If we suppose all tickets are sent by the server at once, I am wondering if we have any case where the server will send more tickets that the number provided by the hint.
Yours, Daniel On Mon, Oct 7, 2019 at 10:46 AM Hubert Kario <hka...@redhat.com> wrote: > On Saturday, 5 October 2019 14:08:45 CEST Christopher Wood wrote: > > On Fri, Oct 4, 2019, at 6:17 AM, Hubert Kario wrote: > > > On Thursday, 3 October 2019 22:15:14 CEST Daniel Migault wrote: > > > > On Thu, Oct 3, 2019 at 7:56 AM Hubert Kario <hka...@redhat.com> > wrote: > > > > > On Wednesday, 2 October 2019 22:42:32 CEST Daniel Migault wrote: > > > > > > I understand the meaning of count is the higher limit of ticket > and > > > > > > the > > > > > > server can provides any tickets between 0 and count. If that is > > > > > > correct, > > > > > > this could be clearly stated and indication to chose an > appropriated > > > > > > > > > > value > > > > > > > > > > > for each cases may be provided. > > > > > > > > > > > > I would rather see the count value as a hard line from the server > > > > > > with a > > > > > > MUST NOT instead of a SHOULD NOT statement. > > > > > > > > > > see my previous comments: this ignores existence of post handshake > > > > > authentication, ticket lifetimes and KeyUpdate > > > > > > > > I think we agree on the problem which is it is hard to have a > specific > > > > count as tickets may be sent at multiple time during the key > exchange or > > > > the later during the session. If the "SHOULD" is addressing this > aspect > > > > I > > > > believe that more text is needed.. From my perspective, what I > proposed > > > > I > > > > imagined that count was related to the maximum number of ticket > provided > > > > **each time the server is sending tickets**. The reason for having > the > > > > same > > > > number is that the server does not know how the client will use these > > > > tickets and the client does not know precisely when the server sends > the > > > > tickets. So at the end the number of tickets sent is likely to be > > > > n*count > > > > when n represents the number of times during the session tickets are > > > > provided. > > > > > > > > My understanding of the SHOULD is that it makes legitimate for a > server > > > > to > > > > ignore the count value provided by the client. > > > > > > yes, it looks like we are in agreement > > > > > > Sounds like something that should be spelled out explicitly in the > draft. > > > I.e. it should talk about groups of tickets, not tickets in connection, > > > and it should definitely not specify the maximum number of tickets a > > > server can send in a connection. > > > > Since it's meant as a hint, removing that clause (maximum number of > tickets > > a server can send in a connection) is reasonable, and sounds like it > should > > address the concerns here. (Assuming we also include text which says > > servers should obviously place a cap on what they send, as they do > today.) > > Would you agree? I'm really trying to avoid this becoming overly complex, > > as it's a very small feature. > > yes, having a "server SHOULD place a cap on the number of tickets it sends > at > once" with a note in security section that "not having such a limit may > expose > the server to DoS or amplification attacks" would be enough > > (note that without client certificate authentication, the server can use > early > exit from handshake method and thus send the extensions just as a result > of a > single ClientHello message) > -- > Regards, > Hubert Kario > Senior Quality Engineer, QE BaseOS Security team > Web: www.cz.redhat.com > Red Hat Czech s.r.o., Purkyňova 115, 612 00 Brno, Czech > Republic_______________________________________________ > TLS mailing list > TLS@ietf.org > https://www.ietf.org/mailman/listinfo/tls >
_______________________________________________ TLS mailing list TLS@ietf.org https://www.ietf.org/mailman/listinfo/tls
