path: root/chapter_03.xml

<?xml version="1.0"?>
<!DOCTYPE chapter PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN"
  "/usr/share/xml/docbook/xml-dtd-4.5/docbookx.dtd">

<chapter>
<title>Booting</title>

<para>
Ok, now that you've gotten Slackware installed on your system, you
should learn exactly what controls the boot sequence of your machine,
and how to fix it should you manage to break it somehow.  If you use Linux 
long enough, sooner or later you will make a mistake that breaks your
bootloader.  Fortunately, this doesn't require a reinstall to fix.  Unlike
many other operating systems that hide the underlying details of how they
work, Linux (and in particular, Slackware) gives you full control over
the boot process.  Simply by editing a configuration file or two and
re-running the bootloader installer, you can quickly and easily change
(or break) your system.  Slackware even makes it easy to dual-boot
multiple operating systems, such as other Linux distributions or Microsoft 
Windows.
</para>

<section>
<title><application>mkinitrd</application></title>

<para>
Before we go any further, a quick discussion on the Linux kernel is
warranted.  Slackware Linux includes at least two, but sometimes more,
different kernels.  While they are all compiled from the same source 
code, and hence are the "same", they are not identical.  Depending on 
your architecture and Slackware version, the installer may have loaded 
your system with several kernels.  There are kernels for single-processor 
systems and kernels for multi-processor systems (on 32bit Slackware).
In the old days, there were lots of kernels for installing on many different 
kinds of hard drive controllers.  More importantly for our discussion, 
there are "huge" kernels and "generic" kernels.
</para>

<para>
If you look inside your <filename>/boot</filename> directory, you'll
see the various kernels installed on your system.
</para>

<screen><prompt>darkstar:~# </prompt><userinput>ls -1 /boot/vmlinuz*</userinput>
/boot/vmlinuz-huge-2.6.29.4
/boot/vmlinuz-generic-2.6.29.4
</screen>

<para>
Here you can see that I have two kernels installed,
<filename>vmlinuz-huge-2.6.29.4</filename> and
<filename>vmlinuz-generic-2.6.29.4</filename>.  Each Slackware release
includes different kernel versions and sometimes even slightly
different names, so don't be alarmed if what you see doesn't exactly
match what I have listed here.
</para>

<para>
Huge kernels are exactly what you might think; they're huge.  However,
that does NOT mean that they have all of the possible drivers and such
compiled into them.  Instead, these kernels are made to boot (and run) on
every conceivable computer on which Slackware is supported (there may very
well be a few out there that won't boot/work with them though).  They most 
certainly contain support for hardware your machine does not (and never 
will) have, but that shouldn't concern you.  These kernels are included for 
several reasons, but probably the most important is their use by Slackware's
installer - these are the kernels that the Slackware installation disks run. 
If you chose to let the installer configure your bootloader for you, it 
chooses to use these kernels due to the incredible variety of hardware they 
support.  In contrast, the generic kernels support very little hardware 
without the use of external modules.  If you want to use one of the generic 
kernels, you'll need to make use of something called an initrd, which is
created using the <application>mkinitrd</application>(8) utility.
</para>

<para>
So why should you use a generic kernel?  Currently the Slackware
development team recommends use of a generic kernel for a variety of
reasons.  Perhaps the most obvious is size.  The huge kernels are
currently about twice the size of the generic kernels before they are
uncompressed and loaded into memory.  If you are running an older
machine, or one with some small ammount of RAM, you will appreciate the
savings the generic kernels offer you.  Other reasons are somewhat more
difficult to quantify.  Conflicts between drivers included in the huge
kernels do appear from time to time, and generally speaking, the huge
kernels may not perform as well as the generic ones.  Also, by using the
generic kernels, special arguments can be passed to hardware drivers
seperately, rather than requiring these options be passed on the kernel
command line.  Some of the tools included with Slackware work better if
your kernel uses some drivers as modules rather than statically building
them into the kernel.  If you're having trouble understanding this, don't
be alarmed: just think "huge kernel = good, generic kernel = better".
</para>

