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: <7113d66a-9792-b026-cece-5c1b21dd989d@gmail.com>
Date:   Thu, 10 Nov 2022 10:48:24 +0800
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:102.0) Gecko/20100101
 Thunderbird/102.4.2
Subject: Re: [PATCH] mm/vmscan: simplify the nr assignment logic for pages to
 scan
To:     Andrew Morton <akpm@linux-foundation.org>
Cc:     linux-mm@kvack.org, linux-kernel@vger.kernel.org,
        Chao Xu <Chao.Xu9@zeekrlife.com>
References: <20221109070416.620887-1-Chao.Xu9@zeekrlife.com>
 <20221109171952.826d991327e07319c5eb8cd3@linux-foundation.org>
From:   Chao Xu <amos.xuchao@gmail.com>
In-Reply-To: <20221109171952.826d991327e07319c5eb8cd3@linux-foundation.org>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk


在 2022/11/10 9:19, Andrew Morton 写道:
> On Wed,  9 Nov 2022 15:04:16 +0800 Chao Xu <amos.xuchao@gmail.com> wrote:
>
>> By default the assignment logic of anonymouns or file inactive
>> pages and active pages to scan using the same duplicated code
>> snippet. To simplify the logic, merge the same part.
>>
>> --- a/mm/vmscan.c
>> +++ b/mm/vmscan.c
>> @@ -5932,14 +5932,11 @@ static void shrink_lruvec(struct lruvec *lruvec, struct scan_control *sc)
>>   		 * scan target and the percentage scanning already complete
>>   		 */
>>   		lru = (lru == LRU_FILE) ? LRU_BASE : LRU_FILE;
>> -		nr_scanned = targets[lru] - nr[lru];
>> -		nr[lru] = targets[lru] * (100 - percentage) / 100;
>> -		nr[lru] -= min(nr[lru], nr_scanned);
>> -
>> -		lru += LRU_ACTIVE;
>> -		nr_scanned = targets[lru] - nr[lru];
>> -		nr[lru] = targets[lru] * (100 - percentage) / 100;
>> -		nr[lru] -= min(nr[lru], nr_scanned);
>> +		for ( ; lru <= lru + LRU_ACTIVE; lru++) {
> The "lru++" implicitly assumes that LRU_ACTIVE=1.  That happens to be
> the case, but a more accurate translation of the existing code would
> use "lru += LRU_ACTIVE" here, yes?
By default the value of LRU_ACTIVE is 1，but if someone change it one day, I
use "lru++" maybe facing some exceptions, which is not robust. So I 
agree with
that "lru += LRU_ACTIVE" is appropriate instead of "lru++". I will send 
a new patch.
>> +			nr_scanned = targets[lru] - nr[lru];
>> +			nr[lru] = targets[lru] * (100 - percentage) / 100;
>> +			nr[lru] -= min(nr[lru], nr_scanned);
>> +		}
>>   
>>   		scan_adjusted = true;
>>   	}