From owner-freebsd-bugs  Mon Feb 10 19:50:06 1997
Return-Path: <owner-bugs>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id TAA09189
          for bugs-outgoing; Mon, 10 Feb 1997 19:50:06 -0800 (PST)
Received: (from gnats@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id TAA09181;
          Mon, 10 Feb 1997 19:50:04 -0800 (PST)
Date: Mon, 10 Feb 1997 19:50:04 -0800 (PST)
Message-Id: <199702110350.TAA09181@freefall.freebsd.org>
To: freebsd-bugs
Cc: 
From: Mike Durkin <mdurkin@tsoft.net>
Subject: Re: bin/2256: PPP process on port will not close when an NT user has diconnected after a session.
Reply-To: Mike Durkin <mdurkin@tsoft.net>
Sender: owner-bugs@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

The following reply was made to PR bin/2256; it has been noted by GNATS.

From: Mike Durkin <mdurkin@tsoft.net>
To: freebsd-gnats-submit@freebsd.org, j@uriah.heep.sax.de
Cc: greg@gcsl.co.uk
Subject: Re: bin/2256: PPP process on port will not close when an NT user has diconnected after a session.
Date: Mon, 10 Feb 1997 19:50:57 -0800

 I've seen the same problem, and modems are definately configured correctly.
 I'm not sure if the bug is in mgetty or the Cyclades driver (I haven't checked
 to see if it exists with sio as well).
 
 My setup: P120, FreeBSD 2.1.6R, Cyclades Ye16 multiport card, mgetty-1.1.0
 
 The root of the problem is thus:
 
 While waiting for a call, mgetty has 'clocal' on the port.
 After the call answers, and Carrier Detect is high, mgetty switches the
    port to '-clocal'.
 When the caller disconnects, pppd does not receive a SIGHUP (so it seems),
    and keeps the port tied up.
 
 It seems that the Cyclades driver won't generate a SIGHUP on an on-to-off
 CD transition if clocal was toggled off while CD was already on/high.
 
 If mgetty is modified to set -clocal while waiting for the call, it works
 fine, but I am only using it for dialin so I don't know if that breaks
 mgetty's fax functions, etc.  A workaroung in mgetty is to change
 the TRUE flag to FALSE in the call to tio_mode_sane() in
 mg_init_device() in mg_m_init.c.  Setting -clocal on the lock-state device
 in /etc/rc.serial would probably also work instead.  Note I believe the
 bug is actually in the Cyclades driver assuming the above behavior wrt
 lack of SIGHUP under the conditions given isn't kosher.
 
 thanks, Mike