linux-kernel - Re: [PATCH v4 2/2] sched/fair: Use helper function to access rd->overutilized

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: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date: Tue, 5 Mar 2024 20:33:14 +0530
From: Shrikanth Hegde <sshegde@...ux.ibm.com>
To: Vincent Guittot <vincent.guittot@...aro.org>
Cc: mingo@...nel.org, peterz@...radead.org, yu.c.chen@...el.com,
        dietmar.eggemann@....com, linux-kernel@...r.kernel.org,
        nysal@...ux.ibm.com, aboorvad@...ux.ibm.com, srikar@...ux.ibm.com,
        vschneid@...hat.com, pierre.gondois@....com, morten.rasmussen@....com,
        qyousef@...alina.io
Subject: Re: [PATCH v4 2/2] sched/fair: Use helper function to access
 rd->overutilized



On 3/5/24 7:35 PM, Vincent Guittot wrote:
> On Fri, 1 Mar 2024 at 16:18, Shrikanth Hegde <sshegde@...ux.ibm.com> wrote:
>
> 
> It seems that is_rd_overutilized() is always used with
> sched_energy_enabled() in the pattern:
> 
> If (sched_energy_enabled() && !is_rd_overutilized(rd))
>        do something
> 
> This pattern includes feec() case where we have in select_task_rq_fair():
> 
> If (sched_energy_enabled())
>        feec():
>        |->  if (is_rd_overutilized())
>        |->       goto unlock
> 
> which could be changed into
> If (sched_energy_enabled() && !is_rd_overutilized(rd))
>        feec()
> 
> Then you can create the function is_rd_not_overutilized() instead of
> is_rd_overutilized()
> 
> -static inline int is_rd_overutilized(struct root_domain *rd)
> +static inline int is_rd_not_overutilized(struct root_domain *rd)
>  {
> -       return READ_ONCE(rd->overutilized);
> +       return sched_energy_enabled() && READ_ONCE(rd->overutilized);
>  }
> 
> and use is_rd_not_overutilized() instead
> 

Ok. Makes sense. I will keep this patch as is. and use the above 
approach in a new patch. 

>> +
>>  static inline void set_rd_overutilized_status(struct root_domain *rd,
>>                                               unsigned int status)
>>  {
>> @@ -6686,13 +6695,14 @@ static inline void check_update_overutilized_status(struct rq *rq)
>>         if (!sched_energy_enabled())
>>                 return;
>>
>> -       if (!READ_ONCE(rq->rd->overutilized) && cpu_overutilized(rq->cpu))
>> +       if (!is_rd_overutilized(rq->rd) && cpu_overutilized(rq->cpu))
>>                 set_rd_overutilized_status(rq->rd, SG_OVERUTILIZED);
>>  }
>>  #else
>>  static inline void check_update_overutilized_status(struct rq *rq) { }
>>  static inline void set_rd_overutilized_status(struct root_domain *rd,
>>                                               unsigned int status) { }
>> +static inline int is_rd_overutilized(struct root_domain *rd) { }
> 
> It should be
> static inline int is_rd_overutilized(struct root_domain *rd) { return 0; }

ok. 

> 
>>  #endif
>>
>>  /* Runqueue only has SCHED_IDLE tasks enqueued */
>> @@ -7974,7 +7984,7 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu)
>>
>>         rcu_read_lock();

>>