No description
Find a file
Daniel Barlow 77922c875b add per-device overlay
presently this is used to reference the appropriate kernel and (if
needed) openwrt source trees, but I would not be surprised if we find
other uses
2022-10-15 18:55:33 +01:00
devices add per-device overlay 2022-10-15 18:55:33 +01:00
kernel add per-device overlay 2022-10-15 18:55:33 +01:00
modules kernel: enlarge offset to rootfs, disable sysfs devlink 2022-10-15 16:11:40 +01:00
pkgs add build-host tftp server 2022-10-04 23:08:43 +01:00
scripts add QEMU_OPTIONS env var for run-qemu 2022-10-02 10:05:28 +01:00
tests add per-device overlay 2022-10-15 18:55:33 +01:00
.gitignore tidy repo 2022-09-22 11:11:31 +01:00
CONTRIBUTING.md add note about contributions 2022-10-01 09:26:15 +01:00
default.nix add per-device overlay 2022-10-15 18:55:33 +01:00
LICENSE add Licence information 2022-10-01 12:13:47 +01:00
merge-modules.nix merge-modules accepts >1 module 2022-09-25 11:52:02 +01:00
overlay.nix make dnsmasq smaller 2022-10-08 20:05:17 +01:00
README.md add build-host tftp server 2022-10-04 23:08:43 +01:00
run-tests.sh run-tests cd into each test directory before running it 2022-09-23 17:21:43 +01:00
STYLE.md explain package/module distinction, add notes on side tracks 2022-09-27 14:11:23 +01:00
THOUGHTS.txt openwrt ag71xx as module 2022-10-15 16:28:19 +01:00

Liminix

A Nix-based system for configuring consumer wifi routers.

What is this?

This is a Nix-based collection of software tailored for domestic wifi router or IoT device devices, of the kind that OpenWrt or DD-WRT or Gargoyle or Tomato run on. It's a reboot/restart/rewrite of NixWRT.

This is not NixOS-on-your-router: it's aimed at devices that are underpowered for the full NixOS experience. It uses busybox tools, musl instead of GNU libc, and s6-rc instead of systemd.

The Liminix name comes from Liminis, in Latin the genitive declension of "limen", or "of the threshold". Your router stands at the threshold of your (online) home and everything you send to/receive from the outside word goes across it.

What about NixWRT?

This is an in-progress rewrite of NixWRT, incorporating Lessons Learned. That said, as of today (September 2022) it is not yet anywhere near feature parity.

Liminix will eventually provide these differentiators over NixWRT:

  • a writable filesystem so that software updates or reconfiguration (e.g. changing passwords) don't require taking the device offline to reflash it.

  • more flexible service management with dependencies, to allow configurations such as "route through PPPoE if it is healthy, with fallback to LTE"

  • a spec for valid configuration options (a la NixOS module options) to that we can detect errors at evaluation time instead of producing a bad image.

  • a network-based mechanism for secrets management so that changes can be pushed from a central location to several Liminix devices at once

  • send device metrics and logs to a monitoring/alerting/o11y infrastructure

Today though, it does approximately none of these things and certainly not on real hardware.

Building

These instructions assume you have nixpkgs checked out in a peer directory of this one.

You need a configuration.nix file pointed to by <liminix-config>, a hardware device definition as argument device, and to choose an appropriate output attribute depending on what your device is and how you plan to install onto it. For example:

NIX_PATH=nixpkgs=../nixpkgs:$NIX_PATH NIXPKGS_ALLOW_UNSUPPORTED_SYSTEM=1 nix-build -I liminix-config=./tests/smoke/configuration.nix --arg device "import ./devices/qemu.nix" -A outputs.default

outputs.default is intended to do something appropriate for the device, whatever that is. For the qemu device, it creates a directory containing a squashfs root image and a kernel.

QEMU

QEMU is useful for developing userland without needing to keep flashing or messing with U-Boot: it also enables testing against emulated network peers using QEMU socket networking, which may be preferable to letting Liminix loose on your actual LAN.

