Linux File System

Question 1

What’s the basic definition of a linux file

Answer 1

A sequence of 0 or more bytes, arbitrary information. No distinction is made between any kind of file.

Question 2

How does linux handle race conditions in read / write operations to files?

Answer 2

Using locks. The locking algorithm requires the caller to specify the file to be locked, the starting byte and the number of bytes.

Question 3

How can a file be created?

Answer 3

Through the creat system call. The file is created and opened for writing.

Question 4

What is returned by the creat system call?

Answer 4

A file descriptor, file is configured in open mode.

Question 5

What are file descriptors 0, 1 and 2?

Answer 5

These are standard input, output and error. When used as part of pipes these are redirected to the corresponding pipes (buffers).

Question 6

Explain what’s an i-node

Answer 6

Representation of a file (including directories and special files) under the linux virtual file system.

Question 7

How are paths represented in the virtual file system?

Answer 7

As D-entries in the central dentry data structure.

Question 8

What information is contained in the Superblock

Answer 8

Critical information about the layout of the file system; number of i-nodes, number of disk blocks and the start of the list of free disk blocks.

Question 9

What are the components of an ext2 block group and their roles?

Answer 9

Superblock: a copy of the file system’s superblock. Group descriptor table: information about the group itself, such as location of bitmaps, number of free blocks and i-nodes in the group. Block bitmap: bitmap that tracks usage status of the blocks in the block group. I-node bitmap: tracks usage status of the nodes in the block group. I-node table: table containing i-nodes for files and directories stored in the block group. Data blocks: the actual blocks that contain data.

Question 10

What is a block?

Answer 10

Fixed-size unit of data storage. Basic unit of data management in a file system.

Question 11

Describe what a disk request looks like

Answer 11

First stage: software queue (device driver), second stage: hardware controller, third stage actual disk movement.

Question 12

Name a few characteristics a FS must provide

Answer 12

Naming, organizing fine names with directories, organization (map files to blocks), protection (enforcing access restrictions) and reliability.

Question 13

Where does protection for a certain file reside?

Answer 13

In the I-node itself. Not in the block.

Question 14

What is a sector in a hard disk?

Answer 14

It’s a physical portion of the disk.

Question 15

What algorithm does the disk use for scheduling?

Answer 15

Elevator algorithm: it chooses a direction for the disk head to move, as the disk moves, it services all the read and write requests that it encounters along the way. If the disk reaches end of disk or there are no more requests, it changes direction. This continues indefinitely.

Question 16

What does t he filesystem need to work?

Answer 16

Track free disk blocks. Track a mapping of file -> blocks. Track what files are in a directory. All stored on disk

Question 17

What is the minimum read/write size?

Answer 17

It’s the size of the whole block (may be 1KB to 4KB)

Question 18

How does a file descriptor reference a file?

Answer 18

It stores the i-number NOT the file name.

Question 19

Give a high level definition of what is a directory

Answer 19

It’s a file containing file_name -> file_number mappings

Question 20

How can you read the contents of a directory?

Answer 20

By using readdir system call

Question 21

How many disk accesses to resolve “/my/book/count/

Answer 21

Read file header (lookup i-node for root dir), read first data block for root, find table of file name/index pairs, search table to find “my”. Repeat until “count”.

Question 22

What happens when you use mv to move a file?

Answer 22

It only changes the name and what directory is pointing to it. The file itself never changes.

Question 23

What does FAT stand for and how does it work in essence?

Answer 23

It stands for File Allocation Table. It’s essentially a collection of files with start indexes. Blocks in a file are stored as a linked list of blocks.

Question 24

What is the purpose of doubly indirect and triply indirect pointers in an i-node?

Answer 24

These are used y files thart are very large. It makes the usage of really small files efficient but also allows the existance of very large files.

Question 25

(ext2) What data structures are required for a process to write or read from a file?

Answer 25

The first data structure is the file descriptor table, it has a pointer to an entry in the open-file-description table, containing a file position (indicating what position we are reading / writing at in the file).

Question 26

How many addresses does an i-node hold?

Answer 26

12 direct disk blocks and then 3 indirect blocks. The 3 indirect blocks is first a single indirect block, that holds addresses for other blocks, and then a doubly and triply indirect block.

Question 27

What is a journaling file system? Provide an example and why they’re useful.

Answer 27

A journaling file system is one that maintain a journal describing all file-system operations in sequential order. Eventually, all operations are committed to the appropriate disk location and corresponding journal entries are discarded. On a crash or power failure, he system detects uncommitted changes and applies them.