01) Linux - Lab Manual

GNU/Linux Lab Manual
for
CSE220 Programming for
Computer Engineering
Kevin Burger
School of Comp. Info. and Dec. Sys. Eng.

Arizona State University
Rev 1.5(a)
1 Sep 2009
Contents
1 Introduction to Unix and GNU/Linux................................................................... 3
1.1 Short History of Unix ................................................................................... 3
1.2 The GNU Project and the Linux Kernel ....................................................... 5
1.3 GNU/Linux Distributions .............................................................................. 5
1.4 Other Unix and Unix-Like Distros ................................................................ 7
1.5 Cygwin ........................................................................................................ 7
2 Logging in to GNU/Linux .................................................................................... 8
2.1 KDE, Konqueror, the Terminal Window ....................................................... 9
2.2 Logging Out ............................................................................................... 11
2.3 Logging In To "research.asu.edu" ............................................................. 11
2.4 Logging Out Of research.asu.edu ............................................................. 12
2.5 Transferring Files with WinSCP................................................................. 12
3 Shells ............................................................................................................... 15
3.1 The Bash Shell .......................................................................................... 15
4 The *nix File System and Common Commands .............................................. 18
4.1 Common Directories in *nix ....................................................................... 18
4.2 Pathnames ................................................................................................ 20
4.3 Commands, Arguments, and Options ....................................................... 20
4.4 Common File and Directory Commands ................................................... 21
4.5 How Do I Know Which Directory I Am In? ................................................. 21
4.6 List Files, File Permissions, The Chmod Command .................................. 22
4.7 Deleting Files............................................................................................. 23
4.8 Creating, Changing, and Removing Directories ........................................ 23
4.9 Filename Globbing .................................................................................... 24
4.10 More on Deleting Subdirectories ............................................................. 25
4.11 Copying Files and Directories.................................................................. 26
4.12 Moving and Renaming Files .................................................................... 27
4.13 Finding a File ........................................................................................... 27
4.14 Displaying a File at the Terminal: Cat and Less ...................................... 28
4.15 Archiving Files ......................................................................................... 29
4.16 Compressing Files ................................................................................... 30
4.17 Hexdump ................................................................................................. 31
5 Getting Help ..................................................................................................... 32
5.1 Man Pages ................................................................................................ 32
5.1.1 Reading a Man Page .......................................................................... 32
5.2 Apropos ..................................................................................................... 36
5.3 Info ............................................................................................................ 36
5.4 Final Notes ................................................................................................ 37
6 Environment Variables, Search Path, Prompt .................................................. 38
6.1 Changing your Prompt .............................................................................. 38
7 Searching Files ................................................................................................ 40
8 I/O and Redirection .......................................................................................... 41
1
8.1 Redirection Example ................................................................................. 41

8.2 Redirection Example ................................................................................. 42
8.3 Redirecting Stderr ..................................................................................... 42
8.4 Sending Output to /dev/null ....................................................................... 43
8.5 Concatenating Files ................................................................................... 43
8.6 Sending Stdout and Stderr to the Same Place .......................................... 44
8.7 Appending to a File ................................................................................... 44
8.8 Pipes ......................................................................................................... 45
9 Processes ........................................................................................................ 46
10 Editing Files ................................................................................................... 48
10.1 Absolutely Painless Introduction to Vim .................................................. 48
10.1.1 Starting VI ......................................................................................... 48
10.1.2 Getting out of VI ................................................................................ 48
10.1.3 Switching Between Editing Mode and Command Mode ................... 48
10.1.4 Other Useful Settings ........................................................................ 48
10.1.5 Moving Around a Character at a Time .............................................. 49
10.1.6 Moving Around a Word at a Time ..................................................... 49
10.1.7 Moving Around the File ..................................................................... 49
10.1.8 Deleting Characters .......................................................................... 49
10.1.9 Copying a Line of Text ...................................................................... 49
10.1.10 Moving a Line of Text...................................................................... 49
10.1.11 Deleting a Line of Text .................................................................... 50
10.1.12 Cutting, Copying, Pasting Multiple Lines of Text ............................. 50
10.1.13 Moving and Copying Text in Visual Mode ....................................... 50
10.1.14 Finding Text .................................................................................... 50
10.1.15 Find and Replace ............................................................................ 50
10.1.16 Undo and Redo ............................................................................... 50
10.1.17 More Help ....................................................................................... 50
10.2 KDE - The Kate Editor ............................................................................. 51
10.3 Other Linux Editors .................................................................................. 51
11 Basic Bash Shell Scripting ............................................................................. 52
11.1 Variables ................................................................................................. 52
11.2 Arithmetic and Relational Operators ........................................................ 53
11.3 Basic Shell Script Example...................................................................... 53
11.4 Command-Line Arguments...................................................................... 54
11.5 Aliases ..................................................................................................... 55
11.6 Displaying Text from a Shell Script.......................................................... 55
11.7 A Simple Backup Script ........................................................................... 55
11.8 Reading From the Keyboard ................................................................... 56
11.9 Capturing Command Output.................................................................... 56
11.10 If Statements ......................................................................................... 56
11.11 For Loops .............................................................................................. 58
11.12 While Loops........................................................................................... 59
11.13 Expr and Substrings .............................................................................. 59
12 Program Development, GCC, and GNU Make............................................... 61
2
12.1 Using GCC .............................................................................................. 61

12.1.1 Putting . In Your PATH...................................................................... 62
12.1.2 Useful GCC Command Line Options ................................................ 62
12.2 Using G++ ............................................................................................... 66
12.3 *nix Libraries............................................................................................ 67
12.3.1 Static Libraries .................................................................................. 67
12.3.2 Shared Libraries ............................................................................... 68
12.3.3 Dynamic Libraries ............................................................................. 69
12.3.4 Linking With a Static Library ............................................................. 69
12.4 Makefiles ................................................................................................. 70
12.4.1 Building a Project the Hard Way ....................................................... 70
12.4.2 Building a Project with a Script File ................................................... 70
12.4.3 Introduction to Make ......................................................................... 71
12.4.4 Make Command Line Options .......................................................... 72
12.4.5 Automatic Variables and Implicit Rules ............................................. 73
12.4.6 Compiler Command Line Options ..................................................... 74
12.4.7 Handling Header Files ...................................................................... 75
12.4.8 Phony Targets .................................................................................. 77
13 Debugging ..................................................................................................... 79
13.1 Text-Based Debugging with GDB............................................................ 79
13.1.1 Launching and Quitting gdb .............................................................. 79
13.1.2 Running Your Program ..................................................................... 79
13.1.3 Listing Source Code.......................................................................... 80
13.1.4 Examining Assembly Language Code .............................................. 80
13.1.5 Examining Data and Variables.......................................................... 81
13.1.6 Examining Memory ........................................................................... 81
13.1.7 Examining Array Variables................................................................ 82
13.1.8 Setting Watches or Automatic Displays ............................................ 82
13.1.9 Changing the Values of Variables ..................................................... 82
13.1.10 Examining Registers ....................................................................... 82
13.1.11 Stepping.......................................................................................... 82
13.1.12 Breakpoints ..................................................................................... 83
13.1.13 Watchpoints .................................................................................... 85
13.1.14 Miscellaneous Gdb Commands ...................................................... 85
13.2 GUI Debugging with DDD........................................................................ 86
1 Introduction to Unix and GNU/Linux

1.1 Short History of Unix
A good history of the Unix operating system is available at Wikipedia1. UNIX was
an operating system2 originally written at AT&T Bell Labs in the late 1960's by
1
2
http://en.wikipedia.org/wiki/Unix.
An operating system is software which manages the resources in a computer system.
Ken Thompson3, Dennis Ritchie4, and others. Thompson had been working on
an earlier operating system project named Multics (Multiplexed Information and
Computing Service) in conjunction with MIT and General Electric. While working
on the project, Thompson had written a game called Space Travel for the GE
mainframe being used for Multics development. After AT&T Bell Labs withdrew
from the Multics project, Thompson began porting his Space Travel game to a
DEC PDP-7 minicomputer at Bell Labs, and in the process began work on a new
operating system he named Unics (for Uniplexed Information and Computing
Service); the spelling was later changed to Unix and when it was owned by AT&T
was written in all capitals as UNIX.
Over time, Thompson, Ritchie, and others at Bell Labs continued developing
UNIX. In 1973 the entire operating system was rewritten in Ritchie's new C programming language, and during this time AT&T began giving the operating system to universities, research institutions, and the US government under licenses,
which included all source code. In 1982, when AT&T realized that they had an
OS on their hands that was worth something, they changed their licensing model
and began selling UNIX as a commercial product without source code.
Meanwhile, at the University of California, Berkeley, Bill Joy and others had begun to develop BSD5 Unix as an alternative to the now-commercialized AT&T
Unix. BSD Unix went on to become immensely popular, especially in the academic world, and Bill Joy eventually left Berkeley to go to work for a small startup
named Sun Microsystems where he created the SunOS version of Unix which
eventually became Sun Solaris.
Throughout the next two decades, many companies and organizations produced
their own versions of Unix, usually under license from AT&T or from the Berkeley
branch. This led to a proliferation of differing operating systems that were called
Unix. For example, at one time, one could acquire "Unix" from various entities
such as: AT&T (System V), UC-Berkeley (BSD Unix), IBM (AIX), SGI (IRIX), Microsoft (Xenix), SCO, HP (HP-UX), and others. During the late 1980's, it was
thought that Unix would become the dominant operating system among PC users, but the divisions among the participants and the wrangling for control of Unix
created an opportunity for Microsoft Windows NT to fill the gap in the market, and
for Microsoft Windows to, instead, dominate the industry6.
Eventually, while getting out of the computer business, AT&T washed their hands
of Unix and sold all rights to Novell, which eventually transferred control to the
Open Group. The Open Group is not a company, but rather is an industry con3
4
5
6
http://en.wikipedia.org/wiki/Ken_Thompson_(computer_programmer).
http://en.wikipedia.org/wiki/Dennis_Ritchie. Dennis Ritchie is the creator of the C programming language. It was based
on an earlier programming language named B that was created by Ken Thompson.
Berkeley Software Distribution.
Supposedly. Sigh.
sortium7 which sets standards for Unix. The Open Group published the Single
Unix Specification (SUS) in 2002 which is a family of standards that defines
what Unix is, is not, and what OS's qualify for the name. The SUS is now maintained by the Austin Group for the Open Group.
1.2 The GNU Project and the Linux Kernel

In 1983, after AT&T "closed" Unix by commercializing it and restricting access to
the source code, Richard Stallman founded the GNU project8, whose goal was to
provide a free, open-source Unix-like operating system which would replace
Unix. The project was named GNU for GNU's Not Unix. However, the most import part of the operating system, the GNU kernel9, was not completed very
quickly, and is in fact, not entirely completed yet10.
In 1991, Linus Torvalds, a student at the University of Helsinki, Finland, started
work on his own Unix-like kernel. It was based on a published, open-source operating system called Minix which was created at the University of Vrije, Netherlands, by Andrew Tanenbaum as an OS for teaching operating system principles
to students; MINIX was written for the IBM PC. Linus began making modifications
to MINIX, and his work eventually led to the creation of the Linux kernel. For
whatever reason, and with much luck, Linux became extremely popular and the
Linux kernel11 is now commonly combined with many of the tools and much of
the GNU source code from the GNU project to form an operating system that is
called "Linux."
However, Richard Stallman, the founder of the GNU project, is adamant12 that
the moniker "Linux" is inappropriate as most of what is called the "Linux" operating system is, in fact, GNU code (the "Linux" kernel actually only comprises
around 1.53% of the total GNU/Linux source code). Henceforth, to avoid any
confusion and because Stallman is correct, we will collectively refer to Unix and
GNU/Linux in this document as *nix (if you're not familiar with regular expressions, you will see where the * comes from shortly).
1.3 GNU/Linux Distributions

Since GNU/Linux is free, open-source software, anyone is free to make changes
to the software and redistribute it as long as they follow the licensing model established by the GNU Project and by Linus Torvalds for the Linux kernel. Many
people and companies have done this by starting with the Linux kernel, and adding necessary GNU software to build a working GNU/Linux operating system.
7
8
9
10
11
12
Members include IBM, HP, and other major Unix players.

http://www.gnu.org
Named Hurd.
http://en.wikipedia.org/wiki/Gnu_hurd
http://en.wikipedia.org/wiki/Linux_kernel; http://www.kernel.org
http://www.gnu.org/gnu/ linux-and-gnu.html
Often, additional software packages are added to complete the particular software distribution or distro as these are called. These additional software packages often include Open Office13 which is a free productivity suite with applications similar to Microsoft Office (Open Office includes a word processor, spreadsheet, presentation, database, and graphics drawing program). The quality of
Open Office may not be as high as Microsoft Office14, but it is very good, and additionally it is free (and open source).
Graphics support in most GNU/Linux distros is based on the X Window System15
first developed at MIT in 1984. Modern, open-source Graphical User Interfaces
(GUI's) such as KDE16 and Gnome17,18 are written to work with X as the underlying graphics system while KDE and Gnome make the interface intuitive and
"pretty" for the usermuch as the GUI does in Microsoft Windows and Apple's
OS X. Many GNU/Linux distros include support for both KDE and Gnome, and
you can choose which to run when logging in. I personally prefer KDE, but many
people swear by Gnome. Truth be told19, they do more-or-less the same thing.
Much information about GNU/Linux distros can be found at 20 where there are
links to hundreds, if not over a thousand, different Linux distros. Currently, the
most popular distros seem to include: Ubuntu (based on Debian), Debian, RedHat Fedora, Mandriva, CentOS, SUSE, Knoppix, PCLinuxOS, Slackware, Gentoo, and Puppy.
A lot of distros are based on other distros. For example, Ubuntu is based on Debian, CentOS on RedHat Enterprise Linux, Mint on Ubuntu, PCLinuxOS on Mandrake (Mandriva) etc. In general, the underlying OS is more-or-less the same
(the kernels may be slightly different versions) but the packages included in the
distro may differ. For example, Ubuntu is "like" Debian, but it is packaged so as
to make it easier to install and configure, which is, of course, important for beginners. I personally have been using Debian for a number of years, and have not
seen a great reason to switch, but if you are starting out you might want to look at
Ubuntu; it has become very popular. Support varies among the distros, and with
the larger ones, you are more likely to find online support, especially through online forums.
Another option, if you don't want to install GNU/Linux to your hard drive is to boot
from a LiveCD21 (Live USB22) which is a version of GNU/Linux which is bootable
13
14
15
16
17
18
19
20
21
22
http://www.openoffice.org.
It's buggy, but then again, so is Office.
http://en.wikipedia.org/wiki/X_Window_System. Commonly called X11 or just X.
http://en.wikipedia.org/wiki/KDE.
http://en.wikipedia.org/wiki/GNOME
There are other GUI's for GNU/Linux but KDE and Gnome are the two most popular.
Don't shoot me.
http://distrowatch.com. More info than you could possibly use.
http://en.wikipedia.org/wiki/Live_CD
http://en.wikipedia.org/wiki/Live_USB
off of a CD-ROM disc (USB flash drive). Another, newer, cool option is to use
Wubi which is an installer which will installed Kubuntu23 directly onto one of your
existing Windows partitions. It will modify the PC bootup screen so you can select to boot unto Windows or Kubuntu. The nice thing about this is that it does not
require you to create a separate hard drive partition for GNU/Linux, and if you
decide you do not want Wubi/Kubuntu on your system anymore, all you have to
do is uninstall it like any other Windows application.
It is possible, of course, to build your own GNU/Linux OS by starting with the
kernel and adding GNU source and other packages to that. One such project
teaching you how to do this is Linux from Scratch24 which provides step-by-step
instructions for building a working GNU/Linux OS from source (i.e., this means
you build/compile all of the necessary programs yourself, including the kernel). I
haven't tried it yet, and although it sounds fun, it also sounds like a lot of work.
1.4 Other Unix and Unix-Like Distros

FreeBSD25 is a free, open-source version of the Berkeley BSD version of Unix. It
is similar in functionality to GNU/Linux, but based on a different code-base. Many
people have very good things to say about FreeBSD and it's variants. Most
GNU/Linux programs distributed in binary form will run directly on FreeBSD; others require the use of the FreeBSD GNU/Linux compatibility layer which makes
it appear to the program that it is running on a GNU/Linux OS rather than
FreeBSD. Related OS's are NetBSD and OpenBSD.
1.5 Cygwin
Cygwin26 is a Linux-like environment for Windows. It consists of a DLL (dynamic
link layer; named cygwin1.dll) which acts as a GNU/Linux API (application programming interface) emulation layer. When a program makes a Linux system
call, this API layer traps the call and transforms it into a functionally equivalent
Windows system call. From the program's perspective, it appears to be running
on GNU/Linux, even though it is actually executing on a Windows machine.
In addition to the Cygwin DLL, there are a large number of native GNU/Linux applications which have been ported to Cygwin (native GNU/Linux applications will
not run directly, but rather, must be modified in order to run with Cygwin at the
API layer).
23
24
25
26
Kubuntu is Ubuntu with the KDE GUI environment.

