News for hackers. Stuff that matters.
Re:Audi Cruise Control Bug by peeters on Monday January 12, 2015 @10:16AM (#8874238)
I call bullshit on this analysis. There's no way the code would be using integers to represent speed (they'd just be rendered as integers). >= and > are essentially the same operation when comparing floating point values.
Re:Audi Cruise Control Bug by wjy on Monday January 12, 2015 @11:01AM (#8874570)
Many processors in a modern car have no floating point capability whatsoever. Fractional numbers are represented using integers - "fixed point" representation. While it's almost certainly true that vehicle speed is calculated as a fractional value, the cruise control set speed may be an integer, as it's set in increments of 1kph (or mph, depending). There's no need to keep smaller increments.

It all depends on which processor is handling the cruise control interface, and how they wrote the software. There's no way to know without more information.

Re:Audi Cruise Control Bug by webXL on Monday January 12, 2015 @11:44AM (#8874824)
The analysis doesn't claim that they're comparing integers, just the fact that it appears that you can only activate once you're very close to 31km/h or higher. It seems to contradict the manual, and therefore might be a bug or at least incomplete documentation.
Re:Audi Cruise Control Bug by nullc on Monday January 12, 2015 @11:04AM (#8874602)
Funny. If the story was different and arguing something on the basis of floating point; I would have been posting saying that its very unlikely that the firmware was using floating point (or even running on hardware with a fast FPU).