From: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
To: Andrew Morton <akpm@linux-foundation.org>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>,
Andi Kleen <andi@firstfloor.org>,
linux-kernel@vger.kernel.org, mpm@selenic.com, "Huang,
Ying" <ying.huang@intel.com>,
Thomas Gleixner <tglx@linutronix.de>,
Christoph Lameter <clameter@sgi.com>,
apw@shadowen.org
Subject: Re: x86 patches was Re: -mm merge plans for 2.6.24
Date: Tue, 02 Oct 2007 13:25:29 -0400 [thread overview]
Message-ID: <1191345930.5026.9.camel@localhost> (raw)
In-Reply-To: <20071002004324.be030b63.akpm@linux-foundation.org>
On Tue, 2007-10-02 at 00:43 -0700, Andrew Morton wrote:
> On Tue, 2 Oct 2007 16:36:24 +0900 KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:
>
> > On Tue, 2 Oct 2007 00:18:09 -0700
> > Andrew Morton <akpm@linux-foundation.org> wrote:
> > > >
> > > > > How come? Memoryless node can and do occur in real-world machines. Kernel
> > > > > should support that?
> > > >
> > > > But a node is just defined by its memory?
> > >
> > > Don't think so. A node is a lump of circuitry which can have zero or more
> > > CPUs, IO and memory.
> > >
> > > It may initially have been conceived as a memory-only concept in the Linux
> > > kernel, but that doesn't fully map onto reality (does it?)
> > >
> > > There was a real-world need for this, I think from the Fujitsu guys. That
> > > should be spelled out in the changelog but isn't.
> >
> > Yes, Fujitsu and HP guys really need this memory-less-node support.
> >
>
> For what reason, please?
For the HP platforms, we can configure each cell with from 0% to 100%
"cell local memory". When we configure with <100% CLM, the "missing
percentages" are interleaved by hardware on a cache-line granularity to
improve bandwidth at the expense of latency for numa-challenged
applications [and OSes, but not our problem ;-)]. When we boot Linux on
such a config, all of the real nodes have no memory--it all resides in a
single interleaved pseudo-node.
When we boot Linux on a 100% CLM configuration [== NUMA], we still have
the interleaved pseudo-node. It contains a few hundred MB stolen from
the real nodes to contain the DMA zone. [Interleaved memory resides at
phys addr 0]. The memoryless-nodes patches, along with the zoneorder
patches, support this config as well.
Also, when we boot a NUMA config with the "mem=" command line,
specifying less memory than actually exists, Linux takes the excluded
memory "off the top" rather than distributing it across the nodes. This
can result in memoryless nodes, as well.
Lee
next prev parent reply other threads:[~2007-10-02 17:25 UTC|newest]
Thread overview: 112+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-10-01 21:22 -mm merge plans for 2.6.24 Andrew Morton
2007-10-01 21:34 ` wibbling over the cpuset shed domain connnection Paul Jackson
2007-10-02 12:36 ` Nick Piggin
2007-10-03 5:21 ` Paul Jackson
2007-10-02 13:12 ` Nick Piggin
2007-10-03 7:00 ` Paul Jackson
2007-10-03 10:57 ` Andrew Morton
2007-10-02 4:21 ` Memory controller merge (was Re: -mm merge plans for 2.6.24) Balbir Singh
2007-10-02 15:46 ` Hugh Dickins
2007-10-03 8:13 ` Balbir Singh
2007-10-03 18:47 ` Hugh Dickins
2007-10-04 4:16 ` Balbir Singh
2007-10-04 13:16 ` Hugh Dickins
2007-10-05 3:07 ` Balbir Singh
2007-10-07 17:41 ` Hugh Dickins
2007-10-08 2:54 ` Balbir Singh
2007-10-04 16:10 ` Paul Menage
2007-10-10 21:07 ` Rik van Riel
2007-10-11 6:33 ` Balbir Singh
2007-10-02 6:18 ` x86 patches was Re: -mm merge plans for 2.6.24 Andi Kleen
2007-10-02 6:32 ` Andrew Morton
2007-10-02 7:01 ` Andi Kleen
2007-10-02 7:18 ` Andrew Morton
2007-10-02 7:36 ` KAMEZAWA Hiroyuki
2007-10-02 7:43 ` Andrew Morton
2007-10-02 8:16 ` KAMEZAWA Hiroyuki
2007-10-02 10:48 ` Yasunori Goto
2007-10-02 18:18 ` Christoph Lameter
2007-10-02 17:25 ` Lee Schermerhorn [this message]
2007-10-02 16:40 ` Nish Aravamudan
