Re: [PATCH] sched: remove rcu_read_lock from wake_affine

[Peter Zijlstra <peterz@infradead.org>]

On Tue, 2011-06-07 at 15:43 +0530, Nikunj A. Dadhania wrote:
> wake_affine is called from one path: select_task_rq_fair, which already has
> rcu read lock held.
> 
> Signed-off-by: Nikunj A. Dadhania <nikunj@linux.vnet.ibm.com>
> ---
>  kernel/sched_fair.c |    3 +--
>  1 files changed, 1 insertions(+), 2 deletions(-)
> 
> diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c
> index 354e26b..0bfec93 100644
> --- a/kernel/sched_fair.c
> +++ b/kernel/sched_fair.c
> @@ -1461,6 +1461,7 @@ static inline unsigned long effective_load(struct task_group *tg, int cpu,
>  
>  #endif
>  
> +/* Assumes rcu_read_lock is held */

Not a big fan of such comments, esp with CONFIG_PROVE_RCU its better to
use those facilities, which is to say: if we're missing a
rcu_read_lock() the thing will yell bloody murder.

>  static int wake_affine(struct sched_domain *sd, struct task_struct *p, int sync)
>  {
>  	s64 this_load, load;
> @@ -1481,7 +1482,6 @@ static int wake_affine(struct sched_domain *sd, struct task_struct *p, int sync)
>  	 * effect of the currently running task from the load
>  	 * of the current CPU:
>  	 */
> -	rcu_read_lock();
>  	if (sync) {
>  		tg = task_group(current);
>  		weight = current->se.load.weight;
> @@ -1517,7 +1517,6 @@ static int wake_affine(struct sched_domain *sd, struct task_struct *p, int sync)
>  		balanced = this_eff_load <= prev_eff_load;
>  	} else
>  		balanced = true;
> -	rcu_read_unlock();
>  
>  	/*
>  	 * If the currently running task will sleep within
> 

OK, took the patch and removed the comment, thanks!