[builtins] migrate C++ String Iterator builtins to baseline TurboFan

src/code-stub-assembler.cc

Index: src/code-stub-assembler.cc
diff --git a/src/code-stub-assembler.cc b/src/code-stub-assembler.cc
index 80c4c77bf6f034f21cd12a2b7d90788aa2fec7d4..64a1ed6b2b9dc4c995909b952ba4d3728e3d7c19 100644
--- a/src/code-stub-assembler.cc
+++ b/src/code-stub-assembler.cc
@@ -2429,6 +2429,251 @@ Node* CodeStubAssembler::StringFromCharCode(Node* code) {
   return var_result.value();
 }
 
+Node* CodeStubAssembler::StringFromCodePointAt(Node* string, Node* length,

[Benedikt Meurer, 2016/09/22 03:48:23]

I'd prefer to have a helper method StringFromCodePoint, similar to
StringFromCharCode, which just takes the 32-bit code point and returns a proper
string, utilizing StringFromCharCode for 16-bit strings, otherwise just
allocating a surrogate pair string.

And since the decision tree to load the actual code point is highly specific to
the string iterator (as it can only deal with sequential/external strings), I'd
prefer to have that either inlined into the iterator next function or as a
helper function in builtins-string.cc. So basically this way:

code, size = LoadCodePointAt(string, string.length, iterator.index);
iterator.index += size;
return StringFromCodePoint(code);

[caitp, 2016/09/22 13:26:21]

The main reason I've put this in CodeStubAssembler is for readability --- I find
it difficult to read what's going on with all of the `assembler->` taking up a
lot of extra space and not really serving much purpose.

Something like the MacroAssembler convention of `#define __ ACCESS_MASM(masm_)`
would help with that a bit, but I didn't want to do that because I figured it
would get shot down :)

I'm in the process of moving this to builtins-string.cc, but it takes up a lot
more lines, and is a lot harder to read this way.

[Benedikt Meurer, 2016/09/23 03:46:24]

I'd be fine with __ if it improves readability (and in fact I think it often
does, as can be seen in the interpreter).

An alternative approach could be a BuiltinAssembler that inherits
CodeStubAssembler, which would also make this easier to read w/o polluting the
CodeStubAssembler.

[caitp, 2016/09/23 10:14:28]

