On Thu, 10 Sep 2009, Minchan Kim wrote:
> >        /*
> >         * When core dumping an enormous anonymous area that nobody
> > -        * has touched so far, we don't want to allocate page tables.
> > +        * has touched so far, we don't want to allocate unnecessary pages or
> > +        * page tables.  Return error instead of NULL to skip handle_mm_fault,
> > +        * then get_dump_page() will return NULL to leave a hole in the dump.
> > +        * But we can only make this optimization where a hole would surely
> > +        * be zero-filled if handle_mm_fault() actually did handle it.
> >         */
> > -       if (flags & FOLL_ANON) {
> > -               page = ZERO_PAGE(0);
> > -               if (flags & FOLL_GET)
> > -                       get_page(page);
> > -               BUG_ON(flags & FOLL_WRITE);
> > -       }
> > +       if ((flags & FOLL_DUMP) &&
> > +           (!vma->vm_ops || !vma->vm_ops->fault))
> 
> How about adding comment about zero page use?

What kind of comment did you have in mind?
We used to use ZERO_PAGE there, but with this patch we're not using it.
I thought the comment above describes what we're doing well enough.

I may have kept too quiet about ZERO_PAGEs, knowing that a later patch
was going to change the story; but I don't see what needs saying here.

Hugh