A sample Incomplete C Program

1
A sample Incomplete C Program

• int AI_Fib 10, I_n 4, AI_Arg105,7,10,13
• int main ( )
• //begin main
• int IF_Fibonacci (int) I_Idx
• for (I_Idx0 I_Idx ltI_n I_Idx)
• //begin for 1
• AI_Fib I_IdxIF_Fibonacci(AI_ArgI_Idx
  )
• printf ( Fib (d) d\n, AI_ArgI_Idx
• AI_FibI_Idx)
• //end for 1
• //end main

2
The Pentium Integer Registers
3
Common Conventions Adopted

• 1)  Constant values are either qualified as
  relevant types quantity and preceded by 0 as well
  as suffixed by the type specifying letter H
  (Hex) or D (Decimal) e.g. 020 H ? 20 Hex, 030 D
  ? 30 Decimal OR specified using relevant pseudo
  Op Code.
•    2) A GPR/CPU Register is preceded by a
  special character / .
•    3) A Memory location is either directly
  Addressed via a Symbol or by the specified
  addressing modes To be Illustrated Later
• 4) Comments are preceded by / and are
  ignored during Assembly translation.
• 5) The relative ordering of the Operands in
  an instruction may vary from assembler to
  assembler even for the same CPU.

4
The Compiled Pentium Assembly Code 1The
Assembler Directives (Pseudo Ops)

• .file main.c The file
  Directive/ Pseudo Op
• .globl I_n Global
  Symbol Directive
• .data Data
  Segment STARTS
• .align 4 Align with
  4 Byte Boundary
• .type I_n , _at_object Details of I_n
• .size I_n, 4 I_n Size is 4
  Bytes
• I_n Label I_n The
  Symbolic Pointer
• .long 4 Initial Value 4
  of 4 Byte Size

5
The Compiled Pentium Assembly Code 2The
Assembler Directives (Pseudo Ops)

• .globl AI_Arg Another Global
  Object
• .align 32 Align with 32
  Byte Boundary
• .type AI_Arg,_at_object Details of AI_Arg
• .size Ai_Arg,40 Size 4 10 40
  Bytes
• AI_Arg Label The Symbolic Base
  Pointer
• .long 5 Initial Value of
  AI_Arg 0
• .long 7 Initial Value of
  AI_Arg 1
• .long 10 Initial Value of
  AI_Arg 2
• .long 13 Initial Value of
  AI_Arg 3
• .zero 24 Remaining 24 Bytes
  initialized to 0

6
The Compiled Pentium Assembly Code 2The
Assembler Directives (Pseudo Ops)

• .LC0 Label LC0 ,
  Symbolic Pointer
• .string fib(d) d \n
  Format Specification
• .text Code Segment
  Starts Here
• .align 2 Align in 2 Byte
  Boundary Pseudo Op
• .globl main Global Symbol main
  Pseudo Op
• .type main,_at_function Function
  main Pseudo Op
• main Instruction Label Entry
  Point
• pushl ebp PUSH CPU register ebp in
  User Stack
• esp ? esp
  MINUS 4
• Mesp ? ebp
  Little Endian Conven.

7
The Compiled Pentium Assembly Code - 4

• pushl ebx PUSH CPU register ebx
  in User Stack
• esp ? esp
  MINUS 4
• Mesp ? ebx
  Little Endian Conven.
• subl 4, esp esp ? esp MINUS 4
  (Decimal)
• Create
  4 Byte Space in Stack
• andl -16,esp esp ? esp -16
  (Decimal)
• Align
  Stack Pointer in 16 Bit Bound
• movl 0,eax eax ? 0 (Decimal)
• subl eax, esp esp ? esp MINUS eax
  Set Flags ?

8
The Compiled Pentium Assembly Code - 5

• movl 0 , -8(ebp) M ebp-8 ? Decimal
  0
• Local
  Variable I_Idx ?0
• 32 bit 2s Complement Value in Little Endian
  Convention
• .L2 Label
• movl 8(ebp) , eax eax ? M ebp 8
  (I_Idx)
• cmpl I_n, eax Compare I_n with
  I_Idx
• jl L5 if I_Idx lt
  I_n go to label .L5
• jmp L3 Otherwise (I_Idx gtI_n) go to
  label .L3 (EXIT)

