Support for multiple listen sockets #27
Comments
Hi, I have a couple of questions for you (since you surely are more experienced with OpenBSD than I am):
use std::net::UdpSocket;
fn main() {
let ipv4 = "0.0.0.0:1024";
let ipv6 = ":::1024";
let v4socket = UdpSocket::bind(ipv4).unwrap();
let v6socket = UdpSocket::bind(ipv6).unwrap();
}/cc @emilengler @koraa |
|
Windows apparently also defaults to not being dual stack. So I see two options: either we introduce dynamic detection of dual-stack and on-demand adding of a second socket, or we just default to either IPv4 or IPv6 all-interface listening, while informing the user on how to change the default with an explicit |
|
|
We're currently not testing under openbsd; we should start to add the BSDs and Mac as supported platforms. The |
|
Ah, this might also explain why I couldn't get a tunnel to work on FreeBSD. It likely has the same behaviour. Treating this as a blocking issue for finishing the port. |
|
I think the following makes it work for me but I am not familiar with rust and as discussed it should break Linux. Index: src/main.rs
--- src/main.rs.orig
+++ src/main.rs
@@ -211,6 +211,7 @@ pub enum Verbosity {
pub struct AppServer {
pub crypt: CryptoServer,
pub sock: UdpSocket,
+ pub sock4: UdpSocket,
pub peers: Vec<AppPeer>,
pub verbosity: Verbosity,
}
@@ -449,6 +450,7 @@ impl AppServer {
Ok(Self {
crypt: CryptoServer::new(sk, pk),
sock: UdpSocket::bind(addr)?,
+ sock4: UdpSocket::bind("0.0.0.0:0")?,
peers: Vec::new(),
verbosity,
})
@@ -520,7 +522,10 @@ impl AppServer {
let p = $peer.get_app(self);
if let Some(addr) = p.tx_addr {
let len = $fn()?;
- self.sock.send_to(&tx[..len], addr)?;
+ match self.sock.send_to(&tx[..len], addr) {
+ Ok(x) => { x },
+ Err(_) => { self.sock4.send_to(&tx[..len], addr)? },
+ };
}
Ok(())
})
@@ -547,7 +552,10 @@ impl AppServer {
info!("error processing incoming message from {:?}: {:?} {}", addr, e, e.backtrace())),
Ok(HandleMsgResult { resp: Some(len), .. }) => {
- self.sock.send_to(&tx[0..len], addr)?
+ match self.sock.send_to(&tx[..len], addr) {
+ Ok(x) => { x },
+ Err(_) => { self.sock4.send_to(&tx[..len], addr)? },
+ };
},
Ok(HandleMsgResult { exchanged_with: Some(p), .. }) => {
@@ -634,7 +642,17 @@ impl AppServer {
}
self.sock
.set_read_timeout(Some(Duration::from_secs_f64(timeout)))?;
+ self.sock4
+ .set_read_timeout(Some(Duration::from_secs_f64(timeout)))?;
match self.sock.recv_from(buf) {
+ Ok(x) => return Ok(Some(x)),
+ Err(e) => match e.kind() {
+ ErrorKind::WouldBlock => {},
+ ErrorKind::TimedOut => {},
+ _ => return Err(anyhow::Error::new(e)),
+ },
+ };
+ match self.sock4.recv_from(buf) {
Ok(x) => Ok(Some(x)),
Err(e) => match e.kind() {
ErrorKind::WouldBlock => Ok(None), |
|
@moritzbuhl Thank you for your code contribution! Would you mind opening a PR? I don't think that code is what we will arrive, but it is a starting point for development. I think the correct thing to do is to have support for multiple I will also create a "BSD Support" meta ticket to track what needs to be done to fully support BSD. |
|
Hello, long-term OpenBSD Desktop user here. I have been learning network development on OpenBSD, so I am quite familiar with that issue :D The problem lies within the fact, that OpenBSDs IPv6 sockets ONLY support IPv6, not the IPv4 backwards compatibility. Unfortunately, most applications use the latter approach to do a dual stack. The OpenBSD way would be to create two sockets and set the IPv6 socket to Years ago, I was working on an OpenBSD-exclusive privilege-separated POP3-Daemon using dual-stack. The code looked like this: static void
