Kiwi expert cracks chip passport

[Peter Gutmann]

Stefan Kelm <stefan.kelm at secorvo.de> writes:

>> The original story was actually the coverage in the UK Times last week,
>
>Which card reader(s) did you use?

Adam and I used the Omnikey Cardman 5321 (I'm not sure what Jeroen used,
probably the same), which is cheap, well-supported with drivers, and cheap.
Oh, and it's cheap too.  The card was a standard NXP JCOP 41, one country's
passport implementation didn't change the ATR so when you ping the passport it
returns the product ID in the response :-).  Having said that, going with the
JCOP 41 was more a case of "OK, we'll use that too then" rather than "now we
know the secret" so having the product ID returned in the ATR isn't really a
security problem.  In practice anything programmable with a 13.56MHz RFID
interface should do it, you don't have to specifically use a JCOP 41 card.  As
with the reader, the card just happened to be available and cheap.  Given that
people have built their own prox card emulators it wouldn't surprise me if
someone did the same for a 13.56MHz card (e.g. using the freely-available
OpenPICC design) so you can return "foo'; DROP TABLE passports; --" as your
passport MRZ when the "card" is read :-).

One thing that wasn't mentioned in the news coverage is that, as with any
SCADA-type software, there are bound to be all manner of bugs and holes in the
various reader implementations just waiting to be exploited.  For example when
I was initially playing with creating signatures I just memcpy()d some fixed
data together to create something to sign and was surprised when the Golden
Reader software accepted invalid signed data that should have been rejected as
valid.  I also managed to crash it at one point, quickly fixed the problem,
and then spent the next day kicking myself for not recording what data I'd fed
in to cause this (all your readers are belong to buffer overflows).  I'm sure
there's going to be many more Black Hat/Defcon talks on this in the future.

Has there ever been any third-party analysis of passport reader software as
there has for voting-machine software?  By "analysis" I don't mean the usual
Common Criteria rubber-stamping, I mean actual independent scrutiny of the
code.

Peter.

---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at metzdowd.com