[PATCH v1] gpiolib: Don't probe gpio_device if it's not the primary device

[PATCH v1] gpiolib: Don't probe gpio_device if it's not the primary device

[Saravana Kannan]

Dmitry reported[1] boot error messages caused by
commit 4731210c09f5 ("gpiolib: Bind gpio_device to a driver to enable fw_devlink=on by default").

gpio-1022 (cpu-pwr-req-hog): hogged as input
max77620-pinctrl max77620-pinctrl: pin gpio4 already requested by max77620-pinctrl; cannot claim for gpiochip1
max77620-pinctrl max77620-pinctrl: pin-4 (gpiochip1) status -22
max77620-pinctrl max77620-pinctrl: could not request pin 4 (gpio4) from group gpio4  on device max77620-pinctrl
gpio_stub_drv gpiochip1: Error applying setting, reverse things back
gpio_stub_drv: probe of gpiochip1 failed with error -22

This happens because when we try to probe a device, driver core calls
into pinctrl to set up the pins. However, if the GPIO DT node already
has a proper device created and probed, trying to probe the gpio_device
with a stub driver makes the pins be claimed twice. pinctrl doesn't like
this and throws an error.

So, this patch makes sure the gpio_stub_drv doesn't match with a
gpio_device if it's not the primary device for the fwnode.

[1] - https://lore.kernel.org/lkml/544ad0e4-0954-274c-8e77-866aaa5661a8@gmail.com/
Fixes: 4731210c09f5 ("gpiolib: Bind gpio_device to a driver to enable fw_devlink=on by default")
Signed-off-by: Saravana Kannan <saravanak@google.com>
Tested-by: Dmitry Osipenko <digetx@gmail.com>
---
Greg/Linus,

This will need to go into driver-core because the Fixes is in
driver-core too.

Thanks,
Saravana

 drivers/gpio/gpiolib.c | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
index 8e0564c50840..8ad679a928b0 100644
--- a/drivers/gpio/gpiolib.c
+++ b/drivers/gpio/gpiolib.c
@@ -56,8 +56,10 @@
 static DEFINE_IDA(gpio_ida);
 static dev_t gpio_devt;
 #define GPIO_DEV_MAX 256 /* 256 GPIO chip devices supported */
+static int gpio_bus_match(struct device *dev, struct device_driver *drv);
 static struct bus_type gpio_bus_type = {
 	.name = "gpio",
+	.match = gpio_bus_match,
 };
 
 /*
@@ -4199,6 +4201,18 @@ void gpiod_put_array(struct gpio_descs *descs)
 }
 EXPORT_SYMBOL_GPL(gpiod_put_array);
 
+
+static int gpio_bus_match(struct device *dev, struct device_driver *drv)
+{
+	/*
+	 * Only match if the fwnode doesn't already have a proper struct device
+	 * created for it.
+	 */
+	if (dev->fwnode && dev->fwnode->dev != dev)
+		return 0;
+	return 1;
+}
+
 static int gpio_stub_drv_probe(struct device *dev)
 {
 	/*
-- 
2.30.0.365.g02bc693789-goog

Re: [PATCH v1] gpiolib: Don't probe gpio_device if it's not the primary device

[Bartosz Golaszewski]

On Fri, Feb 5, 2021 at 3:08 AM Saravana Kannan <saravanak@google.com> wrote:
>
> Dmitry reported[1] boot error messages caused by
> commit 4731210c09f5 ("gpiolib: Bind gpio_device to a driver to enable fw_devlink=on by default").
>
> gpio-1022 (cpu-pwr-req-hog): hogged as input
> max77620-pinctrl max77620-pinctrl: pin gpio4 already requested by max77620-pinctrl; cannot claim for gpiochip1
> max77620-pinctrl max77620-pinctrl: pin-4 (gpiochip1) status -22
> max77620-pinctrl max77620-pinctrl: could not request pin 4 (gpio4) from group gpio4  on device max77620-pinctrl
> gpio_stub_drv gpiochip1: Error applying setting, reverse things back
> gpio_stub_drv: probe of gpiochip1 failed with error -22
>
> This happens because when we try to probe a device, driver core calls
> into pinctrl to set up the pins. However, if the GPIO DT node already
> has a proper device created and probed, trying to probe the gpio_device
> with a stub driver makes the pins be claimed twice. pinctrl doesn't like
> this and throws an error.
>
> So, this patch makes sure the gpio_stub_drv doesn't match with a
> gpio_device if it's not the primary device for the fwnode.
>
> [1] - https://lore.kernel.org/lkml/544ad0e4-0954-274c-8e77-866aaa5661a8@gmail.com/
> Fixes: 4731210c09f5 ("gpiolib: Bind gpio_device to a driver to enable fw_devlink=on by default")
> Signed-off-by: Saravana Kannan <saravanak@google.com>
> Tested-by: Dmitry Osipenko <digetx@gmail.com>
> ---
> Greg/Linus,
>
> This will need to go into driver-core because the Fixes is in
> driver-core too.
>
> Thanks,
> Saravana
>
>  drivers/gpio/gpiolib.c | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
>
> diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
> index 8e0564c50840..8ad679a928b0 100644
> --- a/drivers/gpio/gpiolib.c
> +++ b/drivers/gpio/gpiolib.c
> @@ -56,8 +56,10 @@
>  static DEFINE_IDA(gpio_ida);
>  static dev_t gpio_devt;
>  #define GPIO_DEV_MAX 256 /* 256 GPIO chip devices supported */
> +static int gpio_bus_match(struct device *dev, struct device_driver *drv);
>  static struct bus_type gpio_bus_type = {
>         .name = "gpio",
> +       .match = gpio_bus_match,
>  };
>
>  /*
> @@ -4199,6 +4201,18 @@ void gpiod_put_array(struct gpio_descs *descs)
>  }
>  EXPORT_SYMBOL_GPL(gpiod_put_array);
>
> +
> +static int gpio_bus_match(struct device *dev, struct device_driver *drv)
> +{
> +       /*
> +        * Only match if the fwnode doesn't already have a proper struct device
> +        * created for it.
> +        */
> +       if (dev->fwnode && dev->fwnode->dev != dev)
> +               return 0;
> +       return 1;
> +}
> +
>  static int gpio_stub_drv_probe(struct device *dev)
>  {
>         /*
> --
> 2.30.0.365.g02bc693789-goog
>

Acked-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>

Re: [PATCH v1] gpiolib: Don't probe gpio_device if it's not the primary device

[Geert Uytterhoeven]

Hi Saravana,

On Fri, Feb 5, 2021 at 3:09 AM Saravana Kannan <saravanak@google.com> wrote:
> Dmitry reported[1] boot error messages caused by
> commit 4731210c09f5 ("gpiolib: Bind gpio_device to a driver to enable fw_devlink=on by default").
>
> gpio-1022 (cpu-pwr-req-hog): hogged as input
> max77620-pinctrl max77620-pinctrl: pin gpio4 already requested by max77620-pinctrl; cannot claim for gpiochip1
> max77620-pinctrl max77620-pinctrl: pin-4 (gpiochip1) status -22
> max77620-pinctrl max77620-pinctrl: could not request pin 4 (gpio4) from group gpio4  on device max77620-pinctrl
> gpio_stub_drv gpiochip1: Error applying setting, reverse things back
> gpio_stub_drv: probe of gpiochip1 failed with error -22

Similar issue on Salvator-XS:

    sh-pfc e6060000.pinctrl: pin GP_2_0 already requested by 7-0030;
cannot claim for gpiochip9
    sh-pfc e6060000.pinctrl: pin-64 (gpiochip9) status -22
    sh-pfc e6060000.pinctrl: could not request pin 64 (GP_2_0) from
group intc_ex_irq0  on device sh-pfc
    gpio_stub_drv gpiochip9: Error applying setting, reverse things back
    gpio_stub_drv: probe of gpiochip9 failed with error -22

> This happens because when we try to probe a device, driver core calls
> into pinctrl to set up the pins. However, if the GPIO DT node already
> has a proper device created and probed, trying to probe the gpio_device
> with a stub driver makes the pins be claimed twice. pinctrl doesn't like
> this and throws an error.
>
> So, this patch makes sure the gpio_stub_drv doesn't match with a
> gpio_device if it's not the primary device for the fwnode.
>
> [1] - https://lore.kernel.org/lkml/544ad0e4-0954-274c-8e77-866aaa5661a8@gmail.com/
> Fixes: 4731210c09f5 ("gpiolib: Bind gpio_device to a driver to enable fw_devlink=on by default")
> Signed-off-by: Saravana Kannan <saravanak@google.com>
> Tested-by: Dmitry Osipenko <digetx@gmail.com>

Tested-by: Geert Uytterhoeven <geert+renesas@glider.be>

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: [PATCH v1] gpiolib: Don't probe gpio_device if it's not the primary device

[Linus Walleij]

On Fri, Feb 5, 2021 at 3:07 AM Saravana Kannan <saravanak@google.com> wrote:

> Dmitry reported[1] boot error messages caused by
> commit 4731210c09f5 ("gpiolib: Bind gpio_device to a driver to enable fw_devlink=on by default").
>
> gpio-1022 (cpu-pwr-req-hog): hogged as input
> max77620-pinctrl max77620-pinctrl: pin gpio4 already requested by max77620-pinctrl; cannot claim for gpiochip1
> max77620-pinctrl max77620-pinctrl: pin-4 (gpiochip1) status -22
> max77620-pinctrl max77620-pinctrl: could not request pin 4 (gpio4) from group gpio4  on device max77620-pinctrl
> gpio_stub_drv gpiochip1: Error applying setting, reverse things back
> gpio_stub_drv: probe of gpiochip1 failed with error -22
>
> This happens because when we try to probe a device, driver core calls
> into pinctrl to set up the pins. However, if the GPIO DT node already
> has a proper device created and probed, trying to probe the gpio_device
> with a stub driver makes the pins be claimed twice. pinctrl doesn't like
> this and throws an error.
>
> So, this patch makes sure the gpio_stub_drv doesn't match with a
> gpio_device if it's not the primary device for the fwnode.
>
> [1] - https://lore.kernel.org/lkml/544ad0e4-0954-274c-8e77-866aaa5661a8@gmail.com/
> Fixes: 4731210c09f5 ("gpiolib: Bind gpio_device to a driver to enable fw_devlink=on by default")
> Signed-off-by: Saravana Kannan <saravanak@google.com>
> Tested-by: Dmitry Osipenko <digetx@gmail.com>
> ---
> Greg/Linus,
>
> This will need to go into driver-core because the Fixes is in
> driver-core too.

OK that seems to be the right solution. Maybe I would try to
explain the situation in the comment in gpio_bus_match
a bit more, but anyway:
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>

Yours,
Linus Walleij