crypto: use EVP_MD_fetch and cache EVP_MD for hashes

joyeecheung · richardlau · commit 68e8b2c49296 · 2024-03-25T19:23:38.000Z
On OpenSSL 3, migrate from EVP_get_digestbyname() to EVP_MD_fetch() to get the implementation and use a per-Environment cache for it. The EVP_MDs are freed during Environment cleanup. Drive-by: declare the smart pointer for EVP_MD_CTX as EVPMDCtxPointer instead of EVPMDPointer to avoid confusion with EVP_MD pointers. PR-URL: #51034 Refs: https://www.openssl.org/docs/man3.0/man7/crypto.html#Explicit-fetching Refs: nodejs/performance#136 Reviewed-By: James M Snell <jasnell@gmail.com>
diff --git a/lib/internal/crypto/hash.js b/lib/internal/crypto/hash.js
@@ -19,6 +19,8 @@ const {
   normalizeHashName,
   validateMaxBufferLength,
   kHandle,
+  getCachedHashId,
+  getHashCache,
 } = require('internal/crypto/util');
 
 const {
@@ -59,13 +61,17 @@ const kFinalized = Symbol('kFinalized');
 function Hash(algorithm, options) {
   if (!new.target)
     return new Hash(algorithm, options);
-  if (!(algorithm instanceof _Hash))
+  const isCopy = algorithm instanceof _Hash;
+  if (!isCopy)
     validateString(algorithm, 'algorithm');
   const xofLen = typeof options === 'object' && options !== null ?
     options.outputLength : undefined;
   if (xofLen !== undefined)
     validateUint32(xofLen, 'options.outputLength');
-  this[kHandle] = new _Hash(algorithm, xofLen);
+  // Lookup the cached ID from JS land because it's faster than decoding
+  // the string in C++ land.
+  const algorithmId = isCopy ? -1 : getCachedHashId(algorithm);
+  this[kHandle] = new _Hash(algorithm, xofLen, algorithmId, getHashCache());
   this[kState] = {
     [kFinalized]: false,
   };
diff --git a/lib/internal/crypto/util.js b/lib/internal/crypto/util.js
@@ -29,6 +29,7 @@ const {
   getHashes: _getHashes,
   setEngine: _setEngine,
   secureHeapUsed: _secureHeapUsed,
+  getCachedAliases,
 } = internalBinding('crypto');
 
 const { getOptionValue } = require('internal/options');
@@ -66,6 +67,13 @@ const {
   lazyDOMException,
 } = require('internal/util');
 
+const {
+  namespace: {
+    isBuildingSnapshot,
+    addSerializeCallback,
+  },
+} = require('internal/v8/startup_snapshot');
+
 const {
   isDataView,
   isArrayBufferView,
@@ -87,6 +95,23 @@ function toBuf(val, encoding) {
   return val;
 }
 
+let _hashCache;
+function getHashCache() {
+  if (_hashCache === undefined) {
+    _hashCache = getCachedAliases();
+    if (isBuildingSnapshot()) {
+      // For dynamic linking, clear the map.
+      addSerializeCallback(() => { _hashCache = undefined; });
+    }
+  }
+  return _hashCache;
+}
+
+function getCachedHashId(algorithm) {
+  const result = getHashCache()[algorithm];
+  return result === undefined ? -1 : result;
+}
+
 const getCiphers = cachedResult(() => filterDuplicateStrings(_getCiphers()));
 const getHashes = cachedResult(() => filterDuplicateStrings(_getHashes()));
 const getCurves = cachedResult(() => filterDuplicateStrings(_getCurves()));
@@ -574,4 +599,6 @@ module.exports = {
   getStringOption,
   getUsagesUnion,
   secureHeapUsed,
+  getCachedHashId,
+  getHashCache,
 };
diff --git a/src/crypto/README.md b/src/crypto/README.md
@@ -79,7 +79,7 @@ using SSLPointer = DeleteFnPtr<SSL, SSL_free>;
 using PKCS8Pointer = DeleteFnPtr<PKCS8_PRIV_KEY_INFO, PKCS8_PRIV_KEY_INFO_free>;
 using EVPKeyPointer = DeleteFnPtr<EVP_PKEY, EVP_PKEY_free>;
 using EVPKeyCtxPointer = DeleteFnPtr<EVP_PKEY_CTX, EVP_PKEY_CTX_free>;
-using EVPMDPointer = DeleteFnPtr<EVP_MD_CTX, EVP_MD_CTX_free>;
+using EVPMDCtxPointer = DeleteFnPtr<EVP_MD_CTX, EVP_MD_CTX_free>;
 using RSAPointer = DeleteFnPtr<RSA, RSA_free>;
 using ECPointer = DeleteFnPtr<EC_KEY, EC_KEY_free>;
 using BignumPointer = DeleteFnPtr<BIGNUM, BN_clear_free>;
diff --git a/src/crypto/crypto_hash.cc b/src/crypto/crypto_hash.cc
@@ -14,11 +14,13 @@ namespace node {
 using v8::Context;
 using v8::FunctionCallbackInfo;
 using v8::FunctionTemplate;
+using v8::Int32;
 using v8::Isolate;
 using v8::Just;
 using v8::Local;
 using v8::Maybe;
 using v8::MaybeLocal;
+using v8::Name;
 using v8::Nothing;
 using v8::Object;
 using v8::Uint32;
@@ -34,22 +36,170 @@ void Hash::MemoryInfo(MemoryTracker* tracker) const {
   tracker->TrackFieldWithSize("md", digest_ ? md_len_ : 0);
 }
 
