[PATCH v2 0/5] GPIO driver for Maxim MAX3191x

[PATCH v2 0/5] GPIO driver for Maxim MAX3191x

[Lukas Wunner]

GPIO driver for Maxim MAX31910, MAX31911, MAX31912, MAX31913,
MAX31953 and MAX31963 industrial serializer, a daisy-chainable
chip to make 8 digital 24V inputs available via SPI.  Supports
CRC checksums to guard against electromagnetic interference,
as well as undervoltage and overtemperature detection.

The chip is used by the "Revolution Pi" family of open source PLCs
based on the Raspberry Pi (https://revolution.kunbus.com/).

In a typical SCADA system, all input signals are read periodically,
say, every 5 or 10 ms, and stored in a so-called "process image".
To make this perform well with serializers, add a ->get_multiple
callback to struct gpio_chip, add corresponding consumer functions
and wire it up with linehandle_ioctl().

Changes v1 -> v2:

- Patch [1/5]: Change the argument order of assign_bit() to reflect
  traditional "dst = src" in C. (Peter Zijlstra)

- Patch [2/5]: Update documentation. (Linus Walleij)
  Drop const qualifier from struct gpio_desc ** in all function
  signatures to avoid a cast when passing a non-const array created
  with gpiod_get_array(), which is likely the most common use case.

- Patch [3/5]: Newly inserted patch to introduce common property for
  number of daisy-chained devices.

- Patch [4/5]: Add vendor prefix to GPIO identifiers, use boolean
  instead of integer to select mode, rename boolean to ignore
  undervoltage alarms, separate compatible strings with newlines.
  (Rob Herring)

- Patch [5/5]: Optimize algorithm in max3191x_get_multiple() to
  iterate over the bits in the mask, instead of iterating over every
  chip, thus implicitly skipping chips which are not selected at all
  by the mask.
  Support configurations where all chips in a daisy-chain share the
  same modesel, fault or debounce pin.
  Verify that the number of db0 and db1 GPIOs specified in the DT is
  identical.

Link to v1:

https://www.spinics.net/lists/linux-gpio/msg25067.html

Thanks,

Lukas


Lukas Wunner (5):
  bitops: Introduce assign_bit()
  gpio: Introduce ->get_multiple callback
  dt-bindings: Document common property for daisy-chained devices
  dt-bindings: gpio: max3191x: Document new driver
  gpio: Add driver for Maxim MAX3191x industrial serializer

 .../devicetree/bindings/common-properties.txt      |  26 ++
 .../devicetree/bindings/gpio/gpio-max3191x.txt     |  59 +++
 Documentation/gpio/consumer.txt                    |  41 +-
 drivers/gpio/Kconfig                               |  10 +
 drivers/gpio/Makefile                              |   1 +
 drivers/gpio/gpio-max3191x.c                       | 492 +++++++++++++++++++++
 drivers/gpio/gpiolib.c                             | 179 +++++++-
 drivers/gpio/gpiolib.h                             |   4 +
 drivers/md/dm-mpath.c                              |  22 +-
 include/linux/bitops.h                             |  24 +
 include/linux/gpio/consumer.h                      |  43 ++
 include/linux/gpio/driver.h                        |   5 +
 12 files changed, 869 insertions(+), 37 deletions(-)
 create mode 100644 Documentation/devicetree/bindings/gpio/gpio-max3191x.txt
 create mode 100644 drivers/gpio/gpio-max3191x.c

-- 
2.11.0

[PATCH v2 1/5] bitops: Introduce assign_bit()

[Lukas Wunner]

A common idiom is to assign a value to a bit with:

    if (value)
        set_bit(nr, addr);
    else
        clear_bit(nr, addr);

Likewise common is the one-line expression variant:

    value ? set_bit(nr, addr) : clear_bit(nr, addr);

Commit 9a8ac3ae682e ("dm mpath: cleanup QUEUE_IF_NO_PATH bit
manipulation by introducing assign_bit()") introduced assign_bit()
to the md subsystem for brevity.

Make it available to others, specifically gpiolib and the upcoming
driver for Maxim MAX3191x industrial serializer chips.

As requested by Peter Zijlstra, change the argument order to reflect
traditional "dst = src" in C, hence "assign_bit(nr, addr, value)".

Cc: Bart Van Assche <bart.vanassche@wdc.com>
Cc: Alasdair Kergon <agk@redhat.com>
Cc: Mike Snitzer <snitzer@redhat.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Neil Brown <neilb@suse.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Borislav Petkov <bp@alien8.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
---
Changes v1 -> v2:

- Change the argument order of assign_bit() to reflect
  traditional "dst = src" in C. (Peter Zijlstra)

 drivers/md/dm-mpath.c  | 22 +++++++---------------
 include/linux/bitops.h | 24 ++++++++++++++++++++++++
 2 files changed, 31 insertions(+), 15 deletions(-)

diff --git a/drivers/md/dm-mpath.c b/drivers/md/dm-mpath.c
index 11f273d2f018..0e211de9bc54 100644
--- a/drivers/md/dm-mpath.c
+++ b/drivers/md/dm-mpath.c
@@ -641,14 +641,6 @@ static void process_queued_bios(struct work_struct *work)
 	blk_finish_plug(&plug);
 }
 
-static void assign_bit(bool value, long nr, unsigned long *addr)
-{
-	if (value)
-		set_bit(nr, addr);
-	else
-		clear_bit(nr, addr);
-}
-
 /*
  * If we run out of usable paths, should we queue I/O or error it?
  */
@@ -658,11 +650,11 @@ static int queue_if_no_path(struct multipath *m, bool queue_if_no_path,
 	unsigned long flags;
 
 	spin_lock_irqsave(&m->lock, flags);
-	assign_bit((save_old_value && test_bit(MPATHF_QUEUE_IF_NO_PATH, &m->flags)) ||
-		   (!save_old_value && queue_if_no_path),
-		   MPATHF_SAVED_QUEUE_IF_NO_PATH, &m->flags);
-	assign_bit(queue_if_no_path || dm_noflush_suspending(m->ti),
-		   MPATHF_QUEUE_IF_NO_PATH, &m->flags);
+	assign_bit(MPATHF_SAVED_QUEUE_IF_NO_PATH, &m->flags,
+		   (save_old_value && test_bit(MPATHF_QUEUE_IF_NO_PATH, &m->flags)) ||
+		   (!save_old_value && queue_if_no_path));
+	assign_bit(MPATHF_QUEUE_IF_NO_PATH, &m->flags,
+		   queue_if_no_path || dm_noflush_suspending(m->ti));
 	spin_unlock_irqrestore(&m->lock, flags);
 
 	if (!queue_if_no_path) {
@@ -1588,8 +1580,8 @@ static void multipath_resume(struct dm_target *ti)
 	unsigned long flags;
 
 	spin_lock_irqsave(&m->lock, flags);
-	assign_bit(test_bit(MPATHF_SAVED_QUEUE_IF_NO_PATH, &m->flags),
-		   MPATHF_QUEUE_IF_NO_PATH, &m->flags);
+	assign_bit(MPATHF_QUEUE_IF_NO_PATH, &m->flags,
+		   test_bit(MPATHF_SAVED_QUEUE_IF_NO_PATH, &m->flags));
 	spin_unlock_irqrestore(&m->lock, flags);
 }
 
diff --git a/include/linux/bitops.h b/include/linux/bitops.h
index 8fbe259b197c..9a874deee6e2 100644
--- a/include/linux/bitops.h
+++ b/include/linux/bitops.h
@@ -227,6 +227,30 @@ static inline unsigned long __ffs64(u64 word)
 	return __ffs((unsigned long)word);
 }
 
+/**
+ * assign_bit - Assign value to a bit in memory
+ * @nr: the bit to set
+ * @addr: the address to start counting from
+ * @value: the value to assign
+ */
+static __always_inline void assign_bit(long nr, volatile unsigned long *addr,
+				       bool value)
+{
+	if (value)
+		set_bit(nr, addr);
+	else
+		clear_bit(nr, addr);
+}
+
+static __always_inline void __assign_bit(long nr, volatile unsigned long *addr,
+					 bool value)
+{
+	if (value)
+		__set_bit(nr, addr);
+	else
+		__clear_bit(nr, addr);
+}
+
 #ifdef __KERNEL__
 
 #ifndef set_mask_bits
-- 
2.11.0

[PATCH v2 2/5] gpio: Introduce ->get_multiple callback

[Lukas Wunner]

SPI-attached GPIO controllers typically read out all inputs in one go.
If callers desire the values of multipe inputs, ideally a single readout
should take place to return the desired values.  However the current
driver API only offers a ->get callback but no ->get_multiple (unlike
->set_multiple, which is present).  Thus, to read multiple inputs, a
full readout needs to be performed for every single value (barring
driver-internal caching), which is inefficient.

In fact, the lack of a ->get_multiple callback has been bemoaned
repeatedly by the gpio subsystem maintainer:
http://www.spinics.net/lists/linux-gpio/msg10571.html
http://www.spinics.net/lists/devicetree/msg121734.html

Introduce the missing callback.  Add corresponding consumer functions
such as gpiod_get_array_value().  Amend linehandle_ioctl() to take
advantage of the newly added infrastructure.  Update the documentation.

Cc: Rojhalat Ibrahim <imr@rtschenk.de>
Signed-off-by: Lukas Wunner <lukas@wunner.de>
---
Changes v1 -> v2:

- Update documentation. (Linus Walleij)
  Drop const qualifier from struct gpio_desc ** in all function
  signatures to avoid a cast when passing a non-const array created
  with gpiod_get_array(), which is likely the most common use case.

 Documentation/gpio/consumer.txt |  41 ++++++---
 drivers/gpio/gpiolib.c          | 179 +++++++++++++++++++++++++++++++++++++---
 drivers/gpio/gpiolib.h          |   4 +
 include/linux/gpio/consumer.h   |  43 ++++++++++
 include/linux/gpio/driver.h     |   5 ++
 5 files changed, 250 insertions(+), 22 deletions(-)

diff --git a/Documentation/gpio/consumer.txt b/Documentation/gpio/consumer.txt
index ddbfa775a78a..63e1bd1d88e3 100644
--- a/Documentation/gpio/consumer.txt
+++ b/Documentation/gpio/consumer.txt
@@ -295,9 +295,22 @@ as possible, especially by drivers which should not care about the actual
 physical line level and worry about the logical value instead.
 
