diff mbox series

sched/deadline: fix inactive_task_timer splat with CONFIG_PREEMPT_RT

Message ID 20230104181701.43224-1-wander@redhat.com
State New
Headers
Series sched/deadline: fix inactive_task_timer splat with CONFIG_PREEMPT_RT |

Commit Message

Wander Lairson Costa Jan. 4, 2023, 6:17 p.m. UTC 
  inactive_task_timer() executes in interrupt (atomic) context. It calls
put_task_struct(), which indirectly acquires sleeping locks under
PREEMPT_RT.

Below is an example of a splat that happened in a test environment:

 CPU: 1 PID: 2848 Comm: life Kdump: loaded Tainted: G W ---------
 Hardware name: HP ProLiant DL388p Gen8, BIOS P70 07/15/2012
 Call Trace:
 dump_stack_lvl+0x57/0x7d
 mark_lock_irq.cold+0x33/0xba
 ? stack_trace_save+0x4b/0x70
 ? save_trace+0x55/0x150
 mark_lock+0x1e7/0x400
 mark_usage+0x11d/0x140
 __lock_acquire+0x30d/0x930
 lock_acquire.part.0+0x9c/0x210
 ? refill_obj_stock+0x3d/0x3a0
 ? rcu_read_lock_sched_held+0x3f/0x70
 ? trace_lock_acquire+0x38/0x140
 ? lock_acquire+0x30/0x80
 ? refill_obj_stock+0x3d/0x3a0
 rt_spin_lock+0x27/0xe0
 ? refill_obj_stock+0x3d/0x3a0
 refill_obj_stock+0x3d/0x3a0
 ? inactive_task_timer+0x1ad/0x340
 kmem_cache_free+0x357/0x560
 inactive_task_timer+0x1ad/0x340
 ? switched_from_dl+0x2d0/0x2d0
 __run_hrtimer+0x8a/0x1a0
 __hrtimer_run_queues+0x91/0x130
 hrtimer_interrupt+0x10f/0x220
 __sysvec_apic_timer_interrupt+0x7b/0xd0
 sysvec_apic_timer_interrupt+0x4f/0xd0
 ? asm_sysvec_apic_timer_interrupt+0xa/0x20
 asm_sysvec_apic_timer_interrupt+0x12/0x20
 RIP: 0033:0x7fff196bf6f5

Instead of calling put_task_struct() directly, we defer it using
call_rcu(). A more natural approach would use a workqueue, but since
in PREEMPT_RT, we can't allocate dynamic memory from atomic context,
the code would become more complex because we would need to put the
work_struct instance in the task_struct and initialize it when we
allocate a new task_struct.

Signed-off-by: Wander Lairson Costa <wander@redhat.com>
Cc: Paul McKenney <paulmck@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
---
 kernel/sched/build_policy.c |  1 +
 kernel/sched/deadline.c     | 24 +++++++++++++++++++++++-
 2 files changed, 24 insertions(+), 1 deletion(-)

Comments

Paul E. McKenney Jan. 10, 2023, 1:33 a.m. UTC | #1 
On Wed, Jan 04, 2023 at 03:17:01PM -0300, Wander Lairson Costa wrote:
> inactive_task_timer() executes in interrupt (atomic) context. It calls
> put_task_struct(), which indirectly acquires sleeping locks under
> PREEMPT_RT.
> 
> Below is an example of a splat that happened in a test environment:
> 
>  CPU: 1 PID: 2848 Comm: life Kdump: loaded Tainted: G W ---------
>  Hardware name: HP ProLiant DL388p Gen8, BIOS P70 07/15/2012
>  Call Trace:
>  dump_stack_lvl+0x57/0x7d
>  mark_lock_irq.cold+0x33/0xba
>  ? stack_trace_save+0x4b/0x70
>  ? save_trace+0x55/0x150
>  mark_lock+0x1e7/0x400
>  mark_usage+0x11d/0x140
>  __lock_acquire+0x30d/0x930
>  lock_acquire.part.0+0x9c/0x210
>  ? refill_obj_stock+0x3d/0x3a0
>  ? rcu_read_lock_sched_held+0x3f/0x70
>  ? trace_lock_acquire+0x38/0x140
>  ? lock_acquire+0x30/0x80
>  ? refill_obj_stock+0x3d/0x3a0
>  rt_spin_lock+0x27/0xe0
>  ? refill_obj_stock+0x3d/0x3a0
>  refill_obj_stock+0x3d/0x3a0
>  ? inactive_task_timer+0x1ad/0x340
>  kmem_cache_free+0x357/0x560
>  inactive_task_timer+0x1ad/0x340
>  ? switched_from_dl+0x2d0/0x2d0
>  __run_hrtimer+0x8a/0x1a0
>  __hrtimer_run_queues+0x91/0x130
>  hrtimer_interrupt+0x10f/0x220
>  __sysvec_apic_timer_interrupt+0x7b/0xd0
>  sysvec_apic_timer_interrupt+0x4f/0xd0
>  ? asm_sysvec_apic_timer_interrupt+0xa/0x20
>  asm_sysvec_apic_timer_interrupt+0x12/0x20
>  RIP: 0033:0x7fff196bf6f5
> 
> Instead of calling put_task_struct() directly, we defer it using
> call_rcu(). A more natural approach would use a workqueue, but since
> in PREEMPT_RT, we can't allocate dynamic memory from atomic context,
> the code would become more complex because we would need to put the
> work_struct instance in the task_struct and initialize it when we
> allocate a new task_struct.
> 
> Signed-off-by: Wander Lairson Costa <wander@redhat.com>
> Cc: Paul McKenney <paulmck@kernel.org>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> ---
>  kernel/sched/build_policy.c |  1 +
>  kernel/sched/deadline.c     | 24 +++++++++++++++++++++++-
>  2 files changed, 24 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/sched/build_policy.c b/kernel/sched/build_policy.c
> index d9dc9ab3773f..f159304ee792 100644
> --- a/kernel/sched/build_policy.c
> +++ b/kernel/sched/build_policy.c
> @@ -28,6 +28,7 @@
>  #include <linux/suspend.h>
>  #include <linux/tsacct_kern.h>
>  #include <linux/vtime.h>
> +#include <linux/rcupdate.h>
>  
>  #include <uapi/linux/sched/types.h>
>  
> diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
> index 9ae8f41e3372..ab9301d4cc24 100644
> --- a/kernel/sched/deadline.c
> +++ b/kernel/sched/deadline.c
> @@ -1405,6 +1405,13 @@ static void update_curr_dl(struct rq *rq)
>  	}
>  }
>  
> +static void delayed_put_task_struct(struct rcu_head *rhp)
> +{
> +	struct task_struct *task = container_of(rhp, struct task_struct, rcu);
> +
> +	__put_task_struct(task);

