Re: [avr-gcc-list] Bug? with switch statement - other case !!

[Christoph Plattner]

Hello to all !

I had the same problem about one-to-two year ago !!
In my switch()-case statement, only the first entry was taken.
This was no coding problem !!!
I analysed the assembler code, and it definitly clobbered all other
branches.

So there may be a compiler branch, which have a buggy switch()-case
implementation !!!

I rewrote the code using `if ... else if .... else if .... else' to work
around the bug.

(I cannot remeber the comiler version, it was one cvs checkout in the
beginnig of AVR C being in mainstream).

Bye
Christoph Plattner


Hendrix, Jeff wrote:
> I installed gcc 20021209 and now the code works perfectly (I didn't make any
> changes to the source, only the makefile).
> I'll try changing my variable to unsigned and see if that works (I'll have
> to wait until tonight, I uninstalled the old gcc from my machine here),
> although I don't think it should matter since it's only dealing with this
> data in 8 bits.
>
> To answer Eric's question (or reply to his comment), I have the high nibble
> set as inputs and writing a 0xf0 to portc will turn on the pull up
> resistors. The keypad will then pull the pin to ground when a button is
> pressed and I only want to respond when 1 button is pressed (so I only have
> 4 valid cases).
>
> -jeff
>
> -----Original Message-----
> From: Hendrix, Jeff [mailto:address@hidden
> Sent: Friday, December 13, 2002 12:07 PM
> To: address@hidden
> Subject: [avr-gcc-list] Bug? with switch statement
>
>
> I'm compiling the code below with 2002-06-25 release from avrfreaks.
> I was originally testing this code on a mega128, the I tried in in avr
> studio with the same results. I also tried compiling it to an 8515 target
> and still get the same results.
> What's happening is the switch statement will only ever take the last case.
> To simulate this, you need to turn on the high nibble on port C pins (to
> simulate the pullups), and then turn off one of the bits in the high nibble.
> The only time it takes one of the case paths is when bit 8 is off (0x70),
> all the other times it just ignores the other cases.
> Anyone have any ideas?
>
> thanks
> -jeff
>
> #include <io.h>
>
> char ReadKeyPad(void)
> {
>    char readPort = 0;
>    char curKey = 0;
>
>    // check ROW1
>         PORTC = 0xfe;  // turn on pull ups and ground row 1
>         readPort = PINC&0xf0;   // check just the high nibble
>         switch (readPort) {
>         case 0xe0:     // key 1 held down
>                 curKey = 1;
>                 break;
>         case 0xd0:     // key 2 held down
>                 curKey = 2;
>                 break;
>         case 0xb0:     // key 3 held down
>                 curKey = 3;
>                 break;
>         case 0x70:     // key 4 held down
>                 curKey = 4;
>                 break;
>         }
>
>         PORTC = 0;
>    return curKey;
> }
>
>
> int main(void)
> {
>         char kp;
>         DDRC = 0x0f;
>         PORTC = 0x00;
>
>    while (1) {
>                 kp = ReadKeyPad();
>         }
>         return 0;
> }
> avr-gcc-list at http://avr1.org
> avr-gcc-list at http://avr1.org



--
-------------------------------------------------------
private:        address@hidden
company:        address@hidden

avr-gcc-list at http://avr1.org