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;
Date:   Wed, 22 May 2019 16:35:45 +0200
From:   Peter Zijlstra <peterz@infradead.org>
To:     Viktor Rosendahl <viktor.rosendahl@gmail.com>
Cc:     Steven Rostedt <rostedt@goodmis.org>,
        Ingo Molnar <mingo@redhat.com>, linux-kernel@vger.kernel.org,
        Joel Fernandes <joel@joelfernandes.org>,
        "Rafael J . Wysocki" <rafael.j.wysocki@intel.com>
Subject: Re: [PATCH v4 1/4] ftrace: Implement fs notification for
 tracing_max_latency
Message-ID: <20190522143545.GG16275@worktop.programming.kicks-ass.net>
References: <20190521120142.186487e9@gandalf.local.home>
 <20190522003014.1359-1-viktor.rosendahl@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20190522003014.1359-1-viktor.rosendahl@gmail.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, May 22, 2019 at 02:30:14AM +0200, Viktor Rosendahl wrote:

> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 874c427742a9..440cd1a62722 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -3374,6 +3374,7 @@ static void __sched notrace __schedule(bool preempt)
>  	struct rq *rq;
>  	int cpu;
>  
> +	trace_disable_fsnotify();
>  	cpu = smp_processor_id();
>  	rq = cpu_rq(cpu);
>  	prev = rq->curr;
> @@ -3449,6 +3450,7 @@ static void __sched notrace __schedule(bool preempt)
>  	}
>  
>  	balance_callback(rq);
> +	trace_enable_fsnotify();
>  }
>  
>  void __noreturn do_task_dead(void)
> diff --git a/kernel/sched/idle.c b/kernel/sched/idle.c
> index 80940939b733..1a38bcdb3652 100644
> --- a/kernel/sched/idle.c
> +++ b/kernel/sched/idle.c
> @@ -225,6 +225,7 @@ static void cpuidle_idle_call(void)
>  static void do_idle(void)
>  {
>  	int cpu = smp_processor_id();
> +	trace_disable_fsnotify();
>  	/*
>  	 * If the arch has a polling bit, we maintain an invariant:
>  	 *
> @@ -284,6 +285,7 @@ static void do_idle(void)
>  	smp_mb__after_atomic();
>  
>  	sched_ttwu_pending();
> +	/* schedule_idle() will call trace_enable_fsnotify() */
>  	schedule_idle();
>  
>  	if (unlikely(klp_patch_pending(current)))

I still hate this.. why are we doing this? We already have this
stop_critical_timings() nonsense and are now adding more gunk.


> +static DEFINE_PER_CPU(atomic_t, notify_disabled) = ATOMIC_INIT(0);

> +	atomic_set(&per_cpu(notify_disabled, cpu), 1);

> +	atomic_set(&per_cpu(notify_disabled, cpu), 0);

> +	if (!atomic_read(&per_cpu(notify_disabled, cpu)))

That's just wrong on so many levels..