[PATCH] sched/cputime: Fix steal time accounting vs. cpu hotplug

[PATCH] sched/cputime: Fix steal time accounting vs. cpu hotplug

[Thomas Gleixner]

On cpu hotplug the steal time accounting can keep a stale rq->prev_steal_time
value over cpu down and up. So after the cpu comes up again the delta
calculation in steal_account_process_tick() wreckages itself due to the
unsigned math:

	 u64 steal = paravirt_steal_clock(smp_processor_id());
	 
	 steal -= this_rq()->prev_steal_time;

So if steal is smaller than rq->prev_steal_time we end up with an insane large
value which then gets added to rq->prev_steal_time, resulting in a permanent
wreckage of the accounting. As a consequence the per cpu stats in /proc/stat
become stale.

Nice trick to tell the world how idle the system is (100%) while the cpu is
100% busy running tasks. Though we prefer realistic numbers.

None of the accounting values which use a previous value to account for
fractions is reset at cpu hotplug time. update_rq_clock_task() has a sanity
check for prev_irq_time and prev_steal_time_rq, but that sanity check solely
deals with clock warps and limits the /proc/stat visible wreckage. The
prev_time values are still wrong.

Solution is simple: Reset rq->prev_*_time when the cpu is plugged in again.

Fixes: commit e6e6685accfa "KVM guest: Steal time accounting"
Fixes: commit 095c0aa83e52 "sched: adjust scheduler cpu power for stolen time"
Fixes: commit aa483808516c "sched: Remove irq time from available CPU power"
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org

---
 kernel/sched/core.c  |    1 +
 kernel/sched/sched.h |   13 +++++++++++++
 2 files changed, 14 insertions(+)

Index: b/kernel/sched/core.c
===================================================================
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -5627,6 +5627,7 @@ migration_call(struct notifier_block *nf
 
 	case CPU_UP_PREPARE:
 		rq->calc_load_update = calc_load_update;
+		account_reset_rq(rq);
 		break;
 
 	case CPU_ONLINE:
Index: b/kernel/sched/sched.h
===================================================================
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1738,3 +1738,16 @@ static inline u64 irq_time_read(int cpu)
 }
 #endif /* CONFIG_64BIT */
 #endif /* CONFIG_IRQ_TIME_ACCOUNTING */
+
+static inline void account_reset_rq(struct rq *rq)
+{
+#ifdef CONFIG_IRQ_TIME_ACCOUNTING
+	rq->prev_irq_time = 0;
+#endif
+#ifdef CONFIG_PARAVIRT
+	rq->prev_steal_time = 0;
+#endif
+#ifdef CONFIG_PARAVIRT_TIME_ACCOUNTING
+	rq->prev_steal_time_rq = 0;
+#endif
+}

Re: [PATCH] sched/cputime: Fix steal time accounting vs. cpu hotplug

[Rik van Riel]

[-- Attachment #1: Type: text/plain, Size: 1736 bytes --]

On Fri, 2016-03-04 at 15:59 +0100, Thomas Gleixner wrote:
> On cpu hotplug the steal time accounting can keep a stale rq-
> >prev_steal_time
> value over cpu down and up. So after the cpu comes up again the delta
> calculation in steal_account_process_tick() wreckages itself due to
> the
> unsigned math:
> 
>          u64 steal = paravirt_steal_clock(smp_processor_id());
>          
>          steal -= this_rq()->prev_steal_time;
> 
> So if steal is smaller than rq->prev_steal_time we end up with an
> insane large
> value which then gets added to rq->prev_steal_time, resulting in a
> permanent
> wreckage of the accounting. As a consequence the per cpu stats in
> /proc/stat
> become stale.
> 
> Nice trick to tell the world how idle the system is (100%) while the
> cpu is
> 100% busy running tasks. Though we prefer realistic numbers.
> 
> None of the accounting values which use a previous value to account
> for
> fractions is reset at cpu hotplug time. update_rq_clock_task() has a
> sanity
> check for prev_irq_time and prev_steal_time_rq, but that sanity check
> solely
> deals with clock warps and limits the /proc/stat visible wreckage.
> The
> prev_time values are still wrong.
> 
> Solution is simple: Reset rq->prev_*_time when the cpu is plugged in
> again.
> 
> Fixes: commit e6e6685accfa "KVM guest: Steal time accounting"
> Fixes: commit 095c0aa83e52 "sched: adjust scheduler cpu power for
> stolen time"
> Fixes: commit aa483808516c "sched: Remove irq time from available CPU
> power"
> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
> Cc: stable@vger.kernel.org

