = kSmiTagSize + kSmiShiftSize;
  constexpr Tagged_t kSmiSignBit = Tagged_t{1}
                               << (kSmiShiftBits + kSmiValueSize - 1);
  testl(smi_register, Immediate(static_cast<uint32_t>(kSmiSignBit)));
  Check(zero, AbortReason::kSignBitOfSmiIsNotZero);
}

void MacroAssembler::AssertMap(Register object) {
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  testb(object, Immediate(kSmiTagMask));
  Check(not_equal, AbortReason::kOperandIsNotAMap);
  Push(object);
  LoadMap(object, object);
  CmpInstanceType(object, MAP_TYPE);
  popq(object);
  Check(equal, AbortReason::kOperandIsNotAMap);
}

void MacroAssembler::AssertCode(Register object) {
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  testb(object, Immediate(kSmiTagMask));
  Check(not_equal, AbortReason::kOperandIsNotACode);
  Push(object);
  LoadMap(object, object);
  CmpInstanceType(object, CODE_TYPE);
  popq(object);
  Check(equal, AbortReason::kOperandIsNotACode);
}

void MacroAssembler::AssertSmiOrHeapObjectInMainCompressionCage(
    Register object) {
  if (!PointerCompressionIsEnabled()) return;
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  Label ok;
  // We may not have any scratch registers so we preserve our input register.
  pushq(object);
  j(CheckSmi(object), &ok);
  // Clear the lower 32 bits.
  shrq(object, Immediate(32));
  shlq(object, Immediate(32));
  // Either the value is now equal to the pointer compression cage base or it's
  // zero if we got a compressed pointer register as input.
  j(zero, &ok);
  cmpq(object, kPtrComprCageBaseRegister);
  Check(equal, AbortReason::kObjectNotTagged);
  bind(&ok);
  popq(object);
}

void MacroAssembler::AssertConstructor(Register object) {
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  testb(object, Immediate(kSmiTagMask));
  Check(not_equal, AbortReason::kOperandIsASmiAndNotAConstructor);
  Push(object);
  LoadMap(object, object);
  testb(FieldOperand(object, Map::kBitFieldOffset),
        Immediate(Map::Bits1::IsConstructorBit::kMask));
  Pop(object);
  Check(not_zero, AbortReason::kOperandIsNotAConstructor);
}

void MacroAssembler::AssertFunction(Register object) {
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  testb(object, Immediate(kSmiTagMask));
  Check(not_equal, AbortReason::kOperandIsASmiAndNotAFunction);
  Push(object);
  LoadMap(object, object);
  CmpInstanceTypeRange(object, object, FIRST_JS_FUNCTION_TYPE,
                       LAST_JS_FUNCTION_TYPE);
  Pop(object);
  Check(below_equal, AbortReason::kOperandIsNotAFunction);
}

void MacroAssembler::AssertCallableFunction(Register object) {
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  testb(object, Immediate(kSmiTagMask));
  Check(not_equal, AbortReason::kOperandIsASmiAndNotAFunction);
  Push(object);
  LoadMap(object, object);
  CmpInstanceTypeRange(object, object, FIRST_CALLABLE_JS_FUNCTION_TYPE,
                       LAST_CALLABLE_JS_FUNCTION_TYPE);
  Pop(object);
  Check(below_equal, AbortReason::kOperandIsNotACallableFunction);
}

void MacroAssembler::AssertBoundFunction(Register object) {
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  testb(object, Immediate(kSmiTagMask));
  Check(not_equal, AbortReason::kOperandIsASmiAndNotABoundFunction);
  Push(object);
  IsObjectType(object, JS_BOUND_FUNCTION_TYPE, object);
  Pop(object);
  Check(equal, AbortReason::kOperandIsNotABoundFunction);
}

void MacroAssembler::AssertGeneratorObject(Register object) {
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  testb(object, Immediate(kSmiTagMask));
  Check(not_equal, AbortReason::kOperandIsASmiAndNotAGeneratorObject);

  // Load map
  Register map = object;
  Push(object);
  LoadMap(map, object);

  // Check if JSGeneratorObject
  CmpInstanceTypeRange(map, kScratchRegister, FIRST_JS_GENERATOR_OBJECT_TYPE,
                       LAST_JS_GENERATOR_OBJECT_TYPE);
  // Restore generator object to register and perform assertion
  Pop(object);
  Check(below_equal, AbortReason::kOperandIsNotAGeneratorObject);
}

void MacroAssembler::AssertUndefinedOrAllocationSite(Register object) {
  if (!v8_flags.debug_code) return;
  ASM_CODE_COMMENT(this);
  Label done_checking;
  AssertNotSmi(object);
  Cmp(object, isolate()->factory()->undefined_value());
  j(equal, &done_checking);
  Register map = object;
  Push(object);
  LoadMap(map, object);
  Cmp(map, isolate()->factory()->allocation_site_map());
  Pop(object);
  Assert(equal, AbortReason::kExpectedUndefinedOrCell);
  bind(&done_checking);
}

void MacroAssembler::AssertJSAny(Register object, Register map_tmp,
                                 AbortReason abort_reason) {
  if (!v8_flags.debug_code) return;

  ASM_CODE_COMMENT(this);
  DCHECK(!AreAliased(object, map_tmp));
  Label ok;

  Label::Distance dist = DEBUG_BOOL ? Label::kFar : Label::kNear;

  JumpIfSmi(object, &ok, dist);

  LoadMap(map_tmp, object);
  CmpInstanceType(map_tmp, LAST_NAME_TYPE);
  j(below_equal, &ok, dist);

