0% found this document useful (0 votes)

548 views

Rute User's Tutorial and Exposition

Author: "we exploit the multidimensional nature of the Universe to cut down on manufacturing costs" author: "the way to get things done is to Book a ticket, get a visa, pack a bag and it just happens." author: "it's not a matter of if, But when"

Uploaded by

crucerescu_alex

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

548 views

Rute User's Tutorial and Exposition

Uploaded by

crucerescu_alex

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 330

Pages up to and including this page are not included by Prentice Hall.

LINUX: Rute Users Tutorial and Exposition

Paul Sheer August 14, 2001

The reason we dont sell billions and billions of Guides, continued Harl, after wiping his mouth, is the expense. What we do is we sell one Guide billions and billions of times. We exploit the multidimensional nature of the Universe to cut down on manufacturing costs. And we dont sell to penniless hitchhikers. What a stupid notion that was! Find the one section of the market that, more or less by denition, doesnt have any money, and try to sell to it. No. We sell to the afuent business traveler and his vacationing wife in a billion, billion different futures. This is the most radical, dynamic and thrusting business venture in the entire multidimensional innity of space-time-probability ever. ... Ford was completely at a loss for what to do next. Look, he said in a stern voice. But he wasnt certain how far saying things like Look in a stern voice was necessarily going to get him, and time was not on his side. What the hell, he thought, youre only young once, and threw himself out of the window. That would at least keep the element of surprise on his side. ... In a spirit of scientic inquiry he hurled himself out of the window again. Douglas Adams Mostly Harmless

Strangely, the thing that least intrigued me was how theyd managed to get it all done. I suppose I sort of knew. If Id learned one thing from traveling, it was that the way to get things done was to go ahead and do them. Dont talk about going to Borneo. Book a ticket, get a visa, pack a bag, and it just happens. Alex Garland The Beach

Chapter Summary

Chapter Summary
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Computing Sub-basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . PC Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regular Expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Editing Text Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shell Scripting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Streams and sed The Stream Editor . . . . . . . . . . . . . . . . . . . Processes, Environment Variables . . . . . . . . . . . . . . . . . . . . . . Mail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 5 15 25 49 53 61 73 81 97

27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 A B C D E Index

DNS and Name Resolution . . . . . . Network File System, NFS . . . . . . Services Running Under inetd . . . exim and sendmail . . . . . . . . . . lilo, initrd, and Booting . . . . . init, ?getty, and U NIX Run Levels

. . . . .

. . . . . . . . . . . . . . .

273 285 291 299 317 325 333 337 347 389 409 413 425 437 453

. . . . . . . . . . . . . . Sending Faxes . . . . . . . . . . . . . . . . . . . . . . . . . . . uucp and uux . . . . . . . . . . . . . . . . . . . . . . . . . . . . The L INUX File System Standard . . . . . . . . . . . . . . . . httpd Apache Web Server . . . . . . . . . . . . . . . . . . crond and atd . . . . . . . . . . . . . . . . . . . . . . . . . . . postgres SQL Server . . . . . . . . . . . . . . . . . . . . . . smbd Samba NT Server . . . . . . . . . . . . . . . . . . . . named Domain Name Server . . . . . . . . . . . . . . . . . Point-to-Point Protocol Dialup Networking . . . . . . . . The L INUX Kernel Source, Modules, and Hardware Support

User Accounts and Ownerships . . . . . . . . . . . . . . . . . . . . . . . 101 Using Internet Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 L INUX Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 Permission and Modication Times . . . . . . . . . . . . . . . . . . . . . 123 Symbolic and Hard Links . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 Pre-installed Documentation . . . . . . . . . . . . . . . . . . . . . . . . . 131 Overview of the U NIX Directory Layout . . . . . . . . . . . . . . . . . . 135 U NIX Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Partitions, File Systems, Formatting, Mounting . . . . . . . . . . . . . . 153 Advanced Shell Scripting . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 System Services and lpd . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 Trivial Introduction to C . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 Shared Libraries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 Source and Binary Packages . . . . . . . . . . . . . . . . . . . . . . . . . 237 Introduction to IP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247 TCP and UDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 vii

The X Window System . . . . . . U NIX Security . . . . . . . . . . . Lecture Schedule . . . . . . . . . . LPI Certication Cross-Reference

. . . . . . 463 . . . . . . . . . . . . . . . . . . . . . . 485

. . . . . . . . . . . . . . . . . . . . . . 511 . . . . . . . . . . . . . . . . . . . . . . 525 . . . . . . . . . . . . . . . . . . . . . . 531

RHCE Certication Cross-Reference . . . . . . . . . . . . . . . . . . . . 543 L INUX Advocacy FAQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 551 The GNU General Public License Version 2 . . . . . . . . . . . . . . . . 573 581

viii

Contents

3.3 3.4 3.5 4

CMOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Serial Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Modems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20 20 23 25 25 26 29 29 32 33 34 34 35 36 36 40 40 41 41 42 43 44 45 46 47 49 49 51 51 52 52

Basic Commands 4.1 4.2 4.3 The ls Command, Hidden Files, Command-Line Options . . . . . . . . Error Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wildcards, Names, Extensions, and glob Expressions . . . . . . . . . . . 4.3.1 File naming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Glob expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.2 4.4 4.5 4.6 4.7 4.8 4.9

Contents
Acknowledgments 1 Introduction 1.1 What This Book Covers . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Read This Next. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 What Do I Need to Get Started? . . . . . . . . . . . . . . . . . . . . . 1.4 More About This Book . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 I Get Frustrated with U NIX Documentation That I Dont Understand 1.6 LPI and RHCE Requirements . . . . . . . . . . . . . . . . . . . . . . . 1.7 Not RedHat: RedHat-like . . . . . . . . . . . . . . . . . . . . . . . . . 1.8 Updates and Errata . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Computing Sub-basics 2.1 Binary, Octal, Decimal, and Hexadecimal 2.2 Files . . . . . . . . . . . . . . . . . . . . . 2.3 Commands . . . . . . . . . . . . . . . . . 2.4 Login and Password Change . . . . . . . 2.5 Listing Files . . . . . . . . . . . . . . . . . 2.6 Command-Line Editing Keys . . . . . . . 2.7 Console Keys . . . . . . . . . . . . . . . . 2.8 Creating Files . . . . . . . . . . . . . . . . 2.9 Allowable Characters for File Names . . 2.10 Directories . . . . . . . . . . . . . . . . . . xxxi 1 1 1 1 2 2 2 3 3 5 5 7 8 9 10 10 11 12 12 12 15 15 19

Usage Summaries and the Copy Command . . . . . . . . . . . . . . . . . Directory Manipulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . Relative vs. Absolute Pathnames . . . . . . . . . . . . . . . . . . . . . . . . System Manual Pages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . System info Pages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Some Basic Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . .

4.10 The mc File Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.11 Multimedia Commands for Fun . . . . . . . . . . . . . . . . . . . . . . . 4.12 Terminating Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.13 Compressed Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.14 Searching for Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.15 Searching Within Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.16 Copying to MS-DOS and Windows Formatted Floppy Disks . . . . . . . 4.17 Archives and Backups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.18 The PATH Where Commands Are Searched For . . . . . . . . . . . . . . 4.19 The -- Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Regular Expressions 5.1 5.2 5.3 5.4 5.5 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The fgrep Command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regular Expression \{ \} Notation . . . . . . . . . . . . . . . . . . . . . + ? \< \> ( ) | Notation . . . . . . . . . . . . . . . . . . . . . . . . . . . Regular Expression Subexpressions . . . . . . . . . . . . . . . . . . . . . x

. . . . . . . . . .

PC Hardware 3.1 Motherboard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Master/Slave IDE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

Contents

Editing Text Files 6.1 vi . . . . . . . . . . . 6.2 Syntax Highlighting 6.3 Editors . . . . . . . . 6.3.1 Cooledit . . . 6.3.2 vi and vim . 6.3.3 Emacs . . . . 6.3.4 Other editors

. . . . . . .

53 53 57 57 58 58 59 59 61 61 62 63 65 66 66 67 68 70 70 70 71 73 73 74 74 75 75 77 77 79 81 81 82 82

9.4 9.5 9.6 9.7 9.8 9.9 10 Mail

Creating Background Processes . . . . . . . . . . . . . . . . . . . . . . . . killing a Process, Sending Signals . . . . . . . . . . . . . . . . . . . . . List of Common Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . Niceness of Processes, Scheduling Priority . . . . . . . . . . . . . . . . . Process CPU/Memory Consumption, top . . . . . . . . . . . . . . . . . Environments of Processes . . . . . . . . . . . . . . . . . . . . . . . . . .

83 84 86 87 88 90 97 99 99 101

Shell Scripting 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . 7.2 Looping: the while and until Statements . . . . 7.3 Looping: the for Statement . . . . . . . . . . . . . 7.4 breaking Out of Loops and continueing . . . . 7.5 Looping Over Glob Expressions . . . . . . . . . . 7.6 The case Statement . . . . . . . . . . . . . . . . . 7.7 Using Functions: the function Keyword . . . . 7.8 Properly Processing Command-Line Args: shift 7.9 More on Command-Line Arguments: $@ and $0 . 7.10 Single Forward Quote Notation . . . . . . . . . . . 7.11 Double-Quote Notation . . . . . . . . . . . . . . . 7.12 Backward-Quote Substitution . . . . . . . . . . . . Streams and sed The Stream Editor 8.1 Introduction . . . . . . . . . . . . . . 8.2 Tutorial . . . . . . . . . . . . . . . . . 8.3 Piping Using | Notation . . . . . . . 8.4 A Complex Piping Example . . . . . 8.5 Redirecting Streams with >& . . . . 8.6 Using sed to Edit Streams . . . . . . 8.7 Regular Expression Subexpressions 8.8 Inserting and Deleting Lines . . . .

10.1 Sending and Reading Mail . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 The SMTP Protocol Sending Mail Raw to Port 25 . . . . . . . . . . . . 11 User Accounts and Ownerships

. . . . . . . . . . . .

11.1 File Ownerships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 11.2 The Password File /etc/passwd . . . . . . . . . . . . . . . . . . . . . . 102 11.3 Shadow Password File: /etc/shadow . . . . . . . . . . . . . . . . . . . 103 11.4 The groups Command and /etc/group . . . . . . . . . . . . . . . . . 104 11.5 Manually Creating a User Account . . . . . . . . . . . . . . . . . . . . . . 105 11.6 Automatically: useradd and groupadd . . . . . . . . . . . . . . . . . . 106 11.7 User Logins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 11.7.1 The login command . . . . . . . . . . . . . . . . . . . . . . . . . 106 11.7.2 The set user, su command . . . . . . . . . . . . . . . . . . . . . . . 107 11.7.3 The who, w, and users commands to see who is logged in . . . . 108 11.7.4 The id command and effective UID . . . . . . . . . . . . . . . . . 109 11.7.5 User limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 12 Using Internet Services 111

. . . . . . . .

12.1 ssh, not telnet or rlogin . . . . . . . . . . . . . . . . . . . . . . . . . 111 12.2 rcp and scp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 12.3 rsh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 12.4 FTP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 12.5 finger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 12.6 Sending Files by Email . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 12.6.1 uuencode and uudecode . . . . . . . . . . . . . . . . . . . . . . 114 12.6.2 MIME encapsulation . . . . . . . . . . . . . . . . . . . . . . . . . . 115 xii

Processes, Environment Variables 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 ps List Running Processes . . . . . . . . . . . . . . . . . . . . . . . . . 9.3 Controlling Jobs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

Contents

13 L INUX Resources

117

18.5.6 Tape backups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 18.5.7 Hiding program output, creating blocks of zeros . . . . . . . . . 149 18.6 Creating Devices with mknod and /dev/MAKEDEV . . . . . . . . . . . . 150 19 Partitions, File Systems, Formatting, Mounting 153

13.1 FTP Sites and the sunsite Mirror . . . . . . . . . . . . . . . . . . . . . . 117 13.2 HTTP Web Sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 13.3 SourceForge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 13.4 Mailing Lists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 13.4.1 Majordomo and Listserv . . . . . . . . . . . . . . . . . . . . . . . 119 13.4.2 *-request . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 13.5 Newsgroups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 13.6 RFCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 14 Permission and Modication Times 123

19.1 The Physical Disk Structure . . . . . . . . . . . . . . . . . . . . . . . . . . 153 19.1.1 Cylinders, heads, and sectors . . . . . . . . . . . . . . . . . . . . . 153 19.1.2 Large Block Addressing . . . . . . . . . . . . . . . . . . . . . . . . 154 19.1.3 Extended partitions . . . . . . . . . . . . . . . . . . . . . . . . . . 154 19.2 Partitioning a New Disk . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 19.3 Formatting Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 19.3.1 File systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 19.3.2 mke2fs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 19.3.3 Formatting oppies and removable drives . . . . . . . . . . . . . 161 19.3.4 Creating MS-DOS oppies . . . . . . . . . . . . . . . . . . . . . . 162 19.3.5 mkswap, swapon, and swapoff . . . . . . . . . . . . . . . . . . . 162 19.4 Device Mounting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 19.4.1 Mounting CD-ROMs . . . . . . . . . . . . . . . . . . . . . . . . . 163 19.4.2 Mounting oppy disks . . . . . . . . . . . . . . . . . . . . . . . . 164 19.4.3 Mounting Windows and NT partitions . . . . . . . . . . . . . . . 164 19.5 File System Repair: fsck . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 19.6 File System Errors on Boot . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 19.7 Automatic Mounts: fstab . . . . . . . . . . . . . . . . . . . . . . . . . . 166 19.8 Manually Mounting /proc . . . . . . . . . . . . . . . . . . . . . . . . . . 167 19.9 RAM and Loopback Devices . . . . . . . . . . . . . . . . . . . . . . . . . 167 19.9.1 Formatting a oppy inside a le . . . . . . . . . . . . . . . . . . . 167 19.9.2 CD-ROM les . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 19.10 Remounting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 19.11 Disk sync . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 20 Advanced Shell Scripting 20.2 Special Parameters: $?, $*,. . . 171 . . . . . . . . . . . . . . . . . . . . . . . . 172

14.1 The chmod Command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 14.2 The umask Command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 14.3 Modication Times: stat . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 15 Symbolic and Hard Links 127

15.1 Soft Links . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 15.2 Hard Links . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 16 Pre-installed Documentation 17 Overview of the U NIX Directory Layout 131 135

17.1 Packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 17.2 U NIX Directory Superstructure . . . . . . . . . . . . . . . . . . . . . . . . 136 17.3 L INUX on a Single Floppy Disk . . . . . . . . . . . . . . . . . . . . . . . . 138 18 U NIX Devices 141

18.1 Device Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 18.2 Block and Character Devices . . . . . . . . . . . . . . . . . . . . . . . . . 142 18.3 Major and Minor Device Numbers . . . . . . . . . . . . . . . . . . . . . . 143 18.4 Common Device Names . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 18.5 dd, tar, and Tricks with Block Devices . . . . . . . . . . . . . . . . . . . 147 18.5.1 Creating boot disks from boot images . . . . . . . . . . . . . . . . 147 18.5.2 Erasing disks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 18.5.3 Identifying data on raw disks . . . . . . . . . . . . . . . . . . . . . 148 18.5.4 Duplicating a disk . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 18.5.5 Backing up to oppies . . . . . . . . . . . . . . . . . . . . . . . . . 149 xiii

20.1 Lists of Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 20.3 Expansion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 20.4 Built-in Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 20.5 Trapping Signals the trap Command . . . . . . . . . . . . . . . . . . 176 xiv

Contents

20.6 Internal Settings the set Command . . . . . . . . . 20.7 Useful Scripts and Commands . . . . . . . . . . . . . . 20.7.1 chroot . . . . . . . . . . . . . . . . . . . . . . . 20.7.2 if conditionals . . . . . . . . . . . . . . . . . . . 20.7.3 patching and diffing . . . . . . . . . . . . . . 20.7.4 Internet connectivity test . . . . . . . . . . . . . 20.7.5 Recursive grep (search) . . . . . . . . . . . . . . 20.7.6 Recursive search and replace . . . . . . . . . . . 20.7.7 cut and awk manipulating text le elds . . 20.7.8 Calculations with bc . . . . . . . . . . . . . . . . 20.7.9 Conversion of graphics formats of many les . 20.7.10 Securely erasing les . . . . . . . . . . . . . . . . 20.7.11 Persistent background processes . . . . . . . . . 20.7.12 Processing the process list . . . . . . . . . . . . . 20.8 Shell Initialization . . . . . . . . . . . . . . . . . . . . . 20.8.1 Customizing the PATH and LD LIBRARY PATH 20.9 File Locking . . . . . . . . . . . . . . . . . . . . . . . . . 20.9.1 Locking a mailbox le . . . . . . . . . . . . . . . 20.9.2 Locking over NFS . . . . . . . . . . . . . . . . . 20.9.3 Directory versus le locking . . . . . . . . . . . 20.9.4 Locking inside C programs . . . . . . . . . . . . 21 System Services and lpd 21.1 Using lpr . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Downloading and Installing . . . . . . . . . . . . . . . . 21.3 LPRng vs. Legacy lpr-0.nn . . . . . . . . . . . . . . . 21.4 Package Elements . . . . . . . . . . . . . . . . . . . . . . 21.4.1 Documentation les . . . . . . . . . . . . . . . . 21.4.2 Web pages, mailing lists, and download points 21.4.3 User programs . . . . . . . . . . . . . . . . . . . 21.4.4 Daemon and administrator programs . . . . . . 21.4.5 Conguration les . . . . . . . . . . . . . . . . . 21.4.6 Service initialization les . . . . . . . . . . . . . 21.4.7 Spool les . . . . . . . . . . . . . . . . . . . . . . 21.4.8 Log les . . . . . . . . . . . . . . . . . . . . . . . 21.4.9 Log le rotation . . . . . . . . . . . . . . . . . . xv

. . . . . . . . . . . . . . . . . . . . .

177 178 178 179 179 180 180 181 182 183 183 184 184 185 186 187 187 188 190 190 191 193 193 194 195 195 195 195 196 196 196 196 197 198 198

21.4.10 Environment variables . . . . . . . . . . . . . . . . . . . . . . . . 199 21.5 The printcap File in Detail . . . . . . . . . . . . . . . . . . . . . . . . . 199 21.6 PostScript and the Print Filter . . . . . . . . . . . . . . . . . . . . . . . . . 200 21.7 Access Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 21.8 Printing Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . 203 21.9 Useful Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204 21.9.1 printtool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204 21.9.2 apsfilter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204 21.9.3 mpage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204 21.9.4 psutils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204 21.10 Printing to Things Besides Printers . . . . . . . . . . . . . . . . . . . . . . 205 22 Trivial Introduction to C 207

22.1 C Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208 22.1.1 The simplest C program . . . . . . . . . . . . . . . . . . . . . . . . 208 22.1.2 Variables and types . . . . . . . . . . . . . . . . . . . . . . . . . . 209 22.1.3 Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210 22.1.4 for, while, if, and switch statements . . . . . . . . . . . . . . 211 22.1.5 Strings, arrays, and memory allocation . . . . . . . . . . . . . . . 213 22.1.6 String operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 22.1.7 File operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217 22.1.8 Reading command-line arguments inside C programs . . . . . . 218 22.1.9 A more complicated example . . . . . . . . . . . . . . . . . . . . . 218 22.1.10 #include statements and prototypes . . . . . . . . . . . . . . . . 220 22.1.11 C comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 22.1.12 #define and #if C macros . . . . . . . . . . . . . . . . . . . 222 22.2 Debugging with gdb and strace . . . . . . . . . . . . . . . . . . . . . . 223 22.2.1 gdb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 22.2.2 Examining core les . . . . . . . . . . . . . . . . . . . . . . . . . 227 22.2.3 strace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 22.3 C Libraries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 22.4 C Projects Makefiles . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 22.4.1 Completing our example Makefile . . . . . . . . . . . . . . . . 231 22.4.2 Putting it all together . . . . . . . . . . . . . . . . . . . . . . . . . 231 xvi

. . . . . . . . . . . . .

Contents

23 Shared Libraries

233

23.1 Creating DLL .so Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 23.2 DLL Versioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234 23.3 Installing DLL .so Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235 24 Source and Binary Packages 237

26 TCP and UDP 26.1 The TCP Header . . . . . . . . . 26.2 A Sample TCP Session . . . . . . 26.3 User Datagram Protocol (UDP) . 26.4 /etc/services File . . . . . . 26.5 Encrypting and Forwarding TCP

. . . . .

263 264 265 268 269 270 273 273 274 275 276 277 281 281 281 282 283 284 285 285 286 288 289 289 291 291 291 292 292 293 294 294 295 295

24.1 Building GNU Source Packages . . . . . . . . . . . . . . . . . . . . . . . . 237 24.2 RedHat and Debian Binary Packages . . . . . . . . . . . . . . . . . . . . 240 24.2.1 Package versioning . . . . . . . . . . . . . . . . . . . . . . . . . . 240 24.2.2 Installing, upgrading, and deleting . . . . . . . . . . . . . . . . . 240 24.2.3 Dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241 24.2.4 Package queries . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241 24.2.5 File lists and le queries . . . . . . . . . . . . . . . . . . . . . . . . 242 24.2.6 Package verication . . . . . . . . . . . . . . . . . . . . . . . . . . 243 24.2.7 Special queries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244 24.2.8 dpkg/apt versus rpm . . . . . . . . . . . . . . . . . . . . . . . . . 245 24.3 Source Packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246 25 Introduction to IP 247

27 DNS and Name Resolution 27.1 Top-Level Domains (TLDs) . . . . . . . . 27.2 Resolving DNS Names to IP Addresses . 27.2.1 The Internet DNS infrastructure . 27.2.2 The name resolution process . . . 27.3 Conguring Your Local Machine . . . . . 27.4 Reverse Lookups . . . . . . . . . . . . . . 27.5 Authoritative for a Domain . . . . . . . . . 27.6 The host, ping, and whois Command . 27.7 The nslookup Command . . . . . . . . . 27.7.1 NS, MX, PTR, A and CNAME records 27.8 The dig Command . . . . . . . . . . . . . 28 Network File System, NFS 28.1 Software . . . . . . . . . 28.2 Conguration Example 28.3 Access Permissions . . . 28.4 Security . . . . . . . . . 28.5 Kernel NFS . . . . . . .

. . . . . . . . . . .

25.1 Internet Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247 25.2 Special IP Addresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249 25.3 Network Masks and Addresses . . . . . . . . . . . . . . . . . . . . . . . . 250 25.4 Computers on a LAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 25.5 Conguring Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251 25.6 Conguring Routing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252 25.7 Conguring Startup Scripts . . . . . . . . . . . . . . . . . . . . . . . . . . 254 25.7.1 RedHat networking scripts . . . . . . . . . . . . . . . . . . . . . . 254 25.7.2 Debian networking scripts . . . . . . . . . . . . . . . . . . . . . . 255 25.8 Complex Routing a Many-Hop Example . . . . . . . . . . . . . . . . . 256 25.9 Interface Aliasing Many IPs on One Physical Card . . . . . . . . . . . 259 25.10 Diagnostic Utilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260 25.10.1 ping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260 25.10.2 traceroute . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 25.10.3 tcpdump . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 xvii

. . . . .

29 Services Running Under inetd 29.1 The inetd Package . . . . . . . . . . . . . . . . . . 29.2 Invoking Services with /etc/inetd.conf . . . . 29.2.1 Invoking a standalone service . . . . . . . . 29.2.2 Invoking an inetd service . . . . . . . . . . 29.2.3 Invoking an inetd TCP wrapper service 29.2.4 Distribution conventions . . . . . . . . . . . 29.3 Various Service Explanations . . . . . . . . . . . . . 29.4 The xinetd Alternative . . . . . . . . . . . . . . . . 29.5 Conguration Files . . . . . . . . . . . . . . . . . . . xviii

. . . . . . . . .

Contents

29.5.1 Limiting access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296 29.6 Security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 30 exim and sendmail 30.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 30.1.1 How mail works . . . . . . . . . . . . . . . . 30.1.2 Conguring a POP/IMAP server . . . . . . 30.1.3 Why exim? . . . . . . . . . . . . . . . . . . . 30.2 exim Package Contents . . . . . . . . . . . . . . . . 30.3 exim Conguration File . . . . . . . . . . . . . . . . 30.3.1 Global settings . . . . . . . . . . . . . . . . . 30.3.2 Transports . . . . . . . . . . . . . . . . . . . . 30.3.3 Directors . . . . . . . . . . . . . . . . . . . . 30.3.4 Routers . . . . . . . . . . . . . . . . . . . . . 30.4 Full-blown Mail server . . . . . . . . . . . . . . . . . 30.5 Shell Commands for exim Administration . . . . . 30.6 The Queue . . . . . . . . . . . . . . . . . . . . . . . . 30.7 /etc/aliases for Equivalent Addresses . . . . . 30.8 Real-Time Blocking List Combating Spam . . . . 30.8.1 What is spam? . . . . . . . . . . . . . . . . . . 30.8.2 Basic spam prevention . . . . . . . . . . . . . 30.8.3 Real-time blocking list . . . . . . . . . . . . . 30.8.4 Mail administrator and user responsibilities 30.9 Sendmail . . . . . . . . . . . . . . . . . . . . . . . . . 31 lilo, initrd, and Booting 31.1 Usage . . . . . . . . . . . . . . . . . . . . . . . 31.2 Theory . . . . . . . . . . . . . . . . . . . . . . 31.2.1 Kernel boot sequence . . . . . . . . . 31.2.2 Master boot record . . . . . . . . . . . 31.2.3 Booting partitions . . . . . . . . . . . 31.2.4 Limitations . . . . . . . . . . . . . . . 31.3 lilo.conf and the lilo Command . . . . 31.4 Creating Boot Floppy Disks . . . . . . . . . . 31.5 SCSI Installation Complications and initrd 31.6 Creating an initrd Image . . . . . . . . . . 31.7 Modifying lilo.conf for initrd . . . . . 31.8 Using mkinitrd . . . . . . . . . . . . . . . . xix 299 299 299 301 301 301 302 303 304 305 306 306 308 309 310 311 311 312 313 313 314 317 317 318 318 318 318 319 319 321 322 322 324 324

. . . . . . . . . . . . . . . . . . . .

32 init, ?getty, and U NIX Run Levels 32.1 init the First Process . . . . . . . . . . 32.2 /etc/inittab . . . . . . . . . . . . . . . . 32.2.1 Minimal conguration . . . . . . . . 32.2.2 Rereading inittab . . . . . . . . . 32.2.3 The respawning too fast error 32.3 Useful Run Levels . . . . . . . . . . . . . . 32.4 getty Invocation . . . . . . . . . . . . . . . 32.5 Bootup Summary . . . . . . . . . . . . . . . 32.6 Incoming Faxes and Modem Logins . . . . 32.6.1 mgetty with character terminals . 32.6.2 mgetty log les . . . . . . . . . . . 32.6.3 mgetty with modems . . . . . . . . 32.6.4 mgetty receiving faxes . . . . . . .

. . . . . . . . . . . . .

325 325 326 326 328 328 328 329 329 330 330 330 330 331

33 Sending Faxes 333 33.1 Fax Through Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333 33.2 Setgid Wrapper Binary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335 34 uucp and uux 34.1 Command-Line Operation 34.2 Conguration . . . . . . . . 34.3 Modem Dial . . . . . . . . . 34.4 tty/UUCP Lock Files . . . 34.5 Debugging uucp . . . . . . 34.6 Using uux with exim . . . 34.7 Scheduling Dialouts . . . . 337 338 338 341 342 343 343 346 347 349 349 349 349 351 351 352 352

. . . . . . . . . . . .

. . . . . . .

35 The L INUX File System Standard 35.1 Introduction . . . . . . . . . . 35.1.1 Purpose . . . . . . . . 35.1.2 Conventions . . . . . 35.2 The Filesystem . . . . . . . . 35.3 The Root Filesystem . . . . . 35.3.1 Purpose . . . . . . . . 35.3.2 Requirements . . . . . 35.3.3 Specic Options . . .

. . . . . . . .

Contents

35.3.4 /bin : Essential user command binaries (for use by all users) . . 353 35.3.5 /boot : Static les of the boot loader . . . . . . . . . . . . . . . . . 354 35.3.6 /dev : Device les . . . . . . . . . . . . . . . . . . . . . . . . . . . 355 35.3.7 /etc : Host-specic system conguration . . . . . . . . . . . . . . 355 35.3.8 /home : User home directories (optional) . . . . . . . . . . . . . . 358 35.3.9 /lib : Essential shared libraries and kernel modules . . . . . . . . 358 35.3.10 /lib<qual> : Alternate format essential shared libraries (optional)359 35.3.11 /mnt : Mount point for a temporarily mounted lesystem . . . . 359 35.3.12 /opt : Add-on application software packages . . . . . . . . . . . 360 35.3.13 /root : Home directory for the root user (optional) . . . . . . . . 361 35.3.14 /sbin : System binaries . . . . . . . . . . . . . . . . . . . . . . . . 361 35.3.15 /tmp : Temporary les . . . . . . . . . . . . . . . . . . . . . . . . 362 35.4 The /usr Hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362 35.4.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362 35.4.2 Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363 35.4.3 Specic Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363 35.4.4 /usr/X11R6 : X Window System, Version 11 Release 6 (optional) 363 35.4.5 /usr/bin : Most user commands . . . . . . . . . . . . . . . . . . . 364 35.4.6 /usr/include : Directory for standard include les. . . . . . . . . 365 35.4.7 /usr/lib : Libraries for programming and packages . . . . . . . . 365 35.4.8 /usr/lib<qual> : Alternate format libraries (optional) . . . . . . 366 35.4.9 /usr/local : Local hierarchy . . . . . . . . . . . . . . . . . . . . . 366 35.4.10 /usr/sbin : Non-essential standard system binaries . . . . . . . . 367 35.4.11 /usr/share : Architecture-independent data . . . . . . . . . . . . 367 35.4.12 /usr/src : Source code (optional) . . . . . . . . . . . . . . . . . . 373 35.5 The /var Hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373 35.5.1 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373 35.5.2 Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373 35.5.3 Specic Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374 35.5.4 /var/account : Process accounting logs (optional) . . . . . . . . . 374 35.5.5 /var/cache : Application cache data . . . . . . . . . . . . . . . . 374 35.5.6 /var/crash : System crash dumps (optional) . . . . . . . . . . . . 376 35.5.7 /var/games : Variable game data (optional) . . . . . . . . . . . . 376 35.5.8 /var/lib : Variable state information . . . . . . . . . . . . . . . . 377 35.5.9 /var/lock : Lock les . . . . . . . . . . . . . . . . . . . . . . . . . 379 35.5.10 /var/log : Log les and directories . . . . . . . . . . . . . . . . . 379 xxi

35.5.11 /var/mail : User mailbox les (optional) . . . . . . . . . . . . . . 35.5.12 /var/opt : Variable data for /opt . . . . . . . . . . . . . . . . . . 35.5.13 /var/run : Run-time variable data . . . . . . . . . . . . . . . . . . 35.5.14 /var/spool : Application spool data . . . . . . . . . . . . . . . . . 35.5.15 /var/tmp : Temporary les preserved between system reboots . 35.5.16 /var/yp : Network Information Service (NIS) database les (optional) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.6 Operating System Specic Annex . . . . . . . . . . . . . . . . . . . . . . 35.6.1 Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.7 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.7.1 The FHS mailing list . . . . . . . . . . . . . . . . . . . . . . . . . . 35.7.2 Background of the FHS . . . . . . . . . . . . . . . . . . . . . . . . 35.7.3 General Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . 35.7.4 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.7.5 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . 35.7.6 Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 httpd Apache Web Server 36.1 Web Server Basics . . . . . . . . . . . . . . . 36.2 Installing and Conguring Apache . . . . . 36.2.1 Sample httpd.conf . . . . . . . . 36.2.2 Common directives . . . . . . . . . 36.2.3 User HTML directories . . . . . . . 36.2.4 Aliasing . . . . . . . . . . . . . . . . 36.2.5 Fancy indexes . . . . . . . . . . . . . 36.2.6 Encoding and language negotiation 36.2.7 Server-side includes SSI . . . . . 36.2.8 CGI Common Gateway Interface 36.2.9 Forms and CGI . . . . . . . . . . . . 36.2.10 Setuid CGIs . . . . . . . . . . . . . . 36.2.11 Apache modules and PHP . . . . . 36.2.12 Virtual hosts . . . . . . . . . . . . .

379 380 380 381 382 382 382 382 386 386 386 386 386 387 387 389 389 393 393 394 398 398 399 399 400 401 403 405 406 407 409 409 411 412

. . . . . . . . . . . . . .

37 crond and atd 37.1 /etc/crontab Conguration File . . . . . . . . . . . . . . . . . . . . . 37.2 The at Command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37.3 Other cron Packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii

Contents

38 postgres SQL Server 38.1 Structured Query Language . . . . . . . . . . . . . . . . . . . . . . . . . . 38.2 postgres . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.3 postgres Package Content . . . . . . . . . . . . . . . . . . . . . . . . . . 38.4 Installing and Initializing postgres . . . . . . . . . . . . . . . . . . . . 38.5 Database Queries with psql . . . . . . . . . . . . . . . . . . . . . . . . . 38.6 Introduction to SQL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.6.1 Creating tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.6.2 Listing a table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.6.3 Adding a column . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.6.4 Deleting (dropping) a column . . . . . . . . . . . . . . . . . . . . 38.6.5 Deleting (dropping) a table . . . . . . . . . . . . . . . . . . . . . . 38.6.6 Inserting rows, object relational . . . . . . . . . . . . . . . . . . 38.6.7 Locating rows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.6.8 Listing selected columns, and the oid column . . . . . . . . . . . 38.6.9 Creating tables from other tables . . . . . . . . . . . . . . . . . . . 38.6.10 Deleting rows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.6.11 Searches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.6.12 Migrating from another database; dumping and restoring tables as plain text . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38.6.13 Dumping an entire database . . . . . . . . . . . . . . . . . . . . . 38.6.14 More advanced searches . . . . . . . . . . . . . . . . . . . . . . . 38.7 Real Database Projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 smbd Samba NT Server 39.1 Samba: An Introduction by Christopher R. Hertel 39.2 Conguring Samba . . . . . . . . . . . . . . . . . . 39.3 Conguring Windows . . . . . . . . . . . . . . . . 39.4 Conguring a Windows Printer . . . . . . . . . . . 39.5 Conguring swat . . . . . . . . . . . . . . . . . . 39.6 Windows NT Caveats . . . . . . . . . . . . . . . . 40 named Domain Name Server 40.1 Documentation . . . . . . . . . . 40.2 Conguring bind . . . . . . . . 40.2.1 Example conguration . 40.2.2 Starting the name server

413 413 414 414 415 417 418 418 419 420 420 420 420 421 421 421 421 422 422 423 423 423 425 425 431 433 434 434 435 437 438 438 438 443

40.2.3 Conguration in detail . . . . . . . . . . . . . . . . . . . . . . . . 444 40.3 Round-Robin Load-Sharing . . . . . . . . . . . . . . . . . . . . . . . . . . 448 40.4 Conguring named for Dialup Use . . . . . . . . . . . . . . . . . . . . . . 449 40.4.1 Example caching name server . . . . . . . . . . . . . . . . . . . . 449 40.4.2 Dynamic IP addresses . . . . . . . . . . . . . . . . . . . . . . . . . 450 40.5 Secondary or Slave DNS Servers . . . . . . . . . . . . . . . . . . . . . . . 450 41 Point-to-Point Protocol Dialup Networking 453

41.1 Basic Dialup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453 41.1.1 Determining your chat script . . . . . . . . . . . . . . . . . . . . 455 41.1.2 CHAP and PAP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456 41.1.3 Running pppd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456 41.2 Demand-Dial, Masquerading . . . . . . . . . . . . . . . . . . . . . . . . . 458 41.3 Dialup DNS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460 41.4 Dial-in Servers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 460 41.5 Using tcpdump . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 462 41.6 ISDN Instead of Modems . . . . . . . . . . . . . . . . . . . . . . . . . . . 462 42 The L INUX Kernel Source, Modules, and Hardware Support 463

42.1 Kernel Constitution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 463 42.2 Kernel Version Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464 42.3 Modules, insmod Command, and Siblings . . . . . . . . . . . . . . . . . 464 42.4 Interrupts, I/O Ports, and DMA Channels . . . . . . . . . . . . . . . . . 466 42.5 Module Options and Device Conguration . . . . . . . . . . . . . . . . . 467 42.5.1 Five ways to pass options to a module . . . . . . . . . . . . . . . 467 42.5.2 Module documentation sources . . . . . . . . . . . . . . . . . . . 469 42.6 Conguring Various Devices . . . . . . . . . . . . . . . . . . . . . . . . . 470 42.6.1 Sound and pnpdump . . . . . . . . . . . . . . . . . . . . . . . . . . 470 42.6.2 Parallel port . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 472 42.6.3 NIC Ethernet, PCI, and old ISA . . . . . . . . . . . . . . . . . . 472 42.6.4 PCI vendor ID and device ID . . . . . . . . . . . . . . . . . . . . . 474 42.6.5 PCI and sound . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474 42.6.6 Commercial sound drivers . . . . . . . . . . . . . . . . . . . . . . 474 42.6.7 The ALSA sound project . . . . . . . . . . . . . . . . . . . . . . . 475 42.6.8 Multiple Ethernet cards . . . . . . . . . . . . . . . . . . . . . . . . 475 42.6.9 SCSI disks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 475 xxiv

. . . . . .

. . . .

xxiii

Contents

42.6.10 SCSI termination and cooling 42.6.11 CD writers . . . . . . . . . . 42.6.12 Serial devices . . . . . . . . . 42.7 Modem Cards . . . . . . . . . . . . . 42.8 More on LILO: Options . . . . . . . 42.9 Building the Kernel . . . . . . . . . . 42.9.1 Unpacking and patching . . 42.9.2 Conguring . . . . . . . . . . 42.10 Using Packaged Kernel Source . . . 42.11 Building, Installing . . . . . . . . . .

. . . . . . . . . .

477 477 479 480 481 481 481 482 483 483 485 485 491 491 492 492 493 493 493 494 495 495 496 497 497 497 498 498 498 499 499 500 501 504 505

43.9 Login Screen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 506 43.10 X Font Naming Conventions . . . . . . . . . . . . . . . . . . . . . . . . . 506 43.11 Font Conguration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 508 43.12 The Font Server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509 44 U NIX Security 511

44.1 Common Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511 44.1.1 Buffer overow attacks . . . . . . . . . . . . . . . . . . . . . . . . 512 44.1.2 Setuid programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513 44.1.3 Network client programs . . . . . . . . . . . . . . . . . . . . . . . 514 44.1.4 /tmp le vulnerability . . . . . . . . . . . . . . . . . . . . . . . . . 514 44.1.5 Permission problems . . . . . . . . . . . . . . . . . . . . . . . . . 514 44.1.6 Environment variables . . . . . . . . . . . . . . . . . . . . . . . . 515 44.1.7 Password snifng . . . . . . . . . . . . . . . . . . . . . . . . . . . 515 44.1.8 Password cracking . . . . . . . . . . . . . . . . . . . . . . . . . . . 515 44.1.9 Denial of service attacks . . . . . . . . . . . . . . . . . . . . . . . . 515 44.2 Other Types of Attack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516 44.3 Counter Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 516 44.3.1 Removing known risks: outdated packages . . . . . . . . . . . . 516 44.3.2 Removing known risks: compromised packages . . . . . . . . . . 517 44.3.3 Removing known risks: permissions . . . . . . . . . . . . . . . . 517 44.3.4 Password management . . . . . . . . . . . . . . . . . . . . . . . . 517 44.3.5 Disabling inherently insecure services . . . . . . . . . . . . . . . . 517 44.3.6 Removing potential risks: network . . . . . . . . . . . . . . . . . 518 44.3.7 Removing potential risks: setuid programs . . . . . . . . . . . . . 519 44.3.8 Making life difcult . . . . . . . . . . . . . . . . . . . . . . . . . . 520 44.3.9 Custom security paradigms . . . . . . . . . . . . . . . . . . . . . . 521 44.3.10 Proactive cunning . . . . . . . . . . . . . . . . . . . . . . . . . . . 522 44.4 Important Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523 44.5 Security Quick-Quiz . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 523 44.6 Security Auditing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524 A Lecture Schedule 525

43 The X Window System 43.1 The X Protocol . . . . . . . . . . . . . . . . 43.2 Widget Libraries and Desktops . . . . . . . 43.2.1 Background . . . . . . . . . . . . . . 43.2.2 Qt . . . . . . . . . . . . . . . . . . . 43.2.3 Gtk . . . . . . . . . . . . . . . . . . . 43.2.4 GNUStep . . . . . . . . . . . . . . . 43.3 XFree86 . . . . . . . . . . . . . . . . . . . . 43.3.1 Running X and key conventions . . 43.3.2 Running X utilities . . . . . . . . . . 43.3.3 Running two X sessions . . . . . . . 43.3.4 Running a window manager . . . . 43.3.5 X access control and remote display 43.3.6 X selections, cutting, and pasting . 43.4 The X Distribution . . . . . . . . . . . . . . 43.5 X Documentation . . . . . . . . . . . . . . . 43.5.1 Programming . . . . . . . . . . . . . 43.5.2 Conguration documentation . . . 43.5.3 XFree86 web site . . . . . . . . . . . 43.6 X Conguration . . . . . . . . . . . . . . . . 43.6.1 Simple 16-color X server . . . . . . . 43.6.2 Plug-and-Play operation . . . . . . 43.6.3 Proper X conguration . . . . . . . 43.7 Visuals . . . . . . . . . . . . . . . . . . . . . 43.8 The startx and xinit Commands . . . . xxv

. . . . . . . . . . . . . . . . . . . . . . . .

A.1 Hardware Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525 A.2 Student Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 525 A.3 Lecture Style . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 526 xxvi

Contents

B LPI Certication Cross-Reference 531 B.1 Exam Details for 101 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531 B.2 Exam Details for 102 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 536 C RHCE Certication Cross-Reference C.1 RH020, RH030, RH033, RH120, RH130, and RH133 C.2 RH300 . . . . . . . . . . . . . . . . . . . . . . . . . . C.3 RH220 (RH253 Part 1) . . . . . . . . . . . . . . . . . C.4 RH250 (RH253 Part 2) . . . . . . . . . . . . . . . . . D L INUX Advocacy FAQ D.1 D.2 D.3 D.4 L INUX Overview . . . . . . . L INUX, GNU, and Licensing L INUX Distributions . . . . . L INUX Support . . . . . . . . 543 543 544 547 549

. . . .

551 . . . . . . . . . . . . . . . . . . . . . . . . . 551 . . . . . . . . . . . . . . . . . . . . . . . . . 556 . . . . . . . . . . . . . . . . . . . . . . . . . 560 . . . . . . . . . . . . . . . . . . 563 . . . . . . . . . . . . . . . . . . 563 . . . . . . . . . . . . . . . . . . 567 . . . . . . . . . . . . . . . . . . 569 573 581

. . . . . . . D.5 L INUX Compared to Other Systems . . . D.6 Migrating to L INUX . . . . . . . . . . . . D.7 Technical . . . . . . . . . . . . . . . . . . .

E The GNU General Public License Version 2 Index

xxvii

xxviii

Contents

Preface
When I began working with GNU/L INUX in 1994, it was straight from the DOS world. Though U NIX was unfamiliar territory, L INUX books assumed that anyone using L INUX was migrating from System V or BSDsystems that I had never heard of. It is a sensible adage to create, for others to share, the recipe that you would most like to have had. Indeed, I am not convinced that a single unifying text exists, even now, without this book. Even so, I give it to you desperately incomplete; but there is only so much one can explain in a single volume. I hope that readers will now have a single text to guide them through all facets of GNU/L INUX.

xxix

xxx

Acknowledgments

Acknowledgments
A special thanks goes to my technical reviewer, Abraham van der Merwe, and my production editor, Jane Bonnell. Thanks to Jonathan Maltz, Jarrod Cinman, and Alan Tredgold for introducing me to GNU/Linux back in 1994 or so. Credits are owed to all A TEX, TEX, GhostScript, GhostView, Authe Free software developers that went into L A totrace, XFig, XV, Gimp, the Palatino font, the various L TEX extension styles, DVIPS, DVIPDFM, ImageMagick, XDVI, XPDF, and LaTeX2HTML without which this document would scarcely be possible. To name a few: John Bradley, David Carlisle, Eric Cooper, John Cristy, Peter Deutsch, Nikos Drakos, Mark Eichin, Brian Fox, Carsten Heinz, Spencer Kimball, Paul King, Donald Knuth, Peter Mattis, Frank Mittelbach, Ross Moore, Derek B. Noonburg, Johannes Plass, Sebastian Rahtz, Chet Ramey, Tomas Rokicki, Bob Scheier, Rainer Schoepf, Brian Smith, Supoj Sutanthavibul, Herb Swan, Tim Theisen, Paul Vojta, Martin Weber, Mark Wicks, Masatake Yamato, Ken Yap, Herman Zapf. Thanks to Christopher R. Hertel for contributing his introduction to Samba. An enormous thanks to the GNU project of the Free Software Foundation, to the countless developers of Free software, and to the many readers that gave valuable feedback on the web site.

xxxi

xxxii

1.4. More About This Book

1. Introduction

One hint: try and install as much as possible so that when I mention a software package in this text, you are likely to have it installed already and can use it immediately. Most cities with a sizable IT infrastructure will have a L INUX user group to help you source a cheap CD. These are getting really easy to install, and there is no longer much need to read lengthy installation instructions.

Chapter 1

1.4

More About This Book

Introduction
Whereas books shelved beside this one will get your feet wet, this one lets you actually paddle for a bit, then thrusts your head underwater while feeding you oxygen.

Chapter 16 contains a fairly comprehensive list of all reference documentation available on your system. This book supplements that material with a tutorial that is both comprehensive and independent of any previous U NIX knowledge. The book also aims to satisfy the requirements for course notes for a GNU /L INUX training course. Here in South Africa, I use the initial chapters as part of a 36-hour GNU /L INUX training course given in 12 lessons. The details of the layout for this course are given in Appendix A. Note that all L INUX systems are really composed mostly of GNU softsystem as L INUX in the way ware, but from now on I will refer to the GNU almost everyone (incorrectly) does.

1.1 What This Book Covers

This book covers GNU /L INUX system administration, for popular distributions like RedHat and Debian , as a tutorial for new users and a reference for advanced administrators. It aims to give concise, thorough explanations and practical examples of each aspect of a U NIX system. Anyone who wants a comprehensive text on (what is commercially called) L INUX need look no furtherthere is little that is not covered here.

1.5

I Get Frustrated with U NIX Documentation That I Dont Understand

1.2 Read This Next. . .

The ordering of the chapters is carefully designed to allow you to read in sequence without missing anything. You should hence read from beginning to end, in order that later chapters do not reference unseen material. I have also packed in useful examples which you must practice as you read.

Any system reference will require you to read it at least three times before you get a reasonable picture of what to do. If you need to read it more than three times, then there is probably some other information that you really should be reading rst. If you are reading a document only once, then you are being too impatient with yourself. It is important to identify the exact terms that you fail to understand in a document. Always try to backtrack to the precise word before you continue. Its also probably not a good idea to learn new things according to deadlines. Your U NIX knowledge should evolve by grace and fascination, rather than pressure.

1.6 1.3 What Do I Need to Get Started?

You will need to install a basic L INUX system. A number of vendors now ship pointand-click-install CDs: you should try get a Debian or RedHat-like distribution. 1

Linux Professionals Institute (LPI) and RedHat Certied Engineer (RHCE) Requirements

The difference between being able to pass an exam and being able to do something useful, of course, is huge. 2

1. Introduction

1.7. Not RedHat: RedHat-like

1.8. Updates and Errata

1. Introduction

The LPI and RHCE are two certications that introduce you to L INUX . This book covers far more than both these two certications in most places, but occasionally leaves out minor items as an exercise. It certainly covers in excess of what you need to know to pass both these certications. The LPI and RHCE requirements are given in Appendix B and C. These two certications are merely introductions to U NIX. To earn them, users are not expected to write nifty shell scripts to do tricky things, or understand the subtle or advanced features of many standard services, let alone be knowledgeable of the enormous numbers of non-standard and useful applications out there. To be blunt: you can pass these courses and still be considered quite incapable by the standards of companies that do system integration. System integration is my own term. It refers to the act
of getting L INUX to do nonbasic functions, like writing complex shell scripts; setting up wide-area dialup In networks; creating custom distributions; or interfacing database, web, and email services together.

fact, these certications make no reference to computer programming whatsoever.

1.7 Not RedHat: RedHat-like

Throughout this book I refer to examples specic to RedHat and Debian . What I actually mean by this are systems that use .rpm (redHat package manager) packages as opposed to systems that use .deb (debian) packagesthere are lots of both. This just means that there is no reason to avoid using a distribution like Mandrake, which is .rpm based and viewed by many as being better than RedHat. In short, brand names no longer have any meaning in the Free software community. (Note that the same applies to the word U NIX which we take to mean the common denominator between all the U NIX variants, including RISC, mainframe, and PC variants of both System V and BSD.)

1.8 Updates and Errata

Corrections to this book will be posted on http://www.icon.co.za/psheer/rute-errata.html. Please check this web page before notifying me of errors.

2.1. Binary, Octal, Decimal, and Hexadecimal

2. Computing Sub-basics

the digits 0 through 9 and the letters A through F, whereas octal numbers use only the digits 0 through 7. Hexadecimal is often abbreviated as hex. Consider a 4-digit binary number. It has 24 = 16 possible combinations and can therefore be easily represented by one of the 16 hex digits. A 3-digit binary number has 23 = 8 possible combinations and can thus be represented by a single octal digit. Hence, a binary number can be represented with hex or octal digits without much calculation, as shown in Table 2.1. Table 2.1 Binary hexadecimal, and octal representation Binary Octal Binary Hexadecimal 0000 0 000 0 1 001 1 0001 0010 2 010 2 3 011 3 0011 4 100 4 0100 0101 5 101 5 6 110 6 0110 7 111 7 0111 1000 8 9 1001 1010 A B 1011 C 1100 1101 D E 1110 1111 F

Chapter 2

Computing Sub-basics
This chapter explains some basics that most computer users will already be familiar with. If you are new to U NIX, however, you may want to gloss over the commonly used key bindings for reference. The best way of thinking about how a computer stores and manages information is to ask yourself how you would. Most often the way a computer works is exactly the way you would expect it to if you were inventing it for the rst time. The only limitations on this are those imposed by logical feasibility and imagination, but almost anything else is allowed.

2.1 Binary, Octal, Decimal, and Hexadecimal

When you rst learned to count, you did so with 10 digits. Ordinary numbers (like telephone numbers) are called base ten numbers. Postal codes that include letters and digits are called base 36 numbers because of the addition of 26 letters onto the usual 10 digits. The simplest base possible is base two which uses only two digits: 0 and 1. Now, a 7-digit telephone number has 10 10 10 10 10 10 10 = 107 = 10, 000, 000 possible combinations. A postal code with four characters has 364 = 1, 679, 616 possible combinations. However, an 8-digit binary number only has 28 = 256 possible combinations. Since the internal representation of numbers within a computer is binary and since it is rather tedious to convert between decimal and binary, computer scientists have come up with new bases to represent numbers: these are base sixteen and base eight, known as hexadecimal and octal, respectively. Hexadecimal numbers use 5
7 digits

A binary number 01001011 can be represented in hex as 4B and in octal as 113 by simply separating the binary digits into groups of four or three, respectively. In U NIX administration, and also in many programming languages, there is often the ambiguity of whether a number is in fact a hex, decimal, or octal number. For instance, a hex number 56 is 01010110, but an octal number 56 is 101110, whereas a decimal number 56 is 111000 (computed through a more tedious calculation). To distinguish between them, hex numbers are often prexed with the characters 0x, while octal numbers are prexed with a 0. If the rst digit is 1 through 9, then it is a decimal number that is probably being referred to. We would then write 0x56 for hex, and 056 for octal. Another representation is to append the letter H, D, O, or B (or h, d, o, b) to the number to indicate its base. U NIX makes heavy use of 8-, 16-, and 32-digit binary numbers, often representing them as 2-, 4-, and 8-digit hex numbers. You should get used to seeing numbers like 0xffff (or FFFFh), which in decimal is 65535 and in binary is 1111111111111111. 6

2. Computing Sub-basics

2.2. Files

2.3. Commands

2. Computing Sub-basics

2.2 Files
Oct Dec 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Hex 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F Char NUL SOH STX ETX EOT ENQ ACK BEL BS HT LF VT FF CR SO SI DLE DC1 DC2 DC3 DC4 NAK SYN ETB CAN EM SUB ESC FS GS RS US Oct 040 041 042 043 044 045 046 047 050 051 052 053 054 055 056 057 060 061 062 063 064 065 066 067 070 071 072 073 074 075 076 077

Table 2.2 ASCII character set

Dec 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 Hex 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F Char SPACE ! " # $ % & ( ) * + , . / 0 1 2 3 4 5 6 7 8 9 : ; < = > ? Oct Dec Hex Char 100 101 102 103 104 105 106 107 110 111 112 113 114 115 116 117 120 121 122 123 124 125 126 127 130 131 132 133 134 135 136 137 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z [ \ ] _ Oct 140 141 142 143 144 145 146 147 150 151 152 153 154 155 156 157 160 161 162 163 164 165 166 167 170 171 172 173 174 175 176 177 Dec 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 Hex 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F Char a b c d e f g h i j k l m n o p q r s t u v w x y z { | } DEL 000 001 002 003 004 005 006 007 010 011 012 013 014 015 016 017 020 021 022 023 024 025 026 027 030 031 032 033 034 035 036 037

Common to every computer system invented is the le. A le holds a single contiguous block of data. Any kind of data can be stored in a le, and there is no data that cannot be stored in a le. Furthermore, there is no kind of data that is stored anywhere else except in les. A le holds data of the same type, for instance, a single picture will be stored in one le. During production, this book had each chapter stored in a le. It is uncommon for different types of data (say, text and pictures) to be stored together in the same le because it is inconvenient. A computer will typically contain about 10,000 les that have a great many purposes. Each le will have its own name. The le name on a L INUX or U NIX machine can be up to 256 characters long. The le name is usually explanatoryyou might call a letter you wrote to your friend something like Mary Jones.letter (from now on, whenever you see the typewriter font A style of print: here is typewriter font. , it means that those are words that might be read off the screen of the computer). The name you choose has no meaning to the computer and could just as well be any other combination of letters or digits; however, you will refer to that data with that le name whenever you give an instruction to the computer regarding that data, so you would like it to be descriptive. It

is important to internalize the fact that computers do not have an interpretation for anything. A computer operates with a set of interdependent logical rules. Interdependent means that the rules have no apex, in the sense that computers have no xed or single way of working. For example, the reason a computer has les at all is because computer programmers have decided that this is the most universal and convenient way of storing data, and if you think about it, it really is.

The data in each le is merely a long list of numbers. The size of the le is just the length of the list of numbers. Each number is called a byte. Each byte contains 8 bits. Each bit is either a one or a zero and therefore, once again, there are 2 2 2 2 2 2 2 2 = 256 possible combinations. Hence a byte can only hold a number as large as 255. There is no type of data that cannot be represented as a list of bytes. Bytes are sometimes also called octets. Your letter to Mary will be encoded into bytes for storage on the computer. We all know that a television picture is just a sequence of dots on the screen that scan from left to right. In that way, a picture might be represented in a le: that is, as a sequence of bytes where each byte is interpreted as a level of brightness0 for black and 255 for white. For your letter, the convention is to store an A as 65, a B as 66, and so on. Each punctuation character also has a numerical equivalent. A mapping between numbers and characters is called a character mapping or a character set. The most common character set in use in the world today is the ASCII character set which stands for the American Standard Code for Information Interchange. Table 2.2 shows the complete ASCII mappings between characters and their hex, decimal, and octal equivalents.
8 bits 1 byte

2.3

Commands

The second thing common to every computer system invented is the command. You tell the computer what to do with single words typed into the computer one at a time. Modern computers appear to have done away with the typing of commands by having beautiful graphical displays that work with a mouse, but, fundamentally, all that is happening is that commands are being secretly typed in for you. Using commands is still the only way to have complete power over the computer. You dont really know anything about a computer until you come to grips with the commands it uses. Using , and then waiting a computer will very much involve typing in a word, pressing for the computer screen to spit something back at you. Most commands are typed in to do something useful to a le. 8

2. Computing Sub-basics

2.4. Login and Password Change

2.5. Listing Files

2. Computing Sub-basics

2.4 Login and Password Change

Turn on your L INUX box. After a few minutes of initialization, you will see the login prompt. A prompt is one or more characters displayed on the screen that you are expected to follow with some typing of your own. Here the prompt may state the name of the computer (each computer has a nametypically consisting of about eight lowercase letters) and then the word login:. L INUX machines now come with a graphical desktop by default (most of the time), so you might get a pretty graphical login with the same effect. Now you should type your login namea sequence of about eight lower case letters that would have been assigned to you by your computer ). administratorand then press the Enter (or Return) key (that is, A password prompt will appear after which you should type your password. Your password may be the same as your login name. Note that your password will not be shown on the screen as you type it but will be invisible. After typing your password, press the Enter or Return key again. The screen might show some message and prompt you for a log in againin this case, you have probably typed something incorrectly and should give it another try. From now on, you will be expected to know that the Enter or Return key should be pressed at the end of every line you type in, analogous to the mechanical typewriter. You will also be expected to know that human error is very common; when you type something incorrectly, the computer will give an error message, and you should try again until you get it right. It is uncommon for a person to understand computer concepts after a rst reading or to get commands to work on the rst try. Now that you have logged in you will see a shell prompta shell is the place where you can type commands. The shell is where you will spend most of your time as a system administrator Computer manager. , but it neednt look as bland as you see now. Your rst exercise is to change your password. Type the command passwd. You will be asked for a new password and then asked to conrm that password. The password you choose should consist of letters, numbers, and punctuationyou will see later on why this security measure is a good idea. Take good note of your password for the next time you log in. Then the shell will return. The password you have chosen will take effect immediately, replacing the previous password that you used to log in. The password command might also have given some message indicating what effect it actually had. You may not understand the message, but you should try to get an idea of whether the connotation was positive or negative.

2.5

Listing Files

Type in the command ls. ls is short for list, abbreviated to two letters like most other U NIX commands. ls lists all your current les. You may nd that ls does nothing, but just returns you back to the shell. This would be because you have no les as yet. Most U NIX commands do not give any kind of message unless something went wrong (the passwd command above was an exception). If there were les, you would see their names listed rather blandly in columns with no indication of what they are for.

2.6

Command-Line Editing Keys

The following keys are useful for editing the command-line. Note that U NIX has had a , and other keys may long and twisted evolution from the mainframe, and the not work properly. The following keys bindings are however common throughout many L INUX applications: Ctrl-a Move to the beginning of the line ( Ctrl-e Move to the end of the line ( Ctrl-h Erase backward ( Ctrl-d Erase forward ( ). ). ). ). ). ).

Ctrl-f Move forward one character ( Ctrl-b Move backward one character ( Alt-f Move forward one word. Alt-b Move backward one word. Alt-Ctrl-f Erase forward one word. Alt-Ctrl-b Erase backward one word. Ctrl-p Previous command (up arrow). Ctrl-n Next command (down arrow).

When you are using a computer, it is useful to imagine yourself as being in different places within the computer, rather than just typing commands into it. After you entered the passwd command, you were no longer in the shell, but moved into the password place. You could not use the shell until you had moved out of the passwd command. 9

, Ctrl for , and Shift for , mean to hold the Note that the prexes Alt for key down through the pressing and releasing of the letter key. These are known as key modiers. Note also, that the Ctrl key is always case insensitive; hence Ctrl-D (i.e. ) and Ctrl-d (i.e. ) are identical. The Alt modier (i.e., 10 ?) is

2. Computing Sub-basics

2.7. Console Keys

2.8. Creating Files

2. Computing Sub-basics

in fact a short way of pressing and releasing before entering the key combination; hence Esc then f is the same as Alt-fU NIX is different from other operating systems in this use of Esc. The Alt modier is not case insensitive although some applications will make a special effort to respond insensitively. The Alt key is also sometimes referred to as the Meta key. All of these keys are sometimes referred to by their abbreviations: for example, C-a for Ctrl-a, or M-f for Meta-f and Alt-f. The Ctrl modier is sometimes also designated with a caret: for example, C for Ctrl-C. Your command-line keeps a history of all the commands you have typed in. Ctrlp and Ctrl-n will cycle through previous commands entered. New users seem to gain tremendous satisfaction from typing in lengthy commands over and over. Never type in anything more than onceuse your command history instead. Ctrl-s is used to suspend the current session, causing the keyboard to stop responding. Ctrl-q reverses this condition. Ctrl-r activates a search on your command history. Pressing Ctrl-r in the middle of a search nds the next match whereas Ctrl-s reverts to the previous match (although some distributions have this confused with suspend). The Tab command is tremendously useful for saving key strokes. Typing a partial directory name, le name, or command, and then pressing Tab once or twice in sequence completes the word for you without your having to type it all in full. You can make Tab and other keys stop beeping in the irritating way that they do by editing the le /etc/inputrc and adding the line
set bell-style none

2.8

Creating Files

There are many ways of creating a le. Type cat > Mary Jones.letter and then type out a few lines of text. You will use this le in later examples. The cat command is used here to write from the keyboard into a le Mary Jones.letter. At the end one more time and then press . Now, if you type of the last line, press ls again, you will see the le Mary Jones.letter listed with any other les. Type cat Mary Jones.letter without the >. You will see that the command cat writes the contents of a le to the screen, allowing you to view your letter. It should match exactly what you typed in.

2.9

Allowable Characters for File Names

Although U NIX le names can contain almost any character, standards dictate that only the following characters are preferred in le names: A B C D E F G H I J K L M N O P Q R S T U V W X Y Z a b c d e f g h i j k l m n o p q r s t u v w x y z - 0 1 2 3 4 5 6 7 8 9 . Hence, never use other punctuation characters, brackets, or control characters to name les. Also, never use the space or tab character in a le name, and never begin a le name with a - character.

and then logging out and logging in again. (More about this later.)

2.10 2.7 Console Keys

There are several special keys interpreted directly by the L INUX console or text mode interface. The Ctrl-Alt-Del combination initiates a complete shutdown and hardware reboot, which is the preferred method of restarting L INUX . The Ctrl-PgUp and Ctrl-PgDn keys scroll the console, which is very useful for seeing text that has disappeared off the top of the terminal. You can use Alt-F2 to switch to a new, independent login session. Here you can log in again and run a separate session. There are six of these virtual consolesAltF1 through Alt-F6to choose from; they are also called virtual terminals. If you are in graphical mode, you will have to instead press Ctrl-Alt-F? because the Alt-F? keys are often used by applications. The convention is that the seventh virtual console is graphical, so Alt-F7 will always take you back to graphical mode. 11

Directories

I mentioned that a system may typically contain 10,000 les. Since it would be cumbersome if you were to see all 10,000 of them whenever you typed ls, les are placed in different cabinets so that les of the same type are placed together and can be easily isolated from other les. For instance, your letter above might go in a separate cabinet with other letters. A cabinet in computer terms is actually called a directory. This is the third commonality between all computer systems: all les go in one or another directory. To get an idea of how directories work, type the command mkdir letters, where mkdir stands for make directory. Now type ls. This will show the le Mary Jones.letter as well as a new le, letters. The le letters is not really a le at all, but the name of a directory in which a number of other les can be placed. To go into the directory letters, you can type cd letters where cd stands for change directory. Since the directory is newly created, you would not expect it to contain any les, and typing ls will verify such by not listing anything. You can now create a le by using the cat command as you did before (try this). To go back 12

2. Computing Sub-basics

2.10. Directories

2. Computing Sub-basics

to the original directory that you were in, you can use the command cd .. where the .. has the special meaning of taking you out of the current directory. Type ls again to verify that you have actually gone up a directory. It is, however, bothersome that we cannot tell the difference between les and directories. The way to differentiate is with the ls -l command. -l stands for long format. If you enter this command, you will see a lot of details about the les that may not yet be comprehensible to you. The three things you can watch for are the le name on the far right, the le size (i.e., the number of bytes that the le contains) in the fth column from the left, and the le type on the far left. The le type is a string of letters of which you will only be interested in one: the character on the far left is either a - or a d. A - signies a regular le, and a d signies a directory. The command ls -l Mary Jones.letter will list only the single le Mary Jones.letter and is useful for nding out the size of a single le. In fact, there is no limitation on how many directories you can create within each other. In what follows, you will glimpse the layout of all the directories on the computer. Type the command cd /, where the / has the special meaning to go to the topmost directory on the computer called the root directory. Now type ls -l. The listing may be quite long and may go off the top of the screen; in that case, try ls -l | less (then use PgUp and PgDn, and press q when done). You will see that most, if not all, are directories. You can now practice moving around the system with the cd command, not forgetting that cd .. takes you up and cd / takes you to the root directory. At any time you can type pwd (present working directory) to show the directory you are currently in. When you have nished, log out of the computer by using the logout command.

3.1. Motherboard

3. PC Hardware

Chapter 3

PC Hardware
This chapter explains a little about PC hardware. Readers who have built their own PC or who have conguring myriad devices on Windows can probably skip this section. It is added purely for completeness. This chapter actually comes under the subject of Microcomputer Organization, that is, how your machine is electronically structured.

3.1 Motherboard
Inside your machine you will nd a single, large circuit board called the motherboard (see Figure 3.1). It is powered by a humming power supply and has connector leads to the keyboard and other peripheral devices. Anything that is not the motherboard, not the power

supply and not purely mechanical.

The motherboard contains several large microchips and many small ones. The important ones are listed below. RAM Random Access Memory or just memory. The memory is a single linear sequence of bytes that are erased when there is no power. It contains sequences of simple coded instructions of one to several bytes in length. Examples are: add this number to that; move this number to this device; go to another part of RAM to get other instructions; copy this part of RAM to this other part. When your machine has 64 megs (64 megabytes), it has 64 1024 1024 bytes of RAM. Locations within that space are called memory addresses, so that saying memory address 1000 means the 1000th byte in memory. ROM A small part of RAM does not reset when the computer switches off. It is called ROM, Read Only Memory. It is factory xed and usually never changes through the life of a PC, hence the name. It overlaps the area of RAM close to the end of 15

Figure 3.1 Partially assembled motherboard

3. PC Hardware

3.1. Motherboard

3. PC Hardware

the rst megabyte of memory, so that area of RAM is not physically usable. ROM contains instructions to start up the PC and access certain peripherals. CPU Central Processing Unit. It is the thing that is called 80486, 80586, Pentium, or whatever. On startup, it jumps to memory address 1040475 (0xFE05B) and starts reading instructions. The rst instructions it gets are actually to fetch more instructions from disk and give a Boot failure message to the screen if it nds nothing useful. The CPU requires a timer to drive it. The timer operates at a high speed of hundreds of millions of ticks per second (hertz). Thats why the machine is named, for example, a 400 MHz (400 megahertz) machine. The MHz of the machine is roughly proportional to the number of instructions it can process per second from RAM. I/O ports Stands for Input/Output ports. The ports are a block of RAM that sits in parallel to the normal RAM. There are 65,536 I/O ports, hence I/O is small compared to RAM. I/O ports are used to write to peripherals. When the CPU writes a byte to I/O port 632 (0x278), it is actually sending out a byte through your parallel port. Most I/O ports are not used. There is no specic I/O port chip, though. There is more stuff on the motherboard: ISA slots ISA (eye-sah) is a shape of socket for plugging in peripheral devices like modem cards and sound cards. Each card expects to be talked to via an I/O port (or several consecutive I/O ports). What I/O port the card uses is sometimes congured by the manufacturer, and other times is selectable on the card through jumpers Little pin bridges that you can pull off with your ngers. or switches on the card. Other times still, it can be set by the CPU using a system called Plug and Pray This means that you plug the device in, then beckon your favorite deity for spiritual asindicate that it is ready to send or receive more bytes through an I/O port. They do this through 1 of 16 connectors inside the ISA slot. These are called Interrupt Request lines or IRQ lines (or sometimes just Interrupts), so numbered 0 through 15. Like I/O ports, the IRQ your card uses is sometimes also jumper selectable, sometimes not. If you unplug an old ISA card, you can often see the actual copper thread that goes from the IRQ jumper to the edge connector. Finally, ISA cards can also access memory directly through one of eight Direct Memory Access Channels or DMA Channels, which are also possibly selectable by jumpers. Not all cards use DMA, however.
sistance. Actually, some people complained that this might be taken seriouslyno, its a joke: the real term is Plug n Play or PnP. A card also sometimes needs to signal the CPU to

8-bit ISA slots Old motherboards have shorter ISA slots. You will notice yours is a double slot (called 16-bit ISA) with a gap between them. The larger slot can still take an older 8-bit ISA card: like many modem cards. PCI slots PCI (pee-see-eye) slots are like ISA but are a new standard aimed at highperformance peripherals like networking cards and graphics cards. They also use an IRQ, I/O port and possibly a DMA channel. These, however, are automatically congured by the CPU as a part of the PCI standard, hence there will rarely be jumpers on the card. AGP slots AGP slots are even higher performance slots for Accelerated Graphics Processors, in other words, cards that do 3D graphics for games. They are also autocongured. Serial ports A serial port connection may come straight from your motherboard to a socket on your case. There are usually two of these. They may drive an external modem and some kinds of mice and printers. Serial is a simple and cheap way to connect a machine where relatively slow (less that 10 kilobytes per second) data transfer speeds are needed. Serial ports have their own ISA card built into the motherboard which uses I/O port 0x3F80x3FF and IRQ 4 for the rst serial port (also called COM1 under DOS/Windows) and I/O port 0x2F80x2FF and IRQ 3 for COM2. A discussion on serial port technology proceeds in Section 3.4 below. Parallel port Normally, only your printer would plug in here. Parallel ports are, however, extremely fast (being able to transfer 50 kilobytes per second), and hence many types of parallel port devices (like CD-ROM drives that plug into a parallel port) are available. Parallel port cables, however, can only be a few meters in length before you start getting transmission errors. The parallel port uses I/O port 0x3780x37A and IRQ 7. If you have two parallel ports, then the second one uses I/O port 0x2780x27A, but does not use an IRQ at all. USB port The Universal Serial Bus aims to allow any type of hardware to plug into one plug. The idea is that one day all serial and parallel ports will be scrapped in favor of a single USB socket from which all external peripherals will daisy chain. I will not go into USB here. IDE ribbon The IDE ribbon plugs into your hard disk drive or C: drive on Windows/DOS and also into your CD-ROM drive (sometimes called an IDE CDROM). The IDE cable actually attaches to its own PCI card internal to the motherboard. There are two IDE connectors that use I/O ports 0xF0000xF007 and 0xF0080xF00F, and IRQ 14 and 15, respectively. Most IDE CD-ROMs are also ATAPI CD-ROMs. ATAPI is a standard (similar to SCSI, below) that enables many other kinds of devices to plug into an IDE ribbon cable. You get special oppy drives, tape drives, and other devices that plug into the same ribbon. They will be all called ATAPI-(this or that). 18

In summary, the peripheral and the CPU need to cooperate on three things: the I/O port, the IRQ, and the DMA. If any two cards clash by using either the same I/O port, IRQ number, or DMA channel then they wont work (at worst your machine will crash). Come to a halt and stop responding.

3. PC Hardware

3.2. Master/Slave IDE

3.3. CMOS

3. PC Hardware

SCSI ribbon Another ribbon might be present, coming out of a card (called the SCSI host adaptor or SCSI card) or your motherboard. Home PCs will rarely have SCSI, such being expensive and used mostly for high-end servers. SCSI cables are more densely wired than are IDE cables. They also end in a disk drive, tape drive, CD-ROM, or some other device. SCSI cables are not allowed to just-beplugged-in: they must be connected end on end with the last device connected in a special way called SCSI termination. There are, however, a few SCSI devices that are automatically terminated. More on this on page 477.

then called primary master, primary slave, secondary master, and secondary slave. Their labeling under L INUX is discussed in Section 18.4.

3.3

CMOS

The CMOS

3.2 Master/Slave IDE

Two IDE hard drives can be connected to a single IDE ribbon. The ribbon alone has nothing to distinguish which connector is which, so the drive itself has jumper pins on it (see Figure 3.2) that can be set to one of several options. These are one of Master (MA), Slave (SL), Cable Select (CS), or Master-only/Single-Drive/and-like. The MA option means that your drive is the rst drive of two on this IDE ribbon. The SL option means that your drive is the second drive of two on this IDE ribbon. The CS option means that your machine is to make its own decision (some boxes only work with this setting), and the Master-only option means that there is no second drive on this ribbon.

It is also known as the ROM BIOS conguration. You can start it instead of your operor (or something else) just after you switch your ating system (OS) by pressing machine on. There will usually be a message Press <key> to enter setup to explain this. Doing so will take you inside the CMOS program where you can change your machines conguration. CMOS programs are different between motherboard manufacturers. Inside the CMOS, you can enable or disable built-in devices (like your mouses and serial ports); set your machines hardware clock (so that your machine has the correct time and date); and select the boot sequence (whether to load the operating system off the hard drive or CD-ROMwhich you will need for installing L INUX from a bootable CD-ROM). Boot means to start up the computer. The term comes from the lack
of resources with which to begin: the operating system is on disk, but you might need the operating system to load from the disklike trying to lift yourself up from your bootstraps.

Stands for Complementary Metal Oxide Semiconductor, which has to do with the technology used to store setup information through power-downs. is a small application built into ROM.

You can also congure your hard drive. You should always select Hardrive autodetection Autodetection

refers to a system that, though having incomplete information, congures itself. In this case the CMOS program probes the drive to determine its capacity. Very old CMOS programs required you to enter the drives details manually. whenever installing a new machine or adding/removing disks. Dif-

ferent CMOSs will have different procedures, so browse through all the menus to see what your CMOS can do.

The CMOS is important when it comes to conguring certain devices built into the motherboard. Modern CMOSs allow you to set the I/O ports and IRQ numbers that you would like particular devices to use. For instance, you can make your CMOS switch COM1 with COM2 or use a non-standard I/O port for your parallel port. When it comes to getting such devices to work under L INUX , you will often have to power down your machine to see what the CMOS has to say about that device. More on this in Chapter 42.

Figure 3.2 Connection end of a typical IDE drive There might also be a second IDE ribbon, giving you a total of four possible drives. The rst ribbon is known as IDE1 (labeled on your motherboard) or the primary ribbon, and the second is known as IDE2 or the secondary ribbon. Your four drives are 19

3.4

Serial Devices

Serial ports facilitate low speed communications over a short distance using simple 8 core (or less) cable. The standards are old and communication is not particularly fault tolerant. There are so many variations on serial communication that it has become somewhat of a black art to get serial devices to work properly. Here I give a 20

3. PC Hardware

3.4. Serial Devices

3. PC Hardware

short explanation of the protocols, electronics, and hardware. The Serial-HOWTO and Modem-HOWTO documents contain an exhaustive treatment (see Chapter 16). Some devices that communicate using serial lines are: Ordinary domestic dial-up modems. Some permanent modem-like Internet connections. Mice and other pointing devices. Character text terminals. Printers. Cash registers. Magnetic card readers. Uninterruptible power supply (UPS) units. Embedded microprocessor devices. A device is connected to your computer by a cable with a 9-pin or 25-pin, male
1 2 3 4 5

bytes in a similar way. The computer and the serial device need to agree on a data rate (also called the serial port speed) so that the toggling and reading of voltage levels is properly synchronized. The speed is usually quoted in bps (bits per second). Table 3.2 shows a list of possible serial port speeds.

Table 3.2 Serial port speeds in bps 50 75 110 134 150 200 300 600 1,200 1,800 2,400 4,800 9,600 19,200 38,400 57,600 576,000 2,000,000 115,200 921,600 2,500,000 230,400 1,000,000 3,000,000 460,800 1,152,000 3,500,000 500,000 1,500,000 4,000,000

or female connector at each end. These are known as DB-9 (

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

) or DB-25

A typical mouse communicates between 1,200 and 9,600 bps. Modems communicate at 19,200, 38,400, 57,600, or 115,200 bps. It is rare to nd serial ports or peripherals that support the speeds not blocked in Table 3.2. To further synchronize the peripheral with the computer, an additional start bit proceeds each byte and up to two stop bits follow each byte. There may also be a parity bit which tells whether there is an even or odd number of 1s in the byte (for error checking). In theory, there may be as many as 12 bits sent for each data byte. These additional bits are optional and device specic. Ordinary modems communicate with an 8N1 protocol8 data bits, No parity bit, and 1 stop bit. A mouse communicates with 8 bits and no start, stop, or parity bits. Some devices only use 7 data bits and hence are limited to send only ASCII data (since ASCII characters range only up to 127). Some types of devices use two more pins called the request to send (RTS) and clear to send (CTS) pins. Either the computer or the peripheral pull the respective pin to +12 volts to indicate that it is ready to receive data. A further two pins call the DTR (data terminal ready) pin and the DSR (data set ready) pin are sometimes used instead these work the same way, but just use different pin numbers. In particular, domestic modems make full use of the RTS/CTS pins. This mechanism is called RTS/CTS ow control or hardware ow control. Some simpler devices make no use of ow control at all. Devices that do not use ow control will loose data which is sent without the receivers readiness. Some other devices also need to communicate whether they are ready to receive data, but do not have RTS/CTS pins (or DSR/DTR pins) available to them. These emit special control characters, sent amid the data stream, to indicate that ow should halt or restart. This is known as software ow control. Devices that optionally support either type of ow control should always be congured to use hardware ow control. In particular, a modem used with L INUX must have hardware ow control enabled. 22

(
14 15 16 17 18 19 20 21 22 23 24 25

) connectors. Only eight of the pins are ever used, how-

ever. See Table 3.1.

Table 3.1 Pin assignments for DB-9 and DB-25 sockets DB-9 pin number 3 2 7 8 6 4 1 9 5 DB-25 pin number 2 3 4 5 6 20 8 22 7 Direction Acronym TD RD RTS CTS DSR DTR CD RI Full-Name Transmit Data Receive Data Request To Send Clear To Send Data Set Ready Data Terminal Ready Data Carrier Detect Ring Indicator Signal Ground
PC device

The way serial devices communicate is very straightforward: A stream of bytes is sent between the computer and the peripheral by dividing each byte into eight bits. The voltage is toggled on a pin called the TD pin or transmit pin according to whether a bit is 1 or 0. A bit of 1 is indicated by a negative voltage (-15 to -5 volts) and a bit of 0 is indicated by a positive voltage (+5 to +15 volts). The RD pin or receive pin receives 21

3. PC Hardware

3.5. Modems

3. PC Hardware

Two other pins are the ring indicator (RI) pin and the carrier detect (CD) pin. These are only used by modems to indicate an incoming call and the detection of a peer modem, respectively. The above pin assignments and protocol (including some hard-core electrical specications which I have omitted) are known as RS-232. It is implemented using a standard chip called a 16550 UART (Universal Asynchronous Receiver-Transmitter) chip. RS-232 is easily effected by electrical noise, which limits the length and speed at which you can communicate: A half meter cable can carry 115,200 bps without errors, but a 15 meter cable is reliable at no more than 19,200 bps. Other protocols (like RS-423 or RS-422) can go much greater distances and there are converter appliances that give a more advantageous speed/distance tradeoff.

Converting voltages to pitches and back again is known as modulationdemodulation and is where the word modem comes from. The word baud means the number of possible pitch switches per second, which is sometimes used interchangeably with bps. There are many newer modulation techniques used to get the most out of a telephone line, so that 57,600 bps modems are now the standard (as of this writing). Modems also do other things to the data besides modulating it: They may pack the data to reduce redundancies (bit compression) and perform error detection and compensation (error correction). Such modem protocols are given names like V.90 (57,600 bps), V.34 (33,600 bps or 28,800 bps), V.42 (14,400 bps) or V.32 (14,400 bps and lower). When two modems connect, they need to negotiate a V protocol to use. This negotiation is based on their respective capabilities and the current line quality. A modem can be in one of two states: command mode or connect mode. A modem is connected if it can hear a peer modems carrier signal over a live telephone call (and is probably transmitting and receiving data in the way explained), otherwise it is in command mode. In command mode the modem does not modulate or transmit data but interprets special text sequences sent to it through the serial line. These text sequences begin with the letters AT and are called ATtention commands. AT commands are sent by your computer to congure your modem for the current telephone line conditions, intended function, and serial port capabilityfor example, there are commands to: enable automatic answering on ring; set the ow control method; dial a number; and hang up. The sequence of commands used to congure the modem is called the modem initialization string. How to manually issue these commands is discussed in Section 32.6.3, 34.3, and 41.1 and will become relevant when you want to dial your Internet service provider (ISP). Because each modem brand supports a slightly different set of modem commands, it is worthwhile familiarizing yourself with your modem manual. Most modern modems now support the Hayes command seta generic set of the most useful modem commands. However, Hayes has a way of enabling hardware ow control that many popular modems do not adhere to. Whenever in this book I give examples of modem initialization, I include a footnote referring to this section. It is usually sufcient to congure your modem to factory default settings, but often a second command is required to enable hardware ow control. There are no initialization strings that work on all modems. The web sites http://www.spy.net/dustin/modem/ and http://www.teleport.com/curt/modems.html are useful resources for nding out modem specications.

3.5 Modems
Telephone lines, having been designed to carry voice, have peculiar limitations when it comes to transmitting data. It turns out that the best way to send a binary digit over a telephone line is to beep it at the listener using two different pitches: a low pitch for 0 and a high pitch for 1. Figure 3.3 shows this operation schematically.

Figure 3.3 Communication between two remote computers by modem 23 24

4.2. Error Messages

4. Basic Commands

already familiar with the commandit will never be an exhaustive description of the usage. See the later explanation about man pages. The difference between a hidden le and an ordinary le is merely that the le name of a hidden le starts with a period. Hiding les in this way is not for security, but for convenience.

Chapter 4

The option ls -l is somewhat cryptic for the novice. Its more explanatory version is ls --format=long. Similarly, the all option can be given as ls --all, and means the same thing as ls -a.

Basic Commands
All of U NIX is case sensitive. A command with even a single letters capitalization altered is considered to be a completely different command. The same goes for les, directories, conguration le formats, and the syntax of all native programming languages.

4.2

Error Messages

4.1 The ls Command, Hidden Files, Command-Line Options

In addition to directories and ordinary text les, there are other types of les, although all les contain the same kind of data (i.e., a list of bytes). The hidden le is a le that will not ordinarily appear when you type the command ls to list the contents of a directory. To see a hidden le you must use the command ls -a. The -a option means to list all les as well as hidden les. Another variant is ls -l, which lists the contents in long format. The - is used in this way to indicate variations on a command. These are called command-line options or command-line arguments, and most U NIX commands can take a number of them. They can be strung together in any way that is convenient Commands under the GNU free software license are superior in this way: they have a greater number of options than traditional U NIX commands and are therefore more exible. , for example, ls -a -l, ls -l -a, or ls -al any of these will list all les in long format.

Although commands usually do not display a message when they execute The comsuccessfully, commands do report errors in a consistent format. The format varies from one command to another but often appears as follows: command-name: what was attempted: error message. For example, the command ls -l qwerty gives an error ls: qwerty: No such file or directory. What actually happened was that the command ls attempted to read the le qwerty. Since this le does not exist, an error code 2 arose. This error code corresponds to a situation where a le or directory is not being found. The error code is automatically translated into the sentence No such file or directory. It is important to understand the distinction between an explanatory message that a command gives (such as the messages reported by the passwd command in the previous chapter) and an error code that was just translated into a sentence. The reason is that a lot of different kinds of problems can result in an identical error code (there are only about a hundred different error codes). Experience will teach you that error messages do not tell you what to do, only what went wrong, and should not be taken as gospel.
puter accepted and processed the command.

The le /usr/include/asm/errno.h contains a complete list of basic error codes. In addition to these, several other header les Files ending in .h might dene their own error codes. Under U NIX, however, these are 99% of all the errors you are ever likely to get. Most of them will be meaningless to you at the moment but are included in Table 4.1 as a reference.

Table 4.1 L INUX error codes

Number 0 1 2 3 4 5 6 7 8 9 dene EPERM ENOENT ESRCH EINTR EIO ENXIO E2BIG ENOEXEC EBADF Message Success Operation not permitted No such le or directory No such process Interrupted system call Input/output error Device not congured Argument list too long Exec format error Bad le descriptor

commands take the additional arguments -h and --help. You can All GNU type a command with just this on the command-line and get a usage summary. This is some brief help that will summarize options that you may have forgotten if you are 25

continues...

4. Basic Commands

4.2. Error Messages

4. Basic Commands

Table 4.1 (continued)

Number 10 11 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 35 36 37 38 39 40 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 59 60 61 62 63 64 65 66 67 68 69 dene ECHILD EAGAIN EWOULDBLOCK ENOMEM EACCES EFAULT ENOTBLK EBUSY EEXIST EXDEV ENODEV ENOTDIR EISDIR EINVAL ENFILE EMFILE ENOTTY ETXTBSY EFBIG ENOSPC ESPIPE EROFS EMLINK EPIPE EDOM ERANGE EDEADLK EDEADLOCK ENAMETOOLONG ENOLCK ENOSYS ENOTEMPTY ELOOP EWOULDBLOCK ENOMSG EIDRM ECHRNG EL2NSYNC EL3HLT EL3RST ELNRNG EUNATCH ENOCSI EL2HLT EBADE EBADR EXFULL ENOANO EBADRQC EBADSLT EDEADLOCK EBFONT ENOSTR ENODATA ETIME ENOSR ENONET ENOPKG EREMOTE ENOLINK EADV ESRMNT Message No child processes Resource temporarily unavailable Resource temporarily unavailable Cannot allocate memory Permission denied Bad address Block device required Device or resource busy File exists Invalid cross-device link No such device Not a directory Is a directory Invalid argument Too many open les in system Too many open les Inappropriate ioctl for device Text le busy File too large No space left on device Illegal seek Read-only le system Too many links Broken pipe Numerical argument out of domain Numerical result out of range Resource deadlock avoided Resource deadlock avoided File name too long No locks available Function not implemented Directory not empty Too many levels of symbolic links (same as EAGAIN) No message of desired type Identier removed Channel number out of range Level 2 not synchronized Level 3 halted Level 3 reset Link number out of range Protocol driver not attached No CSI structure available Level 2 halted Invalid exchange Invalid request descriptor Exchange full No anode Invalid request code Invalid slot (same as EDEADLK) Bad font le format Device not a stream No data available Timer expired Out of streams resources Machine is not on the network Package not installed Object is remote Link has been severed Advertise error Srmount error Number 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 dene

Table 4.1 (continued)

Message Communication error on send Protocol error Multihop attempted RFS specic error Bad message Value too large for dened data type Name not unique on network File descriptor in bad state Remote address changed Can not access a needed shared library Accessing a corrupted shared library .lib section in a.out corrupted Attempting to link in too many shared libraries Cannot exec a shared library directly Invalid or incomplete multibyte or wide character Interrupted system call should be restarted Streams pipe error Too many users Socket operation on non-socket Destination address required Message too long Protocol wrong type for socket Protocol not available Protocol not supported Socket type not supported Operation not supported Protocol family not supported Address family not supported by protocol Address already in use Cannot assign requested address Network is down Network is unreachable Network dropped connection on reset Software caused connection abort Connection reset by peer No buffer space available Transport endpoint is already connected Transport endpoint is not connected Cannot send after transport endpoint shutdown Too many references: cannot splice Connection timed out Connection refused Host is down No route to host Operation already in progress Operation now in progress Stale NFS le handle Structure needs cleaning Not a XENIX named type le No XENIX semaphores available Is a named type le Remote I/O error Disk quota exceeded No medium found Wrong medium type ECOMM EPROTO EMULTIHOP EDOTDOT EBADMSG EOVERFLOW ENOTUNIQ EBADFD EREMCHG ELIBACC ELIBBAD ELIBSCN ELIBMAX ELIBEXEC EILSEQ ERESTART ESTRPIPE EUSERS ENOTSOCK EDESTADDRREQ EMSGSIZE EPROTOTYPE ENOPROTOOPT EPROTONOSUPPORT ESOCKTNOSUPPORT EOPNOTSUPP EPFNOSUPPORT EAFNOSUPPORT EADDRINUSE EADDRNOTAVAIL ENETDOWN ENETUNREACH ENETRESET ECONNABORTED ECONNRESET ENOBUFS EISCONN ENOTCONN ESHUTDOWN ETOOMANYREFS ETIMEDOUT ECONNREFUSED EHOSTDOWN EHOSTUNREACH EALREADY EINPROGRESS ESTALE EUCLEAN ENOTNAM ENAVAIL EISNAM EREMOTEIO EDQUOT ENOMEDIUM EMEDIUMTYPE

continues...

4. Basic Commands

4.3. Wildcards, Names, Extensions, and glob Expressions

4. Basic Commands

4.3 Wildcards, Names, Extensions, and glob Expressions

ls can produce a lot of output if there are a large number of les in a directory. Now say that we are only interested in les that ended with the letters tter. To list only these les, you can use ls *tter. The * matches any number of any other characters. So, for example, the les Tina.letter, Mary Jones.letter and the le splatter, would all be listed if they were present, whereas a le Harlette would not be listed. While the * matches any length of characters, the ? matches only one character. For example, the command ls ?ar* would list the les Mary Jones.letter and Harlette.

.csh csh shell script. .c program source code.

.db Database le. .dir X Window System font/other database directory. .deb Debian package for the Debian distribution.

.diff Output of the diff program indicating the difference between les or source trees.
A .dvi Device-independent le. Formatted output of .tex L TEX le.

4.3.1

File naming

.el Lisp program source. .g3 G3 fax format image le. .gif, .giff GIF image le. .gz File compressed with the gzip compression program. .htm, .html, .shtm, .html Hypertext Markup Language. A web page of some sort. .h /C++ program header le. preprocessor output.

When naming les, it is a good idea to choose names that group les of the same type together. You do this by adding an extension to the le name that describes the type of le it is. We have already demonstrated this by calling a le Mary Jones.letter instead of just Mary Jones. If you keep this convention, you will be able to easily list all the les that are letters by entering ls *.letter. The le name Mary Jones.letter is then said to be composed of two parts: the name, Mary Jones, and the extension, letter. Some common U NIX extensions you may see are: .a Archive. lib*.a is a static library. .alias X Window System font alias catalog. .avi Video format. .au Audio format (original Sun Microsystems generic sound le). .awk awk program source le.
A .bib bibtex L TEX bibliography source le.

.i SWIG source, or

.in configure input le. .info Info pages read with the info command. .jpg, .jpeg JPEG image le. .lj LaserJet le. Suitable input to a HP LaserJet printer. .log Log le of a system service. This le grows with status messages of some system program. .lsm L INUX Software Map entry.

.bmp Microsoft Bitmap le image format. .bz2 File compressed with the bzip2 compression program. .cc, .cxx, .C, .cpp C++ program source code. .cf, .cfg Conguration le or script. .cgi Executable script that produces web page output. .conf, .config Conguration le. 29

.lyx LyX word processor document. .man Man page. .mf Meta-Font font program source le. .pbm PBM image le format. .pcf PCF image leintermediate representation for fonts. X Window System font. .pcx PCX image le. 30

4. Basic Commands

4.3. Wildcards, Names, Extensions, and glob Expressions

4. Basic Commands

.pfb X Window System font le. .pdf Formatted document similar to PostScript or dvi. .php PHP program source code (used for web page design). .pl Perl program source code. .ps PostScript le, for printing or viewing. .py Python program source code. .rpm RedHat Package Manager rpm le. .sgml Standard Generalized Markup Language. Used to create documents to be converted to many different formats. .sh sh shell script. .so Shared object le. lib*.so is a Dynamically Linked Library.
code shared by more than one program to save disk space and memory.

.Z File compressed with the compress compression program. .zip File compressed with the pkzip (or PKZIP.EXE for DOS) compression program. .1, .2 . . . Man page. In addition, les that have no extension and a capitalized descriptive name are usually plain English text and meant for your reading. They come bundled with packages and are for documentation purposes. You will see them hanging around all over the place. Some full le names you may see are: AUTHORS List of people who contributed to or wrote a package.

Executable program

ChangeLog List of developer changes made to a package. COPYING Copyright (usually GPL) for a package. INSTALL Installation instructions. README Help information to be read rst, pertaining to the directory the README is in. TODO List of future desired work to be done to package. BUGS List of errata. NEWS Info about new features and changes for the layman about this package. THANKS List of contributors to a package. VERSION Version information of the package.

.spd Speedo X Window System font le. .tar tarred directory tree. .tcl Tcl/Tk source code (programming language). .texi, .texinfo Texinfo source. Info pages are compiled from these.
A A .tex TEX or L TEX document. L TEX is for document processing and typesetting.

.tga TARGA image le. .tgz Directory tree that has been archived with tar, and then compressed with gzip. Also a package for the Slackware distribution. .tiff TIFF image le.
A .tfm L TEX font metric le.

.ttf Truetype font. .txt Plain English text le. .voc Audio format (Soundblasters own format). .wav Audio format (sound les common to Microsoft Windows). .xpm XPM image le. .y yacc source le. 31

4.3.2

Glob expressions

There is a way to restrict le listings to within the ranges of certain characters. If you only want to list the les that begin with A through M, you can run ls [A-M]*. Here the brackets have a special meaningthey match a single character like a ?, but only those given by the range. You can use this feature in a variety of ways, for example, [a-dJW-Y]* matches all les beginning with a, b, c, d, J, W, X or Y; and *[a-d]id matches all les ending with aid, bid, cid or did; and *.{cpp,c,cxx} matches all les ending in .cpp, .c or .cxx. This way of specifying a le name is called a glob expression. Glob expressions are used in many different contexts, as you will see later. 32

4. Basic Commands

4.4. Usage Summaries and the Copy Command

4.5. Directory Manipulation

4. Basic Commands

4.4 Usage Summaries and the Copy Command

The command cp stands for copy. It duplicates one or more les. The format is cp <file> <newfile> cp <file> [<file> ...] <dir> or cp le newle cp le [le ...] dir The above lines are called a usage summary. The < and > signs mean that you dont actually type out these characters but replace <file> with a le name of your own. These are also sometimes written in italics like, cp le newle. In rare cases they are written in capitals like, cp FILE NEWFILE. <file> and <dir> are called parameters. Sometimes they are obviously numeric, like a command that takes <ioport>. Anyone emailing me to ask why typing in literal, <, i, o, p, o, r, t and > characters did not work will get a rude reply. These are common conventions used to specify the usage of a command. The

4.5

Directory Manipulation

The cd command is used to take you to different directories. Create a directory new with mkdir new. You could create a directory one by doing cd new and then mkdir one, but there is a more direct way of doing this with mkdir new/one. You can then change directly to the one directory with cd new/one. And similarly you can get back to where you were with cd ../... In this way, the / is used to represent directories within directories. The directory one is called a subdirectory of new. The command pwd stands for present working directory (also called the current directory) and tells what directory you are currently in. Entering pwd gives some output like /home/<username>. Experiment by changing to the root directory (with cd /) and then back into the directory /home/<username> (with cd /home/<username>). The directory /home/<username> is called your home directory, and is where all your personal les are kept. It can be used at any time with the abbreviation . In other words, entering cd /home/<username> is the same as entering cd . The process whereby a is substituted for your home directory is called tilde expansion. To remove (i.e., erase or delete) a le, use the command rm <filename>. To remove a directory, use the command rmdir <dir>. Practice using these two commands. Note that you cannot remove a directory unless it is empty. To remove a directory as well as any contents it might contain, use the command rm -R <dir>. The -R option species to dive into any subdirectories of <dir> and delete their contents. The process whereby a command dives into subdirectories of subdirectories of . . . is called recursion. -R stands for recursively. This is a very dangerous command. Although you may be used to undeleting les on other systems, on U NIX a deleted le is, at best, extremely difcult to recover. The cp command also takes the -R option, allowing it to copy whole directories. The mv command is used to move les and directories. It really just renames a le to a different directory. Note that with cp you should use the option -p and -d with -R to preserve all attributes of a le and properly reproduce symlinks (discussed later). Hence, always use cp -dpR <dir> <newdir> instead of cp R <dir> <newdir>.

[ and ] brackets are also not actually typed but mean that the contents between them are optional. The ellipses ... mean that <file> can be given repeatedly, and these also are never actually typed. From now on you will be expected to substitute your own parameters by interpreting the usage summary. You can see that the second of the above lines is actually just saying that one or more le names can be listed with a directory name last. From the above usage summary it is obvious that there are two ways to use the cp command. If the last name is not a directory, then cp copies that le and renames it to the le name given. If the last name is a directory, then cp copies all the les listed into that directory.

The usage summary of the ls command is as follows:

ls [-l, --format=long] [-a, --all] <file> <file> ... ls -al

where the comma indicates that either option is valid. Similarly, with the passwd command:
passwd [<username>]

4.6

Relative vs. Absolute Pathnames

You should practice using the cp command now by moving some of your les from place to place. 33

Commands can be given le name arguments in two ways. If you are in the same directory as the le (i.e., the le is in the current directory), then you can just enter the le name on its own (e.g., cp my file new file). Otherwise, you can enter the full path name, like cp /home/jack/my file /home/jack/new file. Very often administrators use the notation ./my file to be clear about the distinction, for instance, 34

4. Basic Commands

4.7. System Manual Pages

4.8. System info Pages

4. Basic Commands

cp ./my file ./new file. The leading ./ makes it clear that both les are relative to the current directory. File names not starting with a / are called relative path names, and otherwise, absolute path names.

information might be incomprehensible to you at this stage. Skim through the pages to get an idea of how they are structured and what headings they usually contain. Man pages are referenced with notation like cp(1), for the cp command in Section 1, which can be read with man 1 cp. This notation will be used from here on.

4.7 System Manual Pages

(See Chapter 16 for a complete overview of all documentation on the system, and also how to print manual pages in a properly typeset format.) The command man [<section>|-a] <command> displays help on a particular topic and stands for manual. Every command on the entire system is documented in so-named man pages. In the past few years a new format of documentation, called info, has evolved. This is considered the modern way to document commands, but most system documentation is still available only through man. Very few packages are not documented in man however. Man pages are the authoritative reference on how a command works because they are usually written by the very programmer who created the command. Under U NIX, any printed documentation should be considered as being second-hand information. Man pages, however, will often not contain the underlying concepts needed for understanding the context in which a command is used. Hence, it is not possible for a person to learn about U NIX purely from man pages. However, once you have the necessary background for a command, then its man page becomes an indispensable source of information and you can discard other introductory material. Now, man pages are divided into sections, numbered 1 through 9. Section 1 contains all man pages for system commands like the ones you have been using. Sections 2-7 contain information for programmers and the like, which you will probably not have to refer to just yet. Section 8 contains pages specically for system administration commands. There are some additional sections labeled with letters; other than these, there are no manual pages besides the sections 1 through 9. The sections are . . . /man1 User programs . . . /man2 System calls . . . /man3 Library calls . . . /man4 Special les . . . /man5 File formats . . . /man6 Games . . . /man7 Miscellaneous . . . /man8 System administration . . . /man9 Kernel documentation You should now use the man command to look up the manual pages for all the commands that you have learned. Type man cp, man mv, man rm, man mkdir, man rmdir, man passwd, man cd, man pwd, and of course man man. Much of the 35

4.8

System info Pages

info pages contain some excellent reference and tutorial information in hypertext linked format. Type info on its own to go to the top-level menu of the entire info hierarchy. You can also type info <command> for help on many basic commands. Some packages will, however, not have info pages, and other U NIX systems do not support info at all. info is an interactive program with keys to navigate and search documentation. Inwill invoke the help screen from where you can learn more comside info, typing mands.

4.9

Some Basic Commands

You should practice using each of these commands. bc A calculator program that handles arbitrary precision (very large) numbers. It is useful for doing any kind of calculation on the command-line. Its use is left as an exercise. cal [[0-12] 1-9999] Prints out a nicely formatted calender of the current month, a specied month, or a specied whole year. Try cal 1 for fun, and cal 9 1752, when the pope had a few days scrapped to compensate for roundoff error. cat <filename> [<filename> ...] Writes the contents of all the les listed to the screen. cat can join a lot of les together with cat <filename> <filename> ... > <newfile>. The le <newfile> will be an end-on-end concatenation of all the les specied. clear Erases all the text in the current terminal. date Prints out the current date and time. (The command time, though, does something entirely different.) df Stands for disk free and tells you how much free space is left on your system. The available space usually has the units of kilobytes (1024 bytes) (although on some other U NIX systems this will be 512 bytes or 2048 bytes). The right-most column 36

4. Basic Commands

4.9. Some Basic Commands

4. Basic Commands

tells the directory (in combination with any directories below that) under which that much space is available. dircmp Directory compare. This command compares directories to see if changes have been made between them. You will often want to see where two trees differ (e.g., check for missing les), possibly on different computers. Run man dircmp (that is, dircmp(1)). (This is a System 5 command and is not present on L INUX . You can, however, compare directories with the Midnight Commander, mc). du <directory> Stands for disk usage and prints out the amount of space occupied by a directory. It recurses into any subdirectories and can print only a summary with du -s <directory>. Also try du --max-depth=1 /var and du x / on a system with /usr and /home on separate partitions. See page 143.

and avoids the possibility of ever running out of memory (which can be quite fatal). head [-n <lines>] <filename> Prints the rst <lines> lines of a le or 10 lines if the -n option is not given. (See also tail below). hostname [<new-name>] With no options, hostname prints the name of your machine, otherwise it sets the name to <new-name>. kbdrate -r <chars-per-second> -d <repeat-delay> Changes the repeat rate of your keys. Most users will like this rate set to kbdrate -r 32 -d 250 which unfortunately is the fastest the PC can go. more Displays a long le by stopping at the end of each page. Run the following: ls -l /bin > bin-ls, and then try more bin-ls. The rst command creates a le with the contents of the output of ls. This will be a long le because the directory /bin has a great many entries. The second command views the le. . Use the space bar to page through the le. When you get bored, just press You can also try ls -l /bin | more which will do the same thing in one go. version of more, but with extra features. On your system, the two less The GNU commands may be the same. With less, you can use the arrow keys to page , and then up and down through the le. You can do searches by pressing typing in a word to search for and then pressing . Found words will be highlighted, and the text will be scrolled to the rst found word. The important commands are: Go to the end of a le. ssss Search backward through a le for the text ssss.

dmesg Prints a complete log of all messages printed to the screen during the bootup process. This is useful if you blinked when your machine was initializing. These messages might not yet be meaningful, however. echo Prints a message to the terminal. Try echo hello there, echo $[10*3+2], echo $[10*3+2]. The command echo -e allows interpretation of certain backslash sequences, for example echo -e "\a", which prints a bell, or in other words, beeps the terminal. echo -n does the same without printing the trailing newline. In other words, it does not cause a wrap to the next line after the text is printed. echo -e -n "\b", prints a back-space character only, which will erase the last character printed. exit Logs you out. expr <expression> Calculates the numerical expression expression. Most arithmetic operations that you are accustomed to will work. Try expr 5 + 10 * 2. Observe how mathematical precedence is obeyed (i.e., the * is worked out before the +). file <filename> Prints out the type of data contained in a le. file portrait.jpg will tell you that portrait.jpg is a JPEG image data, JFIF standard. The command file detects an enormous amount of le types, across every platform. file works by checking whether the rst few bytes of a le match certain tell-tale byte sequences. The byte sequences are called magic numbers. Their complete list is stored in /usr/share/magic.

ssss Search forward through a le for the text ssss.

expression. See Chapter 5 for more info.

Actually ssss is a regular

Scroll forward and keep trying to read more of the le in case some other program is appending to ituseful for log les. nnn Go to line nnn of the le. ). Quit. Used by many U NIX text-based applications (sometimes

The word magic under U

NIX normally refers to byte sequences or numbers that have a specic meaning or implication. So-called magic numbers are invented for source code, le formats, and le systems.

(You can make less stop beeping in the irritating way that it does by editing the le /etc/profile and adding the lines
LESS=-Q export LESS

free Prints out available free memory. You will notice two listings: swap space and physical memory. These are contiguous as far as the user is concerned. The swap space is a continuation of your installed memory that exists on disk. It is obviously slow to access but provides the illusion of much more available RAM 37

and then logging out and logging in again. But this is an aside that will make more sense later.) 38

4. Basic Commands

4.9. Some Basic Commands

4.10. The mc File Manager

4. Basic Commands

lynx <url> Opens a URL URL stands for Uniform Resource Locatora web address. console. Try lynx http://lwn.net/. links <url> Another text-based web browser.

at the

usleep <microseconds> Pauses (1/1,000,000 of a second).

for

microseconds

wc [-c] [-w] [-l] <filename> Counts the number of bytes (with -c for character), or words (with -w), or lines (with -l) in a le. whatis <command> Gives the rst line of the man page corresponding to <command>, unless no such page exists, in which case it prints nothing appropriate. whoami Prints your login name.

nohup <command> & Runs a command in the background, appending any output the command may produce to the le nohup.out in your home directory. nohup has the useful feature that the command will continue to run even after you have logged out. Uses for nohup will become obvious later. sleep <seconds> Pauses for <seconds> seconds. See also usleep. sort <filename> Prints a le with lines sorted in alphabetical order. Create a le called telephone with each line containing a short telephone book entry. Then type sort telephone, or sort telephone | less and see what happens. sort takes many interesting options to sort in reverse (sort -r), to eliminate duplicate entries (sort -u), to ignore leading whitespace (sort -b), and so on. See the sort(1) for details. strings [-n <len>] <filename> Writes out a binary le, but strips any unreadable characters. Readable groups of characters are placed on separate lines. If you have a binary le that you think may contain something interesting but looks completely garbled when viewed normally, use strings to sift out the interesting stuff: try less /bin/cp and then try strings /bin/cp. By default strings does not print sequences smaller than 4. The -n option can alter this limit. split ... Splits a le into many separate les. This might have been used when a le was too big to be copied onto a oppy disk and needed to be split into, say, 360-KB pieces. Its sister, csplit, can split les along specied lines of text within the le. The commands are seldom used on their own but are very useful within programs that manipulate text. tac <filename> [<filename> ...] Writes the contents of all the les listed to the screen, reversing the order of the linesthat is, printing the last line of the le rst. tac is cat backwards and behaves similarly. tail [-f] [-n <lines>] <filename> Prints the last <lines> lines of a le or 10 lines if the -n option is not given. The -f option means to watch the le for lines being appended to the end of it. (See also head above.) uname Prints the name of the U NIX operating system you are currently using. In this case, L INUX . uniq <filename> Prints a le with duplicate lines deleted. The le must rst be sorted. 39

4.10

The mc File Manager

Those who come from the DOS world may remember the famous Norton Commander le manager. The GNU project has a Free clone called the Midnight Commander, mc. It is essential to at least try out this packageit allows you to move around les and directories extremely rapidly, giving a wide-angle picture of the le system. This will drastically reduce the number of tedious commands you will have to type by hand.

4.11 Multimedia Commands for Fun

You should practice using each of these commands if you have your sound card congured. I dont want to give the impression that L INUX does not have graphical applications to do
all the functions in this section, but you should be aware that for every graphical application, there is a textmode one that works better and consumes fewer resources. You may also nd that some of these

packages are not installed, in which case you can come back to this later.

play [-v <volume>] <filename> Plays linear audio formats out through your sound card. These formats are .8svx, .aiff, .au, .cdr, .cvs, .dat, .gsm, .hcom, .maud, .sf, .smp, .txw, .vms, .voc, .wav, .wve, .raw, .ub, .sb, .uw, .sw, or .ul les. In other words, it plays almost every type of basic sound le there is: most often this will be a simple Windows .wav le. Specify <volume> in percent. rec <filename> Records from your microphone into a le. (play and rec are from the same package.) mpg123 <filename> Plays audio from MPEG les level 1, 2, or 3. Useful options are -b 1024 (for increasing the buffer size to prevent jumping) and --2to1 (downsamples by a factor of 2 for reducing CPU load). MPEG les contain sound and/or video, stored very compactly using digital signal processing techniques that the commercial software industry seems to think are very sophisticated. 40

4. Basic Commands

4.12. Terminating Commands

4.14. Searching for Files

4. Basic Commands

cdplay Plays a regular music CD

. cdp is the interactive version.

aumix Sets your sound cards volume, gain, recording volume, etc. You can use it interactively or just enter aumix -v <volume> to immediately set the volume in percent. Note that this is a dedicated mixer program and is considered to be an application separate from any that play music. Preferably do not set the volume from within a sound-playing application, even if it claims this featureyou have much better control with aumix. mikmod --interpolate -hq --renice Y <filename> Plays Mod les. Mod les are a special type of audio format that stores only the duration and pitch of the notes that constitute a song, along with samples of each musical instrument needed to play the song. This makes for high-quality audio with phenomenally small le size. mikmod supports 669, AMF, DSM, FAR, GDM, IMF, IT, MED, MOD, MTM, S3M, STM, STX, ULT, UNI, and XM audio formatsthat is, probably every type in existence. Actually, a lot of excellent listening music is available on the Internet in Mod le format. The most common formats are .it, .mod, .s3m, and .xm. Original .mod les are the product of Commodore-Amiga computers and
had only four tracks. Todays 16 (and more) track Mod les are comparable to any recorded music.

gzip -d <filename>. Now, use more to view the le again. Many les on the system are stored in compressed format. For example, man pages are often stored compressed and are uncompressed automatically when you read them. You previously used the command cat to view a le. You can use the command zcat to do the same thing with a compressed le. Gzip a le and then type zcat <filename>. You will see that the contents of the le are written to the screen. Generally, when commands and les have a z in them they have something to do with compressionthe letter z stands for zip. You can use zcat <filename> | less to view a compressed le proper. You can also use the command zless <filename>, which does the same as zcat <filename> | less. (Note that your less may actually have the functionality of zless combined.) A new addition to the arsenal is bzip2. This is a compression program very much like gzip, except that it is slower and compresses 20%30% better. It is useful for compressing les that will be downloaded from the Internet (to reduce the transfer volume). Files that are compressed with bzip2 have an extension .bz2. Note that the improvement in compression depends very much on the type of data being compressed. Sometimes there will be negligible size reduction at the expense of a huge speed penalty, while occasionally it is well worth it. Files that are frequently compressed and uncompressed should never use bzip2.

4.12 Terminating Commands

to stop an application or command that runs continuously. You usually use You must type this at the same prompt where you entered the command. If this doesnt work, the section on processes (Section 9.5) will explain about signalling a running application to quit.

4.14

Searching for Files

4.13 Compressed Files

Files typically contain a lot of data that one can imagine might be represented with a smaller number of bytes. Take for example the letter you typed out. The word the was probably repeated many times. You were probably also using lowercase letters most of the time. The le was by far not a completely random set of bytes, and it repeatedly used spaces as well as using some letters more than others. English text
in fact contains, on average, only about 1.3 useful bits (there are eight bits in a byte) of data per byte.

You can use the command find to search for les. Change to the root directory, and enter find. It will spew out all the les it can see by recursively descending Goes into into all subdirectories. each subdirectory and all its subdirectories, and repeats the command find. In other words, find, when executed from the root directory, prints all the les on the to system. find will work for a long time if you enter it as you havepress stop it.

Now change back to your home directory and type find again. You will see all your personal les. You can specify a number of options to find to look for specic les. find -type d Shows only directories and not the les they contain. find -type f Shows only les and not the directories that contain them, even though it will still descend into all directories. find -name <filename> Finds only les that have the name <filename>. For instance, find -name *.c will nd all les that end in a .c extension (find -name *.c without the quote characters will not work. You will see why later). find -name Mary Jones.letter will nd the le with the name Mary Jones.letter. 42

Because of this the le can be compressed to take up less space. Compression involves representing the same data by using a smaller number of bytes, in such a way that the original data can be reconstructed exactly. Such usually involves nding patterns in the data. The command to compress a le is gzip <filename>, which stands for GNU zip. Run gzip on a le in your home directory and then run ls to see what happened. Now, use more to view the compressed le. To uncompress the le use 41

4. Basic Commands

4.15. Searching Within Files

4.16. Copying to MS-DOS and Windows Formatted Floppy Disks 4. Basic Commands

find -size [[+|-]]<size> Finds only les that have a size larger (for +) or smaller (for -) than <size> kilobytes, or the same as <size> kilobytes if the sign is not specied. find <directory> [<directory> ...] Starts find in each of the specied directories. There are many more options for doing just about any type of search for a le. See find(1) for more details (that is, run man 1 find). Look also at the -exec option which causes find to execute a command for each le it nds, for example:
find /usr -type f -exec ls -al {} ;

Run grep for the word the to display all lines containing it: the Mary Jones.letter. Now try grep the *.letter.

grep

grep -n <pattern> <filename> shows the line number in the le where the word was found. grep -<num> <pattern> <filename> prints out <num> of the lines that came before and after each of the lines in which the word was found. grep -A <num> <pattern> <filename> prints out <num> of the lines that came After each of the lines in which the word was found. grep -B <num> <pattern> <filename> prints out <num> of the lines that came Before each of the lines in which the word was found. grep -v <pattern> <filename> prints out only those lines that do not contain You may think that the -v option is no longer doing the the word you are searching for.
same kind of thing that grep is advertised to do: i.e., searching for strings. In fact, U NIX commands often suffer from thisthey have such versatility that their functionality often overlaps with that of other commands. One actually never stops learning new and nifty ways of doing things hidden in the dark corners of man pages.

find has the deciency of actively reading directories to nd les. This process is slow, especially when you start from the root directory. An alternative command is locate <filename>. This searches through a previously created database of all the les on the system and hence nds les instantaneously. Its counterpart updatedb updates the database of les used by locate. On some systems, updatedb runs automatically every day at 04h00. Try these (updatedb will take several minutes):
updatedb locate rpm locate deb locate passwd locate HOWTO locate README

grep -i <pattern> <filename> does the same as an ordinary grep but is case insensitive.

4.16

Copying to MS-DOS and Windows Formatted Floppy Disks

4.15 Searching Within Files

Very often you will want to search through a number of les to nd a particular word or phrase, for example, when a number of les contain lists of telephone numbers with peoples names and addresses. The command grep does a line-by-line search through a le and prints only those lines that contain a word that you have specied. grep has the command summary:
grep [options] <pattern> <filename> [<filename> ...]

A package, called the mtools package, enables reading and writing to MSDOS/Windows oppy disks. These are not standard U NIX commands but are packaged with most L INUX distributions. The commands support Windows long le name oppy disks. Put an MS-DOS disk in your A: drive. Try
mdir A: touch myfile mcopy myfile A: mdir A:

The words word, string, or pattern are used synonymously in this context, basically meaning a short length
of letters and-or numbers that you are trying to nd matches for. A pattern can also be a string with kinds of wildcards in it that match different characters, as we shall see later.

Note that there is no such thing as an A: disk under L INUX . Only the mtools package understands A: in order to retain familiarity for MS-DOS users. The complete list of commands is
floppyd mattrib mcopy mdel mformat minfo mmount mmove mshowfat mtoolstest

4. Basic Commands

4.17. Archives and Backups

4.18. The PATH Where Commands Are Searched For

4. Basic Commands

mbadblocks mcat mcd

mdeltree mdir mdu

mkmanifest mlabel mmd

mpartition mrd mren

mtype mzip xcopy

It is obvious that you can call your archive anything you like, however; the common practice is to call it <directory>.tar, which makes it clear to all exactly what it is. Another important option is -p which preserves detailed attribute information of les. Once you have your .tar le, you would probably want to compress it with gzip. This will create a le <directory>.tar.gz, which is sometimes called <directory>.tgz for brevity. A second kind of archiving utility is cpio. cpio is actually more powerful than tar, but is considered to be more cryptic to use. The principles of cpio are quite similar and its use is left as an exercise.

Entering info mtools will give detailed help. In general, any MS-DOS command, put into lower case with an m prexed to it, gives the corresponding L INUX command.

4.17 Archives and Backups

Never begin any work before you have a fail-safe method of backing it up. One of the primary activities of a system administrator is to make backups. It is essential never to underestimate the volatility Ability to evaporate or become chaotic. of information in a computer. Backups of data are therefore continually made. A backup is a duplicate of your les that can be used as a replacement should any or all of the computer be destroyed. The idea is that all of the data in a directory As usual, meaning a directory and all its subdirectories and all the les in those subdirectories, etc. are stored in a separate placeoften compressedand can be retrieved in case of an emergency. When we want to store a number of les in this way, it is useful to be able to pack many les into one le so that we can perform operations on that single le only. When many les are packed together into one, this packed le is called an archive. Usually archives have the extension .tar, which stands for tape archive.

4.18

The PATH Where Commands Are Searched For

When you type a command at the shell prompt, it has to be read off disk out of one or other directory. On U NIX, all such executable commands are located in one of about four directories. A le is located in the directory tree according to its type, rather than according to what software package it belongs to. For example, a word processor may have its actual executable stored in a directory with all other executables, while its font les are stored in a directory with other fonts from all other packages. The shell has a procedure for searching for executables when you type them in. If you type in a command with slashes, like /bin/cp, then the shell tries to run the named program, cp, out of the /bin directory. If you just type cp on its own, then it tries to nd the cp command in each of the subdirectories of your PATH. To see what your PATH is, just type
echo $PATH

To create an archive of a directory, use the tar command:

tar -c -f <filename> <directory>

Create a directory with a few les in it, and run the tar command to back it up. A le of <filename> will be created. Take careful note of any error messages that tar reports. List the le and check that its size is appropriate for the size of the directory you are archiving. You can also use the verify option (see the man page) of the tar command to check the integrity of <filename>. Now remove the directory, and then restore it with the extract option of the tar command:
tar -x -f <filename>

You will see a colon separated list of four or more directories. Note that the current directory . is not listed. It is important that the current directory not be listed for reasons of security. Hence, to execute a command in the current directory, we hence always ./<command>. To append, for example, a new directory /opt/gnome/bin to your PATH, do 46

PATH="$PATH:/opt/gnome/bin" export PATH

You should see your directory recreated with all its les intact. A nice option to give to tar is -v. This option lists all the les that are being added to or extracted from the archive as they are processed, and is useful for monitoring the progress of archiving. 45

L INUX

supports the convenience of doing this in one line:

export PATH="$PATH:/opt/gnome/bin"

4. Basic Commands

4.19. The -- Option

4. Basic Commands

There is a further command, which, to check whether a command is locatable from the PATH. Sometimes there are two commands of the same name in different directories of the PATH. This is more often true of Solaris systems than L INUX . Typing which <command> locates the one that your shell would execute. Try:

which which which which

ls cp mv rm which cranzgots

which is also useful in shell scripts to tell if there is a command at all, and hence check whether a particular package is installed, for example, which netscape.

4.19 The -- Option

If a le name happens to begin with a - then it would be impossible to use that le name as an argument to a command. To overcome this circumstance, most commands take an option --. This option species that no more options follow on the commandlineeverything else must be treated as a literal le name. For instance
touch -- -stupid_file_name rm -- -stupid_file_name

5.1. Overview

5. Regular Expressions

Here are some regular expression examples that will teach you the regular expression basics. We use the grep command to show the use of regular expressions (remember that the -w option matches whole words only). Here the expression itself is enclosed in quotes for reasons that are explained later. grep -w t[a-i]e Matches the words tee, the, and tie. The brackets have a special signicance. They mean to match one character that can be anything from a to i. grep -w t[i-z]e Matches the words tie and toe.

Chapter 5

Regular Expressions
A regular expression is a sequence of characters that forms a template used to search for strings Words, phrases, or just about any sequence of characters. within text. In other words, it is a search pattern. To get an idea of when you would need to do this, consider the example of having a list of names and telephone numbers. If you want to nd a telephone number that contains a 3 in the second place and ends with an 8, regular expressions provide a way of doing that kind of search. Or consider the case where you would like to send an email to fty people, replacing the word after the Dear with their own name to make the letter more personal. Regular expressions allow for this type of searching and replacing.

grep -w cr[a-m]*t Matches the words craft, credit, and cricket. The * means to match any number of the previous character, which in this case is any character from a through m. grep -w kr.*n Matches the words kremlin and krypton, because the . matches any character and the * means to match the dot any number of times. egrep -w (th|sh).*rt Matches the words shirt, short, and thwart. The | means to match either the th or the sh. egrep is just like grep but supports extended regular expressions that allow for the | feature. The | character often denotes
a logical OR, meaning that either the thing on the left or the right of the | is applicable. This is true of Note how the square brackets mean one-of-severalmany programming languages.

characters and the round brackets with |s mean one-of-several-words.

grep -w thr[aeiou]*t Matches the words threat and throat. As you can see, a list of possible characters can be placed inside the square brackets. grep -w thr[a-f]*t Matches the words throughput and thrust. The after the rst bracket means to match any character except the characters listed. For example, the word thrift is not matched because it contains an f. The above regular expressions all match whole words (because of the -w option). If the -w option was not present, they might match parts of words, resulting in a far greater number of matches. Also note that although the * means to match any number of characters, it also will match no characters as well; for example: t[a-i]*e could actually match the letter sequence te, that is, a t and an e with zero characters between them. Usually, you will use regular expressions to search for whole lines that match, and sometimes you would like to match a line that begins or ends with a certain string. The character species the beginning of a line, and the $ character the end of the line. For example, The matches all lines that start with a The, and hack$ matches all lines that end with hack, and *The.*hack *$ matches all lines that begin with The and end with hack, even if there is whitespace at the beginning or end of the line. 50

5.1 Overview
Many utilities use the regular expression to give them greater power when manipulating text. The grep command is an example. Previously you used the grep command to locate only simple letter sequences in text. Now we will use it to search for regular expressions. In the previous chapter you learned that the ? character can be used to signify that any character can take its place. This is said to be a wildcard and works with le names. With regular expressions, the wildcard to use is the . character. So, you can use the command grep .3....8 <filename> to nd the seven-character telephone number that you are looking for in the above example. Regular expressions are used for line-by-line searches. For instance, if the seven characters were spread over two lines (i.e., they had a line break in the middle), then grep wouldnt nd them. In general, a program that uses regular expressions will consider searches one line at a time. 49

5. Regular Expressions

5.2. The fgrep Command

5.4. + ? \< \> ( ) | Notation

5. Regular Expressions

Because regular expressions use certain characters in a special way (these are . \ [ ] * + ?), these characters cannot be used to match characters. This restriction severely limits you from trying to match, say, le names, which often use the . character. To match a . you can use the sequence \. which forces interpretation as an actual . and not as a wildcard. Hence, the regular expression myfile.txt might match the letter sequence myfileqtxt or myfile.txt, but the regular expression myfile\.txt will match only myfile.txt. You can specify most special characters by adding a \ character before them, for example, use \[ for an actual [, a \$ for an actual $, a \\ for and actual \, \+ for an actual +, and \? for an actual ?. (? and + are explained below.)

5.4

Extended Regular Expression + ? \< \> ( ) | Notation with egrep

An enhanced version of regular expressions allows for a few more useful features. Where these conict with existing notation, they are only available through the egrep command. + is analogous to \{1,\}. It does the same as * but matches one or more characters instead of zero or more characters. ? is analogous to 1. It matches zero or one character. \< \> can surround a string to match only whole words.

5.2 The fgrep Command

fgrep is an alternative to grep. The difference is that while grep (the more commonly used command) matches regular expressions, fgrep matches literal strings. In other words you can use fgrep when you would like to search for an ordinary string that is not a regular expression, instead of preceding special characters with \.

( ) can surround several strings, separated by |. This notation will match any of these strings. (egrep only.)  can surround several strings, separated by \|. This notation will match any of these strings. (grep only.) The following examples should make the last two notations clearer. grep trot Matches the words electrotherapist, betroth, and so on, but

5.3 Regular Expression \{ \} Notation

x* matches zero to innite instances of a character x. You can specify other ranges of numbers of characters to be matched with, for example, x\{3,5\}, which will match at least three but not more than ve xs, that is xxx, xxxx, or xxxxx. x\{4\} can then be used to match 4 xs exactly: no more and no less. x\{7,\} will match seven or more xsthe upper limit is omitted to mean that there is no maximum number of xs. As in all the examples above, the x can be a range of characters (like [a-k]) just as well as a single charcter.

grep \<trot\> matches only the word trot. egrep -w (this|that|c[aeiou]*t) Matches the words this, that, cot, coat, cat, and cut.

5.5

Regular Expression Subexpressions

Subexpressions are covered in Chapter 8.

grep -w th[a-t]\{2,3\}t Matches the words theft, thirst, threat, thrift, and throat. grep -w th[a-t]\{4,5\}t Matches the words theorist, thicket, and thinnest. 51 52

6.1. vi

6. Editing Text Files

/usr/doc/vim-common-5.7/tutor, /usr/share/vim/vim56/tutor/tutor, or /usr/share/doc/vim-common-5.7/tutor/tutor, depending on your distribution.

By this you should be getting an idea of the kinds of differences there are between different L INUX distributions. You will then see the following at the top of

Chapter 6
5

your screen:

=============================================================================== = W e l c o m e t o t h e V I M T u t o r Version 1.4 = =============================================================================== Vim is a very powerful editor that has many commands, too many to explain in a tutor such as this. This tutor is designed to describe enough of the commands that you will be able to easily use Vim as an all-purpose editor.

Editing Text Files

The approximate time required to complete the tutor is 25-30 minutes,

To edit a text le means to interactively modify its content. The creation and modication of an ordinary text le is known as text editing. A word processor is a kind of editor, but more basic than that is the U NIX or DOS text editor.

You are supposed to edit the tutor le itself as practice, following through 6 lessons. Copy it rst to your home directory. Table 6.1 is a quick reference for vi. It contains only a few of the many hundreds of available commands but is enough to do all basic editing operations. Take note of the following: vi has several modes of operation. If you press , you enter insert-mode. You then enter text as you would in a normal DOS text editor, but you cannot arbitrarily will get you move the cursor and delete characters while in insert mode. Pressing out of insert mode, where you are not able to insert characters, but can now do things like arbitrary deletions and moves. (i.e., : ) gets you into command-line mode, where you can Pressing do operations like importing les, saving of the current le, searches, and text . processing. Typically, you type : then some text, and then hit The word register is used below. A register is a hidden clipboard. A useful tip is to enter :set ruler before doing anything. This shows, in the bottom right corner of the screen, what line and column you are on.

6.1

The important editor to learn how to use is vi. After that you can read why, and a little more about other, more user-friendly editors. Type simply,
vi <filename>

to edit any le, or the compatible, but more advanced

vim <filename>

To exit vi, press

, then the key sequence :q! and then press

vi has a short tutorial which should get you going in 20 minutes. If you get bored in the middle, you can skip it and learn vi as you need to edit things. To read the tutorial, enter:
vimtutor

which edits the le 53 54

6. Editing Text Files

6.1. vi

6. Editing Text Files

Table 6.1 Common vi commands Key combination h l k j b w { } $ gg G i o O a R :wq :q :q! x X dd :j! Ctrl-J u Ctrl-R de or or or or Function Cursor left Cursor right. Cursor up. Cursor down. Cursor left one word. Cursor right one word. Cursor up one paragraph. Cursor down one paragraph. Cursor to line start. Cursor to line end. Cursor to rst line. Cursor to last line. Get out of current mode. Start insert mode. Insert a blank line below the current line and then start insert mode. Insert a blank line above the current line and then start insert mode. Append (start insert mode after the current character). Replace (start insert mode with overwrite). Save (write) and quit. Quit. Quit forced (without checking whether a save is required). Delete (delete under cursor and copy to register). Backspace (delete left of cursor and copy to register). Delete line (and copy to register). Join line (remove newline at end of current line). Same. Undo. Redo. Delete to word end (and copy to register).
continues...

Table 6.1 (continued) Key combination db d$ d dd 2dd 5dd p Ctrl-G 5G 16G G /search-string ?search-string :-1,$s/search-string/replace-string/gc :,$s/search-string/replace-string/gc :,$s/\<search-string\>/replace-string/gc :8,22s/search-string/replace-string/g :%s/search-string/replace-string/g :w lename :5,20w lename :5,$w! lename :r lename v y d p Press v, then move cursor down a few lines, then, 56 Function Delete to word start (and copy to register). Delete to line end (and copy to register). Delete to line beginning (and copy to register). Delete current line (and copy to register). Delete two lines (and copy to register). Delete ve lines (and copy to register). Paste clipboard (insert register). Show cursor position. Cursor to line ve. Cursor to line sixteen. Cursor to last line. Search forwards for search-string. Search backwards for search-string. Search and replace with conrmation starting at current line. Search and replace with conrmation starting at line below cursor. Search and replace whole words. Search and replace in lines 8 through 22 without conrmation. Search and replace whole le without conrmation. Save to le lename. Save lines 5 through 20 to le lename (use Ctrl-G to get line numbers if needed). Force save lines 5 through to last line to le lename. Insert le lename. Visual mode (start highlighting). Copy highlighted text to register. Delete highlighted text (and copy to register). Paste clipboard (insert register). Search and replace within highlighted text.
continues...

6. Editing Text Files

6.2. Syntax Highlighting

6.3. Editors

6. Editing Text Files

Table 6.1 (continued) Key combination :s/search-string/replace-string/g :help Function Reference manual (open new window with help screen insideprobably the most important command here!). Open new blank window. Open new window with lename. Close current window. Close all windows. Move cursor to window below. Move cursor to window above. Make window smaller. Make window larger.

The editor is the place you will probably spend most of your time. Whether you are doing word processing, creating web pages, programming, or administrating. It is your primary interactive application.

6.3.1

Cooledit

:new :split lename :q :qa Ctrl-W j Ctrl-W k Ctrl-W Ctrl-W +

(Read this if you just-want-to-open-a-le-and-start-typing-like-under-Windows.)

author, I am probably biased in this view.

cooledit The best editor for day-to-day work is Cooledit, As Cooledits available from the Cooledit web page http://cooledit.sourceforge.net/. Cooledit is a graphical (runs under X) editor. It is also a full-featured Integrated Development Environment (IDE) for whatever you may be doing. Those considering buying an IDE for development need look no further than installing Cooledit for free. People coming from a Windows background will nd Cooledit the easiest and most powerful editor to use. It requires no tutelage; just enter cooledit under X and start typing. Its counterpart in text mode is mcedit, which comes with the GNU Midnight Commander package mc. The text-mode version is inferior to other text mode editors like emacs and jed but is adequate if you dont spend a lot of time in text mode. Cooledit has pull-down menus and intuitive keys. It is not necessary to read any documentation before using Cooledit.

6.2 Syntax Highlighting

Something all U NIX users are used to (and have come to expect) is syntax highlighting. This basically means that a bash (explained later) script will look like:

instead of Syntax highlighting is meant to preempt programming errors by colorizing correct keywords. You can set syntax highlighting in vim by using :syntax on (but not in vi). Enable syntax highlighting whenever possibleall good text editors support it.