net_sockets_init(uint16_t port)
{
struct sockaddr_in sa4;
struct sockaddr_in6 sa6;
if ((net_in4 = socket(AF_INET, SOCK_STREAM, 0)) == -1)
fatal("socket AF_INET");
if ((net_in6 = socket(AF_INET6, SOCK_STREAM, 0)) == -1)
fatal("socket AF_INET6");
/* Portable code would require a setsockopt(2) here */
if (net_fd_nonblock(net_in4) == -1)
fatal("net_fd_nonblock AF_INET");
if (net_fd_nonblock(net_in6) == -1)
fatal("net_fd_nonblock AF_INET6");
/* Steps related for bind(2). */
memset(&sa4, 0, sizeof(sa4));
sa4.sin_family = AF_INET;
sa4.sin_port = htons(port);
sa4.sin_addr.s_addr = INADDR_ANY;
memset(&sa6, 0, sizeof(sa6));
sa6.sin6_family = AF_INET6;
sa6.sin6_port = htons(port);
sa6.sin6_addr = in6addr_any;
if (bind(net_in4, (struct sockaddr *)&sa4, sizeof(sa4)) == -1)
fatal("bind AF_INET");
if (bind(net_in6, (struct sockaddr *)&sa6, sizeof(sa6)) == -1)
fatal("bind AF_INET6");
if (listen(net_in4, 128) == -1)
fatal("listen AF_INET");
if (listen(net_in6, 128) == -1)
fatal("listen AF_INET6");
}
Yes
Yes, but in order to achieve compatibility with other systems, you have to make the IPv6 socket IPv6 only (default on OpenBSD). It requires a
I would not introduce conditional compilation. The dual socket approach that OpenBSD uses (one V6-only and one V4-only socket) works cross platform. Sure, it might add a tiny overhead, but having a common codebase across architectures is probably better. |
|
I see some problems with the multi-socket approach, foremost that we use
@koraa what is your call on this? The way I see it, our best option is either platform-dependent code (clean, but a high maintenance burden), async for the frontend or just a single listen address per rosenpass process and no happy non-v4-v6-dualstack users. |
There are |
|
I agree with @clausecker. |
|
I looked a little more into this. The correct way IMHO would be usin |
- adds TOML based configuation files - with example configuratios in config-examples - reimplments arcane CLI argument parser as automaton - adds a new CLI focused arround configuration files - moves all file utility stuff from `main.rs` to `util.rs` - moves all AppServer stuff to dedicated `app_server.rs` - add mio for multi-listen-socket support (should fix #27) - consistency: rename private to secret
- adds TOML based configuation files - with example configuratios in config-examples - reimplments arcane CLI argument parser as automaton - adds a new CLI focused arround configuration files - moves all file utility stuff from `main.rs` to `util.rs` - moves all AppServer stuff to dedicated `app_server.rs` - add mio for multi-listen-socket support (should fix #27) - consistency: rename private to secret
- adds TOML based configuation files - with example configuratios in config-examples - reimplments arcane CLI argument parser as automaton - adds a new CLI focused arround configuration files - moves all file utility stuff from `main.rs` to `util.rs` - moves all AppServer stuff to dedicated `app_server.rs` - add mio for multi-listen-socket support (should fix #27) - consistency: rename private to secret
- adds TOML based configuation files - with example configuratios in config-examples - reimplments arcane CLI argument parser as automaton - adds a new CLI focused arround configuration files - moves all file utility stuff from `main.rs` to `util.rs` - moves all AppServer stuff to dedicated `app_server.rs` - add mio for multi-listen-socket support (should fix #27) - consistency: rename private to secret
