MIPS: Implement loop for global regexps in regexp assembler.

Port r11623 (f153116d)

BUG=
TEST=

Review URL: https://chromiumcodereview.appspot.com/10436012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@11639 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

src/mips/code-stubs-mips.cc

@@ -5043,7 +5043,7 @@ void RegExpExecStub::Generate(MacroAssembler* masm) {
                       1, a0, a2);
 
   // Isolates: note we add an additional parameter here (isolate pointer).
-  const int kRegExpExecuteArguments = 8;
+  const int kRegExpExecuteArguments = 9;
   const int kParameterRegisters = 4;
   __ EnterExitFrame(false, kRegExpExecuteArguments - kParameterRegisters);
 
@@ -5054,27 +5054,33 @@ void RegExpExecStub::Generate(MacroAssembler* masm) {
   // allocating space for the c argument slots, we don't need to calculate
   // that into the argument positions on the stack. This is how the stack will
   // look (sp meaning the value of sp at this moment):
+  // [sp + 5] - Argument 9
   // [sp + 4] - Argument 8
   // [sp + 3] - Argument 7
   // [sp + 2] - Argument 6
   // [sp + 1] - Argument 5
   // [sp + 0] - saved ra
 
-  // Argument 8: Pass current isolate address.
+  // Argument 9: Pass current isolate address.
   // CFunctionArgumentOperand handles MIPS stack argument slots.
   __ li(a0, Operand(ExternalReference::isolate_address()));
+  __ sw(a0, MemOperand(sp, 5 * kPointerSize));
+
+  // Argument 8: Indicate that this is a direct call from JavaScript.
+  __ li(a0, Operand(1));
   __ sw(a0, MemOperand(sp, 4 * kPointerSize));
 
-  // Argument 7: Indicate that this is a direct call from JavaScript.
+  // Argument 7: Start (high end) of backtracking stack memory area.
-  __ li(a0, Operand(1));
-  __ sw(a0, MemOperand(sp, 3 * kPointerSize));
-
-  // Argument 6: Start (high end) of backtracking stack memory area.
   __ li(a0, Operand(address_of_regexp_stack_memory_address));
   __ lw(a0, MemOperand(a0, 0));
   __ li(a2, Operand(address_of_regexp_stack_memory_size));
   __ lw(a2, MemOperand(a2, 0));
   __ addu(a0, a0, a2);
+  __ sw(a0, MemOperand(sp, 3 * kPointerSize));
+
+  // Argument 6: Set the number of capture registers to zero to force global
+  // regexps to behave as non-global.  This does not affect non-global regexps.
+  __ mov(a0, zero_reg);
   __ sw(a0, MemOperand(sp, 2 * kPointerSize));
 
   // Argument 5: static offsets vector buffer.
@@ -5125,7 +5131,9 @@ void RegExpExecStub::Generate(MacroAssembler* masm) {
   // Check the result.
 
   Label success;
-  __ Branch(&success, eq, v0, Operand(NativeRegExpMacroAssembler::SUCCESS));
+  __ Branch(&success, eq, v0, Operand(1));
+  // We expect exactly one result since we force the called regexp to behave
+  // as non-global.
   Label failure;
   __ Branch(&failure, eq, v0, Operand(NativeRegExpMacroAssembler::FAILURE));
   // If not exception it can only be retry. Handle that in the runtime system.

