On Sun, Mar 13, 2016 at 6:50 PM, David Miller <da...@davemloft.net> wrote: > From: Mahesh Bandewar <mah...@bandewar.net> > Date: Wed, 9 Mar 2016 13:49:49 -0800 > >> This does happen as expected for egress traffic however on ingress >> traffic, the IPvlan packet-handler changes the skb->dev and this >> forces packet to be processed with the IPvlan slave and it's >> associated ns. This causes above mentioned problem and few other >> which are not yet reported / attempted. e.g. IPsec with L3 mode or >> even ingress routing. > > And now your changes make it so that we can't run netfilter rules on > the slave's device within the slave's ns. > > If someone is doing that now you're breaking things for them. > Yes, I agree. If someone is using netfilter hooks in slave-ns, then this would break for them. If we do this as a separate mode (e.g. L3s) keeping the current mode as it is. Would that be an acceptable solution?
> It doesn't matter whether doing so or not makes sense. > > You're going to have to find a way to do both, and also I'm concerned > about how you're leaking the source namespace's "stuff" into the > destination's. That's very worrisome to me. > If we add a new mode (e.g. L3s) and preserve current mode as is it, then that should address your first concern. Now about leaking ns "stuff". I do not want to leak stuff across ns. In the current L3 mode there is an issue that Nicolas has pointed out and I'll fix that soon. In the new mode (L3s); right from the beginning the L2 and L3 processing always (and consistently) will happen with master device (proposed skb->dev->l3_dev). > There is no way I'm applying this as-is, sorry.
