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;
Subject: Re: [PATCH 0/3] add page_ext_data to get client data in page_ext
To:     Mike Rapoport <rppt@kernel.org>
Cc:     akpm@linux-foundation.org, pasha.tatashin@soleen.com,
        linux-mm@kvack.org, linux-kernel@vger.kernel.org
References: <20230718145812.1991717-1-shikemeng@huaweicloud.com>
 <20230719094421.GI1901145@kernel.org>
 <aa2134aa-3767-b9fe-1504-d6945e8c2f7b@huaweicloud.com>
 <20230720053736.GM1901145@kernel.org>
From:   Kemeng Shi <shikemeng@huaweicloud.com>
Message-ID: <cef4f633-8d5c-0029-cbac-feed9f106bfc@huaweicloud.com>
Date:   Thu, 20 Jul 2023 16:37:29 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:60.0) Gecko/20100101
 Thunderbird/60.5.0
MIME-Version: 1.0
In-Reply-To: <20230720053736.GM1901145@kernel.org>
Content-Type: text/plain; charset=gbk
Content-Transfer-Encoding: 7bit
Precedence: bulk


on 7/20/2023 1:37 PM, Mike Rapoport wrote:
> Hi,
> 
> On Thu, Jul 20, 2023 at 10:38:39AM +0800, Kemeng Shi wrote:
>>
>> on 7/19/2023 5:44 PM, Mike Rapoport wrote:
>>> On Tue, Jul 18, 2023 at 10:58:09PM +0800, Kemeng Shi wrote:
>>>> Current client get data from page_ext by adding offset which is auto
>>>> generated in page_ext core and expose the data layout design insdie
>>>> page_ext core. This series adds a page_ext_data to hide offset from
>>>> client. Thanks!
>>>
>>> Implementers of page_ext_operations are anyway intimately related to
>>> page_ext, so I'm not convinced this has any value.
>>>  
>> Hi Mike, thanks for reply. I thinks page_ext_operations can be futher splited
>> into public part which used by client to simply register and private part which
>> only page_ext core cares and should not be accessed by client directly
>> to reduce decoupling.
> 
> It would be easier to justify changes in this series if they were a part of
> the refactoring you describe here.
Actually, it's not the refactoring trigger this. I found offset used
in client code while I could not find intialization of offset in client.
After some search, I found how offset is generated in page_ext core and
it's more like a page_ext internal behavior. Feel it's better to reduce
dependency from upper level client to low level internal implementation.

>> This series makes offset to be private which client
>> doesn't really care to hide data layout inside page_ext core from client.
>> There are some concrete gains I can list for now:
>> 1. Future client cound call page_ext_data directly instead of define a
>> new function like get_page_owner to get it's data.
>> 2. No change to client if layout of page_ext data change.
> 
> These should be a part of the changelog.
Yes, it's better to highlight the gains. This series was taken into the tree.
I'm not sure if I need send a v2 to include this or Andrew could add this
when code merged to more stable tree.
>> I hope this could be more convincing to you now.
>> Thanks!
>>
>> -- 
>> Best wishes
>> Kemeng Shi
>>
> 

-- 
Best wishes
Kemeng Shi