[PATCH v2] misc: eeprom: idt_89hpesx: fix fwnode leak in idt_get_fw_data()

[PATCH v2] misc: eeprom: idt_89hpesx: fix fwnode leak in idt_get_fw_data()

[Manush Prajwal]

device_for_each_child_node() takes a reference on each fwnode it
returns, which the loop body must release before it stops iterating.
When idt_ee_match_id() finds a matching EEPROM child, the loop breaks
out immediately without calling fwnode_handle_put() on that fwnode,
leaking the reference.

Rework the loop around device_for_each_child_node_scoped() instead of
adding a manual fwnode_handle_put(), so the reference is released
automatically on every exit path (the early break as well as normal
loop completion), per Markus Elfring's review of v1.

Signed-off-by: Manush Prajwal <manushprajwal555@gmail.com>
---
v2: Use device_for_each_child_node_scoped() instead of a manual
    fwnode_handle_put() before the break, and use my full name in
    Signed-off-by, per Markus Elfring's review.

 drivers/misc/eeprom/idt_89hpesx.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/misc/eeprom/idt_89hpesx.c b/drivers/misc/eeprom/idt_89hpesx.c
index e056d2dea..a1b2c3d4e 100644
--- a/drivers/misc/eeprom/idt_89hpesx.c
+++ b/drivers/misc/eeprom/idt_89hpesx.c
@@ -1080,7 +1080,7 @@ static void idt_get_fw_data(struct idt_89hpesx_dev *pdev)
 	struct device *dev = &pdev->client->dev;
 	struct fwnode_handle *fwnode;

-	device_for_each_child_node(dev, fwnode) {
+	device_for_each_child_node_scoped(dev, fwnode) {
 		ee_id = idt_ee_match_id(fwnode);
 		if (ee_id)
 			break;
--
2.46.2.windows.1

Re: [PATCH v2] misc: eeprom: idt_89hpesx: fix fwnode leak in idt_get_fw_data()

[Greg KH]

On Sun, Aug 23, 2026 at 10:35:46PM +0530, Manush Prajwal wrote:
> device_for_each_child_node() takes a reference on each fwnode it
> returns, which the loop body must release before it stops iterating.
> When idt_ee_match_id() finds a matching EEPROM child, the loop breaks
> out immediately without calling fwnode_handle_put() on that fwnode,
> leaking the reference.
> 
> Rework the loop around device_for_each_child_node_scoped() instead of
> adding a manual fwnode_handle_put(), so the reference is released
> automatically on every exit path (the early break as well as normal
> loop completion), per Markus Elfring's review of v1.
> 
> Signed-off-by: Manush Prajwal <manushprajwal555@gmail.com>
> ---
> v2: Use device_for_each_child_node_scoped() instead of a manual
>     fwnode_handle_put() before the break, and use my full name in
>     Signed-off-by, per Markus Elfring's review.
> 
>  drivers/misc/eeprom/idt_89hpesx.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/misc/eeprom/idt_89hpesx.c b/drivers/misc/eeprom/idt_89hpesx.c
> index e056d2dea..a1b2c3d4e 100644
> --- a/drivers/misc/eeprom/idt_89hpesx.c
> +++ b/drivers/misc/eeprom/idt_89hpesx.c
> @@ -1080,7 +1080,7 @@ static void idt_get_fw_data(struct idt_89hpesx_dev *pdev)
>  	struct device *dev = &pdev->client->dev;
>  	struct fwnode_handle *fwnode;
> 
> -	device_for_each_child_node(dev, fwnode) {
> +	device_for_each_child_node_scoped(dev, fwnode) {
>  		ee_id = idt_ee_match_id(fwnode);
>  		if (ee_id)
>  			break;
> --
> 2.46.2.windows.1
> 

Breaks the build, please be more careful.