We have some tooling to make this easier.

Networks

We observe these conventions for QEMU network sockets, so that we can run multiple emulated instances and have them wired up to each other in the right way

  • multicast 230.0.0.1:1234 : access (interconnect between router and "isp")
  • multicast 230.0.0.1:1235 : lan
  • multicast 230.0.0.1:1236 : world (the internet)

Running instances

./scripts/run-qemu.sh accepts a kernel vmlinux image and a squashfs and runs qemu with appropriate config for two ethernet interfaces hooked up to "lan" and "access" respectively. It connects the Liminix serial console and the QEMU monitor to stdin/stdout. Use ^P (not ^A) to switch to the monitor.

If you run with --background /path/to/unix/socket it will fork into the background and open a Unix socket at that pathname to communicate on. Use ./scripts/connect-qemu.sh to connect to it, and ^O to disconnect.

Emulated upstream connection

In the tests/support/ppp-server directory there are instructions and a script to configure Mikrotik RouterOS as a PPPoE access concentrator connected to the access and world networks, so that Liminix PPPoE client support can be tested. Liminix does not provide RouterOS licences and it is your own responsibility if you use this to ensure you're compliant with the terms of Mikrotik's licencing.

This may be supplemented or replaced in time with configuurations for RP-PPPoE and/or Accel PPP.

Running tests

Assuming you have nixpkgs checked out in a peer directory of this one,

NIX_PATH=nixpkgs=../nixpkgs:$NIX_PATH ./run-tests.sh

Some of the tests require the emulated upstream connection to be running.

Hardware

How you get the thing onto hardware will vary according to the device, but is likely to involve U-Boot and TFTP.

There is a rudimentary TFTP server bundled with the system which runs from the command line, has an allowlist for client connections, and follows symlinks, so you can have your device download images direct from the ./result directory without exposing /nix/store/ to the internet or mucking about copying files to /tftproot. If the permitted device is to be given the IP address 192.168.8.251 you might do something like this:

$ NIX_PATH=nixpkgs=../nixpkgs:$NIX_PATH NIXPKGS_ALLOW_UNSUPPORTED_SYSTEM=1 nix-build -I liminix-config=./tests/smoke/configuration.nix --arg device "import ./devices/gl-ar750.nix" -A outputs.tftpd -o tftpd
$  ./tftpd/bin/tufted -a 192.168.8.251 result

Troubleshooting

Diagnosing unexpectedly large images

Sometimes you can add a package and it causes the image size to balloon because it has dependencies on other things you didn't know about. Build the outputs.manifest attribute, which is a json representation of the filesystem, and you can run nix-store --query on it:

NIX_PATH=nixpkgs=../nixpkgs:$NIX_PATH NIXPKGS_ALLOW_UNSUPPORTED_SYSTEM=1 nix-build -I liminix-config=path/to/your/configuration.nix --arg device "import ./devices/qemu.nix" -A outputs.manifest -o manifest
nix-store -q --tree manifest

Contributing

Contributions are welcome, though in these early days there may be a bit of back and forth involved before patches are merged. Have a read of CONTRIBUTING and STYLE and try to intuit the unarticulated vision :-)

Liminix' primary repo is https://gti.telent.net/dan/liminix. There's a mirror on Github for convenience and visibility: you can open PRs against that but be aware that the process of merging them may be arcane. Some day, we will have federated Gitea using ActivityPub.

Articles of interest

  • Build Safety of Software in 28 Popular Home Routers: "of the access points and routers we reviewed, not a single one took full advantage of the basic application armoring features provided by the operating system. Indeed, only one or two models even came close, and no brand did well consistently across all models tested"

  • A PPPoE Implementation for Linux: "Many DSL service providers use PPPoE for residential broadband Internet access. This paper briefly describes the PPPoE protocol, presents strategies for implementing it under Linux and describes in detail a user-space implementation of a PPPoE client."

  • PPP IPV6CP vs DHCPv6 at AAISP