Please note that BH is disabled here.  Don't you therefore
need to schedule a workqueue handler?  Perhaps directly from
inactive_task_timer(), or maybe from this point.  If the latter, one
way to skip the extra step is to use queue_rcu_work().

							Thanx, Paul

> +}
> +
>  static enum hrtimer_restart inactive_task_timer(struct hrtimer *timer)
>  {
>  	struct sched_dl_entity *dl_se = container_of(timer,
> @@ -1442,7 +1449,22 @@ static enum hrtimer_restart inactive_task_timer(struct hrtimer *timer)
>  	dl_se->dl_non_contending = 0;
>  unlock:
>  	task_rq_unlock(rq, p, &rf);
> -	put_task_struct(p);
> +
> +	if (IS_ENABLED(CONFIG_PREEMPT_RT)) {
> +		/*
> +		 * Decrement the refcount explicitly to avoid unnecessarily
> +		 * calling call_rcu.
> +		 */
> +		if (refcount_dec_and_test(&p->usage))
> +			/*
> +			 * under PREEMPT_RT, we can't call put_task_struct
> +			 * in atomic context because it will indirectly
> +			 * acquire sleeping locks.
> +			 */
> +			call_rcu(&p->rcu, delayed_put_task_struct);
> +	} else {
> +		put_task_struct(p);
> +	}
>  
>  	return HRTIMER_NORESTART;
>  }
> -- 
> 2.39.0
>
Wander Lairson Costa Jan. 10, 2023, 8:52 p.m. UTC | #2 
On Mon, Jan 9, 2023 at 10:40 PM Paul E. McKenney <paulmck@kernel.org> wrote:
>
> On Wed, Jan 04, 2023 at 03:17:01PM -0300, Wander Lairson Costa wrote:
> > inactive_task_timer() executes in interrupt (atomic) context. It calls
> > put_task_struct(), which indirectly acquires sleeping locks under
> > PREEMPT_RT.
> >
> > Below is an example of a splat that happened in a test environment:
> >
> >  CPU: 1 PID: 2848 Comm: life Kdump: loaded Tainted: G W ---------
> >  Hardware name: HP ProLiant DL388p Gen8, BIOS P70 07/15/2012
> >  Call Trace:
> >  dump_stack_lvl+0x57/0x7d
> >  mark_lock_irq.cold+0x33/0xba
> >  ? stack_trace_save+0x4b/0x70
> >  ? save_trace+0x55/0x150
> >  mark_lock+0x1e7/0x400
> >  mark_usage+0x11d/0x140
> >  __lock_acquire+0x30d/0x930
> >  lock_acquire.part.0+0x9c/0x210
> >  ? refill_obj_stock+0x3d/0x3a0
> >  ? rcu_read_lock_sched_held+0x3f/0x70
> >  ? trace_lock_acquire+0x38/0x140
> >  ? lock_acquire+0x30/0x80
> >  ? refill_obj_stock+0x3d/0x3a0
> >  rt_spin_lock+0x27/0xe0
> >  ? refill_obj_stock+0x3d/0x3a0
> >  refill_obj_stock+0x3d/0x3a0
> >  ? inactive_task_timer+0x1ad/0x340
> >  kmem_cache_free+0x357/0x560
> >  inactive_task_timer+0x1ad/0x340
> >  ? switched_from_dl+0x2d0/0x2d0
> >  __run_hrtimer+0x8a/0x1a0
> >  __hrtimer_run_queues+0x91/0x130
> >  hrtimer_interrupt+0x10f/0x220
> >  __sysvec_apic_timer_interrupt+0x7b/0xd0
> >  sysvec_apic_timer_interrupt+0x4f/0xd0
> >  ? asm_sysvec_apic_timer_interrupt+0xa/0x20
> >  asm_sysvec_apic_timer_interrupt+0x12/0x20
> >  RIP: 0033:0x7fff196bf6f5
> >
> > Instead of calling put_task_struct() directly, we defer it using
> > call_rcu(). A more natural approach would use a workqueue, but since
> > in PREEMPT_RT, we can't allocate dynamic memory from atomic context,
> > the code would become more complex because we would need to put the
> > work_struct instance in the task_struct and initialize it when we
> > allocate a new task_struct.
> >
> > Signed-off-by: Wander Lairson Costa <wander@redhat.com>
> > Cc: Paul McKenney <paulmck@kernel.org>
> > Cc: Thomas Gleixner <tglx@linutronix.de>
> > ---
> >  kernel/sched/build_policy.c |  1 +
> >  kernel/sched/deadline.c     | 24 +++++++++++++++++++++++-
> >  2 files changed, 24 insertions(+), 1 deletion(-)
> >
> > diff --git a/kernel/sched/build_policy.c b/kernel/sched/build_policy.c
> > index d9dc9ab3773f..f159304ee792 100644
> > --- a/kernel/sched/build_policy.c
> > +++ b/kernel/sched/build_policy.c
> > @@ -28,6 +28,7 @@
> >  #include <linux/suspend.h>
> >  #include <linux/tsacct_kern.h>
> >  #include <linux/vtime.h>
> > +#include <linux/rcupdate.h>
> >
> >  #include <uapi/linux/sched/types.h>
> >
> > diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
> > index 9ae8f41e3372..ab9301d4cc24 100644
> > --- a/kernel/sched/deadline.c
> > +++ b/kernel/sched/deadline.c
> > @@ -1405,6 +1405,13 @@ static void update_curr_dl(struct rq *rq)
> >       }
> >  }
> >
> > +static void delayed_put_task_struct(struct rcu_head *rhp)
> > +{
> > +     struct task_struct *task = container_of(rhp, struct task_struct, rcu);
> > +
> > +     __put_task_struct(task);
>
> Please note that BH is disabled here.  Don't you therefore
> need to schedule a workqueue handler?  Perhaps directly from
> inactive_task_timer(), or maybe from this point.  If the latter, one
> way to skip the extra step is to use queue_rcu_work().
>

