Hi,

PKCS#1 1.5 signatures are obsolete. New specifications should not
mandate support for them.

RSA signatures in general are difficult for some devices to process
due to their large size. It would be frustrating to have used a pure
ECC infrastructure with no RSA involved at all, only to need to
implement RSA for the purpose of verifying signatures from logs. Thus
I think the group should consider dropping any mention of RSA
signatures from section 10.4.so that log clients do not have to
implement RSA.

If it really is important to have RSA signatures, then RSA PSS should
be used instead. In particular, it would be good to require the same
restricted form specified for TLS, where the same digest algorithm
must be used for all parts of the signature. Note that RSA PSS can be
made deterministic by using a fixed salt, and most implementations of
RSA PSS seem to support fixed salts if the salt length is set to zero.
As mentioned in the RSA PSS specification, PSS signatures are more
secure than PKCS#1 1.5 signatures even with a zero-length salt.

Cheers,
Brian
-- 
https://briansmith.org/

_______________________________________________
Trans mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/trans

Reply via email to