 
-Set multiple GPIO outputs with a single function call
------------------------------------------------------
-The following functions set the output values of an array of GPIOs:
+Access multiple GPIOs with a single function call
+-------------------------------------------------
+The following functions get or set the values of an array of GPIOs:
+
+	int gpiod_get_array_value(unsigned int array_size,
+				  struct gpio_desc **desc_array,
+				  int *value_array);
+	int gpiod_get_raw_array_value(unsigned int array_size,
+				      struct gpio_desc **desc_array,
+				      int *value_array);
+	int gpiod_get_array_value_cansleep(unsigned int array_size,
+					   struct gpio_desc **desc_array,
+					   int *value_array);
+	int gpiod_get_raw_array_value_cansleep(unsigned int array_size,
+					   struct gpio_desc **desc_array,
+					   int *value_array);
 
 	void gpiod_set_array_value(unsigned int array_size,
 				   struct gpio_desc **desc_array,
@@ -312,34 +325,40 @@ The following functions set the output values of an array of GPIOs:
 						struct gpio_desc **desc_array,
 						int *value_array)
 
-The array can be an arbitrary set of GPIOs. The functions will try to set
+The array can be an arbitrary set of GPIOs. The functions will try to access
 GPIOs belonging to the same bank or chip simultaneously if supported by the
 corresponding chip driver. In that case a significantly improved performance
-can be expected. If simultaneous setting is not possible the GPIOs will be set
-sequentially.
+can be expected. If simultaneous access is not possible the GPIOs will be
+accessed sequentially.
 
-The gpiod_set_array() functions take three arguments:
+The functions take three arguments:
 	* array_size	- the number of array elements
 	* desc_array	- an array of GPIO descriptors
-	* value_array	- an array of values to assign to the GPIOs
+	* value_array	- an array to store the GPIOs' values (get) or
+			  an array of values to assign to the GPIOs (set)
 
 The descriptor array can be obtained using the gpiod_get_array() function
 or one of its variants. If the group of descriptors returned by that function
-matches the desired group of GPIOs, those GPIOs can be set by simply using
+matches the desired group of GPIOs, those GPIOs can be accessed by simply using
 the struct gpio_descs returned by gpiod_get_array():
 
 	struct gpio_descs *my_gpio_descs = gpiod_get_array(...);
 	gpiod_set_array_value(my_gpio_descs->ndescs, my_gpio_descs->desc,
 			      my_gpio_values);
 
-It is also possible to set a completely arbitrary array of descriptors. The
+It is also possible to access a completely arbitrary array of descriptors. The
 descriptors may be obtained using any combination of gpiod_get() and
 gpiod_get_array(). Afterwards the array of descriptors has to be setup
-manually before it can be used with gpiod_set_array().
+manually before it can be passed to one of the above functions.
 
 Note that for optimal performance GPIOs belonging to the same chip should be
 contiguous within the array of descriptors.
 
+The return value of gpiod_get_array_value() and its variants is 0 on success
+or negative on error. Note the difference to gpiod_get_value(), which returns
+0 or 1 on success to convey the GPIO value. With the array functions, the GPIO
+values are stored in value_array rather than passed back as return value.
+
 
 GPIOs mapped to IRQs
 --------------------
diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
index 8113929eb2bd..ff46a1b6299e 100644
--- a/drivers/gpio/gpiolib.c
+++ b/drivers/gpio/gpiolib.c
@@ -365,28 +365,28 @@ static long linehandle_ioctl(struct file *filep, unsigned int cmd,
 	struct linehandle_state *lh = filep->private_data;
 	void __user *ip = (void __user *)arg;
 	struct gpiohandle_data ghd;
+	int vals[GPIOHANDLES_MAX];
 	int i;
 
 	if (cmd == GPIOHANDLE_GET_LINE_VALUES_IOCTL) {
-		int val;
+		/* TODO: check if descriptors are really input */
+		int ret = gpiod_get_array_value_complex(false,
+							true,
+							lh->numdescs,
+							lh->descs,
+							vals);
+		if (ret)
+			return ret;
 
 		memset(&ghd, 0, sizeof(ghd));
-
-		/* TODO: check if descriptors are really input */
-		for (i = 0; i < lh->numdescs; i++) {
-			val = gpiod_get_value_cansleep(lh->descs[i]);
-			if (val < 0)
-				return val;
-			ghd.values[i] = val;
-		}
+		for (i = 0; i < lh->numdescs; i++)
+			ghd.values[i] = vals[i];
 
 		if (copy_to_user(ip, &ghd, sizeof(ghd)))
 			return -EFAULT;
 
 		return 0;
 	} else if (cmd == GPIOHANDLE_SET_LINE_VALUES_IOCTL) {
-		int vals[GPIOHANDLES_MAX];
-
 		/* TODO: check if descriptors are really output */
 		if (copy_from_user(&ghd, ip, sizeof(ghd)))
 			return -EFAULT;
@@ -2466,6 +2466,71 @@ static int gpiod_get_raw_value_commit(const struct gpio_desc *desc)
 	return value;
 }
 
+static int gpio_chip_get_multiple(struct gpio_chip *chip,
+				  unsigned long *mask, unsigned long *bits)
+{
+	if (chip->get_multiple) {
+		return chip->get_multiple(chip, mask, bits);
+	} else if (chip->get) {
+		int i, value;
+
+		for_each_set_bit(i, mask, chip->ngpio) {
+			value = chip->get(chip, i);
+			if (value < 0)
+				return value;
+			__assign_bit(i, bits, value);
+		}
+		return 0;
+	}
+	return -EIO;
+}
+
+int gpiod_get_array_value_complex(bool raw, bool can_sleep,
+				  unsigned int array_size,
+				  struct gpio_desc **desc_array,
+				  int *value_array)
+{
+	int i = 0;
+
+	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)];
+		int first, j, ret;
+
+		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);
+
+			__set_bit(hwgpio, mask);
+			i++;
+		} while ((i < array_size) &&
+			 (desc_array[i]->gdev->chip == chip));
+
+		ret = gpio_chip_get_multiple(chip, mask, bits);
+		if (ret)
+			return ret;
+
+		for (j = first; j < i; j++) {
+			const struct gpio_desc *desc = desc_array[j];
+			int hwgpio = gpio_chip_hwgpio(desc);
+			int value = test_bit(hwgpio, bits);
+
+			if (!raw && test_bit(FLAG_ACTIVE_LOW, &desc->flags))
+				value = !value;
+			value_array[j] = value;
+			trace_gpio_value(desc_to_gpio(desc), 1, value);
+		}
+	}
+	return 0;
+}
+
 /**
  * gpiod_get_raw_value() - return a gpio's raw value
  * @desc: gpio whose value will be returned
@@ -2514,6 +2579,51 @@ int gpiod_get_value(const struct gpio_desc *desc)
 }
 EXPORT_SYMBOL_GPL(gpiod_get_value);
 
+/**
+ * gpiod_get_raw_array_value() - read raw values from an array of GPIOs
+ * @array_size: number of elements in the descriptor / value arrays
+ * @desc_array: array of GPIO descriptors whose values will be read
+ * @value_array: array to store the read values
+ *
+ * Read the raw values of the GPIOs, i.e. the values of the physical lines
+ * without regard for their ACTIVE_LOW status.  Return 0 in case of success,
+ * else an error code.
+ *
+ * This function should be called from contexts where we cannot sleep,
+ * and it will complain if the GPIO chip functions potentially sleep.
+ */
+int gpiod_get_raw_array_value(unsigned int array_size,
+			      struct gpio_desc **desc_array, int *value_array)
+{
+	if (!desc_array)
+		return -EINVAL;
+	return gpiod_get_array_value_complex(true, false, array_size,
+					     desc_array, value_array);
+}
+EXPORT_SYMBOL_GPL(gpiod_get_raw_array_value);
+
+/**
+ * gpiod_get_array_value() - read values from an array of GPIOs
+ * @array_size: number of elements in the descriptor / value arrays
+ * @desc_array: array of GPIO descriptors whose values will be read
+ * @value_array: array to store the read values
+ *
+ * Read the logical values of the GPIOs, i.e. taking their ACTIVE_LOW status
+ * into account.  Return 0 in case of success, else an error code.
+ *
+ * This function should be called from contexts where we cannot sleep,
+ * and it will complain if the GPIO chip functions potentially sleep.
+ */
+int gpiod_get_array_value(unsigned int array_size,
+			  struct gpio_desc **desc_array, int *value_array)
+{
+	if (!desc_array)
+		return -EINVAL;
+	return gpiod_get_array_value_complex(false, false, array_size,
+					     desc_array, value_array);
+}
+EXPORT_SYMBOL_GPL(gpiod_get_array_value);
+
 /*
  *  gpio_set_open_drain_value_commit() - Set the open drain gpio's value.
  * @desc: gpio descriptor whose state need to be set.
@@ -2943,6 +3053,53 @@ int gpiod_get_value_cansleep(const struct gpio_desc *desc)
 EXPORT_SYMBOL_GPL(gpiod_get_value_cansleep);
 
 /**
+ * gpiod_get_raw_array_value_cansleep() - read raw values from an array of GPIOs
+ * @array_size: number of elements in the descriptor / value arrays
+ * @desc_array: array of GPIO descriptors whose values will be read
+ * @value_array: array to store the read values
+ *
+ * Read the raw values of the GPIOs, i.e. the values of the physical lines
+ * without regard for their ACTIVE_LOW status.  Return 0 in case of success,
+ * else an error code.
+ *
+ * This function is to be called from contexts that can sleep.
+ */
+int gpiod_get_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 -EINVAL;
+	return gpiod_get_array_value_complex(true, true, array_size,
+					     desc_array, value_array);
+}
+EXPORT_SYMBOL_GPL(gpiod_get_raw_array_value_cansleep);
+
+/**
+ * gpiod_get_array_value_cansleep() - read values from an array of GPIOs
+ * @array_size: number of elements in the descriptor / value arrays
+ * @desc_array: array of GPIO descriptors whose values will be read
+ * @value_array: array to store the read values
+ *
+ * Read the logical values of the GPIOs, i.e. taking their ACTIVE_LOW status
+ * into account.  Return 0 in case of success, else an error code.
+ *
+ * This function is to be called from contexts that can sleep.
+ */
+int gpiod_get_array_value_cansleep(unsigned int array_size,
+				   struct gpio_desc **desc_array,
+				   int *value_array)
+{
+	might_sleep_if(extra_checks);
+	if (!desc_array)
+		return -EINVAL;
+	return gpiod_get_array_value_complex(false, true, array_size,
+					     desc_array, value_array);
+}
+EXPORT_SYMBOL_GPL(gpiod_get_array_value_cansleep);
+
+/**
  * gpiod_set_raw_value_cansleep() - assign a gpio's raw value
  * @desc: gpio whose value will be assigned
  * @value: value to assign
diff --git a/drivers/gpio/gpiolib.h b/drivers/gpio/gpiolib.h
index d003ccb12781..10a48caf8477 100644
--- a/drivers/gpio/gpiolib.h
+++ b/drivers/gpio/gpiolib.h
@@ -180,6 +180,10 @@ static inline bool acpi_can_fallback_to_crs(struct acpi_device *adev,
 #endif
 
 struct gpio_desc *gpiochip_get_desc(struct gpio_chip *chip, u16 hwnum);
+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,
 				   unsigned int array_size,
 				   struct gpio_desc **desc_array,
diff --git a/include/linux/gpio/consumer.h b/include/linux/gpio/consumer.h
index 8f702fcbe485..d4920ec1f1da 100644
--- a/include/linux/gpio/consumer.h
+++ b/include/linux/gpio/consumer.h
@@ -99,10 +99,15 @@ int gpiod_direction_output_raw(struct gpio_desc *desc, int value);
 
 /* Value get/set from non-sleeping context */
 int gpiod_get_value(const struct gpio_desc *desc);
+int gpiod_get_array_value(unsigned int array_size,
+			  struct gpio_desc **desc_array, int *value_array);
 void gpiod_set_value(struct gpio_desc *desc, int value);
 void gpiod_set_array_value(unsigned int array_size,
 			   struct gpio_desc **desc_array, int *value_array);
 int gpiod_get_raw_value(const struct gpio_desc *desc);
+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,
 			       struct gpio_desc **desc_array,
@@ -110,11 +115,17 @@ void gpiod_set_raw_array_value(unsigned int array_size,
 
 /* Value get/set from sleeping context */
 int gpiod_get_value_cansleep(const struct gpio_desc *desc);
+int gpiod_get_array_value_cansleep(unsigned int array_size,
+				   struct gpio_desc **desc_array,
+				   int *value_array);
 void gpiod_set_value_cansleep(struct gpio_desc *desc, int value);
 void gpiod_set_array_value_cansleep(unsigned int array_size,
 				    struct gpio_desc **desc_array,
 				    int *value_array);
 int gpiod_get_raw_value_cansleep(const struct gpio_desc *desc);
+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,
 					struct gpio_desc **desc_array,
@@ -305,6 +316,14 @@ static inline int gpiod_get_value(const struct gpio_desc *desc)
 	WARN_ON(1);
 	return 0;
 }
+static inline int gpiod_get_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 void gpiod_set_value(struct gpio_desc *desc, int value)
 {
 	/* GPIO can never have been requested */
@@ -323,6 +342,14 @@ static inline int gpiod_get_raw_value(const struct gpio_desc *desc)
 	WARN_ON(1);
 	return 0;
 }
+static inline int gpiod_get_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 void gpiod_set_raw_value(struct gpio_desc *desc, int value)
 {
 	/* GPIO can never have been requested */
@@ -342,6 +369,14 @@ static inline int gpiod_get_value_cansleep(const struct gpio_desc *desc)
 	WARN_ON(1);
 	return 0;
 }
+static inline int gpiod_get_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 void gpiod_set_value_cansleep(struct gpio_desc *desc, int value)
 {
 	/* GPIO can never have been requested */
@@ -360,6 +395,14 @@ static inline int gpiod_get_raw_value_cansleep(const struct gpio_desc *desc)
 	WARN_ON(1);
 	return 0;
 }
+static inline int gpiod_get_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 void gpiod_set_raw_value_cansleep(struct gpio_desc *desc,
 						int value)
 {
diff --git a/include/linux/gpio/driver.h b/include/linux/gpio/driver.h
index 1de5dd185118..5fe533408b3a 100644
--- a/include/linux/gpio/driver.h
+++ b/include/linux/gpio/driver.h
@@ -35,6 +35,8 @@ struct module;
  * @direction_input: configures signal "offset" as input, or returns error
  * @direction_output: configures signal "offset" as output, or returns error
  * @get: returns value for signal "offset", 0=low, 1=high, or negative error
+ * @get_multiple: reads values for multiple signals defined by "mask" and
+ *	stores them in "bits", returns 0 on success or negative error
  * @set: assigns output value for signal "offset"
  * @set_multiple: assigns output values for multiple signals defined by "mask"
  * @set_config: optional hook for all kinds of settings. Uses the same
@@ -126,6 +128,9 @@ struct gpio_chip {
 						unsigned offset, int value);
 	int			(*get)(struct gpio_chip *chip,
 						unsigned offset);
+	int			(*get_multiple)(struct gpio_chip *chip,
+						unsigned long *mask,
+						unsigned long *bits);
 	void			(*set)(struct gpio_chip *chip,
 						unsigned offset, int value);
 	void			(*set_multiple)(struct gpio_chip *chip,
-- 
2.11.0

[PATCH v2 3/5] dt-bindings: Document common property for daisy-chained devices

[Lukas Wunner]

Many serially-attached GPIO and IIO devices are daisy-chainable.

    Examples for GPIO devices are Maxim MAX3191x and TI SN65HVS88x:
    https://datasheets.maximintegrated.com/en/ds/MAX31913.pdf
    http://www.ti.com/lit/ds/symlink/sn65hvs880.pdf

    Examples for IIO devices are TI DAC128S085 and TI DAC161S055:
    http://www.ti.com/lit/ds/symlink/dac128s085.pdf
    http://www.ti.com/lit/ds/symlink/dac161s055.pdf

We already have drivers for daisy-chainable devices in the tree but
their devicetree bindings are somewhat inconsistent and ill-named:

    The gpio-74x164.c driver uses "registers-number" to convey the
    number of devices in the daisy-chain.  (Sans vendor prefix,
    multiple vendors sell compatible versions of this chip.)

    The gpio-pisosr.c driver takes a different approach and calculates
    the number of devices in the daisy-chain by dividing the common
    "ngpios" property (Documentation/devicetree/bindings/gpio/gpio.txt)
    by 8 (which assumes that each chip has 8 inputs).

Let's standardize on a common "#daisy-chained-devices" property.
That name was chosen because it's the term most frequently used in
datasheets.  (A less frequently used synonym is "cascaded devices".)

Signed-off-by: Lukas Wunner <lukas@wunner.de>
---
Changes v1 -> v2:

- Newly inserted patch.

 .../devicetree/bindings/common-properties.txt      | 26 ++++++++++++++++++++++
 1 file changed, 26 insertions(+)

diff --git a/Documentation/devicetree/bindings/common-properties.txt b/Documentation/devicetree/bindings/common-properties.txt
index 697714f8d75c..a3448bfa1c82 100644
--- a/Documentation/devicetree/bindings/common-properties.txt
+++ b/Documentation/devicetree/bindings/common-properties.txt
@@ -1,4 +1,8 @@
 Common properties
+=================
+
+Endianness
+----------
 
 The Devicetree Specification does not define any properties related to hardware
 byteswapping, but endianness issues show up frequently in porting Linux to
@@ -58,3 +62,25 @@ dev: dev@40031000 {
 	      ...
 	      little-endian;
 };
+
+Daisy-chained devices
+---------------------
+
+Many serially-attached GPIO and IIO devices are daisy-chainable.  To the
+host controller, a daisy-chain appears as a single device, but the number
+of inputs and outputs it provides is the sum of inputs and outputs provided
+by all of its devices.  The driver needs to know how many devices the
+daisy-chain comprises to determine the amount of data exchanged, how many
+inputs and outputs to register and so on.
+
+Optional properties:
+ - #daisy-chained-devices: Number of devices in the daisy-chain (default is 1).
+
+Example:
+gpio@0 {
+	      compatible = "name";
+	      reg = <0>;
+	      gpio-controller;
+	      #gpio-cells = <2>;
+	      #daisy-chained-devices = <3>;
+};
-- 
2.11.0

[PATCH v2 4/5] dt-bindings: gpio: max3191x: Document new driver

[Lukas Wunner]

Add device tree bindings for Maxim MAX3191x industrial serializer.

Cc: Mathias Duckeck <m.duckeck-XB/JSsFECOqzQB+pC5nmwQ@public.gmane.org>
Signed-off-by: Lukas Wunner <lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org>
---
Changes v1 -> v2:

- Add vendor prefix to GPIO identifiers, use boolean instead of
  integer to select mode, rename boolean to ignore undervoltage
  alarms, separate compatible strings with newlines. (Rob Herring)

 .../devicetree/bindings/gpio/gpio-max3191x.txt     | 59 ++++++++++++++++++++++
 1 file changed, 59 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/gpio/gpio-max3191x.txt

diff --git a/Documentation/devicetree/bindings/gpio/gpio-max3191x.txt b/Documentation/devicetree/bindings/gpio/gpio-max3191x.txt
new file mode 100644
index 000000000000..b3a6444b8f45
--- /dev/null
+++ b/Documentation/devicetree/bindings/gpio/gpio-max3191x.txt
@@ -0,0 +1,59 @@
+GPIO driver for Maxim MAX3191x industrial serializer
+
+Required properties:
+ - compatible:		Must be one of:
+			"maxim,max31910"
+			"maxim,max31911"
+			"maxim,max31912"
+			"maxim,max31913"
+			"maxim,max31953"
+			"maxim,max31963"
+ - reg: 		Chip select number.
+ - gpio-controller:	Marks the device node as a GPIO controller.
+ - #gpio-cells: 	Should be two. For consumer use see gpio.txt.
+
+Optional properties:
+ - #daisy-chained-devices:
+			Number of chips in the daisy-chain (default is 1).
+ - maxim,modesel-gpios: GPIO pins to configure modesel of each chip.
+			The number of GPIOs must equal "#daisy-chained-devices"
+			(if each chip is driven by a separate pin) or 1
+			(if all chips are wired to the same pin).
+ - maxim,fault-gpios: 	GPIO pins to read fault of each chip.
+			The number of GPIOs must equal "#daisy-chained-devices"
+			or 1.
+ - maxim,db0-gpios:	GPIO pins to configure debounce of each chip.
+			The number of GPIOs must equal "#daisy-chained-devices"
+			or 1.
+ - maxim,db1-gpios:	GPIO pins to configure debounce of each chip.
+			The number of GPIOs must equal "maxim,db0-gpios".
+ - maxim,modesel-8bit:	Boolean whether the modesel pin of the chips is
+			pulled high (8-bit mode).  Use this if the modesel pin
+			is hardwired and consequently "maxim,modesel-gpios"
+			cannot be specified.  By default if neither this nor
+			"maxim,modesel-gpios" is given, the driver assumes
+			that modesel is pulled low (16-bit mode).
+ - maxim,ignore-undervoltage:
+			Boolean whether to ignore undervoltage alarms signaled
+			by the "maxim,fault-gpios" or by the status byte
+			(in 16-bit mode).  Use this if the chips are powered
+			through 5VOUT instead of VCC24V, in which case they
+			will constantly signal undervoltage.
+
+For other required and optional properties of SPI slave nodes please refer to
+../spi/spi-bus.txt.
+
+Example:
+	gpio@0 {
+		compatible = "maxim,max31913";
+		reg = <0>;
+		gpio-controller;
+		#gpio-cells = <2>;
+
+		maxim,modesel-gpios = <&gpio2 23>;
+		maxim,fault-gpios   = <&gpio2 24 GPIO_ACTIVE_LOW>;
+		maxim,db0-gpios     = <&gpio2 25>;
+		maxim,db1-gpios     = <&gpio2 26>;
+
+		spi-max-frequency = <25000000>;
+	};
-- 
2.11.0

