division of two signed16 bit numbers

[shomikc]

Hi. I am trying to learn Atmel AVR microcontroller assembly programming.

I am trying to do the division of two signed 16 bit numbers and I have used the code from the AVR200 webpage but when I rum the program it does not go the quotient registers at all. I have no idea what is going on. Please can anyone help. The code I have used is here. And also when the program returns it does not go to the forever loop but instead it goes to start.

.def      d16s      =   r13        ;    sign register
.def      drem16sL  =   r22        ;    remainder low byte
.def      drem16sH  =   r23        ;    remainder high byte
.def      dquo16sL  =   r24        ;    quotient low byte
.def      dquo16sH  =   r25        ;    quotient high byte
.def      dd16sL    =   r26        ;    dividend low byte
.def      dd16sH    =   r27        ;    dividend high byte
.def      dv16sL    =   r18        ;    divisor low byte
.def      dv16sH    =   r19        ;    divisor high byte
.def      dcnt16s   =   r20        ;    loop counter

start:
           ldi      dd16sL,low(00)
           ldi      dd16sH,high(10)

           ldi      dv16sL,low(00)
           ldi      dv16sH,high(10)
           rcall    div16s           ;    result:    $f752 (-2222)
                                     ;    remainder:    $0002 (2)

forever:   rjmp    forever


;***** Code

div16s:    mov   d16s,dd16sH     ;    move dividend High to sign register
               eor   d16s,dv16sH     ;    xor divisor High with sign register
               sbrs  dd16sH,7        ;   if MSB in dividend set
               rjmp  d16s_1
               com   dd16sH                       ;    change sign of dividend
               com   dd16sL
               subi  dd16sL,low(-1)
               sbci  dd16sL,high(-1)
d16s_1:    sbrs  dv16sH,7        ;     if MSB in divisor set
              rjmp  d16s_2
               com   dv16sH          ;     change sign of divisor
               com   dv16sL
               subi  dv16sL,low(-1)
               sbci  dv16sH,high(-1)
d16s_2:    clr   drem16sL        ;     clear remainder Low byte
               sub   drem16sH,drem16sH  ;     clear remainder High byte and carry
                ldi   dcnt16s,17      ;      init loop counter

d16s_3:    rol   dd16sL          ;      shift left dividend
           rol   dd16sH
           dec   dcnt16s         ;      decrement counter
           brne  d16s_5          ;      if done
           sbrs  d16s,7          ;      if MSB in sign register set
           rjmp  d16s_4
           com   dquo16sH        ;      change sign of result
           com   dquo16sL
           subi  dquo16sL,low(-1)
           sbci  dquo16sH,high(-1)
d16s_4:    ret                    ;    return
d16s_5:    rol   drem16sL         ;    shift dividend into remainder
                rol   drem16sH
               sub   drem16sL,dv16sL  ;    remainder = remainder - divisor
                sbc   drem16sH,dv16sH  ;
               brcc  d16s_6           ;      if result negative
                add   drem16sL,dv16sL  ;    restore remainder
               adc   drem16sH,dv16sH
               clc                   ;    clear carry to be shifted into result
               rjmp  d16s_3          ;    else
d16s_6:    sec                   ;    set carry to be shifted into result
                rjmp        d16s_3

 

[barry]

I’m not really that familiar with the AVR, but are you running a debugger? If so, you should be able to trace your code. If you don’t have a debugger or simulator, get one.

 

[KlausST]

Hi,

it does not go to the forever loop but instead it goes to start.

* is the power supply stable
* is the stack pointer correctly initialized
* is the watchdog switched OFF?

****
Simple debugging method to see what's going on:
* use port outputs and output unique patterns at several places in your code. Check the states with LEDs or scope.
A scope gives better information, because you see very short pulses and you can measure the timing.

Klaus

 

[FvM]

Do you actually want to load all zero to the registers, e.g. high(10)? Which result do you expect for 0/0?

 

[shomikc]

Do you actually want to load all zero to the registers, e.g. high(10)? Which result do you expect for 0/0?

I didn't understand the question.

 

[FvM]

Which result see your in your calculation, which do you expect?