Received-SPF: pass (google.com: domain of linux-kernel+bounces-150065-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240418105750.98866-1-ioworker0@gmail.com> <20240418105750.98866-5-ioworker0@gmail.com>
 <2fdcee93-b8ad-4374-a8ab-7c7bed463813@redhat.com>
In-Reply-To: <2fdcee93-b8ad-4374-a8ab-7c7bed463813@redhat.com>
From: Lance Yang <ioworker0@gmail.com>
Date: Thu, 18 Apr 2024 20:33:12 +0800
Message-ID: <CAK1f24m4y0kvmdFtHoJoPZeF9aeRpw4nnr1W5BMRz_OH49dHvg@mail.gmail.com>
Subject: Re: [PATCH v9 4/4] mm/madvise: optimize lazyfreeing with mTHP in madvise_free
To: David Hildenbrand <david@redhat.com>
Cc: akpm@linux-foundation.org, ryan.roberts@arm.com, 21cnbao@gmail.com, 
	mhocko@suse.com, fengwei.yin@intel.com, zokeefe@google.com, 
	shy828301@gmail.com, xiehuan09@gmail.com, wangkefeng.wang@huawei.com, 
	songmuchun@bytedance.com, peterx@redhat.com, minchan@kernel.org, 
	linux-mm@kvack.org, linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Thu, Apr 18, 2024 at 8:03=E2=80=AFPM David Hildenbrand <david@redhat.com=
> wrote:
>
> On 18.04.24 12:57, Lance Yang wrote:
> > This patch optimizes lazyfreeing with PTE-mapped mTHP[1]
> > (Inspired by David Hildenbrand[2]). We aim to avoid unnecessary folio
> > splitting if the large folio is fully mapped within the target range.
> >
> > If a large folio is locked or shared, or if we fail to split it, we jus=
t
> > leave it in place and advance to the next PTE in the range. But note th=
at
> > the behavior is changed; previously, any failure of this sort would cau=
se
> > the entire operation to give up. As large folios become more common,
> > sticking to the old way could result in wasted opportunities.
> >
> > On an Intel I5 CPU, lazyfreeing a 1GiB VMA backed by PTE-mapped folios =
of
> > the same size results in the following runtimes for madvise(MADV_FREE) =
in
> > seconds (shorter is better):
> >
> > Folio Size |   Old    |   New    | Change
> > ------------------------------------------
> >        4KiB | 0.590251 | 0.590259 |    0%
> >       16KiB | 2.990447 | 0.185655 |  -94%
> >       32KiB | 2.547831 | 0.104870 |  -95%
> >       64KiB | 2.457796 | 0.052812 |  -97%
> >      128KiB | 2.281034 | 0.032777 |  -99%
> >      256KiB | 2.230387 | 0.017496 |  -99%
> >      512KiB | 2.189106 | 0.010781 |  -99%
> >     1024KiB | 2.183949 | 0.007753 |  -99%
> >     2048KiB | 0.002799 | 0.002804 |    0%
> >
> > [1] https://lkml.kernel.org/r/20231207161211.2374093-5-ryan.roberts@arm=
com
> > [2] https://lore.kernel.org/linux-mm/20240214204435.167852-1-david@redh=
at.com
> >
> > Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
> > Signed-off-by: Lance Yang <ioworker0@gmail.com>
> > ---
> >   mm/madvise.c | 85 +++++++++++++++++++++++++++------------------------=
-
> >   1 file changed, 44 insertions(+), 41 deletions(-)
> >
> > diff --git a/mm/madvise.c b/mm/madvise.c
> > index 4597a3568e7e..375ab3234603 100644
> > --- a/mm/madvise.c
> > +++ b/mm/madvise.c
> > @@ -643,6 +643,7 @@ static int madvise_free_pte_range(pmd_t *pmd, unsig=
ned long addr,
> >                               unsigned long end, struct mm_walk *walk)
> >
> >   {
> > +     const cydp_t cydp_flags =3D CYDP_CLEAR_YOUNG | CYDP_CLEAR_DIRTY;
> >       struct mmu_gather *tlb =3D walk->private;
> >       struct mm_struct *mm =3D tlb->mm;
> >       struct vm_area_struct *vma =3D walk->vma;
> > @@ -697,44 +698,57 @@ static int madvise_free_pte_range(pmd_t *pmd, uns=
igned long addr,
> >                       continue;
> >
> >               /*
> > -              * If pmd isn't transhuge but the folio is large and
> > -              * is owned by only this process, split it and
> > -              * deactivate all pages.
> > +              * If we encounter a large folio, only split it if it is =
not
> > +              * fully mapped within the range we are operating on. Oth=
erwise
> > +              * leave it as is so that it can be marked as lazyfree. I=
f we
> > +              * fail to split a folio, leave it in place and advance t=
o the
> > +              * next pte in the range.
> >                */
> >               if (folio_test_large(folio)) {
> > -                     int err;
> > +                     bool any_young, any_dirty;
> >
> > -                     if (folio_likely_mapped_shared(folio))
> > -                             break;
> > -                     if (!folio_trylock(folio))
> > -                             break;
> > -                     folio_get(folio);
> > -                     arch_leave_lazy_mmu_mode();
> > -                     pte_unmap_unlock(start_pte, ptl);
> > -                     start_pte =3D NULL;
> > -                     err =3D split_folio(folio);
> > -                     folio_unlock(folio);
> > -                     folio_put(folio);
> > -                     if (err)
> > -                             break;
> > -                     start_pte =3D pte =3D
> > -                             pte_offset_map_lock(mm, pmd, addr, &ptl);
> > -                     if (!start_pte)
> > -                             break;
> > -                     arch_enter_lazy_mmu_mode();
> > -                     pte--;
> > -                     addr -=3D PAGE_SIZE;
> > -                     continue;
> > +                     nr =3D madvise_folio_pte_batch(addr, end, folio, =
pte,
> > +                                                  ptent, &any_young, N=
ULL);
> > +
> > +                     if (nr < folio_nr_pages(folio)) {
> > +                             int err;
> > +
> > +                             if (folio_likely_mapped_shared(folio))
> > +                                     continue;
> > +                             if (!folio_trylock(folio))
> > +                                     continue;
> > +                             folio_get(folio);
> > +                             arch_leave_lazy_mmu_mode();
> > +                             pte_unmap_unlock(start_pte, ptl);
> > +                             start_pte =3D NULL;
> > +                             err =3D split_folio(folio);
> > +                             folio_unlock(folio);
> > +                             folio_put(folio);
> > +                             start_pte =3D pte =3D
> > +                                     pte_offset_map_lock(mm, pmd, addr=
, &ptl);
>
> I'd just put it on a single line.

start_pte =3D pte =3D pte_offset_map_lock(mm, pmd, addr, &ptl);

I suddenly realized that putting it on a single line would exceed the
80-char limit.

Or:

start_pte =3D pte =3D pte_offset_map_lock(
        mm, pmd, addr, &ptl);

Thanks,
Lance
>
> > +                             if (!start_pte)
> > +                                     break;
> > +                             arch_enter_lazy_mmu_mode();
> > +                             if (!err)
> > +                                     nr =3D 0;
> > +                             continue;
> > +                     }
> > +
> > +                     if (any_young)
> > +                             ptent =3D pte_mkyoung(ptent);
> > +                     if (any_dirty)
>
> any_dirty is never set, likely missed to pass it to
> madvise_folio_pte_batch().
>
> Apart from that LGTM and this patch is much easier to review now!
>
>
> With above:
>
> Acked-by: David Hildenbrand <david@redhat.com>
>
> --
> Cheers,
>
> David / dhildenb
>