My initial work was using a workqueue [1,2]. However, I realized I
could reach a much simpler code with call_rcu().
I am afraid my ignorance doesn't allow me to get your point. Does
disabling softirq imply atomic context?


[1] https://gitlab.com/walac/kernel-ark/-/commit/ec8addbe38d5c318f1789b4c0fa480a9d2afdb65
[2] https://gitlab.com/walac/kernel-ark/-/commit/0bde233235ffed233a7466a36a4866bc48064f54


>                                                         Thanx, Paul
>
> > +}
> > +
> >  static enum hrtimer_restart inactive_task_timer(struct hrtimer *timer)
> >  {
> >       struct sched_dl_entity *dl_se = container_of(timer,
> > @@ -1442,7 +1449,22 @@ static enum hrtimer_restart inactive_task_timer(struct hrtimer *timer)
> >       dl_se->dl_non_contending = 0;
> >  unlock:
> >       task_rq_unlock(rq, p, &rf);
> > -     put_task_struct(p);
> > +
> > +     if (IS_ENABLED(CONFIG_PREEMPT_RT)) {
> > +             /*
> > +              * Decrement the refcount explicitly to avoid unnecessarily
> > +              * calling call_rcu.
> > +              */
> > +             if (refcount_dec_and_test(&p->usage))
> > +                     /*
> > +                      * under PREEMPT_RT, we can't call put_task_struct
> > +                      * in atomic context because it will indirectly
> > +                      * acquire sleeping locks.
> > +                      */
> > +                     call_rcu(&p->rcu, delayed_put_task_struct);
> > +     } else {
> > +             put_task_struct(p);
> > +     }
> >
> >       return HRTIMER_NORESTART;
> >  }
> > --
> > 2.39.0
> >
>
Paul E. McKenney Jan. 10, 2023, 10:27 p.m. UTC | #3 
On Tue, Jan 10, 2023 at 05:52:03PM -0300, Wander Lairson Costa wrote:
> On Mon, Jan 9, 2023 at 10:40 PM Paul E. McKenney <paulmck@kernel.org> wrote:
> >
> > On Wed, Jan 04, 2023 at 03:17:01PM -0300, Wander Lairson Costa wrote:
> > > inactive_task_timer() executes in interrupt (atomic) context. It calls
> > > put_task_struct(), which indirectly acquires sleeping locks under
> > > PREEMPT_RT.
> > >
> > > Below is an example of a splat that happened in a test environment:
> > >
> > >  CPU: 1 PID: 2848 Comm: life Kdump: loaded Tainted: G W ---------
> > >  Hardware name: HP ProLiant DL388p Gen8, BIOS P70 07/15/2012
> > >  Call Trace:
> > >  dump_stack_lvl+0x57/0x7d
> > >  mark_lock_irq.cold+0x33/0xba
> > >  ? stack_trace_save+0x4b/0x70
> > >  ? save_trace+0x55/0x150
> > >  mark_lock+0x1e7/0x400
> > >  mark_usage+0x11d/0x140
> > >  __lock_acquire+0x30d/0x930
> > >  lock_acquire.part.0+0x9c/0x210
> > >  ? refill_obj_stock+0x3d/0x3a0
> > >  ? rcu_read_lock_sched_held+0x3f/0x70
> > >  ? trace_lock_acquire+0x38/0x140
> > >  ? lock_acquire+0x30/0x80
> > >  ? refill_obj_stock+0x3d/0x3a0
> > >  rt_spin_lock+0x27/0xe0
> > >  ? refill_obj_stock+0x3d/0x3a0
> > >  refill_obj_stock+0x3d/0x3a0
> > >  ? inactive_task_timer+0x1ad/0x340
> > >  kmem_cache_free+0x357/0x560
> > >  inactive_task_timer+0x1ad/0x340
> > >  ? switched_from_dl+0x2d0/0x2d0
> > >  __run_hrtimer+0x8a/0x1a0
> > >  __hrtimer_run_queues+0x91/0x130
> > >  hrtimer_interrupt+0x10f/0x220
> > >  __sysvec_apic_timer_interrupt+0x7b/0xd0
> > >  sysvec_apic_timer_interrupt+0x4f/0xd0
> > >  ? asm_sysvec_apic_timer_interrupt+0xa/0x20
> > >  asm_sysvec_apic_timer_interrupt+0x12/0x20
> > >  RIP: 0033:0x7fff196bf6f5
> > >
> > > Instead of calling put_task_struct() directly, we defer it using
> > > call_rcu(). A more natural approach would use a workqueue, but since
> > > in PREEMPT_RT, we can't allocate dynamic memory from atomic context,
> > > the code would become more complex because we would need to put the
> > > work_struct instance in the task_struct and initialize it when we
> > > allocate a new task_struct.
> > >
> > > Signed-off-by: Wander Lairson Costa <wander@redhat.com>
> > > Cc: Paul McKenney <paulmck@kernel.org>
> > > Cc: Thomas Gleixner <tglx@linutronix.de>
> > > ---
> > >  kernel/sched/build_policy.c |  1 +
> > >  kernel/sched/deadline.c     | 24 +++++++++++++++++++++++-
> > >  2 files changed, 24 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/kernel/sched/build_policy.c b/kernel/sched/build_policy.c
> > > index d9dc9ab3773f..f159304ee792 100644
> > > --- a/kernel/sched/build_policy.c
> > > +++ b/kernel/sched/build_policy.c
> > > @@ -28,6 +28,7 @@
> > >  #include <linux/suspend.h>
> > >  #include <linux/tsacct_kern.h>
> > >  #include <linux/vtime.h>
> > > +#include <linux/rcupdate.h>
> > >
> > >  #include <uapi/linux/sched/types.h>
> > >
> > > diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
> > > index 9ae8f41e3372..ab9301d4cc24 100644
> > > --- a/kernel/sched/deadline.c
> > > +++ b/kernel/sched/deadline.c
> > > @@ -1405,6 +1405,13 @@ static void update_curr_dl(struct rq *rq)
> > >       }
> > >  }
> > >
> > > +static void delayed_put_task_struct(struct rcu_head *rhp)
> > > +{
> > > +     struct task_struct *task = container_of(rhp, struct task_struct, rcu);
> > > +
> > > +     __put_task_struct(task);
> >
> > Please note that BH is disabled here.  Don't you therefore
> > need to schedule a workqueue handler?  Perhaps directly from
> > inactive_task_timer(), or maybe from this point.  If the latter, one
> > way to skip the extra step is to use queue_rcu_work().
> >
> 
> My initial work was using a workqueue [1,2]. However, I realized I
> could reach a much simpler code with call_rcu().
> I am afraid my ignorance doesn't allow me to get your point. Does
> disabling softirq imply atomic context?

