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: <525298ad-5e6a-2f8d-366d-4dcb7eebd093@redhat.com>
Date:   Fri, 29 Apr 2022 12:07:17 +0200
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:91.0) Gecko/20100101
 Thunderbird/91.6.2
Content-Language: en-US
To:     Miaohe Lin <linmiaohe@huawei.com>, akpm@linux-foundation.org,
        mike.kravetz@oracle.com, naoya.horiguchi@nec.com
Cc:     ying.huang@intel.com, hch@lst.de, dhowells@redhat.com,
        cl@linux.com, linux-mm@kvack.org, linux-kernel@vger.kernel.org
References: <20220425132723.34824-1-linmiaohe@huawei.com>
 <20220425132723.34824-3-linmiaohe@huawei.com>
From:   David Hildenbrand <david@redhat.com>
Organization: Red Hat
Subject: Re: [PATCH v2 2/4] mm/migration: remove unneeded lock page and
 PageMovable check
In-Reply-To: <20220425132723.34824-3-linmiaohe@huawei.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 25.04.22 15:27, Miaohe Lin wrote:
> When non-lru movable page was freed from under us, __ClearPageMovable must
> have been done. Even if it's not done, ClearPageIsolated here won't hurt
> as page will be freed anyway. So we can thus remove unneeded lock page and
> PageMovable check here.
> 
> Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
> Reviewed-by: Christoph Hellwig <hch@lst.de>
> ---
>  mm/migrate.c | 8 ++------
>  1 file changed, 2 insertions(+), 6 deletions(-)
> 
> diff --git a/mm/migrate.c b/mm/migrate.c
> index b779646665fe..0fc4651b3e39 100644
> --- a/mm/migrate.c
> +++ b/mm/migrate.c
> @@ -1093,12 +1093,8 @@ static int unmap_and_move(new_page_t get_new_page,
>  		/* page was freed from under us. So we are done. */
>  		ClearPageActive(page);
>  		ClearPageUnevictable(page);
> -		if (unlikely(__PageMovable(page))) {
> -			lock_page(page);
> -			if (!PageMovable(page))
> -				ClearPageIsolated(page);
> -			unlock_page(page);
> -		}
> +		if (unlikely(__PageMovable(page)))
> +			ClearPageIsolated(page);
>  		goto out;
>  	}

Hm, that code+change raises a couple of questions.

We're doing here the same as in putback_movable_pages(). So I guess the
difference here is that the caller did release the reference while the
page was isolated, while we don't assume the same in
putback_movable_pages().


Shouldn't whoever owned the page have cleared that? IOW, is it even
valid that we see a movable or isolated page here (WARN/BUG?)?

At least for balloon compaction, I remember that __PageMovable() is
properly cleared before freeing it via balloon_page_delete().


Also, I am not sure how reliable that page count check is here: if we'd
have another speculative reference to the page, we might see
"page_count(page) > 1" and not take that path, although the previous
owner released the last reference.


-- 
Thanks,

David / dhildenb