ICS 412

1
Chapter 8

• ICS 412

2
Code Generation

• Final phase of a compiler construction.
• It generates executable code for a target
  machine.
• A compiler may instead generate some form of
  assembly code that must be processed further by
  an assembler, a linker, and a loader.

3
Intermediate Code

• An intermediate representation that looks like
  target code is called intermediate code.
• Intermediate code can take many forms.
• Two popular forms are
• Three-Address code
• P-Code

4
Intermediate Code

• Intermediate code is particularly useful when the
  goal of the compiler is to produce efficient
  code.
• The analysis of the properties of the target code
  can be easily generated from intermediate code.
• Intermediate code can also be useful in making
  the compiler target machine independent.
• To generate code for a different target machine
  we only need to write a translator from the
  intermediate code to the target code.

5
Three-Address Code

• The three address code has the following general
  form
• x y op z
• The name "three-address code" comes from this
  form of instruction.
• x, y, and z represents an address in memory.

6
Example

• Consider the arithmetic expression
• 2 a ( b 3 )
• The corresponding three-address code is
• t1 2 a
• t2 b 3
• t3 t1 t2

7
Three-Address Code

• t1, t2, and t3 are temporaries correspond to the
  interior nodes of the syntax tree and represent
  their computed values, with the final temporary
  (t3, in this example) representing the value of
  the root.
• -
• 2 a b 3

8
Three-Address Code

• The above three-address code represents a
  left-to-right linearization of the syntax tree.
• Another order is possible for this three-address
  code, namely (with a different meaning for the
  temporaries),
• t1 b - 3
• t2 2 a
• t3 t2 t1

9
Three-Address Code

• One form of three-address code is insufficient to
  represent all language features.
• For instance, unary operators uses a
  two-addresses variation of the three-address code
  
• t2 - t1
• It is necessary to vary the form of three-address
  code to represent all the programming constructs.

10
Factorial Example

• read x
• t1 x gt 0
• if_false t1 goto L1
• fact 1
• label L2
• t2 fact x
• fact t2
• t3 x 1
• x t3
• t4 x 0
• if_false t4 goto L2
• write fact
• label L1
• halt

• read x
• if 0 lt x then
• fact 1
• repeat
• fact fact x
• x x 1
• until x 0
• write fact
• end

11
Factorial Example

• This code contains a number of different forms of
  three-address code
• Built-in input/output operations read and write
  have been translated directly into one-address
  instructions.
• Conditional jump instruction if_false that is
  used to translate both if-stmt and repeat-stmt
  ant that contains two addresses.
• One address label instruction used to represent
  the position of the jump.
• Halt instruction to represent the end of the
  code.

12
P-Code

• P-code began as a standard target assembly code
  produced by a number of Pascal compilers of the
  1970s and early 1980s.
• It was designed to be the actual code for a
  hypothetical stack machine, called the P-machine,
  for which an interpreter was written on various
  actual machines.
• The idea was to make Pascal compilers easily
  portable by requiring only that the P-machine
  interpreter be rewritten for a new platform.

13
P-Code

• The P-machine consists of
• A code memory.
• An unspecified data memory for named variables.
• A stack for temporary data, together with
  whatever registers are needed to maintain the
  stack and support.

14
Example 1

• Consider the expression
• 2a(b-3)
• P-code for this expression is as follows
• ldc 2 load constant 2
• lod a load value of variable a
• mpi integer multiplication
• lod b load value of variable b
• ldc 3 load constant 3
• sbi integer subtraction
• adi integer addition

15
Example 1

• These instructions are to be viewed as
  representing the following P-machine operations
• ldc 2 pushes the value 2 onto the temporary
  stack.
• lod a pushes the value of the variable a onto the
  stack.
• mpi pops these two values from the stack,
  multiplies them and pushes the result onto the
  stack.
• lod b and ldc 3 push the value of b and the
  constant 3 onto the stack (there are now three
  values on the stack).
• sbi pops the top two values from the stack,
  subtracts them, and pushes the result.
• adi pops the remaining two values from the stack,
  adds them, and pushes the result.
• The code ends with a single value on the stack,
  representing the result of the computation.

16
Example 2

• Consider the assignment statement
• x y 1
• The corresponding P-Code is
• lda x load address of x
• lod y load value of y
• ldc 1 load constant 1
• adi add
• sto store top to address
• below top pop both.

17
Example 3

• read x
• if 0 lt x then
• fact 1
• repeat
• fact fact x
• x x 1
• until x 0
• write fact
• end

18
P-Code

• lda x load address of x
• rdi read an integer store to address on top of
  stack ( pop it)
• lod x load the value of x
• ldc 0 load constant 0
• grt pop and compare top two values, push
  Boolean result
• fjp Ll pop Boolean value, jump to Ll if false
• lda fact load address of fact
• ldc 1 load constant 1
• sto pop two values, storing first to address
  of second
• lab L2 definition of label L2
• lda fact load address of fact
• lod fact load value of fact
• lod x load value of x
• mpi multiply

19
P-Code

• sto store top to address of second pop
• lda x load address of x
• lod x load value of x
• ldc 1 load constant 1
• sbi subtract
• sto store (as before)
• lod x load value of x
• ldc 0 load constant 0
• equ test for equality
• fjp L2 jump to L2 if false
• lod fact load value of fact
• wri write top of stack pop
• lab Ll definition of label Ll
• stp

20
P-Code vs. Three-Address Code

• P-code
• Closer to machine code.
• Fewer addresses (1 or 0).
• Stack automatically handles temps, so compiler
  does not need to generate name/locations.
• Three-Address
• fewer instructions
• More complex instructions, so less code to
  generate.