<div dir="ltr">Hi,<div><br></div><div>My machine is running CentOS 7 with 3.10.0-1127.19.1.el7.x86_64, and I noticed that my application (it is a dummy C++ application with an infinite tight loop calling rand()) keeps getting rescheduling interrupts per /proc/interrupts and only nr_involuntary_switches is increasing (per /proc/<pid>/sched) by 2 every second.</div><div><br></div><div>My kernel parameters:</div><div>rcu_nocbs=2-17 isolcpus=2-17 rcu_nocb_poll irqaffinity=0 noapic nowatchdog<br></div><div><br></div><div>Tho, bcc tool shows that someone is still calling apic_timer_interrupt() (although kernel parameter above says noapic):</div><div># taskset -c 6 /usr/share/bcc/tools/stackcount -p 2496 native_smp_send_reschedule<br>Tracing 1 functions for "native_smp_send_reschedule"... Hit Ctrl-C to end.<br>^C<br>  native_smp_send_reschedule<br>  activate_task<br>  ttwu_do_activate<br>  try_to_wake_up<br>  wake_up_process<br>  wake_up_worker<br>  insert_work<br>  __queue_work<br>  delayed_work_timer_fn<br>  call_timer_fn<br>  run_timer_softirq<br>  __do_softirq<br>  call_softirq<br>  do_softirq<br>  irq_exit<br>  smp_apic_timer_interrupt<br>  apic_timer_interrupt<br>  [unknown]<br>  [unknown]<br>    1<br><br>  native_smp_send_reschedule<br>  activate_task<br>  ttwu_do_activate<br>  try_to_wake_up<br>  wake_up_process<br>  wake_up_worker<br>  insert_work<br>  __queue_work<br>  delayed_work_timer_fn<br>  call_timer_fn<br>  run_timer_softirq<br>  __do_softirq<br>  call_softirq<br>  do_softirq<br>  irq_exit<br>  smp_apic_timer_interrupt<br>  apic_timer_interrupt<br>  [unknown]<br>  [unknown]<br>    1<br></div><div><br></div><div>Any idea how to turn off this interrupt?  I have another Skylake machine (different CPU from the tested one above but same architecture -- Skylake), and it has no such issue at all.</div><div><br></div><div>Thanks!</div><div><br></div><div><br></div><div>Cheers,</div><div>Henry</div></div>