It appears there are no perfect ways of detecting anything to with hardware-related issues - anything to do with driver support, availability, interfacing with the o/s, etc, it's all a bit sketchy, you just have to do what you can and fake it the rest of the time.
The file idea is good, I am using this also to check for previous crashes as a last resort.
The file idea is good, I am using this also to check for previous crashes as a last resort.