From owner-freebsd-current  Fri Apr 12 10: 8:43 2002
Delivered-To: freebsd-current@freebsd.org
Received: from espresso.q9media.com (espresso.q9media.com [216.254.138.122])
	by hub.freebsd.org (Postfix) with ESMTP
	id 6DF2337B404; Fri, 12 Apr 2002 10:08:37 -0700 (PDT)
Received: (from mike@localhost)
	by espresso.q9media.com (8.11.6/8.11.6) id g3CH1wq53477;
	Fri, 12 Apr 2002 13:01:58 -0400 (EDT)
	(envelope-from mike)
Date: Fri, 12 Apr 2002 13:01:58 -0400
From: Mike Barcroft <mike@FreeBSD.ORG>
To: "David O'Brien" <obrien@FreeBSD.ORG>
Cc: Ruslan Ermilov <ru@FreeBSD.ORG>,
	Dag-Erling Smorgrav <des@FreeBSD.ORG>, current@FreeBSD.ORG
Subject: Re: alpha tinderbox failure
Message-ID: <20020412130158.H28249@espresso.q9media.com>
References: <200204111654.g3BGsmc9069057@beast.freebsd.org> <20020412135421.GC27751@sunbay.com> <20020412121212.E28249@espresso.q9media.com> <20020412095128.A97770@dragon.nuxi.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20020412095128.A97770@dragon.nuxi.com>; from obrien@FreeBSD.ORG on Fri, Apr 12, 2002 at 09:51:28AM -0700
Organization: The FreeBSD Project
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

David O'Brien <obrien@FreeBSD.ORG> writes:
> On Fri, Apr 12, 2002 at 12:12:12PM -0400, Mike Barcroft wrote:
> > 
> > I think <machine/endian.h> would be better here.  It may not have
> > worked because of the missing <sys/cdefs.h> issue.
> 
> We will never get that header included in the GDB sources -- it is
> totally FreeBSD (NetBSD also?) specific.  It almost sounds like we have a
> bug in our headers.

Sort of.  The way it historically built was by taking advantage of the
fact that C allows multiple #define's if they are the same.  My commit
changed the definition of BYTE_ORDER to _BYTE_ORDER instead of
LITTLE_ENDIAN or BIG_ENDIAN.

The reason this didn't show up as a problem on i386 was because of a
<sys/types.h> include (which includes <machine/endian.h>) earlier in
the source file.  <machine/endian.h> (apparently this is spelled
<endian.h> on other systems) should always be included before fiddling
with the byte order manifest constants.

Your commit which included <sys/types.h> on alpha "fixed" this problem
in the same way the i386 case is "fixed".  Where "fixed" is defined as
a side-effect of fixing another problem. :)

Best regards,
Mike Barcroft

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