Port and integrate the irregexp engine from V8

runtime/vm/regexp_assembler.h

Index: runtime/vm/regexp_assembler.h
diff --git a/runtime/vm/regexp_assembler.h b/runtime/vm/regexp_assembler.h
new file mode 100644
index 0000000000000000000000000000000000000000..e1d5b3788f81d5bb8a52a10abc4da57e8c4ed7b3
--- /dev/null
+++ b/runtime/vm/regexp_assembler.h
@@ -0,0 +1,626 @@
+// Copyright (c) 2014, the Dart project authors.  Please see the AUTHORS file
+// for details. All rights reserved. Use of this source code is governed by a
+// BSD-style license that can be found in the LICENSE file.
+
+#ifndef VM_REGEXP_ASSEMBLER_H_
+#define VM_REGEXP_ASSEMBLER_H_
+
+#include "vm/assembler.h"
+#include "vm/intermediate_language.h"
+#include "vm/object.h"
+
+namespace dart {
+
+/// Convenience wrapper around a BlockEntryInstr pointer.
+class BlockLabel : public ValueObject {
+ public:
+  BlockLabel()
+    : block_(new JoinEntryInstr(-1, -1)),
+      is_bound_(false),
+      is_linked_(false) { }
+
+  BlockLabel(const BlockLabel& that)
+    : ValueObject(),
+      block_(that.block_),
+      is_bound_(that.is_bound_),
+      is_linked_(that.is_linked_) { }
+
+  BlockLabel& operator=(const BlockLabel& that) {
+    block_ = that.block_;
+    is_bound_ = that.is_bound_;
+    is_linked_ = that.is_linked_;
+    return *this;
+  }
+
+  JoinEntryInstr* block() const { return block_; }
+
+  bool IsBound() const { return is_bound_; }
+  void SetBound(intptr_t block_id) {
+    ASSERT(!is_bound_);
+    block_->set_block_id(block_id);
+    is_bound_ = true;
+  }
+
+  bool IsLinked() const { return !is_bound_ && is_linked_; }
+  void SetLinked() {
+    is_linked_ = true;
+  }
+
+  intptr_t Position() const {
+    ASSERT(IsBound());
+    return block_->block_id();
+  }
+
+ private:
+  JoinEntryInstr* block_;
+
+  bool is_bound_;
+  bool is_linked_;
+};
+
+
+class RegExpMacroAssembler {
+ public:
+  // The implementation must be able to handle at least:
+  static const intptr_t kMaxRegister = (1 << 16) - 1;
+  static const intptr_t kMaxCPOffset = (1 << 15) - 1;
+  static const intptr_t kMinCPOffset = -(1 << 15);
+
+  static const intptr_t kTableSizeBits = 7;
+  static const intptr_t kTableSize = 1 << kTableSizeBits;
+  static const intptr_t kTableMask = kTableSize - 1;
+
+  enum IrregexpImplementation {
+    kIRImplementation
+  };
+
+  enum StackCheckFlag {
+    kNoStackLimitCheck = false,
+    kCheckStackLimit = true
+  };
+
+  explicit RegExpMacroAssembler(Isolate* isolate);
+  virtual ~RegExpMacroAssembler();
+  // The maximal number of pushes between stack checks. Users must supply
+  // kCheckStackLimit flag to push operations (instead of kNoStackLimitCheck)
+  // at least once for every stack_limit() pushes that are executed.
+  virtual intptr_t stack_limit_slack() = 0;
+  virtual bool CanReadUnaligned() = 0;
+  virtual void AdvanceCurrentPosition(intptr_t by) = 0;  // Signed cp change.
+  virtual void AdvanceRegister(intptr_t reg, intptr_t by) = 0;  // r[reg] += by.
+  // Continues execution from the position pushed on the top of the backtrack
+  // stack by an earlier PushBacktrack(BlockLabel*).
+  virtual void Backtrack() = 0;
+  virtual void BindBlock(BlockLabel* label) = 0;
+  virtual void CheckAtStart(BlockLabel* on_at_start) = 0;
+  // Dispatch after looking the current character up in a 2-bits-per-entry
+  // map.  The destinations vector has up to 4 labels.
