Mgetty+Sendfax with Vgetty Extensions (FAQ)
Lichte...@acm.org
Last-modified: 1998/03/12
URL: http://www.webforum.de/mgetty-faq.txt
Maintainer: Klaus Lichtenwalder <Lichte...@ACM.org>
``mgetty+sendfax''
Answers to Frequently-Asked Questions
regarding Gert Doering's Fax-enabled getty replacement,
with Klaus Weidner and Marc Eberhard's voice extensions
Klaus Lichtenwalder
Lichte...@ACM.org
Posted by: Chris Lewis, cle...@ferret.ocunix.on.ca
------------------------------
Subject: Introduction
This document attempts to answer the most frequently asked questions
about mgetty+sendfax/vgetty, Gert Doering's fax-enabled getty
replacement with Klaus Weidner's and Marc Eberhard's voice processing
extensions.
The official release of mgetty is now (at least) 1.0. Gert adopted
the numbering scheme of the Linux kernel, with even release numbers (1.0,
1.2,...) being production version and uneven numbers being development
versions. Pls note that in 1.0 there's no vgetty support as that is
being revamped. If you need voice, you have to go for older versions or,
far better, for the newer development tree.
Subject: Japanese Version
I'm glad I can announce a Japanese Version made by chie nakatani
<jea...@mbox.kyoto-inet.or.jp>. You can find it at
<URL:http://epsenewsc.gee.kyoto-u.ac.jp/JF/JF-ftp/euc/mgetty_j.euc>
Subject: Legalese
This document is copyrighted by the FAQ Maintainer,
K.Lichtenwalder. It may not be reproduced in part or in full without
the written consent of the FAQ Maintainer.
------------------------------
Subject: Table of Contents
Part I: "Deciding whether to use it" questions
What is it?
What does it look like when it runs?
What do I need to use mgetty+sendfax/vgetty?
What other software do I need?
What terms cover my use of mgetty+sendfax/vgetty?
Where can I get mgetty+sendfax?
How can I integrate it into an office environment?
What's going on with those Class 1 modems?
What modems are recommended?
Where is the list archived?
Part 2: Other questions
How can I contact mgetty/vgetty users and developers?
What image file formats can sendfax send?
Why doesn't mgetty accept other file formats besides G3?
Why doesn't mgetty use the modem's autoanswer capabilities?
Why mgetty ignores /dev/tty* dialin, /dev/cua* dialout conventions:
Troubleshooting questions & answers
Programs generating "invalid" Postscript that can't converted
to the G3 format for sendfax
My ZyXEL doesn't work right after a ROM upgrade: What's wrong?
When is mgetty actually running? (i.e. what can mgetty control?)
The user's shell doesn't get killed after the line drops
AutoPPP appears in "who" listing
Things to think of if using AutoPPP, or, as you might call it, ppp connections
Part 3: Compatibility Issues
Suspicious fax machines
Part 4: Compiling, installing and using vgetty
Compiling vgetty
US Robotics voice format converter
---------------------------
Subject: Part I: "Deciding whether to use it" questions
------------------------------
Subject: What is it?
From: st...@work.bellingham.wa.us (Steve Work)
Mgetty+sendfax is a collection of programs to send and receive faxes
in a unix environment using a class 2.0 or 2 (they're different)
faxmodem. vgetty is an extension to mgetty, distributed with it, that
implements incoming voice call handling for certain voice-capable
modems, with new ones added regularly, if specs are available.
More specifically, the program `mgetty' allows you to use a class 2.0
or 2 fax modem for receiving faxes and handling external logins
without interfering with outgoing calls. `sendfax' is a standalone
program which sends fax files. `vgetty' is an extended version of
mgetty that can answer the telephone like an answering machine and
record a voice-mail message (if it finds one), or perform `mgetty's
fax or data call handling otherwise. The mgetty+sendfax distribution
includes vgetty and a good-sized gob of utility programs that help you
manage faxes and voice messages.
------------------------------
Subject: What does it look like when it runs?
From: st...@work.bellingham.wa.us (Steven Work) and the distribution
CC: cle...@ferret.ocunix.on.ca (Chris Lewis)
Like a smarter `getty'. getty is the program that manages the
first step of the login procedure on a Unix computer; when used
with a modem, it watches for an incoming call and (ordinarily)
prints the "login:" prompt (and reads the username, and passes off
to "login").
Unlike traditional versions of getty or uugetty, which will put a
modem into auto-answer mode, mgetty does not. When an incoming call
occurs, mgetty sees the "RING"s when they occur. When they do occur,
mgetty tells the modem to answer, and the modem will tell mgetty what
kind of connection happens. If it is FAX, mgetty will receive the
FAX. If data, mgetty prompts for a userid, then hands the open line
off to login for a normal data login.
Note that it's the modem's job to distinguish a FAX call from a data
call. Not all fax modems can do this, and if yours _can't_ there is no way
for mgetty to do this for it. mgetty can be used with modems that
cannot distinguish a fax call from a data call, but you must tell it
ahead of time what type of call to expect.
mgetty is also configurable to select programs other than login for
special connections (eg: uucico, fido or other programs) depending
on the login userid.
mgetty also supports caller-id and can deny connections based on
originating telephone number.
vgetty is an extension to mgetty that works with voice-capable
modems to provide additional call-handling capabilities. When the
modem reports a RING, vgetty has the modem pick up the line and
play a voice message (the greeting). If the modem detects a data or
fax calling tone, it reports this back to vgetty with special codes
(DLE-sequences) which causes vgetty to switch to either mode. Else
voice mode is used.
