eap Mailing List: Re: [Issue 278]: lifetimes of keys internal to EAP methods

[Jari Arkko (jari.arkko]

Bernard Aboba wrote:
> The resolution to Issue 261 involved distinguishing between
> transient EAP keys used directly to protect the EAP conversation
> (to be known as Transient EAP Session Keys or TESKs) and other
> keying material internal to the EAP method.  So I think we need
> to use the new language.

I do not see a reason to distinguish between the different
types of keys internal to a method beyond what we have already
done, such as making a difference between the long-term
credential and TEKs. If either endpoint of the EAP conversation
is compromised, a method with a fast reconnect support will
be compromised, no matter how many levels of key derivation
are used between the "semi-long term key" and the TEKs.

> I think the issue is not necessarily the lifetime of the key material, but
> session key reuse.  Note that the definition of "session" typically
> relates to when a replay counter can wrap, which could encompass multiple
> EAP sessions.

Is this reflected somewhere in your proposed text? Perhaps
we could avoid making a new requirement on the separation
of TEKs and fast reconnect keys, but instead have a requirement
that fast reconnect and other usage of TEKs should have
means for avoiding replay counter wraparound.

--Jari