RE: [pkix] Last Call: draft-ietf-pkix-tamp (TrustAnchor ManagementProtocol(TAMP)) to Proposed Standard

["Carl Wallace" <[e-mail]>]

Though we've been through each of these points before responses are inline...

 

From: [e-mail] [mailto:[e-mail]] On Behalf Of Denis Pinkas
Sent: Friday, January 29, 2010 3:19 AM
To: ietf
Cc: pkix
Subject: Re: [pkix] Last Call: draft-ietf-pkix-tamp (TrustAnchor ManagementProtocol(TAMP)) to Proposed Standard

 

Carl,

 

You said: "the current protocol is able to accommodate the web browser model and 
does so for the existing path processing constraints defined in RFC 5280, i.e., 
name constraints, certificate policies and certificate policy constraints".

 

Unfortunately, this is not the case. Applying "name constraints, certificate policies 
and certificate policy constraints" as defined in RFC 5280 is not sufficient to accommodate 
the web browser model.

 

The web browser model controls characteristics which only apply to leaf certificates, 
in practice EKU (Extended Key Usages) and OIDs of Certication Policies.

[CW] Certificate policies and policy constraints are fully supported.  EKU is not processed across a certification path so its utility in a TA is limited.  Independent of TAMP/TAF, the EKU-like mechanism used by some browsers has been the subject of mailing list posts describing interoperability problems.  This could be addressed by defining and using a similar extension that has associated path processing rules.  It has also been suggested that the certificate policies extension could serve this purpose without defining a new extension.  TAMP is not the place to sort out that issue.  

 

You claim that this feature could be provided as an extension to the protocol. 

[CW] I claim this, have given pointers to similar extensions and have offered to co-author or review the new specification.  

 

This is an acknowledgment that the current document does not currently support the web browser model.

[CW] The use of an EKU extension in a TA is not a different model.  It's a different extension that fits within the model that has been defined.  

 

The current draft is in fact covering three use cases, none of them is correctly addressing the web browser model.

 

Should an extension be defined, it would be difficult to use, since extensions, as supported in the draft, 
mandate to use two separate operations: to set the initial content of a trust anchor and then to modify it 
afwterwards using a TAMPUpdate operation (which is solely able to use extensions).

[CW] This is not correct.  A trust anchor can be added to a trust anchor store with a full definition (including extensions) using an add operation.  There is no need for a second message simply to set extensions.

 

The initial content of a Trust Anchor is defined by:

 

    TrustAnchorChoice ::= CHOICE {
        certificate  Certificate,
        tbsCert      [1] EXPLICIT TBSCertificate,
        taInfo       [2] EXPLICIT TrustAnchorInfo }

 

None of these options, include an extension field. 

[CW] All of these options include an extensions field: Certificate.tbsCertificate.extensions, TBSCertificate.extensions, TrustAnchorInfo.exts.

 

Only the TAMP update operation includes an extension field:

 

    TBSCertificateChangeInfo  ::=  SEQUENCE  {
      serialNumber         CertificateSerialNumber OPTIONAL,
      signature            [0] AlgorithmIdentifier OPTIONAL,
      issuer               [1] Name OPTIONAL,
      validity             [2] Validity OPTIONAL,
      subject              [3] Name OPTIONAL,
      subjectPublicKeyInfo [4] SubjectPublicKeyInfo,
      exts                 [5] EXPLICIT Extensions OPTIONAL  }



 

Using a change function to add information is not the right way to proceed.

 

The protocol is unable to support the sending of a full description of a trust anchor, 
including the support of extensions, all in a single exchange.

[CW] The protocol fully supports the sending a full description of a trust anchor, including the support of extensions, all in a single exchange.  You reference the change operation above.  Look at the add operation.

 

As said in the PKIX list, this can be done in a single step. Proposals have been posted to demonstrate how it could be done.

It has been responded that the proposal was correctly adressing the issue in principle, but the editors were not willing 
to make a change which was considered as a major change to the initial proposal.

 

Another major issue for this draft is that it is unable to tell for which usage (e.g. for which application or which purpose) 
each trust anchor may be used.

[CW] A variety of extensions can be included to indicate the intended usage of a trust anchor so it's easy to look at a trust anchor and find this information.  

 

All these issues led me to propose that this document proceeds on the EXPERIMENTAL track, 
thus leaving room for a STANDARD protocol adressing the needs of the Internet community 
when using X.509 self-signed certificates associated with metadata. 

