Given that the patents on point compression expired a few years ago, Libgcrypt should support standard point compression. In particular gcry_mpi_ec_get_mpi should support it, so that we can put a compressed point into the subjectKeyIdentifier as generated by gpgsm.
What kind of API should we offer?
(1) offering something like q@comp name for gcry_mpi_ec_get_mpi
If the intended use case will be in create_request function in gpg/sm/certreqgen.c, the 'q' is already generated in the form of SEXP.
It is up to an application (gpgsm), to convert non-compressed point representation to compressed point representation, here.
(2) keygen supporting (flag comp) for classic curves to ask compressed point representation for q, and support of compressed point representation for other places.
I'm afraid this will introduce requirement of libgcrypt 1.9 for gpg.
- compressed representation of EC point can be used in:
- public key
- (exporting) private key
- ECDH ephemeral key
- Accepting compressed representation,for the initial implementation, I'd like to limit our effort for curves of NIST and Brainpool, except NIST P-224, which p = 3 mod 4.
Reading compressed point (in keys) is supported (except for NIST P-224). When curve point is represented in compressed format, it is correctly interpreted now. So, for example, I think that with 1.9.0, gpgsm can handle certificate which uses compressed format in its curve point representation.
(I was wrong about signature (ECDSA signature (R,S) is composed by both integers. So, it's irrelevant. I had thought about EdDSA, which R is curve point, but it specifically has its own little-endian opaque format for R).)
Key generation never generate a key with compressed point (yet).
Please note that:
Currently, in gpgsm (sm/certreqgen.c), in the create_request function, sigkey is retrieved by gpgsm_agent_readkey, and curve point is extracted by get_ecc_q_from_canon_sexp from SEXP representation, then the curve point is used by ksba_certreq_add_extention in DER format. The curve point is passed with no interpretation or conversion, as is. (No use of gcry_mpi_ec_get_mpi here.)
When key generation with compressed point will be supported and it's used as sigkey, subjectKeyIdentifier will have compressed point representation.
Umm... I'm afraid, I would missed the point or have misunderstanding...