Add expectedMeasurement and measure CLI command

Author: nhynes

src/workerd/server/server.c++

@@ -2245,6 +2245,21 @@ private:
 
 // =======================================================================================
 
+kj::Array<kj::byte> measureConfig(config::Worker::Reader& config) {
+  auto confWords = capnp::canonicalize(config);
+  EVP_MD_CTX* digestCtx = EVP_MD_CTX_new();
+  const EVP_MD* digestAlg = EVP_sha384();
+  KJ_DEFER(EVP_MD_CTX_free(digestCtx));
+  KJ_REQUIRE(EVP_DigestInit_ex(digestCtx, digestAlg, nullptr));
+  KJ_REQUIRE(EVP_DigestUpdate(digestCtx,
+        reinterpret_cast<kj::byte*>(confWords.begin()), confWords.size() * sizeof(capnp::word)));
+  auto digest = kj::heapArray<kj::byte>(EVP_MD_CTX_size(digestCtx));
+  uint digestSize = 0;
+  KJ_REQUIRE(EVP_DigestFinal_ex(digestCtx, digest.begin(), &digestSize));
+  KJ_ASSERT(digestSize == digest.size());
+  return digest;
+}
+
 class Server::WorkerdApiService final: public Service, private WorkerInterface {
   // Service used when the service is configured as network service.
 
@@ -2269,20 +2284,31 @@ private:
       return requestBody.readAllText().then([this, &headers, &response](auto confJson) {
           capnp::MallocMessageBuilder confArena;
           capnp::JsonCodec json;
-          json.handleByAnnotation<config::Worker>();
-          auto conf = confArena.initRoot<config::Worker>();
+          json.handleByAnnotation<config::NewWorker>();
+          auto conf = confArena.initRoot<config::NewWorker>();
           json.decode(confJson, conf);
 
           kj::String id = workerd::randomUUID(kj::none);
 
           server.actorConfigs.insert(kj::str(id), {});
 
+          kj::Maybe<kj::Array<kj::byte>> expectedMeasurement = kj::none;
+          if (conf.hasExpectedMeasurement()) {
+            auto res = kj::decodeHex(conf.getExpectedMeasurement());
+            if (res.hadErrors) {
+              auto out = response.send(400, "Bad Request", headers, kj::none);
+              auto errMsg = "invalid expected measurement"_kjc.asBytes();
+              return out->write(errMsg.begin(), errMsg.size());
+            }
+            expectedMeasurement = kj::mv(res);
+          }
+
           kj::Maybe<kj::String> configError = kj::none;
           auto workerService = server.makeWorker(
-              id, conf.asReader(), {},
-              [&configError](auto err) {
-              configError = kj::mv(err);
-              });
+            id, conf.getWorker().asReader(), {},
+            [&configError](auto err) { configError = kj::mv(err); },
+            kj::mv(expectedMeasurement)
+          );
           KJ_IF_SOME(err, configError) {
             throw KJ_EXCEPTION(FAILED, err);
           }
@@ -2704,8 +2730,19 @@ void Server::abortAllActors() {
 
 kj::Own<Server::Service> Server::makeWorker(kj::StringPtr name, config::Worker::Reader conf,
     capnp::List<config::Extension>::Reader extensions,
-    kj::Function<void(kj::String)> reportConfigError) {
+    kj::Function<void(kj::String)> reportConfigError,
+    kj::Maybe<kj::Array<kj::byte>> expectedMeasurement) {
   TRACE_EVENT("workerd", "Server::makeWorker()", "name", name.cStr());
+
+  KJ_IF_SOME(expected, expectedMeasurement) {
+    auto measurement = measureConfig(conf);
+    if (measurement != expected) {
+      reportConfigError(kj::str("service ", name, ": measurement mismatched.",
+            " expected ", kj::encodeHex(expected), " but got ", kj::encodeHex(measurement)));
+      return makeInvalidConfigService();
+    }
+  }
+
   auto& localActorConfigs = KJ_ASSERT_NONNULL(actorConfigs.find(name));
 
   struct ErrorReporter: public Worker::ValidationErrorReporter {

src/workerd/server/server.h

@@ -173,7 +173,8 @@ class Server: private kj::TaskSet::ErrorHandler {
       kj::HttpHeaderTable::Builder& headerTableBuilder);
   kj::Own<Service> makeWorker(kj::StringPtr name, config::Worker::Reader conf,
       capnp::List<config::Extension>::Reader extensions,
-      kj::Function<void(kj::String)> reportConfigError);
+      kj::Function<void(kj::String)> reportConfigError,
+      kj::Maybe<kj::Array<kj::byte>> expectedMeasurement = kj::none);
   kj::Own<Service> makeService(
       config::Service::Reader conf,
       kj::HttpHeaderTable::Builder& headerTableBuilder,
@@ -257,4 +258,6 @@ class EmptyReadOnlyActorStorageImpl final: public rpc::ActorStorage::Stage::Serv
   };
 };
 
+kj::Array<kj::byte> measureConfig(config::Worker::Reader& config);
+
 }  // namespace workerd::server

src/workerd/server/workerd.c++

@@ -549,6 +549,8 @@ public:
               "run the server")
           .addSubCommand("compile", KJ_BIND_METHOD(*this, getCompile),
               "create a self-contained binary")
+          .addSubCommand("measure", KJ_BIND_METHOD(*this, getMeasure),
+              "measure the provided worker configuration")
           .addSubCommand("test", KJ_BIND_METHOD(*this, getTest),
               "run unit tests")
           .build();
@@ -558,10 +560,13 @@ public:
       // "explain": Produces human-friendly description of the config.
     } else {
       // We already have a config, meaning this must be a compiled binary.
-      auto builder = kj::MainBuilder(context, getVersionString(),
-            "Serve requests based on the compiled config.",
-            "This binary has an embedded configuration.");
-      return addServeOptions(builder);
+      return kj::MainBuilder(context, getVersionString(),
+            "Runs the Workers JavaScript/Wasm runtime.")
+          .addSubCommand("serve", KJ_BIND_METHOD(*this, getServe),
+              "run the server")
+          .addSubCommand("measure", KJ_BIND_METHOD(*this, getMeasure),
+              "measure the provided worker configuration")
+          .build();
     }
   }
 
