[cryptography] ssl/tls splicing attack

Chris Palmer chris at noncombatant.org
Wed Mar 17 00:34:19 EDT 2010


James A. Donald writes:

> In short, none of the three major protocols actually work, all of them
> need complete architectural redesign and full replacement.

As soon as someone creates a fully working and wonderful WEP/WAP/link-layer
gadget, we'll all complain that we really need end-to-end security anwyay,
because we do. Why bother even talking about the link layer? I don't get it.

Any reason you didn't list SSH v2 as successful deployed cryptography?

Also, I'd posit that the true truename of an HTTPS endpoint is the origin,
as defined by the browser
(http://code.google.com/p/browsersec/wiki/Part2#Same-origin_policy) ---
never mind the path. Origins are still tricky for people to understand, but
in their common form, e.g. https://www.wellsfargo.com, they are not *that*
weird. (And some browsers like Chrome and IE try to make this distinction in
the location bar.) When you get into kinky stuff like
http://www.wellsfargo.com@completely-sick-and-wrong.co.us.nz:443, people can
easily get confused. Incidentally, IE has dropped support for the bad old
username:password at host:port syntax...

The biggest security problems are in UI and in browser policy right now.
(And maybe the presistent belief among developers that HTTPS is "slow"
"because of all the crypto". Nice to see SPDY jumping into that debate on a
good footing.) The cryptography is far from the most important problem
facing us. Perhaps we should all stop debating here and join one of the
usability mailing lists and get jobs with browser vendors...




More information about the cryptography mailing list