6.3.2

vi and vim

6.3 Editors
Although U NIX has had full graphics capability for a long time now, most administration of low-level services still takes place inside text conguration les. Word processing is also best accomplished with typesetting systems that require creation of ordinary text les. This is in spite of all the hype regarding the WYSIWYG (what you see is what you get) word
A TEX and the Cooledit text editor. processor. This document itself was typeset with L

Today vi is considered the standard. It is the only editor that will be installed by default on any U NIX system. vim is a Charityware version that (as usual) improves upon the original vi with a host of features. It is important to learn the basics of vi even if your day-to-day editor is not going to be vi. The reason is that every administrator is bound to one day have to edit a text le over some really slow network link and vi is the best for this. On the other hand, new users will probably nd vi unintuitive and tedious and will spend a lot of time learning and remembering how to do all the things they need to. I myself cringe at the thought of vi pundits recommending it to new U NIX users. In defense of vi, it should be said that many people use it exclusively, and it is probably the only editor that really can do absolutely everything. It is also one of the few editors that has working versions and consistent behavior across all U NIX and non-U NIX systems. vim works on AmigaOS, AtariMiNT, BeOS, DOS, MacOS, OS/2, RiscOS, VMS, and Windows (95/98/NT4/NT5/2000) as well as all U NIX variants. 58

Historically, the standard text editor used to be ed. ed allows the user to see only one line of text of a le at a time (primitive by todays standards). Today, ed is mostly used in its streaming version, sed. ed has long since been superseded by vi. 57

6. Editing Text Files

6.3. Editors

6. Editing Text Files

6.3.3

Emacs

Emacs stands for Editor MACroS. It is the monster of all editors and can do almost everything one could imagine that a single software package might. It has become a de facto standard alongside vi. Emacs is more than just a text editor. It is a complete system of using a computer for development, communications, le management, and things you wouldnt even imagine there are programs for. There is even an Window System version available which can browse the web.

6.3.4

Other editors

Other editors to watch out for are joe, jed, nedit, pico, nano, and many others that try to emulate the look and feel of well-known DOS, Windows, or Apple Mac development environments, or to bring better interfaces by using Gtk/Gnome or Qt/KDE. The list gets longer each time I look. In short, dont think that the text editors that your vendor has chosen to put on your CD are the best or only free ones out there. The same goes for other applications.

7.2. Looping: the while and until Statements

7. Shell Scripting

echo "what is your name? (Type your name here and press Enter)" read NM echo "Hello $NM"

Chapter 7

computer to read and act on your list of commands, also called running the program.

Now, exit from your editor and type ./myfile.sh. This will execute Cause the the le. Note that typing ./myfile.sh is no different from typing any other command at the shell prompt. Your le myfile.sh has in fact become a new U NIX command all of its own. Note what the read command is doing. It creates a pigeonhole called NM, and then inserts text read from the keyboard into that pigeonhole. Thereafter, whenever the shell encounters NM, its contents are written out instead of the letters NM (provided you write a $ in front of it). We say that NM is a variable because its contents can vary.

Shell Scripting
This chapter introduces you to the concept of computer programming. So far, you have entered commands one at a time. Computer programming is merely the idea of getting a number of commands to be executed, that in combination do some unique powerful function.

echo echo read echo read echo

You can use shell scripts like a calculator. Try

"I will work out X*Y" "Enter X" X "Enter Y" Y "X*Y = $X*$Y = $[X*Y]"

7.1 Introduction
To execute a number of commands in sequence, create a le with a .sh extension, into which you will enter your commands. The .sh extension is not strictly necessary but serves as a reminder that the le contains special text called a shell script. From now on, the word script will be used to describe any sequence of commands placed in a text le. Now do a which allows the le to be run in the explained way. Edit the le using your favorite text editor. The rst line should be as follows Whitespace are tabs and spaces, and in some contexts, newline (end of line) with no whitespace.
characters.

The [ and ] mean that everything between must be evaluated Substituted, worked out, or as a numerical expression Sequence of numbers with +, -, *, etc. reduced to some simplied form. between them. . You can, in fact, do a calculation at any time by typing at the prompt

echo $[3*6+2*8+9]

use the expr command to get the same effect.

Note that the shell that you are using allows such [ ] notation. On some U

NIX

systems you will have to

chmod 0755 myfile.sh

7.2

Looping to Repeat Commands: the while and until Statements

The line dictates that the following program is a shell script, meaning that it accepts the same sort of commands that you have normally been typing at the prompt. Now enter a number of commands that you would like to be executed. You can start with
echo "Hi there"

#!/bin/sh

The shell reads each line in succession from top to bottom: this is called program ow. Now suppose you would like a command to be executed more than onceyou would like to alter the program ow so that the shell reads particular commands repeatedly. The while command executes a sequence of commands many times. Here is an example (-le stands for less than or equal):
N=1 while test "$N" -le "10" do

7. Shell Scripting

7.3. Looping: the for Statement

7. Shell Scripting

echo "Number $N" N=$[N+1]

X=10 Y=5 if test "$X" -gt "$Y" ; then echo "$X is greater than $Y" fi

The N=1 creates a variable called N and places the number 1 into it. The while command executes all the commands between the do and the done repetitively until the test condition is no longer true (i.e., until N is greater than 10). The -le stands for less than or equal to. See test(1) (that is, run man 1 test) to learn about the other types of tests you can do on variables. Also be aware of how N is replaced with a new value that becomes 1 greater with each repetition of the while loop. You should note here that each line is a distinct commandthe commands are newline-separated. You can also have more than one command on a line by separating them with a semicolon as follows: (Try counting down from 10 with -ge (greater than or equal).) It is easy to see that shell scripts are extremely powerful, because any kind of command can be executed with conditions and loops. The until statement is identical to while except that the reverse logic is applied. The same functionality can be achieved with -gt (greater than):
N=1 ; until test "$N" -gt "10"; do echo "Number $N"; N=$[N+1] ; done N=1 ; while test "$N" -le "10"; do echo "Number $N"; N=$[N+1] ; done

done

The if command in its full form can contain as much as:

X=10 Y=5 if test "$X" -gt "$Y" ; then echo "$X is greater than $Y" elif test "$X" -lt "$Y" ; then echo "$X is less than $Y" else echo "$X is equal to $Y" fi

Now let us create a script that interprets its arguments. Create a new script called backup-lots.sh, containing:
#!/bin/sh for i in 0 1 2 3 4 5 6 7 8 9 ; do cp $1 $1.BAK-$i done

7.3 Looping to Repeat Commands: the for Statement

The for command also allows execution of commands multiple times. It works like this:
for i in cows sheep chickens pigs do echo "$i is a farm animal" done echo -e "but\nGNUs are not farm animals"

Now create a le important data with anything in it and then run ./backuplots.sh important data, which will copy the le 10 times with 10 different extensions. As you can see, the variable $1 has a special meaningit is the rst argument on the command-line. Now lets get a little bit more sophisticated (-e test whether the le exists):
#!/bin/sh if test "$1" = "" ; then echo "Usage: backup-lots.sh <filename>" exit fi for i in 0 1 2 3 4 5 6 7 8 9 ; do NEW_FILE=$1.BAK-$i if test -e $NEW_FILE ; then echo "backup-lots.sh: **warning** $NEW_FILE" echo " already exists - skipping" else cp $1 $NEW_FILE

The for command takes each string after the in, and executes the lines between do and done with i substituted for that string. The strings can be anything (even numbers) but are often le names. The if command executes a number of commands if a condition is met (-gt stands for greater than, -lt stands for less than). The if command executes all the lines between the if and the fi (if spelled backwards). 63
10

7. Shell Scripting

7.4. breaking Out of Loops and continueing

7.5. Looping Over Glob Expressions

7. Shell Scripting

7.5

Looping Over Glob Expressions

done

We know that the shell can expand le names when given wildcards. For instance, we can type ls *.txt to list all les ending with .txt. This applies equally well in any situation, for instance:
#!/bin/sh for i in *.txt ; do echo "found a file:" $i done

7.4

breaking Out of Loops and continueing

A loop that requires premature termination can include the break statement within it:
#!/bin/sh for i in 0 1 2 3 4 5 6 7 8 9 ; do NEW_FILE=$1.BAK-$i if test -e $NEW_FILE ; then echo "backup-lots.sh: **error** $NEW_FILE" echo " already exists - exitting" break else cp $1 $NEW_FILE fi done

The *.txt is expanded to all matching les. These les are searched for in the current directory. If you include an absolute path then the shell will search in that directory:
#!/bin/sh for i in /usr/doc/*/*.txt ; do echo "found a file:" $i done

This example demonstrates the shells ability to search for matching les and expand an absolute path.

which causes program execution to continue on the line after the done. If two loops are nested within each other, then the command break 2 causes program execution to break out of both loops; and so on for values above 2. The continue statement is also useful for terminating the current iteration of the loop. This means that if a continue statement is encountered, execution will immediately continue from the top of the loop, thus ignoring the remainder of the body of the loop:
#!/bin/sh for i in 0 1 2 3 4 5 6 7 8 9 ; do NEW_FILE=$1.BAK-$i if test -e $NEW_FILE ; then echo "backup-lots.sh: **warning** $NEW_FILE" echo " already exists - skipping" continue fi cp $1 $NEW_FILE done
5

7.6

The case Statement

The case statement can make a potentially complicated program very short. It is best explained with an example.
#!/bin/sh case $1 in --test|-t) echo "you used the --test option" exit 0 ;; --help|-h) echo "Usage:" echo " myprog.sh [--test|--help|--version]" exit 0 ;; --version|-v) echo "myprog.sh version 0.0.1" exit 0 ;; -*) echo "No such option $1" echo "Usage:"

Note that both break and continue work inside for, while, and until loops. 65

7. Shell Scripting

7.7. Using Functions: the function Keyword

7.8. Properly Processing Command-Line Args: shift

7. Shell Scripting

echo " exit 1 ;; esac

myprog.sh [--test|--help|--version]"

echo "Error: no such option $1" usage exit 1

;; esac

echo "You typed \"$1\" on the command-line"

Above you can see that we are trying to process the rst argument to a program. It can be one of several options, so using if statements will result in a long program. The case statement allows us to specify several possible statement blocks depending on the value of a variable. Note how each statement block is separated by ;;. The strings before the ) are glob expression matches. The rst successful match causes that block to be executed. The | symbol enables us to enter several possible glob expressions.

echo "You typed \"$1\" on the command-line"

Wherever the usage keyword appears, it is effectively substituted for the two lines inside the { and }. There are obvious advantages to this approach: if you would like to change the program usage description, you only need to change it in one place in the code. Good programs use functions so liberally that they never have more than 50 lines of program code in a row.

7.7 Using Functions: the function Keyword

So far, our programs execute mostly from top to bottom. Often, code needs to be repeated, but it is considered bad programming practice to repeat groups of statements that have the same functionality. Function denitions provide a way to group statement blocks into one. A function groups a list of commands and assigns it a name. For example:
#!/bin/sh function usage () { echo "Usage:" echo " myprog.sh [--test|--help|--version]" } case $1 in --test|-t) echo "you used the --test option" exit 0 ;; --help|-h) usage ;; --version|-v) echo "myprog.sh version 0.0.2" exit 0 ;; -*)

7.8

Properly Processing Command-Line Arguments: the shift Keyword

Most programs we have seen can take many command-line arguments, sometimes in any order. Here is how we can make our own shell scripts with this functionality. The command-line arguments can be reached with $1, $2, etc. The script,
#!/bin/sh

can be run with and prints

echo "The first argument is: $1, second argument is: $2, third argument is: $3"

myfile.sh dogs cats birds

The first argument is: dogs, second argument is: cats, third argument is: birds

Now we need to loop through each argument and decide what to do with it. A script like doesnt give us much exibilty. The shift keyword is meant to make things easier. It shifts up all the arguments by one place so that $1 gets the value of $2, $2 gets the value of $3, and so on. (!= tests that the "$1" is not equal to "", that is, whether it is empty and is hence past the last argument.) Try 68
for i in $1 $2 $3 $4 ; do <statments> done

7. Shell Scripting

7.8. Properly Processing Command-Line Args: shift

7.9. More on Command-Line Arguments: $@ and $0

7. Shell Scripting

while test "$1" != "" ; do echo $1 shift done

7.9

and run the program with lots of arguments.

Whereas $1, $2, $3, etc. expand to the individual arguments passed to the program, $@ expands to all arguments. This behavior is useful for passing all remaining arguments onto a second command. For instance,
if test "$1" = "--special" ; then shift myprog2.sh "$@" fi

Now we can put any sort of condition statements within the loop to process the arguments in turn:
#!/bin/sh function usage () { echo "Usage:" echo " myprog.sh [--test|--help|--version] [--echo <text>]" } while test "$1" != "" ; do case $1 in --echo|-e) echo "$2" shift ;; --test|-t) echo "you used the --test option" ;; --help|-h) usage exit 0 ;; --version|-v) echo "myprog.sh version 0.0.3" exit 0 ;; -*) echo "Error: no such option $1" usage exit 1 ;; esac shift done

$0 means the name of the program itself and not any command-line argument. It is the command used to invoke the current program. In the above cases, it is ./myprog.sh. Note that $0 is immune to shift operations.

7.10

Single Forward Quote Notation

Single forward quotes protect the enclosed text from the shell. In other words, you can place any odd characters inside forward quotes, and the shell will treat them literally and reproduce your text exactly. For instance, you may want to echo an actual $ to the screen to produce an output like costs $1000. You can use echo costs $1000 instead of echo "costs $1000".

7.11 Double-Quote Notation

Double quotes " have the opposite sense of single quotes. They allow all shell interpretations to take place inside them. The reason they are used at all is only to group text containing whitespace into a single word, because the shell will usually break up text along whitespace boundaries. Try,
for i in "henry john mary sue" ; do echo "$i is a person" done

compared to
for i in henry john mary sue ; do echo $i is a person done

myprog.sh can now run with multiple arguments on the command-line. 69

7. Shell Scripting

7.12. Backward-Quote Substitution

7. Shell Scripting

7.12 Backward-Quote Substitution

Backward quotes have a special meaning to the shell. When a command is inside backward quotes it means that the command should be run and its output substituted in place of the backquotes. Take, for example, the cat command. Create a small le, to be catted, with only the text daisy inside it. Create a shell script
X=cat to_be_catted echo $X

The value of X is set to the output of the cat command, which in this case is the word daisy. This is a powerful tool. Consider the expr command:
X=expr 100 + 50 * 3 echo $X

Hence we can use expr and backquotes to do mathematics inside our shell script. Here is a function to calculate factorials. Note how we enclose the * in forward quotes. They prevent the shell from expanding the * into matching le names:
function factorial () { N=$1 A=1 while test $N -gt 0 ; do A=expr $A * $N N=expr $N - 1 done echo $A }

We can see that the square braces used further above can actually sufce for most of the times where we would like to use expr. (However, $[] notation is an extension of the GNU shells and is not a standard feature on all varients of U NIX.) We can now run factorial 20 and see the output. If we want to assign the output to a variable, we can do this with X=factorial 20. Note that another notation which gives the effect of a backward quote is $(command), which is identical to command. Here, I will always use the older backward quote style.

8.2. Tutorial

8. Streams and sed The Stream Editor

8.2

Tutorial

Chapter 8

Create a text le with lots of lines that contain the word GNU and one line that contains the word GNU as well as the word Linux. Then run grep GNU myfile.txt. The result is printed to stdout as usual. Now try grep GNU myfile.txt > gnu lines.txt. What is happening here is that the output of the grep command is being redirected into a le. The > gnu lines.txt tells the shell to create a new le gnu lines.txt and to ll it with any output from stdout instead of displaying the output as it usually does. If the le already exists, it will be truncated.

Streams and sed The Stream Editor

The ability to use pipes is one of the powers of U NIX. This is one of the principle deciencies of some non-U NIX systems. Pipes used on the command-line as explained in this chapter are a neat trick, but pipes used inside programs enormously simplify program interaction. Without pipes, huge amounts of complex and buggy code usually needs to be written to perform simple tasks. It is hoped that this chapter will give the reader an idea of why U NIX is such a ubiquitous and enduring standard.

Shortened to zero length.

Now suppose you want to append further output to this le. Using >> instead of > does not truncate the le, but appends output to it. Try then view the contents of gnu lines.txt.
echo "morestuff" >> gnu_lines.txt

8.3

Piping Using | Notation

The real power of pipes is realized when one program can read from the output of another program. Consider the grep command, which reads from stdin when given no arguments; run grep with one argument on the command-line:
[root@cericon]# grep GNU A line without that word in it Another line without that word in it A line with the word GNU in it A line with the word GNU in it I have the idea now C #

8.1 Introduction
The commands grep, echo, df and so on print some output to the screen. In fact, what is happening on a lower level is that they are printing characters one by one into a theoretical data stream (also called a pipe) called the stdout pipe. The shell itself performs the action of reading those characters one by one and displaying them on the screen. The word pipe itself means exactly that: A program places data in the one end of a funnel while another program reads that data from the other end. Pipes allow two separate programs to perform simple communications with each other. In this case, the program is merely communicating with the shell in order to display some output. The same is true with the cat command explained previously. This command, when run with no arguments, reads from the stdin pipe. By default, this pipe is the keyboard. One further pipe is the stderr pipe to which a program writes error messages. It is not possible to see whether a program message is caused by the program writing to its stderr or stdout pipe because usually both are directed to the screen. Good programs, however, always write to the appropriate pipes to allow output to be specially separated for diagnostic purposes if need be. 73
5

greps default behavior is to read from stdin when no les are given. As you can see, it is doing its usual work of printing lines that have the word GNU in them. Hence, lines containing GNU will be printed twiceas you type them in and again when grep reads them and decides that they contain GNU. Now try grep GNU myfile.txt | grep Linux. The rst grep outputs all lines with the word GNU in them to stdout. The | species that all stdout is to be typed as stdin (as we just did above) into the next command, which is also a grep command. The second grep command scans that data for lines with the word Linux in them. grep is often used this way as a lter Something that screens data. and can be used multiple times, for example,

8. Streams and sed The Stream Editor

8.4. A Complex Piping Example

8.5. Redirecting Streams with >&

8. Streams and sed The Stream Editor

grep L myfile.txt | grep i | grep n | grep u | grep x

touch existing_file rm -f non-existing_file ls existing_file non-existing_file

The < character redirects the contents of a le in place of stdin. In other words, the contents of a le replace what would normally come from a keyboard. Try
grep GNU < gnu_lines.txt

8.4 A Complex Piping Example

In Chapter 5 we used grep on a dictionary to demonstrate regular expressions. This is how a dictionary of words can be created (your dictionary might be under /var/share/ or under /usr/lib/aspell instead):
cat /usr/lib/ispell/english.hash | strings | tr A-Z a-z \ | grep [a-z] | sort -u > mydict

ls will output two lines: a line containing a listing for the le existing file and a line containing an error message to explain that the le non-existing file does not exist. The error message would have been written to stderr or le descriptor number 2, and the remaining line would have been written to stdout or le descriptor number 1. Next we try

ls existing_file non-existing_file 2>A cat A

the \ but then you must leave out the newline as well this is known as line continuation.

A backslash \ as the last character on a line indicates that the line is to be continued. You can leave out

Now A contains the error message, while the remaining output came to the screen. Now try
ls existing_file non-existing_file 1>A cat A

The le english.hash contains the U NIX dictionary normally used for spell checking. With a bit of ltering, you can create a dictionary that will make solving crossword puzzles a breeze. First, we use the command strings, explained previously, to extract readable bits of text. Here we are using its alternate mode of operation where it reads from stdin when no les are specied on its command-line. The command tr (abbreviated from translatesee tr(1)) then converts upper to lower case. The grep command then lters out lines that do not start with a letter. Finally, the sort command sorts the words in alphabetical order. The -u option stands for unique, and species that duplicate lines of text should be stripped. Now try less mydict.

The notation 1>A is the same as >A because the shell assumes that you are referring to le descriptor 1 when you dont specify a le descriptor. Now A contains the stdout output, while the error message has been redirected to the screen. Now try

ls existing_file non-existing_file 1>A 2>&1 cat A

8.5 Redirecting Streams with >&

Try the command ls nofile.txt > A. We expect that ls will give an error message if the le doesnt exist. The error message is, however, displayed and not written into the le A. The reason is that ls has written its error message to stderr while > has only redirected stdout. The way to get both stdout and stderr to both go to the same le is to use a redirection operator. As far as the shell is concerned, stdout is called 1 and stderr is called 2, and commands can be appended with a redirection like 2>&1 to dictate that stderr is to be mixed into the output of stdout. The actual words stderr and stdout are only used in programming, where the number 1, 2 are known as le numbers or le descriptors. Try the following: 75

Now A contains both the error message and the normal output. The >& is called a redirection operator. x>&y tells the shell to write pipe x into pipe y. Redirection is specied from right to left on the command-line. Hence, the above command means to mix stderr into stdout and then to redirect stdout to the le A. Finally,

ls existing_file non-existing_file 2>A 1>&2 cat A

We notice that this has the same effect, except that here we are doing the reverse: redirecting stdout into stderr and then redirecting stderr into a le A. To see what happens if we redirect in reverse order, we can try, 76

8. Streams and sed The Stream Editor

8.6. Using sed to Edit Streams

8.7. Regular Expression Subexpressions

8. Streams and sed The Stream Editor

ls existing_file non-existing_file 2>&1 1>A cat A

sed -e s/$\<[ ]*\>$$[ ]*$$\<[ ]*\>$/\3\2\1/g

which means to redirect stdout into a le A, and then to redirect stderr into stdout. This command will therefore not mix stderr and stdout because the redirection to A came rst.

8.6 Using sed to Edit Streams

ed used to be the standard text editor for U NIX. It is cryptic to use but is compact and programmable. sed stands for stream editor and is the only incarnation of ed that is commonly used today. sed allows editing of les non-interactively. In the way that grep can search for words and lter lines of text, sed can do search-replace operations and insert and delete lines into text les. sed is one of those programs with no man page to speak of. Do info sed to see seds comprehensive info pages with examples. The most common usage of sed is to replace words in a stream with alternative words. sed reads from stdin and writes to stdout. Like grep, it is line buffered, which means that it reads one line in at a time and then writes that line out again after performing whatever editing operations. Replacements are typically done with where <search-regexp> is a regular expression, <replace-text> is the text you would like to replace each occurrence with, and <option> is nothing or g, which means to replace every occurrence in the same line (usually sed just replaces the rst occurrence of the regular expression in each line). (There are other <option>; see the sed info page.) For demonstration, type and type out a few lines of English text.
sed -e s/e/E/g cat <file> | sed -e s/<search-regexp>/<replace-text>/<option> \ > <resultfile>

Here sed is searching for the expression \<.*\>[ ]*\<.*\>. From the chapter on regular expressions, we can see that it matches a whole word, an arbitrary amount of whitespace, and then another whole word. The  groups these three so that they can be referred to in <replace-text>. Each part of the regular expression inside  is called a subexpression of the regular expression. Each subexpression is numberednamely, \1, \2, etc. Hence, \1 in <replace-text> is the rst \<[ ]*\>, \2 is [ ]*, and \3 is the second \<[ ]*\>. Now test to see what happens when you run this:

sed -e s/$\<[ ]*\>$$[ ]*$$\<[ ]*\>$/\3\2\1/g GNU Linux is cool Linux GNU cool is

To return to our ls example (note that this is just an example, to count le sizes you should instead use the du command), think about how we could sum the bytes sizes of all the les in a directory:
expr 0 ls -l | grep - | \ sed s/$[ ]*[ ]*$\{4,4\}$[0-9]*$.*$/ + \2/

We know that ls -l output lines start with - for ordinary les. So we use grep to strip lines not starting with -. If we do an ls -l, we see that the output is divided into four columns of stuff we are not interested in, and then a number indicating the size of the le. A column (or eld) can be described by the regular expression [ ]*[ ]*, that is, a length of text with no whitespace, followed by a length of whitespace. There are four of these, so we bracket it with  and then use the \{ \} notation to specify that we want exactly 4. After that come our number [0-9]*, and then any trailing characters, which we are not interested in, .*$. Notice here that we have neglected to use \< \> notation to indicate whole words. The reason is that sed tries to match the maximum number of characters legally allowed and, in the situation we have here, has exactly the same effect. If you havent yet gured it out, we are trying to get that column of byte sizes into a format like
+ + + + 438 1525 76 92146

8.7 Regular Expression Subexpressions

The section explains how to do the apparently complex task of moving text around within lines. Consider, for example, the output of ls: say you want to automatically strip out only the size columnsed can do this sort of editing if you use the special  notation to group parts of the regular expression together. Consider the following example: 77

so that expr can understand it. Hence, we replace each line with subexpression \2 and a leading + sign. Backquotes give the output of this to expr, which studiously sums 78

8. Streams and sed The Stream Editor

8.8. Inserting and Deleting Lines

8. Streams and sed The Stream Editor

them, ignoring any newline characters as though the summation were typed in on a single line. There is one minor problem here: the rst line contains a + with nothing before it, which will cause expr to complain. To get around this, we can just add a 0 to the expression, so that it becomes 0 + . . . .

The new last line.

and nally, the negation symbol, !, is used to match all lines not specied; for instance,
sed -e 7,11!D

8.8 Inserting and Deleting Lines

sed can perform a few operations that make it easy to write scripts that edit conguration les for you. For instance,
sed -e 7a\ an extra line.\ another one.\ one more.

deletes all lines except lines 7 through 11.

appends three lines after line 7, whereas

sed -e 7i\ an extra line.\ another one.\ one more.

inserts three lines before line 7. Then

sed -e 3,5D

Deletes lines 3 through 5. In sed terminology, the numbers here are called addresses, which can also be regular expressions matches. To demonstrate:
sed -e /Dear Henry/,/Love Jane/D

deletes all the lines starting from a line matching the regular expression Dear Henry up to a line matching Love Jane (or the end of the le if one does not exist). This behavior applies just as well to to insertions:

sed -e /Love Jane/i\ Love Carol\ Love Beth

Note that the $ symbol indicates the last line:

sed -e $i\ The new second last line\

9.2. ps List Running Processes

9. Processes, Environment Variables

The center of this operation is called the U NIX kernel. The kernel is what actually does the hardware access, execution, allocation of process IDs, sharing of CPU time, and ownership management.

Chapter 9

9.2

ps List Running Processes

Processes and Environment Variables

From this chapter you will get an idea about what is happening under the hood of your U NIX system, but go have some coffee rst.

Log in on a terminal and type the command ps. You should get some output like:
PID TTY STAT TIME COMMAND 5995 2 S 0:00 /bin/login -- myname 5999 2 S 0:00 -bash 6030 2 R 0:00 ps

9.1 Introduction
On U NIX, when you run a program (like any of the shell commands you have been using), the actual computer instructions are read from a le on disk from one of the bin/ directories and placed in RAM. The program is then executed in memory and becomes a process. A process is some command/program/shell-script that is being run (or executed) in memory. When the process has nished running, it is removed from memory. There are usually about 50 processes running simultaneously at any one time on a system with one person logged in. The CPU hops between each of them to give a share of its execution time. Time given to carry out the instructions of a particular program. Note this
is in contrast to Windows or DOS where the program itself has to allow the others a share of the CPU: under U NIX, the process has no say in the matter. Each process is given a process number called the PID (process ID). Besides the memory actually occupied by the executable, the process itself seizes additional memory for its operations.

ps with no options shows three processes to be running. These are the only three processes visible to you as a user, although there are other system processes not belonging to you. The rst process was the program that logged you in by displaying the login prompt and requesting a password. It then ran a second process call bash, the Bourne Again shell The Bourne shell was the original U NIX shell where you have been typing commands. Finally, you ran ps, which must have found itself when it checked which processes were running, but then exited immediately afterward.

9.3

Controlling Jobs

The shell has many facilities for controlling and executing processesthis is called job control. Create a small script called proc.sh:
#!/bin/sh echo "proc.sh: is running" sleep 1000

In the same way that a le is owned by a particular user and group, a process also has an ownerusually the person who ran the program. Whenever a process tries to access a le, its ownership is compared to that of the le to decide if the access is permissible. Because all devices are les, the only way a process can do anything is through a le, and hence le permission restrictions are the only kind of restrictions ever needed on U NIX. There are some exceptions to this. This is how U NIX access control and security works.

Run the script with chmod 0755 proc.sh and then ./proc.sh. The shell blocks, waiting for the process to exit. Now press Z. This will cause the process to stop (that is, pause but not terminate). Now do a ps again. You will see your script listed. However, it is not presently running because it is in the condition of being stopped. Type bg (for background). The script will now be unstopped and run in the background. You can now try to run other processes in the meantime. Type fg, and the script returns to the foreground. You can then type C to interrupt the process. 82

9. Processes, Environment Variables

9.4. Creating Background Processes

9.5. killing a Process, Sending Signals

9. Processes, Environment Variables

9.4 Creating Background Processes

Create a program that does something a little more interesting:
#!/bin/sh echo "proc.sh: is running" while true ; do echo -e \a sleep 2 done

signals. Only foreground processes are allowed to read from or write to the terminal. Background processes which attempt to read from (write to) the terminal are sent a SIGTTIN (SIGTTOU) signal by the terminal driver, which, unless caught, suspends the process. If the operating system on which bash is running supports job control, bash allows you to use it. Typing the suspend character (typically Z, Control-Z) while a process is running causes that process to be stopped and returns you to bash. Typing the delayed suspend character (typically Y, Control-Y) causes the process to be stopped when it attempts to read input from the terminal, and control to be returned to bash. You may then manipulate the state of this job, using the bg command to continue it in the background, the fg command to continue it in the foreground, or the kill command to kill it. A Z takes effect immediately, and has the additional side effect of causing pending output and typeahead to be discarded. There are a number of ways to refer to a job in the shell. The character % introduces a job name. Job number n may be referred to as %n. A job may also be referred to using a prex of the name used to start it, or using a substring that appears in its command line. For example, %ce refers to a stopped ce job. If a prex matches more than one job, bash reports an error. Using %?ce, on the other hand, refers to any job containing the string ce in its command line. If the substring matches more than one job, bash reports an error. The symbols %% and %+ refer to the shells notion of the current job, which is the last job stopped while it was in the foreground. The previous job may be referenced using %-. In output pertaining to jobs (e.g., the output of the jobs command), the current job is always agged with a +, and the previous job with a -. Simply naming a job can be used to bring it into the foreground: %1 is a synonym for fg %1, bringing job 1 from the background into the foreground. Similarly, %1 & resumes job 1 in the background, equivalent to bg %1. The shell learns immediately whenever a job changes state. Normally, bash waits until it is about to print a prompt before reporting changes in a jobs status so as to not interrupt any other output. If the -b option to the set builtin command is set, bash reports such changes immediately. (See also the description of notify variable under Shell Variables above.) If you attempt to exit bash while jobs are stopped, the shell prints a message warning you. You may then use the jobs command to inspect their status. If you do this, or try to exit again immediately, you are not warned again, and the stopped jobs are terminated.

Now perform the Z, bg, fg, and C operations from before. To put a process immediately into the background, you can use:
./proc.sh &

The JOB CONTROL section of the bash man page (bash(1)) looks like this : (the footnotes are mine)
JOB CONTROL Job control refers to the ability to selectively stop (suspend) the execution of processes and continue (resume) their execution at a later point. A user typically employs this facility via an interactive interface supplied jointly by the systems terminal driver and bash.
What does this mean? It means The shell associates a job with each pipeline. that each time you execute something in the background, it gets its own unique number, called the job number. It keeps a table of currently executing jobs, which may be

listed with the jobs command. When bash starts a job asynchronously (in the background), it prints a line that looks like: [1] 25647 indicating that this job is job number 1 and that the process ID of the last process in the pipeline associated with this job is 25647. All of the processes in a single pipeline are members of the same job. Bash uses the job abstraction as the basis for job control. To facilitate the implementation of the user interface to job control, the system maintains the notion of a current terminal process group ID. Members of this process group (processes whose process group ID is equal to the current terminal process group ID) receive keyboard-generated signals such as SIGINT. These processes are said to be in the foreground. Background processes are those whose process group ID differs from the terminals; such processes are immune to keyboard-generated
1 Thanks

9.5

killing a Process, Sending Signals

to Brian Fox and Chet Ramey for this material.

To terminate a process, use the kill command: 84

9. Processes, Environment Variables

9.5. killing a Process, Sending Signals

9.6. List of Common Signals

9. Processes, Environment Variables

kill <PID>

9.6

List of Common Signals

and in the le

The kill command actually sends a termination signal to the process. The sending of a signal simply means that the process is asked to execute one of 30 predened functions. In some cases, developers would not have bothered to dene a function for a particular signal number (called catching the signal); in which case the kernel will substitute the default behavior for that signal. The default behavior for a signal is usually to ignore the signal, to stop the process, or to terminate the process. The default behavior for the termination signal is to terminate the process. To send a specic signal to a process, you can name the signal on the commandline or use its numerical equivalent:
kill -SIGTERM 12345

The full list of signals can be gotten from signal(7), /usr/include/asm/signal.h.

SIGHUP (1) Hang up. If the terminal becomes disconnected from a process, this signal is sent automatically to the process. Sending a process this signal often causes it to reread its conguration les, so it is useful instead of restarting the process. Always check the man page to see if a process has this behavior. SIGINT (2) Interrupt from keyboard. Issued if you press C. SIGQUIT (3) Quit from keyboard. Issued if you press D. SIGFPE (8) Floating point exception. Issued automatically to a program performing some kind of illegal mathematical operation. SIGKILL (9) Kill signal. This is one of the signals that can never be caught by a process. If a process gets this signal it must quit immediately and will not perform any clean-up operations (like closing les or removing temporary les). You can send a process a SIGKILL signal if there is no other means of destroying it. SIGUSR1 (10), SIGUSR2 (12) User signal. These signals are available to developers when they need extra functionality. For example, some processes begin logging debug messages when you send them SIGUSR1. SIGSEGV (11) Segmentation violation. Issued automatically when a process tries to access memory outside of its allowable address space, equivalent to a Fatal Exception or General Protection Fault under Windows. Note that programs with bugs or programs in the process of being developed often get these signals. A program receiving a SIGSEGV, however, can never cause the rest of the system to be compromised. If the kernel itself were to receive such an error, it would cause the system to come down, but such is extremely rare. SIGPIPE (13) Pipe died. A program was writing to a pipe, the other end of which is no longer available. SIGTERM (15) Terminate. Cause the program to quit gracefully SIGCHLD (17) Child terminate. Sent to a parent process every time one of its spawned processes dies. 86

or
kill -15 12345

which is the signal that kill normally sends when none is specied on the commandline. To unconditionally terminate a process:
kill -SIGKILL 12345

or
kill -9 12345

which should only be used as a last resort. Processes are prohibited from ever catching the SIGKILL signal. It is cumbersome to have to constantly look up the PID of a process. Hence the GNU utilities have a command, killall, that sends a signal to all processes of the same name:
killall -<signal> <process_name>

This command is useful when you are sure that there is only one of a process running, either because no one else is logged in on the system or because you are not logged in as superuser. Note that on other U NIX systems, the killall command kills all the processes that you are allowed to kill. If you are root, this action would crash the machine. 85

9. Processes, Environment Variables

9.7. Niceness of Processes, Scheduling Priority

9.8. Process CPU/Memory Consumption, top

9. Processes, Environment Variables

9.7 Niceness of Processes, Scheduling Priority

All processes are allocated execution time by the kernel. If all processes were allocated the same amount of time, performance would obviously get worse as the number of processes increased. The kernel uses heuristics Sets of rules. to guess how much time each process should be allocated. The kernel tries to be fairtwo users competing for CPU usage should both get the same amount.

time of a process, as well as the time it takes for a process to wake). There are already some kernel patches that accomplish this goal.

Also useful are the -u and -g options, which set the priority of all the processes that a user or group owns. Further, we have the nice command, which starts a program under a dened niceness relative to the current nice value of the present user. For example,
nice +<priority> <pid> nice -<priority> <pid>

Most processes spend their time waiting for either a key press, some network input, some device to send data, or some time to elapse. They hence do not consume CPU. On the other hand, when more than one process runs at out, it can be difcult for the kernel to decide if it should be given greater priority than another process. What if a process is doing some operation more important than another process? How does the kernel tell? The answer is the U NIX feature of scheduling priority or niceness. Scheduling priority ranges from +20 to -20. You can set a processs niceness with the renice command.
renice <priority> <pid> renice <priority> -u <user> renice <priority> -g <group>

Finally, the snice command can both display and set the current niceness. This command doesnt seem to work on my machine.
snice -v <pid>

9.8

Process CPU/Memory Consumption, top

A typical example is the SETI program.

telligence. SETI is an initiative funded by various obscure sources to scan the skies for radio signals from other civilizations. The data that SETI gathers has to be intensively processed. SETI distributes part of that data to anyone who wants to run a seti program in the background. This puts the idle time of millions of machines to good use. There is even a SETI screen-saver that has become quite popular. Unfortunately for the colleague in my ofce, he runs seti at -19 instead of +19 scheduling priority, so nothing on his machine works right. On the other hand, I have inside information that the millions of other civilizations in this galaxy and others are probably not using radio signals to communicate at all :-)

SETI stands for Search for Extraterrestrial In-

The top command sorts all processes by their CPU and memory consumption and displays the top twenty or so in a table. Use top whenever you want to see whats hogging your system. top -q -d 2 is useful for scheduling the top command itself to a high priority, so that it is sure to refresh its listing without lag. top -n 1 -b > top.txt lists all processes, and top -n 1 -b -p <pid> prints information on one process. top has some useful interactive responses to key presses: f Shows a list of displayed elds that you can alter interactively. By default the only elds shown are USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND which is usually what you are most interested in. (The eld meanings are given below.) r Renices a process. k Kills a process. The top man page describes the eld meanings. Some of these are confusing and assume knowledge of the internals of programs. The main question people ask is: How much memory is a process using? The answer is given by the RSS eld, which stands for Resident Set Size. RSS means the amount of RAM that a process consumes alone. The following examples show totals for all processes running on my system (which had 65536 kilobytes of RAM at the time). They represent the total of the SIZE, RSS, and SHARE elds, respectively. 88

+19 with:
renice +19 <pid>

Set its priority to

to make it disrupt your machine as little as possible. Note that nice values have the reverse meaning that you would expect: +19 means a process that eats little CPU, while -19 is a process that eats lots. Only superuser can set processes to negative nice values. Mostly, multimedia applications and some device utilities are the only processes that need negative renicing, and most of these will have their own command-line options to set the nice value. See, for example, cdrecord(1) and mikmod(1) a negative nice value will prevent skips in your playback. L INUX will soon have so called real time process scheduling. This is a kernel feature that reduces scheduling latency (the gaps between CPU execution

9. Processes, Environment Variables

9.8. Process CPU/Memory Consumption, top

9.9. Environments of Processes

9. Processes, Environment Variables

echo echo 0 ; top -q -n 1 -b | sed -e 1,/PID *USER *PRI/D | \ awk {print "+" $5} | sed -e s/M/\\*1024/ | bc 68016
5

TSIZE The code size of the task. This gives strange values for kernel processes and is broken for ELF processes. DSIZE Data + Stack size. This is broken for ELF processes. TRS Text resident size. SWAP Size of the swapped out part of the task. D Size of pages marked dirty. LIB Size of use library pages. This does not work for ELF processes. RSS The total amount of physical memory used by the task, in kilobytes, is shown here. For ELF processes used library pages are counted here, for a.out processes not. SHARE The amount of shared memory used by the task is shown in this column. STAT The state of the task is shown here. The state is either S for sleeping, D for uninterruptible sleep, R for running, Z for zombies, or T for stopped or traced. These states are modied by a trailing for a process with negative nice value, N for a process with positive nice value, W for a swapped out process (this does not work correctly for kernel processes). WCHAN depending on the availability of either /boot/psdatabase or the kernel link map /boot/System.map this shows the address or the name of the kernel function the task currently is sleeping in. TIME Total CPU time the task has used since it started. If cumulative mode is on, this also includes the CPU time used by the processs children which have died. You can set cumulative mode with the S command line option or toggle it with the interactive command S. The header line will then be changed to CTIME. %CPU The tasks share of the CPU time since the last screen update, expressed as a percentage of total CPU time per processor. %MEM The tasks share of the physical memory. COMMAND The tasks command name, which will be truncated if it is too long to be displayed on one line. Tasks in memory will have a full command line, but swapped-out tasks will only have the name of the program in parentheses (for example, (getty)).

echo echo 0 ; top -q -n 1 -b | sed -e 1,/PID *USER *PRI/D | \ awk {print "+" $6} | sed -e s/M/\\*1024/ | bc 58908 echo echo 0 ; top -q -n 1 -b | sed -e 1,/PID *USER *PRI/D | \ awk {print "+" $7} | sed -e s/M/\\*1024/ | bc 30184

The SIZE represents the total memory usage of a process. RSS is the same, but excludes memory not needing actual RAM (this would be memory swapped to the swap partition). SHARE is the amount shared between processes. Other elds are described by the top man page (quoted verbatim) as follows:
uptime This line displays the time the system has been up, and the three load averages for the system. The load averages are the average number of processes ready to run during the last 1, 5 and 15 minutes. This line is just like the output of uptime(1). The uptime display may be toggled by the interactive l command. processes The total number of processes running at the time of the last update. This is also broken down into the number of tasks which are running, sleeping, stopped, or undead. The processes and states display may be toggled by the t interactive command. CPU states Shows the percentage of CPU time in user mode, system mode, niced tasks, and idle. (Niced tasks are only those whose nice value is negative.) Time spent in niced tasks will also be counted in system and user time, so the total will be more than 100%. The processes and states display may be toggled by the t interactive command. Mem Statistics on memory usage, including total available memory, free memory, used memory, shared memory, and memory used for buffers. The display of memory information may be toggled by the m interactive command. Swap Statistics on swap space, including total swap space, available swap space, and used swap space. This and Mem are just like the output of free(1). PID The process ID of each task. PPID The parent process ID of each task. UID The user ID of the tasks owner. USER The user name of the tasks owner. PRI The priority of the task. NI The nice value of the task. Negative nice values are higher priority. SIZE The size of the tasks code plus data plus stack space, in kilobytes, is shown here.

9.9

Environments of Processes

Each process that runs does so with the knowledge of several var=value text pairs. All this means is that a process can look up the value of some variable that it may have inherited from its parent process. The complete list of these text pairs is called the environment of the process, and each var is called an environment variable. Each process has its own environment, which is copied from the parent processs environment. After you have logged in and have a shell prompt, the process you are using (the shell itself) is just like any other process with an environment with environment variables. To get a complete list of these variables, just type: 90

9. Processes, Environment Variables

9.9. Environments of Processes

9. Processes, Environment Variables

set

This command is useful for nding the value of an environment variable whose name you are unsure of:
set | grep <regexp>

man page lists these (when it talks about unsetting a variable, it means using the command unset <variable>). You may not understand some of these at the moment, but they are included here as a complete reference for later. The following is quoted verbatim from the bash man page. You will see that some variables are of the type that provide special information and are read but never never set, whereas other variables congure behavioral features of the shell (or other programs) and can be set at any time2 .
Shell Variables The following variables are set by the shell: PPID The process ID of the shells parent. PWD The current working directory as set by the cd command. OLDPWD The previous working directory as set by the cd command. REPLY Set to the line of input read by the read builtin command when no arguments are supplied. UID Expands to the user ID of the current user, initialized at shell startup. EUID Expands to the effective user ID of the current user, initialized at shell startup. BASH Expands to the full pathname used to invoke this instance of bash. BASH VERSION Expands to the version number of this instance of bash. SHLVL Incremented by one each time an instance of bash is started. RANDOM Each time this parameter is referenced, a random integer is generated. The sequence of random numbers may be initialized by assigning a value to RANDOM. If RANDOM is unset, it loses its special properties, even if it is subsequently reset. SECONDS Each time this parameter is referenced, the number of seconds since shell invocation is returned. If a value is assigned to SECONDS. the value returned upon subsequent references is the number of seconds since the assignment plus the value assigned. If SECONDS is unset, it loses its special properties, even if it is subsequently reset. LINENO Each time this parameter is referenced, the shell substitutes a decimal number representing the current sequential line number (starting with 1) within a script or function. When not in a script or function, the value substituted is not guaranteed to be meaningful. When in a function, the value is not the number of the source line that the command appears on (that information has been lost by the time the function is executed), but is an approximation of the number of simple commands executed in the current function. If LINENO is unset, it loses its special properties, even if it is subsequently reset. HISTCMD The history number, or index in the history list, of the current command. If HISTCMD is unset, it loses its special properties, even if it is subsequently reset.
2 Thanks

Try set | grep PATH to see the PATH environment variable discussed previously. The purpose of an environment is just to have an alternative way of passing parameters to a program (in addition to command-line arguments). The difference is that an environment is inherited from one process to the next: for example, a shell might have a certain variable set and may run a le manager, which may run a word processor. The word processor inherited its environment from le manager which inherited its environment from the shell. If you had set an environment variable PRINTER within the shell, it would have been inherited all the way to the word processor, thus eliminating the need to separately congure which printer the word processor should use. Try

X="Hi there" echo $X

You have set a variable. But now run

bash

You have now run a new process which is a child of the process you were just in. Type
echo $X

You will see that X is not set. The reason is that the variable was not exported as an environment variable and hence was not inherited. Now type
exit

which breaks to the parent process. Then

export X bash echo $X

You will see that the new bash now knows about X. Above we are setting an arbitrary variable for our own use. bash (and many other programs) automatically set many of their own environment variables. The bash 91

to Brian Fox and Chet Ramey for this material.

9. Processes, Environment Variables

9.9. Environments of Processes

9. Processes, Environment Variables

OPTARG The value of the last option argument processed by the getopts builtin command (see SHELL BUILTIN COMMANDS below). OPTIND The index of the next argument to be processed by the getopts builtin command (see SHELL BUILTIN COMMANDS below). HOSTTYPE Automatically set to a string that uniquely describes the type of machine on which bash is executing. The default is system-dependent. OSTYPE Automatically set to a string that describes the operating system on which bash is executing. The default is system-dependent. The following variables are used by the shell. In some cases, bash assigns a default value to a variable; these cases are noted below. IFS The Internal Field Separator that is used for word splitting after expansion and to split lines into words with the read builtin command. The default value is <space><tab><newline>. PATH The search path for commands. It is a colon-separated list of directories in which the shell looks for commands (see COMMAND EXECUTION below). The default path is system-dependent, and is set by the administrator who installs bash. A common value is /usr/gnu/bin:/usr/local/bin:/usr/ucb:/bin:/usr/bin:.. HOME The home directory of the current user; the default argument for the cd builtin command. CDPATH The search path for the cd command. This is a colon-separated list of directories in which the shell looks for destination directories specied by the cd command. A sample value is .::/usr. ENV If this parameter is set when bash is executing a shell script, its value is interpreted as a lename containing commands to initialize the shell, as in .bashrc. The value of ENV is subjected to parameter expansion, command substitution, and arithmetic expansion before being interpreted as a pathname. PATH is not used to search for the resultant pathname. MAIL If this parameter is set to a lename and the MAILPATH variable is not set, bash informs the user of the arrival of mail in the specied le. MAILCHECK Species how often (in seconds) bash checks for mail. The default is 60 seconds. When it is time to check for mail, the shell does so before prompting. If this variable is unset, the shell disables mail checking. MAILPATH A colon-separated list of pathnames to be checked for mail. The message to be printed may be specied by separating the pathname from the message with a ?. $ stands for the name of the current maille. Example: MAILPATH=/usr/spool/mail/bfox?"You have mail":/shell-mail?"$_ has mail!" Bash supplies a default value for this variable, but the location of the user mail les that it uses is system dependent (e.g., /usr/spool/mail/$USER). MAIL WARNING If set, and a le that bash is checking for mail has been accessed since the last time it was checked, the message The mail in maille has been read is printed.

PS1 The value of this parameter is expanded (see PROMPTING below) and used as the primary prompt string. The default value is bash$ . PS2 The value of this parameter is expanded and used as the secondary prompt string. The default is > . PS3 The value of this parameter is used as the prompt for the select command (see SHELL GRAMMAR above). PS4 The value of this parameter is expanded and the value is printed before each command bash displays during an execution trace. The rst character of PS4 is replicated multiple times, as necessary, to indicate multiple levels of indirection. The default is + . HISTSIZE The number of commands to remember in the command history (see HISTORY below). The default value is 500. HISTFILE The name of the le in which command history is saved. (See HISTORY below.) The default value is /.bash history. If unset, the command history is not saved when an interactive shell exits. HISTFILESIZE The maximum number of lines contained in the history le. When this variable is assigned a value, the history le is truncated, if necessary, to contain no more than that number of lines. The default value is 500. OPTERR If set to the value 1, bash displays error messages generated by the getopts builtin command (see SHELL BUILTIN COMMANDS below). OPTERR is initialized to 1 each time the shell is invoked or a shell script is executed. PROMPT COMMAND If set, the value is executed as a command prior to issuing each primary prompt. IGNOREEOF Controls the action of the shell on receipt of an EOF character as the sole input. If set, the value is the number of consecutive EOF characters typed as the rst characters on an input line before bash exits. If the variable exists but does not have a numeric value, or has no value, the default value is 10. If it does not exist, EOF signies the end of input to the shell. This is only in effect for interactive shells. TMOUT If set to a value greater than zero, the value is interpreted as the number of seconds to wait for input after issuing the primary prompt. Bash terminates after waiting for that number of seconds if input does not arrive. FCEDIT The default editor for the fc builtin command. FIGNORE A colon-separated list of sufxes to ignore when performing lename completion (see READLINE below). A lename whose sufx matches one of the entries in FIGNORE is excluded from the list of matched lenames. A sample value is .o:. INPUTRC The lename for the readline startup le, overriding the default of /.inputrc (see READLINE below). notify If set, bash reports terminated background jobs immediately, rather than waiting until before printing the next primary prompt (see also the -b option to the set builtin command).

9. Processes, Environment Variables

9.9. Environments of Processes

9. Processes, Environment Variables

history control HISTCONTROL If set to a value of ignorespace, lines which begin with a space character are not entered on the history list. If set to a value of ignoredups, lines matching the last history line are not entered. A value of ignoreboth combines the two options. If unset, or if set to any other value than those above, all lines read by the parser are saved on the history list. command oriented history If set, bash attempts to save all lines of a multiple-line command in the same history entry. This allows easy re-editing of multi-line commands. glob dot lenames If set, bash includes lenames beginning with a . in the results of pathname expansion. allow null glob expansion If set, bash allows pathname patterns which match no les (see Pathname Expansion below) to expand to a null string, rather than themselves. histchars The two or three characters which control history expansion and tokenization (see HISTORY EXPANSION below). The rst character is the history expansion character, that is, the character which signals the start of a history expansion, normally !. The second character is the quick substitution character, which is used as shorthand for re-running the previous command entered, substituting one string for another in the command. The default is . The optional third character is the character which signies that the remainder of the line is a comment, when found as the rst character of a word, normally #. The history comment character causes history substitution to be skipped for the remaining words on the line. It does not necessarily cause the shell parser to treat the rest of the line as a comment. nolinks If set, the shell does not follow symbolic links when executing commands that change the current working directory. It uses the physical directory structure instead. By default, bash follows the logical chain of directories when performing commands which change the current directory, such as cd. See also the description of the -P option to the set builtin ( SHELL BUILTIN COMMANDS below). hostname completion le HOSTFILE Contains the name of a le in the same format as /etc/hosts that should be read when the shell needs to complete a hostname. The le may be changed interactively; the next time hostname completion is attempted bash adds the contents of the new le to the already existing database. noclobber If set, bash does not overwrite an existing le with the >, >&, and <> redirection operators. This variable may be overridden when creating output les by using the redirection operator > instead of > (see also the -C option to the set builtin command). auto resume This variable controls how the shell interacts with the user and job control. If this variable is set, single word simple commands without redirections are treated as candidates for resumption of an existing stopped job. There is no ambiguity allowed; if there is more than one job beginning with the string typed, the job most recently accessed is selected. The name of a

stopped job, in this context, is the command line used to start it. If set to the value exact, the string supplied must match the name of a stopped job exactly; if set to substring, the string supplied needs to match a substring of the name of a stopped job. The substring value provides functionality analogous to the %? job id (see JOB CONTROL below). If set to any other value, the supplied string must be a prex of a stopped jobs name; this provides functionality analogous to the % job id. no exit on failed exec If this variable exists, a non-interactive shell will not exit if it cannot execute the le specied in the exec builtin command. An interactive shell does not exit if exec fails. cdable vars If this is set, an argument to the cd builtin command that is not a directory is assumed to be the name of a variable whose value is the directory to change to.

10. Mail

From mands@inetafrica.com Mon Jun 1 21:20:21 1998 Return-Path: <mands@inetafrica.com> Received: from pizza.cranzgot.co.za (root@pizza.cranzgot.co.za [192.168.2.254]) by onion.cranzgot.co.za (8.8.7/8.8.7) with ESMTP id VAA11942 for <psheer@icon.co.za>; Mon, 1 Jun 1998 21:20:20 +0200 Received: from mail450.icon.co.za (mail450.icon.co.za [196.26.208.3]) by pizza.cranzgot.co.za (8.8.5/8.8.5) with ESMTP id VAA19357 for <psheer@icon.co.za>; Mon, 1 Jun 1998 21:17:06 +0200 Received: from smtp02.inetafrica.com (smtp02.inetafrica.com [196.7.0.140]) by mail450.icon.co.za (8.8.8/8.8.8) with SMTP id VAA02315 for <psheer@icon.co.za>; Mon, 1 Jun 1998 21:24:21 +0200 (GMT) Received: from default [196.31.19.216] (fullmoon) by smtp02.inetafrica.com with smtp (Exim 1.73 #1) id 0ygTDL-00041u-00; Mon, 1 Jun 1998 13:57:20 +0200 Message-ID: <357296DF.60A3@inetafrica.com> Date: Mon, 01 Jun 1998 13:56:15 +0200 From: a person <mands@inetafrica.com> Reply-To: mands@inetafrica.com Organization: private X-Mailer: Mozilla 3.01 (Win95; I) MIME-Version: 1.0 To: paul sheer <psheer@icon.co.za> Subject: hello Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Status: RO X-Status: A hey paul its me how r u doing i am well what u been upot hows life hope your well amanda

Chapter 10

Mail
Electronic Mail, or email, is the way most people rst come into contact with the Internet. Although you may have used email in a graphical environment, here we show you how mail was rst intended to be used on a multiuser system. To a large extent what applies here is really what is going on in the background of any system that supports mail. A mail message is a block of text sent from one user to another, using some mail command or mailer program. A mail message will usually also be accompanied by a subject explaining what the mail is about. The idea of mail is that a message can be sent to someone even though he may not be logged in at the time and the mail will be stored for him until he is around to read it. An email address is probably familiar to you, for example: bruce@kangeroo.co.au. This means that bruce has a user account on a computer called kangeroo.co.au, which often means that he can log in as bruce on that machine. The text after the @ is always the name of the machine. Todays Internet does not obey this exactly, but there is always a machine that bruce does have an account on where mail is eventually sent. That machine is also usually a U NIX
machine.

Each mail message begins with a From at the beginning of a line, followed by a space. Then comes the mail header, explaining where the message was routed from to get to your mailbox, who sent the message, where replies should go, the subject of the mail, and various other mail header elds. Above, the header is longer than the mail messages. Examine the header carefully. The header ends with the rst blank line. The message itself (or body) starts right after. The next header in the le will once again start with a From. Froms on the beginning of a line never exist within the body. If they do, the mailbox is considered to be corrupt. Some mail readers store their messages in a different format. However the above format (called the mbox format) is the most common for U NIX. Of interest is a format called Maildir, which is one format that does not store mail messages in a single contiguous le. Instead, Maildir stores each message as a separate le within a directory. The name of the directory is then considered to be the mailbox le; by default Maildir uses a directory Maildir within the users home directory. 98

Sometimes email addresses are written in a more user-friendly form like Bruce Wallaby <bruce@kangeroo.co.au> or bruce@kangeroo.co.au (Bruce Wallaby). In this case, the surrounding characters are purely cosmetic; only bruce@kangeroo.co.au is ever used. When mail is received for you (from another user on the system or from a user from another system) it is appended to the le /var/spool/mail/<username> called the mail le or mailbox le; <username> is your login name. You then run some program that interprets your mail le, allowing you to browse the le as a sequence of mail messages and read and reply to them. An actual addition to your mail le might look like this: 97

10. Mail

10.1. Sending and Reading Mail

10.2. The SMTP Protocol Sending Mail Raw to Port 25

10. Mail

10.1 Sending and Reading Mail

The simplest way to send mail is to use the mail command. Type mail s "hello there" <username>. The mail program will then wait for you to type out your message. When you are nished, enter a . on its own on a single line. The user name will be another user on your system. If no one else is on your system, then send mail to root with mail -s "Hello there" root or mail -s "Hello there" root@localhost (if the @ is not present, then the local machine, localhost, is implied). Sending les over email is discussed in Section 12.6. You can use mail to view your mailbox. This is a primitive utility in comparison with modern graphical mail readers but is probably the only mail reader that can handle arbitrarily sized mailboxes. Sometimes you may get a mailbox that is over a gigabyte in size, and mail is the only way to delete messages from it. To view your mailbox, type mail, and then z to read your next window of messages, and z- to view the previous window. Most commands work like command message number, for example, delete 14 or reply 7. The message number is the left column with an N next to it (for a New message). For the state of the art in terminal-based mail readers (also called mail clients), try mutt and pine. pines license is not Free.

Connected to 192.168.2.1. Escape character is ]. 220 onion.cranzgot.co.za ESMTP Sendmail 8.9.3/8.9.3; Wed, 2 Feb 2000 14:54:47 +0200 HELO cericon.cranzgot.co.za 250 onion.cranzgot.co.za Hello cericon.ctn.cranzgot.co.za [192.168.3.9], pleased to meet yo MAIL FROM:psheer@icon.co.za 250 psheer@icon.co.za... Sender ok RCPT TO:mands@inetafrica.com 250 mands@inetafrica.com... Recipient ok DATA 354 Enter mail, end with "." on a line by itself Subject: just to say hi hi there heres a short message . 250 OAA04620 Message accepted for delivery QUIT 221 onion.cranzgot.co.za closing connection Connection closed by foreign host. [root@cericon]#

There are also some graphical mail readers in various stages of development. At the time I am writing this, I have been using balsa for a few months, which was the best mail reader I could nd.

The above causes the message hi there heres a short message to be delivered to mands@inetafrica.com (the ReCiPienT). Of course, I can enter any address that I like as the sender, and it can be difcult to determine who sent the message. In this example, the Subject: is the only header eld, although I neednt have supplied a header at all. Now, you may have tried this and gotten a rude error message. This might be because the MTA is congured not to relay mail except from specic trusted machines say, only those machines within that organization. In this way anonymous email is prevented. On the other hand, if you are connecting to the users very own mail server, it has to necessarily receive the mail, regardless of who sent it. Hence, the above is a useful way to supply a bogus FROM address and thereby send mail almost anonymously. By almost I mean that the mail server would still have logged the machine from which you connected and the time of connectionthere is no perfect anonymity for properly congured mail servers. The above technique is often the only way to properly test a mail server, and should be practiced for later.

10.2 The SMTP Protocol Sending Mail Raw to Port 25

To send mail, you need not use a mail client at all. The mail client just follows SMTP (Simple Mail Transfer Protocol), which you can type in from the keyboard. For example, you can send mail by telneting to port 25 of a machine that has an MTA (Mail Transfer Agentalso called the mailer daemon or mail server) running. The word daemon denotes programs that run silently without user intervention. This is, in fact, how so-called anonymous mail or spam mail

daemon runs in most small institutions in the world and has the simple task of receiving mail requests and relaying them on to other mail servers. Try this, for example (obviously substituting mail.cranzgot.co.za for the name of a mail server that you normally use):
[root@cericon]# telnet mail.cranzgot.co.za 25 Trying 192.168.2.1...

indicate unsolicited emailthat is, junk mail that is posted in bulk to large numbers of arbitrary email addresses. Sending spam is considered unethical Internet practice. is sent on the Internet. A mailer

Spam is a term used to

100

11.2. The Password File /etc/passwd

11. User Accounts and Ownerships

chown <user>[:<group>] <filename>

Chapter 11

11.2 The Password File /etc/passwd

The only place in the whole system where a user name is registered is in this le.
neednt worry about these for now.

User Accounts and User Ownerships

Once a user is added to this le, that user is said to exist on the system. If you thought that user accounts were stored in some unreachable dark corner, then this should dispel that idea. This is also known as the password le to administrators. View this le with less:
root:x:0:0:Paul Sheer:/root:/bin/bash bin:x:1:1:bin:/bin: daemon:x:2:2:daemon:/sbin: adm:x:3:4:adm:/var/adm: lp:x:4:7:lp:/var/spool/lpd: sync:x:5:0:sync:/sbin:/bin/sync shutdown:x:6:0:shutdown:/sbin:/sbin/shutdown halt:x:7:0:halt:/sbin:/sbin/halt mail:x:8:12:mail:/var/spool/mail: news:x:9:13:news:/var/spool/news: uucp:x:10:14:uucp:/var/spool/uucp: gopher:x:13:30:gopher:/usr/lib/gopher-data: ftp:x:14:50:FTP User:/home/ftp: nobody:x:99:99:Nobody:/: alias:x:501:501::/var/qmail/alias:/bin/bash paul:x:509:510:Paul Sheer:/home/paul:/bin/bash jack:x:511:512:Jack Robbins:/home/jack:/bin/bash silvia:x:511:512:Silvia Smith:/home/silvia:/bin/bash

Exceptions to this rule are several distributed authentication schemes and the Samba package, but you

U NIX intrinsically supports multiple users. Each user has a personal home directory /home/<username> in which the users les are stored, hidden from other users. So far you may have been using the machine as the root user, who is the system administrator and has complete access to every le on the system. The root is also called the superuser. The home directory of the root user is /root. Note that there is an ambiguity here: the root directory is the topmost directory, known as the / directory. The root users home directory is /root and is called the home directory of root. Other than the superuser, every other user has limited access to les and directories. Always use your machine as a normal user. Log in as root only to do system administration. This practice will save you from the destructive power that the root user has. In this chapter we show how to manually and automatically create new users. Users are also divided into sets, called groups. A user can belong to several groups and there can be as many groups on the system as you like. Each group is dened by a list of users that are part of that set. In addition, each user may have a group of the same name (as the users login name), to which only that user belongs.
10

Above is an extract of my own password le. Each user is stored on a separate line. Many of these are not human login accounts but are used by other programs. Each line contains seven elds separated by colons. The account for jack looks like this: jack The users login name. It should be composed of lowercase letters and numbers. Other characters are allowed, but are not preferable. In particular, there should never be two user names that differ only by their capitalization. x The users encrypted password. An x in this eld indicates that it is stored in a separate le, /etc/shadow. This shadow password le is a later addition to U NIX systems. It contains additional information about the user. 102

11.1 File Ownerships

Each le on a system is owned by a particular user and also owned by a particular group. When you run ls -al, you can see the user that owns the le in the third column and the group that owns the le in the fourth column (these will often be identical, indicating that the les group is a group to which only the user belongs). To change the ownership of the le, simply use the chown, change ownerships, command as follows. 101

11. User Accounts and Ownerships

11.3. Shadow Password File: /etc/shadow

11.4. The groups Command and /etc/group

11. User Accounts and Ownerships

511 The users user identication number, UID.

This is used by programs as a short alternative to the users login name. In fact, internally, the login name is never used, only the UID.

512 The users group identication number, GID.

be discussed later.

Similarly applies to the GID. Groups will

Jack Robbins The users full name.

Few programs ever make use of this eld.

/home/jack The users home directory. The HOME environment variable will be set to this when the user logs in. /bin/bash The shell to start when the user logs in.

:CavHIpD1W.cmg:. Hence, trying to guess the password from the hash can only be done by trying every possible password. Such a brute force attack is therefore considered computationally expensive but not impossible. To check if an entered password matches, just apply the identical mathematical algorithm to it: if it matches, then the password is correct. This is how the login command works. Sometimes you will see a * in place of a hashed password. This means that the account has been disabled. 10795 Days since January 1, 1970, that the password was last changed. 0 Days before which password may not be changed. Usually zero. This eld is not often used. 99999 Days after which password must be changed. This is also rarely used, and will be set to 99999 by default.

11.3 Shadow Password File: /etc/shadow

eryone on the system can read the le.

The problem with traditional passwd les is that they had to be world readable Evin order for programs to extract information, such as the users full name, about the user. This means that everyone can see the encrypted password in the second eld. Anyone can copy any other users password eld and then try billions of different passwords to see if they match. If you have a hundred users on the system, there are bound to be several that chose passwords that matched some word in the dictionary. The so-called dictionary attack will simply try all 80,000 common English words until a match is found. If you think you are clever to add a number in front of an easy-to-guess dictionary word, password cracking algorithms know about these as well. And about every other trick you can think of. To solve this problem the shadow password le was invented. The shadow password le is used only for authentication Verifying that the user is the genuine owner of the account. and is not world readablethere is no information in the shadow password le that a common program will ever needno regular user has permission to see the encrypted password eld. The elds are colon separated just like the passwd le.

7 Days before password is to expire that user is warned of pending password expiration. -1 Days after password expires that account is considered inactive and disabled. 1 is used to indicate innitythat is, to mean we are effectively not using this feature. -1 Days since January 1, 1970, when account will be disabled. 134537220 Flag reserved for future use.

11.4 The groups Command and /etc/group

On a U NIX system you may want to give a number of users the same access rights. For instance, you may have ve users that should be allowed to access some privileged le and another ten users that are allowed to run a certain program. You can group these users into, for example, two groups previl and wproc and then make the relevant le and directories owned by that group with, say,
chown root:previl /home/somefile chown root:wproc /usr/lib/wproc

Here is an example line from a /etc/shadow le:

jack:Q,Jpl.or6u2e7:10795:0:99999:7:-1:-1:134537220

jack The users login name. Q,Jpl.or6u2e7 The users encrypted password known as the hash of the password. This is the users 8-character password with a one-way hash function applied to it. It is simply a mathematical algorithm applied to the password that is known to produce a unique result for each password. To demonstrate: the (rather poor) password Loghimin hashes to :lZ1F.0VSRRucs: in the shadow le. An almost identical password loghimin gives a completely different hash 103

Permissions Explained later. dictate the kind of access, but for the meantime, the le/directory must at least be owned by that group. The /etc/group le is also colon separated. A line might look like this:

wproc:x:524:jack,mary,henry,arthur,sue,lester,fred,sally

104

11. User Accounts and Ownerships

11.5. Manually Creating a User Account

11.6. Automatically: useradd and groupadd

11. User Accounts and Ownerships

wproc The name of the group. There should really also be a user of this name as well. x The groups password. This eld is usually set with an x and is not used. 524 The GID group ID. This must be unique in the groups le. jack,mary,henry,arthur,sue,lester,fred,sally The list of users that belong to the group. This must be comma separated with no spaces. You can obviously study the group le to nd out which groups a user belongs to, That is, not which users does a group consist of? which is easy to see at a glance. but when there are a lot of groups, it can be tedious to scan through the entire le. The groups command prints out this information.

11.6 Automatically Creating a User Account: useradd and groupadd

The above process is tedious. The commands that perform all these updates automatically are useradd, userdel, and usermod. The man pages explain the use of these commands in detail. Note that different avors of U NIX have different commands to do this. Some may even have graphical programs or web interfaces to assist in creating users. In addition, the commands groupadd, groupdel, and groupmod do the same with respect to groups.

11.5 Manually Creating a User Account

The following steps are required to create a user account: /etc/passwd entry To create an entry in this le, simply edit it and copy an existing line. When editing conguration les, never write out a line from scratch if it has some kind of
special format. Always copy an existing entry that has proved itself to be correct, and then edit in the appropriate changes. This will prevent you from making errors. Always add users from the

11.7 User Logins

It is possible to switch from one user to another, as well as view your login status and the status of other users. Logging in also follows a silent procedure which is important to understand.

11.7.1 The login command

A user most often gains access to the system through the login program. This program looks up the UID and GID from the passwd and group le and authenticates the user. The following is quoted from the login man page, and explains this procedure in detail:
login is used when signing onto a system. It can also be used to switch from one user to another at any time (most modern shells have support for this feature built into them, however). If an argument is not given, login prompts for the username. If the user is not root, and if /etc/nologin exists, the contents of this le are printed to the screen, and the login is terminated. This is typically used to prevent logins when the system is being taken down. If special access restrictions are specied for the user in /etc/usertty, these must be met, or the login attempt will be denied and a syslog System error log program syslog writes all system messages to the le /var/log/messages. message will be generated. See the section on Special Access Restrictions. If the user is root, then the login must be occuring on a tty listed in /etc/securetty.

bottom and try to preserve the pattern of the lethat is, if you see numbers increasing, make yours t in; if you are adding a normal user, add it after the existing lines of normal users. Each user must have a unique UID and should usually have a unique GID. So if you are adding a line to the end of the le, make your new UID and GID the same as the last line but incremented by 1.

/etc/shadow entry Create a new shadow password entry. At this stage you do not know what the hash is, so just make it a *. You can set the password with the passwd command later. /etc/group entry Create a new group entry for the users group. Make sure the number in the group entry matches that in the passwd le. /etc/skel This directory contains a template home directory for the user. Copy the entire directory and all its contents into /home directory, renaming it to the name of the user. In the case of our jack example, you should have a directory /home/jack. Home directory ownerships You need to now change the ownerships of the home directory to match the user. The command chown -R jack:jack /home/jack will accomplish this change. Setting the password Use passwd <username> to set the users password. 105

If this le is not present, then root logins will be allowed from anywhere. It is worth deleting
106

this le if your machine is protected by a rewall and you would like to easily login from

11. User Accounts and Ownerships

11.7. User Logins

11. User Accounts and Ownerships

another machine on your LAN. If /etc/securetty is present, then logins are only allowed from the terminals it lists. Failures will be logged with the syslog facility.

11.7.3 The who, w, and users commands to see who is logged in

who and w print a list of users logged in to the system, as well as their CPU consumption and other statistics. who --help gives:
Usage: who [OPTION]... [ FILE | ARG1 ARG2 ] -H, -i, -m -q, -s -T, --heading -u, --idle --count -w, --mesg --message --writable --help --version print line of column headings add user idle time as HOURS:MINUTES, . or old only hostname and user associated with stdin all login names and number of users logged on (ignored) add users message status as +, - or ? same as -T same as -T display this help and exit output version information and exit

After these conditions have been checked, the password will be requested and checked (if a password is required for this username). Ten attempts are allowed before login dies, but after the rst three, the response starts to get very slow. Login failures are reported via the syslog facility. This facility is also used to report any successful root logins. If the le .hushlogin exists, then a quiet login is performed (this disables the checking of mail and the printing of the last login time and message of the day). Otherwise, if /var/log/lastlog exists, the last login time is printed (and the current login is recorded). Random administrative things, such as setting the UID and GID of the tty are performed. The TERM environment variable is preserved, if it exists (other environment variables are preserved if the -p option is used). Then the HOME, PATH, SHELL, TERM, MAIL, and LOGNAME environment variables are set. PATH defaults to /usr/local/bin:/bin:/usr/bin:. Note that the . the current directoryis listed in the PATH. This is only the default PATH however. for normal users, and to /sbin:/bin:/usr/sbin:/usr/bin for root. Last, if this is not a quiet login, the message of the day is printed and the le with the users name in /usr/spool/mail will be checked, and a message printed if it has non-zero length. The users shell is then started. If no shell is specied for the user in /etc/passwd, then /bin/sh is used. If there is no directory specied in /etc/passwd, then / is used (the home directory is checked for the .hushlogin le described above).

If FILE is not specified, use /var/run/utmp. /var/log/wtmp as FILE is common. If ARG1 ARG2 given, -m presumed: am i or mom likes are usual.

A little more information can be gathered from the info pages for this command. The idle time indicates how long since the user has last pressed a key. Most often, one just types who -Hiw. w is similar. An extract of the w man page says:
w displays information about the users currently on the machine, and their processes. The header shows, in this order, the current time, how long the system has been running, how many users are currently logged on, and the system load averages for the past 1, 5, and 15 minutes. The following entries are displayed for each user: login name, the tty name, the remote host, login time, idle time, JCPU, PCPU, and the command line of their current process. The JCPU time is the time used by all processes attached to the tty. It does not include past background jobs, but does include currently running background jobs. The PCPU time is the time used by the current process, named in the what eld.

11.7.2 The set user, su command

To temporarily become another user, you can use the su program:
su jack

This command prompts you for a password (unless you are the root user to begin with). It does nothing more than change the current user to have the access rights of jack. Most environment variables will remain the same. The HOME, LOGNAME, and USER environment variables will be set to jack, but all other environment variables will be inherited. su is, therefore, not the same as a normal login. To get the equivalent of a login with su, run

su - jack

This will cause all initialization scripts (that are normally run when the user logs in) to be executed. What actually happens is that the subsequent shell is started with a - in front of the
zeroth argument. This makes the shell read the users personal prole. The login command also does this. Hence, after running su with the - option, you logged in as if with the login

Finally, from a shell script the users command is useful for just seeing who is logged in. You can use in a shell script, for example:
for user in users ; do <etc> done

command.

107

108

11. User Accounts and Ownerships

11.7. User Logins

11. User Accounts and Ownerships

11.7.4 The id command and effective UID

id prints your real and effective UID and GID. A user normally has a UID and a GID but may also have an effective UID and GID as well. The real UID and GID are what a process will generally think you are logged in as. The effective UID and GID are the actual access permissions that you have when trying to read, write, and execute les.

11.7.5 User limits

There is a le /etc/security/limits.conf that stipulates the limitations on CPU usage, process consumption, and other resources on a per-user basis. The documentation for this cong le is contained in /usr/[share/]doc/pam-<version>/txts/README.pam limits.

109

110

12.2. rcp and scp

12. Using Internet Services

Though rlogin and telnet are very convenient, they should never be used across a public network because your password can easily be read off the wire as you type it in.

Chapter 12

12.2

rcp and scp

Using Internet Services

This chapter summarizes remote access and the various methods of transferring les and data over the Internet.

rcp stands for remote copy and scp is the secure version from the ssh package. These two commands copy les from one machine to another using a similar notation to cp.
rcp [-r] [<remote_machine>:]<file> [<remote_machine>:]<file> scp [-l <username>] [-r] [<remote_machine>:]<file> [<remote_machine>:]<file>

Here is an example:
[psheer@cericon]# rcp /var/spool/mail/psheer \ divinian.cranzgot.co.za:/home/psheer/mail/cericon [psheer@cericon]# scp /var/spool/mail/psheer \ divinian.cranzgot.co.za:/home/psheer/mail/cericon The authenticity of host divinian.cranzgot.co.za cant be established. RSA key fingerprint is 43:14:36:5d:bf:4f:f3:ac:19:08:5d:4b:70:4a:7e:6a. Are you sure you want to continue connecting (yes/no)? yes Warning: Permanently added divinian.cranzgot.co.za (RSA) to the list of known hosts. psheer@divinians password: psheer 100% |***************************************| 4266 KB 01:18

12.1

ssh, not telnet or rlogin

telnet is a program for talking to a U NIX network service. It is most often used to do a remote login. Try
telnet <remote_machine> telnet localhost

The -r option copies recursively and copies can take place in either direction or even between two nonlocal machines. scp should always be used instead of rcp for security reasons. Notice also the warning given by scp for this rst-time connection. See the ssh documentation for how to make your rst connection securely. All commands in the ssh package have this same behavior.

to log in to your remote machine. It neednt matter if there is no physical network; network services always work regardless because the machine always has an internal link to itself. rlogin is like a minimal version of telnet that allows login access only. You can type
rlogin -l <username> <remote_machine> rlogin -l jack localhost

12.3

rsh

if the system is congured to support remote logins. These two services are the domain of old world U NIX; for security reasons, ssh is now the preferable service for logging in remotely:
ssh [-l <username>] <remote_machine>

rsh (remote shell) is a useful utility for executing a command on a remote machine. Here are some examples:
[psheer@cericon]# rsh divinian.cranzgot.co.za hostname divinian.cranzgot.co.za [psheer@cericon]# rsh divinian.cranzgot.co.za \ tar -czf - /home/psheer | dd of=/dev/fd0 bs=1024 tar: Removing leading / from member names 20+0 records in 20+0 records out

111

112

12. Using Internet Services

12.4. FTP

12.5. finger

12. Using Internet Services

[psheer@cericon]# cat /var/spool/mail/psheer | rsh divinian.cranzgot.co.za \ sh -c cat >> /home/psheer/mail/cericon

The le will be downloaded into your current directory. You can also cd to the /incoming directory and upload les. Try

The rst command prints the host name of the remote machine. The second command backs up my remote home directory to my local oppy disk. (More about dd and /dev/fd0 come later.) The last command appends my local mailbox le to a remote mailbox le. Notice how stdin, stdout, and stderr are properly redirected to the local terminal. After reading Chapter 29 see rsh(8) or in.rshd(8) to congure this service. Once again, for security reasons rsh should never be available across a public network.

to upload the le that you have just downloaded. Most FTP sites have an /incoming directory that is ushed periodically. FTP allows far more than just uploading of les, although the administrator has the option to restrict access to any further features. You can create directories, change ownerships, and do almost anything you can on a local le system. If you have several machines on a trusted LAN (Local Area Networkthat is, your private ofce or home network), all should have FTP enabled to allow users to easily copy les between machines. How to install and congure one of the many available FTP servers will become obvious later in this book.

put README

12.4 FTP
FTP stands for File Transfer Protocol. If FTP is set up on your local machine, then other machines can download les. Type or
ftp metalab.unc.edu

12.5

finger

ncftp metalab.unc.edu

user program accessing some remote service.

ftp is the traditional command-line U NIX FTP client, client always indicates the while ncftp is a more powerful client that will not always be installed. You will now be inside an FTP session. You will be asked for a login name and a password. The site metalab.unc.edu is one that allows anonymous logins. This means that you can type anonymous as your user name, and then anything you like as a password. You will notice that the session will ask you for an email address as your password. Any sequence of letters with an @ symbol will sufce, but you should put your actual email address out of politeness.

finger is a service for remotely listing who is logged in on a remote system. Try finger @<hostname> to see who is logged in on <hostname>. The nger service will often be disabled on machines for security reasons.

12.6

Sending Files by Email

The FTP session is like a reduced shell. You can type cd, ls, and ls -al to view le lists. help brings up a list of commands, and you can also type help <command> to get help on a specic command. You can download a le by using the get <filename> command, but before you do this, you must set the transfer type to binary. The transfer type indicates whether or not newline characters will be translated to DOS format. Typing ascii turns on this feature, while binary turns it off. You may also want to enter hash which will print a # for every 1024 bytes of download. This is useful for watching the progress of a download. Go to a directory that has a README le in it and enter
get README

Mail is being used more and more for transferring les between machines. It is bad practice to send mail messages over 64 kilobytes over the Internet because it tends to excessively load mail servers. Any le larger than 64 kilobytes should be uploaded by FTP onto some common FTP server. Most small images are smaller than this size, hence sending a small JPEG A common Internet image le format. These are especially compressed and are usually under 100 kilobytes for a typical screen-sized photograph. image is considered acceptable.

12.6.1

uuencode and uudecode

113

If you must send les by mail then you can do it by using uuencode. This utility packs binary les into a format that mail servers can handle. If you send a mail message containing arbitrary binary data, it will more than likely be corrupted on the way because mail agents are only designed to handle a limited range of characters. uuencode represents a binary le with allowable characters, albeit taking up slightly more space. 114

12. Using Internet Services

12.6. Sending Files by Email

12. Using Internet Services

Here is a neat trick to pack up a directory and send it to someone by mail.

tar -czf - <mydir> | uuencode <mydir>.tar.gz \ | mail -s "Here are some files" <user>@<machine>

To unpack a uuencoded le, use the uudecode command:

uudecode <myfile>.uu

12.6.2

MIME encapsulation

Most graphical mail readers have the ability to attach les to mail messages and read these attachments. The way they do this is not with uuencode but in a special format known as MIME encapsulation. MIME (Multipurpose Internet Mail Extensions) is a way of representing multiple les inside a single mail message. The way binary data is handled is similar to uuencode, but in a format known as base64. Each MIME attachment to a mail message has a particular type, known as the MIME type. MIME types merely classify the attached le as an image, an audio clip, a formatted document, or some other type of data. The MIME type is a text tag with the format <major>/<minor>. The major part is called the major MIME type and the minor part is called the minor MIME type. Available major types match all the kinds of les that you would expect to exist. They are usually one of application, audio, image, message, text, or video. The application type means a le format specic to a particular utility. The minor MIME types run into the hundreds. A long list of MIME types can be found in /etc/mime.types. If needed, some useful command-line utilities in the same vein as uuencode can create and extract MIME messages. These are mpack, munpack, and mmencode (or mimencode).

115

116

13.2. HTTP Web Sites

13. L INUX Resources

The generic Linux download directory. It contains innumerable U NIX packages in source and binary form, categorized in a directory tree. For instance, mail clients have their own directory with many mail packages inside. metalab is the place where new developers can host any new software that they have produced. There are instructions on the FTP site to upload software and to request it to be placed into a directory.

Chapter 13

L INUX Resources
Very often it is not even necessary to connect to the Internet to nd the information you need. Chapter 16 contains a description of most of the documentation on a L INUX distribution. It is, however, essential to get the most up-to-date information where security and hardware driver support are concerned. It is also fun and worthwhile to interact with L INUX users from around the globe. The rapid development of Free software could mean that you may miss out on important new features that could streamline IT services. Hence, reviewing web magazines, reading newsgroups, and subscribing to mailing lists are essential parts of a system administrators role.

The kernel sources. This is a mirror of the kernel archives where Linus and other maintainers upload new stable Meaning that the software is well tested and free of serious Meaning that the software is in its development stages. kernel versions bugs. and beta and kernel patches.

The various distributions. RedHat, Debian , and possibly other popular distributions may be present. This list is by no means exhaustive. Depending on the willingness of the site maintainer, there may be mirrors to far more sites from around the world. The FTP site is how you will download free software. Often, maintainers will host their software on a web site, but every popular package will almost always have an FTP site where versions are persistently stored. An example is metalab.unc.edu in the directory /pub/Linux/apps/editors/X/cooledit/ where the authors own Cooledit package is distributed.

13.2

HTTP Web Sites

13.1 FTP Sites and the sunsite Mirror

The metalab.unc.edu FTP site (previously called sunsite.unc.edu) is one of the traditional sites for free software. It is mirrored in almost every country that has a signicant IT infrastructure. If you point your web browser there, you will nd a list of mirrors. For faster access, do pick a mirror in your own country. It is advisable to browse around this FTP site. In particular you should try to nd the locations of: packages are stored. This The directory where all sources for ofcial GNU would be a mirror of the Free Software Foundations FTP archives. These are packages that were commissioned by the FSF and not merely released under the General Public License). The FSF will distribute them in source GPL (GNU form (.tar.gz) for inclusion into various distributions. They will, of course, compile and work under any U NIX. 117

Most users should already be familiar with using a web browser. You should also become familiar with the concept of a web search. Do I need to explain this? You search the web when you point your web browser to a popular search engine like http://www.google.com/, http://www.google.com/linux, http://infoseek.go.com/, http://www.altavista.com/, or http://www.yahoo.com/ and search for particular key words. Searching is a bit of a black art with the billions of web pages out there. Always consult the search engines advanced search options to see how you can do more complex searches than just plain word searches.

The web sites in the FAQ (Frequently Asked Questions) (see Appendix D) should all be consulted to get an overview on some of the primary sites of interest to L INUX users. Especially important is that you keep up with the latest L INUX news. I nd the Linux Weekly News http://lwn.net/ an excellent source. Also, the famous (and infamous) SlashDot http://slashdot.org/ web site gives daily updates about stuff that matters (and therefore contains a lot about free software). Fresh Meat http://freshmeat.net/ is a web site devoted to new software releases. You will nd new or updated packages announced every few hours or so. 118

13. L INUX Resources

13.3. SourceForge

13.5. Newsgroups

13. L INUX Resources

Linux Planet http://www.linuxplanet.com/ seems to be a new (?) web site that I just found while writing this. It looks like it contains lots of tutorial information on L INUX . News Forge http://www.newsforge.net/ also contains daily information about software issues. Lycos http://download.lycos.com/static/advanced search.asp is an efcient FTP search engine for locating packages. It is one of the few search engines that understand regular expressions. Realistically, though, a new L INUX web site is created every week; almost anything prepended or appended to linux is probably a web site already.

This command adds your name to the mailing list <mailing-listname>@<machine>, to which messages are posted. Do the same for listserv-type lists, by sending the same message to listserv@<machine>. For instance, if you are an administrator for any machine that is exposed to the Internet, you should get on bugtraq. Send email to
subscribe bugtraq

to listserv@netspace.org, and become one of the tens of thousands of users that read and report security problems about L INUX . To unsubscribe to a list is just as simple. Send an email message:

13.3 SourceForge
A new phenomenon in the free software community is the SourceForge web site, http://www.sourceforge.net/. Developers can use this service at no charge to host their projects web site, FTP archives, and mailing lists. SourceForge has mushroomed so rapidly that it has come to host the better half of all free software projects.

unsubscribe <mailing-list-name>

Never send subscribe or unsubscribe messages to the mailing list itself. Send subscribe or unsubscribe messages only to to the address majordomo@<machine> or listserv@<machine>.

13.4.2

*-request

13.4 Mailing Lists

A mailing list is a special address that, when posted to, automatically sends email to a long list of other addresses. You usually subscribe to a mailing list by sending some specially formatted email or by requesting a subscription from the mailing list manager. Once you have subscribed to a list, any email you post to the list will be sent to every other subscriber, and every other subscribers posts to the list will be sent to you. There are mostly three types of mailing lists: the majordomo type, the listserv type, and the *-request type.

You subscribe to these mailing lists by sending an empty email message to <mailinglist-name>-request@<machine> with the word subscribe as the subject. The same email with the word unsubscribe removes you from the list. Once again, never send subscribe or unsubscribe messages to the mailing list itself.

13.5

Newsgroups

A newsgroup is a notice board that everyone in the world can see. There are tens of thousands of newsgroups and each group is unique in the world. The client software you use to read a newsgroup is called a news reader (or news client). rtin is a popular text mode reader, while netscape is graphical. pan is an excellent graphical news reader that I use. Newsgroups are named like Internet hosts. One you might be interested in is comp.os.linux.announce. The comp is the broadest subject description for computers; os stands for operating systems; and so on. Many other linux newsgroups are devoted to various L INUX issues. 120

13.4.1

Majordomo and Listserv

To subscribe to the majordomo variety, send a mail message to majordomo@<machine> with no subject and a one-line message:
subscribe <mailing-list-name>

119

13. L INUX Resources

13.6. RFCs

13. L INUX Resources

Newsgroups servers are big hungry beasts. They form a tree-like structure on the Internet. When you send mail to a newsgroup it takes about a day or so for the mail you sent to propagate to every other server in the world. Likewise, you can see a list of all the messages posted to each newsgroup by anyone anywhere. Whats the difference between a newsgroup and a mailing list? The advantage of a newsgroup is that you dont have to download the messages you are not interested in. If you are on a mailing list, you get all the mail sent to the list. With a newsgroup you can look at the message list and retrieve only the messages you are interested in. Why not just put the mailing list on a web page? If you did, then everyone in the world would have to go over international links to get to the web page. It would load the server in proportion to the number of subscribers. This is exactly what SlashDot is. However, your newsgroup server is local, so you retrieve mail over a faster link and save Internet trafc.

13.6 RFCs
An indispensable source of information for serious administrators or developers is the RFCs. RFC stands for Request For Comments. RFCs are Internet standards written by authorities to dene everything about Internet communication. Very often, documentation will refer to RFCs. There are also a few nonsense RFCs out there. For example there is an
RFC to communicate using pigeons, and one to facilitate an innite number of monkeys trying to write the complete works of Shakespeare. Keep a close eye on Slashdot http://slashdot.org/ to catch these.

ftp://metalab.unc.edu/pub/docs/rfc/ (and mirrors) has the complete RFCs archived for download. There are about 2,500 of them. The index le rfc-index.txt is probably where you should start. It has entries like:
2045 Multipurpose Internet Mail Extensions (MIME) Part One: Format of Internet Message Bodies. N. Freed & N. Borenstein. November 1996. (Format: TXT=72932 bytes) (Obsoletes RFC1521, RFC1522, RFC1590) (Updated by RFC2184, RFC2231) (Status: DRAFT STANDARD)
5

2046 Multipurpose Internet Mail Extensions (MIME) Part Two: Media Types. N. Freed & N. Borenstein. November 1996. (Format: TXT=105854 bytes) (Obsoletes RFC1521, RFC1522, RFC1590) (Status: DRAFT STANDARD)

and
2068 Hypertext Transfer Protocol -- HTTP/1.1. R. Fielding, J. Gettys, J. Mogul, H. Frystyk, T. Berners-Lee. January 1997. (Format: TXT=378114 bytes) (Status: PROPOSED STANDARD)

Well, you get the idea.

121

122

14.1. The chmod Command

14. Permission and Modication Times

Table 14.1 File and directory permissions

Possible chars, for unset r Effect for directories Effect for les

Chapter 14

User, u

w x s S

Permission and Modication Times

Every le and directory on a U NIX system, besides being owned by a user and a group, has access ags A switch that can either be on or off. (also called access bits) dictating what kind of access that user and group have to the le.

User can read the contents of the directory. With x or s, user can create and remove les in the directory. User can access the contents of the les in a directory for x or s. S has no effect.

User can read the le. User can write to the le. User can execute the le for x or s. s, known as the setuid bit, means to set the user owner of the subsequent process to that of the le. S has no effect. Group can read the le. Group can write to the le. Group can execute the le for x or s. s, known as the setgid bit, means to set the group owner of the subsequent process to that of the le. S has no effect. Everyone can read the le. Everyone can write to the le. Group can execute the le for x or t. For t, save the process text image to the swap device so that future loads will be faster (I dont know if this has an effect on L INUX ). T has no effect.

r Group, g w x s S

Running ls -ald /bin/cp /etc/passwd /tmp gives you a listing like this:
1 root 1 root 5 root root root root 28628 Mar 24 1999 /bin/cp 1151 Jul 23 22:42 /etc/passwd 4096 Sep 25 15:23 /tmp

Other, o

w x t T

-rwxr-xr-x -rw-r--r-drwxrwxrwt

In the leftmost column are ags which completely describe the access rights to the le. So far I have explained that the furthest ag to the left is either - or d, indicating an ordinary le or directory. The remaining nine have a - to indicate an unset value or one of several possible characters. Table 14.1 gives a complete description of le system permissions. For example,

Group can read the contents of the directory. With x or s, group can create and remove les in the directory. Group can access the contents of the les in a directory for x. For s, force all les in this directory to the same group as the directory. S has no effect. Everyone can read the contents of the directory. With x or t, everyone can create and remove les in the directory. Everyone can access the contents of the les in a directory for x and t. t, known as the sticky bit, prevents users from removing les that they do not own, hence users are free to append to the directory but not to remove other users les. T has no effect.

14.1 The chmod Command

You use the chmod command to change the permissions of a le. Its usually used as follows:
chmod [-R] [u|g|o|a][+|-][r|w|x|s|t] <file> [<file>] ...

124

chmod u+x myfile

adds execute permissions for the user of myfile. And,

chmod a-rx myfile

123

14. Permission and Modication Times

14.2. The umask Command

14.3. Modication Times: stat

14. Permission and Modication Times

removes read and execute permissions for allthat is, user, group, and other. The -R option, once again means recursive, diving into subdirectories as usual. Permission bits are often represented in their binary form, especially in programs. It is convenient to show the rwxrwxrwx set in octal, See Section 2.1. where each digit ts conveniently into three bits. Files on the system are usually created with mode 0644, meaning rw-r--r--. You can set permissions explicitly with an octal number, for example,

14.3

Modication Times: stat

In addition to permissions, each le has three integers associated with it that represent, in seconds, the last time the le was accessed (read), when it was last modied (written to), and when its permissions were last changed. These are known as the atime, mtime, and ctime of a le respectively. To get a complete listing of the les permissions, use the stat command. Here is the result of stat /etc:
File: Size: Mode: Device: Access: Modify: Change: "/etc" 4096 Filetype: Directory (0755/drwxr-xr-x) Uid: ( 0/ 3,1 Inode: 14057 Links: 41 Sat Sep 25 04:09:08 1999(00000.15:02:23) Fri Sep 24 20:55:14 1999(00000.22:16:17) Fri Sep 24 20:55:14 1999(00000.22:16:17) root) Gid: ( 0/ root)

chmod 0755 myfile

gives myfile the permissions rwxr-xr-x. For a full list of octal values for all kinds of permissions and le types, see /usr/include/linux/stat.h. In Table 14.1 you can see s, the setuid or setgid bit. If it is used without execute permissions then it has no meaning and is written as a capitalized S. This bit effectively colorizes an x into an s, so you should read an s as execute with the setuid or setgid bit set. t is known as the sticky bit. It also has no meaning if there are no execute permissions and is written as a capital T. The leading 0 can in be ignored, but is preferred for explicitness. It can take on a value representing the three bits, setuid (4), setgid (2), and sticky (1). Hence a value of 5764 is 101 111 110 100 in binary and gives -rwsrw-r-T.

The Size: quoted here is the actual amount of disk space used to store the directory listing, and is the same as reported by ls. In this case it is probably four disk blocks of 1024 bytes each. The size of a directory as quoted here does not mean the sum of all les contained under it. For a le, however, the Size: would be the exact le length in bytes (again, as reported by ls).

14.2 The umask Command

umask sets the default permissions for newly created les; it is usually 022. This default value means that the permissions of any new le you create (say, with the touch command) will be masked with this number. 022 hence excludes write permissions of group and of other. A umask of 006 would exclude read and write permissions of other, but would allow read and write of group. Try
umask touch <file1> ls -al <file1> umask 026 touch <file2> ls -al <file2>

026 is probably closer to the kind of mask we like as an ordinary user. Check your /etc/profile le to see what umask your login defaults to, when, and also why. 125 126

15.1. Soft Links

15. Symbolic and Hard Links

Notice that the ls -al listing has the letter l on the far left next to myfile2, and the usual - next to myfile. This indicates that the le is a soft link (also known as a symbolic link or symlink) to some other le. A symbolic link contains no data of its own, only a reference to another le. It can even contain a reference to a directory. In either case, programs operating on the link will actually see the le or directory it points to. Try

Chapter 15

Symbolic and Hard Links

Very often, a le is required to be in two different directories at the same time. Think for example of a conguration le that is required by two different software packages that are looking for the le in different directories. The le could simply be copied, but to have to replicate changes in more than one place would create an administrative nightmare. Also consider a document that must be present in many directories, but which would be easier to update at one point. The way two (or more) les can have the same data is with links.

mkdir mydir ln -s mydir mydir2 ls -al . touch ./mydir/file1 touch ./mydir2/file2 ls -al ./mydir ls -al ./mydir2

The directory mydir2 is a symbolic link to mydir2 and appears as though it is a replica of the original. Once again the directory mydir2 does not consume additional disk spacea program that reads from the link is unaware that it is seeing into a different directory. Symbolic links can also be copied and retain their value:

cp mydir2 / ls -al / cd /mydir2

15.1 Soft Links

To demonstrate a soft link, try the following:
touch myfile ln -s myfile myfile2 ls -al cat > myfile a few lines of text D cat myfile cat myfile2

You have now copied the link to the root directory. However, the link points to a relative path mydir in the same directory as the link. Since there is no mydir here, an error is raised. Try

rm -f mydir2 /mydir2 ln -s pwd/mydir mydir2 ls -al

Now you will see mydir2 has an absolute path. You can try
cp mydir2 / ls -al / cd /mydir2

and notice that it now works. One of the common uses of symbolic links is to make mounted (see Section 19.4) le systems accessible from a different directory. For instance, you may have a large 128

127

15. Symbolic and Hard Links

15.2. Hard Links

15. Symbolic and Hard Links

directory that has to be split over several physical disks. For clarity, you can mount the disks as /disk1, /disk2, etc., and then link the various subdirectories in a way that makes efcient use of the space you have. Another example is the linking of /dev/cdrom to, say, /dev/hdc so that programs accessing the device le /dev/cdrom (see Chapter 18) actually access the correct IDE drive.

15.2 Hard Links

U NIX allows the data of a le to have more than one name in separate places in the same le system. Such a le with more than one name for the same data is called a hard-linked le and is similar to a symbolic link. Try
touch mydata ln mydata mydataB ls -al

The les mydata and mydataB are indistinguishable. They share the same data, and have a 2 in second column of the ls -al listing. This means that they are hard-linked twice (that there are two names for this le). The reason why hard links are sometimes used in preference to symbolic links is that some programs are not fooled by a symbolic link: If you have, say, a script that uses cp to copy a le, it will copy the symbolic link instead of the le it points to. cp actually has an option to override this behavior. A hard link, however, will always be seen as a real le.

On the other hand, hard links cannot be made between les on different le systems nor can they be made between directories.

129

130

16. Pre-installed Documentation

/usr/share/texmf/doc

/usr/share/lout/doc

Kernel documentation: /usr/src/linux/Documentation/

Chapter 16

This directory contains information on all hardware drivers except graphic cards. The kernel has built-in drivers for networking cards, SCSI controllers, sound cards, and so on. If you need to nd out if one of these is supported, this is the rst place to look. X Window System graphics hardware support: /usr/X11R6/lib/X11/doc/ (This is the same as /usr/X11R6/doc/.) In this directory you will nd documentation on all of the graphics hardware supported by , how to congure , tweak video modes, cope with incompatible graphics cards, and so on. See Section 43.5 for details.

Pre-installed Documentation
This chapter tells you where to nd documentation on a common L INUX distribution. The paths are derived from a RedHat distribution, but are no less applicable to other distributions, although the exact locations might be different. One difference between distributions is the migration of documentation source from /usr/???? to /usr/share/????the proper place for themon account of their being shareable between different machines. See Chapter 35 for the reason documentation goes where it does. In many cases, documentation may not be installed or may be in completely different locations. Unfortunately, I cannot keep track of what the 20 major vendors are doing, so it is likely that this chapter will quickly become out of date. For many proprietary operating systems, the denitive reference for their operating system is printed texts. For L INUX , much of documentation is written by the authors themselves and is included with the source code. A typical L INUX distribution will package documentation along with the compiled binaries. Common distributions come with hundreds of megabytes of printable, hyperlinked, and plain text documentation. There is often no need to go the the World Wide Web unless something is outdated. If you have not already tried this, run

TEX and Meta-Font reference: /usr/share/texmf/doc/ This directory has an enormous and comprehensive reference to the TEX typesetting language and the Meta-Font font generation package. It is not, however, an exhaustive reference.
A L TEX HTML documentation: /usr/share/texmf/doc/latex/latex2e-html/ A This directory contains a large reference to the L TEX typesetting language. (This book A TEX.) itself was typeset using L

HOWTOs: /usr/doc/HOWTO or /usr/share/doc/HOWTO HOWTOs are an excellent source of layman tutorials for setting up almost any kind of service you can imagine. RedHat seems to no longer ship this documentation with their base set of packages. It is worth listing the contents here to emphasize diversity of topics covered. These are mirrored all over the Internet, so you should have no problem nding them from a search engine (in particular, from http://www.linuxdoc.org/):
3Dfx-HOWTO AX25-HOWTO Access-HOWTO Alpha-HOWTO Assembly-HOWTO Bash-Prompt-HOWTO Benchmarking-HOWTO Beowulf-HOWTO BootPrompt-HOWTO Bootdisk-HOWTO Busmouse-HOWTO Finnish-HOWTO Firewall-HOWTO French-HOWTO Ftape-HOWTO GCC-HOWTO German-HOWTO Glibc2-HOWTO HAM-HOWTO Hardware-HOWTO Hebrew-HOWTO INDEX.html Modem-HOWTO Multi-Disk-HOWTO Multicast-HOWTO NET-3-HOWTO NFS-HOWTO NIS-HOWTO Networking-Overview-HOWTO Optical-Disk-HOWTO Oracle-HOWTO PCI-HOWTO PCMCIA-HOWTO Security-HOWTO Serial-HOWTO Serial-Programming-HOWTO Shadow-Password-HOWTO Slovenian-HOWTO Software-Release-Practice-HOWTO Sound-HOWTO Sound-Playing-HOWTO Spanish-HOWTO TeTeX-HOWTO Text-Terminal-HOWTO

ls -ld /usr//doc /usr///doc /usr/share///doc \ /opt//doc /opt///doc

This is a somewhat unreliable way to search for potential documentation directories, but it gives at least the following list of directories for an ofcial RedHat 7.0 with a complete set of installed packages:
/usr/X11R6/doc /usr/lib/X11/doc /usr/local/doc /usr/share/vim/vim57/doc /usr/share/doc /usr/share/gphoto/doc

131

132

16. Pre-installed Documentation

CD-Writing-HOWTO CDROM-HOWTO COPYRIGHT Chinese-HOWTO Commercial-HOWTO Config-HOWTO Consultants-HOWTO Cyrillic-HOWTO DNS-HOWTO DOS-Win-to-Linux-HOWTO DOS-to-Linux-HOWTO DOSEMU-HOWTO Danish-HOWTO Distribution-HOWTO ELF-HOWTO Emacspeak-HOWTO Esperanto-HOWTO Ethernet-HOWTO

INFO-SHEET IPCHAINS-HOWTO IPX-HOWTO IR-HOWTO ISP-Hookup-HOWTO Installation-HOWTO Intranet-Server-HOWTO Italian-HOWTO Java-CGI-HOWTO Kernel-HOWTO Keyboard-and-Console-HOWTO KickStart-HOWTO LinuxDoc+Emacs+Ispell-HOWTO META-FAQ MGR-HOWTO MILO-HOWTO MIPS-HOWTO Mail-HOWTO

PPP-HOWTO PalmOS-HOWTO Parallel-Processing-HOWTO Pilot-HOWTO Plug-and-Play-HOWTO Polish-HOWTO Portuguese-HOWTO PostgreSQL-HOWTO Printing-HOWTO Printing-Usage-HOWTO Quake-HOWTO README RPM-HOWTO Reading-List-HOWTO Root-RAID-HOWTO SCSI-Programming-HOWTO SMB-HOWTO SRM-HOWTO

Thai-HOWTO Tips-HOWTO UMSDOS-HOWTO UPS-HOWTO UUCP-HOWTO Unix-Internet-Fundamentals-HOWTO User-Group-HOWTO VAR-HOWTO VME-HOWTO VMS-to-Linux-HOWTO Virtual-Services-HOWTO WWW-HOWTO WWW-mSQL-HOWTO XFree86-HOWTO XFree86-Video-Timings-HOWTO XWindow-User-HOWTO

Apache reference: /home/httpd/html/manual or /var/www/html/manual Apache keeps this reference material online, so that it is the default web page shown when you install Apache for the rst time. Apache is the most popular web server. Manual pages: /usr/man/ or /usr/share/man/ Manual pages were discussed in Section 4.7. Other directory superstructures (see page 137) may contain man pageson some other U NIX systems man pages are littered everywhere. To convert a man page to PostScript (for printing or viewing), use, for example (for the cp command),
groff -Tps -mandoc /usr/man/man1/cp.1 > cp.ps ; gv cp.ps groff -Tps -mandoc /usr/share/man/man1/cp.1 > cp.ps ; gv cp.ps

Mini HOWTOs: /usr/doc/HOWTO/mini or /usr/share/doc/HOWTO/mini These are smaller quick-start tutorials in the same vein (also available from http://www.linuxdoc.org/):
3-Button-Mouse ADSL ADSM-Backup AI-Alife Advocacy Alsa-sound Apache+SSL+PHP+fp Automount Backup-With-MSDOS Battery-Powered Boca BogoMips Bridge Bridge+Firewall Bzip2 Cable-Modem Cipe+Masq Clock Coffee Colour-ls Cyrus-IMAP DHCP DHCPcd DPT-Hardware-RAID Diald Diskless Ext2fs-Undeletion Fax-Server Firewall-Piercing GIS-GRASS GTEK-BBS-550 Hard-Disk-Upgrade INDEX INDEX.html IO-Port-Programming IP-Alias IP-Masquerade IP-Subnetworking ISP-Connectivity Install-From-ZIP Kerneld LBX LILO Large-Disk Leased-Line Linux+DOS+Win95+OS2 Linux+FreeBSD Linux+FreeBSD-mini-HOWTO Linux+NT-Loader Linux+Win95 Loadlin+Win95 Loopback-Root-FS Mac-Terminal Mail-Queue Mail2News Man-Page Modules Multiboot-with-LILO NCD-X-Terminal NFS-Root NFS-Root-Client Netrom-Node Netscape+Proxy Netstation News-Leafsite Offline-Mailing PLIP Partition Partition-Rescue Path Pre-Installation-Checklist Process-Accounting Proxy-ARP-Subnet Public-Web-Browser Qmail+MH Quota RCS README RPM+Slackware RedHat-CD Remote-Boot Remote-X-Apps SLIP-PPP-Emulator Secure-POP+SSH Sendmail+UUCP Sendmail-Address-Rewrite Small-Memory Software-Building Software-RAID Soundblaster-AWE StarOffice Term-Firewall TkRat Token-Ring Ultra-DMA Update Upgrade VAIO+Linux VPN Vesafb Visual-Bell Windows-Modem-Sharing WordPerfect X-Big-Cursor XFree86-XInside Xterm-Title ZIP-Drive ZIP-Install

info pages: /usr/info/ or /usr/share/info/ Info pages were discussed in Section 4.8. Individual package documentation: /usr/doc/* or /usr/share/doc/* Finally, all packages installed on the system have their own individual documentation directory. A package foo will most probably have a documentation directory /usr/doc/foo (or /usr/share/doc/foo). This directory most often contains documentation released with the sources of the package, such as release information, feature news, example code, or FAQs. If you have a particular interest in a package, you should always scan its directory in /usr/doc (or /usr/share/doc) or, better still, download its source distribution. Below are the /usr/doc (or /usr/share/doc) directories that contained more than a trivial amount of documentation for that package. In some cases, the package had complete references. (For example, the complete Python references were contained nowhere else.)
ImageMagick-5.2.2 LPRng-3.6.24 XFree86-doc-4.0.1 bash-2.04 bind-8.2.2 P5 cdrecord-1.9 cvs-1.10.8 fetchmail-5.5.0 freetype-1.3.1 gawk-3.0.6 gcc-2.96 gcc-c++-2.96 ghostscript-5.50 gimp-1.1.25 glibc-2.1.92 gtk+-1.2.8 gtk+-devel-1.2.8 ipchains-1.3.9 iproute-2.2.4 isdn4k-utils-3.1 krb5-devel-1.2.1 libtiff-devel-3.5.5 libtool-1.3.5 libxml-1.8.9 lilo-21.4.4 lsof-4.47 lynx-2.8.4 ncurses-devel-5.1 nfs-utils-0.1.9.1 openjade-1.3 openssl-0.9.5a pam-0.72 pine-4.21 pmake-2.1.34 pygtk-0.6.6 python-docs-1.5.2 rxvt-2.6.3 sane-1.0.3 sgml-tools-1.0.9 slang-devel-1.4.1 stylesheets-1.54.13rh tin-1.4.4 uucp-1.06.1 vim-common-5.7

L INUX documentation project: /usr/doc/LDP or /usr/share/doc/ldp The LDP projects home page is http://www.linuxdoc.org/. The LDP is a consolidation of HOWTOs, FAQs, several books, man pages, and more. The web site will have anything that is not already installed on your system.

Web documentation: /home/httpd/html or /var/www/html Some packages may install documentation here so that it goes online automatically if your web server is running. (In older distributions, this directory was /home/httpd/html.) 133

134

17.2. U NIX Directory Superstructure

17. Overview of the U NIX Directory Layout

17.2

U NIX Directory Superstructure

The root directory on a U NIX system typically looks like this:

Chapter 17

Overview of the U NIX Directory Layout

Here is an overview of how U NIX directories are structured. This is a simplistic and theoretical overview and not a specication of the L INUX le system. Chapter 35 contains proper details of permitted directories and the kinds of les allowed within them.

drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x dr-xr-xr-x drwxr-xr-x drwxrwxrwt drwxr-xr-x

2 2 7 41 24 4 2 7 80 3 5 25

root root root root root root root root root root root root

2048 1024 35840 4096 1024 3072 12288 1024 0 3072 4096 1024

Aug Sep Aug Sep Sep May Dec Jun Sep Sep Sep May

25 16 26 24 27 19 15 7 16 23 28 29

14:04 10:36 17:08 20:55 11:01 10:05 1998 11:47 10:36 23:41 18:12 10:23

bin boot dev etc home lib lost+found mnt proc sbin tmp usr

The /usr directory typically looks like this:

root root root root root root root root root root root root root root root root 1024 27648 1024 7168 1024 2048 1024 7168 9216 12288 1024 1024 4096 1024 1024 1024 May Sep May Sep Sep May Mar Sep Sep Sep May May May Sep Sep Dec 15 28 13 26 3 15 21 12 7 28 13 13 15 12 4 16 11:49 17:18 16:46 10:55 08:07 10:02 1999 17:06 09:05 17:17 16:21 16:35 10:02 17:07 14:38 1998 X11R6 bin dict doc etc games i386-redhat-linux include info lib libexec man sbin share src var

17.1 Packages
L INUX systems are divided into hundreds of small packages, each performing some logical group of operations. On L INUX , many small, self-contained packages interoperate to give greater functionality than would large, aggregated pieces of software. There is also no clear distinction between what is part of the operating system and what is an applicationevery function is just a package. A software package on a RedHat type system is distributed in a single RedHat Package Manager (RPM) le that has a .rpm extension. On a Debian distribution, the equivalent is a .deb package le, and on the Slackware distribution there are Slackware .tgz les. Each package will unpack as many les, which are placed all over the system. Packages generally do not create major directories but unpack les into existing, wellknown, major directories. Note that on a newly installed system there are no les anywhere that do not belong to some package. 135

drwxr-xr-x 9 root drwxr-xr-x 6 root drwxr-xr-x 2 root drwxr-xr-x 261 root drwxr-xr-x 7 root drwxr-xr-x 2 root drwxr-xr-x 4 root drwxr-xr-x 36 root drwxr-xr-x 2 root drwxr-xr-x 79 root drwxr-xr-x 3 root drwxr-xr-x 15 root drwxr-xr-x 2 root drwxr-xr-x 39 root drwxr-xr-x 3 root drwxr-xr-x 3 root

The /usr/local directory typically looks like this:

3 2 4 2 5 2 9 12 2 15 root root root root root root root root root root root root root root root root root root root root 4096 1024 1024 1024 1024 1024 2048 1024 1024 1024 Sep 27 13:16 bin Feb 6 1996 doc Sep 3 08:07 etc Feb 6 1996 games Aug 21 19:36 include Sep 7 09:08 info Aug 21 19:44 lib Aug 2 1998 man Feb 6 1996 sbin Sep 7 09:08 share

drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x drwxr-xr-x

136

17. Overview of the U NIX Directory Layout

17.2. U NIX Directory Superstructure

17.3. L INUX on a Single Floppy Disk

17. Overview of the U NIX Directory Layout

and the /usr/X11R6 directory also looks similar. What is apparent here is that all these directories contain a similar set of subdirectories. This set of subdirectories is called a directory superstructure or superstructure. To my knowledge this is a new term not
previously used by U NIX administrators.

share Shared data. Architecture-independent les. Files that are independent of the hardware platform go here. This allows them to be shared across different machines, even though those machines may have a different kind of processor altogether. include src header les. These are for development.

The superstructure always contains a bin and lib subdirectory, but almost all others are optional. Each package will install under one of these superstructures, meaning that it will unpack many les into various subdirectories of the superstructure. A RedHat package would always install under the /usr or / superstructure, unless it is a graphical Window System application, which installs under the /usr/X11R6/ superstructure. Some very large applications may install under a /opt/<package-name> superstructure, and homemade packages usually install under the /usr/local/ superstructure (local means specic to this very machine). The directory superstructure under which a package installs is often called the installation prex. Packages almost never install les across different superstructures. Exceptions to this are conguration les which
are mostly stored in /etc/.

source les. These are sources to the kernel or locally built packages.

tmp Temporary les. A convenient place for a running program to create a le for temporary use.

17.3

L INUX on a Single 1.44 Megabyte Floppy Disk

Typically, most of the system is under /usr. This directory can be read-only, since packages should never need to write to this directoryany writing is done under /var or /tmp (/usr/var and /usr/tmp are often just symlinked to /var or /tmp, respectively). The small amount under / that is not part of another superstructure (usually about 40 megabytes) performs essential system administration functions. These are commands needed to bring up or repair the system in the absence of /usr. The list of superstructure subdirectories and their descriptions is as follows: bin Binary executables. Usually all bin directories are in the PATH environment variable so that the shell will search all these directories for binaries. sbin Superuser binary executables. These are programs for system administration only. Only the root will have these executables in their PATH. lib Libraries. All other data needed by programs goes in here. Most packages have their own subdirectory under lib to store data les into. Dynamically Linked Libraries (DLLs or .so les.) Executable program code shared by more than one program in the bin directory to save disk space and memory. are stored directly in lib.

You can get L INUX to run on a 1.44 megabyte oppy disk if you trim all unneeded les off an old Slackware distribution with a 2.0.3x kernel. You can compile a small 2.0.3x kernel to about 400 kilobytes (compressed) (see Chapter 42). A le system can be reduced to 23 megabytes of absolute essentials and when compressed will t into 1 megabyte. If the total is under 1.44 megabytes, then you have your L INUX on one oppy. The le list might be as follows (includes all links):
/bin /bin/sh /bin/cat /bin/chmod /bin/chown /bin/cp /bin/pwd /bin/dd /bin/df /bin/du /bin/free /bin/gunzip /bin/gzip /bin/hostname /bin/login /bin/ls /bin/mkdir /bin/mv /bin/ps /bin/rm /bin/stty /bin/su /bin/sync /bin/zcat /bin/dircolors /bin/mount /bin/umount /bin/bash /bin/domainname /bin/head /bin/kill /bin/tar /bin/cut /bin/uname /bin/ping /bin/ln /bin/ash /etc /etc/default /etc/fstab /etc/group /etc/host.conf /etc/hosts /etc/inittab /etc/issue /etc/utmp /etc/networks /etc/passwd /etc/profile /etc/protocols /etc/rc.d /etc/rc.d/rc.0 /etc/rc.d/rc.K /etc/rc.d/rc.M /etc/rc.d/rc.S /etc/rc.d/rc.inet1 /etc/rc.d/rc.6 /etc/rc.d/rc.4 /etc/rc.d/rc.inet2 /etc/resolv.conf /etc/services /etc/termcap /etc/motd /etc/magic /etc/DIR COLORS /etc/HOSTNAME /etc/mtools /etc/ld.so.cache /etc/psdevtab /etc/mtab /etc/fastboot /lib /lib/ld.so /lib/libc.so.5 /lib/ld-linux.so.1 /lib/libcurses.so.1 /lib/libc.so.5.3.12 /lib/libtermcap.so.2.0.8 /lib/libtermcap.so.2 /lib/libext2fs.so.2.3 /lib/libcom err.so.2 /lib/libcom err.so.2.0 /lib/libext2fs.so.2 /lib/libm.so.5.0.5 /lib/libm.so.5 /lib/cpp /usr /usr/adm /usr/bin /usr/bin/less /usr/bin/more /usr/bin/sleep /usr/bin/reset /usr/bin/zless /usr/bin/file /usr/bin/fdformat /usr/bin/strings /usr/bin/zgrep /usr/bin/nc /usr/bin/which /usr/bin/grep /usr/sbin /usr/sbin/showmount /usr/sbin/chroot /usr/spool /usr/tmp /sbin /sbin/e2fsck /sbin/fdisk /sbin/fsck /sbin/ifconfig /sbin/iflink /sbin/ifsetup /sbin/init /sbin/mke2fs /sbin/mkfs /sbin/mkfs.minix /sbin/mklost+found /sbin/mkswap /sbin/mount /sbin/route /sbin/shutdown /sbin/swapoff /sbin/swapon /sbin/telinit /sbin/umount /sbin/agetty /sbin/update /sbin/reboot /sbin/netcfg /sbin/killall5 /sbin/fsck.minix /sbin/halt /sbin/badblocks /sbin/kerneld /sbin/fsck.ext2 /var /var/adm /var/adm/utmp /var/adm/cron /var/spool /var/spool/uucp /var/spool/uucp/SYSLOG /var/spool/uucp/ERRLOG /var/spool/locks /var/tmp /var/run /var/run/utmp /home/user /mnt /proc /tmp /dev/<various-devices>

etc Et cetera. Conguration les. var Variable data. Data les that are continually being re-created or updated. doc Documentation. This directory is discussed in Chapter 16. man Manual pages. This directory is discussed in Chapter 16. info Info pages. This directory is discussed in Chapter 16. 137

Note that the etc directory differs from that of a RedHat distribution. The system startup les /etc/rc.d are greatly simplied under Slackware. 138

17. Overview of the U NIX Directory Layout

17.3. L INUX on a Single Floppy Disk

17. Overview of the U NIX Directory Layout

The /lib/modules directory has been stripped for the creation of this oppy. /lib/modules/2.0.36 would contain dynamically loadable kernel drivers (modules). Instead, all needed drivers are compiled into the kernel for simplicity (explained in Chapter 42). At some point, try creating a single oppy distribution as an exercise. This task should be most instructive to a serious system administrator. At the very least, you should look through all of the commands in the bin directories and the sbin directories above and browse through the man pages of any that are unfamiliar. The preceding le system comes from the morecram-1.3 package available from http://rute.sourceforge.net/morecram-1.3.tar.gz. It can be downloaded to provide a useful rescue and setup disk. Note that there are many such rescue disks available which are more current than morecram.

139

140

18.2. Block and Character Devices

18. U NIX Devices

If no programs are currently using your mouse, you can also try:

Chapter 18

If you now move the mouse, the mouse protocol commands will be written directly to your screen (it will look like garbage). This is an easy way to see if your mouse is working, and is especially useful for testing a serial port. Occasionally this test doesnt work because some command has previously congured the serial port in some odd way. In that case, also try:
cu -s 1200 -l /dev/mouse

cat /dev/mouse

U NIX Devices
U NIX was designed to allow transparent access to hardware devices across all CPU architectures. U NIX also supports the philosophy that all devices be accessible using the same set of command-line utilities.

At a lower level, programs that access device les do so in two basic ways: They read and write to the device to send and retrieve bulk data (much like less and cat above). They use the ioctl (IO Control) function to congure the device. (In the case of the sound card, this might set mono versus stereo, recording speed, or other parameters.) Because every kind of device that one can think of (except for network cards) can be twisted to t these two modes of operation, U NIXs scheme has endured since its inception and is the universal method of accessing hardware.

18.1 Device Files

U NIX has a beautifully consistent method of allowing programs to access hardware. Under U NIX, every piece of hardware is a le. To demonstrate this novelty, try viewing the le /dev/hda (you will have to be root to run this command): /dev/hda is not really a le at all. When you read from it, you are actually reading directly from the rst physical hard disk of your machine. /dev/hda is known as a device le, and all of them are stored under the /dev directory. Device les allow access to hardware. If you have a sound card installed and congured, you can try: Say something into your microphone and then type:
cat my_recording > /dev/dsp cat /dev/dsp > my_recording less -f /dev/hda

18.2

Block and Character Devices

Hardware devices can generally be categorized into random access devices like disk and tape drives, and serial devices like mouse devices, sound cards, and terminals. Random access devices are usually accessed in large contiguous blocks of data that are stored persistently. They are read from in discrete units (for most disks, 1024 bytes at a time). These are known as block devices. Running an ls -l /dev/hda shows a b on the far left of the listing, which means that your hard disk is a block device:
brw-r----1 root disk 3, 64 Apr 27 1995 /dev/hdb

The system will play out the sound through your speakers. (Note that this does not always work, since the recording volume or the recording speed may not be set correctly.) 141

Serial devices, on the other hand, are accessed one byte at a time. Data can be read or written only once. For example, after a byte has been read from your mouse, the same byte cannot be read by some other program. Serial devices are called character devices and are indicated by a c on the far left of the listing. Your /dev/dsp (Digital Signal Processorthat is, your sound card) device looks like: 142

18. U NIX Devices

18.3. Major and Minor Device Numbers

18.4. Common Device Names

18. U NIX Devices

crw-r--r-1 root sys 14, 3 Jul 18 1994 /dev/dsp

/dev/sd?? (8) sd stands for SCSI disk, the high-end drives mostly used by servers. sda is the rst physical disk probed, and so on. Probing goes by SCSI ID and has a system completely different from that of IDE devices. /dev/sda1 is the rst partition on the rst drive, etc. /dev/ttyS? (4) These are serial devices numbered from 0 up. /dev/ttyS0 is your rst serial port (COM1 under MS-DOS or Windows). If you have a multiport card, these can go to 32, 64, and up. /dev/psaux (10) PS/2 mouse. /dev/mouse A symlink to /dev/ttyS0 or /dev/psaux. Other mouse devices are also supported. /dev/modem A symlink to /dev/ttyS1 or whatever port your modem is on. /dev/cua? (4) Identical to ttyS? but now fallen out of use. /dev/fd? (2) Floppy disk. fd0 is equivalent to your A: drive and fd1 your B: drive. The fd0 and fd1 devices autodetect the format of the oppy disk, but you can explicitly specify a higher density by using a device name like /dev/fd0H1920, which gives you access to 1.88 MB, formatted, 3.5-inch oppies. Other oppy devices are shown in Table 18.1. See Section 19.3.4 on how to format these devices. /dev/par? (6) Parallel port. /dev/par0 is your rst parallel port or LPT1 under DOS. /dev/lp? (6) Line printer. Identical to /dev/par?. /dev/urandom Random number generator. Reading from this device gives pseudorandom numbers. /dev/st? (9) SCSI tape. SCSI backup tape drive. /dev/zero (1) Produces zero bytes, and as many of them as you need. This is useful if you need to generate a block of zeros for some reason. Use dd (see Section 18.5.2) to read a specic number of zeros. /dev/null (1) Null device. Reads nothing. Anything you write to the device is discarded. This is very useful for discarding output. /dev/pd? Parallel port IDE disk. /dev/pcd? Parallel port ATAPI CD-ROM. /dev/pf? Parallel port ATAPI disk. /dev/sr? SCSI CD-ROM. /dev/scd? SCSI CD-ROM (Identical, alternate name). 144

18.3

Major and Minor Device Numbers

Devices are divided into sets called major device numbers. For instance, all SCSI disks are major number 8. Further, each individual device has a minor device number like /dev/sda, which is minor device 0. Major and minor device numbers identify the device to the kernel. The le name of the device is arbitrary and is chosen for convenience and consistency. You can see the major and minor device number (8, 0) in the ls listing for /dev/sda:
brw-rw---1 root disk 8, 0 May 5 1998 /dev/sda

18.4 Common Device Names

A list of common devices and their descriptions follows. The major numbers are shown in parentheses. The complete reference for devices is the le /usr/src/linux/Documentation/devices.txt. /dev/hd?? hd stands for hard disk, but refers here only to IDE devicesthat is, common hard disks. The rst letter after the hd dictates the physical disk drive: /dev/hda (3) First drive, or primary master. /dev/hdb (3) Second drive, or primary slave. /dev/hdc (22) Third drive, or secondary master. /dev/hdd (22) Fourth drive, or secondary slave. When accessing any of these devices (with, say, less /dev/hda), you would be reading raw from the actual physical disk starting at the rst sector of the rst track, sequentially, until the last sector of the last track. Partitions
typical disk might have 2 to 10 partitions. Each partition acts as a whole disk on its own, giving the effect of having more than one disk. For instance, you might have Windows installed on one partition and L INUX installed on another. More details come in Chapter 19. are named /dev/hda1,

With all operating systems, disk drives are divided into sections called partitions. A

/dev/hda2, etc., indicating the rst, second, etc., partition on physical drive a. 143

18. U NIX Devices

18.4. Common Device Names

18. U NIX Devices

Table 18.1 Floppy device names Floppy devices are named /dev/fdlmnnnn l 0 A: drive 1 B: drive m d double density 360 KB or 5.25 inch h high density 1.2 MB or 5.25 inch q quad density 5.25 inch D double density 720 KB or 3.5 inch H high density 1.44 MB or 3.5 inch E Extra density 3.5 inch. u Any 3.5-inch oppy. Note that u now replaces D, H, and E, thus leaving it up to the user to decide if the oppy has enough density for the format. nnnn 360 410 420 720 The size of the format. With D, H, and E, 3.5800 820 830 880 inch oppies have devices only for the sizes 1040 1120 1200 that are likely to work. For instance, there is no 1440 1476 1494 /dev/fd0D1440 because double density disks 1600 1680 1722 wont manage 1440 KB. /dev/fd0H1440 and 1743 1760 1840 /dev/fd0H1920 are probably the ones you 1920 2880 3200 are most interested in. 3520 3840

Recommended links It is recommended that these links exist on all systems: /dev/core /dev/ramdisk /dev/ftape /dev/bttv0 /dev/radio /dev/i2o* /dev/scd? /proc/kcore ram0 qft0 video0 radio0 /dev/i2o/* sr? symbolic symbolic symbolic symbolic symbolic symbolic hard Backward compatibility Backward compatibility Backward compatibility Backward compatibility Backward compatibility Backward compatibility Alternate SCSI CD-ROM name

Locally dened links The following links may be established locally to conform to the conguration of the system. This is merely a tabulation of existing practice, and does not constitute a recommendation. However, if they exist, they should have the following uses: /dev/mouse /dev/tape /dev/cdrom /dev/cdwriter /dev/scanner /dev/modem /dev/root /dev/swap mouse port tape device CD-ROM device CD-writer scanner modem port root device swap device symbolic symbolic symbolic symbolic symbolic symbolic symbolic symbolic Current mouse device Current tape device Current CD-ROM device Current CD-writer device Current scanner device Current dialout device Current root le system Current swap device

/dev/sg? SCSI generic. This is a general-purpose SCSI command interface for devices like scanners. /dev/fb? (29) Frame buffer. This represents the kernels attempt at a graphics driver. /dev/cdrom A symlink to /dev/hda, /dev/hdb, or /dev/hdc. It can also be linked to your SCSI CD-ROM. /dev/ttyI? ISDN modems. /dev/tty? (4) Virtual console. This is the terminal device for the virtual console itself and is numbered /dev/tty1 through /dev/tty63. /dev/tty?? (3) and /dev/pty?? (2) Other TTY devices used for emulating a terminal. These are called pseudo-TTYs and are identied by two lowercase letters and numbers, such as ttyq3. To nondevelopers, these are mostly of theoretical interest. The le /usr/src/linux/Documentation/devices.txt also has this to say (quoted verbatim): 145

/dev/modem should not be used for a modem which supports dial-in as well as dialout, as it tends to cause lock le problems. If it exists, /dev/modem should point to the appropriate primary TTY device (the use of the alternate callout devices is deprecated). For SCSI devices, /dev/tape and /dev/cdrom should point to the cooked devices (/dev/st* and /dev/sr*, respectively), whereas /dev/cdwriter and /dev/scanner should point to the appropriate generic SCSI devices (/dev/sg*). /dev/mouse may point to a primary serial TTY device, a hardware mouse device, or a socket for a mouse driver program (e.g. /dev/gpmdata).

Sockets and pipes Non-transient sockets and named pipes may exist in /dev. Common entries are: /dev/printer /dev/log /dev/gpmdata socket socket socket lpd local socket syslog local socket mouse multiplexer

146

18. U NIX Devices

18.5. dd, tar, and Tricks with Block Devices

18. U NIX Devices

18.5

dd, tar, and Tricks with Block Devices

will write zeros to the rst 10 megabytes of your rst IDE drive. This will wipe out the partition table as well as any le system information and give you a brand new disk. To zero a oppy disk is just as easy:

dd probably originally stood for disk dump. It is actually just like cat except it can read and write in discrete blocks. It essentially reads and writes between devices while converting the data in some way. It is generally used in one of these ways:
dd if=<in-file> of=<out-file> [bs=<block-size>] \ [count=<number-of-blocks>] [seek=<output-offset>] \ [skip=<input-offset>]
5

dd if=/dev/zero of=/dev/fd0 bs=1024 count=1440

dd if=<in-file> [bs=<block-size>] [count=<number-of-blocks>] \ [skip=<input-offset>] > <outfile> dd of=<out-file> [bs=<block-size>] [count=<number-of-blocks>] \ [seek=<output-offset>] < <infile>

Even writing zeros to a oppy may not be sufcient. Specialized equipment can probably still read magnetic media after it has been erased several times. If, however, you write random bits to the oppy, it becomes completely impossible to determine what was on it:
mknod /dev/urandom c 1 9 for i in 1 2 3 4 ; do dd if=/dev/urandom of=/dev/fd0 bs=1024 count=1440 done

To use dd, you must specify an input le and an output le with the if= and of= options. If the of= option is omitted, then dd writes to stdout. If the if= option is omitted, then dd reads from stdin. If you are confused, remember that dd thinks of in and out
with respect to itself.

18.5.3 Identifying data on raw disks

Here is a nice trick to nd out something about a hard drive: gives x86 boot sector.
dd if=/dev/hda1 count=1 bs=512 | file -

Note that dd is an unforgiving and destructive command that should be used with caution.

18.5.1

Creating boot disks from boot images

To discover what a oppy disk is, try

To create a new RedHat boot oppy, nd the boot.img le on ftp.redhat.com, and with a new oppy, run:
dd if=boot.img of=/dev/fd0

which gives x86 boot sector, system )k?/bIHC, FAT (12 bit) for DOS oppies.

dd if=/dev/fd0 count=1 bs=512 | file -

This command writes the raw disk image directly to the oppy disk. All distributions will have similar disk images for creating installation oppies (and sometimes rescue oppies).