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

@@ -43,27 +43,31 @@ namespace internal {
 #ifndef V8_INTERPRETED_REGEXP
 /*
  * This assembler uses the following register assignment convention
+ * - t7 : Temporarily stores the index of capture start after a matching pass
+ *        for a global regexp.
  * - t1 : Pointer to current code object (Code*) including heap object tag.
  * - t2 : Current position in input, as negative offset from end of string.
  *        Please notice that this is the byte offset, not the character offset!
  * - t3 : Currently loaded character. Must be loaded using
  *        LoadCurrentCharacter before using any of the dispatch methods.
- * - t4 : points to tip of backtrack stack
+ * - t4 : Points to tip of backtrack stack
  * - t5 : Unused.
  * - t6 : End of input (points to byte after last character in input).
  * - fp : Frame pointer. Used to access arguments, local variables and
  *         RegExp registers.
- * - sp : points to tip of C stack.
+ * - sp : Points to tip of C stack.
  *
  * The remaining registers are free for computations.
  * Each call to a public method should retain this convention.
  *
  * The stack will have the following structure:
  *
- *  - fp[56]  direct_call  (if 1, direct call from JavaScript code,
+ *  - fp[64]  Isolate* isolate   (address of the current isolate)
+ *  - fp[60]  direct_call  (if 1, direct call from JavaScript code,
  *                          if 0, call through the runtime system).
- *  - fp[52]  stack_area_base (High end of the memory area to use as
+ *  - fp[56]  stack_area_base (High end of the memory area to use as
  *                             backtracking stack).
+ *  - fp[52]  capture array size (may fit multiple sets of matches)
  *  - fp[48]  int* capture_array (int[num_saved_registers_], for output).
  *  - fp[44]  secondary link/return address used by native call.
  *  --- sp when called ---
@@ -71,16 +75,17 @@ namespace internal {
  *  - fp[36]  old frame pointer   (r11).
  *  - fp[0..32]  backup of registers s0..s7.
  *  --- frame pointer ----
- *  - fp[-4]  end of input       (Address of end of string).
+ *  - fp[-4]  end of input       (address of end of string).
- *  - fp[-8]  start of input     (Address of first character in string).
+ *  - fp[-8]  start of input     (address of first character in string).
  *  - fp[-12] start index        (character index of start).
  *  - fp[-16] void* input_string (location of a handle containing the string).
- *  - fp[-20] Offset of location before start of input (effectively character
+ *  - fp[-20] success counter    (only for global regexps to count matches).
+ *  - fp[-24] Offset of location before start of input (effectively character
  *            position -1). Used to initialize capture registers to a
  *            non-position.
- *  - fp[-24] At start (if 1, we are starting at the start of the
+ *  - fp[-28] At start (if 1, we are starting at the start of the
  *    string, otherwise 0)
- *  - fp[-28] register 0         (Only positions must be stored in the first
+ *  - fp[-32] register 0         (Only positions must be stored in the first
  *  -         register 1          num_saved_registers_ registers)
  *  -         ...
  *  -         register num_registers-1
@@ -201,8 +206,8 @@ void RegExpMacroAssemblerMIPS::CheckCharacterGT(uc16 limit, Label* on_greater) {
 void RegExpMacroAssemblerMIPS::CheckAtStart(Label* on_at_start) {
   Label not_at_start;
   // Did we start the match at the start of the string at all?
-  __ lw(a0, MemOperand(frame_pointer(), kAtStart));
+  __ lw(a0, MemOperand(frame_pointer(), kStartIndex));
-  BranchOrBacktrack(&not_at_start, eq, a0, Operand(zero_reg));
+  BranchOrBacktrack(&not_at_start, ne, a0, Operand(zero_reg));
 
   // If we did, are we still at the start of the input?
   __ lw(a1, MemOperand(frame_pointer(), kInputStart));
@@ -214,8 +219,8 @@ void RegExpMacroAssemblerMIPS::CheckAtStart(Label* on_at_start) {
 
 void RegExpMacroAssemblerMIPS::CheckNotAtStart(Label* on_not_at_start) {
   // Did we start the match at the start of the string at all?
-  __ lw(a0, MemOperand(frame_pointer(), kAtStart));
+  __ lw(a0, MemOperand(frame_pointer(), kStartIndex));
-  BranchOrBacktrack(on_not_at_start, eq, a0, Operand(zero_reg));
+  BranchOrBacktrack(on_not_at_start, ne, a0, Operand(zero_reg));
   // If we did, are we still at the start of the input?
   __ lw(a1, MemOperand(frame_pointer(), kInputStart));
   __ Addu(a0, end_of_input_address(), Operand(current_input_offset()));
@@ -640,6 +645,7 @@ void RegExpMacroAssemblerMIPS::Fail() {
 
 
 Handle<HeapObject> RegExpMacroAssemblerMIPS::GetCode(Handle<String> source) {
+  Label return_v0;
   if (masm_->has_exception()) {
     // If the code gets corrupted due to long regular expressions and lack of
     // space on trampolines, an internal exception flag is set. If this case
@@ -669,8 +675,9 @@ Handle<HeapObject> RegExpMacroAssemblerMIPS::GetCode(Handle<String> source) {
     // Set frame pointer in space for it if this is not a direct call
     // from generated code.
     __ Addu(frame_pointer(), sp, Operand(4 * kPointerSize));
+    __ mov(a0, zero_reg);
+    __ push(a0);  // Make room for success counter and initialize it to 0.
     __ push(a0);  // Make room for "position - 1" constant (value irrelevant).
-    __ push(a0);  // Make room for "at start" constant (value irrelevant).
 
