Research Of The Instrumentation In Embeded Software Coverage Testing

As embedded applications continue to expand, much more concern about the Embedded Software Testing has been raised by the Developer. Compared with other software, the embedded software has its own specificity, which means that it can only be Implemented on the Demand Designated hardware platform, and because of the different Development environment and runtime environment, even the test on the mainframe environment is quite fully, there is no adequate justification for us to guarantee that the software will work well on the Goal machine. Besides, Due to the features of embedded software, such as real-time, memory resources limited, expensive development tools, and so on, this makes the optimization technology of embedded software testing more important. Instrumentation technology is the key in Embedded Software Coverage Testing, the main task of this paper is that research of two questions: code expansion and program executed efficiency in instrumentation technology.In this paper, the research on the optimization technology includes three aspects: reducing testing requirement, reducing test case, reducing the overhead of program instrumentation. Test suite reduction problem is one of important problems in software testing. It is aims to satisfy all testing requirements with the minimum number of test cases, so that the testing efficiency can be improved while the test cost can be decreased. Meanwhile, the location of instruments is also a problem to improve the efficiency for computer; its endeavor is to reduce the number of instruments inserted into program.That simplie the control flow graph for the program is aim to reduced testing requirement. The notion of generalized post-dominators can be used for test suite optimization, The primary idea of the method, called virtual node method (VNM), is to add virtual nodes to the control flow, it can be used to reduce the overhead of program instrumentation for coverage testing by reducing the redundancies.In order to achieve better cross-test results, we used gcc and gcov, in order to handle transmission of data between targets and host, we improved it. To simulate the hardware environment we used, and given an example for the improvement.