Re: VST timer support, comments needed

On Mon, Nov 01, 2004 at 12:00:36PM -0800, Tony Lindgren wrote:
> * Marc Singer <elf <at> buici.com> [041101 11:52]:
> > On Mon, Nov 01, 2004 at 11:27:57AM -0800, Tony Lindgren wrote:
> > > Hi all,
> > > 
> > > We've implemented a VST (Variable Scheduling Timeouts) support on omap, and
> > > I thought this might be of interest to people here too. It should be possible
> > > to implement similar VST support on other machines as well.
> > > 
> > > VST means the system can stay in idle until next timer event happens. This
> > > allows considerable power savings.
> > > 
> > > VST can be easily implemented in the standard 2.6 kernel using the
> > > next_timer_interrupt() function that was introduced about six months ago by
> > > the os390 folks.
> > 
> > Nice.  This is one of those things I've wanted.  Have you checked for
> > skew against the wall-clock?
> 
> It seems to be running at 32KHz timer hardware accuracy :) So the drift is
> something less than a second per day.
> 
> We use two timers; A 32KHz synchronized continuous timer, and a 32KHz 
> programmable timer for the interrupts. The timer interrupt updates time until
> it's in sync with the continuous timer.

Does this mean that the longest delay during idle will be when when
continuous 32KHz would exhaust it's count?  For example, for a 16 bit
timer that would be about two seconds.

-------------------------------------------------------------------
Subscription options: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel
FAQ:       http://www.arm.linux.org.uk/armlinux/mlfaq.php
Etiquette: http://www.arm.linux.org.uk/armlinux/mletiquette.php