Home » Mailing list archives » linux-kernel » 2008 » April » 28

Re: [PATCH 01/11] x86: Modify Kconfig to allow up to 4096 cpus

Previous thread: [PATCH 00/11] x86: cleanup early per cpu variables/accesses v5-folded by Mike Travis on Friday, April 25, 2008 - 8:15 pm. (4 messages)

Next thread: [PATCH 03/11] x86: restore pda nodenumber field by Mike Travis on Friday, April 25, 2008 - 8:15 pm. (1 message)
[view original message]
From: Mike Travis <travis@...>
To: Ingo Molnar <mingo@...>
Cc: Andrew Morton <akpm@...>, Thomas Gleixner <tglx@...>, H. Peter Anvin <hpa@...>, <linux-kernel@...>
Subject: [PATCH 01/11] x86: Modify Kconfig to allow up to 4096 cpus
Date: Friday, April 25, 2008 - 8:15 pm

* Increase the limit of NR_CPUS to 4096 and introduce a boolean

    called "MAXSMP" which when set (e.g. "allyesconfig"), will set

    NR_CPUS = 4096 and NODES_SHIFT = 9 (512).


  * Changed max setting for NODES_SHIFT from 15 to 9 to accurately

    reflect the real limit.


For inclusion into sched-devel/latest tree.


Based on:

	git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git

    +   sched-devel/latest  .../mingo/linux-2.6-sched-devel.git


Signed-off-by: Mike Travis <travis@sgi.com>

---

 arch/x86/Kconfig |   37 ++++++++++++++++++++++++++++++++-----

 1 file changed, 32 insertions(+), 5 deletions(-)


--- linux-2.6.sched.orig/arch/x86/Kconfig

+++ linux-2.6.sched/arch/x86/Kconfig

@@ -513,20 +513,35 @@ config SWIOTLB

 	  access 32-bits of memory can be used on systems with more than

 	  3 GB of memory. If unsure, say Y.


+config MAXSMP

+	bool "Configure Maximum number of SMP Processors and NUMA Nodes"

+	depends on X86_64 && SMP

+	default n

+	help

+	  Configure maximum number of CPUS and NUMA Nodes for this architecture.

+	  If unsure, say N.

+

+if MAXSMP

+config NR_CPUS

+	int

+	default "4096"

+endif


+if !MAXSMP

 config NR_CPUS

-	int "Maximum number of CPUs (2-255)"

-	range 2 255

+	int "Maximum number of CPUs (2-4096)"

+	range 2 4096

 	depends on SMP

 	default "32" if X86_NUMAQ || X86_SUMMIT || X86_BIGSMP || X86_ES7000

 	default "8"

 	help

 	  This allows you to specify the maximum number of CPUs which this

-	  kernel will support.  The maximum supported value is 255 and the

+	  kernel will support.  The maximum supported value is 4096 and the

 	  minimum value which makes sense is 2.


 	  This is purely to save memory - each supported CPU adds

 	  approximately eight kilobytes to the kernel image.

+endif


 config SCHED_SMT

 	bool "SMT (Hyperthreading) scheduler support"

@@ -917,13 +932,25 @@ config NUMA_EMU

 	  into virtual nodes when booted with "numa=fake=N", where N is the

 	  number of nodes. This is only usef...
[ message continues ]
[view original message]
From: Pavel Machek <pavel@...>
To: Mike Travis <travis@...>
Cc: Ingo Molnar <mingo@...>, Andrew Morton <akpm@...>, Thomas Gleixner <tglx@...>, H. Peter Anvin <hpa@...>, <linux-kernel@...>
Subject: Re: [PATCH 01/11] x86: Modify Kconfig to allow up to 4096 cpus
Date: Sunday, April 27, 2008 - 6:39 am

Why is redundant option 'maxsmp' a good idea?

--

(english) http://www.livejournal.com/~pavelmachek

(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

--
[view original message]
From: Ingo Molnar <mingo@...>
To: Pavel Machek <pavel@...>
Cc: Mike Travis <travis@...>, Andrew Morton <akpm@...>, Thomas Gleixner <tglx@...>, H. Peter Anvin <hpa@...>, <linux-kernel@...>
Subject: Re: [PATCH 01/11] x86: Modify Kconfig to allow up to 4096 cpus
Date: Monday, April 28, 2008 - 9:38 am

because this way randconfig can trigger it and can configure all the

otherwise randconfig-invisible [or just plain unlikely] numerics and

options up to their max.


I found 2-3 "large box" bugs via that way already.


	Ingo

--
[view original message]
From: Pavel Machek <pavel@...>
To: Ingo Molnar <mingo@...>
Cc: Mike Travis <travis@...>, Andrew Morton <akpm@...>, Thomas Gleixner <tglx@...>, H. Peter Anvin <hpa@...>, <linux-kernel@...>
Subject: Re: [PATCH 01/11] x86: Modify Kconfig to allow up to 4096 cpus
Date: Monday, April 28, 2008 - 9:54 am

Should we improve randconfig to select numeric values when range is

given, too?

								Pavel

--

(english) http://www.livejournal.com/~pavelmachek

(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

--
[view original message]
From: H. Peter Anvin <hpa@...>
To: Pavel Machek <pavel@...>
Cc: Ingo Molnar <mingo@...>, Mike Travis <travis@...>, Andrew Morton <akpm@...>, Thomas Gleixner <tglx@...>, <linux-kernel@...>
Subject: Re: [PATCH 01/11] x86: Modify Kconfig to allow up to 4096 cpus
Date: Monday, April 28, 2008 - 1:13 pm

Would not be a bad idea; preferrably with a logarithmic distribution.


	-hpa

--
[view original message]
From: Ingo Molnar <mingo@...>
To: Pavel Machek <pavel@...>
Cc: Mike Travis <travis@...>, Andrew Morton <akpm@...>, Thomas Gleixner <tglx@...>, H. Peter Anvin <hpa@...>, <linux-kernel@...>
Subject: Re: [PATCH 01/11] x86: Modify Kconfig to allow up to 4096 cpus
Date: Monday, April 28, 2008 - 11:52 am

definitely - but often there are constraints on the numeric value that

are not spelled out in the Kconfig language. (alignment, granularity,

etc.)


MAXSMP is still useful, as it would still be rather unlikely for all

numeric values to go to the max at once. The other idea is that with it

'make allyesconfig' really puts all build-time limits up to the max.


	Ingo

--
Previous thread: [PATCH 00/11] x86: cleanup early per cpu variables/accesses v5-folded by Mike Travis on Friday, April 25, 2008 - 8:15 pm. (4 messages)

Next thread: [PATCH 03/11] x86: restore pda nodenumber field by Mike Travis on Friday, April 25, 2008 - 8:15 pm. (1 message)