[PATCH] gpio: shared: drop the parent check for reset-gpio devices

[PATCH] gpio: shared: drop the parent check for reset-gpio devices

[Tyler Baker]

gpio_shared_dev_is_reset_gpio() requires the reset-gpio device's parent
to carry the GPIO chip's fwnode. Once the reset core parents that device
to the GPIO controller instead, this fails on controllers with a
gpio_chip and fwnode per bank, so their shared reset lines are no longer
matched.

The check is redundant: only devices compatible with "reset-gpio" get
here, and their reset-gpios reference must already name the chip's
fwnode. Drop it.

Signed-off-by: Tyler Baker <tyler.baker@oss.qualcomm.com>
---
On top of Abel's "reset: gpio: Parent the reset-gpio device to the GPIO
controller":
https://lore.kernel.org/r/20261026-b4-reset-gpio-parent-to-gpio-controller-v1-1-11549e6ea930@oss.qualcomm.com

 drivers/gpio/gpiolib-shared.c | 11 +----------
 1 file changed, 1 insertion(+), 10 deletions(-)

diff --git a/drivers/gpio/gpiolib-shared.c b/drivers/gpio/gpiolib-shared.c
index 8c9ac9d933049..91d91c78fe55d 100644
--- a/drivers/gpio/gpiolib-shared.c
+++ b/drivers/gpio/gpiolib-shared.c
@@ -362,22 +362,13 @@ static bool gpio_shared_dev_is_reset_gpio(struct device *consumer,
 {
 	struct fwnode_handle *reset_fwnode = dev_fwnode(consumer);
 	struct fwnode_reference_args ref_args, aux_args;
-	struct device *parent = consumer->parent;
 	struct gpio_shared_ref *real_ref;
 	bool match;
 	int ret;
 
 	lockdep_assert_held(&ref->lock);
 
-	/* The reset-gpio device must have a parent AND a firmware node. */
-	if (!parent || !reset_fwnode)
-		return false;
-
-	/*
-	 * Parent of the reset-gpio auxiliary device is the GPIO chip whose
-	 * fwnode we stored in the entry structure.
-	 */
-	if (!device_match_fwnode(parent, entry->fwnode))
+	if (!reset_fwnode)
 		return false;
 
 	/*
-- 
2.47.3

Re: [PATCH] gpio: shared: drop the parent check for reset-gpio devices

[Bartosz Golaszewski]

On Fri, Nov 06, 2026 at 12:28:50AM -0500, Tyler Baker wrote:
> gpio_shared_dev_is_reset_gpio() requires the reset-gpio device's parent
> to carry the GPIO chip's fwnode. Once the reset core parents that device
> to the GPIO controller instead, this fails on controllers with a
> gpio_chip and fwnode per bank, so their shared reset lines are no longer
> matched.
>

Again: mixing up of the naming convention for devices. It should be:

"gpio_shared_dev_is_reset_gpio() requires the reset-gpio device's parent
to carry the GPIO controller's fwnode. Once the reset core parents that device
to the GPIO controller's parent instead, this fails on controllers with a
gpio_chip and separate fwnode per bank, so their shared reset lines are no
longer matched."

> The check is redundant: only devices compatible with "reset-gpio" get
> here, and their reset-gpios reference must already name the chip's
> fwnode. Drop it.
> 
> Signed-off-by: Tyler Baker <tyler.baker@oss.qualcomm.com>
> ---
> On top of Abel's "reset: gpio: Parent the reset-gpio device to the GPIO
> controller":
> https://lore.kernel.org/r/20261026-b4-reset-gpio-parent-to-gpio-controller-v1-1-11549e6ea930@oss.qualcomm.com

This passes my tests, though I have it in the back of my head that there
was a reason for this more strict check. I can't remember what it was
anymore so I'm fine with the change, if it introduces a regression, we
can revert it.

Abel's patch introduces a regression that this patch fixes so we should squash
the two, get an Acked-by from Philipp and sent them via the GPIO tree.

Bart