Chapter 2: Using Linux at the command line

Chapter 2
Using Linux at the command line

Last revised: 20/6/2004

Chapter 2 Outline

In this chapter we will learn how to:

Use the linux shell productively

Control access to files
Combine key linux filter programs using pipelines
Navigate the filesystem and manage files
Edit text files with vi
Get help through on-line documentation

Basic use of the shell

Basic use of the shell

Logging in and out
Simple commands
Command options
Command arguments
Command history
Absolute and relative
pathnames
File name completion

Logging in

After booting, a desktop SuSE Linux system will usually present a

graphical login dialog

Enter your login name and password (both are case sensitive)

KDE desktop environment started

Systems (eg servers) which do not run a graphical desktop will

present a command line login

Enter login name and password

A 'shell' (command interpreter) is started

Our machines are currently configured for a command line login

We will reconfigure them to use a graphical login and KDE desktop later

Please log in now:

Log in as the user 'tux'

Supply the password 'penguin'

Do now!

Logging out

To logout from a command-line environment:

Enter the end of file character (usually ^D)

Or type the command exit

Please:

Log out

Log back in again

Do now!

The shell

The 'shell' is the linux command interpreter

The shell operates in a command processing loop:

Displays a 'prompt' and reads a command line

Peforms various substitutions and expansions on the command line

Executes the resulting command and waits for it to finish

Loops back and prompts for another command

Several shells have been written for UNIX and Linux

Bourne shell (sh), Korn Shell, C Shell, Bourne Again Shell (bash)

The core feauture set of all these shells is very similar

We will focus on bash, the most popular shell on Linux

Simple commands

Try these simple commands:

Do now!

$ hostname
Reports the name
snowhite
of this machine
$ date
Fri Apr 16 11:48:33 BST 2004
$ id
uid=500(chris) gid=100(users) groups=100(users),14(uucp)
$ cal
April 2004
Su Mo Tu We Th Fr Sa
For simplicity we show the prompt
1 2 3
as a '$'. The default configuration
4 5 6 7 8 9 10
of SuSE linux uses a longer prompt
11 12 13 14 15 16 17
which includes your login name,
18 19 20 21 22 23 24
hostname and current directory
25 26 27 28 29 30

Command options

Command options modify the behaviour of a command

Usually, an option is a single letter prefixed by '-'

$ cal -y
... calendar for the entire year ...
$ date -I
Options are case sensitive
2004-04-16
The date in ISO format

Some commands also have 'long' options

Begin with '--'

Supplement or replace the traditional single character options

$ date --iso-8601
2004-04-16

Same as '-I'

Command arguments

Most commands accept arguments

The command name, options, and arguments are separated by

whitespace (spaces and tabs)

Arguments are often the names of files or directories on which to operate

$ cal 1955
... calendar for the year 1955 ...
$ ls /home
chris dilbert tux

ls lists the contents of the

specified directory /home

Options and arguments are often used together:

$ ls -l /home
total 6
drwxr-xr-x
66 chris
drwxr-xr-x
17 dilbert
drwxr-xr-x
18 tux

users
users
users

The -l option requests

a 'long' listing
3328 2004-04-16 11:48 chris
1112 2004-02-09 11:52 dilbert
1240 2004-03-30 20:29 tux

Command history

bash remembers the most recent commands you've entered

stored in the file .bash_history in your home directory

survives across logout / login, shared by all instances of bash

size of history file is configurable, set to 500 commands in SuSE Linux

The history command shows your command history

history

shows your entire command history

history 10

shows the last ten commands

history -c

clears your command history

Previous commands can be selected and re-executed

!85

re-execute command 85

!string

re-execute most recent command that began with string

!!

re-execute last command

Command history (continued)

You can also edit your command history on screen.

The following keys are used:

scroll back through history

scroll forward through history

move left along line

move right along line

string

insert text string into line

<DEL>

delete character

<ENTER>

execute the command

Command history (continued)

To provide you with a command history, execute the following

commands:
$$ pwd
pwd
$$ hostname
hostname
$$ cal
cal -y
-y
$$ date
date
$$ id
id
$$ ls
ls /etc/hosts.allow
/etc/hosts.allow
$$ cat
cat /etc/fstab
/etc/fstab

Do now!

Recall your command history with the history command

Recall the date command by command number

Recall the most recent command beginning with 'ho'

Using the arrow keys, recall the 'ls' command, changing the file
name from hosts.allow to hosts.deny

Absolute pathnames

/ (root)

home

etc

tux

code
main.c
edges.c

at.deny
dumpdates
exports

dilbert

reports

July_Conf
progress
quality

expenses

Aug_2003
Sep_2003
Oct_2003

travel

paris
rome

Absolute pathnames (continued)

The linux file system is organised in a tree structure

The top level directory of the tree is called the root directory and is
named '/'
A file or directory can be referenced using an absolute pathname

Starts with a '/'

Traces a path from the root of the tree to the file

Uses '/' (forward slash) to separate components of the pathname

Examples:
/etc/at.deny
/home/dilbert/travel
/home/tux/reports/quality

Relative pathnames

Pathnames not beginning with '/' are relative to the current directory

Examples (assuming /home/tux is the current directory):

reports
code/main.c

Every directory has a special entry named '..' which references the
parent directory

The directory immediately above it in the tree

Use relative pathnames beginning with '..' to move outside the

current directory
Examples (assuming /home/tux/code is the current directory):
../reports/July_Conf
../../dilbert/travel/paris

File name completion

When entering a command, bash will perform filename completion

Press the TAB key

bash will complete as much of the name as is unambiguous, based on the

name completed so far and the names of the existing files

Less typing, less chance for typing mistakes

If there are several possible matches, [TAB] [TAB] will show them all

Example ([T] denotes the TAB key)

$ ls -l /h[T]/t[T]/re[T]/Ju[T] ... completes to:
$ ls -l /home/tux/reports/July_Conf

Using file name completion to minimise typing, do a long directory

listing (ls -l) on these files:
/usr/X11R6/bin/showfont
/usr/share/xscreensaver/screensaver-diagnostic.xpm
/boot/vmlinuz.config

Do now!

