Add new HSeqStringGetChar instruction.

This instruction is required for copying characters from sequential
strings in the hydrogenized StringAddStub.

BUG=v8:2990
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/63863005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@17562 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

src/arm/lithium-arm.cc

@@ -1884,6 +1884,13 @@ LInstruction* LChunkBuilder::DoDateField(HDateField* instr) {
 }
 
 
+LInstruction* LChunkBuilder::DoSeqStringGetChar(HSeqStringGetChar* instr) {
+  LOperand* string = UseRegisterAtStart(instr->string());
+  LOperand* index = UseRegisterOrConstantAtStart(instr->index());
+  return DefineAsRegister(new(zone()) LSeqStringGetChar(string, index));
+}
+
+
 LInstruction* LChunkBuilder::DoSeqStringSetChar(HSeqStringSetChar* instr) {
   LOperand* string = UseRegister(instr->string());
   LOperand* index = UseRegisterOrConstant(instr->index());

src/arm/lithium-arm.h

@@ -156,6 +156,7 @@ class LCodeGen;
   V(Random)                                     \
   V(RegExpLiteral)                              \
   V(Return)                                     \
+  V(SeqStringGetChar)                           \
   V(SeqStringSetChar)                           \
   V(ShiftI)                                     \
   V(SmiTag)                                     \
@@ -1360,6 +1361,21 @@ class LDateField V8_FINAL : public LTemplateInstruction<1, 1, 1> {
 };
 
 
+class LSeqStringGetChar V8_FINAL : public LTemplateInstruction<1, 2, 0> {
+ public:
+  LSeqStringGetChar(LOperand* string, LOperand* index) {
+    inputs_[0] = string;
+    inputs_[1] = index;
+  }
+
+  LOperand* string() const { return inputs_[0]; }
+  LOperand* index() const { return inputs_[1]; }
+
+  DECLARE_CONCRETE_INSTRUCTION(SeqStringGetChar, "seq-string-get-char")
+  DECLARE_HYDROGEN_ACCESSOR(SeqStringGetChar)
+};
+
+
 class LSeqStringSetChar V8_FINAL : public LTemplateInstruction<1, 3, 0> {
  public:
   LSeqStringSetChar(LOperand* string,

src/arm/lithium-codegen-arm.cc

@@ -1948,6 +1948,34 @@ MemOperand LCodeGen::BuildSeqStringOperand(Register string,
 }
 
 
+void LCodeGen::DoSeqStringGetChar(LSeqStringGetChar* instr) {
+  String::Encoding encoding = instr->hydrogen()->encoding();
+  Register string = ToRegister(instr->string());
+  Register result = ToRegister(instr->result());
+
+  if (FLAG_debug_code) {
+    Register scratch = scratch0();
+    __ ldr(scratch, FieldMemOperand(string, HeapObject::kMapOffset));
+    __ ldrb(scratch, FieldMemOperand(scratch, Map::kInstanceTypeOffset));
+
+    __ and_(scratch, scratch,
+            Operand(kStringRepresentationMask | kStringEncodingMask));
+    static const uint32_t one_byte_seq_type = kSeqStringTag | kOneByteStringTag;
+    static const uint32_t two_byte_seq_type = kSeqStringTag | kTwoByteStringTag;
+    __ cmp(scratch, Operand(encoding == String::ONE_BYTE_ENCODING
+                            ? one_byte_seq_type : two_byte_seq_type));
+    __ Check(eq, kUnexpectedStringType);
+  }
+
+  MemOperand operand = BuildSeqStringOperand(string, instr->index(), encoding);
+  if (encoding == String::ONE_BYTE_ENCODING) {
+    __ ldrb(result, operand);
+  } else {
+    __ ldrh(result, operand);
+  }
+}
+
+
 void LCodeGen::DoSeqStringSetChar(LSeqStringSetChar* instr) {
   String::Encoding encoding = instr->hydrogen()->encoding();
   Register string = ToRegister(instr->string());

