Re: [PATCH] tracing/filters: allow event filters to be set only when not tracing

["Paul E. McKenney" <paulmck@linux.vnet.ibm.com>]

On Fri, Apr 03, 2009 at 12:43:15PM -0400, Steven Rostedt wrote:
> 
> On Fri, 3 Apr 2009, Ingo Molnar wrote:
> > > 
> > > My kneejerk reaction is "why would anyone want to trace the idle 
> > > loop?"
> > 
> > heh :-)
> > 
> > Idle enter/exit events are useful to tune power use for example. The 
> > more events we have there, the more we prevent the CPU from slowly 
> > going into deep sleep mode.

But we could in principle trace idle enter/exit from the scheduler, correct?

That said, it would be possible to allow much of the idle loop to contain
RCU read-side critical sections, but this requires putting rcu_qsctr_inc()
in each and every idle loop, plus catching the cases where idle loops
shut down the CPU.  Note that this applies to synchronize_sched() as
well as synchronize_rcu().

> I can say I use it a lot. I'm still needing a way to set the function pid 
> recorder to 0 since the conversion to the pid structure. There are times 
> I only want to trace the interrupts that happen in the idle loop.

It would be OK to trace the interrupts, just not the idle loop itself.

							Thanx, Paul