Re: [PATCH v3 14/20] sched/debug: Provide debugfs to enable/disable steal monitor

[Shrikanth Hegde <sshegde@linux.ibm.com>]

On 6/8/26 11:07 PM, Ilya Leoshkevich wrote:
> 
> 
> On 5/14/26 17:21, Shrikanth Hegde wrote:
>> Add a debugfs "enable" file to enable steal time monitor.
>>
>> Computing steal time and acting on it periodically are to be opted by
>> the user. This helps to avoid any overhead when the feature
>> is disabled.
>>
>> It is disabled by default.
>>
>> Signed-off-by: Shrikanth Hegde <sshegde@linux.ibm.com>
>> ---
>>   kernel/sched/core.c  |  1 +
>>   kernel/sched/debug.c | 29 +++++++++++++++++++++++++++++
>>   kernel/sched/sched.h |  2 ++
>>   3 files changed, 32 insertions(+)
>>
>> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
>> index 0d1995c65ce6..1533a44d1a6b 100644
>> --- a/kernel/sched/core.c
>> +++ b/kernel/sched/core.c
>> @@ -11270,6 +11270,7 @@ void sched_change_end(struct sched_change_ctx 
>> *ctx)
>>   #ifdef CONFIG_PREFERRED_CPU
>>   /* npc - non preferred CPU */
>>   static DEFINE_PER_CPU(struct cpu_stop_work, npc_push_task_work);
>> +DEFINE_STATIC_KEY_FALSE(__sched_sm_enable);
>>   static int sched_non_preferred_cpu_push_stop(void *arg)
>>   {
>> diff --git a/kernel/sched/debug.c b/kernel/sched/debug.c
>> index b1abfdc168bf..be8d223b43fd 100644
>> --- a/kernel/sched/debug.c
>> +++ b/kernel/sched/debug.c
>> @@ -592,6 +592,33 @@ static void debugfs_ext_server_init(void)
>>   #endif /* CONFIG_SCHED_CLASS_EXT */
>>   #ifdef CONFIG_PREFERRED_CPU
>> +__read_mostly bool sched_sm_wr_enable;
>> +
>> +static ssize_t sched_sm_en_write(struct file *filp, const char __user 
>> *ubuf,
>> +                 size_t cnt, loff_t *ppos)
>> +{
>> +    bool orig = sched_sm_wr_enable;
>> +    ssize_t result;
>> +
>> +    result = debugfs_write_file_bool(filp, ubuf, cnt, ppos);
>> +
>> +    if (sched_sm_wr_enable && !orig) {
>> +        static_branch_enable(&__sched_sm_enable);
>> +    } else if (!sched_sm_wr_enable && orig) {
>> +        static_branch_disable(&__sched_sm_enable);
>> +        cpumask_copy(&__cpu_preferred_mask, cpu_online_mask);
>> +    }
> 
> What if two userspace processes write to this debugfs file concurrently?
> I think __sched_sm_enable and __cpu_preferred_mask may go out of sync.
> 

I don't know how realistic is that use-case.

But, similar case may be present if CPU hotplug and sched_sm_en_write
maybe called in parallel.

I can add cpus_read_lock, so that only one can proceed.
Does that make sense? This will protect against parallel hotplug too.