## Answers to Self-Study Questions

### Test Yourself #1

The constant-propagation information is in red, and the live-variables
information is in green.

### Test Yourself #2

The definition of the may be uninitialized problem is as follows:

- An individual dataflow fact is a set of variables (those that
might be uninitialized), so the domain D of dataflow facts is
the set of all sets of variables.
- The combining (meet) operator is set union.
- The initial dataflow fact (what is true at the entry node)
is the set of all variables in the program.
- This is a GEN/KILL problem, so for each CFG node n, the dataflow
function is of the form
f
_{n}(S) = S - KILL_{n} union GEN_{n}.
The KILL set for node n is the set of variales defined at n,
and the GEN set is always empty.

### Test Yourself #3

Here are the dataflow equations for live analysis.
Note that the meet operator (⌈⌉) is set
union.
enter.after = 1.before

1.before = 1.after - {x}

1.after = 2.before

2.before = 2.after - {y} union {x}

2.after = 3.before

3.before = 3.after

3.after = exit.before ⌈⌉ 4.before

4.before = 4.after union {y}

4.after = 3.before
exit.before = empty-set

*Variable* |
*Least Solution* |
*Greatest Solution* |

1.before |
{ } |
{ } |

1.after |
{ x } |
{ x } |

2.before |
{ x } |
{ x } |

2.after |
{ y } |
{ x, y} |

3.before |
{ y } |
{ x, y} |

3.after |
{ y } |
{ x, y } |

4.before |
{ y } |
{ x, y} |

4.after |
{ y } |
{ x, y } |

The solution we want is the least solution.

### Test Yourself #5

### Test Yourself #6