If instead the modem hears nothing following the greeting (a
certain level of silence that continues for a certain number of
seconds) it assumes the caller is a data modem and attempts a data
connection.
vgetty implements the normal answering-machine functions of
remote message playback as well; its operation is driven from shell
scripts, so you can extend it to a full voice-mail jail if you
wish. (This description of voice modem behavior applies to the
ZyXELs; I [st...@work.bellingham.wa.us] assume other voice modems
are similar.)
For an example on how a voice mail system looks like, there is the
voice_mail.sh script in voice/scripts from Marc Schaefer. Since the voice
shell is independent of the real modem hardware, it works on all supported
modems, not just ZyXELs. The hardware drivers hide the modem specific stuff,
so that the voice shell can provide a general interface that is completely
modem independant. Of course the reliability of the whole systems relies on
the reliability of the used voice modem. And there are quite notably
differences between different modems.
vgetty is intended for people who want to share a phone line for
data and voice use, with the main focus being voice calls. It is
*NOT* intended for a dialup system that occasionally gets a voice
call, since some modems are confused by hearing a recorded voice
message and won't connect.
If you have distinctive ring, you still can have one line, but vgetty can
detect the type of the call from the RING message and switch directly
to data/fax mode. In countries where distinctive ring is supported,
you can have dialup and voice on the same line without problems.
Voice extensions were originally written by Klaus Weidner
(kl...@snarc.greenie.muc.de) but are now maintained by Marc Eberhard
(Marc.E...@Uni-Duesseldorf.DE). Direct questions about them to that
address.
More from the distribution (some edits):
This is what you can do with `sendfax' if you have a standard class
2.0 or 2 fax modem:
* send faxes directly or using shell scripts (easily integrated into
other applications).
* do "fax polling", this means you can call the weather station and
get them to send you a fax containing the current weather map.
(Not all modem manufacturers implement this feature in their
modems!)
* create a "fax queue", outgoing faxes get sent automatically, the
user is informed by mail about the result.
`mgetty' allows you to use a single modem line for receiving calls
and dialing out.
* `mgetty' knows about "smart" modems, and will make sure that the
modem is always in a defined state (specific modem initialization
possible)
* Incoming calls are answered manually (`RING' -> `ATA' ->
`CONNECT') instead of using auto-answer (`ATS0=1'), this way the
modem won't pick up the phone when the machine is down or logins
are not allowed.
* mgetty completely replaces getty and/or uugetty. Like uugetty,
supports lock files in a fashion compatible with almost all known
versions of UUCP (HDB/BNU, SVR4, V7, Taylor in various flavours).
uugetty has some features mgetty doesn't support; see "How does
mgetty differ from uugetty?" below.
* mgetty supports System V style gettydefs terminal configurations.
* mgetty can receive class 2 faxes (if your modem supports it).
* mgetty knows about incoming FidoNet calls.
* mgetty has extensive logging / debugging features
* do "fax poll sending", that is, you can setup your machine as fax
poll server, to send some fax pages to "fax poll" callers. (Send
informations about your system, the current wheather map, ...). Be
warned, even less modems support this feature.
* mgetty can selectively refuse calls based upon CallerID, if your
modem supports it, and you're subscribed to the service. CallerID
is also logged.
* mgetty has facilities to allow you to refuse incoming FAXes when
available disk space is low.
* mgetty knows about incoming PPP calls, and can hand them off
to the PPP-daemon, without requiring a login/password sequence. This
feature is also known as AutoPPP
vgetty inherits all of mgettys features, and offers some additional
ones:
* behaves like a normal answering machine for human callers
* automatic fax reception when a T.30 calling tone is detected
* If the caller isn't a human or fax, a data connect is attempted,
if this is successful, the caller will get a normal login
* does not interfere with dialouts
* remote playback of messages via a DTMF code
* toll saver -- if there are new messages, pick up the phone
earlier, this way you can hang up in time to avoid a useless call
* message light - the autoanswer LED of your modem (if it has one)
is turned on if there are new messages
* easy playback - on some modems, you can play back the new messages
just by pressing DATA/VOICE
* using a speech synthesizer is possible - add the date and time to
messages (not included by default). The scripts show how to use a
speech synthesizer like rsynth, but it is not included in the
package. To use this feature, you need a voice modem for that;
a converter from the pvf format to the rmd (raw modem data) format
exists. This is not true for all supported modems.
* voice conversion utilities - play messages on /dev/audio
(Not for all supported modems, some voice modems use a proprietary
format)
------------------------------
Subject: What do I need to use mgetty+sendfax/vgetty?
From: st...@work.bellingham.wa.us (Steve Work), and distribution
CC: cle...@ferret.ocunix.on.ca (Chris Lewis)
Several things. A computer running some (most) variants of the Unix
operating system. (The operating system must support termio.h or
termios.h; this generally rules out "pure BSD" systems.) For support
of dial-in data connections (a la "getty"), you need a modem (probably
one somewhat compatable with the H*yes "AT" command set). For sending
and receiving faxes, you need a modem that understands the Class 2 (or
2.0) fax command set. For voice processing, you need a modem that is
capable of doing voice.
Vgetty currently supports Dolphin, Dr. Neuhaus Cybermod, Elsa, IS 101
compatible, Rockwell, Sierra, US Robotics and all ZyXEL modems.
The Cirrus Logic, ISDN4Linux and UMC drivers are basically working,
but they need to be updated to the new internal interface between the
generic vgetty parts and the hardware dirver. This change was
necessary to be strictly ANSI C compatible. Vgetty now compiles with
gcc -Wall -pedantic without a warning.
Mgetty has been successfully installed and run on at least the
following systems, probably more by the time you read this list:
SCO Unix 3.2.1 (ODT 1.0) (very well tested)
SCO Unix 3.2.4 (ODT 2.0 + 3.0) (very well tested)
SCO Open Server 5.0 (Gert uses it ...)
