amb@gedanken.demon.co.uk
The Loopback Root Filesystem HOWTO Copyright (C) 1998,99 Andrew M. Bishop (amb@gedanken.demon.co.uk).
This documentation is free documentation; 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.
The GNU General Public License is available from http://www.fsf.org/ or, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111 USA
Initial Version (June 1998)
Slight Modifications, kernel version changes, typos etc. (1998 - July 1999)
Added Copyright Information and Re-Submitted (September 1999)
First I will describe some of the general principles that are used in the setting up of a loopback filesystem as the root device.
A loopback device in Linux is a virtual device that can be used like any other media device.
Examples of normal media devices are hard disk partitions like
/dev/hda1, /dev/hda2, /dev/sda1, or entire disks like
the floppy disk /dev/fd0 etc. They are all devices that can be used to
hold a files and directory structures. They can be formatted with the
filesystem that is required (ext2fs, msdos, ntfs etc.) and then mounted.
The loopback filesystem associates a file on another filesystem as a complete
device. This can then be formatted and mounted just like any of the other
devices listed above. To do this the device called /dev/loop0 or
/dev/loop1 etc is associated with the file and then this new virtual
device is mounted.
In Linux it is also possible to have another type of virtual device mounted as a filesystem, this is the ramdisk device.
In this case the device does not refer to any physical hardware, but to a portion of memory that is set aside for the purpose. The memory that is allocated is never swapped out to disk, but remains in the disk cache.
A ramdisk can be created at any time by writing to the ramdisk device
/dev/ram0 or /dev/ram1 etc. This can then be formatted and
mounted in the same way that the loopback device is.
When a ramdisk is used to boot from (as is often done on Linux installation disks or rescue disks) then the disk image (the entire contents of the disk as a single file) can be stored on the boot floppy in a compressed form. This is automatically recognised by the kernel when it boots and is uncompressed into the ramdisk before it is mounted.
The initial ramdisk device in Linux is another important mechanism that we need to be able to use a loopback device as a the root filesystem.
When the initial ramdisk is used the filesystem image is copied into memory and
mounted so that the files on it can be accessed. A program on this ramdisk
(called /linuxrc) is run and when it is finished a different device is
mounted as the root filesystem. The old ramdisk is still present though and is
mounted on the directory /initrd if present or available through the
device /dev/initrd.
This is unusual behaviour since the normal boot sequence boots from the designated root partition and keeps on running. With the initial ramdisk option the root partition is allowed to change before the main boot sequence is started.
The root filesystem is the device that is mounted first so that it appears as
the directory called / after booting.
There are a number of complications about the root filesystem that are due to
the fact that it contains all files. When booting the rc scripts are
run, these are either the files in /etc/rc.d or /etc/rc?.d
depending on the version of the /etc/init program.
When the system has booted it is not possible to unmount the root partition or change it since all programs will be using it to some extent. This is why the initial ramdisk is so useful because it can be used so that the final root partition is not the same as the one that is loaded at boot time.
To show how the initial ramdisk operates in the boot sequence, the order of events is listed below.
LILO or
LOADLIN. You can see the Loading... message as this happens.LILO or LOADLIN. You can see the Loading... message
again as this happens./linuxrc is run on the initial ramdisk./etc/init is run which will perform the user
configurable boot sequence.This is just a simplified version of what happens, but is sufficient to explain how the kernel starts up and where the initial ramdisk is used.
Now that the general principles are explained the method of creating the loopback device can be explained.
To create the loopback root device will require a number of things.
Most important is access to an installed Linux system. This is because the loop device can only be created under Linux. This will mean that it is not possible to bootstrap a working system from nothing. The requirements of the Linux system that you use is that you can compile a kernel on it.
Once the loopback device is created it will be a large file. I have used an 80 MB files, but while this was sufficient for an X terminal it may not be enough if you want to use it for much else. This file must be copied onto the DOS partition, so either a network or a lot of floppy disks must be used.
The software that you will require includes
LOADLIN version 1.6 or abovemount that supports loopback devicesAll of these should be standard for recent Linux installations.
I created the loopback device using Linux kernel version 2.0.31, other versions should also work, but they must have at least the options listed below.
The kernel options that you will need to enable are the following:
CONFIG_BLK_DEV_RAM).CONFIG_BLK_DEV_INITRD).CONFIG_BLK_DEV_LOOP).CONFIG_FAT_FS).CONFIG_MSDOS_FS).The first two are for the RAM disk device itself and the initial ram disk device. The next one is the loop back filesystem option. The last two are the msdos filesystem support which is required to mount the DOS partitition.
