Re: [PATCH] sched/proc: Print user_cpus_ptr for task status

[Waiman Long <longman@redhat.com>]

On 5/7/24 02:57, Xuewen Yan wrote:
>> These changes essentially reverts commit 851a723e45d1c("sched: Always
>> clear user_cpus_ptr in do_set_cpus_allowed()") except the additional
>> caller in the cpuset code.
>>
>> How about the following less invasive change?
>>
>>    diff --git a/kernel/sched/core.c b/kernel/sched/core.c
>> index 7019a40457a6..646837eab70c 100644
>> --- a/kernel/sched/core.c
>> +++ b/kernel/sched/core.c
>> @@ -2796,21 +2796,24 @@ __do_set_cpus_allowed(struct task_struct *p,
>> struct affinity_context *ctx)
>>    }
>>
>>    /*
>> - * Used for kthread_bind() and select_fallback_rq(), in both cases the user
>> - * affinity (if any) should be destroyed too.
>> + * Used for kthread_bind() and select_fallback_rq(). Destroy user affinity
>> + * if no intersection with the new mask.
>>     */
>>    void do_set_cpus_allowed(struct task_struct *p, const struct cpumask
>> *new_mask)
>>    {
>>           struct affinity_context ac = {
>>                   .new_mask  = new_mask,
>>                   .user_mask = NULL,
>> -               .flags     = SCA_USER,  /* clear the user requested mask */
>> +               .flags     = 0,
>>           };
>>           union cpumask_rcuhead {
>>                   cpumask_t cpumask;
>>                   struct rcu_head rcu;
>>           };
>>
>> +       if (current->user_cpus_ptr &&
>> !cpumask_intersects(current->user_cpus_ptr, new_mask))
> Thanks for your suggestion, and I try it and as for me, it works well,
> but I change the "current" to p.
> I think “current” is inappropriate because what is changed here is the
> mask of p.
> It is possible that “p” and “current” are not equal.
>
> I would send the next patch later and add your Suggested-by. Thanks
> again for your advice!

You are right. It should be "p" instead of "current".

Thanks,
Longman