Code generation
- C course
- Expert
- Lesson 16/47
- 5 min
Learn the code generation phase.
Code generation (codegen)
Code generation (codegen) is the stage that turns the AST into a target language. The target can be real machine assembly, bytecode or another high-level language.
In this lesson we emit an assembly-like intermediate language. An emit function writes one instruction per AST node. A number node yields MOV; an operator node yields ADD. Codegen walks the tree in postorder: children's values are loaded into registers first, then the operator applies. This scheme runs on a simple virtual machine. Code quality depends on the target architecture.
emit patterns
Codegen usually runs through an emit function. It takes an instruction name and its arguments, then writes a formatted line. The instruction set depends on the target language.
For a simple virtual machine these suffice: MOV reg, value, ADD reg, reg, LOAD reg, addr, STORE addr, reg. Codegen walks in postorder: left child, right child, then root instruction. When register allocation is hard, the stack manages a small set of registers. Here we assume a single register; R0 holds every value.
Practice Emit assembly instructions
Define void emit(const char *instr, int arg) with body printf("%s %d\n", instr, arg);. In main call it for four instructions: ("MOV", 42), ("ADD", 7), ("STORE", 0), ("LOAD", 1).
Starter code
#include <stdio.h>
// Define emit function
// Call it for MOV, ADD, STORE, LOAD
int main(void) {
return 0;
}Sample solution
#include <stdio.h>
void emit(const char *instr, int arg) {
printf("%s %d\n", instr, arg);
}
int main(void) {
emit("MOV", 42);
emit("ADD", 7);
emit("STORE", 0);
emit("LOAD", 1);
return 0;
}Expected output
MOV 42 ADD 7 STORE 0 LOAD 1
Questions and answers
Review the lesson questions together with their answers. To test yourself, solve the lesson interactively.
1 What does the codegen stage produce?
- Tokens
- AST
- Symbols
- Target code Correct answer
Explanation: Codegen writes code for the target language.
2 Match node types to instructions
| Term | Meaning |
|---|---|
num |
MOV |
binop |
ADD |
var |
LOAD |
assign |
STORE |
3 Complete the emit call
emit("MOV R0, #%d", ____);emit("MOV R0, #%d", value);Explanation: emit carries the numeric value into the MOV instruction.