Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <1553564407-1162-1-git-send-email-laoar.shao@gmail.com>
 <1553564407-1162-3-git-send-email-laoar.shao@gmail.com> <20190325222105.637f7e23@oasis.local.home>
In-Reply-To: <20190325222105.637f7e23@oasis.local.home>
From:   Yafang Shao <laoar.shao@gmail.com>
Date:   Tue, 26 Mar 2019 10:23:26 +0800
Message-ID: <CALOAHbDJgT6Uu8PWJKLxwz0wHJB9=YbkBN-KXqaniAEa=Pf=8Q@mail.gmail.com>
Subject: Re: [PATCH 2/3] sched/fair: do not expose some tracepoints to user if
 CONFIG_SCHEDSTATS is not set
To:     Steven Rostedt <rostedt@goodmis.org>
Cc:     mingo@redhat.com, Peter Zijlstra <peterz@infradead.org>,
        Paul McKenney <paulmck@linux.vnet.ibm.com>,
        josh@joshtriplett.org,
        Mathieu Desnoyers <mathieu.desnoyers@efficios.com>,
        jiangshanlai@gmail.com, joel@joelfernandes.org,
        shaoyafang@didiglobal.com, LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Tue, Mar 26, 2019 at 10:21 AM Steven Rostedt <rostedt@goodmis.org> wrote:
>
> On Tue, 26 Mar 2019 09:40:06 +0800
> Yafang Shao <laoar.shao@gmail.com> wrote:
>
> > The tracepoints trace_sched_stat_{iowait, blocked, wait, sleep} should
> > be not exposed to user if CONFIG_SCHEDSTATS is not set.
> >
> > Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
> > ---
> >  include/trace/events/sched.h | 14 ++++++++++++++
> >  1 file changed, 14 insertions(+)
> >
> > diff --git a/include/trace/events/sched.h b/include/trace/events/sched.h
> > index 9a4bdfa..4686c7f 100644
> > --- a/include/trace/events/sched.h
> > +++ b/include/trace/events/sched.h
> > @@ -336,6 +336,7 @@ static inline long __trace_sched_switch_state(bool preempt, struct task_struct *
> >                 __entry->pid, __entry->old_pid)
> >  );
> >
> > +#ifdef CONFIG_SCHEDSTATS
>
> Instead of doing this #ifdef here, perhaps we should have at the top of the file:
>
> #ifdef CONFIG_SCHEDSTATS
> # define TRACE_EVENT_SCHEDSTAT TRACE_EVENT
> #else
> # define TRACE_EVENT_SCHEDSTAT TRACE_EVENT_NOP
> #endif
>
> And then switch the sched stat tracepoints from TRACE_EVENT to
> use TRACE_EVENT_SCHEDSTAT.
>
> And if my magic macro hat is working while sitting on the shelf (don't
> have time to put it on), then the TRACE_EVENT_SCHEDSTAT will work like
> a TRACE_EVENT when enabled, and a TRACE_EVENT_NOP when not.
>
> Can you give that a try?
>
> If this works, you don't need to duplicate all the parameters again.
>

You proposal seems like a better solution.
I will have a try.
Thanks for your suggestion.

-Yafang

>
> >  /*
> >   * XXX the below sched_stat tracepoints only apply to
> > SCHED_OTHER/BATCH/IDLE
> >   *     adding sched_stat support to SCHED_FIFO/RR would be welcome.
> > @@ -395,6 +396,19 @@ static inline long
> > __trace_sched_switch_state(bool preempt, struct task_struct *
> > TP_PROTO(struct task_struct *tsk, u64 delay), TP_ARGS(tsk, delay));
> >
> > +#else /* CONFIG_SCHEDSTATS */
> > +
> > +TRACE_EVENT_NONE(sched_stat_wait,
> > +              TP_PROTO(struct task_struct *tsk, u64 delay));
> > +TRACE_EVENT_NONE(sched_stat_sleep,
> > +              TP_PROTO(struct task_struct *tsk, u64 delay));
> > +TRACE_EVENT_NONE(sched_stat_iowait,
> > +              TP_PROTO(struct task_struct *tsk, u64 delay));
> > +TRACE_EVENT_NONE(sched_stat_blocked,
> > +              TP_PROTO(struct task_struct *tsk, u64 delay));
> > +
> > +#endif /* CONFIG_SCHEDSTATS */
> > +
> >  /*
> >   * Tracepoint for accounting runtime (time the task is executing
> >   * on a CPU).
>