slaac: add initial stateless address autoconfiguration (SLAAC) implementation #1039

bergzand · 2025-02-05T17:10:11Z

This PR adds

Functions to derive ipv6 addresses from a Cidr and a LL address
An slaac implementation
Glue to hook it into the Interface

This works by having a separate module to track the slaac state machine. The Interface module passes the received router advertisements to this module and polls this to query whether router solicitations must be send and whether the addresses and routes on the interface need to be synced with the slaac state.

Compared to the dhcpv4, this module modifies the address and route state on the Interface within the module itself without any outside glue or socket required.

Currently there's no separate feature for slaac and it is pulled in when proto-ipv6 together with medium-ieee802154 or medium-ethernet is enabled. Done this way because I consider SLAAC to be an essential part of IPv6. However, I'm not opposed to adding a separate feature for slaac that depends on proto-ipv6

An example is provided that prints the addresses and routes on the interface every second for testing with router advertisement daemons such as radvd

Example output

Addresses:
  - fe80::1/64
Routes:

Addresses:
  - fe80::1/64
Routes:

Addresses:
  - fe80::1/64
  - 2001:db8::ff:fe00:1/64
Routes:
  - ::/0 via fe80::d83c:a5ff:fe9c:3a40

Addresses:
  - fe80::1/64
  - 2001:db8::ff:fe00:1/64
Routes:
  - ::/0 via fe80::d83c:a5ff:fe9c:3a40

Addresses:
  - fe80::1/64
  - 2001:db8::ff:fe00:1/64
Routes:
  - ::/0 via fe80::d83c:a5ff:fe9c:3a40

Addresses:
  - fe80::1/64
  - 2001:db8::ff:fe00:1/64
Routes:

Addresses:
  - fe80::1/64
  - 2001:db8::ff:fe00:1/64
Routes:

Alternative to #948

I can split out the initial commits of this PR into separate PRs if needed to ease reviewing.

This commit splits the state maintenance from the poll function to a separate poll_maintenance function.

codecov · 2025-02-05T17:12:44Z

Codecov Report

Attention: Patch coverage is 95.41284% with 25 lines in your changes missing coverage. Please review.

Project coverage is 81.70%. Comparing base (3e3afb6) to head (8786ec8).
Report is 14 commits behind head on main.

Files with missing lines	Patch %	Lines
src/iface/interface/mod.rs	71.11%	13 Missing ⚠️
src/iface/slaac.rs	97.99%	6 Missing ⚠️
src/iface/interface/ipv6.rs	95.90%	5 Missing ⚠️
src/wire/ipv6.rs	98.18%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1039      +/-   ##
==========================================
+ Coverage   80.84%   81.70%   +0.85%     
==========================================
  Files          81       82       +1     
  Lines       28485    29501    +1016     
==========================================
+ Hits        23029    24103    +1074     
+ Misses       5456     5398      -58

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

Dominaezzz

Looks good!

My main comment is here is "Why limit SLAAC to any specific mediums?"

Dominaezzz · 2025-03-03T13:05:36Z

src/iface/mod.rs

@@ -23,4 +28,9 @@ pub use self::interface::{
 };

 pub use self::route::{Route, RouteTableFull, Routes};
+#[cfg(all(
+    feature = "proto-ipv6",
+    any(feature = "medium-ethernet", feature = "medium-ieee802154")


SLAAC is a general IPv6 thing, why restrict it to any particular medium(s)?

SLAAC generates an IP addr based on the MAC addr. With IP medium there's no MAC addr. How is it supposed to work?

as @Dirbaio mentioned, there needs to be support for each link layer with SLAAC, see for example rfc 2464 and rfc 4944.

Sure for link-local address you want to use the MAC address (or equivalent) to generate it, but that's not quite the same for global or temporary addresses where you might want to hide your MAC address. I think in this case the user should be allowed to specify an interface identifier (which would be different from the MAC).

If Linux can do SLAAC for TUN devices, smoltcp can definitely do it for IP mediums.

EDIT: I'm looking at https://datatracker.ietf.org/doc/html/rfc4862#section-5.5

Sure for link-local address you want to use the MAC address (or equivalent) to generate it, but that's not quite the same for global or temporary addresses where you might want to hide your MAC address. I think in this case the user should be allowed to specify an interface identifier (which would be different from the MAC).

I prefer to leave the additional logic required to switch the generation of interface identifiers to a follow up PR as this PR is already complex enough.

