From owner-cvs-sys  Fri Dec  6 23:07:46 1996
Return-Path: <owner-cvs-sys>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id XAA14837
          for cvs-sys-outgoing; Fri, 6 Dec 1996 23:07:46 -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 XAA14832;
          Fri, 6 Dec 1996 23:07:27 -0800 (PST)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.3/8.6.9) id RAA18540; Sat, 7 Dec 1996 17:58:30 +1100
Date: Sat, 7 Dec 1996 17:58:30 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199612070658.RAA18540@godzilla.zeta.org.au>
To: dg@root.com, peter@spinner.dialix.com
Subject: Re: cvs commit: src/sys/i386/include endian.h
Cc: bde@zeta.org.au, cvs-all@freefall.freebsd.org,
        CVS-committers@freefall.freebsd.org, cvs-sys@freefall.freebsd.org,
        dyson@freebsd.org, dyson@freefall.freebsd.org
Sender: owner-cvs-sys@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>>Perhaps we should seriously consider using 'cc -include opt_global.h' and
>>moving things like Ix86_CPU etc into there?  It has the benefits of having
>>every file know about the settings, but without having to modify every file.
>
>   I'd rather see a #include added to cpufunc.h, which is included by systm.h,
>which is included by everything (?) in the kernel.

cpufunc.h shouldn't be included in systm.h.  This has been on my list of
things to fix since FreeBSD-1.1.  It is fixed in NetBSD.  Only the ffs()
and fls() functions in our version are used in machine-independent code
(except possibly in spls), and these functions probably shouldn't be
written in asm anyway (on Pentiums the asm versions are about the same
speed as naive C versions if everything is in the L1 cache).

systm.h is included for 501 out of 521 objects in LINT in -current.  It
is actually used for only 442 of these objects.

Bruce