The archived chapter discusses how an already-linked program is placed in memory and prepared to run. It compares ordinary file reads with virtual-memory mapping, then describes relocation when a program cannot use its assumed addresses.
A substantial section explains position-independent code: keeping executable instructions shareable while handling process-specific data separately. Historical examples include pointer tables, the AIX table of contents and the ELF global offset table. The discussion weighs extra indirection and calling overhead against the ability to share code between processes.
The overlay discussion describes fitting larger programs into limited memory by loading different code segments as needed. It explains how call relationships constrain the layout and how linker-generated glue coordinates loading.
This page is a short historical summary, not a reproduction of the chapter. Its architecture examples reflect the manuscript’s period and are not a compatibility guide for current systems. The original diagrams and code examples are not reproduced here.