Linux 0.99pl1 .. 2.1 (very well tested)
ISC Unix 3.0 (tested)
SVR4 Unix (well tested)
AT&T 3B1 3.51m (well tested)
HP-UX 8.x (well tested)
AIX 3.2.5, 4.1.4, 4.2 (mgetty, not vgetty)
SunOS 4.1.x (well tested)
SunOS 5.x (at least with USR 33.6
Misha Pavlov <o...@interport.net>)
NetBSD / FreeBSD (works)
BSDI v1.1 (under work, not done --
gr...@wwa.com)
It should be possible to run mgetty on any other Unix with
`termio.h' or `termios.h'. For best results, `select(S)' or
`poll(S)' are recommended, but there's a workaround. (Warning: for
Unix SVR3.1 or earlier, *do not use poll()*, it will not work on
tty devices.)
Up to now, it has been successfully used with at least the following
modems, and probably more:
Here's a short list of often used modems. For an up to date list check
with doc/modems.db from the distribution:
* Aceex 1496
* Boca V.32bis
* Creatix LC 288 FC
* Practical Peripherals PM14400FXMT
* TKR Terbo Line
* U.S. Robotics Courier V.34 Fax
* U.S. Robotics Sportster V.34 28.800 Fax Modem
* Zoltrix Platinum Series 14.4
* ZyXEL 1496E+, always recommended
Mgetty *should* work with all class 2 faxmodems. Maybe the DC2
character sent at the beginning of a page by `faxrec.c' must be
changed to XON, for old class 2 modems (implementing very old drafts
of the standard). Unfortunately, each class 2 modem can be a tiny bit
different.
Early USR fax modems did a bad job of conforming to the Class 2.0 (and
maybe Class 2) operating standards. Reports are that current USR
modems (Sportster and Courier) work without excuses.
------------------------------
Subject: What other software do I need?
From: cle...@ferret.ocunix.on.ca (Chris Lewis)
CC: ge...@greenie.muc.de (Gert Doering)
For data only, no other software is needed.
mgetty itself can only send or receive G3 (raster) format. However,
the distribution includes tools to convert raw G3 files to or from the
format used by "pbmplus", the Portable Bitmap Toolkit. The pbmplus
formats support (or are supported by) most raster-image programs and
file formats generally used in the Unix world. pbmplus is available
at this URL (among others):
ftp://sunsite.unc.edu/pub/X11/contrib/pbmplus10dec91.tar
The mgetty+sendfax distribution contains a patch to fix pbmplus's
broken pbmtog3 converter -- using the unpatched pbmtog3 can cause
errors during transmission.
GhostScript, the free Postscript page description language
interpreter, can convert PostScript to G3. Ghostscript is available
at this URL (among others):
ftp://sunsite.unc.edu/pub/gnu/applications/ghostscript-2.6.1.tar.gz
(also check out patch files in same directory.)
Hp2pbm, available from ftp:// ??, can convert text and PCL (HP
Laserjet language) to G3 or PBM. It also contains programs for
converting PBM to PostScript, PCL and Epson printers.
PBMPLUS has converters from most existing raster formats or ASCII
to PBM, and from PBM to most raster formats. You'd use the pbm2g3
and g32pbm utilities in mgetty to convert between PBM and G3.
In essence, you can run with hp2pbm or PBMPLUS alone. With GhostScript,
you also need PBMPLUS or hp2pbm to convert ASCII (used for page headers
etc.) to G3.
Mgetty+sendfax includes some voice processing utilities in the voice/
subdirectory. These tools (pvftools) can convert ZyXEL, Rockwell and
ISDN4Linux voice formats. Other formats are coming. People reported
success in translating GSM encoded voice formats, so support for that
will also be added in the future. This means that vgetty currently
supports the three above and support for more modems is planned.
------------------------------
Subject: What terms cover my use of mgetty+sendfax/vgetty?
>From the distribution:
"The mgetty+sendfax package is Copyright (c) 1993 Gert Doering. You
are permitted to do anything you want with this program -
redistribute it, use parts of the code in your own programs, ...,
but you have to give me credit - do not remove my name.
"If the program works for you, and you want to honour my efforts,
you are invited to donate as much as you want.
"If you make money with mgetty, I want a share. What I mean by that
is: it's perfectly OK with me to get paid for mgetty installation
or support, but if you want to actually sell mgetty, or pack mgetty
with a modem and sell it as "unix fax package", contact me first.
"*WARNING:* This package is still BETA software. Use it at your own
risk, there is *no* warranty. If it erases all the data on your
hard disk, damages your hardware, or kills your dog, that is
entirely your problem. Anyway, the program works for me and quite a
lot of other people."
Marc put the voice part under the GPL. To avoid misunderstandings,
Gert decided to not include the GPL text into the distribution,
though. The copyright for the voice stuff is GPL.
------------------------------
Subject: Where can I get mgetty+sendfax?
The home page is at http://www.leo.org/~doering/mgetty/
The official release sites are these URLs:
ftp://ftp.leo.org/pub/comp/networking/communication/modem/mgetty
ftp://sunsite.unc.edu/pub/Linux/system/Serial/mgetty+sendfax*
ftp://tsx-11.mit.edu/pub/linux/sources/usr.bin/... (or so)
ftp://linux.nrao.edu/pub/packages/mgetty+sendfax/
------------------------------
Subject: How can I integrate it into an office environment?
From: Klaus Lichtenwalder <Lichte...@ACM.org> and the mgetty dist.
