Mailing List Archive

Re-sign subkey binding with changed digest?
So, this SHA-1 mess is "fun".

To get a fresh self-sig user ID signature on the main key, I can do
this:

gpg --expert --cert-digest-algo SHA256 --sign-key ${KEYID:?}

The `--expert` overrides the "already signed" safety check, letting you
confirm that yes you really want this. Alas, it seems that
`--ask-cert-expire` is not enough, it no-ops out.

For sub-key bindings, for encryption keys it's easy: just generate a new
encryption sub-key, let it be signed with a modern hash, and future
messages encrypted to you will just use the new subkey.

For non-encryption subkeys, I'm looking really at signing subkeys: it
seems useful to make sure that existing signatures can continue to be
verified.

How do I re-sign the subkey binding for a [S] signing subkey, to keep
the same key but make the association from the main key be with SHA256
please?

Thanks,
-Phil
Re: Re-sign subkey binding with changed digest? [ In reply to ]
> On 8 Jan 2020, at 20:05, Phil Pennock via Gnupg-users <gnupg-users@gnupg.org> wrote:
>
> How do I re-sign the subkey binding for a [S] signing subkey, to keep
> the same key but make the association from the main key be with SHA256
> please?

Have you tried changing the subkey expiry? Or does that reuse the same hash?

A
_______________________________________________
Gnupg-users mailing list
Gnupg-users@gnupg.org
http://lists.gnupg.org/mailman/listinfo/gnupg-users
Re: Re-sign subkey binding with changed digest? [ In reply to ]
On Wed, 8 Jan 2020 21:37, Andrew Gallagher said:

> Have you tried changing the subkey expiry? Or does that reuse the same hash?

That is what I would also suggest. The expire sub-command is useful for
all such things. It should always use the current default digest
algorithms.

Regarding the SHA-1 collisions: GnuPG 2.2 still considers SHA-1 based
self-signatures (either on a user-id or a subkey) has valid. If we
would disallow that all dsa1024 keys would be rendered useless. dsa1024
requires SHA-1. Compared to the trouble we already had with removing
PGP-2 keys, removing dsa1024 would be a much loader outcry.

Nevertheless, moving away from dsa1024 is important. We just can't
force users to do that.


Shalom-Salam,

Werner

--
Die Gedanken sind frei. Ausnahmen regelt ein Bundesgesetz.