#3916: Mutt 1.8: TOFU host certificate not working
-----------------------+----------------------
  Reporter:  kratem32  |      Owner:  mutt-dev
      Type:  defect    |     Status:  new
  Priority:  major     |  Milestone:  1.8
 Component:  crypto    |    Version:
Resolution:            |   Keywords:  tofu
-----------------------+----------------------
Comment (by m-a):

 Replying to [comment:8 kevin8t8]:
 > I wonder if the "error:0906D06C:PEM routines:PEM_read_bio:no start line"
 is a remnant from ssl_load_certificates() trying to read an empty cert
 file.  Perhaps we need to reset the err on rejecting.

 No, there's a place that should have looked at and purged the error queue,
 but forgot.

 > The new behavior is the default behavior using OpenSSL verification.
 This is definitely a change in behavior, but I'm not convinced it's a bug.
 It's not clear to me that it's worth preserving the previous behavior of
 being able to reject a piece in the chain and continue.   Matthias and
 MichaƂ I'd appreciate your opinion though.

 I don't see it as a bug either, but rather a missing special feature.  I
 acknowledge there is a point in providing **some** way for users to
 restrict what they want to trust, by cutting chains, but I wonder if we
 need to expose it through mutt's user interface.  Ultimately users claim
 to improve security by avoiding rogue or dodgy CAs, but I question that
 TOFU is any better.  If you see a certificate change, is that because (a)
 an eavesdropping has started, (b) an eavesdropping has ended, or (c)
 someone just reissued the host's certificate - which will be rather
 frequent if it's a Let's Encrypt certificate, or behind certain load
 balancing setups.  This looks like self-inflicted pain somewhat.  If we
 add something to the UI, it should be a "skip trusting this certificate".

--
Ticket URL: <https://dev.mutt.org/trac/ticket/3916#comment:10>
Mutt <http://www.mutt.org/>
The Mutt mail user agent

Reply via email to