Re: new IRQ scalability changes in 2.3.48

From: yodaiken@fsmlabs.com
Date: Wed Mar 08 2000 - 22:04:56 EST


On Wed, Mar 08, 2000 at 06:44:51PM -0800, Linus Torvalds wrote:
>
>
> On Wed, 8 Mar 2000 yodaiken@fsmlabs.com wrote:
> > On a UP -- no change except code is more complex
> > On a SMP box performance loss without using affinity.
> > Take two spinlocks instead of one, more cache boucing etc.
>
> You've said that now several times, and you always ignore the answer that
> you are always given: normally you don't need any other spinlock if your
> interrupt controller can do the operations atomically. Which they actually
> usually can, at least the better ones.
>
> Ignoring that answer only makes you look silly. Don't do it.

The way you are getting an atomic operation on the level irqs is by
splitting it into two parts: unless I'm mistaken about it.
So the performance loss comes from the io-apic being frozen
until the interrupt handler returns and does a desc-end
You avoid the second spin lock by freezing the io-apic

Is that false?

-- 
---------------------------------------------------------
Victor Yodaiken 
FSMLabs:  www.fsmlabs.com  www.rtlinux.com
FSMLabs is a servicemark and a service of 
VJY Associates L.L.C, New Mexico.

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Wed Mar 15 2000 - 21:00:15 EST