src: support snapshot in single executable applications

Author: joyeecheung

src/env.h

@@ -493,13 +493,17 @@ struct SnapshotMetadata {
 
 struct SnapshotData {
   enum class DataOwnership { kOwned, kNotOwned };
+  // TODO(joyeecheung): consider setting this in the blob. For now it's
+  // only set at deserialization time.
+  enum class IsForSea { kDefault, kIsSea };
 
   static const uint32_t kMagic = 0x143da19;
   static const SnapshotIndex kNodeVMContextIndex = 0;
   static const SnapshotIndex kNodeBaseContextIndex = kNodeVMContextIndex + 1;
   static const SnapshotIndex kNodeMainContextIndex = kNodeBaseContextIndex + 1;
 
   DataOwnership data_ownership = DataOwnership::kOwned;
+  IsForSea is_for_sea = IsForSea::kDefault;
 
   SnapshotMetadata metadata;
 
@@ -525,10 +529,12 @@ struct SnapshotData {
   bool Check() const;
   static bool FromFile(SnapshotData* out, FILE* in);
   static bool FromBlob(SnapshotData* out, const std::vector<char>& in);
+  static bool FromBlob(SnapshotData* out, std::string_view in);
   static const SnapshotData* FromEmbedderWrapper(
       const EmbedderSnapshotData* data);
   EmbedderSnapshotData::Pointer AsEmbedderWrapper() const;
 
+  static bool IsSnapshotBlob(std::string_view data);
   ~SnapshotData();
 };
 

src/node.cc

@@ -1158,8 +1158,28 @@ ExitCode LoadSnapshotDataAndRun(const SnapshotData** snapshot_data_ptr,
   ExitCode exit_code = result->exit_code_enum();
   // nullptr indicates there's no snapshot data.
   DCHECK_NULL(*snapshot_data_ptr);
-  // --snapshot-blob indicates that we are reading a customized snapshot.
-  if (!per_process::cli_options->snapshot_blob.empty()) {
+
+  bool has_cli_snapshot_blob = !per_process::cli_options->snapshot_blob.empty();
+#ifndef DISABLE_SINGLE_EXECUTABLE_APPLICATION
+  // TODO(joyeecheung): make the snapshot part of the blob, not **the** blob.
+  // https://github.com/nodejs/single-executable/discussions/58
+  if (sea::IsSingleExecutable()) {
+    const std::string_view sea_code = sea::FindSingleExecutableCode();
+    if (SnapshotData::IsSnapshotBlob(sea_code)) {
+      std::unique_ptr<SnapshotData> read_data =
+          std::make_unique<SnapshotData>();
+      if (SnapshotData::FromBlob(read_data.get(), sea_code)) {
+        read_data->is_for_sea = SnapshotData::IsForSea::kIsSea;
+        *snapshot_data_ptr = read_data.release();
+      } else {
+        fprintf(stderr, "Invalid snapshot data in single executable binary\n");
+        return ExitCode::kGenericUserError;
+      }
+    }
+  } else if (has_cli_snapshot_blob) {
+#else
+  if (has_cli_snapshot_blob) {
+#endif
     std::string filename = per_process::cli_options->snapshot_blob;
     FILE* fp = fopen(filename.c_str(), "rb");
     if (fp == nullptr) {
@@ -1239,6 +1259,9 @@ static ExitCode StartInternal(int argc, char** argv) {
     return GenerateAndWriteSnapshotData(&snapshot_data, result.get());
   }
 
