Re: [Ntp] An NTPv5 design sketch

[Daniel Franke <[e-mail]>]

On Thu, Apr 16, 2020 at 4:26 AM Miroslav Lichvar <[e-mail]> wrote:
> The device may be very simple. It may not have an OS and NTP may be
> the only networking it does. It could be measuring intervals in a
> physics experiment, or controlling a robot in a factory. Consider
> where and why PTP originated and that NTPv5 with its correction field
> might be usable there too.

I remain skeptical that systems actually exist, even in these domains,
where unprotected NTPv5 would be a good solution but NTPv5 with NTS
would not be. I've forwarded this thread to a friend who has done a
great deal of work with systems of this nature so that she weigh in
further. At any rate, I second Doug Arnold that if a use case is
already well-served by PTP, then complicating NTPv5 on their behalf is
not solving anyone's problem.

> Protected responses need to be handled in the same way as unprotected
> responses. You never know if the server isn't compromised and trying
> to attack you.

I think you've misunderstood me here. I'm not talking about cases like
"don't do an out-of-bounds read if the length field is longer than the
actual packet length" which yes, need to be handled regardless. I'm
referring to handling NTS stripping attacks and making sure you don't
accept an unprotected packet from a source that should only be sending
protected ones.