Dear OpenSSL Developers,

I am somewhat involved with the BetterCrypto(.org) project that tries to
provide the operations community with a BCP for daemon settings,
references and other recommendations. We've discovered an inconsistency
that could be called a flaw starting with OpenSSL 0.9.7m ending at
1.0.0a. My best guess is this is already known, but since I could not
find an appropriate post to this ML I figured I should send notice anyway.

In particular, given our cipherstring recommendation we encounter that
DHE and ECDHE based ciphersuites and their preference are neglected by
these OpenSSL versions [0] - we are currently discussing updating our
recommendation to an augmented version of this ciphersuite [1].

Can you confirm this is an issue in the OpenSSL codebase? Your release
notes state "Various ciphersuite selection fixes." for 0.9.7l to 0.9.7m.
And if so - what would be the next steps? I am aware that changing this
behavior this might cause problems with legacy software.

Here you can find a (scripted) comparison of a handshake negotiation for
a given OpenSSL version (s_server & s_client):
http://nopaste.narf.at/show/bY1GlbqLyNZhUO27MG5B/

Extending our cipherstring (to a somewhat insane length - that will
cause software configuration problems in daemons - e.g. OpenVPN) yield
the following result:
http://nopaste.narf.at/show/p2uTJ8Gi7RGpBktp6nTr/


Thanks for your time,
Aaron

[0] https://github.com/puppetlabs/puppet/pull/2494#issuecomment-41350861
[1] http://lists.cert.at/pipermail/ach/2014-May/

Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to