THE GNU BOURNE AGAIN SHELL (BASH): USEFUL COMMANDS AND KEYS

(C) 2004 Daniel Brodzik

This document is released under the terms of the GNU Free Documentation License as published by the Free Software Foundation, either version 1.2 or (at your option) any later version, with no front-cover texts, no back-cover texts, and no invariant sections.

INTRODUCTION

This is a guide to using the command line under a shell called GNU BASH. The GNU Bourne Again SHell, or 'bash', is the most commonly-used shell under most Linux-based operating systems. It has numerous features to help you out when you're typing at the prompt. I'll first explain some keys that may be useful to you. After that, I'll explain some useful commands. I hope this is helpful to those who are completely new to GNU or Linux.

Because of graphical desktop environments for Linux like KDE and GNOME, a working knowledge the command line is optional, but I highly recommend knowing how to do at least basic tasks on the command line under Linux because the command line is more flexible than a GUI could ever be. It is also often faster. Want to erase all files that end in '.wav'? Type 'rm *.wav'! There are more examples to this. If you are afraid of DOS prompts because of their user-unfriendliness, prepare to be amazed at how much easier the Linux command prompt is to use than the DOS prompt. I liked the DOS prompt, but I love the Bash prompt! Another reason to learn the command line is that many useful graphical utilities under Linux are really just front-ends to command-line programs. Learning the command line can give you better control of your computer.

Note that this guide does not explain the BSD C-shell (csh; included with many Linux distributions as well as BSD), ash, the Bourne Shell (sh; found on most older and non-Linux Unix systems), zsh, or other shells; it just explains Bash. However, most shells use a similar syntax to most commands, provided the commands in question are not built-ins.

This guide first explains some useful keystrokes in Bash, followed by most useful commands.

KEYSTROKES:

Up and down

Browse the history by going back one command (up) or ahead one command (down). Note that if you edit any commands as you're going through the history, then the change(s) WILL stay in effect until you press Enter. For example, say you edit the third command going up, but instead of pressing Enter, you continue moving through the history. In DOS, this would abandon your changes, but in Bash, this will NOT abandon your changes. (That is, if you 'turn around' and go the other way in the history, then, when you get to the command you edited, your changes will still be there UNTIL YOU PRESS ENTER!)

Home; End; Backspace; Delete; Left; Right

These keys behave as they would in most word processors.

Alt-Backspace

This erases one 'word' to the left. An alternate is Esc, Backspace (press Esc, release it, and press Backspace).

Insert,Ctrl-[character]

Allows you to insert control characters. First you press Insert, and then type your control character.

Tab

This is one of my favorites. It acts like an auto-complete for filenames. If you have a long filename to type, then type the first few characters of it and press Tab. One of two things will happen. Either the rest of the filename will be typed automatically, or it will beep and possibly type some of it. The first case is easy to understand, but the second case needs a little explanation. The second is caused by more than one filename starting out with the same character(s). Bash will type the part until the the part in which the difference occurs in the filename and beep. If you press Tab once or twice again, then it will list the filenames that could fit. This also works for typing the commands themselves. If there is only one file in the directory you're referencing, all you have to do is press Tab without typing the name or any part of it.

Shift-Page Up; Shift+Page Down

These are two other favorite keystrokes of mine. These scroll the screen backwards and forwards, allowing you to see the output of commands that scrolled off-screen or verbose output that was too, well, verbose to fit on-screen. This even works after a 'clear' command has been typed. This does not work after switching virtual terminals, however.

Alt-.

Did you just type a command and want to use the last or other recently-used filename/pathname in the next command? Simply type the new command up to the point at which you wish to reuse the filename/pathname, and press Alt-. until you get to the filename you want. Useful, isn't it?

Ctrl-D

Do you want to log out or exit the shell right away? Push Ctrl-D to exit BASH. This types the 'logout' or 'exit' command for you depending upon whether the current shell is a log-in shell or not, respectively. The origin of this keystroke is that ASCII code 4 (Ctrl-D) is the end-of-file character for text files in Unix and Unix-like systems, including Linux. You only need to know the origin of this if you're a programmer (or are planning to be a programmer) or you wish to use the keyboard to type stdin 'file' input for a command directly. (Each operating system type uses a different end-of-file character for text files. For GNU, Linux, and other Unix-like operating systems, it is the Ctrl-D character, ASCII code 0x04. In MS-DOS, MS-Windows, and other DOS-based operating systems, it is Ctrl-Z, 0x1a. I don't know for sure what the EOF character is on a Macintosh. On all modern operating systems, the end-of-file character is optional, but it is still used for using direct keyboard input as standard input (stdin).)

Alt-P; Alt-N

These keystrokes are used to search the history for a certain string. To use this searching functionality, press Alt-P, type in what you're looking for, and press Enter. If what shows up is not what you're looking for, press Alt-P and then Enter to search again. If you accidentally went too far back, use Alt-N to search the other way.

