src: use Blob{Des|S}erializer for SEA blobs

src/blob_serializer_deserializer-inl.h

@@ -13,8 +13,8 @@
 
 #include "debug_utils-inl.h"
 
-// This is related to the blob that is used in snapshots and has nothing to do
-// with `node_blob.h`.
+// This is related to the blob that is used in snapshots and single executable
+// applications and has nothing to do with `node_blob.h`.
 
 namespace node {
 
@@ -130,22 +130,20 @@ std::vector<T> BlobDeserializer<Impl>::ReadVector() {
 
 template <typename Impl>
 std::string BlobDeserializer<Impl>::ReadString() {
-  size_t length = ReadArithmetic<size_t>();
-
-  if (is_debug) {
-    Debug("ReadString(), length=%d: ", length);
-  }
+  std::string_view view = ReadStringView();
+  Debug("ReadString(), length=%zu: \"%s\"\n", view.size(), view.data());
+  return std::string(view);
+}
 
-  CHECK_GT(length, 0);  // There should be no empty strings.
-  MallocedBuffer<char> buf(length + 1);
-  memcpy(buf.data, sink.data() + read_total, length + 1);
-  std::string result(buf.data, length);  // This creates a copy of buf.data.
+template <typename Impl>
+std::string_view BlobDeserializer<Impl>::ReadStringView() {
+  size_t length = ReadArithmetic<size_t>();
+  Debug("ReadStringView(), length=%zu: ", length);
 
-  if (is_debug) {
-    Debug("\"%s\", read %zu bytes\n", result.c_str(), length + 1);
-  }
+  std::string_view result(sink.data() + read_total, length);
+  Debug("%p, read %zu bytes\n", result.data(), result.size());
 
-  read_total += length + 1;
+  read_total += length;
   return result;
 }
 
@@ -262,26 +260,25 @@ size_t BlobSerializer<Impl>::WriteVector(const std::vector<T>& data) {
 // [  4/8 bytes     ] length
 // [ |length| bytes ] contents
 template <typename Impl>
-size_t BlobSerializer<Impl>::WriteString(const std::string& data) {
-  CHECK_GT(data.size(), 0);  // No empty strings should be written.
+size_t BlobSerializer<Impl>::WriteStringView(std::string_view data) {
   size_t written_total = WriteArithmetic<size_t>(data.size());
-  if (is_debug) {
-    std::string str = ToStr(data);
-    Debug("WriteString(), length=%zu: \"%s\"\n", data.size(), data.c_str());
-  }
+  Debug("WriteStringView(), length=%zu: %p\n", data.size(), data.data());
 
-  // Write the null-terminated string.
-  size_t length = data.size() + 1;
-  sink.insert(sink.end(), data.c_str(), data.c_str() + length);
+  size_t length = data.size();
+  sink.insert(sink.end(), data.data(), data.data() + length);
   written_total += length;
 
-  if (is_debug) {
-    Debug("WriteString() wrote %zu bytes\n", written_total);
-  }
+  Debug("WriteStringView() wrote %zu bytes\n", written_total);
 
   return written_total;
 }
 
+template <typename Impl>
+size_t BlobSerializer<Impl>::WriteString(const std::string& data) {
+  Debug("WriteString(), length=%zu: \"%s\"\n", data.size(), data.data());
+  return WriteStringView(data);
+}
+
 // Helper for writing an array of numeric types.
 template <typename Impl>
 template <typename T>

src/blob_serializer_deserializer.h

@@ -6,8 +6,8 @@
 
 #if defined(NODE_WANT_INTERNALS) && NODE_WANT_INTERNALS
 
-// This is related to the blob that is used in snapshots and has nothing to do
-// with `node_blob.h`.
+// This is related to the blob that is used in snapshots and single executable
+// applications and has nothing to do with `node_blob.h`.
 
 namespace node {
 
@@ -52,7 +52,9 @@ class BlobDeserializer : public BlobSerializerDeserializer {
   template <typename T>
   std::vector<T> ReadVector();
 
