Internet Survival Time

If you are thinking about pointing a new Windows machine at the Internet so that you can do online updates, you need to take into account the Internet Survival time and make sure that you have a firewall installed and turned on. If you haven't, you will only have a few minutes before your machine is attacked and possibly taken over - this is not normally long enough to download and install all of the patches you'll need on a new install. The Survival time is based upon a number of different factors and assumptions. Attacks on UNIX systems will always be added to the UNIX count and the same for Windows but there are also other factors such as the applications that are used on those systems.

If you used Javascript on Internet Explorer (please don't), then you are open to the attack displayed above. The screenshot was taken on a Linux machine using Konqueror. There is a lot more information on the site relating to security, including a world map of where attacks come from. Click here to open the http://isc.sans.org/ home page in a new window.

French Keyboard - Key Locations

Sometimes, we end up with a keyboard that doesn't match the layout that the computer is expecting and we end up pressing keys that are labelled quite clearly but produce something different on the screen. Normally, this happens to people using the KNOPPIX distribution which has a default US keyboard and we (in th UK) get the @ and " in the wrong places although there are other keys that do this as well.

Here, we look at the French Keyboard because it produces problems that are analogous to the problems you will find if you happen to have any of the hundreds of other keyboards. Note that this poses no problem at all if you touch-type in the language of the keyboard you are mapped to - if you can touch-type on a French keyboard, it doesn't matter what legends the keys have on them as it will always come out in French if you see what I mean.

I know this because a number of years ago, I used to run (as well as perform in) a modern jazz sextet and we had to have a technical specification for venues so that they would know how big a stage to provide us with, where acceptable places for power outlets would have to be (along with their ratings), the weight bearing capacity of the stage for various load types (things like speakers, drums and so on which would produce spread-out or concentrated loadings) any requirement for external amplification and so on. Not only was this in English but it also had to be in German on the back. So, I ended up typesetting a complete German translation on an English keyboard. After a while, I got used to the way that German words are spelled in the sense that when anybody types, they perform finger patterns on the keyboard. Whether you are spelling 'necessary' or 'schlagzeuger', with a bit of experience, you are performing a pattern rather than trying to remember how to spell something. Especially when you have to remember the key sequences for characters with accents.

There are many ISOs out there and you can try this one out for yourself using the French MiB KNOPPIX ISO - see below in the 'Other Live CDs' section about getting yourself a copy.

With French, accents dictate that the number row has special characters and the more common accents 'é', 'è', 'ç' and 'à'. These are used more commonly in French than the number keys (in normal use) so you will find that to access the numbers, you need to press [Shift] key. This is just making life easy for the people who would normally use such a keyboard.

Also, we find that the [Alt][Gr] key has a use other than stopping you from using the space bar by accident when you go for the flag key. To get to '#', '{', '[', '@', ']' and '}', hold down [AltGr] and press '3', '4', '5', '0', '-' or '=' respectively (on a UK keyboard).

The left side of the keyboard is different in that: 'q', 'w', 'a' and 'z' are replaced by 'a', 'z', 'q' and 'w' respectively whereas on the right side, 'm' is to the right of the 'l' instead of on the bottom row. As for punctuation, you need to look at the file - see below.

To get a number of other accents and other special characters, one of the keys - the '[' key - has been taken over as a special key. You press this and then press the target key and it modifies the output. Without pressing the shift when pressing '[', you get a caret. If you press the target letter without a shift, you get the lowercase version (such as 'â') and with the shift, the uppercase version. If you press the '[' whilst holding the shift, you get an umlaut (two dots above the letter) in the same way (as in 'ä').

You can see the layout file, as gleaned from my own computer with a French OS, by clicking here - it will open up the file in a new browser. You can also see some of the basics of other keyboard layouts by clicking here although you will need to set the character encoding to UTF8 to see the Cyrillic characters properly.

French Keyboard - Changing Layout

Learning the layout of a foreign keyboard is all right if you are stuck at a terminal or you have no way of changing it - or, if you are learning to use it. However, if you are using a GUI such as the example in the screenshot - KNOPPIX using KDE - then there is an easier way of changing it. In the bottom right, you will see the system tray.

In the system tray, you will see a flag and in this case, it has the French flag in it. If you clicked on it with the left mouse button, you would get any alternatives that are available - usually the US keyboard.

If you right-click on it, you will see the alternatives you have available to you. You could click on the US English but as you might end up in a situation without that option, we will look into how to get around it.

The key to all of this is the icons. We need to configure the keyboard so look for the tool icon - here, in French, it just happens to have the word 'Configurer' next to it but you might not be so lucky on yours.

Next, under the 'Disposition' tab, look in the 'Dispositions supplémentaires' list. The first thing that you will notice is that you can't find the UK English keyboard layout under 'EN', 'GB' or 'UK'. If you don't know anything about the local language, you are faced with the prospect of searching your way through hundreds of icons, looking for a Union Jack with 'gb' or 'en' in it. French though, is a Latin language, just like English and if you remember that a king is a 'Roy' and that it is united, you can find what you are looking for as 'Royaume Uni'.

Check the box and click on the 'OK' button.

Now, when you click on the flag in the system tray, it includes the UK layout and you have back your '@' and '"' in the right place which you wouldn't have had if you had just opted for the US layout. If you never had anything like a US or UK layout, you have made an even larger gain.

French Keyboard - Configuring Ethernet Card (KNOPPIX)

One thing that you need to do is get in the Internet and unless you have a DHCP server on your subnet, you need to do this manually.

Again, you are in a position where the only similarity is the icon set that you can see on other distributions. However, this is not always there as you will see. So, if you are going to configure your network card using French KNOPPIX, first of all, click on the KMenu icon

and then on 'KNOPPIX'. The KNOPPIX menu item contains many of the configuration items that are needed to get KNOPPIX up and running each time so it is a good place to start

Next, click on 'Network/Internet'

and then on 'Configuration de la carte réseau'. So far, we have had icons to guide us.

