Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: quoted-printable
From:   Qian Cai <cai@lca.pw>
Mime-Version: 1.0 (1.0)
Subject: Re: [PATCH] mm/page_alloc: Add a reason for reserved pages in has_unmovable_pages()
Date:   Thu, 3 Oct 2019 08:14:40 -0400
Message-Id: <6A58E80B-7A5F-4CAD-ACF1-89BCCBE4D3B1@lca.pw>
References: <49fa7dea-00ac-155f-e7b7-eeca206556b5@arm.com>
Cc:     linux-mm@kvack.org, Andrew Morton <akpm@linux-foundation.org>,
        Michal Hocko <mhocko@suse.com>,
        Vlastimil Babka <vbabka@suse.cz>,
        Oscar Salvador <osalvador@suse.de>,
        Mel Gorman <mgorman@techsingularity.net>,
        Mike Rapoport <rppt@linux.ibm.com>,
        Dan Williams <dan.j.williams@intel.com>,
        Pavel Tatashin <pavel.tatashin@microsoft.com>,
        linux-kernel@vger.kernel.org
In-Reply-To: <49fa7dea-00ac-155f-e7b7-eeca206556b5@arm.com>
To:     Anshuman Khandual <Anshuman.Khandual@arm.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


> On Oct 3, 2019, at 8:01 AM, Anshuman Khandual <Anshuman.Khandual@arm.com> w=
rote:
>=20
> Will something like this be better ?

Not really. dump_page() will dump PageCompound information anyway, so it is t=
rivial to figure out if went in that path.

> hugepage_migration_supported() has got
> uncertainty depending on platform and huge page size.
>=20
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 15c2050c629b..8dbc86696515 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -8175,7 +8175,7 @@ bool has_unmovable_pages(struct zone *zone, struct p=
age *page, int count,
>        unsigned long found;
>        unsigned long iter =3D 0;
>        unsigned long pfn =3D page_to_pfn(page);
> -       const char *reason =3D "unmovable page";
> +       const char *reason;
>=20
>        /*
>         * TODO we could make this much more efficient by not checking ever=
y
> @@ -8194,7 +8194,7 @@ bool has_unmovable_pages(struct zone *zone, struct p=
age *page, int count,
>                if (is_migrate_cma(migratetype))
>                        return false;
>=20
> -               reason =3D "CMA page";
> +               reason =3D "Unmovable CMA page";
>                goto unmovable;
>        }
>=20
> @@ -8206,8 +8206,10 @@ bool has_unmovable_pages(struct zone *zone, struct p=
age *page, int count,
>=20
>                page =3D pfn_to_page(check);
>=20
> -               if (PageReserved(page))
> +               if (PageReserved(page)) {
> +                       reason =3D "Unmovable reserved page";
>                        goto unmovable;
> +               }
>=20
>                /*
>                 * If the zone is movable and we have ruled out all reserve=
d
> @@ -8226,8 +8228,10 @@ bool has_unmovable_pages(struct zone *zone, struct p=
age *page, int count,
>                        struct page *head =3D compound_head(page);
>                        unsigned int skip_pages;
>=20
> -                       if (!hugepage_migration_supported(page_hstate(head=
)))
> +                       if (!hugepage_migration_supported(page_hstate(head=
))) {
> +                               reason =3D "Unmovable HugeTLB page";
>                                goto unmovable;
> +                       }
>=20
>                        skip_pages =3D compound_nr(head) - (page - head);
>                        iter +=3D skip_pages - 1;
> @@ -8271,8 +8275,10 @@ bool has_unmovable_pages(struct zone *zone, struct p=
age *page, int count,
>                 * is set to both of a memory hole page and a _used_ kernel=

>                 * page at boot.
>                 */
> -               if (found > count)
> +               if (found > count) {
> +                       reason =3D "Unmovable non-LRU page";
>                        goto unmovable;
> +               }
>        }
>        return false;
> unmovable: