From owner-freebsd-current Thu Dec 4 11:11:18 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id LAA17413 for current-outgoing; Thu, 4 Dec 1997 11:11:18 -0800 (PST) (envelope-from owner-freebsd-current) Received: from Ilsa.StevesCafe.com (Ilsa.StevesCafe.com [205.168.119.129]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id LAA17391 for ; Thu, 4 Dec 1997 11:11:13 -0800 (PST) (envelope-from fbsd@Ilsa.StevesCafe.com) Received: from Ilsa.StevesCafe.com (localhost [127.0.0.1]) by Ilsa.StevesCafe.com (8.8.7/8.8.5) with ESMTP id MAA13997; Thu, 4 Dec 1997 12:10:48 -0700 (MST) Message-Id: <199712041910.MAA13997@Ilsa.StevesCafe.com> X-Mailer: exmh version 2.0gamma 1/27/96 From: Steve Passe To: Philippe Regnauld cc: dg@root.com, FreeBSD-current@FreeBSD.ORG Subject: Re: 3.0 -release ? In-reply-to: Your message of "Thu, 04 Dec 1997 12:34:19 +0100." <19971204123419.09270@deepo.prosa.dk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Thu, 04 Dec 1997 12:10:47 -0700 Sender: owner-freebsd-current@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk Hi, > My _personal_ opinion is that merging SMP into -current had one > inevitable side-effect that maybe not many people thought about: > > SMP is one _BIG_ lump to digest, and since it's not likely > to be "retro-fitted" in -stable (unless many little green men > come and help -- and maintain it after), it becomes the biggest > single step to get over to reach the next release bump. > > In a way, merging SMP has had a braking effect on release speed, > but has not slowed down development in other places. > > 3.0 will be one _hell_ of a milestone. Very well stated. My "best wild guess" about SMP is that there is at least 6-9 person-months of work (probably more) before we meet David's benchmark: >There are two things >that need to be done: locks need to be pushed down so that we have at >least per-subsystem locking (networking, filesystems, VM system, etc), >and we need to rewrite the scheduler for process affinity. I am in agreement with this target, but given the current level of work, this will occur about 2-3 years from now! We need more people actually willing to do some coding and/or testing. I have come to the conclusion (from (non)working experiments) that some fundimental changes will need to be made to the kernel, primarily tossing the spl mechanisms and using mutexs. I am trying to get a serious design discussion on SMP going in smp@freebsd.org, anyone is encouraged to join us there if they wish to participate. Although somewhat stale, there are many useful links to whitepapers on SMP in: http://www.freebsd.org/~fsmp/SMP/SMP.html -- Steve Passe | powered by smp@csn.net | Symmetric MultiProcessor FreeBSD