How does the Linux kernel schedule CPU, between user space processes/threads and kernel tasks/jobs

Question

I was trying to understand the Linux process management and scheduling. I know that the scheduler schedules different processes based on priority/time slicing. But there are kernel tasks (I am not mentioning the process' system calls which takes the process to kernel mode) which have to be processed as well (eg: scheduler/timers or some kernel code which runs forever). I did not understand if the scheduler schedules CPU for different processes how the kernel tasks are serviced in between.

Answer 1

Kernel tasks which aren’t run “in process” (to service a system call, or an interrupt) are handled as separate processes themselves, and you can see them in ps’s output:

root 2 0.0 0.0 0 0 ? S Sep16 0:02 [kthreadd] root 3 0.0 0.0 0 0 ? I< Sep16 0:00 [rcu_gp] root 4 0.0 0.0 0 0 ? I< Sep16 0:00 [rcu_par_gp] root 6 0.0 0.0 0 0 ? I< Sep16 0:00 [kworker/0:0H-kblockd] root 8 0.0 0.0 0 0 ? I< Sep16 0:00 [mm_percpu_wq] root 9 0.0 0.0 0 0 ? S Sep16 9:11 [ksoftirqd/0] root 10 0.2 0.0 0 0 ? I Sep16 173:25 [rcu_sched] root 11 0.0 0.0 0 0 ? I Sep16 0:00 [rcu_bh] root 12 0.0 0.0 0 0 ? S Sep16 0:20 [migration/0] root 14 0.0 0.0 0 0 ? S Sep16 0:00 [cpuhp/0] root 15 0.0 0.0 0 0 ? S Sep16 0:00 [cpuhp/1] 

These processes are scheduled in the same way as processes you’re more familiar with.

A common pattern for such tasks is workqueues; the kernel documentation for those is quite good, I encourage you to read it if you’re interested in the topic.