x86/aperfmperf: Erase stale arch_freq_scale values when disabling frequency invarianc...

Once disable_freq_invariance_work is called the scale_freq_tick function
will not compute or update the arch_freq_scale values.
However the scheduler will still read these values and use them.
The result is that the scheduler might perform unfair decisions based on stale
values.

This patch adds the step of setting the arch_freq_scale values for all
cpus to the default (max) value SCHED_CAPACITY_SCALE, Once all cpus
have the same arch_freq_scale value the scaling is meaningless.

Signed-off-by: Yair Podemsky <ypodemsk@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230110160206.75912-1-ypodemsk@redhat.com

diff --git a/arch/x86/kernel/cpu/aperfmperf.c b/arch/x86/kernel/cpu/aperfmperf.c
index 1f60a2b279368e94c79f64ea43f736baf966f39f..fdbb5f07448fad0dd267868fdfb0a6394d9e45b9 100644 (file)
--- a/arch/x86/kernel/cpu/aperfmperf.c
+++ b/arch/x86/kernel/cpu/aperfmperf.c
@@ -330,7 +330,16 @@ static void __init bp_init_freq_invariance(void)
 
 static void disable_freq_invariance_workfn(struct work_struct *work)
 {
+       int cpu;
+
        static_branch_disable(&arch_scale_freq_key);
+
+       /*
+        * Set arch_freq_scale to a default value on all cpus
+        * This negates the effect of scaling
+        */
+       for_each_possible_cpu(cpu)
+               per_cpu(arch_freq_scale, cpu) = SCHED_CAPACITY_SCALE;
 }
 
 static DECLARE_WORK(disable_freq_invariance_work,