google-protobuf 3.25.6 → 4.30.1

data/ext/google/protobuf_c/message.c

@@ -44,9 +44,11 @@ static void Message_mark(void* _self) {
   rb_gc_mark(self->arena);
 }
 
+static size_t Message_memsize(const void* _self) { return sizeof(Message); }
+
 static rb_data_type_t Message_type = {
     "Google::Protobuf::Message",
-    {Message_mark, RUBY_DEFAULT_FREE, NULL},
+    {Message_mark, RUBY_DEFAULT_FREE, Message_memsize},
     .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
 };
 
@@ -78,11 +80,13 @@ const upb_Message* Message_Get(VALUE msg_rb, const upb_MessageDef** m) {
 }
 
 upb_Message* Message_GetMutable(VALUE msg_rb, const upb_MessageDef** m) {
-  rb_check_frozen(msg_rb);
-  return (upb_Message*)Message_Get(msg_rb, m);
+  const upb_Message* upb_msg = Message_Get(msg_rb, m);
+  Protobuf_CheckNotFrozen(msg_rb, upb_Message_IsFrozen(upb_msg));
+  return (upb_Message*)upb_msg;
 }
 
-void Message_InitPtr(VALUE self_, upb_Message* msg, VALUE arena) {
+void Message_InitPtr(VALUE self_, const upb_Message* msg, VALUE arena) {
+  PBRUBY_ASSERT(arena != Qnil);
   Message* self = ruby_to_Message(self_);
   self->msg = msg;
   RB_OBJ_WRITE(self_, &self->arena, arena);
@@ -103,7 +107,7 @@ void Message_CheckClass(VALUE klass) {
   }
 }
 
-VALUE Message_GetRubyWrapper(upb_Message* msg, const upb_MessageDef* m,
+VALUE Message_GetRubyWrapper(const upb_Message* msg, const upb_MessageDef* m,
                              VALUE arena) {
   if (msg == NULL) return Qnil;
 
@@ -114,7 +118,6 @@ VALUE Message_GetRubyWrapper(upb_Message* msg, const upb_MessageDef* m,
     val = Message_alloc(klass);
     Message_InitPtr(val, msg, arena);
   }
-
   return val;
 }
 
@@ -246,7 +249,7 @@ static int extract_method_call(VALUE method_name, Message* self,
 static VALUE Message_oneof_accessor(VALUE _self, const upb_OneofDef* o,
                                     int accessor_type) {
   Message* self = ruby_to_Message(_self);
-  const upb_FieldDef* oneof_field = upb_Message_WhichOneof(self->msg, o);
+  const upb_FieldDef* oneof_field = upb_Message_WhichOneofByDef(self->msg, o);
 
   switch (accessor_type) {
     case METHOD_PRESENCE:
@@ -286,13 +289,42 @@ static void Message_setfield(upb_Message* msg, const upb_FieldDef* f, VALUE val,
   upb_Message_SetFieldByDef(msg, f, msgval, arena);
 }
 
