Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 4 Apr 2012 11:07:05 +0200
From: Steffen Dettmer <>
Subject: Re: Re: [JDBC] CVE DISPUTE notification:
 postgresql-jdbc: SQL injection due improper escaping of JDBC statement parameters

On Mon, Apr 2, 2012 at 6:16 PM, Kevin Grittner
<> wrote:
> ... which shows version 8.1 as having reached end-of-life and gone
> out of support five years after release, in November, 2010.

Thank you for sharing your view. Let's me take this opportunity
to tell my arguments why I think an advisory with this
information could be suited.

Yes, all this is true. However, not everyone knows all that and
can "apply usual expectations".

So you don't have old clients running anywhere? Others may follow
the idea not to change a running system (at least as supported by
vendor) but for some reason upgraded one of their databases to a
newer version.

Is this such an unlikely situation that some few old applications
connect to a database which later is updated because some newer
applications benefit from newer database features? Or because the
database server hard- and software is upgraded to keep increasing
performance requirements?

Of course for an expert focusing PostgreSQL / JDBC all this is
known, clear and obvious, but for the less experienced people,
for example application developers, it might not be the case.

Java developers may follow "run everywhere" as long as the test
suites pass (which might not include security penetration testing
and even if so may fail to detect the problem).

Who is in charge to know that such a combination is insecure? The
application vendors might not know which JDBC / DBMS combination
a customer might use. At the customer site IT professionals may
choose supported distributions and may not know details of
PostgreSQL version requirements, especially if not stated in the

The last 8.1 version (build) is from 2010, so someone could
assume it would still be supported. I think this shouldn't be
mentioned only somewhere on a web page, but in the appropriate
users manuals.

Often so many things are used by projects that it could be very
hard trying to read all related mailing lists.

When using a linux distribution that is still supported by the
vendor, I think it is reasonable to assume the included packages
are still supported at least by the same vendor.

I talked with a (very) few people, no one had assumed that; the
assumption was "install all vendor patches and be safe for known
issues" and that different versions either work securely or not
at all.

IMHO also misleading seems to be the term "protocol version 3". I
think this really suggests that any protocol version 3 client
could connect to any protocol version 3 server.
But this is not true. There are many different protocol
version 3 versions, but I don't think that this is commonly
understood by end users. I know that it is nowhere told that it
would be interoperable, but I think a common assumption would be
that protocol v3 is interoperable with protocol v3.

The current situation seems to be "you cannot use older versions
of your enterprise long term support distribution to connect newer
versions of your enterprise long term support one", which in my
point of view had been worth an advisory, but maintainers seem to
conclude that what is not supported from the project does not
need to be supported by the distribution.

Maybe maintainers could decide to provide an advisory anyway, be
it only summarizing the issue. I understand that creating a patch
would be much effort, is not supported and disliked, but
spreading the information could be suited nevertheless.
Distributors could decide to pack newer versions. Even if not,
Distribution customers then at least could decide to upgrade
their systems and/or manually install a suited driver version.

Knowing the issue is needed in any case, I think.



Powered by blists - more mailing lists

Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.

Powered by Openwall GNU/*/Linux - Powered by OpenVZ