Hot Fuzz: Assisting verification by fuzz testing microelectronic hardware

The task of verifying microelectronic hardware designs is as crucial to the design process as it is tedious. Despite numerous helpful methodologies like constraint random testing, it takes an experienced engineer to find hidden bugs and unintended system behavior. Conventional testing approaches are centered on individual test cases in order to test specific scenarios. Analog to the well-established software technique of fuzz-testing, we present What-The-Fuzz (WTF), a coverage-guided mutation-based fuzzer and demonstrate it on an example circuit. Test cases are generated in an automated fashion by consecutively mutating input stimuli, guiding them to achieve increased coverage. In contrast to purely random test cases, we avoid the vast majority of trivial noise-like invalid inputs and focus on test cases that actually result state transitions of the tested device.