| 1 | |
|
| 2 | |
|
| 3 | |
|
| 4 | |
|
| 5 | |
|
| 6 | |
|
| 7 | |
|
| 8 | |
|
| 9 | |
|
| 10 | | use std::io::{Read, Write}; |
| 11 | | use std::net::{IpAddr, SocketAddr, TcpStream, ToSocketAddrs}; |
| 12 | | use std::os::unix::net::{UnixListener, UnixStream}; |
| 13 | | use std::path::Path; |
| 14 | | use std::sync::atomic::{AtomicBool, AtomicUsize, Ordering}; |
| 15 | | use std::sync::{mpsc, Arc}; |
| 16 | | use std::time::Duration; |
| 17 | |
|
| 18 | |
|
| 19 | | pub const VSOCK_PORT: u32 = 3128; |
| 20 | |
|
| 21 | | pub const GUEST_PROXY: &str = "http://127.0.0.1:3128"; |
| 22 | |
|
| 23 | | const PORT: u16 = 443; |
| 24 | |
|
| 25 | | const MAX_OPEN: usize = 64; |
| 26 | | const CONNECT_TIMEOUT: Duration = Duration::from_secs(10); |
| 27 | |
|
| 28 | | const MAX_HEAD: usize = 8 * 1024; |
| 29 | |
|
| 30 | |
|
| 31 | | #[derive(Debug, PartialEq, Eq)] |
| 32 | | pub enum Ask { |
| 33 | | Connect { host: String, port: u16 }, |
| 34 | | Refused(String), |
| 35 | | } |
| 36 | |
|
| 37 | |
|
| 38 | 5× | pub fn parse_head(head: &str) -> Ask { |
| 39 | 5× | let line = head.lines().next().unwrap_or(""); |
| 40 | 5× | let mut parts = line.split_whitespace(); |
| 41 | 5× | let (method, target, version) = (parts.next().unwrap_or(""), parts.next().unwrap_or(""), parts.next().unwrap_or("")); |
| 42 | 5× | if method != "CONNECT" { |
| 43 | 2× | return Ask::Refused(format!("only https is relayed (CONNECT), not {method:?}")); |
| 44 | 3× | } |
| 45 | 3× | if !version.starts_with("HTTP/1.") { |
| 46 | 0× | return Ask::Refused("not an HTTP/1 request".into()); |
| 47 | 3× | } |
| 48 | 3× | let Some((host, port)) = target.rsplit_once(':') else { return Ask::Refused(format!("CONNECT {target:?} names no port")) }; |
| 49 | 2× | let Ok(port) = port.parse::<u16>() else { return Ask::Refused(format!("CONNECT {target:?} has no valid port")) }; |
| 50 | 2× | let host = host.trim_start_matches('[').trim_end_matches(']').trim_end_matches('.').to_ascii_lowercase(); |
| 51 | 2× | if host.is_empty() { |
| 52 | 0× | return Ask::Refused("CONNECT names no host".into()); |
| 53 | 2× | } |
| 54 | 2× | Ask::Connect { host, port } |
| 55 | 5× | } |
| 56 | |
|
| 57 | |
|
| 58 | |
|
| 59 | 6× | pub fn decide(allow: &[String], host: &str, port: u16) -> Result<(), String> { |
| 60 | 6× | if port != PORT { |
| 61 | 1× | return Err(format!("only port {PORT} is relayed")); |
| 62 | 5× | } |
| 63 | 5× | if host.parse::<IpAddr>().is_ok() { |
| 64 | 1× | return Err("an address, not a name: the allow-list names hosts".into()); |
| 65 | 4× | } |
| 66 | 4× | if !flow_core::egress_allows(allow, host) { |
| 67 | 2× | return Err("not in the sandbox's egress allow-list".into()); |
| 68 | 2× | } |
| 69 | 2× | Ok(()) |
| 70 | 6× | } |
| 71 | |
|
| 72 | |
|
| 73 | |
|
| 74 | |
|
| 75 | 14× | pub fn public(ip: IpAddr) -> bool { |
| 76 | 14× | match ip { |
| 77 | 9× | IpAddr::V4(v4) => { |
| 78 | 9× | let o = v4.octets(); |
| 79 | 9× | !(v4.is_loopback() |
| 80 | 8× | || v4.is_private() |
| 81 | 4× | || v4.is_link_local() |
| 82 | 3× | || v4.is_multicast() |
| 83 | 3× | || v4.is_broadcast() |
| 84 | 3× | || v4.is_unspecified() |
| 85 | 2× | || v4.is_documentation() |
| 86 | 2× | || o[0] == 0 |