+  virtual void CheckCharacter(unsigned c, BlockLabel* on_equal) = 0;
+  // Bitwise and the current character with the given constant and then
+  // check for a match with c.
+  virtual void CheckCharacterAfterAnd(unsigned c,
+                                      unsigned and_with,
+                                      BlockLabel* on_equal) = 0;
+  virtual void CheckCharacterGT(uint16_t limit, BlockLabel* on_greater) = 0;
+  virtual void CheckCharacterLT(uint16_t limit, BlockLabel* on_less) = 0;
+  virtual void CheckGreedyLoop(BlockLabel* on_tos_equals_current_position) = 0;
+  virtual void CheckNotAtStart(BlockLabel* on_not_at_start) = 0;
+  virtual void CheckNotBackReference(
+      intptr_t start_reg, BlockLabel* on_no_match) = 0;
+  virtual void CheckNotBackReferenceIgnoreCase(intptr_t start_reg,
+                                               BlockLabel* on_no_match) = 0;
+  // Check the current character for a match with a literal character.  If we
+  // fail to match then goto the on_failure label.  End of input always
+  // matches.  If the label is NULL then we should pop a backtrack address off
+  // the stack and go to that.
+  virtual void CheckNotCharacter(unsigned c, BlockLabel* on_not_equal) = 0;
+  virtual void CheckNotCharacterAfterAnd(unsigned c,
+                                         unsigned and_with,
+                                         BlockLabel* on_not_equal) = 0;
+  // Subtract a constant from the current character, then and with the given
+  // constant and then check for a match with c.
+  virtual void CheckNotCharacterAfterMinusAnd(uint16_t c,
+                                              uint16_t minus,
+                                              uint16_t and_with,
+                                              BlockLabel* on_not_equal) = 0;
+  virtual void CheckCharacterInRange(uint16_t from,
+                                     uint16_t to,  // Both inclusive.
+                                     BlockLabel* on_in_range) = 0;
+  virtual void CheckCharacterNotInRange(uint16_t from,
+                                        uint16_t to,  // Both inclusive.
+                                        BlockLabel* on_not_in_range) = 0;
+
+  // The current character (modulus the kTableSize) is looked up in the byte
+  // array, and if the found byte is non-zero, we jump to the on_bit_set label.
+  virtual void CheckBitInTable(const TypedData& table,
+                               BlockLabel* on_bit_set) = 0;
+
+  // Checks whether the given offset from the current position is before
+  // the end of the string.  May overwrite the current character.
+  virtual void CheckPosition(intptr_t cp_offset, BlockLabel* on_outside_input) {
+    LoadCurrentCharacter(cp_offset, on_outside_input, true);
+  }
+  // Check whether a standard/default character class matches the current
+  // character. Returns false if the type of special character class does
+  // not have custom support.
+  // May clobber the current loaded character.
+  virtual bool CheckSpecialCharacterClass(uint16_t type,
+                                          BlockLabel* on_no_match) {
+    return false;
+  }
+  virtual void Fail() = 0;
+  virtual Function& GetCode(const String& source) = 0;
+  // Jump to the target label, and continue emission there.
+  virtual void GoTo(BlockLabel* to) = 0;
+  // Check whether a register is >= a given constant and go to a label if it
+  // is.  Backtracks instead if the label is NULL.
+  virtual void IfRegisterGE(
+      intptr_t reg, intptr_t comparand, BlockLabel* if_ge) = 0;
+  // Check whether a register is < a given constant and go to a label if it is.
+  // Backtracks instead if the label is NULL.
+  virtual void IfRegisterLT(
+      intptr_t reg, intptr_t comparand, BlockLabel* if_lt) = 0;
+  // Check whether a register is == to the current position and go to a
+  // label if it is.
+  virtual void IfRegisterEqPos(intptr_t reg, BlockLabel* if_eq) = 0;
+  virtual IrregexpImplementation Implementation() = 0;
+  // The assembler is closed, iff there is no current instruction assigned.
+  virtual bool IsClosed() const = 0;
+  // Jump to the target label without setting it as the current instruction.
