From owner-freebsd-current Mon Mar 2 15:13:15 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id PAA10164 for freebsd-current-outgoing; Mon, 2 Mar 1998 15:13:15 -0800 (PST) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: from critter.freebsd.dk (critter.freebsd.dk [195.8.129.14]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id PAA10061 for ; Mon, 2 Mar 1998 15:13:03 -0800 (PST) (envelope-from phk@critter.freebsd.dk) Received: from critter.freebsd.dk (localhost [127.0.0.1]) by critter.freebsd.dk (8.8.7/8.8.5) with ESMTP id AAA13023; Tue, 3 Mar 1998 00:04:29 +0100 (CET) To: "John S. Dyson" cc: karpen@ocean.campus.luth.se (Mikael Karpberg), current@FreeBSD.ORG Subject: Re: 3.0-RELEASE? In-reply-to: Your message of "Mon, 02 Mar 1998 17:50:17 EST." <199803022250.RAA14700@dyson.iquest.net> Date: Tue, 03 Mar 1998 00:04:28 +0100 Message-ID: <13021.888879868@critter.freebsd.dk> From: Poul-Henning Kamp Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG In message <199803022250.RAA14700@dyson.iquest.net>, "John S. Dyson" writes: >To make the kernel have fine-grained locking, it will be approx 6-12 man >mos. These changes need to be made by people who are very knowledgeable abou >t >the entire kernel. These people are in great demand, and that would push 3.0 > out >into 1999. I doubt even that would be possible, unless somebody actually goes and offers a salary to about 5 or 6 of us heavy-duty people to do just this. I must say, that I belong to the school where we think that it may not actually pay off to go that far. I'm pretty sure that we would get a long way with a much more narrow scope of locks: * One lock for each device driver. To avoid rewriting the existing drivers, this could be done entirely in the generic code, with the few obvious exceptions of pty and other incestous pseudo-drivers. * One lock for the entire network stack * One lock per mountpoint * One lock for "the rest" (the smoking remains of our current giant lock. But since this is a pretty likely step on the way, it would be easy to determine if further performance is derived from going further, so I'm not standing on my soapbox arguing about this. >Mikael Karpberg said: >> Oh well... I'm rambling as usual. Any comments? You're rambling some perfectly valid points as usual :-) We're at the point where it would really be nice to be able to go tell the CEO "off to the bank you go, borrow some money for R&D, you may get them back in increased revenue later. We'll need $1M over the year. Small unmarked used notes would be best." Sponsors most welcome. -- Poul-Henning Kamp FreeBSD coreteam member phk@FreeBSD.ORG "Real hackers run -current on their laptop." "Drink MONO-tonic, it goes down but it will NEVER come back up!" To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message