9
The Compiled Pentium Assembly Code - 6

• .L5 Label L5
• movl 8(ebp), ebx ebx ? Mebp-4
  I_Idx
• subl 12 , esp esp ? esp MINUS 12
  (Decimal)
• Create 12
  Byte Stack Space for Parameters
• movl -8(ebp), eax eax ? Mebp-4
  I_Idx
• pushl AI_Arg(,eax, 4)
• Push MAI_Arg 4 eax (Actually
  I_Idx)
• Push AI_Arg0 the First Parameter
• call IF_Fibonacci User Function Call
• addl 16, esp esp ? esp PLUS 16
  (Decimal) Reclaim
• Stack Space used
  for Fibbonaci Function

10
The Compiled Pentium Assembly Code - 7

• movl eax, IF_Fibonacci ( , ebx, 4)
• Returned Value is in eax Now IF_Fibonacci
  I_Idx ? This Value
• M IF_Fibonacci4ebx (actually I_Idx)
  ? eax
• subl 4 , esp esp ? esp MINUS 4 (Decimal)
  
• Create 4 Byte Stack
  Space for Storing Function Value
• movl -8(ebp), eax eax ? Mebp-4 I_Idx
• pushl IF_Fibonacci( ,eax, 4)
• Push M IF_Fibonacci 4 eax
  (Actually I_Idx)
• Push IF_Fibonacci I_Idx
• movl -8(ebp), eax eax ? Mebp-4 I_Idx
• pushl AI_Arg(,eax, 4)
• Push MAI_Arg 4 eax (Actually
  I_Idx) AI_Arg0
• pushl .LC0 Push the address of the
  Format String
• call printf System Function Call

11
The Compiled Pentium Assembly Code - 8

• addl 16, esp Reclaim Stack Space used for
  printf
• leal -8(ebp) , eax
• eax ? Address obtd by ebp MINUS 8 I_Idx
• incl (eax) Increment the Memory Location
  Pointed to
• by the eax I.e. I_Idx
• jmp .L2 Continue LOOPING
• .L3 Loop Exit Point
• movl -4(ebp) , ebx ebx ? M ebp 4
  Restore old
• leave Clear Stack esp ? ebp
• ret

12
Basic Features of Assembly Language (Generated by
the Compiler after translating the HLL / Source
Code)

•    A language, which is machine/CPU dependent.
•    The humans can perceive it too.
•   Composed of mnemonic Op-Codes (small English
• like words specifying the Operation to be
  performed)
• Labels Operand Addresses.
•   Has got two (2) distinct types of OP-Codes
  namely
• a)    Executable Machine OP-Codes.
• b) Assembler Directives (Non Executable/
• Assembler Dependent only) Pseudo
  OP-Codes.

13
Typical Format of Assembly Instruction(
Assembler Dependent )

• ltSymbolic Labelgt optional useful for usage as
  Target Operand Address of Branching / Jump
  Instructions or Symbolic Address Pointer
• ltMnemonic Machine OP Codegt ltOperand ? 1 Addressgt,
  
• ltOperand ? 2 Addressgt, , ltOperand ? n Addressgt
• Number of Operand Addresses in an instruction(n)
  may vary from 0 to 4 to be illustrated later.
  Here it is 2 . The Relative Ordering of Operands
  ( Source , Destination for 2 Operands , Result ,
  Source 1 , Source 2 for 3 Operands ) is also
  assembler dependent.

14
Typical Format of Assembly Instruction(
Assembler Dependent ) Example

• .L3
• movl -4(ebp) , ebx
• Label .L3
• Mnemonic Op Code movl Data Movement
• Source Operand -4(ebp) Based Offset Addr.
• Destination Operand ebx Register Direct Addr.
  

15
Basic Op Code Types in an Instruction Set
Architecture (ISA) - 1

• Data Movement
• Source
  Destination
• CPU Register Memory /
  Peripheral Register
• Memory/Peripheral Register CPU
  Register
• CPU Register
  CPU Register
• Immediate Data
  CPU Register
• Immediate Data
  Memory / Peripheral

16
Basic Op Code Types in an Instruction Set
Architecture (ISA) - 2

• Data Processing
• Operand 1 Operand2
  Result Destination
• CPU Register Memory /
  Peripheral Register
• Memory/Peripheral Register CPU
  Register
• CPU Register
  CPU Register
• Immediate Data
  CPU Register
• Immediate Data
  Memory / Peripheral

17
Basic Op Code Types in an Instruction Set
Architecture (ISA) - 3

• Control Flow
• Op Code Target Address Memory

18
Operand Addresses in the Instruction Set
Architecture (ISA) - 1

• A. Constants/ Immediate Data Binary Integers of
  various sizes (signed /unsigned) . Sign depends
  on the Interpretation. Usually Hexadecimal (Hex)
  notation is used for compactness. However we
  shall be using decimal , the underlying
  conversion will convert it into binary /hex.
• Typical sizes Byte 1 Character , Word
  2 Bytes , Double Word/ Long 4 Bytes , Quad
  Word 8 Bytes

