COMS W4115
Programming Languages and Translators
Lecture 23: Code Generation Algorithms
April 21, 2008

Lecture Outline

1. Review
2. Next-use information
3. A simple code generator
4. Optimal code generation for expression trees

1. Review

• Instruction selection
• Optimization of basic blocks
• Order of evaluation

2. Next-use Information

• If there is a three-address instruction sequence of the form

       i:  x = y + z

       .
       . no assignments to x
       .

       j:  a = x + b
  

then we say statement j uses the value of x computed at i.
• We also say that x is live at statement i.
• A simple way to find next uses is to scan backward from the end of a basic block keeping track for each name x whether x has a next use in the block and if not whether x is live on exit from that block. See Alg. 8.7, p. 528.

3. A Simple Code Generator

• There is a simple algorithm to generate code for a basic block. It generates code for each three-address instruction in turn, keeping track of what values are in what registers to avoid generating redundant loads and stores.
• It uses a register descriptor to keep track of what value is currently in each register.
• It uses an address descriptor to keep track of locations where the current value of a name can be found at run time.
• It uses a routine getReg to return a location L to hold the value of the variable x for a three-address instruction x = y + z.
• See Section 8.6 (pp. 542-548) for details of the algorithm.

4. Optimal Code Generation for Expression Trees

• Assume a simple register machine with instructions of the form
• LD reg, mem
• ST mem, reg
• OP reg, reg, reg
• Label the nodes in an expression tree with Ershov numbers.
• The Sethi-Ullman algorithm (Alg. 8.24, pp. 568-572) can be used to generate code that minimizes the number of spills to evaluate the expression tree.

5. Reading

• ALSU, Sections 8.4, 8.6, 8.10