src/hydrogen-instructions.cc

@@ -3991,6 +3991,26 @@ HInstruction* HShr::New(
 }
 
 
+HInstruction* HSeqStringGetChar::New(Zone* zone,
+                                     HValue* context,
+                                     String::Encoding encoding,
+                                     HValue* string,
+                                     HValue* index) {
+  if (FLAG_fold_constants && string->IsConstant() && index->IsConstant()) {
+    HConstant* c_string = HConstant::cast(string);
+    HConstant* c_index = HConstant::cast(index);
+    if (c_string->HasStringValue() && c_index->HasInteger32Value()) {
+      Handle<String> s = c_string->StringValue();
+      int32_t i = c_index->Integer32Value();
+      ASSERT_LE(0, i);
+      ASSERT_LT(i, s->length());
+      return H_CONSTANT_INT(s->Get(i));
+    }
+  }
+  return new(zone) HSeqStringGetChar(encoding, string, index);
+}
+
+
 #undef H_CONSTANT_INT
 #undef H_CONSTANT_DOUBLE
 

src/hydrogen-instructions.h

@@ -159,6 +159,7 @@ class LChunkBuilder;
   V(Return)                                    \
   V(Ror)                                       \
   V(Sar)                                       \
+  V(SeqStringGetChar)                          \
   V(SeqStringSetChar)                          \
   V(Shl)                                       \
   V(Shr)                                       \
@@ -7023,6 +7024,56 @@ class HDateField V8_FINAL : public HUnaryOperation {
 };
 
 
+class HSeqStringGetChar V8_FINAL : public HTemplateInstruction<2> {
+ public:
+  static HInstruction* New(Zone* zone,
+                           HValue* context,
+                           String::Encoding encoding,
+                           HValue* string,
+                           HValue* index);
+
+  virtual Representation RequiredInputRepresentation(int index) V8_OVERRIDE {
+    return (index == 0) ? Representation::Tagged()
+                        : Representation::Integer32();
+  }
+
+  String::Encoding encoding() const { return encoding_; }
+  HValue* string() const { return OperandAt(0); }
+  HValue* index() const { return OperandAt(1); }
+
+  DECLARE_CONCRETE_INSTRUCTION(SeqStringGetChar)
+
+ protected:
+  virtual bool DataEquals(HValue* other) V8_OVERRIDE {
+    return encoding() == HSeqStringGetChar::cast(other)->encoding();
+  }
+
+  virtual Range* InferRange(Zone* zone) V8_OVERRIDE {
+    if (encoding() == String::ONE_BYTE_ENCODING) {
+      return new(zone) Range(0, String::kMaxOneByteCharCode);
+    } else {
+      ASSERT_EQ(String::TWO_BYTE_ENCODING, encoding());
+      return  new(zone) Range(0, String::kMaxUtf16CodeUnit);
+    }
+  }
+
+ private:
+  HSeqStringGetChar(String::Encoding encoding,
+                    HValue* string,
+                    HValue* index) : encoding_(encoding) {
+    SetOperandAt(0, string);
+    SetOperandAt(1, index);
+    set_representation(Representation::Integer32());
+    SetFlag(kUseGVN);
+    SetGVNFlag(kDependsOnStringChars);
+  }
+
+  virtual bool IsDeletable() const V8_OVERRIDE { return true; }
+
+  String::Encoding encoding_;
+};
+
+
 class HSeqStringSetChar V8_FINAL : public HTemplateInstruction<3> {
  public:
   DECLARE_INSTRUCTION_FACTORY_P4(HSeqStringSetChar, String::Encoding,

src/ia32/lithium-codegen-ia32.cc

@@ -2078,6 +2078,34 @@ Operand LCodeGen::BuildSeqStringOperand(Register string,
 }
 
