To: Kanoj Sarcar
cc: k-suganuma@mvj.biglobe.ne.jp (Kimio Suganuma), Paul McKenney/Beaverton/IBM@IBMUS, andi@suse.de, andrea@suse.de, aono@ed2.com1.fc.nec.co.jp, beckman@turbolabs.com, bjorn_helgaas@hp.com, Hubertus Franke/Watson/IBM@IBMUS, Jerry.Harrow@Compaq.com, jwright@engr.sgi.com, kanoj@engr.sgi.com, kumon@flab.fujitsu.co.jp, norton@mclinux.com, suganuma@hpc.bs1.fc.nec.co.jp, sunil.saxena@intel.com, tbutler@hi.com, woodman@missioncriticallinux.com, mikek@sequent.com, Kenneth Rozendal/Austin/IBM@IBMUS
Date: 03/29/01 10:54 PM
From: Kimio Suganuma
Subject: Re: NUMA-on-Linux roadmap, version 2
> > I feel this topology descriptor might be also useful
> > for other purposes. For example, for hot-plugging,
> > dynamic patitioning, MCA handling and so on.
>
> If you can provide some logic, let us all know, it is
> always helpful to know a certain feature can be used
> for multiple things while arguing that it should be done.
For example, CPU online/offline function posted to linux-kernel
by Rusty Russel is using like /proc/sys/kernel/cpu/0, 1, 2...
to control CPU status. Like this, hot plugging for CPU and memory
will need /proc/*something*. Guess when each function to show
or control cpu/memory has each /proc/*something*. It's ugly!
I think it is good idea to gather all interfaces regarding
topology into under /proc/node.
Do I make sense?
> > Do you think this /proc information should be generic
> > or NUMA specific?
>
> At least for /proc/nodeinfo, and possibly even for a
> tree/graph under /proc, I think it would be done even
> for non-NUMA. For example, /proc/nodeinfo on regular
> SMP would just list most of the information that
> /proc/meminfo provides, as well as all the cpus in
> the system.
>
> Of course, Linus (or someone else) might object to
> exporting the same information multiple ways.
It's all right. I just wanted to make sure your thought.
Kimi
--
Kimio Suganuma