diff mbox

[PATCHv8] gpio: Remove VLA from gpiolib

Message ID 20180518175304.20414-1-labbott@redhat.com (mailing list archive)
State New, archived
Headers show

Commit Message

Laura Abbott May 18, 2018, 5:53 p.m. UTC 
The new challenge is to remove VLAs from the kernel
(see https://lkml.org/lkml/2018/3/7/621) to eventually
turn on -Wvla.

Using a kmalloc array is the easy way to fix this but kmalloc is still
more expensive than stack allocation. Introduce a fast path with a
fixed size stack array to cover most chip with gpios below some fixed
amount. The slow path dynamically allocates an array to cover those
chips with a large number of gpios.

Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Phil Reid <preid@electromag.com.au>
Reviewed-and-tested-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Signed-off-by: Laura Abbott <labbott@redhat.com>
---
v8: Adjust lower bound
---
 drivers/gpio/Kconfig          | 12 ++++++
 drivers/gpio/gpiolib.c        | 76 +++++++++++++++++++++++++++--------
 drivers/gpio/gpiolib.h        |  2 +-
 include/linux/gpio/consumer.h | 10 +++--
 4 files changed, 78 insertions(+), 22 deletions(-)

Comments

Andy Shevchenko May 18, 2018, 8:07 p.m. UTC | #1 
On Fri, May 18, 2018 at 8:53 PM, Laura Abbott <labbott@redhat.com> wrote:
> The new challenge is to remove VLAs from the kernel
> (see https://lkml.org/lkml/2018/3/7/621) to eventually
> turn on -Wvla.
>
> Using a kmalloc array is the easy way to fix this but kmalloc is still
> more expensive than stack allocation. Introduce a fast path with a
> fixed size stack array to cover most chip with gpios below some fixed
> amount. The slow path dynamically allocates an array to cover those
> chips with a large number of gpios.

> +               unsigned long fastpath[2 * BITS_TO_LONGS(FASTPATH_NGPIO)];
> +               unsigned long *mask, *bits;
>                 int first, j, ret;
>
> +               if (likely(chip->ngpio <= FASTPATH_NGPIO)) {
> +                       mask = fastpath;
> +               } else {
> +                       mask = kmalloc_array(2 * BITS_TO_LONGS(chip->ngpio),
> +                                          sizeof(*mask),
> +                                          can_sleep ? GFP_KERNEL : GFP_ATOMIC);
> +                       if (!mask)
> +                               return -ENOMEM;
> +               }
> +
> +               bits = mask + BITS_TO_LONGS(chip->ngpio);
> +               memset(mask, 0, BITS_TO_LONGS(chip->ngpio) * sizeof(*mask));

Wouldn't be better

bitmap_zero(mask, chip->ngpio);

?

> +               bits = mask + BITS_TO_LONGS(chip->ngpio);
> +               memset(mask, 0, BITS_TO_LONGS(chip->ngpio) * sizeof(*mask));

Ditto.
Geert Uytterhoeven May 18, 2018, 9:05 p.m. UTC | #2 
Hi Andy,

On Fri, May 18, 2018 at 10:07 PM, Andy Shevchenko
<andy.shevchenko@gmail.com> wrote:
> On Fri, May 18, 2018 at 8:53 PM, Laura Abbott <labbott@redhat.com> wrote:
>> The new challenge is to remove VLAs from the kernel
>> (see https://lkml.org/lkml/2018/3/7/621) to eventually
>> turn on -Wvla.
>>
>> Using a kmalloc array is the easy way to fix this but kmalloc is still
>> more expensive than stack allocation. Introduce a fast path with a
>> fixed size stack array to cover most chip with gpios below some fixed
>> amount. The slow path dynamically allocates an array to cover those
>> chips with a large number of gpios.
>
>> +               unsigned long fastpath[2 * BITS_TO_LONGS(FASTPATH_NGPIO)];
>> +               unsigned long *mask, *bits;
>>                 int first, j, ret;
>>
>> +               if (likely(chip->ngpio <= FASTPATH_NGPIO)) {
>> +                       mask = fastpath;
>> +               } else {
>> +                       mask = kmalloc_array(2 * BITS_TO_LONGS(chip->ngpio),
>> +                                          sizeof(*mask),
>> +                                          can_sleep ? GFP_KERNEL : GFP_ATOMIC);
>> +                       if (!mask)
>> +                               return -ENOMEM;
>> +               }
>> +
>> +               bits = mask + BITS_TO_LONGS(chip->ngpio);
>> +               memset(mask, 0, BITS_TO_LONGS(chip->ngpio) * sizeof(*mask));
>
> Wouldn't be better
>
> bitmap_zero(mask, chip->ngpio);
>
> ?

chip->ngpio is never const, so bitmap_zero() would always fall back to
plain memset().

