[RESEND] fix WARNING: at kernel/cpu/idle.c:96

Commit Message

James Bottomley June 14, 2013, 6:55 p.m. UTC

From 48bbf44a96676ce6f520a408378730c976e9a11e Mon Sep 17 00:00:00 2001
From: James Bottomley <[e-mail]>
Date: Wed, 8 May 2013 14:05:34 -0700
Subject: [PATCH] [PARISC] fix WARNING: at kernel/cpu/idle.c:96

On PA-RISC (and presumably any other arch that doesn't implement its own
arch_cpu_idle), we get this spurious boot warning.  The problem is that the
way the idle task is selected initially using the weak arch_cpu_idle() in
idle.c causes us to enter this place once with interrupts enabled.  Fix this
by disabling interrupts in the weak arch_cpu_idle() code.

Reviewed-by: Srivatsa S. Bhat <[e-mail]>
Cc: [e-mail]
Signed-off-by: James Bottomley <[e-mail]>

---

Thomas, I'm getting a bit impatient: this is a clear bug in the cpu idle
code and we keep getting reports of this as a boot crash on parisc.  If
you don't push it through your tree, I'll take it through the parisc
one.




--
To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Comments

David Daney June 14, 2013, 7:11 p.m. UTC | #1

On 06/14/2013 11:55 AM, James Bottomley wrote:
>>From 48bbf44a96676ce6f520a408378730c976e9a11e Mon Sep 17 00:00:00 2001
> From: James Bottomley <[e-mail]>
> Date: Wed, 8 May 2013 14:05:34 -0700
> Subject: [PATCH] [PARISC] fix WARNING: at kernel/cpu/idle.c:96
>
> On PA-RISC (and presumably any other arch that doesn't implement its own
> arch_cpu_idle), we get this spurious boot warning.  The problem is that the
> way the idle task is selected initially using the weak arch_cpu_idle() in
> idle.c causes us to enter this place once with interrupts enabled.  Fix this
> by disabling interrupts in the weak arch_cpu_idle() code.

Is this changelog correct?  It looks to me like you are enabling 
interrupts down there.

David Daney


>
> Reviewed-by: Srivatsa S. Bhat <[e-mail]>
> Cc: [e-mail]
> Signed-off-by: James Bottomley <[e-mail]>
>
> ---
>
> Thomas, I'm getting a bit impatient: this is a clear bug in the cpu idle
> code and we keep getting reports of this as a boot crash on parisc.  If
> you don't push it through your tree, I'll take it through the parisc
> one.
>
>
> diff --git a/kernel/cpu/idle.c b/kernel/cpu/idle.c
> index d5585f5..0a4d11e 100644
> --- a/kernel/cpu/idle.c
> +++ b/kernel/cpu/idle.c
> @@ -58,6 +58,7 @@ void __weak arch_cpu_idle_dead(void) { }
>   void __weak arch_cpu_idle(void)
>   {
>   	cpu_idle_force_poll = 1;
> +	local_irq_enable();

Here     ^^^^^^^^^^^^^^^^


>   }
>
>   /*
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [e-mail]
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>
>

--
To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

James Bottomley June 14, 2013, 7:13 p.m. UTC | #2

On Fri, 2013-06-14 at 12:11 -0700, David Daney wrote:
> On 06/14/2013 11:55 AM, James Bottomley wrote:
> >>From 48bbf44a96676ce6f520a408378730c976e9a11e Mon Sep 17 00:00:00 2001
> > From: James Bottomley <[e-mail]>
> > Date: Wed, 8 May 2013 14:05:34 -0700
> > Subject: [PATCH] [PARISC] fix WARNING: at kernel/cpu/idle.c:96
> >
> > On PA-RISC (and presumably any other arch that doesn't implement its own
> > arch_cpu_idle), we get this spurious boot warning.  The problem is that the
> > way the idle task is selected initially using the weak arch_cpu_idle() in
> > idle.c causes us to enter this place once with interrupts enabled.  Fix this
> > by disabling interrupts in the weak arch_cpu_idle() code.
> 
> Is this changelog correct?  It looks to me like you are enabling 
> interrupts down there.