Acked-by: Rik van Riel <riel@redhat.com>

-- 
All Rights Reversed.


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

[tip:sched/urgent] sched/cputime: Fix steal time accounting vs. CPU hotplug

[tip-bot for Thomas Gleixner]

Commit-ID:  e9532e69b8d1d1284e8ecf8d2586de34aec61244
Gitweb:     http://git.kernel.org/tip/e9532e69b8d1d1284e8ecf8d2586de34aec61244
Author:     Thomas Gleixner <tglx@linutronix.de>
AuthorDate: Fri, 4 Mar 2016 15:59:42 +0100
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Sat, 5 Mar 2016 09:17:20 +0100

sched/cputime: Fix steal time accounting vs. CPU hotplug

On CPU hotplug the steal time accounting can keep a stale rq->prev_steal_time
value over CPU down and up. So after the CPU comes up again the delta
calculation in steal_account_process_tick() wreckages itself due to the
unsigned math:

	 u64 steal = paravirt_steal_clock(smp_processor_id());

	 steal -= this_rq()->prev_steal_time;

So if steal is smaller than rq->prev_steal_time we end up with an insane large
value which then gets added to rq->prev_steal_time, resulting in a permanent
wreckage of the accounting. As a consequence the per CPU stats in /proc/stat
become stale.

Nice trick to tell the world how idle the system is (100%) while the CPU is
100% busy running tasks. Though we prefer realistic numbers.

None of the accounting values which use a previous value to account for
fractions is reset at CPU hotplug time. update_rq_clock_task() has a sanity
check for prev_irq_time and prev_steal_time_rq, but that sanity check solely
deals with clock warps and limits the /proc/stat visible wreckage. The
prev_time values are still wrong.

Solution is simple: Reset rq->prev_*_time when the CPU is plugged in again.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Rik van Riel <riel@redhat.com>
Cc: <stable@vger.kernel.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Glauber Costa <glommer@parallels.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Fixes: commit 095c0aa83e52 "sched: adjust scheduler cpu power for stolen time"
Fixes: commit aa483808516c "sched: Remove irq time from available CPU power"
Fixes: commit e6e6685accfa "KVM guest: Steal time accounting"
Link: http://lkml.kernel.org/r/alpine.DEB.2.11.1603041539490.3686@nanos
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 kernel/sched/core.c  |  1 +
 kernel/sched/sched.h | 13 +++++++++++++
 2 files changed, 14 insertions(+)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index ab814bf..406182a 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -5627,6 +5627,7 @@ migration_call(struct notifier_block *nfb, unsigned long action, void *hcpu)
 
 	case CPU_UP_PREPARE:
 		rq->calc_load_update = calc_load_update;
+		account_reset_rq(rq);
 		break;
 
 	case CPU_ONLINE:
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 30ea2d8..4f6598a 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1738,3 +1738,16 @@ static inline u64 irq_time_read(int cpu)
 }
 #endif /* CONFIG_64BIT */
 #endif /* CONFIG_IRQ_TIME_ACCOUNTING */
+
+static inline void account_reset_rq(struct rq *rq)
+{
+#ifdef CONFIG_IRQ_TIME_ACCOUNTING
+	rq->prev_irq_time = 0;
+#endif
+#ifdef CONFIG_PARAVIRT
+	rq->prev_steal_time = 0;
+#endif
+#ifdef CONFIG_PARAVIRT_TIME_ACCOUNTING
+	rq->prev_steal_time_rq = 0;
+#endif
+}