+  virtual void Jump(BlockLabel* to) = 0;
+  virtual void LoadCurrentCharacter(intptr_t cp_offset,
+                                    BlockLabel* on_end_of_input,
+                                    bool check_bounds = true,
+                                    intptr_t characters = 1) = 0;
+  virtual void PopCurrentPosition() = 0;
+  virtual void PopRegister(intptr_t register_index) = 0;
+  // Prints string within the generated code. Used for debugging.
+  virtual void Print(const char* str) = 0;
+  // Prints all emitted blocks.
+  virtual void PrintBlocks() = 0;
+  // Pushes the label on the backtrack stack, so that a following Backtrack
+  // will go to this label. Always checks the backtrack stack limit.
+  virtual void PushBacktrack(BlockLabel* label) = 0;
+  virtual void PushCurrentPosition() = 0;
+  virtual void PushRegister(intptr_t register_index,
+                            StackCheckFlag check_stack_limit) = 0;
+  virtual void ReadCurrentPositionFromRegister(intptr_t reg) = 0;
+  virtual void ReadStackPointerFromRegister(intptr_t reg) = 0;
+  virtual void SetCurrentPositionFromEnd(intptr_t by) = 0;
+  virtual void SetRegister(intptr_t register_index, intptr_t to) = 0;
+  // Return whether the matching (with a global regexp) will be restarted.
+  virtual bool Succeed() = 0;
+  virtual void WriteCurrentPositionToRegister(
+      intptr_t reg, intptr_t cp_offset) = 0;
+  virtual void ClearRegisters(intptr_t reg_from, intptr_t reg_to) = 0;
+  virtual void WriteStackPointerToRegister(intptr_t reg) = 0;
+
+  // Controls the generation of large inlined constants in the code.
+  void set_slow_safe(bool ssc) { slow_safe_compiler_ = ssc; }
+  bool slow_safe() { return slow_safe_compiler_; }
+
+  enum GlobalMode { NOT_GLOBAL, GLOBAL, GLOBAL_NO_ZERO_LENGTH_CHECK };
+  // Set whether the regular expression has the global flag.  Exiting due to
+  // a failure in a global regexp may still mean success overall.
+  inline void set_global_mode(GlobalMode mode) { global_mode_ = mode; }
+  inline bool global() { return global_mode_ != NOT_GLOBAL; }
+  inline bool global_with_zero_length_check() {
+    return global_mode_ == GLOBAL;
+  }
+
+  Isolate* isolate() const { return isolate_; }
+
+ private:
+  bool slow_safe_compiler_;
+  bool global_mode_;
+  Isolate* isolate_;
+};
+
+
+class IRRegExpMacroAssembler: public RegExpMacroAssembler {
+ public:
+  // Type of input string to generate code for.
+  enum Mode { ASCII = 1, UC16 = 2 };
+
+  // Result of calling generated native RegExp code.
+  // RETRY: Something significant changed during execution, and the matching
+  //        should be retried from scratch.
+  // EXCEPTION: Something failed during execution. If no exception has been
+  //        thrown, it's an internal out-of-memory, and the caller should
+  //        throw the exception.
+  // FAILURE: Matching failed.
+  // SUCCESS: Matching succeeded, and the output array has been filled with
+  //        capture positions.
+  enum Result { RETRY = -2, EXCEPTION = -1, FAILURE = 0, SUCCESS = 1 };
+
+  IRRegExpMacroAssembler(Mode mode, intptr_t capture_count, Isolate* isolate);
+  virtual ~IRRegExpMacroAssembler();
+
+  virtual bool CanReadUnaligned();
+
+  // Compares two-byte strings case insensitively.
+  // Called from generated RegExp code.
