Chapter 2: Operating System Overview
1. Operating System Objectives and Functions
Def: An Operating System is a program that controls the execution of application programs and acts as an interface between applications and the computer hardware. It has three main objectives:
- Convenience (方便): Makes a computer more convenient to use for the end user,.
- Efficiency (高效): Allows the computer system resources to be used in an efficient manner,.
- Ability to evolve (可演进能力): Permits the effective development, testing, and introduction of new system functions without interfering with service,.
1.1 The OS as a User/Computer Interface
The OS masks the details of the hardware from the programmer and provides a convenient interface. It typically provides services in the following areas:
- Program development: Provides tools like editors and debuggers (often via utilities).
- Program execution: Handles scheduling duties, loads instructions/data into main memory, and initializes files and I/O devices.
- Access to I/O devices: Hides specific hardware details and provides a uniform interface for reading and writing.
- Controlled access to files: Understands storage media and data structures, and provides protection mechanisms in multi-user environments.
- System access: Resolves resource conflicts and protects resources/data from unauthorized access.
- Error detection and response: Responds to internal/external hardware errors or software errors with minimal impact on running applications-.
- Accounting (记账): Collects usage statistics to tune system performance and bill users.
- It provides an Application Programming Interface (API, 应用程序编程接口), allowing software to be ported easily across systems-.
1.2 The OS as Resource Manager
The OS is responsible for controlling the use of a computer's resources (I/O, main/secondary memory, processor execution time).
- Unlike typical control mechanisms, the OS functions as an ordinary computer program executed by the processor.
- It frequently relinquishes control to let the processor perform "useful" work and depends on the processor to regain control.
- The Kernel (内核), or nucleus, contains the most frequently used OS functions and typically remains permanently in main memory,-.
1.3 Ease of Evolution
An OS will evolve over time due to hardware upgrades, new types of hardware, new services, and software fixes-. This requires the OS to be modular in construction with clearly defined interfaces.
2. The Evolution of Operating Systems
2.1 Serial Processing (串行处理)
In early computers (late 1940s to mid-1950s), there was no OS; programmers interacted directly with hardware.
- Scheduling: Users used a hardcopy sign-up sheet, which often resulted in wasted computer processing time-.
- Setup time: A single job (作业) involved mounting tapes and setting up card decks manually, causing a considerable waste of time.
2.2 Simple Batch Systems (简单批处理系统)
To maximize processor utilization, early batch OSs used a Monitor (监控程序)-.
- Users submitted jobs on cards/tape to an operator, who batched them together on an input device for the monitor.
- The monitor automatically loaded the next program after a job was completed, avoiding idle time-.
- Job Control Language (JCL, 作业控制语言) was used to provide instructions to the monitor.
- Hardware Features Required: Memory protection (内存保护), Timer (定时器), Privileged instructions (特权指令), and Interrupts (中断).
- Execution modes were split into User mode (用户态) (with restricted memory and instruction access) and Kernel mode / System mode (内核态/系统态) (for the monitor to execute privileged instructions),.
2.3 Multiprogrammed Batch Systems (多道批处理系统)
In a simple batch system, the processor is often idle waiting for slow I/O devices.
- Multiprogramming / Multitasking (多道程序设计/多任务处理): Expands memory to hold multiple programs; when one job waits for I/O, the processor switches to another job,.
- This relies on hardware supporting I/O interrupts and Direct Memory Access (DMA, 直接内存访问).
- It maximizes processor utilization but requires advanced memory management and scheduling algorithms,.
2.4 Time-Sharing Systems (分时系统)
Used to handle multiple interactive jobs by interleaving execution in short bursts or time slices.
- Multiple users simultaneously access the system through terminals.
- While multiprogramming maximizes processor use, time-sharing aims to minimize Response time (响应时间),.
- Examples include the Compatible Time-Sharing System (CTSS).
3. Major Achievements
3.1 The Process (进程)
A Process (进程) is defined as a program in execution, an instance of a running program, or a unit of activity characterized by a single sequential thread of execution, a current state, and an associated set of system resources,.
- It consists of three components: an executable program, associated data, and the Execution context (执行上下文) (or Process state (进程状态))-,-.
- The execution context includes contents of processor registers and information the OS needs to manage the process, allowing the OS to suspend and resume the process correctly,.
- Thread (线程): A single process can be broken up into multiple, concurrent threads that execute cooperatively.
3.2 Memory Management (内存管理)
The OS has five principal storage management responsibilities-,:
- Process isolation (进程隔离): Preventing independent processes from interfering with each other's memory,.
- Automatic allocation and management: Dynamically allocating memory across the hierarchy transparently to the programmer,.
- Support of modular programming: Allowing dynamic creation and alteration of program modules,.
- Protection and access control: Permitting shared memory while protecting the integrity of programs/OS,.
- Long-term storage: Storing information persistently (e.g., via file systems),.
- Virtual memory (虚拟内存): Allows programs to address memory logically without worrying about physical limitations. It maps a Virtual address (虚拟地址) to a Real address / Physical address (实地址/物理地址) dynamically-,.
3.3 Information Protection and Security (信息保护与安全)
Security policies generally target four categories:
- Availability (可用性): Protecting against interruption.
- Confidentiality (机密性): Assuring unauthorized users cannot read data.
- Data integrity (数据完整性): Protecting data from unauthorized modification.
- Authenticity (真实性): Verifying user identity and message validity.
3.4 Scheduling and Resource Management (调度和资源管理)
The OS must schedule process execution and allocate resources based on three factors-:
- Fairness (公平性): Giving processes equal and fair access to resources.
- Differential responsiveness (差异响应性): Discriminating among different classes of jobs to meet different service requirements dynamically.
- Efficiency (高效): Maximizing throughput, minimizing response time, and accommodating the maximum number of users.
4. Developments Leading to Modern Operating Systems
- Microkernel architecture (微内核架构): Assigns only a few essential functions (e.g., address space management, Interprocess Communication (IPC), basic scheduling) to the kernel,. Other services run in user mode, unlike a Monolithic kernel (单体内核) where all elements share the same address space,.
- Multithreading (多线程): A technique where a process is divided into threads that run concurrently,. A thread is a dispatchable unit of work; a process is a collection of threads and associated resources.
- Symmetric Multiprocessing (SMP, 对称多处理): The OS schedules threads across multiple processors, offering advantages in Performance, Availability, Incremental growth, and Scaling-.
- Distributed Operating Systems (分布式操作系统): Provides the illusion of a single main and secondary memory space across a cluster of separate computers.
- Object-oriented design (面向对象设计): Allows programmers to customize the OS without disrupting system integrity.
5. Fault Tolerance (容错)
Fault tolerance (容错) is the ability of a system to continue normal operations despite the presence of hardware or software faults. This involves redundancy to increase reliability, balancing the costs against the critical nature of the resources.
6. OS Design Considerations for Multiprocessor and Multicore
6.1 SMP OS Considerations
An SMP OS must manage processor and computer resources to provide the illusion of a uniprocessor system. Key design issues include-,:
- Simultaneous concurrent processes or threads,.
- Scheduling,.
- Synchronization,.
- Memory management,.
- Reliability and fault tolerance,.
6.2 Multicore OS Considerations
With "many-core" systems, the challenge is extracting parallelism from workloads-.
- Parallelism within Applications: Developers divide applications into parallel tasks.
- OS approaches like Apple's Grand Central Dispatch (GCD) map tasks onto a pool of available threads.
- Future architectures might drop the distinction between user and kernel mode, treating the OS more like a hypervisor while applications self-manage resources-.