Implement --omit, --raw_[context_]file=... for mksnapshot tool.

These switches are analogous to --omit and --raw for js2c.py.
The goal it to allow embedders to externally load all (or generate) the V8 startup data.

There's a bit of gratuitous refactoring with the writing logic that helped me understand the code. Let me know if it's not an improvement.

BUG=355539
LOG=N
R=jochen@chromium.org, svenpanne@chromium.org

Review URL: https://codereview.chromium.org/249283002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@20941 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

diff --git a/src/flag-definitions.h b/src/flag-definitions.h
index 989e285..4e8791b 100644 (file)
--- a/src/flag-definitions.h
+++ b/src/flag-definitions.h
@@ -653,7 +653,13 @@ DEFINE_string(testing_serialization_file, "/tmp/serdes",
 
 // mksnapshot.cc
 DEFINE_string(extra_code, NULL, "A filename with extra code to be included in"
-                  " the snapshot (mksnapshot only)")
+                                " the snapshot (mksnapshot only)")
+DEFINE_string(raw_file, NULL, "A file to write the raw snapshot bytes to. "
+                              "(mksnapshot only)")
+DEFINE_string(raw_context_file, NULL, "A file to write the raw context "
+                                      "snapshot bytes to. (mksnapshot only)")
+DEFINE_bool(omit, false, "Omit raw snapshot bytes in generated code. "
+                         "(mksnapshot only)")
 
 // code-stubs-hydrogen.cc
 DEFINE_bool(profile_hydrogen_code_stub_compilation, false,

diff --git a/src/mksnapshot.cc b/src/mksnapshot.cc
index 47b5b3c..9f91b83 100644 (file)
--- a/src/mksnapshot.cc
+++ b/src/mksnapshot.cc
@@ -56,148 +56,173 @@ class Compressor {
 };
 
 
-class PartialSnapshotSink : public i::SnapshotByteSink {
+class ListSnapshotSink : public i::SnapshotByteSink {
  public:
-  PartialSnapshotSink() : data_(), raw_size_(-1) { }
-  virtual ~PartialSnapshotSink() { data_.Free(); }
-  virtual void Put(int byte, const char* description) {
-    data_.Add(byte);
+  explicit ListSnapshotSink(i::List<char>* data) : data_(data) { }
+  virtual ~ListSnapshotSink() {}
+  virtual void Put(int byte, const char* description) { data_->Add(byte); }
+  virtual int Position() { return data_->length(); }
+ private:
+  i::List<char>* data_;
+};
+
+
+class SnapshotWriter {
+ public:
+  explicit SnapshotWriter(const char* snapshot_file)
+      : fp_(GetFileDescriptorOrDie(snapshot_file))
+      , raw_file_(NULL)
+      , raw_context_file_(NULL)
+      , compressor_(NULL)
+      , omit_(false) {
   }
-  virtual int Position() { return data_.length(); }
-  void Print(FILE* fp) {
-    int length = Position();
-    for (int j = 0; j < length; j++) {
-      if ((j & 0x1f) == 0x1f) {
-        fprintf(fp, "\n");
-      }
-      if (j != 0) {
-        fprintf(fp, ",");
-      }
-      fprintf(fp, "%u", static_cast<unsigned char>(at(j)));
-    }
+
+  ~SnapshotWriter() {
+    fclose(fp_);
+    if (raw_file_) fclose(raw_file_);
+    if (raw_context_file_) fclose(raw_context_file_);
   }
-  char at(int i) { return data_[i]; }
-  bool Compress(Compressor* compressor) {
-    ASSERT_EQ(-1, raw_size_);
-    raw_size_ = data_.length();
-    if (!compressor->Compress(data_.ToVector())) return false;
-    data_.Clear();
-    data_.AddAll(*compressor->output());
-    return true;
+
+  void SetCompressor(Compressor* compressor) {
+    compressor_ = compressor;
   }
-  int raw_size() { return raw_size_; }
 
- private:
-  i::List<char> data_;
-  int raw_size_;
-};
+  void SetOmit(bool omit) {
+    omit_ = omit;
+  }
 
+  void SetRawFiles(const char* raw_file, const char* raw_context_file) {
+    raw_file_ = GetFileDescriptorOrDie(raw_file);
+    raw_context_file_ = GetFileDescriptorOrDie(raw_context_file);
+  }
 