Gr{oetje,eeting}s,

                        Geert
Andy Shevchenko May 18, 2018, 9:12 p.m. UTC | #3 
On Sat, May 19, 2018 at 12:05 AM, Geert Uytterhoeven
<geert@linux-m68k.org> wrote:
> On Fri, May 18, 2018 at 10:07 PM, Andy Shevchenko
> <andy.shevchenko@gmail.com> wrote:
>> On Fri, May 18, 2018 at 8:53 PM, Laura Abbott <labbott@redhat.com> wrote:

>>> +               bits = mask + BITS_TO_LONGS(chip->ngpio);
>>> +               memset(mask, 0, BITS_TO_LONGS(chip->ngpio) * sizeof(*mask));
>>
>> Wouldn't be better
>>
>> bitmap_zero(mask, chip->ngpio);
>>
>> ?
>
> chip->ngpio is never const, so bitmap_zero() would always fall back to
> plain memset().

Yes. But it also much more readable and understandable here.
Laura Abbott May 21, 2018, 5:30 p.m. UTC | #4 
On 05/18/2018 02:12 PM, Andy Shevchenko wrote:
> On Sat, May 19, 2018 at 12:05 AM, Geert Uytterhoeven
> <geert@linux-m68k.org> wrote:
>> On Fri, May 18, 2018 at 10:07 PM, Andy Shevchenko
>> <andy.shevchenko@gmail.com> wrote:
>>> On Fri, May 18, 2018 at 8:53 PM, Laura Abbott <labbott@redhat.com> wrote:
> 
>>>> +               bits = mask + BITS_TO_LONGS(chip->ngpio);
>>>> +               memset(mask, 0, BITS_TO_LONGS(chip->ngpio) * sizeof(*mask));
>>>
>>> Wouldn't be better
>>>
>>> bitmap_zero(mask, chip->ngpio);
>>>
>>> ?
>>
>> chip->ngpio is never const, so bitmap_zero() would always fall back to
>> plain memset().
> 
> Yes. But it also much more readable and understandable here.
> 

I wasn't aware of that API. I'll switch to that.
diff mbox

Patch

diff --git a/drivers/gpio/Kconfig b/drivers/gpio/Kconfig
index b960f6f35abd..71c0ab46f216 100644
--- a/drivers/gpio/Kconfig
+++ b/drivers/gpio/Kconfig
@@ -22,6 +22,18 @@  menuconfig GPIOLIB
 
 if GPIOLIB
 
+config GPIOLIB_FASTPATH_LIMIT
+	int "Maximum number of GPIOs for fast path"
+	range 32 512
+	default 512
+	help
+	   This adjusts the point at which certain APIs will switch from
+	   using a stack allocated buffer to a dynamically allocated buffer.
+
+	   You shouldn't need to change this unless you really need to
+	   optimize either stack space or performance. Change this carefully
+	   since setting an incorrect value could cause stack corruption.
+
 config OF_GPIO
 	def_bool y
 	depends on OF
diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
index d8ccb500872f..d29916cd3b74 100644
--- a/drivers/gpio/gpiolib.c
+++ b/drivers/gpio/gpiolib.c
@@ -61,6 +61,11 @@  static struct bus_type gpio_bus_type = {
 	.name = "gpio",
 };
 
+/*
+ * Number of GPIOs to use for the fast path in set array
+ */
+#define FASTPATH_NGPIO CONFIG_GPIOLIB_FASTPATH_LIMIT
+
 /* gpio_lock prevents conflicts during gpio_desc[] table updates.
  * While any GPIO is requested, its gpio_chip is not removable;
  * each GPIO's "requested" flag serves as a lock and refcount.
@@ -450,12 +455,11 @@  static long linehandle_ioctl(struct file *filep, unsigned int cmd,
 			vals[i] = !!ghd.values[i];
 
 		/* Reuse the array setting function */
-		gpiod_set_array_value_complex(false,
+		return gpiod_set_array_value_complex(false,
 					      true,
 					      lh->numdescs,
 					      lh->descs,
 					      vals);
-		return 0;
 	}
 	return -EINVAL;
 }
@@ -1244,6 +1248,10 @@  int gpiochip_add_data_with_key(struct gpio_chip *chip, void *data,
 		goto err_free_descs;
 	}
 
