Chapter 3

Chapter 3: Operating-System Structures
􀂄 System Components
􀂄 System Calls
􀂄 System Programs
􀂄 System Structure

Common System Components
􀂄 Process Management
􀂄 Main-Memory Management
􀂄 File Management
􀂄 I/O-System Management
􀂄 Secondary-Storage Management
􀂄 Networking
􀂄 Protection System
􀂄 Command-Interpreter System

Process Management
􀂄 A process is a program in execution.
􀂄 A process needs resources - CPU time, memory, files,and I/O devices
􀂄 The operating system is responsible for
􀀩 Process creation and deletion.
􀀩 Process suspension and resumption.
􀀩 Provision of mechanisms for:
􀀗 process synchronization
􀀗 process communication
􀀗 deadlock handling
􀂄 Ch4 - Ch8

Main-Memory Management
􀂄 A program must be loaded into memory to be executed
􀂄 Keep multiple processes in memory
􀀩 Improve CPU utilization
􀀩 Improve response speed to users
􀂄 The operating system is responsible for:
􀀩 Keeping track of who is using what parts of memory
􀀩 Deciding which processes to load when memory space becomes available.
􀀩 Allocating and de-allocating memory space as needed.
􀂄 Ch9, Ch10

File Management
􀂄 A file is a collection of related information defined by its creator.
􀂄 OS maps files onto physical media and accesses files via the devices that control the physical media
􀂄 OS is responsible for:
􀀩 File creation and deletion.
􀀩 Directory creation and deletion.
􀀩 Supporting primitives for manipulating files and directories.
􀀩 Mapping files onto secondary storage.
􀀩 Backing up files on nonvolatile storage media
􀀩 File protection: who can access a file in what ways
􀂄 Ch11, Ch12

I/O-System Management
􀂄 Manage and control I/O operations and I/O devices
􀂄 The I/O subsystem consists of:
􀀩 A general device-driver interface
􀀩 Drivers for specific hardware devices
􀀩 Buffering, caching, and spooling
􀂄 Ch13 (skip)

Secondary-Storage Management
􀂄 The operating system is responsible for:
􀀩 Free space management
􀀩 Storage allocation
􀀩 Disk scheduling
􀂄 Ch14 (skip)

Networking
􀂄 Processors in a distributed system are connected through a communication network.
􀂄 Communication among processors takes place using a set of protocols implemented by OS.
􀂄 Ch15 - Ch17

Protection System
􀂄 Protection refers to a mechanism for controlling access by processes or users to system resources.
􀂄 Ch18

Command-Interpreter System
􀂄 Command interpreter (or shell): get the next command statement and execute it
􀀩 Graphical shell: Microsoft Windows and Mac OS
􀀩 Command-line shell: UNIX
􀂄 Can be part of kernel or just an application program

System Calls
􀂄 System calls provide the interface between a process and the operating system.
􀀩 Usually available as assembly-language instructions.
􀀩 Some languages allow system calls to be made directly (e.g., C, C++)
􀂄 Three methods to pass parameters to OS:
􀀩 Pass parameters in registers.
􀀩 Store the parameters in a table in memory, the table address is passed as a parameter in a register.
􀀩 Parameters are pushed onto the stack by the program, and popped off the stack by operating system

Categories of System Calls
􀂄 Process control: load & execute, create, end, abort, get/set process attributes, wait, etc.
􀂄 File management: open, close, create, delete, read, write, reposition, get/set file attributes, etc.
􀂄 Device management: request, release, read, write, reposition, get/set file attributes, etc.
􀀩 UNIX: devices handled as virtual files, identified by special file names
􀂄 Information maintenance: get/set time or date, get/set system data, get/set process/file/device attributes, etc.
􀂄 Communications: open/close connection, send/receive message, etc.

Communication Models
􀂄 Message-passing model: processes exchange info though an interprocess-communication facility provided by OS
􀀩 Open connection before communication
􀀩 Communication done by exchanging messages
􀀩 Close connection after communication
􀀩 Easier to implement for intercomputer communication
􀂄 Shared-memory model: processes exchange info by reading/writing data in shared areas in memory
􀀩 Fast and convenient within a computer
􀀩 Must ensure processes do not write to the same location simultaneously

System Programs
􀂄 System programs provide a convenient environment for program development and execution.
􀂄 Categories of system programs:
􀀩 File management: cp, ls, rm
􀀩 Status information: df, date
􀀩 File modification: vi, pico, emacs
􀀩 Programming language support: compilers, assemblers,interpreters
􀀩 Program loading and execution: linkers, loaders, debuggers
􀀩 Communications: web browsers, email, remote login
􀂄 System programs define the user interface

Shell
􀂄 A system program that gets and executes the next user specified command
􀂄 Two ways to implement commands
􀀩 Shell contains the code to execute the command
􀀗 Size of shell determined by number of commands
􀀗 Difficult to add a new command
􀀩 Commands are system programs
+ Shell can be small
+ Can easily add commands
- Slower
- Need pass parameters from shell to the system program
- Interpretation of parameters may be inconsistent
􀂄 Unix: some commands are built in, the rest are system programs

System Structure
􀂄 Simple structure
􀀩 MS-DOS, Unix
􀂄 Layered approach
􀀩 Windows NT, OS/2
􀂄 Microkernels
􀀩 Mach, MacOS X server

UNIX System Structure
􀂄 OS consists of two parts:
􀀩 System programs
􀀩 The kernel
􀀗 Consists of everything below the system-call interface and above the physical hardware
􀀗 Provides the file system, CPU scheduling, memory management, I/O, etc.
􀂄 A large number of functions in one level􀃎difficult to enhance


Layered Approach
􀂄 The operating system is divided into a number of layers
􀀩 The bottom layer (layer 0) is the hardware
􀀩 The highest (layer N) is the user interface
􀂄 Each layer uses functions and services of only lower-level layers.
􀂄 Benefits:
􀀩 Simplify debugging and system verification: can debug one layer at a time
􀀩 Can change the implementation of a layer without affecting other layers, provided that the interface and functions of the layer do not change
􀂄 Problems:
􀀩 Design of layers can be tricky
􀀩 Less efficient: each layer may add overhead to a system call

Microkernel System Structure
􀂄 Remove all non-essential components from kernel, implementing them as services run in user space
􀂄 Usually, microkernel provides:
􀀩 Minimal process and memory management
􀀩 Communication facility
􀂄 Communication between client program and various services provided by message passing
􀀩 System calls become messages sent via kernel to appropriate service
􀂄 Benefits:
􀀩 Adding services doesn’t require kernel modification
􀀩 Kernel codifications simpler when required
􀀩 More secure: less code is running in kernel mode
􀀩 More reliable: kernel OK when a service fails
􀂄 Examples: Mach, MacOS X server