April 21, 2017
4:17 p.m.
Hello all, I was wondering if anyone had some information on how CPU time used by the kernel is accounted for, e.g. in top or htop. I would imagine that kernel code running in process code would be "billed" toward that process. What about kernel worker threads (for example, work_structs queued for later)? My overall goal is that I'm doing some large-scale network emulation, and I want to ensure that the kernel is not using too much CPU time, otherwise the CPU overhead may be messing with the results. Thanks, Stephen