From owner-freebsd-alpha Sun Jul 8 9:53:17 2001
Delivered-To: freebsd-alpha@freebsd.org
Received: from wall.polstra.com (rtrwan160.accessone.com [206.213.115.74])
by hub.freebsd.org (Postfix) with ESMTP id D337037B401
for <alpha@freebsd.org>; Sun, 8 Jul 2001 09:53:13 -0700 (PDT)
(envelope-from jdp@wall.polstra.com)
Received: from vashon.polstra.com (vashon.polstra.com [206.213.73.13])
by wall.polstra.com (8.11.3/8.11.1) with ESMTP id f68GrAD90267;
Sun, 8 Jul 2001 09:53:11 -0700 (PDT)
(envelope-from jdp@wall.polstra.com)
Received: (from jdp@localhost)
by vashon.polstra.com (8.11.3/8.11.0) id f68GrAJ24108;
Sun, 8 Jul 2001 09:53:10 -0700 (PDT)
(envelope-from jdp)
Date: Sun, 8 Jul 2001 09:53:10 -0700 (PDT)
Message-Id: <200107081653.f68GrAJ24108@vashon.polstra.com>
To: alpha@freebsd.org
From: John Polstra <jdp@polstra.com>
Cc: mjacob@feral.com
Subject: Re: this a gcc bug?
In-Reply-To: <Pine.BSF.4.21.0107062130090.48822-100000@beppo>
References: <Pine.BSF.4.21.0107062130090.48822-100000@beppo>
Organization: Polstra & Co., Seattle, WA
Sender: owner-freebsd-alpha@FreeBSD.ORG
Precedence: bulk
List-ID: <freebsd-alpha.FreeBSD.ORG>
List-Archive: <http://docs.freebsd.org/mail/> (Web Archive)
List-Help: <mailto:majordomo?subject=help> (List Instructions)
List-Subscribe: <mailto:majordomo?subject=subscribe%20freebsd-alpha>
List-Unsubscribe: <mailto:majordomo?subject=unsubscribe%20freebsd-alpha>
X-Loop: FreeBSD.org
In article <Pine.BSF.4.21.0107062130090.48822-100000@beppo>,
Matthew Jacob <mjacob@feral.com> wrote:
>
> I was tracking down unaligned faults in ping...
>
> They come from an unaligned reference in that:
>
> struct sockaddr whereto;
> ...
> bzero((char *)&whereto, sizeof(struct sockaddr));
> to = (struct sockaddr_in *)&whereto;
> ...
> (void)printf("PING %s (%s)", hostname,
> inet_ntoa(to->sin_addr));
>
> ...
> pid 528 (ping): unaligned access: va=0x120017ad3 pc=0x120003850 ra=0x12000335c
> op=ldl
>
> Why?
>
> Well, in test compiling ping.c I found:
>
> yorp.feral.com > nm /tmp/ping |grep where
> 0000000120027b0f B whereto
>
> But was it the linker?
>
>
> No. Assembler output shows:
> ...
> .comm options,4,4
> .comm rcvd_tbl,128,1
> .comm whereto,16,1
> ...
>
>
> Say &what&? Here we have a structure (struct sockaddr) that has only
> byte required alignment?
>
> Structs, no matter what there contents, have been ALDOUBLE since Portable C
> compiler days- or so I thought. Maybe I'm just showing my ignorance/age. Have
> I
> gone nuts?
Well, that's for you and your shrink to decide. But you are wrong
about struct alignment. The compiler aligns a struct such that all
of its members are properly aligned. In the case of struct sockaddr,
every member is a signed or unsigned char, or an array of char. Chars
don't have to be aligned at all, so neither does struct sockaddr.
I bet that if "whereto" were declared as a struct sockaddr_in, this
problem would go away.
John
--
John Polstra jdp@polstra.com
John D. Polstra & Co., Inc. Seattle, Washington USA
"Disappointment is a good sign of basic intelligence." -- Chögyam Trungpa
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-alpha" in the body of the message