Structured Testing of WorstCase Execution Time Analysis Methods.pdfVIP

Structured Testing of Worst-Case Execution Time Analysis Metho ds Jakob Engblomy  Andreas Ermedahly Friedhelm Stapp ert z IAR Systems AB Dept. of Information Technology C-LAB Box 23051, SE-750 23 Uppsala Uppsala University Furstenallee 11 Sweden Box 325, SE-751 05 Uppsala 33102 Paderb orn email: jakob.engblom@iar.se Sweden Germany email: andreas.ermedahl@docs.uu.se email: fst@c-lab.de Abstract met for p erio dic tasks, and to check that interrupts have suciently short reaction times. Knowing the Worst-Case Exe cution Time (WCET) To b e valid, WCET estimates must b e safe , i.e. guar- of a progr am is necessary when designing and verifying anteed not to underestimate the execution time. To b e real-time systems. When WCET analysis tools are used useful, they must b e tight, i.e. provide low overestima- to estimate the WCET, the tool is a critical part of the tions. The goal is to come as close as p ossible, but not system design and must be correct. b elow, the execution time on the target hardware. In this p aper we pr esent a methodology for system- The WCET dep ends b oth on the program ow (like atical ly testing WCET analysis tools. The methodology lo op iterations and function calls), and on architectural is based on a decomposition of W