RTC: Avoid races between RTC alarm wakeup and suspend.

PATCH Applied NeilBrown 2 participants 4 messages Jul 30, 2012 – Aug 5, 2012 Real time clock (RTC)

lore

Share & tools

lore

NeilBrown

author

2012-07-30 01:36 UTC #0 lore raw

PATCH 1/1 RTC: Avoid races between RTC alarm wakeup and suspend.

If an RTC alarm fires just as suspend is happening, it is possible for suspend to complete and the alarm to be missed.

To avoid the race, we must register the event with the PM core.

As the event is made visible to userspace through a thread which is only scheduled by the interrupt, we need a pm_stay_awake/pm_relax pair preventing suspend from the interrupt until the thread completes its work.

Signed-off-by NeilBrown <[email protected]>

--
This makes the pm_wakeup_event() call in cmos_interrupt unnecessary as it provides suspend protection for all RTCs that use rtc_update_irq.

I think the pm_stay_awake//pm_relax is needed - just pm_wakup_event() is theoretically not sufficient.

This is because there is no guarantee (that I know of) that the workqueue thread will actually get scheduled before 'suspend' takes over.

Thanks,
NeilBrown

drivers/rtc/interface.c +2 -0

index eb415bd..9592b93 100644
@@ -582,6 +582,7 @@ enum hrtimer_restart rtc_pie_update_irq(struct hrtimer *timer)
 void rtc_update_irq(struct rtc_device *rtc,
 		unsigned long num, unsigned long events)
 {
+	pm_stay_awake(rtc->dev.parent);
 	schedule_work(&rtc->irqwork);
 }
 EXPORT_SYMBOL_GPL(rtc_update_irq);

@@ -844,6 +845,7 @@ void rtc_timer_do_work(struct work_struct *work)
 
 	mutex_lock(&rtc->ops_lock);
 again:
+	pm_relax(rtc->dev.parent);
 	__rtc_read_time(rtc, &tm);
 	now = rtc_tm_to_ktime(tm);
 	while ((next = timerqueue_getnext(&rtc->timerqueue))) {

Rafael J. Wysocki

2012-07-30 20:56 UTC #1 lore raw

Quoted 11 lines · NeilBrown

If an RTC alarm fires just as suspend is happening, it is possible for suspend to complete and the alarm to be missed.
To avoid the race, we must register the event with the PM core.
As the event is made visible to userspace through a thread which is only scheduled by the interrupt, we need a pm_stay_awake/pm_relax pair preventing suspend from the interrupt until the thread completes its work.

Quoted 5 lines

Signed-off-by NeilBrown <[email protected]>
--
This makes the pm_wakeup_event() call in cmos_interrupt unnecessary as it provides suspend protection for all RTCs that use rtc_update_irq.

Care to remove the call in cmos_interrupt(), then?

Quoted 5 lines

I think the pm_stay_awake//pm_relax is needed - just pm_wakup_event() is theoretically not sufficient.
This is because there is no guarantee (that I know of) that the workqueue thread will actually get scheduled before 'suspend' takes over.

I think you are right.

Thanks,
Rafael

Quoted 21 lines

drivers/rtc/interface.c +2 -0

index eb415bd..9592b93 100644
@@ -582,6 +582,7 @@ enum hrtimer_restart rtc_pie_update_irq(struct hrtimer *timer)
 void rtc_update_irq(struct rtc_device *rtc,
 		unsigned long num, unsigned long events)
 {
+	pm_stay_awake(rtc->dev.parent);
 	schedule_work(&rtc->irqwork);
 }
 EXPORT_SYMBOL_GPL(rtc_update_irq);

@@ -844,6 +845,7 @@ void rtc_timer_do_work(struct work_struct *work)
 
 	mutex_lock(&rtc->ops_lock);
 again:
+	pm_relax(rtc->dev.parent);
 	__rtc_read_time(rtc, &tm);
 	now = rtc_tm_to_ktime(tm);
 	while ((next = timerqueue_getnext(&rtc->timerqueue))) {

NeilBrown

author

2012-07-31 02:51 UTC #2 lore raw

Quoted 20 lines · On Mon, 30 Jul 2012 23:01:49 +0200 "Rafael J. Wysocki" <[email protected]> wrote:

Quoted 11 lines · NeilBrown
If an RTC alarm fires just as suspend is happening, it is possible for suspend to complete and the alarm to be missed.
To avoid the race, we must register the event with the PM core.
As the event is made visible to userspace through a thread which is only scheduled by the interrupt, we need a pm_stay_awake/pm_relax pair preventing suspend from the interrupt until the thread completes its work.
Quoted 5 lines
Signed-off-by NeilBrown <[email protected]>
--
This makes the pm_wakeup_event() call in cmos_interrupt unnecessary as it provides suspend protection for all RTCs that use rtc_update_irq.
Care to remove the call in cmos_interrupt(), then?

Quoted 7 lines

Quoted 5 lines
I think the pm_stay_awake//pm_relax is needed - just pm_wakup_event() is theoretically not sufficient.
This is because there is no guarantee (that I know of) that the workqueue thread will actually get scheduled before 'suspend' takes over.
I think you are right.

Thanks.
Here is the revised patch.

