Re: [PATCH v2] genirq: Fix nested thread vs synchronize_hardirq() deadlock

[Vincent Whitchurch]

On Fri, 2023-06-30 at 11:07 +0200, Thomas Gleixner wrote:
> On Tue, Jun 20 2023 at 13:16, Vincent Whitchurch wrote:
> > --- a/kernel/irq/chip.c
> > +++ b/kernel/irq/chip.c
> > @@ -476,7 +476,7 @@ void handle_nested_irq(unsigned int irq)
> >  	}
> >  
> > 
> >  	kstat_incr_irqs_this_cpu(desc);
> > -	irqd_set(&desc->irq_data, IRQD_IRQ_INPROGRESS);
> > +	atomic_inc(&desc->threads_active);
> >  	raw_spin_unlock_irq(&desc->lock);
> >  
> > 
> >  	action_ret = IRQ_NONE;
> > @@ -487,7 +487,8 @@ void handle_nested_irq(unsigned int irq)
> >  		note_interrupt(desc, action_ret);
> >  
> > 
> >  	raw_spin_lock_irq(&desc->lock);
> > -	irqd_clear(&desc->irq_data, IRQD_IRQ_INPROGRESS);
> > +	if (atomic_dec_and_test(&desc->threads_active))
> > +		wake_up(&desc->wait_for_threads);
> 
> This breaks on RT. The wakeup cannot be inside the raw spin-locked
> region.

OK.  I will remove the acquisition of the spin lock at the end of this
function since it was only used for protecting the irqd flags.

> Also this is open coding wake_threads_waitq().

OK, that's in manage.c so I'll make it non-static and add it to
internals.h and use it from here.

> > +static void __synchronize_irq(struct irq_desc *desc)
> > +{
> > +	__synchronize_hardirq(desc, true);
> > +	/*
> > +	 * We made sure that no hardirq handler is
> > +	 * running. Now verify that no threaded handlers are
> > +	 * active.
> > +	 */
> > +	wait_event(desc->wait_for_threads,
> > +		   !atomic_read(&desc->threads_active));
> 
> Splitting this out is fine. Not reformatting it not so much.

Will fix.