Re: PREEMPT_RT vs ADEOS: the numbers, part 1

[Ingo Molnar <mingo@elte.hu>]

* Karim Yaghmour <karim@opersys.com> wrote:

> > -				 0, p->name, p)) {
> > +				 SA_NODELAY, p->name, p)) {
> >  			printk (KERN_WARNING "%s: irq %d in use, "
> >  				"resorting to polled operation\n",
> >  				p->name, p->irq);
> 
> Thanks for the patch. However, we actually wrote our own driver 
> requesting the parport int instead of using the one in Linux. We just 
> wanted to really customize the driver in as much as possible for 
> benchmarking purposes.

in this case you'll still have to use SA_NODELAY - otherwise you'll get 
an interrupt thread allocated, whose priority could, depending on the 
order of IRQ requests, be lower than the priority of some other
interrupt threads. In that case not only do scheduling latencies get
added to your latency value, but also the worst-case latencies of other
IRQ handlers!

	Ingo