From owner-freebsd-current  Thu Jan  2 05:24:43 1997
Return-Path: <owner-current>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id FAA03675
          for current-outgoing; Thu, 2 Jan 1997 05:24:43 -0800 (PST)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.8.4/8.8.4) with ESMTP id FAA03667
          for <current@freebsd.org>; Thu, 2 Jan 1997 05:24:37 -0800 (PST)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.3/8.6.9) id AAA05486; Fri, 3 Jan 1997 00:16:18 +1100
Date: Fri, 3 Jan 1997 00:16:18 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199701021316.AAA05486@godzilla.zeta.org.au>
To: phk@critter.dk.tfs.com, wollman@lcs.mit.edu
Subject: Re: <sys/kernel.h> and MAKE_SET
Cc: current@freebsd.org
Sender: owner-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>>>>> Isn't it wrong to have MAKE_SET() in <sys/kernel.h> ?  What if you want
>>>>> to use it from userland ?

Probably.  It not very conveniently placed for use in the kernel.  In
userland, you declare the magic parts of MAKE_SET() in a user header.

>>I would just as soon see linker sets not used in user-land code, so as
>>to not diminish its value to other people who are attempting to do
>>things with it that don't involve using our linker/build environment.

I agree.

>>One of the nice things about most of the user-land code as it stands
>>is that I can take most of the utilities together with some useful
>>glue from the C library and have things work on other platforms.

Except for cruft like __P.

>I don't care much about this, but it prevents me from testing my
>kernel code in a userland harness :-(
>
>I'm going to stick in this patch:
>
>Index: kernel.h
> ...
>+#ifdef KERNEL

kernel.h was supposed to be kernel-only.  In the CSRG version it has
only the kernel variables that you ifdefed (less a few).

Bruce