<para>
Unfortunately, using the generic kernels isn't as straightforward as
using the huge kernels.  In order for the generic kernel to boot your
system, you must also include a few basic modules in an initird. 
Modules are pieces of compiled kernel code that can be inserted or removed 
from a running kernel (ideally using <application>modprobe</application>(8).
This makes the system somewhat more flexible at the cost of a tiny bit of 
added complexity.  You might find it easier to think of modules as device 
drivers, at least for this section.  Typically you will need to add the 
module for whatever filesystem you chose to use for your root partition 
during the installer, and if your root partition is located on a SCSI disk 
or RAID controller, you'll need to add those modules as well.  Finally, if 
you're using software RAID, disk encryption, or LVM, you'll also need to 
create an initrd regardless of whether you're using the generic kernel or not.
</para>

<para>
An initrd is a compressed <application>cpio</application>(1) archive, so
creating one isn't very straightforward.  Fortunately for you, Slackware 
includes a tool that makes this very easy:
<application>mkinitrd</application>.  A full discussion of
<application>mkinitrd</application> is a bit beyond the scope of this
book, but we'll show you all the highlights.  For a more complete
explanation, check the manpage or run
<application>mkinitrd</application> with the <arg>--help</arg>
argument.
</para>

<screen><prompt>darkstar:~# </prompt><userinput>mkinitrd --help</userinput>
mkinitrd creates an initial ramdisk (actually an initramfs cpio+gzip
archive) used to load kernel modules that are needed to mount the
root filesystem, or other modules that might be needed before the
root filesystem is available.  Other binaries may be added to the
initrd, and the script is easy to modify.  Be creative.  :-)
.... many more lines deleted ....
</screen>

<para>
When using <application>mkinitrd</application>, you'll need to know a
few items of information: your root partition, your root filesystem,
any hard disk controllers you're using, and whether or not you're using
LVM, software RAID, or disk encryption.  Unless you're using some kind
of SCSI controller (and have your root partition located on the SCSI
controller), you should only need to know your root filesystem and
partition type.  Assuming you've booted into your Slackware installation
using the huge kernel, you can easily find this information with the
<application>mount</application> command or by viewing the contents of
<filename>/proc/mounts</filename>.
</para>

<screen><prompt>darkstar:~# </prompt><userinput>mount</userinput>
/dev/sda1 on / type ext4 (rw,barrier=1,data=ordered)
proc on /proc type proc (rw)
sysfs on /sys type sysfs (rw)
usbfs on /proc/bus/usb type usbfs (rw)
/dev/sda2 on /home type jfs (rw)
tmpfs on /dev/shm type tmpfs (rw)
</screen>

<para>
In the example provided, you can see that the root partition is located
on <filename>/dev/sda1</filename> and is an ext4 type partition. If we
want to create an initrd for this system, we simply need to tell this
information to <application>mkinitrd</application>.
</para>

<screen><prompt>darkstar:~# </prompt><userinput>mkinitrd -f ext4 -r /dev/sda1</userinput>
</screen>

<para>
Note that in most cases, <application>mkinitrd</application> is smart
enough to determine this information on its own, but it never hurts to
specify it manually. Now that we've created our initrd, we simply need
to tell LILO where to find it.  We'll focus on that in the next section.
</para>

<para>
Looking up all those different options for
<application>mkinitrd</application> or worse, memorizing them, can be a
real pain though, especially if you try out different kernels
consistently. This became tedious for the Slackware development team,
so they came up with a simple configuration file,
<filename>mkinitrd.conf</filename>(5). You can find a sample file that
can be easily customized for your system at 
<filename>/etc/mkinitrd.conf.sample</filename> directory. Here's mine.
</para>

