Re: [pkix] Proposed resolution to non-issued certificates - 2560bis

[Simon Tardell <[e-mail]>]

On Fri, Nov 2, 2012 at 11:16 AM, Peter Rybar <[e-mail]> wrote:

> ** **
>
> Stefan,****
>
> ** **
>
> I agree with David A. Cooper and Piyush.****
>
> ** **
>
> And also retroactive revocation is not allowed in X.509.****
>
> The certificate status "revoked" in OCSP must be consistent with the
> status in CRL.****
>
> In this case it is not consistent because status "revoked" is proposed to
> be in OCSP but in the same time it is not included in CRL.****
>
> Validation will be different by CRL/OCSP!
>
So a CA that does answer revoked for never issued MUST NOT use CRLs.

> **
>
> Don't forget that also thisUpdate MUST be correctly used and included in
> OCSP response.****
>
> ** **
>
>    - thisUpdate: The time at which the status being indicated is known to
> be correct.****
>
> ** **
>
> It mans any new "correct" revocation MUST be with revocation value which
> is greater than value thisUpdate of any issued CRL/OCSP.
>
Including revocation date/time before value thisUpdate (of any already
> issued CRL/OCSP) is destruction of X.509 fundamental rules for validations.
> ****
>
> Date and time of revocation MUST be only after value thisUpdate which is
> included in last issued CRL or OCSP.****
>
> **
>

I don't quite understand. The revocationTime of a RevokedInfo must
logically be earlier than the thisUpdate of the response, unless we assume
that we are dealing with a precog OCSP responder. The OCSP responder should
try to keep the thisUpdate as close to "now" as it can, to limit the window
where compromised keys can be used.

/Simon