Lecture: Structures of Operating Systems

Introduction

• Discussion on operating system structures.
• Importance of engineering operating systems for functionality and easy modification.
• Historical evolution of OS structures.

Simple Structure

• Used by: Early operating systems, e.g., MS-DOS.
• Characteristics:
  • Not well-defined structure.
  • Layers include ROM/BIOS, device drivers, system programs, and application programs.
  • Direct access to base hardware by all upper layers.
• Disadvantages:
  • Vulnerable to system crashes due to errant programs accessing hardware.
  • Lack of hardware protection and dual-mode in early processors like Intel 8088.

Monolithic Structure

• Used by: Early Unix systems.
• Characteristics:
  • Kernel contains all functionalities (signal handling, file systems, CPU scheduling, etc.).
  • Users and system call interfaces above kernel, hardware below.
• Disadvantages:
  • Difficult to maintain and debug due to packed functionalities.
  • Changes require modifying the entire kernel.

Layered Structure

• Characteristics:
  • Divides OS into layers (hardware at layer 0, user interface at the topmost layer).
  • Functions are separated into different layers.
• Advantages:
  • Easier to implement and debug.
  • Hardware protection from upper layers.
• Disadvantages:
  • Complex to design specific layer order.
  • Efficiency loss due to layered communication (service requests passing through multiple layers).

Microkernels

• Characteristics:
  • Minimizes kernel by removing non-essential components.
  • Core functionalities remain in the kernel; other services implemented as system programs.
  • Communication through message passing.
• Advantages:
  • Core functionalities in user mode minimize system crashes.
• Disadvantages:
  • Potential performance decrease due to overhead from frequent message passing.

Modular Structure

• Current best methodology for OS design.
• Characteristics:
  • Core kernel with essential functionalities.
  • Use of loadable modules for additional functionalities (device drivers, file systems, etc.).
• Advantages:
  • Flexible, defined interfaces.
  • Modules can communicate directly, unlike layered systems.
  • Avoids message passing overhead seen in microkernels.
  • Combines benefits of both layered and microkernel approaches.