Safe worst-case execution time analysis by abstract interpretation of executable code

Abstract

To verify that a safety-critical application meets all requirements lies in the responsibility of the system designers. Classical software validation methods like code review and testing with debugging are expensive and cannot really guarantee the absence of errors. In constrast, abstract interpretation provides a methodology for static program analysis that yields statements valid for all program runs with all inputs. Employing static analyzers is thus orthogonal to classical testing. A possible cause of catastrophic failure is the violation of timing constraints. Recent trends, e.g. in automotive industries (X-by-wire, time-triggered protocols) require knowledge on the Worst-Case Execution Time (WCET) of each task. Yet WCET determination is a difficult problem because of the characteristics of modern software and hardware. This talk focuses on AbsInt's tool aiT that combines abstract interpretation and integer linear programming to determine upper bounds for the WCETs of code snippets given as routines in executables. The hardware architecture of the underlying processor is precisely taken into account, including cache and pipeline behavior. Bounds are determined without the need to change the code and hold for all executions. aiT is used among others by Airbus in the development of various safety-critical applications for the A380.