Next, a form will appear asking; 'Utiliser un serveur DHCP?'. Fortunately, DHCP is an acronym used fairly universally as it is an Internet standard that is fundamental to the way things work. If you use DHCP, click on 'Oui' and that's it, otherwise, click on 'Non'.

Next, we get a form with the legend 'Tapez l'adresse IP pour eth0'. Fortunately, it uses the term 'l'adresse IP' so we can guess that it wants the IP address of the machine we are using. So, enter the IP address and click on 'OK'.

Following this, we see 'Tapez le masque réseau pour eth0'. Again, we recognise the word 'masque' so we can assume that it wants to know the subnet mask. If you typed in the IP address correctly, you should get the correct subnet mask as it is able to work this out from the first few bits of the IP address so this should be correct. A 192.168.x.x will give 255.255.255.0 as its mask without you having to type anything. Click on 'OK'

Next, in 'Tapez l'adresse de broadcast pour eth0', it asks for the broadcast address - it has the word broadcast in it. This should be your network address (in this case 192.168.1.0) ORed with the inverse of your subnet mask (ie, 255.255.255.0 inverted is 0.0.0.255 so OR this with the network address and you get 192.168.1.255). This number should already be in the box as it can calculate it from the two pieces of information you have already supplied it with (as we have done in the calculation above).

This one has no clues for the non-French speaker: 'Tapez l'adresse de la passerelle par défault'. It is clearly an address and it turns out that it wants the gateway address that you use and, just to be helpful, it suggests a number - 254.

Finally, in 'Tapez l'adresse du serveur DNS' it asks for your DNS address and suggests 254.

After you have clicked OK for the last time, it configures your network card for you.

KNOPPIX always asks for things in this order so, if you are experienced with KNOPPIX (or have read this), you should be able to do it regardless of the language used (be it French, German, Dutch or whatever) or how familiar you are with it.

If you have your network configured properly - your network address is '0', your broadcast address is '255', your gateway address is '254' and your DNS is also on '254' - it should be able to give you good suggestions once you have typed in your IP address - all you have to do is click on 'OK'.

Remember though that if you are typing in IP addresses and masks, that either you need to know where the requisite characters are on the keyboard or you should have already changed the keyboard.

For example, on a French Keyboard,

192.168.1 80

turns into

&çé:&-_:&:_à

so unless you know how to get numbers and full stops (dots), you will be stuck with a job of trial and error.

Choosing a Printer

A printer, isn't just for Christmas, if you choose carefully and maintain it well, it can last for life (well, a decade).

If you choose a good printer, you can not only make it last for a long time but also, you can have it work with the hardware that you are likely to have in five years or so. This rather begs the question: 'what operating systems will I be running and what will they be capable of?' The answer to this depends on what you are (or are going to be) doing.

If you are running a small office or a home office, you will probably need a printer that is capable of producing several hundred sheets within a given month and therefore one of these small printers that only cost around £50 will probably cost a small fortune to run in terms of replacement ink. So ideally, you will be looking at a printer that costs between £150 and £300 that is built to last and, whilst the replacement cartridges for an InkJet would cost, say, £25, you get a lot more mileage out of them and so the running cost is a lot less.

There are other considerations though. Many printers today can cost as little as only £50 because they have a chip missing. This started around a decade ago when normal printers of the time would - if you sent an ASCII 65 to them - print out an 'A'. Manufacturers realised that there was a fair amount of computing power in the computer itself and also that when people were printing, they usually weren't doing anything more than watching the printer. With this in mind, they decided that the driver could get the computer to do all of the calculations and just send raw head matrix data down the cable to the printer. By doing this, there was no need for the printer chip that did the work. At the time, these so-called 'GDI printers' cost around £50 less than the ones with a brain.

So, at that time, people would generally have a printer plugged into their machine and whenever they wanted to print, they would switch it on and print. The users were on a single-user, standalone machine that wasn't networked at all.

Things have changed though. Today, people use networks and you can either buy a network server for a single printer (or up to 3 of them) or, a network printer and, being networked, you can have the printer anywhere and have anybody using it. The idea of having to have a computer switched on all of the time just so that you can use a printer is an abhorrent waste of energy and the planet's future.

Also, gone are the days when Microsoft (or Apple) was the only type of OS plugged into your printer. Nowadays, users are more intelligent and use a variety of OSes. In the future, there will be less Microsoft-driven hardware on your networks as people realise that there are cheaper and more reliable alternatives and they switch over to Linux and the BSDs. We don't even know if there will still be a Microsoft in five years time. GDI printing only works with Microsoft (although there are some printer manufacturers that appear to provide MacOSX drivers for these as well) so if you get a GDI ('Brainless') printer, you will be stuck with having to boot up a Windows machine just to use it or having one switched on all of the time. Another thing to consider is whether or not Microsoft will abandon GDI printing - their increasingly inefficient operating systems drive the upgrade cycle as previous hardware fails to provide enough punch to make the new bloated OSes work adequately.

So, what printer spec should you be looking for if you want to make your office hardware reasonably future-proof?

You need a printer:

• with a sturdy build;
• hopefully with a sloping top (if it is flat, people will put cups of coffee on it and it only takes one spill to ruin it);
• that has a Centronic interface as a minimum so that you can connect it with a printer server or alternatively, with an RJ45 socket so that you can plug it directly into your network;
• with that vital chip so that you can send data to it from a variety of OSes - if you go for a GDI printer, you will be faced with OS-Lock-In and you will have to have a computer switched on all of the time in order to run your print jobs;
• ideally, it should use some sort of common language such as PostScript or have drivers on your OSes so that you can take PostScript files to it;
• the capacity of the ink cartridges should be large otherwise you are going to spend a lot of time buying new ones; and,
• it should have some of the green, power-saving options such as some equivalent to a 'sleep' mode.

