To clarify, our interpretation of the spec was that it is the encrypted
data, not unencrypted data.

Separately, we limit unencrypted data that we skip over, as I mentioned in
that thread. These are not the same units as max_early_data_size are
currently specified, so we intend to advertise a smaller value in the
ticket to compensate for this difference.

David

On Mon, Mar 6, 2017 at 1:06 AM Martin Thomson <[email protected]>
wrote:

> (Adding Filippo, who wrote the original change.)
>
> I just did some spelunking of the archives, and poking at boring SSL.
> I found that David Benjamin mentions unencrypted data, which seems to
> be consistent with what boring implements:
>
> <https://mailarchive.ietf.org/arch/msg/tls/b5GpGR9QQpBV3tbxCspdHVJs8HU>
>
> I don't think that this makes sense: the true cost to the server is in
> the data it has to store, not process (a client has many better
> options for causing the server to expend CPU resources).  Any data
> that can be ignored is cheap.
>
> On 6 March 2017 at 11:17, Martin Thomson <[email protected]> wrote:
> > The section on the maximum early data size says this:
> >
> > "Only Application Data payload is counted."
> >
> > I don't know how to interpret that.  I can see arguments for counting
> > TLSInnerPlaintext.content or all of TLSInnerPlaintext.
>
> _______________________________________________
> TLS mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/tls
>
_______________________________________________
TLS mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tls

Reply via email to