提升版本:1.68
C ++标准:C ++ 17
开发平台:MSVC 2017
操作系统:Windows 10 Professional
PC体系结构:x64
我正在使用Boost :: Asio创建异步TCP连接。在首次成功连接期间,一切正常。由于某些问题,套接字中断,它尝试重新连接,那时候我收到运行时错误。即使出现运行时错误,程序仍然可以接收数据。
起初,我试图在我的主函数的while(无限)循环(主线程)中重新连接套接字,但是我遇到了错误。
出现错误D:\ vcpkg \ installed \ x64-windows \ include \ boost \ asio \ detail \ impl \ win_iocp_io_context.ipp
size_t win_iocp_io_context::run(boost::system::error_code& ec)
{
if (::InterlockedExchangeAdd(&outstanding_work_, 0) == 0)
{
stop();
ec = boost::system::error_code();
return 0;
}
win_iocp_thread_info this_thread;
thread_call_stack::context ctx(this, this_thread);
size_t n = 0;
while (do_one(INFINITE, ec))
if (n != (std::numeric_limits<size_t>::max)())
++n;
return n;
}
当n = 13时在线
同时(do_one(INFINITE,ec))
我通过在接收和发送处理程序中检测到连接断开或套接字关闭的情况下添加连接调用来解决该错误。
有人可以解释为什么我尝试从主线程重新连接时遇到问题,并且在io_context线程中的套接字断开后立即尝试重新连接时,该问题得到解决。
在连接丢失并且io_context线程退出之后,我在调用run之前确实重启了上下文。
在我的主要功能中循环:
while (true)
{
fmt::print("Socket Alive : {}\n", as.isSocketAlive());
while(not as.isSocketAlive() and not as.isConnectionInProcess())
as.connectSocket();
if (not as.isSocketAlive())
continue;
if (as.isReadComplete())
as.sendDataSocket(as.getRecievedData());
}
我的异步套接字函数:
void AsyncSocket::runContext(void)
{
std::atomic_store(std::addressof(this->contextExitted), false);
this->context.run();
std::atomic_store(std::addressof(this->contextExitted), true);
}
void AsyncSocket::restartContext(void)
{
if (std::atomic_load(std::addressof(this->contextExitted)))
this->context.restart();
}
void AsyncSocket::connectSocket(void)
{
std::atomic_store(std::addressof(this->socketAlive), false);
std::atomic_store(std::addressof(this->connectionInProcess), true);
//this->socket.async_connect(this->endpoint, std::bind(&AsyncSocket::connectHandler, this, boost::asio::placeholders::error));
this->socket.async_connect(this->endpoint, boost::bind(&AsyncSocket::connectHandler, this, boost::asio::placeholders::error));
//this->socket.connect(this->endpoint, );
if (this->context.stopped())
{
if (this->isContextExitted())
this->restartContext();
this->t = std::thread(&AsyncSocket::runContext, this);
}
std::call_once(this->firstRun, [this]() {
t = std::thread(&AsyncSocket::runContext, this);
});
}
void AsyncSocket::recieveHandler(const boost::system::error_code& ec, std::size_t bytes_transferred)
{
fmt::print("In recieve handler, Error Code : {}\nBytes recieved : {}\n", ec.message(), bytes_transferred);
try
{
std::atomic_store(std::addressof(this->receivingData), false);
if (ec not_eq boost::system::errc::success)
{
#ifdef _DEBUG
fmt::print("Error in WebSocket::recieveHandler. Error : {0}\n", ec.message());
#endif // _DEBUG
LOG_ERROR << ec.message();
switch (ec.value())
{
case boost::asio::error::eof :
case boost::asio::error::connection_reset :
case boost::asio::error::connection_aborted :
case boost::asio::error::network_reset :
case boost::asio::error::network_down :
case boost::asio::error::network_unreachable :
if (this->isSocketAlive() and this->socket.is_open())
this->socket.close();
std::atomic_store(std::addressof(this->socketAlive), false);
this->connectSocket();// If I comment this line and try to reconnect in my main function (infinite while loop), I get mentioned run-time error
return;
default:
break;
}
}
else
{
this->readDataQueue.push(std::string(reinterpret_cast <const char *> (this->readDataBuffer.data()), bytes_transferred));
std::atomic_store(std::addressof(this->readComplete), true);
}
}
catch (const std::exception& ex)
{
#ifdef _DEBUG
fmt::print("Error in WebSocket::sendHandler. Error : {0}\n", ex.what());
#endif // _DEBUG
LOG_ERROR << "Exception : " << ex.what() << "Data : " << this->writeDataQueue.front();
}
this->recieveDataSocket();
}
void AsyncSocket::sendHandler(const boost::system::error_code& ec, std::size_t bytes_transferred)
{
fmt::print("In send handler, Error Code : {}\nBytes recieved : {}\n", ec.message(), bytes_transferred);
try
{
if (ec not_eq boost::system::errc::success)
{
#ifdef _DEBUG
fmt::print("Error in WebSocket::recieveHandler. Error : {0}\n", ec.message());
#endif // _DEBUG
LOG_ERROR << ec.message();
switch (ec.value())
{
case boost::asio::error::eof:
case boost::asio::error::connection_reset:
case boost::asio::error::connection_aborted:
case boost::asio::error::network_reset:
case boost::asio::error::network_down:
case boost::asio::error::network_unreachable:
if (this->isSocketAlive() and this->socket.is_open())
this->socket.close();
std::atomic_store(std::addressof(this->socketAlive), false);
this->connectSocket();// If I comment this line and try to reconnect in my main function (infinite while loop), I get mentioned run-time error
return;
default:
break;
}
}
this->writeDataQueue.pop();
std::atomic_init(std::addressof(this->sendingData), false);
this->writeSocket();
}
catch (const std::exception& ex)
{
#ifdef _DEBUG
fmt::print("Error in WebSocket::sendHandler. Error : {0}\n", ex.what());
#endif // _DEBUG
LOG_ERROR << "Exception : " << ex.what() << "Data : " << this->writeDataQueue.front();
}
}
答案 0 :(得分:1)
abort
被std::terminate
称为默认操作。当线程被销毁或被terminate
覆盖且处于 joinable 状态时,将调用operator=(thread&&)
。
此代码段不安全:
connectSocket() {
...
if (this->context.stopped())
{
if (this->isContextExitted())
this->restartContext();
this->t = std::thread(&AsyncSocket::runContext, this); // [1]
}
std::call_once(this->firstRun, [this]() {
t = std::thread(&AsyncSocket::runContext, this); // [2]
});
在 connectSocket 第[2]行的第一次调用中执行并启动了线程。如果context.stopped()
返回true,则在 connectSocket 的第N次调用时,您正在创建新线程并将其分配给t
-可连接状态的线程,此时{{1 }} 叫做。在进行移动分配之前,您应该加入terminate
线程。