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: <20140217151626.5ff99027@hananiah.suse.cz>
Date:	Mon, 17 Feb 2014 15:16:26 +0100
From:	Petr Tesarik <ptesarik@...e.cz>
To:	Henrique de Moraes Holschuh <hmh@....eng.br>
Cc:	Jan Beulich <JBeulich@...e.com>, Borislav Petkov <bp@...en8.de>,
	x86@...nel.org, Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...hat.com>, linux-kernel@...r.kernel.org,
	"H. Peter Anvin" <hpa@...or.com>
Subject: Re: [PATCH] x86: Issue a warning if number of present CPUs >
 maxcpus and CONFIG_HOTPLUG=n

On Mon, 17 Feb 2014 10:40:07 -0300
Henrique de Moraes Holschuh <hmh@....eng.br> wrote:

> On Mon, 17 Feb 2014, Petr Tesarik wrote:
> > Well, if the user passes both nr_cpus and maxcpus parameters to the
> > kernel, I think it's fair to issue two warnings. But if everyone agrees
> > that only the maxcpus warning should be printed in that case, I can
> > send a version 2 of my patch.
> 
> Please remember that the market is full of motherboards with the extremely
> annoying behaviour of declaring ACPI objects for CPU cores that will never
> be available.  This includes a large number of workstation and server boards
> at the very least, from at least one rather large vendor.
> 
> As far as I know, we still don't have a way to realiably detect this and get
> rid of the ghost processors which will *NEVER* become online.  Setting
> maxcpus or nr_cpus manually is the current way to avoid wasting runtime
> resources because of phantom cores that will never become reality.
> 
> So, when you fix the bug that always supress the warnings, you will at the
> same time cause a regression on those boxes, which will now print undesired
> warnings.  If the user has manually set nr_cpus or maxcpus, maybe it would
> be best to not print any warnings or alternatively to downgrade them to
> debug level?

While I appreciate your concerns, I fail to see how they are related.

First, please keep in mind that my patch does not alter the (more
common) case with CONFIG_HOTPLUG_CPU=y in any way. For the (presumably
less common) !HOTPLUG_CPU case, it adds a warning that is always issued
in the HOTPLUG_CPU case.

Second, regarding your use case, I don't think it changes anything. So,
let's say you have a board with 16 CPUs and a MADT that describes 1024
CPUs. You know that there can (physically) be at most 48 CPUs, so you
boot with nr_cpus=48:

  num_processors = 16  /* online CPUs at boot */
  disabled_cpus = 992  /* 1008 - 16 */

This results in:

  total_cpus = 1008   /* this is purely informative, it is *NOT* used
                         to size anything */
  possible = 48       /* clamped to nr_cpu_ids */

A warning message (with or without my patch):
  1024 Processors exceeds NR_CPUS limit of 48

Informative message:
  Allowing 16 CPUs, 32 hotplug CPUs

No other warning (with or without my patch).

Petr Tesarik
SUSE L3 Team 1
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