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH v2 5/5] gpio: Add driver for Maxim MAX3191x industrial serializer

[Lukas Wunner]

The driver was developed for and tested with the MAX31913 built into
the Revolution Pi by KUNBUS, but should work with all members of the
MAX3191x family:

MAX31910: low power
MAX31911: LED drivers
MAX31912: LED drivers + 2nd voltage monitor + low power
MAX31913: LED drivers + 2nd voltage monitor
MAX31953: LED drivers + 2nd voltage monitor + isolation
MAX31963: LED drivers + 2nd voltage monitor + isolation + buck regulator

Cc: Mathias Duckeck <m.duckeck-XB/JSsFECOqzQB+pC5nmwQ@public.gmane.org>
Signed-off-by: Lukas Wunner <lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org>
---
Changes v1 -> v2:

- Optimize algorithm in max3191x_get_multiple() to iterate over the
  bits in the mask, instead of iterating over every chip, thus
  implicitly skipping chips which are not selected at all by the mask.
  Support configurations where all chips in a daisy-chain share the
  same modesel, fault or debounce pin.
  Verify that the number of db0 and db1 GPIOs specified in the DT is
  identical.

 drivers/gpio/Kconfig         |  10 +
 drivers/gpio/Makefile        |   1 +
 drivers/gpio/gpio-max3191x.c | 492 +++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 503 insertions(+)
 create mode 100644 drivers/gpio/gpio-max3191x.c

diff --git a/drivers/gpio/Kconfig b/drivers/gpio/Kconfig
index 3388d54ba114..d5282cac6ea6 100644
--- a/drivers/gpio/Kconfig
+++ b/drivers/gpio/Kconfig
@@ -1255,6 +1255,16 @@ config GPIO_74X164
 	  shift registers. This driver can be used to provide access
 	  to more gpio outputs.
 
