From owner-freebsd-current  Sun Oct 20 16:29:16 2002
Delivered-To: freebsd-current@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP
	id BEABF37B401; Sun, 20 Oct 2002 16:29:14 -0700 (PDT)
Received: from mailman.zeta.org.au (mailman.zeta.org.au [203.26.10.16])
	by mx1.FreeBSD.org (Postfix) with ESMTP
	id 5D89543E88; Sun, 20 Oct 2002 16:29:13 -0700 (PDT)
	(envelope-from bde@zeta.org.au)
Received: from bde.zeta.org.au (bde.zeta.org.au [203.2.228.102])
	by mailman.zeta.org.au (8.9.3/8.8.7) with ESMTP id JAA25178;
	Mon, 21 Oct 2002 09:28:48 +1000
Date: Mon, 21 Oct 2002 09:39:40 +1000 (EST)
From: Bruce Evans <bde@zeta.org.au>
X-X-Sender: bde@gamplex.bde.org
To: Peter Wemm <peter@wemm.org>
Cc: Mike Barcroft <mike@FreeBSD.ORG>,
	Kris Kennaway <kris@obsecurity.org>, <current@FreeBSD.ORG>
Subject: Re: Conflicting declarations for ffs() 
In-Reply-To: <20021020195557.972FD2A88D@canning.wemm.org>
Message-ID: <20021021093120.I8562-100000@gamplex.bde.org>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-freebsd-current@FreeBSD.ORG
Precedence: bulk
List-ID: <freebsd-current.FreeBSD.ORG>
List-Archive: <http://docs.freebsd.org/mail/> (Web Archive)
List-Help: <mailto:majordomo@FreeBSD.ORG?subject=help> (List Instructions)
List-Subscribe: <mailto:majordomo@FreeBSD.ORG?subject=subscribe%20freebsd-current>
List-Unsubscribe: <mailto:majordomo@FreeBSD.ORG?subject=unsubscribe%20freebsd-current>
X-Loop: FreeBSD.ORG

On Sun, 20 Oct 2002, Peter Wemm wrote:

> Mike Barcroft wrote:
>
> > Kris Kennaway <kris@obsecurity.org> writes:
> > > Take a look at:
> > >
> > > http://bento.freebsd.org/errorlogs/5-full/cqcam-0.91_1.log
> > >
> > > This port includes headers that declare the ffs() function twice: once
> > > with an inline version and once with a prototype.
> > >
> > > Is the bug in the application, or the headers?
> >
> > It looks like a bug in our headers.  I don't see why this is a new bug
> > though.  It looks like <string.h> (which <strings.h> used to include)
> > and i386's <machine/cpufunc.h> have been defining conflicting ffs()
> > prototypes since at least 1999.
>
> machine/cpufunc.h is really meant to be a kernel header and isn't really
> meant for consumption by userland.  However, it is sortof useful.

Right, it is a bug in the application.

> In this particular case, gcc provides a respectable builtin for ffs().  We
> probably shouldn't be overriding it for userland with our own inline for
> userland.

The gcc inline may now be OK in the kernel too.  However, we use -fno-inline
in the kernel, so we don't even get inlining for memcpy().

> Maybe it would just be better to stick a #ifdef _KERNEL around the inline
> version?  That should leave gcc to use the builtin and fall back to the
> libc version if -fno-builtin is used [or if non-gcc is used].

I prefer:

    #ifndef _KERNEL
    #error "no user-serviceable parts inside"
    #endif

near the beginning of cpufunc.h.  I have added this to several other
kernel-only headers.

BSD/OS arranges kernel-only MD headers better by not putting them in
<machine>.  This also inhibits them being abused in MI code.

Bruce


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message