ringroute

A minimal traceroute built on io_uring.

Goal

ringroute exists to demonstrate io_uring usage. At this time, it is not meant to be a replacement for traceroute, mtr, or any other production network diagnostic tool. Not only is there no guarantee of feature parity, but the use of modern kernel features limits which environments this can be run in.

How it works

At its core, ringroute works like any other traceroute tool. Each router that forwards an IP packet decrements its time-to-live (TTL) by one. If decrementing the TTL would reduce it to zero, the router discards the packet and will typically send an ICMP Time Exceeded message back to the sender. These responses reveal the router's own IP address. To discover the path to a target, ringroute sends probes with increasing TTL values. A probe sent with TTL N expires at the Nth hop, allowing us to identify that device from its ICMP response.

Usage

To build from source:

git clone git@github.com:pchaseh/ringroute.git
cd ringroute
cargo build --release

Run with --help to see usage:

./ringroute --help

ringroute uses env_logger for application logging. The RUST_LOG environment variable can be used to set the log level:

RUST_LOG=debug ./ringroute . . .

When not set, RUST_LOG defaults to error.

Requirements

• Linux 6.0 or newer compiled with CONFIG_IO_URING=y
• Rust toolchain (including Cargo)

Limitations

• Limited support for different probing methods.
• All hops are probed at once.
• Options to influence routing (eg. specifying source address, network device) are not implemented.