NTPsec 1.2.5 released
James Browning
jamesb192 at jamesb192.com
Thu Aug 6 13:25:02 UTC 2026
> On 08/06/2026 5:16 AM PDT Hal Murray via devel <devel at ntpsec.org> wrote:
>
>
> This area is a big mess. I think we have to figure out what we mean by
> "support".
>
> We need crypto in 3 areas:
> MD5 to hash IPv6 addresses to fit into 32 bits.
> Shared key
> NTS
Leap file signature check?
> What does support mean if it can't build with NTS?
We lose the a feature we have and classic ntpd lacks.
> FreeBSD ships with one version of OpenSSL, but you can install a newer (or
> older) version from their packages collection. NTPsec knows how to find
> it.
>
> $ openssl version
> OpenSSL 3.5.6 7 Apr 2026 (Library: OpenSSL 3.5.6 7 Apr 2026)
>
> $pkg search
> openssl-3.0.21,1 TLSv1.3 capable SSL and crypto library
> openssl34-3.4.6 TLSv1.3 capable SSL and crypto library
> openssl35-3.5.7 TLSv1.3 capable SSL and crypto library
> openssl36-3.6.3 TLSv1.3 capable SSL and crypto library
> openssl40-4.0.1 TLSv1.3 capable SSL and crypto library
>
> -------
>
> You can build yourself a version of OpenSSL from source. I do it for
> testing upcoming releases. Directions are in HOWTO-OpenSSL I haven't
> tested those directions and any of their tarballs on any really old
> systems.
>
> -------
>
> > I think we wanted to keep supporting OpenSSL 1.1.1 for a bit longer since
> > several Linux distributions still have it. Eg:
>
> How many of the people running old systems want modern NTPsec? I assume
> the reason they are running such old code is that they don't want to rock
> the boat.
>
> > - Debian 11 shipped OpenSSL 1.1.1. Standard LTS support ends August 2026.
>
> From apt show ntpsec on Debian 11/bullseye
> Package: ntpsec
> Version: 1.2.0+dfsg1-4
>
> ----------
>
> The reason that I want to get rid of supporting 1.1.1 is to simplify the
> code.
>
> There are 28 places that start with
> #if OPENSSL_VERSION_NUMBER
>
> There is more in the attic, but they don't come under the support umbrella.
>
> We did all the early NTS development on 1.1.1. I don't remember any major
> changes when 3.0 came out.
I'm sure there are some; I'm too lazy to dig up the release notes.
> Most of the ugliness is in the shared key code where I've been working recently. I'll look into cleaning that up when things calm down. I have the HMAC code close to ready to merge, but there is no rush for that.
>
> In case anybody hasn't noticed yet, NIST now supports good crypto on their authenticated (aka shared key) service.
> https://www.nist.gov/pml/time-and-frequency-division/time-services/nist-authenticated-ntp-service
>
> SHA256 already works so I don't see any rush for HMAC-SHA256.
{aes,aria,camellia}-{128,192,256} ciphers "work".
the following digests, "work".
sha{,3}-{224,256,384,512} blake2{b512,s256} md5 sha1 rmd160
I see more things on the lists that I don't remember checking.
contrib/keygone.py "works" for generating keys for all the above.
More information about the devel
mailing list