kthread: Fix the race condition when kthread is parked

PATCH In review Subbaraman Narayanamurthy 3 participants 6 messages Jun 25, 2014 – Jun 26, 2014

lore

Share & tools

lore

Subbaraman Narayanamurthy

author

2014-06-25 19:43 UTC #0 lore raw

PATCH 1/1 kthread: Fix the race condition when kthread is parked

While stressing the CPU hotplug path, sometimes we hit a problem as shown below.

[57056.416774] ------------[ cut here ]------------
[57056.489232] ksoftirqd/1 (14): undefined instruction: pc=c01931e8
[57056.489245] Code: e594a000 eb085236 e15a0000 0a000000 (e7f001f2)
[57056.489259] ------------[ cut here ]------------
[57056.492840] kernel BUG at kernel/kernel/smpboot.c:134!
[57056.513236] Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM
[57056.519055] Modules linked in: wlan(O) mhi(O)
[57056.523394] CPU: 0 PID: 14 Comm: ksoftirqd/1 Tainted: G        W O 3.10.0-g3677c61-00008-g180c060 #1
[57056.532595] task: f0c8b000 ti: f0e78000 task.ti: f0e78000
[57056.537991] PC is at smpboot_thread_fn+0x124/0x218
[57056.542750] LR is at smpboot_thread_fn+0x11c/0x218
[57056.547528] pc : [<c01931e8>]    lr : [<c01931e0>] psr: 200f0013
[57056.547528] sp : f0e79f30  ip : 00000000  fp : 00000000
[57056.558983] r10: 00000001  r9 : 00000000  r8 : f0e78000
[57056.564192] r7 : 00000001  r6 : c1195758  r5 : f0e78000  r4 : f0e5fd00
[57056.570701] r3 : 00000001  r2 : f0e79f20  r1 : 00000000  r0 : 00000000

This issue was always seen in the context of "ksoftirqd". It seems to be happening because of a potential race condition in __kthread_parkme where just after completing the parked completion, before the ksoftirqd task has been scheduled again, it can go into running state.

Fix this by waiting for the task state to parked after waiting for the parked completion.

Signed-off-by Subbaraman Narayanamurthy <[email protected]>

---
 kernel/kthread.c | 2 ++
 1 file changed, 2 insertions(+)

kernel/kthread.c +2 -0

index 760e86d..c56c6f8 100644
@@ -398,6 +398,8 @@ int kthread_park(struct task_struct *k)
             if (k != current) {
                 wake_up_process(k);
                 wait_for_completion(&kthread->parked);
+                while (k->state != TASK_PARKED)
+                    cond_resched();
             }
         }
         ret = 0;

Signature · 5 lines

1.8.2.1

-- 
The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation

Thomas Gleixner

2014-06-26 00:44 UTC #1 lore raw

Quoted 21 lines · Subbaraman Narayanamurthy

While stressing the CPU hotplug path, sometimes we hit a problem as shown below.

[57056.416774] ------------[ cut here ]------------
[57056.489232] ksoftirqd/1 (14): undefined instruction: pc=c01931e8
[57056.489245] Code: e594a000 eb085236 e15a0000 0a000000 (e7f001f2)
[57056.489259] ------------[ cut here ]------------
[57056.492840] kernel BUG at kernel/kernel/smpboot.c:134!
[57056.513236] Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM
[57056.519055] Modules linked in: wlan(O) mhi(O)
[57056.523394] CPU: 0 PID: 14 Comm: ksoftirqd/1 Tainted: G        W O
3.10.0-g3677c61-00008-g180c060 #1
[57056.532595] task: f0c8b000 ti: f0e78000 task.ti: f0e78000
[57056.537991] PC is at smpboot_thread_fn+0x124/0x218
[57056.542750] LR is at smpboot_thread_fn+0x11c/0x218
[57056.547528] pc : [<c01931e8>]    lr : [<c01931e0>] psr: 200f0013
[57056.547528] sp : f0e79f30  ip : 00000000  fp : 00000000
[57056.558983] r10: 00000001  r9 : 00000000  r8 : f0e78000
[57056.564192] r7 : 00000001  r6 : c1195758  r5 : f0e78000  r4 : f0e5fd00
[57056.570701] r3 : 00000001  r2 : f0e79f20  r1 : 00000000  r0 : 00000000

Quoted 4 lines

This issue was always seen in the context of "ksoftirqd". It seems to be happening because of a potential race condition in __kthread_parkme where just after completing the parked completion, before the ksoftirqd task has been scheduled again, it can go into running state.

