Re: [PATCH v2 4/4] sched_ext: Fix cgroup exit ordering by moving sched_ext_free() to finish_task_switch()

[Peter Zijlstra]

On Mon, Nov 03, 2025 at 10:31:36AM -1000, Tejun Heo wrote:
> Hello,
> 
> On Mon, Nov 03, 2025 at 09:28:43PM +0100, Peter Zijlstra wrote:
> > > @@ -5222,6 +5222,12 @@ static struct rq *finish_task_switch(str
> > >  		if (prev->sched_class->task_dead)
> > >  			prev->sched_class->task_dead(prev);
> > 
> > ^^^ can you not use task_dead_scx() ?
> 
> Unfortunately not. Because sched class switches are atomic infallible
> operations, all tasks in the system must be prepped on scheduler attach &
> fork regardless of their current sched class and thus have to be cleaned up
> in the same way on detach & exit.

Ah, I see scx_post_fork() and sched_ext_free^H^H^H^Hdead() effectively
duplicate the tasklist.

Oh well.