点击次数：

论文类型：期刊论文

发表时间：2022-06-29

发表刊物：清华大学学报 自然科学版

所属单位：软件学院

卷号：51

期号：10

页面范围：1363-1368

ISSN号：1000-0054

摘要：Embedded software needs to deal with many interrupts, but random
   interrupts lead to test difficulties in embedded systems which are
   becoming more and more complicated. A method is given to enable dynamic
   detection of interrupt overloads based on a genetic algorithm. The
   algorithm handles diverse types of interrupts with non-uniform searches
   giving improved local searches to construct an interrupt handling
   sequence to generate the stack depth and detect stack overflow errors
   during interrupt execution. Tests on a SPARC V8 architecture
   Virtualsparc simulation platform show that this method quickly detects
   the loss rate of non-executed interrupts in the stack. Dynamic changes
   in the stack depth greatly reduce the interrupt overload loss rate with
   increased detection efficiency of interrupt faults in embedded software.

备注：新增回溯数据