+VALUE Message_getfield_frozen(const upb_Message* msg, const upb_FieldDef* f,
+                              VALUE arena) {
+  upb_MessageValue msgval = upb_Message_GetFieldByDef(msg, f);
+  if (upb_FieldDef_IsMap(f)) {
+    if (msgval.map_val == NULL) {
+      return Map_EmptyFrozen(f);
+    }
+    const upb_FieldDef* key_f = map_field_key(f);
+    const upb_FieldDef* val_f = map_field_value(f);
+    upb_CType key_type = upb_FieldDef_CType(key_f);
+    TypeInfo value_type_info = TypeInfo_get(val_f);
+    return Map_GetRubyWrapper(msgval.map_val, key_type, value_type_info, arena);
+  }
+  if (upb_FieldDef_IsRepeated(f)) {
+    if (msgval.array_val == NULL) {
+      return RepeatedField_EmptyFrozen(f);
+    }
+    return RepeatedField_GetRubyWrapper(msgval.array_val, TypeInfo_get(f),
+                                        arena);
+  }
+  VALUE ret;
+  if (upb_FieldDef_IsSubMessage(f)) {
+    const upb_MessageDef* m = upb_FieldDef_MessageSubDef(f);
+    ret = Message_GetRubyWrapper(msgval.msg_val, m, arena);
+  } else {
+    ret = Convert_UpbToRuby(msgval, TypeInfo_get(f), Qnil);
+  }
+  return ret;
+}
+
 VALUE Message_getfield(VALUE _self, const upb_FieldDef* f) {
   Message* self = ruby_to_Message(_self);
-  // This is a special-case: upb_Message_Mutable() for map & array are logically
-  // const (they will not change what is serialized) but physically
-  // non-const, as they do allocate a repeated field or map. The logical
-  // constness means it's ok to do even if the message is frozen.
-  upb_Message* msg = (upb_Message*)self->msg;
+  if (upb_Message_IsFrozen(self->msg)) {
+    return Message_getfield_frozen(self->msg, f, self->arena);
+  }
+  upb_Message* msg = Message_GetMutable(_self, NULL);
   upb_Arena* arena = Arena_get(self->arena);
   if (upb_FieldDef_IsMap(f)) {
     upb_Map* map = upb_Message_Mutable(msg, f, arena).map;
@@ -305,12 +337,12 @@ VALUE Message_getfield(VALUE _self, const upb_FieldDef* f) {
     upb_Array* arr = upb_Message_Mutable(msg, f, arena).array;
     return RepeatedField_GetRubyWrapper(arr, TypeInfo_get(f), self->arena);
   } else if (upb_FieldDef_IsSubMessage(f)) {
-    if (!upb_Message_HasFieldByDef(self->msg, f)) return Qnil;
+    if (!upb_Message_HasFieldByDef(msg, f)) return Qnil;
     upb_Message* submsg = upb_Message_Mutable(msg, f, arena).msg;
     const upb_MessageDef* m = upb_FieldDef_MessageSubDef(f);
     return Message_GetRubyWrapper(submsg, m, self->arena);
   } else {
-    upb_MessageValue msgval = upb_Message_GetFieldByDef(self->msg, f);
+    upb_MessageValue msgval = upb_Message_GetFieldByDef(msg, f);
     return Convert_UpbToRuby(msgval, TypeInfo_get(f), self->arena);
   }
 }
@@ -330,7 +362,8 @@ static VALUE Message_field_accessor(VALUE _self, const upb_FieldDef* f,
       if (!upb_FieldDef_HasPresence(f)) {
         rb_raise(rb_eRuntimeError, "Field does not have presence.");
       }
-      return upb_Message_HasFieldByDef(Message_Get(_self, NULL), f);
+      return upb_Message_HasFieldByDef(Message_Get(_self, NULL), f) ? Qtrue
+                                                                    : Qfalse;
     case METHOD_WRAPPER_GETTER: {
       Message* self = ruby_to_Message(_self);
       if (upb_Message_HasFieldByDef(self->msg, f)) {
@@ -434,7 +467,6 @@ static VALUE Message_method_missing(int argc, VALUE* argv, VALUE _self) {
       if (argc != 2) {
         rb_raise(rb_eArgError, "Expected 2 arguments, received %d", argc);
       }
-      rb_check_frozen(_self);
       break;
     default:
       if (argc != 1) {
@@ -661,29 +693,12 @@ static VALUE Message_dup(VALUE _self) {
   Message* self = ruby_to_Message(_self);
   VALUE new_msg = rb_class_new_instance(0, NULL, CLASS_OF(_self));
   Message* new_msg_self = ruby_to_Message(new_msg);
-  size_t size = upb_MessageDef_MiniTable(self->msgdef)->size;
-
-  // TODO
-  // TODO
-  memcpy((upb_Message*)new_msg_self->msg, self->msg, size);
+  const upb_MiniTable* m = upb_MessageDef_MiniTable(self->msgdef);
+  upb_Message_ShallowCopy((upb_Message*)new_msg_self->msg, self->msg, m);
   Arena_fuse(self->arena, Arena_get(new_msg_self->arena));
   return new_msg;
 }
 
