[rfc] disable scheduler warnings during oopses.

[rfc] disable scheduler warnings during oopses.

[Dave Jones]

The panic-on-framebuffer code seems to cause a schedule
to occur during an oops. This causes a bunch of extra
spew as can be seen in https://bugzilla.redhat.com/attachment.cgi?id=549230

Would this (untested) patch be sufficient here, or does the drm
code need rearchitecting to not perform allocations during panic ?

Signed-off-by: Dave Jones <davej@redhat.com>

diff --git a/kernel/sched.c b/kernel/sched.c
index d6b149c..627f788 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -4321,6 +4321,9 @@ static noinline void __schedule_bug(struct task_struct *prev)
 {
 	struct pt_regs *regs = get_irq_regs();
 
+	if (oops_in_progress)
+		return;
+
 	printk(KERN_ERR "BUG: scheduling while atomic: %s/%d/0x%08x\n",
 		prev->comm, prev->pid, preempt_count());
 

Re: [rfc] disable scheduler warnings during oopses.

[Ingo Molnar]

* Dave Jones <davej@redhat.com> wrote:

> The panic-on-framebuffer code seems to cause a schedule
> to occur during an oops. This causes a bunch of extra
> spew as can be seen in https://bugzilla.redhat.com/attachment.cgi?id=549230
> 
> Would this (untested) patch be sufficient here, or does the drm
> code need rearchitecting to not perform allocations during panic ?
> 
> Signed-off-by: Dave Jones <davej@redhat.com>
> 
> diff --git a/kernel/sched.c b/kernel/sched.c
> index d6b149c..627f788 100644
> --- a/kernel/sched.c
> +++ b/kernel/sched.c
> @@ -4321,6 +4321,9 @@ static noinline void __schedule_bug(struct task_struct *prev)
>  {
>  	struct pt_regs *regs = get_irq_regs();
>  
> +	if (oops_in_progress)
> +		return;
> +
>  	printk(KERN_ERR "BUG: scheduling while atomic: %s/%d/0x%08x\n",
>  		prev->comm, prev->pid, preempt_count());
>  

Doing that makes sense - if an oops in going on we don't want to 
produce recursive messages. Note that scheduler code has been 
moved to kernel/sched/, so i've changed the patch accordingly.

Note that despite that, the DRM code *definitely* needs to be 
fixed to not allocate and schedule while printing on the 
framebuffer in KMS mode ...

Thanks,

	Ingo

Re: [rfc] disable scheduler warnings during oopses.

[Peter Zijlstra]

On Fri, 2011-12-23 at 11:19 +0100, Ingo Molnar wrote:
> * Dave Jones <davej@redhat.com> wrote:
> 
> > The panic-on-framebuffer code seems to cause a schedule
> > to occur during an oops. This causes a bunch of extra
> > spew as can be seen in https://bugzilla.redhat.com/attachment.cgi?id=549230
> > 
> > Would this (untested) patch be sufficient here, or does the drm
> > code need rearchitecting to not perform allocations during panic ?
> > 
> > Signed-off-by: Dave Jones <davej@redhat.com>
> > 
> > diff --git a/kernel/sched.c b/kernel/sched.c
> > index d6b149c..627f788 100644
> > --- a/kernel/sched.c
> > +++ b/kernel/sched.c
> > @@ -4321,6 +4321,9 @@ static noinline void __schedule_bug(struct task_struct *prev)
> >  {
> >  	struct pt_regs *regs = get_irq_regs();
> >  
> > +	if (oops_in_progress)
> > +		return;
> > +
> >  	printk(KERN_ERR "BUG: scheduling while atomic: %s/%d/0x%08x\n",
> >  		prev->comm, prev->pid, preempt_count());
> >  
> 
> Doing that makes sense - if an oops in going on we don't want to 
> produce recursive messages. Note that scheduler code has been 
> moved to kernel/sched/, so i've changed the patch accordingly.

Urgh I would actually argue against merging this. What the KSM console
does is horridly wrong, and simply papering over it doesn't help its
cause. Its up there with USB-Serial console on wrongness.

> Note that despite that, the DRM code *definitely* needs to be 
> fixed to not allocate and schedule while printing on the 
> framebuffer in KMS mode ...

For sure, its plain insane and completely unreliable.

[tip:sched/core] sched: Disable scheduler warnings during oopses

[tip-bot for Dave Jones]

Commit-ID:  664dfa65e84429d0b68694483e1de7365c7c56fb
Gitweb:     http://git.kernel.org/tip/664dfa65e84429d0b68694483e1de7365c7c56fb
Author:     Dave Jones <davej@redhat.com>
AuthorDate: Thu, 22 Dec 2011 16:39:30 -0500
Committer:  Ingo Molnar <mingo@elte.hu>
CommitDate: Fri, 23 Dec 2011 11:20:50 +0100

sched: Disable scheduler warnings during oopses

The panic-on-framebuffer code seems to cause a schedule
to occur during an oops. This causes a bunch of extra
spew as can be seen in:

   https://bugzilla.redhat.com/attachment.cgi?id=549230

Don't do scheduler debug checks when we are oopsing already.

Signed-off-by: Dave Jones <davej@redhat.com>
Link: http://lkml.kernel.org/r/20111222213929.GA4722@redhat.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
---
 kernel/sched/core.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 081ece2..8ffe523 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -3077,6 +3077,9 @@ static noinline void __schedule_bug(struct task_struct *prev)
 {
 	struct pt_regs *regs = get_irq_regs();
 
+	if (oops_in_progress)
+		return;
+
 	printk(KERN_ERR "BUG: scheduling while atomic: %s/%d/0x%08x\n",
 		prev->comm, prev->pid, preempt_count());
 

Re: [rfc] disable scheduler warnings during oopses.

[Ingo Molnar]

* Peter Zijlstra <a.p.zijlstra@chello.nl> wrote:

> On Fri, 2011-12-23 at 11:19 +0100, Ingo Molnar wrote:
> > * Dave Jones <davej@redhat.com> wrote:
> > 
> > > The panic-on-framebuffer code seems to cause a schedule
> > > to occur during an oops. This causes a bunch of extra
> > > spew as can be seen in https://bugzilla.redhat.com/attachment.cgi?id=549230
> > > 
> > > Would this (untested) patch be sufficient here, or does the drm
> > > code need rearchitecting to not perform allocations during panic ?
> > > 
> > > Signed-off-by: Dave Jones <davej@redhat.com>
> > > 
> > > diff --git a/kernel/sched.c b/kernel/sched.c
> > > index d6b149c..627f788 100644
> > > --- a/kernel/sched.c
> > > +++ b/kernel/sched.c
> > > @@ -4321,6 +4321,9 @@ static noinline void __schedule_bug(struct task_struct *prev)
> > >  {
> > >  	struct pt_regs *regs = get_irq_regs();
> > >  
> > > +	if (oops_in_progress)
> > > +		return;
> > > +
> > >  	printk(KERN_ERR "BUG: scheduling while atomic: %s/%d/0x%08x\n",
> > >  		prev->comm, prev->pid, preempt_count());
> > >  
> > 
> > Doing that makes sense - if an oops in going on we don't 
> > want to produce recursive messages. Note that scheduler code 
> > has been moved to kernel/sched/, so i've changed the patch 
> > accordingly.
> 
> Urgh I would actually argue against merging this. What the KSM 
> console does is horridly wrong, and simply papering over it 
> doesn't help its cause. Its up there with USB-Serial console 
> on wrongness.

Well, i agree in principle, but not escallating an already bad 
scenario would be a basic robustness rule. We do skip certain 
things on oops_in_progress already.

Linus, Andrew, what would be your preference?

Thanks,

	Ingo