Page faults and the algorithm of paging used in the kernel are an important aspect of the overall functioning of the system. This does add an overhead to the overall processing. Numerous algorithms and their improvements exist for paging but miss an important point. The ability to statistically assess and gather data about major and minor page faults at run time are missing. In this paper, we propose an approach to solve this problem by the mechanism of adding our device driver in the kernel. The driver when invoked from kernel space gathers data from numerous data structures which the kernel maintains and computes. This computation then delivers the major and minor page faults occurred by user space applications at that instant. We believe by using our unique approach in measuring page faults by paging algorithms, these can be statistically backed and will help in optimizing the overall performance of the Linux kernel.
展开▼