With mgetty, you get a subdirectory called frontends/. There are a lot
of more or less documented programs for viewing/printing/organizing
received faxes, for entering them into the queue from all kinds of
different programs.
------------------------------
Subject: What's going on with those Class 1 modems?
From: Klaus Lichtenwalder <Lichte...@ACM.org>, modifying comments
from Gert Doering
While Class 2 and Class 2.0 modems do most of the dirty work
themselves, clever modem manufacturers seem to drift toward implementing
only class 1 fax modes. Probably so they can use smaller proms while
also integrating voice functionality. Well, they want to see $$$,
too..
The bad thing with Class 1 is, the modem expects all the dirty work to
be done by the host computer. After all, why did you buy a Pentium II
300MHz? Just for hitting on a few keys from time to time? So you have
a very fast cpu and there should be no problem to do something every n
ms? Right? Wrong! Most of the Unix systems out there don't have
realtime extensions (this *is* a real time job, after all), and if
they have, there's no real standard to these extensions, so generally
it's not possible to maintain that strain. Windows knows how to use
class 1 modems? Yeah, sure, the multitasking is quite different, you
can quite easily exclude other processes from running, because you
just want to have that high priority process. And if you insist in
doing some other time consuming tasks, your fax might no be very
readable... So, so much prosa for this situation, you still might see
some class 1 support in the near future. Just hang on and in the
meantime, boycott modem manufacturers who only support class 1 modes.
------------------------------
Subject: What modems are recommended?
From: Gert Doering <ge...@greenie.muc.de> and the List
----------------------------------------------------
MODEM Recommendations for use with mgetty/vgetty
------------------------------------------------
What modem to recommend depends on your needs. There are very few modems
that handle fax + data + voice perfectly, but quite a number that are
well suited for two of the categories, and less good (or sometimes not
at all) for the third.
For Fax+Data:
* USR Courier V.34/X.2 (no voice)
For FAX+Voice:
* ZyXEL 1496 (data only up to 19200, but best fax implementation)
For Data+Voice:
* USR Sportster VI series (fax implementation is VERY bad)
For FAX+DATA+Voice
* ZyXEL 2864
* MultiTech MT2834ZDXv
* ELSA MicroLink TQV series (fax is not perfect, but works ok)
Last updated: February 1998
------------------------------
Subject: Where is the list archived?
From: Gert Doering <ge...@greenie.muc.de>
A pointer to the archive as well as lots of other documentation can be
found at <URL:http://www.leo.org/~doering/mgetty>. The archive itself
can be found at <URL:http://eli.wariat.org/mgarc/index.html>
------------------------------
Subject: Part 2: Other questions
------------------------------
Subject: How can I contact mgetty/vgetty users and developers?
From: ezmlm at crynwr
A mailing list exists. Due to its international character, the only
supported language on this list is *English*, though it's gated to
some local newsgroups, de.alt.comm.mgetty for example. To subscribe to
the list, send mail to mgetty-s...@crynwr.com, to unsubscribe,
send a mail to mgetty-un...@crynwr.com, NOT to mge...@muc.de.
If you want to specify a different e-mail address for your subscribtion
than the address you send your subscribe request, send it to
mgetty-subscribe-<your e-mail address with @ replaced by =>@crynwr.com.
This address gets checked by sending a confirmation message.
Example (from mgett...@crynwr.com): to specify G...@heaven.af.mil as
your subscription address, send mail to
<mgetty-subscribe-God=heaven...@crynwr.com>. I'll send a
confirmation message to that address; when you receive that message,
simply reply to it to complete your subscription.
The volume on mge...@muc.de varies from one or two messages a week to
ten or so a day, depending on development activity.
If you have questions because of a misbehaving mgetty or vgetty, be
sure to include the relevant parts of your log file, usually obtained
with a loglevel of 6 The default location of these log files is /tmp/
for mgetty and /var/log for vgetty.
------------------------------
Subject: What image file formats can sendfax send?
From: ge...@greenie.muc.de (Gert Doering)
Fax input format:
raw G3 data (according to CCITT standard T.4, 1-dimensionally
compressed). Can be created by GhostScript's "dfaxhigh" or "dfaxlow"
drivers (they will create raw G3 data with a 64 byte header, that
sendfax + g3cat + g32pbm will recognize and skip) or by the "pbm2g3"
program.
Warning: the pbmtog3 program from the "pbmplus" distribution does *not*
create valid G3 data according to T.4, to be precise, the lines are
shorter than 1728 pixels and the leading EOL code is missing. To fix
it, use the patch that is provided in "patches/pbmtog3.p1". Even
better, use my own utility. (The patch is *NOT* needed for my pbm2g3
program!).
Use of a unpatched pbmtog3 will most likely cause +FHNG:50 or +FHNG:54
error codes.
Do not use "tiff-G3" data as input. Though the data itself is G3 encoded,
it's wrapped into a complex layer of TIFF headers that would
require non-trivial parsing that's simply not needed for sendfax.
Faxing a TIFF file will result in a warning message from sendfax, and,
most likely, in a failed transmission.
Fax output format:
The files that mgetty places into FAX_SPOOL_IN are in the same format
as the files that sendfax wants to send, raw G3 data as of CCITT T.4.
To convert them to "pbm", use the g32pbm program. To convert them to
X-Windows xwd format, use the "contrib/g3toxwd" program.
If the files are received without transmission errors, it is possibly
to send received fax files without any additional conversion. Since
some modems insert filling zero-bits, a run through "g3cat" is
recommended anyway, this will remove any surplus stuff, and clean up
garbled lines.
------------------------------
Subject: Why doesn't mgetty accept other file formats besides G3?
