Re: [PATCH v3] sched: do not call __put_task_struct() on rt if pi_blocked_on is set

["Luis Claudio R. Goncalves" <lgoncalv@redhat.com>]

On Thu, Apr 10, 2025 at 02:40:59PM +0200, Peter Zijlstra wrote:
> On Thu, Apr 10, 2025 at 09:10:12AM -0300, Luis Claudio R. Goncalves wrote:
> > With PREEMPT_RT enabled, some of the calls to put_task_struct() coming
> > from rt_mutex_adjust_prio_chain() could happen in preemptible context and
> > with a mutex enqueued. That could lead to this sequence:
> > 
> > 	rt_mutex_adjust_prio_chain()
> > 	  put_task_struct()
> > 	    __put_task_struct()
> > 	      sched_ext_free()
> > 	        spin_lock_irqsave()
> > 	          rtlock_lock() --->  TRIGGERS
> > 	                              lockdep_assert(!current->pi_blocked_on);
> > 
> > Fix that by unconditionally resorting to the deferred call to
> > __put_task_struct().
> > 
> > v2: (Rostedt) remove the #ifdef from put_task_struct() and create
> >     tsk_is_pi_blocked_on() in sched.h to make the change cleaner.
> > 
> > v3: (Sebastian and PeterZ) always call the RCU deferred __put_task_struct().
> 
> Changelog goes below the --- line.
> 
> > Suggested-by: Crystal Wood <crwood@redhat.com>
> > Signed-off-by: Luis Claudio R. Goncalves <lgoncalv@redhat.com>
> > ---
> >  include/linux/sched/task.h |   20 +++++---------------
> >  1 file changed, 5 insertions(+), 15 deletions(-)
> > 
> > diff --git a/include/linux/sched/task.h b/include/linux/sched/task.h
> > index 0f2aeb37bbb04..49847efe5559e 100644
> > --- a/include/linux/sched/task.h
> > +++ b/include/linux/sched/task.h
> > @@ -134,22 +134,12 @@ static inline void put_task_struct(struct task_struct *t)
> >  		return;
> >  
> >  	/*
> > -	 * In !RT, it is always safe to call __put_task_struct().
> > -	 * Under RT, we can only call it in preemptible context.
> > -	 */
> > -	if (!IS_ENABLED(CONFIG_PREEMPT_RT) || preemptible()) {
> > -		static DEFINE_WAIT_OVERRIDE_MAP(put_task_map, LD_WAIT_SLEEP);
> > -
> > -		lock_map_acquire_try(&put_task_map);
> > -		__put_task_struct(t);
> > -		lock_map_release(&put_task_map);
> > -		return;
> > -	}
> 
> I don't think you've substantiated why the !PREEMPT_RT case needs to go.

That was my misunderstanding of "unconditionally call the deferred
function". I see I took it too far and made the patch wrong.

I am testing v4 (closer to the original code with fixed comments) that is
basically:

	if !RT
		__put_task_struct (original code)
	else
		call_rcu(__put_task_struct_rcu_cb)

With the corrected comments Sebastian pointed out.

As soon as the tests complete I will post v4.

Thanks,
Luis
 
> > -
> > -	/*
> > -	 * under PREEMPT_RT, we can't call put_task_struct
> > +	 * In !RT, it is always safe to call __put_task_struct(),
> > +	 * but under PREEMPT_RT, we can't call put_task_struct
> >  	 * in atomic context because it will indirectly
> > -	 * acquire sleeping locks.
> > +	 * acquire sleeping locks. The same is true if the
> > +	 * current process has a mutex enqueued (blocked on
> > +	 * a PI chain).
> >  	 *
> >  	 * call_rcu() will schedule delayed_put_task_struct_rcu()
> >  	 * to be called in process context.
> > -- 
> > 2.49.0
> > 
> 
---end quoted text---