Given that this problem occurred in PREEMPT_RT, I am assuming that the
appropriate definition of "atomic context" is "cannot call schedule()".
And you are in fact not permitted to call schedule() from a bh-disabled
region.

This also means that you cannot acquire a non-raw spinlock in a
bh-disabled region of code in a PREEMPT_RT kernel, because doing
so can invoke schedule.

Of course, using a workqueue does incur needless overhead in
non-PREEMPT_RT kernels.  So one alternative approach is to use
the workqueue only in PREEMPT_RT kernels and to just invoke
__put_task_struct() directly (without call_rcu() along the way)
otherwise.

Does that help, or am I missing your point?

						Thanx, Paul

> [1] https://gitlab.com/walac/kernel-ark/-/commit/ec8addbe38d5c318f1789b4c0fa480a9d2afdb65
> [2] https://gitlab.com/walac/kernel-ark/-/commit/0bde233235ffed233a7466a36a4866bc48064f54
> 
> 
> >                                                         Thanx, Paul
> >
> > > +}
> > > +
> > >  static enum hrtimer_restart inactive_task_timer(struct hrtimer *timer)
> > >  {
> > >       struct sched_dl_entity *dl_se = container_of(timer,
> > > @@ -1442,7 +1449,22 @@ static enum hrtimer_restart inactive_task_timer(struct hrtimer *timer)
> > >       dl_se->dl_non_contending = 0;
> > >  unlock:
> > >       task_rq_unlock(rq, p, &rf);
> > > -     put_task_struct(p);
> > > +
> > > +     if (IS_ENABLED(CONFIG_PREEMPT_RT)) {
> > > +             /*
> > > +              * Decrement the refcount explicitly to avoid unnecessarily
> > > +              * calling call_rcu.
> > > +              */
> > > +             if (refcount_dec_and_test(&p->usage))
> > > +                     /*
> > > +                      * under PREEMPT_RT, we can't call put_task_struct
> > > +                      * in atomic context because it will indirectly
> > > +                      * acquire sleeping locks.
> > > +                      */
> > > +                     call_rcu(&p->rcu, delayed_put_task_struct);
> > > +     } else {
> > > +             put_task_struct(p);
> > > +     }
> > >
> > >       return HRTIMER_NORESTART;
> > >  }
> > > --
> > > 2.39.0
> > >
> >
>
Valentin Schneider Jan. 18, 2023, 3:57 p.m. UTC | #4 
On 10/01/23 14:27, Paul E. McKenney wrote:
> On Tue, Jan 10, 2023 at 05:52:03PM -0300, Wander Lairson Costa wrote:
>> On Mon, Jan 9, 2023 at 10:40 PM Paul E. McKenney <paulmck@kernel.org> wrote:
>> >
>> > On Wed, Jan 04, 2023 at 03:17:01PM -0300, Wander Lairson Costa wrote:
>> > > inactive_task_timer() executes in interrupt (atomic) context. It calls
>> > > put_task_struct(), which indirectly acquires sleeping locks under
>> > > PREEMPT_RT.
>> > >
>> > > Below is an example of a splat that happened in a test environment:
>> > >
>> > >  CPU: 1 PID: 2848 Comm: life Kdump: loaded Tainted: G W ---------
>> > >  Hardware name: HP ProLiant DL388p Gen8, BIOS P70 07/15/2012
>> > >  Call Trace:
>> > >  dump_stack_lvl+0x57/0x7d
>> > >  mark_lock_irq.cold+0x33/0xba
>> > >  ? stack_trace_save+0x4b/0x70
>> > >  ? save_trace+0x55/0x150
>> > >  mark_lock+0x1e7/0x400
>> > >  mark_usage+0x11d/0x140
>> > >  __lock_acquire+0x30d/0x930
>> > >  lock_acquire.part.0+0x9c/0x210
>> > >  ? refill_obj_stock+0x3d/0x3a0
>> > >  ? rcu_read_lock_sched_held+0x3f/0x70
>> > >  ? trace_lock_acquire+0x38/0x140
>> > >  ? lock_acquire+0x30/0x80
>> > >  ? refill_obj_stock+0x3d/0x3a0
>> > >  rt_spin_lock+0x27/0xe0
