By Carsten Eiram
Normally, we spend our time buried in deeply technical analyses of vulnerabilities, but that doesn't prevent us from enjoying the simple things in life like when a vendor designs a security feature completely without considering the "security" aspect of it.
This morning, I was talking to a co-worker outside the Secunia Research department. He had just bought a shiny, new Google Android-based HTC Desire cellular phone and was quite excited about the gadget and all the stuff it could be used for. Cell phones have come a long way and are sometimes even invaluable tools in companies for keeping track of ones contacts and schedule as well as storing various personal and business sensitive information.
Considering all the valuable information stored on such devices, one would hope that they were developed with security in mind to e.g. prevent people from gaining unauthorised access; it should at least be a worthy challenge.
When seeing the "connect the dots" screen-lock feature used for preventing access to the device, it immediately seemed like the idea failed on so many levels. For people, who are not familiar with how this works (including myself about a couple of hours ago), the screen-lock presents you with 9 dots with three rows of dots, having three dots in each row. To gain access, the user has to correctly connect the dots.
This happy owner is by no means a security unconscious person so he had actually given the correct combination (or "unlock pattern" as it's called) some thought. While the combination has to at least consist of four dots, it is, unfortunately, not possible to re-use a dot. This seriously reduces the number of different ways of connecting the dots. Considering most right-handed would start drawing from the left-hand side and most users would pick a gesture that "lies in the wrist", it didn't seem unlikely to gain access without too much of a hassle; it took only five tries…
After a bit of gloating, I was ready for round two to try out a couple of other ideas for gaining access in a straight-forward manner.
This time, I didn't even worry about going for a "brute-force" approach (if casually dragging your finger over a screen can be considered brute force). Instead, I looked at the screen and took note of the smudge marks. Normal use is to press the screen, dragging upwards and downwards, or dragging from side to side. There is a good chance that any other smudge marks may be from the unlock pattern – especially if the user only briefly used it to e.g. make a call before locking it again. It proved to be a simple, but effective approach for gaining access as well. Don't forget to wipe off the screen after use!
The last approach may have limited effectiveness in real life, but was sweet nonetheless and implemented a bit of psychology to unconsciously make the user give me his "password". For this one I needed a new prey and quickly found one. Instead of immediately throwing myself at the phone, I faked interest in how it worked and upon being presented with the screen-lock, I casually asked how it worked. The phone's owner answered: "Oh, you just have to connect the right dots" while unconsciously making a gesture with a finger, matching his unlock pattern. Afterwards, I bet him that I would be able to access his phone, which he didn't believe that I could – he was wrong.
Ultimately, this is a perfect case of a vendor designing a feature (even a security mechanism), but thinking more about the "gimmick factor" than security. We often focus on really elaborate, technical schemes to bypass security mechanisms, but sometimes there may be an awfully simple approach.
In this case, the idea of a unlock pattern screen-lock may not be all that bad. However, it should at least allow re-use of dots and preferably also support separate dot presses instead of just dot connecting to ensure a reasonable number of combinations.
Chief Security Specialist