Worst-Case Execution Time Analysis In Real-Time System

Whether Real-Time System is correct or not is not only built on the result of logic computation , but also related to the time for producing of it . Real-Time System should be response to internal event in safety and predictable time period. The Worst-Case Execution time refer to the maximum execution time of Real-Time program. It is important in the following fields: System Schema; task prior arbitration; resource conflict arbitration; communication between tasks and partition between software and hardware. Concerning to a given processor -ARM7TDMI, this paper is going to evaluate the Worst-Case Execution Time of MP3 Sub-System , offer a impersonal opinion about the optimization of MP3 programs , meanwhile, give an important base for the software and hardware partition in MP3 Module.There are two important issues in calculating WCET: High-Level Analysis and Low-Level modeling, that is, program path analysis and microarchitecture modeling. Under the advanced program language, High-Level Analysis most purchase in analyzing program structure, determining what sequence of instructions will be executed in the worst case, and constructing corresponding control flow graph. Low-Level modeling will build up hardware model under the assemble language, then obtain the WCET of a known sequence of instructions. The compLEXity of calculating WCET limits the accuracy of the estimated bound and the size of the program that can be analyzed.In this paper, we present an effective method to determine a tight bound on a program's worst case execution time. The method includes a direct-mapped instruction Cache analysis and pipeline units analysis, and use an integer linear programming formulation to solve WCET. This solution is implemented in the tool CRYINGCAT which is developed by LEX and Bison and currently targets the ARM7TDMI processor. Considering pipelined instruction execution units and Cached memory systems, the tool CRYINGCAT constructs Pipeline Model and Cache Model. Using these model, we can obtain the precise instruction execution time. Finally, Since the linear constraints are mostly derived from the network flow graphs, the ILP problems are typically solved efficiently by ILP Solver. As an experiment, we calculated MP3 Sub-System WCET using this tool.