Re: [PATCH] hugetlb: Do not account hugetlb pages as NR_FILE_PAGES

[Michal Hocko]

On Tue 02-06-15 11:38:05, Michal Hocko wrote:
> On Tue 02-06-15 11:33:05, Vlastimil Babka wrote:
> > On 06/02/2015 11:25 AM, Michal Hocko wrote:
[...]
> > >diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
> > >index 91b7f9b2b774..bb8a70e8fc77 100644
> > >--- a/include/linux/page-flags.h
> > >+++ b/include/linux/page-flags.h
> > >@@ -547,7 +547,13 @@ static inline void ClearPageCompound(struct page *page)
> > >  #endif /* !PAGEFLAGS_EXTENDED */
> > >
> > >  #ifdef CONFIG_HUGETLB_PAGE
> > >-int PageHuge(struct page *page);
> > >+int __PageHuge(struct page *page);
> > >+static inline int PageHuge(struct page *page)
> > >+{
> > >+	if (!PageCompound(page))
> > 
> > Perhaps the above as likely()?
> 
> I have added it already when writing the changelog.
> 
> > [...]
> > 
> > >-EXPORT_SYMBOL_GPL(PageHuge);
> > >+EXPORT_SYMBOL_GPL(__PageHuge);
> > >
> > >  /*
> > >   * PageHeadHuge() only returns true for hugetlbfs head page, but not for
> > >
> > 
> > Do the same thing here by inlining the PageHead() test?
> > I guess the page_to_pgoff and __compound_tail_refcounted callers are rather
> > hot?
> 
> Yes, that sounds like a good idea.

So the overal codesize (with defconfig) has still grown with the patch:
   text    data     bss     dec     hex filename
 443075   59217   25604  527896   80e18 mm/built-in.o.before
 443477   59217   25604  528298   80faa mm/built-in.o.PageHuge
 443653   59217   25604  528474   8105a mm/built-in.o.both

It is still not ~1K with the full inline but quite large on its own.
So I am not sure it makes sense to fiddle with this without actually
seeing some penalty in profiles.

Here is what I have if somebody wants to play with it:
---