From owner-freebsd-current  Fri Dec 19 10:02:09 1997
Return-Path: <owner-freebsd-current>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id KAA03978
          for current-outgoing; Fri, 19 Dec 1997 10:02:09 -0800 (PST)
          (envelope-from owner-freebsd-current)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id KAA03967
          for <current@freebsd.org>; Fri, 19 Dec 1997 10:01:57 -0800 (PST)
          (envelope-from bde@zeta.org.au)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.7/8.6.9) id EAA21467; Sat, 20 Dec 1997 04:52:10 +1100
Date: Sat, 20 Dec 1997 04:52:10 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199712191752.EAA21467@godzilla.zeta.org.au>
To: bde@zeta.org.au, hasty@rah.star-gate.com
Subject: Re: cvs commit: src/sys/i386/linux linux.h linux_ioctl.c
Cc: current@freebsd.org, gjp@erols.com, gjp@erols.net, pb@fasterix.freenix.org
Sender: owner-freebsd-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>I know what ioperm does and I first implemented it for 386bsd 0.0 8)

That was more like Linux iopl than Linux ioperm.  Linux iopl is already
implemented (incorrectly) in the Linux emulator.  In Linux, it sets
the iopl bitfield to the specified value (0-3). In the Linux emulator,
it always sets the iopl bitfield to 3.

>Not sure if it is worth any hazzle to implement it the way that
>that svgalib uses it . The problem comes in when they want to

There is no choice.  Anything else would not be Linux compatible.

>do any kind of video acceleration how to then enable all those
>i/o registers for something like an S3 chipset.

Svgalib presumably uses iopl for the hard cases.  Linux has a restriction
on the bitmap size, at least in old versions, so it may be impossible to
access all the registers using ioperm.  FreeBSD VM86-ioperm doesn't have
this restriction.

>At any rate, there may be other valid uses for ioperm other than
>for svgalib.

Linux ioperm gives some chance of secure access to device registers.
It needs at least a big immutable database of which registers are
secure.  Where are all those i/o registers in your graphics card today?

Bruce