Re: [PATCH 01/14] sched: Employ sched_change guards

[Shrikanth Hegde]

On 10/6/25 11:44 PM, Peter Zijlstra wrote:
> On Mon, Oct 06, 2025 at 08:51:27PM +0530, Shrikanth Hegde wrote:
> > On 9/10/25 9:14 PM, Peter Zijlstra wrote:
> > > As proposed a long while ago -- and half done by scx -- wrap the
> > > scheduler's 'change' pattern in a guard helper.
> > [...]>   		put_task_struct(p);
> > > --- a/kernel/sched/sched.h
> > > +++ b/kernel/sched/sched.h
> > > @@ -3860,23 +3860,22 @@ extern void check_class_changed(struct r
> > >    extern struct balance_callback *splice_balance_callbacks(struct rq *rq);
> > >    extern void balance_callbacks(struct rq *rq, struct balance_callback *head);
> > > -#ifdef CONFIG_SCHED_CLASS_EXT
> > > -/*
> > > - * Used by SCX in the enable/disable paths to move tasks between sched_classes
> > > - * and establish invariants.
> > > - */
> > > -struct sched_enq_and_set_ctx {
> > > +struct sched_change_ctx {
> > >    	struct task_struct	*p;
> > > -	int			queue_flags;
> > > +	int			flags;
> > >    	bool			queued;
> > >    	bool			running;
> > >    };
> > > -void sched_deq_and_put_task(struct task_struct *p, int queue_flags,
> > > -			    struct sched_enq_and_set_ctx *ctx);
> > > -void sched_enq_and_set_task(struct sched_enq_and_set_ctx *ctx);
> > > +struct sched_change_ctx *sched_change_begin(struct task_struct *p, unsigned int flags);
> > > +void sched_change_end(struct sched_change_ctx *ctx);
> > > -#endif /* CONFIG_SCHED_CLASS_EXT */
> > > +DEFINE_CLASS(sched_change, struct sched_change_ctx *,
> > > +	     sched_change_end(_T),
> > > +	     sched_change_begin(p, flags),
> > > +	     struct task_struct *p, unsigned int flags)
> > > +
> > > +DEFINE_CLASS_IS_UNCONDITIONAL(sched_change)
> > >    #include "ext.h"
> > could you please add a comment on matching flags on dequeue/enqueue
> > here?
>
> Would something like so be okay? This assumes at least the second patch
> is applied as well.
>
> ---
>
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -10783,6 +10783,12 @@ struct sched_change_ctx *sched_change_be
>   	struct sched_change_ctx *ctx = this_cpu_ptr(&sched_change_ctx);
>   	struct rq *rq = task_rq(p);
>   
> +	/*
> +	 * Must exclusively use matched flags since this is both dequeue and
> +	 * enqueue.
> +	 */

yes. Something like that. Unless callsites explicitly change the flags using
the scope, enqueue will happen with matching flags.

> +	WARN_ON_ONCE(flags & 0xFFFF0000);
> +

A mythical example:
scope_guard(sched_change, p, DEQUEUE_THROTTLE)
	scope->flags &= ~DEQUEUE_THROTTLE;
	scope->flags |= ENQUEUE_HEAD;

But, One could still do this right? for such users the warning may be wrong.

>   	lockdep_assert_rq_held(rq);
>   
>   	if (!(flags & DEQUEUE_NOCLOCK)) {