This explanation does not make any sense. You completely fail to explain the details of the race. And your patch does not make any sense either, because the real issue is this:

Task CPU 0 CPU 1

T1 unplug cpu1

       kthread_park(T2)
       set_bit(KTHREAD_SHOULD_PARK);
	  wait_for_completion()
T2					parkme(X)
				   	  __set_current_state(TASK_PARKED);
				   	  while (test_bit(KTHREAD_SHOULD_PARK)) {
				     	    if (!test_and_set_bit(KTHREAD_IS_PARKED))
				              complete();
			             	    schedule();

T1 plug cpu1

--> premature wakeup of T2, i.e. before unpark, so T2 gets scheduled on
CPU 0

T2 __set_current_state(TASK_PARKED);

--> Preemption by the plug thread

T1 thread_unpark(T2)

         clear_bit(KTHREAD_SHOULD_PARK);

--> Preemption by the softirq thread which breaks out of the
while(test_bit(KTHREAD_SHOULD_PARK)) loop because KTHREAD_SHOULD_PARK is not longer set.

T2 }

    clear_bit(KTHREAD_IS_PARKED);

--> Now T2 happily continues to run on CPU0 which rightfully casues
the BUG to trigger.

	__kthread_bind(T2)

--> Too late ....

So the real issue is that the park/unpark code is not able to handle the premature wakeup of T2 and that needs to be fixed.

Your changelog says:

Quoted 1 line

It seems to be happening because of a potential race condition in

Potential is wrong to begin with. A race condition is either real and explainable or it does not exist.

Quoted 3 lines

__kthread_parkme where just after completing the parked completion, before the ksoftirqd task has been scheduled again, it can go into running state.

What exactly has this to do with state RUNNING or PARKED?

Nothing, the task state is completely irrelevant as the real issue is the task->*PARK flags state.

So what is your patch solving here ?

You put a wait for task->state == TASK_PARKED after the wait_for_completion. What does it solve? Actually nothing. It just changes the propability of that issue. Go and apply it between any step of the above and figure out what it solves. Nothing, really.

Now just as an extra thought experiment assume that you have only two cpus and T1 is a SCHED_FIFO task and T2 is SCHED_OTHER ....

Please do not misunderstand me, but "fixing" races without proper understanding them is plain wrong. Providing a vague changelog which does neither explain what the issue is and why the fix works is even more wrong.

The next time you hit something like this, please take the time and sit down, get out the old fashioned piece of paper and a pencil and draw the picture so you can actually understand what the root cause of the observed issue is before sending out halfarsed duct tape fixes which just paper over the root cause. If you cannot figure it out, send a proper bug report.

Thanks,

tglx
------------------>

Subject: kthread: Plug park/ unplug race
From: Thomas Gleixner <[email protected]>
Date: Thu, 26 Jun 2014 01:24:36 +0200

The kthread park/unpark logic has the following issue:

Task CPU 0 CPU 1

T1 unplug cpu1

       kthread_park(T2)
       set_bit(KTHREAD_SHOULD_PARK);
	  wait_for_completion()
T2					parkme(X)
				   	  __set_current_state(TASK_PARKED);
				   	  while (test_bit(KTHREAD_SHOULD_PARK)) {
				     	    if (!test_and_set_bit(KTHREAD_IS_PARKED))
				              complete();
			             	    schedule();

T1 plug cpu1

--> premature wakeup of T2, i.e. before unpark, so T2 gets scheduled on
CPU 0

T2 __set_current_state(TASK_PARKED);

--> Preemption by the plug thread

T1 thread_unpark(T2)

         clear_bit(KTHREAD_SHOULD_PARK);

--> Preemption by the softirq thread which breaks out of the
while(test_bit(KTHREAD_SHOULD_PARK)) loop because KTHREAD_SHOULD_PARK is not longer set.

T2 }

    clear_bit(KTHREAD_IS_PARKED);

--> Now T2 happily continues to run on CPU0 which rightfully causes

    the BUG_ON(T2->cpu != smp_processor_id()) to trigger.

	__kthread_bind(T2)

--> Too late ....

