Tutorial For Lpi Exam 201: Part 2

Topic 202: Understanding System Startup

David Mertz, Ph.D.
Professional Neophyte
July, 2005

Welcome to "Understanding System Startup", the second of eight tutorials designed to prepare you for LPI exam 201. In this tutorial you will learn the steps a Linux system goes through during system initialization, and how to modify and customize those behaviors for you specific needs.

Before You Start

About this series

The Linux Professional Institute (LPI) certifies Linux system administrators at junior and intermediate levels. There are two exams at each certification level. This series of eight tutorials helps you prepare for the first of the two LPI intermediate level system administrator exams--LPI exam 201. A companion series of tutorials is available for the other intermediate level exam--LPI exam 202. Both exam 201 and exam 202 are required for intermediate level certification. Intermediate level certification is also known as certification level 2.

Each exam covers several or topics and each topic has a weight. The weight indicate the relative importance of each topic. Very roughly, expect more questions on the exam for topics with higher weight. The topics and their weights for LPI exam 201 are:

Topic 201: Linux Kernel (5) Topic 202: System Startup (5) Topic 203: Filesystems (10) Topic 204: Hardware (8) Topic 209: File Sharing Servers (8) Topic 211: System Maintenance (4) Topic 213: System Customization and Automation (3) Topic 214: Troubleshooting (6)

About this tutorial

Welcome to "Understanding the System Startup", the second of eight tutorials designed to prepare you for LPI exam 201. In this tutorial you will learn the steps a Linux system goes through during system initialization, and how to modify and customize those behaviors for you specific needs.


To get the most from this tutorial, you should already have a basic knowledge of Linux and a working Linux system on which you can practice the commands covered in this tutorial.

About the Linux system startup

This tutorial is at the border of Linux, strictly speaking. Topic 201 addressed the kernel itself, which is the core of Linux. This tutorial moves on to the ancillary tools and scripts that are necessary both to get the kernel running, and to initialize a system to the point it does something meaningful. Hower the scripts and tools associated with initialization are maintained by the creators of Linux distributions, or individualized by system administrators, rather than developed as part of the Linux kernel per se. Still, every Linux system--even an embedded one--will require some basic initialization steps similar to those discussed here.

In later tutorials, we will look at a variety of tools for networking, system maintenance, manipulating files and data, and so on, which are important for a working Linux installation and part of almost every Linux distribution, but are even less part of Linux per se than are initialization scripts.

The Linux Boot Process

What happens when you turn a Linux computer on? (part one)

It is useful to break the Linux boot process into 9 steps that will occur in almost every Linux configuration. Steps 10 and beyond might involve launching additional services, logging into a graphical environment, restoring UI settings, or other more personalized details that are outside this tutorial.

1. Hardware/firmware: The BIOS or firmware system reads the master boot record on the harddisk or other boot device (e.g. CD, floppy, netboot, etc).

2. A boot loader runs. Linux systems on x86 systems typically use

LILO or GRUB. Some older systems might use loadlin to boot via an intermediate DOS partition. On PowerPC systems, this might be BootX or yaboot. In general, a boot loader is a simple program that knows where to look for the Linux kernel, perhaps choosing among several versions, or even selecting other operating systems on the same machine.

3. The kernel loads.

4. The root filesystem is mounted. In some cases, a temporary ramdisk image is loaded before the true root filesystem to enable special drivers or modules that might be necessary for the true root filesystem.

Keep reading boot steps on the next page

What happens when you turn a Linux computer on? (part two)

One we have a root filesystem in place (see last page), we are ready for initialization proper.

5. Start the process init, the parent of all other Linux processes.

6. Read the contents of /etc/inittab to configure the remaining boot steps. Of special importance is the line in /etc/inittab that controls the runlevel the system will boot to (and therefore, which further steps will be taken during initialization).

Actually, everything after this point is completely controlled by the content of the file /etc/inittab. Specifically, the scripts and tools that run generally follow some conventions, but in theory you could completely change /etc/inittab to run different scripts.

One specific setting in /etc/inittab is particularly crucial. A line similar to:


Generally occurs near the top of the file, and sets the runlevel. This runlevel controls what actions are taken in the remainder on the /etc/inittab script.

What happens when you turn a Linux computer on? (part three)

Just what happens as an /etc/inittab script is processed? And specifically, what conventional files and directories are involved in the process?

