Received-SPF: pass (google.com: domain of linux-kernel+bounces-199712-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Precedence: bulk
MIME-Version: 1.0
References: <20240603140745.83880-1-ioworker0@gmail.com> <Zl3Wjh9_aGY8Xxm7@casper.infradead.org>
 <c0309ab6-8bae-42b7-8d27-1df895689fb8@redhat.com> <Zl3axbhcljTg_X0C@casper.infradead.org>
 <db8d4d78-a519-4642-a882-2a84f7a1366b@redhat.com> <CAK1f24=Xa-n2J4_sYL161PMB3hhJ8tcQtJ0qdV7ChPtjMTrPQg@mail.gmail.com>
In-Reply-To: <CAK1f24=Xa-n2J4_sYL161PMB3hhJ8tcQtJ0qdV7ChPtjMTrPQg@mail.gmail.com>
From: Barry Song <21cnbao@gmail.com>
Date: Tue, 4 Jun 2024 09:00:45 +1200
Message-ID: <CAGsJ_4wwz9gBT35PSffcjQUwPCA2bE=_8Z75RuhsjuQHZnRCVA@mail.gmail.com>
Subject: Re: [PATCH v2 1/1] mm/mlock: implement folio_mlock_step() using folio_pte_batch()
To: Lance Yang <ioworker0@gmail.com>
Cc: David Hildenbrand <david@redhat.com>, Matthew Wilcox <willy@infradead.org>, akpm@linux-foundation.org, 
	ryan.roberts@arm.com, baolin.wang@linux.alibaba.com, ziy@nvidia.com, 
	fengwei.yin@intel.com, ying.huang@intel.com, libang.li@antgroup.com, 
	linux-mm@kvack.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Tue, Jun 4, 2024 at 3:46=E2=80=AFAM Lance Yang <ioworker0@gmail.com> wro=
te:
>
> On Mon, Jun 3, 2024 at 11:26=E2=80=AFPM David Hildenbrand <david@redhat.c=
om> wrote:
> >
> > On 03.06.24 17:01, Matthew Wilcox wrote:
> > > On Mon, Jun 03, 2024 at 04:56:05PM +0200, David Hildenbrand wrote:
> > >> On 03.06.24 16:43, Matthew Wilcox wrote:
> > >>> On Mon, Jun 03, 2024 at 10:07:45PM +0800, Lance Yang wrote:
> > >>>> +++ b/mm/mlock.c
> > >>>> @@ -307,26 +307,15 @@ void munlock_folio(struct folio *folio)
> > >>>>    static inline unsigned int folio_mlock_step(struct folio *folio=
,
> > >>>>                    pte_t *pte, unsigned long addr, unsigned long e=
nd)
> > >>>>    {
> > >>>> -  unsigned int count, i, nr =3D folio_nr_pages(folio);
> > >>>> -  unsigned long pfn =3D folio_pfn(folio);
> > >>>> +  const fpb_t fpb_flags =3D FPB_IGNORE_DIRTY | FPB_IGNORE_SOFT_DI=
RTY;
> > >>>> +  unsigned int count =3D (end - addr) >> PAGE_SHIFT;
> > >>>
> > >>> This is a pre-existing bug, but ... what happens if you're on a 64-=
bit
> > >>> system and you mlock() a range that is exactly 2^44 bytes?  Seems t=
o me
> > >>> that count becomes 0.  Why not use an unsigned long here and avoid =
the
> > >>> problem entirely?
> > >>>
> > >>> folio_pte_batch() also needs to take an unsigned long max_nr in tha=
t
> > >>> case, because you aren't restricting it to folio_nr_pages().
> > >>
> > >> Yeah, likely we should also take a look at other folio_pte_batch() u=
sers
> > >> like copy_present_ptes() that pass the count as an int. Nothing shou=
ld
> > >> really be broken, but we might not batch as much as we could, which =
is
> > >> unfortunate.
> > >
> > > You did include:
> > >
> > >          VM_WARN_ON_FOLIO(!folio_test_large(folio) || max_nr < 1, fol=
io);
> > >
> > > so at the least we have a userspace-triggerable warning.
> >
> > Yes, and max_nr =3D=3D 0 would likely not be healthy to the system.
> >
> > But
> >
> > For copy_pte_range(), zap_pte_range() and the madvise users, we should
> > always have:
> >         next =3D pmd_addr_end(addr, end);
> >
> > and use "next" as the actual "end" -- not the VMA end. So "end - addr" =
=3D
> > "next - addr" should never exceed a single PMD size.
> >
> >
> > mlock_pte_range() is also called from walk_page_range(), which uses
> >         next =3D pmd_addr_end(addr, end);
> >
> > So likely exceeding PMD size is not possible here and all is working as
> > expected.
>
> Thanks for clarifying!
>
> I agree that currently all is fine, so perhaps we don't worry about that =
:)

I agree with this point.

These functions are all scanning PTEs under a PMD. Any value exceeding
the PTE entries of one PMD has been a bug of callers but not the callee.

>
> >
> > Will double check later.
>
> I did a double-check and you're correct.
>
> Thanks,
> Lance
>
> >
> > --
> > Cheers,
> >
> > David / dhildenb
> >

Thanks
Barry