Why does mgetty only send raw G3 fax files, instead of converting
arbitrary image files (like TIFF) on the fly?
>From Chris Lewis:
"As I understand it, in addition to its own formats, TIFF can encapsulate
some number of other formats. Put another way, TIFF is, at least to some
extent, simply a way of getting magic numbers into other formats so that
TIFF-capable converters can handle multiple formats transparently.
"Yes, you certainly can have TIFF encapsulate a G3, and mgetty wouldn't
have much trouble with that. However, that leaves you with the question -
what does mgetty do if it's not a G3 that's been encapsulated? It
would have to convert it. And then we would encounter situations where
mgetty's conversion speeds couldn't meet the class II FAX transmission
timeouts, and you've wasted telephone time... Ditto on reception. Indeed,
there is a real possibility that mgetty would not be able to keep up
if the input or output file was anything other than a G3.
"Approaching this from another viewpoint, we should also remember that
mgetty is a transfer protocol and implementation. *Not* conversion
software. mgetty needs to read and write G3s, and that's all. Leave
conversions to other software."
And from Gert Doering:
"Well, TIFF is a very complex file format, that can support dozens of
different page data encodings, different byte / bit orderings,
multiple pages per file, and so on. While TIFF is a flexible format,
parsing it is a complex task.
"In contrast, mgetty's "g3" files are raw G3 data. No headers, no
formatting, no need for the transmission code to dig around in the file
to find the actual page data. One page per file, simplest possible.
"Ignoring TIFF (leaving TIFF conversions to outside software)
simplifies mgetty and sendfax enormously, but doesn't complicate the
user interface much, as long as "faxspool" or similar tools are used
that will hide the internals, that is, how the fax backend expect its
data, from the user. To be precise, leaving out TIFF support
*simplifies* mgetty. Many Unix programs can read pbm and converting
g3 -> pbm is very easy, while I haven't seen a good *multipage*-Tiff -
to - PBM converter yet."
------------------------------
Subject: Why doesn't mgetty use the modem's autoanswer capabilities?
1. Because it isn't possible to distinguish a fax from a data call if you
don't have full modem control. Besides, it will make sure that the modem
doesn't answer the phone while the host isn't ready for it.
2. And callerid won't work without extreme difficulty.
------------------------------
Subject: Why mgetty ignores /dev/tty* dialin, /dev/cua* dialout conventions:
[Under Linux and most other Unices (excepting SunOS) mgetty should be
set on the tty* devices, and the cua* devices should be ignored
entirely:]
From: "Theodore Y. Ts'o" <ty...@mit.edu>
/dev/ttySxx devices are fully POSIX-compliant TTY devices. If you are
only going to be using one set of tty devices, you should be using
/dev/ttySxx.
/dev/cuaXX devices are different from /dev/ttySXX in two ways --- first
of all, they will allow you to open the device even if CLOCAL is not set
and the O_NONBLOCK flag was not given to the open device. This allows
programs that don't use the POSIX-mondated interface for opening
/dev/ttySxx devices to be able to use /dev/cuaXX to make outgoing phone
calls on their modem (cu stands for "callout", and is taken from SunOS).
The second way in which /dev/cuaXX differs from /dev/ttySXX is that if
they are used, they will trigger a simplistic kernel-based locking
scheme: If /dev/ttySXX is opened by one or more processes, then an
attempt to open /dev/cuaXX will return EAGAIN. If /dev/cuaXX is opened
by one or more processes, then an attempt to open /dev/ttySXX will
result the open blocking until /dev/cuaXX is closed, and the carrier
detect line goes high.
While this will allow for simple lockouts between a user using a modem
for callout and a getty listening on the line for logins, it doesn't
work if you need to arbitrate between multiple programs wanting to do
dialout --- for example, users wanting to do dialout and UUCP.
I originally implemented the cuaXX/ttySXX lockout mechanism back before
FSSTND established a standard convention for the use of tty lock files.
Now that it's there, people should use the tty lock files and not try
using /dev/cuaXX. The only reason why /dev/cuaXX hasn't disappeared yet
is for backwards compatibility reasons.
- Ted
[Under SunOS *everything* should use cua*, as follows:]
From: Gert Doering <ge...@muc.de>
The two-device scheme is meant to prevent multiple processes from
accessing the same physical device at the same time. Since mgetty
opens the port with O_NDELAY, the kernel sees a process on tty*
(mgetty) and prevents any open() on cua* (uucico, cu, ...). So, you
have to use the same device for both program types, and that's cua*.
------------------------------
Subject: Troubleshooting questions & answers
From: ge...@greenie.muc.de
Q: I keep getting the error code +FHNG:50 or +FHNG:54 after sending a
page. Sometimes it says "page bad, retrain requested" and infinitely
resends the page.
A: This error means that something went wrong between the two machines,
while your end was sending the page data. In the case of +FHNG:50 or
+FHNG:54, the other end most likely simply hung up (so your modem
couldn't get any page transfer status at all), in the other case, the
receiver complained that it didn't like the data on the page.
The most common reason for this behaviour is that you used a copy
of ``pbmtog3'' that came from the ``pbmplus'' distribution and doesn't
have my patch included (Mind you, the pbmtog3 program is required for
the page headers that faxspool puts on top of each page!).
If that is not the reason, there may be flow control problems, or the
line have simply has been very noisy, or so. Get in touch with the
receiver, and find out whether the page looks good or whether there are
lines missing, others corrupted, ... - if there are errors, check your
flow control setting.
If there are no errors whatsoever, and you're sure that you use my
version of pbmtog3 (called pbm2g3 or a patched version of pbmplus',
then I'm out of wits - something's broken in the modem. Maybe you
should upgrade your ROM version.