7. Runlevel-neutral system initialization. Generally there are some initialization actions that are performed regardless of runlevel. These steps are indicated in /etc/inittab with a line like:

# System initialization.

     On some Linux systems (mostly Debian based), you will see something
     more like:


     If the latter case, '/etc/init.d/rcS' is a script that simply runs
     each of the scripts matching '/etc/rcS.d/[Ss]??*'. On the other
     hand, if your system uses '/etc/rc.d/rc.sysinit', that file
     contains a single long script to perform *all* the initialization.

What happens when you turn a Linux computer on? (part four)

8. Runlevel-specific system initialization. You may actually define as many actions as you like related to runlevel, and each action may pertain to one or more runlevels. As a rule, /etc/inittab will contain some lines like:

l0:0:wait:/etc/rc.d/rc 0
# ...
l5:5:wait:/etc/rc.d/rc 5
l6:6:wait:/etc/rc.d/rc 6

     In turn, the script '/etc/rc.d/rc' will run all the files
     matched by the pattern '/etc/rc$1.d/[KkSs]??*'.  For example, on
     the sample system described, that start at runlevel 5, we would run
     (in order):


     The files(s) starting with 'K' or 'k' are "kill scripts" that end
     processes or cleanup their actions.  Those starting with 'S' or 's'
     are startup scripts that generally launch new processes or
     otherwise prepare the system to run at that runlevel.  Most of
     these files ran will be shell scripts, and most will be links
     (often to files in '/etc/init.d/')

Logging into the running Linux system

Most of the time, once a Linux system is running at a runlevel, you wish to log into the system as a user. To let that happen, a program called getty runs to handle the login process. A number of variations on the basic getty are used by distribution creators, such as agetty, mgetty and mingetty. All do basically the same thing, however.

9. Login at the prompt. Our good friend /etc/inittab/ will usually launch getty programs on one or more virtual terminals, and do so in for several different runlevels. Those are configured with lines like:

# Run gettys in standard runlevels
1:2345:respawn:/sbin/mingetty tty1
2:2345:respawn:/sbin/mingetty tty2
3:2345:respawn:/sbin/mingetty tty3
4:2345:respawn:/sbin/mingetty tty4
5:2345:respawn:/sbin/mingetty tty5
6:2345:respawn:/sbin/mingetty tty6

     The first number is reminds us of the virtual terminal where the
     getty runs, the next set of number are the several runlevels where
     this will happen. E.g. we launch 'mingetty' in all of the runlevels
     2, 3, 4 and 5.

Understanding runlevels

The concept of runlevel is somewhat arbitrary, or at least it is not hardcoded into a Linux kernel. Valid runlevel numbers to set with the initdefault option (or override otherwise) are 0-6. By convention, the following meanings are given to each number:

# Default runlevel. The runlevels used by Mandrake Linux are:
#   0 - halt (Do NOT set initdefault to this)
#   1 - Single user mode
#   2 - Multiuser, without NFS (The same as 3, if you do not have networking)
#   3 - Full multiuser mode
#   4 - unused
#   5 - X11
#   6 - reboot (Do NOT set initdefault to this)

This convention, as you can see, is as used in the Mandrake Linux distribution, but most distributions obey the same convention. Text only or embedded distributions might omit usage of some of the levels, but will still reserve their hypothetical use at the given numbers.

Configuration lines in /etc/inittab

We have seen a number of /etc/inittab lines in examples, but it is worth understanding explicitly what these lines do. Each one has the format:


The id field is simply a short abbreviation naming the configuration line (1-4 character in recent versions of init, 1-2 in ancient ones). The runlevels we have explained well enough. Next is the action taken by the line. Some actions are "special", such as:

ca::ctrlaltdel:/sbin/shutdown -t3 -r now

Which traps the ctrl-alt-delete key sequence (regardless of runlevel). But most actions simply relate to spawning. A non-exhaustive list of actions includes:

* respawn: The process will be restarted whenever it terminates (e.g.


* wait: The process will be started once when the specified runlevel

is entered and init will wait for its termination.

* once: The process will be executed once when the specified

runlevel is entered.

* boot: The process will be executed during system boot (but after

sysinit). The runlevels field is ignored.

Customizing System Startup

What is a boot loader?

A few years ago, a program called lilo was pretty much universally used to boot Linux on x86 systems. The name lilo is short for "LInux LOader." Nowadays, another program called grub (GRand Unified Bootloader) is more popular. On non-x86 Linux systems, other boot loaders are used, but they are generally configured in the same manner as lilo or grub.

