Boost C++ Libraries of the most highly regarded and expertly designed C++ library projects in the world. Herb Sutter and Andrei Alexandrescu, C++ Coding Standards

basic_datagram_socket::native_non_blocking (1 of 3 overloads)

Inherited from basic_socket.

Gets the non-blocking mode of the native socket implementation.

bool native_non_blocking() const;

This function is used to retrieve the non-blocking mode of the underlying native socket. This mode has no effect on the behaviour of the socket object's synchronous operations.

Return Value

true if the underlying socket is in non-blocking mode and direct system calls may fail with boost::asio::error::would_block (or the equivalent system error).


The current non-blocking mode is cached by the socket object. Consequently, the return value may be incorrect if the non-blocking mode was set directly on the native socket.


This function is intended to allow the encapsulation of arbitrary non-blocking system calls as asynchronous operations, in a way that is transparent to the user of the socket object. The following example illustrates how Linux's sendfile system call might be encapsulated:

template <typename Handler>
struct sendfile_op
  tcp::socket& sock_;
  int fd_;
  Handler handler_;
  off_t offset_;
  std::size_t total_bytes_transferred_;

  // Function call operator meeting WriteHandler requirements.
  // Used as the handler for the async_write_some operation.
  void operator()(boost::system::error_code ec, std::size_t)
    // Put the underlying socket into non-blocking mode.
    if (!ec)
      if (!sock_.native_non_blocking())
        sock_.native_non_blocking(true, ec);

    if (!ec)
      for (;;)
        // Try the system call.
        errno = 0;
        int n = ::sendfile(sock_.native_handle(), fd_, &offset_, 65536);
        ec = boost::system::error_code(n < 0 ? errno : 0,
        total_bytes_transferred_ += ec ? 0 : n;

        // Retry operation immediately if interrupted by signal.
        if (ec == boost::asio::error::interrupted)

        // Check if we need to run the operation again.
        if (ec == boost::asio::error::would_block
            || ec == boost::asio::error::try_again)
          // We have to wait for the socket to become ready again.
          sock_.async_wait(tcp::socket::wait_write, *this);

        if (ec || n == 0)
          // An error occurred, or we have reached the end of the file.
          // Either way we must exit the loop so we can call the handler.

        // Loop around to try calling sendfile again.

    // Pass result back to user's handler.
    handler_(ec, total_bytes_transferred_);

template <typename Handler>
void async_sendfile(tcp::socket& sock, int fd, Handler h)
  sendfile_op<Handler> op = { sock, fd, h, 0, 0 };
  sock.async_wait(tcp::socket::wait_write, op);