[cabfpub] Private key control

Bruce Morton bruce.morton at entrust.com
Fri Oct 24 10:14:17 MST 2014


Ben,

I do not believe that we technically validate private key control. Obtaining a CSR does not verify the control of the private key. You would need to compare the signature on the CSR to something.

I think what all CAs do is receive authorization to issue a certificate for a certain domain. We receive authorization by contacting the applicant, the domain owner, or accepting a two-factor authorization.

So I would suggest that we either use some wording about authorization or put in No Stipulation. I do not believe that addressing this requirement covers any policy that has been provided from a browser.

Bruce.

From: Ben Wilson [mailto:ben.wilson at digicert.com]
Sent: Friday, October 24, 2014 12:47 PM
To: Bruce Morton; Rick Andrews; Jeremy Rowley; CABFPub
Subject: RE: Private key control

Bruce,
I assume you're not opposed to coming up with wording that is less specific but would at least eventually fill in content for section 3.2.1 of an RFC-3647-formatted CP, which is what I'd like us to be working toward.    If we really want to strip this down, even though I think we should address the MITM concern, then maybe we could just say something like, "Prior to issuing a Certificate, the CA MUST verify that the Applicant possesses the Private Key associated with the Public Key to be included in the Certificate.  The CA MAY verify this association by obtaining a CSR from the Applicant."  ?

In other words, I don't think anyone wants to disrupt internal practices of CAs that comply with this textbook CA baseline practice, unless you are saying that for the Web PKI, proof of possession isn't necessary.  If you are, then do you have an alternative you could propose?  That's what I'm looking for-something that we can all agree on that will fill a hole.
Thanks,
Ben


From: public-bounces at cabforum.org<mailto:public-bounces at cabforum.org> [mailto:public-bounces at cabforum.org] On Behalf Of Bruce Morton
Sent: Friday, October 24, 2014 8:19 AM
To: Rick Andrews; Jeremy Rowley; CABFPub
Subject: Re: [cabfpub] Private key control

I think the requirement should be dropped.

If we only validate the signature on the CSR, then we do not know if there is a man-in-the-middle. You need some other data.

If we want to technically validate private key control, then we should take some action such as sending the Subscriber some information out-of-band for signature. The signature would be compared to the signature on the CSR to see if the same key was used.

The softer way that an OV/EV private key control is confirmed is by contacting the Subscriber out of band to confirm that they made/authenticate the request. I don't think this works for DV. It will also not work when the Subscriber can approve a certificate issuance with dual-factor login.

I don't think we should have the requirement unless we suggest methods that will actually work for all certificate types and our current certificate management methods.

This requirement is not in the BRs or the EV guidelines and we have not been suffering from an incidents, so again, I think the requirement should be dropped.

Bruce.

From: public-bounces at cabforum.org<mailto:public-bounces at cabforum.org> [mailto:public-bounces at cabforum.org] On Behalf Of Rick Andrews
Sent: Thursday, October 23, 2014 2:46 PM
To: Jeremy Rowley; CABFPub
Subject: Re: [cabfpub] Private key control

Jeremy,

How about "The CA MAY verify this association by obtaining a CSR from the Applicant and validating the signature on the CSR."

-Rick

From: public-bounces at cabforum.org<mailto:public-bounces at cabforum.org> [mailto:public-bounces at cabforum.org] On Behalf Of Jeremy Rowley
Sent: Wednesday, October 22, 2014 6:57 PM
To: CABFPub
Subject: [cabfpub] Private key control

During the Code Signing BR discussion a few weeks ago, we noticed that the Baseline Requirements lack a definitive requirement for the CA to confirm that the Application is properly associated with the Public Key being included in the certificate.  We'd like to remedy this oversight.  What does everyone thing about adding a section similar to the following to the BRs?
Section 11.1.5    Verification of Key Pair Association
Prior to issuing a Certificate, the CA MUST verify that the Applicant's Private Key is properly associated with the Public Key and a subject name to be included in the Certificate. The CA MAY verify this association by obtaining a CSR from the Applicant.

Jeremy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://cabforum.org/pipermail/public/attachments/20141024/0ce093f6/attachment.html 


More information about the Public mailing list