L33_Ch12d

Chapter 12: File Management (Part D: Sections 12.8-End)

Applications of General Concepts in Various Operating Systems

Page 1: Overview

• This section covers file management principles as they pertain to various operating systems.

Page 2: UNIX File Management

Types of Files in UNIX

• Regular files: Contains arbitrary data in zero or more data blocks.

• Directory files: Contains a list of file names and pointers to associated INodes (index nodes).

• Special files: Provide mechanisms to map physical devices to file names.

• Named pipes: An inter-process communication facility.

• Links: Alternative file names for existing files.

• Symbolic links: A data file that points to another file name.

• Note: Remember the Producer-Consumer Problem!

Page 3: INodes in UNIX

• INodes administer all UNIX file types.

• INode Definition: A control structure containing the essential information for a file.

• Multiple file names can link to a single INode.

• Every active INode is related to only one file, while each file is controlled by just one INode.

Page 4: Structure of INodes

• INodes contain attributes such as:

  • Mode

  • Owner

  • Timestamps

  • Size

  • Reference counts

  • Pointers indicating data locations in blocks.

• Each INode also tracks various types of data and location pointers.

Page 5: File Allocation in UNIX

• Dynamic Allocation: File allocation occurs based on need and not pre-allocated.

• Indexed Method: Utilized to track files with references in the INode.

• All UNIX implementations include direct and indirect pointers in the INode (single, double, triple).

• Look into the FreeBSD variant for implementation details.

Page 6: Capacity of a FreeBSD File

• Table of block size and capacity for FreeBSD:

  • Direct: 12 blocks = 48K bytes

  • Single Indirect: 512 blocks = 2M bytes

  • Double Indirect: 512 x 512 = 256K blocks = 1G bytes

  • Triple Indirect: 512 x 256K = 128M blocks = 512G bytes

Page 7: UNIX Directories and INodes

• INode Table Example:

  • Directory mapping, showing file names and their associated INodes.

Page 8: Volume Structure in UNIX

• The UNIX file system resides on a logical disk laid out with the following:

  • Boot Block: Contains OS boot code.

  • Superblock: Contains file system attributes.

  • INode Table: Collection of INodes for each file.

  • Data Blocks: Storage for data files and subdirectories.

Page 9: LINUX Virtual File System (VFS)

• Components of LINUX VFS:

  • Individual file systems, device drivers, buffer cache, system call interface, user applications.

Page 10: VFS Concept in LINUX

• Illustrates mapping of user processes to secondary storage through system calls and disk I/O calls using VFS system calls.

Page 11: Primary Object Types in VFS

• DEntry Object: Represents a directory entry.

• File Object: Represents an open file associated with a process.

• Superblock Object: Represents a mounted file system.

• INode Object: Represents a specific file.

• Implementation follows Object-Oriented principles.

Page 12: The Superblock Object

• Functionality: Stores information about the specific file system.

• Attributes include: device, block size, dirty flag, file system type, open files list, and mutual exclusion semaphores.

Page 13: The INode Object

• Holds information on file characteristics excluding the file name and content.

• Attributes might include owner, permissions, access times, size, and number of links.

Page 14: The DEntry Object

• Purpose: Represents components in a path for faster lookup.

• Links to INodes and superblocks, including parent DEntry pointers.

Page 15: The File Object

• Represents an open file by a process.

• Created via open() and destroyed via close().

• Methods include read, write, open, release, and lock, along with various informational items.

Page 16: Caches in VFS

• INode Cache: Stores recent INodes for quicker access.

• Directory Cache: Links directory names with their INode numbers to speed listings.

• Buffer Cache: Optimizes data buffer management, reducing disk access frequency.

Page 17: Windows File System - NTFS

• Designed for high-end workstation and server requirements.

• Key features include recoverability, security, handling large disks/files, multiple data streams, journaling, compression, and linking.

Page 18: NTFS Volume and File Structure

• Storage Concepts:

  • Sectors: Smallest physical unit (512 bytes).

  • Clusters: Contiguous sectors; power of 2.

  • Volumes: Logical partitions of one or more clusters. Maximum volume size is 2^64 clusters.

Page 19: NTFS Partition and Cluster Sizes

• Table outlining sector and cluster sizes corresponding to various volume sizes, highlighting the relationship between cluster count and byte size.

Page 20: NTFS Volume Layout

• Diagram representing how NTFS structures volume with system files, boot sector, and Master File Table.

Page 21: Master File Table (MFT)

• Central to the Windows file system; a structured table containing records describing each file.

• Distinction between MFT records and actual data file records highlighted.

Page 22: NTFS File and Directory Attribute Types

• Attributes of files/folders include standard info, attribute lists, file names, security descriptors, data contents, index roots, and bitmap records.

• Note distinctions between required and optional attributes.

Page 23: NTFS Components

• Overview of the integrated functionality of various components such as the I/O Manager, Cache Manager, and disk drivers for system performance and fault tolerance.

Page 24: ANDROID File Structure

• Typical directory tree layout for Android systems, showcasing root and mounted storage points.

Page 25: Database Management Interface

• User perspective focuses on data abstraction rather than physical files.

• Overview of the interaction between DBMS and files, alongside the role of the filesystem and storage drivers.