+  // ReadString() creates a copy of the data. ReadStringView() doesn't.
   std::string ReadString();
+  std::string_view ReadStringView();
 
   // Helper for reading an array of numeric types.
   template <typename T>
@@ -77,11 +79,7 @@ template <typename Impl>
 class BlobSerializer : public BlobSerializerDeserializer {
  public:
   explicit BlobSerializer(bool is_debug_v)
-      : BlobSerializerDeserializer(is_debug_v) {
-    // Currently the snapshot blob built with an empty script is around 4MB.
-    // So use that as the default sink size.
-    sink.reserve(4 * 1024 * 1024);
-  }
+      : BlobSerializerDeserializer(is_debug_v) {}
   ~BlobSerializer() {}
 
   Impl* impl() { return static_cast<Impl*>(this); }
@@ -102,6 +100,7 @@ class BlobSerializer : public BlobSerializerDeserializer {
   // The layout of a written string:
   // [  4/8 bytes     ] length
   // [ |length| bytes ] contents
+  size_t WriteStringView(std::string_view data);
   size_t WriteString(const std::string& data);
 
   // Helper for writing an array of numeric types.

src/debug_utils.h

@@ -48,6 +48,7 @@ void NODE_EXTERN_PRIVATE FWrite(FILE* file, const std::string& str);
   V(INSPECTOR_PROFILER)                                                        \
   V(CODE_CACHE)                                                                \
   V(NGTCP2_DEBUG)                                                              \
+  V(SEA)                                                                       \
   V(WASI)                                                                      \
   V(MKSNAPSHOT)
 

src/node_main_instance.cc

@@ -87,14 +87,16 @@ ExitCode NodeMainInstance::Run() {
 
 void NodeMainInstance::Run(ExitCode* exit_code, Environment* env) {
   if (*exit_code == ExitCode::kNoFailure) {
-    bool is_sea = false;
+    bool runs_sea_code = false;
 #ifndef DISABLE_SINGLE_EXECUTABLE_APPLICATION
     if (sea::IsSingleExecutable()) {
-      is_sea = true;
-      LoadEnvironment(env, sea::FindSingleExecutableCode());
+      runs_sea_code = true;
+      sea::SeaResource sea = sea::FindSingleExecutableResource();
+      std::string_view code = sea.code;
+      LoadEnvironment(env, code);
     }
 #endif
-    if (!is_sea) {
+    if (!runs_sea_code) {
       LoadEnvironment(env, StartExecutionCallback{});
     }
 

src/node_sea.cc

@@ -1,5 +1,6 @@
 #include "node_sea.h"
 
+#include "blob_serializer_deserializer-inl.h"
 #include "debug_utils-inl.h"
 #include "env-inl.h"
 #include "json_parser.h"
@@ -34,16 +35,6 @@ namespace node {
 namespace sea {
 
 namespace {
-// A special number that will appear at the beginning of the single executable
-// preparation blobs ready to be injected into the binary. We use this to check
-// that the data given to us are intended for building single executable
-// applications.
-const uint32_t kMagic = 0x143da20;
-
-enum class SeaFlags : uint32_t {
-  kDefault = 0,
-  kDisableExperimentalSeaWarning = 1 << 0,
-};
 
 SeaFlags operator|(SeaFlags x, SeaFlags y) {
   return static_cast<SeaFlags>(static_cast<uint32_t>(x) |
@@ -59,47 +50,91 @@ SeaFlags operator|=(/* NOLINT (runtime/references) */ SeaFlags& x, SeaFlags y) {
   return x = x | y;
 }
 
-struct SeaResource {
-  SeaFlags flags = SeaFlags::kDefault;
-  std::string_view code;
-  static constexpr size_t kHeaderSize = sizeof(kMagic) + sizeof(SeaFlags);
+class SeaSerializer : public BlobSerializer<SeaSerializer> {
+ public:
+  SeaSerializer()
+      : BlobSerializer<SeaSerializer>(
+            per_process::enabled_debug_list.enabled(DebugCategory::SEA)) {}
+
+  template <typename T,
+            std::enable_if_t<!std::is_same<T, std::string>::value>* = nullptr,
+            std::enable_if_t<!std::is_arithmetic<T>::value>* = nullptr>
+  size_t Write(const T& data);
 };
 
