Re: [PATCH 31/35] Optimistically check the first page on the PCP free list is suitable

[Mel Gorman <mel@csn.ul.ie>]

On Mon, Mar 16, 2009 at 12:33:44PM -0400, Christoph Lameter wrote:
> On Mon, 16 Mar 2009, Mel Gorman wrote:
> 
> > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > index bb5bd5e..8568284 100644
> > --- a/mm/page_alloc.c
> > +++ b/mm/page_alloc.c
> > @@ -1212,6 +1212,12 @@ again:
> >  				if (pcp_page_suit(page, order))
> >  					break;
> >  		} else {
> > +			/* Optimistic before we start a list search */
> > +			page = list_entry(list->next, struct page, lru);
> > +			if (pcp_page_suit(page, order))
> > +				goto found;
> > +
> > +			/* Do the search */
> >  			list_for_each_entry(page, list, lru)
> >  				if (pcp_page_suit(page, order))
> >  					break;
> 
> I am not convinced that this is beneficial. If it would then the compiler
> would unroll the loop.
> 

It hit a large number of times when I checked (although I don't have the
figures any more) and the list_for_each_entry() does a prefetch
possibly fetching in a line we don't need.

-- 
Mel Gorman
Part-time Phd Student                          Linux Technology Center
University of Limerick                         IBM Dublin Software Lab

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>