     // Check if we have space on the stack for registers.
     Label stack_limit_hit;
@@ -689,12 +696,12 @@ Handle<HeapObject> RegExpMacroAssemblerMIPS::GetCode(Handle<String> source) {
     // Exit with OutOfMemory exception. There is not enough space on the stack
     // for our working registers.
     __ li(v0, Operand(EXCEPTION));
-    __ jmp(&exit_label_);
+    __ jmp(&return_v0);
 
     __ bind(&stack_limit_hit);
     CallCheckStackGuardState(a0);
     // If returned value is non-zero, we exit with the returned value as result.
-    __ Branch(&exit_label_, ne, v0, Operand(zero_reg));
+    __ Branch(&return_v0, ne, v0, Operand(zero_reg));
 
     __ bind(&stack_ok);
     // Allocate space on stack for registers.
@@ -715,16 +722,25 @@ Handle<HeapObject> RegExpMacroAssemblerMIPS::GetCode(Handle<String> source) {
     // position registers.
     __ sw(a0, MemOperand(frame_pointer(), kInputStartMinusOne));
 
-    // Determine whether the start index is zero, that is at the start of the
+    // Initialize code pointer register
-    // string, and store that value in a local variable.
+    __ li(code_pointer(), Operand(masm_->CodeObject()), CONSTANT_SIZE);
-    __ mov(t5, a1);
-    __ li(a1, Operand(1));
-    __ Movn(a1, zero_reg, t5);
-    __ sw(a1, MemOperand(frame_pointer(), kAtStart));
 
+    Label load_char_start_regexp, start_regexp;
+    // Load newline if index is at start, previous character otherwise.
+    __ Branch(&load_char_start_regexp, ne, a1, Operand(zero_reg));
+    __ li(current_character(), Operand('\n'));
+    __ jmp(&start_regexp);
+
+    // Global regexp restarts matching here.
+    __ bind(&load_char_start_regexp);
+    // Load previous char as initial value of current character register.
+    LoadCurrentCharacterUnchecked(-1, 1);
+    __ bind(&start_regexp);
+
+    // Initialize on-stack registers.
     if (num_saved_registers_ > 0) {  // Always is, if generated from a regexp.
       // Fill saved registers with initial value = start offset - 1.
-
+      if (num_saved_registers_ > 8) {
         // Address of register 0.
         __ Addu(a1, frame_pointer(), Operand(kRegisterZero));
         __ li(a2, Operand(num_saved_registers_));
@@ -734,20 +750,16 @@ Handle<HeapObject> RegExpMacroAssemblerMIPS::GetCode(Handle<String> source) {
         __ Addu(a1, a1, Operand(-kPointerSize));
         __ Subu(a2, a2, Operand(1));
         __ Branch(&init_loop, ne, a2, Operand(zero_reg));
+      } else {
+        for (int i = 0; i < num_saved_registers_; i++) {
+          __ sw(a0, register_location(i));
+        }
+      }
     }
 
     // Initialize backtrack stack pointer.
     __ lw(backtrack_stackpointer(), MemOperand(frame_pointer(), kStackHighEnd));
-    // Initialize code pointer register
+
-    __ li(code_pointer(), Operand(masm_->CodeObject()), CONSTANT_SIZE);
-    // Load previous char as initial value of current character register.
-    Label at_start;
-    __ lw(a0, MemOperand(frame_pointer(), kAtStart));
-    __ Branch(&at_start, ne, a0, Operand(zero_reg));
-    LoadCurrentCharacterUnchecked(-1, 1);  // Load previous char.
-    __ jmp(&start_label_);
-    __ bind(&at_start);
-    __ li(current_character(), Operand('\n'));
     __ jmp(&start_label_);
 
