linux-kernel - Re: [PATCH] raid10: avoid spin_lock from fastpath from raid10

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]

Date:   Tue, 20 Jun 2023 10:17:48 -0700
From:   Song Liu <song@...nel.org>
To:     Yu Kuai <yukuai1@...weicloud.com>
Cc:     Paul Menzel <pmenzel@...gen.mpg.de>, aligrudi@...il.com,
        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] raid10: avoid spin_lock from fastpath from raid10_unplug()

On Mon, Jun 19, 2023 at 6:00 AM Yu Kuai <yukuai1@...weicloud.com> wrote:
>
> Hi,
>
> 在 2023/06/19 18:26, Paul Menzel 写道:
> > Dear Yu,
> >
> >
> > Thank you for your patch. Some minor nits from my side, you can also
> > ignore.
> >
> > Am 18.06.23 um 16:25 schrieb Yu Kuai:
> >> From: Yu Kuai <yukuai3@...wei.com>
> >>
> >> Commit 0c0be98bbe67 ("md/raid10: prevent unnecessary calls to wake_up()
> >> in fast path") missed one place, for example, while testing with:
> >
> > … one place. For example, with
> >
> >>
> >> fio -direct=1 -rw=write/randwrite -iodepth=1 ...
> >>
> >> Then plug and unplug will be called for each io, then wake_up() from
> >
> > Maybe:
> >
> >      fio -direct=1 -rw=write/randwrite -iodepth=1 ...
> >
> > plug und unplug are called for each io, then …
> >
> >> raid10_unplug() will cause lock contention as well.
> >
> > Maybe paste the perf command and output?
>
> Related test and test result and perf result can be found in the below
> Link.

We don't need a lot of details of the test results in the commit log. But a
quick summary of the performance result can be really helpful here.
Something like: on array X, fio job Y, before the patch we got AAA, with
the patch, we got BBB.

Other than this, the patch looks good to me.

Thanks,
Song