On Wed, Jul 31, 2024 at 09:14:52AM -1000, Tejun Heo wrote:
> p->scx.disallow provides a way for the BPF scheduler to reject certain tasks
> from attaching. It's currently allowed for both the load and fork paths;
> however, the latter doesn't actually work as p->sched_class is already set
> by the time scx_ops_init_task() is called during fork.
> 
> This is a convenience feature which is mostly useful from the load path
> anyway. Allow it only from the load path.
> 
> Signed-off-by: Tejun Heo <tj@kernel.org>
> Reported-by: "Zhangqiao (2012 lab)" <zhangqiao22@huawei.com>
> Link: http://lkml.kernel.org/r/20240711110720.1285-1-zhangqiao22@huawei.com
> Fixes: 7bb6f0810ecf ("sched_ext: Allow BPF schedulers to disallow specific tasks from joining SCHED_EXT")
> ---
>  include/linux/sched/ext.h |   11 ++++++-----
>  kernel/sched/ext.c        |   14 ++++++++------
>  2 files changed, 14 insertions(+), 11 deletions(-)
> 
> --- a/include/linux/sched/ext.h
> +++ b/include/linux/sched/ext.h
> @@ -181,11 +181,12 @@ struct sched_ext_entity {
>  	 * If set, reject future sched_setscheduler(2) calls updating the policy
>  	 * to %SCHED_EXT with -%EACCES.
>  	 *
> -	 * If set from ops.init_task() and the task's policy is already
> -	 * %SCHED_EXT, which can happen while the BPF scheduler is being loaded
> -	 * or by inhering the parent's policy during fork, the task's policy is
> -	 * rejected and forcefully reverted to %SCHED_NORMAL. The number of
> -	 * such events are reported through /sys/kernel/debug/sched_ext::nr_rejected.
> +	 * Can be set from ops.init_task() while the BPF scheduler is being
> +	 * loaded (!scx_init_task_args->fork). If set and the task's policy is
> +	 * already %SCHED_EXT, the task's policy is rejected and forcefully
> +	 * reverted to %SCHED_NORMAL. The number of such events are reported
> +	 * through /sys/kernel/debug/sched_ext::nr_rejected. Setting this flag
> +	 * during fork is not allowed.
>  	 */
>  	bool			disallow;	/* reject switching into SCX */
>  
> --- a/kernel/sched/ext.c
> +++ b/kernel/sched/ext.c
> @@ -3399,18 +3399,17 @@ static int scx_ops_init_task(struct task
>  
>  	scx_set_task_state(p, SCX_TASK_INIT);
>  
> -	if (p->scx.disallow) {
> +	if (!fork && p->scx.disallow) {
>  		struct rq *rq;
>  		struct rq_flags rf;
>  
>  		rq = task_rq_lock(p, &rf);
>  
>  		/*
> -		 * We're either in fork or load path and @p->policy will be
> -		 * applied right after. Reverting @p->policy here and rejecting
> -		 * %SCHED_EXT transitions from scx_check_setscheduler()
> -		 * guarantees that if ops.init_task() sets @p->disallow, @p can
> -		 * never be in SCX.
> +		 * We're in the load path and @p->policy will be applied right
> +		 * after. Reverting @p->policy here and rejecting %SCHED_EXT
> +		 * transitions from scx_check_setscheduler() guarantees that if
> +		 * ops.init_task() sets @p->disallow, @p can never be in SCX.
>  		 */
>  		if (p->policy == SCHED_EXT) {
>  			p->policy = SCHED_NORMAL;
> @@ -3418,6 +3417,9 @@ static int scx_ops_init_task(struct task
>  		}
>  
>  		task_rq_unlock(rq, p, &rf);
> +	} else if (p->scx.disallow) {

Just to make it a bit easier on schedulers, should we do this:

} else if (p->scx.disallow && p->policy == SCHED_EXT)

That way if you have a task that isn't running with SCHED_EXT and forks,
the scheduler can set p->scx.disallow without having to check that it's
being set in a fork. Seems unnecessary to enforce that given that the
end result is the same.

Otherwise this LG. Feel free to apply if you agree, and add my ack:

Acked-by: David Vernet <void@manifault.com>

Thanks,
David