Re: page cache: Store only head pages in i_pages

[Matthew Wilcox <willy@infradead.org>]

On Fri, Apr 05, 2019 at 04:37:42PM +0300, Kirill A. Shutemov wrote:
> On Thu, Apr 04, 2019 at 05:28:02PM -0400, Qian Cai wrote:
> > On Thu, 2019-04-04 at 16:45 +0300, Kirill A. Shutemov wrote:
> > > What about this:
> > > 
> > > diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
> > > index f939e004c5d1..2e8438a1216a 100644
> > > --- a/include/linux/pagemap.h
> > > +++ b/include/linux/pagemap.h
> > > @@ -335,12 +335,15 @@ static inline struct page *grab_cache_page_nowait(struct
> > > address_space *mapping,
> > >  
> > >  static inline struct page *find_subpage(struct page *page, pgoff_t offset)
> > >  {
> > > -	unsigned long index = page_index(page);
> > > +	unsigned long mask;
> > > +
> > > +	if (PageHuge(page))
> > > +		return page;
> > >  
> > >  	VM_BUG_ON_PAGE(PageTail(page), page);
> > > -	VM_BUG_ON_PAGE(index > offset, page);
> > > -	VM_BUG_ON_PAGE(index + (1 << compound_order(page)) <= offset, page);
> > > -	return page - index + offset;
> > > +
> > > +	mask = (1UL << compound_order(page)) - 1;
> > > +	return page + (offset & mask);
> > >  }
> > >  
> > >  struct page *find_get_entry(struct address_space *mapping, pgoff_t offset);
> > 
> > It works fine.
> 
> Nice.
> 
> Matthew, does it look fine to you?

Yes, this looks good to me.  I think we'll have opportunities to improve
it later (eg when unifying THP and hugetlbfs uses of the page cache).
But for now, Andrew can you add this version as -fix-fix?