首页> 外文会议>International conference on Hardware/software codesign and system synthesis >Hardware assisted pre-emptive control flow checking for embedded processors to improve reliability
【24h】

Hardware assisted pre-emptive control flow checking for embedded processors to improve reliability

机译:硬件辅助的抢先式控制流检查,用于嵌入式处理器,以提高可靠性

获取原文

摘要

Reliability in embedded processors can be improved by control flow checking and such checking can be conducted using software or hardware. Proposed software-only approaches suffer from significant code size penalties, resulting in poor performance. Proposed hardware-assisted approaches are not scalable and therefore cannot be implemented in real embedded systems. This paper presents a scalable, cost effective and novel fault detection technique, to ensure proper control flow of a program. This technique includes architectural changes to the processor and software modifications. While architectural refinement incorporates additional instructions, the software transformation utilizes these instructions into the program flow. Applications from an embedded systems benchmark suite are used for testing and evaluation. The overheads are compared with the state of the art approach that performs the same error coverage using software-only techniques. Our method has greatly reduced overheads compared to thestate of the art. Our approach increased code size by between 3.85-11.2% and reduced performance by just 0.24-1.47% for eight different industry standard applications. The additional hardware (gates) overhead in this approach was just 3.59%. In contrast, the state of the art software-only approach required 50-150% additional code, and reduced performance by 53.5-99.5% when error detection was inserted.
机译:通过控制流检查可以提高嵌入式处理器的可靠性,并且可以使用软件或硬件进行检查。提议的仅软件方法会遭受重大的代码大小损失,从而导致性能不佳。提议的硬件辅助方法是不可扩展的,因此不能在实际的嵌入式系统中实现。本文提出了一种可扩展,具有成本效益的新颖故障检测技术,以确保程序的正确控制流程。该技术包括对处理器的体系结构更改和软件修改。虽然体系结构改进包含了附加指令,但是软件转换会将这些指令利用到程序流中。嵌入式系统基准套件中的应用程序用于测试和评估。将开销与使用仅软件技术执行相同错误覆盖率的最新技术方法进行比较。与现有技术相比,我们的方法大大减少了开销。对于八个不同的行业标准应用程序,我们的方法将代码大小增加了3.85-11.2%之间,而性能仅降低了0.24-1.47%。这种方法的额外硬件(门)开销仅为3.59%。相反,最新的纯软件方法需要50-150%的附加代码,并且在插入错误检测时将性能降低了53.5-99.5%。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号