linux-kernel - Re: [PATCH] cpu/hotplug: Create SMT sysfs interface for all arches

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: <alpine.DEB.2.21.1903242104280.1798@nanos.tec.linutronix.de>
Date:   Sun, 24 Mar 2019 21:13:18 +0100 (CET)
From:   Thomas Gleixner <tglx@...utronix.de>
To:     Josh Poimboeuf <jpoimboe@...hat.com>
cc:     linux-kernel@...r.kernel.org, x86@...nel.org,
        Andrea Arcangeli <aarcange@...hat.com>,
        Waiman Long <longman@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Jiri Kosina <jikos@...nel.org>
Subject: Re: [PATCH] cpu/hotplug: Create SMT sysfs interface for all arches

On Fri, 1 Mar 2019, Josh Poimboeuf wrote:
> Make the /sys/devices/system/cpu/smt/* files available on all arches, so
> user space has a consistent way to detect whether SMT is enabled.
> 
> The 'control' file now shows 'notsupported' for architectures which
> don't yet have CONFIG_HOTPLUG_SMT.

I'm slowly crawling through my backlog ...

> --- a/Documentation/ABI/testing/sysfs-devices-system-cpu
> +++ b/Documentation/ABI/testing/sysfs-devices-system-cpu
> @@ -514,7 +514,8 @@ Description:	Control Symetric Multi Threading (SMT)
>  			 "on"		SMT is enabled
>  			 "off"		SMT is disabled
>  			 "forceoff"	SMT is force disabled. Cannot be changed.
> -			 "notsupported" SMT is not supported by the CPU
> +			 "notsupported" Runtime SMT toggling is not currently
> +					supported for the architecture

Second thoughts. I'm not really convinced that changing the meaning of
notsupported and in fact overloading it, is the right thing to do.
notsupported means now:

  CPU does not support it - OR - architecture does not support it

That's not pretty and we are surely not short of state space. There are
several options for handling this:

 1) Do not expose the state file, just expose the active file

 2) Expose the state file, but return -ENOTSUPP or some other sensible error
    code

 3) Expose the state file and let show return 'notimplemented' which is
    more accurate. That wouldn't even require to expand the state space
    enum. It just can be returned unconditionally.

Thanks,

	tglx