Struct TcpSocket

pub struct TcpSocket { /* private fields */ }

Expand description

A TCP socket that has not yet been converted to a TcpStream or TcpListener.

TcpSocket wraps an operating system socket and enables the caller to configure the socket before establishing a TCP connection or accepting inbound connections. The caller is able to set socket option and explicitly bind the socket with a socket address.

The underlying socket is closed when the TcpSocket value is dropped.

TcpSocket should only be used directly if the default configuration used by TcpStream::connect and TcpListener::bind does not meet the required use case.

Calling TcpStream::connect("127.0.0.1:8080") is equivalent to:

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "127.0.0.1:8080".parse().unwrap();

    let socket = TcpSocket::new_v4()?;
    let stream = socket.connect(addr).await?;

    Ok(())
}

Calling TcpListener::bind("127.0.0.1:8080") is equivalent to:

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "127.0.0.1:8080".parse().unwrap();

    let socket = TcpSocket::new_v4()?;
    // On platforms with Berkeley-derived sockets, this allows to quickly
    // rebind a socket, without needing to wait for the OS to clean up the
    // previous one.
    //
    // On Windows, this allows rebinding sockets which are actively in use,
    // which allows “socket hijacking”, so we explicitly don't set it here.
    // https://docs.microsoft.com/en-us/windows/win32/winsock/using-so-reuseaddr-and-so-exclusiveaddruse
    socket.set_reuseaddr(true)?;
    socket.bind(addr)?;

    let listener = socket.listen(1024)?;

    Ok(())
}

Setting socket options not explicitly provided by TcpSocket may be done by accessing the RawFd/RawSocket using AsRawFd/AsRawSocket and setting the option with a crate like socket2.

Implementations§

§

impl TcpSocket

pub fn new_v4() -> Result<TcpSocket, Error>

Creates a new socket configured for IPv4.

Calls socket(2) with AF_INET and SOCK_STREAM.

§Returns

On success, the newly created TcpSocket is returned. If an error is encountered, it is returned instead.

§Examples

Create a new IPv4 socket and start listening.

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "127.0.0.1:8080".parse().unwrap();
    let socket = TcpSocket::new_v4()?;
    socket.bind(addr)?;

    let listener = socket.listen(128)?;
    Ok(())
}

pub fn new_v6() -> Result<TcpSocket, Error>

Creates a new socket configured for IPv6.

Calls socket(2) with AF_INET6 and SOCK_STREAM.

§Returns

On success, the newly created TcpSocket is returned. If an error is encountered, it is returned instead.

§Examples

Create a new IPv6 socket and start listening.

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "[::1]:8080".parse().unwrap();
    let socket = TcpSocket::new_v6()?;
    socket.bind(addr)?;

    let listener = socket.listen(128)?;
    Ok(())
}

pub fn set_keepalive(&self, keepalive: bool) -> Result<(), Error>

Sets value for the SO_KEEPALIVE option on this socket.

pub fn keepalive(&self) -> Result<bool, Error>

Gets the value of the SO_KEEPALIVE option on this socket.

pub fn set_reuseaddr(&self, reuseaddr: bool) -> Result<(), Error>

Allows the socket to bind to an in-use address.

Behavior is platform specific. Refer to the target platform’s documentation for more details.

§Examples

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "127.0.0.1:8080".parse().unwrap();

    let socket = TcpSocket::new_v4()?;
    socket.set_reuseaddr(true)?;
    socket.bind(addr)?;

    let listener = socket.listen(1024)?;

    Ok(())
}

pub fn reuseaddr(&self) -> Result<bool, Error>

Retrieves the value set for SO_REUSEADDR on this socket.

§Examples

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "127.0.0.1:8080".parse().unwrap();

    let socket = TcpSocket::new_v4()?;
    socket.set_reuseaddr(true)?;
    assert!(socket.reuseaddr().unwrap());
    socket.bind(addr)?;

    let listener = socket.listen(1024)?;
    Ok(())
}

pub fn set_reuseport(&self, reuseport: bool) -> Result<(), Error>

Allows the socket to bind to an in-use port. Only available for unix systems (excluding Solaris & Illumos).

Behavior is platform specific. Refer to the target platform’s documentation for more details.