Q: When receiving faxes, I get the +FCON message logged, then mgetty says
"starting fax receiver". fax_wait_for(OK) is called, logs some random
junk bytes, and gives up after a minute, complaining about "timeout".
A: Most likely, you have a modem that switches baud rate to 19200 bps
right after sending the +FCON message to the host, and the normal port
speed is something else. Check policy.h whether FAX_RECEIVE_SWITCHBD
is defined, and change the setting (if the modem does *not* change
speed, but the define is *set*, the effect will be similar).
Better yet, with the runtime configurable stuff, is to add the entry
switchbd <nnn>" in mgetty.config
Q: I keep changing values in policy.h, but nothing changes.
A: maybe you've had an older version of mgetty installed to
/usr/local/bin/mgetty and are calling this from /etc/init? Newer
versions are installed in /usr/local/sbin/mgetty. Check the time
stamp on the mgetty you just compiled vs. the mgetty listed in
/etc/inittab.
Q: Half of the faxes that I receive are too short, they look as if every
second pixel line is missing.
A: Well, they *are* short: most likely, they are received in normal
resolution (204x98 dpi) instead of fine resolution (204x196 dpi). You
can see it in the filename, if it starts with "ff*", it's fine, if it
starts with "fn*", it's normal resolution. To ``stretch'' a normal
resolution fax to proper proportions, use ``g32pbm -stretch fn...''
Q: login complains with ``no utmp entry, must execute login from the
lowest level sh''
A: I *told* you not to fiddle with the ``utmp'' field... - most likely,
in your ``login.config'' file, the utmp field for the entry calling
/bin/login isn't "-".
Another reason could be a faulty /etc/init (hits only Linux users) or a
corrupted /etc/utmp file. In that case, reboot your machine.
Q: When mgetty is running and I dial out, I do not get "CONNECT" but only
junk, as "+FCO", "+FTI:", "+FHS:20"
A: Well, yes, that's a problem with the 2.0 implementation in mgetty. That
is: while mgetty is running, the modem is in "AT+FCLASS=2.0" mode and
expects to connect to a fax on the remote side. (With class 2, we
worked around this by setting +FCLASS=0;+FAA=1, but that will make the
modem answer in class 2, not 2.0 [subject to further testing!])
Solution: in the program dialing out, initialize the modem with
"AT+FCLASS=0". Most likely, a modem reset (ATZ) will also help.
Q: every time mgetty starts up, the permissions of my tty device get
changed and I have to issue "chmod +w /dev/ttySx" to be able to
dial out.
A: that's not a bug, that's a feature. You don't *want* to allow anybody
using your machine to be able to dial out (think of your phone costs!),
so it's a security issue.
If you *want* to allow dialout for everyone, #define FILE_MODE 0666
in policy.h. I would not recommend it, I would give access to a
special group, and put every one that may dial out into this group.
Q: I have a Linux system, and while trying to dial out on /dev/cua1
(mgetty is running on /dev/ttyS1), it says "device busy" (EBUSY)???
A: use the same device (always!!) for dial-in and dial-out.
On Linux, use /dev/ttySx, on SunOS and *BSD use /dev/cuax.
Q: If I create a fax file with "gs -sDEVICE=dfaxhigh ..." and send it with
sendfax, everything works *great*. If I run it through "faxspool", the
receiving side reports an error. Is the "g3cat" program broken?
A: No, g3cat isn't the problem. The real problem is "pbmtog3", and I bet
you have the pbmtog3 program from the pbmplus distribution installed.
This program is *broken* (patch is in mgetty/patches/pbmtog3.p1), that
is, it doesn't create proper T.4/G3 fax data. Thus, the receiving fax
machine will get a fax that has some corrupt lines (the page header)
and will complain about it.
Patch pbmtog3, or use mgetty's pbm2g3. It's faster anyway.
Q: mgetty doesn't accept FidoNet calls. I get log entries like this:
10/30 01:54:54 ##### data dev=ttyS1, pid=3401, caller=none,
conn='38400/V32b 14400/V42b', name='', cmd='/bin/login',
user='**EMSI_INQC816**EMSI_INQC816q.'
or this:
10/30 05:31:03 ##### data dev=ttyS1, pid=7238, caller=none,
conn='38400/ZyX 16800/V42b', name='', cmd='/bin/login', user='q.q.q.'
A: did you compile mgetty with -DFIDO defined? I don't think so. If
-DFIDO isn't set, mgetty doesn't know about fido.
Q: Some of my programs use binary lockfiles and some use ASCII
lockfiles. Why does mgetty complain? Can't it recognize both?
A: Mgetty complains because your system configuration is _wrong_.
These error messages are there to help the system administrator
notice a *severe configuration error* on his site.
If all programs would understand both types of Locking, the
messages would be silly, but since kermit usually simply ignores
ascii locks, and uucico does so for binary locks, the situation is
*highly* error-prone, and sysadmins should *SEE* this.
Recompile your applications that use the modem so that all agree on
the lockfile types.
Q: My modem and I share one phone line. Now I answered the phone
and a modem greets me. How can I make mgetty take over?
A: Send the signal SIGUSR1 to the mgetty process. It will then answer
the phone.
Q: I can fax only one page, then I get an error
A: When you see something like this in your log:
> 03/22 19:15:58 yS1 fax_wait_for: string 'OK'** found **
> 03/22 19:15:58 yS1 fax_send: 'AT+FET=0'
> 03/22 19:15:58 yS1 fax_wait_for(OK)
> 03/22 19:15:58 yS1 fax_read_byte: read returned 0: Unknown error
> 03/22 19:15:58 yS1 fax_get_line: cannot read byte, return: Unknown error
> 03/22 19:15:58 ##### failed transmitting f1.g3: +FHS:-4, time=55s
you should define FAX_SEND_IGNORE_CARRIER in policy.h and recompile
your binaries. There are some modems that drop the DCD line between
pages. Alternatively, you can define ''ignore-carrier true'' in your
config file.