-SeaResource FindSingleExecutableResource() {
+template <>
+size_t SeaSerializer::Write(const SeaResource& sea) {
+  sink.reserve(SeaResource::kHeaderSize + sea.code.size());
+
+  size_t written_total = WriteArithmetic<uint32_t>(kMagic);
+  written_total += WriteArithmetic<uint32_t>(static_cast<uint32_t>(sea.flags));
+  DCHECK_EQ(written_total, SeaResource::kHeaderSize);
+
+  written_total += WriteStringView(sea.code);
+  return written_total;
+}
+
+class SeaDeserializer : public BlobDeserializer<SeaDeserializer> {
+ public:
+  explicit SeaDeserializer(std::string_view v)
+      : BlobDeserializer<SeaDeserializer>(
+            per_process::enabled_debug_list.enabled(DebugCategory::SEA), v) {}
+
+  template <typename T,
+            std::enable_if_t<!std::is_same<T, std::string>::value>* = nullptr,
+            std::enable_if_t<!std::is_arithmetic<T>::value>* = nullptr>
+  T Read();
+};
+
+template <>
+SeaResource SeaDeserializer::Read() {
+  uint32_t magic = ReadArithmetic<uint32_t>();
+  CHECK_EQ(magic, kMagic);
+  SeaFlags flags(static_cast<SeaFlags>(ReadArithmetic<uint32_t>()));
+  Debug("Read SEA flag %d", static_cast<uint32_t>(flags));
+  CHECK_EQ(read_total, SeaResource::kHeaderSize);
+
+  std::string_view code = ReadStringView();
+  Debug("Read SEA resource code %p %zu\n", code.data(), code.size());
+  return {flags, code};
+}
+
+std::string_view FindSingleExecutableBlob() {
   CHECK(IsSingleExecutable());
-  static const SeaResource sea_resource = []() -> SeaResource {
+  static const std::string_view result = []() -> std::string_view {
     size_t size;
 #ifdef __APPLE__
     postject_options options;
     postject_options_init(&options);
     options.macho_segment_name = "NODE_SEA";
-    const char* code = static_cast<const char*>(
+    const char* blob = static_cast<const char*>(
         postject_find_resource("NODE_SEA_BLOB", &size, &options));
 #else
-    const char* code = static_cast<const char*>(
+    const char* blob = static_cast<const char*>(
         postject_find_resource("NODE_SEA_BLOB", &size, nullptr));
 #endif
-    uint32_t first_word = reinterpret_cast<const uint32_t*>(code)[0];
-    CHECK_EQ(first_word, kMagic);
-    SeaFlags flags{
-        reinterpret_cast<const SeaFlags*>(code + sizeof(first_word))[0]};
-    // TODO(joyeecheung): do more checks here e.g. matching the versions.
-    return {
-        flags,
-        {
-            code + SeaResource::kHeaderSize,
-            size - SeaResource::kHeaderSize,
-        },
-    };
+    return {blob, size};
   }();
-  return sea_resource;
+  per_process::Debug(DebugCategory::SEA,
+                     "Found SEA resource %p %zu\n",
+                     result.data(),
+                     result.size());
+  return result;
 }
 
-}  // namespace
+}  // anonymous namespace
 
-std::string_view FindSingleExecutableCode() {
-  SeaResource sea_resource = FindSingleExecutableResource();
-  return sea_resource.code;
+SeaResource FindSingleExecutableResource() {
+  static const SeaResource sea_resource = []() -> SeaResource {
+    std::string_view blob = FindSingleExecutableBlob();
+    per_process::Debug(DebugCategory::SEA,
+                       "Found SEA resource %p %zu\n",
+                       blob.data(),
+                       blob.size());
+    SeaDeserializer deserializer(blob);
+    return deserializer.Read<SeaResource>();
+  }();
+  return sea_resource;
 }
 
 bool IsSingleExecutable() {
@@ -194,51 +229,46 @@ std::optional<SeaConfig> ParseSingleExecutableConfig(
   return result;
 }
 