- adds TOML based configuation files - with example configuratios in config-examples - reimplments arcane CLI argument parser as automaton - adds a new CLI focused arround configuration files - moves all file utility stuff from `main.rs` to `util.rs` - moves all AppServer stuff to dedicated `app_server.rs` - add mio for multi-listen-socket support (should fix #27) - consistency: rename private to secret
- adds TOML based configuation files - with example configuratios in config-examples - reimplments arcane CLI argument parser as automaton - adds a new CLI focused arround configuration files - moves all file utility stuff from `main.rs` to `util.rs` - moves all AppServer stuff to dedicated `app_server.rs` - add mio for multi-listen-socket support (should fix #27) - consistency: rename private to secret
- adds TOML based configuation files - with example configuratios in config-examples - reimplments arcane CLI argument parser as automaton - adds a new CLI focused arround configuration files - moves all file utility stuff from `main.rs` to `util.rs` - moves all AppServer stuff to dedicated `app_server.rs` - add mio for multi-listen-socket support (should fix #27) - consistency: rename private to secret
|
Looking forwards to the next release! Will try another attempt at packaging then. |
I tried following this setup: https://rosenpass.eu/#start
Server: 192.168.2.107/24
Client: 192.168.2.106/24
However, I am getting this output from the client:
To my understanding this is because the socket the rosenpass binary is using for sendig is somehow bound for IPv6:
98150 rosenpass CALL socket(AF_INET6,0x8002,0) 98150 rosenpass RET socket 3 98150 rosenpass CALL bind(3,0x7f7ffffd19a8,28) 98150 rosenpass STRU struct sockaddr { AF_INET6, [::]:0 } 98150 rosenpass RET bind 0 ... 98150 rosenpass CALL sendto(3,0x7f7ffffd24b0,0x444,0x400,0x7f7ffffd1a30,0x10) 98150 rosenpass STRU struct sockaddr { AF_INET, 192.168.2.107:9999 } 98150 rosenpass RET sendto -1 errno 47 Address family not supported by protocol family 98150 rosenpass CALL write(2,0x878b524c500,0xd9) 98150 rosenpass GIO fd 2 wrote 217 bytes "\^[[0m\^[[38;5;8m[\^[[0m2023-03-14T20:59:57Z \^[[0m\^[[1m\^[[31mERROR\ \^[[0m rosenpass\^[[0m\^[[38;5;8m]\^[[0m unexpected error after proces\ sing 0 messages: Address family not supported by protocol family (os e\ rror 47) disabled backtrace " 98150 rosenpass RET write 217/0xd9 98150 rosenpass CALL write(2,0x878b524c500,0x8c) 98150 rosenpass GIO fd 2 wrote 140 bytes "\^[[0m\^[[38;5;8m[\^[[0m2023-03-14T20:59:57Z \^[[0m\^[[1m\^[[31mERROR\ \^[[0m rosenpass\^[[0m\^[[38;5;8m]\^[[0m reinitializing networking in \ 0.01! 10 tries left. "This unwrap_or makes the socket IPv6 since `listen` is not set for the client.To fix this, the client should bind a v4 socket too and depending on endpoint it should use that socket for sending.
Alternatively it could just try the v4 socket in case the v6 one failed.
I will look into implementing that but I have little knowledge on rust.
https://github.com/rosenpass/rosenpass/blob/main/src/main.rs#L356
Adding
listenfor the client makes it work with IPv4.Because the client is trying to send via an IPv6 socket, I tried just using IPv6 to connect to the server:
Server:
Client:
Now everything works, I can assign IP addresses to the wg/rosenpass interfaces and ping through them.
I had to adjust the rp script sligthly to make it work with OpenBSD, but I did not add any changes to the rosenpass binary:
https://github.com/moritzbuhl/mystuff/tree/master/net/rosenpass
The text was updated successfully, but these errors were encountered: