From owner-freebsd-bugs  Wed Aug 27 11:40:15 1997
Return-Path: <owner-freebsd-bugs>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.7/8.8.7) id LAA26785
          for bugs-outgoing; Wed, 27 Aug 1997 11:40:15 -0700 (PDT)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id LAA26779
          for <bugs@freebsd.org>; Wed, 27 Aug 1997 11:40:13 -0700 (PDT)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.5/8.6.9) id EAA03347 for bugs@freebsd.org; Thu, 28 Aug 1997 04:36:53 +1000
Date: Thu, 28 Aug 1997 04:36:53 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199708271836.EAA03347@godzilla.zeta.org.au>
To: bugs@freebsd.org
Subject: rlogin misses child death
Sender: owner-freebsd-bugs@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

Why does rlogin ignore SIGCHLD when it is stopped?  Doing so causes it
to miss the death of the child.  To see the problem, rlogin somewhere,
suspend rlogin, then kill the remote shell.  Then de-suspend rlogin.
Sometimes (more often than not here) the child starts up first.  Then
the parent doesn't exit until it attempts to write something and gets
a SIGPIPE.

telnet notices that the connection has gone away when it is de-suspended
after being suspended using ^]z, but not when it has been suspended using
^]^Z.

Bruce

diff -c2 rlogin.c~ rlogin.c
*** rlogin.c~	Wed Aug  6 09:51:11 1997
--- rlogin.c	Thu Aug 28 04:15:37 1997
***************
*** 585,591 ****
  {
  	mode(0);
- 	(void)signal(SIGCHLD, SIG_IGN);
  	(void)kill(cmdc == defltc.t_suspc ? 0 : getpid(), SIGTSTP);
- 	(void)signal(SIGCHLD, catch_child);
  	mode(1);
  	sigwinch(0);			/* check for size changes */
--- 585,589 ----