Compiling a kernel without modules is the easiest option, although if you do want modules then it should be possible although I have not tried it. If modules are used then you should make sure that you have the options above compiled in and not as modules themselves.
Depending on the kernel version that you have you may need to apply a kernel patch. It is a very simple one that allows the loopback device to be used as the root filesystem.
For 2.0.x kernels the file /init/main.c needs to have a single line
added to it as shown by the modified version below. The line that says
"loop", 0x0700 is the one that was added.
static void parse_root_dev(char * line)
{
int base = 0;
static struct dev_name_struct {
const char *name;
const int num;
} devices[] = {
{ "nfs", 0x00ff },
{ "loop", 0x0700 },
{ "hda", 0x0300 },
...
{ "sonycd", 0x1800 },
{ NULL, 0 }
};
...
}
For 2.2.x kernels the file /init/main.c needs to have three lines added
to it as shown by the modified version below. The line that says "loop",
0x0700 and the ones either side of it are the ones that were added.
static struct dev_name_struct {
const char *name;
const int num;
} root_dev_names[] __initdata = {
#ifdef CONFIG_ROOT_NFS
{ "nfs", 0x00ff },
#endif
#ifdef CONFIG_BLK_DEV_LOOP
{ "loop", 0x0700 },
#endif
#ifdef CONFIG_BLK_DEV_IDE
{ "hda", 0x0300 },
...
{ "ddv", DDV_MAJOR << 8},
#endif
{ NULL, 0 }
};
Once the kernel is configured it should be compiled to produce a zImage
file (make zImage). This file will be arch/i386/boot/zImage
when compiled.
The initial ramdisk is most easily created as a loopback device from the start.
You will need to do this as root, the commands that you need to execute are
listed below, they are assumed to be run from root's home directory
(/root).
mkdir /root/initrd
dd if=/dev/zero of=initrd.img bs=1k count=1024
mke2fs -i 1024 -b 1024 -m 5 -F -v initrd.img
mount initrd.img /root/initrd -t ext2 -o loop
cd initrd
[create the files]
cd ..
umount /root/initrd
gzip -c -9 initrd.img > initrdgz.img
There are a number of steps to this, but they can be described as follows.
Contents Of The Initial Ramdisk
The files that you will need on the ramdisk are the minimum requirements to be able to execute any commands.
/linuxrc The script that is run to mount the msdos file system
(see below)./lib/* The dynamic linker and the libraries that the programs
need./etc/* The cache used by the dynamic linker (not strictly needed,
but does stop it complaining)./bin/* A shell interpreter (ash because it is smaller
than bash. The mount and losetup programs for
handling the DOS disk and setting up the loopback devices./dev/* The devices that will be used. You need
/dev/zero for ld-linux.so, /dev/hda* to mount the
msdos disk and /dev/loop* for the lopback device./mnt An empty directory to mount the msdos disk on.The initial ramdisk that I used is listed below, the contents come to about 800kB when the overhead of the filesystem are taken into account.
total 18
drwxr-xr-x 2 root root 1024 Jun 2 13:57 bin
drwxr-xr-x 2 root root 1024 Jun 2 13:47 dev
drwxr-xr-x 2 root root 1024 May 20 07:43 etc
drwxr-xr-x 2 root root 1024 May 27 07:57 lib
-rwxr-xr-x 1 root root 964 Jun 3 08:47 linuxrc
drwxr-xr-x 2 root root 12288 May 27 08:08 lost+found
drwxr-xr-x 2 root root 1024 Jun 2 14:16 mnt
./bin:
total 168
-rwxr-xr-x 1 root root 60880 May 27 07:56 ash
-rwxr-xr-x 1 root root 5484 May 27 07:56 losetup
-rwsr-xr-x 1 root root 28216 May 27 07:56 mount
lrwxrwxrwx 1 root root 3 May 27 08:08 sh -> ash
./dev:
total 0
brw-r--r-- 1 root root 3, 0 May 20 07:43 hda
brw-r--r-- 1 root root 3, 1 May 20 07:43 hda1
brw-r--r-- 1 root root 3, 2 Jun 2 13:46 hda2
brw-r--r-- 1 root root 3, 3 Jun 2 13:46 hda3
brw-r--r-- 1 root root 7, 0 May 20 07:43 loop0
brw-r--r-- 1 root root 7, 1 Jun 2 13:47 loop1
crw-r--r-- 1 root root 1, 3 May 20 07:42 null
crw-r--r-- 1 root root 5, 0 May 20 07:43 tty
crw-r--r-- 1 root root 4, 1 May 20 07:43 tty1
crw-r--r-- 1 root root 1, 5 May 20 07:42 zero
./etc:
total 3
-rw-r--r-- 1 root root 2539 May 20 07:43 ld.so.cache
./lib:
total 649
lrwxrwxrwx 1 root root 18 May 27 08:08 ld-linux.so.1 -> ld-linux.so.1.7.14
-rwxr-xr-x 1 root root 21367 May 20 07:44 ld-linux.so.1.7.14
lrwxrwxrwx 1 root root 14 May 27 08:08 libc.so.5 -> libc.so.5.3.12
-rwxr-xr-x 1 root root 583795 May 20 07:44 libc.so.5.3.12
./lost+found:
total 0
./mnt:
total 0
The only complex steps about this are the devices in dev. Use the
mknod program to create them, use the existing devices in /dev
as a template to get the required parameters.
