On Tue, Sep 15, 2020 at 12:02:03AM +0200, Daniel Borkmann wrote: > today it is possible and > perfectly fine to e.g. redirect to a host-facing veth from tc ingress which > then goes into container. Only traffic that goes up the host stack is seen > by nf ingress hook in that case. Likewise, reply traffic can be redirected > from host-facing veth to phys dev for xmit w/o any netfilter interference. > This means netfilter in host ns really only sees traffic to/from host as > intended. This is fine today, however, if 3rd party entities (e.g. distro > side) start pushing down rules on the two nf hooks, then these use cases will > break on the egress one due to this asymmetric layering violation. Hence my > ask that this needs to be configurable from a control plane perspective so > that both use cases can live next to each other w/o breakage. Most trivial > one I can think of is (aside from the fact to refactor the hooks and improve > their performance) a flag e.g. for skb that can be set from tc/BPF layer to > bypass the nf hooks. Basically a flexible opt-in so that existing use-cases > can be retained w/o breakage. I argue that being able to filter traffic coming out of the container is desirable because why should the host trust the software running in the container to never send malicious packets. As for the flag you're asking for, it already exists in the form of skb->mark. Just let tc set the mark when the packet exits the container and add a netfilter rule to accept packets carrying that mark. Or do not set any netfilter egress rules at all to disable the egress filtering and avoid the performance impact it would imply. Thanks, Lukas
