Am Tage des Herren Tue, 17 Jun 2025 15:52:34 +0100 Lexi Winter <i...@freebsd.org> schrieb:
> FreeBSD User: > > net.link.bridge.inherit_mac: 1 > > [...] > > > > ether 4c:52:62:2b:05:eb > > inet6 fe80::5a9c:fcff:fe00:393a%bridge0 prefixlen 64 scopeid 0x6 > > i was able to reproduce this: First of all, thank you very much for checking this. > > # sysctl net.link.bridge.inherit_mac=1 > net.link.bridge.inherit_mac: 0 -> 1 > # ifconfig bridge1 create inet6 -ifdisabled auto_linklocal up > # ifconfig bridge1 | egrep 'ether|inet6' > ether 58:9c:fc:10:43:03 > inet6 fe80::5a9c:fcff:fe10:4303%bridge1 prefixlen 64 scopeid 0xa > # ifconfig epair create > epair2a > # ifconfig bridge1 addm epair2a > # ifconfig bridge1 | egrep 'ether|inet6' > ether 02:36:0f:34:6b:0a > inet6 fe80::5a9c:fcff:fe10:4303%bridge1 prefixlen 64 scopeid 0xa > > some observations: > > - i don't think this is caused directly by the member_ifaddrs changes. > > - you can work around the problem by manually setting the bridge link > address when creating it instead of using inherit_mac, i.e.: > ifconfig bridge1 create link 02:00:00:00:00:01 > > - i don't think this is a bug per se, it's more that we simply don't > update the IPv6 lladdr when an interface's MAC address changes. > but i've copied net@ in case anyone else has an opinion on this. I'd dare to disaggree, we have running several recent CURRENT and 14-STABLE boxes (almost every 2nd day recompiled world/kernel) with very close configuration schemes (according to kernel config and NIC/bridge config as far it is possible to say so with respect to the different OS and different network addresses). On 14-STABLE with very similar jail configuration I do NOT see this (mis)behaviour of if_bridge(). -- A FreeBSD user
pgp68whOFifK8.pgp
Description: OpenPGP digital signature
