Re: 5.6-rc3: WARNING: CPU: 48 PID: 17435 at kernel/sched/fair.c:380 enqueue_task_fair+0x328/0x440

[Christian Borntraeger <borntraeger@de.ibm.com>]

On 02.03.20 19:17, Christian Borntraeger wrote:
> On 02.03.20 12:16, Christian Borntraeger wrote:
>>
>>
>> On 28.02.20 17:35, Vincent Guittot wrote:
>>> Le vendredi 28 févr. 2020 à 16:42:27 (+0100), Christian Borntraeger a écrit :
>>>>
>>>>
>>>> On 28.02.20 16:37, Vincent Guittot wrote:
>>>>> On Fri, 28 Feb 2020 at 16:08, Christian Borntraeger
>>>>> <borntraeger@de.ibm.com> wrote:
>>>>>>
>>>>>> Also happened with 5.4:
>>>>>> Seems that I just happen to have an interesting test workload/system size interaction
>>>>>> on a newly installed system that triggers this.
>>>>>
>>>>> you will probably go back to 5.1 which is the version where we put
>>>>> back the deletion of unused cfs_rq from the list which can trigger the
>>>>> warning:
>>>>> commit 039ae8bcf7a5 : (Fix O(nr_cgroups) in the load balancing path)
>>>>>
>>>>> AFAICT, we haven't changed this since
>>>>
>>>> So you do know what is the problem? If not is there any debug option or
>>>> patch that I could apply to give you more information?
>>>
>>> No I don't know what is happening. Your test probably goes through an unexpected path
>>>
>>> Would it be difficult for me to reproduce your test env ?
>>
>> Not sure. Its a 32CPU (SMT2 -> 64) host. I have about 10 KVM guests running doing different
>> things.
>>
>>>
>>> There is an optimization in the code which could generate problem if assumption is not
>>> true. Could you try the patch below ?
>>>
>>> ---
>>>  kernel/sched/fair.c | 2 +-
>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
>>> index 3c8a379c357e..beb773c23e7d 100644
>>> --- a/kernel/sched/fair.c
>>> +++ b/kernel/sched/fair.c
>>> @@ -4035,8 +4035,8 @@ enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
>>>  		__enqueue_entity(cfs_rq, se);
>>>  	se->on_rq = 1;
>>>  
>>> +	list_add_leaf_cfs_rq(cfs_rq);
>>>  	if (cfs_rq->nr_running == 1) {
>>> -		list_add_leaf_cfs_rq(cfs_rq);
>>>  		check_enqueue_throttle(cfs_rq);
>>>  	}
>>>  }
>>
>> Now running for 3 hours. I have not seen the issue yet. I can tell tomorrow if this fixes 
>> the issue.
> 
> 
> Still running fine. I can tell for sure tomorrow, but I have the impression that this makes the
> WARN_ON go away.

So I guess this change "fixed" the issue. If you want me to test additional patches, let me know.