linux-kernel - Re: [PATCH for-6.12 0/4] block, bfq: fix corner cases related to bfqq merging

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <52711505-53db-315c-0e39-143fe8379514@huaweicloud.com>
Date: Thu, 5 Sep 2024 09:48:23 +0800
From: Yu Kuai <yukuai1@...weicloud.com>
To: Bart Van Assche <bvanassche@....org>, Yu Kuai <yukuai1@...weicloud.com>,
 Jens Axboe <axboe@...nel.dk>, jack@...e.cz, tj@...nel.org,
 josef@...icpanda.com, paolo.valente@...more.it, mauro.andreolini@...more.it,
 avanzini.arianna@...il.com
Cc: cgroups@...r.kernel.org, linux-block@...r.kernel.org,
 linux-kernel@...r.kernel.org, yi.zhang@...wei.com, yangerkun@...wei.com,
 "yukuai (C)" <yukuai3@...wei.com>
Subject: Re: [PATCH for-6.12 0/4] block, bfq: fix corner cases related to bfqq
 merging

Hi,

在 2024/09/05 1:17, Bart Van Assche 写道:
> On 9/3/24 7:45 PM, Yu Kuai wrote:
>> 在 2024/09/04 10:28, Bart Van Assche 写道:
>>> On 9/3/24 6:32 PM, Yu Kuai wrote:
>>>> We do have customers are using bfq in downstream kernels, and we are
>>>> still running lots of test for bfq.
>>>
>>> It may take less time to add any missing functionality to another I/O
>>> scheduler rather than to keep maintaining BFQ.
>>>
>>> If Android device vendors would stop using BFQ, my job would become
>>> easier.
>>
>> I'm confused now, I think keep maintaining BFQ won't stop you from
>> adding new functionality to another scheduler, right? Is this something
>> that all scheduler have to support?
> 
> As long as the BFQ I/O scheduler does not get deprecated, there will be
> Android device vendors that select it for their devices. BFQ bug reports
> are either sent to one of my colleagues or to myself.

Then, you can share them to me now, I'll like to help.
> 
> For Android devices that use UFS storage, we noticed that the
> mq-deadline scheduler is good enough. The device boot time is shorter
> and I'm not aware of any significant differences in application startup
> time.

We're using bfq for HDD, performance overhead in bfq is not less,
like you said, if bfq doen't show better results in UFS storage, and you
don't want to use the io control feature, you can choose not to use it,
however, remove bfq will be too aggressive.

Thanks,
Kuai

> 
> Thanks,
> 
> Bart.
> 
> .
>