[kwlug-disc] Truecrypt

[Bob Jonkman]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Even if a piece of software passes both a code review and a security
audit to verify that its code properly implements the security
algorithms (assuming those algorithms are sound in the first place),
it is possible that the audit team will miss some minor flaw that can
be exploited later. After all, the OpenSSL code passed a code
review[1] that completely missed the Heartbleed bug. And GotoFail. And
Debian PRNG. And any flavour of Windows. Think Microsoft doesn't do
code reviews/audits?

Even if the code is flawless, bugfree and properly implemented there
may be bugs in the compiler that cause the binaries to behave in ways
that the source code doesn't predict.  Will there be a code
review/audit of the compiler? AFAIK the targeted compiler for
TrueCrypt is a 10-year-old version of Microsoft C. Good luck in
getting the Open Crypto Audit Project access to that source code.  And
even then, it's worth re-reading Reflections On Trusting Trust[2].

And even if the compiler is flawless, bugfree and creates flawless and
bugfree binaries, there will be updates to the Operating System that
runs the binaries.  It is entirely possible that a shift in the OS
paradigm (init.d? Upstart? SystemD?) invalidates some compiler
assumption, so that the binaries no longer behave as expected. Or
worse, that the compiler is designed to work around some flaw in the
OS, so that when that OS flaw is fixed the compiled binaries exhibit a
new flaw that doesn't exist today.

So, the assumption that if a piece of code is OK now then it will be
OK forever is deeply and dangerously flawed.

- --Bob.

[1] http://veridicalsystems.com/blog/of-money-responsibility-and-pride/#9

[2] http://cm.bell-labs.com/who/ken/trust.html



On 14-05-30 05:57 PM, Khalid Baheyeldin wrote:
> If I read Bob's comment correctly, he is objecting to the
> extrapolation on that the future audit is most likely positive.
> 
> On Fri, May 30, 2014 at 4:40 PM, unsolicited <unsolicited at swiz.ca>
> wrote:
> 
>> And others from here below:
>> 
>> On 14-05-29 12:03 PM, Khalid Baheyeldin wrote:
>>> ... If the previously audited version was safe, i.e. works as
>>> designed,
>> then ...
> 
> 
> The initial audit was to verify that the binaries do indeed
> correspond to the published source code, and that no backdoors have
> been slipped in the binaries. That much was verified.
> 
> But all that applies to the 2012 version (7.1a). The one that was
> just published this month removes a lot of features, and has not
> been audited in any way (beyond the diff between it and the
> previous version).
> 
> Another note: someone is saying that a flaw will disclosed soon.
> 
> http://soylentnews.org/article.pl?sid=14/05/30/1318243
> 
> 
> 
> _______________________________________________ kwlug-disc mailing
> list kwlug-disc at kwlug.org 
> http://kwlug.org/mailman/listinfo/kwlug-disc_kwlug.org
> 
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Ensure confidentiality, authenticity, non-repudiability

iEYEARECAAYFAlOJCiAACgkQuRKJsNLM5ergZACeJxowFr5kYHD+aJi2xDUFW4uW
woAAoIF55AbEuAWQJ279sod9d/I7N/Sq
=dTBU
-----END PGP SIGNATURE-----