Previous | Next

Section 10 TERMINALS

Objectives

By the end of this section you should

have been introduced to procedures for connecting terminals to a UNIX computer,
be able to describe the login procedure for a terminal,
be able to list and explain the configuration files used by terminals,
be able to explain the purpose of the commands tty stty,
explain the purpose and relation between termcap and terminfo,

UNIX is a multi-user operating system. To make use of this attribute multiple users must be able to connect to the system at the same time. This implies that there must be multiple access points. Dumb terminals are the cheapest method for providing multiple access points to a UNIX machine.

In most cases a dumb terminal is connected to a UNIX machine using a serial line. A dumb terminal does little more than present text to the user and transfer keystrokes from the terminal back to the central computer. It is dumb because the terminal does no processing on the data.

Even though the interface on such beasts is primitive they are still one of the most used methods for adding extra access points to a UNIX computer. Businesses wanting to use dumb terminals have two options

purpose built dumb terminals, or
modern personal computers using communications programs to simulate dumb terminals.

Hardware

The process of connecting terminals to a UNIX machine follows a fairly generic set of steps that can be divided into two separate sections, hardware and software. Figure 10.1 lists the steps in the hardware process.

Choose a port on the computer.
The device has to be connected to the computer . To do so an I/O port and the associated device file must be chosen.
Obtain the correct cable.
Find a cable of the right type that has the right plugs for your device and computer.
Configure the device.
The device will have to be configured to work correctly with your machine.
Make the connection.
Connect the device and prepare it for operation.
Test the connection.
Send output of some description to the device file to test whether or not the hardware connection is working.
Figure 10.1. Generic Hardware Process for Adding a Device.

Ports and Devices

There are two steps to choosing a port to which to connect a device

choosing the physical port, and
A typical UNIX computer is likely to have many different ports of various configurations. Some computers will be connected to multi-port cards that supply multiple (4, 20 and more) ports.
finding the corresponding logical device file.
Each physical connector has a corresponding device file through which the operating system passes information to the port.

Table 10.1 lists some of the common device files on a UNIX machine.

	Device File	Purpose

	/dev/ttyn	Generic device file used for terminals,
			  usually corresponds to a serial port
			  (n is a number e.g. /dev/tty0)
	/dev/modem	modem device file
	/dev/lp		line printer device file
	/dev/par	parallel port
	/dev/ser	serial port

		Table 10.1. Example Device Filenames.

Terminals are generally connected using serial ports. The number, type and location of the serial ports will be machine specific. Some machines will have multi-port serial cards that provide from four or more serial ports.

Most serial ports will be associated with device files of the format /dev/ttyn, where n is some number.

Cables

The next step is obtaining the correct cable to connect the device to the computer. To do so the cable must be correctly configured and have connectors that are the

right sex, and
Plugs are either female (small holes) or male (small prongs).
the correct size
Serial plugs for example can be either 9 pin, 25 pin or a couple of other configurations.

Even though serial cables are meant to follow the RS-232 standard there are a number of differences that mean no two serial cables are the same.
Serial cables are used to connect two different types of device

DTE, data terminal equipment
Most terminals and computers fall into this category.
DCE, data communications equipment
Modems and serial printers fall into this category.

The type of serial cable you use depends on the type of equipment you are connecting. Table 10.2 explains.

		Connection	Cable Type

		DTE <-> DCE	straight modem cable
		DTE <-> DTE	null modem cable

			Table 10.2. Types of Serial Cable.

Reading.

UNIX System Administration Handbook (2nd Ed.) Chapter 8.

Purpose.

An optional reading that explains in more detail the ins and outs of RS-232 cables. This section will not be examined and is simply included for reference.

Terminal Configuration

For a dumb terminal to work correctly it must be configured properly. Characteristics of a dumb terminal that need to be configured include

bits per second,
The speed at which information can be transferred. Typical values (for today) range from 9600 bps up to 38,400 bps.
parity,
Typically be set off.
duplex,
This should be set to full and signifies that data can be transferred in both directions simultaneously.
auto linefeed,
Should be turned off. The end of a line under UNIX is signified by a newline character. MS-DOS and other systems use a combination of a newline and a carriage return character.
data bits, and
Suggested values are either 7 or 8 with 7 being the preference.
stop bits.
With 7 data bits use 2 stop bits. With 8 data bits use 1 stop bit.

In the case of purpose built dumb terminals, configuration will generally be performed by setting dip switches. In the case of a personal computer and a communications package these settings are set using the options within the communications program.

Testing the Connection

Once all the hardware is configured, connected, and turned on, the next step is to test whether or not you can actually transmit data through the connection. The simplest method to do this is to send some information directly to the device file associated with the device.