2007-10-02 17:17 ` Lee Schermerhorn
2007-10-02 18:16 ` Christoph Lameter
2007-10-02 7:55 ` Matt Mackall
2007-10-02 7:59 ` Andi Kleen
2007-10-02 9:26 ` Andy Whitcroft
2007-10-02 7:37 ` Ingo Molnar
2007-10-02 7:46 ` Andi Kleen
2007-10-02 7:58 ` Thomas Gleixner
2007-10-02 7:59 ` v4l-stk11xx* [Was: -mm merge plans for 2.6.24] Jiri Slaby
[not found] ` <4701FC79.3060608@gmail.com>
2007-10-02 8:10 ` Wireless damage " Jiri Slaby
2007-10-02 8:17 ` per BDI dirty limit (was Re: -mm merge plans for 2.6.24) Peter Zijlstra
[not found] ` <20071002082831.GA19954@mail.ustc.edu.cn>
2007-10-02 8:28 ` Fengguang Wu
2007-10-02 8:31 ` Andrew Morton
2007-10-02 8:48 ` Peter Zijlstra
2007-10-02 10:31 ` Kay Sievers
2007-10-02 10:44 ` Peter Zijlstra
[not found] ` <20071002104734.GA9410@mail.ustc.edu.cn>
2007-10-02 10:47 ` Fengguang Wu
2007-10-02 11:22 ` Kay Sievers
[not found] ` <20071002112802.GA12607@mail.ustc.edu.cn>
2007-10-02 11:28 ` Fengguang Wu
2007-10-02 11:21 ` Kay Sievers
2007-10-02 11:40 ` Peter Zijlstra
2007-10-02 12:05 ` Nick Piggin
2007-10-03 10:15 ` Kay Sievers
2007-10-03 10:37 ` Peter Zijlstra
2007-10-03 13:35 ` Kay Sievers
2007-10-03 13:58 ` Peter Zijlstra
2007-10-26 14:48 ` Peter Zijlstra
2007-10-26 15:06 ` Miklos Szeredi
2007-10-26 15:10 ` Kay Sievers
2007-10-26 15:22 ` Peter Zijlstra
2007-10-26 15:33 ` Kay Sievers
2007-10-26 15:33 ` Peter Zijlstra
2007-10-26 15:55 ` Kay Sievers
2007-10-26 20:04 ` Peter Zijlstra
2007-10-27 1:18 ` Peter Zijlstra
2007-10-27 2:40 ` Greg KH
2007-10-27 8:39 ` Peter Zijlstra
2007-10-27 16:02 ` Greg KH
2007-10-27 16:07 ` Peter Zijlstra
2007-10-27 21:08 ` Kay Sievers
2007-10-27 21:35 ` Peter Zijlstra
2007-10-28 7:10 ` Greg KH
2007-11-02 13:15 ` Peter Zijlstra
2007-11-02 13:50 ` Kay Sievers
2007-11-02 13:54 ` Peter Zijlstra
2007-11-02 14:17 ` Peter Zijlstra
2007-11-02 14:32 ` Kay Sievers
2007-11-02 14:59 ` [PATCH] mm: sysfs: expose the BDI object in sysfs Peter Zijlstra
2007-11-02 15:13 ` Kay Sievers
2007-10-26 16:37 ` per BDI dirty limit (was Re: -mm merge plans for 2.6.24) Trond Myklebust
2007-12-14 14:50 ` Peter Zijlstra
2007-12-14 15:14 ` Miklos Szeredi
2007-12-14 15:54 ` Peter Zijlstra
2007-10-02 14:38 ` Kay Sievers
2007-10-03 11:00 ` Martin Knoblauch
[not found] ` <20071002083922.GA28892@mail.ustc.edu.cn>
2007-10-02 8:39 ` writeback fixes Fengguang Wu
2007-10-02 16:06 ` kswapd min order, slub max order [was Re: -mm merge plans for 2.6.24] Hugh Dickins
2007-10-02 9:10 ` Nick Piggin
2007-10-02 18:38 ` Mel Gorman
2007-10-02 18:28 ` Christoph Lameter
2007-10-03 0:37 ` Christoph Lameter
2007-10-02 16:12 ` -mm merge plans for 2.6.24 Pekka Enberg
2007-10-02 16:21 ` new aops merge [was Re: -mm merge plans for 2.6.24] Hugh Dickins
2007-10-02 17:45 ` remove zero_page (was Re: -mm merge plans for 2.6.24) Nick Piggin
2007-10-03 10:58 ` Andrew Morton
2007-10-03 15:21 ` Linus Torvalds
2007-10-08 15:17 ` Nick Piggin
2007-10-09 13:00 ` Hugh Dickins
2007-10-09 14:52 ` Linus Torvalds
2007-10-09 9:31 ` Nick Piggin
2007-10-10 2:22 ` Linus Torvalds
2007-10-09 10:15 ` Nick Piggin
2007-10-10 3:06 ` Linus Torvalds
2007-10-10 4:06 ` Hugh Dickins
2007-10-10 5:20 ` Linus Torvalds
2007-10-09 14:30 ` Nick Piggin
2007-10-10 15:04 ` Linus Torvalds
2007-10-03 19:50 ` A kernel Tracing interface " David Wilder
2007-10-09 9:19 ` r/o bind mounts, was Re: -mm merge plans for 2.6.24 Christoph Hellwig
2007-10-13 8:44 ` Borislav Petkov
2007-10-13 8:52 ` Andrew Morton
2007-10-13 11:45 ` Borislav Petkov
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1191345930.5026.9.camel@localhost \
--to=lee.schermerhorn@hp.com \
--cc=akpm@linux-foundation.org \
--cc=andi@firstfloor.org \
--cc=apw@shadowen.org \
--cc=clameter@sgi.com \
--cc=kamezawa.hiroyu@jp.fujitsu.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mpm@selenic.com \
--cc=tglx@linutronix.de \
--cc=ying.huang@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox