Dan Harkins wrote:
>   Is it the intention of the community that each individual EAP method
> will have to define ciphers to negotiate, how to negotiate a cipher,
> how to get a key for the cipher, how to use the cipher to protect a new
> payload that contains arbitrary TLVs? I personally think that's a bad
> idea. Each EAP method would have to basically duplicate a whole bunch
> of capabilities.

  I agree that it's a bad idea.  Where additional data needs to be
transferred, it would be best to have a way that is EAP method agnostic.

>   Would it not be better to do this with two new EAP codes to pass TLVs
> in each direction and the have a single definition of the cipher and a
> single definition of how to get the key and a single definition of how
> to use the cipher and key to protect packets with these new EAP codes?
> That leaves EAP methods to do authentication and generation of the MSK
> and EMSK, period. The (optional) exchange of packets with these new EAP
> codes would happen after the EAP method has finished sending requests
> and responses and before "success" is declared.

  There are multiple ways of doing this.  That is one approach.

>   Disregarding the EMU charter for a minute, would this be a better
> architectural solution?

  Yes.

  Alan DeKok.
_______________________________________________
Emu mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/emu

Reply via email to