The /linuxrc file
The /linuxrc file on the initial ramdisk is required to do all of the
preparations so that the loopback device can be used for the root partition when
it exits.
The example below tries to mount /dev/hda1 as an msdos partition and
if it succeeds then sets up the files /linux/linuxdsk.img as
/dev/loop0 and /linux/linuxswp.img as /dev/loop1.
#!/bin/sh
echo INITRD: Trying to mount /dev/hda1 as msdos
if /bin/mount -n -t msdos /dev/hda1 /mnt; then
echo INITRD: Mounted OK
/bin/losetup /dev/loop0 /mnt/linux/linuxdsk.img
/bin/losetup /dev/loop1 /mnt/linux/linuxswp.img
exit 0
else
echo INITRD: Mount failed
exit 1
fi
The first device /dev/loop0 will become the root device and the second
one /dev/loop1 will become the swap space.
If you want to be able to write to the DOS partition as a non-root user when you
have finished then you should use mount -n -t msdos /dev/hda1 /mnt -o
uid=0,gid=0,umask=000,quiet instead. This will map all accesses to the DOS
partition to root and set the permissions appropriately.
The root device that you will be using is the file linuxdsk.img. You
will need to create this in the same way that the initial ramdisk was created,
but bigger. You can install any Linux installation that you like onto this
disk.
The easiest way might be to copy an existing Linux installation into it. An alternative is to install a new Linux installation onto it.
Assuming that you have done this, there are some minor changes that you must make.
The /etc/fstab file must reference the root partition and the
swap using the two loopback devices that are setup on the initial ramdisk.
/dev/loop0 / ext2 defaults 1 1
/dev/loop1 swap swap defaults 1 1
This will ensure that when the real root device is used the kernel will not be confused about where the root device is. It will also allow the swap space to be added in the same way a swap partition is normally used. You should remove any other reference to a root disk device or swap partition.
If you want to be able to read the DOS partition after Linux has started then you will need to make a number of extra small changes.
Create a directory called /initrd, this is where the initial ramdisk
will be mounted once the loopback root filesystem is mounted.
Create a symbolic link called /DOS that points to /initrd/mnt
where the real DOS parition will be mounted.
Add a line into the rc file that mounts the disks. This should run the command
mount -f -t msdos /dev/hda1 /initrd/mnt, this will create a 'fake'
mount of the DOS partition so that all programs (like df) will know that
the DOS partition is mounted and where to find it. If you used different
options in the /linuxrc file that obviously you should use them here
also.
There is no need to have a Linux kernel on this root device since that is already loaded earlier. If you are using modules however then you should include them on this device as normal.
The root device that you will be using is the file linuxswap.img. The
swap device is very simple to create. Create an empty file as was done for the
initial ramdisk and then run mkswap linuxswap.img to intialise it.
The size of the swap space will depend on what you plan to do with the installed system, but I would recommend between 8MB and the amount of RAM that you have.
The files that are going to be used need to be moved onto the DOS partition.
The files that are required in the DOS directory called C:\LINUX are
the following:
LINUXDSK.IMG The disk image that will become the root device.LINUXSWP.IMG The swap space.
The boot floppy that is used is just a normal DOS format bootable floppy.
This is created using format a: /s from DOS.
Onto this disk you will need to create an AUTOEXEC.BAT file (as below)
and copy the kernel, compressed initial ramdisk and LOADLIN executable.