-bool GenerateSingleExecutableBlob(const SeaConfig& config) {
+ExitCode GenerateSingleExecutableBlob(const SeaConfig& config) {
   std::string main_script;
   // TODO(joyeecheung): unify the file utils.
   int r = ReadFileSync(&main_script, config.main_path.c_str());
   if (r != 0) {
     const char* err = uv_strerror(r);
     FPrintF(stderr, "Cannot read main script %s:%s\n", config.main_path, err);
-    return false;
+    return ExitCode::kGenericUserError;
   }
 
-  std::vector<char> sink;
-  // TODO(joyeecheung): reuse the SnapshotSerializerDeserializer for this.
-  sink.reserve(SeaResource::kHeaderSize + main_script.size());
-  const char* pos = reinterpret_cast<const char*>(&kMagic);
-  sink.insert(sink.end(), pos, pos + sizeof(kMagic));
-  pos = reinterpret_cast<const char*>(&(config.flags));
-  sink.insert(sink.end(), pos, pos + sizeof(SeaFlags));
-  sink.insert(
-      sink.end(), main_script.data(), main_script.data() + main_script.size());
-
-  uv_buf_t buf = uv_buf_init(sink.data(), sink.size());
+  SeaResource sea{config.flags, main_script};
+
+  SeaSerializer serializer;
+  serializer.Write(sea);
+
+  uv_buf_t buf = uv_buf_init(serializer.sink.data(), serializer.sink.size());
   r = WriteFileSync(config.output_path.c_str(), buf);
   if (r != 0) {
     const char* err = uv_strerror(r);
     FPrintF(stderr, "Cannot write output to %s:%s\n", config.output_path, err);
-    return false;
+    return ExitCode::kGenericUserError;
   }
 
   FPrintF(stderr,
           "Wrote single executable preparation blob to %s\n",
           config.output_path);
-  return true;
+  return ExitCode::kNoFailure;
 }
 
 }  // anonymous namespace
 
 ExitCode BuildSingleExecutableBlob(const std::string& config_path) {
   std::optional<SeaConfig> config_opt =
       ParseSingleExecutableConfig(config_path);
-  if (!config_opt.has_value() ||
-      !GenerateSingleExecutableBlob(config_opt.value())) {
-    return ExitCode::kGenericUserError;
+  if (config_opt.has_value()) {
+    ExitCode code = GenerateSingleExecutableBlob(config_opt.value());
+    return code;
   }
 
-  return ExitCode::kNoFailure;
+  return ExitCode::kGenericUserError;
 }
 
 void Initialize(Local<Object> target,

src/node_sea.h

@@ -11,9 +11,26 @@
 
 namespace node {
 namespace sea {
+// A special number that will appear at the beginning of the single executable
+// preparation blobs ready to be injected into the binary. We use this to check
+// that the data given to us are intended for building single executable
+// applications.
+const uint32_t kMagic = 0x143da20;
+
+enum class SeaFlags : uint32_t {
+  kDefault = 0,
+  kDisableExperimentalSeaWarning = 1 << 0,
+};
+
+struct SeaResource {
+  SeaFlags flags = SeaFlags::kDefault;
+  std::string_view code;
+
+  static constexpr size_t kHeaderSize = sizeof(kMagic) + sizeof(SeaFlags);
+};
 
 bool IsSingleExecutable();
-std::string_view FindSingleExecutableCode();
+SeaResource FindSingleExecutableResource();
 std::tuple<int, char**> FixupArgsForSEA(int argc, char** argv);
 node::ExitCode BuildSingleExecutableBlob(const std::string& config_path);
 }  // namespace sea