Chapter 9: Intermediate Memory Topics

How a Computer Views Memory

Let's review how memory within a computer works. You may also want to re-read Chapter 2.

A computer looks at memory as a long sequence of numbered storage locations. A sequence of millions of numbered storage locations. Everything is stored in these locations. Your programs are stored there, your data is stored there, everything. Each storage location looks like every other one. The locations holding your program are just like the ones holding your data. In fact, the computer has no idea which are which, except that the executable file tells it where to start executing.

These storage locations are called bytes. The computer can combine up to four of them together into a single word. Normally numeric data is operated on a word at a time. As we mentioned, instructions are also stored in this same memory. Each instruction is a different length. Most instructions take up one or two storage locations for the instruction itself, and then storage locations for the instruction's arguments. For example, the instruction

 movl data_items(,%edi,4), %ebx

takes up 7 storage locations. The first two hold the instruction, the third one tells which registers to use, and the next four hold the storage location of data_items. In memory, instructions look just like all the other numbers, and the instructions themselves can be moved into and out of registers just like numbers, because that's what they are.

This chapter is focused on the details of computer memory. To get started let's review some basic terms that we will be using in this chapter:

Byte

• This is the size of a storage location. On x86 processors, a byte can hold numbers between 0 and 255.

Word

• This is the size of a normal register. On x86 processors, a word is four bytes long. Most computer operations handle a word at a time.

Address

• An address is a number that refers to a byte in memory. For example, the first byte on a computer has an address of 0, the second has an address of 1, and so on. ^[1] Every piece of data on the computer not in a register has an address. The address of data which spans several bytes is the same as the address of its first byte.

• Normally, we don't ever type the numeric address of anything, but we let the assembler do it for us. When we use labels in code, the symbol used in the label will be equivalent to the address it is labelling. The assembler will then replace that symbol with its address wherever you use it in your program. For example, say you have the following code:

   .section .data
  my_data:
   .long 2, 3, 4
  

• Now, any time in the program that my_data is used, it will be replaced by the address of the first value of the .long directive.

Pointer

• A pointer is a register or memory word whose value is an address. In our programs we use %ebp as a pointer to the current stack frame. All base pointer addressing involves pointers. Programming uses a lot of pointers, so it's an important concept to grasp.