The ever-increasing use of multi-core chips with many heterogeneous processors in the domains of ADAS and autonomous driving poses new challenges for the verification of the system behavior. However, the increase in hardware/software complexity also increased the demand for deeply embedded logic that supports runtime profiling and monitoring of the application. Even though this technology is capable of tracking which parts of an application have been executed it was only very rarely use for the measurement of code coverage in the context of ISO 26262 due to the complexity of the required code coverage criteria. Until this day code instrumentation is still the predominant approach in this domain.

We present a new approach for the measurement of code coverage on modern automotive chips that combines built-in hardware capabilities with static code analysis at the source code level to satisfy the code coverage objectives of ASIL-D applications like modified condition/decision coverage or call coverage. We explore a workflow that is compliant with the guidelines of ISO 26262 and enables testing in the target environment throughout all development stages and offers multiple benefits over the use of instrumentation. Moreover, our presented approach can be easily extended to become viable for long-term testing and complement compiler qualification efforts.

Was lernen die Zuhörer in dem Vortrag?

Hardware-gestützte Code-Coverage-Messung ohne Instrumentierung