http://www.linuxfromscratch.org/
http://en.wikipedia.org/wiki/FreeBSD
http://www.cygwin.com
To use Cygwin, you would want to install at a minimum the following packages
and tools:
Base
Devel
Doc
Shells
Utils
everything
binutils, gcc-core, gcc-g++, gdb, make
cygwin-doc, man
bash
bzip2, cygutils
It will work best if you let Cygwin install itself where it wants to in C:\Cygwin; the
main thing is to not put it in C:\Program Folders or any other Windows folder
with a space in the folder name.
The default Cygwin shell is Bash which we will be discussing later in Chapter 3.
2 Logging in to GNU/Linux
The machines in the Brickyard BY214 engineering lab are set up to dual boot
both Windows and GNU/Linux (the particular GNU/Linux distribution, that is, "distro" is CentOS27 which is based on RedHat's Enterprise Linux distro). To boot
up to GNU/Linux, turn off the machine, turn it back on, and select CentOS from
the boot window when it appears.
After CentOS boots, the login screen will be displayed as shown below. Your user identifier and password are your ASURITE login identifier and password.
27
http://www.centos.org; It stands for Community Enterprise Operating System.
Historically, the Unix user interface was text-based28 as it was developed in the
early 1970's before graphical terminals were common. Today, most GNU/Linux
distros support a graphical user interface. The two GUI's that are most commonly
used are the K Desktop Environment, KDE29 and Gnome30.
Some distros include KDE but not Gnome. Others support Gnome, but not KDE.
Some include both and give you the option of selecting which one you wish to
use when you log in. In CentOS, You can select which desktop environment you
wish to use through the Session link in the lower left corner of the login screen.
By default, Linux will start with the desktop manager you last used. In the remainder of this Chapter we assume you are using KDE.
2.1 KDE, Konqueror, the Terminal Window

KDE is a desktop environment similar to the Windows or Mac OS X environments. You select the programs you want to run through the
icon in the lower
left corner of the screen, see below.
The two main applications you need to learn to use for this course are the file
manager and the terminal. The KDE file manager is Konqueror and it is
icon and then Home. Konqueror allows you to
launched by clicking on the
view the files in your directories, move/copy/delete/rename files, etc.
28
29
30
http://en.wikipedia.org/wiki/Text-based
http://www.kde.org
http://www.gnome.org; pronounced "gnome" and not "guh-nome".
c
To launch the terminal window (i.e., the shell) click on the
icon, then select
System and then Terminal. Note that you can have multiple terminals running at
one time.
10
2.2 Logging Out

To log out of the system, simply click on the
pop-up menu.
icon and select Log Out from the
2.3 Logging In To "research.asu.edu"

ASU maintains a RedHat Linux cluster with the domain name research.asu.edu.
You can get an account on this cluster by loading your browser and navigating to
the URL www.asu.edu/selfsub. Log in with your ASURITE id and password
when requested. Select the Research Unix Cluster radio button and then click
on the Subscribe to Selected Service Button. (Note that your account will not
be available for anywhere from a few minutes to perhaps one hour).
Once your account is activated, you can remotely log in to this cluster using a
secure shell program31. If you are using one of the campus Windows computers,
the SSH Secure Shell program should be installed. Launch it from the Start
menu. In the dialog box that appears enter research.asu.edu as the host name,
your ASURITE user identifier for the user name, and make sure port number is
set to 22. The first time you connect a dialog box may appear saying something
about the server's host key and the cache. Just click OK or Yes to accept it.
Then, once connected you will be asked to enter your password. Enter your
ASURITE password. Once logged in, you will be at the shell prompt (see Chapter
3 for a discussion of the shell). If you want to access research.asu.edu off campus, another very good free, secure shell program is putty32. I highly recommend
it.
31
32
http://en.wikipedia.org/wiki/Secure_Shell
http://www.chiark.greenend.org.uk/~sgtatham/putty/download.html
11
2.4 Logging Out Of research.asu.edu

To log out of the system, at the command prompt enter the word logout or exit
or press the Ctrl+D (^D) key combination.
2.5 Transferring Files with WinSCP

If you are working at home and you want to transfer files from or to your account
on one of the Linux systems, I recommend a program named WinSCP33. It is a
free, open-source secure FTP program with a very easy to use GUI. Download it
and install it.
When you launch it, you can set up a saved session for research.asu.edu. You
will see dialog box that looks like the following,
Click the New button and the following dialog box will appear,
33
http://winscp.net/eng/index.php
12
In the textfield for Host Name, enter research.asu.edu. In the textfield for User
name enter your ASURITE user identifier. You may enter you ASURITE password in the Password textfield if you do not want to be prompted to log in every
time you connect. Make sure the Port Number is set to 22, and that SFTP is selected for the File protocol. When you click Save you may be presented with a
dialog box that says something about saving your password (this only occurs if
you entered a password). Click OK. You are then asked to give the session a
name, you might use ASU - Research Cluster for example.
Then to connect to research.asu.edu, select ASU - Research Cluster from the
list of saved sessions, and click the Login button. You may get a dialog box that
says something about the server's host key and the cache. Just click Yes to accept it. If everything succeeds the following window will be displayed,
In the left pane of the window are the files on your local machine; the right pane
displays the files in your home directory on the remote machine. You can copy
and move files between these two machines by drag-and-drop. You can also delete files, rename files, create directories, etc.
13
WinSCP can be programmed to automatically upload and download files using

scripts, see 34.
When you are finished transferring files, click Session | Disconnect on the
menu bar. To exit WinSCP click Close on the session list window.
34
http://winscp.net/eng/docs/scripting
14
3 Shells
In *nix the shell35 is a text-based interface to the operating system. The kernel36
is the lowest-level innermost part of the operating system, so the shell is so
named because it surrounds the kernel.
Over the years, various shell programs have been written. The first was the
Thompson shell (commonly referred to as simply sh37) written by Ken Thompson at Bell Labs. Other popular shells have that have been written over the years
include the Berkeley Unix C shell (csh), the TC shell (tcsh; the T comes from the
TENEX operating system), the Korn shell (ksh), the Z shell (zsh), the Bourne
shell (sh), and the Bourne-again shell (Bash38).
Bash was the shell originally developed by GNU for the GNU operating system. It
is commonly available in most GNU/Linux distributions and this manual is written
assuming you are using Bash. Extensive online documentation for Bash is available39. Another reference is available at 40.
You can see which shells are installed on your system by typing cat /etc/shells
(the chsh -l command simply displays this file). Most likely if you are using Linux,
your default shell is bash. An easy way to tell what shell you are running is to
type less /etc/passwd and look for the line containing your username. The last
entry on the line will be your default login shell.
If you wish to change your default login shell you can use the chsh -s shell
command. This will modify the shell entry for your username in /etc/passwd so
the next time you login you will be running the new shell.
3.1 The Bash Shell

See Section 2.3 of this manual for instructions on how to log in to research.asu.edu. Once logged in, you will see a window that looks similar to the
one on the top of the next page. This is called a terminal window or just terminal; you may also hear the term console window or just console.
35
36
37
38
39
40
http://en.wikipedia.org/wiki/Shell_(computing)
http://en.wikipedia.org/wiki/Kernel_(computer_science)
http://en.wikipedia.org/wiki/Thompson_shell
http://www.gnu.org/software/bash
http://www.gnu.org/software/bash/manual/bashref.html
http://tiswww.case.edu/php/ chet/bash/bashref.html
15
The $ prompt is the Bash shell prompt. It is Bash's way of letting you know that it
is waiting for you to type a command. The research.asu.edu:~ part lets me
know the name of the machine I am logged in to and my current directory. The ~
symbol in Bash always refers to a user's home directory. Your home directory is
where you are placed when you first log in and it is the area in the file system
where you can store your files. The prompt string can be changed, see Section
6.1.
When you type commands at the prompt and press the Enter key, Bash will
read the command string you typed in, parse it to break it into words and operators, perform some post-processing, and will then execute the command. Unless
you tell it otherwise, Bash will wait for the command to finish executing before
displaying the $ prompt again (see Chapter 9 on Processes for how to run a
command in the background).
Some of the commands that you may use are built-in to Bash, i.e., the code that
gets executed is inside the Bash program itself. However, most *nix commands
are standalone programs that are found in the directories /bin, /usr/bin, and
/sbin directories (see Section 4.1 for a discussion of these directories). For example, we can determine where the GNU C compiler is installed on research.
asu.edu by typing,
$ whereis gcc
gcc: /usr/bin/gcc /usr/lib/gcc /usr/libexec/gcc /usr/share/man/man1/gcc.1.gz
whereis is a command which will search for the program file given to it. In this
case we are searching for the program file named gcc. whereis tells us that a file
named "gcc" can be found in four separate directories. In fact, the first one
16
/usr/bin/gcc is the actual executable file which is the GNU C compiler. For fun,
try to determine where the whereis program is installed,
$ whereis whereis
whereis: /usr/bin/whereis /usr/share/man/man1/whereis.1.gz
We can see that it is also installed in /usr/bin. In fact, most of the basic *nix
commands (i.e. executable programs) are located in this directory.
Bash will also execute special programs called scripts written in the Bash shell
scripting language. See Chapter 11 for a basic introduction to Bash shell scripting.
Overall, Bash is a very complex program, and we will only learn a small part of it
in this course. However, what we learn in this course is supposed to provide you
a solid foundation on which you can learn and obtain more knowledge of *nix and
Bash.
17
4 The *nix File System and Common Commands

The *nix file system41 is the part of the OS that manages the files and directories
stored on secondary storage devices such as tape, optical and hard disks. Files
and directories in the file system are stored in a hierarchical fashion42.
At the top level is the root directory indicated by a forward slash symbol /.
"Root" is at the top of the file system hierarchy. Every file and directory on the
system is stored under root and there is nothing above it in the directory tree.
4.1 Common Directories in *nix

On a typical *nix system you will find several common directories directly under
root. Some of these common directories are discussed below. A very good discussion of the *nix file system is available at 43.
/bin
Binary images (executables) for various *nix commands used by users. These
would include programs for commands such as ls, cp, mkdir and so on. In a typical distribution there will be a large number of commands files located in this directory, but only a small subset of them are technically required to be installed
there.
/boot
Contains the files necessary to boot the *nix operating system. This includes the
binary image of the kernel (i.e., the compiled kernel source code). In GNU/Linux
The kernel file is commonly named vmlinux (uncompressed) or vmlinuz (compressed) followed by the version and release numbers. If you want to know
which version of the kernel you are using, type uname -r at the shell prompt (discussed later).
/dev
The device directory contains files corresponding to different I/O devices connected to the system. These would include tape drives, cdrom drives, hard disks,
terminals, printers, etc. A special device is /dev/null. Anything written to /dev/null
disappears into the ether. One of the features that made Unix popular in the early
days was that all devices are treated as files, and could be accessed using file
I/O commands.
/etc
This directory contains host-specific files and directories which are used primarily
for system configuration purposes. Executable programs are not located here. If
the X window system is installed, the files for it will be installed in a subdirectory
of /etc, e.g., /etc/X11.
41
42
43
http://en.wikipedia.org/wiki/Berkeley_Fast_File_System
http://amath.colorado.edu/computing/ unix/tree.html
http://www.pathname.com/fhs/pub/fhs-2.3.html
18
/home
Users' personal directories are located here. Normally each user is given a home
directory where he/she may store his/her files. This is the directory where, as a
user, you would spend most of your time. A typical home directory for user fredf
would be /home/fredf. Note that the special symbol ~ is used to refer to the
home directory in the shell (discussed later).
/lib
Contains shared library images needed to boot the system and to run the commands in bin and sbin.
/mnt
Used to temporarily mount another file system under the current file system,
e.g., to connect the file system for a floppy disk or a cdrom to the current file system. How these devices are mounted is not something ordinary users care about.
/opt
Third-party application programs can be installed here, each within its own subdirectory.
/root
This is root's home directory. The special login identifier root is the master login
for the entire system. Root has privileges that other users do not have, and he or
she is typically the system administrator who is responsible for administering the
system, installing programs, adding users, mounting file systems, etc.
/sbin
This directory contains system binary files, i.e., executables that are normally
used only by root in administering the system. Other system binaries may be
found in /usr/sbin and /usr/local/sbin.
/tmp
Temporary files are often placed here.
/usr
The files and directories under usr are mounted read-only (except to root). Underneath usr are several important subdirectories.
/usr/bin
Other user-specific binaries. For example, the gcc program might be installed here.
/usr/include
Include header files for the C compiler are placed here.
/usr/lib
Contains object files and library files used in software development.
/usr/local
Another place where root can install user programs. For example, if you
install the Firefox web browser, you may install it under /usr/local/firefox
19
for example. Multiple networked host systems may access the files in
/usr/local.
/usr/share
Shared data files, e.g., font files, man pages, other documentation, icon
images files, etc.
/usr/src
Source code, generally for reference only (i.e., not for development purposes). On many GNU/Linux systesms, the Linux source code will be installed in /usr/src/linux (during installation, you usually have the choice to
download and install the Linux source code).
/usr/X11
Files used by X Window (X is part of the graphical system of your Linux
installation).
/var
Variable data files (i.e., data files that contain data that are changing) are stored
here. These would include temporary files and log files.
4.2 Pathnames
When referring to files in the file system, the shell maintains a current directory
pointer which also "points" to the current directory. When you log in to the system, you will be placed in your home directory, e.g., /home/fredf if your log in
identifier were fredf. That is now your current directory and is called an absolute
pathname because it refers to the complete path that must be followed from root
to get to the directory fredf. Another example of an absolute pathname might be
/usr/local/bin/firefox/v1.4.0/bin/firefox.
A relative pathname is a path to a file or directory which is specified relative to
your current directory. Two special relative pathnames are . (pronounced dot)
and .. (pronounced dot dot). Dot always refers to your current directory and dot
dot refers to the directory above your current directory (i.e., the parent directory). For example, if my current directory is /usr/local/bin then . is /usr/local/ bin
and .. is /usr/local. These are quite useful as we will see.
4.3 Commands, Arguments, and Options

A command is a word that is typed at the terminal window prompt in order to
perform some action. For example, the ls command is used to list the contents of
a directory (specifically, the current directory). If you would like to view the contents of a different directory, then you can supply an argument to the ls command which is the location of the desired directory, e.g., ls /bin would display the
contents of the /bin directory where /bin is the argument to the ls command.
20
In addition to arguments, many (most) *nix commands accept command options

as well. Options are specified either with a hyphen followed by a letter, e.g., -l, or
a pair of hyphens followed by a word, e.g., --help.
For example, ls -l /usr/bin is a command (ls) with the option -l and the argument
/usr/bin. We'll talk about what -l does when used with the ls command in a bit.
Options modify the way commands work. When no options are specified we
would say that is the default behavior of the command. Options can be used to
alter the default behavior. Note that like everything else in *nix, options are casesensitive. There is a difference between ls -l and ls -L as there is between ls and
LS.
A command can have multiple options, e.g., the ls command supports two options -l and -a that alter it in different ways. A legal ls command using these two
options would be ls -a -l. But often times, options can be combined so this same
command could be typed as ls -al and it would perform the same action.
A useful option you should know is --help. Many commands will display a screen
of text describing the commands, its action, and its options and arguments in response to the --help option.
Whitespace surrounding options is important. The command ls -a-l is erroneous
because there is no space in between the -a and -l options. The correct command is ls<space>-a<space>-l.
In conclusion, know that the general format of a *nix command is command options argumentsin that order.
4.4 Common File and Directory Commands

In this section, assume we are logged in as user fredf in our home directory
/home/fredf and we have the files as displayed below,
4.5 How Do I Know Which Directory I Am In?

Use the command pwd (print working directory),
21
$ pwd
/home/fredf
4.6 List Files, File Permissions, The Chmod Command

To list the files in a directory type ls,
$ ls
a.out file1.c file3 src
The option -a will show all the files in the directory,

$ ls -a
.bash_profile a.out file1.c file3 src
The special file .bash_profile is a configuration file which is used to configure

the Bash shell when a user logs in; we will discuss it later. Note that files starting
with a dot are hidden files and are not displayed during a normal ls command.
Hidden files are usually used to store configuration information for various programs. The option -l will display a long listing,
$ ls -l
-rwxr-xr-x
-rwxr-xr-x
-rwxr-xr-x
drwxr-xr-x+
1
1
1
1
fredf
fredf
fredf
fredf
fredf
fredf
fredf
fredf
20987
1098
654
0
Oct 10 10:04 a.out

Mar 7 2005 file1.c
Jun 12 18:44 file3
Jan 01 2005 src
The first column displays the permissions for the files44. Permission attributes
are changed with the chmod command45.
$ chmod go=
$ ls -l
-rwx------rwxr-xr-x
-rwxr-xr-x
drwxr-xr-x+
a.out
1
1
1
1
fredf
fredf
fredf
fredf
fredf
fredf
fredf
fredf
20987
1098
654
0
Oct 10 10:04 a.out

Mar 7 2005 file1.c
Jun 12 18:44 file3
Jan 01 2005 src
The above command removes all of the group and other's permission bits.
$ chmod ugo=rwx a.out
$ ls -l
-rwxrwxrwx 1 fredf fredf
-rwxr-xr-x 1 fredf fredf
-rwxr-xr-x 1 fredf fredf
20987 Oct 10 10:04 a.out

1098 Mar 7 2005 file1.c
654 Jun 12 18:44 file3
The above command gives everyone in the group and all other's full permissions.
See the online website46 for more explanation.
44
45
46
http://zzee.com/solutions/unix-permissions.shtml
http://catcode.com/teachmod
http://catcode.com/teachmod
22
4.7 Deleting Files

To delete a file use the rm (remove) command,
$ rm a.out
$ ls
file1.c file3
$ rm file3
$ ls -al
-rwxrwxrwx 1
-rwxr-xr-x 1
drwxr-xr-x+ 1
src
fredf fredf
fredf fredf
fredf fredf
36 Sep 23
1098 Mar 7
0 Jan 01
7:53 .bash_profile
2005 file1.c
2005 src
4.8 Creating, Changing, and Removing Directories

To create a new directory use the mkdir (make directory) command and to remove it use the rmdir (remove directory) command,
$ mkdir cse220
$ ls
a.out cse220 file1.c file3 src
$ rmdir cse220
$ ls
$ mkdir cse220
$ mkdir cse220/assgn02
$ ls
a.out cse220 file1.c file3 src
$ ls cse220
assgn02
To change your working (current) directory, use the cd (change dir) command,
$ cd cse220
$ pwd
/home/fredf/cse220
$ ls
assgn02
Suppose we have created directories cse220 and cse220/assgn02 and that

cse220/assgn02 has the files file1.c, file2.c, file1.o, file2.o, and a.out in it.
$ cd assgn02
$ pwd
/home/fredf/cse220/assgn02
$ ls
a.out file1.c file1.o file2.c file2.o
$ cd .
$ pwd
$ cd ..
$ pwd
/home/fredf/cse220
23
$ cd ..
$ pwd
/home/fredf
Before deleting a directory, it must be empty,

$ rmdir cse220
rmdir: `cse220': Directory not empty
$ cd cse220
$ rmdir assgn02
rmdir: àssgn02': Directory not empty
$ cd assgn02
$ rm file1.c
$ rm file2.c
$ rm file1.o
$ rm file2.o
$ rm a.out
$ ls
$ cd ..
$ rmdir assgn02
$ cd ..
$ rmdir cse220
$ ls
The command cd by itself will always change you back to your home directory.
The symbol ~ also always refers to your home directory as well.
$ cd /usr/local/bin
$ pwd
/usr/local/bin
$ cd
$ pwd
/home/fredf
$ cd /usr/local/bin
$ pwd
/usr/local/bin
$ cd ~
$ pwd
/home/fredf
$ cd cse220
$ cd assgn02
$ pwd
$ rm ~/file1.c
The last command would remove the file /home/fredf/file1.c.
4.9 Filename Globbing

Sometimes you want to do something to a group of files, e.g., delete all of them,
without having to perform the command on each file individually. For example,
suppose we want to delete all of the .c files in a directory.
24
A wildcard is a pattern which matches something else. Two commonly used

*nix wildcards are * (called star) and ? (question mark). Star means 0 or more
characters and ? means exactly one character.
Commands involving filenames specified with wildcards are expanded by the
shell (this is called globbing after the name of a former program called glob
which used to do this outside the shell). Thus, file* will match any filename which
starts with the characters 'f', 'i', 'l', 'e', and then is followed by zero or more occurrences of any character. For example, in fredf's home directory the wildcard file*
would match file1.c and file3. The wildcard file1* would match just file1.c.
A single * matches every filename in a directory, so the command rm * will attempt to delete every file in a directory. To delete all of the .c files, type rm *.c.
If I have a group of files named file1.c, file2.c, file3.c, file1.o, file2.o, file3.o,
file01.c, file02.c, file03.c, file01.o, file02.o, file03.o, file01.xyz, file02.xyz, and
file03.xyz, then the wildcard pattern file??.c would match file01.c, file02.c, and
file03.c but not file1.c, file2.c, or file3.c; it would also not match the .o files or
the .xyz file.
The wildcard pattern file01.? would match file01.c and file01.o but not file01.
xyz. The wildcard pattern file01.??? would match any filename with the name
file01 and a three character extension, specifically, file01.xyz.
A set of characters can be specified with brackets [ ]. For example, the command
rm file0[23].c would delete file02.c and file03.c but not file01.c. To delete all
files which begin with the character f type rm f*. To delete all files which begin
with either f or F (*nix is case-sensitive), type rm [fF]*. To delete all files which
start with the string file followed by a single letter type rm file[a-zA-Z]. The a-z
and A-Z in the last example means all the letters in the range lowercase a-z or
uppercase A-Z.
There's more to wildcard matching than this, but this is probably enough for our
purposes. See 47 if you are interested in learning more.
4.10 More on Deleting Subdirectories

To delete all the files in a subdirectory and all of its subdirectories and then delete the subdirectory itself, use the -r (recursive) command option with the rm
command,
$ rmdir cse220
rmdir: `cse220': Directory not empty
$ rm -r cse220
47
this deletes cse220 and everything in it
http://man-wiki.net/index.php/7: glob
25
$ ls cse220
ls: cannot access cse220: No such file or directory
The following is a very dangerous command,

$ cd
$ pwd
/home/fredf
$ rm -rf *
$ ls
This last rm command deletes everything in your home directory (all files, all directories, all subdirectories, and all files in all subdirectories). The option -f
means force and will cause even files marked as read-only to be deleted.
4.11 Copying Files and Directories

To copy a file use the cp command,
$ cd
$ pwd
/home/fredf
$ cp file1.c file1.c.backup
$ ls
a.out file1.c file1.c.backup file3 src
$ mkdir cse220
$ cp file3 cse220
$ ls
a.out cse220 file1.c file1.c.backup file3 src
$ ls cse220
file3
$ rm file3
$ ls
a.out cse220 file1.c file1.c.backup src
$ cd cse220
$ ls
file3
$ cp file3 ..
$ ls
file3
$ cd ..
$ ls
a.out cse220 file1.c file1.c.backup file3 src
To copy all the .c files from one directory to another,

$ cd
$ pwd
/home/fredf
$ mkdir src-backup
$ cd src
$ cp *.c ../src-backup
$ ls ../src-backup
26
file01.c file02.c file03.c file01.o file02.o file03.o
To copy all the files in a directory (and its subdirectories) to another directory,
use the -r command line option with cp (-r stands for recursive),
$ cd
$ cp -r src src-backup
$ ls src-backup
To copy a file from another directory to the current directory,

$ cp ../dir/file .
Note that the . is necessary in the command to specify the destination for the file
that is being copied. Without it you will get an error message,
$ cp ../dir/file
cp: missing destination file operand after `/dir/file'
Try `cp --help' for more information.
4.12 Moving and Renaming Files

To move a file from one location to another use the mv (move) command. Note
that move is different from copy in that copy "makes a copy" so the original file
remains where it was at. Move actually "moves the file" from one place to another, so the original is removed.
$ cd
$ ls
$ mv file1.c src
$ ls
a.out file3 src
$ ls src
file01.c file02.c file03.c file01.o file02.o file03.o file1.c
$ mv src/file1.c ../../tmp
$ ls src
To rename a file in *nix you move it to a new file, e.g., to rename file1.c to
file01.c,
$ mv file1.c file01.c
4.13 Finding a File

The find command can be used to find a file that is stored somewhere in the file
system. It is a very complicated command with lots of options, so I will only show
you the most basic command. Suppose you know there is a file named mlo27
cate.db somewhere in the file system but you don't know where. The command
to type would be,
$ find / -name mlocate.db
This command tells the find program to start at the root directory (the / part) and
search for a file a filed named (-name) mlocate.db.
It should be noted that a *nix installation might have 10,000 or more files in the
file system, so this command could run for quite a long time before it finds the
file. If you have a hunch that the file you are looking for is in a subdirectory below
root, say /usr, then you could narrow down your search by typing,
$ find /usr name mlocate.db
Now, only the directory /usr and its subdirectories will be searched. To perform a
case-insensitive search on the file use -iname rather than -name.
4.14 Displaying a File at the Terminal: Cat and Less

To displays the contents of a file at the terminal window, use the cat (concatenate) command,
$ cat file1.c
#include <stdio.h>
int main() {
printf("Hello world.\n");
return 0;
}
With cat, the contents of the file will scroll right off the top of the window if the file
is too large. You can use the less command to display a file on the window and
then use the Page Up (or b for backwards) and Page Down (or f for forwards)
keys to scroll through the file.
$ less file1.c
file1.c is displayed on window
Frequently-Used Less Commands

h
display help
q
quit less
g
go to first line of file
G
go to last line of file
=
display name of file and current line
ng
go to line number n
/
search forward for a pattern
?
search backward for a pattern
n
repeat search in same direction
N
repeat search in opposite direction
28
!cmd
v
run the shell and execute cmd

load current file in VI
4.15 Archiving Files

An archive is a collection of one or more files grouped together into a single file.
The program tar (tape archive) is used in *nix to form an archive. This archive file
is commonly called a tar file. Note that the tar file is not compressed. A compressed or uncompressed tar file is also commonly called a tarball. We will see
in the next section how to compress a tar file. To form a tar archive of one or
more files,
$ cd
$ ls
$ tar cvf archive.tar file1.c file3
file1.c
file3
$ ls
a.out archive.tar file1.c file3 src
Note that the "tarred" files are not removed. This tar command would create an
archive file named archive.tar which contains the files file1.c and file3. The v
option to tar specifies that the program be verbose and display the files it is archiving as they are being archived. The c option means create an archive file. The
f option means that what follows the options (the word "archive.tar") is to be the
name of the archive file. It is common to give this file a .tar extension. Following
the name of the archive file is a list of one or more filenames (or directories) to be
archived. To tar all of the .c files in a directory,
$ ls src
$ cd src
$ tar cvf archive-c.tar *.c
file01.c
file02.c
file03.c
$ ls
archive-c.tar file01.c file02.c file03.c file01.o file02.o file03.o
To list the contents of a tar archive to the screen, i.e., to see what is in the archive, use the t option (t stands for table of contents),
$ tar tf archive-c.tar
file01.c
file02.c
file03.c
To extract the files from a tar archive use the x (extract) option,
29
$ cd
$ mkdir gromulate
$ mv src/archive-c.tar gromulate
$ cd gromulate
$ tar xvf archive-c.tar
file01.c
file02.c
file03.c
$ ls
file01.c file02.c file03.c
The file names will be displayed as each of them is extracted (because of the v
option). Tar is often use to archive an entire directory and its subdirectories,
$ cd
$ ls src
$ tar cvf src.tar src
file01.c
file02.c
file03.c
file01.o
file02.o
file03.o
$ ls
a.out file1.c file3 src src.tar
$ tar tf src.tar
file01.c
file02.c
file03.c
file01.o
file02.o
file03.o
4.16 Compressing Files

The most common file compression programs in *nix are gzip and bzip2. To
compress a file with gzip,
$ gzip file1.c
$ ls
a.out file1.c.gz file3 src
This compresses the file and leaves file1.c.gz in the current directory; the original file file1.c is replaced by the .gz file. To decompress the file,
$ gzip -d file1.c.gz
$ ls
This uncompresses the file and leaves file1.c in the current directory; the .gz file
is removed. The program bzip2 is an alternative compression program which
30
uses a different algorithm than gzip; often compressing with bzip2 will result in a
smaller compressed file than if gzip were used. It works similarly to gzip,
$ bzip2 file1.c
This compresses the file and leaves file1.c.bz2 in the current directory; the original file file1.c is replaced by the .bz2 file. To decompress the file,
$ bzip2 -d file1.c.bz2
This uncompresses the file and leaves file1.c in the current directory; the .bz2
file is removed. To compress every .c file in a directory,
$ cd
$ ls src
$ cd src
$ gzip *.c
$ ls
file01.c.gz file02.c.gz file03.c.gz file01.o file02.o file03.o
To decompress every gzipped .c file in a directory,

$ gzip -d *.c.gz
$ ls
To zip a tar file, just gzip or bzip2 it like a normal file,

