Control

A bit pattern that is fetched as an instruction is interpretted as an instruction. The bits determine what is done in the next machine cycle. If the pattern makes no sense as an instruction then normal control flow is interrupted. If the pattern can be interpretted as an instruction then it will be executed, whatever it does.

The control point of an executing program is the address in memory of the instruction being executed. When an instructiion is being executed (in the third step of the machine cycle) the program counter holds the address of the instruction after the control point.

Normally the control point moves sequentially through the machine instructions. On the MIPS this means it normally moves through memory in steps of four bytes (32 bits) at a time. Usually "control point" is shortened to "control" and the phrase flow of control means how the control point moves through memory.

If control flow leads to an address in memory, then the four bytes starting at that address are fetched as a machine instruction. The processor has no other way to tell instructions from data. Whatever bit pattern gets pulled in from memory as an instruction will be executed as an instruction. It is common for the control point of a buggy program to enter a section of data. This sometimes leads to mystifying results.

By software convention, data and instructions are placed in different sections of memory. (This helps prevent mystifying results). But this is not a requirement of the architecture.