ietf-dkim
[Top] [All Lists]

Re: [ietf-dkim] RFC4871bis - whether to drop -- k: Key type

2009-06-02 20:34:34

On Jun 2, 2009, at 3:36 PM, Murray S. Kucherawy wrote:

Without this feature, people may soon find their inbox flooded by  
bogus messages indicating the use of new algorithm, that could have  
been mitigated extensively by having the key feature.

As opposed to what?  What would you expect a verifier or assessor to  
do if the hash used to sign was not in the key's approved hash list?

It would be reasonable to assume the lack of a matching algorithm  
within the key record would be a strong indication of a spoofed  
signature.  The duration before new algorithms are universally  
supported could be fairly long.

 Wouldn't it get delivered anyway, but perhaps with a slightly  
different annotation?

It would be preferred to have "Unsupported Signature Algorithm"  
reserved for domains that indicate they actually use the unsupported  
algorithm.  Without the key being able to make this indication, all  
such messages might then receive the same indication, or be rejected  
where the recipient could have taken steps that check the signature's  
validity independently from the non-updated MTA.

I don't see any value here other than disabling verification using a  
known-compromised hash algorithm.  But even that wouldn't inhibit  
delivery, only change annotation.

Huh? This is not about known compromised hash algorithms.  This is  
about asserting the proper annotation for new algorithms not yet  
supported by the receiver.

When a message is received that has the signature:
   DKIM-Signature: v=1; a=ecdsa-md6;  d=example.com; s=sel; ...

Annotation A "Signature algorithm not supported by receiver" can be  
noted when the key record is:
  sel._domainkey.example.com. TXT "k=edcsa\;. h=md6\; ...


Annotation B "Signature algorithm not supported by either the sender  
or the receiver!" can be noted when the key record is:
  sel._domainkey.example.com. TXT "k=rsa\;. h=sha256\; ...

By being able communicate what a sender supports for a specific key  
limits the number of domains that might receive the Annotation A  
notation or some ambiguous signature failure indication.

When the message receives the annotation: "Signature algorithm not  
supported by either the sender or the receiver!" this should cause the  
message to be treated with prejudice where users will know that  
additional verification steps will prove futile.

Not having this feature will mean users are unable to differentiate  
between the Annotation A or Annotation B cases.  Being unable to  
communication this significant difference allows bad actors to benefit  
by the confusion thereby created.

Without this feature being in place, there will be a greater  
reluctance to upgrade DKIM algorithms due to the potential for  
confusion.

-Doug



_______________________________________________
NOTE WELL: This list operates according to 
http://mipassoc.org/dkim/ietf-list-rules.html

<Prev in Thread] Current Thread [Next in Thread>