Horton

Re: SCRAM with channel binding downgrade attack

Bruce Momjian <bruce@momjian.us>

View thread View on postgr.es
From: Bruce Momjian <bruce@momjian.us>
To: Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
Cc: Heikki Linnakangas <hlinnaka@iki.fi>, Michael Paquier <michael@paquier.xyz>, Magnus Hagander <magnus@hagander.net>, PostgreSQL-development <pgsql-hackers@postgresql.org>, Stephen Frost <sfrost@snowman.net>
Date: 2018-10-08T15:23:48Z
Lists: pgsql-hackers 
On Mon, Oct  8, 2018 at 12:42:39PM +0200, Peter Eisentraut wrote:
> On 05/10/2018 19:01, Bruce Momjian wrote:
> > On Fri, Oct  5, 2018 at 04:53:34PM +0200, Peter Eisentraut wrote:
> >> On 23/05/2018 08:46, Heikki Linnakangas wrote:
> >>> "tls-unique" and "tls-server-end-point" are overly technical to users. 
> >>> They don't care which one is used, there's no difference in security. 
> >>
> >> A question was raised about this in a recent user group meeting.
> >>
> >> When someone steals the server certificate from the real database server
> >> and sets up a MITM with that certificate, this would pass
> >> tls-server-end-point channel binding, because both the MITM and the real
> >> server have the same certificate.  But with tls-unique they would have
> >> different channel binding data, so the channel binding would detect this.
> >>
> >> Is that not correct?
> > 
> > Not correct.  First, they need to steal the server certificate and
> > _private_ key that goes with the certificate to impersonate the owner of
> > the certificate.
> 
> Right, I meant to imply that.

Right --- that is often confused so I wanted to clarify.

> > If that happens, with tls-server-end-point, a MITM
> > could replay what the real server sends to the MITM.  You are right that
> > tls-unique makes it harder for a MITM to reproduce the TLS shared key
> > which is mixed with the password hash to prove the server knows the
> > password hash.
> 
> So you appear to be saying the above *is* correct?

Yep, I am afraid so.  tls-unique seems technically superior, but I guess
not operationally superior, e.g., Java can't access the TLS shared
secret.  Supporting both causes the kind of channel binding mode
confusion that we have been dealing with, and the new security trend is
not to support too many options because their interaction is often
surprising, and insecure.

-- 
  Bruce Momjian  <bruce@momjian.us>        http://momjian.us
  EnterpriseDB                             http://enterprisedb.com

+ As you are, so once was I.  As I am, so you will be. +
+                      Ancient Roman grave inscription +

Commits

  1. doc: update PG 11 release notes

  2. Fix misspelled pg_trgm contrib name in PostgreSQL 11 release notes

  3. Doc: clarify release note text about v11's new window function features.

  4. Improve wording of release notes item

  5. Fix typos in release notes

  6. Doc: preliminary list of PG11 major features.

  7. Make numeric power() handle NaNs according to the modern POSIX spec.

  8. Various improvements of skipping index scan during vacuum technics

  9. Revert back-branch changes in power()'s behavior for NaN inputs.

  10. Avoid wrong results for power() with NaN input on more platforms.

  11. Avoid wrong results for power() with NaN input on some platforms.

  12. Skip full index scan during cleanup of B-tree indexes when possible

  13. Rewrite the code that applies scan/join targets to paths.

  14. Postpone generate_gather_paths for topmost scan/join rel.

  15. Add casts from jsonb

  16. Make plpgsql use its DTYPE_REC code paths for composite-type variables.

  17. Don't allow VACUUM VERBOSE ANALYZE VERBOSE.

  18. Pass InitPlan values to workers via Gather (Merge).

  19. Account for the effect of lossy pages when costing bitmap scans.

  20. Allow no-op GiST support functions to be omitted.

  21. Rearm statement_timeout after each executed query.

  22. Push limit through subqueries to underlying sort, where possible.