Just a thought in line with David C?s second message. What would the implications be if we redefine the semantics of the extension, where the current semantics applies for an EKU in an end-entity public-key certificate and the MS/Mozilla definition applies to CA-certificates? Erik Fra: x500standard-bounce@xxxxxxxxxxxxx [mailto:x500standard-bounce@xxxxxxxxxxxxx] På vegne af Santosh Chokhani Sendt: 2. november 2012 19:21 Til: x500standard@xxxxxxxxxxxxx Emne: [x500standard] Re: Processing of extended key usage extension Dave, I see some benefit in using an extension that constrains the path. In order to ensure backward compatibility, that should be done using a new extension. I wonder what the response would be if we approached Mozilla and MSFT with this recommendation. From: x500standard-bounce@xxxxxxxxxxxxx [mailto:x500standard-bounce@xxxxxxxxxxxxx] On Behalf Of David A. Cooper Sent: Friday, November 02, 2012 2:00 PM To: x500standard@xxxxxxxxxxxxx Subject: [x500standard] Re: Processing of extended key usage extension These seem like good ideas. I've never seen an intermediate certificate with an extended key usage extension, and this may be part of the reason that Microsoft's implementation of extended key usage hasn't caused much of a problem. Unfortunately, the reason that I found out about Mozilla's plans is that they intend to actively encourage the inclusion of extended key usage extensions in intermediate certificates that include currently defined key purpose OIDs (http://www.mozilla.org/projects/security/certs/policy/WorkInProgress/Inclus ionPolicy.html): * For a certificate to be considered technically constrained, the certificate MUST include an Extended Key Usage (EKU) extension specifying all extended key usages that the subordinate CA is authorized to issue certificates for. The anyExtendedKeyUsage KeyPurposeId MUST NOT appear within this extension. * If the certificate includes the id-kp-serverAuth extended key usage, then the certificate MUST include the Name Constraints X.509v3 extension. The Name Constraints extension MUST contain a dNSName permittedSubtrees constraint that only contains domains for which the issuing CA has confirmed that the subordinate CA has registered or has been authorized by the domain registrant to act on the registrant's behalf. If there are no such dNSNames (e.g. because the certificate is for issuing IP-address-based certificates), then the certificate MUST contain a dNSNames constraint that prohibits all DNS names. * If the certificate includes the id-kp-emailProtection extended key usage, then all end-entity certificates MUST only include e-mail addresses or mailboxes that the issuing CA has confirmed (via technical and/or business controls) that the subordinate CA is authorized to use. * If the certificate includes the id-kp-codeSigning extended key usage, then the certificate MUST contain a directoryName permittedSubtrees constraint where each permittedSubtree contains the organizationName, localityName (where relevant), stateOrProvinceName (where relevant) and countryName fields of an address that the issuing CA has confirmed belongs to the subordinate CA. Mozilla's draft policy imposes additional requirements on CA certificates that are not "technically constrained." The general response on the Mozilla web site (https://bugzilla.mozilla.org/show_bug.cgi?id=725351) seems to be "everyone else is ignoring the X.509 semantics for extended key usage, so why shouldn't Mozilla?". I asked why, if everyone feels that the X.509 semantics are wrong, no one has proposed a change to the standard. The only response I've gotten so far cam from Brian Smith who said "I expect that will happen as part of the WebPKI effort at the IETF." I would certainly hope the IETF wouldn't try to use the Web PKI Operations working group (assuming one forms) to try to change the semantics of an X.509 extension. Any change would have to be made in X.509, not in the IETF, and get the impression from the discussions on this list that there isn't interest in changing X.509. Dave On 10/27/2012 10:43 AM, Michael Ströder wrote: Basically for PKI practitioners this means: 1. Never ever set extendedKeyUsage in intermediate certs. 2. Never ever sign certs with a key for which the accompanying public-key cert contains the extendedKeyUsage extension. On 10/22/2012 05:46 PM, David Chadwick wrote: However, one could possible use this line of argument (I am not condoning their current approach, only playing devil's advocate for them). EKU is a bucket which allows new extended key uses to be defined. If all the EKUs that they define, specify in their definition that this EKU does not apply to the certificate in question, but to the certificate of the subject, then one could argue that it is OK, since you have to understand the value of the EKU in order to implement it, and when you understand the value you understand the semantics that are applied to the value ----- www.x500standard.com: The central source for information on the X.500 Directory Standard.