[core] Request-Tag and Block2 with present but unfragmented payload draft-ietf-core-attacks-on-coap-03.txt

[Christian Amsüss <[e-mail]>]

Hi group,

Jon's mail at [1] triggered some re-reading of block-wise transfer, and
made me realize an error in some assumptions underlying 9175: I was
assuming that if the request body fit in a single payload, that payload
would be repeated in the Block2 phase.

Apparantly that is not the case; this also makes me moderately unhappy
with the state of block-wise because it means that a FETCH can not be
processed as statelessly as block-wise promised.

I have no clear plan for mitigation yet; my current assumption is that
while the mechanisms of Request-Tag still work (especially the parts
that enable the Request-Tag to almost never be sent), it would not only
need to apply to all requests where a Block1 is set, but to all that can
possibly have fragmentation on either side. Further consideration
needed; also, I don't yet know whether that will fit the format of an
erratum or will need more.

BR
c

who is now even more motivated to evolve block-wise to fix current
quirks and have some features of Q-Block without its caveats, but also
doesn't see when that would happen.

[1]: https://mailarchive.ietf.org/arch/msg/core/KKgoaU8B3T8mmtACKCvlx8mKvPw

-- 
To use raw power is to make yourself infinitely vulnerable to greater powers.
  -- Bene Gesserit axiom