  CmpInstanceType(map_tmp, FIRST_JS_RECEIVER_TYPE);
  j(above_equal, &ok, dist);

  CompareRoot(map_tmp, RootIndex::kHeapNumberMap);
  j(equal, &ok, dist);

  CompareRoot(map_tmp, RootIndex::kBigIntMap);
  j(equal, &ok, dist);

  CompareRoot(object, RootIndex::kUndefinedValue);
  j(equal, &ok, dist);

  CompareRoot(object, RootIndex::kTrueValue);
  j(equal, &ok, dist);

  CompareRoot(object, RootIndex::kFalseValue);
  j(equal, &ok, dist);

  CompareRoot(object, RootIndex::kNullValue);
  j(equal, &ok, dist);

  Abort(abort_reason);

  bind(&ok);
}

void MacroAssembler::Assert(Condition cc, AbortReason reason) {
  if (v8_flags.debug_code) Check(cc, reason);
}

void MacroAssembler::AssertUnreachable(AbortReason reason) {
  if (v8_flags.debug_code) Abort(reason);
}
#endif  // V8_ENABLE_DEBUG_CODE

void MacroAssembler::LoadWeakValue(Register in_out, Label* target_if_cleared) {
  cmpl(in_out, Immediate(kClearedWeakHeapObjectLower32));
  j(equal, target_if_cleared);

  andq(in_out, Immediate(~static_cast<int32_t>(kWeakHeapObjectMask)));
}

void MacroAssembler::EmitIncrementCounter(StatsCounter* counter, int value) {
  DCHECK_GT(value, 0);
  if (v8_flags.native_code_counters && counter->Enabled()) {
    ASM_CODE_COMMENT(this);
    Operand counter_operand =
        ExternalReferenceAsOperand(ExternalReference::Create(counter));
    // This operation has to be exactly 32-bit wide in case the external
    // reference table redirects the counter to a uint32_t dummy_stats_counter_
    // field.
    if (value == 1) {
      incl(counter_operand);
    } else {
      addl(counter_operand, Immediate(value));
    }
  }
}

void MacroAssembler::EmitDecrementCounter(StatsCounter* counter, int value) {
  DCHECK_GT(value, 0);
  if (v8_flags.native_code_counters && counter->Enabled()) {
    ASM_CODE_COMMENT(this);
    Operand counter_operand =
        ExternalReferenceAsOperand(ExternalReference::Create(counter));
    // This operation has to be exactly 32-bit wide in case the external
    // reference table redirects the counter to a uint32_t dummy_stats_counter_
    // field.
    if (value == 1) {
      decl(counter_operand);
    } else {
      subl(counter_operand, Immediate(value));
    }
  }
}

void MacroAssembler::InvokeFunction(
    Register function, Register new_target, Register actual_parameter_count,
    InvokeType type, ArgumentAdaptionMode argument_adaption_mode) {
  ASM_CODE_COMMENT(this);
  DCHECK_EQ(function, rdi);
  LoadTaggedField(rsi, FieldOperand(function, JSFunction::kContextOffset));
  InvokeFunctionCode(rdi, new_target, actual_parameter_count, type,
                     argument_adaption_mode);
}

void MacroAssembler::InvokeFunctionCode(
    Register function, Register new_target, Register actual_parameter_count,
    InvokeType type, ArgumentAdaptionMode argument_adaption_mode) {
  ASM_CODE_COMMENT(this);
  // You can't call a function without a valid frame.
  DCHECK_IMPLIES(type == InvokeType::kCall, has_frame());
  DCHECK_EQ(function, rdi);
  DCHECK_IMPLIES(new_target.is_valid(), new_target == rdx);

  Register dispatch_handle = kJavaScriptCallDispatchHandleRegister;
  movl(dispatch_handle,
       FieldOperand(function, JSFunction::kDispatchHandleOffset));

  AssertFunction(function);

  // On function call, call into the debugger if necessary.
  Label debug_hook, continue_after_hook;
  {
    ExternalReference debug_hook_active =
        ExternalReference::debug_hook_on_function_call_address(isolate());
    Operand debug_hook_active_operand =
        ExternalReferenceAsOperand(debug_hook_active);
    cmpb(debug_hook_active_operand, Immediate(0));
    j(not_equal, &debug_hook);
  }
  bind(&continue_after_hook);

  // Clear the new.target register if not given.
  if (!new_target.is_valid()) {
    LoadRoot(rdx, RootIndex::kUndefinedValue);
  }

  if (argument_adaption_mode == ArgumentAdaptionMode::kAdapt) {
    Register expected_parameter_count = rbx;
    LoadParameterCountFromJSDispatchTable(expected_parameter_count,
                                          dispatch_handle);
    InvokePrologue(expected_parameter_count, actual_parameter_count, type);
  }

  // We call indirectly through the code field in the function to
  // allow recompilation to take effect without changing any of the
  // call sites.
  static_assert(kJavaScriptCallCodeStartRegister == rcx, "ABI mismatch");
  LoadEntrypointFromJSDispatchTable(rcx, dispatch_handle);
  switch (type) {
    case InvokeType::kCall:
      call(rcx);
      break;
    case InvokeType::kJump:
      jmp(rcx);
      break;
  }
  Label done;
  jmp(&done, Label::kNear);

