> > of course I found a new topic where problems could be: > > radvd can send packets on unicast tunnels, too. > > so auto configuration should be possible even via the tunnel? > > Yes, in theory at least. In practice there might be all kinds of nasty > routing problems if you start sending RAs over the tunnel. > > Since the tunnel isn't a real device it will need a new route to the > tunnel endpoint after encapsulation. If we then have default routes both > over the tunnel interface and the underlying (physical) interface, the > routing might behave quite strangely (if at all). my plan is to not have a default route on the underlying interface - it is a wireless lan segment, not routed, all I do is ping, ssh and build up an ipsec tunnel (ike, and ip6ip6 packets). So adding a default route to the tunnel interface will not hurt anything. > Can you reproduce it, or at least describe what you did to get the panic? started racoon on the client. it was a wierd testing cycle, so I guess i can't reproduce it. But my normal setup gives me about 20-50 Call Traces during the boot sequence. Once someone has found those issues and fixed them, I will run bigger tests again. Right now with still so many xfrm related problems testing is a bit pointless. Regards, Andreas - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