+config GPIO_MAX3191X
+	tristate "Maxim MAX3191x industrial serializer"
+	select CRC8
+	help
+	  GPIO driver for Maxim MAX31910, MAX31911, MAX31912, MAX31913,
+	  MAX31953 and MAX31963 industrial serializer, a daisy-chainable
+	  chip to make 8 digital 24V inputs available via SPI.  Supports
+	  CRC checksums to guard against electromagnetic interference,
+	  as well as undervoltage and overtemperature detection.
+
 config GPIO_MAX7301
 	tristate "Maxim MAX7301 GPIO expander"
 	select GPIO_MAX730X
diff --git a/drivers/gpio/Makefile b/drivers/gpio/Makefile
index aeb70e9de6f2..cf9ecfb24ef4 100644
--- a/drivers/gpio/Makefile
+++ b/drivers/gpio/Makefile
@@ -69,6 +69,7 @@ obj-$(CONFIG_ARCH_LPC32XX)	+= gpio-lpc32xx.o
 obj-$(CONFIG_GPIO_LP873X)	+= gpio-lp873x.o
 obj-$(CONFIG_GPIO_LP87565)	+= gpio-lp87565.o
 obj-$(CONFIG_GPIO_LYNXPOINT)	+= gpio-lynxpoint.o
+obj-$(CONFIG_GPIO_MAX3191X)	+= gpio-max3191x.o
 obj-$(CONFIG_GPIO_MAX730X)	+= gpio-max730x.o
 obj-$(CONFIG_GPIO_MAX7300)	+= gpio-max7300.o
 obj-$(CONFIG_GPIO_MAX7301)	+= gpio-max7301.o
diff --git a/drivers/gpio/gpio-max3191x.c b/drivers/gpio/gpio-max3191x.c
new file mode 100644
index 000000000000..f74b1072e84b
--- /dev/null
+++ b/drivers/gpio/gpio-max3191x.c
@@ -0,0 +1,492 @@
+/*
+ * gpio-max3191x.c - GPIO driver for Maxim MAX3191x industrial serializer
+ *
+ * Copyright (C) 2017 KUNBUS GmbH
+ *
+ * The MAX3191x makes 8 digital 24V inputs available via SPI.
+ * Multiple chips can be daisy-chained, the spec does not impose
+ * a limit on the number of chips and neither does this driver.
+ *
+ * Either of two modes is selectable: In 8-bit mode, only the state
+ * of the inputs is clocked out to achieve high readout speeds;
+ * In 16-bit mode, an additional status byte is clocked out with
+ * a CRC and indicator bits for undervoltage and overtemperature.
+ * The driver returns an error instead of potentially bogus data
+ * if any of these fault conditions occur.  However it does allow
+ * readout of non-faulting chips in the same daisy-chain.
+ *
+ * MAX3191x supports four debounce settings and the driver is
+ * capable of configuring these differently for each chip in the
+ * daisy-chain.
+ *
+ * If the chips are hardwired to 8-bit mode ("modesel" pulled high),
+ * gpio-pisosr.c can be used alternatively to this driver.
+ *
+ * https://datasheets.maximintegrated.com/en/ds/MAX31910.pdf
+ * https://datasheets.maximintegrated.com/en/ds/MAX31911.pdf
+ * https://datasheets.maximintegrated.com/en/ds/MAX31912.pdf
+ * https://datasheets.maximintegrated.com/en/ds/MAX31913.pdf
+ * https://datasheets.maximintegrated.com/en/ds/MAX31953-MAX31963.pdf
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License (version 2) as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/bitmap.h>
+#include <linux/crc8.h>
+#include <linux/gpio/consumer.h>
+#include <linux/gpio/driver.h>
+#include <linux/module.h>
+#include <linux/spi/spi.h>
+
+enum max3191x_mode {
+	STATUS_BYTE_ENABLED,
+	STATUS_BYTE_DISABLED,
+};
+
+/**
+ * struct max3191x_chip - max3191x daisy-chain
+ * @gpio: GPIO controller struct
+ * @lock: protects read sequences
+ * @nchips: number of chips in the daisy-chain
+ * @mode: current mode, 0 for 16-bit, 1 for 8-bit;
+ *	for simplicity, all chips in the daisy-chain are assumed
+ *	to use the same mode
+ * @modesel_pins: GPIO pins to configure modesel of each chip
+ * @fault_pins: GPIO pins to detect fault of each chip
+ * @db0_pins: GPIO pins to configure debounce of each chip
+ * @db1_pins: GPIO pins to configure debounce of each chip
+ * @mesg: SPI message to perform a readout
+ * @xfer: SPI transfer used by @mesg
+ * @crc_error: bitmap signaling CRC error for each chip
+ * @overtemp: bitmap signaling overtemperature alarm for each chip
+ * @undervolt1: bitmap signaling undervoltage alarm for each chip
+ * @undervolt2: bitmap signaling undervoltage warning for each chip
+ * @fault: bitmap signaling assertion of @fault_pins for each chip
+ * @ignore_uv: whether to ignore undervoltage alarms;
+ *	set by a device property if the chips are powered through
+ *	5VOUT instead of VCC24V, in which case they will constantly
+ *	signal undervoltage;
+ *	for simplicity, all chips in the daisy-chain are assumed
+ *	to be powered the same way
+ */
+struct max3191x_chip {
+	struct gpio_chip gpio;
+	struct mutex lock;
+	u32 nchips;
+	enum max3191x_mode mode;
+	struct gpio_descs *modesel_pins;
+	struct gpio_descs *fault_pins;
+	struct gpio_descs *db0_pins;
+	struct gpio_descs *db1_pins;
+	struct spi_message mesg;
+	struct spi_transfer xfer;
+	unsigned long *crc_error;
+	unsigned long *overtemp;
+	unsigned long *undervolt1;
+	unsigned long *undervolt2;
+	unsigned long *fault;
+	bool ignore_uv;
+};
+
+#define MAX3191X_NGPIO 8
+#define MAX3191X_CRC8_POLYNOMIAL 0xa8 /* (x^5) + x^4 + x^2 + x^0 */
+
+DECLARE_CRC8_TABLE(max3191x_crc8);
+
+static int max3191x_get_direction(struct gpio_chip *gpio, unsigned int offset)
+{
+	return 1; /* always in */
+}
+
+static int max3191x_direction_input(struct gpio_chip *gpio, unsigned int offset)
+{
+	return 0;
+}
+
+static int max3191x_direction_output(struct gpio_chip *gpio,
+				     unsigned int offset, int value)
+{
+	return -EINVAL;
+}
+
+static void max3191x_set(struct gpio_chip *gpio, unsigned int offset, int value)
+{ }
+
+static void max3191x_set_multiple(struct gpio_chip *gpio, unsigned long *mask,
+				  unsigned long *bits)
+{ }
+
+static unsigned int max3191x_wordlen(struct max3191x_chip *max3191x)
+{
+	return max3191x->mode == STATUS_BYTE_ENABLED ? 2 : 1;
+}
+
+static int max3191x_readout_locked(struct max3191x_chip *max3191x)
+{
+	struct device *dev = max3191x->gpio.parent;
+	struct spi_device *spi = to_spi_device(dev);
+	int val, i, ot = 0, uv1 = 0;
+
+	val = spi_sync(spi, &max3191x->mesg);
+	if (val) {
+		dev_err_ratelimited(dev, "SPI receive error %d\n", val);
+		return val;
+	}
+
+	for (i = 0; i < max3191x->nchips; i++) {
+		if (max3191x->mode == STATUS_BYTE_ENABLED) {
+			u8 in	  = ((u8 *)max3191x->xfer.rx_buf)[i * 2];
+			u8 status = ((u8 *)max3191x->xfer.rx_buf)[i * 2 + 1];
+
+			val = (status & 0xf8) != crc8(max3191x_crc8, &in, 1, 0);
+			__assign_bit(i, max3191x->crc_error, val);
+			if (val)
+				dev_err_ratelimited(dev,
+					"chip %d: CRC error\n", i);
+
+			ot  = (status >> 1) & 1;
+			__assign_bit(i, max3191x->overtemp, ot);
+			if (ot)
+				dev_err_ratelimited(dev,
+					"chip %d: overtemperature\n", i);
+
+			if (!max3191x->ignore_uv) {
+				uv1 = !((status >> 2) & 1);
+				__assign_bit(i, max3191x->undervolt1, uv1);
+				if (uv1)
+					dev_err_ratelimited(dev,
+						"chip %d: undervoltage\n", i);
+
+				val = !(status & 1);
+				__assign_bit(i, max3191x->undervolt2, val);
+				if (val && !uv1)
+					dev_warn_ratelimited(dev,
+						"chip %d: voltage warn\n", i);
+			}
+		}
+
+		if (max3191x->fault_pins && !max3191x->ignore_uv) {
+			/* fault pin shared by all chips or per chip */
+			struct gpio_desc *fault_pin =
+				(max3191x->fault_pins->ndescs == 1)
+					? max3191x->fault_pins->desc[0]
+					: max3191x->fault_pins->desc[i];
+
+			val = gpiod_get_value_cansleep(fault_pin);
+			if (val < 0) {
+				dev_err_ratelimited(dev,
+					"GPIO read error %d\n", val);
+				return val;
+			}
+			__assign_bit(i, max3191x->fault, val);
+			if (val && !uv1 && !ot)
+				dev_err_ratelimited(dev,
+					"chip %d: fault\n", i);
+		}
+	}
+
+	return 0;
+}
+
+static bool max3191x_chip_is_faulting(struct max3191x_chip *max3191x,
+				      unsigned int chipnum)
+{
+	/* without status byte the only diagnostic is the fault pin */
+	if (!max3191x->ignore_uv && test_bit(chipnum, max3191x->fault))
+		return true;
+
+	if (max3191x->mode == STATUS_BYTE_DISABLED)
+		return false;
+
+	return test_bit(chipnum, max3191x->crc_error) ||
+	       test_bit(chipnum, max3191x->overtemp)  ||
+	       (!max3191x->ignore_uv &&
+		test_bit(chipnum, max3191x->undervolt1));
+}
+
+static int max3191x_get(struct gpio_chip *gpio, unsigned int offset)
+{
+	struct max3191x_chip *max3191x = gpiochip_get_data(gpio);
+	int ret, chipnum, wordlen = max3191x_wordlen(max3191x);
+	u8 in;
+
+	mutex_lock(&max3191x->lock);
+	ret = max3191x_readout_locked(max3191x);
+	if (ret)
+		goto out_unlock;
+
+	chipnum = offset / MAX3191X_NGPIO;
+	if (max3191x_chip_is_faulting(max3191x, chipnum)) {
+		ret = -EIO;
+		goto out_unlock;
+	}
+
+	in = ((u8 *)max3191x->xfer.rx_buf)[chipnum * wordlen];
+	ret = (in >> (offset % MAX3191X_NGPIO)) & 1;
+
+out_unlock:
+	mutex_unlock(&max3191x->lock);
+	return ret;
+}
+
+static int max3191x_get_multiple(struct gpio_chip *gpio, unsigned long *mask,
+				 unsigned long *bits)
+{
+	struct max3191x_chip *max3191x = gpiochip_get_data(gpio);
+	int ret, bit = 0, wordlen = max3191x_wordlen(max3191x);
+
+	mutex_lock(&max3191x->lock);
+	ret = max3191x_readout_locked(max3191x);
+	if (ret)
+		goto out_unlock;
+
+	while ((bit = find_next_bit(mask, gpio->ngpio, bit)) != gpio->ngpio) {
+		unsigned int chipnum = bit / MAX3191X_NGPIO;
+		unsigned long in, shift, index;
+
+		if (max3191x_chip_is_faulting(max3191x, chipnum)) {
+			ret = -EIO;
+			goto out_unlock;
+		}
+
+		in = ((u8 *)max3191x->xfer.rx_buf)[chipnum * wordlen];
+		shift = round_down(bit % BITS_PER_LONG, MAX3191X_NGPIO);
+		index = bit / BITS_PER_LONG;
+		bits[index] &= ~(mask[index] & (0xff << shift));
+		bits[index] |= mask[index] & (in << shift); /* copy bits */
+
+		bit = (chipnum + 1) * MAX3191X_NGPIO; /* go to next chip */
+	}
+
+out_unlock:
+	mutex_unlock(&max3191x->lock);
+	return ret;
+}
+
+static int max3191x_set_config(struct gpio_chip *gpio, unsigned int offset,
+			       unsigned long config)
+{
+	struct max3191x_chip *max3191x = gpiochip_get_data(gpio);
+	u32 debounce, chipnum, db0_val, db1_val;
+
+	if (pinconf_to_config_param(config) != PIN_CONFIG_INPUT_DEBOUNCE)
+		return -ENOTSUPP;
+
+	if (!max3191x->db0_pins || !max3191x->db1_pins)
+		return -EINVAL;
+
+	debounce = pinconf_to_config_argument(config);
+	switch (debounce) {
+	case 0:
+		db0_val = 0;
+		db1_val = 0;
+		break;
+	case 1 ... 25:
+		db0_val = 0;
+		db1_val = 1;
+		break;
+	case 26 ... 750:
+		db0_val = 1;
+		db1_val = 0;
+		break;
+	case 751 ... 3000:
+		db0_val = 1;
+		db1_val = 1;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if (max3191x->db0_pins->ndescs == 1)
+		chipnum = 0; /* all chips use the same pair of debounce pins */
+	else
+		chipnum = offset / MAX3191X_NGPIO; /* per chip debounce pins */
+
+	mutex_lock(&max3191x->lock);
+	gpiod_set_value_cansleep(max3191x->db0_pins->desc[chipnum], db0_val);
+	gpiod_set_value_cansleep(max3191x->db1_pins->desc[chipnum], db1_val);
+	mutex_unlock(&max3191x->lock);
+	return 0;
+}
+
+static void gpiod_set_array_single_value_cansleep(unsigned int ndescs,
+						  struct gpio_desc **desc,
+						  int value)
+{
+	int i, values[ndescs];
+
+	for (i = 0; i < ndescs; i++)
+		values[i] = value;
+
+	gpiod_set_array_value_cansleep(ndescs, desc, values);
+}
+
+static struct gpio_descs *devm_gpiod_get_array_optional_count(
+				struct device *dev, const char *con_id,
+				enum gpiod_flags flags, unsigned int expected)
+{
+	struct gpio_descs *descs;
+	int found = gpiod_count(dev, con_id);
+
+	if (found == -ENOENT)
+		return NULL;
+
+	if (found != expected && found != 1) {
+		dev_err(dev, "ignoring %s-gpios: found %d, expected %u or 1\n",
+			con_id, found, expected);
+		return NULL;
+	}
+
+	descs = devm_gpiod_get_array_optional(dev, con_id, flags);
+
+	if (IS_ERR(descs)) {
+		dev_err(dev, "failed to get %s-gpios: %ld\n",
+			con_id, PTR_ERR(descs));
+		return NULL;
+	}
+
+	return descs;
+}
+
+static int max3191x_probe(struct spi_device *spi)
+{
+	struct device *dev = &spi->dev;
+	struct max3191x_chip *max3191x;
+	int n, ret;
+
+	max3191x = devm_kzalloc(dev, sizeof(*max3191x), GFP_KERNEL);
+	if (!max3191x)
+		return -ENOMEM;
+	spi_set_drvdata(spi, max3191x);
+
+	max3191x->nchips = 1;
+	device_property_read_u32(dev, "#daisy-chained-devices",
+				 &max3191x->nchips);
+
+	n = BITS_TO_LONGS(max3191x->nchips);
+	max3191x->crc_error   = devm_kcalloc(dev, n, sizeof(long), GFP_KERNEL);
+	max3191x->undervolt1  = devm_kcalloc(dev, n, sizeof(long), GFP_KERNEL);
+	max3191x->undervolt2  = devm_kcalloc(dev, n, sizeof(long), GFP_KERNEL);
+	max3191x->overtemp    = devm_kcalloc(dev, n, sizeof(long), GFP_KERNEL);
+	max3191x->fault       = devm_kcalloc(dev, n, sizeof(long), GFP_KERNEL);
+	max3191x->xfer.rx_buf = devm_kcalloc(dev, max3191x->nchips,
+								2, GFP_KERNEL);
+	if (!max3191x->crc_error || !max3191x->undervolt1 ||
+	    !max3191x->overtemp  || !max3191x->undervolt2 ||
+	    !max3191x->fault     || !max3191x->xfer.rx_buf)
+		return -ENOMEM;
+
+	max3191x->modesel_pins = devm_gpiod_get_array_optional_count(dev,
+				 "maxim,modesel", GPIOD_ASIS, max3191x->nchips);
+	max3191x->fault_pins   = devm_gpiod_get_array_optional_count(dev,
+				 "maxim,fault", GPIOD_IN, max3191x->nchips);
+	max3191x->db0_pins     = devm_gpiod_get_array_optional_count(dev,
+				 "maxim,db0", GPIOD_OUT_LOW, max3191x->nchips);
+	max3191x->db1_pins     = devm_gpiod_get_array_optional_count(dev,
+				 "maxim,db1", GPIOD_OUT_LOW, max3191x->nchips);
+
+	max3191x->mode = device_property_read_bool(dev, "maxim,modesel-8bit")
+				 ? STATUS_BYTE_DISABLED : STATUS_BYTE_ENABLED;
+	if (max3191x->modesel_pins)
+		gpiod_set_array_single_value_cansleep(
+				 max3191x->modesel_pins->ndescs,
+				 max3191x->modesel_pins->desc, max3191x->mode);
+
+	max3191x->ignore_uv = device_property_read_bool(dev,
+						  "maxim,ignore-undervoltage");
+
+	if (max3191x->db0_pins && max3191x->db1_pins &&
+	    max3191x->db0_pins->ndescs != max3191x->db1_pins->ndescs) {
+		dev_err(dev, "ignoring maxim,db*-gpios: array len mismatch\n");
+		devm_gpiod_put_array(dev, max3191x->db0_pins);
+		devm_gpiod_put_array(dev, max3191x->db1_pins);
+		max3191x->db0_pins = NULL;
+		max3191x->db1_pins = NULL;
+	}
+
+	max3191x->xfer.len = max3191x->nchips * max3191x_wordlen(max3191x);
+	spi_message_init_with_transfers(&max3191x->mesg, &max3191x->xfer, 1);
+
+	max3191x->gpio.label = spi->modalias;
+	max3191x->gpio.owner = THIS_MODULE;
+	max3191x->gpio.parent = dev;
+	max3191x->gpio.base = -1;
+	max3191x->gpio.ngpio = max3191x->nchips * MAX3191X_NGPIO;
+	max3191x->gpio.can_sleep = true;
+
+	max3191x->gpio.get_direction = max3191x_get_direction;
+	max3191x->gpio.direction_input = max3191x_direction_input;
+	max3191x->gpio.direction_output = max3191x_direction_output;
+	max3191x->gpio.set = max3191x_set;
+	max3191x->gpio.set_multiple = max3191x_set_multiple;
+	max3191x->gpio.get = max3191x_get;
+	max3191x->gpio.get_multiple = max3191x_get_multiple;
+	max3191x->gpio.set_config = max3191x_set_config;
+
+	mutex_init(&max3191x->lock);
+
+	ret = gpiochip_add_data(&max3191x->gpio, max3191x);
+	if (ret) {
+		mutex_destroy(&max3191x->lock);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int max3191x_remove(struct spi_device *spi)
+{
+	struct max3191x_chip *max3191x = spi_get_drvdata(spi);
+
+	gpiochip_remove(&max3191x->gpio);
+	mutex_destroy(&max3191x->lock);
+
+	return 0;
+}
+
+static int __init max3191x_register_driver(struct spi_driver *sdrv)
+{
+	crc8_populate_msb(max3191x_crc8, MAX3191X_CRC8_POLYNOMIAL);
+	return spi_register_driver(sdrv);
+}
+
+#ifdef CONFIG_OF
+static const struct of_device_id max3191x_of_id[] = {
+	{ .compatible = "maxim,max31910" },
+	{ .compatible = "maxim,max31911" },
+	{ .compatible = "maxim,max31912" },
+	{ .compatible = "maxim,max31913" },
+	{ .compatible = "maxim,max31953" },
+	{ .compatible = "maxim,max31963" },
+	{ }
+};
+MODULE_DEVICE_TABLE(of, max3191x_of_id);
+#endif
+
+static const struct spi_device_id max3191x_spi_id[] = {
+	{ "max31910" },
+	{ "max31911" },
+	{ "max31912" },
+	{ "max31913" },
+	{ "max31953" },
+	{ "max31963" },
+	{ }
+};
+MODULE_DEVICE_TABLE(spi, max3191x_spi_id);
+
+static struct spi_driver max3191x_driver = {
+	.driver = {
+		.name		= "max3191x",
+		.of_match_table	= of_match_ptr(max3191x_of_id),
+	},
+	.probe	  = max3191x_probe,
+	.remove	  = max3191x_remove,
+	.id_table = max3191x_spi_id,
+};
+module_driver(max3191x_driver, max3191x_register_driver, spi_unregister_driver);
+
+MODULE_AUTHOR("Lukas Wunner <lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org>");
+MODULE_DESCRIPTION("GPIO driver for Maxim MAX3191x industrial serializer");
+MODULE_LICENSE("GPL v2");
-- 
2.11.0

