::Though I wonder how often one wants to read the input pins, then write the output latch with A, and then write the input pin value into A::

Are you by chance designing a MCU core or architecture by chance ? The RMW stuff is internal to the MCU and the Philips document very clearly lists out the instructions that use this process. As such cannot understand your statement,

::Oh well, have to implement it that way anyhow::

Raghu