19
Operand Addresses in the Instruction Set
Architecture (ISA) - 2

• B. CPU Registers
• 1) General Purpose Registers GPRs can
  be used to store Data , Address, Result of
  Computation. Usually programmer accessible partly
  / as a whole.
• 2) Special Purpose Registers Specific
  Purpose of usage like Instruction Pointer, Stack
  Pointer, Segment Registers Restricted
  accessibility.
• 3) Condition Flags Stores status of
  computation like ZERO Flag , SIGN Flag, CARRY
  Flag. Useful in conditional Control Flow
  Instructions.

20
Operand Addresses in the Instruction Set
Architecture (ISA) - 3

• C. Main Memory / RAM / RWM Locations
• A location ? ( Address , Data Content ).
• There are several ways to address a memory
  location depending on the Addressing Modes as
  provided by the CPU. This will be illustrated in
  due course. All memory accesses happens through
  Cache Memories ( is ignored for the time being).
• D. Stack LIFO Memory Grows from Higher
  Address towards lower Address. Accessed via some
  special purpose registers termed as Stack
  Pointer. Useful for passing parameters , storage
  of activation records . Will be illustrated in
  due course.

21
Machine Language Features

• 1)    Obtained by translating an assembly
  language program by an Assembler
• 2) Everything i.e. Machine OP Codes, Operands as
  well as constant values are represented in Binary
  Form i.e. strings of 1s 0s or in Hex notation
  (for compactness sake).
• 3)   A language that can be directly interpreted
  by the machine but difficult to interpret by the
  humans
• e.g.
• movl esp, ebp ? ebp ? esp gt 1000 1001
  1110 0101
• 
  ? 89E5

22
Machine Language Conventions

• 1)  Each Memory Location 8 bit wide
• i.e. 1 byte wide (2 Hex digits)
•    2) Each Memory Address 32 bit wide
• i.e. 4 byte long 8 Hex digits.
• a) Hence any memory variable/symbolic
  operand will be
• represented by 8 Hex digits
• b) Any statement symbolic label will
  translate to 8 Hex digits.
• c) Any memory location having a 32 bit
  address will either
• contain 8 bit data (Operand Value)
  belonging
• Data Memory Segment or will contain an
  8/ 16/32Bit Machine
• Op-Code Program Memory / Code Segment
  

23
Assembly Machine Code ( List File) - 1

• Location Machine Code
  Symbol/Label Directive Value
• (in HEX)
  
  .data
• 
  
  .align 4
• 
  .type
  I_n,_at_object
• 
  .size I_n,4
• 
  I_n
• 0000 04000000
  .long 4
• 
  .globl
  AI_Arg
• 0004 00000000
  .align 32
• 
  .type
  AI_Arg,_at_object
• 
  .size
  AI_Arg, 40
• 
  AI_Arg
• 0020 05000000
  .long 5
• 0024 07000000
  .long 7
• 
  
• 
  
• 
  

24
Assembly Machine Code ( List File) - 2

• Location Machine Code
  Symbol/Label Directive Value
• (in HEX)
• 0020 0A000000
  .long 10
• 0024 0D000000
  .long 13
• 0030 00000000
  .zero 24
• 
  00000000
• 
  00000000
• 
  00000000
• 
  00000000
• 
  
  .section .rodata
• 
  .LC0
• 0000 66696228
  .string
  fib(d)d\n
• 
  25642920
• 
  3D202564
• 0A00
  
• 
  
• 
  

25
Assembly Machine Code ( List File) - 3

• Location Machine Code
  Symbol/Label Directive Value
• (in HEX)
  
  .text
• 
  
  .align 2
• 
  
  .globl main
• 
  .type
  main,_at_function
• 
  main START of Execution
• 0000 55
  pushl ebp
• 0001 89E5
  movl esp,ebp
• 0003 53
  pushl ebx
• 0004 83EC04
  subl 4,esp
• 0007 83E4F0
  andl -16.esp
• 000a B800000000
  movl 0, eax
• 000f 29C4
  subl eax,esp
• 0011 C745f800
  movl 0,-8(ebp)
• 000000
• 
  
  
  

26
Assembly Machine Code ( List File) - 4

• Location Machine Code
  Symbol/Label Directive Value
• (in HEX)
• 
  .L2
• 0018 8B45F8
  movl -8(ebp), eax
• 001B 3B050000
  cmpl I_n, eax
• 0000
• 0021 7C02
  jl .L5
  
•