Unit:
Department of Informatics and TelecommunicationsΠληροφορική
Author:
Prokopaki-Kostopoulou Nefeli
Supervisors info:
Σμαραγδάκης Γιάννης, Καθηγητής, Πληροφορικής και Τηλεπικοινωνιών, Εθνικό και Καποδιστριακό Πανεπιστήμιο Αθηνών
Original Title:
Graph-based data structure for representation of sets of must-alias analysis inferences
Translated title:
Graph-based data structure for representation of sets of must-alias analysis inferences
Summary:
In this thesis we present a graph-based data structure for representing alias pairs
that hold in each program point, which is useful for optimizations and program
understanding.
The purpose of this project was to re-implement in Java, the original implementation
of a declarative model of a must-alias analysis over access paths, written in Datalog and in
use in the Doop framework. The new implementation manufactures an optimized data
structure that encodes multiple alias sets, and aliasing relations over longer access paths,
in a single graph and outperforms in speed the original implementaion.
We use as input and output files with relations that correspond to Doop's
intermediate language features.
Main subject category:
Science
Keywords:
must alias, context-insensitive points-to analysis, Java