-class CppByteSink : public PartialSnapshotSink {
- public:
-  explicit CppByteSink(const char* snapshot_file) {
-    fp_ = i::OS::FOpen(snapshot_file, "wb");
-    if (fp_ == NULL) {
-      i::PrintF("Unable to write to snapshot file \"%s\"\n", snapshot_file);
-      exit(1);
-    }
+  void WriteSnapshot(const i::List<char>& snapshot_data,
+                     const i::Serializer& serializer,
+                     const i::List<char>& context_snapshot_data,
+                     const i::Serializer& context_serializer) const {
+    WriteFilePrefix();
+    WriteData("", snapshot_data, raw_file_);
+    WriteData("context_", context_snapshot_data, raw_context_file_);
+    WriteMeta("context_", context_serializer);
+    WriteMeta("", serializer);
+    WriteFileSuffix();
+  }
+
+ private:
+  void WriteFilePrefix() const {
     fprintf(fp_, "// Autogenerated snapshot file. Do not edit.\n\n");
     fprintf(fp_, "#include \"v8.h\"\n");
     fprintf(fp_, "#include \"platform.h\"\n\n");
     fprintf(fp_, "#include \"snapshot.h\"\n\n");
-    fprintf(fp_, "namespace v8 {\nnamespace internal {\n\n");
-    fprintf(fp_, "const byte Snapshot::data_[] = {");
+    fprintf(fp_, "namespace v8 {\n");
+    fprintf(fp_, "namespace internal {\n\n");
   }
 
-  virtual ~CppByteSink() {
-    fprintf(fp_, "const int Snapshot::size_ = %d;\n", Position());
-#ifdef COMPRESS_STARTUP_DATA_BZ2
-    fprintf(fp_, "const byte* Snapshot::raw_data_ = NULL;\n");
-    fprintf(fp_,
-            "const int Snapshot::raw_size_ = %d;\n\n",
-            raw_size());
-#else
-    fprintf(fp_,
-            "const byte* Snapshot::raw_data_ = Snapshot::data_;\n");
-    fprintf(fp_,
-            "const int Snapshot::raw_size_ = Snapshot::size_;\n\n");
-#endif
-    fprintf(fp_, "} }  // namespace v8::internal\n");
-    fclose(fp_);
+  void WriteFileSuffix() const {
+    fprintf(fp_, "}  // namespace internal\n");
+    fprintf(fp_, "}  // namespace v8\n");
+  }
+
+  void WriteData(const char* prefix,
+                 const i::List<char>& source_data,
+                 FILE* raw_file) const {
+    const i::List <char>* data_to_be_written = NULL;
+    i::List<char> compressed_data;
+    if (!compressor_) {
+      data_to_be_written = &source_data;
+    } else if (compressor_->Compress(source_data.ToVector())) {
+      compressed_data.AddAll(*compressor_->output());
+      data_to_be_written = &compressed_data;
+    } else {
+      i::PrintF("Compression failed. Aborting.\n");
+      exit(1);
+    }
+
+    ASSERT(data_to_be_written);
+    MaybeWriteRawFile(data_to_be_written, raw_file);
+    WriteData(prefix, source_data, data_to_be_written);
   }
 
-  void WriteSpaceUsed(
-      const char* prefix,
-      int new_space_used,
-      int pointer_space_used,
-      int data_space_used,
-      int code_space_used,
-      int map_space_used,
-      int cell_space_used,
-      int property_cell_space_used) {
-    fprintf(fp_,
-            "const int Snapshot::%snew_space_used_ = %d;\n",
-            prefix,
-            new_space_used);
-    fprintf(fp_,
-            "const int Snapshot::%spointer_space_used_ = %d;\n",
-            prefix,
-            pointer_space_used);
-    fprintf(fp_,
-            "const int Snapshot::%sdata_space_used_ = %d;\n",
-            prefix,
-            data_space_used);
-    fprintf(fp_,
-            "const int Snapshot::%scode_space_used_ = %d;\n",
-            prefix,
-            code_space_used);
-    fprintf(fp_,
-            "const int Snapshot::%smap_space_used_ = %d;\n",
-            prefix,
-            map_space_used);
-    fprintf(fp_,
-            "const int Snapshot::%scell_space_used_ = %d;\n",
-            prefix,
-            cell_space_used);
-    fprintf(fp_,
-            "const int Snapshot::%sproperty_cell_space_used_ = %d;\n",
-            prefix,
-            property_cell_space_used);
+  void MaybeWriteRawFile(const i::List<char>* data, FILE* raw_file) const {
+    if (!data || !raw_file)
+      return;
+
+    // Sanity check, whether i::List iterators truly return pointers to an
+    // internal array.
+    ASSERT(data->end() - data->begin() == data->length());
+
+    size_t written = fwrite(data->begin(), 1, data->length(), raw_file);
+    if (written != (size_t)data->length()) {
+      i::PrintF("Writing raw file failed.. Aborting.\n");
+      exit(1);
+    }
   }
 
