Re: Series of IPv6 Improvements
Marek Zarychta <[email protected]>
| Newsgroups | gmane.os.freebsd.devel.net |
|---|---|
| Message-ID | <[email protected]> |
On 23.02.2026 at 10:12, Pouria Mousavizadeh Tehrani wrote: > Hi, > > I've recently completed a series of changes focused on enhancing our > IPv6 stack, particularly around several important RFCs related to > Neighbor Discovery Protocol (NDP). These changes not only resolve > long-standing issues but also introduce new features that align with > modern networking standards. > > Here's an overview of what I've done: > - Implemented Gratuitous Neighbor Discovery (RFC 9131 + RFC 4861 > Section 7.2.6, which is also addresses one of the IPv6 deployment > issues in RFC 9898 Section 3.9.) > - Implemented Optimistic DAD (RFC 4429) > - Added support for delayed anycast and proxy addresses during > Neighbor Advertisement (RFC4861 7.2.7, + 7.2.8) > - Introduced support for multiple PREF64 in Router Advertisement (RFC > 8781) > - Implemented support for Route Information option (RFC 4191) > - Simplified `nd6_ra_input()` > > While implementing the RFC 4191, I found out some of the route > expiration facilities are broken. So I fixed it myself: > - rtnetlink(4): Add support for nexthop expiration in new route > - route(8): Fixed the -expire argument when using netlink > - netstat(1): Fixed expire column values in -r flag when using netlink > - snl(3): Remove duplicate `rta_expires` member in the route parser > > These improvements should enhance our IPv6 capabilities significantly > and resolves many /* XXX */ comments + some long-standing issues like > lack of route option support (see PR263982). > > You can find those reviews here: > https://reviews.freebsd.org/search/query/FpCNDQh8WgIb > > I'd appreciate any review or testing these proposed changes on > Phabricator. > Hi Pouria, That’s really good news - it’s great to hear that our long-neglected IPv6 stack is finally getting some fixes and enhancements. It’s a bit unfortunate that the reviews are progressing so slowly, but that’s the reality sometimes. Thank you very much for your work. I’ll keep my fingers crossed, and I promise to test at least some of these patches. From a tester’s perspective, it would be much easier if the changes were available as ready-to-apply commits (for example, something that could be applied with |git am|). Picking patches one by one from Phabricator and committing them manually into a local repository isn’t the most convenient workflow. Thanks again for the effort - it’s very much appreciated. Best regards, -- Marek Zarychta