Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <1598331582-19923-1-git-send-email-iamjoonsoo.kim@lge.com> <20200824221049.edb3c540bbfc859a6806600d@linux-foundation.org>
In-Reply-To: <20200824221049.edb3c540bbfc859a6806600d@linux-foundation.org>
From:   Joonsoo Kim <js1304@gmail.com>
Date:   Tue, 25 Aug 2020 14:34:32 +0900
Message-ID: <CAAmzW4MYQsRoR-q8tUmwPYLvCbiMq_zt19_AgorAQHnbwnu37g@mail.gmail.com>
Subject: Re: [PATCH for v5.9] mm/page_alloc: handle a missing case for
 memalloc_nocma_{save/restore} APIs
To:     Andrew Morton <akpm@linux-foundation.org>
Cc:     Linux Memory Management List <linux-mm@kvack.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Michal Hocko <mhocko@kernel.org>,
        Vlastimil Babka <vbabka@suse.cz>,
        "Aneesh Kumar K . V" <aneesh.kumar@linux.ibm.com>,
        kernel-team@lge.com, Joonsoo Kim <iamjoonsoo.kim@lge.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

2020=EB=85=84 8=EC=9B=94 25=EC=9D=BC (=ED=99=94) =EC=98=A4=ED=9B=84 2:10, A=
ndrew Morton <akpm@linux-foundation.org>=EB=8B=98=EC=9D=B4 =EC=9E=91=EC=84=
=B1:
>
> On Tue, 25 Aug 2020 13:59:42 +0900 js1304@gmail.com wrote:
>
> > From: Joonsoo Kim <iamjoonsoo.kim@lge.com>
> >
> > memalloc_nocma_{save/restore} APIs can be used to skip page allocation
> > on CMA area, but, there is a missing case and the page on CMA area coul=
d
> > be allocated even if APIs are used. This patch handles this case to fix
> > the potential issue.
> >
> > Missing case is an allocation from the pcplist. MIGRATE_MOVABLE pcplist
> > could have the pages on CMA area so we need to skip it if ALLOC_CMA isn=
't
> > specified.
> >
> > This patch implements this behaviour by checking allocated page from
> > the pcplist rather than skipping an allocation from the pcplist entirel=
y.
> > Skipping the pcplist entirely would result in a mismatch between waterm=
ark
> > check and actual page allocation. And, it requires to break current cod=
e
> > layering that order-0 page is always handled by the pcplist. I'd prefer
> > to avoid it so this patch uses different way to skip CMA page allocatio=
n
> > from the pcplist.
> >
> > ...
> >
> > --- a/mm/page_alloc.c
> > +++ b/mm/page_alloc.c
> > @@ -3341,6 +3341,22 @@ static struct page *rmqueue_pcplist(struct zone =
*preferred_zone,
> >       pcp =3D &this_cpu_ptr(zone->pageset)->pcp;
> >       list =3D &pcp->lists[migratetype];
> >       page =3D __rmqueue_pcplist(zone,  migratetype, alloc_flags, pcp, =
list);
> > +#ifdef CONFIG_CMA
> > +     if (page) {
> > +             int mt =3D get_pcppage_migratetype(page);
> > +
> > +             /*
> > +              * pcp could have the pages on CMA area and we need to sk=
ip it
> > +              * when !ALLOC_CMA. Free all pcplist and retry allocation=
.
> > +              */
> > +             if (is_migrate_cma(mt) && !(alloc_flags & ALLOC_CMA)) {
> > +                     list_add(&page->lru, &pcp->lists[migratetype]);
> > +                     pcp->count++;
> > +                     free_pcppages_bulk(zone, pcp->count, pcp);
> > +                     page =3D __rmqueue_pcplist(zone, migratetype, all=
oc_flags, pcp, list);
> > +             }
> > +     }
> > +#endif
> >       if (page) {
> >               __count_zid_vm_events(PGALLOC, page_zonenum(page), 1);
> >               zone_statistics(preferred_zone, zone);
>
> That's a bunch more code on a very hot path to serve an obscure feature
> which has a single obscure callsite.
>
> Can we instead put the burden on that callsite rather than upon
> everyone?  For (dumb) example, teach __gup_longterm_locked() to put the
> page back if it's CMA and go get another one?

Hmm... Unfortunately, it cannot ensure that we eventually get the non-CMA p=
age.
I think that the only way to ensure it is to implement the
functionality here. We can
use 'unlikely' or 'static branch' to reduce the overhead for a really
rare case but
for now I have no idea how to completely remove the overhead.

Thanks.