The linux security model

The linux security model

Users and groups
The superuser
Standard file permissions
Changing access permissions
Representing file permissions
in octal
setuid and setgid programs
Changing ownership with chown

Users and groups

Every user has an account name (e.g. tux) along with an associated
numeric user ID (e.g. 500)
Every user is associated with one named group which is their primary
group

Groups allow additional flexibility in assigning access permissions

Users can also be associated with one or more secondary groups

The command id shows your user identity and group memberships

Numeric user ID and user name
Primary group ID and group name

$ id
uid=500(tux) gid=100(users) groups=100(users),
14(uucp),16(dialout),17(audio),33(video)
Secondary groups

The user root

Linux has a priviledged user account called the super-user

The account name is usually root

The numeric user ID is zero

root can access or modify any file or directory and run any command

Only log in as root if you are doing something that requires it

You can start a new shell as root using the su command

$$ su
su -Password:
Password: suseroot
suseroot
earth:
earth: ~~ ##
The '#' in the prompt warns
you that you are root

The '- flag causes root's normal login

environment to be established
You are prompted for the password
It is not echoed to the screen

User identity and the super user

Log in as tux, with password penguin

Do now!

What is your prompt string? _______________________________

Run the id command

What is your numeric user ID? _____________

What is your primary group name? __________________

What are the names of your secondary groups? ___________________

Run the command 'su -' to start a shell as the superuser

Supply the password suseroot

What is your prompt string? _______________________________

What is your numeric user ID now? ____________________

What is your primary group ID? ________________

Type exit (or enter ^D) to exit from your superuser shell

Confirm (from your prompt) that you are no longer logged in as root

Standard file permissions

Every file and directory has a set of attributes:

An owner (by default, the person who created it)

A group (by default, the primary group of the person who created it)

Three sets of access permissions, relating to:

The owner of the file

Users who are members of the file's group

Everyone else (others)

These attributes are shown in a long directory listing:

$ ls -l etclist
-rw-r--r--

1 tux

users 65584 2004-03-16 11:30 etclist

Group
Owner
Permissions for others
Permissions for members of group
Permissions for owner

Standard file permissions (continued)

There are three access permissions in each set

The meanings of these permissions differ slightly depending on whether

they are applied to a regular file or a directory
Permission Meaning for a regular file

Meaning for a directory

r (read)

Able to see the contents of the Able to list the contents of the
file
directory

w (write)

Able to change the contents of

the file

Able to create or delete files or

subdirectories

x (execute)

Able to run the file as a

program or a script

Able to make the directory

current or use it in a path name

The permissions are shown as a group of nine characters,

for example:

rwxr-xr-x

Changing access permissions with chmod

The command chmod is used to change file permissions

How to build yourself a chmod command in five easy lessons

chmod

1. Type the
command
name!

2. Select any
combination of:
u User (owner) of file
g Group
o Other (rest-of-world)
Or a (same as ugo)

file1

4. Select any
combination
of r, w, x
3.
+
=

file2

5. List the files whose

permissions you want
to change

Select one of:

Add to existing permissions
Remove from existing permissions
Assign absolute permissions

Changing access permissions (continued)

Only the owner of a file (or the superuser) can change the files
permissions
$
$
$
$

chmod
chmod
chmod
chmod

u+x
go-w
a-wx
u=rw

hello.txt
display.object
opensource openwindows
open*

To set different access permissions for different users, either use two
chmod commands or separate changes with a comma:
$ chmod
$ chmod

u=rwx index
go=r index

Or:
$ chmod u=rwx,go=r index
Question: What do you think 'chmod u=rwx go=r index' does?

Representing file permissions in octal

Since each of the file permissions (rwxrwxrwx) is either on or off, a

files permissions can be represented by 3 groups of 3 binary digits

Then each set of 3 bits can be written as an octal digit (0-7)

rwx

rw-

r--

111

110

100

This notation can be used by chmod; e.g.

$ chmod

644

hello.txt

$ chmod

400

hello.txt

Some early versions of chmod, and a few other commands which deal
with access permissions, only understand the octal notation

Special file permissions

In addition to the standard 9 permission bits (rwxrwxrwx) there are

three more bits in a file's 'mode':
Octal
Value

Shown as

1000

t as the
'other
execute'
permission

2000

s as the
Files created in the directory
If the file is executed, the
'group
belong to the directory's
Set Group
effective group ID is set to the
execute'
group and not to the primary
ID
group of the file
permission
group of the user

4000

s as the
'user
execute'
permission

Name

Meaning for files

Meaning for Directories

Sticky bit

Originally this bit indicated

that executable programs
should stay in memory after
they have terminated. This
meaning is now obsolete

You can only delete files if

you own the file or you own
the directory. Often used for
communal directories such
as /tmp

Set User
ID

If the file is executed, the

effective user ID is set to the
owner of the file

Special file permissions (continued)

There are special notations in chmod to allow the special file

permissions to be set or unset
When listed using 'ls -l' the special permissions show up as an 's' or
a 't' in the execute permission positions
$$ touch
touch temp3
temp3
$$ chmod
777
chmod 777 temp3
temp3
$$ ls
-l
temp3
ls -l temp3
-rwxrwxrwx
11 chris
-rwxrwxrwx
chris
$$ chmod
chmod u+s
u+s temp3
temp3
$$ ls
-l
temp3
ls -l temp3
-rwsrwxrwx
11 chris
-rwsrwxrwx
chris
$$ chmod
g+s
temp3
chmod g+s temp3
$$ ls
ls -l
-l temp3
temp3
-rwsrwsrwx
11 chris
-rwsrwsrwx
chris
$$ chmod
o+t
temp3
chmod o+t temp3
$$ ls
ls -l
-l temp3
temp3
-rwsrwsrwt
11 chris
-rwsrwsrwt
chris
$$

users
users

00 2004-03-16
2004-03-16 14:56
14:56 temp3
temp3

users
users

00 2004-03-16
2004-03-16 14:56
14:56 temp3
temp3

users
users

00 2004-03-16
2004-03-16 14:56
14:56 temp3
temp3

users
users

00 2004-03-16
2004-03-16 14:56
14:56 temp3
temp3

setuid and setgid programs

The 'setuid' and 'setgid' permissions allow a user to run a command

with the effective identity of the owner / group of the file

Grants the user different priviledges for the duration of the command

Example:

