From owner-freebsd-current  Fri Sep 27 03:54:34 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id DAA28766
          for current-outgoing; Fri, 27 Sep 1996 03:54:34 -0700 (PDT)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id DAA28719
          for <current@freebsd.org>; Fri, 27 Sep 1996 03:54:27 -0700 (PDT)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.7.6/8.6.9) id UAA20273; Fri, 27 Sep 1996 20:49:19 +1000
Date: Fri, 27 Sep 1996 20:49:19 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199609271049.UAA20273@godzilla.zeta.org.au>
To: bde@zeta.org.au, phk@critter.tfs.com
Subject: Re: BLOAT in minimal programs
Cc: current@freebsd.org, julian@whistle.com
Sender: owner-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>>Even better: don't use malloc():
>>
>>	main() {} atexit() {}
>>
>>atexit() is normally linked to and references malloc(), although it
>>only calls malloc() when there are a lot of atexits.  malloc() now
>>uses stdio and stdio uses half the universe.
>
>Malloc doesn't use stdio if you don't define MALLOC_STATS in it...
>
>I'm not to happy about that being the default, it doesn't matter for
>shlib cases, but for static programs it's a pain.

atexit() could use sbrk() instead of malloc().  This would work fairly
well since atexit() rarely needs to expands its default of 32 pointers
(even 32 is wasteful) and never frees its memory.

Bruce