[PATCH 0/2] pinctrl: apple: Fixup and RFC GPIO mode patch

[PATCH 0/2] pinctrl: apple: Fixup and RFC GPIO mode patch

[Linus Walleij]

This fixes some uses of the "unsigned" type to "unsigned int"
then propose to implement the .function_is_gpio() callback.

The Apple pin control maintainers can comment on this: I
am not sure that "mode 0" is GPIO on this hardware but I
find it likely.

Toggling a pin between a certain function mode and GPIO
mode happens on any sufficiently advanced system sooner or
later and this callback was implemented because Qualcomm
ran into it, so let's add it to the Apple driver before
the users turn up.

Signed-off-by: Linus Walleij <linusw@kernel.org>
---
Linus Walleij (2):
      pinctrl: apple: Use unsigned int instead of unsigned
      RFC: pinctrl: apple: Implement GPIO func check callback

 drivers/pinctrl/pinctrl-apple-gpio.c | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)
---
base-commit: 8f0b4cce4481fb22653697cced8d0d04027cb1e8
change-id: 20260111-apple-req-gpio-func-3010ca45a118

Best regards,
-- 
Linus Walleij <linusw@kernel.org>

[PATCH 1/2] pinctrl: apple: Use unsigned int instead of unsigned

[Linus Walleij]

It is discouraged to use the ambiguous "unsigned" type, use
explicit unsigned int in the driver.

Signed-off-by: Linus Walleij <linusw@kernel.org>
---
 drivers/pinctrl/pinctrl-apple-gpio.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/pinctrl/pinctrl-apple-gpio.c b/drivers/pinctrl/pinctrl-apple-gpio.c
index e1a7bc8cf765..a4161d59ab9f 100644
--- a/drivers/pinctrl/pinctrl-apple-gpio.c
+++ b/drivers/pinctrl/pinctrl-apple-gpio.c
@@ -102,9 +102,9 @@ static u32 apple_gpio_get_reg(struct apple_gpio_pinctrl *pctl,
 static int apple_gpio_dt_node_to_map(struct pinctrl_dev *pctldev,
 				     struct device_node *node,
 				     struct pinctrl_map **map,
-				     unsigned *num_maps)
+				     unsigned int *num_maps)
 {
-	unsigned reserved_maps;
+	unsigned int reserved_maps;
 	struct apple_gpio_pinctrl *pctl;
 	u32 pinfunc, pin, func;
 	int num_pins, i, ret;
@@ -170,8 +170,8 @@ static const struct pinctrl_ops apple_gpio_pinctrl_ops = {
 
 /* Pin multiplexer functions */
 
-static int apple_gpio_pinmux_set(struct pinctrl_dev *pctldev, unsigned func,
-				 unsigned group)
+static int apple_gpio_pinmux_set(struct pinctrl_dev *pctldev, unsigned int func,
+				 unsigned int group)
 {
 	struct apple_gpio_pinctrl *pctl = pinctrl_dev_get_drvdata(pctldev);
 
@@ -202,7 +202,7 @@ static int apple_gpio_get_direction(struct gpio_chip *chip, unsigned int offset)
 	return GPIO_LINE_DIRECTION_IN;
 }
 
-static int apple_gpio_get(struct gpio_chip *chip, unsigned offset)
+static int apple_gpio_get(struct gpio_chip *chip, unsigned int offset)
 {
 	struct apple_gpio_pinctrl *pctl = gpiochip_get_data(chip);
 	unsigned int reg = apple_gpio_get_reg(pctl, offset);

-- 
2.52.0

[PATCH 2/2] RFC: pinctrl: apple: Implement GPIO func check callback

[Linus Walleij]

This function will check if the indicated function enumerator
i.e. the value set into bits 5 and 6 of the pin configuration
register, is zero and thus corresponds to the GPIO mode of the
pin.

This may be necessary to know since the pin controller is
flagged as "strict": once you set a pin *explicitly* to GPIO
mode using function 0, the core is unaware that this is
actually the GPIO mode, and clients may be denied to
retrieve the pin as a GPIO.

Currently none of the in-kernel device trees sets any pin to
GPIO mode, but this may happen any day.

