We trust that the processors in our computers run code faithfully, but it is essentially impossible to verify that a nontrivial piece of silicon will behave according to its speci cation. We know that processors have shipped with bugs before, because such bugs have been found by users| most famously, the Pentium oating-point division bug of 1994. But what if a processor ships with a malicious backdoor?
a. What is the threat model for malicious processors? Assume that a processor modi ed by the attacker has made its way into vulnerable computers. How will the attacker access those computers to be able to influence their behavior? How might she want to cause those computers to misbehave?
b. From the attacker's perspective, the best processor backdoor would allow the attacker (who knows of its existence) to subvert the running system, while not being easily detected in testing.
What is the very simplest backdoor you can think of to a CPU that satis es these criteria?