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;
Subject: Re: [PATCH] mm/madvise: add MADV_WILLNEED to process_madvise()
To:     zhangkui <zhangkui@oppo.com>, akpm@linux-foundation.org,
        linux-mm@kvack.org, linux-kernel@vger.kernel.org
References: <20210804082010.12482-1-zhangkui@oppo.com>
From:   David Hildenbrand <david@redhat.com>
Organization: Red Hat
Message-ID: <480fd747-d8d7-0830-0b97-bf2069dbef07@redhat.com>
Date:   Wed, 4 Aug 2021 21:08:57 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.11.0
MIME-Version: 1.0
In-Reply-To: <20210804082010.12482-1-zhangkui@oppo.com>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 04.08.21 10:20, zhangkui wrote:
> There is a usecase in Android that an app process's memory is swapped out
> by process_madvise() with MADV_PAGEOUT, such as the memory is swapped to
> zram or a backing device. When the process is scheduled to running, like
> switch to foreground, multiple page faults may cause the app dropped
> frames.
> To reduce the problem, SMS can read-ahead memory of the process immediately
> when the app switches to forground.
> Calling process_madvise() with MADV_WILLNEED can meet this need.
> 
> Signed-off-by: zhangkui <zhangkui@oppo.com>
> ---
>   mm/madvise.c | 1 +
>   1 file changed, 1 insertion(+)
> 
> diff --git a/mm/madvise.c b/mm/madvise.c
> index 6d3d348b17f4..b9681fb3fbb5 100644
> --- a/mm/madvise.c
> +++ b/mm/madvise.c
> @@ -1046,6 +1046,7 @@ process_madvise_behavior_valid(int behavior)
>          switch (behavior) {
>          case MADV_COLD:
>          case MADV_PAGEOUT:
> +       case MADV_WILLNEED:
>                  return true;
>          default:
>                  return false;

I guess this should be fine. In contrast to other (e.g., destructive) 
madvise calls, an application could merely notice the difference that 
for example, in the pagemap something is suddenly no longer swapped, 
even thought the page wasn't touched by the process. But that can also 
happen e.g., during swapoff.

So I think this is fine, but let's hear if there are concerns.

-- 
Thanks,

David / dhildenb