Research And Implementation Of Software Defect Detection Technology For Embedded C

In this thesis, we will introduce the research work on software testing in Embedded-C language, including the defects model description and the defects analysis using extended defect model state machine, the unified framework for defect model description and detection, and the and the algorithm implement for practical defects, which proved that the system performance well in practice.Static testing method is an important branch in software testing field, because it can test a program just by analyzing its source code and not by running it. It is applied broadly in software testing field and is gradually developing. There are two advantages in static testing method:it is targeted to specific defects and can detect defects with lower probability.The kernel of software static analysis technology based on defect pattern is the definition and description of defects and the design of a unified detection framework. To resolve the problem, a method of using extended defect pattern state machine to detect defects is given. We first introduce the design and implement of defect pattern state machine, including the machine structure, the description rules for defects and the transition conditions of state machine, and to improve system efficiency we discuss some special transition conditions, like path sensitive and variable sensitive. Then we give algorithms for some specific defects under the framework, including the definition of these defects and the design of their state machine and detection algorithms.The defect detection system DTSEmbed applies our research for the detection of some defects in Embedded-C language source code, and the experiment shows that our system has lower false alarm rate and satisfied testing result. Our research provides some useful technologies and approaches for the static analysis methods based on defects pattern.