+  static intptr_t CaseInsensitiveCompareUC16(
+      uint8_t* byte_offset1, uint8_t* byte_offset2, size_t byte_length);
+
+  static Result Execute(const Function& function,
+                        const String& input,
+                        const Smi& start_offset,
+                        Array* output,
+                        Isolate* isolate);
+
+  virtual bool IsClosed() const { return (current_instruction_ == NULL); }
+
+  virtual intptr_t stack_limit_slack();
+  virtual void AdvanceCurrentPosition(intptr_t by);
+  virtual void AdvanceRegister(intptr_t reg, intptr_t by);
+  virtual void Backtrack();
+  virtual void BindBlock(BlockLabel* label);
+  virtual void CheckAtStart(BlockLabel* on_at_start);
+  virtual void CheckCharacter(uint32_t c, BlockLabel* on_equal);
+  virtual void CheckCharacterAfterAnd(uint32_t c,
+                                      uint32_t mask,
+                                      BlockLabel* on_equal);
+  virtual void CheckCharacterGT(uint16_t limit, BlockLabel* on_greater);
+  virtual void CheckCharacterLT(uint16_t limit, BlockLabel* on_less);
+  // A "greedy loop" is a loop that is both greedy and with a simple
+  // body. It has a particularly simple implementation.
+  virtual void CheckGreedyLoop(BlockLabel* on_tos_equals_current_position);
+  virtual void CheckNotAtStart(BlockLabel* on_not_at_start);
+  virtual void CheckNotBackReference(intptr_t start_reg,
+                                     BlockLabel* on_no_match);
+  virtual void CheckNotBackReferenceIgnoreCase(intptr_t start_reg,
+                                               BlockLabel* on_no_match);
+  virtual void CheckNotCharacter(uint32_t c, BlockLabel* on_not_equal);
+  virtual void CheckNotCharacterAfterAnd(uint32_t c,
+                                         uint32_t mask,
+                                         BlockLabel* on_not_equal);
+  virtual void CheckNotCharacterAfterMinusAnd(uint16_t c,
+                                              uint16_t minus,
+                                              uint16_t mask,
+                                              BlockLabel* on_not_equal);
+  virtual void CheckCharacterInRange(uint16_t from,
+                                     uint16_t to,
+                                     BlockLabel* on_in_range);
+  virtual void CheckCharacterNotInRange(uint16_t from,
+                                        uint16_t to,
+                                        BlockLabel* on_not_in_range);
+  virtual void CheckBitInTable(const TypedData& table, BlockLabel* on_bit_set);
+
+  // Checks whether the given offset from the current position is before
+  // the end of the string.
+  virtual void CheckPosition(intptr_t cp_offset, BlockLabel* on_outside_input);
+  virtual bool CheckSpecialCharacterClass(
+      uint16_t type, BlockLabel* on_no_match);
+  virtual void Fail();
+  virtual Function& GetCode(const String& source);
+  virtual void GoTo(BlockLabel* to);
+  virtual void IfRegisterGE(intptr_t reg,
+                            intptr_t comparand, BlockLabel* if_ge);
+  virtual void IfRegisterLT(intptr_t reg,
+                            intptr_t comparand, BlockLabel* if_lt);
+  virtual void IfRegisterEqPos(intptr_t reg, BlockLabel* if_eq);
+  virtual IrregexpImplementation Implementation();
+  virtual void Jump(BlockLabel* to);
+  virtual void LoadCurrentCharacter(intptr_t cp_offset,
+                                    BlockLabel* on_end_of_input,
+                                    bool check_bounds = true,
+                                    intptr_t characters = 1);
+  virtual void PopCurrentPosition();
+  virtual void PopRegister(intptr_t register_index);
+  virtual void Print(const char* str);
+  virtual void PushBacktrack(BlockLabel* label);
+  virtual void PushCurrentPosition();
+  virtual void PushRegister(intptr_t register_index,
+                            StackCheckFlag check_stack_limit);
+  virtual void ReadCurrentPositionFromRegister(intptr_t reg);
+  virtual void ReadStackPointerFromRegister(intptr_t reg);
+  virtual void SetCurrentPositionFromEnd(intptr_t by);
+  virtual void SetRegister(intptr_t register_index, intptr_t to);
+  virtual bool Succeed();
+  virtual void WriteCurrentPositionToRegister(intptr_t reg, intptr_t cp_offset);
+  virtual void ClearRegisters(intptr_t reg_from, intptr_t reg_to);
+  virtual void WriteStackPointerToRegister(intptr_t reg);
+
+  virtual void PrintBlocks();
+
+  // Called from RegExp if the stack-guard is triggered.
