Instruction cache locking for multi-task real-time embedded systems

摘要

In a multi-task embedded system, a cache is shared by different tasks, which increases the complexity of cache management and the unpredictability of cache behavior. This unpredictability in turn brings an overestimation of application's worst-case execution time (WCET) and worst-case CPU utilization (WCU) which are two of the most important criteria for real-time embedded systems. Modern processors often provide cache locking capability, which can be applied statically and dynamically to manage cache in a predictable manner. The selection of instructions to be locked in the instruction cache (I-Cache) has dramatic influence on the system performance. This paper focuses on applying cache locking techniques to the shared I-Cache to minimize WCU for multi-task embedded systems. We analyze and compare three different strategies to perform I-Cache locking: static locking, semi-dynamic locking, and dynamic locking. Different algorithms are proposed utilizing the foreknown information of embedded applications. Experimental results show that the proposed algorithms can reduce WCU compared to previous techniques.