Segmentation fault during Serialization: caused by message as a field in C++17

[danilyef]

This post is related to this one: https://stackoverflow.com/questions/79576605/segmentation-fault-during-serialization-by-using-protobuf-in-c17

But I have narrowed it down to the specific problem and decided to make separated post.

My goal is to serialize information by using protobuf in C++.

proto file from ./proto folder:

syntax="proto3";

package PhoneBookSerialize;

message Date {
    int32 year = 1;
    int32 month = 2;
    int32 day = 3;
}

message Contact {
    string name = 1;
    Date birthday = 2;
    repeated string phone_number = 3;
}

message ContactList {
    repeated Contact contact = 1;
}

main.cpp in ./src folder

#include "contact.pb.h"
#include <sstream>
#include <iostream>

using namespace std;

int main() {

  // Example 2: Serialize single Date
  PhoneBookSerialize::Date date;
  date.set_year(1990);
  date.set_month(6);
  date.set_day(15);
  
  // Serialize Date
  ostringstream date_output(std::ios::binary);
  date.SerializeToOstream(&date_output);
  std::string serialized_date = date_output.str();
  std::cout << "Serialized date size: " << serialized_date.size() << std::endl;

  // Deserialize Date
  PhoneBookSerialize::Date parsed_date;
  istringstream date_input(serialized_date, std::ios::binary);
  parsed_date.ParseFromIstream(&date_input);
  std::cout << "Deserialized date: " 
            << parsed_date.year() << "-"
            << parsed_date.month() << "-"
            << parsed_date.day() << std::endl;

  std::cout << "================================================" << std::endl;
  // Example 3: Serialize single Contact
  PhoneBookSerialize::Contact contact;
  contact.set_name("John Doe");

  contact.mutable_birthday()->set_year(1990);
  contact.mutable_birthday()->set_month(6);
  contact.mutable_birthday()->set_day(15);
  contact.add_phone_number("+1234567890");
  contact.add_phone_number("+1234567891");

  // Serialize Contact
  ostringstream contact_output(std::ios::binary);
  if (!contact.SerializeToOstream(&contact_output)) {
    std::cerr << "Failed to serialize contact!" << std::endl;
  } else {
    std::cout << "Successfully serialized contact (size: " << contact_output.str().size() << " bytes)" << std::endl;
  }

  std::string serialized_contact = contact_output.str();
  std::cout << "Serialized contact size: " << serialized_contact.size() << std::endl;

  // Deserialize Contact
  PhoneBookSerialize::Contact parsed_contact;
  istringstream contact_input(serialized_contact, std::ios::binary);
  if (!parsed_contact.ParseFromIstream(&contact_input)) {
    std::cerr << "Failed to parse contact!" << std::endl;
  }
  std::cout << "Deserialized contact:" << std::endl
            << "Name: " << parsed_contact.name() << std::endl
            << "Birthday: " 
            << parsed_contact.birthday().year() << "-"
            << parsed_contact.birthday().month() << "-"
            << parsed_contact.birthday().day() << std::endl
            << "Phone number: " << parsed_contact.phone_number(0) << std::endl
            << "Phone number: " << parsed_contact.phone_number(1) << std::endl;


  return 0;
}

CMakeLists.txt file in . folder:

cmake_minimum_required(VERSION 3.20)
project(PhoneBookProtobuf LANGUAGES CXX)

set(CMAKE_CXX_STANDARD 17)
set(CMAKE_CXX_STANDARD_REQUIRED ON)

# Add debug flags
set(CMAKE_BUILD_TYPE Debug)
set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} -g -fcolor-diagnostics -fansi-escape-codes -fsanitize=address -fsanitize=undefined -DDEBUG -fno-omit-frame-pointer -fno-optimize-sibling-calls -Werror")

# Find Protocol Buffers package
find_package(Protobuf REQUIRED)


# Find Abseil package
find_package(absl REQUIRED)
include_directories(${Protobuf_INCLUDE_DIRS})
include_directories(${CMAKE_CURRENT_BINARY_DIR})

# Generate protobuf files from proto directory
protobuf_generate_cpp(PROTO_SRCS PROTO_HDRS proto/contact.proto)

# Add all source files
add_executable(main
    src/main.cpp
    ${PROTO_SRCS}
    ${PROTO_HDRS}
)

# Link necessary libraries
target_link_libraries(main 
    ${Protobuf_LIBRARIES}
    absl::log
    absl::log_internal_message
    absl::log_internal_check_op
)

# Add include directories
target_include_directories(main PRIVATE
    ${CMAKE_CURRENT_SOURCE_DIR}/src
    ${CMAKE_CURRENT_BINARY_DIR}
    ${Protobuf_INCLUDE_DIRS}
)

after I build (creates a ./build folder) and run the code I am getting the following error message:

 ./build/main                
Serialized date size: 7
Deserialized date: 1990-6-15
================================================
Successfully serialized contact (size: 45 bytes)
Serialized contact size: 45
AddressSanitizer:DEADLYSIGNAL
=================================================================
==76867==ERROR: AddressSanitizer: SEGV on unknown address 0x000000000000 (pc 0x000000000000 bp 0x7ff7b890a050 sp 0x7ff7b890a008 T0)
==76867==Hint: pc points to the zero page.
==76867==The signal is caused by a READ memory access.
==76867==Hint: address points to the zero page.
    #0 0x000000000000  (<unknown module>)
    #1 0x000107a3dbdb in bool google::protobuf::internal::MergeFromImpl<false>(google::protobuf::io::ZeroCopyInputStream*, google::protobuf::MessageLite*, google::protobuf::internal::TcParseTableBase const*, google::protobuf::MessageLite::ParseFlags)+0xd1 (libprotobuf.29.3.0.dylib:x86_64+0x10cbdb)
    #2 0x000107a3eda2 in google::protobuf::MessageLite::ParseFromIstream(std::__1::basic_istream<char, std::__1::char_traits<char>>*)+0x32 (libprotobuf.29.3.0.dylib:x86_64+0x10dda2)
    #3 0x000107603cd7 in main main.cpp:57
    #4 0x7ff8165ad52f in start+0xbef (dyld:x86_64+0xfffffffffff1f52f) 

The error is cased by the birthday field of the type Date in the Contact message.
For some reason when I use:

  contact.mutable_birthday()->set_year(1990);
  contact.mutable_birthday()->set_month(6);
  contact.mutable_birthday()->set_day(15);

or CopyFrom method it causes a segmentation fault.

Summary:

1. If I remove Date field from the Contact message (and all the code in main() related to setting Date field), then problem disappears.

2. When I leave Date field in the Contact message BUT I don't use set methods for this field in main(), then problem dissappears as well.

Basically: problem occurs due to set methods (contact.mutable_birthday()->set_year(1990) etc.)

• OS: macOS
• protoc version: libprotoc 29.3
• generated files contact.pb.cc and contact.pb.h are in the build folder, if it's important.