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2 1/5] bitops: Introduce assign_bit()

[Andrew Morton]

On Thu, 12 Oct 2017 12:40:10 +0200 Lukas Wunner <lukas@wunner.de> wrote:

> A common idiom is to assign a value to a bit with:
> 
>     if (value)
>         set_bit(nr, addr);
>     else
>         clear_bit(nr, addr);
> 
> Likewise common is the one-line expression variant:
> 
>     value ? set_bit(nr, addr) : clear_bit(nr, addr);
> 
> Commit 9a8ac3ae682e ("dm mpath: cleanup QUEUE_IF_NO_PATH bit
> manipulation by introducing assign_bit()") introduced assign_bit()
> to the md subsystem for brevity.
> 
> Make it available to others, specifically gpiolib and the upcoming
> driver for Maxim MAX3191x industrial serializer chips.
> 
> As requested by Peter Zijlstra, change the argument order to reflect
> traditional "dst = src" in C, hence "assign_bit(nr, addr, value)".

Acked-by: Andrew Morton <akpm@linux-foundation.org>

Re: [PATCH v2 5/5] gpio: Add driver for Maxim MAX3191x industrial serializer

[Lukas Wunner]

[cc += William Breathitt Gray, Geert Uytterhoeven, Phil Reid,
       David Daney, Iban Rodriguez]

The drivers

       gpio-104-dio-48e.c, gpio-74x164.c, gpio-gpio-mm.c,
       gpio-pca953x.c, gpio-thunderx.c, gpio-ws16c48.c,
       gpio-xilinx.c

currently use an inefficient algorithm for their ->set_multiple
callback:  They iterate over every chip (or bank or gpio pin),
check if any bits are set in the mask for this particular chip,
and if so, update the affected GPIOs.  If the mask is sparsely
populated, you'll waste CPU time checking chips even though
they're not affected by the operation at all.

Iterating over the chips is backwards, it is more efficient to
iterate over the bits set in the mask, identify the corresponding
chip and update its affected GPIOs.  The gpio-max3191x.c driver
I posted yesterday contains an example for such an algorithm and
you may want to improve your ->set_mutiple implementation accordingly:


> +static int max3191x_get_multiple(struct gpio_chip *gpio, unsigned long *mask,
> +				 unsigned long *bits)
> +{
> +	struct max3191x_chip *max3191x = gpiochip_get_data(gpio);
> +	int ret, bit = 0, wordlen = max3191x_wordlen(max3191x);
> +
> +	mutex_lock(&max3191x->lock);
> +	ret = max3191x_readout_locked(max3191x);
> +	if (ret)
> +		goto out_unlock;
> +
> +	while ((bit = find_next_bit(mask, gpio->ngpio, bit)) != gpio->ngpio) {
> +		unsigned int chipnum = bit / MAX3191X_NGPIO;
> +		unsigned long in, shift, index;
> +
> +		if (max3191x_chip_is_faulting(max3191x, chipnum)) {
> +			ret = -EIO;
> +			goto out_unlock;
> +		}
> +
> +		in = ((u8 *)max3191x->xfer.rx_buf)[chipnum * wordlen];
> +		shift = round_down(bit % BITS_PER_LONG, MAX3191X_NGPIO);
> +		index = bit / BITS_PER_LONG;
> +		bits[index] &= ~(mask[index] & (0xff << shift));
> +		bits[index] |= mask[index] & (in << shift); /* copy bits */
> +
> +		bit = (chipnum + 1) * MAX3191X_NGPIO; /* go to next chip */
> +	}
> +
> +out_unlock:
> +	mutex_unlock(&max3191x->lock);
> +	return ret;
> +}


I've used a while loop, obviously the same can be achieved with a
for loop but I found that harder to read and it didn't save any LoC.

This particular chip has 8 inputs (= MAX3191X_NGPIO) that can be read
concurrently.

The series containing this driver introduces a ->get_multiple callback.
Since you've implemented a ->set_multiple callback, you may want to
add a ->get_multiple callback to your driver as well.  (If supported
by the hardware, which is not the case for output-only chips such as
gpio-74x164.c.)

You can find the full series here:
https://www.spinics.net/lists/linux-gpio/msg25997.html

If you want to fetch the series from a git repo, use this branch:
https://github.com/RevolutionPi/linux/commits/revpi-4.9

Thanks,

Lukas

Re: [PATCH v2 1/5] bitops: Introduce assign_bit()

[Linus Walleij]

On Thu, Oct 12, 2017 at 12:40 PM, Lukas Wunner <lukas@wunner.de> wrote:

> A common idiom is to assign a value to a bit with:
>
>     if (value)
>         set_bit(nr, addr);
>     else
>         clear_bit(nr, addr);
>
> Likewise common is the one-line expression variant:
>
>     value ? set_bit(nr, addr) : clear_bit(nr, addr);
>
> Commit 9a8ac3ae682e ("dm mpath: cleanup QUEUE_IF_NO_PATH bit
> manipulation by introducing assign_bit()") introduced assign_bit()
> to the md subsystem for brevity.
>
> Make it available to others, specifically gpiolib and the upcoming
> driver for Maxim MAX3191x industrial serializer chips.
>
> As requested by Peter Zijlstra, change the argument order to reflect
> traditional "dst = src" in C, hence "assign_bit(nr, addr, value)".
>
> Cc: Bart Van Assche <bart.vanassche@wdc.com>
> Cc: Alasdair Kergon <agk@redhat.com>
> Cc: Mike Snitzer <snitzer@redhat.com>
> Cc: Linus Walleij <linus.walleij@linaro.org>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Cc: Neil Brown <neilb@suse.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Ingo Molnar <mingo@redhat.com>
> Cc: Theodore Ts'o <tytso@mit.edu>
> Cc: Borislav Petkov <bp@alien8.de>
> Cc: "H. Peter Anvin" <hpa@zytor.com>
> Cc: Denys Vlasenko <dvlasenk@redhat.com>
> Signed-off-by: Lukas Wunner <lukas@wunner.de>

This v2 applied with Andrew's ACK.

Yours,
Linus Walleij

Re: [PATCH v2 2/5] gpio: Introduce ->get_multiple callback

[Linus Walleij]

On Thu, Oct 12, 2017 at 12:40 PM, Lukas Wunner <lukas@wunner.de> wrote:

> SPI-attached GPIO controllers typically read out all inputs in one go.
> If callers desire the values of multipe inputs, ideally a single readout
> should take place to return the desired values.  However the current
> driver API only offers a ->get callback but no ->get_multiple (unlike
> ->set_multiple, which is present).  Thus, to read multiple inputs, a
> full readout needs to be performed for every single value (barring
> driver-internal caching), which is inefficient.
>
> In fact, the lack of a ->get_multiple callback has been bemoaned
> repeatedly by the gpio subsystem maintainer:
> http://www.spinics.net/lists/linux-gpio/msg10571.html
> http://www.spinics.net/lists/devicetree/msg121734.html
>
> Introduce the missing callback.  Add corresponding consumer functions
> such as gpiod_get_array_value().  Amend linehandle_ioctl() to take
> advantage of the newly added infrastructure.  Update the documentation.
>
> Cc: Rojhalat Ibrahim <imr@rtschenk.de>
> Signed-off-by: Lukas Wunner <lukas@wunner.de>
> ---
> Changes v1 -> v2:

Applied this v2 version.

