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

[Frederic Weisbecker <fweisbec@gmail.com>]

On Fri, May 21, 2010 at 12:34:03PM +0200, Peter Zijlstra wrote:
> On Fri, 2010-05-21 at 12:21 +0200, Frederic Weisbecker wrote:
> > On Fri, May 21, 2010 at 12:19:09PM +0200, Peter Zijlstra wrote:
> > > On Fri, 2010-05-21 at 12:13 +0200, Frederic Weisbecker wrote:
> > > > > I assumed that after probe unregister a tracepoint callback doesn't
> > > > > happen, which then guarantees we should never get !head.
> > > 
> > > > I'm not sure about this. The tracepoints are called under rcu_read_lock(),
> > > > but there is not synchronize_rcu() after we unregister a tracepoint, which
> > > > means you can have a pending preempted one somewhere.
> > > > 
> > > > There is a call_rcu that removes the callbacks, but that only protect
> > > > the callback themselves. 
> > > 
> > > Ah, ok, so we should do probe_unregister + synchronize_sched().
> > > That should ensure __DO_TRACE() doesn't call into it anymore.
> > > 
> > > /me goes make a patch
> > > 
> > 
> > 
> > Yep. But that also means we need to rcu_dereference_sched() to access
> > the per cpu list of events.
> 
> Why?
> 
> The per-cpu vars are allocated and freed in a fully serialized manner,
> there should be no races what so ever.


Ah right, I was confused.