Security hardening, concurrency fixes, and expanded test coverage

This commit introduces a comprehensive set of improvements to enhance
the security, reliability, and configurability of the proxy server,
specifically targeting adversarial resilience and high-load concurrency.

Security & Cryptography:
- Zeroize MTProto cryptographic key material (`dec_key`, `enc_key`)
  immediately after use to prevent memory leakage on early returns.
- Move TLS handshake replay tracking after full policy/ALPN validation
  to prevent cache poisoning by unauthenticated probes.
- Add `proxy_protocol_trusted_cidrs` configuration to restrict PROXY
  protocol headers to trusted networks, rejecting spoofed IPs.

Adversarial Resilience & DoS Mitigation:
- Implement "Tiny Frame Debt" tracking in the middle-relay to prevent
  CPU exhaustion from malicious 0-byte or 1-byte frame floods.
- Add `mask_relay_max_bytes` to strictly bound unauthenticated fallback
  connections, preventing the proxy from being abused as an open relay.
- Add a 5ms prefetch window (`mask_classifier_prefetch_timeout_ms`) to
  correctly assemble and classify fragmented HTTP/1.1 and HTTP/2 probes
  (e.g., `PRI * HTTP/2.0`) before routing them to masking heuristics.
- Prevent recursive masking loops (FD exhaustion) by verifying the mask
  target is not the proxy's own listener via local interface enumeration.

Concurrency & Reliability:
- Eliminate executor waker storms during quota lock contention by replacing
  the spin-waker task with inline `Sleep` and exponential backoff.
- Roll back user quota reservations (`rollback_me2c_quota_reservation`)
  if a network write fails, preventing Head-of-Line (HoL) blocking from
  permanently burning data quotas.
- Recover gracefully from idle-registry `Mutex` poisoning instead of
  panicking, ensuring isolated thread failures do not break the proxy.
- Fix `auth_probe_scan_start_offset` modulo logic to ensure bounds safety.

Testing:
- Add extensive adversarial, timing, fuzzing, and invariant test suites
  for both the client and handshake modules.
This commit is contained in:
David Osipov
2026-03-22 23:06:26 +04:00
parent 6fc188f0c4
commit 91be148b72
65 changed files with 7473 additions and 210 deletions

