At the risk failing to tread lightly into a debate, I agree Eric (and others) are right about the port description. The best explanation is in the old Intel data books, which actually shows the port I/O design at the transistor level. Which brings me to my point.

The I/O configuration was patented by Intel (which may now be expired, I’m not sure). Historically, vendors had to pay a royalty to Intel to use it, some did (such as Phillips and OKI) and some did not. I personally worked on an instantiation of the 8052 from a 3rd party vendor that did not use the Intel I/O design and the application programmer did have to specify each I/O, as a different buffer (output or input) would be connected to the pin. This was not standard but not difficult to deal with once you understood it.

That processor is still out there and others (such as “soft” versions) may not use the standard I/O buffer for similar reasons. It may also apply to “some modern derivatives” as well.

It’s always best to read the datasheet.

Stephan