Received-SPF: pass (google.com: domain of linux-kernel+bounces-79461-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Precedence: bulk
MIME-Version: 1.0
References: <20240129054551.57728-1-ioworker0@gmail.com> <CAHbLzkrScCvmDOPF0_+CieOtfcaP4JE=Pn29eYS_ZKiX-o4=-g@mail.gmail.com>
In-Reply-To: <CAHbLzkrScCvmDOPF0_+CieOtfcaP4JE=Pn29eYS_ZKiX-o4=-g@mail.gmail.com>
From: Lance Yang <ioworker0@gmail.com>
Date: Sat, 24 Feb 2024 11:54:20 +0800
Message-ID: <CAK1f24=2z9tk4cCCSobVpExZ+uV6N8gbGx+xiJ7p_UjrZvSv3g@mail.gmail.com>
Subject: Re: [PATCH 1/1] mm/khugepaged: bypassing unnecessary scans with
 MMF_DISABLE_THP check
To: Yang Shi <shy828301@gmail.com>
Cc: akpm@linux-foundation.org, mhocko@suse.com, zokeefe@google.com, 
	david@redhat.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

Hey Yang,

Thanks for taking time to review!

On Sat, Feb 24, 2024 at 9:46=E2=80=AFAM Yang Shi <shy828301@gmail.com> wrot=
e:
>
> On Sun, Jan 28, 2024 at 9:46=E2=80=AFPM Lance Yang <ioworker0@gmail.com> =
wrote:
> >
> > khugepaged scans the entire address space in the
> > background for each given mm, looking for
> > opportunities to merge sequences of basic pages
> > into huge pages. However, when an mm is inserted
> > to the mm_slots list, and the MMF_DISABLE_THP flag
> > is set later, this scanning process becomes
> > unnecessary for that mm and can be skipped to avoid
> > redundant operations, especially in scenarios with
> > a large address space.
> >
> > This commit introduces a check before each scanning
> > process to test the MMF_DISABLE_THP flag for the
> > given mm; if the flag is set, the scanning process
> > is bypassed, thereby improving the efficiency of
> > khugepaged.
> >
> > Signed-off-by: Lance Yang <ioworker0@gmail.com>
> > ---
> >  mm/khugepaged.c | 18 ++++++++++++------
> >  1 file changed, 12 insertions(+), 6 deletions(-)
> >
> > diff --git a/mm/khugepaged.c b/mm/khugepaged.c
> > index 2b219acb528e..d6a700834edc 100644
> > --- a/mm/khugepaged.c
> > +++ b/mm/khugepaged.c
> > @@ -410,6 +410,12 @@ static inline int hpage_collapse_test_exit(struct =
mm_struct *mm)
> >         return atomic_read(&mm->mm_users) =3D=3D 0;
> >  }
> >
> > +static inline int hpage_collapse_test_exit_or_disable(struct mm_struct=
 *mm)
> > +{
> > +       return hpage_collapse_test_exit(mm) ||
> > +              test_bit(MMF_DISABLE_THP, &mm->flags);
> > +}
> > +
> >  void __khugepaged_enter(struct mm_struct *mm)
> >  {
> >         struct khugepaged_mm_slot *mm_slot;
> > @@ -1422,7 +1428,7 @@ static void collect_mm_slot(struct khugepaged_mm_=
slot *mm_slot)
> >
> >         lockdep_assert_held(&khugepaged_mm_lock);
> >
> > -       if (hpage_collapse_test_exit(mm)) {
> > +       if (hpage_collapse_test_exit_or_disable(mm)) {
>
> I'm not quite sure whether we should remove the mm from mm_slot and
> drop mm_count or not since clearing MMF_THP_DISABLE flag doesn't add
> the mm back. Creating new vma may add the mm back, but I don't think
> the behavior should depend on this.

Thanks for bringing this up! I agree with your point.
I'll remove the MMF_THP_DISABLE flag check in
collect_mm_slot().

>
> >                 /* free mm_slot */
> >                 hash_del(&slot->hash);
> >                 list_del(&slot->mm_node);
> > @@ -2360,7 +2366,7 @@ static unsigned int khugepaged_scan_mm_slot(unsig=
ned int pages, int *result,
> >                 goto breakouterloop_mmap_lock;
> >
> >         progress++;
> > -       if (unlikely(hpage_collapse_test_exit(mm)))
> > +       if (unlikely(hpage_collapse_test_exit_or_disable(mm)))
> >                 goto breakouterloop;
> >
> >         vma_iter_init(&vmi, mm, khugepaged_scan.address);
> > @@ -2368,7 +2374,7 @@ static unsigned int khugepaged_scan_mm_slot(unsig=
ned int pages, int *result,
> >                 unsigned long hstart, hend;
> >
> >                 cond_resched();
> > -               if (unlikely(hpage_collapse_test_exit(mm))) {
> > +               if (unlikely(hpage_collapse_test_exit_or_disable(mm))) =
{
> >                         progress++;
> >                         break;
> >                 }
> > @@ -2390,7 +2396,7 @@ static unsigned int khugepaged_scan_mm_slot(unsig=
ned int pages, int *result,
> >                         bool mmap_locked =3D true;
> >
> >                         cond_resched();
> > -                       if (unlikely(hpage_collapse_test_exit(mm)))
> > +                       if (unlikely(hpage_collapse_test_exit_or_disabl=
e(mm)))
> >                                 goto breakouterloop;
> >
> >                         VM_BUG_ON(khugepaged_scan.address < hstart ||
> > @@ -2408,7 +2414,7 @@ static unsigned int khugepaged_scan_mm_slot(unsig=
ned int pages, int *result,
> >                                 fput(file);
> >                                 if (*result =3D=3D SCAN_PTE_MAPPED_HUGE=
PAGE) {
> >                                         mmap_read_lock(mm);
> > -                                       if (hpage_collapse_test_exit(mm=
))
> > +                                       if (hpage_collapse_test_exit_or=
_disable(mm))
> >                                                 goto breakouterloop;
> >                                         *result =3D collapse_pte_mapped=
_thp(mm,
> >                                                 khugepaged_scan.address=
, false);
> > @@ -2450,7 +2456,7 @@ static unsigned int khugepaged_scan_mm_slot(unsig=
ned int pages, int *result,
> >          * Release the current mm_slot if this mm is about to die, or
> >          * if we scanned all vmas of this mm.
> >          */
> > -       if (hpage_collapse_test_exit(mm) || !vma) {
> > +       if (hpage_collapse_test_exit_or_disable(mm) || !vma) {
>
> Same like the above comment.

I=E2=80=98ll also remove the MMF_THP_DISABLE flag check here.

>
> In addition, didn't you need to change hpage_collapse_test_exit() to
> hpage_collapse_test_exit_or_disable() in hugepage_vma_revalidate()?

Thanks for your suggestion! It makes sense to change
hpage_collapse_test_exit() to hpage_collapse_test_exit_or_disable()
in hugepage_vma_revalidate().

Thanks again for your time!
Lance

>
> >                 /*
> >                  * Make sure that if mm_users is reaching zero while
> >                  * khugepaged runs here, khugepaged_exit will find
> > --
> > 2.33.1
> >