From owner-freebsd-hackers  Tue Jan 21 20:31:25 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id UAA11709
          for hackers-outgoing; Tue, 21 Jan 1997 20:31:25 -0800 (PST)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id UAA11698
          for <freebsd-hackers@FreeBSD.org>; Tue, 21 Jan 1997 20:31:16 -0800 (PST)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.3/8.6.9) id PAA01813; Wed, 22 Jan 1997 15:29:20 +1100
Date: Wed, 22 Jan 1997 15:29:20 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199701220429.PAA01813@godzilla.zeta.org.au>
To: freebsd-hackers@FreeBSD.org, jdp@polstra.com
Subject: Re: Commerical applications (was: Development and validation
Sender: owner-hackers@FreeBSD.org
X-Loop: FreeBSD.org
Precedence: bulk

>I realized later that this is not true.  ELF double-maps the page
>containing the boundary between text and data.  That wastes a page
>(4K), making ELF's memory usage in execution the same on average
>as that of a.out.

Isn't this just wasteful on (flat model) i386's?  It provides no
protection against writing the text in the boundary page via the data
mapping, and isn't necessary for execution because the i386 doesn't have
an execution bit in its page tables.  Protection can only be provided
by starting the data segment at the boundary, but this would break many
(broken) programs that assume that text pointers are interchangeable
with data pointers, and doesn't require double mapping anyway.

Bruce