While looking into the failure of test 48272 (addn), I realised that we
were using both target_address and target_spec in the operation struct.
The pblock is a bit inconsistent about their usage also.
I've attached a list of the places we use them.
I think that this is a mistake, we really only need "one" target value
I want to propose that we drop target_address for target_spec, update
the pblock to use operation_set_target_spec as needed.
However, I don't know the full history of these values, so I would like
your input to this design. Is there a difference between target_spec and
Red Hat, Brisbane