Message-ID: <20140131125719.73340f6e@gandalf.local.home>
Date:	Fri, 31 Jan 2014 12:57:19 -0500
From:	Steven Rostedt <rostedt@...dmis.org>
To:	paulmck@...ux.vnet.ibm.com
Cc:	Sebastian Andrzej Siewior <bigeasy@...utronix.de>,
	linux-rt-users@...r.kernel.org, linux-kernel@...r.kernel.org,
	tglx@...utronix.de, Clark Williams <williams@...hat.com>
Subject: Re: [PATCH 2/2] timer: really raise softirq if there is irq_work to
 do

On Fri, 31 Jan 2014 09:42:27 -0800
"Paul E. McKenney" <paulmck@...ux.vnet.ibm.com> wrote:

> On Fri, Jan 31, 2014 at 12:07:57PM -0500, Steven Rostedt wrote:
> > On Fri, 31 Jan 2014 15:34:05 +0100
> > Sebastian Andrzej Siewior <bigeasy@...utronix.de> wrote:
> > 
> > > from looking at the code, it seems that the softirq is only raised (in
> > > the !base->active_timers case) if we have also an expired timer
> > > (time_before_eq() is true). This patch ensures that the timer softirq is
> > > also raised in the !base->active_timers && no timer expired.
> > 
> > A couple of things. If there is no active timers, we do not need to
> > check the expired timers. That may contain a deferred timer that does
> > not need to be raised if the system is idle. This will just
> > re-introduce the problems that other people have been seeing.
> > 
> > The bug that I found is that if there *are* active timers, but they
> > have not expired yet. Why is this a problem? Because in that case we do
> > not check if there is irq_work to be done. That means the irq_work will
> > have to wait till the timer expires, and since RCU depends on this,
> > that can take a while. I've had a synchronize_sched() take up to 5
> > seconds to complete due to this!
> > 
> > 
> > The real fix is the following:
> > 
> > timer/rt: Always raise the softirq if there's irq_work to be done
> > 
> > It was previously discovered that some systems would hang on boot up
> > with a previous version of 3.12-rt. This was due to RCU using irq_work,
> > and RT defers the irq_work to a softirq. But if there's no active
> > timers, the softirq will not be raised, and RCU work will not get done,
> > causing the system to hang.  The fix was to check that if there was no
> > active timers but irq_work to be done, then we should raise the softirq.
> > 
> > But this fix was not 100% correct. It left out the case that there were
> > active timers that were not expired yet. This would have the softirq
> > not get raised even if there was irq work to be done.
> > 
> > If there is irq_work to be done, then we must raise the timer softirq
> > regardless of if there is active timers or whether they are expired or
> > not. The softirq can handle those cases. But we can never ignore
> > irq_work.
> > 
> > As it is only PREEMPT_RT_FULL that requires irq_work to be done in the
> > softirq, we can pull out the check in the active_timers condition, and
> > make the code a bit cleaner by having the irq_work check separate, and
> > put the code in with the other #ifdef PREEMPT_RT. If there is irq_work
> > to be done, there's no need to check the active timers or if they are
> > expired. Just raise the time softirq and be done with it. Otherwise, we
> > can do the timer checks just like we do with non -rt.
> > 
> > Signed-off-by: Steven Rostedt <rostedt@...dmis.org>
> > 
> > diff --git a/kernel/timer.c b/kernel/timer.c
> > index 106968f..426d114 100644
> > --- a/kernel/timer.c
> > +++ b/kernel/timer.c
> > @@ -1461,18 +1461,20 @@ void run_local_timers(void)
> >  	 * the timer softirq.
> >  	 */
> >  #ifdef CONFIG_PREEMPT_RT_FULL
> > +	/* On RT, irq work runs from softirq */
> > +	if (irq_work_needs_cpu()) {
> > +		raise_softirq(TIMER_SOFTIRQ);
> 
> OK, I'll bite...  What if the IRQ work that needs doing is something
> other than TIMER_SOFTIRQ?

Heh, don't let the timer part confuse you. The only reason that softirq
is relevant to irq_work is that is the softirq that we placed the
irq_work to be done. If you look at the code that is called for that
softirq (in -rt) you'll see:

static void run_timer_softirq(struct softirq_action *h)
{
	struct tvec_base *base = __this_cpu_read(tvec_bases);

#if defined(CONFIG_IRQ_WORK) && defined(CONFIG_PREEMPT_RT_FULL)
	irq_work_run();
#endif

	if (time_after_eq(jiffies, base->timer_jiffies))
		__run_timers(base);
}

And we also have:

void update_process_times(int user_tick)
{
	struct task_struct *p = current;
	int cpu = smp_processor_id();

	/* Note: this timer irq context must be accounted for as well. */
	account_process_tick(p, user_tick);
	scheduler_tick();
	run_local_timers();
	rcu_check_callbacks(cpu, user_tick);
#if defined(CONFIG_IRQ_WORK) && !defined(CONFIG_PREEMPT_RT_FULL)
	if (in_irq())
		irq_work_run();
#endif
	run_posix_cpu_timers(p);
}


In vanilla Linux, irq_work_run() is called from update_process_times()
when it is called from the timer interrupt. In -rt, there's reasons we
can't do the irq work from hard irq, so we push it off to the timer
softirq, and run it there.

That means if we have *any* irq work to do, we raise the timer softirq,
even if the work to be done has nothing to do with timers. As you can
see from the softirq timer code, in -rt, irq_work_run() is always
called, without having to look at any timers.

-- Steve



> 
> 							Thanx, Paul
> 
> > +		return;
> > +	}
> > +
> >  	if (!spin_do_trylock(&base->lock)) {
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists