Feature/structured logging msgpack - #23
Merged
Merged
Conversation
Replace the custom verbose! macro with tracing spans that provide
hierarchical context per session and tunnel. Log lines now show which
session and tunnel they belong to using the remote's Display format
(e.g. session{id=1}: tunnel{remote=3000=>127.0.0.1:8080/tcp}).
Also adds version bump check to CI and simplifies --verbose/--debug
log level mapping.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Split tests across focused files (combinations, concurrent, edge_cases, large_transfer) sharing helpers in tests/common/mod.rs, and refactor tunnels.rs onto the same helpers. The new suites deliberately leave 9 tests failing to surface real bugs: the remote_start handshake race that silently truncates TCP payloads written immediately after connect, the UDP tunnel's lack of per-datagram framing, and the UDP forward client latching onto the first sender's source address. Made-with: Cursor
QUIC streams are byte-oriented, so the previous code's single
recv.read(&mut [u8; 1024]) calls in server_receive_remote_request and
server_receive_remote_start were unsafe: any client write that landed
in the same QUIC frame as the next message would be silently consumed
along with the discarded buffer. The remote_start step in particular
caused exactly 1012 bytes of payload loss on every TCP connection
because the client wrote payload immediately after the sentinel.
Replace the raw reads with a u32 little-endian length prefix and drop
the remote_start round trip entirely (also saves one RTT per TCP
connection). The wire format changes; client and server must be
deployed together.
Fixes the 7 failing TCP integration tests in tests/{large_transfer,
concurrent,edge_cases}.rs. The 2 remaining UDP-related failures are
tracked separately (issue #18 sections 3 and 5).
Made-with: Cursor
The UDP tunnel previously copied raw bytes onto a QUIC byte-stream and read them back with a fixed 1024-byte buffer, with no per-datagram framing. Two consecutive datagrams could be coalesced into a single far-side read and re-emitted as one oversized packet (silently corrupting protocols like DNS, WireGuard, RTP), and any datagram larger than 1024 bytes was truncated. Frame each datagram on the wire as a u16 little-endian length plus payload, and bump the buffer to 65535 bytes so real-world UDP payloads (up to the IPv4 cap of 65507) round-trip intact. Hoist the buffer out of the hot loop while we're here, and log dropped unexpected-source datagrams instead of silently discarding them. Fixes large_transfer::test_udp_forward_many_packets. The remaining multi-source UDP bug (issue #18 §5) is tracked separately. Made-with: Cursor
The forward UDP tunnel previously latched onto the first sender's SocketAddr in tunnel_udp_client and silently dropped every packet from any other source — see issue #18 §5. Multi-client UDP scenarios (e.g. several apps using the tunnel as a DNS forwarder) were broken with no log to diagnose. Replace the single-stream design with a per-source session: each new SocketAddr observed on the local UDP socket gets its own QUIC bi-stream and its own RemoteRequest, fed by a bounded mpsc channel from the recv loop. Sessions self-terminate after 60 s of inactivity to bound memory under churn. Backpressure drops are logged at debug level instead of silent. Fixes concurrent::test_udp_forward_multiple_senders. Full test suite (30 tests across 5 files) is now green. Made-with: Cursor
Three small correctness wins surfaced by issues #20: - tunnel_socks_client previously did `tokio::spawn(...).await?` per accepted connection, which serialized the entire SOCKS5 accept loop on each handshake. Drop the outer .await? and merge the two nested spawns into one fire-and-forget task so handshake + tunnel both run concurrently with future accepts (#20 §1). - tunnel_tcp_stream and tunnel_udp_stream now use tokio::join! instead of try_join!, so an error in one direction does not cancel the still in-flight copy in the other and lose buffered application data (#20 §3). Errors are still logged at debug. - Drop the stale `// TODO - support udp over socks5 :)` comment in socks.rs (#22 §2). Made-with: Cursor
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.