Keep natives source code in external strings instead of putting

it in regular flat strings that are part of the snapshot.
After this change we don't need libraries-empty.cc any more.  In
this change libraries-empty.cc is just a the same as libraries.cc
and the scons build builds it but does not use it.  We can move
in stages to a situation where it is not generated at all for all
the build systems that we have.
Review URL: http://codereview.chromium.org/360050

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3238 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

src/SConscript

@@ -264,7 +264,6 @@ def ConfigureObjectFiles():
     else:
       snapshot_cc = Command('snapshot.cc', [], [])
     snapshot_obj = context.ConfigureObject(env, snapshot_cc, CPPPATH=['.'])
-    libraries_obj = context.ConfigureObject(env, libraries_empty_src, CPPPATH=['.'])
   else:
     snapshot_obj = empty_snapshot_obj
   library_objs = [non_snapshot_files, libraries_obj, snapshot_obj]

src/bootstrapper.cc

@@ -36,6 +36,7 @@
 #include "global-handles.h"
 #include "macro-assembler.h"
 #include "natives.h"
+#include "snapshot.h"
 
 namespace v8 {
 namespace internal {
@@ -92,15 +93,35 @@ class SourceCodeCache BASE_EMBEDDED {
 
 static SourceCodeCache natives_cache(Script::TYPE_NATIVE);
 static SourceCodeCache extensions_cache(Script::TYPE_EXTENSION);
+// This is for delete, not delete[].
+static List<char*>* delete_these_non_arrays_on_tear_down = NULL;
 
 
 Handle<String> Bootstrapper::NativesSourceLookup(int index) {
   ASSERT(0 <= index && index < Natives::GetBuiltinsCount());
   if (Heap::natives_source_cache()->get(index)->IsUndefined()) {
+    if (!Snapshot::IsEnabled || FLAG_new_snapshot) {
+      if (delete_these_non_arrays_on_tear_down == NULL) {
+        delete_these_non_arrays_on_tear_down = new List<char*>(2);
+      }
+      // We can use external strings for the natives.
+      NativesExternalStringResource* resource =
+          new NativesExternalStringResource(
+              Natives::GetScriptSource(index).start());
+      // The resources are small objects and we only make a fixed number of
+      // them, but lets clean them up on exit for neatness.
+      delete_these_non_arrays_on_tear_down->
+          Add(reinterpret_cast<char*>(resource));
+      Handle<String> source_code =
+          Factory::NewExternalStringFromAscii(resource);
+      Heap::natives_source_cache()->set(index, *source_code);
+    } else {
+      // Old snapshot code can't cope with external strings at all.
       Handle<String> source_code =
         Factory::NewStringFromAscii(Natives::GetScriptSource(index));
       Heap::natives_source_cache()->set(index, *source_code);
     }
+  }
   Handle<Object> cached_source(Heap::natives_source_cache()->get(index));
   return Handle<String>::cast(cached_source);
 }
@@ -125,6 +146,16 @@ void Bootstrapper::Initialize(bool create_heap_objects) {
 
 
 void Bootstrapper::TearDown() {
+  if (delete_these_non_arrays_on_tear_down != NULL) {
+    int len = delete_these_non_arrays_on_tear_down->length();
+    ASSERT(len < 20);  // Don't use this mechanism for unbounded allocations.
+    for (int i = 0; i < len; i++) {
+      delete delete_these_non_arrays_on_tear_down->at(i);
+    }
+    delete delete_these_non_arrays_on_tear_down;
+    delete_these_non_arrays_on_tear_down = NULL;
+  }
+
   natives_cache.Initialize(false);  // Yes, symmetrical
   extensions_cache.Initialize(false);
 }

src/bootstrapper.h

@@ -76,6 +76,25 @@ class Bootstrapper : public AllStatic {
   static void FreeThreadResources();
 };
 
+
+class NativesExternalStringResource
+    : public v8::String::ExternalAsciiStringResource {
+ public:
+  explicit NativesExternalStringResource(const char* source)
+      : data_(source), length_(strlen(source)) { }
+
+  const char* data() const {
+    return data_;
+  }
+
+  size_t length() const {
+    return length_;
+  }
+ private:
+  const char* data_;
+  size_t length_;
+};
+
 }}  // namespace v8::internal
 
 #endif  // V8_BOOTSTRAPPER_H_

src/objects-inl.h

@@ -163,11 +163,6 @@ bool Object::IsConsString() {
 }
 
 
-#ifdef DEBUG
-// These are for cast checks.  If you need one of these in release
-// mode you should consider using a StringShape before moving it out
-// of the ifdef
-
 bool Object::IsSeqString() {
   if (!IsString()) return false;
   return StringShape(String::cast(this)).IsSequential();
@@ -214,9 +209,6 @@ bool Object::IsSlicedString() {
 }
 