+  // If the code object is relocated, the return address is fixed before
+  // returning.
+  static intptr_t CheckStackGuardState(Address* return_address,
+                                  Code* re_code,
+                                  Address re_frame);
+
+ private:
+  static const intptr_t kPointerSize = sizeof(void*);
+
+  // Offsets from ebp of function parameters and stored registers.
+  static const intptr_t kFramePointer = 0;
+  // Above the frame pointer - function parameters and return address.
+  static const intptr_t kReturn_eip = kFramePointer + kPointerSize;
+  static const intptr_t kFrameAlign = kReturn_eip + kPointerSize;
+  // Parameters.
+  static const intptr_t kInputString = kFrameAlign;
+  static const intptr_t kStartIndex = kInputString + kPointerSize;
+  static const intptr_t kInputStart = kStartIndex + kPointerSize;
+  static const intptr_t kInputEnd = kInputStart + kPointerSize;
+  static const intptr_t kRegisterOutput = kInputEnd + kPointerSize;
+  // For the case of global regular expression, we have room to store at least
+  // one set of capture results.  For the case of non-global regexp, we ignore
+  // this value.
+  static const intptr_t kNumOutputRegisters = kRegisterOutput + kPointerSize;
+  static const intptr_t kStackHighEnd = kNumOutputRegisters + kPointerSize;
+  static const intptr_t kDirectCall = kStackHighEnd + kPointerSize;
+  static const intptr_t kIsolate = kDirectCall + kPointerSize;
+  // Below the frame pointer - local stack variables.
+  // When adding local variables remember to push space for them in
+  // the frame in GetCode.
+  static const intptr_t kBackup_esi = kFramePointer - kPointerSize;
+  static const intptr_t kBackup_edi = kBackup_esi - kPointerSize;
+  static const intptr_t kBackup_ebx = kBackup_edi - kPointerSize;
+  static const intptr_t kSuccessfulCaptures = kBackup_ebx - kPointerSize;
+  static const intptr_t kInputStartMinusOne =
+      kSuccessfulCaptures - kPointerSize;
+  // First register address. Following registers are below it on the stack.
+  static const intptr_t kRegisterZero = kInputStartMinusOne - kPointerSize;
+
+  // Initial size of code buffer.
+  static const size_t kRegExpCodeSize = 1024;
+
+  // Generate the contents of preset blocks. The entry block is the entry point
+  // of the generated code.
+  void GenerateEntryBlock();
+  // Performs backtracking, i.e. popping an offset from the stack and doing
+  // an indirect goto.
+  void GenerateBacktrackBlock();
+  // Copies capture indices into the result area and returns true.
+  void GenerateSuccessBlock();
+  // Returns false.
+  void GenerateExitBlock();
+
+  enum ComparisonKind {
+    kEQ,
+    kNE,
+    kLT,
+    kGT,
+    kLTE,
+    kGTE,
+  };
+
+  LocalVariable* Local(const char* name);
+  LocalVariable* Parameter(const char* name, intptr_t index) const;
+
+  ConstantInstr* Int64Constant(int64_t value) const;
+  ConstantInstr* Uint64Constant(uint64_t value) const;
+  ConstantInstr* BoolConstant(bool value) const;
+  ConstantInstr* StringConstant(const char* value) const;
+
+  ComparisonInstr* Comparison(ComparisonKind kind,
+                              Definition* lhs, Definition* rhs);
+
+  InstanceCallInstr* InstanceCall(const char *name,
+                                  PushArgumentInstr* arg1) const;
+  InstanceCallInstr* InstanceCall(const char *name,
+                                  PushArgumentInstr* arg1,
+                                  PushArgumentInstr* arg2) const;
+  InstanceCallInstr* InstanceCall(const char *name,
+                                  PushArgumentInstr* arg1,
+                                  PushArgumentInstr* arg2,
+                                  PushArgumentInstr* arg3) const;
+  InstanceCallInstr* InstanceCall(
+      const char* name,
+      ZoneGrowableArray<PushArgumentInstr*>* arguments) const;
+
+  StaticCallInstr* StaticCall(const Function& function) const;
+  StaticCallInstr* StaticCall(const Function& function,
+                              PushArgumentInstr* arg1) const;
+  StaticCallInstr* StaticCall(
+      const Function& function,
+      ZoneGrowableArray<PushArgumentInstr*>* arguments) const;
+
+  // Creates a new block consisting simply of a goto to dst.
