The internal pull-down nFET at port output works similar to a voltage controlled resistor: The less drain source current flows, the less voltage drops across the channel. So, without any load (only the high ohmic oscilloscope or voltmeter connected) the VOL should be less 0.1V, as stated in the datasheet. Seeing in this case a VOL of 0.92V is way too high!

There can be many reasons for that: When measuring the output voltage by connecting the probes to the corresponding output pin and ground, you might have chosen an improper ground point, which by itself is contaminated by noise, hum or voltage drops resulting from other (high) currents. So, connect the voltmeter directly to the pins of micro for the measurement.

But even then, if there are high currents sinked into other ports at the same time, then the internal ground on die can be lifted by some hundreds of mV, leading you to believe that your port output shows a wrong VOL. Make no mistake, if this is the case, i.e., if into another port(s) such a high current is sinked, that your port can no longer emit regular TTL levels, then this current is way too high and you must change your design.

Another possible reason for your strange measurement could be, that you forgot, that the output isn't emitting low state all the time, but additionally brief periods of high state. I.e., that your output is emitting a somehow shaped high low pulse train, where the slowly voltmeter can only take the average, which then must be much higher than a plain VOL only, of course.

There are several ways to find out what is going on:

1. Take a fresh micro into your application and check the same potentials. If you again observe the strange reading, then the micro is probaly ok, but your measurement or design is faulty.

2. Disconnect all loads from the micro. Write a short test code, which permanently emits low state at all the port outputs. This code might contain a few globally configuring instructions (watchdog, or whatsoever), instructions to write "0" into the port SFRs and afterwards an endlees loop, preventing the micro from changing the ports again.
Now check the potentials at all port outputs. If you measure anything higher than 0.1V (corresponding to a load current of less than 10µA) then your micro is very probably damaged. But take care, that you measure properly!

Of course, the VOL of ports are not identical, usually. But the variation on the same chip is rather small. To see a variation of more than 0.2V at load currents of less than 4mA is rather improbable.

Kai