Other Live CDs

A live CD distribution is where you can get an operating system up and running from just a CD - you insert the CD and then boot up (providing that your BIOS is set so that it queries the CD before the hard drive).

The Live CD doesn't write to your hard drives (you can run them without a hard drive if you so wish) because it doesn't need to. You can, in effect, test out an operating system without damaging your existing, installed system at all. If you do want to write to your hard drive, you must explicitly go through a number of actions and you cannot do it by accident.

CDs are, of course, write only as far as booting an operating system is concerned so it takes a special type of operating system to be put onto a live CD. You cannot have a normal version of an operating system that needs to write to a disk when it is booting so you won't find Windows as a live CD. UNIX-like operating systems such as Linux and the BSD families are far more flexible when they boot up and can be made to boot up without writing anything to a hard disk. This makes them ideal.

KNOPPIX is also a very good way of recovering files from dead Windows boxes. With KNOPPIX, you can mount another drive or use network storage to copy files from the Windows machine before you try to recover Windows - this normally ends up with a clean install as Windows isn't particularly repairable if something fundamental has gone wrong with one of the files such as a Registry file that wasn't backed up. It is usually better to take the opportunity to upgrade to one of the Linux or BSD distributions than resintall Windows.

One of the big differences between Windows and the UNIX-like operating systems is that Windows has a megalithic kernel and you cannot have it without the GUI - you are always tied into the GUI. You cannot change the way that Windows boots in any fundamental way. On the other hand, the UNIX-like operating systems such as Linux use a number of scripts to tell them what to do when booting up and loading X-Windows is only one part of that. In that way, you can edit what the UNIX-like operating systems load and how - it is far more flexible. So, if you want to make your own live CD you can, just by getting all of the relevant parts and assembling it all so that it fits on a CD and doesn't write anything. This flexible approach has led to many different types of live CD for many different uses from playing games in French to forensic analysis of discs. These are some of them.

This is KNOPPIX (in English - it is also available in German) and it is the most common live CD by far. It is available from the KNOPPIX sites: http://www.knoppix.net/ and http://knopper.net/knoppix/index-en.html. KNOPPIX has many programs on it because it uses a high level of compression. This means that it might take a bit longer to get a program you have selected to get up and running (many are quite quick but the larger ones have to be read from the CD which is intrinsically slower than a HDD). In effect, KNOPPIX allows you to try out Linux and the Open Source programs that go with it without making any commitment to trashing your hard drive as it is left untouched. One way of working with this is to run KNOPPIX and then go back to using your normal OS, then try KNOPPIX again and so on. You will find that, after a while, you will want to install a permanent Linux distribution because Linux provides everything you want, does it better and for free.

Many other live CDs are based on KNOPPIX and whilst KNOPPIX is officially in English and German, there are others including Spanish and Japanese: This one is KNOPPIX in French. It has a fair number of programs and games on it but all in French. Even the operating system is in French so if you have a job that means that you go abroad and use a French OS, you can get used to the commands on this first - without making any commitment to your hard drive. This distribution is called KNOPPIX-3.2 MiB-11b Privacy Edition and you can get it from http://www.bouissou.net/knoppix-mib/doc-html/Knoppix-Mib.html.

Traditionally, KNOPPIX uses KDE as its desktop environment. This is highly configurable and you can make it look like MacOSX or any version of Windows if you are worried that other users will not be able to cope with something that looks different. There are alternatives to KDE - one of these being Gnome. Gnome is equivalent in many ways (certainly in terms of the flexibility of the interface) to KDE and many say that it looks better and performs better. Then there is the opinion of those that use KDE instead. With the live GNOPPIX CD, you can make up your own mind. You can download it by clicking on the downloads link close to the top of this page ( http://www.gnoppix.org/ ) and selecting a mirror that is close to you.

Of course, KNOPPIX - whether it is the KDE or Gnome version - isn't the only Linux distribution that has a live CD. Here, we see Slax - the Kill Bill version - which you can get from http://slax.linux-live.org/?lang=en. This detects your hardware quite nicely and if you have several optical drives, you can use the KIIIb CD/DVD burner. There are several versions of this of which Kill Bill is one. If that wasn't enough, you can make your own Slax distribution live CD.

Getting away from the heavyweight GUIs and using the fact that Linux is very flexible and doesn't write to the hard drive, we get to another important area of live CD distributions - Security. There are various live CDs that turn a box into a firewall - the advantage being that as the CD can only be read, the OS cannot be corrupted. Any configuration details can be saved to a floppy diskette (and also to another one as a backup if you want to) so that when you boot up, your firewall is already configured. These are not simple firewalls such as the ones you find on Windows, these can be application layer gateways (the most secure as it checks what is in the data part of the packet as well as the header - SPI and SMLI filters just check the header information unless instructed explicitly to check the data part of each packet for a particular protocol). Application gateways will take packets, assemble them as the application would see them, inspect the data and then build new packets. So, if you ban JavaScript on html, the application gateway will strip out everything bracketed by the JavaScript tags, assemble new packets and send on the data without the JavaScript. This particular release is not a firewall but a forensics version of KNOPPIX called KNOPPIX STD (Security Tools Distribution). It has a variety of tools falling into the following categories: Authentication; Encryption; Forensics; Firewall; Honeypots; Intrusion detection systems; Network utilities; Password tools; Servers; Packet sniffers; TCP tools; Tunnels; Vulnerability assessment; and, Wireless tools. You can download this with wget from http://www.knoppix-std.org/.

So far, we've looked at Linux but Linux isn't the only OS that can be used on a live CD. Here, we see NetBSD. If you want to prod the BSDs to see what they do, this is an ideal opportunity. NetBSD is not the only one either. You can download the current NetBSD live CD from http://www.netbsd.org/gallery/products.html#netbsd-live.

You can see a full list (if you can get a 'full list' of something as dynamic as this) at FrozenTech's LiveCD List at http://www.frozentech.com/content/livecd.php. Getting it running To get hold of and use a live CD, you need either: to get the already-burned CD; get the ISO from a magazine cover disc (PC Plus usually has quite a few of these over any given year); or, download the ISO from a site yourself. See 'Downloading Large Files Effectively' below for more about this. Use your favourite CD burning tool to burn the ISO. With your image burned, make sure that your BIOS is set to read the CD before the HDD (otherwise you will just end up running your normal OS) and boot up the machine. The live CD will probe your hardware, select which drivers it needs (something that it normally does quite well in my experience) and then take you into the GUI (if that is what your choice if live CD does - most of them do). When you have finished, you can shut it down in the normal way (something that you should get into the habit of doing if you are going to use an installed version at some point and definitely something that you should do if you have attained write access to any hardware such as flash drives or other HDDs) or, if you haven't mounted any hardware with write access, you can just switch it off.

Digital Rights Management Rootkit Trojan

Security presents a steep learning curve for anybody who is not used to the concept. For Digital Rights Management, that curve is pointing downwards. DRM software tries to limit the amount of copying going on from any CD - there are of course legitimate reasons for copying of which backup is one.

The big problem for any software that attempts to find out how many times a CD has been recorded is that the CD itself is read-only. You can put as many programs on there are you like but they can't save anything on the CD itself. This means that if you want to keep this data then you need to save it somewhere and on the user's machine seems to be a logical place.

The problem with using the user's computer to run a program that monitors the system for any copying activity and logs any that is found is that you need to save the files on the system somewhere. Under normal circumstances, you can see the files on the system and then anybody who wants to play unfair can just delete them. So, how can they make it so that people can't see them? They alter the kernel - the heart of the operating system - so that it doesn't display them.

The solution to Sony/Bertelsmann Media Group Music Entertainment's DRM problem that they are using in this case is a software product from First4Internet called 'XCP software'. This installs a kernel mode driver rootkit which patches the kernel service descriptor table making itself invisible. All right if they asked you if they could compromise your kernel in this way, except that they didn't and, to make it worse, the program obfuscates its true identity by having a mundane, system-like file name.

So, there is definitely a problem and like many large companies, Sony BMG initially denied that there was a problem. They said that the XCP rootkit trojan; 'is not malicious and does not compromise security.' In an interview on US national public radio, president of Sony BMG's global digital business division said; 'Most people, I think, don't even know what a rootkit is, so why should they care about it?' Res ipsa loquitor.

The sloppy implementation of this has raised a few eyebrows in the security community, especially as it slows down your computer and makes it less stable. It consumes around two per cent of processor power - polling processes around eight times per second. Another problem is that it is susceptible to race conditions which can bring down Microsoft's megalithic kernel. If that wasn't bad enough, it installs itself in such a way that it even runs when you boot into safemode. This particular version hides files, directories, Registry keys and processes that start with '$SYS$'.

Just when you thought it couldn't get worse than that, if you try to uninstall it by almost any means, you may well lose your CD player drives. Sony has given people an uninstaller but apparently, instead of uninstalling the rootkit, it installs another patch and leaves the system in an even worse condition. They have now withdrawn that.

So, how far has it spread? It is estimated from DNS cache queries that the Sony DRM rootkit trojan has contaminated at least one machine on at least 568,200 networks in 169 countries. and from comments on the Internet, it seems to have done a lot to make people decide to switch to Linux as it also affects Mac OSX systems as well.

With their Windows installations destroyed by the installation and possibly an attempted uninstall of the Sony rootkit trojan, there seems no future and because of the way it has gained access to their machines - being illegal in many places - there is an increasing number of law suits against Sony for this including (currently) at least six class action cases in the US in addition to the State of Texas and the State of California and then there are those in other countries as well.

It gets worse

If you thought that Sony's problems were limited to the XCP software, there is another DRM program in use by Sony - some people reckon that there are four such programs. MediaMax is reckoned to be installed on over 20 million CDs and is also outstandingly bad. This one installs files on the user's computer whether they accept or decline the terms of the End User License Agreement. Like the XCP software, it doesn't include any way for the user to uninstall it.

Wait for it... The software then sends data about the user back to SunnComm via their Internet connection every time they listen to CDs even though the EULA is explicit that the software won't collect personal information. SunnComm's website even says says; 'no information is ever collected about you or your computer.'

In order to obtain an uninstaller for the MediaMax software, users have to request it repeatedly and supply further information about themselves. And, like the XCP uninstaller, security researchers have uncovered the fact that SunnComm's uninstaller opens up more security risks. Even the US-CERT (Computer Emergency Response Team) recommends 'Do not install software from sources that you do not expect to contain software, such as an audio CD'.

As Sony has it interests in many pieces of hardware as well as music, people are starting to ask if this type of pernicious DRM software is already in those other product lines. Only time will tell.

Rootkit Trojan Exploit Viruses

Sony/Bertelsmann Media Group Music Entertainment's DRM rootkit trojan attempts to protect itself by hiding files, directories, Registry keys and processes whose names start with '$SYS$'. If you ignore the egregious implementation of this with its stability side effects issues, one problem is that it hides from the user, anything that starts with '$SYS$'. So, it was not long before somebody decided to write some malware that exploited this feature.

Infection by this malware is not limited to Sony DRM rootkit trojan infected machines but on those machines that are infected by the rootkit trojan, detection of the offending files will be more difficult.

At the time of writing, there are already Sony rootkit trojan viruses in the wild so, if you have used a Sony BMG disc that has the trojan, you might not even know it. Fortunately for those in the UK, the CDs with the trojan were aimed at the US market although it is possible for them to have been bought through Internet resellers. If this has happened to anybody in the UK, whether or not you have been infected with a virus, there may be a case against Sony under the Computer Misuse Act.

Fingerprint Drive

A normal USB-style drive can be used to store a great amount of data - you can, if you are using a UNIX-like system, mount one of these in your home directory and use it for that although the access speed isn't what you would be used to if you were using a normal HDD that was connected directly to your computer.

As the data on these is portable, you need to make sure that other people cannot access it. One of the problems with password systems is that the passwords can be picked up by devices as they travel from the keyboard to the computer. In addition to this, password-based devices are limited to using only systems that have the authentication software installed on them. So, ideally, we need some sort of USB flash drive that has self-contained authentication. Looking at the different methods of authentication - passwords, tokens and biometrics - one of the better solutions is to use a fingerprint to authenticate the user(s).

Normally, several fingerprints would be enrolled so that if you had an accident (I'm just talking about a cut finger - nothing too horrible), you could still access the data. The iTube USB token (manufactured by Technoimagia http://www.technoimagia.com/) is one such example of a self-contained fingerprint biometric device.

When you plug it into the PC, you have 30 seconds to put your thumb or other enrolled finger on the fingerprint device and it recognise you. Until then, the storage is not readable by the computer. All of the authentication process is dealt with on the flash drive itself meaning that there are no external data streams to intercept. The enrolment manager program (Windows XP) runs directly from the CD and is not installed anywhere although you do need admin rights in order to use it.

So, being self-contained, this biometric USB drive should be cross-platform - running on any operating system that accepts ordinary USB drives. I tried this one out of Windows XP (on which it worked well) but on SuSE 8.2, SUSE 10.0, OpenBSD and FreeBSD, there were serious problems. Although it was recognised as a device, and mounted in the file system, the read/write rates were very slow on SUSE 10.0.

The next version is currently being developed and hopefully, this will be ironed out.

Testing Fingerprint Devices

Fingerprints have been used for many years to identify people. The first use of a fingerprint in a crime was a bloody thumb print on a wall at a crime scene. Since then, they have been used and modern electronic sensors do a pretty good job of authenticating the right people. However, the right people are not necessarily the only ones trying to get at your data - if only the right people ever were, you wouldn't have bothered with the extra security in the first place.

So, if you go for a biometric device and you feel that you need to test it out yourself, you need to use the methods that are likely to be used by those wishing to break in.

One of the easiest methods is to use a finger print that is already there. If you put your finger on the device and then carefully lift it off again, you should be able to see a fingerprint on the chip. This is not enough to spoof the sensor but if you plug the device in (ie, make it read the sensor) and then breathe on the chip, in some cases, the moisture that condenses on the chip where the fingerprint is will be enough to form a successful authentication.

This used to work on some of the older chips such as the Siemens mouse but they have changed the software so that this is no longer the case but that doesn't mean that it isn't with other chips. A simple work around is (and you really want to use that product) to instruct the users that they should always drag there finger off the device once they have been authenticated so that they smudge any fingerprint that is there. For the iTube, I found that breathing on an existing fingerprint did not work (that is to say that it did not authenticate the user).

So, if you want to try out another method of attack, try the following. First of all though, there are several types of sensor and this doesn't necessarily work with all of them (or any if you don't do it right).

There are essentially two types of attack (assuming that the thing is cryptographically secure enough and your enrolment process is not open to enrolment fraud to a dangerous extent).

• An enrolled user is unaware that they have supplied a copy of their fingerprint; or,
• An enrolled user co-operates by supplying a copy of their fingerprint (by co-operate, I mean either: they are in on it; or, they have been coerced into it using blackmail, been knocked out and had their finger print stolen and so on).

In the former case, you can take a fingerprint either by smoking it with cyanoacrylate or dusting it with aluminium, then photograph it, process the image and print it out then mask a piece of etchable surface (PCB will do, use acid to develop the print) and then use that to cast the dummy finger.

In the latter case, you can do the following (the last part is the same as in the former case) to test out just how good any given fingerprint system is. To do this realistically, you need to have access to the sort of materials that anybody who wants to break in is going to use so...

First of all, you need to make a mould of the enrolled finger so you need something that, when casting takes place, a finger can survive. Clearly there are some materials that take too long a time to set or have a setting temperature that is too high. In addition to that, another thing to consider is the amount of time that it takes for any mould material to set. Ideally, it also needs to be reasonably available as well. You could use a substance that sets rigid or flexible - only the casting made from this mould has to be flexible. Taking these points into consideration candle wax is fairly good and it is what we shall use here. It can be melted in situ and when you press your finger or thumb into it, you do not burn your skin. Also, it is quite good at picking out details and only takes a minute to set. You can get it in just about any shop and it is cheap.

If you are going to move any moulds around - ie, you don't want them stuck to whatever surface you are casting them on - you need to use some sort of oven paper or similar. This is baking paper from the local supermarket. Just tear off a piece around 4" square.

Next, melt some wax to make a small pool. There doesn't need to be that much wax there as you are going to have to wait until it solidifies before you can remove the finger. On the other hand, it doesn't need to be so shallow that the finger pushed right through it when it is making the casting. Also, don't make it too hot or it will burn you. Around 1cm3 will do.

Next, wet the finger and wipe off any excess water - you only need to dampen the finger a little. This has two effects: it hydrates the skin and makes it repel the wax a little more; and, it cools down your finger and speeds up the setting process. When you put the cold, damp finger into the well of molten wax, don't push it all of the way down to the paper because doing so will form holes in the casting when it is too thin. Try to keep the finger around 1mm above the surface of the paper and when it starts to set solid, press a little harder so that any fine details are impressed into the still soft surface. Let it go solid and when it starts to harden properly, you will notice that it starts to separate from the skin spontaneously - there is no need to have to pry it away from the skin.

When the mould is done, there is a nice sharp negative casting of the thumb or finger. Any creases or cuts or any details are all there. It is a good idea to make several moulds as I have found that whilst you are learning how to do this, you will be learning and it is not necessarily the case that your first attempt will work - we don't want to approve a device that will fail, do we? Now, we just need to make a positive casting of this image.

There are no doubt many substances that you can use but here we will concentrate on silicone (on the left of the two pictures on the right) and latex (far right). The silicone is a caulking sealant and the latex is a fabric glue. They are both easy to obtain and both fairly cheap. They are flexible, easy to use and reasonably safe. Like any of these things, you need to be careful that they don't contaminate other things as they are difficult to clean off (fabric, especially carpet - just think of the telling off you will get if you get it on any carpet - use a tiled room). One other thing you need to consider is that whatever you use isn't going to damage the mould until you take the false fingerprint out of it (or preferable not at all). If you choose something that has a working temperature greater than a temperature that would damage the mould then it isn't going to work. One example of this would be using a gelatine melt in a candle wax mould as the mould would start to melt or at least lose its surface detail before the gelatine would have had a chance to set. If you did want to use gelatine, you could make up a solution and paint it on - allowing it to dry out between coats. Both silicone and latex are very good insulators and so some of the sensors that look at capacitance or conductivity might not be fooled by them so you would get better results with gelatine which is essentially a protein framework with water in the interstices (a scanning electron microscope image of gelatine looks like a plastic scourer. Gelatine is actually made from boiling pig and cow bones and skin just in case you were thinking you were enjoying that trifle or marshmallow). One problem with gelatine is that it shrinks when it dries out so you will have to be careful about moisture with gelatine.

For the silicone, squeeze around 1cm3 of silicone out of the tube onto your finger and then smear it across the wax fingerprint mould so that it is fairly even. The left image is the silicone after it has had a while to dry out. For the latex, paint on a thinnish coat and allow to dry. Follow this with another coat. For both of them, you might like to consider having an extra piece - a little like a handle - so that you can peal it away from the mould easier when it is dry. Even though the wax repels water, there was no problem with it repelling the silicone or the latex - they both painted on well and did not try to form a globule. Allow them both to dry before attempting to peal them off the mould and don't let them come into contact with other castings you are making as they will stick together.

When they are dry, peal them off carefully from an edge that isn't too close to the centre (or, if you have cast a 'handle', use that). Make sure that the casting doesn't flop around and stick to itself - keep it flat. When you look at the casting and compare that against the finger you have used, you will see that it is a worrying likeness.

All of the find details are there - even the pores in the centre of the ridges. Both the silicone (left) and the latex (right) have the same amount of detail as far as this experiment is concerned.

Finally, you need to test the sensor. For the iTube, the normal procedure is to plug it into the computer and then put your enrolled thumb on the sensor within 30 seconds. The LEDs on it will be red until you are authenticated upon which point then will turn green. This works with both damp and dry fingers. I put the false thumbprint on my thumb (both the latex and the silicone are sticky), plugged the device into the computer, waited a few seconds and then tried putting the false print on the sensor. I found that when the false thumbprint was dry, neither the silicone or the latex worked. However, if I made them moist (think of someone without access to a source of water such as a tap - they are going to lick the silicone or latex), they both worked. I had to change the pressure that I was applying and move the print around a little before it did though. This could be due to the software looking for a real print moving around slightly whereas a breathed upon grease/epithelial print is inherently static. It could be that as I released the pressure on the print, water/spit was forced to the places where the ridges were in contact with the sensor (capillary attraction) and because we are almost all water, the water has a similar set of physical properties to the genuine skin print. Both of these are speculation but the thing that matters is that the false prints were authenticated on this system. Note that this is with a chip that you place your fingerprint on. For systems that look at your print pressed against a piece of glass or plastic, you might need to use a material that has a similar refractive index to skin (silicone and/or latex might already fall into that category so it is probably worth a try). If you are going to investigate this yourself for a system that you are thinking of using, it would probably be wise to use only your own prints in a system on which only you are enrolled.

You can find out more about this in the interesting document on this at http://wwwcryptome.org/gummy.htm

Biometrics - Thoughts About User Risk

Biometric devices such as fingerprint readers cannot on their own distinguish between:

• somebody who requires access because they need to; or,
• somebody who needs access because they have a gun pointed at their head.

Unlike passwords, fingerprints can be used when the enrolled user is unconscious. This represents a big problem. The usual way around this is to protect the access with a PIN.

Whilst a PIN on its own is very insecure, using it to add to the biometric element of the authentication helps the user. If you have two PINs, one a normal one and one to be used under duress, then if they have a gun pointed at their head, the system can be told to cough up a lump of bogus information without making the assailant suspicious that anything is wrong. Meanwhile the system can contact the police.

This has been done with one example being a pharmaceutical company that produces a large quantity of data from dead-end research to act as a honeypot to the assailant. The data is of no commercial value but the assailant would have to be a pharmaceutical chemist to know that it was of no value.

So, whilst biometrics can protect your data from certain types of attack on the storage medium, it can put your users at risk and you need to think of ways of making it safer.

Downloading Large Files Effectively

ISOs or film DVD images are likely to be the largest files you will download for the time being any way. Many sites will allow your browser to download the image but if the download speed is set to be slow, the browser will often time out with an incomplete download. Also, if the download is interrupted, the download will fail and any attempt to restart the download will likely end with the transfer starting from the beginning again thus wasting the bandwidth that has already been used.

wget is a program that operates from the command line and can be used not only to download large files but also whole websites. It respects the robots.txt files and, running from the command line, means that you can schedule it to run in the middle of the night. Unlike many programs, that run on Windows, wget is free and, it is also cross-platform so what you learn for using it on Windows will not be wasted if your company ditches Windows.

