datatracker.ietf.org
Sign in
Version 5.3.0, 2014-04-12
Report a bug

Attribute Certificate (AC) Policies Extension
RFC 4476

Note: This ballot was opened for revision 06 and is now closed.

Summary: Needs a YES. Needs 10 more YES or NO OBJECTION positions to pass.

[Brian Carpenter]

Comment (2006-01-03)

Comments from Gen-ART review by Spencer Dawkins
(these are not blocking comments):

2. AC Policies Extension Semantics

  It should thus be noticed that an Attribute Authority (AA) does not
  necessarily support one single ACP. However, for each AC that is
  delivered it SHALL make sure that the policy applies to all the
  attributes that are contained in it.

Spencer: darned pronouns... (1) you lost me on who/what "it" is, in "it SHALL
make sure", and (2) I'm not totally clear on what the "it" in "contained in it"
is, either. Is this text saying something like "Each Attribute Authority (AA)
SHALL ensure that the ACP applies to all the attributes which are included in
the ACs that the AA issues" - which is almost exactly repeated in the Security
considerations, anyway? ("AAs SHALL ensure that the ACP applies to all the
attributes which are included in the ACs they issue")

2.1 AC Policy Extension Syntax

Spencer: is there a bad paragraph break in the following text (should it be all
one sentence)?

  This specification defines two policy qualifier types for use by
  ACP writers and AAs.  The qualifier types are the ACPS Pointer and
  the AC User

  Notice qualifiers.

Spencer: is it clear to those skilled in the art what "local" scope is, in the
following text? Within an AA, or ...? Since an URI could end up anywhere on the
Internet, I'm not sure what this text is saying.

  The ACPS Pointer qualifier contains a pointer to an Attribute
  Certification Practice Statement (ACPS) published by the AA.
  The pointer is in the form of a URI.  Processing requirements for
  this qualifier are a local matter.

3. Security Considerations

Spencer: there's a lot of general good operational advice for AAs on page 4 -
but I'm wondering if the Security Considerations section of the ACP extensions
specification is the right place to publish this information (it's not where
*I* would have first looked for operational BCP)... perhaps in its own
document, called "Recommended Operational Practices for Certificate Attribute
Authorities"?

  If an AC is tied to the holder's PKC using the baseCertificateID
  component of the Holder field and the PKI in use includes a rogue
  CA with the same issuer name specified in the baseCertificateID
  component, this rogue CA could issue a PKC to a malicious party,
  using the same issuer name and serial number as the proper
  holder's PKC.  Then the malicious party could use this PKC in
  conjunction with the AC.  This scenario SHOULD be avoided by
  properly managing and configuring the PKI so that there cannot be
  two CAs with the same name.  Another alternative is to tie ACs to
  PKCs using the publicKeyCert type in the ObjectDigestInfo field.
  Failing this, AC verifiers have to establish (using other means)
  that the potential collisions cannot actually occur, for example,
  the CPSs of the CAs involved may make it clear that no such name
  collisions can occur.

Spencer: "ACPS" is spelled out, earlier in the document, but CPS is not - is
this more or less the same thing?

  Implementers MUST ensure that following validation of an AC, only
  attributes that the issuer is trusted to issue are used in
  authorization decisions.  Other attributes, which MAY be present
  MUST be ignored. AC verifiers SHALL support means of being provided
  with this information.  The AA controls PKC extension is one
  possibility, but is optional to implement.

Spencer: got a reference for "the AA controls PKC extension", or do all skilled
in the art know where this is defined? ...

Spencer: from idnits 1.84

...
   Checking conformance with RFC 3978/3979 boilerplate...

 * Found RFC 3978 Section 5.4 paragraph 1 boilerplate (on line 324), which
   is fine, but *also* found RFC 2026 Section 10.4C paragraph 1 boilerplate on
   line 34. It should be removed.
 * There is 1 instance of too long lines in the document, the longest one
   being 2 characters in excess of 72.

[Scott Hollenbeck]

Comment (2006-01-03)

It might be helpful if the PKC acronym were expanded on first use in the
Introduction.