Received-SPF: pass (google.com: domain of linux-kernel+bounces-143518-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Subject: Re: [PATCH RFC v2 6/6] blk-throtl: switch to use rq_qos
To: Tejun Heo <tj@kernel.org>, Yu Kuai <yukuai1@huaweicloud.com>
Cc: axboe@kernel.dk, chenhuacai@kernel.org, josef@toxicpanda.com,
 jhs@mojatatu.com, svenjoac@gmx.de, raven@themaw.net, pctammela@mojatatu.com,
 qde@naccy.de, zhaotianrui@loongson.cn, linux-block@vger.kernel.org,
 linux-kernel@vger.kernel.org, loongarch@lists.linux.dev,
 cgroups@vger.kernel.org, yi.zhang@huawei.com, yangerkun@huawei.com,
 "yukuai (C)" <yukuai3@huawei.com>
References: <20240406080059.2248314-1-yukuai1@huaweicloud.com>
 <20240406080059.2248314-7-yukuai1@huaweicloud.com>
 <Zhl5ONFlPg4vqjGj@slm.duckdns.org>
From: Yu Kuai <yukuai1@huaweicloud.com>
Message-ID: <b522c2b7-efae-a7ca-ee6c-197a4b9b54ff@huaweicloud.com>
Date: Sat, 13 Apr 2024 10:17:51 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:60.0) Gecko/20100101
 Thunderbird/60.8.0
Precedence: bulk
MIME-Version: 1.0
In-Reply-To: <Zhl5ONFlPg4vqjGj@slm.duckdns.org>
Content-Type: text/plain; charset=gbk; format=flowed
Content-Transfer-Encoding: 8bit

Hi,

?? 2024/04/13 2:11, Tejun Heo д??:
> Hello,
> 
> On Sat, Apr 06, 2024 at 04:00:59PM +0800, Yu Kuai wrote:
>> From: Yu Kuai <yukuai3@huawei.com>
>>
>> To avoid exposing blk-throttle internal implementation to general block
>> layer.
> ...
>> @@ -832,7 +832,7 @@ void submit_bio_noacct(struct bio *bio)
>>   		goto not_supported;
>>   	}
>>   
>> -	if (blk_throtl_bio(bio))
>> +	if (rq_qos_throttle_bio(q, bio))
>>   		return;
>>   	submit_bio_noacct_nocheck(bio);
>>   	return;
> 
> This is a half-way conversion, right? You're adding a dedicated hook to
> rq_qos and none of the other hooks can be used by blk-throtl. Even the name,
> rq_qos_throttle_bio(), becomes a misnomer. I'm not really sure this makes
> things better or worse. It makes certain things a bit cleaner but other
> things nastier. I don't know.

Yes, the final goal is making all blk-cgroup policies modular, and this
patch use rq-qos to prevent exposing blk-throtle to block layer, like
other policies.

There is another choice that I think is feasible:

Let blk-throttle ping a policy id, and use the id to call throttle
function directly, this will require initializing the 'plid' from
blkcg_policy() during definition instead of blkcg_policy_register().

Thanks,
Kuai

> 
> Thanks.
>