Re: [PATCH 7/8] kernel/fork: Only cache the VMAP stack in finish_task_switch().

[Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>]

On 2022-02-14 17:54:48 [+0100], To Andy Lutomirski wrote:
> index fcf0c180617c2..defe31036930a 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -4895,8 +4895,11 @@ static struct rq *finish_task_switch(struct task_struct *prev)
>  		if (prev->sched_class->task_dead)
>  			prev->sched_class->task_dead(prev);
>  
> -		/* Task is done with its stack. */
> -		put_task_stack(prev);
> +		/*
> +		 * Task is done with its stack. Try to cache VMAP stack and
> +		 * delay free it otherwise.
> +		 */
> +		put_task_stack_sched(prev);

Now that I write the commit message, there is probably nothing wrong
with unconditionally delaying it via RCU if caching failed. Then I don't
have to explain that there is one function is for the atomic context and
the other for non-atomic.

>  		put_task_struct_rcu_user(prev);
>  	}
> -- 
Sebastian