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: <f67e8d9b-dae8-fced-ce7b-b0f8fff2d127@deltatee.com>
Date:   Tue, 11 Apr 2023 09:46:14 -0600
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.9.0
Content-Language: en-CA
To:     Song Liu <song@kernel.org>, Yu Kuai <yukuai1@huaweicloud.com>
Cc:     linux-kernel@vger.kernel.org, linux-raid@vger.kernel.org,
        yi.zhang@huawei.com, yangerkun@huawei.com,
        "yukuai (C)" <yukuai3@huawei.com>
References: <20230410113559.1610455-1-yukuai1@huaweicloud.com>
 <20230410113559.1610455-2-yukuai1@huaweicloud.com>
 <CAPhsuW7rUQ9KFxCQw-iAAA64=x==b-Qu0nnp32TXn-85Okb00w@mail.gmail.com>
 <daa344fe-9d8d-16f9-aded-513f250928b9@huaweicloud.com>
 <CAPhsuW7bBaVsneBHYqwN3yd5k3sg6ZUFqEJad3yOD0=N-e411g@mail.gmail.com>
From:   Logan Gunthorpe <logang@deltatee.com>
In-Reply-To: <CAPhsuW7bBaVsneBHYqwN3yd5k3sg6ZUFqEJad3yOD0=N-e411g@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Subject: Re: [PATCH -next v5 1/6] md: pass a md_thread pointer to
 md_register_thread()
Precedence: bulk


On 2023-04-10 23:26, Song Liu wrote:
> On Mon, Apr 10, 2023 at 6:34 PM Yu Kuai <yukuai1@huaweicloud.com> wrote:
>>
>> Hi,
>>
>> 在 2023/04/11 9:15, Song Liu 写道:
>>> On Mon, Apr 10, 2023 at 4:37 AM Yu Kuai <yukuai1@huaweicloud.com> wrote:
>>>>
>>>> From: Yu Kuai <yukuai3@huawei.com>
>>>>
>>>> Prepare to protect md_thread with rcu, there are no functional changes.
>>>
>>> Why do we need this change? To add __rcu later?
>>
>> Add __rcu is one reason, more importantly is to assign md_thread inside
>> md_register_thread in patch 6:
>>
>> rcu_assign_pointer(*threadp, thread);
> 
> Got it.
> 
>>>
>>> Can we do something like:
>>>
>>> struct md_thread __rcu *md_register_thread(void (*run) (struct md_thread *),
>>>                 struct mddev *mddev, const char *name)
>>
>> I think this is not necessary, if we don't want to change api, we must
>> use rcu_assign_pointer for each caller to set md_thread.
> 
> I think it is better to use rcu_assign_pointer at the caller side.

I agree.

Logan