On Tue, 2019-02-19 at 13:42 +0100, Mike Massonnet wrote: > Package: iproute2 > Version: 4.9.0-1+deb9u1 > Severity: normal > Tags: ipv6 > > > I stepped on a bug that is triggered when enabling ipv6 forwarding. > > Whenever I request the route for my subnet, with ipv6 forwarding > enabled, it responds with the looopback. > > For example a server with IPv6 1234:1234:1:2::40/64 on eth0. > > When I run the commands: > `ip route get 1234:1234:1:2::/64` > `ip route get 1234:1234:1:2::0` > > I get the route with the right interface: > > ``` > # sysctl net.ipv6.conf.all.forwarding > net.ipv6.conf.all.forwarding = 0 > > # ip route get 1234:1234:1:2::/64 > 1234:1234:1:2:: from :: dev eth0 proto kernel src 1234:1234:1:2::40 > metric 256 pref medium ``` > > But as soon as I enable forwarding it returns a bad value: > > ``` > # sysctl net.ipv6.conf.all.forwarding=1 > net.ipv6.conf.all.forwarding = 1 > > # ip route get 1234:1234:1:2::/64 > local 1234:1234:1:2:: from :: dev lo table local proto none src > 1234:1234:1:2::40 metric 0 pref medium ``` > > I can "cheat" around, for example by requesting the route for > 1234:1234:1:2::1/64 it will return the expected route. > > Regards, > Mike
Hi, I can't reproduce this: $ sudo sysctl -w net.ipv6.conf.all.forwarding=1 net.ipv6.conf.all.forwarding = 1 $ ip route get 1234:1234:1:2::/64 1234:1234:1:2:: from :: via fe80::<foo> dev eth0 proto static src <foobar> metric 100 pref medium $ ip route get 1234:1234:1:2::1/64 1234:1234:1:2::1 from :: via fe80::<foo> dev eth0 proto static src <foobar> metric 100 pref medium This is just my desktop with one interface and ipv6 configure via dhcp. -- Kind regards, Luca Boccassi
signature.asc
Description: This is a digitally signed message part
