Volume 2: The Virtual Memory System

virtual memory system AUTHOR William F. Jolitz / Lynne Greer Jolitz
PAGES 400 pages
TYPE Hardcover
ISBN 1-57398-027-7
PUBLISHED Due Date Uncertain
PRICE To Be Announced

Email us at info@peerllc.com to let us know which book or books in the series you would like to be notified about upon publication.


Highlights

  • Extensively describes a modified, thoroughly documented implementation of the Mach virtual memory system integrated into the x86 BSD environment.
  • Covers topics such as X86 mmu control, copy on write, mapping, swapping, paging, and fault handling.
  • Emphasizes themes of system performance, clustering, security, and extensibility in all subject treatments.
  • Unix Review named Volume 1: The Basic Kernel a “Top 10 Book of the Year” in 1996.

About Virtual Memory Source Code Secrets

The “brain” of an operating system is the virtual memory system. Like a human brain, the virtual memory system is the means through which temporary and permanent memory is stored and accessed. Temporary memory is used as a workspace to assemble a new item before it is stored into permanent memory. Among these most fundamental of functions are:

  • Management of Kernel and Program Virtual Address Spaces: Creation, Replication, Modification and Destruction
  • Management of Store Backing an Address Space: Naming, Protection, Sharing, and Inheritance of Memory Contents
  • Persistent and Temporary Memory
  • Machine Dependent (X86) Memory Management Unit (MMU) Control
  • Copy-on-write Semantics
  • Main Store (RAM) and Secondary Store (swap) Management
  • Page Replacement Algorithm
  • Memory-Mapped I/O (mmap)
  • Address Space Fault Handling
  • Demand Loading from Secondary Store
  • Lazy versus Industrious Evaluation
  • UNIX and BSD API’s for Address Space Control

The virtual memory module is not only a key subsystem of the operating system kernel but is itself composed of layers and levels of abstraction. These layers are discussed in terms of future evolution to support the needs of:

  • clustering
  • security
  • extensibility
  • performance

About the Authors

William F. Jolitz and Lynne Greer Jolitz have been principals in a number of Silicon Valley start-ups and have written over 40 feature articles on operating system and network design for major computer magazines, including an unprecedented 17-part serialization of the 386BSD project for Dr. Dobb’s Journal. William was one of the original architects of Berkeley UNIX. Lynne has been involved in the 386BSD project since its inception and has worked with technical operating system and data center issues for over thirty years.