18.5.4 Duplicating a disk

If you have two IDE drives that are of identical size, and provided that you are sure they contain no bad sectors and provided neither are mounted, you can run to copy the entire disk and avoid having to install an operating system from scratch. It doesnt matter what is on the original (Windows, L INUX , or whatever) since each sector is identically duplicated; the new system will work perfectly. (If they are not the same size, you will have to use tar or mirrordir to replicate the le system exactly.) 148
dd if=/dev/hdc of=/dev/hdd

18.5.2

Erasing disks

If you have ever tried to repartition a L INUX disk back into a DOS/Windows disk, you will know that DOS/Windows FDISK has bugs in it that prevent it from recreating the partition table. A quick
dd if=/dev/zero of=/dev/hda bs=1024 count=10240

147

18. U NIX Devices

18.5. dd, tar, and Tricks with Block Devices

18.6. Creating Devices with mknod and /dev/MAKEDEV

18. U NIX Devices

18.5.5

Backing up to oppies

You can use tar to back up to any device. Consider periodic backups to an ordinary IDE drive instead of a tape. Here we back up to the secondary slave:
tar -cvzf /dev/hdd /bin /boot /dev /etc /home /lib /sbin /usr /var

le from a security point of view. It is often used when a le is required for some feature in a conguration script, and you would like the particular feature disabled. For instance, specifying the users shell to /dev/null inside the password le will certainly prevent insecure use of a shell, and is an explicit way of saying that that account does not allow shell logins. You can also use /dev/null to create a le containing nothing:

tar can also back up across multiple oppy disks:

tar -cvMf /dev/fd0 /home/simon

or alternatively, to create a le containing only zeros. Try

dd if=/dev/zero bs=1024 count=<number-of-kilobytes> > myfile

cat /dev/null > myfile

18.5.6

Tape backups

tar traditionally backs up onto tape drives. The commands

mt -f /dev/st0 rewind tar -cvf /dev/st0 /home

18.6

Creating Devices with mknod and /dev/MAKEDEV

Although all devices are listed in the /dev directory, you can create a device anywhere in the le system by using the mknod command:
mknod [-m <mode>] <file-name> [b|c] <major-number> <minor-number>

rewind scsi tape 0 and archive the /home directory onto it. You should not try to use compression with tape drives because they are error prone, and a single error could make the entire archive unrecoverable. The mt command stands for magnetic tape and controls generic SCSI tape devices. See also mt(1).

The letters b and c are for creating a block or character device, respectively. To demonstrate, try

18.5.7

Hiding program output, creating blocks of zeros

If you dont want to see any program output, just append > /dev/null to the command. For example, we arent often interested in the output of make. make is discussed later. Here we absorb everything save for error messages.

my-floppy can be used just like /dev/fd0 Note carefully the mode (i.e., the permissions) of /dev/fd0. /dev/fd0 should be readable and writable only to root and to users belonging to the floppy group, since we obviously dont want an arbitrary user to be able to log in (remotely) and overwrite a oppy disk. In fact, this is the reason for having devices represented as les in the rst place. U NIX les naturally support group access control, and therefore so do devices. To create devices that are missing from your /dev directory (some esoteric devices will not be present by default), simply look up the devices major and minor number in /usr/src/linux/Documentation/devices.txt and use the mknod command. This procedure is, however, somewhat tedious, and the script /dev/MAKEDEV is usually available for convenience. You must be in the /dev directory before you run this script. 150

mknod -m 0600 /my-floppy b 2 0 ls -al /dev/fd0 /my-floppy

make > /dev/null

Then, of course, we can absorb all output including error messages with either or
make >& /dev/null

make > /dev/null 2>&1

The device /dev/null nds innumerable uses in shell scripting to suppress the output of a command or to feed a command dummy (empty) input. /dev/null is a safe 149

18. U NIX Devices

18.6. Creating Devices with mknod and /dev/MAKEDEV

18. U NIX Devices

Typical usage of MAKEDEV is

cd /dev ./MAKEDEV -v fd0 ./MAKEDEV -v fd1

to create a complete set of oppy disk devices. The man page for MAKEDEV contains more details. In particular, it states:
Note that programs giving the error ENOENT: No such le or directory normally means that the device le is missing, whereas ENODEV: No such device normally means the kernel does not have the driver congured or loaded.

151

152

19.1. The Physical Disk Structure

19. Partitions, File Systems, Formatting, Mounting

Partition

Sector Cylinder Side 0

Chapter 19

Side 1 Side 2

Partitions, File Systems, Formatting, Mounting

19.1 The Physical Disk Structure
Physical disks are divided into partitions. See /dev/hd?? under Section 18.4. Information as to how the disk is partitioned up is stored in a partition table, which is a small area of the disk separate from the partitions themselves.

Side 3 Side 4

Side 5

Figure 19.1 Hard drive platters and sector layout

19.1.2 Large Block Addressing

The system above is quite straightforward except for the curious limitation that partition tables have only 10 bits in which to store the partitions cylinder offset. This means that no disk can have more than 1024 cylinders. This limitation was overcome by multiplying up the number of heads in software to reduce the number of cylinders, Called LBA (Large Block Addressing) mode. hence portraying a disk of impossible proportions. The user, however, need never be concerned that the physical disk is completely otherwise.

19.1.1

Cylinders, heads, and sectors

The physical drive itself usually comprises several actual disks of which both sides are used. The sides are labelled 0, 1, 2, 3, and so on, and are also called heads because one magnetic head per side does the actual reading and writing. Each side/head has tracks, and each track is divided into segments called sectors. Each sector typically holds 512 bytes. The total amount of space on the drive in bytes is therefore: 512 (sectors-per-track) (tracks-per-side) (number-of-sides)

19.1.3 Extended partitions

The partition table has room for only four partitions. For more partitions, one of these four partitions can be divided into many smaller partitions, called logical partitions. The original four are then called primary partitions. If a primary partition is subdivided in this way, it is known as an extended primary or extended partition. Typically, the rst primary partition will be small (/dev/hda1, say). The second primary partition will ll the rest of the disk as an extended partition (/dev/hda2, say). In this case, the entries in the partition table of /dev/hda3 and /dev/hda4 will be blank. The 154

A single track and all the tracks of the same diameter (on all the sides) are called a cylinder. Disks are normally talked about in terms of cylinders and sectors instead of sides, tracks, and sectors. Partitions are (usually) divided along cylinder boundaries. Hence, disks do not have arbitrarily sized partitions; rather, the size of the partition is usually a multiple of the amount of data held in a single cylinder. Partitions therefore have a denite inner and outer diameter. Figure 19.1 illustrates the layout of a hard disk. 153

19. Partitions, File Systems, Formatting, Mounting

19.2. Partitioning a New Disk

19. Partitions, File Systems, Formatting, Mounting

extended partition can be subdivided repeatedly to give /dev/hda5, /dev/hda6, and so on.

Table 19.1 Which directories should have their own partitions, and their partitions sizes
Directory swap Size (Megabytes) Why? Twice the size of your RAM This is where memory is drawn from when you run out. The swap partition gives programs the impression that you have more RAM than you actually do, by swapping data in and out of this partition. Swap partitions cannot be over 128 MB, but you can have many of them. This limitation has been removed in newer kernels. Disk access is obviously slow compared to direct RAM, but when a lot of idle programs are running, swapping to disk allows more real RAM for needy programs. This directory need not be on a different partition to your / partition (below). Whatever you choose, there must be no chance that a le under /boot could span sectors that are over the 1024 cylinder boundary (i.e., outside of the rst 500 megabytes of your hard drive). This is why /boot (or /) is often made the rst primary partition of the hard drive. If this requirment is not met, you get the famous LI prompt on a nonbooting system. See Section 31.2.4. Here is variable data, like log les, mail spool les, database les, and your web proxy cache (web cache and databases may need to be much bigger though). For newer distributions, this directory also contains any local data that this site serves (like FTP les or web pages). If you are going to be using a web cache, either store the stuff in a separate partition/disk or make your /var partition huge. Also, log les can grow to enormous sizes when there are problems. You dont want a full or corrupted /var partition to effect the rest of your disk. This is why it goes in its own partition. Here is temporary data. Programs access this frequently and need it to be fast. It goes in a separate partition because programs really need to create a temporary le sometimes, and this should not be affected by other partitions becoming full. This partition is also more likely to be corrupted. Here is your distribution (Debian , RedHat, Mandrake, etc.). It can be mounted readonly. If you have a disk whose write access can physically be disabled (like some SCSI drives), then you can put /usr on a separate drive. Doing so will make for a much more secure system. Since /usr is stock standard, this is the partition you can most afford to lose. Note however that /usr/local/ may be important to youpossibly link this elsewhere. Here are your users home directories. For older distributions, this directory also contains any local data that this site serves (like FTP les or web pages). Anything not in any of the other directories is directly under your / directory. These are the /bin (5MB), (possibly) /boot (3MB), /dev (0.1MB), /etc (4MB), /lib (20MB), /mnt (0MB), /proc (0MB), and /sbin (4MB) directories. They are essential for the system to start up and contain minimal utilities for recovering the other partitions in an emergency. As stated above, if the /boot directory is in a separate partition, then / must be below the 1024 cylinder boundary (i.e., within the rst 500 megabytes of your hard drive).

19.2 Partitioning a New Disk

A new disk has no partition information. Typing fdisk will start an interactive partitioning utility. The command fdisks your primary master. What follows is an example of the partitioning of a new hard drive. Most distributions these days have a simpler graphical system for creating partitions, so using fdisk will not be necessary at installation time. However, adding a new drive or transferring/copying a L INUX system to new hardware will require partitioning. On U NIX, each partition has its own directory. Files under one directory might be stored on a different disk or a different partition to les in another directory. Typically, the /var directory (and all subdirectories beneath it) is stored on a different partition from the /usr directory (and all subdirectories beneath it). Table 19.2 offers a general guideline as to how a server machine should be set up (with home computers, you can be far more liberalmost home PCs can do with merely a swap and / partition.). When you install a new server, your distribution should allow you to customize your partitions to match this table. If another operating system is already installed in the rst partition, you can type p and might see:
Command (m for help): p Disk /dev/hda: 255 heads, 63 sectors, 788 cylinders Units = cylinders of 16065 * 512 bytes
5

fdisk /dev/hda

/boot

510

/var

1001000

/tmp

/usr

5001500

/home /

Remainder of disk 50100

Device Boot /dev/hda1

Start 1

End 312

Blocks 2506108+

Id c

System Win95 FAT32 (LBA)

In such a case, you can just start adding further partitions. The exact same procedure applies in the case of SCSI drives. The only difference is that /dev/hd? changes to /dev/sd?. (See Chapter 42 for SCSI device driver information.) Here is a partitioning session with fdisk:
5

Building a new DOS disklabel. Changes will remain in memory only, until you decide to write them. After that, of course, the previous content wont be recoverable.

First, we use the p option to print current partitions

Command (m for help): p Disk /dev/hda: 255 heads, 63 sectors, 788 cylinders Units = cylinders of 16065 * 512 bytes

[root@cericon /root]# fdisk /dev/hda Device contains neither a valid DOS partition table, nor Sun or SGI disklabel

Device Boot

Start

End

Blocks

System

155

156

19. Partitions, File Systems, Formatting, Mounting

19.2. Partitioning a New Disk

19. Partitions, File Systems, Formatting, Mounting

of which there are clearly none. Now n lets us add a new partition:
Command (m for help): n Command action e extended p primary partition (1-4) p

Command (m for help): n Command action l logical (5 or over) p primary partition (1-4) l First cylinder (34-788, default 34): 34 Last cylinder or +size or +sizeM or +sizeK (34-788, default 788): +200M Command (m for help): n Command action l logical (5 or over) p primary partition (1-4) l First cylinder (60-788, default 60): 60 Last cylinder or +size or +sizeM or +sizeK (60-788, default 788): +1500M Command (m for help): n Command action l logical (5 or over) p primary partition (1-4) l First cylinder (252-788, default 252): 252 Last cylinder or +size or +sizeM or +sizeK (252-788, default 788): 788

We want to dene the rst physical partition starting at the rst cylinder:

Partition number (1-4): 1 First cylinder (1-788, default 1): 1

We would like an 80-megabyte partition. fdisk calculates the last cylinder automatically with:
Last cylinder or +size or +sizeM or +sizeK (1-788, default 788): +80M

Our next new partition will span the rest of the disk and will be an extended partition:
Command (m for help): n Command action e extended p primary partition (1-4) e Partition number (1-4): 2 First cylinder (12-788, default 12): 12 Last cylinder or +size or +sizeM or +sizeK (12-788, default 788): 788

The default partition type is a single byte that the operating system will look at to determine what kind of le system is stored there. Entering l lists all known types:
Command (m for help): l 0 Empty [...] 8 AIX 9 AIX bootable [...] 12 Compaq diagnost 14 Hidden FAT16 <3 16 4d 4e 56 5c Hidden FAT16 61 SpeedStor Linux swap Linux BSD/386 a6 db e1 ff OpenBSD CP/M / CTOS / . DOS access BBT

QNX4.x 82 QNX4.x 2nd part 83 Golden Bow Priam Edisk a5

Our remaining logical partitions t within the extended partition:

Command (m for help): n Command action l logical (5 or over) p primary partition (1-4) l First cylinder (12-788, default 12): 12 Last cylinder or +size or +sizeM or +sizeK (12-788, default 788): +64M Command (m for help): n Command action l logical (5 or over) p primary partition (1-4) l First cylinder (21-788, default 21): 21 Last cylinder or +size or +sizeM or +sizeK (21-788, default 788): +100M

fdisk will set the type to Linux by default. We only need to explicitly set the type of the swap partition:
Command (m for help): t Partition number (1-9): 5 Hex code (type L to list codes): 82 Changed system type of partition 5 to 82 (Linux swap)

Now we need to set the bootable ag on the rst partition, since BIOSs will not boot a disk without at least one bootable partition:
Command (m for help): a

157

158

19. Partitions, File Systems, Formatting, Mounting

19.2. Partitioning a New Disk

19.3. Formatting Devices

19. Partitions, File Systems, Formatting, Mounting

Partition number (1-10): 1

19.3

Formatting Devices

19.3.1 File systems

Displaying our results gives:
Command (m for help): p Disk /dev/hda: 255 heads, 63 sectors, 788 cylinders Units = cylinders of 16065 * 512 bytes
5

Disk drives are usually read in blocks of 1024 bytes (two sectors). From the point of view of anyone accessing the device, blocks are stored consecutivelythere is no need to think about cylinders or headsso that any program can read the disk as though it were a linear tape. Try
less /dev/hda1 less -f /dev/hda1

Device Boot /dev/hda1 * /dev/hda2 /dev/hda5 /dev/hda6 /dev/hda7 /dev/hda8 /dev/hda9

Start 1 12 12 21 34 60 252

End 11 788 20 33 59 251 788

Blocks 88326 6241252+ 72261 104391 208813+ 1542208+ 4313421

Id 83 5 82 83 83 83 83

System Linux Extended Linux swap Linux Linux Linux Linux

Now a complex directory structure with many les of arbitrary size needs to be stored in this contiguous partition. This poses the problem of what to do with a le that gets deleted and leaves a data hole in the partition, or a le that has to be split into parts because there is no single contiguous space big enough to hold it. Files also have to be indexed in such a way that they can be found quickly (consider that there can easily be 10,000 les on a system). U NIXs symbolic/hard links and devices les also have to be stored. To cope with this complexity, operating systems have a format for storing les called the le system (fs). Like MS-DOS with its FAT le system or Windows with its FAT32 le system, L INUX has a le system called the 2nd extended le system, or ext2. Whereas ext2 is the traditional native L INUX le system, three other native le systems have recently become available: SGIs XFS le system, the ext3fs le system, and the reiserfs le system. These three support fast and reliable recovery in the event of a power failure, using a feature called journaling. A journaling le system prewrites disk alterations to a separate log to facilitate recovery if the le system reaches an incoherent state. (See Section 19.5.)

At this point, nothing has been committed to disk. We write it as follows (Note: this step is irreversible):
Command (m for help): w The partition table has been altered! Calling ioctl() to re-read partition table. Syncing disks. WARNING: If you have created or modified any DOS 6.x partitions, please see the fdisk manual page for additional information.

19.3.2
Even having written the partition, fdisk may give a warning that the kernel does not know about the new partitions. This happens if the disk is already in use. In this case, you will need to reboot. For the above partition, the kernel will give the following information at boot time:
Partition check: hda: hda1 hda2 < hda5 hda6 hda7 hda8 hda9 >

mke2fs

To create a le system on a blank partition, use the command mkfs (or one of its variants). To create a L INUX ext2 le system on the rst partition of the primary master run: or, alternatively
mkfs -t ext2 -c /dev/hda1

The < . . . > shows that partition hda2 is extended and is subdivided into ve smaller partitions. 159

The -c option means to check for bad blocks by reading through the entire disk rst. 160

mke2fs -c /dev/hda1

19. Partitions, File Systems, Formatting, Mounting

19.3. Formatting Devices

19. Partitions, File Systems, Formatting, Mounting

This is a read-only check and causes unreadable blocks to be agged as such and not be used. To do a full read-write check, use the badblocks command. This command writes to and veries every bit in that partition. Although the -c option should always be used on a new disk, doing a full read-write test is probably pedantic. For the above partition, this test would be:
badblocks -o blocks-list.txt -s -w /dev/hda1 88326 mke2fs -l blocks-list.txt /dev/hda1

19.3.4 Creating MS-DOS oppies

Accessing les on MS-DOS/Windows oppies is explained in Section 4.16. The command mformat A: will format a oppy, but this command merely initializes the le system; it does not check for bad blocks or do the low-level formatting necessary to reformat oppies to odd storage sizes. formats a oppy in any way that you like. A more common (but less thorough) command is fdformat from the util-linux package. It veries that each track is working properly and compensates for variations between the mechanics of different oppy drives. To format a 3.5-inch 1440-KB, 1680-KB, or 1920-KB oppy, respectively, run:
nd this package on the Internet. See Chapter 24 for how to compile and install source packages.

A command, called superformat, from the fdutils package

You may have to

After running mke2fs, we will nd that

dd if=/dev/hda1 count=8 bs=1024 | file -

gives Linux/i386 ext2 filesystem.

19.3.3

Formatting oppies and removable drives

cd /dev ./MAKEDEV -v fd0 superformat /dev/fd0H1440 superformat /dev/fd0H1690 superformat /dev/fd0H1920

New kinds of removable devices are being released all the time. Whatever the device, the same formatting procedure is used. Most are IDE compatible, which means you can access them through /dev/hd?. The following examples are a parallel port IDE disk drive, a parallel port ATAPI CD-ROM drive, a parallel port ATAPI disk drive, and your A: oppy drive, respectively:
mke2fs mke2fs mke2fs mke2fs -c -c -c -c /dev/pda1 /dev/pcd0 /dev/pf0 /dev/fd0

Note that these are long le name oppies (VFAT), not old 13-characterlename MS-DOS oppies. Most users would have only ever used a 3.5-inch oppy as a 1.44 MB oppy. In fact, the disk media and magnetic head can write much more densely than this specication, allowing 24 sectors per track to be stored instead of the usual 18. This is why there is more than one device le for the same drive. Some inferior disks will, however, give errors when trying to format that denselysuperformat will show errors when this happens. See Table 18.1 on page 145 for the naming conventions of oppy devices, and their many respective formats.

Actually, using an ext2 le system on a oppy drive wastes a lot of space. Rather, use an MS-DOS le system, which has less overhead and can be read by anyone (see Section 19.3.4). You often will not want to be bothered with partitioning a device that is only going to have one partition anyway. In this case, you can use the whole disk as one partition. An example is a removable IDE drive as a primary slave LS120 disks and Jazz drives as well as removable IDE brackets are commercial examples. :

19.3.5

mkswap, swapon, and swapoff

The mkswap command formats a partition to be used as a swap device. For our disk, -c has the same meaning as previouslyto check for bad blocks.
mkswap -c /dev/hda5

mke2fs -c /dev/hdb

161

Once the partition is formatted, the kernel can be signalled to use that partition as a swap partition with
swapon /dev/hda5

162

19. Partitions, File Systems, Formatting, Mounting

19.4. Device Mounting

19. Partitions, File Systems, Formatting, Mounting

and to stop usage,

swapoff /dev/hda5

Swap partitions cannot be larger than 128 MB, although you can have as many of them as you like. You can swapon many different partitions simultaneously.

will notice that it is no longer empty, but contains the CD-ROMs les. What is happening is that the kernel is redirecting all lookups from the directory /mnt/cdrom to read from the CD-ROM disk. You can browse around these les as though they were already copied onto your hard drive. This is one of the things that makes U NIX cool. When you are nished with the CD-ROM unmount it with

19.4 Device Mounting

The question of how to access les on an arbitrary disk (without C:, D:, etc., notation, of course) is answered here. In U NIX, there is only one root le system that spans many disks. Different directories may actually exist on a different physical disk. To bind a directory to a physical device (like a partition or a CD-ROM) so that the devices le system can be read is called mounting the device. The mount command is used as follows:

umount /dev/hdb eject /dev/hdb

19.4.2 Mounting oppy disks

Instead of using mtools, you could mount the oppy disk with
mkdir /mnt/floppy mount -t vfat /dev/fd0 /mnt/floppy

or, for older MS-DOS oppies, use

mkdir /mnt/floppy mount -t msdos /dev/fd0 /mnt/floppy

mount [-t <fstype>] [-o <option>] <device> <directory> umount [-f] [<device>|<directory>]

Before you eject the oppy, it is essential to run

umount /dev/fd0

The -t option species the kind of le system, and can often be omitted since L INUX can autodetect most le systems. <fstype> can be one of adfs, affs, autofs, coda, coherent, devpts, efs, ext2, hfs, hpfs, iso9660, minix, msdos, ncpfs, nfs, ntfs, proc, qnx4, romfs, smbfs, sysv, ufs, umsdos, vfat, xenix, or xiafs. The most common le systems are discussed below. The -o option is not usually used. See mount(8) for all possible options.

in order that cached data is committed to the disk. Failing to umount a oppy before ejecting will probably corrupt its le system.

19.4.3 Mounting Windows and NT partitions 19.4.1 Mounting CD-ROMs

Mounting a Windows partition can also be done with the vfat le system, and NT partitions (read-only) with the ntfs le system. VAT32 is also supported (and autodetected). For example,
mkdir mount mkdir mount /windows -t vfat /dev/hda1 /windows /nt -t ntfs /dev/hda2 /nt

Put your distribution CD-ROM disk into your CD-ROM drive and mount it with
ls /mnt/cdrom mount -t iso9660 -o ro /dev/hdb /mnt/cdrom

(Your CD-ROM might be /dev/hdc or /dev/hdd, howeverin this case you should make a soft link /dev/cdrom pointing to the correct device. Your distribution may also prefer /cdrom over /mnt/cdrom.) Now cd to your /mnt/cdrom directory. You 163

164

19. Partitions, File Systems, Formatting, Mounting

19.5. File System Repair: fsck

19.7. Automatic Mounts: fstab

19. Partitions, File Systems, Formatting, Mounting

19.5 File System Repair: fsck

fsck stands for le system check. fsck scans the le system, reporting and xing errors. Errors would normally occur only if the kernel halted before the le system was umounted. In this case, it may have been in the middle of a write operation which left the le system in an incoherent state. This usually happens because of a power failure. The le system is then said to be unclean. fsck is used as follows:
5

19.7

Automatic Mounts: fstab

Manual mounts are explained above for new and removable disks. It is, of course necessary for le systems to be automatically mounted at boot time. What gets mounted and how is specied in the conguration le /etc/fstab. /etc/fstab will usually look something like this for the disk we partitioned above:
/dev/hda1 /dev/hda6 /dev/hda7 /dev/hda8 /dev/hda9 /dev/hda5 /dev/fd0 /dev/cdrom none none / /tmp /var /usr /home swap /mnt/floppy /mnt/cdrom /proc /dev/pts ext2 ext2 ext2 ext2 ext2 swap auto iso9660 proc devpts defaults defaults defaults defaults defaults defaults noauto,user noauto,ro,user defaults mode=0622 1 1 1 1 1 0 0 0 0 0 1 2 2 2 2 0 0 0 0 0

fsck [-V] [-a] [-t <fstype>] <device>

-V means to produce verbose output. -a means to check the le system noninteractivelymeaning to not ask the user before trying to make any repairs. Here is what you would normally do with L INUX lot about the ext2 le system:
fsck -a -t ext2 /dev/hda1

if you dont know a whole

although you can omit the -t option because L INUX autodetects the le system. Note that you should not run fsck on a mounted le system. In exceptional circumstances it is permissible to run fsck on a le system that has been mounted read-only. fsck actually just runs a program specic to that le system. In the case of ext2, the command e2fsck (also known as fsck.ext2) is run. See e2fsck(8) for exhaustive details. During an interactive check (without the -a option, or with the -r option the default), various questions may be asked of you, as regards xing and saving things. Its best to save stuff if you arent sure; it will be placed in the lost+found directory below the root directory of the particular device. In the example system further below, there would exist the directories /lost+found, /home/lost+found, /var/lost+found, /usr/lost+found, etc. After doing a check on, say, /dev/hda9, list the /home/lost+found directory and delete what you think you dont need. These will usually be temporary les and log les (les that change often). Its rare to lose important les because of an unclean shutdown.

For the moment we are interested in the rst six lines only. The rst three elds (columns) dictate the partition, the directory where it is to be mounted, and the le system type, respectively. The fourth eld gives options (the -o option to mount). The fth eld tells whether the le system contains real les. The eld is used by the dump command to decide if it should be backed up. This is not commonly used. The last eld tells the order in which an fsck should be done on the partitions. The / partition should come rst with a 1, and all other partitions should come directly after. Placing 2s everywhere else ensures that partitions on different disks can be checked in parallel, which speeds things up slightly at boot time. The floppy and cdrom entries enable you to use an abbreviated form of the mount command. mount will just look up the corresponding directory and le system type from /etc/fstab. Try These entries also have the user option, which allows ordinary users to mount these devices. The ro option once again tells to mount the CD-ROM read only, and the noauto command tells mount not to mount these le systems at boot time. (More comes further below.) proc is a kernel information database that looks like a le system. For example /proc/cpuinfo is not any kind of le that actually exists on a disk somewhere. Try cat /proc/cpuinfo. Many programs use /proc to get dynamic information on the status and conguration of your machine. More on this is discussed in Section 42.4. 166
mount /dev/cdrom

19.6 File System Errors on Boot

Just read Section 19.5 again and run fsck on the le system that reported the error. 165

19. Partitions, File Systems, Formatting, Mounting

19.8. Manually Mounting /proc

19.10. Remounting

19. Partitions, File Systems, Formatting, Mounting

The devpts le system is another pseudo le system that generates terminal master/slave pairs for programs. This is mostly of concern to developers.

umount /mnt losetup -d /dev/loop0

19.8 Manually Mounting /proc

You can mount the proc le system with the command
mount -t proc /proc /proc

To dump the le system to a oppy, run

dd if=/file-floppy of=/dev/fd0 count=1440 bs=1024

A similar procedure for RAM devices is

dd if=/dev/zero of=/dev/ram0 count=1440 bs=1024 mke2fs /dev/ram0 mkdir /mnt mount /dev/ram0 /mnt ls -al /mnt

This is an exception to the normal mount usage. Note that all common L INUX installations require /proc to be mounted at boot time. The only times you will need this command are for manual startup or when doing a chroot. (See page 178.)
5

19.9 RAM and Loopback Devices

A RAM device is a block device that can be used as a disk but really points to a physical area of RAM. A loopback device is a block device that can be used as a disk but really points to an ordinary le somewhere. If your imagination isnt already running wild, consider creating a oppy disk with le system, les and all, without actually having a oppy disk, and being able to dump this creation to oppy at any time with dd. You can also have a whole other L INUX system inside a 500 MB le on a Windows partition and boot into itthus obviating having to repartition a Windows machine just to run L INUX . All this can be done with loopback and RAM devices.

When you are nished copying the les that you want into /mnt, merely run

umount /mnt

To dump the le system to a oppy or le, respectively, run:

dd if=/dev/ram0 of=/dev/fd0 count=1440 bs=1024 dd if=/dev/ram0 of=/file-floppy count=1440 bs=1024

19.9.2 CD-ROM les

Another trick is to move your CD-ROM to a le for high-speed access. Here, we use a shortcut instead of the losetup command:
dd if=/dev/cdrom of=some_name.iso mount -t iso9660 -o ro,loop=/dev/loop0 some_name.iso /cdrom

19.9.1

Formatting a oppy inside a le

The operations are quite trivial. To create an ext2 oppy inside a 1440 KB le, run:
dd if=/dev/zero of=/file-floppy count=1440 bs=1024 losetup /dev/loop0 /file-floppy mke2fs /dev/loop0 mkdir /mnt mount /dev/loop0 /mnt ls -al /mnt

19.10

Remounting from Read-Only to Read-Write

When you are nished copying the les that you want into /mnt, merely run 167

A le system that is already mounted as read-only can be remounted as read-write, for example, with 168

19. Partitions, File Systems, Formatting, Mounting

19.11. Disk sync

19. Partitions, File Systems, Formatting, Mounting

mount -o rw,remount /dev/hda1 /

This command is useful when you log in in single-user mode with no write access to your root partition.

19.11 Disk sync

The kernel caches write operations in memory for performance reasons. These ush (physically commit to the magnetic media) every so often, but you sometimes want to force a ush. This is done simply with
sync

169

170

20.2. Special Parameters: $?, $*,. . .

20. Advanced Shell Scripting

20.2

Special Parameters: $?, $*,. . .

Chapter 20

An ordinary variable can be expanded with $V ARNAME. Commonly used variables like PATH and special variables like PWD and RANDOM were covered in Chapter 9. Further special expansions are documented in the following section, quoted verbatim from the bash man page (the footnotes are mine).1
Special Parameters The shell treats several parameters specially. referenced; assignment to them is not allowed. These parameters may only be

Advanced Shell Scripting

This chapter completes our discussion of sh shell scripting begun in Chapter 7 and expanded on in Chapter 9. These three chapters represent almost everything you can do with the bash shell.

$* Expands to the positional parameters (i.e., the command-line arguments passed to the shell script, with $1 being the rst argument, $2 the second etc.), starting from one. When the expansion occurs within double quotes, it expands to a single word with the value of each parameter separated by the rst character of the IFS special variable. That is, $* is equivalent to $1c$2c..., where c is the rst character of the value of the IFS variable. If IFS is unset, the parameters are separated by spaces. If IFS is null, the parameters are joined without intervening separators. $@ Expands to the positional parameters, starting from one. When the expansion occurs within double quotes, each parameter expands to a separate word. That is, $@ is equivalent to $1 $2 ... When there are no positional parameters, $@ and $@ expand to nothing (i.e., they are removed). Hint: this
is very useful for writing wrapper shell scripts that just add one argument.

20.1 Lists of Commands

The special operator && and || can be used to execute functions in sequence. For instance:
grep harry: /etc/passwd || useradd harry

$# Expands to the number of positional parameters in decimal (i.e. the number of command-line arguments). $? Expands to the status of the most recently executed foreground pipeline.
the exit code of the last command.

I.e.,

The || means to only execute the second command if the rst command returns an error. In the above case, grep will return an exit code of 1 if harry is not in the /etc/passwd le, causing useradd to be executed. An alternate representation is
grep -v harry: /etc/passwd && useradd harry

$- Expands to the current option ags as specied upon invocation, by the set builtin command, or those set by the shell itself (such as the -i option).

$$ Expands to the process ID of the shell. In a () subshell, it expands to the process ID of the current shell, not the subshell. $! Expands to the process ID of the most recently executed background (asynI.e., after executing a background command with comchronous) command.
mand &, the variable $! will give its process ID.

where the -v option inverts the sense of matching of grep. && has the opposite meaning to ||, that is, to execute the second command only if the rst succeeds. Adept script writers often string together many commands to create the most succinct representation of an operation:
grep -v harry: /etc/passwd && useradd harry || \ echo "date: useradd failed" >> /var/log/my_special_log

$0 Expands to the name of the shell or shell script. This is set at shell initialization. If bash is invoked with a le of commands, $0 is set to the name of that le. If bash is started with the -c option, then $0 is set to the rst argument after the string to be executed, if one is present. Otherwise, it is set to the le name used to invoke bash, as given by argument zero. Note that basename $0 is a
useful way to get the name of the current command without the leading path.
1 Thanks

to Brian Fox and Chet Ramey for this material.

171

172

20. Advanced Shell Scripting

20.3. Expansion

20. Advanced Shell Scripting

$- At shell startup, set to the absolute le name of the shell or shell script being executed as passed in the argument list. Subsequently, expands to the last argument to the previous command, after expansion. Also set to the full le name of each command executed and placed in the environment exported to that command. When checking mail, this parameter holds the name of the mail le currently being checked.

TEXT=scripting_for_phun echo ${TEXT:10:3} echo ${TEXT:10}

${#VAR} Gives the length of $V AR. ${!PRE*} Gives a list of all variables whose names begin with PRE. ${VAR#pattern} $V AR is returned with the glob expression pattern removed from the leading part of the string. For instance, ${TEXT#scr} in the above example will return ripting for phun. ${VAR##pattern} This is the same as the previous expansion except that if pattern contains wild cards, then it will try to match the maximum length of characters. ${VAR%pattern} The same as ${V AR#pattern} except that characters are removed from the trailing part of the string. ${VAR%%pattern} The same as ${V AR##pattern} except that characters are removed from the trailing part of the string. ${VAR/search/replace} $V AR is returned with the rst occurrence of the string search replaced with replace. ${VAR/#search/replace} Same as ${V AR/search/replace} except that the match is attempted from the leading part of $V AR. ${VAR/%search/replace} Same as ${V AR/search/replace} except that the match is attempted at the trailing part of $V AR. ${VAR//search/replace} Same as ${V AR/search/replace} except that all instances of search are replaced. Backquote expansion We have already shown backquote expansion in 7.12. Note that the additional notation $(command) is equivalent to command except that escapes (i.e., \) are not required for special characters. Arithmetic expansion We have already shown arithmetic expansion on page 62. Note that the additional notation $((expression)) is equivalent to $[expression]. Finally The last modications to the command-line are the splitting of the commandline into words according to the white space between them. The IFS (Internal Field Separator) environment variable determines what characters delimit command-line words (usually whitespace). With the command-line divided into words, path names are expanded according to glob wild cards. Consult bash(1) for a comprehensive description of the pattern matching options that most people dont know about. 174

20.3 Expansion
Expansion refers to the way bash modies the command-line before executing it. bash performs several textual modications to the command-line, proceeding in the following order: Brace expansion We have already shown how you can use, for example, the shorthand touch file {one,two,three}.txt to create multiple les file one.txt, file two.txt, and file three.txt. This is known as brace expansion and occurs before any other kind of modication to the command-line. Tilde expansion The special character is replaced with the full path contained in the HOME environment variable or the home directory of the users login (if $HOME is null). + is replaced with the current working directory and - is replaced with the most recent previous working directory. The last two are rarely used. Parameter expansion This refers to expanding anything that begins with a $. Note that $V AR and ${V AR} do exactly the same thing, except in the latter case, V AR can contain non-whole word characters that would normally confuse bash. There are several parameter expansion tricks that you can use to do string manipulation. Most shell programmers never bother with these, probably because they are not well supported by other U NIX systems. ${VAR:-default} This will result in $V AR unless V AR is unset or null, in which case it will result in default. ${VAR:=default} Same as previous except that default is also assigned to VAR if it is empty. ${VAR:-default} This will result in an empty string if V AR is unset or null; otherwise it will result in default. This is the opposite behavior of ${V AR:default}. ${VAR:?message} This will result in $V AR unless V AR is unset or null, in which case an error message containing message is displayed. ${VAR:offset} or ${VAR:n:l} This produces the nth character of $V AR and then the following l characters. If l is not present, then all characters to the right of the nth character are produced. This is useful for splitting up strings. Try: 173

20. Advanced Shell Scripting

20.4. Built-in Commands

20.5. Trapping Signals the trap Command

20. Advanced Shell Scripting

20.4 Built-in Commands

Many commands operate some built-in functionality of bash or are especially interpreted. These do not invoke an executable off the le system. Some of these were described in Chapter 7, and a few more are discussed here. For an exhaustive description, consult bash(1). : A single colon by itself does nothing. It is useful for a no operation line such as:
if <command> ; then : else echo "<command> was unsuccessful" fi

printf format args ... This is like the printf function. It outputs to the terminal like echo but is useful for more complex formatting of output. See printf(3) for details and try printf "%10.3e\n" 12 as an example. pwd Prints the present working directory. set Prints the value of all environment variables. See also Section 20.6 on the set command. source lename args ... Reads lename into the current current shell environment. This is useful for executing a shell script when environment variables set by that script must be preserved. times Prints the accumulated user and system times for the shell and for processes run from the shell. type command Tells whether command is an alias, a built-in or a system executable. ulimit Prints and sets various user resource limits like memory usage limits and CPU limits. See bash(1) for details. umask See Section 14.2. unset VAR Deletes a variable or environment variable. unset -f func Deletes a function. wait Pauses until all background jobs have completed. wait PID Pauses until background process with process ID of PID has exited, then returns the exit code of the background process. wait %job Same with respect to a job spec.

. lename args ... A single dot is the same as the source command. See below. alias command=value Creates a pseudonym for a command. Try:
alias necho="echo -n" necho "hello"

Some distributions alias the mv, cp, and rm commands to the same pseudonym with the -i (interactive) option set. This prevents les from being deleted without prompting, but can be irritating for the administrator. See your /.bashrc le for these settings. See also unalias. unalias command Removes an alias created with alias. alias -p Prints list of aliases. eval arg ... Executes args as a line of shell script. exec command arg ... Begins executing command under the same process ID as the current script. This is most often used for shell scripts that are mere wrapper scripts for real programs. The wrapper script sets any environment variables and then execs the real program binary as its last line. exec should never return. local var=value Assigns a value to a variable. The resulting variable is visible only within the current function. pushd directory and popd These two commands are useful for jumping around directories. pushd can be used instead of cd, but unlike cd, the directory is saved onto a list of directories. At any time, entering popd returns you to the previous directory. This is nice for navigation since it keeps a history of wherever you have been. 175

20.5

Trapping Signals the trap Command

You will often want to make your script perform certain actions in response to a signal. A list of signals can be found on page 86. To trap a signal, create a function and then use the trap command to bind the function to the signal.
#!/bin/sh function on_hangup () { echo Hangup (SIGHUP) signal recieved }

176

20. Advanced Shell Scripting

20.6. Internal Settings the set Command

20.7. Useful Scripts and Commands

20. Advanced Shell Scripting

trap on_hangup SIGHUP

set -n Read commands without executing them. This command is useful for syntax checking. set -o posix Comply exactly with the POSIX 1003.2 standard. set -u Report an error when trying to reference a variable that is unset. Usually bash just lls in an empty string. set -v Print each line of script as it is executed. set -x Display each command expansion as it is executed. set -C Do not overwrite existing les when using >. You can use >| to force overwriting.

while true ; do sleep 1 done

Run the above script and then send the process ID the -HUP signal to test it. (See Section 9.5.) An important function of a program is to clean up after itself on exit. The special signal EXIT (not really a signal) executes code on exit of the script:
#!/bin/sh function on_exit () { echo I should remove temp files now } trap on_exit EXIT
10

exit 0

20.7

Useful Scripts and Commands

Here is a collection of useful utility scripts that people are always asking for on the mailing lists. See page 517 for several security check scripts.

while true ; do sleep 1 done

20.7.1

chroot

Breaking the above program will cause it to print its own epitaph. If - is given instead of a function name, then the signal is unbound (i.e., set to its default value).

exit 0

The chroot command makes a process think that its root le system is not actually /. For example, on one system I have a complete Debian installation residing under a directory, say, /mnt/debian. I can issue the command
chroot /mnt/debian bash -i

20.6 Internal Settings the set Command

The set command can modify certain behavioral settings of the shell. Your current options can be displayed with echo $-. Various set commands are usually entered at the top of a script or given as command-line options to bash. Using set +option instead of set -option disables the option. Here are a few examples: set -e Exit immediately if any simple command gives an error. set -h Cache the location of commands in your PATH. The shell will become confused if binaries are suddenly inserted into the directories of your PATH, perhaps causing a No such file or directory error. In this case, disable this option or restart your shell. This option is enabled by default. 177

to run the bash shell interactively, under the root le system /mnt/debian. This command will hence run the command /mnt/debian/bin/bash -i. All further commands processed under this shell will have no knowledge of the real root directory, so I can use my Debian installation without having to reboot. All further commands will effectively behave as though they are inside a separate U NIX machine. One caveat: you may have to remount your /proc le system inside your chrootd le system see page 167. This useful for improving security. Insecure network services can change to a different root directoryany corruption will not affect the real system. Most rescue disks have a chroot command. After booting the disk, you can manually mount the le systems on your hard drive, and then issue a chroot to begin using your machine as usual. Note that the command chroot <new-root> without arguments invokes a shell by default. 178

20. Advanced Shell Scripting

20.7. Useful Scripts and Commands

20. Advanced Shell Scripting

20.7.2

if conditionals

cd <old-dir> patch -p1 -s < <patch-file>.diff

The if test ... was used to control program ow in Chapter 7. Bash, however, has a built-in alias for the test function: the left square brace, [. Using [ instead of test adds only elegance:

if [ 5 -le 3 ] ; then echo 5 < 3 fi

which makes <old-dir> identical to <new-dir>. The -p1 option strips the leading directory name from the patch le. The presence of a leading directory name in the patch le often confuses the patch command.

20.7.4 Internet connectivity test

You may want to leave this example until you have covered more networking theory. The acid test for an Internet connection is a successful DNS query. You can use ping to test whether a server is up, but some networks lter ICMP messages and ping does not check that your DNS is working. dig sends a single UDP packet similar to ping. Unfortunately, it takes rather long to time out, so we fudge in a kill after 2 seconds. This script blocks until it successfully queries a remote name server. Typically, the next few lines of following script would run fetchmail and a mail server queue ush, or possibly uucico. Do set the name server IP to something appropriate like that of your local ISP; and increase the 2 second time out if your name server typically takes longer to respond.
MY_DNS_SERVER=197.22.201.154 while true ; do ( dig @$MY_DNS_SERVER netscape.com IN A & DIG_PID=$! { sleep 2 ; kill $DIG_PID ; } & sleep 1 wait $DIG_PID ) 2>/dev/null | grep -q [;]*netscape.com && break done

It is important at this point to realize that the if command understands nothing of arithmetic. It merely executes a command test (or in this case [) and tests the exit code. If the exit code is zero, then the command is considered to be successful and if proceeds with the body of the if statement block. The onus is on the test command to properly evaluate the expression given to it. if can equally well be used with any command:

conditionally adds /usr/local/bin if grep does not nd it in your PATH.

if echo "$PATH" | grep -qwv /usr/local/bin ; then export PATH="$PATH:/usr/local/bin" fi

20.7.3

patching and diffing

You may often want to nd the differences between two les, for example to see what changes have been made to a le between versions. Or, when a large batch of source code may have been updated, it is silly to download the entire directory tree if there have been only a few small changes. You would want a list of alterations instead. The diff utility dumps the lines that differ between two les. It can be used as follows: You can also use diff to see difference netween two directory trees. diff recursively compares all corresponding les: The output is known as a patch le against a directory tree, that can be used both to see changes, and to bring <old-dir> up to date with <new-dir>. Patch les may also end in .patch and are often gzipped. The patch le can be applied to <old-dir> with 179
diff -u --recursive --new-file <old-dir> <new-dir> > <patch-file>.diff diff -u <old-file> <new-file>

20.7.5 Recursive grep (search)

Recursively searching through a directory tree can be done easily with the find and xargs commands. You should consult both these man pages. The following command pipe searches through the kernel source for anything about the pcnet Ethernet card, printing also the line number: 180

20. Advanced Shell Scripting

20.7. Useful Scripts and Commands

20. Advanced Shell Scripting

find /usr/src/linux -follow -type f | xargs grep -iHn pcnet

(You will notice how this command returns rather a lot of data. However, going through it carefully can be quite instructive.) Limiting a search to a certain le extension is just another common use of this pipe sequence.
find /usr/src/linux -follow -type f -name *.[ch] | xargs grep -iHn pcnet

Note that new versions of grep also have a -r option to recursively search through directories.
40

D="$?" if [ "$D" = "0" ] ; then if diff -q $T $i >/dev/null ; then : else if [ "$VERBOSE" = "-v" ] ; then echo $i fi cat $T > $i fi rm -f $T fi done else find . -type f -name "$1" | xargs $0 $VERBOSE "$S" "$R"

20.7.6

Recursive search and replace

Often you will want to perform a search-and-replace throughout all the les in an entire source tree. A typical example is the changing of a function call name throughout lots of source. The following script is a must for any /usr/local/bin/. Notice the way it recursively calls itself.
#!/bin/sh N=basename $0
5

20.7.7

cut and awk manipulating text le elds

The cut command is useful for slicing les into elds; try
cut -d: -f1 /etc/passwd cat /etc/passwd | cut -d: -f1

if [ "$1" = "-v" ] ; then VERBOSE="-v" shift fi if [ "$3" = "" -o "$1" = "-h" -o "$1" = "--help" ] ; then echo "$N: Usage" echo " $N [-h|--help] [-v] <regexp-search> \ <regexp-replace> <glob-file>" echo exit 0 fi S="$1" ; shift ; R="$1" ; shift T=$$replc

The awk program is an interpreter for a complete programming language call AWK. A common use for awk is in eld stripping. It is slightly more exible than cut
cat /etc/passwd | awk -F : {print $1}

especially where whitespace gets in the way,

ls -al | awk {print $6 " " $7 " " $8} ls -al | awk {print $5 " bytes"}

which isolates the time and size of the le respectively. Get your nonlocal IP addresses with:

ifconfig | grep inet addr: | fgrep -v 127.0.0. | \ cut -d: -f2 | cut -d -f1

if echo "$1" | grep -q / ; then for i in "$@" ; do SEARCH=echo "$S" | sed s,/,\\\\/,g REPLACE=echo "$R" | sed s,/,\\\\/,g cat $i | sed "s/$SEARCH/$REPLACE/g" > $T

Reverse an IP address with:

echo 192.168.3.2 | awk -F . {print $4 "." $3 "." $2 "." $1 }

181

182

20. Advanced Shell Scripting

20.7. Useful Scripts and Commands

20. Advanced Shell Scripting

Print all common user names (i.e., users with UID values greater than 499 on RedHat and greater than 999 on Debian ):
awk -F: $3 >= 500 {print $1} /etc/passwd ( awk -F: $3 >= 1000 {print $1} /etc/passwd )

Incidentally, the above nicely compresses high-resolution pcx lespossibly the A output of a scanning operation, or a L TEX compilation into PostScript rendered with GhostScript (i.e. gs -sDEVICE=pcx256 -sOutputFile=page%d.pcx le.ps).

20.7.10 Securely erasing les

Removing a le with rm only unlinks the le name from the data. The le blocks may still be on disk, and will only be reclaimed when the le system reuses that data. To erase a le proper, requires writing random bytes into the disk blocks occupied by the le. The following overwrites all the les in the current directory:
for i in * ; do dd if=/dev/urandom \ of="$i" \ bs=1024 \ count=expr 1 + \ \stat "$i" | grep Size: | awk {print $2}\ / 1024 done

20.7.8

Calculations with bc

Scripts can easily use bc to do calculations that expr cant handle. For example, convert to decimal with
echo -e ibase=16;FFFF | bc

to binary with
echo -e obase=2;12345 | bc

or work out the SIN of 45 degrees with

pi=echo "scale=10; 4*a(1)" | bc -l echo "scale=10; s(45*$pi/180)" | bc -l

You can then remove the les normally with rm.

20.7.11 Persistent background processes

Consider trying to run a process, say, the rxvt terminal, in the background. This can be done simply with:
rxvt &

20.7.9

Conversion of graphics formats of many les

The convert program of the ImageMagick package is a command many Windows users would love. It can easily be used to convert multiple les from one format to another. Changing a les extension can be done with echo lename | sed e s/\.old$/.new/. The convert command does the rest:
for i in *.pcx ; do CMD="convert -quality 625 $i echo $i | sed -e s/\.pcx$/.png/" # Show the command-line to the user: echo $CMD # Execute the command-line: eval $CMD done

However, rxvt still has its output connected to the shell and is a child process of the shell. When a login shell exits, it may take its child processes with it. rxvt may also die of its own accord from trying to read or write to a terminal that does not exist without the parent shell. Now try:
{ rxvt >/dev/null 2>&1 </dev/null & } &

This technique is known as forking twice, and redirecting the terminal to dev null. The shell can know about its child processes but not about the its grand child processes. We have hence create a daemon process proper with the above command. Now, it is easy to create a daemon process that restarts itself if it happens to die. Although such functionality is best accomplished within (which you will get a taste of in Chapter 22), you can make do with: 184

Note that the search-and-replace expansion mechanism could also be used to replace the extensions: ${i/%.pcx/.png} produces the desired result. 183

20. Advanced Shell Scripting

20.7. Useful Scripts and Commands

20.8. Shell Initialization

20. Advanced Shell Scripting

You will notice the effects of all these tricks with:

ps awwwxf { { while true ; do rxvt ; done ; } >/dev/null 2>&1 </dev/null & } &

PID 1 2 3 4 5 6 262 272 341 447 480 506 507 508 509 514 515 524 748 749 5643 5645 25292 11780 11814 15534 15535 TTY ? ? ? ? ? ? ? ? ? ? ? tty1 tty2 tty3 ? ? ? ? ? pts/0 pts/0 pts/6 pts/6 ? ? pts/6 pts/6 STAT S SW SW SW SW SW< S S S S S S S S S S S S S S S S R S S S S TIME 0:05 0:02 0:02 0:00 0:01 0:00 0:02 0:00 0:00 0:00 0:02 0:00 0:00 0:00 0:00 7:04 0:00 0:18 0:08 0:00 0:09 0:02 0:00 0:16 0:00 3:12 6:03 COMMAND init [5] [kflushd] [kupdate] [kpiod] [kswapd] [mdrecoveryd] syslogd -m 0 klogd xinetd -reuse -pidfile /var/run/xinetd.pid crond xfs -droppriv -daemon /sbin/mingetty tty1 /sbin/mingetty tty2 /sbin/mingetty tty3 /usr/bin/gdm -nodaemon \_ /etc/X11/X -auth /var/gdm/:0.Xauth :0 \_ /usr/bin/gdm -nodaemon \_ /opt/icewm/bin/icewm \_ rxvt -bg black -cr green -fg whi | \_ bash | \_ mc | \_ bash -rcfile .bashrc | \_ ps awwxf \_ /usr/lib/netscape/netscape-commu \_ (dns helper) cooledit -I /root/.cedit/projects/Rute \_ aspell -a -a

20.7.12

Processing the process list

The following command uses the custom format option of ps to print every conceivable attribute of a process:
ps -awwwxo %cpu,%mem,alarm,args,blocked,bsdstart,bsdtime,c,caught,cmd,comm,\ command,cputime,drs,dsiz,egid,egroup,eip,esp,etime,euid,euser,f,fgid,fgroup,\ flag,flags,fname,fsgid,fsgroup,fsuid,fsuser,fuid,fuser,gid,group,ignored,\ intpri,lim,longtname,lstart,m_drs,m_trs,maj_flt,majflt,min_flt,minflt,ni,\ nice,nwchan,opri,pagein,pcpu,pending,pgid,pgrp,pid,pmem,ppid,pri,rgid,rgroup,\ rss,rssize,rsz,ruid,ruser,s,sess,session,sgi_p,sgi_rss,sgid,sgroup,sid,sig,\ sig_block,sig_catch,sig_ignore,sig_pend,sigcatch,sigignore,sigmask,stackp,\ start,start_stack,start_time,stat,state,stime,suid,suser,svgid,svgroup,svuid,\ svuser,sz,time,timeout,tmout,tname,tpgid,trs,trss,tsiz,tt,tty,tty4,tty8,ucomm,\ uid,uid_hack,uname,user,vsize,vsz,wchan

The output is best piped to a le and viewed with a nonwrapping text editor. More interestingly, the awk command can print the process ID of a process with which prints all the processes having httpd in the command name or command-line. This lter is useful for killing netscape as follows: (Note that the [a] in the regular expression prevents grep from nding itself in the process list.)
ps ps ps ps ps kill -9 ps awx | grep netsc[a]pe | awk {print $1} ps awwx | grep -w htt[p]d | awk {print $1}

The u option shows the useful user format, and the others show virtual memory, signal and long format.

20.8

Shell Initialization

Other useful ps variations are:

awwxf awwxl awwxv awwxu awwxs

Here I will briey discuss what initialization takes place after logging in and how to modify it. The interactive shell invoked after login will be the shell specied in the last eld of the users entry in the /etc/passwd le. The login program will invoke the shell after authenticating the user, placing a - in front of the the command name, which indicates to the shell that it is a login shell, meaning that it reads and execute several scripts to initialize the environment. In the case of bash, the les it reads are: /etc/profile, /.bash profile, /.bash login and /.profile, in that order. In addition, an interactive shell that is not a login shell also reads /.bashrc. Note that traditional sh shells only read /etc/profile and /.profile. 186

The f option is most useful for showing parent-child relationships. It stands for forest, and shows the full process tree. For example, here I am running an desktop with two windows: 185

20. Advanced Shell Scripting

20.9. File Locking

20. Advanced Shell Scripting

20.8.1

Customizing the PATH and LD LIBRARY PATH

also sometimes want to ensure that a program can never be run twice at the same time; this insurance is another use for locking. In the case of a mail folder, if the le is being written to, then no other process should try read it or write to it: and we would like to create a write lock on the le. However if the le is being read from, no other process should try to write to it: and we would like to create a read lock on the le. Write locks are sometimes called exclusive locks; read locks are sometimes called shared locks. Often, exclusive locks are preferred for simplicity. Locking can be implemented by simply creating a temporary le to indicate to other processes to wait before trying some kind of access. U NIX also has some more sophisticated builtin functions.

Administrators can customise things like the environment variables by modifying these startup scripts. Consider the classic case of an installation tree under /opt/. Often, a package like /opt/staroffice/ or /opt/oracle/ will require the PATH and LD LIBRARY PATH variables to be adjusted accordingly. In the case of RedHat, a script,
for i in /opt/*/bin /usr/local/bin ; do test -d $i || continue echo $PATH | grep -wq "$i" && continue PATH=$PATH:$i export PATH done if test id -u -eq 0 ; then for i in /opt/*/sbin /usr/local/sbin ; do test -d $i || continue echo $PATH | grep -wq "$i" && continue PATH=$PATH:$i export PATH done fi for i in /opt/*/lib /usr/local/lib ; do test -d $i || continue echo $LD_LIBRARY_PATH | grep -wq "$i" && continue LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$i export LD_LIBRARY_PATH done

20.9.1 Locking a mailbox le

There are currently four methods of le locking.
research in this area, so this is what I am going on.

The exim sources seem to indicate thorough

1. dot lock le locking. Here, a temporary le is created with the same name as the mail folder and the extension .lock added. So long as this le exists, no program should try to access the folder. This is an exclusive lock only. It is easy to write a shell script to do this kind of le locking. 2. MBX le locking. Similar to 1, but a temporary le is created in /tmp. This is also an exclusive lock. 3. fcntl locking. Databases require areas of a le to be locked. fcntl is a system call to be used inside programs. 4. flock le locking. Same as fcntl, but locks whole les. The following shell function does proper mailbox le locking.
function my_lockfile () { TEMPFILE="$1.$$" LOCKFILE="$1.lock" echo $$ > $TEMPFILE 2>/dev/null || { echo "You dont have permission to access dirname $TEMPFILE" return 1 } ln $TEMPFILE $LOCKFILE 2>/dev/null && { rm -f $TEMPFILE return 0 } STALE_PID=< $LOCKFILE

can be placed as /etc/profile.d/my local.sh with execute permissions. This will take care of anything installed under /opt/ or /usr/local/. For Debian , the script can be inserted directly into /etc/profile. Page 235 of Section 23.3 contains details of exactly what LD LIBRARY PATH is. (Unrelated, but you should also edit your /etc/man.config to add man page paths that appear under all installation trees under /opt/.)
5

20.9 File Locking

Often, one would like a process to have exclusive access to a le. By this we mean that only one process can access the le at any one time. Consider a mail folder: if two processes were to write to the folder simultaneously, it could become corrupted. We 187
10

188

20. Advanced Shell Scripting

20.9. File Locking

20. Advanced Shell Scripting

test "$STALE_PID" -gt "0" >/dev/null || { return 1 } kill -0 $STALE_PID 2>/dev/null && { rm -f $TEMPFILE return 1 } rm $LOCKFILE 2>/dev/null && { echo "Removed stale lock file of process $STALE_PID" } ln $TEMPFILE $LOCKFILE 2>/dev/null && { rm -f $TEMPFILE return 0 } rm -f $TEMPFILE return 1

ously, executed line 4 at the same time. Both would think that the lock did not exist and proceed to line 5. Then both would successfully create the lock lenot what you wanted. The kill command is then useful for checking whether a process is running. Sending the 0 signal does nothing to the process, but the signal fails if the process does not exist. This technique can be used to remove a lock of a process that died before removing the lock itself: that is, a stale lock.

20.9.2 Locking over NFS

The preceding script does not work if your le system is mounted over NFS (network le systemsee Chapter 28). This is obvious because the script relies on the PID of the process, which is not visible across different machines. Not so obvious is that the ln function does not work exactly right over NFSyou need to stat the le and actually check that the link count has increased to 2. The commands lockfile (from the procmail package) and mutt dotlock (from the mutt email reader but perhaps not distributed) do similar le locking. These commands, however, but do not store the PID in the lock le. Hence it is not possible to detect a stale lock le. For example, to search your mailbox, you can run:
lockfile /var/spool/mail/mary.lock grep freddy /var/spool/mail/mary rm -f /var/spool/mail/mary.lock

(Note how instead of cat $LOCKFILE, we use < $LOCKFILE, which is faster.) You can include the above function in scripts that need to lock any kind le. Use the function as follows:
# wait for a lock until my_lockfile /etc/passwd ; do sleep 1 done
5

# The body of the program might go here # [...] # Then to remove the lock, rm -f /etc/passwd.lock

This script is of academic interest only but has a couple of interesting features. Note how the ln function is used to ensure exclusivity. ln is one of the few U NIX functions that is atomic, meaning that only one link of the same name can exist, and its creation excludes the possibility that another program would think that it had successfully created the same link. One might naively expect that the program
function my_lockfile () { LOCKFILE="$1.lock" test -e $LOCKFILE && return 1 touch $LOCKFILE return 0 }

This sequence ensures that you are searching a clean mailbox even if /var is a remote NFS share.

20.9.3 Directory versus le locking

File locking is a headache for the developer. The problem with U NIX is that whereas we are intuitively thinking about locking a le, what we really mean is locking a le name within a directory. File locking per se should only be used on perpetual les, such as database les. For mailbox and passwd les we need directory locking My own term. , meaning the exclusive access of one process to a particular directory entry. In my opinion, lack of such a feature is a serious deciency in U NIX, but because it will require kernel, NFS, and (possibly) library extensions, will probably not come into being any time soon.

is sufcient for le locking. However, consider if two programs, running simultane189

190

20. Advanced Shell Scripting

20.9. File Locking

20. Advanced Shell Scripting

20.9.4

Locking inside C programs

This topic is certainly outside of the scope of this text, except to say that you should consult the source code of reputable packages rather than invent your own locking scheme.

191

192

21.2. Downloading and Installing

21. System Services and lpd

Printing typies the client/server nature of U NIX services. The lpd background process is the server and is initiated by the root user. The remaining commands are client programs, and are run mostly by users.

Chapter 21

21.2

Downloading and Installing

System Services and lpd the Printer Service

This chapter covers a wide range of concepts about the way U NIX services function. Every function of U NIX is provided by one or another package. For instance, mail is often handled by the sendmail or other package, web by the apache package. Here we examine how to obtain, install, and congure a package, using lpd as an example. You can then apply this knowledge to any other package, and later chapters assume that you know these concepts. This discussion will also sufce as an explanation of how to set up and manage printing.

The following discussion should relieve the questions of Where do I get xxx service/package? and How do I install it?. Full coverage of package management comes in Section 24.2, but here you briey see how to use package managers with respect to a real system service. Let us say we know nothing of the service except that it has something to do with a le /usr/sbin/lpd. First, we use our package managers to nd where the le comes from (Debian commands are shown in parentheses):
rpm -qf /usr/sbin/lpd ( dpkg -S /usr/sbin/lpd )

Returns lpr-0.nn-n (for RedHat 6.2, or LPRng-n.n.nn-n on RedHat 7.0, or lpr on Debian ). On RedHat you may have to try this on a different machine because rpm does not know about packages that are not installed. Alternatively, if we would like to see whether a package whose name contains the letters lpr is installed:
rpm -qa | grep -i lpr ( dpkg -l *lpr* )

21.1 Using lpr

Printing under U NIX on a properly congured machine is as simple as typing lpr -Plp <filename> (or cat <filename> | lpr -Plp). The lp in -Plp is the name of the printer queue on the local machine you would like to print to. You can omit it if you are printing to the default (i.e., the rst listed) queue. A queue belongs to a physical printer, so users can predict where paper will come spewing out, by what queue they print to. Queues are conventionally named lp, lp0, lp1, and so on, and any number of them may have been redirected to any other queue on any other machine on the network. The command lprm removes pending jobs from a print queue; lpq reports jobs in progress. The service that facilitates all this is called lpd. The lpr user program makes a network connection to the lpd background process, sending it the print job. lpd then queues, lters, and feeds the job until it appears in the print tray. 193
5

If the package is not present, the package le will be on your CD-ROM and is easily installable with (RedHat 7.0 and Debian in braces):
rpm -i lpr-0.50-4.i386.rpm ( rpm -i LPRng-3.6.24-2 ) ( dpkg -i lpr_0.48-1.deb )

(Much more about package management is covered in Chapter 24.) The list of les which the lpr package is comprises (easily obtained with rpm ql lpr or dpkg -L lpr) is approximately as follows:
/etc/init.d/lpd /etc/cron.weekly/lpr /usr/sbin/lpf /usr/sbin/lpc /usr/sbin/lpd /usr/sbin/pac /usr/bin/lpq /usr/share/man/man1/lprm.1.gz /usr/share/man/man5/printcap.5.gz /usr/share/man/man8/lpc.8.gz /usr/share/man/man8/lpd.8.gz /usr/share/man/man8/pac.8.gz /usr/share/man/man8/lpf.8.gz /usr/share/doc/lpr/README.Debian

194

21. System Services and lpd

21.3. LPRng vs. Legacy lpr-0.nn

21.4. Package Elements

21. System Services and lpd

/usr/bin/lpr /usr/bin/lprm /usr/bin/lptest /usr/share/man/man1/lpr.1.gz /usr/share/man/man1/lptest.1.gz /usr/share/man/man1/lpq.1.gz

/usr/share/doc/lpr/copyright /usr/share/doc/lpr/examples/printcap /usr/share/doc/lpr/changelog.gz /usr/share/doc/lpr/changelog.Debian.gz /var/spool/lpd/lp /var/spool/lpd/remote

21.3

LPRng vs. Legacy lpr-0.nn

maintenance on it themselves. A better example is the lprNG package. Go to The LPRng Web Page http://www.astart.com/lprng/LPRng.html with your web browser. There you can see the authors, mailing lists, and points of download. If a particular package is of much interest to you, then you should become familiar with these resources. Good web pages will also have additional documentation like troubleshooting guides and FAQs (Frequently Asked Questions). Some may even have archives of their mailing lists. Note that some web pages are geared more toward CD vendors who are trying to create their own distribution and so will not have packages for download that beginner users can easily install.

(The word legacy with regard to software means outdated, superseded, obsolete, or just old.) RedHat 7.0 has now switched to using LPRng rather than the legacy lpr that Debian and other distributions use. LPRng is a more modern and comprehensive package. It supports the same /etc/printcap le and identical binaries as did the legacy lpr on RedHat 6.2. The only differences are in the control les created in your spool directories, and a different access control mechanism (discussed below). Note that LPRng has strict permissions requirements on spool directories and is not trivial to install from source.

21.4.3 User programs

User programs are found in one or another bin directory. In this case, we can see lpq, lpr, lprm, and lptest, as well as their associated man pages.

21.4.4 Daemon and administrator programs

Daemon and administrator command will an sbin directory. In this case we can see lpc, lpd, lpf, and pac, as well as their associated man pages. The only daemon (background) program is really the lpd program itself, which is the core of the whole package.

21.4 Package Elements

A packages many les can be loosely grouped into functional elements. In this sectiom, each element will be explained, drawing on the lpr package as an example. Refer to the list of les in Section 21.2.

21.4.5 Conguration les

The le /etc/printcap controls lpd. Most system services will have a le in /etc. printcap is a plain text le that lpd reads on startup. Conguring any service primarily involves editing its conguration le. Several graphical conguration tools are available that avoid this inconvenience (printtool, which is especially for lpd, and linuxconf), but these actually just silently produce the same conguration le. Because printing is so integral to the system, printcap is not actually provided by the lpr package. Trying rpm -qf /etc/printcap gives setup-2.3.4-1, and dpkg -S /etc/printcap shows it to not be owned (i.e., it is part of the base system).

21.4.1

Documentation les

Documentation should be your rst and foremost interest. Man pages will not always be the only documentation provided. Above we see that lpr does not install very much into the /usr/share/doc directory. However, other packages, like rpm -ql apache, reveal a huge user manual (in /home/httpd/html/manual/ or /var/www/html/manual/), and rpm -ql wuftpd shows lots inside /usr/doc/wu-ftpd-?.?.?.

21.4.2

Web pages, mailing lists, and download points

21.4.6 Service initialization les

The les in /etc/rc.d/init.d/ (or /etc/init.d/) are the startup and shutdown scripts to run lpd on boot and shutdown. You can start lpd yourself on the commandline with 196

Every package will probably have a team that maintains it as well as a web page. In the case of lpd, however, the code is very old, and the various CD vendors do 195

21. System Services and lpd

21.4. Package Elements

21. System Services and lpd

/usr/sbin/lpd

but it is preferably to use the given script:

/etc/rc.d/init.d/lpd start /etc/rc.d/init.d/lpd stop

lpd has a spool directory /var/spool/lpd, which may have been created on installation. You can create spool directories for the two printers in the example below, with
mkdir -p /var/spool/lpd/lp /var/spool/lpd/lp0

(or /etc/init.d/lpd). The script has other uses as well:

/etc/rc.d/init.d/lpd status /etc/rc.d/init.d/lpd restart

21.4.8 Log les

U NIX has a strict policy of not reporting error messages to the user interface whenever there might be no user around to read those messages. Whereas error messages of interactive commands are sent to the terminal screen, error or information messages produced by non-interactive commands are logged to les in the directory /var/log/. A log le is a plain text le that continually has one-liner status messages appended to it by a daemon process. The usual directory for log les is /var/log. The main log les are /var/log/messages and possibly /var/log/syslog. It contains kernel messages and messages from a few primary services. When a service would produce large log les (think web access with thousands of hits per hour), the service would use its own log le. sendmail, for example, uses /var/log/maillog. Actually, lpd does not have a log le of its ownone of its failings. View the system log le with the follow option to tail:

(or /etc/init.d/lpd). To make sure that lpd runs on startup, you can check that it has a symlink under the appropriate run level. The symlinks can be explained by running
ls -al find /etc -name *lpd* find /etc -name *lpd* -ls

showing,
5 -rw-r--r--rw-r--r--rwxr-xr-x lrwxrwxrwx lrwxrwxrwx lrwxrwxrwx lrwxrwxrwx lrwxrwxrwx lrwxrwxrwx lrwxrwxrwx 1 1 1 1 1 1 1 1 1 1 root root root root root root root root root root root root root root root root root root root root 17335 10620 2277 13 13 13 13 13 13 13 Sep Sep Sep Mar Mar Mar Mar Mar Mar Mar 25 25 25 21 21 21 24 21 28 21 2000 2000 2000 14:03 14:03 14:03 01:13 14:03 23:13 14:03 /etc/lpd.conf /etc/lpd.perms /etc/rc.d/init.d/lpd /etc/rc.d/rc0.d/K60lpd /etc/rc.d/rc1.d/K60lpd /etc/rc.d/rc2.d/S60lpd /etc/rc.d/rc3.d/S60lpd /etc/rc.d/rc4.d/S60lpd /etc/rc.d/rc5.d/S60lpd /etc/rc.d/rc6.d/K60lpd -> -> -> -> -> -> -> ../init.d/lpd ../init.d/lpd ../init.d/lpd ../init.d/lpd ../init.d/lpd ../init.d/lpd ../init.d/lpd

tail -f /var/log/messages tail -f /var/log/syslog

The 3 in rc3.d is the what are interested in. Having S60lpd symlinked to lpd under rc3.d means that lpd will be started when the system enters run level 3, which is the systems state of usual operation. Note that under RedHat the command setup has a menu option System Services. The Services list will allow you to manage what services come alive on boot, thus creating the symlinks automatically. For Debian , check the man page for the update-rc.d command. More details on bootup are in Chapter 32.

Restarting the lpd service gives messages like:

Not all distributions log this information.

Jun 27 16:06:43 cericon lpd: lpd shutdown succeeded Jun 27 16:06:45 cericon lpd: lpd startup succeeded

21.4.9 Log le rotation

Log les are rotated daily or weekly by the logrotate package. Its conguration le is /etc/logrotate.conf. For each package that happens to produce a log le, there is an additional conguration le under /etc/logrotate.d/. It is also easy to write your ownbegin by using one of the existing les as an example. Rotation means that the log le is renamed with a .1 extension and then truncated to zero length. The service is notied by the logrotate program, sometimes with a SIGHUP. Your /var/log/ may contain a number of old log les named .2, .3, etc. The point of log le rotation is to prevent log les from growing indenitely. 198

21.4.7

Spool les

Systems services like lpd, innd, sendmail, and uucp create intermediate les in the course of processing each request. These are called spool les and are stored somewhere under the /var/spool/ directory, usually to be processed and then deleted in sequence. 197

21. System Services and lpd

21.5. The printcap File in Detail

21.6. PostScript and the Print Filter

21. System Services and lpd

21.4.10

Environment variables

Most user commands of services make use of some environment variables. These can be dened in your shell startup scripts as usual. For lpr, if no printer is specied on the command-line, the PRINTER environment variable determines the default print queue. For example, export PRINTER=lp1 will force use of the lp1 print queue.

if Input lter. This is an executable script into which printer data is piped. The output of this script is fed directly to the printing device or remote machine. This lter will translate from the applications output into the printers native code. rm Remote machine. If the printer queue is not local, this is the machine name. rp Remote printer queue name. The remote machine will have its own printcap le with possibly several printers dened. This species which printer to use.

21.5 The printcap File in Detail

The printcap (printer capabilities) le is similar to (and based on) the termcap (terminal capabilities) le. Conguring a printer means adding or removing text in this le. printcap contains a list of one-line entries, one for each printer. Lines can be broken by a \ before the newline. Here is an example of a printcap le for two printers.
lp:\ :sd=/var/spool/lpd/lp:\ :mx#0:\ :sh:\ :lp=/dev/lp0:\ :if=/var/spool/lpd/lp/filter: lp0:\ :sd=/var/spool/lpd/lp0:\ :mx#0:\ :sh:\ :rm=edison:\ :rp=lp3:\ :if=/bin/cat:

21.6

PostScript and the Print Filter

On U NIX the standard format for all printing is the PostScript le. PostScript .ps les are graphics les representing arbitrary scalable text, lines, and images. PostScript is actually a programming language specically designed to draw things on a page; hence, .ps les are really PostScript programs. The last line in any PostScript program is always showpage, meaning that all drawing operations are complete and that the page can be displayed. Hence, it is easy to see the number of pages inside a PostScript le by grepping for the string showpage. The procedure for printing on U NIX is to convert whatever you would like to print into PostScript. PostScript les can be viewed with a PostScript emulator, like the gv (GhostView) program. A program called gs (GhostScript) is the standard utility for converting the PostScript into a format suitable for your printer. The idea behind PostScript is that it is a language that can easily be built into any printer. The so-called PostScript printer is one that directly interprets a PostScript le. However, these printers are relatively expensive, and most printers only understand the lesser PCL (printer control language) dialect or some other format. In short, any of the hundreds of different formats of graphics and text have a utility that will convert a le into PostScript, whereafter gs will convert it for any of the hundreds of different kinds of printers. There are actually many printers not supported by
gs at the time of this writing. This is mainly because manufacturers refuse to release specications to their proprietary printer communication protocols . The print lter is the workhorse of this whole

Printers are named by the rst eld: in this case lp is the rst printer and lp0 the second printer. Each printer usually refers to a different physical device with its own queue. The lp printer should always be listed rst and is the default print queue used when no other is specied. Here, lp refers to a local printer on the device /dev/lp0 (rst parallel port). lp0 refers to a remote print queue lp3 on the machine edison. The printcap has a comprehensive man page. However, the following elds are most of what you will ever need: sd Spool directory. This directory contains status and spool les. mx Maximum le size. In the preceding example, unlimited. sh Suppress headers. The header is a few informational lines printed before or after the print job. This option should always be set to off. lp Line printer device. 199

operation.

Most applications conveniently output PostScript whenever printing. For example, netscapes menu selection shows

200

21. System Services and lpd

21.6. PostScript and the Print Filter

21.7. Access Control

21. System Services and lpd

which sends PostScript through the stdin of lpr. All applications without their own printer drivers will do the same. This means that we can generally rely on the fact that the print lter will always receive PostScript. gs, on the other hand, can convert PostScript for any printer, so all that remains is to determine its command-line options. If you have chosen Print To: File, then you can view the resulting output with the gv program. Try gv netscape.ps, which shows a print preview. On U NIX, most desktop applications do not have their own preview facility because the PostScript printer itself is emulated by gv. Note that lter programs should not be used with remote lters; remote printer queues can send their PostScript les as is with :if=/bin/cat: (as in the example printcap le above). This way, the machine connected to the device need be the only one especially congured for it. The lter program we are going to use for the local print queue will be a shell script /var/spool/lpd/lp/filter. Create the lter with then edit it so that it looks like
touch /var/spool/lpd/lp/filter chmod a+x /var/spool/lpd/lp/filter

echo hello | enscript -p - | lpr

21.7

Access Control

You should be very careful about running lpd on any machine that is exposed to the Internet. lpd has had numerous security alerts See Chapter 44. and should really only be used within a trusted LAN.

To prevent any remote machine from using your printer, lpd rst looks in the le /etc/hosts.equiv. This is a simple list of all machines allowed to print to your printers. My own le looks like this:
192.168.3.8 192.168.3.9 192.168.3.10 192.168.3.11

#!/bin/bash cat | gs -sDEVICE=ljet4 -sOutputFile=- -sPAPERSIZE=a4 -r600x600 -q exit 0

The -sDEVICE option describes the printer, in this example a Hewlett Packard LaserJet 1100. Many printers have similar or compatible formats; hence, there are far fewer DEVICEs than different makes of printers. To get a full list of supported devices, use gs -h and also consult one of the following les (depending on your distribution): /usr/doc/ghostscript-?.??/devices.txt /usr/share/doc/ghostscript-?.??/Devices.htm /usr/share/doc/gs/devices.txt.gz The -sOutputFile=- sets to write to stdout (as required for a lter). The sPAPERSIZE can be set to one of 11x17, a3, a4, a5, b3, b4, b5, halfletter, ledger, legal, letter, note, and others listed in the man page. You can also use -g<width>x<height> to set the exact page size in pixels. -r600x600 sets the resolution, in this case, 600 dpi (dots per inch). -q means to set quiet mode, suppressing any informational messages that would otherwise corrupt the PostScript output, and - means to read from stdin and not from a le. Our printer conguration is now complete. What remains is to start lpd and test print. You can do that on the command-line with the enscript package. enscript is a program to convert plain text les into nicely formatted PostScript pages. The man page for enscript shows an enormous number of options, but we can simply try: 201

The le /etc/hosts.lpd does the same but doesnt give administrative control by those machines to the print queues. Note that other services, like sshd and rshd (or in.rshd), also check the hosts.equiv le and consider any machine listed to be equivalent. This means that they are completed trusted and so rshd will not request user logins between machines to be authenticated. This behavior is hence a grave security concern. LPRng on RedHat 7.0 has a different access control facility. It can arbitrarily limit access in a variety of ways, depending on the remote user and the action (such as who is allowed to manipulate queues). The le /etc/lpd.perms contains the conguration. The le format is simple, although LPRngs capabilities are rather involvedto make a long story short, the equivalent hosts.equiv becomes in lpd.perms
ACCEPT SERVICE=* ACCEPT SERVICE=* ACCEPT SERVICE=* ACCEPT SERVICE=* DEFAULT REJECT REMOTEIP=192.168.3.8 REMOTEIP=192.168.3.9 REMOTEIP=192.168.3.10 REMOTEIP=192.168.3.11

Large organizations with many untrusted users should look more closely at the LPRng-HOWTO in /usr/share/doc/LPRng-n.n.nn. It explains how to limit access in more complicated ways. 202

21. System Services and lpd

21.8. Printing Troubleshooting

21.9. Useful Programs

21. System Services and lpd

21.8 Printing Troubleshooting

Here is a convenient order for checking what is not working. 1. Check that your printer is plugged in and working. All printers have a way of printing a test page. Read your printer manual to nd out how. 2. Check your printer cable. 3. Check your CMOS settings for your parallel port. 4. Check your printer cable. 5. Try echo hello > /dev/lp0 to check that the port is operating. The printer should do something to signify that data has at least been received. Chapter 42 explains how to install your parallel port kernel module. 6. Use the lpc program to query the lpd daemon. Try help, then status lp, and so on. 7. Check that there is enough space in your /var and /tmp devices for any intermediate les needed by the print lter. A large print job may require hundreds of megabytes. lpd may not give any kind of error for a print lter failure: the print job may just disappear into nowhere. If you are using legacy lpr, then complain to your distribution vendor about your print lter not properly logging to a le. 8. For legacy lpr, stop lpd and remove all of lpds runtime At or pertaining to the les from /var/spool/lpd and from any of its program being in a running state. subdirectories. (New LPRng should never require this step.) The unwanted les are .seq, lock, status, lpd.lock, and any left over spool les that failed to disappear with lprm (these les are recognizable by long le names with a host name and random key embedded in the le name). Then, restart lpd. 9. For remote queues, check that you can do forward and reverse lookups on both machines of both machines host names and IP address. If not, you may get Host name for your address (ipaddr) unknown error messages when trying an lpq. Test with the command host <ip-address> and also host <machinename> on both machines. If any of these do not work, add entries for both machines in /etc/hosts from the example on page 278. Note that the host command may be ignorant of the le /etc/hosts and may still fail. Chapter 40 will explain name lookup conguration. 10. Run your print lter manually to check that it does, in fact, produce the correct output. For example, echo hello | enscript -p - | /var/spool/lpd/lp/filter > /dev/lp0. 11. Legacy lpd is a bit of a quirky packagemeditate. 203

21.9
21.9.1

Useful Programs
printtool

printtool is a graphical printer setup program that helps you very quickly set up lpd. It immediately generates a printcap le and magic lter, and you need not know anything about lpd conguration.

21.9.2

apsfilter

apsfilter stands for any to PostScript lter. The setup described above requires everything be converted to PostScript before printing, but a lter could foreseeably use the file command to determine the type of data coming in and then invoke a program to convert it to PostScript before piping it through gs. This would enable JPEG, GIF, plain text, DVI les, or even gzipped HTML to be printed directly, since PostScript converters have been written for each of these. apsfilter is one of a few such lters, which are generally called magic lters. This is because the file command uses magic numbers.
See page 37.

I personally nd this feature a gimmick rather than a genuine utility, since most of the time you want to lay out the graphical object on a page before printing, which requires you to preview it, and hence convert it to PostScript manually. For most situations, the straight PostScript lter above will work adequately, provided users know to use enscript instead of lpr when printing plain text.

21.9.3

mpage

mpage is a useful utility for saving the trees. It resizes PostScript input so that two, four or eight pages t on one. Change your print lter to:
#!/bin/bash cat | mpage -4 | gs -sDEVICE=ljet4 -sOutputFile=- -sPAPERSIZE=a4 -r600x600 -q exit 0

21.9.4

psutils

The package psutils contains a variety of command-line PostScript manipulation programsa must for anyone doing fancy things with lters. 204

21. System Services and lpd

21.10. Printing to Things Besides Printers

21. System Services and lpd

21.10 Printing to Things Besides Printers

The printcap allows anything to be specied as the printer device. If we set it to /dev/null and let our lter force the output to an alternative device, then we can use lpd to redirect print jobs to any kind of service imaginable. Here, my filter.sh is a script that might send the print job through an SMB (Windows NT) print share (using smbclientsee Chapter 39), to a printer previewer, or to a script that emails the job somewhere.
lp1:\ :sd=/var/spool/lpd/lp1:\ :mx#0:\ :sh:\ :lp=/dev/null:\ :if=/usr/local/bin/my_filter.sh:

We see a specic example of redirecting print jobs to a fax machine in Chapter 33.

205

206

22.1. C Fundamentals

22. Trivial Introduction to C

22.1

C Fundamentals

We start with a simple program and then add fundamental elements to it. Before going too far, you may wish to review bash functions in Section 7.7.

Chapter 22

22.1.1 The simplest C program

A simple program is:

#include <stdlib.h> #include <stdio.h>

Trivial Introduction to C
was invented for the purpose of writing an operating system that could be recompiled (ported) to different hardware platforms (different CPUs). Because the operating system is written in , this language is the rst choice for writing any kind of application that has to communicate efciently with the operating system. Many people who dont program very well in think of as an arbitrary language out of many. This point should be made at once: is the fundamental basis of all computing in the world today. U NIX, Microsoft Windows, ofce suites, web browsers and device drivers are all written in . Ninety-nine percent of your time spent at a computer is probably spent using an application written in . About 70% of all open source software is written in , and the remaining 30% written in languages whose compilers or interpreters are written in . C++ is also quite popular. It is, however, not as
fundamental to computing, although it is more suitable in many situations.

int main (int argc, char *argv[]) { printf ("Hello World!\n"); return 3; }

Compiling Save this program in a le hello.c. We will now compile the program. is the process of turning C code into assembler instructions. Assembler instructions are the program code that your 80?86/SPARC/RS6000 CPU understands directly. The resulting binary executable is fast because it is executed natively by your processorit is the very chip that you see on your motherboard that does fetch Hello byte for byte from memory and executes each instruction. This is what is meant by million instructions per second (MIPS). The megahertz of the machine quoted by hardware vendors is very roughly the number of MIPS. Interpreted languages (like shell scripts) are much slower because the code itself is written in something not understandable to the CPU. The /bin/bash program has to interpret the shell program. /bin/bash itself is written in , but the overhead of interpretation makes scripting languages many orders of magnitude slower than compiled languages. Shell scripts do not need to be compiled.

Run the command

gcc -Wall -o hello hello.c

Further, there is no replacement for . Since it fullls its purpose almost awlessly, there will never be a need to replace it. Other languages may fulll other purposes, but fullls its purpose most adequately. For instance, all future operating systems will probably be written in for a long time to come. It is for these reasons that your knowledge of U NIX will never be complete until you can program in . On the other hand, just because you can program in does not mean that you should. Good programming is a ne art which many veteran programmers never manage to master, even after many years. It is essential to join a Free software project to properly master an effective style of development. 207

The -o hello option tells gcc GNU Compiler. cc on other U NIX systems. to produce the binary le hello instead of the default binary le named a.out. Called a.out for historical reasons. The -Wall option means to report all Warnings during the compilation. This is not strictly necessary but is most helpful for correcting possible errors in your programs. More compiler options are discussed on page 239.

Then, run the program with

./hello

Previously you should have familiarized yourself with bash functions. In all code is inside a function. The rst function to be called (by the operating system) is the main function. 208

22. Trivial Introduction to C

22.1. C Fundamentals

22. Trivial Introduction to C

Type echo $? to see the return code of the program. You will see it is 3, the return value of the main function. Other things to note are the " on either side of the string to be printed. Quotes are required around string literals. Inside a string literal, the \n escape sequence indicates a newline character. ascii(7) shows some other escape sequences. You can also see a proliferation of ; everywhere in a program. Every statement in is terminated by a ; unlike statements in shell scripts where a ; is optional. Now try:
5

The int x is a variable declaration. It tells the program to reserve space for one integer variable that it will later refer to as x. int is the type of the variable. x = 10 assigned a value of 10 to the variable. There are types for each kind of number you would like to work with, and format speciers to convert them for printing:
#include <stdlib.h> #include <stdio.h> int main (int argc, char *argv[]) { char a; short b; int c; long d; float e; double f; long double g; a = A; b = 10; c = 10000000; d = 10000000; e = 3.14159; f = 10e300; g = 10e300; printf ("%c, %hd, %d, %ld, %f, %f, %Lf\n", a, b, c, d, e, f, g); exit (3); }

#include <stdlib.h> #include <stdio.h> int main (int argc, char *argv[]) { printf ("number %d, number %d\n", 1 + 2, 10); exit (3); }

printf can be thought of as the command to send output to the terminal. It is also what is known as a standard library function. In other words, it is specied that a implementation should always have the printf function and that it should behave in a certain way. The %d species that a decimal should go in at that point in the text. The number to be substituted will be the rst argument to the printf function after the string literalthat is, the 1 + 2. The next %d is substituted with the second argumentthat is, the 10. The %d is known as a format specier. It essentially converts an integer number into a decimal representation. See printf(3) for more details.

22.1.2

Variables and types

You will notice that %f is used for both floats and doubles. The reason is that a float is always converted to a double before an operation like this. Also try replacing %f with %e to print in exponential notationthat is, less signicant digits.

With bash, you could use a variable anywhere, anytime, and the variable would just be blank if it had never been assigned a value. In , however, you have to explicitly tell the compiler what variables you are going to need before each block of code. You do this with a variable declaration:
#include <stdlib.h> #include <stdio.h> int main (int argc, char *argv[]) { int x; int y; x = 10; y = 2: printf ("number %d, number %d\n", 1 + y, x); exit (3);

22.1.3 Functions
Functions are implemented as follows:
#include <stdlib.h> #include <stdio.h> void mutiply_and_print (int x, int y) { printf ("%d * %d = %d\n", x, y, x * y); } int main (int argc, char *argv[]) { mutiply_and_print (30, 5);

209

210

22. Trivial Introduction to C

22.1. C Fundamentals

22. Trivial Introduction to C

mutiply_and_print (12, 3); exit (3);

printf ("No, x is not equal to 10 or 20\n"); }

Here we have a non-main function called by the main function. The function is rst declared with
void mutiply_and_print (int x, int y)

if (x > 10) { printf ("Yes, x is more than 10\n"); } while (x > 0) { printf ("x is %d\n", x); x = x - 1; } for (x = 0; x < 10; x++) { printf ("x is %d\n", x); } switch (x) { case 9: printf break; case 10: printf break; case 11: printf break; default: printf break; } return 0;

This declaration states the return value of the function (void for no return value), the function name (mutiply and print), and then the arguments that are going to be passed to the function. The numbers passed to the function are given their own names, x and y, and are converted to the type of x and y before being passed to the function in this case, int and int. The actual code that comprises the function goes between curly braces { and }. In other words, the above function is equivalent to:

("x is nine\n");

("x is ten\n");

void mutiply_and_print () { int x; int y; x = <first-number-passed> y = <second-number-passed> printf ("%d * %d = %d\n", x, y, x * y); }

("x is eleven\n");

("x is huh?\n");

22.1.4

for, while, if, and switch statements

It is easy to see the format that these statements take, although they are vastly different from shell scripts. code works in statement blocks between curly braces, in the same way that shell scripts have dos and dones. Note that with most programming languages when we want to add 1 to a variable we have to write, say, x = x + 1. In , the abbreviation x++ is used, meaning to increment a variable by 1. The for loop takes three statements between ( . . . ): a statement to start things off, a comparison, and a statement to be executed on each completion of the statement block. The statement block after the for is repeatedly executed until the comparison is untrue. The switch statement is like case in shell scripts. switch considers the argument inside its ( . . . ) and decides which case line to jump to. In this example it will obviously be printf ("x is ten\n"); because x was 10 when the previous for loop exited. The break tokens mean that we are through with the switch statement and that execution should continue from Line 46.

As with shell scripting, we have the for, while, and if statements:

#include <stdlib.h> #include <stdio.h> int main (int argc, char *argv[]) { int x; x = 10;
10

if (x == 10) { printf ("x is exactly 10\n"); x++; } else if (x == 20) { printf ("x is equal to 20\n"); } else {

211

212

22. Trivial Introduction to C

22.1. C Fundamentals

22. Trivial Introduction to C

Note that in the comparison == is used instead of =. The symbol = means to assign a value to a variable, whereas == is an equality operator.

(Note that the rst item in the array is y[0], not y[1], as with some other programming languages.) In the preceding example, the line char y[11] reserved 11 bytes for the string. But what if you want a string of 100,000 bytes? allows you to request memory from the kernel. This is called allocate memory. Any non-trivial program will allocate memory for itself and there is no other way of getting large blocks of memory for your program to use. Try:
#include <stdlib.h> #include <stdio.h> int main (int argc, char *argv[]) { int x; char *y; y = malloc (11); printf ("%ld\n", y); for (x = 0; x < 10; x++) { y[x] = 65 + x * 2; } y[10] = 0; printf ("string is %s\n", y); free (y); return 0; }

22.1.5

Strings, arrays, and memory allocation

You can dene a list of numbers with: This list is called an array:
#include <stdlib.h> #include <stdio.h> int main (int argc, char *argv[]) { int x; int y[10]; for (x = 0; x < 10; x++) { y[x] = x * 2; } for (x = 0; x < 10; x++) { printf ("item %d is %d\n", x, y[x]); } return 0; } int y[10];

If an array is of type character, then it is called a string:

#include <stdlib.h> #include <stdio.h> int main (int argc, char *argv[]) { int x; char y[11]; for (x = 0; x < 10; x++) { y[x] = 65 + x * 2; } for (x = 0; x < 10; x++) { printf ("item %d is %d\n", x, y[x]); } y[10] = 0; printf ("string is %s\n", y); return 0; }

The declaration char *y means to declare a variable (a number) called y that points to a memory location. The * (asterisk) in this context means pointer. For example, if you have a machine with perhaps 256 megabytes of RAM + swap, then y potentially has a range of this much. The numerical value of y is also printed with printf ("%ld\n", y);, but is of no interest to the programmer. When you have nished using memory you must give it back to the operating system by using free. Programs that dont free all the memory they allocate are said to leak memory. Allocating memory often requires you to perform a calculation to determine the amount of memory required. In the above case we are allocating the space of 11 chars. Since each char is really a single byte, this presents no problem. But what if we were allocating 11 ints? An int on a PC is 32 bitsfour bytes. To determine the size of a type, we use the sizeof keyword:

#include <stdlib.h> #include <stdio.h> int main (int argc, char *argv[]) { int a; int b;

Note that a string has to be null-terminated. This means that the last character must be a zero. The code y[10] = 0 sets the 11th item in the array to zero. This also means that strings need to be one char longer than you would think. 213

214

22. Trivial Introduction to C

22.1. C Fundamentals

22. Trivial Introduction to C

int c; int d; int e; int f; int g; a = sizeof (char); b = sizeof (short); c = sizeof (int); d = sizeof (long); e = sizeof (float); f = sizeof (double); g = sizeof (long double); printf ("%d, %d, %d, %d, %d, %d, %d\n", a, b, c, d, e, f, g); return 0;

{ int length_of_word; int i; int length_of_sentence; char p[256]; char *q; strcpy (p, "hello there, my name is fred.");
15

length_of_sentence = strlen (p); length_of_word = 0; for (i = 0; i <= length_of_sentence; i++) { if (p[i] == || i == length_of_sentence) { q = malloc (length_of_word + 1); if (q == 0) { perror ("malloc failed"); abort (); } strncpy (q, p + i - length_of_word, length_of_word); q[length_of_word] = 0; printf ("word: %s\n", q); free (q); length_of_word = 0; } else { length_of_word = length_of_word + 1; } } return 0;

Here you can see the number of bytes required by all of these types. Now we can easily allocate arrays of things other than char.
#include <stdlib.h> #include <stdio.h> int main (int argc, char *argv[]) { int x; int *y; y = malloc (10 * sizeof (int)); printf ("%ld\n", y); for (x = 0; x < 10; x++) { y[x] = 65 + x * 2; } for (x = 0; x < 10; x++) { printf ("%d\n", y[x]); } free (y); return 0; }

Here we introduce three more standard library functions. strcpy stands for stringcopy. It copies bytes from one place to another sequentially, until it reaches a zero byte (i.e., the end of string). Line 13 of this program copies text into the character array p, which is called the target of the copy. strlen stands for stringlength. It determines the length of a string, which is just a count of the number of characters up to the null character. We need to loop over the length of the sentence. The variable i indicates the current position in the sentence. Line 20 says that if we nd a character 32 (denoted by ), we know we have reached a word boundary. We also know that the end of the sentence is a word boundary even though there may not be a space there. The token || means OR. At this point we can allocate memory for the current word and copy the word into that memory. The strncpy function is useful for this. It copies a string, but only up to a limit of length of word characters (the last argument). Like strcpy, the rst argument is the target, and the second argument is the place to copy from. To calculate the position of the start of the last word, we use p + i length of word. This means that we are adding i to the memory location p and 215 216

On many machines an int is four bytes (32 bits), but you should never assume this. Always use the sizeof keyword to allocate memory.

22.1.6

String operations

programs probably do more string manipulation than anything else. Here is a program that divides a sentence into words:
#include <stdlib.h> #include <stdio.h> #include <string.h>
5

int main (int argc, char *argv[])

22. Trivial Introduction to C

22.1. C Fundamentals

22. Trivial Introduction to C

then going back length of word counts thereby pointing strncpy to the exact position. Finally, we null-terminate the string on Line 27. We can then print q, free the used memory, and begin with the next word. For a complete list of string operations, see string(3).

The command fgetc gets a character from the le. It retrieves consecutive bytes from the le until it reaches the end of the le, when it returns a -1. The break statement says to immediately terminate the for loop, whereupon execution will continue from line 21. break statements can appear inside while loops as well. You will notice that the for statement is empty. This is allowable means to loop forever. code and

22.1.7

File operations

Some other le functions are fread, fwrite, fputc, fprintf, and fseek. See fwrite(3), fputc(3), fprintf(3), and fseek(3).

Under most programming languages, le operations involve three steps: opening a le, reading or writing to the le, and then closing the le. You use the command fopen to tell the operating system that you are ready to begin working with a le: The following program opens a le and spits it out on the terminal:

22.1.8 Reading command-line arguments inside C programs

Up until now, you are probably wondering what the (int argc, char *argv[]) are for. These are the command-line arguments passed to the program by the shell. argc is the total number of command-line arguments, and argv is an array of strings of each argument. Printing them out is easy:
#include <stdlib.h> #include <stdio.h> #include <string.h>
5

#include <stdlib.h> #include <stdio.h> #include <string.h>

int main (int argc, char *argv[]) { int c; FILE *f; f = fopen ("mytest.c", "r"); if (f == 0) { perror ("fopen"); return 1; } for (;;) { c = fgetc (f); if (c == -1) break; printf ("%c", c); } fclose (f); return 0;

int main (int argc, char *argv[]) { int i; for (i = 0; i < argc; i++) { printf ("argument %d is %s\n", i, argv[i]); } return 0; }

22.1.9 A more complicated example

Here we put this altogether in a program that reads in lots of les and dumps them as words. Here are some new notations you will encounter: != is the inverse of == and tests if not-equal-to; realloc reallocates memoryit resizes an old block of memory so that any bytes of the old block are preserved; \n, \t mean the newline character, 10, or the tab character, 9, respectively (see ascii(7)).
#include <stdlib.h> #include <stdio.h> #include <string.h>
5

A new type is presented here: FILE *. It is a le operations variable that must be initialized with fopen before it can be used. The fopen function takes two arguments: the rst is the name of the le, and the second is a string explaining how we want to open the lein this case "r" means reading from the start of the le. Other options are "w" for writing and several more described in fopen(3). If the return value of fopen is zero, it means that fopen has failed. The perror function then prints a textual error message (for example, No such file or directory). It is essential to check the return value of all library calls in this way. These checks will constitute about one third of your program. 217

void word_dump (char *filename) { int length_of_word; int amount_allocated;

218

22. Trivial Introduction to C

22.1. C Fundamentals

22. Trivial Introduction to C

char q; FILE f; int c; c = 0;

}
75

return 0;

f = fopen (filename, "r"); if (f == 0) { perror ("fopen failed"); exit (1); } length_of_word = 0; amount_allocated = 256; q = malloc (amount_allocated); if (q == 0) { perror ("malloc failed"); abort (); } while (c != -1) { if (length_of_word >= amount_allocated) { amount_allocated = amount_allocated * 2; q = realloc (q, amount_allocated); if (q == 0) { perror ("realloc failed"); abort (); } } c = fgetc (f); q[length_of_word] = c; if (c == -1 || c == || c == \n || c == \t) { if (length_of_word > 0) { q[length_of_word] = 0; printf ("%s\n", q); } amount_allocated = 256; q = realloc (q, amount_allocated); if (q == 0) { perror ("realloc failed"); abort (); } length_of_word = 0; } else { length_of_word = length_of_word + 1; } }

This program is more complicated than you might immediately expect. Reading in a le where we are sure that a word will never exceed 30 characters is simple. But what if we have a le that contains some words that are 100,000 characters long? GNU programs are expected to behave correctly under these circumstances. To cope with normal as well as extreme circumstances, we start off assuming that a word will never be more than 256 characters. If it appears that the word is growing over 256 characters, we reallocate the memory space to double its size (lines 32 amd 33). When we start with a new word, we can free up memory again, so we realloc back to 256 again (lines 48 and 49). In this way we are using the minimum amount of memory at each point in time. We have hence created a program that can work efciently with a 100-gigabyte le just as easily as with a 100-byte le. This is part of the art of programming. Experienced programmers may actually scoff at the above listing because it really isnt as minimalistic as is absolutely possible. In fact, it is a truly excellent listing for the following reasons: The program is easy to understand. The program uses an efcient algorithm (albeit not optimal). The program contains no arbitrary limits that would cause unexpected behavior in extreme circumstances. The program uses no nonstandard functions or notations that would prohibit it compiling successfully on other systems. It is therefore portable. Readability in is your rst priorityit is imperative that what you do is obvious to anyone reading the code.

fclose (f); } int main (int argc, char *argv[]) { int i; if (argc < 2) { printf ("Usage:\n\twordsplit <filename> ...\n"); exit (1); } for (i = 1; i < argc; i++) { word_dump (argv[i]);

22.1.10

#include statements and prototypes

At the start of each program will be one or more #include statements. These tell the compiler to read in another program. Now, raw does not have a whole lot in the way of protecting against errors: for example, the strcpy function could just as well be used with one, three, or four arguments, and the program would still compile. It would, however, wreak havoc with the internal memory and cause the program to crash. These other .h programs are called header les. They contain templates for 220

219

22. Trivial Introduction to C

22.1. C Fundamentals

22. Trivial Introduction to C

how functions are meant to be called. Every function you might like to use is contained in one or another template le. The templates are called function prototypes. C++ has
something called templates. This is a special C++ term having nothing to do with the discussion here.

int sqr (int x) { <...>

A function prototype is written the same as the function itself, but without the code. A function prototype for word dump would simply be: The trailing ; is essential and distinguishes a function prototype from a function. After a function prototype is dened, any attempt to use the function in a way other than intendedsay, passing it to few arguments or arguments of the wrong typewill be met with erce opposition from gcc. You will notice that the #include <string.h> appeared when we started using string operations. Recompiling these programs without the #include <string.h> line gives the warning message which is quite to the point.
mytest.c:21: warning: implicit declaration of function strncpy
5

void word_dump (char *filename);

which is liable to cause nausea. In C++, the additional comment // is allowed, whereby everything between the // and the end of the line is ignored. It is accepted under gcc, but should not be used unless you really are programming in C++. In addition, programmers often comment out lines by placing a #if 0 . . . #endif around them, which really does exactly the same thing as a comment (see Section 22.1.12) but allows you to have comments within comments. For example
int x; x = 10; #if 0 printf ("debug: x is %d\n", x); #endif y = x + 10; <...>

/* print debug information */

comments out Line 4.

The function prototypes give a clear denition of how every function is to be used. Man pages will always rst state the function prototype so that you are clear on what arguments are to be passed and what types they should have.

22.1.12

#define and #if C macros

22.1.11 C comments
A comment is denoted with /* <comment lines> */ and can span multiple lines. Anything between the /* and */ is ignored. Every function should be commented, and all nonobvious code should be commented. It is a good maxim that a program that needs lots of comments to explain it is badly written. Also, never comment the obvious, and explain why you do things rather that what you are doing. It is advisable not to make pretty graphics between each function, so rather:
/* returns -1 on error, takes a positive integer */ int sqr (int x) { <...>

Anything starting with a # is not actually , but a preprocessor directive. A program is rst run through a preprocessor that removes all spurious junk, like comments, #include statements, and anything else beginning with a #. You can make programs much more readable by dening macros instead of literal values. For instance,
#define START_BUFFER_SIZE 256

in our example program, #defines the text START BUFFER SIZE to be the text 256. Thereafter, wherever in the program we have a START BUFFER SIZE, the text 256 will be seen by the compiler, and we can use START BUFFER SIZE instead. This is a much cleaner way of programming because, if, say, we would like to change the 256 to some other value, we only need to change it in one place. START BUFFER SIZE is also more meaningful than a number, making the program more readable. Whenever you have a literal constant like 256, you should replace it with a macro dened near the top of your program. You can also check for the existence of macros with the #ifdef and #ifndef directive. # directives are really a programming language all on their own:
/* Set START_BUFFER_SIZE to fine-tune performance before compiling: */ #define START_BUFFER_SIZE 256 /* #define START_BUFFER_SIZE 128 */ /* #define START_BUFFER_SIZE 1024 */ /* #define START_BUFFER_SIZE 16384 */

than

/***************************----SQR----****************************** * x = argument to make the square of * * return value = * * -1 (on error) * * square of x (on success) * ********************************************************************/

221

222

22. Trivial Introduction to C

22.2. Debugging with gdb and strace

22. Trivial Introduction to C

#ifndef START_BUFFER_SIZE #error This code did not define START_BUFFER_SIZE. Please edit #endif
10

The rst of these is run which executes the program as though it had been started from a regular shell:
(gdb) r Starting program: /homes/src/wordsplit/wordsplit Usage: wordsplit <filename> ...
5

#if START_BUFFER_SIZE <= 0 #error Wooow! START_BUFFER_SIZE must be greater than zero #endif
15

#if START_BUFFER_SIZE < 16 #warning START_BUFFER_SIZE to small, program may be inefficient #elif START_BUFFER_SIZE > 65536 #warning START_BUFFER_SIZE to large, program may be inefficient #else /* START_BUFFER_SIZE is ok, do not report */ #endif void word_dump (char *filename) { <...> amount_allocated = START_BUFFER_SIZE; q = malloc (amount_allocated); <...>

Program exited with code 01.

Obviously, we will want to set some trial command-line arguments. This is done with the special command, set args:
(gdb) set args readme.txt readme2.txt

The break command is used like b [[<file>:]<line>|<function>], and sets a break point at a function or line number:
(gdb) b main Breakpoint 1 at 0x8048796: file wordsplit.c, line 67.

22.2 Debugging with gdb and strace

Programming errors, or bugs, can be found by inspecting program execution. Some developers claim that the need for such inspection implies a sloppy development process. Nonetheless it is instructive to learn by actually watching a program work.

A break point will interrupt execution of the program. In this case the program will stop when it enters the main function (i.e., right at the start). Now we can run the program again:
(gdb) r Starting program: /home/src/wordsplit/wordsplit readme.txt readme2.txt Breakpoint 1, main (argc=3, argv=0xbffff804) at wordsplit.c:67 67 if (argc < 2) { (gdb)

22.2.1

gdb

As specied, the program stops at the beginning of the main function at line 67. If you are interested in viewing the contents of a variable, you can use the print command:
(gdb) p argc $1 = 3 (gdb) p argv[1] $2 = 0xbffff988 "readme.txt"

The GNU debugger, gdb, is a replacement for the standard U NIX debugger, db. To debug a program means to step through its execution line-by-line, in order to nd programming errors as they happen. Use the command gcc -Wall -g -O0 -o wordsplit wordsplit.c to recompile your program above. The -g option enables debugging support in the resulting executable and the -O0 option disables compiler optimization (which sometimes causes confusing behavior). For the following example, create a test le readme.txt with some plain text inside it. You can then run gdb -q wordsplit. The standard gdb prompt will appear, which indicates the start of a debugging session:
(gdb)

which tells us the value of argc and argv[1]. The list command displays the lines about the current line:
(gdb) l 63 int main (int argc, char *argv[]) 64 { 65 int i; 66

At the prompt, many one letter commands are available to control program execution. 223

224

22. Trivial Introduction to C

22.2. Debugging with gdb and strace

22. Trivial Introduction to C

The list command can also take an optional le and line number (or even a function name):
(gdb) l wordsplit.c:1 1 #include <stdlib.h> 2 #include <stdio.h> 3 #include <string.h> 4 5 void word_dump (char *filename) 6 { 7 int length_of_word; 8 int amount_allocated;

67 68 69 70

if (argc < 2) { printf ("Usage:\n\twordsplit <filename> ...\n"); exit (1); }

(gdb) n 49 (gdb) n 50 (gdb) n 54 q = realloc (q, amount_allocated); if (q == 0) { length_of_word = 0;

This activity is called stepping through your program. The s command is identical to n except that it dives into functions instead of running them as single line. To see the difference, step over line 73 rst with n, and then with s, as follows:
(gdb) set args readme.txt readme2.txt (gdb) b main Breakpoint 1 at 0x8048796: file wordsplit.c, line 67. (gdb) r Starting program: /home/src/wordsplit/wordsplit readme.txt readme2.txt Breakpoint 1, main (argc=3, argv=0xbffff814) at wordsplit.c:67 67 if (argc < 2) { (gdb) n 72 for (i = 1; i < argc; i++) { (gdb) n 73 word_dump (argv[i]); (gdb) n Zaphod has two heads 72 for (i = 1; i < argc; i++) { (gdb) s 73 word_dump (argv[i]); (gdb) s word_dump (filename=0xbffff993 "readme2.txt") at wordsplit.c:13 13 c = 0; (gdb) s 15 f = fopen (filename, "r"); (gdb)

Next, we can try setting a break point at an arbitrary line and then using the continue command to proceed with program execution:
(gdb) b wordsplit.c:48 Breakpoint 2 at 0x804873e: file wordsplit.c, line 48. (gdb) c Continuing. Zaphod Breakpoint 2, word_dump (filename=0xbffff988 "readme.txt") at wordsplit.c:48 48 amount_allocated = 256;

Execution obediently stops at line 48. At this point it is useful to run a backtrace. This prints out the current stack which shows the functions that were called to get to the current line. This output allows you to trace the history of execution.
(gdb) bt #0 word_dump (filename=0xbffff988 "readme.txt") at wordsplit.c:48 #1 0x80487e0 in main (argc=3, argv=0xbffff814) at wordsplit.c:73 #2 0x4003db65 in __libc_start_main (main=0x8048790 <main>, argc=3, ubp_av=0xbf fff814, init=0x8048420 <_init>, fini=0x804883c <_fini>, rtld_fini=0x4000df24 <_dl_fini>, stack_end=0xbffff8 0c) at ../sysdeps/generic/libc-start.c:111

An interesting feature of gdb is its ability to attach onto running programs. Try the following sequence of commands:
[root@cericon]# lpd [root@cericon]# ps awx | grep lpd 28157 ? S 0:00 lpd Waiting 28160 pts/6 S 0:00 grep lpd [root@cericon]# gdb -q /usr/sbin/lpd (no debugging symbols found)... (gdb) attach 28157 Attaching to program: /usr/sbin/lpd, Pid 28157 0x40178bfe in __select () from /lib/libc.so.6 (gdb)

The clear command then deletes the break point at the current line:

(gdb) clear Deleted breakpoint 2

The most important commands for debugging are the next and step commands. The n command simply executes one line of code: 225

226

22. Trivial Introduction to C

22.3. C Libraries

22. Trivial Introduction to C

The lpd daemon was not compiled with debugging support, but the point is still made: you can halt and debug any running process on the system. Try running a bt for fun. Now release the process with
(gdb) detach Detaching from program: /usr/sbin/lpd, Pid 28157

libraries. The Standard functions, run:

library is the le /lib/libc.so.6. To list all the

library

nm /lib/libc.so.6 nm /lib/libc.so.6 | grep T | cut -f3 -d | grep -v _ | sort -u | less

The debugger provides copious amounts of online help. The help command can be run to explain further. The gdb info pages also elaborate on an enormous number of display features and tracing features not covered here.

many of these have man pages, but some will have no documentation and require you to read the comments inside the header les (which are often most explanatory). It is better not to use functions unless you are sure that they are standard functions in the sense that they are common to other systems. To create your own library is simple. Lets say we have two les that contain several functions that we would like to compile into a library. The les are simple math sqrt.c
#include <stdlib.h> #include <stdio.h> static int abs_error (int a, int b) { if (a > b) return a - b; return b - a; } int simple_math_isqrt (int x) { int result; if (x < 0) { fprintf (stderr, "simple_math_sqrt: taking the sqrt of a negative number\n"); abort (); } result = 2; while (abs_error (result * result, x) > 1) { result = (x / result + result) / 2; } return result; }

22.2.2

Examining core les

If your program has a segmentation violation (segfault) then a core le will be written to the current directory. This is known as a core dump. A core dump is caused by a bug in the programits response to a SIGSEGV signal sent to the program because it tried to access an area of memory outside of its allowed range. These les can be examined using gdb to (usually) reveal where the problem occurred. Simply run gdb <executable> ./core and then type bt (or any gdb command) at the gdb prompt. Typing file ./core will reveal something like
/root/core: ELF 32-bit LSB core file of <executable> (signal 11), Intel 80386, version 1

22.2.3

strace

The strace command prints every system call performed by a program. A system call is a function call made by a library function to the L INUX kernel. Try
strace ls strace ./wordsplit

If a program has not been compiled with debugging support, the only way to inspect its execution may be with the strace command. In any case, the command can provide valuable information about where a program is failing and is useful for diagnosing errors.
5

and simple math pow.c

#include <stdlib.h> #include <stdio.h> int simple_math_ipow (int x, int y) { int result; if (x == 1 || y == 0) return 1; if (x == 0 && y < 0) { fprintf (stderr, "simple_math_pow: raising zero to a negative power\n");

22.3

C Libraries
10

We made reference to the Standard library. The language on its own does almost nothing; everything useful is an external function. External functions are grouped into 227

228

22. Trivial Introduction to C

22.3. C Libraries

22.4. C Projects Makefiles

22. Trivial Introduction to C

abort (); } if (y < 0) return 0; result = 1; while (y > 0) { result = result * x; y = y - 1; } return result;

The rst command compiles the le mytest.c into mytest.o, and the second function is called linking the program, which assimilates mytest.o and the libraries into a single executable. The option L. means to look in the current directory for any libraries (usually only /lib and /usr/lib are searched). The option -lsimple math means to assimilate the library libsimple math.a (lib and .a are added automatically). This operation is called static Nothing to do with the static keyword. linking because it happens before the program is run and includes all object les into the executable.

As an aside, note that it is often the case that many static libraries are linked into the same program. Here order is important: the library with the least dependencies should come last, or you will get so-called symbol referencing errors. We can also create a header le simple math.h for using the library.

We would like to call the library simple math. It is good practice to name all the functions in the library simple math ??????. The function abs error is not going to be used outside of the le simple math sqrt.c and so we put the keyword static in front of it, meaning that it is a local function. We can compile the code with:
5

/* calculates the integer square root, aborts on error */ int simple_math_isqrt (int x); /* calculates the integer power, aborts on error */ int simple_math_ipow (int x, int y);

gcc -Wall -c simple_math_sqrt.c gcc -Wall -c simple_math_pow.c

Add the line #include "simple math.h" to the top of mytest.c:

#include <stdlib.h> #include <stdio.h> #include "simple_math.h"

The -c option means compile only. The code is not turned into an executable. The generated les are simple math sqrt.o and simple math pow.o. These are called object les. We now need to archive these les into a library. We do this with the ar command (a predecessor of tar):
ar libsimple_math.a simple_math_sqrt.o simple_math_pow.o ranlib libsimple_math.a

This addition gets rid of the implicit declaration of function warning messages. Usually #include <simple math.h> would be used, but here, this is a header le in the current directoryour own header leand this is where we use "simple math.h" instead of <simple math.h>.

The ranlib command indexes the archive. The library can now be used. Create a le mytest.c:
#include <stdlib.h> #include <stdio.h> int main (int argc, char *argv[]) { printf ("%d\n", simple_math_ipow (4, 3)); printf ("%d\n", simple_math_isqrt (50)); return 0; }

22.4

C Projects Makefiles

What if you make a small change to one of the les (as you are likely to do very often when developing)? You could script the process of compiling and linking, but the script would build everything, and not just the changed le. What we really need is a utility that only recompiles object les whose sources have changed: make is such a utility. make is a program that looks inside a Makefile in the current directory then does a lot of compiling and linking. Makefiles contain lists of rules and dependencies describing how to build a program. Inside a Makefile you need to state a list of what-depends-on-what dependencies that make can work through, as well as the shell commands needed to achieve each goal. 230

and run
gcc -Wall -c mytest.c gcc -o mytest mytest.o -L. -lsimple_math

229

22. Trivial Introduction to C

22.4. C Projects Makefiles

22. Trivial Introduction to C

22.4.1

Completing our example Makefile

all: libsimple_math.a mytest

Our rst (last?) dependency in the process of completing the compilation is that mytest depends on both the library, libsimple math.a, and the object le, mytest.o. In make terms we create a Makefile line that looks like:
mytest: libsimple_math.a mytest.o

The all: target is the rule that make tries to satisfy when make is run with no command-line arguments. This just means that libsimple math.a and mytest are the last two les to be built, that is, they are the top-level dependencies. Makefiles also have their own form of environment variables, like shell scripts. You can see that we have used the text simple math in three of our rules. It makes sense to dene a macro for this so that we can easily change to a different library name. Our nal Makefile is:

meaning simply that the les libsimple math.a mytest.o must exist and be updated before mytest. mytest: is called a make target. Beneath this line, we also need to state how to build mytest:
gcc -Wall -o $@ mytest.o -L. -lsimple_math

# Comments start with a # (hash) character like shell scripts. # Makefile to build libsimple_math.a and mytest program. # Paul Sheer <psheer@icon.co.za> Sun Mar 19 15:56:08 2000 OBJS = simple_math_sqrt.o simple_math_pow.o LIBNAME = simple_math CFLAGS = -Wall all:
10

The $@ means the name of the target itself, which is just substituted with mytest. Note that the space before the gcc is a tab character and not 8 space characters. The next dependency is that libsimple math.a depends on simple math sqrt.o simple math pow.o. Once again we have a dependency, along with a shell script to build the target. The full Makefile rule is:
libsimple_math.a: simple_math_sqrt.o simple_math_pow.o rm -f $@ ar rc $@ simple_math_sqrt.o simple_math_pow.o ranlib $@

lib$(LIBNAME).a mytest lib$(LIBNAME).a mytest.o gcc $(CFLAGS) -o $@ mytest.o -L. -l${LIBNAME}

mytest:

Note again that the left margin consists of a single tab character and not spaces. The nal dependency is that the les simple math sqrt.o and simple math pow.o depend on the les simple math sqrt.c and simple math pow.c. This requires two make target rules, but make has a short way of stating such a rule in the case of many source les,
.c.o:
20

lib$(LIBNAME).a: $(OBJS) rm -f $@ ar rc $@ $(OBJS) ranlib $@ .c.o: gcc $(CFLAGS) -c -o $*.o $< clean:

make

rm -f .o .a mytest

gcc -Wall -c -o $*.o $<

We can now easily type

which means that any .o les needed can be built from a .c le of a similar name by means of the command gcc -Wall -c -o $*.o $<, where $*.o means the name of the object le and $< means the name of the le that $*.o depends on, one at a time.

in the current directory to cause everything to be built.

22.4.2

Putting it all together

You can see we have added an additional disconnected target clean:. Targets can be run explictly on the command-line like this: which removes all built les. Makefiles have far more uses than just building programs. Anything that needs to be built from sources can employ a Makefile to make things easier. 232
make clean

Makefiles can, in fact, have their rules put in any order, so its best to state the most obvious rules rst for readability. There is also a rule you should always state at the outset: 231

23.2. DLL Versioning

23. Shared Libraries

mytest: lib$(LIBNAME).so mytest.o gcc $(CFLAGS) -o $@ mytest.o -L. -l${LIBNAME} lib$(LIBNAME).so: $(OBJS) gcc -shared $(CFLAGS) $(OBJS) -lc -Wl,-soname -Wl,$(SOVERSION) \ -o $(SONAME) && \ ln -sf $(SONAME) $(SOVERSION) && \ ln -sf $(SONAME) lib$(LIBNAME).so .c.o: gcc -fPIC -DPIC $(CFLAGS) -c -o $*.o $<

Chapter 23
20

clean:

Shared Libraries
This chapter follows directly from our construction of static .a libraries in Chapter 22. It discusses creation and installation of Dynamically Linked Libraries (DLLs). Here I show you both so that you have a good technical overview of how DLLs work on U NIX. You can then promptly forget everything except ldconfig and LD LIBRARY PATH discussed below. The .a library le is good for creating functions that many programs can include. This practice is called code reuse. But note how the .a le is linked into (included) in the executable mytest in Chapter 22. mytest is enlarged by the size of libsimple math.a. When hundreds of programs use the same .a le, that code is effectively duplicated all over the le system. Such inefciency was deemed unacceptable long before L INUX , so library les were invented that only link with the program when it runsa process known as dynamic linking. Instead of .a les, similar .so (shared object) les live in /lib/ and /usr/lib/ and are automatically linked to a program when it runs.

rm -f .o .a *.so mytest

The -shared option to gcc builds our shared library. The -W options are linker options that set the version number of the library that linking programs will load at runtime. The -fPIC -DPIC means to generate position-independent code, that is, code suitable for dynamic linking. After running make we have
1 1 1 1 root root root root root root root root 23 23 6046 13677 Sep Sep Sep Sep 17 17 17 17 22:02 22:02 22:02 22:02 libsimple_math.so -> libsimple_math.so.1.0.0 libsimple_math.so.1.0 -> libsimple_math.so.1.0.0 libsimple_math.so.1.0.0 mytest

lrwxrwxrwx lrwxrwxrwx -rwxr-xr-x -rwxr-xr-x

23.2

DLL Versioning

You may observe that our three .so les are similar to the many les in /lib/ and /usr/lib/. This complicated system of linking and symlinking is part of the process of library versioning. Although generating a DLL is out of the scope of most system admin tasks, library versioning is important to understand. DLLs have a problem. Consider a DLL that is outdated or buggy: simply overwriting the DLL le with an updated le will affect all the applications that use it. If these applications rely on certain behavior of the DLL code, then they will probably crash with the fresh DLL. U NIX has elegantly solved this problem by allowing multiple versions of DLLs to be present simultaneously. The programs themselves have their required version number built into them. Try
ldd mytest

23.1 Creating DLL .so Files

Creating a DLL requires several changes to the Makefile on page 232:
OBJS LIBNAME SONAME SOVERSION CFLAGS all: = = = = = simple_math_sqrt.o simple_math_pow.o simple_math libsimple_math.so.1.0.0 libsimple_math.so.1.0 -Wall

lib$(LIBNAME).so mytest

which will show the DLL les that mytest is scheduled to link with:
libsimple_math.so.1.0 => ./libsimple_math.so.1.0 (0x40018000)

233

234

23. Shared Libraries

23.3. Installing DLL .so Files

23. Shared Libraries

At the moment, we are interested in libsimple math.so.1.0. Note how it matches the SOVERSION variable in the Makefile. Note also how we have chosen our symlinks. We are effectively allowing mytest to link with any future libsimple math.so.1.0.? (were our simple math library to be upgraded to a new version) purely because of the way we have chosen our symlinks. However, it will not link with any library libsimple math.so.1.1.?, for example. As developers of libsimple math, we are deciding that libraries of a different minor For this example we are considering libraries to be named libname.so.major.minor.patch version number will be incompatible, whereas libraries of a different patch level will not be incompatible.

libc.so.6 => /lib/libc.so.6 (0x40022000) /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)

The LD LIBRARY PATH environment variable is relevant to every executable on the system and similar to the PATH environment variable. LD LIBRARY PATH dictates what directories should be searched for library les. Here, we appended /usr/local/lib to the search path in case it was missing. Note that even with LD LIBRARY PATH unset, /lib and /usr/lib will always be searched.

We could also change SOVERSION to libsimple math.so.1. This would effectively be saying that future libraries of different minor version numbers are compatible; only a change in the major version number would dictate incompatibility.

23.3 Installing DLL .so Files

If you run ./mytest, you will be greeted with an error while loading shared libraries message. The reason is that the dynamic linker does not search the current directory for .so les. To run your program, you will have to install your library:
mkdir -p /usr/local/lib install -m 0755 libsimple_math.so libsimple_math.so.1.0 \ libsimple_math.so.1.0.0 /usr/local/lib

Then, edit the /etc/ld.so.conf le and add a line Then, recongure your libraries with Finally, run your program with
ldconfig /usr/local/lib

export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/usr/local/lib" ./mytest

ldconfig congures all libraries on the system. It recreates appropriate symlinks (as we did) and rebuilds a lookup cache. The library directories it considers are /lib, /usr/lib, and those listed in /etc/ld.so.config. The ldconfig command should be run automatically when the system boots and manually whenever libraries are installed or upgraded. 235 236

24.1. Building GNU Source Packages

24. Source and Binary Packages

Being GNU standards compliant should also mean that the package can be installed with only the three following commands:
./configure make make install

Chapter 24

It also usually means that packages will compile on any U NIX system. Hence, this section should be a good guide to getting L INUX software to work on non-L INUX machines. An example will illustrate these steps. Begin by downloading cooledit from metalab.unc.edu in the directory /pub/Linux/apps/editors/X/cooledit, using ftp. Make a directory /opt/src in which to build such custom packages. Now run
cd /opt/src tar -xvzf cooledit-3.17.2.tar.gz cd cooledit-3.17.2

Source and Binary Packages

In this chapter you will, rst and foremost, learn to build packages from source, building on your knowledge of Makefiles in Chapter 22. Most packages, however, also come as .rpm (RedHat) or .deb (Debian ) les, which are discussed further below.

24.1 Building GNU Source Packages

Almost all packages originally come as sources, tared and available from one of the many public FTP sites, like metalab.unc.edu. Thoughtful developers would standards compliant. This means that untarring the have made their packages GNU package will reveal the following les inside the top-level directory: INSTALL This is a standard document beginning with the line These are packages are generic installation instructions. Since all GNU installed in the same way, this le should always be the same. NEWS News of interest to users. README Any essential information. This is usually an explanation of what the package does, promotional material, and anything special that need be done to install the package. COPYING The GNU General Public License.

You will notice that most sources have the name package-major.minor.patch.tar.gz. The major version of the package is changed when the developers make a substantial feature update or when they introduce incompatibilities to previous versions. The minor version is usually updated when small features are added. The patch number (also known as the patch level) is updated whenever a new release is made and usually signies bug xes. At this point you can apply any patches you may have. See Section 20.7.3. You can now ./configure the package. The ./configure script is generated by autoconfa package used by developers to create source that will compile on any type of U NIX system. The autoconf package also contains the GNU Coding Standards to which all software should comply. autoconf is the remarkable work of David
MacKenzie. I often hear the myth that U NIX systems have so diverged that they are no longer compatible. The fact that sophisticated software like cooledit (and countless others) compiles on almost any U NIX machine should dispel this nonsense. There is also hype surrounding developers porting commercial software from other U NIX systems to L INUX. If they had written their software in the least bit properly to begin with, there would be no porting to be done. In short, all L INUX software runs on all U NIXs. The only exceptions are a few packages that use some custom features of the L INUX kernel.

AUTHORS A list of major contributors. ChangeLog A specially formatted list containing a history of all changes ever done to the package, by whom, and on what date. Used to track work on the package. 237

Here, --prefix indicates the top-level directory under which the package will be installed. (See Section 17.2.). Always also try to see package-specic options. 238
./configure --help

./configure --prefix=/opt/cooledit

24. Source and Binary Packages

24.1. Building GNU Source Packages

24.2. RedHat and Debian Binary Packages

24. Source and Binary Packages

Another trick sets compile options:

24.2

RedHat and Debian Binary Packages

CFLAGS=-O2 -fomit-frame-pointer -s -pipe ./configure --prefix=/opt/cooledit

In this section, we place Debian examples inside parentheses, ( . . . ). Since these are examples from actual systems, they do not always correspond.

-O2 Sets compiler optimizations to be as fast as possible without making the binary larger. (-O3 almost never provides an advantage.) -fomit-frame-pointer Permits the compiler to use one extra register that would normally be used for debugging. Use this option only when you are absolutely sure you have no interest in analyzing any running problems with the package. -s Strips the object code. This reduces the size of the object code by eliminating any debugging data. -pipe Instructs not to use temporary les. Rather, use pipes to feed the code through the different stages of compilation. This usually speeds compilation. Compile the package. This can take up to several hours depending on the amount of code and your CPU power. cooledit will compile in under 10 minutes on any
entry-level machine at the time of writing.

24.2.1 Package versioning

The package numbering for RedHat and Debian though this is far from a rule): For example, packages is often as follows (al

<package-name>-<source-version>-<package-version>.<hardware-platform>.rpm ( <package-name>_<source-version>-<package-version>.deb )

make

is the Bourne Again Shell you are using, major version 1, minor version 14, patch 7, package version 22, compiled for an Intel 386 processor. Sometimes, the Debian package will have the architecture appended to the version number, in the above case, perhaps bash 2.03-6 i386.deb. The <source-version> is the version on the original .tar le (as above). The <package-version>, also called the release, refers to the .rpm le itself; in this case, bash-1.14.7-22.i386.rpm has been packed together for the 8th time, possibly with minor improvements to the way it installs with each new number. The i386 is called the architecture and could also be sparc for a SPARC Type of processor used in Another non-Intel workstation , Sun Microsystems workstations machine, ppc for a PowerPC alpha for a DEC Alpha High-end 64 bit server/workstation machine, or several others.

bash-1.14.7-22.i386.rpm ( bash_2.03-6.deb )

You can also run

make CFLAGS=-O0 -g

if you decide that you would rather compile with debug support after all. Install the package with

make install

A nice trick to install into a different subdirectory is

mkdir /tmp/cooledit make install prefix=/tmp/cooledit

Not always supported.

24.2.2 Installing, upgrading, and deleting

To install a package, run the following command on the .rpm or .deb le:
rpm -i mirrordir-0.10.48-1.i386.rpm ( dpkg -i mirrordir_0.10.48-2.deb )

You can use these commands to pack up the completed build for untaring onto a different system. You should, however, never try to run a package from a directory different from the one it was --prefixed to install into, since most packages compile in this location and then access installed data from beneath it. Using a source package is often the best way to install when you want the package to work the way the developers intended. You will also tend to nd more documentation, when vendors have neglected to include certain les. 239

Upgrading (Debian automatically chooses an upgrade if the package is already present) can be done with the following command, and then completely uninstalling with
rpm -U mirrordir-0.10.49-1.i386.rpm ( dpkg -i mirrordir_0.10.49-1.deb )

240

24. Source and Binary Packages

24.2. RedHat and Debian Binary Packages

24. Source and Binary Packages

rpm -e mirrordir ( dpkg --purge mirrordir )

( dpkg -l * )

With Debian , a package removal does not remove conguration les, thus allowing you to revert to its current setup if you later decide to reinstall:
dpkg -r mirrordir

Try,

To search for a package name,

rpm -qa | grep <regular-expression> ( dpkg -l <glob-expression> )

If you need to reinstall a package (perhaps because of a le being corrupted), use

rpm -i --force python-1.6-2.i386.rpm

rpm -qa | grep util ( dpkg -l util )

Debian

reinstalls automatically if the package is present.

24.2.3

Dependencies

To query for the existence of a package, say, textutils (query, list),

rpm -q textutils ( dpkg -l textutils )

Packages often require other packages to already be installed in order to work. The package database keeps track of these dependencies. Often you will get an error: failed dependencies: (or dependency problems for Debian ) message when you try to install. This means that other packages must be installed rst. The same might happen when you try to remove packages. If two packages mutually require each other, you must place them both on the command-line at once when installing. Sometimes a package requires something that is not essential or is already provided by an equivalent package. For example, a program may require sendmail to be installed even though exim is an adequate substitute. In such cases, the option --nodeps skips dependency checking.
rpm -i --nodeps <rpm-file> ( dpkg -i --ignore-depends=<required-package> <deb-file> )

gives the name and version

textutils-2.0e-7 ( ii textutils 2.0-2 The GNU text file processing utilities. )

To get info on a package (query info, status),

rpm -qi <package> ( dpkg -s <package> )

To list libraries and other packages required by a package,

rpm -qR <package> ( dpkg -s <package> | grep Depends )

Note that Debian is far more fastidious about its dependencies; override them only when you are sure what is going on underneath.

