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

[FUSTE Emmanuel <[e-mail]>]

Le 18/01/2021 à 16:46, Magnus Danielson a écrit :
> Hi,
>
> On 2021-01-18 12:38, Hal Murray wrote:
>> [e-mail] said:
>>> And all bootstrapping problems could be solved with CD / DO flags  controls
>>> and validation back checking with the first returned time of day.
>> How is that back checking going to work?  You don't know the time yet.
>>
>> Sure, it will say "OK" if you are talking to an honest server, but if you are talking to a bad guy, it can lie consistently.  Am I missing something?
> So, you only need to have sufficiently accurate time that DNSSEC will
> give thumbs up.
>
> Also, it only really need to match of that handshake.
>
> So, if you talk to server A, you can query which NTP server A say I
> should be able to trust.
>
> 1) Query A for NTP server B for A.
>
> 2) Query B for NTP time to use in initial exchange.
>
> 3) Query A for DNSSEC verified answer, using the time as initiated from B.
>
> 4) With verified answer, now handshake B using verified DNS records. If
> not verified, it fails.
>
> 5) Query B for NTP time.
>
> 6) If verified time matches that of the initial handshake request (as
> updated using local time progess), then unauthenticated time in step 2
> and 3 was valid, and thus the verification was valid and progress, else
> fail.
>
> So, we see how unauthenticated time can be used to bootstrap the
> handshake. Notice that this is not used to initiate node time, but only
> for that handshake. Only after things is secured, you accept time as
> being secured.
>
> This will work if the DNSSEC part and the consistency check in 6 leave
> enough margin.
>
> If it fails, it fails. This is why you want multiple sources for redundancy.
> mailman/listinfo/ntp
Exactly what I have in mind.
My reference to CD/DO flags are misleading, and is implementation details.
It for the case where your talk to a local DNS Server do do validation 
when you not embed the complete DNSSEC validation in your code (SMTP 
current usage deformation...).
With a proper embedded validation library (as Kurt said) the dependence 
on a local  (or non-local) validating resolver is no longer relevant 
(this library will do the necessary CD/DO flags dance if/as needed).

Emmanuel.