+#if OPENSSL_VERSION_MAJOR >= 3
+void PushAliases(const char* name, void* data) {
+  static_cast<std::vector<std::string>*>(data)->push_back(name);
+}
+
+EVP_MD* GetCachedMDByID(Environment* env, size_t id) {
+  CHECK_LT(id, env->evp_md_cache.size());
+  EVP_MD* result = env->evp_md_cache[id].get();
+  CHECK_NOT_NULL(result);
+  return result;
+}
+
+struct MaybeCachedMD {
+  EVP_MD* explicit_md = nullptr;
+  const EVP_MD* implicit_md = nullptr;
+  int32_t cache_id = -1;
+};
+
+MaybeCachedMD FetchAndMaybeCacheMD(Environment* env, const char* search_name) {
+  const EVP_MD* implicit_md = EVP_get_digestbyname(search_name);
+  if (!implicit_md) return {nullptr, nullptr, -1};
+
+  const char* real_name = EVP_MD_get0_name(implicit_md);
+  if (!real_name) return {nullptr, implicit_md, -1};
+
+  auto it = env->alias_to_md_id_map.find(real_name);
+  if (it != env->alias_to_md_id_map.end()) {
+    size_t id = it->second;
+    return {GetCachedMDByID(env, id), implicit_md, static_cast<int32_t>(id)};
+  }
+
+  // EVP_*_fetch() does not support alias names, so we need to pass it the
+  // real/original algorithm name.
+  // We use EVP_*_fetch() as a filter here because it will only return an
+  // instance if the algorithm is supported by the public OpenSSL APIs (some
+  // algorithms are used internally by OpenSSL and are also passed to this
+  // callback).
+  EVP_MD* explicit_md = EVP_MD_fetch(nullptr, real_name, nullptr);
+  if (!explicit_md) return {nullptr, implicit_md, -1};
+
+  // Cache the EVP_MD* fetched.
+  env->evp_md_cache.emplace_back(explicit_md);
+  size_t id = env->evp_md_cache.size() - 1;
+
+  // Add all the aliases to the map to speed up next lookup.
+  std::vector<std::string> aliases;
+  EVP_MD_names_do_all(explicit_md, PushAliases, &aliases);
+  for (const auto& alias : aliases) {
+    env->alias_to_md_id_map.emplace(alias, id);
+  }
+  env->alias_to_md_id_map.emplace(search_name, id);
+
+  return {explicit_md, implicit_md, static_cast<int32_t>(id)};
+}
+
+void SaveSupportedHashAlgorithmsAndCacheMD(const EVP_MD* md,
+                                           const char* from,
+                                           const char* to,
+                                           void* arg) {
+  if (!from) return;
+  Environment* env = static_cast<Environment*>(arg);
+  auto result = FetchAndMaybeCacheMD(env, from);
+  if (result.explicit_md) {
+    env->supported_hash_algorithms.push_back(from);
+  }
+}
+
+#else
+void SaveSupportedHashAlgorithms(const EVP_MD* md,
+                                 const char* from,
+                                 const char* to,
+                                 void* arg) {
+  if (!from) return;
+  Environment* env = static_cast<Environment*>(arg);
+  env->supported_hash_algorithms.push_back(from);
+}
+#endif  // OPENSSL_VERSION_MAJOR >= 3
+
+const std::vector<std::string>& GetSupportedHashAlgorithms(Environment* env) {
+  if (env->supported_hash_algorithms.empty()) {
+    MarkPopErrorOnReturn mark_pop_error_on_return;
+#if OPENSSL_VERSION_MAJOR >= 3
+    // Since we'll fetch the EVP_MD*, cache them along the way to speed up
+    // later lookups instead of throwing them away immediately.
+    EVP_MD_do_all_sorted(SaveSupportedHashAlgorithmsAndCacheMD, env);
