With this, we can: * use profilers to estimate how many cycles we spend on these checks (subject to caveats), * more easily see why we crashed.