Project 4 - File Systems
Administrivia: For this project, you may work with the same partners you
had in project 3. However, if you want to be matched with a group, please send
email to Adrienne by Sunday, November 23.
Out: Friday, November 21
Due: Friday, December 5, 4:00pm
Assignment Goals
-
To understand the problems that file system implementations must solve, and the
range of approaches that might be taken
-
To practice design (in this case of file systems)
-
To experience working in a "more sophisticated" environment: complex software,
a variety of tools, and teams of programmers
-
To gain further experience with concurrent software
Overview
The starting point for this assignment is a simplified file system, cse451fs,
the design of which imposes strict limits on both the number of files that can
be stored and the maximum size of any one file. In particular, no matter
how big a disk you might have, this file system can hold only about 8,000
distinct files, no file can be larger than 13KB, and file names cannot be
longer than 30 characters. These restrictions result from the choice of
on-disk data structures used to find files and the data blocks of a given file,
that is, the superblock and inode representations.
Here are the major steps involved in this assignment:
-
Make sure that you, individually, understand the mechanical aspects of the
"development environment" you'll be working in. These include how to
build the file system code, how to configure the raw disk device provided by
VMware to host your file system, and how to run and test your file
system. A description of these mechanical aspects is
here.
-
Of the three limitations cited above you will need to improve the following
two:
-
Increase the maximum size of files.
-
Allow for longer file names.
Design how you want to implement these file system modifications on disk: how
you will represent your directories on disk, how file data is indexed,
etc. There can still be a limit on any of these properties, but your
improvement needs to be more than simply altering a program constant.
If you still have time left over, you may do
additional improvements for up to 15% extra
credit on this project. for example, you can
fix the limit on the number of files, or
make the file system more efficient in
terms of space usage or in terms of CPU
usage.
-
Alter the skeleton code (/cse451/projects/Proj4FS.tar.gz) to
implement your file system. There are two major components to this.
One is that the user level program mkfs.cse451fs must be changed to
initialize the raw disk device with a valid, empty file system using your new
on-disk data structures. The other is to change the file system source (fsSource/)
itself.
Details
-
While real file systems are very concerned with performance, in your
implementation you can largely ignore it. That is, do not
spend a great deal of effort to produce a faster implementation. (For one
thing, because we're running on virtual machines on top of Windows, it's
unlikely you'd be able to measure much difference.)
-
A description of the skeleton version of the cse451fs file system is
here. (necessary reading)
-
A description of the ext2 file system and vfs (Virtual File
System) is
here. (strongly recommended)
-
A description of how dynamically loaded modules are handled in Linux is
here. (not required reading but may answer odd questions that
arise)
Hints/Starting Points
-
Large Files:
-
An important function for creating and accessing the blocks of a file (that you
will almost certainly need to modify) is get_block() in super.c.
-
Look at cse451_truncate() in file.c in order to handle file
truncations (e.g., deleting a file uses this). You don't need to worry
about this until you are sure that you can create and access your files.
-
Long File Names:
-
All the functions that you will need to modify for your kernel module are in dir.c.
-
Start with cse451_add_entry() and cse451_readdir() to create
and read directory entries with long file names. If listing with ls
doesn't appear correct, check how the directory structure looks on disk (using hexdump
or xxd).
-
General:
-
If you modify any of the data structures in cse451fs.h, you will
probably need to modify mkfs.cse451fs in addition to the kernel module
sources. Otherwise you probably can leave mkfs alone.
Schedule and What to Hand In
By Friday, December 5 at 4:00 PM, please turn in a file that
includes the following.
-
Describe the design for your file system modifications. This might
include a discussion of other approaches you considered but rejected.
-
What concurrency-related issues did you encounter? How did you deal with them?
-
What methodology did you follow in order to test your file system (for
functionality)?
-
Does your implementation work? If not, what parts work and what parts don't?
How would you fix it if you had more time?
-
What do you like best about your design? What do you like least about
it? How would you improve your design?
The report should be no more than 4 pages long.
The file may be plain (ASCII) text or HTML, and should
contain the names of all the people who worked on the project
as well as your group name.
You will be graded primarily on your write-up. We will be looking at the clarity
and viability of your ideas, depth of your implementation, and completeness of
your testing, based on your report. However, please do turn in a copy of your
code as well, in case we need to refer to it.
Turnin
You will be turning in 4 files: a tar.gz file containing all of your
modified sources, your compiled mkfs.cse451fs and cse451fs.o files,
and a single write-up file for both partners.
To create the source archive file, use the following command: tar -cvzf
sources.tar.gz Proj4FS/
Use the turnin(1L) program under project name proj4 by 4:00pm
on the day it is due. Note: turnin will not work on coredump/spinlock,
so you'll need to use attu.