RE: [PATCH v1] mm: migration: fix the FOLL_GET failure on following huge page

From: Wang, Haiyue
Date: Sun Aug 14 2022 - 02:49:34 EST


> -----Original Message-----
> From: Wang, Haiyue
> Sent: Sunday, August 14, 2022 14:20
> To: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> Cc: linux-mm@xxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; david@xxxxxxxxxx; linmiaohe@xxxxxxxxxx; Huang,
> Ying <ying.huang@xxxxxxxxx>; songmuchun@xxxxxxxxxxxxx; naoya.horiguchi@xxxxxxxxx
> Subject: RE: [PATCH v1] mm: migration: fix the FOLL_GET failure on following huge page
>
> > -----Original Message-----
> > From: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> > Sent: Sunday, August 14, 2022 07:29
> > To: Wang, Haiyue <haiyue.wang@xxxxxxxxx>
> > Cc: linux-mm@xxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; david@xxxxxxxxxx; linmiaohe@xxxxxxxxxx; Huang,
> > Ying <ying.huang@xxxxxxxxx>; songmuchun@xxxxxxxxxxxxx; naoya.horiguchi@xxxxxxxxx
> > Subject: Re: [PATCH v1] mm: migration: fix the FOLL_GET failure on following huge page
> >
> > On Fri, 12 Aug 2022 16:49:21 +0800 Haiyue Wang <haiyue.wang@xxxxxxxxx> wrote:
> >
> > > Not all huge page APIs support FOLL_GET option, so the __NR_move_pages
> > > will fail to get the page node information for huge page.
> >


> > Is a -stable backport warranted?
>
> Yes.
>
> Since the mainline has introduced the new patch:
>
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=3218f8712d6bb
>

Looks like 'is_zone_device_page' will cause 'FOLL_GET' page reference count issue, which should
call 'put_page' before call branch jump.

try_grab_page -->
if (flags & FOLL_GET)
folio_ref_inc(folio);

Or I misunderstood the 'is_zone_device_page' ? ;-)

> The backported needs to rebase like for 5.19:
>
> - if (page && !is_zone_device_page(page)) {
> + if (page) {
>
> >
> > >