+  TargetEntryInstr* TargetWithJoinGoto(JoinEntryInstr* dst);
+
+  // Adds, respectively subtracts lhs and rhs and returns the result.
+  Value* Add(PushArgumentInstr* lhs, PushArgumentInstr* rhs);
+  Value* Sub(PushArgumentInstr* lhs, PushArgumentInstr* rhs);
+
+  LoadLocalInstr* LoadLocal(LocalVariable* local) const;
+  void StoreLocal(LocalVariable* local, Value* value);
+
+  PushArgumentInstr* PushArgument(Value* value);
+
+  // Returns the character within the passed string at the specified index.
+  Value* CharacterAt(Definition* index);
+
+  // Load a number of characters at the given offset from the
+  // current position, into the current-character register.
+  void LoadCurrentCharacterUnchecked(intptr_t cp_offset,
+                                     intptr_t character_count);
+
+  // Check whether preemption has been requested.
+  void CheckPreemption();
+
+  // Generate a call to CheckStackGuardState.
+  void CallCheckStackGuardState(Register scratch);
+
+  // Byte size of chars in the string to match (decided by the Mode argument)
+  inline intptr_t char_size() { return static_cast<int>(mode_); }
+
+  // Equivalent to a conditional branch to the label, unless the label
+  // is NULL, in which case it is a conditional Backtrack.
+  void BranchOrBacktrack(ComparisonInstr* comparison,
+                         BlockLabel* true_successor);
+
+  // Set up all local variables and parameters.
+  void InitializeLocals();
+
+  // A table mapping block ids to block offsets, used to look up offsets
+  // for indirect goto instructions.
+  void FinalizeBlockOffsetTable(const GrowableArray<BlockEntryInstr*>& blocks);
+
+  // When backtrack blocks are pushed, they are not necessarily bound and thus
+  // do not have a block id yet. When we are done generating the IR, insert
+  // final block ids into all such places.
+  void RewriteBacktrackPushes();
+
+  // All blocks that are the target of indirect gotos must be added as an
+  // explicit edge to the graph entry in order for flow analysis to succeed
+  // (similar to catch blocks).
+  void AttachIndirectTargets();
+
+  // Bookkeeping for block ids.
+  intptr_t AllocateBlockId() { return next_block_id_++; }
+
+  // Bookkeeping for temp local ids.
+  intptr_t temp_count() const { return temp_count_; }
+  intptr_t AllocateTemp() { return temp_count_++; }
+  void DeallocateTemps(intptr_t count) {
+    ASSERT(temp_count_ >= count);
+    temp_count_ -= count;
+  }
+
+  // Bookkeeping for the number of pushed arguments.
+  intptr_t args_pushed() const { return args_pushed_; }
+  void add_args_pushed(intptr_t n) { args_pushed_ += n; }
+
+  // Bookkeeping for the number of stack locals.
+  intptr_t GetNextLocalIndex();
+  intptr_t AllocateStackLocal() { return num_stack_locals_++; }
+  intptr_t num_stack_locals() const {
+    return num_stack_locals_;
+  }
+
+  // We never have any copied parameters.
+  intptr_t num_copied_params() const {
+    return 0;
+  }
+
+  // Return the position register at the specified index, creating it if
+  // necessary. Note that the number of such registers can exceed the amount
+  // required by the number of output captures.
+  LocalVariable* position_register(intptr_t index);
+
+  void set_current_instruction(Instruction* instruction);
+
+  // The following functions are responsible for appending instructions
+  // to the current instruction in various ways. The most simple one
+  // is AppendInstruction, which simply appends an instruction and performs
+  // bookkeeping.