User's (encrypted) passwords are stored in the file /etc/shadow; only

root has write permission on this file

Users can change their passwords using the command /usr/bin/passwd

To allow this, the ownerships and permissions are like this:

$$ ls
ls -l
-l /etc/shadow
/etc/shadow
-rw-r-----rw-r----- 11 root
root shadow
shadow 768
768 2004-03-05
2004-03-05 10:03
10:03 /etc/shadow
/etc/shadow
$$ ls
-l
/usr/bin/passwd

ls -l /usr/bin/passwd
-rwsr-xr-x
-rwsr-xr-x 33 root
root shadow
shadow 77204
77204 2003-09-24
2003-09-24 00:04
00:04 /usr/bin/passwd
/usr/bin/passwd

It's important to be sure that setuid programs are trustworthy,

especially those owned by root

Changing ownership with chown

The chown command can change the ownership and group of a file
chown

root.wheel

file2 ...

foo

bar

You can change just the owner:

$ chown root

file1

Example:
$ chown

owner.group

foo

or you can change just the group:

$ chown .wheel

bar

$ chgrp

bar

wheel

Only root can change a file's owner

Ordinary users can change a file's group only if they are members of both
the original and the new group

Exercise: File permissions and ownerships

1. You should initially be logged in as tux for this exercise
2. Using the touch command, create a file called sample
$ touch sample

Who owns the file sample? _____________________

What group does the file belong to? ____________________

What are the initial access permissions on the file? _________________

2. Using chmod, create the following sets of access permissions, in turn,

on the file sample. After each change, verify the permissions by doing
a long listing of the file
rw------rw-rw-rwrwxrwxrwx

Exercise continued
4. As the user tux, try to change the ownership of the file sample to be
owned by the user dilbert.

What happens?

5. Use the su command to switch to a superuser shell

Try again to change sample to be owned by dilbert

(It should work this time)

Change the group ownership of sample to the group trusted

6. Exit from the superuser shell

7. Do a long listing of sample and verify the ownership and group

End of Exercise

Filter programs

Filter programs
Six useful commands
Standard input and output
Redirecting standard output
Redirecting standard error
Filter programs
Redirecting standard input
Using programs in combination
Pipelines

Six useful commands

There are hundreds of command line tools for linux. There are only
about 30 that you need to know to be proficient at the command line

We have already met date, id, ls, touch, chmod and chown

In this section we'll meet another six useful commands:

less

Browse text files

grep

Search for patterns in files

wc

Count characters, words, lines in a file

head

Display the beginning of a file

tail

Display the end of a file

sort

Sort the contents of a file

Individually, each command does a fairly simple job

Much of the power of the linux command line comes from using tools in
combination

Browsing text files with less

The program less (an extension of an earlier program called more)

provides a simple way to display a text file
less is an interactive program and waits for you to enter a command

Bidirectional scrolling

Searching

Command
SPACE
b
Down arrow
Up arrow
/string
?string
n
5G
h
q

Meaning
Scroll forward one screen
Scroll backward one screen
Scroll forward one line
Scroll backward one line
Search forward for string
Search backward for string
Repeat previous search
Go to line 5
Display help screen
Quit (back to command prompt)

Browsing text files with less (continued)

Use less to browse the file /etc/profile

Do now!

Use the down arrow key and/or the SPACE bar to

browse through the file

Return to the beginning of the file (i.e. go to line 1)

Search forward for the string 'SuSE'

Repeatedly search forward for further occurrences of the string

How many times does the string appear in the file? _______________

Display the help screen

Quit from less back to the shell prompt

Searching for patterns with grep

The program grep searches one or more text files for lines that
match a specified pattern. At its simplest, it is used like this

grep

'string'

file1 file2 ...

Lines in the file(s) that contain a match for the string are displayed

Note: putting the string inside single quotes is not always necessary but
is good practice, for reasons we will examine in chapter 8

Examples:
$ grep 'tux' /etc/passwd
tux:x:504:100:Tux Penguin:/home/tux:/bin/bash
$ grep 'Clothes'
Supermarket 50
Clothes
1
Clothes
6
Clothes
2

shopping
Clothespegs 1.25
Trousers
24.99
Socks
9.00
Skirt
28.00

Options for grep

Command line arguments for grep include:

Option Meaning
-r

Recursive: search all files in and below a given directory

-i

Ignore upper / lower case distinctions

-l

Show only the names of files that contain a match; not the matching lines

-v

Display lines that do not contain a match for the pattern

-Cn

Show n lines of context before and after each matching line

-An

Show n lines of context after each matching line

-Bn

Show n lines of context before each matching line

Anchoring the search

The string that grep looks for is called a regular expression

Can contain special characters that match specific patterns in the text

Not covered in depth in this course

The characters '^' and '$' anchor the search to the beginning and end
of the line respectively
$ grep 'Clothes' shopping
Supermarket 50
Clothespegs
Clothes
1
Trousers
Clothes
6
Socks
Clothes
2
Skirt
$ grep '^Clothes' shopping
Clothes
1
Trousers
Clothes
6
Socks
Clothes
2
Skirt

1.25
24.99
9.00
28.00
24.99
9.00
28.00

Counting characters, words and lines with wc

The command wc counts lines, words and characters in its input files

$ wc

/etc/passwd
29
64
15
56
44
120

shopping
1510 /etc/passwd
491 shopping
2001 total

Command line options include:

Option
-l
-w

Meaning
Show only the line count
Show only the word count

-c

Show only the character count

-L

Show the length of the longest line

Displaying the start of a file with head

The head command displays the beginning of one or more files

$ head

-N

file1

file2

...

Displays the first N lines of each file (default = 10 lines)

Example:
$ wc shopping
15
56
491 shopping
$ head -4 shopping
Supermarket 1
Chicken
4.55
Supermarket 50
Clothespegs 1.25
Bakers
3
Bread
2.40
DIY
1
Hosepipe
15.00

The file shopping

has 15 lines
Here are the first 4

Displaying the end of a file with tail

The tail command displays the end of a file

$ tail

-N

file1 file2 ...