You can get wget from http://ftp.gnu.org/pub/gnu/wget/ (UNIX-like systems), http://space.tin.it/computer/hherold/ (Windows) or http://www.rahul.net/dkaufman/ (DOS)

Using MD5 Hashes

One of the main areas where the accuracy of the contents of a file is required is with a file that has been downloaded. Another is where you have a file of a known state and you want to check that the contents has not change at all. For very short files, it is easy to inspect a file visually or compare it with a backed up file - bit for bit - but this becomes impractical for large files that take a long time to download for for those that it is not practical to keep two copies of.

Disc images such as ISOs need to be perfect for the ISO to perform properly and if you have a disc image that is part of evidence, you need to know that it has not been tampered with. So, instead of comparing two images bit-for-bit (imagine doing that for a 700MB download) you can make a hash of the image and compare that with a hash on the server.

An md5 hash is created by breaking up any length of file into small blocks and then allowing the bits within that block to interact with each other and with any previous block. In this way any change cascades through the numbers and changes them all.

For example, the sentence 'How quickly daft jumping zebras vex' produces the md5 hash '8e6797c98dfac9741c1a9ad464d3bb7d'. If we change just one binary digit, making vex into vax - 'e' is 1100101 and 'a' is 1100001 - we now get a hash value of '8c6acf03b7dc0603d54d72107bfe1c9b'. Note that this is not the same as compression.

You can do your own md5 hashes on your system using the command line. For a Windows system, just use 'md5 filename'. For a UNIX-like system, use the following or look up the man page for md5.

openssl dgst -md5 filename

So, if you download your file and the md5 sum, you can check that one with the one you generate from your ISO instead of downloading the whole thing again.

For the DOS/Windows md5 program, you can open the directory in another window by clicking here.

Blocking Used Ports on LAN with Hardware Firewall

Many people use a desktop firewall to protect their machines but if you have several machines and you use network shares, whilst these machines need to be able to talk to each other, you also need to make sure that they don't talk to any machine that is not on your network.

The best way to do this is to use a hardware firewall. By default, the cheaper ones will block any connection that starts on the Internet so as long as your computer doesn't start a connection to an external machine using one of these ports, you are all right. The problem is that there are ways of getting people to start connections from inside the firewall-protected LAN and these usually involve using a vulnerable browser such as IE or any browser configured to run scripts by default; or, by opening email attachments (or just viewing the email in a browser that has HTML main viewing enabled).

So, how do you stop these connections? Just block them off at the firewall from the inside of the network as well as the outside. Any firewall worth having should be configurable in this way and you should be able to limit connections based on ports both inward- and outward-bound.

To get a port list, look at http://www.iana.org/assignments/port-numbers.

Network Time - Posix and Right

Different times can be particularly confusing, especially when it comes to changing clocks. Perhaps this is most confusing when we live in the time zone at the centre of the definition of time; and coping with time zones is something that other people do. Concepts such as the date line are alien and the idea that if your computer breaks down on a Sunday afternoon in the US, it needs fixing there and then because it is already Monday morning in Japan are beyond our experience.

For our annual changes between our basic time and our summer time or daylight savings time - GMT and BST in the UK - we have to rely upon little rhymes such as: 'In March, march forward; in the Fall, fall back' and so on. However, most of us can remember leap seconds - those curious times when the Greenwich pips have an extra pip - and at the time of writing (November 2005) the last one was in 1998 although by the time you read this, there will have been another one at the end of 2005.

UTC forgets about leap seconds as soon as they have finished and therefore loses them. UTC keeps up with the Greenwich pips. Atomic clock time (TAI) holds onto the leap seconds so it gradually goes out of time from the rest of the world. It makes more sense to have a terrestrial time code that keeps with the seasons (albeit with modern accuracy only a few seconds) so UTC is used for everyday time as we all use the sun. So, sundials and other sun-related time events (such as crop growth and so on) all keep time with a system that forgets about leap seconds.

TAI, on the other hand, is useful in you want astronomical time - say you want to know about solar eclipses going back thousands of years where UTC/TAI-type errors can build up to several hours. It makes more sense to have a time-scale that is not related to what goes on on Earth. So, although there are many other time scales, there are essentially two different ones for your computer: local time and atomic clock time.

Windows was only designed to work on a non-networked, single-user basis and so - even with networking bolted on and all of the so-called multi-user Windows stuff added, it still uses local time. One interesting thing I found out about Windows is that it is considered all right for the machines to be several hours out with respect to their time - the program in Windows XP Home only checks the time server around once a week which is all right if you are not in a networked environment (ironic that it then uses a network protocol to check it though). As we all listen to the radio and therefore use local time, Windows machines trot along in an okay manner for computer beginners who probably aren't particularly aware of what their hardware is really capable of (they're running Windows after all).

For an OS that was designed from the beginning to be multi-user and networked, such as Linux, the BSDs and others too numerous to mention here, having the correct time is important and you are given a choice.

On SUSE Linux and a few others, you aren't given a chance to select anything other than then UTC-based time. However, on OpenBSD - which is designed to be a secure, multi-purpose OS - you are given a choice of different time standards.

If you find your computer clock is around 22 seconds different, when you have set it to your local time, you might not have actually set it to local time as there are two representation of it on OpenBSD. If you click on the 'Control Center' icon and select 'Data & Time', or open 'Control Center' and select the 'Date & Time' section, you will see the dialogue with a number of time options

Towards the bottom on this dialogue box, you can see a combo box with various time zones to select.

If you click on it you will see many and this is where the confusion arises. On the right, you can see 'right/Europe/London' which would seem to be the one you want to select for UK time - after all, it does say that it is 'right' and we all want the right time, don't we?

If you select that, your time will be out by 23 seconds (22 seconds at the time of writing which is why you can see only a 22 second difference in teh screenshot) which might not be what you are after.