  // Deferred debug hook.
  bind(&debug_hook);
  CallDebugOnFunctionCall(function, new_target, dispatch_handle,
                          actual_parameter_count);
  jmp(&continue_after_hook);

  bind(&done);
}

Operand MacroAssembler::StackLimitAsOperand(StackLimitKind kind) {
  DCHECK(root_array_available());
  intptr_t offset = kind == StackLimitKind::kRealStackLimit
                        ? IsolateData::real_jslimit_offset()
                        : IsolateData::jslimit_offset();

  CHECK(is_int32(offset));
  return Operand(kRootRegister, static_cast<int32_t>(offset));
}

void MacroAssembler::StackOverflowCheck(
    Register num_args, Label* stack_overflow,
    Label::Distance stack_overflow_distance) {
  ASM_CODE_COMMENT(this);
  DCHECK_NE(num_args, kScratchRegister);
  // Check the stack for overflow. We are not trying to catch
  // interruptions (e.g. debug break and preemption) here, so the "real stack
  // limit" is checked.
  movq(kScratchRegister, rsp);
  // Make kScratchRegister the space we have left. The stack might already be
  // overflowed here which will cause kScratchRegister to become negative.
  subq(kScratchRegister, StackLimitAsOperand(StackLimitKind::kRealStackLimit));
  // TODO(victorgomes): Use ia32 approach with leaq, since it requires less
  // instructions.
  sarq(kScratchRegister, Immediate(kSystemPointerSizeLog2));
  // Check if the arguments will overflow the stack.
  cmpq(kScratchRegister, num_args);
  // Signed comparison.
  // TODO(victorgomes):  Save some bytes in the builtins that use stack checks
  // by jumping to a builtin that throws the exception.
  j(less_equal, stack_overflow, stack_overflow_distance);
}

void MacroAssembler::InvokePrologue(Register expected_parameter_count,
                                    Register actual_parameter_count,
                                    InvokeType type) {
    ASM_CODE_COMMENT(this);
    if (expected_parameter_count == actual_parameter_count) {
      Move(rax, actual_parameter_count);
      return;
    }
    Label regular_invoke;

    // If overapplication or if the actual argument count is equal to the
    // formal parameter count, no need to push extra undefined values.
    subq(expected_parameter_count, actual_parameter_count);
    j(less_equal, &regular_invoke, Label::kFar);

    Label stack_overflow;
    StackOverflowCheck(expected_parameter_count, &stack_overflow);

    // Underapplication. Move the arguments already in the stack, including the
    // receiver and the return address.
    {
      Label copy, check;
      Register src = r8, dest = rsp, num = r9, current = r11;
      movq(src, rsp);
      leaq(kScratchRegister,
           Operand(expected_parameter_count, times_system_pointer_size, 0));
      AllocateStackSpace(kScratchRegister);
      // Extra words are for the return address (if a jump).
      int extra_words =
          type == InvokeType::kCall ? 0 : kReturnAddressStackSlotCount;

      leaq(num, Operand(rax, extra_words));  // Number of words to copy.
      Move(current, 0);
      // Fall-through to the loop body because there are non-zero words to copy.
      bind(&copy);
      movq(kScratchRegister,
           Operand(src, current, times_system_pointer_size, 0));
      movq(Operand(dest, current, times_system_pointer_size, 0),
           kScratchRegister);
      incq(current);
      bind(&check);
      cmpq(current, num);
      j(less, &copy);
      leaq(r8, Operand(rsp, num, times_system_pointer_size, 0));
    }
    // Fill remaining expected arguments with undefined values.
    LoadRoot(kScratchRegister, RootIndex::kUndefinedValue);
    {
      Label loop;
      bind(&loop);
      decq(expected_parameter_count);
      movq(Operand(r8, expected_parameter_count, times_system_pointer_size, 0),
           kScratchRegister);
      j(greater, &loop, Label::kNear);
    }
    jmp(&regular_invoke);

    bind(&stack_overflow);
    {
      FrameScope frame(
          this, has_frame() ? StackFrame::NO_FRAME_TYPE : StackFrame::INTERNAL);
      CallRuntime(Runtime::kThrowStackOverflow);
      int3();  // This should be unreachable.
    }
    bind(&regular_invoke);
}

void MacroAssembler::CallDebugOnFunctionCall(Register fun, Register new_target,
                                             Register dispatch_handle,
                                             Register actual_parameter_count) {
  ASM_CODE_COMMENT(this);
  // Load receiver to pass it later to DebugOnFunctionCall hook.
  // Receiver is located on top of the stack if we have a frame (usually a
  // construct frame), or after the return address if we do not yet have a
  // frame.
  movq(kScratchRegister, Operand(rsp, has_frame() ? 0 : kSystemPointerSize));

  FrameScope frame(
      this, has_frame() ? StackFrame::NO_FRAME_TYPE : StackFrame::INTERNAL);

  // We must not Smi-tag the dispatch handle, because its top bits are
  // meaningful; and we also don't need to, because its low bits are zero.
  static_assert(kJSDispatchHandleShift >= 1);
  Push(dispatch_handle);