Displays the last N lines of each file (default = 10 lines)

Example: the last line of the passwd file:

$ tail -1 /etc/passwd
tux:x:504:100:Tux Penguin:/home/tux:/bin/bash

The -f option causes tail to wait after reaching the end of the file

Any text subsequently appended to the file is displayed

Very useful for monitoring the growth of log files

Sorting a file with sort

The sort command sorts its input line by line

By default, does alphanumeric sort on entire line

Command line options include:

Option
-f
-n
-r
-k N

Meaning
Ignore upper/lower case distinction
Numeric sort
Reverse sort
Sort on field N (first field is 1)

Example of using sort

$ sort -n -r -k 4 shopping
Clothes
2
Skirt
Clothes
1
Trousers
DIY
1
Hosepipe
DIY
20
Sandpaper
Clothes
6
Socks
DIY
2
Doorknob
Bakers
2
Quiche
Supermarket 1
Chicken
Bakers
3
Bread
DIY
2
Screwdriver
Bakers
10
Muffin
Supermarket 2
Milk
Supermarket 50
Clothespegs
DIY
50
Nails

28.00
24.99
15.00
10.00
9.00
8.40
6.50
4.55
2.40
2.00
1.95
1.25
1.25
0.95

Reverse numeric
sort on fourth field

More sort examples for you to try

Try the following commands; make sure you

understand the results
$ sort

Do now!

shopping

$ sort -r shopping
$ sort -k 3 shopping
$ sort -k 2 shopping
$ sort -n -k 2 shopping
$ sort -n -r -k 2 shopping

Most commands allow you to combine multiple options, e.g.

$ sort -nr -k 2 shopping
$ sort -nrk 2 shopping

Standard input and standard output

Every program started from the command line has three standard
streams:

Stream 0: Standard input (stdin): from the keyboard by default

Stream 1: Standard output (stdout): to the terminal by default

Stream 2: Standard error (stderr): to the terminal by default

Standard
Input

(0)Any linux
program

(1)

Standard
Output

(2)
Standard
Error

Redirecting standard output

The normal output from a program is written to standard output

The shell can be told to redirect standard output to a file

$ date > myfile
$ ls /opt > myfile

Beware! myfile will be overwritten if it exists

$ ls /boot >> myfile

The output is appended to the file

myfile

Standard
Input

Standard
Output
(0)Any linux
program

(1)
(2)
Standard
Error

Redirecting standard error

Error messages are written to the standard error stream

The notation 2> redirects standard error

$$ ls
ls /opt
/opt /test
/test >> mylist
mylist
/bin/ls:
/bin/ls: /test:
/test: No
No such
such file
file or
or directory
directory
$$ ls
ls /opt
/opt /test
/test >> mylist
mylist 2>
2> myerrors
myerrors
$$ cat
myerrors
cat myerrors
/bin/ls:
/bin/ls: /test:
/test: No
No such
such file
file or
or directory
directory

Standard
Input

(1)
(0)

Standard
Output

mylist

ls
(2)

myerrors
Standard
Error

Combining standard error with standard output

The notation '2>&1' says 'send output stream 2 (standard error) to

wherever output stream 1 (standard output) is going'
$$ ls
ls /opt
/opt /test
/test >> mylist
mylist 2>&1
2>&1
$$ cat
mylist
cat mylist
/bin/ls:
/bin/ls: /test:
/test: No
No such
such file
file or
or directory
directory
/opt:
/opt:
gnome
gnome
kde3
kde3
mozilla
mozilla

Standard
Input

(1)
(0)

Standard
Output
mylist

ls
(2)
Standard
Error

Filter programs

Programs such as grep, wc, head, tail and sort read their
standard input if they are not given a filename argument
Programs that read standard input, process it, and write the result to
standard output are called filters

$ sort
apple
orange
banana
^D
apple
banana
orange

Since no file name is given, sort reads

from standard input (the keyboard)

The user enters ^D to

signify the end of input
The sorted output is written
to standard output

Redirecting standard input

The notation '<' redirects a program's standard input

This example shows another filter, tr, performing lower to upper
case conversion
Here, standard input comes from
$ tr a-z A-Z
the keyboard; used ^D to terminate
Hello World
HELLO WORLD
$ tr a-z A-Z < /etc/motd
Standard input comes from a file
WELCOME TO FIRST TECHNOLOGY TRANSFER

Standard input and standard output can both be redirected

$ tr a-z A-Z < /etc/motd > outfile
$ cat outfile
WELCOME TO FIRST TECHNOLOGY TRANSFER

Using programs in combination

The output of one program may be used as input to another

An intermediate temporary file is one way to do this

$ grep DIY shopping > temp

$ sort -n -k 4 < temp
DIY
50
Nails
0.95
DIY
2
Screwdriver 2.00
DIY
2
Doorknob
8.40
DIY
20
Sandpaper
10.00
DIY
1
Hosepipe
15.00

Get list of all DIY items

List sorted
on price

Pipelines

A pipe connects the standard output of one program directly to the

standard input of another

The two programs are run concurrently

$ grep DIY shopping | sort -n -k 4

Do now!

shopping

(1)
(0)

Standard
Output

Standard
Input

(1)
(0)

sort

(2)

grep

(2)

Standard
Error

More pipeline examples

How many items in the shopping list are from the Baker's?

$ grep Bakers shopping | wc -l

Do a long listing of /etc, browse the output through less

$ ls -l /etc

less

Find the most expensive item in the shopping list

$ sort -n -r -k 4 shopping

head -1

How many files in /lib are actually directories?

$ ls -l /lib | grep '^d' | wc -l

Exercise: Using filters

Display the first 10 lines of the file /etc/sysconfig/network/config

Hint: use filename completion to reduce typing

Display the first 20 lines of the same file

Hint: use command history to reduce typing

Show the last line (only) of the same file

Search /etc/sysconfig/network/config for the string FIREWALL

Hint: use grep

Redo the search, but ignore upper/lower case distinctions i.e.

search for firewall, FIREWALL, etc
(Harder) Display all lines in /etc/sysconfig/network/config that
are not comments (i.e. do not start with a '#')

Exercise: I/O Redirection and pipelines

Create a file called stuff1 containing a long listing of the files in /etc

Search stuff1 for lines containing the string rw-------, putting

the output in a file called stuff2
Count the number of lines in stuff2

Hint: Redirect standard output

Don't count them yourself, make the computer do it!