Reorder the logic so that the unplug code binds the thread to the target cpu before clearing the KTHREAD_SHOULD_PARK bit.

Reported-by Subbaraman Narayanamurthy <[email protected]>

Signed-off-by Thomas Gleixner <[email protected]>

Cc [email protected]

---
 kernel/kthread.c |   14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)

kernel/kthread.c +10 -4

Index: linux/kernel/kthread.c
===================================================================
@@ -382,6 +382,15 @@ struct task_struct *kthread_create_on_cp
 
 static void __kthread_unpark(struct task_struct *k, struct kthread *kthread)
 {
+	/*
+	 * Rebind the thread to the target cpu first if it is a per
+	 * cpu thread unconditionally because it must be bound to the
+	 * target cpu before it can observe the KTHREAD_SHOULD_PARK
+	 * bit cleared.
+	 */
+	if (test_bit(KTHREAD_IS_PER_CPU, &kthread->flags))
+		__kthread_bind(k, kthread->cpu, TASK_PARKED);
+
 	clear_bit(KTHREAD_SHOULD_PARK, &kthread->flags);
 	/*
 	 * We clear the IS_PARKED bit here as we don't wait

@@ -389,11 +398,8 @@ static void __kthread_unpark(struct task
 	 * park before that happens we'd see the IS_PARKED bit
 	 * which might be about to be cleared.
 	 */
-	if (test_and_clear_bit(KTHREAD_IS_PARKED, &kthread->flags)) {
-		if (test_bit(KTHREAD_IS_PER_CPU, &kthread->flags))
-			__kthread_bind(k, kthread->cpu, TASK_PARKED);
+	if (test_and_clear_bit(KTHREAD_IS_PARKED, &kthread->flags))
 		wake_up_state(k, TASK_PARKED);
-	}
 }
 
 /**

Steven Rostedt

2014-06-26 02:00 UTC #2 lore raw

Quoted 7 lines · Thomas Gleixner

Subject: kthread: Plug park/ unplug race
From: Thomas Gleixner <[email protected]>
Date: Thu, 26 Jun 2014 01:24:36 +0200
The kthread park/unpark logic has the following issue:

Quoted 7 lines

Task CPU 0 CPU 1
T1 unplug cpu1
       kthread_park(T2)
       set_bit(KTHREAD_SHOULD_PARK);
	  wait_for_completion()
T2 parkme(X)

But with your patch, isn't it possible for T1 to call thread_unpark here?

Then looking at the code I see this turn of events:

  if (test_bit(KTHREAD_IS_PER_CPU, &kthread->flags))
        __kthread_bind(k, kthread->cpu, TASK_PARKED);

Which in __kthread_bind() (state == TASK_PARKED)

	if (!wait_task_inactive(p, state)) {
		WARN_ON(1);
		return;
	}

Where wait_task_inactive() does:

		while (task_running(rq, p)) {
			if (match_state && unlikely(p->state != match_state))
				return 0;

As match_state is non zero and p->state != match_state because it hasn't been set yet. The wait_task_inactive() returns zero, and then we hit the WARN_ON() in __kthread_bind().

-- Steve

Quoted 87 lines

				   	  __set_current_state(TASK_PARKED);
				   	  while (test_bit(KTHREAD_SHOULD_PARK)) {
				     	    if (!test_and_set_bit(KTHREAD_IS_PARKED))
				              complete();
			             	    schedule();
T1 plug cpu1
--> premature wakeup of T2, i.e. before unpark, so T2 gets scheduled on
CPU 0
T2 __set_current_state(TASK_PARKED);
--> Preemption by the plug thread
T1 thread_unpark(T2)
         clear_bit(KTHREAD_SHOULD_PARK);
--> Preemption by the softirq thread which breaks out of the
while(test_bit(KTHREAD_SHOULD_PARK)) loop because KTHREAD_SHOULD_PARK is not longer set.
T2 }
    clear_bit(KTHREAD_IS_PARKED);
--> Now T2 happily continues to run on CPU0 which rightfully causes
    the BUG_ON(T2->cpu != smp_processor_id()) to trigger.
T1
	__kthread_bind(T2)
--> Too late ....
Reorder the logic so that the unplug code binds the thread to the target cpu before clearing the KTHREAD_SHOULD_PARK bit.
Reported-by Subbaraman Narayanamurthy <[email protected]>
Signed-off-by Thomas Gleixner <[email protected]>
Cc [email protected]
---
 kernel/kthread.c |   14 ++++++++++----
 1 file changed, 10 insertions(+), 4 deletions(-)
Index: linux/kernel/kthread.c
===================================================================
kernel/kthread.c +10 -5
@@ -382,6 +382,15 @@ struct task_struct *kthread_create_on_cp
 
