[stable 4.4: PATCH] sched/cputime: Fix prev steal time accouting during CPU hotplug

[stable 4.4: PATCH] sched/cputime: Fix prev steal time accouting during CPU hotplug

[Andres Oportus]

commit 3d89e5478bf550a50c99e93adf659369798263b0 upstream.

Commit:

  e9532e69b8d1 ("sched/cputime: Fix steal time accounting vs. CPU hotplug")

... set rq->prev_* to 0 after a CPU hotplug comes back, in order to
fix the case where (after CPU hotplug) steal time is smaller than
rq->prev_steal_time.

However, this should never happen. Steal time was only smaller because of the
KVM-specific bug fixed by the previous patch.  Worse, the previous patch
triggers a bug on CPU hot-unplug/plug operation: because
rq->prev_steal_time is cleared, all of the CPU's past steal time will be
accounted again on hot-plug.

Since the root cause has been fixed, we can just revert
commit e9532e69b8d1 ("sched/cputime: Fix steal time accounting vs. CPU hotplug").

Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Fixes: 'commit e9532e69b8d1 ("sched/cputime: Fix steal time accounting vs. CPU hotplug")'
Link: http://lkml.kernel.org/r/1465813966-3116-3-git-send-email-wanpeng.li@hotmail.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andres Oportus <andresoportus@google.com>
---
 kernel/sched/core.c  |  1 -
 kernel/sched/sched.h | 13 -------------
 2 files changed, 14 deletions(-)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index c436426..b103b9c 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -5553,7 +5553,6 @@ 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 4e5db65..55d92a1 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1770,16 +1770,3 @@ 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
-}
-- 
1.9.1

Re: [stable 4.4: PATCH] sched/cputime: Fix prev steal time accouting during CPU hotplug

[Paolo Bonzini]

On 02/08/2017 18:36, Andres Oportus wrote:
> +Paolo Bonzini
> 
> On Mon, Jul 31, 2017 at 9:18 PM, Andres Oportus
> <andresoportus@android.com <mailto:andresoportus@android.com>> wrote:
> 
>     commit 3d89e5478bf550a50c99e93adf659369798263b0 upstream.

Not an easy call because this is a guest patch that depends on a host
patch.  But since it is a corner case (CPU hotplug) and it was
backported to 4.7 already,

Acked-by: Paolo Bonzini <pbonzini@redhat.com>

Paolo

>     Commit:
> 
>       e9532e69b8d1 ("sched/cputime: Fix steal time accounting vs. CPU
>     hotplug")
> 
>     ... set rq->prev_* to 0 after a CPU hotplug comes back, in order to
>     fix the case where (after CPU hotplug) steal time is smaller than
>     rq->prev_steal_time.
> 
>     However, this should never happen. Steal time was only smaller
>     because of the
>     KVM-specific bug fixed by the previous patch.  Worse, the previous patch
>     triggers a bug on CPU hot-unplug/plug operation: because
>     rq->prev_steal_time is cleared, all of the CPU's past steal time will be
>     accounted again on hot-plug.
> 
>     Since the root cause has been fixed, we can just revert
>     commit e9532e69b8d1 ("sched/cputime: Fix steal time accounting vs.
>     CPU hotplug").
> 
>     Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com
>     <mailto:wanpeng.li@hotmail.com>>
>     Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org
>     <mailto:peterz@infradead.org>>
>     Acked-by: Paolo Bonzini <pbonzini@redhat.com
>     <mailto:pbonzini@redhat.com>>
>     Cc: Frederic Weisbecker <fweisbec@gmail.com <mailto:fweisbec@gmail.com>>
>     Cc: Linus Torvalds <torvalds@linux-foundation.org
>     <mailto:torvalds@linux-foundation.org>>
>     Cc: Mike Galbraith <efault@gmx.de <mailto:efault@gmx.de>>
>     Cc: Peter Zijlstra <peterz@infradead.org <mailto:peterz@infradead.org>>
>     Cc: Radim Krčmář <rkrcmar@redhat.com <mailto:rkrcmar@redhat.com>>
>     Cc: Rik van Riel <riel@redhat.com <mailto:riel@redhat.com>>
>     Cc: Thomas Gleixner <tglx@linutronix.de <mailto:tglx@linutronix.de>>
>     Fixes: 'commit e9532e69b8d1 ("sched/cputime: Fix steal time
>     accounting vs. CPU hotplug")'
>     Link:
>     http://lkml.kernel.org/r/1465813966-3116-3-git-send-email-wanpeng.li@hotmail.com
>     <http://lkml.kernel.org/r/1465813966-3116-3-git-send-email-wanpeng.li@hotmail.com>
>     Signed-off-by: Ingo Molnar <mingo@kernel.org <mailto:mingo@kernel.org>>
>     Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org
>     <mailto:gregkh@linuxfoundation.org>>
>     Signed-off-by: Andres Oportus <andresoportus@google.com
>     <mailto:andresoportus@google.com>>
>     ---
>      kernel/sched/core.c  |  1 -
>      kernel/sched/sched.h | 13 -------------
>      2 files changed, 14 deletions(-)
> 
>     diff --git a/kernel/sched/core.c b/kernel/sched/core.c
>     index c436426..b103b9c 100644
>     --- a/kernel/sched/core.c
>     +++ b/kernel/sched/core.c
>     @@ -5553,7 +5553,6 @@ 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 4e5db65..55d92a1 100644
>     --- a/kernel/sched/sched.h
>     +++ b/kernel/sched/sched.h
>     @@ -1770,16 +1770,3 @@ 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
>     -}
>     --
>     1.9.1
> 
> 

Re: [stable 4.4: PATCH] sched/cputime: Fix prev steal time accouting during CPU hotplug

[Greg Kroah-Hartman]

On Wed, Aug 02, 2017 at 06:46:46PM +0200, Paolo Bonzini wrote:
> On 02/08/2017 18:36, Andres Oportus wrote:
> > +Paolo Bonzini
> > 
> > On Mon, Jul 31, 2017 at 9:18 PM, Andres Oportus
> > <andresoportus@android.com <mailto:andresoportus@android.com>> wrote:
> > 
> >     commit 3d89e5478bf550a50c99e93adf659369798263b0 upstream.
> 
> Not an easy call because this is a guest patch that depends on a host
> patch.  But since it is a corner case (CPU hotplug) and it was
> backported to 4.7 already,
> 
> Acked-by: Paolo Bonzini <pbonzini@redhat.com>

Thanks for the review, now queued up.

greg k-h