Create a pipeline (no intermediate files) that displays a count of the

number of files in /etc that have access mode rw-------
Create a file called stuff3 which contains, in order:

The current date and time

A calendar for the current year

The hostname of your machine, converted to upper case

Hint: append standard output to the file

Exercise: I/O Redirection and pipelines (contd)

Bonus Exercises:
Using a pipeline, display the line that describes the most expensive
DIY item in the shopping list file
Run the command ps aux and examine the output. It gives a list of
all processes running on the computer. You're not expected to
understand all of this output, but note that the first field shows the
owner and the fifth field shows how much memory the process is
using
Using pipelines, devise commands to answer the following:

1. How many processes are there altogether?

2. How many processes are owned by root?

3. How many processes are not owned by root?

4. Which of root's processes is using the most memory? (Your pipeline

should just display the line describing this process)

Managing files

Managing files
The current directory
Creating and deleting directories
Listing directory contents
Copying files
Renaming files
Deleting files
Updating files
Links and the inode table
Creating links to a file
File and directory permissions
revisited

Setting and querying the current directory

Every process (including the shell) has a current directory

Where it will look for 1relative path names

The cd command changes the current directory.

The pwd command displays the name of the current directory

$$ cd
cd
$$ pwd
pwd
/home/tux
/home/tux
$$ cd
cd pics
pics
$$ pwd
pwd
/home/tux/pics
/home/tux/pics
$$ cd
cd /etc
/etc
$$ pwd
pwd
/etc
/etc
$$ cd
cd -/home/tux/pics
/home/tux/pics
$$ cd
cd ..
..
$$ pwd
pwd
/home/tux
/home/tux

cd with no arguments takes you

to your home directory
Descend into a subdirectory
Change directory using an
absolute pathname
'-' takes you back to your
previous directory
'..' takes you up one level

Creating and deleting directories

The mkdir command creates new directories

-p option creates additional directories, if required, along the path

The rmdir command deletes directories

Only if they are empty

$$ mkdir
mkdir proposals/january
proposals/january
mkdir:
mkdir: cannot
cannot create
create directory
directory `proposals/january':
`proposals/january':
No
such
file
or
directory
No such file or directory
$$ mkdir
mkdir -p
-p proposals/january
proposals/january
$$ rmdir
rmdir proposals
proposals
rmdir:
rmdir: `proposals':
`proposals': Directory
Directory not
not empty
empty
$$ rmdir
proposals/january
rmdir proposals/january
$$ rmdir
rmdir proposals
proposals
$$

Listing directory contents with the ls command

The ls command lists files in a directory

If a directory name is given, the contents of the directory are shown

With no arguments, the current directory is listed

ls has many options. Here are a few:

Option Meaning
None

Display filenames only, in a multi-column listing

-l

Display a 'long' listing, including file type, permissions, modification time, and
size, in addition to the name

-a

Display 'hidden' files (ones whose names begin with a '.')

-F

After each name, append a character to indicate the file type: '/' indicates a
directory, '*' indicates an executable file, '@' indicates a symbolic link

-i

Display inode numbers (discussed later)

-t

Sort by time of last alteration (by default, sort is alphabetic on file name)

-u

Sort by time of last access

-R

Recursive: descend into any subdirectories

-d

When listing a directory, list just the directory entry, not the contents

Hidden files

Files whose names begin with a '.' are hidden

They do not show up on a normal directory listing

Most of them are configuration and startup files, for example:

.bash_history:

Where the bash shell stores its command history

.xinitrc:

Startup file for the X window system

.bashrc:

Startup file for the bash shell

.profile:

Startup file for all shells

Use ls -a to show hidden files

Copying files with the cp command

The cp command copies files

$ cp file1 file2

This form makes a copy of file1 under the name file2

$ cp

file1

file2 ...

dir

This form makes copies of file1 file2 ... in (existing) directory dir

Beware: the destination files will be replaced if they already exist

Options include:
Option
-i
-u
-l
-s
-r, -R

Meaning
Interactive mode: ask for confirmation before replacing an
existing file
If the destination file already exists, perform the operation
only if the source file is newer than the destination file
Create links instead of making copies
Create symbolic links instead of making copies
Copy directories recursively

Renaming files with the mv command

The mv command renames or moves files

$ mv file1 file2
This form renames file1 as file2.

$ mv file1 file2 ...

dir

This form moves file1 file2 ... into (existing) directory dir

Beware: the destination files will be replaced if they already exist

Options include:
Option
-i
-u

Meaning
Interactive mode: ask for confirmation before replacing an
existing file
If the destination file already exists, perform the operation
only if the source file is newer than the destination file

Deleting files with the rm command

The rm command deletes files

$ rm file1 file2 ...

Beware: there is no 'undelete' command!

Options include:

Option
-i
-f

-r

Meaning
Interactive mode: ask for confirmation before deleting
Normally rm will prompt for confirmation before deleting
a file on which you do not have write permission. The -f
flag suppresses this prompt and forces rm to delete the
file
Recursive mode: delete all subdirectories and contents
(Use with caution!)

Updating files with the touch command

The command touch updates the access and modification

timestamps on a file to the current time

Makes it appear that the file has just been modified

If the file does not exist it is created with zero length

$ touch file1 file 2 ...

Options for touch include:

Option
-a
-m
-r file
-t time

Meaning
Update only the access timestamp
Update only the modification timestamp
Update the timestamps to match those of file, not the
current time
Update the timestamps to the specified time, in the
format [[CC]YY]MMDDhhmm[.ss]

Links and the inode table

The filesystem associates a structure called an inode with each file

Contains file's attributes and pointers to the actual data blocks

space for inodes (inode table) is pre-allocated when filesystem created

1
2
3
Directory /home/tux
4
5
Name
inum
6
.
55
7
8
..
93
9
stuff
7
10
11
shopping
10
12
13
A link associates a name with inode table
an inode number. A directory
contains a list of links.
Link to parent

Type = regular file

Owner = tux
Group = student
Mode = rw-r--r-Time last accessed
Time last modified
Pointers to data blocks

Supermarket

Chicken ..

Creating links to a file with ln

Creating additional links to a file allows the file to be referenced by

more than one name
General form of ln command is:

Directory /home/tux

$ ln existing_name new_name

Example:

Name

inum

55

..

93

$ cd /home/dilbert

foo

$ ln ../tux/shopping mylist

shopping

7
10

Directory /home/dilbert
Name

The system keeps count (in

the inode) of the number of
links. When the last link is
removed, the inode and the
data blocks are freed.

inum
1236

..

