Re: [PATCH v3 06/20] sched/core: allow only preferred CPUs in is_cpu_allowed

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

On 6/4/26 10:36 AM, K Prateek Nayak wrote:
> Hello Shrikanth,
> 
> On 5/14/2026 8:51 PM, Shrikanth Hegde wrote:
>> +	/*
>> +	 * This is essential to maintain user affinities when preferred
>> +	 * CPUs change. A task pinned on non-preferred CPU should continue
>> +	 * to run there, since this is non-user triggered.
>> +	 *
>> +	 * For majority of the cases this would still keep select_fallback_rq
>> +	 * as O(N). task_has_preferred_cpus which is O(N) is called only if
>> +	 * !cpu_preferred. Then task running there is expected to move out.
>> +	 * So subsequent it should run on preferred CPU. This becomes O(N**2)
>> +	 * only for tasks pinned only non preferred CPUs. That is rare case.
>> +	 */
>> +	task_has_preferred_cpu = !cpu_preferred(cpu) && task_has_preferred_cpus(p);
> 
> For this O(N**2) bit, can't we cache the result of cpumask_intersect()
> in the task_struct similar to "p->sched_task_hot" when entering
> select_fallback_rq() and clear it at the end?
> 

Ok. Let me try.

>> +
>>   	/* Non kernel threads are not allowed during either online or offline. */
>> -	if (!(p->flags & PF_KTHREAD))
>> -		return cpu_active(cpu);
>> +	if (!(p->flags & PF_KTHREAD)) {
>> +		if (!cpu_active(cpu))
>> +			return false;
>> +		if (task_has_preferred_cpu)
>> +			return false;
> 
> Shouldn't this return true at the end here for
> cpu_active() && !task_has_preferred_cpu?
> 
> Essentially, this:
> 
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 77a172f31862..3bf7e3c4d518 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -2489,6 +2489,7 @@ static inline bool is_cpu_allowed(struct task_struct *p, int cpu)
>   			return false;
>   		if (task_has_preferred_cpu)
>   			return false;
> +		return true;
>   	}
>   

Yes. Good catch. I missed it. it would have returned cpu_online()

>   	/* KTHREAD_IS_PER_CPU is always allowed. */
> ---
> 
> Otherwise, it starts evaluating kthread bits for user tasks.
> 
>> +	}
>>   
>>   	/* KTHREAD_IS_PER_CPU is always allowed. */
>>   	if (kthread_is_per_cpu(p))
>