$ ls src
$ tar cf src.tar src
$ ls
a.out file1.c file3 src src.tar
$ bzip2 src.tar
$ ls
a.out file1.c file3 src src.tar.bz2
4.17 Hexdump
A hex dump is a display of the contents of a file in hexadecimal format. When
we want to "see inside" a binary file, we often will use a program that produces a
hex dump. In GNU/Linux the program is, aptly, named hexdump. In its simplest
form, you would type hexdump file to view the contents of file in hex. More
commonly, it is useful to use "canonical" mode by typing a -C option, e.g.,
hexdump -C file. Note that hexdump will display the entire contents of the file on
the terminal; you will probably want to pipe the output through less, for example,
hexdump -C file | less. There are many other options you can explore by reading the man page.
31
5 Getting Help
5.1 Man Pages
Historically the online Unix manual was divided into eight sections:
1.
2.
3.
4.
5.
6.
7.
8.
Commands
System Calls
Library Functions
Special Files
File Formats
Games
Miscellaneous Information
System Administration
The man (manual) command can be used in *nix to get help on many commands. For example, type man cd to display the "man page" for the cd command. Typing man kill would display information on the kill command from section 1 of the manual. However, if there is also information about "kill" in section 2,
one could type man 2 kill to display that entry (on some systems you may have
to enter man -s 2 kill instead).
When reading man pages you may see a command or phrase followed by a
number in parentheses, such as chmod(2). This tells you that man information
about chmod can be found in section 2.
Man output is displayed with less so you can use the Page Up and Page Down
keys to scroll through the manual page (or use the b and f keys; b for backward; f
for forward). Hit q to get out of less.
5.1.1 Reading a Man Page
Man pages can be difficult to read sometimes because of the complexity of the
commands and the various options and arguments. Nonetheless, if you really
want to become a proficient *nix user, you will have to learn how to read them.
There is no standard format for a man page, but many of them are organized in a
a similar format:
Heading
Name
Synopsis
Description
Environment
Author
Files
Copyright
See Also
Meaning
name and purpose of the command
syntax of the command; shows whether it accepts options or args
full description of the command; may be quite lengthy (e.g., bash)
environment variables used by the command
the person or persons who wrote the program
list of important files to this command
who holds the copyright to the program
where to look for related information
32
Heading
Diagnostics
Bugs
Meaning
possible errors and warnings
known mistakes or shortcomings
For example, let's consider the man page for the cp command.
CP(1)
User Commands
CP(1)
NAME
cp - copy files and directories
SYNOPSIS
cp [OPTION]... [-T] SOURCE DEST
cp [OPTION]... SOURCE... DIRECTORY
cp [OPTION]... -t DIRECTORY SOURCE...
DESCRIPTION
Copy SOURCE to DEST, or multiple SOURCE(s) to DIRECTORY.
Mandatory arguments to long options are mandatory for short options too.
-a, --archive
same as -dpR
--backup[=CONTROL]
make a backup of each existing destination file
-b
like --backup but does not accept an argument
--copy-contents
copy contents of special files when recursive
-d
same as --no-dereference --preserve=links
-f, --force
if an existing destination file cannot be opened, remove it and try again
-i, --interactive
prompt before overwrite
-H
follow command-line symbolic links in SOURCE
-l, --link
link files instead of copying
-L, --dereference
always follow symbolic links in SOURCE
-P, --no-dereference
never follow symbolic links in SOURCE
-p
same as --preserve=mode,ownership,timestamps
--preserve[=ATTR_LIST]
preserve the specified attributes (default: mode,ownership,timestamps),
if possible additional attributes: context, links, all
--no-preserve=ATTR_LIST
don't preserve the specified attributes
--parents
use full source file name under DIRECTORY
-R, -r, --recursive
copy directories recursively
--remove-destination
remove each existing destination file before attempting to open it
(contrast with --force)
--sparse=WHEN
control creation of sparse files
--strip-trailing-slashes
remove any trailing slashes from each SOURCE argument
-s, --symbolic-link
make symbolic links instead of copying
33
-S, --suffix=SUFFIX
override the usual backup suffix
-t, --target-directory=DIRECTORY
copy all SOURCE arguments into DIRECTORY
-T, --no-target-directory
treat DEST as a normal file
-u, --update
copy only when the SOURCE file is newer than the destination file or when
the destination file is missing
-v, --verbose
explain what is being done
-x, --one-file-system
stay on this file system
--help display this help and exit
--version
output version information and exit
By default, sparse SOURCE files are detected by a crude heuristic and the
corresponding DEST file is made sparse as well.
That is the behavior
selected by --sparse=auto. Specify --sparse=always to create a sparse
DEST file whenever the SOURCE file contains a long enough sequence
of zero bytes. Use --sparse=never to inhibit creation of sparse files.
The backup suffix is `~', unless set with --suffix or SIMPLE_BACKUP_SUFFIX.
The version control method may be selected via the --backup option or through
the VERSION_CONTROL environment variable. Here are the values:
none, off
never make backups (even if --backup is given)
numbered, t
make numbered backups
existing, nil
numbered if numbered backups exist, simple otherwise
simple, never
always make simple backups
As a special case, cp makes a backup of SOURCE when the force and backup options
are given and SOURCE and DEST are the same name for an existing, regular file.
AUTHOR
Written by Torbjorn Granlund, David MacKenzie, and Jim Meyering.
REPORTING BUGS
Report bugs to <bug-coreutils@gnu.org>.
COPYRIGHT
Copyright (C) 2008 Free Software Foundation, Inc. License GPLv3+: GNU GPL
version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
SEE ALSO
The full documentation for cp is maintained as a Texinfo manual.
If the info and cp programs are properly installed at your site, the command
info cp
should give you access to the complete manual.
The following sections are present in this man page: Name, Synopsis, Description, Author, Reporting bugs, Copyright, See Also.
The Name section is very simple: it states that the name of the command is cp
and that it performs the action, "copy file and directories."
34
The Synopsis section states that there are three different "forms" of the cp command. The brackets indicates that the item is optional and the ellipses indicate
or more (as in one or more). Therefore each of these three forms of the cp
command start with the letters "cp" followed by zero or more options.
The first form (cp [OPTION]... [-T] SOURCE DEST) states that the options
can be followed by a -T option and then a source (SOURCE) and a destination
(DEST). To know what -T does you have to read about it in the Description section. There it tells us that -T tells cp to treat DEST as a normal file and not as a
destination directory. Thus we are copying SOURCE to a file and not to a directory. What is SOURCE? It's a single file. An example of this command would be cp
-T afile ../someotherfile. Note that if ../someotherfile is a directory, you will get
an error message.
The second form (cp [OPTION]... SOURCE... DIRECTORY) states that multiple source files can be copied to a single directory. Note that the directory must
be the last argument on the command line. An example of this command would
be cp file1 file2 file3 somedir.
The final form (cp [OPTION]... -t DIRECTORY SOURCE...) requires the -t
option. Looking in the Description section we see that -t means, "copy all
SOURCE arguments into DIRECTORY". The target directory is specified following the -t option, and the list of one or more source code files is on the end of the
command line. An example of this command would be cp t ../../somedir file1
file2 file3.
Next is the Description section. This is usually the lengthiest section in a man
page. Among other things, the Description section will provide a description of
each of the options. We're not going to go through all of them here, but let's just
look at a couple of them.
The -i (or --interactive) option will prompt you before it overwrites the destination
file, e.g.,
$ cp i file1 file2
cp: overwrite `file2'?
At this point you hit 'y' or 'n' to either allow or disallow the overwrite.
The -u (--update) option will copy the source to the destination only when the
source is newer or when the destination is missing.
The Author section is self-obvious. Note that if you find a genuine bug in the cp
program, you are told whom to contact in the Reporting Bugs section. Finally, the
35
See Also section tells you that there is more information about cp in info (described Section 5.3).
5.2 Apropos
Sometimes you know what you would like to do but you are not sure what command to use. For example, suppose you want to rename a file but you don't know
the command. You could type man -k rename and man would display a list of
manual pages with a one-line description that contains the word "rename". You
could look through that list to see if there is an entry for a command that renames
files. The apropos foo command is equivalent to man -k foo. Note that apropos
is pronounced "ap-pro-po" and not "ap-pro-pose".
5.3 Info
Info is an online help system, separate from the *nix manual, that is used to document GNU utilities. To get help on the bzip2 command you would type info
bzip2. Where info differs from man is that info documentation is organized into a
hierarchical document of nodes (it's a tree structure) and the user uses keypresses to navigate around the document. Info is, like the Emacs editor, incredibly complicated to use at first. Important info commands are summarized below.
General Commands
q
quit
z
start help tutorial
?
display command summary
^x 0
get out of the command summary window pane (that's a zero)
Reading a Node
Pg Dn
display next screenful of text
Pg Up
display previous screenful of text
b
jump to first line of text in node
Up
move cursor up one line of text
Down
move cursor down one line of text
Left
move cursor one char to the left
Right
move cursor one char to the right
Moving Around Nodes
n
jump to next node in document
p
jump to previous node in document
t
jump to the top node in the document
u
move up from this node to the parent node
Home
move to top line of node
End
move to bottom line of node
36
Jumping From Node File to Another

Tab
move cursor to next hyperlink in current node
Return
follow hyperlink to a new node
l
jump to previously displayed node
d
jump to directory node (the root node of info)
Hyperlinks to other nodes are of the form * node:. For example, shown below is
part of the info node for the bzip2 command. The first hyperlink in this node is *
Menu:. The cursor is currently in column one of row 1. Pressing tab will move the
cursor to the Menu link. Pressing Return, then, would display the "Menu" node.
5.4 Final Notes

There is extensive online documentation about Unix, GNU, and GNU/Linux as
well as Bash and other shells. An excellent starting point for learning more about
GNU/Linux is the Linux Documentation Project48.
The Bash reference manual is available from GNU49. A good reference on various *nix commands can be found at 50. All of the man pages are online at various
cites; just use Google to locate them. The GNU Compiler Collection51 (GCC), is
the standard suite of compilers used in the *nix world. The manuals for GCC are
available at 52.
48
49
50
51
52
http://tldp.org
http://www.gnu.org/software/ bash/manual/bashref.html
http://www.computer hope.com/unix.htm
http://gcc.gnu.org
http://gcc.gnu. org/onlinedocs
37
6 Environment Variables, Search Path, Prompt

The shell maintains several environment variables53 which are used to store
information used by the shell. The PATH environment variable lists the directories in the order that the shell will search when trying to locate an executable file.
You can display the environment variables with the set command (the env and
printenv commands do the same thing),
$ set
a lot of stuff scrolls by
$ set | less
In the second set command, the output is being piped through the less program.
We will discuss piping in more detail in Chapter 8. To change the value of an environment variable, use varname=value,
$ MYNAME=Kevin
$ set | grep MYNAME
MYNAME=Kevin
To add . and sbin to your path,

$ PATH=/sbin:$PATH:.
$ set | grep PATH
PATH='/sbin:/usr/local/bin:/usr/bin:/bin:/usr/X11R6/bin:.'
Setting an environment variable at the Bash command prompt only changes the
value of the environment variable for this session. To make the change permanent (so the variable will be defined when you log in again), edit your .bashrc file
in your home directory, and add the lines,
MYNAME=Kevin; export MYNAME
PATH=/sbin:$PATH:.; export PATH
The special file .bashrc contains commands that are executed whenever a "new
environment" is created, i.e., whenever a new shell is started54. The next time
you log in the environment variables will be updated. To delete an environment
variable, use the unset command, e.g., unset MYNAME.
6.1 Changing your Prompt

The prompt is the string that is displayed by Bash when it is waiting for your input. By default your prompt may be some string ending with a $. You can change
the prompt. An excellent online reference is at 55. As an example, my prompt
53
54
55
http://tldp.org/LDP/Bash-Beginners-Guide/html/sect_03_02.html
The file .bash_profile contains commands that are executed when you log in to the *nix system. It differs from .bashrc
in that the commands in .bash_profile are only executed once, during log in. On the other hand the commands in
.bashrc are executed by the shell every time a new shell is started.
http://tldp.org/HOWTO/Bash-Prompt-HOWTO
38
string is defined in my .bashrc file as PS1='\[\033[1;32m\]\w $\[\033 [0m\] '.

The environment variable PS1 is your main prompt string. This funky-looking
command sets my prompt to display the current directory (the \w part) followed
by a $ in bright green. Setting prompt colors is described at 56. To permanently
change your prompt in Bash, edit .bashrc and enter a line such as,
PS1=desired-prompt-string; export PS1
56
http://tldp.org/HOWTO/Bash-Prompt-HOWTO/x329.html
39
7 Searching Files
It is very common to want to search a file for a certain string. The grep (get regular expression) command57 can be used to do this,
$ cd
$ grep MYNAME .bash_profile
MYNAME=Kevin
The string you are searching for is typed first and then you type the filename of
the file in which you wish to search. To search for the string in every file of the
current directory try grep string * (or grep string .* to include hidden files).
Grep can be used for more than simply searching a file or group of files for a
fixed string. Grep allows searching files for patterns which are specified using a
particularly obtuse syntax. We are not going to discuss grep patterns in this
course, but remember there is a lot of good online documentation that can be
used to learn more. For example, see 58 or 59.
57
58
59
http://en.wikipedia.org/wiki/Grep
http://www.gnu.org/software/grep/doc/grep.html
http://www. panix.com/~elflord/unix/grep.html
40
8 I/O and Redirection

Unix was created in the days of text-based computing before graphical user interfaces were common. One of the design philosophies was that users would interact with the system through small programs which performed some simple
command or purpose. Dennis Ritchie and Ken Thompson designed these simple
programs so they would accept input from the keyboard, a file, or as the output of
another program. Thus, there was a need to be able to direct the output of a program to either a file or some other program and to also specify where a program
was to get its input.
In *nix, the kernel opens and makes available three files for every program that
is executed. These three files are called standard input, standard output, and
standard error, usually abbreviated to stdin, stdout, and stderr.
By default, stdin is attached to the keyboard, and stdout and stderr to the console
window. If a program prints something to stdout, then it will be displayed on the
console window. If it reads something from stdin, the program will wait for input
from the keyboard.
Stdin and stdout can be changed so that rather than reading input from the keyboard, the program will read input from a file. Traditionally in other operating systems of that era, this was done by changing the source code to read the data
from a file rather than from the keyboard. One of the most beautiful things about
Unix was that this could be accomplished without changing the source code. Rather, the user would redirect stdin from the keyboard to a data file when invoking
the program and the source code would remain unchanged. The same could be
done for stdout: rather than sending the output of a program to the console window it could be redirected to a file.
8.1 Redirection Example

An example may be helpful. The program cat normally reads a file and sends the
output to the console window. When invoking cat we can redirect its output to a
file rather than the console window,
$ cat file01.c > fooie
The > symbol is used to send cat's output to the file fooie rather than to the console window. Note that we have just made a copy of file01.c. The < symbol is
used to redirect stdin. Consider a C program that reads in two integers and prints
out the larger one.
41
/* larger.c */
#include <stdio.h>
int main() {
int n1, n2;
fprintf(stdout,"Enter an integer: "); fscanf(stdin,"%d",&n1);
fprintf(stdout,"Enter an integer: "); fscanf(stdin,"%d",&n2);
fprintf("The largest integer was the ");
if (n1 > n2) fprintf(stdout,"first one: %d\n",n1);
else fprintf(stdout,"second one: %d\n",n2);
return 0;
}
Since this program reads its input from stdin and writes its output to stdout, we
can redirect its input and output so it reads its input from a file and sends its output to a file. For example, suppose we create a text file named larger.in which
has two lines, the first with a 2 and the second with a 4. The following command
will execute our program and send its output to larger.out.
$ cat larger.in
2
4
$ larger < larger.in > larger.out
$ cat larger.out
Enter an integer: Enter an integer: The largest integer was the second
one: 4
8.2 Redirection Example

For a different example, consider that cat normally sends its output to the console window. We can redirect cat's output to a file,
$ cat > larger.in
2
4
<Ctrl+D>
$ cat larger.in
2
4
user types 2 and presses enter

user types 4 and presses enter
user hits Ctrl and D keys simultaneously
This is a commonly used quick-and-dirty way to create a small, simple text file
without invoking a text editor.
8.3 Redirecting Stderr

Stderr can be redirected as well,
$ larger < larger.in > larger.out 2> larger.err
42
Note that < is really 0< and > is really 1>, so the above command is equivalent
to,
$ larger 0< larger.in 1> larger.out 2> larger.err
Stderr is not used in C and C++ programs as much as stdout is. The idea with
stderr is that error messageswhich are special in that they are not part of the
normal program outputcan be sent to a different location than the normal output. For example, when running a program, you might send stdout to an output
file and stderr to the console window,
$ blobx < data.in 1> data.out
Since the normal connection for stderr is to the console window, we only need to
redirect stdout to the output data file. Any error messages printed to stderr will
appear on the console window.
8.4 Sending Output to /dev/null

*nix interfaces to "devices" through files, e.g., the CD-ROM drive, the mouse, etc.
A special device named /dev/null is created at startup. Anything written to
/dev/null simply "goes away". So, we can use /dev/null and redirect stdout when,
for example, we don't want to see the output of a program on the terminal window,
$ prog
output is displayed on the terminal
$ prog > /dev/null
no output
Anything written to stdout within the program will be redirected to /dev/null and
will not be displayed. Stderr can also be redirected to /dev/null.
8.5 Concatenating Files

Cat can read input from multiple input files and it will send its output to stdout,
$ cat file1
cookies
pizza
$ cat file2
dogs
cats
$ cat file3
red blue
$ cat file1 file2 file3 > file4
$ cat file4
cookies
pizza
dogs
cats
43
red blue
Here, the contents of the files file1, file2, and file3 have been concatenated together when creating the new file file4. Now you can see where the name cat
came from (it is short for concatenate)
8.6 Sending Stdout and Stderr to the Same Place

Sometimes you want to send stdout and stderr to the same output file. This can
be done using the command line,
$ blobx 1> data.out 2> data.out
The drawback is that you have to specify the same filename twice. To send
stdout and stderr to the same file without having to specify the filename twice,
use 1>&2 or 2>&1,
$ blobx 1> data.out 2>&1
The notation 2>&1 tells the shell to send stderr's output (that is, 2>'s output) to
wherever stdout (that is 1>) is going. Note that if 2>&1 is used then 1> must appear before 2>&1. If 1>&2 is used (to send stdout to wherever stderr is going)
then 2> must appear before 1>&2. If you do not do that, then something else
happens. For example, $ blobx 2>&1 1> file.err would send stdout to file.err
and stderr would go to wherever stdout normally goes, i.e., the console window.
8.7 Appending to a File

