Re: [PATCH 10/19] Calculate the preferred zone for allocation onlyonce

[Mel Gorman]

On Tue, Feb 24, 2009 at 12:31:41PM -0500, Christoph Lameter wrote:
> On Tue, 24 Feb 2009, Mel Gorman wrote:
> 
> > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > index 6f26944..074f9a6 100644
> > --- a/mm/page_alloc.c
> > +++ b/mm/page_alloc.c
> > @@ -1399,24 +1399,19 @@ static void zlc_mark_zone_full(struct zonelist *zonelist, struct zoneref *z)
> >   */
> >  static struct page *
> >  get_page_from_freelist(gfp_t gfp_mask, nodemask_t *nodemask, unsigned int order,
> > -		struct zonelist *zonelist, int high_zoneidx, int alloc_flags)
> > +		struct zonelist *zonelist, int high_zoneidx, int alloc_flags,
> > +		struct zone *preferred_zone)
> >  {
> 
> This gets into a quite a number of parameters now. Pass a structure like in
> vmscan.c?

I considered it, but thought that multiple offsets into structures might
exceed the cost of pushing the parameters onto the stack. I never
actually looked at the generated assembly though to make a proper
assessment.

> Or simplify things to be able to run get_page_from_freelist with
> less parameters? The number of parameters seem to be too high for a
> fastpath function.
> 

Which is why I ended up inlining get_page_from_freelist() in V1. It's a rock
and a hard place basically. Passing parameters is expensive, but calculating
the information multiple times is too.

-- 
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/