For example:

	ls -l / > /dev/tty1
	If the connection is correct and working you should see the output
	 appear on the device.

Be careful when you are choosing device files to send output to.
Sending output to the wrong device can be disastrous.

For example, one of the previous students for this subject tried the command ls > /dev/hda on a Linux box. Under Linux /dev/hda is the device file for the first hard disk. Since he was the root user at the time the operating system happily sent the data to the hard disk.

By writing directly to the device file he bypassed the file system code that places the data on the drive into the correct format. As a result the data on the disk was lost.

Reasons Why It Won't Work

There are a number or reasons why a connection may not work including

incorrect permissions,
For the test to work you must have write permission on the device file. Check the permissions. Typically you have to be the root user to perform the test. Don't change the permissions on the device file to something like world write. This can be a security hole.
the wrong device file,
You've picked the wrong device file and the information isn't being sent to the new device.
incorrect configuration, and
The hardware configuration on the device is not correct. Don't expect the output you send to the device to appear picture perfect. Since you are bypassing the normal mechanism for using the device it may not work 100%.
incorrect cabling.
Is the device turned on? Are you sure you have the correct type of cable.

Exercise 10.1. Beg, borrow or steal a dumb terminal (another PC with a communications program will suffice). Perform all the steps listed above for connecting the terminal to your UNIX machine.

Software

Terminal configuration files is one area in which the diversity of UNIX platforms rears its ugly head. System V based machines will use different configuration files than BSD based systems. Early BSD systems use different configuration files again.

Terminal configuration files can be divided along the lines of their purpose

enabling the login process,
For a terminal to work users have to login. For users to login particular processes have to be executed for each terminal. There are configuration files that control which device files have the login process enabled.
setting line configuration, and
The operating system has to know about and set the characteristics of the serial line that the terminal is connected to, speed, data bits, parity etc.
terminal characteristics.
Different terminals have different keyboard layouts, different capabilities (colour etc) and different special character codes to do things like clear the screen. In order to use a particular type of terminal to its fullest ability UNIX must know about its capabilities. To do this the terminal must have an entry in the database of terminal characteristics that UNIX maintains.

The Login Process

For a user to actually be able to use a terminal it is necessary for the UNIX login process to be running for that terminal.

The login process on a terminal involves

/etc/init,
The first process run on a UNIX system. It is init's responsibility to ensure that each terminal has a getty process running for it.
getty,
Is responsible for setting up the characteristics of the line, displaying the login prompt, getting a username, and executing the login process with the username as an argument.
login, and
Is responsible for logging the user into the system, obtaining and validating the password, and executing the user's login shell.
a login shell.
The user's entry in the /etc/passwd file lists the program to execute as the user's login shell. The shell will in turn execute the appropriate startup scripts.

When the user quits their login shell the init process picks up this event and will rerun the getty process for that terminal.

The SysV Login Process

Under SysV the init process is controlled by the /etc/inittab configuration file (the format of /etc/inittab is discussed in the previous section). The inittab file must have an entry for each terminal that requires a getty process. Typical entries look like

	c6:23:respawn:/etc/getty 9600 tty6
	c7:23:respawn:/etc/getty 9600 ttys1

The getty process here takes two arguments a label used as a lookup value for the file /etc/gettydefs, and the device name for which the getty process is being executed. The /etc/gettydefs file is used for setting various line options. /etc/gettydefs has the following format

	label#initial flags#final flags#login prompt#next label

Table 10.3 explains each field from the gettydefs file.

	Field		Purpose

	label		the "key" to the particular entry
	initial flags	flags to set before running the login process
	final flags	flags to set after running the login process
	login prompt	the prompt to be displayed by getty before running login
	next label	points to the next label to use if this one is not valid (used to cycle through baud rates for modems)

		Table 10.3. Fields for the /etc/gettydefs file.

The process for enabling logins for a new terminal on a SysV system includes

have an entry in gettydefs file,
The system should come with most of the standard settings required for normal use so a new addition is usually not required.
add a new entry to the inittab file to start the getty process for the terminal, and
force init to reread the inittab file.
The new getty process will not be started until init reenters the specified run levels. Under SysV this is typically done using the command init -q

The BSD Login Process

UNIX versions based on pre 4.3 BSD use a different format than subsequent versions. In this section discussion will be restricted to the later versions. From 4.3 on BSD uses a single configuration file /etc/ttys with the following format.

	device program terminal_type status


	Field	Purpose

	device		device file terminal is connected through
	program		program to be run if terminal is on (usually getty)
	terminal_type	typically the name of a terminal type listed
			 in /etc/termcap
	status		a list of key words separated by spaces,
			 keywords include
				on     should run program
				off    ignore the entry
				secure root logins are allowed on
					 this terminal

		Table 10.4. Format of the BSD /etc/ttys File.

