Chapter 4 INTERNAL REPRESENTATION OF FILES THE DESIGN

Chapter 4. INTERNAL REPRESENTATION OF FILES THE DESIGN OF THE UNIX OPERATING SYSTEM Maurice J. bach Prentice Hall 1

Contents n n n n Inodes Structure of a regular file Directories Conversion of a path name to an inode Super block Inode assignment to a new file Allocation of disk blocks Other file types 2

File System Algorithms Lower Level File system Algorithms namei alloc free iget ialloc ifree iput bmap buffer allocation algorithms getblk brelse breada bwrite 3

4. 1 Inode n n contains the information necessary for a process to access a file exits in a static form on disk and the kernel reads them into an in-core inode 4

4. 1 Inodes n consists of - file owner identifier - file type - file access permissions - file access times - number of links to the file - table of contents for the disk address of data in a file - file size 5

4. 1 Inodes n in-core copy of the inode contains - status of the in-core inode - logical device number of file system - inode number - pointers to other in-core inodes - reference count 6

Algorithm iget 1. The kernel finds the inode in inode cache and it is on inode free list remove from free list increment inode reference count 2. The kernel cannot find the inode in inode cache so it allocates a inode from inode free list remove new inode from free list reset inode number and file system remove inode from old hash queue, place on new one read inode from disk(algorithm bread) initialize inode 7

Algorithm iget 3. The kernel cannot find the inode in inode cache but finds the free list empty error : process have control over the allocation of inodes at user level via execution of open and close system calls and consequently the kernel cannot guarantee when an inode will become available 4. The kernel finds the inode in inode cache but it was locked sleep until inode becomes unlocked 8

iget (inode_no) //get. Incore. Inode n while (not done) n if (inode in inode cache) n if (inode locked) n n n if (inode on inode free list) n n n n n sleep(event inode becomes unlocked) continue remove from free list return locked inode if (no inode on free list) return error remove new inode from free list set inode number remove inode from old hash queue and place on new one read inode from disk set reference count 1 return locked indoe 9

Algorithm iput - The kernel locks the inode if it has not been already locked - The kernel decrements inode reference count - The kernel checks if reference count is 0 or not - If the reference count is 0 and the number of links to the file is 0, then the kernel releases disk blocks for file(algorithm free), free the inode(algorithm ifree) • If the file was accessed or the inode was changed or the file was changed , then the kernel updates the disk inode • The kernel puts the inode on free list - If the reference count is not 0, the kernel releases the inode lock 10

iput (inode_no) //release. Incore. Inode n n n lock inode if not locked decrement inode refernece count if (refernce count==0) n if (inode link==0) n n if (file accessed or inode changed or file changed) n n n free disk block set file type to 0 free inode update disk inode put inode on free list Release inode lock 11

4. 2 Structure of a regular file Inode Data Blocks direct 0 direct 1 direct 2 direct 3 direct 4 direct 5 direct 6 direct 7 direct 8 direct 9 single indirect double indirect triple indirect 12

4. 2 Structure of a regular file Suppose System V UNIX Assume that a logical on the file system holds 1 K bytes and that a block number is addressable by a 32 bit integer, then a block can hold up to 256 block numbers 10 direct blocks with 1 K bytes each=10 K bytes 1 indirect block with 256 direct blocks= 1 K*256=256 K bytes 1 double indirect block with 256 indirect blocks= 256 K*256=64 M bytes 1 triple indirect block with 256 double indirect blocks= 64 M*256=16 G bytes 13

4. 2 Structure of a regular file Processes access data in a file by byte offset and view a file as a stream of bytes n The kernel accesses the inode and converts the logical file block into the appropriate disk block n algorithm bmap - The kernel calculates logical block number in file from byte offset - The kernel calculates start byte in block for I/O - The kernel calculates number of bytes to copy to user - The kernel checks if read-ahead is applicable, then marks inode - The kernel determines level of indirection n 14

4. 2 Structure of a regular file - While it’s not at necessary level of indirection, the kernel calculates index into inode or indirect block from logical block number in file, gets disk block number from inode or indirect block and release buffer from previous disk read • If there is no more levels of indirection , the kernel stops conversing • Otherwise the kernel reads indirect disk block(bread) and adjusts logical block number in file according to level of indirection 15

4. 3 Directories n n A directory is a file whose data is a sequence of entries, each consisting of an inode number and the name of a file contained in the directory Path name is a null terminated character string divided by slash (“/”) Each component except the last must be the name of a directory, last component may be a non-directory file Directory layout for /etc Byte Offset in Directory Inode Number File Name 0 83. 16 2. . 32 1798 init 16

4. 4 Path conversion to an inode n if (path name starts with root) n n else n n working inode= root inode working inode= current directory inode while (there is more path name) n n n read next component from input read directory content if (component matches an entry in directory) n n n get inode number for matched component release working inode=inode of matched component else return no inode return (working inode) 17

Algorithm namei - If path name starts from root, then the kernel assigns root inode(iget) to working inode - Otherwise, the kernel assigns current directory inode to working inode - While there is more path name, the kernel reads next path name component from input, and verifies that working inode is of directory, access permissions OK • If working inode is of root and component is ‘. . ’, then the kernel checks whethere is more path name or not • Otherwise the kernel reads directory by repeated use of bmap, bread, brelse 18