+#else
+    EVP_MD_do_all_sorted(SaveSupportedHashAlgorithms, env);
+#endif
+  }
+  return env->supported_hash_algorithms;
+}
+
 void Hash::GetHashes(const FunctionCallbackInfo<Value>& args) {
-  Environment* env = Environment::GetCurrent(args);
-  MarkPopErrorOnReturn mark_pop_error_on_return;
-  CipherPushContext ctx(env);
-  EVP_MD_do_all_sorted(
+  Local<Context> context = args.GetIsolate()->GetCurrentContext();
+  Environment* env = Environment::GetCurrent(context);
+  const std::vector<std::string>& results = GetSupportedHashAlgorithms(env);
+
+  Local<Value> ret;
+  if (ToV8Value(context, results).ToLocal(&ret)) {
+    args.GetReturnValue().Set(ret);
+  }
+}
+
+void Hash::GetCachedAliases(const FunctionCallbackInfo<Value>& args) {
+  Isolate* isolate = args.GetIsolate();
+  Local<Context> context = args.GetIsolate()->GetCurrentContext();
+  Environment* env = Environment::GetCurrent(context);
+  std::vector<Local<Name>> names;
+  std::vector<Local<Value>> values;
+  size_t size = env->alias_to_md_id_map.size();
 #if OPENSSL_VERSION_MAJOR >= 3
-    array_push_back<EVP_MD,
-                    EVP_MD_fetch,
-                    EVP_MD_free,
-                    EVP_get_digestbyname,
-                    EVP_MD_get0_name>,
+  names.reserve(size);
+  values.reserve(size);
+  for (auto& [alias, id] : env->alias_to_md_id_map) {
+    names.push_back(OneByteString(isolate, alias.c_str(), alias.size()));
+    values.push_back(v8::Uint32::New(isolate, id));
+  }
 #else
-    array_push_back<EVP_MD>,
+  CHECK(env->alias_to_md_id_map.empty());
+#endif
+  Local<Value> prototype = v8::Null(isolate);
+  Local<Object> result =
+      Object::New(isolate, prototype, names.data(), values.data(), size);
+  args.GetReturnValue().Set(result);
+}
+
+const EVP_MD* GetDigestImplementation(Environment* env,
+                                      Local<Value> algorithm,
+                                      Local<Value> cache_id_val,
+                                      Local<Value> algorithm_cache) {
+  CHECK(algorithm->IsString());
+  CHECK(cache_id_val->IsInt32());
+  CHECK(algorithm_cache->IsObject());
+
+#if OPENSSL_VERSION_MAJOR >= 3
+  int32_t cache_id = cache_id_val.As<Int32>()->Value();
+  if (cache_id != -1) {  // Alias already cached, return the cached EVP_MD*.
+    return GetCachedMDByID(env, cache_id);
+  }
+
+  // Only decode the algorithm when we don't have it cached to avoid
+  // unnecessary overhead.
+  Isolate* isolate = env->isolate();
+  Utf8Value utf8(isolate, algorithm);
+
+  auto result = FetchAndMaybeCacheMD(env, *utf8);
+  if (result.cache_id != -1) {
+    // Add the alias to both C++ side and JS side to speedup the lookup
+    // next time.
+    env->alias_to_md_id_map.emplace(*utf8, result.cache_id);
+    if (algorithm_cache.As<Object>()
+            ->Set(isolate->GetCurrentContext(),
+                  algorithm,
+                  v8::Int32::New(isolate, result.cache_id))
+            .IsNothing()) {
+      return nullptr;
+    }
+  }
+
+  return result.explicit_md ? result.explicit_md : result.implicit_md;
+#else
+  Utf8Value utf8(env->isolate(), algorithm);
+  return EVP_get_digestbyname(*utf8);
 #endif
-    &ctx);
-  args.GetReturnValue().Set(ctx.ToJSArray());
 }
 
