From owner-freebsd-questions  Sat Oct 14 04:52:28 1995
Return-Path: owner-questions
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id EAA05539
          for questions-outgoing; Sat, 14 Oct 1995 04:52:28 -0700
Received: from late5.e-technik.uni-erlangen.de (late5.e-technik.uni-erlangen.de [131.188.254.132])
          by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id EAA05529
          for <questions@freebsd.org>; Sat, 14 Oct 1995 04:52:23 -0700
Received: (from tuucp@localhost) by late5.e-technik.uni-erlangen.de (8.6.10/8.6.6) id HAA21406 for questions@freebsd.org; Sat, 14 Oct 1995 07:52:04 -0400
Received: from odin (iwte01-2.dialin.rrze.uni-erlangen.de [192.44.83.18]) by iwte01.dialin.rrze.uni-erlangen.de (8.6.11/8.6.6) with ESMTP id MAA03399; Sat, 14 Oct 1995 12:52:13 +0100
Received: (from eilts@localhost) by odin (8.6.8/8.6.6) id MAA00431; Sat, 14 Oct 1995 12:52:11 +0100
Date: Sat, 14 Oct 1995 12:52:11 +0100
From: Hinrich Eilts <eilts@iwte01.dialin.rrze.uni-erlangen.de>
Message-Id: <199510141152.MAA00431@odin>
Received: by NeXT.Mailer (1.100)
Received: by NeXT Mailer (1.100)
To: questions@freebsd.org
Subject: Re: hylafax unable to spawn getty
Cc: brian@MediaCity.com (Brian Litzinger)
Sender: owner-questions@freebsd.org
Precedence: bulk

> I'm running hylafax on a 2.1 SNAP 951005 with a generic
> ROCKWELL modem and it mostly works.  However, when
> I make a data call into the system, the log shows
> that it spawns getty on the port, put no login
> ever shows up.  I can see (via ps)  getty on the right port
> just sitting there.  It sits there until I kill the getty
> at which point I get hung up upon and the hylafax system
> takes over again.

If a data-call happens, a Rockwell based modems (ELSA Microlink 28.8 TQV in my case)
picks up line, recognize the data-call and sends +FDM while it negotiate speed,
compression etc with remote modem. Because HylaFax spawns getty on receiving +FDM,
getty got /dev/cuaa* before the connection is enabled (modem is still negotating).
To solve this problem, getty should wait for the CONNECT message and not for +FDM,
there is an option in HylaFax conf named ModemWaitForConnect for this. In my case,
because this option does not work (for me), I installed a little hacked version of
getty named fgetty waiting for CONNECT for HylaFax (only HylaFax uses it, of course):

  fd_set readset;
  struct timeval connectwait;
  [...]

  if (argc <= 2 || strcmp(argv[2], "-") == 0) {
    strcpy(ttyn, ttyname(0));
    FD_ZERO(&readset);
    FD_SET(STDIN_FILENO, &readset);
    connectwait.tv_sec = MAXCONNECTWAIT;
    connectwait.tv_usec = 0;
    n = select(STDIN_FILENO +1,&readset,NULL,NULL,&connectwait);
    if (n == 0) {
      syslog(LOG_ERR, "%s: timeout on waiting for CONNECT", ttyn);
      exit(1);
    }
    else if (n < 0) {
      syslog(LOG_ERR, "%s: select: %m");
      exit(1);
    }
  }
  [...]

You must tell the fgetty at config-time to HylaFax.
Because I used an old getty for this, I cannot provide an actual patch, but I think
it is easy for you.

Hinrich