Creating an X25519 client certificate

[Robert Moskowitz]

On 3/17/21 8:17 PM, Viktor Dukhovni wrote:
> On Wed, Mar 17, 2021 at 07:44:05PM -0400, Robert Moskowitz wrote:
>
>>>> I have created my X25519 pub/priv keypair with:
>>>>
>>>> openssl genpkey -algorithm X25519\
>>>>        -out $dir/private/$clientemail-X.key.$format
>>> Are you sure you didn't want ed25519 instead?  X25519 is a key agreement
>>> menthod, not a signature method.
>> My limited understanding is that for encrypted S/MIME, and ECDH cert
>> is needed.  Thus if ED25519 is used for signing said S/MIME, X25519 is
>> used for encrypting.
> I see, this is for CMS, so you're trying to support:
>
>      https://tools.ietf.org/html/draft-ietf-curdle-cms-ecdh-new-curves-10
>
>>>> openssl req -config $dir/openssl-intermediate.cnf\
>>>>        -key $dir/private/$clientemail-X.key.$format \
>>>>        -subj "$DN" -new -out $dir/csr/$clientemail-X.csr.$format
>>>>
>>>> which is what I used for ED25519 client certs.  But I get an error:
>>>>
>>>> 140487683954496:error:0608D096:digital envelope
>>>> routines:EVP_PKEY_sign_init:operation not supported for this
>>>> keytype:crypto/evp/pmeth_fn.c:39:
>>> Not surprising, why do you expect this to work?
>> Shooting from the hip, a bit.  If I am going to have an X25519 cert,
>> then I need a csr and this is the command to make one.  So try it and
>> see what it does.  Not too well, it turns out.
> Well, CSRs are self-signed, and X25519 does not support signing, so
> you CANNOT have an X25519 CSR.

Slap myself on the forehead....

Of course I know that.  But did not stop to think this through.   :(

Will read through all this and get back here....

> You can however create an X25519
> certificate directly for a given key:
>
> https://crypto.stackexchange.com/questions/19452/static-dh-static-ecdh-certificate-using-openssl
>
> with other details (names, extensions, ...) pulled from a CSR.
> What can't get is proof of posession.
>
>>>> Oh, and I am ASSuMEing that a CA cert of ED25519 signs an X25519 client
>>>> cert.  Haven't found instructions on this, but it seems reasonable...
>>> https://crypto.stackexchange.com/questions/27866/why-curve25519-for-encryption-but-ed25519-for-signatures
> Well the CA can use any algorithm that supports signing, and is widely
> supported.  It does not have to be Ed25519.
>