Line Configuration

Every terminal connected to a UNIX machine has an associated terminal driver process. This process maintains

a list of characteristics about the current terminal, and
a list of special characters and how they should be handled.

Probably the most common complaint from users is that when they hit particular keys the terminal doesn't do what is expected. Hitting the backspace key might produce a weird character or the cursor keys might not work under vi. These problems are caused by the terminal driver not being configured properly.

	The tty command is sometimes useful.  It displays the name
	of the device file being used by the current terminal.

Initially these settings are set up by the system from the entries in the system's terminal configuration database. The stty command can be used to view and modify these settings.

Table 10.5 lists some of the terminal characteristics and Table 10.6 lists some of the special characters. To view the current settings try stty -a (the command might be stty all or stty everything depending on your system).

For example:

	bash > stty -a
	speed 9600 baud; rows 24; columns 80; line = 0;
	intr = ^C; quit = ^\; erase = ^?; kill = ^U; eof = ^D; eol =
	  < undef>;
	eol2 = < undef>; start = ^Q; stop = ^S; susp = ^Z; rprnt = ^R;
	  werase = ^W;
	lnext = ^V; flush = ^O; min = 1; time = 0;
	-parenb -parodd cs8 hupcl -cstopb cread -clocal -crtscts
	-ignbrk -brkint -ignpar -parmrk -inpck -istrip -inlcr -igncr
	  icrnl ixon ixoff
	-iuclc -ixany -imaxbel
	opost -olcuc -ocrnl onlcr -onocr -onlret -ofill -ofdel nl0 cr0
	  tab0 bs0 vt0 ff0
	isig icanon iext


		Option	Meaning

		n		bits per second
		rows n		lines to the screen
		columns n	columns on the screen
		oddp		odd parity
		evenp		even parity
		-parity		no parity

			Table 10.5. Characteristics affected by stty.


Symbolic Name	SysV default	BSD default	Meaning

ERASE		#		^H	erase one character of input
WERASE		N/A		^W	erase one word of input
KILL		@		^U	erase entire line
EOF		^D		^D	end of file
INTR		^?		^C	interrupt current process
QUIT		^\\		^\\	kill current process with core dump
STOP		^S		^S	stop output to screen
START		^Q		^Q	restart output to screen
SUSPEND		N/A		^Z	suspend current process

		Table 10.6. Special Characters.

In these tables you will see character combinations like ^H and ^?. The ^ symbol is used in this case to signify the control key. So ^H could be rewritten CTRL-H.

A useful option of the stty command is sane. Entering stty sane when the terminal is behaving strangely will solve many problems.

It is possible to use I/O redirection to affect the settings of terminals other than the one you are currently using. Which form of I/O redirection (input or output) you use depends on your system. For BSD redirect the output of stty. For SysV redirect the input. (This will only work if you have the correct permissions on the device file associated with the terminal.

Exercise 10.2. Use the stty program to examine the current status for your terminal. Change the values for EOF ERASE and INTR experiment using the system with the changes.

Exercise 10.3. Change the values for rows and columns and observe the difference.

Exercise 10.4. Complete the process of connecting the terminal you started in question 10.1.

Exercise 10.5. If you are using a PC and a communications program you should be able to change the type of terminal you are using as part of the programs options. Change the type of terminal you are using. How does it effect using the terminal? Modify the appropriate files so that the system uses the correct entry from the terminal configuration database.

Terminal Characteristics

Different terminals have different keyboard layouts, escape codes and capabilities. For example one terminal will use one combination of characters to signify clearing the screen while another terminal will use another combination of characters.

If programs that wish to be able to clear the screen want to work on different terminals they must be able to find out how each terminal performs the operation. Under the UNIX operating system programs discover this information using

the TERM (term if you're using csh) environment variable, and
a system specific terminal database.

The shell variable TERM is usually initialised when a user first logs in. It will hold a unique identifier that signifies the type of terminal being used. This identifier is used to access the information about the terminal from the system's terminal database.

If the TERM variable is set incorrectly or the terminal does not have an entry in the terminal database problems likely to occur include

keys not performing the expected task,
Hitting the backspace key doesn't do anything or displays a weird key combination.
screen output not being written properly, or
The screen not scrolling properly, unexpected colours or characters are appearing.
programs not working properly.
Full screen programs, vi for example, make use of special characteristics offered by most terminals. If the particular terminal you have doesn't have an entry in the terminal characteristics file it can't make use of these special characteristics.

It is the responsibility of various startup files (typically /etc/profile) to make sure that the TERM variable is initialised to the correct value.

