Hi,
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.
My kernel parameters:
rcu_nocbs=2-17 isolcpus=2-17 rcu_nocb_poll irqaffinity=0 noapic nowatchdog
Tho, bcc tool shows that someone is still calling apic_timer_interrupt()
(although kernel parameter above says noapic):
# taskset -c 6 /usr/share/bcc/tools/stackcount -p 2496
native_smp_send_reschedule
Tracing 1 functions for "native_smp_send_reschedule"... Hit Ctrl-C to end.
^C
native_smp_send_reschedule
activate_task
ttwu_do_activate
try_to_wake_up
wake_up_process
wake_up_worker
insert_work
__queue_work
delayed_work_timer_fn
call_timer_fn
run_timer_softirq
__do_softirq
call_softirq
do_softirq
irq_exit
smp_apic_timer_interrupt
apic_timer_interrupt
[unknown]
[unknown]
1
native_smp_send_reschedule
activate_task
ttwu_do_activate
try_to_wake_up
wake_up_process
wake_up_worker
insert_work
__queue_work
delayed_work_timer_fn
call_timer_fn
run_timer_softirq
__do_softirq
call_softirq
do_softirq
irq_exit
smp_apic_timer_interrupt
apic_timer_interrupt
[unknown]
[unknown]
1
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.
Thanks!
Cheers,
Henry
_______________________________________________
Kernelnewbies mailing list
Kernelnewbies@kernelnewbies.org
https://lists.kernelnewbies.org/mailman/listinfo/kernelnewbies