  SmiTag(actual_parameter_count);
  Push(actual_parameter_count);
  SmiUntag(actual_parameter_count);

  if (new_target.is_valid()) {
    Push(new_target);
  }
  Push(fun);
  Push(fun);
  Push(kScratchRegister);
  CallRuntime(Runtime::kDebugOnFunctionCall);
  Pop(fun);
  if (new_target.is_valid()) {
    Pop(new_target);
  }
  Pop(actual_parameter_count);
  SmiUntag(actual_parameter_count);
  Pop(dispatch_handle);
}

void MacroAssembler::StubPrologue(StackFrame::Type type) {
  ASM_CODE_COMMENT(this);
  pushq(rbp);  // Caller's frame pointer.
  movq(rbp, rsp);
  Push(Immediate(StackFrame::TypeToMarker(type)));
}

void MacroAssembler::Prologue() {
  ASM_CODE_COMMENT(this);
  pushq(rbp);  // Caller's frame pointer.
  movq(rbp, rsp);
  Push(kContextRegister);                 // Callee's context.
  Push(kJSFunctionRegister);              // Callee's JS function.
  Push(kJavaScriptCallArgCountRegister);  // Actual argument count.
}

void MacroAssembler::EnterFrame(StackFrame::Type type) {
  ASM_CODE_COMMENT(this);
  pushq(rbp);
  movq(rbp, rsp);
  if (!StackFrame::IsJavaScript(type)) {
    static_assert(CommonFrameConstants::kContextOrFrameTypeOffset ==
                  -kSystemPointerSize);
    Push(Immediate(StackFrame::TypeToMarker(type)));
  }
#if V8_ENABLE_WEBASSEMBLY
  if (type == StackFrame::WASM) Push(kWasmImplicitArgRegister);
#endif  // V8_ENABLE_WEBASSEMBLY
}

void MacroAssembler::LeaveFrame(StackFrame::Type type) {
  ASM_CODE_COMMENT(this);
  // TODO(v8:11429): Consider passing BASELINE instead, and checking for
  // IsJSFrame or similar. Could then unify with manual frame leaves in the
  // interpreter too.
  if (v8_flags.debug_code && !StackFrame::IsJavaScript(type)) {
    cmpq(Operand(rbp, CommonFrameConstants::kContextOrFrameTypeOffset),
         Immediate(StackFrame::TypeToMarker(type)));
    Check(equal, AbortReason::kStackFrameTypesMustMatch);
  }
  movq(rsp, rbp);
  popq(rbp);
}

#if defined(V8_TARGET_OS_WIN) || defined(V8_TARGET_OS_MACOS)
void MacroAssembler::AllocateStackSpace(Register bytes_scratch) {
  ASM_CODE_COMMENT(this);
  // On Windows and on macOS, we cannot increment the stack size by more than
  // one page (minimum page size is 4KB) without accessing at least one byte on
  // the page. Check this:
  // https://msdn.microsoft.com/en-us/library/aa227153(v=vs.60).aspx.
  Label check_offset;
  Label touch_next_page;
  jmp(&check_offset);
  bind(&touch_next_page);
  subq(rsp, Immediate(kStackPageSize));
  // Just to touch the page, before we increment further.
  movb(Operand(rsp, 0), Immediate(0));
  subq(bytes_scratch, Immediate(kStackPageSize));

  bind(&check_offset);
  cmpq(bytes_scratch, Immediate(kStackPageSize));
  j(greater_equal, &touch_next_page);

  subq(rsp, bytes_scratch);
}

void MacroAssembler::AllocateStackSpace(int bytes) {
  ASM_CODE_COMMENT(this);
  DCHECK_GE(bytes, 0);
  while (bytes >= kStackPageSize) {
    subq(rsp, Immediate(kStackPageSize));
    movb(Operand(rsp, 0), Immediate(0));
    bytes -= kStackPageSize;
  }
  if (bytes == 0) return;
  subq(rsp, Immediate(bytes));
}
#endif

void MacroAssembler::EnterExitFrame(int extra_slots,
                                    StackFrame::Type frame_type) {
  ASM_CODE_COMMENT(this);
  DCHECK(frame_type == StackFrame::EXIT ||
         frame_type == StackFrame::BUILTIN_EXIT ||
         frame_type == StackFrame::API_NAMED_ACCESSOR_EXIT ||
         frame_type == StackFrame::API_CALLBACK_EXIT);

  // Set up the frame structure on the stack.
  // All constants are relative to the frame pointer of the exit frame.
  DCHECK_EQ(kFPOnStackSize + kPCOnStackSize,
            ExitFrameConstants::kCallerSPDisplacement);
  DCHECK_EQ(kFPOnStackSize, ExitFrameConstants::kCallerPCOffset);
  DCHECK_EQ(0 * kSystemPointerSize, ExitFrameConstants::kCallerFPOffset);
  pushq(rbp);
  movq(rbp, rsp);

  Push(Immediate(StackFrame::TypeToMarker(frame_type)));
  DCHECK_EQ(-2 * kSystemPointerSize, ExitFrameConstants::kSPOffset);
  Push(Immediate(0));  // Saved entry sp, patched below.

  DCHECK(!AreAliased(rbp, kContextRegister));

  movq(AsMemOperand(IsolateFieldId::kCEntryFP), rbp);
  movq(AsMemOperand(IsolateFieldId::kContext), kContextRegister);

#ifdef V8_TARGET_OS_WIN
  // Note this is only correct under the assumption that the caller hasn't
  // considered home stack slots already.
  // TODO(jgruber): This is a bit hacky since the caller in most cases still
  // needs to know about the home stack slots in order to address reserved
  // slots. Consider moving this fully into caller code.
  extra_slots += kWindowsHomeStackSlots;
#endif
  AllocateStackSpace(extra_slots * kSystemPointerSize);

  AlignStackPointer();

  // Patch the saved entry sp.
  movq(Operand(rbp, ExitFrameConstants::kSPOffset), rsp);
}

void MacroAssembler::LeaveExitFrame() {
  ASM_CODE_COMMENT(this);

  leave();

  // Restore the current context from top and clear it in debug mode.
  movq(rsi, AsMemOperand(IsolateFieldId::kContext));
#ifdef DEBUG
  Move(AsMemOperand(IsolateFieldId::kContext), Context::kNoContext);
#endif

