Re: [PATCH v3 4/4] sched/fair: Check the SD_ASYM_PACKING flag in sched_use_asym_prio()

[kuiliang Shi]

On 2/6/24 6:38 AM, Ricardo Neri wrote:
> On Thu, Feb 01, 2024 at 07:54:47PM +0800, alexs@xxxxxxxxxx wrote:
>> From: Alex Shi <alexs@xxxxxxxxxx>
>>
>> sched_use_asym_prio() checks whether CPU priorities should be used. It
>> makes sense to check for the SD_ASYM_PACKING() inside the function.
>> Since both sched_asym() and sched_group_asym() use sched_use_asym_prio(),
>> remove the now superfluous checks for the flag in various places"
> 
> s/places"/places./
> 
>>
>> Signed-off-by: Alex Shi <alexs@xxxxxxxxxx>
>> To: Ricardo Neri <ricardo.neri-calderon@xxxxxxxxxxxxxxx>
>> To: Ben Segall <bsegall@xxxxxxxxxx>
>> To: Steven Rostedt <rostedt@xxxxxxxxxxx>
>> To: Dietmar Eggemann <dietmar.eggemann@xxxxxxx>
>> To: Valentin Schneider <vschneid@xxxxxxxxxx>
>> To: Daniel Bristot de Oliveira <bristot@xxxxxxxxxx>
>> To: Vincent Guittot <vincent.guittot@xxxxxxxxxx>
>> To: Juri Lelli <juri.lelli@xxxxxxxxxx>
>> To: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
>> To: Ingo Molnar <mingo@xxxxxxxxxx>
>> ---
>>  kernel/sched/fair.c | 15 +++++++--------
>>  1 file changed, 7 insertions(+), 8 deletions(-)
>>
>> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
>> index 44fd5e2ca642..bd32efbea688 100644
>> --- a/kernel/sched/fair.c
>> +++ b/kernel/sched/fair.c
>> @@ -9741,6 +9741,9 @@ group_type group_classify(unsigned int imbalance_pct,
>>   */
>>  static bool sched_use_asym_prio(struct sched_domain *sd, int cpu)
>>  {
>> +	if (!(sd->flags & SD_ASYM_PACKING))
>> +		return false;
>> +
>>  	if (!sched_smt_active())
>>  		return true;
>>  
>> @@ -9940,11 +9943,9 @@ static inline void update_sg_lb_stats(struct lb_env *env,
>>  	sgs->group_weight = group->group_weight;
>>  
>>  	/* Check if dst CPU is idle and preferred to this group */
>> -	if (!local_group && env->sd->flags & SD_ASYM_PACKING &&
>> -	    env->idle != CPU_NOT_IDLE && sgs->sum_h_nr_running &&
>> -	    sched_group_asym(env, sgs, group)) {
>> +	if (!local_group && env->idle != CPU_NOT_IDLE && sgs->sum_h_nr_running &&
>> +			sched_group_asym(env, sgs, group))
> 
> You should align sched_group_asym() to !local_group.

Thanks for all suggestion. I will take them in next version.

> 
>>  		sgs->group_asym_packing = 1;
>> -	}
>>  
>>  	/* Check for loaded SMT group to be balanced to dst CPU */
>>  	if (!local_group && smt_balance(env, sgs, group))
>> @@ -11040,9 +11041,7 @@ static struct rq *find_busiest_queue(struct lb_env *env,
>>  		 * If balancing between cores, let lower priority CPUs help
>>  		 * SMT cores with more than one busy sibling.
>>  		 */
>> -		if ((env->sd->flags & SD_ASYM_PACKING) &&
>> -		    sched_asym(env->sd, i, env->dst_cpu) &&
>> -		    nr_running == 1)
>> +		if (sched_asym(env->sd, i, env->dst_cpu) && nr_running == 1)
>>  			continue;
>>  
>>  		switch (env->migration_type) {
>> @@ -11138,7 +11137,7 @@ asym_active_balance(struct lb_env *env)
>>  	 * the lower priority @env::dst_cpu help it. Do not follow
>>  	 * CPU priority.
>>  	 */
>> -	return env->idle != CPU_NOT_IDLE && (env->sd->flags & SD_ASYM_PACKING) &&
>> +	return env->idle != CPU_NOT_IDLE &&
>>  	       sched_use_asym_prio(env->sd, env->dst_cpu) &&
>>  	       (sched_asym_prefer(env->dst_cpu, env->src_cpu) ||
> 
> Perhaps you can rearrange the spaghetti of conditions to make better use of
> the full 80-column line.

80-column doesn't work here, will try 100 column.

Thanks
Alex