[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: encoding an X.509 certificate


At 8:22 PM -0500 11/12/08, Tom Gindin wrote:

        "Conservative in what you produce" - CA code should generate DER.
Not all such code does so, but it should.
        "Liberal in what you accept" - RP code should verify signatures on
the supplied binary TBSCertificate, rather than re-encoding.  For parsing
code to accept BER which isn't valid DER is almost harmless.    There may
well have been a good reason for believing that DER was more secure than
BER against various digest collision or pre-image attacks when X.509v1 was
in use and all certificate content had syntax verifiable by every RP (at
least in theory).  With non-critical extensions, that is no longer the
case.

                Tom Gindin


Tom,
I have to disagree. The standards call for DER to be used for signature generation and validation. They are generally silent on transport encoding. If we say that an RP should just try to verify the signature on whatever it receives, then we encourage (more) sloppy behavior by CAs. The preferred approach is for the RP to encode the received cert into DER before checking. 

Steve