 static void __kthread_unpark(struct task_struct *k, struct kthread *kthread)
 {
+	/*
+	 * Rebind the thread to the target cpu first if it is a per
+	 * cpu thread unconditionally because it must be bound to the
+	 * target cpu before it can observe the KTHREAD_SHOULD_PARK
+	 * bit cleared.
+	 */
+	if (test_bit(KTHREAD_IS_PER_CPU, &kthread->flags))
+		__kthread_bind(k, kthread->cpu, TASK_PARKED);
+
 	clear_bit(KTHREAD_SHOULD_PARK, &kthread->flags);
 	/*
 	 * We clear the IS_PARKED bit here as we don't wait
@@ -389,11 +398,8 @@ static void __kthread_unpark(struct task
 	 * park before that happens we'd see the IS_PARKED bit
 	 * which might be about to be cleared.
 	 */
-	if (test_and_clear_bit(KTHREAD_IS_PARKED, &kthread->flags)) {
-		if (test_bit(KTHREAD_IS_PER_CPU, &kthread->flags))
-			__kthread_bind(k, kthread->cpu, TASK_PARKED);
+	if (test_and_clear_bit(KTHREAD_IS_PARKED, &kthread->flags))
 		wake_up_state(k, TASK_PARKED);
-	}
 }
 
 /**
 
 
 
 
 
 
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Steven Rostedt

2014-06-26 02:03 UTC #3 lore raw

Quoted 10 lines · Steven Rostedt

Quoted 9 lines · Thomas Gleixner
Subject: kthread: Plug park/ unplug race
From: Thomas Gleixner <[email protected]>
Date: Thu, 26 Jun 2014 01:24:36 +0200
The kthread park/unpark logic has the following issue:
Task CPU 0 CPU 1

Quoted 7 lines

Quoted 5 lines
T1 unplug cpu1
       kthread_park(T2)
       set_bit(KTHREAD_SHOULD_PARK);
	  wait_for_completion()
T2 parkme(X)
But with your patch, isn't it possible for T1 to call thread_unpark here?

Let me answer that.... No, it can't.

I missed the wait_for_completion() above, which will prevent this from happening.

Nevermind, I'll go work on something less brain intensive.

-- Steve

Quoted 24 lines

Then looking at the code I see this turn of events:
  if (test_bit(KTHREAD_IS_PER_CPU, &kthread->flags))
        __kthread_bind(k, kthread->cpu, TASK_PARKED);
Which in __kthread_bind() (state == TASK_PARKED)
	if (!wait_task_inactive(p, state)) {
		WARN_ON(1);
		return;
	}
Where wait_task_inactive() does:
		while (task_running(rq, p)) {
			if (match_state && unlikely(p->state != match_state))
				return 0;
As match_state is non zero and p->state != match_state because it hasn't been set yet. The wait_task_inactive() returns zero, and then we hit the WARN_ON() in __kthread_bind().
-- Steve

Subbaraman Narayanamurthy

author

2014-06-26 21:31 UTC #4 lore raw

In reply to Thomas Gleixner

Quoted 14 lines · Thomas Gleixner

The kthread park/unpark logic has the following issue:

Task CPU 0 CPU 1

T1 unplug cpu1

        kthread_park(T2)
        set_bit(KTHREAD_SHOULD_PARK);
	  wait_for_completion()
T2					parkme(X)
				   	  __set_current_state(TASK_PARKED);
				   	  while (test_bit(KTHREAD_SHOULD_PARK)) {
				     	    if (!test_and_set_bit(KTHREAD_IS_PARKED))
				              complete();
			             	    schedule();

Quoted 4 lines

T1 plug cpu1
--> premature wakeup of T2, i.e. before unpark, so T2 gets scheduled on
CPU 0

I understood the explanation above. But still I don't understand how this premature wakeup of T2 is happening/possible? Also, what will happen if the task state is not in TASK_PARKED when __kthread_unpark is called? __kthread_bind will fail silently causing the same problem.

