From owner-freebsd-hackers  Fri Mar 10 16:28:11 2000
Delivered-To: freebsd-hackers@freebsd.org
Received: from voyager.fisicc-ufm.edu (ip-198-202.guate.net [209.198.197.202])
	by hub.freebsd.org (Postfix) with ESMTP
	id 7B1FF37BB32; Fri, 10 Mar 2000 16:27:50 -0800 (PST)
	(envelope-from obonilla@voyager.fisicc-ufm.edu)
Received: (from obonilla@localhost)
	by voyager.fisicc-ufm.edu (8.9.3/8.9.3) id SAA94301;
	Fri, 10 Mar 2000 18:26:06 -0600 (CST)
	(envelope-from obonilla)
Date: Fri, 10 Mar 2000 18:26:02 -0600
From: Oscar Bonilla <obonilla@fisicc-ufm.edu>
To: Chuck Robey <chuckr@picnic.mat.net>
Cc: Oscar Bonilla <obonilla@fisicc-ufm.edu>,
	Alfred Perlstein <bright@wintelcom.net>,
	Kris Kennaway <kris@hub.freebsd.org>, freebsd-hackers@FreeBSD.ORG
Subject: Re: inner workings of the C compiler
Message-ID: <20000310182602.A94174@fisicc-ufm.edu>
References: <20000310172706.A93660@fisicc-ufm.edu> <Pine.BSF.4.21.0003101849420.32633-100000@picnic.mat.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 1.0pre1i
In-Reply-To: <Pine.BSF.4.21.0003101849420.32633-100000@picnic.mat.net>
Sender: owner-freebsd-hackers@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

On Fri, Mar 10, 2000 at 06:51:20PM -0500, Chuck Robey wrote:
> I wasn't reading this too closely, but if you're trying to hand feed in
> the object files, the C startup object file *MUST* come first in the list
> of object files, because it's gotta link at the lowest address ...
> 
> Is that it?

Ok, I'm even more puzzled than before...

If I link them in this order:  nss-test.o crt1.o crti.o

I get the following:

files called
files called
retval = 1
NS_SUCCESS
Bus error (core dumped)

Notice the duplicate line that says "files called"? This had me puzzled
before, but now I see that somehow _init point to the first function
in the first object file

(gdb) p _init 
$1 = {<text variable, no debug info>} 0x8048074 <files> 

Now if I like it in this order: crt1.o crti.o nss-test.o

I get an infinite recursion!!!!

(gdb) p _init 
$1 = {<text variable, no debug info>} 0x8048074 <_start> 

since _start calls _init() *before* calling main() it just loops
(and starts swapping like hell)

Let me try this order: crti.o crt1.o nss-test.o

(gdb) p _init 
$1 = {<text variable, no debug info>} 0x8048074 <_start> 

nope, same as before. infinite recursion and swapping like hell.

maybe this order: crt1.o nss-test.o crti.o

(gdb) print _init 
$1 = {<text variable, no debug info>} 0x8048074 <_start> 

same thing.

I created a small C file:

#include <stdio.h>
foo() {
	printf("foo\n");
}

and tried this order: foo.o nss-test.o crt1.o crti.o

(gdb) print _init 
$1 = {<text variable, no debug info>} 0x8048074 <foo> 

and if I just run it...

foo
files called
retval = 1
NS_SUCCESS
Bus error (core dumped)

If I try gdb with a normal program (i.e. one linked with the standard
C library in /usr/lib and not my own C library)

(gdb) print _init
$1 = {<text variable, no debug info>} 0x804838c <_init>

What could be the problem?

regards,

-oscar

-- 
pgp public key: finger obonilla@fisicc-ufm.edu
pgp fingerprint: 6D 18 8C 90 4C DF F0 4B  DF 35 1F 69 A1 33 C7 BC


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