On Mon, Jan 16, 2023 at 7:36 AM Alan DeKok <[email protected]>
wrote:

> On Jan 16, 2023, at 9:53 AM, Alexander Clouter <[email protected]>
> wrote:
> > I was wondering what to do with A-ID[1] (and everything around PAC-Info)
> but starting to figure that as you can stuff anything you want into the
> opaque SessionTicket it really does not matter.
>
>   I think so, yes.  For one, A-ID is defined as:
>
> The A-ID is the identity of the authority that issued the PAC.
>
>   Since the Authority-ID is already sent (or should be sent) in the outer
> TLVs, this A-ID seems often superfluous.  But that's another issue. :(
>
>
[Joe] The authority-ID was meant to be sent from the server to the client
so the client could choose the right PAC to use to establish the session
with the server.  I don't think there is an equivalent for TLS tickets, so
perhaps it could be useful in this case.  However, since it only applies to
PAC, I would suggest that it shares the fate of the PAC text unless there
is a need for it.  A mechanism can be added later if the need arises.




> > Not aware this has even been implemented by anyone even for RADIUS
> (D)TLS yet, right?
>
>   FreeRADIUS supports TLS-PSK for both inbound and outbound
> RADIUS/(D)TLS.  It also supports TLS-PSK for EAP-TLS, and any other
> TLS-enabled EAP types.
>
> > Maybe time to hammer out a TLS-PSK implementation just to see what this
> looks and smells like? Previous RFC's have described that TLS-PSK is
> supported but someone here (or was it radext) pointed out that there is no
> mention of a PSK Identity which is needed not just the keying material.
>
>   Yes, that's been discussed in RADEXT.  (D)TLS-bis should describe how to
> do PSK.
>
> > Maybe I am mis-reading it but I think TLS-PSK has effectively been given
> the kibosh by RFC9190 section 2.1.1: "Pre-Shared Key (PSK) authentication
> SHALL NOT be used except for resumption"?
>
>   That's for EAP-TLS, and not for other EAP methods.  EAP-TLS has no other
> way to authenticate users, whereas TEAP could use additional methods inside
> of the TLS tunnel.
>
>   But it is a good reason to frown on TLS-PSK in TEAP, I think.
>

[Joe]  I can imagine use cases where TLS-PSK is used to establish a tunnel
for the initial provisioning of credentials,  but it might be good to limit
its use until a need is explicitly defined.


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

Reply via email to