Re: [dhcwg] preliminary comments on draft-ietf-dhc-sedhcpv6-17

[神明達哉 <[e-mail]>]

At Tue, 15 Nov 2016 00:16:41 +0800,
Lishan Li <[e-mail]> wrote:

> >> Alternatively, we might add both an EA-id and SA-id fields to the
> >> option:
> >>
> >>     0                   1                   2                   3
> >>     0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
> >>    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >>    |      OPTION_CERTIFICATE       |         option-len            |
> >>    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >>    |            EA-id              |            SA-id              |
> >>    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >>    |                                                               |
> >>    .                  Certificate List(variable length)            .
> >>    |                                                               |
> >>    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
> >> (I'm not sure if this has to be a list of certificates instead of one
> >> certificate, but that's a different question).
> >
> >
> >> And we can use a value of 0 for EA-id and SA-id to mean this
> >> certificate is not supposed to be used for encryption and signature,
> >> respectively.  (The combination of 0, 0 makes no sense so we should
> >> probably prohibit the use of it explicitly).
> >>
> > [LS]: So, there is no need to define a new field. The EA-id and SA-id are
> > used to identify the certificate type.

If you use this approach, correct.  Note, however, that I'm not
necessarily pushing a particular approach at this time.  I'm just
showing various possible approaches for discussion.  You may want to
consider pros and cons of these approaches yourself.

> > And the certificate list field should be changed to certificate field. If
> > multiple
> > certificates are contained, then multiple certificate option is contained.

Regarding this, I'm not sure.  Basically the choice of EA-id and SA-id
should be fixed (it should be the most preferred one for the server,
and it's known that the client supports it), so the question is
whether we want to make it possible to include multiple different
certificates for that combination of EA and SA.  That sounds to me
like an unlikely scenario in practice, although being flexible/generic
itself is not necessarily bad.

> [LS]: In this way, the certificate option and Signature option all contain
> the SA-id field. And the content of the two SA-ids are same.

Yes (otherwise the recipient should treat it as an error condition).

BTW this makes me notice one related issue: it doesn't seem to be
possible for a server to identify the private key to decrypt the
message contained in an Encrypted-message Option contained in the
Encrypted-Query message unless it tries all private keys it might be
used.

--
jinmei