>> > >  ? refill_obj_stock+0x3d/0x3a0
>> > >  refill_obj_stock+0x3d/0x3a0
>> > >  ? inactive_task_timer+0x1ad/0x340
>> > >  kmem_cache_free+0x357/0x560
>> > >  inactive_task_timer+0x1ad/0x340
>> > >  ? switched_from_dl+0x2d0/0x2d0
>> > >  __run_hrtimer+0x8a/0x1a0
>> > >  __hrtimer_run_queues+0x91/0x130
>> > >  hrtimer_interrupt+0x10f/0x220
>> > >  __sysvec_apic_timer_interrupt+0x7b/0xd0
>> > >  sysvec_apic_timer_interrupt+0x4f/0xd0
>> > >  ? asm_sysvec_apic_timer_interrupt+0xa/0x20
>> > >  asm_sysvec_apic_timer_interrupt+0x12/0x20
>> > >  RIP: 0033:0x7fff196bf6f5
>> > >
>> > > Instead of calling put_task_struct() directly, we defer it using
>> > > call_rcu(). A more natural approach would use a workqueue, but since
>> > > in PREEMPT_RT, we can't allocate dynamic memory from atomic context,
>> > > the code would become more complex because we would need to put the
>> > > work_struct instance in the task_struct and initialize it when we
>> > > allocate a new task_struct.
>> > >
>> > > Signed-off-by: Wander Lairson Costa <wander@redhat.com>
>> > > Cc: Paul McKenney <paulmck@kernel.org>
>> > > Cc: Thomas Gleixner <tglx@linutronix.de>
>> > > ---
>> > >  kernel/sched/build_policy.c |  1 +
>> > >  kernel/sched/deadline.c     | 24 +++++++++++++++++++++++-
>> > >  2 files changed, 24 insertions(+), 1 deletion(-)
>> > >
>> > > diff --git a/kernel/sched/build_policy.c b/kernel/sched/build_policy.c
>> > > index d9dc9ab3773f..f159304ee792 100644
>> > > --- a/kernel/sched/build_policy.c
>> > > +++ b/kernel/sched/build_policy.c
>> > > @@ -28,6 +28,7 @@
>> > >  #include <linux/suspend.h>
>> > >  #include <linux/tsacct_kern.h>
>> > >  #include <linux/vtime.h>
>> > > +#include <linux/rcupdate.h>
>> > >
>> > >  #include <uapi/linux/sched/types.h>
>> > >
>> > > diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
>> > > index 9ae8f41e3372..ab9301d4cc24 100644
>> > > --- a/kernel/sched/deadline.c
>> > > +++ b/kernel/sched/deadline.c
>> > > @@ -1405,6 +1405,13 @@ static void update_curr_dl(struct rq *rq)
>> > >       }
>> > >  }
>> > >
>> > > +static void delayed_put_task_struct(struct rcu_head *rhp)
>> > > +{
>> > > +     struct task_struct *task = container_of(rhp, struct task_struct, rcu);
>> > > +
>> > > +     __put_task_struct(task);
>> >
>> > Please note that BH is disabled here.  Don't you therefore
>> > need to schedule a workqueue handler?  Perhaps directly from
>> > inactive_task_timer(), or maybe from this point.  If the latter, one
>> > way to skip the extra step is to use queue_rcu_work().
>> >
>>
>> My initial work was using a workqueue [1,2]. However, I realized I
>> could reach a much simpler code with call_rcu().
>> I am afraid my ignorance doesn't allow me to get your point. Does
>> disabling softirq imply atomic context?
>
> Given that this problem occurred in PREEMPT_RT, I am assuming that the
> appropriate definition of "atomic context" is "cannot call schedule()".
> And you are in fact not permitted to call schedule() from a bh-disabled
> region.
>
> This also means that you cannot acquire a non-raw spinlock in a
> bh-disabled region of code in a PREEMPT_RT kernel, because doing
> so can invoke schedule.
>

But per the PREEMPT_RT lock "replacement", non-raw spinlocks end up
invoking schedule_rtlock(), which should be safe vs BH disabled
(local_lock() + rcu_read_lock()):

  6991436c2b5d ("sched/core: Provide a scheduling point for RT locks")

