From mboxrd@z Thu Jan  1 00:00:00 1970
From: peterz@infradead.org (Peter Zijlstra)
Date: Wed, 10 Dec 2008 14:14:32 +0100
Subject: [ltt-dev] [patch] add tracepoints to trace activate/deactivate
 task
In-Reply-To: <20081210123440.GD28524@Krystal>
References: <20081208194948.GC27166@redhat.com>
	<1228766050.6939.7.camel@twins> <20081208223840.GA30314@redhat.com>
	<1228776169.12729.2.camel@twins> <20081209210048.GA4440@redhat.com>
	<20081209221012.GA4673@Krystal> <1228892902.6978.23.camel@twins>
	<20081210123440.GD28524@Krystal>
Message-ID: <1228914872.12143.11.camel@twins>

On Wed, 2008-12-10 at 07:34 -0500, Mathieu Desnoyers wrote:

> >  include/trace/sched.h |    4 ++--
> >  kernel/sched.c        |    3 ++-
> >  2 files changed, 4 insertions(+), 3 deletions(-)
> > 
> > diff --git a/include/trace/sched.h b/include/trace/sched.h
> > index 9b2854a..f4549d5 100644
> > --- a/include/trace/sched.h
> > +++ b/include/trace/sched.h
> > @@ -30,8 +30,8 @@ DECLARE_TRACE(sched_switch,
> >  		TPARGS(rq, prev, next));
> >  
> >  DECLARE_TRACE(sched_migrate_task,
> > -	TPPROTO(struct rq *rq, struct task_struct *p, int dest_cpu),
> > -		TPARGS(rq, p, dest_cpu));
> > +	TPPROTO(struct task_struct *p, int orig_cpu, int dest_cpu),
> > +		TPARGS(p, orig_cpu, dest_cpu));
> >  
> >  DECLARE_TRACE(sched_process_free,
> >  	TPPROTO(struct task_struct *p),
> > diff --git a/kernel/sched.c b/kernel/sched.c
> > index 0eff15b..3dc54cd 100644
> > --- a/kernel/sched.c
> > +++ b/kernel/sched.c
> > @@ -1861,6 +1861,8 @@ void set_task_cpu(struct task_struct *p, unsigned int new_cpu)
> >  
> >  	clock_offset = old_rq->clock - new_rq->clock;
> >  
> > +	trace_sched_migrate_task(p, task_cpu(p), new_cpu);
> 
> Hrm, looking at it, I think that :
> 
> - task_cpu(p) will add some code output outside of the conditional
>   branch, which I think we would like to avoid.
> - We can easily get the "from" cpu within the tracepoint probe.
>   Therefore, I don't see why we would extract this information
>   explicitly ?

Right, just nuke it - thanks!