From owner-freebsd-hackers  Tue Feb 25 19:13:15 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id TAA09280
          for hackers-outgoing; Tue, 25 Feb 1997 19:13:15 -0800 (PST)
Received: from bunyip.cc.uq.edu.au (daemon@bunyip.cc.uq.edu.au [130.102.2.1])
          by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA09271
          for <freebsd-hackers@freebsd.org>; Tue, 25 Feb 1997 19:12:58 -0800 (PST)
Received: (from daemon@localhost)
	by bunyip.cc.uq.edu.au (8.8.5/8.8.5) id NAA19728;
	Wed, 26 Feb 1997 13:10:46 +1000
Received: by ogre.devetir.qld.gov.au (8.7.5/DEVETIR-E0.3a)
	id NAA24882; Wed, 26 Feb 1997 13:11:14 +1000 (EST)
Date: Wed, 26 Feb 1997 13:11:14 +1000 (EST)
From: Stephen McKay <syssgm@devetir.qld.gov.au>
Message-Id: <199702260311.NAA24882@ogre.devetir.qld.gov.au>
To: j@uriah.heep.sax.de (J Wunsch), Naoki Hamada <nao@sbl.cl.nec.co.jp>
cc: freebsd-hackers@freebsd.org, syssgm@devetir.qld.gov.au
Subject: Re: MALLOC(3) writes to stderr...
X-Newsreader: NN version 6.5.0 #1 (NOV)
Sender: owner-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

j@uriah.heep.sax.de (J Wunsch) wrote:

>As Naoki Hamada wrote:
>
>> Malloc(3) of 2.2-GAMMA writes error/warning messages to stderr. There
>> is a daemon named jserver, which provides translations between
>> Japanese characters, which does close stderr before opening its
>> dictionary files. Sometimes jserver gets its dictionary file
>> completely damaged by malloc's error/warning message.
>
>Hmm, i think it's a common misconception in Unix that everybody
>blindly assumes that file descriptors 0, 1, and 2 refer to stdin,
>stdout, and stderr respectively.  OTOH, there's no opportunity for a
>library module to check where stderr really goes (or whether it has
>been closed).  I bet malloc isn't the only library module that would
>write there.
>
>Your safest bet is to never close these descriptors, but redirect
>/dev/null on them.  So, they remain `reserved'.

Although not enforced by the kernel, file descriptors 0, 1 and 2 have a
conventional use that should always be respected.  If you don't want to
see stderr output, open /dev/null as descriptor 2.

This should work:

    fd = open("/dev/null", O_WRONLY);
    if (fd == -1)
	{
	perror("/dev/null");
	exit(1);
	}
    if (fd != 2)
	{
	dup2(fd, 2);
	close(fd);
	}

The test of fd != 2 is for the unlikely case that descriptor 2 is closed
already.  If the open of "/dev/null" fails, exiting is probably the kindest
result; you've got some repair work to do.

Some others responded that stderr and descriptor 2 are logically different
and recommend using fileno(stderr) instead of 2.  We of the old school all
know that stdio is an upstart young library, and will never be allowed to
use anything but descriptor 2 on Unix. :-)

Feel free to use fileno(stderr) if the hairy revolutionaries have got to you.
You might like to use STDERR_FILENO from <unistd.h> and/or _PATH_DEVNULL
from <paths.h> and/or err() instead of perror()/exit(), but these are upstart
newbies too.

BTW, the dup2() man page implies that newd will be closed even if it is
equal to oldd.  In reality, it is not, and (as far as I can remember) never
was for any version of Unix.  Similarly, newd is not closed if oldd is
invalid.  I'll update the man page as soon as I get a chance.

Stephen.