[Rpm-maint] [rpm-software-management/rpm] Ignore most unhashed subpackets in OpenPGP signatures (Issue #1886)

[Justus Winter]

> Agreed. I would go further and reject a signature that has a subpacket which cannot validly be in the unhashed section.

I think that is too far.  At the very least, allow unknown and private subpackets there, as they may be self-authenticating.

> Is there ever a reason to have an embedded signature in a type 0 or type 1 signature? The only use-case I can think of is timestamping.

The predominant use case are primary key binding signatures, but RPM doesn't want to even attempt to canonicalize certificates.  Timestamping is a great use case though :) But, for timestamping purposes, we'd want the signature to end up in the hashed subpacket area.

> At the very least, an embedded signature must not also have an embedded signature to avoid recursion.

That is only interesting if you actually attempt to verify the embedded signature.  I don't see RPM doing that anytime soon.  First, they don't want to canonicalize certs.  Second, for timestamping purposes you'd need the certificate of the timestamping entity.  That requires non-local reasoning, and I don't see that on the horizon either, given the state of RPM's OpenPGP implementation.

-- 
Reply to this email directly or view it on GitHub:
https://github.com/rpm-software-management/rpm/issues/1886#issuecomment-1014606583
You are receiving this because you are subscribed to this thread.

Message ID: <rpm-software-management/rpm/issues/1886/1014606583 at github.com>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rpm.org/pipermail/rpm-maint/attachments/20220117/ad5082ef/attachment.html>