Re: [PATCH] tracing: Add trace_trigger kernel command line option

[Paul E. McKenney]

On Wed, Oct 19, 2022 at 08:07:45PM -0400, Steven Rostedt wrote:
> On Wed, 19 Oct 2022 20:01:37 -0400
> Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:
> 
> > @@ -90,6 +90,10 @@ int unregister_tracepoint_module_notifier(struct notifier_block *nb)
> >  #ifdef CONFIG_TRACEPOINTS
> >  static inline void tracepoint_synchronize_unregister(void)
> >  {
> > +	/* Early updates do not need synchronization */
> > +	if (early_boot_irqs_disabled)
> > +		return;
> > +
> >  	synchronize_srcu(&tracepoint_srcu);
> >  	synchronize_rcu();
> 
> I wonder if this check should be just moved to the RCU synchronization
> code? That is, if early_boot_irqs_disabled is set, do nothing, as there
> should be nothing to synchronize against.

There already is a similar check, but it follows the lockdep check.

Does the following patch help?

							Thanx, Paul

------------------------------------------------------------------------

diff --git a/kernel/rcu/srcutree.c b/kernel/rcu/srcutree.c
index ca4b5dcec675b..356ef70d5442c 100644
--- a/kernel/rcu/srcutree.c
+++ b/kernel/rcu/srcutree.c
@@ -1267,14 +1267,15 @@ static void __synchronize_srcu(struct srcu_struct *ssp, bool do_norm)
 {
 	struct rcu_synchronize rcu;
 
+	if (rcu_scheduler_active == RCU_SCHEDULER_INACTIVE)
+		return;
+
 	RCU_LOCKDEP_WARN(lockdep_is_held(ssp) ||
 			 lock_is_held(&rcu_bh_lock_map) ||
 			 lock_is_held(&rcu_lock_map) ||
 			 lock_is_held(&rcu_sched_lock_map),
 			 "Illegal synchronize_srcu() in same-type SRCU (or in RCU) read-side critical section");
 
-	if (rcu_scheduler_active == RCU_SCHEDULER_INACTIVE)
-		return;
 	might_sleep();
 	check_init_srcu_struct(ssp);
 	init_completion(&rcu.completion);