244

OK. We can mount such file systems, but how are they constructed in the

first place? This, of course requires a *nix environment. The tools are dd and

losetup, and the process goes something like this:

First it is necessary to create an empty file of the desired size. This is done

with dd and the zero device in a command like:

dd if=/dev/zero of=/../image.file bs=1k count=100000

This will create a file with 100 MB of space. Note that the file size is equivalent

to the partition size when creating partitions on a block device like the hard

disk. This file will thus hold a file system that can reach 100 MB in size before

 device full  errors occur.

Before a file system can be created on this image file and mounted, it needs

to be made to look like a block device. This is done with the losetup program

in the following fashion:

losetup /dev/loop0 /../image.file

There are 8 loop devices to choose from, so you may need to check to see

if a particular device is already in use. losetup /dev/loopN, where N ranges

between 0 and 7, will return the status of the Nth loop device if it is mounted

and will return an error message if it is not.

Now that /dev/loop0 is a legitimate block device, mke2fs (or any of the other

file system creation utilities) can be used to create a file system on the looped

image file with a command like:

mke2fs  c /dev/loop() 100000

The  c option checks the device for bad blocks and the value on the end

specifies how many blocks are on the file system. These two options can

be left off the command line and results will be seen faster, but there is no

guarantee that the resultant file system will be useful. Take the time to check

for bad blocks. It is time you will not spend looking for the cause of problems