[lamps] Re: Mike Bishop's No Objection on draft-ietf-lamps-pq-composite-sigs-15: (with COMMENT)

[Mike Ounsworth <[e-mail]>]

Thanks Mike!

> It MUST be freshly generated, but if there's concern about potential
collisions due to limited sources of randomness

Yeah, so the reason behind wanting to bolt an ML-DSA key to an existing RSA
key is not to do with limited randomness, but rather to do with process and
audits, like, I've heard stories like "We've had this RSA key in production
for 85 years, and 35 government officials watched it be generated, and
we've syncronized it between datacentres by military briefcase. So bolding
an ML-DSA key to it is basically free (ie the audit requirements don't
prohibit that), but generating a fresh RSA key requires us to re-do all
that expensive and time-consuming ceremony". I'm exaggerating of course,
but sadly not by much.

So I'm not sure what you mean by "fallback". We have people who we know are
going to do exactly the bad thing and lose basically all of the security
properties of this document. I'm not sure what "fallback" means in that
context? Does it mean "fall through the floor to your death"?


> If someone's operational constraints dictate disabling core security
features of the protocol, you've explained why that's a bad idea and also
discussed how to mitigate the fallout.

Have we not done that already in 9.3?

"key reuse MUST be avoided to prevent the introduction of EUF-CMA
vulnerabilities."
"In addition, there is a further implication to key reuse regarding
certificate revocation. ... Therefore, ..., CAs performing revocation
checks on a composite key SHOULD also check..."
"the weakening of security from doing so can be mitigated by using an
appropriate ctx value, such as ctx=Foobar-dual-cert-sig to indicate..."


PS -- I am treating this as a DISCUSS because I think it needs to be
discussed because Section 9.3 is in fact deeply uncomfortable for exactly
the reasons you're raising. And thank you for raising them. If we can
improve this text, I'd like to do that.

On Thu, 9 Apr 2026 at 11:08, Mike Bishop <[e-mail]> wrote:

> I'm thinking of something like
> https://www.ietf.org/archive/id/draft-ietf-core-oscore-groupcomm-28.html#section-14.8,
> which says "We take this security precaution, because if we didn't, the
> following attack would work: ... The attack above is prevented because
> (thing) leads to detection of the attack." You're wanting to hint at a
> fallback position, so maybe also "Doing (alternative) would reduce the odds
> of the attack succeeding, but not eliminate it."
>
> That is, document the reason the requirement is there, discuss the
> fallback, and why it's helpful but not sufficient. If someone's operational
> constraints dictate disabling core security features of the protocol,
> you've explained why that's a bad idea and also discussed how to mitigate
> the fallout.
>
> Another approach might be to phrase it as uncertainty. It MUST be freshly
> generated, but if there's concern about potential collisions due to limited
> sources of randomness, this additional step can mitigate the fallout of a
> collision; if a "collision" is made more likely by someone's operational
> setup, you've at least helped.
>
> ------------------------------
> *From:* Mike Ounsworth <[e-mail]>
> *Sent:* Wednesday, April 8, 2026 8:41 PM
> *To:* Mike Bishop <[e-mail]>
> *Cc:* The IESG <[e-mail]>;
> [e-mail] <
> [e-mail]>; [e-mail] <
> [e-mail]>; [e-mail] <[e-mail]>;
> [e-mail] <[e-mail]>
> *Subject:* Re: [lamps] Mike Bishop's No Objection on
> draft-ietf-lamps-pq-composite-sigs-15: (with COMMENT)
>
> Hi @Mike Bishop
>
> Thanks for the careful review.
>
> I have made these changes here:
>
> https://github.com/lamps-wg/draft-composite-sigs/commit/c2d3ca12014874ab5adbc3a4750ed59e36c54011
>
>
> I think this COMMENT of yours is really astute and probably actually
> should be a DISCUSS. So let's discuss ;)
>
> In Section 9.3:
>
> >   designers are aware that some implementers may be forced to break
> >   this rule due to operational constraints.  This section documents the
> >   implications of doing so.
>
> This statement makes me nervous. Maybe this section should explain why the MUST
> exists, and not touch any implied "permission" to violate it?
>
>
> That section basically says "We are aware that people are going to have to
> violate the core security requirement of the draft, so let's analyse
> exactly what happens when you do".
> Trust me, it also makes me nervous. Very nervous.
> I have a customer in this situation, and trust me, I have tried my hardest
> to talk them out of it.
> The thinking here is that if we can't stop it, then at least we can
> document what sharp pointy edges you need to watch out for.
> Thoughts?
>
> On Wed, 1 Apr 2026 at 09:50, Mike Bishop via Datatracker <[e-mail]>
> wrote:
>
> Mike Bishop has entered the following ballot position for
> draft-ietf-lamps-pq-composite-sigs-15: No Objection
>
> When responding, please keep the subject line intact and reply to all
> email addresses included in the To and CC lines. (Feel free to cut this
> introductory paragraph, however.)
>
>
> Please refer to
> https://www.ietf.org/about/groups/iesg/statements/handling-ballot-positions/
> for more information about how to handle DISCUSS and COMMENT positions.
>
>
> The document, along with other ballot positions, can be found here:
> https://datatracker.ietf.org/doc/draft-ietf-lamps-pq-composite-sigs/
>
>
>
> ----------------------------------------------------------------------
> COMMENT:
> ----------------------------------------------------------------------
>
> Thank you for this work. This is an important step forward in our handling
> of
> PQ crypto.
>
> Section 3.1.1: There are a lot of MAY-examples in this section, when the
> whole
> thing is an overarching "MAY do whatever internally so long as the
> external is
> consistent." Consider not using MAY for every single example.
>
> ----
>
> Section 4 has:
>
> >   example, a stand-alone RSA private key can be encoded in Chinese
> >   Remainder Theorem form.  In order to obtain interoperability,
>
> Consider an informative reference?
>
> ----
>
> Section 6 has:
>
> >   Labels are represented here as ASCII strings, but implementers MUST
> >   convert them to byte strings using the obvious ASCII conversions
>
> While they may be obvious, calling them so comes across a bit oddly. Maybe
> just
> say something like "Labels are represented here as ASCII strips, but are
> octet
> sequences when used in..."
>
> ----
>
> Consider moving Section 6.2 to an appendix; it's not part of the core
> protocol,
> just design background.
>
> ----
>
> In Section 9.1:
>
> >   algorithm security or to provide migration flexibility.  Let's
> >   quickly explore both.
>
> It's typically advised to avoid first- or second-person language in
> specifications. I'd just drop this final sentence, personally.
>
> ----
>
> In Section 9.3:
>
> >   designers are aware that some implementers may be forced to break
> >   this rule due to operational constraints.  This section documents the
> >   implications of doing so.
>
> This statement makes me nervous. Maybe this section should explain why the
> MUST
> exists, and not touch any implied "permission" to violate it?
>
> ----
>
> In Section 9.4:
>
> >   message which happens to start with this string.  The designers
> >   accepted this trade-off.
>
> I would remove this last sentence. You've explained the trade-off; it's up
> to
> the implementers to decide whether it's worth making, no?
>
> ----
>
> Consider moving Section 10 to an appendix.
>
>
>
> _______________________________________________
> Spasm mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
>
>