 
-#endif  // DEBUG
-
-
 StringShape::StringShape(String* str)
   : type_(str->map()->instance_type()) {
   set_valid();

src/objects.cc

@@ -1138,6 +1138,15 @@ void HeapObject::IterateBody(InstanceType type, int object_size,
       case kSlicedStringTag:
         reinterpret_cast<SlicedString*>(this)->SlicedStringIterateBody(v);
         break;
+      case kExternalStringTag:
+        if ((type & kStringEncodingMask) == kAsciiStringTag) {
+          reinterpret_cast<ExternalAsciiString*>(this)->
+              ExternalAsciiStringIterateBody(v);
+        } else {
+          reinterpret_cast<ExternalTwoByteString*>(this)->
+              ExternalTwoByteStringIterateBody(v);
+        }
+        break;
     }
     return;
   }
@@ -4398,6 +4407,23 @@ void SlicedString::SlicedStringIterateBody(ObjectVisitor* v) {
   IteratePointer(v, kBufferOffset);
 }
 
+#define FIELD_ADDR(p, offset) \
+  (reinterpret_cast<byte*>(p) + offset - kHeapObjectTag)
+
+void ExternalAsciiString::ExternalAsciiStringIterateBody(ObjectVisitor* v) {
+  typedef v8::String::ExternalAsciiStringResource Resource;
+  v->VisitExternalAsciiString(
+      reinterpret_cast<Resource**>(FIELD_ADDR(this, kResourceOffset)));
+}
+
+
+void ExternalTwoByteString::ExternalTwoByteStringIterateBody(ObjectVisitor* v) {
+  typedef v8::String::ExternalStringResource Resource;
+  v->VisitExternalTwoByteString(
+      reinterpret_cast<Resource**>(FIELD_ADDR(this, kResourceOffset)));
+}
+
+#undef FIELD_ADDR
 
 uint16_t SlicedString::SlicedStringGet(int index) {
   ASSERT(index >= 0 && index < this->length());

src/objects.h

@@ -790,7 +790,6 @@ class Object BASE_EMBEDDED {
   inline bool IsHeapNumber();
   inline bool IsString();
   inline bool IsSymbol();
-#ifdef DEBUG
   // See objects-inl.h for more details
   inline bool IsSeqString();
   inline bool IsSlicedString();
@@ -799,7 +798,6 @@ class Object BASE_EMBEDDED {
   inline bool IsExternalAsciiString();
   inline bool IsSeqTwoByteString();
   inline bool IsSeqAsciiString();
-#endif  // DEBUG
   inline bool IsConsString();
 
   inline bool IsNumber();
@@ -4438,6 +4436,9 @@ class ExternalAsciiString: public ExternalString {
   // Casting.
   static inline ExternalAsciiString* cast(Object* obj);
 
+  // Garbage collection support.
+  void ExternalAsciiStringIterateBody(ObjectVisitor* v);
+
   // Support for StringInputBuffer.
   const unibrow::byte* ExternalAsciiStringReadBlock(unsigned* remaining,
                                                     unsigned* offset,
@@ -4473,6 +4474,9 @@ class ExternalTwoByteString: public ExternalString {
   // Casting.
   static inline ExternalTwoByteString* cast(Object* obj);
 
+  // Garbage collection support.
+  void ExternalTwoByteStringIterateBody(ObjectVisitor* v);
+
   // Support for StringInputBuffer.
   void ExternalTwoByteStringReadBlockIntoBuffer(ReadBlockBuffer* buffer,
                                                 unsigned* offset_ptr,
@@ -5104,6 +5108,12 @@ class ObjectVisitor BASE_EMBEDDED {
   // Visits a runtime entry in the instruction stream.
   virtual void VisitRuntimeEntry(RelocInfo* rinfo) {}
 
+  // Visits the resource of an ASCII or two-byte string.
+  virtual void VisitExternalAsciiString(
+      v8::String::ExternalAsciiStringResource** resource) {}
+  virtual void VisitExternalTwoByteString(
+      v8::String::ExternalStringResource** resource) {}
+
   // Visits a debug call target in the instruction stream.
   virtual void VisitDebugTarget(RelocInfo* rinfo);
 

src/serialize.cc

@@ -39,6 +39,7 @@
 #include "stub-cache.h"
 #include "v8threads.h"
 #include "top.h"
+#include "bootstrapper.h"
 
