[nfsv4] Id-sha1 as REQUIRED hashing algorithm

[David Noveck <[e-mail]>]

In following up on the comments on rfc5661sesqui, I've been reminded of the
fact that reviews of a previous iteration of the document had raised the
issue of the (server-) REQUIRED status of id-sha1 as a hashing algorithm in
rfc5661.   It turns out that I indicated that the working group woud need
to be consulted about potential compatibility issues and then just forgot
about it :-(.

If anyone knows of compatibility issues or is concerned about the
possibility of them arising from dropping this as hash algorithm for which
server support is REQUIRED, please let me know as soon as you can.
 Although I could be wrong, it seems to me that such compatibility issues
could arise only if a client were to do an EXCHANGE_ID with ssp_hash_alg<>
including id-sha1 and no other server-REQUIRED hash algorithm.

People in the security directorate are understandably concerned about
approving a Proposed Standard specifying, as REQUIRED, server support for
hashing using a hashing algorithm for which the generation of collisions
has been demonstrated.  However, given the limited scope of the review for
this document (according to the rules in draft-roach-bis-documents), we
could defer this to rfc5661bis, which I would prefer to do, since I think
it is better to address NFSv4's security weaknesses together rather than in
a piecemeal fashion.   Nevertheless, I expect this issue to keep coming up,
and I want to be clear whether the potential compatibility issue is of any
concern.