-// Support function for Message_eq, and also used by other #eq functions.
-bool Message_Equal(const upb_Message* m1, const upb_Message* m2,
-                   const upb_MessageDef* m) {
-  upb_Status status;
-  upb_Status_Clear(&status);
-  bool return_value = shared_Message_Equal(m1, m2, m, &status);
-  if (upb_Status_IsOk(&status)) {
-    return return_value;
-  } else {
-    rb_raise(cParseError, "Message_Equal(): %s",
-             upb_Status_ErrorMessage(&status));
-  }
-}
-
 /*
  * call-seq:
  *     Message.==(other) => boolean
@@ -700,7 +715,10 @@ static VALUE Message_eq(VALUE _self, VALUE _other) {
   Message* other = ruby_to_Message(_other);
   assert(self->msgdef == other->msgdef);
 
-  return Message_Equal(self->msg, other->msg, self->msgdef) ? Qtrue : Qfalse;
+  const upb_MiniTable* m = upb_MessageDef_MiniTable(self->msgdef);
+  const int options = 0;
+  return upb_Message_IsEqual(self->msg, other->msg, m, options) ? Qtrue
+                                                                : Qfalse;
 }
 
 uint64_t Message_Hash(const upb_Message* msg, const upb_MessageDef* m,
@@ -769,58 +787,34 @@ static VALUE Message_CreateHash(const upb_Message* msg,
   if (!msg) return Qnil;
 
   VALUE hash = rb_hash_new();
-  int n = upb_MessageDef_FieldCount(m);
-  bool is_proto2;
-
-  // We currently have a few behaviors that are specific to proto2.
-  // This is unfortunate, we should key behaviors off field attributes (like
-  // whether a field has presence), not proto2 vs. proto3. We should see if we
-  // can change this without breaking users.
-  is_proto2 = upb_MessageDef_Syntax(m) == kUpb_Syntax_Proto2;
-
-  for (int i = 0; i < n; i++) {
-    const upb_FieldDef* field = upb_MessageDef_Field(m, i);
-    TypeInfo type_info = TypeInfo_get(field);
-    upb_MessageValue msgval;
-    VALUE msg_value;
-    VALUE msg_key;
-
-    if (!is_proto2 && upb_FieldDef_IsSubMessage(field) &&
-        !upb_FieldDef_IsRepeated(field) &&
-        !upb_Message_HasFieldByDef(msg, field)) {
-      // TODO: Legacy behavior, remove when we fix the is_proto2 differences.
-      msg_key = ID2SYM(rb_intern(upb_FieldDef_Name(field)));
-      rb_hash_aset(hash, msg_key, Qnil);
-      continue;
-    }
+  size_t iter = kUpb_Message_Begin;
+  const upb_DefPool* pool = upb_FileDef_Pool(upb_MessageDef_File(m));
+  const upb_FieldDef* field;
+  upb_MessageValue val;
 
-    // Do not include fields that are not present (oneof or optional fields).
-    if (is_proto2 && upb_FieldDef_HasPresence(field) &&
-        !upb_Message_HasFieldByDef(msg, field)) {
+  while (upb_Message_Next(msg, m, pool, &field, &val, &iter)) {
+    if (upb_FieldDef_IsExtension(field)) {
+      // TODO: allow extensions once we have decided what naming scheme the
+      // symbol should use.  eg. :"[pkg.ext]"
       continue;
     }
 
-    msg_key = ID2SYM(rb_intern(upb_FieldDef_Name(field)));
-    msgval = upb_Message_GetFieldByDef(msg, field);
-
-    // Proto2 omits empty map/repeated filds also.
+    TypeInfo type_info = TypeInfo_get(field);
+    VALUE msg_value;
 
     if (upb_FieldDef_IsMap(field)) {
       const upb_MessageDef* entry_m = upb_FieldDef_MessageSubDef(field);
       const upb_FieldDef* key_f = upb_MessageDef_FindFieldByNumber(entry_m, 1);
       const upb_FieldDef* val_f = upb_MessageDef_FindFieldByNumber(entry_m, 2);
       upb_CType key_type = upb_FieldDef_CType(key_f);
-      msg_value = Map_CreateHash(msgval.map_val, key_type, TypeInfo_get(val_f));
+      msg_value = Map_CreateHash(val.map_val, key_type, TypeInfo_get(val_f));
     } else if (upb_FieldDef_IsRepeated(field)) {
-      if (is_proto2 &&
-          (!msgval.array_val || upb_Array_Size(msgval.array_val) == 0)) {
-        continue;
-      }
-      msg_value = RepeatedField_CreateArray(msgval.array_val, type_info);
+      msg_value = RepeatedField_CreateArray(val.array_val, type_info);
     } else {
-      msg_value = Scalar_CreateHash(msgval, type_info);
+      msg_value = Scalar_CreateHash(val, type_info);
     }
 