While there are differences in their configuration syntaxes, lilo and grub perform largely the same task. Basically, either one presents a choice of operating systems (including, perhaps, multiple Linux kernels), and loads the selected OS kernel into memory. Both programs let you pass arguments on to a Linux kernel along the way, and both can be configured to boot non-Linux operating systems on the same machine.

Either lilo or grub (or other boot loaders) generally live in the MBR (Master Boot Record) of the primary harddisk, which is automatically loaded by system BIOS. lilo was restricted to loading specific raw sector from a harddisk. grub is more sophisticated in itself understanding a number of filesystem types such as ext2/3, ReiserFS, VFAT, UFS. This means that grub doesn't need to rewrite the MBR every time a configuration file is changes, as lilo requires.

Configuring the LILO boot loader (part one)

The lilo boot loader is configured with the contents of the file /etc/lilo.conf. For full details on configuration options, read the man page on lilo.conf. Several initial options control general behavior. For example, you will often see boot=/dev/hda or similar; this means to install lilo to the MBR of the first IDE harddisk. You might also install lilo within a particular partition, usually because you use a different main boot loader. For example, boot=/dev/sda3 installs lilo to the third partition of the first SCSI disk. Other options control the appearance and wait time of lilo.

The thing to keep in mind is that after you have edited a /etc/lilo.conf configuration, you need to run lilo to actually install a new boot sector used during initialization. It is easy to forget to install new settings, but the boot loader itself cannot read the configuration, except as encoded as raw sector offsets (which lilo calculates when run).

Configuring the LILO boot loader (part two)

In the main, when using lilo you are interested in the one or more image= lines, and perhaps in some other= lines if you multiboot to other operating systems. A sample /etc/lilo.conf might contain:

      append="devfs=mount acpi=off quiet"

This would allow you to choose at runtime either a 2.7.4 development kernel, or a stable kernel (the latter happens to utilize an initial ramdrive during boot). You can also select a DOS installation installed to partition 3 on the first IDE drive.

Configuring the GRUB boot loader

A nice thing about grub is that it does not need to be reinstalled each time you change boot configuration. Of course, you do need to install it once in the first place, usually using a command like grub-install /dev/hda. Generally, distributions will do this for you during installation, so you may never explicitly run this.

However, since grub knows how to read many filesystems, normally you can simply change the contents of /boot/grub/menu.lst to change the options for the next bootup. Let us look at a sample configuration:

timeout 5
color black/yellow yellow/black
default 0
password secretword

title linux
kernel (hd0,1)/boot/vmlinuz root=/dev/hda2 quiet
vga=788 acpi=off
initrd (hd0,1)/boot/initrd.img

title experimental
kernel (hd0,1)/boot/bzImage-2.7.4 root=/dev/hda2 quiet

title dos
root (hd0,4)
chainloader +1

Changing options within the boot loader (LILO)

Both lilo and grub allow you to pass special parameters to the kernel you select. If you use lilo you may pass boot prompt arguments by appending them to your kernel selection. For example, you might type:

LILO: linux ether=9,0x300,0xd0000 root=/dev/ha2 vga=791 acpi=on

For a custom boot setting (special options to the ethernet module, specify the root partition, choose video mode, etc). Of course, it is not all that friendly, since you need to know the exact options available, and type them exactly right.

Of particular importance is the option to change the runlevel from the boot loader. For example, for recovery purposes you may want to run in single user mode, which you can do with, e.g.:

LILO: experimental single


LILO: linux 1

Another rather special option is the init= argument that lets you use a program other than init as the first process. An option for a fallback situation might be init=/bin/sh, which at least gets you an Linux shell if init fails catastrophically.

Changing options within the boot loader (GRUB)

With grub you have even more flexibility. In fact, grub is a whole basic shell that lets you change boot loader configurations and even read filesystems. For custom boot options, simply press e in the grub shell, then add options (such as a numeric runlevel, or the keyword "single", as with lilo). All the other boot prompt arguments you might type under lilo can be edited in a grub boot command, using simple readlines style editing.

For some real sense of the power, you can open a grub command line. For example, suppose you think your /etc/inittab might be misconfigured, and you need to examine it before booting. You might type:

grub> cat (hd0,2)/etc/inittab

This would let you manually view your initialization without even launching any operating system. If there was a problem there, perhaps you would want to boot into single user mode and fix it.