Path conversion to an inode • If the kernel finds a match, it records the inode number of the matched directory entry, releases the block and the old working inode, and allocates the inode of the match component • If the kernel does not match the path name in the block, it releases the block, adjusts the byte offset by the number of bytes in a block, converts the new offset to a disk block number and reads the new block 19

4. 5 Super block n File System boot block super block n inode list data blocks consists of - the size of the file system - the number of free blocks in the file system - a list of free blocks available on the file system - the 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 - a flag indicating that the super block has been modified 20

4. 6 Inode assignment to a new file n Locked inode illoc() n while (not done) n If (super blocked) n Sleep (event super block becomes free) n Continue n If (inode list in super block empty) n Lock super block n Get remember inode for free inode search n Search until super block full or no more free inode n Unlock super block and wake up (event super block free) n If no free inode found on disk return (no inode) n Set remmbered inode for next free inode search n Get inode number from super block inode list n Get inode n Write inode to disk n Decrement free inode count n Return inode 21

4. 6 Inode assignment to a new file n algorithm ialloc : assigns a disk inode to a newly created file -super block is unlocked 1. There are inodes in super block inode list and inode is free get inode number from super block inode list get inode (iget) initialize inode write inode to disk decrement file system free inode count 2. There are inodes in super block inode list but inode is not free get inode number from super block inode list get inode (iget) write inode to disk release inode (iput) 22

4. 6 Inode assignment to a new file 3. Inode list in super block is empty lock super block get remembered inode for free inode search disk for free inode until super block full or no more free inodes(bread and brelse) unlock super block becomes free if no free inodes found on disk , stop otherwise, set remembered inode for next free inode search - If super block is locked, sleep 23

4. 6 Inode assignment to a new file Super Block Free Inode List free inodes 83 48 18 array 1 19 empty 20 index Super Block Free Inode List free inodes 83 18 array 2 empty 19 20 index Assigning Free Inode from Middle of List 24

4. 6 Inode assignment to a new file Super Block Free Inode List 470 empty 0 array 1 index remembered inode Super Block Free Inode List 535 free inodes 0 array 2 476 475 471 48 49 50 index Assigning Free Inode – Super Block List Empty 25

Algorithm ifree - The kernel increments file system free inode count - If super block is locked, avoids race conditions by returning - If super block is unlock and inode list is full , • If inode number is less than remembered inode for search, then the kernel remembers the newly freed inode number, discarding the old remembered inode number from the super block - If super block is unlock and inode list is not full, then the kernel stores inode number in inode list 26

Freeing inode n ifree(inode_no) n n n Increment free inode count If super blocked return If (inode list full) //at super block n if (inode number <remembered inode) n n Else n n Set remembered inode as input inode Store inode number in inode list return 27

4. 6 Inode assignment to a new file 535 476 475 471 free inodes remembered inode index Original Super Block List of Free Inodes 499 476 475 471 free inodes remembered inode index Free Inode 601 28

4. 6 Inode assignment to a new file n n A Race Condition Scenario in Assigning Inodes Consider three processes A, B, and C and suppose that the kernel, acting on behalf of process A, assigns inode I but goes to sleep before it copies the disk inode into the in-core copy. While process A is asleep, suppose process B attempts to assign a new inode but free inode list is empty, and attempts assign free inode at an inode number lower than that of the inode that A is assigning. Suppose process C later requests an inode and happens to pick inode I from the super block free list 29

4. 6 Inode assignment to a new file Process A Process B Process C Assigns inode I from super block Sleeps while reading inode(a) Tries to assign inode from super block use Super block empty(b) the lock Search for free inodes on disk, puts inode I in super block (c) Inode I in core Does usual activity Completes search, assigns another inode(d) Assigns inode I from super block I is in use! Assign another inode(e) Race Condition in Assigning Inodes 30

4. 7 Allocation of disk blocks 109 106 103 100 ……………. . 109 211 208 205 202 ………………… 211 310 307 304 301 112 ………………… 214 310 409 406 403 400 ………………… 313 linked list of free disk block number 31

Algorithm alloc - The kernel wants to allocate a block from a file system it allocates the next available block in the super block list - Once allocated , the block cannot be reallocated until it becomes free - If the allocated block is the last block , the kernel treats it as a pointer to a block that contains a list of free blocks • The kernel locks super block, reads block jut taken from free list, copies block numbers in block into super block, releases block buffer, and unlocks super block - Otherwise, • The kernels gets buffer for block removed from super block list , zero buffer contents, decrements total count of free blocks, and marks super block modified 32

4. 7 Allocation of disk blocks super block list 109 …………………………… 109 211 208 205 202 ………………. . 112 original configuration 109 949 …………………………. . 109 211 208 205 202 ………………. 112 After freeing block number 949 33

4. 7 Allocation of disk blocks 109 ……………………………. . 109 211 208 205 202 ………………. 112 After assigning block number(949) 211 208 205 202 ……………… 112 211 344 341 338 335 ………………. 243 After assigning block number(109) replenish super block free list 34

4. 8 Other file types n pipe - fifo(first-in-first-out) - its data is transient: Once data is read from a pipe, it cannot be read again, no deviation from that order - use only direct block n special file - include block device, character device - the inode contains the major and minor device number - major number indicates a device type such as terminal or disk - minor number indicates the unit number of the device 35