Aleksandr Bogomiakov
2b0c938487
A long-running rc.33 daemon reached OOM with roughly 22 GiB resident and 24 GiB swapped. Live inspection found 40 GiB and 10/20 GiB anonymous container mappings; the active worker stack and tracing metadata identified iroh's scheduled path-open retry loop. In iroh 1.2.0 every failed connection requeues the same path, while each retry visits every connection again. Separate control and data connections multiply the queue geometrically. Vendor the published iroh 1.2.0 source and select it with a workspace Cargo patch. Deduplicate retry admission and cap distinct pending paths at 64, evicting the oldest before inserting a new candidate. Keep the current and next retry batches bounded without combining control/data connections or changing wire protocols, identity derivation, authentication or state. Document source provenance, the live evidence and removal conditions. Upstream PR 4551 merged on October 6, but inspection of the official 1.3.0 crate still shows the unconditional enqueue, so that upgrade alone is not a fix. No version bump, release tag or installed-service change. The default workspace suite includes the exact vendored admission module. Three regressions failed with the original enqueue and pass with the fix: repeated failures across connections, duplicate admission at capacity, and 10,000 distinct candidates with bounded length/capacity and FIFO eviction. Transport integration tests continue using real endpoints and encryption. Checks on Linux in nix develop: - cargo fmt --all -- --check - cargo clippy --locked --workspace --all-targets -- -D warnings - cargo test --locked --workspace --all-targets, with only the inherited TSUNAGI_CONTROL_SOCKET removed from the test environment - git diff --cached --check Initial full-suite runs failed six DNS readiness checks because the inherited system socket path overrode their temporary state directories. Clearing that environment variable resolves those failures without changing application code. Root/network-dependent ignored tests and other operating systems were not run.
95 lines
3.1 KiB
Rust
95 lines
3.1 KiB
Rust
//! Force iroh to negotiate ONLY a post-quantum key exchange (X25519MLKEM768).
|
|
//!
|
|
//! Requires `tls-aws-lc-rs` (the only rustls backend with ML-KEM today).
|
|
//! Stripping `kx_groups` to the PQ group makes PQ *required* rather than
|
|
//! *preferred*: peers without it fail the TLS handshake.
|
|
//!
|
|
//! Note: iroh's `crypto_provider` is shared with relay/discovery TLS, and
|
|
//! n0's infra does not support PQ key exchange yet — so a PQ-only endpoint
|
|
//! can't use n0 public relays or discovery servers.
|
|
//!
|
|
//! ## Usage
|
|
//!
|
|
//! `tls-aws-lc-rs` is required (the example explicitly constructs an
|
|
//! aws-lc-rs `CryptoProvider`, so the backend must be linked):
|
|
//!
|
|
//! cargo run --example pq-key-exchange --features=tls-aws-lc-rs
|
|
//!
|
|
//! With iroh's default features still on, both `ring` and `aws-lc-rs` get
|
|
//! linked. That's harmless — we wire the aws-lc-rs provider in directly via
|
|
//! `Builder::crypto_provider`.
|
|
use std::sync::Arc;
|
|
|
|
use iroh::{
|
|
RelayMode,
|
|
endpoint::{Endpoint, presets::Empty},
|
|
};
|
|
use n0_error::{Result, StdResultExt};
|
|
use rustls::crypto::aws_lc_rs;
|
|
|
|
const ALPN: &[u8] = b"iroh-example/pq-key-exchange/0";
|
|
|
|
#[tokio::main]
|
|
async fn main() -> Result<()> {
|
|
tracing_subscriber::fmt::init();
|
|
|
|
let pq = pq_only_provider();
|
|
|
|
let server = Endpoint::builder(Empty)
|
|
.crypto_provider(pq.clone())
|
|
.alpns(vec![ALPN.to_vec()])
|
|
.relay_mode(RelayMode::Disabled)
|
|
.bind()
|
|
.await?;
|
|
let server_addr = server.addr();
|
|
|
|
let server_task = tokio::spawn({
|
|
let server = server.clone();
|
|
async move {
|
|
let conn = server
|
|
.accept()
|
|
.await
|
|
.expect("incoming")
|
|
.accept()
|
|
.anyerr()?
|
|
.await
|
|
.anyerr()?;
|
|
let mut recv = conn.accept_uni().await.anyerr()?;
|
|
let msg = recv.read_to_end(1024).await.anyerr()?;
|
|
println!("server received: {:?}", String::from_utf8_lossy(&msg));
|
|
let mut send = conn.open_uni().await.anyerr()?;
|
|
send.write_all(b"pong over PQ").await.anyerr()?;
|
|
send.finish().anyerr()?;
|
|
conn.closed().await;
|
|
n0_error::Ok(())
|
|
}
|
|
});
|
|
|
|
let client = Endpoint::builder(Empty)
|
|
.crypto_provider(pq)
|
|
.relay_mode(RelayMode::Disabled)
|
|
.bind()
|
|
.await?;
|
|
let conn = client.connect(server_addr, ALPN).await.anyerr()?;
|
|
println!("client handshake done (X25519MLKEM768 was the only kx offered)");
|
|
|
|
let mut send = conn.open_uni().await.anyerr()?;
|
|
send.write_all(b"ping over PQ").await.anyerr()?;
|
|
send.finish().anyerr()?;
|
|
let mut recv = conn.accept_uni().await.anyerr()?;
|
|
let msg = recv.read_to_end(1024).await.anyerr()?;
|
|
println!("client received: {:?}", String::from_utf8_lossy(&msg));
|
|
|
|
conn.close(0u32.into(), b"done");
|
|
server_task.await.anyerr()?.anyerr()?;
|
|
client.close().await;
|
|
server.close().await;
|
|
Ok(())
|
|
}
|
|
|
|
fn pq_only_provider() -> Arc<rustls::crypto::CryptoProvider> {
|
|
let mut p = aws_lc_rs::default_provider();
|
|
p.kx_groups = vec![aws_lc_rs::kx_group::X25519MLKEM768];
|
|
Arc::new(p)
|
|
}
|