Re: [PATCH v2] sched: cpuset: Don't rebuild sched domains on suspend-resume

[Waiman Long <longman@redhat.com>]

On 1/29/23 21:49, Waiman Long wrote:
> On 1/25/23 11:35, Qais Yousef wrote:
>> On 01/20/23 17:16, Waiman Long wrote:
>>> On 1/20/23 14:48, Qais Yousef wrote:
>>>> Commit f9a25f776d78 ("cpusets: Rebuild root domain deadline 
>>>> accounting information")
>>>> enabled rebuilding sched domain on cpuset and hotplug operations to
>>>> correct deadline accounting.
>>>>
>>>> Rebuilding sched domain is a slow operation and we see 10+ ms delay on
>>>> suspend-resume because of that.
>>>>
>>>> Since nothing is expected to change on suspend-resume operation; skip
>>>> rebuilding the sched domains to regain the time lost.
>>>>
>>>> Debugged-by: Rick Yiu <rickyiu@google.com>
>>>> Signed-off-by: Qais Yousef (Google) <qyousef@layalina.io>
>>>> ---
>>>>
>>>>       Changes in v2:
>>>>           * Remove redundant check in update_tasks_root_domain() 
>>>> (Thanks Waiman)
>>>>       v1 link:
>>>> https://lore.kernel.org/lkml/20221216233501.gh6m75e7s66dmjgo@airbuntu/
>>>>
>>>>    kernel/cgroup/cpuset.c  | 3 +++
>>>>    kernel/sched/deadline.c | 3 +++
>>>>    2 files changed, 6 insertions(+)
>>>>
>>>> diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
>>>> index a29c0b13706b..9a45f083459c 100644
>>>> --- a/kernel/cgroup/cpuset.c
>>>> +++ b/kernel/cgroup/cpuset.c
>>>> @@ -1088,6 +1088,9 @@ static void rebuild_root_domains(void)
>>>>        lockdep_assert_cpus_held();
>>>>        lockdep_assert_held(&sched_domains_mutex);
>>>> +    if (cpuhp_tasks_frozen)
>>>> +        return;
>>>> +
>>>>        rcu_read_lock();
>>>>        /*
>>>> diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
>>>> index 0d97d54276cc..42c1143a3956 100644
>>>> --- a/kernel/sched/deadline.c
>>>> +++ b/kernel/sched/deadline.c
>>>> @@ -2575,6 +2575,9 @@ void dl_clear_root_domain(struct root_domain 
>>>> *rd)
>>>>    {
>>>>        unsigned long flags;
>>>> +    if (cpuhp_tasks_frozen)
>>>> +        return;
>>>> +
>>>>        raw_spin_lock_irqsave(&rd->dl_bw.lock, flags);
>>>>        rd->dl_bw.total_bw = 0;
>>>>        raw_spin_unlock_irqrestore(&rd->dl_bw.lock, flags);
>>> cpuhp_tasks_frozen is set when thaw_secondary_cpus() or
>>> freeze_secondary_cpus() is called. I don't know the exact 
>>> suspend/resume
>>> calling sequences, will cpuhp_tasks_frozen be cleared at the end of 
>>> resume
>>> sequence? Maybe we should make sure that rebuild_root_domain() is 
>>> called at
>>> least once at the end of resume operation.
>> Very good questions. It made me look at the logic again and I realize 
>> now that
>> the way force_build behaves is causing this issue.
>>
>> I *think* we should just make the call rebuild_root_domains() only if
>> cpus_updated in cpuset_hotplug_workfn().
>>
>> cpuset_cpu_active() seems to be the source of force_rebuild in my 
>> case; which
>> seems to be called only after the last cpu is back online (what you 
>> suggest).
>> In this case we can end up with cpus_updated = false, but 
>> force_rebuild = true.
>>
>> Now you added a couple of new users to force_rebuild in 
>> 4b842da276a8a; I'm
>> trying to figure out what the conditions would be there. It seems we 
>> can have
>> corner cases for cpus_update might not trigger correctly?
>>
>> Could the below be a good cure?
>>
>> AFAICT we must rebuild the root domains if something has changed in 
>> cpuset.
>> Which should be captured by either having:
>>
>>     * cpus_updated = true
>>     * force_rebuild && !cpuhp_tasks_frozen
>>
>> /me goes to test the patch
>>
>> --->8---
>>
>>     diff --git a/kernel/cgroup/cpuset.c b/kernel/cgroup/cpuset.c
>>     index a29c0b13706b..363e4459559f 100644
>>     --- a/kernel/cgroup/cpuset.c
>>     +++ b/kernel/cgroup/cpuset.c
>>     @@ -1079,6 +1079,8 @@ static void update_tasks_root_domain(struct 
>> cpuset *cs)
>>         css_task_iter_end(&it);
>>      }
>>
>>     +static bool need_rebuild_rd = true;
>>     +
>>      static void rebuild_root_domains(void)
>>      {
>>         struct cpuset *cs = NULL;
>>     @@ -1088,6 +1090,9 @@ static void rebuild_root_domains(void)
>>         lockdep_assert_cpus_held();
>>         lockdep_assert_held(&sched_domains_mutex);
>>
>>     +       if (!need_rebuild_rd)
>>     +               return;
>>     +
>>         rcu_read_lock();
>>
>>         /*
>>     @@ -3627,7 +3632,9 @@ static void cpuset_hotplug_workfn(struct 
>> work_struct *work)
>>         /* rebuild sched domains if cpus_allowed has changed */
>>         if (cpus_updated || force_rebuild) {
>>             force_rebuild = false;
>>     +               need_rebuild_rd = cpus_updated || (force_rebuild 
>> && !cpuhp_tasks_frozen);
>>             rebuild_sched_domains();
>>     +               need_rebuild_rd = true;
>
> You do the force_check check after it is set to false in the previous 
> statement which is definitely not correct. So it will be false 
> whenever cpus_updated is false.
>
> If you just want to skip rebuild_sched_domains() call for hotplug, why 
> don't just skip the call here if the condition is right? Like
>
>         /* rebuild sched domains if cpus_allowed has changed */
>         if (cpus_updated || (force_rebuild && !cpuhp_tasks_frozen)) {
>                 force_rebuild = false;
>                 rebuild_sched_domains();
>         }
>
> Still, we will need to confirm that cpuhp_tasks_frozen will be cleared 
> outside of the suspend/resume cycle.

BTW, you also need to expand the comment to explain why we need to check 
for cpuhp_tasks_frozen.

Cheers,
Longman