CS 221 IT 221 Lecture 06

1
CS 221/ IT 221Lecture 06

• Assembly Instructions
• and Intel 80x86
• Dr. Jim Holten

2
Assembly Language Overview

• Assemblers
• Inline assembly in C programs
• Debuggers
• Machine Code
• Memory Maps
• Assembly Mnemonics

3
Assemblers

• Masm.exe MS Windows, requires Visual C
• ML.exe Detmers disk, MS Windows
• as Linux and other OSs, assembles to many
  architectures (IA64 especially)
• nasm Linux, widely used 80x86 assembler
• MANY others, as people write their own.

4
Assemblers (cont)

• Input assembly language code
• Assembler directives
• Code and data definition mnemonics
• Converts to relocatable machine code (object
  files), which may have external references to
  other routines and data
• Linker combines with other object files to define
  external references
• Loader reads into memory, relocates to memory
  locations, and starts its execution.

5
Assembly Inline in C Program

• Start with normal C code file
• Use the assembly function call
• __asm__ __volatile__(ltstring containing assembly
  codegt)
• Any C variables to be referenced inside the
  assembly code need to be declared, for example
• static int apples asm(apple) 25
• The compile line needs to be told to use intel
  assembly conventions
• gcc masmintel test.c o test
• OR generate the assembly source code file (test.s
  here)
• gcc masmintel test.c S

6
Debuggers

• Load the code (uses the loader, but does not run
  the code yet)
• Give interactive access to the code and data
  locations and their contents
• Allow interactive control of the code execution
  through breakpoints and single stepping
• May have a GUI (Windbg and IDEs) or be
  command-line driven (gdb)

7
Windbg Windows Debugger

• GUI-based IDE for Microsoft systems (only)
• Loads files from ML.exe and reassembles them
  using ML.exe as requested
• Allows interactive control of program execution
• Allows interactive viewing of intermediate
  processor, register, and memory states
• Emulates and controls the execution as needed

8
gdb (GNUs Linux debugger)

• Command line-based
• Enables interactive control
• Enables code loading and execution
• Allows breakpoints and single stepping
• View or modify variable values and register
  contents at any stop
• View code as desired
• Compatible with embedded assembly in C

9
Review For ReadingDebugger Outputs

• Binary 1010111011010001100001101001
• Octal
• 1 010 111 011 010 001 100 001 101 001
• 1 2 7 3 2 1 4 1 5 1
• Hexadecimal
• 1010 1110 1101 0001 1000 0110 1001
• A E C 1 8 6 9
• Character codes
• Floating point

10
Other Debugging Support

• Assembler listing of the generated machine code
• as a test.s gttest_s.txt
• od get an octal dump of a file
• -x -- output hex format
• -c -- output ASCII characters (if convertible)
• od xc test.o
• Find ways to print intermediate values to a file
• Store into C variables and print
• Call C fprintf routine

11
Machine Code

• Initializing the code (load, initial data values)
• Code and data in memory in BINARY
• Debugging the code break, step, view
• Really large programs gt really large problems

12
Machine CodeOrganizing It In Memory

• Instructions (.code or .text segment or
  .section)
• Data (and stack) segments (or .section)
• Flow control Following the sequences of
  execution?
• Allowing room for large data what data types,
  how much space?
• Memory mapping keeping track of it all by
  addresses

13
Machine CodeDoes It Run Right?

• Routine starts and ends
• Loop starts and ends
• Jumps and their destinations
• Data locations
• Reserving room for future (run-time) data
• Arrays
• Stacks
• Heaps

14
Machine Code Debugging Code

• Address switches and front panel lights?
• Emulators
• Viewing memory locations (machine
  representations)
• Single step
• Run with breakpoints
• Modern debuggers -- emulators plus
• Data conversion for display
• Display names for variables and routines
• Disassemble the machine code instructions to
  assembly mnemonics

15
Intel 80x86 Assembly

• Pseudo code for mapping and organization
• Mnemonics for easy recognition of references to
  operations and registers
• Name assignments for easy recognition of
  variables and function blocks of code
• Standardizes file organization for specifying
  programs

16
Mnemonics

• Control for the assembler
• Names for operators
• Names for registers
• Names for flags
• Names for data memory addresses
• Names for code memory addresses
• Acronyms for combinations of microcode flags

17
Control for the Assembler

• Code sequence and memory mapping
• Start of code and data
• Ordered memory placement
• End of code and data
• Other control constructs
• Embedded mode control
• Macro definitions and evocations
• Designations of stack, data, and code segments
• Comments

18
Mnemonics Operators

• Moving data among registers and memory locations
  (mov)
• Arithmetic (add, sub, inc, dec, neg, mul, imul,
  div, idiv)
• Jumps, conditional jumps (pg 118), calls, and
  returns (jmp, js, jns, jz, jnz, jc, jnc, , call,
  ret)
• Comparisons, conditionals, and flags (cmp, etc.)
• Manipulating the stack (push, pop, pushad, popad,
  , call, ret)

19
Mnemonics Operators (cont)

• Logical (and, or, xor, not, test)
• Shifts and rotates (shl, shr, sal, sar, ror, rol)
• Floating point arithmetic
• I/O and interrupts
• Extended instructions

20
Mnemonics Registers

• General data registers

21
Mnemonics Registers (cont)

• esi source index register
• edi destination index register
• esp stack pointer
• ebp stack base pointer

22
Status and Results Flags

• EFLAGS register access

23
MnemonicsMemory Locations/Addresses

• Stack start and reserved locations (.STACK)
• Code segment (.CODE, label on code lines)
• Data locations (.DATA, label on data
  declarations)
• Jump point (label on code line)
• Loop back point (label on code line)
• Procedure declaration (PROC and ENDP)

24
Difficulties to Overcome

• There are multiple assembly language conventions
  for the same 80x86 architecture, even for a
  single assembler.
• Proprietary control means NO compatibility
  between assembly language or run-time
  environments, even with the same hardware.
• Open source is free but less controlled.
• I am working out a usable combination for our
  class, meanwhile try what is available and feel
  free to ask questions!