[PATCH] ARM: cns3xxx: fix allmodconfig panic in pci driver

[PATCH] ARM: cns3xxx: fix allmodconfig panic in pci driver

[Xia Kaixu]

From: Xia Kaixu <kaixu.xia@linaro.org>

The kernel panic occurs when running an allmodconfig kernel on 
OMAP4460. The inicall "cns3xxx_pcie_init" does not check which
hardware it's running on and just tries to access to its specific 
registers. Now call it from .init_late callback from the two
machine descriptors.

Signed-off-by: Xia Kaixu <kaixu.xia@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Anton Vorontsov <anton@enomsg.org>
Cc: Felix Fietkau <nbd@openwrt.org>
Cc: Imre Kaloz <kaloz@openwrt.org>
Cc: linaro-kernel at lists.linaro.org
Cc: linux-arm-kernel at lists.infradead.org
---
 arch/arm/mach-cns3xxx/cns3420vb.c | 1 +
 arch/arm/mach-cns3xxx/core.c      | 1 +
 arch/arm/mach-cns3xxx/core.h      | 9 +++++++++
 arch/arm/mach-cns3xxx/pcie.c      | 3 +--
 4 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/arch/arm/mach-cns3xxx/cns3420vb.c b/arch/arm/mach-cns3xxx/cns3420vb.c
index d863d87..99c74cc 100644
--- a/arch/arm/mach-cns3xxx/cns3420vb.c
+++ b/arch/arm/mach-cns3xxx/cns3420vb.c
@@ -250,5 +250,6 @@ MACHINE_START(CNS3420VB, "Cavium Networks CNS3420 Validation Board")
 	.init_irq	= cns3xxx_init_irq,
 	.init_time	= cns3xxx_timer_init,
 	.init_machine	= cns3420_init,
+	.init_late      = cns3xxx_pcie_init_late,
 	.restart	= cns3xxx_restart,
 MACHINE_END
diff --git a/arch/arm/mach-cns3xxx/core.c b/arch/arm/mach-cns3xxx/core.c
index f85449a..c2363bb 100644
--- a/arch/arm/mach-cns3xxx/core.c
+++ b/arch/arm/mach-cns3xxx/core.c
@@ -404,5 +404,6 @@ DT_MACHINE_START(CNS3XXX_DT, "Cavium Networks CNS3xxx")
 	.init_irq	= cns3xxx_init_irq,
 	.init_time	= cns3xxx_timer_init,
 	.init_machine	= cns3xxx_init,
+	.init_late	= cns3xxx_pcie_init_late,
 	.restart	= cns3xxx_restart,
 MACHINE_END
diff --git a/arch/arm/mach-cns3xxx/core.h b/arch/arm/mach-cns3xxx/core.h
index 5218b61..780689b 100644
--- a/arch/arm/mach-cns3xxx/core.h
+++ b/arch/arm/mach-cns3xxx/core.h
@@ -21,6 +21,15 @@ void __init cns3xxx_l2x0_init(void);
 static inline void cns3xxx_l2x0_init(void) {}
 #endif /* CONFIG_CACHE_L2X0 */
 
+#ifdef CONFIG_PCI
+extern int __init cns3xxx_pcie_init_late(void);
+#else
+static inline int __init cns3xxx_pcie_init_late(void)
+{
+	return -ENXIO;
+}
+#endif
+
 void __init cns3xxx_map_io(void);
 void __init cns3xxx_init_irq(void);
 void cns3xxx_power_off(void);
diff --git a/arch/arm/mach-cns3xxx/pcie.c b/arch/arm/mach-cns3xxx/pcie.c
index 413134c..bf2a4e8 100644
--- a/arch/arm/mach-cns3xxx/pcie.c
+++ b/arch/arm/mach-cns3xxx/pcie.c
@@ -318,7 +318,7 @@ static int cns3xxx_pcie_abort_handler(unsigned long addr, unsigned int fsr,
 	return 0;
 }
 
-static int __init cns3xxx_pcie_init(void)
+int __init cns3xxx_pcie_init_late(void)
 {
 	int i;
 
@@ -340,4 +340,3 @@ static int __init cns3xxx_pcie_init(void)
 
 	return 0;
 }
-device_initcall(cns3xxx_pcie_init);
-- 
1.8.5.5

[PATCH] ARM: cns3xxx: fix allmodconfig panic in pci driver

[Arnd Bergmann]

On Wednesday 03 September 2014 21:18:12 Xia Kaixu wrote:
> From: Xia Kaixu <kaixu.xia@linaro.org>
> 
> The kernel panic occurs when running an allmodconfig kernel on 
> OMAP4460. The inicall "cns3xxx_pcie_init" does not check which
> hardware it's running on and just tries to access to its specific 
> registers. Now call it from .init_late callback from the two
> machine descriptors.
> 
> Signed-off-by: Xia Kaixu <kaixu.xia@linaro.org>
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> Cc: Anton Vorontsov <anton@enomsg.org>
> Cc: Felix Fietkau <nbd@openwrt.org>
> Cc: Imre Kaloz <kaloz@openwrt.org>
> Cc: linaro-kernel at lists.linaro.org
> Cc: linux-arm-kernel at lists.infradead.org

Hi Kaixu,

it seems this time all the Cc's worked. One thing I just noticed is that
you confused the 'Signed-off-by' lines. I did look at the patch earlier
before you sent it out but unfortunately missed that.