Unless I'm missing something else?
Paul E. McKenney Jan. 18, 2023, 6:11 p.m. UTC | #5 
On Wed, Jan 18, 2023 at 03:57:38PM +0000, Valentin Schneider wrote:
> On 10/01/23 14:27, Paul E. McKenney wrote:
> > On Tue, Jan 10, 2023 at 05:52:03PM -0300, Wander Lairson Costa wrote:
> >> On Mon, Jan 9, 2023 at 10:40 PM Paul E. McKenney <paulmck@kernel.org> wrote:
> >> >
> >> > On Wed, Jan 04, 2023 at 03:17:01PM -0300, Wander Lairson Costa wrote:
> >> > > inactive_task_timer() executes in interrupt (atomic) context. It calls
> >> > > put_task_struct(), which indirectly acquires sleeping locks under
> >> > > PREEMPT_RT.
> >> > >
> >> > > Below is an example of a splat that happened in a test environment:
> >> > >
> >> > >  CPU: 1 PID: 2848 Comm: life Kdump: loaded Tainted: G W ---------
> >> > >  Hardware name: HP ProLiant DL388p Gen8, BIOS P70 07/15/2012
> >> > >  Call Trace:
> >> > >  dump_stack_lvl+0x57/0x7d
> >> > >  mark_lock_irq.cold+0x33/0xba
> >> > >  ? stack_trace_save+0x4b/0x70
> >> > >  ? save_trace+0x55/0x150
> >> > >  mark_lock+0x1e7/0x400
> >> > >  mark_usage+0x11d/0x140
> >> > >  __lock_acquire+0x30d/0x930
> >> > >  lock_acquire.part.0+0x9c/0x210
> >> > >  ? refill_obj_stock+0x3d/0x3a0
> >> > >  ? rcu_read_lock_sched_held+0x3f/0x70
> >> > >  ? trace_lock_acquire+0x38/0x140
> >> > >  ? lock_acquire+0x30/0x80
> >> > >  ? refill_obj_stock+0x3d/0x3a0
> >> > >  rt_spin_lock+0x27/0xe0
> >> > >  ? refill_obj_stock+0x3d/0x3a0
> >> > >  refill_obj_stock+0x3d/0x3a0
> >> > >  ? inactive_task_timer+0x1ad/0x340
> >> > >  kmem_cache_free+0x357/0x560
> >> > >  inactive_task_timer+0x1ad/0x340
> >> > >  ? switched_from_dl+0x2d0/0x2d0
> >> > >  __run_hrtimer+0x8a/0x1a0
> >> > >  __hrtimer_run_queues+0x91/0x130
> >> > >  hrtimer_interrupt+0x10f/0x220
> >> > >  __sysvec_apic_timer_interrupt+0x7b/0xd0
> >> > >  sysvec_apic_timer_interrupt+0x4f/0xd0
> >> > >  ? asm_sysvec_apic_timer_interrupt+0xa/0x20
> >> > >  asm_sysvec_apic_timer_interrupt+0x12/0x20
> >> > >  RIP: 0033:0x7fff196bf6f5
> >> > >
> >> > > Instead of calling put_task_struct() directly, we defer it using
> >> > > call_rcu(). A more natural approach would use a workqueue, but since
> >> > > in PREEMPT_RT, we can't allocate dynamic memory from atomic context,
> >> > > the code would become more complex because we would need to put the
> >> > > work_struct instance in the task_struct and initialize it when we
> >> > > allocate a new task_struct.
> >> > >
> >> > > Signed-off-by: Wander Lairson Costa <wander@redhat.com>
> >> > > Cc: Paul McKenney <paulmck@kernel.org>
> >> > > Cc: Thomas Gleixner <tglx@linutronix.de>
> >> > > ---
> >> > >  kernel/sched/build_policy.c |  1 +
> >> > >  kernel/sched/deadline.c     | 24 +++++++++++++++++++++++-
> >> > >  2 files changed, 24 insertions(+), 1 deletion(-)
> >> > >
> >> > > diff --git a/kernel/sched/build_policy.c b/kernel/sched/build_policy.c
> >> > > index d9dc9ab3773f..f159304ee792 100644
> >> > > --- a/kernel/sched/build_policy.c
> >> > > +++ b/kernel/sched/build_policy.c
> >> > > @@ -28,6 +28,7 @@
> >> > >  #include <linux/suspend.h>
> >> > >  #include <linux/tsacct_kern.h>
> >> > >  #include <linux/vtime.h>
> >> > > +#include <linux/rcupdate.h>
> >> > >
> >> > >  #include <uapi/linux/sched/types.h>
> >> > >
> >> > > diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
> >> > > index 9ae8f41e3372..ab9301d4cc24 100644
> >> > > --- a/kernel/sched/deadline.c
> >> > > +++ b/kernel/sched/deadline.c
> >> > > @@ -1405,6 +1405,13 @@ static void update_curr_dl(struct rq *rq)
> >> > >       }
> >> > >  }
> >> > >
> >> > > +static void delayed_put_task_struct(struct rcu_head *rhp)
> >> > > +{
> >> > > +     struct task_struct *task = container_of(rhp, struct task_struct, rcu);
> >> > > +
> >> > > +     __put_task_struct(task);
> >> >
> >> > Please note that BH is disabled here.  Don't you therefore
> >> > need to schedule a workqueue handler?  Perhaps directly from
> >> > inactive_task_timer(), or maybe from this point.  If the latter, one
> >> > way to skip the extra step is to use queue_rcu_work().
> >> >
> >>
> >> My initial work was using a workqueue [1,2]. However, I realized I
> >> could reach a much simpler code with call_rcu().
> >> I am afraid my ignorance doesn't allow me to get your point. Does
> >> disabling softirq imply atomic context?
> >
> > Given that this problem occurred in PREEMPT_RT, I am assuming that the
> > appropriate definition of "atomic context" is "cannot call schedule()".
> > And you are in fact not permitted to call schedule() from a bh-disabled
> > region.
> >
> > This also means that you cannot acquire a non-raw spinlock in a
> > bh-disabled region of code in a PREEMPT_RT kernel, because doing
> > so can invoke schedule.
> 
> But per the PREEMPT_RT lock "replacement", non-raw spinlocks end up
> invoking schedule_rtlock(), which should be safe vs BH disabled
> (local_lock() + rcu_read_lock()):
> 
>   6991436c2b5d ("sched/core: Provide a scheduling point for RT locks")
> 
> Unless I'm missing something else?