 
@@ -776,6 +788,10 @@ Handle<HeapObject> RegExpMacroAssemblerMIPS::GetCode(Handle<String> source) {
         for (int i = 0; i < num_saved_registers_; i += 2) {
           __ lw(a2, register_location(i));
           __ lw(a3, register_location(i + 1));
+          if (global()) {
+            // Keep capture start in a4 for the zero-length check later.
+            __ mov(t7, a2);
+          }
           if (mode_ == UC16) {
             __ sra(a2, a2, 1);
             __ Addu(a2, a2, a1);
@@ -791,10 +807,51 @@ Handle<HeapObject> RegExpMacroAssemblerMIPS::GetCode(Handle<String> source) {
           __ Addu(a0, a0, kPointerSize);
         }
       }
+
+      if (global()) {
+        // Restart matching if the regular expression is flagged as global.
+        __ lw(a0, MemOperand(frame_pointer(), kSuccessfulCaptures));
+        __ lw(a1, MemOperand(frame_pointer(), kNumOutputRegisters));
+        __ lw(a2, MemOperand(frame_pointer(), kRegisterOutput));
+        // Increment success counter.
+        __ Addu(a0, a0, 1);
+        __ sw(a0, MemOperand(frame_pointer(), kSuccessfulCaptures));
+        // Capture results have been stored, so the number of remaining global
+        // output registers is reduced by the number of stored captures.
+        __ Subu(a1, a1, num_saved_registers_);
+        // Check whether we have enough room for another set of capture results.
+        __ mov(v0, a0);
+        __ Branch(&return_v0, lt, a1, Operand(num_saved_registers_));
+
+        __ sw(a1, MemOperand(frame_pointer(), kNumOutputRegisters));
+        // Advance the location for output.
+        __ Addu(a2, a2, num_saved_registers_ * kPointerSize);
+        __ sw(a2, MemOperand(frame_pointer(), kRegisterOutput));
+
+        // Prepare a0 to initialize registers with its value in the next run.
+        __ lw(a0, MemOperand(frame_pointer(), kInputStartMinusOne));
+        // Special case for zero-length matches.
+        // t7: capture start index
+        // Not a zero-length match, restart.
+        __ Branch(&load_char_start_regexp, ne, current_input_offset(), Operand(t7));
+        // Offset from the end is zero if we already reached the end.
+        __ Branch(&exit_label_, eq, current_input_offset(), Operand(zero_reg));
+        // Advance current position after a zero-length match.
+        __ Addu(current_input_offset(),
+                current_input_offset(),
+                Operand((mode_ == UC16) ? 2 : 1));
+        __ Branch(&load_char_start_regexp);
+      } else {
         __ li(v0, Operand(SUCCESS));
       }
+    }
     // Exit and return v0.
     __ bind(&exit_label_);
+    if (global()) {
+      __ lw(v0, MemOperand(frame_pointer(), kSuccessfulCaptures));
+    }
+
+    __ bind(&return_v0);
     // Skip sp past regexp registers and local variables..
     __ mov(sp, frame_pointer());
     // Restore registers s0..s7 and return (restoring ra to pc).
@@ -820,7 +877,7 @@ Handle<HeapObject> RegExpMacroAssemblerMIPS::GetCode(Handle<String> source) {
       __ MultiPop(regexp_registers_to_retain);
       // If returning non-zero, we should end execution with the given
       // result as return value.
-      __ Branch(&exit_label_, ne, v0, Operand(zero_reg));
+      __ Branch(&return_v0, ne, v0, Operand(zero_reg));
 
       // String might have moved: Reload end of string from frame.
       __ lw(end_of_input_address(), MemOperand(frame_pointer(), kInputEnd));
@@ -864,7 +921,7 @@ Handle<HeapObject> RegExpMacroAssemblerMIPS::GetCode(Handle<String> source) {
       __ bind(&exit_with_exception);
       // Exit with Result EXCEPTION(-1) to signal thrown exception.
       __ li(v0, Operand(EXCEPTION));
-      __ jmp(&exit_label_);
+      __ jmp(&return_v0);
     }
   }
 
@@ -1012,8 +1069,9 @@ void RegExpMacroAssemblerMIPS::SetRegister(int register_index, int to) {
 }
 
 
-void RegExpMacroAssemblerMIPS::Succeed() {
+bool RegExpMacroAssemblerMIPS::Succeed() {
   __ jmp(&success_label_);
+  return global();
 }
 
 
@@ -1280,8 +1338,9 @@ void RegExpMacroAssemblerMIPS::LoadCurrentCharacterUnchecked(int cp_offset,
                                                              int characters) {
   Register offset = current_input_offset();
   if (cp_offset != 0) {
-    __ Addu(a0, current_input_offset(), Operand(cp_offset * char_size()));
+    // t7 is not being used to store the capture start index at this point.
-    offset = a0;
+    __ Addu(t7, current_input_offset(), Operand(cp_offset * char_size()));
+    offset = t7;
   }
   // We assume that we cannot do unaligned loads on MIPS, so this function
   // must only be used to load a single character at a time.

