[PATCH] Give gpiolib an implementation of gpio_is_valid()

[PATCH] Give gpiolib an implementation of gpio_is_valid()

[Ben Nizette]

Platforms which use generic (NOP) gpio routines get an implementation of
gpio_is_valid() (in linux/gpio.h) but the people who actually use gpio
don't get one.  Roll an implementation for gpiolib based on
gpio_request(); this ought to do for most people.

Signed-off-by: Ben Nizette <bn@niasdigital.com>
---
diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
index d8db2f8..137688c 100644
--- a/drivers/gpio/gpiolib.c
+++ b/drivers/gpio/gpiolib.c
@@ -158,6 +158,34 @@ int gpiochip_remove(struct gpio_chip *chip)
 }
 EXPORT_SYMBOL_GPL(gpiochip_remove);
 
+/**
+ * gpio_is_valid() - check if a gpio number actually corresponds to a gpio
+ * @number: the gpio number to test
+ *
+ * Returns 1 if the gpio is valid, 0 otherwise.
+ */
+int gpio_is_valid(int number)
+{
+	struct gpio_desc	*desc;
+	int			ret = 0;
+	unsigned long		flags;
+
+	spin_lock_irqsave(&gpio_lock, flags);
+
+	if (number >= ARCH_NR_GPIOS)
+		goto done;
+
+	desc = &gpio_desc[number];
+	if (desc->chip == NULL)
+		goto done;
+
+	ret = 1;
+
+done:
+	spin_unlock_irqrestore(&gpio_lock, flags);
+	return ret;
+}
+EXPORT_SYMBOL_GPL(gpio_is_valid);
 
 /* These "optional" allocation calls help prevent drivers from stomping
  * on each other, and help provide better diagnostics in debugfs.
diff --git a/include/asm-generic/gpio.h b/include/asm-generic/gpio.h
index f29a502..8c26523 100644
--- a/include/asm-generic/gpio.h
+++ b/include/asm-generic/gpio.h
@@ -75,6 +75,8 @@ extern int __must_check gpiochip_remove(struct gpio_chip *chip);
 /* Always use the library code for GPIO management calls,
  * or when sleeping may be involved.
  */
+extern int gpio_is_valid(int number);
+
 extern int gpio_request(unsigned gpio, const char *label);
 extern void gpio_free(unsigned gpio);
 

Re: [PATCH] Give gpiolib an implementation of gpio_is_valid()

[David Brownell]

On Saturday 26 April 2008, Ben Nizette wrote:
> Platforms which use generic (NOP) gpio routines get an implementation of
> gpio_is_valid() (in linux/gpio.h) but the people who actually use gpio
> don't get one.  Roll an implementation for gpiolib based on
> gpio_request(); this ought to do for most people.

There's already a gpio_is_valid() patch in the MM merge queue;
it comes with the documentation, and covers implementations which
include the <asm-generic/gpio.h> header (for cansleep stubs) but
don't use gpiolib.

So, please wait till RC1, by which time I expect Andrew's queue
will be relatively depleted.  :)

- Dave

Re: [PATCH] Give gpiolib an implementation of gpio_is_valid()

[Ben Nizette]

On Sat, 2008-04-26 at 09:12 -0700, David Brownell wrote:
> On Saturday 26 April 2008, Ben Nizette wrote:
> > Platforms which use generic (NOP) gpio routines get an implementation of
> > gpio_is_valid() (in linux/gpio.h) but the people who actually use gpio
> > don't get one.  Roll an implementation for gpiolib based on
> > gpio_request(); this ought to do for most people.
> 
> There's already a gpio_is_valid() patch in the MM merge queue;
> it comes with the documentation, and covers implementations which
> include the <asm-generic/gpio.h> header (for cansleep stubs) but
> don't use gpiolib.
> 
> So, please wait till RC1, by which time I expect Andrew's queue
> will be relatively depleted.  :)
> 
Righteo, thanks and sorry for being impatient :-)

	--Ben.