Re: [Curdle] I-D Action: draft-ietf-curdle-ssh-ed25519-ed448-01.txt

[Daniel Migault <[e-mail]>]

Hi,

In addition to Mark's comment, please find my review of the document.
Comments are indicative, so you may find better ways to address them than
what I suggest. Please update the next version by the end of August.

Yours,
Daniel






  Ed25519 and Ed 448 public key algorithms for the Secure Shell (SSH)
                                protocol
                 draft-ietf-curdle-ssh-ed25519-ed448-01

[...]
1.  Introduction

   Secure Shell (SSH) [RFC4251] is a secure remote-login protocol.  It
   provides for an extensible variety of public key algorithms for
   identifying servers and users to one another.  Ed25519 [RFC8032] is a
   digital signature system.  OpenSSH 6.5 [OpenSSH-6.5] introduced
   support for using Ed25519 for server and user authentication.
   Compatible support for Ed25519 has since been added to other SSH
   implementations. Ed448 [RFC8032] is another digital signature
   system.

<mglt>
I am reading the sentence "Compatible support" as variants, and a such the
current document may describe on eor the other variant. I am not sure the
Ed448 sentence adds much information. I would rather propose the following
wording:

OLD:
 OpenSSH 6.5 [OpenSSH-6.5] introduced
   support for using Ed25519 for server and user authentication.
   Compatible support for Ed25519 has since been added to other SSH
   implementations. Ed448 [RFC8032] is another digital signature
   system.

NEW:

 OpenSSH 6.5 [OpenSSH-6.5] introduced
   support for using Ed25519 for server and user authentication and was
then followed by other implementations.
</mglt>

   This document describes the method implemented by OpenSSH and others,
   and formalizes its use of the name "ssh-ed25519".  Additionally, it
   also describes the use of Ed448 and formalizes its use of the name
   "ssh-ed448".

   [TO BE REMOVED: Please send comments on this draft to
   [e-mail].]

2.  Conventions Used in This Document
3.  Public Key Algorithm
4.  Public Key Format
5.  Signature Algorithm
6.  Signature Format
7.  Verification Algorithm
8.  SSHFP DNS resource records

<mglt>
The main purpose of this section is the IANA registration for the "SSHFP RR
Types for public
   key algorithms" registry. I would thus start the section as follows:

Usage and generation of SSHFP DNS resource record is described in RFC 4255.
The generation of SSHFP resource records for "ssh-ed25519" keys is
described in [RFC7479]. This section illustrates the generation of SSHFP
resource records for "ssh-ed448" keys and the document specifies the
corresponding Ed448 code point to the "SSHFP RR Types for public key
algorithms" IANA registry.
</mglt>

   The generation of SSHFP resource records for "ssh-ed25519" keys is
   described in [RFC7479].

   The generation of SSHFP resource records for "ssh-ed448" keys is
   described as follows.





Harris & Velvindron      Expires January 3, 2019                [Page 3]

Internet-Draft               Ed25519 for SSH                   July 2018


   the SSHFP Resource Record for the Ed448 public key with SHA-256
   fingerprint would be example be: ssh.hackers.mu IN SSHFP 5 2 (
   a87f1b687ac0e57d2a081a2f2826723 34d90ed316d2b818ca9580ea384d924 01 )

<mglt>
probably having the DNS in a figure would ease the reading.
I believe it would also be nice to have the key used to generate the
example.
5 is 'To be defined' TBD
Maybe it would be good to also indicate that 2 indicates the SHA2 as a
complete example. (rfc6594).
</mglt>

9.  IANA Considerations

   This document augments the Public Key Algorithm Names in [RFC4250],
   Section 4.6.2 [RFC4250].


   IANA is requested to add to the Public Key Algorithm Names registry
   [IANA-PKA] with the following entry:

<mglt>
It is also better to indicate the type of review. In this case the addition
of the code point requires an IETF review.
</mglt>
                   Public Key Algorithm Name Reference
                   ------------------------- ----------
                   ssh-ed25519               This Draft
                   ssh-ed448                 This Draft

   IANA must add the following entry to the "SSHFP RR Types for public
   key algorithms" registry:
<mglt>
is requested may be better ;-)
The registry got a Public Key Format column. This should also be mentioned
in this section.
</mglt>

   +--------+-------------+------------+

   | Value | Description | Reference |

   +--------+-------------+------------+

   | 5 | Ed448 | [this-draft]  |

   +--------+-------------+------------+

   [TO BE REMOVED: This registration should take place at the following
   location: <http://www.iana.org/assignments/ssh-parameters/ssh-
   parameters.xhtml#ssh-parameters-19>]

<mglt>
Maybe that would be good to have the URL as an informational reference. The
reference is for the previous registry ( i.e. Public Key Algorithm name
reference.

The latest is
https://www.iana.org/assignments/dns-sshfp-rr-parameters/dns-sshfp-rr-parameters.xhtml

The review is also an IETF review.

Though 5 is the most likely value. It might be better to have it as TBD and
then suggest TBD to be 5 to the IANA.
</mglt>

10.  Security Considerations

   The security considerations in [RFC4251], Section 9 [RFC4251] apply
   to all SSH implementations, including those using Ed25519 and Ed448.

<mglt>
RFC6469 may also be references here.
</mglt>

   The security considerations in [RFC8032], Section 8 [RFC8032] apply
   to all uses of Ed25519 and Ed448 including those in SSH.



On Fri, Jul 6, 2018 at 5:43 PM, Mark D. Baushke <[e-mail]> wrote:

> The https://tools.ietf.org/html/draft-ietf-curdle-ssh-ed25519-ed448-01
> looks better than the -00 revision.
>
> A very minor correction should be the table in section 9 which adds the
> SSHFP RR Type value 5 for "Ed448" for SSHFP RR Types for public key
> algorithms.
>
> I believe there should be an informative reference to [IANA-SSHFP]
> https://www.iana.org/assignments/dns-sshfp-rr-parameters/dns-sshfp-rr-
> parameters.xhtml#dns-sshfp-rr-parameters-1
>
> which is the IANA.ORG parameter table that should be updated.
>
>         Thank you,
>         -- Mark
>
> _______________________________________________
> Curdle mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/curdle
>