Font Size: a A A

Research On Java’s Synchronization Optimization Based On Soot Framwork

Posted on:2014-06-30Degree:MasterType:Thesis
Country:ChinaCandidate:Y S GuanFull Text:PDF
GTID:2268330425951829Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
With the help of attributes in Java class file, this paper proposes a method for Java program optimization. In the method, we can find out the redundant synchronization operations on the objects. And then it attached these tags into Jimple statements in Soot. At the same time, the method transforms the tags into attributes of the class file. At last, makes the Java class file through the Jasmin assembler. We accomplished the algorithm based on the Soot. Soot is a framework which supports and transforms the Java class file. Focusing on the synchronization optimization, this paper discusses and depicts the following work:(1) Describe a modified escape analysis algorithm.In this paper, we use the Points-to as a basic tool, focusing on static field, thread object and thread filed. We put the analysis operations on the objects in a single while and reduce the consumption of the algorithm. With the help of analysis rules and analysis status, we make some iteration. Then we can find out all local objects in the whole application and attach the information on the attributes of the byte-code file. Finally, when the JVM runs the class file, it can make some optimizations with the information in the attributes.(2) Achieve the algorithm based on the Soot framework.We achieve the algorithm based on the soot framework. Firstly, make some pointers graphs analysis, with the tool of points-to framework-Soot Pointer Analysis Research Kit(SPARK); then we use this basement and extend the flow analysis in the Soot to achieve the escape analysis algorithm. The important step in our work is that extend the annotation framework in Soot and reduce the redundant operations on while in the former research work. At the same time, we get the information about the local objects after the escape analysis and make the information into the tags which are attached to the Jimple statements. After the above operations, we aggregate the tags through the aggregator. Finally, it generates the byte-code file which contains the optimization attributes. It is useful for the running in the JVM.
Keywords/Search Tags:Java, escape analysis, synchronization, Soot framework, attribute, byte-code
PDF Full Text Request
Related items