Customizing what comes after the boot loader

In the main, once you understand the steps in a Linux post-kernel boot process (i.e. the init process, and everything it calls), you also understand how to customize it. Basically, customization is just a matter of editing /etc/inittab and the various scripts in /etc/rc?.d/ directories.

For example, I recently needed to customize the video bios on a Debian-based Linux laptop using a third-party tool. If this didn't run before X11 ran, my XOrg driver would not detect the correct video modes. Once I figured out what the issue actually was, the solution is as simple as creating the script /etc/rcS.d/S56-resolution.sh. In other words, I run an extra script during every system startup. Notably, I made sure this ran before /etc/rcS.d/S70xorg-common by the simple convention that scripts run in alphabetical order (if I wanted it to run later, I might have named it S98-resolution.sh instead. Arguably, I might have put this script only in the /etc/rc5.d/ directory to run when X11 does--but my way lets me manually run startx from other runlevels.

Everthing in the initialization process is out in the open, right in the filesystem; and almost all of it is in editable text scripts.

System Recovery

About recovery

The nicest thing about Linux from a maintenance perspective is that everything is a file. Of course, it can be perplexing at times to know which file something lives in. But as a rule, Linux recovery amounts to using basic filesystem utilities like cp, mv, rm, and a text editor like vi. Of course, to automate these activities, tools like grep, awk, bash are helpful; or at a higher level perl or python. This particular tutorial does not address basic file manipulation.

Assuming you know how to manipulate and edit files, the only "gotcha" perhaps remaining for a broken system is not being able to use the filesystems at all.


Your best friend in repairing a broken filesystem is fsck. Topic 203 in this series has more information, so we will just introduce the tool here.

The tool called fsck is actually just a frontend for a number of more narrow fsck.* tools, such as fsck.ext2, fsck.ext3 or fsck.reiser. You may specify the type explicitly using the -t option, but fsck will make an effort to figure it out on its own. Read the manpage for fsck or fsck.* for more details. The main think you want to know is that the -a option will try to fix everything it can automatically.

You can check an unmounted filesystem by mentioning its raw device. For example: fsck /dev/hda8 to check a partition not in use. You can also check a rooted filesystem, such as fsck /home; but generally you only want to do that if the filesystem is already mounted as read-only, not as read-write.

mount and umount

A flexible feature of Linux systems is the fine tuned control you have over mounting and unmounting filesystems. Unlike under Windows and some other operating systems, partitions are not automatically assigned locations by the Linux kernel, but are instead attached to the single / root hierarchy by the mount command. Moreover, different filesystems types (on different drives, even) may be mounted within the same hierarchy. Unmounting a particular partition is done with the umount command, specifying either the mount point (e.g. /home) or the raw device (e.g. /dev/hda7).

For recovery purposes, the ability to control mount points lets you perform forensic analysis on partitions--using fsck or other tools--without risking further damage to a damaged filesystem. You may also custom mount a filesystem using various options; the most important of these is mounting read-only using either of the synonyms -r or -o ro.

As a quick example, you might wish to substitute one user directory location for another, either because of damage to one, or simply to expand disk space or move to a faster disk. You might perform this switch using something like:

# umount /home      # old /dev/hda7 home dir
# mount -t xfs /dev/sda1 /home  # new SCSI disk using XFS
# mount -t ext3 /dev/sda2 /tmp  # also put the /tmp on SCSI


For recovery, system upgrades, and special purposes, it is useful to be able to mount and unmount filesystems at will. But for day-to-day operation, you generally want a pretty fixed set of mounts to happen at every system boot. You control the mounts that happen at bootup by putting configuration lines in the file /etc/fstab. A typical configuration might look something like the below example. The Topic 203 tutorial contains more details.

/dev/hda7 / ext3 defaults 1 1
none /dev/pts devpts mode=0620 0 0
/dev/hda9 /home ext3 defaults 1 2
none /mnt/cdrom supermount dev=/dev/hdc,fs=auto,ro,--,iocharset=iso8859-1,codepage=850,umask=0 0 0
none /mnt/floppy supermount dev=/dev/fd0,fs=auto,--,iocharset=iso8859-1,sync,codepage=850,umask=0 0 0
none /proc proc defaults 0 0
/dev/hda8 swap swap defaults 0 0