View File

@@ -62,7 +62,7 @@ use std::sync::{Arc, Mutex, OnceLock};
use std::task::{Context, Poll};
use std::time::Duration;
use tokio::io::{AsyncRead, AsyncWrite, AsyncWriteExt, ReadBuf, copy_bidirectional_with_sizes};
use tokio::time::Instant;
use tokio::time::{Instant, Sleep};
use tracing::{debug, trace, warn};
// ============= Constants =============
@@ -209,12 +209,16 @@ struct StatsIo<S> {
counters: Arc<SharedCounters>,
stats: Arc<Stats>,
user: String,
quota_lock: Option<Arc<Mutex<()>>>,
cross_mode_quota_lock: Option<Arc<Mutex<()>>>,
quota_limit: Option<u64>,
quota_exceeded: Arc<AtomicBool>,
quota_read_wake_scheduled: bool,
quota_write_wake_scheduled: bool,
quota_read_retry_active: Arc<AtomicBool>,
quota_write_retry_active: Arc<AtomicBool>,
quota_read_retry_sleep: Option<Pin<Box<Sleep>>>,
quota_write_retry_sleep: Option<Pin<Box<Sleep>>>,
quota_read_retry_attempt: u8,
quota_write_retry_attempt: u8,
epoch: Instant,
}
@@ -230,30 +234,29 @@ impl<S> StatsIo<S> {
) -> Self {
// Mark initial activity so the watchdog doesn't fire before data flows
counters.touch(Instant::now(), epoch);
let quota_lock = quota_limit.map(|_| quota_user_lock(&user));
let cross_mode_quota_lock = quota_limit
.map(|_| crate::proxy::quota_lock_registry::cross_mode_quota_user_lock(&user));
Self {
inner,
counters,
stats,
user,
quota_lock,
cross_mode_quota_lock,
quota_limit,
quota_exceeded,
quota_read_wake_scheduled: false,
quota_write_wake_scheduled: false,
quota_read_retry_active: Arc::new(AtomicBool::new(false)),
quota_write_retry_active: Arc::new(AtomicBool::new(false)),
quota_read_retry_sleep: None,
quota_write_retry_sleep: None,
quota_read_retry_attempt: 0,
quota_write_retry_attempt: 0,
epoch,
}
}
}
impl<S> Drop for StatsIo<S> {
fn drop(&mut self) {
self.quota_read_retry_active.store(false, Ordering::Relaxed);
self.quota_write_retry_active
.store(false, Ordering::Relaxed);
}
}
#[derive(Debug)]
struct QuotaIoSentinel;
@@ -281,20 +284,52 @@ fn is_quota_io_error(err: &io::Error) -> bool {
const QUOTA_CONTENTION_RETRY_INTERVAL: Duration = Duration::from_millis(1);
#[cfg(not(test))]
const QUOTA_CONTENTION_RETRY_INTERVAL: Duration = Duration::from_millis(2);
#[cfg(test)]
const QUOTA_CONTENTION_RETRY_MAX_INTERVAL: Duration = Duration::from_millis(16);
#[cfg(not(test))]
const QUOTA_CONTENTION_RETRY_MAX_INTERVAL: Duration = Duration::from_millis(64);
fn spawn_quota_retry_waker(retry_active: Arc<AtomicBool>, waker: std::task::Waker) {
tokio::task::spawn(async move {
loop {
if !retry_active.load(Ordering::Relaxed) {
break;
}
tokio::time::sleep(QUOTA_CONTENTION_RETRY_INTERVAL).await;
if !retry_active.load(Ordering::Relaxed) {
break;
}
waker.wake_by_ref();
}
});
#[inline]
fn quota_contention_retry_delay(retry_attempt: u8) -> Duration {
let shift = u32::from(retry_attempt.min(5));
let multiplier = 1_u32 << shift;
QUOTA_CONTENTION_RETRY_INTERVAL
.saturating_mul(multiplier)
.min(QUOTA_CONTENTION_RETRY_MAX_INTERVAL)
}
#[inline]
fn reset_quota_retry_scheduler(
sleep_slot: &mut Option<Pin<Box<Sleep>>>,
wake_scheduled: &mut bool,
retry_attempt: &mut u8,
) {
*wake_scheduled = false;
*sleep_slot = None;
*retry_attempt = 0;
}
fn poll_quota_retry_sleep(
sleep_slot: &mut Option<Pin<Box<Sleep>>>,
wake_scheduled: &mut bool,
retry_attempt: &mut u8,
cx: &mut Context<'_>,
) {
if !*wake_scheduled {
*wake_scheduled = true;
*sleep_slot = Some(Box::pin(tokio::time::sleep(quota_contention_retry_delay(
*retry_attempt,
))));
}
if let Some(sleep) = sleep_slot.as_mut()
&& sleep.as_mut().poll(cx).is_ready()
{
*sleep_slot = None;
*wake_scheduled = false;
*retry_attempt = retry_attempt.saturating_add(1);
cx.waker().wake_by_ref();
}
}
static QUOTA_USER_LOCKS: OnceLock<DashMap<String, Arc<Mutex<()>>>> = OnceLock::new();
@@ -357,6 +392,11 @@ fn quota_user_lock(user: &str) -> Arc<Mutex<()>> {
}
}
#[cfg(test)]
pub(crate) fn cross_mode_quota_user_lock_for_tests(user: &str) -> Arc<Mutex<()>> {
crate::proxy::quota_lock_registry::cross_mode_quota_user_lock(user)
}
impl<S: AsyncRead + Unpin> AsyncRead for StatsIo<S> {