Instead, you need to select the 'posix' version (Portable Operating System Interface) which uses UTC.

Now, the time is correct (assuming that that is what you want).

Setting Time from the Command Line

By design, there are sanity checks in the NTP client so that if it is more than a few minutes out, it will bail out by design (letting you know why as well - this isn't an NTPD crash). So, if you just type /usr/local/sbin/ntpd, whilst your ntp client will chug along, making note of the time difference and drift every 64 seconds (usually), it will only move slowly towards the actual time - it does this using adjtime because ntpd is usually invoked with the -x option (ie, by the system as /usr/local/sbin/ntpd -x). It does this so that it doesn't trash the system log files and increments or decrements the time by 128ms at a time. However, if you want it to step to the normal time, just don't use the -x switch. In that case, it will be some time before it decides that it is confident enough to set the time itself (if it is within the sanity check). but when it does, it is all of the way there. So, this is how to set the time and run ntpd (assuming that you have set up the ntpd configuration file (usually /etc/ntpdconf or similar) if you need the time to be correct now as opposed to later (so to speak).

If ntpd is already running, you need to kill it off because if you try to use ntpdate, the socket will already be in use and ntpdate will bail out. First of all, find its PID. You can do this by typing 'top' and if you have hardly any processes running, you will see ntpd running even when it isn't doing anything (you could wait for it to appear when it does), or you can type ps -x | grep ntpd and that will give you the PID you need like so... 24843 ?? Is 0:00.03 /usr/local/sbin/ntpd -x Next, kill it off by typing kill 24843 and now that socket is clear, you can use it to set the time using ntpdate. If you don't like typing the same path all of the time, you need to cd to the directory where ntpd is. This is usually in '/usr/local/sbin'. So, type the following (in red) and you should get what is in green (the prompt is in blue). Here, the local ntp server is at 192.168.1.254. bash-2.05b# cd /usr/local/sbin bash-2.05b# ./ntpdate 192.168.1.254 Looking for host 192.168.1.254 and service 123 host found : obsidian.gem 26 Nov 23:01:31 ntpdate[4090]: step time server 192.168.1.254 offset 26.970277 sec bash-2.05b# ./ntpdate 192.168.1.254 Looking for host 192.168.1.254 and service 123 host found : obsidian.gem 26 Nov 23:01:36 ntpdate[16637]: adjust time server 192.168.1.254 offset -0.000200 sec bash-2.05b# ./ntpd bash-2.05b# Nov 26 23:01:45 emerald ntpd[21608]: ntpd 4.1.74@1.998-r Wed Mar 24 05:29:33 MST 2004 (1) bash-2.05b# Note that I repeated the ./ntpdate command so that I could see immediately that the client's time had been set.

Proposed Changes in Leap Seconds and Effects

One of the nice things about the time system that we use is that you can use the sun to see the time. Things like sundials and so on can be quite accurate with the larger ones displaying the time to within a quarter of a minute.

There is a proposal that the UTC standard should stop dropping leap seconds. This has the advantage that if you want to get to TAI, you don't have to look up the number of leap seconds that have happened in a table. Although it must be said that they happen at the most only twice a year and there was a substantial gap between 1998 and 2005 which didn't really inconvenience anybody who could remember things that happened more than a few minutes ago.

The disadvantage is that time will drift further away from reality - something that it is supposed to represent. The number of people on this planet that need time in relation to things that happen on a solar basis (going to work, watching TV and so on) is far greater than the number of people who need to do things in an astronomical way (astronomers for example).

This has happened once before and it would do for the people making the proposal to take this into account. It was the case that we didn't take into account leap years properly (remember the year 2000 problem - it was, then it wasn't, and so on with ignorance eventually winning out). By 1752, measured time had become so far out relative to the seasons (and the sun) that it was decided that in September, we would miss out a few days in order to bring the calendar back in line with the real world. As a result, the date went from 2nd September 1752 over to the 14th September the following day, thus missing out 11 days. People who thought that their death was determined by the calendar, thought that their lives would be cut short by 11 days and there was rioting.

If we stop dropping the dropping of leap seconds, our terrestrial time would become meaningless over a period of time.

PostScript on Windows and Examples

PostScript is a platform-independent and printer-independent printing language. If you have a program that can output PostScript, you can print it just about anywhere. There are some parts of it that can end up printer specific but you can make it independent if you want.

The problem is that it is largely a printing industry, Mac and UNIX system and Windows, it appears, doesn't like to acknowledge anything that is non-Windows. Even though we are now entering a phase in the development of home and small office PCs where there is more than one type of machine on a LAN, this is still the case. So, if you have a .ps file on your Windows system, how do you print it or preview it?

Well, even if you are stuck with a brainless GDI printer (a Windows printer without the necessary processing to make it platform independent), you can still use its Windows driver on the computer that it needs to be plugged into in order to work if you use ghostview Here, I've included some PostScript files for you to test your printer with - Sudoku puzzles for you to have a go at. You can get to the directory with the examples in it by clicking here. There are 10 PostScript files with the Sudoku puzzles in them (the print out for each has the reference number at the bottom left as you can see in the example image on the right) and you can see the answers on next month's SuperDisc in the form of PostScript files. If you have a PostScript driver, you should be able to preview the files on your screen or print them out to a printer. If you have a real PostScript printer, you just need to copy it to the printer as a raw file.

To see PostScript files, you need ghostview (free) which you can get from http://www.cs.wisc.edu/~ghost/doc/AFPL/get851.htm and to view files, you need GSview. With these installed (read the instructions on the web site), you can read and print postscript files on Windows systems. This is what GhostView looks like on a Linux system. You can choose what paper size to view it on and print it out if you wish.

PDF documents are also based upon PostScript so GhostView will also let you see these as well. This version of GhostView uses the old UNIX interface but newer version uses prettier interfaces (although the functionality is still the same).