On Fri, Aug 1, 2014 at 2:25 PM, Nico Williams <[email protected]> wrote:

> On Fri, Aug 1, 2014 at 4:14 PM, Eric Rescorla <[email protected]> wrote:
> > On Fri, Aug 1, 2014 at 2:11 PM, Nico Williams <[email protected]>
> wrote:
> >> +1.  Above all: integrity protection for the entire pair of data octet
> >> streams.
> >>
> >> Required as an option, if not alway: confidentiality protection
> >> (encryption).
> >>
> >> Obviously required: protection for any TCP options where not
> >> protecting them implies failure to protect the data streams.
> >
> > Can you elaborate here?
>
> No, it's a corollary to the "integrity protection for the entire pair
> of data octet streams".  I could have left it out as implied.  I could
> go look at all the TCP options and make a list of which might need
> protection and which might not, but that should be unnecessary at this
> point: we're stating high-level requirements.
>
> >> Highly desirable: integrity protection for close/ EOF / RST.
> >
> > For reasons that people have already gone onto on the list,
> > I think this minimally needs to be optional.
>
> Perhaps so.  If middleboxes make that too hard then yes, it should be
> optional, and probably default to off.  (Though middlebox presence
> could be detected by exchanging hashes of the SYN handshake, no?)
>

The issue isn't middleboxes; it's that if you require integrity protection
for RSTs, then there's no way for a box that reboots to send you an
RST.

-Ekr
_______________________________________________
Tcpinc mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tcpinc

Reply via email to