21. Could you explain in more detail just how you interfaced the device to the sms controller? Does the sms controller include instructions to help achieve this? I hate carrying around those things (netkey, one time password, token, etc) as the banks here in Mexico require a different one per bank so it no longer an unobtrusive keychain dangle, it is several!

Posted at 2:40PM on Nov 27th 2007 by JC

22. I got one of these calculator type things from Barclays. It's called PinSentry.

It clearly states in the instructions that if you enter your pin incorrectly three times your card will be locked. As there is no connection to the outside world this must be done via a pin stored on the card on the chip. Presumably also to lock your card the device must be able to write to the chip to tell it that it is now locked.

Presumably there is some encryption. Whether this can be reverse engineered to allow the PIN to be determined for any card I don't know. However I wonder if you could use the PinSentry itself to (laboriously) hack cards to find their PIN.

All it would need is a way of bypassing the card locking. Maybe this could be done by simply putting tape over the write contacts on the chip so that the device can't write to the chip to tell the card that its locked. Maybe its a little more complex than that. Maybe they check they can write to the card as part of some authentication procedure?

I'll do some investigation for sure. Anyone know if I'm barking up the wrong tree completely?

Posted at 7:04PM on Dec 1st 2007 by mj

23. to #13.

The entered PIN will be hashed into the 8 digit code the PINSentry device generates.. the card won't have the PIN on it. If the decoded PIN (from the submitted 8 digit code) turns out not to match the one held on the server 3 times then it will trigger a lockout for the next time that card appears at an ATM or in a shop.. it wont actually write the lock to the card just as it didnt read the PIN from there.

It does accept any code I just tried it on mine.

Posted at 1:02PM on Dec 3rd 2007 by sam

24. The PINsentry isn't connected to the outside world, and the card does not store the PIN.

What I think happens is the card will store a validation number. The PINsentry can use the PIN to generate a validation number, and see if this matches the one on the card.

So yes, there might be a way to use the PINsentry to still calculate the correct (or a number of possible) pin numbers. So I do wonder if overall this is a secure idea.

Posted at 6:36PM on Dec 5th 2007 by Danny

25. my pinsentry does not accept any PIN. if i enter a wrong PIN after pressing Identify, it knows and says the code is wrong (i just tried it). it only generates the 8 digit one time login code if i enter the correct PIN.

i would think that the way it does this is a one way encryption process, like normal unix passwords - presumably the card itself stores an encrypted version of the PIN, which is tested against whatever value is entered via the keypad after putting it through the same one way encryption. (i think that chip-and-pin readers work the same way?)

so you could use the pinsentry to brute force the PIN for a card, but only if you could figure out a way to disable the card lockout after 3 wrong tries. i would hope that the bank knows this and that the pinsentry is designed in such a way that the lockout mechanism couldn't be disabled without disabling the PIN test method.

Posted at 8:45AM on Dec 17th 2007 by g

26. I did have a similar idea as soon as I realised I'd have to carry the bloody thing around, but there's something I don't get: surely you have to leave your card at home, which is far more inconvenient than not being able to get into online banking at will?

Posted at 12:29PM on Dec 17th 2007 by Alex

27. to #23 and a host of others.

The card contains the PIN. It works the same as Chip & PIN. In fact, it is a Chip & PIN card application!

You enter the PIN, the card verifies the PIN (yes, the PIN is stored in the card!) then generates a cryptogram based on some other variable input data and the crypto keys (also in the card). The calc device puts part of the cryptogram into the code - the other part of the code is the variable data the bank needs to recreate the code for validation purposes. The bank also has the same crypto keys. Hey presto!

The PIN is not hashed. Nor does it, or a hash, or anything else to do with the PIN, get sent to the bank.

If you enter the PIN incorrectly three times, the card locks and blam - no more cryptograms. At least, not until you phone the bank and go down to the ATM for a card unlock.

As for the "hack", how is the PIN entered?

Posted at 10:03AM on Dec 18th 2007 by Webster

28. Isn't there a big flaw in this that he has to leave his card plugged into his pin sentry device at home. What if he wants to get money out of an ATM or buy something at the shop and he does not have his card?!

I have heard anecdotally that these particular devices are not time critical, so for log in only (not payments where you have to also enter the amount and recipient account no) you can go through the process three times, generate three codes and take them with you to allow three log ins while you are away from your pin sentry. No idea whether this is correct though.

Posted at 8:36PM on Jan 23rd 2008 by tom p