0

I'm not claiming that this has already happened. My question is, how can I make sure that there is not an undocumented CPU instruction designed intentionally to be used as a backdoor? An illegal opcode or a sequence of instructions, which, when executed, would lead to, for example, jumping from user mode to Real Mode, which further can be used for privilege escalation or arbitrary access to physical memory from user mode.

This can be used as a weapon in emergency cases.

Also, for preventing other people from using that accidentally or after the discovery, they can define a public key and also a private one, which allows to execute that behavior, when it comes with the sequence of the key stored in some CPU registers and/or a sequence of stack memory values and/or a sequence of future instructions. I'm well-aware this may sound like a conspiracy theory!

cadaniluk
  • 15,027
  • 2
  • 39
  • 67
Mehrdad Nazmdar
  • 184
  • 10
  • Generally speaking you can't and while I'm categorically ruling out the existence of intentional loopholes I'd be more worried about [silicon bugs](https://en.wikipedia.org/wiki/Pentium_F00F_bug) and left-over test features allowing privilege escalation. Validating that a processor implementation lacks any suck holes is the hardware equivalent of proving program correctness. That to say it is only practically doable with great care for limited cases, on systems specially designed to make validation feasible. – doynax Nov 01 '15 at 09:10
  • 1
    This is a good question but it should be asked on [security.stackexchange](https://security.stackexchange.com/) or, even better, [skeptics](https://skeptics.stackexchange.com/). Here on SO there's almost nothing to discuss. – edmz Nov 01 '15 at 09:43

0 Answers0