Quoted 40 lines

Reorder the logic so that the unplug code binds the thread to the target cpu before clearing the KTHREAD_SHOULD_PARK bit.

Reported-by Subbaraman Narayanamurthy<[email protected]>

Signed-off-by Thomas Gleixner<[email protected]>

Cc [email protected]

---
  kernel/kthread.c |   14 ++++++++++----
  1 file changed, 10 insertions(+), 4 deletions(-)

Index: linux/kernel/kthread.c
===================================================================

kernel/kthread.c +9 -0

@@ -382,6 +382,15 @@ struct task_struct *kthread_create_on_cp
 
  static void __kthread_unpark(struct task_struct *k, struct kthread *kthread)
  {
+	/*
+	 * Rebind the thread to the target cpu first if it is a per
+	 * cpu thread unconditionally because it must be bound to the
+	 * target cpu before it can observe the KTHREAD_SHOULD_PARK
+	 * bit cleared.
+	 */
+	if (test_bit(KTHREAD_IS_PER_CPU, &kthread->flags))
+		__kthread_bind(k, kthread->cpu, TASK_PARKED);
+
  	clear_bit(KTHREAD_SHOULD_PARK, &kthread->flags);
  	/*
  	 * We clear the IS_PARKED bit here as we don't wait

@@ -389,11 +398,8 @@ static void __kthread_unpark(struct task
  	 * park before that happens we'd see the IS_PARKED bit
  	 * which might be about to be cleared.

kernel/kthread.c static void __kthread_unpark(struct task · line 405

Quoted 10 lines

kernel/kthread.c +1 -4

  	 */
-	if (test_and_clear_bit(KTHREAD_IS_PARKED, &kthread->flags)) {
-		if (test_bit(KTHREAD_IS_PER_CPU, &kthread->flags))
-			__kthread_bind(k, kthread->cpu, TASK_PARKED);
+	if (test_and_clear_bit(KTHREAD_IS_PARKED, &kthread->flags))
  		wake_up_state(k, TASK_PARKED);
-	}
  }
 
  /**

Thanks for the patch. I've tested (running hotplug tests) it for sometime and looks good so far. Can you please submit it?

Signature · 2 lines

The Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
hosted by The Linux Foundation

Thomas Gleixner

2014-06-26 23:50 UTC #5 lore raw

Quoted 19 lines · Subbaraman Narayanamurthy

Quoted 18 lines · Thomas Gleixner

The kthread park/unpark logic has the following issue:

Task CPU 0 CPU 1

T1 unplug cpu1

        kthread_park(T2)
        set_bit(KTHREAD_SHOULD_PARK);
	  wait_for_completion()
T2					parkme(X)
				   	  __set_current_state(TASK_PARKED);
				   	  while
(test_bit(KTHREAD_SHOULD_PARK)) {
				     	    if
(!test_and_set_bit(KTHREAD_IS_PARKED))
				              complete();
			             	    schedule();

T1 plug cpu1

Quoted 5 lines

Quoted 2 lines
--> premature wakeup of T2, i.e. before unpark, so T2 gets scheduled on
CPU 0
I understood the explanation above. But still I don't understand how this premature wakeup of T2 is happening/possible?

Come on. You have a machine which reproduces the issue. So some moderate tracing should tell you that ...

Without using my lost crystal ball, I bet that it's a premature per cpu timer interrupt.

Quoted 3 lines

Also, what will happen if the task state is not in TASK_PARKED when __kthread_unpark is called? __kthread_bind will fail silently causing the same problem.

Right you are, but thinking more about it:

Nothing is supposed to wakeup a parked thread except the unpark machinery. So the real question is: What causes the premature wakeup?

Darn, I should have thought about that before, but you tricked my overloaded brain into believing that this is a real issue.

No, it's not.

The parked state is not any different from creating a new kthread, advertise the thread to possible wakers and then do the bind.

So yes, the code is fine and the BUG_ON() is rightfully asserting here.

Quoted 2 lines

Thanks for the patch. I've tested (running hotplug tests) it for sometime and looks good so far. Can you please submit it?

So you have a legitimate question about the correctness of the patch and then you ask me to apply it?

Again, we do not apply patches which "fix" an issue just because we do not observe it anymore. We apply them when the problem at hand is fully understood and the solution solves all aspects.

Thanks,

tglx