To list what other packages require this one (with Debian a removal with the --no-act option to merely test),
rpm -q --whatrequires <package> ( dpkg --purge --no-act <package> )

we can check by attempting

24.2.4

Package queries

.rpm and .deb packages are more than a way of archiving les; otherwise, we could just use .tar les. Each package has its le list stored in a database that can be queried. The following are some of the more useful queries that can be done. Note that these are queries on already installed packages only: To get a list of all packages (query all, llist),

24.2.5 File lists and le queries

installed.

rpm -qa

To get a le list contained by a package ,

Once again, not for les but packages already

241

242

24. Source and Binary Packages

24.2. RedHat and Debian Binary Packages

24. Source and Binary Packages

rpm -ql <package> ( dpkg -L <package> )

( md5sum openssh_2.1.1p4-1_i386.deb )

Package le lists are especially useful for nding what commands and documentation a package provides. Users are often frustrated by a package that they dont know what to do with. Listing les owned by the package is where to start. To nd out what package a le belongs to,

md5sum uses the MD5 mathematical algorithm to calculate a numeric hash value based on the le contents, in this case, 8e8d8e95db7fde99c09e1398e4dd3468. This is identical to password hashing described on page 103. There is no feasible computational method of forging a package to give the same MD5 hash; hence, packagers will often publish their md5sum results on their web page, and you can check these against your own as a security measure.

rpm -qf <filename> ( dpkg -S <filename> )

24.2.7 Special queries

To query a package le that has not been installed, use, for example:
rpm -qp --qf [%{VERSION}\n] <rpm-file> ( dpkg -f <deb-file> Version )

For example, rpm -qf /etc/rc.d/init.d/httpd (or rpm -qf /etc/init.d/httpd) gives apache-mod ssl-1.3.12.2.6.6-1 on my system, and rpm -ql fileutils-4.0w-3 | grep bin gives a list of all other commands from fileutils. A trick to nd all the sibling les of a command in your PATH is:
rpm -ql rpm -qf \which --skip-alias <command> \ ( dpkg -L dpkg -S \which <command> \ | cut -f1 -d: )

Here, VERSION is a query tag applicable to .rpm les. Here is a list of other tags that can be queried: BUILDHOST BUILDTIME CHANGELOG CHANGELOGTEXT CHANGELOGTIME COPYRIGHT DESCRIPTION DISTRIBUTION GROUP LICENSE NAME OBSOLETES OS PACKAGER PROVIDES RELEASE REQUIREFLAGS REQUIRENAME REQUIREVERSION RPMTAG POSTIN RPMTAG POSTUN RPMTAG PREIN RPMTAG PREUN RPMVERSION SERIAL SIZE SOURCERPM SUMMARY VENDOR VERIFYSCRIPT VERSION

24.2.6

Package verication

You sometimes might want to query whether a packages les have been modied since installation (possibly by a hacker or an incompetent system administrator). To verify all packages is time consuming but provides some very instructive output:
rpm -V rpm -qa ( debsums -a )

However, there is not yet a way of saying that the package installed is the real package (see Section 44.3.2). To check this, you need to get your actual .deb or .rpm le and verify it with:
rpm -Vp openssh-2.1.1p4-1.i386.rpm ( debsums openssh_2.1.1p4-1_i386.deb )

For Debian , Version is a control eld. Others are Conffiles Conflicts Depends Description Essential Installed-Size Maintainer Package Pre-Depends Priority Provides Recommends Replaces Section Source Status Suggests Version

Finally, even if you have the package le, how can you be absolutely sure that it is the package that the original packager created, and not some Trojan substitution? Use the md5sum command to check:
md5sum openssh-2.1.1p4-1.i386.rpm

It is further possible to extract all scripts, cong, and control les from a .deb le with: 244

243

24. Source and Binary Packages

24.2. RedHat and Debian Binary Packages

24.3. Source Packages

24. Source and Binary Packages

dpkg -e <deb-file> <out-directory>

24.3

This command creates a directory <out-directory> and places the les in it. You can also dump the package as a tar le with:
dpkg --fsys-tarfile <deb-file>

Source Packages Building RedHat and Debian Packages

or for an .rpm le,

rpm2cpio <rpm-file>

Both RedHat and Debian binary packages begin life as source les from which their binary versions are compiled. Source RedHat packages will end in .src.rpm, and Debian packages will always appear under the source tree in the distribution. The RPM-HOWTO details the building of RedHat source packages, and Debian s dpkgdev and packaging-manual packages contain a complete reference to the Debian package standard and packaging methods (try dpkg -L dpkg-dev and dpkg L packaging-manual). The actual building of RedHat and Debian edition. source packages is not covered in this

Finally, package le lists can be queried with

rpm -qip <rpm-file> ( dpkg -I <deb-file> ) rpm -qlp <rpm-file> ( dpkg -c <deb-file> )

which is analogous to similar queries on already installed packages.

24.2.8

dpkg/apt versus rpm

Only a taste of Debian package management was provided above. Debian has two higher-level tools: APT (Advanced Package Toolwhich comprises the commands aptcache, apt-cdrom, apt-config, and apt-get); and dselect, which is an interactive text-based package selector. When you rst install Debian , I suppose the rst thing you are supposed to do is run dselect (there are other graphical front-ends search on Fresh Meat http://freshmeat.net/), and then install and congure all the things you skipped over during installation. Between these you can do some sophisticated time-saving things like recursively resolving package dependencies through automatic downloadsthat is, just mention the package and APT will nd it and what it depends on, then download and install everything for you. See apt(8), sources.list(5), and apt.conf(5) for more information. There are also numerous interactive graphical applications for managing RPM packages. Most are purely cosmetic. Experience will clearly demonstrate the superiority of Debian packages over most others. You will also notice that where RedHat-like distributions have chosen a selection of packages that they thought you would nd useful, Debian has hundreds of volunteer maintainers selecting what they nd useful. Almost every free U NIX package on the Internet has been included in Debian . 245 246

25.1. Internet Communication

25. Introduction to IP

Table 25.1 (continued) 67 8 9 1011 1215 1619 20IHL*4-1 bits 0-3: Flags, bits 4-15: Offset Time to live (TTL) Type Checksum Source IP address Destination IP address Options + padding to round up to four bytes Data begins at IHL*4 and ends at Length-1

Chapter 25

Introduction to IP
IP stands for Internet Protocol. It is the method by which data is transmitted over the Internet. Version for the mean time is 4, although IP Next Generation (version 6) is in the (slow) process of deployment. IHL is the length of the header divided by 4. TOS (Type of Service) is a somewhat esoteric eld for tuning performance and is not explained here. The Length eld is the length in bytes of the entire packet including the header. The Source and Destination are the IP addresses from and to which the packet is coming/going. The above description constitutes the view of the Internet that a machine has. However, physically, the Internet consists of many small high-speed networks (like those of a company or a university) called Local Area Networks, or LANs. These are all connected to each other by lower-speed long distance links. On a LAN, the raw medium of transmission is not a packet but an Ethernet frame. Frames are analogous to packets (having both a header and a data portion) but are sized to be efcient with particular hardware. IP packets are encapsulated within frames, where the IP packet ts within the Data part of the frame. A frame may, however, be too small to hold an entire IP packet, in which case the IP packet is split into several smaller packets. This group of smaller IP packets is then given an identifying number, and each smaller packet will then have the Identication eld set with that number and the Offset eld set to indicate its position within the actual packet. On the other side of the connection, the destination machine will reconstruct a packet from all the smaller subpackets that have the same Identication eld. The convention for writing an IP address in human readable form is dotted decimal notation like 152.2.254.81, where each number is a byte and is hence in the range of 0 to 255. Hence the entire address space is in the range of 0.0.0.0 to 255.255.255.255. To further organize the assignment of addresses, each 32-bit address is divided into two parts, a network and a host part of the address, as shown in Figure 25.1. 248

25.1 Internet Communication

At a hardware level, network cards are capable of transmitting packets (also called datagrams) of data between one another. A packet contains a small block of, say, 1 kilobyte of data (in contrast to serial lines, which transmit continuously). All Internet communication occurs through transmission of packets, which travel intact, even between machines on opposite sides of the world. Each packet contains a header of 24 bytes or more which precedes the data. Hence, slightly more than the said 1 kilobyte of data would be found on the wire. When a packet is transmitted, the header would obviously contain the destination machine. Each machine is hence given a unique IP addressa 32-bit number. There are no machines on the Internet that do not have an IP address. The header bytes are shown in Table 25.1. Table 25.1 IP header bytes Bytes 0 1 23 45 Description bits 03: Version, bits 47: Internet Header Length (IHL) Type of service (TOS) Length Identication
continues...

247

25. Introduction to IP

25.2. Special IP Addresses

25.3. Network Masks and Addresses

25. Introduction to IP

25.3
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

Network Masks and Addresses

Class A: 0 network part Class B: 1 0 Class C: 1 1 0 network part network part

host part host part host part

Consider again the example of a university with a Class B address. It might have an IP address range of 137.158.0.0 through 137.158.255.255. Assume it was decided that the astronomy department should get 512 of its own IP addresses, 137.158.26.0 through 137.158.27.255. We say that astronomy has a network address of 137.158.26.0. The machines there all have a network mask of 255.255.254.0. A particular machine in astronomy may have an IP address of 137.158.27.158. This terminology is used later. Figure 25.2 illustrates this example.

Figure 25.1 IP address classes The network part of the address designates the LAN, and the host part the particular machine on the LAN. Now, because it was unknown at the time of specication whether there would one day be more LANs or more machines per LAN, three different classes of address were created. Class A addresses begin with the rst bit of the network part set to 0 (hence, a Class A address always has the rst dotted decimal number less than 128). The next 7 bits give the identity of the LAN, and the remaining 24 bits give the identity of an actual machine on that LAN. A Class B address begins with a 1 and then a 0 (rst decimal number is 128 through 191). The next 14 bits give the LAN, and the remaining 16 bits give the machine. Most universities, like the address above, are Class B addresses. Lastly, Class C addresses start with a 1 1 0 (rst decimal number is 192 through 223), and the next 21 bits and then the next 8 bits are the LAN and machine, respectively. Small companies tend use Class C addresses. In practice, few organizations require Class A addresses. A university or large company might use a Class B address but then would have its own further subdivisions, like using the third dotted decimal as a department (bits 16 through 23) and the last dotted decimal (bits 24 through 31) as the machine within that department. In this way the LAN becomes a micro-Internet in itself. Here, the LAN is called a network and the various departments are each called a subnet.
Netmask Network address IP address Host part Dotted IP 255 . 255 . 254 . Binary 1111 1111 1111 1111 1111 1110 0000 0000 1000 1001 1001 1110 0001 1010 0000 0000 1000 1001 1001 1110 0001 1011 1001 1110 0000 0000 0000 0000 0000 0001 1001 1110

137 . 158 . 26 . 0 137 . 158 . 27 . 158 0 . 0 . 1 . 158

Figure 25.2 Dividing an address into network and host portions

25.4

Computers on a LAN

In this section we will use the term LAN to indicate a network of computers that are all more or less connected directly together by Ethernet cables (this is common for small businesses with up to about 50 machines). Each machine has an Ethernet card which is referred to as eth0 throughout all command-line operations. If there is more than one card on a single machine, then these are named eth0, eth1, eth2, etc., and are each called a network interface (or just interface, or sometimes Ethernet port) of the machine. LANs work as follows. Network cards transmit a frame to the LAN, and other network cards read that frame from the LAN. If any one network card transmits a frame, then all other network cards can see that frame. If a card starts to transmit a frame while another card is in the process of transmitting a frame, then a clash is said to have occurred, and the card waits a random amount of time and then tries again. Each network card has a physical address of 48 bits called the hardware address (which is inserted at the time of its manufacture and has nothing to do with IP addresses). Each frame has a destination address in its header that tells what network card it is destined for, so that network cards ignore frames that are not addressed to them. Since frame transmission is governed by the network cards, the destination hardware address must be determined from the destination IP address before a packet is sent to a particular machine. This is done is through the Address Resolution Protocol 250

25.2 Special IP Addresses

Some special-purposes IP addresses are never used on the open Internet. 192.168.0.0 through 192.168.255.255 are private addresses perhaps used inside a local LAN that does not communicate directly with the Internet. 127.0.0.0 through 127.255.255.255 are used for communication with the localhostthat is, the machine itself. Usually, 127.0.0.1 is an IP address pointing to the machine itself. Further, 172.16.0.0 through 172.31.255.255 are additional private addresses for very large internal networks, and 10.0.0.0 through 10.255.255.255 are for even larger ones. 249

25. Introduction to IP

25.5. Conguring Interfaces

25.6. Conguring Routing

25. Introduction to IP

(ARP). A machine will transmit a special packet that asks What hardware address is this IP address? The guilty machine then responds, and the transmitting machine stores the result for future reference. Of course, if you suddenly switch network cards, then other machines on the LAN will have the wrong information, so ARP has timeouts and re-requests built into the protocol. Try typing the command arp to get a list of hardware address to IP mappings.

Interrupt:11 Base address:0xe400 lo

Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 UP LOOPBACK RUNNING MTU:3924 Metric:1 RX packets:53175 errors:0 dropped:0 overruns:0 frame:0 TX packets:53175 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0

25.5 Conguring Interfaces

Most distributions have a generic way to congure your interfaces. Here, however, we rst look at a complete network conguration using only raw networking commands. We rst create a lo interface. This is called the loopback device (and has nothing to do with loopback block devices: /dev/loop? les). The loopback device is an imaginary network card that is used to communicate with the machine itself; for instance, if you are telneting to the local machine, you are actually connecting via the loopback device. The ifconfig (interface configure) command is used to do anything with interfaces. First, run
/sbin/ifconfig lo down /sbin/ifconfig eth0 down

which shows various interesting bits, like the 48-bit hardware address of the network card (hex bytes 00:00:E8:3B:2D:A2).

25.6

Conguring Routing

The interfaces are now active. However, nothing tells the kernel what packets should go to what interface, even though we might expect such behavior to happen on its own. With U NIX, you must explicitly tell the kernel to send particular packets to particular interfaces. Any packet arriving through any interface is pooled by the kernel. The kernel then looks at each packets destination address and decides, based on the destination, where it should be sent. It doesnt matter where the packet came from; once the kernel has the packet, its what its destination address says that matters. It is up to the rest of the network to ensure that packets do not arrive at the wrong interfaces in the rst place. We know that any packet having the network address 127.???.???.??? must go to the loopback device (this is more or less a convention). The command,
/sbin/route add -net 127.0.0.0 netmask 255.0.0.0 lo

to delete any existing interfaces, then run

/sbin/ifconfig lo 127.0.0.1

which creates the loopback interface. Create the Ethernet interface with:

/sbin/ifconfig eth0 192.168.3.9 broadcast 192.168.3.255 netmask 255.255.255.0

adds a route to the network 127.0.0.0, albeit an imaginary one. The eth0 device can be routed as follows:

The broadcast address is a special address that all machines respond to. It is usually the rst or last address of the particular network. Now run

/sbin/route add -net 192.168.3.0 netmask 255.255.255.0 eth0

/sbin/ifconfig

The command to display the current routes is

/sbin/route -n

to view the interfaces. The output will be

eth0 Link encap:Ethernet HWaddr 00:00:E8:3B:2D:A2 inet addr:192.168.3.9 Bcast:192.168.3.255 Mask:255.255.255.0 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:1359 errors:0 dropped:0 overruns:0 frame:0 TX packets:1356 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:100

(-n causes route to not print IP addresses as host names) with the following output:
Kernel IP routing table Destination Gateway 127.0.0.0 0.0.0.0 192.168.3.0 0.0.0.0 Genmask 255.0.0.0 255.255.255.0 Flags Metric Ref U 0 0 U 0 0 Use Iface 0 lo 0 eth0

251

252

25. Introduction to IP

25.6. Conguring Routing

25.7. Conguring Startup Scripts

25. Introduction to IP

This output has the meaning, packets with destination address 127.0.0.0/255.0.0.0 The notation network/mask is often used to denote ranges of IP must be sent to the loopback device, and packets with destination address. address 192.168.3.0/255.255.255.0 must be sent to eth0. Gateway is zero, hence, is not set (see the following commands).

25.7

Conguring Startup Scripts

Most distributions will have a modular and extensible system of startup scripts that initiate networking.

The routing table now routes 127. and 192.168.3. packets. Now we need a route for the remaining possible IP addresses. U NIX can have a route that says to send packets with particular destination IP addresses to another machine on the LAN, from whence they might be forwarded elsewhere. This is sometimes called the gateway machine. The command is:
/sbin/route add -net <network-address> netmask <netmask> gw \ <gateway-ip-address> <interface>

25.7.1 RedHat networking scripts

RedHat systems contain the directory /etc/sysconfig/, which contains conguration les to automatically bring up networking. The le /etc/sysconfig/network-scripts/ifcfg-eth0 contains:

This is the most general form of the command, but its often easier to just type:
/sbin/route add default gw <gateway-ip-address> <interface>

when we want to add a route that applies to all remaining packets. This route is called the default gateway. default signies all packets; it is the same as
/sbin/route add -net 0.0.0.0 netmask 0.0.0.0 gw <gateway-ip-address> \ <interface>

DEVICE=eth0 IPADDR=192.168.3.9 NETMASK=255.255.255.0 NETWORK=192.168.3.0 BROADCAST=192.168.3.255 ONBOOT=yes

The le /etc/sysconfig/network contains:

but since routes are ordered according to netmask, more specic routes are used in preference to less specic ones. Finally, you can set your host name with:

NETWORKING=yes HOSTNAME=cericon.cranzgot.co.za GATEWAY=192.168.3.254

hostname cericon.cranzgot.co.za

You can see that these two les are equivalent to the example conguration done above. These two les can take an enormous number of options for the various protocols besides IP, but this is the most common conguration. The le /etc/sysconfig/network-scripts/ifcfg-lo for the loopback device will be congured automatically at installation; you should never need to edit it. To stop and start networking (i.e., to bring up and down the interfaces and routing), type (alternative commands in parentheses):
/etc/init.d/network stop ( /etc/rc.d/init.d/network stop ) /etc/init.d/network start ( /etc/rc.d/init.d/network start )

A summary of the example commands so far is

/sbin/ifconfig lo down /sbin/ifconfig eth0 down /sbin/ifconfig lo 127.0.0.1 /sbin/ifconfig eth0 192.168.3.9 broadcast 192.168.3.255 netmask 255.255.255.0 /sbin/route add -net 127.0.0.0 netmask 255.0.0.0 lo /sbin/route add -net 192.168.3.0 netmask 255.255.255.0 eth0 /sbin/route add default gw 192.168.3.254 eth0 hostname cericon.cranzgot.co.za

which will indirectly read your /etc/sysconfig/ les. You can add further les, say, ifcfg-eth1 (under /etc/sysconfig/network-scripts/) for a secondary Ethernet device. For example, ifcfg-eth1 could contain 254

Although these 7 commands will get your network working, you should not do such a manual conguration. The next section explains how to congure your startup scripts. 253

25. Introduction to IP

25.7. Conguring Startup Scripts

25.8. Complex Routing a Many-Hop Example

25. Introduction to IP

DEVICE=eth1 IPADDR=192.168.4.1 NETMASK=255.255.255.0 NETWORK=192.168.4.0 BROADCAST=192.168.4.255 ONBOOT=yes

iface lo inet loopback iface eth0 inet static address 192.168.3.9 netmask 255.255.255.0 gateway 192.168.3.254 iface eth1 inet static address 192.168.4.1 netmask 255.255.255.0

and then run echo "1" > /proc/sys/net/ipv4/ip forward to enable packet forwarding between your two interfaces.

and then set ip forward=yes in your /etc/network/options le. Finally, whereas RedHat sets its host name from the line HOSTNAME=. . . in /etc/sysconfig/network, Debian sets it from the contents of the le /etc/hostname, which, in the present case, would contain just
cericon.cranzgot.co.za

25.7.2

Debian networking scripts

Debian , on the other hand, has a directory /etc/network/ containing a le /etc/network/interfaces. As usual, Debian has a neat and clean approach. (See also interfaces(5).) For the same conguration as above, this le would contain:

iface lo inet loopback iface eth0 inet static address 192.168.3.9 netmask 255.255.255.0 gateway 192.168.3.254

25.8

Complex Routing a Many-Hop Example

The le /etc/network/options contains the same forwarding (and some other) options:
ip_forward=no spoofprotect=yes syncookies=no

Consider two distant LANs that need to communicate. Two dedicated machines, one on each LAN, are linked by some alternative method (in this case, a permanent serial line), as shown in Figure 25.3. This arrangement can be summarized by ve machines X, A, B, C, and D. Machines X, A, and B form LAN 1 on subnet 192.168.1.0/26. Machines C and D form LAN 2 on subnet 192.168.1.128/26. Note how we use the /26 to indicate that only the rst 26 bits are network address bits, while the remaining 6 bits are host address bits. This means that we can have at most 26 = 64 IP addresses on each of LAN 1 and 2. Our dedicated serial link comes between machines B and C. Machine X has IP address 192.168.1.1. This machine is the gateway to the Internet. The Ethernet port of machine B is simply congured with an IP address of 192.168.1.2 with a default gateway of 192.168.1.1. Note that the broadcast address is 192.168.1.63 (the last 6 bits set to 1). The Ethernet port of machine C is congured with an IP address of 192.168.1.129. No default gateway should be set until serial line is congured. We will make the network between B and C subnet 192.168.1.192/26. It is effectively a LAN on its own, even though only two machines can ever be connected. Machines B and C will have IP addresses 192.168.1.252 and 192.168.1.253, respectively, on their facing interfaces. 256

type

To stop and start networking (i.e., bring up and down the interfaces and routing),

/etc/init.d/networking stop /etc/init.d/networking start

which will indirectly read your /etc/network/interfaces le. Actually, the /etc/init.d/networking script merely runs the ifup and ifdown commands. See ifup(8). You can alternatively run these commands directly for ner control. We add further interfaces similar to the RedHat example above by appending to the /etc/network/interfaces le. The Debian equivalent is, 255

25. Introduction to IP

25.8. Complex Routing a Many-Hop Example

25. Introduction to IP

Routing on machine B is achieved with the following script, provided the link is up. This script must be executed whenever pppd has negotiated the connection and can therefore be placed in the le /etc/pppd/ip-up, which pppd executes automatically as soon as the ppp0 interface is available:
/sbin/route /sbin/route /sbin/route /sbin/route
5

del add add add

default -net 192.168.1.192 netmask 255.255.255.192 dev ppp0 -net 192.168.1.128 netmask 255.255.255.192 gw 192.168.1.253 default gw 192.168.1.1

echo 1 > /proc/sys/net/ipv4/ip_forward

Our full routing table and interface list for machine B then looks like this RedHat 6 likes to add (redundant) explicit routes to each device. These may not be necessary on your system :
Kernel IP routing table Destination Gateway 192.168.1.2 0.0.0.0 192.168.1.253 0.0.0.0 192.168.1.0 0.0.0.0 192.168.1.192 0.0.0.0 192.168.1.128 192.168.1.253 127.0.0.0 0.0.0.0 0.0.0.0 192.168.1.1 eth0 lo Link inet Link inet Link inet Genmask 255.255.255.255 255.255.255.255 255.255.255.192 255.255.255.192 255.255.255.192 255.0.0.0 0.0.0.0 Flags UH UH U U UG U UG Metric 0 0 0 0 0 0 0 Ref 0 0 0 0 0 0 0 Use 0 0 0 0 0 0 0 Iface eth0 ppp0 eth0 ppp0 ppp0 lo eth0

Figure 25.3 Two remotely connected networks

This is a real-life example with an unreliable serial link. To keep the link up requires pppd and a shell script to restart the link if it dies. The pppd program is covered in Chapter 41. The script for Machine B is:
#!/bin/sh while true ; do pppd lock local mru 296 mtu 296 nodetach nocrtscts nocdtrcts \ 192.168.1.252:192.168.1.253 /dev/ttyS0 115200 noauth \ lcp-echo-interval 1 lcp-echo-failure 2 lcp-max-terminate 1 lcp-restart 1 done

ppp0

encap:Ethernet HWaddr 00:A0:24:75:3B:69 addr:192.168.1.2 Bcast:192.168.1.63 Mask:255.255.255.192 encap:Local Loopback addr:127.0.0.1 Mask:255.0.0.0 encap:Point-to-Point Protocol addr:192.168.1.252 P-t-P:192.168.1.253 Mask:255.255.255.255

On machine C we can similarly run the script,

Note that if the link were an Ethernet link instead (on a second Ethernet card), and/or a genuine LAN between machines B and C (with subnet 192.168.1.252/26), then the same script would be just
/sbin/ifconfig eth1 192.168.1.252 broadcast 192.168.1.255 netmask \ 255.255.255.192

#!/bin/sh while true ; do pppd lock local mru 296 mtu 296 nodetach nocrtscts nocdtrcts \ 192.168.1.253:192.168.1.252 /dev/ttyS0 115200 noauth \ lcp-echo-interval 1 lcp-echo-failure 2 lcp-max-terminate 1 lcp-restart 1 done

and then create routes with

/sbin/route del default /sbin/route add -net 192.168.1.192 netmask 255.255.255.192 dev ppp0 /sbin/route add default gw 192.168.1.252
5

in which case all ppp0 would change to eth1 in the scripts that follow. 257

echo 1 > /proc/sys/net/ipv4/ip_forward

258

25. Introduction to IP

25.9. Interface Aliasing Many IPs on One Physical Card

25.10. Diagnostic Utilities

25. Introduction to IP

Our full routing table for machine C then looks like:

Genmask 255.255.255.255 255.255.255.255 255.255.255.192 255.255.255.192 255.0.0.0 0.0.0.0 Flags UH UH U U U UG Metric 0 0 0 0 0 0 Ref 0 0 0 0 0 0 Use 0 0 0 0 0 0 Iface eth0 ppp0 ppp0 eth0 lo ppp0

Kernel IP routing table Destination Gateway 192.168.1.129 0.0.0.0 192.168.1.252 0.0.0.0 192.168.1.192 0.0.0.0 192.168.1.128 0.0.0.0 127.0.0.0 0.0.0.0 0.0.0.0 192.168.1.252 eth0 lo ppp0 Link inet Link inet Link inet

If you have one network card which you would like to double as several different IP addresses, you can. Simply name the interface eth0:n where n is from 0 to some large integer. You can use ifconfig as before as many times as you like on the same network card
/sbin/ifconfig eth0:0 192.168.4.1 broadcast 192.168.4.255 netmask 255.255.255.0 /sbin/ifconfig eth0:1 192.168.5.1 broadcast 192.168.5.255 netmask 255.255.255.0 /sbin/ifconfig eth0:2 192.168.6.1 broadcast 192.168.6.255 netmask 255.255.255.0

encap:Ethernet HWaddr 00:A0:CC:D5:D8:A7 addr:192.168.1.129 Bcast:192.168.1.191 Mask:255.255.255.192 encap:Local Loopback addr:127.0.0.1 Mask:255.0.0.0 encap:Point-to-Point Protocol addr:192.168.1.253 P-t-P:192.168.1.252 Mask:255.255.255.255

in addition to your regular eth0 device. Here, the same interface can communicate to three LANs having networks 192.168.4.0, 192.168.5.0, and 192.168.6.0. Dont forget to add routes to these networks as above.

25.10

Diagnostic Utilities

Machine D can be congured like any ordinary machine on a LAN. It just sets its default gateway to 192.168.1.129. Machine A, however, has to know to send packets destined for subnet 192.168.1.128/26 through machine B. Its routing table has an extra entry for the 192.168.1.128/26 LAN. The full routing table for machine A is:
Kernel IP routing table Destination Gateway 192.168.1.0 0.0.0.0 192.168.1.128 192.168.1.2 127.0.0.0 0.0.0.0 0.0.0.0 192.168.1.1 Genmask 255.255.255.192 255.255.255.192 255.0.0.0 0.0.0.0 Flags U UG U UG Metric 0 0 0 0 Ref 0 0 0 0 Use 0 0 0 0 Iface eth0 eth0 lo eth0

It is essential to know how to inspect and test your network to resolve problems. The standard U NIX utilities are explained here.

25.10.1

ping

To avoid having to add this extra route on machine A, you can instead add the same route on machine X. This may seem odd, but all that this means is that packets originating from A destined for LAN 2 rst try to go through X (since A has only one route), and are then redirected by X to go through B. The preceding conguration allowed machines to properly send packets between machines A and D and out through the Internet. One caveat: ping sometimes did not work even though telnet did. This may be a peculiarity of the kernel version we were using, **shrug**.

The ping command is the most common network utility. IP packets come in three types on the Internet, represented in the Type eld of the IP header: UDP, TCP, and ICMP. (The rst two, discussed later, represent the two basic methods of communication between two programs running on different machines.) ICMP stands for Internet Control Message Protocol and is a diagnostic packet that is responded to in a special way. Try:
ping metalab.unc.edu

or specify some other well-known host. You will get output like:
PING metalab.unc.edu (152.19.254.81) from 192.168.3.9 : 56(84) bytes of data. 64 bytes from 152.19.254.81: icmp_seq=0 ttl=238 time=1059.1 ms 64 bytes from 152.19.254.81: icmp_seq=1 ttl=238 time=764.9 ms 64 bytes from 152.19.254.81: icmp_seq=2 ttl=238 time=858.8 ms 64 bytes from 152.19.254.81: icmp_seq=3 ttl=238 time=1179.9 ms 64 bytes from 152.19.254.81: icmp_seq=4 ttl=238 time=986.6 ms 64 bytes from 152.19.254.81: icmp_seq=5 ttl=238 time=1274.3 ms 64 bytes from 152.19.254.81: icmp_seq=6 ttl=238 time=930.7 ms

25.9 Interface Aliasing Many IPs on One Physical Card

(The le /usr/src/linux/Documentation/networking/alias.txt contains the kernel documentation on this.) 259

What is happening is that ping is sending ICMP packets to metalab.unc.edu, which is automatically responding with a return ICMP packet. Being able to ping a machine is often the acid test of whether you have a correctly congured and working network interface. Note that some sites explicitly lter out ICMP packets, so, for example, ping cnn.com wont work. 260

25. Introduction to IP

25.10. Diagnostic Utilities

25. Introduction to IP

ping sends a packet every second and measures the time it takes to receive the return packetlike a submarine sonar ping. Over the Internet, you can get times in excess of 2 seconds if the place is remote enough. On a local LAN this delay will drop to under a millisecond. If ping does not even get to the line PING metalab.unc.edu. . . , it means that ping cannot resolve the host name. You should then check that your DNS is set up correctlysee Chapter 27. If ping gets to that line but no further, it means that the packets are not getting there or are not getting back. In all other cases, ping gives an error message reporting the absence of either routes or interfaces.

puts the card into promiscuous mode, meaning that the card is to retrieve all frames regardless of their destination hardware address. Try
tcpdump -n -N -f -i eth0

tcpdump is also discussed in Section 41.5. Deciphering the output of tcpdump is left for now as an exercise for the reader. More on the tcp part of tcpdump in Chapter 26.

25.10.2

traceroute

traceroute is a rather fascinating utility to identify where a packet has been. It uses UDP packets or, with the -I option, ICMP packets to detect the routing path. On my machine, gives
5

traceroute metalab.unc.edu

traceroute to metalab.unc.edu (152.19.254.81), 30 hops max, 38 byte packets 1 192.168.3.254 (192.168.3.254) 1.197 ms 1.085 ms 1.050 ms 2 192.168.254.5 (192.168.254.5) 45.165 ms 45.314 ms 45.164 ms 3 cranzgate (192.168.2.254) 48.205 ms 48.170 ms 48.074 ms 4 cranzposix (160.124.182.254) 46.117 ms 46.064 ms 45.999 ms 5 cismpjhb.posix.co.za (160.124.255.193) 451.886 ms 71.549 ms 173.321 ms 6 cisap1.posix.co.za (160.124.112.1) 274.834 ms 147.251 ms 400.654 ms 7 saix.posix.co.za (160.124.255.6) 187.402 ms 325.030 ms 628.576 ms 8 ndf-core1.gt.saix.net (196.25.253.1) 252.558 ms 186.256 ms 255.805 ms 9 ny-core.saix.net (196.25.0.238) 497.273 ms 454.531 ms 639.795 ms 10 bordercore6-serial5-0-0-26.WestOrange.cw.net (166.48.144.105) 595.755 ms 595.174 ms * 11 corerouter1.WestOrange.cw.net (204.70.9.138) 490.845 ms 698.483 ms 1029.369 ms 12 core6.Washington.cw.net (204.70.4.113) 580.971 ms 893.481 ms 730.608 ms 13 204.70.10.182 (204.70.10.182) 644.070 ms 726.363 ms 639.942 ms 14 mae-brdr-01.inet.qwest.net (205.171.4.201) 767.783 ms * * 15 * * * 16 * wdc-core-03.inet.qwest.net (205.171.24.69) 779.546 ms 898.371 ms 17 atl-core-02.inet.qwest.net (205.171.5.243) 894.553 ms 689.472 ms * 18 atl-edge-05.inet.qwest.net (205.171.21.54) 735.810 ms 784.461 ms 789.592 ms 19 * * * 20 * * unc-gw.ncren.net (128.109.190.2) 889.257 ms 21 unc-gw.ncren.net (128.109.190.2) 646.569 ms 780.000 ms * 22 * helios.oit.unc.edu (152.2.22.3) 600.558 ms 839.135 ms

You can see that there were twenty machines mine and metalab.unc.edu.

enterprise-level web servers have no use in non-U.S. markets: there isnt even the network speed to load (or hops) between such servers, thus making any kind of server speed comparisons superuous.

This is actually a good argument for why

25.10.3

tcpdump

tcpdump watches a particular interface for all the trafc that passes itthat is, all the trafc of all the machines connected to the same hub (also called the segment or network segment). A network card usually grabs only the frames destined for it, but tcpdump 261 262

26.1. The TCP Header

26. TCP and UDP

send a SYN (SYNchronization) packet to the remote TCP server (in this case, a web server). The web server responds with a SYN-ACK packet (ACKnowledge), and nally the client responds with a nal SYN packet. This packet negotiation is unbeknown to the programmer.

Chapter 26

Transmission Control Protocol (TCP) and User Datagram Protocol (UDP)

In the previous chapter we talked about communication between machines in a generic sense. However, when you have two applications on opposite sides of the Atlantic Ocean, being able to send a packet that may or may not reach the other side is not sufcient. What you need is reliable communication. Ideally, a programmer wants to be able to establish a link to a remote machine and then feed bytes in one at a time and be sure that the bytes are being read on the other end, and vice-versa. Such communication is called reliable stream communication. If your only tools are discrete, unreliable packets, implementing a reliable, continuous stream is tricky. You can send single packets and then wait for the remote machine to conrm receipt, but this approach is inefcient (packets can take a long time to get to and from their destination)you really want to be able to send as many packets as possible at once and then have some means of negotiating with the remote machine when to resend packets that were not received. What TCP (Transmission Control Protocol) does is to send data packets one way and then acknowledgment packets the other way, saying how much of the stream has been properly received. We therefore say that TCP is implemented on top of IP. This is why Internet communication is sometimes called TCP/IP. TCP communication has three stages: negotiation, transfer, and detachment.
is all my own terminology. This is also somewhat of a schematic representation.

Transfer: The programmer will use the send() (send(2)) and recv() (recv(2)) function calls to send and receive an actual stream of bytes. The stream of bytes will be broken into packets, and the packets sent individually to the remote application. In the case of the web server, the rst bytes sent would be the line GET /index.html HTTP/1.0<CR><NL><CR><NL>. On the remote side, reply packets (also called ACK packets) are sent back as the data arrives, indicating whether parts of the stream went missing and require retransmission. Communication is full-duplexmeaning that there are streams in both directionsboth data and acknowledge packets are going both ways simultaneously. Detachment: The programmer will use the function call shutdown() and close() (see shutdown() and close(2)) to terminate the connection. A FIN packet will be sent and TCP communication will cease.

26.1

The TCP Header

TCP packets are obviously encapsulated within IP packets. The TCP packet is inside the Data begins at. . . part of the IP packet. A TCP packet has a header part and a data part. The data part may sometimes be empty (such as in the negotiation stage). Table 26.1 shows the full TCP/IP header. Table 26.1 Combined TCP and IP header Bytes (IP) 0 1 23 45 67 8 9 1011 1215 1619 20IHL*4-1 Bytes (TCP) Description Bits 03: Version, Bits 47: Internet Header Length (IHL) Type of service (TOS) Length Identication Bits 0-3: Flags, bits 4-15: Offset Time to live (TTL) Type Checksum Source IP address Destination IP address Options + padding to round up to four bytes Description
continues...

This

Negotiation The client application (say, a web browser) rst initiates the connection by using a connect() (see connect(2)) function. This causes the kernel to 263

264

26. TCP and UDP

26.2. A Sample TCP Session

26. TCP and UDP

Table 26.1 (continued) 01 Source port 23 Destination port 47 Sequence number 811 Acknowledgment number 12 Bits 03: number of bytes of additional TCP options / 4 13 Control 1415 Window 1617 Checksum 1819 Urgent pointer 20(20 + options * 4) Options + padding to round up to four bytes TCP data begins at IHL * 4 + 20 + options * 4 and ends at Length - 1 The minimum combined TCP/IP header is thus 40 bytes. With Internet machines, several applications often communicate simultaneously. The Source port and Destination port elds identify and distinguish individual streams. In the case of web communication, the destination port (from the clients point of view) is port 80, and hence all outgoing trafc will have the number 80 lled in this eld. The source port (from the clients point of view) is chosen randomly to any unused port number above 1024 before the connection is negotiated; these, too, are lled into outgoing packets. No two streams have the same combinations of source and destination port numbers. The kernel uses the port numbers on incoming packets to determine which application requires those packets, and similarly for the remote machine. Sequence number is the offset within the stream that this particular packet of data belongs to. The Acknowledge number is the point in the stream up to which all data has been received. Control is various other ag bits. Window is the maximum amount that the receiver is prepared to accept. Checksum is used to verify data integrity, and Urgent pointer is for interrupting the stream. Data needed by extensions to the protocol are appended after the header as options.
5

[root@cericon]# host cnn.com cnn.com has address 207.25.71.20

Now, in one window we run

which says to list all packets having source (src) or destination (dst) addresses of either us or CNN. Then we use the HTTP protocol to grab the page. Type in the HTTP command GET / HTTP/1.0 and then press twice (as required by the HTTP protocol). The rst and last few lines of the sessions are shown below:
[root@cericon root]# telnet 207.25.71.20 80 Trying 207.25.71.20... Connected to 207.25.71.20. Escape character is ]. GET / HTTP/1.0 HTTP/1.0 200 OK Server: Netscape-Enterprise/2.01 Date: Tue, 18 Apr 2000 10:55:14 GMT Set-cookie: CNNid=cf19472c-23286-956055314-2; expires=Wednesday, 30-Dec-2037 16:00:00 GMT; path=/; domain=.cnn.com Last-modified: Tue, 18 Apr 2000 10:55:14 GMT Content-type: text/html <HTML> <HEAD> <TITLE>CNN.com</TITLE> <META http-equiv="REFRESH" content="1800">
20

[root@cericon]# tcpdump \ ( src 192.168.3.9 and dst 207.25.71.20 ) or ( src 207.25.71.20 and dst 192.168.3.9 ) Kernel filter, protocol ALL, datagram packet socket tcpdump: listening on all devices

<SCRIPT src="/virtual/2000/code/main.js" language="javascript"></SCRIPT> <LINK rel="stylesheet" href="/virtual/2000/style/main.css" type="text/css"> <SCRIPT language="javascript" type="text/javascript"> <!--// if ((navigator.platform==MacPPC)&&(navigator.ap .............. ..............

26.2 A Sample TCP Session

It is easy to see TCP working by using telnet. You are probably familiar with using telnet to log in to remote systems, but telnet is actually a generic program to connect to any TCP socket as we did in Chapter 10. Here we will try connect to cnn.coms web page. We rst need to get an IP address of cnn.com: 265

</BODY> </HTML> Connection closed by foreign host.

The above commands produce the front page of CNNs web site in raw HTML. This is easy to paste into a le and view off-line. In the other window, tcpdump is showing us what packets are being exchanged. tcpdump nicely shows us host names instead of IP addresses and the letters www instead of the port number 80. The local random port in this case was 4064. 266

26. TCP and UDP

26.2. A Sample TCP Session

26.3. User Datagram Protocol (UDP)

26. TCP and UDP

[root@cericon]# tcpdump \ ( src 192.168.3.9 and dst 207.25.71.20 ) or ( src 207.25.71.20 and dst 192.168.3.9 ) Kernel filter, protocol ALL, datagram packet socket tcpdump: listening on all devices 12:52:35.467121 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: S 2463192134:2463192134(0) win 32120 <mss 1460,sackOK,timestamp 154031689 0,nop,wscale 0 12:52:35.964703 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: S 4182178234:4182178234(0) ack 2463192135 win 10136 <nop,nop,timestamp 1075172823 154031 12:52:35.964791 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: . 1:1(0) ack 1 win 32120 <nop,nop,timestamp 154031739 1075172823> (DF) 12:52:46.413043 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: P 1:17(16) ack 1 win 32120 <nop,nop,timestamp 154032784 1075172823> (DF) 12:52:46.908156 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: . 1:1(0) ack 17 win 10136 <nop,nop,timestamp 1075173916 154032784> 12:52:49.259870 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: P 17:19(2) ack 1 win 32120 <nop,nop,timestamp 154033068 1075173916> (DF) 12:52:49.886846 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: P 1:278(277) ack 19 win 10136 <nop,nop,timestamp 1075174200 154033068> 12:52:49.887039 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: . 19:19(0) ack 278 win 31856 <nop,nop,timestamp 154033131 1075174200> (DF) 12:52:50.053628 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: . 278:1176(898) ack 19 win 10136 <nop,nop,timestamp 1075174202 154033068> 12:52:50.160740 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: P 1176:1972(796) ack 19 win 10136 <nop,nop,timestamp 1075174202 154033068> 12:52:50.220067 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: . 19:19(0) ack 1972 win 31856 <nop,nop,timestamp 154033165 1075174202> (DF) 12:52:50.824143 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: . 1972:3420(1448) ack 19 win 10136 <nop,nop,timestamp 1075174262 154033131> 12:52:51.021465 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: . 3420:4868(1448) ack 19 win 10136 <nop,nop,timestamp 1075174295 154033165> .............. ..............
35

. 65201:65201(0) ack 20 win 10136 <nop,nop,timestamp 1075177315 154035895> 103 packets received by filter

The preceding output requires some explanation: Line 5, 7, and 9 are the negotiation stage. tcpdump uses the format <Sequence number>:<Sequence number + data length>(<data length>) on each line to show the context of the packet within the stream. Sequence number, however, is chosen randomly at the outset, so tcpdump prints the relative sequence number after the rst two packets to make it clearer what the actual position is within the stream. Line 11 is where I pressed Enter the rst time, and Line 15 was Enter with an empty line. The ack 19s indicates the point to which CNNs web server has received incoming data; in this case we only ever typed in 19 bytes, hence the web server sets this value in every one of its outgoing packets, while our own outgoing packets are mostly empty of data. Lines 61 and 63 are the detachment stage. More information about the tcpdump output can be had from tcpdump(8) under the section TCP Packets.

26.3

User Datagram Protocol (UDP)

12:53:13.856919 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: . 19:19(0) ack 53204 win 30408 <nop,nop,timestamp 154035528 1075176560> (DF) 12:53:14.722584 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: . 53204:54652(1448) ack 19 win 10136 <nop,nop,timestamp 1075176659 154035528> 12:53:14.722738 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: . 19:19(0) ack 54652 win 30408 <nop,nop,timestamp 154035615 1075176659> (DF) 12:53:14.912561 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: . 54652:56100(1448) ack 19 win 10136 <nop,nop,timestamp 1075176659 154035528> 12:53:14.912706 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: . 19:19(0) ack 58500 win 30408 <nop,nop,timestamp 154035634 1075176659> (DF) 12:53:15.706463 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: . 58500:59948(1448) ack 19 win 10136 <nop,nop,timestamp 1075176765 154035634> 12:53:15.896639 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: . 59948:61396(1448) ack 19 win 10136 <nop,nop,timestamp 1075176765 154035634> 12:53:15.896791 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: . 19:19(0) ack 61396 win 31856 <nop,nop,timestamp 154035732 1075176765> (DF) 12:53:16.678439 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: . 61396:62844(1448) ack 19 win 10136 <nop,nop,timestamp 1075176864 154035732> 12:53:16.867963 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: . 62844:64292(1448) ack 19 win 10136 <nop,nop,timestamp 1075176864 154035732> 12:53:16.868095 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: . 19:19(0) ack 64292 win 31856 <nop,nop,timestamp 154035829 1075176864> (DF) 12:53:17.521019 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064: FP 64292:65200(908) ack 19 win 10136 <nop,nop,timestamp 1075176960 154035829> 12:53:17.521154 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: . 19:19(0) ack 65201 win 31856 <nop,nop,timestamp 154035895 1075176960> (DF) 12:53:17.523243 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: F 19:19(0) ack 65201 win 31856 <nop,nop,timestamp 154035895 1075176960> (DF) 12:53:20.410092 eth0 > cericon.cranzgot.co.za.4064 > www1.cnn.com.www: F 19:19(0) ack 65201 win 31856 <nop,nop,timestamp 154036184 1075176960> (DF) 12:53:20.940833 eth0 < www1.cnn.com.www > cericon.cranzgot.co.za.4064:

You dont always need reliable communication. Sometimes you want to directly control packets for efciency, or because you dont really mind if packets get lost. Two examples are name server communications, for which single packet transmissions are desired, or voice transmissions for which reducing lag time is more important than data integrity. Another is NFS (Network File System), which uses UDP to implement exclusively high bandwidth data transfer. With UDP the programmer sends and receives individual packets, again encapsulated within IP. Ports are used in the same way as with TCP, but these are merely identiers and there is no concept of a stream. The full UDP/IP header is listed in Table 26.2. Table 26.2 Combined UDP and IP header Bytes (IP) 0 1 23 45 67 Description bits 03: Version, bits 47: Internet Header Length (IHL) Type of service (TOS) Length Identication bits 0-3: Flags, bits 4-15: Offset
continues...

267

268

26. TCP and UDP

26.4. /etc/services File

26.5. Encrypting and Forwarding TCP

26. TCP and UDP

Table 26.2 (continued) 8 Time to live (TTL) 9 Type 1011 Checksum 1215 Source IP address 1619 Destination IP address 20(IHL * 4 - 1) Options + padding to round up to four bytes Bytes (UDP) Description 01 Source port 23 Destination port 45 Length 67 Checksum UDP data begins at IHL * 4 + 8 and ends at Length - 1
15

26.4

/etc/services File

Various standard port numbers are used exclusively for particular types of services. Port 80 is for web as shown earlier. Port numbers 1 through 1023 are reserved for such standard services and each is given a convenient textual name. All services are dened for both TCP as well as UDP, even though there is, for example, no such thing as UDP FTP access. Port numbers below 1024 are used exclusively for root uid programs such as mail, DNS, and web services. Programs of ordinary users are not allowed to bind to Port binding is where a program reserves a port for listening for an incoming ports below 1024. The place where connection, as do all network services. Web servers, for example, bind to port 80. these ports are dened is in the /etc/services le. These mappings are mostly for descriptive purposesprograms can look up port names from numbers and visa versa. The /etc/services le has nothing to do with the availability of a service.

ftp-data ftp fsp ssh ssh telnet smtp time time rlp nameserver whois domain domain mtp bootps bootps bootpc bootpc tftp gopher gopher rje finger www www

20/tcp 21/tcp 21/udp 22/tcp 22/udp 23/tcp 25/tcp 37/tcp 37/udp 39/udp 42/tcp 43/tcp 53/tcp 53/udp 57/tcp 67/tcp 67/udp 68/tcp 68/udp 69/udp 70/tcp 70/udp 77/tcp 79/tcp 80/tcp 80/udp

fspd # SSH Remote Login Protocol # SSH Remote Login Protocol mail timserver timserver resource name nicname nameserver nameserver

# resource location # IEN 116 # name-domain server # deprecated # BOOTP server # BOOTP client

# Internet Gopher netrjs http # WorldWideWeb HTTP # HyperText Transfer Protocol

26.5

Encrypting and Forwarding TCP

The TCP stream can easily be reconstructed by anyone listening on a wire who happens to see your network trafc, so TCP is known as an inherently insecure service. We would like to encrypt our data so that anything captured between the client and server will appear garbled. Such an encrypted stream should have several properties: 1. It should ensure that the connecting client really is connecting to the server in question. In other words it should authenticate the server to ensure that the server is not a Trojan. 2. It should prevent any information being gained by a snooper. This means that any trafc read should appear cryptographically garbled. 3. It should be impossible for a listener to modify the trafc without detection. The above is relatively easily accomplished with at least two packages. Take the example where we would like to use POP3 to retrieve mail from a remote machine. First, we can verify that POP3 is working by logging in on the POP3 server. Run a telnet to port 110 (i.e., the POP3 service) as follows: 270

Here is an extract of the /etc/services.

1/tcp 7/tcp 7/udp 9/tcp 9/udp 11/tcp 13/tcp 13/udp 15/tcp 17/tcp 18/tcp 18/udp # TCP port service multiplexer

tcpmux echo echo discard discard systat daytime daytime netstat qotd msp msp

sink null sink null users

quote # message send protocol # message send protocol

269

26. TCP and UDP

26.5. Encrypting and Forwarding TCP

26. TCP and UDP

telnet localhost 110 Connected to localhost.localdomain. Escape character is ]. +OK POP3 localhost.localdomain v7.64 server ready QUIT +OK Sayonara Connection closed by foreign host.

Client
(telnet locahost 12345)

(ssh . . . 12345:pop:110 pop) 12345

For our rst example, we use the OpenSSH package. We can initialize and run the sshd Secure Shell daemon if it has not been initialized before. The following commands would be run on the POP3 server:
ssh-keygen -b 1024 -f /etc/ssh/ssh_host_key -q -N ssh-keygen -d -f /etc/ssh/ssh_host_dsa_key -q -N sshd

POP3 Server

(sshd)

To create an encrypted channel shown in Figure 26.1, we use the ssh client login program in a special way. We would like it to listen on a particular TCP port and then encrypt and forward all trafc to the remote TCP port on the server. This is known as (encrypted) port forwarding. On the client machine we choose an arbitrary unused port to listen on, in this case 12345: where <user> is the name of a shell account on the POP3 server. Finally, also on the client machine, we run:
telnet localhost 12345 Connected to localhost.localdomain. Escape character is ]. +OK POP3 localhost.localdomain v7.64 server ready QUIT +OK Sayonara Connection closed by foreign host.
ssh -C -c arcfour -N -n -2 -L 12345:<pop3-server.doma.in>:110 \ <pop3-server.doma.in> -l <user> -v

(ipop3d)

110

Figure 26.1 Forwarding between two machines

tocol has not been validated by the community at large (and therefore should be used with caution). On the server machine you can just type secure-mcserv. On the client run
forward <user>@<pop3-server.doma.in> <pop3-server.doma.in>:110 \ 12345 --secure -z -K 1024

Here we get results identical to those above, because, as far as the server is concerned, the POP3 connection comes from a client on the server machine itself, unknowing of the fact that it has originated from sshd, which in turn is forwarding from a remote ssh client. In addition, the -C option compresses all data (useful for low-speed connections). Also note that you should generally never use any encryption besides arcfour and SSH Protocol 2 (option -2). The second method is the forward program of the mirrordir package. It has a unique encryption protocol that does much of what OpenSSH can, although the pro271

and then run telnet 12345 to test as before. With forwarding enabled you can use any POP3 client as you normally would. Be sure, though, to set your host and port addresses to localhost and 12345 within your POP3 client. This example can, of course, be applied to almost any service. Some services will not work if they do special things like create reverse TCP connections back to the client (for example, FTP). Your luck may vary.

272

27.2. Resolving DNS Names to IP Addresses

27. DNS and Name Resolution

.gov A U.S. government organization. .edu A U.S. university. .mil A U.S. military department. .int An organization established by international treaties.

Chapter 27

.org A U.S. or nonprot organization. In fact, anyone can have a .org domain. .net An Internet service provider (ISP). In fact, any bandwidth reseller, IT company, or any company at all might have a .net domain.

DNS and Name Resolution

We know that each computer on the Internet has its own IP address. Although this address is sufcient to identify a computer for purposes of transmitting packets, it is not particularly accommodating to people. Also, if a computer were to be relocated, we would like to still identify it by the same name. Hence, each computer is given a descriptive textual name. The basic textual name of a machine is called the unqualied host name This is my own terminology. and is usually less than eight characters and contains only lowercase letters and numbers (and especially no dots). Groups of computers have a domain name. The full name of machine is unqualied host name.domain name and is called the fully qualied host name Standard or the qualied host name. My terminology. For example, my computer terminology. is cericon. The domain name of my company is cranzgot.co.za, and hence the qualied host name of my computer is cericon.cranzgot.co.za, although the IP address might be 160.123.76.9.

Besides the above, the domain could end in a two-letter country code. The complete list of country codes is given in Table 27.1. The .us domain is rarely used, since in the United States .com, .edu, .org, .mil, .gov, .int, or .net are mostly used. Within each country, a domain may have things before it for better description. Each country may implement a different structure. Some examples are: .co.za A South African company. (za = Zuid Afrika, from Dutch.) .org.za A South African nonprot organization. .ac.za A South African academic university. .edu.au An australian tertiary educational institution. .gov.za A South African government organization. Note that a South African company might choose a .com domain instead of a .co.za domain. The Internet has become more commercialized than organized, meaning that anyone can pretty much register any domain that is not already taken.

Often the word domain is synonymous with domain name, and the host name on its own can mean either the qualied or unqualied host name. This system of naming computers is called the Domain Name System (DNS)

27.2 27.1 Top-Level Domains (TLDs)

Domains always end in a standard set of things. Here is a complete list of things that the last bit of a domain can be. .com A U.S. or international company proper. In fact, any organization might have a .com domain. 273

Resolving DNS Names to IP Addresses

In practice, a user will type a host name (say, www.cranzgot.co.za) into some application like a web browser. The application has to then try nd the IP address associated with that name, in order to send packets to it. This section describes the query structure used on the Internet so that everyone can nd out anyone elses IP address. An obvious lookup infrastructure might involve distributing a long table of host name vs. IP numbers to every machine on the Internet. But as soon as you have more than a few thousand machines, this approach becomes impossible. 274

27. DNS and Name Resolution

27.2. Resolving DNS Names to IP Addresses

27. DNS and Name Resolution

Table 27.1 ISO country codes

.af Afghanistan .al Albania .dz Algeria .as American samoa .ad Andorra .ao Angola .ai Anguilla .aq Antarctica .ag Antigua and barbuda .ar Argentina .am Armenia .aw Aruba .au Australia .at Austria .az Bzerbaijan .bs Bahamas .bh Bahrain .bd Bangladesh .bb Barbados .be Belgium .bz Belize .bj Benin .bm Bermuda .bt Bhutan .bo Bolivia .ba Bosnia Hercegovina .bw Botswana .bv Bouvet Island .br Brazil .io British Indian Ocean Territory .bn Brunei Darussalam .bg Bulgaria .bf Burkina Faso .bi Burundi .by Celarus .kh Cambodia .cm Cameroon .ca Canada .cv Cape Verde .ky Cayman Islands .cf Central African Rep. .td Chad .cl Chile .cn China .cx Christmas Island .cc Cocos (Keeling) Islands .co Colombia .km Comoros .cg Congo .ck Cook Islands .cr Costa Rica .ci Cote Divoire .hr Croatia .cu Cuba .cy Cyprus .cz Czech Rep. .cs Dzechoslovakia .dk Denmark .dj Djibouti .dm Dominica .do Eominican Rep. .tp East Timor .ec Ecuador .eg Egypt .sv El Salvador .gq Equatorial Guinea .ee Estonia .et Fthiopia .fk Falkland Islands (Malvinas) .fo Faroe Islands .fj Fiji .fi Finland .fr France .gf French Guiana .pf French Polynesia .tf Grench Southern Territories .ga Gabon .gm Gambia .ge Georgia .de Germany .gh Ghana .gi Gibraltar .gr Greece .gl Greenland .gd Grenada .gp Guadeloupe .gu Guam .gt Guatemala .gn Guinea .gw Guinea-Bissau .gy Huyana .ht Haiti .hm Heard and Mc Donald Islands .hn Honduras .hk Hong Kong .hu Iungary .is Iceland .in India .id Indonesia .ir Iran (Islamic Rep. of) .iq Iraq .ie Ireland .il Israel .it Jtaly .jm Jamaica .jp Japan .jo Kordan .kz Kazakhstan .ke Kenya .ki Kiribati .kp Korea, Demo. Peoples Rep.of .kr Korea, Rep. of .kw Kuwait .kg Lyrgyzstan .la Lao Peoples Demo. Rep. .lv Latvia .lb Lebanon .ls Lesotho .lr Liberia .ly Libyan Arab Jamahiriya .li Liechtenstein .lt Lithuania .lu Muxembourg .mo Macau .mg Madagascar .mw Malawi .my Malaysia .mv Maldives .ml Mali .mt Malta .mh Marshall Islands .mq Martinique .mr Mauritania .mu Mauritius .mx Mexico .fm Micronesia .md Moldova, Rep. of .mc Monaco .mn Mongolia .ms Montserrat .ma Morocco .mz Mozambique .mm Nyanmar .na Namibia .nr Nauru .np Nepal .nl Netherlands .an Netherlands Antilles .nt Neutral Zone .nc New Caledonia .nz New Zealand .ni Nicaragua .ne Niger .ng Nigeria .nu Niue .nf Norfolk Island .mp Northern Mariana Islands .no Oorway .om Pman .pk Pakistan .pw Palau .pa Panama .pg Papua New Guinea .py Paraguay .pe Peru .ph Philippines .pn Pitcairn .pl Poland .pt Portugal .pr Querto Rico .qa Ratar .re Reunion .ro Romania .ru Russian Federation .rw Swanda .sh St. Helena .kn Saint Kitts and Nevis .lc Saint Lucia .pm St. Pierre and Miquelon .vc St. Vincent and the Grenadines .ws Samoa .sm San Marino .st Sao Tome and Principe .sa Saudi Arabia .sn Senegal .sc Seychelles .sl Sierra Leone .sg Singapore .sk Slovakia .si Slovenia .sb Solomon Islands .so Somalia .za South Africa .es Spain .lk Sri Lanka .sd Sudan .sr Suriname .sj Svalbard and Jan Mayen Is. .sz Swaziland .se Sweden .ch Switzerland .sy Tyrian Arab Rep. .tw Taiwan, Province of China .tj Tajikistan .tz Tanzania, United Rep. of .th Thailand .tg Togo .tk Tokelau .to Tonga .tt Trinidad and Tobago .tn Tunisia .tr Turkey .tm Turkmenistan .tc Turks and Caicos Islands .tv Uuvalu .ug Uganda .ua Ukraine .ae United Arab Emirates .gb United Kingdom .us United States .um US Minor Outlying Islands .uy Uruguay .su USSR .uz Vzbekistan .vu Vanuatu .va Vatican City State (Holy See) .ve Venezuela .vn Viet Nam .vg Virgin Islands (British) .vi Wirgin Islands (U.S.) .wf Wallis and Futuna Islands .eh Yestern Sahara .ye Yemen, Rep. of .yu Zugoslavia .zr Zaire .zm Zambia .zw Zimbabwe

There are computers that service requests for IP numbersmillions of them. They are called name servers (or DNS servers), and a request is called a DNS lookup (or just a lookup). However, each name server only has information about a specic part of the Internet, and they constantly query each other. There are 13 root name servers on the Internet.
ftp://ftp.rs.internic.net/domain/named.root. a.root-servers.net b.root-servers.net c.root-servers.net d.root-servers.net e.root-servers.net f.root-servers.net g.root-servers.net h.root-servers.net i.root-servers.net j.root-servers.net k.root-servers.net l.root-servers.net m.root-servers.net 198.41.0.4 128.9.0.107 192.33.4.12 128.8.10.90 192.203.230.10 192.5.5.241 192.112.36.4 128.63.2.53 192.36.148.17 198.41.0.10 193.0.14.129 198.32.64.12 202.12.27.33

This list can be gotten from

Each country also has a name server, and in turn each organization has a name server. Each name server only has information about machines in its own domain, as well as information about other name servers. The root name servers only have information on the IP addresses of the name servers of .com, .edu, .za, etc. The .za name server only has information on the IP addresses of the name servers of .org.za, .ac.za, .co.za, etc. The .co.za name server only has information on the name servers of all South African companies, like .cranzgot.co.za, .icon.co.za, .mweb.co.za, etc. The .cranzgot.co.za, name server only has info on the machines at Cranzgot Systems, like www.cranzgot.co.za. Your own machine will dened in its conguration les a name server that is geographically close to it. The responsibilities of this name server will be to directly answer any queries about its own domain that it has information about and to answer any other queries by querying as many other name servers on the Internet as is necessary.

Another imaginary infrastructure might have one huge computer on the Internet somewhere whose IP address is known by everyone. This computer would be responsible for servicing requests for IP numbers, and the said application running on your local machine would just query this big machine. Of course, with billions of machines out there, this approach will obviously create far too much network trafc. Actually,
some Microsoft LANs kind of work this waythat is, not very well.

27.2.2 The name resolution process

Now our application is presented with www.cranzgot.co.za. The following sequence of lookups takes place to resolve this name into an IP address. This procedure is called host name resolution and the algorithm that performs this operation is called the resolver. 276

27.2.1

The Internet DNS infrastructure

The DNS structure on the Internet actually works like this. 275

27. DNS and Name Resolution

27.3. Conguring Your Local Machine

27. DNS and Name Resolution

1. The application checks certain special databases on the local machine. If it can get an answer directly from them, it proceeds no further. 2. The application looks up a geographically close name server from the local machines conguration le. Lets say this machine is called ns. 3. The application queries ns with www.cranzgot.co.za?. 4. ns determines whether that IP has been recently looked up. If it has, there is no need to ask further, since the result would be stored in a local cache. 5. ns checks whether the domain is local. That is, whether it is a computer about which it has direct information. In this case, this would only be true if the ns were cranzgot.co.zas very own name server. 6. ns strips out the TLD (top level domain) .za. It queries a root name server, asking what name server is responsible for .za. The answer will be ucthpx.uct.ac.za of IP address 137.158.128.1. 7. ns strips out the next highest domain co.za It queries 137.158.128.1, asking what name server is responsible for .co.za. The answer will be secdns1.posix.co.za of IP address 160.124.112.10. 8. ns strips out the next highest domain cranzgot.co.za. It queries 160.124.112.10, asking what name server is responsible for cranzgot.co.za. The answer will be pizza.cranzgot.co.za of IP address 196.28.123.1. 9. ns queries 196.28.123.1 asking for the IP address of www.cranzgot.co.za. The answer will be 160.123.176.1. 10. ns returns the result to the application. 11. ns stores each of these results in a local cache with an expiration date, to avoid having to look them up a second time.

gethostbyname(), hence all these conguration les really belong to the C library packages glibc or The following are equivalent libc. However, this is a detail you need not be concerned about.

to steps 1, 2, and 3 above, with the details of the conguration les lled in. The conguration les that follow are taken from an actual installation. 1. The application checks the le /etc/host.conf. This le will usually have a line order hosts,bind in it, specifying that it should rst (hosts) check the local database le /etc/hosts, and then (bind) query the name server specied in /etc/resolv.conf. The le /etc/hosts contains a plain text list of IP addresses and names. An example is given below. If the application can get an answer directly from /etc/hosts, it proceeds no further. 2. The application checks in the le /etc/resolv.conf for a line nameserver <nameserver>. There can actually be three of these lines so that if one name server fails, the application can try the next in turn. 3. The application sends to the name server a query with the host name. If the host name is unqualied, then the application, before trying the query, appends to the host name a local domain name. A line search <domain1> <domain2> ... <domainN> may appear in the conguration le to facilitate this. A query is made with each of <domain1>, <domain2> etc. appended in turn until the query successfully returns an IP. This just saves you having to type in the full host name for computers within your own organization. 4. The name server proceeds with the hierarchical queries described from step 4 onward.

The /etc/hosts le should look something like this:

localhost.localdomain cericon.cranzgot.co.za pepper.cranzgot.co.za onion.cranzgot.co.za localhost cericon pepper onion

127.0.0.1 192.168.3.9 192.168.3.10 192.168.2.1

27.3 Conguring Your Local Machine

We referred to conguration les above. These are actually the les: /etc/host.conf, /etc/hosts, and /etc/resolv.conf. These are the three and only les that specify how all applications are going to look up IP numbers; and have nothing to do with the conguration les of the name server daemon itself, even though a name server daemon might be running on the local machine. When an application needs to look up a host name, it goes through the following procedure. What is actually happening is that the application is making a library call to the function

The hosts pepper, cericon, and onion are the hosts that this machine has the most communication with, and hence are listed here. cericon is the local machine and must be listed. You can list any hosts to which you want fast lookups, or hosts that might need to be known in spite of name servers being down. The /etc/host.conf might look like this. All of the lines are optional:

order hosts, bind, nis trim some.domain spoofalert nospoof

277

278

27. DNS and Name Resolution

27.3. Conguring Your Local Machine

27. DNS and Name Resolution

multi reorder

options ndots:1 timeout:30 attempts:2 rotate no-check-names inet6

order The order in which lookups are done. Dont try ddling with this value. It never seems to have any effect. You should leave it as order hosts,bind (or order hosts,bind,nis if you are using NISsearch for the NISHOWTO on the web.) Once again, bind means to then go and check the /etc/resolv.conf which holds the name server query options. trim Strip the domain some.domain from the end of a host name before trying a lookup. You will probably never require this feature. spoofalert Try reverse lookups on a host name after looking up the IP (i.e., do a query to nd the name from the IP). If this query does not return the correct result, it could mean that some machine is trying to make it look like it is someone it really isnt. This is a hackers trick called spoong. spoofalert warns you of such attempts in your log le /var/log/messages. nospoof Disallow results that fail the spoof test. multi on Return more than one result if there are aliases. Actually, a host can have several IP numbers, and an IP number can have several host names. Consider a computer that might want more than one name (ftp.cranzgot.co.za and www.cranzgot.co.za are the same machine.) Or a machine that has several networking cards and an IP address for each. This option should always be turned on. multi off is the alternative. Most applications use only the rst value returned. reorder If more than one IP is returned by a lookup, then sort that list according to the IP that has the most convenient network route. Despite this array of options, an /etc/host.conf le almost always looks simply like
order multi hosts, bind on

nameserver Species a name server to query. No more than three may be listed. The point of having more than one is to safeguard against a name server being down; the next in the list will then be queried. search If given a host name with less than ndots dots (i.e., 1 in this case), add each of the domains in turn to the host name, trying a lookup with each. This option allows you to type in an unqualied host name and the application work out what organization it is belongs to from the search list. You can have up to six domains, but then queries would be time consuming. domain The line domain ct.cranzgot.co.za is the same as search ct.cranzgot.co.za cranzgot.co.za co.za. Always use search explicitly instead of domain to reduce the number of queries to a minimum. sortlist If more than one host is returned, sort them according to the following network/masks. options Various additional parameters can be specied in this one line: ndots Explained under search above. The default is 1. timeout How long to wait before considering a query to have failed. The default is 30 seconds. attempts Number of attempts to make before failing. The default is 2. This means that a down name server will cause your application to wait 1 full minute before deciding that it cant resolve the IP. rotate Try the name servers in round robin fashion. This distributes load across name servers. no-check-names Dont check for invalid characters in host names. inet6 The man page for resolv.conf (resolver(5)) says:
inet6 sets RES_USE_INET6 in _res.options . This has the effect of trying a AAAA query before an A query inside the gethostbyname function, and of mapping IPv4 responses in IPv6 tunnelled form if no AAAA records are found but an A record set exists.

An AAAA query is a 128-bit next generation, or IPV6 Internet address. Despite this array of options, an /etc/resolv.conf le almost always looks simply like:
nameserver 192.168.2.254 search cranzgot.co.za

The /etc/resolv.conf le could look something like this:

nameserver 192.168.2.1 nameserver 160.123.76.1 nameserver 196.41.0.131 search cranzgot.co.za ct.cranzgot.co.za uct.ac.za sortlist 192.168.3.0/255.255.255.0 192.168.2.0/255.255.255.0

280

279

27. DNS and Name Resolution

27.4. Reverse Lookups

27.7. The nslookup Command

27. DNS and Name Resolution

27.4 Reverse Lookups

A reverse lookup, mentioned under nospoof, is the determining of the host name from the IP address. The course of queries is similar to forward lookups using part of the IP address to nd out what machines are responsible for what ranges of IP address. A forward lookup is an ordinary lookup of the IP address from the host name.

Now enter:

(Note that original BSD whois worked like whois -h <host> <user>.) You will get a response like this:
[rs.internic.net] Whois Server Version 1.1

whois cnn.com@rs.internic.net

27.5

Authoritative for a Domain

I have emphasized that name servers only hold information for their own domains. Any other information they may have about another domain is cached, temporary data that has an expiration date attached to it. The domain that a name server has information about is said to be the domain that a name server is authoritative for. Alternatively we say: a name server is authoritative for the domain. For instance, the server ns2.cranzgot.co.za is authoritative for the domain cranzgot.co.za. Hence, lookups from anywhere on the Internet having the domain cranzgot.co.za ultimately are the responsibility of ns2.cranzgot.co.za, and originate (albeit through a long series of caches) from the host ns2.cranzgot.co.za.

Domain names in the .com, .net, and .org domains can now be registered with many different competing registrars. Go to http://www.internic.net for detailed information. Domain Name: CNN.COM Registrar: NETWORK SOLUTIONS, INC. Whois Server: whois.networksolutions.com Referral URL: www.networksolutions.com Name Server: NS-01A.ANS.NET Name Server: NS-01B.ANS.NET Name Server: NS-02A.ANS.NET Name Server: NS-02B.ANS.NET Updated Date: 22-sep-1999

>>> Last update of whois database: Thu, 20 Jan 00 01:39:07 EST <<< The Registry database contains ONLY .COM, .NET, .ORG, .EDU domains and Registrars.

27.6 The host, ping, and whois Command

The command host looks up a host name or an IP address, by doing a name server query. Try for an example of a host with lots of IP address. Keep typing host over and over. Notice that the order of the hosts keeps changing randomly. This reordering distributes load among the many cnn.com servers. Now, pick one of the IP addresses and type
host www.cnn.com

(Internic happens to have this database of .com, .net, .org, and .edu domains.)

27.7

The nslookup Command

nslookup is a program to interactively query a name server. If you run

nslookup

This command will return the host name cnn.com. Note that the host command is not available on all U NIX systems. The ping command has nothing directly to do with DNS but is a quick way of getting an IP address and at the same time checking whether a host is responding. It is often used as the acid test for network and DNS connectivity. See Section 25.10.1. 281

host <ip-address>

you will get a > prompt at which you can type commands. If you type in a host name, nslookup will return its IP address(s), and visa versa. Also, typing any time will return a complete list of commands. By default, nslookup uses the rst name server listed in /etc/resolv.conf for all its queries. However, the command
server <nameserver> help

282

27. DNS and Name Resolution

27.7. The nslookup Command

27.8. The dig Command

27. DNS and Name Resolution

will force nslookup to connect to a name server of your choice.

set type=MX cnn.com

27.7.1

NS, MX, PTR, A and CNAME records

The word record is a piece of DNS information. Now enter the command:

This tells nslookup to return the second type of information that a DNS can deliver: the authoritative name server for a domain or the NS record of the domain. You can enter any domain here. For instance, if you enter nslookup returns
Non-authoritative answer: cnn.com nameserver = NS-02B.ANS.NET cnn.com nameserver = NS-02A.ANS.NET cnn.com nameserver = NS-01B.ANS.NET cnn.com nameserver = NS-01A.ANS.NET Authoritative answers can be found NS-02B.ANS.NET internet address = NS-02A.ANS.NET internet address = NS-01B.ANS.NET internet address = NS-01A.ANS.NET internet address = from: 207.24.245.178 207.24.245.179 199.221.47.8 199.221.47.7 set type=NS cnn.com

set type=NS

to get the so-called MX record for that domain. The MX record is the server responsible for handling mail destined to that domain. MX records also have a priority (usually 10 or 20). This tells any mail server to try the 20 one should the 10 one fail, and so on. There are usually only one or two MX records. Mail is actually the only Internet service handled by DNS. (For instance, there is no such thing as a NEWSX record for news, or a WX record for web pages, whatever kind of information we may like such records to hold.) Also try

So-called PTR records are reverse lookups, or PoinTeRs to host names. So-called A records are forward lookups (the default type of lookup when you rst invoke nslookup and the type of lookup the rst half of this chapter was most concerned with), or Address lookups. So-called CNAME records are lookups of Canonical NAMEs. DNS allows you to alias a computer to many different names, even though each has one real name (called the canonical name). CNAME lookups returns the machine name proper.

set type=PTR <ip-address> set type=A <hostname> set type=CNAME <hostname>

This output tells us that four name servers are authoritative for the domain cnn.com (one plus three backups). It also tells us that it did not get this answer from an authoritative source, but through a cached source. It also tells us what name servers are authoritative for this very information. Now, switch to a name server that is authoritative for cnn.com:

27.8

The dig Command

dig stands for domain information groper. It sends single requests to a DNS server for testing or scripting purposes (it is similar to nslookup, but non-interactive). It is usually used like,

and run the same query:

cnn.com

server NS-02B.ANS.NET

The new result is somewhat more emphatic, but no different. Try There are only a few other kinds of records that you can get from a name server.

where <server> is the machine running the DNS daemon to query, <domain> is the domain of interest and <query-type> is one of A, ANY, MX, NS, SOA, HINFO, or AXFRof these, you can read about the non-obvious ones in dig(1). dig can also be used to test an Internet connection. See Section 20.7.4. Useful is the AXFR record. For instance

dig @<server> <domain> <query-type>

lists the entire domain of one of our local ISPs. 284

dig @dns.dial-up.net icon.co.za AXFR

283

28.2. Conguration Example

28. Network File System, NFS

rpc.mountd (also sometimes called mountd) This handles the initial incoming request from a client to mount a le system and check that the request is allowable. rpc.nfsd (also sometimes called nfsd) This is the corethe le-server program itself.

Chapter 28

rpc.lockd (also sometimes called lockd) This handles shared locks between different machines on the same le over the network. The acronym RPC stands for Remote Procedure Call. RPC was developed along with NFS by Sun Microsystems. It is an efcient way for a program to call a function on another machine and can be used by any service that needs to have efcient distributed processing. These days, its not really used for much except NFS, having been superseded by technologies like CORBA. The Object-Oriented version of RPC You can however, still write distributed applications with L INUXs RPC implementation.

Network File System, NFS

This chapter covers NFS, the le-sharing capabilities of U NIX, and describes how to set up directories shareable to other U NIX machines. As soon as one thinks of high-speed Ethernet, the logical possibility of sharing a le system across a network comes to mind. MS-DOS, OS/2, Apple Macintosh, and Windows have their own le-sharing schemes (IPX, SMB etc.), and NFS is the U NIX equivalent. Consider your hard drive with its 10,000 or so les. Ethernet is fast enough that you should be able to entirely use the hard drive of another machine, transferring needed data as network packets as required; or you should be able to make a directory tree visible to several computers. Doing this efciently is a complex task. NFS is a standard, a protocol, and (on L INUX ) a software suite that accomplishes this task in an efcient manner. It is really easy to congure as well. Unlike some other sharing protocols, NFS merely shares les and does not facilitate printing or messaging.

28.2

Conguration Example

Sharing a directory with a remote machine requires that forward and reverse DNS lookups be working for the server machine as well as all client machines. DNS is covered in Chapter 27 and Chapter 40. If you are just testing NFS and you are sharing directories to your local machine (which we do now), you may nd NFS to still work without a proper DNS setup. You should at least have proper entries in your /etc/hosts le for your local machine (see page 278). The rst step is deciding on the directory you would like to share. A useful trick is to share your CD-ROM to your whole LAN. This is perfectly safe considering that CDs are read-only. Create an /etc/exports le with the following in it:
/mnt/cdrom 192.168.1.0/24(ro) localhost(ro)

28.1 Software
Depending on your distribution, the following programs may be located in any of the bin or sbin directories. These are all daemon processes. To get NFS working, they should be started in the order given here. portmap (also sometimes called rpc.portmap) This maps service names to ports. Client and server processes may request a TCP port number based on a service name, and portmap handles these requests. It is basically a network version of your /etc/services le. 285

You can immediately see that the format of the /etc/exports le is simply a line for each shareable directory. Next to each directory name goes a list of hosts that are allowed to connect. In this case, those allowed access are all IP addresses having the upper 24 bits matching 192.168.1, as well as the localhost. Next, mount your CD-ROM as usual with

mkdir -p /mnt/cdrom mount -t iso9660 -o ro /dev/cdrom /mnt/cdrom

Now start each of the NFS processes in sequence: 286

28. Network File System, NFS

28.2. Conguration Example

28.3. Access Permissions

28. Network File System, NFS

portmap rpc.mountd rpc.nfsd rpc.lockd

Whenever you make changes to your /etc/exports le you should also follow by running which causes a rereading of the /etc/exports le. Entering the exportfs command with no options should then show which lists directories and hosts allowed to access them.
/mnt/cdrom /mnt/cdrom 192.168.1.0/24 localhost.localdomain exportfs -r

is failing because of a forward or reverse DNS lookup, so double-check beforehand that these are workingmount will not usually be more eloquent than the classic NFS error message: mount: <xyz> failed, reason given by server: Permission denied. A faulty DNS is also indicated by whole-minute pauses in operation. Most distributions will not require you to manually start and stop the daemon processes above. Like most services, RedHats NFS implementation can be invoked simply with:
/etc/init.d/nfs start /etc/init.d/nfslock start

(or /etc/rc.d/init.d/). On Debian , similarly,

/etc/init.d/nfs-common start /etc/init.d/nfs-kernel-server start

It is useful to test mounts from your local machine before testing from a remote machine. Here we perform the NFS mounting operation proper: You can see that the mount command sees the remote machines directory as a device of sorts, although the type is nfs instead of ext2, vfat, or iso9660. The remote host name is followed by a colon followed by the directory on that remote machine relative to the root directory. This syntax is unlike that for other kinds of services that name all les relative to some top level directory (eg., FTP and web servers). The acid test now is to run ls on the /mnt/nfs directory to verify that its contents are indeed the same as /mnt/cdrom. Supposing our server is called cdromserver, we can run the same command on all client machines:
mkdir /mnt/nfs mount -t nfs cdromserver:/mnt/cdrom /mnt/nfs mkdir /mnt/nfs mount -t nfs localhost:/mnt/cdrom /mnt/nfs

28.3

Access Permissions

Above, we used 192.168.1.0/24(ro) to specify that we want to give read-only access to a range of IP addresses. You can actually put host names with wildcards also; for example:
/mnt/cdrom *.mynet.mydomain.co.za(ro)

Then also allow read-write access with, say:

*.mynet.mydomain.co.za(rw)

/home

If anything went wrong, you might like to search your process list for all processes with an rpc, mount, nfs, or portmap in them. Completely stopping NFS means clearing all of these processes (if you really want to start from scratch). It is useful to also keep running in a separate console to watch for any error (or success) messages (actually true of any conguration you are doing). Note that it is not always obvious that NFS 287
tail -f /var/log/messages tail -f /var/log/syslog

One further option, no root squash, disables NFSs special treatment of rootowned les. This option is useful if you are nding certain les strangely inaccessible. no root squash is really only for systems (like diskless workstations) that need full root access to a le system. An example is:
/ *.very.trusted.net(rw,no_root_squash)

The man page for /etc/exports, exports(5), contains an exhaustive list of options. 288

28. Network File System, NFS

28.4. Security

28.5. Kernel NFS

28. Network File System, NFS

28.4 Security
NFS requires that a number of services be running that have no use anywhere else. Many naive administrators create directory exports with impunity, thus exposing those machines to opportunistic hackers. An NFS server should be well hidden behind a rewall, and any Internet server exposed to the Internet should never run the portmap or RPC services. Preferably uninstall all of these services if you are not actually running an NFS server.

28.5 Kernel NFS

There are actually two versions of the NFS implementation for L INUX . Although this is a technical caveat, it is worth understanding that the NFS server was originally implemented by an ordinary daemon process before the L INUX kernel itself supported NFS. Debian supports both implementations in two packages, nfs-server and nfs-kernel-server, although the conguration should be identical. Depending on the versions of these implementations and the performance you require, one or the other may be better. You are advised to at least check the status of the kernel NFS implementation on the kernel web pages. Of course, NFS as a client must necessarily be supported by the kernel as a regular le system type in order to be able to mount anything.

289

290

29.2. Invoking Services with /etc/inetd.conf 29. Services Running Under inetd

TCP wrapper-moderated. However, some services will run using only one method. Here, we will give an example showing all three methods. You will need to have an ftp package installed for this example (either wuftpd on RedHat or ftpd on Debian ).

Chapter 29

29.2.1 Invoking a standalone service

Try the following (alternative commands in parentheses):
/usr/sbin/in.ftpd -D ( /usr/sbin/in.wuftpd -s )

Services Running Under inetd

There are some hundred odd services that a common L INUX distribution supports. For all of these to be running simultaneously would be a strain. Hence, a special daemon process watches for incoming TCP connections and then starts the relevant executable, saving that executable from having to run all the time. This is used only for sparsely used servicesthat is, not web, mail, or DNS. The daemon that performs this function is traditionally called inetd: the subject of this chapter. (Section 36.1 contains an example of writing your own network service in shell script to run under inetd.)

The -D option instructs the service to start in Daemon mode (or standalone mode). This represents the rst way of running an Internet service.

29.2.2 Invoking an inetd service

With this method we can let inetd run the service for us. Edit your /etc/inetd.conf le and add or edit the line (alternatives in parentheses): Then, restart the inetd service with
ftp ( ftp stream tcp nowait root /usr/sbin/in.ftpd in.ftpd stream tcp nowait root /usr/sbin/in.wuftpd in.wuftpd )

29.1 The inetd Package

Which package contains inetd depends on the taste of your distribution. Indeed, under RedHat, version 7.0 switched to xinetd, a move that departs radically from the traditional U NIX inetd. xinetd is discussed below. The important inetd les are the conguration le /etc/inetd.conf, the executable /usr/sbin/inetd, the inetd and inetd.conf man pages, and the startup script /etc/init.d/inet (or /etc/rc.d/init.d/inetd or /etc/init.d/inetd). Another important le is /etc/services, discussed in Section 26.4.

and test with

/etc/init.d/inet restart ( killall -1 inetd ) ( /etc/rc.d/init.d/inet restart )

ps awx | grep ftp ftp localhost

The elds in the /etc/inetd.conf le have the following meanings: ftp The name of the service. Looking in the /etc/services le, we can see that this is TCP port 21. stream tcp Socket type and protocol. In this case, a TCP stream socket, and hardly ever anything else. nowait Do not wait for the process to exit before listening for a further incoming connection. Compare to wait and respawn in Chapter 32. 292

29.2 Invoking Services with /etc/inetd.conf

Most services can be started in one of three ways: rst as a standalone (resource hungry, as discussed) daemon; second, under inetd; or third as an inetd service which is 291

29. Services Running Under inetd 29.2. Invoking Services with /etc/inetd.conf

29.3. Various Service Explanations

29. Services Running Under inetd

root The initial user ID under which the service must run. /usr/sbin/in.ftpd ( /usr/sbin/in.wuftpd ) The actual executable. in.ftpd The command-line. In this case, just the program name and no options.

29.2.4 Distribution conventions

Note that the above methods cannot be used simultaneously. If a service is already running one way, trying to start it another way will fail, possibly with a port in use error message. Your distribution would have already decided whether to make the service an inetd entry or a standalone daemon. In the former case, a line in /etc/inetd.conf will be present; in the latter case, a script /etc/init.d/<service> (or /etc/rc.d/init.d/<service>) will be present to start or stop the daemon. Typically, there will be no /etc/init.d/ftpd script, but there will be /etc/init.d/httpd and /etc/init.d/named scripts. Note that there will always be a /etc/init.d/inet script.

29.2.3

Invoking an inetd TCP wrapper service

With this last method we let inetd run the service for us under the tcpd wrapper command. This is almost the same as before, but with a slight change in the /etc/inetd.conf entry:
ftp ( ftp stream tcp nowait root /usr/sbin/tcpd /usr/sbin/in.ftpd stream tcp nowait root /usr/sbin/tcpd /usr/sbin/in.wuftpd )

29.3

Various Service Explanations

All these services are potential security holes. Dont take chances: disable them all by commenting out all lines in /etc/inetd.conf.

Then, restart the inetd service as before. These alternative lines allow tcpd to invoke in.ftpd (or in.wuftpd) on inetds behalf. The tcpd command does various tests on the incoming connection to decide whether it should be trusted. tcpd checks what host the connection originates from and compares that host against entries in the le /etc/hosts.allow and /etc/hosts.deny. It can refuse connections from selected hosts, thus giving you ner access control to services. Consider the preceding /etc/inetd.conf entry against the following line in your /etc/hosts.allow le:
in.ftpd: LOCAL, .my.domain ( in.wuftpd: LOCAL, .my.domain )
5

A typical /etc/inetd.conf le (without the comment lines) looks something like:

ftp telnet shell login talk ntalk pop-3 imap uucp tftp bootps finger auth stream stream stream stream dgram dgram stream stream stream dgram dgram stream stream tcp tcp tcp tcp udp udp tcp tcp tcp udp udp tcp tcp nowait nowait nowait nowait wait wait nowait nowait nowait wait wait nowait wait root /usr/sbin/tcpd root /usr/sbin/tcpd root /usr/sbin/tcpd root /usr/sbin/tcpd nobody.tty /usr/sbin/tcpd nobody.tty /usr/sbin/tcpd root /usr/sbin/tcpd root /usr/sbin/tcpd uucp /usr/sbin/tcpd root /usr/sbin/tcpd root /usr/sbin/tcpd nobody /usr/sbin/tcpd root /usr/sbin/in.identd in.ftpd -l -a in.telnetd in.rshd in.rlogind in.talkd in.ntalkd ipop3d imapd /usr/sbin/uucico -l in.tftpd bootpd in.fingerd in.identd -e -o

as well as the following line in the le /etc/hosts.deny:

in.ftpd: ALL ( in.wuftpd: ALL )

This example will deny connections from all machines with host names not ending in .my.domain but allow connections from the local The same machine on which inetd is running machine. It is useful at this point to try make an ftp connection from different machines to test access control. A complete explanation of the /etc/hosts.allow and /etc/hosts.deny le format can be obtained from hosts access(5). Another example is (/etc/hosts.deny):

The above services have the following purposes (port numbers in parentheses): ftp (21) File Transfer Protocol, as shown above. telnet (23) Telnet login access. shell (514) rsh Remote shell script execution service. login (513) rlogin Remote Login login service. 294

ALL: .snake.oil.com, 146.168.160.0/255.255.240.0

which would deny access for ALL services to all machines inside the 146.168.160.0 (rst 20 bits) network, as well as all machines under the snake.oil.com domain. 293

29. Services Running Under inetd

29.4. The xinetd Alternative

29.5. Conguration Files

29. Services Running Under inetd

talk (517), ntalk User communication gimmick. pop-3 (110) Post Ofce Protocol mail retrieval servicehow most people get their mail through their ISP. imap (143) Internet Mail Access Protocola more sophisticated and dangerously insecure version of POP. uucp (540) Unix-to-Unix copy operating over TCP. tftp (69) Trivial FTP service used, for example, by diskless workstations to retrieve a kernel image. bootpd (67) BOOTP IP conguration service for LANs that require automatic IP assignment. finger (79) User lookup service. auth (113) A service that determines the owner of a particular TCP connection. If you run a machine with lots of users, administrators of other machines can see which users are connecting to them from your machine. For tracking purposes, some IRC and FTP servers require that a connecting client run this service. Disable this service if your box does not support shell logins for many users.
5

includedir /etc/xinetd.d

The le dictates, respectively, that xinetd does the following: limits the number of simultaneous connections of each service to 60; logs to the syslog facility, using syslogs authpriv channel; logs the HOST and Process ID for each successful connection; and logs the HOST (and also RECORD information about the connection attempt) for each failed connection. In other words, /etc/xinetd.conf really says nothing interesting at all. The last line says to look in /etc/xinetd.d/ for more (service-specic) les. Our FTP service would have the le /etc/xinetd.d/wu-ftpd containing:
service ftp { socket_type server server_args wait user log_on_success log_on_failure nice }

= stream = /usr/sbin/in.ftpd = -l -a = no = root += DURATION USERID += USERID = 10

29.4 The xinetd Alternative

Instead of the usual inetd + tcpd combination, RedHat switched to the xinetd package as of version 7.0. The xinetd package combines the features of tcpd and inetd into one neat package. The xinetd package consists of a top-level cong le, /etc/xinetd.conf; an executable /usr/sbin/xinetd; and then a cong le for each service under the directory /etc/xinetd.d/. This arrangement allows a package like ftpd control over its own conguration through its own separate le.

This le is similar to our /etc/inetd.conf line above, albeit more verbose. Respectively, this le dictates these actions: listen with a stream TCP socket; run the executable /usr/sbin/in.ftpd on a successful incoming connection; pass the arguments -l -a on the command-line to in.ftpd (see ftpd(8)); never wait for in.ftpd to exit before accepting the next incoming connection; run in.ftpd as user root; additionally log the DURATION and USERID of successful connections; additionally log the USERID of failed connections; and be nice to the CPU by running in.ftpd at a priority of 10.

29.5.1 Limiting access

29.5 Conguration Files

The default top-level cong le, /etc/xinetd.conf, looks simply like this:
defaults { instances log_type log_on_success log_on_failure }

The security options of xinetd allow much exibility. Most important is the only from option to limit the remote hosts allowed to use a service. The most extreme use is to add only from 127.0.0.1 to the top-level cong le:
defaults { only_from . . .

= = = =

60 SYSLOG authpriv HOST PID HOST RECORD

= 127.0.0.1 mymachine.local.domain

295

296

29. Services Running Under inetd

29.6. Security

29. Services Running Under inetd

which allows no remote machines to use any xinetd service at all. Alternatively, you can add an only from line to any of the les in /etc/xinetd.d/ to restrict access on a per-service basis. only from can also take IP address ranges of the form nnn.nnn.nnn.nnn/bits, as well as domain names. For example,
only_from = 127.0.0.1 192.168.128.0/17 .somewhere.friendly.com

which in the last case allows access from all machines with host names ending in .somewhere.friendly.com. Finally there is the no access option that works identically to only from, dictating hosts and IP ranges from which connections are not allowed:
no_access = .snake.oil.net

29.6 Security
It may be thought that using /etc/hosts.deny ( or only from = ) to deny access to all remote machines should be enough to secure a system. This is not true: even a local user being able to access a local service is a potential security hole, since the service usually has higher privileges than the user. It is best to remove all services that are not absolutely necessary. For Internet machines, do not hesitate to hash out every last service or even uninstall inetd ( or xinetd ) entirely. See also Chapter 44.

297

298

30.1. Introduction

30. exim and sendmail

message (sent by a typical interactive mail client) ends up on a distant users personal workstation is as follows: 1. A user congures his mail client (Outlook Express, Netscape, etc.) to use a particular SMTP host (for outgoing mail, also called the SMTP gateway) and POP host (or IMAP host) for incoming mail.

Chapter 30

2. The user composes a message to, say, rrabbit@toonland.net and then clicks on Send. 3. The mail client initiates an outgoing TCP connection to port 25 of the SMTP host. An MTA running on the SMTP host and listening on port 25 services the request. The mail client uses the SMTP protocol exactly as in Section 10.2. It lls in rrabbit@toonland.net as the recipient address and transfers a properly composed header (hopefully) and message body to the MTA. The mail client then terminates the connection and reports any errors. 4. The MTA queues the message as a spool le, periodically considering whether to process the message further according to a retry schedule. 5. Should the retry schedule permit, the MTA considers the recipient address rrabbit@toonland.net. It strips out the domain part of the email address that is, everything after the @. It then performs a DNS MX query (or MX lookupindexiiMX recordDNS) for the domain toonland.net. DNS resolution for toonland.net follows the procedure listed in Section 27.2.2. In short, this means (approximately) that it looks for the name server that is authoritative for the domain toonland.net. It queries that name server for the MX record of the domain toonland.net. The name server returns a host name, say, mail.toonland.net with corresponding IP address, say, 197.21.135.82.
how to set up your name server to return such an MX record.

exim and sendmail

This chapter effectively explains how to get L INUX up and running as a mail server. I have also included discussion on the process of mail delivery right through to retrieval of mail using POP and IMAP.

30.1 Introduction
exim and sendmail are MTAs (mail transfer agents). An MTA is just a daemon process that listens on port 25 for incoming mail connections, spools See page 197 about spooling in general. that mail in a queue (for exim, the /var/spool/exim/input/ directory, for sendmail, the /var/spool/mqueue/ directory), then resends that mail to some other MTA or delivers it locally to some users mailbox. In other words, the MTA is the very package that handles all mail spooling, routing, and delivery. We saw in Section 10.2 how to manually connect to an MTA with telnet. In that example, sendmail version 8.9.3 was the MTA running on machine mail.cranzgot.co.za.

Section 27.7.1 shows you how you can manually lookup the MX record. Chapter 40 shows you

sendmail is the original and popular U NIX MTA. It is probably necessary to learn how to congure it because so many organizations standardize on it. However, because exim is so easy to congure, it is worthwhile replacing sendmail wherever you see itthere are at least three MTAs that are preferable to sendmail. I explain the minimum of what you need to know about sendmail later on and explain exim in detail.

6. The MTA makes an SMTP connection to port 25 of 197.21.135.82. Another MTA running on mail.toonland.net services the request. A recipient address, message header, and message body are transferred using the SMTP protocol. The MTA then terminates the connection. 7. The MTA running on mail.toonland.net considers the recipient address rrabbit@toonland.net. It recognizes toonland.net as a domain for which it hosts mail (that is, a local domain). It recognizes rrabbit as a user name within its own /etc/passwd le. 8. The MTA running on mail.toonland.net appends the message to the users personal mailbox le, say, /var/spool/mail/rrabbit or /home/rrabbit/Maildir/. The delivery is now complete. How the email gets from the mailbox on mail.toonland.net to Mr Rabbits personal workstation is not the responsibility of the MTA and does not happen through SMTP. 300

30.1.1

How mail works

Before we get into MTA conguration, a background in mail delivery and indexiiMX recordDNSMX record handling is necessary. The sequence of events whereby a mail 299

30. exim and sendmail

30.2. exim Package Contents

30.3. exim Conguration File

30. exim and sendmail

9. Mr Rabbit would have congured his mail client (running on his personal workstation) to use a POP/IMAP host mail.toonland.net for incoming mail. mail.toonland.net runs a POP or IMAP service on port 110 or 143, respectively. 10. Mr Rabbits mail client makes a TCP connection to port 110 (or 143) and communicates using the POP or IMAP protocol. The POP or IMAP service is responsible for feeding the message to the mail client and deleting it from the mailbox le. 11. Mr Rabbits mail client stores the message on his workstation using its own methods and displays the message as a new message.
5

/etc/aliases /usr/bin/mailq /usr/bin/newaliases /usr/bin/rmail /usr/lib/sendmail /usr/sbin/sendmail

Finally, there is the exim binary itself, /usr/sbin/exim, and conguration le /etc/exim/config, /etc/exim.conf, or /etc/exim/exim.conf, dependdistribution. Then there are the usual start/stop scripts, ing on your L INUX /etc/init.d/exim. or /etc/rc.d/init.d/exim

30.1.2

Conguring a POP/IMAP server

30.3

exim Conguration File

POP and IMAP are invoked by inetd or xinetdsee Chapter 29. Except for limiting the range of clients that are allowed to connect (for security reasons), no conguration is required. Client connections authenticate themselves using the normal U NIX login name and password. There are specialized POP and IMAP packages for supporting different mailbox types (like Maildir).

As a preliminary example, here we create a simple spooling mail server for a personal workstation, cericon.cranzgot.co.za. Client applications (especially non-U NIX ones) are usually congured to connect to an MTA running on a remote machine, however, using a remote SMTP host can be irritating if the host or network go down. Running exim on the local workstation enables all applications to use localhost as their SMTP gateway: that is, exim takes care of queuing and periodic retries. Here is the conguration. The difference between this and a full-blown mail server is actually very slight.
#################### MAIN CONFIGURATION SETTINGS ##################### log_subject errors_address = postmaster freeze_tell_mailmaster = yes queue_list_requires_admin = false prod_requires_admin = false trusted_users = psheer local_domains = localhost : ${primary_hostname} never_users = root # relay_domains = my.equivalent.domains : more.equivalent.domains host_accept_relay = localhost : *.cranzgot.co.za : 192.168.0.0/16 exim_user = mail exim_group = mail end ###################### TRANSPORTS CONFIGURATION ###################### remote_smtp: driver = smtp hosts = 192.168.2.1 hosts_override local_delivery: driver = appendfile

30.1.3

Why exim?

The exim home page http://www.exim.org/ gives you a full rundown. Here I will just say that exim is the simplest MTA to congure. Moreover, its conguration le works the same way you imagine mail to work. Its really easy to customize the exim conguration to do some really weird things. The whole package ts together cleanly, logically, and intuitively. This is in contrast to sendmails sendmail.cf le, which is widely considered to be extremely cryptic and impractical. exim also seems to have been written with proper security considerations, although many people argue that postfix and qmail are the last word in secure mail.

30.2

exim Package Contents

You can get exim as a .rpm or .deb le. After installation, the le /usr/share/doc/exim-?.??/doc/spec.txt or /usr/doc/ contains the complete exim documentation; there is also an HTML version on the exim web page, whereas the man page contains only command-line information. exim is a drop-in replacement for sendmail, meaning that for every critical sendmail command, there is an exim command of the same name that takes the same options, so that needy scripts wont know the difference. These are:

301

302

30. exim and sendmail

30.3. exim Conguration File

30. exim and sendmail

file = /var/spool/mail/${local_part} delivery_date_add envelope_to_add return_path_add group = mail mode_fail_narrower = mode = 0660 end ###################### DIRECTORS CONFIGURATION ####################### localuser: driver = localuser transport = local_delivery end ###################### ROUTERS CONFIGURATION ######################### lookuphost: driver = lookuphost transport = remote_smtp literal: driver = ipliteral transport = remote_smtp end ###################### RETRY CONFIGURATION ########################### * * F,2h,15m; G,16h,1h,1.5; F,4d,8h end

frozen messages sometimes mean that something is wrong with your system or mail conguration. local domains Each mail message received is processed in one of two ways: by either a local or remote delivery. A local delivery is one to a user on the local machine, and a remote delivery is one to somewhere else on the Internet. local domains distinguishes between these two. For example, according to the cong line above, a message destined to psheer@localhost or psheer@cericon.cranzgot.co.za is local, whereas a message to psheer@elsewhere.co.za is remote. Note that the list is colon delimited. never users Never become this user. Just for security. exim user Species the user that exim should run as. exim group Species the group that exim should run as. It is important to understand the host accept relay and relay domains options for security. host accept relay This option species machines that are allowed to use cericon.cranzgot.co.za as a relay. A relay is a host that sends mail on another machines behalf: that is, we are acting as a relay when we process a mail message that neither originated from our machine nor is destined for a user on our machine. We never want to relay from an untrusted host. Why? Because it may, for example, allow someone to send 100,000 messages to 100,000 different addresses, each with us in the message header. host accept relay species a list of trusted hosts that are allowed to send such arbitrary messages through us. Note again that the list is colon delimited. In this example, we dont even need to put in addresses of other machines on our LAN, except if we are feeling friendly. relay domains relay domains gives an additional condition for which an arbitrary host is allowed to use us as a relay. Consider that we are a backup mail server for a particular domain; mail to the domain does not originate from us nor is destined for us yet must be allowed only if the destination address matches the domains for which we are a backup. We put such domains under relay domains.

###################### REWRITE CONFIGURATION ######################### *@cericon.cranzgot.co.za psheer@icon.co.za

30.3.1

Global settings

The exim cong le is divided into six logical sections separated by the end keyword. The top or MAIN section contains global settings. The global settings have the following meanings: log subject Tells exim to log the subject in the mail log le. For example, T="I LOVE YOU" will be added to the log le. errors address The mail address where errors are to be sent. It doesnt matter what you put here, because all mail will get rewritten to psheer@icon.co.za, as we see later. freeze tell mailmaster Tells errors address about frozen messages. frozen messages are messages that could not be delivered for some reason (like a permissions problem, or a failed message whose return address is invalid) and are agged to sit idly in the mail queue, and not be processed any further. Note that 303

30.3.2 Transports
The transport section comes immediately after the main conguration options. It denes various methods of delivering mail. We are going to refer to these methods later in 304

30. exim and sendmail

30.3. exim Conguration File

30.4. Full-blown Mail server

30. exim and sendmail

the conguration le. Our manual telneting to port 25 was transporting a mail message by SMTP. Appending a mail message to the end of a mail folder is also a transport method. These are represented by the remote smtp: and local delivery: labels, respectively. remote smtp: This transport has the following suboptions: driver The actual method of delivery. driver = always species the kind of transport, director, or router. hosts override and hosts Using these two options together overrides any list of hosts that may have been looked up by DNS MX queries. By list of hosts we mean machines established from the recipients email address to which we might like to make an SMTP delivery, but which we are not going to use. Instead we send all mail to 192.168.2.1, which is this companys internal mail server. local delivery: This transport has the following suboptions: driver The actual method of delivery. driver = always species the kind of transport, director, or router. file The le to append the mail message to. ${local part} is replaced with everything before the @ character of the recipients address. delivery date add, envelope to add, and return path add Various things to add to the header. group, mode fail narrower and mode Various permission settings.
5

30.3.4 Routers
If a message arrives and it is not listed in local domains, exim attempts a remote delivery. Similarly, this means exim works through the list of routers until it nds one that does not fail. Two routers are listed here. The rst is for common email addresses. It uses the lookuphost driver, which does a DNS MX query on the domain part of the email address (i.e., everything after the @). The MX records found are then passed to the remote smtp transport (and in our case, then ignored). The lookuphost driver will fail if the domain part of the email address is a bracketed literal IP address. The second router uses the ipliteral driver. It sends mail directly to an IP address in the case of bracketed, literal email addresses. For example, root@[111.1.1.1]. A router routes mail to another host.

30.4

Full-blown Mail server

An actual mail server cong le contains very little extra. This one is the example cong le that comes by default with exim-3.16:
#################### MAIN CONFIGURATION SETTINGS ##################### # primary_hostname = # qualify_domain = # qualify_recipient = # local_domains = never_users = root # host_accept_relay = localhost # host_accept_relay = my.friends.host : 131.111.0.0/16 # relay_domains = my.equivalent.domains : more.equivalent.domains host_lookup = 0.0.0.0/0 # receiver_unqualified_hosts = # sender_unqualified_hosts = rbl_domains = rbl.maps.vix.com no_rbl_reject_recipients sender_reject = "*@*.sex*.net:*@sex*.net" host_reject = "open-relay.spamming-site.com" rbl_warn_header # rbl_domains = rbl.maps.vix.com:dul.maps.vix.com:relays.orbs.org # percent_hack_domains = * end ###################### TRANSPORTS CONFIGURATION ###################### remote_smtp: driver = smtp # procmail transport goes here <--local_delivery: driver = appendfile

(It should be obvious at this stage what these two transports are going to be used for. As far as MTAs are concerned, the only two things that ever happen to an email message are that it either (a) gets sent through SMTP to another host or (b) gets appended to a le.)

30.3.3

Directors

If a message arrives and it is listed in local domains, exim will attempt a local delivery. This means exim works through the list of directors until it nds one that does not fail. The only director listed here is the one labeled localuser: with local delivery as its transport. So quite simply, email messages having recipient addresses that are listed under local domains are appended to a users mailbox le not very complicated. A director directs mail to a mailbox. 305

306

30. exim and sendmail

30.4. Full-blown Mail server

30.5. Shell Commands for exim Administration

30. exim and sendmail

file = /var/spool/mail/${local_part} delivery_date_add envelope_to_add return_path_add group = mail mode = 0660 address_pipe: driver = pipe return_output address_file: driver = appendfile delivery_date_add envelope_to_add return_path_add address_reply: driver = autoreply end ###################### DIRECTORS CONFIGURATION ####################### # routers because of a "self=local" setting (not used in this configuration). system_aliases: driver = aliasfile file = /etc/aliases search_type = lsearch user = mail group = mail file_transport = address_file pipe_transport = address_pipe userforward: driver = forwardfile file = .forward no_verify no_expn check_ancestor # filter file_transport = address_file pipe_transport = address_pipe reply_transport = address_reply # procmail director goes here <--localuser: driver = localuser transport = local_delivery end ###################### ROUTERS CONFIGURATION ######################### # widen_domains = "sales.mycompany.com:mycompany.com" lookuphost: driver = lookuphost transport = remote_smtp # widen_domains = literal: driver = ipliteral transport = remote_smtp end ###################### RETRY CONFIGURATION ########################### * * F,2h,15m; G,16h,1h,1.5; F,4d,8h end

######################################################################

For procmail support (see procmail(1), procmailrc(6), and procmailex(5)), simply add
procmail: driver = pipe command = "/usr/bin/procmail -Y -d ${local_part}"

after your remote smtp transport, and then also,

procmail: driver = localuser transport = procmail require_files = /usr/bin/procmail

after your user forward director.

30.5

Shell Commands for exim Administration

As with other daemons, you can stop exim, start exim, and cause exim to reread its conguration le with:
/etc/init.d/exim stop /etc/init.d/exim start /etc/init.d/exim reload

You should always do a reload to cause cong le changes to take effect. The startup script actually just runs exim -bd -q30m, which tells exim to start as a standalone daemon, listening for connections on port 25, and then execute a runq (explained below) every 30 minutes. To cause exim and many other MTAs for that matter to loop through the queue of pending messages and consider each one for deliver, run
runq

To list mail that is queued for delivery, use

which is the same as exim -q.

mailq

which is the same as exim -bp. To forcibly attempt delivery on any mail in the queue, use 308

307

30. exim and sendmail

30.6. The Queue

30.7. /etc/aliases for Equivalent Addresses

30. exim and sendmail

exim -qf

[root@cericon]#

and then to forcibly retry even frozen messages in the queue, use
exim -qff

work even better.

30.7

/etc/aliases for Equivalent Addresses

To delete a message from the queue, use

exim -Mrm <message-id>

Often, we would like certain local addresses to actually deliver to other addresses. For instance, we would like all mail destined to user MAILER-DAEMON to actually go to user postmaster; or perhaps some user has two accounts but would like to read mail from only one of them. The /etc/aliases le performs this mapping. This le has become somewhat of an institution; however you can see that in the case of exim, aliasing is completely arbitrary: you can specify a lookup on any le under the system aliases: director provided that le is colon delimited. A default /etc/aliases le could contain as much as the following; you should check that the postmaster account does exist on your system, and test whether you can read, send, and receive mail as user postmaster.
# This is a combination of what I found in the Debian # and RedHat distributions. MAILER-DAEMON: abuse: anonymous: backup: backup-reports: bin: daemon: decode: dns: dns-admin: dumper: fetchmail-daemon: games: gnats: ingres: info: irc: list: listmaster: lp: mail: mailer-daemon: majordom: man: manager: msql: news: postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster

The man page exim(8) contains exhaustive treatment of command-line options. Those above are most of what you will use, however.

30.6 The Queue

It is often useful to check the queue directory /var/spool/exim/input/ for mail messages, just to get an inside look at whats going on. The simple session
[root@cericon]# mailq 0m 320 14Epss-0008DY-00 <psheer@icon.co.za> freddy@elmstreet.org
5 5

304 14Ept8-0008Dg-00 <psheer@icon.co.za> igor@ghostbusters.com ls -l /var/spool/exim/input/ root root root root root root root root 25 550 25 530 Jan Jan Jan Jan 6 6 6 6 11:43 11:43 11:43 11:43 14Epss-0008DY-00-D 14Epss-0008DY-00-H 14Ept8-0008Dg-00-D 14Ept8-0008Dg-00-H
10

[root@cericon]# total 16 -rw------1 -rw------1 -rw------1 -rw------1

clearly shows that two messages are queued for delivery. The les ending in -H are envelope headers, and those ending in -D are message bodies. The spec.txt document will show you how to interpret the contents of the header les. Dont be afraid to manually rm les from this directory, but always delete them in pairs (i.e., remove the both the header and the body le), and make sure exim is not running at the time. In the above example, the commands,
[root@cericon]# exim -Mrm 14Epss-0008DY-00 14Ept8-0008Dg-00 Message 14Epss-0008DY-00 has been removed Message 14Ept8-0008Dg-00 has been removed [root@cericon]# mailq

309

310

30. exim and sendmail

30.8. Real-Time Blocking List Combating Spam

30. exim and sendmail

nobody: operator: postgres: proxy: root: sync: support: sys: system: toor: uucp: warnings: web-master: www-data:

postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster postmaster

To send spam is easy. Work your way around the Internet till you nd a mail server that allows relaying. Then send it 10,000 email addresses and a message about where to get pictures of naked underage girls. Now you are a genuine worthy-ofbeing-arrested spammer. Unfortunately for the unsuspecting administrator of that machine and provided you have even a little clue what youre doing, he will probably never be able to track you down. Several other tricks are employed to get the most out of your $100-for-1,000,000-genuine-email-addresses. Note that spam is not merely email you are not interested in. People often confuse mail with other types of communication. . . like telephone calls: if you get a telephone call, you have to pick up the phone then and therethe call is an an invasion of your privacy. The beauty of email is that you never need to have your privacy invaded. You can simply delete the mail. If you never want to get email from a particular person again, you can simply add a lter that blocks mail from that persons address (see procmailex(5)). If you are irritated by the presumption of the sender, then thats your problem.
Replying to that person with Please dont email me... not only shows that you are insecure, but also that you are clueless, dont get much mail, and are therefore also unpopular.

# some users who want their mail redirected arny: mail@swartzneger.co.us larry: lking@cnn.com

You can remove a lot of these aliases, since they assume services to be running that might not be installedgames, ingres, for example. Aliases can do two things: rstly, anticipate what mail people are likely to use if they need to contact the administrator; and secondly, catch any mail sent by system daemons: for example the, email address of the DNS administrator is dictated by the DNS cong les, as explained on page 445. Note that an alias in the /etc/aliases le does not have to have an account on the systemlarry and arny need not have entries in the /etc/passwd le.

The point at which email becomes intrusive is purely a question of volume, much like airwave advertisements. Because it comes from a different place each time, you cannot protect yourself against it with a simple mail lter. Typical spam mail will begin with a spammer subject like Create Wealth From Home Now!! and then the spammer will audaciously append the footer:
This is not a SPAM. You are receiving this because you are on a list of email addresses that I have bought. And you have opted to receive information about business opportunities. If you did not opt in to receive information on business opportunities then please accept our apology. To be REMOVED from this list simply reply with REMOVE as the subject. And you will NEVER receive another email from me.

30.8 Real-Time Blocking List Combating Spam

30.8.1 What is spam?

Need I say that you should be wary of replying with REMOVE, since it clearly tells the sender that your email is a valid address.

Spam refers to unsolicited Not looked for or requested; unsought bulk mail sent to users usually for promotional purposes. That is, mail is sent automatically to many people with whom the sender has no relationship, and where the recipient did nothing to prompt the mail: all on the chance that the recipient might be interested in the subject matter. Alternatively, spam can be thought of as any mail sent to email addresses, where those addresses were obtained without their owners consent. More practically, anyone who has had an email account for very long will have gotten messages like Subject: Fast way to earn big $$$!, which clutters my mailbox. The longer you have an email address, the more of these messages you will get, and the more irritated you will get. 311

30.8.2 Basic spam prevention

You can start by at least adding the following lines to your MAIN conguration section:
headers_check_syntax headers_sender_verify sender_verify receiver_verify

The option headers check syntax causes exim to check all headers of incoming mail messages for correct syntax, failing them otherwise. The next three options check 312

30. exim and sendmail

30.8. Real-Time Blocking List Combating Spam

30.9. Sendmail

30. exim and sendmail

that one of the Sender:, Reply-To: or From: headers, as well as both the addresses in the SMTP MAIL and RCPT commands, are genuine email addresses. The reasoning here is that spammers will often use malformed headers to trick the MTA into sending things it ordinarily wouldnt, I am not sure exactly how this applies in exims case, but these are for the good measure of rejecting email messages at the point where the SMTP exchange is being initiated.

Spam is evil. Spam is caused by poorly congured mail servers. It is the responsibility of the mail administrator to ensure that proper measures have been taken to prevent spam. Even as a user, you should follow up spam by checking where it came from and complaining to those administrators. Many mail administrators are not aware there is an issue. Remind them.

30.8.3

Real-time blocking list

To nd out a lot more about spamming, banning hosts, reporting spam and email usage in general, see MAPS (Mail Abuse Prevention System LLC) http://www.mail-abuse.org/, as well as Open Relay Behavior-modication System http://www.orbs.org/. If this site is not working, there is also http://www.orbl.org/ and http://www.ordb.org/. Real-time Blocking Lists or RBLs are a not-so-new idea that has been incorporated into exim as a feature. It works as follows. The spammer has to use a host that allows relays. The IP of that relay host will be clear to the MTA at the time of connection. The MTA can then check that against a database of publicly available banned IP addresses of relay hosts. For exim, this means the list under rbl domains. If the rbl domains friendly has this IP blacklisted, then exim denies it also. You can enable this capability with This example comes from exims

30.9

Sendmail

sendmails conguration le is /etc/sendmail.cf. This le format was inherited from the rst U NIX servers and references simpler les under the directory /etc/mail/. You can do most ordinary things by editing one or another le under /etc/mail/ without having to deal with the complexities of /etc/sendmail.cf. Like most stock MTAs shipped with L INUX distributions, the sendmail package will work by default as a mailer without any conguration. However, as always, you will have to add a list of relay hosts. This is done in the le /etc/mail/access for sendmail-8.10 and above. To relay from yourself and, say, the hosts on network 192.168.0.0/16, as well as, say, the domain hosts.trusted.com, you must have at least:
localhost.localdomain localhost 127.0.0.1 192.168 trusted.com RELAY RELAY RELAY RELAY RELAY

front web page.

# reject messages whose sending host is in MAPS/RBL # add warning to messages whose sending host is in ORBS rbl_domains = blackholes.mail-abuse.org/reject : \ dialups.mail-abuse.org/reject : \ relays.mail-abuse.org/reject : \ relays.orbs.org/warn # check all hosts other than those on internal network rbl_hosts = !192.168.0.0/16:0.0.0.0/0 # but allow mail to postmaster@my.dom.ain even from rejected host recipients_reject_except = postmaster@my.dom.ain # change some logging actions (collect more data) rbl_log_headers # log headers of accepted RBLed messages rbl_log_rcpt_count # log recipient info of accepted RBLed messages

which is exactly what the host accept relay option does in the case of exim. The domains for which you are acting as a backup mail server must be listed in the le /etc/mail/relay-domains, each on a single line. This is analogous to the relay domains option of exim. Then, of course, the domains for which sendmail is going to receive mail must also be specied. This is analogous to the local domains option of exim. These are listed in the le /etc/mail/local-host-names, each on a single line. The same /etc/aliases le is used by exim and sendmail. Having congured anything under /etc/mail/, you should now run make in this directory to rebuild lookup tables for these les. You also have to run the command 314

in your MAIN conguration section. Also remember to remove the line no rbl reject recipients; otherwise, exim will only log a warning message and not actually refuse email.

30.8.4

Mail administrator and user responsibilities

Mail administrator and email users are expected to be aware of the following: 313

30. exim and sendmail

30.9. Sendmail

30. exim and sendmail

newaliases whenever you modify the /etc/aliases le. In both cases, you must restart sendmail. sendmail has received a large number of security alerts in its time. It is imperative that you install the latest version. Note that older versions of sendmail have congurations that allowed relaying by defaultanother reason to upgrade. A useful resource to for nding out more tricks with sendmail is The Sendmail FAQ http://www.sendmail.org/faq/.

315

316

31.2. Theory

31. lilo, initrd, and Booting

31.2

Theory

31.2.1 Kernel boot sequence

A U NIX kernel, to be booted, must be loaded into memory from disk and be executed. The execution of the kernel causes it to uncompress itself and then run. The word boot
itself comes from the concept that a computer cannot begin executing without program code, and program code cannot get into memory without other program codelike trying to lift yourself up by your bootstraps, and hence the name. The rst thing the kernel does after it runs is initialize various hard-

Chapter 31

lilo, initrd, and Booting

lilo stands for linux loader. LILO: is the prompt you rst see after boot up, from which you can usually choose the OS you would like to boot and give certain boot options to the kernel. This chapter explains how to congure lilo and kernel boot options, and to get otherwise non-booting systems to boot. The lilo package itself contains the les
/sbin/lilo /usr/bin/keytab-lilo

ware devices. It then mounts the root le system on a specied partition. Once the root le system is mounted, the kernel executes /sbin/init to begin the U NIX operating system. This is how all U NIX systems begin life.

31.2.2 Master boot record

PCs begin life with a small program in the ROM BIOS that loads the very rst sector of the disk into memory, called the boot sector of the master boot record or MBR. This piece of code is up to 512 bytes long and is expected to start the operating system. In the case of L INUX , the boot sector loads the le /boot/map, which contains a list of the precise location of the disk sectors that the L INUX kernel image (usually the le /boot/vmlinuz) spans. It loads each of these sectors, thus reconstructing the kernel image in memory. Then it jumps to the kernel to execute it. You may ask how it is possible to load a le from a le system when the le system is not mounted. Further, the boot partition is a small and simple program and certainly does not support the many types of le systems and devices that the kernel image may reside in. Actually, lilo doesnt have to support a le system to access a le, as long as it has a list of the sectors that the le spans and is prepared to use the BIOS interrupts Nothing to do with interrupting or hardware interrupts, but refers to BIOS
functions that are available for use by the operating system. Hardware devices may insert custom BIOS functions to provided rudimentary support needed for themselves at startup. This support is distinct from that provided by the hardware device drivers of the booted kernel. to read those sectors. If the le

/boot/boot.b /boot/message /boot/chain.b /boot/os2_d.b /usr/share/doc/lilo-<version>

which is not that interesting, except to know that the technical and user documentation is there if hard-core details are needed.

31.1 Usage
When you rst start your L INUX system, the LILO: prompt, at which you can enter displays a list of things to type. The purboot options, is displayed. Pressing pose is to allow the booting of different L INUX installations on the same machine, or different operating systems stored in different partitions on the same disk. Later, you can actually view the le /proc/cmdline to see what boot options (including default boot options) were used. 317

is never modied, that sector list will never change; this is how the /boot/map and /boot/vmlinuz les are loaded.

31.2.3 Booting partitions

In addition to the MBR, each primary partition has a boot sector that can boot the operating system in that partition. MS-DOS (Windows) partitions have this, and hence lilo can optionally load and execute these partition boot sectors to start a Windows installation in another partition. 318

31. lilo, initrd, and Booting

31.3. lilo.conf and the lilo Command

31. lilo, initrd, and Booting

31.2.4

Limitations
15

BIOSs have inherited several limitations because of lack of foresight of their designers.
documentation.

First, some BIOSs do not support more than one IDE. At least according to the lilo I myself have not come across this as a problem.

The second limitation is most important to note. As explained, lilo uses BIOS functions to access the IDE drive, but the BIOS of a PC is often limited to accessing the rst 1024 cylinders of the disk. Hence, whatever LILO reads must reside within the rst 1024 cylinders (the rst 500 megabytes of disk space). Here is the list of things whose sectors are required to be within this space: 1. /boot/vmlinuz 2. Various lilo les /boot/*.b 3. Any non-L INUX partition boot sector you would like to boot However a L INUX root partition can reside anywhere because the boot sector program never reads this partition except for the abovementioned les. A scenario where the /boot/ directory is a small partition below the 500 megabyte boundary and the / partition is above the 500 megabyte boundary, is quite common. See page 155. Note that newer LBA BIOSs support more than the rst 512 megabyteseven up to 8 Gigabytes. I personally do not count on this.

image = /boot/vmlinuz-2.0.38 label = linux-old root = /dev/hda6 read-only other = /dev/hda2 label = win table = /dev/hda

Running lilo will install into the MBR a boot loader that understands where to get the /boot/map le, which in turn understands where to get the /boot/vmlinuz2.2.12-20 le. It gives output like:
[root@cericon]# lilo Added linux * Added linux-old Added win

It also backs up your existing MBR, if this has not previously been done, into a le /boot/boot.0300 (where 0300 refers to the devices major and minor number). Lets go through the options: boot Device to boot. It will most always be /dev/hda or /dev/sda. prompt Display a prompt where the user can enter the OS to boot. timeout How many tenths of a seconds to display the prompt (after which the rst image is booted). compact String together adjacent sector reads. This makes the kernel load much faster. vga We would like 80 50 text mode. Your startup scripts may reset this to 80 25 search /etc/rc.d recursively for any le containing textmode. lock Always default to boot the last OS booted used. . password Require a password to boot. restricted Require a password only if someone attempts to enter special options at the LILO: prompt. append A kernel boot option. Kernel boot options are central to lilo and kernel modules and are discussed in Chapter 42.5. They are mostly not needed in simple installations. image A L INUX kernel to boot.

31.3

lilo.conf and the lilo Command

To do a lilo means running the lilo command as root, with a correct /etc/lilo.conf le. The lilo.conf le will doubtless have been set up by your distribution (check yours). A typical lilo.conf le that allows booting of a Windows partition and two L INUX partitions is as follows:
boot=/dev/hda prompt timeout = 50 compact vga = extended lock password = jAN]")Wo restricted append = "ether=9,0x300,0xd0000,0xd4000,eth0 hisax=1,3,5,0xd8000,0xd80,HiSax" image = /boot/vmlinuz-2.2.17 label = linux root = /dev/hda5 read-only

A very useful feature which is seldom

label The text to type at the boot prompt to cause this kernel/partition to boot. 320

319

31. lilo, initrd, and Booting

31.4. Creating Boot Floppy Disks

31.5. SCSI Installation Complications and initrd

31. lilo, initrd, and Booting

root The root le system that the kernel must mount. read-only Flag to specify that the root le system must initially be mounted readonly. other Some other operating system to boot: in this case, a Windows partition. table Partition table info to be passed to the partition boot sector. Further other = partitions can follow, and many image = kernel images are allowed. The preceding lilo.conf le assumed a partition scheme as follows: /dev/hda1 10-megabyte ext2 partition to be mounted on /boot. /dev/hda2 Windows 98 partition over 500 megabytes in size. /dev/hda3 Extended partition. /dev/hda4 Unused primary partition. /dev/hda5 ext2 root le system. /dev/hda6 Second ext2 root le system containing an older distribution. /dev/hda? L INUX swap, /home, and other partitions.

31.5

SCSI Installation Complications and initrd

Some of the following descriptions may be difcult to understand without knowledge of kernel modules explained in Chapter 42. You may want to come back to it later. Consider a system with zero IDE disks and one SCSI disk containing a L INUX installation. There are BIOS interrupts to read the SCSI disk, just as there were for the IDE, so LILO can happily access a kernel image somewhere inside the SCSI partition. However, the kernel is going to be lost without a kernel module See Chapter 42. The kernel
doesnt support every possible kind of hardware out there all by itself. It is actually divided into a main part (the kernel image discussed in this chapter) and hundreds of modules (loadable parts that reside in that /lib/modules/) that support the many type of SCSI, network, sound etc., peripheral devices.

understands the particular SCSI driver. So although the kernel can load and execute, it wont be able to mount its root le system without loading a SCSI module rst. But the module itself resides in the root le system in /lib/modules/. This is a tricky situation to solve and is done in one of two ways: either (a) using a kernel with preenabled SCSI support or (b) using what is known as an initrd preliminary root le system image.

31.4 Creating Boot Floppy Disks

If LILO is broken or absent, we require an alternative boot method. A oppy disk capable of booting our system must contain a kernel image, the means to load the image into memory, and the means to mount /dev/hda5 as the root le system. To create such a oppy, insert a new oppy disk into a running L INUX system, and overwrite it with the following commands:
dd if=/boot/vmlinuz-2.2.17 of=/dev/fd0 rdev /dev/fd0 /dev/hda5

The rst method is what I recommend. Its a straightforward (though timeconsuming) procedure to create a kernel with SCSI support for your SCSI card built-in (and not in a separate module). Built-in SCSI and network drivers will also autodetect cards most of the time, allowing immediate access to the devicethey will work without being given any options Discussed in Chapter 42. and, most importantly, without your having to read up on how to congure them. This setup is known as compiled-in support for a hardware driver (as opposed to module support for the driver). The resulting kernel image will be larger by an amount equal to the size of module. Chapter 42 discusses such kernel compiles. The second method is faster but trickier. L INUX supports what is known as an initrd image (initial rAM disk image). This is a small, 1.5 megabyte le system that is loaded by LILO and mounted by the kernel instead of the real le system. The kernel mounts this le system as a RAM disk, executes the le /linuxrc, and then only mounts the real le system.

31.6

Creating an initrd Image

Then simply boot the oppy. This procedure requires a second L INUX installation at least. If you only have an MS-DOS or Windows system at your disposal then you will have to download the RAWRITE.EXE utility as well as a raw boot disk image. Many of these are available and will enable you to create a boot oppy from a DOS prompt. I will not go into detail about this here. 321

Start by creating a small le system. Make a directory /initrd and copy the following les into it.
drwxr-xr-x drwxr-xr-x -rwxr-xr-x -rwxr-xr-x 7 2 1 1 root root root root root root root root 1024 1024 436328 424680 Sep Sep Sep Sep 14 14 14 14 20:12 20:12 20:12 20:12 initrd/ initrd/bin/ initrd/bin/insmod initrd/bin/sash

322

31. lilo, initrd, and Booting

31.6. Creating an initrd Image

31.7. Modifying lilo.conf for initrd

31. lilo, initrd, and Booting

drwxr-xr-x crw-r--r-crw-r--r-brw-r--r-crw-r--r-crw-r--r-crw-r--r-crw-r--r-crw-r--r-drwxr-xr-x drwxr-xr-x -rwxr-xr-x drwxr-xr-x

2 1 1 1 1 1 1 1 1 2 2 1 2

root root root root root root root root root root root root root

5, 1, 1, 4, 4, 4, 4, 4,

1024 1 3 1 0 1 1 1 1 1024 1024 76 1024

Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep Sep

14 14 14 14 14 14 14 14 14 14 14 14 14

20:12 20:12 20:12 20:12 20:12 20:12 20:12 20:12 20:12 20:12 20:12 20:12 20:12

initrd/dev/ initrd/dev/console initrd/dev/null initrd/dev/ram initrd/dev/systty initrd/dev/tty1 initrd/dev/tty2 initrd/dev/tty3 initrd/dev/tty4 initrd/etc/ initrd/lib/ initrd/linuxrc initrd/loopfs/

Now, create a le system image similar to that in Section 19.9:

dd if=/dev/zero of=/file-inird count=2500 bs=1024 losetup /dev/loop0 /file-inird mke2fs /dev/loop0 mkdir /mnt mount /dev/loop0 /mnt cp -a initrd/* /mnt/ umount /mnt losetup -d /dev/loop0

/linuxrc

On my system, the le initrd/bin/insmod is the statically linked meaning it version copied from /sbin/insmod.statica member of the modutils-2.3.13 package. initrd/bin/sash is a statically linked shell from the sash-3.4 package. You can recompile insmod from source if you dont have a statically linked version. Alternatively, copy the needed DLLs from /lib/ to initrd/lib/. (You can get the list of required DLLs by running ldd /sbin/insmod. Dont forget to also copy symlinks and run strip -s <lib> to reduce the size of the DLLs.)
does not require shared libraries.

Finally, gzip the le system to an appropriately named le:

gzip -c /file-inird > initrd-<kernel-version>

31.7

Modifying lilo.conf for initrd

Now copy into the initrd/lib/ directory the SCSI modules you require. For example, if we have an Adaptec AIC-7850 SCSI adapter, we would require the aic7xxx.o module from /lib/modules/<version>/scsi/aic7xxx.o. Then, place it in the initrd/lib/ directory.
-rw-r--r-1 root root 129448 Sep 27 1999 initrd/lib/aic7xxx.o

Your lilo.conf le can be changed slightly to force use of an initrd le system. Simply add the initrd option. For example:
boot=/dev/sda prompt timeout = 50 compact vga = extended linear image = /boot/vmlinuz-2.2.17 initrd = /boot/initrd-2.2.17 label = linux root = /dev/sda1 read-only

The le initrd/linuxrc should contain a script to load all the modules needed for the kernel to access the SCSI partition. In this case, just the aic7xxx module insmod can take options such as the IRQ and IO-port for the device. See Chapter 42. :

#!/bin/sash aliasall
5

echo "Loading aic7xxx module" insmod /lib/aic7xxx.o

Notice the use of the linear option. This is a BIOS trick that you can read about in lilo(5). It is often necessary but can make SCSI disks nonportable to different BIOSs (meaning that you will have to rerun lilo if you move the disk to a different computer).

31.8

Using mkinitrd

Now double-check all your permissions and then chroot to the le system for testing.
chroot /initrd /bin/sash

Now that you have learned the manual method of creating an initrd image, you can read the mkinitrd man page. It creates an image in a single command. This is command is peculiar to RedHat. 324

323

32.2. /etc/inittab

32. init, ?getty, and U NIX Run Levels

5 6 7 8 9

Window System Workstation (usually identical to run level 3). Reboot. Undened. Undened. Undened.

Chapter 32

init, ?getty, and U NIX Run Levels

This chapter explains how L INUX (and a U NIX system in general) initializes itself. It follows on from the kernel boot explained in Section 31.2. We also go into some advanced uses for mgetty, like receiving of faxes.

The idea here is that init begins at a particular run level that can then be manually changed to any other by the superuser. init uses a list of scripts for each run level to start or stop each of the many services pertaining to that run level. These scripts are /etc/rc?.d/KNNservice or /etc/rc?.d/SNNservice On some systems /etc/rc.d/rc?.d/. . . . , where NN, K, or S is a prex to force the order of execution (since the les are executed in alphabetical order).

These scripts all take the options start and stop on the command-line, to begin or terminate the service. For example, when init enters, say, run level 5 from run level 3, it executes the particular scripts from /etc/rc3.d/ and /etc/rc5.d/ to bring up or down the appropriate services. This may involve, say, executing
/etc/rc3.d/S20exim stop

and similar commands.

32.1

init the First Process

32.2

/etc/inittab

After the kernel has been unpacked into memory, it begins to execute, initializing hardware. The last thing it does is mount the root le system, which necessarily contains a program /sbin/init, which the kernel executes. init is one of the only programs the kernel ever executes explicitly; the onus is then on init to bring the U NIX system up. init always has the process ID 1. For the purposes of init, the (rather arbitrary) concept of a U NIX run level was invented. The run level is the current operation of the machine, numbered run level 0 through run level 9. When the U NIX system is at a particular run level, it means that a certain selection of services is running. In this way, the machine could be a mail server or an Window workstation depending on what run level it is in. The traditionally dened run levels are: 0 1 2 3 4 Halt. Single-user mode. Multiuser, without network le system (NFS). Full multiuser mode. Unused. 325
5

init has one cong le: /etc/inittab which is scanned once on bootup.

32.2.1 Minimal conguration

A minimal inittab le might consist of the following.
id:3:initdefault: si::sysinit:/etc/rc.d/rc.sysinit l0:0:wait:/etc/rc.d/rc l1:1:wait:/etc/rc.d/rc l2:2:wait:/etc/rc.d/rc l3:3:wait:/etc/rc.d/rc l4:4:wait:/etc/rc.d/rc l5:5:wait:/etc/rc.d/rc l6:6:wait:/etc/rc.d/rc 0 1 2 3 4 5 6

326

32. init, ?getty, and U NIX Run Levels

32.2. /etc/inittab

32.3. Useful Run Levels

32. init, ?getty, and U NIX Run Levels

32.2.2 Rereading inittab

ud::once:/sbin/update
15

1:2345:respawn:/sbin/getty 2:2345:respawn:/sbin/getty 3:2345:respawn:/sbin/getty 4:2345:respawn:/sbin/getty

38400 38400 38400 38400

tty1 tty2 tty3 tty4

If you modify the inittab le, init will probably not notice until you issue it a SIGHUP. This is the same as typing
telinit q

S0:2345:respawn:/sbin/mgetty -n 3 -s 115200 ttyS0 57600 S4:2345:respawn:/sbin/mgetty -r -s 19200 ttyS4 DT19200

which causes init to reread /etc/inittab.

