The patch is updated: http://www.netbsd.org/~ozaki-r/run-mcast-tests-on-rump.diff
Now all tests pass on rump kernels :) ozaki-r On Mon, May 25, 2015 at 1:10 PM, Ryota Ozaki <[email protected]> wrote: > Hi, > > I found it. The problem is related to host's environments, but the > real problem is in t_mcast.c. It explicitly specifies an interface > of index=1: > > #else > ifc = 1; > #endif > m6.ipv6mr_interface = ifc; > m6.ipv6mr_multiaddr = s6->sin6_addr; > return setsockopt(s, IPPROTO_IPV6, IPV6_JOIN_GROUP, > &m6, sizeof(m6)); > > However, the interface indexes depend on host's environments. > In martin's environment, the interface of index=1 is awge0 and > it works in luck, but in my rump environment the interface of > index=1 is lo0 and it doesn't work with no luck. > > If we pass index=0, the kernel selects an appropriate interface > for mcast and IPv6 mcast tests pass on any environments if its > IPv6 setup is proper (inet6 addr, default gw, etc.). > > Thank you for your help, > ozaki-r
