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;
MIME-Version: 1.0
References: <CAPcyv4hh5PmF8qU+p7Q903PhX+ho9yHMzLFncmh6psW5YOLU_w@mail.gmail.com>
 <B23A05D9-40E2-4862-979D-C6DA69DDDC80@redhat.com>
In-Reply-To: <B23A05D9-40E2-4862-979D-C6DA69DDDC80@redhat.com>
From:   Dan Williams <dan.j.williams@intel.com>
Date:   Wed, 12 Feb 2020 07:48:15 -0800
Message-ID: <CAPcyv4gLL654xtuiLJUwTGAwzZkn1CVT3i_zCjK9N2X=4nsdpw@mail.gmail.com>
Subject: Re: [PATCH 3/7] mm/sparse.c: only use subsection map in VMEMMAP case
To:     David Hildenbrand <david@redhat.com>
Cc:     Baoquan He <bhe@redhat.com>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Linux MM <linux-mm@kvack.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        Wei Yang <richardw.yang@linux.intel.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Feb 12, 2020 at 1:40 AM David Hildenbrand <david@redhat.com> wrote:
>
>
>
> > Am 11.02.2020 um 21:15 schrieb Dan Williams <dan.j.williams@intel.com>:
> >
> > =EF=BB=BFOn Sun, Feb 9, 2020 at 2:48 AM Baoquan He <bhe@redhat.com> wro=
te:
> >>
> >> Currently, subsection map is used when SPARSEMEM is enabled, including
> >> VMEMMAP case and !VMEMMAP case. However, subsection hotplug is not
> >> supported at all in SPARSEMEM|!VMEMMAP case, subsection map is unneces=
sary
> >> and misleading. Let's adjust code to only allow subsection map being
> >> used in SPARSEMEM|VMEMMAP case.
> >>
> >> Signed-off-by: Baoquan He <bhe@redhat.com>
> >> ---
> >> include/linux/mmzone.h |   2 +
> >> mm/sparse.c            | 231 ++++++++++++++++++++++-------------------
> >> 2 files changed, 124 insertions(+), 109 deletions(-)
> >>
> >> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> >> index 462f6873905a..fc0de3a9a51e 100644
> >> --- a/include/linux/mmzone.h
> >> +++ b/include/linux/mmzone.h
> >> @@ -1185,7 +1185,9 @@ static inline unsigned long section_nr_to_pfn(un=
signed long sec)
> >> #define SUBSECTION_ALIGN_DOWN(pfn) ((pfn) & PAGE_SUBSECTION_MASK)
> >>
> >> struct mem_section_usage {
> >> +#ifdef CONFIG_SPARSEMEM_VMEMMAP
> >>        DECLARE_BITMAP(subsection_map, SUBSECTIONS_PER_SECTION);
> >> +#endif
> >
> > This was done deliberately so that the SPARSEMEM_VMEMMAP=3Dn case ran a=
s
> > a subset of the SPARSEMEM_VMEMMAP=3Dy case.
> >
> > The diffstat does not seem to agree that this is any clearer:
> >
> >    124 insertions(+), 109 deletions(-)
> >
>
> I don=E2=80=98t see a reason to work with subsections (+store them) if su=
bsections are not supported.
>
> I do welcome this cleanup. Diffstats don=E2=80=98t tell the whole story.

I'll take a look at the final result and see if my opinion changes,
but I just wanted to clarify upfront that making sparsemem run some of
the subsection logic was deliberate.