Re: [avr-chat] Interrupts from assync timer in sleepmode

[Bjarne Laursen]

Erik Christiansen wrote:

> On Thu, Sep 22, 2005 at 08:15:46AM +0200, Bjarne Laursen wrote:
>  
>
> > What happens is that sometimes the output-compare-match happens just 
> > before the sleep while the interrupt is disabled. Then the processor 
> > will go to sleep with the interrupt flag set, and not wake up. (Can 
> > anyone confirm that this will happen ?).
> >    
>
> Bjarne,
>
>   Page 13 of the ATmega32 datasheet shows example code for exactly this
> case. The pending interrupt will then not be able to occur before the
> cpu is in sleep mode, so your problem cases would seem to be excluded,
> IIUC. (i.e. The interrupt is processed from sleep mode, after check
> completion.)
>
>   Surely, the cpu will "not wake up" only if the sei is not executed,
> or the interrupt is individually masked?
>
>  
I think this is the examble:
_SEI(); /* set global interrupt enable */
_SLEEP(); /* enter sleep, waiting for interrupt */
/* note: will enter sleep before any pending interrupt(s) */

Yes, that is how I did it first. When the sleepmode is idle, it works fine.
But when sleepmode is power-save, and the output-compare-match happens 
in the interrupt-disabled part just before sleep. The mega128 will go 
into sleep, with interrupt enabled, but not wake until the next 
output-compare-match.
It appears that the match must happen after the sleepmode is entered to 
wake the mcu from power-save.

-Bjarne Laursen, RoseTechnology A/S