+	if (chip->ngpio > FASTPATH_NGPIO)
+		chip_warn(chip, "line cnt %u is greater than fast path cnt %u\n",
+		chip->ngpio, FASTPATH_NGPIO);
+
 	gdev->label = kstrdup_const(chip->label ?: "unknown", GFP_KERNEL);
 	if (!gdev->label) {
 		status = -ENOMEM;
@@ -2719,16 +2727,28 @@  int gpiod_get_array_value_complex(bool raw, bool can_sleep,
 
 	while (i < array_size) {
 		struct gpio_chip *chip = desc_array[i]->gdev->chip;
-		unsigned long mask[BITS_TO_LONGS(chip->ngpio)];
-		unsigned long bits[BITS_TO_LONGS(chip->ngpio)];
+		unsigned long fastpath[2 * BITS_TO_LONGS(FASTPATH_NGPIO)];
+		unsigned long *mask, *bits;
 		int first, j, ret;
 
+		if (likely(chip->ngpio <= FASTPATH_NGPIO)) {
+			mask = fastpath;
+		} else {
+			mask = kmalloc_array(2 * BITS_TO_LONGS(chip->ngpio),
+					   sizeof(*mask),
+					   can_sleep ? GFP_KERNEL : GFP_ATOMIC);
+			if (!mask)
+				return -ENOMEM;
+		}
+
+		bits = mask + BITS_TO_LONGS(chip->ngpio);
+		memset(mask, 0, BITS_TO_LONGS(chip->ngpio) * sizeof(*mask));
+
 		if (!can_sleep)
 			WARN_ON(chip->can_sleep);
 
 		/* collect all inputs belonging to the same chip */
 		first = i;
-		memset(mask, 0, sizeof(mask));
 		do {
 			const struct gpio_desc *desc = desc_array[i];
 			int hwgpio = gpio_chip_hwgpio(desc);
@@ -2739,8 +2759,11 @@  int gpiod_get_array_value_complex(bool raw, bool can_sleep,
 			 (desc_array[i]->gdev->chip == chip));
 
 		ret = gpio_chip_get_multiple(chip, mask, bits);
-		if (ret)
+		if (ret) {
+			if (mask != fastpath)
+				kfree(mask);
 			return ret;
+		}
 
 		for (j = first; j < i; j++) {
 			const struct gpio_desc *desc = desc_array[j];
@@ -2752,6 +2775,9 @@  int gpiod_get_array_value_complex(bool raw, bool can_sleep,
 			value_array[j] = value;
 			trace_gpio_value(desc_to_gpio(desc), 1, value);
 		}
+
+		if (mask != fastpath)
+			kfree(mask);
 	}
 	return 0;
 }
@@ -2935,7 +2961,7 @@  static void gpio_chip_set_multiple(struct gpio_chip *chip,
 	}
 }
 
-void gpiod_set_array_value_complex(bool raw, bool can_sleep,
+int gpiod_set_array_value_complex(bool raw, bool can_sleep,
 				   unsigned int array_size,
 				   struct gpio_desc **desc_array,
 				   int *value_array)
@@ -2944,14 +2970,26 @@  void gpiod_set_array_value_complex(bool raw, bool can_sleep,
 
 	while (i < array_size) {
 		struct gpio_chip *chip = desc_array[i]->gdev->chip;
-		unsigned long mask[BITS_TO_LONGS(chip->ngpio)];
-		unsigned long bits[BITS_TO_LONGS(chip->ngpio)];
+		unsigned long fastpath[2 * BITS_TO_LONGS(FASTPATH_NGPIO)];
+		unsigned long *mask, *bits;
 		int count = 0;
 
+		if (likely(chip->ngpio <= FASTPATH_NGPIO)) {
+			mask = fastpath;
+		} else {
+			mask = kmalloc_array(2 * BITS_TO_LONGS(chip->ngpio),
+					   sizeof(*mask),
+					   can_sleep ? GFP_KERNEL : GFP_ATOMIC);
+			if (!mask)
+				return -ENOMEM;
+		}
+
+		bits = mask + BITS_TO_LONGS(chip->ngpio);
+		memset(mask, 0, BITS_TO_LONGS(chip->ngpio) * sizeof(*mask));
+
 		if (!can_sleep)
 			WARN_ON(chip->can_sleep);
 
-		memset(mask, 0, sizeof(mask));
 		do {
 			struct gpio_desc *desc = desc_array[i];
 			int hwgpio = gpio_chip_hwgpio(desc);
@@ -2982,7 +3020,11 @@  void gpiod_set_array_value_complex(bool raw, bool can_sleep,
 		/* push collected bits to outputs */
 		if (count != 0)
 			gpio_chip_set_multiple(chip, mask, bits);
+
+		if (mask != fastpath)
+			kfree(mask);
 	}
+	return 0;
 }
 