 

Denis  

	De : pkix-bounces 

	À : denis.pinkas,ietf 

	Date : 2010-01-25, 16:20:06

	Sujet : Re: [pkix] Last Call: draft-ietf-pkix-tamp (Trust Anchor ManagementProtocol(TAMP)) to Proposed Standard

	 

	Denis,

	As we have discussed on the PKIX mailing list, the current protocol is able to accommodate the web browser model and does so for the existing path processing constraints defined in RFC 5280, i.e., name constraints, certificate policies and certificate policy constraints.  The problem you are referring to is really with the current EKU extension, which is not processed across a certification path.  Were one to define an EKU variant that has path processing semantics, TAMP would convey this information just fine.  Other specifications have defined extensions for inclusion in trust anchors to extend the RFC 5280 set, including RFC 3779 and CCC.  Something similar is appropriate for this purpose.

	Carl

	From: [e-mail] [mailto:[e-mail]] On Behalf Of Denis Pinkas
	Sent: Monday, January 25, 2010 3:49 AM
	To: ietf
	Cc: pkix
	Subject: Re: [pkix] Last Call: draft-ietf-pkix-tamp (Trust Anchor ManagementProtocol (TAMP)) to Proposed Standard

	The current protocol has severe limitations.

	They have been pointed during the last call at the PKIX WG level, but the protocol 
	has not been modified to address them.The end result has only been to add text 
	to explain the limitations without removing these limitations.

	See section 3: "When using these structures without any additional extension, 
	for which purposes the trust anchor info shall be used to verify 
	certification paths needs to be locally defined; this means that different 
	usages for the same or different trust anchors placed in the same TAS 
	are not possible either.

	One way to have different usages for different trust anchors without 
	using extensions is to use a different TAS for every different usage".

	The consequences are as follows:

	All web browser providers currently use a different model to manage trust anchors. 
	They are able to associate different key usages for every leaf certificate 
	with any trust anchor (all placed in the same trust anchor store). This can be done 
	in a single operation.

	Furthermore, with the introduction of EV SSL Certificates 
	(i.e. Extended Validation SSL certificates) the Certification Policy OIDs of 
	leaf certificates that fulfills the requirements of EV SL certificates 
	are added to the trust anchor to which the EV SSL certificate relates.

	This means that supporting the web browser model mandates to be able to add 
	key usages (e.g. EKU extended key usages) for leaf certificates 
	as well as Certification Policies for leaf certificates.

	This is not possible with the proposed protocol.

	As a consequence, the current protocol is unable to accomodate the web browser model.

	Since the protocol seems to be sufficient for another community 

	(but not to the Internet community), it is proposed to place this document 

	on the EXPERIMENTAL track rather than on the standards track.

	Denis

		Date : 2010-01-14, 18:34:14

		Sujet : [pkix] Last Call: draft-ietf-pkix-tamp (Trust Anchor Management Protocol (TAMP)) toProposed Standard

		The IESG has received a request from the Public-Key Infrastructure 
		(X.509) WG (pkix) to consider the following document:
		
		- 'Trust Anchor Management Protocol (TAMP) '
		   <draft-ietf-pkix-tamp-05.txt> as a Proposed Standard
		
		This document includes a downref to draft-ietf-pkix-new-asn1, which
		is under consideration by the IESG for publication as an Informational RFC.
		This document updates ASN.1 modules for PKIX specifications to conform to
		the 2002 version of ASN.1, but makes no changes to the bits on the wire.
		The community is specifically requested to consider whether down refs
		to draft-ietf-pkix-new-asn1 are appropriate in the general case, 
		in addition to the specific case of draft-ietf-pkix-tamp.
		
		The IESG plans to make a decision in the next few weeks, and solicits
		final comments on this action. Please send substantive comments to the
		[e-mail] <mailto:[e-mail]>  mailing lists by 2010-01-28. Exceptionally, 
		comments may be sent to [e-mail] <mailto:[e-mail]>  instead. In either case, please 
		retain the beginning of the Subject line to allow automated sorting.
		
		The file can be obtained via
		http://www.ietf.org/internet-drafts/draft-ietf-pkix-tamp-05.txt
		
		
		IESG discussion can be tracked via
		https://datatracker.ietf.org/public/pidtracker.cgi?command=view_id&dTag=17760&rfc_flag=0
		
		_______________________________________________
		pkix mailing list
		[e-mail] <mailto:[e-mail]> 
		https://www.ietf.org/mailman/listinfo/pkix