  // Clear the top frame.
  Move(AsMemOperand(IsolateFieldId::kCEntryFP), 0);
}

void MacroAssembler::LoadNativeContextSlot(Register dst, int index) {
  ASM_CODE_COMMENT(this);
  // Load native context.
  LoadMap(dst, rsi);
  LoadTaggedField(
      dst,
      FieldOperand(dst, Map::kConstructorOrBackPointerOrNativeContextOffset));
  // Load value from native context.
  LoadTaggedField(dst, Operand(dst, Context::SlotOffset(index)));
}

void MacroAssembler::TryLoadOptimizedOsrCode(Register scratch_and_result,
                                             CodeKind min_opt_level,
                                             Register feedback_vector,
                                             FeedbackSlot slot,
                                             Label* on_result,
                                             Label::Distance distance) {
  ASM_CODE_COMMENT(this);
  Label fallthrough, on_mark_deopt;
  LoadTaggedField(
      scratch_and_result,
      FieldOperand(feedback_vector,
                   FeedbackVector::OffsetOfElementAt(slot.ToInt())));
  LoadWeakValue(scratch_and_result, &fallthrough);

  // Is it marked_for_deoptimization? If yes, clear the slot.
  {
    // The entry references a CodeWrapper object. Unwrap it now.
    LoadCodePointerField(
        scratch_and_result,
        FieldOperand(scratch_and_result, CodeWrapper::kCodeOffset),
        kScratchRegister);

    TestCodeIsMarkedForDeoptimization(scratch_and_result);

    if (min_opt_level == CodeKind::TURBOFAN_JS) {
      j(not_zero, &on_mark_deopt, Label::Distance::kNear);

      TestCodeIsTurbofanned(scratch_and_result);
      j(not_zero, on_result, distance);
      jmp(&fallthrough);
    } else {
      DCHECK_EQ(min_opt_level, CodeKind::MAGLEV);
      j(equal, on_result, distance);
    }

    bind(&on_mark_deopt);
    StoreTaggedField(
        FieldOperand(feedback_vector,
                     FeedbackVector::OffsetOfElementAt(slot.ToInt())),
        ClearedValue());
  }

  bind(&fallthrough);
  Move(scratch_and_result, 0);
}

int MacroAssembler::ArgumentStackSlotsForCFunctionCall(int num_arguments) {
  DCHECK_GE(num_arguments, 0);
#ifdef V8_TARGET_OS_WIN
  return std::max(num_arguments, kWindowsHomeStackSlots);
#else
  return std::max(num_arguments - kRegisterPassedArguments, 0);
#endif
}

void MacroAssembler::PrepareCallCFunction(int num_arguments) {
  ASM_CODE_COMMENT(this);
  int frame_alignment = base::OS::ActivationFrameAlignment();
  DCHECK_NE(frame_alignment, 0);
  DCHECK_GE(num_arguments, 0);

  // Make stack end at alignment and allocate space for arguments and old rsp.
  movq(kScratchRegister, rsp);
  DCHECK(base::bits::IsPowerOfTwo(frame_alignment));
  int argument_slots_on_stack =
      ArgumentStackSlotsForCFunctionCall(num_arguments);
  AllocateStackSpace((argument_slots_on_stack + 1) * kSystemPointerSize);
  andq(rsp, Immediate(-frame_alignment));
  movq(Operand(rsp, argument_slots_on_stack * kSystemPointerSize),
       kScratchRegister);
}

int MacroAssembler::CallCFunction(ExternalReference function, int num_arguments,
                                  SetIsolateDataSlots set_isolate_data_slots,
                                  Label* return_location) {
  // Note: The "CallCFunction" code comment will be generated by the other
  // CallCFunction method called below.
  LoadAddress(rax, function);
  return CallCFunction(rax, num_arguments, set_isolate_data_slots,
                       return_location, CodeSandboxingMode::kUnsandboxed);
}

int MacroAssembler::CallCFunction(Register function, int num_arguments,
                                  SetIsolateDataSlots set_isolate_data_slots,
                                  Label* return_location,
                                  CodeSandboxingMode target_sandboxing_mode) {
  ASM_CODE_COMMENT(this);
  DCHECK_LE(num_arguments, kMaxCParameters);
  DCHECK(has_frame());
  // Check stack alignment.
  if (v8_flags.debug_code) {
    CheckStackAlignment();
  }

  CodeSandboxingMode previous_mode =
      SwitchSandboxingModeBeforeCallIfNeeded(target_sandboxing_mode);

  // Save the frame pointer and PC so that the stack layout remains iterable,
  // even without an ExitFrame which normally exists between JS and C frames.
  Label get_pc;

  if (set_isolate_data_slots == SetIsolateDataSlots::kYes) {
    DCHECK(!AreAliased(kScratchRegister, function));
    leaq(kScratchRegister, Operand(&get_pc, 0));

    CHECK(root_array_available());
    movq(ExternalReferenceAsOperand(IsolateFieldId::kFastCCallCallerPC),
         kScratchRegister);
    movq(ExternalReferenceAsOperand(IsolateFieldId::kFastCCallCallerFP), rbp);
  }

  call(function);
  int call_pc_offset = pc_offset();
  bind(&get_pc);
  if (return_location) bind(return_location);