 void Hash::Initialize(Environment* env, Local<Object> target) {
@@ -65,6 +215,7 @@ void Hash::Initialize(Environment* env, Local<Object> target) {
   SetConstructorFunction(context, target, "Hash", t);
 
   SetMethodNoSideEffect(context, target, "getHashes", GetHashes);
+  SetMethodNoSideEffect(context, target, "getCachedAliases", GetCachedAliases);
 
   HashJob::Initialize(env, target);
 
@@ -77,24 +228,24 @@ void Hash::RegisterExternalReferences(ExternalReferenceRegistry* registry) {
   registry->Register(HashUpdate);
   registry->Register(HashDigest);
   registry->Register(GetHashes);
+  registry->Register(GetCachedAliases);
 
   HashJob::RegisterExternalReferences(registry);
 
   registry->Register(InternalVerifyIntegrity);
 }
 
+// new Hash(algorithm, algorithmId, xofLen, algorithmCache)
 void Hash::New(const FunctionCallbackInfo<Value>& args) {
   Environment* env = Environment::GetCurrent(args);
 
   const Hash* orig = nullptr;
   const EVP_MD* md = nullptr;
-
   if (args[0]->IsObject()) {
     ASSIGN_OR_RETURN_UNWRAP(&orig, args[0].As<Object>());
     md = EVP_MD_CTX_md(orig->mdctx_.get());
   } else {
-    const Utf8Value hash_type(env->isolate(), args[0]);
-    md = EVP_get_digestbyname(*hash_type);
+    md = GetDigestImplementation(env, args[0], args[2], args[3]);
   }
 
   Maybe<unsigned int> xof_md_len = Nothing<unsigned int>();
@@ -284,7 +435,7 @@ bool HashTraits::DeriveBits(
     Environment* env,
     const HashConfig& params,
     ByteSource* out) {
-  EVPMDPointer ctx(EVP_MD_CTX_new());
+  EVPMDCtxPointer ctx(EVP_MD_CTX_new());
 
   if (UNLIKELY(!ctx ||
                EVP_DigestInit_ex(ctx.get(), params.digest, nullptr) <= 0 ||
@@ -357,6 +508,5 @@ void InternalVerifyIntegrity(const v8::FunctionCallbackInfo<v8::Value>& args) {
     args.GetReturnValue().Set(rc.FromMaybe(Local<Value>()));
   }
 }
-
 }  // namespace crypto
 }  // namespace node
diff --git a/src/crypto/crypto_hash.h b/src/crypto/crypto_hash.h
@@ -25,6 +25,7 @@ class Hash final : public BaseObject {
   bool HashUpdate(const char* data, size_t len);
 
   static void GetHashes(const v8::FunctionCallbackInfo<v8::Value>& args);
+  static void GetCachedAliases(const v8::FunctionCallbackInfo<v8::Value>& args);
 
  protected:
   static void New(const v8::FunctionCallbackInfo<v8::Value>& args);
@@ -34,7 +35,7 @@ class Hash final : public BaseObject {
   Hash(Environment* env, v8::Local<v8::Object> wrap);
 
  private:
-  EVPMDPointer mdctx_ {};
+  EVPMDCtxPointer mdctx_{};
   unsigned int md_len_ = 0;
   ByteSource digest_;
 };
diff --git a/src/crypto/crypto_sig.cc b/src/crypto/crypto_sig.cc
@@ -73,7 +73,7 @@ bool ApplyRSAOptions(const ManagedEVPPKey& pkey,
 }
 
 std::unique_ptr<BackingStore> Node_SignFinal(Environment* env,
-                                             EVPMDPointer&& mdctx,
+                                             EVPMDCtxPointer&& mdctx,
                                              const ManagedEVPPKey& pkey,
                                              int padding,
                                              Maybe<int> pss_salt_len) {
@@ -391,7 +391,7 @@ Sign::SignResult Sign::SignFinal(
   if (!mdctx_)
     return SignResult(kSignNotInitialised);
 
-  EVPMDPointer mdctx = std::move(mdctx_);
+  EVPMDCtxPointer mdctx = std::move(mdctx_);
 
   if (!ValidateDSAParameters(pkey.get()))
     return SignResult(kSignPrivateKey);
@@ -511,7 +511,7 @@ SignBase::Error Verify::VerifyFinal(const ManagedEVPPKey& pkey,
   unsigned char m[EVP_MAX_MD_SIZE];
   unsigned int m_len;
   *verify_result = false;
-  EVPMDPointer mdctx = std::move(mdctx_);
+  EVPMDCtxPointer mdctx = std::move(mdctx_);
 
   if (!EVP_DigestFinal_ex(mdctx.get(), m, &m_len))
     return kSignPublicKey;
@@ -696,7 +696,7 @@ bool SignTraits::DeriveBits(
     const SignConfiguration& params,
     ByteSource* out) {
   ClearErrorOnReturn clear_error_on_return;
-  EVPMDPointer context(EVP_MD_CTX_new());
+  EVPMDCtxPointer context(EVP_MD_CTX_new());
   EVP_PKEY_CTX* ctx = nullptr;
 
   switch (params.mode) {
diff --git a/src/crypto/crypto_sig.h b/src/crypto/crypto_sig.h
diff --git a/src/crypto/crypto_util.h b/src/crypto/crypto_util.h
diff --git a/src/env.h b/src/env.h
