• Recently Written

  • Manifesto
  • Garbage
  • Natural
  • Symbian
  • My
  • It
  • Can
  • In
  • Mocking
  • import

• Categories

  • Blogroll
  • Coding
  • General
  • History
  • Personal
  • Rants

• Archives

  • June 2010
  • September 2008
  • February 2008
  • January 2008
  • December 2007
  • November 2007
  • October 2007
  • September 2007
  • August 2007
  • April 2007
  • March 2007
  • February 2007
  • January 2007

• Blogroll

• Admin

  • RSS
  • Comments RSS
  • Wordpress
  • Login

  Search

Our precious little computers live in a dangerous word. Viruses, attacs, malicious software of different kind is out there. If user is not diligent in updating his system, antivirus, turning firewall on and guarding himself not to run a suspicious software sent by mail or downloaded from web, (s)he can hope that the computer will be safe. It's such a pain, to make computer safe. Couldn't we do better?

Current security model is based on what Thompson and Ritchie invented for their UNICS operating system in early 70s. In simplest words, it allows to declare who can use a file: user himself, his group(s) or everybody. That means that if you trust a user, you trust all his programs as well.

Unfortunately the world is not this simple any more. Given omnipresent Internet, malicious programs can disguise as a well-behaving one or abuse another application to make it run nasty code. Can't we do better?

The guys over at OLPC are apparently trying to make security better. Since they don't need to be backward compatible with anything, they invented whole new security system. It is called Bitfrost and can be found here.

There are some important assumptions they make. There are three which in my opinion are much more important that the rest. One is assuming user does not know anything about the security and is not even able to remember a password. The second one is that applications cannot be trusted. The third on is promoting open collaborating, including sharing code and applications.

The target group for OLPC projects are small children, ranging from age 5 to 10. That means the security system cannot assume that user will expose any security awareness. It's the system that should protect the computer.

Experience shows that the application cannot be trusted. They can be taken over or otherwise exploited to run any code. On the other hand, third assumptions requires that any piece of code can be run, even those malicious ones. From first assumption we deduce that the user cannot tell the difference from well- and ill-behaving code.

The constrains are interesting. I think Bitfrost provides the security infrastructure which "just works". OLPC guys even managed to fit in some additional security features, like anti-theft measures. Go ahead and read the details.

Comments