 /**
@@ -3057,13 +3099,13 @@  EXPORT_SYMBOL_GPL(gpiod_set_value);
  * This function should be called from contexts where we cannot sleep, and will
  * complain if the GPIO chip functions potentially sleep.
  */
-void gpiod_set_raw_array_value(unsigned int array_size,
+int gpiod_set_raw_array_value(unsigned int array_size,
 			 struct gpio_desc **desc_array, int *value_array)
 {
 	if (!desc_array)
-		return;
-	gpiod_set_array_value_complex(true, false, array_size, desc_array,
-				      value_array);
+		return -EINVAL;
+	return gpiod_set_array_value_complex(true, false, array_size,
+					desc_array, value_array);
 }
 EXPORT_SYMBOL_GPL(gpiod_set_raw_array_value);
 
@@ -3383,14 +3425,14 @@  EXPORT_SYMBOL_GPL(gpiod_set_value_cansleep);
  *
  * This function is to be called from contexts that can sleep.
  */
-void gpiod_set_raw_array_value_cansleep(unsigned int array_size,
+int gpiod_set_raw_array_value_cansleep(unsigned int array_size,
 					struct gpio_desc **desc_array,
 					int *value_array)
 {
 	might_sleep_if(extra_checks);
 	if (!desc_array)
-		return;
-	gpiod_set_array_value_complex(true, true, array_size, desc_array,
+		return -EINVAL;
+	return gpiod_set_array_value_complex(true, true, array_size, desc_array,
 				      value_array);
 }
 EXPORT_SYMBOL_GPL(gpiod_set_raw_array_value_cansleep);
diff --git a/drivers/gpio/gpiolib.h b/drivers/gpio/gpiolib.h
index ad456b6f9d8b..1a8e20363861 100644
--- a/drivers/gpio/gpiolib.h
+++ b/drivers/gpio/gpiolib.h
@@ -188,7 +188,7 @@  int gpiod_get_array_value_complex(bool raw, bool can_sleep,
 				  unsigned int array_size,
 				  struct gpio_desc **desc_array,
 				  int *value_array);
-void gpiod_set_array_value_complex(bool raw, bool can_sleep,
+int gpiod_set_array_value_complex(bool raw, bool can_sleep,
 				   unsigned int array_size,
 				   struct gpio_desc **desc_array,
 				   int *value_array);
diff --git a/include/linux/gpio/consumer.h b/include/linux/gpio/consumer.h
index dbd065963296..243112c7fa7d 100644
--- a/include/linux/gpio/consumer.h
+++ b/include/linux/gpio/consumer.h
@@ -116,7 +116,7 @@  int gpiod_get_raw_array_value(unsigned int array_size,
 			      struct gpio_desc **desc_array,
 			      int *value_array);
 void gpiod_set_raw_value(struct gpio_desc *desc, int value);
-void gpiod_set_raw_array_value(unsigned int array_size,
+int gpiod_set_raw_array_value(unsigned int array_size,
 			       struct gpio_desc **desc_array,
 			       int *value_array);
 
@@ -134,7 +134,7 @@  int gpiod_get_raw_array_value_cansleep(unsigned int array_size,
 				       struct gpio_desc **desc_array,
 				       int *value_array);
 void gpiod_set_raw_value_cansleep(struct gpio_desc *desc, int value);
-void gpiod_set_raw_array_value_cansleep(unsigned int array_size,
+int gpiod_set_raw_array_value_cansleep(unsigned int array_size,
 					struct gpio_desc **desc_array,
 					int *value_array);
 
@@ -369,12 +369,13 @@  static inline void gpiod_set_raw_value(struct gpio_desc *desc, int value)
 	/* GPIO can never have been requested */
 	WARN_ON(1);
 }
-static inline void gpiod_set_raw_array_value(unsigned int array_size,
+static inline int gpiod_set_raw_array_value(unsigned int array_size,
 					     struct gpio_desc **desc_array,
 					     int *value_array)
 {
 	/* GPIO can never have been requested */
 	WARN_ON(1);
+	return 0;
 }
 
 static inline int gpiod_get_value_cansleep(const struct gpio_desc *desc)
@@ -423,12 +424,13 @@  static inline void gpiod_set_raw_value_cansleep(struct gpio_desc *desc,
 	/* GPIO can never have been requested */
 	WARN_ON(1);
 }
-static inline void gpiod_set_raw_array_value_cansleep(unsigned int array_size,
+static inline int gpiod_set_raw_array_value_cansleep(unsigned int array_size,
 						struct gpio_desc **desc_array,
 						int *value_array)
 {
 	/* GPIO can never have been requested */
 	WARN_ON(1);
+	return 0;
 }
 
 static inline int gpiod_set_debounce(struct gpio_desc *desc, unsigned debounce)