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 -next 0/3] md/raid10: reduce lock contention for io
To:     Paul Menzel <pmenzel@molgen.mpg.de>,
        Yu Kuai <yukuai1@huaweicloud.com>
Cc:     song@kernel.org, linux-raid@vger.kernel.org,
        linux-kernel@vger.kernel.org, yi.zhang@huawei.com,
        "yukuai (C)" <yukuai3@huawei.com>
References: <20220829131502.165356-1-yukuai1@huaweicloud.com>
 <70172016-8ae5-92b8-2a43-8f0d4a2f6668@molgen.mpg.de>
From:   Yu Kuai <yukuai1@huaweicloud.com>
Message-ID: <0e22c51b-d36a-2216-4bfd-522049d3ed44@huaweicloud.com>
Date:   Tue, 30 Aug 2022 09:09:49 +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: <70172016-8ae5-92b8-2a43-8f0d4a2f6668@molgen.mpg.de>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk

Hi, Paul!

在 2022/08/29 21:58, Paul Menzel 写道:
> Dear Yu,
> 
> 
> Thank you for your patches.
> 
> Am 29.08.22 um 15:14 schrieb Yu Kuai:
>> From: Yu Kuai <yukuai3@huawei.com>
>>
>> patch 1 is a small problem found by code review.
>> patch 2 avoid holding resync_lock in fast path.
>> patch 3 avoid holding lock in wake_up() in fast path.
>>
>> Test environment:
>>
>> Architecture: aarch64
>> Cpu: Huawei KUNPENG 920, there are four numa nodes
>>
>> Raid10 initialize:
>> mdadm --create /dev/md0 --level 10 --bitmap none --raid-devices 4 
>> /dev/nvme0n1 /dev/nvme1n1 /dev/nvme2n1 /dev/nvme3n1
>>
>> Test cmd:
>> fio -name=0 -ioengine=libaio -direct=1 -group_reporting=1 
>> -randseed=2022 -rwmixread=70 -refill_buffers -filename=/dev/md0 
>> -numjobs=16 -runtime=60s -bs=4k -iodepth=256 -rw=randread
>>
>> Test result:
>> before this patchset:    2.9 GiB/s
>> after this patchset:    6.6 Gib/s
> 
> Could you please give more details about the test setup, like the drives 
> used?

test setup is described above, four nvme disks is used.
> 
> Did you use some tools like ftrace to figure out the bottleneck?

Yes, I'm sure the bottleneck is spin_lock(), specifically threads from
multiple nodes try to grab the same lock. By the way, if I bind the
threads to the same node, performance can also improve to 6.6 Gib/s
without this patchset.

Thanks,
Kuai
> 
>> Please noted that in kunpeng-920, memory access latency is very bad
>> accross nodes compare to local node, and in other architecture
>> performance improvement might not be significant.
>>
>> Yu Kuai (3):
>>    md/raid10: fix improper BUG_ON() in raise_barrier()
>>    md/raid10: convert resync_lock to use seqlock
>>    md/raid10: prevent unnecessary calls to wake_up() in fast path
>>
>>   drivers/md/raid10.c | 88 +++++++++++++++++++++++++++++----------------
>>   drivers/md/raid10.h |  2 +-
>>   2 files changed, 59 insertions(+), 31 deletions(-)
> 
> 
> Kind regards,
> 
> Paul
> .
>