[Top] [All Lists]


2008-03-18 15:02:15

I would prefer not to say anything at all.  But that may not be realistic.

I do understand Jari's concern that there *could* be some issues if no used 

I think the main objection as I understand it is that Network Access 
Authentication should be decoupled for Application Authentication.  That is, if 
we use the EMSK derived from Network Access Authentication to derive keys for 
applications we are coupling the two authentications.

This can be bad as Jari puts it because those applications that are coupled to 
one Network may not work when the user is connect to another network.

I agree but not generally. That is, there are some applications for which 
coupling makes perfect sense.  Applications that strictly run within the 
context of a Network Access session could benefit by having their application 
keys bootstrapped from EMSK.

For example, Over-the-Air provisioning application and location based 
applications that are part of the network.

Bootstrapping the keys from EMSK creates a single-sign-on and elliminates the 
need to have to provision keys for each of the applications and elliminates the 
need to run extra messaging to create session keys etc.

Application like a web server may not be appropriate to have its keys derived 
from EMSK because these application are network agnostic.

So we need to say something.  But that something needs to indicate the issues 
so that SDOs understand when to use the EMSK thingy and when not to use it.  I 
do not support making blanket statements such as Application MUST not derive 
their keys from EMSK.

Also note that there are ways to use the EMSK even in the case of the 
Applications that are access network agnostic.  And we should not block this 

-----Original Message-----
From: hokey-bounces(_at_)ietf(_dot_)org 
On Behalf Of Glen Zorn
Sent: Tuesday, March 18, 2008 11:31 AM
To: Charles Clancy
Cc: ietf(_at_)ietf(_dot_)org; hokey(_at_)ietf(_dot_)org; Bernard Aboba
Subject: Re: [HOKEY] EMSK Issue

Charles Clancy <> scribbled on :


From Bernard's "walled garden" LC comments, I've created
the following
issue below in the issue tracker.

Some of us don't subscribe to the IETF list (due to the
extremely poor S/N ratio).  Someone did forward me Bernard's
original message & to me it appears to fall squarely into the
N category (either that or it is an early April 1 RFC
candidate).  I understand, though, that it is actually
receiving serious discussion on the IETF list, so I'm happy
that you are bringing some of that discussion to this forum.
Of course, common courtesy would have required that the WG
the work of which is being disparaged in outrageous fashion
be included in the discussion but courtesy seems to be in
short supply.
EMSK: applicability statement, scope

The EMSK document, as-is, allows (or more precisely does not
disallow) for broader use of keys than it should.

That may be somebody's claim; however, RFC 3748 says:

   Extended Master Session Key (EMSK)
      Additional keying material derived between the EAP client and
      server that is exported by the EAP method.  The EMSK is at least
      64 octets in length.  The EMSK is not shared with the
      authenticator or any other third party.  The EMSK is
reserved for
      future uses that are not defined yet.

This doesn't appear to me to constrain the uses of the EMSK;
furthermore, since Bernard is not just one of the authors of
3748 but also a co-Chair of the working group that published
it, it seems like the time & place to constrain the usage of
the EMSK would have been during the drafting of the RFC and
in the eap WG.

There could
be interoperability issues if applications require
EAP-generated keys,
breaking the layering of the Internet protocol stack.

What does that mean?  How can _keys_ cause layer violations?

Some sort of statement regarding applicability and scoping
of derived
keys is necessary for this document.

t. charles clancy, ph.d.       
electrical & computer engineering, university of maryland
HOKEY mailing list
HOKEY mailing list

IETF mailing list

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