Answer 2—Unfortunately, there is no longer a single answer to this question.

The terms were first used in the late 1970s during the days of the IBM mainframe operating systems for the System/360 and System/370. These were batch-processing systems that divided the available resources (mainly memory) into partitions, usually three to four foreground partitions that had the highest priority levels, and a background partition that had the lowest priority. You’d put I/O-bound jobs in the high-priority partitions and compute-bound tasks in the background partition. By extension, because interrupt service routines (ISRs) have the highest priority by definition, they are the “most foreground” code.

Things got really confused when the IBM PC and primitive (single-task) operating systems (e.g., MSDOS) became popular. User code was always operating in Non-interrupt mode (background), but this was in the foreground in terms of the user interface. People started writing TSR utilities that were hooked into one interrupt or another (usually the timer tick) to do tasks that were in the background in terms of the user interaction. As a result, people starting to think of interrupt code in general as being background code and, therefore, the noninterrupt code as foreground code.

These days, it’s probably better to explicitly describe code as “ISR” or “non-ISR.”

Contributor: David Tweed

Published December 2003