| 87 | 2× | || (o[0] == 100 && (o[1] & 0xc0) == 64) |
| 88 | 1× | || (o[0] == 192 && o[1] == 0 && o[2] == 0) |
| 89 | 1× | || o[0] >= 240) |
| 90 | | } |
| 91 | 5× | IpAddr::V6(v6) => { |
| 92 | 5× | if let Some(v4) = v6.to_ipv4_mapped() { |
| 93 | 1× | return public(IpAddr::V4(v4)); |
| 94 | 4× | } |
| 95 | 4× | let s = v6.segments(); |
| 96 | 4× | !(v6.is_loopback() |
| 97 | 3× | || v6.is_unspecified() |
| 98 | 3× | || v6.is_multicast() |
| 99 | 3× | || (s[0] & 0xfe00) == 0xfc00 |
| 100 | 2× | || (s[0] & 0xffc0) == 0xfe80 |
| 101 | 1× | || (s[0] == 0x64 && s[1] == 0xff9b) |
| 102 | 1× | || (s[0] == 0x2001 && s[1] == 0x0db8)) |
| 103 | | } |
| 104 | | } |
| 105 | 14× | } |
| 106 | |
|
| 107 | |
|
| 108 | | pub struct Proxy { |
| 109 | | stop: Arc<AtomicBool>, |
| 110 | | path: std::path::PathBuf, |
| 111 | | } |
| 112 | |
|
| 113 | | impl Drop for Proxy { |
| 114 | 1× | fn drop(&mut self) { |
| 115 | 1× | self.stop.store(true, Ordering::SeqCst); |
| 116 | | |
| 117 | 1× | let _ = UnixStream::connect(&self.path); |
| 118 | 1× | let _ = std::fs::remove_file(&self.path); |
| 119 | 1× | } |
| 120 | | } |
| 121 | |
|
| 122 | |
|
| 123 | |
|
| 124 | 1× | pub fn start(socket: &Path, allow: Vec<String>, log: mpsc::Sender<String>) -> std::io::Result<Proxy> { |
| 125 | 1× | let _ = std::fs::remove_file(socket); |
| 126 | 1× | let listener = UnixListener::bind(socket)?; |
| 127 | 1× | let stop = Arc::new(AtomicBool::new(false)); |
| 128 | 1× | let open = Arc::new(AtomicUsize::new(0)); |
| 129 | 1× | let allow = Arc::new(allow); |
| 130 | 1× | let s = stop.clone(); |
| 131 | 1× | std::thread::spawn(move || { |
| 132 | 3× | for conn in listener.incoming() { |
| 133 | 3× | if s.load(Ordering::SeqCst) { |
| 134 | 1× | break; |
| 135 | 2× | } |
| 136 | 2× | let Ok(conn) = conn else { continue }; |
| 137 | 2× | if open.load(Ordering::SeqCst) >= MAX_OPEN { |
| 138 | 0× | let _ = refuse(conn, "too many connections at once"); |
| 139 | 0× | let _ = log.send(format!("egress: refused a connection: {MAX_OPEN} already open")); |
| 140 | 0× | continue; |
| 141 | 2× | } |
| 142 | 2× | open.fetch_add(1, Ordering::SeqCst); |
| 143 | 2× | let (allow, log, open) = (allow.clone(), log.clone(), open.clone()); |
| 144 | 2× | std::thread::spawn(move || { |
| 145 | 2× | serve(conn, &allow, &log); |
| 146 | 2× | open.fetch_sub(1, Ordering::SeqCst); |
| 147 | 2× | }); |
| 148 | | } |
| 149 | 1× | }); |
| 150 | 1× | Ok(Proxy { stop, path: socket.to_path_buf() }) |
| 151 | 1× | } |
| 152 | |
|
| 153 | 2× | fn refuse(mut conn: UnixStream, why: &str) -> std::io::Result<()> { |
| 154 | 2× | let body = format!("{why}\n"); |
| 155 | 2× | write!(conn, "HTTP/1.1 403 Forbidden\r\nContent-Type: text/plain\r\nContent-Length: {}\r\nConnection: close\r\n\r\n{body}", body.len()) |
| 156 | 2× | } |
| 157 | |
|
| 158 | |
|
| 159 | 2× | fn read_head(conn: &mut UnixStream) -> std::io::Result<String> { |
| 160 | 2× | let mut head = Vec::new(); |
| 161 | 2× | let mut byte = [0u8; 1]; |
| 162 | 85× | while !head.ends_with(b"\r\n\r\n") { |
| 163 | 83× | if head.len() >= MAX_HEAD || conn.read(&mut byte)? == 0 { |
| 164 | 0× | break; |
| 165 | 83× | } |
| 166 | 83× | head.push(byte[0]); |
| 167 | | } |