NeilBrown

==========
Subject: [PATCH] RTC: Avoid races between RTC alarm wakeup and suspend.

If an RTC alarm fires just as suspend is happening, it is possible for suspend to complete and the alarm to be missed.

To avoid the race, we must register the event with the PM core.

This makes the pm_wakeup_event() call in cmos_interrupt unnecessary as it provides suspend protection for all RTCs that use rtc_update_irq.

Cc Paul Fox <[email protected]>

Signed-off-by NeilBrown <[email protected]>

drivers/rtc/interface.c +2 -0

index eb415bd..9592b93 100644
@@ -582,6 +582,7 @@ enum hrtimer_restart rtc_pie_update_irq(struct hrtimer *timer)
 void rtc_update_irq(struct rtc_device *rtc,
 		unsigned long num, unsigned long events)
 {
+	pm_stay_awake(rtc->dev.parent);
 	schedule_work(&rtc->irqwork);
 }
 EXPORT_SYMBOL_GPL(rtc_update_irq);

@@ -844,6 +845,7 @@ void rtc_timer_do_work(struct work_struct *work)
 
 	mutex_lock(&rtc->ops_lock);
 again:
+	pm_relax(rtc->dev.parent);
 	__rtc_read_time(rtc, &tm);
 	now = rtc_tm_to_ktime(tm);
 	while ((next = timerqueue_getnext(&rtc->timerqueue))) {

drivers/rtc/rtc-cmos.c +0 -1

index 132333d..4267789 100644
@@ -568,7 +568,6 @@ static irqreturn_t cmos_interrupt(int irq, void *p)
 		hpet_mask_rtc_irq_bit(RTC_AIE);
 
 		CMOS_READ(RTC_INTR_FLAGS);
-		pm_wakeup_event(cmos_rtc.dev, 0);
 	}
 	spin_unlock(&rtc_lock);

Rafael J. Wysocki

Applied

2012-08-05 21:05 UTC #3 lore raw

Quoted 49 lines · NeilBrown

Quoted 20 lines · On Mon, 30 Jul 2012 23:01:49 +0200 "Rafael J. Wysocki" <[email protected]> wrote:
Quoted 11 lines · NeilBrown
If an RTC alarm fires just as suspend is happening, it is possible for suspend to complete and the alarm to be missed.
To avoid the race, we must register the event with the PM core.
As the event is made visible to userspace through a thread which is only scheduled by the interrupt, we need a pm_stay_awake/pm_relax pair preventing suspend from the interrupt until the thread completes its work.
Quoted 5 lines
Signed-off-by NeilBrown <[email protected]>
--
This makes the pm_wakeup_event() call in cmos_interrupt unnecessary as it provides suspend protection for all RTCs that use rtc_update_irq.
Care to remove the call in cmos_interrupt(), then?
Quoted 7 lines
Quoted 5 lines
I think the pm_stay_awake//pm_relax is needed - just pm_wakup_event() is theoretically not sufficient.
This is because there is no guarantee (that I know of) that the workqueue thread will actually get scheduled before 'suspend' takes over.
I think you are right.
Thanks.
Here is the revised patch.
NeilBrown
==========
Subject: [PATCH] RTC: Avoid races between RTC alarm wakeup and suspend.
If an RTC alarm fires just as suspend is happening, it is possible for suspend to complete and the alarm to be missed.
To avoid the race, we must register the event with the PM core.
As the event is made visible to userspace through a thread which is only scheduled by the interrupt, we need a pm_stay_awake/pm_relax pair preventing suspend from the interrupt until the thread completes its work.

Quoted 5 lines

This makes the pm_wakeup_event() call in cmos_interrupt unnecessary as it provides suspend protection for all RTCs that use rtc_update_irq.
Cc Paul Fox <[email protected]>
Signed-off-by NeilBrown <[email protected]>

Applied to the linux-next branch of the linux-pm.git tree. I'm going to push it as a fix for v3.6.

Thanks,
Rafael

Quoted 33 lines

drivers/rtc/interface.c +2 -0

index eb415bd..9592b93 100644
@@ -582,6 +582,7 @@ enum hrtimer_restart rtc_pie_update_irq(struct hrtimer *timer)
 void rtc_update_irq(struct rtc_device *rtc,
 		unsigned long num, unsigned long events)
 {
+	pm_stay_awake(rtc->dev.parent);
 	schedule_work(&rtc->irqwork);
 }
 EXPORT_SYMBOL_GPL(rtc_update_irq);

@@ -844,6 +845,7 @@ void rtc_timer_do_work(struct work_struct *work)
 
 	mutex_lock(&rtc->ops_lock);
 again:
+	pm_relax(rtc->dev.parent);
 	__rtc_read_time(rtc, &tm);
 	now = rtc_tm_to_ktime(tm);
 	while ((next = timerqueue_getnext(&rtc->timerqueue))) {

drivers/rtc/rtc-cmos.c +0 -1

index 132333d..4267789 100644
@@ -568,7 +568,6 @@ static irqreturn_t cmos_interrupt(int irq, void *p)
 		hpet_mask_rtc_irq_bit(RTC_AIE);
 
 		CMOS_READ(RTC_INTR_FLAGS);
-		pm_wakeup_event(cmos_rtc.dev, 0);
 	}
 	spin_unlock(&rtc_lock);