Re: [PATCH] sched/cache: honor migrate_llc_task semantics in active load balance

["Chen, Yu C" <yu.c.chen@intel.com>]

Hi Lu Wang, Tim,

On 8/1/2026 8:22 PM, Lu Wang wrote:
> A passive load-balance pass marks group_llc_balance as migrate_llc_task
> and queues active balance when it cannot move a task. The CPU stopper
> callback constructs a fresh lb_env, so preserve the migration type on
> the runqueue across the asynchronous boundary.
> 
> For CAS-directed active balance, reject a candidate whose preferred LLC
> does not match the destination LLC. This keeps the fallback from moving
> a task away from its preferred LLC.
> 

It looks like this proposal provides fine-grain control on per-task base
migration strategy is promising.

> +static inline bool
> +migrate_llc_task_wrong_dst(struct task_struct *p, struct lb_env *env)
> +{
> +	return sched_cache_enabled() &&
> +	       env->migration_type == migrate_llc_task &&
> +	       READ_ONCE(p->preferred_llc) != llc_id(env->dst_cpu);
> +}
> +

[ ... ]

> @@ -13654,6 +13672,7 @@ static int active_load_balance_cpu_stop(void *data)
>   			.src_rq		= busiest_rq,
>   			.idle		= CPU_IDLE,
>   			.flags		= LBF_ACTIVE_LB,
> +			.migration_type	= (enum migration_type)busiest_rq->active_balance_type,

If we overwrite migration_type for ALB (default is 0, i.e. migrate_load),
then in can_migrate_task() a delayed task might not be migrated in ALB:

if ((p->se.sched_delayed) && (env->migration_type != migrate_load))
	return 0;
So an enhanced approach I'm thinking of is to pass
migrate_llc_task information via env->flags:
>   		};
>   
>   		schedstat_inc(sd->alb_count);
> diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
> index 56acf502b..82084d405 100644
> --- a/kernel/sched/sched.h
> +++ b/kernel/sched/sched.h
> @@ -1266,6 +1266,7 @@ struct rq {
>   	/* For active balancing */
>   	int			active_balance;
>   	int			push_cpu;
> +	int			active_balance_type; /* enum migration_type */

We can set env->flags by invoking different callbacks of 
stop_one_cpu_nowait(),
thus avoiding the need to introduce active_balance_type into rq - which 
could
cause false sharing if cache-line alignment is broken.

something like:

#define LBF_ACTIVE_LB_LLC      0x40

-static int active_load_balance_cpu_stop(void *data)
+static int __active_load_balance_cpu_stop(void *data, unsigned int 
lb_flags)
-static int active_load_balance_cpu_stop(void *data)
+static int __active_load_balance_cpu_stop(void *data, unsigned int 
lb_flags)
  {
         struct rq *busiest_rq = data;
         int busiest_cpu = cpu_of(busiest_rq);
@@ -13659,7 +13687,7 @@ static int active_load_balance_cpu_stop(void *data)
                         .src_cpu        = busiest_rq->cpu,
                         .src_rq         = busiest_rq,
                         .idle           = CPU_IDLE,
-                       .flags          = LBF_ACTIVE_LB,
+                       .flags          = LBF_ACTIVE_LB | lb_flags,
                 };

and in migrate_llc_task_wrong_dst(), we check env->flags & LBF_ACTIVE_LB_LLC
instead.

static int active_load_balance_cpu_stop(void *data)
{
         return __active_load_balance_cpu_stop(data, 0);
}

static int active_load_balance_llc_cpu_stop(void *data)
{
         return __active_load_balance_cpu_stop(data, LBF_ACTIVE_LB_LLC); 
<-- new flag
}

static inline cpu_stop_fn_t alb_stop_fn(struct lb_env *env)
{
         if (env->migration_type == migrate_llc_task)
                 return active_load_balance_llc_cpu_stop;

         return active_load_balance_cpu_stop;
}

if (active_balance) {
	stop_one_cpu_nowait(cpu_of(busiest),
		alb_stop_fn(&env), busiest,
		&busiest->active_balance_work);
}


thanks,
Chenyu

>   	struct cpu_stop_work	active_balance_work;
>   
>   	/* CPU of this runqueue: */