BITS Pilani

Tutorial session 5: Internal

Representation of File system
(Continued)
 Review of last tutorial
• Free list, lock information, reference count
• iget() algorithm
• Releasing nodes
• Structure of a regular file
• Allocation of disk space for a file in non-contiguous spaces
• Inode structure of table of contents
 Single indirect block
 Double indirect block
 Triple indirect block
• Maximum file size referenced by the table
• Accessing given file’s byte offset and computing the block
number using table structure of system V Unix system

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 2

 Algorithms used in File handling
internally by the operating system
• iget(): allocation for in-core inodes
• iput(): releasing an inode
• bmap(): conversion of byte offset to block
number in file system
• namei(): conversion of a path name to an
inode.
• ialloc(): for assigning new nodes
• ifree(): for freeing inode
• alloc(): allocating disk block

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 3

 Directories
• Directories are the files that give the file
system its hierarchical structure.
• Its data is a sequence of entries, each
consisting of an inode number and the name
of a file contained in the directory.
• The last component of the path may be a non-
directory file.
• UNIX system V restricts the components name
to a maximum of 14 characters (2 bytes for
inode number)
September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 4
 Directory layout for /etc

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 5

 Directories
• Directories are maintained by the kernel in the same
way it maintains the files
• Each directory has an associated inode.
• Processes read the directories in the same way as they
read an ordinary file
• A process must have permission to search the contents
of the directory
• A directory’s data is the information about other sub-
directories and files.
• The inode corresponding to the directory name has
information about other sub-directories in its data
block table.

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 6

 Conversion of a path name to an
inode
• The kernel uses an algorithm namei() for converting
the path name to a inode
• Example path name is ../etc/subdir1/subdir2/file
The path has components separated by /
• The algorithm parses the components one by one and
converts each component to an inode based on its
name and directory
• Processes can change their directory using chdir
command
• Namei() uses intermediate inodes as it parses a path
name
• The intermediate inodes are called as working inodes

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 7

 Working inodes
• The root .. Is the first working inode.
• The byte offsets of the subdirectories are
placed in the data blocks of the parent
directory.
• The search for the working inode terminates
after the leaf node of the path in the file’s
hierarchical structure
• The namei() algorithm uses bmap() algorithm
to reach the block containing data of working
inode iteratively for each component.
September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 8
 Superblock
The record structure of the superblock is
• Size of file system
• Number of free blocks in the file system
• List of free blocks available on the file system
• Index of the next free block in the free block list
• The size of the inode list
• The number of free inodes in the file system
• A list of free inodes in the file system
• The index of the next free inode in the free inode list
• Lock fields for the free block and free inode lists
• The flag indicating the the super block has been modified

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 9

 Inode assignment to a new file
• For given path name, the kernel uses namei()
algorithm to convert this into corresponding
inode.
• If the inode is previously determined, the kernel
uses iget() algorithm to allocate the inode to the
file.
• If the file is newly created, then the algorithm
ialloc() assigns a disk inode to the file.
• The superblock free inode information is
accessed by the kernel for getting the free inode.

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 10

 Inode assignment to a new file
• The kernel first verifies that no other processes have
locked access to the super block free inode list
• If the list of inode numbers in the superblock is not
empty,
 the kernel assigns the next inode number to the new file
 Allocates a free in-core inode for the newly assigned disk
inode using iget()
 Copies the disk inode to the in-core copy
 Initializes the fields in the inode
 And returns the locked inode (to indicate that the inode is
in use)

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 11

 Assigning free inode from middle
of list

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 12

 When the list of inodes is empty
• If the list of inode numbers in the superblock is empty
 The kernel searches the disk
 Places as many free inode numbers as possible into the
superblock
 The kernel reads the inode list on disk, block by block, and
fills the superblock list of inode numbers to the full
capacity
 It remembers the highest numbered inode that it finds and
calls that “remembered node”
 The remembered node is the last one saved in the
superblock
 Next time when kernel needs to search the free list on
disk, it starts its search from the remembered inode

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 13

 Assigning free inode when
superblock list is empty

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 14

 ialloc()
algorithm

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 15

 Allocation of disk blocks
• When a process writes data to a file, the kernel
allocates disk blocks from the file system for direct and
indirect data blocks.
• File system super block contains an arraythat is used to
cache the numbers of free disk blocks in the file
system.
• The utility program mkfs (make file system) organizes
the data blocks of a file system in a linked list.
• The linked list has address of the next block containing
free disk block numbers
• The first entry of the block can be the address of the
next block containing the addresses of free blocks

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 16

Linked list of free
Disk Block
Numbers

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 17

 Alloc()
algorithm

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 18

 Original configuration

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 19

 When a block is freed, its number is added to
the free list of disk blocks

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 20

 An other block is needed, then
taken from the super block list

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 21

 When the super block list is
empty

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 22

 Freeing a disk block
• It is the reverse of allocating a block for an
inode
• When a block is freed, its number is written in
the super block list
 If the super block list is not full, then the block
number of the freed block is written onto the
super block list
 If the super block list is full, then the newly freed
block becomes a link block. Later all free blocks
numbers are further written in this block

September 25, 2020 OS Tutorial 5 (Sections 3 and 4) 23