On Jan 28, 2019, at 10:29 AM, John Mattsson <john.matts...@ericsson.com> wrote: > > I think you accidently took the key derivation from > draft-mattsson-eap-tls13-00.
Ah, yes. Every time I look for the draft, Google etc. seem to prefer the -02 for some reason. I'll remember to avoid that. > According to Section 6.2 of RFC 3748 the EAP Method Type values are between > 1-4294967295 (2^32 - 1) so while the Method Types of current TLS-based EAP > methods are a single byte, future TLS-based EAP methods could use 2 or more > bytes. Sure. The question then becomes one of encoding. For types < 256, 1 octet is enough. For others, it should be a 32-bit number in network byte order. That allows for other EAP types to use similar derivations. > https://tools.ietf.org/html/rfc3748#page-41 > https://www.iana.org/assignments/eap-numbers/eap-numbers.xhtml#eap-numbers-4 > > I think your suggestion of adding the EAP Method Type as context_value seems > like a good idea. draft-mattsson-eap-tls13 should then be changed to: > > Method_Type = 0x0D > Key_Material = TLS-Exporter("EXPORTER_EAP_TLS_Key_Material", Method_Type, > 128) > IV = TLS-Exporter("EXPORTER_EAP_TLS_IV", Method_Type, 64) > Method-Id = TLS-Exporter("EXPORTER_EAP_TLS_Method-Id", Method_Type, 64) > Session-Id = Method_Type || Method-Id Thanks. My only remaining nit here is that there should really be a sentence alluding to other TLS-based EAP methods. So we don't have to rev all of those documents, too. I'm not sure that this document is the best place to do it, but it's only 1-2 sentences. And I think we're not going to rev TTLS / PEAP / FAST in any reasonable time frame. Maybe: --- The inclusion of Method_Type in the above derivations is intended to enable the use of TLS 1.3 in other TLS-based EAP methods. While we do not explicitly update those methods here, TLS-based EAP methods SHOULD use the above key derivation for TLS 1.3, changing only Method_Type. Method_Types that are larger than 255 SHOULD use Method_Type as a 32-bit number in network byte order. --- Without such text, there will be problems. People will want to use TLS 1.3 with other EAP methods. And if there is no standards guidance, the implementors *will* choose something, so that meet real-world demand. Alan DeKok. _______________________________________________ Emu mailing list Emu@ietf.org https://www.ietf.org/mailman/listinfo/emu