+  // TODO(joyeecheung): parse SEA config here and generate a blob that
+  // can be injected into SEA if a certain flag is set.
+
   // Without --build-snapshot, we are in snapshot loading mode.
   return LoadSnapshotDataAndRun(&snapshot_data, result.get());
 }

src/node_main_instance.cc

@@ -89,7 +89,20 @@ void NodeMainInstance::Run(ExitCode* exit_code, Environment* env) {
   if (*exit_code == ExitCode::kNoFailure) {
     bool is_sea = false;
 #ifndef DISABLE_SINGLE_EXECUTABLE_APPLICATION
-    if (sea::IsSingleExecutable()) {
+    if (snapshot_data_->is_for_sea == SnapshotData::IsForSea::kIsSea) {
+      is_sea = true;
+      if (env->snapshot_deserialize_main().IsEmpty()) {
+        fprintf(
+            stderr,
+            "No deserialized main function found for the snapshot blob found"
+            " in single executable binary.\n");
+        *exit_code = ExitCode::kStartupSnapshotFailure;
+        return;
+      }
+      // LoadEnvironment would load and run the deserialized snapshot main
+      // function.
+      LoadEnvironment(env, StartExecutionCallback{});
+    } else if (sea::IsSingleExecutable()) {
       is_sea = true;
       LoadEnvironment(env, sea::FindSingleExecutableCode());
     }

src/node_sea.h

@@ -14,7 +14,7 @@ namespace sea {
 bool IsSingleExecutable();
 std::string_view FindSingleExecutableCode();
 std::tuple<int, char**> FixupArgsForSEA(int argc, char** argv);
-
+std::string_view FindSingleExecutableCode();
 }  // namespace sea
 }  // namespace node
 

src/node_snapshotable.cc

@@ -188,7 +188,7 @@ class SnapshotSerializerDeserializer {
 
 class SnapshotDeserializer : public SnapshotSerializerDeserializer {
  public:
-  explicit SnapshotDeserializer(const std::vector<char>& s)
+  explicit SnapshotDeserializer(const std::string_view s)
       : SnapshotSerializerDeserializer(), sink(s) {}
   ~SnapshotDeserializer() {}
 
@@ -247,7 +247,7 @@ class SnapshotDeserializer : public SnapshotSerializerDeserializer {
   }
 
   size_t read_total = 0;
-  const std::vector<char>& sink;
+  const std::string_view sink;
 
  private:
   // Helper for reading an array of numeric types.
@@ -884,10 +884,19 @@ bool SnapshotData::FromFile(SnapshotData* out, FILE* in) {
 }
 
 bool SnapshotData::FromBlob(SnapshotData* out, const std::vector<char>& in) {
+  return FromBlob(out, std::string_view(in.data(), in.size()));
+}
+
+bool SnapshotData::FromBlob(SnapshotData* out, const std::string_view in) {
   SnapshotDeserializer r(in);
   r.Debug("SnapshotData::FromBlob()\n");
 
   DCHECK_EQ(out->data_ownership, SnapshotData::DataOwnership::kOwned);
+  // For now we don't serialize this bit.
+  // XXX(joyeecheung): should we require snapshots being injected into
+  // SEA to be explicitly built for SEA, and check this bit at SEA
+  // start up?
+  DCHECK_EQ(out->is_for_sea, SnapshotData::IsForSea::kDefault);
 
   // Metadata
   uint32_t magic = r.Read<uint32_t>();
@@ -910,6 +919,11 @@ bool SnapshotData::FromBlob(SnapshotData* out, const std::vector<char>& in) {
   return true;
 }
 
+bool SnapshotData::IsSnapshotBlob(const std::string_view data) {
+  const uint32_t* ptr = reinterpret_cast<const uint32_t*>(data.data());
+  return (ptr[0] == kMagic);
+}
+
 bool SnapshotData::Check() const {
   if (metadata.node_version != per_process::metadata.versions.node) {
     fprintf(stderr,
@@ -1041,6 +1055,9 @@ static const int v8_snapshot_blob_size = )"
   // -- data_ownership begins --
   SnapshotData::DataOwnership::kNotOwned,
   // -- data_ownership ends --
+  // -- is_for_sea begins --
+  SnapshotData::IsForSea::kDefault,
+  // -- is_for_sea ends --
   // -- metadata begins --
 )" << data->metadata
      << R"(,