Re: [PATCH 0/4] Prevent UDP tunnels from operating on garbage socket

[Hannes Frederic Sowa <hannes@stressinduktion.org>]

On Tue, Apr 7, 2015, at 06:45, Tom Herbert wrote:
> On Mon, Apr 6, 2015 at 8:51 PM, David Miller <davem@davemloft.net> wrote:
> > I don't see what is convoluted about using the correct socket for
> > sending L3 protocol frames.  That's in fact how it's _supposed_ to
> > work.  And consistently having a proper matching socket available
> > makes it so that, long-term, we'll never have to deal with this issue
> > ever again.
> 
> I guess this is where I'm confused. We can send just about anything
> over GRE also, but have never needed a transmit socket for that. Is
> UDP encapsulation so different, or is GRE equally broken also? Also,
> will we need to add the socket to FOU and GUE then?

GRE, FOU, GUE in case of sk_mc_loop and the destination is a multicast
address, we imply sk_mc_loop() == true, what is not what we want. Tunnel
sockets deliberately set mc_loop to false but we cannot adhere to them,
yet. David's patchset changes that.

Also, I saw one inconsistency with sk_bound_dev_if in netfilter which
doesn't get solved by this patchset (it uses skb->sk->sk_bound_dev_if
unconditionally).

I think that having a struct-sock carrying over meta-information is a
good thing.

Thanks,
Hannes