COMMANDS:

Here are some useful commands at the bash prompt. These will be organized by 'normal users' and 'root only'. To get help on a command, there are some things you can type, substituting the command for command.

• command --help

• command -h

• man command

• info command

The first two may cause part of the help to scroll off-screen. To counter this, add '| more' or '| less' to the end of the command. I prefer '| less' because it lets you go up and down as you wish, while '| more' only lets you go forward. Less also has a search function, in addition to many more features. NOTE THAT ALL COMMANDS AND FILENAMES UNDER LINUX AND OTHER UNIX SYSTEMS ARE CASE-SENSITIVE!

Normal Users:

uname

Tells you the Unix name. This includes the kernel name (which is probably Linux), the kernel version number, the kernel build time in your local time zone, and other stuff. This is kind of like Micro$oft Windows' ver command, only more verbose. Type 'uname -a' to see everything.

history

Displays your command history. 'history -c' clears the history.

clear

Clears the console screen.

cd

Changes directories. If a directory is specified, then it goes to that directory. If '..' is specified, then it goes to the parent of the current directory. If nothing is specified, then cd takes you to your home directory.

pwd

This built-in shell command will display your present working directory.

mkdir

Makes a directory.

rmdir

Removes an empty directory.

ls

Used for the same purpose as DIR under DOS. Lists the contents of the current directory or the one specified on the command line. Here are some examples of useful command line switches:

• ls – Shows a list of only filenames and directory names.

• ls --color – Forces 'ls' to color-code the file and directory names by type.

• ls -lh – Shows the long format (-l) and gives sizes in 'human-readable' format (-h) instead of just bytes. 'Human-readable' means sizes like 23k, 4M, and 1.2G instead of long numbers.

dir

Same as 'ls' except that backslashes are shown where you need to type them on the command line in filenames that need them. Filenames with spaces or apostrophes are mainly the ones that need these.

df

This nifty little command shows a nicely-formatted table of all the mounted file systems, how much is used on each file system, and how much space is free. Use the -h option to tell it to use 'human-readable' sizes (like 45M, 256k, and 2.3G) instead of long numbers.

mount

Mounts a disk or disk image. Some common variants are:

• mount /cdrom -- Mount the CD-ROM drive

• mount /floppy -- Mount the floppy drive.

Note that on some Linux distributions, these commands should be:

• mount /mnt/cdrom – Mount the CD-ROM drive

• mount /mnt/floppy – Mount the floppy drive

umount

Unmounts a disk or disk image, providing the device is NOT in use. Add a 'u' to the beginning of the command examples in the 'mount' section above for some examples. Note that this command is not 'unmount' but 'umount' without an 'n' after the first 'u'!

nano

Starts the GNU nano text editor. GNU nano is a replacement for the non-free (as in freedom) popular text editor Pico. (For some odd reason, Pico's license agreement allows distribution in source code, but not binary, form.) Specify a filename on the command line to open and edit the text file. If you installed the 'ne' package for your distribution, then I'd recommend that over nano.

emacs

Starts GNU Emacs, a fancy text editor, to say the least. :) This editor is very popular in the Unix world, ranking second only to vi. If you want to be like the prestigious Unix people to the point of using a popular text editor, use either Emacs or vim (Vi Improved), but stay away from nvi if you are a new user! In text mode, the menus are accessed through F10. In the X Window System, Emacs has a menu bar you can click. Emacs even has an easy-to-follow tutorial and full documentation within itself! To top it all off, Emacs has a LISP interpreter embedded in it, so you can write programs for it! This is one cool editor!

vi

Vi is a very terse but very powerful editor. It was one of the first screen-oriented editors available for Unix. (TECO was available for other platforms for a while before vi came out, and it was screen-oriented.) If you are a new user, please install Vim (it comes with most non-live-CD Linux distributions), which is Vi Improved. Vim has a very good interactive tutorial. In fact, after one hour in the tutorial, I already felt quite comfortable with it. If you have Vim installed, you should be able to type "vi" to access it; if not, use "vim" instead. Vim has a GUI, so you should be able to use "gvim" under the X Window System to get a graphical interface. I would consider writing a tutorial on using Vim, but the tutorial that comes with Vim, which can be accessed by typing "vimtutor" at the command prompt, is very good.

less

This is a file viewer and text pager. You can scroll through the output of a command by typing 'command | less', or you can use it to view text files by typing 'less filename', replacing command with the command you wish to scroll through the output of or filename with the text file's name.

cp

Copies files. Multiple files can be specified on the command line, like this:

cp zork gork bork /home/user

If you want to copy a whole directory structure, use the '-r' switch, like this:

cp -r directory ~/foo

mv

Moves or renames files. Like Micro$oft Wincrap, only the directory entries get moved and/or renamed if the source and target are on the same file system. This command uses the same syntax as 'cp', except that you don't need to use '-r' to move a directory.