fn poll_read(
self: Pin<&mut Self>,
@@ -368,26 +408,47 @@ impl<S: AsyncRead + Unpin> AsyncRead for StatsIo<S> {
return Poll::Ready(Err(quota_io_error()));
}
let quota_lock = this
.quota_limit
.is_some()
.then(|| quota_user_lock(&this.user));
let _quota_guard = if let Some(lock) = quota_lock.as_ref() {
let _quota_guard = if let Some(lock) = this.quota_lock.as_ref() {
match lock.try_lock() {
Ok(guard) => {
this.quota_read_wake_scheduled = false;
this.quota_read_retry_active.store(false, Ordering::Relaxed);
reset_quota_retry_scheduler(
&mut this.quota_read_retry_sleep,
&mut this.quota_read_wake_scheduled,
&mut this.quota_read_retry_attempt,
);
Some(guard)
}
Err(_) => {
if !this.quota_read_wake_scheduled {
this.quota_read_wake_scheduled = true;
this.quota_read_retry_active.store(true, Ordering::Relaxed);
spawn_quota_retry_waker(
Arc::clone(&this.quota_read_retry_active),
cx.waker().clone(),
);
}
poll_quota_retry_sleep(
&mut this.quota_read_retry_sleep,
&mut this.quota_read_wake_scheduled,
&mut this.quota_read_retry_attempt,
cx,
);
return Poll::Pending;
}
}
} else {
None
};
let _cross_mode_quota_guard = if let Some(lock) = this.cross_mode_quota_lock.as_ref() {
match lock.try_lock() {
Ok(guard) => {
reset_quota_retry_scheduler(
&mut this.quota_read_retry_sleep,
&mut this.quota_read_wake_scheduled,
&mut this.quota_read_retry_attempt,
);
Some(guard)
}
Err(_) => {
poll_quota_retry_sleep(
&mut this.quota_read_retry_sleep,
&mut this.quota_read_wake_scheduled,
&mut this.quota_read_retry_attempt,
cx,
);
return Poll::Pending;
}
}
@@ -460,27 +521,47 @@ impl<S: AsyncWrite + Unpin> AsyncWrite for StatsIo<S> {
return Poll::Ready(Err(quota_io_error()));
}
let quota_lock = this
.quota_limit
.is_some()
.then(|| quota_user_lock(&this.user));
let _quota_guard = if let Some(lock) = quota_lock.as_ref() {
let _quota_guard = if let Some(lock) = this.quota_lock.as_ref() {
match lock.try_lock() {
Ok(guard) => {
this.quota_write_wake_scheduled = false;
this.quota_write_retry_active
.store(false, Ordering::Relaxed);
reset_quota_retry_scheduler(
&mut this.quota_write_retry_sleep,
&mut this.quota_write_wake_scheduled,
&mut this.quota_write_retry_attempt,
);
Some(guard)
}
Err(_) => {
if !this.quota_write_wake_scheduled {
this.quota_write_wake_scheduled = true;
this.quota_write_retry_active.store(true, Ordering::Relaxed);
spawn_quota_retry_waker(
Arc::clone(&this.quota_write_retry_active),
cx.waker().clone(),
);
}
poll_quota_retry_sleep(
&mut this.quota_write_retry_sleep,
&mut this.quota_write_wake_scheduled,
&mut this.quota_write_retry_attempt,
cx,
);
return Poll::Pending;
}
}
} else {
None
};
let _cross_mode_quota_guard = if let Some(lock) = this.cross_mode_quota_lock.as_ref() {
match lock.try_lock() {
Ok(guard) => {
reset_quota_retry_scheduler(
&mut this.quota_write_retry_sleep,
&mut this.quota_write_wake_scheduled,
&mut this.quota_write_retry_attempt,
);
Some(guard)
}
Err(_) => {
poll_quota_retry_sleep(
&mut this.quota_write_retry_sleep,
&mut this.quota_write_wake_scheduled,
&mut this.quota_write_retry_attempt,
cx,
);
return Poll::Pending;
}
}
@@ -791,3 +872,27 @@ mod relay_quota_waker_storm_adversarial_tests;
#[cfg(test)]
#[path = "tests/relay_quota_wake_liveness_regression_tests.rs"]
mod relay_quota_wake_liveness_regression_tests;
#[cfg(test)]
#[path = "tests/relay_quota_lock_identity_security_tests.rs"]
mod relay_quota_lock_identity_security_tests;
#[cfg(test)]
#[path = "tests/relay_cross_mode_quota_lock_security_tests.rs"]
mod relay_cross_mode_quota_lock_security_tests;
#[cfg(test)]
#[path = "tests/relay_quota_retry_scheduler_tdd_tests.rs"]
mod relay_quota_retry_scheduler_tdd_tests;
#[cfg(test)]
#[path = "tests/relay_cross_mode_quota_fairness_tdd_tests.rs"]
mod relay_cross_mode_quota_fairness_tdd_tests;
#[cfg(test)]
#[path = "tests/relay_quota_retry_backoff_security_tests.rs"]
mod relay_quota_retry_backoff_security_tests;
#[cfg(test)]
#[path = "tests/relay_quota_retry_backoff_benchmark_security_tests.rs"]
mod relay_quota_retry_backoff_benchmark_security_tests;