Q: The message "WARING: DSR off - modem turned off or bad cable?" keeps
showing up in the mgetty logfile.
A: If you see this message, it basically means that mgetty can't get
the status of dsr (data set ready). This can be because the cable
is bad or because you are using mgetty on solaris. Solaris doesn't
give you the status of dsr. If you are running mgetty on a different
platform, you might consider checking your cable or the modem
setting.
------------------------------
Subject: Programs generating "invalid" Postscript that can't converted
to the G3 format for sendfax
From: Gert Doering <ge...@greenie.muc.de>
Right now the list of programs generating "invalid" postscript (causing
Ghostscript to create G3 files with wrong line width, in turn causing
sendfax to fail) includes:
FrameMaker 4.0
WinWord (dunno which version)
dvipsk 5.58f
Gert Doering wrote on Sun, 22 Dec 1996 00:26:55 +0100 in message
<m0vbaoh...@greenie.muc.de>:
Interesting enough, I tried dvipsk 5.58f today, and it does *NOT* emit any
of those bad "setpagedevice" commands, and thus the resulting postscript
*will* generate correct page sizes. So maybe it's sufficient to upgrade
your dvips program to the most recent version.
Maybe it's necessary to tell *ghostscript* what the "right" paper size is
supposed to be (A4 -- but you must make sure that dvips makes "a4" as
well!), by specifying "gs -sPAPERSIZE=a4".
For some people, even that does not suffice (some strange interaction
between different paper sizes in different programs), so it may help to
post-process the G3 files with "g3cat" (from mgetty 0.99*, versions
after July 96!), or with "g32pbm badfile.g3 | pbm2g3 >goodfile.g3". The
latter will definitely help.
------------------------------
Subject: My ZyXEL doesn't work right after a ROM upgrade: What's wrong?
From: fe...@escape.vsse.in-berlin.de
Do a full modem reset after upgrading the firmware. This is not
described in the German ZyXEL manual (is it described in the
English one?) but should be done in any case.
------------------------------
Subject: Why the occasional "tcsetattr failed: I/O error" message?
From: ge...@greenie.muc.de
Q: Occasionally, mainly after "clean" user logouts (that is, the user
typed "exit" instead of just hanging up), I get the message
09/08 21:26:26 yS2 lowering DTR to reset Modem
09/08 21:26:27 yS2 tcsetattr failed: I/O error
in the mgetty log file, and a similar I/O error message in the syslog
file.
A: Well, this is a Linux and SunOS specific problem: if the modem is still
connected to the other end when mgetty starts, mgetty will force it to
hangup by lowering DTR (and sending +++ATH, in case DTR drop won't
suffice). This will make the modem lower the DCD (carrier detect)
line. Unfortunately, this will trigger a security mechanism in the
Linux kernel, which will prevent all further access via that file
descriptor. This is done to prevent one well-known password hack
(I won't that explain in detail).
Mgetty knows about that problem, and, upon noticing an error at this
point during modem initialization, will simply reopen the port and
redo all modem / port setup stuff.
Because suppressing that error message would be messy, it keeps
appearing, but it is harmless (... "trying again").
------------------------------
Subject: When is mgetty actually running? (i.e. what can mgetty control?)
From: Klaus Lichtenwalder <Lichte...@ACM.org>
At what moments can mgetty exert control over the line? Let's
consider mgetty's life span: at some time (e.g. system boot, init
assumes the specified run level and starts mgetty) mgetty starts and
(simplified) checks for the lock file on the device it has to
control. If there's no lock file, it initializes the modem and wait's
for an event on the line. If a character arrives, mgetty does not read
it but first checks the lock file.
* If a lock file is present, mgetty knows some{body,thing} wants to
dial out and periodically checks for the existence of the lock
file. After the lock file is gone, the device is free and mgetty
simply exits. It'll get restarted by init, so the modem line will
get reinitialized.
* If no lock file is present, the modem most probably sent a RING, so
we go check for the expected word (= RING) and send our
answer_chat. If it's a fax call (and the modem can receive faxes and
we allow receiving them) we get the pages, store 'em away and exit,
thus restarting mgetty, see above. If it's a data call (I ignore
DISTINGIVE_RING and Caller_ID features for now), we prompt for the
login and wait for an answer. After the answer is read, it's checked
against the login.config file and the appropriate program get's exec'd
by mgetty, which means, there's *no* mgetty for that line. Only after
the termination of the connection, when the other process(es) exit(s),
mgetty will be restarted by init.
------------------------------
Subject: The user's shell doesn't get killed after the line drops
From: Gert Doering <ge...@greenie.muc.de>
> why doesn't mgetty kill the user's shell when the user just hangs up
> the modem, without doing 'logout'? [Mod: or the line simply crashes]
How should mgetty kill a user's shell? While the user is logged in, mgetty
is *not running*.
The kernel will signal the shell via the SIGHUP signal when the DCD line
on the modem drops. Then the shell will exit, and init will re-start
mgetty. Unfortunately, the BASH shell is very broken in various versions
and will ignore the SIGHUP signal, so that could be one reason why it
isn't exiting.
Another reason could be an improper modem setup (AT&C0), make sure
that the modem raises and lowers the DCD line properly (check with the
modem manual) and that your serial cable is OK (swap it with another
one for testing).
