Java baitkoodi sünkroniseerimise analüüs raamistikus Põder

Abstract

Complex applications demand more and more performance from computers. One way to add performance is to add more cores to the processor. To utilize those added cores, the use of multithreading in programs is needed. However, multithreading adds additional complexity to programs and one problem that arises with multithreading is data races. Data races can be discovered with static analysis. The aim of this thesis is to add a module to the Põder framework that analyses Java bytecode synchronization and that could rule out the presence of data races. The result of the analysis produces a report that states if it could rule out data races. If not it brings to attention the variables that could have data races and to points out from where they were accessed and which monitors were guarding them at that time. With this information it is possible to fix synchronization flaws so that data races can be eliminated.