Re: [PATCH 02/10] perf, trace: Use per-tracepoint-per-cpu hlist to track events

[Peter Zijlstra <peterz@infradead.org>]

On Fri, 2010-05-21 at 10:04 -0400, Steven Rostedt wrote:
> On Fri, 2010-05-21 at 11:02 +0200, Peter Zijlstra wrote:
> > plain text document attachment (perf-trace-per-cpu-fold.patch)
> > Avoid the swevent hash-table by using per-tracepoint hlists.
> > 
> > Also, avoid conditionals on the fast path by ordering with probe unregister
> > so that we should never get on the callback path without the data being there.
> > 
> 
> > -void perf_trace_disable(int event_id)
> > +void perf_trace_destroy(struct perf_event *p_event)
> >  {
> > -	struct ftrace_event_call *event;
> > +	struct ftrace_event_call *tp_event = p_event->tp_event;
> > +	int i;
> >  
> > -	mutex_lock(&event_mutex);
> > -	list_for_each_entry(event, &ftrace_events, list) {
> > -		if (event->id == event_id) {
> > -			perf_trace_event_disable(event);
> > -			module_put(event->mod);
> > -			break;
> > +	if (--tp_event->perf_refcount > 0)
> > +		return;
> > +
> 
> You remove the event_mutex and then update the tp_event->perf_refcount
> without any protection.
> 
> That tp_event->perf_refcount is global to the event not to the perf
> event.

Oh, bugger, yes. I seem to have lost it on destroy..

Thanks!