Inefficient IR List Construction (O(n²))

[SAMBA8695]

Description

IR instructions are stored in a singly linked list. Each call to append_ir() traverses the entire list to find the end, resulting in quadratic time complexity during IR generation.

while (temp->next != NULL) {
    temp = temp->next;
}

This becomes inefficient for large programs.

Current Behavior

• Appending IR instructions is O(n)
• Generating n instructions results in O(n²) time complexity
• Performance degrades rapidly as program size grows

Expected Behavior

• IR construction should be linear time
• Appending instructions should be O(1)

Suggested Direction

• Maintain a tail pointer for the IR list
• Or use a dynamic array/vector structure
• Optionally refactor IR into basic blocks