ln -s

This is one of my favorite Unix features. This command creates a symbolic link. Coming from using Wincrap, I am familiar with shortcuts. Think of this as going one step further than a shortcut. Whereas a shortcut under Micro$oft Wincrap takes you to the original under Windows Explorer, a symbolic link actually functions as the original item when referred to in a path, and any writes go to the original item in its original location. For example, if you created a shortcut under Windows to C:\WINDOWS\SYSTEM under C:\ and called it 'SYSTEM', then references to C:\SYSTEM or even C:\SYSTEM.LNK would not refer to C:\WINDOWS\SYSTEM; they'd refer to the shortcut file itself. Under Linux, however, creating a link in ~/ to /usr/local called 'local' would make all references to ~/local refer to the directory /usr/local. In fact, if there was a file called 'foo' in /usr/local, then ~/local/foo would be a valid path! (By the way, ~, when used in a Unix pathname, refers to the current user's home directory.) This applies to everything except deleting the link; 'rm'ming a symbolic link will delete the link and not the original file/directory.

rm

Deletes a file, but not a directory (unless used with -r).

rm -r

Deletes a directory and ALL its contents. Same as DELTREE under DOS, except that rm does NOT prompt first! BE CAREFUL!

touch

What 'touch' does depends on whether the filename you give it exists or not. If the file exists, 'touch' updates the time and date stamp to 'today, right now'. In other words, this has the same effect as 'copy /b filename+,,' under DOS. However, if the file does not exist, 'touch' creates a zero-byte file with the name you gave 'touch'. This command is often used by the operating system tools for 'stamp' files and lockfiles.

su

After typing this, you'll be asked for the root password. If you get it right, then you'll become root (the 'superuser'). To exit su, type 'exit'. Note that if you type a user name after 'su', then you will be prompted for that user's password instead and, if the password is correct, you'll become that user.

exit

Exits a shell. Also works as a replacement for 'logout'.

logout

Logs you out. I usually type 'clear;logout' to clear the screen when logging out.

clear

Clears the screen.

startx

Starts an X Window System (X11) GUI session. You don't need this if you're using a display manager like kdm, xdm, gdm, or wdm. In other words, if you see a graphical login prompt when you start up, you don't need startx.

killall

This kills process(es) by name. Use '-s KILL' to forcefully terminate the process(es). Note that unless you are root, you can only kill your own processes.

kill

This ends a process. You must specify the PID(s) of the process to be killed. Use '-s KILL' to forcefully terminate the process(es). Note that unless you are root, you can only kill your own processes.

top

This will show a table of processes and some system statistics. It's pretty interesting....

pidof

This reports the Process Identifier(s) (PID(s)) of a process when given the name.

ps ax

This shows a list of all the running processes.

command &

When you type a command and end the command line with '&', the process will run in the background, and you will be returned to the command prompt. You will be given a status line, like this:

[1] 665

This means, in the example, that the job number is 1, and the PID is 665. The job number will be useful to you in the next two commands. Note that in some console-based programs, you can also press [CTRL][Z] to put the job in the background. If you do this, however, the program will be stopped. The next two commands will make it go again.

fg

This brings a program that's in the background to the foreground again. To use this command, enter 'fg' followed by the job number you got when the program went into the background.

bg

This pauses/resumes a program that's in the background. To use this command, give the job number that you received when the program went into the background.

whoami

Tells you who you're logged in as.

whereis

Would you like to know where a certain program/command's binary and manual page are stored? Type 'whereis command'.

Commands most useful as the superuser (root):

Note that some of these, like 'chmod', are also useful as a normal user if you use them on your own files.

poweroff

As you may have guessed, this is used to shut down the computer safely. You could also type 'init 0' to get exactly the same effect. If you have APM or ACPI support installed, then this may turn off your computer automagically. This is great for virus scanners, etc.

halt

This command is equivalent to 'poweroff'.

reboot

Reboots the computer. You can also use 'init 6' to get exactly the same effect.

fsck

This is Linux's equivalent to scandisk. This works on FAT12, FAT16, FAT32, ext2, ext3, and more. Since ext3 is a journaling file system, fsck is almost never needed to fix problems on ext3 disks/partitions. Anyway, here are some command-line examples:

• fsck /dev/hda1 --- checks the first partition of the first hard drive for errors if it's not marked clean

• fsck -f /dev/hda3 --- Forces a filesystem check on the third partition on the first hard disk

• fsck -fy /dev/fd0 --- Forces a filesystem check on the first floppy drive and automatically fixes any problems found without prompting you first.