Good work! I'm happy :)

We should take a sweep over existing drivers and see
which in-tree drivers that can exploit this and send some
patches for testing.

Yours,
Linus Walleij

Re: [PATCH v2 0/5] GPIO driver for Maxim MAX3191x

[Linus Walleij]

On Thu, Oct 12, 2017 at 12:40 PM, Lukas Wunner <lukas@wunner.de> wrote:

> - Patch [1/5]: Change the argument order of assign_bit() to reflect
>   traditional "dst = src" in C. (Peter Zijlstra)
>
> - Patch [2/5]: Update documentation. (Linus Walleij)
>   Drop const qualifier from struct gpio_desc ** in all function
>   signatures to avoid a cast when passing a non-const array created
>   with gpiod_get_array(), which is likely the most common use case.

Applied these two so we get the infrastructure in place so others
can use it too.

> - Patch [3/5]: Newly inserted patch to introduce common property for
>   number of daisy-chained devices.
>
> - Patch [4/5]: Add vendor prefix to GPIO identifiers, use boolean
>   instead of integer to select mode, rename boolean to ignore
>   undervoltage alarms, separate compatible strings with newlines.
>   (Rob Herring)
>
> - Patch [5/5]: Optimize algorithm in max3191x_get_multiple() to
>   iterate over the bits in the mask, instead of iterating over every
>   chip, thus implicitly skipping chips which are not selected at all
>   by the mask.

These are pending device tree maintainer review.

Yours,
Linus Walleij

Re: [PATCH v2 5/5] gpio: Add driver for Maxim MAX3191x industrial serializer

[David Daney]

On 10/13/2017 04:11 AM, Lukas Wunner wrote:
> [cc += William Breathitt Gray, Geert Uytterhoeven, Phil Reid,
>         David Daney, Iban Rodriguez]
> 
> The drivers
> 
>         gpio-104-dio-48e.c, gpio-74x164.c, gpio-gpio-mm.c,
>         gpio-pca953x.c, gpio-thunderx.c, gpio-ws16c48.c,
>         gpio-xilinx.c
> 
> currently use an inefficient algorithm for their ->set_multiple
> callback:  They iterate over every chip (or bank or gpio pin),
> check if any bits are set in the mask for this particular chip,
> and if so, update the affected GPIOs.  If the mask is sparsely
> populated, you'll waste CPU time checking chips even though
> they're not affected by the operation at all.
> 
> Iterating over the chips is backwards, it is more efficient to
> iterate over the bits set in the mask, identify the corresponding
> chip and update its affected GPIOs.  The gpio-max3191x.c driver
> I posted yesterday contains an example for such an algorithm and
> you may want to improve your ->set_mutiple implementation accordingly:
> 
> 


Do you have any profiling results that demonstrate significant 
system-wide performance improvements by making such a change?

For the gpio-thunderx.c driver, the words in the bits/mask exactly match 
the banks, so the number of iterations doesn't change with the approach 
you suggest.

In fact, an argument could be made in the other direction.  The 
increased ICache pressure from code bloat and branch prediction misses 
resulting from extra testing of the mask bits could easily make the 
system slower using your suggestion.

Thanks,
David Daney

Re: [PATCH v2 5/5] gpio: Add driver for Maxim MAX3191x industrial serializer

[Lukas Wunner]

On Fri, Oct 13, 2017 at 09:53:06AM -0700, David Daney wrote:
> On 10/13/2017 04:11 AM, Lukas Wunner wrote:
> > The drivers
> >
> >        gpio-104-dio-48e.c, gpio-74x164.c, gpio-gpio-mm.c,
> >        gpio-pca953x.c, gpio-thunderx.c, gpio-ws16c48.c,
> >        gpio-xilinx.c
> >
> > currently use an inefficient algorithm for their ->set_multiple
> > callback:  They iterate over every chip (or bank or gpio pin),
> > check if any bits are set in the mask for this particular chip,
> > and if so, update the affected GPIOs.  If the mask is sparsely
> > populated, you'll waste CPU time checking chips even though
> > they're not affected by the operation at all.
> > 
> > Iterating over the chips is backwards, it is more efficient to
> > iterate over the bits set in the mask, identify the corresponding
> > chip and update its affected GPIOs.  The gpio-max3191x.c driver
> > I posted yesterday contains an example for such an algorithm and
> > you may want to improve your ->set_mutiple implementation accordingly:
>
> Do you have any profiling results that demonstrate significant system-wide
> performance improvements by making such a change?

Sorry, no.


> For the gpio-thunderx.c driver, the words in the bits/mask exactly match the
> banks, so the number of iterations doesn't change with the approach you
> suggest.

No, I was specifically referring to a sparsely populated mask above.
Let's say the mask consists of several 64-bit words and only the last
word has set bits.  With the existing algorithm you have as many
iterations as you have banks:

	for (bank = 0; bank <= chip->ngpio / 64; bank++) {
		set_bits = bits[bank] & mask[bank];
		clear_bits = ~bits[bank] & mask[bank];
		writeq(set_bits, txgpio->register_base + (bank * GPIO_2ND_BANK) + GPIO_TX_SET);
		writeq(clear_bits, txgpio->register_base + (bank * GPIO_2ND_BANK) + GPIO_TX_CLR);
	}

Whereas with the algorithm I'm proposing you only have a single iteration:

	int bit = 0;
	while ((bit = find_next_bit(mask, chip->ngpio, bit)) != chip->ngpio) {
		unsigned int bank = bit / 64;
		set_bits = bits[bank] & mask[bank];
		clear_bits = ~bits[bank] & mask[bank];
		writeq(set_bits, txgpio->register_base + (bank * GPIO_2ND_BANK) + GPIO_TX_SET);
		writeq(clear_bits, txgpio->register_base + (bank * GPIO_2ND_BANK) + GPIO_TX_CLR);
		bit = (bank + 1) * 64; /* continue search from next bank */
	}

find_next_bit() uses the clz instruction on ARM, similar instructions
exist on most arches, so the search is very fast.

Or did you mean that find_next_bit() iterates over the words to find
the first one with set bits?

BTW, why isn't there a check in your code whether mask[bank] != 0,
n which case the mmio writes can be skipped?


> In fact, an argument could be made in the other direction.  The increased
> ICache pressure from code bloat and branch prediction misses resulting from
> extra testing of the mask bits could easily make the system slower using
> your suggestion.

I don't quite follow, where should the branch prediction misses come from,
the number of branch instructions is the same with both algorithms, and
fewer with the proposed algorithm on sparsely populated masks.

Also, 2 additional LoC is hardly code bloat.

Are you referring to the call to find_next_bit() specifically?

Thanks,

Lukas

Re: [PATCH v2 5/5] gpio: Add driver for Maxim MAX3191x industrial serializer

[David Daney]

On 10/13/2017 02:29 PM, Lukas Wunner wrote:
> On Fri, Oct 13, 2017 at 09:53:06AM -0700, David Daney wrote:
>> On 10/13/2017 04:11 AM, Lukas Wunner wrote:
>>> The drivers
>>>
>>>         gpio-104-dio-48e.c, gpio-74x164.c, gpio-gpio-mm.c,
>>>         gpio-pca953x.c, gpio-thunderx.c, gpio-ws16c48.c,
>>>         gpio-xilinx.c
>>>
>>> currently use an inefficient algorithm for their ->set_multiple
>>> callback:  They iterate over every chip (or bank or gpio pin),
>>> check if any bits are set in the mask for this particular chip,
>>> and if so, update the affected GPIOs.  If the mask is sparsely
>>> populated, you'll waste CPU time checking chips even though
>>> they're not affected by the operation at all.
>>>
>>> Iterating over the chips is backwards, it is more efficient to
>>> iterate over the bits set in the mask, identify the corresponding
>>> chip and update its affected GPIOs.  The gpio-max3191x.c driver
>>> I posted yesterday contains an example for such an algorithm and
>>> you may want to improve your ->set_mutiple implementation accordingly:
>>
>> Do you have any profiling results that demonstrate significant system-wide
>> performance improvements by making such a change?
> 
> Sorry, no.
> 
> 
>> For the gpio-thunderx.c driver, the words in the bits/mask exactly match the
>> banks, so the number of iterations doesn't change with the approach you
>> suggest.
> 
> No, I was specifically referring to a sparsely populated mask above.
> Let's say the mask consists of several 64-bit words and only the last
> word has set bits.  With the existing algorithm you have as many
> iterations as you have banks:
> 
> 	for (bank = 0; bank <= chip->ngpio / 64; bank++) {
> 		set_bits = bits[bank] & mask[bank];
> 		clear_bits = ~bits[bank] & mask[bank];
> 		writeq(set_bits, txgpio->register_base + (bank * GPIO_2ND_BANK) + GPIO_TX_SET);
> 		writeq(clear_bits, txgpio->register_base + (bank * GPIO_2ND_BANK) + GPIO_TX_CLR);
> 	}
> 
> Whereas with the algorithm I'm proposing you only have a single iteration:
> 
> 	int bit = 0;
> 	while ((bit = find_next_bit(mask, chip->ngpio, bit)) != chip->ngpio) {
> 		unsigned int bank = bit / 64;
> 		set_bits = bits[bank] & mask[bank];
> 		clear_bits = ~bits[bank] & mask[bank];
> 		writeq(set_bits, txgpio->register_base + (bank * GPIO_2ND_BANK) + GPIO_TX_SET);
> 		writeq(clear_bits, txgpio->register_base + (bank * GPIO_2ND_BANK) + GPIO_TX_CLR);
> 		bit = (bank + 1) * 64; /* continue search from next bank */
> 	}
> 
> find_next_bit() uses the clz instruction on ARM, similar instructions
> exist on most arches, so the search is very fast.
> 

There are a maximum of two banks.

For some thunderx SoCs there is only a single bank, so my code has 
exactly one iteration.  For the SoCs with two banks, my code has exactly 
two iterations.  There are no function calls in the loop.

Your suggested code does a function call to the find_next_bit() function 
for each iteration.  I assert that your method will be slower.

Unless you can show an instruction trace where your code results in 
fewer instructions being executed, I would NAK the patch.

I can't believe I have spent so much time on this issue, for which no 
matter what you do, will not impact system performance in any measurable 
way.


David.

