Luigi Rizzo <luigi@LABINFO.IET.UNIPI.IT> writes:

> Two students of mine have written a 16C84 simulator, in C++.  It
> simulates correctly all peripherals (timer, interrupts, io ports,
> EEPROM writes), and CPU instructions (including the delay slots
> after every modification to the PC -- something that the Microchip
> simulator does not do).

Do you mean that the Microchip simulator only counts one cycle for instructions
that write to PCL?  Are they really that inept?

Of course, the Microchip documentation erroneously states

        If an instruction causes the program counter to change (e.g. GOTO)
        then two cycles are required to complete the instruction

This would imply that the goto in the sequence

        goto    next
next:

or the addwf in the sequence

        movlw   0
        addwf   pcl

should only take one cycle, since the program counter isn't changed.

What they really should have said is that instructions that *may* change
the program counter always take two cycles.

While we're on the subject, has anyone tested the behavior (on real parts,
not a simulator) of writing to the PCL *indirectly*?  I.e.,

        movlw   pcl
        movwf   fsr
        movlw   label
        movwf   ind     ; should take two cycles
        incf    foo     ; should not execute
        ...

label:  incf    bar     ; should continue here

It wouldn't surprise me one bit if the part didn't notice that the indirect
write affected pcl, and actually executed the incf foo.



Cheers,
Eric