 
+void LCodeGen::DoSeqStringGetChar(LSeqStringGetChar* instr) {
+  String::Encoding encoding = instr->hydrogen()->encoding();
+  Register result = ToRegister(instr->result());
+  Register string = ToRegister(instr->string());
+
+  if (FLAG_debug_code) {
+    __ push(string);
+    __ mov(string, FieldOperand(string, HeapObject::kMapOffset));
+    __ movzx_b(string, FieldOperand(string, Map::kInstanceTypeOffset));
+
+    __ and_(string, Immediate(kStringRepresentationMask | kStringEncodingMask));
+    static const uint32_t one_byte_seq_type = kSeqStringTag | kOneByteStringTag;
+    static const uint32_t two_byte_seq_type = kSeqStringTag | kTwoByteStringTag;
+    __ cmp(string, Immediate(encoding == String::ONE_BYTE_ENCODING
+                             ? one_byte_seq_type : two_byte_seq_type));
+    __ Check(equal, kUnexpectedStringType);
+    __ pop(string);
+  }
+
+  Operand operand = BuildSeqStringOperand(string, instr->index(), encoding);
+  if (encoding == String::ONE_BYTE_ENCODING) {
+    __ movzx_b(result, operand);
+  } else {
+    __ mov_w(result, operand);
+  }
+}
+
+
 void LCodeGen::DoSeqStringSetChar(LSeqStringSetChar* instr) {
   String::Encoding encoding = instr->hydrogen()->encoding();
   Register string = ToRegister(instr->string());

src/ia32/lithium-ia32.cc

@@ -1873,6 +1873,13 @@ LInstruction* LChunkBuilder::DoDateField(HDateField* instr) {
 }
 
 
+LInstruction* LChunkBuilder::DoSeqStringGetChar(HSeqStringGetChar* instr) {
+  LOperand* string = UseRegisterAtStart(instr->string());
+  LOperand* index = UseRegisterOrConstantAtStart(instr->index());
+  return DefineAsRegister(new(zone()) LSeqStringGetChar(string, index));
+}
+
+
 LInstruction* LChunkBuilder::DoSeqStringSetChar(HSeqStringSetChar* instr) {
   LOperand* string = UseRegisterAtStart(instr->string());
   LOperand* index = UseRegisterOrConstantAtStart(instr->index());

src/ia32/lithium-ia32.h

@@ -156,6 +156,7 @@ class LCodeGen;
   V(PushArgument)                               \
   V(RegExpLiteral)                              \
   V(Return)                                     \
+  V(SeqStringGetChar)                           \
   V(SeqStringSetChar)                           \
   V(ShiftI)                                     \
   V(SmiTag)                                     \
@@ -1337,6 +1338,21 @@ class LDateField V8_FINAL : public LTemplateInstruction<1, 1, 1> {
 };
 
 
+class LSeqStringGetChar V8_FINAL : public LTemplateInstruction<1, 2, 0> {
+ public:
+  LSeqStringGetChar(LOperand* string, LOperand* index) {
+    inputs_[0] = string;
+    inputs_[1] = index;
+  }
+
+  LOperand* string() const { return inputs_[0]; }
+  LOperand* index() const { return inputs_[1]; }
+
+  DECLARE_CONCRETE_INSTRUCTION(SeqStringGetChar, "seq-string-get-char")
+  DECLARE_HYDROGEN_ACCESSOR(SeqStringGetChar)
+};
+
+
 class LSeqStringSetChar V8_FINAL : public LTemplateInstruction<1, 3, 0> {
  public:
   LSeqStringSetChar(LOperand* string,

src/x64/assembler-x64.cc

@@ -1366,6 +1366,15 @@ void Assembler::movb(const Operand& dst, Immediate imm) {
 }
 
 
+void Assembler::movw(Register dst, const Operand& src) {
+  EnsureSpace ensure_space(this);
+  emit(0x66);
+  emit_optional_rex_32(dst, src);
+  emit(0x8B);
+  emit_operand(dst, src);
+}
+
+
 void Assembler::movw(const Operand& dst, Register src) {
   EnsureSpace ensure_space(this);
   emit(0x66);

src/x64/assembler-x64.h

@@ -698,6 +698,7 @@ class Assembler : public AssemblerBase {
 
   // Move the low 16 bits of a 64-bit register value to a 16-bit
   // memory location.
+  void movw(Register dst, const Operand& src);
   void movw(const Operand& dst, Register src);
   void movw(const Operand& dst, Immediate imm);
 

src/x64/lithium-codegen-x64.cc

@@ -1650,6 +1650,34 @@ Operand LCodeGen::BuildSeqStringOperand(Register string,
 }
 
 
+void LCodeGen::DoSeqStringGetChar(LSeqStringGetChar* instr) {
+  String::Encoding encoding = instr->hydrogen()->encoding();
+  Register result = ToRegister(instr->result());
+  Register string = ToRegister(instr->string());
+
+  if (FLAG_debug_code) {
+    __ push(string);
+    __ movq(string, FieldOperand(string, HeapObject::kMapOffset));
+    __ movzxbq(string, FieldOperand(string, Map::kInstanceTypeOffset));
+
+    __ andb(string, Immediate(kStringRepresentationMask | kStringEncodingMask));
+    static const uint32_t one_byte_seq_type = kSeqStringTag | kOneByteStringTag;
+    static const uint32_t two_byte_seq_type = kSeqStringTag | kTwoByteStringTag;
+    __ cmpq(string, Immediate(encoding == String::ONE_BYTE_ENCODING
+                              ? one_byte_seq_type : two_byte_seq_type));
+    __ Check(equal, kUnexpectedStringType);
+    __ pop(string);
+  }
+
+  Operand operand = BuildSeqStringOperand(string, instr->index(), encoding);
+  if (encoding == String::ONE_BYTE_ENCODING) {
+    __ movzxbl(result, operand);
+  } else {
+    __ movw(result, operand);
+  }
+}
+
+
 void LCodeGen::DoSeqStringSetChar(LSeqStringSetChar* instr) {
   String::Encoding encoding = instr->hydrogen()->encoding();
   Register string = ToRegister(instr->string());

src/x64/lithium-x64.cc

@@ -1756,6 +1756,13 @@ LInstruction* LChunkBuilder::DoDateField(HDateField* instr) {
 }
 
 
+LInstruction* LChunkBuilder::DoSeqStringGetChar(HSeqStringGetChar* instr) {
+  LOperand* string = UseRegisterAtStart(instr->string());
+  LOperand* index = UseRegisterOrConstantAtStart(instr->index());
+  return DefineAsRegister(new(zone()) LSeqStringGetChar(string, index));
+}
+
+
 LInstruction* LChunkBuilder::DoSeqStringSetChar(HSeqStringSetChar* instr) {
   LOperand* string = UseRegisterAtStart(instr->string());
   LOperand* index = UseRegisterOrConstantAtStart(instr->index());

src/x64/lithium-x64.h

@@ -154,6 +154,7 @@ class LCodeGen;
   V(Random)                                     \
   V(RegExpLiteral)                              \
   V(Return)                                     \
+  V(SeqStringGetChar)                           \
   V(SeqStringSetChar)                           \
   V(ShiftI)                                     \
   V(SmiTag)                                     \
@@ -1281,6 +1282,21 @@ class LDateField V8_FINAL : public LTemplateInstruction<1, 1, 0> {
 };
 
 
+class LSeqStringGetChar V8_FINAL : public LTemplateInstruction<1, 2, 0> {
+ public:
+  LSeqStringGetChar(LOperand* string, LOperand* index) {
+    inputs_[0] = string;
+    inputs_[1] = index;
+  }
+
+  LOperand* string() const { return inputs_[0]; }
+  LOperand* index() const { return inputs_[1]; }
+
+  DECLARE_CONCRETE_INSTRUCTION(SeqStringGetChar, "seq-string-get-char")
+  DECLARE_HYDROGEN_ACCESSOR(SeqStringGetChar)
+};
+
+
 class LSeqStringSetChar V8_FINAL : public LTemplateInstruction<1, 3, 0> {
  public:
   LSeqStringSetChar(LOperand* string,