You might reasonably be wondering why exec and its brethren are implemented in this counterintuitive way. You might also be wondering how a UNIX shell manages to continue operating during and after it launches a command.
We can start to work toward a solution by more closely parsing two words from the manual excerpt I quoted earlier: program and process. Although I’ve tried to use both terms precisely thus far, we don’t have a proper working definition of either yet, but now we’re ready to take a proper stab at it.
A program is a file on disk that can be executed; it’s a series of machine instructions loadable by our operating system. A process is an instance of a program that’s in the process of execution (for example, its stack, its heap, and the loaded contents of the program text) and the instruction pointer, which is the current position in the program.
What we want to do is run execve in a new process so that our code can continue whatever it needs to do. However, since processes are executed separately by the operating system, we also need some way to wait for our new process to complete before we continue. In other words, once we’re concerned with multiple processes, we have to deal with concurrency.