+    VALUE msg_key = ID2SYM(rb_intern(upb_FieldDef_Name(field)));
     rb_hash_aset(hash, msg_key, msg_value);
   }
 
@@ -848,43 +842,42 @@ static VALUE Message_to_h(VALUE _self) {
 
 /*
  * call-seq:
- *     Message.freeze => self
+ *     Message.frozen? => bool
  *
- * Freezes the message object. We have to intercept this so we can pin the
- * Ruby object into memory so we don't forget it's frozen.
+ * Returns true if the message is frozen in either Ruby or the underlying
+ * representation. Freezes the Ruby message object if it is not already frozen
+ * in Ruby but it is frozen in the underlying representation.
  */
-static VALUE Message_freeze(VALUE _self) {
+VALUE Message_frozen(VALUE _self) {
   Message* self = ruby_to_Message(_self);
-  if (!RB_OBJ_FROZEN(_self)) {
-    Arena_Pin(self->arena, _self);
-    RB_OBJ_FREEZE(_self);
+  if (!upb_Message_IsFrozen(self->msg)) {
+    PBRUBY_ASSERT(!RB_OBJ_FROZEN(_self));
+    return Qfalse;
   }
-  return _self;
+
+  // Lazily freeze the Ruby wrapper.
+  if (!RB_OBJ_FROZEN(_self)) RB_OBJ_FREEZE(_self);
+  return Qtrue;
 }
 
 /*
- * Deep freezes the message object recursively.
- * Internal use only.
+ * call-seq:
+ *     Message.freeze => self
+ *
+ * Freezes the message object. We have to intercept this so we can freeze the
+ * underlying representation, not just the Ruby wrapper.
  */
-VALUE Message_internal_deep_freeze(VALUE _self) {
+VALUE Message_freeze(VALUE _self) {
   Message* self = ruby_to_Message(_self);
-  Message_freeze(_self);
-
-  int n = upb_MessageDef_FieldCount(self->msgdef);
-  for (int i = 0; i < n; i++) {
-    const upb_FieldDef* f = upb_MessageDef_Field(self->msgdef, i);
-    VALUE field = Message_getfield(_self, f);
-
-    if (field != Qnil) {
-      if (upb_FieldDef_IsMap(f)) {
-        Map_internal_deep_freeze(field);
-      } else if (upb_FieldDef_IsRepeated(f)) {
-        RepeatedField_internal_deep_freeze(field);
-      } else if (upb_FieldDef_IsSubMessage(f)) {
-        Message_internal_deep_freeze(field);
-      }
-    }
+  if (RB_OBJ_FROZEN(_self)) {
+    PBRUBY_ASSERT(upb_Message_IsFrozen(self->msg));
+    return _self;
   }
+  if (!upb_Message_IsFrozen(self->msg)) {
+    upb_Message_Freeze(Message_GetMutable(_self, NULL),
+                       upb_MessageDef_MiniTable(self->msgdef));
+  }
+  RB_OBJ_FREEZE(_self);
   return _self;
 }
 
@@ -990,7 +983,7 @@ VALUE Message_decode_bytes(int size, const char* bytes, int options,
     rb_raise(cParseError, "Error occurred during parsing");
   }
   if (freeze) {
-    Message_internal_deep_freeze(msg_rb);
+    Message_freeze(msg_rb);
   }
   return msg_rb;
 }
@@ -1012,9 +1005,6 @@ static VALUE Message_decode_json(int argc, VALUE* argv, VALUE klass) {
   int options = 0;
   upb_Status status;
 