  DCHECK_NE(base::OS::ActivationFrameAlignment(), 0);
  DCHECK_GE(num_arguments, 0);
  int argument_slots_on_stack =
      ArgumentStackSlotsForCFunctionCall(num_arguments);
  // Restoring the stack pointer has to happen right after the call. The
  // deoptimizer may overwrite everything after restoring the SP.
  movq(rsp, Operand(rsp, argument_slots_on_stack * kSystemPointerSize));

  if (set_isolate_data_slots == SetIsolateDataSlots::kYes) {
    // We don't unset the PC; the FP is the source of truth.
    movq(ExternalReferenceAsOperand(IsolateFieldId::kFastCCallCallerFP),
         Immediate(0));
  }

  SwitchSandboxingModeAfterCallIfNeeded(previous_mode);

  return call_pc_offset;
}

void MacroAssembler::MemoryChunkHeaderFromObject(Register object,
                                                 Register header) {
  constexpr intptr_t alignment_mask =
      MemoryChunk::GetAlignmentMaskForAssembler();
  if (header == object) {
    andq(header, Immediate(~alignment_mask));
  } else {
    movq(header, Immediate(~alignment_mask));
    andq(header, object);
  }
}

void MacroAssembler::CheckPageFlag(Register object, Register scratch, int mask,
                                   Condition cc, Label* condition_met,
                                   Label::Distance condition_met_distance) {
  ASM_CODE_COMMENT(this);
  DCHECK(cc == zero || cc == not_zero);
  MemoryChunkHeaderFromObject(object, scratch);
  if (mask < (1 << kBitsPerByte)) {
    testb(Operand(scratch, MemoryChunk::FlagsOffset()),
          Immediate(static_cast<uint8_t>(mask)));
  } else {
    testl(Operand(scratch, MemoryChunk::FlagsOffset()), Immediate(mask));
  }
  j(cc, condition_met, condition_met_distance);
}

void MacroAssembler::JumpIfMarking(Label* is_marking,
                                   Label::Distance condition_met_distance) {
  testb(Operand(kRootRegister, IsolateData::is_marking_flag_offset()),
        Immediate(static_cast<uint8_t>(1)));
  j(not_zero, is_marking, condition_met_distance);
}

void MacroAssembler::JumpIfNotMarking(Label* not_marking,
                                      Label::Distance condition_met_distance) {
  testb(Operand(kRootRegister, IsolateData::is_marking_flag_offset()),
        Immediate(static_cast<uint8_t>(1)));
  j(zero, not_marking, condition_met_distance);
}

void MacroAssembler::PreCheckSkippedWriteBarrier(Register object,
                                                 Register value,
                                                 Register scratch, Label* ok) {
  ASM_CODE_COMMENT(this);
  DCHECK(!AreAliased(object, scratch));
  DCHECK(!AreAliased(value, scratch));

  // The most common case: Static write barrier elimination is allowed on the
  // last young allocation.
  leaq(scratch, Operand(object, -kHeapObjectTag));
  cmpq(scratch,
       Operand(kRootRegister, IsolateData::last_young_allocation_offset()));
  j(Condition::equal, ok);

#if CONTIGUOUS_COMPRESSED_READ_ONLY_SPACE_BOOL
  JumpIfUnsignedLessThan(value, kContiguousReadOnlyReservationSize, ok);
#else   // !CONTIGUOUS_COMPRESSED_READ_ONLY_SPACE_BOOL
  // Write barier can also be removed if value is in read-only space.
  CheckPageFlag(value, scratch, MemoryChunk::kIsInReadOnlyHeapMask, not_zero,
                ok);
#endif  // !CONTIGUOUS_COMPRESSED_READ_ONLY_SPACE_BOOL

  Label not_ok;

  // Handle allocation folding, allow WB removal if:
  //   LAB start <= last_young_allocation_ < (object address+1) < LAB top
  // Note that object has tag bit set, so object == object address+1.

  // Check LAB start <= last_young_allocation_.
  movq(scratch,
       Operand(kRootRegister, IsolateData::last_young_allocation_offset()));
  cmpq(scratch,
       Operand(kRootRegister, IsolateData::new_allocation_info_start_offset()));
  j(Condition::kUnsignedLessThan, &not_ok);

  // Check last_young_allocation_ < (object address+1).
  cmpq(scratch, object);
  j(Condition::kUnsignedGreaterThanEqual, &not_ok);

  // Check (object address+1) < LAB top.
  cmpq(object,
       Operand(kRootRegister, IsolateData::new_allocation_info_top_offset()));
  j(Condition::kUnsignedLessThan, ok);

  // Slow path: Potentially check more cases in C++.
  bind(&not_ok);
}

void MacroAssembler::CheckMarkBit(Register object, Register scratch0,
                                  Register scratch1, Condition cc,
                                  Label* condition_met,
                                  Label::Distance condition_met_distance) {
  ASM_CODE_COMMENT(this);
  DCHECK(cc == carry || cc == not_carry);
  DCHECK(!AreAliased(object, scratch0, scratch1));