For example:

	The following is an example of how the TERM variable might be set.
	if [ 'tty' = /dev/tty1 ]
	then
	  TERM=vt100
	elsif [ 'tty' = /dev/tty2 ]
	then
	  TERM=tvi912b
	else
	  TERM=console
	fi
	On this system the terminal connected to /dev/tty1 is a vt100 so that 
	is the value TERM is set to.  The terminal on /dev/tty2 is a tvi912b 
	and it assumes that any other type of terminal is a console.  The tty 
	command used here returns the device file used by the current terminal.

Once the TERM variable is set processes can use its value to access information in the terminal database. SysV and BSD based systems use different terminal databases.

Exercise 10.6. Make up some name for a terminal, e.g. myterm. Set the TERM shell variable to this value. Attempt to use the vi editor. What happens?

The SysV Terminal Database

The SysV terminal database is called terminfo. It is a collection of separate binary files. Each different terminal type will have its own separate binary terminfo file that describes its characteristics.

terminfo files are usually stored under the /usr/lib/terminfo directory. The terminfo sub-directory is further divided into directories with single character names. The actual terminfo files are stored under the separate directory corresponding to the first letter of the terminal identifier.

For example:

	The terminfo file for the terminal type vt100 is stored in the 
	directory
		/usr/lib/terminfo/v/vt100
	The file for the terminal type tvi912b is in 
		/usr/lib/terminfo/t/tvi912b

terminfo files are compiled from a terminfo source file (an ASCII file) using the command tic. The format of terminfo source files will not be discussed here.

Some of the commands used by the terminfo database are listed in Table 10.7.

	Command		Purpose

	tic		compile a terminfo source file
	infocmp		list the source for a compiled terminfo entry
	infocmp -C	translate from a terminfo file to the termcap
			 format used by BSD systems
	captoinfo	translate from termcap format to terminfo format

		Table 10.7. terminfo Related Commands.

The BSD Terminal Database

BSD systems use one central text based file as the terminal database. The file is /etc/termcap. It contains colon delimited entries for each type of terminal the system recognises.

The first field of every entry is a list of terminal names (separated by |) by which the system recognises a particular terminal. These names are what appears as the value for the TERM variable and is used by the system to look up an entry.

The rest of the entry for a terminal consists of various options that describe the way in which the terminal works. The various options will not be discussed here. They are described in the manual pages for the system if needed.

It is advisable to put the entries for the most used terminals on your site at the front of the termcap file to speed searching.

For example:

	vt100|dec-vt100|vt100-am|vt100am|dec vt100:\
	:do=^J:co#80:li#24:cl=50\E[;H\E[2J:sf=2*\ED:\
	:le=^H:bs:am:cm=5\E[%i%d;%dH:nd=2\E[C:up=2\E[A:\
	:ce=3\E[K:cd=50\E[J:so=2\E[7m:se=2\E[m:us=2\E[4m:ue=2\
	  E[m:md=2\E[1m:mr=2\E[7m:mb=2\E[5m:me=2\E[m:is=\E[1;24r\
	  E[24;1H:if=/usr/share/tabset/vt100:\
	:rs=\E>\E[?3l\E[?4l\E[?5l\E[?7h\E[?8h:ks=\E[?1h\E=:ke=\E[?1l\E>:\
	:ku=\EOA:kd=\EOB:kr=\EOC:kl=\EOD:kb=^H:\
	:ho=\E[H:k1=\EOP:k2=\EOQ:k3=\EOR:k4=\EOS:pt:sr=2*\
	  EM:vt#3:xn:\:sc=\E7:rc=\E8:cs=\E[%i%d;%dr:

Exercise 10.7. Determine the type of terminal you are using and examine the entry for your terminal that is stored in your system's terminal database files. (On SysV based machines you'll have to use one of the commands listed in Table 10.5 to obtain the ASCII source.)

Conclusions

The process of adding terminals to a UNIX box is a reasonably straight forward process that can be plagued by strange and mysterious problems. Hopefully this section has introduced you to

the hardware and software processes for adding terminals to a UNIX machine,
the BSD and SysV configuration files for terminals,
the commands tty and stty, and
some of the problems that can arise while connecting a terminal.

The steps involved in connecting a terminal are

choose a port on the computer,
obtain the correct cable to connect the terminal to the port,
configure the terminal,
make and test the connection,
ensure a correct entry for the terminal in the systems terminal configuration database,
ensure that the TERM shell variable is set correctly,
start the login process for the port.

Review Questions

10.1. List the different types of configuration files used in controlling a terminal.

10.2. List the specifics of the configuration files for the terminals on your system.

10.3. List and describe the hardware steps involved in connecting a terminal to a UNIX machine.

10.4. Describe the steps involved when a user logs onto a UNIX machine.

Previous | Next

David Jones (author)
Chris Hanson (html 08/09/96)