AUTOEXEC.BAT The DOS automatically executed batch file.LOADLIN.EXE The LOADLIN program executable.ZIMAGE The Linux kernel.INITRDGZ.IMG The compressed initial ramdisk image.The AUTOEXEC.BAT file should contain just one line as below.
\loadlin \zImage initrd=\initrdgz.img root=/dev/loop0 ro
This specifies the kernel image to use, the initial ramdisk image, the root device after the initial ramdisk has finished and that the root partition is to be mounted read-only.
To boot from this new root device all that is required is that the floppy disk prepared as described above is inserted for the PC to boot from.
You will see the following sequence of events.
/linuxrc file on the initial ramdisk is runWhen this is complete you can remove the boot floppy and use the Linux system.
There are a number of stages where this process could fail, I will try to explain what they are and what to check.
DOS booting is easy to recognise by the message that it prints MS-DOS
Starting ... on the screen. If this is not seen then the floopy disk is
either not-bootable or the PC is not bootable from the floppy disk drive.
When the AUTOEXEC.BAT file is run the commands in it should be echoed
to the screen by default. In this case there is just the single line in the
file that starts LOADLIN.
When LOADLIN executes it will do two very visible things, firstly it
will load the kernel into memory, secondly it will copy the ramdisk into memory.
Both of these are indicated by a Loading... message.
The kernel starts by uncompressing itself, this can give crc errors if the kernel image is corrupted. Then it will start running the initialisation sequence which is very verbose with diagnostic messages. Loading of the initial ramdisk device is also visible during this phase.
When the /linuxrc file is run there is no diagnostic messages, but you
can add these yourself as an aid to debugging. If this stage fails to set up
the loopback device as the root device then you may see a message that there is
no root device and the kernel aborts.
The normal boot sequence of the new root device will now continue and this is
quite verbose. There may be problems about the root device being mounted
read-write, but the LOADLIN command line option 'ro' should
stop that. Other problems that can occur are that the boot sequence is confused
about where the root device is, this is probably due to a problem with
/etc/fstab.
When the boot sequence has completed, the remaining problem is that programs are confused about whether the DOS partition is mounted or not. This is why it is a good idea to use the fake mount command described earlier. This makes life a lot easier if you want to access the files on the DOS device.
The documents that I used to create my first loopback root filesystem were:
init/main.cDocumentation/initrd.txt and Documentation/ramdisk.txt.LILO documentation.LOADLIN documentation
Once the principle of booting a filesystem in a file on a DOS partition has been established there are many other things that you can now do.
If it is possible to boot Linux from a file on a DOS harddisk by using a boot floppy then it is obviously also possible to do it using the harddisk itself.
A configuration boot menu can be used to give the option of running
LOADLIN from within the AUTOEXEC.BAT. This will give a much
faster boot sequence, but is otherwise identical.
Using LOADLIN is only one option for booting a Linux kernel. There is
also LILO that does much the same but without needing DOS.
In this case the DOS format floppy disk can be replaced by an ext2fs format one. Otherwise the details are very similar, with the kernel and the initial ramdisk being files on that disk.
The reason that I chose the LOADLIN method is that the arguments that
need to be given to LILO are slightly more complex. Also it is more
obvious to a casual observer what the floppy disk is since it can be read under
DOS.
I have tried the NTFS method, and have had no problems with it. The NTFS filesystem driver is not a standard kernel option in version 2.0.x, but is available as a patch from http://www.informatik.hu-berlin.de/~loewis/ntfs/. In version 2.2.x the NTFS driver is included as standard in the kernel.
The only changes for the VFAT or NTFS options are in the initial ramdisk, the
file /linuxrc needs to mount a file system of type vfat or ntfs rather
that msdos.
I know of no reason why this should not also work on a VFAT partition.
The process of installing Linux on a PC from a standard distribution requires booting from a floppy disk and re-partitioning the disk. This stage could instead be accomplished by a boot floppy that creates an empty loopback device and swap file. This would allow the installation to proceed as normal, but it would install into the loopback device rather than a partition.
This could be used as an alternative to a UMSDOS installation, it would
be more efficient in disk usage since the minimum allocation unit in the ext2
filesystem is 1kB instead of up to 32kB on a DOS partition. It can also be used
on VFAT and NTFS formatted disks which are otherwise a problem.
This method can also be used to boot a Linux system from a device that is not normally bootable.
Obviously there are many other devices that could be used, NFS root filesystems are already included in the kernel as an option, but the method described here might also be used instead.