Files and file allocation
Copyright ©: Nahrstedt, Angrave, Abdelzaher, Caccamo
1
Copyright ©: Nahrstedt, Angrave, Abdelzaher
What is an inode?


An inode (index node) is a control structure that contains
key information needed by the OS to access a particular
file. Several file names may be associated with a single
inode, but each file is controlled by exactly ONE inode.
On the disk, there is an inode table that contains the
inodes of all the files in the filesystem. When a file is
opened, its inode is brought into main memory and stored
in a memory-resident inode table.
[classbook at pag.700]
2
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Information in the inode
3
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Directories

In Unix a directory is simply a file that contains a list of file
names plus pointers to associated inodes
Inode table
Directory
i1
Name1
i2
Name2
i3
Name3
i4
Name4
…
…
4
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Directories
DIR *opendir(const char *name);

The opendir() function opens a directory stream
corresponding to the directory name, and returns a pointer
to the directory stream. The stream is positioned at the
first entry in the directory.
5
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Directories
struct dirent *readdir(DIR *dir);

The readdir() function returns a pointer to a dirent structure
representing the next directory entry in the directory stream
pointed to by dir. It returns NULL on reaching the end-offile or if an error occurred. The dirent structure is defined as
follows:
struct dirent {
ino_t
d_ino;
char
d_name[256];
…
};
/* inode number */
/* filename */
6
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Directories
// error handling is not included!
struct dirent *entry;
DIR *dirp;
dirp = opendir(".");
while((entry = readdir(dirp)) != NULL)
printf("%s\n",entry->d_name);
closedir(dirp);
7
Copyright ©: Nahrstedt, Angrave, Abdelzaher
readdir() is not thread safe


readdir() returns a pointer to a statically allocated structure;
hence the returned pointer points to data which may be
overwritten by another call to readdir() on the same directory
stream.
The data is not overwritten by another call to readdir() on a
different directory stream.
pointer to storage
allocated by the user
The reentrant version of readdir() is:
int readdir_r(DIR *dirp, struct dirent *entry, struct dirent
**result);

8
Copyright ©: Nahrstedt, Angrave, Abdelzaher
UNIX file structure implementation




Each user has a "file descriptor table“. Each entry in the "file
descriptor table" is a pointer to an entry in the system-wide
"open file table"
Each entry in the open file table contains a file offset (file
pointer) and a pointer to an entry in the "memory-resident inode table"
If a process opens an already-open file, a new open file table
entry is created (with a new file offset), pointing to the same
entry in the memory-resident i-node table
If a process forks, the child gets a copy of the "file descriptor
table" (and thus the same file offset)
9
Copyright ©: Nahrstedt, Angrave, Abdelzaher
UNIX file structure implementation
[figure from http://pages.cs.wisc.edu/~swift/classes/cs537-sp09/lectures/14-unix-fs.pdf]
10
Copyright ©: Nahrstedt, Angrave, Abdelzaher
File opening


A call to open() creates a new entry (open file description) in
the system-wide table of open files. This entry records the file
offset and the file status flags.
A file descriptor is a reference to one of these entries. The new
open file description is initially not shared with any other
process, but sharing may arise via fork.
11
Copyright ©: Nahrstedt, Angrave, Abdelzaher
File Allocation on Disk

Low level access methods for a file depend
upon the disk allocation scheme used to
store file data



Contiguous
Linked list
Block or indexed
12
Copyright ©: Nahrstedt, Angrave, Abdelzaher
#1. Contiguous Allocation
13
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Contiguous Allocation Issues




Access method suits sequential and direct access
Directory table maps files into starting physical
address and length
Easy to recover in event of system crash
Fast, often requires no head movement and when it
does, head only moves one track
14
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Contiguous Allocation Issues




File is allocated large contiguous chunks
Expanding the file requires copying
Dynamic storage allocation - first fit, best fit
External fragmentation occurs on disk
15
Copyright ©: Nahrstedt, Angrave, Abdelzaher
External Fragmentation
Solution: Linked allocation
16
Copyright ©: Nahrstedt, Angrave, Abdelzaher
#2. Linked Allocation
17
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Linked List Allocation





Each file is a linked list of nodes
Pointers in list are not accessible to user
Directory table maps files into head of list for a file
A node in the list can be a fixed size physical block
or a contiguous collection of blocks
Easy to use - no estimation of size necessary
18
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Linked List Allocation




Can grow in middle and at ends
Space efficient, little fragmentation
Slow - defies the principle of locality. Need to read
through linked list of nodes sequentially to find the
needed blocks of data
Suited for sequential access but not direct access
19
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Linked List Allocation Issues



Disk space must be used to store pointers (if disk
block is 512 bytes, and disk address requires 4
bytes, then the user sees blocks of 508 bytes)
Not very reliable. System crashes can scramble
files being updated
Important variation on linked allocation method:
`file-allocation table' (FAT) - OS/2 and MS-DOS
20
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Linked List Allocation Issues


Summary: linked allocation solves the
external fragmentation and sizedeclaration problems of contiguous
allocation,
However, it can't support efficient direct
access
21
Copyright ©: Nahrstedt, Angrave, Abdelzaher
#3. Indexed Allocation
22
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Indexed Allocation



Solves external fragmentation
Supports sequential, direct and indexed
access
Access requires at most one access to
index block first. This can be cached in
main memory
23
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Indexed Allocation


Requires extra space for index block,
possible wasted space
How to extend to big files?

A file can be extended by using linked
indexed files or multilevel indexed files
24
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Linked Indexed Files
Link full index
blocks together
using last entry.
25
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Multilevel Indexed File
Multiple levels of index blocks
26
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Layout of a UNIX file on disk


File allocation is done on a block basis and allocation is
dynamic (as needed).
UNIX uses a multilevel indexing mechanism for file allocation
on disk. Addresses of first 10 data blocks + 3 index blocks
(first, second, and third level of indexing)

In UNIX System V the length of a block is 1 Kbyte and each
block can hold a total of 256 block addresses

According to above parameters, maximum size for a file is
slightly over 16Gbytes
27
Copyright ©: Nahrstedt, Angrave, Abdelzaher
Layout of a UNIX file on disk
First ten addresses
point to the first 10
data blocks of the file
The inode includes 39 bytes
of address information that
is organized as thirteen 3-byte
addresses
28