-  void WritePartialSnapshot() {
-    int length = partial_sink_.Position();
-    fprintf(fp_, "};\n\n");
-    fprintf(fp_, "const int Snapshot::context_size_ = %d;\n",  length);
-#ifdef COMPRESS_STARTUP_DATA_BZ2
-    fprintf(fp_,
-            "const int Snapshot::context_raw_size_ = %d;\n",
-            partial_sink_.raw_size());
-#else
-    fprintf(fp_,
-            "const int Snapshot::context_raw_size_ = "
-            "Snapshot::context_size_;\n");
-#endif
-    fprintf(fp_, "const byte Snapshot::context_data_[] = {\n");
-    partial_sink_.Print(fp_);
-    fprintf(fp_, "};\n\n");
-#ifdef COMPRESS_STARTUP_DATA_BZ2
-    fprintf(fp_, "const byte* Snapshot::context_raw_data_ = NULL;\n");
-#else
-    fprintf(fp_, "const byte* Snapshot::context_raw_data_ ="
-            " Snapshot::context_data_;\n");
-#endif
+  void WriteData(const char* prefix,
+                 const i::List<char>& source_data,
+                 const i::List<char>* data_to_be_written) const {
+    fprintf(fp_, "const byte Snapshot::%sdata_[] = {\n", prefix);
+    if (!omit_)
+      WriteSnapshotData(data_to_be_written);
+    fprintf(fp_, "};\n");
+    fprintf(fp_, "const int Snapshot::%ssize_ = %d;\n", prefix,
+            data_to_be_written->length());
+
+    if (data_to_be_written == &source_data && !omit_) {
+      fprintf(fp_, "const byte* Snapshot::%sraw_data_ = Snapshot::%sdata_;\n",
+              prefix, prefix);
+      fprintf(fp_, "const int Snapshot::%sraw_size_ = Snapshot::%ssize_;\n",
+              prefix, prefix);
+    } else {
+      fprintf(fp_, "const byte* Snapshot::%sraw_data_ = NULL;\n", prefix);
+      fprintf(fp_, "const int Snapshot::%sraw_size_ = %d;\n",
+              prefix, source_data.length());
+    }
+    fprintf(fp_, "\n");
   }
 
-  void WriteSnapshot() {
-    Print(fp_);
+  void WriteMeta(const char* prefix, const i::Serializer& ser) const {
+    WriteSizeVar(ser, prefix, "new", i::NEW_SPACE);
+    WriteSizeVar(ser, prefix, "pointer", i::OLD_POINTER_SPACE);
+    WriteSizeVar(ser, prefix, "data", i::OLD_DATA_SPACE);
+    WriteSizeVar(ser, prefix, "code", i::CODE_SPACE);
+    WriteSizeVar(ser, prefix, "map", i::MAP_SPACE);
+    WriteSizeVar(ser, prefix, "cell", i::CELL_SPACE);
+    WriteSizeVar(ser, prefix, "property_cell", i::PROPERTY_CELL_SPACE);
+    fprintf(fp_, "\n");
   }
 
-  PartialSnapshotSink* partial_sink() { return &partial_sink_; }
+  void WriteSizeVar(const i::Serializer& ser, const char* prefix,
+                    const char* name, int space) const {
+    fprintf(fp_, "const int Snapshot::%s%s_space_used_ = %d;\n",
+            prefix, name, ser.CurrentAllocationAddress(space));
+  }
+
+  void WriteSnapshotData(const i::List<char>* data) const {
+    for (int i = 0; i < data->length(); i++) {
+      if ((i & 0x1f) == 0x1f)
+        fprintf(fp_, "\n");
+      if (i > 0)
+        fprintf(fp_, ",");
+      fprintf(fp_, "%u", static_cast<unsigned char>(data->at(i)));
+    }
+    fprintf(fp_, "\n");
+  }
+
+  FILE* GetFileDescriptorOrDie(const char* filename) {
+    FILE* fp = i::OS::FOpen(filename, "wb");
+    if (fp == NULL) {
+      i::PrintF("Unable to open file \"%s\" for writing.\n", filename);
+      exit(1);
+    }
+    return fp;
+  }
 
