On Tue, Jan 30, 2024 at 5:42 PM Paul Wouters <[email protected]> wrote:

> > Not necessarily. A VPN client might know for sure that the server it
> wants to talk to supports ESP ping. Before the IKE
> > handshake, it could send the ping, and if no response came back, it
> simply wouldn't bother with negotiating ESP or IPv6
> > at all and just go back to IPv4.
>
> That would be a poor implementation. A man-in-the-middle could quickly
> reply with an ICMP message before the ESP ping reply would come back.
> It would be a handy way to disable IKEv2/IPsec entirely.
>

So what if it did? In such a situation the client could simply say it will
fall back IKEv2 over IPv4 with UDP encap, which pretty all networks have to
support anyway. So it's not a DOS or a downgrade attack.


> The IKEv2 part should be much easier to get updated compared to the
> kernel support part. I would think it not very common to have kernel
> support without IKE support. So making it part of IKE makes sense to me.
>

Not necessarily. There are implementations that don't support IPv6 UDP
encap (e.g., Linux using kernels < 5.10, Android), and such implementations
can simply skip IKEv2 over IPv6 completely.
_______________________________________________
IPsec mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ipsec

Reply via email to