-  // TODO: use this message's pool instead.
-  const upb_DefPool* symtab = DescriptorPool_GetSymtab(generated_pool);
-
   if (argc < 1 || argc > 2) {
     rb_raise(rb_eArgError, "Expected 1 or 2 arguments.");
   }
@@ -1048,11 +1038,19 @@ static VALUE Message_decode_json(int argc, VALUE* argv, VALUE klass) {
   }
 
   upb_Status_Clear(&status);
-  if (!upb_JsonDecode(RSTRING_PTR(data), RSTRING_LEN(data),
-                      (upb_Message*)msg->msg, msg->msgdef, symtab, options,
-                      Arena_get(msg->arena), &status)) {
-    rb_raise(cParseError, "Error occurred during parsing: %s",
-             upb_Status_ErrorMessage(&status));
+  const upb_DefPool* pool = upb_FileDef_Pool(upb_MessageDef_File(msg->msgdef));
+
+  int result = upb_JsonDecodeDetectingNonconformance(
+      RSTRING_PTR(data), RSTRING_LEN(data), (upb_Message*)msg->msg,
+      msg->msgdef, pool, options, Arena_get(msg->arena), &status);
+
+  switch (result) {
+    case kUpb_JsonDecodeResult_Ok:
+      break;
+    case kUpb_JsonDecodeResult_Error:
+      rb_raise(cParseError, "Error occurred during parsing: %s",
+               upb_Status_ErrorMessage(&status));
+      break;
   }
 
   return msg_rb;
@@ -1128,9 +1126,6 @@ static VALUE Message_encode_json(int argc, VALUE* argv, VALUE klass) {
   size_t size;
   upb_Status status;
 
-  // TODO: use this message's pool instead.
-  const upb_DefPool* symtab = DescriptorPool_GetSymtab(generated_pool);
-
   if (argc < 1 || argc > 2) {
     rb_raise(rb_eArgError, "Expected 1 or 2 arguments.");
   }
@@ -1165,8 +1160,9 @@ static VALUE Message_encode_json(int argc, VALUE* argv, VALUE klass) {
   }
 
   upb_Status_Clear(&status);
-  size = upb_JsonEncode(msg->msg, msg->msgdef, symtab, options, buf,
-                        sizeof(buf), &status);
+  const upb_DefPool* pool = upb_FileDef_Pool(upb_MessageDef_File(msg->msgdef));
+  size = upb_JsonEncode(msg->msg, msg->msgdef, pool, options, buf, sizeof(buf),
+                        &status);
 
   if (!upb_Status_IsOk(&status)) {
     rb_raise(cParseError, "Error occurred during encoding: %s",
@@ -1176,7 +1172,7 @@ static VALUE Message_encode_json(int argc, VALUE* argv, VALUE klass) {
   VALUE ret;
   if (size >= sizeof(buf)) {
     char* buf2 = malloc(size + 1);
-    upb_JsonEncode(msg->msg, msg->msgdef, symtab, options, buf2, size + 1,
+    upb_JsonEncode(msg->msg, msg->msgdef, pool, options, buf2, size + 1,
                    &status);
     ret = rb_str_new(buf2, size);
     free(buf2);
@@ -1402,6 +1398,7 @@ static void Message_define_class(VALUE klass) {
   rb_define_method(klass, "==", Message_eq, 1);
   rb_define_method(klass, "eql?", Message_eq, 1);
   rb_define_method(klass, "freeze", Message_freeze, 0);
+  rb_define_method(klass, "frozen?", Message_frozen, 0);
   rb_define_method(klass, "hash", Message_hash, 0);
   rb_define_method(klass, "to_h", Message_to_h, 0);
   rb_define_method(klass, "inspect", Message_inspect, 0);

data/ext/google/protobuf_c/message.h

@@ -36,7 +36,7 @@ const upb_Message* Message_GetUpbMessage(VALUE value, const upb_MessageDef* m,
 
 // Gets or constructs a Ruby wrapper object for the given message. The wrapper
 // object will reference |arena| and ensure that it outlives this object.
-VALUE Message_GetRubyWrapper(upb_Message* msg, const upb_MessageDef* m,
+VALUE Message_GetRubyWrapper(const upb_Message* msg, const upb_MessageDef* m,
                              VALUE arena);
 
 // Gets the given field from this message.
@@ -54,10 +54,6 @@ uint64_t Message_Hash(const upb_Message* msg, const upb_MessageDef* m,
 upb_Message* Message_deep_copy(const upb_Message* msg, const upb_MessageDef* m,
                                upb_Arena* arena);
 
-// Returns true if these two messages are equal.
-bool Message_Equal(const upb_Message* m1, const upb_Message* m2,
-                   const upb_MessageDef* m);
-
 // Checks that this Ruby object is a message, and raises an exception if not.
 void Message_CheckClass(VALUE klass);
 