 namespace v8 {
 namespace internal {
@@ -2062,6 +2063,14 @@ void Deserializer2::ReadChunk(Object** current,
         pages_[space].Add(last_object_address_);
         break;
       }
+      case NATIVES_STRING_RESOURCE: {
+        int index = source_->Get();
+        Vector<const char> source_vector = Natives::GetScriptSource(index);
+        NativesExternalStringResource* resource =
+            new NativesExternalStringResource(source_vector.start());
+        *current++ = reinterpret_cast<Object*>(resource);
+        break;
+      }
       default:
         UNREACHABLE();
     }
@@ -2253,7 +2262,7 @@ void Serializer2::ObjectSerializer::VisitPointers(Object** start,
   Object** current = start;
   while (current < end) {
     while (current < end && (*current)->IsSmi()) current++;
-    OutputRawData(reinterpret_cast<Address>(current));
+    if (current < end) OutputRawData(reinterpret_cast<Address>(current));
 
     while (current < end && !(*current)->IsSmi()) {
       serializer_->SerializeObject(*current, TAGGED_REPRESENTATION);
@@ -2300,6 +2309,33 @@ void Serializer2::ObjectSerializer::VisitCodeTarget(RelocInfo* rinfo) {
 }
 
 
+void Serializer2::ObjectSerializer::VisitExternalAsciiString(
+    v8::String::ExternalAsciiStringResource** resource_pointer) {
+  Address references_start = reinterpret_cast<Address>(resource_pointer);
+  OutputRawData(references_start);
+  for (int i = 0; i < Natives::GetBuiltinsCount(); i++) {
+    // Use raw_unchecked when maps are munged.
+    Object* source = Heap::raw_unchecked_natives_source_cache()->get(i);
+    if (!source->IsUndefined()) {
+      // Don't use cast when maps are munged.
+      ExternalAsciiString* string =
+          reinterpret_cast<ExternalAsciiString*>(source);
+      typedef v8::String::ExternalAsciiStringResource Resource;
+      Resource* resource = string->resource();
+      if (resource == *resource_pointer) {
+        sink_->Put(NATIVES_STRING_RESOURCE, "NativesStringResource");
+        sink_->PutSection(i, "NativesStringResourceEnd");
+        bytes_processed_so_far_ += sizeof(resource);
+        return;
+      }
+    }
+  }
+  // One of the strings in the natives cache should match the resource.  We
+  // can't serialize any other kinds of external strings.
+  UNREACHABLE();
+}
+
+
 void Serializer2::ObjectSerializer::OutputRawData(Address up_to) {
   Address object_start = object_->address();
   int up_to_offset = up_to - object_start;

src/serialize.h

@@ -441,7 +441,8 @@ class SerDes: public GenericDeserializer {
     EXTERNAL_BRANCH_TARGET_SERIALIZATION = 35,
     SYNCHRONIZE = 36,
     START_NEW_PAGE_SERIALIZATION = 37,
-    // Free: 38-47.
+    NATIVES_STRING_RESOURCE = 38,
+    // Free: 39-47.
     BACKREF_SERIALIZATION = 48,
     // One per space, must be kSpaceMask aligned.
     // Free: 57-63.
@@ -567,6 +568,14 @@ class Serializer2 : public SerDes {
     void VisitExternalReferences(Address* start, Address* end);
     void VisitCodeTarget(RelocInfo* target);
     void VisitRuntimeEntry(RelocInfo* reloc);
+    // Used for seralizing the external strings that hold the natives source.
+    void VisitExternalAsciiString(
+        v8::String::ExternalAsciiStringResource** resource);
+    // We can't serialize a heap with external two byte strings.
+    void VisitExternalTwoByteString(
+        v8::String::ExternalStringResource** resource) {
+      UNREACHABLE();
+    }
 
    private:
     void OutputRawData(Address up_to);

test/cctest/cctest.status

@@ -35,9 +35,9 @@ test-api/ApplyInterruption: PASS || TIMEOUT
 
 # This is about to go away anyway since new snapshot code is on the way.
 test-serialize/Deserialize: FAIL
-test-serialize/DeserializeAndRunScript: FAIL
-test-serialize/DeserializeNatives: FAIL
-test-serialize/DeserializeExtensions: FAIL
+test-serialize/DeserializeAndRunScript: FAIL || CRASH
+test-serialize/DeserializeNatives: FAIL || CRASH
+test-serialize/DeserializeExtensions: FAIL || CRASH
 
 # These tests always fail.  They are here to test test.py.  If
 # they don't fail then test.py has failed.

tools/js2c.py

@@ -301,7 +301,7 @@ def JS2C(source, target, env):
     else:
       ids.append((id, len(lines)))
     source_lines.append(SOURCE_DECLARATION % { 'id': id, 'data': data })
-    source_lines_empty.append(SOURCE_DECLARATION % { 'id': id, 'data': 0 })
+    source_lines_empty.append(SOURCE_DECLARATION % { 'id': id, 'data': data })
 
   # Build delay support functions
   get_index_cases = [ ]