linux-kernel - Re: [PATCH 4/7] sched/fair: Take care of group/affinity/sched

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ad8197f1-548d-4fad-abd0-e8f6e9dbe12a@linux.dev>
Date: Fri, 23 May 2025 17:13:35 +0800
From: Chengming Zhou <chengming.zhou@...ux.dev>
To: Aaron Lu <ziqianlu@...edance.com>
Cc: Valentin Schneider <vschneid@...hat.com>, Ben Segall
 <bsegall@...gle.com>, K Prateek Nayak <kprateek.nayak@....com>,
 Peter Zijlstra <peterz@...radead.org>, Josh Don <joshdon@...gle.com>,
 Ingo Molnar <mingo@...hat.com>, Vincent Guittot
 <vincent.guittot@...aro.org>, Xi Wang <xii@...gle.com>,
 linux-kernel@...r.kernel.org, Juri Lelli <juri.lelli@...hat.com>,
 Dietmar Eggemann <dietmar.eggemann@....com>,
 Steven Rostedt <rostedt@...dmis.org>, Mel Gorman <mgorman@...e.de>,
 Chuyi Zhou <zhouchuyi@...edance.com>, Jan Kiszka <jan.kiszka@...mens.com>,
 Florian Bezdeka <florian.bezdeka@...mens.com>
Subject: Re: [PATCH 4/7] sched/fair: Take care of group/affinity/sched_class
 change for throttled task

On 2025/5/23 15:56, Aaron Lu wrote:
> On Fri, May 23, 2025 at 10:43:53AM +0800, Chengming Zhou wrote:
>> On 2025/5/20 18:41, Aaron Lu wrote:
>>> On task group change, for tasks whose on_rq equals to TASK_ON_RQ_QUEUED,
>>> core will dequeue it and then requeued it.
>>>
>>> The throttled task is still considered as queued by core because p->on_rq
>>> is still set so core will dequeue it, but since the task is already
>>> dequeued on throttle in fair, handle this case properly.
>>>
>>> Affinity and sched class change is similar.
>>
>> How about setting p->on_rq to 0 when throttled? which is the fact that
>> the task is not on cfs queue anymore, does this method cause any problem?
>>
> 
> On task group change/affinity change etc. if the throttled task is
> regarded as !on_rq, then it will miss the chance to be enqueued to the
> new(and correct) cfs_rqs, instead, it will be enqueued back to its
> original cfs_rq on unthrottle which breaks affinity or task group

Yeah, this is indeed a problem, I was thinking to delete the throttled task
from the cfs_rq limbo list, then add it to another cfs_rq limbo list or cfs_rq
runnable tree based on the new cfs_rq's throttle status.

But it's much complex compared with your current method.

> settings. We may be able to do something in tg_unthrottle_up() to take
> special care of these situations, but it seems a lot of headaches.
> 
> Also, for task group change, if the new task group does not have throttle
> setting, that throttled task should be allowed to run immediately instead
> of waiting for its old cfs_rq's unthrottle event. Similar is true when
> this throttled task changed its sched class, like from fair to rt.
> 
> Makes sense?

Ok, the another problem of the current method I can think of is the PELT maintenance,
we skip the actual dequeue_task_fair() process, which includes PELT detach, we just
delete it from the cfs_rq limbo list, so it can result in PELT maintenance error.

Thanks!

> 
> Thanks,
> Aaron
> 
>>>
>>> Signed-off-by: Aaron Lu <ziqianlu@...edance.com>
>>> ---
>>>    kernel/sched/fair.c | 24 ++++++++++++++++++++++++
>>>    1 file changed, 24 insertions(+)
>>>
>>> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
>>> index 74bc320cbc238..4c66fd8d24389 100644
>>> --- a/kernel/sched/fair.c
>>> +++ b/kernel/sched/fair.c
>>> @@ -5866,6 +5866,10 @@ static void throttle_cfs_rq_work(struct callback_head *work)
>>>    		update_rq_clock(rq);
>>>    		WARN_ON_ONCE(!list_empty(&p->throttle_node));
>>>    		dequeue_task_fair(rq, p, DEQUEUE_SLEEP | DEQUEUE_SPECIAL);
>>> +		/*
>>> +		 * Must not add it to limbo list before dequeue or dequeue will
>>> +		 * mistakenly regard this task as an already throttled one.
>>> +		 */
>>>    		list_add(&p->throttle_node, &cfs_rq->throttled_limbo_list);
>>>    		resched_curr(rq);
>>>    	}
>>> @@ -5881,6 +5885,20 @@ void init_cfs_throttle_work(struct task_struct *p)
>>>    	INIT_LIST_HEAD(&p->throttle_node);
>>>    }
>>> +static void dequeue_throttled_task(struct task_struct *p, int flags)
>>> +{
>>> +	/*
>>> +	 * Task is throttled and someone wants to dequeue it again:
>>> +	 * it must be sched/core when core needs to do things like
>>> +	 * task affinity change, task group change, task sched class
>>> +	 * change etc.
>>> +	 */
>>> +	WARN_ON_ONCE(p->se.on_rq);
>>> +	WARN_ON_ONCE(flags & DEQUEUE_SLEEP);
>>> +
>>> +	list_del_init(&p->throttle_node);
>>> +}
>>> +
>>>    static void enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags);
>>>    static int tg_unthrottle_up(struct task_group *tg, void *data)
>>>    {
>>> @@ -6834,6 +6852,7 @@ static inline void sync_throttle(struct task_group *tg, int cpu) {}
>>>    static __always_inline void return_cfs_rq_runtime(struct cfs_rq *cfs_rq) {}
>>>    static void task_throttle_setup_work(struct task_struct *p) {}
>>>    static bool task_is_throttled(struct task_struct *p) { return false; }
>>> +static void dequeue_throttled_task(struct task_struct *p, int flags) {}
>>>    static inline int cfs_rq_throttled(struct cfs_rq *cfs_rq)
>>>    {
>>> @@ -7281,6 +7300,11 @@ static int dequeue_entities(struct rq *rq, struct sched_entity *se, int flags)
>>>     */
>>>    static bool dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags)
>>>    {
>>> +	if (unlikely(task_is_throttled(p))) {
>>> +		dequeue_throttled_task(p, flags);
>>> +		return true;
>>> +	}
>>> +
>>>    	if (!(p->se.sched_delayed && (task_on_rq_migrating(p) || (flags & DEQUEUE_SAVE))))
>>>    		util_est_dequeue(&rq->cfs, p);