| 168 | 2× | Ok(String::from_utf8_lossy(&head).into_owned()) |
| 169 | 2× | } |
| 170 | |
|
| 171 | 2× | fn serve(mut conn: UnixStream, allow: &[String], log: &mpsc::Sender<String>) { |
| 172 | 2× | let _ = conn.set_read_timeout(Some(Duration::from_secs(30))); |
| 173 | 2× | let head = match read_head(&mut conn) { |
| 174 | 2× | Ok(h) => h, |
| 175 | 0× | Err(_) => return, |
| 176 | | }; |
| 177 | 2× | let (host, port) = match parse_head(&head) { |
| 178 | 1× | Ask::Connect { host, port } => (host, port), |
| 179 | 1× | Ask::Refused(why) => { |
| 180 | 1× | let _ = log.send(format!("egress: refused: {why}")); |
| 181 | 1× | let _ = refuse(conn, &why); |
| 182 | 1× | return; |
| 183 | | } |
| 184 | | }; |
| 185 | 1× | if let Err(why) = decide(allow, &host, port) { |
| 186 | 1× | let _ = log.send(format!("egress: refused {host}:{port}: {why}")); |
| 187 | 1× | let _ = refuse(conn, &format!("{host}:{port}: {why}")); |
| 188 | 1× | return; |
| 189 | 0× | } |
| 190 | 0× | let addrs: Vec<SocketAddr> = match (host.as_str(), port).to_socket_addrs() { |
| 191 | 0× | Ok(a) => a.collect(), |
| 192 | 0× | Err(e) => { |
| 193 | 0× | let _ = log.send(format!("egress: refused {host}:{port}: it could not be resolved ({e})")); |
| 194 | 0× | let _ = refuse(conn, &format!("{host} could not be resolved")); |
| 195 | 0× | return; |
| 196 | | } |
| 197 | | }; |
| 198 | 0× | if let Some(bad) = addrs.iter().find(|a| !public(a.ip())) { |
| 199 | 0× | let why = format!("it resolves to {}, which is not a public address", bad.ip()); |
| 200 | 0× | let _ = log.send(format!("egress: refused {host}:{port}: {why}")); |
| 201 | 0× | let _ = refuse(conn, &format!("{host}: {why}")); |
| 202 | 0× | return; |
| 203 | 0× | } |
| 204 | 0× | let out = addrs.iter().find_map(|a| TcpStream::connect_timeout(a, CONNECT_TIMEOUT).ok()); |
| 205 | 0× | let Some(out) = out else { |
| 206 | 0× | let _ = log.send(format!("egress: {host}:{port} is allowed but did not answer")); |
| 207 | 0× | let _ = conn.write_all(b"HTTP/1.1 502 Bad Gateway\r\nConnection: close\r\n\r\n"); |
| 208 | 0× | return; |
| 209 | | }; |
| 210 | 0× | let _ = log.send(format!("egress: allowed {host}:{port}")); |
| 211 | 0× | if conn.write_all(b"HTTP/1.1 200 Connection established\r\n\r\n").is_err() { |
| 212 | 0× | return; |
| 213 | 0× | } |
| 214 | 0× | let _ = conn.set_read_timeout(None); |
| 215 | 0× | relay(conn, out); |
| 216 | 2× | } |
| 217 | |
|
| 218 | |
|
| 219 | 0× | fn relay(guest: UnixStream, out: TcpStream) { |
| 220 | 0× | let (Ok(mut guest_r), Ok(mut out_w)) = (guest.try_clone(), out.try_clone()) else { return }; |
| 221 | 0× | let up = std::thread::spawn(move || { |
| 222 | 0× | let _ = std::io::copy(&mut guest_r, &mut out_w); |
| 223 | 0× | let _ = out_w.shutdown(std::net::Shutdown::Write); |
| 224 | 0× | }); |
| 225 | 0× | let (mut out_r, mut guest_w) = (out, guest); |
| 226 | 0× | let _ = std::io::copy(&mut out_r, &mut guest_w); |
| 227 | 0× | let _ = guest_w.shutdown(std::net::Shutdown::Write); |
| 228 | 0× | let _ = up.join(); |
| 229 | 0× | } |
| 230 | |
|
| 231 | | #[cfg(test)] |
| 232 | | mod tests { |
| 233 | | use super::*; |
| 234 | |
|
| 235 | 6× | fn allow() -> Vec<String> { |
