Hello Daniel,
Even though the 42-pin version of the device does not have PTC0 and PTC1 bonded out, these bits still exist internally. It is perfectly reasonable to set these as outputs, (and a very good idea so as to prevent a floating input situation). I can't see that this would be the cause of your previous problem.
When you experienced the problem, were you using the actual hardware for the test, or were you simulating the hardware in full chip simulation mode?
I would also suggest that, while you are experimenting, that you disable COP timer operation. Unless properly handled, this can cause a reset.
Regards,
Mac