On Jul 23, 2025, at 11:26 AM, Alexander Clouter 
<[email protected]> wrote:
> For where topological information (eg. IP address assignment) is deemed to be 
> within scope.

  I would argue that IP address assignment should not be in scope.  That would 
go down the path of replacing DHCP, which seems a bit much to do.

> Only spit balling but maybe there is something we can instead do to extend 
> the TLS binding to carry over to the DHCP.
> 
> This would then no longer be limited to TEAP, you could maybe even retrofit 
> it to EAP-(T)TLS.

  TTLS would need to define a TTLS-specific attribute but sure.

> My thinking is the DHCP client would then include an attribute to the server 
> saying "I expect something binding here tied back to the TLS session of my 
> EAP dance" and the DHCP server would include it as an option in the response. 
> The client decide what to do in the presence (or non-presence) of it based on 
> a local policy.

  Hmm... I'll have to think about that.

> With this, you could now do topological (eg. IP) assignment and support DHCP 
> snooping.
> 
> With the proposal TEAP options approach, there would need to be some 
> additionally special OOB protocol between the switchport and your policy 
> server to communicate these DHCP assignments and make DHCP snooping work in 
> practice. Of course the other option is to leave this at "use this only for 
> assigning the WPAD server" :)

  Yes, it would be useful to send these options in the final Access-Accept, too.

  Alan DeKok.

_______________________________________________
Emu mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to