You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently boringtun leaks file descriptors. If we add this to src/device/integration_tests/mod.rs:
#[test]
/// Test if wireguard leaks resources on closing
fn test_fd_leaks() {
let n_before = count_file_descriptors_currently_in_use();
let wg = WGHandle::init("192.0.2.0".parse().unwrap(), "::2".parse().unwrap());
let response = wg.wg_get();
assert!(response.ends_with("errno=0\n\n"));
drop(wg); // call destructor
let n_before = count_file_descriptors_currently_in_use();
assert_eq!(n_before, n_after);
}
This test will fail.
Comments:
count_file_descriptors_currently_in_use() would count only file descriptors in use by the current process.
The implementation for the count_file_descriptors_currently_in_use() method is platform-specific but a naive implementation for Linux would be
pub fn count_file_descriptors_currently_in_use() -> u16 {
use std::process;
use std::process::Command;
let path = format!("/proc/{}/fd/", process::id());
let output = Command::new("ls")
.args(&["-l", &path])
.output()
.expect("failed to get fd info");
let stdout = String::from_utf8_lossy(&output.stdout);
let mut n: u16 = 0;
for x in stdout.lines() {
n += 1;
}
n
}
The text was updated successfully, but these errors were encountered:
Having investigated this a bit further the main leak appears to be in epoll/kqueue where file descriptors in use by the queue aren't freed when the queue is dropped.
Currently boringtun leaks file descriptors. If we add this to src/device/integration_tests/mod.rs:
This test will fail.
Comments:
count_file_descriptors_currently_in_use()
would count only file descriptors in use by the current process.count_file_descriptors_currently_in_use()
method is platform-specific but a naive implementation for Linux would beThe text was updated successfully, but these errors were encountered: