Re: [RFC PATCH 1/3] mm, numa: rework do_pages_move

From: Michal Hocko
Date: Fri Dec 15 2017 - 04:58:03 EST


On Fri 15-12-17 12:51:25, Kirill A. Shutemov wrote:
> On Fri, Dec 15, 2017 at 10:28:59AM +0100, Michal Hocko wrote:
> > On Thu 14-12-17 18:35:58, Kirill A. Shutemov wrote:
> > > On Wed, Dec 13, 2017 at 03:39:48PM +0100, Michal Hocko wrote:
> > [...]
> > > > + err = 0;
> > > > + if (page_to_nid(page) == node)
> > > > + goto out_putpage;
> > > > +
> > > > + err = -EACCES;
> > > > + if (page_mapcount(page) > 1 &&
> > > > + !migrate_all)
> > >
> > > Non-sensible line break.
> >
> > fixed
> >
> > > > + goto out_putpage;
> > > > +
> > > > + if (PageHuge(page)) {
> > > > + if (PageHead(page)) {
> > > > + isolate_huge_page(page, pagelist);
> > > > + err = 0;
> > > > + }
> > > > + } else {
> > >
> > > Hm. I think if the page is PageTail() we have to split the huge page.
> > > If an user asks to migrate part of THP, we shouldn't migrate the whole page,
> > > otherwise it's not transparent anymore.
> >
> > Well, as I've said in the cover letter. There are more things which are
> > worth considering but I've tried to keep the original semantic so
> > further changes should be done in separete patches. I will work on those
> > but I would prefer this to stay smaller if you do not mind.
>
> Sure.
>
> Fill free to use my ack.

Thanks a lot Kirill! I will wait for some more feedback and then
resubmit later next week.

--
Michal Hocko
SUSE Labs