From owner-freebsd-current  Thu Sep 26 10:01:14 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id KAA05272
          for current-outgoing; Thu, 26 Sep 1996 10:01:14 -0700 (PDT)
Received: from spinner.DIALix.COM (root@spinner.DIALix.COM [192.203.228.67])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id KAA05222
          for <freebsd-current@freebsd.org>; Thu, 26 Sep 1996 10:01:08 -0700 (PDT)
Received: from spinner.DIALix.COM (peter@localhost.DIALix.oz.au [127.0.0.1])
          by spinner.DIALix.COM (8.7.6/8.7.3) with ESMTP id BAA03640;
          Fri, 27 Sep 1996 01:00:54 +0800 (WST)
Message-Id: <199609261700.BAA03640@spinner.DIALix.COM>
X-Mailer: exmh version 1.6.7 5/3/96
To: Garrett Wollman <wollman@lcs.mit.edu>
cc: freebsd-current@freebsd.org
Subject: Re: BLOAT in minimal programs 
In-reply-to: Your message of "Thu, 26 Sep 1996 11:48:33 -0400."
             <9609261548.AA30202@halloran-eldar.lcs.mit.edu> 
Date: Fri, 27 Sep 1996 01:00:54 +0800
From: Peter Wemm <peter@spinner.DIALix.COM>
Sender: owner-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

Garrett Wollman wrote:
> <<On 26 Sep 1996 15:33:00 GMT, peter@spinner.DIALix.COM (Peter Wemm) said:
> 
> > Yes, we do have a special version of crt0.o, it's /usr/lib/scrt.o and gcc
> > correctly links with it:
> 
> > /usr/lib/libgcc.a(__main.o) needed due to ___main
> 
> We really should fix our crt0 and libc to do things right and not need
> __main or the functions it calls.
> 
> -GAWollman

Well, having crt0.c calling __do_global_ctors before main() and having 
exit.c in libc calling __do_global_dtors() seems a bit gcc-centric to me, 
but I guess there's not a lot of alternative if we want to keep the 
-static overheads minimal.

Sigh, and the ELF file format solves it so nicely (or any other file formt 
with multiple arbitary sections) by contatenating a call lists and 
crt0.o's task simply becomes "call .init" and for exit(), it's "call 
.fini" (or is that __init and __fini? I never can remember).  All the work 
is done once-off at link time and there's only any constructor/destructor 
glue *IF* there are any C++ object's present somewhere.  It simply is left 
out otherwise.  It's not quite like linker sets, the result is an 
executable chunk of code.

Cheers,
-Peter