@@ -78,7 +74,7 @@ VALUE Message_decode_bytes(int size, const char* bytes, int options,
                            VALUE klass, bool freeze);
 
 // Recursively freeze message
-VALUE Message_internal_deep_freeze(VALUE _self);
+VALUE Message_freeze(VALUE _self);
 
 // Call at startup to register all types in this module.
 void Message_register(VALUE protobuf);

data/ext/google/protobuf_c/protobuf.c

@@ -7,8 +7,6 @@
 
 #include "protobuf.h"
 
-#include <ruby/version.h>
-
 #include "defs.h"
 #include "map.h"
 #include "message.h"
@@ -164,11 +162,23 @@ static void Arena_free(void *data) {
   xfree(arena);
 }
 
+static size_t Arena_memsize(const void *data) {
+  const Arena *arena = data;
+  size_t fused_count;
+  size_t memsize = upb_Arena_SpaceAllocated(arena->arena, &fused_count);
+  if (fused_count > 1) {
+    // If other arena were fused we attribute an equal
+    // share of memory usage to each one.
+    memsize /= fused_count;
+  }
+  return memsize + sizeof(Arena);
+}
+
 static VALUE cArena;
 
 const rb_data_type_t Arena_type = {
     "Google::Protobuf::Internal::Arena",
-    {Arena_mark, Arena_free, NULL},
+    {Arena_mark, Arena_free, Arena_memsize},
     .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
 };
 
@@ -209,15 +219,6 @@ void Arena_fuse(VALUE _arena, upb_Arena *other) {
 
 VALUE Arena_new() { return Arena_alloc(cArena); }
 
-void Arena_Pin(VALUE _arena, VALUE obj) {
-  Arena *arena;
-  TypedData_Get_Struct(_arena, Arena, &Arena_type, arena);
-  if (arena->pinned_objs == Qnil) {
-    RB_OBJ_WRITE(_arena, &arena->pinned_objs, rb_ary_new());
-  }
-  rb_ary_push(arena->pinned_objs, obj);
-}
-
 void Arena_register(VALUE module) {
   VALUE internal = rb_define_module_under(module, "Internal");
   VALUE klass = rb_define_class_under(internal, "Arena", rb_cObject);
@@ -241,16 +242,17 @@ static void ObjectCache_Init(VALUE protobuf) {
   item_try_add = rb_intern("try_add");
 
   rb_gc_register_address(&weak_obj_cache);
+  VALUE internal = rb_const_get(protobuf, rb_intern("Internal"));
 #if SIZEOF_LONG >= SIZEOF_VALUE
-  VALUE cache_class = rb_const_get(protobuf, rb_intern("ObjectCache"));
+  VALUE cache_class = rb_const_get(internal, rb_intern("ObjectCache"));
 #else
-  VALUE cache_class = rb_const_get(protobuf, rb_intern("LegacyObjectCache"));
+  VALUE cache_class = rb_const_get(internal, rb_intern("LegacyObjectCache"));
 #endif
 
   weak_obj_cache = rb_class_new_instance(0, NULL, cache_class);
-  rb_const_set(protobuf, rb_intern("OBJECT_CACHE"), weak_obj_cache);
-  rb_const_set(protobuf, rb_intern("SIZEOF_LONG"), INT2NUM(SIZEOF_LONG));
-  rb_const_set(protobuf, rb_intern("SIZEOF_VALUE"), INT2NUM(SIZEOF_VALUE));
+  rb_const_set(internal, rb_intern("OBJECT_CACHE"), weak_obj_cache);
+  rb_const_set(internal, rb_intern("SIZEOF_LONG"), INT2NUM(SIZEOF_LONG));
+  rb_const_set(internal, rb_intern("SIZEOF_VALUE"), INT2NUM(SIZEOF_VALUE));
 }
 
 static VALUE ObjectCache_GetKey(const void *key) {
@@ -284,7 +286,8 @@ VALUE ObjectCache_Get(const void *key) {
 static VALUE Google_Protobuf_discard_unknown(VALUE self, VALUE msg_rb) {
   const upb_MessageDef *m;
   upb_Message *msg = Message_GetMutable(msg_rb, &m);
-  if (!upb_Message_DiscardUnknown(msg, m, 128)) {
+  const upb_DefPool* ext_pool = upb_FileDef_Pool(upb_MessageDef_File(m));
+  if (!upb_Message_DiscardUnknown(msg, m, ext_pool, 128)) {
     rb_raise(rb_eRuntimeError, "Messages nested too deeply.");
   }
 
