Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <4be7946c-54f5-80a9-4139-5ae07a89e781@redhat.com>
Date:   Mon, 20 Jun 2022 10:56:03 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.9.0
Subject: Re: [PATCH v4 2/2] mm: memory_hotplug: make hugetlb_optimize_vmemmap
 compatible with memmap_on_memory
Content-Language: en-US
To:     Oscar Salvador <osalvador@suse.de>,
        Muchun Song <songmuchun@bytedance.com>
Cc:     akpm@linux-foundation.org, corbet@lwn.net, mike.kravetz@oracle.com,
        paulmck@kernel.org, linux-doc@vger.kernel.org,
        linux-kernel@vger.kernel.org, linux-mm@kvack.org,
        duanxiongchun@bytedance.com, smuchun@gmail.com
References: <20220619133851.68184-1-songmuchun@bytedance.com>
 <20220619133851.68184-3-songmuchun@bytedance.com>
 <YrAgUtV6wD6CIrad@FVFYT0MHHV2J.usts.net>
 <226243a9-b4f5-182e-1a5b-7b8d5c28f3b3@redhat.com>
 <YrAv18GnMOcQaAxz@FVFYT0MHHV2J.usts.net>
 <YrAzeHbYt1mAs9ue@localhost.localdomain>
From:   David Hildenbrand <david@redhat.com>
Organization: Red Hat
In-Reply-To: <YrAzeHbYt1mAs9ue@localhost.localdomain>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 20.06.22 10:44, Oscar Salvador wrote:
> On Mon, Jun 20, 2022 at 04:29:11PM +0800, Muchun Song wrote:
>>>> Although it works, I think PageVmemmapSelfHosted() check for the 1st pfn's
>>>> vmemmap page is not always reliable.  Since we reused PG_owner_priv_1
>>>> as PG_vmemmap_self_hosted, the test is noly reliable for vmemmap page's
>>>> vmemmap page.  Other non-vmemmap page can be flagged with PG_owner_priv_1.
>>>> So this check can be false-positive. Maybe the following code snippet is
>>>> the solution.
>>>
>>> How could that happen for pages used for backing a vmemmap?
>>>
>>
>> It cannot happen for memmap_on_memory case. Howwver, it can happen for other
>> cases. E.g. the 1st pfn (of boot memory block) whose vmemmap page may be flagged
>> as PG_owner_priv_1 (if PG_swapcache is set). Then, the check is false-positive.
> 
> If this can really happen, which I am not that sure tbh, maybe a way out would be
> to just define a new page-type as we did in previous versions of memmap_on_memory.
> In that way we would not for flags, but for its type.
> 
> But as I said, I am not entirely sure about the potential fallout of what you mention.

We are talking about the memmap of a page, who's page content is the
memmap of pages actually exposed to other users (file-backed, anonymous,
whatsoever).

In other words, while setting PG_swapcache on the memmap of a page
exposed to the user is possible, it shouldn't be possible for the memmap
of the page "hosting" these memmaps.

I know, it's confusing and I keep confusing myself. I tried creating a
picture and it doesn't really clarify the situation :D

-- 
Thanks,

David / dhildenb