Acknowledged, but maybe that will be a cleanup followup cl

+                                               Node* index) {
+  Label return_result(this);
+  Variable var_result(this, MachineRepresentation::kTagged);
+  var_result.Bind(UndefinedConstant());
+
+  Node* string_instance_type = LoadInstanceType(string);
+
+  Label if_stringissequential(this), if_stringisexternal(this);
+  Branch(Word32Equal(Word32And(string_instance_type,
+                               Int32Constant(kStringRepresentationMask)),
+                     Int32Constant(kSeqStringTag)),
+         &if_stringissequential, &if_stringisexternal);
+
+  Bind(&if_stringissequential);
+  {
+    Label if_stringisonebyte(this), if_stringistwobyte(this);
+    Branch(Word32Equal(Word32And(string_instance_type,
+                                 Int32Constant(kStringEncodingMask)),
+                       Int32Constant(kOneByteStringTag)),
+           &if_stringisonebyte, &if_stringistwobyte);
+
+    Bind(&if_stringisonebyte);
+    {
+      Node* ch =
+          Load(MachineType::Uint8(), string,
+               IntPtrAdd(index, IntPtrConstant(SeqOneByteString::kHeaderSize -
+                                               kHeapObjectTag)));
+      Node* result = AllocateSeqOneByteString(1);
+      StoreNoWriteBarrier(
+          MachineRepresentation::kWord8, result,
+          IntPtrConstant(SeqOneByteString::kHeaderSize - kHeapObjectTag), ch);
+      var_result.Bind(result);
+      Goto(&return_result);
+    }
+
+    Bind(&if_stringistwobyte);
+    {
+      Node* lead = Load(MachineType::Uint16(), string,
+                        IntPtrAdd(WordShl(index, IntPtrConstant(1)),
+                                  IntPtrConstant(SeqTwoByteString::kHeaderSize -
+                                                 kHeapObjectTag)));
+      Node* next_pos = Int32Add(index, Int32Constant(1));
+
+      Label if_issinglecodeunit(this), if_isdoublecodeunit(this);
+      GotoIf(Int32GreaterThanOrEqual(next_pos, length), &if_issinglecodeunit);
+      GotoIf(Int32LessThan(lead, Int32Constant(0xD800)), &if_issinglecodeunit);
+      BranchIf(Int32LessThan(lead, Int32Constant(0xDC00)), &if_isdoublecodeunit,
+               &if_issinglecodeunit);
+
+      Bind(&if_isdoublecodeunit);
+      {
+        Node* trail =
+            Load(MachineType::Uint16(), string,
+                 IntPtrAdd(WordShl(next_pos, IntPtrConstant(1)),
+                           IntPtrConstant(SeqTwoByteString::kHeaderSize -
+                                          kHeapObjectTag)));
+        GotoIf(Int32LessThan(trail, Int32Constant(0xDC00)),
+               &if_issinglecodeunit);
+        GotoIf(Int32GreaterThan(trail, Int32Constant(0xDFFF)),
+               &if_issinglecodeunit);
+
+        Node* result = AllocateSeqTwoByteString(2);
+#if defined(V8_TARGET_LITTLE_ENDIAN)
+        Node* ch = Word32Or(WordShl(trail, Int32Constant(16)), lead);
+#else
+        Node* ch = Word32Or(WordShl(lead, Int32Constant(16)), trail);
+#endif
+        StoreNoWriteBarrier(
+            MachineRepresentation::kWord32, result,
+            IntPtrConstant(SeqTwoByteString::kHeaderSize - kHeapObjectTag), ch);
+        var_result.Bind(result);
+        Goto(&return_result);
+      }
+
+      Bind(&if_issinglecodeunit);
+      {
+        Node* result = AllocateSeqTwoByteString(1);
+        StoreNoWriteBarrier(
+            MachineRepresentation::kWord16, result,
+            IntPtrConstant(SeqOneByteString::kHeaderSize - kHeapObjectTag),
+            lead);
+        var_result.Bind(result);
+        Goto(&return_result);
+      }
+    }
+  }
+
+  Bind(&if_stringisexternal);
+  {
+    Assert(Word32Equal(Word32And(string_instance_type,
+                                 Int32Constant(kStringRepresentationMask)),
+                       Int32Constant(kExternalStringTag)));
+    Label if_stringisshort(this), if_stringisnotshort(this);
+
+    Branch(Word32Equal(Word32And(string_instance_type,
+                                 Int32Constant(kShortExternalStringMask)),
+                       Int32Constant(0)),
+           &if_stringisshort, &if_stringisnotshort);
+
+    Bind(&if_stringisshort);
+    {
+      // Load the actual resource data from the {string}.
+      Node* string_resource_data = LoadObjectField(
+          string, ExternalString::kResourceDataOffset, MachineType::Pointer());
+
+      Label if_stringistwobyte(this), if_stringisonebyte(this);
+      Branch(Word32Equal(Word32And(string_instance_type,
+                                   Int32Constant(kStringEncodingMask)),
+                         Int32Constant(kTwoByteStringTag)),
+             &if_stringistwobyte, &if_stringisonebyte);
+
+      Bind(&if_stringisonebyte);
+      {
+        Node* ch = Load(MachineType::Uint8(), string_resource_data, index);
+        Node* result = AllocateSeqOneByteString(1);
+        StoreNoWriteBarrier(
+            MachineRepresentation::kWord8, result,
