Re: [PATCH] For preventing kstat overflow

[Kingsley Cheung <kingsley@aurema.com>]

On Tue, Mar 09, 2004 at 04:57:04PM +1100, Kingsley Cheung wrote:
> On Mon, Mar 08, 2004 at 06:53:54PM -0800, Andrew Morton wrote:
> > Kingsley Cheung <kingsley@aurema.com> wrote:
> > >
> > > Hi All,
> > > 
> > > What do people think of a patch to change the fields in cpu_usage_stat
> > > from unsigned ints to unsigned long longs?  And the same change for
> > > nr_switches in the runqueue structure too?
> > 
> > Sounds unavoidable.
> > 
> > > Its actually worse for context
> > > switches on a busy system, for we've been seeing an average of ten
> > > switches a tick for some of the statistics we have.
> > 
> > Sounds broken.  What CPU scheduler are you using?
> 
> Um, what do you mean by broken?
> 
> Well, as for the scheduler, its the Entitlement Based Scheduler, but
> it doesn't look like its got anything to do with the scheduler.  Some
> work loads we have been testing just have processes that come and go
> so frequently that the context switch rate is high.  Even when Ingo
> posted his original O(1) patch (see
> http://marc.theaimsgroup.com/?l=linux-kernel&m=101010394225604&w=2) he
> claimed high context switch rates.

Oh, just in case there's some confusion... even though we've been
seeing it for EBS, the patch is for 2.6.3.

-- 
		Kingsley