WARNING!!! RUNNING FSCK ON A FILE SYSTEM THAT IS MOUNTED READ-WRITE MAY, AND PROBABLY WILL, CAUSE SEVERE FILE SYSTEM DAMAGE!! If you forget this and try to fsck a mounted file system, fsck will give you a warning before it does anything. To unmount a file system, use 'umount mount-point'. To mount all your automatically-mounted file systems again, type 'mount -a'. To remount an already-mounted file system read-only, type 'mount -o remount,rw mount-point'. To remount an already-mounted file system read-write, type 'mount -o remount,rw mount-point'. It's recommended (and usually easier) to unmount the file system and mount it again later, but you will need to remount the root filesystem as read-only to check that because you cannot unmount the root...

It is likely that you will have to go into single-user mode to change the access mode of the root file system to read-only. To do this, type "init 1" as root. To get back to normal, you can type "exit" at the command prompt. Before returning to the normal runlevel, don't forget to return your root file system to read-write mode and to run "mount -a" to mount everything else again.

cfdisk

Runs Linux's easier-to-use disk partitioning utility. USE WITH CAUTION!! This program can PERMANENTLY erase data on your disk if you aren't careful!

fdformat

Formats a floppy disk. Unlike DOS and Wincrap, you must also place a filesystem on the disk by using the next command.

superformat

This is supposedly a better version of fdformat. It has many more command-line options.

mkfs.type

Makes a filesystem in either a device (/dev/x) or a file. Type 'mkfs' BUT DON'T PRESS ENTER YET. Then, press Tab twice or thrice to see what filesystems you can make. (Yes, thrice is a word; it means 'three times'.)

chmod

Changes file permissions. The syntax is:

chmod mode file

Mode can be an octal number. It can also be in the form of the user (u for user, g for group, o for others, a for all), a + or - sign, and then an action (r for read, w for write, x for execute, X to make it so that execute is changed only if the file is a directory OR if the execute permission is already set). A common code for the octal number is 666 to allow ALL users to read and write the file. File can be a wildcard. For more information on file permissions, read my file system guide, which is another one of my Linux guides.

Examples:

• chmod 666 fred.txt --- Enables everyone to read and write to 'fred.txt'

• chmod ug+rwx jim --- Enables the file owner and members of the group that's associated with the file 'jim' to read, write, and execute it.

• chmod u+w,go-w,a+x sheila.sh --- Enables the owner to write to 'sheila.sh', prevents others (except root) from writing to it, and allows everybody to execute it. 'sh' is a common extension used for shell scripts that aren't used frequently. (If the shell script has an extension, you must type the extension every time you run the script at the command line.)

• chmod a-r,a+w,a-x barney.foo --- Sets some weird permissions for 'barney.foo': nobody except root can read the file, everybody can write to the file, and nobody can execute the file. This set of permissions is not very common except on some device nodes in /dev. 'foo' is a common metasyntactic variable, and so are these names.

chown

Changes the owner of a file. Syntax is:

chown user:group file

Either the user or the group can be omitted. (Sorry, you cannot omit both.) If the user is omitted, there must be a colon before the group.

adduser

This one adds a user to a group.

useradd

groupadd

These add users and groups. I'm not sure how to use them, so read the manual pages ('man useradd').

modprobe

Loads a kernel module. With -r, it removes a module. You may want to use modconf instead. It's explained later on in this article.

insmod

This is the 'simpler' version of modprobe. Unlike modprobe, it does not automatically load any dependencies.

rmmod

Unloads a kernel module. Again, you may want to look at the section for modconf later on in this article.

lsmod

Do you want to see what modules are loaded? Type this command as root.

MAKEDEV

This command is to be entered in all uppercase. It makes device files that you don't have. To use it, first type 'cd /dev'. Then, type 'man MAKEDEV' to see some of the different device files you can make. Note that the man page for MAKEDEV was written in 1994, and, as such, is "woefully out of date". You don't need this command if you use devfs. (See my file system hierarchy guide to learn more about devfs and why I highly recommend using it. In a nutshelll, Devfs is a way to automatically manage your device files so that you don't have to manually create and delete them; this job gets offloaded to the kernel and a program called devfsd.)

Debian-specific commands:

These commands must be run as root in order to be effective.

dselect

Starts Debian's dselect program, the easy way to add and remove software that comes with Debian.

modconf

This utility is a menu-driven program for configuring modules. Changes take effect immediately and will remain in effect after rebooting. I'd recommend using this over modprobe, insmod, and rmmod if you use Debian.

dpkg

The Debian packaging tool. I use it to install packages that I downloaded from the Internet. If you have a package file (they end in '.deb'), then you can type 'dpkg -i <filename>' to install it. This may cause dependency problems, so if you get any error messages about missing dependencies, run dselect A.S.A.P.! Choose '[S]elect' from the menu, press [Enter] to leave '[S]elect', and select '[I]nstall'. Until you do that, you won't be able to use the software in the package you installed.