linux-kernel - Re: [External Mail]Re: [PATCH v3] block: move non sync requests complete flow to softirq

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

Message-ID: <31b99746-badc-4304-877d-790c6ff95c80@xiaomi.com>
Date: Wed, 4 Sep 2024 03:25:55 +0000
From: 章辉 <zhanghui31@...omi.com>
To: Jens Axboe <axboe@...nel.dk>, "bvanassche@....org" <bvanassche@....org>
CC: "linux-block@...r.kernel.org" <linux-block@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [External Mail]Re: [PATCH v3] block: move non sync requests
 complete flow to softirq

On 2024/9/4 1:49, Jens Axboe wrote:
> On 9/3/24 5:54 AM, ZhangHui wrote:
>> From: zhanghui <zhanghui31@...omi.com>
>>
>> Currently, for a controller that supports multiple queues, like UFS4.0,
>> the mq_ops->complete is executed in the interrupt top-half. Therefore,
>> the file system's end io is executed during the request completion process,
>> such as f2fs_write_end_io on smartphone.
>>
>> However, we found that the execution time of the file system end io
>> is strongly related to the size of the bio and the processing speed
>> of the CPU. Because the file system's end io will traverse every page
>> in bio, this is a very time-consuming operation.
>>
>> We measured that the 80M bio write operation on the little CPU will
>> cause the execution time of the top-half to be greater than 100ms.
>> The CPU tick on a smartphone is only 4ms, which will undoubtedly affect
>> scheduling efficiency.
> The elephant in the room here is why an 80M completion takes 100 msec?
> That seems... insane.
>
> That aside, doing writes that big isn't great for latencies in general,
> even if they are orders of magnitude smaller (as they should be). Maybe
> this is solvable by just limiting the write size here.
>
> But it really seems out of line for a write that size to take 100 msec
> to process.
>
> --
> Jens Axboe
>
hi Jens,

This problem is strongly related to whether the CPU is a large
core or a little core and the CPU frequency. On a large core, the time
will obviously be shorter, but we cannot assume which core the IO will
be completed on and the current CPU frequency...

Limiting the IO size is also a method, but how large to limit it is a
problem, and I am worried whether it will cause bandwidth loss in
some scenarios?

Thanks
Zhang