src/mips/regexp-macro-assembler-mips.h

@@ -115,7 +115,7 @@ class RegExpMacroAssemblerMIPS: public NativeRegExpMacroAssembler {
   virtual void ReadStackPointerFromRegister(int reg);
   virtual void SetCurrentPositionFromEnd(int by);
   virtual void SetRegister(int register_index, int to);
-  virtual void Succeed();
+  virtual bool Succeed();
   virtual void WriteCurrentPositionToRegister(int reg, int cp_offset);
   virtual void ClearRegisters(int reg_from, int reg_to);
   virtual void WriteStackPointerToRegister(int reg);
@@ -141,7 +141,8 @@ class RegExpMacroAssemblerMIPS: public NativeRegExpMacroAssembler {
   static const int kStackFrameHeader = kReturnAddress + kPointerSize;
   // Stack parameters placed by caller.
   static const int kRegisterOutput = kStackFrameHeader + 20;
-  static const int kStackHighEnd = kRegisterOutput + kPointerSize;
+  static const int kNumOutputRegisters = kRegisterOutput + kPointerSize;
+  static const int kStackHighEnd = kNumOutputRegisters + kPointerSize;
   static const int kDirectCall = kStackHighEnd + kPointerSize;
   static const int kIsolate = kDirectCall + kPointerSize;
 
@@ -153,10 +154,10 @@ class RegExpMacroAssemblerMIPS: public NativeRegExpMacroAssembler {
   static const int kInputString = kStartIndex - kPointerSize;
   // When adding local variables remember to push space for them in
   // the frame in GetCode.
-  static const int kInputStartMinusOne = kInputString - kPointerSize;
+  static const int kSuccessfulCaptures = kInputString - kPointerSize;
-  static const int kAtStart = kInputStartMinusOne - kPointerSize;
+  static const int kInputStartMinusOne = kSuccessfulCaptures - kPointerSize;
   // First register address. Following registers are below it on the stack.
-  static const int kRegisterZero = kAtStart - kPointerSize;
+  static const int kRegisterZero = kInputStartMinusOne - kPointerSize;
 
   // Initial size of code buffer.
   static const size_t kRegExpCodeSize = 1024;

src/mips/simulator-mips.h

@@ -50,16 +50,16 @@ namespace internal {
   entry(p0, p1, p2, p3, p4)
 
 typedef int (*mips_regexp_matcher)(String*, int, const byte*, const byte*,
-                                   void*, int*, Address, int, Isolate*);
+                                   void*, int*, int, Address, int, Isolate*);
 
 
 // Call the generated regexp code directly. The code at the entry address
 // should act as a function matching the type arm_regexp_matcher.
 // The fifth argument is a dummy that reserves the space used for
 // the return address added by the ExitFrame in native calls.
-#define CALL_GENERATED_REGEXP_CODE(entry, p0, p1, p2, p3, p4, p5, p6, p7) \
+#define CALL_GENERATED_REGEXP_CODE(entry, p0, p1, p2, p3, p4, p5, p6, p7, p8) \
   (FUNCTION_CAST<mips_regexp_matcher>(entry)( \
-      p0, p1, p2, p3, NULL, p4, p5, p6, p7))
+      p0, p1, p2, p3, NULL, p4, p5, p6, p7, p8))
 
 #define TRY_CATCH_FROM_ADDRESS(try_catch_address) \
   reinterpret_cast<TryCatch*>(try_catch_address)
@@ -403,9 +403,9 @@ class Simulator {
     reinterpret_cast<Object*>(Simulator::current(Isolate::Current())->Call( \
       FUNCTION_ADDR(entry), 5, p0, p1, p2, p3, p4))
 
-#define CALL_GENERATED_REGEXP_CODE(entry, p0, p1, p2, p3, p4, p5, p6, p7) \
+#define CALL_GENERATED_REGEXP_CODE(entry, p0, p1, p2, p3, p4, p5, p6, p7, p8) \
     Simulator::current(Isolate::Current())->Call( \
-        entry, 9, p0, p1, p2, p3, NULL, p4, p5, p6, p7)
+        entry, 10, p0, p1, p2, p3, NULL, p4, p5, p6, p7, p8)
 
 #define TRY_CATCH_FROM_ADDRESS(try_catch_address)                              \
   try_catch_address == NULL ?                                                  \