"tty" is an abbreviation for "Teletype". The first terminals were Teletypes (like remotely controlled typewriters). See subsection Teletypes. A list of Linux devices (the stuff in the /dev directory) may be found in "Linux Allocated Devices" which should be included with kernel sources. It "describes" what each device used for in only a word or two but doesn't tell you how to use them.
The computer considers each serial port to be a "device". It's sometimes called a terminal device since at one time terminals were the most common use for a serial port. For each such serial port there is a special file in the /dev (device) directory. /dev/ttyS0) is the special file for the serial port known as COM1 in the DOS/Windows world.
To send text to a terminal you may redirect standard output of some command-line command to the appropriate special file. For example typing "echo test > /dev/ttyS1" at the command prompt should send the word "test" to the terminal on ttyS1 (COM2) provided you have write permission on /dev/ttyS1. Similarly, typing "cat my_file > /dev/ttyS0" will send the contents of the file my_file to COM1 (ttyS0).
Pseudo terminals are pairs of devices such as /dev/ptyp3 and /dev/ttyp3. There is no physical device directly associated with either of them, not even a serial port connector. But if a program treats ttyp3 like it was a serial port, what is read and written to that port appears on the other member of the pair ptyp3 which another program uses to read and write to. Thus two programs talk to each other via this method and one program on ttyp3 thinks it's talking to a serial port. It's something like a "pipe" between these two tty's.
For talking to ttyp3, any program designed to talk to a serial port will do. But for the other program that talks to ptyp3, it must have been specially written to talk to it. It's mainly programmers that must concern themselves with pseudo terminals and most users don't need to worry about them.
Here's an example: If someone connects via telnet to your computer over a network (you are a telnet server), the part of the telnet program handling this connection on your computer may wind up connected to the pseudo terminal ptyp2. A getty program should be running on the corresponding ttyp2. Getty thinks it's talking to a terminal. When telnet gets a character from the remote client, it goes thru ptyp2 to ttyp2 to getty which then sends back "login:" routed via ttyp2, ptyp2, your server telnet program, and then out to the network back to the client. Here the login program and the telnet server program talk to each other via a "pseudo terminal". Note that there is no pseudo terminal used on the client computer, just telnet. Of course the server allocates a pseudo terminal (on the server) for each client.
In X Window, the terminal emulator programs (such as xterm) use pseudo terminals. Ham radio programs under Linux also use them. By using certain application software, it is possible to have 2 or more pseudo terminals attached to the same physical serial port.
For a pseudo terminal pair such as ptyp3 and ttyp3, the pty... is the master or controlling terminal and the tty... is the slave.
There are only 16 ttyp's: ttyp0-ttypf (f is a hexadecimal digit). To get more pairs, more letters such as q, r, s are used instead of p. For example the pair ttys8, ptys8 is a pseudo terminal pair. Later on, even more letters were added so as to allow even more pseudo terminals. And when z was reached, they wrapped around to a. This is confusing but old habits are difficult to change. Today Linux allow say ttyp189 but it's not used. The device file system, which was abandoned in 2004, would have used tty/s189. Be sure not to type say ttys2 if you mean ttyS2 (a real serial port).
The master and slave are really the same "port" but the slave is used by the application program and the master is used by a network program (or the like) which supplies (and gets) data to/from the slave port. The program using the slave port can run "as is" since it thinks it is talking to a serial port.
Unix98 pseudo terminals (available on Linux) is more advanced than the above but the basic concepts are the same (only the device names and methods of creating them change). It creates pseudo terminal devices on request so there is no need to check if the pseudo terminal you might want to use in in use. By opening /dev/ptmx a new pseudo terminal pair is created. The master doesn't show up as a device but is just a file descriptor number passed to the computer program that opened /dev/ptmx. But the slave is put into the /dev/pts directory: for example" /dev/pts/3.
The /dev/pts directory is considered to be a file system of type devpts and appears in the lists of mounted filesystems. While the "file" /dev/pts/3 looks like an entry in the now obsolete device filesystem, /dev/pts Is really a wholly different kind of filesystem.
See the Linux manual pages "pty" and "pts" (Unix 98 style) for more details. For programmers there's the man-page openpty/forkpty (either name displays the same man-page) which assumes that you already understand pseudo terminals. There is a usr/include/pty.h file for use by programmers. In earlier versions of Linux there was a pty.o module, but it now seems that it's been built into the kernel. Here's an example of some options available when you are compiling a Linux 2.6 kernel:
CONFIG_UNIX98_PTYS=y CONFIG_LEGACY_PTYS=y CONFIG_LEGACY_PTY_COUNT=256
/dev/tty stands for the controlling terminal (if any) for the current process. To find out which tty's are attached to which processes use the "ps -a" command at the shell prompt (command line). Look at the "tty" column. For the shell process you're in, /dev/tty is the terminal you are now using. Type "tty" at the shell prompt to see what it is (see manual pg. tty(1)). /dev/tty is something like a link to the actually terminal device name with some additional features for C-programmers: see the manual page tty(4).
N stands for an integer. One use of these in Linux is with the ISDN driver package: isdn4linux. The ttyIN is something like ttySN but it emulates a modem and can be given modem commands.
In Linux the PC monitor is usually called the console and has several device special files associated with it: vc/0 (tty0), vc/1 (tty1), vc/2 (tty2), etc. When you log in you are on vc/1. To go to vc/2 (on the same screen) press down the 2 keys Alt(left)-F3. For vc/3 use Left Alt-F3, etc. These (vc/1, vc/2, vc/3, etc.) are called "virtual terminals". vc/0 (tty0) is just an alias for the current virtual terminal and it's where messages from the system are sent. Thus messages from the system will be seen on the console (monitor) regardless of which virtual terminal it is displaying.
You may log in to different virtual terminals and thus have a few different sessions with the computer going on at the same time. Only the system or the root user may write to /dev/vc/0 to which /dev/console is sometimes linked. For more info on the console see The Linux Console.
The /dev directory comes supplied with many device special files. If you need something that's not there you may try to create it with the "mknod" command. See the manual page ttys(4) for how to do this for serial ports. To use
mknod you must know the major and minor device numbers. You might be able to infer the numbers you need by using the "
ls -l" command in the /dev directory. It will display the major and minor numbers of existing special files.