Yes, obvious typo, sorry.  The WARN_ON check is for disabled interrupts
the fix is to enable them.

James


--
To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Thomas Gleixner June 14, 2013, 8:39 p.m. UTC | #3

On Fri, 14 Jun 2013, James Bottomley wrote:

> >From 48bbf44a96676ce6f520a408378730c976e9a11e Mon Sep 17 00:00:00 2001
> From: James Bottomley <[e-mail]>
> Date: Wed, 8 May 2013 14:05:34 -0700
> Subject: [PATCH] [PARISC] fix WARNING: at kernel/cpu/idle.c:96
> 
> On PA-RISC (and presumably any other arch that doesn't implement its own
> arch_cpu_idle), we get this spurious boot warning.  The problem is that the
> way the idle task is selected initially using the weak arch_cpu_idle() in
> idle.c causes us to enter this place once with interrupts enabled.  Fix this
> by disabling interrupts in the weak arch_cpu_idle() code.
> 
> Reviewed-by: Srivatsa S. Bhat <[e-mail]>
> Cc: [e-mail]

What's the stable tag for? This code got merged in 3,10, so stable is
totally irrelevant.

> Signed-off-by: James Bottomley <[e-mail]>
> 
> ---
> 
> Thomas, I'm getting a bit impatient: this is a clear bug in the cpu idle
> code and we keep getting reports of this as a boot crash on parisc.  If
> you don't push it through your tree, I'll take it through the parisc
> one.

Hold your breath. I was not even CC'ed on the original patch and I
admit that I ignored the patch which starts with [PARISC].

If the subject line would have started with [idle], [core/idle] I
definitely would have paid attention.

Aside of that the rest of the subject line is just annoyingly
sloppy. We do not fix a WARNING. That's not what this patch is
about. The patch fixes a problem which got introduced with the idle
rework, period.

I'll pick it up and fix the changelog.

Thanks,

	tglx

> 
> diff --git a/kernel/cpu/idle.c b/kernel/cpu/idle.c
> index d5585f5..0a4d11e 100644
> --- a/kernel/cpu/idle.c
> +++ b/kernel/cpu/idle.c
> @@ -58,6 +58,7 @@ void __weak arch_cpu_idle_dead(void) { }
>  void __weak arch_cpu_idle(void)
>  {
>  	cpu_idle_force_poll = 1;
> +	local_irq_enable();
>  }
>  
>  /*
> 
> 
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Thomas Gleixner June 14, 2013, 8:49 p.m. UTC | #4

On Fri, 14 Jun 2013, Thomas Gleixner wrote:
> On Fri, 14 Jun 2013, James Bottomley wrote:
> 
> > >From 48bbf44a96676ce6f520a408378730c976e9a11e Mon Sep 17 00:00:00 2001
> > From: James Bottomley <[e-mail]>
> > Date: Wed, 8 May 2013 14:05:34 -0700
> > Subject: [PATCH] [PARISC] fix WARNING: at kernel/cpu/idle.c:96
> > 
> > On PA-RISC (and presumably any other arch that doesn't implement its own
> > arch_cpu_idle), we get this spurious boot warning.  The problem is that the
> > way the idle task is selected initially using the weak arch_cpu_idle() in
> > idle.c causes us to enter this place once with interrupts enabled.  Fix this
> > by disabling interrupts in the weak arch_cpu_idle() code.
> > 
> > Reviewed-by: Srivatsa S. Bhat <[e-mail]>
> > Cc: [e-mail]
> 
> What's the stable tag for? This code got merged in 3,10, so stable is
> totally irrelevant.
> 
> > Signed-off-by: James Bottomley <[e-mail]>
> > 
> > ---
> > 
> > Thomas, I'm getting a bit impatient: this is a clear bug in the cpu idle
> > code and we keep getting reports of this as a boot crash on parisc.  If
> > you don't push it through your tree, I'll take it through the parisc
> > one.
> 
> Hold your breath. I was not even CC'ed on the original patch and I
> admit that I ignored the patch which starts with [PARISC].
> 
> If the subject line would have started with [idle], [core/idle] I
> definitely would have paid attention.
> 
> Aside of that the rest of the subject line is just annoyingly
> sloppy. We do not fix a WARNING. That's not what this patch is
> about. The patch fixes a problem which got introduced with the idle
> rework, period.
> 
> I'll pick it up and fix the changelog.

