Re: [PATCH 7/7 v3] tracing: Make sure rcu_irq_enter() can work for trace_*_rcuidle() trace events

From: Steven Rostedt
Date: Mon Apr 10 2017 - 14:33:00 EST


On Mon, 10 Apr 2017 11:29:01 -0700
"Paul E. McKenney" <paulmck@xxxxxxxxxxxxxxxxxx> wrote:

> On Mon, Apr 10, 2017 at 02:10:12PM -0400, Steven Rostedt wrote:
> > From: "Steven Rostedt (VMware)" <rostedt@xxxxxxxxxxx>
> >
> > Stack tracing discovered that there's a small location inside the RCU
> > infrastructure where calling rcu_irq_enter() does not work. As trace events
> > use rcu_irq_enter() it must make sure that it is functionable. A check
> > against rcu_irq_enter_disabled() is added with a WARN_ON_ONCE() as no trace
> > event should ever be used in that part of RCU. If the warning is triggered,
> > then the trace event is ignored.
> >
> > Restructure the __DO_TRACE() a bit to get rid of the prercu and postrcu,
> > and just have an rcucheck that does the work from within the _DO_TRACE()
> > macro. gcc optimization will compile out the rcucheck=0 case.
> >
> > Link: http://lkml.kernel.org/r/20170405093207.404f8deb@xxxxxxxxxxxxxxxxxx
> >
> > Acked-by: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
> > Signed-off-by: Steven Rostedt (VMware) <rostedt@xxxxxxxxxxx>
>
> Acked-by: Paul E. McKenney <paulmck@xxxxxxxxxxxxxxxxxx>
>
> As an aside, it looks like the rcu_irq_enter_disabled() settings
> in the RCU idle-entry code could be placed under CONFIG_PROVE_RCU=y
> should my idle-entry-overhead concerns prove to be well-founded.
> The errors would be caught during testing, but no production-side
> overhead.

Even with the underscored __this_cpu_*() calls?

>
> Again, I am not necessarily agitating for this change now, just getting
> this possibility on the record. ;-)

That would be easy to add :-) But we can do that at a later time.

-- Steve