Research Of The Runtime Errors Site Analysis Tool Based On The Variable Tracking

The runtime errors are a kind of program errors which are difficult to detect in Cprogram, because runtime errors are generated in the course of the program’s dynamicexecution, with concealed and unforeseen features. For example, the common runtimeerrors include array bounds, dynamic memory allocation, memory overflow, illegalpointer references, the type of inconsistency implicit conversion, etc. These errors arenot found at compile time.With the scale of the program increase and the functions morecomplex,these runtime errors are even hard to be found in the testing phase, especiallyin the actual running environment. Those potential and unknown errors, generated alongwith the program execution, can’t be detected by the current errors detection tools.Such errors as are caused by programs’ stopping unexpectedly, multi-threadedprograms’ abnormal behaviors, unreliable data sources and logical errors should beanalyzed and located with traditional debug tools when occurring. However,thetraditional debug analysis methodexist some problems as follows: relying entirely onhuman intervention, it needs a large workload. And for the real execution environmentof some exception and multi-thread to be reproduced, it is not obvious. If we cancapture the variable information during the program execution, the runtime error sitewill be captured, which is used as a basis for analyzing the source of the errors. In thispaper, the technology of capturing runtime error site based on variable tracking isproposed.The main contents of this paper have been summarized as follows:1. Introduce the research background, and the present condition and deficiencies ofruntime error detection techniques and tools, and present the method of capturingruntime error site based on variable tracking.2. Use source code instrumentation technology and implement a lightweightinstrumentation tool which support user-selected ranging from specified variable and allthe variables to function.3. Utilize and improve existing program analysis techniques, providing twomethods of analyzing and detecting errors. On one hand, use program information flowanalysis techniques to slice the source program and reduce the search scope, aiming toguide the users to find runtime errors. On the other hand, we use testing technology toget runtime dependency information and build appropriate model to calculate statements suspicious, further narrowing the scope of the errors.4. Implement a tool named runtime fault site analysis, providing multiple data viewwindows.This paper surveys automatic instrumentation technology to capture dynamicvariable information and presents two methods of error analysis and debug. Based onthe above techniques, runtime fault site analysis tool can be used to analyze the causesof errors and provide error site and aided analysis approach as well. We can observedata flows and running trace and other information presented in data view window oftool, which can detect errors more quickly and accurately, combining with aidedanalysis methods. In the application, strong practicability is a powerful supplement ofexisting error detection tools.