<screen>
<prompt>darkstar:~# >/prompt><userinput>cat /etc/mkinitrd.conf.sample</userinput>
# See "man mkinitrd.conf" for details on the syntax of this file
#
SOURCE_TREE="/boot/initrd-tree"
CLEAR_TREE="0"
OUTPUT_IMAGE="/boot/initrd.gz"
KERNEL_VERSION="$(uname -r)"
#KEYMAP="us"
MODULE_LIST="ext3:ext4:jfs"
#LUKSDEV="/dev/hda1"
ROOTDEV="/dev/sda1
ROOTFS="ext4"
#RESUMEDEV="/dev/hda2"
#RAID="0"
LVM="1"
#WAIT="1"
</prompt>
</screen>

<para>
For a complete description of each of these lines and what they do,
you'll need to consult the man page for <filename>mkinitrd.conf</filename>.
Copy the sample file to to <filename>/etc/mkinitrd.conf</filename> and
edit it as desired.  Once it is setup properly, you need only run 
<application>mkinitrd</application> with the <arg>-F</arg> argument. 
A proper initrd file will be constructed and installed for you without 
you having to remember all those obscure arguments.
</para>

<para>
If you're unsure what options to specify in the configuration file or
on the command-line, there is one final option. Slackware includes a
nifty little utility that can tell what options are required for your
currently running kernel
<application>/usr/share/mkinitrd/mkinitrd_command_generator.sh</application>.
When you run this script, it will generate a command line for
<application>mkinitrd</application> that should work
for your computer, but you may wish to check everything anyway.
</para>

<screen><prompt>darkstar:~# </prompt><userinput>/usr/share/mkinitrd/mkinitrd_command_generator.sh</userinput>
mkinitrd -c -k 2.6.33.4 -f ext3 -r /dev/sda3 -m \
  usbhid:ehci-hcd:uhci-hcd:ext3 -o /boot/initrd.gz
</screen>

</section>

<section>
<title>LILO</title>

<para>
LILO is the Linux Loader and is currently the default boot loader
installed with Slackware Linux.  If you've used other Linux
distributions before, you may be more familiar with GRUB.  If you prefer
to use GRUB instead, you can easily find it in the 
<filename>extra/</filename> directory on one of your Slackware CDs. 
However, since LILO is the default Slackware bootloader, we'll focus 
exclusively on it.
</para>

<para>
Configuring LILO can be a little daunting for new users, so Slackware
comes with a special setup tool called <application>liloconfig</application>.
Normally, <application>liloconfig</application> is first run by the
installer, but you can run it at any time from a terminal.
</para>

<para>

<imagedata fileref="img/liloconfig01.png" format="PNG" />

</para>

<para>
<application>liloconfig</application> has two modes of operation:
simple and expert.  The "simple" mode tries to automatically configure
lilo for you.  If Slackware is the only operating system installed on
your computer, the "simple" mode will almost always do the right thing
quickly and easily.  It is also very good at detecting Windows
installations and adding them to <filename>/etc/lilo.conf</filename>
so that you can choose which operating system to boot when you
turn your computer on.
</para>

<para>
In order to use "expert" mode, you'll need to know Slackware's root
partition.  You can also setup other linux operating systems if you know
their root partitions, but this may not work as well as you expect.
<application>liloconfig</application> will try to boot each linux
operating system with Slackware's kernel, and this is probably not what
you want.  Fortunately, setting up Windows partitions in expert mode is
trivial.  One hint when using expert mode: you should almost always
install LILO to the Master Boot Record (MBR).  Once upon a time, it was
recommended to install the boot loader onto the root partition and set
that partition as bootable.  Today, LILO has matured greatly and is safe
to install on the MBR.  In fact, you will encounter fewer problems if
you do so.
</para>

<para>
<application>liloconfig</application> is a great way to quickly setup
your boot loader, but if you really need to know what's going on, you'll
need to look at LILO's configuration file:
<filename>lilo.conf</filename>(5) under the <filename>/etc</filename>
directory.  <filename>/etc/lilo.conf</filename> is separated into
several sections.  At the top, you'll find a "global" section where you
specify things like where to install LILO (generally the MBR), any
special images or screens to show on boot, and the timeout after which
LILO will boot the default operating system.  Here's what the global
section of my lilo.conf file looks like in part.
</para>