| 236 | 6× | vec!["*.googleapis.com".into(), "registry.opentofu.org".into()] |
| 237 | 6× | } |
| 238 | |
|
| 239 | | #[test] |
| 240 | 1× | fn only_connect_to_an_allowed_name_on_443_is_let_through() { |
| 241 | 1× | assert_eq!( |
| 242 | 1× | parse_head("CONNECT storage.googleapis.com:443 HTTP/1.1\r\nHost: x\r\n\r\n"), |
| 243 | 1× | Ask::Connect { host: "storage.googleapis.com".into(), port: 443 } |
| 244 | | ); |
| 245 | 1× | assert!(matches!(parse_head("GET http://example.com/ HTTP/1.1\r\n\r\n"), Ask::Refused(w) if w.contains("only https"))); |
| 246 | 1× | assert!(matches!(parse_head("CONNECT example.com HTTP/1.1\r\n\r\n"), Ask::Refused(w) if w.contains("no port"))); |
| 247 | 1× | assert_eq!(decide(&allow(), "storage.googleapis.com", 443), Ok(())); |
| 248 | 1× | assert_eq!(decide(&allow(), "registry.opentofu.org", 443), Ok(())); |
| 249 | 1× | assert!(decide(&allow(), "example.com", 443).unwrap_err().contains("allow-list")); |
| 250 | 1× | assert!(decide(&allow(), "registry.opentofu.org", 22).unwrap_err().contains("only port 443")); |
| 251 | 1× | assert!(decide(&allow(), "142.250.1.1", 443).unwrap_err().contains("names hosts")); |
| 252 | 1× | } |
| 253 | |
|
| 254 | | #[test] |
| 255 | 1× | fn private_addresses_are_not_public() { |
| 256 | 11× | for ip in [ |
| 257 | 1× | "127.0.0.1", |
| 258 | 1× | "10.1.2.3", |
| 259 | 1× | "192.168.1.1", |
| 260 | 1× | "172.16.0.1", |
| 261 | 1× | "169.254.169.254", |
| 262 | 1× | "100.64.0.1", |
| 263 | 1× | "0.0.0.0", |
| 264 | 1× | "::1", |
| 265 | 1× | "fd00::1", |
| 266 | 1× | "fe80::1", |
| 267 | 1× | "::ffff:10.0.0.1", |
| 268 | 1× | ] { |
| 269 | 11× | assert!(!public(ip.parse().unwrap()), "{ip}"); |
| 270 | | } |
| 271 | 2× | for ip in ["142.250.1.1", "2a00:1450:4009::200e"] { |
| 272 | 2× | assert!(public(ip.parse().unwrap()), "{ip}"); |
| 273 | | } |
| 274 | 1× | } |
| 275 | |
|
| 276 | | #[test] |
| 277 | 1× | fn the_proxy_refuses_and_logs_a_host_not_allowed() { |
| 278 | 1× | let dir = tempfile::tempdir().unwrap(); |
| 279 | 1× | let sock = dir.path().join("v.sock_3128"); |
| 280 | 1× | let (tx, rx) = mpsc::channel(); |
| 281 | 1× | let proxy = start(&sock, allow(), tx).unwrap(); |
| 282 | 2× | let ask = |head: &str| { |
| 283 | 2× | let mut c = UnixStream::connect(&sock).unwrap(); |
| 284 | 2× | c.write_all(head.as_bytes()).unwrap(); |
| 285 | 2× | let mut answer = String::new(); |
| 286 | 2× | c.read_to_string(&mut answer).unwrap(); |
| 287 | 2× | answer |
| 288 | 2× | }; |
| 289 | 1× | let a = ask("CONNECT example.com:443 HTTP/1.1\r\n\r\n"); |
| 290 | 1× | assert!(a.starts_with("HTTP/1.1 403") && a.contains("not in the sandbox's egress allow-list"), "{a}"); |
| 291 | 1× | assert_eq!( |
| 292 | 1× | rx.recv_timeout(Duration::from_secs(5)).unwrap(), |
| 293 | | "egress: refused example.com:443: not in the sandbox's egress allow-list" |
| 294 | | ); |
| 295 | 1× | let a = ask("GET http://storage.googleapis.com/ HTTP/1.1\r\n\r\n"); |
| 296 | 1× | assert!(a.starts_with("HTTP/1.1 403"), "{a}"); |
| 297 | 1× | assert!(rx.recv_timeout(Duration::from_secs(5)).unwrap().contains("only https")); |
| 298 | 1× | drop(proxy); |
| 299 | 1× | assert!(!sock.exists(), "the socket goes with the job"); |
| 300 | 1× | } |
| 301 | | } |