§Examples

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "127.0.0.1:8080".parse().unwrap();

    let socket = TcpSocket::new_v4()?;
    socket.set_reuseport(true)?;
    socket.bind(addr)?;

    let listener = socket.listen(1024)?;
    Ok(())
}

pub fn reuseport(&self) -> Result<bool, Error>

Allows the socket to bind to an in-use port. Only available for unix systems (excluding Solaris & Illumos).

Behavior is platform specific. Refer to the target platform’s documentation for more details.

§Examples

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "127.0.0.1:8080".parse().unwrap();

    let socket = TcpSocket::new_v4()?;
    socket.set_reuseport(true)?;
    assert!(socket.reuseport().unwrap());
    socket.bind(addr)?;

    let listener = socket.listen(1024)?;
    Ok(())
}

pub fn set_send_buffer_size(&self, size: u32) -> Result<(), Error>

Sets the size of the TCP send buffer on this socket.

On most operating systems, this sets the SO_SNDBUF socket option.

pub fn send_buffer_size(&self) -> Result<u32, Error>

Returns the size of the TCP send buffer for this socket.

On most operating systems, this is the value of the SO_SNDBUF socket option.

Note that if set_send_buffer_size has been called on this socket previously, the value returned by this function may not be the same as the argument provided to set_send_buffer_size. This is for the following reasons:

Most operating systems have minimum and maximum allowed sizes for the send buffer, and will clamp the provided value if it is below the minimum or above the maximum. The minimum and maximum buffer sizes are OS-dependent.
Linux will double the buffer size to account for internal bookkeeping data, and returns the doubled value from getsockopt(2). As per man 7 socket:

Sets or gets the maximum socket send buffer in bytes. The kernel doubles this value (to allow space for bookkeeping overhead) when it is set using setsockopt(2), and this doubled value is returned by getsockopt(2).

pub fn set_recv_buffer_size(&self, size: u32) -> Result<(), Error>

Sets the size of the TCP receive buffer on this socket.

On most operating systems, this sets the SO_RCVBUF socket option.

pub fn recv_buffer_size(&self) -> Result<u32, Error>

Returns the size of the TCP receive buffer for this socket.

On most operating systems, this is the value of the SO_RCVBUF socket option.

Note that if set_recv_buffer_size has been called on this socket previously, the value returned by this function may not be the same as the argument provided to set_send_buffer_size. This is for the following reasons:

Most operating systems have minimum and maximum allowed sizes for the receive buffer, and will clamp the provided value if it is below the minimum or above the maximum. The minimum and maximum buffer sizes are OS-dependent.
Linux will double the buffer size to account for internal bookkeeping data, and returns the doubled value from getsockopt(2). As per man 7 socket:

Sets or gets the maximum socket send buffer in bytes. The kernel doubles this value (to allow space for bookkeeping overhead) when it is set using setsockopt(2), and this doubled value is returned by getsockopt(2).

pub fn set_linger(&self, dur: Option<Duration>) -> Result<(), Error>

Sets the linger duration of this socket by setting the SO_LINGER option.

This option controls the action taken when a stream has unsent messages and the stream is closed. If SO_LINGER is set, the system shall block the process until it can transmit the data or until the time expires.

If SO_LINGER is not specified, and the socket is closed, the system handles the call in a way that allows the process to continue as quickly as possible.

pub fn linger(&self) -> Result<Option<Duration>, Error>

Reads the linger duration for this socket by getting the SO_LINGER option.

For more information about this option, see set_linger.

pub fn set_nodelay(&self, nodelay: bool) -> Result<(), Error>

Sets the value of the TCP_NODELAY option on this socket.

If set, this option disables the Nagle algorithm. This means that segments are always sent as soon as possible, even if there is only a small amount of data. When not set, data is buffered until there is a sufficient amount to send out, thereby avoiding the frequent sending of small packets.

§Examples

use tokio::net::TcpSocket;

let socket = TcpSocket::new_v4()?;

println!("{:?}", socket.nodelay()?);

pub fn nodelay(&self) -> Result<bool, Error>

Gets the value of the TCP_NODELAY option on this socket.

For more information about this option, see set_nodelay.

§Examples

use tokio::net::TcpSocket;

let stream = TcpSocket::new_v4()?;

stream.set_nodelay(true)?;

pub fn tos(&self) -> Result<u32, Error>

Gets the value of the IP_TOS option for this socket.

