Call Stack - Description

Description

Since the call stack is organized as a stack, the caller pushes the return address onto the stack, and the called subroutine, when it finishes, pops the return address off the call stack and transfers control to that address. If a called subroutine calls on to yet another subroutine, it will push another return address onto the call stack, and so on, with the information stacking up and unstacking as the program dictates. If the pushing consumes all of the space allocated for the call stack, an error called a stack overflow occurs, generally causing the program to crash. Adding a subroutine's entry to the call stack is sometimes called winding; conversely, removing entries is unwinding.

There is usually exactly one call stack associated with a running program (or more accurately, with each task or thread of a process), although additional stacks may be created for signal handling or cooperative multitasking (as with setcontext). Since there is only one in this important context, it can be referred to as the stack (implicitly, "of the task"); however, in the Forth programming language the data stack or parameter stack is accessed more explicitly than the call stack and is commonly referred to as the stack (see below).

In high-level programming languages, the specifics of the call stack are usually hidden from the programmer. They are given access only to a set of functions, and not the memory on the stack itself. Most assembly languages, on the other hand, require programmers to be involved with manipulating the stack. The actual details of the stack in a programming language depend upon the compiler, operating system, and the available instruction set.