Received-SPF: pass (google.com: domain of linux-kernel+bounces-77419-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Precedence: bulk
MIME-Version: 1.0
References: <20240222080815.46291-1-zhengqi.arch@bytedance.com> <ad82668d-10c6-4670-97f3-77c60a391c5f@redhat.com>
In-Reply-To: <ad82668d-10c6-4670-97f3-77c60a391c5f@redhat.com>
From: Suren Baghdasaryan <surenb@google.com>
Date: Thu, 22 Feb 2024 13:41:22 -0800
Message-ID: <CAJuCfpGznL_4PJXq6JfYTR6VZ7H=xbdQYNN_DzS1QR3OtRHGEQ@mail.gmail.com>
Subject: Re: [PATCH] mm: userfaultfd: fix unexpected change to src_folio when
 UFFDIO_MOVE fails
To: David Hildenbrand <david@redhat.com>
Cc: Qi Zheng <zhengqi.arch@bytedance.com>, akpm@linux-foundation.org, 
	aarcange@redhat.com, linux-mm@kvack.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Thu, Feb 22, 2024 at 12:43=E2=80=AFAM David Hildenbrand <david@redhat.co=
m> wrote:
>
> On 22.02.24 09:08, Qi Zheng wrote:
> > After ptep_clear_flush(), if we find that src_folio is pinned we will f=
ail
> > UFFDIO_MOVE and put src_folio back to src_pte entry, but the change to
> > src_folio->{mapping,index} is not restored in this process. This is not
> > what we expected, so fix it.
> >
> > Fixes: adef440691ba ("userfaultfd: UFFDIO_MOVE uABI")
> > Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
> > ---
> >   mm/userfaultfd.c | 6 +++---
> >   1 file changed, 3 insertions(+), 3 deletions(-)
> >
> > diff --git a/mm/userfaultfd.c b/mm/userfaultfd.c
> > index 4744d6a96f96..503ea77c81aa 100644
> > --- a/mm/userfaultfd.c
> > +++ b/mm/userfaultfd.c
> > @@ -1008,9 +1008,6 @@ static int move_present_pte(struct mm_struct *mm,
> >               goto out;
> >       }
> >
> > -     folio_move_anon_rmap(src_folio, dst_vma);
> > -     WRITE_ONCE(src_folio->index, linear_page_index(dst_vma, dst_addr)=
);
> > -
> >       orig_src_pte =3D ptep_clear_flush(src_vma, src_addr, src_pte);
> >       /* Folio got pinned from under us. Put it back and fail the move.=
 */
> >       if (folio_maybe_dma_pinned(src_folio)) {
> > @@ -1019,6 +1016,9 @@ static int move_present_pte(struct mm_struct *mm,
> >               goto out;
> >       }
> >
> > +     folio_move_anon_rmap(src_folio, dst_vma);
> > +     WRITE_ONCE(src_folio->index, linear_page_index(dst_vma, dst_addr)=
);
> > +
> >       orig_dst_pte =3D mk_pte(&src_folio->page, dst_vma->vm_page_prot);
> >       /* Follow mremap() behavior and treat the entry dirty after the m=
ove */
> >       orig_dst_pte =3D pte_mkwrite(pte_mkdirty(orig_dst_pte), dst_vma);
>
> Indeed, LGTM.
>
> Reviewed-by: David Hildenbrand <david@redhat.com>

Thanks for catching this! Makes total sense to check before modification.

Reviewed-by: Suren Baghdasaryan <surenb@google.com>

>
> --
> Cheers,
>
> David / dhildenb
>