Re: [PATCH 3/7] irqbypass: Take ownership of producer/consumer token tracking

[Sean Christopherson]

On Thu, Apr 10, 2025, Kevin Tian wrote:
> > From: Sean Christopherson <seanjc@xxxxxxxxxx>
> > +int irq_bypass_register_consumer(struct irq_bypass_consumer *consumer,
> > +				 struct eventfd_ctx *eventfd)
> >  {
> >  	struct irq_bypass_consumer *tmp;
> >  	struct irq_bypass_producer *producer;
> >  	int ret;
> > 
> > -	if (!consumer->token ||
> > -	    !consumer->add_producer || !consumer->del_producer)
> > +	if (WARN_ON_ONCE(consumer->token))
> > +		return -EINVAL;
> > +
> > +	if (!consumer->add_producer || !consumer->del_producer)
> >  		return -EINVAL;
> > 
> >  	mutex_lock(&lock);
> > 
> >  	list_for_each_entry(tmp, &consumers, node) {
> > -		if (tmp->token == consumer->token || tmp == consumer) {
> > +		if (tmp->token == eventfd || tmp == consumer) {
> >  			ret = -EBUSY;
> >  			goto out_err;
> >  		}
> >  	}
> 
> the 2nd check 'tmp == consumer' is redundant. If they are equal 
> consumer->token is not NULL then the earlier WARN_ON will be
> triggered already.

Oh, nice.  Good catch!  That check subtly gets dropped on the conversion to
xarray, so it definitely makes sense to remove it in this patch.