I made all of the suggested changes except one as noted below.

On 2/17/16, 1:55 PM, "Elwyn Davies" <[email protected]> wrote:

><snip>
>
>s1, para 2: It would be useful to provide a reference for Certificate
>Signing Requests, as well as clarifying that CSR is the relevant acronym
>- one possibility might be "Certificate Signing Request (CSR; See, for
>example, CertificationRequestInfo in [RFC2986])".

I think all that was missing here was placing (CSR) following the first
use of Certificate Signing Request. In context, EST is the relevant spec.
I added (CSR) after the first use of Certificate Signing Request in the
first paragraph and replaced the second use of Certificate Signing Request
with CSR in the second paragraph to tie these references together. The
first two paragraphs now read:

PKCS (Public-Key Cryptography Standards) #9 [RFC2985] defined a
   challengePassword attribute that has been overloaded by modern
   protocol usage with the appropriate interpretation being provided by
   context rather than OID definition.  PKCS #9 defines the
   challengePassword attribute as "a password by which an entity may
   request certificate revocation".  The parsing and embedding of this
   attribute within Certificate Signing Requests is well supported by
   common PKI tool sets, but many work-flows leverage this supported
   field as a one-time password for authentication.  For example this is
   codified in many Simple Certificate Enrollment Protocol (SCEP)
   implementations as indicated by [I-D.gutmann-scep].  Continuing this
   trend, Enrollment over Secure Transport (EST) [RFC7030] defines an
   additional semantic for the challengePassword attribute in Section
   3.5, in order to provide a linking of the Certificate Signing Request
   (CSR) to the secure transport.

   Where the context of the protocol operation fully defined the proper
   semantic, and when only one use was required at a time, the
   overloading of this field did not cause difficulties.  Implementation
   experience with EST has shown this to be a limitation though.  There
   are plausible use cases where it is valuable to use either of the
   existing methods separately or in concert.  For example an EST server
   might require the client to authenticate itself using the existing
   client X.509 certificate, the user's username and password and to
   include a one-time password within the CSR all while maintaining
   identity linking to bind the CSR to the secure transport.  The
   overloading of a single attribute type should not be the limiting
   factor for administrators attempting to meet their security
   requirements.


>
><snip>


_______________________________________________
Gen-art mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/gen-art

Reply via email to