linux-kernel - Re: [PATCH] driver core: shut down devices asynchronously

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: <20230816155435.GA18922@wunner.de>
Date:   Wed, 16 Aug 2023 17:54:35 +0200
From:   Lukas Wunner <lukas@...ner.de>
To:     Stuart Hayes <stuart.w.hayes@...il.com>
Cc:     linux-kernel@...r.kernel.org,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        "Rafael J . Wysocki" <rafael@...nel.org>,
        Tanjore Suresh <tansuresh@...gle.com>,
        Martin Belanger <Martin.Belanger@...l.com>,
        Oliver O'Halloran <oohall@...il.com>,
        Daniel Wagner <dwagner@...e.de>,
        Keith Busch <kbusch@...nel.org>
Subject: Re: [PATCH] driver core: shut down devices asynchronously

On Wed, Aug 16, 2023 at 10:45:18AM -0500, Stuart Hayes wrote:
> Attempt to shut down devices asynchronously, by making a tree of devices with
> associated work and completion structs, to ensure that child devices are shut
> down before parents.
> 
> This can dramatically reduce system shutdown/reboot time on systems that have
> devices that take many seconds to shut down, such as some NVMe drives.  On once
> system tested, the shutdown time went from 11 minutes before the patch to 55
> seconds with the patch.
> 
> The code could be simplified by adding the work and completion structs to
> struct device, but it may make more sense to not burden it with that when there
> is likely enough memory to allocate this at shutdown time, and if there isn???t,
> it just falls back to the current synchronous shutdown.

Please wrap the commit message at 72 chars.

Is there a particular reason why you're not using the infrastructure
provided by kernel/async.c and <async.h>, such as async_schedule()?
It wraps all the work_struct plumbing and also has helpers to await
completion.  I imagine using that might reduce LoC in this patch.

Thanks,

Lukas