  // Computing cell.
  MemoryChunkHeaderFromObject(object, scratch0);
#ifdef V8_ENABLE_SANDBOX
  movl(scratch0, Operand(scratch0, MemoryChunk::MetadataIndexOffset()));
  andl(scratch0,
       Immediate(MemoryChunkConstants::kMetadataPointerTableSizeMask));
  shll(scratch0, Immediate(kSystemPointerSizeLog2));
  LoadAddress(scratch1,
              ExternalReference::memory_chunk_metadata_table_address());
  movq(scratch0, Operand(scratch1, scratch0, times_1, 0));
#else   // !V8_ENABLE_SANDBOX
  movq(scratch0, Operand(scratch0, MemoryChunk::MetadataOffset()));
#endif  // !V8_ENABLE_SANDBOX
  if (v8_flags.slow_debug_code) {
    Push(object);
    movq(scratch1, Operand(scratch0, BasePage::AreaStartOffset()));
    MemoryChunkHeaderFromObject(scratch1, scratch1);
    MemoryChunkHeaderFromObject(object, object);
    cmpq(object, scratch1);
    Check(equal, AbortReason::kMetadataAreaStartDoesNotMatch);
    Pop(object);
  }
  addq(scratch0, Immediate(MutablePage::MarkingBitmapOffset()));

  movq(scratch1, object);
  andq(scratch1, Immediate(MemoryChunk::GetAlignmentMaskForAssembler()));
  // It's important not to fold the next two shifts.
  shrq(scratch1, Immediate(kTaggedSizeLog2 + MarkingBitmap::kBitsPerCellLog2));
  shlq(scratch1, Immediate(kBitsPerByteLog2));
  addq(scratch0, scratch1);

  // Computing mask.
  movq(scratch1, object);
  andq(scratch1, Immediate(MemoryChunk::GetAlignmentMaskForAssembler()));
  shrq(scratch1, Immediate(kTaggedSizeLog2));
  andq(scratch1, Immediate(MarkingBitmap::kBitIndexMask));
  btq(Operand(scratch0, 0), scratch1);

  j(cc, condition_met, condition_met_distance);
}

void MacroAssembler::ComputeCodeStartAddress(Register dst) {
  Label current;
  bind(&current);
  int pc = pc_offset();
  // Load effective address to get the address of the current instruction.
  leaq(dst, Operand(&current, -pc));
}

void MacroAssembler::AssertNotDeoptimized(Register scratch) {
  int offset = InstructionStream::kCodeOffset - InstructionStream::kHeaderSize;
  LoadProtectedPointerField(scratch,
                            Operand(kJavaScriptCallCodeStartRegister, offset));
  TestCodeIsMarkedForDeoptimization(scratch);
  Assert(zero, AbortReason::kInvalidDeoptimizedCode);
}

void MacroAssembler::CallForDeoptimization(Builtin target, int, Label* exit,
                                           DeoptimizeKind kind, Label* ret,
                                           Label*) {
  ASM_CODE_COMMENT(this);
  // Note: Assembler::call is used here on purpose to guarantee fixed-size
  // exits even on Atom CPUs; see MacroAssembler::Call for Atom-specific
  // performance tuning which emits a different instruction sequence.
  call(EntryFromBuiltinAsOperand(target));
  DCHECK_EQ(SizeOfCodeGeneratedSince(exit),
            (kind == DeoptimizeKind::kLazy ||
             kind == DeoptimizeKind::kLazyAfterFastCall)
                ? Deoptimizer::kLazyDeoptExitSize
                : Deoptimizer::kEagerDeoptExitSize);
}

void MacroAssembler::Trap() { int3(); }
void MacroAssembler::DebugBreak() { int3(); }

// Calls an API function. Allocates HandleScope, extracts returned value
// from handle and propagates exceptions. Clobbers C argument registers
// and C caller-saved registers. Restores context. On return removes
//   (*argc_operand + slots_to_drop_on_return) * kSystemPointerSize
// (GCed, includes the call JS arguments space and the additional space
// allocated for the fast call).
void CallApiFunctionAndReturn(MacroAssembler* masm, bool with_profiling,
                              Register function_address,
                              ExternalReference thunk_ref, Register thunk_arg,
                              int slots_to_drop_on_return,
                              MemOperand* argc_operand,
                              MemOperand return_value_operand,
                              bool handle_interceptor_result) {
  ASM_CODE_COMMENT(masm);
  Label propagate_exception;
  Label delete_allocated_handles;
  Label leave_exit_frame;

  using ER = ExternalReference;

  Isolate* isolate = masm->isolate();
  MemOperand next_mem_op = __ ExternalReferenceAsOperand(
      ER::handle_scope_next_address(isolate), no_reg);
  MemOperand limit_mem_op = __ ExternalReferenceAsOperand(
      ER::handle_scope_limit_address(isolate), no_reg);
  MemOperand level_mem_op = __ ExternalReferenceAsOperand(
      ER::handle_scope_level_address(isolate), no_reg);

  Register return_value = rax;
  Register scratch = kCArgRegs[3];

  // Allocate HandleScope in callee-saved registers.
  // We will need to restore the HandleScope after the call to the API function,
  // by allocating it in callee-saved registers it'll be preserved by C code.
  Register prev_next_address_reg = r12;
  Register prev_limit_reg = r15;

  // C arguments (kCArgRegs[0/1/2]) are expected to be initialized outside, so
  // this function must not corrupt them. kScratchRegister might be used
  // implicitly by the macro assembler.
  DCHECK(!AreAliased(kCArgRegs[0], kCArgRegs[1], kCArgRegs[2],  // C args
                     return_value, scratch, kScratchRegister,
                     prev_next_address_reg, prev_limit_reg));
  // function_address and thunk_arg might overlap but this function must not
  // corrupt them until the call is made (i.e. overlap with return_value is
  // fine).
  DCHECK(!AreAliased(function_address,  // incoming parameters
                     scratch, kScratchRegister, prev_next_address_reg,
                     prev_limit_reg));
  DCHECK(!AreAliased(thunk_arg,  // incoming parameters
                     scratch, kScratchRegister, prev_next_address_reg,
                     prev_limit_reg));
  {
    ASM_CODE_COMMENT_STRING(masm,
                            "Allocate HandleScope in callee-save registers.");
    __ movq(prev_next_address_reg, next_mem_op);
    __ movq(prev_limit_reg, limit_mem_op);
    __ addl(level_mem_op, Immediate(1));
  }

