From owner-freebsd-hackers  Sun Nov 16 17:05:13 1997
Return-Path: <owner-freebsd-hackers>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id RAA28258
          for hackers-outgoing; Sun, 16 Nov 1997 17:05:13 -0800 (PST)
          (envelope-from owner-freebsd-hackers)
Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id RAA28245
          for <hackers@FreeBSD.ORG>; Sun, 16 Nov 1997 17:05:06 -0800 (PST)
          (envelope-from toor@dyson.iquest.net)
Received: (from root@localhost)
	by dyson.iquest.net (8.8.7/8.8.5) id UAA26711;
	Sun, 16 Nov 1997 20:04:55 -0500 (EST)
From: "John S. Dyson" <toor@dyson.iquest.net>
Message-Id: <199711170104.UAA26711@dyson.iquest.net>
Subject: Re: Is fork() copy-on-write ?
In-Reply-To: <Pine.BSF.3.96.971116171305.10703B-100000@darkstar.home> from Charles Mott at "Nov 16, 97 05:19:07 pm"
To: cmott@srv.net (Charles Mott)
Date: Sun, 16 Nov 1997 20:04:54 -0500 (EST)
Cc: hackers@FreeBSD.ORG
X-Mailer: ELM [version 2.4ME+ PL31 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

Charles Mott said:
> I was curious to know when a process is forked, are the new vm pages which
> are created copy-on-write?
> 
> [I know, I know I should just go look at the source code, but I haven't
> quite got up the nerve to start looking at that area.  It is sort of like
> peering over a cliff into an infinite abyss.]
> 
In the FreeBSD kernel, ALL pages that can be COW are COW -- if anyone finds the
contrary PLEASE tell me ASAP (and I'll fix it.)  One of the original
problems in the BSD 4.4VM, fixed in FreeBSD,  was that originally, during a fork,
as many as three (3) pages would exist for every page COW after a fork.  We have
reduced that to two (2) in the normal case.  Of course, that is what is expected,
but when figuring out what that code does, expect the unexpected!!! :-).

(Note that many versions of UNIX don't bother COW on modified .data actually
 performing explicit copies for .data pages, but the FreeBSD/Mach VM doesn't do
 any unnecessary copies.)

-- 
John
dyson@freebsd.org
jdyson@nc.com