On 2010-08-05 14:34, Aleksi Suhonen wrote: > Hi, > > Remi Despres a ecrit: >>> If this this approach is retained, I could contribute on detailed >>> changes to RFC 3679, with whoever is interested. > > Steven Blake wrote: >> I agree with this in principle, but there are still a few issues: >> >> - If the sending host sets FL=0, and an intermediate router resets it >> non-zero, the receiving host cannot determine whether the sending host or >> an intermediate router set the FL. This may break some e2e >> applications of >> the FL. > > Actually, having had a couple of nights more to sleep thinking > about this, I have a couple more questions regarding this argument: > > > If the sending host sets FL=0, then the receiving host is clearly > not even expecting anything special in that field? > > If some e2e application is using the FL, it will surely be using some > signalling method associated with the FL values? > > So when a host receives an FL that doesn't match earlier signalling, > it can determine that it was set by an intermediate node, and that > the original FL was zero?
That doesn't prevent a case where the intermediate node, for whatever reason (including MITM), sets a label in the format that does match the signalling. We don't know what the signalling is like... > > > Is there an internet-draft out there that would answer anything > other than "yes" to the above questions? I think this only works if we assign at least one bit in the label to mean 'e2e or mutable'. And even that is not immune to MITM forgery. > > Can it be altered with reasonable effort so that it too would > yield "yes"es? I am pessimistic at the moment, which is why I suggested we have a binary choice. Brian -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------