21.04.2017, 19:19, "Stephen Brennan" <stephen@brennan.io>:

Hello all,

Hello,

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.

I written to the code for myself, you can use and develop this code (open source). https://github.com/mueddib/cpustat

Thanks, Stephen

Regards, Ozgur