Prev: [PATCH] Fix CONFIG_CROSS_COMPILE issue in .config
Next: Slabinfo: Fix display format
From: Jesper Nilsson on 17 Jun 2010 09:50
Breaking here dropped us to the default code which always sends
a SIGILL to the current process, no matter what the CU2 notifier says.

Signed-off-by: Jesper Nilsson <jesper(a)jni.nu>
---
traps.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/mips/kernel/traps.c b/arch/mips/kernel/traps.c
index 8bdd6a6..8527808 100644
--- a/arch/mips/kernel/traps.c
+++ b/arch/mips/kernel/traps.c
@@ -976,7 +976,7 @@ asmlinkage void do_cpu(struct pt_regs *regs)

case 2:
raw_notifier_call_chain(&cu2_chain, CU2_EXCEPTION, regs);
- break;
+ return;

case 3:
break;

/^JN - Jesper Nilsson
--
Jesper Nilsson -- jesper_at_jni.nu
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: David Daney on 17 Jun 2010 13:20
On 06/17/2010 06:25 AM, Jesper Nilsson wrote:
> Breaking here dropped us to the default code which always sends
> a SIGILL to the current process, no matter what the CU2 notifier says.
>
> Signed-off-by: Jesper Nilsson<jesper(a)jni.nu>
> ---
> traps.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/mips/kernel/traps.c b/arch/mips/kernel/traps.c
> index 8bdd6a6..8527808 100644
> --- a/arch/mips/kernel/traps.c
> +++ b/arch/mips/kernel/traps.c
> @@ -976,7 +976,7 @@ asmlinkage void do_cpu(struct pt_regs *regs)
>
> case 2:
> raw_notifier_call_chain(&cu2_chain, CU2_EXCEPTION, regs);
> - break;
> + return;
>

What happens when the call chain is empty, and the proper action *is*
SIGILL?

David Daney
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Jesper Nilsson on 17 Jun 2010 15:10
On Thu, Jun 17, 2010 at 10:13:18AM -0700, David Daney wrote:
> On 06/17/2010 06:25 AM, Jesper Nilsson wrote:
> >Breaking here dropped us to the default code which always sends
> >a SIGILL to the current process, no matter what the CU2 notifier says.
> >
> >Signed-off-by: Jesper Nilsson<jesper(a)jni.nu>
> >---
> > traps.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> >diff --git a/arch/mips/kernel/traps.c b/arch/mips/kernel/traps.c
> >index 8bdd6a6..8527808 100644
> >--- a/arch/mips/kernel/traps.c
> >+++ b/arch/mips/kernel/traps.c
> >@@ -976,7 +976,7 @@ asmlinkage void do_cpu(struct pt_regs *regs)
> >
> > case 2:
> > raw_notifier_call_chain(&cu2_chain, CU2_EXCEPTION, regs);
> >- break;
> >+ return;
> >
>
> What happens when the call chain is empty, and the proper action *is*
> SIGILL?

Well, since there is a default notifier installed at the end, it will
correctly return SIGILL.

See the definition of default_cu2_call in the same file.

> David Daney

/^JN - Jesper Nilsson
--
Jesper Nilsson -- jesper_at_jni.nu
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Gleb O. Raiko on 18 Jun 2010 04:30
On 17.06.2010 21:13, David Daney wrote:
> On 06/17/2010 06:25 AM, Jesper Nilsson wrote:
>> Breaking here dropped us to the default code which always sends
>> a SIGILL to the current process, no matter what the CU2 notifier says.
>>
>> Signed-off-by: Jesper Nilsson<jesper(a)jni.nu>
[...]
>> case 2:
>> raw_notifier_call_chain(&cu2_chain, CU2_EXCEPTION, regs);
>> - break;
>> + return;
>>
>
> What happens when the call chain is empty, and the proper action *is*
> SIGILL?

It's never empty, in fact. The default notifier declared at top of
traps.c sends SIGILL. The problem that current code is sending SIGILL in
all cases.

Gleb.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Ralf Baechle on 18 Jun 2010 06:10
On Fri, Jun 18, 2010 at 11:54:38AM +0400, Gleb O. Raiko wrote:

> >What happens when the call chain is empty, and the proper action *is*
> >SIGILL?
>
> It's never empty, in fact. The default notifier declared at top of
> traps.c sends SIGILL. The problem that current code is sending
> SIGILL in all cases.

That's not really a problem. The design idea is that a the default
notifier has the lowest priority, that is any user notifier installed
should have higher priority resulting in it getting run first. To avoid
the default notifier from getting executed such an extra notifier should
set NOTIFY_STOP_MASK in its return like:

static int default_cu2_call(struct notifier_block *nfb, unsigned long action,
void *data)
{
...

return NOTIFY_OK | NOTIFY_STOP;
}

The notifier list could also be used for example by perf but there it
we'd want the notifier function not to return NOTIFY_STOP as the result;

Arguably sending the signal for CU2 instructions has been delegated to the
hook so the I agree that the break stateement should be replaced with a
return and will apply the patch.

Ralf
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
 |  Next  |  Last
Pages: 1 2
Prev: [PATCH] Fix CONFIG_CROSS_COMPILE issue in .config
Next: Slabinfo: Fix display format