Well, not exactly ... It's not pointless, but I haven't seen any microcontroller with "broken" ports over the last 30+ years. I've encountered some that were defective as shipped and would have been caught in normal incoming inspection, but I wonder what would motivate your boss to have you expend effort and commit resources on your product for the purpose of checking port function. Moreover, there are probably other ways of determining whether the ports are functioning correctly without dedicating hardware specific to that task. After all, those ports are doing something, and that's probably observable from the MCU. Since I know nothing about the rest of your application circuit, I can't suggest any other way to ensure that the ports work, but I'd bet that there are ways available that don't require dedicated hardware for this power-on-self-test alone.

However, the fact is that, just because I personally have not encountered it, doesn't mean that ports don't fail.

You, as the design engineer, have the task of ensuring that the ports are not driven beyond their nominal limits, as stated in the datasheet. I'd point out that driving them to the "absolute maximum" spec's in the datasheet is not advisable. There are limits on Vih/Vol and Ioh/Iol, that, if adhered to, will prevent damaging your MCU.

Perhaps this is a matter of economics. If it is cheaper to test the devices in-circuit each time it is power on, it may cost less than having you devise the built-in-self-test features. It may be a matter of reliability, meaning that it is extremely important to know whether the ports are functioning properly, so one doesn't rely on information that may be faulty.

Do you perform incoming inspection on your components? How thorough is it? Is it a spot-test, or is it an exhaustive test? How extensive is the incoming inspection? Surely you don't simply rely on the distributor to ship you "good" parts.

RE