x86 Assembly Tutorial

COS 318: Fall 2020

 Project 1 Schedule
Deadline to
Design Review Precept 1 Use
x86 Assembly register teams or Project 1
Sessions (Design details andPiazza
Tutorial sign-up for random Due
(15 min per team) Q&A) and OHs.
team assignment

3-7pm ET 7:30-8:20 pm ET 11:55 pm ET

Mon, 9/7 Fri, 9/11 Mon, 9/14 Mon, 9/14 Sun, 9/20
Tue, 9/8
Tue, 9/15 Tue, 9/15

The link will open on 9/12 or 9/13. 2

 Overview
● x86 (technically IA-32) Assembly Overview
○ Registers, Flags, Memory Addressing, Instructions, Stack,
Calling Conventions, Directives, Segments

● BIOS (Basic Input/Output System) +

GDB (GNU Debugger)
● Design Review
3
 Registers
Segment Registers: 16 bits
General Purpose Registers: 8,16,32 bits (hold 16-bit segment selectors
to identify memory segment)
31 15 7 0
EAX
AH AL AX = AH | AL CS code segment

BH BL BX = BH | BL EBX DS data segment

CH CL CX = CH | CL ECX SS stack segment

DH DL DX = DH | DL EDX ES data segment

BP EBP FS data segment

SI ESI GS data segment

DI EDI Instruction Pointer (EIP): 32 bits

SP ESP Flags (EFLAGS): 32 bits 4
Flags
The 32-bit EFLAGS register:

Flags types:

- Status
- Control
- System

5
● Important Flags
○ CF: Carry flag

○ ZF: Zero flag

○ SF: Sign flag

○ IF: Interrupt flag (sti, cli)

[details: https://en.wikipedia.org/wiki/Interrupt_flag]

○ DF: Direction flag (std, cld)

6
Instruction Syntax Conventions

7
source: http://flint.cs.yale.edu/cs422/doc/pc-arch.html
 Memory Addressing
● Memory addressing modes:
1. Real address (unprotected) 2. Protected 3. System Management

● Real address mode:

○ 1MB of memory (20-bit addresses)

○ Valid address range: 0x00000 ~ 0xFFFFF

○ 16-bit segment add. [times 16, i.e., +4 bits] + 16-bit offset add.

8
 Memory Addressing (Real Mode)
● Format (AT&T syntax):
○ segment:displacement(base,index,scale)

● Offset = Base + Index * Scale + Displacement

● Address = (Segment * 16) + Offset

● Displacement: Constant

● Base: %bx, %bp

● Index: %si, %di

● Segment: %cs, %ds, %ss, %es, %fs, %gs

9
 Data Types

Name Size (bits)

byte 8
word 16
double-word
32
(long in gnu assembler)
quad-word 64

10
 Instructions: Arithmetic & Logic
● Arithmetic, such as:
○ add/sub{l,w,b} source,dest
○ inc/dec/neg{l,w,b} dest
○ cmp{l,w,b} source,dest
● Logic, such as:
○ and/or/xor{l,w,b} source,dest ...
● Restrictions
○ No more than one memory operand 11
 Instructions: Data Transfer
● mov{l,w,b} source, dest
● xchg{l,w,b} source, dest (exchange)
● movsb/movsw (move byte/word)
○ %es:(%di) ← %ds:(%si)
○ Often used with %cx to move a number of bytes
■ movw $0x10,%cx
■ rep movsw (repeat)

12
 Stack Layout
...
● Grows from high to low (function arg. n …)
function arg. 1
○ Lowest address = “top” of stack %bp + 8
return address
%bp + 4
old %ebp
● %sp points to top of the stack local var. 1
%bp
%bp - 4
(local var. n …)
○ Used to reference temporary variables
(callee-save regs)
● %bp points to bottom of stack frame
callee-save reg 1

○ Used for local vars + function args. (temp var. n …)

%sp + 4
temp var. 1
%sp 13
 Calling Convention
...

● When calling a function: (function arg. n …)

function arg. 1
○ 1. Push caller-save regs onto stack return address
%bp + 8
%bp + 4
○ 2. Push function args on to stack old %ebp
%bp
local var. 1
○ 3. Push return address + branch %bp - 4
(local var. n …)

● In subroutine: (callee-save regs)

○ 1. Push old %bp + set %bp = %sp callee-save reg 1

○ 2. Allocate space for local variables (temp var. n …)

%sp + 4
○ 3. Push callee-save regs if necessary temp var. 1
%sp 14
 Instructions: Stack Access
...

● pushl source (function arg. n …)

function arg. 1
%bp + 8
○ %sp ← %sp - 4 return address
%bp + 4
old %ebp
○ %ss:(%sp) ← source local var. 1
%bp
%bp - 4
● popl dest (local var. n …)

(callee-save regs)
○ dest ← %ss:(%sp)
callee-save reg 1
○ %sp ← %sp + 4 (temp var. n …)
%sp + 4
temp var. 1
%sp 15
 Instructions: Control Flow
● jmp label ● call label
○ %eip ← label ○ push %eip
○ %eip ← label
● ljmp NEW_CS, offset
○ %cs ← NEW_CS ● ret
○ %eip ← offset ○ pop %eip

16
 Instructions: Conditional Jump
● Relies on %eflags bits
○ Most arithmetic operations change %eflags

● j{e,ne,l,le,g,ge}
○ Jump to label if {=,!=,<,<=,>,>=}

17
 Assembler Directives
● Commands that speak directly to the assembler
○ Are not instructions

● Examples:
○ .globl - defines a list of symbols as global
○ .equ - defines a constant (like #define)
○ .bytes, .word, .asciz - reserve space in memory

https://docs.oracle.com/cd/E26502_01/html/E28388/eoiyg.html
18
 Assembler Segments
● Organize memory by data properties
○ .text - holds executable instructions
○ .bss - holds zero-initialized data (e.g. static int i;)
○ .data - holds initialized data (e.g. char c = ‘a’;)
○ .rodata - holds read-only data
● Stack / Heap - Set up by linker / loader / programmer
19
Basic Input/Output System (BIOS) Services

● Use BIOS services through int instruction

○ Must store parameters in specified registers
○ Triggers a software interrupt

● int INT_NUM sending a character to the

display at the current
cursor position
○ int $0x10: Video services
○ int $0x13: Disk services
20
 Useful GDB Commands
● r - show register values ● b - set a breakpoint

● sreg - show segment registers ● d <n> - delete a breakpoint

● s - step into instruction ● bpd / bpe <n> - disable /

enable a breakpoint
● n - next instruction
● x/Nx addr - display hex dump
● c - continue
of N words, starting at addr
● u <start> <stop> - disassembles
● x/Ni addr - display N
C code into assembly
instructions, starting at addr 21
 Design Review
● Write print_char and print_string assembly functions
● Be ready to describe:
○ How to move the kernel from disk to memory
○ How to create disk image
○ (More specific guidelines are provided on the project page)

22