Schneier on Security

Clive Robinson • May 29, 2012 7:26 AM

I’ve never met a safe or lock that kept me out for very long. Not saying I can’t be stumped

And in the physical world he never will….

It’s almost a given that all mechanical devices have “slop” in their design, it’s actuall more important than “lubrication” to the long and successful working of a mechanical device.

Without it the lock would due to a whole host of reasons “bind” up and be usless.

Also at the end of the day all mechanical things break no matter how well you design them “entropy” will have it’s way. And if you. are designing a safe it’s important that you have weak points so you can still get it open even if the lock does break.

Anyone who things otherwise is going to be in for a sad and sorry awakening at some point unless they are short lived.

We people who work with information think we can design unbreakable information systems and in certain respects we can. But to be usefull all of these fancy information systems have to work in the physical universe we inhabit that we sometimes call “the real world”. This means even fancy information systems are vulnerable to times arrow and it’s hand maiden entropy.

As has been observed before “-ou can’t cheat Old Father Time”…

Which is why those dealing in physical security talk of security absoluts but time. In France they have one word for both security and safety, and we know very well from safty systems we don’t talk about fire proof but time ratings for resistance such as A30 and A60, where those numbers mean minutes.

Clive Robinson • May 29, 2012 8:00 AM

@ Bruce,

You should also have provided a link to another page on that site,

http://www.mcsweeneys.net/articles/airplane-passengers-as-explained-by-their-pants

It’s a funny take on “profiling” so atleast vaguly relevant :o)

Clive Robinson • May 29, 2012 12:49 PM

@ Mike B,

Shame this guy didn’t cover new digita technologies which are a lot more resistan against many of the physical cracking attacks…

They are and they are not…

The mechanical parts of all types of lock are just as vulnerable as they have always been (in fact more so in many “digital” or electronic locks).

Now a traditional mechanical lock has a lot more mechanical parts to do with “key detection” so you would thus expect more “mechanicaly vulnerable” points.

A modern digital lock however does the “key detection” electronicaly, and this introduces new mechanical and electrical attacks on the electronics.

Now there are two basic types of electronic lock based on their power source (battery or mains) and this has very very significant design effects not just on the electronic but on the mechanics as well, and the additional “electro-mechanical” interface be it a motor, solenoid, speaker coil, or piezo clutch plate etc.

In a mains powered system such as a “door entry system” you can use an “open solenoid” where you have a feromagnetic metal plate bolted to a strong point on the wall/door and a bracket bolted to another strong point on the door/wall, bolted onto the bracket is a low current high intensity “electro magnet”. If you cut the current in any way then the electromagnet “fails open” good for an emergency door, bad for a safe door.

Thus most safes are designed to “fail in the last state” that is they remain in the previous locked or open state irespective of if the power is there or not. That is they have some kind of mechanical latch that is bi-stable it is either open or it is closed in it’s rest or unpowered state and is only powered to transition the latch from one state to the other.

Now obviously such a bi-stable latch can be fairly easily and robustly designed if you have effectivly unlimited “mains power” available. You don’t have to overly worry about friction or weight or orientation of the bi-stable latch as there will be sufficient electrical power to overcome it.

This is not true of a battery powered lock where you might be looking at a specification calling for a years life time or minimum of 4000 state changes before a battery change (or both). This is a very tough spec to meet on a battery even though it’s only 11 state changes a day.

This means you have to use quite clever mechanical designs using very low powered motors or solenoids. Almost invariably they do not actually have the power to move the latch, what they do is put a “dogtooth” into a “drive chain” or act as a “clutch” in a drive chain. In almost all cases the drive chain is between the “human driven” handle and the latch.

Because of the low power of the solenoid the “dogtooth” is almost without exception very light and held back out of the drive chain by a very weak spring, just sufficient to pull back from the unloaded drive chain and against gravity. Usually the drive chain will have a stronger spring on the handle end sufficient to open the chain in the unloaded state so little or no friction is present on the dog tooth and it can be thus pulled back by the weak spring or pulled in by the almost as weak solenoid.

Thus all sorts of new avenues for attack come up including,

1, hitting the lock with a rubber hammer and twisting the handle, this is just like “bumping” a mechanical lock.

2, Spin and jerk on the handle, as some rotational drive chains will due to “centrifugal” (or centripetal depending on design) force catch without the dog tooth being activated.

3, A powerfull magnet, in practice you can not cost effectivly shield the solenoid from external magnetic fields so the solenoid will “pull in” with a sufficiently strong applied magnetic force (the way to stop it is to have another dog tooth that has no solenoid but when activated by a magnet locks out the drive chain so it won’t work).

4, Drill the lock, all locks have deliberate “weak pointss” so they can be opened, even if they don’t if you know where to make a small hole you can poke in a small piece of wire to activate the drive chain.

Oh and many other attacks on the mechanics.

Then of course you can attack the electronics in a whole vatiety of ways that I won’t detail simply because way way to many electronic “digital” locks fail to very very simple and very easy to implement attacks that usually leave neither physical or electronic evidence, usually because the designer does not know what they are doing as they have little or no experiance in designing “security electronics” and have to deal with parts of the specification that include “emergancy opening proceadures”.