File Allocation and Directory Implementation Techniques in Operating Systems - Prof. Jeffr, Study notes of Operating Systems

Download File Allocation and Directory Implementation Techniques in Operating Systems - Prof. Jeffr and more Study notes Operating Systems in PDF only on Docsity!

Announcements

z

Reading Chapter 12

Allocation Methods

z^

How do we select a free disk block to use? z^

Contiguous allocation–

allocate a contiguous chunk of space to a file

-^

directory entry indicates the starting block and the length of the file

-^

easy to implement, but

-^

how to satisfy a given sized request from a list of free holes?

-^

two options

• first fit (find the first gap that fits)– best fit (find the smallest gaps that is large enough) -^

What happens if one wants to append to file?

-^

from time to time, one will need to repack files

Indexed Allocation

z^

Bring all pointers together in an

index block

-^

Each file has its own index block -

i th entry of index block points to

i th block making up the file

z^

How large to make an index block?–

To avoid a fixed maximum file size, index block must be extensible

z^

Linked scheme:–

maintain a linked list of indexed blocks

z^

Multilevel index:–

Index block can point to other index blocks (which point to indexblocks ....), which point to files

z^

Hybrid multi-level index–

first n blocks are from a fixed index

-^

next m blocks from an indirect index

-^

next o blocks from a double indirect index

Hybrid Multi-level Index (UNIX)

z

Observations– most files are small– most of the space on the disk is consumed by large files

z

Want a flexible way to support different sized– assume 4096 byte block– first 12 blocks (48 KB) are from a fixed index– next 1024 blocks (4 MB) from an indirect index– next 1024

2

blocks (4 GB) from a double indirect index

• final 1024

3

blocks (4 TB) from a triple indirect index

directory entry

IndirectIndex

double indirect index

Performance Issues

z

FAT^ 

simple, easy to implement 

faster to traverse than linked allocation

• random access requires following links– files can’t have holes in them

z

Hybrid indirect^ 

fast access to any part of the file 

files can have holes in them

• more complex

Free Space Management

z

How do we find a disk block to allocate?

z

Bit Vectors– array of bits (one per block) that indicates if a block is free– compact so can keep in memory

• 100 GB disk, 4K blocks -> 6MB per disk (0.003%)
  • easy to find long runs of free blocks

z

Linked lists– each disk block contains the pointer to the next free block– pointer to first free block is keep in a special location on disk

z

Run length encoding (called counting in book)– pointer to first free block is keep in a special location on disk– each free block also includes a count of the number of

consecutive blocks that are free

DOS Directories

z

Root directory– immediately follows the FAT

z

Directory is a table of 32 byte entries– 8 byte file name, 3 byte filename extension– size of file, data and time stamp, starting cluster number of

the file, file attribute codes

• Fixed size and capacity

z

Subdirectory– This is just a file– Record of where the subdirectory is located is stored in the

FAT