Re: [PATCH] rcu: Use IRQ_WORK_INIT_HARD for srcu's irq_work

On Fri, 21 Aug 2026 13:01:08 +0200
Sebastian Andrzej Siewior <[email protected]> wrote:

> > > > > > +	ssp->srcu_sup->irq_work = IRQ_WORK_INIT_HARD(srcu_irq_work);    
> > > > > 
> > > > > It may need a comment to not lose that.    
> > > 
> > > I agree a comment would be useful here.  
> > Urgh.  
> 
> What do you want me to cover in the comment? The arguing pro
> IRQ_WORK_INIT_HARD is because it is "just" a workqueue schedule and it
> avoids the jump over irq_work/ thread. This makes sense regardless of
> the tracing issue.
> If it would be not compatible with hard-irq context I would argue for
> the makefile change which is still sane since there is no other way to
> express dependencies (which we clearly have here).

Basically just a word to why it is HARD.

 	atomic_set(&ssp->srcu_sup->srcu_barrier_cpu_cnt, 0);
 	INIT_DELAYED_WORK(&ssp->srcu_sup->work, process_srcu);
-	init_irq_work(&ssp->srcu_sup->irq_work, srcu_irq_work);
+	/*
+	 * trace events started on the command line require SRCU before
+	 * the irq_work kthread starts. Since all it does is a simple
+	 * wakeup, having it as a hard irq, even on PREEMPT_RT is fine.
+	 */
+	ssp->srcu_sup->irq_work = IRQ_WORK_INIT_HARD(srcu_irq_work);
 	ssp->srcu_sup->sda_is_static = is_static;

-- Steve