Subject: Re: halving context switch time
To: Jason R Thorpe <thorpej@wasabisystems.com>
From: Bill Studenmund <wrstuden@netbsd.org>
List: tech-perform
Date: 08/22/2002 10:58:07
On Wed, 21 Aug 2002, Jason R Thorpe wrote:

> On Wed, Aug 21, 2002 at 09:00:04PM -0700, Bill Studenmund wrote:
>
>  > Why not just change cpu_switch to return if it entered or not? That seems
>  > to be the cleanest solution.
>
> There's a slight problem -- in the nathanw_sa branch, cpu_switch() returns
> 1 or 0 depending on whether or not it switched to another process.. this
> is not quite the same, because in nathanw_sa, A -> idle -> A returns "didn't
> switch", while this is precisely what David wants to differentiate.

So then for nathanw_sa, we need to return two flags, one to indicate we
entered the idle loop, and one to indicate if we switched processes. ??

It sounds like this change speeds up context switching a fair bit, so I
think we should see what we can do to do it right. :-)

Take care,

Bill