From owner-freebsd-current  Sun Oct 27 14:51:00 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id OAA22119
          for current-outgoing; Sun, 27 Oct 1996 14:51:00 -0800 (PST)
Received: from whistle.com (s205m131.whistle.com [207.76.205.131])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id OAA22112
          for <current@freebsd.org>; Sun, 27 Oct 1996 14:50:56 -0800 (PST)
Received: (from smap@localhost) by whistle.com (8.7.5/8.6.12) id OAA15839; Sun, 27 Oct 1996 14:50:25 -0800 (PST)
Received: from bubba.whistle.com(207.76.205.7) by whistle.com via smap (V1.3)
	id sma015837; Sun Oct 27 14:50:22 1996
Received: (from archie@localhost) by bubba.whistle.com (8.7.5/8.6.12) id OAA14535; Sun, 27 Oct 1996 14:50:22 -0800 (PST)
From: Archie Cobbs <archie@whistle.com>
Message-Id: <199610272250.OAA14535@bubba.whistle.com>
Subject: Strange bug related to fstat()
To: current@freebsd.org
Date: Sun, 27 Oct 1996 14:50:22 -0800 (PST)
X-Mailer: ELM [version 2.4ME+ PL25 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-current@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk


I'm trying to track down a bug that's completely baffling to me...
any help would be greatly appreciated. This is in a pretty -current
environment.

Within a little background daemon, the system() command is used
to execute a command looking something like this:

  system("tar cf - file1 file2 | gzip -cf | uuencode files > output");

What's happening is that gzip is failing, and the place it fails
is where it tries to fstat() standard input:

  if (fstat(fileno(stdin), &istat) != 0) {
      error("fstat(stdin)");
  }

The error it's getting is "Bad file descriptor".

Of course, if there's any way the daemon could be causing this
to happen I'm sure it is! But I can't figure out how it could.

Can this error possibly be correct, unless there is a bug in
gzip itself? Even if the "tar" command failed, standard input
should still exist until closed, no?

The only other possibilities seem to be a bug in /bin/sh or the
kernel.  As you can see, I need a little hint as to how to pursue
this one...

Thanks for any help.

-Archie

___________________________________________________________________________
Archie Cobbs   *   Whistle Communications, Inc.  *   http://www.whistle.com