On Wed, Jan 31, 2024 at 10:44:07PM -0600, Yan Zhai wrote: > > > > diff --git a/tools/testing/selftests/bpf/prog_tests/lwt_redirect.c b/tools/testing/selftests/bpf/prog_tests/lwt_redirect.c > > index beeb3ac1c361..b5b9e74b1044 100644 > > --- a/tools/testing/selftests/bpf/prog_tests/lwt_redirect.c > > +++ b/tools/testing/selftests/bpf/prog_tests/lwt_redirect.c > > @@ -203,6 +203,7 @@ static int setup_redirect_target(const char *target_dev, bool need_mac) > > if (!ASSERT_GE(target_index, 0, "if_nametoindex")) > > goto fail; > > > > + SYS(fail, "sysctl -w net.ipv6.conf.all.disable_ipv6=1"); > > Thanks for digging into this! I was totally unprepared for that many > router solicitations when I wrote the wait logic. For now disable v6 > is totally good to unblock similar scenarios. But think it twice it is > probably still worthwhile to incorporate v6 later since lwt hooks mess > with both v4/v6 routing. So I'll try to fix up the wait logic later > this week. An exact packet filter is probably best suited to make > icmpv6/arp/nd happy. > Greatly appreciated Yan. I definitely agree that it is desirable to have V6 support to. In case that helps... [0] should help you repro locally on your favorite architecture. [0] https://chantra.github.io/bpfcitools/bpf-local-development.html Manu > best > Yan > > > SYS(fail, "ip link add link_err type dummy"); > > SYS(fail, "ip link set lo up"); > > SYS(fail, "ip addr add dev lo " LOCAL_SRC "/32"); > > -- > > 2.39.3 > >