32.2.3 The respawning too fast error

x:5:respawn:/usr/bin/X11/xdm -nodaemon

You get a respawning too fast error when an inittab line makes no sense
tion on it.

The lines are colon-separated elds and have the following meaning (lots more can be gotten from inittab(5)): id:3:initdefault: This dictates that the default run level is 3. It is the run level that the system will boot up into. This eld usually has a 3 or a 5, which are most often the only two run levels that the system ever sits in. si::sysinit:/etc/rc.d/rc.sysinit This says to run a script on bootup to initialize the system. If you view the le /etc/rc.d/rc.sysinit, you will see a fairly long script that does the following: mounts the proc le system; initializes the keyboard maps, console font, NIS domain, host name, and swap partition; runs isapnp and depmod -a; cleans the utmp le; as well as other things. This script is only run once on bootup. On Debian this is a script, /etc/init.d/rcS, that runs everything under /etc/rcS.d/. As usual, Debian gravitated to the most
clean, elegant and extensible solution.

: like a getty running on a non-functioning serial port. Simply comment out or delete the appropriate line and then run
telinit q

These errors are common and very irritating when you are doing console work, hence an explicit sec-

32.3

Useful Run Levels

Switching run levels manually is something that is rarely done. The most common way of shutting down the machine is to use:
shutdown -h now

l3:3:wait:/etc/rc.d/rc 3 The rst eld is a descriptive tag and could be anything. The second is a list of run levels under which the particular script (last eld) is to be invoked: in this case, /etc/rc.d/rc 3 is to be run when entering run level 3. The wait means to pause until /etc/rc.d/rc has nished execution. If you view the le /etc/rc.d/rc, you will see it merely executes scripts under /etc/rc?.d/ as appropriate for a run level change. ud::once:/sbin/update This ushes the disk cache on each run level change. 1:2345:respawn:/sbin/getty 38400 tty1 This says to run the command /sbin/getty 38400 tty1 when in run level 2 through 5. respawn means to restart the process if it dies. x:5:respawn:/usr/bin/X11/xdm -nodaemon This says to run the command /usr/bin/X11/xdm -nodaemon when in run level 5. This is the Window System graphical login program. 327

which effectively goes to run level 0, and

shutdown -r now

which effectively goes to run level 6. or You can also specify the run level at the LILO: prompt. Type

linux 1

linux single

to enter single-user mode when booting your machine. You change to single-user mode on a running system with: 328

32. init, ?getty, and U NIX Run Levels

32.4. getty Invocation

32.6. Incoming Faxes and Modem Logins

32. init, ?getty, and U NIX Run Levels

telinit S

9. getty prompts for login. 10. getty executes /bin/login to authentic user. 11. login starts shell.

You can forcefully enter any run level with

telinit <N>

32.6
32.6.1

Incoming Faxes and Modem Logins

mgetty with character terminals

32.4

getty Invocation

The getty man page begins with:

getty opens a tty port, prompts for a login name and invokes the /bin/login command. It is normally invoked by init(8).

The original purpose of getty was to manage character terminals on mainframe computers. mgetty is a more comprehensive getty that deals with proper serial devices. A typical inittab entry is which would open a login on a terminal connected to a serial line on /dev/ttyS4. See page 479 for information on conguring multiport serial cards. (The L INUX devices /dev/tty1 through /dev/tty12 as used by getty emulate classic terminals in this way.)
S4:2345:respawn:/sbin/mgetty -r -s 19200 ttyS4 DT19200

Note that getty, agetty, fgetty and mingetty are just different implementations of getty. The most noticeable effect of init running at all is that it spawns a login to each of the L INUX virtual consoles. It is the getty (or sometimes mingetty) command as specied in the inittab line above that displays this login. Once the login name is entered, getty invokes the /bin/login program, which then prompts the user for a password. The login program (discussed in Section 11.7) then executes a shell. When the shell dies (as a result of the user exiting the session) getty is just respawned.

32.6.2

mgetty log les

mgetty will log to /var/log/mgetty.log.ttyS?. This log le contains everything you need for troubleshooting. It is worthwhile running tail -f on these les while watching a login take place.

32.5 Bootup Summary

Together with Chapter 31 you should now have a complete picture of the entire bootup process: 1. 2. 3. 4. 5. 6. 7. 8. First sector loaded into RAM and executedLILO: prompt appears. Kernel loaded from sector list. Kernel executed; unpacks. Kernel initializes hardware. Kernel mounts root le system, say /dev/hda1. Kernel executes /sbin/init as PID 1. init executes all scripts for default run level. init spawns getty programs on each terminal. 329

32.6.3

mgetty with modems

Running mgetty (see mgetty(8)) is a common and trivial way to get a dial login to a L INUX machine. Your inittab entry is just where -n 3 says to answer the phone after the 3rd ring. Nothing more is needed than to plug your modem into a telephone. You can then use dip -t, as done in Section 41.1.1, to dial this machine from another L INUX machine. Here is an example session:
S0:2345:respawn:/sbin/mgetty -n 3 -s 115200 ttyS0 57600

This example assumes that an initialization string of AT&F1 is sufcient. See Section 3.5.
[root@cericon]# dip -t DIP: Dialup IP Protocol Driver version 3.3.7o-uri (8 Feb 96) Written by Fred N. van Kempen, MicroWalt Corporation.

330

32. init, ?getty, and U NIX Run Levels

32.6. Incoming Faxes and Modem Logins

32. init, ?getty, and U NIX Run Levels

DIP> port ttyS0 DIP> speed 57600 DIP> term [ Entering TERMINAL mode. Use CTRL-] to get back ] AT&F1 OK ATDT5952521 CONNECT 19200/ARQ/V34/LAPM/V42BIS Red Hat Linux release 6.1 (Cartman) Kernel 2.2.12-20 on an i686

/etc/mgetty+sendfax/new_fax

which is a script that mgetty secretly runs when new faxes arrive. It can be used to convert faxes into something (like .gif graphics les I recommend .png over .gif any day, however. ) readable by typical ofce programs. The following example /etc/mgetty+sendfax/new fax script puts incoming faxes into /home/fax/ as .gif les that all users can access. Modied from the mgetty contribs. Note how it uses the CPU-intensive convert program from the ImageMagic package.

remote.dialup.private login:

#!/bin/sh # you must have pbm tools and they must be in your PATH PATH=/usr/bin:/bin:/usr/X11R6/bin:/usr/local/bin

Note that this is purely a login session having nothing to do with PPP dialup.

HUP="$1" SENDER="$2" PAGES="$3"

32.6.4

mgetty receiving faxes

mgetty receives faxes by default, provided your modem supports faxing

it has not been explicitly disabled with the -D option. An appropriate inittab line is,
S0:2345:respawn:/sbin/mgetty -x 4 -n 3 -s 57600 -I 27 21 7654321 ttyS0 57600

dem says it supports faxing, and this still does not work, you will have to spend a lot of time reading through your modems AT command set manual, as well as the mgetty info documentation. and provided

If your mo-

shift 3 P=1 while [ $P -le $PAGES ] ; do FAX=$1 BASENAME=basename $FAX RES=echo $BASENAME | sed s/.$.$.*/\1/ if [ "$RES" = "n" ] ; then STRETCH="-s" else STRETCH="" fi nice g32pbm $STRETCH $FAX > /tmp/$BASENAME.pbm \ && rm -f $FAX \ && nice convert -colorspace gray -colors 16 -geom \ 50%x50% /tmp/$BASENAME.pbm /home/fax/$BASENAME.gif \ && rm -f /tmp/$BASENAME.pbm \ && chmod 0666 /home/fax/$BASENAME.gif shift P=expr $P + 1 done

The options mean, respectively, to set the debug level to 4, answer after 3 rings, set the port speed to 57600, and set the fax ID number to 27 21 7654321. Alternatively, you can use the line
S0:2345:respawn:/sbin/mgetty ttyS0 57600

and instead put your conguration options in the le mgetty.config under /etc/mgetty+sendfax/:
debug 4 rings 3 speed 57600 fax-id 27 21 7654321

exit 0

Faxes end up in /var/spool/fax/incoming/ as useless .g3 format les, but note how the command gives
strings /sbin/mgetty | grep new_fax

331

332

33.1. Fax Through Printing

33. Sending Faxes

:sh:\ :lp=/dev/null:\ :if=/var/spool/lpd/fax/fax_filter.sh:

Chapter 33

the /var/log/fax log le, see page 198. #!/bin/sh exec 1>>/var/log/fax exec 2>>/var/log/fax

The le fax filter.sh itself could contain a script like this for a modem on /dev/ttyS0:

Remember to rotate

Sending Faxes
This chapter discusses the sendfax program, with reference to the specic example of setting up an articial printer that will automatically use a modem to send its print jobs to remote fax machines.

echo echo echo $@

echo "Starting fax date: I am id" export DISPLAY=localhost:0.0 export HOME=/home/lp

33.1 Fax Through Printing

Continuing from Section 21.10. . . You should go now and read the sendfax section of the info page for mgetty. The sendfax command is just one program that sends faxes through the modem. Like mgetty, it reads a cong le in /etc/mgetty+sendfax/. This cong le is just sendfax.config and can contain as little as
verbose y debug 5 fax-devices ttyS0 fax-id 27 21 7654321 max-tries 3 max-tries-continue y
25

function error() { gdialog --title "Send Fax" --msgbox "$1" 10 75 || \ echo Huh? no gdialog on this machine cd / rm -Rf /tmp/$$fax || \ gdialog \ --title "Send Fax" \ --msgbox "rm -Rf /tmp/$$fax failed" \ 10 75 exit 1 } mkdir /tmp/$$fax || error "mkdir /tmp/$$fax failed" cd /tmp/$$fax || error "cd /tmp/$$fax failed"

cat > fax.ps if /usr/bin/gdialog \ --title "Send Fax" \ --inputbox "Enter the phone number to fax:" \ 10 75 "" 2>TEL ; then : else echo "gdialog failed < TEL" rm -Rf /tmp/$$fax exit 0 fi TEL=< TEL test -z "$TEL" && error no telephone number given

Below, fax filter.sh is a script that sends the print job through the fax machine after requesting the telephone number through gdialog. gdialog is part of the gnome-utils package. An appropriate /etc/printcap entry is:

fax:\ :sd=/var/spool/lpd/fax:\ :mx#0:\

333

334

33. Sending Faxes

33.2. Setgid Wrapper Binary

33. Sending Faxes

cat fax.ps | gs -r204x98 -sOutputFile=- -sDEVICE=faxg3 -dBATCH -q - \ 1>fax.ps.g3 || error gs failed

gcc sendfax_wrapper.c -o /usr/sbin/sendfax_wrapper -Wall chown lp:uucp /usr/sbin/sendfax_wrapper chmod g+s,o-rwx /usr/sbin/sendfax_wrapper

ls -al /var/lock/ /usr/sbin/sendfax -x 5 -n -l ttyS0 $TEL fax.ps.g3 || \ error "sendfax failed" rm -Rf /tmp/$$fax

exit 0

Then, replace sendfax with sendfax wrapper in the lter script. You can see that sendfax wrapper just executes sendfax after changing the group ID to the effective group ID (GID) as obtained from the getegid function on line 12. The effective group ID is uucp because of the setgid group bit (i.e., g+s) in the chmod command, and hence sendfax runs under the uucp group with full access to the modem device. On your own system it may be cleaner to try implement this without a wrapper. Debian , for example, has a dialout group for the purposes of accessing modems. Also be aware that some distributions may not use the uucp user in the way RedHat does and you may have to create an alternative user especially for this task.

33.2 Setgid Wrapper Binary

The above script is not enough however. Above, sendfax requires access to the /dev/ttyS0 device as well as the /var/lock/ directory (to create a modem lock lesee Section 34.4). It cannot do that as the lp user (under which the above lter runs). On RedHat, the command ls -ald /var/lock /dev/ttyS0 reveals that only uucp is allowed to access modems. We can get around this restriction by creating a setgid (see Chapter 14) binary that runs as the uucp user. Do this by compiling the program,
#include #include #include #include
5

<stdlib.h> <string.h> <stdio.h> <unistd.h>

int main (int argc, char argv) { char a; int i;

/* set the real group ID to that of the effective group ID */ if (setgid (getegid ())) { perror ("sendfax_wrapper: setgid failed"); exit (1); } /* copy all arguments */ a = (char **) malloc ((argc + 1) * sizeof (char *)); for (i = 1; i < argc; i++) a[i] = (char *) strdup (argv[i]); a[argc] = NULL; /* execute sendfax */ a[0] = "/usr/sbin/sendfax"; execvp (a[0], a); /* exit on failure */ perror ("sendfax_wrapper: failed to exececute /usr/sbin/sendfax"); exit (1); }

using the commands, 335 336

34.1. Command-Line Operation

34. uucp and uux

34.1

Command-Line Operation

To copy a le from one machine to another, simply enter

uucp <filename> <machine>!<path>

Chapter 34

You can also run commands on the remote system, like

uux - cericon!rmail john

echo -n Hi, this is a short message\n\n-paul | \

uucp and uux

uucp is a command to copy a le from one U NIX system to another. uux executes a command on another U NIX system, even if that command is receiving data through stdin on the local system. uux is extremely useful for automating many kinds of distributed functions, like mail and news. The uucp and uux commands both come as part of the uucp (Unix-to-Unix Copy) package. uucp may sound ridiculous considering the availability of modern commands like rcp, rsh, or even FTP transfers (which accomplish the same thing), but uucp has features that these do not, making it an essential, albeit antiquated, utility. For instance, uucp never executes jobs immediately. It will, for example, queue a le copy for later processing and then dial the remote machine during the night to complete the operation. uucp predates the Internet: It was originally used to implement a mail system, using only modems and telephone lines. It hence has sophisticated protocols for ensuring that your le/command really does get there, with the maximum possible fault tolerance and the minimum of retransmission. This is why it should always be used for automated tasks wherever there are unreliable (i.e., modem) connections. The uucp version that comes with most L INUX distributions is called Taylor UUCP after its author. Especially important is that when a uucp operation is interrupted by a line break, the connection time is not wasted: uucp will not have discarded any partially transmitted data. This means that no matter how slow or error prone the connection, progress is always made. Compare this to an SMTP or POP3/IMAP connection: Any line break halfway through a large mail message will necessitate that the entire operation to be restarted from scratch. 337

which runs rmail on the remote system cericon, feeding some text to the rmail program. Note how you should quote the ! character to prevent it from being interpreted by the shell. (These commands will almost always fail with permission denied by remote. The error will come in a mail message to the user that ran the command.)

34.2

Conguration

uucp comes with comprehensive documentation in HTML format (/usr/doc/uucpversion/uucp.html or /usr/share/. . . ) on RedHat, and info format on Debian and RedHat. Here, I sketch a basic and typical conguration. The uucp package has a long history of revisions, beginning with the rst modem-based mail networks. The latest GNU editions that come with L INUX distributions have a conguration le format that will probably differ from that which old uucp hands are used to. Dialup networks today typically use uucp in combination with normal PPP dialup, probably not using uucps dial-in facilities at all. For example, if you are deploying a number of remote hosts that are using modems, these hosts should always use uucp to upload and retrieve mail, rather than POP3/IMAP or straight SMTP, because of the retransmission problem discussed above. In other words, uucp is really working as an ordinary TCP service, albeit with far more fault tolerance.
uucp

To make uucp into a TCP server, place it into /etc/inetd.conf as follows

stream tcp nowait uucp /usr/sbin/tcpd /usr/lib/uucp/uucico -l

being also very careful to limit the hosts that can connect by using the techniques discussed in Chapter 29. Similarly for xinetd, create a le /etc/xinetd.d/uucp containing, 338

34. uucp and uux

34.2. Conguration

34. uucp and uux

service uucp { only_from socket_type wait user server server_args disable }

= = = = = = =

127.0.0.1 192.168.0.0/16 stream no uucp /usr/lib/uucp/uucico -l no

up the user and password in the /etc/uucp/passwd le, and protocol t means to use a simple non-error, correcting protocol (as appropriate for use over TCP). The commands option takes a space-separated list of permitted commandsfor security reasons, commands not in this list cannot be executed. (This is why I stated above that commands will almost always fail with permission denied by remotethey are usually not listed under commands.) The /etc/uucp/sys le on machine1 will contain:

uucp conguration les are stored under /etc/uucp/. Now we congure a client machine, machine1.cranzgot.co.za, to send mail through server1.cranzgot.co.za, where server1.cranzgot.co.za is running the uucico service above. uucp has an antiquated authentication mechanism that uses its own list of users and passwords completely distinct from those of ordinary U NIX accounts. We must rst add a common user and password to both machines for authentication purposes. For machine1.cranzgot.co.za, we can add to the le /etc/uucp/call the line
server1 machine1login pAsSwOrD123

Here time any species which times of the day uucp may make calls to server1. The default is time Never. See the uucp documentation under Time Strings for more info. The option port TCP means that we are using a modem named TCP to execute the dialout. All modems are dened in the le /etc/uucp/port. We can add our modem entry to /etc/uucp/port as follows,

system server1 call-login * call-password * time any port TCP address 192.168.3.2 protocol t

which tells uucp to use the login machine1login whenever trying to speak to server1. On server1.cranzgot.co.za we can add to the le /etc/uucp/passwd the line,
machine1login pAsSwOrD123

which clearly is not really a modem at all. Finally, we can queue a mail transfer job with

port TCP type tcp

Note that the uucp name server1 was chosen for the machine server1.cranzgot.co.za for convenience. uucp names, however, have nothing to do with domain names. Next, we need to tell uucp about the intentions of machine1. Any machine that you might connect to or from must be listed in the /etc/uucp/sys le. Our entry looks like
system machine1 call-login * call-password * commands rmail protocol t

and copy a le with

echo -e Hi Jack\n\nHow are you?\n\n-jill" | \ uux - --nouucico server1!rmail jack@beanstalk.com

Note that /var/spool/uucppublic/ is the only directory you are allowed access to by default. You should probably keep it this way for security.

uucp --nouucico README cericon!/var/spool/uucppublic

uucico
Although we have queued a job for processing, nothing will transfer until the program uucico (which stands for Unix-to-Unix copy in copy out) is run. The idea is that both server1 and machine1 may have queued a number of jobs; then when uucico is running on both machines and talking to each other, all jobs on both machines are processed in turn, regardless of which machine initiated the connection. 340

and can have as many entries as we like. The only things server1 has to know about machine1 are the user and password and the preferred protocol. The *s mean to look 339

34. uucp and uux

34.3. Modem Dial

34.4. tty/UUCP Lock Files

34. uucp and uux

Usually uucico is run from a crond script every hour. (Even having run uucico, nothing will transfer if the time of day does not come within the ranges specied under time ....) Here we can run tail -f /var/log/uucp/Log while running uucico manually as follows:
uucico --debug 3 --force --system server1

ACU is antiquated terminology and stands for Automatic Calling Unit (i.e., a modem). We have to specify the usual types of things for serial ports, like the device (/dev/ttyS0 for a modem on COM1) and speed of the serial line. We also must specify a means to initialize the modem: the dialer mymodem option. A le /etc/uucp/dial should then contain an entry for our type of modem matching mymodem as follows: This example assumes that an initialization string of AT&F1 is sufcient.
See Section 3.5.

The higher the debug level, the more verbose output you will see in the Log le. This will --forceably dial the --system server1 regardless of when it last dialed (usually there are constraints on calling soon after a failed call: --force overrides this). If your mail server on server1 is congured correctly, it should now have queued the message on the remote side.
5

34.3 Modem Dial

If you are really going to use uucp the old-fashioned way, you can use mgetty to answer uucp calls on server1 by adding the following to your /etc/inittab le:
S0:2345:respawn:/sbin/mgetty -s 57600 ttyS0

dialer mymodem chat "" AT&F1\r\d\c OK\r ATDT\D CONNECT chat-fail RING chat-fail NO\sCARRIER chat-fail ERROR chat-fail NO\sDIALTONE chat-fail BUSY chat-fail NO\sANSWER chat-fail VOICE complete \d\d+++\d\dATH\r\c abort \d\d+++\d\dATH\r\c

More about modems and dialing is covered with pppd in Chapter 41. With the modem properly specied, we can change our entry in the sys le to

And then add the line

machine1login uucp machine1login /usr/sbin/uucico -l -u machine1login

to the le /etc/mgetty+sendfax/login.config (/etc/mgetty/login.config for Debian ). You will then also have to add a U NIX account machine1login with password pAsSwOrD123. This approach works is because mgetty and uucico have the same login prompt and password prompt, but mgetty uses /etc/passwd instead of /etc/uucp/passwd to authenticate. Also, for a modem connection, protocol t is error prone: change it to protocol g, which has small packet sizes and error correction. Note that the above conguration also supports faxes, logins, voice, and PPP (see Section 41.4) on the same modem, because mgetty only starts uucico if the user name is machine1login. To dial out from machine1, you rst need to add a modem device (besides TCP) to your /etc/uucp/port le:
port ACU type modem device /dev/ttyS0 dialer mymodem speed 57600

system server1 call-login * call-password * time any port ACU phone 555-6789 protocol g

The same uux commands should now work over dialup.

34.4

tty/UUCP Lock Files

I hinted about lock les in Section 33.2. A more detailed explanation follows. You will have noticed by now that several services use serial devices, and many of them can use the same device at different times. This creates a possible conict should two services wish to use the same device at the same time. For instance, what if someone wants to send a fax, while another person is dialing in? The solution is the UUCP lock le. This is a le created by a process in /var/lock/ of the form LCK..device that indicates the serial port is being used by that process. For instance, when running sendfax through a modem connected on 342

341

34. uucp and uux

34.5. Debugging uucp

34.6. Using uux with exim

34. uucp and uux

/dev/ttyS0, a le /var/lock/LCK..ttyS0 suddenly appears. This is because sendfax, along with all other mgetty programs, obeys the UUCP lock le convention. The contents of this le actually contain the process ID of the program using the serial device, so it is easy to check whether the lock le is bogus. A lock le of such a dead process is called a stale lock le and can be removed manually.

#################### MAIN CONFIGURATION SETTINGS ##################### log_subject errors_address = admin local_domains = localhost : ${primary_hostname} : machine1 : \ machine1.cranzgot.co.za host_accept_relay = 127.0.0.1 : localhost : ${primary_hostname} : \ machine1 : machine1.cranzgot.co.za never_users = root exim_user = mail exim_group = mail end ###################### TRANSPORTS CONFIGURATION ###################### uucp: driver = pipe user = nobody command = "/usr/bin/uux - --nouucico ${host}!rmail \ ${local_part}@${domain}" return_fail_output = true local_delivery: driver = appendfile file = /var/spool/mail/${local_part} delivery_date_add envelope_to_add return_path_add group = mail mode_fail_narrower = mode = 0660 end ###################### DIRECTORS CONFIGURATION ####################### localuser: driver = localuser transport = local_delivery end ###################### ROUTERS CONFIGURATION ######################### touucp: driver = domainlist route_list = "* server1" transport = uucp end ###################### RETRY CONFIGURATION ########################### * * F,2m,1m end

34.5 Debugging uucp

uucp implementations rarely run smoothly the rst time. Fortunately, you have available a variety of verbose debugging options. uucico takes the --debug option to specify the level of debug output. You should examine the les /var/log/uucp/Log, /var/log/uucp/Debug, and /var/log/uucp/Stats to get an idea about what is going on in the background. Also important is the spool directory /var/spool/uucp/. You can specify the debugging level with -debug level where level is in the range of 0 through 11. You can also use -debug chat to only see modem communication details. A full description of other options follows Credits to the uucp documentation. :

--debug abnormal Output debugging messages for abnormal situations, such as recoverable errors. --debug chat Output debugging messages for chat scripts. --debug handshake Output debugging messages for the initial handshake. --debug uucp protocol Output debugging messages for the UUCP session protocol. --debug proto Output debugging messages for the individual link protocols. --debug port Output debugging messages for actions on the communication port. --debug config Output debugging messages while reading the conguration les. --debug spooldir Output debugging messages for actions in the spool directory. --debug execute Output debugging messages whenever another program is executed. --debug incoming List all incoming data in the debugging le. --debug outgoing List all outgoing data in the debugging le. --debug all All of the above.

34.6 Using uux with exim

On machine1 we would like exim to spool all mail through uucp. Using uucp requires a pipe transport (exim transports are discussed in Section 30.3.2). exim merely sends mail through stdin of the uux command and then forgets about it. uux is then responsible for executing rmail on server1. The complete exim.conf le is simply as follows. 343

On machine server1, exim must however be running as a full-blown mail server to properly route the mail elsewhere. Of course, on server1, rmail is the sender; hence, it appears to exim that the mail is coming from the local machine. This means that no extra conguration is required to support mail coming from a uux command. Note that you can add further domains to your route list so that your dialouts occur directly to the recipients machine. For instance: 344

34. uucp and uux

34.6. Using uux with exim

34.7. Scheduling Dialouts

34. uucp and uux

route_list = "machine2.cranzgot.co.za machine2 machine3.cranzgot.co.za machine3 * machine2 machine2 machine3 machine3 server1" ; ; ; ; \ \ \ \

34.7

Scheduling Dialouts

Above, we used uucico only manually. uucico does not operate as a daemon process on its own and must be invoked by crond. All systems that use uucp have a /etc/crontab entry or a script under /etc/cron.hourly. A typical /etc/crontab for machine1 might contain:

You can then add further entries to your /etc/uucp/sys le as follows:

system machine2 call-login * call-password * time any port ACU phone 555-6789 protocol g system machine3 call-login * call-password * time any port ACU phone 554-3210 protocol g

45 * * * * uucp /usr/lib/uucp/uucico --master 40 8,13,18 * * * root /usr/bin/uux -r server1!

The option --master tells uucico to loop through all pending jobs and call any machines for which jobs are queued. It does this every hour. The second line queues a null command three times daily for the machine server1. This will force uucico to dial out to server1 at least three times a day on the appearance of real work to be done. The point of this to pick up any jobs coming the other way. This process is known as creating a poll le. Clearly, you can use uucp over a TCP link initiated by pppd. If a dial link is running in demand mode, a uucp call will trigger a dialout and make a straight TCP connection through to the remote host. A common situation occurs when a number of satellite systems are dialing an ISP that has no uucp facility. To service the satellite machines, a separate uucp server is deployed that has no modems of its own. The server will have a permanent Internet connection and listen on TCP for uucp transfers.

The exim.conf le on server1 must also have a router to get mail back to machine1. The router will look like this:
###################### ROUTERS CONFIGURATION ######################### touucp: driver = domainlist route_list = "machine2.cranzgot.co.za machine2 ; \ machine2 machine2 ; \ machine3.cranzgot.co.za machine3 ; \ machine3 machine3" transport = uucp lookuphost: driver = lookuphost transport = remote_smtp end

This router sends all mail matching our dial-in hosts through the uucp transport while all other mail (destined for the Internet) falls through to the lookuphost router. 345 346

35. The L INUX File System Standard

All trademarks and copyrights are owned by their owners, unless specically noted otherwise. Use of a term in this document should not be regarded as affecting the validity of any trademark or service mark.

Chapter 35

The L INUX File System Standard

A This chapter reproduces the Filesystem Hierarchy Standard, translated into L TEX with some minor formatting changes and the addition of this books chapter number to all the section headers. An original can be obtained from the FHS home page http://www.pathname.com/fhs/.

If you have ever asked the questions Where in my le system does le xxx go? or What is directory yyy for?, then consult this document. It can be considered to provide the nal word on such matters. Although this is mostly a reference for people creating new L INUX distributions, all administrators can benet from an understanding of the rulings and explanations provided here.

Filesystem Hierarchy Standard Version 2.2 nal

Filesystem Hierarchy Standard Group edited by Rusty Russell and Daniel Quinlan Copyright c 1994-2000 Daniel Quinlan Copyright c 2001 Paul Rusty Russell Permission is granted to make and distribute verbatim copies of this standard provided the copyright and this permission notice are preserved on all copies. ABSTRACT This standard consists of a set of requirements and guidelines for le and directory placement under U NIX-like operating systems. The guidelines are intended to support interoperability of applications, system administration tools, development tools, and scripts as well as greater uniformity of documentation for these systems. May 23, 2001 Permission is granted to copy and distribute modied versions of this standard under the conditions for verbatim copying, provided also that the title page is labeled as modied including a reference to the original standard, provided that information on retrieving the original standard is included, and provided that the entire resulting derived work is distributed under the terms of a permission notice identical to this one. Permission is granted to copy and distribute translations of this standard into another language, under the above conditions for modied versions, except that this permission notice may be stated in a translation approved by the copyright holder.

347

348

35. The L INUX File System Standard

35.1. Introduction

35.2. The Filesystem

35. The L INUX File System Standard

35.1 Introduction
35.1.1 Purpose
This standard enables Software to predict the location of installed les and directories, and Users to predict the location of installed les and directories. We do this by Specifying guiding principles for each area of the lesystem, Specifying the minimum les and directories required, Enumerating exceptions to the principles, and Enumerating specic cases where there has been historical conict. The FHS document is used by Independent software suppliers to create applications which are FHS compliant, and work with distributions which are FHS complaint, OS creators to provide systems which are FHS compliant, and Users to understand and maintain the FHS compliance of a system.

Static data includes binaries, libraries, documentation, and anything that does not change without system administrator intervention; variable data is anything else that does change without system administrator intervention.
BEGIN RATIONALE

The distinction between shareable and unshareable data is needed for several reasons: In a networked environment (i.e., more than one host at a site), there is a good deal of data that can be shared between different hosts to save space and ease the task of maintenance. In a networked environment, certain les contain information specic to a single host. Therefore these lesystems cannot be shared (without taking special measures). Historical implementations of U NIX-like lesystems interspersed shareable and unshareable data in the same hierarchy, making it difcult to share large portions of the lesystem. The shareable distinction can be used to support, for example: A /usr partition (or components of /usr) mounted (read-only) through the network (using NFS). A /usr partition (or components of /usr) mounted from read-only media. A CDROM is one copy of many identical ones distributed to other users by the postal mail system and other methods. It can thus be regarded as a read-only lesystem shared with other FHS-compliant systems by some kind of network. The static versus variable distinction affects the lesystem in two major ways: Since / contains both variable and static data, it needs to be mounted read-write. Since the traditional /usr contains both variable and static data, and since we may want to mount it read-only (see above), it is necessary to provide a method to have /usr mounted read-only. This is done through the creation of a /var hierarchy that is mounted read-write (or is a part of another read-write partition, such as /), taking over much of the /usr partitions traditional functionality. Here is a summarizing chart. This chart is only an example for a common FHS-compliant system, other chart layouts are possible within FHS-compliance. shareable /usr /opt /var/mail /var/spool/news unshareable /etc /boot /var/run /var/lock

35.1.2

Conventions

A constant-width font is used for displaying the names of les and directories. Components of lenames that vary are represented by a description of the contents enclosed in < and > characters, <thus>. Electronic mail addresses are also enclosed in < and > but are shown in the usual typeface. Optional components of lenames are enclosed in [ and ] characters and may be combined with the < and > convention. For example, if a lename is allowed to occur either with or without an extension, it might be represented by <filename>[.<extension>]. Variable substrings of directory names and lenames are indicated by *.

35.2 The Filesystem

This standard assumes that the operating system underlying an FHS-compliant le system supports the same basic security features found in most U NIX lesystems. It is possible to dene two independent categories of les: shareable vs. unshareable and variable vs. static. There should be a simple and easily understandable mapping from directories to the type of data they contain: directories may be mount points for other lesystems with different characteristics from the lesystem on which they are mounted. Shareable data is that which can be shared between several different hosts; unshareable is that which must be specic to a particular host. For example, user home directories are shareable data, but device lock les are not.

static variable

END RATIONALE

349

350

35. The L INUX File System Standard

35.3. The Root Filesystem

35. The L INUX File System Standard

35.3 The Root Filesystem

35.3.1 Purpose
The contents of the root lesystem must be adequate to boot, restore, recover, and/or repair the system. To boot a system, enough must be present on the root partition to mount other lesystems. This includes utilities, conguration, boot loader information, and other essential startup data. /usr, /opt, and /var are designed such that they may be located on other partitions or lesystems. To enable recovery and/or repair of a system, those utilities needed by an experienced maintainer to diagnose and reconstruct a damaged system must be present on the root lesystem. To restore a system, those utilities needed to restore from system backups (on oppy, tape, etc.) must be present on the root lesystem.
BEGIN RATIONALE

35.3.2 Requirements
The following directories, or symbolic links to directories, are required in /. / the root directory bin boot dev etc lib mnt opt sbin tmp usr var Essential command binaries Static les of the boot loader Device les Host-specic system conguration Essential shared libraries and kernel modules Mount point for mounting a lesystem temporarily Add-on application software packages Essential system binaries Temporary les Secondary hierarchy Variable data

The primary concern used to balance these considerations, which favor placing many things on the root lesystem, is the goal of keeping root as small as reasonably possible. For several reasons, it is desirable to keep the root lesystem small: It is occasionally mounted from very small media. The root lesystem contains many system-specic conguration les. Possible examples include a kernel that is specic to the system, a specic hostname, etc. This means that the root lesystem isnt always shareable between networked systems. Keeping it small on servers in networked systems minimizes the amount of lost space for areas of unshareable les. It also allows workstations with smaller local hard drives. While you may have the root lesystem on a large partition, and may be able to ll it to your hearts content, there will be people with smaller partitions. If you have more les installed, you may nd incompatibilities with other systems using root lesystems on smaller partitions. If you are a developer then you may be turning your assumption into a problem for a large number of users. Disk errors that corrupt data on the root lesystem are a greater problem than errors on any other partition. A small root lesystem is less prone to corruption as the result of a system crash. Software must never create or require special les or subdirectories in the root directory. Other locations in the FHS hierarchy provide more than enough exibility for any package. There are several reasons why introducing a new subdirectory of the root lesystem is prohibited: It demands space on a root partition which the system administrator may want kept small and simple for either performance or security reasons. It evades whatever discipline the system administrator may have set up for distributing standard le hierarchies across mountable volumes.
END RATIONALE

Each directory listed above is specied in detail in separate subsections below. /usr and /var each have a complete section in this document due to the complexity of those directories.

35.3.3 Specic Options

The following directories, or symbolic links to directories, must be in /, if the corresponding subsystem is installed: / the root directory home lib<qual> root User home directories (optional) Alternate format essential shared libraries (optional) Home directory for the root user (optional)

Each directory listed above is specied in detail in separate subsections below.

351

352

35. The L INUX File System Standard

35.3. The Root Filesystem

35. The L INUX File System Standard

35.3.4
35.3.4.1

/bin : Essential user command binaries (for use by all users)

Purpose

true umount uname

Utility to do nothing, successfully Utility to unmount le systems Utility to print system information

/bin contains commands that may be used by both the system administrator and by users, but which are required when no other lesystems are mounted (e.g. in single user mode). It may also contain commands which are used indirectly by scripts.1

If /bin/sh is not a true Bourne shell, it must be a hard or symbolic link to the real shell command. The [ and test commands must be placed together in either /bin or /usr/bin.
BEGIN RATIONALE

35.3.4.2

Requirements

There must be no subdirectories in /bin. The following commands, or symbolic links to commands, are required in /bin. cat chgrp chmod chown cp date dd df dmesg echo false hostname kill ln login ls mkdir mknod more mount mv ps pwd rm rmdir sed sh stty su sync Utility to concatenate les to standard output Utility to change le group ownership Utility to change le access permissions Utility to change le owner and group Utility to copy les and directories Utility to print or set the system data and time Utility to convert and copy a le Utility to report lesystem disk space usage Utility to print or control the kernel message buffer Utility to display a line of text Utility to do nothing, unsuccessfully Utility to show or set the systems host name Utility to send signals to processes Utility to make links between les Utility to begin a session on the system Utility to list directory contents Utility to make directories Utility to make block or character special les Utility to page through text Utility to mount a lesystem Utility to move/rename les Utility to report process status Utility to print name of current working directory Utility to remove les or directories Utility to remove empty directories The sed stream editor The Bourne command shell Utility to change and print terminal line settings Utility to change user ID Utility to ush lesystem buffers

For example bash behaves differently when called as sh or bash. The use of a symbolic link also allows users to easily see that /bin/sh is not a true Bourne shell. The requirement for the [ and test commands to be included as binaries (even if implemented internally by the shell) is shared with the POSIX.2 standard.
END RATIONALE

35.3.4.3

Specic Options

The following programs, or symbolic links to programs, must be in /bin if the corresponding subsystem is installed: csh ed tar cpio gzip gunzip zcat netstat ping The C shell (optional) The ed editor (optional) The tar archiving utility (optional) The cpio archiving utility (optional) The GNU compression utility (optional) The GNU uncompression utility (optional) The GNU uncompression utility (optional) The network statistics utility (optional) The ICMP network test utility (optional)

If the gunzip and zcat programs exist, they must be symbolic or hard links to gzip. /bin/csh may be a symbolic link to /bin/tcsh or /usr/bin/tcsh.
BEGIN RATIONALE

The tar, gzip and cpio commands have been added to make restoration of a system possible (provided that / is intact). Conversely, if no restoration from the root partition is ever expected, then these binaries might be omitted (e.g., a ROM chip root, mounting /usr through NFS). If restoration of a system is planned through the network, then ftp or tftp (along with everything necessary to get an ftp connection) must be available on the root partition.
END RATIONALE

35.3.5 /boot : Static les of the boot loader

35.3.5.1 Purpose
This directory contains everything required for the boot process except conguration les and the map installer. Thus /boot stores data that is used before the kernel begins executing user-

1 Command binaries that are not essential enough to place into /bin must be placed in /usr/bin, instead. Items that are required only by non-root users (the X Window System, chsh, etc.) are generally not essential enough to be placed into the root partition.

353

354

35. The L INUX File System Standard

35.3. The Root Filesystem

35. The L INUX File System Standard

mode programs. This may include saved master boot sectors, sector map les, and other data that is not directly edited by hand.2

35.3.7.3

Specic Options

35.3.5.2

Specic Options

The following directories, or symbolic links to directories must be in /etc, if the corresponding subsystem is installed: /etc Host-specic system conguration X11 Conguration for the X Window System (optional) Conguration for SGML and XML (optional)

The operating system kernel must be located in either / or /boot.3

35.3.6
35.3.6.1

/dev : Device les

Purpose

sgml

The following les, or symbolic links to les, must be in /etc if the corresponding subsystem is installed:5 csh.login exports fstab ftpusers gateways gettydefs group host.conf hosts hosts.allow hosts.deny hosts.equiv hosts.lpd inetd.conf inittab issue ld.so.conf motd mtab mtools.conf networks passwd printcap profile protocols resolv.conf rpc securetty services shells syslog.conf Systemwide initialization le for C shell logins (optional) NFS lesystem access control list (optional) Static information about lesystems (optional) FTP daemon user access control list (optional) File which lists gateways for routed (optional) Speed and terminal settings used by getty (optional) User group le (optional) Resolver conguration le (optional) Static information about host names (optional) Host access le for TCP wrappers (optional) Host access le for TCP wrappers (optional) List of trusted hosts for rlogin, rsh, rcp (optional) List of trusted hosts for lpd (optional) Conguration le for inetd (optional) Conguration le for init (optional) Pre-login message and identication le (optional) List of extra directories to search for shared libraries (optional) Post-login message of the day le (optional) Dynamic information about lesystems (optional) Conguration le for mtools (optional) Static information about network names (optional) The password le (optional) The lpd printer capability database (optional) Systemwide initialization le for sh shell logins (optional) IP protocol listing (optional) Resolver conguration le (optional) RPC protocol listing (optional) TTY access control for root login (optional) Port names for network services (optional) Pathnames of valid login shells (optional) Conguration le for syslogd (optional)

The /dev directory is the location of special or device les.

35.3.6.2

Specic Options

If it is possible that devices in /dev will need to be manually created, /dev must contain a command named MAKEDEV, which can create devices as needed. It may also contain a MAKEDEV.local for any local devices. If required, MAKEDEV must have provisions for creating any device that may be found on the system, not just those that a particular implementation installs.

35.3.7
35.3.7.1

/etc : Host-specic system conguration

Purpose

/etc contains conguration les and directories that are specic to the current system.4

35.3.7.2

Requirements

No binaries may be located under /etc. The following directories, or symbolic links to directories are required in /etc: /etc Host-specic system conguration opt Conguration for /opt

2 Programs necessary to arrange for the boot loader to be able to boot a le must be placed in /sbin. Conguration les for boot loaders must be placed in /etc. 3 On some i386 machines, it may be necessary for /boot to be located on a separate partition located completely below cylinder 1024 of the boot device due to hardware constraints. Certain MIPS systems require a /boot partition that is a mounted MS-DOS lesystem or whatever other lesystem type is accessible for the rmware. This may result in restrictions with respect to usable lenames within /boot (only for affected systems). 4 The setup of command scripts invoked at boot time may resemble System V, BSD or other models. Further specication in this area may be added to a future version of this standard.

5 Systems that use the shadow password suite will have additional conguration les in /etc (/etc/shadow and others) and programs in /usr/sbin (useradd, usermod, and others).

355

356

35. The L INUX File System Standard

35.3. The Root Filesystem

35. The L INUX File System Standard

mtab does not t the static nature of /etc: it is excepted for historical reasons.6

35.3.7.6

/etc/sgml : Conguration les for SGML and XML (optional)

35.3.7.4 35.3.7.4.1

/etc/opt : Conguration les for /opt Purpose

35.3.7.6.1 Purpose
Generic conguration les dening high-level parameters of the SGML or XML systems are installed here. Files with names *.conf indicate generic conguration les. File with names *.cat are the DTD-specic centralized catalogs, containing references to all other catalogs needed to use the given DTD. The super catalog le catalog references all the centralized catalogs.

Host-specic conguration les for add-on application software packages must be installed within the directory /etc/opt/<package>, where <package> is the name of the subtree in /opt where the static data from that package is stored.

35.3.7.4.2

Requirements

35.3.8 /home : User home directories (optional)

35.3.8.1 Purpose
/home is a fairly standard concept, but it is clearly a site-specic lesystem.8 The setup will differ from host to host. Therefore, no program should rely on this location.9

No structure is imposed on the internal arrangement of /etc/opt/<package>. If a conguration le must reside in a different location in order for the package or system to function properly, it may be placed in a location other than /etc/opt/<package>.
BEGIN RATIONALE

Refer to the rationale for /opt.

END RATIONALE

35.3.9 /lib : Essential shared libraries and kernel modules

35.3.9.1 Purpose
The /lib directory contains those shared library images needed to boot the system and run the commands in the root lesystem, ie. by binaries in /bin and /sbin.10

35.3.7.5 35.3.7.5.1

/etc/X11 : Conguration for the X Window System (optional) Purpose

/etc/X11 is the location for all X11 host-specic conguration. This directory is necessary to allow local control if /usr is mounted read only.

35.3.9.2

Requirements

35.3.7.5.2

Specic Options

The following les, or symbolic links to les, must be in /etc/X11 if the corresponding subsystem is installed: Xconfig XF86Config Xmodmap The conguration le for early versions of XFree86 (optional) The conguration le for XFree86 versions 3 and 4 (optional) Global X11 keyboard modication le (optional)

At least one of each of the following lename patterns are required (they may be les, or symbolic links): libc.so.* ld* The dynamically-linked C library (optional) The execution time linker/loader (optional)

Subdirectories of /etc/X11 may include those for xdm and for any other programs (some window managers, for example) that need them.7 We recommend that window managers with only one conguration le which is a default .*wmrc le must name it system.*wmrc (unless there is a widely-accepted alternative name) and not use a subdirectory. Any window manager subdirectories must be identically named to the actual window manager binary.

6 On some Linux systems, this may be a symbolic link to /proc/mounts, in which case this exception is not required. 7 /etc/X11/xdm holds the conguration les for xdm. These are most of the les previously found in /usr/lib/X11/xdm. Some local variable data for xdm is stored in /var/lib/xdm.

8 Different people prefer to place user accounts in a variety of places. This section describes only a suggested placement for user home directories; nevertheless we recommend that all FHS-compliant distributions use this as the default location for home directories. On small systems, each users directory is typically one of the many subdirectories of /home such as /home/smith, /home/torvalds, /home/operator, etc. On large systems (especially when the /home directories are shared amongst many hosts using NFS) it is useful to subdivide user home directories. Subdivision may be accomplished by using subdirectories such as /home/staff, /home/guests, /home/students, etc. 9 If you want to nd out a users home directory, you should use the getpwent(3) library function rather than relying on /etc/passwd because user information may be stored remotely using systems such as NIS. 10 Shared libraries that are only necessary for binaries in /usr (such as any X Window binaries) must not be in /lib. Only the shared libraries required to run binaries in /bin and /sbin may be here. In particular, the library libm.so.* may also be placed in /usr/lib if it is not required by anything in /bin or /sbin.

357

358

35. The L INUX File System Standard

35.3. The Root Filesystem

35. The L INUX File System Standard

If a C preprocessor is installed, /lib/cpp must be a reference to it, for historical reasons.11

35.3.12 /opt : Add-on application software packages

35.3.12.1 Purpose
/opt is reserved for the installation of add-on application software packages. A package to be installed in /opt must locate its static les in a separate /opt/<package> directory tree, where <package> is a name that describes the software package.

35.3.9.3

Specic Options

The following directories, or symbolic links to directories, must be in /lib, if the corresponding subsystem is installed: /lib essential shared libraries and kernel modules modules Loadable kernel modules (optional)

35.3.12.2
/opt

Requirements
Add-on application software packages <package> Static package objects

35.3.10
35.3.10.1

/lib<qual> : Alternate format essential shared libraries (optional)

Purpose

The directories /opt/bin, /opt/doc, /opt/include, /opt/info, /opt/lib, and /opt/man are reserved for local system administrator use. Packages may provide front-end les intended to be placed in (by linking or copying) these reserved directories by the local system administrator, but must function normally in the absence of these reserved directories. Programs to be invoked by users must be located in the directory /opt/<package>/bin. If the package includes U NIX manual pages, they must be located in /opt/<package>/man and the same substructure as /usr/share/man must be used. Package les that are variable (change in normal operation) must be installed in /var/opt. See the section on /var/opt for more information. Host-specic conguration les must be installed in /etc/opt. See the section on /etc for more information. No other package les may exist outside the /opt, /var/opt, and /etc/opt hierarchies except for those package les that must reside in specic locations within the lesystem tree in order to function properly. For example, device lock les must be placed in /var/lock and devices must be located in /dev. Distributions may install software in /opt, but must not modify or delete software installed by the local system administrator without the assent of the local system administrator.
BEGIN RATIONALE

There may be one or more variants of the /lib directory on systems which support more than one binary format requiring separate libraries.12

35.3.10.2

Requirements

If one or more of these directories exist, the requirements for their contents are the same as the normal /lib directory, except that /lib<qual>/cpp is not required.13

35.3.11 /mnt : Mount point for a temporarily mounted lesystem

35.3.11.1 Purpose
This directory is provided so that the system administrator may temporarily mount a lesystem as needed. The content of this directory is a local issue and should not affect the manner in which any program is run. This directory must not be used by installation programs: a suitable temporary directory not in use by the system must be used instead.
11 The usual placement of this binary is /usr/lib/gcc-lib/<target>/<version>/cpp. /lib/cpp can either point at this binary, or at any other reference to this binary which exists in the lesystem. (For example, /usr/bin/cpp is also often used.) 12 This is commonly used for 64-bit or 32-bit support on systems which support multiple binary formats, but require libraries of the same name. In this case, /lib32 and /lib64 might be the library directories, and /lib a symlink to one of them. 13 /lib<qual>/cpp is still permitted: this allows the case where /lib and /lib<qual> are the same (one is a symbolic link to the other).

The use of /opt for add-on software is a well-established practice in the U NIX community. The System V Application Binary Interface [AT&T 1990], based on the System V Interface Denition (Third Edition), provides for an /opt structure very similar to the one dened here. The Intel Binary Compatibility Standard v. 2 (iBCS2) also provides a similar structure for /opt. Generally, all data required to support a package on a system must be present within /opt/<package>, including les intended to be copied into /etc/opt/<package> and /var/opt/<package> as well as reserved directories in /opt. The minor restrictions on distributions using /opt are necessary because conicts are possible between distribution-installed and locally-installed software, especially in the case of xed pathnames found in some binary software.
END RATIONALE

359

360

35. The L INUX File System Standard

35.3. The Root Filesystem

35.4. The /usr Hierarchy

35. The L INUX File System Standard

35.3.13
35.3.13.1

/root : Home directory for the root user (optional)

Purpose

The root accounts home directory may be determined by developer or local preference, but this is the recommended default location.14

35.3.14
35.3.14.1

/sbin : System binaries

Purpose

Utilities used for system administration (and other root-only commands) are stored in /sbin, /usr/sbin, and /usr/local/sbin. /sbin contains binaries essential for booting, restoring, recovering, and/or repairing the system in addition to the binaries in /bin.15 Programs executed after /usr is known to be mounted (when there are no problems) are generally placed into /usr/sbin. Locally-installed system administration programs should be placed into /usr/local/sbin.16

fastboot fasthalt fdisk fsck fsck.* getty halt ifconfig init mkfs mkfs.* mkswap reboot route swapon swapoff update

Reboot the system without checking the disks (optional) Stop the system without checking the disks (optional) Partition table manipulator (optional) File system check and repair utility (optional) File system check and repair utility for a specic lesystem (optional) The getty program (optional) Command to stop the system (optional) Congure a network interface (optional) Initial process (optional) Command to build a lesystem (optional) Command to build a specic lesystem (optional) Command to set up a swap area (optional) Command to reboot the system (optional) IP routing table utility (optional) Enable paging and swapping (optional) Disable paging and swapping (optional) Daemon to periodically ush lesystem buffers (optional)

35.3.14.2

Requirements

The following commands, or symbolic links to commands, are required in /sbin. shutdown Command to bring the system down.

35.3.15 /tmp : Temporary les

35.3.15.1 Purpose

The /tmp directory must be made available for programs that require temporary les. Programs must not assume that any les or directories in /tmp are preserved between invocations of the program.
BEGIN RATIONALE

35.3.14.3

Specic Options

The following les, or symbolic links to les, must be in /sbin if the corresponding subsystem is installed:

IEEE standard P1003.2 (POSIX, part 2) makes requirements that are similar to the above section. Although data stored in /tmp may be deleted in a site-specic manner, it is recommended that les and directories located in /tmp be deleted whenever the system is booted. FHS added this recommendation on the basis of historical precedent and common practice, but did not make it a requirement because system administration is not within the scope of this standard.
END RATIONALE

the home directory of the root account is not stored on the root partition it will be necessary to make certain it will default to / if it can not be located. We recommend against using the root account for tasks that can be performed as an unprivileged user, and that it be used solely for system administration. For this reason, we recommend that subdirectories for mail and other applications not appear in the root accounts home directory, and that mail for administration roles such as root, postmaster, and webmaster be forwarded to an appropriate user. 15 Originally, /sbin binaries were kept in /etc. 16 Deciding what things go into "sbin" directories is simple: if a normal (not a system administrator) user will ever run it directly, then it must be placed in one of the "bin" directories. Ordinary users should not have to place any of the sbin directories in their path. For example, les such as chfn which users only occasionally use must still be placed in /usr/bin. ping, although it is absolutely necessary for root (network recovery and diagnosis) is often used by users and must live in /bin for that reason. We recommend that users have read and execute permission for everything in /sbin except, perhaps, certain setuid and setgid programs. The division between /bin and /sbin was not created for security reasons or to prevent users from seeing the operating system, but to provide a good partition between binaries that everyone uses and ones that are primarily used for administration tasks. There is no inherent security advantage in making /sbin off-limits for users.

14 If

35.4

The /usr Hierarchy

35.4.1 Purpose
/usr is the second major section of the lesystem. /usr is shareable, read-only data. That means that /usr should be shareable between various FHS-compliant hosts and must not be written to. Any information that is host-specic or varies with time is stored elsewhere. Large software packages must not use a direct subdirectory under the /usr hierarchy.

361

362

35. The L INUX File System Standard

35.4. The /usr Hierarchy

35. The L INUX File System Standard

35.4.2

Requirements

The following directories, or symbolic links to directories, are required in /usr. /usr Secondary Hierarchy bin include lib local sbin share Most user commands Header les included by C programs Libraries Local hierarchy (empty after main installation) Non-vital system binaries Architecture-independent data

/usr/bin/X11 -> /usr/X11R6/bin /usr/lib/X11 -> /usr/X11R6/lib/X11 /usr/include/X11 -> /usr/X11R6/include/X11 In general, software must not be installed or managed via the above symbolic links. They are intended for utilization by users only. The difculty is related to the release version of the X Window System in transitional periods, it is impossible to know what release of X11 is in use.

35.4.4.2

Specic Options

Host-specic data in /usr/X11R6/lib/X11 should be interpreted as a demonstration le. Applications requiring information about the current host must reference a conguration le in /etc/X11, which may be linked to a le in /usr/X11R6/lib.17

35.4.3
/usr

Specic Options
Secondary Hierarchy X11R6 games lib<qual> src X Window System, version 11 release 6 (optional) Games and educational binaries (optional) Alternate Format Libraries (optional) Source code (optional)

35.4.5 /usr/bin : Most user commands

35.4.5.1 Purpose
This is the primary directory of executable commands on the system.

35.4.5.2

Specic Options

The following directories, or symbolic links to directories, must be in /usr/bin, if the corresponding subsystem is installed: /usr/bin Binaries that are not needed in single-user mode mh Commands for the MH mail handling system (optional)

An exception is made for the X Window System because of considerable precedent and widelyaccepted practice. The following symbolic links to directories may be present. This possibility is based on the need to preserve compatibility with older systems until all implementations can be assumed to use the /var hierarchy. /usr/spool -> /var/spool /usr/tmp -> /var/tmp /usr/spool/locks -> /var/lock Once a system no longer requires any one of the above symbolic links, the link may be removed, if desired.

/usr/bin/X11 must be a symlink to /usr/X11R6/bin if the latter exists. The following les, or symbolic links to les, must be in /usr/bin, if the corresponding subsystem is installed: perl python tclsh wish expect The Practical Extraction and Report Language (optional) The Python interpreted language (optional) Simple shell containing Tcl interpreter (optional) Simple Tcl/Tk windowing shell (optional) Program for interactive dialog (optional)

35.4.4
35.4.4.1

/usr/X11R6 : X Window System, Version 11 Release 6 (optional)

Purpose

BEGIN RATIONALE

This hierarchy is reserved for the X Window System, version 11 release 6, and related les. To simplify matters and make XFree86 more compatible with the X Window System on other systems, the following symbolic links must be present if /usr/X11R6 exists:

Because shell script interpreters (invoked with #!<path> on the rst line of a shell script) cannot rely on a path, it is advantageous to standardize their locations. The Bourne shell and C-shell interpreters are already xed in /bin, but Perl, Python, and Tcl are often found
17 Examples

of such conguration les include Xconfig, XF86Config, or system.twmrc)

363

364

35. The L INUX File System Standard

35.4. The /usr Hierarchy

35. The L INUX File System Standard

in many different places. They may be symlinks to the physical location of the shell interpreters.
END RATIONALE

If /lib/X11 exists, /usr/lib/X11 must be a symbolic link to /lib/X11, or to whatever /lib/X11 is a symbolic link to.21

35.4.6
35.4.6.1

/usr/include : Directory for standard include les.

Purpose

35.4.8 /usr/lib<qual> : Alternate format libraries (optional)

35.4.8.1 Purpose
/usr/lib<qual> performs the same role as /usr/lib for an alternate binary format, except that the symbolic links /usr/lib<qual>/sendmail and /usr/lib<qual>/X11 are not required.22

This is where all of the systems general-use include les for the C programming language should be placed.

35.4.6.2

Specic Options

35.4.9 /usr/local : Local hierarchy

35.4.9.1 Purpose
The /usr/local hierarchy is for use by the system administrator when installing software locally. It needs to be safe from being overwritten when the system software is updated. It may be used for programs and data that are shareable amongst a group of hosts, but not found in /usr. Locally installed software must be placed within /usr/local rather than /usr unless it is being installed to replace or upgrade software in /usr.23

The following directories, or symbolic links to directories, must be in /usr/include, if the corresponding subsystem is installed: /usr/include Include les bsd BSD compatibility include les (optional)

The symbolic link /usr/include/X11 must link to /usr/X11R6/include/X11 if the latter exists.

35.4.7
35.4.7.1

/usr/lib : Libraries for programming and packages

Purpose

35.4.9.2

Requirements

The following directories, or symbolic links to directories, must be in /usr/local /usr/local Local hierarchy bin games include lib man sbin must be a symbolic link to Local binaries Local game binaries Local C header les Local libraries Local online manuals Local system binaries Local architecture-independent hierarchy

/usr/lib includes object les, libraries, and internal binaries that are not intended to be executed directly by users or shell scripts.18 Applications may use a single subdirectory under /usr/lib. If an application uses a subdirectory, all architecture-dependent data exclusively used by the application must be placed within that subdirectory.19

35.4.7.2

Specic Options

For historical reasons, /usr/lib/sendmail /usr/sbin/sendmail if the latter exists.20

18 Miscellaneous architecture-independent application-specic static les and subdirectories must be placed in /usr/share. 19 For example, the perl5 subdirectory for Perl 5 modules and libraries. 20 Some executable commands such as makewhatis and sendmail have also been traditionally placed in /usr/lib. makewhatis is an internal binary and must be placed in a binary directory; users access only catman. Newer sendmail binaries are now placed by default in /usr/sbin. Additionally, systems using a sendmail-compatible mail transfer agent must provide /usr/sbin/sendmail as a symbolic link to the appropriate executable.

21 Host-specic data for the X Window System must not be stored in /usr/lib/X11. Host-specic conguration les such as Xconfig or XF86Config must be stored in /etc/X11. This includes conguration data such as system.twmrc even if it is only made a symbolic link to a more global conguration le (probably in /usr/X11R6/lib/X11). 22 The case where /usr/lib and /usr/lib<qual> are the same (one is a symbolic link to the other) these les and the per-application subdirectories will exist. 23 Software placed in / or /usr may be overwritten by system upgrades (though we recommend that distributions do not overwrite data in /etc under these circumstances). For this reason, local software must not be placed outside of /usr/local without good reason.

365

366

35. The L INUX File System Standard

35.4. The /usr Hierarchy

35. The L INUX File System Standard

src

Local source code

35.4.11.3 Specic Options

The following directories, or symbolic links to directories, must be in /usr/share, if the corresponding subsystem is installed: /usr/share Architecture-independent data dict doc games info locale nls sgml terminfo tmac zoneinfo Word lists (optional) Miscellaneous documentation (optional) Static data les for /usr/games (optional) GNU Info systems primary directory (optional) Locale information (optional) Message catalogs for Native language support (optional) SGML and XML data (optional) Directories for terminfo database (optional) troff macros not distributed with groff (optional) Timezone information and conguration (optional)

No other directories, except those listed below, may be in /usr/local after rst installing a FHS-compliant system.

35.4.9.3

Specic Options

If directories /lib<qual> or /usr/lib<qual> exist, the equivalent directories must also exist in /usr/local.

35.4.10
35.4.10.1

/usr/sbin : Non-essential standard system binaries

Purpose

This directory contains any non-essential binaries used exclusively by the system administrator. System administration programs that are required for system repair, system recovery, mounting /usr, or other essential functions must be placed in /sbin instead.24

35.4.11 /usr/share : Architecture-independent data

35.4.11.1 Purpose
The /usr/share hierarchy is for all read-only architecture independent data les.
25

This hierarchy is intended to be shareable among all architecture platforms of a given OS; thus, for example, a site with i386, Alpha, and PPC platforms might maintain a single /usr/share directory that is centrally-mounted. Note, however, that /usr/share is generally not intended to be shared by different OSes or by different releases of the same OS. Any program or package which contains or requires data that doesnt need to be modied should store that data in /usr/share (or /usr/local/share, if installed locally). It is recommended that a subdirectory be used in /usr/share for this purpose. Game data stored in /usr/share/games must be purely static data. Any modiable les, such as score les, game play logs, and so forth, should be placed in /var/games.

It is recommended that application-specic, architecture-independent directories be placed here. Such directories include groff, perl, ghostscript, texmf, and kbd (Linux) or syscons (BSD). They may, however, be placed in /usr/lib for backwards compatibility, at the distributors discretion. Similarly, a /usr/lib/games hierarchy may be used in addition to the /usr/share/games hierarchy if the distributor wishes to place some game data there.

35.4.11.4 /usr/share/dict : Word lists (optional) 35.4.11.4.1 Purpose

This directory is the home for word lists on the system; Traditionally this directory contains only the English words le, which is used by look(1) and various spelling programs. words may use either American or British spelling.
BEGIN RATIONALE

35.4.11.2 Requirements
The following directories, or symbolic links to directories, must be in /usr/share /usr/share Architecture-independent data man misc Online manuals Miscellaneous architecture-independent data

The reason that only word lists are located here is that they are the only les common to all spell checkers.
END RATIONALE

35.4.11.4.2 Specic Options

The following les, or symbolic links to les, must be in /usr/share/dict, if the corresponding subsystem is installed: words List of English words (optional)

24 Locally

installed system administration programs should be placed in /usr/local/sbin. 25 Much of this data originally lived in /usr (man, doc) or /usr/lib (dict, terminfo, zoneinfo).

367

368

35. The L INUX File System Standard

35.4. The /usr Hierarchy

35. The L INUX File System Standard

Sites that require both American and British spelling may link words /usr/share/dict/american-english or /usr/share/dict/british-english.

Word lists for other languages may be added using the English name for that language, e.g., /usr/share/dict/french, /usr/share/dict/danish, etc. These should, if possible, use an ISO 8859 character set which is appropriate for the language in question; if possible the Latin1 (ISO 8859-1) character set should be used (this is often not possible). Other word lists must be included here, if present.

man8: System administration Programs used by system administrators for system operation and maintenance are documented here. Some of these programs are also occasionally useful for normal users.

35.4.11.5.2 Specic Options

The following directories, or symbolic links to directories, /usr/share/<mandir>/<locale>, unless they are empty:28 <mandir>/<locale> A manual page hierarchy man1 User programs (optional) System calls (optional) Library calls (optional) Special les (optional) File formats (optional) Games (optional) Miscellaneous (optional) System administration (optional) must be in

35.4.11.5 /usr/share/man : Manual pages 35.4.11.5.1 Purpose

This section details the organization for manual pages throughout the system, including /usr/share/man. Also refer to the section on /var/cache/man. The primary <mandir> of the system is /usr/share/man. /usr/share/man contains manual information for commands and data under the / and /usr lesystems.26 Manual pages are stored in <mandir>/<locale>/man<section>/<arch>. An explanation of <mandir>, <locale>, <section>, and <arch> is given below. A description of each section follows: man1: User programs Manual pages that describe publicly accessible commands are contained in this chapter. Most program documentation that a user will need to use is located here. man2: System calls This section describes all of the system calls (requests for the kernel to perform operations). man3: Library functions and subroutines Section 3 describes program library routines that are not direct calls to kernel services. This and chapter 2 are only really of interest to programmers. man4: Special les Section 4 describes the special les, related driver functions, and networking support available in the system. Typically, this includes the device les found in /dev and the kernel interface to networking protocol support. man5: File formats The formats for many data les are documented in the section 5. This includes various include les, program output les, and system les. man6: Games This chapter documents games, demos, and generally trivial programs. Different people have various notions about how essential this is. man7: Miscellaneous Manual pages that are difcult to classify are designated as being section 7. The troff and other text processing macro packages are found here.
26 Obviously, there are no manual pages in / because they are not required at boot time nor are they required in emergencies.27 27 Really.

man2 man3 man4 man5 man6 man7 man8

The component <section> describes the manual section. Provisions must be made in the structure of /usr/share/man to support manual pages which are written in different (or multiple) languages. These provisions must take into account the storage and reference of these manual pages. Relevant factors include language (including geographical-based differences), and character code set. This naming of language subdirectories of /usr/share/man is based on Appendix E of the POSIX 1003.1 standard which describes the locale identication string the most well-accepted method to describe a cultural environment. The <locale> string is: <language>[ <territory>][.<character-set>][,<version>] The <language> eld must be taken from ISO 639 (a code for the representation of names of languages). It must be two characters wide and specied with lowercase letters only. The <territory> eld must be the two-letter code of ISO 3166 (a specication of representations of countries), if possible. (Most people are familiar with the two-letter codes used for the country codes in email addresses.29 ) It must be two characters wide and specied with uppercase letters only. The <character-set> eld must represent the standard describing the character set. If the <character-set> eld is just a numeric specication, the number represents the number of
28 For example, if /usr/local/man has no manual pages in section 4 (Devices), then /usr/local/man/man4 may be omitted. 29 A major exception to this rule is the United Kingdom, which is GB in the ISO 3166, but UK for most email addresses.

369

370

35. The L INUX File System Standard

35.4. The /usr Hierarchy

35. The L INUX File System Standard

the international standard describing the character set. It is recommended that this be a numeric representation if possible (ISO standards, especially), not include additional punctuation symbols, and that any letters be in lowercase. A parameter specifying a <version> of the prole may be placed after the <character-set> eld, delimited by a comma. This may be used to discriminate between different cultural needs; for instance, dictionary order versus a more systems-oriented collating order. This standard recommends not using the <version> eld, unless it is necessary. Systems which use a unique language and code set for all manual pages may omit the <locale> substring and store all manual pages in <mandir>. For example, systems which only have English manual pages coded with ASCII, may store manual pages (the man<section> directories) directly in /usr/share/man. (That is the traditional circumstance and arrangement, in fact.) Countries for which there is a well-accepted standard character code set may omit the <character-set> eld, but it is strongly recommended that it be included, especially for countries with several competing standards. Various examples: Language English English English French French German German German German Japanese Japanese Japanese Territory United Kingdom United States Canada France Germany Germany Germany Switzerland Japan Japan Japan Character Set ASCII ASCII ASCII ISO 8859-1 ISO 8859-1 ISO 646 ISO 6937 ISO 8859-1 ISO 646 JIS SJIS UJIS (or EUC-J) Directory /usr/share/man/en /usr/share/man/en /usr/share/man/en /usr/share/man/fr /usr/share/man/fr /usr/share/man/de /usr/share/man/de /usr/share/man/de /usr/share/man/de /usr/share/man/ja /usr/share/man/ja /usr/share/man/ja

In addition, some large sets of application-specic manual pages have an additional sufx appended to the manual page lename. For example, the MH mail handling system manual pages must have mh appended to all MH manuals. All X Window System manual pages must have an x appended to the lename. The practice of placing various language manual pages in appropriate subdirectories of /usr/share/man also applies to the other manual page hierarchies, such as /usr/local/man and /usr/X11R6/man. (This portion of the standard also applies later in the section on the optional /var/cache/man structure.)

35.4.11.6 /usr/share/misc : Miscellaneous architecture-independent data

This directory contains miscellaneous architecture-independent les which dont require a separate subdirectory under /usr/share.

35.4.11.6.1 Specic Options

The following les, or symbolic links to les, must be in /usr/share/misc, if the corresponding subsystem is installed: ascii magic termcap termcap.db ASCII character set table (optional) Default list of magic numbers for the le command (optional) Terminal capability database (optional) Terminal capability database (optional)

GB US CA FR DE.646 DE.6937 DE.88591 CH.646 JP.jis JP.sjis JP.ujis

Other (application-specic) les may appear here,30 but a distributor may place them in /usr/lib at their discretion.

35.4.11.7 /usr/share/sgml : SGML and XML data (optional) 35.4.11.7.1 Purpose

/usr/share/sgml contains architecture-independent les used by SGML or XML applications, such as ordinary catalogs (not the centralized ones, see /etc/sgml), DTDs, entities, or style sheets.

Similarly, provision must be made for manual pages which are architecture-dependent, such as documentation on device-drivers or low-level system administration commands. These must be placed under an <arch> directory in the appropriate man<section> directory; for example, a man page for the i386 ctrlaltdel(8) command might be placed in /usr/share/man/<locale>/man8/i386/ctrlaltdel.8. Manual pages for commands and data under /usr/local are stored in /usr/local/man. Manual pages for X11R6 are stored in /usr/X11R6/man. It follows that all manual page hierarchies in the system must have the same structure as /usr/share/man. The cat page sections (cat<section>) containing formatted manual page entries are also found within subdirectories of <mandir>/<locale>, but are not required nor may they be distributed in lieu of nroff source manual pages. The numbered sections 1 through 8 are traditionally dened. In general, the le name for manual pages located within a particular section end with .<section>.

35.4.11.7.2 Specic Options

The following directories, or symbolic links to directories, must be in /usr/share/sgml, if the corresponding subsystem is installed:

30 Some

such les include:

{ airport, birthtoken, eqnchar, getopt, gprof.callg, gprof.flat, inter.phone, ipfw.samp.filters, ipfw.samp.scripts, keycap.pcvt, mail.help, mail.tildehelp, man.template, map3270, mdoc.template, more.help, na.phone, nslookup.help, operator, scsi modes, sendmail.hf, style, units.lib, vgrindefs, vgrindefs.db, zipcodes }

371

372

35. The L INUX File System Standard

35.5. The /var Hierarchy

35. The L INUX File System Standard

/usr/share/sgml

SGML and XML data docbook tei html mathml docbook DTD (optional) tei DTD (optional) html DTD (optional) mathml DTD (optional)

local lock log opt run spool tmp

Variable data for /usr/local Lock les Log les and directories Variable data for /opt Data relevant to running processes Application spool data Temporary les preserved between system reboots

Other les that are not specic to a given DTD may reside in their own subdirectory.

35.4.12
35.4.12.1

/usr/src : Source code (optional)

Purpose

Several directories are reserved in the sense that they must not be used arbitrarily by some new application, since they would conict with historical and/or local practice. They are: /var/backups /var/cron /var/msgs /var/preserve

Any non-local source code should be placed in this subdirectory.

35.5 The /var Hierarchy

35.5.1 Purpose
/var contains variable data les. This includes spool directories and les, administrative and logging data, and transient and temporary les. Some portions of /var are not shareable between different systems. For instance, /var/log, /var/lock, and /var/run. Other portions may be shared, notably /var/mail, /var/cache/man, /var/cache/fonts, and /var/spool/news. /var is specied here in order to make it possible to mount /usr read-only. Everything that once went into /usr that is written to during system operation (as opposed to installation and software maintenance) must be in /var. If /var cannot be made a separate partition, it is often preferable to move /var out of the root partition and into the /usr partition. (This is sometimes done to reduce the size of the root partition or when space runs low in the root partition.) However, /var must not be linked to /usr because this makes separation of /usr and /var more difcult and is likely to create a naming conict. Instead, link /var to /usr/var. Applications must generally not add directories to the top level of /var. Such directories should only be added if they have some system-wide implication, and in consultation with the FHS mailing list.

35.5.3 Specic Options

The following directories, or symbolic links to directories, must be in /var, if the corresponding subsystem is installed: /var Variable data account crash games mail yp Process accounting logs (optional) System crash dumps (optional) Variable game data (optional) User mailbox les (optional) Network Information Service (NIS) database les (optional)

35.5.4 /var/account : Process accounting logs (optional)

35.5.4.1 Purpose
This directory holds the current active process accounting log and the composite process usage data (as used in some U NIX-like systems by lastcomm and sa).

35.5.2

Requirements

The following directories, or symbolic links to directories, are required in /var. /var Variable data cache lib Application cache data Variable state information

35.5.5 /var/cache : Application cache data

35.5.5.1 Purpose
/var/cache is intended for cached data from applications. Such data is locally generated as a result of time-consuming I/O or calculation. The application must be able to regenerate or

373

374

35. The L INUX File System Standard

35.5. The /var Hierarchy

35. The L INUX File System Standard

restore the data. Unlike /var/spool, the cached les can be deleted without data loss. The data must remain valid between invocations of the application and rebooting the system. Files located under /var/cache may be expired in an application specic manner, by the system administrator, or both. The application must always be able to recover from manual deletion of these les (generally because of a disk space shortage). No other requirements are made on the data format of the cache directories.
BEGIN RATIONALE

Allow no caching of formatted man pages, and require formatting to be done each time a man page is brought up. Allow local caching of formatted man pages in /var/cache/man. The structure of /var/cache/man needs to reect both the fact of multiple man page hierarchies and the possibility of multiple language support. Given an unformatted manual page that normally appears in <path>/man/<locale>/man<section>, the directory to place formatted man pages in is /var/cache/man/<catpath>/<locale>/cat<section>, where <catpath> is derived from <path> by removing any leading usr and/or trailing share pathname components.32 (Note that the <locale> component may be missing.) Man pages written to /var/cache/man may eventually be transferred to the appropriate preformatted directories in the source man hierarchy or expired; likewise formatted man pages in the source man hierarchy may be expired if they are not accessed for a period of time.

The existence of a separate directory for cached data allows system administrators to set different disk and backup policies from other directories in /var.
END RATIONALE

35.5.5.2

Specic Options
Cache directories fonts man www <package> Locally-generated fonts (optional) Locally-formatted manual pages (optional) WWW proxy or cache data (optional) Package specic cache data (optional)

/var/cache

If preformatted manual pages come with a system on read-only media (a CD-ROM, for instance), they must be installed in the source man hierarchy (e.g. /usr/share/man/cat<section>). /var/cache/man is reserved as a writable cache for formatted manual pages.
BEGIN RATIONALE

35.5.5.3 35.5.5.3.1

/var/cache/fonts : Locally-generated fonts (optional) Purpose

Release 1.2 of the standard specied /var/catman for this hierarchy. The path has been moved under /var/cache to better reect the dynamic nature of the formatted man pages. The directory name has been changed to man to allow for enhancing the hierarchy to include post-processed formats other than cat, such as PostScript, HTML, or DVI.
END RATIONALE

The directory /var/cache/fonts should be used to store any dynamically-created fonts. In particular, all of the fonts which are automatically generated by mktexpk must be located in appropriately-named subdirectories of /var/cache/fonts.31

35.5.6 /var/crash : System crash dumps (optional)

35.5.6.1 Purpose

35.5.5.3.2

Specic Options

Other dynamically created fonts may also be placed in this tree, under appropriately-named subdirectories of /var/cache/fonts.

This directory holds system crash dumps. As of the date of this release of the standard, system crash dumps were not supported under Linux.

35.5.5.4 35.5.5.4.1

/var/cache/man : Locally-formatted manual pages (optional) Purpose

35.5.7 /var/games : Variable game data (optional)

35.5.7.1 Purpose

This directory provides a standard location for sites that provide a read-only /usr partition, but wish to allow caching of locally-formatted man pages. Sites that mount /usr as writable (e.g., single-user installations) may choose not to use /var/cache/man and may write formatted man pages into the cat<section> directories in /usr/share/man directly. We recommend that most sites use one of the following options instead: Preformat all manual pages alongside the unformatted versions.
standard does not currently incorporate the TEX Directory Structure (a document that describes the layout TEX les and directories), but it may be useful reading. It is located at ftp://ctan.tug.org/tex/.
31 This

Any variable data relating to games in /usr should be placed here. /var/games should hold the variable data previously found in /usr; static data, such as help text, level descriptions, and so on, must remain elsewhere, such as /usr/share/games.
BEGIN RATIONALE
32 For example, /usr/share/man/man1/ls.1 is formatted into /var/cache/man/cat1/ls.1, and /usr/X11R6/man/<locale>/man3/XtClass.3x into /var/cache/man/X11R6/<locale>/cat3/XtClass.3x.

375

376

35. The L INUX File System Standard

35.5. The /var Hierarchy

35. The L INUX File System Standard

/var/games has been given a hierarchy of its own, rather than leaving it merged in with the old /var/lib as in release 1.2. The separation allows local control of backup strategies, permissions, and disk usage, as well as allowing inter-host sharing and reducing clutter in /var/lib. Additionally, /var/games is the path traditionally used by BSD.
END RATIONALE

hwclock xdm

State directory for hwclock (optional) X display manager variable data (optional)

35.5.8
35.5.8.1

/var/lib : Variable state information

Purpose

35.5.8.4

/var/lib/<editor> : Editor backup les and state (optional)

35.5.8.4.1 Purpose
These directories contain saved les generated by any unexpected termination of an editor (e.g., elvis, jove, nvi). Other editors may not require a directory for crash-recovery les, but may require a welldened place to store other information while the editor is running. This information should be stored in a subdirectory under /var/lib (for example, GNU Emacs would place lock les in /var/lib/emacs/lock). Future editors may require additional state information beyond crash-recovery les and lock les this information should also be placed under /var/lib/<editor>.
BEGIN RATIONALE

This hierarchy holds state information pertaining to an application or the system. State information is data that programs modify while they run, and that pertains to one specic host. Users must never need to modify les in /var/lib to congure a packages operation. State information is generally used to preserve the condition of an application (or a group of inter-related applications) between invocations and between different instances of the same application. State information should generally remain valid after a reboot, should not be logging output, and should not be spooled data. An application (or a group of inter-related applications) must use a subdirectory of /var/lib for its data.33 There is one required subdirectory, /var/lib/misc, which is intended for state les that dont need a subdirectory; the other subdirectories should only be present if the application in question is included in the distribution. /var/lib/<name> is the location that must be used for all distribution packaging support. Different distributions may use different names, of course.

Previous Linux releases, as well as all commercial vendors, use /var/preserve for vi or its clones. However, each editor uses its own format for these crash-recovery les, so a separate directory is needed for each editor. Editor-specic lock les are usually quite different from the device or resource lock les that are stored in /var/lock and, hence, are stored under /var/lib.
END RATIONALE

35.5.8.2

Requirements 35.5.8.5 /var/lib/hwclock : State directory for hwclock (optional)

The following directories, or symbolic links to directories, are required in /var/lib: /var/lib Variable state information misc Miscellaneous state data

35.5.8.5.1 Purpose
This directory contains the le /var/lib/hwclock/adjtime.
BEGIN RATIONALE

35.5.8.3

Specic Options

In FHS 2.1, this le was /etc/adjtime, but as hwclock updates it, that was obviously incorrect.
END RATIONALE

The following directories, or symbolic links to directories, must be in /var/lib, if the corresponding subsystem is installed: /var/lib Variable state information <editor> <pkgtool> <package> Editor backup les and state (optional) Packaging support les (optional) State data for packages and subsystems (optional)

35.5.8.6

/var/lib/misc : Miscellaneous variable data

35.5.8.6.1 Purpose
This directory contains variable data not placed in a subdirectory in /var/lib. An attempt should be made to use relatively unique names in this directory to avoid namespace conicts.34
34 This hierarchy should contain les stored in /var/db in current BSD releases. These include locate.database and mountdtab, and the kernel symbol database(s).

33 An important difference between this version of this standard and previous ones is that applications are now required to use a subdirectory of /var/lib.

377

378

35. The L INUX File System Standard

35.5. The /var Hierarchy

35. The L INUX File System Standard

35.5.9
35.5.9.1

/var/lock : Lock les

Purpose

Lock les should be stored within the /var/lock directory structure. Lock les for devices and other resources shared by multiple applications, such as the serial device lock les that were originally found in either /usr/spool/locks or /usr/spool/uucp, must now be stored in /var/lock. The naming convention which must be used is LCK.. followed by the base name of the device le. For example, to lock /dev/ttyS0 the le LCK..ttyS0 would be created.
35

The logical location for this directory was changed from /var/spool/mail in order to bring FHS in-line with nearly every U NIX implementation. This change is important for inter-operability since a single /var/mail is often shared between multiple hosts and multiple U NIX implementations (despite NFS locking issues). It is important to note that there is no requirement to physically move the mail spool to this location. However, programs and header les must be changed to use /var/mail.
END RATIONALE

35.5.12 /var/opt : Variable data for /opt

35.5.12.1 Purpose
Variable data of the packages in /opt must be installed in /var/opt/<package>, where <package> is the name of the subtree in /opt where the static data from an add-on software package is stored, except where superseded by another le in /etc. No structure is imposed on the internal arrangement of /var/opt/<package>.
BEGIN RATIONALE

The format used for the contents of such lock les must be the HDB UUCP lock le format. The HDB format is to store the process identier (PID) as a ten byte ASCII decimal number, with a trailing newline. For example, if process 1230 holds a lock le, it would contain the eleven characters: space, space, space, space, space, space, one, two, three, zero, and newline.

35.5.10
35.5.10.1

/var/log : Log les and directories

Purpose

Refer to the rationale for /opt.

END RATIONALE

This directory contains miscellaneous log les. Most logs must be written to this directory or an appropriate subdirectory.

35.5.13 /var/run : Run-time variable data

35.5.13.1 Purpose
This directory contains system information data describing the system since it was booted. Files under this directory must be cleared (removed or truncated as appropriate) at the beginning of the boot process. Programs may have a subdirectory of /var/run; this is encouraged for programs that use more than one run-time le.37 Process identier (PID) les, which were originally placed in /etc, must be placed in /var/run. The naming convention for PID les is <program-name>.pid. For example, the crond PID le is named /var/run/crond.pid.

35.5.10.2

Specic Options

The following les, or symbolic links to les, must be in /var/log, if the corresponding subsystem is installed: lastlog messages wtmp record of last login of each user system messages from syslogd record of all logins and logouts

35.5.13.2

Requirements

35.5.11 /var/mail : User mailbox les (optional)

35.5.11.1 Purpose
The mail spool must be accessible through /var/mail and the mail spool les must take the form <username>.36 User mailbox les in this location must be stored in the standard U NIX mailbox format.
BEGIN RATIONALE
35 Then, anything wishing to use /dev/ttyS0 can read the lock le and act accordingly (all locks in /var/lock should be world-readable). 36 Note that /var/mail may be a symbolic link to another directory.

The internal format of PID les remains unchanged. The le must consist of the process identier in ASCII-encoded decimal, followed by a newline character. For example, if crond was process number 25, /var/run/crond.pid would contain three characters: two, ve, and newline. Programs that read PID les should be somewhat exible in what they accept; i.e., they should ignore extra whitespace, leading zeroes, absence of the trailing newline, or additional lines in the PID le. Programs that create PID les should use the simple specication located in the above paragraph. The utmp le, which stores information about who is currently using the system, is located in this directory. Programs that maintain transient U NIX-domain sockets must place them in this directory.
37 /var/run should be unwritable for unprivileged users (root or users running daemons); it is a major security problem if any user can write in this directory.

379

380

35. The L INUX File System Standard

35.5. The /var Hierarchy

35.6. Operating System Specic Annex

35. The L INUX File System Standard

35.5.14
35.5.14.1

/var/spool : Application spool data

Purpose

Some BSD releases use /var/rwho for this data; given its historical location in /var/spool on other systems and its approximate t to the denition of spooled data, this location was deemed more appropriate.
END RATIONALE

/var/spool contains data which is awaiting some kind of later processing. Data in /var/spool represents work to be done in the future (by a program, user, or administrator); often data is deleted after it has been processed.38

35.5.15 /var/tmp : Temporary les preserved between system reboots

35.5.15.1 Purpose
The /var/tmp directory is made available for programs that require temporary les or directories that are preserved between system reboots. Therefore, data stored in /var/tmp is more persistent than data in /tmp. Files and directories located in /var/tmp must not be deleted when the system is booted. Although data stored in /var/tmp is typically deleted in a site-specic manner, it is recommended that deletions occur at a less frequent interval than /tmp.

35.5.14.2

Specic Options

The following directories, or symbolic links to directories, must be in /var/spool, if the corresponding subsystem is installed: /var/spool Spool directories lpd mqueue news rwho uucp Printer spool directory (optional) Outgoing mail queue (optional) News spool directory (optional) Rwhod les (optional) Spool directory for UUCP (optional)

35.5.16 /var/yp : Network Information Service (NIS) database les (optional)

35.5.16.1 Purpose
Variable data for the Network Information Service (NIS), formerly known as the Sun Yellow Pages (YP), must be placed in this directory.
BEGIN RATIONALE

35.5.14.3 35.5.14.3.1

/var/spool/lpd : Line-printer daemon print queues (optional) Purpose

The lock le for lpd, lpd.lock, must be placed in /var/spool/lpd. It is suggested that the lock le for each printer be placed in the spool directory for that specic printer and named lock.

/var/yp is the standard directory for NIS (YP) data and is almost exclusively used in NIS documentation and systems.39
END RATIONALE

35.5.14.3.2

Specic Options
Printer spool directory <printer> Spools for a specic printer (optional)

35.6

Operating System Specic Annex

/var/spool/lpd

This section is for additional requirements and recommendations that only apply to a specic operating system. The material in this section should never conict with the base standard.

35.6.1 Linux
35.5.14.4 35.5.14.4.1 /var/spool/rwho : Rwhod les (optional) Purpose
This is the annex for the Linux operating system.

This directory holds the rwhod information for other systems on the local net.
BEGIN RATIONALE
38 UUCP

35.6.1.1

/ : Root directory

On Linux systems, if the kernel is located in /, we recommend using the names vmlinux or vmlinuz, which have been used in recent Linux kernel source packages.
39 NIS

lock les must be placed in /var/lock. See the above section on /var/lock.

should not be confused with Sun NIS+, which uses a different directory, /var/nis.

381

382

35. The L INUX File System Standard

35.6. Operating System Specic Annex

35. The L INUX File System Standard

35.6.1.2

/bin : Essential user command binaries (for use by all users)

Optional les for /sbin:

Static binaries: { ldconfig, sln, ssync } Static ln (sln) and static sync (ssync) are useful when things go wrong. The primary use of sln (to repair incorrect symlinks in /lib after a poorly orchestrated upgrade) is no longer a major concern now that the ldconfig program (usually located in /usr/sbin) exists and can act as a guiding hand in upgrading the dynamic libraries. Static sync is useful in some emergency situations. Note that these need not be statically linked versions of the standard ln and sync, but may be. The ldconfig binary is optional for /sbin since a site may choose to run ldconfig at boot time, rather than only when upgrading the shared libraries. (Its not clear whether or not it is advantageous to run ldconfig on each boot.) Even so, some people like ldconfig around for the following (all too common) situation: 1. Ive just removed /lib/<file>. 2. I cant nd out the name of the library because ls is dynamically linked, Im using a shell that doesnt have ls built-in, and I dont know about using echo * as a replacement. 3. I have a static sln, but I dont know what to call the link. Miscellaneous: { ctrlaltdel, kbdrate } So as to cope with the fact that some keyboards come up with such a high repeat rate as to be unusable, kbdrate may be installed in /sbin on some systems. Since the default action in the kernel for the Ctrl-Alt-Del key combination is an instant hard reboot, it is generally advisable to disable the behavior before mounting the root lesystem in read-write mode. Some init suites are able to disable Ctrl-Alt-Del, but others may require the ctrlaltdel program, which may be installed in /sbin on those systems.

Linux systems which require them place these additional les into /bin. { setserial }

35.6.1.3

/dev : Devices and special les

All devices and special les in /dev should adhere to the Linux Allocated Devices document, which is available with the Linux kernel source. It is maintained by H. Peter Anvin <hpa@zytor.com>. Symbolic links in /dev should not be distributed with Linux systems except as provided in the Linux Allocated Devices document.
BEGIN RATIONALE

The requirement not to make symlinks promiscuously is made because local setups will often differ from that on the distributors development machine. Also, if a distribution install script congures the symbolic links at install time, these symlinks will often not get updated if local changes are made in hardware. When used responsibly at a local level, however, they can be put to good use.
END RATIONALE

35.6.1.4

/etc : Host-specic system conguration

Linux systems which require them place these additional les into /etc. { lilo.conf }

35.6.1.5

/proc : Kernel and process information virtual lesystem

35.6.1.7

/usr/include : Header les included by C programs

The proc lesystem is the de-facto standard Linux method for handling process and system information, rather than /dev/kmem and other similar methods. We strongly encourage this for the storage and retrieval of process information as well as other kernel and memory information.

These symbolic links are required if a C or C++ compiler is installed and only for systems not based on glibc. /usr/include/asm -> /usr/src/linux/include/asm-<arch> /usr/include/linux -> /usr/src/linux/include/linux

35.6.1.6

/sbin : Essential system binaries

Linux systems place these additional les into /sbin. Second extended lesystem commands (optional): { badblocks, dumpe2fs, e2fsck, mke2fs, mklost+found, tune2fs } Boot-loader map installer (optional): { lilo }

35.6.1.8

/usr/src : Source code

For systems based on glibc, there are no specic guidelines for this directory. For systems based on Linux libc revisions prior to glibc, the following guidelines and rationale apply: The only source code that should be placed in a specic location is the Linux kernel source code. It is located in /usr/src/linux. If a C or C++ compiler is installed, but the complete Linux kernel source code is not installed, then the include les from the kernel source code must be located in these directories:

383

384

35. The L INUX File System Standard

35.6. Operating System Specic Annex

35.7. Appendix

35. The L INUX File System Standard

/usr/src/linux/include/asm-<arch> /usr/src/linux/include/linux <arch> is the name of the system architecture. Note: /usr/src/linux may be a symbolic link to a kernel source code tree.
BEGIN RATIONALE

35.7

Appendix

35.7.1 The FHS mailing list

The FHS mailing list is located at <fhs-discuss@ucsd.edu>. To subscribe to the list send mail to <listserv@ucsd.edu> with body ADD fhs-discuss. Thanks to Network Operations at the University of California at San Diego who allowed us to use their excellent mailing list server. As noted in the introduction, please do not send mail to the mailing list without rst contacting the FHS editor or a listed contributor.

It is important that the kernel include les be located in /usr/src/linux and not in /usr/include so there are no problems when system administrators upgrade their kernel version for the rst time.
END RATIONALE

35.6.1.9

/var/spool/cron : cron and at jobs

35.7.2 Background of the FHS

The process of developing a standard lesystem hierarchy began in August 1993 with an effort to restructure the le and directory structure of Linux. The FSSTND, a lesystem hierarchy standard specic to the Linux operating system, was released on February 14, 1994. Subsequent revisions were released on October 9, 1994 and March 28, 1995. In early 1995, the goal of developing a more comprehensive version of FSSTND to address not only Linux, but other U NIX-like systems was adopted with the help of members of the BSD development community. As a result, a concerted effort was made to focus on issues that were general to U NIX-like systems. In recognition of this widening of scope, the name of the standard was changed to Filesystem Hierarchy Standard or FHS for short. Volunteers who have contributed extensively to this standard are listed at the end of this document. This standard represents a consensus view of those and other contributors.

This directory contains the variable data for the cron and at programs.

35.7.3 General Guidelines

Here are some of the guidelines that have been used in the development of this standard: Solve technical problems while limiting transitional difculties. Make the specication reasonably stable. Gain the approval of distributors, developers, and other decision-makers in relevant development groups and encourage their participation. Provide a standard that is attractive to the implementors of different U NIX-like systems.

35.7.4 Scope
This document species a standard lesystem hierarchy for FHS lesystems by specifying the location of les and directories, and the contents of some system les. This standard has been designed to be used by system integrators, package developers, and system administrators in the construction and maintenance of FHS compliant lesystems. It is primarily intended to be a reference and is not a tutorial on how to manage a conforming lesystem hierarchy.

385

386

35. The L INUX File System Standard

35.7. Appendix

35. The L INUX File System Standard

The FHS grew out of earlier work on FSSTND, a lesystem organization standard for the Linux operating system. It builds on FSSTND to address interoperability issues not just in the Linux community but in a wider arena including 4.4BSD-based operating systems. It incorporates lessons learned in the BSD world and elsewhere about multi-architecture support and the demands of heterogeneous networking. Although this standard is more comprehensive than previous attempts at lesystem hierarchy standardization, periodic updates may become necessary as requirements change in relation to emerging technology. It is also possible that better solutions to the problems addressed here will be discovered so that our solutions will no longer be the best possible solutions. Supplementary drafts may be released in addition to periodic updates to this document. However, a specic goal is backwards compatibility from one release of this document to the next. Comments related to this standard are welcome. Any comments or suggestions for changes may be directed to the FHS editor (Daniel Quinlan <quinlan@pathname.com>) or the FHS mailing list. Typographical or grammatical comments should be directed to the FHS editor. Before sending mail to the mailing list it is requested that you rst contact the FHS editor in order to avoid excessive re-discussion of old topics. Questions about how to interpret items in this document may occasionally arise. If you have need for a clarication, please contact the FHS editor. Since this standard represents a consensus of many participants, it is important to make certain that any interpretation also represents their collective opinion. For this reason it may not be possible to provide an immediate response unless the inquiry has been the subject of previous discussion.

Eric S. Raymond Rusty Russell Mike Sangrey David H. Silber Thomas Sippel-Dau Theodore Tso Stephen Tweedie Fred N. van Kempen Bernd Warken

<esr@thyrsus.com> <rusty@rustcorp.com.au> <mike@sojurn.lns.pa.us> <dhs@glowworm.firefly.com> <t.sippel-dau@ic.ac.uk> <tytso@athena.mit.edu> <sct@dcs.ed.ac.uk> <waltje@infomagic.com> <bwarken@mayn.de>

35.7.5

Acknowledgments

The developers of the FHS wish to thank the developers, system administrators, and users whose input was essential to this standard. We wish to thank each of the contributors who helped to write, compile, and compose this standard. The FHS Group also wishes to thank those Linux developers who supported the FSSTND, the predecessor to this standard. If they hadnt demonstrated that the FSSTND was benecial, the FHS could never have evolved.

35.7.6

Contributors
<bsa@kf8nh.wariat.org> <bostic@cs.berkeley.edu> <drew@colorado.edu> <faith@cs.unc.edu> <sweh@spuddy.mew.co.uk> <ijackson@cus.cam.ac.uk> <jmartin@acm.org> <ian@ucsd.edu> <metcalf@lcs.mit.edu> <imurdock@debian.org> <niemidc@clark.net> <quinlan@pathname.com>

Brandon S. Allbery Keith Bostic Drew Eckhardt Rik Faith Stephen Harris Ian Jackson John A. Martin Ian McCloghrie Chris Metcalf Ian Murdock David C. Niemi Daniel Quinlan

387

388

36.1. Web Server Basics

36. httpd Apache Web Server

You will also have to create a log le (/usr/local/var/log/sh-httpd.log) and at least one web page (/usr/local/var/sh-www/index.html) for your server to serve. It can contain, say:
<HTML> <HEAD> <TITLE>My First Document</TITLE> </HEAD> <BODY bgcolor=#CCCCCC text="#000000"> This is my first document<P> Please visit <A HREF="http://rute.sourceforge.net/"> The Rute Home Page </A> for more info.</P> </BODY> </HTML>

Chapter 36

httpd Apache Web Server

In this chapter, we will show how to set up a web server running virtual domains and dynamic CGI web pages. HTML is not covered, and you are expected to have some understanding of what HTML is, or at least where to nd documentation about it.

36.1 Web Server Basics

In Section 26.2 we showed a simple HTTP session with the telnet command. A web server is really nothing more than a program that reads a le from the hard disk whenever a GET /<filename>.html HTTP/1.0 request comes in on port 80. Here, we will show a simple web server written in shell script. Not by me. The author did not put his name in the source, so if you are out there, please drop me an email. You will need to add the line

Note that the server runs as nobody, so the log le must be writable by the nobody user, and the index.html le must be readable. Also note the use of the getpeername command, which can be changed to PEER="" if you do not have the netpipes package installed. I am not completely sure if other commands used here are unavailable on other U NIX systems. .

#!/bin/sh VERSION=0.1 NAME="ShellHTTPD" DEFCONTENT="text/html" DOCROOT=/usr/local/var/sh-www DEFINDEX=index.html LOGFILE=/usr/local/var/log/sh-httpd.log log() { local local local local

to your /etc/inetd.conf le. If you are running xinetd, then you will need to add a le containing
service www { socket_type wait user server }

www stream

tcp

nowait

nobody

/usr/local/sbin/sh-httpd

REMOTE_HOST=$1 REFERRER=$2 CODE=$3 SIZE=$4

echo "$REMOTE_HOST $REFERRER - [$REQ_DATE] \ \"${REQUEST}\" ${CODE} ${SIZE}" >> ${LOGFILE} } print_header() { echo -e "HTTP/1.0 200 OK\r" echo -e "Server: ${NAME}/${VERSION}\r" echo -e "Date: date\r" } print_error() { echo -e "HTTP/1.0 $1 $2\r"

= = = =

stream no nobody /usr/local/sbin/sh-httpd

to your /etc/xinetd.d/ directory. Then, you must stop any already running web servers and restart inetd (or xinetd). 389

390

36. httpd Apache Web Server

36.1. Web Server Basics

36. httpd Apache Web Server

echo echo echo echo echo exit }

-e -e -e -e -e 1

"Content-type: $DEFCONTENT\r" "Connection: close\r" "Date: date\r" "\r" "$2\r"

cat ${URL} sleep 3 }

read_request() { local DIRT local COMMAND

guess_content_type() { local FILE=$1 local CONTENT case ${FILE##*.} in html) CONTENT=$DEFCONTENT ;; gz) CONTENT=application/x-gzip ;; *) CONTENT=application/octet-stream ;; esac echo -e "Content-type: $CONTENT" } do_get() { local DIR local NURL local LEN if [ ! -d $DOCROOT ]; then log ${PEER} - 404 0 print_error 404 "No such file or directory" fi if [ -z "${URL##*/}" ]; then URL=${URL}${DEFINDEX} fi DIR="dirname $URL" if [ ! -d ${DOCROOT}/${DIR} ]; then log ${PEER} - 404 0 print_error 404 "Directory not found" else cd ${DOCROOT}/${DIR} NURL="pwd/basename ${URL}" URL=${NURL} fi if [ ! -f ${URL} ]; then log ${PEER} - 404 0 print_error 404 "Document not found" fi print_header guess_content_type ${URL} LEN="ls -l ${URL} | tr -s | cut -d -f 5" echo -e "Content-length: $LEN\r\n\r" log ${PEER} - 200 ${LEN}

read REQUEST read DIRT REQ_DATE="date +"%d/%b/%Y:%H:%M:%S %z"" REQUEST="echo ${REQUEST} | tr -s [:blank:]" COMMAND="echo ${REQUEST} | cut -d -f 1" URL="echo ${REQUEST} | cut -d -f 2" PROTOCOL="echo ${REQUEST} | cut -d -f 3" case $COMMAND in HEAD) print_error 501 "Not implemented (yet)" ;; GET) do_get ;; *) print_error 501 "Not Implemented" ;; esac } # # It was supposed to be clean - without any non-standard utilities # but I want some logging where the connections come from, so # I use just this one utility to get the peer address # # This is from the netpipes package PEER="getpeername | cut -d -f 1" read_request

100

105

110

115

120

exit 0

Now run telnet localhost 80, as in Section 26.2. If that works and your log les are being properly appended (use tail -f . . . ), you can try to connect to http://localhost/ with a web browser like Netscape. Notice also that the command getsockname (which tells you which of your own IP addresses the remote client connected to) could allow the script to serve pages from a different directory for each IP address. This is virtual domains in a nutshell. Groovy,
baby, Im in a giant nutshell.... how do I get out?

391

392

36. httpd Apache Web Server

36.2. Installing and Conguring Apache

36. httpd Apache Web Server

36.2 Installing and Conguring Apache

Because all distributions package Apache in a different way, here I assume Apache to have been installed from its source tree, rather than from a .deb or .rpm package. You can refer to Section 24.1 on how to install Apache from its source package. (You can even install it under Win.tar.gz le like any other GNU dows, Windows NT, or OS/2.) The source tree is, of course, available from The Apache Home Page http://www.apache.org. Here I assume you have installed it in -prefix=/opt/apache/. In the process, Apache will have dumped a huge reference manual into /opt/apache/htdocs/manual/.

Order allow,deny Allow from all </Directory> <Directory "/opt/apache/htdocs/home/*/www"> Options Indexes MultiViews AllowOverride None Order allow,deny Allow from all </Directory> UserDir /opt/apache/htdocs/home/*/www

36.2.1

Sample httpd.conf

With the cong le ready, you can move the index.html le above to /opt/apache/htdocs/. You will notice the complete Apache manual and a demo page already installed there; you can move them to another directory for the time being. Now run
/opt/apache/bin/httpd -X

Apache has several legacy conguration les: access.conf and srm.conf are two of them. These les are now deprecated and should be left empty. A single conguration le /opt/apache/conf/httpd.conf may contain at minimum:
ServerType standalone ServerRoot "/opt/apache" PidFile /opt/apache/logs/httpd.pid ScoreBoardFile /opt/apache/logs/httpd.scoreboard Port 80 User nobody Group nobody HostnameLookups Off ServerAdmin webmaster@cranzgot.co.za UseCanonicalName On ServerSignature On DefaultType text/plain ErrorLog /opt/apache/logs/error_log LogLevel warn LogFormat "%h %l %u %t \"%r\" %>s %b" common CustomLog /opt/apache/logs/access_log common DocumentRoot "/opt/apache/htdocs" DirectoryIndex index.html AccessFileName .htaccess <Directory /> Options FollowSymLinks AllowOverride None Order Deny,Allow Deny from All </Directory> <Files "\.ht"> Order allow,deny Deny from all </Files> <Directory "/opt/apache/htdocs"> Options Indexes FollowSymLinks MultiViews AllowOverride All

and then point your web browser to http://localhost/ as before.

36.2.2 Common directives

Here is a description of the options. Each option is called a directive in Apache terminology. A complete list of basic directives is in the le /opt/apache/htdocs/manual/mod/core.html. ServerType As discussed in Section 29.2, some services can run standalone or from inetd (or xinetd). This directive can be exactly standalone or inetd. If you choose inetd, you will need to add an appropriate line into your inetd conguration, although a web server should almost certainly choose standalone mode. ServerRoot This is the directory superstructure See page 137. under which Apache is installed. It will always be the same as the value passed to --prefix=. PidFile Many system services store the process ID in a le for shutdown and monitoring purposes. On most distributions, the le is /var/run/httpd.pid. ScoreBoardFile This option is used for communication between Apache parent and child processes on some non-U NIX systems. Port This is the TCP port for standalone servers to listen on. User, Group This option is important for security. It forces httpd to user nobody privileges. If the web server is ever hacked, the attack will not be able to gain more than the privileges of the nobody user. 394

393

36. httpd Apache Web Server

36.2. Installing and Conguring Apache

36. httpd Apache Web Server

HostnameLookups To force a reverse DNS lookup on every connecting host, set this directive to on. To force a forward lookup on every reverse lookup, set this to double. This option is for logging purposes since access control does a reverse and forward reverse lookup anyway if required. It should certainly be off if you want to reduce latency. ServerAdmin Error messages include this email address. UseCanonicalName If Apache has to return a URL for any reason, it will normally return the full name of the server. Setting to off uses the very host name sent by the client. ServerSignature Add the server name to HTML error messages. DefaultType All les returned to the client have a type eld specifying how the le should be displayed. If Apache cannot deduce the type, it assumes the MIME Type to be text/plain. See Section 12.6.2 for a discussion of MIME types. ErrorLog Where errors get logged, usually /var/log/httpd/error log LogLevel How much info to log. LogFormat Dene a new log format. Here we dened a log format and call it common. Multiple lines are allowed. Lots of interesting information can actually be logged: See /opt/apache/htdocs/manual/mod/mod log config.html for a full description. CustomLog The log le name and its (previously dened) format. DocumentRoot This directive species the top-level directory that client connections will see. The string /opt/apache/htdocs/ is prepended to any le lookup, and hence a URL http://localhost/manual/index.html.en will return the le /opt/apache/htdocs/manual/index.html.en. DirectoryIndex This directive gives the default le to try serve for URLs that contain only a directory name. If a le index.html does not exist under that directory, an index of the directory is sent to the client. Other common congurations use index.htm or default.html. AccessFileName Before serving a le to a client, Apache reads additional directives from a le .htaccess in the same directory as the requested le. If a parent directory contains a .htaccess instead, this one will take priority. The .htaccess le contains directives that limit access to the directory, as discussed below. The above is merely the general conguration of Apache. To actually serve pages, you need to dene directories, each with a particular purpose, containing particular HTML or graphic les. The Apache conguration le is very much like an HTML document. Sections are started with <section parameter> and ended with </section>. 395
5

The most common directive of this sort is <Directory /directory> which does such directory denition. Before dening any directories, we need to limit access to the root directory. This control is critical for security.
<Directory /> Options FollowSymLinks Deny from All Order Deny,Allow AllowOverride None </Directory>

This conguration tells Apache about the root directory, giving clients very restrictive access to it. The directives are Some of these are extracted from the Apache manual. :

Options The Options directive controls which server features are available in a particular directory. There is also the syntax +option or -option to include the options of the parent directory, for example, Options +FollowSymLinks -Indexes. FollowSymLinks The server will follow any symbolic links beneath the directory. Be careful about what symbolic links you have beneath directories with FollowSymLinks. You can, for example, give everyone access to the root directory by having a link ../../../ under htdocsnot what you want. ExecCGI Execution of CGI scripts is permitted. Includes Server-side includes are permitted (more on this later). IncludesNOEXEC Server-side includes are permitted, but the #exec command and #include of CGI scripts are disabled. Indexes If a client asks for a directory by name and no index.html le (or whatever DirectoryIndex le you specied) is present, then a pretty listing of the contents of that directory is created and returned. For security you may want to turn this option off. MultiViews Content-negotiated MultiViews are allowed (more on this later). SymLinksIfOwnerMatch The server will only follow symbolic links for which the target le or directory is owned by the same user ID as the link (more on this later). All All options except for MultiViews. This is the default setting. Deny Hosts that are not allowed to connect. You can specify a host name or IP address, for example, as:
Deny from 10.1.2.3 Deny from 192.168.5.0/24 Deny from cranzgot.co.za

396

36. httpd Apache Web Server

36.2. Installing and Conguring Apache

36. httpd Apache Web Server

which will deny access to 10.1.2.3, all hosts beginning with 192.168.5., and all hosts ending in .cranzgot.co.za, including the host cranzgot.co.za. Allow Hosts that are allowed to connect. This directive uses the same syntax as Deny. Order If order is Deny,Allow, then the Deny directives are checked rst and any client that does not match a Deny directive or does match an Allow directive will be allowed access to the server. If order is Allow,Deny, then the Allow directives are checked rst and any client that does not match an Allow directive or does match a Deny directive will be denied access to the server. AllowOverride In addition to the directives specied here, additional directives will be read from the le specied by AccessFileName, usually called .htaccess. This le would usually exist alongside your .html les or otherwise in a parent directory. If the le exists, its contents are read into the current <Directory . . . > directive. AllowOverride says what directives the .htaccess le is allowed to squash. The complete list can be found in /opt/apache/htdocs/manual/mod/core.html. You can see that we give very restrictive Options to the root directory, as well as very restrictive access. The only server feature we allow is FollowSymLinks, then we Deny any access, and then we remove the possibility that a .htaccess le could override our restrictions. The <Files . . . > directive sets restrictions on all les matching a particular regular expression. As a security measure, we use it to prevent access to all .htaccess les as follows:
<Files "\.ht"> Order allow,deny Deny from all </Files>
5

36.2.3 User HTML directories

Our users may require that Apache know about their private web page directories /www/. This is easy to support with the special UserDir directive:
<Directory "/opt/apache/htdocs/home/*/www"> Options Indexes MultiViews AllowOverride None Order allow,deny Allow from all </Directory> UserDir /opt/apache/htdocs/home/*/www

For this feature to work, you must symlink /opt/apache/htdocs/home to /home, and create a directory www/ under each users home directory. Hitting the URL http://localhost/jack/index.html will then retrieve the le /opt/apache/htdocs/home/jack/www/index.html. You will nd that Apache gives a Forbidden error message when you try to do this. This is probably because jacks home directorys permissions are too restrictive. Your choices vary between now making jacks home directory less restricted or increasing the privileges of Apache. Running Apache under the www group by using Group www, and then running
groupadd -g 65 www chown jack:www /home/jack /home/jack/www chmod 0750 /home/jack /home/jack/www

is a reasonable compromise.

36.2.4 Aliasing
Sometimes, HTML documents will want to refer to a le or graphic by using a simple prex, rather than a long directory name. Other times, you want two different references to source the same le. The Alias directive creates virtual links between directories. For example, adding the following line, means that a URL /icons/bomb.gif will serve the le /opt/apache/icons/bomb.gif:
Alias /icons/ "/opt/apache/icons/"

We are now nally ready to add actual web page directories. These take a less restrictive set of access controls:
<Directory "/opt/apache/htdocs"> Options Indexes FollowSymLinks MultiViews AllowOverride All Order allow,deny Allow from all </Directory>

We do, of course, need to tell Apache about this directory:

<Directory "/opt/apache/icons"> Options None AllowOverride None Order allow,deny

397

398

36. httpd Apache Web Server

36.2. Installing and Conguring Apache

36. httpd Apache Web Server

Allow from all </Directory>

36.2.5

Fancy indexes
10

You will nd the directory lists generated by the preceding conguration rather bland. The directive
IndexOptions FancyIndexing

AddLanguage da .dk AddLanguage nl .nl AddLanguage et .ee AddLanguage fr .fr AddLanguage de .de AddLanguage el .el AddLanguage ja .ja AddLanguage ru .ru LanguagePriority en da nl et fr de el ja ru

The LanguagePriority directive indicates the preferred language if the browser did not specify any. Some les might contain a .koi8-r extension, indicating a Russian character set encoding for this le. Many languages have such custom character sets. Russian les are named webpage.html.ru.koi8-r. Apache must tell the web browser about the encoding type, based on the extension. Here are directives for Japanese, Russian, and UTF-8 UTF-8 is a Unicode character set encoding useful for any language. , as follows:

causes nice descriptive icons to be printed to the left of the le name. What icons match what le types is a trick issue. You can start with:
AddIconByEncoding (CMP,/icons/compressed.gif) x-compress x-gzip AddIconByType (TXT,/icons/text.gif) text/* AddIconByType (IMG,/icons/image2.gif) image/* AddIconByType (SND,/icons/sound2.gif) audio/* AddIconByType (VID,/icons/movie.gif) video/* AddIcon /icons/compressed.gif .Z .z .tgz .gz .zip AddIcon /icons/a.gif .ps .eps AddIcon /icons/layout.gif .html .shtml .htm

AddCharset ISO-2022-JP AddCharset KOI8-R AddCharset UTF-8

.jis .koi8-r .utf8

This requires the Alias directive above to be present. The default Apache conguration contains a far more extensive map of le types.

Once again, the default Apache conguration contains a far more extensive map of languages and character sets.

36.2.6

Encoding and language negotiation

36.2.7 Server-side includes SSI

Apache actually has a built-in programming language that interprets .shtml les as scripts. The output of such a script is returned to the client. Most of a typical .shtml le will be ordinary HTML, which will be served unmodied. However, lines like

You can get Apache to serve gzipped les with this:

AddEncoding x-compress Z AddEncoding x-gzip gz

Now if a client requests a le index.html, but only a le index.html.gz exists, Apache decompresses it on-the-y. Note that you must have the MultiViews options enabled. The next options cause Apache to serve index.html.language-code when index.html is requested, lling in the preferred language code sent by the web browser. Adding these directives causes your Apache manual to display correctly and will properly show documents that have non-English translations. Here also, the MultiViews must be present.
AddLanguage en .en

will be interpreted, and their output included into the HTMLhence the name serverside includes. Server-side includes are ideal for HTML pages that contain mostly static HTML with small bits of dynamic content. To demonstrate, add the following to your httpd.conf:
AddType text/html .shtml AddHandler server-parsed .shtml <Directory "/opt/apache/htdocs/ssi"> Options Includes AllowOverride None Order allow,deny

399

400

36. httpd Apache Web Server

36.2. Installing and Conguring Apache

36. httpd Apache Web Server

Allow from all </Directory>

Create a directory /opt/apache/htdocs/ssi with the index le index.shtml:

<HTML> The date today is .<P> Here is a directory listing:<br> <PRE>  </PRE>  </HTML>

echo echo echo echo echo echo

Make this script executable with chmod a+x test.cgi and test the output by running it on the command-line. Add the line to your httpd.conf le. Next, modify your Options for the directory /opt/apache/htdocs to include ExecCGI, like this:
<Directory "/opt/apache/htdocs"> Options Indexes FollowSymLinks MultiViews ExecCGI AllowOverride All Order allow,deny Allow from all </Directory> AddHandler cgi-script .cgi

and then a le footer.html containing anything you like. It is obvious how useful this procedure is for creating many documents with the same banner by means of a #include statement. If you are wondering what other variables you can print besides DATE LOCAL, try the following:
<HTML> <PRE>  </PRE> </HTML>
5

You can also goto http://localhost/manual/howto/ssi.html to see some other examples.

After restarting Apache you should be able to visit the URL http://localhost/test.cgi. If you run into problems, dont forget to run tail /opt/apache/logs/error log to get a full report. To get a full list of environment variables available to your CGI program, try the following script:
#!/bin/sh

36.2.8

CGI Common Gateway Interface

(I have actually never managed to gure out why CGI is called CGI.) CGI is where a URL points to a script. What comes up in your browser is the output of the script (were it to be executed) instead of the contents of the script itself. To try this, create a le /opt/apache/htdocs/test.cgi:
#!/bin/sh echo echo echo echo echo echo echo echo echo Content-type: text/html <HTML> <HEAD> <TITLE>My First CGI</TITLE> </HEAD> <BODY bgcolor=#CCCCCC text="#000000"> This is my first CGI<P> Please visit

The script will show ordinary bash environment variables as well as more interesting variables like QUERY STRING: Change your script to
#!/bin/sh echo Content-type: text/html echo

echo echo echo echo set echo echo

Content-type: text/html <HTML> <PRE> </PRE> </HTML>

401

402

36. httpd Apache Web Server

36.2. Installing and Conguring Apache

36. httpd Apache Web Server

echo echo echo echo echo

<HTML> <PRE> $QUERY_STRING </PRE> </HTML>

and then go to the URL http://localhost/test/test.cgi?xxx=2&yyy=3. It is easy to see how variables can be passed to the shell script. The preceding example is not very interesting. However, it gets useful when scripts have complex logic or can access information that Apache cant access on its own. In Chapter 38 we see how to deploy an SQL database. When you have covered SQL, you can come back here and replace your CGI script with,
#!/bin/sh echo Content-type: text/html echo
5 25

which looks like:

psql -d template1 -H -c "SELECT * FROM pg_tables;"

This script will dump the table list of the template1 database if it exists. Apache will have to run as a user that can access this database, which means changing User nobody to User postgres. Note that for security you should really limit who can connect to the
postgres database. See Section 38.4.

36.2.9

Forms and CGI

A le
5

Note how this form calls our existing test.cgi script. Here is a script that adds the entered data to a postgres SQL table:
#!/bin/sh echo Content-type: text/html echo opts=echo "$QUERY_STRING" | \ sed -e s/[A-Za-z0-9 %&+,.\/:=@_-]//g -e s/&/ /g -e q for opt in $opts ; do case $opt in name=*) name=${opt/name=/} ;; email=*) email=${opt/email=/} ;; tel=*) tel=${opt/tel=/} ;; esac

To create a functional form, use the HTTP <FORM> tag as follows. /opt/apache/htdocs/test/form.html could contain:
<HTML> <FORM name="myform" action="test.cgi" method="get"> <TABLE> <TR> <TD colspan="2" align="center"> Please enter your personal details: </TD> </TR> <TR> <TD>Name:</TD><TD><INPUT type="text" name="name"></TD> </TR> <TR> <TD>Email:</TD><TD><INPUT type="text" name="email"></TD> </TR> <TR>

403

404

36. httpd Apache Web Server

36.2. Installing and Conguring Apache

36. httpd Apache Web Server

done if psql -d template1 -H -c "\ INSERT INTO people (name, email, tel) \ VALUES ($name, $email, $tel)" 2>&1 | grep -q INSERT ; then echo "<HTML>Your details \"$name\", \"$email\" and \"$tel\"<BR>" echo "have been succesfully recorded.</HTML>" else echo "<HTML>Database error, please contact our webmaster.</HTML>" fi

Then run chown postgres:www test.cgi and chmod a-w,o-rx,u+s test.cgi (or chmod 4550 test.cgi). Recreate your shell script as test.sh and go to the URL again. Apache runs test.cgi, which becomes user postgres, and then executes the script as the postgres user. Even with Apache as User nobody your script will still work. Note how your setuid program is insecure: it takes no arguments and performs only a single function, but it takes environment variables (or input from stdin) that could inuence its functionality. If a login user could execute the script, that user could send data via these variables that could cause the script to behave in an unforeseen way. An alternative is:
#include <unistd.h>

exit 0

Note how the rst lines of script remove all unwanted characters from QUERY STRING. Such processing is imperative for security because shell scripts can easily execute commands should characters like $ and be present in a string. To use the alternative POST method, change your FORM tag to
<FORM name="myform" action="test.cgi" method="post">

The POST method sends the query text through stdin of the CGI script. Hence, you need to also change your opts= line to
opts=cat | \ sed -e s/[A-Za-z0-9 %&+,.\/:=@_-]//g -e s/&/ /g -e q

This script nullies the environment before starting the CGI, thus forcing you to use the POST method only. Because the only information that can be passed to the script is a single line of text (through the -e q option to sed) and because that line of text is carefully stripped of unwanted characters, we can be much more certain of security.

int main (int argc, char *argv[]) { char *envir[] = {0}; setreuid (geteuid (), geteuid ()); execle ("/opt/apache/htdocs/test/test.sh", "test.sh", 0, envir); return 0; }

36.2.11 Apache modules and PHP

CGI execution is extremely slow if Apache has to invoke a shell script for each hit. Apache has a number of facilities for built-in interpreters that will parse script les with high efciency. A well-known programming language developed specically for the Web is PHP. PHP can be downloaded as source from The PHP Home Page http://www.php.net and contains the usual GNU installation instructions. Apache has the facility for adding functionality at runtime using what it calls DSO (Dynamic Shared Object) les. This feature is for distribution vendors who want to ship split installs of Apache that enable users to install only the parts of Apache they like. This is conceptually the same as what we saw in Section 23.1: To give your program some extra feature provided by some library, you can either statically link the library to your program or compile the library as a shared .so le to be linked at run time. The difference here is that the library les are (usually) called mod name and are stored in /opt/apache/libexec/. They are also only loaded if a LoadModule name module appears in httpd.conf. To enable DSO support, rebuild and reinstall Apache starting with:
./configure --prefix=/opt/apache --enable-module=so

36.2.10

Setuid CGIs

Running Apache as a privileged user has security implications. Another way to get this script to execute as user postgres is to create a setuid binary. To do this, create a le test.cgi by compiling the following program similar to that in Section 33.2.
#include <unistd.h> int main (int argc, char *argv[]) { setreuid (geteuid (), geteuid ()); execl ("/opt/apache/htdocs/test/test.sh", "test.sh", 0); return 0; }

405

406

36. httpd Apache Web Server

36.2. Installing and Conguring Apache

36. httpd Apache Web Server

Any source package that creates an Apache module can now use the Apache utility /opt/apache/bin/apxs to tell it about the current Apache installation, so you should make sure this executable is in your PATH. You can now follow the instructions for installing PHP, possibly beginning with ./configure --prefix=/opt/php --with-apxs=/opt/apache/bin/apxs --with-pgsql=/usr. (This assumes that you want to enable support for the postgres SQL database and have postgres previously installed as a package under /usr.) Finally, check that a le libphp4.so eventually ends up in /opt/apache/libexec/. lines Your httpd.conf then needs to know about PHP scripts. Add the following
5

If our machine has one IP address, 196.123.45.1, we may need to congure a separate IP address on the same network card as follows (see Section 25.9):
ifconfig eth0:1 196.123.45.2 netmask 255.255.255.0 up

For each domain /opt/apache/htdocs/www.domain?.com/, we now create a top-level directory. We need to tell Apache that we intend to use the IP address 196.123.45.1 for several hosts. We do that with the NameVirtualHost directive. Then for each host, we must specify a top-level directory as follows:
NameVirtualHost 196.123.45.1 <VirtualHost 196.123.45.1> ServerName www.domain1.com DocumentRoot /opt/apache/htdocs/www.domain1.com/ </VirtualHost> <VirtualHost 196.123.45.1> ServerName www.domain2.com DocumentRoot /opt/apache/htdocs/www.domain2.com/ </VirtualHost> <VirtualHost 196.123.45.2> ServerName www.domain3.com DocumentRoot /opt/apache/htdocs/www.domain3.com/ </VirtualHost>

and then create a le /opt/apache/htdocs/hello.php containing

<html> <head> <title>Example</title> </head> <body> <?php echo "Hi, Im a PHP script!"; ?> </body> </html>

LoadModule php4_module /opt/apache/libexec/libphp4.so AddModule mod_php4.c AddType application/x-httpd-php .php

and test by visiting the URL http://localhost/hello.php. Programming in the PHP language is beyond the scope of this book.

All that remains is to congure a correct DNS zone for each domain so that lookups of www.domain1.com and www.domain2.com return 196.123.45.1 while lookups of www.domain3.com return 196.123.45.2. You can then add index.html les to each directory.

36.2.12

Virtual hosts

Virtual hosting is the use of a single web server to serve the web pages of multiple domains. Although the web browser seems to be connecting to a web site that is an isolated entity, that web site may in fact be hosted alongside many others on the same machine. Virtual hosting is rather trivial to congure. Let us say that we have three domains: www.domain1.com, www.domain2.com, and www.domain3.com. We want domains www.domain1.com and www.domain2.com to share IP address 196.123.45.1, while www.domain3.com has its own IP address of 196.123.45.2. The sharing of a single IP address is called name-based virtual hosting, and the use of a different IP address for each domain is called IP-based virtual hosting. 407 408

37.1. /etc/crontab Conguration File

37. crond and atd

The time pattern gives the minute, hour, day of the month, month, and weekday that the current time is compared. The comparison is done at the start of every single minute. If crond gets a match, it will execute the command. A simple time pattern is as follows.

Chapter 37

50 13 2 9 6 root /usr/bin/play /etc/theetone.wav

which will playen WAV Sat Sep 2 13:50:00 every year, and
50 13 2 * * root /usr/bin/play /etc/theetone.wav

crond and atd

crond and atd are two very simple and important services that everyone should be familiar with. crond does the job of running commands periodically (daily, weekly), and atds main feature is to run a command once at some future time. These two services are so basic that we are not going to detail their package contents and invocation.

will play it at 13:50:00 on the 2nd of every month, and

50 13 * * 6 root /usr/bin/play /etc/theetone.wav

will do the same on every Saturday. Further,

50 13,14 * * 5,6,7 root /usr/bin/play /etc/theetone.wav

will play at 13:50:00 and at 14:50:00 on Friday, Saturday, and Sunday, while
*/10 * * * 6 root /usr/bin/play /etc/theetone.wav

will play every 10 minutes the whole of Saturday. The / is a special notation meaning in steps of. Note that in the above examples, the play command is executed as root. The following is an actual /etc/crontab le:

37.1

/etc/crontab Conguration File

The /etc/crontab le dictates a list of periodic jobs to be runlike updating the locate (see page 43) and whatis (see page 40) databases, rotating logs (see Section 21.4.9), and possibly performing backup tasks. If anything needs to be done periodically, you can schedule that job in this le. /etc/crontab is read by crond on startup. crond will already be running on all but the most broken of U NIX systems. After modifying /etc/crontab, you should restart crond with /etc/rc.d/init.d/crond restart (or /etc/init.d/crond restart, or /etc/init.d/cron restart). /etc/crontab consists of single line denitions for the time of the day/week/month at which a particular command should be run. Each line has the form,
<time> <user> <executable>

# Environment variables first SHELL=/bin/bash PATH=/sbin:/bin:/usr/sbin:/usr/bin MAILTO=root HOME=/ # Time specs 30 20 * * * 35 19 * * * 58 18 * * * 01 * * * * 02 4 * * * 22 4 * * 0 42 4 1 * *

root root root root root root root

/etc/cron-alarm.sh /etc/cron-alarm.sh /etc/cron-alarm.sh run-parts /etc/cron.hourly run-parts /etc/cron.daily run-parts /etc/cron.weekly run-parts /etc/cron.monthly

where <time> is a time pattern that the current time must match for the command to be executed, <user> tells under what user the command is to be executed, and <executable> is the command to be run. 409

Note that the # character is used for comments as usual. crond also allows you to specify environment variables under which commands are to be run. 410

37. crond and atd

37.2. The at Command

37.3. Other cron Packages

37. crond and atd

Your time additions should come like mine have, to remind me of the last three Metro trains of the day. The last four entries are vendor supplied. The run-parts command is a simple script to run all the commands listed under /etc/cron.hourly, /etc/cron.daily, etc. Hence, if you have a script that needs to be run every day but not at a specic time, you neednt edit your crontab le: rather just place the script with the others in /etc/cron.<interval>. My own /etc/cron.daily/ directory contains:
2 59 1 1 1 1 1 1 1 root root root root root root root root root root root root root root root root root root 1024 6144 140 51 390 459 99 103 104 Sep Aug Aug Jun Sep Mar Jul Sep Aug 2 31 13 16 14 25 23 25 30 13:22 13:11 16:16 1999 1999 1999 23:48 1999 1999 . .. backup logrotate makewhatis.cron radiusd.cron.daily slocate.cron tetex.cron tmpwatch

3 2000-09-02 14:19 =

The at and atd man pages contain additional information. Note that atd should generally be disabled for security.

37.3

Other cron Packages

total 14 drwxr-xr-x drwxr-xr-x -rwxr-xr-x -rwxr-xr-x -rwxr-xr-x -rwxr-xr-x -rwxr-xr-x -rwxr-xr-x -rwxr-xr-x

There are many crond implementations. Some have more exible cong les, and others have functionality cope with job schedules that run when the machine is typically switched off (like home PCs). Your distribution may have chosen one of these packages instead.

It is advisable to go through each of these now to see what your system is doing to itself behind your back.

37.2 The at Command

at will execute a command at some future time, and only once. I suppose it is essential to know, although I never used it myself until writing this chapter. at is the front end to the atd daemon which, like crond will almost denitely be running. Try our wave le example, remembering to press Of Text):
[root@cericon /etc]# at 14:19 at> /usr/bin/play /etc/theetone.wav at> <EOT> warning: commands will be executed using /bin/sh job 3 at 2000-09-02 14:19

to get the <EOT> (End

You can type atq to get a list of current jobs:

2000-09-02 14:19 a

a means is the queue name, 3 is the job number, and 2000-09-02 14:19 is the scheduled time of execution. While play is executing, atq will display: 411 412

38.2. postgres

38. postgres SQL Server

38.2

postgres

Chapter 38

postgres (PostGreSQL) is a free SQL server written under the BSD license. postgres supports an extended subset of SQL92. The denitive SQL standard. It does a lot of very nifty things that no other database can (it seems). About the only commercial equivalent worth buying over postgres is a certain very expensive industry leader. postgres runs on every avor of U NIX and also on Windows NT.

The postgres documentation proudly states: The Object-Relational Database Management System now known as PostgreSQL (and briey called Postgres95) is derived from the Postgres package written at Berkeley. With over a decade of development behind it, PostgreSQL is the most advanced open-source database available anywhere, offering multi-version concurrency control, supporting almost all SQL constructs (including subselects, transactions, and user-dened types and functions), and having a wide range of language bindings available (including C, C++, Java, Perl, Tcl, and Python). postgres is also fairly dry. Most people ask why it doesnt have a graphical frontend. Considering that it runs on so many different platforms, it makes sense for it to be purely a back-end engine. A graphical interface is a different kind of software project that would probably support more than one type of database server at the back and possibly run under only one kind of graphical interface. The postgres package consists of the les described in the next two sections:

postgres SQL Server

This chapter will show you how to set up an SQL server for free.

38.1 Structured Query Language

Structured Query Language (SQL) is a programming language developed specically to access data arranged in tables of rows and columnsas in a databaseas well as do searching, sorting and cross-referencing of that data. Typically, the database tables will sit in les managed by an SQL server daemon process. The SQL server will listen on a TCP socket for incoming requests from client machines and will service those requests. SQL has become a de facto industry standard. However, the protocols (over TCP/IP) by which those SQL requests are sent are different from implementation to implementation. SQL requests can usually be typed in manually from a command-line interface. This is difcult for most users, so a GUI interface will usually hide this process from the user. SQL servers and SQL support software is major institution. Management of database tables is actually a complicated affair. A good SQL server will properly streamline multiple simultaneous requests that may access and modify rows in the same table. Doing this efciently, along with the many types of complex searches and cross-referencing, while also ensuring data integrity, is a complex task. 413

38.3

postgres Package Content

The postgres packages consists of the user programs

createdb createlang createuser dropdb droplang dropuser pg_dump pg_dumpall pg_id psql vacuumdb

and the server programs

initdb initlocation ipcclean

Each of these programs has a man page which you should get an inkling of. Further man pages provide references to actual SQL commands. Try man l select (explained further on): 414

pg_ctl pg_encoding pg_passwd

pg_upgrade pg_version postgres

postgresql-dump postmaster

38. postgres SQL Server

38.4. Installing and Initializing postgres

38. postgres SQL Server

SELECT(l) NAME SELECT - Retrieve rows from a table or view.

