Good points, Jean-Luc. Open source is not about free software, as Rich Stallman has noted repeatedly. Nor is it about easily modified software. It is about understandable and therefore trustworthy software.
I wonder how many smart card manufacturers would be willing
to give their source code to any government that requested it as
Microsoft is doing with their source code.
There will always be a certain element of distrust as regards commercial smartcard products. The reason for this is simple - the government of a particular country where a smartcard is manufactured would be unlikely to permit the release of a smartcard which its technical intelligence services could not defeat or compromise.
Where the primary threat in the threat environment came from disorganised amateurs, then a software based backdoor was probably viable. However with the organised professional piracy of pay television in the early 1990s, the whole threat model changed. Most of those hacks on smartcards were hardware based. Though one very interesting example was the hack on the Sky 09 card. It had a hashing function that used the card's ROM/EEPROM as an input. By generating the correct packets, it was possible to dump the card via software. However at that time, the card code itself had been dumped and reimplemented by commercial pirates. This was not so much a backdoor as a very dangerous, and ultimately fatal for the card, mistake in security. With one valid set of keys (even an old set) and a simple implementation of the hashing algorithm, the whole card memory was dumped. (This happened about nine years ago.)
The comparison of Microsoft with an SC manufacturer is not a good one. The user can always reinstall a Microsoft operating system. Unless a card is completely reprogrammable, then the OS cannot be reinstalled. The disclosure of the source code by Microsoft is one of sheer desperation as more governments consider switching to more robust and more trustworthy (due to their Open Source natures) *nix based operating systems.
Furthermore, it must be noted that the half-life of physical attack
expertise is about two years so a relevant question is how quickly can the manufactureres respond. In the case of both timing and power
attacks, the chip manufacturers responded much more quickly
and more effectively than the smart card manufacturers.
It also depends on the card users and the size of the market that they would have to replace with a more secure card. This is the attacker's window of opportunity.
Which brings another point. Hasn't the physical security of the token
moved by and large from software to hardware? It the early days
of smart cards, the card software did have to worry about these things.
But aren't those days long gone and isn't the expertise that created
this software obsolete? As far as today's environmental attacks go,
the expertise is with the chip manufactureres and the counter measures are largely in the silicon, not in the software.
It would not be unthinkable for a manufacturer to make it as difficult as possible to reimplement the code in another microcontroller. A few technical avenues exist for this - making the code heavily dependent on a feature of the particular microcontroller in use (timing, opcodes etc). The extreme of this would be to include an ASIC in the package which would give any attacker two targets. In some ways, security is more hardware dependent than ever before.
Without full disclosure, any smartcard operating system developer is at the mercy of the smartcard manufacturer. The standard way around this is to adopt the moving target scenario - reissue smartcards often enough so that the attacker's window of opporunity is too narrow to make a hack commercially viable.
The GSM issue though was an interesting example. The primary attack on the GSM card would have been in hardware first - after all the GSM card at the time was not more complex than the typcial Pay TV smartcard and the techniques were already tried and tested. It was a hardware hack that uncovered a crypto crock. Once the code was out of the card, there was very little anyone could do to stop it.
However the GSM and the Humpich cases and the Pay TV cases are very different. Banks and GSM cards operate in a closed loop environment where the control over the backend is maintained and exploiting such a system has very high risks. With a low risk case like Pay TV, the conditional access side of things was removed so that there was no backend control. When the conditional access element can be separated from the encryption element successfully, then the system has failed catastrophically. If the hardware can be compromised in an open loop system, no amount of nifty software tricks will save it. I guess the best way to design a system is so that it can recover easily from security failure instead of trying to design a totally secure system.
Regards...jmcc
_______________________________________________ Muscle mailing list [EMAIL PROTECTED] http://lists.musclecard.com/mailman/listinfo/muscle