+            IntPtrConstant(SeqOneByteString::kHeaderSize - kHeapObjectTag), ch);
+        var_result.Bind(result);
+        Goto(&return_result);
+      }
+
+      Bind(&if_stringistwobyte);
+      {
+        Label if_issinglecodeunit(this), if_isdoublecodeunit(this);
+        Node* lead = Load(MachineType::Uint16(), string_resource_data,
+                          WordShl(index, IntPtrConstant(1)));
+        Node* next_pos = Int32Add(index, Int32Constant(1));
+
+        GotoIf(Int32GreaterThanOrEqual(next_pos, length), &if_issinglecodeunit);
+        GotoIf(Int32LessThan(lead, Int32Constant(0xD800)),
+               &if_issinglecodeunit);
+        Branch(Int32LessThan(lead, Int32Constant(0xDC00)), &if_isdoublecodeunit,
+               &if_issinglecodeunit);
+
+        Bind(&if_isdoublecodeunit);
+        {
+          Node* trail =
+              Load(MachineType::Uint16(), string,
+                   IntPtrAdd(WordShl(next_pos, IntPtrConstant(1)),
+                             IntPtrConstant(SeqTwoByteString::kHeaderSize -
+                                            kHeapObjectTag)));
+          GotoIf(Int32LessThan(trail, Int32Constant(0xDC00)),
+                 &if_issinglecodeunit);
+          GotoIf(Int32GreaterThan(trail, Int32Constant(0xDFFF)),
+                 &if_issinglecodeunit);
+
+          Node* result = AllocateSeqTwoByteString(2);
+#if defined(V8_TARGET_LITTLE_ENDIAN)
+          Node* ch = Word32Or(WordShl(trail, Int32Constant(16)), lead);
+#else
+          Node* ch = Word32Or(WordShl(lead, Int32Constant(16)), trail);
+#endif
+          StoreNoWriteBarrier(
+              MachineRepresentation::kWord32, result,
+              IntPtrConstant(SeqTwoByteString::kHeaderSize - kHeapObjectTag),
+              ch);
+          var_result.Bind(result);
+          Goto(&return_result);
+        }
+
+        Bind(&if_issinglecodeunit);
+        {
+          Node* result = AllocateSeqTwoByteString(1);
+          StoreNoWriteBarrier(
+              MachineRepresentation::kWord16, result,
+              IntPtrConstant(SeqOneByteString::kHeaderSize - kHeapObjectTag),
+              lead);
+          var_result.Bind(result);
+          Goto(&return_result);
+        }
+      }
+    }
+
+    Bind(&if_stringisnotshort);
+    {
+      Label if_issinglecodeunit(this), if_isdoublecodeunit(this);
+      Node* lead =
+          SmiToWord32(CallRuntime(Runtime::kExternalStringGetChar,
+                                  NoContextConstant(), string, SmiTag(index)));
+      Node* next_pos = Int32Add(index, Int32Constant(1));
+
+      GotoIf(Int32GreaterThanOrEqual(next_pos, length), &if_issinglecodeunit);
+      GotoIf(Int32LessThan(lead, Int32Constant(0xD800)), &if_issinglecodeunit);
+      Branch(Int32GreaterThan(lead, Int32Constant(0xDBFF)),
+             &if_issinglecodeunit, &if_isdoublecodeunit);
+
+      Bind(&if_isdoublecodeunit);
+      {
+        Node* trail = SmiToWord32(CallRuntime(Runtime::kExternalStringGetChar,
+                                              NoContextConstant(), string,
+                                              SmiTag(next_pos)));
+        GotoIf(Int32LessThan(trail, Int32Constant(0xDC00)),
+               &if_issinglecodeunit);
+        GotoIf(Int32GreaterThan(trail, Int32Constant(0xDFFF)),
+               &if_issinglecodeunit);
+        Node* result = AllocateSeqTwoByteString(2);
+#if defined(V8_TARGET_LITTLE_ENDIAN)
+        Node* ch = Word32Or(WordShl(trail, Int32Constant(16)), lead);
+#else
+        Node* ch = Word32Or(WordShl(lead, Int32Constant(16)), trail);
+#endif
+        StoreNoWriteBarrier(
+            MachineRepresentation::kWord32, result,
+            IntPtrConstant(SeqTwoByteString::kHeaderSize - kHeapObjectTag), ch);
+        var_result.Bind(result);
+        Goto(&return_result);
+      }
+
+      Bind(&if_issinglecodeunit);
+      {
+        Label if_isonebyte(this), if_istwobyte(this);
+        Branch(Int32LessThan(lead, Int32Constant(0x100)), &if_isonebyte,
+               &if_istwobyte);
+
+        Bind(&if_isonebyte);
+        {
+          Node* result = AllocateSeqOneByteString(1);
+          StoreNoWriteBarrier(
+              MachineRepresentation::kWord8, result,
+              IntPtrConstant(SeqOneByteString::kHeaderSize - kHeapObjectTag),
+              lead);
+          var_result.Bind(result);
+          Goto(&return_result);
+        }
+
+        Bind(&if_istwobyte);
+        {
+          Node* result = AllocateSeqTwoByteString(1);
+          StoreNoWriteBarrier(
+              MachineRepresentation::kWord8, result,
+              IntPtrConstant(SeqTwoByteString::kHeaderSize - kHeapObjectTag),
+              lead);
+          var_result.Bind(result);
+          Goto(&return_result);
+        }
+      }
+    }
+  }
+
+  Bind(&return_result);
+  return var_result.value();
+}
+
 Node* CodeStubAssembler::StringToNumber(Node* context, Node* input) {
   Label runtime(this, Label::kDeferred);
   Label end(this);