For more information about this option, see set_tos.

NOTE: On Windows, IP_TOS is only supported on Windows 8+ or Windows Server 2012+.

pub fn set_tos(&self, tos: u32) -> Result<(), Error>

Sets the value for the IP_TOS option on this socket.

This value sets the type-of-service field that is used in every packet sent from this socket.

NOTE: On Windows, IP_TOS is only supported on Windows 8+ or Windows Server 2012+.

pub fn device(&self) -> Result<Option<Vec<u8>>, Error>

Gets the value for the SO_BINDTODEVICE option on this socket

This value gets the socket binded device’s interface name.

pub fn bind_device(&self, interface: Option<&[u8]>) -> Result<(), Error>

Sets the value for the SO_BINDTODEVICE option on this socket

If a socket is bound to an interface, only packets received from that particular interface are processed by the socket. Note that this only works for some socket types, particularly AF_INET sockets.

If interface is None or an empty string it removes the binding.

pub fn local_addr(&self) -> Result<SocketAddr, Error>

Gets the local address of this socket.

Will fail on windows if called before bind.

§Examples

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "127.0.0.1:8080".parse().unwrap();

    let socket = TcpSocket::new_v4()?;
    socket.bind(addr)?;
    assert_eq!(socket.local_addr().unwrap().to_string(), "127.0.0.1:8080");
    let listener = socket.listen(1024)?;
    Ok(())
}

pub fn take_error(&self) -> Result<Option<Error>, Error>

Returns the value of the SO_ERROR option.

pub fn bind(&self, addr: SocketAddr) -> Result<(), Error>

Binds the socket to the given address.

This calls the bind(2) operating-system function. Behavior is platform specific. Refer to the target platform’s documentation for more details.

§Examples

Bind a socket before listening.

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "127.0.0.1:8080".parse().unwrap();

    let socket = TcpSocket::new_v4()?;
    socket.bind(addr)?;

    let listener = socket.listen(1024)?;

    Ok(())
}

pub async fn connect(self, addr: SocketAddr) -> Result<TcpStream, Error>

Establishes a TCP connection with a peer at the specified socket address.

The TcpSocket is consumed. Once the connection is established, a connected TcpStream is returned. If the connection fails, the encountered error is returned.

This calls the connect(2) operating-system function. Behavior is platform specific. Refer to the target platform’s documentation for more details.

§Examples

Connecting to a peer.

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "127.0.0.1:8080".parse().unwrap();

    let socket = TcpSocket::new_v4()?;
    let stream = socket.connect(addr).await?;

    Ok(())
}

pub fn listen(self, backlog: u32) -> Result<TcpListener, Error>

Converts the socket into a TcpListener.

backlog defines the maximum number of pending connections are queued by the operating system at any given time. Connection are removed from the queue with TcpListener::accept. When the queue is full, the operating-system will start rejecting connections.

This calls the listen(2) operating-system function, marking the socket as a passive socket. Behavior is platform specific. Refer to the target platform’s documentation for more details.

§Examples

Create a TcpListener.

use tokio::net::TcpSocket;

use std::io;

#[tokio::main]
async fn main() -> io::Result<()> {
    let addr = "127.0.0.1:8080".parse().unwrap();

    let socket = TcpSocket::new_v4()?;
    socket.bind(addr)?;

    let listener = socket.listen(1024)?;

    Ok(())
}

pub fn from_std_stream(std_stream: TcpStream) -> TcpSocket

Converts a std::net::TcpStream into a TcpSocket. The provided socket must not have been connected prior to calling this function. This function is typically used together with crates such as socket2 to configure socket options that are not available on TcpSocket.

§Notes

The caller is responsible for ensuring that the socket is in non-blocking mode. Otherwise all I/O operations on the socket will block the thread, which will cause unexpected behavior. Non-blocking mode can be set using set_nonblocking.

§Examples

use tokio::net::TcpSocket;
use socket2::{Domain, Socket, Type};

#[tokio::main]
async fn main() -> std::io::Result<()> {
    let socket2_socket = Socket::new(Domain::IPV4, Type::STREAM, None)?;
    socket2_socket.set_nonblocking(true)?;

    let socket = TcpSocket::from_std_stream(socket2_socket.into());

    Ok(())
}