To append a file to another file use the >> symbol,
$ cat file1
cookies
pizza
$ cat file2
dogs
cats
$ cat file1 file2 >> file1
$ cat file1
cookies
pizza
cookies
pizza
dogs
cats
The notation >> is equivalent to 1>> and the notation 2>> can be used to concatenate to stderr.
44
8.8 Pipes
To connect the stdout of one program to the stdin of another program, in *nix we
use a pipe (the output flows out one program through the pipe to the input of the
other program). The pipe symbol is | and the way to send the output of one program to the input of another program is program1 | program2. For example, to
display a long listing of the files in a directory and sort them based on the time
the file was last modified,
$ ls -l | sort +7
The program sort sorts its input (which is a text file) and sends the sorted output
to stdout. The option +7 tells sort to sort the input file based on the data in field 7
which in the ls -l output is the column which contains the timestamp. To sort on
size try ls -l | sort + 4. To display the contents of a directory and display the output using less,
$ ls -l | less
45
9 Processes
When you run a program in *nix, the OS loads it into memory and sets up some
other information concerning your program. The loaded and running program is
called a process. When the program terminates, so does the process.
Every process has a system-wide unique identifier called the process id, or pid;
it is just an integer. The process is owned by the user who launched it. The ps
command will show you a list of the running processes of which you are the
owner.
$ ps
PID TTY
17776 pts/18
19041 pts/18
TIME CMD
00:00:00 bash
00:00:00 ps
I am running two programs right now: bash and the ps program itself. The PID of
each process is shown. The time column is the cumulative amount of CPU time
the process has consumedthis is not the same as wall time, i.e., how many
seconds it has been since you ran the program.
To see all of the processes running on the system, type ps -A or ps -e. The
command ps -f will display full process information including the user id of the
process owner. It can be combined with -A or -e. To sort this list by user id and
view it in less, use ps -ef | sort | less.
When you run a program (i.e., a process) you can ask the OS to run it in the
background. Do this by putting a & on the end of the command line,
$ some-program &
$ now-you-can-type-another-command
Now some-program will run in the background and you can get back to work
typing commands. To bring a running program back into the foreground, type the
fg command,
$ fg
Now some-program would be running and if it is displaying output, the output
would be going to the console window. To put the program back into the background, hit Ctrl+Z to pause it, and then type bg to send it back to the background.
To see a list of your processes running in the background use the jobs command. If you have more than one background job they will be numbered with a
46
job number 1, 2, 3, .... To bring job number 3 to the foreground, type fg %3. To
send this job back to the background, hit Ctrl+Z, and type bg.
To terminate a runaway process, use the kill command. The syntax is kill pidnumber. You cannot kill processes that you do not own.
To kill a process running in the background, issue a jobs command to see your
running processes and the determine the job number. Then use kill job number
to kill the job; e.g., kill %3 to kill job number 3.
The top command will display a table of the processes which are currently using
the most CPU time. It is a highly configurable program; try man top. (Be aware
that top is a fairly resource-intensive program; it's not cool or nice to run top a
lot).
The nice command was designed to allow a user to run a program with lower
than normal priority, so it will not hog the CPU as much. The idea is that other
people may be peeved if you're sucking up too much CPU time, so you can be
nice about it and still run your program. The syntax is nice -n adjust command
where command is the program you want to run (e.g., it could be a big make
process). Adjust is an integer which determines the adjusted process priority:
-20 is highest, to 19 lowest. No one ever uses nice.
47
10 Editing Files
10.1 Absolutely Painless Introduction to Vim
The VI editor60 is the standard editor available on most *nix systems. Some
people who know it very well, love it. I don't know it very well; I don't love it. I've
never taken the time to learn it very well because I would rather gouge my eyes
out with infected hypodermic syringes found in the dumpster behind the plasma
donation center until bloody pus oozes out of them than learn VI61; I simply do
not have the time for it. Nonetheless, learning a bit of VI can be useful, if for nothing else, you may find yourself in a situation where it is the only editor that is
available on the system you are on. At the least, knowing how to load a file,
make some simple changes to it, and save it can be useful.
The version of VI found on most GNU/Linux systems is Vim (VI Improved)62.
Purdue University has a good VI tutorial63. Also see 64.
10.1.1 Starting VI
At the Bash command prompt, type vim or vim some-filename. VI is a modal
editor in which you are either in editing mode or command mode.
10.1.2 Getting out of VI
This is the most important thing you need to know about VI: how to get out of it. If
you are in editing mode, hit ESC to enter command mode. Enter :wq to write
your file and quit. To quit without writing the file, hit :q. If you really want to quit
without saving the file you have made changes to hit :q!. To save the file under a
different name, try :wq new-filename.
10.1.3 Switching Between Editing Mode and Command Mode
If you are in editing mode, hit ESC to enter command mode. If you are in command mode hit i to enter insert (editing) mode. If you cannot tell which mode you
are in, trying hitting ESC several times to make sure you are in command mode,
then type :set showmode. This may tell VI to let you know when you are in insert mode (it depends on the version of VI you are using).
10.1.4 Other Useful Settings
All of these are recommended in addition to :set showmode. In command mode,
type :set nocompatible to enable advanced VIM features. Type :set ruler to tell
VI to display your current cursor position. Type :set number to display line numbers.
60
61
62
63
64
http://en.wikipedia.org/wiki/Vi
It would be preferable if I were also on fire and being eaten alive by rabid hyenas at the same time. That would still be
more pleasurable than using VI.
http://en.wikipedia.org/wiki/Vim_(text_editor). It's really not improved because it's still VI.
http://engineering.purdue.edu/ECN/Support/KnowledgeBase/Docs/20020202121609
http://thomer.com/vi/vi.html
48
10.1.5 Moving Around a Character at a Time

The original VI was written by Bill Joy (at UC Berkeley) using a computer system
and keyboard that lacked arrow, page up, and page down keys. This is the reason for the stupid assignment of the keys for moving around: h (move left one
character), j (move down one line), k (move up one line), and l (move right one
character). On Joy's terminal, these four keys also had arrow keys on them and
that is the historical reason they are still mapped that way today.
10.1.6 Moving Around a Word at a Time
Forget about your Del and Backspace keys. They may or may not work. Switch
to command mode. Press w to move forward one word. Press b to move backward one word. Press nw to move forward n words, e.g., 3w to move forward
three words. Press nw to move backward n words.
10.1.7 Moving Around the File
Switch to command mode. To move to the end of the line, hit $. To move to the
beginning, hit 0. Hit 1G to go the first line of text. hit nG to go line number n. Hit
G to go to the end of file. To display line numbers, in command mode type :set
number. Note that :13 is equivalent to, e.g., 13G. To page down, hit Ctrl+F. To
page up hit Ctrl+B.
10.1.8 Deleting Characters
Switch to command mode, move to the character you want to delete, and hit x to
delete that character. To delete n characters hit nx, e.g., 17x to delete 17 characters. To delete all the characters from the cursor position to the end of the line
hit D.
10.1.9 Copying a Line of Text
To copy a line of text from one place in the file to another, switch to command
mode. Move to the line you want to copy. Hit yy (yank something or other, I dunno). Move the cursor to the location where you want to make the copy. Hit p (lowercase p) to put the copied line after the current line. Hit P (uppercase P) to put
the copied line before the current line.
10.1.10 Moving a Line of Text
To move a line of text from one place in the file to another, switch to commandmode. Move to the line you want to copy. Hit dd. Move the cursor to the location
where you want to make the move. Hit p (lowercase p) to move the line after the
current line. Hit P (uppercase P) to move the line before the current line.
49
10.1.11 Deleting a Line of Text

Switch to command mode. Move to the line you want to delete. Hit dd.
10.1.12 Cutting, Copying, Pasting Multiple Lines of Text
Use nyy to copy n lines of text to the copy buffer. Move to where you want the
lines to be and hit p or P to copy them. Use ndd to move n lines of text somewhere. Hit ndd to delete n lines of text, e.g., 3dd will delete the line of text the
cursor is on and the next two lines.
10.1.13 Moving and Copying Text in Visual Mode
To cut, copy, and paste text that is not an entire line, enter visual mode by hitting
v in command mode. Use the h, j, k, l keys to move around and select the text
you want to copy or move. Once selected, hit y to copy the text to a buffer or d to
delete the text to the buffer. Once copied (with y) or deleted with (with d) move to
where you want the text to be. Hit p to paste the text.
10.1.14 Finding Text
Switch to command mode. Hit / (forward slash) and enter the string you are
searching for, e.g., /cookies. The cursor will be placed on the first occurrence of
cookies following the current location. Hitting n over-and-over will continue
searching.
10.1.15 Find and Replace
Switch to command mode. To replace all occurrences of homer by bart, try
something like :1,100 s/homer/bart/g. The 1,100 part means search from line 1
to line 100 inclusive (I think 1,$ will search the entire file). The s stands for
search. We are searching for homer and replacing all occurrences by bart. The
g stands for global which means it will replace all occurrences without prompting
you for each one. If you want to be prompted for each one, use c instead (for
confirmation).
10.1.16 Undo and Redo
Since you will use it the most because of VI's horrible user interface the most
useful command in VI (other than :q!) is the undo command. Hitting u will undo
your last change. Hitting u multiple times will undo all of your changes back to
the last time you saved the file (only if you are in :set nocompatible mode). Redo is Ctrl+R.
10.1.17 More Help
Let's say you're either insane or a masochist and want to learn more about VI. In
command mode, hit :h to bring up help. The most useful thing you need to know
here is how to get out of help, hit :q (like in less).
50
10.2 KDE - The Kate Editor

Kate (KDE advanced text editor) is a GUI text editor for KDE. Its features include
color syntax highlighting, search and replace, spell checker, copy-and-paste,
adding comments, etc. To run Kate, at the Bash command prompt type kate or
kate filename. The Kate editor window will appear, see below. From here on you
can probably figure out what you need to know about Kate as it does not have a
user interface which is as absolutely awful as VI.
10.3 Other Linux Editors

There are a plethora of editors for GNU/Linux and Unix: VI, VIM, Kate, Kedit,
GEdit, Joe, Pico, Nano, Elvis, Emacs, X Emacs, NEdit, it goes on and on and on.
Editors are a matter of personal preference. Some people feel very strongly
about the merits of their personal preference. Some VI people will never quit talking about how great VI is and how much Emacs sucks. There are Emacs people
who will tell you that Emacs is the only editor in the Universe and God himself
uses Emacs65. If you have the need or desire to learn to use another editor go
ahead, but honestly, ed66 is all you need.
65
66
http://en.wikipedia.org/wiki/Editor_wars
http://en.wikipedia.org/wiki/ Ed_(Unix)
51
11 Basic Bash Shell Scripting

One of the more powerful features of Unix systems and the shells is the ability to
write shell programs, i.e., programs which are executed by the shell interpreter
in order to perform useful commands or perform some task. These programs are
called shell scripts, scripts, or script files. Shell scripts are a way of automating commonly performed commands or a complex set of commands.
A good basic introduction to Bash shell scripting is available at 67 and a more advanced guide at 68. Complete books have been written about shell scripting, and
it is not something we can cover in a day or two and five pages of notes. This will
be a basic introduction, and you are referred to the online guides and tutorials for
more information.
A script file is a text file, and the first line is usually supposed to be #!/bin/bash if
the shell script is designed to be executed by the Bash shell. If you are using a
different shell to write your script (e.g., the Korn shell), then set the first line to
point to the Korn shell's binary image (usually it is in the /bin directory; e.g.,
/bin/ksh). This will ensure that the correct shell is loaded and used to execute the
shell script regardless of what particular shell the user may be running.
Note that in shell scripts, # lines are comments, just as // are comments in C++.
Any character from the # symbol to the end of the line is ignored by the shell interpreter.
11.1 Variables
Shell variables are used to store values. A variable can be defined at the same
time it is assigned by the = operator, e.g.,
LONG_NAME="Flintstone Fred"
SHORTNAME=Fred
Note in the first case that quotation marks are required because the string contains spaces. The backquote operator will execute the enclosed command and
assign the output to the designated variable, e.g.,
LS_OUT=`ls`
When referring to the variable, its name must be preceded by $, e.g., X1 = $Y2.
Here we are assigning the value of Y2 to X1. Note that Y2 is preceded by $ whereas X1 is not.
67
68
http://tldp.org/HOWTO/ Bash-Prog-Intro-HOWTO.html
http://tldp.org/LDP/abs/html
52
11.2 Arithmetic and Relational Operators

The standard operators are available as shown in the table (note: there are more
operators than those shown here; read the Bash man page for more information
about the complete set of operators).
Within an expression, the normal precedence rules are applied; in the table precedence is from highest (unary minus, plus) to lowest (logical or).
!
**
*
+
<<
>
==
&
^
|
&&
||
+
~
/
-
%
>>
<
>=
!=
<=
Unary minus, plus

Logical negation, bitwise negation
Exponentiation
Multiplication, division, remainder
Addition, Subtraction
Left and right bitwise shifts
Greater than, less than, greater than or equal, less than or equal
Equal, not equal
Bitwise and
Bitwise exclusive or
Bitwise or
Logical and
Logical or
Variables can be assigned integer values using normal assignment, or using let,
e.g.,
x=33
let y=12
let z = 95
# error
The last statement will cause the shell script to fail; there should be no spaces to
the left and right of the = operator. Expressions are evaluated by enclosing them
in $[ ], e.g.,
x=33
let y=12
z=$[x/y]
echo $z
Displays 2 (note integer division is performed).
11.3 Basic Shell Script Example

A script file can contain regular Bash commands, e.g., rm, cp, mv, ls, etc. A
script file can also define variables and assign them values. As we will see
shortly, the Bash shell programming language is quite sophisticated and allows if
statements, loops, and procedures. Here is a sample shell script which moves to
directory tmp and deletes all of the .o files (the script file is named rmo.sh),
53
#!/bin/bash
# usage: rmo
mv tmp
rm *.o
$ ./rmo.sh
./rmo.sh: Permission denied
The shell script failed to run because shell scripts are not executable files by default. To make the shell script file executable, you must change the permissions
using the chmod command to add x,
$ ls -l rmo.sh
-rw-r--r-- 1 kburger2 kburger 122 Oct 17 11:35 rmo.sh
$ chmod 744 rmo.sh
$ ls -l rmo.sh
-rwxr--r-- 1 kburger2 kburger 122 Oct 17 11:35 rmo.sh
$ ./rmo.sh
A slightly more sophisticated program might use variables,

#!/bin/bash
# usage: rmo
dirname=tmp
pattern=*.o
cd $dirname
rm $pattern
Here dirname and pattern are variables. Note that when the variable is assigned
to (using =), we do not put a $ in front of the variable name, but we do use a $
when referring to the variable later on. If you don't, the shell will think that the variable name is the filename of a program to be executed and it will probably fail
with an error message.
11.4 Command-Line Arguments

When a Bash shell script is launched, arguments may be passed to the script,
e.g., as in $ ./mymake.sh foo bar install. Here mymake.sh is the name of the
shell script file, and foo, bar, and install are arguments. Special shell variables,
$0, $1, $2, , are used to refer to the command line arguments. The variable $0
refers to the name of the shell script (./mymake.sh), $1 the first argument (foo),
$2 the second argument (bar) and so on. $# is the number of command line arguments following the script filename. Our script above can be rewritten,
#!/bin/bash
# usage: rmwild dir pattern
dirname=$1
pattern=$2
cd $dirname
rm $pattern
$ ./rmwild tmp *.o
54
11.5 Aliases
An alias is another name for something. In Bash, you can define aliases for
commands using the syntax alias name=command,
$ alias rmwild="./rmwild"
$ rmwild tmp *.o
$ alias gohome=cd ~
$ pwd
/usr/bin
$ gohome
$ pwd
/home/fredf
Aliases defined on the command line only exist for the current Bash session. If
you want the alias name to be available every time you log in, define the alias in
your .bashrc file.
To delete an alias that was defined at the command prompt, use the unalias
command, e.g., unalias rmwild. To delete all of your aliases, type unalias -a.
11.6 Displaying Text from a Shell Script

An executing script can display text on the console window using the echo command. For example,
#!/bin/bash
# usage: rmwild dir pattern
dirname=$1
pattern=$2
echo "Removing all $2 files from $1..."
cd $dirname
rm $pattern 2> /dev/null
echo "Done."
$ ls tmp
file1.c file1.o file2.c file2.o
$ rmwild tmp *.o
Removing all *.o files from tmp...
Done.
$ ls tmp
file1.c file2.c
11.7 A Simple Backup Script

This script will tar and compress the files in the specified directory which is located beneath the current directory,
#!/bin/bash
# usage: bu dir
tar cvf $1.tar $1
bzip2 $1.tar
55
mv $1.tar.bz2 mybackups
echo "$1 backed up."
$ bu src
src backed up.
$ ls mybackups
src.tar.bz2
11.8 Reading From the Keyboard

To read a string from the keyboard and store it in a shell variable use the read
command, e.g.,
#!/bin/bash
echo "Enter your name: "
read name
echo "Pleased to meet you $name."
11.9 Capturing Command Output

The output of a command can be assigned to a variable,
#!/bin/bash
LS_OUT=$(ls)
echo $LS_OUT
Or
#!/bin/bash
DOW=$(date +%a)
MONTH=$(date +%b)
DAY=$(date +%d)
YEAR=$(date +%Y)
echo "Today is ${DOW} ${MONTH}-${DAY}-${YEAR}"
$ alias today="./today.sh"
$ today
Today is Wed Oct-17-2007
11.10 If Statements
The syntax of the if statement is,
if [ conditional-expression ]; then
some commands
fi
or
if [ conditional-expression ]; then
some commands
else
some other commands
fi
56
The character "[" actually refers to a built-in Bash commandit is a synonym for
another command named test. The result of the conditional expression is 0 or
nonzero with 0 being false and nonzero being true. Some examples,
Use VI to edit if.sh
-------------------------------today=0
if [ $today = 1 ]; then
echo "today = 1"
else
echo "today <> 1"
fi
-------------------------------$ chmod 755 if.sh
$ ./if.sh
today <> 1
Since using [ ] is equivalent to using test,

today=0
if test $today = 1; then
echo "today = 1"
else
echo "today <> 1"
fi
The test command has command line options for checking the status of files,
e.g.,
test
test
test
test
test
test
-d
-e
-r
-s
-w
-x
file
file
file
file
file
file
True if file is a directory

