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 v2 1/2] blk-iocost: add refcounting for iocg
To:     Tejun Heo <tj@kernel.org>, Yu Kuai <yukuai1@huaweicloud.com>
Cc:     hch@infradead.org, josef@toxicpanda.com, axboe@kernel.dk,
        cgroups@vger.kernel.org, linux-block@vger.kernel.org,
        linux-kernel@vger.kernel.org, yi.zhang@huawei.com,
        "yukuai (C)" <yukuai3@huawei.com>
References: <20221227125502.541931-1-yukuai1@huaweicloud.com>
 <20221227125502.541931-2-yukuai1@huaweicloud.com>
 <Y7XzUee5Bq+DoIC1@slm.duckdns.org>
From:   Yu Kuai <yukuai1@huaweicloud.com>
Message-ID: <c63ee2ad-23d5-3be0-c731-28494398b391@huaweicloud.com>
Date:   Thu, 5 Jan 2023 09:14:07 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:60.0) Gecko/20100101
 Thunderbird/60.8.0
MIME-Version: 1.0
In-Reply-To: <Y7XzUee5Bq+DoIC1@slm.duckdns.org>
Content-Type: text/plain; charset=gbk; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hi,

?? 2023/01/05 5:44, Tejun Heo д??:
> On Tue, Dec 27, 2022 at 08:55:01PM +0800, Yu Kuai wrote:
>> From: Yu Kuai <yukuai3@huawei.com>
>>
>> iocost requires that child iocg must exit before parent iocg, otherwise
>> kernel might crash in ioc_timer_fn(). However, currently iocg is exited
>> in pd_free_fn(), which can't guarantee such order:
>>
>> 1) remove cgroup can concurrent with deactivate policy;
>> 2) blkg_free() triggered by remove cgroup is asynchronously, remove
>> child cgroup can concurrent with remove parent cgroup;
>>
>> Fix the problem by add refcounting for iocg, and child iocg will grab
>> reference of parent iocg, so that parent iocg will wait for all child
>> iocg to be exited.
> 
> Wouldn't it be better to do this refcnting in the blk-cgroup core code
> rather than in blk-iocost?
> 

The problem is that I can't find a proper way to fix the competition
that pd_free_fn() can be called from different context:

1) from blkg_free() that is called asynchronously from removing cgroup;
2) from blkcg_deactivate_policy() that is called from removing device;

1) is related to blkg, while 2) is not, hence refcnting from blkg can't
fix the problem. refcnting from blkcg_policy_data should be ok, but I
see that bfq already has the similar refcnting, while other policy
doesn't require such refcnting.

Any suggestions?

Thanks,
Kuai
> Thanks.
>