@@ -341,3 +344,14 @@ __attribute__((visibility("default"))) void Init_protobuf_c() {
   rb_define_singleton_method(protobuf, "deep_copy", Google_Protobuf_deep_copy,
                              1);
 }
+
+// -----------------------------------------------------------------------------
+// Utilities
+// -----------------------------------------------------------------------------
+
+// Raises a Ruby error if val is frozen in Ruby or UPB.
+void Protobuf_CheckNotFrozen(VALUE val, bool upb_frozen) {
+  if (RB_UNLIKELY(rb_obj_frozen_p(val)||upb_frozen)) {
+    rb_error_frozen_object(val);
+  }
+}

data/ext/google/protobuf_c/protobuf.h

@@ -50,13 +50,6 @@ upb_Arena* Arena_get(VALUE arena);
 // possible.
 void Arena_fuse(VALUE arena, upb_Arena* other);
 
-// Pins this Ruby object to the lifetime of this arena, so that as long as the
-// arena is alive this object will not be collected.
-//
-// We use this to guarantee that the "frozen" bit on the object will be
-// remembered, even if the user drops their reference to this precise object.
-void Arena_Pin(VALUE arena, VALUE obj);
-
 // -----------------------------------------------------------------------------
 // ObjectCache
 // -----------------------------------------------------------------------------
@@ -105,6 +98,9 @@ extern VALUE cTypeError;
   rb_bug("Assertion failed at %s:%d, expr: %s", __FILE__, __LINE__, #expr)
 #endif
 
+// Raises a Ruby error if val is frozen in Ruby or upb_frozen is true.
+void Protobuf_CheckNotFrozen(VALUE val, bool upb_frozen);
+
 #define PBRUBY_MAX(x, y) (((x) > (y)) ? (x) : (y))
 
 #define UPB_UNUSED(var) (void)var

data/ext/google/protobuf_c/repeated_field.c

@@ -44,8 +44,9 @@ static RepeatedField* ruby_to_RepeatedField(VALUE _self) {
 }
 
 static upb_Array* RepeatedField_GetMutable(VALUE _self) {
-  rb_check_frozen(_self);
-  return (upb_Array*)ruby_to_RepeatedField(_self)->array;
+  const upb_Array* array = ruby_to_RepeatedField(_self)->array;
+  Protobuf_CheckNotFrozen(_self, upb_Array_IsFrozen(array));
+  return (upb_Array*)array;
 }
 
 VALUE RepeatedField_alloc(VALUE klass) {
@@ -56,9 +57,32 @@ VALUE RepeatedField_alloc(VALUE klass) {
   return TypedData_Wrap_Struct(klass, &RepeatedField_type, self);
 }
 
-VALUE RepeatedField_GetRubyWrapper(upb_Array* array, TypeInfo type_info,
+VALUE RepeatedField_EmptyFrozen(const upb_FieldDef* f) {
+  PBRUBY_ASSERT(upb_FieldDef_IsRepeated(f));
+  VALUE val = ObjectCache_Get(f);
+
+  if (val == Qnil) {
+    val = RepeatedField_alloc(cRepeatedField);
+    RepeatedField* self;
+    TypedData_Get_Struct(val, RepeatedField, &RepeatedField_type, self);
+    self->arena = Arena_new();
+    TypeInfo type_info = TypeInfo_get(f);
+    self->array = upb_Array_New(Arena_get(self->arena), type_info.type);
+    self->type_info = type_info;
+    if (self->type_info.type == kUpb_CType_Message) {
+      self->type_class = Descriptor_DefToClass(type_info.def.msgdef);
+    }
+    val = ObjectCache_TryAdd(f, RepeatedField_freeze(val));
+  }
+  PBRUBY_ASSERT(RB_OBJ_FROZEN(val));
+  PBRUBY_ASSERT(upb_Array_IsFrozen(ruby_to_RepeatedField(val)->array));
+  return val;
+}
+
+VALUE RepeatedField_GetRubyWrapper(const upb_Array* array, TypeInfo type_info,
                                    VALUE arena) {
   PBRUBY_ASSERT(array);
+  PBRUBY_ASSERT(arena != Qnil);
   VALUE val = ObjectCache_Get(array);
 
