On Jun 8, 2012, at 04:54, Tony Finch wrote:

> Matt Miller <[email protected]> wrote:
>> On Jun 7, 2012, at 15:12, Peter Saint-Andre wrote:
> 
> Glad we're all pretty much agreeing :-) One further question:
> 
>> If this specification is about DNSSEC with DANE goodness, then I do
>> think the A/AAAA verification is necessary, or at least recommended.
> 
> What advantage does the verification have, if the client is going to check
> the server certificate? (I'm assuming that bogus == DNS lookup failure, so
> the question is about secure vs. insecure.) The client is knows the SRV
> record is secure so the link from source domain to derived domain is
> secure. If the address records are insecure, the client needs to check a
> certificate to be sure the server is the right one; the client gets the
> same level of assurance whether it matches the source or derived domain.
> 

I was thinking of a case where a hosting provider ("hosting.example.net" in our 
text) has their DNS entries hijacked, and the hijacker finds a CA that will 
issue them certs for the hijacked name.  My thinking was this can be mitigated 
by verifying the A/AAAA records.

That was my thinking, but I'm less convinced it's worth covering now, and will 
probably remove in the next revision.


- m&m

Matt Miller - <[email protected]>
Cisco Systems, Inc.

Attachment: smime.p7s
Description: S/MIME cryptographic signature

Attachment: PGP.sig
Description: This is a digitally signed message part

_______________________________________________
dane mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dane

Reply via email to