Received-SPF: pass (google.com: domain of linux-kernel+bounces-212893-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240610120809.66601-1-ioworker0@gmail.com> <20240613072858.16512-1-ioworker0@gmail.com>
 <d48bcbcc-0935-4d04-b6a8-f1ceec8190bf@linux.alibaba.com>
In-Reply-To: <d48bcbcc-0935-4d04-b6a8-f1ceec8190bf@linux.alibaba.com>
From: Lance Yang <ioworker0@gmail.com>
Date: Thu, 13 Jun 2024 16:56:34 +0800
Message-ID: <CAK1f24nr4khVun9eSanywYhA3-0LGn-ZDV2kLVkToRTekVfG2g@mail.gmail.com>
Subject: Re: [PATCH v7 4/4] mm/vmscan: avoid split lazyfree THP during shrink_folio_list()
To: Baolin Wang <baolin.wang@linux.alibaba.com>
Cc: akpm@linux-foundation.org, 21cnbao@gmail.com, david@redhat.com, 
	fengwei.yin@intel.com, libang.li@antgroup.com, linux-kernel@vger.kernel.org, 
	linux-mm@kvack.org, maskray@google.com, mhocko@suse.com, minchan@kernel.org, 
	peterx@redhat.com, ryan.roberts@arm.com, shy828301@gmail.com, sj@kernel.org, 
	songmuchun@bytedance.com, wangkefeng.wang@huawei.com, willy@infradead.org, 
	xiehuan09@gmail.com, ziy@nvidia.com, zokeefe@google.com
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Thu, Jun 13, 2024 at 4:20=E2=80=AFPM Baolin Wang
<baolin.wang@linux.alibaba.com> wrote:
>
>
>
> On 2024/6/13 15:28, Lance Yang wrote:
> > Hi Andrew,
> >
> > I'd like to fix the bug[1] I mentioned previously. Could you please fol=
d the
> > following changes into this patch?
> >
> > [1] https://lore.kernel.org/linux-mm/20240613065521.15960-1-ioworker0@g=
mail.com/
> >
> > diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> > index f409ea9fcc18..425374ae06ed 100644
> > --- a/mm/huge_memory.c
> > +++ b/mm/huge_memory.c
> > @@ -2707,10 +2707,8 @@ static bool __discard_anon_folio_pmd_locked(stru=
ct vm_area_struct *vma,
> >       if (unlikely(page_folio(page) !=3D folio))
> >               return false;
> >
> > -     if (folio_test_dirty(folio) || pmd_dirty(orig_pmd)) {
> > -             folio_set_swapbacked(folio);
> > +     if (folio_test_dirty(folio) || pmd_dirty(orig_pmd))
> >               return false;
> > -     }
> >
> >       orig_pmd =3D pmdp_huge_clear_flush(vma, addr, pmdp);
> >
> > @@ -2737,10 +2735,8 @@ static bool __discard_anon_folio_pmd_locked(stru=
ct vm_area_struct *vma,
> >        *
> >        * The only folio refs must be one from isolation plus the rmap(s=
).
> >        */
> > -     if (folio_test_dirty(folio) || pmd_dirty(orig_pmd))
> > -             folio_set_swapbacked(folio);
> > -
> > -     if (folio_test_swapbacked(folio) || ref_count !=3D map_count + 1)=
 {
> > +     if (folio_test_dirty(folio) || pmd_dirty(orig_pmd) ||
> > +         ref_count !=3D map_count + 1) {
> >               set_pmd_at(mm, addr, pmdp, orig_pmd);
> >               return false;
> >       }
> > diff --git a/mm/rmap.c b/mm/rmap.c
> > index b9e5943c8349..393e2c11c44c 100644
> > --- a/mm/rmap.c
> > +++ b/mm/rmap.c
> > @@ -1824,12 +1824,7 @@ static bool try_to_unmap_one(struct folio *folio=
, struct vm_area_struct *vma,
> >                        */
> >                       if (unlikely(folio_test_swapbacked(folio) !=3D
> >                                       folio_test_swapcache(folio))) {
> > -                             /*
> > -                              * unmap_huge_pmd_locked() will unmark a
> > -                              * PMD-mapped folio as lazyfree if the fo=
lio or
> > -                              * its PMD was redirtied.
> > -                              */
> > -                             WARN_ON_ONCE(!pmd_mapped);
> > +                             WARN_ON_ONCE(1);
> >                               goto walk_done_err;
> >                       }
> >
>
> You can also drop the unused 'pmd_mapped' variable now.

Good catch!

Will drop it in the next version.

Thanks,
Lance