@@ -688,6 +693,16 @@ public:
         .build();
   }
 
+  kj::MainFunc getMeasure() {
+    return kj::MainBuilder(context, getVersionString(),
+          "Measures the provided worker config and outputs the hash.",
+          "Loads a worker's code and config in the same way as would be done by the "
+          "`workerd.createWorker` method, hashes the full config, and returns the hash.")
+        .expectArg("<config-file>", CLI_METHOD(parseWorkerConfigFile))
+        .callAfterParsing(CLI_METHOD(measure))
+        .build();
+  }
+
   void addImportPath(kj::StringPtr pathStr) {
     auto path = fs->getCurrentPath().evalNative(pathStr);
     if (fs->getRoot().tryOpenSubdir(path) != kj::none) {
@@ -811,6 +826,19 @@ public:
   }
 
   void parseConfigFile(kj::StringPtr pathStr) {
+    config = parseCapnpConfig<config::Config>(pathStr);
+    // We'll fail at getConfig() if there are multiple top level Config objects.
+    // The error message says that you have to specify which config to use, but
+    // it's not clear that there is any mechanism to do that.
+    util::Autogate::initAutogate(getConfig().getAutogates());
+  }
+
+  void parseWorkerConfigFile(kj::StringPtr pathStr) {
+    workerConfig = parseCapnpConfig<config::Worker>(pathStr);
+  }
+
+  template <typename T>
+  kj::Maybe<typename T::Reader> parseCapnpConfig(kj::StringPtr pathStr) {
     if (pathStr == "-") {
       // Read from stdin.
 
@@ -826,8 +854,8 @@ public:
 #else
       auto reader = kj::heap<capnp::StreamFdMessageReader>(STDIN_FILENO, CONFIG_READER_OPTIONS);
 #endif
-      config = reader->getRoot<config::Config>();
       configOwner = kj::mv(reader);
+      return reader->getRoot<T>();
     } else {
       // Read file from disk.
       auto path = fs->getCurrentPath().evalNative(pathStr);
@@ -840,11 +868,11 @@ public:
                                   mapping.size() / sizeof(capnp::word));
         auto reader = kj::heap<capnp::FlatArrayMessageReader>(words, CONFIG_READER_OPTIONS)
             .attach(kj::mv(mapping));
-        config = reader->getRoot<config::Config>();
         configOwner = kj::mv(reader);
+        return reader->getRoot<T>();
       } else {
         // Interpret as schema file.
-        schemaParser.loadCompiledTypeAndDependencies<config::Config>();
+        schemaParser.loadCompiledTypeAndDependencies<T>();
 
         parsedSchema = schemaParser.parseFile(
             kj::heap<SchemaFileImpl>(fs->getRoot(), fs->getCurrentPath(),
@@ -857,18 +885,14 @@ public:
             auto constSchema = nested.asConst();
             auto type = constSchema.getType();
             if (type.isStruct() &&
-                type.asStruct().getProto().getId() == capnp::typeId<config::Config>()) {
-              topLevelConfigConstants.add(constSchema);
+                type.asStruct().getProto().getId() == capnp::typeId<T>()) {
+              return constSchema.as<T>();
             }
           }
         }
+        return kj::none;
       }
     }
-
-    // We'll fail at getConfig() if there are multiple top level Config objects.
-    // The error message says that you have to specify which config to use, but
-    // it's not clear that there is any mechanism to do that.
-    util::Autogate::initAutogate(getConfig().getAutogates());
   }
 
   void setConstName(kj::StringPtr name) {
@@ -1075,6 +1099,15 @@ public:
     }
   }
 
+  void measure() {
+    if (hadErrors) context.exit();
+    auto measurement = workerd::server::measureConfig(
+        KJ_UNWRAP_OR(workerConfig, CLI_ERROR("no worker config provided")));
+    auto measurementHex = kj::encodeHex(measurement);
+    kj::FdOutputStream out{STDOUT_FILENO};
+    out.write(measurementHex.asBytes().begin(), measurementHex.size());
+  }
+
   [[noreturn]] void serve() noexcept {
     serveImpl([&](jsg::V8System& v8System, config::Config::Reader config) {
 #if _WIN32
@@ -1164,6 +1197,7 @@ private:
 
   kj::Own<void> configOwner;  // backing object for `config`, if it's not `schemaParser`.
   kj::Maybe<config::Config::Reader> config;
+  kj::Maybe<config::Worker::Reader> workerConfig;
 
   kj::Vector<int> inheritedFds;
 

src/workerd/server/workerd.capnp

@@ -915,3 +915,15 @@ struct Extension {
     # Raw source code of ES module.
   }
 }
+
+# ========================================================================================
+# Unsafe Evalualuation
+
+struct NewWorker {
+  worker @0 :Worker;
+
+  expectedMeasurement @1 :Text;
+  # The expected measurement of the worker. The measurement is computed as the SHA-512 hash of the
+  # binary format of the worker's capnp config. If this field is set and the measurement does not
+  # match, the worker will not run.
+}