From owner-freebsd-smp  Thu Jul  8 10:36: 6 1999
Delivered-To: freebsd-smp@freebsd.org
Received: from cs.rice.edu (cs.rice.edu [128.42.1.30])
	by hub.freebsd.org (Postfix) with ESMTP id 1B54A14E5F
	for <freebsd-smp@freebsd.org>; Thu,  8 Jul 1999 10:35:50 -0700 (PDT)
	(envelope-from alc@cs.rice.edu)
Received: (from alc@localhost)
	by cs.rice.edu (8.9.0/8.9.0) id MAA11200;
	Thu, 8 Jul 1999 12:35:47 -0500 (CDT)
Date: Thu, 8 Jul 1999 12:35:46 -0500
From: Alan Cox <alc@cs.rice.edu>
To: "Robert S. Sciuk" <rob@controlq.com>
Cc: Cosmic 665 <the_hermit665@hotmail.com>, freebsd-smp@freebsd.org
Subject: Re: SMP comparisons
Message-ID: <19990708123546.H10611@cs.rice.edu>
References: <19990708162724.16604.qmail@hotmail.com> <Pine.UW2.3.96.990708123245.26753A-100000@fatlady.controlq.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 0.95.5us
In-Reply-To: <Pine.UW2.3.96.990708123245.26753A-100000@fatlady.controlq.com>; from Robert S. Sciuk on Thu, Jul 08, 1999 at 12:34:57PM -0400
Sender: owner-freebsd-smp@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.org

On Thu, Jul 08, 1999 at 12:34:57PM -0400, Robert S. Sciuk wrote:
> I think we're not where we should be ... what with the GreatBigLock in the
> Kernel ... however, I've seen some -smp traffic which leads me to believe
> some very talented individuals are working to rectify this situation ...
> then FreeBSD should kick some proverbial butt 8-).
> 

Here's what you'll see shortly:

1. Bruce Evans is about to commit some changes to the interrupt management
code that removes one impediment to moving (or removing) the giant lock.

2. Luoqi Chen is working on the next step.  He's moving some
of the interrupt management variables from shared memory to
per processor memory.

Once these pieces are in place, the *body* of many simple system calls can
be executed without the giant lock.  Returning from the system call
to user level will still, however, require the giant lock.  Tackling
that problem and making some further changes to the interrupt management
code will probably be the next steps, but in the meantime people will be
able to experiment with multithreading various system calls.

Alan


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