And it needs fixing. It says:

"... way the idle task is selected initially using the weak
 arch_cpu_idle() in idle.c causes us to enter this place once with
 interrupts enabled.  Fix this by disabling interrupts in the weak
 arch_cpu_idle() code."

And the patch does:

 void __weak arch_cpu_idle(void)
 {
        cpu_idle_force_poll = 1;
+       local_irq_enable();
 }

Instead of bullying around you might consider to read
Documentation/SubmittingPatches.

Thanks,

	tglx
--
To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

James Bottomley June 14, 2013, 9:50 p.m. UTC | #5

On Fri, 2013-06-14 at 22:39 +0200, Thomas Gleixner wrote:
> On Fri, 14 Jun 2013, James Bottomley wrote:
> 
> > >From 48bbf44a96676ce6f520a408378730c976e9a11e Mon Sep 17 00:00:00 2001
> > From: James Bottomley <[e-mail]>
> > Date: Wed, 8 May 2013 14:05:34 -0700
> > Subject: [PATCH] [PARISC] fix WARNING: at kernel/cpu/idle.c:96
> > 
> > On PA-RISC (and presumably any other arch that doesn't implement its own
> > arch_cpu_idle), we get this spurious boot warning.  The problem is that the
> > way the idle task is selected initially using the weak arch_cpu_idle() in
> > idle.c causes us to enter this place once with interrupts enabled.  Fix this
> > by disabling interrupts in the weak arch_cpu_idle() code.
> > 
> > Reviewed-by: Srivatsa S. Bhat <[e-mail]>
> > Cc: [e-mail]
> 
> What's the stable tag for? This code got merged in 3,10, so stable is
> totally irrelevant.

Hm, OK, it's been so long I'm misremembering which kernel versions need
it.

> > Signed-off-by: James Bottomley <[e-mail]>
> > 
> > ---
> > 
> > Thomas, I'm getting a bit impatient: this is a clear bug in the cpu idle
> > code and we keep getting reports of this as a boot crash on parisc.  If
> > you don't push it through your tree, I'll take it through the parisc
> > one.
> 
> Hold your breath. I was not even CC'ed on the original patch and I
> admit that I ignored the patch which starts with [PARISC].

Oh, you were ... I made sure of that. It's thread with subject

Re: [PATCH] parisc: avoid WARNING: at kernel/cpu/idle.c:96

You were cc'd from the one dated Wed, 08 May 2013 14:05:34 -0700

> If the subject line would have started with [idle], [core/idle] I
> definitely would have paid attention.
> 
> Aside of that the rest of the subject line is just annoyingly
> sloppy. We do not fix a WARNING. That's not what this patch is
> about. The patch fixes a problem which got introduced with the idle
> rework, period.
> 
> I'll pick it up and fix the changelog.

Sure, whatever you think is best ... Given Linus' current mood I think
leading with a description of the actual user visible problem being
fixed is a good way to make sure he doesn't get annoyed, but it's your
call.

James



--
To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/kernel/cpu/idle.c b/kernel/cpu/idle.c
index d5585f5..0a4d11e 100644
--- a/kernel/cpu/idle.c
+++ b/kernel/cpu/idle.c
@@ -58,6 +58,7 @@  void __weak arch_cpu_idle_dead(void) { }
 void __weak arch_cpu_idle(void)
 {
 	cpu_idle_force_poll = 1;
+	local_irq_enable();
 }
 
 /*