Schneier on Security

Clive Robinson • January 7, 2020 7:08 AM

@ ALL,

How many have,taken apart a cheap lock such as a padlock to see how it works?

You might be shocked to learn that most of the key is not used as part of the opening process, but only the key exclusion process. Thus most of the lock keyway is a series of cheap “static wards” not expensive “active levers”. So if you know where the active lever is you simply leave that on your “skeleton key” and remove the rest of the key that would block against the static wards. Hey presto you now have a “skeleton key” that opens all locks of that type made by that manufacturer.

There is one type of padlock where this is easy to see just by looking at the lock… It looks like a stack of metal plates held together with long rivits at it’s corners, and the key looks like a simple flat metal plate that has a V-fold down the middle of the key to give it a spine for mechanical strength. With the key combination made by simple slits sawn into the key at ninty degrees to the spine.

Well those metal plates are punched out by a machine press and there are only two types of plate for the warding section of the key way. One has a hole the diameter of the outer width of the key, the other has a hole the diameter of the key spine with a cut across it the width of the key. This second type of plate forms the static wards to stop the wrong key being turned in the lock. The first forms what becomes the barrel or tube of the keyway.

Most people would think incorrectly that the longer the section of the warding keyway the more secure the lock… Whilst it will give more combinations to play with the warding idea is flawed and there is a “default open” cut for each ward position that is always the same. Thus you cut your key with all those ward positions in the default open position.

To see this you need to know there are only two metal plate stampings for the warded section of the keyway, and that they are always going to be slightly visually different. Thus you can actually see the static ward pattern from the outside of the lock… So with a little time you can cut an exact duplicate key without ever having to see the key or impression the lock you wish to open…

All you have to know is where the active lever is, which you can do the hard way by buying a lock of the same type and manufacture and taking it appart which is actually not that difficult. Or the easier way because the active lever section is thicker than the static warded keyway plates thus can be observed from the outside of the lock. Or even easier just guess it’s going to be right on the end of the keyway, and by buying two different locks work out where it is by simply puting the two different keys on top of each other and then cut away the area where the static wards are you end up with your skeleton key…

Most other low cost padlocks have similar issues where working out how to make skelton keys for them can be seen by having knowledge of the lock and seeing the outside of the lock you wish to open.

The reason for this is “mechanical bind and slop” the more precision in machining the parts of the lock not just the more expensive it is going to be, but the more likely it is the lock will bind up. Thus the more slop the manufacturer adds the less likeky the lock is to bind up in the field. But slop is a lockpikers friend, as it gives more “feel room” or “wiggle” to feal locking levers and pins into place.

Which has an unfortunate side effect for the consumer. Because less slop means tighter tolerances in machining and beter quality materials it means much higher manufacturing price. However as the lock design is in effect the same that extra price realy does not buy you anymore security…

This problem is also true of very high security mechanical locks, better materials might make “drilling” a lock or similar more difficult but it does not increase the security of the actual design. To increase the mechanical security of the design usually makes bind and wear more likely, thus the lock less reliable mechanically…

Yes that means there is a very definite issue in that for locks to be reliable there are wuite hard limits on what level of mechanical security can be offered.

But any design with “static wards” is going to be only superficialy secure because they have a default open cut that is known. Thus the actual security falls to just knowing where the opening lever is in the keyway. So you cut a key where all the warding positions are “cut to the default open” and only the positions where the opening lever is are left uncut. So the number of warding positions realy is irrelevant…

If you want a “programers note” explanation look at it each ward position as being effectively an “Inclusive OR” gate where cutting out the ward or cutting out the key gives an “open” rather than the position acting like an “Exclusive OR” where there is no default open state.

Thus such locks realy are “security by obscurity” at best, and effectively “no security” against anyone with eyes, a brain and a little basic mechanical knowledge which I’ve now given to those reading who did not know it. But as it’s fairly common knowledge, I suspect some readers here already knew it or would have realised it if they thought about it for a few minutes 😉

But importantly the knowledge whilst given for a specific lock type is way way more general and applies to many lock types not just those that use “fixed wards”.

Thus the take away lesson if you like, is when buying a lock, price is not realy a guide to it’s security, but knowing it’s internal design is… Which the “programers note” also tells you the same logic applys to way more than locks be they mechanical or electronic, it also applies to encryption and other information security practices such as “Human engineering” where there is no effective penalty for “testing” the security system.

Clive Robinson • January 7, 2020 11:01 PM

@ SpaceLifeForm,

You diss Masterlock, rightly, but they are actually tough to beat with a bolt cutter.

I did not mention any manufacturers on the “No name, no pack drill” principle. But “If the cap fits” and quite a few do as I believe the patents have long expired…

However I did mention the locks made with beter materials are harder to drill, but cost a lot more conundrum.

Security like life is a series of trade offs, the real problem is the “hidden hand of the marketplace”. To be able to make valid trade offs you have to be aware of all the implications.

In the locksmithing world there is a “Guild secret” mentality which would put magicians to shame. That is they try to keep many known weaknesses secret from customers, including using lawyers etc.

Take “split pin” locks that support “master keys” the manufacturers and locksmiths that sell such locks fail to tell you something that you will say “that’s obvious” when told.

Lets assume you have five pins in the lock barrel each with only one cut. No matter how many hight combinations there are for the key, only one key profile works in that lock barrel. Which is what most people would expect of a lock.

But if you put two cuts in one pin you now have two key profiles that will open the lock. Which is what a customer who wants a master key would like to hear. What they perhaps don’t realise is that they can only have a very few locks that work with that master key because there is only so many places you can make the second cut in the pin. The side effect of this is often the “master key” cut is at the top or bottom of the pin, thus leaking a little information to a lock picker.

But when you put two cuts in a second pin you now have four key profiles that work in the lock. That is for each extra pin with two cuts in it the number of key profiles that will open the lock doubles up. Therefor with five pins there are thirty two profiles that will open the lock, which is not what most people would want to hear.

Worse because of the likely hood of one of the cuts being at the top or bottom position of the pin, some manufactures “master key locks” used to be “all opened with one of two keys” either a key cut to the top cut or a key cut to the bottom cut…

Word got out about this quite some time ago last century so manufacturers and locksmiths jiggled things around a bit to cover this failing up a little bit. But as a general rule “master cuts” tend to be in the same place on the pins with the second cuts being used for the combinations to give single room keys. This is because it makes cutting of hierarchical master keys as used in multi floor hotels and offices easier to manage and maintain. Oh and much easier for pickers etc to make the master-master or “golden” key that even the hotel or building manager might not have…

Clive Robinson • January 8, 2020 6:27 AM

@ ame,

How do you send a letter securely without using a computer or mobile phone?

The personal computer and mobile phones are relatively new from maube the last decade of the last century, though smart devices are this century.

Mankind communicated privately for centuries and millennia before that.

Whilst we don’t shave slaves heads any more, the principle still applies in stenography. Likewise we don’t wrap pieces of parchment around genetals batons as an early Ceaser was alledged to have done.

If real security is what you want then it can be yours as it was in times past with pencil paper and a few dice and a lot of laborious work.

All you have to do is,

1, Make One Time Pads.
2, Find a secure way to transport them to the person you wish to communicate with.
3, Have some way to stop that person leaking either the keys or plaintexts.

Arguably the third is not possible, the second unlikely in this modern world, and as for the first just way to much work for everyone… But you can do it if you want to.

There is howrver no reason why you can not use a PC or phone to communicate. But to do it securely you have to seperate the communications from the security function so that the security function is not reachable by an attacker via either the communications channel or some other route be it physical or energy.

So as the first party in a secure message exchange you could have one computer that is “energy gapped” and physically protected always, on which you have the security function which turns the human readable “plaintext” into secure “ciphertext”. You then need a secure mechanism by which you take the ciphertext across the energy gap to a second computer or smartphone that is the end of the communications channel. At the other end the second party does the opposit.

However there is a fly in the ointment. It is possible for a message to also be malware, which you will not know untill it’s ciphertext has become plaintext on the energy gapped computer potentialy infecting it (think hidden malware in PDF’s, image files, word processor documents etc). So the only files you should send are “plain ASCII text” using CVS or similar human readable format to transfer data. You therefore turn the ciphertext to a plaintext and scan it in various ways to find potential malware.

But no scanning process is perfect. Which is why some years ago several people then frequent posters on this blog talked about how to reduce the problem further whilst still making it usable.

The result was Markus Ottela started “Tin Foil Chat” now know more commonly known as “TFC”. It has pages up on GitHub that you can read a lot lot more on not just TFC but how it works and the threats / attacks it was designed to stop,

https://github.com/maqp/tfc/wiki

Clive Robinson • January 13, 2020 7:47 AM

@ SpaceLifeForm,

Fitst off an appology for not getting back sooner, I’ve damaged the rotor cuff in my other shoulder, of the dominant hand, and it’s making typing even on a smartphone screen more than a little unplesant. Worse as I walk on sticks I’m house bound, and would if I could be “climbing the walls”.

With regards,

With current technology, I believe it is possible to create an electro-mechanical key (battery driven), that can open *ANY* lock.

Whilst locks have certainly been openable with just magnets in the past and other locks have fallen to motorized “pick guns” and “Newtons keys” and their simpler cousins “bump keys” and similar attacks there are ways to stop the attacks if you “think hinky” and design it in.

Take the first case of magnets, most electronic locks that run off of batteries don’t work the way most people would think.

That is the energy to pull back the latch does not come from the battery, but by the human turning the handle.

When you pull electronic locks of this type appart you usually find their are two distinct sets of mechanisms, one for the door handle rotation and one for the latch pull back. Joining them is a simple “pull in cog” gear chain or “clutch plate” which are “normally open” so you can spin the door handle as much as you like because there is no connection between the two sets of mechanics.

When the electronics is activated a small low current solenoid either pulls in the cog or closes the clutch plate thus conecting the two sets of mechanisms together so that turning the handle causes the now pulled in cog/plate to turn the latch mechanism pulling the latch back out of the strike plate and door jamb.

The problem with solenoids especially low current ones is that any correctly aligned magnetic field be it from current through the windings or from a sufficiently powerfull external magnet will cause it to “pull in”.

Thus as described the lock is vulnerable even if you put in magnetic shielding. Because shielding never stops the unwanted field, it only attenuates it to a lesser or greater extent (which is why if you want rid of the earths magnetic field in an experiment or high sensitivity instrumment you balance it out with linear field Helmholtz Coils). Thus if you are not smart you enter a battle between the thickness of expensive shielding and ever bigger rare earth magnets. Which in most cases the lock designer will lose.

And rightly so, because the lock designer is not “thinking hinky”, or even sensibly about the lock design.

Logically if you have a normally open mechanism that pulls in a cog or plate under a magnetic field, you can invert the state and have one that is normally closed that pulls out a cog or plate under a magnetic field. If you put them a small distance appart and orient them orthagonaly to each other then the solenoid will only actuate the “pull in” whilst any external magnetic field will actuate both thus pulling out the other cog or plate disabling the link between the handle and latch mechanisms.

A similar stratagie applies to all the other types of attack, you just have to be able to recognise an attack type characterize it and apply an appropriate defence.

The problem of course is two fold firstly we don’t know all the instances or even classes of attack, thus defending against the unknown is to a certain extent probabilistic[1]. But importantly except in rare cases you can not aford to design against all classes of known attack, but even then due to the nature of things mechanical there will be ware, bind and slop working in an attackers favour giving rise to edge and corner cases they can then use as toe holds. But in most cases attackers won’t bother they will find some other attack such as “jacking the door back” or more simply going up and over the room walls via the ceiling crawl space in office blocks etc.

[1] As a designer you can tip the probabilities more in your favour by makeing each defence measure “generic” and relate to one or more classes of attack, not “specific” to individual instances of attack.

Clive Robinson • January 14, 2020 8:04 PM

@ SpaceLifeForm,

I remember the Bi-lock being mentioned before, and I looked up the design.

From what I remember whilst it used pins none of them were split, they had holes into which a side bar had to be pushed by the key but only when fully inserted.

I know locksmiths won’t like me saying it but “pick-guns” especially the electronic ones are ment more for the inexperienced not skilled.

I don’t think those that need to use pick-guns would be able to get them to work on the Bi-lock. However Untill I get my hands on a couple of Bi-Locks to pull apart etc I could not be certain, but the cut-away drawings I’ve seen would suggest they would require someone quite a bit more than just “skilled” to pick them.

That is I’m not saying they are pick proof –I don’t think any mechanical lock is– but unless they have a design flaw that’s not obvious on “artists impression” “cut-away drawings” they will take skill and time.