- private:
   FILE* fp_;
-  PartialSnapshotSink partial_sink_;
+  FILE* raw_file_;
+  FILE* raw_context_file_;
+  Compressor* compressor_;
+  bool omit_;
 };
 
 
@@ -369,46 +394,32 @@ int main(int argc, char** argv) {
       i::Heap::kNoGCFlags, "mksnapshot");
   i::Object* raw_context = *v8::Utils::OpenPersistent(context);
   context.Reset();
-  CppByteSink sink(argv[1]);
+
   // This results in a somewhat smaller snapshot, probably because it gets rid
   // of some things that are cached between garbage collections.
-  i::StartupSerializer ser(internal_isolate, &sink);
+  i::List<char> snapshot_data;
+  ListSnapshotSink snapshot_sink(&snapshot_data);
+  i::StartupSerializer ser(internal_isolate, &snapshot_sink);
   ser.SerializeStrongReferences();
 
-  i::PartialSerializer partial_ser(
-      internal_isolate, &ser, sink.partial_sink());
-  partial_ser.Serialize(&raw_context);
-
+  i::List<char> context_data;
+  ListSnapshotSink contex_sink(&context_data);
+  i::PartialSerializer context_ser(internal_isolate, &ser, &contex_sink);
+  context_ser.Serialize(&raw_context);
   ser.SerializeWeakReferences();
 
+  {
+    SnapshotWriter writer(argv[1]);
+    writer.SetOmit(i::FLAG_omit);
+    if (i::FLAG_raw_file && i::FLAG_raw_context_file)
+      writer.SetRawFiles(i::FLAG_raw_file, i::FLAG_raw_context_file);
 #ifdef COMPRESS_STARTUP_DATA_BZ2
-  BZip2Compressor compressor;
-  if (!sink.Compress(&compressor))
-    return 1;
-  if (!sink.partial_sink()->Compress(&compressor))
-    return 1;
+    BZip2Compressor bzip2;
+    writer.SetCompressor(&bzip2);
 #endif
-  sink.WriteSnapshot();
-  sink.WritePartialSnapshot();
-
-  sink.WriteSpaceUsed(
-      "context_",
-      partial_ser.CurrentAllocationAddress(i::NEW_SPACE),
-      partial_ser.CurrentAllocationAddress(i::OLD_POINTER_SPACE),
-      partial_ser.CurrentAllocationAddress(i::OLD_DATA_SPACE),
-      partial_ser.CurrentAllocationAddress(i::CODE_SPACE),
-      partial_ser.CurrentAllocationAddress(i::MAP_SPACE),
-      partial_ser.CurrentAllocationAddress(i::CELL_SPACE),
-      partial_ser.CurrentAllocationAddress(i::PROPERTY_CELL_SPACE));
-  sink.WriteSpaceUsed(
-      "",
-      ser.CurrentAllocationAddress(i::NEW_SPACE),
-      ser.CurrentAllocationAddress(i::OLD_POINTER_SPACE),
-      ser.CurrentAllocationAddress(i::OLD_DATA_SPACE),
-      ser.CurrentAllocationAddress(i::CODE_SPACE),
-      ser.CurrentAllocationAddress(i::MAP_SPACE),
-      ser.CurrentAllocationAddress(i::CELL_SPACE),
-      ser.CurrentAllocationAddress(i::PROPERTY_CELL_SPACE));
+    writer.WriteSnapshot(snapshot_data, ser, context_data, context_ser);
+  }
+
   isolate->Exit();
   isolate->Dispose();
   V8::Dispose();

diff --git a/src/serialize.h b/src/serialize.h
index 9e1d3d5..2d65436 100644 (file)
--- a/src/serialize.h
+++ b/src/serialize.h
@@ -464,7 +464,7 @@ class Serializer : public SerializerDeserializer {
   void VisitPointers(Object** start, Object** end);
   // You can call this after serialization to find out how much space was used
   // in each space.
-  int CurrentAllocationAddress(int space) {
+  int CurrentAllocationAddress(int space) const {
     ASSERT(space < kNumberOfSpaces);
     return fullness_[space];
   }