On Thu, May 08, 2014 at 08:37:27AM -0700, Paul E. McKenney wrote:
> On Wed, May 07, 2014 at 08:06:21PM +0200, Peter Zijlstra wrote:
> > On Wed, May 07, 2014 at 02:35:26PM +0200, Peter Zijlstra wrote:
> > >  static void ring_buffer_attach(struct perf_event *event,
> > >  			       struct ring_buffer *rb)
> > >  {
> > > +	struct ring_buffer *old_rb = NULL;
> > >  	unsigned long flags;
> > >  
> > > +	if (event->rb) {
> > > +		/*
> > > +		 * Should be impossible, we set this when removing
> > > +		 * event->rb_entry and wait/clear when adding event->rb_entry.
> > > +		 */
> > > +		WARN_ON_ONCE(event->rcu_pending);
> > >  
> > > +		old_rb = event->rb;
> > > +		event->rcu_batches = get_state_synchronize_rcu();
> > > +		event->rcu_pending = 1;
> > >  
> > > +		spin_lock_irqsave(&rb->event_lock, flags);
> > > +		list_del_rcu(&event->rb_entry);
> > > +		spin_unlock_irqrestore(&rb->event_lock, flags);
> > 
> > This all works a whole lot better if you make that old_rb->event_lock.
> > 
> > > +	}
> > >  
> > > +	if (event->rcu_pending && rb) {
> > > +		cond_synchronize_rcu(event->rcu_batches);
> 
> There is not a whole lot of code between the get_state_synchronize_rcu()
> and the cond_synchronize_rcu(), so I would expect this to do a
> synchronize_rcu() almost all the time.  Or am I missing something here?

From the Changelog:

 2) an event that has a buffer attached, the buffer is destroyed
    (munmap) and then the event is attached to a new/different buffer
    using PERF_EVENT_IOC_SET_OUTPUT.

    This case is more complex because the buffer destruction does:
      ring_buffer_attach(.rb = NULL)
    followed by the ioctl() doing:
      ring_buffer_attach(.rb = foo);

    and we still need to observe the grace period between these two
    calls due to us reusing the event->rb_entry list_head.