Occasionnally crash in fastrtps

[BaiCai1990]

Bug report

Required Info:

• Operating System:
  Debian 10 buster
• Installation type:
  from source
• Version or commit hash:
• DDS implementation:
  Fast-RTPS
• Client library (if applicable):
  rclcpp

Steps to reproduce issue

backtrace:
#0  0x00007ffff2159704 in eprosima::fastrtps::rtps::StatefulReader::send_acknack(eprosima::fastrtps::rtps::WriterProxy const*, eprosima::fastrtps::rtps::RTPSMessageSenderInterface const&, bool) () from /opt/ros2_foxy/install/fastrtps/lib/libfastrtps.so.2
#1  0x00007ffff2156c7c in std::_Function_handler<bool (), eprosima::fastrtps::rtps::WriterProxy::WriterProxy(eprosima::fastrtps::rtps::StatefulReader*, eprosima::fastrtps::rtps::RemoteLocatorsAllocationAttributes const&, eprosima::fastrtps::ResourceLimitedContainerConfig const&)::{lambda()#1}>::_M_invoke(std::_Any_data const&) () from /opt/ros2_foxy/install/fastrtps/lib/libfastrtps.so.2
#2  0x00007ffff212a080 in eprosima::fastrtps::rtps::TimedEventImpl::trigger(std::chrono::time_point<std::chrono::_V2::steady_clock, std::chrono::duration<long, std::ratio<1l, 1000000000l> > >, std::chrono::time_point<std::chrono::_V2::steady_clock, std::chrono::duration<long, std::ratio<1l, 1000000000l> > >) () from /opt/ros2_foxy/install/fastrtps/lib/libfastrtps.so.2
#3  0x00007ffff2128bbc in eprosima::fastrtps::rtps::ResourceEvent::do_timer_actions() ()
   from /opt/ros2_foxy/install/fastrtps/lib/libfastrtps.so.2
#4  0x00007ffff21290bb in eprosima::fastrtps::rtps::ResourceEvent::event_service() ()
   from /opt/ros2_foxy/install/fastrtps/lib/libfastrtps.so.2
#5  0x00007ffff67c5b2f in ?? () from /usr/lib/x86_64-linux-gnu/libstdc++.so.6
#6  0x00007ffff7e6ffa3 in start_thread (arg=<optimized out>) at pthread_create.c:486
#7  0x00007ffff64a34cf in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95

Expected behavior

run normal

Actual behavior

occasionally crash.

Additional information

  // controller.cpp, **key code** 
  auto publish_interval_ms =
      std::chrono::milliseconds(static_cast<uint64_t>(1000.0 / 100));
  control_timer_ = ros_node_hdr_->create_wall_timer(publish_interval_ms, [=](){
    RunMain();
  });

void ControllerBridge::RunMain()
{
  // 获取当前状态
  GlobalMotionState current_state = motion_state_mechine_hdr_->GetCurrentMotionState();

  // 执行控制
  switch(current_state){
  case GlobalMotionState::IDLE:{
    idle_controller_hdr_->Init();
    idle_controller_hdr_->RunOnce();

    // switch to WALK status
    mutex_.lock();
    if(super_controller_hdr_->GetRemainRouteNum() > 0) {
      motion_state_mechine_hdr_->SignalWalk();
    }
    mutex_.unlock();

    data_sender_ptr_->PublishTaskStatus();
    break;
  }
  case GlobalMotionState::WALK:{
    // core control pragram
    mutex_.lock();
    LOG_INFO("walk begin");
    super_controller_hdr_->SetInput(msg_localizer_, msg_obs_info_, msg_axis_info_);
    if(super_controller_hdr_->RunOnce() && super_controller_hdr_->GetRemainRouteNum() <= 0){
      motion_state_mechine_hdr_->SignalWalkComplete();
      super_controller_hdr_->Init();                         
      LOG_INFO("task complete.");
    }
    mutex_.unlock();

    data_sender_ptr_->PublishTaskStatus();
    break;
  }
  case GlobalMotionState::MANUAL:{
    manual_controller_hdr_->RunOnce();

    // clear route list
    mutex_.lock();
    super_controller_hdr_->ClearRouteList();
    super_controller_hdr_->Init();
    mutex_.unlock();

    // reset task status
    TaskStateMotion::Instance()->ResetTaskStatus();

    data_sender_ptr_->PublishTaskStatus();
    break;
  }
  case GlobalMotionState::SEMI_AUTO:{
    semi_auto_controller_hdr_->RunOnce();
    // clear rouelist
    mutex_.lock();
    super_controller_hdr_->ClearRouteList();
    super_controller_hdr_->Init();
    mutex_.unlock();

    data_sender_ptr_->PublishTaskStatus();
    break;
  }
  case GlobalMotionState::ERROR:{
    error_controller_hdr_->RunOnce();
    data_sender_ptr_->PublishTaskStatus();
    break;
  }
  default:{
    LOG_ERROR("错误状态.");
  }
  }
}


***main.cpp
int main(int argc, char** argv) {
  rclcpp::init(argc, argv);
  std::shared_ptr<rclcpp::Node> node = std::make_shared<rclcpp::Node>("controller");

   ......

   ControllerHdr controller_hdr = std::make_shared<Controller>(node);

  controller_hdr->RunMain();
  rclcpp::shutdown();
  return 0;
}

---

Feature request

Feature description

Implementation considerations

[clalancette]

@BaiCai1990 Can you provide a reproducing example here? That is, how do you get into this situation?

@MiguelCompany @EduPonz This seems to be somewhere in the Fast-DDS stack. Could you take a look?

[BaiCai1990]

@clalancette I switched to cyclonedds， the probability of problems is greatly reduced.

[BaiCai1990]

The key code I have provide @clalancette

[hidmic]

@BaiCai1990 I believe @clalancette is asking for an example we can actually compile and execute to reproduce. It's quite difficult to track down the issue otherwise.

[hidmic]

@BaiCai1990 friendly ping

[BaiCai1990]

@hidmic The complete code i can not provide, Because it involves company secrets。But i will provide a simplified version recently, thinks for your attention.

[clalancette]

Since we cannot reproduce this and there is no example we can run, I'm going to close this out. Please feel free to reopen if you are still having problems and can provide an example showing the problem.