Re: [PATCH v7 1/3] sched/fair: Co-locate cfs_rq and sched_entity in cfs_tg_state

[K Prateek Nayak <kprateek.nayak@amd.com>]

Hello Zecheng,

On 1/18/2026 9:04 AM, Zecheng Li wrote:
> @@ -13617,10 +13617,11 @@ void free_fair_sched_group(struct task_group *tg)
>  	int i;
>  
>  	for_each_possible_cpu(i) {
> -		if (tg->cfs_rq)
> -			kfree(tg->cfs_rq[i]);
> -		if (tg->se)
> -			kfree(tg->se[i]);
> +		if (tg->cfs_rq && tg->cfs_rq[i]) {

nit. Since the cfs_tg_state uses __no_randomize_layout now and, "cfs_rq"
is the first member, you can just do kfree(tg->cfs_rq[i]) since it is
the beginning of cfs_tg_state too.

This is how kfree() on tg->se[i] would have freed the entire
"sched_entity_stats" previously.

I know this is a transient change since later you switch to using
free_percpu() but it would still be nice to have a simplified hunk here.

> +			struct cfs_tg_state *state =
> +				container_of(tg->cfs_rq[i], struct cfs_tg_state, cfs_rq);
> +			kfree(state);
> +		}
>  	}
>  
>  	kfree(tg->cfs_rq);

-- 
Thanks and Regards,
Prateek