Some string optimizations:

 * Inlined checks for strings and regexps.
 * Rewrote split for the non-regexp case.
 * Implemented one-char case for lastIndexOf.

Review URL: http://codereview.chromium.org/596122

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3917 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

src/arm/codegen-arm.cc

@@ -3420,6 +3420,25 @@ void CodeGenerator::GenerateIsArray(ZoneList<Expression*>* args) {
 }
 
 
+void CodeGenerator::GenerateIsRegExp(ZoneList<Expression*>* args) {
+  VirtualFrame::SpilledScope spilled_scope;
+  ASSERT(args->length() == 1);
+  LoadAndSpill(args->at(0));
+  JumpTarget answer;
+  // We need the CC bits to come out as not_equal in the case where the
+  // object is a smi.  This can't be done with the usual test opcode so
+  // we use XOR to get the right CC bits.
+  frame_->EmitPop(r0);
+  __ and_(r1, r0, Operand(kSmiTagMask));
+  __ eor(r1, r1, Operand(kSmiTagMask), SetCC);
+  answer.Branch(ne);
+  // It is a heap object - get the map. Check if the object is a regexp.
+  __ CompareObjectType(r0, r1, r1, JS_REGEXP_TYPE);
+  answer.Bind();
+  cc_reg_ = eq;
+}
+
+
 void CodeGenerator::GenerateIsObject(ZoneList<Expression*>* args) {
   // This generates a fast version of:
   // (typeof(arg) === 'object' || %_ClassOf(arg) == 'RegExp')

src/arm/codegen-arm.h

@@ -359,6 +359,7 @@ class CodeGenerator: public AstVisitor {
   void GenerateIsSmi(ZoneList<Expression*>* args);
   void GenerateIsNonNegativeSmi(ZoneList<Expression*>* args);
   void GenerateIsArray(ZoneList<Expression*>* args);
+  void GenerateIsRegExp(ZoneList<Expression*>* args);
   void GenerateIsObject(ZoneList<Expression*>* args);
   void GenerateIsFunction(ZoneList<Expression*>* args);
   void GenerateIsUndetectableObject(ZoneList<Expression*>* args);

src/codegen.cc

@@ -360,6 +360,7 @@ CodeGenerator::InlineRuntimeLUT CodeGenerator::kInlineRuntimeLUT[] = {
   {&CodeGenerator::GenerateIsSmi, "_IsSmi"},
   {&CodeGenerator::GenerateIsNonNegativeSmi, "_IsNonNegativeSmi"},
   {&CodeGenerator::GenerateIsArray, "_IsArray"},
+  {&CodeGenerator::GenerateIsRegExp, "_IsRegExp"},
   {&CodeGenerator::GenerateIsConstructCall, "_IsConstructCall"},
   {&CodeGenerator::GenerateArgumentsLength, "_ArgumentsLength"},
   {&CodeGenerator::GenerateArgumentsAccess, "_Arguments"},

src/ia32/codegen-ia32.cc

@@ -5422,6 +5422,25 @@ void CodeGenerator::GenerateIsArray(ZoneList<Expression*>* args) {
 }
 
 
+void CodeGenerator::GenerateIsRegExp(ZoneList<Expression*>* args) {
+  ASSERT(args->length() == 1);
+  Load(args->at(0));
+  Result value = frame_->Pop();
+  value.ToRegister();
+  ASSERT(value.is_valid());
+  __ test(value.reg(), Immediate(kSmiTagMask));
+  destination()->false_target()->Branch(equal);
+  // It is a heap object - get map.
+  Result temp = allocator()->Allocate();
+  ASSERT(temp.is_valid());
+  // Check if the object is a regexp.
+  __ CmpObjectType(value.reg(), JS_REGEXP_TYPE, temp.reg());
+  value.Unuse();
+  temp.Unuse();
+  destination()->Split(equal);
+}
+
+
 void CodeGenerator::GenerateIsObject(ZoneList<Expression*>* args) {
   // This generates a fast version of:
   // (typeof(arg) === 'object' || %_ClassOf(arg) == 'RegExp')
@@ -6370,13 +6389,10 @@ void CodeGenerator::VisitCompareOperation(CompareOperation* node) {
       __ movzx_b(temp.reg(), FieldOperand(temp.reg(), Map::kBitFieldOffset));
       __ test(temp.reg(), Immediate(1 << Map::kIsUndetectable));
       destination()->false_target()->Branch(not_zero);
-      __ mov(temp.reg(), FieldOperand(answer.reg(), HeapObject::kMapOffset));
-      __ movzx_b(temp.reg(),
-                 FieldOperand(temp.reg(), Map::kInstanceTypeOffset));
-      __ cmp(temp.reg(), FIRST_NONSTRING_TYPE);
+      __ CmpObjectType(answer.reg(), FIRST_NONSTRING_TYPE, temp.reg());
       temp.Unuse();
       answer.Unuse();
-      destination()->Split(less);
+      destination()->Split(below);
 