<screen># LILO configuration file

boot = /dev/sda
  bitmap = /boot/slack.bmp
  bmp-colors = 255,0,255,0,255,0
  bmp-table = 60,6,1,16
  bmp-timer = 65,27,0,255

append=" vt.default_utf8=0"
prompt
timeout = 50

# VESA framebuffer console @ 1024x768x256
vga = 773
.... many more lines ommitted ....
</screen>

<para>
For a complete listing of all the possible LILO options, you should
consult the man page for <filename>lilo.conf</filename>.  We'll 
briefly discuss the most common options in this document.
</para>

<para>
The first thing that should draw your attention is the "boot" line.  This
determines where the bootloader is installed.  In order to install to
the Master Boot Record (MBR) of your hard drive, you simply list the hard 
drive's device entry on this line.  In my case, I'm using a SATA hard drive 
that shows up as a SCSI device <filename>/dev/sda</filename>.  In order
to install to the boot block of a partition, you'll have to list the 
partition's device entry.  For example, if you are installing to the first 
partition on the only SATA hard drive in your computer, you would probably 
use <filename>/dev/sda1</filename>.
</para>

<para>
The "prompt" option simply tells LILO to ask (prompt) you for which
operating system to boot.  Operating systems are each listed in their
own section deeper in the file.  We'll get to them in a minute.  The
timeout option tells LILO how long to wait (in tenths of seconds)
before booting the default OS.  In my case, this is 5 seconds.  Some
systems seem to take a very long time to display the boot screen, so
you may need to use a larger timeout value than I have set.  This is in
part why the simple LILO installation method utilizes a very long
timeout (somewhere around 2 whole minutes).  The append line in my case
was set up by <application>liloconfig</application>.  You may (and
probably should) see something similar when looking at your own
<filename>/etc/lilo.conf</filename>. I won't go into the details of why 
this line is needed, so you're just going to have to trust me that things
work better if it is present. :^)
</para>

<para>
Now that we've looked into the global section, let's take a look at the
operating systems section.  Each linux operating system section begins
with an "image" line.  Microsoft Windows operating systems are specified
with an "other" line. Let's take a look at a sample
<filename>/etc/lilo.conf</filename> that boots both Slackware and 
Microsoft Windows.
</para>

<screen># LILO configuration file
... global section ommitted ....
# Linux bootable partition config begins
image = /boot/vmlinuz-generic-2.6.29.4
  root = /dev/sda1
  initrd = /boot/initrd.gz
  label = Slackware64
  read-only
# Linux bootable partition config ends
# Windows bootable partition config begins
other = /dev/sda3
  label = Windows
  table = /dev/sda
# Windows bootable partition config ends
</screen>

<para>
For Linux operating systems like Slackware, the image line specifies
which kernel to boot.  In this case, we're booting
<filename>/boot/vmlinuz-generic-2.6.29.4</filename>. The remaining
sections are pretty self-explainatory.  They tell LILO where to find the
root filesystem, what initrd (if any) to use, and to initially mount
the root filesystem read-only.  That initrd line is very important for
anyone running a generic kernel or using LVM or software RAID.  It
tells LILO (and the kernel) where to find the initrd you created using
<application>mkinitrd</application>.
</para>

<para>
Once you've gotten <filename>/etc/lilo.conf</filename> set up for your
machine, simply run <application>lilo</application>(8) to install it.
Unlike GRUB and other bootloaders, LILO requires you re-run
<application>lilo</application> anytime you make changes to its
configuration file, or else the new (changed) bootloader image will 
not be installed, and those changes will not be reflected.
</para>


<screen><prompt>darkstar:~# </prompt><userinput>lilo</userinput>
Warning: LBA32 addressing assumed
Added Slackware *
Added Backup
6 warnings were issued.
</screen>

<para>
Don't be too scared by many of the warnings you may see when running
<application>lilo</application>. Unless you see a fatal error, things
should be just fine. In particular, the LBA32 addressing warning is
commonplace.
</para>

</section>

</chapter>