Re: [PATCH 04/10] hugetlb: hugepage migration core

[Naoya Horiguchi]

On Fri, Sep 03, 2010 at 09:10:03AM -0500, Christoph Lameter wrote:
> On Fri, 3 Sep 2010, Naoya Horiguchi wrote:
> 
> > diff --git v2.6.36-rc2/mm/vmscan.c v2.6.36-rc2/mm/vmscan.c
> > index c391c32..69ce2a3 100644
> > --- v2.6.36-rc2/mm/vmscan.c
> > +++ v2.6.36-rc2/mm/vmscan.c
> > @@ -40,6 +40,7 @@
> >  #include <linux/memcontrol.h>
> >  #include <linux/delayacct.h>
> >  #include <linux/sysctl.h>
> > +#include <linux/hugetlb.h>
> >
> >  #include <asm/tlbflush.h>
> >  #include <asm/div64.h>
> > @@ -508,6 +509,10 @@ void putback_lru_page(struct page *page)
> >
> >  	VM_BUG_ON(PageLRU(page));
> >
> > +	if (PageHuge(page)) {
> > +		put_page(page);
> > +		return;
> > +	}
> >  redo:
> >  	ClearPageUnevictable(page);
> >
> > @@ -1112,7 +1117,9 @@ int isolate_lru_page(struct page *page)
> >  {
> >  	int ret = -EBUSY;
> >
> > -	if (PageLRU(page)) {
> > +	if (PageHuge(page) && get_page_unless_zero(compound_head(page)))
> > +		ret = 0;
> > +	else if (PageLRU(page)) {
> >  		struct zone *zone = page_zone(page);
> >
> >  		spin_lock_irq(&zone->lru_lock);
> >
> 
> Huge pages are not on the LRU right?

Yes.

> So why use the lru functions for
> them and then not use the lru? Its a bit stranger. The caller must aware
> of the different handling of huge pages since there is no toying around
> with the lru. So just have the caller do a put or get page instead.

So, you mean we should copy migrate_huge_pages() from migrate_pages() and
make two changes on new function
  1. unmap_and_move() -> unmap_and_move_huge_page()
  2. putback_lru_pages() -> put_page()
, right?

Thanks,
Naoya Horiguchi
--
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/