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;
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
From:   David Hildenbrand <david@redhat.com>
Mime-Version: 1.0 (1.0)
Subject: Re: [PATCH v1 05/29] virtio-mem: generalize check for added memory
Date:   Sat, 17 Oct 2020 09:39:38 +0200
Message-Id: <2E12AC3C-872A-4B30-8FD7-12420FA1D14E@redhat.com>
References: <20201016223811.GJ44269@L-31X9LVDL-1304.local>
Cc:     David Hildenbrand <david@redhat.com>, linux-kernel@vger.kernel.org,
        linux-mm@kvack.org, virtualization@lists.linux-foundation.org,
        Andrew Morton <akpm@linux-foundation.org>,
        "Michael S . Tsirkin" <mst@redhat.com>,
        Jason Wang <jasowang@redhat.com>,
        Pankaj Gupta <pankaj.gupta.linux@gmail.com>
In-Reply-To: <20201016223811.GJ44269@L-31X9LVDL-1304.local>
To:     Wei Yang <richard.weiyang@linux.alibaba.com>
Precedence: bulk


> Am 17.10.2020 um 00:38 schrieb Wei Yang <richard.weiyang@linux.alibaba.com=
>:
>=20
> =EF=BB=BFOn Fri, Oct 16, 2020 at 12:32:50PM +0200, David Hildenbrand wrote=
:
>>>>> Ok, I seems to understand the logic now.
>>>>>=20
>>>>> But how we prevent ONLINE_PARTIAL memory block get offlined? There are=
 three
>>>>> calls in virtio_mem_set_fake_offline(), while all of them adjust page'=
s flag.
>>>>> How they hold reference to struct page?
>>>>=20
>>>> Sorry, I should have given you the right pointer. (similar to my other
>>>> reply)
>>>>=20
>>>> We hold a reference either via
>>>>=20
>>>> 1. alloc_contig_range()
>>>=20
>>> I am not familiar with this one, need to spend some time to look into.
>>=20
>> Each individual page will have a pagecount of 1.
>>=20
>>>=20
>>>> 2. memmap init code, when not calling generic_online_page().
>>>=20
>>> I may miss some code here. Before online pages, memmaps are allocated in=

>>> section_activate(). They are supposed to be zero-ed. (I don't get the ex=
act
>>> code line.) I am not sure when we grab a refcount here.
>>=20
>> Best to refer to __init_single_page() -> init_page_count().
>>=20
>> Each page that wasn't onlined via generic_online_page() has a refcount
>> of 1 and looks like allocated.
>>=20
>=20
> Thanks, I see the logic.
>=20
>    online_pages()
>        move_pfn_range_to_zone()  --- 1)
>    online_pages_range()      --- 2)
>=20
> At 1), __init_single_page() would set page count to 1. At 2),
> generic_online_page() would clear page count, while the call back would no=
t.
>=20
> Then I am trying to search the place where un-zero page count prevent offl=
ine.
> scan_movable_pages() would fail, since this is a PageOffline() and has 1 p=
age
> count.
>=20
> So the GUARD we prevent offline partial-onlined pages is
>=20
>    (PageOffline && page_count)
>=20
> And your commit aa218795cb5fd583c94f
>=20
> mm: Allow to offline unmovable PageOffline() pages via MEM_GOING_OFFLINE
>=20
> is introduced to handle this case.
>=20
> That's pretty clear now.
>=20

I=E2=80=98m happy to see that I am no longer the only person that understand=
s all this magic :)

Thanks for having a look / reviewing!

>> --=20
>> Thanks,
>>=20
>> David / dhildenb
>=20
> --=20
> Wei Yang
> Help you, Help me
>=20