   if (val == Qnil) {
@@ -78,7 +102,6 @@ VALUE RepeatedField_GetRubyWrapper(upb_Array* array, TypeInfo type_info,
   PBRUBY_ASSERT(ruby_to_RepeatedField(val)->type_info.def.msgdef ==
                 type_info.def.msgdef);
   PBRUBY_ASSERT(ruby_to_RepeatedField(val)->array == array);
-
   return val;
 }
 
@@ -473,36 +496,47 @@ VALUE RepeatedField_eq(VALUE _self, VALUE _other) {
 
 /*
  * call-seq:
- *     RepeatedField.freeze => self
+ *     RepeatedField.frozen? => bool
  *
- * Freezes the repeated field. We have to intercept this so we can pin the Ruby
- * object into memory so we don't forget it's frozen.
+ * Returns true if the repeated field is frozen in either Ruby or the underlying
+ * representation. Freezes the Ruby repeated field object if it is not already
+ * frozen in Ruby but it is frozen in the underlying representation.
  */
-static VALUE RepeatedField_freeze(VALUE _self) {
+VALUE RepeatedField_frozen(VALUE _self) {
   RepeatedField* self = ruby_to_RepeatedField(_self);
-  if (!RB_OBJ_FROZEN(_self)) {
-    Arena_Pin(self->arena, _self);
-    RB_OBJ_FREEZE(_self);
+  if (!upb_Array_IsFrozen(self->array)) {
+    PBRUBY_ASSERT(!RB_OBJ_FROZEN(_self));
+    return Qfalse;
   }
-  return _self;
+
+  // Lazily freeze the Ruby wrapper.
+  if (!RB_OBJ_FROZEN(_self)) RB_OBJ_FREEZE(_self);
+  return Qtrue;
 }
 
 /*
- * Deep freezes the repeated field and values recursively.
- * Internal use only.
+ * call-seq:
+ *     RepeatedField.freeze => self
+ *
+ * Freezes the repeated field object. We have to intercept this so we can freeze
+ * the underlying representation, not just the Ruby wrapper.
  */
-VALUE RepeatedField_internal_deep_freeze(VALUE _self) {
+VALUE RepeatedField_freeze(VALUE _self) {
   RepeatedField* self = ruby_to_RepeatedField(_self);
-  RepeatedField_freeze(_self);
-  if (self->type_info.type == kUpb_CType_Message) {
-    int size = upb_Array_Size(self->array);
-    int i;
-    for (i = 0; i < size; i++) {
-      upb_MessageValue msgval = upb_Array_Get(self->array, i);
-      VALUE val = Convert_UpbToRuby(msgval, self->type_info, self->arena);
-      Message_internal_deep_freeze(val);
+  if (RB_OBJ_FROZEN(_self)) {
+    PBRUBY_ASSERT(upb_Array_IsFrozen(self->array));
+    return _self;
+  }
+
+  if (!upb_Array_IsFrozen(self->array)) {
+    if (self->type_info.type == kUpb_CType_Message) {
+      upb_Array_Freeze(RepeatedField_GetMutable(_self),
+                       upb_MessageDef_MiniTable(self->type_info.def.msgdef));
+    } else {
+      upb_Array_Freeze(RepeatedField_GetMutable(_self), NULL);
     }
   }
+  RB_OBJ_FREEZE(_self);
   return _self;
 }
 
@@ -649,6 +683,7 @@ void RepeatedField_register(VALUE module) {
   rb_define_method(klass, "==", RepeatedField_eq, 1);
   rb_define_method(klass, "to_ary", RepeatedField_to_ary, 0);
   rb_define_method(klass, "freeze", RepeatedField_freeze, 0);
+  rb_define_method(klass, "frozen?", RepeatedField_frozen, 0);
   rb_define_method(klass, "hash", RepeatedField_hash, 0);
   rb_define_method(klass, "+", RepeatedField_plus, 1);
   rb_define_method(klass, "concat", RepeatedField_concat, 1);