Thanks for the additional comments!


On Thu, 17 Jul 1997 02:17:16 -0400 Mike Keitz <mkeitz@JUNO.COM> writes:
>On Wed, 16 Jul 1997 21:36:22 EDT Harold Hallikainen
><haroldhallikainen@JUNO.COM> writes:
>Using gotos, calls, etc. in the ISR means saveContext needs to handle
>PCLATH if you're using more than the first 2K code page (unless copies
>of the routines are stored in the same place in every page).

        Highest code address is 0x01a7, so it doesn't look like that's
the problem.

> Also it
>may be overflowing the stack, if the main program is doing a lot of
>calling around or the ISR continues to call deeper and deeper.   Since
>these routines are almost certainly only called from one place, it may
>work better to goto them, then go back to the next bit test when done
>rather than return.  This will save on stack space but not help the
>PCLATH problem.

        I don't think this is the problem in that the new enableIRQ is at
the same subroutine
depth (one call from the main loop).  It seems that if the problem were
stack overflow, it would still be there.  I would need to move UP a level
to eliminate the problem.  Here's another code fragment.

wn0                             ; Waiting for -DataRq to go low
        btfsc   porta,4         ; test bit
        return                  ; if not yet low, return
        movfw   hostReadAddressHi ; Set up RamPort address
        disableIRQ              ; RamPort code not reentrant
        movwf   ReadAddressHi
        movfw   hostReadAddressLo
        movwf   ReadAddressLo
        call    ReadRamPort     ;read it
        movfw   ReadData        ; get results
;       enableIRQ               ; OLD enable location... moved down a
bit
        movwf   wnData          ; and save it
        iorlw   0xf0            ; Set high half so we don't make
-DataAck low right away
        movwf   porta           ; Output the data
        enableIRQ               ; Enable IRQ after outputting data in
case IRQ messes up W (shouldn't!)
        bcf     porta,5         ; Set -DataAck low
        incf    wnState,1       ; Go wait for -DataRq to go high
;       return
wn1                             ; Waiting for -dataRq to go high
        btfss   porta,4         ; test bit
        return                  ; Exit if host has not yet taken
nybble



        So... I got it working by moving the enableIRQ.  BUT, it
should've worked before, I think.  I'd sure like to understand what's
going on!

Harold