On Tue, 2006-12-26 at 11:21 -0500, DKIM Chair wrote:
> In discussions with the IESG to sort through their "discuss" comments, 
> I had a talk with Lisa Dusseault, and she had one point that I want to 
> bring back to the mailing list:  I don't think we considered, in our 
> discussions of multiple signatures, multiple *linked* signatures, which 
> could work TOGETHER to convey information, and the protocol doesn't 
> allow that sort of thing.  The way dkim-base is set up, I don't think 
> this could easily be added as an extension, and it'd be a significant 
> change at this point.  Here's the concept:
> * Signer puts on two signatures (maybe as two header records, maybe as 
> one that contains two sigs).

There is a situation that could lead to a DoS related to multiple
signatures.  This problem would be in conjunction with a discovery
process checking associations of an email-address with that of a list of
signatures contained within a message.  The signature itself offers
little assistance in determining which email-address induced the
signature when the signing-domain differs from that of the email-address
domain.

Lack of assistance in a signature selection process in this case is due
to a limitation placed upon the 'i=' syntax.  This syntax can be
upwardly changed to allow explicit associations asserted between an
email-address-in-question and any signature.  It will remain apparent
when the email-address domain and the signing-domain differ.  With older
implementations, such a signature will be invalid and ignored.

The basis for the 'i=' syntax limitation assumes policy blocks messages
not signed by a parent domain.  This concept is based upon visual
recognition and remains highly flawed for many reasons.  Once MUA
"recognition" of email-addresses based on Address-Books or trusted lists
is used, then extending the 'i=' syntax makes a great deal of sense.
Having the MUA preform verification also overcomes requiring a trust
relationship between the MDA and the MUA, while also thwarting common
look-alike attacks.


> * One of the signatures has minimal scope, maybe signing only "from:",
>   with l=0.
> * The other signature covers as much of the message as possible...
>   most headers, all the body.
> * The two signatures work together.  If one verifies and the other 
>   doesn't, the verifier can consider what was changed in the message,
>   and possibly use that information to deal with mailing list
>   modifications or whatnot.

There is also the body hash that can provide a similar function.  The
use of multiple signature will likely occur, but not for this reason.
At least there should not be any valid reason for using the 'l=0'
parameter. 

> One way this might be used is to have one signature that covers the 
> subject header and one that doesn't, to allow the verifier to detect a 
> subject change and decide whether it's OK.  As the spec is now, the 
> verifier would just find the one signature (that doesn't cover the 
> subject) that works, and use that, not considering the other.
 
A change to the limitations on the 'i=' syntax could clarify the
specific role of the signature, and prevent various DoS exploits that
might attempt to overload the resource intensive validation process.  An
'i=' syntax change would be beneficial when validations are limited to
"recognized" email-addresses.  Such a change would limit the number of
bogus signatures checked.  As it is now, any or all signatures would
need to be evaluated for a possible relationship.  Header ordering
should not be assumed.

This issue will prove more significant when the MUA or MDA is used to
annotate "recognized" messages.  Annotation is the only method where
recipients are offered protection by DKIM.  Allowing the email-address
domain and the signing-domain to differ also retains the normal use of
email.  Without there being any correlation between the signing-domain
and the transmitting entity, abuse-reporting will be far less useful.

Before DKIM improves upon email integrity, an ability to associate
various email-originating domain with that of the signing-domain still
needs to be developed.

-Doug


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

Reply via email to