COMPSCI 340 Operating Systems
Assignment 2 - User space file system
Worth 10%
due date 9:30pm Friday 4th of October, 2019
Introduction
In lectures the FUSE (file system in user space) library and eduFUSE were introduced and
explained. In this assignment you need to write your own user space file system using these tools.
Note the due date change. I will apply my usual penalties for late submissions: 5% off your earned
score for every day late up until and including 3 days late (15%).
Setup
Do the assignment either on Ubuntu in the Computer Science labs or on your own machine
(virtual machines work too, but not the Windows subsystem for Linux version 1). The markers will
use the lab image.
• Install the FUSE development package. (This step should already have been done on the CS lab
machines.)
sudo apt-get install libfuse-dev
• Install the maven package. (This step should also already have been done on the lab machines.)
sudo apt-get install maven
• Download the file MemoryFS.zip from the A2 files section of Canvas into a directory and
uncompress it.
It is easiest if you use an IDE or editor like Visual Studio Code (VSC). The following instructions
are for VSC. You can use other editors but you do need to create a project based on the pom.xml
file you have unpacked.
• Add the Maven extension to VSC. Maven is a project management tool. You will probably want
to install the Java Extension Pack and this includes the Maven for Java extension.
• Select File / Open Folder... in VSC. Navigate to the unzipped folder MemoryFS.
• In the Explorer panel you should see a section called Maven Projects. • Right click on MemoryFS and select Compile.
This should compile without problems. You have to be connected to internet for this. You can
ignore WARNINGs.
• Run this by navigating to src/main/java in the Explorer panel and right clicking on
MemoryFS.java and selecting Run. The program will crash. This is because you need to pass a
directory as a command line argument.
• Make a directory called memMount somewhere in your home directory. This directory is the
mount point for your user space file system.
mkdir /home/whatever/memMount
The memMount directory will normally be empty, but when your file system is active it will contain
files.
Page 1 of 4
• Pass this directory name as a command line argument to the program. In VSC you open the
launch.json file in the .vscode directory and add the following line inside
configurations:
"args": "-d -f /fullpathnameTo/memMount"
Where /fullpathnameTo/memMount is the full path to the memMount directory. The markers
will replace this with a path to a directory they have created.
When you run the program now it should open up the eduFUSE window. At the moment it won't
show very much but you can still click on things in the eduFUSE window to see what is there.
Examine the "Charts", "Directory", "Logger" and "Inodes" panels.
In the "Charts" panel you can select the system calls you want to chart, the main ones are
already selected.
In the "Directory" panel you will eventually see the files created in your file system and be
able to click on them in order to see information about them.
In the "Logger" panel you can see the information being sent to your file system and back to
the operating system. There can be a lot of information here and so you can filter it using the
options at the top of the panel. It also helps to "Clear" the information when you want to focus
on particular events.
The "Inodes" panel is useful as you get the assignment going.
Closing the eduFUSE window does not stop your file system from running. You may have to
explicitly unmount the file system with:
fusermount -u memMount
Further Documentation
There are alternative installation instructions for IntelliJ IDEA in the assignment 2 area of Canvas.
The name of the file system in this document is BlockFileSystem, but you should use
MemoryFS.
Documentation about eduFUSE including the C and Java user guides can be found at https://
github.com/lukethompsxn/edufuse-student . This includes the user guide to the eduFUSE gui
https://github.com/lukethompsxn/edufuse-student/tree/master/gui/ .
Other important documentation is in the following man pages (this is not a complete listing):
man 2 stat
man inode
man utimensat
Feel free to ask for help on Piazza for anything you are unsure of.
What you have to do
You need to complete MemoryFS.java so that in conjunction with MemoryINode.java and
MemoryInodeTable.java it runs a memory based file system. There are versions of memory
based file systems on the web but we will be checking submissions against these and these
existing versions will not get you any marks.
While the file system is running you can store files in it and access them in all the ways files can be
used.
Page 2 of 4
You can use MemoryINode.java and MemoryInodeTable.java as they are. Read through
these files and understand what they are doing. A real inode is more complicated than the
MemoryINode class but in this case we are really storing the information about the file in a
FileStat object. The other big simplification is that we can store all of the data for a file as the
content byte array instance variable.
The FileStat objects will need to have values for most of their fields, but you can ignore
st_ino, st_dev, st_rdev.
All steps below are worth two marks each.
Step 1
Complete the readdir method.
readdir gets information about all of the files in the directory.
When this is working you should be able to see the hello file in the directory, but it won't have all
necessary information.
Step 2
Complete the gettatr method. In all of the methods you can safely ignore any FuseFileInfo
parameters.
This method returns all of the file attributes in the stat parameter, the first few are done for you.
You should also fix up the init method so that all useful FileStat fields are filled in for the
hello file.
Check that when you select the hello file in the directory panel you see something like this:
The file mode will be different and of course the times will be different. The group and owner IDs
should be yours. You can use the unix instance variable with getUid and getGid to retrieve
these values.
Page 3 of 4
In all that follows you need to make sure that the time fields are correctly updated when files are
accessed and modified.
Step 3
Complete the read method so that you can read the hello file from the command line.
cat memMount/hello
Step 4
Complete the write method so that data can be written to the hello file from the command line.
echo "more" >> memMount/hello
cat memMount/hello
Step 5
Complete the mknod method so that files can be added to the directory from the command line.
touch memMount/newfile
Ensure that new files can be written to and read from.
Step 6
Complete the unlink method so that files can be deleted from the directory from the command
line.
rm memMount/newfile
Step 7
Complete the link method so that hard links can be made to files from the command line. Both
the source and destination files will be inside your file system. You will probably have to modify
the iNode classes to do this.
ln memMount/hello memMount/hi
When you modify the hello file you should see the changes in the hi file as they are different
names for the same file. Make sure that removing files works correctly with hard links.
Step 8
By now you should have a file system which works in one directory. There are two more marks if
you can add the functionality to have nested directories inside your memory file system.
The markers will go no deeper than directories inside your mounted directory. All of the preceding
operations should work with multiple files, multiple directories and files up to any size (within
reason, no files will be more than 100KiBs in size).
Submission, questions and marking guide
Information on how to submit your assignment along with a number of questions to complete and
the accompanying marking guide will be on Canvas shortly.
N.B. All submitted work must be your work alone. You may discuss assignments with others
but by submitting any work you are claiming you did that work without the contributions of
others (except for work you clearly identify as being from another source).