+  void AppendInstruction(Instruction* instruction);
+  // Similar to AppendInstruction, but closes the current block by
+  // setting current_instruction_ to NULL.
+  void CloseBlockWith(Instruction* instruction);
+  // Appends definition and allocates a temp index for the result.
+  Value* Bind(Definition* definition);
+  // Appends the definition.
+  void Do(Definition* definition);
+  // Closes the current block with a jump to the specified block.
+  void Jump(JoinEntryInstr* to);
+
+  // Accessors for our local stack_.
+  void PushStack(Definition* definition);
+  Value* PopStack();
+
+  // Prints the specified argument. Used for debugging.
+  void Print(PushArgumentInstr* argument);
+
+  // Used to keep track of each (block, block reference) pair created
+  // during pushes to the backtracking stack. These are required since
+  // such pushes must be rewritten to contain correct block ids once they
+  // are available.
+  struct BacktrackReference : public ValueObject {
+    BacktrackReference(JoinEntryInstr* block, ConstantInstr* reference)
+      : block(block),
+        reference(reference) { }
+
+    BacktrackReference(const BacktrackReference& that)
+      : ValueObject(),
+        block(that.block),
+        reference(that.reference) { }
+
+    BacktrackReference& operator=(const BacktrackReference& that) {
+      block = that.block;
+      reference = that.reference;
+      return *this;
+    }
+
+    JoinEntryInstr* block;
+    ConstantInstr* reference;
+  };
+
+  // Which mode to generate code for (ASCII or UC16).
+  Mode mode_;
+
+  // Counters keeping track of the number of blocks, temps, pushed arguments,
+  // and stack locals.
+  intptr_t next_block_id_;
+  intptr_t temp_count_;
+  intptr_t args_pushed_;
+  intptr_t num_stack_locals_;
+
+  // Block entries used internally.
+  GraphEntryInstr* entry_block_;
+  JoinEntryInstr* start_block_;
+  JoinEntryInstr* success_block_;
+  JoinEntryInstr* backtrack_block_;
+  JoinEntryInstr* exit_block_;
+
+  // All created blocks are contained within this set. Used for printing
+  // the generated code.
+  GrowableArray<BlockEntryInstr*> blocks_;
+
+  // The current instruction to link to when new code is emitted.
+  Instruction* current_instruction_;
+
+  // A list, acting as the runtime stack for both backtrack locations and
+  // stored positions within the string.
+  LocalVariable* stack_;
+
+  // Stores the current character within the string.
+  LocalVariable* current_character_;
+
+  // Stores the current location within the string as a negative offset
+  // from the end of the string.
+  LocalVariable* current_position_;
+
+  // The string being processed, passed as a function parameter.
+  LocalVariable* string_param_;
+
+  // Stores the length of string_param_.
+  LocalVariable* string_param_length_;
+
+  // The start index within the string, passed as a function parameter.
+  LocalVariable* start_index_param_;
+
+  // The array of matches to be filled in, passed as a function parameter.
+  LocalVariable* matches_param_;
+
+  // The word character map static member of the RegExp class.
+  // Byte map of one byte characters with a 0xff if the character is a word
+  // character (digit, letter or underscore) and 0x00 otherwise.
+  // Used by generated RegExp code.
+  LocalVariable* word_character_map_;
+
+  // An assortment of utility variables.
+  LocalVariable* capture_length_;
+  LocalVariable* stack_ptr_;
+  LocalVariable* match_start_index_;
+  LocalVariable* capture_start_index_;
+  LocalVariable* match_end_index_;
+  LocalVariable* char_in_capture_;
+  LocalVariable* char_in_match_;
+
+  // Stored positions containing group bounds. Generated as needed.
+  const intptr_t position_registers_count_;
+  GrowableArray<LocalVariable*> position_registers_;
+
+  // Stores code offsets for all blocks.
+  // Used to implement backtracking.
+  GrowableObjectArray& block_offsets_;
+  GrowableArray<BacktrackReference> backtrack_references_;
+};
+
+}  // namespace dart
+
+#endif  // VM_REGEXP_ASSEMBLER_H_