True if file exists
True if file exists and is readable
True if file exists and has size > 0
True if file exists and is writable.
True if file exists and is executable
Multiple conditions can be checked using -a (and) and -o (or), e.g.,

if test -e if.sh -a -x if.sh; then
echo "if.sh is an executable"
else
echo "if.sh either does not exist or it is not an executable"
fi
57
In a schell script, the ! symbol is used as the NOT logical operator, e.g.,
if [ ! -d src ]; then
echo '"'src'"' does not exist, creating directory.
mkdir src
fi
To learn more about conditional expressions I suggest you try $ man test.
11.11 For Loops

Bash implements for, while, and until loops. The basic syntax of the for loop is:
for var in list; do
some commands
done
Note the semicolon; it is important to not omit it. The commands are executed
once for every item in the list. The current item is accessed through the variable
$var, e.g.,
Use VI to edit for.sh
-------------------------------for i in 1 2 3 4 5 6 7 8 9 10; do
echo -n "$i "
done
echo
-------------------------------$ chmod 755 for.sh
$ ./for.sh
1 2 3 4 5 6 7 8 9 10
The option -n to echo means do not print a newline character after outputting the
text.
Another example,
for i in $(ls); do
echo $i
done
This script display the name of each of the files in the current directory one per
output line. Another example,
Use VI to edit ls-dirs.sh
-------------------------------for i in *; do
if [ -d "$i" ]; then
echo "$i [dir]"
else
echo $i
58
fi
done
-------------------------------$ mkdir foo1
$ mkdir foo2
$ chmod 755 ls-dirs.sh
$ alias ls-dirs=./ls-dirs.sh
$ ls
foo1 foo2 for.sh if.sh ls-dirs.sh
$ ls-dirs
foo1 [dir]
foo2 [dir]
The seq command (type man seq for help) can be used to generate an integer
sequence. The syntax is,
seq first increment last
which will generate an integer sequence starting at first, incrementing by increment, and stopping at last. For example, Here is a loop which displays the first
100 odd integers,
for i in `seq 1 2 100`; do
echo -n $i " "
done
Note that the seq command must be enclosed in backquotes `seq` so the shell
will execute it to generate the desired output. This loop prints the first 100 odd
integers in reverse order,
for i in `seq 99 -2 1`; do
echo -n $i " "
done
For more on for loops consult 69.
11.12 While Loops

The basic syntax of the while loop is,
while [ conditional-expression ]; do
statements
done
11.13 Expr and Substrings

Sometimes in a script, you would like to evaluate an expression. For example, to
find the sum of two variables. The expr command can be used to do this,
69
http://tldp.org/LDP/abs/html/loops1.html
59
$ cat testsum.sh
#!/bin/bash
v1=12
v2=39
v3=èxpr $v1 + $v2`
echo $v3
$ ./testsum.sh
51
$
Note that the expr command must be enclosed in backquotes `` for this to
properly work. Typing man expr will display a help screen about the expr command.
The expr substr command can be used to extract a substring from a string variable, e.g., this program prints the individual characters of a string stored in a variable v,
v="foobar"
len=èxpr length $v`
for i in `seq 1 1 $len`; do
echo -n èxpr substr $v $i 1`
done
echo
The syntax of expr substr is,

expr substr string pos len
which will extract len characters from the string string starting at position pos.
Note that the first character of the string string is at position 1.
60
12 Program Development, GCC, and GNU Make

*nix is the OS of choice for serious programmers because it was designed by,
and for, programmers. Any OS that was designed so that my mother could easily
use it (aka Windoze) is obviously of no use to someone who prefers to toggle his
programs into the computer on the front panel70 in binary.
Just as *nix is the OS of choice for serious
software development, GCC is the compiler
of choice for serious programmers.
GCC is an acronym for the Gnu Compiler
Collection (I used to think it meant GNU C
Compiler; that's false). It is a free, opensource set of compilers for building software,
and includes compilers for the C, C++, Ada,
Java, and FORTRAN languages.
GCC is part of the GNU toolchain71 and is the
name given to the set of programming tools
produced by the GNU project for developing
the GNU operating system. Programs included in the toolchain include:
GCC
GNU Make
GNU Binutils
GDB
A Real ComputerThe Altair 8800

(it did not come with an OS; why
would you need one when you've
got all those LED's and toggle
switches on the front panel?)
Compilers for various languages.

A program which facilitates automated building of software.
Includes an assembler, linker, profiler, and an archiver.
The GNU debugger.
In this course we will focus on learning to use the GCC C/C++ compilers, GNU
Make, and GDB. The compilers and GNU make will be discussed in this chapter.
The GNU debugger GDB is discussed in the next chapter.
12.1 Using GCC

The online documentation72 for GCC is worthwhile reading. In this section we will
discuss how to use the GCC C compiler to compile a program written in C. The
basic command syntax is:
$ gcc file.c
where file.c is the filename of the C program source code file. This command will
compile the code and link it with the C libraries to create an executable file. This
70
71
72
http://en.wikipedia.org/wiki/Front_panel
http://en.wikipedia.org/wiki/GNU_toolchain
http://gcc.gnu.org/onlinedocs
61
executable will typically be named a.exe in Cygwin and a.out in *nix. To run the
program,
$ ./a.out
12.1.1 Putting . In Your PATH

Note that the "./" is not necessary if you put "." in your PATH, although this is not
recommended.
If you do want to put "." in your path, you must edit the file ~/.bashrc. Note that ~
refers to your home directory, so .bashrc is a hidden file (because it starts with a
".") in your home directory. Hidden files do not appear when you do an ls command by default; use ls -a to see hidden files. Edit .bash_profile and add the following line to it.
PATH=${PATH}:.
Then log out, and log back in. This line appends "." to the end of the list of directories already set in the PATH environment variable. The shell will examine the
directories in the order they are listed in the path when you type a command. It
will execute the program from the directory in which it first finds the program executable file. Hence, because "." appears last, it will only run the program in the
current directory if it is not present in any of the other directories in the PATH.
Now to execute your program you can type:
$ a.out
12.1.2 Useful GCC Command Line Options

When you invoke the GCC C compiler to compile a C source code file, you can
also specify command line options73 which modify the default behavior of the
compiler. In this section we will discuss some of the more useful command line
options.
12.1.2.1 --version
Prints the version number of the compiler. Note the double hyphens.
$ gcc --version
gcc (GCC) 3.4.4 (cygming special, gdc 0.12, using dmd 0.125)
Copyright (C) 2004 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
73
http://en.wikipedia.org/wiki/Command_line_argument
62
The output above tells me that the version of the GCC C compiler I have installed
is 3.4.4, and that it was compiled for the Cygwin/Mingw platform (I am using
Cygwin right now).
12.1.2.2 -ansi
Tells GCC to compile the C code as if it conforms to ANSI Standard C. This is
equivalent to the command line option: -std=c89. Non ANSI Standard C code
should be rejected as syntactically incorrect.
12.1.2.3 -Dname[=def]
This option defines a macro named name. These macros are interpreted by the
C preprocessor74. One common use of preprocessor macros is to include code
that is conditionally compiled only if the macro is defined. For example,
/* file.c */
int main() {
#ifdef DEBUG
printf("Calling init()\n");
#endif
init();
#ifdef DEBUG
printf("Calling process()\n");
#endif
process();
#ifdef DEBUG
printf("Calling terminate()\n");
#endif
terminate();
return 0;
}
$ gcc -DDEBUG file.c
If we compile with DEBUG defined, then the printf() statements will be compiled
as well, so when we run the program, the debugging printf() statements will be
executed. If we compile with DEBUG not defined, then the printf() statements will
not be compiled. This is a handy way to put debugging printf() statements in the
program and decide at compile time if we want them to be output or not.
As another example, consider writing C code for a program which depends on
the endianness75 of the bytes in an integer. Generally in the Intel x86 world, multibyte integers are stored in little-endian format. Motorola's PowerPC architecture uses big-endian. In order to make our C code portable (i.e., so it will run on
both an x86 system or a PowerPC system), we might do something like the following:
74
75
http://en.wikipedia.org/wiki/Preprocessor
http://en.wikipedia.org/wiki/Endianness
63
/* file.c */
...
#if ARCH == x86
/* little-endian specific code here */
#elif ARCH == ppc
/* big-endian specific code here */
#endif
...
$ gcc -DARCH=x86 file.c
Or, in building the PowerPC version of the program,

$ gcc -DARCH=ppc file.c
12.1.2.4 -g
This option tells the C compiler to include additional debugging information in the
executable file so we can debug the program using GDB. If you do not compile a
file with -g, you can still use GDB to debug the program but symbolic information
(names and types of variables, line numbers, etc) will be missing.
12.1.2.5 -o file
Put the output in file. For example,
$ gcc file.c -o file
$ ./file
This command will compile and link the source code file file.c and will name the
executable file file rather than a.out.
12.1.2.6 -S (capital)
Stops the compiler after the assembly code generation phase, e.g.,
$ gcc -S file.c
This command will convert the code in file.c into functionally equivalent assembly language in the file named file.s. It will not assemble, link, and produce an
executable. This is a handy way to help yourself learn assembly language: write
some C code, compile it to assembly code, and study the assembly code to see
how it is implemented.
12.1.2.7 -c
Compile only, do not link.
$ gcc -c file.c
This command will compile file.c and produce the output file file.o. However, it
will stop at that point and not attempt to link to produce an executable.
64
Normally, when you make a project containing multiple C source code files, you
will compile all of the source code files individually using -c and compile the main
source code file last without using -c. We will discuss this in more detail when we
talk about makefiles later in this chapter. For example, suppose we have a
project with four C source code files: file1.c, file2.c, file3.c, and file_main.c:
$
$
$
$
gcc
gcc
gcc
gcc
-c
-c
-c
-o
file1.c
file2.c
file3.c
file file_main.c file1.o file2.o file3.o
produces file1.o
produces file2.o
produces file3.o
produces executable named file
12.1.2.8 -w
Inhibits all warnings. Warnings are messages from the compiler that something
about the code it is compiling seems fishy but it is not syntactically incorrect. The
canonical example would be,
if (a = b) {
...
}
This is perfectly legal C, but often it is a mistake because the programmer really
intended to write a == b. Some compilers will issue a warning message when
they compile this code to let the programmer know that what the compiler is seeing may not be what the programmer expected.
If you compile with -w these sorts of warning messages will not be displayed. In
general, warning messages are a good thing and your code should compile successfully without any errors or warnings. If you do get warnings from the compiler
you should fix them. For warnings which are issued by the compiler for code
which is okay, then pragmas76 or __attributes__77 should be used to inform the
compiler that what you wrote is what you intended.
12.1.2.9 -Wall
Turns on all warning checks. I highly advise you use this option and check all of
the warning messages generated by using it. This may be helpful in tracking
down difficult-to-find bugs which are the result of syntactically correct code.
12.1.2.10 Optimization Options
-O0 (i.e., dash capital-oh zero) turns off all optimization. This is the default. -O1,
-O2, and -O3 successively increase the amount of optimization that the compiler
performs on the final machine language code. -Os optimizes for size over speed.
76
77
http://gcc.gnu.org/onlinedocs/cpp/Pragmas.html
http://unixwiz.net/techtips/gnu-c-attributes.html
65
12.1.2.11 -Idir
(That is: dash capital-eye followed by dir). Add dir to the list of directories
searched for included header files. By default the GCC system directories and
the current directory will be searched.
12.1.2.12 -M and -MM
Useful for generating dependencies for makefiles. We will discuss this more
when we discuss makefiles.
12.1.2.13 -std=standard
Treats the code as conforming to C/C++ standard standard.
-std=c89
-std=c99
-std=gnu89
-std=gnu99
-std=c++98
-std=gnu++98
Is equivalent to -ansi; 1990 Standard C definition.

The 1999 Standard C definition.
1990 Standard C plus GNU extensions. Default for C code.
1999 Standard C plus GNU extensions
1998 ISO Standard C++
1998 ISO Standard C++ plus GNU extensions. Default for C++.
Nonstandard code elements will be treated as syntactically incorrect.

12.1.2.14 -llibrary
(That is: dash lowercase-ell followed by library). Search the library named library
when linking. We will discuss libraries and this command line option a bit later.
12.1.2.15 -Ldir
Search the directory named dir for library files in addition to the standard directories and the current directory.
12.1.2.16 --help
Prints help information about the compiler.
12.2 Using G++

The command g++ invokes the GCC C++ compiler. Most of the options for gcc
are the same as for g++, e.g.,
$ g++ -Ddbg=YES -I./include -L./lib -o foo -O3 -w foo.c bar.cpp baz.cpp
$ ./foo
Compile using the g++ compiler. Define the macro dbg to YES. Search the directory include located under the current directory for header files. Search the directory lib located under the current directory for library files. Send the output
executable to the file named foo. Optimize the heck out of the code. Disable all
warnings. And compile the files foo.c, bar.cpp, and baz.cpp.
66
Note that C++ programs commonly have an extension of .cc, .cpp, or occasionally .c++. C++ header files are usually either .h or .hpp or do not have a filename extension.
12.3 *nix Libraries

A good reference on libraries is available at 78.
A library contains compiled (binary) code and data that can be incorporated into
a program. Generally, libraries are created by adding one or more object files to
the library file. You can create your own library files, but more commonly in *nix
development, you will be linking your code with preexisting code contained in
various library files (which are usually installed in /lib and /usr/lib).
The advantages of libraries are that they contain code that has already been developed and presumably debugged so the functions in the libraries can be used
over and over again in different programs. This increases programmer productivity which is a good thing.
In *nix systems there are three main types of library files: static libraries,
shared libraries, and dynamic libraries. In current Linux distros, library files
and executable files are stored in a file format named ELF79.
12.3.1 Static Libraries
Static libraries are the simplest. A static library (file) contains just a collection of
one or more object code files (much as a zip file contains one or more files). Static libraries are identified with a .a suffix on the filename, e.g., /usr/lib/libm.a is
the math library in the Standard C library.
Static libraries are created with the ar command (ar standards for archive), e.g.,
$ pwd
/home/fredf/proj01
$ cd lib
$ pwd
/home/fredf/proj01/lib
$ gcc -c stack.c
produces stack.o
$ gcc -c list.c
produces list.o
$ gcc -c math.c
produces math.o
$ ar -rcs mylib stack.o list.o math.o
creates archive named mylib
$ cd ../src
$ pwd
/home/fredf/proj01/src
$ gcc -o myexe my-program.c some-other-file.c -L../lib -lmylib
$ ./myexe
78
79
http://www.faqs.org/docs/Linux-HOWTO/ Program-Library-HOWTO.html
http://en.wikipedia.org/wiki/Executable_and_Linking_Format
67
The option r to the ar command means insert the named file(s) into the archive
and replace any files already in the archive with the same filename. The c option
tells ar to create the archive without complaining about it if it does not exist. The
s option writes an index of the object files contained in the archive. There are additional command line options but these are sufficient to create a simple archive.
For more help, type man ar at the $ prompt.
The major disadvantage of static libraries is that each executable file that is produced by linking with the library will end up with identical copies of the object
code linked from the library. For example, in libm.a there is a function named
sin() that can be used to obtain the sine of an angle given in radians. If Bob
compiles a program executable named foo which uses sin() by linking with
libm.a then Bob's executable file will contain the machine language code for the
sin() function. If Alice does the same thing, then her program executable will also
contain the same machine language code.
Now consider when Bob and Alice both run their programs at the same time.
Each program will be loaded into memory with two identical copies of the machine language code for sin() taking up room in memory. This is not a very efficient allocation of resources.
12.3.2 Shared Libraries
A shared library is designed to overcome the major problem with static libraries.
When Bob links his code against the shared library, rather than including the machine language code for the sin() function in Bob's executable, a reference to the
machine language code will be inserted. This reference does not take up much
space. Alice's program will end up with the same sort of reference.
Now suppose Bob runs his program first. When it is loaded into memory, the OS
will check to see if the machine language code for the sin() function is loaded. If
it is not, it will be loaded. Then Bob's program code will be modified to point to
the address of the sin() function so that when Bob's program makes a call to
sin() it will jump to that address and execute the code of the shared sin() function. Alice's program will work similarly.
Shared libraries have a more complex naming scheme that static libraries. We
won't get into it in this course, but if you are interested, I would recommend reading the FAQ referenced at the beginning of this section.
The gcc command can be used to compile files and create a shared library. Also,
as with static libraries, we can link our code with existing shared libraries. In fact,
this happens automatically behind the scenes when we compile C programs calling common library functions such as printf(); gcc handles this for us.
68
12.3.3 Dynamic Libraries

