Re: [avr-gcc-list] avr-libc-20020612

[Christoph Plattner]

Hello !

Is this a good strategy to set all not-defined interrupt vectors to
a reset instead of handling this with a common dummy routine ?

Even in safety-critical applications a correct handled "wrong" interrupt
should not lead to a reset, because the availibility is reduces, and it
is not an impact to the safety.

The quite best soultion is to introduce a "dummy" routine which calls
a waek symbol, so that a developer can set an default handler for
non-defined interrupts to do 
- reset the machine (if the design should do so ...)
- do nothing
- do an error report, log, whatever...
- hold the machine (fail safe/fail silent).

Christoph



Michael Tsao wrote:
> 
> >I've just fixed the interrupt vector bug (affecting all <=8K devices).
> >
> >   http://www.amelek.gda.pl/avr/test/avr-libc-20020612.tar.gz
> 
> I just built an object file using this new code. I believe the vector table
> is correct: all unimplemented interrupts forward-jump to an instruction at
> 0x002D that then jumps back to address zero (reset), and implemented
> interrupts appear to jump to the interrupt code.
> 
> I'll test on a real device tonight but so far the .hex file looks good.
> 
> Thank you for such a fast turnaround, Marek!
> 
> _________________________________________________________________
> Send and receive Hotmail on your mobile device: http://mobile.msn.com
> 
> avr-gcc-list at http://avr1.org

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

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