Assignment: Filesystem Integrity

Assigned

• October 5, 2026

Due

• October 12, 2026 11:59pm

Collaboration

You are welcome to discuss general concepts related to this assignment with other students in the class, but you may not discuss answers to any of the assignment’s questions or any of the provided code snippets with anyone except the instructor.

Submitting

Find this assignment on Gradescope and submit your responses to each question there.

In our discussion of filesystems in class, we considered the “simple” operation of creating a new file in the user’s home directory. We worked as a class to build a list of all the accesses and updates required to carry out that operation on a filesystem that does not use journaling. The list of steps we came up with in class is shown below, with the write operations highlighted in bold text:

• Read the inode bitmap to find an available inode entry
• Write the inode bitmap to claim the selected inode entry
• Read the data block bitmap to find an available data block
• Write the data block bitmap to claim the selected data block
• Write the new file’s inode entry, setting the file permissions, size, and point to its associated data block
• Write the new file’s data block, storing the new file’s contents
• Read the home directory inode entry to find its data block
• Read the home directory data block so we can add a new entry to its content
• Write the directory data block, adding the newly created file to its list of files

We generally allow disks and operating systems to reorder storage operations to help with performance, which means the writes highlighted above could be carried out in any order. If something goes wrong during this operation, we may be left with a filesystem that has received only a subset of the writes. We’ll consider three scenarios where a different subset of writes are lost. For each scenario, answer the following questions:

Part 1: What is the state of the filesystem after the failed update?

Describe what this state would look like to a user looking for the file that was just created, and explain how that state may change over time if nothing is done to correct the issue.

Part 2: Is the filesystem in a recoverable state?

If so, explain what the OS could do to restore the filesystem to a consistent state and what that state would be. If not, explain why the filesystem is not recoverable. Keep in mind, a “consistent state” does not necessarily mean the new file has to exist; if removing the remnants of the partially-created new file can fix the issue, that would still be consistent. A filesystem is in a consistent state when its on-disk data is in agreement; every inode entry and data block reachable from a reachable directory is marked as in-use in the corresponding bitmap, and all unreachable inode entries and data blocks are marked free in their respective bitmaps.

Scenarios

For each scenario, answer the two questions above and explain your answers. Your explanations must clearly describe the filesystem state and/or relevant operations to justify your initial answer. A few sentences should be sufficient in most cases, but there is no specific minimum or limit on the length of your response.

Scenario 1

The writes to the inode and data block bitmaps are both lost. The other three writes succeed.

Scenario 2

The write to the directory data block is lost, but all other writes succeed.

Scenario 3

The writes to the new file’s inode entry and the data block bitmap are lost. The other three writes succeed.