Simple problem with if-then statements, where am I going wrong?

[aaron800]

I am very confused why the code is doing the following in a simple if-then structure:

interrupt 12 void MTIM_ISR(void){

    MTIMSC_TOF = 0;
 
    if(i== 0)
    {
      PTBD_PTBD4 = 0;
      PTBD_PTBD5 = 1;
      PTBD_PTBD6 = 1;      // JUMPING FROM HERE
      i++;
    }
    else if(i==1)
    {
      PTBD_PTBD4 = 1;
      PTBD_PTBD5 = 0; 
      PTBD_PTBD5 = 1;
      i++;                             // TO HERE, AND THEN OUT
    }
    else
    {
      PTBD_PTBD4 = 1;
      PTBD_PTBD5 = 1;
      PTBD_PTBD6 = 0;
      i= 0;
    }
    
   
}

This is just a interrupt handler for a timer-based delay.  Is the compiler optimizing the code in some weird way?  This basically blinks three led's in order based on the i index, and then starts back over at the first led (i=0).

Thanks,
Aaron

[aaron800]

Hello, yes i corrected the mistake, and everything works now.  Thanks again, sorry for my newbie question.

[aaron800]

tony, you're right, i am an idiot, I think i'm dyslexic sometimes :-P

I still don't see why it is jumping though, unless the optimizer says that is the same thing

Thanks

[bigmac]

Hello Aaron,

I tried compiling your sample code, and then single stepping through the ISR under full chip simulation.  I found that the code does work as intended (apart from the typographical error previously observed).

The compiler seems to generate the following assembly code for the ISR (I have added the labels):

    PSHH

    LDA   i

    BNE   B1

    ; Case: i = 0

    BCLR  4,PTBD

    BSET  5,PTBD

    BSET  6,PTBD

    BRA   B2

B1: DBNZA B3

    ; Case: i = 1

    BSET  4,PTBD

    BCLR  5,PTBD

    BSET  5,PTBD  ; Typographical error

B2: LDHX  #i

    INC   ,X

    PULH

    RTI

B3: ; Case: i > 1

   BSET  4,PTBD

    BSET  5,PTBD

    BCLR  6,PTBD

    CLRA

    STA   i

    PULH

    RTI

Because of the typograqhical error, the LED connected to PTBD5 will never flash.

Regards,

Mac

[aaron800]

Tony, in the next case (i==2), the PTBD5 is turned back off, so I dont see the minor bug??

As far as debugging, yes I can step through the code in the debugger, I do a run from point at the beginning of the interrupt handler, I can step through and see it execute each time, the middle case (i==1) is never executed.  Yes i is initialized as a global variable before the handler routine. 

Thank you for helping a noobie!  I am more familiar with C development on the Atmel AVR series, but these processors seem more powerful and are considerably cheaper too.

Aaron

[tonyp]

---

aaron800 wrote:
This basically blinks three led's in order based on the i index, and then starts back over at the first led (i=0).

---

Indeed, that's what it should do (if it didn't have a minor bug)!

    else if(i==1)
    {
      PTBD_PTBD4 = 1;
      PTBD_PTBD5 = 0; 
      PTBD_PTBD5 = 1;

should be


    else if(i==1)
    {
      PTBD_PTBD4 = 1;
      PTBD_PTBD5 = 0; 
      PTBD_PTBD6 = 1;

[fabio]

Hi Aaron,

Yes, this is due to compiler optimization and I do not consider it weird.

The compiler is, in fact, reusing common code, instead of two increment instructions, it is using just one. Look at the disassembly listing and you will notice that.

Best regards,

[Ake]

Hi,

Assuming that i = 0 when the program starts.

Thís means that the first test will be true and PTB4 will turn low.

i is incremented.

The next time, i = 1, which means that PTB5 drops low.

i is incremented.

Next time i = 2, so neither of two earlier comparisons are OK, which makes PTB6 go low.

i is set to 0

Is there something hidden that I have not understood?

Regards,

Ake