Re: [Lse-tech] per cpu time statistics

[Erich F. <ef...@es...>]

On Tuesday 17 December 2002 15:44, Ravikiran G Thirumalai wrote:
> On Mon, Dec 16, 2002 at 04:40:43PM +0000, Christoph Hellwig wrote:
> > On Mon, Dec 16, 2002 at 05:30:50PM +0100, Erich Focht wrote:
> > > :-( I find it ugly as a separately configurable thing. Otherwise it=
 is
> > >
> > > just a normal monitoring feature which we had long time before
> > > starting to talk about NUMA. Eliminating this feature saves 16 byte=
s
> > > in the task-struct for 2 CPU boxes, GREAT!!!
> >
> > Redo it using the per-cpu slab allocator?
=2E..
>
> This is all theory (maybe wrong theory) .. but I can make a patch for
> these stats to use kmalloc_percpu if ppl are interested...

It makes sense to have a new patch if there is a realistic chance to
get it accepted. Andrew??? If not, we probably can live with the
additional patch which just reverts the lost functionality (and
without CONFIG_*).

My opinion... Others?

Regards,
Erich