[PATCH] ARM: OMAP: Fix kernel panic in dmtimer probe

[PATCH] ARM: OMAP: Fix kernel panic in dmtimer probe

[Jon Hunter]

When booting with device-tree the kernel is panicing in the probe of the
DMTIMER driver. The panic is caused because the pointer to platform_data
structure is NULL when booting with device-tree and the driver is
attempting to access the structure without checking if the pointer is
valid.

Fix this by moving the code that accesses the platform data structure
under the "else" clause of the "if (dev->of_node)" statement because
here the pointer to platform_data is guaranteed to be valid. The code
accessing the "timer_capability" member of the platform data is simply
removed as this is already handled under the else clause.

This regression was introduced while integrating commit "ARM: OMAP: Add
DT support for timer driver" to add device-tree support to the DMTIMER
driver and commit "ARM: OMAP: Move omap-pm-noop.c local to mach-omap2"
to prepare for single zImage support.

Signed-off-by: Jon Hunter <jon-hunter@ti.com>
---
 arch/arm/plat-omap/dmtimer.c |    3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/arch/arm/plat-omap/dmtimer.c b/arch/arm/plat-omap/dmtimer.c
index 38c12ef..9dca23e 100644
--- a/arch/arm/plat-omap/dmtimer.c
+++ b/arch/arm/plat-omap/dmtimer.c
@@ -799,12 +799,11 @@ static int __devinit omap_dm_timer_probe(struct platform_device *pdev)
 		timer->id = pdev->id;
 		timer->capability = pdata->timer_capability;
 		timer->reserved = omap_dm_timer_reserved_systimer(timer->id);
+		timer->get_context_loss_count = pdata->get_context_loss_count;
 	}
 
 	timer->irq = irq->start;
 	timer->pdev = pdev;
-	timer->capability = pdata->timer_capability;
-	timer->get_context_loss_count = pdata->get_context_loss_count;
 
 	/* Skip pm_runtime_enable for OMAP1 */
 	if (!(timer->capability & OMAP_TIMER_NEEDS_RESET)) {
-- 
1.7.9.5

[PATCH] ARM: OMAP: Fix kernel panic in dmtimer probe

[Jon Hunter]

On 11/08/2012 06:17 PM, Jon Hunter wrote:
> When booting with device-tree the kernel is panicing in the probe of the
> DMTIMER driver. The panic is caused because the pointer to platform_data
> structure is NULL when booting with device-tree and the driver is
> attempting to access the structure without checking if the pointer is
> valid.
> 
> Fix this by moving the code that accesses the platform data structure
> under the "else" clause of the "if (dev->of_node)" statement because
> here the pointer to platform_data is guaranteed to be valid. The code
> accessing the "timer_capability" member of the platform data is simply
> removed as this is already handled under the else clause.
> 
> This regression was introduced while integrating commit "ARM: OMAP: Add
> DT support for timer driver" to add device-tree support to the DMTIMER
> driver and commit "ARM: OMAP: Move omap-pm-noop.c local to mach-omap2"
> to prepare for single zImage support.


By the way, I meant to add ...

Reported-by: Tony Lindgren <tony@atomide.com>

Jon

> Signed-off-by: Jon Hunter <jon-hunter@ti.com>
> ---
>  arch/arm/plat-omap/dmtimer.c |    3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/arch/arm/plat-omap/dmtimer.c b/arch/arm/plat-omap/dmtimer.c
> index 38c12ef..9dca23e 100644
> --- a/arch/arm/plat-omap/dmtimer.c
> +++ b/arch/arm/plat-omap/dmtimer.c
> @@ -799,12 +799,11 @@ static int __devinit omap_dm_timer_probe(struct platform_device *pdev)
>  		timer->id = pdev->id;
>  		timer->capability = pdata->timer_capability;
>  		timer->reserved = omap_dm_timer_reserved_systimer(timer->id);
> +		timer->get_context_loss_count = pdata->get_context_loss_count;
>  	}
>  
>  	timer->irq = irq->start;
>  	timer->pdev = pdev;
> -	timer->capability = pdata->timer_capability;
> -	timer->get_context_loss_count = pdata->get_context_loss_count;
>  
>  	/* Skip pm_runtime_enable for OMAP1 */
>  	if (!(timer->capability & OMAP_TIMER_NEEDS_RESET)) {
> 

[PATCH] ARM: OMAP: Fix kernel panic in dmtimer probe

[Tony Lindgren]

* Jon Hunter <jon-hunter@ti.com> [121108 16:19]:
> When booting with device-tree the kernel is panicing in the probe of the
> DMTIMER driver. The panic is caused because the pointer to platform_data
> structure is NULL when booting with device-tree and the driver is
> attempting to access the structure without checking if the pointer is
> valid.
> 
> Fix this by moving the code that accesses the platform data structure
> under the "else" clause of the "if (dev->of_node)" statement because
> here the pointer to platform_data is guaranteed to be valid. The code
> accessing the "timer_capability" member of the platform data is simply
> removed as this is already handled under the else clause.
> 
> This regression was introduced while integrating commit "ARM: OMAP: Add
> DT support for timer driver" to add device-tree support to the DMTIMER
> driver and commit "ARM: OMAP: Move omap-pm-noop.c local to mach-omap2"
> to prepare for single zImage support.

Thanks, looks like this needs to be fixed in the merge, so I'll
merge omap-for-v3.8/cleanup-headers-prepare-multiplatform-v3 into
omap-for-v3.8/dt and resolve the conflict according to your patch.

Regards,

Tony

 
> Signed-off-by: Jon Hunter <jon-hunter@ti.com>
> ---
>  arch/arm/plat-omap/dmtimer.c |    3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> diff --git a/arch/arm/plat-omap/dmtimer.c b/arch/arm/plat-omap/dmtimer.c
> index 38c12ef..9dca23e 100644
> --- a/arch/arm/plat-omap/dmtimer.c
> +++ b/arch/arm/plat-omap/dmtimer.c
> @@ -799,12 +799,11 @@ static int __devinit omap_dm_timer_probe(struct platform_device *pdev)
>  		timer->id = pdev->id;
>  		timer->capability = pdata->timer_capability;
>  		timer->reserved = omap_dm_timer_reserved_systimer(timer->id);
> +		timer->get_context_loss_count = pdata->get_context_loss_count;
>  	}
>  
>  	timer->irq = irq->start;
>  	timer->pdev = pdev;
> -	timer->capability = pdata->timer_capability;
> -	timer->get_context_loss_count = pdata->get_context_loss_count;
>  
>  	/* Skip pm_runtime_enable for OMAP1 */
>  	if (!(timer->capability & OMAP_TIMER_NEEDS_RESET)) {
> -- 
> 1.7.9.5
>