Re: [avr-gcc-list] rcall should be call?

[Tyler Hall]

Without simply emailing my entire project over, I just started another 
project with three functions laid out like this:

void beefedUpFunc(void);
void targetFunc(void);

void main(void) {
    targetFunc();
}
void beefedUpFunc(void) {
    ...lots of bogus code here...
}
void targetFunc(void) {
    int i;
    for (i=0; i<10; i++);
}

And the goal was to quickly beef up beefedUpFunc() with dummy code. I 
figured the quickest way (that could also be easily followed in a 
dissassemble) was to actually use 32-bit instructions in an asm() 
directive, so I used   asm("call 0"::);   over and over.  And behold, 
the make failed saying that was invalid with cpu type avr2 (gcc actually 
made the complaint)! Sure enough, my mistake was that I forgot to 
specify -mmcu=avr5 in my CFLAGs, even though I had remembered to put 
-mavr5 in LDFLAGS.

Thanks,
Tyler


Marek Michalkiewicz wrote:

> On Sun, Apr 06, 2003 at 04:04:04PM -0500, Tyler Hall wrote:
>  
>
> > Hi, I working on a growing project written in C, targetting the mega323. 
> > I've hit a strange problem and narrowed it down to this: making a 
> > function call from main() to one of my other functions (which gets 
> > compiled/linked way past the 2k distance) the thing dies.  Looking at 
> > the objdump I see that gcc felt it should attempt an rcall to the 
> > function even though the opcode operand could never reach the address. 
> >    
>
> Please send me a simple test case, complete with avr-gcc options used.
>
> Marek
>
>
> _______________________________________________
> avr-gcc-list mailing list
> address@hidden
> http://www.avr1.org/mailman/listinfo/avr-gcc-list
>
>