Re: linux-next: Tree for May 26 (RCU stalls)

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

On Thu, May 26, 2011 at 08:31:28PM +0200, Sedat Dilek wrote:
> On Thu, May 26, 2011 at 7:31 PM, Paul E. McKenney
> <paulmck@linux.vnet.ibm.com> wrote:
> > On Thu, May 26, 2011 at 05:48:32PM +0200, Sedat Dilek wrote:
> >> On Thu, May 26, 2011 at 8:39 AM, Stephen Rothwell <sfr@canb.auug.org.au> wrote:
> >> > Hi all,
> >> >
> >> > [The kernel.org mirroring is being slow today]
> >> >
> >> > Changes since 20110525:
> >> >
> >> > Linus' tree gained a build failure for which I applied a patch.
> >> >
> >> > The m68knommu tree lost its conflicts.
> >> >
> >> > The hwmon-staging lost its conflict.
> >> >
> >> > The wireless lost its conflict.
> >> >
> >> > The mmc lost its conflict.
> >> >
> >> > The dwmw2-iommu tree lost its conflict.
> >> >
> >> > The kvm tree still had its build failure so I used the version from
> >> > next-20110524.
> >> >
> >> > The namespace lost its conflicts.
> >> >
> >> > ----------------------------------------------------------------------------
> >> >
> >>
> >> Hi,
> >>
> >> I see these call-traces on x86 UP machine:
> >>
> >> [  240.268061] INFO: task rcun0:8 blocked for more than 120 seconds.
> >> [  240.268069] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
> >> disables this message.
> >> [  240.268072] rcun0           D 00000000     0     8      2 0x00000000
> >> [  240.268079]  f6473fb8 00000046 013131b6 00000000 c1461ac0 00000000
> >> 00000000 c1461ac0
> >> [  240.268089]  00000000 00000000 f645dc70 f645bf60 00000003 f6473f78
> >> c102a570 f6473f9c
> >> [  240.268097]  c1021476 00000000 f645bf6c 00000001 00000000 00000286
> >> f6473f9c c129b35a
> >> [  240.268106] Call Trace:
> >> [  240.268121]  [<c102a570>] ? default_wake_function+0xb/0xd
> >> [  240.268127]  [<c1021476>] ? __wake_up_common+0x33/0x5b
> >> [  240.268134]  [<c129b35a>] ? _raw_spin_unlock_irqrestore+0xe/0x10
> >> [  240.268140]  [<c10234ed>] ? complete+0x34/0x3e
> >> [  240.268147]  [<c1074d23>] ? cpumask_weight+0xc/0xc
> >> [  240.268157]  [<c1044c97>] kthread+0x53/0x67
> >> [  240.268162]  [<c1044c44>] ? kthread_worker_fn+0x111/0x111
> >> [  240.268169]  [<c12a123e>] kernel_thread_helper+0x6/0xd
> >>
> >> dmesg and kernel-config are attached.
> >
> > Hello, Sedat,
> >
> > Does the following patch clear things up?
> >
> >                                                        Thanx, Paul
> >
> > ------------------------------------------------------------------------
> >
> > rcu: Start RCU kthreads in TASK_INTERRUPTIBLE state
> >
> > Upon creation, kthreads are in TASK_UNINTERRUPTIBLE state, which can
> > result in softlockup warnings.  Because some of RCU's kthreads can
> > legitimately be idle indefinitely, start them in TASK_INTERRUPTIBLE
> > state in order to avoid those warnings.
> >
> > Suggested-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
> > Signed-off-by: Paul E. McKenney <paul.mckenney@linaro.org>
> > Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> > Tested-by: Yinghai Lu <yinghai@kernel.org>
> >
> > diff --git a/kernel/rcutree.c b/kernel/rcutree.c
> > index a1a8bb6..40aab8d 100644
> > --- a/kernel/rcutree.c
> > +++ b/kernel/rcutree.c
> > @@ -1647,6 +1647,7 @@ static int __cpuinit rcu_spawn_one_cpu_kthread(int cpu)
> >        if (IS_ERR(t))
> >                return PTR_ERR(t);
> >        kthread_bind(t, cpu);
> > +       set_task_state(t, TASK_INTERRUPTIBLE);
> >        per_cpu(rcu_cpu_kthread_cpu, cpu) = cpu;
> >        WARN_ON_ONCE(per_cpu(rcu_cpu_kthread_task, cpu) != NULL);
> >        per_cpu(rcu_cpu_kthread_task, cpu) = t;
> > @@ -1754,6 +1755,7 @@ static int __cpuinit rcu_spawn_one_node_kthread(struct rcu_state *rsp,
> >                if (IS_ERR(t))
> >                        return PTR_ERR(t);
> >                raw_spin_lock_irqsave(&rnp->lock, flags);
> > +               set_task_state(t, TASK_INTERRUPTIBLE);
> >                rnp->node_kthread_task = t;
> >                raw_spin_unlock_irqrestore(&rnp->lock, flags);
> >                sp.sched_priority = 99;
> > diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h
> > index 049f278..a767b7d 100644
> > --- a/kernel/rcutree_plugin.h
> > +++ b/kernel/rcutree_plugin.h
> > @@ -1295,6 +1295,7 @@ static int __cpuinit rcu_spawn_one_boost_kthread(struct rcu_state *rsp,
> >        if (IS_ERR(t))
> >                return PTR_ERR(t);
> >        raw_spin_lock_irqsave(&rnp->lock, flags);
> > +       set_task_state(t, TASK_INTERRUPTIBLE);
> >        rnp->boost_kthread_task = t;
> >        raw_spin_unlock_irqrestore(&rnp->lock, flags);
> >        sp.sched_priority = RCU_KTHREAD_PRIO;
> >
> 
> Thanks for the quick reply and patch!
> 
> On 1st look at dmesg the RCU stalls are gone.
> I tested against linux-next (next-20110526).
> 
> Feel free to add:
> 
>      Tested-by: Sedat Dilek <sedat.dilek@gmail.com>

Thank you for testing, Sedat!

							Thanx, Paul