From owner-cvs-sys  Wed Dec 17 06:47:10 1997
Return-Path: <owner-cvs-sys>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id GAA15292
          for cvs-sys-outgoing; Wed, 17 Dec 1997 06:47:10 -0800 (PST)
          (envelope-from owner-cvs-sys)
Received: from mutara.noc.erols.net (gjp@mutara.noc.erols.net [207.172.25.12])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id GAA15236;
          Wed, 17 Dec 1997 06:46:14 -0800 (PST)
          (envelope-from gjp@mutara.noc.erols.net)
Received: (from gjp@localhost)
	by mutara.noc.erols.net (8.8.8/8.8.7) id JAA12759;
	Wed, 17 Dec 1997 09:46:12 -0500 (EST)
Date: Wed, 17 Dec 1997 09:46:12 -0500 (EST)
Message-Id: <199712171446.JAA12759@mutara.noc.erols.net>
Mime-Version: 1.0
X-Newsreader: knews 0.9.8
References: <20185.882182566@time.cdrom.com>
    <199712151301.XAA00302@word.smith.net.au>
From: gjp@erols.net (Gary Palmer)
Subject: Re: cvs commit: src/sys/i386/linux linux.h linux_ioctl.c
X-Original-Newsgroups: lists.freebsd.cvs.all,lists.freebsd.cvs.sys
To: cvs-all@FreeBSD.ORG, cvs-sys@FreeBSD.ORG
Content-Type: text/plain; charset=us-ascii
Sender: owner-cvs-sys@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

In article <199712151301.XAA00302@word.smith.net.au>,
	mike@smith.net.au (Mike Smith) writes:
>> How close does this bring us to being able to run SVGAlib binaries
>> in general?  Thanks also to Amancio for doign this!
> 
> Amancio has a modified SVGAlib which when built linux-mode allows Linux 
> SVGALlib-using-binaries to run on FreeBSD.  It's not unreasonable to 
> suggest that these changes might make it into the "mainstream" SVGAlib 
> at some stage; until then we can just include it as part of the 
> linux_lib port I guess.

There are 3 remaining problems:

- Linux's ioperm() isn't implimented, but I can fix that

- normal svgalib checks the major number of the device to see if it is
  a known console, there is a hack in linux_newfstat that can fix that.

- quake has a rather unusual ``bug'' in dealing with the console file
  descriptors. It opens /dev/console twice (from memory). It then
  proceeds to set one non-blocking, and then promply reads from the
  *other*, expecting it to be non-blocking *also*. This, of course,
  *fails* (surprise surprise). An oddity to note is that this works
  on NetBSD as Linux expects.

I've gotten SVGALib Quake to run in the past by a *gross* hack, which
re-mapped the FD that the ioctl was done on so that it did the
non-blocking options on the one it actually reads from. This is,
of course, not the Correct Solution (TM). I am not sure what is

Gary