Received-SPF: pass (google.com: domain of linux-kernel+bounces-128809-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: <20240327144537.4165578-1-ryan.roberts@arm.com>
 <20240327144537.4165578-6-ryan.roberts@arm.com> <CAGsJ_4x40DxoukgRuEt3OKP7dESj3w+HXz=dHYR+PH8LjtCnEA@mail.gmail.com>
 <63c9caf4-3af4-4149-b3c2-e677788cb11f@arm.com> <47fa11db-3c26-48a6-bb7e-5ea59739b691@arm.com>
In-Reply-To: <47fa11db-3c26-48a6-bb7e-5ea59739b691@arm.com>
From: Barry Song <21cnbao@gmail.com>
Date: Wed, 3 Apr 2024 11:54:00 +1300
Message-ID: <CAGsJ_4zCTLjJOX3Cs-KDVgqytUykzRJQncHbOEHzyEbyVmt2=g@mail.gmail.com>
Subject: Re: [PATCH v5 5/6] mm: vmscan: Avoid split during shrink_folio_list()
To: Ryan Roberts <ryan.roberts@arm.com>
Cc: Andrew Morton <akpm@linux-foundation.org>, David Hildenbrand <david@redhat.com>, 
	Matthew Wilcox <willy@infradead.org>, Huang Ying <ying.huang@intel.com>, Gao Xiang <xiang@kernel.org>, 
	Yu Zhao <yuzhao@google.com>, Yang Shi <shy828301@gmail.com>, Michal Hocko <mhocko@suse.com>, 
	Kefeng Wang <wangkefeng.wang@huawei.com>, Chris Li <chrisl@kernel.org>, 
	Lance Yang <ioworker0@gmail.com>, linux-mm@kvack.org, linux-kernel@vger.kernel.org, 
	Barry Song <v-songbaohua@oppo.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Wed, Apr 3, 2024 at 2:22=E2=80=AFAM Ryan Roberts <ryan.roberts@arm.com> =
wrote:
>
> On 02/04/2024 14:10, Ryan Roberts wrote:
> > On 28/03/2024 08:18, Barry Song wrote:
> >> On Thu, Mar 28, 2024 at 3:45=E2=80=AFAM Ryan Roberts <ryan.roberts@arm=
com> wrote:
> >>>
> >>> Now that swap supports storing all mTHP sizes, avoid splitting large
> >>> folios before swap-out. This benefits performance of the swap-out pat=
h
> >>> by eliding split_folio_to_list(), which is expensive, and also sets u=
s
> >>> up for swapping in large folios in a future series.
> >>>
> >>> If the folio is partially mapped, we continue to split it since we wa=
nt
> >>> to avoid the extra IO overhead and storage of writing out pages
> >>> uneccessarily.
> >>>
> >>> Reviewed-by: David Hildenbrand <david@redhat.com>
> >>> Reviewed-by: Barry Song <v-songbaohua@oppo.com>
> >>> Signed-off-by: Ryan Roberts <ryan.roberts@arm.com>
> >>> ---
> >>>  mm/vmscan.c | 9 +++++----
> >>>  1 file changed, 5 insertions(+), 4 deletions(-)
> >>>
> >>> diff --git a/mm/vmscan.c b/mm/vmscan.c
> >>> index 00adaf1cb2c3..293120fe54f3 100644
> >>> --- a/mm/vmscan.c
> >>> +++ b/mm/vmscan.c
> >>> @@ -1223,11 +1223,12 @@ static unsigned int shrink_folio_list(struct =
list_head *folio_list,
> >>>                                         if (!can_split_folio(folio, N=
ULL))
> >>>                                                 goto activate_locked;
> >>>                                         /*
> >>> -                                        * Split folios without a PMD=
 map right
> >>> -                                        * away. Chances are some or =
all of the
> >>> -                                        * tail pages can be freed wi=
thout IO.
> >>> +                                        * Split partially mapped fol=
ios right
> >>> +                                        * away. We can free the unma=
pped pages
> >>> +                                        * without IO.
> >>>                                          */
> >>> -                                       if (!folio_entire_mapcount(fo=
lio) &&
> >>> +                                       if (data_race(!list_empty(
> >>> +                                               &folio->_deferred_lis=
t)) &&
> >>>                                             split_folio_to_list(folio=
,
> >>>                                                                 folio=
_list))
> >>>                                                 goto activate_locked;
> >>
> >> Hi Ryan,
> >>
> >> Sorry for bringing up another minor issue at this late stage.
> >
> > No problem - I'd rather take a bit longer and get it right, rather than=
 rush it
> > and get it wrong!
> >
> >>
> >> During the debugging of thp counter patch v2, I noticed the discrepanc=
y between
> >> THP_SWPOUT_FALLBACK and THP_SWPOUT.
> >>
> >> Should we make adjustments to the counter?
> >
> > Yes, agreed; we want to be consistent here with all the other existing =
THP
> > counters; they only refer to PMD-sized THP. I'll make the change for th=
e next
> > version.
> >
> > I guess we will eventually want equivalent counters for per-size mTHP u=
sing the
> > framework you are adding.
> >
> >>
> >> diff --git a/mm/vmscan.c b/mm/vmscan.c
> >> index 293120fe54f3..d7856603f689 100644
> >> --- a/mm/vmscan.c
> >> +++ b/mm/vmscan.c
> >> @@ -1241,8 +1241,10 @@ static unsigned int shrink_folio_list(struct
> >> list_head *folio_list,
> >>                                                                 folio_=
list))
> >>                                                 goto activate_locked;
> >>  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> >> -
> >> count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
> >> -                                       count_vm_event(THP_SWPOUT_FALL=
BACK);
> >> +                                       if (folio_test_pmd_mappable(fo=
lio)) {
>
> This doesn't quite work because we have already split the folio here, so =
this
> will always return false. I've changed it to:
>
> if (nr_pages >=3D HPAGE_PMD_NR) {

make sense to me.

>
>
> >> +
> >> count_memcg_folio_events(folio, THP_SWPOUT_FALLBACK, 1);
> >> +
> >> count_vm_event(THP_SWPOUT_FALLBACK);
> >> +                                       }
> >>  #endif
> >>                                         if (!add_to_swap(folio))
> >>                                                 goto activate_locked_s=
plit;
> >>
> >>
> >> Because THP_SWPOUT is only for pmd:
> >>
> >> static inline void count_swpout_vm_event(struct folio *folio)
> >> {
> >> #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> >>         if (unlikely(folio_test_pmd_mappable(folio))) {
> >>                 count_memcg_folio_events(folio, THP_SWPOUT, 1);
> >>                 count_vm_event(THP_SWPOUT);
> >>         }
> >> #endif
> >>         count_vm_events(PSWPOUT, folio_nr_pages(folio));
> >> }
> >>
> >> I can provide per-order counters for this in my THP counter patch.
> >>
> >>> --
> >>> 2.25.1
> >>>
> >>

Thanks
Barry