A dynamic library is a library which is not linked with the program during the build
process. Rather, a running program can call the OS to open a dynamic library file
for reading, and can load in the machine language code for a desired function at
run-time. Then the running program can jump to the address where this code
was loaded which effectively calls the function. That's all I am going to say about
it.
12.3.4 Linking With a Static Library
The library file libm.a is the math library in the C Standard library. It contains
various mathematical functions such as sin(), cos(), etc. If we write a program
which calls the sin() function then we must link our code with the math library,
e.g.,
$ cat myprogram.c
#include <math.h>
int main() {
double f = sin(.434);
return 0;
}
$ gcc myprogram.c -o myprogram -lm
Note that gcc already looks in /lib for library files by default so we do not need to
use -L/lib on the command line to tell gcc to look there. Note also that the syntax
-lm means link with the library named libm.a. In general, the syntax -lfoo means
link with the static library named libfoo.a (gcc will look in /lib and /usr/lib). Finally,
note that because the linking stage comes after the compilation stage in the build
process, the library file must be listed after the C source code file and output file.
It is an error to do this,
$ gcc -lm myprogram.c -o myprogram
Most of the standard *nix libraries are found in the /lib and /usr/lib directories,
e.g., /usr/lib/libc.a is the Standard C librarywhich is linked automatically so we
do not need to include -lc on the command line of the gcc command.
Many Standard C library header files are in /usr/include, e.g., /usr/include/stdio.h.
Use #include <> for system libraries. The compiler will know to look in /usr/ include and its subdirectories for the header file.
On the other hand, using #include "" will tell the compiler to look in the current
directory and any directories specified with the -I command line option for the
header file. In general, <> is used for system header files, and "" for your own
header files.
69
12.4 Makefiles
Most software projects of a reasonable size consist of multiple source code files,
for very large projects perhaps thousands. Consider these numbers,
Debian 4.0
Vista
Mac OS X 10.4
283 MLOC (million lines of code)

> 50 MLOC
86 MLOC80
The project may also include libraries (prebuilt as well as built) and many other
types of resource files.
Dependencies: One source code file (A) may depend on another source code
file (B). Changes to B may cause A to change. In GNU terminology dependencies are called prerequisites.
One of the problems with large software projects containing hundreds or thousands of files is that we don't want to recompile every source code file every time
we do a build. Therefore, some automated way of recompiling only those source
code files that need to be recompiled is desired.
12.4.1 Building a Project the Hard Way
Now the hard way to build a multifile program would be,
$ gcc -c file1.c
wait
$ gcc -c file2.c
wait
$ gcc -c file3.c
wait
...
$ gcc -c file100.c
wait
$ gcc file1.o file2.o ... file100.o -I./include -o myprog -L./lib -lmylib
$ ./myprog
Oops, found a bug. Edit file4.c and correct the bug. Start typing again. Sigh.
12.4.2 Building a Project with a Script File
A slightly better approach would be to put all of these commands in a shell script:
#!/bin/bash
#File: b.sh
gcc -c file1.c
gcc -c file2.c
gcc -c file3.c
...
gcc -c file100.c
gcc file1.o file2.o ... file100.o -I./include -o myprog -L./lib -lmylib
80
According to Wikipedia (http://en.wikipedia.org/wiki/Source_lines_of_code); these numbers may be questionable.
70
$ alias b=./b.sh
$ chmod 755 b.sh
$ b
go get some coffee
You would know the build is complete when the $ prompt comes back. Note that
you could send the build to the background,
$ b&
$ tetris
do something productive now
Problems with this approach? Suppose our program grows to contain 1000 files.
Every time a new source code file is added to the project, someone has to edit
the b.sh script to add the new file. Or, every time a source code file is removed
from the project, someone has to edit the b.sh script to delete the file. Or, every
time we rename a source code file... I hope you get the idea.
Furthermore, there is nothing in the script file that determines if a source code file
really has to be recompiled. Note that the only time a source code file must be
recompiled is when it's text is modified. Recompiling 100 files over and over
every time we perform a build, when only one or two of those files has changed
is simply too inefficient (unless you're on a contract and getting paid by the hour).
Surely there has to be a better way.
12.4.3 Introduction to Make
Make is a common *nix utility that was designed to automate the build process.
According to Wikipedia81, the first version was created at Bell Labs in 1977.
Make has evolved over the years and there are many variants of it (e.g., Apache
ANT and Maven for Java programmers). GNU Make82 is the GNU project's version of make (online documentation83).
Make is an extremely complicated program and would require us to spend much
more time than we have to fully discuss it. Therefore, we will focus on a very
simplified make process to give you a feel for how it is used. I highly recommend
you read the online documentation for further understanding.
Make processes one or more makefiles which contain rules (or instructions) for
how to build the program. The syntax of these makefiles is, in usual *nix fashion,
weird if you don't understand it84.
81
82
83
84
http://en.wikipedia.org/wiki/Make_(software)
http://www.gnu.org/software/make/
http://www.gnu.org/software/make/manual/make.html
It's weird even if you do understand it.
71
We will start with a simple makefile and build upon it to make (ha ha) a complete,
final makefile. Suppose my program is comprised of two C source code files
main.c and mutils.c. Here is an initial, reasonably simple makefile,
#!/bin/bash
# File: Makefile
myprog: main.o mutils.o
gcc main.o mutils.o -o myprog
main.o: main.c
gcc -c main.c
mutils.o: mutils.c
gcc -c mutils.c
Note that the lines starting with gcc are indented with a hard tab, i.e., not spaces. The words myprog, main.o, and mutils.o are called targets. They specify
what make is to build. For the target myprog, main.o and mutils.o are dependencies (or prerequisites). This says in English that the target myprog depends
on main.o and mutils.o. If either one of main.o or mutils.o changes, then myprog is "out of date" and must be rebuilt.
How does make know when a file has changed? It looks at the timestamp, i.e.,
the date and time the file was last modified. If main.o was last modified (written)
at 10/01/2007 11:15:23pm and main.c was last modified at 10/01/2007 11:18:21
then make assumes main.c has changed since the last time it was rebuilt (because the timestamp on main.c is more recent).
The target myprog is the default target because it appears first in the file. If we
do not specify a target for make, it will attempt to build the default target.
$ make Makefile
gcc -c main.c
gcc -c mutils.c
$ make makefile
make: 'myprog' is up to date.
$
Note that typing make is equivalent to typing make Makefile, i.e., make will always try to read commands from Makefile; if that file cannot be found it will look
for makefile.
12.4.4 Make Command Line Options
Make's default behavior can be modified with command line options. Some of the
more useful ones are described here.
-f file
Read make commands from named file rather than from Makefile.
72
-d
Display debugging info. Makefiles can be extremely complex, so

this option is useful when your makefile is not working correctly.
-n
Do everything but actually execute each of the make commands.

Useful to see what make is going to do before you run it.
-s
Silent mode, no output.
--version
Print the version of make. Sometimes when you are building a

program on *nix from source the instructions will tell you that you
need a specific version of make.
-B
Make everything unconditionally without regard to timestamps.
-i
If a command generates an error keep going.
-k
If a target cannot be made keep going.
--help
Display help information.
12.4.5 Automatic Variables and Implicit Rules

Suppose we have a project with 1000 C source code files:
#!/bin/bash
# File: Makefile
myprog: file1.o file2.o file3.o ... file1000.o
gcc file1.o file2.o ... file1000.o -l<libs> -o myprog
file1.o: file1.c
gcc -c file1.c
file2.o: file2.c
gcc -c file2.c
...
file1000.o: file1000.c
gcc -c file1000.c
There has to be a better way to specify that every .c file should be recompiled to
a .o file when the .c file is out of date. There is, using implicit or suffix rules. Try,
$ make -p > make-defaults.txt
$ less make-defaults.txt
CC = cc
OUTPUT_OPTION = -o $@
...
COMPILE.c = $(CC) $(CFLAGS) $(CPPFLAGS) -c
...
%.o: %.c
$(COMPILE.c) $(OUTPUT_OPTION) $<
The rule starting with %.o is a suffix rule. The characters %.o specify that this is
a pattern rule. The characters $@ refers to the name of the target .o file. The
characters $< refers to the name of the prerequisite .c file. Collectively, $@ and
$< are called automatic variables in make terminology (there are other automat73
ic variables as well, you should read the online manual about $? which is also
very useful).
Therefore, since make already has a predefined (or built-in) rule for building a .o
file from a .c file, we can modify our makefile:
#!/bin/bash
# File: Makefile
myprog: file1.o file2.o file3.o ... file1000.o
gcc file1.o file2.o ... file1000.o -l<libs> -o myprog
Or,
#!/bin/bash
# File: makefile
Now make,
$ make
cc
-c -o main.o main.c
cc
-c -o mutils.o mutils.c
$
12.4.6 Compiler Command Line Options

We see a macro variable $(CFLAGS) was defined and used in the implicit rule
for building a .c file. We can use this variable to specify our own compiler command line options,
#!/bin/bash
# File: makefile
CFLAGS=-O2 -g
Then make,
$ !m
make
cc -O2 -g
cc -O2 -g
$
-c -o main.o main.c
-c -o mutils.o mutils.c
Suffix rules for C++,

CXX = g++
COMPILE.cpp = $(COMPILE.cc)
COMPILE.cc = $(CXX) $(CXXFLAGS) $(CPPFLAGS) -c
74
Thus, we can modify $(CPPFLAGS) to specify g++ compiler command line options.
12.4.7 Handling Header Files
Notice that mutils.h is not listed in the makefile. Header files are not compiled
separately, but .c files do depend on them, e.g., if we change mutils.h then we
have to recompile main.c because main.c includes mutils.h (in other words,
changing the text of mutils.h is equivalent to changing the text of any .c file
which includes mutils.h).
Our makefile does not work correctly then because if I change mutils.h and perform a make, then make tells me that myprog is up to date. So how do we specify that changes to .h files should cause .c files that depend on them to be recompiled? We modify the prerequisites for the .o file to include the .h file(s) as well as
the .c file(s),
#!/bin/bash
# File: makefile
CFLAGS=-O2 -g
main.o: main.c mutils.h
mutils.o: mutils.c mutils.h
The line starting main.o basically says that main.o depends on main.c and mutils.h. A change to one or both of those files would cause main.c to be recompiled to produce an up-to-date main.o.
But, wouldn't having to remember and list which .h files a .o file depends on get
tedious for a large project? (Trust me, it would.) What if I add one or more new
#include header files to a .c source code file, or remove one or more header
files? Sounds like a headache. Never fear, gcc to the rescue. The -M option to
gcc will generate a dependency rule listing all of the .h files that a .o file depends on,
$ gcc -M main.c
main.o: main.c /usr/include/stdio.h /usr/include/_ansi.h \
/usr/include/newlib.h /usr/include/sys/config.h \
/usr/include/machine/ieeefp.h /usr/include/cygwin/config.h \
/usr/lib/gcc/i686-pc-cygwin/3.4.4/include/stddef.h \
/usr/lib/gcc/i686-pc-cygwin/3.4.4/include/stdarg.h \
/usr/include/sys/reent.h /usr/include/_ansi.h
/usr/include/sys/_types.h \
/usr/include/sys/lock.h /usr/include/sys/types.h \
/usr/include/machine/_types.h /usr/include/machine/types.h \
/usr/include/sys/features.h /usr/include/cygwin/types.h \
/usr/include/sys/sysmacros.h /usr/include/stdint.h \
/usr/include/endian.h /usr/include/sys/stdio.h
/usr/include/sys/cdefs.h \
mutils.h
$
75
Note that system header files not do change that often (usually only when a
newer version of the compiler is installed). Hence, they do not need to be listed in
the makefile. Use the -MM command line option to gcc to omit system header
files:
$ gcc -MM main.c
$
Our new modified makefile is:

#!/bin/bash
# File: makefile
CFLAGS
= -O2 -g
SOURCES = main.c
\
mutils.c
%.d: %.c
rm -f $@; gcc -MM $< > $@
include $(SOURCES:.c=.d)
The %.d: %.c rule says that a file with a .d extension depends on a file with a .c
extension, i.e., foo.d depends on foo.c. The automatic variable $@ refers to the
target dependency (the .d file) and $< to the prerequisite dependency (the .c file).
When this rule is applied to a .c file, the corresponding .d file is deleted (because
of the "rm -f $@" command. Then "gcc -MM" is performed on the .c file and the
output is redirected to the .d file (this is the "> $@" part). For main.c and mutils.c
this would create the files main.d and mutils.d:
$ cat main.d
$ cat mutils.d
mutils.o mutils.d : mutils.c
$
The SOURCES= line defines a macro variable which contains the names of all of
the .c files in the project. The include line states that in the $(SOURCES) macro
variable, every filename that ends with .c should be changed to .d. Then everyone of those .d files are included in the makefile. This is equivalent to writing the
line,
include main.d mutils.d
but is more flexible, because now if we add a new C source code file to the
project, all we have to do is add the name of the source code file to the
SOURCES= line. Now that we know that trick, we can make another change,
76
#!/bin/bash
# File: makefile
BINARY = myprog
CFLAGS = -O2 g
SOURCES = main.c
\
mutils.c
OBJECTS = $(SOURCES:.c=.o)
# the name of the executable file

# options for the C compiler
# a list of the .c files in the project
# a list of the .o files in the project
$(BINARY): $(OBJECTS)
# the binary depends on the object files
gcc $(OBJECTS) -o $(BINARY) # link the object files to produce the binary
%.d: %.c
rm -f $@; gcc -MM $< > $@
# the .d files depend on the .c files

# create a .d file by running gcc MM
# include the contents of all of the .d files
Now, that's a pretty nice looking makefile isn't it?

12.4.8 Phony Targets
A phony target is a target that is not built, but rather is placed in the makefile to
do something useful, e.g.,
#!/bin/bash
# File: makefile
BINARY = myprog
CFLAGS = -O2 g
SOURCES = main.c
\
mutils.c
OBJECTS = $(SOURCES:.c=.o)
# the name of the executable file

# options for the C compiler
# a list of the .c files in the project
# a list of the .o files in the project
$(BINARY): $(OBJECTS)
# the binary depends on the object files
gcc $(OBJECTS) -o $(BINARY) # link the object files to produce the binary
%.d: %.c
rm -f $@; gcc -MM $< > $@
# the .d files depend on the .c files

# create a .d file by running gcc MM
# include the contents of all of the .d files
.PHONY: clean
clean:
rm -f $(OBJECTS)
rm -f *.d
rm -f myprog
# "clean" is a phony target

# delete all of the .o files
# delete all of the .d files
# delete the binary
Now run,
$ make clean
rm -f main.o mutils.o
rm -f main.d mutils.d
rm -f myprog
$
Open-source programs that you might download from the Internet and build on
your GNU/Linux box often come with makefiles provided. A common way to build
the software and install it is to perform the commands,
$ ./configure
$ make
$ make install
77
where configure is a shell script that actually creates the Makefile file and install is a phony target which causes make to copy/move the executable and other installation files to their final destinations (e.g., to /usr/local/bin).
As I mentioned earlier, make is a very complicated program. We have only
scratched the surface of what it can do here, but I urge you to read the online tutorials and manuals and learn more about it. Someday you will make your employer happy that you did.
78
13 Debugging
13.1 Text-Based Debugging with GDB
13.1.1 Launching and Quitting gdb
To start gdb and debug the executable a.out, at the command prompt type gdb
a.out. Note that in order to debug a.out you must have built a.out with the gcc -g
option; it also best if you do not compile with optimization, i.e., use -O0. You will
see a screen similar to the following. The gdb prompt is (gdb) which means it is
waiting for input.
To leave gdb, use the command quit (abbreviated q) or hit Ctrl+D.