No, you miss nothing.  Apologies for my confusion!

(I could have sworn that someone else corrected me on this earlier,
but I don't see it right off hand.)

							Thanx, Paul
Valentin Schneider Jan. 18, 2023, 6:28 p.m. UTC | #6 
On 18/01/23 10:11, Paul E. McKenney wrote:
> On Wed, Jan 18, 2023 at 03:57:38PM +0000, Valentin Schneider wrote:
>> > Given that this problem occurred in PREEMPT_RT, I am assuming that the
>> > appropriate definition of "atomic context" is "cannot call schedule()".
>> > And you are in fact not permitted to call schedule() from a bh-disabled
>> > region.
>> >
>> > This also means that you cannot acquire a non-raw spinlock in a
>> > bh-disabled region of code in a PREEMPT_RT kernel, because doing
>> > so can invoke schedule.
>>
>> But per the PREEMPT_RT lock "replacement", non-raw spinlocks end up
>> invoking schedule_rtlock(), which should be safe vs BH disabled
>> (local_lock() + rcu_read_lock()):
>>
>>   6991436c2b5d ("sched/core: Provide a scheduling point for RT locks")
>>
>> Unless I'm missing something else?
>
> No, you miss nothing.  Apologies for my confusion!
>
> (I could have sworn that someone else corrected me on this earlier,
> but I don't see it right off hand.)
>
>                                                       Thanx, Paul

Heh, I had a smidge of doubt myself, but since we've cleared this up:

Reviewed-by: Valentin Schneider <vschneid@redhat.com>
Valentin Schneider Jan. 19, 2023, 6:03 p.m. UTC | #7 
On 04/01/23 15:17, Wander Lairson Costa wrote:
> inactive_task_timer() executes in interrupt (atomic) context. It calls
> put_task_struct(), which indirectly acquires sleeping locks under
> PREEMPT_RT.
>
> Below is an example of a splat that happened in a test environment:
>
>  CPU: 1 PID: 2848 Comm: life Kdump: loaded Tainted: G W ---------
>  Hardware name: HP ProLiant DL388p Gen8, BIOS P70 07/15/2012
>  Call Trace:
>  dump_stack_lvl+0x57/0x7d
>  mark_lock_irq.cold+0x33/0xba
>  ? stack_trace_save+0x4b/0x70
>  ? save_trace+0x55/0x150
>  mark_lock+0x1e7/0x400
>  mark_usage+0x11d/0x140
>  __lock_acquire+0x30d/0x930
>  lock_acquire.part.0+0x9c/0x210
>  ? refill_obj_stock+0x3d/0x3a0
>  ? rcu_read_lock_sched_held+0x3f/0x70
>  ? trace_lock_acquire+0x38/0x140
>  ? lock_acquire+0x30/0x80
>  ? refill_obj_stock+0x3d/0x3a0
>  rt_spin_lock+0x27/0xe0
>  ? refill_obj_stock+0x3d/0x3a0
>  refill_obj_stock+0x3d/0x3a0
>  ? inactive_task_timer+0x1ad/0x340
>  kmem_cache_free+0x357/0x560
>  inactive_task_timer+0x1ad/0x340
>  ? switched_from_dl+0x2d0/0x2d0
>  __run_hrtimer+0x8a/0x1a0
>  __hrtimer_run_queues+0x91/0x130
>  hrtimer_interrupt+0x10f/0x220
>  __sysvec_apic_timer_interrupt+0x7b/0xd0
>  sysvec_apic_timer_interrupt+0x4f/0xd0
>  ? asm_sysvec_apic_timer_interrupt+0xa/0x20
>  asm_sysvec_apic_timer_interrupt+0x12/0x20
>  RIP: 0033:0x7fff196bf6f5
>
> Instead of calling put_task_struct() directly, we defer it using
> call_rcu(). A more natural approach would use a workqueue, but since
> in PREEMPT_RT, we can't allocate dynamic memory from atomic context,
> the code would become more complex because we would need to put the
> work_struct instance in the task_struct and initialize it when we
> allocate a new task_struct.
>

Sorry to come back on this; Juri reminded me offline that put_task_struct()
is invoked in other non-sleepable contexts, not just inactive_task_timer().

e.g.

  rto_push_irq_work_func() // hard irq work so hardirq context
  `\
    push_rt_task()
    `\
       put_task_struct()

Or

  cpu_stopper_thread() // stopper callbacks must not sleep
  `\
    push_cpu_stop()
    `\
      put_task_struct()

