Polling Loops and Interrupts

Polling loops and interrupts are both ways of dealing with events that do not fit neatly into the ordinary flow of a program. An interrupt, in particular, is a way of handling an event that arrives outside the CPU's regular instruction sequence. But you cannot understand an interruption until you have a clear picture of the sequence being interrupted. This article builds that picture. It covers what the CPU normally does, where the instructions it runs actually live, and why the ordinary path is the foundation for everything that follows.

Because the CPU works with patterns rather than explanations, a single pattern has to carry enough structure for the CPU to tell what it should do. A useful way to picture this is to think of the pattern as being split into fields. One field indicates the operation, and other fields supply the data the operation works on. The source material refers to these parts as an operation code and operands. The diagram below shows the idea using an invented pattern. It is not a real instruction set, and the split shown is only there to make the concept concrete.

Before the CPU can run anything, the instructions have to be somewhere it can reach. The source material separates two roles. Main memory is the temporary execution workspace, the place where a running program's instructions are held while the CPU works through them. Storage devices, such as hard disks and solid state drives, provide permanent storage. They keep information even when the machine is switched off, and they offer room for much larger amounts of information than main memory does.

The practical consequence is that a program moves through a fixed route. It is stored on storage, loaded into main memory, and only then executed. The ordinary path always runs through main memory, which is why the source material treats main memory as central to the CPU's behavior.

The CPU's normal behavior is a sequential fetch-and-execute process. Think of it as a repeating loop with two stages. In the fetch stage, the CPU retrieves an instruction from main memory. In the execute stage, it carries out the operation that the instruction describes. Once that instruction has been executed, the cycle moves on to the next instruction in the sequence and repeats. The diagram below shows this loop. Notice that it is a loop in which each step leads to the next, not a set of separate actions.

An interrupt is studied as a way of handling events that do not fit into the CPU's ordinary instruction sequence. To see why it counts as a departure, you need the ordinary sequence firmly in mind. The fetch-execute cycle is the steady path, and main memory is where that path runs. An event that arrives from outside this path has to be understood relative to it.