13.1.2 Running Your Program
To run your program non-stop, type run or just r. Your program will either run to
completion or crash. If it runs to completion and produces the correct output, then
what are you using the debugger for? On the other hand, mine crashed with a
segmentation fault,
The normal way to generate a segmentation fault is to try to read from or write to
memory that does not belong to you. Seems simple enough to avoid that, but
when you're using pointers, crazy things can happen. The dump tells me which
program I was running, ".../Homework02/a.out". I was executing code in a function named set_elem() and you can see the input parameters to the function
were array which was assigned the value 0x501010, row which was 1, col which
was 0, and value which was 1. This code is on line 30 of the source code file
Assgn03-04-Bonus.c. The very last line in the display (starting with 30) is the
source code statement on line 30 which caused the exception.
79
Now that I know the line number of the statement which caused the exception, I
can go back to my source code and examine it to see if I can find the bug. Note
that just because that is the line we were executing when the segmentation fault
occurred does not mean that that is the line where the actually bug lies. C is an
extremely powerful and exceedingly dangerous language. It is entirely possible
that some other statement somewhere accidentally modified this block of memory I am now accessing and trashed it. Hence, the bug may not be on this line of
code but somewhere else. These bugs are hard as heck to find; I once spent two
8-hr work days tracking one down.
13.1.3 Listing Source Code
Gdb is a source code debugger, which means you can display your source code
as you are debugging and also view the assembly language code. You can step
through your program line by line executing instructions and viewing the contents
of registers and memory. This is extremely powerful when attempting to locate
bugs. To print source code lines, use one of the variants of the list command
(abbreviated l; that is, lowercase ell):
list linenum
Prints lines centered around line number linenum. The number of lines is controlled by set listsize count. The default
value for count is 10. To see the current value of count, use
show listsize.
list function
Prints lines centered around the beginning of function. If

count is ten, five lines before the function will be printed, and
the first five lines of the function will be printed.
list
Prints more lines following the lines printed by the previous

list command, i.e., scroll down.
list -
Prints more lines preceding the lines printed by the previous

list command, i.e., scroll up.
list first,last
Prints lines from first to last, e.g., list 1,100.
list file:number
Prints line number from source file file.
list file:function
Prints lines centered around function function from source

file file.
13.1.4 Examining Assembly Language Code

Your C or C++ source code is compiled into assembly language. You can examine this assembly language code in gdb. To see the assembly language code
you use the disassemble (or disas) command.
disas
Disassembles code centered around the current PC.
disas addr1
Displays disassembled code starting at memory address

addr1.
80
disas addr1,addr2
Displays disassembled code from memory address addr1

to memory address addr2.
To find the memory address of the beginning of a line of source code, use info
line line, e.g., info line 30. Gdb will respond with the starting memory address of
the assembly language code for the statement and the ending memory address
(a single HLL statement may compile down to multiple assembly language instructions). The command x/i will begin displaying instructions or you can use
disas addr1, addr2 to display the entire range. Pressing x repeatedly after x/i
will display one instruction at a time.
13.1.5 Examining Data and Variables
When your program is running, or when it crashes, you may want to see what
values were in certain memory locations at the time of the crash. This is done
with the print or p command.
p file::var
Will print the value stored in global variable var defined in

file.
p var
Will print the value of the variable var if it is in scope. If it is

not in scope, then strange things may happen.
13.1.6 Examining Memory

Sometimes you want to look at memory directly as a bunch of bytes than as variables. For example, I might want to know what values are stored at memory locations 0x501010 through 0x501030. The command to do this is x (for examine).
x addr
Display the value at memory location addr.
x/5 addr
Display five words (a word is 4-bytes) starting at addr. In

general use x/n to display n words.
x/3d addr
Display three words starting at addr. The d means display

the values as signed decimal numbers. The default is to display in hex (you would use x for that). Use u for unsigned integers, o for octal, t for binary, c for character, f for float, and
s for string. Use i to interpret the words as machine language instructions.
x/4b 0x501010
Display four bytes starting at 0x501010 in hex format.
x/4t 0x501010
Display four bytes starting at 0x501010 in binary format.
x/fg 0x501010
Displays a double (8-bytes) at 0x501010. g is for giant (a

double to gdb is a giant float).
x/5fg 0x501010
Displays five doubles start at 0x501010.
Continuing to hit x will display successive memory locations.

81
13.1.7 Examining Array Variables

If a is an array then p a will print the address in memory of the beginning of the
array. Remember that the name of an array variable is equivalent to the address
in memory of the beginning of the array. To display the entire array use the
command p *array@len where len is the number of words in the array, e.g., if a
is an array of 10 integers, then use p*a@10 to display the 10 integers in a.
13.1.8 Setting Watches or Automatic Displays
If you want to view the contents of a variable or memory location each time your
program stops (e.g., if you are single-stepping through the program), then add
the variable or address to the automatic display list: display var or display addr.
Each display in the list is numbered starting at 1.
display/5t 0x501010
Displays five words starting at 0x501010 in binary

format every time the program stops.
display/i $pc
Displays the machine language instruction pointed to

by the PC register. This is the instruction to be executed next.
delete display 1
Delete display number 1.
disable display 2
Disable display number 2 but don't delete it.
enable display 2
Enable display number 2.
info display
Display the display list.
13.1.9 Changing the Values of Variables

Use set var varname=value if the variable is in scope. If the variable is not in
scope, you cannot do this.
13.1.10 Examining Registers
To display the register names and their values, use info registers. To print the
program counter register, use p $pc. The command x/i $pc will print the instruction pointed to by $pc. The stack register is referred to by $sp. To set a register
use set regname=value, e.g., set rcx=0x100.
13.1.11 Stepping
Normally we use a debugger because our program has bugs in it that we need to
locate and fix. In these cases, you often want to step through the program executing one line at a time, examining the values of variables and memory, until
you find the point where something bad is happening. When you load your program in gdb, if you hit r or run it will run non-stop until it finishes or crashes.
To step through the program one line at a time, use the step or s commandbut
only after typing start to start program execution. Step will execute one source
code line and then stop (remember that each source code line is compiled into
82
one or more assembly language instructions). The actual bug will be found during execution of one of these assembly language instructions.
step n
Abbreviated s n. Execute n source code instructions.
next
Abbreviated n. Executes the next source code instruction,

but jumps over function calls. Normally step will step into
function calls. Next does notever. This is useful if you
know the function you are calling does not have any bugs in
it, and you just want the function to execute so you can get
on with your life.
next n
Or n n. Like step n but doesn't execute into functions.
finish
Run until the function you are in returns, then stop and display the return value.
until
Typically used after you begin stepping through a loop. It will

run nonstop until the end of the loop. Typing until will continue execution until the loop terminates, whereas next
would execute the loop again by going back to the beginning.
stepi
Abbreviated si. Execute one machine language instruction. Using display/i $pc in conjunction with si is an easy
way to display each machine language instruction before it is
executed, and then execute the instruction and stop. That is,
you would type display/i $pc once, and then you could do
si, si, ... to single-step through the program executing one
machine language instruction at a time. Note that stepi will
step into functions.
si n
stepi n. Execute n assembly language statements. Will step

into functions.
nexti
Abbreviated ni. Execute one assembly language statement

and execute over function calls (i.e., don't step into it).
ni n
Execute n assembly language statements executing over

function calls.
Once you enter one of these commands, for example, s or si, continuing to hit
ENTER will execute the same command again.
13.1.12 Breakpoints
A breakpoint is a marker which tells the debugger to stop at that marker. For
example, you can set a breakpoint on a certain source code line, and then run
your program with r. When that line is reached during execution, the debugger
will stop. You can then examine the state of memory and variables to try and find
the bug. Then you could single step from that point on, or you could continue
83
running until the next breakpoint is reached, e.g., if the line containing the breakpoint is reached again during execution, the debugger will break again.
Breakpoints can be set on source code lines or machine language instructions.
Often it is helpful to identify the function where the bug is, then to find the line
within the function, and then to finally identify the machine language instruction
containing the bug if need be. Breakpoints are set with the break or b command.
Breakpoints are disabled with the disable command. Break points are enabled
with the enable command. Breakpoints are deleted with the delete (d) commands. Information on breakpoints is obtained with the info (i) command.
continue
Abbreviated c. Continue execution after a breakpoint

is triggered until the next breakpoint.
b function
Set a breakpoint at the entry point of function.
b linenum
Set a breakpoint at source code line linenum in the

current source code file.
b file:linenum
Set a breakpoint at linenum in source code file file.
b file:function
Set a breakpoint at function in source code file file.
b *addr
Set a breakpoint at memory address addr.
tbreak args
The argument args are the same as for the break

commands discussed above, but in this case the
breakpoint is temporary, i.e., the debugger will break
once but then the breakpoint is deleted.
rbreak rexpr
Argument rexpr is a regular expression. This command will set a breakpoint on all functions matching
the rexpr pattern. Since the regular expression .
matches every function, rbreak . will set a breakpoint
on every function in your program. Then you can hit c
to continue execution until the next function gets
called.
info break
Abbreviated i b. Displays a table of breakpoints.
disable bps
Abbreviated dis. Disable all breakpoints in the range,

e.g., disable 2 would disable breakpoint number 2
(see info break). disable 2 3 5 would disable breakpoints 2, 3, and 5. Disable [2-5] would disable breakpoints 2, 3, 4, and 5. If bps is not given then all
breakpoints are disabled.
enable bps
Abbreviated en. Arguments are similar to disable.

Enables breakpoints.
84
en bps once
Enables the breakpoints once, then once each breakpoint is hit, disables the breakpoint again.
en bps delete
Enables the breakpoints once, then once each breakpoint is hit, deletes the breakpoint.
delete bps
Abbreviated d. Arguments are similar to disable and

enable. Deletes the breakpoints.
condition bnum expr
Abbreviated cond. Sets a condition on break number

bnum. The condition might specify that the value of a
variable local to a function be equal to a certain value,
e.g., if we do b foo to set a breakpoint (numbered 2)
on function foo(), and foo() has a local variable named
x, then we can set a condition on the breakpoint for
foo(): condition 2 x==4 says the breakpoint is triggered only if x is equal to 4.
cond bnum
Remove the condition on breakpoint number bnum.
An especially useful thing is to be able to break on a breakpoint only after it has

been triggered n times. For example, set a breakpoint on function foo(), but specify that the breakpoint is to be triggered only on the fifth call to foo(). This is
done by ignore bnum count. For example, ig 2 5 if the breakpoint on function
foo() is breakpoint number 2.
13.1.13 Watchpoints
A watchpoint is similar to a breakpoint. You instruct the debugger to stop execution if the value of variable changes (we could call these data breakpoints).
watch foo
Watches variable foo and breaks if it is written to.
rwatch foo
Breaks if foo is read.
awatch foo
Breaks if foo is read or written (a means access).
info wat
Displays a table of watchpoints.
The expression to watch does not have to be a simple variable. We could tell gdb
that we want it to watch a certain region of memory and break when that region is
read from or written to.
watch 0x501010
Watch memory location 0x501010 for a write.
watch 0x500-0x600
Watch memory range 0x500 through 0x600 for writes.
13.1.14 Miscellaneous Gdb Commands

The command shell (she) can be used to execute a shell command without leaving gdb, e.g., try shell ls. The command shell make will rebuild the project in the
85
current directory. If the make succeeds, you can reload the executable using file
a.out.
To log the output of a debugging session, use the set logging on command. The
default log file is gdb.txt. Use set logging file file to change the logging file from
gdb.txt to file. By default gdb appends to the logging file; use set logging
overwrite on to switch to overwrite mode. Use set logging overwrite off to
switch back to append mode.
Gdb will repeat the previous command if you just press the RETURN key. For
example, if you are single-stepping through the program using si, just type si
once, and from then on you can just keep hitting the RETURN key to continue
single stepping.
By default, gdb assumes numbers you enter are in decimal. Use 0x to specify
hex. To change the input radix to hex permanently, use set input-radix 16. From
now until the end of your debugging session, gdb will assume numbers you enter
are in hex without you having to type 0x.
Getting help is accomplished by using help or Google.
13.2 GUI Debugging with DDD

Gdb is a powerful debugger, but it's
interface is relatively primitive compared to today's GUI's. It is possible to
write a graphical front-end for gdb
which interacts with gdb on the backend by submitting text commands in
response to the user's actions with the
mouse and keyboard while interacting
with the GUI. One such program is
DDD, the Data Display Debugger85.
command tool
Since DDD is a graphical program,
you cannot run it while logged in to re
search.asu.edu in text mode. You can
run it on the GNU/Linux systems in the
Brickyard computer lab under Fedora,
or if you are running Cygwin/X on your Windows machine, then you can open a
ssh connection to research.asu.edu, and run ddd from there; ASU has version
3.3.9 installed. The latest version on the DDD website is 3.3.12. (Of course if
you're running Cygwin, there is also a port of DDD for Cygwin that you would
probably be better off installing; it is version 3.3.9 as well).
85
http://www.gnu.org/software/ddd
86
In DDD you can interact with gdb through the console window at the bottom of
the DDD window. The command tool is turned on by selecting View | Command
Tool.
To set a breakpoint, move the mouse pointer to
the left margin of the source line where you want
to set a breakpoint, and right-click. Select Set
Breakpoint from the popup menu. A stop sign
icon will be displayed on the line.
To start running the program, click Program |
Run from the main menu, or click Run on the
command tool. This will run the program non-stop. To start running and stop on
the first line of main(), type start in the console window, or set a breakpoint on
the first line of the code where you want to stop.
To inspect the value of a variable, right click on it and select Print. It's value will
be printed in the console window. To add the variable to the display, right-click on
it, and select Display. If you don't see the display pane, click View | Data Window on the main menu. If you hold the mouse pointer over a variable, after .5
seconds a popup box will appear with the value of the variable.
Clicking View | Machine Code Window will show you the disassembly listing.
You can set breakpoints here as in the source code listing window.
To see the contents of the registers, click Status | Registers.
To display line numbers in the source code listing window, click Source | Display Line Numbers.
If you click Source | Edit Source, VI will be loaded with your source code. You
can make changes to the code, save it in VI, exit VI, and the code will be reloaded into DDD.
If you want to remake the project without exiting DDD, you can type make in the
gdb console window to make your project (assuming you have a makefile). The
issue the file your-exec-file command to reload the remade binary file.
To view information about the variables of the current stack frame, click Data |
Status Displays. Then click Local Variables of Current Stack Frame. The local variables will be displayed in the data window and will be updated as the program is being executed step-by-step, see below.
87
To examine memory, click Data | Memory. In the box that pops up you can select the format for the display (e.g., decimal, binary, hex, char), the number of
words to display, and the starting memory address. Clicking print will dump the
memory contents to the gdb console window. Clicking display will display the
memory contents in the data window (see below).
Well, that's it for this basic introduction to gdb and DDD. For more info, consult
the online documentation.
88

01) Linux - Lab Manual

Uploaded by

01) Linux - Lab Manual

Uploaded by

GNU/Linux Lab Manual

School of Comp. Info. and Dec. Sys. Eng.

8.1 Redirection Example ................................................................................. 41

12.1 Using GCC .............................................................................................. 61

1 Introduction to Unix and GNU/Linux

1.2 The GNU Project and the Linux Kernel

1.3 GNU/Linux Distributions

Members include IBM, HP, and other major Unix players.

1.4 Other Unix and Unix-Like Distros

Kubuntu is Ubuntu with the KDE GUI environment.

http://www.centos.org; It stands for Community Enterprise Operating System.

2.1 KDE, Konqueror, the Terminal Window

2.2 Logging Out

icon and select Log Out from the

2.3 Logging In To "research.asu.edu"

2.4 Logging Out Of research.asu.edu

2.5 Transferring Files with WinSCP

WinSCP can be programmed to automatically upload and download files using

3.1 The Bash Shell

4 The *nix File System and Common Commands

4.1 Common Directories in *nix

4.3 Commands, Arguments, and Options

In addition to arguments, many (most) *nix commands accept command options

4.4 Common File and Directory Commands

4.5 How Do I Know Which Directory I Am In?

4.6 List Files, File Permissions, The Chmod Command

The option -a will show all the files in the directory,

The special file .bash_profile is a configuration file which is used to configure

Oct 10 10:04 a.out

Oct 10 10:04 a.out

20987 Oct 10 10:04 a.out

4.7 Deleting Files

4.8 Creating, Changing, and Removing Directories

Suppose we have created directories cse220 and cse220/assgn02 and that

Before deleting a directory, it must be empty,

The last command would remove the file /home/fredf/file1.c.

4.9 Filename Globbing

A wildcard is a pattern which matches something else. Two commonly used

4.10 More on Deleting Subdirectories

this deletes cse220 and everything in it

The following is a very dangerous command,

4.11 Copying Files and Directories

To copy all the .c files from one directory to another,

file01.c file02.c file03.c file01.o file02.o file03.o

To copy a file from another directory to the current directory,

4.12 Moving and Renaming Files

4.13 Finding a File

4.14 Displaying a File at the Terminal: Cat and Less

Frequently-Used Less Commands

run the shell and execute cmd

4.15 Archiving Files

4.16 Compressing Files

To decompress every gzipped .c file in a directory,

To zip a tar file, just gzip or bzip2 it like a normal file,

like --backup but does not accept an argument

same as --no-dereference --preserve=links

follow command-line symbolic links in SOURCE

Jumping From Node File to Another

5.4 Final Notes

6 Environment Variables, Search Path, Prompt

To add . and sbin to your path,

6.1 Changing your Prompt

string is defined in my .bashrc file as PS1='\[\033[1;32m\]\w $\[\033 [0m\] '.

8 I/O and Redirection

8.1 Redirection Example

8.2 Redirection Example

user types 2 and presses enter

8.3 Redirecting Stderr

8.4 Sending Output to /dev/null

8.5 Concatenating Files

8.6 Sending Stdout and Stderr to the Same Place

8.7 Appending to a File

10.1.5 Moving Around a Character at a Time

10.1.11 Deleting a Line of Text

10.2 KDE - The Kate Editor

10.3 Other Linux Editors