------------------------------
Subject: AutoPPP appears in the "who" listing
From: Gert Doering <ge...@greenie.muc.de>
If you use the /AutoPPP/ option for automatically creating a ppp
connection, the user name does not appear in the who list. Instead,
something like AutoPPP or ppp is displayed. Solution: you have to
patch pppd, that's not mgetty's issue and thus can't resolved patching
mgetty.
If you have the patched pppd, this is how it works. You have
to add the option "login" to pppd in login.config and change the third
column to "-" (in the example it's a_ppp or ppp).
------------------------------
Subject: Things to think of if using AutoPPP, or, as you might call it,
ppp connections
From: Gert Doering <ge...@greenie.muc.de>
You can't get a ppp connection going. So, what should you check before
calling for help in the list.
Run mgetty with a debug level high enough to see what login it will
try. It might not be wrong to just use level 9 to see everything
happening.
Scenario 1: mgetty recognizes AutoPPP:
03/11 12:06:31 yS1 match: user='/AutoPPP/', key='/AutoPPP/'*** hit!
03/11 12:06:31 yS1 login: utmp entry: ppp
03/11 12:06:31 yS1 looking for utmp entry... (my PID: 5313)
03/11 12:06:31 yS1 utmp + wtmp entry made
03/11 12:06:31 yS1 calling login: cmd='/usr/sbin/pppd', argv[]='pppd...
03/11 12:06:31 ##### data dev=ttyS1, pid=5313, caller=none, conn='LAPM', name=''
At this point, mgetty executes pppd and has nothing to do with
connection failures, failing authorization, the police banging at your
door or anything else. Check pppd's logs for any trouble.
If you use "who" and the only thing you can see is "a_ppp", go check
your pppd sources. It's pppd's job to forge a utmp entry
It looks like some unix clients should explicitely send the string
"/AutoPPP/" as there seems to be a problem in connecting. Mgetty
doesn't seem to get ppp-frames and waits for something it can decide
what to do. As you are most probably using a script for the
connection, there should be no problem to send that string.
------------------------------
Subject: Part 3: Compatibility Issues
------------------------------
Subject: Suspicious fax machines
From: h...@ix.de (Harald Milz)
I'm collecting all data concerning suspective fax
machines, i.e. those which made problems in cooperating
with sendfax. The main reason is to find out whether
there are specific fax machines that refuse to work
with sendfax and/or your fax modem. As a goal, we will
be able to track down the bug(s).
To contribute, please fill in the following template
and send it to me (h...@ix.de):
1. <fax machine's brand and model>
2. <corresponding fax number> (optional)
3. <fax modem brand and model>
4. <fax modem's firmware revision> # tbd from ATI1
5. <protocol parameters> # tbd from Faxlog
6. <errlog line from Faxlog> # tbd from Faxlog
7. <remarks>
If you encounter problems with a fax machine, please
call the receiving party and ask them for their fax
machine's brand & model and if they are willing to
offer their machine for some (limited) testing.
The more exact your data is (the first 3 entries aren't
too good :-} ), the better the result will be,
hopefully.
This list is posted once a month (automatically) and if
five new entries were added to it (manually).
Here's what's already in the list:
1. Panasonic Panafax UF311
2. +49 89 74824899
3. ZyXEL U1496EG+
4. U1496EG V 6.10g P
5. +FDCS:1,3,0,2,0,0,0,4
6. +FHNG:50 (Unspecified Transmit Phase D error)
7. when sending 15 pg, connection broke after 6 pg.
1. NEC Nefax 17
2. +49 2242 82114
3. ZyXEL U1496EG+
4. U1496EG V 6.10g P
5. +FDCS:1,3,0,2,1,0,0,4
6. +FHNG:50 (Unspecified Transmit Phase D error)
7. machine didn't refuse when sending only 3 pages
earlier. This time, 15 pg were sent.
1. Telekom AF-310
2. +49 7231 560851
3. ZyXEL U1496 E / 6.10a, E+ / 6.01, E+ / 6.11a
4.
5. +FDCS:1,3,0,2,0,0,0,4
6. +FTPS:2 -> page bad, retrain requested
7. sendfax hangs up after three tries.
received fax shows black and white boxes at the
footer, such as,
### ### ### ### ###
### ### ### ### ### ...
### ### ### ### ###
------------------------------
Subject: Part 4: Compiling, installing and using vgetty
------------------------------
Subject: Compiling vgetty
From: Marc Eberhard <ma...@poseidon.thphy.uni-duesseldorf.de>,
Klaus Lichtenwalder <Lichte...@ACM.org>
After having set up mgetty for compile (i.e. copying policy.h-dist to
policy.h and editing it), you change to the voice/ directory and type
make. For installation (make install) be sure to copy voice.conf-dist
to the mgetty configuration directory (usually
/usr/local/etc/mgetty+sendfax) as voice.conf and edit it, using the
comments in that file.
------------------------------
Subject: US Robotics voice format converter
From: Scott Hutton <shu...@indiana.edu>
There's a script I wrote in voice/pvftools called "extract_gsm" that
extracts the GSM data from a US Robotics "rmd" file. Once you have
that, and you have a version of "sox" that can convert .gsm files in
to .au (or other format) files, you're in business.
I found the necessary patches for sox after a bit of digging:
http://kbs.cs.tu-berlin.de/~jutta/toast.html
You'll need the GSM library before you try to apply the sox patches.
It's too bad sox doesn't come with GSM support--it seems to have
about everything else...
Also, Thomas Hellstroem (tho...@vtd.volvo.se) had some C programs to
extract GSM data and to "pack" GSM data into rmd files off of his page.
Unforutnately, I didn't save the URL.
-Scott Hutton, UCS Messaging Team, Indiana Univeristy