If Linux can do SLAAC for TUN devices, smoltcp can definitely do it for IP mediums.

If there's a way to add an interface identifier to a Medium::Ip Interface, it can be hooked in somehow. The whole process_ndisc where this code hooks into is skipped on Medium::Ip interfaces, so that would need to be modified. Not impossible, but also not sure if that must go in this PR.

Do you know if there's some special handling for SLAAC on Linux tun devices? Or does it always add a privacy address to that interface for slaac?

Or does it always add a privacy address to that interface for slaac?

Yes

I prefer to leave the additional logic required to switch the generation of interface identifiers to a follow up PR as this PR is already complex enough.

Sure the additional logic isn't necessary for this PR, I'm just saying the any(feature = "medium-ethernet", feature = "medium-ieee802154") isn't needed.

Dominaezzz · 2025-03-03T13:09:24Z

src/wire/ndiscoption.rs

+impl PrefixInformation {
+    /// Validates the prefix information option against check a, b, c in
+    /// https://www.rfc-editor.org/rfc/rfc4862#section-5.5.3
+    pub fn valid_prefix_info(&self) -> bool {


This probably reads better?

Suggested change

pub fn valid_prefix_info(&self) -> bool {

pub fn is_valid_prefix_info(&self) -> bool {

interface: Split state maintenance from poll function

c069d40

This commit splits the state maintenance from the poll function to a separate poll_maintenance function.

bergzand added 5 commits February 5, 2025 19:18

IPv6: Add IPv6 autogenerated address functions

29f5db2

ndisc: add router advertisement validity check

4627a5c

slaac: add initial SLAAC implementation

74bd6d2

interface: Hook up SLAAC to interface

a4deb31

examples/slaac: add SLAAC example

3973877

bergzand force-pushed the pr/slaac branch from a651c9a to 3973877 Compare February 5, 2025 18:18

Dominaezzz reviewed Mar 3, 2025

View reviewed changes

bergzand added 2 commits March 3, 2025 17:28

fixup! slaac: add initial SLAAC implementation

6eb2406

fixup! ndisc: add router advertisement validity check

8786ec8

emmanuelsearch mentioned this pull request Mar 12, 2025

feat: ipv6 support ariel-os/ariel-os#724

Open

2 tasks

Dirbaio mentioned this pull request Mar 21, 2025

embassy_net: Add neighbors to cache manually. embassy-rs/embassy#3988

Open

chrysn added this to Delta Mar 26, 2025

chrysn moved this to Delta Suggestions in Delta Mar 26, 2025

chrysn mentioned this pull request Mar 26, 2025

SLAAC support #570

Open

kaspar030 moved this from Delta Suggestions to Todo Next in Delta Apr 8, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

slaac: add initial stateless address autoconfiguration (SLAAC) implementation #1039

slaac: add initial stateless address autoconfiguration (SLAAC) implementation #1039

bergzand commented Feb 5, 2025

codecov bot commented Feb 5, 2025 •

edited

Loading

Dominaezzz left a comment

Dominaezzz Mar 3, 2025

Dirbaio Mar 3, 2025

bergzand Mar 3, 2025

Dominaezzz Mar 3, 2025 •

edited

Loading

bergzand Mar 3, 2025

Dominaezzz Mar 3, 2025

Dominaezzz Mar 3, 2025

	pub fn valid_prefix_info(&self) -> bool {
	pub fn is_valid_prefix_info(&self) -> bool {

slaac: add initial stateless address autoconfiguration (SLAAC) implementation #1039

Are you sure you want to change the base?

slaac: add initial stateless address autoconfiguration (SLAAC) implementation #1039

Conversation

bergzand commented Feb 5, 2025

codecov bot commented Feb 5, 2025 • edited Loading

Codecov Report

Dominaezzz left a comment

Choose a reason for hiding this comment

Dominaezzz Mar 3, 2025

Choose a reason for hiding this comment

Dirbaio Mar 3, 2025

Choose a reason for hiding this comment

bergzand Mar 3, 2025

Choose a reason for hiding this comment

Dominaezzz Mar 3, 2025 • edited Loading

Choose a reason for hiding this comment

bergzand Mar 3, 2025

Choose a reason for hiding this comment

Dominaezzz Mar 3, 2025

Choose a reason for hiding this comment

Dominaezzz Mar 3, 2025

Choose a reason for hiding this comment

codecov bot commented Feb 5, 2025 •

edited

Loading

Dominaezzz Mar 3, 2025 •

edited

Loading