  DCHECK_EQ(__ sandboxing_mode(), CodeSandboxingMode::kSandboxed);
  __ ExitSandbox();

  Label profiler_or_side_effects_check_enabled, done_api_call,
      done_reading_result;
  if (with_profiling) {
    __ RecordComment("Check if profiler or side effects check is enabled");
    __ cmpb(__ ExternalReferenceAsOperand(IsolateFieldId::kExecutionMode),
            Immediate(0));
    __ j(not_zero, &profiler_or_side_effects_check_enabled);
#ifdef V8_RUNTIME_CALL_STATS
    __ RecordComment("Check if RCS is enabled");
    __ Move(scratch, ER::address_of_runtime_stats_flag());
    __ cmpl(Operand(scratch, 0), Immediate(0));
    __ j(not_zero, &profiler_or_side_effects_check_enabled);
#endif  // V8_RUNTIME_CALL_STATS
  }

  __ RecordComment("Call the api function directly.");
  __ call(function_address);
  __ bind(&done_api_call);

  if (handle_interceptor_result) {
    // Skip reading return value if the callback returned kInterceptedNo,
    // this would make the builtin return kNotInterceptedSentinel value.
    // Size is important here, otherwise the C++ function could have returned
    // one- or two-byte value with junk in the upper part.
    static_assert(kInterceptedNo == 1 && kInterceptedSize == 4);
    static_assert(kInterceptedNo == kNotInterceptedSentinel);
    static_assert(kInterceptedYes == 0);
    __ testb(return_value, return_value);
    __ j(not_zero, &done_reading_result);
  }

  __ RecordComment("Load the value from ReturnValue");
  __ movq(return_value, return_value_operand);
  __ bind(&done_reading_result);

  {
    ASM_CODE_COMMENT_STRING(
        masm,
        "No more valid handles (the result handle was the last one)."
        "Restore previous handle scope.");
    __ subl(level_mem_op, Immediate(1));
    __ Assert(above_equal, AbortReason::kInvalidHandleScopeLevel);
    __ movq(next_mem_op, prev_next_address_reg);
    __ cmpq(prev_limit_reg, limit_mem_op);
    __ j(not_equal, &delete_allocated_handles);
  }

  __ RecordComment("Leave the API exit frame.");
  __ bind(&leave_exit_frame);

  Register argc_reg = prev_limit_reg;
  if (argc_operand != nullptr) {
    __ movq(argc_reg, *argc_operand);
  }
  __ LeaveExitFrame();

  __ EnterSandbox();

  {
    ASM_CODE_COMMENT_STRING(masm,
                            "Check if the function scheduled an exception.");
    __ CompareRoot(
        __ ExternalReferenceAsOperand(ER::exception_address(isolate), no_reg),
        RootIndex::kTheHoleValue);
    __ j(not_equal, &propagate_exception);
  }

  if (v8_flags.debug_code) {
    Label ok;
    if (handle_interceptor_result) {
      __ Cmp(return_value, kNotInterceptedSentinel);
      __ j(equal, &ok);
    }
    __ AssertJSAny(return_value, scratch,
                   AbortReason::kAPICallReturnedInvalidObject);
    __ bind(&ok);
  }

  if (argc_operand == nullptr) {
    DCHECK_NE(slots_to_drop_on_return, 0);
    __ Ret(slots_to_drop_on_return * kSystemPointerSize, scratch);
  } else {
    __ PopReturnAddressTo(scratch);
    // {argc_operand} was loaded into {argc_reg} above.
    __ leaq(rsp, Operand(rsp, argc_reg, times_system_pointer_size,
                         slots_to_drop_on_return * kSystemPointerSize));
    // Push and ret (instead of jmp) to keep the RSB and the CET shadow stack
    // balanced.
    __ PushReturnAddressFrom(scratch);
    __ ret(0);
  }
  if (with_profiling) {
    ASM_CODE_COMMENT_STRING(masm, "Call the api function via thunk wrapper.");
    // Call the api function via thunk wrapper.
    __ bind(&profiler_or_side_effects_check_enabled);
    // Additional parameter if provided.
    if (thunk_arg.is_valid()) {
      MemOperand thunk_arg_mem_op = __ ExternalReferenceAsOperand(
          IsolateFieldId::kApiCallbackThunkArgument);
      __ movq(thunk_arg_mem_op, thunk_arg);
    }
    __ Call(thunk_ref);
    __ jmp(&done_api_call);
  }
  __ RecordComment("An exception was thrown. Propagate it.");
  __ bind(&propagate_exception);
  __ TailCallRuntime(Runtime::kPropagateException);
  {
    ASM_CODE_COMMENT_STRING(
        masm, "HandleScope limit has changed. Delete allocated extensions.");
    __ bind(&delete_allocated_handles);
    __ movq(limit_mem_op, prev_limit_reg);
    // Save the return value in a callee-save register.
    Register saved_result = prev_limit_reg;
    __ movq(saved_result, return_value);
    __ LoadAddress(kCArgRegs[0], ER::isolate_address());
    __ Call(ER::delete_handle_scope_extensions());
    __ movq(return_value, saved_result);
    __ jmp(&leave_exit_frame);
  }
}

}  // namespace internal
}  // namespace v8

#undef __

#endif  // V8_TARGET_ARCH_X64