Re: [dnsext] Clarifying the mandatory algorithm rules

[Matthijs Mekking <[e-mail]>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

On November 30, Wouter acknowledges that changes need to be made to the
Unbound implementation and asks for guidance:

  http://www.ietf.org/mail-archive/web/dnsext/current/msg10115.html

Presenting two options:
* One signature is enough (the lenient way)
* Check the algorithms.

But when checking the algorithms, thou should not use the DNSKEY RRset,
but the DS RRset.

I think the general consensus is that a validator should at least be
able to check the algorithms in the DS RRset (Please correct me if I am
being to hasty in my conclusion). There is still debate whether the
validator SHOULD or MAY do this (Ed Lewis argued the term 'could', I
think that translates to the RFC2119 term MAY).

 Proposed text would then be:

  "The validator SHOULD or MAY check (choice here) that the algorithms
   signaled in the DS-set work (but only for algorithms supported by
   the validator, of course)."

Best regards,

Matthijs




On 03/10/2011 11:34 PM, Mark Andrews wrote:
> 
> In message <[e-mail]>, Samuel Weil
> er writes:
>> On Tue, 30 Nov 2010, W.C.A. Wijngaards wrote:
>>
>>> It is clear that checking the set of algorithms present in the DNSKEY
>>> set is not a good idea, and checking the set of algorithms from the DS
>>> set is the right, more lenient way to go.
>>
>> I apologize for checking out of this discussion last fall.
>>
>> I would like the WG's help understanding where you want to go with 
>> this topic.  I don't fully understand the argument in favor of not 
>> checking the algorithms on the child side of the zone cut (= the ones 
>> in the DNSKEY RRset), nor am I sure that was the direction everyone 
>> seemed to want to go.  Could someone summarize the current state of 
>> this?
>>
>> My own inclination is (still) to treat this as a clarification, saying 
>> that validators are not required to enforce these rules.  (In other 
>> words, the extra checks Unbound did were just fine, though 
>> unnecessary.  BIND's lenient approach was also fine.)  Two specific 
>> pieces of proposed text can be found in the first message in this 
>> thread, dated 18 November 2010.
> 
> While we think about this ISC has also had bug reports claiming
> that is we don't publish DNSKEY prior to signing with them we are
> break RFC 4035 because it allows verification of every RRSIG as a
> policy and the only way to do that is to publish the DNSKEY prior
> to use.
> 
>    If other RRSIG RRs also cover this RRset, the local resolver security
>    policy determines whether the resolver also has to test these RRSIG
>    RRs and how to resolve conflicts if these RRSIG RRs lead to differing
>    results.
> 
>> -- Sam
>> _______________________________________________
>> dnsext mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/dnsext
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iQEcBAEBAgAGBQJNed35AAoJEA8yVCPsQCW52EcIALwitOh1IOVX6AqV28GNEYYM
VSSlPZyVWYxD0LCVTWq5aleh3EAJ3T4BdRio795mlNrc/vLi0VNL8FqRI8U7mYk8
I+acLNJGhoFpEFeVF+rOXpoZ3yfTerUctv1mmeyeO/2iW+PF++BJh67bcUV34h1p
UvP3lMZCBqIpbR3uhRITjF5JsZ2yaqwoARMyIw58MM72M2Y0X3IuJCYSKSl3ANzJ
nwBdrtZZcRmfmzJAY8PvDO0/mcyws1iK4JtiMMyXPNAfgxyBEeyxIe8YUR1yrQgL
tcxHaDyTsaPp1GnqAbqucYWdhr17QZ1b6SUmh9qyYW1leKXZSecFUp0tjdhCYQo=
=qOVC
-----END PGP SIGNATURE-----