Re: [PATCH v2 5/5] gpio: Add driver for Maxim MAX3191x industrial serializer

[Lukas Wunner]

On Fri, Oct 13, 2017 at 03:12:41PM -0700, David Daney wrote:
> There are a maximum of two banks.
> 
> For some thunderx SoCs there is only a single bank, so my code has exactly
> one iteration.  For the SoCs with two banks, my code has exactly two
> iterations.  There are no function calls in the loop.

Fair enough.  So as you've pointed out, if the number of GPIOs per bank
is identical to the number of bits in an unsigned long, the iteration
count is the same with both algorithms and just iterating over the
banks results in simpler and less code.  Doubly so if there's just
one or two banks.  Those are good points.

The situation is different with e.g. gpio-max3191x.c and gpio-74x164.c
where each bank has only 8 GPIOs.


> I can't believe I have spent so much time on this issue, for which no matter
> what you do, will not impact system performance in any measurable way.

The use case that I'm referring to in the cover letter of this series
are PLCs which poll all inputs and write all outputs in cycles of e.g.
every 500 us to 10 ms, so optimizations pay off.

Thanks,

Lukas

Re: [PATCH v2 3/5] dt-bindings: Document common property for daisy-chained devices

[Jonathan Cameron]

On Thu, 12 Oct 2017 12:40:10 +0200
Lukas Wunner <lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org> wrote:

> Many serially-attached GPIO and IIO devices are daisy-chainable.
> 
>     Examples for GPIO devices are Maxim MAX3191x and TI SN65HVS88x:
>     https://datasheets.maximintegrated.com/en/ds/MAX31913.pdf
>     http://www.ti.com/lit/ds/symlink/sn65hvs880.pdf
> 
>     Examples for IIO devices are TI DAC128S085 and TI DAC161S055:
>     http://www.ti.com/lit/ds/symlink/dac128s085.pdf
>     http://www.ti.com/lit/ds/symlink/dac161s055.pdf
> 
> We already have drivers for daisy-chainable devices in the tree but
> their devicetree bindings are somewhat inconsistent and ill-named:
> 
>     The gpio-74x164.c driver uses "registers-number" to convey the
>     number of devices in the daisy-chain.  (Sans vendor prefix,
>     multiple vendors sell compatible versions of this chip.)
> 
>     The gpio-pisosr.c driver takes a different approach and calculates
>     the number of devices in the daisy-chain by dividing the common
>     "ngpios" property (Documentation/devicetree/bindings/gpio/gpio.txt)
>     by 8 (which assumes that each chip has 8 inputs).
> 
> Let's standardize on a common "#daisy-chained-devices" property.
> That name was chosen because it's the term most frequently used in
> datasheets.  (A less frequently used synonym is "cascaded devices".)
> 
> Signed-off-by: Lukas Wunner <lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org>
Makes sense to me.
Acked-by: Jonathan Cameron <Jonathan.Cameron-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>
> ---
> Changes v1 -> v2:
> 
> - Newly inserted patch.
> 
>  .../devicetree/bindings/common-properties.txt      | 26 ++++++++++++++++++++++
>  1 file changed, 26 insertions(+)
> 
> diff --git a/Documentation/devicetree/bindings/common-properties.txt b/Documentation/devicetree/bindings/common-properties.txt
> index 697714f8d75c..a3448bfa1c82 100644
> --- a/Documentation/devicetree/bindings/common-properties.txt
> +++ b/Documentation/devicetree/bindings/common-properties.txt
> @@ -1,4 +1,8 @@
>  Common properties
> +=================
> +
> +Endianness
> +----------
>  
>  The Devicetree Specification does not define any properties related to hardware
>  byteswapping, but endianness issues show up frequently in porting Linux to
> @@ -58,3 +62,25 @@ dev: dev@40031000 {
>  	      ...
>  	      little-endian;
>  };
> +
> +Daisy-chained devices
> +---------------------
> +
> +Many serially-attached GPIO and IIO devices are daisy-chainable.  To the
> +host controller, a daisy-chain appears as a single device, but the number
> +of inputs and outputs it provides is the sum of inputs and outputs provided
> +by all of its devices.  The driver needs to know how many devices the
> +daisy-chain comprises to determine the amount of data exchanged, how many
> +inputs and outputs to register and so on.
> +
> +Optional properties:
> + - #daisy-chained-devices: Number of devices in the daisy-chain (default is 1).
> +
> +Example:
> +gpio@0 {
> +	      compatible = "name";
> +	      reg = <0>;
> +	      gpio-controller;
> +	      #gpio-cells = <2>;
> +	      #daisy-chained-devices = <3>;
> +};

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2 4/5] dt-bindings: gpio: max3191x: Document new driver

[Rob Herring]

On Thu, Oct 12, 2017 at 12:40:10PM +0200, Lukas Wunner wrote:
> Add device tree bindings for Maxim MAX3191x industrial serializer.
> 
> Cc: Mathias Duckeck <m.duckeck@kunbus.de>
> Signed-off-by: Lukas Wunner <lukas@wunner.de>
> ---
> Changes v1 -> v2:
> 
> - Add vendor prefix to GPIO identifiers, use boolean instead of
>   integer to select mode, rename boolean to ignore undervoltage
>   alarms, separate compatible strings with newlines. (Rob Herring)
> 
>  .../devicetree/bindings/gpio/gpio-max3191x.txt     | 59 ++++++++++++++++++++++
>  1 file changed, 59 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/gpio/gpio-max3191x.txt

Acked-by: Rob Herring <robh@kernel.org>

Re: [PATCH v2 3/5] dt-bindings: Document common property for daisy-chained devices

[Rob Herring]

On Thu, Oct 12, 2017 at 12:40:10PM +0200, Lukas Wunner wrote:
> Many serially-attached GPIO and IIO devices are daisy-chainable.
> 
>     Examples for GPIO devices are Maxim MAX3191x and TI SN65HVS88x:
>     https://datasheets.maximintegrated.com/en/ds/MAX31913.pdf
>     http://www.ti.com/lit/ds/symlink/sn65hvs880.pdf
> 
>     Examples for IIO devices are TI DAC128S085 and TI DAC161S055:
>     http://www.ti.com/lit/ds/symlink/dac128s085.pdf
>     http://www.ti.com/lit/ds/symlink/dac161s055.pdf
> 
> We already have drivers for daisy-chainable devices in the tree but
> their devicetree bindings are somewhat inconsistent and ill-named:
> 
>     The gpio-74x164.c driver uses "registers-number" to convey the
>     number of devices in the daisy-chain.  (Sans vendor prefix,
>     multiple vendors sell compatible versions of this chip.)
> 
>     The gpio-pisosr.c driver takes a different approach and calculates
>     the number of devices in the daisy-chain by dividing the common
>     "ngpios" property (Documentation/devicetree/bindings/gpio/gpio.txt)
>     by 8 (which assumes that each chip has 8 inputs).
> 
> Let's standardize on a common "#daisy-chained-devices" property.
> That name was chosen because it's the term most frequently used in
> datasheets.  (A less frequently used synonym is "cascaded devices".)
> 
> Signed-off-by: Lukas Wunner <lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org>
> ---
> Changes v1 -> v2:
> 
> - Newly inserted patch.
> 
>  .../devicetree/bindings/common-properties.txt      | 26 ++++++++++++++++++++++
>  1 file changed, 26 insertions(+)

Acked-by: Rob Herring <robh-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2 3/5] dt-bindings: Document common property for daisy-chained devices

[Linus Walleij]

On Thu, Oct 12, 2017 at 12:40 PM, Lukas Wunner <lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org> wrote:

> Many serially-attached GPIO and IIO devices are daisy-chainable.
>
>     Examples for GPIO devices are Maxim MAX3191x and TI SN65HVS88x:
>     https://datasheets.maximintegrated.com/en/ds/MAX31913.pdf
>     http://www.ti.com/lit/ds/symlink/sn65hvs880.pdf
>
>     Examples for IIO devices are TI DAC128S085 and TI DAC161S055:
>     http://www.ti.com/lit/ds/symlink/dac128s085.pdf
>     http://www.ti.com/lit/ds/symlink/dac161s055.pdf
>
> We already have drivers for daisy-chainable devices in the tree but
> their devicetree bindings are somewhat inconsistent and ill-named:
>
>     The gpio-74x164.c driver uses "registers-number" to convey the
>     number of devices in the daisy-chain.  (Sans vendor prefix,
>     multiple vendors sell compatible versions of this chip.)
>
>     The gpio-pisosr.c driver takes a different approach and calculates
>     the number of devices in the daisy-chain by dividing the common
>     "ngpios" property (Documentation/devicetree/bindings/gpio/gpio.txt)
>     by 8 (which assumes that each chip has 8 inputs).
>
> Let's standardize on a common "#daisy-chained-devices" property.
> That name was chosen because it's the term most frequently used in
> datasheets.  (A less frequently used synonym is "cascaded devices".)
>
> Signed-off-by: Lukas Wunner <lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org>
> ---
> Changes v1 -> v2:
>
> - Newly inserted patch.

Patch applied with the ACKs.

Yours,
Linus Walleij
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2 4/5] dt-bindings: gpio: max3191x: Document new driver

[Linus Walleij]

On Thu, Oct 12, 2017 at 12:40 PM, Lukas Wunner <lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org> wrote:

> Add device tree bindings for Maxim MAX3191x industrial serializer.
>
> Cc: Mathias Duckeck <m.duckeck-XB/JSsFECOqzQB+pC5nmwQ@public.gmane.org>
> Signed-off-by: Lukas Wunner <lukas-JFq808J9C/izQB+pC5nmwQ@public.gmane.org>
> ---
> Changes v1 -> v2:

Patch applied with Rob's ACK.

Yours,
Linus Walleij
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v2 5/5] gpio: Add driver for Maxim MAX3191x industrial serializer

[Linus Walleij]

On Thu, Oct 12, 2017 at 12:40 PM, Lukas Wunner <lukas@wunner.de> wrote:

> The driver was developed for and tested with the MAX31913 built into
> the Revolution Pi by KUNBUS, but should work with all members of the
> MAX3191x family:
>
> MAX31910: low power
> MAX31911: LED drivers
> MAX31912: LED drivers + 2nd voltage monitor + low power
> MAX31913: LED drivers + 2nd voltage monitor
> MAX31953: LED drivers + 2nd voltage monitor + isolation
> MAX31963: LED drivers + 2nd voltage monitor + isolation + buck regulator
>
> Cc: Mathias Duckeck <m.duckeck@kunbus.de>
> Signed-off-by: Lukas Wunner <lukas@wunner.de>

Patch applied.

Yours,
Linus Walleij