| java.lang.Object soot.Transformer soot.BodyTransformer soot.jimple.toolkits.scalar.pre.BusyCodeMotion
BusyCodeMotion | public class BusyCodeMotion extends BodyTransformer (Code) | | Performs a partial redundancy elimination (= code motion). This is
done, by moving everycomputation as high as possible (it is
easy to show, that they are computationally optimal), and then
replacing the original computation by a reference to this new high
computation. This implies, that we introduce many new
helper-variables (that can easily be eliminated afterwards). In
order to catch every redundant expression, this transformation must
be done on a graph without critical edges. Therefore the first
thing we do, is removing them. A subsequent pass can then easily
remove the synthetic nodes we have introduced. The term "busy"
refers to the fact, that we always move computations as high
as possible. Even, if this is not necessary.
See Also: soot.jimple.toolkits.graph.CriticalEdgeRemover |
BusyCodeMotion | public BusyCodeMotion(Singletons.Global g)(Code) | | |
internalTransform | protected void internalTransform(Body b, String phaseName, Map opts)(Code) | | performs the busy code motion.
|
|
|