I should have included the context-saving code.  As Michael guessed, the
code snippet I posted earlier is CALLed from the actual interrupt
vector, and the RETURN shown in the snippet comes back to this
context-saving routine.  The RETFIE is at the end of it, so that isn't
the problem.



;**********************************************************************


        org     4             ; Interrupt Vector

;**********************************************************************

save  movwf      temp_w      ;save w contents

       swapf      STATUS,0    ;swap status, result in w

       movwf      temp_s      ;save status contents (swapped)



       call inter



intret swapf      temp_s,0    ;re-swap status, result in w

       movwf      STATUS      ;restore status

       swapf      temp_w,1    ;swap, back in file

       swapf      temp_w,0    ;re-swap, result in w

       retfie                 ;return with proper w contents

;**********************************************************************





>PREVIOUS POST:



Wise Ones:



I am a moderately experienced programmer, but I am completely puzzled by
the behavior of this code fragment.

I've been stuck on this all day. I'm obviously overlooking something.
Can anybody help?



This is inside an ISR and is intended to set a flag and clear a register
after counting down a number of TMR0 overflows.



  inter

            (snip)



            decfsz   ir_cnt     ;counts down 8.9ms

             return             ;return if not yet counted down to 0

  irtst     nop                 ;test point for IR countdown

            movlw   b'01110000' ;clear any latent external interrupt
(from IR detector)

            movwf   INTCON      ;  and re-enable external interrupt

            nop

            bsf        flags_5,1      ;set this flag denoting end of
8.9ms period

            clrf        ir_dat

            return



========================================================



This is what happens, as seen by single-stepping in MPLAB IDE v5.61:



  irtst     nop                 ;test point for IR countdown



When the program reaches this point, watch window shows INTCON =
b'01100110'



            movlw   b'01110000' ;clear any latent external interrupt
(from IR detector)

            movwf   INTCON      ;  and re-enable external interrupt



SURPRISINGLY after the movwf the watch-window shows INTCON = b'11110100'


How can b2 (TOIF) still be set ?



            nop



SURPRISINGLY after executing this NOP, the watch-window shows INTCON =
b'11110110'

Now b1, (INTF) is ALSO set, even if I (temporarily) physically strap the
RB0/INT pin HIGH

in order to be sure that no external interrupt can occur !



            clrf        ir_dat

            return



I just don't understand how this can be happening.



========================================================



Here are my INTCON and OPTION_REG setups:



INTCON setup

        movlw b'11110000'         ; GIE=on TOIE=on (peripheral AND
timer0 overflow

        movwf INTCON              ;    AND ext-int set up)



             ;  x-------          ; 7, GIE: Global Internet Enable (1 =
ints enabled)

             ;  -x------          ; 6, PEIE: Peripheral Interupt Enable
(1 = p. ints enabled)

             ;  --x-----          ; 5, TOIE: TMR0 Overflow Interrupt
Enable (1 enables interrupt)

             ;  ---x----          ; 4, INTE: RB0/INT Ext. int (1 enables
ext ints)

             ;  ----x---          ; 3, RBIE: RB Change Interrupt Enable
(1 enables port change int)

             ;  -----x--          ; 2, TOIF: TMR0 Overflow Interrupt
Flag (1 = overflowed)

             ;  ------x-          ; 1, INTF: RB0/INT External Interrupt
Flag (1=external int.)

             ;  -------x          ; 0, RBIF: RB Port Change Interrupt
Flag (at least one of RB7:RB4

                                  ;               has changed if this
flag is set



OPTION_REG setup

        movlw        b'00001000'

        movwf OPTION_REG          ; load data into OPTION_REG

           ;

                ;  x-------       ; 7, 0=enable, 1=disable, portb
pullups

                ;  -x------       ; 6, 1=/, int edge select bit
0=falling

                ;  --x-----       ; 5, timer0 source, 0=internal clock,
1=ext pin.

                ;  ---x----       ; 4, timer0 ext edge, 1=\

                ;  ----x---       ; 3, prescaler assign, 1=wdt, 0=timer0

                ;  -----x--       ; 2,1,0, timer0 prescaler rate select

                ;  ------x-       ;   000=2, 001=4, 010=8, 011=16, etc.

                ;  -------x       ;

                                  ; Note! We set the prescaler to the
wdt, so timer0 overflows

                                                ; every 256 instructions
to make an interrupt every 102.4 uS

                                                ; (for 10 MHz clock)



TRISB is set for all outputs, except for RBO/INT which is used as an
external interrupt



===========================================================



Obviously I have to be able to clear the TOIF and INTF flags, but I
can't seem to do it.

I have tried different '877's on the off chance that the chip was bad,
but no joy.

Power Supply voltage looks good, with appropriate 0.1uF bypass at chip
pins

Interrupt context-saving is implemented



I would be so thankful if someone can tell me what I am missing here.



Thanks,



  Foster






--
http://www.piclist.com hint: To leave the PICList
mailto:piclist-unsubscribe-request@mitvma.mit.edu