SELECT(l)

-p /usr/bin/postmaster -o -i -o -e start >/dev/null 2>&1"

SYNOPSIS SELECT [ ALL | DISTINCT [ ON ( expression [, ...] ) expression [ AS name ] [, ...] [ INTO [ TEMPORARY | TEMP ] [ TABLE ] new_table [ FROM table [ alias ] [, ...] ] [ WHERE condition ] [ GROUP BY column [, ...] ] [ HAVING condition [, ...] ] [ { UNION [ ALL ] | INTERSECT | EXCEPT } select [ ORDER BY column [ ASC | DESC | USING operator [ FOR UPDATE [ OF class_name [, ...] ] ] LIMIT { count | ALL } [ { OFFSET | , } start ]

] ] ]

which also (with the -o -e option) forces European date formats (28/4/1984 instead of 4/28/1984). Note that hosts will not be able to connect unless you edit your /var/lib/pgsql/data/pg hba.conf (/etc/postgresql/pg hba.conf on Debian ) le, and add lines like
host mydatabase 192.168.4.7 255.255.255.255 trust

] ] [, ...] ]

In either case, you should check this le to ensure that only trusted hosts can connect to your database, or remove the -i option altogether if you are only connecting from the local machine. To a limited extent, you can also limit what users can connect within this le. It would be nice if the U NIX domain socket that postgres listens on (i.e., /tmp/.s.PGSQL.5432) had permissions 0770 instead of 0777. That way, you could limit connections to only those users belonging to the postgres group. You can add this feature by searching for the chmod command within src/backend/libpq/pqcomm.c inside the postgres-7.0 sources. Later versions may have added a feature to set the permissions on this socket. To run postgres, you need a user of that name. If you do not already have one then enter
/usr/sbin/useradd postgres

Most important is the enormous amount of HTML documentation that comes with postgres. Point your web browser to /usr/doc/postgresql-?.?.? (or /usr/share/doc/. . . ), then dive into the admin, user, programmer, tutorial, and postgres directories. Finally, there are the start and stop scripts in /etc/rc.d/init.d/ (or /etc/init.d/) and the directory in which the database tables themselves are stored: /var/lib/pgsql/.

and restart the server with

/etc/rc.d/init.d/postgresql restart

38.4 Installing and Initializing postgres

postgres can be gotten prepackaged for your favorite distribution. Simply install the package using rpm or dpkg and then follow the instructions given below. Stop the postgres server if it is running; the init.d script may be called postgres or postgresql (Debian commands in parentheses):
/etc/rc.d/init.d/postgres stop ( /etc/init.d/postgresql stop )

The postgres init.d script initializes a template database on rst run, so you may have to start it twice. Now you can create your own database. The following example creates a database finance as well as a postgres user finance. It does these creations while being user postgres (this is what the -U option is for). You should run these commands as user root or as user postgres without the -U postgres.
[root@cericon]# /usr/sbin/useradd finance [root@cericon]# createuser -U postgres --adduser --createdb finance CREATE USER [root@cericon]# createdb -U finance finance CREATE DATABASE [root@cericon]#

Edit the init.d script to support TCP requests. There will be a line like the following to which you can add the -i option. Mine looks like:
su -l postgres -c "/usr/bin/pg_ctl -D $PGDATA \

415

416

38. postgres SQL Server

38.5. Database Queries with psql

38.6. Introduction to SQL

38. postgres SQL Server

38.5 Database Queries with psql

Now that the database exists, you can begin running SQL queries.
[root@cericon]# psql -U finance Welcome to psql, the PostgreSQL interactive terminal. Type:
5 5

datname | datdba | encoding | datpath -----------+--------+----------+----------template1 | 24 | 0 | template1 finance | 26 | 0 | finance (2 rows)

\copyright for distribution terms \h for help with SQL commands \? for help on internal slash commands \g or terminate with semicolon to execute query \q to quit

38.6

Introduction to SQL

finance=# select * from pg_tables; tablename | tableowner | hasindexes | hasrules | hastriggers ----------------+------------+------------+----------+------------pg_type | postgres | t | f | f pg_attribute | postgres | t | f | f pg_proc | postgres | t | f | f pg_class | postgres | t | f | f pg_group | postgres | t | f | f pg_database | postgres | f | f | f pg_variable | postgres | f | f | f pg_log | postgres | f | f | f pg_xactlock | postgres | f | f | f pg_attrdef | postgres | t | f | f pg_relcheck | postgres | t | f | f pg_trigger | postgres | t | f | f pg_inherits | postgres | t | f | f pg_index | postgres | t | f | f pg_statistic | postgres | t | f | f pg_operator | postgres | t | f | f pg_opclass | postgres | t | f | f pg_am | postgres | t | f | f pg_amop | postgres | t | f | f pg_amproc | postgres | f | f | f pg_language | postgres | t | f | f pg_aggregate | postgres | t | f | f pg_ipl | postgres | f | f | f pg_inheritproc | postgres | f | f | f pg_rewrite | postgres | t | f | f pg_listener | postgres | t | f | f pg_description | postgres | t | f | f pg_shadow | postgres | f | f | t (28 rows)

The following are 99% of the commands you are ever going to use. (Note that all SQL commands require a semicolon at the endyou wont be the rst person to ask why without the semicolon.) nothing happens when you press

38.6.1 Creating tables

To create a table called people, with three columns:
CREATE TABLE people ( name text, gender bool, address text );

The created table will title the columns, name, gender, and address. Columns are typed. This means that only the kind of data that was specied at the time of creation can go in that column. In the case of gender, it can only be true or false for the boolean type, which we will associate to the male and female genders. There is probably no reason to use the boolean value here: using an integer or text eld can often be far more descriptive and exible. In the case of name and address, these can hold anything, since they are of the text type, which is the most encompassing type of all. Note that in the postgres documentation, a column is called an attribute for historical reasons. You should try to choose types according to the kind of searches you are going to do and not according to the data it holds. Table 38.1 lists the most of the useful types as well as their SQL92 equivalents. The types in bold are to be used in preference to other similar types for greater range or precision: Postgres Type bool box char(n) cidr Table 38.1 Common postgres types SQL92 or SQL3 Type boolean character(n) Description logical boolean (true/false) rectangular box in 2D plane xed-length character string IP version 4 network or host address
continues...

The preceeding rows are postgress internal tables. Some are actual tables, and some are views of tables. A selective representation of an actual table.

To get a list of databases, try:

417

finance=# select * from pg_database;

418

38. postgres SQL Server

38.6. Introduction to SQL

38. postgres SQL Server

Table 38.1 (continued) Postgres Type circle date decimal oat4 oat8 inet int2 int4 int8 interval line lseg money numeric path point polygon serial time text timetz timestamp varchar(n) time time with time zone timestamp with time zone character varying(n) SQL92 or SQL3 Type date decimal(p,s) oat(p), p 7 oat(p), 7 = p 16 smallint int, integer interval Description circle in 2D plane calendar date without time of day exact numeric for p = 9, s = 0 oating-point number with precision p oating-point number with precision p IP version 4 network or host address signed 2-byte integer signed 4-byte integer signed 8-byte integer general-use time span innite line in 2D plane line segment in 2D plane U.S.-style currency exact numeric for p == 9, s = 0 open and closed geometric path in 2D plane geometric point in 2D plane closed geometric path in 2D plane unique ID for indexing and crossreference time of day arbitrary length text (up to 8k for postgres 7) time of day, including time zone accurate high range, high precision date/time with zone variable-length character string

38.6.3 Adding a column

The ALTER statement changes something:
finance=# ALTER TABLE people ADD COLUMN phone text; ALTER finance=# SELECT * FROM people; name | gender | address | phone ------+--------+---------+------(0 rows)

38.6.4 Deleting (dropping) a column

You cannot drop columns in postgres; you must create a new table from the old table without the column. How to do this will become obvious further on.

decimal(9,2) numeric(p,s)

38.6.5 Deleting (dropping) a table

Use the DROP command to delete most things:
DROP TABLE people;

38.6.6 Inserting rows, object relational

Insert a row with (you can continue typing over multiple lines):

38.6.2

Listing a table

The SELECT statement is the most widely used statement in SQL. It returns data from tables and can do searches:
finance=# SELECT * FROM PEOPLE; name | gender | address ------+--------+--------(0 rows)

finance=# INSERT INTO people (name, gender, address, phone) finance-# VALUES (Paul Sheer, true, Earth, 7617224); INSERT 20280 1

419

The return value is the oid (Object ID) of the row. postgres is an Object Relational database. This term gets thrown around a lot, but it really means that every table has a hidden column called the oid column that stores a unique identity number for each row. The identity number is unique across the entire database. Because it uniquely identies rows across all tables, you could call the rows objects. The oid feature is most useful to programmers. 420

38. postgres SQL Server

38.6. Introduction to SQL

38. postgres SQL Server

38.6.7

Locating rows

38.6.11 Searches
About the simplest search you can do with postgres is Or alternatively, case insensitively and across the address eld:
SELECT * FROM people WHERE lower(name) LIKE %paul% OR lower(address) LIKE %paul%;

The oid of the above row is 20280. To nd it:

SELECT * FROM people WHERE name LIKE %Paul%;

The rst % is a wildcard that matches any length of text before the Paul, and the nal % matches any text after. It is the usual way of searching with a eld, instead of trying to get an exact match. The possibilities are endless:

38.6.8

Listing selected columns, and the oid column

SELECT * FROM people WHERE gender = true AND phone = 8765432;

To list selected columns, try:

SELECT name, address FROM people; SELECT oid, name FROM people; SELECT oid, * FROM people;

It should be obvious what these do.

38.6.12 Migrating from another database; dumping and restoring tables as plain text
Migrating from another database; dumping and restoring tables as plain text

38.6.9

Creating tables from other tables

The command

Here we create a new table and ll two of its columns from columns in our original table:
finance=# CREATE TABLE sitings (person text, place text, siting text); CREATE finance=# INSERT INTO sitings (person, place) SELECT name, address FROM people; INSERT 20324 1

dumps the people table to /tmp/people.txt, as tab delimeter, newline terminated rows. The command,

COPY people TO /tmp/people.txt;

dumps the people table to /tmp/people.txt, as comma-delimited, newlineterminated rows, with (null) whereever there is supposed to be a zero byte. Similarly, the command

COPY people WITH OIDS TO /tmp/people.txt DELIMITERS , WITH NULL AS (null);

38.6.10

Deleting rows

Delete selected rows, like

finance=# DELETE FROM people WHERE name = Paul Sheer; DELETE 1

inserts into the table people the rows from /tmp/people.txt. It assumes one line per row and the tab character between each cell. Note that unprintable characters are escaped with a backslash \ in both output and the interpretation of input. 422

COPY people FROM /tmp/people.txt;

421

38. postgres SQL Server

38.7. Real Database Projects

38. postgres SQL Server

Hence, it is simple to get data from another database. You just have to work out how to dump it as text.

38.6.13

Dumping an entire database

The command pg dump <database-name> dumps your entire database as plain text. If you try this on your database, you will notice that the output contains straightforward SQL commands. Your database can be reconstructed from scratch by piping this output through stdin of the psql command. In other words, pg dump merely produces the exact sequence of SQL commands necessary to reproduce your database. Sometimes a new version of postgres will switch to a database le format that is incompatible with your previous les. In this case it is prudent to do a pg dumpall (and carefully save the output) before upgrading. The output of pg dumpall can once again be fed through stdin of the psql command and contains all the commands necessary to reconstruct all your databases as well as all the data they contain.

38.6.14

38.7 Real Database Projects

University Computer Science majors learn about subjects like Entity Modeling, Relational Algebra, and Database Normalization. These are formal academic methods according to which good databases are designed. You should not venture into constructing any complex database without these methods. Most university book shops will have academic books that teach formal database theory.

423

424

39.1. Samba: An Introduction

39. smbd Samba NT Server

NetBIOS, or Network Basic Input Output System. NetBIOS was a chunk of software that was loaded into memory to provide an interface between programs and the network hardware. It included an addressing scheme that used 16-byte names to identify workstations and networkenabled applications. Next, Microsoft added features to DOS that allowed disk I/O to be redirected to the NetBIOS interface, which made disk space sharable over the LAN. The le-sharing protocol that they used eventually became known as SMB, and now CIFS.

Chapter 39

smbd Samba NT Server

The following introduction is quoted from the Samba online documentation.

Lots of other software was also written to use the NetBIOS API (Application Programmers Interface), which meant that it would never, ever, ever go away. Instead, the workings beneath the API were cleverly gutted and replaced. NetBEUI (NetBIOS Enhanced User Interface), introduced by IBM, provided a mechanism for passing NetBIOS packets over Token Ring and Ethernet. Others developed NetBIOS LAN emulation over higher-level protocols including DECnet, IPX/SPX and, of course, TCP/IP. NetBIOS and TCP/IP made an interesting team. The latter could be routed between interconnected networks (internetworks), but NetBIOS was designed for isolated LANs. The trick was to map the 16-byte NetBIOS names to IP addresses so that messages could actually nd their way through a routed IP network. A mechanism for doing just that was described in the Internet RFC1001 and RFC1002 documents. As Windows evolved, Microsoft added two additional pieces to the SMB package. These were service announcement, which is called browsing, and a central authentication and authorization service known as Windows NT Domain Control.

39.1 Samba: An Introduction by Christopher R. Hertel

A lot of emphasis has been placed on peaceful coexistence between U NIX and Windows. Unfortunately, the two systems come from very different cultures and they have difculty getting along without mediation. . . . and that, of course, is Sambas job. Samba http://samba.org/ runs on U NIX platforms, but speaks to Windows clients like a native. It allows a U NIX system to move into a Windows Network Neighborhood without causing a stir. Windows users can happily access le and print services without knowing or caring that those services are being offered by a U NIX host. All of this is managed through a protocol suite which is currently known as the Common Internet File System, or CIFS http://www.cifs.com. This name was introduced by Microsoft, and provides some insight into their hopes for the future. At the heart of CIFS is the latest incarnation of the Server Message Block (SMB) protocol, which has a long and tedious history. Samba is an open source CIFS implementation, and is available for free from the http://samba.org/ mirror sites. Samba and Windows are not the only ones to provide CIFS networking. OS/2 supports SMB le and print sharing, and there are commercial CIFS products for Macintosh and other platforms (including several others for U NIX). Samba has been ported to a variety of non-U NIX operating systems, including VMS, AmigaOS, and NetWare. CIFS is also supported on dedicated le server platforms from a variety of vendors. In other words, this stuff is all over the place.

Meanwhile, on the Other Side of the Planet. . .

Andrew Tridgell, who is both tall and Australian, had a bit of a problem. He needed to mount disk space from a U NIX server on his DOS PC. Actually, this wasnt the problem at all because he had an NFS (Network File System) client for DOS and it worked just ne. Unfortunately, he also had an application that required the NetBIOS interface. Anyone who has ever tried to run multiple protocols under DOS knows that it can be...er...quirky. So Andrew chose the obvious solution. He wrote a packet sniffer, reverse engineered the SMB protocol, and implemented it on the U NIX box. Thus, he made the U NIX system appear to be a PC le server, which allowed him to mount shared lesystems from the U NIX server while concurrently running NetBIOS applications. Andrew published his code in early 1992. There was a quick, but short succession of bug-x releases, and then he put the project aside. Occasionally he would get email about it, but he otherwise ignored it. Then one day, almost two years later, he decided to link his wifes Windows PC with his own Linux system. Lacking any better options, he used his own server code. He was actually surprised when it worked. Through his email contacts, Andrew discovered that NetBIOS and SMB were actually (though nominally) documented. With this new information at his ngertips he set to work again, but soon ran into another problem. He was contacted by a company claiming trademark on the name that he had chosen for his server software. Rather than cause a fuss, Andrew did a quick scan against a spell-checker dictionary, looking for words containing the letters smb. Samba was in the list. Curiously, that same word is not in the dictionary le that he uses today. (Perhaps they know its been taken.) The Samba project has grown mightily since then. Andrew now has a whole team of programmers, scattered around the world, to help with Samba development. When a new release

History the (hopefully) Untedious Version

It started a long time ago, in the early days of the PC, when IBM and Sytec co-developed a simple networking system designed for building small LANs. The system included something called

425

426

39. smbd Samba NT Server

39.1. Samba: An Introduction

39. smbd Samba NT Server

is announced, thousands of copies are downloaded within days. Commercial systems vendors, including Silicon Graphics, bundle Samba with their products. There are even Samba T-shirts available. Perhaps one of the best measures of the success of Samba is that it was listed in the Halloween Documents, a pair of internal Microsoft memos that were leaked to the Open Source community. These memos list Open Source products which Microsoft considers to be competitive threats. The absolutely best measure of success, though, is that Andrew can still share the printer with his wife.

The other type of name resolution involves the use of an NBNS (NetBIOS Name Service) server. (Microsoft called their NBNS implementation WINS, for Windows Internet Name Service, and that acronym is more commonly used today.) The NBNS works something like the wall of an old-fashioned telephone booth. (Remember those?) Machines can leave their name and number (IP address) for others to see. Hi, Im node Voomba. Call me for a good time! 192.168.100.101

What Samba Does

Samba consists of two key programs, plus a bunch of other stuff that well get to later. The two key programs are smbd and nmbd. Their job is to implement the four basic modern-day CIFS services, which are: File and print services Authentication and Authorization Name resolution Service announcement (browsing)

It works like this: The clients send their NetBIOS names and IP addresses to the NBNS server, which keeps the information in a simple database. When a client wants to talk to another client, it sends the other clients name to the NBNS server. If the name is on the list, the NBNS hands back an IP address. Youve got the name, look up the number. Clients on different subnets can all share the same NBNS server so, unlike broadcast, the point-to-point mechanism is not limited to the local LAN. In many ways the NBNS is similar to the DNS, but the NBNS name list is almost completely dynamic and there are few controls to ensure that only authorized clients can register names. Conicts can, and do, occur fairly easily. Finally, theres browsing. This is a whole nother kettle of worms, but Sambas nmbd handles it anyway. This is not the web browsing we know and love, but a browsable list of services (le and print shares) offered by the computers on a network. On a LAN, the participating computers hold an election to decide which of them will become the Local Master Browser (LMB). The winner then identies itself by claiming a special NetBIOS name (in addition to any other names it may have). The LMBs job is to keep a list of available services, and it is this list that appears when you click on the Windows Network Neighborhood icon. In addition to LMBs, there are Domain Master Browsers (DMBs). DMBs coordinate browse lists across NT Domains, even on routed networks. Using the NBNS, an LMB will locate its DMB to exchange and combine browse lists. Thus, the browse list is propagated to all hosts in the NT Domain. Unfortunately, the synchronization times are spread apart a bit. It can take more than an hour for a change on a remote subnet to appear in the Network Neighborhood.

File and print services are, of course, the cornerstone of the CIFS suite. These are provided by smbd, the SMB daemon. Smbd also handles share mode and user mode authentication and authorization. That is, you can protect shared le and print services by requiring passwords. In share mode, the simplest and least recommended scheme, a password can be assigned to a shared directory or printer (simply called a share). This single password is then given to everyone who is allowed to use the share. With user mode authentication, each user has their own username and password and the System Administrator can grant or deny access on an individual basis. The Windows NT Domain system provides a further level of authentication renement for CIFS. The basic idea is that a user should only have to log in once to have access to all of the authorized services on the network. The NT Domain system handles this with an authentication server, called a Domain Controller. An NT Domain (which should not be confused with a Domain Name System (DNS) Domain) is basically a group of machines which share the same Domain Controller. The NT Domain system deserves special mention because, until the release of Samba version 2, only Microsoft owned code to implement the NT Domain authentication protocols. With version 2, Samba introduced the rst non-Microsoft-derived NT Domain authentication code. The eventual goal, of course, it to completely mimic a Windows NT Domain Controller. The other two CIFS pieces, name resolution and browsing, are handled by nmbd. These two services basically involve the management and distribution of lists of NetBIOS names. Name resolution takes two forms: broadcast and point-to-point. A machine may use either or both of these methods, depending upon its conguration. Broadcast resolution is the closest to the original NetBIOS mechanism. Basically, a client looking for a service named Trillian will call out Yo! Trillian! Where are you?, and wait for the machine with that name to answer with an IP address. This can generate a bit of broadcast trafc (a lot of shouting in the streets), but it is restricted to the local LAN so it doesnt cause too much trouble.

Other Stuff
Samba comes with a variety of utilities. The most commonly used are: smbclient A simple SMB client, with an interface similar to that of the FTP utility. It can be used from a U NIX system to connect to a remote SMB share, transfer les, and send les to remote print shares (printers). nmblookup A NetBIOS name service client. Nmblookup can be used to nd NetBIOS names on a network, look up their IP addresses, and query a remote machine for the list of names the machine believes it owns. swat The Samba Web Administration Tool. Swat allows you to congure Samba remotely, using a web browser. There are more, of course, but describing them would require explaining even more bits and pieces of CIFS, SMB, and Samba. Thats where things really get tedious, so well leave it alone for now.

427

428

39. smbd Samba NT Server

39.1. Samba: An Introduction

39. smbd Samba NT Server

SMB Filesystems for Linux

One of the cool things that you can do with a Windows box is use an SMB le share as if it were a hard disk on your own machine. The N: drive can look, smell, feel, and act like your own disk space, but its really disk space on some other computer somewhere else on the network. Linux systems can do this too, using the smbfs lesystem. Built from Samba code, smbfs (which stands for SMB Filesystem) allows Linux to map a remote SMB share into its directory structure. So, for example, the /mnt/zarquon directory might actually be an SMB share, yet you can read, write, edit, delete, and copy the les in that directory just as you would local les. The smbfs is nifty, but it only works with Linux. In fact, its not even part of the Samba suite. It is distributed with Samba as a courtesy and convenience. A more general solution is the new smbsh (SMB shell, which is still under development at the time of this writing). This is a cool gadget. It is run like a U NIX shell, but it does some funky ddling with calls to U NIX libraries. By intercepting these calls, smbsh can make it look as though SMB shares are mounted. All of the read, write, etc. operations are available to the smbsh user. Another feature of smbsh is that it works on a per user, per shell basis, while mounting a lesystem is a system-wide operation. This allows for much ner-grained access controls.

Internet support is, of course, still available via the comp.protocols.smb newsgroup and the samba@samba.org mailing list. The Samba Team continues to work on new goodies. Current interests include NT ACLs (Access Control Lists), support for LDAP (the Lightweight Directory Access Protocol), NT Domain Control, and Microsofts DFS (Distributed File System).

The Future
Windows 2000 looms on the horizon like a lazy animal peeking its head over the edge of its burrow while trying to decide whether or not to come out. No one is exactly sure about the kind of animal it will be when it does appear, but folks are fairly certain that it will have teeth. Because of their dominance on the desktop, Microsoft gets to decide how CIFS will grow. Windows 2000, like previous major operating system releases, will give us a whole new critter to study. Based on the beta copies and the things that Microsoft has said, here are some things to watch for: CIFS Without NetBIOS Microsoft will attempt to decouple CIFS and NetBIOS. NetBIOS wont go away, mind you, but it wont be required for CIFS networking either. Instead, the SMB protocol will be carried natively over TCP/IP. Name lookups will occur via the DNS. Dynamic DNS Microsoft will implement Dynamic DNS, a still-evolving system designed by the IETF (Internet Engineering Task Force). Dynamic DNS allows names to be added to a DNS server on-the-y. Kerberos V Microsoft has plans to use Kerberos V. The Microsoft K5 tickets are supposed to contain a Privilege Attribute Certicate (PAC) http://www.usenix.org/publications/login/199711/embraces.html, which will include user and group ID information from the Active Directory. Servers will be looking for this PAC when they grant access to the services that they provide. Thus, Kerberos may be used for both authentication and authorization. Active Directory The Active Directory appears to be at the heart of Windows 2000 networking. It is likely that legacy NetBIOS services will register their names in the Active Directory. Hierarchical NT Domains Instead of isolated Domain Controllers, the NT Domain system will become hierarchical. The naming system will change to one that is remarkably similar to that of the DNS. One certainty is that W2K (as it is often called) is, and will be, under close scrutiny. Windows has already attracted the attention of some of the Internet Wonderlands more curious inhabitants, including security analysts, standards groups, crackers dens, and general all-purpose geeks. The business world, which has nally gotten a taste of the freedom of Open Source Software, may be reluctant to return to the world of proprietary, single-vendor solutions. Having the code in your hands is both reassuring and empowering. Whatever the next Windows animal looks like, it will be Sambas job to help it get along with its peers in the diverse world of the Internet. The Samba Team, a microcosm of the Internet community, are among those watching W2K to see how it develops. Watching does not go handin-hand with waiting, though, and Samba is an on-going and open effort. Visit the Samba web site, join the mailing lists, and see whats going on. Participate in the future.

Setup and Management

Samba is congured using the smb.conf le. This is a simple text le designed to look a lot like those *.ini les used in Windows. The goal, of course, is to give network administrators familiar with Windows something comfortable to play with. Over time, though, the number of things that can be congured in Samba has grown, and the percentage of Network Admins willing to edit a Windows *.ini le has shrunk. For some people, that makes managing the smb.conf le a bit daunting. Still, learning the ins and outs of smb.conf is a worthwhile penance. Each of the smb.conf variables has a purpose, and a lot of ne-tuning can be accomplished. The le structure contents are fully documented, so as to give administrators a running head start, and smb.conf can be manipulated using swat, which at least makes it nicer to look at.

The Present
Samba 2.0 was released in January 1999. One of the most signicant and cool features of the 2.0 release was improved speed. Ziff-Davis Publishing used their Netbench software to benchmark Samba 2.0 on Linux against Windows NT4. They ran all of their tests on the same PC hardware, and their results showed Sambas throughput under load to be at least twice that of NT. Samba is shipped with all major Linux distributions, and Ziff-Davis tested three of those. Another milestone was reached when Silicon Graphics (SGI) became the rst commercial U NIX vendor to support Samba. In their December 1998 press release, they claimed that their Origin series servers running Samba 2.0 were the most powerful line of le servers for Windows clients available. SGI now offers commercial support for Samba as do several other providers, many of which are listed on the Samba web site (see http://samba.org/). Traditional

429

430

39. smbd Samba NT Server

39.2. Conguring Samba

39. smbd Samba NT Server

39.2 Conguring Samba

That said, conguring smbd is really easy. A typical LAN will require a U NIX machine that can share /home/* directories to Windows clients, where each user can log in as the name of their home directory. It must also act as a print share that redirects print jobs through lpr; and then in PostScript, the way we like it. Consider a Windows machine divinian.cranzgot.co.za on a local LAN 192.168.3.0/24. The user of that machine would have a U NIX login psheer on the server cericon.cranzgot.co.za. The usual place for Sambas conguration le is /etc/samba/smb.conf on most distributions. A minimalist conguration le to perform the above functions might be:
[global] workgroup = MYGROUP server string = Samba Server hosts allow = 192.168. 127. printcap name = /etc/printcap load printers = yes printing = bsd log file = /var/log/samba/%m.log max log size = 0 security = user socket options = TCP_NODELAY SO_RCVBUF=8192 SO_SNDBUF=8192 encrypt passwords = yes smb passwd file = /etc/samba/smbpasswd [homes] comment = Home Directories browseable = no writable = yes [printers] comment = All Printers path = /var/spool/samba browseable = no guest ok = no printable = yes

Note that with SMB there are all sorts of issues with case interpretationan incorrectly typed password could still work with Samba but obviously wont with U NIX. To start Samba, run the familiar

/etc/init.d/smbd start ( /etc/rc.d/init.d/smbd start ) ( /etc/init.d/samba start )

For good measure, there should also be a proper DNS conguration with forward and reverse lookups for all client machines. At this point you can test your Samba server from the U NIX side. L INUX has native support for SMB shares with the smbfs le system. Try mounting a share served by the local machine:
mkdir -p /mnt/smb mount -t smbfs -o username=psheer,password=12345 //cericon/psheer /mnt/smb

You can now run tail -f /var/log/samba/cericon.log. It should contain messages like:
cericon (192.168.3.2) connect to service psheer as user psheer (uid=500, gid=500) (pid 942)

where a service means either a directory share or a print share. The useful utility smbclient is a generic tool for running SMB requests, but is mostly useful for printing. Make sure your printer daemon is running (and working) and then try
echo hello | smbclient //cericon/lp 12345 -U psheer -c print -

The SMB protocol stores passwords differently from U NIX. It therefore needs its own password le, usually /etc/samba/smbpasswd. There is also a mapping between U NIX logins and Samba logins in /etc/samba/smbusers, but for simplicity we will use the same U NIX name as the Samba login name. We can add a new U NIX user and Samba user and set both their passwords with
smbadduser psheer:psheer useradd psheer smbpasswd psheer passwd psheer

which will create a small entry in the lp print queue. Your log le will be appended with:
cericon (192.168.3.2) connect to service lp as user psheer (uid=500, gid=500) (pid 1014)

431 432

39. smbd Samba NT Server

39.3. Conguring Windows

39.4. Conguring a Windows Printer

39. smbd Samba NT Server

39.3 Conguring Windows

Conguration from Windows begins with a working TCP/IP conguration:

This should bring up your home directory like you have probably never seen it before.

39.4

Conguring a Windows Printer

Under Settings in your Start menu, you can add new printers. Your U NIX lp print queue is visible as the \\cericon\lp network printer and should be entered as such in the conguration wizard. For a printer driver, you should choose Apple Color Laserwriter, since this driver just produces regular PostScript output. In the printer driver options you should also select to optimize for portability. Next, you need to Log Off from the Start menu and log back in as your Samba user.

39.5

Conguring swat

swat is a service, run from inetd, that listens for HTTP connections on port 901. It allows complete remote management of Samba from a web browser. To congure, add the service swat 901/tcp to your /etc/services le, and the following to your /etc/inetd.conf le. being very careful who you allow connections from. If you are running xinetd, create a le /etc/xinetd.d/swat:
service swat { port socket_type wait only_from user server server_args log_on_failure = 901 = stream = no = localhost 192.168.0.0/16 = root = /usr/sbin/swat = -s /etc/samba/smb.conf += USERID swat stream tcp nowait root /usr/sbin/tcpd /usr/sbin/swat

Finally, go to Run. . . in the Start menu and enter \\cericon\psheer. You will be prompted for a password, which you should enter as for the smbpasswd program above. 433

434

39. smbd Samba NT Server

39.6. Windows NT Caveats

disable = no

root (swat does not use smbpasswd to authenticate this login). The web page interface is extremely easy to use

http://cericon:901/. Netscape will request a user and password. You should login as

After restarting inetd (or xinetd), you can point your web browser to

and, being written by the Samba developers themselves, can be trusted to produce working congurations. The web page also gives a convenient interface to all the documentation. Do note that it will completely overwrite your existing conguration le.

39.6 Windows NT Caveats

Windows SMB servers compete to be the name server of their domain by version number and uptime. By this we again mean the Windows name service and not the DNS service. How exactly this works I will not cover here, Probably because I have no idea what I am talking about. but do be aware that conguring a Samba server on a network of many NT machines and getting it to work can be a nightmare. A solution once attempted was to shut down all machines on the LAN, then pick one as the domain server, then bring it up rst after waiting an hour for all possible timeouts to have elapsed. After verifying that it was working properly, the rest of the machines were booted.

Then of course, dont forget your nmblookup command. 435 436

40.1. Documentation

40. named Domain Name Server

Keep this window throughout the entire setup and testing procedure. From now on, when I refer to messages, I am referring to a message in this window.

40.1

Documentation

Chapter 40

named Domain Name Server

In Chapter 27 we dealt with the client side of DNS. In this chapter we congure the name server that services such requests. There seems to be a lot of hype that elevates the name server to something mystical and illusive. In fact, setting up a name server is a standard and trivial exercise. A name server daemon is also no heavyweight service: The named executable is 500 KB and consumes little CPU. The package that the name server comes in is called bind. This chapter assumes a bind of approximately bind-8.2 or later. bind stands for Berkeley Internet Name Domain. The difculty with setting up a name server is that the conguration les are impossible to construct from the specication without some kind of typing error being made. The solution is quite simple: Never create a name server cong le from scratch. Always copy one from an existing working name server. Here we give more example conguration les than explanation. You can copy these examples to create your own name server. Please note before running bind that it has security vulnerabilities. Hence, it may be possible for someone to hack your machine if you are running an old version. Many people are also skeptical about even the latest versions of bind (9.1 at the time of writing) even though no security holes had been announced for this version. An alternative is djbdns, which is purported to be the ultimate DNS server. Before you even start working on name server conguration, you should start a new terminal window with the command (Debian alternative in parentheses):
tail -f /var/log/messages ( tail -f /var/log/syslog )

The man pages for named are hostname(7), named-xfer(8), named(8), and ndc(8). These pages reference a document called the Name Server Operations Guide for BIND. What they actually mean is the PostScript le /usr/[share/]doc/bind<version>/bog/file.psf (or /usr/share/doc/bind/bog.ps). The problem with some of this documentation is that it is still based on the old (now deprecated) named.boot conguration le. There is a script /usr/doc/bind<version>/named-bootconf/named-bootconf (or /usr/sbin/namedbootconf) that reads a named.boot le from stdin and writes a named.conf le to stdout. I found it useful to echo "old config line" | named-bootconf to see what a new style equivalent would be. The directory /usr/[share/]doc/bind[-<version>]/html/ contains the most important general information. It is a complete reference to bind conguration. Parallel directories also contain FAQ documents and various theses on security. A le style.txt contains the recommended layout of the conguration les for consistent spacing and readability. Finally an rfc/ directory contains the relevant RFCs (see Section 13.6).

40.2

Conguring bind

There is only one main conguration le for named: /etc/named.conf (or /etc/bind/named.conf on Debian here we assume a /etc/named.conf le for simplicity). The named service once used a le /etc/named.boot, but this has been scrapped. If there is a named.boot le in your /etc directory, then it is not being used, except possibly by a very old version of bind. Here we will show example congurations necessary for typical scenarios of a name server.

40.2.1 Example conguration

The named.conf le will have in it the line directory "/var/named"; (or directory "/etc/named"; or directory "/var/cache/bind";). This directory holds various les containing textual lists of the name to IP address mappings that bind will serve. The following example is a name server for a company that has been 438

437

40. named Domain Name Server

40.2. Conguring bind

40. named Domain Name Server

given a range of IP addresses 196.28.144.16/29 (i.e., 196.28.144.1623), as well as one single IP address (160.123.181.44). This example also must support a range of internal IP addresses (192.168.2.0255) The trick is not to think about how everything works. If you just copy and edit things in a consistent fashion, carefully reading the comments, bind will work ne. I will now list all necessary les. Local client conguration: /etc/resolv.conf
domain localdomain nameserver 127.0.0.1

type master; file "named.192.168.2";

}; /* Reverse lookups of 196.28.144.* Internet IP numbers: */ zone "144.28.196.in-addr.arpa" { type master; file "named.196.28.144"; }; /* Reverse lookup of 160.123.181.44 only: */ zone "44.181.123.160.in-addr.arpa" { type master; file "named.160.123.181.44"; };

Top-level cong le: /etc/named.conf

/* * The directory line tells named that any further file names * given are under the /var/named/ directory. */ options { directory "/var/named"; /* * If there is a firewall between you and nameservers you want * to talk to, you might need to uncomment the query-source * directive below. Previous versions of BIND always asked * questions using port 53, but BIND 8.1 uses an unprivileged * port by default. */ // query-source address * port 53; }; /* The list of root servers: */ zone "." { type hint; file "named.ca"; }; /* Forward lookups of the localhost: */ zone "localdomain" { type master; file "named.localdomain"; }; /* Reverse lookups of the localhost: */ zone "1.0.0.127.in-addr.arpa" { type master; file "named.127.0.0.1"; }; /* Forward lookups of hosts in my domain: */ zone "cranzgot.co.za" { type master; file "named.cranzgot.co.za"; }; /* Reverse lookups of local IP numbers: */ zone "2.168.192.in-addr.arpa" {

Root name server list: /var/named/named.ca

; Get the original of this file from ftp://ftp.rs.internic.net/domain/named.root ; ; formerly ns.internic.net . 3600000 IN NS a.root-servers.net. a.root-servers.net. 3600000 A 198.41.0.4 . 3600000 NS b.root-servers.net. b.root-servers.net. 3600000 A 128.9.0.107 . 3600000 NS c.root-servers.net. c.root-servers.net. 3600000 A 192.33.4.12 . 3600000 NS d.root-servers.net. d.root-servers.net. 3600000 A 128.8.10.90 . 3600000 NS e.root-servers.net. e.root-servers.net. 3600000 A 192.203.230.10 . 3600000 NS f.root-servers.net. f.root-servers.net. 3600000 A 192.5.5.241 . 3600000 NS g.root-servers.net. g.root-servers.net. 3600000 A 192.112.36.4 . 3600000 NS h.root-servers.net. h.root-servers.net. 3600000 A 128.63.2.53 . 3600000 NS i.root-servers.net. i.root-servers.net. 3600000 A 192.36.148.17 . 3600000 NS j.root-servers.net. j.root-servers.net. 3600000 A 198.41.0.10 . 3600000 NS k.root-servers.net. k.root-servers.net. 3600000 A 193.0.14.129 . 3600000 NS l.root-servers.net. l.root-servers.net. 3600000 A 198.32.64.12 . 3600000 NS m.root-servers.net. m.root-servers.net. 3600000 A 202.12.27.33

Local forward lookups: /var/named/named.localdomain

$TTL 259200 @ IN SOA 2000012101 10800 3600 localhost.localdomain. dns-admin.localhost.localdomain. ( ; Serial number ; Refresh every 3 hours ; Retry every hour

40 5

439

440

40. named Domain Name Server

40.2. Conguring bind

40. named Domain Name Server

3600000 259200 ) IN
10

; Expire after 42 days ; Minimum Time to Live (TTL) of 3 days localhost.localdomain. 127.0.0.1
35

NS A

localhost

; Reserved for future web servers: unused18 IN A 196.28.144.18 unused19 IN A 196.28.144.19 unused20 IN A 196.28.144.20 unused21 IN A 196.28.144.21 unused22 IN A 196.28.144.22 unused23 IN A 196.28.144.23 ; local LAN: pc1 IN A pc2 IN A pc3 IN A pc4 IN A ; and so on... to 192.168.2.255

Local reverse lookups: /var/named/named.127.0.0.1

$TTL 259200 @ IN SOA 2000012101 10800 3600 3600000 259200 ) IN
10

localhost. dns-admin.localhost. ( ; Serial number ; Refresh every 3 hours ; Retry every hour ; Expire after 42 days ; Minimum Time to Live (TTL) of 3 days localhost. localhost.

192.168.2.1 192.168.2.2 192.168.2.3 192.168.2.4

LAN reverse lookups: /var/named/named.192.168.2

NS PTR

IN SOA 2000012101 10800 3600 3600000 259200 ) IN NS PTR PTR PTR PTR ns1.cranzgot.co.za. dns-admin.ns1.cranzgot.co.za. ( ; Serial number ; Refresh every 3 hours ; Retry every hour ; Expire after 42 days ; Minimum Time to Live (TTL) of 3 days ns1.cranzgot.co.za. pc1.cranzgot.co.za. pc2.cranzgot.co.za. pc3.cranzgot.co.za. pc4.cranzgot.co.za.

$TTL 259200 @

Authoritative domain le: /var/named/named.cranzgot.co.za

$TTL 259200 @ IN SOA 2000012101 10800 3600 3600000 259200 ) IN IN IN IN IN
15

ns1.cranzgot.co.za. dns-admin.ns1.cranzgot.co.za. ( ; Serial number ; Refresh every 3 hours ; Retry every hour ; Expire after 42 days ; Minimum Time to Live (TTL) of 3 days ns1.cranzgot.co.za. ns2.cranzgot.co.za. 160.123.181.44 10 mail1.cranzgot.co.za. 20 mail2.cranzgot.co.za.

1 IN 2 IN 3 IN 4 IN ; and so on... to 255

NS NS A MX MX

Authoritative reverse lookups (1): /var/named/named.196.28.144

$TTL 259200 @ IN SOA 2000012101 10800 3600 3600000 259200 ) IN NS NS NS NS NS NS NS NS NS ns1.cranzgot.co.za. dns-admin.ns1.cranzgot.co.za. ( ; Serial number ; Refresh every 3 hours ; Retry every hour ; Expire after 42 days ; Minimum Time to Live (TTL) of 3 days dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net.

; We will use the first IP address for the name server itself: ns1 IN A 196.28.144.16
5 20

; our backup name server is faaar away: ns2 IN A 146.143.21.88 ; FTP server: ftp

196.28.144.17

; Aliases: www mail1 mail2 gopher pop proxy

IN IN IN IN IN IN

CNAME CNAME CNAME CNAME CNAME CNAME

cranzgot.co.za. ns1.cranzgot.co.za. ns2.cranzgot.co.za. ftp.cranzgot.co.za. mail1.cranzgot.co.za. ftp.cranzgot.co.za.

0 1 2 3 4 5 6 7

IN IN IN IN IN IN IN IN

441

442

40. named Domain Name Server

40.2. Conguring bind

40. named Domain Name Server

8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

IN IN IN IN IN IN IN IN IN IN IN IN IN IN IN IN

NS NS NS NS NS NS NS NS PTR PTR PTR PTR PTR PTR PTR PTR

dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net. ns1.cranzgot.co.za. ftp.cranzgot.co.za. unused18.cranzgot.co.za. unused19.cranzgot.co.za. unused20.cranzgot.co.za. unused21.cranzgot.co.za. unused22.cranzgot.co.za. unused23.cranzgot.co.za. dns.big-isp.net. dns.big-isp.net. dns.big-isp.net.
5

Jul Jul Jul Jul Jul Jul Jul Jul Jul Jul Jul Jul Jul Jul Jul Jul

You should get messages like:

8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 8 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 15:45:23 ns1 ns1 ns1 ns1 ns1 ns1 ns1 ns1 ns1 ns1 ns1 ns1 ns1 ns1 ns1 ns1 named[17656]: starting. named 8.2.2-P5 Sat Aug 5 13:21:24 EDT 2000 I named[17656]: hint zone "" (IN) loaded (serial 0) named[17656]: master zone "localhost" (IN) loaded (serial 2000012101) named[17656]: master zone "1.0.0.127.in-addr.arpa" (IN) loaded (serial named[17656]: master zone "cranzgot.co.za" (IN) loaded (serial 20000121 named[17656]: master zone "myisp.co.za" (IN) loaded (serial 2000012101) named[17656]: master zone "2.168.192.in-addr.arpa" (IN) loaded (serial named[17656]: master zone "144.28.196.in-addr.arpa" (IN) loaded (serial named[17656]: master zone "44.181.123.160.in-addr.arpa" (IN) loaded (se named[17656]: listening on [127.0.0.1].53 (lo) named[17656]: listening on [196.28.144.16].53 (eth0) named[17656]: Forwarding source address is [0.0.0.0].1041 named: named startup succeeded named[17657]: group = 25 named[17657]: user = named named[17657]: Ready to answer queries.

24 IN NS 25 IN NS 26 IN NS ; and so on... up to 255

Authoritative reverse lookups (2): /var/named/named.160.123.181.44

$TTL 259200 @ IN SOA 2000012101 10800 3600 3600000 259200 ) IN IN NS NS PTR ns1.cranzgot.co.za. dns-admin.ns1.cranzgot.co.za. ( ; Serial number ; Refresh every 3 hours ; Retry every hour ; Expire after 42 days ; Minimum Time to Live (TTL) of 3 days ns1.cranzgot.co.za. ns2.cranzgot.co.za. cranzgot.co.za.

If you have made typing errors, or named les incorrectly, you will get appropriate error messages. Novice administrators are wont to edit named conguration les and restart named without checking /var/log/messages (or /var/log/syslog) for errors. NEVER do this.

40.2.3 Conguration in detail

If there are no apparent errors in your cong les, you can now more closely examine the contents of the les. 40.2.3.1 Top-level named.conf style format.

The top-level conguration le /etc/named.conf has an obvious Comments are designated by /* */ or //.

40.2.2

Starting the name server

If you have created a conguration similar to that above, you can then run the bind package initialization commands. The actions available are (alternative commands in parentheses):
/etc/rc.d/init.d/named start ( /etc/init.d/named start ) ( /etc/init.d/bind start ) /etc/rc.d/init.d/named stop /etc/rc.d/init.d/named restart /etc/rc.d/init.d/named status

The options section in our case species only one parameter: the directory for locating any les. The le options.html under the bind documentation directories has a complete list of options. Some of these are esoteric, but a few have common uses. The lines zone "." {. . . will be present in all name server congurations. They tell named that the whole Internet is governed by the le named.ca. named.ca in turn contains the list of root name servers. The lines zone "localdomain" {. . . are common. They specify that forward lookups for host.localdomain are contained in the le /var/named/named.localdomain. This le gives a correct result for any lookup for localhost. Many applications query the name server for this name and a fastidious conguration ought to return it correctly. Note that such a lookup works together 444

443

40. named Domain Name Server

40.2. Conguring bind

40. named Domain Name Server

with resolv.confit has a line search localdomain so that a query for localhost gives the same result as a query for localhost.localdomain. The lines zone "1.0.0.127.in-addr.arpa" {. . . resolve reverse lookups for the IP address 127.0.0.1 (stored in the le named.127.0.0.1). Note that 1.0.0.127 is 127.0.0.1 written backwards. In fact, reverse lookups are just forward lookups under the domain .in-addr.arpa. Many applications reverse lookup any received connection to check its authenticity, even from localhost, so you may want to have these lines present to prevent such applications failing or blocking. The rest of the le is the conguration specic to our domain. The lines zone "cranzgot.co.za" {. . . say that information for forward lookups is located in the le named.cranzgot.co.za. The lines zone "1.168.192.in-addr.arpa" {. . . say that information for reverse lookups on the IP address range 192.168.1.0255 is located in the le named.192.168.1. The lines zone "44.182.124.160.in-addr.arpa" {. . . says that information for reverse lookups on the IP address 160.124.182.44 is located in the le named.160.124.182.44. 40.2.3.2 Domain SOA records

40.2.3.3

Dotted and non-dotted host names

If a host name ends in a . then the dot signies a fully qualied host name. If it does not end in a . then the absence of a dot signies that the domain should be appended to the host name. This feature is purely to make les more elegant.
ftp

For instance, the line

IN A 196.28.144.17

A 196.28.144.17

could just as well be written

ftp.cranzgot.co.za. IN

Always be careful to properly end qualied host names with a dot, since failing to do so causes named to append a further domain. 40.2.3.4 Empty host names

If a host name is omitted from the start of the line, then the domain is substituted. The purpose of this notation is also for elegance. For example, is the same as
cranzgot.co.za. IN NS ns1.cranzgot.co.za. IN NS ns1.cranzgot.co.za.

Each of the other named. les has a similar format. They begin with $TTL line and then an @ IN SOA. TTL stands for Time To Live, the default expiration time for all subsequent entries. This line not only prevents a No default TTL set. . . warning message, but really tells the rest of the Internet how long to cache an entry. If you plan on moving your site soon or often, set this to a smaller value. SOA stands for Start of Authority. The host name on the second line species the authority for that domain, and the adjacent <user>.<hostname> species the email address of the responsible person. The next few lines contain timeout specications for cached data and data propagation across the net. These are reasonable defaults, but if you would like to tune these values, consult the relevant documentation listed on page 438. The values are all in seconds. The serial number for the le (i.e., 2000012101) is used to tell when a change has been made and hence that new data should be propagated to other servers. When updating the le in any way, you must increment this serial number. The format is conventionally YYYYMMDDxxexactly ten digits. xx begins with, say, 01 and is incremented with each change made during a day. It is absolutely essential that the serial number be updated whenever a le is edited. If not, the changes will not be reected through the rest of the Internet. 445

40.2.3.5 NS, MX, PTR, A, and CNAME records Each DNS record appears on a single line, associating some host name / domain or IP address with some other host name or IP address. Hence, it is easy to construct a le that makes the Internet think anything you want it to about your organization. The most basic types of record are the A and PTR records. They simply associate a host name with an IP number, or an IP number with a host name, respectively. You should not have more than one host associated to a particular IP number. The CNAME record says that a host is just an alias to another host. So have
IN IN A CNAME 196.28.144.1 ns1.cranzgot.co.za.

ns1 mail1

rather than 446

40. named Domain Name Server

40.2. Conguring bind

40.3. Round-Robin Load-Sharing

40. named Domain Name Server

ns1 mail1 IN IN A A 196.28.144.1 196.28.144.1

Finally, NS and MX records

IN IN NS MX <nameserver> <mailserver>

8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

IN IN IN IN IN IN IN IN IN IN IN IN IN IN IN IN

NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS NS

ns1.picasso.co.za. ns1.picasso.co.za. ns1.picasso.co.za. ns1.picasso.co.za. ns1.picasso.co.za. ns1.picasso.co.za. ns1.picasso.co.za. ns1.picasso.co.za. ns1.cranzgot.co.za. ns1.cranzgot.co.za. ns1.cranzgot.co.za. ns1.cranzgot.co.za. ns1.cranzgot.co.za. ns1.cranzgot.co.za. ns1.cranzgot.co.za. ns1.cranzgot.co.za. ns1.matisse.co.za. ns1.matisse.co.za. ns1.matisse.co.za. ns1.matisse.co.za. ns1.matisse.co.za. ns1.matisse.co.za. ns1.matisse.co.za. ns1.matisse.co.za.

just state that domain <domain> has a name server or mail server <nameserver> or <mailserver>, respectively. MTAs can now locate your mail server as being responsible for email addresses of the form user@cranzgot.co.za. 40.2.3.6 Reverse lookups conguration

The le /var/named/named.196.28.144 contains reverse lookup data on all 255 IP addresses under 196.28.144.. It is, however, our ISP (called big-isp.net) that is responsible for this address range, possibly having bought all 65536 addresses under 196.28.. The Internet is going to query big-isp.net when trying to do a reverse lookup for 196.28.144.?. The problem here is that there are many companies comprising the 196.28.144.? range, each with their own name server, so no single name server can be authoritative for the whole domain 144.28.196.in-addr.arpa. This is the reason for lines in /var/named/named.196.28.144 like
5 IN NS dns.big-isp.net.

24 IN NS 25 IN NS 26 IN NS 27 IN NS 28 IN NS 29 IN NS 30 IN NS 31 IN NS ; and so on... up to 255

Here, Matisse, Dali, and Picasso are other companies that have bought small IP address blocks from big-isp. Each of these lines will redirect queries to the appropriate name server.

IP address 196.28.144.5 is not our responsibility, and hence we refer any such query to a more authoritative name server. On the ISP side, the name server dns.bigisp.net must have a le /var/named/named.196.28.144 that contains something like:
$TTL 259200 @ IN SOA 2000012101 10800 3600 3600000 259200 ) IN
10

40.3

Round-Robin Load-Sharing

dns.dns.big-isp.net. dns-admin.dns.big-isp.net. ( ; Serial number ; Refresh every 3 hours ; Retry every hour ; Expire after 42 days ; Minimum Time to Live (TTL) of 3 days dns.big-isp.net. ns1.dali.co.za. ns1.dali.co.za. ns1.dali.co.za. ns1.dali.co.za. ns1.dali.co.za. ns1.dali.co.za. ns1.dali.co.za. ns1.dali.co.za.

If you have more than one A record for a particular machine, then named will return multiple IP addresses upon a lookup. Load sharing between several web servers is now possiblethe record ordering is randomized with each new lookup and your web browser will only choose the rst listed IP address. For instance, host cnn.com returns several IP addresses. Their zone le conguration might look like
cnn.com. cnn.com. . . . cnn.com. cnn.com. IN IN A A 207.25.71.5 207.25.71.6

NS NS NS NS NS NS NS NS NS

0 1 2 3 4 5 6 7

IN IN IN IN IN IN IN IN

IN IN

A A

207.25.71.29 207.25.71.30

447

448

40. named Domain Name Server

40.4. Conguring named for Dialup Use

40.5. Secondary or Slave DNS Servers

40. named Domain Name Server

40.4 Conguring named for Dialup Use

If you have a dialup connection, the name server should be congured as what is called a caching-only name server. Of course, there is no such thing as a caching-only name serverthe term just means that the name. les have only a few essential records in them. The point of a caching server is to prevent spurious DNS lookups that may eat modem bandwidth or cause a dial-on-demand server to initiate a dialout. It also prevents applications blocking, waiting for a DNS lookup. (Typical examples are sendmail, which blocks for a couple of minutes when a machine is turned on without the network plugged in; and netscape 4, which tries to look up the IP address of news.<localdomain>.)

$TTL 259200 @ IN SOA localhost. root.localhost. ( 2000012101 10800 3600 3600000 259200 ) IN NS localhost. 1.1 IN PTR cericon.priv.ate.

The remaining les are the same as before. In addition to the above, your host name has to be congured as in Chapter 27.

40.4.2 Dynamic IP addresses

The one contingency of dialup machines is that IP addresses are often dynamically assigned, so your 192.168. addresses arent going to apply. Probably one way to get around this is to dial in a few times to get a feel for what IP addresses you are likely to get. Assuming you know that your ISP always gives you 196.26.x.x, you can have a reverse lookup le named.196.26 with nothing in it. This will just cause reverse lookups to fail instead of blocking. Such a hack is probably unnecessary. It is best to identify the particular application that is causing a spurious dialout or causing a block, and then apply your creativity to the particular case.

40.4.1

Example caching name server

For a caching name server, the /etc/name.conf le should look as follows. Replace <nameserver> with the IP address of the name server your ISP has given you. Your local machine name is assumed to be cericon.priv.ate. (The following listings are minus superuous comments and newlines for brevity):
options { forwarders { <nameserver>; }; directory "/var/named"; }; zone zone zone zone zone "." { type hint; file "named.ca"; }; "localdomain" { type master; file "named.localdomain"; }; "1.0.0.127.in-addr.arpa" { type master; file "named.127.0.0.1";}; "priv.ate" { type master; file "named.priv.ate"; }; "168.192.in-addr.arpa" { type master; file "named.192.168"; };

40.5

Secondary or Slave DNS Servers

named can operate as a backup server to another server, also called a slave or secondary server. Like the caching-only server there, is no such thing as a secondary server. Its just the same named running with reduced capacity. Lets say we would like ns2.cranzgot.co.za to be a secondary to ns1.cranzgot.co.za. The named.conf le would look as follows:
options { directory "/var/named"; };
5

The /var/named/named.priv.ate le should look like:

$TTL 259200 @ IN SOA cericon.priv.ate. root.cericon.priv.ate. ( 2000012101 10800 3600 3600000 259200 ) IN NS cericon.priv.ate. cericon IN A 192.168.1.1 news IN A 192.168.1.2

zone "." { type hint; file "named.ca"; }; zone "localdomain" {

The /var/named/named.192.168 le should look like: 449

450

40. named Domain Name Server

40.5. Secondary or Slave DNS Servers

40. named Domain Name Server

type master; file "named.localdomain"; };

zone "1.0.0.127.in-addr.arpa" { type master; file "named.127.0.0.1"; }; zone "cranzgot.co.za" { type slave; file "named.cranzgot.co.za"; masters { 196.28.144.16; }; }; zone "2.168.192.in-addr.arpa" { type slave; file "named.192.168.2"; masters { 196.28.144.16; }; }; zone "144.28.196.in-addr.arpa" { type slave; file "named.196.28.144"; masters { 196.28.144.16; }; }; zone "44.181.123.160.in-addr.arpa" { type slave; file "named.160.123.181.44"; masters { 196.28.144.16; }; };

When an entry has a master in it, you must supply the appropriate le. When an entry has a slave in it, named will automatically download the le from 196.28.144.16 (i.e., ns1.cranzgot.co.za) the rst time a lookup is required from that domain. And thats DNS! 451 452

41.1. Basic Dialup

41. Point-to-Point Protocol Dialup Networking

con: ppp" \ /dev/<modem> 57600 debug crtscts modem lock nodetach \ hide-password defaultroute \ user <username> \ noauth

Chapter 41

This is a minimalists dial-in command and its specic to my ISP only. Dont use the exact command unless you have an account with the Internet Solution ISP in South Africa, before January 2000. The command-line options are explained as follows:

Point-to-Point Protocol Dialup Networking

Dialup networking is unreliable and difcult to congure. The reason is simply that telephones were not designed for data. However, considering that the telephone network is by far the largest electronic network on the globe, it makes sense to make use of it. This is why modems were created. On the other hand, the advent of ISDN is slightly more expensive and a better choice for all but home dialup. See Section 41.6 for more information.

connect <script> Species the script that pppd must use to start things up. When you use a modem manually (as is shown further below), you need to go through the steps of initializing the modem, causing a dial, connecting, logging in, and nally telling the remote computer that you would like to set the connection to data communication mode, called the point-to-point protocol, or PPP. The <script> is the automation of this manual procedure. chat -S -s -v <expect> <send> <expect> <send> ... The <script> proper. chat has a man page and uses other than modem communication. -S means to log messages to the terminal and not to syslog; -s means to log to stderr; -v means verbose output. After the options comes a list of things the modem is likely to say, alternated with appropriate responses. This is called an expectsend sequence. The sequence AT&F1 is the modem initialization string. This example assumes that an initialization string of AT&F1 is sufcient. See Section 3.5. \q means to not print the password amid the debug outputvery important.

41.1 Basic Dialup

For home use, dialup networking is not all that difcult to congure. The PPP HOWTO contains lots on this (see Section 16). For my machine this boils down to creating the les /etc/ppp/chap-secrets and /etc/ppp/pap-secrets, both containing the following line of text:
<username> * <password> *

/dev/tty?? Species the device you are going to use. This will usually be /dev/ttyS0, /dev/ttyS1, /dev/ttyS2, or /dev/ttyS3. 57600 The speed the modem is to be set to. This is only the speed between the PC and the modem and has nothing to do with the actual data throughput. It should be set as high as possible except in the case of very old machines whose serial ports may possibly only handle 38400. Its best to choose 115200 unless this doesnt work. debug Output debug information. This option is useful for diagnosing problems. crtscts Use hardware ow control. modem Use modem control lines. This is actually the default. lock Create a UUCP lock le in /var/lock/. As explained in Section 34.4, this is a le of the form /var/lock/LCK..tty?? that tells other applications that the serial device is in use. For this reason, you must not call the device /dev/modem or /dev/cua?. 454

although only one of the les will be used, then running the following command at a shell prompt: This example assumes that an initialization string of AT&F1 is sufcient. See Section
3.5.

pppd connect \ "chat -S -s -v \ AT&F1 \ OK ATDT<tel-number> CONNECT \ name: <username> assword: \q<password> \

453

41. Point-to-Point Protocol Dialup Networking

41.1. Basic Dialup

41. Point-to-Point Protocol Dialup Networking

nodetach Remain always a foreground process. This allows you to watch pppd run and stop it with C. defaultroute Create an IP route after PPP comes alive. Henceforth, packets will go to the right place. hide-password Hide the password from the logs. This is important for security. user <username> Species the line from the /etc/ppp/chap-secrets and /etc/ppp/pap-secrets le to use. For a home PC there is usually only one line.

[root@cericon]#

Now you can modify the above chat script as you need. The kinds of things that will differ are trivial: like having login: instead of name:. Some systems also require you to type something instead of ppp, and some require nothing to be typed after your password. Some further require nothing to be typed at all, thus immediately entering PPP mode. Note that dip also creates UUCP lock les as explained in Section 34.4.

41.1.1

Determining your chat script

41.1.2 CHAP and PAP

You may ask why there are /etc/ppp/chap-secrets and /etc/ppp/papsecrets les if a user name and password are already specied inside the the chat script. CHAP (Challenge Handshake Authentication Protocol) and PAP (Password Authentication Protocol) are authentication mechanisms used after logging inin other words, somewhere amid the y}#A!}!e} }3}"}&} }*} } }}&4}2Iq}}"}(}"N$y}#A!}!r} }4}"}&} }.

To determine the list of expectsend sequences, you need to do a manual dial-in. The command
dip -t

stands for dial-IP and talks directly to your modem. The following session demonstrates a manual dial for user psheer. Using dip manually like this is a game of trying to get the garbage lines you see below: this is PPP starting to talk. When you get this junk, you have won and can press C. Then, copy and paste your session for future reference.
[root@cericon]# dip -t DIP: Dialup IP Protocol Driver version 3.3.7o-uri (8 Feb 96) Written by Fred N. van Kempen, MicroWalt Corporation.
5

41.1.3 Running pppd

If you run the pppd command above, you will get output something like this:
send (AT&F1M) expect (OK) AT&F1MM OK -- got it send (ATDT4068500M) expect (CONNECT) M ATDT4068500MM CONNECT -- got it send (M) expect (name:) 45333/ARQ/V90/LAPM/V42BISM Checking authorization, Please wait...M username: -- got it send (psheerM) expect (assword:) psheerM password: -- got it send (??????) expect (con:)

DIP> port ttyS0 DIP> speed 57600 DIP> term [ Entering TERMINAL mode. Use CTRL-] to get back ] AT&F1 OK ATDT4068500 CONNECT 26400/ARQ/V34/LAPM/V42BIS Checking authorization, please wait... name:psheer password: c2-ctn-icon:ppp Entering PPP mode. Async interface address is unnumbered (FastEthernet0) Your IP address is 196.34.157.148. MTU is 1500 bytes y}#A!}!e} }3}"}&} }*} } }}&4}2Iq}}"}(}"N$y}#A!}!r} }4}"}&} } [ Back to LOCAL mode. ] DIP> quit

455

456

41. Point-to-Point Protocol Dialup Networking

41.1. Basic Dialup

41.2. Demand-Dial, Masquerading 41. Point-to-Point Protocol Dialup Networking

M M c2-ctn-icon: -- got it send (pppM) Serial connection established. Using interface ppp0 Connect: ppp0 <--> /dev/ttyS0 sent [LCP ConfReq id=0x1 <asyncmap 0x0> <magic 0x88c5a54f> <pcomp> <accomp>] rcvd [LCP ConfReq id=0x3d <asyncmap 0xa0000> <magic 0x3435476c> <pcomp> <accomp>] sent [LCP ConfAck id=0x3d <asyncmap 0xa0000> <magic 0x3435476c> <pcomp> <accomp>] rcvd [LCP ConfAck id=0x1 <asyncmap 0x0> <magic 0x88c5a54f> <pcomp> <accomp>] sent [IPCP ConfReq id=0x1 <addr 192.168.3.9> <compress VJ 0f 01>] sent [CCP ConfReq id=0x1 <deflate 15> <deflate(old#) 15> <bsd v1 15>] rcvd [IPCP ConfReq id=0x45 <addr 168.209.2.67>] sent [IPCP ConfAck id=0x45 <addr 168.209.2.67>] rcvd [IPCP ConfRej id=0x1 <compress VJ 0f 01>] sent [IPCP ConfReq id=0x2 <addr 192.168.3.9>] rcvd [LCP ProtRej id=0x3e 80 fd 01 01 00 0f 1a 04 78 00 18 04 78 00 15 03 2f] rcvd [IPCP ConfNak id=0x2 <addr 196.34.157.131>] sent [IPCP ConfReq id=0x3 <addr 196.34.157.131>] rcvd [IPCP ConfAck id=0x3 <addr 196.34.157.131>] local IP address 196.34.25.95 remote IP address 168.209.2.67 Script /etc/ppp/ip-up started (pid 671) Script /etc/ppp/ip-up finished (pid 671), status = 0x0 Terminating on signal 2. Script /etc/ppp/ip-down started (pid 701) sent [LCP TermReq id=0x2 "User request"] rcvd [LCP TermAck id=0x2]

it.

This clearly shows what pppd has done: created a network device and a route to

If your name server is congured, you should now be able to ping metalab.unc.edu or some well-known host.

41.2

Demand-Dial, Masquerading

Dial-on-demand really just involves adding the demand option to the pppd commandline above. The other way of doing dial-on-demand is to use the diald package, but here we discuss the pppd implementation. The diald package is, however, a far more thorough solution. With the demand option, you will notice that spurious dialouts take place. You need to add some ltering rules to ensure that only the services you are interested in cause a dialout. These services should only make outgoing connections when absolutely necessary. A rewall script might look as follows. This example uses the old ipfwadm command, possibly called /sbin/ipfwadm-wrapper on your machine. The newer
ipchains command is now superseded by a completed different packet ltering system in kernel 2.4.

You can see the expectsend sequences working, so its easy to correct them if you made a mistake somewhere. At this point you might want to type route -n and ifconfig in another terminal:
[root@cericon]# route -n Kernel IP routing table Destination Gateway Genmask Flags Metric Ref Use 168.209.2.67 0.0.0.0 255.255.255.255 UH 0 0 0 127.0.0.0 0.0.0.0 255.0.0.0 U 0 0 0 0.0.0.0 168.209.2.69 0.0.0.0 UG 0 0 0 [root@cericon]# ifconfig lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 UP LOOPBACK RUNNING MTU:3924 Metric:1 RX packets:2547933 errors:0 dropped:0 overruns:0 frame:0 TX packets:2547933 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 ppp0 Iface ppp0 lo ppp0

See the Firewall-HOWTO for more information on building a rewall.

# Enable ip forwarding and dynamic address changing: echo 1 > /proc/sys/net/ipv4/ip_forward echo 1 > /proc/sys/net/ipv4/ip_dynaddr
5

# Clear all firewall rules: /sbin/ipfwadm -O -f /sbin/ipfwadm -I -f /sbin/ipfwadm -F -f /sbin/ipfwadm -O -p deny /sbin/ipfwadm -I -p deny # Allow all local communications: /sbin/ipfwadm -O -a accept -D 192.168.0.0/16 /sbin/ipfwadm -O -a accept -D 127.0.0.0/24 /sbin/ipfwadm -O -a accept -S 192.168.0.0/16 /sbin/ipfwadm -O -a accept -S 192.168.0.0/16 /sbin/ipfwadm -I -a accept -S 192.168.0.0/16 /sbin/ipfwadm -I -a accept -S 127.0.0.0/24 /sbin/ipfwadm -I -a accept -D 192.168.0.0/16 /sbin/ipfwadm -I -a accept -D 192.168.0.0/16

Link encap:Point-to-Point Protocol inet addr:196.34.25.95 P-t-P:168.209.2.67 Mask:255.255.255.255 UP POINTOPOINT RUNNING NOARP MULTICAST MTU:1500 Metric:1 RX packets:7 errors:0 dropped:0 overruns:0 frame:0 TX packets:7 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:10

-S -S -D -D -D -D -S -S

0.0.0.0/0 127.0.0.0/24 127.0.0.0/24 192.168.0.0/16 0.0.0.0/0 127.0.0.0/24 127.0.0.0/24 192.168.0.0/16

# Allow ports outgoing: /sbin/ipfwadm -O -a accept -P tcp -S 0.0.0.0/0 \ -D 0.0.0.0/0 20 21 22 25 53 80 110 119 143

457

458

41. Point-to-Point Protocol Dialup Networking 41.2. Demand-Dial, Masquerading

41.3. Dialup DNS

41. Point-to-Point Protocol Dialup Networking

/sbin/ipfwadm -O -a accept -P udp -S 0.0.0.0/0 -D 0.0.0.0/0 53 # # Add this line to allow FTP from masqueraded machines: # /sbin/ipfwadm -O -a accept -P tcp -S 0.0.0.0/0 -D 0.0.0.0/0 1024:65535
30

# Allow ports /sbin/ipfwadm /sbin/ipfwadm /sbin/ipfwadm

incoming: -I -a accept -P tcp -S 0.0.0.0/0 -D 0.0.0.0/0 20 113 -I -a accept -P tcp -S 0.0.0.0/0 -D 0.0.0.0/0 1024:65535 -I -a accept -P udp -S 0.0.0.0/0 -D 0.0.0.0/0 1024:65535

/dev/ttyS0 57600 debug crtscts modem lock nodetach \ hide-password defaultroute \ user <username> \ demand \ :10.112.112.112 \ idle 180 \ holdoff 30

The ports we are using are 20 21 22 25 53 80 110 113 119 143 ftp-data ftp ssh smtp domain www pop3 auth nntp imap2

41.3

Dialup DNS

Your DNS service, to be used on a dialup server, requires some customization. Replace your options section from the DNS congurations in Chapter 40 with the following:
options { forwarders { 196.7.173.2; /* example only */ }; listen-on { 192.168.2.254; }; directory "/var/cache/bind"; dialup yes; notify no; forward only; };

The auth service is not needed but should be kept open so that connecting services get a failure instead of waiting for a timeout. You can comment out the auth line in /etc/inetd.conf for security. If you have a LAN of machines that needs to share the same dialup link, then you can give them all 192.168. addresses and masquerade the LAN through the PPP interface. IP masquerading or NAT (network address translation) can be done with:
# Masquerading for ftp requires special handling on older kernels: /sbin/modprobe ip_masq_ftp # Masquerade the /sbin/ipfwadm -F /sbin/ipfwadm -F /sbin/ipfwadm -F domain 192.168.2.0/255.255.128.0 -f -p deny -a m -S 192.168.0.0/17 -D 0.0.0.0/0

The options dialup yes; notify no; forward only; tell bind to use the link as little as possible; not send notify messages (there are no slave servers on our LAN to notify) and to forward requests to 192.168.2.254 rather than trying to answer them itself; respectively. The option listen-on causes the name server to bind to the network interface 192.168.2.254 only. In this example, the interface 192.168.2.254 is our Ethernet card which routes packets from the local LAN. This is important for security, because it prevents any possible connection from the outside. There is also a DNS package written specically for use by dialup servers. It is called dnrd and is much easier to congure than bind.

41.4

Dial-in Servers

The pppd script becomes (note that you need pppd-2.3.11 or later for this to work as I have it here):
pppd connect \ "chat -S -s -v \ AT&F1 \ OK ATDT<tel-number> CONNECT \ name: <username> assword: \q<password> \ con: ppp" \

pppd is really just a way to initiate a network device over a serial port, regardless of whether you initiate or listen for a connection. As long as there is a serial connection between two machines, pppd will negotiate a link. To listen for a pppd dial-in, you need just add the following line to your /etc/inittab le: and then the line
S0:2345:respawn:/sbin/mgetty -s 115200 ttyS0

459

460

41. Point-to-Point Protocol Dialup Networking

41.4. Dial-in Servers

41.5. Using tcpdump

41. Point-to-Point Protocol Dialup Networking

/AutoPPP/ - a_ppp /usr/sbin/pppd

41.5

Using tcpdump

to the le /etc/mgetty+sendfax/login.config (/etc/mgetty/login.config for Debian). For security, you would probably want to run chmod as /usr/sbin/pppd, since mgetty runs pppd as root anyway. Your /etc/ppp/options le could contain
proxyarp mtu 552 mru 552 require-chap <hostname>:

If a dialout does occur unexpectedly, you can run tcpdump to dump packets going to your ppp0 device. This output will probably highlight the error. You can then look at the TCP port of the service and try to gure out what process the packet might have come from. The command is:
tcpdump -n -N -f -i ppp0

Note that we dispense with the serial line options (i.e., speed and ow control) because mgetty would have already initialized the serial line. <hostname> is just the name of the local machine. The proxyarp setting adds the remote client to the ARP tables. This enables your client to connect through to the Internet on the other side of the line without extra routes. The le /etc/ppp/chap-secrets can be lled with lines like,
dialup * <passwd> 192.168.254.123

tcpdump is also discussed in Section 25.10.3.

41.6

ISDN Instead of Modems

to specify the IP address and password of each user. Next, add a user dialup and perhaps set its password to that in the chapsecrets le. You can then test your conguration from a remote machine with dip t as above. If that works (i.e., mgetty answers, and you get your garbage lines as on page 456), then a proper pppd dial-in should also work. The /etc/ppp/chapsecrets le can contain:
dialup * <passwd> *

and you can dial out using a typical pppd command, like this:
pppd \ connect "chat -S -s -v AT&F1 OK ATDT<telephone> CONNECT " /dev/<modem> 57600 debug crtscts modem lock nodetach hide-password defaultroute \ user dialup \ noauth

For those who are not familiar with ISDN, this paragraph gives you a quick summary. ISDN stands for Integrated Services Digital Network. ISDN lines are like regular telephone lines, except that an ISDN line comes with two analog and two digital channels. The analog channels are regular telephone lines in every respectjust plug your phone in and start making calls. The digital lines each support 64 kilobits/second data transfer; only ISDN communication equipment is meant to plug in to these and the charge rate is the same as that of a telephone call. To communicate over the digital line, you need to dial an ISP just as with a regular telephone. PPP runs over ISDN in the same way as a modem connection. It used to be that only very expensive ISDN routers could work with ISDN, but ISDN modems and ISDN ISA/PCI cards have become cheap enough to allow anyone to use ISDN, and most telephone companies will install an ISDN line as readily as a regular telephone line. So you may ask whats with the Integrated Services. I suppose it was thought that this service, in allowing both data and regular telephone, would be the ubiquitous communications service. It remains to be seen, however, if video conferencing over 64-Kb lines becomes mainstream. ISDN is not covered in detail here, although ample HOWTOs exists on the subject. Be wary when setting up ISDN. ISDN dials really fast. It can dial out a thousand times in a few minutes, which is expensive.

You should be carefully to have a proper DNS conguration for forward and reverse lookups of your pppd IP addresses. This is so that no services block with long timeouts and also so that other Internet machines will be friendly to your users connections. Note that the above also supports faxes, logins, voice, and uucp (see Section 34.3) on the same modem because mgetty only starts pppd if it sees an LCP request (part of the PPP protocol). If you just want PPP, read the cong les in /etc/mgetty+sendfax/ (Debian /etc/mgetty/) to disable the other services. 461 462

42.2. Kernel Version Numbers

42. Kernel

Chapter 42

not generally possibly to use a module on a kernel other than the one it was compiled forhence the name module is used instead of driver. Modules are separated out of the kernel image purely to save RAM. Modules are sometimes compiled into the kernel in the same way that our test program was statically linked on page 230. In this case, they would be absent from /lib/modules/<version>/ and should not really be called modules. In this chapter I show how to create compiled-in or compiled-out versions of modules when rebuilding the kernel. Next is the System.map le, also in /boot. It is used by klogd to resolve kernel address references to symbols, so as to write logs about them, and then also by depmod to work out module dependencies (what modules need what other modules to be loaded rst). Finally, the kernel headers /usr/src/linux/include are used when certain packages are built. The various support daemons should be running already. Since 2.2, these have been reduced to klogd only. The other kernel daemons that appear to be running are generated by the kernel itself.

The L INUX Kernel Source, Modules, and Hardware Support

This chapter explains how to congure, patch, and build a kernel from source. The conguration of device drivers and modules is also discussed in detail.

42.2 42.1 Kernel Constitution

A kernel installation consists of the kernel boot image, the kernel modules, the System.map le, the kernel headers (needed only for development), and various support daemons (already provided by your distribution). These constitute everything that is called Linux under L INUX , and are built from about 50 megabytes of code of around 1.5 million lines. The L INUX kernel image is a 400 to 600-KB le that sits in /boot/ (see Chapter 31). If you look in this directory, you might see several kernels. The choice of which to boot is probably available to you at boot time, through lilo. The kernel in /boot/ is compressed. That is, it is gzip compressed and is actually about twice the size when unpacked into memory on boot. The kernel also has detached parts called modules. These all sit in /lib/modules/<version>/. They are categorized into the subdirectories below this directory. In kernel 2.2 there were about 400, modules totaling about 9 megabytes. Modules are actually just shared object les, like the .o les we created in Section 23.1. They are not quite the same as Windows device drivers, in that it is 463

Kernel Version Numbers

The kernel is versioned like other packages: linux-major.minor.patch. Development kernels are given odd minor version numbers; stable kernels are given even minor version numbers. At the time of writing, the stable kernel was 2.2.17, and 2.4.0 was soon to be released. By the time you read this, 2.4.0 will be available. This chapter should be entirely applicable to future stable releases of 2.4.

42.3

Modules, insmod Command, and Siblings

A module is usually a device driver pertaining to some device node generated with the mknod command or already existing in the /dev/ directory. For instance, the SCSI driver automatically locks onto device major = 8, minor = 0, 1,. . . , when it loads; and the Sound module onto device major = 14, minor = 3 (/dev/dsp), and others. The modules people most often play with are SCSI, Ethernet, and Sound modules. There are also many modules that support extra features instead of hardware. Modules are loaded with the insmod command, and removed with the rmmod command. This is somewhat like the operation of linking shown in the Makefile on page 233. To list currently loaded modules, use lsmod. Try (kernel 2.4 paths are different and are given in braces) 464

42. Kernel

42.3. Modules, insmod Command, and Siblings

42.4. Interrupts, I/O Ports, and DMA Channels

42. Kernel

insmod /lib/modules/<version>/fs/fat.o ( insmod /lib/modules/<version>/kernel/fs/fat/fat.o ) lsmod rmmod fat lsmod

42.4

Interrupts, I/O Ports, and DMA Channels

rmmod -a further removes all unused modules. Modules sometimes need other modules to be present in order to load. If you try to load a module and it gives <module-name>: unresolved symbol <symbolname> error messages, then it requires something else to be loaded rst. The modprobe command loads a module along with other modules that it depends on. Try
insmod /lib/modules/2.2.12-20/fs/vfat.o ( insmod /lib/modules/<version>/kernel/fs/vfat/vfat.o ) modprobe vfat

A loaded module that drives hardware will often consume I/O ports, IRQs, and possibly a DMA channel, as explained in Chapter 3. You can get a full list of occupied resources from the /proc/ directory:
[root@cericon]# cat /proc/ioports 0000-001f 0020-003f 0040-005f 0060-006f 0070-007f 0080-008f 00a0-00bf 00c0-00df 00f0-00ff 0170-0177 01f0-01f7 0220-022f 02f8-02ff 0330-0333 0376-0376 0378-037a 0388-038b 03c0-03df 03f0-03f5 03f6-03f6 03f7-03f7 03f8-03ff e400-e47f f000-f007 f008-f00f : : : : : : : : : : : : : : : : : : : : : : : : : dma1 pic1 timer keyboard rtc dma page reg pic2 dma2 fpu ide1 ide0 soundblaster serial(auto) MPU-401 UART ide1 parport0 OPL3/OPL2 vga+ floppy ide0 floppy DIR serial(auto) DC21140 (eth0) ide0 ide1

modprobe, however, requires a table of module dependencies. This table is the le /lib/modules/<version>/modules.dep and is generated automatically by your startup scripts with the command
/sbin/depmod -a

although you can run it manually at any time. The lsmod listing also shows module dependencies in brackets.
Module de4x5 parport_probe parport_pc lp parport slip slhc sb uart401 sound soundlow soundcore loop nls_iso8859-1 nls_cp437 vfat fat Size 41396 3204 5832 4648 7320 7932 4504 33812 6224 57464 420 2596 7872 2272 3748 9372 30656 Used by 1 (autoclean) 0 (autoclean) 1 (autoclean) 0 (autoclean) 1 (autoclean) [parport_probe parport_pc lp] 2 (autoclean) 1 (autoclean) [slip] 0 0 [sb] 0 [sb uart401] 0 [sound] 6 [sb sound] 2 (autoclean) 1 (autoclean) 1 (autoclean) 1 (autoclean) 1 (autoclean) [vfat]

[root@cericon]# cat /proc/interrupts

0: 1: 2: 3: 5: 6: 7: 8: 11: 13:

CPU0 8409034 157231 0 104347 2 82 2 1 8 1

XT-PIC XT-PIC XT-PIC XT-PIC XT-PIC XT-PIC XT-PIC XT-PIC XT-PIC XT-PIC

timer keyboard cascade serial soundblaster floppy parport0 rtc DC21140 (eth0) fpu

465

466

42. Kernel

42.5. Module Options and Device Conguration

42. Kernel

14: 15: NMI:

237337 16919 0

XT-PIC XT-PIC

ide0 ide1

[root@cericon]# cat /proc/dma 1: 2: 4: 5: SoundBlaster8 floppy cascade SoundBlaster16

to initialize the Adaptec 1542 SCSI driver. What these options are and exactly what goes in them can be learned from the le /usr/src/linux<version>/drivers/scsi/aha1542.c. Near the top of the le are comments explaining the meaning of these options. 2. If you are using LOADLIN.EXE or some other DOS or Windows kernel loader, then it, too, can take similar options. I will not go into these.
append = aha1542=<portbase>[,<buson>,<busoff>[,<dmaspeed>]]

3. /etc/lilo.conf can take the append = option, as discussed on page 320. This options passes options to the kernel as though you had typed them at the LILO: prompt. The equivalent lilo.conf line is

The above conguration is typical. Note that the second column of the IRQ listing shows the number of interrupts signals received from the device. Moving my mouse a little and listing the IRQs again gives me
3: 104851 XT-PIC serial

This is the most common way of giving kernel boot options. 4. The insmod and modprobe commands can take options that are passed to the module. These are vastly different from the way you pass options with append =. For instance, you can give options to a compiled-in Ethernet module with the commands
append = ether=9,0x300,0xd0000,0xd4000,eth0 append = ether=0,0,eth1

showing that several hundred interrupts were since received. Another useful entry is /proc/devices, which shows what major devices numbers were allocated and are being used. This le is extremely useful for seeing what peripherals are alive on your system.

42.5 Module Options and Device Conguration

Device modules often need information about their hardware conguration. For instance, ISA device drivers need to know the IRQ and I/O port that the ISA card is physically congured to access. This information is passed to the module as module options that the module uses to initialize itself. Note that most devices will not need options at all. PCI cards mostly autodetect; it is mostly ISA cards that require these options.

from within /etc/lilo.conf. But then, using modprobe on the same compiled-out modules, these options have to be specied like this:
modprobe wd irq=9 io=0x300 mem=0xd0000 mem_end=0xd4000 modprobe de4x5

42.5.1

Five ways to pass options to a module

1. If a module is compiled into the kernel, then the module will be initialized at boot time. lilo passes module options to the kernel from the command-line at the LILO: prompt. For instance, at the LILO: prompt, you can type See Section 4.4 :

Note that the 0xd0000,0xd4000 are only applicable to a few Ethernet modules and are usually omitted. Also, the 0s in ether=0,0,eth1 mean to try autodetect. To nd out what options a module will take, you can use the modinfo command which shows that the wd driver is one of the few Ethernet drivers where you can set their RAM usage. This has not been discussed, but cards can sometimes use
areas of memory directly.

linux aha1542=<portbase>[,<buson>,<busoff>[,<dmaspeed>]]

467

[root@cericon]# modinfo -p /lib/modules/<version>/net/wd.o ( [root@cericon]# modinfo -p /lib/modules/<version>/kernel/drivers/net/wd.o ) io int array (min = 1, max = 4) irq int array (min = 1, max = 4) mem int array (min = 1, max = 4) mem_end int array (min = 1, max = 4)

468

42. Kernel

42.5. Module Options and Device Conguration

42.6. Conguring Various Devices

42. Kernel

5. The le /etc/modules.conf Also sometimes called /etc/conf.modules, but now deprecated. contains default options for modprobe, instead of our giving them on the modprobe command-line. This is the preferred and most common way of giving module options. Our Ethernet example becomes:
alias eth0 wd alias eth1 de4x5 options wd irq=9 io=0x300 mem=0xd0000 mem_end=0xd4000

42.6

Conguring Various Devices

Further examples on getting common devices to work now follow but only a few devices are discussed. See the documentation sources above for more info. We concentrate here on what is normally done.

42.6.1 Sound and pnpdump

Plug-and-Play (PnP) ISA sound cards (like SoundBlaster cards) are possibly the more popular of any cards that people have gotten to work under L INUX . Here, we use the sound card example to show how to get a PnP ISA card working in a few minutes. This is, of course, applicable to cards other than sound. A utility called isapnp takes one argument, the le /etc/isapnp.conf, and congures all ISA Plug-and-Play devices to the IRQs and I/O ports specied therein. /etc/isapnp.conf is a complicated le but can be generated with the pnpdump utility. pnpdump outputs an example isapnp.conf le to stdout, which contains IRQ and I/O port values allowed by your devices. You must edit these to unused values. Alternatively, you can use pnpdump --config to get a /etc/isapnp.conf le with the correct IRQ, I/O port, and DMA channels automatically guessed from an examination of the /proc/ entries. This comes down to
[root@cericon]# pnpdump --config | grep -v $#.*\|$$ > /etc/isapnp.conf [root@cericon]# isapnp /etc/isapnp.conf

Having set up an /etc/modules.conf le allows module dynamic loading to take place. This means that the kernel automatically loads the necessary module whenever the device is required (as when ifconfig is rst used for Ethernet devices). The kernel merely tries an /sbin/modprobe eth0, and the alias line hints to modprobe to actually run /sbin/modprobe wd. Further, the options line means to run /sbin/modprobe wd irq=9 io=0x300 mem=0xd0000 mem end=0xd4000. In this way, /etc/modules.conf maps devices to drivers.

42.5.2

Module documentation sources

You might like to see a complete summary of all module options with examples of each of the ve ways of passing options. No such summary exists at this point, simply because there is no overall consistency and because people are mostly interested in getting one particular device to work, which will doubtless have peculiarities best discussed in a specialized document. Further, some specialized modules are mostly used in compiled-out form, whereas others are mostly used in compiled-in form. To get an old or esoteric device working, it is best to read the appropriate HOWTO documents: BootPrompt-HOWTO, Ethernet-HOWTO, and Sound-HOWTO. The device could also be documented in /usr/linux<version>/Documentation/ or under one of its subdirectories like sound/ and networking/. This is documentation written by the driver authors themselves. Of particular interest is the le /usr/src/linux/Documentation/networking/net-modules.txt, which, although outdated, has a fairly comprehensive list of networking modules and the module options they take. Another source of documentation is the driver code itself, as in the aha1542.c example above. It may explain the /etc/lilo.conf or /etc/modules.conf options to use but will often be quite cryptic. A driver is often written with only one of compiled-in or compiled-out support in mind (even though it really supports both). Choose whether to compile-in or compiled-out based on what is implied in the documentation or source. 469

Board 1 has Identity c9 00 00 ab fa 29 00 8c 0e: CTL0029 Serial No 44026 [checksum c9] CTL0029/44026[0]{Audio }: Ports 0x220 0x330 0x388; IRQ5 DMA1 DMA5 --- Enabled OK CTL0029/44026[1]{IDE }: Ports 0x168 0x36E; IRQ10 --- Enabled OK CTL0029/44026[2]{Game }: Port 0x200; --- Enabled OK

which gets any ISA PnP card congured with just two commands. Note that the /etc/isapnp.gone le can be used to make pnpdump avoid using certain IRQ and I/O ports. Mine contains
IO 0x378,2 IRQ 7

to avoid conicting with my parallel port. isapnp /etc/isapnp.conf must be run each time at boot and is probably already in your startup scripts. Now that your ISA card is enabled, you can install the necessary modules. You can read the /etc/isapnp.conf le and also isapnps output above to reference the I/O ports to the correct module options:
alias sound-slot-0 sb alias sound-service-0-0 sb alias sound-service-0-1 sb

470

42. Kernel

42.6. Conguring Various Devices

42. Kernel

alias sound-service-0-2 sb alias sound-service-0-3 sb alias sound-service-0-4 sb alias synth0 sb post-install sb /sbin/modprobe "-k" "adlib_card" options sb io=0x220 irq=5 dma=1 dma16=5 mpu_io=0x330 options adlib_card io=0x388 # FM synthesizer

means a card that should supply all of these. The post-install option means to run an additional command after installing the sb module; this takes care of the Adlib sequencer driver. I was tempted to try removing the post-install line and adding a alias
sound-service-0-1 adlib card. This works, but not if you run aumix before playmidi, **shrug**.

42.6.2 Parallel port

The parallel port module is much less trouble:
alias parport_lowlevel parport_pc options parport_lowlevel io=0x378 irq=7

Now run tail -f /var/log/messages /var/log/syslog, and then at another terminal type:
depmod -a modprobe sb

If you get no kernel or other errors, then the devices are working. Now we want to set up dynamic loading of the module. Remove all the sound and other modules with rmmod -a (or manually), and then try:
aumix

Merely make sure that your IRQ and I/O port match those in your CMOS (see Section 3.3), and that they do not conict with any other devices.

42.6.3 NIC Ethernet, PCI, and old ISA

Here I demonstrate non-PnP ISA cards and PCI cards, using Ethernet devices as an example. (NIC stands for Network Interface Card, that is, an Ethernet 10 or 100 Mb card.) For old ISA cards with jumpers, you will need to check your /proc/ les for unused IRQ and I/O ports and then physically set the jumpers. Now you can do a modprobe as usual, for example:
modinfo -p ne modprobe ne io=0x300 irq=9

You should get a kernel log like this:

Sep 24 00:45:19 cericon kernel: Soundblaster audio driver Copyright (C) by Hannu Savolainen 1993-1996 Sep 24 00:45:19 cericon kernel: SB 4.13 detected OK (240)

Then try:

playmidi <somefile>.mid

You should get a kernel log like this one:

kernel: Soundblaster audio driver Savolainen 1993-1996 kernel: SB 4.13 detected OK (240) kernel: YM3812 and OPL-3 driver Savolainen, Rob Hooft 1993-1996

lines:

Of course, for dynamic loading, your /etc/modules.conf le must have the

Sep 24 00:51:34 cericon Copyright (C) by Hannu Sep 24 00:51:34 cericon Sep 24 00:51:35 cericon Copyright (C) by Hannu

alias eth0 ne options ne io=0x300 irq=9

If you had to comment out the alias lines, then a kernel message like modprobe: Cant locate module sound-slot-0 would result. This indicates that the kernel is attempting a /sbin/modprobe sound-slot-0: a cue to insert an alias line. Actually, sound-service-0-0,1,2,3,4 are the /dev/mixer,sequencer,midi,dsp,audio devices, respectively. sound-slot-0 471

On some occasions you will come across a card that has software congurable jumpers, like PnP, but that can only be congured with a DOS utility. In this case compiling the module into the kernel will cause it to be autoprobed on startup without needing any other conguration. A worst case scenario is a card whose make is unknown, as well its IRQ and I/O ports. The chip number on the card can sometimes give you a hint (grep the kernel sources for this number), but not always. To get this card working, compile in support for 472

42. Kernel

42.6. Conguring Various Devices

42. Kernel

several modules, one of which the card is likely to be. Experience will help you make better guesses. If one of your guesses is correct, your card will almost certainly be discovered on reboot. You can nd its IRQ and I/O port values in /proc/ or you can run dmesg to see the autoprobe message line; the message will begin with eth0: . . . and contain some information about the driver. This information can be used if you decide later to use modules instead of your custom kernel. As explained, PCI devices almost never require IRQ or I/O ports to be given as options. As long as you have the correct module, a simple
modprobe <module>

and the same for. . . grep. . . Accton gave rtl8139.o and tulip.o (the former of which was correct), and for. . . grep. . . Macronix (or even 987) gave tulip.o, which hung the machine. I have yet to get that card working, although Eddie across the room claims he got a similar card working ne. Cards are cheapthere are enough working brands so that you dont have to waist your time on difcult ones.

42.6.4 PCI vendor ID and device ID

PCI supports the useful concept that every vendor and device have unique hex IDs. For instance, Intel has chosen to represent themselves by the completely random number 0x8086 as their vendor ID. PCI cards will provide their IDs on request. You will see numerical values listed in the output of lspci, scanpci, and cat /proc/pci, especially if the respective utility cannot look up the vendor name from the ID number. The le /usr/share/pci.ids (/usr/share/misc/pci.ids on Debian ) from the pciutils package contains a complete table of all IDs and their corresponding names. The kudzu package also has a table /usr/share/kudzu/pcitable containing the information we are really looking for: ID to kernel module mappings. This enables you to use the intended scientic method for locating the correct PCI module from the kernels /proc/pci data. The le format is easy to understand, and as an exercise you should try writing a shell script to do the lookup automatically.

will always work. Finding the correct module can still be a problem, however, because suppliers will call a card all sorts of marketable things besides the actual chipset it is compatible with. The utility scanpci (which is actually part of ) checks your PCI slots for PCI devices. Running scanpci might output something like:
. . . pci bus 0x0 cardnum 0x09 function 0x0000: vendor 0x1011 device 0x0009 Digital DC21140 10/100 Mb/s Ethernet pci bus 0x0 cardnum 0x0b function 0x0000: vendor 0x8086 device 0x1229 Intel 82557/8/9 10/100MBit network controller
10

pci bus 0x0 cardnum 0x0c function 0x0000: vendor 0x1274 device 0x1371 Ensoniq es1371

42.6.5 PCI and sound

The scanpci output just above also shows the popular Ensoniq sound card, sometimes built into motherboards. Simply adding the line
alias sound es1371

Another utility is lspci from the pciutils package, which gives comprehensive information where scanpci sometimes gives none. Then a simple script (kernel 2.4 paths in parentheses again),
for i in /lib/modules/<version>/net/* ; do strings $i \ | grep -q -i 21140 && echo $i ( for i in /lib/modules/<version>/kernel/drivers/net/* \ ; do strings $i | grep -q -i 21140 && echo $i ; for i in /lib/modules/<version>/net/* ; do strings $i \ | grep -q -i 8255 && echo $i ( for i in /lib/modules/<version>/kernel/drivers/net/* \ ; do strings $i | grep -q -i 8255 && echo $i ; ; done done ) ; done done )

to your modules.conf le will get this card working. It is relatively easy to nd the type of card from the card itselfEnsoniq cards actually have es1371 printed on one of the chips.

42.6.6 Commercial sound drivers

If your card is not listed in /usr/src/<version>/Documentation/sound/, then you might be able to get a driver from Open Sound http://www.opensound.com. If you still cant nd a driver, complain to the manufacturer by email.
There are a lot of sound (and other) cards whose manufacturers refuse to supply the Free software community with specs. Disclosure of programming information would enable L INUX users to buy their cards; Free software developers would produce a driver at no cost. Actually, manufacturers reasons are often just pig-headedness.

faithfully outputs three modules de4x5.o, eepro100.o, and tulip.o, of which two are correct. On another system lspci gave
. . . 00:08.0 Ethernet controller: Macronix, Inc. [MXIC] MX987x5 (rev 20) 00:0a.0 Ethernet controller: Accton Technology Corporation SMC2-1211TX (rev 10)

473

474

42. Kernel

42.6. Conguring Various Devices

42. Kernel

42.6.7

The ALSA sound project

this implemented in practice. Because many U NIX hardware platforms only support SCSI, it has become an integral part of U NIX operating systems. SCSIs also introduce the concept of LUNs (which stands for Logical Unit Number), Buses, and ID. These are just numbers given to each device in order of the SCSI cards you are using (if more than one), the SCSI cables on those cards, and the SCSI devices on those cablesthe SCSI standard was designed to support a great many of these. The kernel assigns each SCSI drive in sequence as it nds them: /dev/sda, /dev/sdb, and so on, so these details are usually irrelevant. An enormous amount should be said on SCSI, but the bare bones is that for 90% of situations, insmod <pci-scsi-driver> is all you are going to need. You can then immediately begin accessing the device through /dev/sd? for disks, /dev/st? for tapes, /dev/scd? for CD-ROMs, or /dev/sg? for scanners. Scanner user programs will have docs on what devices they access. SCSIs often also come with their own BIOS that you can enter on startup (like your CMOS). This will enable you to set certain things. In some cases, where your distribution compiles-out certain modules, you may have to load one of sd mod.o, st.o, sr mod.o, or sg.o, respectively. The core scsi mod.o module may also need loading, and /dev/ devices may need to be created. A safe bet is to run

The ALSA (Advanced Linux Sound Architecture http://www.alsa-project.org/) project aims to provide better kernel sound support. If your card is not supported by the standard kernel or you are not getting the most out of the standard kernel drivers, then do check this web site.

42.6.8

Multiple Ethernet cards

If you have more than one Ethernet card, you can easily specify both in your modules.conf le, as shown in Section 42.5 above. Modules compiled into the kernel only probe a single card (eth0) by default. Adding the line
append = "ether=0,0,eth1 ether=0,0,eth2 ether=0,0,eth3"

will cause eth1, eth2, and eth3 to be probed as well. Further, replacing the 0s with actual values can force certain interfaces to certain physical cards. If all your cards are PCI, however, you will have to get the order of assignment by experimentation. If you have two of the same card, your kernel may complain when you try to load the same module twice. The -o option to insmod species a different internal name for the driver to trick the kernel into thinking that the driver is not really loaded:
alias eth0 3c509 alias eth1 3c509 options eth0 -o 3c509-0 io=0x280 irq=5 options eth1 -o 3c509-1 io=0x300 irq=7
5

cd /dev ./MAKEDEV ./MAKEDEV ./MAKEDEV ./MAKEDEV

-v -v -v -v

sd st0 st1 st2 st3 scd0 scd1 scd2 scd3 sg

to ensure that all necessary device les exist in the rst place. It is recommended that you compile into your kernel support for your SCSI card (also called the SCSI Host Adapter) that you have, as well as support for tapes, CDROMs, etc. When your system next boots, everything will just autoprobe. An example system with a SCSI disk and tape gives the following at bootup:
(scsi0) <Adaptec AIC-7895 Ultra SCSI host adapter> found at PCI 0/12/0 (scsi0) Wide Channel A, SCSI ID=7, 32/255 SCBs (scsi0) Cables present (Int-50 YES, Int-68 YES, Ext-68 YES) (scsi0) Illegal cable configuration!! Only two (scsi0) connectors on the SCSI controller may be in use at a time! (scsi0) Downloading sequencer code... 384 instructions downloaded (scsi1) <Adaptec AIC-7895 Ultra SCSI host adapter> found at PCI 0/12/1 (scsi1) Wide Channel B, SCSI ID=7, 32/255 SCBs (scsi1) Downloading sequencer code... 384 instructions downloaded scsi0 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 5.1.28/3.2.4 <Adaptec AIC-7895 Ultra SCSI host adapter> scsi1 : Adaptec AHA274x/284x/294x (EISA/VLB/PCI-Fast SCSI) 5.1.28/3.2.4 <Adaptec AIC-7895 Ultra SCSI host adapter> scsi : 2 hosts. (scsi0:0:0:0) Synchronous at 40.0 Mbyte/sec, offset 8. Vendor: FUJITSU Model: MAE3091LP Rev: 0112

However, with the following two PCI cards that deception was not necessary:
alias eth0 rtl8139 alias eth1 rtl8139

42.6.9

SCSI disks

SCSI (pronounced scuzzy) stands for Small Computer System Interface. SCSI is a ribbon, a specication, and an electronic protocol for communication between devices and computers. Like your IDE ribbons, SCSI ribbons can connect to their own SCSI hard disks. SCSI ribbons have gone through some versions to make SCSI faster, the latest UltraWide SCSI ribbons are thin, with a dense array of pins. Unlike your IDE, SCSI can also connect tape drives, scanners, and many other types of peripherals. SCSI theoretically allows multiple computers to share the same device, although I have not seen 475

476

42. Kernel

42.6. Conguring Various Devices

42. Kernel

Type: Direct-Access ANSI SCSI revision: 02 Detected scsi disk sda at scsi0, channel 0, id 0, lun 0 (scsi0:0:3:0) Synchronous at 10.0 Mbyte/sec, offset 15. Vendor: HP Model: C1533A Rev: A708 Type: Sequential-Access ANSI SCSI revision: 02 Detected scsi tape st0 at scsi0, channel 0, id 3, lun 0 scsi : detected 1 SCSI tape 1 SCSI disk total. SCSI device sda: hdwr sector= 512 bytes. Sectors= 17826240 [8704 MB] [8.7 GB] . . . Partition check: sda: sda1 hda: hda1 hda2 hda3 hda4 hdb: hdb1

hda: FUJITSU MPE3084AE, ATA DISK drive hdb: CD-ROM 50X L, ATAPI CDROM drive hdd: Hewlett-Packard CD-Writer Plus 9300, ATAPI CDROM drive

Note that these devices should give BIOS messages before LILO: starts to indicate that they are correctly installed. The /etc/modules.conf lines to get the CD-writer working are:
# load sr_mod upon access of /dev/scd0 # SCSI hostadaptor emulation # Our normal IDE CD is on /dev/hdb

You should also check Section 31.5 to nd out how to boot SCSI disks when the needed module. . . is on a le system. . . inside a SCSI disk. . . that needs the module. For actually using a tape drive, see page 149.

42.6.10

SCSI termination and cooling

This is the most important section to read regarding SCSI. You may be used to IDE ribbons that just plug in and work. SCSI ribbons are not of this variety; they need to be impedance matched and terminated. These are electrical technicians terms. Basically, it means that you must use high-quality SCSI ribbons and terminate your SCSI device. SCSI ribbons allow many SCSI disks and tapes to be connected to one ribbon. Terminating means setting certain jumpers or switches on the last devices on the ribbon. It may also mean plugging the last cable connector into something else. Your adapter documentation and disk documentation should explain what to do. If you terminate incorrectly, everything may work ne, but you may get disk errors later in the life of the machine. Also note that some newer SCSI devices have automatic termination. Cooling is another important consideration. When the documentation for a disk drive recommends forced air cooling for that drive, it usually means it. SCSI drives get extremely hot and can burn out in time. Forced air cooling can mean as little as buying a cheap circuit box fan and tying it in a strategic position. You should also use very large cases with several inches of space between drives. Anyone who has opened up an expensive high end server will see the attention paid to air cooling.
5

The alias scd0 line must be omitted if sr mod is compiled into the kernelsearch your /lib/modules/<version>/ directory. Note that the kernel does not support ATAPI CD-Writers directly. The ide-scsi module emulates a SCSI adapter on behalf of the ATAPI CD-ROM. CD-Writer software expects to speak to /dev/scd?, and the ide-scsi module makes this device appear like a real SCSI CD writer. Real SCSI CD writers are much more expensive. There is one caveat: your ordinary IDE CD-ROM driver, ide-cd, will also want to probe your CD writer as if it were a normal CD-ROM. The ignore option makes the ide-cd module overlook any drives that should not be probedon this system, these would be the hard disk, CD writer, and non-existent secondary master. However, there is no way of giving an ignore option to a compiledin ide-cd module (which is how many distributions ship), so read on.

alias scd0 sr_mod alias scsi_hostadapter ide-scsi options ide-cd ignore="hda hdc hdd"

An alternative is to compile in support for ide-scsi and completely leave out support for ide-cd. Your normal CD-ROM will work perfectly as a read-only CDROM under SCSI emulation. This means setting the relevant Even with music CDs. sections of your kernel conguration menu:

<*> Enhanced IDE/MFM/RLL disk/cdrom/tape/floppy support < > Include IDE/ATAPI CDROM support <*> SCSI emulation support <*> SCSI support <*> SCSI CD-ROM support [*] Enable vendor-specific extensions (for SCSI CDROM) <*> SCSI generic support

No further conguration is needed, and on bootup, you will nd messages like:

42.6.11 CD writers
A system with an ATAPI (IDE CD message at bootup like, -Writer and ordinary CD-ROM will display a
5

scsi0 : SCSI host adapter emulation for IDE ATAPI devices scsi : 1 host. Vendor: E-IDE Model: CD-ROM 50X L Rev: 12 Type: CD-ROM ANSI SCSI revision: 02 Detected scsi CD-ROM sr0 at scsi0, channel 0, id 0, lun 0 Vendor: HP Model: CD-Writer+ 9300 Rev: 1.0b

477

478

42. Kernel

42.6. Conguring Various Devices

42.7. Modem Cards

42. Kernel

Type: CD-ROM ANSI SCSI revision: 02 Detected scsi CD-ROM sr1 at scsi0, channel 0, id 1, lun 0 scsi : detected 2 SCSI generics 2 SCSI cdroms total. sr0: scsi3-mmc drive: 4x/50x cd/rw xa/form2 cdda tray Uniform CD-ROM driver Revision: 3.10 sr1: scsi3-mmc drive: 32x/32x writer cd/rw xa/form2 cdda tray

To get these cards working requires the use of the setserial command. It is used to congure the kernels built-in serial driver. A typical example is an 8-port nonPnP ISA card with jumpers set to unused IRQ 5 and ports 0x1800x1BF. Note that unlike most devices, many serial devices can share the same IRQ. The reason is that serial
devices set an I/O port to tell which device is sending the interrupt. The CPU just checks every serial device whenever an interrupt comes in. The card is congured with this script:

If you do have a real SCSI writer, compiling in support for your SCSI card will detect it in a similar fashion. Then, for this example, the device on which to mount your CD-ROM is /dev/scd0 and your CD-Writer, /dev/scd1. , the cdrecord command-line program is simple For actually recording a CD and robust, although there are also many pretty graphical front ends. To locate your ID, run CD which will give a comma-separated numeric sequence. You can then use this sequence as the argument to cdrecords dev= option. On my machine I type
mkisofs -a -A Paul Sheer -J -L -r -P PaulSheer \ -p www.icon.co.za/psheer/ -o my_iso /my/directory cdrecord dev=0,1,0 -v speed=10 -isosize -eject my_iso cdrecord -scanbus
5

cd /dev/ ./MAKEDEV -v ttyS4 ./MAKEDEV -v ttyS5 ./MAKEDEV -v ttyS6 ./MAKEDEV -v ttyS7 ./MAKEDEV -v ttyS8 ./MAKEDEV -v ttyS9 ./MAKEDEV -v ttyS10 ./MAKEDEV -v ttyS11 /bin/setserial -v /dev/ttyS4 irq 5 port 0x180 uart 16550A skip_test /bin/setserial -v /dev/ttyS5 irq 5 port 0x188 uart 16550A skip_test /bin/setserial -v /dev/ttyS6 irq 5 port 0x190 uart 16550A skip_test /bin/setserial -v /dev/ttyS7 irq 5 port 0x198 uart 16550A skip_test /bin/setserial -v /dev/ttyS8 irq 5 port 0x1A0 uart 16550A skip_test /bin/setserial -v /dev/ttyS9 irq 5 port 0x1A8 uart 16550A skip_test /bin/setserial -v /dev/ttyS10 irq 5 port 0x1B0 uart 16550A skip_test /bin/setserial -v /dev/ttyS11 irq 5 port 0x1B8 uart 16550A skip_test

to create an ISO9660 CD-ROM out of everything below a directory /my/directory. This is most useful for backups. (The -a option should be omitted in newer versions of this command.) Beware not to exceed the speed limit of your CD writer.

42.6.12

Serial devices

You should immediately be able to use these devices as regular ports. Note that you would expect to see the interrupt in use under /proc/interrupts. For serial devices this is only true after data actually starts to ow. However, you can check /proc/tty/driver/serial to get more status information. The setserial man page contains more about different UARTs and their compatibility problems. It also explains autoprobing of the UART, IRQ, and I/O ports (although it is better to be sure of your card and never use autoprobing). Serial devices give innumerable problems. There is a very long Serial-HOWTO that will help you solve most of them; It goes into more technical detail. It will also explain special kernel support for many nonstandard cards.

You dont need to load any modules to get your mouse and modem to work. Regular serial devices (COM1 through COM4 under DOS/Windows) will autoprobe on boot and are available as /dev/ttyS0 through /dev/ttyS3. A message on boot, like
Serial driver version 4.27 with MANY_PORTS MULTIPORT SHARE_IRQ enabled ttyS00 at 0x03f8 (irq = 4) is a 16550A ttyS01 at 0x02f8 (irq = 3) is a 16550A

will testify to their correct detection.

42.7

Modem Cards

On the other hand, multiport serial cards can be difcult to congure. These devices are in a category all of their own. Most use a chip called the 16550A UART (Universal Asynchronous Receiver Transmitter), which is similar to that of your builtin serial port. The kernels generic serial code supports them, and you will not need a separate driver. The UART really is the serial port and comes in the avors 8250, 16450, 16550, 16550A, 16650, 16650V2, and 16750. 479

Elsewhere in this book I refer only to ordinary external modems that connect to your machines auxiliary serial port. However, internal ISA modem cards are cheaper and include their own internal serial port. This card can be treated as above, like an ISA multiport serial card with only one port: just set the I/O port and IRQ jumpers and then run setserial /dev/ttyS3. . . . Beware that a new variety of modem has been invented called the win-modem. These cards are actually just sound cards. Your operating system has to generate the 480

42. Kernel

42.8. More on LILO: Options

42.9. Building the Kernel

42. Kernel

signals needed to talk the same protocol as a regular modem. Because the CPU has to be very fast to do this, such modems were probably not viable before 1997 or so. http://linmodems.technion.ac.il/, http://www.idir.net/gromitkc/winmodem.html, and http://www.linmodems.org/ are three resources that cover these modems.
5

bzip2 -cd ../patch-2.4.0-test7.bz2 | patch -s -p1 cd .. mv linux-2.4.0-test6 linux-2.4.0-test7 ln -sf linux-2.4.0-test7 linux cd linux make mrproper

42.8 More on LILO: Options

The BootPrompt-HOWTO contains an exhaustive list of things that can be typed at the boot prompt to do interesting things like NFS root mounts. This document is important to read if only to get an idea of the features that L INUX supports.

Your 2.4.0-test6 kernel source tree is now a 2.4.0-test7 kernel source tree. You will often want to patch the kernel with features that Linus did not include, like security patches or commercial hardware drivers. Important is that the following include directories point to the correct directories in the kernel source tree:
[root@cericon]# ls -al /usr/include/{linux,asm} /usr/src/linux/include/asm lrwxrwxrwx 1 root root 24 Sep 4 13:45 /usr/include/asm -> ../src/linux/include/asm lrwxrwxrwx 1 root root 26 Sep 4 13:44 /usr/include/linux -> ../src/linux/include/linux lrwxrwxrwx 1 root root 8 Sep 4 13:45 /usr/src/linux/include/asm -> asm-i386

42.9 Building the Kernel

Summary:
cd /usr/src/linux/ make mrproper make menuconfig make dep make clean make bzImage make modules make modules_install cp /usr/src/linux/arch/i386/boot/bzImage /boot/vmlinuz-<version> cp /usr/src/linux/System.map /boot/System.map-<version>

Before continuing, you should read the Changes le (under /usr/src/linux/Documentation/) to nd out what is required to build the kernel. If you have a kernel source tree supplied by your distribution, everything will already be up-to-date.

42.9.2 Conguring
(A kernel tree that has suffered from previous builds may need you to run
make mrproper

Finally, edit /etc/lilo.conf and run lilo. Details on each of these steps follow.

42.9.1

Unpacking and patching

The L INUX kernel is available from various places as linux-?.?.?.tar.gz, but primarily from the L INUX kernels home ftp://ftp.kernel.org/pub/linux/kernel/. The kernel can easily be unpacked with

before anything else. This completely cleans the tree, as though you had just unpacked it.) There are three kernel conguration interfaces. The old line-for-line y/n interface is painful to use. For a better text mode interface, you can type
make menuconfig

and possibly patched with (see Section 20.7.3): 481

cd /usr/src mv linux linux-OLD tar -xzf linux-2.4.0-test6.tar.gz mv linux linux-2.4.0-test6 ln -s linux-2.4.0-test6 linux cd linux

enter

otherwise, under
make xconfig

to get the graphical congurator. For this discussion, I assume that you are using the text-mode interface. 482

42. Kernel

42.10. Using Packaged Kernel Source

42.11. Building, Installing

42. Kernel

The congure program enables you to specify an enormous number of features. It is advisable to skim through all the sections to get a feel for the different things you can do. Most options are about specifying whether you want a feature [*] compiled into the kernel image, [M] compiled as a module, or [ ] not compiled at all. You can also turn off module support altogether from Loadable module support --->. The kernel conguration is one L INUX program that offers lots of helpselect < Help > on any feature. The raw help le is /usr/src/linux/Documentation/Configure.help (nearly 700 kilobytes) and is worth reading. When you are satised with your selection of options, select < Exit > and select save your new kernel configuration. The kernel conguration is saved in a le /usr/src/linux/.config. Next time you run make menuconfig, your conguration will default to these settings. The le /usr/src/linux/arch/i386/defconfig contains defaults to use in the absence of a .config le. Note that the command make mrproper removes the .config le.

make make make make make dep && \ clean && \ bzImage && \ modules && \ modules_install

The command make modules install would have installed all modules into /lib/modules/<version>. You may like to clear out this directory at some point and rerun
make modules install, since stale modules cause problems with depmod -a.

The kernel image itself, /usr/src/linux/arch/i386/boot/bzImage, and /usr/src/linux/System.map are two other les produced by the build. These must be copied to /boot/, possibly creating neat symlinks:
cp cp ln ln /usr/src/linux/arch/i386/boot/bzImage /boot/vmlinuz-<version> /usr/src/linux/System.map /boot/System.map-<version> -sf System.map-<version> /boot/System.map -sf /boot/vmlinuz-<version> vmlinuz

42.10 Using Packaged Kernel Source

Your distribution will probably have a kernel source package ready to build. This package is better to use than downloading the source yourself because all the default build options will be present; for instance, RedHat 7.0 comes with the le /usr/src/linux-2.2.16/configs/kernel-2.2.16-i586smp.config, which can be copied over the /usr/src/linux-2.2.16/.config to build a kernel optimized for SMP (Symmetric Multiprocessor Support) with all of RedHats defaults enabled. It also comes with a custom defconfig le to build kernels identical to those of RedHat. Finally, RedHat would have applied many patches to add features that may be time consuming to do yourself. The same goes for Debian . You should try to enable or compile-in features rather than disable anything, since the default RedHat kernel supports almost every kernel feature, and later it may be more convenient to have left it that way. On the other hand, a minimal kernel will compile much faster.

Finally, your lilo.conf may be edited as described in Chapter 31. Most people now forget to run lilo and nd their system unbootable. Do run lilo, making sure that you have left your old kernel in as an option, in case you need to return to it. Also make a boot oppy from your kernel, as shown in Section 31.4.

42.11 Building, Installing

Run the following commands to build the kernel; this process may take anything from a few minutes to several hours, depending on what you have enabled. After each command completes, check the last few messages for errors (or check the return code, $?), rather than blindly typing the next commands. 483 484

43.1. The X Protocol

43. The X Window System

needs to know is that if it sends a particular sequence of bytes down the TCP/IP link, the server will interpret them to mean that a line, circle, font, box, or other graphics entity should be drawn on its screen. In the other direction, the application needs to know that particular sequences of bytes mean that a keyboard key was pressed or that a mouse has moved. This TCP communication is called the X protocol.

Chapter 43

The X Window System

Before The X Window System (from now on called ), U NIX was terminal based and had no proper graphical environment, sometimes called a GUI. Graphical User Interface. was designed to fulll that need and to incorporate into graphics all the power of a networked computer.

When you are using , you will probably not be aware that this interaction is happening. The server and the application might very well be on the same machine. The real power of is evident when they are not on the same machine. Consider, for example, that 20 users can be logged in to a single machine and be running different programs that are displayed on 20 different remote servers. It is as though a single machine was given multiple screens and keyboards. It is for this reason that is called a network transparent windowing system. The developer of a graphical application can then dispense with having to know anything about the graphics hardware itself (consider DOS applications where each had to build in support for many different graphics cards), and that developer can also dispense with having to know what machine the graphics will be displayed on. The precise program that performs this miracle is /usr/X11/bin/X. A typical sequence of events to get a graphical program to run is as follows. (This is an illustration. In practice, numerous utilities perform these functions in a more generalized and user-friendly way.) 1. The program /usr/X11R6/bin/X is started and run in the background. will detect through conguration les (/etc/XF86Config or /etc/X11/XF86Config on L INUX ), and possibly through hardware autodetection, what graphics hardware (like a graphics add-on card) is available. It then initializes that hardware into graphics mode. 2. It then opens a socket connection to listen for incoming requests on a specic port (usually TCP port 6000), being ready to interpret any connection as a stream of graphics commands. 3. An application is started on the local machine or on a remote machine. All programs have a conguration option by which you can specify (with an IP address or host name) where you would like the program to connect, that is, on which server you would like the resulting output to display. 4. The application opens a socket connection to the specied server over the network. This is the most frequent source of errors. Applications fail to connect to a server because the server is not running, because the server was specied incorrectly, or because the server refuses a connection from an untrusted host. 5. The application begins sending protocol requests, waiting for them to be proprotocol responses. cessed, and then receiving and processing the resulting From the users point of view, the application now appears to be running on the servers display. 486

was developed in 1985 at the Massachusetts Institute of Technology by the X Consortium and is now owned by the Open Software Foundation (OSF). It comprises over 2 million lines of code that run on every variant of U NIX. You might imagine that allowing an application to put graphics on a screen involves nothing more than creating a user library that can perform various graphical functions like line drawing, font drawing, and so on. To understand why is more than merely this, consider the example of character terminal applications: these are programs that run on a remote machine while displaying to a character terminal and receiving feedback (keystrokes) from that character terminal. There are two distinct entities at workthe application and the users character terminal display; these two are connected by some kind of serial or network link. Now what if the character terminal could display windows and other graphics (in addition to text), while giving feedback to the application with a mouse (as well as a keyboard)? This is what achieves.

43.1 The X Protocol

is a protocol of commands that are sent and received between an application and a special graphical terminal called an X Server (from now on called the server). The word
server is confusing, because there are lots of servers for each client machine, and the user sits on the server side. This is in the opposite sense to what we usually mean by a server. How the server actu-

ally draws graphics on the hardware is irrelevant to the developer; all the application 485

43. The X Window System

43.1. The X Protocol

43. The X Window System

Communication between the application and the server is somewhat more complex than the mere drawing of lines and rectangles and reporting of mouse and key events. The server must be able to handle multiple applications connecting from multiple machines, and these applications may interact with each other (think of cut and paste operations between applications that are actually running on different machines.) Some examples of the fundamental X Protocol requests that an application can make to a server are the following:
5

waits for a key press or mouse click before exiting. You can compile it with gcc o splash splash.c -lX11 -L/usr/X11R6/lib. (You can see right away why there are few applications written directly in .) Notice that all library functions are prexed by an X.
/* splash.c - display an image */ #include <stdlib.h> #include <stdio.h> #include <string.h> #include <X11/Xlib.h> /* XPM */ static char *graham_splash[] = { /* columns rows colors chars-per-pixel */ "28 32 16 1", " c #34262e", ". c #4c3236", "X c #673a39", "o c #543b44", "O c #724e4e", "+ c #6a5459", "@ c #6c463c", "# c #92706c", "$ c #92685f", "% c #987e84", "& c #aa857b", "n c #b2938f", "= c #bca39b", "- c #a89391", "; c #c4a49e", ": c #c4a8a4", /* pixels */ "--%#%%nnnn#-nnnnnn=====;;=;:", "--------n-nnnnnn=n==;==;=:;:", "----n--n--n-n-n-nn===:::::::", "-----&------nn-n=n====::::::", "----------------n===;=::::::", "----%&-%--%##%---n===:::::::", "------%#%+++o+++----=:::::::", "--#-%%#+++oo. oo+#--=:::::::", "-%%%%++++o.. .++&-==:::::", "---%#+#+++o. oo+&n=::::", "--%###+$+++Oo. o+#-:=::", "-&%########++Oo @$-==:", "####$$$+###$++OX .O+&==", "&##$O+OXo+++$#+Oo. ..O&&-", "&##+OX..... .oOO@@... o@+&&", "&###$Oo.o++ ..oX@oo@O$&-", "n###$$$$O$o ...X.. .XXX@$$$&", "nnn##$$#$OO. .XX+@ .XXX@$$#&", "nnn&&%####$OX.X$$@. XX$$$$&", "nnnnn&&###$$$OX$$X..XXX@O$&n", "nnnnnn&&%###$$$$@XXXXX@O$&&n", ";n=;nnnn&&&#$$$$$@@@@@@O$&n;", ";n;=nn;nnnn#&$$$@X@O$@@$$&n;", "=n=;;;n;;nn&&&$$$$OO$$$$$&;;", "n;=n;;=nn&n&&&&&&$$$$$##&&n;", "n;=;;;;;;;;&&&n&&&&&&&&#&n=;", ";n;n;;=n;&;&;&n&&&&&&&#nn;;;", "n;=;;;;;;;;n;&&n&&&n&nnnn;;;", "n=;;:;;=;;nn;&n;&n&nnnnnnn=;", "nn;;;;;;;;;;;;;;n&nnnnnn===;", "=nn;;:;n;;;;&&&&n&&nnnnnn;=;", "n====;;;;&&&&&&&nnnnnnnnnn;;" }; int main (int argc, char **argv) { int i, j, x, y, width, height, n_colors; XSetWindowAttributes xswa; XGCValues gcv; Display *display; char *display_name = 0; int depth = 0; Visual *visual; Window window; Pixmap pixmap; XImage *image; Colormap colormap; GC gc; int bytes_per_pixel; unsigned long colors[256]; unsigned char **p, *q; for (i = 1; i < argc - 1; i++) if (argv[i]) if (!strcmp (argv[i], "-display")) display_name = argv[i + 1]; display = XOpenDisplay (display_name); if (!display) { printf ("splash: cannot open display\n");

Create Window A window is a logical rectangle on the screen, owned by particular application, into which graphics can be drawn. List Fonts To list fonts available to the application. Allocate Color Will dene a color of the specied name or RGB value for later use. Create Graphics Context A Graphics Context is a denition of how graphics are to be drawn within a windowfor example, the default background color, line style, clipping, and font. Get Selection Owner Find which window (possibly belonging to another application) owns the selection (i.e., a cut of text). In return, the server replies by sending events back to the application. The application is required to constantly poll the server for these events. Besides events detailing the users mouse and keyboard input, there are other events, for example, that indicate that a window has been exposed (a window on top of another window was moved, thus exposing the window beneath it. The application should then send the appropriate commands needed to redraw the graphics within the window now on top). Another example is a notication to request a paste from another application. The le /usr/include/X11/Xproto.h contains the full list of protocol requests and events. The programmer of an application need not be directly concerned with these requests. A high-level library handles the details of the server interaction. This library is called the X Library, /usr/X11R6/lib/libX11.so.6. One of the limitations of such a protocol is that developers are restricted to the set of commands that have been dened. overcame this problem by making the protocol extensible Being able to add extensions and enhancements without complicating or breaking from the start. These days there are extensions to to allow, for excompatibility. ample, the display of 3D graphics on the server, the interpretation of PostScript commands, and many other capabilities that improve aesthetic appeal and performance. Each extension comes with a new group of protocol requests and events, as well as a programmers library interface.
15 10

An example of a real program follows. This is about the simplest an program is ever going to get. The program displays a small XPM image le in a window and 487

488

43. The X Window System

43.1. The X Protocol

43. The X Window System

exit (1); } depth = DefaultDepth (display, DefaultScreen (display)); visual = DefaultVisual (display, DefaultScreen (display)); p = (unsigned char **) graham_splash; q = p[0]; width = atoi ((const char *) q); q = (unsigned char *) strchr (q, ); height = atoi ((const char *) ++q); q = (unsigned char *) strchr (q, ); n_colors = atoi ((const char *) ++q); colormap = DefaultColormap (display, DefaultScreen (display)); pixmap = XCreatePixmap (display, DefaultRootWindow (display), width, height, depth); gc = XCreateGC (display, pixmap, 0, &gcv); image = XCreateImage (display, visual, depth, ZPixmap, 0, 0, width, height, 8, 0); image->data = (char *) malloc (image->bytes_per_line * height + 16); /* create color palette */ for (p = p + 1, i = 0; i < n_colors; p++, i++) { XColor c, c1; unsigned char *x; x = *p + 4; if (*x == #) { unsigned char *h = (unsigned char *) "0123456789abcdef"; x++; c.red = ((unsigned long) strchr (h, *x++) (unsigned long) h) << 12; c.red |= ((unsigned long) strchr (h, *x++) (unsigned long) h) << 8; c.green = ((unsigned long) strchr (h, *x++) (unsigned long) h) << 12; c.green |= ((unsigned long) strchr (h, *x++) (unsigned long) h) << 8; c.blue = ((unsigned long) strchr (h, *x++) (unsigned long) h) << 12; c.blue |= ((unsigned long) strchr (h, *x++) (unsigned long) h) << 8; if (!XAllocColor (display, colormap, &c)) printf ("splash: could not allocate color cell\n"); } else { if (!XAllocNamedColor (display, colormap, (char *) x, &c, &c1)) printf ("splash: could not allocate color cell\n"); } colors[(*p)[0]] = c.pixel; } bytes_per_pixel = image->bytes_per_line / width;

125

130

135

140

145

150

155

160

100

165

105

170

110

175

115

180

120

/* cope with servers having different byte ordering and depths */ for (j = 0; j < height; j++, p++) { unsigned char *r; unsigned long c; q = image->data + image->bytes_per_line * j;

185

r = *p; if (image->byte_order == MSBFirst) { switch (bytes_per_pixel) { case 4: for (i = 0; i < width; i++) { c = colors[*r++]; *q++ = c >> 24; *q++ = c >> 16; *q++ = c >> 8; *q++ = c; } break; case 3: for (i = 0; i < width; i++) { c = colors[*r++]; *q++ = c >> 16; *q++ = c >> 8; *q++ = c; } break; case 2: for (i = 0; i < width; i++) { c = colors[*r++]; *q++ = c >> 8; *q++ = c; } break; case 1: for (i = 0; i < width; i++) *q++ = colors[*r++]; break; } } else { switch (bytes_per_pixel) { case 4: for (i = 0; i < width; i++) { c = colors[*r++]; *q++ = c; *q++ = c >> 8; *q++ = c >> 16; *q++ = c >> 24; } break; case 3: for (i = 0; i < width; i++) { c = colors[*r++]; *q++ = c; *q++ = c >> 8; *q++ = c >> 16; } break; case 2: for (i = 0; i < width; i++) { c = colors[*r++]; *q++ = c; *q++ = c >> 8; } break; case 1: for (i = 0; i < width; i++) *q++ = colors[*r++]; break; } } }

489

490

43. The X Window System

43.2. Widget Libraries and Desktops

43. The X Window System

190

XPutImage (display, pixmap, gc, image, 0, 0, 0, 0, width, height); x = (DisplayWidth (display, DefaultScreen (display)) - width) / 2; y = (DisplayHeight (display, DefaultScreen (display)) - height) / 2;
195

xswa.colormap = colormap; xswa.background_pixmap = pixmap; window = XCreateWindow (display, DefaultRootWindow (display), x, y, width, height, 0, depth, InputOutput, visual, CWColormap | CWBackPixmap, &xswa); XSelectInput (display, window, KeyPressMask | ButtonPressMask); XMapRaised (display, window); while (1) { XEvent event; XNextEvent (display, &event); if (event.xany.type == KeyPress || event.xany.type == ButtonPressMask) break; } XUnmapWindow (display, window); XCloseDisplay (display); return 0;

Motif (libM.so) is a modern, full-featured widget library that had become an industry standard. Motif is, however, bloated, slow, and dependent on the toolkit. It has always been an expensive proprietary library. Tk (tee-kay, libtk.so) is a library that is primarily used with the Tcl scripting language. It was probably the rst platform-independent library (running on Windows, all U NIX variants, and the Apple Mac). It is, however, slow and has limited features (this is progressively changing). Both Tcl and Motif are not very elegant-looking.
fact used to do the diagrams in this book.

200

205

Around 1996, we saw a lot of widget libraries popping up with different licenses. V , xforms, and graphix come to mind. (This was when I started to write coolwidgetsmy own widget library.) There was no efcient, multipurpose, Free, and elegant-looking widget library for U NIX. This was a situation that sucked and was retarding Free software development.

210

43.2.2 Qt
At about that time, a new GUI library was released. It was called Qt and was developed by Troll Tech. It was not free, but it was an outstanding technical accomplishment in that it worked efciently and cleanly on many different platforms. It was shunned by some factions of the Free software community because it was written in C++, Which
is not considered to be the standard development language by the Free Software Foundation because it is and was only free for noncommercial not completely portable and possibly for other reasons.

215

You can learn to program from the documentation in the Window System sourcessee below. The preceding program is said to be written directly in X-lib because it links only with the lowest-level library, libX11.so. The advantage of developing this way is that your program will work across every variant of U NIX without any modications. Notice also that the program deals with any type of display device regardless of its resolution (width height or pixels-per-inch), color capacity, or hardware design.

applications to link with.

Nevertheless, advocates of Qt went ahead and began producing the outstanding KDE desktop projecta set of higher-level development libraries, a window manager, and many core applications that together make up the KDE Desktop. The licensing issues with Qt have relaxed somewhat, and it is now available under both the GPL and a proprietary license.

43.2 Widget Libraries and Desktops

To program in is tedious. Therefore, most developers will use a higher-level widget library. Most users of GUIs will be familiar with widgets: buttons, menus, text input programmers have to implement these manually. The reason boxes, and so on. protocol is to allow different user interfaces to be widgets were not built into the built on top of . This exibility makes the enduring technology that it is.

43.2.3 Gtk
At one point, before KDE was substantially complete, Qt antagonists reasoned that since there were more lines of Qt code than of KDE code, it would be better to develop a widget library from scratchbut that is an aside. The Gtk widget library was written especially for gimp (GNU Image Manipulation Program), is GPLd and written entirely in in low-level calls (i.e., without the X Toolkit), object oriented, fast, clean, extensible and having a staggering array of features. It comprises Glib, a library meant to extend standard , providing higher-level functions usually akin only to scripting languages, like hash tables and lists; Gdk, a wrapper around raw Library to give GNU naming conventions to , and to give a slightly higher level interface to ; and the Gtk library itself. Using Gtk, the Gnome project began, analogous to KDE, but written entirely in . 492

43.2.1

Background

The X Toolkit (libXt.so) is a widget library that has always come free with . It is crude-looking by todays standards. It doesnt feature 3D (shadowed) widgets, although it is comes free with . The excellent xfig application, an X Toolkit application, was in 491

43. The X Window System

43.3. XFree86

43. The X Window System

43.2.4

GNUStep

To switch to the L INUX 2.7).

console, use

. The seven common virtual consoles of terminal (as explained in Section

OpenStep (based on NeXTStep) was a GUI specication published in 1994 by Sun Microsystems and NeXT Computers, meant for building applications. It uses the Objective-C language, which is an object-oriented extension to , that is arguably more suited to this kind of development than is C++. OpenStep requires a PostScript display engine that is analogous to the protocol, because all graphics are independent of the pixel but it is considered superior to resolution of the screen. In other words, high-resolution screens would improve the picture quality without making the graphics smaller. The GNUStep project has a working PostScript display engine and is meant as a Free replacement to OpenStep.

are 16 as text terminals, and 7 as an

To zoom in or out of your

session, use

and

43.3.2 Running X utilities

/usr/X11R6/bin/ contains a large number of utilities that most other operating systems have based theirs on. Most of these begin with an x. The basic XFree86 programs are:
SuperProbe X XFree86 Xmark Xprt Xwrapper appres atobm bdftopcf beforelight bitmap bmtoa dga editres fsinfo fslsfonts fstobdf iceauth ico lbxproxy listres lndir makepsres makestrs mergelib mkcfm mkdirhier mkfontdir oclock pcitweak proxymngr resize revpath rstart rstartd scanpci sessreg setxkbmap showfont showrgb smproxy startx twm viewres x11perf x11perfcomp xauth xbiff xcalc xclipboard xclock xcmsdb xconsole xcutsel xditview xdm xdpyinfo xedit xev xeyes xf86config xfd xfindproxy xfontsel xfs xfwp xgamma xgc xhost xieperf xinit xkbbell xkbcomp xkbevd xkbprint xkbvleds xkbwatch xkill xload xlogo xlsatoms xlsclients xlsfonts xmag xman xmessage xmodmap xon xprop xrdb xrefresh xset xsetmode xsetpointer xsetroot xsm xstdcmap xterm xvidtune xwd xwininfo xwud

43.3 XFree86
was developed by the X Consortium as a standard as well as a reference implementation of that standard. There are ports to every platform that supports graphics. The current version of the standard is 11 release 6 (hence the directory /usr/X11R6/). There will probably never be another version. XFree86 http://www.xfree86.org/ is a free port of that includes L INUX Intel machines among its supported hardware. has some peculiarities that are worth noting if you are a Windows user, and XFree86 has some over those. XFree86 has its own versioning system beneath the 11R6 as explained below.

