[PATCH 0/2] ARM: highbank: add missing of_node_put calls

[PATCH 0/2] ARM: highbank: add missing of_node_put calls

[Martin Kaiser]

Here's two simple patches to call of_node_put for device nodes that were
returned by of_find_compatible_node.

Both patches were compile-tested only as I don't have the hardware to run
the code.

Martin Kaiser (2):
  ARM: highbank: add missing of_node_put call
  ARM: highbank: fix of node handling in highbank_init_irq

 arch/arm/mach-highbank/highbank.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

-- 
2.30.2

[PATCH 1/2] ARM: highbank: add missing of_node_put call

[Martin Kaiser]

A node that is returned by of_find_compatible_node has its refcount
incremented. We have to call of_node_put when the node is no longer
needed.

Add a missing of_node_put call in function highbank_init.

Signed-off-by: Martin Kaiser <martin@kaiser.cx>
---
 arch/arm/mach-highbank/highbank.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm/mach-highbank/highbank.c b/arch/arm/mach-highbank/highbank.c
index 5d4f977ac7d2..97ccb8c1b3cd 100644
--- a/arch/arm/mach-highbank/highbank.c
+++ b/arch/arm/mach-highbank/highbank.c
@@ -141,6 +141,7 @@ static void __init highbank_init(void)
 	/* Map system registers */
 	np = of_find_compatible_node(NULL, NULL, "calxeda,hb-sregs");
 	sregs_base = of_iomap(np, 0);
+	of_node_put(np);
 	WARN_ON(!sregs_base);
 
 	pm_power_off = highbank_power_off;
-- 
2.30.2

Re: [PATCH 1/2] ARM: highbank: add missing of_node_put call

[Andre Przywara]

On Sun, 18 Dec 2022 15:26:46 +0100
Martin Kaiser <martin@kaiser.cx> wrote:

Hi Martin,

> A node that is returned by of_find_compatible_node has its refcount
> incremented. We have to call of_node_put when the node is no longer
> needed.
> 
> Add a missing of_node_put call in function highbank_init.

Thanks for the patch.
So Ruyi sent the exact same one earlier this year already:
https://lore.kernel.org/linux-arm-kernel/20220408094817.2494756-1-lv.ruyi@zte.com.cn/

I asked Arnd and Olof to take it back then (see my reply), but this somehow
got lost.

I will try to push on both Ruyi's and your second patch again - but I
guess only after the holidays.

Cheers,
Andre

> Signed-off-by: Martin Kaiser <martin@kaiser.cx>
> ---
>  arch/arm/mach-highbank/highbank.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/arch/arm/mach-highbank/highbank.c b/arch/arm/mach-highbank/highbank.c
> index 5d4f977ac7d2..97ccb8c1b3cd 100644
> --- a/arch/arm/mach-highbank/highbank.c
> +++ b/arch/arm/mach-highbank/highbank.c
> @@ -141,6 +141,7 @@ static void __init highbank_init(void)
>  	/* Map system registers */
>  	np = of_find_compatible_node(NULL, NULL, "calxeda,hb-sregs");
>  	sregs_base = of_iomap(np, 0);
> +	of_node_put(np);
>  	WARN_ON(!sregs_base);
>  
>  	pm_power_off = highbank_power_off;

[PATCH 2/2] ARM: highbank: fix of node handling in highbank_init_irq

[Martin Kaiser]

highbank_init calls of_find_compatible_node and discards the returned
node. We should save the node and call of_node_put on it when finished.

Signed-off-by: Martin Kaiser <martin@kaiser.cx>
---
 arch/arm/mach-highbank/highbank.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/arch/arm/mach-highbank/highbank.c b/arch/arm/mach-highbank/highbank.c
index 97ccb8c1b3cd..54eb1422ef8f 100644
--- a/arch/arm/mach-highbank/highbank.c
+++ b/arch/arm/mach-highbank/highbank.c
@@ -50,10 +50,16 @@ static void highbank_l2c310_write_sec(unsigned long val, unsigned reg)
 
 static void __init highbank_init_irq(void)
 {
+	struct device_node *np;
+
 	irqchip_init();
 
-	if (of_find_compatible_node(NULL, NULL, "arm,cortex-a9"))
-		highbank_scu_map_io();
+	np = of_find_compatible_node(NULL, NULL, "arm,cortex-a9");
+	if (!np)
+		return;
+
+	highbank_scu_map_io();
+	of_node_put(np);
 }
 
 static void highbank_power_off(void)
-- 
2.30.2

Re: [PATCH 2/2] ARM: highbank: fix of node handling in highbank_init_irq

[Andre Przywara]

On Sun, 18 Dec 2022 15:26:47 +0100
Martin Kaiser <martin@kaiser.cx> wrote:

Hi Martin,

thanks for taking care of this.

> highbank_init calls of_find_compatible_node and discards the returned
> node. We should save the node and call of_node_put on it when finished.
> 
> Signed-off-by: Martin Kaiser <martin@kaiser.cx>
> ---
>  arch/arm/mach-highbank/highbank.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/arm/mach-highbank/highbank.c b/arch/arm/mach-highbank/highbank.c
> index 97ccb8c1b3cd..54eb1422ef8f 100644
> --- a/arch/arm/mach-highbank/highbank.c
> +++ b/arch/arm/mach-highbank/highbank.c
> @@ -50,10 +50,16 @@ static void highbank_l2c310_write_sec(unsigned long val, unsigned reg)
>  
>  static void __init highbank_init_irq(void)
>  {
> +	struct device_node *np;
> +
>  	irqchip_init();
>  
> -	if (of_find_compatible_node(NULL, NULL, "arm,cortex-a9"))
> -		highbank_scu_map_io();
> +	np = of_find_compatible_node(NULL, NULL, "arm,cortex-a9");
> +	if (!np)
> +		return;
> +
> +	highbank_scu_map_io();
> +	of_node_put(np);

Technically the of_node_put() could happen even earlier, since we
use the of_find_compatible_node() call only to figure out if this is the
Cotex-A15 or the A9 part. But it doesn't really matter, and is correct
anyways:

Reviewed-by: Andre Przywara <andre.przywara@arm.com>

I also quickly boot tested this on a Midway(A15), but don't have a
Highbank(A9) readily set up at the moment.

Cheers,
Andre


>  }
>  
>  static void highbank_power_off(void)