93

bar

12

mylist

10

1
2
3
4
5
6
7
8
9
10
11
12
13
inode table

Familiarising with links

The -i option to ls shows the inode numbers

$ ls -li
total 4
190116 -rw-r--r-- 1
$ ln shopping mylist
$ ls -li
total 8
190116 -rw-r--r-- 2
190116 -rw-r--r-- 2
$ rm shopping
$ ls -li
total 4
190116 -rw-r--r-- 1

The inode number

tux

users

491 2004-04-19 11:22 shopping

tux
tux

users
users

491 2004-04-19 11:22 mylist

491 2004-04-19 11:22 shopping

tux

users

491 2004-04-19 11:22 mylist

The link count

Symbolic links

A symbolic link is a special type of file that simply contains the

pathname of a target file

Any references to the symbolic link are automatically translated into

references to the target

The '-s' flag tells ln to create symbolic links

$ ln -s shopping mylist
$ ls -l shopping mylist
lrwxrwxrwx 1 chris users
8 2004-04-23 11:43 mylist -> shopping
-rw-r--r-- 1 chris users 491 2004-04-23 09:24 shopping

Deleting the target file breaks the symbolic link and causes confusing
behaviour:
$ rm shopping
$ less mylist
mylist: No such file or directory

File and directory permissions revisited

The ways in which file access permissions control what you can do
derive from a few simple rules

To access the data in a file, you need read permission on the file

To modify the data in a file, you need write permission on the file

To list the names (only) of the files in a directory, you need read
permission on the directory

To list the attributes of the files in a directory (e.g. to perform an ls -l), or

to use the directory in a pathname, or to make it your current directory,
you need execute permission on the directory

(You need both read and execute permission to have useful access to a
directory)

To add a link to a directory, or to remove a link from a directory, you

need write permission on the directory

File permission examples

Operation

Situation

Required permissions

cp file1 file2

file2 doesnt
exist

r permission on file1
w permission on file2s directory

cp file1 file2

file2 already
exists

r permission on file1
w permission on file2

mv file1 file2

w permission on file1s directory

w permission on file2s directory

rm file1

w permission on file1s directory

ln file1 file2

rx permission on file2s directory; w

permission on file2s directory

File permission quiz

Using the ls command, and your knowledge of file permissions,

predict which of the following operations would be allowed if you
were logged in as tux
Operation
Make a copy of
/etc/at.deny in your
home directory
Make a copy of
/etc/group in your home
directory
Make a copy of your
shopping file in the
/etc directory
Delete your own home
directory

Allowed?

Why or why not?

Exercise: Managing files

1. Log in as tux
2. Copy the files passwd, group, hosts and fstab from /etc to your
home directory
3. In your home directory, rename the file fstab to table
4. Create a directory called private in your home directory
5. Change the permissions on private so that only you can access it

Hint: the permissions should be rwx------

6. Move the files passwd and group into private

7. Change directory into the private directory and list the files there
8. Log out and log back in as dilbert with password adams

Can you list tux's home directory? ___________________________

Can you list tux's private directory? __________________________

Exercise: Managing files (continued)

9. Log out as dilbert and log back in as tux
10. Create a hard link called mygroup in your home directory to the file
group in the private directory

What is the inode number associated with this link? _________

11. Create a symbolic link called mypasswd in your home directory to the
file passwd in the private directory
12.Delete the private directory and its contents
13.Try to access the files mygroup and mypasswd in your home directory

What happens? ______________________________________________

Can you explain why? ________________________________________

14.Delete the symbolic link mypasswd

End of Exercise

Miscellaneous features

Miscellaneous
features

File name expansion using

wildcards
Editing with vi
Finding files with find
Manual pages
Builtin help
HOWTO documents

Filename expansion using wildcards

The shell uses several meta-characters for matching patterns in file

names

Commonly known as wildcards

The command as
typed

$ ls

-l

The shell looks

in the current
directory for
matching file
names

*.c

The command as
executed

$ ls

-l

edges.c

and substitutes
them back into the
command line

main.c

alias.pl
edges.c
edges.o
main.c
main.o
wireframe

Filename expansion using wildcards (continued)

Matches zero or more characters

Matches exactly one character

[x-y]

Matches any one of the enclosed characters, e.g. [AaTt]

Matches any character in range
e.g. [a-m]

* matches all names, like *.* in DOS

Multiple wildcards can be used

[0-9]

Linux has no notion of filename extensions

[A-Z]

E.g.

rm

[A-Z]*.html

Wildcards can be used in multiple components of a pathname

E.g.

rm

backups/*.199[7-9]/expenses??

Wildcard quiz
Given these files in the current directory

410-chap1.doc
410-chap2.doc
410-chap3.doc
410-chap4.doc
410-chap5.doc
410-CHAPS.doc
display
display.c
display.h
display.object
ideas
ideas.old
ideas.older
index
intro
A directory

intro.old
meetings.June
meetings.July
meetings.Aug
oldstuff
opensource
openwindows
project6
project45
project46
project346
training
venues
windows
x-windows

What do these commands do?

rm *.old
ls -l 410-chap?.doc
less 410-chap[2-5].doc
mv ideas* training
mv ideas.* training
rm *old*
rm *
cp [v-z]* training
cp [a-z]* training
ls display.?
rm proj*6
rm proj*[a-z]6

Exercise: Wildcards

Do a long listing of all the files in /usr/bin whose names contain a

digit

Create a subdirectory called config in your home directory

Copy all the files whose name ends in .conf from /etc to your
config directory

How many such files are there?

Can you find a way to supress the error messages due to unreadable
files?

Change the access permissions of all files in your config directory to

be rw-------
Are there any files in your config directory whose name begins with
a vowel?
Delete any files in your config directory whose name begins with a
vowel

Editing with vi

The editor vi is the 'standard' editor on UNIX and Linux systems

Advantages:

It is available on every UNIX and Linux system

It is available on the SuSE rescue disk

It works on character terminals, without a graphical user interface

It is fast and powerful once you know it well

Disadvantages:

It is a 'moded' editor which makes it difficult to learn initially

There are a lot of commands to remember

Professional linux users and administrators benefit in the long term

by learning vi

Budget several hours of learning time, over a period of time

Working modes in vi

vi has three major operating modes

Start
i = insert before cursor

Command Mode
Most editing operations such
as deleting and searching are
done in command mode.
Most use single character
commands

':'

A = append to line
others ...

ESC

Insert Mode
The characters you
type are entered into
the edit buffer

Newline
Bottom Line Mode

Most commands that require arguments use this mode.

e.g. global search and replace, write buffer to file, go to
given line number. The command is terminated by a newline

Command mode in vi

In command mode, most 'ordinary' characters invoke a command

This table shows a minimal command set to survive with vi. It is nowhere
near complete and does not show the full power of the command set.
Command Meaning
i
Switch to insert mode, insert text before cursor position
A
Switch to insert mode, append to current line
x
Delete the character under the cursor
dd

Delete the current line (and put in the paste buffer)

Delete to end of line

/pattern

Search for pattern, forwards from current cursor position

?pattern

Search for pattern, backwards from current cursor position

Repeat the search in the same direction

Repeat the search in the opposite direction

yy

Yank' (copy) the current line into the paste buffer

Insert the paste buffer before the current cursor position

ZZ
.

Save the file and exit (this is the normal way to exit vi)
Repeat the last change at the current cursor position

Bottom line mode in vi

The table below shows some important 'bottom line' commands

There are also powerful global 'search and replace' commands that are
not shown here

Command Meaning
:q
Quit the editor (only works if no changes have been made)
:q!
Quit the editor, abandoning any changes that have been made
:wq
:w file
:41

Write the file out and exit (same as 'ZZ' in command mode)
Write the edit buffer out to the specified file (instead of the
original file)
Go to line 41

Exercise: Using vi

Using vi, enter the text:

A linux sysadmin called Pete
Typed all his commands with his feet
After rm -r, his toe hit a star,
A sysadmin no longer is Pete

Save the text to the file limerick in your home directory

Make the following changes:

Change 'Pete' to 'Joe', (twice) and 'feet' to 'toe'

Swap the first line with the last line. (Use cut-and-paste, don't retype!)

Delete the third line

Save the result in the file junk (not to the limerick file)

Finding files with the find command

The find command searches for files meeting specified criteria

Name, owner, group, timestamps, permissions, size, etc.

find has a complicated syntax; the general framework is:

find

Where to look

List of directories to
start looking in. find
will search
recursively down
through the file
system from here

What to look for

Set of criteria
which files must
match. Can be
used in AND and
OR combinations

What to do with it

The action to take

when a matching file is
found. The default is to
print the pathname to
the file

Search criteria for find

Search criteria for find include:

Syntax

Example

File name matches string.

Wildcards are allowed

-name

-iname 'string'

Same as -name but not case

sensitive

-name 'greet*'

-user

File is owned by username

-user

-group groupname

File belongs to groupname

-group root

-type

File is a directory, regular file,

or symbolic link

-type f

-size +N

File is bigger than N blocks

suffix c = bytes, k = kbytes

-size +1000k

-size -N

File is smaller than N blocks

-size -50c

-name

Description

'string'

username

d, f, or l

'*.old'

dilbert

The default action for find is simply to display the names of the
matching files

Some examples of using find

Show all files ending in '.c' in (and below) the current directory
$ find . -name '*.c'

Find all files under /home owned by tux

$ find /home -user tux

Find ordinary files in /usr/bin which are bigger than 1 Mbyte

$ find /usr/bin

-type f

-size +1000k

Find all ordinary files owned by root which have zero length
$ find / -type f -user root -size 0 2> /dev/null

More search criteria for find

Syntax

Description

Example

-perm xxx

File permissions exactly

match octal digits xxx

-perm 644

-perm +xxx

File has at least one of the

permissions xxx

-perm +222

-perm -xxx

File has all of the permissions

xxx

-perm -001

-mtime +n

File last modified more than n

days ago

-mtime +14

-mtime -n

File last modified less than n

days ago

-mtime -2

-atime +n

File last accessed more than

n days ago

-atime +7

-atime -n

File last accessed less than n

days ago

-atime -1

Actions for find

An action specifies what to do with each matching file

-print (also the default) just writes out the pathname of the file

-ls writes output similar to ls -li for the file

Any arbitrary command may be executed using -exec

The syntax is messy; here's an example:

-exec

rm

The command you

want to execute

{}

\;
Marks the end of the
-exec command

The name of the

matching file is
substituted in here

More examples of using find

Find all files in /home/tux or /home/dilbert which are world-writable

and give a detailed listing
$ find /home/tux /home/dilbert -perm +002 -ls

Delete files under /home with names ending '.bak' which have not
been accessed for two weeks
$ find /home -name '*.bak' -atime +14 -exec rm {} \;

Find all files which are not symbolic links but have mode 777
$ find / ! -type l -perm 777 2> /dev/null

'!' negates the sense

of the following test

Discard reports of
unreadable directories, etc.

Exercise: Using find

List all the directories under /home that belong to dilbert

How many symbolic links are there under /usr/bin?

Does root own any zero-length regular files? How many?

Hint: don't count them yourself!

What is the largest file in the filesystem?

Hint: it is bigger than 10 Mbytes

Manual pages

The traditional way of providing online help is the manual page

accessed via the man command

tux@earth:~> man mount

tux@earth:~> man mount
MOUNT(8)
Linux Programmer's Manual
MOUNT(8)
Linux Programmer's Manual
NAME
NAME

MOUNT(8)
MOUNT(8)

mount - mount a file system

mount - mount a file system

SYNOPSIS
SYNOPSIS
mount [-lhV]
mount [-lhV]
mount -a [-fFnrsvw] [-t vfstype] [-O optlist]
mount -a [-fFnrsvw] [-t vfstype] [-O optlist]
mount [-fnrsvw] [-o options [,...]] device | dir
mount [-fnrsvw] [-o options [,...]] device | dir
mount [-fnrsvw] [-t vfstype] [-o options] device dir
mount [-fnrsvw] [-t vfstype] [-o options] device dir
DESCRIPTION
DESCRIPTION
All files accessible in a Unix system are arranged in one big
All files accessible in a Unix system are arranged in one big
tree, the file hierarchy, rooted at /.
These files can be
tree, the file hierarchy, rooted at /.
These files can be
spread out over several devices. The mount command serves to
spread out over several devices. The mount command serves to
attach the file system found on some device to the big file
attach the file system found on some device to the big file
tree. Conversely, the umount(8) command will detach it again.
tree. Conversely, the umount(8) command will detach it again.