... But then again I'm not aware of any splats happening in these paths. Is
there something special about inactive_task_timer(), or could it be the
issue is there for those other paths but we just haven't had them reported
yet?
Wander Lairson Costa Jan. 19, 2023, 7:53 p.m. UTC | #8 
On Thu, Jan 19, 2023 at 3:03 PM Valentin Schneider <vschneid@redhat.com> wrote:
>
> On 04/01/23 15:17, Wander Lairson Costa wrote:
> > inactive_task_timer() executes in interrupt (atomic) context. It calls
> > put_task_struct(), which indirectly acquires sleeping locks under
> > PREEMPT_RT.
> >
> > Below is an example of a splat that happened in a test environment:
> >
> >  CPU: 1 PID: 2848 Comm: life Kdump: loaded Tainted: G W ---------
> >  Hardware name: HP ProLiant DL388p Gen8, BIOS P70 07/15/2012
> >  Call Trace:
> >  dump_stack_lvl+0x57/0x7d
> >  mark_lock_irq.cold+0x33/0xba
> >  ? stack_trace_save+0x4b/0x70
> >  ? save_trace+0x55/0x150
> >  mark_lock+0x1e7/0x400
> >  mark_usage+0x11d/0x140
> >  __lock_acquire+0x30d/0x930
> >  lock_acquire.part.0+0x9c/0x210
> >  ? refill_obj_stock+0x3d/0x3a0
> >  ? rcu_read_lock_sched_held+0x3f/0x70
> >  ? trace_lock_acquire+0x38/0x140
> >  ? lock_acquire+0x30/0x80
> >  ? refill_obj_stock+0x3d/0x3a0
> >  rt_spin_lock+0x27/0xe0
> >  ? refill_obj_stock+0x3d/0x3a0
> >  refill_obj_stock+0x3d/0x3a0
> >  ? inactive_task_timer+0x1ad/0x340
> >  kmem_cache_free+0x357/0x560
> >  inactive_task_timer+0x1ad/0x340
> >  ? switched_from_dl+0x2d0/0x2d0
> >  __run_hrtimer+0x8a/0x1a0
> >  __hrtimer_run_queues+0x91/0x130
> >  hrtimer_interrupt+0x10f/0x220
> >  __sysvec_apic_timer_interrupt+0x7b/0xd0
> >  sysvec_apic_timer_interrupt+0x4f/0xd0
> >  ? asm_sysvec_apic_timer_interrupt+0xa/0x20
> >  asm_sysvec_apic_timer_interrupt+0x12/0x20
> >  RIP: 0033:0x7fff196bf6f5
> >
> > Instead of calling put_task_struct() directly, we defer it using
> > call_rcu(). A more natural approach would use a workqueue, but since
> > in PREEMPT_RT, we can't allocate dynamic memory from atomic context,
> > the code would become more complex because we would need to put the
> > work_struct instance in the task_struct and initialize it when we
> > allocate a new task_struct.
> >
>
> Sorry to come back on this; Juri reminded me offline that put_task_struct()
> is invoked in other non-sleepable contexts, not just inactive_task_timer().
>

I guess there is no splat because the usage count doesn't reach zero
in those code paths.

> e.g.
>
>   rto_push_irq_work_func() // hard irq work so hardirq context
>   `\
>     push_rt_task()
>     `\
>        put_task_struct()
>

This is paired with a get_task_struct() a few lines above inside in
the same function.

> Or
>
>   cpu_stopper_thread() // stopper callbacks must not sleep
>   `\
>     push_cpu_stop()
>     `\
>       put_task_struct()
>

This is paired with a get_task_struct() from get_push_task()

> ... But then again I'm not aware of any splats happening in these paths. Is
> there something special about inactive_task_timer(), or could it be the
> issue is there for those other paths but we just haven't had them reported
> yet?
>

Given that those calls have corresponding get_task_struct() calls that
are close in time, there is a low probability of the usage count
reaching zero and triggering the splat. In any case, I will work in a
v2 that also addresses those call sites.
diff mbox series

Patch

diff --git a/kernel/sched/build_policy.c b/kernel/sched/build_policy.c
index d9dc9ab3773f..f159304ee792 100644
--- a/kernel/sched/build_policy.c
+++ b/kernel/sched/build_policy.c
@@ -28,6 +28,7 @@ 
 #include <linux/suspend.h>
 #include <linux/tsacct_kern.h>
 #include <linux/vtime.h>
+#include <linux/rcupdate.h>
 
 #include <uapi/linux/sched/types.h>
 
diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
index 9ae8f41e3372..ab9301d4cc24 100644
--- a/kernel/sched/deadline.c
+++ b/kernel/sched/deadline.c
@@ -1405,6 +1405,13 @@  static void update_curr_dl(struct rq *rq)
 	}
 }
 
+static void delayed_put_task_struct(struct rcu_head *rhp)
+{
+	struct task_struct *task = container_of(rhp, struct task_struct, rcu);
+
+	__put_task_struct(task);
+}
+
 static enum hrtimer_restart inactive_task_timer(struct hrtimer *timer)
 {
 	struct sched_dl_entity *dl_se = container_of(timer,
@@ -1442,7 +1449,22 @@  static enum hrtimer_restart inactive_task_timer(struct hrtimer *timer)
 	dl_se->dl_non_contending = 0;
 unlock:
 	task_rq_unlock(rq, p, &rf);
-	put_task_struct(p);
+
+	if (IS_ENABLED(CONFIG_PREEMPT_RT)) {
+		/*
+		 * Decrement the refcount explicitly to avoid unnecessarily
+		 * calling call_rcu.
+		 */
+		if (refcount_dec_and_test(&p->usage))
+			/*
+			 * under PREEMPT_RT, we can't call put_task_struct
+			 * in atomic context because it will indirectly
+			 * acquire sleeping locks.
+			 */
+			call_rcu(&p->rcu, delayed_put_task_struct);
+	} else {
+		put_task_struct(p);
+	}
 
 	return HRTIMER_NORESTART;
 }