     } else if (check->Equals(Heap::boolean_symbol())) {
       __ cmp(answer.reg(), Factory::true_value());

src/ia32/codegen-ia32.h

@@ -551,6 +551,7 @@ class CodeGenerator: public AstVisitor {
   void GenerateIsSmi(ZoneList<Expression*>* args);
   void GenerateIsNonNegativeSmi(ZoneList<Expression*>* args);
   void GenerateIsArray(ZoneList<Expression*>* args);
+  void GenerateIsRegExp(ZoneList<Expression*>* args);
   void GenerateIsObject(ZoneList<Expression*>* args);
   void GenerateIsFunction(ZoneList<Expression*>* args);
   void GenerateIsUndetectableObject(ZoneList<Expression*>* args);

src/macros.py

@@ -74,6 +74,10 @@ const kYearShift          = 9;
 const kMonthShift         = 5;
 
 # Type query macros.
+#
+# Note: We have special support for typeof(foo) === 'bar' in the compiler.
+#       It will *not* generate a runtime typeof call for the most important
+#       values of 'bar'.
 macro IS_NULL(arg)              = (arg === null);
 macro IS_NULL_OR_UNDEFINED(arg) = (arg == null);
 macro IS_UNDEFINED(arg)         = (typeof(arg) === 'undefined');
@@ -83,7 +87,7 @@ macro IS_BOOLEAN(arg)           = (typeof(arg) === 'boolean');
 macro IS_OBJECT(arg)            = (%_IsObject(arg));
 macro IS_ARRAY(arg)             = (%_IsArray(arg));
 macro IS_FUNCTION(arg)          = (%_IsFunction(arg));
-macro IS_REGEXP(arg)            = (%_ClassOf(arg) === 'RegExp');
+macro IS_REGEXP(arg)            = (%_IsRegExp(arg));
 macro IS_DATE(arg)              = (%_ClassOf(arg) === 'Date');
 macro IS_NUMBER_WRAPPER(arg)    = (%_ClassOf(arg) === 'Number');
 macro IS_STRING_WRAPPER(arg)    = (%_ClassOf(arg) === 'String');
@@ -97,9 +101,11 @@ macro FLOOR(arg)                = $floor(arg);
 
 # Inline macros. Use %IS_VAR to make sure arg is evaluated only once.
 macro NUMBER_IS_NAN(arg) = (!%_IsSmi(%IS_VAR(arg)) && !(arg == arg));
-macro TO_INTEGER(arg)    = (%_IsSmi(%IS_VAR(arg)) ? arg : ToInteger(arg));
-macro TO_INT32(arg)      = (%_IsSmi(%IS_VAR(arg)) ? arg : (arg >> 0));
-macro TO_UINT32(arg)     = (arg >>> 0);
+macro TO_INTEGER(arg) = (%_IsSmi(%IS_VAR(arg)) ? arg : ToInteger(arg));
+macro TO_INT32(arg) = (%_IsSmi(%IS_VAR(arg)) ? arg : (arg >> 0));
+macro TO_UINT32(arg) = (arg >>> 0);
+macro TO_STRING_INLINE(arg) = (IS_STRING(%IS_VAR(arg)) ? arg : NonStringToString(arg));
+
 
 # Macros implemented in Python.
 python macro CHAR_CODE(str) = ord(str[1]);

src/mips/codegen-mips.cc

@@ -305,6 +305,11 @@ void CodeGenerator::GenerateIsArray(ZoneList<Expression*>* args) {
 }
 
 
+void CodeGenerator::GenerateIsRegExp(ZoneList<Expression*>* args) {
+  UNIMPLEMENTED_MIPS();
+}
+
+
 void CodeGenerator::GenerateIsConstructCall(ZoneList<Expression*>* args) {
   UNIMPLEMENTED_MIPS();
 }
@@ -365,6 +370,11 @@ void CodeGenerator::GenerateRegExpExec(ZoneList<Expression*>* args) {
 }
 
 
+void CodeGenerator::GenerateNumberToString(ZoneList<Expression*>* args) {
+  UNIMPLEMENTED_MIPS();
+}
+
+
 void CodeGenerator::VisitCallRuntime(CallRuntime* node) {
   UNIMPLEMENTED_MIPS();
 }
@@ -498,4 +508,3 @@ int CompareStub::MinorKey() {
 #undef __
 
 } }  // namespace v8::internal
-

src/mips/codegen-mips.h

@@ -210,6 +210,7 @@ class CodeGenerator: public AstVisitor {
   void GenerateIsSmi(ZoneList<Expression*>* args);
   void GenerateIsNonNegativeSmi(ZoneList<Expression*>* args);
   void GenerateIsArray(ZoneList<Expression*>* args);
+  void GenerateIsRegExp(ZoneList<Expression*>* args);
 
   // Support for construct call checks.
   void GenerateIsConstructCall(ZoneList<Expression*>* args);
@@ -241,6 +242,8 @@ class CodeGenerator: public AstVisitor {
   void GenerateSubString(ZoneList<Expression*>* args);
   void GenerateStringCompare(ZoneList<Expression*>* args);
   void GenerateRegExpExec(ZoneList<Expression*>* args);
+  void GenerateNumberToString(ZoneList<Expression*>* args);
+
 
   // Fast support for Math.sin and Math.cos.
   inline void GenerateMathSin(ZoneList<Expression*>* args);
@@ -308,4 +311,3 @@ class CodeGenerator: public AstVisitor {
 } }  // namespace v8::internal
 
 #endif  // V8_MIPS_CODEGEN_MIPS_H_
-

src/runtime.cc

@@ -2276,6 +2276,20 @@ static int SingleCharIndexOf(Vector<const schar> string,
   return -1;
 }
 
+
+template <typename schar>
+static int SingleCharLastIndexOf(Vector<const schar> string,
+                                 schar pattern_char,
+                                 int start_index) {
+  for (int i = start_index; i >= 0; i--) {
+    if (pattern_char == string[i]) {
+      return i;
+    }
+  }
+  return -1;
+}
+
+
 // Trivial string search for shorter strings.
 // On return, if "complete" is set to true, the return value is the
 // final result of searching for the patter in the subject.
@@ -2459,16 +2473,17 @@ static Object* Runtime_StringLastIndexOf(Arguments args) {
   NoHandleAllocation ha;
   ASSERT(args.length() == 3);
 
-  CONVERT_CHECKED(String, sub, args[0]);
-  CONVERT_CHECKED(String, pat, args[1]);
+  CONVERT_ARG_CHECKED(String, sub, 0);
+  CONVERT_ARG_CHECKED(String, pat, 1);
   Object* index = args[2];
 
-  sub->TryFlattenIfNotFlat();
-  pat->TryFlattenIfNotFlat();
-
   uint32_t start_index;
   if (!Array::IndexFromObject(index, &start_index)) return Smi::FromInt(-1);
 
+  if (!sub->IsFlat()) {
+    FlattenString(sub);
+  }
+
   uint32_t pattern_length = pat->length();
   uint32_t sub_length = sub->length();
 
@@ -2476,6 +2491,25 @@ static Object* Runtime_StringLastIndexOf(Arguments args) {
     start_index = sub_length - pattern_length;
   }
 
+  if (pattern_length == 1) {
+    AssertNoAllocation no_heap_allocation;  // ensure vectors stay valid
+    if (sub->IsAsciiRepresentation()) {
+      uc16 pchar = pat->Get(0);
+      if (pchar > String::kMaxAsciiCharCode) {
+        return Smi::FromInt(-1);
+      }
+      return Smi::FromInt(SingleCharLastIndexOf(sub->ToAsciiVector(),
+                                                static_cast<char>(pat->Get(0)),
+                                                start_index));
+    } else {
+      return Smi::FromInt(SingleCharLastIndexOf(sub->ToUC16Vector(),
+                                                pat->Get(0),
+                                                start_index));
+    }
+  }
+
+  pat->TryFlattenIfNotFlat();
+
   for (int i = start_index; i >= 0; i--) {
     bool found = true;
     for (uint32_t j = 0; j < pattern_length; j++) {

src/runtime.js

@@ -529,6 +529,13 @@ function ToString(x) {
   return (IS_NULL(x)) ? 'null' : %ToString(%DefaultString(x));
 }
 
+function NonStringToString(x) {
+  if (IS_NUMBER(x)) return %NumberToString(x);
+  if (IS_BOOLEAN(x)) return x ? 'true' : 'false';
+  if (IS_UNDEFINED(x)) return 'undefined';
+  return (IS_NULL(x)) ? 'null' : %ToString(%DefaultString(x));
+}
+
 
 // ECMA-262, section 9.9, page 36.
 function ToObject(x) {

src/x64/codegen-x64.cc

@@ -3627,6 +3627,22 @@ void CodeGenerator::GenerateIsArray(ZoneList<Expression*>* args) {
 }
 
 
+void CodeGenerator::GenerateIsRegExp(ZoneList<Expression*>* args) {
+  ASSERT(args->length() == 1);
+  Load(args->at(0));
+  Result value = frame_->Pop();
+  value.ToRegister();
+  ASSERT(value.is_valid());
+  Condition is_smi = masm_->CheckSmi(value.reg());
+  destination()->false_target()->Branch(is_smi);
+  // It is a heap object - get map.
+  // Check if the object is a regexp.
+  __ CmpObjectType(value.reg(), JS_REGEXP_TYPE, kScratchRegister);
+  value.Unuse();
+  destination()->Split(equal);
+}
+
+
 void CodeGenerator::GenerateIsObject(ZoneList<Expression*>* args) {
   // This generates a fast version of:
   // (typeof(arg) === 'object' || %_ClassOf(arg) == 'RegExp')

src/x64/codegen-x64.h

@@ -536,6 +536,7 @@ class CodeGenerator: public AstVisitor {
   void GenerateIsSmi(ZoneList<Expression*>* args);
   void GenerateIsNonNegativeSmi(ZoneList<Expression*>* args);
   void GenerateIsArray(ZoneList<Expression*>* args);
+  void GenerateIsRegExp(ZoneList<Expression*>* args);
   void GenerateIsObject(ZoneList<Expression*>* args);
   void GenerateIsFunction(ZoneList<Expression*>* args);
   void GenerateIsUndetectableObject(ZoneList<Expression*>* args);