How a manual page is organised

Each manual page is divided up into a number of sections

Not all sections are present for all commands

Section

Contents

NAME

Name and short description of the command

SYNOPSIS

Description of the syntax

DESCRIPTION

Detailed description of the command

OPTIONS

Description of all available options

COMMANDS

Instructions that can be given to the program while it is running

FILES

Files referenced by the command

SEE ALSO

List of related commands

DIAGNOSTICS

Possible error messages and what they mean

EXAMPLE

Examples of usage

AUTHOR

Who wrote it

BUGS or WARNINGS Known errors and problems

Manual page section numbering

Each manual page is allocated a 'section number' depending on what

kind of thing it describes:
Section

Contents

Used by

Commands for end users

End users

System Calls

Developers

Functions and Library Routines

Developers

Device Files

Administrators / developers

Configuration files and file formats

Administrators

Games

Macro packages and file formats

System Administration Commands

Administrators

The output of man is piped through less for ease of browsing

Manual page section numbering (continued)

By default the man command finds the requested page in the lowestnumbered section

If the same name appears in more than one section, you may have to
specify the section number explicitly to get the right man page

$$ man
man crontab
crontab
...
... man
man page
page for
for crontab
crontab command
command in
in section
section 11 ...
...
$$ man
man 55 crontab
crontab
...
man
... man page
page for
for crontab
crontab file
file format
format in
in section
section 55 ...
...
$$ man
man uname
uname
...
... man
man page
page for
for uname
uname command
command in
in section
section 11 ...
...
$$ man
man 22 uname
uname
...
... man
man page
page for
for uname
uname system
system call
call in
in section
section 22 ...
...

Searching the manual pages by keyword

You can search for manual pages by keyword

or

man -k keyword

keyword must appear in NAME section of man page

$$ apropos
apropos partition
partition
mpartition
mpartition (1)
(1)
sfdisk
(8)
sfdisk (8)
gpart
gpart (8)
(8)
ntfsfix
ntfsfix (8)
(8)
mkfs.jfs
mkfs.jfs (8)
(8)
lvmdiskscan
lvmdiskscan (8)
(8)
jfs_mkfs
(8)
jfs_mkfs (8)
pvcreate
pvcreate (8)
(8)
cfdisk
(8)
cfdisk (8)
partprobe
partprobe (8)
(8)
fdisk
(8)
fdisk (8)
parted
parted (8)
(8)

apropos keyword

-- partition
partition an
an MSDOS
MSDOS hard
hard disk
disk
-- Partition
table
manipulator
Partition table manipulator for
for Linux
Linux
-- guess
guess PC-type
PC-type hard
hard disk
disk partitions
partitions
-- tool
for
fixing
NTFS
partitions
tool for fixing NTFS partitions altered
altered by
by ...
...
-- create
a
JFS
formatted
partition
create a JFS formatted partition
-- scan
scan for
for all
all disks
disks // multiple
multiple devices
devices ...
...
-- create
a
JFS
formatted
partition
create a JFS formatted partition
-- initialize
initialize aa disk
disk or
or partition
partition for
for use
use by
by LVM
LVM
-- Curses
based
disk
partition
table
manipulator
Curses based disk partition table manipulator
-- inform
inform the
the OS
OS of
of partition
partition table
table changes
changes
-- Partition
table
manipulator
for
Linux
Partition table manipulator for Linux
-- aa partition
partition manipulation
manipulation program
program

Exercises: Using the manual pages

1. For this exercise, log in as tux
2. Look up the man page for the ls command

What option gives a 'long' listing? _________________

What option gives a color listing? __________________

3. Look up the man page for the write command

What command did you use? __________________________
4. Look up the man page for the write system call
What command did you use? __________________________
5. Find the name of the command for formatting floppy disks

Hint: Use apropos (and maybe grep)

6. Find the name of (a) the command and (b) the system call for changing
permissions on a file

Built-in Help

Most of the GNU command line tools have built-in help, using the
--help option

tux@earth:~> cat --help

tux@earth:~> cat --help
Usage: cat [OPTION] [FILE]...
Usage: cat [OPTION] [FILE]...
Concatenate FILE(s), or standard input, to standard output.
Concatenate FILE(s), or standard input, to standard output.
-A, --show-all
equivalent to -vET
-A, --show-all
equivalent to -vET
-b, --number-nonblank
number nonblank output lines
-b, --number-nonblank
number nonblank output lines
-e
equivalent to -vE
-e
equivalent to -vE
-E, --show-ends
display $ at end of each line
-E, --show-ends
display $ at end of each line
-n, --number
number all output lines
-n, --number
number all output lines
-s, --squeeze-blank
never more than one single blank line
-s, --squeeze-blank
never more than one single blank line
-t
equivalent to -vT
-t
equivalent to -vT
-T, --show-tabs
display TAB characters as ^I
-T, --show-tabs
display TAB characters as ^I
-u
(ignored)
-u
(ignored)
-v, --show-nonprinting
use ^ and M- notation, except for LFD and TAB
-v, --show-nonprinting
use ^ and M- notation, except for LFD and TAB
--help
display this help and exit
--help
display this help and exit
--version output version information and exit
--version output version information and exit
With no FILE, or when FILE is -, read standard input.
With no FILE, or when FILE is -, read standard input.

3.3 HOWTOs

A huge number of 'HOWTO' documents provide help on a very wide

variety of topics

From Astronomy to ZIP drives

The HTML versions are available in the package howtoenh

In the directory /usr/share/doc/howto/en/html

May not be installed by default

Be aware that HOWTOs are maintained by a very loosely knit group

of people on a best effort basis. They may be

Out of date, poorly written, or wrong

Intended for a Linux distribution other than SuSE

Online repository of HOWTOs at www.linuxdoc.org

3.4 SuSE Support Database and Hardware

Database

SuSE maintain an online support database at:

There is also a hardware compatibility database at:

http://sdb.suse.de/sdb/en/html/index.html

http://cdb.suse.de

Other useful web sites include:

http://www.linux.org

http://www.linux.com

http://en.tldp.org

http://www.linuxplanet.com

http://www.cert.org

(Security related)

http://www.kernel.org

(Kernel related)

End of chapter quiz

What command would you use to:

Browse through a text file

Delete a file

Display the first 5 lines of a file

Find a file based on its size and owner

List a directory

Look up on-line documentation

Display the name of your current

directory

Display your user ID

Search a file for a string

Count the number of lines in a file

Delete an (empty) directory

Create a symbolic link

Delete a non-empty directory

Create a directory
Copy a file