Re: [Ntp] NTP Security (was NTPv5: big picture)

[Kurt Roeckx <[e-mail]>]

On Tue, Jan 19, 2021 at 06:26:24PM +0100, Miroslav Lichvar wrote:
> On Tue, Jan 19, 2021 at 05:50:20PM +0100, Magnus Danielson wrote:
> > On 2021-01-19 14:51, Miroslav Lichvar wrote:
> > > The assumption is that both your DNSSEC and NTS was compromised at
> > > some point. You found out and rebuilt your infrastructure from
> > > scratch, but the attacker can still perform a MITM attack on your
> > > device if it doesn't know the current date in order to validate the
> > > certificate/records.
> > >
> > Outside of our scope. Way outside. We assume we can trust whatever these
> > mechanisms provide and our trust becomes tied to them. If you have ways
> > to attack that, this is not really the place to document those.
> 
> I don't agree. The mechanisms should be trusted only if used as
> intended, i.e. a full validation is performed. Skipping the time
> checks is a security risk. You seem to be saying that's ok, that you
> can combine outputs from two separate mechanisms using untrusted data
> and get a trusted output. I don't think that's the case. We can ask
> the DNS and CFRG groups.

I agree that you can't combine two things that are not validated
and end up with something that's validated. But everything really
is about probabilities and amount of work needed to break it. The
more keys that need to be compromised before you accept something
wrong, the more unlikely it is. I think it's really up to the
users to decide what is an acceptable risk.


Kurt