From owner-freebsd-hackers  Fri Oct  4 16:11:59 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id QAA29463
          for hackers-outgoing; Fri, 4 Oct 1996 16:11:59 -0700 (PDT)
Received: from alpo.whistle.com (s204m38.isp.whistle.com [207.76.204.38])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id QAA29453
          for <hackers@freebsd.org>; Fri, 4 Oct 1996 16:11:56 -0700 (PDT)
Received: from current1.whistle.com (current1.whistle.com [207.76.205.22]) by alpo.whistle.com (8.7.5/8.7.3) with SMTP id QAA24520 for <hackers@freebsd.org>; Fri, 4 Oct 1996 16:10:18 -0700 (PDT)
Message-ID: <325598E4.167EB0E7@whistle.com>
Date: Fri, 04 Oct 1996 16:08:20 -0700
From: Julian Elischer <julian@whistle.com>
Organization: Whistle Communications
X-Mailer: Mozilla 3.0b6 (X11; I; FreeBSD 2.2-CURRENT i386)
MIME-Version: 1.0
To: hackers@freebsd.org
Subject: Running FreeBSD with NO CONSOLE (AT ALL)
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: owner-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

I'm running freeBSD with no available sio device and no videa
cards.
(there IS an sio device but it's attached to the modem)

this device is htrefore running with no console device.
By this I mean that all calls to such routines as cnwrite
cnopen etc.
do not call any hardware, but return immediatly.
because the cn_tab == NULL

the only problem I've seen so far is that in exit()
there is an unguarded call to p->pg_session->s_ttyp
which is NULL.
there is a check for p->pg_session->s_ttyvp and the assumption is
being made that if s_ttyvp (points to /dev/console's vnode) is valid,
then the tty struct in the physical device is also valid.

this is not so because the TIOSCTTY ioctl sets them
in different places, and one refers to teh device /dev/console,
whisch always exists, and the other refers to the tty struct for the
underlying physical device.

I have hacked the exit code to also check s_ttyp
which gets arount that problem, but it leaves open the question
"What IS the right answer"?
should I
1/ make a dummy console device that supplies a tty struct
2/ make the fix I've made
3/ Other suggestions?

Eventually the aim is to be able to run with no console, but in 
some special case, attach the console to sio0
but only with some special checks.

julian