Re: [secdir] Secdir review of draft-ietf-tls-record-limit

[Martin Thomson <[e-mail]>]

On Mon, Mar 5, 2018 at 2:57 AM, Hannes Tschofenig
<[e-mail]> wrote:
> [Hannes] I am not sure I fully understand the approach. Even if someone uses TCP a RAM limitation will not go away. TCP is likely to make the situation worse since it requires some amount of RAM as well even with the best possible configuration. (There is this work in LWIG on TCP implementation guidance where the authors are supposed to provide some information about the actual RAM usage of various TCP features and settings.)

Let me try to explain more.

An endpoint that can't handle a big packet (whether that be a big TCP
segment or a big UDP datagram) can always pretend that this is a link
issue and send the appropriate ICMP message.  That limits the size of
packets they receive, at least to the point that the minimum for the
IP version in use is reached (576 for v4, 1280 for v6).  I'm not sure
what fragmentation does here, other than make things far worse.

If there are multiple packets arriving and no space for more than one,
then the endpoint can pretend that it didn't receive the packet.  TCP
also has ACKs, which the endpoint can withhold until it has space.

Thus, the endpoint can hold as little as a single MTU of data at once.
Given that it is not encrypted, my understanding is that there is no
need to constrain how it is turned into records.

Note that a constrained server has to handle a full ClientHello - the
client won't know the limit at the server.