divisor by 0 exception error

[beyondsociety]

When I enable interrupts, I get a divisor by 0 exception error.

How would I check if there is a divisor by 0 or not?

[Ozguxxx]

If you are sure about that you are not dividing anything by 0, maybe problem is in that you are getting irq 0 rather than int 0 which is timer hardware interrupt. Are you properly handling hardware interrupts AND/OR did you disable unhandled hardware interrupts via 8259?

[Curufir]

Remap the PICS.

[beyondsociety]

I fixed it. I had disabled all ints except the irq0 (timer) and this was causing the computer to throw a divisor by 0 exception error.

Thats for the help Ozgunh82!

[Schol-R-LEA]

When I enable interrupts, I get a divisor by 0 exception error.

How would I check if there is a divisor by 0 or not?

Well, first off, I would be suspect of the exception arising at all. Division simply isn't a common operation in system code, and if it is coming up that quickly and regularly, I would definitely wonder why it would be happening. What is telling you that it is a divide-by-zero error - your interrupt handler, or the simulator you're running it in (Bochs, Plex86, VMware, etc)? Are you sure that you have the PIC reprogrammed correctly, that the IDT has the correct interrupt vectors, etc.?

In any case, it is still a good idea to check for possible division-by-zero error any time you have a variable divisor, in order to avoid that error. you need to determine each instance in your OS code where you are performing a divison. If the divisor is a constant, then there shouldn't be a problem; but if the divisor is a variable, then you'll want to add a check to make sure it isn't zero. Thus instead of

Code: Select all

foo = bar / (baz - quux);  // will cause DivZ error if baz == quux

you would use an intermediate value to test the divisor, and add a follow-up to handle the problem cases:

Code: Select all

xoom = baz - quux;  // intermediate value holding the the divisor of the next expression

if (xoom != 0)
   foo = bar / xoom;
else 
   {
     // do whatever you need to do instead of the division
   } 

This is actually easier in x86 assembly, as you can usually take advantage of the order of operations, so that for

Code: Select all

   mov ebx, [baz]     ; move baz into ebx register
   sub ebx, [quux]   ; and subtract quux by it
   mov eax, [bar]    ; move bar into eax register
   div ebx           ; and divide it by the result of baz - quux
   mov [foo], eax    ; save result in foo
   ; ...    

you would only need to add a jump-on-zero after the subtraction:

Code: Select all

   mov ebx, [baz]   ; move baz into ebx register
   sub ebx, [quux]   ; and subtract quux by it    

   ; because a subtraction resulting in zero sets the clear flag 
   ; there is no need for the intermediate value 'xoom' 
   jz handle_zero_case  ; if the result is zero, don't do the divsion

   mov eax, [bar]   ; move bar into eax register
   div ebx          ; and divide it by the result of baz - quux
   mov [foo], eax   ; save result in foo

handle_zero_case:
   ; ...    

HTH.

[beyondsociety]

What is telling you that it is a divide-by-zero error - your interrupt handler, or the simulator you're running it in (Bochs, Plex86, VMware, etc)?

My interrupt handler is throwing the error. I noticed that it also happens when I try to install my exceptions. Because of this, I am in the process of re-writing it.

In any case, it is still a good idea to check for possible division-by-zero error any time you have a variable divisor

I will check for them and apply your method you posted to them.