PPRuNe Forums - View Single Post - FMS vulnerabilities highlighed at Net Security conference
Old 13th Apr 2013, 14:50
  #57 (permalink)  
Ian W
 
Join Date: Dec 2006
Location: Florida and wherever my laptop is
Posts: 1,350
Likes: 0
Received 0 Likes on 0 Posts
Sciolistes

You are making a series of assumptions of the design of the CCS. Unfortunately, from what TESO said the designers made the assumption that no bad guys would have access. So all the normal protections that the average paranoid designer would put in may not have been put in. Remember every single line of code put into a system being certified is a huge cost in testing so why put hacker defensive code into a system or more complex design 'when it is unnecessary"?

"In addition, I would have thought the operating system at the hardware level would not permit any code to be executed that has not been verified or validated through some kind of checksum to guard against memory corruption (for whatever reason). Never mind the basic issue of illegal memory address access."

The CCS would appear to be a single computer that allows the programs to play in the one big sand-box. This is not a dedicated firmware system on a VME board.

"Even disregarding all that, the whole fundamental basis for Teso's assertions, as he states in the slide pack, is the ability to audit the code to look for vulnerabilities, and I can't see how he or anyone else without a role directly related to development of the relevant elements of the source code, would ever get access to the source."

He actually bought the systems on Ebay. The systems he bought included the FMS, simulation systems and simulators of ACARS message generators. In other words all the bits he needed were available for pennies on line. The code was the same as the operational code. At one stage I used to work on system maintenance starting at the machine code level and work up through assembler to the high(er) level code. So you have all the bits all you need is the patience to see what ACARS hack works. Then repeat that on another FMS type and so on see if you can find a common exploit. If I can not only get in but get in as a 'super' user or maintenance level then I can start convincing the software in the aircraft that is really software running in a simulator so take these inputs not those. Or put out these ADS values instead.

The best way out as I said earlier is to have some gatekeeper firmware watching over the comms links and bouncing and reporting any broken messages before they get to the normal too-trusting avionics. My worry would now be that someone before Tesa could have been in and 'done a malicious update' of the FMS code. And that malicious code is now just waiting for a specific legal message or a particular date - in other words a standard trojan timebomb. So perhaps airlines should think about doing a complete clean software reload just in case I don't know what the normal maintenance cycle is for CCS software. Now the exploit idea is publicized there will be those trying their hand at it and there are several things that could be easily done that could cause chaos. I would write more but I don't want to provide hints to the 'black hats'
Ian W is offline