linux-kernel - Re: [PATCH -next v2 0/7] limit the number of plugged bio

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: <88d2b016-376c-d65d-ef04-d9cdf5d7b096@huaweicloud.com>
Date:   Sat, 13 May 2023 10:03:50 +0800
From:   Yu Kuai <yukuai1@...weicloud.com>
To:     Song Liu <song@...nel.org>, Yu Kuai <yukuai1@...weicloud.com>
Cc:     akpm@...l.org, neilb@...e.de, linux-raid@...r.kernel.org,
        linux-kernel@...r.kernel.org, yi.zhang@...wei.com,
        yangerkun@...wei.com, "yukuai (C)" <yukuai3@...wei.com>
Subject: Re: [PATCH -next v2 0/7] limit the number of plugged bio

Hi,

在 2023/05/13 8:50, Song Liu 写道:
> On Fri, May 12, 2023 at 2:43 AM Yu Kuai <yukuai1@...weicloud.com> wrote:
>>
>> Hi,
>>
>> 在 2023/04/26 16:20, Yu Kuai 写道:
>>> From: Yu Kuai <yukuai3@...wei.com>
>>>
>>> Changes in v2:
>>>    - remove the patch to rename raid1-10.c
>>>
>>> This patchset tries to limit the number of plugged bio for raid1 and
>>> raid10, which is done in the last patch, other patches are some refactor
>>> and optimizations.
>>>
>>> This patchset is tested with a new test [1], this test triggers dirty
>>> pages writeback for 10s, and in the meantime checks disk inflight.
>>>
>>> Before this patchset, test will fail because inflight exceed threshold
>>> (threshold is set to 4096 in the test, in theory this can be mutch
>>>    greater as long as there are enough dirty pages and memory).
>>>
>>> After this patchset, inflight is within 96 (MAX_PLUG_BIO * copies).
>>>
>>> [1] https://lore.kernel.org/linux-raid/20230426073447.1294916-1-yukuai1@huaweicloud.com/
>>
>> Friendly ping...
> 
> I am sorry for the delay.
> 
> The set looks good overall, but I will need some more time to take a closer
> look. A few comments/questions:
> 
> 1. For functions in raid1-10.c, let's prefix them with raid1_ instead of md_*.
Ok, I'll change that in v3.

> 2. Do we need unplug_wq to be per-bitmap? Would a shared queue work?

I think this can work, the limitation is that global workqueue can
support 256 queued work at a time, but this should be enough.

Thanks,
Kuai
> 
> Thanks,
> Song
> .
>