From owner-freebsd-smp  Tue Feb  4 12:33:37 2003
Delivered-To: freebsd-smp@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP
	id 5ABF137B401; Tue,  4 Feb 2003 12:33:36 -0800 (PST)
Received: from heron.mail.pas.earthlink.net (heron.mail.pas.earthlink.net [207.217.120.189])
	by mx1.FreeBSD.org (Postfix) with ESMTP
	id BA42343F75; Tue,  4 Feb 2003 12:33:35 -0800 (PST)
	(envelope-from tlambert2@mindspring.com)
Received: from pool0327.cvx21-bradley.dialup.earthlink.net ([209.179.193.72] helo=mindspring.com)
	by heron.mail.pas.earthlink.net with asmtp (SSLv3:RC4-MD5:128)
	(Exim 3.33 #1)
	id 18g9l7-00012O-00; Tue, 04 Feb 2003 12:33:33 -0800
Message-ID: <3E402346.F90982F3@mindspring.com>
Date: Tue, 04 Feb 2003 12:32:06 -0800
From: Terry Lambert <tlambert2@mindspring.com>
X-Mailer: Mozilla 4.79 [en] (Win98; U)
X-Accept-Language: en
MIME-Version: 1.0
To: Julian Elischer <julian@elischer.org>
Cc: John Baldwin <jhb@FreeBSD.org>, freebsd-smp@freebsd.org,
	Chris Luke <chrisy@flix.net>, Brendon and Wendy <brendy33@attbi.com>
Subject: Re: mp_machdep.c patch
References: <Pine.BSF.4.21.0302041135190.84482-100000@InterJet.elischer.org>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-ELNK-Trace: b1a02af9316fbb217a47c185c03b154d40683398e744b8a414520413692b38ee10906ef970a045cf350badd9bab72f9c350badd9bab72f9c350badd9bab72f9c
Sender: owner-freebsd-smp@FreeBSD.ORG
Precedence: bulk
List-ID: <freebsd-smp.FreeBSD.ORG>
List-Archive: <http://docs.freebsd.org/mail/> (Web Archive)
List-Help: <mailto:majordomo@FreeBSD.ORG?subject=help> (List Instructions)
List-Subscribe: <mailto:majordomo@FreeBSD.ORG?subject=subscribe%20freebsd-smp>
List-Unsubscribe: <mailto:majordomo@FreeBSD.ORG?subject=unsubscribe%20freebsd-smp>
X-Loop: FreeBSD.org

Julian Elischer wrote:
> > Correct, BIOS settings have no effect on whether or not
> > FreeBSD "sees" HTT CPUs.
> 
> Can we have a way of 'nobbling' certain cpus?
> e.g. don't use "#1 and #3?"

The limit is 32.  A bitmap and a lookup table of entries to bits
is enough for now, probably.

We all talked about this approach before, when we were talking
about a simple negaffinity feature in the scheduler, so that new
threads ran on different CPUs from exisitng threads; the same
principle should work for picking CPUs to start/not start, and
it would be easy to set the integer value as a tunable (i.e. in
the boot loader).

The other alternative is to allow control of whether or not the
CPUID patch probe happens or not (see other posting).  It's
probably a longer term solution; if you need something fast,
don't wait for it, unless you want to hack it yourself.

-- Terry

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-smp" in the body of the message