43.3.1

Running X and key conventions

). To run an program, you need to tell the program what remote server to connect to. Most programs take the option -display to specify the server. With running in your seventh virtual console, type into your rst virtual console:
xterm -display localhost:0.0

(See Section 43.6 for conguring

At a terminal prompt, you can type:

(provided is not already running). If you have congured properly to start (including putting /usr/X11R6/bin in your PATH), then this command will initiate the graphics hardware and a black-and-white stippled background will appear with a single as the mouse cursor. Contrary to intuition, this means that is actually working properly. To kill the server, use the key combination 493 ... . .

localhost refers to the machine on which the server is runningin this case, our own. The rst 0 means the screen we want to display on ( supports multiple physical screens in its specication). The second 0 refers to the root window we want to display on. Consider a multiheaded For example, two adjacent monitors that behave as one continuous screen. display: we would like to specify which monitor the application pops up on.

While xterm is running, switching to your minal where you can type commands.

session will reveal a character ter-

To switch to the text console, use

A better way to specify the display is to use the DISPLAY environment variable: 494

43. The X Window System

43.3. XFree86

43. The X Window System

DISPLAY=localhost:0.0 export DISPLAY

applications to display to localhost:0.0, although a causes all subsequent display on the command-line takes rst priority. utilities listed above are pretty ugly and unintuitive. Try, for example, The xclock, xcalc, and xedit. For fun, try xbill. Also run
rpm -qa | grep x

buttons, as well as possibly a task bar and a Start button that you may be familiar with. A window manager is just another application that has the additional task of managing the positions of basic applications on your desktop. Window managers executables are usually sufxed by a wm. If you dont have icewm, the minimalists twm window manager will almost always be installed. Clicking on the background is common convention of user interfaces. Different mouse buttons may bring up a menu or a list of actions. It is often analogous to a Start button. An enormous amount of religious attention is given to window managers. There are about 20 useful choices to date. Remember that any beautiful graphics are going to irritate you after you sit in front of the computer for a few hundred hours. You also dont want a window manager that eats too much memory or uses too much space on the screen.

43.3.3

Running two X sessions

server on your machine. The command session in the virtual console 8. You can switch to it by using . server within your current display:

You can start up a second

/usr/X11R6/bin/X :1

starts up a second or

43.3.5 X access control and remote display

The way we described an server may leave you wondering if anyone on the Internet can start an application on your display. By default, prohibits access from all machines except your own. The xhost command enables access from particular machines. For instance, you can run xhost +192.168.5.7 to allow that host to display to your machine. The command xhost + completely disables access control. A typical procedure is the running of an application on a remote machine to a local machine. A sample session follows:
[psheer@divinian]# xhost +192.168.3.2 192.168.3.2 being added to access control list [psheer@divinian]# ifconfig | grep inet inet addr:192.168.3.1 Bcast:192.168.3.255 Mask:255.255.255.0 inet addr:127.0.0.1 Mask:255.0.0.0 [psheer@divinian]# telnet 192.168.3.2 Trying 192.168.3.2... Connected to 192.168.3.2. Escape character is ]. Debian GNU/Linux 2.2 cericon cericon login: psheer Password: Last login: Fri Jul 13 18:46:43 2001 from divinian on pts/1 [psheer@cericon]# export DISPLAY=192.168.3.1:0.0 [psheer@cericon]# nohup rxvt & [1] 32573 nohup: appending output to nohup.out [psheer@cericon]# exit Connection closed by foreign host.

You can also start up a second

/usr/X11R6/bin/Xnest :1 &

A smaller server that uses a subwindow as a display device will be started. You can easily create a third server within that, ad innitum. or To get an application to display to this second server, use, as before,
10 5

DISPLAY=localhost:1.0 export DISPLAY xterm

xterm -display localhost:1.0

43.3.4

Running a window manager

Manually starting and then running an application is not the way to use . We want a window manager to run applications properly. The best window manager available (sic) is icewm, available from icewm.cjb.net http://icewm.cjb.net/. Window managers en, , and close each application inside a resizable bounding box and give you the 495

496

43. The X Window System

43.4. The X Distribution

43.5. X Documentation

43. The X Window System

43.3.6

X selections, cutting, and pasting

43.5.1 Programming
All the books describing all of the programming APIs are included inside the distribution. Most of these are of specialized interest and will not be including in your distribution by defaultdownload the complete distribution if you want them. You can then look inside xc/doc/specs (especially xc/doc/specs/X11) to begin learning how to program under . Debian also comes with the xbooks package, and RedHat with the XFree86doc packages.

Start an xterm to demonstrate the following mouse operations. predates the cut-and-paste conventions of Windows and the Mac. requires a three-button mouse, although pushing the two outer buttons simultaneously is equivalent to pushing the middle button. That is, provided X has been congured for thissee the Emulate3Buttons option in the conguration le example below. Practice the following:

Dragging the left mouse button is the common way to select text. This automatically places the highlighted text into a cut buffer, also sometimes called the clipboard. Dragging the right mouse button extends the selection, that is, enlarges or reduces the selection. Clicking the middle mouse button pastes the selection. Note that ally unusable without the capability of pasting in this way. becomes virtu-

43.5.2 Conguration documentation

Important to conguring is the directory /usr/X11R6/lib/X11/doc/ or /usr/share/doc/xserver-common/. It may contain, for example,
AccelCards.gz Devices.gz Monitors.gz QuickStart.doc.gz README.3DLabs.gz README.Config.gz README.DGA.gz README.Debian README.I128.gz README.Linux.gz README.MGA.gz README.Mach32.gz README.Mach64.gz README.NVIDIA.gz README.Oak.gz README.P9000.gz README.S3.gz README.S3V.gz README.SiS.gz README.Video7.gz README.W32.gz README.WstDig.gz README.agx.gz README.apm.gz README.ark.gz README.ati.gz README.chips.gz README.cirrus.gz README.clkprog.gz README.cyrix.gz README.epson.gz README.fbdev.gz README.gz README.i740.gz README.i810.gz README.mouse.gz README.neo.gz README.r128.gz README.rendition.gz README.trident.gz README.tseng.gz RELNOTES.gz changelog.Debian.gz copyright examples xinput.gz

Modern Gtk and Qt applications have tried to retain compatibility with these mouse conventions.

43.4 The X Distribution

The ofcial distribution comes as an enormous source package available in tgz format at http://www.xfree86.org/. It is traditionally packed as three tgz les to be unpacked over each otherthe total of the three is about 50 megabytes compressed. This package has nothing really to do with the version number X11R6it is a subset of X11R6. Downloading and installing the distribution is a major undertaking, but you should do it if you are interested in development. All U NIX distributions come with a compiled and (mostly) congured installation; hence, the ofcial distribution should never be needed except by developers.

As you can see, there is documentation for each type of graphics card. Learning how to congure is a simple matter of reading the QuickStart guide and then checking the specics for your card.

43.5.3 XFree86 web site

Any missing documentation can be found on the XFree86 http://www.xfree86.org/ web site. New graphics cards are coming out all the time. XFree86 http://www.xfree86.org/ contains FAQs about cards and the latest binaries, should you not be able to get your card working from the information below. Please always search the XFree86 web site for information on your card and for newer releases before reporting a problem. 498

43.5 X Documentation
The Window System comes with tens of megabytes of documentation. 497

43. The X Window System

43.6. X Conguration

43. The X Window System

43.6 X Conguration
Conguring involves editing XFree86s conguration le /etc/X11/XF86Config. Such a le may have been produced at installation time but will not always be correct. You will hence frequently nd yourself having to make manual changes to get running in full resolution. Note that XFree86 has a slightly different conguration le format for the new version 4. Differences are explained below.
35

43.6.1

Simple 16-color X server

The documentation discussed above is a lot to read. The simplest possible way working is to determine what mouse you have, and then create a le, to get /etc/X11/XF86Config (back up your original) containing the following. Adjust the "Pointer" section for your correct Device and Protocol. If you are running version 3.3, you should also comment out the Driver "vga" line. You may also have to switch the line containing 25.175 to 28.32 for some laptop displays.
Section "Files" RgbPath "/usr/X11R6/lib/X11/rgb" FontPath "/usr/X11R6/lib/X11/fonts/misc/" EndSection Section "ServerFlags" EndSection Section "Keyboard" Protocol "Standard" AutoRepeat 500 5 XkbDisable XkbKeymap "xfree86(us)" EndSection Section "Pointer" # Protocol "Busmouse" # Protocol "IntelliMouse" # Protocol "Logitech" Protocol "Microsoft" # Protocol "MMHitTab" # Protocol "MMSeries" # Protocol "MouseMan" # Protocol "MouseSystems" # Protocol "PS/2" Device "/dev/ttyS0" # Device "/dev/psaux" Emulate3Buttons Emulate3Timeout 150 EndSection Section "Monitor" Identifier "My Monitor" VendorName "Unknown" ModelName "Unknown"

HorizSync 31.5 - 57.0 VertRefresh 50-90 Modeline "640x480" 28.32 640 664 760 800 Modeline "640x480" 25.175 640 664 760 800 EndSection Section "Device" Identifier "Generic VGA" VendorName "Unknown" BoardName "Unknown" Chipset "generic" # Driver "vga" Driver "vga" EndSection Section "Screen" Driver "vga16" Device "Generic VGA" Monitor "My Monitor" Subsection "Display" Depth 4 Modes "640x480" Virtual 640 480 EndSubsection EndSection #

480 491 493 525 480 491 493 525

You can then start

. For XFree86 version 3.3, run

/usr/X11R6/bin/XF86_VGA16 -cc 0

or for XFree86 version 4, run

/usr/X11R6/bin/XFree86 -cc 0

Both of these will print out a status line containing clocks: . . . conrming This is the speed, in Megahertz, that pixels whether your choice of 25.175 was correct.
can come from your card and is the only variable to conguring a 16-color display.

You should now have a working gray-level display that is actually almost usable. It has the advantage that it always works.

43.6.2 Plug-and-Play operation

XFree86 version 4 has Plug-and-Play support. Simply run
/usr/X11R6/bin/XFree86 -configure

to produce a working XF86Config le. You can copy this le to /etc/X11/XF86Config and immediately start running . However, the le you get may be less than optimal. Read on for detailed conguration. 500

499

43. The X Window System

43.6. X Conguration

43. The X Window System

43.6.3

Proper X conguration

A simple and reliable way to get working is given by the following steps (if this fails, then you will have to read some of the documentation described above). There is also a tool called Xconfigurator which provides a user-friendly graphical front-end. 1. Back up your /etc/X11/XF86Config to /etc/X11/XF86Config.ORIG. 2. Run SuperProbe at the character console. It will blank your screen and then spit out what graphics card you have. Leave that information on your screen and switch to a different virtual terminal. If SuperProbe fails to recognize your card, it usually means that XFree86 will also fail.
5

These are timing settings for different monitors and screen resolutions. Choosing one that is too fast could blow an old monitor but will usually give you a lot of garbled fuzz on your screen. We are going to eliminate all but the three above; we do that by commenting them out with # or deleting the lines entirely. (You may want to back up the le rst.) You could leave it up to to choose the correct Modeline to match the capabilities of the monitor, but this doesnt always work. I always like to explicitly choose a selection of Modelines. If you dont nd modelines in your XF86Config you can use this as your monitor section:

3. Run xf86config. This is the ofcial conguration script. Run through all the options, being very sure not to guess. You can set your monitor to 31.5, 35.15, 35.5; Super VGA. . . if you have no other information to go on. Vertical sync can be set to 5090. Select your card from the card database (check the SuperProbe output), and check which server the program recommends this will be one of XF86 SVGA, XF86 S3, XF86 S3V, etc. Whether you set the symbolic link or not, or modify the /etc/X11/Xserver le is irrelevant. Note that you do not need a RAM DAC setting with most modern PCI graphics cards. The same goes for the Clockchip setting. 4. Do not run at this point.

Section "Monitor" Identifier "My Monitor" VendorName "Unknown" ModelName "Unknown" HorizSync 30-40 VertRefresh 50-90 Modeline "320x200" 12.588 320 336 384 400 ModeLine "400x300" 18 400 416 448 512 Modeline "512x384" 20.160 512 528 592 640 Modeline "640x480" 25.175 640 664 760 800 ModeLine "800x600" 36 800 824 896 1024 Modeline "1024x768" 44.9 1024 1048 1208 1264 EndSection

200 300 384 480 600 768

204 301 385 491 601 776

205 302 388 493 603 784

225 312 404 525 625 817

Doublescan Doublescan -HSync -VSync

Interlace

6. Edit your "Device" section. You can make it as follows for XFree86 version 3.3, and there should be only one "Device" section.

5. The xf86config command should have given you an example /etc/X11/XF86Config le to work with. You need not run it again. You will notice that the le is divided into sections, like
Section "<section-name>" <config-line> <config-line> <config-line> EndSection

Section "Device" Identifier "My Video Card" VendorName "Unknown" BoardName "Unknown" VideoRam 4096 EndSection

For XFree86 version 4, you must add the device driver module. On my laptop, this is ati:

Search for the "Monitor" section. A little further down you will see lots of lines like:
# 640x480 @ 60 Hz, 31.5 kHz hsync Modeline "640x480" 25.175 640 664 760 800 # 800x600 @ 56 Hz, 35.15 kHz hsync ModeLine "800x600" 36 800 824 896 1024 # 1024x768 @ 87 Hz interlaced, 35.5 kHz hsync Modeline "1024x768" 44.9 1024 1048 1208 1264 480 600 768 491 601 776 493 603 784 525 625 817 Interlace

Section "Device" Identifier "My Video Card" Driver "ati" VendorName "Unknown" BoardName "Unknown" VideoRam 4096 EndSection

Several options that can also be added to the "Device" section to tune your card. Three possible lines are
Option Option Option "no_accel" "sw_cursor" "no_pixmap_cache"

501

502

43. The X Window System

43.6. X Conguration

43.7. Visuals

43. The X Window System

which disable graphics hardware acceleration, hardware cursor support, and video memory pixmap caching, respectively. The last refers to the use of the cards unused memory for intermediate operations. You should try these options if there are glitches or artifacts in your display. 7. Your "Screen" section should properly order the modes specied in the "Monitor" section. It should use your single "Device" section and single "Monitor" section, "My Video Card" and "My Monitor", respectively. Note that XFree86 version 3.3 does not take a DefaultDepth option.
Section "Screen" Identifier "My Screen" Device "My Video Card" Monitor "My Monitor"
5

For XFree86 version 4, card support is compiled as separate modules named /usr/X11R6/lib/modules/drivers/cardname drv.o. A single binary executable /usr/X11R6/bin/XFree86 loads the appropriate module based on the Driver "cardname" line in the "Device" section. Having added this, you can run

/usr/X11R6/bin/XFree86

where the depth is set from the DefaultDepth 16 line in the "Screen" section. You can nd what driver to use by greping the modules with the name of your graphics card. This is similar to what we did with kernel modules on page 473. 9. A good idea is to now create a script, /etc/X11/X.sh, containing your -bpp option with the server you would like to run. For example,
#!/bin/sh exec /usr/X11R6/bin/<server> -bpp 16

DefaultDepth 16 Subsection "Display" ViewPort 0 0 Virtual 1024 768 Depth 16 Modes "1024x768" "800x600" "640x480" "512x384" "400x300" "320x240" EndSubsection Subsection "Display" ViewPort 0 0 Virtual 1024 768 Depth 24 Modes "1024x768" "800x600" "640x480" "512x384" "400x300" "320x240" EndSubsection Subsection "Display" ViewPort 0 0 Virtual 1024 768 Depth 8 Modes "1024x768" "800x600" "640x480" "512x384" "400x300" "320x240" EndSubsection EndSection

10. You can then symlink /usr/X11R6/bin/X to this script. It is also worth symlinking /etc/X11/X to this script since some congurations look for it there. There should now be no chance that could be started except in the way you want. Double-check by running X on the command-line by itself.

43.7

Visuals

introduces the concept of a visual. A visual is the hardware method used to represent colors on your screen. There are two common and four specialized types: TrueColor(4) The most obvious way of representing a color is to use a byte for each of the red, green, and blue values that a pixel is composed of. Your video buffer will hence have 3 bytes per pixel, or 24 bits. You will need 800 600 3 = 1440000 bytes to represent a typical 800 by 600 display. Another way is to use two bytes, with 5 bits for red, 6 for green, and then 5 for blue. This gives you 32 shades of red and blue, and 64 shades of green (green should have more levels because it has the most inuence over the pixels overall brightness). Displays that use 4 bytes usually discard the last byte, and are essentially 24-bit displays. Note also that most displays using a full 8 bits per color discard the trailing bits, so there is often no appreciable difference between a 16-bit display and a 32-bit display. If you have limited memory, 16 bits is preferable; it is also faster. 504

8. At this point you need to run the program itself. For XFree86 version 3.3, there will be a separate package for each video card, as well as a separate binary with the appropriate driver code statically compiled into it. These binaries are of the form /usr/X11R6/bin/XF86 cardname. The relevant packages can be found with the command dpkg -l xserver-* for Debian , and rpm -qa | grep XFree86 for RedHat 6 (or RedHat/RPMS/XFree86-* on your CD-ROM). You can then run
/usr/X11R6/bin/XFree86-<card> -bpp 16

which also sets the display depth to 16, that is, the number of bits per pixel, which translates to the number of colors. 503

43. The X Window System

43.8. The startx and xinit Commands

43.10. X Font Naming Conventions

43. The X Window System

PseudoColor(3) If you want to display each pixel with only one byte and still get a wide range of colors, the best way is to make that pixel index a dynamic table of 24-bit palette values: 256 of them exactly. 8-bit depths work this way. You will have just as many possible colors, but applications will have to pick what colors they want to display at once and compete for entries in the color palette. StaticGray(0) These are gray-level displays usually with 1 byte or 4 bits per pixel, or monochrome displays with 1 byte per pixel, like the legacy Hercules Graphics Card (HGC, or MDAmonochrome graphics adapter). Legacy VGA cards can be set to 640 480 in 16-color black and white. is almost usable in this mode and has the advantage that it always works, regardless of what hardware you have. StaticColor(2) This usually refers to 4-bit displays like the old (and obsolete) CGA and EGA displays having a small xed number of colors. DirectColor(5) This is rarely used and refers to displays that have a separate palette for each of red, green, and blue. GrayScale(1) These are like StaticGray, but the gray levels are programmable like PseudoColor. This is also rarely used. You can check the visuals that your display supports with the xdpyinfo command. You will notice more than one visual listed, since can effectively support a simple StaticColor visual with PseudoColor, or a DirectColor visual with TrueColor. The default visual is listed rst and can be set with the -cc option as we did above for the 16-color server. The argument to the -cc option is the number code above in parentheses. Note that good applications check the list of available visuals and choose an appropriate one. There are also those that require a particular visual, and some that take a -visual option on the command-line.

43.9

init runs mgetty, which displays a login: prompt on every attached character terminal. init can also run xdm, which displays a graphical login box on every server. Usually, there will only be one server: the one on your own machine. and
x:5:respawn:/usr/X11R6/bin/xdm -nodaemon

The interesting lines inside your inittab le are

id:5:initdefault:

which state that the default run level is 5 and that xdm should be started at run level 5. This should only be attempted if you are sure that works (by running X on the command-line by itself). If it doesnt, then xdm will keep trying to start , effectively disabling the console. On systems besides RedHat and Debian , these may be run levels 2 versus 3, where run level 5 is reserved for something else. In any event, there should be comments in your /etc/inittab le to explain your distributions convention.

43.10

X Font Naming Conventions

Most applications take a -fn or -font option to specify the font. In this section, I give a partial guide to font naming. A font name is a list of words and numbers separated by hyphens. A typical font name is -adobe-courier-medium-r-normal--12-120-75-75-m-60-iso8859-1. Use the xlsfonts command to obtain a complete list of fonts. The font name elds have the following meanings:

43.8 The startx and xinit Commands

The action of starting an server, then a window manager should obviously be auis to run the xinit command on its own. On tomated. The classic way to start L INUX this has been superseded by which is a script that runs xinit after setting some environment variables. These commands indirectly call a number of conguration scripts in /etc/X11/xinit/ and your home directory, where you can specify your window manager and startup applications. See xinit(1) and startx(1) for more information. 505
startx

adobe The name of the fonts maker. Others are

abisource adobe arabic b&h bitstream cronyx daewoo dec dtp gnu isas jis macromedia microsoft misc monotype mutt schumacher software sony sun urw xfree86

courier The font family. This is the real name of the font. Some others are
arial arial black arioso avantgarde bitstream charter bookman dingbats fangsong ti fixed goth gothic helmet lucidux serif marlett mincho new century schoolbook newspaper nil starbats starmath symbol tahoma tera special terminal

506

43. The X Window System

43.10. X Font Naming Conventions

43.11. Font Conguration

43. The X Window System

century schoolbook charter chevara chevaraoutline clean comic sans ms conga courier courier new cursor

helmetcondensed helvetic helvetica impact lucida lucida console lucidabright lucidatypewriter lucidux mono lucidux sans

nimbus mono nimbus roman nimbus sans nimbus sans condensed open look cursor open look glyph palatino palladio song ti standard symbols

times times new roman timmons unifont utopia verdana webdings wingdings zapf chancery zapf dingbats

43.11 Font Conguration

Fonts used by are conventionally stored in /usr/X11R6/lib/X11/fonts/. Each directory contains a fonts.alias le that maps full font names to simpler names, and a fonts.alias le which lists the fonts contained in that directory. To create these les, you must cd to each directory and run mkfontdir as follows: You can rerun this command at any time for good measure.
mkfontdir -e /usr/X11R6/lib/X11/fonts/encodings -e /usr/X11R6/lib/X11/fonts/encodings/large

medium The font weight: it can also be bold, demibold, or regular. r Indicate that the font is roman; i is for italic and o is for oblique. normal Character width and intercharacter spacing. It can also be condensed, semicondensed, narrow, or double. 12 The pixel size. A zero means a scalable font that can be selected at any pixel size. The largest xed sized font is about 40 points. 120 The size in tenths of a printers point. This is usually 10 times the pixel size. 75-75 Horizontal and vertical pixel resolution for which the font was designed. Most monitors today are 75 pixels per inch. The only other possible values are 72-72 or 100-100. m The font spacing. Other values are monospaced, proportional, or condensed. 60 The average width of all characters in the font in tenths of a pixel. iso8859-1 The ISO character set. In this case, the 1 indicates ISO Latin 1, a superset of the ASCII character set. This last bit is the locale setting, which you would normally omit to allow to determine it according to your locale settings. As an example, start cooledit with
-font -font -font -font -*-times-medium-r-*--20-*-*-*-p-*-iso8859-1 -*-times-medium-r-*--20-*-*-*-p-* -*-helvetica-bold-r-*--14-*-*-*-p-*-iso8859-1 -*-helvetica-bold-r-*--14-*-*-*-p-*

To tell to use these directories add the following lines to your "Files" section. A typical conguration will contain
Section "Files" RgbPath "/usr/X11R6/lib/X11/rgb" FontPath "/usr/X11R6/lib/X11/fonts/misc/:unscaled" FontPath "/usr/X11R6/lib/X11/fonts/75dpi/:unscaled" FontPath "/usr/X11R6/lib/X11/fonts/Speedo/" FontPath "/usr/X11R6/lib/X11/fonts/Type1/" FontPath "/usr/X11R6/lib/X11/fonts/misc/" FontPath "/usr/X11R6/lib/X11/fonts/75dpi/" EndSection

Often you will add a directory without wanting to restart add a directory to the font path is: and to remove a directory, use To set the font path, use and reset it with
xset +fp /usr/X11R6/lib/X11/fonts/<new-directory>

. The command to

xset -fp /usr/X11R6/lib/X11/fonts/<new-directory>

xset fp= /usr/X11R6/lib/X11/fonts/misc,/usr/X11R6/lib/X11/fonts/75dpi

cooledit cooledit cooledit cooledit

xset fp default

These invoke a newspaper font and an easy-reading font respectively. A * means that the server can place default values into those elds. That way, you do not have to specify a font exactly. The xfontsel command is the traditional showfont command dumps fonts as ASCII text. 507 utility for displaying fonts and the

If you change anything in your font directories, you should run

xset fp rehash

to cause

to reread the font directories. 508

43. The X Window System

43.12. The Font Server

43. The X Window System

The command chkfontpath prints out your current font path setting. Note that XFree86 version 4 has a TrueType engine. TrueType (.ttf) fonts are common to Windows. They are high-quality, scalable fonts designed for graphical displays. You can add your TrueType directory alongside your other directories above, and run inside each one. Note that the ttmkfdir is needed to catalog TrueType fonts as scalable fonts.
ttmkfdir > fonts.scale mkfontdir -e /usr/X11R6/lib/X11/fonts/encodings -e /usr/X11R6/lib/X11/fonts/encodings/large

Section "Files" RgbPath "/usr/X11R6/lib/X11/rgb" FontPath "/usr/X11R6/lib/X11/fonts/misc/:unscaled" FontPath "unix/:7100" EndSection

Or otherwise use xset:

xset +fp unix/:7100

43.12 The Font Server

Having all fonts stored on all machines is expensive. Ideally, you would like a large font database installed on one machine and fonts to be read off this machine, over the network and on demand. You may also have an that does not support a particular font type; if it can read the font from the network, built-in support will not be necessary. The daemon xfs ( font server) facilitates all of this. xfs reads its own simple conguration le from /etc/X11/fs/config or /etc/X11/xfs/config. It might contain a similar list of directories:
client-limit = 10 clone-self = on catalogue = /usr/X11R6/lib/X11/fonts/misc:unscaled, /usr/X11R6/lib/X11/fonts/75dpi:unscaled, /usr/X11R6/lib/X11/fonts/ttf, /usr/X11R6/lib/X11/fonts/Speedo, /usr/X11R6/lib/X11/fonts/Type1, /usr/X11R6/lib/X11/fonts/misc, /usr/X11R6/lib/X11/fonts/75dpi default-point-size = 120 default-resolutions = 75,75,100,100 deferglyphs = 16 use-syslog = on no-listen = tcp

Note that no other machines can use your own font server because of the nolisten = tcp option. Deleting this line (and restarting xfs) allows you to instead use
FontPath "inet/127.0.0.1:7100"

which implies an open TCP connection to your font server, along with all its security implications. Remote machines can use the same setting after changing 127.0.0.1 to your IP address. Finally, note that for XFree86 version 3.3, which does not have TrueType support, the font server name xfstt is available on Fresh Meat http://freshmeat.net/.

You can start the font server by using:

and change your font paths in /etc/X11/XF86Config to include only a minimal set of fonts: 509 510

/etc/init.d/xfs start ( /etc/rc.d/init.d/xfs start )

44.1. Common Attacks

44. U NIX Security

is, say, only used for mail and web, shell logins may not be allowed at all; hence, the former type of security breach is of less concern. Here are some of the ways security is compromised, just to give an idea of what U NIX security is about. In some cases, I indicate when it is of more concern to multiuser systems. Note also that attacks from users become an issue when a remote attack succeeds and a hacker gains user privileges to your system (even as a nobody user). This is an issue even if you do not host logins.

Chapter 44

44.1.1 Buffer overow attacks

U NIX Security
This is probably the most important chapter of this book.1 L INUX has been touted as both the most secure and insecure of all operating systems. The truth is both. Take no heed of advice from the L INUX community, and your server will be hacked eventually. Follow a few simple precautions, and it will be safe for years without much maintenance. The attitude of most novice administrators is Since the U NIX system is so large and complex and since there are so many millions of them on the Internet, it is unlikely that my machine will get hacked. Of course, it wont necessarily be a person targeting your organization that is the problem. It could be a person who has written an automatic scanner that tries to hack every computer in your city. It could also be a person who is not an expert in hacking at all, but who has merely downloaded a small utility to do it for him. Many seasoned experts write such utilities for public distribution, while so-called script kiddies (because the means to execute a script is all the expertise needed) use these to do real damage. The word hack means gaining unauthorized access to a
computer. However, programmers sometimes use the term to refer to enthusiastic work of any kind. Here we refer to the malicious denition.
5

Consider the following program. If you dont understand that well, it doesnt matterits the concept that is important. (Before trying this example, you should unplug your computer from the network.)
#include <stdio.h> void do_echo (void) { char buf[256]; gets (buf); printf ("%s", buf); fflush (stdout); } int main (int argc, char **argv) { for (;;) { do_echo (); } }

In this chapter you will get an idea of the kinds of ways a U NIX system gets hacked. Then you will know what to be wary of, and how you can minimize risk.

You can compile this program with gcc -o /usr/local/sbin/myechod myechod.c. Then, make a system service out of it as follows: For xinetd, create le /etc/xinetd.d/myechod containing:
service myechod { flags socket_type wait user server log_on_failure }

44.1 Common Attacks

I personally divide attacks into two types: attacks that can be attempted by a user on the system, and network attacks that come from outside of a system. If a server
1 Thanks

= REUSE = stream = no = root = /usr/local/sbin/myechod += USERID

to Ryan Rubin for reviewing this chapter.

512

511

44. U NIX Security

44.1. Common Attacks

44. U NIX Security

while for inetd add the following line to your /etc/inetd.conf le:
myechod stream tcp nowait root /usr/local/sbin/myechod

to refuse privileges to anyone who isnt trusted. Hence, su requests a password and checks it against the passwd le before doing anything. Once again, the logic of the program has to hold up to ensure security, as well as to provide insurance against buffer overow attacks. Should su have a aw in the authentication logic, it would enable someone to change to a UID that they were not privileged to hold. Setuid programs should hence be considered with the utmost suspicion. Most setuid programs try be small and simple, to make it easy to verify the security of their logic. A vulnerability is more likely to be found in any setuid program that is large and complex. (Of slightly more concern in systems hosting many untrusted user logins.)

Of course, the service myechod does not exist. Add the following line to your /etc/services le: and then restart xinetd (or inetd) as usual.
myechod 400/tcp # Temporary demo service

You can now run netstat -na. You should see a line like this somewhere in the output:
tcp 0 0 0.0.0.0:400 0.0.0.0:* LISTEN

44.1.3 Network client programs

Consider when your FTP client connects to a remote untrusted site. If the site server returns a response that the FTP client cannot handle (say, a response that is too long a buffer overow), it could allow malicious code to be executed by the FTP client on behalf of the server. Hence, it is quite possible to exploit a security hole in a client program by just waiting for that program to connect to your site. (Mostly a concern in systems that host user logins.)

You can now run telnet localhost 400 and type away happily. As you can see, the myechod service simply prints lines back to you. Someone reading the code will realize that typing more than 256 characters will write into uncharted memory of the program. How can they use this effect to cause the program to behave outside of its design? The answer is simple. Should they be able to write processor instructions into an area of memory that may get executed later, they can cause the program to do anything at all. The process runs with root privileges, so a few instructions sent to the kernel could, for example, cause the passwd le to be truncated, or the le system superblock to be erased. A particular technique that works on a particular program is known as an exploit for a vulnerability. In general, an attack of this type is known as a buffer overow attack. To prevent against such attacks is easy when you are writing new programs. Simply make sure that any incoming data is treated as being dangerous. In the above case, the fgets function should preferably be used, since it limits the number of characters that could be written to the buffer. There are, however, many functions that behave in such a dangerous way: even the strcpy function writes up to a null character that may not be present; sprintf writes a format string that could be longer than the buffer. getwd is another function that also does no bound checking. However, when programs grow long and complicated, it becomes difcult to analyze where there may be loopholes that could be exploited indirectly. A program is a legal contract with an impartial jury.

44.1.4

/tmp le vulnerability

If a program creates a temporary le in your /tmp/ directory and it is possible to predict the name of the le it is going to create, then it may be possible to create that le in advance or quickly modify it without the programs knowledge. Programs that create temporary les in a predictable fashion or those that do not set correct permissions (with exclusive access) to temporary les are liable to be exploited. For instance, if a program running as superuser truncates a le /tmp/9260517.TMP and it was possible to predict that le name in advance, then a hacker could create a symlink to /etc/passwd of the same name, resulting in the superuser program actually truncating the passwd le. (Of slightly more concern in systems that host many untrusted user logins.)

44.1.2

Setuid programs

44.1.5 Permission problems

It is easy to see that a directory with permissions 660 and ownerships root:admin cannot be accessed by user jsmith if he is outside of the admin group. Not so easy 514

A program like su must be setuid (see Chapter 14). Such a program has to run with root privileges in order to switch UIDs to another user. The onus is, however, on su 513

44. U NIX Security

44.1. Common Attacks

44.2. Other Types of Attack

44. U NIX Security

to see is when you have thousands of directories and hundreds of users and groups. Who can access what, when, and why becomes complicated and often requires scripts to be written to do permission tests and sets. Even a badly set /dev/tty* device can cause a users terminal connection to become vulnerable. (Of slightly more a concern in systems that host many untrusted user logins.)

44.2

Other Types of Attack

The preceding lists are far from exhaustive. It never ceases to amaze me how new loopholes are discovered in program logic. Not all of these exploits can be classied; indeed, it is precisely because new and innovative ways of hacking systems are always being found, that security needs constant attention.

44.1.6

Environment variables

There are lots of ways of creating and reading environment variables to either exploit a vulnerability or obtain some information that will compromise security. Environment variables should never hold secret information like passwords. On the other hand, when handling environment variables, programs should consider the data they contain to be potentially malicious and do proper bounds checking and verication of their contents. (Of more concern in systems that host many untrusted user logins.)

44.3

Counter Measures

Security rst involves removing known risks, then removing potential risks, then (possibly) making life difcult for a hacker, then using custom U NIX security paradigms, and nally being proactively cunning in thwarting hack attempts.

44.3.1 Removing known risks: outdated packages

It is especially sad to see naive administrators install packages that are well known to be vulnerable and for which script kiddy exploits are readily available on the Internet. If a security hole is discovered, the package will usually be updated by the distribution vendor or the author. The bugtraq http://www.securityfocus.com/forums/bugtraq/intro.html mailing list announces the latest exploits and has many thousands of subscribers worldwide. You should get on this mailing list to be aware of new discoveries. The Linux Weekly News http://lwn.net/ is a possible source for security announcements if you only want to read once a week. You can then download and install the binary or source distribution provided for that package. Watching security announcements is critical. I often ask administrators if they have upgraded the xxx service and get the response, that
either they are not sure if they need it, do not believe it is vulnerable, do not know if it is running, where to get a current package, or even how to perform the upgrade; as if their ignorance absolves them of their responsibility. If the janitor were to duct-tape your safe keys to a window pane, would you re him?

44.1.7

Password snifng

When telnet, ftp, rlogin, or in fact any program at all that authenticates over the network without encryption is used, the password is transmitted over the network in plain text, that is, human-readable form. These programs are all common network utilities that old U NIX hands were accustomed to using. The sad fact is that what is being transmitted can easily be read off the wire with the most elementary tools (see tcpdump on page 266). None of these services should be exposed to the Internet. Use within a local LAN is safe, provided the LAN is rewalled, and your local users are trusted.

44.1.8

Password cracking

This concept is discussed in Section 11.3.

44.1.9

Denial of service attacks

A denial of service (DoS) attack is one which does not compromise the system but prevents other users from using a service legitimately. It can involve repetitively loading a service to the point that no one else can use it. In each particular case, logs or TCP trafc dumps might reveal the point of origin. You might then be able to deny access with a rewall rule. There are many types of DoS attacks that can be difcult or impossible to protect against. 515

This goes equally for new systems that you install: never install outdated packages. Some vendors ship updates to their older distributions. This means that you can install from an old distribution and then upgrade all your packages from an update package list. Your packages would be then as secure as the packages of the distribution that has the highest version number. For instance, you can install RedHat 6.2 from a 6-month-old CD, then download a list of RedHat 6.2 update packages. Alternatively, you can install the latest RedHat version 7.? which has a completely different set of packages. On the other hand, some other vendors may no longer support an older distribution, meaning that those packages will never be updated. In this case, you should be sure to install or upgrade with the vendors most current distribution or manually recompile vulnerable packages by yourself. 516

44. U NIX Security

44.3. Counter Measures

44. U NIX Security

Over and above this, remember that vendors are sometimes slow to respond to security alerts. Hence, trust the free software communitys alerts over anything vendors may fail to tell you. Alternatively, if you discover that a service is insecure, you may just want to disable it (or better still, uninstall it) if its not really needed.

are ftp, telnet, rlogin, uucp, imap, pop3, and any service that does not use encryption and yet authenticates with a password. Instead, you should use ssh and scp. There are secure versions of POP and IMAP (SPOP3 and SIMAP), but you may not be able to nd good client programs. If you really have to use a service, you should limit the networks that are allowed to connect to it, as described on page 293 and 296. Old U NIX hands are notorious for exporting NFS shares (/etc/exports) that are readable (and writable) from the Internet. The group of functions to do Sun Microsystems port mapping and NFSthe nfs-utils (rpc.. . . ) and portmap packagesdont give me a warm, fuzzy feeling. Dont use these on machines exposed to the Internet.

44.3.2

Removing known risks: compromised packages

Packages that are modied by a hacker can allow him a back door into your system: so called Trojans. Use the package verication commands discussed in Section 24.2.6 to check package integrity.

44.3.3

Removing known risks: permissions

44.3.6 Removing potential risks: network

Install libsafe. This is a library that wraps all those vulnerable functions discussed above, thus testing for a buffer overow attempt with each call. It is trivial to install, and sends email to the administrator upon hack attempts. Go to http://www.avayalabs.com/project/libsafe/index.html for more information, or send email to libsafe@research.avayalabs.com. The libsafe library effectively solves 90% of the buffer overow problem. There is a very slight performance penalty, however. Disable all services that you are not using. Then, try to evaluate whether the remaining services are really needed. For instance, do you really need IMAP or would POP3 sufce? IMAP has had a lot more security alerts than POP3 because it is a much more complex service. Is the risk worth it? xinetd (or inetd) runs numerous services, of which only a few are needed. You should trim your /etc/xinetd.d directory (or /etc/inetd.conf le) to a minimum. For xinetd, you can add the line disable = yes to the relevant le. Only one or two les should be enabled. Alternatively, your /etc/inetd.conf should have only a few lines in it. A real-life example is:
ftp pop-3 imap stream stream stream tcp tcp tcp nowait nowait nowait root root root /usr/sbin/tcpd /usr/sbin/tcpd /usr/sbin/tcpd in.ftpd -l -a ipop3d imapd

It is easy to locate world-writable les. There should be only a few in the /dev and /tmp directories:
find / -perm -2 ! -type l -ls

Files without any owner are an indication of mismanagement or compromise of your system. Use the find command with
find / -nouser -o -nogroup -ls

44.3.4

Password management

It is obvious that variety in user passwords is more secure. It is a good idea to rather not let novice users choose their own passwords. Create a randomizing program to generate completely arbitrary 8 character passwords for them. You should also use the pwconv utility from the shadow-utils package to create the shadow password les (explained in Section 11.3). See pwconv(8) for information.

44.3.5

Disabling inherently insecure services

This advice should be taken quite literally. The rule of thumb is that if you dont know what a service does, you should disable it. See also Section 29.6. In the above real-life case, the services were additionally limited to permit only certain networks to connect (see page 293 and 296). xinetd (or inetd) is not the only problem. There are many other problematic services. Entering netstat -nlp gives initial output, like 518

Services that are inherently insecure are those that allow the password to be sniffed over the Internet or provide no proper authentication to begin with. Any service that does not encrypt trafc should not be used for authentication over the Internet. These 517

44. U NIX Security

44.3. Counter Measures

44. U NIX Security

(Not all processes could be identified, non-owned process info will not be shown, you would have to be root to see it all.) Active Internet connections (only servers) Proto Recv-Q Send-Q Local Address Foreign Address State tcp 0 0 0.0.0.0:25 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:400 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:21 0.0.0.0:* LISTEN tcp 0 0 172.23.80.52:53 0.0.0.0:* LISTEN tcp 0 0 127.0.0.1:53 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:6000 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:515 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:22 0.0.0.0:* LISTEN udp 0 0 0.0.0.0:1045 0.0.0.0:* udp 0 0 172.23.80.52:53 0.0.0.0:* udp 0 0 127.0.0.1:53 0.0.0.0:* raw 0 0 0.0.0.0:1 0.0.0.0:* 7 raw 0 0 0.0.0.0:6 0.0.0.0:* 7

even the ping command. If you do allow any shell logins on your system, then you should remove setuid permissions from all shell commands.

PID/Program name 2043/exim 32582/xinetd 32582/xinetd 30604/named 30604/named 583/X 446/ 424/sshd 30604/named 30604/named 30604/named -

44.3.8 Making life difcult

There is much that you can do that is not security per se but that will make life considerably more difcult for a hacker, and certainly impossible for a stock standard attack, even if your system is vulnerable. A hack attempt often relies on a system being congured a certain way. Making your system different from the standard can go a long way. Read-only partitions: It is allowable to mount your /usr partition (and critical toplevel directories like /bin) read-only since these are, by denition, static data. Of course, anyone with root access can remount it as writable, but a generic attack script may not know this. Some SCSI disks can be congured as read-only by using dip switches (or so I hear). The /usr partition can be made from an ISO 9660 partition (CD-ROM le system) which is read-only by design. You can also mount your CD-ROM as a /usr partition: access will be slow, but completely unmodiable. Finally, you can manually modify your kernel code to fail writemount attempts on /usr. Read-only attributes: L INUX has additional le attributes to make a le unmodiable over and above the usual permissions. These attributes are controlled by the commands chattr and lsattr. You can make a log le append-only with chatter +a /var/log/messages /var/log/syslog or make les immutable with, chatter +i /bin/login: both actions are a good idea. The command
chattr -R +i /bin /boot /lib /sbin /usr

but doesnt show that PID 446 is actually lpd. For that information just type ls al /proc/446/. You can see that ten services are actually open: 1, 6, 21, 22, 25, 53, 400, 515, 1045, and 6000. 1 and 6 are kernel ports, and 21 and 400 are FTP and our echo daemon, respectively. Such a large number of open ports provides ample opportunity for attack. At this point, you should go through each of these services and (1), decide whether you really need them. Then (2), make sure you have the latest version; nally (3), consult the packages documentation so that you can limit the networks that are allowed to connect to those services. It is interesting that people are wont to make assumptions about packages to the tune of This service is so popular it cant possibly be vulnerable. The exact opposite is, in fact, true: The more obscure and esoteric a service is, the less likely that someone has taken the trouble to nd a vulnerability. In the case of named (i.e., bind), a number of most serious vulnerabilities were made public as regards every Bind release prior to 9. Hence, upgrading to the latest version (9.1 at the time of writing) from source was prudent for all the machines I administered (a most-time consuming process).

44.3.7

Removing potential risks: setuid programs

is a better idea still. Of course, anyone with superuser privileges can switch them back. Periodic system monitoring: It is useful to write your own crond scripts to check whether les have changed. the scripts can check for new setuid programs, permissions, or changes to binary les; or you can reset permissions to what you think is secure. Just remember that cron programs can be modied by anyone who hacks into the system. A simple command
find / -mtime 2 -o -ctime 2

It is easy to nd all the setuid programs on your system:

find / -type f -perm +6000 -ls

Disabling them is just as easy:

chmod -s /bin/ping

There is nothing wrong with the decision that ordinary users are not allowed to use 519

searches for all les that have been modied in the last two days. 520

44. U NIX Security

44.3. Counter Measures

44. U NIX Security

Nonstandard packages: If you notice many security alerts for a package, switch to a different one. There are alternatives to bind, wu-ftpd, sendmail (as covered in Chapter 30), and almost every service you can think of. You can also try installing an uncommon or security-specialized distribution. Switching entirely to FreeBSD is also one way of reducing your risk considerably. This is not a joke.

Access control lists: These lists extend the simple user/group/other permissions of U NIX les to allow arbitrary lists of users to access particular les. This really does nothing for network security but is useful if you have many users on the system and you would like to restrict them in odd ways. (ACL is a little out of place in this list.) DTE: Domain and Type Enforcement works like this: When a program is executed, it is categorized and only allowed to do certain things even if it is running as root. These limitations are extended to child processes that it may execute. This is real security; there are kernel patches to do this. The National Security Agency (of the U.S.) (NSA) actually has a L INUX distribution built around DTE. medusa: This is a security system that causes the kernel to query a user daemon before letting any process on the system do anything. It is the most ubiquitous security system out because it is entirely congurableyou can make the user daemon restrict anything however you like. VXE: Virtual eXecuting Environment dictates that a program executes in its own protected space while VXE executes a Lisp program to check whether a system call is allowed. This is effectively a lot like medusa. MAC: Mandatory Access Controls. This is also about virtual environments for processes. MAC is a POSIX standard. RSBAC and RBAC: Rule-Set-Based Access Controls and Role-Based Access Controls. These look like a combination of some of the above. LIDS: Linux Intrusion Detection System does some meager preventive measures to restrict module loading, le modications, and process information. Kernel patches exist to do all of the above. Many of these projects are well out of the test phase but are not in the mainstream kernel, possibly because developers are not sure of the most enduring approach to U NIX security. They all have one thing in common: double-checking what a privileged process does, which can only be a good thing.

Nonstandard messages: Many services provide banners and informational messages which give away the version of your software. For example, mail servers have default HELO responses to advertise themselves; and login and FTP banners often display the operating system you are running. These messages should be customized to provide less information on which to base an attack. You can begin by editing /etc/motd. Minimal kernels: Its easy to compile your kernel without module support, with an absolutely minimal set of features. Loading of Trojan modules has been a source of insecurity in the past. Such a kernel can only make you safer. Non-Intel architecture: Hackers need to learn assembly language to exploit many vulnerabilities. The most common assembly language is that of Intel 80?86 processors. Using a non-Intel platform adds that extra bit of obscurity. Removing ngerprints: Your system identies itself to OpenWall project: This has a kernel patch that makes the stack of a process nonexecutable (which will thwart most kinds of buffer overow attempts) and does some other cute things with the /tmp directory and process I/O.

44.3.9

Custom security paradigms

Hackers have limited resources. Take oneupmanship away and security is about the cost of hacking a system versus the reward of success. If you feel the machine you administer is bordering on this category then you need to start billing far more for your hours and doing things like those described below. It is possible to go to lengths that will make a L INUX system secure against a large governments defense budget. Capabilities: This is a system of security that gives limited kinds of superuser access to programs that would normally need to be full-blown setuid root executables. Think: Most processes that run with root (setuid) privileges do so because of the need to access only a single privileged function. For instance, the ping program does not need complete superuser privileges (run ls -l /bin/ping and note the setuid bit). Capabilities are a ne-grained set of privileges that say that a process can do particular things that an ordinary user cant, without ever having full root access. In the case of ping, its capability would be certain networking access that only root is normally allowed to do. 521

44.3.10 Proactive cunning

Proactive cunning means attack monitoring and reaction, and intrusion monitoring and reaction. Utilities that do this come under a general class called network intrusion detection software. The idea that one might detect and react to a hacker has an emotional appeal, but it automatically implies that your system is insecure to begin withwhich is probably true, considering the rate at which new vulnerabilities are being reported. I am weary of so-called intrusion detection systems that administrators implement even before the most elementary of security measures. Really, one 522

44. U NIX Security

44.4. Important Reading

44.6. Security Auditing

44. U NIX Security

must implement all of the above security measures before thinking about intrusion monitoring. To picture the most basic form of monitoring, consider this: To hack a system, one usually needs to test for open services. To do this, one tries to connect to every port on the system to see which are open. This is known as a port scan. There are simple tools to detect a port scan, which will then start a rewall rule that will deny further access from the offending host although this can work against you if the hacker has spoofed your own IP address. More importantly, the tools will report the IP address from which the attack arose. A reverse lookup will give the domain name, and then a whois query on the appropriate authoritative DNS registration site will reveal the physical address and telephone number of the domain owner. Port scan monitoring is the most elementary form of monitoring and reaction. From there up, you can nd innumerable bizarre tools to try and read into all sorts of network and process activity. I leave this to your own research, although you might want to start with the Snort trafc scanner http://www.snort.org/, the Tripwire intrusion detection system http://www.tripwiresecurity.com/, and IDSA http://jade.cs.uct.ac.za/idsa/. A point to such monitoring is also as a deterrent to hackers. A network should be able to nd the origins of an attack and thereby trace the attacker. The threat of discovery makes hacking a far less attractive pastime, and you should look into the legal recourse you may have against people who try to compromise your system.

44.6

Security Auditing

This chapter is mostly concerned with securing your own L INUX server. However, if you have a large network, security auditing is a more extensive evaluation of your systems vulnerabilities. Security auditing becomes an involved procedure when multiple administrators maintain many different platforms across a network. There are companies that specialize in this work: Any network that does not dedicate an enlightened staff member should budget generously for their services. Auditing your network might involve the following: Doing penetration testing of rewalls. Port scanning. Installing intrusion detection software. Analyzing and reporting on Internet attack paths. Evaluating service access within your local LAN. Tracking your administrators maintenance activities. Trying password cracking on all authentication services. Monitoring the activity of legitimate user accounts.

44.4 Important Reading

The preceding is a practical guide. It gets much more interesting than this. A place to start is the comp.os.linux.security FAQ. This FAQ gives the most important U NIX security references available on the net. You can download it from http://www.memeticcandiru.com/colsfaq.html, http://www.linuxsecurity.com/docs/The Linux Security colsfaq.html or http://www.geocities.com/swan daniel/colsfaq.html. http://www.linuxsecurity.com/ web page also has a security quick reference card that summarizes most everything you need to know in two pages.

44.5 Security Quick-Quiz

How many security reports have you read? How many packages have you upgraded because of vulnerabilities? How many services have you disabled because you were unsure of their security? How many access limit rules do you have in your hosts.*/xinetd services?

Network attacks cost companies billions of dollars each year in service downtime and repair. Failing to pay attention to security is a false economy.

If your answer to any of these questions is fewer than 5, you are not being conscientious about security. 523 524

A.3. Lecture Style

A. Lecture Schedule

who, at the very least, are experienced in assembling hardware and installing operating systems. At the other end of the scale, end users with no knowledge of commandline interfaces, programming, hardware assembly, or networking, would require a far less intensive lecture schedule and would certainly not cope with the abstraction of a shell interface.

Appendix A

Lecture Schedule
The following sections describe a 36-hour lecture schedule in 12 lessons, 2 per week, of 3 hours each. The lectures are interactive, following the text closely, but sometimes giving straightforward chapters as homework.

Of course, people of high intelligence can cover this material quite quickly, regardless of their IT experience, and it is smoothest when the class is at the same level. The most controversial method would be to simply place a tape measure around the cranium (since the latest data puts the correlation between IQ and brain size at about 0.4). A less intensive lecture schedule would probably cover about half of the material, with more personalized tuition, and having more in-class assignments.

A.3

Lecture Style

A.1 Hardware Requirements

The course requires that students have a L INUX system to use for their homework assignments. For past courses, most people were willing to repartition their home machines, buy a new hard drive, or use a machine of their employer. The classroom itself should have 4 to 10 places. It is imperative that students have their own machine, since the course is highly interactive. The lecturer need not have a machine. I myself prefer to write everything on a whiteboard. The machines should be networked with Ethernet and congured so that machines can telnet to each others IPs. A full L INUX installation is preferredeverything covered by the lectures must be installed. This would include all services, several desktops, and and kernel development packages. L INUX computers. CDs should also be available for those who need to set up their home

Lessons are three hours each. In my own course, these were in the evenings from 6 to 9, with two 10 minute breaks on the hour. It is important that there are a few days between each lecture for students to internalize the concepts and practice them by themselves. The course is completely interactive, following a type this now class... genre. The text is replete with examples, so these should be followed in sequence. In some cases, repetitive examples are skipped. Examples are written on the whiteboard, perhaps with slight changes for variety. Long examples are not written out: Now class, type in the example on page.... The motto of the lecture style is: keep em typing. Occasional diversions from the lecturers own experiences are always fun when the class gets weary. The lecturer will also be aware that students get stuck occasionally. I check their screens from time to time, typing in the odd command for them, to speed the class along.

Most notably, each student should have his own copy of this text.

Lesson 1
A background to U NIX and L INUX history is explained, crediting the various responsible persons and organizations. The various copyrights are explained, with emphasis on the GPL. Chapter 4 then occupies the remainder of the rst three hours. 526

A.2 Student Selection

This lecture layout is designed for seasoned administrators of MS-DOS or Windows systems, those who at least have some kind of programming background, or those 525

A. Lecture Schedule

A.3. Lecture Style

A. Lecture Schedule

Homework: Appendix D and E to be read. Students to install their own L INUX distributions. Chapter 6 should be covered to learn basic operations with vi.

Lesson 6
Lectured coverage of Chapter 25 and Chapter 26. Also demonstrate an attempt to sniff the password of a telnet session with tcpdump. Then the same attempt with ssh.

Lesson 2
Chapter 5 (Regular Expressions) occupies the rst hour, then Chapter 7 (Shell Scripting) the remaining time. Lecturers should doubly emphasize to the class the importance of properly understanding regular expressions, as well as their wide use in U NIX. Homework: Research different desktop congurations and end-user applications. Students should become familiar with the different desktops and major applications that they offer.

Homework: Read Chapter 27 through Chapter 29 in preparation for next lesson.

Lesson 7
Chapters 27 through 29 covered in rst and second hour. A DNS server should be up for students to use. Last hour explains how Internet mail works, in theory only, as well as the structure of the exim conguration le. Homework: Read through Chapter 30 in preparation for next lesson.

Lesson 3
First hour covers Chapter 8. Second hour covers Chapters 9 and 10. Third hour covers Chapter 11. Homework: Research L INUX 16 and 13 should be accessed. on the Internet. All resources mentioned in Chapters

Lesson 8
First and second hours cover Chapter 30. Students to congure their own mail server. A DNS server should be present to test MX records for their domain. Last hour covers Chapters 31 and 32, excluding anything about modems. Homework: Experiment with Chapter 33. Chapter 34 not covered. Chapter 35 to be studied in detail. Students to set up a web server from Chapter 36 and report back with problems. Apache itself is not covered in lectures.

Lesson 4
First two hours cover Chapters 12, 13, 14, 15. Third hour covers Chapters 16 and 17. Homework: Chapters 18 through 21 to be covered. Students will not be able to modify the houses partitions, and printers will not be available, so these experiments are given for homework. Chapter 20 is not considered essential. Students are to attempt to congure their own printers and report back with any problems.

Lesson 9
First hour covers Chapter 37. Second and third hours cover Chapter 40. Students to congure their own name servers with forward and reverse lookups. Note that Samba is not covered if there are no Windows machines or printers to properly demonstrate it. An alternative would be to set up printing and le-sharing using smbmount. Homework: Chapter 41 for homeworkstudents to congure dialup network for themselves. Read through Chapter 42 in preparation for next lesson.

Lesson 5
First hour covers Chapter 22, second hour covers Chapter 24. For the third hour, student read Chapter 25 and Chapter 26, asking questions about any unclear points. Homework: Optionally, Chapters 23, then rereading of Chapter 25 and 26. 527

Lesson 10
First and second hours cover Chapter 42. Students to at least congure their own network card if no other hardware devices are available. Build a kernel with some 528

A. Lecture Schedule

A.3. Lecture Style

A. Lecture Schedule

customizations. Third hour covers the Window System in theory and use of the DISPLAY environment variable to display applications to each others servers. Homework: Study Chapter 28.

Lesson 11
First hour covers conguring of NFS, noting the need for a name server with forward and reverse lookups. Second and third hours cover Chapter 38. Homework: Download and read the Python tutorial. View the weeks security reports online. Study Chapter 44.

Lesson 12
First and second hours cover the security chapter and an introduction to the Python programming language. Last hour comprises the course evaluation. The nal lesson could possibly hold an examination if a certication is offered for this particular course.

529

530

B.1. Exam Details for 101

B. LPI Certication Cross-Reference

Interact with shells and commands using the command line (Chapter 4). Includes typing valid commands and command sequences (Chapter 4), dening, referencing and exporting environment variables (Chapter 9), using command history and editing facilities (Section 2.6), invoking commands in the path and outside the path (Section 4.6), using command substitution, and applying commands recursively through a directory tree (Section 20.7.5).

Appendix B

Obj 2: Process text streams using text processing lters Weight of objective: 7
Send text les and output streams through text utility lters to modify the output in a useful way (Chapter 8). Includes the use of standard U NIX commands found in the GNU textutils package such as sed, sort, cut, expand, fmt, head, join, nl, od, paste, pr, split, tac, tail, tr, and wc (see the man pages for each of these commands in conjunction with Chapter 8).

L INUX Professionals Institute Certication Cross-Reference

These requirements are quoted verbatim from the LPI web page http://www.lpi.org/. For each objective, the relevant chapter or section from this book is referenced in parentheses: these are my additions to the text. In some cases, outside references are given. Note that the LPI level 2 exams have not been nalized as of this writing. However, the preliminary draft of the level 2 curricula is mostly covered by this book.

Obj 3: Perform basic le management Weight of objective: 2

Use the basic U NIX commands to copy and move les and directories (Chapter 4). Perform advanced le management operations such as copying multiple les recursively and moving les that meet a wildcard pattern (Chapter 4). Use simple and advanced wildcard specications to refer to les (Chapter 4.3).

Obj 4: Use U NIX streams, pipes, and redirects Weight of objective: 3

Connect les to commands and commands to other commands to efciently process textual data. Includes redirecting standard input, standard output, and standard error; and piping one commands output into another command as input or as arguments (using xargs); sending output to stdout and a le (using tee) (Chapter 8).

Each objective is assigned a weighting value. The weights range roughly from 1 to 8, and indicate the relative importance of each objective. Objectives with higher weights will be covered by more exam questions.

Obj 5: Create, monitor, and kill processes Weight of objective: 5

Includes running jobs in the foreground and background (Chapter 9), bringing a job from the background to the foreground and vise versa, monitoring active processes, sending signals to processes, and killing processes. Includes using commands ps, top, kill, bg, fg, and jobs (Chapter 9).

B.1 Exam Details for 101 General L INUX, part I

This is a required exam for certication level I. It covers fundamental system administration activities that are common across all avors of L INUX.

Obj 6: Modify process execution priorities Weight of objective: 2

Run a program with higher or lower priority, determine the priority of a process, change the priority of a running process (Section 9.7). Includes the command nice and its relatives (Section 9.7).

Topic 1.3: GNU and U NIX Commands

Obj 1: Work Effectively on the U NIX command line Weight of objective: 4 531

Obj 7: Perform searches of text les making use of regular expressions Weight of objective: 3
Includes creating simple regular expressions and using related tools such as grep and sed to perform searches (Chapters 5 and 8).

532

B. LPI Certication Cross-Reference

B.1. Exam Details for 101

B. LPI Certication Cross-Reference

Topic 2.4: Devices, L INUX File Systems, Filesystem Hierarchy Standard

Obj 1: Create partitions and lesystems Weight of objective: 3
Create disk partitions using fdisk, create hard drive and other media lesystems using mkfs (Chapter 19).

Understand the lesystem hierarchy standard, know standard le locations, know the purpose of various system directories, nd commands and les. Involves using the commands: nd, locate, which, updatedb . Involves editing the le: /etc/updatedb.conf (Section 4.14 and Chapters 17 and 35).

Topic 2.6: Boot, Initialization, Shutdown, Run Levels

Obj 1: Boot the system Weight of objective: 3
Guide the system through the booting process, including giving options to the kernel at boot time, and check the events in the log les. Involves using the commands: dmesg (lilo). Involves reviewing the les: /var/log/messages, /etc/lilo.conf, /etc/conf.modules /etc/modules.conf (Sections 21.4.8 and 42.5.1 and Chapters 31 and 32).

Obj 2: Maintain the integrity of lesystems Weight of objective: 5

Verify the integrity of lesystems, monitor free space and inodes, x simple lesystem problems. Includes commands fsck, du, df (Chapter 19).

Obj 3: Control lesystem mounting and unmounting Weight of objective: 3

Mount and unmount lesystems manually, congure lesystem mounting on bootup, congure user-mountable removable le systems. Includes managing le /etc/fstab (Chapter 19).

Obj 2: Change runlevels and shutdown or reboot system Weight of objective: 3

Securely change the runlevel of the system, specically to single user mode, halt (shutdown) or reboot. Make sure to alert users beforehand, and properly terminate processes. Involves using the commands: shutdown, init (Chapter 32).

Obj 4: Set and view disk quota Weight of objective: 1

Setup disk quota for a lesystem, edit user quota, check user quota, generate reports of user quota. Includes quota, edquota, repquota, quotaon commands. (Quotas are not covered but are easily learned form the Quota mini-HOWTO.)

Topic 1.8: Documentation

Obj 1: Use and Manage Local System Documentation Weight of objective: 5
Use and administer the man facility and the material in /usr/doc/. Includes nding relevant man pages, searching man page sections, nding commands and manpages related to one, conguring access to man sources and the man system, using system documentation stored in /usr/doc/ and related places, determining what documentation to keep in /usr/doc/ (Section 4.7 and Chapter 16; you should also study the man page of the man command itself).

Obj 5: Use le permissions to control access to les Weight of objective: 3

Set permissions on les, directories, and special les, use special permission modes such as suid and sticky bit, use the group eld to grant le access to workgroups, change default le creation mode. Includes chmod and umask commands. Requires understanding symbolic and numeric permissions (Chapter 14).

Obj 6: Manage le ownership Weight of objective: 2

Change the owner or group for a le, control what group is assigned to new les created in a directory. Includes chown and chgrp commands (Chapter 11).

Obj 2: Find L INUX documentation on the Internet Weight of objective: 2

Find and use L INUX documentation at sources such as the L INUX Documentation Project, vendor and third-party websites, newsgroups, newsgroup archives, mailing lists (Chapter 13).

Obj 7: Create and change hard and symbolic links Weight of objective: 2
Create hard and symbolic links, identify the hard links to a le, copy les by following or not following symbolic links, use hard and symbolic links for efcient system administration (Chapter 15).

Obj 3: Write System Documentation Weight of objective: 1

Write documentation and maintain logs for local conventions, procedures, conguration and conguration changes, le locations, applications, and shell scripts. (You should learn how to write a man page yourself. There are many man pages to copy as examples. It is difcult to say what the LPI had in mind for this objective.)

Obj 8: Find system les and place les in the correct location Weight of objective: 2 533

534

B. LPI Certication Cross-Reference

B.1. Exam Details for 101

B.2. Exam Details for 102

B. LPI Certication Cross-Reference

Obj 4: Provide User Support Weight of objective: 1

Provide technical assistance to users via telephone, email, and personal contact. (This is not covered. Providing user support can be practiced by answering questions on the newsgroups or mailing lists.)

B.2 Exam Details for 102 General L INUX, part II

Topic 1.1: Hardware and Architecture
Obj 1: Congure fundamental system hardware Weight of objective: 3
Demonstrate a proper understanding of important BIOS settings, set the date and time, ensure IRQs and I/O addresses are correct for all ports including serial and parallel, make a note of IRQs and I/Os, be aware of the issues associated with drives larger than 1024 cylinders (Chapters 3 and 42).

Topic 2.11: Administrative Tasks

Obj 1: Manage users and group accounts and related system les Weight of objective: 7
Add, remove, suspend user accounts, add and remove groups, change user/group info in passwd/group databases, create special purpose and limited accounts. Includes commands useradd, userdel, groupadd, gpasswd, passwd, and le passwd, group, shadow, and gshadow. (Chapter 11. You should also study the useradd and groupadd man pages in detail.)

Obj 2: Setup SCSI and NIC devices Weight of objective: 4

Manipulate the SCSI BIOS to detect used and available SCSI IDs, set the SCSI ID to the correct ID number for the boot device and any other devices required, format the SCSI drivelow level with manufacturers installation toolsand properly partition and system format with L INUX fdisk and mke2fs, set up NIC using manufacturers setup tools setting the I/O and the IRQ as well as the DMA if required. (Sections 42.6.3 and 42.6.9. Each hardware vendor has their own specic tools. There are few such NICs still left to practice on.)

Obj 2: Tune the user environment and system environment variables Weight of objective: 4
Modify global and user proles to set environment variable, maintain skel directories for new user accounts, place proper commands in path. Involves editing /etc/prole and /etc/skel/ (Chapter 11 and Section 20.8).

Obj 3: Congure and use system log les to meet administrative and security needs Weight of objective: 3
Congure the type and level of information logged, manually scan log les for notable activity, arrange for automatic rotation and archiving of logs, track down problems noted in logs. Involves editing /etc/syslog.conf (Sections 21.4.8 and 21.4.9).

Obj 3: Congure modem, sound cards Weight of objective: 3

Ensure devices meet compatibility requirements (particularly that the modem is NOT a winmodem), verify that both the modem and sound card are using unique and correct IRQs, I/O, and DMA addresses, if the sound card is PnP install and run sndcong and isapnp, congure modem for outbound dialup, congure modem for outbound PPP SLIP CSLIP connection, set serial port for 115.2 Kbps (Sections 42.6.1, 42.6.12, and 42.7 and Chapters 34 and 41).

Obj 4: Automate system administration tasks by scheduling jobs to run in the future Weight of objective: 4
Use cron to run jobs at regular intervals, use at to run jobs at a specic time, manage cron and at jobs, congure user access to cron and at services (Chapter 37).

Topic 2.2: L INUX Installation and Package Management

Obj 1: Design hard-disk layout Weight of objective: 2
Design a partitioning scheme for a L INUX system, depending on the hardware and system use (number of disks, partition sizes, mount points, kernel location on disk, swap space). (Chapter 19.)

Obj 5: Maintain an effective data backup strategy Weight of objective: 3

Plan a backup strategy, backup lesystems automatically to various media, perform partial and manual backups, verify the integrity of backup les, partially or fully restore backups (Section 4.17 and Chapter 18).

Obj 2: Install a boot manager Weight of objective: 3 536

535

B. LPI Certication Cross-Reference

B.2. Exam Details for 102

B. LPI Certication Cross-Reference

Select, install and congure a boot loader at an appropriate disk location. Provide alternative and backup boot options (like a boot oppy disk). Involves using the command: lilo . Involves editing the le: /etc/lilo.conf (Chapter 31).

Obj 3: Make and install programs from source Weight of objective: 5

Manage (compressed) archives of les (unpack tarballs), specically GNU source packages. Install and congure these on your systems. Do simple manual customization of the Makele if necessary (like paths, extra include dirs) and make and install the executable. Involves using the commands: gunzip, tar, ./congure, make, make install . Involves editing the les: ./Makele (Chapter 24).

Learn which functionality is available through loadable kernel modules, and manually load and unload the modules as appropriate. Involves using the commands: lsmod, insmod, rmmod, modinfo, modprobe. Involves reviewing the les: /etc/modules.conf /etc/conf.modules (* depends on distribution *), /lib/modules/{kernel-version}/modules.dep (Chapter 42).

Obj 2: Recongure, build, and install a custom kernel and modules Weight of objective: 4
Obtain and install approved kernel sources and headers (from a repository at your site, CD, kernel.org, or your vendor); customize the kernel conguration (i.e., recongure the kernel from the existing .cong le when needed, using oldcong, menucong or xcong); Make a new L INUX kernel and modules; Install the new kernel and modules at the proper place; Recongure and run lilo. N.B.: This does not require to upgrade the kernel to a new version (full source nor patch). Requires the commands: make (dep, clean, menucong, bzImage, modules, modules install), depmod, lilo. Requires reviewing or editing the les: /usr/src/linux/.cong , /usr/src/linux/Makele, /lib/modules/{kernelversion}/modules.dep, /etc/conf.modules /etc/modules.conf, /etc/lilo.conf (Chapter 42).

Obj 4: Manage shared libraries Weight of objective: 3

Determine the dependencies of executable programs on shared libraries, and install these when necessary. Involves using the commands: ldd, ldcong . Involves editing the les: /etc/ld.so.conf (Chapter 23).

Obj 5: Use Debian package management Weight of objective: 5

Use the Debian package management system, from the command line (dpkg) and with interactive tools (dselect). Be able to nd a package containing specic les or software; select and retrieve them from archives; install, upgrade or uninstall them; obtain status information like version, content, dependencies, integrity, installation status; and determine which packages are installed and from which package a specic le has been installed. Be able to install a non-Debian package on a Debian system (Chapter 24). Involves using the commands and programs: dpkg, dselect, apt, apt-get, alien . Involves reviewing or editing the les and directories: /var/lib/dpkg/* .

Topic 1.7: Text Editing, Processing, Printing

Obj 1: Perform basic le editing operations using vi Weight of objective: 2
Edit text les using vi. Includes vi navigation, basic modes, inserting, editing and deleting text, nding text, and copying text (Chapter 6).

Obj 2: Manage printers and print queues Weight of objective: 2

Monitor and manage print queues and user print jobs, troubleshoot general printing problems. Includes the commands: lpc, lpq, lprm and lpr . Includes reviewing the le: /etc/printcap (Chapter 21).

Obj 6:Use Red Hat Package Manager (rpm) Weight of objective: 6

Use rpm from the command line. Familiarize yourself with these tasks: Install a package, uninstall a package, determine the version of the package and the version of the software it contains, list the les in a package, list documentation les in a package, list conguration les or installation or uninstallation scripts in a package, nd out for a certain le from which package it was installed, nd out which packages have been installed on the system (all packages, or from a subset of packages), nd out in which package a certain program or le can be found, verify the integrity of a package, verify the PGP or GPG signature of a package, upgrade a package. Involves using the commands and programs: rpm, grep (Chapter 24).

Obj 3: Print les Weight of objective: 1

Submit jobs to print queues, convert text les to postscript for printing. Includes lpr command (Section 21.6).

Obj 4: Install and congure local and remote printers Weight of objective: 3
Install a printer daemon, install and congure a print lter (e.g.: apslter, magiclter). Make local and remote printers accessible for a L INUX system, including postscript, non-postscript, and Samba printers. Involves the daemon: lpd . Involves editing or reviewing the les and directories: /etc/printcap , /etc/apslterrc , /usr/lib/apslter/lter/*/ , /etc/magiclter/*/ , /var/spool/lpd/*/ (why not to use apsfilter is discussed in Section 21.9.2).

Topic 1.5: Kernel

Obj 1: Manage kernel modules at runtime Weight of objective: 3 537

538

B. LPI Certication Cross-Reference

B.2. Exam Details for 102

B. LPI Certication Cross-Reference

Topic 1.9: Shells, Scripting, Programming, Compiling

Obj 1: Customize and use the shell environment Weight of objective: 4
Customize your shell environment: set environment variables (e.g. PATH) at login or when spawning a new shell; write bash functions for frequently used sequences of commands. Involves editing these les in your home directory: .bash prole .bash login .prole ; .bashrc ; .bash logout ; .inputrc (Chapter 20).

aterm etc.), verify and resolve library dependency issues for X applications, export an X-display to a client workstation. Commands: Files: .xinitrc, .Xdefaults, various .rc les. (The xinit, startx, and xdm man pages provide this information.)

Topic 1.12: Networking Fundamentals

Obj 1: Fundamentals of TCP/IP Weight of objective: 4
Demonstrate an understanding of network masks and what they mean (i.e. determine a network address for a host based on its subnet mask), understand basic TCP/IP protocols (TCP, UDP, ICMP) and also PPP, demonstrate an understanding of the purpose and use of the more common ports found in /etc/services (20, 21, 23, 25, 53, 80, 110, 119, 139, 143, 161), demonstrate an correct understanding of the function and application of a default route. Execute basic TCP/IP tasks: FTP, anonymous FTP, telnet, host, ping, dig, traceroute, whois (Chapters 25 and 26).

Obj 2: Customize or write simple scripts Weight of objective: 5

Customize existing scripts (like paths in scripts of any language), or write simple new (ba)sh scripts. Besides use of standard sh syntax (loops, tests), be able to do things like: command substitution and testing of command return values, test of le status, and conditional mailing to the superuser. Make sure the correct interpreter is called on the rst (#!) line, and consider location, ownership, and execution- and suid-rights of the script (Chapter 20; setuid is covered in Sections 33.2 and 36.2.10 from a slightly more utilitarian angle).

Topic 2.10: X
Obj 1: Install and congure XFree86 Weight of objective: 4
Verify that the video card and monitor are supported by an X server, install the correct X server, congure the X server, install an X font server, install required fonts for X (may require a manual edit of /etc/X11/XF86Cong in the Files section), customize and tune X for videocard and monitor. Commands: XF86Setup, xf86cong. Files: /etc/X11/XF86Cong, .xresources (Chapter 43).

Obj 2: (superseded) Obj 3: TCP/IP troubleshooting and conguration Weight of objective: 10

Demonstrate an understanding of the techniques required to list, congure and verify the operational status of network interfaces, change, view or congure the routing table, check the existing route table, correct an improperly set default route, manually add/start/stop/restart/delete/recongure network interfaces, and congure L INUX as a DHCP client and a TCP/IP host and debug associated problems. May involve reviewing or conguring the following les or directories: /etc/HOSTNAME /etc/hostname, /etc/hosts, /etc/networks, /etc/host.conf, /etc/resolv.conf, and other network conguration les for your distribution. May involve the use of the following commands and programs: dhcpd, host, hostname (domainname, dnsdomainname), ifcong, netstat, ping, route, traceroute, the network scripts run during system initialization (Chapters 25 and 27).

Obj 2: Setup XDM Weight of objective: 1

Turn xdm on and off, change the xdm greeting, change default bitplanes for xdm, set-up xdm for use by X-stations (see the xdm man page for comprehensive information).

Obj 4: Congure and use PPP Weight of objective: 4

Dene the chat sequence to connect (given a login example), setup commands to be run automatically when a PPP connection is made, initiate or terminate a PPP connection, initiate or terminate an ISDN connection, set PPP to automatically reconnect if disconnected (Chapter 41).

Obj 3: Identify and terminate runaway X applications Weight of objective: 1

Identify and kill X applications that wont die after user ends an X-session. Example: netscape, tkrat, etc.

Topic 1.13: Networking Services

Obj 1: Congure and manage inetd and related services Weight of objective: 5
Congure which services are available through inetd, use tcpwrappers to allow or deny services on a host-by-host basis, manually start, stop, and restart Internet services, congure ba-

Obj 4: Install and customize a Window Manager Environment Weight of objective: 4

Select and customize a system-wide default window manager and/or desktop environment, demonstrate an understanding of customization procedures for window manager menus, congure menus for the window manager, select and congure the desired x-terminal (xterm, rxvt,

539

540

B. LPI Certication Cross-Reference

B.2. Exam Details for 102

B. LPI Certication Cross-Reference

sic network services including telnet and ftp. Includes managing inetd.conf, hosts.allow, and hosts.deny (Chapter 29).

Obj 2: Operate and perform basic conguration of sendmail Weight of objective: 5

Modify simple parameters in sendmail cong les (modify the DS value for the Smart Host if necessary), create mail aliases, manage the mail queue, start and stop sendmail, congure mail forwarding (.forward), perform basic troubleshooting of sendmail. Does not include advanced custom conguration of sendmail. Includes commands mailq, sendmail, and newaliases. Includes aliases and mail/ cong les (Chapter 30).

Implement shadowed passwords, turn off unnecessary network services in inetd, set the proper mailing alias for root and setup syslogd, monitor CERT and BUGTRAQ, update binaries immediately when security problems are found (Chapter 44).

Obj 3: Setup user level security Weight of objective: 2

Set limits on user logins, processes, and memory usage (Section 11.7.5).

Obj 3: Operate and perform basic conguration of apache Weight of objective: 3

Modify simple parameters in apache cong les, start, stop, and restart httpd, arrange for automatic restarting of httpd upon boot. Does not include advanced custom conguration of apache. Includes managing httpd conf les (Chapter 36).

Obj 4: Properly manage the NFS, smb, and nmb daemons Weight of objective: 4
Mount remote lesystems using NFS, congure NFS for exporting local lesystems, start, stop, and restart the NFS server. Install and congure Samba using the included GUI tools or direct edit of the /etc/smb.conf le (Note: this deliberately excludes advanced NT domain issues but includes simple sharing of home directories and printers, as well as correctly setting the nmbd as a WINS client). (Chapters 28 and 39.)

Obj 5: Setup and congure basic DNS services Weight of objective: 3

Congure hostname lookups by maintaining the /etc/hosts, /etc/resolv.conf, /etc/host.conf, and /etc/nsswitch.conf les, troubleshoot problems with local caching-only name server. Requires an understanding of the domain registration and DNS translation process. Requires understanding key differences in cong les for bind 4 and bind 8. Includes commands nslookup, host. Files: named.boot (v.4) or named.conf (v.8) (Chapters 27 and 40).

Topic 1.14: Security

Obj 1: Perform security admin tasks Weight of objective: 4
Congure and use TCP wrappers to lock down the system, list all les with SUID bit set, determine if any package (.rpm or .deb) has been corrupted, verify new packages prior to install, use setgid on dirs to keep group ownership consistent, change a users password, set expiration dates on users passwords, obtain, install and congure ssh (Chapter 44).

Obj 2: Setup host security Weight of objective: 4 541 542

C.2. RH300

C. RHCE Certication Cross-Reference

C.2 RH300
This certication seems to be for administrators of non-L INUX systems who want to extend their knowledge. The requirements below lean toward understanding available L INUX alternatives and features, rather than expecting the user to actually congure anything complicated. Note that I abbreviate the RedHat Installation Guide(s) as RHIG. This refers to the install help in the installation program itself or, for RedHat 6.2 systems, the HTML installation guide on the CD. It also refers to the more comprehensive online documentation at http://www.redhat.com/support/manuals/.

Appendix C

Unit 1: Hardware selection and RedHat installation

RedHat Certied Engineer Certication Cross-Reference

RedHat has encouraged a larger number of overlapping courses, some of which contain lighter and more accessible material. They concentrate somewhat on RedHat specic issues that are not always applicable to other distributions. In some areas they expect more knowledge than the LPI, so it is worth at least reviewing RedHats requirements for purposes of self-evaluation. The information contained in this appendix was gathered from discussions with people who had attended the RedHat courses. This is intended purely for cross-referencing purposes and is possibly outdated. By no means should it be taken as denitive. Visit http://redhat.com/training/rhce/courses/ for the ofcial guide.

For each objective, the relevant chapter or section from this book is referenced in parentheses.

Finding Web docs. Using HOWTOs to locate supported hardware (Chapter 16). Knowledge of supported architectures and SMP support (Chapter 42). Use of kudzu (I do not cover kudzu and recommend that you uninstall it). Hardware conceptsIRQ, PCI, EISA, AGP, and I/O ports (Chapters 3 and 42). isapnp, pciscan (Chapter 42). Concepts of L INUX support for PCMCIA, PS/2, tapes, scanners, USB (Chapter 42). Concepts of serial, parallel, SCSI, IDE, CD-ROM and oppy devices, and their /dev/ listings (Chapter 18). hdparm (hdparm(8)). Concepts of IDE geometry, BIOS limitations (Chapter 19). Disk sector and partition structure. Use of fdisk, cfdisk, and diskdruid (Chapter 19). Creation of a partitioning structure (Chapter 19). Management of swap, native, and foreign partitions during installation (RHIG). Concept of distribution of directories over different partitions (Chapter 19). Conguring lilo on installation (Chapter 31 refers to general use of lilo). BIOS conguration (Chapter 3). Conceptual understanding of different disk images. Creating and booting disk images from their boot.img, bootnet.img, or pcmcia.img (RHIG). Use of the installer to create RAID devices (RHIG). Package selection (RHIG). video conguration (Chapter 43 and RHIG).

C.1 RH020, RH030, RH033, RH120, RH130, and RH133

These courses are beneath the scope of this book: They cover L INUX from a user and desktop perspective. Although they include administrative tasks, they keep away from technicalities. They often prefer graphical conguration programs to do administrative tasks. One objective of one of these courses is conguring Gnome panel applets; another is learning the pico text editor.

Unit 2: Conguring and administration

- Using setup, mouseconfig, Xconfigurator, kbdconfig, timeconfig, netconfig, authconfig, sndconfig. (These are higher level interactive utilities than the ones I cover in Chapter 42 and elsewhere. Run each of these commands for a demo.) - Understanding /etc/sysconfig/network-scripts/ifcfg-* (Chapter 25). - Using netcfg or ifconfig (Chapter 25). - Using ifup, ifdown, rp3, usernet, and usernetctl (Chapter 25).

543

544

C. RHCE Certication Cross-Reference

C.2. RH300

C. RHCE Certication Cross-Reference

- Using pnpdump, isapnp and editing /etc/isapnp.conf (Chapter 42). - Conceptual understanding of /etc/conf.modules, esd, and kaudioserver (Chapter 42; man pages for same). - Using mount, editing /etc/fstab (Chapter 19). - Using lpr, lpc, lpq, lprm, printtool and understanding concepts of /etc/printcap (Chapter 21). - Virtual consoles concepts: changing in /etc/inittab (Chapter 32). - Using useradd, userdel, usermod, and passwd (Chapter 11). - Creating accounts manually and with userconf and with linuxconf. (The use of graphical tools is discouraged by this book.) - Understanding concepts of the /etc/passwd and /etc/group les and /etc/skel and contents (Chapter 11). - Editing bashrc, .bashrc, /etc/profile, /etc/profile.d (Chapter 20). - General use of linuxconf. (The use of graphical tools is discouraged by this book.) - Using cron, anacron, editing /var/spool/cron/<username> and /etc/crontab. tmpwatch, logrotate, and locate cron jobs. - Using syslogd, klogd, /etc/syslog.conf, swatch, logcheck. - Understanding and using rpm. Checksums, le listing, forcing, dependencies, querying, verifying querying tags, provides, and requires. FTP and HTTP installs, rpmfind, gnorpm, and kpackage (Chapter 24). - Building .src.rpm les. Customizing and rebuilding packages. (See the RPM-HOWTO.) - /usr/sbin/up2date. (The use of package is discouraged by this book.) - Finding documentation (Chapter 16).

- System startup scripts initialization sequences. inittab, switching run levels. Conceptual understanding of various /etc/rc.d/ les. SysV scripts, chkconfig, ntsysv, tksysv, ksysv (Chapter 32). - Conguring software RAID. Using raidtools to activate and test RAID devices (see the RAID-HOWTO). - Modules Management. modprobe, depmod, lsmod, insmod, rmmod commands. kernelcfg. Editing of /etc/conf.modules, aliasing and optioning modules (Chapter 42). - Concepts of kernel source, .rpm versions, kernel versioning system. Conguring, compiling and installing kernels (Chapter 42).

Unit 5: Basic network services

TCP/IP concepts. inetd. Port concepts and service-port mappings (Chapters 25 and 26). apache, cong les, virtual hosts (Chapter 36). sendmail, cong les, mailconf, m4 macro concepts (Chapter 30). POP and IMAP concepts (Chapters 29 and 30). named conguration (Chapter 40). FTP conguration. (I did not cover FTP because of the huge number of FTP services available. It is recommended that you try the vsftpd package.) conguration, /etc/rc.d/init.d/netfs (Chapter 28). smbd, le-sharing and print-sharing concepts. Security concepts cong le overview. Use of testparam, smbclient, nmblookup, smbmount, Windows authentication concepts (Chapter 39). dhcpd and BOOTP, cong les and concepts. Conguration with netcfg, netconfig or linuxconf. using pump (see the DHCP mini-HOWTO). Understanding squid caching and forwarding concepts. (The squid conguration le /etc/squid/squid.conf provides ample documentation for actually setting up squid.) Overview of lpd, mars-nwe, time services, and news services (Chapter 21).

Unit 3: Alternative installation methods

- Laptops, PCMCIA, cardmanager, and apm. (See the RHIG, PCMCIA-HOWTO and Laptop-HOWTO.) - Multiboot systems, boot options, and alternative boot image conguration (Chapter 31). - Network installations using netboot.img (RHIG). - Serial console installation (RHIG?). - Kickstart concepts.

Unit 6: X Window System

- X client server architecture (Section 43.1). - Use of Xconfigurator, xf86config, XF86Setup, and concepts of /etc/X11/XF86Config (Section 43.6.3). - Knowledge of various window managers, editing /etc/sysconfig/desktop. Understanding of concepts of different user interfaces: Gnome, KDE. Use of switchdesk (Section 43.3.4). - init run level 5 concepts, xdm, kdm, gdm, prefdm alternatives (Section 43.9). - xinit, xinitrc concepts. User cong les .xsession and .Xclients (see xinit(1), xdm(1), startx(1), and read the scripts under /etc/X11/xinit/ and /etc/X11/xdm).

Unit 4: Kernel
- /proc le system concepts and purpose of various subdirectories (see Section 42.4 and the index entries for /proc/). Tuning parameters with /etc/sysctl.conf (see sysctl.conf(5)). - Disk quotas. quota, quotaon, quotaoff, edquota, repquota, quotawarn, quotastats. (Quotas are not covered but are easily learned form the Quota miniHOWTO.)

545

546

C. RHCE Certication Cross-Reference

C.3. RH220 (RH253 Part 1)

C. RHCE Certication Cross-Reference

- Use of xhost (Section 43.3.5). Security issues. DISPLAY environment variable. Remote displays (Section 43.3.2). - xfs concepts (Section 43.12).

Unit 1: DNS
A treatment of bind, analogous to Topic 1.13, Obj 5 of LPI (page 541). Expects exhaustive understanding of the Domain Name System, an understanding of SOA, NS, A, CNAME, PTR, MX and HINFO records, ability to create master domain servers from scratch, caching-only servers, and round-robin load sharing conguration (Chapter 40).

Unit 7: Security
- Use of tcp wrappers (Chapter 29). User and host based access restrictions. PAM access. Port restriction with ipchains (see the Firewall-HOWTO). - PAM concepts. Editing of /etc/pam.d, /etc/security cong les. PAM documentation (see /usr/share/doc/pam-0.72/txts/pam.txt). - NIS concepts and cong les. ypbind, yppasswd ypserv, yppasswdd, makedbm, yppush (see the NIS-HOWTO). - LDAP concepts. OpenLDAP package, slapd, ldapd, slurpd, and cong les. PAM integration. - inetd concepts. Editing of /etc/inetd.conf, interface to tcp wrappers. Editing of /etc/hosts.allow and /etc/hosts.deny. portmap, tcpdchk, tcpdmatch, twist (see the LDAP-HOWTO). - ssh client server and security concepts (Chapters 12 and 44).

Unit 2: Samba
Overview of SMB services and concepts. Conguring Samba for le and print sharing. Using Samba client tools. Using linuxconf and swat. Editing /etc/smb.conf. Understanding types of shares. Support Wins. Setting authentication method. Using client utilities (Chapter 39).

Unit 3: NIS
Conceptual understanding of NIS. NIS master and slave congure. Use of client utilities. LDAP concepts. OpenLDAP package, slapd, ldapd, slurpd, and cong les (see the NIS-HOWTO).

Unit 8: Firewalling, routing and clustering, troubleshooting

- Static and dynamic routing with concepts. /etc/sysconfig/static-routes. Use of linuxconf and netcfg to edit routes. (Use of graphical tools is discouraged by this book.) - Forwarding concepts. Concepts of forwarding other protocols: X.25, frame-relay, ISDN, and PPP. (By concepts of I take it to mean that mere knowledge of these features is sufcient. See also Chapter 41.) - ipchains and ruleset concepts. Adding, deleting, listing, ushing rules. Forwarding, masquerading. Protocol-specic kernel modules (see the Firewall-HOWTO). - High availability concepts. Concepts of lvs, pulse, nanny, cong les, and web-based conguration. Piranha, failover concepts. (A conceptual understanding again.) - High performance clustering concepts. Parallel virtual machine for computational research (conceptual understanding only). - Troublshooting: Networking (Chapter 25), X (Chapter 43), booting (Chapter 31), DNS (Chapters 27 and 40), authentication (Chapter 11), le system corruption (Section 19.5). - mkbootdisk and rescue oppy concepts. Use of the rescue disk environment and available commands (see mkbootdisk(8)).

Unit 4: Sendmail and procmail

Understanding of mail spooling and transfer. Understanding the purpose of all sendmail cong les. Editing cong le for simple client (i.e., forwarding) conguration. Editing /etc/sendmail.mc, /etc/mail/virtusertable, /etc/mail/access. Restricting relays. Viewing log les. Creating simple .procmail folder and email redirectors. (Chapter 30. Also see The Sendmail FAQ http://www.sendmail.org/faq/ as well as procmail(1), procmailrc(6), and procmailex(5).)

Unit 5: Apache
Conguring virtual hosts. Adding MIME types. Manipulating directory access and directory aliasing. Allowing restricting of CGI access. Setting up user and password databases. Understanding important modules (Chapter 36).

Unit 6: pppd and DHCP

C.3 RH220 (RH253 Part 1)

RH220 is the networking module. It covers services sparsely, possibly intending that the student learn only the bare bones of what is necessary to congure a service.

Setting up a basic pppd server. Adding dial-in user accounts. Restricting users. Understanding dhcpd and BOOTP cong les and concepts. Conguring with netcfg, netconfig, or linuxconf. Using pump. Editing /etc/dhcpd.conf. (Chapter 41. See also the DHCPHOWTO.)

547

548

C. RHCE Certication Cross-Reference

C.4. RH250 (RH253 Part 2)

C. RHCE Certication Cross-Reference

C.4 RH250 (RH253 Part 2)

RH250 is the security module. It goes through basic administration from a security perspective.

Unit 7: Security tools

Concepts of nessus, SAINT, SARA, SATAN. Concepts of identd. Use of sniffit, tcpdump, traceroute, ping -f, ethereal, iptraf, mk-ftp-stats, lurkftp, mrtg, netwatch, webalizer, trafshow. (These tools may be researched on the web.)

Unit 1: Introduction
Understanding security requirements. Basic terminology: hacker, cracker, denial of service, virus, trojan horse, worm. Physical security and security policies (Chapter 44).

Unit 2: Local user security

Understanding user accounts concepts, restricting access based on groups. Editing pam cong les. /etc/nologin; editing /etc/security/ les. Using console group, cug; conguring and using clobberd and sudo. Checking logins in log les. Using last (Chapters 11 and 44).

Unit 3: Files and le system security

Exhaustive treatment of groups and permissions. chattr and lsattr commands. Use of find to locate permission problems. Use of tmpwatch. Installation of tripwire. Managment of NFS exports for access control (Chapters 14, 28, and 44).

Unit 4: Password security and encryption

Encryption terms: Public/Private Key, GPG, one-way hash, MD5. xhost, xauth. ssh concepts and features. Password-cracking concepts (Section 11.3 and Chapter 12).

Unit 5: Process security and monitoring

Use PAM to set resource limits (Section 11.7.5). Monitor process memory usage and CPU consumption; top, gtop, kpm, xosview, xload, xsysinfo. last, ac, accton, lastcomm (Chapter 9). Monitor logs with swatch (see swatch(5) and swatch(8)).

Unit 6: Building rewalls

ipchains and ruleset concepts. Adding, deleting, listing, ushing rules. Forwarding, manyto-one and one-to-one masquerading. Kernels options for rewall support. Static and dynamic routing with concepts (see the Firewall-HOWTO). /etc/sysconfig/static-routes. Use of linuxconf and netcfg to edit routes. tcp wrappers (Chapter 29).

549

550

D.1. L INUX Overview

D. L INUX Advocacy FAQ

for the DEC Alpha machine and AIX for the PowerPC/RS6000. Because the U NIX name is a registered trademark, most systems are not called U NIX. Some freely available U NIX systems are NetBSD, FreeBSD, and OpenBSD and also enjoy widespread popularity. U NIX systems are multitasking and multiuser systems, meaning that multiple concurrent users running multiple concurrent programs can connect to and use the same machine.

Appendix D

What are U NIX systems used for? What can L INUX do?

L INUX Advocacy Frequently-Asked-Questions

The capabilities of L INUX are constantly expanding. Please consult the various Internet resources listed for up-to-date information.

U NIX systems are the backbone of the Internet. Heavy industry, mission-critical applications, and universities have always used U NIX systems. High-end servers and multiuser mainframes are traditionally U NIX based. Today, U NIX systems are used by large ISPs through to small businesses as a matter of course. A U NIX system is the standard choice when a hardware vendor comes out with a new computer platform because U NIX is most amenable to being ported. U NIX systems are used as database, le, and Internet servers. U NIX is used for visualization and graphics rendering (as for some Hollywood productions). Industry and universities use U NIX systems for scientic simulations and U NIX clusters for number crunching. The embedded market (small computers without operators that exist inside appliances) has recently turned toward L INUX systems, which are being produced in the millions. L INUX itself can operate as a web, le, SMB (WinNT), Novell, printer, FTP, mail, SQL, masquerading, rewall, and POP server to name but a few. It can do anything that any other network server can do, more efciently and reliably. L INUXs up-and-coming graphical user interfaces (GUI) are the most functional and aesthetically pleasing ever to have graced the computer screen. L INUX has now moved into the world of the desktop.

D.1

L INUX Overview

This section covers questions that pertain to L INUX as a whole.

What other platforms does L INUX run on including the PC? What is L INUX?
L INUX is the core of a free U NIX operating system for the PC and other hardware platforms. Developement of this operating system started in 1984; it was called the GNU project of the Free Software Foundation (FSF). The L INUX core (or kernel), named after its author, Linus Torvalds, began development in 1991the rst usable releases where made in 1993. L INUX is often called GNU/L INUX because much of the OS (operating system) results from the efforts of the GNU project. U NIX systems have been around since the 1960s and are a proven standard in industry. L INUX is said to be POSIX compliant, meaning that it conrms to a certain denite computing standard laid down by academia and industry. This means that L INUX is largely compatible with other U NIX systems (the same program can be easily ported to run on another U NIX system with few (sometimes no) modications) and will network seamlessly with other U NIX systems. Some commercial U NIX systems are IRIX (for Silicon Graphics); Solaris or SunOS for Sun Microsystem SPARC workstations; HP U NIX for Hewlett Packard servers; SCO for the PC; OSF L INUX runs on 386/486/Pentium processors. DEC 64-bit Alpha processors. Motorola 680x0 processors, including Commodore Amiga, Atari-ST/TT/Falcon and HP Apollo 68K. Sun Microsystems SPARC workstations, including sun4c, sun4m, sun4d, and sun4u architectures. Multiprocessor machines are supported as is full 64-bit support on the UltraSPARC. Advanced Risc Machine (ARM) processors. MIPS R3000/R4000 processors, including Silicon Graphics machines. PowerPC machines. Intel Architecture 64-bit processors.

551

552

D. L INUX Advocacy FAQ

D.1. L INUX Overview

D. L INUX Advocacy FAQ

IBM 390 mainframe. ETRAX-100 processor. Other projects are in various stages of completion. For example, you may get L INUX up and running on many other hardware platforms, but it would take some time and expertise to install, and you might not have graphics capabilities. Every month or so support is announced for some new esoteric hardware platform. Watch the Linux Weekly News http://lwn.net/ to catch these.

Scientic Applications for Linux (SAL) http://SAL.KachinaTech.COM/index.shtml Announcements for new software are mostly made on Fresh Meat http://freshmeat.net/ The Linux Weekly News brings up-to-date info covering a wide range of L INUX issues: Linux Weekly News http://lwn.net/ Three major L INUX desktop projects are:

What is meant by GNU/L INUX as opposed to L INUX?

(See also What is GNU? and What is L INUX?.) In 1984 the Free Software Foundation (FSF) set out to create a free U NIX-like system. It is only because of their efforts that the many critical packages that go into a U NIX distribution are available. It is also because of them that a freely available, comprehensive, legally denitive, free-software license is available. Because many of the critical components of a typical L INUX distribution are really just GNU tools developed long before L INUX, it is unfair to merely call a distribution L INUX. The term GNU/L INUX is more accurate and gives credit to the larger part of L INUX.

Gnome Desktop http://www.gnome.org/ KDE Desktop http://www.kde.org/ GNUstep http://gnustep.org/ But dont stop therethere are hundreds more.

What are Debian, RedHat, Caldera, SuSE? Explain the different L INUX distributions.
All applications, network server programs, and utilities that go into a full L INUX machine are free software programs recompiled to run under the L INUX kernel. Most can (and do) actually work on any other of the U NIX systems mentioned above. Hence, many efforts have been made to package all of the utilities needed for a U NIX system into a single collection, usually on a single easily installable CD. Each of these efforts combines hundreds of packages (e.g., the Apache web server is one package, the Netscape web browser is another) into a L INUX distribution. Some of the popular L INUX distributions are: Caldera OpenLinux http://www.calderasystems.com/ Debian GNU/ L INUX http://www.debian.org/ Mandrake http://www.linux-mandrake.com/ RedHat http://www.redhat.com/ Slackware http://www.slackware.com/ SuSE http://www.suse.com/ TurboLinux http://www.turbolinux.com/ There are now about 200 distributions of L INUX. Some of these are single oppy routers or rescue disks, and others are modications of popular existing distributions. Still others have a specialized purpose, like real time work or high security.

What web pages should I look at?

Hundreds of web pages are devoted to L INUX. Thousands of web pages are devoted to different free software packages. A net search will reveal the enormous amount of information available. Three places for general L INUX information are: Alan Coxs Linux web page http://www.linux.org.uk/ Linux Online http://www.linux.org/ Linux International http://www.li.org/ For kernel information, see Linux Headquarters http://www.linuxhq.com/ A very important site is FSF Home Pages http://www.gnu.org/ which is the home page of the Free Software Foundation and explains their purpose and the philosophy of software that can be freely modied and redistributed. Some large indexes of reviewed free and proprietary L INUX software are: Fresh Meat http://freshmeat.net/ Source Forge http://www.sourceforge.net/ Tu Cows http://linux.tucows.com/

553

554

D. L INUX Advocacy FAQ

D.1. L INUX Overview

D.2. L INUX, GNU, and Licensing

D. L INUX Advocacy FAQ

Who developed L INUX?

L INUX was largely developed by the Free Software Foundation http://www.gnu.org/. The Orbiten Free Software Survey http://www.orbiten.org/ came up with the following breakdown of contributors after surveying a wide array of open source packages. The following lists the top 20 contributors by amount of code written: Serial 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Author Free Software Foundation, Inc. Sun Microsystems, Inc. The Regents of the University of California Gordon Matzigkeit Paul Houle Thomas G. Lane The Massachusetts Institute of Technology Ulrich Drepper Lyle Johnson Peter Miller Eric Young login-belabas Lucent Technologies, Inc. Linus Torvalds (uncredited-gdb) Aladdin Enterprises Tim Hudson Carnegie Mellon University James E. Wilson, Robert A. Koeneke ID Software, Inc. Bytes 125565525 20663713 15192791 13599203 11647591 8746848 8513597 6253344 5906249 5871392 5607745 5429114 4991582 4898977 4806436 4580332 4454381 4272613 4272412 4038969 Percentage (11.246%) (1.85%) (1.36%) (1.218%) (1.043%) (0.783%) (0.762%) (0.56%) (0.528%) (0.525%) (0.502%) (0.486%) (0.447%) (0.438%) (0.43%) (0.41%) (0.398%) (0.382%) (0.382%) (0.361%) Projects 546 66 156 267 1 17 38 142 1 3 48 2 5 10 1 27 26 23 2 1

15 16 17 18 19 20

David J. Mackenzie Rich Salz Jean-Loup Gailly eggert@twinsun Josh Macdonald Peter Mattis, Spencer Kimball

337388 365595 2256335 387923 1994755 1981094

(0.03%) (0.032%) (0.202%) (0.034%) (0.178%) (0.177%)

37 35 31 30 28 28

The preceding tables are rough approximations. They do, however, give an idea of the spread of contributions.

Why should I not use L INUX?

If you are a private individual with no U NIX expertise available to help you when you run into problems and you are not interested in learning about the underlying workings of a U NIX system, then you shouldnt install L INUX.

D.2

L INUX, GNU, and Licensing

This section answers questions about the nature of free software and the concepts of GNU.

What is L INUXs license?

The L INUX kernel is distributed under the GNU General Public License (GPL) which is reproduced in Appendix E and is available from the FSF Home Page http://www.gnu.org/. Most of all other software in a typical L INUX distribution is also under the GPL or the LGPL (see below). There are many other types of free software licenses. Each of these is based on particular commercial or moral outlooks. Their acronyms are as follows (as dened by the L INUX Software Map database) in no particular order: PD: Placed in public domain. Shareware: Copyrighted, no restrictions, contributions solicited. MIT: MIT X Consortium license (like that of BSDs but with no advertising requirement). BSD: Berkeley Regents copyright (used on BSD code). Artistic License: Same terms as Perl Artistic License. FRS: Copyrighted, freely redistributable, might have some restrictions on redistribution of modied sources. GPL: GNU General Public License. GPL+LGPL: GNU GPL and Library GPL. restricted: Less free than any of the above. More information on these licenses can be had from the Metalab license List ftp://metalab.unc.edu/pub/Linux/LICENSES

This listing contains the top 20 contributors by number of projects contributed to: Serial 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Author Free Software Foundation, Inc. Gordon Matzigkeit The Regents of the University of California Ulrich Drepper Roland Mcgrath Sun Microsystems, Inc. RSA Data Security, Inc. Martijn Pieterse Eric Young login-vern jot@cray Alfredo K. Kojima The Massachusetts Institute of Technology Digital Equipment Corporation Bytes 125565525 13599203 15192791 6253344 2644911 20663713 898817 452661 5607745 3499616 691862 280990 8513597 2182333 Percentage (11.246%) (1.218%) (1.36%) (0.56%) (0.236%) (1.85%) (0.08%) (0.04%) (0.502%) (0.313%) (0.061%) (0.025%) (0.762%) (0.195%) Projects 546 267 156 142 99 66 59 50 48 47 47 40 38 37

555

556

D. L INUX Advocacy FAQ

D.2. L INUX, GNU, and Licensing

D. L INUX Advocacy FAQ

What is GNU?
GNU (pronounced with a hard G) is an acronym for GNUs Not U NIX. A gnu is a large beast and is the motif of the Free Software Foundation (FSF). GNU is a recursive acronym. Richard Stallman is the founder of the FSF and the creator of the GNU General Public License. One of the purposes of the FSF is to promote and develop free alternatives to proprietary software. The GNU project is an effort to create a free U NIX-like operating system from scratch; the project was started in 1984. GNU represents this software licensed under the GNU General Public Licenseit is called Free software. GNU software is software designed to meet a higher set of standards than its proprietary counterparts. GNU has also become a movement in the computing world. When the word GNU is mentioned, it usually evokes feelings of extreme left-wing geniuses who in their spare time produce free software that is far superior to anything even large corporations can come up with through years of dedicated development. It also means distributed and open development, encouraging peer review, consistency, and portability. GNU means doing things once in the best way possible, providing solutions instead of quick xes and looking exhaustively at possibilities instead of going for the most brightly colored or expedient approach. GNU also means a healthy disrespect for the concept of a deadline and a release schedule.

Fails to build on existing software because of licensing issues. GNU software, on the other hand, is open for anyone to scrutinize. Users can (and do) freely x and enhance software for their own needs, and then allow others the benet of their extensions. Many developers of different areas of expertise collaborate to nd the best way of doing things. Open industry and academic standards are adhered to, to make software consistent and compatible. Collaborated effort between different developers means that code is shared and effort is not replicated. Users have close and direct contact with developers, ensuring that bugs are xed quickly and that user needs are met. Because source code can be viewed by anyone, developers write code more carefully and are more inspired and more meticulous. Possibly the most important reason for the superiority of Free software is peer review. Sometimes this means that development takes longer as more people quibble over the best way of doing things. However, most of the time peer review results in a more reliable product. Another partial reason for this superiority is that GNU software is often written by people from academic institutions who are in the center of IT research and are most qualied to dictate software solutions. In other cases, authors write software for their own use out of their own dissatisfaction for existing proprietary softwarea powerful motivation.

Explain the restrictions of L INUXs freeGNU General Public License.

The following is quoted from the GPL itself. When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things. To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it. For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights.

Why is GNU software better than proprietary software?

Proprietary software is often looked down upon in the free software world for many reasons: The development process is closed to external scrutiny. Users are unable to add features to the software. Users are unable to correct errors (bugs) in the software. Users are not allowed to share the software. The result of these limitations is that proprietary software Does not conform to good standards for information technology. Is incompatible with other proprietary software. Is buggy. Cannot be xed. Costs far more than it is worth. Can do anything behind your back without your knowing. Is insecure. Tries to be better than other proprietary software without meeting real technical and practical needs. Wastes a lot of time duplicating the effort of other proprietary software.

If L INUX is free, where do companies have the right to make money from selling CDs?
See Where do I get L INUX? on page 562.

557

558

D. L INUX Advocacy FAQ

D.2. L INUX, GNU, and Licensing

D.3. L INUX Distributions

D. L INUX Advocacy FAQ

What if Linus Torvalds decided to change the copyright on the kernel? Could he sell out to a company?
This situation is not possible. Because of the legal terms of the GPL, for L INUX to be distributed under a different copyright would require the consent of all 200+ persons that have ever contributed to the L INUX source code. These people come from such a variety of places, that such a task is logistically infeasible. Even if it did happen, new developers would probably rally in deance and continue to work on the kernel as it is. This free kernel would amass more followers and would quickly become the standard, with or without Linus.

D.3

L INUX Distributions

This section covers questions that about how L INUX software is packaged and distributed and how to obtain L INUX.

If everyone is constantly modifying the source, isnt this bad for the consumer? How is the user protected from bogus software?
You as the user are not going to download arbitrary untested software any more than you would if you were using Windows.

What if Linus Torvalds stopped supporting L INUX? What if kernel development split?
There are many kernel developers who have sufcient knowledge to do the job of Linus. Most probably, a team of core developers would take over the task if Linus no longer worked on the kernel. L INUX might even split into different development teams if a disagreement did break out about some programming issue, and it might rejoin later on. This is a process that many GNU software packages are continually going through, to no ill effect. It doesnt really matter much from the end users perspective, since GNU software by its nature always tends to gravitate towards consistency and improvement, one way or another. It is also doesnt matter to the end user because the end user has selected a popular L INUX distribution packaged by someone who has already dealt with these issues.

When you get L INUX, it will be inside a standard distribution, probably on a CD. Each of these packages is selected by the distribution vendors to be a genuine and stable release of that package. This is the responsibility taken on by those who create L INUX distributions. Note that no corporate body oversees L INUX. Everyone is on their own mission. But a package will not nd its way into a distribution unless someone feels that it is a useful one. For people to feel it is useful means that they have to have used it over a period of time; in this way only good, thoroughly reviewed software gets included. Maintainers of packages ensure that ofcial releases are downloadable from their home pages and will upload original versions onto well-established FTP servers. It is not the case that any person is free to modify original distributions of packages and thereby hurt the names of the maintainers of that package. For those who are paranoid that the software they have downloaded is not the genuine article distributed by the maintainer of that software, digital signatures can verify the packager of that software. Cases where vandals have managed to substitute a bogus package for a real one are extremely rare and entirely preventable.

What is Open Source vs. Free vs. Shareware?

Open Source is a new catch phrase that is ambiguous in meaning but is often used synonymously with Free. It sometimes refers to any proprietary vendor releasing source code to their package, even though that source code is not free in the sense of users being able to modify it and redistribute it. Sometimes it means public domain software that anyone can modify but which can be incorporated into commercial packages where later versions will be unavailable in source form. Open Source advocates vie for the superiority of the Open Source development model. GNU supporters dont like to use the term Open Source. Free software, in the sense of freedom to modify and redistribute is the preferred term and necessitates a copyright license along the same vein as the GPL. Unfortunately, its not a marketable term because it requires this very explanation, which tends to bore people who dont really care about licensing issues. Free software advocates vie for the ethical responsibility of making source code available and encouraging others to do the same. Shareware refers to completely nonfree software that is encouraged to be redistributed at no charge, but which requests a small fee if it happens to land on your computer. It is not Free software at all.

There are so many different L INUX versions is this not confusion and incompatibility?
(See also next question.) The L INUX kernel is now on release version 2.4.3 as of this writing. The only other stable release of the kernel was the previous 2.2 series which was the standard for more than a year. The L INUX kernel version does not affect the L INUX user. L INUX programs will work regardless of the kernel version. Kernel versions speak of features, not compatibility. Each L INUX distribution has its own versioning system. RedHat has just released version 7.0 of its distribution, Caldera, 2.2, Debian, 2.1, and so forth. Each new incarnation of a distribution will have newer versions of packages contained therein and better installation software. There may also have been subtle changes in the le system layout. The L INUX U NIX library implementation is called glibc. When RedHat brought out version 5.0 of its distribution, it changed to glibc from the older libc5 library. Because all

559

560

D. L INUX Advocacy FAQ

D.3. L INUX Distributions

D. L INUX Advocacy FAQ

packages require this library, this was said to introduce incompatibility. It is true, however, that multiple versions of libraries can coexist on the same system, and hence no serious compatibility problem was ever introduced in this transition. Other vendors have since followed suit in making the transition to glibc (also known as libc6). The L INUX community has also produced a document called the L INUX Filesystem Standard. Most vendors try to comply with this standard, and hence L INUX systems will look very similar from one distribution to another. There are hence no prohibitive compatibility problems between L INUX distributions.

Whats nice about RPM based distributions (RedHat, Mandrake, and others) is that almost all developers provide RedHat .rpm les (the le that a RedHat package comes in). Debian .deb package les are usually provided, but not as often as .rpm. On the other hand, Debian packages are mostly created by people on the Debian development team, who have rigorous standards to adhere to. TurboLinux, SuSE, and some others are also very popular. You can nd reviews on the Internet. Many other popular distributions are worth installation. Especially worthwhile are distributions developed in your own country that specialize in the support of your local language.

Will a program from one L INUX Distribution run on another? How compatible are the different distributions?
The different distributions are very similar and share binary compatibility (provided that they are for the same type of processor of course)that is, L INUX binaries compiled on one system will work on another. This is in contrast to the differences between, say, two U NIX operating systems (compare Sun vs. IRIX). Utilities also exist to convert packages meant for one distribution to be installed on a different distribution. Some distributions are, however, created for specic hardware, and thus their packages will only run on that hardware. However, all software specifically written for L INUX will recompile without any modications on another L INUX platform in addition to compiling with few modications on other U NIX systems. The rule is basically this: If you have three packages that you would need to get working on a different distribution, then it is trivial to make the adjustments to do this. If you have a hundred packages that you need to get working, then you have a problem.

Where do I get L INUX?

Once you have decided on a distribution (see previous question), you need to download that distribution or buy or borrow it on CD. Commercial distributions may contain proprietary software that you may not be allowed to install multiple times. However, Mandrake, RedHat, Debian, and Slackware are all committed to freedom and hence will not have any software that is not redistributable. Hence, if you get one of these on CD, feel free to install it as many times as you like. Note that the GPL does not say that GNU software is without cost. You are allowed to charge for the service of distributing, installing, and maintaining software. It is the nonprohibition to redistribute and modify GNU software that is meant by the word free. An international mirror for L INUX distributions is Metalab distributions mirror ftp://metalab.unc.edu/pub/Linux/distributions/. Also consult the resources in Chapter 13, What web pages should I look at? on page 553, and the Web sites entry in the index. Downloading from an FTP site is going to take a long time unless you have a really fast link. Hence, rather ask around who locally sells L INUX on CD. Also make sure you have the latest version of whatever it is youre buying or downloading. Under no circumstance install from a distribution that has been superseded by a newer version.

What is the best distribution to use?

If you are an absolute beginner and dont really feel like thinking about what distribution to get, one of the most popular and easiest to install is Mandrake. RedHat is also supported quite well in industry. The attributes of some distributions are: Mandrake: Mandrake is RedHat with some packages added and updated. It has recently become the most popular and may be worth using in preference to RedHat. Debian: This is probably the most technically advanced. It is completely free and very well structured as well as standards conformant. It is slightly less elegant to install. Debian package management is vastly superior to any other. The distribution has legendary technical excellence and stability. RedHat: This is possibly the most popular. Slackware: This was the rst L INUX distribution and is supposed to be the most current (software is always the latest). Its a pain to install and manage, although school kids who dont know any better love it.

How do I install L INUX?

It helps to think more laterally when trying to get information about L INUX: Would-be L INUX users everywhere need to know how to install L INUX. Surely the Free software community has long since generated documentation to help them? Where is that documentation? Most distributions have very comprehensive installation guides, which is the reason I do not cover installation in this book. Browse around your CD to nd it or consult your vendors web site. Also try see what happens when you do a net search with linux installation guide. You need to read through the install guide in detail. It will explain everything you need to know about setting up partitions, dual boots, and other installation goodies. The installation procedure will be completely different for each distribution.

561

562

D. L INUX Advocacy FAQ

D.4. L INUX Support

D.5. L INUX Compared to Other Systems

D. L INUX Advocacy FAQ

D.4

L INUX Support

This section explains where to get free and commercial help with L INUX.

What is clear is that the number of L INUX users is doubling consistently every year. This is evident from user interest and industry involvement in L INUX; journal subscriptions, web hits, media attention, support requirements, software ports, and other criteria. Because it is easy to survey online machines, it is well-established that over 25% of all web servers run L INUX.

Where does a person get L INUX support? My purchased software is supported; how does L INUX compete?
L INUX is supported by the community that uses L INUX. With commercial systems, users are too stingy to share their knowledge because they feel that they owe nothing for having spent money on software. L INUX users, on the other hand, are very supportive of other L INUX users. People can get far better support from the Internet community than they would from their commercial software vendors. Most packages have email lists where the very developers are available for questions. Most cities have mailing lists where responses to email questions are answered within hours. New L INUX users discover that help abounds and that they never lack friendly discussions about any computing problem they may have. Remember that L INUX is your operating system. Newsgroups provide assistance where L INUX issues are discussed and help is given to new users; there are many such newsgroups. Using a newsgroup has the benet of the widest possible audience. The web is also an excellent place for support. Because users constantly interact and discuss L INUX issues, 99% of the problems a user is likely to have would have already been documented or covered in mailing list archives, often obviating the need to ask anyone at all. Finally, many professional companies provide assistance at comparable hourly rates.

What is the total cost of installing and running L INUX compared to a proprietary non-U NIX system?
Although L INUX is free, a good knowledge of U NIX is required to install and congure a reliable server. This tends to cost you in time or support charges. On the other hand, your Windows or OS/2 server, for example, has to be licensed. Many arguments put forward regarding server costs fail to take into account the complete lifetime of the server. This has resulted in contrasting reports that either claim that L INUX costs nothing or claim that it is impossible to use because of the expense of the expertise required. Neither of these extreme views is true. The total cost of a server includes the following: Cost of the OS license. Cost of dedicated software that provides functions not inherently supported by the operating system. Cost of hardware.

D.5

L INUX Compared to Other Systems

Availability of used hardware and the OSs capacity to support it. Cost of installation. Cost of support. Implicit costs of server downtime because of software bugs.

This section discusses the relative merits of different U NIX systems and NT.

What is the most popular U NIX in the world?

L INUX has several times the installed base of any U NIX system.

Implicit costs of server downtime because of security breaches. Cost of maintenance. Cost of repair. Cost of essential upgrades. Negative cost of multiple servers: L INUX can run many services (mail, le, Web) from the same server rather than requiring dedicated servers, and this can be a tremendous saving. When all these factors are considered, any company should probably make a truly enormous saving by choosing a L INUX server over a commercial operating system.

How many L INUX systems are there out there?

This is an answer nobody really knows. Various estimates have been put forward based on statistical considerations. As of early 2001 the gure was about 1020 million. As L INUX begins to dominate the embedded market, that number will soon surpass the number of all other operating systems combined.

563

564

D. L INUX Advocacy FAQ

D.5. L INUX Compared to Other Systems

D. L INUX Advocacy FAQ

What is the total cost of installing and running a L INUX system compared to a proprietary U NIX system?
(See previous question.) Proprietary U NIX systems are not as user friendly as L INUX. L INUX is also considered far easier to maintain than any commercial U NIX system because of its widespread use and hence easy access to L INUX expertise. L INUX has a far more dedicated and beginner friendly documentation project than any commercial U NIX, and many more user-friendly interfaces and commands. The upshot of this is that although your proprietary U NIX system will perform as reliably as L INUX, it will be more time consuming to maintain. U NIX systems that run on specialized hardware are almost never worth what you paid for them in terms of a cost/performance ratio. That is doubly if you are also paying for an operating system.

L INUX has a proper journalling le system called ReiserFS. This means that in the event of a power failure, there is very little chance that the le system would ever be corrupted, or that manual intervention would be required to x the le system.

Does L INUX only support 2 Gigs of memory and 128 Meg of swap?
L INUX supports a full 64 gigabytes of memory, with 1 gigabyte of unshared memory per process. If you really need this much memory, you should be using a 64-bit system, like a DEC Alpha, or Sun UltraSPARC machine. On 64-bit systems, L INUX supports more memory than most rst-world governments can afford to buy. L INUX supports as much swap space as you like. For technical reasons, however, the swap space formerly required division into separate partitions of 128 megabytes each.

How does L INUX compare to other operating systems in performance?

L INUX typically performs 50% to 100% better than other operating systems on the same hardware. There are no commercial exceptions to this rule for a basic PC. There have been a great many misguided attempts to show that L INUX performs better or worse than other platforms. I have never read a completely conclusive study. Usually these studies are done with one or other competing system having better expertise at its disposal and are, hence, grossly biased. In some supposedly independent tests, L INUX tended to outperform NT as a web server, le server, and database server by an appreciable margin. In general, the performance improvement of a L INUX machine is quite visible to users and administrators. It is especially noticeable how fast the le system access is and how it scales smoothly when multiple services are being used simultaneously. L INUX also performs well when loaded by many services simultaneously. There is also criticism of L INUXs SMP (multiprocessor) support, and lack of a journalling le system. These two issues are discussed in the next question. In our experience (from both discussions and development), L INUXs critical operations are always pedantically optimizedfar more than would normally be encouraged in a commercial organization. Hence, if your hardware is not performing the absolute best it can, its by a very small margin. Its also probably not worthwhile debating these kinds of speed issues when there are so many other good reasons to prefer L INUX.

Isnt U NIX antiquated? Isnt its security model outdated?

The principles underlying OS development have not changed since the concept of an OS was invented some 40+ years ago. It is really academia that develops the theoretical models for computer scienceindustry only implements these. There are a great many theoretical paradigms of operating system that vary in complexity and practicality. Of the popular server operating systems, U NIX certainly has the most versatile, exible, and applicable security model and le system structure.

How does FreeBSD compare to L INUX?

FreeBSD is like a L INUX distribution in that it also relies on a large number of GNU packages. Most of the packages available in L INUX distributions are also available for FreeBSD. FreeBSD is not merely a kernel but also a distribution, a development model, an operating system standard, and a community infrastructure. FreeBSD should rather be compared to Debian than L INUX. The arguments comparing the FreeBSD kernel to the L INUX kernel center around the differences between how various kernel functions are implemented. Depending on the area you look at, either L INUX or FreeBSD will have a better implementation. On the whole, FreeBSD is thought to have a better architecture, although L INUX has had the benet of having been ported to many platforms, has a great many more features, and supports far more hardware. It is questionable whether the performance penalties we are talking about are of real concern in most practical situations. Another important consideration is that the FreeBSD maintainers go to far more effort securing FreeBSD than does any L INUX vendor. This makes FreeBSD a more trustworthy alternative.

What about SMP and a journalling le system? Is L INUX enterpriseready?

L INUX is supposed to lack proper SMP support and therefore not be as scalable as other OSs. This is somewhat true and has been the case until kernel 2.4 was released in January 2001.

565

566

D. L INUX Advocacy FAQ

D.6. Migrating to L INUX

D. L INUX Advocacy FAQ

GPL advocates take issue with FreeBSD because its licensing allows a commercial organization to use FreeBSD without disclosing additions to the source code. None of these arguments offset the fact that either of these systems is preferable to a proprietary one.

Managers will typically request that L INUX skills be taught to their employees through a training course. What is often missed, is that their staff have little basic U NIX experience to begin with. For instance, it is entirely feasible to run Apache (a web server package) on a SCO, IRIX, or Sun systems, yet managers will request, for example, that their staff be taught how to congure a L INUX web server in order to avoid web server licensing fees. It is important to gauge whether your staff have a real understanding of the TCP/IP networks and U NIX systems that you are depending on, rather then merely using a trial-and-error approach to conguring your machines. Fundamentally, L INUX is just a U NIX system, and a very user-friendly one at that, so any difculties with L INUX ought not to be greater than those with your proprietary U NIX system. Should their basic U NIX knowledge be incomplete, a book like this one will provide a good reference. Many companies also develop in-house applications specic to their corporations services. Being an in-house application, the primary concern of the developers was to get it working, and that might have been accomplished only by a very small margin. Suddenly running the code on a different platform will unleash havoc, especially if it was badly written. In this case, it will be essential to hire an experienced developer who is familiar with the GNU compiler tools. Well written U NIX applications (even GUI applications) will, however, port very easily to L INUX and of course to other U NIX systems.

D.6 Migrating to L INUX

What are the principal issues when migrating to L INUX from a nonU NIX system?
Most companies tend to underestimate how entrenched they are in Windows skills. An ofce tends to operate organically with individuals learning tricks from each other over long periods of time. For many people, the concept of a computer is synonymous with the Save As and My Documents buttons. L INUX departs completely from every habit they might have learned about their computer. The average secretary will take many frustrating weeks gaining condence with a different platform, while the system administrator will battle for much longer. Whereas Windows does not offer a wide range of options with regards to desktops and ofce suites, the look-and-feel of a L INUX machine can be as different between the desktops of two users as is Windows 98 different from an Apple Macintosh. Companies will have to make careful decisions about standardizing what people use, and creating customizations peculiar to their needs. Note that Word and Excel documents can be read by various L INUX ofce applications but complex formatting will not convert cleanly. For instance, document font sizes, page breaking, and spacing will not be preserved exactly. L INUX can interoperate seamlessly with Windows shared le systems, so this is one area where you will have few migration problems. GUI applications written specically for Windows are difcult to port to a U NIX system. The Wine project now allows pure Windows applications to be recompiled under U NIX, and Borland has developed Kylix (a L INUX version of Delphi). There are more examples of L INUX versions of Windows languages, however, any application that interfaces with many proprietary tools and is written in a proprietary language is extremely difcult to port. The developer who does the porting will need to be an expert in U NIX development and an expert in Windows development. Such people are rare and expensive to hire.

How should a supervisor proceed after making the decision to migrate to L INUX?
Before installing any L INUX machines, you should identify what each person in your organization does with their computer. This undertaking is difcult but very instructive. If you have any custom applications, you need to identify what they do and create a detailed specication of their capabilities. The next step is to encourage practices that lean toward interoperability. You may not be able to migrate to L INUX immediately, but you can save yourself enormous effort by taking steps in anticipation of that possibility. For instance, make a policy that all documents must be saved in a portable format that is not bound to a particular wordprocessor package. Wean people off tools and network services that do not have U NIX equivalents. SMTP and POP/IMAP servers are an Internet standard and can be replaced with L INUX servers. SMB le servers can be replaced by L INUX Samba servers. There are web mail and web groupware services that run on L INUX servers that can be used from Internet Explorer. There are some word processors that have both U NIX and Windows versions whose operation is identical on both OSs. Force your developers to test their Web pages on Netscape/Mozilla as well as Internet Explorer. Do not develop using tools that are tied very closely to the operating system and are therefore unlikely to ever have U NIX versions; there are Free cross platform development tools that are more effective than popular commercial IDEs: Use these languages instead. If you are developing using a compiler language, your developers should ensure that code compiles

What are the principal issues when migrating to L INUX from another U NIX system?
The following is based on my personal experience during the migration of three large companies to L INUX. Commercial U NIX third party software that has been ported to L INUX will pose very little problem at all. You can generally rely on performance improvements and reduced costs. You should have no hesitation to install these on L INUX.

567

568

D. L INUX Advocacy FAQ

D.7. Technical

D. L INUX Advocacy FAQ

cleanly with independent brands of compiler. This will not only improve code quality but will make the code more portable. Be aware that people will make any excuse to avoid having to learn something new. Make the necessary books available to them. Identify common problems and create procedures for solving them. Learn about the capabilities of L INUX by watching Internet publications: A manager who is not prepared to do this much should not expect their staff to do better.

can run on a 386 with 4 megabytes of RAM and a 200-megabyte hard drive. Note that scrap hardware can be very time consuming to congure. Note that recently some distributions are coming out with Pentium-only compilations. This means that your old 386 will no longer work. You will then have to compile your own kernel for the processor you are using and possibly recompile packages.

D.7 Technical
This section covers various specic and technical questions.

What hardware is supported? Will my sound/graphics/network card work?

About 90% of all hardware available for the PC is supported under L INUX. In general, wellestablished brand names will always work, but will tend to cost more. New graphics/network cards are always being released onto the market. If you buy one of these, you might have to wait many months before support becomes available (if ever). To check on hardware support, see the Hardware-HOWTO http://users.bart.nl/patrickr/hardware-howto/Hardware-HOWTO.html This may not be up-to-date, so its best to go to the various references listed in this document and get the latest information.

Are L INUX CDs readable from Windows?

Yes. You can browse the installation documentation on the CD (if it has any) using Internet Explorer. L INUX software tends to prefer Windows oppy disk formats, and ISO9660 CD formats, even though almost everything else uses a different format.

Can I run L INUX and Windows on the same machine?

Yes, L INUX will occupy two or more partitions, while Windows will sit in one of the primary partitions. At boot time, a boot prompt will ask you to select which operating system you would like to boot into.

Can I view my Windows, OS/2, and MS-DOS les under L INUX?

L INUX has read and write support for all these le systems. Hence, your other partitions will be readable from L INUX. In addition, L INUX supports a wide range of other le systems like those of OS/2, Amiga, and other U NIX systems.

How much space do I need to install L INUX?

A useful distribution of packages that includes the Window System (U NIXs graphical environment) will occupy less than 1 gigabyte. A network server that does not have to run X can get away with about 100-300 megabytes. L INUX can run on as little as a single stiffy diskthats 1.4 megabytesand still perform various network services.

Can I run DOS programs under L INUX?

L INUX contains a highly advanced DOS emulator. It will run almost any 16-bit or 32-bit DOS application. It runs a great number of 32-bit DOS games as well. The DOS emulator package for L INUX is called dosemu. It typically runs applications much faster than does normal DOS because of L INUXs faster le system access and system calls. It can run in an window just like a DOS window under Windows.

What are the hardware requirements?

L INUX runs on many different hardware platforms, as explained above. Typical users should purchase an entry-level PC with at least 16 megabytes of RAM if they are going to run the X Window System (U NIXs graphical environment) smoothly. A good L INUX machine is a PII 300 (or AMD, K6, Cyrix, etc.) with 64 megabytes of RAM and a 2-megabyte graphics card (i.e., capable of run 1024x768 screen resolution in 15/16 bit color). One gigabyte of free disk space is necessary. If you are using scrap hardware, an adequate machine for the Window System should not have less than an Intel 486 100 MHz processor and 8 megabytes of RAM. Network servers

Can I recompile Windows programs under L INUX?

Yes. WineLib is a part of the Wine package (see below) and allows Windows applications to be recompiled to work under L INUX. Apparently this works extremely well, with virtually no changes to the source code being necessary.

569

570

D. L INUX Advocacy FAQ

D.7. Technical

D. L INUX Advocacy FAQ

Can I run Windows programs under L INUX?

Yes and no. There are commercial emulators that will run a virtual 386 machine under L INUX. This enables mostly awless running of Windows under L INUX if you really have to and at a large performance penalty. You still have to buy Windows though. There are also some Free versions of these. There is also a project called Wine (WINdows Emulator) which aims to provide a free alternative to Windows by allowing L INUX to run Windows 16 or 32 bit binaries with little to no performance penalty. It has been in development for many years now, and has reached the point where many simple programs work quite awlessly under L INUX. Get a grip on what this means: you can run Minesweep under L INUX and it will come up on your Window screen next to your other L INUX applications and look exactly like what it does under Windowsand you dont have to buy Windows. You will be able to cut and paste between Windows and L INUX application. However, many applications (especially large and complex ones) do not display correctly under L INUX or crash during operation. This has been steadily improving to the point where Microsoft Ofce 2000 is said to be actually usable. Many Windows games do, however, work quite well under L INUX, including those with accelerated 3D graphics. See the Wine Headquarters http://www.winehq.com/faq.html for more information.

At some point in the future, a large number of users may be using the same proprietary desktop application that has some security vulnerability in it. If this were to support a virus, it would only be able to damage the users restricted space, but then it would be the application that is insecure, not L INUX per se. Remember also that with L INUX, a sufcient understanding of the system makes it possible to easily detect and repair the corruption, without have to do anything drastic, like reinstalling or buying expensive virus detection software.

Is L INUX as secure as other servers?

L INUX is as secure as or more secure than typical U NIX systems. Various issues make it more and less secure. Because GNU software is open source, any hacker can easily research the internal workings of critical system services. On one hand, they may nd a aw in these internals that can be indirectly exploited to compromise the security of a server. In this way, L INUX is less secure because security holes can be discovered by arbitrary individuals. On the other hand, individuals may nd a aw in these internals that they can report to the authors of that package, who will quickly (sometimes within hours) correct the insecurity and release a new version on the Internet. This makes L INUX more secure because security holes are discovered and reported by a wide network of programers. It is therefore questionable whether free software is more secure or not. I personally prefer to have access to the source code so that I know what my software is doing. Another issue is that L INUX servers are often installed by lazy people who do not take the time to follow the simplest of security guidelines, even though these guidelines are widely available and easy to follow. Such systems are sitting ducks and are often attacked. (See the previous question.) A further issue is that when a security hole is discovered, system administrators fail to heed the warnings announced to the L INUX community. By not upgrading that service, they leave open a window to opportunistic hackers. You can make a L INUX system completely airtight by following a few simple guidelines, like being careful about what system services you expose, not allowing passwords to be compromised, and installing utilities that close possible vulnerabilities. Because of the community nature of L INUX users, there is openness and honesty with regard to security issues. It is not found, for instance, that security holes are covered up by maintainers for commercial reasons. In this way, you can trust L INUX far more than commercial institutions that think they have a lot to lose by disclosing aws in their software.

I have heard that L INUX does not suffer from virus attacks. Is it true that there is no threat of viruses with U NIX systems?
A virus is a program that replicates itself by modifying the system on which it runs. It may do other damage. Viruses are small programs that exploit social engineering, logistics, and the inherent exibility of a computer system to do undesirable things. Because a U NIX system does not allow this kind of exibility in the rst place, there is categorically no such thing as a virus for it. For example, U NIX inherently restricts access to les outside the users privilege space, so a virus would have nothing to infect. However, although L INUX cannot itself execute a virus, it may be able to pass on a virus meant for a Windows machine should a L INUX machine act as a mail or le server. To avoid this problem, numerous virus detection programs for L INUX are now becoming available. Its what is meant by virus-software-for-L INUX. On the other hand, conditions sometimes allow an intelligent hacker to target a machine and eventually gain access. The hacker may also mechanically try to attack a large number of machines by using custom programs. The hacker may go one step further to cause those machines that are compromised to begin executing those same programs. At some point, this crosses the denition of what is called a worm. A worm is a thwarting of security that exploits the same security hole recursively through a network. See the question on security below.

