Re: __update_max_tr: rcu_read_lock() used illegally while idle!

[Steven Rostedt <rostedt@goodmis.org>]

On Tue, 2012-07-31 at 08:18 -0700, Paul E. McKenney wrote:
> On Tue, Jul 31, 2012 at 10:56:23AM -0400, Steven Rostedt wrote:
> > On Tue, 2012-07-31 at 10:51 -0400, Steven Rostedt wrote:
> > 
> > > > OK, I will bite.  How about using something like RCU_NONIDLE(), either
> > > > directly or open-coded, to make it a legal call site?
> > > 
> > > OK, then something like:
> > > 
> > > 	RCU_NONIDLE(max_data = task_uid(tsk));
> > > 
> > > would work when called normally or with idle?
> > > 
> > 
> > The comment above RCU_NONIDLE() says:
> > 
> > 	This macro may be used from process-level code only.
> > 
> > Although I'm not sure what a 'level' is. Do you mean process-context? If
> > so, then this will not work because it can be called from non process
> > level code (return from interrupt), or any interrupt that enables
> > interrupts.
> 
> Yep, process context.  It seems that I was naively expecting the
> calls in interrupt context to be covered by rcu_irq_enter() and
> rcu_irq_exit().  I take it that these calls are happening outside of
> the rcu_irq_enter()-protected region?

Yep, it is called by the same hooks that lockdep uses. TRACE_IRQS_ON
inside the irq handler in assembly.

-- Steve