What were you trying to do that didn't work?

On SMP systems, inter-processor interrupts are triggered by TuneD when reloading TuneD even when using the same profile after the reload. This causes unwanted disruption (latency spikes) on latency-sensitive workloads.

Please provide the package NVR for which bug is seen:

1. rpm -q tuned
   tuned-2.22.0-0.1.rc1.el9.noarch
2. uname -r
   5.14.0-284.30.1.rt14.315.el9_2.x86_64

How reproducible:

Always when using the affected TuneD plug-in and plug-in options.

Steps to reproduce

1. systemctl enable tuned --now
2. stress-ng -c 1 --taskset 1
3. trace-cmd record -e 'call_function*' -f 'CPU == 1' sleep 10
4. pkill -e -HUP tuned
5. trace-cmd report

Expected results

No unnecessary IPIs during TuneD reload/restart.

Actual results

No IPIs found during reload in the following plug-ins: [acpi], [bootloader], [disk], [irqbalance], [net], [scheduler], [scsi_host], [selinux], [service], [systemd]
Plug-ins not tested: [audio], [eeepc_she], [modules], [mounts], [script], [usb], [video], [vm]

Plug-ins with IPIs found during profile reload.

[rtentsk]
=========
When reloading the same profile while using the [rtentsk] plug-in triggers
a lagre amount of IPIs, but they all seem to be in 0-2us range.

[cpu]
=====
1) governor; [1-30us] even when writing the same/existing value causes IPIs

stress-ng-cpu-7772 [001] 12211.893188: call_function_single_entry: vector=251
stress-ng-cpu-7772 [001] 12211.893218: call_function_single_exit: vector=251

echo performance > /sys/devices/system/cpu/cpufreq/policy1/scaling_governor

2) energy_perf_bias; [3-4us] even when writing the same/existing value causes IPIs; IPIs are also caused when reading the value of energy_perf_bias of the CPU the workload is running on.
echo performance > /sys/devices/system/cpu/cpu1/power/energy_perf_bias

3) min_perf_pct; [0-6us] even when writing the same/existing value causes IPIs;
echo 23 > /sys/devices/system/cpu/intel_pstate/min_perf_pct

4) force_latency; [3-6us] calling x86_energy_perf_policy [-r] causes IPIs

5) energy_performance_preference; [1-2us] even when writing the same/existing value causes IPIs; IPIs are also caused when reading the value of energy_performance_preference of the CPU the workload is running on.
echo performance > /sys/devices/system/cpu/cpufreq/policy1/energy_performance_preference

6) no_turbo; [1-7us] even when writing the same/existing value
echo 0 > /sys/devices/system/cpu/intel_pstate/no_turbo

[sys]
=====
changing a value of sysctls can also cause [1-4us] IPIs; examples:
kernel.nmi_watchdog
kernel.timer_migration

[sysfs]
=======
changing a value of a sysfs file can also cause [1-6us] IPIs; examples:
/sys/devices/system/cpu/cpufreq/policy1/scaling_min_freq
/sys/devices/system/cpu/cpufreq/policy1/scaling_max_freq