571

572

E. The GNU General Public License Version 2

Software Foundation software is covered by the GNU Library General Public License instead.) You can apply it to your programs, too. When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things. To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it. For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights. We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software. Also, for each authors protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modied by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reect on the original authors reputations. Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyones free use or not licensed at all. low. The precise terms and conditions for copying, distribution and modication fol-

Appendix E

The GNU General Public License Version 2

Most of the important components of a Free U NIX system (like L INUX ) were developed by the Free Software Foundation http://www.gnu.org/ (FSF). Further, most of a typical L INUX distribution comes under the FSFs copyright, called the GNU General Public License. It is therefore important to study this license in full to understand the ethos of Free Meaning the freedom to be modied and redistributed. development, and the culture under which L INUX continues to evolve.

GNU GENERAL PUBLIC LICENSE Version 2, June 1991

Copyright (C) 1989, 1991 Free Software Foundation, Inc. 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.

Preamble
The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change free softwareto make sure the software is free for all its users. This General Public License applies to most of the Free Software Foundations software and to any other program whose authors commit to using it. (Some other Free 573

GNU GENERAL PUBLIC LICENSE TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
0. This License applies to any program or other work which contains a notice placed by the copyright holder saying it may be distributed under the terms of this General Public License. The Program, below, refers to any such program or work, 574

E. The GNU General Public License Version 2

and a work based on the Program means either the Program or any derivative work under copyright law: that is to say, a work containing the Program or a portion of it, either verbatim or with modications and/or translated into another language. (Hereinafter, translation is included without limitation in the term modication.) Each licensee is addressed as you. Activities other than copying, distribution and modication are not covered by this License; they are outside its scope. The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does. 1. You may copy and distribute verbatim copies of the Programs source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and give any other recipients of the Program a copy of this License along with the Program. You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee. 2. You may modify your copy or copies of the Program or any portion of it, thus forming a work based on the Program, and copy and distribute such modications or work under the terms of Section 1 above, provided that you also meet all of these conditions: a) You must cause the modied les to carry prominent notices stating that you changed the les and the date of any change. b) You must cause any work that you distribute or publish, that in whole or in part contains or is derived from the Program or any part thereof, to be licensed as a whole at no charge to all third parties under the terms of this License. c) If the modied program normally reads commands interactively when run, you must cause it, when started running for such interactive use in the most ordinary way, to print or display an announcement including an appropriate copyright notice and a notice that there is no warranty (or else, saying that you provide a warranty) and that users may redistribute the program under these conditions, and telling the user how to view a copy of this License. (Exception: if the Program itself is interactive but does not normally print such an announcement, your work based on the Program is not required to print an announcement.) These requirements apply to the modied work as a whole. If identiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and 575

its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it. Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Program. In addition, mere aggregation of another work not based on the Program with the Program (or with a work based on the Program) on a volume of a storage or distribution medium does not bring the other work under the scope of this License. 3. You may copy and distribute the Program (or a work based on it, under Section 2) in object code or executable form under the terms of Sections 1 and 2 above provided that you also do one of the following: a) Accompany it with the complete corresponding machine-readable source code, which must be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, b) Accompany it with a written offer, valid for at least three years, to give any third party, for a charge no more than your cost of physically performing source distribution, a complete machine-readable copy of the corresponding source code, to be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, c) Accompany it with the information you received as to the offer to distribute corresponding source code. (This alternative is allowed only for noncommercial distribution and only if you received the program in object code or executable form with such an offer, in accord with Subsection b above.) The source code for a work means the preferred form of the work for making modications to it. For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface denition les, plus the scripts used to control compilation and installation of the executable. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable. If distribution of executable or object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place counts as distribution of the source code, even though third parties are not compelled to copy the source along with the object code. 576

E. The GNU General Public License Version 2

4. You may not copy, modify, sublicense, or distribute the Program except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense or distribute the Program is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance. 5. You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Program or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Program (or any work based on the Program), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Program or works based on it. 6. Each time you redistribute the Program (or any work based on the Program), the recipient automatically receives a license from the original licensor to copy, distribute or modify the Program subject to these terms and conditions. You may not impose any further restrictions on the recipients exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties to this License. 7. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Program at all. For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program. If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply and the section as a whole is intended to apply in other circumstances. It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system, which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice. This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License. 577

8. If the distribution and/or use of the Program is restricted in certain countries either by patents or by copyrighted interfaces, the original copyright holder who places the Program under this License may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License. 9. The Free Software Foundation may publish revised and/or new versions of the General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns. Each version is given a distinguishing version number. If the Program species a version number of this License which applies to it and any later version, you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation. 10. If you wish to incorporate parts of the Program into other free programs whose distribution conditions are different, write to the author to ask for permission. For software which is copyrighted by the Free Software Foundation, write to the Free Software Foundation; we sometimes make exceptions for this. Our decision will be guided by the two goals of preserving the free status of all derivatives of our free software and of promoting the sharing and reuse of software generally.

NO WARRANTY
11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM AS IS WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING 578

E. The GNU General Public License Version 2

OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.

The hypothetical commands show w and show c should show the appropriate parts of the General Public License. Of course, the commands you use may be called something other than show w and show c; they could even be mouse-clicks or menu itemswhatever suits your program. You should also get your employer (if you work as a programmer) or your school, if any, to sign a copyright disclaimer for the program, if necessary. Here is a sample; alter the names: Yoyodyne, Inc., hereby disclaims all copyright interest in the program Gnomovision (which makes passes at compilers) written by James Hacker. <signature of Ty Coon>, 1 April 1989 Ty Coon, President of Vise

END OF TERMS AND CONDITIONS

How to Apply These Terms to Your New Programs

If you develop a new program, and you want it to be of the greatest possible use to the public, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms. To do so, attach the following notices to the program. It is safest to attach them to the start of each source le to most effectively convey the exclusion of warranty; and each le should have at least the copyright line and a pointer to where the full notice is found.
<one line to give the programs name and a brief idea of what it does.> Copyright (C) 19yy <name of author> This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307

This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Library General Public License instead of this License.

USA

Also add information on how to contact you by electronic and paper mail. If the program is interactive, make it output a short notice like this when it starts in an interactive mode:
Gnomovision version 69, Copyright (C) 19yy name of author Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type show w. This is free software, and you are welcome to redistribute it under certain conditions; type show c for details.

579

580

Index

Index
Symbols
*, 214 */, 221 + key, 494 ++, 212 - key, 494 -, 201 ., 175 . character, 49 /*, 221 :, 175 : key, 54 ;, 209, 221 =, 213 ==, 213, 218 >|, 178 >, 178 >& notation, 76 ? character, 49 ? key, 38 [, 179 #, 222 $, 79, 173 $( ) notation, 71 $*, 172 $-, 172, 173 $0, 172 $1, 172 $?, 172, 209 $#, 172 $$, 172 %, 96 %?, 96 %CPU, 90 %MEM, 90 %d, 209 %e, 210 %f, 210 &&, 171 C, 11, 82, 86 D, 86 Z, 82 key, 10, 55, 493 key, 10, 55 key, 55 key, 55 !, 161 $@, 172 $!, 172 !=, 218 !, 80 ", 209 ||, 171, 216 |, 74 {, 211 }, 211 +, 173 -, 173 , 173

Numbers
0.0.0.0, 248 0x8086, 474 .1, 32 1, port, 519 2.2, kernel, 464 2.4, kernel, 464 2>&1 notation, 76 2nd extended le system, 160 .3, 30 3.5-inch, oppy, 144, 162 3.5-inch oppy, 145 3D graphics, 487 6, port, 519 8-bit, ISA slots, 18 8N1 protocol, 22 9-pin, 21 10.0.0.0, 249 11x17, 201 16-color X server, X, 499, 505 20, port, 459 21, port, 294, 459, 519 22, port, 459, 519 23, port, 294 25-pin, 21 25, port, 99, 299, 300, 459, 519 32 bits, 214

32-bit, 247 32-bit address, 248 53, port, 459, 519 64 bit server, 240 64-Kb line, 462 64-bit, 552 67, port, 295 69, port, 295 79, port, 295 80?86, 208 80 50, 320 80, port, 265, 389, 459 110, port, 271, 295, 301, 459 113, port, 295, 459 119, port, 459 127.0.0.0, 249, 252 127.0.0.1, 445, 510 128-bit, 280 143, port, 295, 301, 459 172.16.0.0, 249 192.168.0.0, 249 255.255.255.255, 248 386, 552 390 mainframe, 553 400, port, 519 486, 552 513, port, 294 514, port, 294 515, port, 519 517, port, 295 540, port, 295 680x0, 552 901, port, 435 1024 cylinder boundary, 156, 319 1024 cylinders, 536 1024, port, 265 6000, port, 486 8250, UART, 479 16450, UART, 479 16550A, UART, 479 16550, UART, 23, 479 16650V2, UART, 479 16650, UART, 479 16750, UART, 479

A
A record, 548 DNS, 283, 441443, 446 .a, 29, 230, 233 a.out, 208 a3, 201 a4, 201 a5, 201 A:, 144 A: disk, 44

AAAA query, 280 absolute, path, 34, 128 ac, 549 .ac.za, 276 access, remote, 113 access bits, 123 access control, 293, 296 Apache, 397 printer, 202 Access Control Lists, see ACL access ags, 123 access permissions, 109 NFS, 288 access rights, 104 access.conf, 393 AccessFileName, 395 accton, 549 ACK, TCP, 264 acknowledgment number, TCP, 265 acknowledgment packet, TCP, 263 ACL, 430 security, 522 Active Directory, 430 ACU, 342 adapter, SCSI, 478 AddEncoding, 399 adding partition, 157 swap, 162 adding a column, postgres, 420 adding to, PATH, 46 address, IP, 247, 250, 252, 256, 273, 277, 300 address classes, IP, 249 Address Resolution Protocol, 250 address space, 248 addresses, 79 sed, 79 adfs, 163 administration, U NIX, 6 administrator, responsibilities, 313 administrator programs, 196 Advanced Linux Sound Architecture, see ALSA Advanced Package Tool, 245 Advanced Risc Machine, see ARM affs, 163 agetty, 329 aggregation of another, GPL, 576 AGP, RHCE, 544 aic7xxx.o, 323 AIX, 552 -al, ls, 25 alerts, security, 202, 517 Alias, 398 .alias, 29 alias, 175

581

582

Index

aliases, 301 aliasing Apache, 398 interface, 259 alien, 537 All, 396 allocate memory, 214 Allow, 397 allow null glob expansion, 95 AllowOverride, 397 alpha, 240 ALSA, sound, 475 Alt key, 10, 493495 Alt-F1, 11 altavista.com, 118 ALTER TABLE, postgres, 420 American Standard Code for Information Interchange, see ASCII Amiga, 570 AmigaOS, 58, 425 anacron, 545 announcements, security, 516 anonymity, 100 anonymous email, 99, 100 anonymous logins, 113 ANY record, DNS, 284 Apache access control, 397 aliasing, 398 CGI, 401 DNS lookup, 395 DSO, 406 encoding, 399 fancy indexes, 399 forms, 403 indexes, 399 installing, 393 IP-based virtual hosting, 407 language negotiation, 399 log format, 395 name-based virtual hosting, 407 PHP, with, 406 reference, 393 RHCE, 548 Server-side includes, 400 SQL, with, 403 SSI, 400 top-level directory, 395 user directories, 398 virtual hosting, 407 apache, 193, 546 Apache reference, 134 Apache, with Windows, 393 API, X, 498

append, 320 append-only permissions, security, 520 append =, 468 Apple, 285 Apple Mac, 492 Apple Macintosh, 59, 567 application, 135 application, 115 application or command, stop, 41 applying to new programs, GPL, 579 appres, 494 apsfilter, 204 APT, 245 apt, 537 apt(8), 245 apt-cache, 245 apt-cdrom, 245 apt-config, 245 apt-get, 245, 537 apt.conf(5), 245 apxs, 407 ar, 229 arcfour, 271 architecture, 240 archive, 45, 229 backup, 45 archive indexing, 229 archiving les, 241 argc, 218, 224 arguments, 211 argv, 218, 224 arithmetic expansion, 174 ARM, 552 ARP, 251 re-request, 251 time-out, 251 arp, 251 array, 213 artifacts, X, 503 Artistic License, 556 ASCII, 7, 22, 113, 209, 218, 507 ascii(7), 209, 218 AT, 331 AT commands, 24 modem, 342, 453 at, 411, 535 AT&F1, 454 ATAPI, 18 CD-ROM, 144, 161 kernel, 477 ATAPI disk, 144 AtariMiNT, 58 atd, 409, 411, 412 aterm, 539 atime, 126

atobm, 494 atomic, 189 atq, 411 attach, 226 attach onto running programs, 226 attaching les, 115 attacks, security, 511 attempts, 280 attribute, postgres, 418 .au, 29 audio, 41 audio, 115 audio format, 29, 31, 40 mod, 41 auditing, security, 524 aumix, 41 auth, 295 auth service, 459 authenticating, 186 authentication uucp, 339 login, 330 authentication logic, security, 514 authoritative, 281, 300 DNS, 441 AUTHORS, 32, 237 authpriv, 296 auto resume, 95 autoconf, 238 autodetection, 20, 486 autofs, 163 Automatic Calling Unit, see ACU automatically, mounting, 166 .avi, 29 .awk, 29 awk, 29, 182, 185 AWK programming language, 182 AXFR record, DNS, 284

B
b command, 224 b3, 201 b4, 201 b5, 201 B:, 144 background, 82 jobs, 108, 176, 532 X, 496 background command, 172 backquote expansion, 174 backspace key, 10, 493 backtrace, 225 backup, 45, 535 postgres, 423 tar, 45

archive, 45 backups, tape, 149 backward, quotes, 71 badblocks, 161 balsa, 99 banned IP addresses, 313 base64, 115 BASH, 92 bash, 82, 91, 171, 186, 539 bash functions, 208, 539 bash(1), 83, 174176 .bash login, 186, 539 .bash logout, 539 .bash profile, 186, 539 BASH VERSION, 92 .bashrc, 93, 175, 186, 545 bashrc, 545 basic editing operations, vi, 54 baud rate, 24 bc, 36, 183 bdftopcf, 494 beeping less, 38 shell, 11 Tab, 11 beforelight, 494 BeOS, 58 Berkeley Internet Name Domain, see bind Berkeley Regents, see BSD beta, 118 bg, 82, 532 .bib, 29 /bin, 137, 156, 520 bin, 196 /bin/login, 329, 330 /bin/sash, 323 /bin/sh, 107 binary, 113, 183 binary executables, 137 binary le, 208 bind, port, 269 bind, 279, 437, 460, 548 binding signals, 176 BIOS, 476 BIOS, 20, 318 functions, 318 interrupts, 322 limitations, 319 ROM, 318 BIOS conguration, RHCE, 544 BIOS limitations, RHCE, 544 BIOS settings, LPI, 536 bitmap, 494 Bitmap le, 29 bits, 7

583

584

Index

bits per pixel, see bpp black and white, X, 505 block devices, 142 .bmp, 29 bmtoa, 494 body, 98 bool, postgres, 418 boot, 20, 317, 318 disk, 147 kernel, 325 partition, 318 /boot, 156 boot, 320 boot device, 536 boot disks, creating, 147 boot oppy, 147, 321 kernel, 484 boot image, kernel, 463 boot loader, 320 boot options, kernel, 317, 320 boot password, 320 boot sector, 318, 321 boot sectors, partition, 318 boot sequence, CMOS, 20 boot up message, partition, 159 boot.img, 147 /boot/, 463 /boot/boot.0300, 320 /boot/map, 318, 320 /boot/vmlinuz, 318, 320 bootable CD-ROM, 20 partition, 158 booting partition, 317 Windows 98, 321 booting process, LPI, 534 BOOTP, 295, 546 bootpc, 269 bootpd, 294, 295 bootps, 269 bootstraps, 20, 318 bootup messages, log, 37 bootup process, 329 Bourne shell, 82 -bpp, X, 503, 504 bps, 22, 24 brace expansion, 173 brand names, 3 break, 224 break, 65, 218 break point, 224 Brian, Fox, xxxi BROADCAST, 254 broadcast, Samba, 427

broadcast, 251 broadcast address, 256 brute force attack, 104 BSD, 556 BSD License, 414 bt command, 225 buffer overow, security, 518, 521 buffer overow attack, security, 512, 513 BUGS, 32 bugs, 223 BUGTRAQ, LPI, 541 building kernel, 481, 483 package, 237 builtin devices, CMOS, 20 bulk mail, 99 bus, SCSI, 476 buttons, 491 byte, 208 encoding, 7 byte sequences, 37 .bz2, 29 bzImage, 484 make, 538 kernel, 484 bzip2, 29, 42

C
C, 414 comment, 221 library, 227 library function, 209, 216 preprocessor, 222 projects, 230 simple program, 208 source, 237, 238 standard C, 209, 216 .C, 29 .c, 30, 231 c command, 225 C header les, 138 C key, 41, 42 C program, 30 C programming language, 2628, 30, 73, 75, 88, 138, 142, 176, 181, 184, 188, 190, 191, 207209, 211218, 220223, 225, 227, 228, 230232, 237, 238, 263, 264, 277, 335, 405, 416, 444, 463, 469, 485, 492, 493, 512, 518, 525, 560, 567, 570 C source les, 138 C++, 29, 207, 221, 222, 414, 492, 493 cache, 277 caching, DNS, 281 caching name server, DNS, 449 cal, 36

Caldera OpenLinux, 554 canonical name, DNS, 284 capabilities, security, 521 card, SCSI, 476 card database, 501 cards, peripheral, 17, 18 carrier detect, 23 carrier signal, modem, 24 case, 66, 212 case sensitive, U NIX, 25 cat, 12, 36, 42, 73, 147 concatenate, 12 -cc, X, 505 .cc, 29 cc, 208 cd, 34, 175 change directory, 12 CD pin, 23 CD writer, 479 kernel, 477, 478 SCSI, 478 CD-ROM, 18, 146, 168, 194, 286, 479, 520 ATAPI, 144, 161 bootable, 20 IDE, 18, 478 kernel, 477 mounting, 163 RHCE, 544 SCSI, 19, 145, 146 CD-writer, 146 cdable vars, 96 CDPATH, 93 cdplay, 41 cdrecord, 87, 479 cdrecord(1), 87 /cdrom, 163 CERT, LPI, 541 certication LPI, 2 RHCE, 2 .cf, 29 cfdisk, 544 CGA, X, 505 CGI, 389, 401, 406 Apache, 401 RHCE, 548 .cgi, 29 CGI script, 404 Challenge Handshake Authentication Protocol, see CHAP change directory, cd, 12 change ownerships, 101 ChangeLog, 32, 237 CHAP, 456 char, 213, 215

character sets, 507 character terminals, 330, 506 characters le names, 12 user name, 102 chargen, 269 Charityware, 58 chat, pppd, 454 chat script, pppd, 455 chat script, 342, 453 chattr, 520, 549 checksum IP, 248, 264 TCP, 265 Chet, Ramey, xxxi chgrp, 533 child, process, 91 child process, 91, 184 child terminate, 86 chkconfig, 546 chkfontpath, 509 chmod, 123, 336, 533 chown, 101, 533 chroot, 167, 178, 323 CIFS, 425, 427 clash, network, 250 Class A/B/C address, 249 clean, make, 538 clear, 36, 225 clear to send, see CTS client, 194 client machine, X, 485 client programs, security, 518 client/server, 194 clients, mail, 99 clipboard, X, 497 clobberd, 549 Clockchip setting, X, 501 clocks:, X, 500 close, 264 close(2), 264 closing les, 217 CMOS, 20, 203, 472, 476 boot sequence, 20 builtin devices, 20 conguration, 20 Harddrive auto-detection, 20 hardware clock, 20 CNAME record, 548 DNS, 283 .co.za, 276 coda, 163 code reuse, 233 coherent, 163 column typing, postgres, 418

585

586

Index

.com, 273 COM port, Windows, 144, 479 COM1, 18, 20, 144, 342, 479 COM2, 18, 20 COM4, 479 combating, spam, 311 COMMAND, 90 command alias, 175 command history, LPI, 531 command list, mtools, 44 command mode, modem, 24 command pseudonym, 175 command summary, grep, 43 command-line, 173, 174, 218 pppd, 454 LPI, 531 command-line arguments, 25, 172, 224 processing, 68 command-line options, 25 command oriented history, 95 commands, 8 U NIX, 10, 25 GNU, 25 modem, 342, 453 periodic, 409 scheduling, 409 comment, C, 221 comment out, 222 commenting code, 221 commercial drivers, kernel, 482 common devices, 143 Common Gateway Interface, see CGI Common Internet File System, see CIFS comp.os.linux.announce, 120 compact, 320 comparing les, 179 compatibility L INUX, 551 U NIX, 561 UART, 480 X, 487 compile, 208, 220, 238 kernel, 481 compile options, 239 compiled-in modules, kernel, 464, 483 compiled-in support, 322 compiled-out modules, kernel, 464 compiler, 207 compiler optimizations, 223, 239 complete list, error codes, 26 completion, 11 compress, 32 compressed, 114 compressing images, 184 compression, 24, 42, 399

le, 41 compromise, security, 512 computer, programming, 3, 61 concatenate, cat, 12 .conf, 29 conguration, 463 exim, 302 uucp, 338 CMOS, 20 kernel, 482 NFS, 286 package, 193 conguration le, 29, 127 X, 486, 499 conguration les, 137, 196, 241 conguration scripts, X, 505 ./configure, 238, 406, 537 configure, 30 Configure.help, kernel, 483 conguring DNS, 438 Samba, 431 X, 498 conguring and administration, RHCE, 544 conguring libraries, 235 conguring printers, Samba, 434 conguring windows, Samba, 433 connect, 263 connect(2), 263 connect mode, modem, 24 connection, TCP, 296, 300 console, 11 L INUX, 11 continue, 225 continue, 65 control, TCP, 265 control eld, package, 244 conventions, X, 496 convert, 183, 332 convert to binary, 183 convert to decimal, 183 converting image les, 332 Cooledit, 118 cooledit, 58, 238, 507 cooling, SCSI, 477 Coolwidgets, 492 copy directories, 532 les, 532 recursive, 112 wildcards, 532 COPYING, 32, 237 copying recursively, 34 software, 574

CORBA, 286 core le, 227 core dump, 227 core.html, 394 costing, L INUX, 564 counter measures, security, 516 country codes, 274 course notes, 2 training, 2 cp, 34, 112, 175, 324 usage summaries, 33 cp(1), 36 cpio, 46 .cpp, 29 CPU, 17, 81, 141, 207, 208, 239, 437, 480, 481 priority, 87, 296 usage, 87 CPU, 89 CPU consumption, 88, 108 CPU limits, 176 CPU time, 82 cracking, 103 CREATE TABLE, postgres, 418 createdb, 414 createlang, 414 createuser, 414 creating boot disks, 147 DLL, 233 les, 12 creating tables, postgres, 418 cron, 535, 545 cron packages, 412 cron.daily, 410 cron.hourly, 410 cron.monthly, 410 cron.weekly, 410 crond, 341, 346, 409, 410, 520 cross platform, 568 cryptography, RHCE, 549 .csh, 30 CSLIP, 536 ctime, 126 Ctrl key, 10, 12, 41, 42, 411, 493495 Ctrl-Alt-Del, 11 Ctrl-PgDn, 11 Ctrl-PgUp, 11 CTS, 22 cu, 142 cug, 549 CustomLog, 395 cut, 182, 532 cut buffer, X, 497 cutting, X, 497

.cxx, 29 cylinder, disk, 153

D
D, 90 D key, 10, 12, 411 daemon, 99, 196, 299 daemon process, 184 data, le, 7 data packet, 263 data rate, serial, 22 data set ready, see DSR data terminal ready, see DTR data transmission, 247 database, 413 postgres, 414 database les, 30 database normalization, 423 database of les, 43 database table directory, 415 database theory, 423 datagrams, 247 date, 36 daytime, 269 .db, 30 db, 223 dd, 113, 147, 167, 168, 321, 324 .deb, 30, 135, 237, 241, 244, 301, 562 Debian, 3 Debian, 1, 3, 30, 118, 135, 156, 178, 183, 187, 195, 197, 237, 240242, 244246, 256, 288, 289, 292, 327, 336, 338, 415, 416, 437, 438, 461, 474, 483, 503, 506, 561 .deb, 3 Debian package, 30 debsums, 243 debug support, 239 debugger, 223 debugging, uucp, 343 debugging level, uucp, 343 debugging session, 223 DEC Alpha, 240, 552 decimal, 183, 209 DECnet, 426 decompression, 399 default, print queue, 199 default PATH, 107 default gateway, 253, 256 default visual, X, 505 DefaultDepth, X, 503 defaultroute, pppd, 455 DefaultType, 395 defconfig, kernel, 483 denition, spam, 311

194, 255, 341, 498,

587

588

Index

delete/dropping a column, postgres, 420 delete/dropping a table, postgres, 420 deleting les, 184 deleting interfaces, 251 deleting lines, sed, 79 delivery, mail, 299 delivery date add, 305 demodulation, 24 denial of service, see DoS density, oppy disk, 144 Deny, 396 deny access, 293 dep, make, 538 dependencies, package, 241, 245 dependency, 230, 231 dependency checking, 241 depmod, 327, 546 make, 538 kernel, 464 depmod -a, kernel, 484 destination address, IP, 248, 253, 264 destination port TCP, 265 UDP, 269 detachment, TCP, 263, 268 detection software, security, 524 /dev, 141, 156, 517 /dev/audio, 471 /dev/bttv0, 146 /dev/cdrom, 129, 145, 146, 163 /dev/cdwriter, 146 /dev/core, 146 /dev/cua?, 144, 454 /dev/dsp, 141, 142, 464, 471 /dev/fb?, 145 /dev/fd?, 144 /dev/fd0, 113, 147, 148, 321 /dev/fd0D1440, 145 /dev/fd0H1440, 145 /dev/fd0H1920, 144, 145 /dev/ftape, 146 /dev/gpmdata, 146 /dev/hd?, 155 /dev/hda, 141, 143, 320 /dev/hda1, 143, 154, 329 /dev/i2o*, 146 /dev/i2o/*, 146 /dev/log, 146 /dev/loop0, 324 /dev/lp?, 144 /dev/midi, 471 /dev/mixer, 471 /dev/modem, 144, 146, 454 /dev/mouse, 142, 144, 146 /dev/null, 144, 205

/dev/par?, 144 /dev/pcd?, 144 /dev/pd?, 144 /dev/pf?, 144 /dev/printer, 146 /dev/psaux, 144 /dev/radio, 146 /dev/ramdisk, 146 /dev/root, 146 /dev/scanner, 146 /dev/scd0, 479 /dev/scd?, 144, 146 /dev/sd?, 155 /dev/sd??, 144 /dev/sda, 320, 476 /dev/sequencer, 471 /dev/sg*, 146 /dev/sg?, 145 /dev/sr?, 144 /dev/st?, 144 /dev/swap, 146 /dev/tape, 146 /dev/tty?, 145 /dev/tty1, 330 /dev/ttyI?, 145 /dev/ttyS?, 144 /dev/ttyS0, 334, 335, 342, 454, 479 /dev/ttyS4, 330 /dev/urandom, 144, 148 /dev/zero, 144, 147, 148 DEVICE, 201, 254 Device section, X, 502 device, SCSI, 143, 146 device driver module, X, 502 device drivers, 463 Windows, 463 device le, 141 device ID, PCI, 474 device independent le, 30 device names, 143 devices IDE, 143 serial, 20 devices.txt, 143, 145 devpts, 163, 167 df, 36, 533 DFS, 430 dga, 494 DHCP, RHCE, 548 dhcpd, 540, 546, 548 diagnostic packet, 260 diagnostic utilities, IP, 260 dial-in, pppd, 460 dial-on-demand, pppd, 458 diald, 458

dialout, pppd, 454 dialout group, 336 dialup, 536 uucp, 338 DNS, 449, 460 networking, 453 PPP, 338 dialup server, 460 dictionary, 75 dictionary attack, 103 .diff, 30, 179 diff, 179 dig, 180, 284, 540 dig(1), 284 digital signal processing, 40 Digital Signal Processor, 142 dip, 461 dip -t, 330, 455 .dir, 30 dircmp, 37 dircmp(1), 37 DirectColor, X, 505 directive, 394 directories, 12 copy, 532 removing, 34 directors, exim, 305 Directory, <Directory /directory>, 396 directory, 128 directory locking, 190 directory permissions, 123 directory superstructure, 136, 137 directory trees, 179 DirectoryIndex, 395 discard, 269 disk boot, 147 cylinder, 153 duplication, 148 erasing, 147 oppy, 144 hard drive, 143 heads, 153 identifying, 148 parallel port ATAPI, 144 parallel port CD-ROM, 144 parallel port IDE, 144 paritions, 153 read-write test, 161 SCSI, 144 sector, 153 sides, 153 tracks, 153 disk drive IDE, 18

SCSI, 19 disk dump, 147 disk partitions, LPI, 533 disk space, 128 diskdruid, 544 diskless workstations, 295 DISPLAY, 529, 547 X, 494 display, X, 486 -display option, X, 494 distinguishing directories, ls, 13 distribute, software, 574 distributed, mail, 337 Distributed File System, see DFS distribution, X, 497 distributions, L INUX, 54, 117, 291, 302, 347, 522, 554, 560, 573 security, 516 djbdns, DNS, 437 DLL, 137, 233 creating, 233 dynamically linked library, 31 installation, 235 symlink, 235 versioning, 234 DMA, 17, 466, 536 DMB, Samba, 428 dmesg, 37, 473, 534 dnrd, 460 DNS, 269, 273, 291, 427, 430, 437, 528, 541 .in-addr.arpa, 445 /etc/named.conf, 444 /usr/sbin/named-bootconf, 438 ANY record, 284 AXFR record, 284 A record, 283, 441443, 446 CNAME record, 283 HINFO record, 284 MX record, 284, 441443, 446 NS record, 283, 441443, 446 PTR record, 283, 441443, 446 SOA record, 284, 441443, 445, 446 djbdns, 437 hostname, 438 named-xfer, 438 named.boot, 438 named, 437, 438 ndc, 438 style.txt, 438 authoritative, 441 caching, 281 caching name server, 449 canonical name, 284 conguring, 438 dialup, 449, 460

589

590

Index

dynamic IP, 450 example conguration, 438 forward lookups, 281, 286 infrastructure, 275 lookup, 276 master setting, 451 messages, 444 Name Server Operations Guide, 438 record, 283 reverse lookup, 286, 445 reverse lookups, 281, 442, 447 RHCE, 548 secondary server, 450 security, 437 serial number, 445 servers, 276 slave server, 450 slave setting, 451 starting bind, 443 TTL, 445 zone, 444 DNS conguration, 432 DNS lookup, Apache, 395 DNS lookup conguration, 277 DNS query, 180 dnsdomainname, 540 doc/, 137 documentation, 131, 137, 534 U NIX, 35 info, 35 man, 35 postgres, 414, 418 uucp, 338 module, 469 reference, 2 Samba, 435 tutorial, 2 X, 491, 497 documentation les, 195 DocumentRoot, 395 domain, 273, 407 domain, 269, 280 Domain and Type Enforcement, see DTE domain information groper, 284 Domain Master Browsers, see DMB domain name, 273 Domain Name System, see DNS domain part, 300 domain servers, RHCE, 548 domainname, 540 done, 63 DOS, see MS-DOS DoS, security, 515 dot lock le locking, 188 dots per inch, 201

dotted decimal, 248 double, quotes, 70 double, 210, 215 double density, 145 down arrow key, 55 download packages, 195 dpi, 201 dpkg, 194, 240, 242, 243, 537 dpkg-dev, 246 drive, SCSI, 476, 536 driver, 305 driver support, 117 drivers, 132 DROP TABLE, postgres, 420 dropdb, 414 droplang, 414 dropuser, 414 dselect, 245, 537 DSIZE, 90 DSO, Apache, 406 DTE, security, 522 du, 37, 78, 533 dump, 166 dumping and restoring tables, postgres, 422 duplicating disks, 148 duplication, disk, 148 DVI, 204 .dvi, 30 dynamic CGI, 389 Dynamic DNS, 430 dynamic IP, DNS, 450 dynamic linker, 235 dynamic linking, 233, 234 dynamic loading, kernel, 469 Dynamic Shared Object, see DSO Dynamically Linked Libraries, see DLL dynamically linked library, DLL, 31

E
e2fsck, 165 e2fsck(8), 165 echo, 37, 176, 269 ed, 57, 77 editing, sed, 77 editors U NIX, 57 jed, 58 mcedit, 58 vim, 58 vi, 58 Emacs, 59 graphical, 58 IDE, 58 MS-DOS, 54 editres, 494

edquota, 533, 545 .edu, 274 effective GID, 109, 336 effective UID, 109 efs, 163 EGA, X, 505 egrep, 52 EISA, RHCE, 544 .el, 30 electrical noise, 23 Electronic Mail, 97 Emacs editors, 59 X, 59 email Internet, 97 message body, 98 message header, 98 email address, 97 email assistance, LPI, 535 embedded, market, 552 Emulate3Buttons, X, 497 emulation, SCSI, 478 encapsulated, 264, 268 encoding Apache, 399 byte, 7 octet, 7 encrypted password, 102 encrypted stream, 270 encryption security, 515, 518 TCP, 270 End key, 10 english.hash, 75 enscript, 201, 203, 204 Enter Key, 8, 9, 12, 38, 53, 54, 266, 418 entity modeling, 423 ENV, 93 envelope to add, 305 environment, 90, 186 process, 90 environment variable, 90, 91, 176, 187, 236 environment variables LPI, 531 security, 515 equality operator, 213 equivalent host, 202 erasing, disk, 147 erasing les, 184 errno.h, 26 error checks, 217 error codes L INUX, 26 complete list, 26

error correction, 24 error log, 106 error messages, 26, 73, 198 ErrorLog, 395 errors address, 303 Esc key, 11, 5355 escape sequence, 209 esd, 545 et cetera, 137 /etc, 137, 156, 196 /etc/aliases, 301, 310, 314 /etc/conf.modules, 534, 545, 546 kernel, 469 /etc/cron.daily/, 411 /etc/cron.hourly, 346 /etc/crontab, 346, 409, 410, 545 /etc/dhcpd.conf, 548 /etc/exim.conf, 302 /etc/exim/config, 302 /etc/exim/exim.conf, 302 /etc/exports, 286, 518 /etc/fstab, 166, 533, 545 /etc/group, 104, 545 /etc/host.conf, 277279, 540, 541 /etc/HOSTNAME, 540 /etc/hostname, 256, 540 /etc/hosts, 277, 278, 286, 540, 541 /etc/hosts.allow, 293, 540, 547 /etc/hosts.deny, 293, 297, 540, 547 /etc/hosts.equiv, 202 /etc/hosts.lpd, 202 /etc/inetd.conf, 291, 338, 389, 434, 459, 513, 518, 540, 547 /etc/init.d/, 196 /etc/init.d/exim, 302, 308 /etc/init.d/ftpd, 294 /etc/init.d/httpd, 294 /etc/init.d/inet, 294 /etc/init.d/inetd, 291 /etc/init.d/named, 294 /etc/init.d/rcS, 327 /etc/init.d/smbd, 432 /etc/inittab, 326, 328, 341, 460, 506, 545 /etc/inputrc, 11 /etc/isapnp.conf, 470, 545 /etc/isapnp.gone, 470 /etc/ld.so.conf, 235, 537 /etc/lilo.conf, 319, 468, 481, 534 /etc/logrotate.conf, 198 /etc/lpd.perms, 202 /etc/mail/, 314 /etc/mail/access, 548 /etc/mail/local-host-names, 314 /etc/mail/relay-domains, 314 /etc/mail/virtusertable, 548

591

592

Index

/etc/man.config, 187 /etc/mgetty+sendfax/, 333, 461 /etc/mgetty+sendfax/login.config, 341, 461 /etc/mgetty+sendfax/mgetty.config, 331 /etc/mgetty+sendfax/new fax, 331 /etc/mgetty/, 461 /etc/mgetty/login.config, 341, 461 /etc/mime.types, 115 /etc/modules.conf, 472, 534, 537 kernel, 469 /etc/motd, 521 /etc/named.boot, 438, 541 /etc/named.conf, 438, 541 DNS, 444 /etc/network/interfaces, 255 /etc/network/options, 255 /etc/networks, 540 /etc/nologin, 106 /etc/nsswitch.conf, 541 /etc/pam.d, 547 /etc/passwd, 102, 107, 171, 186, 300, 341, 514, 545 le format, 102 /etc/ppp/chap-secrets, 453, 455, 461 /etc/ppp/options, 461 /etc/ppp/pap-secrets, 453, 455 /etc/printcap, 195, 196, 333, 538, 545 /etc/profile, 38, 186, 187, 535 /etc/profile.d, 187, 545 /etc/rc?.d/KNNservice, 326 /etc/rc?.d/SNNservice, 326 /etc/rc.d/, 546 /etc/rc.d/init.d/, 196 /etc/rc.d/rc, 327 /etc/rc.d/rc?, 326 /etc/resolv.conf, 277, 278, 540, 541 /etc/samba/smb.conf, 431 /etc/samba/smbpasswd, 431 /etc/samba/smbusers, 431 /etc/securetty, 107 /etc/security, 547 /etc/security/, 549 /etc/sendmail.cf, 314 /etc/sendmail.mc, 548 /etc/services, 269, 285, 291, 434, 513, 540 /etc/shadow, 102, 103 /etc/skel, 545 /etc/skel/, 535 /etc/smb.conf, 541, 548 /etc/squid/squid.conf, 546 /etc/sysconfig/, 254 /etc/sysconfig/desktop, 546 /etc/sysconfig/network-scripts/, 544

/etc/sysconfig/static-routes, 547, 549 /etc/sysctl.conf, 545 /etc/syslog.conf, 535, 545 /etc/usertty, 106 /etc/uucp/call, 339 /etc/uucp/passwd, 339341 /etc/uucp/port, 340, 341 /etc/uucp/sys, 339, 340, 342, 345 /etc/X11/fs/config, 509 /etc/X11/XF86Config, 486, 500, 509, 539, 546 /etc/X11/xfs/config, 509 /etc/X11/xinit/, 505 /etc/XF86Config, 486 /etc/xinetd.conf, 295 /etc/xinetd.d, 297, 389, 512, 518 /etc/xinetd.d/swat, 434 /etc/xinetd.d/wu-ftpd, 296 eth0, 250, 252, 475 eth1, 250 ethereal, 550 Ethernet, 426, 464 Ethernet device, 254 ethernet device, 472 Ethernet frame, 248, 250 Ethernet interface, 251 Ethernet port, 250, 256 ETRAX-100, 553 EUID, 92 European date formats, postgres, 416 eval, 175 events, X, 487 evolution, U NIX, 10 exam questions, LPI, 531 example, partitioning, 155 example conguration, DNS, 438 example session, TCP, 265 Excel, 567 exclusive access, 187 exclusive lock, 188 exclusivity, 189 -exec option, find, 43 #exec, 401 exec, 96, 175 ExecCGI, 396, 402 executables, 137, 233 execute permission, 124 executed, 81 executing remote command, 112 execution, 208 kernel, 318 execution time, 81 exim, 241, 299 package, 301 procmail support, 308 route list, 344

spec.txt, 309 conguration, 302 directors, 305 full blown mail server, 306 routers, 306 transports, 304, 343 Why?, 301 exim sources, 188 exim(8), 309 exim, with uucp, 343 exim.conf, 343, 345 exim group, 304 exim user, 304 EXIT, 177 exit, 37, 329 exit code, 176, 179 expand, 532 expansion, 173 expectsend sequence, 454 exploit, security, 513, 515, 516 exponential notation, 210 export display, LPI, 539 exportfs, 287 exporting a le system, 286 exports(5), 288 exposed window, X, 487 exposure, Internet, 515, 518 expr, 37, 71, 78, 183 ext2, 160, 163, 287 ext3fs, le system, 160 extended partition, 154, 321 regular expressions, 52 extensible, 492 X, 487 extension, le name, 29 extensions, list of common, 29 extra density, 145

F
F key, 38 F1 key, 493 F2 key, 20 F6 key, 493 F7 key, 494 F8 key, 495 F9 key, 10, 20 factory default settings, 24 fancy indexes, Apache, 399 FAQ, 118, 196, 438 security, 523 sendmail, 315 X, 498 FAT, le system, 160

FAT le system, Windows, 160 FAT32, le system, 160 fault tolerance, 337, 338 fax image le, 30 fax filter.sh example, 333 faxes, 341, 461 incoming, 330 receiving, 325, 330, 331 faxing, 333 FCEDIT, 94 fcntl locking, 188 fdformat, 162 FDISK, 147 fdisk, 155, 158, 159, 536, 544 fdutils, 162 fetchmail, 180 fg, 82, 532 fgetc, 218 fgets, 513 fgetty, 329 fgrep, 51 FHS, 347 LPI, 533 FHS home page, 347 FIGNORE, 94 le compression, 41 data, 7 Samba, 425 file, 37, 227, 305 FILE *, 217 le descriptors, 75 le extension, 181, 183 le format, /etc/passwd, 102 le formats, 37 le length, 126 le lists, package, 243, 245 le locking, 188, 342 le management, LPI, 532 le manager, 40 le name extension, 29 length, 7 name, 29 le name completion, 11 le name extensions, U NIX, 29 le name, allowable U NIX, 7 le names U NIX, 12 characters, 12 le numbers, 75 le operations, 217 le size, 182, 199 le system, 37, 160, 233, 318

593

594

Index

L INUX, 135 ext3fs, 160 FAT, 160 FAT32, 160 journaling, 160 mounting, 163 MS-DOS, 161 read-only, 168 reiserfs, 160 XFS, 160 le system check, 165 le system image, 324 le system permissions, 123 le system standard, L INUX, 347 le time, 182 File Transfer Protocol, see FTP, 113 le-server, 286 le-sharing, NFS, 285 Files, <Files . . . >, 397 Files section, X, 508 les, 7 copy, 532 creating, 12 hidden, 25 removing, 34 Filesystem Hierarchy Standard, see FHS lter, 74, 333, 336 LPI, 532 print, 200 program, 201 FIN, TCP, 264 nd, 49 find, 42, 180, 517, 533, 549 -exec option, 43 -name option, 43 -type option, 43 find(1), 43 finger, 114, 269, 295 fingerd, 294 rewall, 458 security, 515 rewalling, RHCE, 547, 549 rst sector, 143 xing le systems, 165 float, 210, 215 Floating Point Exception, 86 flock le locking, 188 oppies, MS-DOS, 148 oppy 3.5-inch, 144, 162 disk, 144 formatting, 161 mounting, 164 MS-DOS, 44, 162, 164 oppy devices, RHCE, 544

oppy disk, 113, 144, 321 density, 144 oppy disk boot, LPI, 536 oppy drives, 18 ow control, serial, 22 fmt, 532 -fn, X, 506 FollowSymLinks, 396 -fomit-frame-pointer, 239 font, X Window System, 29 font conguration, X, 508 font database, X, 509 font directory, X, 508 font family, X, 506 font generation, 132 font metric, 31 font name elds, X, 506 font naming, X, 506 font path, X, 508 font pixel resolution, X, 507 font server, X, 509 font size, X, 507 font weight, X, 507 fonts.alias, 508 fopen, 217 fopen(3), 217 for, 63, 211, 218 for les, searching, 42 force overwriting, 178 foreground, 82 jobs, 532 forging a package, 244 forking twice, 184 FORM, 405 format, 144 formatting oppy, 161 partition, 160 formatting of output, 176 forms, Apache, 403 forward, quotes, 70 .forward, 541 forward, 271 forward lookups, 203, 528 DNS, 281, 286 forwarding IP, 253, 255 RHCE, 547 TCP, 270 Fox, Brian, xxxi Fox, Brian, 83, 92, 172 -fPIC -DPIC, 234 fprintf(3), 218 fputc(3), 218 frame, 250

frame buffer, 145 frame-relay, 547 free, 37, 214, 217 free memory, 37 Free software, 559 Free Software Foundation, see FSF, 573 free software licenses, 556 Free SQL server, 414 FreeBSD, 552 freedom, 573 Frequently Asked Questions, see FAQ, 118, 196 Fresh Meat, 118, 245 From, 98 FRS, 556 fsck, 165, 533 fsck.ext2, 165 fseek(3), 218 FSF, 551, 553, 556, 557, 573 fsinfo, 494 fslsfonts, 494 fsp, 269 fstobdf, 494 FTP, 113, 272, 294, 514, 562 RHCE, 546 ftp, 269, 294, 515, 518, 540 FTP archives, 117 FTP mirror, 117 FTP service, 296 FTP site, 117, 237 FTP upload, 114 ftp://, see web sites ftpd, 292, 294 ftpd(8), 296 full blown mail server, exim, 306 full path name, paths, 34 full-duplex, 264 fully qualied host name, 273 function, 176, 177, 224 function, 67 function arguments, 211 function prototype, 221 function prototypes, 221 functions, 210 BIOS, 318 fundamental system administration, LPI, 531 fwrite(3), 218

G
-g, 223 G key, 38 .g3, 331 -g<width>x<height>, 201 GATEWAY, 254 gateway, 253 gateway device, 253

gateway machine, 253 gcc, 208, 222, 231, 234, 488 gdb, 223, 226 gdialog, 333 Gdk, 492 gdm, 546 generic device, SCSI, 145 GET, 264 gethostbyname, 278 getpeername, 390 getsockname, 392 getty, 326, 328330 getwd, 513 GET /, 266, 389 GhostScript, 184, 200 GhostView, 200 GID, 103, 105107 GIF, 204 .gif, 30, 332 gimp, 492 Glib, 492 glibc, 278, 561 glob expressions, 32 glob wild cards, 174 glob dot filenames, 95 Gnome, 59 RHCE, 546 gnome-utils, 333 gnorpm, 545 GNU, 2, 553, 557 Licensing, 556 maintaining software, 562 movement, 557 project, 551, 557 software, 557, 558 software cost, 562 GNU General Public License, 237 GNU debugger, 223 GNU standards compliant, 237 GNU Coding Standards, 238 GNU General Public License, 573 GNU Image Manipulation Program, see gimp GNU source packages, 537 LPI, 537 GNU/L INUX, 1, 551, 553 GNUStep, 493 GNU, commands, 25 GNU C Compiler, 208 google.com, 118 gopher, 269 .gov, 274 government organization, 274 gpasswd, 535 GPF, fatal exception Windows, 86

595

596

Index

GPL, 117, 492, 556, 558 aggregation of another, 576 applying to new programs, 579 incorporating parts of programs, 578 lack of warranty, 578 modifying, 575 source code, 576 terms and conditions, 574 version number, 578 gpm mouse, 146 graphic cards, 132 graphical, editors, 58 graphical desktop, L INUX, 9 graphical login box, X, 506 graphical mail readers, 99 graphical mode, 11 Graphical User Interface, see GUI graphics, X, 486 graphics card, 498 X, 504 graphics context, X, 487 graphics driver, 145 graphics les, 200 graphics hardware, 132 X, 486 graphics hardware acceleration, X, 503 graphix, 492 gray-level display, X, 500, 505 GrayScale, X, 505 grep, 43, 49, 74, 75, 171, 179, 181, 185, 200, 226, 537 command summary, 43 Group, 394 group, 123 LPI, 533 group, 106, 305, 535 group owner, 124 groupadd, 106, 535 groupdel, 106 groupmod, 106 groups, 101, 104 gs, 184, 200, 201, 204 gshadow, 535 Gtk, 59, 492, 497 gtop, 549 GUI, 491493, 552 X, 485 GUI interface, 413 guideline, partitioning, 155 gunzip, 537 gv, 200, 201 .gz, 30 gzip, 30, 41, 179, 204, 324, 399, 463

H
597

.h, 30 H key, 36 hack, 450, 511 hacking, 511 halfletter, 201 halt, 325 hang up, 86 hard disk, 143 hard drive, disk, 143 hard links, 129, 533 Harddrive auto-detection, CMOS, 20 hardrive autodetection, 20 hardware address, 250 hardware clock, CMOS, 20 hardware cursor, X, 503 hardware devices, 141, 318 hardware drivers, 132 hardware ow control, 22, 454 hardware mouse, 146 hardware platform, 207 hardware requirements, L INUX, 569 hardware selection, RHCE, 544 hardware support, L INUX, 570 hardware vendor, 552 hash MD5, 244 password, 103 Hayes command set, 24 hdparm, RHCE, 544 hdparm(8), 544 head, 38, 532 header, 98 IP, 247 mail, 300 malformed, 313 print, 199 rejecting, 313 TCP, 264 verication, 313 header les, 26, 30, 220, 230 headers, kernel, 463 headers check syntax, 312 heads, disk, 153 Heavy industry, 552 --help, 25 help command, 227 Hercules Graphics Card, see HGC Hertel, Christopher, xxxi Hewlett Packard, 201, 551 hex, 252 hfs, 163 HGC, X, 505 hidden, les, 25 hidden les, vs. ordinary, 26 Hierarchical NT Domains, 430

high availability, RHCE, 547 high density, 145 high performance, RHCE, 547 HINFO record, DNS, 284 HINFO records, 548 histchars, 95 HISTCMD, 92 HISTCONTROL, 95 HISTFILESIZE, 94 history control, 95 HISTSIZE, 94 hole, security, 514 HOME, 93, 103, 107, 173 /home, 156 home computers, 155 home directory, 101, 103, 107 Home key, 10 /home/*, 431 /home/fax/, 332 homework assignments, 525 hops, 256, 261 host, 203, 281, 540, 541 host name, 113, 273 unqualied, 273 Host name for. . . , 203 host part, 248, 250 host accept relay, 304 host access, 293 HOSTFILE, 95 hosting new software, 118 HOSTNAME, 254 hostname, 38, 540 DNS, 438 hostname(7), 438 hostname completion file, 95 HostnameLookups, 395 hosts, 305 hosts.deny, 297 hosts.equiv, 202 hosts access(5), 293 hosts override, 305 HOSTTYPE, 93 HOWTO, 132 HP, 551 hpfs, 163 .htaccess, 395 .htm, 30 HTML, 204, 389 HTML documentation, postgres, 415 HTTP, 264, 389 HTTP protocol, 266 http://, see web sites httpd, 185, 394, 541 httpd.conf, 402, 407 .hushlogin, 107

I
.i, 30 I key, 54 I/O, 426 I/O port, 20 I/O ports, 17, 18, 20, 466, 470, 480, 536 RHCE, 544 i386, 240 IBM, 425 IBM 390, 553 IBM personal computer, see PC iceauth, 494 icewm, 495 ICMP, 260, 261 LPI, 540 ICMP messages, 180 ico, 494 ID, SCSI, 476 id, 109 IDE, 58, 322, 568 CD-ROM, 18, 478 devices, 143 disk drive, 18 editors, 58 kernel, 477 master/slave, 19 RHCE, 544 ribbon, 18 IDE devices, 143 IDE drive, 148 IDE drives, removable, 161 IDE geometry, RHCE, 544 IDE labeling, L INUX, 20 identd, 294, 550 identication eld, IP, 247, 264 identifying, disk, 148 idle time, 108 IETF, 430 if, 63, 179, 200, 211 #if 0, 222 ifcfg-eth0, 254 ifcfg-eth1, 254 ifcfg-lo, 254 ifconfig, 251, 457, 540, 544 #ifdef, 222 ifdown, 255, 544 #ifndef, 222 IFS, 172 IFS, 93 ifup, 255, 544 ifup(8), 255 IGNOREEOF, 94 IHL, 247, 264 image, 318, 320 image, 115, 320

598

Index

image le, 30, 183 image le conversion, 332 ImageMagic, 332 ImageMagick, 183 IMAP, 295, 299, 301, 337, 338, 546 secure, 518 security, 301 imap, 295, 518 IMAP host, 300 imapd, 294 immutable permissions, security, 520 implicit declarations, 230 improving security, 178 .in, 30 in, 63 .in-addr.arpa, DNS, 445 in.fingerd, 294 in.ftpd, 294, 296 in.identd, 294 in.ntalkd, 294 in.rlogind, 294 in.rshd, 202, 294 in.rshd(8), 113 in.talkd, 294 in.telnetd, 294 in.tftpd, 294 #include, 220, 222, 230 include/, 138 Includes, 396 IncludesNOEXEC, 396 incoming, faxes, 330 /incoming, 114 incoming mail, 300 incorporating parts of programs, GPL, 578 increment, 212 index.html, 390, 394 index.shtml, 401 Indexes, 396 indexes, Apache, 399 indexing, 229 inet6, 280 inetd, 291, 301, 394, 434, 513, 518, 540, 546, 547 security, 297 services, 294 TCP, 291 inetd.conf, 291, 294 .info, 30 info, 36, 227 documentation, 35 INFO pages, 137 info/, 137 infoseek.go.com, 118 infrastructure, DNS, 275 init, 325, 326, 328, 506, 534, 546 init.d, 196

init.d script, postgres, 415 initdb, 414 initialization, 186 initialization les, 196 initialization string, modem, 24, 330, 342, 453, 454 initializing, postgres, 415 initlocation, 414 initrd, 324 initrd image, 322 inittab, 326, 328, 330, 331, 506, 546 inittab(5), 327 innd, 197 INPUTRC, 94 .inputrc, 94, 539 INSERT INTO, postgres, 420 inserting lines, sed, 79 inserting rows, postgres, 420 inside a loopback device, L INUX, 167 insmod, 323, 537, 546 kernel, 464, 468 inspecting program execution, 223 INSTALL, 32, 237 install count, L INUX, 563 installation, 245 DLL, 235 kernel, 463 package, 193 installation oppy, 147 installation instructions, 237 installation methods, RHCE, 545 installation prex, 137 installation size, L INUX, 569 installing L INUX, 20 postgres, 415 Apache, 393 kernel, 483 package, 240 instruction, 208 .int, 274 int, 210, 215 integer, 210 Integrated Services Digital Network, see ISDN Intel, 474, 493 Intel Architecture 64-bit, 552 interactive, 175 interactive shell, 186 intercharacter spacing, X, 507 interface, 250 aliasing, 259 testing, 260 interface conguration, 251 interfaces(5), 255 Interim Mail Access Protocol, see IMAP internal modems, 480

internal tables, postgres, 417 international treaties, 274 Internet, 117, 120, 132, 247, 273275, 337, 437, 444447, 461, 516518, 552 email, 97 exposure, 515, 518 security, 511 servers, 552 spam, 99 Internet communication, 247 Internet connection, 21, 180 Internet Control Message Protocol, 260 Internet Engineering Task Force, see IETF Internet Explorer, 568, 569 Internet header length, 247, 264 Internet Protocol, see IP, 247 Internet Service Provider, see ISP Internic, 282 Internic, 276 interoperability, Samba, 425 interpreted languages, 208 interpreter, 207 interrupt, 86 interrupt execution, 224 Interrupt Request lines, IRQ lines, 17 interrupts, 17, 467 BIOS, 322 IO, 536 IO Control, 142 IO-port, 323 ioctl, 142 IP, 180, 247, 254 address, 247, 250, 252, 256, 273, 277, 300 address classes, 249 checksum, 248, 264 destination address, 248, 253, 264 diagnostic utilities, 260 forwarding, 253, 255 header, 247 identication eld, 247, 264 length, 247, 264 Next Generation, 248, 280 offset, 248, 264 options, 248, 264 source address, 248, 264 special-purposes addresses, 249 testing, 260 time to live, 248, 264 TOS, 247, 264 TTL, 248, 264 type of service, 247, 264 version 6, 248, 280 IP address, 182, 203, 392, 408, 438, 445 Samba, 428 IP masquerading, 459

IP packet, 248 IP-based virtual hosting, Apache, 407 ip forward, 255 IPADDR, 254 ipcclean, 414 ipchains, 458, 547, 549 ipfwadm, 458 ipliteral, 306 ipop3d, 294 iptraf, 550 IPX, 285 IPX/SPX, 426 IRIX, 551, 561 IRQ, 18, 20, 323, 466, 467, 470, 480, 536 RHCE, 544 IRQ lines, Interrupt Request lines, 17 IRQ port, 536 ISA, 462, 467, 470, 480 slots, 17 ISA cards, 467 ISA device drivers, 467 ISA jumpers, 472 ISA slots, 8-bit, 18 isapnp, 536 isapnp, 327, 470, 545 RHCE, 544 isapnp.conf, 470 ISDN, 462, 547 LPI, 540 PPP, 453, 462 ISDN modems, 145, 462 ISO 9660 partition, security, 520 ISO Latin 1, 507 ISO9660, 479 iso9660, 163, 287 ISP, 24, 180, 274, 462, 552

J
Java, 414 Jazz, 161 JCPU, 108 jed, 58, 59 editors, 58 JFIF, 37 job control, 82 job number, 83 jobs, 82 background, 108, 176, 532 foreground, 532 LPI, 532 scheduling, 409 jobs, 532 joe, 59 join, 532 journaling, le system, 160

599

600

Index

JPEG, 37, 114, 204 .jpg, 30 jumpers, 17 junk mail, 99

K
kaudioserver, 545 kbdrate, 38 KDE, 59, 492 RHCE, 546 kdm, 546 Kerberos V, 430 kernel, 82, 214, 227, 238, 320 L INUX, 551, 560 /etc/conf.modules, 469 /etc/modules.conf, 469 2.2, 464 2.4, 464 Configure.help, 483 Makefile, 464 System.map, 464, 484 bzImage, 484 defconfig, 483 depmod -a, 484 depmod, 464 insmod, 464, 468 klogd, 464 lsmod, 464, 465 make menuconfig, 483 make targets, 483 modinfo , 468 modprobe, 465, 468 rmmod -a, 465, 471 rmmod, 464 ATAPI, 477 boot, 325 boot oppy, 484 boot image, 463 boot options, 317, 320 building, 481, 483 CD writer, 477, 478 CD-ROM, 477 commercial drivers, 482 compile, 481 compiled-in modules, 464, 483 compiled-out modules, 464 conguration, 482 dynamic loading, 469 execution, 318 headers, 463 IDE, 477 installation, 463 installing, 483 module, 463 multiple Ethernet cards, 475

NFS, 289 Plug-and-Play, 470 RHCE, 545 SCSI, 475, 476 SMP, 483 sound, 469 sound cards, 470 source, 463 source code, 463 support daemons, 463 unpacking, 481 versioning, 464 kernel 2.4, 473 kernel 2.4, 458 kernel image, 295, 318 kernel loader, Windows, 468 kernel module documentation, 469 kernel source, RHCE, 546 kernel source tree, 482 kernel sources, 118 kernelcfg, 546 key conventions, X, 493 key press, X, 488 keyboard repeat rate, 38 keys, less, 38 keys, common bindings L INUX, 10 kill, 86 kill, 84, 180, 190, 532 kill the X server, X, 493 killall, 85, 292 killing netscape, 185 klogd, 545 kernel, 464 kpackage, 545 kpm, 549 ksysv, 546

L
l command, 224 label, 320 lack of warranty, GPL, 578 LAN, 107, 114, 248250, 256, 425, 426 language, 207 language negotiation, Apache, 399 LanguagePriority, 400 Large Block Addressing, 154 laserjet printer, 30 last, 549 last sector, 143 lastcomm, 549 latency, 87 A L TEX, 57 A L TEX, 29, 31, 132 LBA, 154, 319

lbxproxy, 494 LCK..device, 342 LD LIBRARY PATH, 187, 233, 236 LDAP, 430, 548 ldapd, 547, 548 ldconfig, 233, 235, 537 ldd, 234, 537 LDP, 133 leak memory, 214 lecture schedule, 525 ledger, 201 left arrow key, 10, 55 legacy, 195, 203 legal, 201 length le name, 7 IP, 247, 264 UDP, 269 less, 38, 143 beeping, 38 keys, 38 usage, 38 lessons, 525 letter, 201 level I, certication LPI, 531 LI prompt, 156 LIB, 90 /lib, 156, 230 lib, 137 /lib/, 233 /lib/libc.so.6, 228 /lib/modules/, 322 /lib/modules/<version>/, 463 /lib/modules/<version>/modules.dep, 465 libc, 278 libc6, 561 library, 227, 233 C, 227 static, 29 library calls, 217 library compatibility, 235 library le, 233 library function C, 209, 216 X, 488 library name, 232 library versioning, 234 libsafe, security, 518 libX11.so, X, 491 libXt.so, X, 491 Licensing, GNU, 556 LIDS, 522 Lightweight Directory Access Protocol, see LDAP

lilo, 317, 318, 463, 467, 481 make, 538 lilo(5), 324 lilo.conf, 319, 321, 324, 468, 484, 536 LILO:, 317, 320, 328 prompt, 329, 467, 478 limitations, BIOS, 319 limits, 176 line break, 337 line continuation, 75 line number, 180 line printer, see printer linear, 324 LINENO, 92 link count, 190 linked libraries, 234 linking, 230 links, 127, 128, 146 Linus Torvalds, 551 L INUX, costing, 564 L INUX compatibility, 551 console, 11 distributions, 54, 117, 291, 302, 347, 522, 554, 560, 573 error codes, 26 le system, 135 le system standard, 347 graphical desktop, 9 hardware requirements, 569 hardware support, 570 IDE labeling, 20 inside a loopback device, 167 install count, 563 installation size, 569 installing, 20 kernel, 551, 560 keys, common bindings, 10 migration, 567, 568 native le system, 160 news, 118 number of users, 563 on one oppy, 138 partition, 319 porting, 551 resources, 117, 527 restarting, 11 running Windows programs on, 571 support, 563 users, 117, 118 virtual consoles, 329 Windows in another partition, 143 L INUX Software Map, 556 Linux Documentation Project, 133 Linux Intrusion Detection System, see LIDS

601

602

Index

Linux Loader, 317 Linux Planet, 118 Linux Software Map, 30 Linux Weekly News, 118 linuxconf, 196, 545549 /linuxrc, 322 Lisp program source, 30 list, 224 ls, 10 listing a table, postgres, 419 listing databases, postgres, 417 listing domains, 284 listres, 494 listserv, 119 literal constant, 222 .lj, 30 LMB, Samba, 428 ln, 189, 190 lndir, 494 lo, 251 LOADLIN.EXE, 468 LoadModule, 406 local, 137, 175 Local Area Network, see LAN local domain, 300 local function, 229 Local Master Browser, see LMB local delivery, 305 local delivery:, 305 local domains, 304, 305 locale settings, 507 localhost, 286, 445, 494 localhost.localdomain, 445 localuser:, 305 locate, 43, 409, 533, 545 locating PCI modules, 474 locating rows, postgres, 421 .lock, 188 lock, 320 lock le, modem, 335 lockd, 286 lockfile, 190 locks, 342 log, bootup messages, 37 .log, 30 log le rotation, 535 log les, 30, 156, 198 rotation, 198 log format, Apache, 395 log program, 106 log subject, 303 logcheck, 545 LogFormat, 395 logging, 296 logging in, 9

logical, partition, 154 login authentication, 330 name, 9 password, 9 prompt, 9, 330 terminal, 330 login, 106, 107, 186, 294 login limits, LPI, 542 login name, 40, 97 login shell, 184, 186 login: prompt, 506 LogLevel, 395 LOGNAME, 107 logout, 13 logrotate, 198, 545 long, 215 long double, 215 long le name, 44, 162 look-and-feel, 567 lookup, DNS, 276 lookuphost, 306 loop, 212, 216, 218 loopback device, 167, 251 loopback interface, 251 loopholes, 513 security, 516 looping, 62 loops, LPI, 539 losetup, 167, 168, 324 /lost+found, 165 lp, 193, 199 lp print queue, Samba, 432 lpc, 196, 203, 538, 545 lpd, 146, 193, 196198, 203205, 227 RHCE, 546 lpf, 196 LPI BIOS settings, 536 booting process, 534 BUGTRAQ, 541 CERT, 541 certication, 2 command history, 531 command-line, 531 disk partitions, 533 email assistance, 535 environment variables, 531 exam questions, 531 export display, 539 FHS, 533 le management, 532 lter, 532 oppy disk boot, 536 fundamental system administration, 531

GNU source packages, 537 group, 533 ICMP, 540 ISDN, 540 jobs, 532 level I, certication, 531 login limits, 542 loops, 539 mail aliases, 541 mail forwarding, 541 mailing lists, 534 monitor, 539 network interfaces, 540 network masks, 540 newsgroups, 534 NT domain, 541 objective weighting value, 531 package corruption, 541 password expiry dates, 541 permissions, 533 PPP, 540 process limits, 542 program priority, 532 quotas, 533 redirecting standard input/output, 532 regular expressions, 532 requirements, 3 SCSI BIOS, 536 security, 541, 542 shadowed passwords, 541 shell commands, 531 simple lesystem problems, 533 streams, 532 SUID bit, 541 suid-rights, 539 TCP, 540 TCP wrappers, 540, 541 telephone assistance, 535 test, 539 UDP, 540 update binaries, 541 video card, 539 web page, 531 web site, 534 wildcards, 532 WINS client, 541 X font server, 539 X server, 539 lpq, 193, 196, 203, 538, 545 lpr, 193196, 204, 538, 545 lprm, 193, 196, 538, 545 LPRng, 194, 202 lptest, 196 ls -al, 25

ls -l, 13 distinguishing directories, 13 list, 10 usage summaries, 33 ls -l, ls, 13 LS120, 161 lsattr, 520, 549 .lsm, 30 lsmod, 537, 546 kernel, 464, 465 lspci, 473, 474 LUN, SCSI, 476 lurkftp, 550 lwn.net, 118 Lycos, 119 lynx, 39 LyX, 30 .lyx, 30

M
m4 macro, 546 MAC, security, 522 Macintosh, 425 MacOS, 58 macros, 222 magic lter, 204 magic numbers, 37, 204 magnetic head, 153 MAIL, 93, 107 RCPT, 313 mail, 193, 291, 528 uucp, 343 clients, 99 delivery, 299 distributed, 337 header, 300 new message, 301 queue, 299 queue deletions, 309 queuing, 302 relay, 304 removing messages, 309 retries, 302 retry schedule, 300 routing, 299, 306 testing, 100 mail, 99 Mail Abuse Prevention, 313 mail aliases, LPI, 541 mail box, 97 mail client, 300, 301 mail le, 173 mail folder, 187 mail forwarding, LPI, 541 mail header, 98

603

604

Index

mail header elds, 98 mail message, 97 mail queue, 180 mail readers, terminal, 99 mail retrieval, 295 mail server, 99, 299, 302 mail spool les, 156 mail spooling, RHCE, 548 Mail Transfer Agent, see MTA, 99 MAIL WARNING, 93 mailbox, 113, 299 mailbox le, 97, 301 mailbox le locking, 188 MAILCHECK, 93 Maildir, 300, 301 mailer daemon, 99 MAILER-DAEMON, 310 mailing list, 119 mailing lists, 117, 196, 534 LPI, 534 Samba, 430 MAILPATH, 93 mailq, 301, 309, 541 main function, 208, 211, 224 mainframes, 552 maintaining software, GNU, 562 major device numbers, 143 major MIME type, 115 major version, 238, 240 majordomo, 119 make, 230, 232, 314, 537 bzImage, 538 clean, 538 depmod, 538 dep, 538 lilo, 538 menuconfig, 538 modules install, 538 modules, 538 make targets, 231 kernel, 483 make directory, mkdir, 12 make install, 537 make menuconfig, kernel, 483 makedbm, 547 MAKEDEV, 476, 480 Makefile, 230, 233, 237, 537 kernel, 464 Makefile rules, 231 Makefile targets, 231 makepsres, 494 makestrs, 494 malformed, header, 313 .man, 30 man, 187, 201, 228

documentation, 35 sections, 35 man page, 32, 134, 137, 196 man pages to, PostScript, 134 man/, 137 Mandatory Access Controls, see MAC Mandrake, 156, 554, 561 .rpm, 3 map le, 318 MAPS, 313 market, embedded, 552 markup language, 31 mars-nwe, 546 masquerading, PPP, 459 Master Boot Record, see MBR master IDE, 19 master setting, DNS, 451 master/slave, IDE, 19 MBR, 318, 320 MBX le locking, 188 mc, 40 mcedit, 58 editors, 58 MD5, 244 hash, 244 md5sum, 243 MDA, X, 505 medusa, security, 522 megahertz, 208 Mem, 89 memory, 208, 217 memory allocation, 214 memory consumption, 88 memory leak, 214 memory location, 214 memory reallocation, 218 menuconfig, make, 538 menus, 491 mergelib, 494 message, 115 message body, 300 email, 98 message header, email, 98 messages, DNS, 444 Meta-Font, 30, 132 metalab.unc.edu, 117, 237, 260 .mf, 30 mgetty, 325, 326, 330, 331, 343, 461, 506 mgetty(8), 330 mgetty, with uucp, 341 mice, 21 Microcomputer Organization, 15 microphone, 40, 141 microseconds, 40

Microsoft, 427 Midnight Commander, 40, 58 migrating from, Windows, 567 migrating from another database, postgres, 422 migration, L INUX, 567, 568 mikmod, 41, 87 mikmod(1), 87 .mil, 274 military department, 274 million instructions per second, 208 MIME, 115 RHCE, 548 MIME types, 115, 395 mimencode, 115 miner MIME type, 115 mingetty, 329 Mini HOWTOs, 133 minimal cong le, Samba, 431 minimum header, TCP, 265 minix, 163 minor device number, 143 minor version, 238, 240 minus key, 494 MIPS, 208, 552 mirror, 117 mirrordir, 271 mission-critical applications, 552 MIT, 556 mk-ftp-stats, 550 mkbootdisk, 547 mkbootdisk(8), 547 mkcfm, 494 mkdir, 34, 167, 287, 324 make directory, 12 mkdirhier, 494 mke2fs, 161, 167, 324, 536 mkfontdir, 494, 508 mkfs, 160 mkinitrd, 324 mknod, 148, 464 mkswap, 162 mmencode, 115 /mnt, 156 /mnt/cdrom, 163 mod, audio format, 41 mod les, 41 mode, 305 mode fail narrower, 305 Modeline, X, 502 modem, 2123, 330, 337, 342, 536 AT commands, 342, 453 uucp, 341 carrier signal, 24 command mode, 24

commands, 342, 453 connect mode, 24 initialization string, 24, 330, 342, 453, 454 lock le, 335 speed, 454 modem cards, 480 modem port, 146 modem protocols, 24 modes, vi, 54 modication times, 126 modifying GPL, 575 software, 574 modinfo, 537 modinfo , kernel, 468 modprobe, 537, 546 kernel, 465, 468 modulation, 24 module, 322, 463 documentation, 469 kernel, 463 options, 467 RHCE, 546 SCSI, 322, 323 module support, 322 modules, make, 538 modules.conf, 474 modules install, make, 538 Monitor section, X, 501503 monitor, LPI, 539 monochrome displays, 505 monochrome graphics adapter, see MDA more, 38 motherboard, 15, 18, 208 Motif, 492 Motorola, 552 mount, 163, 166, 167, 286, 287, 324, 533, 545 Samba, 432 mount(8), 163 mount points, 536 mountd, 286 mounting, 163 automatically, 166 CD-ROM, 163 le system, 163 oppy, 164 read only, 166 Windows NT partitions, 164 Windows partition, 164 mounting read-only, 156 mouse, 142 mouse and keyboard input, X, 487 mouse click, X, 488 mouse conventions, X, 497 mouse operations, 497

605

606

Index

mouse port, 146 mouse protocol, 142 mouseconfig, 544 movement, GNU, 557 Mozilla, 568 mpack, 115 mpage, 204 MPEG, 40 mpg123, 40 mrtg, 550 MS-DOS, 18, 44, 58, 59, 81, 144, 147, 160, 285, 318, 321, 426, 479, 525, 570 editors, 54 le system, 161 oppies, 148 oppy, 44, 162, 164 msdos, 163 msp, 269 mt, 149 mt(1), 149 MTA, 99, 299 mtime, 126 mtools, 44, 164 command list, 44 mtp, 269 multi off, 279 multi on, 279 multiheaded display, X, 494 multimedia commands, 40 multiple Ethernet cards, kernel, 475 multiple IP addresses, 448 multiple network cards, 279 multiport card, 144 multiport serial card, 480 multiport serial cards, 330, 479 multiprocessor machines, 552 Multipurpose Internet Mail Extensions, see MIME multitasking, 552 multiuser, 552 multiuser mode, 325 multiuser system, 97 multiuser systems, security, 512 MultiViews, 396 munpack, 115 music, 41 mutt, 99, 190 mutt dotlock, 190 mv, 175 MX record, DNS, 284, 441443, 446 MX records, 528, 548 mx, 199

N
n command, 225

N: drive, 429 name le name, 29 login, 9 resolution, 276 -name option, find, 43 name resolution, Samba, 427 name server, 180, 268, 276, 300, 437, 541 Name Server Operations Guide, DNS, 438 name-based virtual hosting, Apache, 407 named, 519, 546 DNS, 437, 438 named(8), 438 named pipes, 146 named-xfer, DNS, 438 named-xfer(8), 438 named.boot, 438 DNS, 438 nameserver, 269, 280 NameVirtualHost, 408 nano, 59 NAT, see masquerading native execution, 208 native le system, L INUX, 160 native printer code, 200 nausea, 222 navigation, 175 NBNS, Samba, 428 ncpfs, 163 ndc, DNS, 438 ndc(8), 438 ndots, 280 nedit, 59 negotiation, TCP, 263, 268 nessus, 550 .net, 274 NetBEUI, Samba, 426 NetBIOS, Samba, 426, 427, 430 NetBIOS Enhanced User Unterface, see NetBEUI NetBIOS Name Service, see NBNS NetBSD, 552 netcfg, 544, 546549 netconfig, 544, 546, 548 NETMASK, 254 netmask, 253 netpipes, 390 Netscape, 300, 568 netscape, 120, 185, 200, 539 netstat, 269, 513, 518, 540 NetWare, 425 netwatch, 550 NETWORK, 254 network, 249 clash, 250 PPP, 453

network address, 250 network address translation, see NAT network attacks, security, 511 Network Basic Input Output System, see NetBIOS network card, 247, 408 network client programs, security, 514 Network File System, see NFS network le system, 190 Network Information Services, see NIS network interface, 250 Network Interface Cards, see NIC network interfaces, LPI, 540 network intrusion detection, security, 522 network mask, 250 network masks, LPI, 540 Network Neighborhood, 428 Windows, 425 network part, 248, 250 network segment, 261 network services, 178 network transparent, X, 486 network/mask notation, 253 NETWORKING, 254 networking dialup, 453 RHCE, 546 networking cards, 132 never users, 304 new disk, partition, 155 new message, mail, 301 newaliases, 301, 315, 541 newline, 218 newline character, 209 NEWS, 32, 237 news, 337 L INUX, 118 News Forge, 119 news reader, 120 newsgroups, 117, 120, 534 LPI, 534 next, 225 NeXT Computers, 493 Next Generation, IP, 248, 280 NeXTStep, 493 NFS, 190, 268, 285, 286, 325, 426, 529 access permissions, 288 conguration, 286 le-sharing, 285 kernel, 289 processes, 286 RHCE, 546, 549 root mounts, 481 security, 289, 518 nfs, 163, 287

nfs-utils, 518 nfsd, 286 NI, 89 NIC, 472, 536 nice, 88, 296, 332, 532 nice values, 87 niceness, 87 process, 87 NIS, 279 RHCE, 547, 548 NIS domain, 327 nl, 532 nmb, 541 nmbd, Samba, 427, 428 nmblookup, 546 Samba, 428 no operation, 175 no-check-names, 280 no access, 297 no exit on failed exec, 96 no root squash, 288 nobody user, 390 noclobber, 95 --nodeps, 241 nohup, 39 noise, 23 nolinks, 95 non-booting systems, 317 nonprot organization, 274 Norton Commander, 40 nospoof, 279 note, 201 notes, course, 2 notify, 94 NS, 440 NS record, 548 DNS, 283, 441443, 446 nslookup, 282, 284, 541 NT domain, LPI, 541 ntalk (518), 295 ntalkd, 294 ntfs, 163, 164 ntsysv, 546 null-terminated, 213 number of users, L INUX, 563

O
.o, 231 -O0, 223 -O2, 239 -O3, 239 object les, 229 object ID, postgres, 420 object oriented, 492 object relational, postgres, 420

607

608

Index

object relational database, 420 object-oriented, 286, 493 object-relational, 414 objective weighting value, LPI, 531 Objective-C, 493 oclock, 494 octal, 125 octet, encoding, 7 od, 532 ofce suites, 567 offset, IP, 248, 264 oid, postgres, 421 OLDPWD, 92 on one oppy, L INUX, 138 one-way hash, 103 only from, 296 Open Relay Behavior-modication System, 313 Open Software Foundation, see OSF Open Sound project, sound, 474 Open Source, 427 open source, 207 OpenBSD, 552 opening les, 217 OpenSSH, 271 OpenStep, 493 openWall project, 521 Operating System, see OS opportunity for attack, security, 519 /opt, 187 /opt/src, 238 OPTARG, 93 OPTERR, 94 optimization, 223 OPTIND, 93 Options, 396 options IP, 248, 264 module, 467 TCP, 265 options, 280 OR, 216 ORBS, 313 Order, 397 order, 279 .org, 274 .org.za, 276 OS, 20, 207, 317, 320, 551, 564, 566 OS/2, 58, 285, 393, 425, 564, 570 OSF, 551 X, 485 OSTYPE, 93 other, 321 outdated packages, security, 516 outgoing mail, 300 Outlook Express, 300

output, 209 overwriting, 178 ownerships, 514

P
p command, 224 PAC, 430 pac, 196 package, 135, 137, 193, 240, 516 postgres, 414 building, 237 conguration, 193 control eld, 244 dependencies, 241, 245 le lists, 243, 245 installation, 193 installing, 240 querying, 241 removing, 241 selector, 245 source, 246 status, 242 tag, 244 uninstalling, 240 upgrading, 240 versioning, 238, 240 package, exim, 301 package corruption, LPI, 541 package management, 245 package numbering, 240 package verication, 517 package-specic options, 238 packaging-manual, 246 packet, 247, 261, 263, 266 packet ltering, 458 packet forwarding, 255 page size, 201 palette, X, 505 PAM, RHCE, 547, 549 pan, 120 PAP, 456 PAPERSIZE, 201 Parallel port, 18 parallel port, 472, 536 RHCE, 544 parallel port ATAPI, disk, 144 parallel port ATAPI CD-ROM, 144 parallel port ATAPI disk, 144 parallel port CD-ROM, disk, 144 parallel port IDE, disk, 144 parallel port IDE disk, 144 parallel port kernel module, 203 parameter expansion, 172, 173 parent, process, 91 parent process, 91

parent shell, 184 paritions, disk, 153 parity bit, 22 partition, 143 L INUX, 319 adding, 157 boot, 318 boot sectors, 318 boot up message, 159 bootable, 158 booting, 317 extended, 154, 321 formatting, 160 logical, 154 new disk, 155 primary, 154, 156, 318, 321 read-only, 168 swap, 327 type, 158 Windows, 143, 318 partition sizes, 536 partition table, 147, 153, 154, 321 partitioning example, 155 guideline, 155 RHCE, 544 partitioning session, 155 passwd, 9, 105, 106, 514, 535, 545 usage summaries, 33 password, 9, 104, 320 hash, 103 login, 9 prompt, 329 password, 320 Password Authentication Protocol, see PAP password cracking, 103 security, 524 password expiry dates, LPI, 541 password le, 102 password snifng, security, 517 paste, 532 X, 487, 497 .patch, 179 patch, 179, 180 patch le, 179 patch level, 235 patch number, 238, 240 patches, security, 482 PATH, 46, 91, 93, 107, 137, 172, 177, 179, 187, 236, 243, 407, 493 adding to, 46 path absolute, 34, 128 relative, 128 paths

full path name, 34 relative, 34 pause, 176 .pbm, 30 PC, 15, 38, 214, 318, 412, 425, 426, 429, 454, 455, 551, 565, 570 PC hardware, 15 .pcf, 30 PCI, 18, 462, 467, 473, 475 device ID, 474 RHCE, 544 vendor ID, 474 PCI graphics cards, 501 PCI slots, 18, 473 pci.ids, 474 pciscan, RHCE, 544 pcitable, 474 pcitweak, 494 pciutils, 473, 474 PCL, 200 PCMCIA, 544 PCPU, 108 .pcx, 30 PD, 556, 559 .pdf, 31 penetration testing, security, 524 Pentium, 552 performance, 487 periodic, commands, 409 peripheral, cards, 17, 18 peripherals devices, 15 Perl, 31, 414 permission bits, 125 permission denied by remote, 338, 340 permissions, 123, 126, 514 LPI, 533 perror, 217 personal computer, see PC personalized tuition, 526 .pfb, 31 pg ctl, 414 pg dump, 414 postgres, 423 pg dumpall, 414 postgres, 423 pg encoding, 414 pg id, 414 pg passwd, 414 pg upgrade, 414 pg version, 414 phone, 330 PHP, 31 .php, 31 PHP scripts, 407 PHP, with

609

610

Index

Apache, 406 physical disk, 154 physical drive, 153 physical memory, 37 pico, 59 PID, 81, 85, 172, 175, 177, 185, 190, 296, 325, 329, 519 PID, 89 PidFile, 394 pine, 99 ping, 180, 259, 260, 281, 458, 520, 540 -pipe, 239 pipeline, 83 pipes, 73, 86, 146, 181 pkzip, 32 .pl, 31 plain text, 131, 198, 201, 204, 223, 278, 422 security, 515 platform-independent library, 492 play, 40, 410 Plug-and-Play kernel, 470 PnP, 17 plus key, 494 .png, 332 PnP, 536 Plug-and-Play, 17 pnpdump, 545 point to point protocol, see PPP point-and-click-install, 1 point-to-point, Samba, 427 Pointer section, X, 499 pointer, 214 poll le, 346 POP, 295, 299, 301, 546 secure, 518 security, 301 POP host, 300 pop-3, 295 POP3, 270, 295, 337, 338 pop3, 518 popd, 175 Port, 394 port 1, 519 6, 519 20, 459 21, 294, 459, 519 22, 459, 519 23, 294 25, 99, 299, 300, 459, 519 53, 459, 519 67, 295 69, 295 79, 295

80, 265, 389, 459 110, 271, 295, 301, 459 113, 295, 459 119, 459 143, 295, 301, 459 400, 519 513, 294 514, 294 515, 519 517, 295 540, 295 901, 435 1024, 265 6000, 486 bind, 269 serial, 142 TCP, 292 port scan, security, 523 port speed, serial, 22 portable, 220 ported, 207 porting, 238 L INUX, 551 U NIX, 552 portmap, 285, 518, 547 ports, serial, 20 position-independent code, 234 positional parameters, 172 POSIX, 551 POST, 405 Post Ofce Protocol, see POP postfix, 301 postgres, 404, 405, 414 ALTER TABLE, 420 CREATE TABLE, 418 DROP TABLE, 420 INSERT INTO, 420 SELECT, 414, 417, 419, 421 bool, 418 init.d script, 415 oid, 421 pg dumpall, 423 pg dump, 423 psql, 417 adding a column, 420 attribute, 418 backup, 423 column typing, 418 creating tables, 418 database, 414 delete/dropping a column, 420 delete/dropping a table, 420 documentation, 414, 418 dumping and restoring tables, 422 European date formats, 416

HTML documentation, 415 initializing, 415 inserting rows, 420 installing, 415 internal tables, 417 listing a table, 419 listing databases, 417 locating rows, 421 migrating from another database, 422 object ID, 420 object relational, 420 package, 414 searches, 422 searching, 423 server program, 414 SQL commands, 414 start stop scripts, 415 subselects, 414 supported types, 418 template database, 416 transactions, 414 user commands, 414 user-dened types, 414 views, 417 postgres, with Windows, 414 PostGreSQL, 414 postgresql0dump, 414 postmaster, 310, 414 PostScript, 184, 200 man pages to, 134 X, 487 PostScript display engine, 493 PostScript emulator, 200 PostScript printer, 200 PowerPC, 240, 552 ppc, 240 PPID, 89, 92 PPP, 341, 536, 547 dialup, 338 ISDN, 453, 462 LPI, 540 masquerading, 459 network, 453 protocol, 454 RHCE, 548 PPP protocol, 461 ppp0 device, 462 pppd, 257, 258, 454, 461, 548 chat, 454 chat script, 455 defaultroute, 455 command-line, 454 dial-in, 460 dial-on-demand, 458

dialout, 454 pr, 532 prefdm, 546 --prefix, 238, 239, 393 preprocessor, C, 222 preprocessor directive, 222 present working directory, 176 pwd, 13 prevention, spam, 312 previous working directory, 173 PRI, 89 primary, partition, 154, 156, 318, 321 primary master, 143, 155, 160 primary slave, 143 print, 224 lter, 200 header, 199 Samba, 425 print lter, 200, 203, 204, 333, 336 print preview, 201, 205 print queue, default, 199 print share, Windows, 205 print tray, 193 printcap, 196, 199, 204, 205 #printenv, 401 PRINTER, 199 printer, 527 access control, 202 queue, 193 supported devices, 201 printer cable, 203 printer capabilities, 199 printer conguration, 201 printer control language, 200 printer device, 199 printer drivers, 201 printer management, 193 printer native code, 200 printer spool directory, 198 printers, serial, 21 printf, 176, 209, 211, 214 printf(3), 176, 209 printing, 193 printing troubleshooting, 203 printtool, 196, 204, 545 priority CPU, 87, 296 process, 87 Privilege Attribute Certicate, see PAC /proc, 156, 166, 255, 545 /proc le system, 178 proc, 163 proc le system, 167 /proc/, 466 /proc/cmdline, 317

611

612

Index

/proc/devices, 467 /proc/dma, 466 /proc/interrupts, 480 /proc/ioports, 466 /proc/kcore, 146 /proc/pci, 474 /proc/tty/driver/serial, 480 process, 81 child, 91 environment, 90 niceness, 87 parent, 91 priority, 87 scheduling, 87 process ID, see PID, 394, 519 process limits, LPI, 542 process list, 185 process tree, 185 processes, NFS, 286 processes, 89 processing, command-line arguments, 68 processor, 208 processor instructions, 513 .procmail, 548 procmail, 190 RHCE, 548 procmail support, exim, 308 procmail(1), 308, 548 procmailex(5), 308, 312, 548 procmailrc(6), 308, 548 .profile, 186, 539 program, lter, 201 program priority, LPI, 532 program source code, 29 programing, X, 498 programming computer, 3, 61 X, 498 programming language, 207 programs, X, 494 project, GNU, 551, 557 projects, C, 230 promiscuous, 262 prompt, 223 LILO:, 329, 467, 478 login, 9, 330 password, 329 shell, 9 prompt, 320 PROMPT COMMAND, 94 proprietary protocols, 200 protocol, 487 PPP, 454 X, 486 protocol g, uucp, 341

protocol responses, X, 486 protocol t, uucp, 340, 341 proxyarp, 461 proxymngr, 494 .ps, 31, 200 ps, 82, 185, 226, 532 ps variations, 185 PS/2, 544 PS/2 mouse, 144 PS1, 94 PS2, 94 PS3, 94 PS4, 94 pseudo-TTY, 145 PseudoColor, X, 505 pseudonym, 175 psql, 414, 423 postgres, 417 psutils, 204 PTR record, DNS, 283, 441443, 446 PTR records, 548 public domain, see PD pump, 546, 548 pushd, 175 pwconv(8), 517 PWD, 92, 172 pwd, 34, 176 present working directory, 13 .py, 31 Python, 31, 134, 414

quotas LPI, 533 RHCE, 545 quotastats, 545 quotawarn, 545 quote notation, 70 quotes, 209 backward, 71 double, 70 forward, 70 shell, 70

R
r, 224 R3000, 552 R4000, 552 -r600x600, 201 radio0, 146 RAID, RHCE, 546 RAID devices, RHCE, 544 RAM, 15, 37, 81, 156, 214, 329, 464, 468 RAM DAC, X, 501 RAM device, 167 RAM devices, 168 ram0, 146 Ramey, Chet, xxxi Ramey, Chet, 83, 92, 172 RANDOM, 92, 172 random access devices, 142 random bytes, 184 random key, 203 ranlib, 229 RAWRITE.EXE, 321 RBAC, security, 522 RBL, spam, 313 rbl domains, 313 rbl reject recipients, 313 rc3.d, 197 rcp, 112, 337 RCPT, 100 MAIL, 313 RD pin, 21 rdev, 321 re-mail-ck, 269 re-request, ARP, 251 read, 62 read lock, 188 read only, mounting, 166 read permission, 124 read-only le system, 168 partition, 168 read-only, 321 read-only partitions, security, 520 read-write test, disk, 161

Q
Q key, 38 qft0, 146 qmail, 301 qnx4, 163 qotd, 269 Qt, 59, 492, 497 quad density, 145 qualied host name, 273 QUERY STRING, 402, 405 querying, package, 241 queue mail, 299 printer, 193 queue deletions, mail, 309 queuing, mail, 302 quick-quiz, security, 523 QuickStart guide, X, 498 quiet mode, 201 Quinlan, Daniel, 347 quit, 86 quota, 533, 545 quotaoff, 545 quotaon, 533, 545

readability, 220 reading les, 217 reading resources, security, 523 README, 32, 237 real time OS, 554 real time scheduling, 87 real UID, 109 real-time blocking list, see RBL realloc, 218 reboot, 11, 326 rec, 40 receive pin, 21 receiving, faxes, 325, 330, 331 recipient, 100 recipient address, 300 recipients reject except, 313 recompiled, 207 record, DNS, 283 recording volume, 141 recursive, 42, 125, 179 copy, 112 recursive call, 181 recursive searches, 180 recursively copying, 34 removing, 34 recv, 264 recv(2), 264 RedHat, 1, 135, 156, 554, 561 RedHat Package Manager, 135 RedHat-like, 1 .rpm, 3 redirecting, streams, 75 redirecting standard input/output, LPI, 532 redirecting the terminal, 184 redirection operator, 76 reference Apache, 393 documentation, 2 Regents of the University of California, see BSD regular expression examples, 50 regular expressions, 49, 75, 185 extended, 52 LPI, 532 subexpressions, 52, 78 reiserfs, le system, 160 rejecting, header, 313 relational algebra, 423 relative path, 128 paths, 34 relay, 100 mail, 304 untrusted hosts, 304 relay domains, 304, 314

613

614

Index

release, 240 reliable communication, 263 reliable stream, 263 remote access, 113 X, 486 remote attack, security, 512 remote copy, see rcp, 112, 337 remote displays, RHCE, 547 remote login, 111 remote machine, 112 remote print queue, 199 Remote Procedure Call, 286 remote shell, see rsh, 112, 294, 337 remote smtp, 305, 306 remote smtp:, 305 remounting, 168 removable, IDE drives, 161 removing directories, 34 les, 34 package, 241 recursively, 34 removing messages, mail, 309 renice, 87 reorder, 279 repartition, 147 repeat rate, 38 replacing, 49 REPLY, 92 repquota, 533, 545 Request For Comments, 121 request to send, see RTS requirements LPI, 3 RHCE, 3 rescue disks, 178, 554 rescue oppy, 147 RHCE, 547 Resident Set Size, 88 resize, 494 resizing PostScript, 204 resolution, 491 name, 276 resolv.conf, 445 resolver, 276, 280 resolver(5), 280 resolving host names, 274 resources, L INUX, 117, 527 respawn, 327, 329 respawning too fast error, 328 responsibilities administrator, 313 spam, 313 restarting, L INUX, 11

restrict access, 297 restricted, 320 resume, 83 retransmission, 337, 338 retries, mail, 302 retry schedule, mail, 300 return code, 209 Return Key, 9 return value, 217 return path add, 305 reverse lookup, DNS, 286, 445 reverse lookups, 203, 445, 528 DNS, 281, 442, 447 revpath, 494 rewinding, tape, 149 RFC, 121, 438 RGB value, X, 487 RHCE hdparm, 544 isapnp, 544 lpd, 546 pciscan, 544 procmail, 548 sendmail, 548 squid, 546 AGP, 544 Apache, 548 BIOS conguration, 544 BIOS limitations, 544 CD-ROM, 544 certication, 2 CGI, 548 conguring and administration, 544 cryptography, 549 DHCP, 548 DNS, 548 domain servers, 548 EISA, 544 rewalling, 547, 549 oppy devices, 544 forwarding, 547 FTP, 546 Gnome, 546 hardware selection, 544 high availability, 547 high performance, 547 I/O ports, 544 IDE, 544 IDE geometry, 544 installation methods, 545 IRQ, 544 KDE, 546 kernel, 545 kernel source, 546 mail spooling, 548

MIME, 548 module, 546 networking, 546 NFS, 546, 549 NIS, 547, 548 PAM, 547, 549 parallel port, 544 partitioning, 544 PCI, 544 PPP, 548 quotas, 545 RAID, 546 RAID devices, 544 remote displays, 547 requirements, 3 rescue oppy, 547 round-robin load-sharing, 548 routing, 547 run levels, 546 Samba, 546, 548 SCSI, 544 security, 547, 549 security tools, 550 serial port, 544 SMP support, 544 TCP/IP, 546 troubleshooting, 547 USB, 544 X, 546 X client server, 546 RI pin, 23 ribbon IDE, 18 SCSI, 19, 477 Richard Stallman, 557 right arrow key, 10, 55 ring indicator, 23 RiscOS, 58 risks, security, 511 rje, 269 rlogin, 111, 294, 515, 518 rlogind, 294 rlp, 269 rm, 34, 175, 184, 200 rmail, 301, 338, 343 rmmod, 537, 546 kernel, 464 rmmod -a, kernel, 465, 471 Role-Based Access Controls, see RBAC ROM, 15 BIOS, 318 romfs, 163 /root, 101 root, 99, 101, 137, 141, 194, 269, 321 root login, 106

root device, 146 root directory, 136, 178 root le system, 329 root mounts, NFS, 481 root name servers, 276 root window, X, 494 rotate, 280 rotation, log les, 198 round-robin load-sharing, 448 RHCE, 548 route, 252, 253, 279 route, 252, 540 route -n, 457 route list, exim, 344 routers, exim, 306 routing, 254 mail, 299, 306 RHCE, 547 routing table, 252, 253, 258, 259 rp, 200 rp3, 544 RPC, 286 rpc.lockd, 286 rpc.mountd, 286 rpc.nfsd, 286 rpc.portmap, 285 RPM, 135, 562 .rpm, 3, 31, 135, 237, 241, 301, 546, 562 Mandrake, 3 RedHat-like, 3 rpm, 194, 240, 242, 243, 537, 545 RPM-HOWTO, 246 rpmfind, 545 RS-232, 23 RS-422, 23 RS-423, 23 RS6000, 208, 552 RSBAC, security, 522 rsh, 112, 294, 337 rsh(8), 113 rshd, 202, 294 RSS, 88, 90 rstart, 494 rstartd, 494 rtin, 120 RTS, 22 RTS/CTS ow control, 22 Rubin, Ryan, 511 Rule-Set-Based Access Controls, see RSBAC rules, 231 run, 224 run level, 325, 326, 329 run level 0, 328 run level 3, 197 run level 6, 328

615

616

Index

run levels, RHCE, 546 run-parts, 411 running Windows programs on, L INUX, 571 runq, 308 runtime, 203, 234, 406, 537 Russell, Rusty, 347 rxvt, 184, 539

S
s command, 225 S60lpd, 197 SAINT, 550 Samba, 528 lp print queue, 432 mount, 432 nmbd, 427, 428 nmblookup, 428 smbclient, 428, 432 smbd, 427, 431 swat, 428, 429 broadcast, 427 conguring, 431 conguring printers, 434 conguring windows, 433 DMB, 428 documentation, 435 le, 425 interoperability, 425 IP address, 428 LMB, 428 mailing lists, 430 minimal cong le, 431 name resolution, 427 NBNS, 428 NetBEUI, 426 NetBIOS, 426, 427, 430 point-to-point, 427 print, 425 RHCE, 546, 548 share, 427, 432 SMB, 425, 426 web administration, 428, 434 Windows NT/2000, 435 Samba 2.0, 429 SARA, 550 sash, 323 SATAN, 550 /sbin, 156 sbin, 137, 196 /sbin/getty, 326 /sbin/init, 318, 325, 329 /sbin/insmod, 323 /sbin/ipfwadm-wrapper, 458 /sbin/mgetty, 326, 330 /sbin/update, 327

scalable fonts, X, 509 scanner, 146 scanners, 145, 544 scanning, 184 scanpci, 473, 474, 494 scheduling uucp, 346 commands, 409 jobs, 409 process, 87 scheduling priority, 87 scientic simulations, 552 SCO, 551 scope, 1 scp, 112, 518 Screen section, X, 503, 504 script kiddies, 511 security, 516 scripting, 61 scripting language, Tcl, 492 scripts, 178 SCSI, 18, 322, 464 adapter, 478 bus, 476 card, 476 CD writer, 478 CD-ROM, 19, 145, 146 cooling, 477 device, 143, 146 disk, 144 disk drive, 19 drive, 476, 536 emulation, 478 generic device, 145 ID, 476 kernel, 475, 476 LUN, 476 module, 322, 323 RHCE, 544 ribbon, 19, 477 security, 520 tape, 144, 149 termination, 19, 477 SCSI BIOS, 536 LPI, 536 SCSI CD-ROM, 144 SCSI controllers, 132 SCSI drives, 155, 156 SCSI ID, 536 SCSI partition, 323 sd, 199 -sDEVICE, 201 search, 49 search, 280 search and replace, sed, 77

search list, 280 search pattern, 49 search through directories, 181 searches, postgres, 422 searching, 413 postgres, 423 for les, 42 web site, 118 within les, 43 searching and replacing, 49, 181, 183 second X server, X, 495 secondary Ethernet device, 254 secondary master, 143 secondary server, DNS, 450 secondary slave, 143 SECONDS, 92 seconds, 39 sections, man, 35 sector, 143, 318, 320 disk, 153 sector list, 318, 329 sector map, 318 secure IMAP, 518 POP, 518 secure erasure, 184 secure shell, see ssh, 271 secure-mcserv, 272 security, 9, 26, 46, 81, 111114, 117, 120, 150, 202, 244, 294, 296, 301, 304, 320, 394, 396, 397, 403, 405, 406, 412, 438, 455, 459461, 510, 511, 535, 572 /tmp le vulnerability, 514 inetd, 297 libsafe, 518 medusa, 522 sendmail, 315 ACL, 522 alerts, 202, 517 announcements, 516 append-only permissions, 520 attacks, 511 auditing, 524 authentication logic, 514 buffer overow, 518, 521 buffer overow attack, 512, 513 capabilities, 521 client programs, 518 compromise, 512 counter measures, 516 detection software, 524 distributions, 516 DNS, 437 DoS, 515

178, 340, 430, 554,

DTE, 522 encryption, 515, 518 environment variables, 515 exploit, 513, 515, 516 FAQ, 523 rewall, 515 hole, 514 IMAP, 301 immutable permissions, 520 Internet, 511 ISO 9660 partition, 520 loopholes, 516 LPI, 541, 542 MAC, 522 multiuser systems, 512 network attacks, 511 network client programs, 514 network intrusion detection, 522 NFS, 289, 518 opportunity for attack, 519 outdated packages, 516 password cracking, 524 password snifng, 517 patches, 482 penetration testing, 524 plain text, 515 POP, 301 port scan, 523 quick-quiz, 523 RBAC, 522 read-only partitions, 520 reading resources, 523 remote attack, 512 RHCE, 547, 549 risks, 511 RSBAC, 522 script kiddies, 516 SCSI, 520 security hole, 516 setuid, 519 spoof, 523 system monitoring, 520 tripwire, 523 Trojans, 517 untrusted user logins, 514 upgrading, 516 vulnerable packages, 516 VXE, 522 world-writable les, 517 Security Focus web site, 516 security hole, security, 516 security tools, RHCE, 550 sed, 57, 77, 532 addresses, 79 deleting lines, 79

617

618

Index

editing, 77 inserting lines, 79 search and replace, 77 segfault, 227 segment, 261 segmentation violation, 86, 227 SELECT, postgres, 414, 417, 419, 421 selecting text, X, 497 selections, X, 497 selector, package, 245 send, 264 send(2), 264 sendfax, 333, 336, 342 sendfax.config, 333 sendmail, FAQ, 315 sendmail, 193, 197, 198, 241, 299, 301, 315, 541, 546 RHCE, 548 security, 315 sendmail.cf, 301 sequence number, TCP, 265 serial data rate, 22 devices, 20 ow control, 22 port, 142 port speed, 22 ports, 20 printers, 21 serial devices, 142, 144, 330 serial line, 247, 256 serial number, DNS, 445 serial port, 18, 142, 328, 342, 480, 536 RHCE, 544 server, 194 X, 485, 506 Server Message Block, see SMB server program, postgres, 414 Server-side includes, Apache, 400 server-side includes, see SSI ServerAdmin, 395 ServerRoot, 394 servers DNS, 276 Internet, 552 ServerSignature, 395 ServerType, 394 service, TCP, 338 service descriptions, TCP, 294 Services list, 197 services, 193, 291 inetd, 294 services, 269 sessreg, 494 set, 172, 176, 177

set args, 224 setgid, 125 setgid binary, 335 setgid group bit, 336 setgid wrapper, 335 setserial, 480 setuid, 124, 125, 513 security, 519 setuid binary, 335 setup, 197, 544 setxkbmap, 494 SGI, 429 .sgml, 31 .sh, 31 sh, 199 shadow, 535 shadow password le, 102, 103 shadowed passwords, LPI, 541 SHARE, 88, 90 share, Samba, 427, 432 share, 131, 134 share/, 138 -shared, 234 shared data, 138 shared libraries, 233 shared locks, 188 shared object les, 233 Shareware, 556 sharing of CPU, 82 SHELL, 107 shell, 103 beeping, 11 prompt, 9 quotes, 70 shell, 294 shell commands, LPI, 531 shell environment, 176 shell initialization, 172, 186 shell script, 30, 108, 176, 201, 208, 257, 389 shell scripting, 61, 171, 211 Shift key, 10, 38, 54 SHLVL, 92 short, 215 showfont, 494, 507 showpage, 200 showrgb, 494 shutdown, 11 shutdown, 264, 328, 534 sides, disk, 153 SIGCHLD, 86 SIGFPE, 86 SIGHUP, 86, 198, 328 SIGINT, 83, 86 SIGKILL, 85, 86 signal, 85, 176

signal(7), 86 signicant digits, 210 SIGPIPE, 86 SIGQUIT, 86 SIGSEGV, 86, 227 SIGTERM, 85, 86 SIGTTIN, 84 SIGTTOU, 84 SIGUSR1, 86 Silicon Graphics, see SGI, 551, 552 SIMAP, 518 simple lesystem problems, LPI, 533 Simple Mail Transfer Protocol, see SMTP simple program, C, 208 simultaneous connections, 296 single-user mode, 325, 328 SIZE, 88, 89 size, 182 sizeof, 214, 215 Slackware, 31, 135, 554, 561 slapd, 547, 548 SlashDot, 118, 121 slave IDE, 19 slave server, DNS, 450 slave setting, DNS, 451 sleep, 39 SLIP, 536 slots, ISA, 17 slurpd, 547, 548 Small Computer System Interface, see SCSI SMB, 285 Samba, 425, 426 smb, 541 SMB vs. NFS, Windows, 285 smb.conf, 429 smbclient, 205, 546 Samba, 428, 432 smbd, 546 Samba, 427, 431 smbfs, 163, 429, 432 smbmount, 546 smbsh, 429 SMP, kernel, 483 SMP support, RHCE, 544 smproxy, 494 SMTP, 99, 302, 337, 338 smtp, 269 SMTP gateway, 300 SMTP host, 300 SMTP protocol, 300 sndcong, 536 snice, 88 sniffit, 550 .so, 31, 233, 235, 406 SOA record, 548

DNS, 284, 441443, 445, 446 socket connection, X, 486 sockets, 146 soft link, 127 software copying, 574 distribute, 574 GNU, 557, 558 modifying, 574 software cost, GNU, 562 software ow control, 22 software package, 135 Solaris, 47, 551 sort, 39, 75, 532 sort(1), 39 sorting, 75, 413 uniquely, 75 sortlist, 280 Sound, 464 sound ALSA, 475 kernel, 469 Open Sound project, 474 sound card, 40, 141, 536 sound cards, 132, 480 kernel, 470 SoundBlaster, 470 source C, 237, 238 kernel, 463 package, 246 source, 176 source address, IP, 248, 264 source code, 37 GPL, 576 kernel, 463 source package, 239 source port TCP, 265 UDP, 269 source tree, 181 SourceForge, 119 sources.list(5), 245 South Africa, 2 space bar, 38 spam combating, 311 denition, 311 Internet, 99 prevention, 312 RBL, 313 responsibilities, 313 spam mail, 99 -sPAPERSIZE, 201 SPARC, 208, 240, 551, 552

619

620

Index

sparc, 240 .spd, 31 speakers, 141 spec.txt, exim, 309 special parameters, 172 special-purposes addresses, IP, 249 speed, modem, 454 Speed font, 31 speeding compilation, 239 spell checking, 75 split, 39, 532 spoof, security, 523 spoofalert, 279 spoong, 279 spool directory, 198 spool les, 197, 199, 203, 300 spooling, 197, 299 spooling mail server, 302 SPOP3, 518 sprintf, 513 SQL, 413 SQL commands, postgres, 414 SQL database, 403 SQL programming language, 413 SQL requests, 413 SQL server, 413 SQL, with Apache, 403 SQL92 standard, 414 squid, 546 RHCE, 546 sr?, 146 .src.rpm les, 545 src/, 138 srm.conf, 393 ssh, 111, 112, 269, 271, 518, 528, 541, 547, 549 SSH Protocol 2, 271 sshd, 202, 271 SSI, Apache, 400 stable, 118 stack, 225 stale, 190 stale lock le, 343 standalone, 394 standard C, C, 209, 216 standard functions, 228 standards compliant, 238 start bit, 22 Start button, X, 496 start of authority, see SOA start stop scripts, postgres, 415 starting bind, DNS, 443 startup script, 187 startup scripts, 253, 254, 465 startx, 494, 505

startx(1), 505, 546 STAT, 90 stat, 126, 190 statement, 209 statement blocks, 212 statements, 211 static, library, 29 static, 229 static linking, 230 statically link, 406 StaticColor, X, 505 StaticGray, X, 505 status, package, 242 stderr, 113 stderr pipe, 73 stdin, 113, 201, 337 stdout, 113 stdout pipe, 73 step, 225 sticky bit, 124, 125 stop, application or command, 41 stop bits, 22 strace, 227 strcpy, 216, 220, 513 stream, 73 TCP, 292 streams LPI, 532 redirecting, 75 string, 209, 213 string(3), 217 string manipulation, 173 string operations, 215, 217 string substition, 173 strings, 39, 75 strip, 323 strlen, 216 strncpy, 216, 217 Structured Query Language, see SQL students, 525 stuff, 17, 39, 78, 118, 156, 165, 425, 427 style.txt, DNS, 438 su, 107, 415, 513 subdirectory, 137 subexpression, 78 subexpressions, regular expressions, 52, 78 subject, 97 subnet, 249, 256 subscribe, 119 subselects, postgres, 414 substition, 173 sudo, 549 SUID bit, LPI, 541 suid-rights, LPI, 539 Sun Microsystems, 240, 286, 493, 551, 552

SunOS, 551 sunsite.unc.edu, 117 superformat, 162 SuperProbe, 494, 501 superstructure, 136, 137 superuser, 101 Superuser binary executables, 137 support, L INUX, 563 support daemons, kernel, 463 supported devices, printer, 201 supported types, postgres, 418 SuSE, 554, 562 suspend, 83 SWAP, 90 Swap, 89 swap, 156 adding, 162 partition, 327 swap device, 146 swap partition, 158 swap space, 37, 536 swapoff, 163 swapon, 162 swat, 434, 548 Samba, 428, 429 swatch, 545, 549 swatch(5), 549 swatch(8), 549 SWIG, 30 switch, 212 switchdesk, 546 symbol referencing, 230 symbolic link, 128, 129, 533 symlink, 137, 144, 145, 197 DLL, 235 SymLinksIfOwnerMatch, 396 Symmetric Multiprocessor Support, see SMPT SYN, TCP, 264 SYN-ACK, TCP, 264 sync, 169 syntax checking, 178 syntax highlighting, 57 vi, 57 sysctl.conf(5), 545 syslog, 107, 146, 296 syslogd, 545 systat, 269 system administrator, 101 system call, 227 system load average, 108 system log le, 198 system monitoring, security, 520 System Services, 197 system times, 176 System.map, kernel, 464, 484

system aliases:, 310 sysv, 163 Sytec, 425

T
Tab, beeping, 11 tab character, 218, 231 Tab completion, 11 Tab key, 11, 317 table, 321 tac, 532 tag, package, 244 tail, 39, 198, 392, 532 tail -f, 330 talk, 295 talkd, 294 tape backups, 149 rewinding, 149 SCSI, 144, 149 tape device, 146 tape drives, 18 tapes, 544 .tar, 31, 45, 240 tar, 39, 45, 115, 229, 239, 245, 537 backup, 45 verify, 45 with gzip, 46 tar les, 237 .tar.gz, 393 TARGA, 31 targets, 231 Taylor UUCP, 337 Tcl, 414 scripting language, 492 .tcl, 31 Tcl/Tk, 31 TCP, 260, 263, 519 inetd, 291 ACK, 264 acknowledgment number, 265 acknowledgment packet, 263 checksum, 265 connection, 296, 300 control, 265 destination port, 265 detachment, 263, 268 encryption, 270 example session, 265 FIN, 264 forwarding, 270 header, 264 LPI, 540 minimum header, 265 negotiation, 263, 268

621

622

Index

options, 265 port, 292 sequence number, 265 service, 338 service descriptions, 294 source port, 265 stream, 292 SYN, 264 SYN-ACK, 264 transfer, 263 urgent pointer, 265 window, 265 TCP connection, 510 TCP wrappers, 293 LPI, 540, 541 TCP/IP, 263, 413, 426 RHCE, 546 X, 486 tcp wrappers, 547, 549 tcpdchk, 547 tcpdmatch, 547 tcpdump, 261, 262, 268, 462, 515, 528, 550 tcpdump(8), 268 tcpmux, 269 TD pin, 21 tee, 532 telephone assistance, LPI, 535 telephone lines, 23, 337 telephone support, 535 telinit, 328 telnet, 99, 111, 251, 259, 265, 269, 272, 294, 299, 389, 392, 513, 515, 518, 528, 540 telnetd, 294 template database, postgres, 416 temporary data, 156 temporary le, 138, 188 TERM, 107 termcap, 199 terminal, 184, 209 login, 330 mail readers, 99 terminal capabilities, 199 terminal device, 145 terminal messages, 198 terminal process group, 83 terminal redirection, 184 terminal-based mail readers, 99 terminals, 21 terminate, 86 terminating commands, 41 termination, SCSI, 19, 477 terms and conditions, GPL, 574 test, LPI, 539 test, 62, 179 test(1), 63

test page, 203 test print, 201 test.cgi, 404 testing interface, 260 IP, 260 mail, 100 testing the serial port, 142 testparam, 546 TEX, 31, 132 .tex, 31 .texi, 31 .texinfo, 31 Texinfo source, 31 text, 115 text editing, 53 text editor, 185 text input boxes, 491 text/plain, 395 .tfm, 31 TFTP, 295 tftp, 269, 295 tftpd, 294 .tga, 31 .tgz, 31, 135 THANKS, 32 three-button mouse, X, 497 .tiff, 31 tilde expansion, 173 TIME, 90 time, 182 time, 269 time to live, IP, 248, 264 time to tive, see TTL time-out, ARP, 251 timeconfig, 544 timeout, 320 times, 176 Tk, 492 Tk widgets, Windows, 492 tkrat, 539 tksysv, 546 TLD, 273 TMOUT, 94 /tmp, 137, 138, 156, 188, 203, 517 /tmp le vulnerability, security, 514 tmpwatch, 545, 549 TODO, 32 Token Ring, 426 top, 88, 532, 549 top-level directory, Apache, 395 top-level domains, see TLD TOS, IP, 247, 264 touch, 125 tr, 75, 532

tr(1), 75 trace, 225 traceroute, 261, 540, 550 tracks, disk, 153 trafshow, 550 training, 2 course, 2 transactions, postgres, 414 transfer, TCP, 263 transfer type, 113 transferring les, 111 translate, 75 Transmission Control Protocol, see TCP transmit pin, 21 transports, exim, 304, 343 trap, 176 trapping signals, 176 trim, 279 tripwire, security, 523 tripwire, 549 trivial FTP, see TFTP Trojans, security, 517 Troll Tech, 492 troubleshooting, RHCE, 547 troubleshooting guides, 196 TRS, 90 TrueColor, X, 504, 505 TrueType font, 31 X, 509 truncated, 74, 198 TSIZE, 90 .ttf, 31, 509 TTL DNS, 445 IP, 248, 264 ttmkfdir, X, 509 tty, 106, 108 TTY devices, 145, 146 tty name, 108 TurboLinux, 554, 562 tutorial, documentation, 2 twist, 547 twm, 494 .txt, 31 type, partition, 158 -type option, find, 43 type, 176 type of service, IP, 247, 264 typesetting, 132

U
UART, 480 16450, 479 16550A, 479 16550, 23, 479

16650V2, 479 16650, 479 16750, 479 8250, 479 compatibility, 480 UDP, 260, 261, 268 destination port, 269 length, 269 LPI, 540 source port, 269 ufs, 163 UID, 103, 105107, 513, 514 UID, 89, 92 ulimit, 176 UltraSPARC, 552 umask, 125, 176, 533 umount, 164, 167 umsdos, 163 unalias, 175 uname, 39 unbootable, 484 unclean umounts, 165 uncompress, 318 uninstall, 297 uninstalling, package, 240 uniq, 39 uniquely, sorting, 75 Universal Asynchronous Receiver-Transmitter, see UART Universal Serial Bus, see USB universities, 274, 552 U NIX administration, 6 case sensitive, 25 commands, 10, 25 compatibility, 561 documentation, 35 editors, 57 evolution, 10 le name extensions, 29 le name, allowable, 7 le names, 12 porting, 552 U NIX compatible, 238 U NIX, with Windows, 425 U NIX, 58 U NIX initialization, 325 U NIX, 5 U NIX debugger, 223 U NIX directories, 135 Unix-to-Unix copy, see UUCP Unix-to-Unix copy in copy out, see uucico unpack, 135 unpacking, kernel, 481

623

624

Index

unqualied, host name, 273 unreliable connections, 337 unset, 92, 176 unsolicited, 311 unsolicited email, 99 unsubscribe, 120 until, 63 untrusted hosts, relay, 304 untrusted user logins, security, 514 up arrow key, 55 update binaries, LPI, 541 update-rc.d, 197 updatedb, 43, 533 upgrading package, 240 security, 516 upload, 114 uptime, 89 urgent pointer, TCP, 265 URL, 39, 406 usage less, 38 CPU, 87 usage limits, 176 usage summaries cp, 33 ls, 33 passwd, 33 USB, 18 RHCE, 544 use of C, Windows, 207 UseCanonicalName, 395 USER, 89, 107 User, 394 user, 123 user commands, postgres, 414 User Datagram Protocol, see UDP user directories, Apache, 398 user name, characters, 102 user programs, 196 user signal, 86 user statistics, 108 user-dened types, postgres, 414 user forward, 308 useradd, 106, 171, 535, 545 userconf, 545 userdel, 535 userdel, 106, 535, 545 UserDir, 398 usermod, 106, 545 usernet, 544 usernetctl, 544 users, L INUX, 117, 118 users, 108 usleep, 40

/usr, 136, 137, 155, 156, 520 /usr/bin/mailq, 301 /usr/bin/newaliases, 301 /usr/bin/rmail, 301 /usr/bin/X11/xdm, 327 /usr/doc/, 534 /usr/include/asm/errno.h, 26 /usr/lib, 230 /usr/lib/, 233 /usr/lib/sendmail, 301 /usr/local, 136 /usr/local/, 137, 156 /usr/local/bin, 179 /usr/local/bin/, 181 /usr/sbin/exim, 302 /usr/sbin/inetd, 291 /usr/sbin/lpd, 194 /usr/sbin/named-bootconf, DNS, 438 /usr/sbin/sendmail, 301 /usr/sbin/up2date, 545 /usr/sbin/uucico, 294 /usr/share/doc, 195 /usr/share/doc/xserver-common/, 498 /usr/share/kudzu/pcitable, 474 /usr/share/pci.ids, 474 /usr/src/linux/arch/i386/defconfig, 483 /usr/src/linux/Documentation/, 143, 482, 483 /usr/src/linux/System.map, 484 /usr/tmp, 137 /usr/X11/bin/X, 486 /usr/X11R6, 137 /usr/X11R6/, 137, 493 /usr/X11R6/lib/libX11.so.6, 487 /usr/X11R6/lib/X11/doc/, 498 /usr/X11R6/lib/X11/fonts/, 508 util-linux, 162 utilities, X, 494 utility scripts, 178 utmp, 327 uucico, 180, 294, 339341, 346 uucp, 197, 295, 337, 461, 518 exim, with, 343 mgetty, with, 341 protocol g, 341 protocol t, 340, 341 authentication, 339 conguration, 338 debugging, 343 debugging level, 343 dialup, 338 documentation, 338 mail, 343 modem, 341

scheduling, 346 uucp user, 335, 336 UUCP lock le, 342, 343, 454 uudecode, 115 uuencode, 114 uux, 337, 342

V
V, 492 V.32, 24 V.34, 24 V.42, 24 V.90, 24 vacuumdb, 414 /var, 137, 155, 156, 190, 203 var/, 137 /var/lock/, 335, 342 /var/lock/LCK..tty??, 454 /var/log, 198 /var/log/, 198 /var/log/fax, 334 /var/log/lastlog, 107 /var/log/maillog, 198 /var/log/messages, 106, 279, 287, 444, 471, 520, 534 /var/log/mgetty.log.ttyS?, 330 /var/log/samba/, 432 /var/log/syslog, 198, 444, 471, 520 /var/log/uucp/Debug, 343 /var/log/uucp/Log, 341, 343 /var/log/uucp/Stats, 343 /var/named, 438 /var/named/named.127.0.0.1, 441 /var/named/named.ca, 440 /var/named/named.localdomain, 440 /var/run/httpd.pid, 394 /var/spool/, 197 /var/spool/exim/input/, 299, 309 /var/spool/fax/incoming/, 331 /var/spool/lpd, 198 /var/spool/mqueue/, 299 /var/spool/uucp/, 343 /var/spool/uucppublic/, 340 variable, 62, 216 variable data, 137, 156 VAT32, 164 vendor ID, PCI, 474 verication, header, 313 verify, tar, 45 VERSION, 32 version 6, IP, 248, 280 version number, 238, 240 GPL, 578 versioning DLL, 234

kernel, 464 package, 238, 240 Vertical sync, X, 501 VFAT, 162 vfat, 163, 164, 287 vga, 320 vi, 53, 538 basic editing operations, 54 editors, 58 modes, 54 syntax highlighting, 57 vi tutorial, 53 video, 115 video buffer, X, 504 video card, LPI, 539 video conferencing, 462 Video format, 29 video memory pixmap caching, X, 503 video modes, 132 video0, 146 viewres, 494 views, postgres, 417 vim, 53, 58 editors, 58 virtual console, 145, 494, 495 virtual consoles, 11 L INUX, 329 virtual domains, 389, 392 Virtual eXecuting Environment, see VXE virtual hosting, Apache, 407 virtual memory, 186 virtual terminals, 11 VirtualHost, 408 viruses, 571 -visual, X, 505 visuals, X, 504 VMS, 58, 425 .voc, 31 void, 211 vulnerable packages, security, 516 VXE, security, 522

W
w, 108 W2K, 430 wait, 176 -Wall, 208 warning messages, 208, 230 WAV, 410 .wav, 31 wc, 40, 532 WCHAN, 90 web, 193, 291 web administration, Samba, 428, 434 web browser, 39, 118, 196, 274

625

626

Index

web cache, 156 web documentation, 133 web page, 30, 195 LPI, 531 web pages, 389 web search, 118 web server, 133, 264, 389 web site LPI, 534 searching, 118 X, 498 Web sites ftp://ftp.kernel.org/pub/linux/kernel/, 481 ftp://ftp.rs.internic.net/domain/named.root, 276 ftp://metalab.unc.edu/pub/Linux/distributions/, 562 ftp://metalab.unc.edu/pub/Linux/LICENSES, 556 ftp://metalab.unc.edu/pub/docs/rfc/, 121 http://www.memeticcandiru.com/colsfaq.html, 523 http://cericon:901/, 435 http://cooledit.sourceforge.net/, 58 http://download.lycos.com/static/advanced search.asp, 119 http://freshmeat.net/, 118, 245, 510, 553, 554 http://gnustep.org/, 554 http://icewm.cjb.net/, 495 http://infoseek.go.com/, 118 http://jade.cs.uct.ac.za/idsa/, 523 http://linmodems.technion.ac.il/, 481 http://linux.tucows.com/, 553 http://localhost/, 392, 394 http://localhost/jack/index.html, 398 http://localhost/hello.php, 407 http://localhost/manual/howto/ssi.html, 401 http://localhost/manual/index.html.en, 395 http://localhost/test.cgi, 402 http://localhost/test/test.cgi?xxx=2&yyy=3, 403 http://lwn.net/, 118, 516, 553, 554 http://redhat.com/training/rhce/courses/, 543 http://rute.sourceforge.net/morecram1.3.tar.gz, 139 http://SAL.KachinaTech.COM/index.shtml, 554 http://samba.org/, 425, 429 http://slashdot.org/, 118, 121

http://users.bart.nl/patrickr/hardwarehowto/Hardware-HOWTO.html, 570 http://www.altavista.com/, 118 http://www.idir.net/gromitkc/winmodem.html, 481 http://www.linmodems.org/, 481 http://www.sourceforge.net/, 119 http://www.yahoo.com/, 118 http://www.alsa-project.org/, 475 http://www.apache.org, 393 http://www.astart.com/lprng/LPRng.html, 196 http://www.avayalabs.com/project/libsafe/index.html, 518 http://www.calderasystems.com/, 554 http://www.cifs.com, 425 http://www.debian.org/, 554 http://www.exim.org/, 301 http://www.geocities.com/swan daniel/colsfaq.html, 523 http://www.gnome.org/, 554 http://www.gnu.org/, 553, 555, 556, 573 http://www.google.com/, 118 http://www.google.com/linux, 118 http://www.icon.co.za/psheer/ruteerrata.html, 3 http://www.kde.org/, 554 http://www.li.org/, 553 http://www.linux-mandrake.com/, 554 http://www.linux.org.uk/, 553 http://www.linux.org/, 553 http://www.linuxdoc.org/, 132, 133 http://www.linuxhq.com/, 553 http://www.linuxplanet.com/, 119 http://www.linuxsecurity.com/, 523 http://www.linuxsecurity.com/docs/colsfaq.html, 523 http://www.lpi.org/, 531 http://www.mail-abuse.org/, 313 http://www.newsforge.net/, 119 http://www.opensound.com, 474 http://www.orbiten.org/, 555 http://www.orbl.org/, 313 http://www.orbs.org/, 313 http://www.ordb.org/, 313 http://www.pathname.com/fhs/, 347 http://www.php.net, 406 http://www.redhat.com/, 554 http://www.redhat.com/support/manuals/, 544 http://www.securityfocus.com/forums/bugtraq/intro.html, 516 http://www.sendmail.org/faq/, 315, 548 http://www.slackware.com/, 554

http://www.snort.org/, 523 http://www.sourceforge.net/, 553 http://www.spy.net/dustin/modem/, 24 http://www.suse.com/, 554 http://www.teleport.com/curt/modems.html, 24 http://www.tripwiresecurity.com/, 523 http://www.turbolinux.com/, 554 http://www.usenix.org/publications/login/1997-11/embraces.html, 430 http://www.winehq.com/faq.html, 571 http://www.xfree86.org/, 493, 497, 498 web sites, 534 webalizer, 550 whatis, 409 which, 47, 533 while, 62, 63, 211 whiteboard, 526 whitespace, 78 who, 108 whoami, 40 whois, 269, 281, 523, 540 Why?, exim, 301 widget library, 491 X, 491 wildcards, 29, 49, 66 copy, 532 LPI, 532 win-modem, 536 win-modems, 480 window, TCP, 265 window managers, X, 492, 495, 496, 505 windowing system, X, 486 Windows, 15, 18, 31, 40, 44, 58, 59, 81, 147, 183, 321, 525, 528, 560, 564, 567, 570 U NIX, with, 425 postgres, with, 414 Apache, with, 393 COM port, 144, 479 device drivers, 463 FAT le system, 160 GPF, fatal exception, 86 kernel loader, 468 migrating from, 567 Network Neighborhood, 425 partition, 143, 318 print share, 205 SMB vs. NFS, 285 Tk widgets, 492 use of C, 207 X, vs., 493 Windows 2000, 430 Windows 98, booting, 321 Windows oppies, 44 Windows oppy, see MS-DOS oppy

Windows in another partition, L INUX, 143 Windows Internet Name Service, see WINS Windows NT, 393, 427 Windows NT partitions, mounting, 164 Windows NT/2000, Samba, 435 Windows partition, 321 mounting, 164 WINS, 428 WINS client, LPI, 541 with gzip, tar, 46 within les, searching, 43 Word, 567 Word processing, 57 World Wide Web, 131 world-writable les, security, 517 wrapper scripts, 175 write lock, 188 write permission, 124 writing les, 217 writing random bytes, 184 wuftpd, 292 www, 266, 269, 389 WYSIWYG, 57

X
X, 59, 132, 137, 185, 325327, 473, 482, 485488, 491509, 529, 544, 569571 -bpp, 503, 504 -cc, 505 -display option, 494 -fn, 506 -visual, 505 DISPLAY, 494 DefaultDepth, 503 Device section, 502 Emulate3Buttons, 497 Files section, 508 Modeline, 502 Monitor section, 501503 Pointer section, 499 QuickStart guide, 498 Screen section, 503, 504 clocks:, 500 libX11.so, 491 libXt.so, 491 ttmkfdir, 509 16-color X server, 499, 505 API, 498 artifacts, 503 background, 496 black and white, 505 CGA, 505 client machine, 485 clipboard, 497 Clockchip setting, 501

627

628

Index

compatibility, 487 conguration le, 486, 499 conguration scripts, 505 conguring, 498 conventions, 496 cut buffer, 497 cutting, 497 default visual, 505 device driver module, 502 DirectColor, 505 display, 486 distribution, 497 documentation, 491, 497 EGA, 505 Emacs, 59 events, 487 exposed window, 487 extensible, 487 FAQ, 498 font conguration, 508 font database, 509 font directory, 508 font family, 506 font name elds, 506 font naming, 506 font path, 508 font pixel resolution, 507 font server, 509 font size, 507 font weight, 507 graphical login box, 506 graphics, 486 graphics card, 504 graphics context, 487 graphics hardware, 486 graphics hardware acceleration, 503 gray-level display, 500, 505 GrayScale, 505 GUI, 485 hardware cursor, 503 HGC, 505 intercharacter spacing, 507 key conventions, 493 key press, 488 kill the X server, 493 library function, 488 MDA, 505 mouse and keyboard input, 487 mouse click, 488 mouse conventions, 497 multiheaded display, 494 network transparent, 486 OSF, 485 palette, 505 paste, 487, 497

PostScript, 487 programing, 498 programming, 498 programs, 494 protocol, 486 protocol responses, 486 PseudoColor, 505 RAM DAC, 501 remote, 486 RGB value, 487 RHCE, 546 root window, 494 scalable fonts, 509 second X server, 495 selecting text, 497 selections, 497 server, 485, 506 socket connection, 486 Start button, 496 StaticColor, 505 StaticGray, 505 TCP/IP, 486 three-button mouse, 497 TrueColor, 504, 505 TrueType font, 509 utilities, 494 Vertical sync, 501 video buffer, 504 video memory pixmap caching, 503 visuals, 504 web site, 498 widget library, 491 window managers, 492, 495, 496, 505 windowing system, 486 zooming, 494 X, 494 X client server, RHCE, 546 X Consortium, 485, 493 X font server, LPI, 539 X server, LPI, 539 X Toolkit, 491, 492 X Toolkit application, 491 X Window System, see X font, 29 X, vs. Windows, 493 X.25, 547 x11perf, 494 x11perfcomp, 494 x86 boot sector, 148 xargs, 180, 532 xauth, 494, 549 xbiff, 494 xbill, 495 xbooks, 498

xcalc, 494, 495 .Xclients, 546 xclipboard, 494 xclock, 494, 495 xcmsdb, 494 Xconfigurator, 501, 544, 546 xconsole, 494 xcutsel, 494 .Xdefaults, 539 xditview, 494 xdm, 494, 506, 546 xdm(1), 546 xdpyinfo, 494, 505 xedit, 494, 495 xenix, 163 xev, 494 xeyes, 494 XF86 S3, 501 XF86 S3V, 501 XF86 SVGA, 501 XF86Config, 486, 500 xf86config, 494, 501, 539, 546 XF86Setup, 546 xfd, 494 xfig, 491 xfindproxy, 494 xfontsel, 494, 507 xforms, 492 XFree86, 494 XFree86-doc, 498 XFS, le system, 160 xfs, 494, 509, 547 xfstt, 510 xfwp, 494 xgamma, 494 xgc, 494 xhost, 494, 547, 549 xiafs, 163 xieperf, 494 xinetd, 291, 295, 297, 301, 338, 389, 394, 434, 512, 518 xinetd.conf, 295 xinit, 494, 505, 546 xinit(1), 505, 546 .xinitrc, 539 xinitrc, 546 xkbbell, 494 xkbcomp, 494 xkbevd, 494 xkbprint, 494 xkbvleds, 494 xkbwatch, 494 xkill, 494 xload, 494, 549 xlogo, 494

xlsatoms, 494 xlsclients, 494 xlsfonts, 494 xmag, 494 xman, 494 Xmark, 494 xmessage, 494 xmodmap, 494 Xnest, 495 xon, 494 xosview, 549 .xpm, 31 XPM image, 487 xprop, 494 Xprt, 494 xrdb, 494 xrefresh, 494 .xsession, 546 xset, 494, 508, 510 xsetmode, 494 xsetpointer, 494 xsetroot, 494 xsm, 494 xstdcmap, 494 xsysinfo, 549 xterm, 494, 497, 539 xvidtune, 494 xwd, 494 xwininfo, 494 Xwrapper, 494 xwud, 494

Y
.y, 31 yahoo.com, 118 Yellow Pages, see Network Information Services ypbind, 547 yppasswd, 547 yppasswdd, 547 yppush, 547 ypserv, 547

Z
.Z, 32 .za, 276 zcat, 42 .zip, 32 zless, 42 zone, DNS, 444 zooming, X, 494

629

630