Signed-off-by: Linus Walleij <linusw@kernel.org>
---
 drivers/pinctrl/pinctrl-apple-gpio.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/drivers/pinctrl/pinctrl-apple-gpio.c b/drivers/pinctrl/pinctrl-apple-gpio.c
index a4161d59ab9f..fd20d0e49f48 100644
--- a/drivers/pinctrl/pinctrl-apple-gpio.c
+++ b/drivers/pinctrl/pinctrl-apple-gpio.c
@@ -170,6 +170,13 @@ static const struct pinctrl_ops apple_gpio_pinctrl_ops = {
 
 /* Pin multiplexer functions */
 
+static int apple_gpio_pinmux_func_is_gpio(struct pinctrl_dev *pctldev,
+					  unsigned int selector)
+{
+	/* Function selector 0 is always the GPIO mode */
+	return (selector == 0);
+}
+
 static int apple_gpio_pinmux_set(struct pinctrl_dev *pctldev, unsigned int func,
 				 unsigned int group)
 {
@@ -186,6 +193,7 @@ static const struct pinmux_ops apple_gpio_pinmux_ops = {
 	.get_functions_count = pinmux_generic_get_function_count,
 	.get_function_name = pinmux_generic_get_function_name,
 	.get_function_groups = pinmux_generic_get_function_groups,
+	.function_is_gpio = apple_gpio_pinmux_func_is_gpio,
 	.set_mux = apple_gpio_pinmux_set,
 	.strict = true,
 };

-- 
2.52.0

Re: [PATCH 1/2] pinctrl: apple: Use unsigned int instead of unsigned

[Sven Peter]

On 11.01.26 21:29, Linus Walleij wrote:
> It is discouraged to use the ambiguous "unsigned" type, use
> explicit unsigned int in the driver.
> 
> Signed-off-by: Linus Walleij <linusw@kernel.org>
> ---

Reviewed-by: Sven Peter <sven@kernel.org>


Thanks,


Sven

Re: [PATCH 2/2] RFC: pinctrl: apple: Implement GPIO func check callback

[Sven Peter]

On 11.01.26 21:29, Linus Walleij wrote:
> This function will check if the indicated function enumerator
> i.e. the value set into bits 5 and 6 of the pin configuration
> register, is zero and thus corresponds to the GPIO mode of the
> pin.
> 
> This may be necessary to know since the pin controller is
> flagged as "strict": once you set a pin *explicitly* to GPIO
> mode using function 0, the core is unaware that this is
> actually the GPIO mode, and clients may be denied to
> retrieve the pin as a GPIO.
> 
> Currently none of the in-kernel device trees sets any pin to
> GPIO mode, but this may happen any day.
> 
> Signed-off-by: Linus Walleij <linusw@kernel.org>
> ---

Just took a quick look: When we use them as GPIO we always set those two 
bits to zero and all "special" functions (like i2c or spi) always start 
at 1. As far as I can tell we can still read the pin value even when bit 
5/6 aren't zero (and e.g. observe i2c traffic that way) but we can't 
drive them anymore.

Since we don't have any documentation for this hardware that's probably 
as good as it'll get:

Reviewed-by: Sven Peter <sven@kernel.org>


Sven

Re: [PATCH 0/2] pinctrl: apple: Fixup and RFC GPIO mode patch

[Bartosz Golaszewski]

On Sun, 11 Jan 2026 21:29:20 +0100, Linus Walleij <linusw@kernel.org> said:
> This fixes some uses of the "unsigned" type to "unsigned int"
> then propose to implement the .function_is_gpio() callback.
>
> The Apple pin control maintainers can comment on this: I
> am not sure that "mode 0" is GPIO on this hardware but I
> find it likely.
>
> Toggling a pin between a certain function mode and GPIO
> mode happens on any sufficiently advanced system sooner or
> later and this callback was implemented because Qualcomm
> ran into it, so let's add it to the Apple driver before
> the users turn up.
>
> Signed-off-by: Linus Walleij <linusw@kernel.org>
> ---
> Linus Walleij (2):
>       pinctrl: apple: Use unsigned int instead of unsigned
>       RFC: pinctrl: apple: Implement GPIO func check callback
>
>  drivers/pinctrl/pinctrl-apple-gpio.c | 18 +++++++++++++-----
>  1 file changed, 13 insertions(+), 5 deletions(-)
> ---
> base-commit: 8f0b4cce4481fb22653697cced8d0d04027cb1e8
> change-id: 20260111-apple-req-gpio-func-3010ca45a118
>
> Best regards,
> --
> Linus Walleij <linusw@kernel.org>
>
>

Reviewed-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>