You can read again in Documentation/SubmittingPatches about how
they work, but the short version is that you must never put someone
else's name in Signed-off-by under a patch you write yourself.
When we apply the patch, I (or whoever else does) will put that
line below yours to document who handled the patch.

Otherwise, the patch looks good to me, thanks a lot for doing it.

	Arnd

[PATCH] ARM: cns3xxx: fix allmodconfig panic in pci driver

[xiakaixu]

? 2014/9/3 21:27, Arnd Bergmann ??:
> On Wednesday 03 September 2014 21:18:12 Xia Kaixu wrote:
>> From: Xia Kaixu <kaixu.xia@linaro.org>
>>
>> The kernel panic occurs when running an allmodconfig kernel on 
>> OMAP4460. The inicall "cns3xxx_pcie_init" does not check which
>> hardware it's running on and just tries to access to its specific 
>> registers. Now call it from .init_late callback from the two
>> machine descriptors.
>>
>> Signed-off-by: Xia Kaixu <kaixu.xia@linaro.org>
>> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
>> Cc: Anton Vorontsov <anton@enomsg.org>
>> Cc: Felix Fietkau <nbd@openwrt.org>
>> Cc: Imre Kaloz <kaloz@openwrt.org>
>> Cc: linaro-kernel at lists.linaro.org
>> Cc: linux-arm-kernel at lists.infradead.org
> 
> Hi Kaixu,
> 
> it seems this time all the Cc's worked. One thing I just noticed is that
> you confused the 'Signed-off-by' lines. I did look at the patch earlier
> before you sent it out but unfortunately missed that.
> 
> You can read again in Documentation/SubmittingPatches about how
> they work, but the short version is that you must never put someone
> else's name in Signed-off-by under a patch you write yourself.
> When we apply the patch, I (or whoever else does) will put that
> line below yours to document who handled the patch.

  Sure, you are right. Thanks for the reminder!
> 
> Otherwise, the patch looks good to me, thanks a lot for doing it.
> 
> 	Arnd
> 
> _______________________________________________
> linaro-kernel mailing list
> linaro-kernel at lists.linaro.org
> http://lists.linaro.org/mailman/listinfo/linaro-kernel
> 

[PATCH] ARM: cns3xxx: fix allmodconfig panic in pci driver

[Arnd Bergmann]

On Wednesday 03 September 2014, Xia Kaixu wrote:
> From: Xia Kaixu <kaixu.xia@linaro.org>
> 
> The kernel panic occurs when running an allmodconfig kernel on 
> OMAP4460. The inicall "cns3xxx_pcie_init" does not check which
> hardware it's running on and just tries to access to its specific 
> registers. Now call it from .init_late callback from the two
> machine descriptors.
> 
> Signed-off-by: Xia Kaixu <kaixu.xia@linaro.org>
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> Cc: Anton Vorontsov <anton@enomsg.org>
> Cc: Felix Fietkau <nbd@openwrt.org>
> Cc: Imre Kaloz <kaloz@openwrt.org>
> Cc: linaro-kernel at lists.linaro.org
> Cc: linux-arm-kernel at lists.infradead.org
> ---
>  arch/arm/mach-cns3xxx/cns3420vb.c | 1 +
>  arch/arm/mach-cns3xxx/core.c      | 1 +
>  arch/arm/mach-cns3xxx/core.h      | 9 +++++++++
>  arch/arm/mach-cns3xxx/pcie.c      | 3 +--
>  4 files changed, 12 insertions(+), 2 deletions(-)

I've applied this to next/fixes-non-critical now. I'm fairly confident
that this helps more than it can hurt, but not confident enough
to put it into 3.17 or stable for now.

It would be nice to get some testing from someone with
hardware access. If necessary, I'll revert the change or
do a fixup on top.

	Arnd

[PATCH] ARM: cns3xxx: fix allmodconfig panic in pci driver

[Arnd Bergmann]

On Wednesday 03 September 2014 21:18:12 Xia Kaixu wrote:
> 
> +#ifdef CONFIG_PCI
> +extern int __init cns3xxx_pcie_init_late(void);
> +#else
> +static inline int __init cns3xxx_pcie_init_late(void)
> +{
> +       return -ENXIO;
> +}
> +#endif
> +
>  void __init cns3xxx_map_io(void);
> 

I found another problem: we get a new compile-time warning
because the function returns an 'int' while the .init_late
callback expects a function returning 'void'.

I've fixed it up locally, but please be more careful about
introducing warnings in the future.

	Arnd